Volume 37, Issue 15S May 20, 2019

2019 ASCO Annual Meeting Proceedings

2019 ASCO ANNUAL MEETING PROCEEDINGS VOLUME 37/ISSUE 15S/May 20, 2019/PP 1S-638S)

Support for this program is funded through 55th Annual Meeting

May 31-June 4, 2019
McCormick Place
Chicago, IL
 55th
Annual Meeting of the
American Society of Clinical Oncology
May 31-June 4, 2019
Chicago, Illinois

2019 Annual Meeting Proceedings

(a supplement to Journal of Clinical Oncology)

Copyright 2019 American Society of Clinical Oncology

Editor: Michael A. Carducci, MD
Managing Editor: Jennifer Giblin
Production Manager: Donna Dottellis

Requests for permission to reprint abstracts should be directed to Intellectual Property Rights
Manager, American Society of Clinical Oncology, 2318 Mill Road, Suite 800, Alexandria, VA 22314.
Tel: 571-483-1300; Fax: 571-366-9530; Email: [email protected]. Editorial correspondence
and production questions should be addressed to Managing Editor, Annual Meeting Proceedings,
American Society of Clinical Oncology, 2318 Mill Road, Suite 800, Alexandria, VA 22314. Email:
[email protected].

Copyright © 2019 American Society of Clinical Oncology. All rights reserved. No part of this
publication may be reproduced or transmitted in any form or by any means, electronic or mechanical,
including photocopy, recording, or any information storage and retrieval system, without written
permission by the Society.

The American Society of Clinical Oncology assumes no responsibility for errors or omissions in this
document. The reader is advised to check the appropriate medical literature and the product
information currently provided by the manufacturer of each drug to be administered to verify the
dosage, the method and duration or administration, or contraindications. It is the responsibility of the
treating physician or other health care professional, relying on independent experience and
knowledge of the patient, to determine drug, disease, and the best treatment for the patient.

Abstract management and indexing provided by CONFEX, Cumberland, RI. Editorial services
provided by KWF Editorial. Composition services and print production provided by Sheridan.
 May 20, 2019
Volume 37, Issue 15_suppl

Contents
2019 ASCO ANNUAL MEETING PROCEEDINGS ABSTRACTS
Special Award Lecture Abstracts ....................................................................................................................................... 1
Plenary Session ................................................................................................................................................................ 5
Special Clinical Science Symposia ................................................................................................................................... 6
Breast Cancer—Local/Regional/Adjuvant ........................................................................................................................... 9
Breast Cancer—Metastatic ............................................................................................................................................... 35
Cancer Prevention, Hereditary Genetics, and Epidemiology ............................................................................................... 64
Central Nervous System Tumors ........................................................................................................................................ 87
Developmental Immunotherapy and Tumor Immunobiology ................................................................................................ 106
Developmental Therapeutics and Tumor Biology (Nonimmuno) .......................................................................................... 148
Gastrointestinal (Colorectal) Cancer .................................................................................................................................. 191
Gastrointestinal (Noncolorectal) Cancer ............................................................................................................................ 222
Genitourinary (Nonprostate) Cancer ................................................................................................................................... 264
Genitourinary (Prostate) Cancer ........................................................................................................................................ 288
Gynecologic Cancer .......................................................................................................................................................... 314
Head and Neck Cancer ..................................................................................................................................................... 342
Health Services Research, Clinical Informatics, and Quality of Care .................................................................................. 367

continued on following page

Journal of Clinical Oncology (ISSN 0732-183X) is published 36 times a year, three times monthly, by the American Society of Clinical Oncology, 2318 Mill Road,
Suite 800, Alexandria, VA 22314. Periodicals postage is paid at Alexandria, VA, and at additional mailing offices. Publication Mail Agreement Number 863289.
Editorial correspondence should be addressed to Stephen A. Cannistra, MD, Journal of Clinical Oncology, 2318 Mill Road, Suite 800, Alexandria, VA 22314.
Phone: 703-797-1900; Fax: 703-684-8720. E-mail: [email protected]. Internet: jco.org.
Journal of Clinical Oncology® is a registered trademark of American Society of Clinical Oncology, Inc.
POSTMASTER: Send address changes to American Society of Clinical Oncology, 2318 Mill Road, Suite 800, Alexandria, VA 22314. ASCO members should send
changes of address to American Society of Clinical Oncology, 2318 Mill Road, Suite 800, Alexandria, VA 22314. Nonmembers should send changes of address to
Journal of Clinical Oncology Customer Service, 2318 Mill Road, Suite 800, Alexandria, VA 22314.
2019 annual subscription rates, effective September 1, 2018: United States and possessions: individual, $625 one year, $1,188 two years; single issue, $40.
International: individual, $867 one year, $1,647 two years; single issue, $50. Institutions: bundled (print/online): Tier 1: $1,110 US, $1,540 Int’l; Tier 2: $1,280 US,
$1,700 Int’l; Tier 3: $1,849 US, $2,250 Int’l; Tier 4: contact JCO for quote. Institutions: online only, worldwide: Tier 1: $936; Tier 2: $1,071; Tier 3: $1,545; Tier 4:
contact JCO for quote. See http://ascopubs.org/pb-assets/pdfs/ASCO-institutional-catalog-1536072550257.pdf for descriptions of each tier. Student and resident:
United States and possessions: $303; all other countries, $421. To receive student/resident rate, orders must be accompanied by name of affiliated institution, date of
term, and the signature of program/residency coordinator on institution letterhead. Orders will be billed at individual rate until
proof of status is received. Current prices are in effect for back volumes and back issues. Back issues sold in conjunction with
a subscription rate are on a prorated basis. Subscriptions are accepted on a 12-month basis. Prices are subject to change
without notice. Single issues, both current and back, exist in limited quantities and are offered for sale subject to availability.
JCO Legacy Archive (electronic back issues from January 1983 through December 1998) is also available; please inquire.
 May 20, 2019
Volume 37, Issue 15_suppl

Hematologic Malignancies—Leukemia, Myelodysplastic Syndromes, and Allotransplant ................................................... 406

Hematologic Malignancies—Lymphoma and Chronic Lymphocytic Leukemia ..................................................................... 424
Hematologic Malignancies—Plasma Cell Dyscrasia .......................................................................................................... 443
Lung Cancer—Non-Small Cell Local-Regional/Small Cell/Other Thoracic Cancers ............................................................. 457
Lung Cancer—Non-Small Cell Metastatic .......................................................................................................................... 477
Melanoma/Skin Cancers ................................................................................................................................................... 508
Pediatric Oncology ............................................................................................................................................................ 535
Professional Development ................................................................................................................................................. 551
Sarcoma ........................................................................................................................................................................... 562
Symptoms and Survivorship .............................................................................................................................................. 582

jco.org
@JCO_ASCO
 American Society of Clinical Oncology
55th Annual Meeting
Descriptions of Scientific Sessions
Plenary Session
The Plenary Session includes abstracts selected by the Scientific Program Committee as having
practice-changing findings of the highest scientific merit.
Highlights of the Day Sessions
Highlights of the Day Sessions invite expert discussants to provide an overview of the previous day’s Oral
Abstract presentations, focusing on key findings and putting abstracts into clinical context.
Oral Abstract Sessions
Oral Abstract Sessions include didactic presentations of abstracts of the highest scientific merit, as
determined by the Scientific Program Committee. Experts in the field serve as discussants and provide
comprehensive themed discussions of the findings from the abstracts.
Clinical Science Symposia
Clinical Science Symposia provide a forum for science in oncology, combining didactic lectures on a specific
topic with abstract presentations. Experts in the field serve as discussants, placing studies in the appropriate
context and critically discussing the applicability of the conclusions in clinical practice. Three special Clinical
Science Symposia will be designated around specific topics that cut across cancer types.
Poster Discussion Sessions
Select posters from the Poster Sessions will be discussed by expert discussants, with the abstract authors
participating in a question and answer period as panel members. These sessions will be followed by
networking with the discussants and authors.
Poster Sessions
Poster Sessions include selected abstracts of clinical research in poster format. Trials in Progress (TPS)
abstracts are presented within a track’s Poster Session.

Publication-Only Abstracts
Publication-only abstracts were selected to be published online in conjunction with the Annual Meeting, but
will not be presented at the Meeting.

All presented and publication-only abstracts are citable to this Journal of Clinical Oncology supplement. For
citation examples, please see the Letter From the Editor.
This publication contains abstracts selected by the ASCO Scientific Program Committee for presentation at
the 2019 Annual Meeting. Abstracts selected for electronic publication only are available in full-text versions
online through ASCO.org and JCO.org. The type of session, the day, and the session start/end times are located
to the right of the abstract number for scheduled presentations. To determine the location of the abstract
session, refer to the Annual Meeting Program or the iPlanner, the online version of the Annual Meeting Program,
available at am.asco.org.

Dates and times are subject to change.

All modifications will be posted on am.asco.org.
Letter From the Editor

T he 2019 ASCO Annual Meeting Proceedings (a sup-

plement to Journal of Clinical Oncology) is an enduring
record of the more than 2,400 abstracts selected by the
abstracts will be publicly released during the Annual
Meeting. Print versions of these abstracts will be available
onsite at the Annual Meeting in the ASCO Daily News.
ASCO Scientific Program Committee for presentation at
the 55th ASCO Annual Meeting. Accepted abstracts not All abstracts carry Journal of Clinical Oncology citations. The
presented at the meeting are included in the online sup- following are citation examples for print and electronic
abstracts:
plement to the May 20 issue of Journal of Clinical Oncology
at JCO.org. J Clin Oncol 37:5s, 2019 (suppl; abstr LBA1)
The majority of abstracts selected for presentation are in- J Clin Oncol 37, 2019 (suppl; abstr e12000)
cluded here in full and are categorized by scientific track.
Abstracts can be also accessed online through the ASCO Should you have any questions or comments about this
Meeting Library (meetinglibrary.asco.org). Online abstracts publication, we encourage you to provide feedback by con-
include the full list of abstract authors and their disclosure tacting us at [email protected].
information.
Late-Breaking Abstracts are represented here by abstract Michael A. Carducci, MD
title and first author only. The full-text versions of these Editor, ASCO Annual Meeting Proceedings
Journal of Clinical Oncology (ISSN 0732-183X) is published 36 times a Important Tiers and Pricing Notes
year, three times monthly, by the American Society of Clinical Oncology, Additional rates along with tier descriptions are available online at
2318 Mill Road, Suite 800, Alexandria, VA 22314. Periodicals postage is http://ascopubs.org/pb-assets/pdfs/ASCO-institutional-catalog-
paid at Alexandria, VA, and at additional mailing offices. 1536072550257.pdf
Postmaster 1. Prices are in effect from September 1, 2018, through August 31,
Send all changes of address for Journal of Clinical 2019. Prices are subject to change.
Oncology subscribers to: 2. Print-only subscriptions or additional print subscriptions are
JCO Customer Service available for $940 in the US and $1,359 outside the US.
2318 Mill Road, Suite 800
3. For multisite licenses, please contact the appropriate agent
Alexandria, VA 22314
for a quote.
Editorial Correspondence (manuscript-related inquiries):
4. Subscribers outside the US, add $150 per print subscription
Stephen A. Cannistra, MD, Editor-in-Chief
for expedited delivery.
Journal of Clinical Oncology
2318 Mill Road, Suite 800 5. Single-issue price: $40 US, $50 international.
Alexandria, VA 22314 6. Prices quoted are in US dollars, and payments must be made in
Phone: 703-797-1900; Fax: 703-684-8720 US dollars.
E-mail: [email protected]; Internet: jco.org
7. Except on consortia orders, the publisher allows for a 5% discount on
American Society of Clinical Oncology (membership-related Tiers 1-3 to recognized subscription agents.
inquiries):
Prices are subject to change without notice. Current prices are in
ASCO Member Services
effect for back volumes and back issues. Single issues, both current and
2318 Mill Road, Suite 800
back, exist in limited quantities and are offered for sale subject to
Alexandria, VA 22314
availability. Back issues sold in conjunction with a subscription are on
Phone: 703-299-0158; Toll-free: 888-282-2552
a prorated basis.
Fax: 703-299-0255
E-mail: [email protected]; Internet: www.asco.org Advertising Sales
Hours: Monday-Friday, 8:30 a.m.-5:00 p.m. Eastern Time Harborside
94 North Woodhull Road
Customer Service, Subscriptions, and Changes of Address: Huntington, NY 11743
JCO Customer Service National Accounts Representatives
2318 Mill Road, Suite 800 Nicholas Vroom
Alexandria, VA 22314 Phone: (631) 935-7677; e-mail: [email protected]
Phone: 703-519-1430; Toll-free: 888-273-3508; Fax: 703-518-8155 Ariana Aston
E-mail: [email protected] Phone: (631) 935-7678; e-mail: [email protected]
Internet orders/renewals: ascopubs.org/jco/site/subscriptions/ http://harborsidepress.com/advertising/asco-journals
2019 SUBSCRIPTION RATES Business-to-Business Sales
Individual Prices Rick Werdann
Domestic (US) Print + Online Springer Healthcare, LLC
Individuals in training $303 233 Spring Street
Individuals (1 year) $625 New York, NY 10013
International Print + Online Phone: 212-460-1523; Mobile: 646-209-1840
Individuals in training $421 E-mail: [email protected]
Individuals (1 year) $867 Internet: www.SpringerHealthcare.com
Institutional Prices
LICENSES AND CONSORTIA
Domestic (US) Print + Online Online Only US, Canada, and Europe-Single Site License
Tier 1 $1,110 $936 ASCO Customer Service
2318 Mill Road, Suite 800
Tier 2 $1,280 $1,071
Alexandria, VA 22314
Tier 3 $1,849 $1,545
Phone: 1-888-282-2552 or 1-703-299-0158
Tier 4 Call for quote Call for quote
Fax: 703-299-0255
International Print + Online Online Only E-mail: [email protected]
Tier 1 $1,540 $936
US, Canada, and Europe-Multisite Licenses and Consortia
Tier 2 $1,700 $1,071 David Charles
Tier 3 $2,250 $1,545 eLicensing
Tier 4 Call for quote Call for quote 92 Avenue du General de Gaulle
78600 Maisons-Laffitte, France
Orders and Payments Phone/Fax: +33-1-39-12-29-29
P.O. Box 37211 Baltimore, MD 21279-3211 E-mail: [email protected]
Japan Permissions Requests
USACO Corporation Licensing, Rights, and Permissions Division
2-17-12 Higashi-Azuba Minato-ku American Society of Clinical Oncology
Tokyo, Japan 106-0044 2318 Mill Road, Suite 800
Phone: +81-3-3505-3256; Fax: +81-3-3505-6282 Alexandria, VA 22314
E-mail: [email protected]; Phone: 571-483-1722; Fax: 703-518-5094
Internet: www.usaco.co.jp E-mail: [email protected]

China Free Public Access

Charlesworth China Journal of Clinical Oncology (JCO) provides free online access to
Room 1105, Building 9, Jianwai SOHO original research articles older than one year at jco.org. In addition,
No. 39 Dongsanhuan Zhonglu all ASCO Special Articles, Rapid Communications, Editorials,
Chaoyang District, Beijing 100022 Comments and Controversies articles, the Art of Oncology series, and
PR China Correspondence articles are free immediately upon publication.
Phone: +86-10-58696201;
Fax: +86-10-58696201 Disclaimer
E-mail: [email protected] The ideas and opinions expressed in JCO do not necessarily reflect
Internet: www.charlesworth.com.cn (in Mandarin) and those of the American Society of Clinical Oncology (ASCO). The
www.charlesworth.com mention of any product, service, or therapy in this publication or in
any advertisement in this publication should not be construed as
India
an endorsement of the products mentioned. It is the responsibility
BMJ
of the treating physician or other health care provider, relying on
Mindmill Corporate Tower
independent experience and knowledge of the patient, to determine
6th Floor, 24 A, Film City, Sector 16 A
drug dosages and the best treatment for the patient. Readers are
Noida, 201301, Uttar Pradesh
advised to check the appropriate medical literature and the product
India
information currently provided by the manufacturer of each drug
Phone: +91-120-4345733-38
to be administered to verify approved uses, the dosage, method,
E-mail: [email protected]
and duration of administration, or contraindications. Readers
South Korea are also encouraged to contact the manufacturer with questions
EBSCO Korea about the features or limitations of any products. ASCO assumes
Phone: +82-2-598-2571 no responsibility for any injury or damage to persons or property
E-mail: [email protected] arising out of or related to any use of the material contained in this
publication or to any errors or omissions.
Taiwan
EBSCO Taiwan Copyright
Phone: +886 2-27751596 Copyright © 2019 by American Society of Clinical Oncology unless
E-mail: [email protected] otherwise indicated. All rights reserved. No part of this publication
Latin America, Caribbean, Middle East, and North Africa may be reproduced or transmitted in any form or by any means now
Accucoms (US), Inc. or hereafter known, electronic or mechanical, including photocopy,
PO Box 1651 recording, or any information storage and retrieval system, without
Lansdale, PA 19446 permission in writing from the Publisher. Printed in the United States
Phone: 1-267-646-1118 of America.
Fax: 215-660-5042 The appearance of the code at the bottom of the left column of the
E-mail (Latin America): [email protected] first page of an article in this journal indicates the copyright owner’s
E-mail (Middle East): [email protected] consent that copies of the article may be made for personal or
Internet: www.accucoms.com internal use, or for the personal or internal use of specific clients,
for those registered with the Copyright Clearance Center, Inc. (222
Australia and New Zealand
Rosewood Drive, Danvers, MA 01923; 978-750-8400; www.
BMJ
copyright.com). This consent is given on the condition that the copier
Phone: +61-414-234-686 pay the stated per-copy fee for that article through the Copyright
E-mail: [email protected] Clearance Center, Inc., for copying beyond that permitted by Sections
Indonesia, Malaysia, Philippines, Singapore, Thailand, and Vietnam 107 or 108 of the US Copyright Law. This consent does not extend
BMJ to other kinds of copying, such as copying for general distribution,
9 Temasek Boulevard, #29-01 for advertising or promotional purposes, for creating new collective
Suntec Tower 2 works, or for resale. Absence of the code indicates that the material
Singapore 038983 may not be processed through the Copyright Clearance Center, Inc.
Phone: +65-3157-1351 Journal of Clinical Oncology® is a registered trademark of American
E-mail: [email protected] Society of Clinical Oncology, Inc.
 ASCO Abstracts Policy

Public Release of Abstracts

The abstracts published in the 2019 ASCO Annual Meeting Proceedings, including those abstracts published but not
presented at the Meeting, were publicly released by ASCO at 5:00 PM (EDT) on Wednesday, May 15, 2019. These
abstracts are publicly available online through ASCO.org, the official website of the Society. Late-Breaking Abstracts,
which include all Plenary Abstracts, will be publicly released according to the following schedule:

· Late-Breaking Abstracts presented in a press briefing or scientific presentation (whichever comes first) on Saturday,
June 1, will be publicly released Saturday, June 1, through ASCO.org at 7:30 AM (EDT).
· Late-Breaking Abstracts presented in a press briefing or scientific presentation (whichever comes first) on Sunday,
June 2, will be publicly released Sunday, June 2, through ASCO.org at 7:30 AM (EDT).
· Late-Breaking Abstracts presented in a press briefing or scientific presentation (whichever comes first) on Monday,
June 3, or Tuesday, June 4, will be publicly released Monday, June 3, through ASCO.org at 7:30 AM (EDT).

Late-Breaking Abstracts will be available in ASCO Daily News on the day of their scientific presentation, with the
exception of abstracts presented on Friday (these will appear in the Saturday issue) and those presented on Tuesday
(these will appear in the Monday issue).
In the unlikely event that ASCO publicly releases an abstract in advance of the scheduled time, the release will be
publicly announced on ASCO.org.

Copyright 2019 American Society of Clinical Oncology

 Conflict of Interest Disclosure

As the CE provider for the Meeting, ASCO is committed to balance, objectivity, and scientific rigor in the management of
financial interactions with for-profit health care companies that could create real or perceived conflicts of interest.
Participants in the Meeting have disclosed their financial relationships in accordance with ASCO’s Policy for
Relationships with Companies; review the policy at asco.org/rwc.
ASCO offers a comprehensive disclosure management system, using one disclosure for all ASCO activities. Members and
participants in activities use coi.asco.org to disclose all interactions with companies. Their disclosure is kept on file and
can be confirmed or updated with each new activity.
Please email [email protected] with specific questions or concerns.
2019 ASCO Annual Meeting Proceedings Special Awards 1s

ABSTRACTS
American Society of Clinical Oncology
55th Annual Meeting
May 31-June 4, 2019
McCormick Place
Chicago, IL

SPECIAL AWARD LECTURE ABSTRACTS

David A. Karnofsky Memorial Award and Lecture
Breast cancer: 40 years of research and progress.
Gabriel N. Hortobagyi, MD, FACP, FASCO; University of Texas MD Anderson Cancer Center
More progress was made in breast cancer management over the past 40 years than ever before. Surgical
management transitioned from radical procedures to breast-conserving surgery with sentinel lymph node biopsy,
resulting in reduced morbidity. Progress in radiation therapy included well-defined indications for postoperative
radiotherapy, hypofractionated schedules, and partial breast irradiation. Effective systemic therapy (chemo-
therapy, endocrine therapy, and targeted treatments) led to improvements in control of metastatic breast cancer
and reductions in breast cancer mortality for patients with primary breast cancer. My group contributed to the
development of anthracyclines, taxanes, bisphosphonates, and multiple endocrine agents. Adjuvant and neo-
adjuvant systemic therapies became an integral part of combined modality management, today’s standard of care.
Having pioneered neoadjuvant chemotherapy, our group led to enhancing the application of limited surgical
excisions resulting in the adoption of an outstanding research tool. Improvements in supportive care significantly
reduced toxicity of treatments, enhanced quality of life, and improved treatment adherence. The incorporation of
bisphosphonates and later, denosumab, into the management and prevention of bone metastases had a major
impact on complications of advanced cancer, reduced morbidity, and later contributed to improved recurrence-
free survival in primary breast cancer. Genetic and genomic studies improved our understanding of the biology of
breast cancer, defined various subtypes, and opened the door to focused interventions, the initial steps toward
personalized therapy. Our group contributed to the definition of familial breast cancer syndromes, empowering
other investigators to identify germline mutations in several genes associated with increased risk of developing
breast (and other) cancers. Our early work in gene expression profiling led to the development of prognostic and
predictive gene panels, today an integral part of the standard of care. These changes brought a .40% reduction in
breast cancer mortality and significantly improved quality of life to women with breast cancer.

ASCO–American Cancer Society Award and Lecture

Progress and challenges in clinical cancer genetics.
Judy Garber, MD, MPH, FASCO; Dana-Farber Cancer Institute
Cancer genetics has become an intrinsic component of much of cancer care. ASCO has been a leader in the
incorporation of germline cancer genetics into oncology care, recognizing early its potential impact independent
of and integrated with somatic genetics. Recognition of technical and clinical aspects of genetics in more
common high-penetrance cancer predisposition syndromes began with breast/ovarian, Lynch syndrome, and
a series of less common conditions (MEN1, Li Fraumeni etc). Genetic testing was expensive, focused on in-
dividual genes or conditions, and often conducted in research or in highly specialized environments. Today, with
next generation sequencing and other forces, genetic testing has become multigene, across a spectrum of more
than 100 syndromes, with expanding diversity of component tumors. There is increasing recognition that delivery
of genetics services also requires innovation to drive inclusion of more diverse populations, as well as a broader
range of individuals at risk, targeting cancer patients to enable increased efforts in Cascade testing to the relatives
of carriers, and education of providers to utilize germline information most effectively in cancer surveillance and
treatment. The future directions for cancer genetics remain exciting and multidimensional. For family members,
there are novel approaches to the development and implementation of imaginative and accessible testing
strategies, of novel early detection strategies including cfDNA, of new approaches to cancer prevention/in-
terception in high risk populations, and deep investigation of the tumors occurring in carriers to continue to
2s Special Awards

identify mechanisms that drive therapy. It is also now recognized that the somatic/germline paired analyses of
tumors provides significant information over and above tumor-only approaches. Cancer genetics research is
certainly a team activity, and this work is the result of many wonderful collaborations with true leaders in the field,
which will also be highlighted in this talk.

B. J. Kennedy Award and Lecture for Scientific Excellence in Geriatric Oncology

Systemic treatment of elderly patients with metastatic stage IV non-small cell lung cancer: From nihilism to reasonable hope.
Elisabeth Quiox, MD; Hopitaux Universitaires de Strasbourg
Median age at diagnosis of non-small cell lung cancer (NSCLC) is now 70 years, meaning that the management of
this frequent disease, in patients age .5 70 years is a daily practice concern. Nihilism, as well from patients but
also doctors, was common but also false notions such as the fact that the disease progresses more slowly in elderly,
justifying therapeutic abstention. Elderly patients are under-represented in clinical trials. Even if there is no more
age limit, the median age of patients included did not increase significantly and very old patients remain seldom.
Thus, it is difficult to generalize the results obtained and dedicated clinical trials to elderly patients are mandatory.
The first one dedicated to patients age .570 years with metastatic NSCLC was published in 1999 by the ELVIS
Italian group comparing vinorelbine to best supportive care. There was a significant survival benefit in the che-
motherapy arm and thus the recommendations from scholarly societies were to treat elderly patients with vinorelbine
or gemcitabine. However, subgroup analyses of clinical trials showed that carboplatine-based doublets might be
appropriate and thus the French intergroup IFCT conducted a trial comparing single agent therapy to carboplatine 1
paclitaxel showing a highly significant survival benefit in favor of the doublet, in PS 0-2 patients with age .570
years. This finding changed the paradigm of treatment of these patients. On the other hand, maintenance therapy is
not to be recommended in this setting (IFCT trial to be published). Elderly patients with EGFR mutations benefit
from targeted therapies like their younger counterparts. Checkpoint inhibitors remain to be investigated specifically
as the concept of immunosenescence may prevent their efficacy. As a conclusion, there is still a long way to go… but
already many improvements in the management of elderly patients with metastatic NSCLC.

Pediatric Oncology Award and Lecture

The molecular pathology of pediatric cancer.
James R. Downing, MD; St. Jude Children’s Research Hospital
The treatment of pediatric cancer is one of modern medicine’s success stories. A half century of advances have led
to cure rates of more than 80 percent. Recent progress has been fueled, in large part, to increased understanding of
the genetic underpinnings that cause these diseases to arise, spread, and resist treatment. This information has
emerged from a variety of different lines of investigation, including cytogenetics and genomic efforts to identify the
underlying lesions, as well as functional studies in cellular and in vivo murine model systems to characterize the
biological consequences of the identified genetic alterations. These studies serve as the foundation for in-
dividualized treatments, which offer patients the best possible chance for a cure. Individualization of therapy has
resulted from improvements in the ability to diagnosis the specific tumor type, identify the genetic alterations that
allow accurate risk stratification, and for some tumor types, treat using drugs that are targeted to the underlying
genetic alteration. Moreover, a significantly improved understanding of pharmacogenomics allows drug doses to
be individualized, giving each child the optimal treatment available. Although the scientific community has
gained a better perspective of the landscape of somatic and germ-line mutations underlying pediatric cancer,
questions remain. Why do some patients respond to current therapeutic approaches, while other relapse and
succumb to their disease? Recent advancements in the ability to sequence pediatric cancer samples—at both
a population and single cell level—are beginning to provide answers about the diverse mechanisms that can
lead to refractory disease. This lecture will focus on how progress made in understanding the molecular
landscape of childhood cancer can be used to benefit the pediatric and adult cancer care communities as well
as to raise cure rates in countries near and far.

Gianni Bonadonna Breast Cancer Award and Lecture

The contribution of NSABP clinical trials to the management of early breast cancer: A 60-year odyssey.
Norman Wolmark, MD; Allegheny General Hospital
Since 1958, the National Surgical Adjuvant Breast and Bowel Project (NSABP) has been conducting clinical trials to
determine optimal therapy in patients with early-stage breast cancer. These trials were performed in a chronologic
 Special Awards 3s

sequential manner and provide a well-documented historical record of prevalent biologic hypotheses, prescient and
otherwise. This endeavor has established numerous standards of care, a tribute to the 120,000 women who have
participated, and is emblematic of this year’s ASCO theme, “Caring for Every Patient, Learning from Every Patient.”
Two landmark studies started in the 1970s (B04, B06) initiated the retreat from radical mastectomy and established
the propriety of breast-preserving operations. They convinced surgeons that treatment failures were not a conse-
quence of inattention to operative detail but were due to the presence of micro-metastatic disease. Thus, the decline
of the radical mastectomy and the ascent of adjuvant therapy are inextricably intertwined. Between 1973 and 2005,
NSABP consecutive adjuvant therapy trials increased the 10-year DFS from 30% to 75% in node-positive patients.
Perhaps viewed in hindsight, some of these studies might seem banal; however, the incremental gains in DFS and the
number of lives saved are incontrovertible. In 1988, the initiation of preoperative chemotherapy trials (B18, B27)
established the association of pCR and improved survival and non-pCR as a high-risk prognostic discriminate. The
findings were instrumental in the FDA considering accelerated approval based on pCR in 2012 and emphasizing the
advantages of the non-pCR setting on the basis of the results of the Katherine (B50) study in 2019. Protocols B14
(1982-1988) and B20 (1988-1993) established the value of tamoxifen in node-negative ER-positive patients and
demonstrated that chemotherapy used in combination with tamoxifen was superior to tamoxifen alone. Both of these
trials were used to validate the 21-gene RS panel (2004). This assay has significantly reduced the overall use of
adjuvant chemotherapy in the United States. Perhaps the most dramatic results in the 60-year history of the group
were obtained in NSABP B31 where trastuzumab was added to chemotherapy in N+ HER2+ patients. The initial joint
analysis from NSABP B31 and North Central N9831 disclosed a hazard ratio of .48 and an absolute 18% increase in
DFS (2005). These results were hailed as the harbinger of a completely altered approach to the treatment of breast
cancer by an enthusiastic editorialist who may yet be proven correct. Symbolically the age of therapy targeting genetic
aberrations in the adjuvant setting was initiated.

The Allen S. Lichter Visionary Leader Award and Lecture

Clinical trials and their application: Should we believe what we read?
Ian Tannock, MD, PhD, FASCO; Princess Margaret Cancer Centre and University of Toronto
Well-designed randomized controlled trials (RCTs) prevent bias in comparing treatments and can provide a sound
basis for changing practice. However, the design and reporting of many RCTs limits their relevance to clinical
practice. The aim of all medical treatments is to improve survival and/or its quality. RCTs should avoid non-
validated surrogate endpoints. Progression-free survival has been validated as a surrogate for overall survival only
rarely and its impact on quality of life depends on balance between possible delay in new symptoms and added
toxicity of a drug. Often, experimental drugs add toxicity and if patients discontinuing treatment are censored prior
to disease progression, higher dropout in the experimental arm introduces bias and essentially invalidates
randomization. RCTs should ask questions of clinical rather than commercial interest; clinical value is more
important than the p value. Development of the ASCO Value Framework and the ESMO-Magnitude of Clinical
Benefit Scales (MCBS) facilitate evaluation of clinical benefit. Many registered anticancer drugs neither satisfy
MCBS thresholds for substantial benefit nor have shown improvements in survival or its quality. Entry criteria for
trials have become more restrictive leading to an efficacy-effectiveness gap: lesser benefit and higher toxicity
when results are extrapolated from selected patients in trials with high performance status and minimal
comorbidity to routine practice. Clinical outcome studies using large databases are important to verify benefit of
new treatments in the real world. Reports of RCTs should address the primary endpoint and provide complete
accounting of toxicities, with updated information to capture chronic toxicity. Strict guidelines should ensure
freedom from bias, with recognition that declaring potential conflict-of-interest does not prevent it. The Annual
Meeting 2019 theme “Caring for every patient, learning from every patient” should obligate oncologists from every
country to lobby (and, if necessary, shame) pharma, insurers, and governments to provide access to all treatments
shown to convey true clinical benefit.

Ellen Stovall Award and Lecture

The art and science of cancer survivorship.
Ann H. Partridge, MD, MPH, FASCO; Dana-Farber Cancer Institute
As a growing number of people live for many years after their cancer diagnosis, cancer survivorship has emerged
as an important part of the cancer care continuum. The life-changing consequences of cancer for patients and
their loved ones pose unique medical and psychosocial challenges in the aftermath of initial treatment. There
are four overarching components of survivorship care and research: (1) surveillance for recurrence and
screening for second primary cancers including personalized assessment of risk; (2) identification and
management of the long-term medical and psychosocial effects of cancer and cancer treatments; (3) promotion
of improvements of modifiable health behaviors; and (4) coordination of care and communication among
providers and with survivors to ensure that their individual needs are met. Although great strides have been
4s Special Awards

made in each of these areas, there is much work to be done. We need to continue to improve our understanding
of risks faced by cancer survivors and identify how to best intervene when indicated. Ongoing research is
focusing on the identification of biomarkers of risk for secondary cancers as well as complications from
treatment. Integrative therapy, psychotherapeutic cognitive behavioral techniques, and energy balance in-
terventions have demonstrated particular promise for a wide range of symptoms and have great potential for
long-term risk reduction. At the same time, we need to optimize cancer survivorship care delivery so that all
survivors can benefit from state-of-the-art care. Harnessing the potential of electronic health record and in-
ternet-based tools can facilitate the implementation of evidence-based guidelines to inform appropriate follow-
up, while enhancing the uptake of resources to support survivors. Providers can endeavor to incorporate these
tools routinely into clinical care, enabling individual patients to communicate their symptoms and concerns
effectively, ensuring that the diverse needs of survivors are met.
 Plenary Session 5s

LBA1 Plenary Session, Sun, 1:00 PM-4:00 PM LBA2 Plenary Session, Sun, 1:00 PM-4:00 PM
Affordable Care Act (ACA) Medicaid expansion impact on racial disparities in Overall survival (OS) results of a phase III randomized trial of standard of care
time to cancer treatment. First Author: Blythe J.S. Adamson, Flatiron Health, therapy with or without enzalutamide for metastatic hormone sensitive
New York, NY prostate cancer (mHSPC): ENZAMET (ANZUP 1304), an ANZUP-led
international co-operative group trial. First Author: Christopher Sweeney,
Lank Center for Genitourinary Oncology, Dana-Farber Cancer Institute,
Boston, MA

The full, final text of this abstract will be available at The full, final text of this abstract will be available at
abstracts.asco.org at 7:30 a.m. ET on Sunday, June 2. abstracts.asco.org at 7:30 a.m. ET on Sunday, June 2.
Onsite at the Meeting, this abstract will be printed in the Onsite at the Meeting, this abstract will be printed in the
Sunday edition of ASCO Daily News. Sunday edition of ASCO Daily News.

LBA3 Plenary Session, Sun, 1:00 PM-4:00 PM LBA4 Plenary Session, Sun, 1:00 PM-4:00 PM
ANNOUNCE: A randomized, placebo (PBO)-controlled, double-blind, phase Olaparib as maintenance treatment following first-line platinum-based
(Ph) III trial of doxorubicin (dox) + olaratumab versus dox + PBO in patients chemotherapy (PBC) in patients (pts) with a germline BRCA mutation
(pts) with advanced soft tissue sarcomas (STS). First Author: William D. Tap, and metastatic pancreatic cancer (mPC): Phase III POLO trial. First Au-
Memorial Sloan Kettering Cancer Center, New York, NY thor: Hedy L. Kindler, The University of Chicago, Chicago, IL

The full, final text of this abstract will be available at The full, final text of this abstract will be available at
abstracts.asco.org at 7:30 a.m. ET on Sunday, June 2. abstracts.asco.org at 7:30 a.m. ET on Sunday, June 2.
Onsite at the Meeting, this abstract will be printed in the Onsite at the Meeting, this abstract will be printed in the
Sunday edition of ASCO Daily News. Sunday edition of ASCO Daily News.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
6s Special Clinical Science Symposia

100 Clinical Science Symposium, Sat, 8:00 AM-9:30 AM 101 Clinical Science Symposium, Sat, 8:00 AM-9:30 AM
Serum IL-6 and CRP as prognostic factors in melanoma patients receiving Biomarker analyses from JAVELIN Renal 101: Avelumab + axitinib (A+Ax)
single agent and combination checkpoint inhibition. First Author: Jeffrey S. versus sunitinib (S) in advanced renal cell carcinoma (aRCC). First Author:
Weber, Laura and Isaac Perlmutter Cancer Center, NYU Langone Medical Toni K. Choueiri, The Lank Center for Genitourinary Oncology, Dana-Farber
Center, New York, NY Cancer Institute and Brigham and Women’s Hospital, Boston, MA
Background: Acute phase reactants including C-reactive protein (CRP), and Background: The phase 3 JAVELIN Renal 101 trial in previously untreated
chronic inflammatory proteins including IL-6, which induces production of patients (pts) with aRCC demonstrated a progression-free survival (PFS)
CRP from the liver, have been associated with a poor outcome in a variety of benefit and higher objective response rate with A+Ax vs S (Motzer, ESMO
cancers. Methods: Sera from the CheckMate 064, 066 and 067 studies were 2018; LBA6_PR). Here, we report outcomes from biomarker analyses of
assessed at baseline and on treatment for levels of IL-6 and CRP by Luminex. baseline tumor samples. Methods: We correlated efficacy with the results of
Associations between IL-6 and/or CRP levels and response or survival were molecular analyses of tissue samples from all 886 pts enrolled in JAVELIN
determined. Purified endotoxin- and azide-free CRP was also tested for its Renal 101. Nephrectomy or tumor samples were characterized by immu-
immune effects in vitro using human T cells. Results: In patients receiving nohistochemistry (CD8 and PD-L1), whole-exome sequencing (WES), and
sequential nivolumab then ipilimumab in CheckMate 064 (cohort A), RNAseq. WES and RNAseq were used to examine somatic mutations and
baseline serum IL-6 was associated with response (p = 0.03); serum IL-6 at analyze relevant gene expression signatures (GES) in relation to clinical
week 12 after nivolumab alone (cohort A) or ipilimumab alone (cohort B), outcomes. GES analyses included published and de novo signatures: ef-
was also associated with response (p = 0.004 and 0.006, respectively). fector T cell (Teff), angiogenesis (angio),T cell-inflamed (Tinf), and a novel
Baseline IL-6 above the median was associated with shorter survival in immune-related signature incorporating pathway indicators for T- and NK-
cohort A (p = 0.003) and cohort B (p = 0.0001). Serum CRP above the cell activation and IFNg signaling, among others. Results: PD-L1 expression
median was associated with shorter survival in cohort A (p = 0.001). Baseline ($1% immune cells) was associated with the longest PFS in the A+Ax arm
IL-6 and CRP in cohort A were associated with one another, rho = 0.71 and and the shortest in the S arm (HR, 0.63; 95% CI, 0.49, 0.81). Significant
p , 0.0001 In the randomized CheckMate 067 study, higher baseline serum treatment arm–specific differences in PFS were observed relative to wildtype
CRP above the median was associated with shorter survival for ipilimumab, when mutations in genes such as CD1631L, PTEN, or DNMT1 were present.
nivolumab or the combination with p , 0.0001, p , 0.11, and p = 0.0034, Tumor mutational burden did not distinguish pts with respect to PFS. High-
respectively. In the randomized CheckMate 066 study, higher baseline angio GES was associated with significantly improved PFS in the S arm but
serum CRP was associated with shorter survival for nivolumab or dacarbazine did not differentiate PFS in the A+Ax arm. In the low-angio subset, A+Ax
(p = 0.131 and P , 0.0001 respectively). CRP suppressed T cell immunity improved PFS vs S. Pts with high Teff and Tinf in the A+Ax arm had longer PFS
and function, and levels above 10 micrograms/mL suppressed T cell pro- vs the S arm. In the A+Ax arm, PFS was enhanced in patients with immune
liferation (p = 0.005) and altered T cell signaling as well as calcium flux (p = GES–positive tumors vs those in the negative group (HR, 0.63; 95% CI,
0.01), suggesting that CRP affected the earliest steps in T cell signaling and 0.46, 0.86; 2-sided p = 0.004), as well as in an independent dataset
activation. Conclusions: High levels of CRP and IL-6 at baseline were as- (JAVELIN Renal 100; Choueiri, Lancet Oncol, 2018) (HR, 0.46; 95% CI,
sociated with a poor response and shorter survival after nivolumab alone, and 0.20, 1.05; 2-sided p = 0.064). Updated efficacy, including overall survival,
CRP with ipilimumab alone or the combination of both drugs. High levels of will be presented. Conclusions: These findings define molecular features
CRP were also associated with shorter overall survival in the randomized that differentiate therapy-specific outcomes in first-line aRCC and may
CheckMate 066 study of chemotherapy compared to nivolumab. CRP and inform personalized therapy strategies for pts with aRCC. Funding: Pfizer and
IL-6 are prognostic factors for checkpoint inhibition. Merck KGaA. Clinical trial information: NCT02684006.

102 Clinical Science Symposium, Sat, 8:00 AM-9:30 AM 103 Clinical Science Symposium, Sat, 8:00 AM-9:30 AM
Association of STK11/LKB1 genomic alterations with lack of benefit from Association of polybromo-associated BAF (PBAF) complex mutations with
the addition of pembrolizumab to platinum doublet chemotherapy in non- overall survival (OS) in cancer patients (pts) treated with checkpoint inhibitors
squamous non-small cell lung cancer. First Author: Ferdinandos Skoulidis, (ICIs). First Author: Sarah Abou Alaiwi, Dana–Farber Cancer Institute, Boston,
The University of Texas MD Anderson Cancer Center, Houston, TX MA
Background: Addition of pembrolizumab (P) to platinum-doublet chemo- Background: ICIs have shown benefit across several metastatic carcinomas, yet
therapy [carboplatin (or cisplatin) and pemetrexed (CP)] prolongs overall predictive biomarkers are still lacking. 20% of malignancies harbor alterations
survival and is a standard of care (SOC) for the 1st line treatment of met- in $1gene that is part of PBAF complex. With recent data suggesting an as-
astatic EGFR/ALK wild-type (wt) non-squamous non-small cell lung cancer sociation between PBRM1 mutations (mts) and outcomes in renal cell carci-
(mnsNSCLC). Despite widespread adoption of the CPP regimen, molecular noma (RCC) pts treated with ICIs (Miao, Science, 2018), we examined the
determinants of clinical benefit from the addition of P to CP remain poorly association between PBAF mts and OS in ICI-treated patients across several
defined. We previously identified genomic alterations in STK11/LKB1 as a solid cancer (ca) types. Methods: Of 6007 pts with different ca histologies and
major driver of primary resistance to PD-1/PD-L1 blockade in mnsNSCLC. targeted exome sequencing (Oncopanel) at Dana Farber Cancer institute (DFCI),
Here, we examine the impact of STK11/LKB1 alterations on clinical out- 138 pts had truncating mts in any PBAF gene (SMARCA4, PBRM1, and ARID2)
comes with CPP chemo-immunotherapy. Methods: 497 pts with mnsNSCLC or oncogenic missense mts in SMARCA4 and were treated with ICIs. 138
and tumor genomic profiling encompassing STK11/LKB1 from 17 academic histology-matched DFCI pts had none. A publicly-available cohort (2:1 histology
institutions in the US and Europe were included in this study. Clinical matched) from Memorial Sloan Kettering (MSKCC) (Samstein et al., Nature
Genetics, 2019) of 621 ca pts (PBAF mutant [MT] = 207, PBAF wild type [WT] =
outcomes were collected for two distinct patient cohorts: a) 377 pts treated
414) treated with ICIs was analyzed for association between PBAF mts and OS.
with first-line CPP (or . 1st line following FDA-approved TKIs) that were alive
OS was defined from time from ICI initiation. OS was compared by Cox re-
for 14 days thereafter and b) 120 STK11/LKB1-mt pts that received CP prior
gression between PBAF MT and PBAF WT. Hazard ratio (HR) was derived using
to regulatory approval of CPP. Results: Among 377 CPP-treated pts, STK11/ univariable and multivariable analysis (MVA) adjusted for ICI regimen (single vs
LKB1 genomic alterations (N = 102) were associated with significantly combination) and age. Results: Median (Md) follow-up for the combined cohort
shorter PFS (mPFS 4.8m vs 7.2m, HR 1.5, 95% CI 1.1 to 2.0; P = 0.0063) (n = 897) was 27 months (m). Major histologies were non-small cell lung ca
and shorter OS (mOS 10.6m vs 16.7m, HR 1.58, 95% CI 1.09 to 2.27; P = (268; 29.9%), melanoma (220; 24.5%), RCC (181; 20.2%), and bladder ca
0.0083) compared with STK11/LKB1-wt tumors (N = 275). ORR also (65; 7.2%). Results on univariable and MVA analyses from individual and
differed significantly between the two groups (32.6% vs 44.7%, P = 0.049). combined cohorts are presented below. Conclusions: PBAF mts are associated
Similar results were obtained when limiting the analysis to EGFR and ALK-wt with survival in ICI-treated ca pts. Work in progress with non-ICI treated pts will
tumors (N = 333). Importantly, in pts with STK11/LKB1-mt mnsNSCLC, determine if this is prognostic or predictive of response.
addition of pembrolizumab to CP did not improve PFS (mPFS 4.8m vs 4.3m,
Univariable MVA
HR 1.13, 95% CI 0.83 to 1.54, P = 0.75) or OS (mOS 10.6m vs 10.3m, HR PBAF MT PBAF WT (HR) (adjusted HR)
1.03, 95% CI 0.71 to 1.49, P = 0.79) compared to CP alone. Conclusions: In
DFCI: Md OS 35.9 (15.2-56.5) 18.8 (13.4-24.1) 0.69 (0.48-0.98) 0.68 (0.48-0.97)
mnsNSCLC, STK11/LKB1 alterations define a subgroup of pts with inferior (132/276 events) *
clinical outcomes with CPP and lack of benefit from the addition of pem- MSKCC: Md OS 27.0 (18.6-35.4) 18.0 (13.8-22.2) 0.76 (0.60-0.96) 0.73 (0.57-0.92)
brolizumab to CP chemotherapy. Novel therapeutic strategies are required to (339/621 events) *
Combined: Md OS 31.0 (23.5-38.5) 18.0 (14.2-21.8) 0.73 (0.60-0.89) 0.72 (0.59-0.87)
establish effective antitumor immunity in STK11/LKB1-mutant NSCLC. (471/897 events) *
* adjusted for age and single or combination ICI

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Special Clinical Science Symposia 7s

104 Clinical Science Symposium, Sun, 9:45 AM-11:15 AM 106 Clinical Science Symposium, Sun, 9:45 AM-11:15 AM
Genetic predisposition to breast cancer among African American women. The Affordable Care Act and cost-related medication non-compliance in
First Author: Julie R Palmer, Boston University, Boston, MA cancer survivors. First Author: Justin Barnes, Saint Louis University School
of Medicine, St. Louis, MO
Background: The identification of pathogenic mutations in breast cancer
susceptibility genes through clinical genetic testing leads to focused Background: Cost-related medication non-compliance (CRN), which is as-
screening and prevention strategies for women at increased risk of cancer. sociated with access-to-care barriers and poorer health outcomes, is more
However, the frequency of mutations and the risks of cancer associated with prevalent among cancer survivors than other adults. While CRN in survivors
breast cancer predisposition genes has not been established for the African has been decreasing recently, evidence for a change driven by the Affordable
American population. Methods: Germline DNA samples from African Ameri- Care Act (ACA) is limited. We aimed to quantify the impact of the ACA on CRN
can women (5,054 breast cancer cases and 4,993 age-matched unaffected in non-elderly cancer survivors using population-based data and a quasi-
controls) from 10 U.S. studies were tested for mutations in 20 established experimental design. Methods: We utilized 2011-2017 National Health In-
breast cancer predisposition genes using a QIAseq multiplex amplicon panel as terview Survey data. CRN was defined as not being able to afford medication or
part of the “CAnceR RIsk Estimates Related to Susceptibility” (CARRIERS) taking less than prescribed, skipping doses, or delaying prescription filling due
study. The frequency of mutations in each gene and associations between to cost. Linear probability models applied to difference-in-difference analyses
mutations and breast cancer risk, adjusted for study design, age, and first- were used to compare CRN changes after the ACA in non-elderly ( , 65 years)
degree family history of breast cancer, were evaluated. Results: The mean age cancer survivors relative to control groups expected to be impacted less by ACA
at diagnosis of breast cancer cases was 54.4 years and the mean age of provisions--non-elderly adults without a cancer history, elderly survivors, and
controls was 55.2 years. 18.2% of cases and 10.8% of controls reported a non-elderly survivors with high income. Results: We identified 6,176 non-
first-degree family history of breast cancer. Pathogenic mutations in any of the elderly and 8,508 elderly cancer survivors and 142,732 other non-elderly
20 breast cancer predisposition genes were identified in 7.6% of breast cancer adults. Non-elderly cancer survivors had a 6.31 (95% CI = 3.47, 9.15; p ,
cases and 2.4% of controls. In multivariable analyses, mutations in BRCA1, .001) percentage point (PP) decrease in CRN relative to non-elderly adults
BRCA2, and PALB2 were associated with high risks of breast cancer (odds ratio without a cancer history, particularly for those earning 125-249% of the
(OR) . 5.0). Mutations in CHEK2 were associated with moderate risks of federal poverty limit (FPL) (8.46 PP; 95% CI = 0.45, 16.46; p = .038) and ,
breast cancer (OR . 2.0), whereas mutations in ATM had lower clinical 125% FPL (11.8 PP; 95% CI = 4.51 to 19.1; p = .002). Relative to elderly
relevance (OR = 1.8). Mutations in BRCA1, BRCA2, PALB2, and RAD51D, but survivors, CRN decreased 6.01 PP (95% CI = 3.12, 8.90; p , .001) in non-
not CHEK2 or ATM, were associated with increased risks of estrogen receptor elderly survivors after the ACA, especially for individuals earning , 125% FPL
negative breast cancer. Conclusions: Cancer predisposition genes confer (15.7 PP; 95% CI = 7.7, 23.7; p , .001). Relative to non-elderly survivors
similar risks of breast cancer in the African American population as in non- earning . 400% FPL, those earning , 125% FPL had an 8.36 PP (95% CI =
Hispanic Whites. These studies provide important insights into the risks of 1.44, 15.3; p = .018) reduction in CRN. Conclusions: CRN decreased in non-
breast cancer associated with predisposition gene mutations in the African elderly cancer survivors after the ACA relative to both elderly survivors and
American population. adults without a history of cancer. Furthermore, reductions in CRN were
observed in low relative to high income survivors. Thus, the ACA is associated
with decreasing CRN in cancer survivors, especially those with low income, and
hence may improve health care access and affordability for this vulnerable
population.

LBA107 Clinical Science Symposium, Sun, 9:45 AM-11:15 AM LBA108 Clinical Science Symposium, Mon, 9:45 AM-11:15 AM
Insurance status and survival of multiple myeloma (MM) patients. First Impact of broadening clinical trial eligibility criteria for advanced non-small
Author: Kamal Chamoun, Adult Hematologic Malignancies and Stem Cell cell lung cancer patients: Real-world analysis. First Author: R. Donald
Transplant Program, University Hospitals Seidman Cancer Center, Cleve- Harvey, Winship Cancer Institute of Emory University, Atlanta, GA
land, OH

The full, final text of this abstract will be available at The full, final text of this abstract will be available at
abstracts.asco.org at 7:30 a.m. ET on Sunday, June 2. abstracts.asco.org at 7:30 a.m. ET on Monday, June 3.
Onsite at the Meeting, this abstract will be printed in the Onsite at the Meeting, this abstract will be printed in the
Sunday edition of ASCO Daily News. Monday edition of ASCO Daily News.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
8s Special Clinical Science Symposia

109 Clinical Science Symposium, Mon, 9:45 AM-11:15 AM 110 Clinical Science Symposium, Mon, 9:45 AM-11:15 AM
A predictive model for survival in non-small cell lung cancer (NSCLC) based Real-world outcomes of patients with advanced non-small cell lung cancer
on electronic health record (EHR) and tumor sequencing data at the (aNSCLC) and autoimmune disease (AD) receiving immune checkpoint
Department of Veterans Affairs (VA). First Author: Nathanael Fillmore, VA inhibitors (ICIs). First Author: Sean Khozin, U.S. Food and Drug Adminis-
Boston Healthcare System and Dana-Farber Cancer Institute, Boston, MA tration, Silver Spring, MD
Background: Machine learning tools based on EHR data hold promise to help Background: Anecdotal and early evidence suggest ICIs are being used in
avoid unnecessary risks associated with lung cancer and its treatment. patients with advanced malignancies and history of AD, despite such patients
Additionally, molecular genetic profiling is becoming an integral tool for being typically excluded from traditional clinical trials. We compared the
clinicians to individualize treatment for lung cancer. However, relatively few outcomes of patients with or without AD, all of whom had ICI treatment for
survival models have been built that integrate this data in individualized aNSCLC. Methods: We conducted a retrospective, observational cohort study
predictive models. Here, we combine real-world EHR and tumor sequencing using de-identified, curated data in ASCO’s CancerLinQ. Patients with Stage
data from the VA Precision Oncology Data Repository (PODR) to build ac- III or IV NSCLC who received $1 dose of an ICI and had $2 visits from Jan
curate individualized survival predictions in newly-diagnosed NSCLC pa- 2011 to Nov 2018 were included. AD status prior to ICI treatment was
tients. Methods: We identified a cohort of 356 VA patients newly diagnosed identified using ICD-9/ICD-10 codes or AD medications (including steroids).
with NSCLC for whom EHR, cancer registry, and targeted tumor sequencing Symphony claims data were linked via tokenization to build cohorts. Time to
data is available in PODR. We defined 41 features reflecting 15 baseline treatment discontinuation (TTD), time to next treatment (TTNT), real-world
clinical and demographic characteristics from the EHR and registry, such as progression-free survival (rwPFS) and overall survival (OS) were compared
age, race, stage, histology, and therapy. We also defined features reflecting across the two cohorts using the log-rank test. Cox Proportional Hazards Model
206 clinically actionable somatic variants. We selected 5 important variants was used to adjust for covariates. Adverse events (AEs) were compared using
for inclusion in the model, as well as the total number of mutations. We Chi-Square and Fisher’s Exact Test. Active AD was defined as evidence of
trained a random forests algorithm to predict 1-year survival. Precision, autoimmune disease in the year prior to starting ICIs. Results: Among 2425
recall, and area under the ROC curve (AUC) were assessed using 5-fold cross patients with aNSCLC treated with ICIs, AD was present in 22% (N=538).
validation. Results: Mean age at diagnosis was 66 years. The majority of Median OS in all patients was 12.4 months (95% CI 11.3-13.5). TTD, TTNT,
patients had late stage disease (15% stage I, 6% II, 15% III, 44% IV), and rwPFS and OS did not differ between the two cohorts (Table). There was no
59% of patients received systemic therapy. 45% died within 1 year of di- association between AD status and outcomes. There was no increased in-
agnosis, and 55% survived past 1 year. Our predictive model for 1-year cidence of AEs in the AD group; however a sub-analysis among patients with
survival achieves strong results. Cross-validated AUC is 0.79 (SD 0.08), active AD showed higher rates of select AEs including endocrine, GI and blood
precision is 0.79 (SD 0.07), recall is 0.74 (SD 0.07), suggesting that the disorders. Conclusions: This analysis demonstrates that patients with evi-
trained model combining clinical and genomic features is effective at dence of AD prior to receiving ICI have similar outcomes compared to patients
predicting 1-year survival. Conclusions: By integrating real-world EHR and with no evidence of AD. Further research is needed to better understand the
sequencing data, we built a highly accurate predictive model of 1-year impact of active AD on the risk of AEs and patient outcomes.
survival in NSCLC patients at the VA. Such a model, after ongoing validation Evidence of AD No Evidence of AD
in a larger cohort, offers the ability to make individualized predictions that (N=538) (N=1871)
could inform patient care to improve outcomes. Endpoints Median (95%CI) Median (95%CI) P-value
TTD 3.68 (3.35 - 4.6) 4.24 (3.94 - 4.60) 0.1
TTNT 12.2 (9.0 - 15.5) 12.5 (11.1 - 13.8) 0.6
rwPFS 4.24 (3.58 - 4.90) 4.17 (3.78 - 4.64) 0.7
OS 11.5 (9.63 - 13.2) 12.8 (11.51 - 14.2) 0.2

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Breast Cancer—Local/Regional/Adjuvant 9s

500 Oral Abstract Session, Mon, 9:45 AM-12:45 PM 501 Oral Abstract Session, Mon, 9:45 AM-12:45 PM
Neoadjuvant trastuzumab (H), pertuzumab (P), and chemotherapy versus Docetaxel, trastuzumab, pertuzumab versus trastuzumab emtansine as
trastuzumab emtansine (T-DM1) and P in human epidermal growth factor neoadjuvant treatment of HER2-positive breast cancer: Results from the
receptor 2 (HER2)-positive breast cancer (BC): Final outcome results from Swedish PREDIX HER2 trial identifying a new potential de-escalation standard?
the phase III KRISTINE study. First Author: Sara A. Hurvitz, David Geffen First Author: Jonas C. S. Bergh, Karolinska Institutet and University Hospital,
School of Medicine, University of California Los Angeles, Los Angeles, CA Stockholm, Sweden
Background: KRISTINE compared neoadjuvant chemotherapy plus dual HER2- Background: Neoadjuvant therapy produces high rates of pathological
blockade (HP) with T-DM1 plus P (T-DM1+P), a targeted regimen that omits standard complete response (pCR) and is the standard of care in HER2 positive breast
chemotherapy. T-DM1+P resulted in a lower pathologic complete response (pCR) rate, cancer; however, the optimal treatment regimen remains to be established.
but a more favorable safety profile. Here we present the final outcomes from KRISTINE.
Methods: KRISTINE (NCT02131064) was a randomized study of T-DM1+P versus
Methods: In this randomized phase II study patients $18 years with HER2
docetaxel, carboplatin, and H plus P (TCHP). Patients with HER2-positive stage II–III BC positive breast cancer . 20mm or verified lymph node metastases were
received 6 cycles of neoadjuvant T-DM1+P or TCHP q3w. Patients receiving T-DM1+P randomized to 6 courses of docetaxel, trastuzumab and pertuzumab (DTP,
continued adjuvant T-DM1+P; patients receiving TCHP received adjuvant HP, for 12 group A) or trastuzumab emtansine (T-DM1, group B), q 21 days. The
cycles in each arm. Patients in the T-DM1+P arm without pCR were encouraged to protocol allowed switch to the competing treatment upon lack of response or
receive standard adjuvant chemotherapy before adjuvant T-DM1+P. Secondary end- drug-related severe toxicity. Patients received postoperative epirubicin+
points, analyzed with descriptive statistics, included event-free survival (EFS; all events cyclophosphamide, trastuzumab for a total of one year and endocrine
pre- and post-surgery), invasive disease-free survival (IDFS; invasive events post-
therapy. Accrual was completed in October 2018 after randomization of 202
surgery), overall survival and safety. Results: At median follow-up of 37 months, EFS
favored TCHP (HR = 2.61 [95% CI: 1.36–4.98]), due to more locoregional progression patients, data on pCR were available for 190 at the time for this abstract
events in the T-DM1+P arm before surgery (6.7% vs 0; Table). pCR was associated with submission. Median age, 52 years (26-74), menopausal status, histological
reduced risk of an IDFS event (HR = 0.24 [95% CI: 0.09– 0.60]) regardless of treatment type and grade were well balanced between the treatment groups. 62.6% of
arm. There were 5 deaths (2.3%) in the TCHP arm and 6 (2.7%) in the T-DM1+P arm. the tumors were hormone receptor (HR) positive. Results: Primary endpoint
There were more grade $3 AEs with TCHP but a higher rate of AEs leading to treatment was pathological objective response. 190 patients completed the protocol-
discontinuation with T-DM1+P. Conclusions: EFS numerically favors TCHP due to specified preoperative treatment. pCR was achieved in 45.3% of patients,
locoregional progression events with T-DM1+P prior to surgery. T-DM1+P was associated 46.4% in patients treated with DTP and 44.1% with T-DM1 (chi-sq., p =
with fewer grade $3 AEs but increased treatment discontinuation. Clinical trial in-
formation: NCT02131064.
0.75). In HR-positive tumors, pCR was obtained in 35.3% of patients,
35.9% in group A vs. 34.6% in group B (p = 0.87); in HR-negative tumors,
TCHP T-DM1+P the overall pCR rate was 62.0%, 66.7% in group A vs. 57.9% in group B (p =
(n = 221) (n = 223) 0.45). Severe (grade 3/4) toxicity was reported at 68 occasions related to
EFS events, n (%) 13 (5.9) 31 (13.9) DTP, compared with 16 related to T-DM1, 26 vs. 3 caused by febrile
Locoregional progression before surgery 0 (0) 15 (6.7) neutropenia. Significantly better quality of life was reported by patients
Invasive disease recurrence 11 (5.0) 11 (4.9) treated with T-DM1. Conclusions: Our data on TDM-1 demonstrates similar
Non-invasive recurrence (DCIS) 0 (0) 3 (1.3) efficacy and less toxicity, in particular for patients with HER2 and HR
Death without prior EFS event 2 (0.9) 2 (0.9) positive cancers, being a potential new standard for neoadjuvant therapy.
3-yr IDFS in patients with pCR (95% CI) 97.5% 96.7% Clinical trial information: NCT02568839.
(94.7–100.0) (93.0–100.0)
3-yr IDFS in patients without pCR (95% CI) 84.2% 89.4%
(72.5–96.0) (83.1–95.6)
Grade ‡3 AEs, n (%) 148 (67.6) 71 (31.8)
AEs leading to any treatment discontinuation, n (%) 24 (11.0) 45 (20.2)

502 Oral Abstract Session, Mon, 9:45 AM-12:45 PM 503 Oral Abstract Session, Mon, 9:45 AM-12:45 PM
HER2 heterogeneity as a predictor of response to neoadjuvant T-DM1 plus Impact of clinical risk category on prognosis and prediction of chemotherapy
pertuzumab: Results from a prospective clinical trial. First Author: Otto benefit in early breast cancer (EBC) by age and the 21-gene recurrence score
Metzger Filho, Dana-Farber Cancer Institute, Boston, MA (RS) in TAILORx. First Author: Joseph A. Sparano, Montefiore Medical
Center, Albert Einstein College of Medicine, Albert Einstein Cancer Center,
Background: HER2 targeted therapy without chemotherapy may be in-
Bronx, NY
sufficient to completely eradicate a HER2+ cancer in cases of significant
intratumor HER2 heterogeneity (ITH-HER2). Methods: We conducted a Background: TAILORx established that endocrine therapy (ET) alone is non-
single-arm phase II study enrolling centrally confirmed HER2+ breast inferior to adjuvant chemotherapy (CT) plus ET in EBC and a 21-gene RS of
cancer. Pts received 6 cycles of T-DM1 plus Pertuzumab before surgery. 11-25, with some benefit if #50 years (y) with RS 16-25 (PMID:
Central ITH-HER2 was assessed on baseline ultrasound-guided core bi- 29860917). We evaluated whether clinical risk (tumor size & histologic
opsies from 2 distinct areas of each tumor (3 cores/site). ITH-HER2 was grade) provides additional prognostic information to RS, a secondary trial
defined as at least one of the six areas demonstrating either 1) HER2 objective. Methods: Clinical risk by was assessed by Adjuvant! (version 8.0)
positivity by FISH in . 5% and , 50% of tumor cells (i.e., CAP guideline) or using MINDACT criteria (PMID 27557300), defined as low clinical risk (LCR -
2) an area of tumor that tested HER2 negative. The primary objective is the tumor #3 cm and low grade, ,2 cm and intermediate grade, or #1 cm and
association between pathologic complete response (pCR) and ITH, stratified high grade) or high clinical risk (HCR -not meeting LCR criteria). Results: Of
by ER status. pCR defined as residual cancer burden (RCB) 0. Results: 164 9427 women with RS and clinical risk information, 70% were LCR and 30%
pts with centrally confirmed HER2+ tumors were enrolled from 1/2015 to HCR, with comparable distribution by age ( #50 vs. .50). The RS was 26-
1/2018. 2 pts withdrew consent. Median tumor size by imaging was 2.8 cm 100 in 9% of LCR and 27% of HCR patients, with similar distributions by
(IQR 2.1-3.8cm); 111 (69%) were ER+ and 51 (32%) ER-. 8 pts dis- age. Although LCR/HCR provided additional prognostic information in each
continued tx (6 due to disease progression, 2 due to toxicity). 49% of pts RS category for iDFS, including RS 0-10 (9-year rates 86.7% vs. 75.7% LCR
had a pCR (RCB-0), 14% RCB-I, 26% RCB-II and 11% RCB-III. Higher rates vs. HCR), 11-25 (85.4% vs. 78.9%), and 26-100 (82.0% vs. 70.4%), iDFS
of RCB-0 were seen in ER- (65%) versus ER+ (42%). ITH-HER2 was de- rates were similar irrespective of CT (no vs. yes) in the entire RS 11-25 cohort
tected in 10% (16/157) of evaluable cases. No pCR was observed among whether LCR (85.8% vs. 85.1%) or HCR (79.8% vs. 77.9%). DRFI rates
cases classified as heterogeneous (RCB-I 25%, RCB-II 25%, RCB-III 50%). were also similar irrespective of CT in the RS 11-25 cohort or . 50y group
The study met its primary endpoint by demonstrating a significant associ- whether LCR (96.0% vs. 96.1% overall; 96.5% vs. 96.0% . 50y) or HCR
ation between ITH-HER2 and pCR stratified by ER status (p , .0001). (92.3% vs. 89.9% overall; 91.7% vs. 90.7% .50y). For women #50y,
Secondary analysis also demonstrated a significant association between the absolute reduction in distant recurrence from CT with a RS of 16-
ITH-HER2 and pathologic response defined as RCB 0 or I (OR = 5.6, p = 20 (N=923) was -0.2% (standard error [SE]62.1%) for LCR vs.
0.004). Exploratory analysis revealed higher rates of RCB-0 among tumors 6.5%(SE64.9%) for HCR (vs. 1.6%[SE61.9%] overall), whereas for a RS
centrally classified as HER2 3+ (56% [66/118]) versus HER2 2+ (27% [10/ 21-25 (N=492) it was 6.4% (SE64.9%) for LCR vs. 8.6% (SE66.2%) for
37]), (OR = 3.4, p = 0.002). The association of ITH-HER2 and pCR was HCR (vs. 6.5%[SE63.7%] overall). Conclusions: Clinical risk stratification
maintained when stratifying by ER status and HER2 IHC (2+ vs. 3+), (p = provides additional prognostic information to the 21-gene RS, but not
0.002). Conclusions: ITH-HER2 assessed by routine pathology evaluation prediction of CT benefit in the overall TAILORx population or those . 50y,
is a strong predictor of pCR to a dual-HER2 targeted therapy regimen. If and facilitates more refined estimates of absolute CT benefit for
validated, ITH-HER2 may need to be considered in selection of pts for women #50y with a RS 16-25. (Funded by National Cancer Institute,
HER2-targeted regimens without chemotherapy in the curative setting. Komen Foundation, Breast Cancer Research Foundation). Clinical trial in-
Clinical trial information: NCT02326974. formation: NCT00310180.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
10s Breast Cancer—Local/Regional/Adjuvant

504 Oral Abstract Session, Mon, 9:45 AM-12:45 PM 505 Oral Abstract Session, Mon, 9:45 AM-12:45 PM
Benefit from letrozole as extended adjuvant therapy after sequential endocrine Trans-aTTom: Breast Cancer Index for prediction of endocrine benefit and
therapy: A randomized, phase III study of Gruppo Italiano Mammella (GIM). late distant recurrence (DR) in patients with HR+ breast cancer treated in the
First Author: Lucia Del Mastro, Ospedale Policlinico San Martino-Oncologia adjuvant tamoxifen—To offer more? (aTTom) trial. First Author: John Bartlett,
Medica, Genova, Italy Ontario Institute for Cancer Research, Toronto, ON, Canada
Background: The effect of extended adjuvant endocrine therapy (ET) with Background: The aTTom study is a prospective phase III trial that randomized
aromatase inhibitors (AI) after sequential ET with tamoxifen (Tam) followed 6953 HR+ women to stop or continue tamoxifen (TAM) for 5 more years after
by AI for 5 years is still controversial. We conduct a clinical trial to assess completing at least 4 years of prior TAM. Results at 9 years of median follow-up
different durations of ET of letrozole after tam. Methods: The GIM4 LEAD demonstrated fewer breast cancer recurrences (21% vs 25%; RR= 0.86 [95%
(Gruppo Italiano Mammella 4- Letrozole adjuvant therapy duration study, CI 0.77-0.96]; P = 0.006) and reduced breast cancer mortality (13% vs 15%;
ClinicalTrials.gov:NCT01064635) was a prospective, randomized, Italian HR= 0.91 [95% CI 0.80-1.04]; P = 0.18) but increased incidence of en-
multicentric trial. Post-menopausal patients (pts) with hormone receptor dometrial cancer with longer TAM use (P , 0.0001). The Breast Cancer Index
positive early breast cancer free of recurrence after 2-3 years of adjuvant (BCI) is a gene expression–based signature that stratifies patients based on the
tam, were randomized in a 1:1 ratio to receive 3-2 years (short arm, S) or 5 risk of overall (0-10y) and late (post-5y) DR and predicted the likelihood of
years (long arm, L) of letrozole. The primary study end point was disease-free benefit from extended endocrine therapy in MA.17. This Translational aTTom
survival (DFS). Results: Between August 2005 and May 2010, 2056 pts (Trans-aTTom) study is a large-scale validation of the predictive ability of BCI
were randomly assigned to receive 3-2 years (n=1030) or 5 years (n=1026) for extended endocrine therapy (EET) benefit. Methods: Patients treated in the
of letrozole. Main patients characteristics in the S and L arms were, re- aTTom trial with available primary tumor tissue were eligible for this multi-
spectively: median age 60 vs 61 years, node negative 56 vs 56%, (neo) institutional prospective-retrospective study. Primary and secondary endpoints
adjuvant chemotherapy 53.4 vs 54.1%. The median follow-up was 10 years were recurrence-free interval (RFI) and disease-free interval (DFI), respec-
(IQR range: 8.6-11.4). The 8-year DFS was 80% (95% CI:77.3-82.7) and tively. Statistical significance level for RFI was set at 0.0336 as per statistical
85% (95% CI:82.9-87.6) in the S and L arm, respectively (hazard ratio, HR plan. Kaplan-Meier and Cox proportional hazards regression analysis with
0.82; 95% CI:0.68-0.98; p=0.031). This effect did not change in a mul- time-varying coefficients were used to test the predictive activity of BCI by
tivariate Cox model that included nodal status, grading and age. No evidence HoxB13/IL17BR (H/I) status (High vs Low). Likelihood ratio test based on
of interaction between random assignment and nodal status, age and grading Cox regression was used to evaluate treatment by biomarker interaction.
was observed. Among 1960 pts evaluable for toxicity, osteoporosis was di- Results: 2637 tumors were centrally assessed for ER, PR and HER2 status
agnosed in 47 (4.8%) in S arm and 81 (8.3%) pts in L arm (chi-square=9.88; leading to 1822 HR+ patients analyzed (1018 N0, 583 N+). Initial results
p=0.002). Bone fractures occurred in 5 (0.5%) and 9 (0.9%) pts in S and L from patients with N+ disease at 12 years of median follow-up showed 287
arm, respectively (p=0.29, Fisher exact test). Conclusions: After 2-3 years of (49%) were classified as H/I-High and 296 (51%) were classified as H/I-
adjuvant tam, extended treatment with 5 years of letrozole resulted in sig- Low. H/I High patients showed a statistically significant benefit of 9.8% in
nificant improvement in DFS compared to the standard duration of 2-3 years of RFI with 10y vs 5y of TAM (HR=0.35 [95% CI 0.15-0.85]; P=0.027), whereas
letrozole. Clinical trial information: NCT01064635. H/I Low patients showed no benefit (-0.2% RFI; HR=1.07 [95% CI 0.69-
1.65]; P=0.77). A statistically significant interaction between continuous BCI
and treatment was demonstrated (P = 0.02). Conclusions: These data provide
further validation and establish level 1B evidence for BCI as a predictive
biomarker for preferential benefit from EET in HR+ breast cancer.

506 Oral Abstract Session, Mon, 9:45 AM-12:45 PM 507 Oral Abstract Session, Mon, 9:45 AM-12:45 PM
GeparOLA: A randomized phase II trial to assess the efficacy of paclitaxel and TBCRC 030: A randomized phase II study of preoperative cisplatin versus
olaparib in comparison to paclitaxel/carboplatin followed by epirubicin/ paclitaxel in TNBC—Evaluating the homologous recombination deficiency
cyclophosphamide as neoadjuvant chemotherapy in patients (pts) with (HRD) biomarker. First Author: Erica L. Mayer, Dana-Farber Cancer Institute,
HER2-negative early breast cancer (BC) and homologous recombination Boston, MA
deficiency (HRD). First Author: Peter A. Fasching, University Hospital Background: Cisplatin (C) and paclitaxel (T) have activity in TNBC, however predictive
Erlangen, Comprehensive Cancer Center Erlangen-EMN, Erlangen, Germany biomarkers are lacking. The HRD assay detects impaired dsDNA break repair and may
Background: The efficacy and toxicity of olaparib in early BC pts with ho- identify BRCA1/2-proficient tumors for treatment with DNA targeting therapies.
mologous DNA repair deficiency (here defined as HRD score high tumors +/- TBCRC 030 was designed to determine the association between HRD and response to
preoperative chemotherapy (CT) in TNBC. Methods: This phase II study randomized
germline (g) or tumor (t) BRCA mutation) is not well described. GeparOLA patients (pts) with BRCA1/2-proficient/unknown stage I-III TNBC to 12 weeks (wks) of
investigates olaparib in HER2 negative early BC with HRD. Methods: GeparOLA preoperative C or T, followed by surgery. HRD was performed on baseline tissue, with
(NCT02789332) randomized 102 pts to either paclitaxel 80 mg/m² weekly positive scores . 33. Non-responders at 12 wks could crossover to alternative CT. The
(Pw) plus olaparib 100 mg twice daily for 12 weeks (PwO n = 65) or Pw plus co-primary objectives were to detect a positive association of HRD with pathologic
carboplatin (Cb) AUC2 weekly for 12 weeks (PwCb n = 37), both followed by EC. response (RCB 0-1) vs not (RCB 2-3) to C and a negative association to T. Target accrual
Randomization was stratified by hormone receptor-status (HR+ vs HR-) of 160 pts was planned to yield 140 evaluable specimens for HRD, providing 90% power
and age ( , 40 vs $40 years). Pts with untreated primary cT2 - cT4a-d or for the primary objectives. Analyses used logistic models and likelihood ratio tests with
cT1c with either cN+ or pNSLN+ or triple negative or Ki-67 . 20% were one-sided Type I errors of alpha = 0.05. Results: 140 pts initiated treatment, (72 Arm C,
included, with either g/tBRCA mutation and/or high HRD score. The 68 Arm T; 81% T1-2, 62% node negative); 138 were evaluable for response at 12 wks.
Post-enrollment testing showed 8 pts (5.8%) with germline DNA-repair pathway mu-
primary endpoint is pathological complete response (pCR; ypT0/is ypN0).
tations. HRD results were available for 95 pts (68.8%, 23 inadequate tissue, 22
A one group x2-test was planned to exclude the pCR rate of #55% in pending); 68 (71.6%) were HRD positive: 38 in Arm C, 30 in Arm T. In response-
PwO→EC arm. Secondary endpoints are other pCR definitions, breast evaluable pts, 87 (63.0%) had surgery at 12 wks, and 51 (37.0%) crossed over.
conservation rate, clinical and imaging response, tolerability and safety. Response outcomes are shown in the Table. No association was seen between HRD
Results: A total of 107 pts were randomized between 9/2016 and 7/2018; score and RCB response to either neoadjuvant C (OR 2.78, [CI 0.61, 17.74]) or T (OR
106 started treatment. Median age was 47.0 years [range 25.0-71.0]; 36.2% 0.98, CI [0.20, 5.06]). There was no evidence of an interaction between HRD and CT
of pts had cT1, 61.0% cT2, 2.9% cT3, and 31.8% cN-positive tumors; G3: arms. Similarly, no association was observed between HRD score and pCR to either C
86.8%; Ki-67 . 20%: 89.6%; TNBC 72.6%; confirmed g/tBRCA 1/2 mu- (OR 1.47, CI [0.40, 5.59]) or T (OR 0.61, CI [0.14, 2.52]). There were no new safety
tation: 60.4%. pCR rate with PwO was 55.1% (90%CI 44.5%-65.3%) vs signals. Conclusions: In this mostly BRCA1/2 proficient TNBC cohort, 12 wks of pre-
PwCb 48.6% (90%CI 34.3%-63.2%). Analysis for the stratified subgroups operative C or T led to a similar response rate of about 40%; baseline HRD was not
predictive of response to preoperative CT, defined either by RCB 0-1 or pCR. Further data
showed higher pCR rates with PwO in the cohorts of pts , 40 years and HR+
will be presented. Correlative analyses of research tissues for markers predictive of response
pts Clinical trial information: NCT02789332. Conclusions: GeparOla could to specific CT in TNBC is ongoing. Clinical trial information: NCT01982448.
not exclude a pCR rate of #55% in the PwO arm. Subgroup analysis is hy-
pothesis generating and need further confirmation. Outcome Arm C N = 71 Arm T N = 67 Total N = 138
Olaparib+Paclitaxel Carboplatin+Paclitaxel RCB 0-1 27 (38.0%) 29 (43.3%) 56 (40.6%)
pCR rate (90%CI) pCR rate (90%CI) -HRD + 15 (21.1%) 13 (19.4%) 28 (20.3%)
-HRD - 3 (4.2%) 5 (7.5%) 8 (5.8%)
HR+ patients (n = 29) 52.6% (32.0%, 72.6%) 20.0% (3.7%, 50.7%) RCB 2-3/crossover 44 (54.3%) 38 (46.3%) 82 (59.4%)
HR- patients (n = 77) 56.0% (43.4%, 68.0%) 59.3% (41.7%, 75.2%) -HRD + 23 (32.0%) 16 (23.9%) 39 (28.3%)
Patients age < 40 (n = 32) 76.2% (56.3%, 90.1%) 45.5% (20.0%, 72.9%) -HRD - 13 (18.3%) 6 (9.0%) 19 (13.8%)
Patients age ‡ 40 (n = 74) 45.8% (33.4%, 58.6%) 50.0% (32.7%, 67.3%) pCR 15% 13%

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Breast Cancer—Local/Regional/Adjuvant 11s

508 Oral Abstract Session, Mon, 9:45 AM-12:45 PM 509 Poster Discussion Session; Displayed in Poster Session (Board #1),
Patient-reported outcomes (PROs) in NRG oncology/NSABP B-39/RTOG Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
0413: A randomized phase III study of conventional whole breast irradiation Sun, 4:30 PM-6:00 PM
(WBI) versus partial breast irradiation (PBI) in stage 0, I, or II breast cancer. Exome analysis of oncogenic pathways and tumor mutational burden (TMB)
First Author: Patricia A. Ganz, NRG Oncology, and The UCLA Jonsson in triple-negative breast cancer (TNBC): Results of the translational biomarker
Comprehensive Cancer Center at UCLA, Los Angeles, CA program of the neoadjuvant double-blind placebo controlled GeparNuevo trial.
First Author: Sibylle Loibl, German Breast Group (GBG) and Centre for Hae-
Background: PBI is an alternative to WBI, with potentially greater therapy (tx)
matology and Oncology Bethanien, Frankfurt, Neu-Isenburg, Germany
compliance, and better integration with chemotherapy (CTX). NSABP B-39/
RTOG 0413 clinical outcome results from 2018 did not show equivalence of Background: GeparNuevo (G9) showed a numerical increase in pCR rate of
PBI to WBI in local tumor control; PBI was statistically inferior, but with 53% vs 44%; p = 0.287 compared to placebo in TNBC with the addition of the
clinically small differences. PBI may be an acceptable alternative to WBI for anti-PD-L1 antibody durvalumab (D) to a neoadjuvant anthracycline-taxane
some women. Understanding cosmesis and quality of life (QOL) treatment containing chemotherapy (Loibl S et al. ASCO 2018). Somatic mutations in
outcomes is important. Methods: B-39/0413 included a prospective QOL malignant cells manifest over the evolutionary history of a tumor. Reports in
substudy with PRO evaluation of breast cancer treatment outcomes (cosmesis, selected tumor types suggest that TMB may predict clinical outcomes on
function, pain) and fatigue using BCTOS and SF-36 vitality scales. Secondary immune-checkpoint inhibitors (ICI). The clinical relevance of TMB in breast
QOL parameters included treatment related symptoms, perceived convenience cancer has not been studied widely. Here, we investigated the hypothesis
of care, and the BPI pain scale. The study sample was stratified by CTX or not, that TMB predicts response to ICI. Methods: Whole exome sequencing was
as CTX is given before WBI but after PBI. PRO assessments occurred before conducted on patient-matched fresh-frozen core biopsies and blood samples
randomization, the last day of adjuvant tx [CTX or radiation], 4 wks later, and 6, with Illumina (n = 149/174). SNVs and indels were called with Mutect and
12, 24, and 36 mo later. Primary aims included comparisons of change in pureCN was used for copy number calls. Mutational signatures were iden-
fatigue from baseline to end of tx and equivalency of change in cosmesis from tified as described by Alexandrov et al. (Cell Rep. 3, 2013). Data from G9
baseline to 36 mo for PBI v WBI. Separate analyses were done for CTX and non- were compared to The Cancer Genome Atlas (TCGA) TNBC cohort. Results: A
CTX pts, controlling for axillary dissection. Each comparison used a=0.0125. similar genomic landscape was observed between G9 and TCGA with pri-
Planned sample size was 964. Results: From 3-23-05 to 5-27-09, 975 pts mary mutations in TP53 (69%), c-MYC (26%), BRCA1 (13%), BRCA2 (6%),
were enrolled in the PRO study; 950 had follow-up data. 504 did not receive PIK3CA (11%) and PTEN (11%). Median TMB was 1.52 mut/MB. TMB in
CTX and 446 received CTX. In non-CTX pts, PBI had less fatigue (p=0.011) G9 was slightly lower than TCGA TNBC. TMB correlated with older age,
and did not meet criteria for cosmesis equivalence (97.5% CI, -0.02 to 0.22; higher mutation rates in BRCA2, ARID1A, and TP53, and higher burden in
Δ=0.20). In CTX pts, PBI had worse fatigue (p=0.011) and equivalent cosmesis variant signatures such as DDR, HRD, GFRs, APOBEC and Alexandrov’s
to WBI (97.5% CI, -0.09 to 0.21; Δ=0.24). In both groups, PBI pts reported less signatures 3 and 6. Continuous TMB predicted pCR in univariate (OR =
pain at end of tx. In non-CTX pts, PBI had more pain at 36 mo but in CTX pts, 1.62, 95%-CI: 1.20 - 2.20, p = 0.0018) and multivariate (OR = 2.06,
there was no difference. Convenience of care and treatment related symptom 95%-CI: 1.33 - 3.20, p = 0.0012) logistic regression models, but did not
outcomes will be presented. Conclusions: In non-CTX pts, PBI is more con- predict a D effect. After dichotomisation of TMB at the top tertile, 50
venient with less fatigue and slightly poorer cosmesis at 36 mo. Cosmesis patients had high TMB and 29 of these (58%) achieved a pCR, while 99
was equivalent at 36 mo in CTX pts. Support: U10CA180868, -180822, had low TMB and only 38 of these (38%) had a pCR (p = 0.0242).
UG1CA189867. Clinical trial information: NCT00103181. Conclusions: Results show that TMB may predict pCR in primary TNBC.
The trial was financially supported by Astra Zeneca and Celgene. Clinical
trial information: NCT02685059.

510 Poster Discussion Session; Displayed in Poster Session (Board #2), 511 Poster Discussion Session; Displayed in Poster Session (Board #3),
Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Sun, 4:30 PM-6:00 PM Sun, 4:30 PM-6:00 PM
Immunophenotype and proliferation to predict for response to neoadjuvant NSABP B-41, a randomized neoadjuvant trial: Genes and signatures associated
chemotherapy in TNBC: Results from BrighTNess phase III study. First with pathologic complete response (pCR). First Author: Sandra M. Swain,
Author: Otto Metzger Filho, Dana-Farber Cancer Institute, Boston, MA NSABP, and Georgetown Lombardi Comprehensive Cancer Center, Georgetown
University Medical Center, Washington, DC
Background: In TNBC, the interplay between immunophenotype, tumor
proliferation (prolif) and achievement of pathologic complete response (pCR) Background: NSABP B-41 randomly assigned 529 patients (pts) with HER2
with neoadjuvant chemotherapy (NAC) remains unknown. Methods: RNA positive breast cancer to neoadjuvant trastuzumab (T), lapatinib (L), or com-
seq was performed on pre-tx research biopsies of stage II/III TNBC enrolled in bination (T+L), with weekly paclitaxel following doxorubicin and cyclophos-
BrighTNess. NAC regimens included paclitaxel alone or with carboplatin phamide. No significant differences in pCR were found, but overall survival was
(Cb) or Cb plus veliparib, followed by AC. Computational analysis included significantly increased for pCR. Methods: RNA was extracted from FFPE tumor
subtyping (i.e. PAM50, Pietenpol), prolif (PAM50) and GeparSixto immune specimens, run on the NanoString Breast Cancer 360 Plus panel. Gene counts
signature (GSIS). Cb-containing arms were combined due to similar pCR. were normalized to include housekeeping genes, 33 biological signatures
Results: High quality RNA seq data was obtained from 482 of 634 pts. PAM50 from 776 genes across 23 pathways and transformed into logarithm scale
classified 80.1% of tumors as basal-like. TNBC subtypes were mostly BL1 or with base two. Univariate logistic regression was used to screen candidate
BL2 (23.3%), IM (22.4%) or M/MSL (31.7%); 6% were LAR. pCR was higher genes and signatures that are prognostic of pCR, with false discovery rate
for basal vs non-basal tumors (52.3% vs 35.4%, p = 0.003). IM had the controlled at 0.1. Multivariable logistic regression with lasso regularization
highest pCR rate (64.2%, 95% CI 59.9%,68.5%). Basal-like was not a was used for model selection. Results: 194 core biopsy samples were
significant predictor for Cb benefit (p-interaction = 0.8). Prolif (OR = 0.30 p , available; 69 treated with T, 64 with L and 61 with T+L. 20 prognostic
0.001) and GSIS (OR 0.68 p , 0.001) were significantly correlated with pCR genes are selected for trastuzumab-based regimens (TBR), including the
but did not correlate with each other (Pearson’s r2 = 0.027). In multivariate epithelial-mesenchymal transition (HEMK1, GRB7, ERBB2, TMPRSS4),
analysis, prolif (HR = 0.36 95% CI, 0.21-0.61 p = 0.0002) and GSIS (HR = adhesion and migration (ITGB6, COL27A1, NRCAM), JAK-STAT (SOCS2),
0.62 95% CI, 0.49-0.79 p , 0.0001) increased the ability to predict pCR Hedgehog (LRP2), ER signaling (ELOVL2, MAPT), DNA damage and repair
beyond standard clinico-pathologic variables (likelihood ratio = 14.9, p = (NPEPPS, PRKDC), MAPK (DUSP6, PRKCB), Apoptosis (BCL2), pro-
0.0001115). Among all pts, those above the median for both prolif. and GSIS liferation (TFDP1). ERBB2 expression are associated with pCR in patients
had the highest pCR (67%; 84/125) while those below the median for both on TBR (OR = 1.73), but not for patients on L (interaction p = 0.01). HER2-
had the lowest pCR rate (34%; 42/125). Tumors with higher inferred CD8+ Enriched correlation (p , 0.001), ESR1 (OR = 0.78, 95% CI = 0.69-0.88,
T-cell infiltration demonstrated greater benefit from Cb using either TIMER p , 0.001), PD1 (OR = 1.68, 95% CI = 1.12-2.52, p = 0.01) and Tumor
(HR = 0.83 [0.73-0.95]) or CIBERSORT (HR = 0.83 [0.76-0.91]). Tumors Inflammation Score (OR = 1.58, 95% CI = 1.18-2.11, p = 0.002) are
with higher inferred total macrophages, particularly immune suppressive associated with pCR in TBR. No genes or signatures were found to be
M2 macrophages had a higher pCR rate on the non-Cb arm (AC-T) using predictive of treatment benefit from L added to T. Conclusions: BC360
CIBERSORT (HR = 1.27 [1.07,1.50]). Conclusions: Immunophenotype highlighted tumor progression and signaling genes prognostic for TBR.
and proliferation are independent predictors of pCR to standard NAC regimens HER2-Enriched correlation, ERBB2 and PD1 expression, and immune
in TNBC. PAM50 is not a significant predictor of Cb benefit. Exploratory activation signatures were associated with pCR in TBR and may provide
findings suggest that tumor infiltrating immunophenotype (i.e. CD8 T cells personalized treatment guidance.
and macrophages) may predict response to specific NAC regimens in
TNBC. Clinical trial information: NCT02032277.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
12s Breast Cancer—Local/Regional/Adjuvant

512 Poster Discussion Session; Displayed in Poster Session (Board #4), 513 Poster Discussion Session; Displayed in Poster Session (Board #5),
Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Sun, 4:30 PM-6:00 PM Sun, 4:30 PM-6:00 PM
Differential impact of endocrine therapy (ET) and chemotherapy (CT) on Patient-reported outcomes (PROs) from KATHERINE: A phase III study of
quality of life (QoL) of 4,262 breast cancer (BC) survivors: A prospective adjuvant trastuzumab emtansine (T-DM1) versus trastuzumab (H) in patients
patient-reported outcomes (PRO) analysis. First Author: Arlindo R. Ferreira, (pts) with residual invasive disease after neoadjuvant therapy for HER2-positive
Institut Gustave Roussy, Villejuif, France breast cancer. First Author: Andreas Schneeweiss, National Center for Tumor
Diseases, Heidelberg University Hospital and German Cancer Research Center,
Background: We recently witnessed a trend to de-escalate CT and escalate ET in
adjuvant BC treatment (tx). However, there has been limited prior research in- Heidelberg, Germany
vestigating the differential impact on QoL of tx classes. We aimed to test the impact Background: The phase 3 KATHERINE (NCT01772472) study, met its
of CT and ET on QoL PROs 2 yrs after diagnosis (dx). Methods: CANTO primary endpoint by demonstrating significantly improved invasive disease-
(NCT01993498) is a multicenter prospective longitudinal study of stage I-III BC free survival with adjuvant T-DM1 compared to H in pts with residual invasive
pts that characterizes long-term toxicities of BC tx. For this analysis we included
disease after neoadjuvant chemotherapy plus HER2-targeted therapy. PROs
4262 pts recruited from 2012-14. QoL was extensively evaluated using the EORTC
QLQ C30 and BR23. Linear regression modeling was performed, adjusting for
are reported here. Methods: Eligible pts had HER2-positive early breast
demographic and clinical factors, with use of CT and/or ET as independent vari- cancer, received taxane- and H-containing neoadjuvant therapy (with/
ables. Analyses were stratified by menopausal status due to different tx patterns without anthracyclines) followed by surgery, and had residual invasive dis-
and sequelae of CT. Results: Median age at dx was 56 yrs, 63% of pts were post ease in the breast and/or axillary nodes. Pts were randomized to 14 cycles of
(PostM) and 37% premenopausal (PreM), 80% had Charlson score 0, 91% stage I- adjuvant T-DM1 (3.6 mg/kg IV q3w) or H (6 mg/kg IV q3w) and adjuvant
II. 26% received mastectomy, 52% CT (preM 68%, postM 44%; 86% anthra- endocrine and radiation therapy per standard of care. The EORTC Quality of
cycline+taxane) and 82% ET (preM 89% tamoxifen; postM 88% aromatase in- Life Questionnaire–Core 30 (QLQ-C30) and QLQ–Breast Cancer (QLQ-BR23)
hibitor). 32% preM pts had menses 1 year after ET initiation. Overall, QoL were completed at screening, at day 1 of cycles 5 and 11, within 30 days after
deteriorates 2 yrs after dx. ET negatively impacts more QoL domains than CT at 2 study drug completion, and at 6 and 12 months’ follow-up. Results: Of 1,486
yrs. Also, young age, smoking, income, aggressive local tx and physiological pts randomized (T-DM1, n = 743; H, n = 743), 612 (82%) and 640 (86%),
distress are often associated with low QoL. In the stratified analyses, in postM pts, respectively, had valid baseline and $1 post-baseline PRO assessments.
mostly ET (not CT) is associated with deteriorated QoL. In contrast, in preM pts, mostly During the study, pts in both arms had similar mean scores on the QLQ-C30
CT (not ET) is associated with deteriorated QoL. Table shows eg of associations.
and QLQ-BR23 function and symptom scales. There was no clinically
Conclusions: In a large prospective cohort of BC survivors, detrimental QoL 2 yrs after
meaningful change ($10 points) from baseline in the mean scores in either
dx is mostly associated with ET; however, negative effects of CT persist on preM. This
differential effect on QOL should be considered when choosing optimal adjuvant arm, including on symptoms similar to AEs seen with T-DM1 (eg, fatigue).
therapy and appropriate selection of pts for ET escalation should be a research priority. While more pts in the T-DM1 arm reported clinically meaningful deterioration
in role functioning (49% vs 41%), appetite loss (38% vs 28%), constipation
Adjusted linear coefficients (95% CI). (47% vs 38%), fatigue (66% vs 61%), nausea/vomiting (39% vs. 30%), and
PostM PreM systemic therapy side effects (49% vs 36%) at $1 assessment, the proportion
ET vs no CT vs no reporting clinically meaningful change in functioning was similar between
Functions (negative worse) arms at any given assessment. Conclusions: Mean scores showed only small
Physical -2.3 (-3.8, -0.9) -3.5 (-5.2, -1.7) deterioration from baseline in patient-reported treatment-related symptoms
Role -4.1 (-6.5, -1.7) -4.5 (-7.6, -1.4) in both study arms. While more pts in the T-DM1 arm reported deterioration
Sexual -3.4 (-5.5, -1.3) -4.2 (-7.6, -0.8) at some point in several symptoms, baseline global health status and
Symptoms (positive worse) functioning were generally maintained in both arms over the treatment
Pain 5.5 (2.8, 8.3) 5.8 (2.2, 9.3)
course. Clinical trial information: NCT01772472.

514 Poster Discussion Session; Displayed in Poster Session (Board #6), 515 Poster Discussion Session; Displayed in Poster Session (Board #7),
Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Sun, 4:30 PM-6:00 PM Sun, 4:30 PM-6:00 PM
Clinical validity of CTS5 for estimating risk of late recurrence in unselected, Event-free survival analysis of the prospectively randomized phase III ETNA
non-trial patients with early ER+ breast cancer. First Author: Juliet Richman, study with neoadjuvant nab-paclitaxel (nab-P) versus paclitaxel (P) followed
Royal Marsden Hospital, London, United Kingdom by anthracycline regimens in women with HER2-negative high-risk breast
cancer. First Author: Luca Gianni, IRCCS San Raffaele Hospital, Milan, Italy
Background: The Clinical Treatment Score at 5 years (CTS5) is a prognostic tool
using clinicopathological data to estimate distant recurrence (DR) risk after 5 Background: The ETNA study showed that substituting P with nab-P did not
years of endocrine therapy for postmenopausal women with estrogen receptor significantly increase the overall rate of pathological complete response
positive (ER+) breast cancer. It was developed and validated in the ATAC and BIG (pCR) (P 18.6%, nab-P 22.5%, p = 0.19). The multivariate analysis revealed
1-98 trials. Methods: The validity of CTS5 was tested in a retrospective cohort that tumor subtype (triple negative vs luminal B-like) was the most signif-
of unselected, non-trial patients diagnosed with early ER+ breast cancer at the icant factor (OR 4.85) influencing treatment outcome (Gianni L et al, JAMA
Royal Marsden Hospital from 2000-2007 who were alive and distant recurrence- Oncol 2018). Methods: This multicenter open label study (NCT01822314)
free at 5 years. The primary endpoint was time to late DR (5-10 years). Cox in collaboration with GEICAM and BCRC-WA randomized 695 patients with
regression models were used to determine the prognostic value of CTS5 and to centrally-confirmed HER2-negative breast cancer to nab-P 125 mg/m2 (346
produce Kaplan-Meier curves with associated 10-year DR risks (%). Results: A patients) or P 90 mg/m2 (349 patients). The two drugs were given on weeks
total of 2428 women were included with a median follow-up of 9.34 years from 1, 2 and 3 followed by 1-week rest for 4 cycles before 4 cycles of an anthracycline
diagnosis. The CTS5 was significantly prognostic for late DR in post- and pre- regimen as per investigator choice. The primary endpoint was pCR (absence of
menopausal women (Table). Risk stratification by CTS5 of the postmenopausal invasive cells in breast and nodes). A secondary endpoint is event-free survival
cohort was comparable with the development cohort. 42.1% of postmenopausal (EFS) defined as the time from randomization to the first date of disease pro-
women were categorised into the low risk group with a late DR risk of 4.9% and gression while on primary therapy or disease recurrence (local, regional, distant,
these women had significantly lower risk of late DR compared to those in the invasive contralateral breast cancer) after surgery or death due to any cause.
intermediate or high-risk groups (Table). Amongst the premenopausal cohort, Results: The ITT analysis of the secondary endpoint EFS at 5 years is reported
41% were categorised as low risk with a late DR risk of 4.9%. The prognostic below: Clinical trial information: NCT01822314. Overall 5-year survival was
effect of CTS5 was seen for chemotherapy treated (HR=2.26, 95% CI (1.68-
84.8% after P and 87.3% for nab-P. No serious adverse events were docu-
3.03)) and untreated patients (HR=1.93, 95% (1.32-2.82)). Conclusions: CTS5
mented during the follow-up. Conclusions: The improved 5-year EFS after nab-
demonstrated clinical validity for predicting late DR within a large cohort of un-
P failed to reach statistical significance (unadjusted P = 0.245). In the analysis
selected, non-trial patients that included premenopausal women. The low risk
by subgroup the numerical improvement was almost exclusively observed in
cohort identified by the CTS5 represents a group of women whose risk of late DR is
so low as to not warrant extended endocrine therapy to ten years.
luminal B and not in TN tumors. So far the data do not support substitution of P
with nab-P in the schedule and doses adopted in the ETNA trial. Additional
HR for late distant recurrence (95% CI) P-value analyses will be based on ongoing molecular studies.
Postmenopausal CTS5 (continuous) 1.95 (1.59-2.39) ,0.0001
(N=1662, DR=107) P% nab-P % HR
CTS5 low Reference # patients (95% CI) (95% CI) (95% CI)
CTS5 intermediate 2.28 (1.32-3.93) 0.003
CTS5 high 3.81 (2.27-6.41) ,0.0001
Overall 695 68.1 75.6 0.83
Premenopausal CTS5 (continuous) 1.72 (1.23-2.40) 0.001 (60.5-74.6) (69.2-80.9) (0.60-1.14)
(N=776, DR=51) Luminal B 476 72.1 80.3 0.89
CTS5 low Reference (62.7-79.5) (73.5-85.5) (0.58-1.37)
CTS5 intermediate 1.69 (0.84-3.51) 0.16 Triple negative 219 61.0 63.5 0.76
CTS5 high 2.63 (1.29-5.34 0.008 (48.3-71.5) (47.1-76.0) (0.47-1.23)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Breast Cancer—Local/Regional/Adjuvant 13s

516 Poster Discussion Session; Displayed in Poster Session (Board #8), 517 Poster Discussion Session; Displayed in Poster Session (Board #9),
Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Sun, 4:30 PM-6:00 PM Sun, 4:30 PM-6:00 PM
Results of randomized phase II trial of neoadjuvant carboplatin plus docetaxel Avoiding peg-filgrastim (Peg-F) prophylaxis during the paclitaxel (T) portion
or carboplatin plus paclitaxel followed by AC in stage I-III triple-negative breast of the dose-dense (DD) doxorubicin-cyclophosphamide (AC)-T regimen: A
cancer (NCT02413320). First Author: Priyanka Sharma, University of Kansas prospective study. First Author: Romualdo Barroso-Sousa, Hospital Sı́rio-Libanês,
Medical Center, Kansas City, KS Brası́lia, Brazil
Background: Addition of neoadjuvant carboplatin (Cb) to paclitaxel (T) fol- Background: Use of growth factors (GF) adds considerable expense and
lowed by doxorubicin + cyclophosphamide (AC) improves pathologic complete some toxicity to adjuvant breast cancer chemotherapy. We tested the fea-
response (pCR) rate compared to T/AC in TNBC. An anthracycline-free regimen sibility and safety of omitting routine GF use during the T portion of DD AC-T.
of Cb plus docetaxel (D) also yields high pCR rates in TNBC, and patients Methods: This is a prospective, single-arm study in which patients (pts) who
achieving pCR with this regimen demonstrate excellent 3-year outcomes completed 4 cycles of DD-AC proceeded to DD-T 175 mg/m2 every two
without adjuvant anthracycline. This study was designed to compare the ef- weeks (wks) without routine GF (NCT02698891). Key inclusion: age# 65,
ficacy of neoadjuvant regimens CbT→AC and CbD in TNBC. Methods: In this ECOG PS#1, absolute neutrophil count (ANC) $1500/mm3, and no febrile
multicenter study, eligible patients with stage I–III TNBC were randomized neutropenia (FN) during DD-AC. Criteria to treat for T included ANC $1000/
(1:1) to either paclitaxel 80 mg/m2 every week X 12 + carboplatin (AUC 6) every mm3. Peg-F was given only if pts had FN in a prior cycle, or at investigator
3 weeks X 4, followed by doxorubicin 60 mg/m2 + cyclophosphamide 600 mg/ discretion if infection or treatment delay . 1 wk. Once Peg-F was given, pts
m2 every 2 weeks X 4 (CbT→AC, Arm A), or to carboplatin (AUC 6) + docetaxel received it in all future cycles. The primary endpoint was the rate of T
(75 mg/m2) every 21 days X 6 cycles (CbD, Arm B). The primary endpoint was completion # 7 wks from cycle 1 day 1 (C1D1) to C4D1. Secondary endpoints
pCR (no evidence of invasive tumor in the breast and axilla). The two regimens included total use of Peg-F, rates of hematologic toxicity and FN, reasons for
were compared for differences in pCR, residual cancer burden (RCB), treatment dose modification or hold. If $85% of pts completed T on time, the regimen
delivery, and toxicity. Results: Between 2015 and 2018, 100 patients were would be considered feasible. If the true on-time completion rate is 75%, the
randomized; 48 to Arm A and 52 to Arm B. Median age was 52 years, median chance the regimen would be declared infeasible is 91%, and if it is 85% the
tumor size was 2.7 cm, 30% were lymph node-positive and 17% carried a chance that the regimen is falsely declared infeasible is 10% (power =
BRCA1/2 mutation. Baseline demographic and tumor characteristics were 0.899). $100/125 pts had to complete T on time for the regimen to be
balanced between two arms. pCR was 55% (95%CI: 41%-59%) in Arm A and deemed successful. Results: Among 127 pts enrolled, 125 received $1
52% (95%CI: 39%-65%) in Arm B, p=0.84. RCB 0+1 rate was 67% in both dose of protocol therapy and are included in the analysis. Median age at
arms. Grade 3/4 adverse events were more common in Arm A compared to Arm registration was 46 (range 21-65). Median C1D1 ANC was 7500/mm3
B (73% vs 21%, p , 0.0001), with most notable differences in rates of G3/4 (range 1500-20500). 112 (90%) (95% CI 83-94%) pts completed DD-
neutropenia (Arm A = 60%, Arm B = 8%, p = 0.0001), febrile neutropenia T # 7 wks, and 3 (2%) completed within . 7 wks (2 due to neutropenia); 10
(Arm A = 18%, Arm B = 0%, p = 0.0001), and G3/4 anemia (Arm A = 46%, (8%) did not complete all cycles of T. Omission of Peg-F was not causally
Arm B = 4%, p = 0.0001). 81% of Arm A patients completed all 4 doses of AC related to non-completion of T in any pts. The most common reasons for dose
and 4 doses of Cb, and 71% completed . 9 doses of T. 90% of Arm B patients reduction or delays were non-hematologic. One pt had FN but was able to
completed all 6 doses of CbD (p = 0.034). Conclusions: The non-anthracycline complete T on time. Eight (6.4%) pts received Peg-F during the trial.
platinum regimen of CbD yields pCR and RCB 0+1 rates similar to 4-drug Conclusions: Omission of routine GF use during DD-T according to a pre-
regimen of CbTàAC, but with a more favorable toxicity profile and higher specified algorithm appears safe, feasible, and was associated with a 95.7%
treatment completion rate. The CbD regimen should be further explored as a reduction in use of Peg-F, relative to the current standard of care. Additional
way to de-escalate therapy in TNBC. Clinical trial information: NCT02413320. analyses including cost implications are ongoing. Clinical trial information:
NCT02698891.

518 Poster Discussion Session; Displayed in Poster Session (Board #10), 519 Poster Discussion Session; Displayed in Poster Session (Board #11),
Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Sun, 4:30 PM-6:00 PM Sun, 4:30 PM-6:00 PM
Association between tumor biology and occult lymph node metastases before Adjuvant endocrine monotherapy (ET) versus adjuvant breast radiation (RT)
and after primary neoadjuvant therapy (NAT) for patients with early breast alone in healthy older women with stage I, estrogen receptor-positive (ER+)
cancer. First Author: Hans-Christian Kolberg, Marienhospital, Bottrop, breast cancer: An analysis of the National Cancer Database (NCDB). First
Germany Author: Anthony H. Bui, Icahn School of Medicine at Mount Sinai, New York,
NY
Background: Scientific efforts aim at a reduction of axillary morbidity through
reduced axillary intervention among patients with early breast cancer. How- Background: The NCCN guidelines state that breast RT may be omitted in
ever, it is still unclear if this approach is feasible in all subtypes based on their patients . 70 years of age with ER+, clinically node-negative, T1 breast
risk of axillary involvement. We analyzed the association of tumor biology and cancer (BC) who receive adjuvant ET. Available data on older patients notes
occult axillary involvement with data from arms A and B of the SENTINA trial that local relapses are the most frequent site of failure, and distant relapse
(Kühn T et al., Lancet Oncol 2013). Methods: Patients were included if they rates are low. The side effects of ET are not inconsequential and negatively
presented with a clinically negative axilla before NAT (arms A and B) and affect QOL. The objectives of this study are to examine clinical outcomes
stratified according to tumor biology. All patients received SLNB before NAT, including overall survival (OS) in women $70 years of age treated by lump-
in cases of negative SLNB without further axillary surgery (Arm A) and in cases ectomy(L)+ET and L+RT in the NCDB. Methods: The 2004-2013 NCDB
of positive SLNB (Arm B) with SLNB and axillary dissection after NAT. Logistic includes 76,431 women $70 years with ER+ stage I BC who underwent L, and
and linear regression analyses were carried out to evaluate the association had a minimum one year follow up. Women who received no adjuvant therapy,
between tumor biology and axillary involvement before and after NAT. both ET+RT, or any chemotherapy were excluded. To limit the analysis to
Results: Of the 1022 patients in arms A and B of the SENTINA trial 926 were healthy women, we excluded subjects with a Charlson comorbidity index . 0.
evaluable for this analysis. Of these, 27.9% had triple negative (TN), 16.3% We identified 24,572 patients who received either adjuvant ET monotherapy
hormone receptor (HR) and HER2 positive (triple positive = TP), 47.6% HR or adjuvant RT alone. Among these, 46% (11,313) received ET and 54%
positive and HER2 negative (luminal) and 8.2% HR negative and HER2 (13,259) breast RT. Overall median follow up was 57 months (range: 12-
positive (HER2) tumors. 39.7% of the luminal, 28.9% of the HER2, 19% of 143 months). Analysis of OS between the 2 treatment groups was performed
the TN and 47% of the TP tumors had involved SLN before NAT. Subgroup using Kaplan-Meier statistics and Cox proportional hazards regression;
comparisons showed a significant difference between luminal and TN (p , propensity weighting was used to balance covariates across the 2 treatment
0.0001), whereas the differences between luminal and TP (p = 0.115) and groups. Results: After propensity weighting, demographic covariates in-
HER2 (p = 0.077) were not statistically significant. The 317 patients with cluding age, race, insurance, and facility type were balanced between the
involved SLN prior to NAT received SLNB and axillary dissection after 2 treatment groups. The median OS for ET was 125.9 months (95% CI
completion of NAT. The analysis after NAT showed trends for lower rates of 120.1-131.8), and 127.2 months for RT (95% CI 124.5-131.7) (p ,
involved lymph nodes for the high-risk groups (TN 20% / TP 14.3% / HER2 0.0001). The weighted hazard of death was 11.7% less in women receiving
8.7%) compared to luminal tumors (27.6%) without reaching statistical RT compared to ET (HR 0.883, 95% CI 0.834-0.936, p , 0.0001).
significance. Conclusions: Our analysis demonstrates that among patients Conclusions: To our knowledge, this is the first large study comparing RT and
enrolled in the SENTINA trial, patients with triple negative disease have ET monotherapy in healthy older women with stage I, ER+ BC. The OS with
the lowest risk for occult lymph node metastases at initial presentation. RT alone is not inferior to ET alone, and in this study population is noted to be
Our results do not justify more intense local intervention among patients better. While this analysis has various limitations not dissimilar from other
with triple negative breast cancer. NCDB database studies, our observations are encouraging and warrant
further research with prospective studies.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
14s Breast Cancer—Local/Regional/Adjuvant

520 Poster Discussion Session; Displayed in Poster Session (Board #12), 521 Poster Session (Board #13), Sun, 8:00 AM-11:00 AM
Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Impact of the extent of resection on primary breast angiosarcoma survival.
Sun, 4:30 PM-6:00 PM First Author: Timur Mitin, Oregon Health and Science University, Portland,
Low-fat dietary pattern and long-term breast cancer incidence and mortality: OR
The Women’s Health Initiative randomized clinical trial. First Author:
Background: The widely accepted standard of care in treating primary breast
Rowan T. Chlebowski, Los Angeles BioMedical Research Institute, Torrance,
angiosarcoma involves surgical resection, often followed by adjuvant therapy
CA
(radiation and/or chemotherapy). The rarity of this disease has precluded
Background: Observational studies of dietary fat intake and breast cancer have large-scale analyses. The question regarding the impact of resection extent
inconsistent findings. To address this issue, the Women’s Health Initiative on survival has yet to be examined on a nationwide scale. Methods: The
(WHI) Dietary Modification (DM) clinical trial assessed a low-fat dietary pattern National Cancer Data Base (NCDB) from 2004-2014 identified primary
influence on breast cancer incidence and outcome. Methods: The WHI DM breast angiosarcoma patients throughout the United States having un-
trial is a randomized, controlled clinical trial conducted at 40 US centers, dergone surgical resection. The extent of resection (mastectomy versus
where 48,835 postmenopausal women, aged 50-79 years, with no previous lumpectomy) was adjusted for several variables (including patient age, race,
breast cancer and dietary fat intake $32% of total energy, were randomly income, primary payer for care, tumor size, adjuvant therapies, and medical
assigned, from 1993-1998, to a usual diet comparison group (60%) or dietary comorbidities) to assess its impact on breast angiosarcoma-related mortality.
intervention group (40%) with goals to reduce fat intake to 20% of energy and Results: Over this eleven-year span, 826 resected primary breast angio-
increase vegetables, fruit, and grain intake. This study is registered as: sarcoma patients were identified in the United States. Mastectomy was by far
NCT00000611. Results: The dietary intervention significantly reduced fat the most common surgical modality for primary breast angiosarcoma (86% of
intake; increased fruit, vegetable and grain intake with modest weight loss patients). Increasing tumor size was predictive for mastectomy over lump-
(3%) (all P, 0.001). During 8.5 years of dietary intervention, there were 8% ectomy (p , 0.0001), and for involvement of adjuvant radiation therapy
fewer breast cancers and deaths from breast cancer were somewhat lower in (p = 0.001). The extent of surgical resection was inversely predictive of
the intervention group but the rates were not significantly different. However, radiation usage (p = 0.017). However, surgical modality was not significantly
deaths after breast cancer (breast cancer followed by death from any cause) predictive of breast angiosarcoma-related mortality. Conclusions: Despite
were significantly reduced in the intervention group, both during in- the frequent preference of mastectomy for primary breast angiosarcoma
tervention (hazard ratio [HR] 0$65 95% confidence interval [CI] 0$45- treatment (more than 6 of every 7 patients), there is no survival benefit of
0$95) and through 16.1 year (median) cumulative follow-up. Now, after mastectomy versus lumpectomy. This lack of benefit should be discussed
long- term, cumulative 19.6 year (median) follow-up, with 3,374 incident with patients, given the reduced operative morbidity of lumpectomy versus
breast cancers, the significant reduction in deaths after breast cancer mastectomy. The Class IIB evidence provided from this analysis represents
continued (with 1,011 deaths, HR 0$85 95% CI 0$74-0$96) and a sig- the highest level of evidence to-date governing management of this disease.
nificant reduction in deaths from breast cancer (breast cancer followed by
death attributed to the breast cancer) emerged (with 383 deaths, HR 0$79
95% CI 0$64-0$97). Conclusions: Adoption of a low-fat dietary pattern
associated with increased vegetable, fruit, and grain intake, demonstrably
achievable by many, significantly reduced the risk of death from breast
cancer in postmenopausal women. To our review, these findings provide
the first randomized clinical trial evidence that a dietary change can
reduce a postmenopausal woman’s risk of dying from breast cancer.
Clinical trial information: NCT00000611.

522 Poster Session (Board #14), Sun, 8:00 AM-11:00 AM 523 Poster Session (Board #15), Sun, 8:00 AM-11:00 AM
Do all patients with HER2-positive breast cancer require one year of adjuvant Adherence to adjuvant endocrine therapy after breast cancer in Sweden
trastuzumab?: A systematic review and meta-analysis. First Author: Paul 2008-2010: A nationwide survey. First Author: Anne Andersson, Department
Stewart, Western University, Schulich School of Medicine and Dentistry, of Radiation Sciences, Oncology, Umeaa, Sweden
London, ON, Canada
Background: In estrogen receptor (ER) positive early breast cancer (EBC)
Background: One year of adjuvant trastuzumab (T) remains the standard adjuvant endocrine therapy (AET) is crucial to reduce recurrence and
treatment for patients with HER2 positive breast cancer. Results from mortality. Previous studies have shown that adherence to AET is lower than
randomized trials with diverse non-inferiority margins comparing one year expected and could negatively affect outcome. Since the year of 2000, BC
to a shorter duration of adjuvant T were not consistent, particularly with the patients in Sweden are treated in accordance to national guidelines.
PERSEPHONE and the final PHARE and Short-HER trials’ results. Our Treatment is offered at a low cost for the patient. The aim of the study was
objective was to conduct a systematic review and meta-analysis of ran- to estimate the adherence to AET in Sweden by regions and age groups.
domized trials in patients with HER2 positive breast cancer to assess Methods: Women with a first primary EBC diagnosis 2008-2010 were
whether a shorter duration of adjuvant T was non-inferior to one year of identified through the Swedish Cancer Registry (SCR). Individual tumour
treatment. Methods: PubMed, EMBASE and The Cochrane Library were and treatment data were retrieved from the Swedish National Breast Cancer
searched for eligible randomized trials. Hazard ratios (HR) for disease free Registry (SNBCR). Patients with ER negative tumours, small tumours
and overall survival (DFS, OS) were weighted using generic inverse variance (# 10 mm) and metastatic disease was excluded from the study since there
and pooled in a meta-analysis using random-effects models. The median of were no indication to AET. Likewise, were individuals with AET registered to
non-inferiority margins derived from each trial was calculated to set a non- be administered by a third part excluded. Dispensed treatment from
inferiority margin of 1.29 for the pooled analysis. Subgroup analyses pharmacies was obtained through the Swedish Prescription Registry and
compared survival outcomes by estrogen receptor (ER) status, nodal status, medication possession rate (MPR) was calculated as number of dispensed
length and timing of trastuzumab treatment. Results: Data of 11,376 pa- doses divided by treatment duration in days. Good adherence to treatment
tients from 5 trials were analyzed. A shorter duration of T was non-inferior to in a patient was set at MPR $ 80 %. Adherence was calculated for 3 and 5
one year of therapy for DFS (HR 1.13, 95%CI 1.03-1.24) but worse for OS years. Results: Twenty-one thousand sixteen (21 016) individuals with a first
(HR 1.16, 95%CI 1.01-1.32). In addition, the non-inferiority for DFS was primary BC between 2008 and 2010 was identified through SCR of which 20
met for patients with ER positive disease (HR 1.1, 95%CI 0.95-1.28) and 596 were registered in the SNBCR. A total of 10 176 met the inclusion
patients treated with 6 months (HR 1.09, 95%CI 0.98-1.22) or sequential T criteria in the study. Adherence after 3 years was 88.0 % and after 5 years
(HR 0.97, 95%CI 0.75-1.27). Conversely, the non-inferiority for DFS was 82.5 %. Adherence differed between regions in Sweden and was positively
not met for patients with ER negative disease (HR 1.22, 95%CI 1.06-1.41), associated with age at diagnosis between 41-74 years. Urban areas had
patients treated with 9 weeks (HR 1.26, 95%CI 1.02-1.55) or concomitant a lower adherence than rural areas (80.7 % vs 83.6 %; p= ,0.001).
T (HR 1.25, 95%CI 1.07-1.45) and patients with node negative (HR 1.12, Conclusions: Adherence to AET in Sweden was good, although there were
95% 0.93-1.35) or positive (HR 1.16, 95%CI 0.99-1.36) disease. differences by age and urban and rural areas. Further studies are needed to
Conclusions: Within the limitations of the available data and the different identify factors affecting differences in adherence, with the purpose of
non-inferiority margins used in randomized trials, a shorter duration of initiate actions to increase adherence to AET in ER positive EBC patients.
adjuvant T is non-inferior to one year of therapy for DFS in patients with
HER2 positive breast cancer, particularly in patients with ER positive dis-
ease. Further trials with appropriately chosen non-inferiority margins are
needed to confirm the optimal duration of T in patients with low-risk disease.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Breast Cancer—Local/Regional/Adjuvant 15s

524 Poster Session (Board #16), Sun, 8:00 AM-11:00 AM 525 Poster Session (Board #17), Sun, 8:00 AM-11:00 AM
De-escalating adjuvant trastuzumab in human epidermal growth factor Effect of tart cherry on aromatase inhibitor-induced arthralgia (AIA) in
receptor 2 (HER2)-positive early-stage breast cancer: A systemic review nonmetastatic hormone-positive breast cancer patients: A randomized
and meta-analysis. First Author: Hadar Goldvaser, Rabin Medical Center, double-blind placebo-controlled trial. First Author: Mina Shenouda, Mar-
Beilinson Hospital, Davidoff Center, Kyriat Ono, Israel Marshall University, Huntington, WV
Background: One year of adjuvant trastuzumab in combination with che- Background: Aromatase inhibitors (AI) are the standard treatment for hor-
motherapy is the standard of care in early-stage HER2 positive breast cancer. mone receptor-positive breast cancer in postmenopausal women. About half
Existing data on shortening trastuzumab treatment show conflicting results. of patients (pts) taking AI suffer from AIA which can be severe enough to cause
Methods: A search of PubMed and conferences identified randomized trials noncompliance. Suboptimal AI adherence is associated with decreased
that compared abbreviated trastuzumab therapy to one year of treatment in disease-free and overall survival, suggesting that improving adherence will
early-stage HER2 positive breast cancer. Hazard ratios (HRs) and 95% lead to improved breast cancer outcomes. Effective interventions for AIA
confidence intervals (CI) were extracted for disease free survival (DFS) and are still limited. In clinical trials, tart cherry (TC) showed beneficial effect
overall survival (OS). Data on the number of DFS and distant relapse events on musculoskeletal pain associated with osteoarthritis, gout, and stren-
were also collected as were the number of patients at risk in each group. uous exercise. The flavonoids and anthocyanins in TC reportedly exert an
Subgroup analyses evaluated the effect of nodal involvement, estrogen anti-inflammatory effect that may lessen adverse effects of estrogen de-
receptor (ER) expression and the duration of abbreviated trastuzumab (9- ficiency. This trial aimed to investigate whether TC can reduce AIA in non-
12 weeks versus 6 months). Odds ratios (ORs) and 95% CI were computed metastatic hormone positive breast cancer (NMHPBC) pts. Methods: This
for pre-specified cardiotoxicity events including cardiac dysfunction and is a randomized, placebo-controlled, double-blind trial. Eligible Pts with
congestive heart failure (CHF). Results: Analysis included 6 trials comprising NMHPBC on AI for at least 4 weeks were randomized to TC concentrate
11603 patients. In most studies adjuvant chemotherapy included anthracy- (equivalent to 50 tart cherries) versus placebo (P) [syrup] in a 1:1 model.
clines and taxanes. Shorter trastuzumab treatment was associated with worse Pts were instructed to take 1 Oz of TC or P in 8 Oz water daily for 6 weeks,
DFS (HR = 1.14, 95% CI 1.05-1.25, p = 0.002) and OS (HR = 1.15, 95% CI and to document their pain intensity at baseline, weekly and at study
1.02-1.29. p = 0.02). The effect on DFS was not influenced by ER status (p for completion in a diary using a Visual Analog Scale (VAS), with 0 mm in-
the subgroup difference = 0.23), nodal involvement (p = 0.44) or the different dicating no pain, and 100 mm indicating highest pain. Results: 60 pts
duration of trastuzumab in the experimental arm (p = 0.08). In absolute terms, were enrolled from May 2016 to August 2018. 2 pts did not complete the
after an estimated median follow-up of 71 months, shorter treatment with study due to diarrhea. 10 pts were non-compliant. 48 pts were included in
trastuzumab was associated with an absolute increase in DFS events of 2.3%. the final analysis. TC group (23 pts) had 34.7% mean decrease in pain
Shorter trastuzumab treatment was associated with lower odds of cardiac compared to 1.4% in P group (25 pts). This difference was statistically
dysfunction (OR = 0.67, 95% CI 0.55-0.81, p , 0.001) and CHF (OR = 0.66, significant (Mann-Whitney U Test P = 0.034). Conclusions: Tart cherry
95% CI 0.50-0.86, p = 0.003). Conclusions: Compared to one year, shorter can significantly improve AIA in non-metastatic breast cancer patients.
duration of adjuvant trastuzumab is associated with significantly worse DFS
and OS, despite favorable cardiotoxicity profile. One year of trastuzumab
should remain the standard adjuvant treatment in early-stage HER2 positive
breast cancer with appropriate cardiac monitoring.

526 Poster Session (Board #18), Sun, 8:00 AM-11:00 AM 527 Poster Session (Board #19), Sun, 8:00 AM-11:00 AM
Simulation modeling of the effects of adjuvant chemotherapy in early-stage Ki67 during and after neoadjuvant trastuzumab, pertuzumab and palbociclib
breast cancer. First Author: Jinani Jayasekera, Lombardi Cancer Center plus or minus fulvestrant in HER2 and ER-positive breast cancer: The
MedStar Georgetown University Hospital, Washington, DC NA-PHER2 Michelangelo study. First Author: Luca Gianni, IRCCS San Raffaele
Hospital, Milan, Italy
Background: The recent Trial Assigning Individualized Options for Treatment
(TAILORx) was practice changing. However, several important questions remain Background: Downregulation of Ki67 by neoadjuvant endocrine therapy
unanswered about therapy for women with early stage, hormone receptor positive predicts activity of endocrine treatments in hormone receptors positive
(HR+), HER2- breast cancers, including chemotherapy effects by age and different breast cancer. NA-PHER2 is an exploratory phase II study (NCT02530424)
Oncotype recurrence score (RS) cut-points, and longer follow-up. We developed a
assessing Ki67 changes in patients with HER2+ and ER+ breast cancer
simulation model to extend the trial results to begin to fill these gaps. Methods: We
developed a simulation model using an empirical Bayesian approach to simulate
undergoing dual HER2 block and palbociclib. Cohort A of the NA-PHER2
women eligible for the TAILORx trial. The joint distributions of patient and tumor showed significant decrease of Ki67 at week 2 and at surgery and patho-
characteristics were derived from de-identified TAILORx data. The remaining inputs logical complete response (pCR) in 27% of patients (Lancet Oncol 2018).
used data independent of the trial, including SEER, the Oxford Overview, and other Methods: After completing cohort A two additional cohorts were started. In
trials. TAILORx was simulated to examine the effects of chemotherapy (+ hormonal Rx) Cohort B cases with HER2 3+ or amplified unilateral breast cancer received
vs. hormonal Rx alone on distant recurrence-free survival (DRFS) at 9- and 20-years by therapy with dual block and palbociclib without fulvestrant. In Cohort C
age (#50 and .50 years) and RS (11-25 and 16-25). We also evaluated the effects of tumors with Ki67 .20% and HER2 low (1+/2+, no amplification) received
chemotherapy in women with RS 26-30. Hazard ratios (HR) were determined using Cox also fulvestrant. Trastuzumab and pertuzumab q3 wks were dosed for 6
regressions and DRFS by treatment were derived from Kaplan-Meier curves. We report cycles and palbociclib for 5 cycles (125 mg po q.d. 3q4 wks). Fulvestrant in
the mean results from 1000 trial simulations, where each simulation randomly sampled Cohort C was given im 500 mg q4 wks for 5 cycles. Primary endpoint was
values for each parameter from their observed joint distribution. Finally, the original trial Ki67 change from baseline to 2 weeks and at surgery. Results: 26 eligible
was replicated for model validation. Results: The model closely replicated actual trial
results. Sample sizes ranged from 7000-10000. The model estimated that chemo-
patients in cohort B and 23 in cohort C with centrally confirmed HER2 and
therapy improved DFRS in women aged #50 with RS 16-25 (Table). The 20-year event ER status were recruited. Ki67 was centrally assessed. Main results are
rates remained low in the 11-25 category. Among women with RS 26-30, HR for no reported in the table. Clinical trial information: NCT02530424. The most
chemo vs. chemo was 1.38 (95% CI:1.18-1.58). Conclusions: Simulation suggests frequent G .=3 adverse events were neutropenia (36%) and gastrointestinal
that chemotherapy may reduce distant recurrence in younger women at different cut disorders (12%). Conclusions: Dual block of HER2 and palbociclib caused
points between 11-25. Simulation modeling may be useful to translate trial results to robust persistent decrease of Ki67 as in cohort A. In cohort B without en-
broader population subgroups than those possible to test in RCTs. Nine-year distant docrine therapy there also were pCR and high objective response rate. Ef-
recurrence-free survival by chemotherapy. fects on Ki67 and ORR were similar in HER2 low tumors. The chemo-free
RS 11-25 RS 16-25 approach of NA-PHER2 leads to promising therapeutic effects and deserves
Rate at 9-Yr (95% CI) Rate at 9-Yr (95% CI)
investigation in ER+ HER2+ tumors to spare the toxicity of chemotherapy,
Mean HR* % Mean HR* % and in HER2-low tumors, in which functional activation of HER2 may lead to
Age (SE) No Chemo Chemo difference (SE) No Chemo Chemo difference
resistance to endocrine therapy. Supported in part by unrestricted grants of
£50 1.32 94.5% 95.7% 1.2% 1.64 92.1% 94.9% 2.8% Pfizer Italia S.r.l. and Roche S.p.a. Italia.
(0.15) (93.4- (94.7- (0.10) (91.1- (94.1-
95.4) 96.5) 93.0%) 95.6%)
>50 1.21 94.3% 95.3% 1.0% 1.04 92.7% 92.8% 0.1% Ki67 geometric mean at KI67 mean change after KI67 mean change at pCR ORR
(0.14) (93.3- (94.4- (0.11) (91.7-93.7) (91.8-93.7) baseline (range) 2 weeks surgery % %
95.1) 96.0)
Total 1.14 94.2% 94.7% 0.5% 1.63 91.0% 94.1% 3.1% Cohort 33.4 (11.0-89.0) - 25.7 - 9.5 19.2 88.5
(0.13) (93.0- (93.6- (0.11) (89.8- (93.1- B P , 0.0001 P = 0.033
95.1) 95.7) 92.0%) 94.9%) Cohort 32.4 (21.0-78.0) - 29.5 - 19.3 0 78.3
C P , 0.0001 P , 0.0001
* No Chemo vs. Chemo

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
16s Breast Cancer—Local/Regional/Adjuvant

528 Poster Session (Board #20), Sun, 8:00 AM-11:00 AM 529 Poster Session (Board #21), Sun, 8:00 AM-11:00 AM
A randomized phase III trial comparing dose-dense epirubicin and The effect of metformin on sex hormones in non-diabetic breast cancer
cyclophosphamide (ECdd) followed by paclitaxel (T) with paclitaxel plus patients in CCTG MA.32: A Phase III randomized adjuvant trial of metformin
carboplatin (PCdd) as adjuvant chemotherapy for early triple-negative breast versus placebo in addition to standard therapy. First Author: Isabel Pimentel,
cancer patients with high-recurrence risk. First Author: Jiani Wang, Mount Sinai Hospital- University of Toronto, Toronto, ON, Canada
Department of Medical Oncology, National Cancer Center/National Clinical
Background: The effect of metformin on sex hormones (SHs) levels that may
Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical
impact breast cancer (BC) outcome, is unclear. We evaluated the effect of
Sciences and Peking Union Medical College, Beijing, China
metformin on SHs in a subgroup of women enrolled in the placebo-controlled
Background: There are no well-established adjuvant chemotherapy (AC) reg- MA.32 trial, a phase III randomized study of nondiabetic subjects with
imens for early triple negative breast cancer (TNBC). Our randomized phase III T1-3, N0-3 BC randomized to receive metformin 850 mg po bid or placebo
trial was designed to compare dose dense paclitaxel plus carboplatin (PCdd) for 5 years. Methods: Our substudy was conducted on the group of post-
with commonly used dose dense epirubicin and cyclophosphamide, followed menopausal women with estrogen receptor (ER) and progesterone receptor
by paclitaxel (ECdd-T) regimen as AC for TNBC with high recurrence risk. (PR) negative BC and not receiving endocrine treatment enrolled in the trial.
Methods: Between May 2011 and November 2015, TNBC patients were Post-menopausal was defined as prior bilateral oophorectomy or . 12 months
randomized in 1:1 ratio to receive PCdd or ECdd-T regimen as AC every two since last menses without prior hysterectomy. Fasting blood and adherence to
weeks for 8 cycles with administration of granulocyte stimulating factor (G- study drug at baseline and 6 months was required. Sex hormone binding
CSF) support. The primary endpoint was 3-year disease free survival globulin (SHBG), free testosterone and estradiol serum levels were evaluated
(DFS).The secondary endpoints included overall survival (OS) and safety. using Roche Modular competitive ECLIA (electrochemiluminescense immu-
Survival analyses were also performed for different subgroups stratified by noassay). We compared change from baseline to 6 months in estradiol, SHBG
age status (#40 years vs .40 years), Ki 67(,50 vs $50), tumor size and free testosterone between study arms using Wilcoxon sum rank tests and
(,2cm vs $2cm), nodal status (N- vs N+) and treatment free survival (TFS) regression models to adjust for baseline BMI and weight change. Results: 304
(,30 days vs $30 days). Results: In total, 132 patients with a median age women were eligible, 135 metformin vs 169 placebo; tumor stage and prior
of 49 years (PCdd 64 patients, ECdd-T 68 patients) were enrolled. After a (neo)adjuvant chemotherapy (98% in both arms) and HER2 targeted treat-
median follow-up of 57.3 months, 23 events were observed (18 in ECdd-T, ment were well balanced between arms. At baseline mean age was 58.266.9
5 in PCdd). Patients in the PCdd arm had significantly higher DFS rate than vs 57.667.9 years, mean BMI 26.964.9 vs 28.666.2 Kg/m2, median es-
that in the ECdd-T arm (log-rank p = 0.0046, hazard ratio (HR) 0.305, tradiol 29.82 vs 31.01 pmol/L, SHBG 80.6 vs 73.7 nmol/L and free tes-
95% confidence interval (CI) = 0.134-0.693). The 3-year DFS rate was tosterone 0.02 vs 0.03 nmol/L in metformin vs placebo patients, respectively.
93.7% with PCdd versus 77.9% with ECdd-T,respectively. Difference in In univariable analysis, median estradiol decreased significantly between
3-year OS rate was observed between the two arms (98.4% vs 92.6%), baseline and 6 months on the metformin vs placebo arm (-4.81 vs 0 pmol/L,
significantly higher in the PCdd arm (p = 0.0268). Both regimens were well p = , 0.0001); this difference remained significant after controlling for baseline
tolerated with manageable adverse events(AEs). Worse neutropenia (Grade BMI (p = 0.0001) and weight change (mean weight change -1.863.5 vs
3/4: 48.5% in ECdd-T vs. 21.9% in PCdd, p=0.002) was found in ECdd-T 0.463.5 Kg, p = , 0.0001 for metformin vs placebo respectively) between
arm. 3-year DFS rate for PCdd was superior in the following subgroups, baseline and 6 months (p = 0.0007). In contrast, there was no significant change
age.40 years, clinically evaluated lymph nodes, TFS ,30 days, with in SHBG or free testosterone (median change SHBG -5.5 vs -5.9 nmol/L, p =
statistical significance (p ,0.05). Conclusions: Our data suggested that 0.33; median change free testosterone 0 vs 0 nmol/L, p = 0.33 for metformin vs
PCdd was superior to ECdd-T as AC for early TNBC in terms of improving placebo respectively). Conclusions: Metformin lowered estradiol levels, in-
3-year DFS and OS. PCdd with lower hematological toxicity might be an dependent of weight change in non-diabetic post-menopausal women with ER
appropriate regimen for early TNBC patients with high recurrence risk. and PR negative BC enrolled onto MA.32 trial. This observation suggests a new
Clinical trial information: NCT01378533. metformin action that has potential relevance to BC management.

530 Poster Session (Board #22), Sun, 8:00 AM-11:00 AM 531 Poster Session (Board #23), Sun, 8:00 AM-11:00 AM
Prognostic value of a new clinical-genomic model to predict 10-year risk of Abandonment trajectories of conventionally fractionated adjuvant radiotherapy
recurrence in patients with operable breast cancer. First Author: Lei Lei, The in breast cancer care. First Author: Xiao Xu, Yale School of Medicine, New
University of Mississippi, Jackson, MS Haven, CT
Background: Searching for a specific biomarker to predict long-term risk of Background: Hypofractionated radiotherapy is the recommended approach
recurrence for all breast cancer subtypes is challenging. DGM-CM6 (Distant for adjuvant breast cancer care. Yet physicians have been slow to abandon
Genetic Model-Clinical variable Model 6) is a new clinical-genomic prog- conventionally fractionated radiotherapy (CFRT). We examined distinct
nostic model developed from the 18-gene panel which was reported pre- trajectories of abandoning CFRT among physician patient-sharing peer
viously. This study aims to validate the long-term prognostic value of this new groups and the associated cost implications. Methods: Using 2011-2014
model in all subtypes of operable breast cancer patients. Methods: We in- national Medicare claims, we constructed peer groups of physicians (radi-
cluded 752 operable breast cancer patients with stage I-III in all subtypes ation and medical oncologists, surgeons, and primary care physicians) who
treated in a Cancer Center from 2005 to 2014 as the internal validation (IV) cared for women with breast cancer. Women $66 years of age who un-
cohort. The median follow-up was 94.1 months. Meanwhile, Affymetrix derwent lumpectomy plus adjuvant radiotherapy were included. Peer groups
U133P2 (n = 1139) data obtained from GEO (GSE9195/16391/17907/ represented physicians who frequently shared patients with one another. For
19615/20711/21653/42568, EMTAB365) were collected as the external each peer group, we calculated risk-adjusted rate of CFRT use in 2011-2012
validation (EV) dataset. The prognostic effect of DGM-CM6 was then eval- (T1) and 2013-2014 (T2) after accounting for patient risk factors. Based
uated by uni- and multivariate analyses. The low- and high-risk patients on these utilization rates and a latent growth curve analysis, we identified
( , 33 or $ 33 as cut-off value) classified by DGM-CM6 were evaluated by distinct trajectories of abandoning CFRT among peer groups and estimated
the 10-year distant relapse-free interval (DRFI), relapse-free interval (RFI), their cost implications from the Medicare perspective. Results: The 215
relapse-free survival (RFS) and distant relapse-free survival (DRFS), re- physician peer groups (caring for 16,988 patients) exhibited four distinct
spectively. We further compared the predictive performance between DGM- trajectories of CFRT use: 1) persistent high use (mean adjusted utilization:
CM6/DGM and PAM50-ROR score in our IV dataset. Results: In the IV T1 = 94.3%, T2 = 90.6%); 2) decreased high use (T1 = 81.3%, T2 =
dataset, DGM-CM6 was proved to be an independent prognostic factor by 65.3%); 3) decreased medium use (T1 = 60.1%, T2 = 44.0%); and 4)
multivariate analysis with hazard ratios of 3.1 (1.6-6.0) for RFS (P = 0.0009) decreased low use (T1 = 31.6%, T2 = 23.6%). They accounted for 33.0%,
and 3.2 (1.6-6.3) for DRFS (P = 0.0009). Significant differences were 35.3%, 25.6% and 6.0% of the peer groups, respectively. Compared to
observed between low- and high-risk groups with 10-year RFI (94.0% vs. “persistent high use” of CFRT, peer groups with “decreased high use” and
83.5%, P , 0.0001), RFS (90.0% vs. 80.5%, P = 0.0003), DRFI (94.1% “decreased medium use” of CFRT had a smaller proportion of patients
vs. 85.0%, P , 0.0001), and DRFS (90.1% vs. 81.9%, P = 0.0004), receiving radiotherapy at free-standing (vs. hospital-based) facilities (ad-
respectively. The prognostic value of RFS was convinced in the EV dataset justed odds ratio = 0.89, p = 0.01; and 0.78, p , 0.01; respectively). Ac-
(HR = 1.34, P = 0.00052) by the DGM only. According to C-index estimate celerating abandonment of CFRT in the three higher utilization trajectories
analysis, DGM appeared to have better performance comparing with PAM50 to “decreased low use” for the 2011-2014 patient cohort could save Medi-
ROR score in prediction of long-term DR, DRFS, RFI, and RFS in N0 patients care $342.7 million (95% confidence interval: $232.2-$457.6 million).
(C index for distant recurrence: 0.582 by DGM, 0.528 by ROR). Conclusions: Physician peer groups had distinct trajectories in abandoning
Conclusions: DGM-CM6 could be a new long-term prognostic model to be CFRT exhibiting different baseline levels and rates of change in utilization.
applied in all subtypes of operable breast cancer patients. Further validation Efforts to reduce overuse of CFRT could generate substantial savings.
in a large scale of clinical trials is needed.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Breast Cancer—Local/Regional/Adjuvant 17s

532 Poster Session (Board #24), Sun, 8:00 AM-11:00 AM 533 Poster Session (Board #25), Sun, 8:00 AM-11:00 AM
EPHA genetics for prediction of paclitaxel-induced peripheral neuropathy Influence of competing risks on estimates of recurrence risk and breast
sensitivity. First Author: Lauren A Marcath, University of Michigan College of cancer-specific mortality in analyses of the Early Breast Cancer Trialists
Pharmacy, Ann Arbor, MI Collaborative Group. First Author: Ramy Saleh, Princess Margaret Cancer
Centre and University of Toronto, Toronto, ON, Canada
Background: Chemotherapy induced peripheral neuropathy (CIPN) is a
common, debilitating paclitaxel side effect that has been primarily attributed Background: Early stage breast cancer is a curable disease with the majority
to cumulative systemic paclitaxel exposure. Paclitaxel dose-adjustment to of patients dying of causes other than breast cancer. The influence of these
achieve target systemic exposure decreases but does not eliminate severe competing risks of death on the interpretation of Kaplan-Meier(KM)-based
CIPN, suggesting some patients are inherently CIPN-sensitive. Ephrin analyses such as those performed by the Early Breast Cancer Trialists
(EPHA) polymorphisms have been reported to increase CIPN occurrence Collaborative Group (EBCTCG) are unknown. Methods: We searched the
(Baldwin 2012, Leandro-Garcia 2013, Boora 2016) but replication has Clinical Trial Service Unit and Epidemiological Studies Unit website at
been challenging, perhaps due to the inability to isolate CIPN-sensitivity Oxford University to identify all meta-analyses published by the EBCTCG
by accounting for systemic exposure differences. The study purpose was to between 2005 and 2018. Studies were included if they contained KM
determine if EPHA genetic variants previously associated with CIPN oc- curves with risk estimates for either breast cancer mortality and/or breast
currence are associated with CIPN-sensitivity. Methods: PN was assessed cancer recurrence. The potential influence of competing risks was estimated
at baseline and weekly using the 8-item sensory subscale (CIPN8) of the using a validated multivariate linear model that predicts the amount that the
patient-reported EORTC CIPN20 in patients receiving paclitaxel 80 mg/m2 KM risk estimates are biased relative to outcome risk measured with the
weekly x 12. EPHA4, EPHA5, EPHA6, and EPHA8 were sequenced in cumulative incidence function (CIF). Results: The initial search identified
germline DNA. Associations with higher PN-sensitivity were tested for 14 analyses published by the EBCTCG with 10 of the 14 studies (71%)
three genetic models (total variants, coding variants, and rs7349683) by susceptible to competing risk bias cited both the number of events of interest
incorporating genetics into previously published CIPN8 predictive models and competing events. Eight of the ten studies (80%) had a relative dif-
that included measured paclitaxel exposure. Significant associations were ference between the KM estimate and the competing risk adjusted estimate
tested for association with higher risk of PN-related treatment disruption of more than 10% while 2 of 10 (20%) had a difference of less than 10%.
(i.e. dose decrease, delay, or discontinuation). Results: In the 59 included The relative difference between the KM and adjusted estimates was 28.4%
patients, carrying EPHA5 rs7349683 was associated with greater CIPN for local recurrence, 16.8% for distant recurrence, and 6.7% for breast
sensitivity (beta coefficient: 0.40, standard error: 0.14, p = 0.005), in- cancer-specific mortality. There was 2.2% difference between KM and
dicating these patients had a greater increase in CIPN8 during treatment adjusted analyses between 0-4 years and 18.9% beyond 10 years of follow
than would be predicted based on cumulative paclitaxel exposure. Rs7349683 up. Use of KM and CIF-based analysis did not influence treatment effect in
was not associated with increased PN-induced treatment disruption, perhaps the majority of included studies. Conclusions: This study provides estimates
due to the low number of events (n = 19). Conclusions: Using a novel approach for the overestimation of risk in Kaplan-Meier analyses resulting from failure
that isolates CIPN-sensitivity by accounting for measured paclitaxel exposure, to address competing risk bias. CIFs are more appropriate to measure
our results provide further evidence that EPHA5 rs7349683 may be a promising outcome risk over time and should be used especially for long-term follow-up
marker for CIPN. If additional validation studies confirm this association, ge- studies and for analysis of rare events.
netic testing could enable personalized treatment strategies to prevent CIPN in
patients with breast cancer.

534 Poster Session (Board #26), Sun, 8:00 AM-11:00 AM 535 Poster Session (Board #27), Sun, 8:00 AM-11:00 AM
Breast Cancer Index and risk stratification in luminal subtypes: A trans-ATAC Additional prognostic value of OncoMasTR multigene prognostic signature
study. First Author: Ivana Sestak, Centre for Cancer Prevention, Wolfson to clinicopathological information in patients with HR-positive, HER2-
Institute of Preventive Medicine, Queen Mary University of London, London, negative, lymph node-negative breast cancer from the TAILORx Tissue Bank,
United Kingdom Ireland. First Author: Darran O’Connor, Department of Molecular and Cel-
lular Therapeutics, Royal College of Surgeons in Ireland, Dublin, Ireland
Background: The Breast Cancer Index (BCI) is a gene-expression based sig-
nature that provides prognostic information for overall (0-10 years) and late Background: Multigene prognostic signatures (MGPS) can identify early stage
(5-10 years) distant recurrence (DR) and prediction of extended endocrine breast cancer patients who may require less aggressive treatment. To be
benefit in hormone receptor positive (HR+) early stage breast cancer. The clinically useful, MGPSs must provide additional prognostic information to
current analysis aims to further characterize, correlate and compare the clinicopathological information routinely used by clinicians. The OncoMasTR
prognostic performance of BCI in luminal subtypes based on immunohisto- MGPS was discovered via a novel bioinformatic transcriptional network
chemical classification. Methods: 670 postmenopausal women with HR+, analysis. OncoMasTR consists of genes – Master Transcription Regulators
LN- disease from the TransATAC cohort were included in this analysis. Luminal (MTRs) – that regulate previously known prognostic genes and have
A-like tumors (LumA) were identified as those with ER+ and/or PR+ and HER2 -, identified functional roles in several hallmarks of cancer including pro-
and Ki67 , 20% by IHC. All other tumors were classified as Luminal B-like liferation, invasion and metastasis. The OncoMasTR Molecular Score (OM)
(LumB) for this analysis. Primary endpoint was DR. Cox regression models were consists of just 3 MTRs. The OncoMasTR Risk Score (OncoMasTR)
used to examine BCI prognostic performance according to luminal subtype, combines OM with lymph node status and tumour size, and categorises
adjusting for the clinicopathological model Clinical Treatment Score (CTS). patients as low or high risk. The OncoMasTR Test has been analytically and
Results: 452 (67.5%) patients were classified as LumA and 218 (32.5%) as clinically validated. Methods: MTR expression was measured by RT-qPCR
LumB. BCI was highly prognostic in LumA cancers (adjusted HR = 1.57 (1.23- in tissue from 404 patients enrolled in an independent translational trial
1.96), P , 0.001, DLR-x2= 14.09), but not in LumB tumors (adjusted HR = (NCT02050750) that collected tissue and clinical data from patients enrolled
1.20 (0.94-1.52, P = 0.14, DLR-x2= 2.23). In LumA, 10-year DR risks in BCI in TAILORx in Ireland. OM, OncoMasTR and Oncotype DX Recurrence Score
intermediate and high risk groups were very similar (25.6% (16.4-38.6) and (RS) were compared on the additional prognostic value they provided to Ki67,
25.3% (13.5-44.3), respectively) and significantly different from BCI low Nottingham Prognostic Index (NPI) and other clinicopathological information
(3.9% (2.1-7.0); HR = 7.47 (3.50-15.96) and HR = 8.13 (3.27-20.23), for distant recurrence (DR) and invasive disease (ID). Results: OM (LRx2 =
respectively). In LumB, 10-year DR risks in BCI low and BCI intermediate risk 20.3, p , 0.00001, c-index = 0.84) and OncoMasTR (LRx2 = 22.6, p ,
groups (13.8% (6.8-26.9) and 14.6% (8.3-24.9), respectively) were very 0.00001, c-index = 0.85) were significantly prognostic, and more prognostic
similar and significantly lower than for the BCI high (29.1% (20.0-41.1)). Lum than RS (LRx2 = 8.4, p = 0.004, c-index = 0.73) for DR. OM and OncoMasTR
subtyping was only prognostic in the BCI low risk group (LumA vs. LumB: HR = provided more additional prognostic information than RS to Ki67, NPI, tumour
4.27 (1.65-11.02)) but not in the other two BCI risk groups. Conclusions: BCI size, tumour grade, and age for DR. Similar results were found when OM,
provided significant prognostic information in Lum A subtype. These results OncoMasTR and RS were compared on prognostic performance for ID.
show that BCI intermediate and high risk had similar risk of DR in LumA Conclusions: OM and OncoMasTR were significantly prognostic for DR and
tumors, while shared similarly low risk of DR as BCI-low in LumB tumors. ID and added significant prognostic value to Ki67, NPI, and other clini-
Further evaluation is needed to elucidate the distinct mechanisms underlying copathological information. Furthermore, OM and OncoMasTR showed
each classification system. superior prognostic performance to RS.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
18s Breast Cancer—Local/Regional/Adjuvant

536 Poster Session (Board #28), Sun, 8:00 AM-11:00 AM 537 Poster Session (Board #29), Sun, 8:00 AM-11:00 AM
Adjuvant chemotherapy in small node-negative triple-negative breast cancer Breast cancer-specific survival outcomes among patients based on 21-gene
(TNBC). First Author: Tessa Gerjanne Steenbruggen, Department of Medical recurrence score. First Author: Wade T. Swenson, Lake Region Healthcare,
Oncology, Netherlands Cancer Institute, Amsterdam, Netherlands Fergus Falls, MN
Background: International guidelines differ in their recommendation for Background: The 21-gene recurrence assay predicts cancer recurrence rates
adjuvant chemotherapy in small node negative TNBC. We evaluated asso- and benefit from adjuvant chemotherapy among patients with hormone-
ciations of chemotherapy with long-term outcome in a large population- receptor-positive breast cancer. The National Cancer Institute and Genomic
based TNBC cohort. Methods: All patients diagnosed with pT1N0M0 TNBC Health collaborated to provide Recurrence Score (RS) results to complement
between 2005 and 2015 were identified from the Netherlands Cancer data from 17 population-based cancer registries in the Surveillance, Epi-
Registry. Patient, tumor and therapy characteristics were recorded. Date and demiology, and End Results (SEER) Program. Methods: Using the SEER
cause of death were obtained from Statistics Netherlands. We used mul- database with linked RS results, a cohort of female breast cancer patients,
tivariable cox regression models to evaluate associations of chemotherapy age greater than 20 years, was identified by RS (low: , 18, intermediate:
with overall survival (OS) and breast-cancer specific survival (BCSS), ad- 18-30, high: . 30) diagnosed between the years 2004 and 2007. A ret-
justed for baseline characteristics. Subgroup analyses were performed by rospective analysis was conducted to determine eight-year breast cancer-
tumor size and grade. Results: We identified 4393 patients: 284 with T1a, specific survival rates based on RS and other variables. Results: 10,318
924 with T1b, and 3185 with T1c tumors. Chemotherapy was adminis- patients were identified in the cohort. 5,194 had a low risk RS; 4,282 had an
tered in 53% of patients: 6% with T1a, 17% with T1b and 67% with T1c. intermediate risk RS; 872 had a high risk RS. Histologic subtypes were:
Chemotherapy use increased over time and varied by geographic region. 7,459 infiltrating ductal carcinoma, 941 mixed infiltrating and lobular
Patients receiving chemotherapy were younger, had larger tumors, higher carcinoma, 933 lobular carcinoma, 327 mixed infiltrating ductal and other
tumor grade, and more often isolated tumor cells (itc) in the lymph nodes. histology, 244 mucinous adenocarcinoma, 101 tubular adenocarcinoma,
At a median follow-up of 7 years (IQR 5-10 years), 611 patients had died, of 45 mixed lobular and other histology, 268 other histologies. 644 low risk RS
whom 287 due to breast cancer. Chemotherapy was associated with im- patients received chemotherapy (12.4%); 1,608 intermediate risk RS pa-
proved OS in the whole group (adjusted hazard ratio [aHR] 0.55; 95% CI tients received chemotherapy (37.8%); 593 high risk patients received
0.44–0.69), in the pT1c subgroup (aHR 0.53, 95% CI 0.41-0.67), and in chemotherapy (68.0%). The eight-year breast cancer-specific survival rates
grade 3 tumors (aHR 0.50, 95% CI 0.39-0.65). Associations were not (with 95% confidence intervals) among low risk RS patients without known
significant for pT1ab or grade 1-2 tumors (table). Findings for BCSS were in chemotherapy administration was 98.9% (98.5, 99.2), and 98.4% (97.0,
line with OS results (table). To illustrate the absolute difference we esti- 99.1) with chemotherapy; intermediate risk RS patients without known
mated 10-year OS and BCSS for a 60-year old woman with a pT1cN0(itc+) chemotherapy was 97.0% (96.2, 97.6), and 96.9% (95.9, 97.7) with
grade 3 TNBC. The predicted 10-year OS was 67% with chemotherapy and chemotherapy; high risk RS patients without known chemotherapy was
49% without; predictions for 10-year BCSS were 80% and 66%, re- 89.7% (85.4, 92.8), and 92.9% (90.4, 94.7) with chemotherapy.
spectively. Conclusions: Adjuvant chemotherapy is associated with higher Conclusions: Among a large cohort of patients identified in a population-
OS and BCSS in small node negative TNBC. Benefit is most evident in grade based cancer registry between 2004 and 2007, there was no statistically
3 tumors and tumors . 1cm and not evident in tumors #1cm and grade 1-2. significant difference in eight-year breast cancer-specific survival rates
among those who received chemotherapy and those who did not, regardless
aHR OS 95% CI aHR BCSS 95% CI
of RS risk group.
all patients 0.55 0.44-0.69 0.55 0.42-0.73
pT1ab 1.52 0.80-2.90 1.17 0.55-2.49
pT1c 0.53 0.41-0.67 0.57 0.43-0.76
grade 1-2 1.03 0.63-1.67 0.99 0.57-1.71
grade 3 0.50 0.39-0.65 0.54 0.40-0.74

538 Poster Session (Board #30), Sun, 8:00 AM-11:00 AM 539 Poster Session (Board #31), Sun, 8:00 AM-11:00 AM
Clinical behavior of recurrent hormone receptor-positive breast cancer by Eflapegrastim, a novel and potent long-acting GCSF for reducing
adjuvant endocrine therapy: A Breast International Group (BIG) 1-98 sub- chemotherapy-induced neutropenia: Integrated results from two phase III
analyses. First Author: Jose Pablo Leone, Dana-Farber Cancer Institute, trials in breast cancer patients. First Author: Lee S. Schwartzberg, University
Boston, MA of Tennessee Health Sciences Center, Memphis, TN
Background: Endocrine therapy resistance is a major cause of distant re- Background: Eflapegrastim (E) is a novel long-acting GCSF comprised of
currence (DR) in HR+ breast cancer. Currently, no data exists evaluating recombinant human GCSF covalently linked to human IgG4 Fc fragment
differences in clinical behavior after DR between patients (pts) treated in via a PEG linker (MW, 72 kDa). E showed increased potency vs pegfilgrastim
the adjuvant setting with different endocrine therapy regimens. The aim of (P) in preclinical and Phase I and II trials. Two identically designed Phase III
this study was to analyze post-DR survival of pts treated on BIG 1-98. pivotal trials (NCT02643420, NCT02953340) were conducted globally
Methods: BIG 1-98 compared 5 years of adjuvant treatment between 4 arms: with a fixed dose of 13.2 mg E containing 3.6 mg GCSF to evaluate E vs P
tamoxifen (T), letrozole (L), T followed by L (TL) and L followed by T (LT). (6 mg) in pts receiving chemotherapy for early-stage breast cancer.
After 8.1 years median follow-up (follow-up through 2010), 911 (T n = 302, Methods: Each open-label trial randomized pts 1:1 to a single subcutaneous
L n = 285, TL n = 170, LT n = 154) of 8010 pts had DR as site of first dose of E 13.2 mg/0.6 mL or P 6 mg/0.6 mL on Day 2 of each of four 21-day
recurrence. Univariate and multivariable Cox analyses were performed to cycles following Day 1 adj/neoadj docetaxel 75 mg/m2 + cyclophosphamide
determine features associated with post-DR survival. With 661 total ob- 600mg/m2 (TC). The primary endpoint was to demonstrate E non-inferiority
served deaths, statistical power was 0.8 to detect a hazard ratio (HaR) (NI) to P as measured by mean duration of severe neutropenia (DSN) in Cycle
. 1.24 at the 2-sided 0.05 level of significance. Results: Median follow up 1. Results: A total 643 intent-to-treat pts (314 E/329 P) with median age
time after DR was 59 months (IQR: 29-88). Among all pts with DR, 38.1% 60 yrs (24–88) were enrolled. Cycle 1 mean (SD) DSN was 0.24 (0.581) vs
were $ 65 years at study enrollment, 61.9% had tumor size . 2 cm, 69.7% 0.36 (0.789) days for E and P, confirming NI (p , .0001) and suggesting
were node positive. Neo/adjuvant chemotherapy was administered to 35.6% statistical superiority (p , .029). DSN NI was also shown across cycles 2–4.
of pts. There was no difference in post-DR survival by treatment arm (me- Among subgroups, including elderly ($65 yrs) and overweight ( . 75kg) pts,
dian survival: T 20.8, L 17.9, TL 17.3, LT 20.8 months; p = 0.21). In DSN was reduced for E vs P. In Cycle 1, E showed an absolute risk reduction
multivariable analysis, older pts (HaR 1.36; p = 0.0002), tumors . 2cm for severe neutropenia of 6.5% vs P (27.1% relative risk reduction, p ,
(HaR 1.2; p = 0.04), $ 4 positive nodes (HaR 1.32; p = 0.05) and PR- .043). Neutropenic complications (hospitalization and/or anti-infective use)
tumors (HaR 1.28; p = 0.001) had significantly worse post-DR survival. were 2.9% and 4.0% for E and P (p = ns). Incidence of FN was low for both E
Endocrine treatment arm, type of surgery, radiotherapy and neo/adjuvant and P, 1.6% vs 1.8% in Cycle 1 and 3.2% vs 3.0% overall. ANC profiles
chemotherapy were not associated with post-DR survival in the multi- showed sustained increased levels for E vs P in the recovery phase across all
variate model. Conclusions: Treatment with adjuvant T, L or their sequence cycles. Safety profiles were similar for E and P, including primarily for ex-
were not associated with differences in survival after DR. We observed sig- pected hematologic AEs and for bone pain and other musculoskeletal pain.
nificant differences in survival by primary tumor size, nodal and PR status, Conclusions: These integrated pivotal trial results confirm a similar safety
which suggest that traditional high-risk features remain prognostic in the profile and non-inferiority in reducing neutropenic risk for E at a lower GCSF
metastatic setting. Clinical trial information: NCT00004205. dose vs P. The data also suggests the potential for increased potency of E to
deliver improved clinical benefit, a possibility that warrants further clinical
trials. Clinical trial information: NCT02643420, NCT02953340.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Breast Cancer—Local/Regional/Adjuvant 19s

540 Poster Session (Board #32), Sun, 8:00 AM-11:00 AM 541 Poster Session (Board #33), Sun, 8:00 AM-11:00 AM
Factors associated with twenty-year (y) risks of breast cancer-specific Effects of Non-Linear or Linear Aerobic Training (AT) Dosing Regimens on
mortality (BCSM) in the Surveillance, Epidemiology, and End Results (SEER) Impaired Cardiovascular (CV) Function in Patients with Operable Breast
Registry. First Author: Jose Pablo Leone, Dana-Farber Cancer Institute, Boston, Cancer: A Randomized Controlled Trial (RCT). First Author: Jessica Scott,
MA Memorial Sloan Kettering Cancer Center, New York, NY
Background: Most reports describing the risk of late relapse in breast cancer Background: Breast cancer therapy causes marked impairments in CV
have been based on selected patients (pts) enrolled into clinical trials. The function predisposing to elevated risk of CV morbidity. We investigated the
aims of this study were to report on population-based long-term risks of effects of two AT dosing regimens on CV function in post-treatment patients
BCSM, and the risks of BCSM conditional on having survived 5 y. Addi- with operable breast cancer. Methods: In a three-arm, parallel-group RCT,
tionally, we aimed to identify factors associated with late deaths from breast 174 post-menopausal patients (2.8 years post primary adjuvant therapy)
cancer. Methods: Using SEER data, we identified women with breast cancer with impaired age/sex-matched peak oxygen consumption (VO2peak) were
(T1-T2, N0-N2, M0) between 1990-2005, with one primary cancer in their randomized to: (1) conventional linear AT (uniform dose-intensity / session),
lifetime, and known hormone receptor (HR) status. We used Kaplan-Meier (2) nonlinear AT (variable dose-intensity / session), or (3) stretching (attention
analyses to determine the effect of baseline variables on cumulative risks of control). AT consisted of 64 supervised treadmill walking sessions delivered
BCSM, we estimated annual rate of events per 100 person-years, and four times weekly at either ~70% VO2peak for 40 mins/session (linear) or
performed Cox regression stratified by HR status. Results: We included 55% to 100% VO2peak for 20-45 mins/session (nonlinear) for 16 consecutive
202,080 pts (median follow-up = 12.25 y). Of all breast cancer deaths, the weeks. Stretching was matched to AT on the basis of location, frequency,
proportion after 5 y was 65% for HR+ vs 28% for HR- (p , 0.001). The table duration, and treatment length. The primary end point was change in VO2peak.
shows risks of BCSM by HR and N status, and annual event rates. The Secondary end points were other markers of CV risk profile (biochemical CV
cumulative risk of BCSM in y 5-20 ranged from 7.9% in HR-N0 to 38% in risk profile, cardiac function, body composition), patient-reported outcomes
HR+N2. Among HR+ pts, adjusted risks of BCSM conditional on having (PROs), tolerability (e.g., relative dose intensity), and safety. All analyses
survived 5 y were higher for T2 vs T1a (Hazard ratio [HzR] 3.3, p , 0.001), followed the intention-to-treat principle. Results: Rates of lost-to-follow
N2 vs N0 (HzR 3.5, p , 0.001), age at diagnosis (dx) . 64 y vs , 50 y (HzR were , 10% in all arms. Relative dose intensity of AT was 73% 6 27%
1.4, p , 0.001), black race vs white (HzR 1.3, p , 0.001) and grade III vs I and 80% 6 21% in linear and nonlinear arms, respectively. No serious
(HzR 2.3, p , 0.001). For HR- pts, adjusted risks of BCSM conditional on adverse events were observed. In adjusted analysis, compared to control,
having survived 5 y were higher for T2 vs T1a (HzR 2.0, p , 0.001), N2 vs N0 VO2peak (ml O.2kg-1.min-1) increased 0.7 (6 0.3) ml O.2kg-1.min-1 (p = 0.06)
(HzR 2.8, p , 0.001) and age at dx . 64 y vs , 50 y (HzR 1.6, p , 0.001). and 0.8 (6 0.4) ml O.2kg-1.min-1 (p = 0.02) in linear and nonlinear AT,
Conclusions: For HR+ breast cancer, risks of BCSM remain high beyond 5 y respectively. Rates of VO2peak improvement greater than the technical error
from dx and depend on T-N status, age, race and grade. In HR- breast cancer, of measurement (i.e., $1.32ml O.2kg-1.min-1) were 33% and 39% in linear
the risk of BCSM is highest within 5 y from dx; however, risks beyond 5 y are and nonlinear AT (p = 0.03), respectively. Both AT regimens were associated
still considerable and depend on T-N status and age. Our results underscore with improvements in several secondary CV end points but only nonlinear
the need for better adjuvant therapies in both HR+ and HR- breast cancer. AT improved PROs compared with control (all p’s , 0.05). There were no
Annual rate (%) Cumulative risk (%)
differences between the two AT regimens. Conclusion: AT significantly
improves CV function and PROs in post-treatment breast cancer patients. The
No. % Event-Free at 5 y y 5- < 10 y 10-20 y 5-20 y 0-20
efficacy-tolerability ratio favors the non-linear regimen over the conventional
HR+ N0 110,108 97.3 0.7 0.6 9.9 12.4 linear prescription approach. Clinical trial information: NCT01186367.
HR+ N1 40,015 92.8 1.9 1.5 21.9 27.5
HR+ N2 11,167 83.5 4.4 2.8 38 48.3
HR- N0 27,014 89.4 0.9 0.3 7.9 17.7
HR- N1 10,191 76.1 1.7 0.5 12.2 33.2
HR- N2 3,585 61.8 3.1 0.8 19.9 50.6

542 Poster Session (Board #34), Sun, 8:00 AM-11:00 AM 543 Poster Session (Board #35), Sun, 8:00 AM-11:00 AM
A novel 95-gene signature (Curebest 95GC Breast) that predicts recurrence- Distinct mutational landscape between HR+ and HR- HER2+ early-stage
risk in patients with ER-positive, HER2-negative, node-negative, early-stage breast cancer patients. First Author: Bo Chen, Guangdong General Hospital
primary invasive breast cancer with an intermediate Oncotype DX Recurrence & Guangdong Academy of Medical Sciences, Department of Breast Cancer,
Score. First Author: Takeo Fujii, Department of Breast Medical Oncology, The Guangzhou, China
University of Texas MD Anderson Cancer Center, Houston, TX
Background: HER2 targeted therapy has revolutionized the survival outcomes
Background: The TAILORx trial demonstrated that adjuvant endocrine and of early and advanced HER2+ breast cancer (BC). However, among HER2+
chemoendocrine therapies had similar efficacy in patients with hormone patients, the therapeutic response to HER2 inhibitors vary. To understand the
receptor-positive, HER2-negative, node-negative breast cancer with an molecular mechanism of the variability in therapeutic efficacies, the muta-
Oncotype DX recurrence score (RS) of 11-25. However, a predictive strategy tional landscape of HER2+ tumors need to be elucidated. Methods: 107
is needed to identify patients with intermediate RS who may benefit from HER2+ Chinese stage I-III BC patients were included in the study, including
adjuvant chemoendocrine therapy. Curebest 95GC Breast (95GC) is a 95- 64 HR+ and 43 HR- patients. A majority of the patients were diagnosed with
gene signature that can stratify patients into two groups with high (95GC-H) infiltrating ductal carcinoma (99/107). Capture-based targeted sequencing
and low (95GC-L) groups to predict the risk of recurrence. Our primary was performed using a panel consisting of 520 cancer-related genes spanning
objective was to show that 95GC can classify patients with intermediate RS 1.64 MB of the human genome. Results: 1,119 alterations were detected,
into binary recurrence risk groups. Methods: Patients with ER-positive, including 478 single nucleotide variants (SNVs), 14 insertions or deletions,
HER2-negative, node-negative invasive breast cancer and RS 11-30 who 29 fusions, 593 copy number amplifications (CNA), 2 large genomic rear-
underwent definitive surgery and adjuvant endocrine therapy were included. rangements and 3 CN deletions in 267 genes. Alterations in 99 genes were
RNA was derived from archived formalin-fixed, paraffin-embedded sam- shared between HR+/HER2+ and HR-/HER2+ tumors; while 123 and 45
ples, and 95GC was calculated as reported previously. The Fisher exact and genes were only detected in either HR+/HER2+ or HR-/HER2+ tumors, re-
Brunner-Munzel tests were used to compare variables between 95GC groups. spectively. CNA, splice site and frameshift mutations were significantly more
A Kaplan-Meier estimate with a log-rank test was used for recurrence-free in HR+/HER2+ patients (p= 0.017). Specifically, CNA in SPOP, CCND1,
survival (RFS) analysis. Results: The analysis included 178 patients from five FGF19, FGF3, FGF4, RNF43, RAD51C, ADGRA4 and MDM4 and various
institutions. The 5-year RFS rate in patients with RS 18-30 was higher in the alterations in GATA3 were significantly more among HR+/HER2+ tumors (p,
95GC-L group (n = 129, 96.3%) than in the 95GC-H group (n = 49, 90.9%; 0.05). In addition to HER2 amplifications, concurrent fusions in ERBB2 (67%,
p = 0.002), which was consistent with results in an independent Japanese 4/6), SNVs in ERBB3 (100%, 3/3) and ERBB4 (100%, 1/1) were more likely to
population (n = 224; p , 0.001). RFS rates significantly differed between the be detected in HR+/HER2+ tumors, while concurrent EGFR amplifications were
groups among patients with RS 11-25 as well (95GC-L, 97.4%; 95GC-H, exclusively detected in HR-/HER2+ tumors. The trend of concurrent mutations
87.1%; p = 0.001). RFS rates did not differ between patients with RS 18-25 was consistent with mutation types detected in HER2- tumors based on HR
(94.8%) and those with RS 26-30 (93.8%; p = 0.33). Conclusions: 95GC can status, wherein EGFR amplifications were more frequent in HR-/HER2- tumors,
predict recurrence risk in patients with ER-positive, HER2-negative, node- while SNVs in EGFR, ERBB2, ERBB3 and ERBB4 were more predominant in HR+/
negative invasive breast cancer and intermediate RS. Further prospective HER2- tumors. Based on KEGG pathway analysis, HR+/HER2+ tumors had
retrospective studies in the TAILORx population are warranted to confirm that more frequent alterations in TGFb (p= 0.007), WNT (p= 0.002) and homologous
95GC can identify patients who may benefit from adjuvant chemoendocrine recombination (p= 0.004) pathways than HR-/HER2+ tumors. Furthermore, our
therapy. data revealed that HR+/HER2+ and HR-/HER2+ patients had comparable TMB
(p= 0.24), with a median TMB of 4.0 mutations/Mb for both. Conclusions: Our
study revealed genetic heterogeneity between HR+ and HR- HER2+ tumors. The
distinct genetic alterations are potentially relevant in the development of optimal
treatment strategies for such patients.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
20s Breast Cancer—Local/Regional/Adjuvant

544 Poster Session (Board #36), Sun, 8:00 AM-11:00 AM 545 Poster Session (Board #37), Sun, 8:00 AM-11:00 AM
PAM50 HER2-enriched subtype as an independent prognostic factor in The prognostic role of progesterone receptor in patients with rapidly proliferating,
early-stage HER2+ breast cancer following adjuvant chemotherapy plus hormone receptor-positive early breast cancer. First Author: Sara Bravaccini,
trastuzumab in the ShortHER trial. First Author: Pier Franco Conte, Istituto Scientifico Romagnolo per lo Studio e la Cura dei Tumori (IRST)
Department of Surgery, Oncology and Gastroenterology, University of Padova, IRCCS, Meldola, Italy
Medical Oncology 2, Istituto Oncologico Veneto IRCCS, Padova, Italy
Background: The prognostic role of progesterone receptor (PgR) in highly
Background: We investigated the prognostic role of the PAM50 HER2- proliferating early breast cancer (BC) is not well established. We retrospectively
enriched (HER2-E) subtype in HER2+ early breast cancer enrolled in the explored this biomarker in a cohort of patients with highly proliferating tumors
randomized Phase III ShortHER trial. Methods: The ShortHER study ran- enrolled in a phase III trial of adjuvant therapy. Methods: 1066 patients with
domized 1254 HER2+ early breast cancer patients to receive 9 weeks vs 1 N- or 1-3 N+ BC were randomized to receive: epirubicin followed by CMF, CMF
year of adjuvant trastuzumab combined with chemotherapy. Gene expression followed by epirubicin, or CMF alone. Rapidly proliferating tumors were defined
measured using nCounter platform was available for 438 surgical samples. by thymidine labeling index (TLI) . 3% or histological grade 3 or S-phase . 10%
Intrinsic subtyping was determined using the research-based PAM50 pre- or Ki67 . 20%. We analyze the subgroup of 466 patients with hormone receptor
dictor. Metastasis-free survival (MFS) was calculated from randomization to (HR)-positive tumors treated with sequential epirubicin/CMF regimens followed
distant disease recurrence or death (median follow up 72 months). Uni- and by tamoxifen and for whom immunohistochemical assessments of estrogen
multi-variable analysis were performed using Cox models. Results: PAM50 receptor (ER), PgR, HER2 and Ki67 were available. Disease-free (DFS) and
subtype distribution was: HER2-E 53% (N = 233), Luminal A 20% (N = 87), overall survival (OS) curves were built with the Kaplan–Meier estimator and
Luminal B 10% (N = 43), Normal-like 11% (N = 48) and Basal-like 6% (N = compared by logrank test and Cox regression models. Results: PgR expression
27). HER2-E subtype was associated with hormone receptor-negative status was significantly associated with ER expression, HER2 status, age and meno-
(p , 0.001) and TILs $20% (p , 0.001), but not with stage and age ( , or pausal status, but not with Ki67, tumor size and nodal status. PgR cutoff values
$60 yrs). HER2-E subtype was associated with worse MFS vs other PAM50 of 10% and 20% were chosen based on a Receiver Operating Characteristics
subtypes overall (HR 2.78, p = 0.001), in the short (HR 2.24, p = 0.046), and analysis and the literature data. DFS and OS figures at 5 and 10 years, as well as
in the long arm (HR 4.04, p = 0.011). Multivariable Cox model confirmed the the relative hazard ratios, according to the different PgR cutoff values, are re-
independent prognostic value of HER2-E subtype (Table). HER2-E subtype ported in the table. Conclusions: Our results confirm the prognostic relevance of
added significant prognostic value on top of clinicopathological variables PgR expression in a cohort of patients with highly proliferating, HR-positive early
(Likelihood ratio test p , 0.001). Conclusions: HER2-E intrinsic subtype is an BC treated with adjuvant chemotherapy and endocrine therapy.
independent prognostic factor for HER2+ early breast cancer patients treated % 5-year DFS % 10-year DFS HR % 5-year OS % 10-year OS HR
with adjuvant chemotherapy and trastuzumab. Integration of PAM50 subtype n. pts (95% CI) (95% CI) (95% CI) p (95% CI) (95% CI) (95% CI) p
in prognostic algorithms can help refine risk stratification. These findings Overall 466 85 70 - - 94 85 - -
warrant independent validation. Clinical trial information: NCT00629278. Cutoff
(81-88) (65-75) (92-96) (81-89)

10%
Univariate Cox Models for MFS Univariate Cox Model for MFS PgR ‡10 338 89 73 1.00 96 88 1.00
Factors HR (95% CI) p HR (95% CI) p (85-92) (67-79) (93-98) (83-92)
PgR < 10 128 78 64 1.48 0.045 89 79 1.84 0.023
Stage III vs I-II 4.33 (2.50-7.51) , 0.001 4.07 (2.33-7.13) , 0.001 (70-85) (55-73) (1.01- (84-95) (71-87) (1.09-
TILs 10% increments 0.63 (0.43-0.91) 0.013 0.55 (0.37-0.81) 0.003 2.18) 3.11)
Intrinsic subtype: HER2E vs other 2.78 (1.49-5.22) 0.001 3.49 (1.84-6.61) , 0.001 Cutoff
Hormone-receptor negative vs positive 1.08 (0.60-1.94) 0.804 20%
PgR ‡20 293 89 74 1.00 95 89 1.00
Grade 3 vs 1-2 1.97 (0.96-4.05) 0.605
(85-93) (67-80) (92-98) (84-93)
Age: ‡60 yrs vs 18-59 yrs 0.97 (0.55-1.72) 0.917 PgR < 20 173 71 65 1.51 0.030 92 80 1.87 0.018
(74-87) (56-73) (1.04- (88-96) (73-87) (1.11-
2.19) 3.16)

546 Poster Session (Board #38), Sun, 8:00 AM-11:00 AM 547 Poster Session (Board #39), Sun, 8:00 AM-11:00 AM
Adjuvant enzalutamide for the treatment of early-stage androgen receptor- Patterns of systemic treatment utilization in ER+/PgR+/HER2+, early-stage
positive (AR+) TNBC. First Author: Tiffany A. Traina, Memorial Sloan Ket- breast cancer (BC): An analysis of the National Cancer Database. First
tering Cancer Center, New York, NY Author: Zeina A. Nahleh, Cleveland Clinic Florida, El Paso, TX
Background: A subset of TNBC is dependent on AR signaling. Enzalutamide Background: The preferences and trends of treatment utilization of adjuvant
(ENZA), an AR-antagonist, has activity in patients (pts) with metastatic AR+ endocrine therapy (ET) versus chemotherapy (CH) for small node-negative
TNBC, with a clinical benefit rate of 33%. This study tests the feasibility of triple positive (TP) BC are unclear. We sought to determine these preferences
adjuvant ENZA for the treatment (tx) of early stage, AR+ TNBC. We now report and assess the impact on outcome. Methods: This is a retrospective study from
the primary endpoint (endpt) and safety. Methods: Eligible pts have centrally the National Cancer Database including patients with TP stage I BC, 2004-
confirmed, Stage I-III, ER/PR , 1%, HER2(-), AR $1% BC and completed all 2015. Treatment selection was evaluated for association with patient clinical
planned surgery, chemotx and radiation (RT) , 6 months of tx start. AR testing and demographic characteristics using logistic regression. Overall survival (OS)
by IHC per MSK methods. Tx consists of ENZA 160mg daily for 1 year (y) with was estimated using the Kaplan-Meier method and compared among patient
the option to extend tx to 2y. Toxicity per NCI CTCAEv4 every (q) 4 weeks (wk) for and treatment cohorts by log-rank test and Cox regression. Results: Of 37,777
12 wk, then q3 months. Primary endpt: feasibility of 1y ENZA defined as the patients analyzed, 79% were White (Non-Hispanics), 10% African Americans,
discontinuation rate due to toxicity, consent withdrawal or tolerability. 50 pts are and 5% Hispanic/Latinos. 57% were 50-70 years old. 86% received adjuvant
enrolled to have 46 evaluable pts required to discriminate between feasibility of endocrine therapy versus 14% CH first. Around 40 % of all patients received
50% and 70%, with type I error 5% and 88% power. Pts who have disease anti-Her2 therapy. Patients younger than 70 years, with male BC, diagnosed
progression (PD) or die during 1st y of ENZA and do not have tx discontinuation with poorly differentiated BC, African Americans and Hispanics were more
due to the above will not be included in the primary analysis. If 29 pts complete likely to be treated with chemotherapy. OS rate at 5-year was 92.3% (95% CI:
1y, adjuvant ENZA will be deemed feasible. Secondary endpts: safety and 0.918-0.928). In multivariate analysis for patients with survival data, an
3y DFS and OS. Exploratory endpts: PROs and biomarker development. increased rate of death was associated with: treatment in community versus
Results: Between 5/2016-6/2018, 50 pts were enrolled. Pt and tumor char- academic/research centers, CH first versus ET, no treatment with anti-Her2
acteristics (N = 50): Median age 55y (33-81); Stage: I 20 (40%), II 23 (46%), therapy, government versus private /no insurance, Native American ethnicity. A
III 7 (14%); Grade (gr): 2 = 26%, 3 = 74%. AR . 10% = 35 (70%), AR #10% = slight but statistically significant reduction in the in the risk of death at 5 years
15 (30%). Chemotx 47/50 (94%): Neoadjuvant (neo) 40%, Adjuvant (adj) was evident for patients receiving anti-Her2 therapy plus ET therapy, 5-year OS
60%; Anthracycline/Taxane-based 38/47 (81%), Platinum 1/47 (2%), 93.5% (CI: 89.2-98%), when compared to patients receiving anti-Her2 therapy
Docetaxel/Cyclophosphamide 3/47 (6%), other 5/47 (11%). 13/19 who re- plus CH 92.7 % (CI: 89.4-96). Conclusions: This study provides real world data
ceived neo tx failed pCR; 9/13 (69.2%) received adj capecitabine. RT: 38/50 of common practices in the US . The majority of patients with node negative
(76%). 27 pts completed 1y of tx. 7 pts will be evaluable by 6/1/19. 1 pt to Stage I, ER+/PR+/Her2+ BC received adjuvant ET and anti-Her2 therapy, not
complete 1y 6/21/19. 15 pts are off tx: PD (3), toxicity (5), noncompliance chemotherapy. These patients had a similar to slightly improved 5 year- survival
(4), withdrawal of consent (3). Tx-related AEs, any gr, . 10% (N = 50): fatigue when compared to anti-Her2 therapy plus CH, supporting the use ET plus anti-
(48%), hot flashes (22%), headache (18%), hyperglycemia (18%), nausea Her2 therapy in this setting. Future studies should focus on better selecting
(18%), WBC decreased (16%), dizziness (14%), arthralgia (12%), dyspnea (12%). patients with hormone receptor positive and Her 2 + early stage BC who would
Tx-related, gr 3 AEs: fatigue (6%), hyperglycemia (2%), hypertension (2%). No gr 4/ benefit from adjuvant CH. Disparity in outcome also warrants further evaluation.
5 AEs or seizures. 11 pts had dose reduction. Conclusions: Feasibility of adjuvant
ENZA will be fully evaluable in 4/2019 and is anticipated to meet the prespecified Level Total N Number of events 5 years OS Years (95%CI) P-value
statistical expectations for primary endpt. ENZA is well tolerated following CH , no Anti Her2 2333 159 91.2 (89.7, 92.7) ,0.0001
locoregional tx and standard of care systemic tx. Secondary analyses and correl- CH + Anti-Her2 1335 56 92.7 ( 89.4, 96.1)
atives are ongoing to define the role of AR in TNBC. Clinical trial information: ET, no Anti-Her2 17160 904 92.2 (91.6, 92.7)
ET + Anti-Her2 7728 93 93.5 ( 89.2, 98)
NCT02750358.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Breast Cancer—Local/Regional/Adjuvant 21s

548 Poster Session (Board #40), Sun, 8:00 AM-11:00 AM 549 Poster Session (Board #41), Sun, 8:00 AM-11:00 AM
Effect of prophylaxis on neratinib-associated diarrhea and tolerability in Oncotype DX testing in early-stage node-positive breast cancer and impact
patients with HER2+ early-stage breast cancer: Phase II CONTROL trial. First on chemotherapy use at a comprehensive cancer center. First Author: Katya
Author: Carlos Hernando Barcenas, The University of Texas MD Anderson Losk, Department of Medical Oncology, Dana-Farber Cancer Institute,
Cancer Center, Houston, TX Boston, MA
Background: CONTROL (clinicaltrials.gov: NCT02400476) is an open-label, Background: The 21-gene Oncotype DX Recurrence Score (RS) is widely
sequential-cohort, phase II study investigating the effectiveness of prophylaxis used to guide adjuvant chemotherapy decisions in hormone receptor positive
or dose escalation in preventing diarrhea and improving tolerability of ner- (HR+), HER2-negtive (HER2-), lymph node negative (LN-) breast cancer. It’s
atinib, an irreversible pan-HER tyrosine kinase inhibitor. Neratinib has dem- adoption in lymph node positive (LN+) disease remains controversial. In 2016,
onstrated benefit as an extended adjuvant therapy for early-stage HER2+ breast we implemented ‘reflex’ RS testing for patients #65 years with HR+/HER2-
cancer. Various prophylactic agents are being studied in adult patients treated breast cancer including T1/T2 N1 (grade 1 or 2) tumors. Providers can also
with oral neratinib (240 mg/day for 1 year) after trastuzumab-based ad- order Oncotype DX outside of reflex criteria. We sought to assess RS distri-
juvant therapy. Methods: Patients (n = 485) were enrolled into cohorts bution and factors associated with chemotherapy use in HR+/HER2-/LN+
investigating antidiarrheal prophylaxis for 1–2 cycles (28 days) with the breast cancer patients at our center. Methods: Patients with non-metastatic
following agents: loperamide (n = 137); loperamide + budesonide (n = 64); HR+/HER2-/LN+ breast cancer who underwent primary surgery at our center
loperamide + colestipol (n = 136); loperamide prn + colestipol (n = 104). were identified from our prospective database. We examined the distribution of
An additional cohort assessing loperamide prn + neratinib dose escalation low (RS , 18), intermediate (RS 18-30) and high (RS . 30) RS and identified
with no mandatory prophylaxis (n = 44) is ongoing. Adverse events were characteristics for those who did not meet reflex criteria. A multinomial logistic
graded according to NCI-CTCAE v4.0. The primary endpoint was incidence regression model was performed to identify factors associated with chemo-
of grade $3 diarrhea. Data cut-off: 14 Jan 2019. Results: As shown in the therapy receipt among all LN+ patients. Results: From 1/2016-3/2018, we
table, all prophylactic regimens reduce the incidence of Grade 3 diarrhea identified 296 consecutive patients with HR+/HER2-/LN+ breast cancer. 200
and drug discontinuation compared with the prior ExteNET trial [Martin (68%) patients had RS testing and 128 (64%) met reflex criteria. Reasons for
et al. 2017]. The median cumulative duration of Grade 3 or higher diarrhea not meeting RS reflex criteria included age . 65 (n = 35), grade III disease (n =
spanned from 2.0 to 3.5 days across regimens for the entire treatment period. 35) and N2/N3 tumors (n = 10). Among the 200 patients with RS, 122 (61%)
No Grade 4 diarrhea was reported. Conclusions: The addition of budesonide or had RS , 18, 67 (34%) had RS 18-30, and 11 (6%) had RS . 30. Only 68/
colestipol to loperamide prophylaxis given for 1–2 cycles reduces the severity 200 (34%) patients with RS received chemotherapy as compared to 54/96
and duration of diarrhea in patients treated with neratinib, thereby improving (56%) patients without RS (p = 0.0004). Compared to patients without RS
tolerability. Updated data for the dose-escalation cohort will be presented at testing, the odds of receiving chemotherapy were less with RS , 18 (OR =
the meeting. Clinical trial information: NCT02400476. 0.46). The odds of receiving chemotherapy were greater with $3 positive LNs
Loperamide + Loperamide + Loperamide Neratinib dose
versus 1 positive LN (OR = 3.40). Conclusions: The majority of HR+/HER2-/LN+
Loperamide budesonide colestipol prn + colestipol escalation patients undergoing upfront surgery at our center receive RS testing (200/296),
(n = 137) (n = 64) (n = 136) (n = 104) (n = 44) with 122 (61%) resulting in low risk RS. Patients with low risk scores (RS , 18)
Diarrhea, % were less likely to receive chemotherapy. While nodal involvement remains a
Grade 1 24.1 25.0 27.2 29.8 43.2
Grade 2 24.8 32.8 34.6 32.7 25.0 common driver of chemotherapy use, our study demonstrates that RS testing
Grade 3 30.7 28.1 20.6 31.7 13.6 provides clinically useful information in this population.
Diarrhea leading to 20.4 10.9 3.7 5.8 2.3
discontinuation, %
Hospitalization (due 1.5 0 0 0 0
to diarrhea), %
Discontinuation of study 44.6 20.3 28.7 26.0 13.6
treatment (any cause), %

550 Poster Session (Board #42), Sun, 8:00 AM-11:00 AM 551 Poster Session (Board #43), Sun, 8:00 AM-11:00 AM
Prognostic value of PD-L1 gene expression with Recurrence Score and 70-gene Prediction of neoadjuvant chemotherapeutic efficacy by CTC and cfDNA in
signature in patients with ER+/HER2- early breast cancer. First Author: Ioannis patients with locally advanced breast cancer. First Author: Ge Ma, The First
Zerdes, Karolinska Institutet, Stockholm, Sweden Affiliated Hospital of Nanjing Medical University, Nanjing, China
Background: We have previously demonstrated that PD-L1 mRNA expression Background: Neoadjuvant chemotherapy (NCT) is the standard treatment for
can serve as prognostic biomarker in breast cancer (BC). In ER+/HER2- BC, RS patients with locally advanced breast cancer (LABC). Liquid biopsy, in-
and 70-gene signature are used to predict the risk of recurrence and benefit cluding circulating tumor cells (CTCs) and cell free DNA (cfDNA) represent
from chemotherapy. Methods: Discovery cohort (cohort 1) included 302 pa- an important paradigm shift in precision medicine. The aim of this study was
tients diagnosed with primary ER+/HER2- BC (1997-2005) in Stockholm to estimate the value of CTCs and cfDNA in efficacy prediction of the re-
health care region. Gene expression profiling has been performed using sponse to NCT in patients with LABC. Methods: Patients with LABC received
DNA microarrays (GSE48091) while information regarding tumor char- EC4-T4 regimen NCT. CTCs and cfDNA obtained at time of biopsy, after first
acteristics, treatment and follow-up have been obtained. TCGA’s dataset course of NCT and after the last course of NCT. All patients were divided into
including 590 ER+/HER2- patients, was used as validation cohort (cohort two groups according to pathological reactivity. A novel SE-iFISH strategy,
2). Kaplan–Meier estimates and Cox regression univariate/multivariable improved for detection of CTCs, was applied. CTCs(CD45-/CD31-) with dif-
analyses were performed using breast cancer-specific survival(BCSS) and ferent cytogenetic abnormality of aneuploid chromosome 8 and small cell size
progression-free interval (PFI) as endpoints in cohorts 1 and 2, respec- CTCs (#5 mm of WBCs) were analyzed separately in LABC patients subjected
tively. Gene signature scores were calculated using the R genefu package. to NCT for the first time. Plasma DNA biomarkers ALU 111 and ALU 260
Likelihood ratio (LR) tests and concordance indices (c-indices) were used elements were evaluated using qRT-PCR. DNA integrity was calculated relative
to assess each score’s added prognostic value. Results: PD-L1 mRNA to the breast cancer cell line MCF-7. Clinical significance of diverse subtypes of
expression (treated as a continuous variable) was independently associ- CTCs and cfDNA was systematically investigated. Results: A total of 45 pa-
ated with better BCSS in cohort 1 (HR = 0.72; 95% CI = 0.58-0.90;p = tients was enrolled in this study. According to the therapy response, 6/45
0.003) and with better PFI in cohort 2 (HR = 0.67; 95% CI = 0.50-0.90; patients had high response (High-R) and 39/45 patients had low response
p = 0.008) in the multivariable analysis. PD-L1 provided significant ad- (Low-R). There were no significant differences in CTC number and small cell
ditional prognostic information beyond that of both RS alone (LR-Dx2= 9.6; size CTC number between High-R and Low-R groups in all three detections.
p = 0.002 and LR-Dx2= 9.7; p = 0.002, in cohorts 1 and 2, respectively), However, the CTC number kept stable in the High-R group, but increased
and 70-gene signature score alone (LR-Dx2= 10.4; p = 0.001 and LR- continually during NCT in Low-R group. In 45 patients, the percentage of CTCs
Dx2= 9.2; p = 0.002 in cohort 1 and 2, respectively). C-indices for PD-L1 + with trisomy 8, which were related to cancer metastasis, incresed in the Low-R
RS vs RS were 0.65 vs 0.60 (cohort 1) and 0.66 vs 0.60 (cohort 2), and for group at the third dectection. The concentration of cfDNA in all three de-
PD-L1 + 70-gene vs 70-gene were 0.65 vs 0.59 (cohort 1) and 0.64 vs tections did not indicate outcome of NCT. However, concentration of ALU 111
0.54 (cohort 2), respectively. Conclusions: PD-L1 gene expression was increased in Low-R patients during NCT. In High-R patients, no significant
correlated with better outcomes and can provide added prognostic value to increase was observed. A CTC and cfDNA panel were constructed to dis-
RS and 70-gene signature scores in ER+/HER2- BC. criminate High-R patients from Low-R patients. The areas under the receiver-
operating characteristic (ROC) curves of the pannel for the three times were
0.803, 0.859 and 0.667, respectively. DNA integrity index(CFDI) was sig-
nificantly higher in High-R group than Low-R after first course of NCT. The
areas under the ROC curves of the CFDI for the three times were 0.675, 0.863
and 0.697, respectively. Conclusions: The trend of cfDNA concentration
changed resembled to the number of CTCs, small cell size CTCs and triploid
CTCs during NCT, and could predict tumor response to ongoing treatment.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
22s Breast Cancer—Local/Regional/Adjuvant

552 Poster Session (Board #44), Sun, 8:00 AM-11:00 AM 553 Poster Session (Board #45), Sun, 8:00 AM-11:00 AM
Factors associated with osteonecrosis of the jaw in women with breast cancer Tailored dose-dense chemotherapy in combination with trastuzumab as
receiving high-dose bisphosphonates to prevent breast cancer metastases as adjuvant therapy for HER2-positive breast cancer: A secondary analysis
part of the SWOG 0307 trial. First Author: Darya Kizub, Everett Clinic, of the phase III PANTHER trial. First Author: Theodoros Foukakis, Karolinska
Everett, WA Institutet and University Hospital, Stockholm, Sweden
Background: Bisphosphonates reduce the risk of bone metastases in post- Background: Dose-dense (DD) adjuvant chemotherapy improves outcomes
menopausal women with early-stage breast cancer but carry the risk of in early breast cancer (BC). However, there are no data to inform on the
osteonecrosis of the jaw (ONJ). We used the data collected in the S0307 trial combination of DD chemotherapy with trastuzumab for patients with HER2-
to describe factors associated with provoked and unprovoked ONJ. positive disease. Methods: This is a protocol predefined secondary analysis
Methods: In S0307, 6097 patients with Stage I-III breast cancer who had of the randomized phase 3 PANTER trial. Women 65 years old or younger
surgery were randomized to receive zoledronic acid (ZA) 4mg IV monthly for with node-positive or high-risk node negative BC were randomized 1:1 to either
6 months, then every 3 months, clodronate (CL) 1600mg daily, or ibandronate tailored according to hematologic nadirs and DD epirubicin/cyclophosphamide
(IB) 50mg daily for three years, with no difference in bone metastases or (4 cycles) followed by 4 cycles of docetaxel (tDD EC/D) or standard 3-weekly
disease-free survival. Patients completed dental procedures prior to and had a 5-fluorouracil/E/C (3 cycles) and D (3 cycles); Patients with HER2-positive
dental exam at enrollment. Pearson’s Chi-squared and Student’s T-test were disease received 1 year of adjuvant trastuzumab. In addition, HER2-positive
used to test differences in categorical and continuous variables, respectively; and an equal number of matched for age, treatment group and institution,
logistic regression was used test independent association. Results: Of 5836 HER2-negative patients that were enrolled in Swedish sites were enrolled in a
evaluable women, 48 developed ONJ, which was associated with bisphospho- predefined study of cardiac safety and underwent echocardiography or MUGA
nate type (28/2124 (1.26%) for ZA, 8/2185 (0.36%) for CL and 12/1527 and electrocardiography at baseline and at four and six years of follow-up. The
(0.77%) for IB (p = 0.002). Median time to onset of ONJ was 24.9 (1.4-66.6) for primary endpoint was BC relapse-free survival (BCRFS). Results: There were
ZA, 41.2 months (range 33.8-67.4) for CL, 23.9 (2.1-75.3) for IB (p = 0.0447). 342 HER2-positive patients; 335 received at least one dose of trastuzumab,
Infection was present in 21 (43.8%) and absent in 20. ONJ was considered while 29 patients discontinued trastuzumab prematurely. BCRFS was not
unprovoked in 20 (41.7%) and provoked by dental extraction in 20 (41.7%), statistically significantly in favor of tDD EC/D (HR = 0.68, 95% CI 0.37 – 1.27,
periodontal disease in 14 (29.2%), denture trauma in 6 (12.5%), other dental P= 0.231). Cardiac outcomes after four and six years of follow-up did not differ
surgery in 3 (6.3%). ONJ was associated with dental calculus (OR 2.03 (95% CI: significantly between HER2-positive and HER2-negative patients, nor be-
1.08-3.81), gingivitis (OR 2.11, 95% CI: 1.12-3.98), and periodontal disease tween tDD and standard treatment. Conclusions: To our knowledge, these are
(OR 2.87, 95% CI 1.45-5.53) that were moderate/severe and . 4mm peri- the only data on combining DD adjuvant chemotherapy and trastuzumab in
odontitis (OR 2.20, 95% CI 1.18-4.08). Patients with provoked and unprovoked BC. The combination decreased the risk for BC relapse by 32% compared to
ONJ had similar amounts of dental disease and lesion location. ONJ was not standard treatment, a statistically non-significant difference. Its efficacy and
associated with dentures, plaque, chemotherapy, corticosteroids or renal adverse safety merit further evaluation as part of both escalation and de-escalation
events. In multivariate analysis (limited by small sample), only bisphosphonate strategies. Clinical trial information: NCT00798070.
type was associated with ONJ. Conclusions: ONJ prevalence was low, likely due
to patients completing dental procedures before enrollment. ZA carried a higher
risk of ONJ (though current adjuvant dosing interval recommendations are less
frequent), with time to onset similar to IB. Oral CL and IB (not currently available
in the United States) are thus likely more appropriate for patients with poor dental
health. Clinical trial information: NCT00127205.

554 Poster Session (Board #46), Sun, 8:00 AM-11:00 AM 555 Poster Session (Board #47), Sun, 8:00 AM-11:00 AM
Genomics of HER2+ breast cancer in young women before and after Comparative performance of Breast Cancer Index (BCIN+) and CTS5 in
exposure to chemotherapy (chemo) plus trastuzumab (H). First Author: hormone receptor-positive (HR+) lymph node-positive (N+) breast cancer
Adrienne Gropper Waks, Dana-Farber Cancer Institute, Boston, MA patients with one to three positive nodes (N1). First Author: Dennis Sgroi,
Center for Cancer Research, Massachusetts General Hospital, Boston, MA
Background: HER2+ breast cancer (BC) is particularly common in young
women. Genomic features of HER2+ tumors before and after H-based therapy Background: Identification of N+ breast cancer patients with a limited risk of
have not been described in a population of young women and may point to recurrence improves selection of those for which chemotherapy and/or ex-
clinically targetable mechanisms of resistance. Methods: From a large prospective tended endocrine therapy (EET) may be most appropriate to reduce over-
cohort of women diagnosed with BC age #40 years, we identified those with HER2+ treatment. BCIN+ integrates gene expression with tumor size and grade, and
BC and tumor tissue available for sequencing before and after chemo+H. Whole is highly prognostic for overall (0-10yr) and late (5-10yr) distant recur-
exome sequencing (WES) was performed on each tumor and on germline DNA rence (DR) in N1 patients. Clinical Treatment Score post-5-years (CTS5) is a
from blood. Tumor-normal pairs were analyzed for mutations and copy number prognostic model based on clinicopathological factors (nodes, age, tumor
(CN) changes. Evolutionary analysis was performed for patients with both pre- and size and grade) and significantly prognostic for late DR. The current analysis
post-treatment (tx) samples. Results: 22 women had successful WES samples compares BCIN+ and CTS5 for risk of late DR in N1 patients. Methods: 349
from at least one timepoint; 13 of these had paired sequencing results both before women with HR+, N1 disease and recurrence-free for $5 years were in-
and after chemo+H. For the majority of women, post-tx sample was following cluded. BCIN+ results were determined blinded to clinical outcome. CTS5
neoadjuvant chemo + H, though post-tx timepoint for other women represented
was calculated as previously described (Dowsett et al, JCO 2018; 36:1941).
locoregional or distant metastasis (Table). TP53 was the only gene that was
Kaplan-Meier analysis and Cox proportional hazards regression for late DR
significantly recurrently mutated in both pre- and post-tx samples. Comparison of
matched pre-tx and post-tx samples demonstrated that large changes in HER2 CN
(5-15y) were evaluated. Results: 64% of patients were . 50 years old, 34%
over the course of tx were uncommon, only 2/13 pts had . 2-fold change in HER2 with tumors . 2cm, 79% received adjuvant chemotherapy and 64% re-
CN. Other clonal and subclonal genomic alterations were found to be acquired in ceived up to 5 years of ET. BCIN+ stratified 23% of patients as low-risk
the post-tx sample compared to the pre-tx sample. One patient acquired a putative with 1.3% risk for late DR vs those classified as high-risk with 16.1% [HR
activating mutation in ERBB2. Another patient acquired a clonal hotpsot mutation in 12.4 (1.7-90.4), p = 0.0014]. CTS5 classified patients into 3 risk groups:
TP53. MYC gene amplification was observed in 4 post-tx tumors. NOTCH2 alter- 29% of patients as low-risk (4.2% DR), 37% as intermediate-risk (10.6%
ations were found in post-tx biopsies from 2 different patients, and mutations in STIL DR), and 34% as high-risk (22.1% DR) [HR intermediate vs. low: 2.3
were also found in post-tx biopsies from 2 patients, though the function of these (0.7-7.0), p = 0.16; high vs. low: 5.3 (1.8-15.5), p = 0.002]. In a subset of
mutations is not known. Conclusions: HER2+ breast tumors in young women display patients who completed 5 years of ET (N = 223), BCIN+ identified 22% of
genomic evolution following tx with chemo+H. HER2 CN changes are uncommon, patients as low-risk with a late DR rate of 2.1%, while CTS5 identified 29%
but we identified several genes that warrant exploration as potential mechanisms of and 37% of patients as low- and intermediate-risk with late DR rates of 5.2%
resistance to therapy in this population. and 10.3%, respectively. Conclusions: BCIN+ classified N1 patients into
binary risk groups and identified 20% patients with limited risk of late DR
Parameter # (%)
( , 2%) that may be advised to forego EET and its attendant toxicities/side
Age at dx Median 36 yo (range 27-39)
Stage at dx I 1 (5%) effects. In comparison, CTS5 classified patients into 3 risk groups, with low-
II 10 (45%) and intermediate-risk of late DR of 4-5% and 10%, wherein the risk-benefit
III 9 (41%)
IV 2 (9%)
profile for extension of endocrine therapy is less clear.
Receptor status at dx ER+ and/or PR+ 17 (77%)
ER- and PR- 5 (23%)
Disease status after chemo+H Surgery post-neoadjuvant tx 13 (59%)
Locoregional recurrence 2 (9%)
Metastatic 7 (32%)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Breast Cancer—Local/Regional/Adjuvant 23s

556 Poster Session (Board #48), Sun, 8:00 AM-11:00 AM 557 Poster Session (Board #49), Sun, 8:00 AM-11:00 AM
Immunologic responses in triple-negative breast cancer patients in a randomized Cardiac safety of the trastuzumab biosimilar ABP 980 in women with HER2-positive
phase IIb trial of nelipepimut-S plus trastuzumab versus trastuzumab alone to early breast cancer in the LILAC study. First Author: Hans-Christian Kolberg,
prevent recurrence. First Author: Jessica Campf, San Antonio Military Medical Marienhospital, Bottrop, Germany
Center, Fort Sam Houston, TX
Background: Although trastuzumab is generally well-tolerated, cardiotox-
Background: Breast cancer (BC) patients (pts) expressing low levels of HER2 icity is the main limitation in its use, leading to a severe heart failure in 2-4%
by (immunohistochemistry (IHC) 1-2+) are not eligible for trastuzumab (Tz). of patients in adjuvant trials. In the phase 3 LILAC trial, trastuzumab bio-
However, in a randomized phase 2b trial, triple negative BC (TNBC) pts similar ABP 980 demonstrated similar efficacy, safety, and immunogenicity
demonstrated a significantly better DFS with nelipepimut-S (NPS) + Tz vs Tz to trastuzumab reference product (RP) in women with HER2-positive early
alone. Here, we assess the ex vivo and in vivo immune responses (IR) in both breast cancer. Here we report analyses comparing cardiac safety of ABP
arms. Methods: Disease-free pts (n = 275) with HER2 IHC 1-2+, non-amplified 980 and RP. Methods: In the neoadjuvant phase, all 725 patients received
BC who were node positive and/or had TNBC were randomized 1:1 to 4 cycles of chemotherapy with epirubicin + cyclophosphamide Q3W and
granulocyte-macrophage-colony stimulating factor (GM-CSF) or NPS+GM- were then randomized 1:1 to ABP 980 (n = 364) or RP (n = 361) with a
CSF. 6NPS was given every 3 weeks x 6 followed by 4 boosters every loading dose of 8 mg/kg, then 6 mg/kg IV Q3W for 3 cycles plus paclitaxel
6 months (mo). All pts received Tz concurrently for 1 year per label regimen Q3W (4 cycles) or QW (12 cycles). After surgery, patients received inves-
and were followed for recurrence. IR were evaluated ex vivo by clonal tigational product (IP) Q3W for up to 1 year; ABP 980-treated patients
expansion of NPS-specific cytotoxic T lymphocytes (CTL) by dextramer- continued ABP 980, and RP-treated patients either continued RP (n = 190)
staining/flow cytometry at time points over 3 years. In vivo IR were assessed or switched to ABP 980 (RP/ABP 980; n = 171). AEs were assessed every
by delayed type hypersensitivity (DTH) reactions periodically. Results: The 3 weeks and cardiac safety every 3 months. Cardiac safety was monitored by
trial enrolled 97 TNBC pts; 60 had 4 timepoints available for analysis computerized 12-lead ECG; LVEF was assessed by 2D ECHO. LVEF decline
(37 NPS + Tz pts; 23 Tz pts). The NPS+Tz group exhibited increases in was defined as LVEF value decrease from study baseline by $10 percentage
CTL frequencies vs baseline: 208%, 303%, 379% at 18, 24 and 30 mo, points and to , 50%. Results: Treatment groups were well balanced with
respectively. NPS+Tz pts’ mean CTL frequencies increased from 0.029 60.001% regard to IP disposition. Over the entire study, 22 (3.1%) patients had LVEF
at baseline to 0.11260.026% at 30 mo (p = 0.01) compared to Tz pts who were decline by $10 percentage points compared to baseline and to , 50%; no
0.027 60.001% at baseline and 0.057 60.016% at 30 mo (p = 0.71). Only meaningful between-group differences were observed (ABP 980:10/359
4 NPS+Tz pts recurred as compared to 13 in the Tz arm. While limited by low [2.8%], RP: 6/184 [3.3%], RP/ABP 980: 6/171 [3.5%]). The incidence
numbers, recurrent NPS + Tz pt did not mount an IR by ex vivo assessment of cardiac AEs was low and comparable in treatment groups. One grade
(range: 0.0 - 0.026%) or by DTH (all measurements: 0 mm), while non- 3 cardiac failure event was reported in the RP/ABP 980 arm; another in the
recurrent pts mounted both clonal CTL expansion (range: 0.000- 0.33%) and RP arm was coincident with LVEF decline. No patient discontinued IP during
enhanced DTH (range: 0.0- 80.5mm). Conclusions: NPS+Tz combination is adjuvant phase because of cardiac failure. Conclusions: These pre-specified
more efficacious in generating time-dependent antigen (NPS)-specific CTL by analyses confirm the tolerability of ABP 980 and demonstrate clinical
both ex vivo and in vivo measures vs Tz. Based on these preliminary data, it similarity of ABP 980 and RP with respect to cardiac safety. The incidence of
appears that both ex vivo and in vivo IR to NPS were attenuated in pts with LVEF decline was consistent with the known cardiac safety profile of the RP.
TNBC recurrence. Further analysis of read-outs from these assays to validate No new cardiac safety signals were observed whether patients were on
the relationship of IRs to clinical effect seen with NPS+Tz in TNBC pts is ABP 980 or switched from RP to ABP 980. Clinical trial information:
underway. Clinical trial information: NCT02297698. NCT01901146.

558 Poster Session (Board #50), Sun, 8:00 AM-11:00 AM 559 Poster Session (Board #51), Sun, 8:00 AM-11:00 AM
Glutaminase (GLS) expression in primary breast cancer (BC): Correlations A prospective validation cohort study of a prediction model on non-sentinel
with clinical and tumor characteristics. First Author: Neelima Vidula, lymph node involvement in early breast cancer. First Author: Xingfei Yu,
Massachusetts General Hospital, San Francisco, CA Zhejiang Cancer Hospital, Hangzhou, China
Background: Tumor cells rely on glutamine for growth. GLS is a mitochondrial Background: Early breast cancer (cT1-2N0) with one or two sentinel lymph
enzyme that is necessary for glutamine catabolism, and is present as isoforms node (SLN) involved may avoid axillary lymph node dissection (ALND) if
GLS1 and GLS2. A GLS1 inhibitor is being studied in triple-negative (TN) BC. follow by radiotherapy supported by Z0011 and AMAROS trials. However,
We studied GLS1 expression in primary BC to understand associations with clinical only less than one-third of those patients have positive non-sentinel lymph
and tumor characteristics in publically available databases. Methods: GLS1 node (nSLN) and can truly benefit from radiotherapy or ALND in those two
mRNA levels were evaluated using expression data from the TCGA (n = 817) trials. It is necessary to identify the risk of nSLN metastasis before local
dataset, with confirmation in METABRIC (n = 1992). Associations between treatment decision. We previously developed a predictive model for nSLN
GLS1 levels and tumor subtype were assessed using ANOVA, followed by the involvement using circulating CK19 mRNA level combined with contrast-
post-hoc Tukey test for pairwise comparisons. Pearson correlations were used enhanced ultrasound (CEUS) score (ASCO2017 poster 239, NCT02992067)
to study associations between GLS1 and selected genes. Correlations with in a training set. To evaluate the predict effect of this model, we designed a
overall survival (OS) were studied with Cox proportional hazard model. For all further study using the model prospectively in a validation set (NCT03280134).
analyses, p , 0.05 was considered significant. Results: In TCGA, the ex- Methods: We identified early breast cancer cases in Zhejiang Cancer Hospital
pression of GLS1 and its isoform GLS2 were significantly inversely correlated from July 2017 to June 2018. The level of circulating CK19 mRNA tested by
(r = -0.32). GLS1 expression was highest in TN compared to hormone receptor qRT-PCR and CEUS scores were collected before surgery in each case. Pa-
(HR)+ and HER2+ BC (p , 0.001). In addition, GLS1 expression was higher tients with 1~2 SLN involved were enrolled and continued for ALND. The
in basal vs luminal A, luminal B, and HER2 enriched BC (p , 0.001). GLS1 estimated percentage of nSLN-involved were calculated both by our model
expression was significantly inversely correlated with ER (r = -0.45), PR formula and the Memorial Sloan-Kettering Cancer Center (MSKCC) nomo-
(r = -0.34), and AR (r = -0.34), and these inverse correlations remained gram. The predictive accuracy and false negative rates (FNR) were evaluated
significant when restricted to TNBC (ER: r = -0.25, PR: r = -0.25, AR: r = and the area under curve (AUC) was compared between two predictive models.
-0.30). Consistent with previous reports of MYC upregulation of GLS1, Results: Totally, 235 patients diagnosed as early breast cancer with 1~2 SLN
GLS1 expression was significantly positively correlated with MYC (r = involved were enrolled and 35.36% of them were nSLN involved after ALND.
0.26). Similarly, in METABRIC, GLS1 was most highly expressed in basal The total accuracy and FNR by our model for nSLN-involved prediction was
and TNBC, significantly inversely correlated with the expression of GLS2, 94.89% and 6.02%, respectively. The AUC was 0.952 (95%CI, 0.922~0.982),
ER, PR, and AR, and positively correlated with MYC expression. In METABRIC, significantly higher than that in MSKCC model 0.880 (95%CI, 0.833~0.927).
higher GLS1 expression was associated with better OS (HR 0.91, p = 0.005); Furthermore, only CK19 mRNA level (HR = 40.091, 95%CI, 13.663~117.635)
this association remained significant in the TN subset (HR 0.83, p = 0.03). and CEUS score (HR = 2.009, 95%CI 1.158~3.485) are significantly related to
Correlations between GLS1 and genes involved in metabolism and immune nSLN involvement in both univariate and multivariate analysis, adjusted by
activation will be presented at the meeting. Conclusions: GLS1 expression is age, menopause statue, tumor size, histological grade, estrogen receptor,
highest in basal and TNBC, is associated with MYC expression, and may have progesterone receptor and human epidermalgrowth factor receptor-2 ex-
prognostic implications. These findings support ongoing trials of GLS1 in- pression. Conclusions: Our model using CK19 mRNA and CEUS score
hibition in TNBC. showed potential predictive value of nSLN before surgery in early breast
cancer patients. Further validation in larger multicenter cohort is warranted
before changing clinical practice.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
24s Breast Cancer—Local/Regional/Adjuvant

560 Poster Session (Board #52), Sun, 8:00 AM-11:00 AM 561 Poster Session (Board #53), Sun, 8:00 AM-11:00 AM
Breast cancer treatment according to pathogenic variants in cancer susceptibility Cognitive impairment in breast cancer patients before surgery?: Results of a
genes in a population-based cohort. First Author: Allison W. Kurian, Stanford subgroup of the French CANTO cohort. First Author: Marie Lange, Centre
School of Medicine, Stanford, CA François Baclesse, Caen, France
Background: Increasing use of germline genetic testing may have unintended Background: Cognitive impairment has been reported among breast cancer
consequences on breast cancer treatment. We do not know whether treatment (BC) patients (pts) after adjuvant chemotherapy. However, very few studies
deviates from guidelines for women with pathogenic variants (PV) in cancer focused on cognitive function at diagnosis. Here we aimed to describe
susceptibility genes. Methods: SEER data for all women aged $20 years, cognitive impairment among recently diagnosed BC before any treatment.
diagnosed with breast cancer in 2014-15 and reported to Georgia and Cal- Methods: A predefined sub-study of the French national prospective cohort
ifornia registries (N = 77,588) by December 1, 2016 were linked to germline of cancer and toxicities performed extensive objective and subjective cog-
genetic testing results from 4 laboratories that did nearly all clinical testing. We nitive assessment before any BC treatment (surgery or neo-adjuvant treat-
examined first course of therapy (before recurrence or progression) of stage ment). This study included a group of pts with newly diagnosed invasive
, IV patients who linked to a genetic test: bilateral mastectomy (BLM) in Stage I-III BC and a group of healthy control (HC) women matched on age and
candidates for surgery (unilateral, stages 0-III); post-lumpectomy radia- education level. Episodic and working memory, executive functions, processing
tion in those with an indication (all but age $70, stage I, hormone receptor speed, attention, cognitive complaints (FACT-COG), anxiety and depression
(HR)-positive and HER2-negative); and chemotherapy in those without a (HADS) and fatigue dimensions (FA12) were assessed with neuropsychological
definitive indication (stage I-II, HR-positive, HER2-negative and 21-gene tests and the referred self-report questionnaires. Objective and cognitive im-
recurrence score , 30). We report the percent treated based on multivariable pairment were defined according to International Cognition and Cancer Task
modeling, adjusted for age, race, stage, grade, insurance and socioeconomic Force recommendations. Results: 264 women (median age 54611 years) re-
status. Results: The table shows that 9% of patients who linked to a genetic cently diagnosed (average of 37 days after initial diagnosis) with invasive BC
test result had a PV (N = 1,283). Compared to women with negative (stage I-II, 69%) and 132 matched HC participated in this study. Impaired
results,women with BRCA1/2 PVs were more likely to receive BLM, more likely working memory (20% vs 4%), information processing speed (36% vs 17%),
to receive chemotherapy without definitive indication, and less likely to receive attention (16% vs 1%) and executive function (21% vs 8%) were higher
indicated radiation in their first course of therapy. Lower-magnitude effects among pts than in HC (p , 0.001). In addition, 24% (n = 64) of pts reported
were seen with other PVs but not variants of uncertain significance (VUS). cognitive complaints versus 12% of HC (n = 16, p , 0.01). Emotional and
Conclusions: In a population-based setting, women with PVs in BRCA1/2 or cognitive fatigue were higher in pts than HC (mean 24 vs 15 and 18 vs 11,
other cancer susceptibility genes may have a higher risk of receiving locore- p , 0.01). Similarly, higher levels of anxiety and depression were observed
gional and systemic treatment that does not follow guidelines. in patients when compared with HC (respectively in 41% and 3% of patients
Treatment Use: % (95% CI). vs 10 and 1% for HC, p , 0.001). Objective cognitive impairment was not
Test Results (mutually exclusive BLM (N = Radiation (N = Chemotherapy (N =
associated with cognitive complains. Both objective and subjective cognitive
categories) 10,144) 4,575) 5,504) impairment were not associated with anxiety or depression. However cog-
Negative (N = 10,418) 23.0 (22.0- 74.5 (73.1-75.9) 30.5 (29.4-31.6)
nitive complain was associated with fatigue (p , 0.001). Conclusions: In
24.0) this large study, compared to HC, patients recently diagnosed with a lo-
VUS (N = 2,539) 24.6 (22.5- 74.2 (71.2-77.2) 30.4 (28.1-32.8) calized BC reported more cognitive complains and objective cognitive im-
26.7)
BRCA1/2 PV (N = 850) 55.4 (51.0- 48.0 (38.1-57.8) 40.2 (35.2-45.1)
pairment before surgery, without link with emotional status, but with fatigue.
59.9) Further understanding of the biology and correlates of cognitive dysfunction
Other PV* (N = 433) 34.5 (28.9- 68.1 (59.8-76.3) 33.7 (27.9-39.5) at BC diagnosis is needed (CANTO-NCT01993498).
40.1)
*ATM, CHEK2, PALB2, NBN, TP53, BARD1 and others

562 Poster Session (Board #54), Sun, 8:00 AM-11:00 AM 563 Poster Session (Board #55), Sun, 8:00 AM-11:00 AM
Does the innovative 4R Care Delivery Model improve timing and sequencing TARGIT E(lderly): Prospective phase II trial of intraoperative radiotherapy
of guideline recommended breast cancer care in safety net and non-safety (IORT) in elderly patients with small breast cancer. First Author: Frederik
net centers? First Author: Christine B. Weldon, Northwestern University Wenz, University Hospital Freiburg, Freiburg, Germany
Feinberg School of Medicine, Chicago, IL
Background: TARGIT E is a prospective, multicentric single-arm phase II
Background: Under the NCI ASCO Teams Project, we proposed a 4R Model trial (NCT01299987; Neumaier et al. BMC Cancer 2012) that is based
which enables patient (pt) and care team to manage timing and sequencing of on the experimental arm of the protocol of the international randomized
interdependent care with a novel multimodality 4R Care Project Plan (Trosman TARGIT A study. The trial was designed to investigate the efficiency of a risk-
JOP ’16). 4R (Right Info/Care/Patient/Time) was previously piloted at 3 Chicago adapted approach consisting of a single dose of intraoperative radiotherapy
centers (Weldon ASCO ‘18). Methods: A new study tested impact of 4R on timing (IORT) followed by whole breast radiotherapy (WBRT) only when risk fac-
and sequencing of guideline recommended care at 4 safety net and 3 non safety tors are present in elderly patients. Methods: Patients with low-risk breast
net US centers. 4R Plans were provided to stage 0-III breast cancer pts Jan- cancer ($70 years, cT1, cN0, cM0, invasive carcinoma of no special type)
Nov’18, 4R cohort. Clinical and pt reported data analyses compared 4R cohort were enrolled between February 2011 and September 2014. During breast
(N=105) to a historical control cohort of pts who received care pre-4R, Jan - Dec ’17 conserving surgery, a single dose of 20 Gy of low-energy IORT was given
(N=190). Results: We significantly improved 3 referral metrics and 4 referral (Intrabeam, Carl Zeiss Meditec, Germany). Additional postoperative WBRT
completion metrics - receipt of care by pts who were referred (Table). After (46 – 50 Gy) was applied in case risk factors (larger size, other histology,
referrals, safety net pts had a significant increase in 4R vs control cohort in free margin , 1 cm, lymphatic vessel invasion (L1), positive nodes,
receiving genetic consult (72%, 21/29 vs. 42%, 18/43, p=.02) and dental visit multifocality/multicentricity, extensive intraductal component (EIC)) were
(100%, 6/6 vs. 20%, 1/5, p=.02). They had lower increases in flu shot referrals present. Systemic therapy was applied according to international standards
(41%, 24/58, vs 36%, 37/104, NS) and dental referrals (10%, 6/58, vs 5%, 5/ and guidelines. The primary outcome was the local relapse rate. Discontin-
104, NS) than non safety net pts who had significant increases. Other metrics uation of the trial was judged to be necessary if the local relapse rates exceed 3/
improved at a similar rate for safety net and non safety net pts. Conclusions: 4R 4/6% at 2.5/5/7.5 years. Results: A total of 541 patients were screened. Of
markedly improved referral and receipt of interdependent guideline recom- those 474 patients were enrolled, whereas 347 (73.1%) received IORT only,
mended breast cancer care. For most metrics safety net pts benefited from 4R
99 (10.8%) IORT plus WBRT, 22 (4.6 %) WBRT only and 7 (1.5%) surgery
at a similar or higher rate than non safety net pts, indicating that 4R may reduce
only with no subsequent adjuvant therapies. After a median follow-up of 3.25
care disparities. Low increases in referrals for safety net pts and in trial referral/
years, four ipsilateral in-breast recurrences were observed (after 11, 33, 42
enrollment for all pts must be addressed. An expansion of 4R across the US
and 43 months) resulting in an actuarial local relapse-free survival of 99.8%
continues this work.
after 2.5 years and 98.5% after 5 years. Conclusions: The results of the
% of pts % of pts re- prospective TARGIT E trial consolidate earlier reports from the randomized
referred,4R ferred, control % of referral comple- % of referral completion
cohort cohort tion by referred pts, 4R by referred pts, control TARGIT A trial, supporting the use of accelerated partial breast radiotherapy
Metric Guideline N=105 N=190 P cohort cohort P (APBI) in selected patients. The observed local relapse rates at 2.5 and 5 years
PCP visit be- NCCN 35 24 .04 86, N=37 58, N=45 .02 are far below the predefined stopping rules. Clinical trial information:
fore treat- OAO
ment (bf Tx) NCT01299987.
Genetic con- NCCN 50 45 NS 77, N=53 56, N=86 .02
sult bf BR/OV
surgery
Flu shot bf Tx NCCN 50 37 .03 85, N=53 66, N=70 .02
INF
Dental visit bf ADA 17 5 .0006 83, N=18 33, N=9 .03
Tx
Trial NCCN 35 26 NS 41, N=37 38, N=50 NS
enrollment BC

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Breast Cancer—Local/Regional/Adjuvant 25s

564 Poster Session (Board #56), Sun, 8:00 AM-11:00 AM 565 Poster Session (Board #57), Sun, 8:00 AM-11:00 AM
No relationship of axillary total tumor load (TTL) by PCR (OSNA) in early Gene expression comparison between primary triple-negative breast cancer
breast cancer and local and distant clinical outcomes. First Author: Jose and matched axillary lymph node metastasis. First Author: Marissa K. Srour,
Ales-Martinez, Complejo Asistencial de Ávila, Ávila, Spain Cedars-Sinai Medical Center, Los Angeles, CA
Background: The study of sentinel lymph nodes (SLN) assessed by One Step Background: Sentinel lymph node (SLN) biopsy guides breast cancer treat-
Nucleic Acid Amplification (OSNA, Sysmex, Kobe, Japan) creates a new ment and prognostication. To date, there have been few studies examining the
variable, Total Tumor Load (TTL). This variable is defined as the total number genetics of SLN metastasis in triple negative breast cancer (TNBC). The aim of
of CK19 mRNA copies in all positive SLN (copies/microL). The latest edition this study is to characterize and compare gene expression patterns of primary
of the Spanish Oncological Gynecology Society (SEGO) Guideline (2017) breast cancers and autologous matched SLN metastases. Methods: Patients
proposes a complete axillary lymph node dissection (ALND) when TTL is with stage 2-3 ER/PR negative, HER2 negative TNBC with macrometastasis to
15,000 copies or more in early breast cancer. In our center we are using the SLN who did not have neoadjuvant therapy were selected. The tumor-
OSNA to ascertain if there is axillary node involvement but the decision to specific area was isolated from breast and SLN paraffin embedded tissue
proceed to ALND is based on Z0011 criteria. We want to determine if there sections. Gene expression of a panel of 2567 cancer-associated genes was
is a correlation between clinical outcomes and TTL values, between TTL and analyzed with the HTG EdgeSeq system coupled with the Illumina Next
pathological variables and if TTL is a useful tool to decide when to complete Generation Sequencing (NGS) platform. Results were validated with RNA-
an ALND. Methods: Clinicopathological and follow up data were obtained scope assays for RNA in situ detection. Results: 18 pairs of TNBC and
from all patients with invasive breast cancer and SLN assessed by OSNA matched SLN metastasis were analyzed for 2567 genes. Compared with the
between 2011 and 2017 at our center. Results: A total of 321 patients primary TNBC, SLN metastasis had 34 statistically significant upregulated
underwent SNB assessed by OSNA with an average follow-up of 56 months. genes and 31 downregulated genes. SLN metastasis had at least a 5-fold
320 were female and 1 male. Age range 27-89 years (mean 58.9). 85 % change (FC) in upregulation in 3 genes and downregulation in 3 genes,
were ductal, 10 % lobular and 5 % other. 53.5% were luminal A, 28.66% compared to primary TNBC [Table]. Notably, there was an upregulation of anti-
luminal B, 7.78%, triple negative, 4.3%, Her2 positive and 4.3%. luminal apoptosis and survival signaling genes (i.e. BIRC3) in the SLN metastasis.
B-Her2 positive.TTL was equal to 0 in 183 cases and greater than zero in There was also an upregulation of chemotaxis genes (CCL13, CXCL19, CXCL21,
138 cases.71 cases showed a TTL higher than 15,000 copies. Only 21 cases CXCR5, TNFSF11, p,0.001). The most striking feature is the downregulation of
met Z0011 criteria and had ALND. As of now, 3 patients have had genes known to regulate cell microenvironment interaction (MMP2, MMP14,
locoregional relapse and 8 metastatic disease. 12 have died, only two from COL1A1, COL1A2, COL5A2, COL6A6, COL11A1, COL17A1). Conclusions: In
metastatic breast cancer. Conclusions: Using Z0011 criteria, we have ad- TNBC, SLN metastasis has a distinct gene expression profile. Genes associated
equate clinical outcomes with a low rate of ALND; If we had based the axillary with anti-apoptosis, survival responses, and chemotaxis are upregulated, and
management on TTL values we would have multiplied the number of ALND genes associated with regulation of extracellular matrix are downregulated.
by a factor of 3.3 (from 21 to 71); We have observed a tendency to higher TTL Upregulated and downregulated genes with at least a 5-fold change in gene
in luminal phenotypes and to lower TTL in HER2 positive and triple negative expression in lymph node metastasis compared with TNBC.
subtypes; Work is in progress to increase our sample size.
Gene Fold-Change P-Value
CYP2A6 19.71 0.002
CCL21 14.13 ,0.001
FCER2 7.33 ,0.001
COL6A6 -5.24 ,0.001
FGF5 -10.89 ,0.001
COL17A1 -11.98 ,0.001

566 Poster Session (Board #58), Sun, 8:00 AM-11:00 AM 567 Poster Session (Board #59), Sun, 8:00 AM-11:00 AM
Prediction of occult axillary metastases in treatment-naı̈ve patients with Screening for cancer-related distress among women with newly diagnosed
breast cancer: A transSENTINA analysis. First Author: Cornelia Kolberg- breast cancer (BC). First Author: Lauren Z. Rynar, Loyola University Medical
Liedtke, Charité - Universitätsmedizin Berlin, Berlin, Germany Center, Maywood, IL
Background: Prediction of occult axillary metastases through clinical / bi- Background: The Loyola University Chicago Cardinal Bernardin Cancer Center
ological parameters may allow reduction of axillary staging. This is particularly multidisciplinary Breast Oncology Center evaluates new patients (pts) for
important, as systemic therapies have become more efficient. We have cancer-related distress using a needs-based screening tool, in accordance with
conducted a systematic analysis among patients undergoing axillary sentinel Commission on Cancer (CoC) Standard 3.2. Identifying distress among newly
lymph-node biopsy (SLNB) before initiation of primary systemic therapy as part diagnosed pre-surgical and pre-neoadjuvant pts allows for comprehensive
of a clinical trial (SENTINA) with the goal to identify predictors of sentinel treatment planning and establishment of a baseline for repeated assessments.
lymph node status in a well-defined patient cohort. Methods: Patients with a Methods: Pts with newly diagnosed BC between May 2017 and June 2018
clinically negative axillary status who underwent SLNB as part of the pro- completed the “Patient Screening Questions for Supportive Care” (Coleman
spective SENTINA trial were included. Univariate and multivariate analyses Supportive Oncology Collaborative, 2017), a consolidated screening tool
were carried out to identify clinical / pathological parameters associated based on validated instruments (NCCN Distress, PHQ-4, PROMIS), prior to
with SLN status, using logistic regression models. Model performance was initial provider visit. Cancer staging, demographics, and supportive oncology
assessed by ROC analyses. Calculations were performed using R version referrals were obtained from medical records. Descriptive statistics and chi-
3.5.2. Results: Arms A and B of the SENTINA study contained 1022 pa- square were used. Results: 100 pts aged 30-94 (mean(SD) = 61.56(12.03))
tients. Among 805 cN0 patients, all parameters considered relevant for this completed the screening tool; 14.9% had Stage 0, 43.6% Stage I, 34.0%
analysis were available. 527 and 278 patients presented with negative and Stage II, 3.2% Stage III, and 4.3% Stage IV disease. 39% of pts screened
positive lymph nodes upon SLN biopsy, respectively. Univariate regression positive for anxiety on the PHQ-4, and over 20% for depression. Anxiety was
models identified largest tumor diameter (odds ratio (OR) 1.016, p-value associated with cancer stage (X2(df) = 12.20(4), p = .016). The most common
0.0041), tumor type (ductal vs. lobular, OR 2.004, p 0.00234), tumor practical concerns included living alone (19%), issues with work/school
grading (low vs. high, OR 0.537, p , 0.001), hormone receptor (HR) status (16%), and paying for medical care (12%). Common physical concerns in-
(negative vs. positive, OR 2.668, p , 0.001), HER2 status (negative vs. cluded difficulty with sleep (40%) and concentration/memory (17%), and tin-
positive, OR 1.462, p 0.0158) as being associated with SLN status with gling hands or feet (14%). Poor sleep was associated with depression (X2(df) =
an a , 0.1. Multivariate analysis resulted in tumor diameter, HR status, 6.50(1), p = .011) and anxiety (X2(df) = 7.17(1), p , .01). 57.7% reported at
HER2 status and tumor type being independently associated. These pa- least "a little bit" of fatigue and 17.7% reported moderate to very severe pain.
rameters were combined using stepwise (backward and forward) selection Nearly all pts wanted to better understand their diagnosis (87.8%), prognosis
into a prediction model. This model predicted SLN status with an AUC of (87.8%), or treatment (91.8%). Conclusions: Pts with newly diagnosed, early
only 0.65. Conclusions: Using data obtained as part of the SENTINA trial we stage BC experience high levels of physical and emotional distress at the
were able to build a prediction model that was able to predict SLN me- earliest point in the treatment trajectory. This study captures BC patients at a
tastases in treatment naı̈ve cN0 patients with limited accuracy. Additional unique time point and provides support for conducting routine screening for
(biological) parameters, such as response to systemic therapies (i.e. axillary supportive oncology needs at initiation of care. Further studies should reassess
conversion through PST) may be more appropriate to predict presence of needs sequentially to determine changes across the care continuum.
occult sentinel lymph node metastases among patients with breast cancer.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
26s Breast Cancer—Local/Regional/Adjuvant

568 Poster Session (Board #60), Sun, 8:00 AM-11:00 AM 569 Poster Session (Board #61), Sun, 8:00 AM-11:00 AM
Risk stratification in earl- stage luminal breast cancer patients treated with Clinical pharmacogenomic testing impacts therapy decisions and supportive
and without RT. First Author: Charlotta Wadsten, Umea University, Department medication choices in breast cancer. First Author: Kathleen Kiernan Harnden,
of Surgical and Perioperative Science, Umea, Sweden Inova Schar Cancer Institute, Fairfax, VA
Background: The goal was to develop and validate a biologic signature for Background: Pharmacogenomics, the study of the interaction between the
10-year ipsilateral invasive breast event (IBE) risk in luminal Stage 1 breast patient’s genome and therapeutic drug response, evaluates the associations
cancer (BC) patients treated surgically and either with or without radiation between efficacy and toxicity through analysis of drug metabolizing en-
therapy (RT). Methods: This cohort was from Uppsala University and zymes. As personalized medicine advances to the forefront of cancer care,
Västerås Hospitals diagnosed with Stage 1 BC and treated surgically between pharmacogenomics can evaluate the individual’s ability to metabolize key
1987 and 2004. Treatment was neither randomized nor strictly rules based, medications in breast cancer treatment including anti-emetics, opioids, and
including adjuvant RT, Hormone Therapy (HT), and Chemotherapy (CT). tamoxifen. Women who do not achieve optimal levels of the active me-
Biomarkers (HER2, PR, Ki67, COX2, p16/INK4A, FOXA1 and SIAH2) were tabolites of tamoxifen are at higher risk of recurrence. Patients on che-
assessed on tissue microarrays in PreludeDx’s CLIA lab by board-certified motherapy who do not respond to anti-emetics can suffer from nausea and
pathologists. Risk groups were calculated using biomarkers and clinical vomiting resulting in dehydration and hospitalization. This project evaluates
factors age and size. A multivariate Cox proportional hazards analysis was the feasibility and therapeutic impact of real time pharmacogenomics in a
used to determine hazard ratio for biologic signature. 10-year IBE risk was selection of patients at the Inova Schar Cancer Institute (ISCI). Methods: An
assessed using Kaplan-Meier survival analysis. Results: There were 423 interdisciplinary team was created through the ISCI and the Inova Trans-
luminal cases with biomarker data having 54 IBEs, and a median follow-up lational Medicine Institute to implement cheek swab based pharmacoge-
of 11.8 years. There were 372 patients treated with BCS and 51 with nomic testing in 50 new patients undergoing mastectomy or neoadjuvant
Mastectomy, and 325 received RT, 169 received HT, and 47 received CT. chemotherapy for breast cancer. Study patients were assessed for genotypic
In a multivariate analysis, the biologic signature (HR = 1.6, p = 0.019) and variability of key CYP enzymes and resulting impact on anti-emetic choices,
RT (HR = 0.51, p = 0.027) were associated with IBE risk adjusting for other perioperative pain control, and tamoxifen use. Results: Data was collected
treatments (HT and CT) and Luminal A status (p = 0.37). For patients over 50 in a RedCap database. The 50 women enrolled were ages 28-83. Cheek
yrs of age with luminal A disease and treated without CT (n = 205), an swabs were performed in clinic and median turn around time was 7 days. 24
elevated biologic signature identified a subset of patients with a 15% (+/- distinct genotypes were found in the 50 patients. 20% had abnormal
14%) 10-year IBE risk without RT (n = 38) compared to a 4% (+/-6%) IBE CYP2D6 phenotypes indicating abnormalities in tamoxifen metabolism.
risk with RT (n = 72), while patients with a low biologic signature had a 10- 28% of patients had results leading to changes in dose or medication choice
year IBE risk of 4% (+/- 4%) without RT (n = 26) and 3% (+/-5%) IBE risk of perioperative pain control. 6% of patients had a CYP2D6 ultra-rapid
with RT (n = 69). Conclusions: With further prospective validation, the bi- metabolizer phenotype and were given granisetron in lieu of ondansetron.
ologic signature identified herein may provide a tool enabling improved These patients had no documented nausea or vomiting requiring dose ad-
management for women diagnosed with early luminal BC. justments to the treatment plan or medical intervention. 40% of patients had
results recommending avoidance of tamoxifen, 75% of which have ER+
breast cancer. 25% of patients had recommended changes to the dose of
tamoxifen. Conclusions: Pharmacogenomic testing is feasible and available
real-time for immediate use in the clinic. CYP mutations impact treatment
decisions in a significant proportion of patients. Individualized treatment
plans tailored to pharmacogenomic recommendations can be created in the
multi-disciplinary setting and may decrease side effects of treatment and
improve efficacy of curative therapy.

570 Poster Session (Board #62), Sun, 8:00 AM-11:00 AM 571 Poster Session (Board #63), Sun, 8:00 AM-11:00 AM
Biomarker analysis of PALLET: A neoadjuvant trial of letrozole (L) 6 palbociclib Genomic-based predictive biomarkers to anti-HER2 therapies: A combined
(P). First Author: Vera Martins, ICR, London, United Kingdom analysis of CALGB 40601 (Alliance) and PAMELA clinical trials. First Author:
Aranzazu Fernandez-Martinez, Lineberger Comprehensive Center. Department
Background: PALLET randomized 307 postmenopausal women with ER+
of Genetics. University of North Carolina, Chapel Hill, NC
primary breast cancer to one of 4 treatment groups (3:2:2:2 ratio): A: L for
14wks; B: L for 2wks then L+P to 14wks; C: P for 2wks then L+P to 14wks; D: Background: In HER2-positive breast cancer, new biomarkers of response
L+P for 14wks. This allowed a randomized 1:2 comparison of L (Group A) vs are needed in order to direct multi-agent anti-HER2 combinations towards
L+P (Groups B+C+D) at 14wks. P was given 125mg/d PO (21 days on, 7 days patients in whom they are truly needed. CALGB 40601 and PAMELA trials
off). Adding P to L markedly enhanced Ki67 suppression and Complete Cell tested neoadjuvant dual HER2 blockade and included gene expression
Cycle Arrest (CCCA, Ki67 , 2.7%) by 14wks but did not substantially in- analysis aimed to evaluate different genomic biomarkers of trastuzumab
crease clinical response. We now report exploratory analysis of the associ- and/or lapatinib benefit. Methods: Gene expression by mRNA sequencing
ation of baseline expression of 6 pre-specified biomarkers involved in (RNAseq) was performed on 265 and 142 pre-treatment tumors of the
estrogen and CDK4/6 signaling with CCCA at 14wks and changes in their CALGB 40601 and the PAMELA clinical trials respectively. Intrinsic sub-
expression during therapy. Methods: Estrogen receptor (ER), progesterone types were determined by nCounter PAM50-predictor on the PAMELA
receptor (PgR), RB and CCNE1 were measured by IHC and CCND1 by IHC samples. A new HER2-positive specific gene-centering method was trained
and FISH (CCND1/CEP11 ratio$2.0 amplified). Baseline biomarker values on the PAMELA RNAseq data, and showed a higher concordance with
were available with 14wk Ki67 values in up to 64 patients for L alone and up PAM50 predictions obtained from nCounter platform. This method was then
to 124 patients for L+P. Of these 59% and 90%, respectively, achieved applied to CALGB 40601 samples. Results: In the combined cohort, the
CCCA. Results: With L alone CCCA was significantly less frequent (indicating subtype distribution was 10% Luminal A, 8% Luminal B, 62% HER2-enriched
relative resistance) with low baseline PgR (odds ratio [OR] 0.22, 95%CI (HER2-E), 10% Basal and 10% Normal-like. The pCR rate was significantly
0.05-0.96, p = 0.04) or high CCNE1 levels (OR 10.39, 95%CI 1.19-90.48, higher in HER2-E vs. not HER2-E subtypes (48.6% vs. 20.7%; P , 0.001).
p = 0.03). With L+P CCCA was also significantly less frequent with high HER2-E subtype correlation, ERBB2 amplicon and B-cell genomic signatures
CCNE1 (OR 50.34 95%CI 5.12-495.34, p = 0.001) or with low baseline ER were associated with pCR, while luminal signatures were associated with non-
(OR 0.21 95%CI 0.08-0.60, p = 0.004). CCCA was not significantly dif- responders. In multivariate analysis HER2-E subtype, ERBB2 mRNA and IgG
ferent with either treatment according to CCND1 amplification status or signature expression were independent predictors of response to paclitaxel +
expression overall. However, CCCA showed a tendency to being less frequent trastuzumab +/-lapatinib (OR = 1.98, OR = 1.51, OR = 1.48, respectively,
in non-amplified cases with low baseline cyclin-D1 expression when treated P ,0.05). The event free survival analysis at 5 years in the CALGB 40601
with L+P (p = 0.10). There were no significant changes in ER levels or cohort showed a benefit of dual vs single anti-HER2-blockade (HR 0.35,
CCND1 amplification over 14wks. By 14 wks PgR, RB, CCND1 and CCNE1 P ,0.05). Within the HER2-E, ERBB2-high and IgG–high subpopulations,
levels were significantly suppressed by L or L+P (geomeans PgR: -96.4% vs there were also a benefit of dual vs. single anti-HER2 treatment (HR = 0.32,
-94.9%; CCND1: -79.9% vs -70.7%; CCNE1: -68.2% vs -74.7%; RB: HR = 0.15, HR =0.15, respectively, P ,0.05). Conclusions: Intrinsic subtype,
-23.5% vs 26.1%, respectively) and there was no significant difference ERBB2 mRNA levels, and IgG genomic signature are independent predictive
between the treatments. Conclusions: These data support low ER, possibly biomarkers of response in the combined cohort. The clinical implementation of
indicating limited luminal status, and high CCNE1 as markers of poor Ki67 these biomarkers could help to design future escalation/de-escalation clinical
response to L+P in primary disease and are consistent with findings in trials in the HER2-positive neoadjuvant setting. Support: U10CA180821,
studies in advanced disease. Clinical trial information: NCT02296801. U10CA180882, U24CA196171, P50-CA58223, GSK, SPORE, BCRF and
SEOM. https://acknowledgments.alliancefound.org.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Breast Cancer—Local/Regional/Adjuvant 27s

572 Poster Session (Board #64), Sun, 8:00 AM-11:00 AM 573 Poster Session (Board #65), Sun, 8:00 AM-11:00 AM
Beyond TILs: Predictors of pathologic complete response (pCR) in triple-negative Germline mutation status and therapy response in high-risk early breast
breast cancer (TNBC) patients with moderate tumor-infiltrating lymphocytes cancer: Results of the GeparOcto study (NCT02125344). First Author:
(TIL) receiving neoadjuvant therapy. First Author: Nour Abuhadra, MD Anderson Esther Pohl-Rescigno, 1 Center for Familial Breast and Ovarian Cancer and
Hematology/Oncology Fellowship, Houston, TX Center for Integrated Oncology (CIO), Cologne, Faculty of Medicine and
University Hospital Cologne, Cologne, Germany
Background: Increased TIL in TNBC is associated with higher rates of pCR.
High TIL is also associated with improved disease free survival and overall Background: GeparOcto compared the efficacy of two neoadjuvant treatment
survival. The aim of this study is to identify data cut-points of pre-treatment (NAT) regimens in high-risk early breast cancer (BC): Sequential intense dose-
low, moderate and high TIL count based on pCR and to identify clinical and dense epirubicin, paclitaxel, and cyclophosphamide (iddEPC) and weekly
pathological predictors of pCR in patients with moderate TIL. Methods: We paclitaxel plus non-pegylated liposomal doxorubicin (PM), plus carboplatin
evaluated the relationship between pCR and TIL in 180 patients with stage I- (PMCb) in triple-negative BC (TNBC). There was no difference in pathologic
III TNBC enrolled in the ARTEMIS trial (NCT02276443). Recursive portioning complete response (pCR) rates (Schneeweiss et al. Eur J Cancer 2019). Here, we
was used to identify cut-points. Clinical and pathological variables such as age stratified pCR rates according to germline mutation status. Methods: Germline
at diagnosis, stage, race, histology as well as Ki-67, vimentin, and androgen (g) mutation analysis of BRCA1/2 and 16 further BC predisposition genes in 914
receptor (AR) by immunohistochemistry, were evaluated in pts with moderate patients (pts) enrolled (393 pts with TNBC, iddEPC n = 194, PMCb n = 199; 156
TIL. A multivariable logistic regression model identified variables indepen- pts with HER2-/HR+ BC, iddEPC n = 75, PM n = 81; and 365 pts with HER2+
dently, significantly associated with pCR. Results: Four TIL groups were BC, iddEPC n = 182, PM n = 183). Results: The gBRCA1/2 mutation prevalence
identified with pCR rates of 23%, 31%, 48% and 78% respectively (p , was 17.6% in TNBC, 14.1% in HER2-/HR+ BC and 1.4% in HER2+ BC. Overall,
0.0001) (Table A). In the two combined moderate TIL groups, 90 (97%) pts pts with gBRCA1/2 mutations achieved higher pCR rates than gBRCA1/2 wildtype
were evaluable for the multivariate model. Stage I-II disease, high Ki-67 and low pts (60.4% vs 46.7%, OR 1.74, P = 0.012), with more pronounced effects in the
AR were associated with increased probability of pCR (Table B). The multi- PM(Cb) arm (68.1% vs 45.7%, OR 2.53, P = 0.005). Among gBRCA1/2 wildtype
variable logistic regression model area under the ROC curve was 0.78 (95% pts, 76 carried mutations in non-BRCA1/2 predisposition genes. pCR rates were
CI=0.68-0.88; p,0.0001). A model of computed risk score [Stage I-II (score similar to those observed in pts without any mutation. Conclusions: Pts with
2)+Ki-67$50% (score 1)+AR,10% (score 1)] predicted a probability of 67% gBRCA1/2 mutations benefitted most from NAT with highest pCR rates achieved
for pCR when all three variables were favorable (Table). Conclusions: Four TIL in the gBRCA1/2 TNBC / PMCb group. The role of Cb for NAT of gBRCA1/2 TNBC
groups were identified. In pts with moderate TIL levels, early stage disease, high should be further explored. Clinical trial information: NCT02125344.
Ki-67 and low AR were associated with increased probability of pCR with pCR both arms [%] pCR iddEPC [%] pCR PM(Cb) [%]
neoadjuvant therapy. OR* OR* OR*

A: TIL Groups. Overall (n = 914) 48.1 48.3 47.9

mt (n = 96) vs wt (n = 818) 60.4 vs 46.7 53.1 vs 47.8 68.1 vs 45.7
% TIL (Group) N #pCR (%) 1.74 (1.13-2.68; 0.012) 1.24 (0.68-2.24; 0.484) 2.53 (1.33-4.83; 0.005)
< 5 (Low) 69 16 (23) TNBC (n = 393) 50.1 48.5 51.8
‡5-10 (Moderate-Low) 51 16 (31) mt (n = 69) vs wt (n = 324) 69.6 vs 46.0 64.7 vs 45.0 74.3 vs 47.0
> 10-30 (Moderate-High) 42 20 (48) 2.69 (1.54-4.69; 0.001) 2.24 (1.04-4.84; 0.040) 3.26 (1.44-7.39; 0.005)
>30 (High) 18 14 (78) interaction arm*mutation p =
B: Variables Predictive of pCR 0.330
Variable OR (95% CI) p-value HER2-/HR+ (n = 156) 14.7 14.7 14.8
Stage I-II 7.12 (1.76 – 28.9) 0.007 mt (n = 22) vs wt (n = 134) 31.8 vs 11.9 25.0 vs 12.7 40.0 vs 11.3
High Ki-67 (‡50%) 5.32 (1.53 - 18.5) 0.01 3.44 (1.22-9.72; 0.020) 2.29 (0.51-10.30; 5.25 (1.22-22.69;
AR low (<10%) 2.93 (1.02 – 8.44) 0.05 0.279) 0.026)
C: Computed Risk Score HER2+ (n = 365) 60.3 62.1 58.5
Computed Risk Score N #pCR (%) mt (n = 5) vs wt (n = 360) 60.0 vs 60.3 33.3 (in n = 3) vs 62.6 100 (in n = 2) vs 58.0
0-2 32 4 (12%) 0.99 (0.16-5.99; 0.990) 0.30 (0.03-3.36; 0.328) n.a.
3 28 10 (36%)
4 30 20 (67%) mt = mutant, wt = wildtype; *OR (95%CI; P-value); univariate logistic regression

574 Poster Session (Board #66), Sun, 8:00 AM-11:00 AM 575 Poster Session (Board #67), Sun, 8:00 AM-11:00 AM
On-treatment changes in tumor-infiltrating lymphocytes (TIL) during neo- Prognostic value of PAM50 in residual breast cancer following neoadjuvant
adjuvant HER2 therapy (NAT) and clinical outcome. First Author: endocrine therapy (NET): A retrospective analysis with long follow-up. First
Stephen James Luen, Peter MacCallum Cancer Centre, East Melbourne, Author: Miguel J. Gil Gil, Breast Cancer Unit & Medical Oncology Department,
Australia Institut Català d’Oncologia, IDIBELL, Barcelona, Spain
Background: Higher quantity of pretreatment TIL (PT) is associated with Background: NET is gaining more acceptances for the management of
improved pCR and EFS in HER2+ early breast cancer (BC). The value of on- hormonal receptors (HR)-positive breast cancer (BC). To date, the decrease
treatment TIL is unknown. Methods: The NeoALTTO trial randomized 455 of Ki-67 and PEPI score are the only prognostic factors associated with
women with HER2+ BC to 12 weeks NAT with trastuzumab, lapatinib or relapse-free survival after NET. PAM50 is a validated prognostic test in newly
combination with paclitaxel, followed by FEC after surgery. In the PAMELA diagnosed BC; however, its value in residual tumors after NET is currently
trial 151 women received 18 weeks NAT with lapatinib and trastuzumab unknown. Methods: We took tumor tissues from patients of a retrospective
(6hormonal therapy). TIL were quantified on PT and on-treatment (W2) study of 119 postmenopausal women with HR-positive stage II-III BC.
biopsies using the published method on H&E slides, and tested for associ- Patients were diagnosed from 1997 to 2009 and were treated with NET for a
ations with pCR (logistic regression), EFS and OS (Cox models) in univariate median duration of 8.5 months. Median age was 74 (63-88). After NET all
(UV) and multivariate (MV) analyses. The likelihood ratio test assessed added patients underwent surgery (73% conservative). Adjuvant treatment were
prognostic value to clinicopathological (CP) variables. pCR associations were endocrine therapy in 100%, radiotherapy in 76.5% and chemotherapy 7%.
validated in PAMELA. We investigated enrichment of immune cell subsets Median follow-up from surgery was 112 months. Median follow-up from
using previously published RNAseq data from NeoALTTO. Results: In Neo- surgery was 112 months. We observed 26 (24%) of distant relapses and 75
ALTTO, PT and W2 TIL were evaluable in 277/455 (61%). We defined two deaths (44 without cancer). Median overall survival was 134.8 months. RNA
groups: immune-poor (L+F) and immune-enriched (II+P), see Table. Immune- was extracted from FFPE tumor tissues of surgical specimens. A panel of 55
enriched (41%; 134) vs poor (59%; 164) patients had significantly higher pCR BC-related genes, including the research-based PAM50 assay (subtypes,
rates (40% vs 21%; UV OR 2.24; 95%CI 1.31-3.85; P = .003; MV P = .009), ROR-S and ROR-P pre-defined cutpoints), androgen receptor (AR), immune
and added significant value to CP + PT TIL for prediction of pCR (P = .003). This genes (CD8A, CD4, PDL1 and PD1). Uni- and multi-variable Cox models
was further confirmed in PAMELA (N = 94/151) (26% vs 6%; UV P = .021; MV were used to evaluate the association of each variable with distance re-
P = .028). In NeoALTTO, the immune-enriched vs poor patients had signifi- currence free interval (DRFI). Results: PAM50 subtype distribution: Luminal
cantly improved EFS (5 yr est 85% vs 60%; UV HR 0.31; 95%CI 0.18-0.54; A 54.3%, Normal-like 24.3%; HER2-enriched 16,5%, Luminal B 1% and
P , .001; MV P , .001) and OS (5 yr est 91% vs 77%; UV HR 0.40; 95%CI basal 1%. Distribution of ROR-S groups was Low 64%, medium 30%and
0.20-0.82; P = .012; MV P = 0.026), and provided significant added prognostic high 6%. Distribution of PEPI score was: 0 in 43%, 3 in 37% and 6 in 20%.
value beyond CP + pCR + PT TIL (EFS P , .001) In NeoALTTO PT samples, II vs Among the different variables explored, PEPI score 0 (HR 0.27 [95%IC
F patients had enrichment of DCs, NKs and CD8+ including tissue resident 0.09-0.79] p=0.001), low ROR-S (HR 0.39 [95%CI 0.17-0.91] p=0.001)
memory cells (P = .009) suggesting requirement of key immune subsets. and high AR expression (HR 0.71 [95CI 0.53-0.96] p=0.007) were sig-
Further validation by IHC is ongoing. Conclusions: On-treatment TIL identifies nificantly associated with lower DRFI in univariate analyses. After adjusting
patients more likely to achieve pCR and have improved EFS in early-stage HER2+ for PEPI (with or without Normal-like tumors), ROR-S and AR remained
BC, beyond CP + PT TIL. This information could aid future trial design. Clinical significantly associated with outcome. Conclusions: PAM50 ROR-S and AR
trial information: NCT00553358, NCT01973660. expression in residual tumors after NET provide independent prognostic
PT TIL W2 TIL TIL change information beyond PEPI. With further validation, these biomarkers could
Persistent low (L) #10% , 10% rise help clinicians in the decision-making of adjuvant chemotherapy.
Fall in immunity (F) . 10% #10%
Immune induction (II) $10% rise
Immune persistent (P) . 10% . 10%

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
28s Breast Cancer—Local/Regional/Adjuvant

576 Poster Session (Board #68), Sun, 8:00 AM-11:00 AM 577 Poster Session (Board #69), Sun, 8:00 AM-11:00 AM
Phase II randomized study of neoadjuvant metformin plus letrozole versus Role of anthracyclines in neoadjuvant anti-HER2 regimens for HER2+ breast
placebo plus letrozole for ER-positive postmenopausal breast cancer [METEOR cancer (BC): A network meta-analysis (NMA). First Author: Giacomo Pelizzari,
Study]. First Author: Jisun Kim, Department of Surgery, Asan Medical Center, Department of Medicine (DAME), University of Udine; Dipartimento di
Seoul, South Korea Oncologia Medica, Centro di Riferimento Oncologico di Aviano (CRO),
IRCCS, Udine, Italy
Background: Neoadjuvant endocrine therapy with an aromatase inhibitor has
shown efficacy comparable to that of neoadjuvant chemotherapy in post- Background: It is matter of current debate which would be the best chemo-
menopausal breast cancer. Pre-/Clinical data have shown that metformin, a widely therapy backbone of neoadjuvant HER2-targeted therapy for HER2+ BC. The
used anti-diabetic drug also one of mTOR inhibitor have shown anti-tumor activity. TRAIN 2 trial showed no significant difference in terms of pathological complete
We report the result of prospective, multicenter, phase II randomized, placebo response (pCR) when anthracyclines–based (CTA) or anthracyclines–free regi-
controlled trial aiming to evaluate the direct anti-tumor effect of metformin in non- mens (CT) were combined with dual HER2 blockade. However, it remains
diabetic postmenopausal women with hormone-receptor (HR) positive breast unclear how anthracyclines may influence the relative benefit across different
cancer. Methods: 203 postmenopausal women diagnosed with hormone anti-HER2 treatments. Methods: A systematic review was conducted which
receptor positive, T1-3/N0-2 invasive breast cancer were randomized to included all phase II/III randomized clinical trials (RCTs) comparing different
24 weeks of neoadjuvant letrozole (2.5 mg/day) and either metformin (2000 neoadjuvant regimens for HER2+ BC. pCR (yT0/isN0) was the outcome of in-
mg/day) or placebo. Women with history of diabetes were excluded. Primary terest. Indirect comparisons of all combination of anti-HER2 agents with CTA or
endpoint was clinical response rate (complete, partial response by caliper). CT were estimated with a random-effects frequentist NMA. Estimated pCR rates
Secondary endpoint was pathologic complete response rate, breast conser- were inferred adopting a Bayesian NMA. Results: 17 RCTs (3933 patients) were
vation rate, percent mammographic density change. PEPI score and toxicity included. Overall, 8 arms were identified, comprising all possible combinations
profile were compared between two groups. Results: 153 intention-to treat of CTA and CT with trastuzumab (H), lapatinib (L) and dual HER2 blockade (D)
population were analyzed (72 metformin, 75 placebo group). Overall clinical but also CTA and D only. Odds ratios (OR) for pCR and 95% confidence interval
response rate was 61.4% (94/153) by caliper and did not reach statistical (CI) of selected NMA comparisons are shown in the table. Estimated rates of pCR
significance between metformin versus placebo groups (66.7% versus for each treatment and 95% credible interval (CrI) are reported in the table.
56.4%, p = 0.193). Breast conservation rate was 68.0% (100/147) (66.7% Conclusions: Through indirect comparisons, no significant pCR gain was found
versus 69.3%). Overall, 87.3% (103/118) displayed Ki67 , 10% at surgical for CTA vs CT when combined to D, H and L. In particular, considering double vs
specimen and 16.7% (21/126) had zero PEPI score. Neither Ki67% nor PEPI single-agent anti-HER2 regimens, D-CT remains superior to H-CTA, supporting a
score was different between two groups. However, among the 20 patients with possible omission of anthracyclines when dual anti-HER2 block is used. On the
core-needle biopsy after 4 weeks of medication, greater number of patients contrary, our pooled estimate suggests a more relevant role for anthracyclines
displayed Ki67 , 10% in metformin group than in placebo group (87.5% versus when comparing H-CT/A vs CTA. Moreover, we estimated a 4% pCR gain for
33.3%, p = 0.017). Patients with 4week Ki67 , 10% had higher clinical re- D-CTA vs D-CT, and an 8% higher pCR rate for H-CTA vs H-CT.
sponse rate (100% versus 57.1%, p = 0.038). Grade 3 side effects were reported
in three patients (vomiting, high blood pressure, weight loss) and no hypoglycemia NMA Comparisons OR (95% CI) Estimated pCR rates (95% CrI)
event was observed. Conclusions: 61.7% overall clinical response was achieved D-CTA vs D-CT 0.88 (0.64-1.20) D-CTA 58% (45-71%)
with 24-weeks of neoadjuvant letrozole, with numerically . 10% higher response H-CTA vs H-CT 1.22 (0.77-1.92) D-CT 54% (40-67%)
rate in letrozole+metformin group (66.7% versus 56.4%). 4-weeks Ki67 , 10% L-CTA vs L-CT 1.33 (0.77-2.30) H-CTA 44% (35-54%)
level was predictive of clinical response. With , 2% grade 3 side effects, pre- D-CTA vs H-CTA 1.39 (1.03-1.87) H-CT 36% (23-53%)
operative letrozole (with/without metformin) followed by 4-week Ki67 evaluation D-CT vs H-CTA 1.58 (1.06-2.38) L-CTA 35% (23-50%)
may indeed serve as primary choice to postmenopausal hormone receptor positive H-CTA vs CTA 2.21 (1.47-3.33) L-CT 26% (12-46%)
breast cancers. Clinical trial information: NCT01589367. H-CT vs CTA 1.82 (0.99-3.35) CTA 24% (13-42%)

578 Poster Session (Board #70), Sun, 8:00 AM-11:00 AM 579 Poster Session (Board #71), Sun, 8:00 AM-11:00 AM
Race and response to neoadjuvant chemotherapy according to MammaPrint Correlation between mutation landscape and clinical outcomes of neoadjuvant
risk. First Author: Raquel Nunes, Sidney Kimmel Cancer Center, Johns therapy in HER2-positive breast cancer patients. First Author: Ning Liao,
Hopkins University, Baltimore, MD Department of Breast Cancer, Cancer Center, Guangdong General Hospital &
Guangdong Academy of Medical Sciences, Guangzhou, China
Background: African-American (AA) women with breast cancer have a less
favorable prognosis, likely due to differences in tumor biology. This is not only Background: The standard management of early stage human epidermal
driven by the higher rate of triple negative/basal tumors in patients with AA growth factor receptor 2 (HER2) positive (+) breast cancer (BC) involves
ancestry, as worse outcome has also been seen in patients with luminal tumors. neoadjuvant therapy with combination of chemotherapy and HER2-targeted
The Neoadjuvant BReast Cancer Symphony Trial (NBRST, NCT01479101) therapy followed by surgery. However, diverse pathologic responses were
was a prospective trial that has shown an association of MammaPrint/BluePrint observed. We interrogated whether baseline genomic heterogeneity con-
(MP/BP) with a rate of pathologic Complete Response (pCR) of 2% in Luminal A tributes to the varied therapeutic responses. Methods: Capture-based tar-
with 95% Distant Metastasis Free Interval at 3 years. Here, we determine the geted sequencing using a panel consisting of 520 cancer-related genes,
MP/BP risk distribution, response to therapy, and outcome in African American spanning 1.6MB of human genome, was performed on tissue biopsy sam-
(AA) and Caucasian (Cau) patients. Methods: NBRST enrolled 1,072 breast ples, obtained prior to neoadjuvant therapy, of 33 HER2+ women with stage
cancer patients (pts) in the US (June 2011 and December 2014), median I-III BC. The median age of the cohort was 53. The correlation between
follow-up 34.9 months. The current unplanned analysis compared clinico- genomic alterations and pathologic response were analyzed by multivariate
pathological characteristics, molecular risk assignment and outcome with analysis. Results: A majority of them was diagnosed with stage II (67%, 22/
neoadjuvant chemotherapy (NACT) in AA and Cau pts. Molecular subtyping 33), while 30% (10/33) had stage III and 3% (1/33) had stage I disease.
groups were assessed by MP/BP as follows: Luminal A (MammaPrint Low Risk), 58% (19/33) were HR+ and 42% (14/33) were HR-. Mutation profiling of
Luminal B (MammaPrint High Risk), HER2 and Basal types. Results: Out of baseline samples revealed 349 mutations spanning 145 genes, with TP53,
1,072 pts, 157 (15%) were AA, and 780 (73%) were Cau. AA patients were CDK12 and PIK3CA being the top 3 most frequently mutated genes.
younger at diagnosis (52 vs 54 yrs; p = 0.016), had a higher likelihood of having Neoadjuvant regimen was comprised of trastuzumab and HER2 inhibitor
higher grade (gr 3, 65% vs 53%; p = 0.005), ER-negative (45% vs 33%; (i.e. pertuzumab or lapatinib). 15 patients used single HER2 inhibitor;18
p = 0.005) and lymph node positive tumors (71% vs 51; p , 0.001). MP/BP used dual HER2 inhibitors. Endocrine therapy was also administered to HR+
classified more AA patients as Basal type, 45% compared to 33% of Cau patients (19/33) in combination with trastuzumab and HER2 inhibitor.
patients (p = 0.004). Fewer AA patients were classified as Luminal A (15%) Complete pathologic response (pCR) was observed in 45.5% (15/33) of
compared to Cau pts (33%; p = 0.004). In multivariate analysis race was a patients. Interestingly, ROS1 copy number amplifications (CANs) were only
significant factor for higher pCR rates to NACT in AA compared to Cau pts, identified in patients achieved pCR (p = 0.033). In contrast, missense
together with PR, HER2, T-stage and Grade (HR = 1.679, 95% CI = (1.057, mutations in PIK3CA and CNAs in CCND1, FGF19, FGF3, FGF4, SPOP,
2.67), p = 0.028). The pCR rate to NACT in patients with Basal tumors was 38% HNF1B and BRIP1 showed a trend of being less likely to mutate in pCR
and similar in AA and Cau patients. In patients with hormone receptor positive patients (p values between 0.05-0.1). Previous reports have suggested that
and HER2 negative tumors, patients classified by MP/BP as Luminal A had pCR rates in HER2+ patients are associated with HR status. However, our
lower pCR (2%) compared to non-luminal A (13%) (p = 0.0015). MP low risk data revealed comparable pathologic response of patients based on either
patients had higher 3 yr DMFS (97%) than MP high risk patients (86%; HR status or neoadjuvant regimen. Conclusions: Our data revealed a distinct
p = 0.010). DMFS for AA MP Low Risk patients was 100%. Conclusions: In this mutational profile between patients achieved pCR vs patients did not.
study, MP was able to identify patients with hormone receptor positive tumors Further studies with a larger cohort are required to confirm these findings.
with low sensitivity to chemotherapy and good outcome, irrespective of race,
suggesting that this test can be helpful to characterize the tumor’s biology and
select patients who will not benefit from chemotherapy independently of their
ancestry. Clinical trial information: NCT01479101.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Breast Cancer—Local/Regional/Adjuvant 29s

580 Poster Session (Board #72), Sun, 8:00 AM-11:00 AM 583 Poster Session (Board #75), Sun, 8:00 AM-11:00 AM
Evaluation of survival by ADCC status: Subgroup analysis of SB3 (Trastuzumab Health-related quality of life in the Swedish PREDIX HER2 trial, evaluating
Biosimilar) and reference trastuzumab in patients with HER2-positive early docetaxel, trastuzumab, pertuzumab versus trastuzumab emtansine as neoadjuvant
breast cancer at three-year follow-up. First Author: Xavier Pivot, Centre Paul treatment of HER2-positive breast cancer. First Author: Yvonne Brandberg,
Strauss, INSERM 110, Strasbourg, France Karolinska Institutet, Department of Oncology-Pathology (OnkPat), Karolinska
University Hospital, Stockholm, Sweden
Background: SB3 is an approved biosimilar of reference trastuzumab (TRZ).
At additional 2-year follow-up after completing neoadjuvant and adjuvant Background: Neoadjuvant therapy combining docetaxel, trastuzumab and
treatment, there was a difference in event-free survival (EFS), but no dif- pertuzumab (DTP) was compared to trastuzumab emtansine (T-DM1) in the
ference in overall survival (OS) between SB3 and TRZ. Upon monitoring randomized phase 2 PREDIX HER2 trial. Patients, $18 years with HER2
quality attributes of TRZ, a marked downward shift in antibody-dependent positive breast cancer, $20mm or with verified lymph node metastases,
cell-mediated cytotoxicity activities (ADCC) was observed in TRZ lots with were randomized to six courses of DTP (Standard arm) or T-DM1 (Experi-
expiry dates from Aug 2018 to Dec 2019. Some of the lots were used in the mental arm). Primary endpoint was pathological objective response to pri-
Phase III study. This is a post-hoc analysis of EFS and OS by ADCC status mary medical therapy at post-treatment surgery. Health related quality of life
from a 3-year follow-up to investigate the difference in EFS between SB3 and (HRQoL) was a secondary outcome, and is of specific interest as there was no
TRZ. Methods: After completion of neoadjuvant and adjuvant therapy in difference between the randomization groups regarding the main endpoint
patients with HER2 positive early breast cancer, patients from selected (results presented in a separate abstract sent to ASCO 2019, Bergh et al.).
countries participated in a 5-year follow-up study (NCT02771795). Within the Methods: Of 202 randomized patients, 190 are available for evaluation at
TRZ group, patients exposed to at least one shifted ADCC lot and those never this point. HRQoL was measured, using EORTC QLQ-C30 + EORTC QLQ-
exposed to shifted ADCC lot during neoadjuvant period were considered as BR23, at baseline before randomization and after six courses. Results: No
“Exposed” and “Unexposed,” respectively. EFS and OS after 3-year follow-up differences between the randomization arms were found at baseline. Results
was analyzed by ADCC status in the long-term follow-up set. Results: 367 after six courses, based on 163 patients (86%) and adjusted to baseline
patients (SB3, N = 186; TRZ, N = 181) were enrolled in the follow-up study. values, revealed statistical significant differences (p#0.01), favoring the
Within TRZ, 55 patients were Unexposed and 126 patients were Exposed. At a experimental T-DM1 arm on 7 out of 15 of the EORTC QLQ-C30 variables
median follow-up duration of 40.8 months in SB3 and 40.5 months in TRZ, (Physical functioning, Role functioning, Social functioning, Global quality of
3-year EFS rates were 92.5% in SB3, 94.5% in Unexposed, and 82.5% in Life, Fatigue, Dyspnea, and Diarrhea). For the breast cancer specific
Exposed and OS rates were 97.0%, 100%, and 90.6%, respectively. Exposed questionnaire (EORTC-BR23), the experimental arm scored statistically
was associated with decreased EFS compared to Unexposed (HR 0.14, 95% significantly better on 5 out of 7 subscales (Body image, Sexual functioning,
CI 0.04-0.51, p= 0.003). There was a trend of decreased OS in Exposed Sexual enjoyment, Systemic therapy side effects and Upset by hair loss). All
compared to Unexposed, however, there was no significant difference (HR of the statistical significant differences were of moderate or large clinical
0.14, 95% CI 0.02-1.15, p= 0.068). Between SB3 and Unexposed, no significance ($10 scale scores). No differences between the randomization
difference was observed in EFS (HR 1.06, 95% CI 0.33-3.44, p= 0.923), or arms were found for the remaining HRQoL variables. Conclusions: The ex-
OS (HR 0.54, 95% CI 0.05-5.44, p= 0.600). Conclusions: Within the TRZ perimental arm reported better HRQoL than the control arm after six courses.
group, Exposed showed significantly lower EFS compared to Unexposed, Trastuzumab emtansine may be a useful treatment alternative due to better
and a similar trend was observed in OS with no statistical significance. HRQoL and lower toxicity. Clinical trial information: NCT02568839.
Between SB3 and Unexposed, no significant difference in EFS or OS was
observed. Clinical trial information: NCT02771795.

584 Poster Session (Board #76), Sun, 8:00 AM-11:00 AM 585 Poster Session (Board #77), Sun, 8:00 AM-11:00 AM
Open label, phase II trial of neoadjuvant TAK-228 plus tamoxifen in patients Tumor immune microenvironment (TiME) changes by multiplex IF staining
with estrogen receptor (ER)-positive, human epidermal growth factor receptor in a pilot study of neoadjuvant talazoparib for early-stage breast cancer
type 2 (HER2)-negative breast cancer-ANETT. First Author: Emre Koca, patients with a BRCA mutation. First Author: Evthokia Hobbs, University of
Houston Methodist Hospital, Houston, TX Texas MD Anderson Cancer Center, Houston, TX
Background: Neoadjuvant endocrine therapy is standard care for women Background: We previously reported a median tumor volume loss of 88%
with hormone receptor-positive breast cancer. However, both primary and (range 30-98%) in 13 patients with early stage BRCA1/2 mutant breast
acquired endocrine resistance is not uncommon, thereby limiting efficacy. cancer treated on a neoadjuvant trial of the PARP inhibitor talazoparib. The
[1] The PI3K-Akt-mTOR pathway is a major mediator of endocrine re- effects of PARP inhibition on immune aspects of the TiME in early-stage
sistance. [2,3] Therefore, we determined the efficacy and safety of the breast cancer has not been well described. The goal of this study was to
mTORC1/2 inhibitor TAK-228 in combination with tamoxifen in neoadjuvant evaluate the TiME in pre and post-treatment core biopsies from enrolled
setting. Methods: In this single-arm, open-label phase II trial, newly di- patients. Methods: Eleven paired core biopsies were available for exami-
agnosed patients with stage I2III ER-positive, HER2-negative breast cancer nation. Tumor infiltrating lymphocytes (TILs) were quantified by H&E
received TAK-228 (30 mg weekly) and tamoxifen (20 mg daily) for 16 weeks stained slides by a central pathologist. Specimens were assessed by mul-
until 2-4 weeks prior to surgery. The primary endpoint was the change in tiplex immunofluorescence (mIF) using a panel of 6 biomarkers (PD-1, PD-
Ki67 after 6 weeks. Secondary endpoints included pathological complete L1, CD3, CD8, CD68 and CK) with the Opal 7-color Kit in LEICA BOND auto
response rate (pCR), preoperative endocrine prognostic index (PEPI) score, stainer, Vectra automated quantitative pathology imaging system and in-
and safety. Results: Of the 28 patients enrolled in the study, 3 were excluded Form software (PerkinElmer). Results: In the analyzed core biopsies, there
due to non-compliance. Mean patient age was 51.7 years. Most patients was an increase in TILs evaluated by H&E in post-treatment compared to
had stage I or II disease (12 [43%] each); 4 (14%) had stage III disease. baseline (mean 36 vs 11%). By mIF there was an increase in CD3+ T cell and
Mean Ki67 was significantly lowered from baseline to Week 6 (17.2% vs. CD3+CD8+ cytotoxic T cell density in post-treatment samples compared to
15.2%, p = 0.0023). Interestingly, mean Ki67 increased to 20.1% from baseline, summarized in table. PD-L1 expression in tumor cells was rare in
baseline to the time of surgery. This may have been due to a rebound effect, the cohort. There was no difference in CD3+PD-1+ or CD3+CD8+ PD-1+
as TAK-228 was discontinued 2-4 weeks prior to surgery. Tumor size also lymphocytes in pre and post-treatment specimens. There was also no dif-
significantly decreased from baseline to surgery, with a median decrease of ferences in macrophages (CD68+). Evaluation of immune phenotype and
0.75 centimeters (p , 0.0001). PEPI score was intermediate risk (score imaging response will be presented in the final analysis. Conclusions: This is
123) in 6 patients and high risk group (score $4) in 15 patients. No patients the first study phenotyping the immune response to neoadjuvant talazoparib
achieved a PEPI score of 0 and no pCR was achieved. Overall, the com- in BRCA-mutant breast cancer patients. In this small cohort, intratumoral
bination was well tolerated, the most common side effects were nausea and stromal CD3+ T cells and CD3+CD8+ cytotoxic T cells increased after
(72%), vomiting (72%), fatigue (72%), mucositis (45%), and headache two months of talazoparib. Clinical trial information: NCT02282345.
(45%). The any Grade 3 AE rate was 7.7%. Conclusions: The TAK-228 and
tamoxifen combination was found to be an effective neoadjuvant strategy Pre-treatment Post-treatment
(mean density/mm2) (mean density/mm2)
with a favorable safety profile in newly diagnosed patients with hormone
receptor-positive breast cancer. Further molecular analysis (PI3K-Akt-mTOR Tumor CD3+ 164.22 668.61
pathway) are pending and will be presented. Clinical trial information: Tumor CD3+CD8+ 84.65 341.48
NCT02988986. Stroma CD3+ 743.78 1070.88
Stroma CD3+CD8+ 163.82 429.92
Tumor & stroma CD3+ 348.30 975.80
Tumor & stroma CD3+CD8+ 109.27 395.08

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
30s Breast Cancer—Local/Regional/Adjuvant

586 Poster Session (Board #78), Sun, 8:00 AM-11:00 AM 587 Poster Session (Board #79), Sun, 8:00 AM-11:00 AM
Delineating longitudinal patterns of response to neoadjuvant systemic therapy A neutrophil to lymphocyte ratio is predictive of response to neoadjuvant
(NAST) in triple-negative breast cancer (TNBC): Profiling results from a HER2-targeted therapies in the patients with HER2-positive breast cancer.
randomized, TNBC enrolling trial to confirm molecular profiling improves First Author: Soong June Bae, Gangnam Severance Hospital, Seoul, South
survival (ARTEMIS; NCT02276443). First Author: Sahil Seth, The University Korea
of Texas MD Anderson Cancer Center, Houston, TX
Background: The neutrophil to lymphocyte ratio (NLR) has been reported
Background: The heterogeneity of TNBC results in varied responses to NAST: 30-40% that is associated with response to treatment and prognosis in breast cancer,
of patients (pts) have pathologic complete response (pCR) with excellent prognosis. but its role is unclear in HER2 positive breast cancer. In this study, the
Those with residual disease, have a much higher risk of recurrence. Longitudinal relevance of NLR for treatment efficacy was analyzed in HER2 positive
profiling assesses biologic response to NAST and mechanisms of resistance.
breast cancer patients underwent neoadjuvant therapy. Methods: Pre-
Methods: Pts with stage I-III TNBC began a planned 4 cycles of Adriamycin-based
chemo (AC). Biopsies were performed pre (mandatory) and post (optional) AC. Volu- treatment NLR was assessed in 546 HER2 positive breast cancer pa-
metric change by ultrasound (VUS) at completion of AC (or progression) was calculated. tients divided into three groups according to neoadjuvant treatment regi-
Pts with sensitive disease received subsequent taxane-based (T) therapy. Pts with mens: i) chemotherapy alone, ii) chemotherapy plus trastuzumab, and iii)
insensitive disease were offered phase II trials. Pathologic response was assessed at chemotherapy plus trastuzumab and pertuzumab. The cutoff value of NLR
surgical resection in 47 pts. Matched samples, pre and post AC (N = 48 pts) underwent was defined as 2.75. We evaluated the correlation of NLR with pathologic
transcriptomic and genomic profiling. Samples were classified into six previously complete response (pCR) rate to neoadjuvant treatment regimen. Results: Of
identified ARTEMIS subtypes of TNBC (ART-Type). Immune deconvolution and esti- all patients, 422 (77.3%) patients were classified as low NLR group
mation was performed using RNA-Seq profiles. Differential pathway-level analysis was (NLR , 2.75) and 124 (22.7%) patients as high NLR group (NLR$2.75). In
performed comparing pre and post AC samples. Results: There was heterogeneity in
response to AC with 4 predominate patterns of biologic response (Table). In 48% of cases
the low NLR group, a pCR was achieved in 59 (25.7%) of 230 patients with
the ART-Type of the tumor switched after AC, with androgen receptor like (LAR) and chemotherapy alone, 69 (57.5%) of 120 patients with chemotherapy plus
immune modulatory (IM) showing greatest stability. Tumors with enrichment in EMT or trastuzumab, and 47 (65.3%) of 72 patients with chemotherapy plus
those with no significant dysregulation after AC (Groups C + D) were associated with less trastuzumab and pertuzumab (P, 0.001). In the high NLR group, a pCR was
immune modulation and lower rates of pCR compared to those with depleted EMT (A and B) achieved 13 (18.8%) of 69 patients with chemotherapy alone, 5 (20.0%) of
(8.7% vs 45.8%, p = 0.0078). Conclusions: Molecular profiling of longitudinal TNBC 25 patients with chemotherapy plus trastuzumab, and 18 (60.0%) of 30
samples reveals distinct response patterns in tumors and their micro-environments upon patients with chemotherapy plus trastuzumab and pertuzumab in the high
treatment with AC. These patterns were indicative of pathologic response in this cohort; NLR group (P, 0.001). The pCR rate of chemotherapy plus trastuzumab and
however, they require validation in a separate cohort. Clinical trial information:
chemotherapy plus trastuzumab and pertuzumab was similar, but higher than
NCT02276443.
chemotherapy alone in patients with low NLR. However, only chemotherapy
Biologic Pathologic plus trastuzumab and pertuzumab showed high pCR rate compared to che-
Response Response
Class Depleted Hallmark pCR/total % pCR motherapy alone and chemotherapy plus trastuzumab in patients with high
(n = 48) Enriched Hallmark Pathways Pathways (n = 47) (p-value) NLR, regardless of hormone receptor status. The elevated NLR was an in-
A Heme metabolism EMT, MYC, G2M 7/18 38.9% dependent predictor of low pCR rate in patients with chemotherapy plus
checkpoint
A Basophil/Neutrophil lineage (immune deconv.) (0.19) trastuzumab (OR 0.18, 95% CI, 0.07 to 0.52; P= 0.002), but not in those with
B EMT 4/6 66.7% chemotherapy alone (OR 0.67; 95% CI, 0.34 to 1.32; P= 0.248; Pinteraction=
(0.04)
C No significant change No significant 2/17 11.8% 0.041) and chemotherapy plus trastuzumab and pertuzumab (OR 0.80; 95%
change
(0.09)
CI, 0.33 to 1.92; P= 0.614; Pinteraction= 0.031). Conclusions: This study
D EMT, Inflammatory response, IL6_JAK_STAT3, MYC, G2M 0/6 0.0% identified a possibility of NLR as an easily accessible predictive marker to
Allograft rejection, Angiogenesis, Coagulation checkpoint, E2F
(0.16)
guide neoadjuvant HER2 target therapy in HER2 positive early breast cancer.
total 13/47 27.7% Further study with other cohort is needed for validation.

588 Poster Session (Board #80), Sun, 8:00 AM-11:00 AM 589 Poster Session (Board #81), Sun, 8:00 AM-11:00 AM
Correlation of the tumor mutational burden with the composition of the Heterogeneity in signaling pathway activity within primary breast cancer and
immune cell subpopulations in peripheral blood of triple-negative breast between primary and metastases. First Author: Wim Verhaegh, Philips
cancer patients undergoing neoadjuvant therapy with durvalumab: Results Research Europe, Eindhoven, Netherlands
from the prospectively randomized GeparNuevo trial. First Author: Barbara
Background: Treatment with targeted drugs aims to block tumor driving
Seliger, Martin-Luther-University Halle-Wittenberg, Halle, Germany
signaling pathway(s). Drug choice is often based on a single preoperative
Background: The GeparNuevo trial is a randomized, double-blind, multi- primary breast cancer biopsy. It is important that biopsied cancer tissue is
center phase II trial of neoadjuvant therapy in patients with early-stage triple representative for the primary tumor (PT) or metastases to treat. Little is
negative breast cancer (TNBC) investigating the role of durvalumab, an anti- known about pathway heterogeneity within the PT, and between PT and
PD-L1 antibody, which blocks PD-L1 binding to PD1 and CD80, in addition metastatic tumors. A novel analysis method was developed to identify and
to standard chemotherapy with nab-Paclitaxel (nab-Pac) followed by Epi- quantify activity of signal transduction pathways in cancer tissue, based on
rubicin plus Cyclophosphamid (EC; Loibl S et al. ASCO 2018). Since the Bayesian models that infer a pathway activity score from transcription factor
tumor mutational burden (TMB) has been suggested to be associated with a target gene mRNA levels (Cancer Res 2014;74:2936-45). Methods: Pathway
better outcome of patients undergoing immunotherapy and an increased analysis, originally developed for AffymetrixU133Plus2.0, was adapted to RT-
T cell response, we determined whether there exists a link between TMB and qPCR for use on formalin fixed paraffin embedded tissue. Estrogen (ER) and
immune cell composition, frequency and function in patients of the androgen (AR) receptor, PI3K, Hedgehog (HH), TGFb and Wnt pathway ac-
GeparNuevo trial. Methods: In order to determine possible predictive and / or tivities were analyzed. Samples were from multiple locations (“quadrants”) in
prognostic biomarkers, tumor biopsies taken at recruitment from 149 pa- 15 luminal A, 9 luminal B, and 8 ER-negative primary breast cancers; from
tients out of the 174 enrolled patients underwent deep sequencing in order subdivided quadrant samples (4 “subquadrants”) of respectively 9, 4, and 4
to determine the TMB. In addition, for 120 patients blood samples were PTs; and from 13 distant and 24 lymph node (LN) metastases of respectively 9
taken at recruitment and during different time points of treatment (after and 7 matched luminal PTs. Analysis of pathway activity score (PAS) variance
durvalumab pre-treatment, after Nab-Pac and at surgery after EC) and was performed with linear mixed models with subgroup-dependent standard
evaluated using multicolor flow cytometry by monitoring the absolute cell deviations. Results: In primary breast cancer intra-tumor PAS variance was not
counts of T cells, B cells and NK cells as well as the frequency, composition larger at macroscale (“quadrant”) than at microscale (“subquadrant”). For ER,
and functionality of different immune cell populations. Results: The TMB of AR, HH, and Wnt pathways, PAS variation was higher between distant me-
the GeparNuevo cohort was in line with published data with a mean of 1.8 tastases and PT than within the PT (p , 0.0002). For HH and Wnt pathways,
mutations/MB (range 0.02 – 7.65), respectively. Preliminary evaluation PAS variation was higher between LN metastases than within the PT (p ,
demonstrated a significant correlation of TMB with blood parameters, in 0.002). Correlation between primary and metastatic pathway activities ranged
particular with subsets of CD8+ T cells. Interestingly, the data suggest a from -0.34 for ER to 0.47/0.50 for TGFb/HH pathways. Conclusions: A single
negative correlation of TMB with the frequency of effector cells while a location tissue sample was representative for the whole primary tumor with
positive correlation exists with the effector memory cells at recruitment. In respect to signaling pathway activity, suggesting one biopsy as generally
depth analyses of a correlation with treatment arm and clinical responses are sufficient for (neo)adjuvant therapy choice. Pathway activities varied between
currently performed. Conclusions: Using this approach we hope to identify primary cancer and metastases, indicating the necessity of metastatic sample
biomarkers, which will allow a better selection of TNBC patients undergoing analysis (biopsies or liquid biopsy) to improve therapy choice.
specific immunotherapies. Clinical trial information: NCT02685059.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Breast Cancer—Local/Regional/Adjuvant 31s

590 Poster Session (Board #82), Sun, 8:00 AM-11:00 AM 591 Poster Session (Board #83), Sun, 8:00 AM-11:00 AM
Evidence that neoadjuvant anthracycline based combination chemotherapy CALGB (Alliance) 40603: Long-term outcomes (LTOs) after neoadjuvant
(NACT) in breast cancer (BC) induces phenotypical changes which guides chemotherapy (NACT) +/- carboplatin (Cb) and bevacizumab (Bev) in triple-
the optimal adjuvant therapy. First Author: Tarek Mohamed Ahmed Abdel- negative breast cancer (TNBC). First Author: William M. Sikov, Women and
Fatah, Nottingham University City Hospital NHS Trust, Nottingham, United Infants Hospital of Rhode Island, Providence, RI
Kingdom
Background: Both Cb and Bev demonstrate activity when combined with
Background: We hereby evaluated the histopathological and radiological alter- standard chemotherapy in TNBC. CALGB 40603 is a 2x2 randomized trial
ations of tumor characteristics after receiving NACT and the clinical significance of that previously demonstrated that adding Cb to NACT significantly increased
the changes of adjuvant therapy based on these findings. Methods: A pathological pathologic complete responses in the breast/axilla (pCR), while adding Bev
assessment of tumor features including ER, PR, HER2 and proliferation markers did not (Sikov, JCO 2015). Here we report 5-year LTOs and assess factors
(Ki67 and SPAG5) status in pre and post NACT tumors tissue have been centrally that influenced them. Methods: 443 patients with clinical stage II-III pre-
evaluated in two cohorts [Nottingham University Hospital (NUH; n=850) and viously untreated TNBC received 12 weeks of paclitaxel (wP) +/- Cb then
Australian cohort (n=250 patients)]. Since 2013 any change in the ER and HER2
dose-dense AC, +/- Bev before surgery. The primary endpoint was pCR.
status from negative (-) [in the pre NACT biopsies] to positive (+) [in the post NACT
surgical specimens] received additional adjuvant therapy (Endocrine therapy (ET) Analyses of LTOs (event-free survival (EFS), distant recurrence-free interval
for ER+ and Trastuzumab for HER2+ cases) in NUH. MRI volumetric and texture (DRFI) and overall survival (OS)), impact of residual cancer burden and other
changes have been assessed in 400 cases. The primary end point was disease free variables were secondary. Results: Median follow-up was 5.7 years (y); 5y
survival (DFS; median follow-up = 62 months). Results: 10% of pre NACT HER2- EFS was 70.9% (95% CI; 66.7%-75.4%), DRFI 76.3% (72.3%-80.5%)
cases had been converted to post NACT HER2+ and those cases who subsequently and OS 76.9% (72.9%-81.2%). Pretreatment clinical stage and achieving
received adjuvant Trastuzumab had achieved 92% 5-year DFS compared to those pCR correlated with LTOs, while age, race, subtype (basal-like vs. not) and
who remained HER- in post NACT specimens (58% 5-year DFS); (HR (95% CI)= 0.25 tumor grade did not. Among pCR 5y EFS was 86.4% vs. 57.5% for non-pCR
(0.08-0.80); p=0.016). While 13% of pre NACT HER2+ tumors were converted into (HR 0.28, 0.19-0.43), OS was 88.7% vs 66.5% (HR = 0.28, 0.17-0.44).
HER2- in post NACT surgical specimens and had similar 5-year DFS to those who This relationship was similar in all trial arms. Any residual disease conferred
remained post NACT HER2+ (5-year DFS= 94% vs., 87%; p=0.613). Loss of PR in poorer outcome; compared with pCR/Residual Cancer Burden (RCB) 0, EFS
the residual disease of pre NACT ER+ BC was associated with shorter 5-year DFS after HRs were 2.29 (1.32-3.97), 3.01 (1.90-4.74), and 9.67 (5.66-16.51) for
ET compared to those who remained post NACT PR+ (HR (95% CI)=2.1 (1.25-3.46); RCBI, II and III, respectively. There were no improvements in LTOs with
p=0.005). After NACT, 40% of pre NACT SPAG5+ cases were converted into post NACT Cb (EFS HR 0.99, 0.70-1.40) or Bev (EFS HR 0.91, 0.64-1.29). In an
SPAG5- and these patients had prolonged DFS compared to those who remained
exploratory analysis, receipt of $11 doses of wP was associated with better
SPAG5+ in post NACT specimens (27%) [5-year DFS=84% vs 49%; (HR (95% CI)=
3.8 (2.1-6.9); p,0.0001). A prognostic model has been generated including factors in EFS (HR 1.92, 1.33-2.77); this was particularly notable in Cb-treated arms.
table (AUC = 0.854 (95% CI) = 0.777-.0.931; p= 0.00000001]. Conclusions: We Conclusions: As expected, regardless of treatment arm pCR was associated
hereby showed evidences a change of treatment strategy based on the changes in the with markedly better LTOs, and pts with any residual disease had signifi-
tumor post NACT phenotype gives the optimal choice of treatment eg., the introduction cantly worse outcomes. The addition of Cb or Bev to standard NACT for TNBC
of HER2 targeting therapy for the conversion of HER2– to HER2+ phenotype after NACT did not improve LTOs in this trial, although it should be noted that the trial
improved DFS. Multivariate Cox regression model for 5-year DFS. was not powered for this endpoint. Omission of chemotherapy doses may
95% CI
result in poorer outcomes, especially among Cb-treated pts, which may
warrant further evaluation. Support: U10CA180821; U10CA180882;
Variables HR Lower Upper P value
Genentech; https://acknowledgments.alliancefound.org; NCT00861705
SPAG5 post NACT (high) 2.58 1.19 5.63 0.017*
MRI tumor volume reduction ( >30%) 0.38 0.16 0.89 0.026* Clinical trial information: NCT00861705.
Lymphovascular invasion 2.92 1.33 6.40 0.008*
Fibrosis 0.41 0.19 0.91 0.029
Grade (high grade) 2.62 1.19 5.74 0.016*

592 Poster Session (Board #84), Sun, 8:00 AM-11:00 AM 593 Poster Session (Board #85), Sun, 8:00 AM-11:00 AM
Preoperative checkpoint inhibition (CPI) and cryoablation (Cryo) in women Development and external validation of a deep learning model for predicting
with early-stage breast cancer (ESBC). First Author: Elizabeth Anne Comen, response to HER2-targeted neoadjuvant therapy from pretreatment breast
Memorial Sloan Kettering Cancer Center, New York, NY MRI. First Author: Manasa Vulchi, Cleveland Clinic, Cleveland, OH
Background: Checkpoint inhibition (CPI) combined with local strategies that Background: HER2-targeted neoadjuvant chemotherapy (NAC) possesses
cause local tumor destruction, such as cryo may augment tumor specific heterogeneous outcomes and currently lacks clinically-accepted markers of
immunity and improve survival. We previously demonstrated in 18 ESBC response. A means of predicting which patients will benefit prior to the
patients (pts) that pre-operative (pre-op) cryo with ipilimumab (ipi) is not only treatment could reduce toxicity and the delay to effective intervention.
safe but also generates robust local and systemic immune responses Computational analysis of MRI via a deep neural network has shown promise in
(NCT01502592). Given the added activity of dual CPI in other tumors, we identifying NAC responders among mixed receptor subtype and treatment
undertook a second pilot study of pre-op ipi/nivolumab (nivo)/cryo to confirm regimen cohorts, but faces challenges due to reproducibility across institutions
the safety of this combination and the impact on immune biomarkers. and has not yet been explored in the context of HER2-targeted therapy. Here
Methods: In both pilot studies, eligible pts had operable $1.5cm invasive we present a deep learning approach for predicting response to HER2-targeted
HER2 negative ESBC. CPI was administered 8-15d prior to, and cryo was NAC from pre-treatment MRI. Methods: 100 HER2+ breast cancer patients
performed 7-10d prior to, standard-of-care (SOC) surgery. Toxicity evalua- who received NAC with docetaxel, carboplatin, trastuzumab, and pertuzumab
tion continued for 12wks after drug administration. Blood for immune at Cleveland Clinic (CCF) and had pre-treatment contrast-enhanced MRI’s were
correlates was obtained at baseline, cryo, surgery and 2-4 weeks thereafter. included in this analysis. 49 patients achieved pathological complete response
Tumor samples were obtained at cryo and surgery. Flow-cytometry of (pCR, ypT0/is), while 51 patients retained presence of residual disease following
peripheral lymphocytes was compared to previously reported ipi/cryo re- NAC (non-pCR). 85 patients were used to train a convolutional neural network to
sponses. Results: After a median follow-up of 66 months all 18 ESBC ipi/ predict pCR based on pre- and post-contrast MRI images, and the model design
cryo pts, including 3 TNBC pts, are recurrence free. In the ipi/nivo/cryo study, was optimized based on performance within a 15 patient internal validation
the safety primary endpoint was met when 5 pts underwent SOC surgery cohort. An external, held-out testing dataset consisting of 28 patients (16 pCR,
without delay. Ipi/nivo/cryo was well tolerated overall. One pt on an aro- 12 non-pCR) imaged and treated at University Hospitals (UH) Cleveland Medical
matase inhibitor had grade 4 liver toxicity 8 weeks after surgery. One pt, Center was used to validate the performance of the model. Performance was
3 weeks after her SOC surgery, developed grade 1 hyperthyroidism, assessed by area under the receiver operating characteristic curve (AUC), accuracy,
preventing a secondary axillary dissection from proceeding as scheduled. sensitivity, and specificity. A multivariable model incorporating age, hormone re-
Robust activation of peripheral CD4+ and CD8+ T cells peaked at week 2 ceptor status, stage, and tumor size was developed and similarly evaluated.
post ipi/nivo with the majority of activated CD8+ T cells expressing PD1. Results: The neural network was able to predict the response to HER2-targeted NAC
Comparing the correlatives of the ipi/nivo/cryo study with the prior ipi/cryo in the internal validation cohort (AUC = 0.93) as well as in an independent cohort
study, we observed higher expression of activation markers (Ki-67, ICOS, from a separate institution (AUC = 0.85). This model offered superior performance
CTLA-4, LAG-3) on peripheral T cells and downregulation of suppressor compared to a multivariate clinical model, which achieved AUC = 0.67 and AUC =
cells. Conclusions: Ipi/cryo-treated pts, including 3 TNBC pts, remain re- 0.52, in internal validation and external held-out testing cohorts, respectively.
currence free after . 5y. Combining cryo with ipi/nivo preop is feasible, safe, Conclusions: Deep learning analysis of contrast-enhanced MRI could be used to
and associated with greater T cell activation than ipi/cryo alone. These better target anti-HER2 therapy by pre-treatment prediction of response.
results informed an ongoing randomized phase 2 study of pre-op ipi/nivo/cryo
versus SOC in women with residual TNBC after neoadjuvant chemotherapy Cohort AUC Accuracy Sensitivity Specificity
(NCT03546686). Clinical trial information: NCT02833233. Internal Validation Cohort - CCF .93 92% 93% 87%
External Held-out Testing Cohort – UH .85 84% 84% 87%

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
32s Breast Cancer—Local/Regional/Adjuvant

594 Poster Session (Board #86), Sun, 8:00 AM-11:00 AM 595 Poster Session (Board #87), Sun, 8:00 AM-11:00 AM
Circulating tumor DNA (ctDNA) during and after neoadjuvant chemotherapy Evaluation of predictive biomarkers for AR therapy and to identify the LAR
and prior to surgery is a powerful prognostic factor in triple-negative breast subtype of TNBC. First Author: Sahil Seth, The University of Texas MD
cancer (TNBC). First Author: Luca Cavallone, Segal Cancer Centre, Lady Anderson Cancer Center, Houston, TX
Davis Institute, Jewish General Hospital, Montreal, QC, Canada
Background: Androgen-receptor-like (LAR) triple-negative breast cancer
Background: TNBC, the most aggressive form of breast cancer, is treated (TNBC) is a subtype identified using Vanderbilt’s molecular signature. LAR
primarily with chemotherapy, even before surgery (neoadjuvant chemo- subtype has the lowest pCR rate for NACT among all TNBC subtypes (10% vs.
therapy or NAC). The prognosis and need for adjuvant therapy depends 28% for TNBC in general). We launched a clinical trial to determine the ef-
greatly on the tumor response assessed by pathology (pCR). Highly sensitive fectiveness of enzalutamide and paclitaxel (ZT) in improving this poor chemo.
and specific ctDNA assays have been shown to be of prognostic value in the response in the neoadjuvant setting for pts with anthracycline-refractory,
metastatic settingbut not yet in earlier settings. Methods: Tissue was col- androgen receptor (AR)+ TNBC (NCT02689427). However, we do not yet
lected from 26 Q-CROC-03 clinical trial TNBC patients before, during and have a robust predictive biomarker to detect an activated AR pathway and have
after NAC, prior to surgery. Whole exome sequencing on tumor tissues was not seen a robust correlation between molecular LAR subtype and AR IHC
used to select single nucleotide variants with high allele frequency (VAF), staining intensity. Methods: Molecular profiling and immunohistochemical
prioritizing TP53, to generateindividual digital droplet PCR (ddPCR) assays. analysis of key biomarkers (LAR, Ki67, and vimentin) was performed for all pts
An average of 5 variants (range 1-12) per patient were tested, for a total of enrolled in A Randomized triple negative breast cancer enrolling Trial to Confirm
121 variants. A detection threshold was defined for each variant from a pool Molecular Profiling Improves Survival (ARTEMIS; NCT02276443). Patients
of normal controls. Median follow-up was 55 months. Results: ctDNA was receive 4 cycles of AC, followed by an experimental arm or standard taxane,
detectable in 96% of patients at baseline, but 20% of the 121 variants were tailored using nuclear IHC staining. IHC staining of $30% AR+ was used as a
not detectable at any time point. At baseline, the mean VAF of all analyzed threshold for selection for enzalutamide combination arm. We evaluated the
variants, but not of TP53 variants alone, was significantly correlated (p , concordance between LAR-subtype using molecular profiling vs % AR+ cells via
0.05) with tumor factors (tumor size, stage, grade, nodal status before and at IHC. Results: As part of the clinical trial, tumors with $30% AR+ cells were
classified as LAR. In addition, we used RNA profiling to assign Vanderbilt
surgery, RCB score) but not with patient age or BRCA1/2 mutation status. 87
subtype scores, resulting in classification of 15 tumors as LAR+. We observed a
variants (74%) were detected at baseline and their VAF fell by 86% after 1
significant correlation (r=0.75) between LAR score and %AR+ cells, with 13 of
cycle of chemotherapy (T1). The detection of ctDNA at T1 was associated
15 LAR tumors having $30% AR+ cells. Among patients with high % of AR+
with DFS (p = 0.027) while the detection of ctDNA at the last post-
tumor cells, 11 received enzalutamide, with 43% (3/7) having responses (pCR or
chemotherapy pre-surgery time point (T4) was strongly associated with RCB-I). Conclusions: Comparison on numerical scores for Vanderbilt subtype
pathological complete response (pCR) and both DFS (p = 0.013) and OS(p = and IHC scores suggests $30% AR+ IHC staining as the threshold (ppv=0.65,
0.006). At this time point, 5 of 41 variants (12%) were detected in pCR npv=0.98, Table) to identify the molecular LAR subtype. We observed a trend
patients vs 42 of 80 (53%) in non-pCR, while only 6 of the 15 (40%) non- where response rate was higher in patients with $ AR+ IHC scores treated with
pCR patients had detectable TP53 variants. Interestingly, for variants de- enzalutamide; however, these results need confirmation in a larger cohort of
tected at baseline, the positive predictive value of T4 ctDNA for disease patients. Clinical trial information: NCT02689427, NCT02276443.
recurrence was 69%, similar to that of non-pCR, while the negative pre-
dictive value of no ctDNA at T4 was 89% for disease recurrence vs 80% for AR >=30% Other LAR
pCR. Conclusions: ctDNA detection after NAC prior to surgery is strongly No 127 2
predictive of disease-free survival and overall survival and is comparable to Yes 7 13
pCR as a prognostic factor in our cohort (NCT01276899). PPV 0.65
NPV 0.98
PPV, positive predictive value; NPV, negative predictive value.

TPS596 Poster Session (Board #88a), Sun, 8:00 AM-11:00 AM TPS597 Poster Session (Board #88b), Sun, 8:00 AM-11:00 AM
ADAPTcycle: Adjuvant dynamic marker-adjusted personalized therapy comparing NATALEE: Phase III study of ribociclib (RIBO) + endocrine therapy (ET) as
endocrine therapy plus ribociclib versus chemotherapy in intermediate-risk adjuvant treatment in hormone receptor–positive (HR+), human epidermal
HR+/HER2- early breast cancer. First Author: Nadia Harbeck, Breast Center, growth factor receptor 2–negative (HER2–) early breast cancer (EBC). First
University of Munich (LMU), Munich, Germany Author: Dennis J. Slamon, David Geffen School of Medicine, University of
California Los Angeles, Los Angeles, CA
Background: WSG (West German Study Group)-ADAPTcycle is a prospec-
tive, multi-center, interventional, two-arm, open-label, controlled (neo)ad- Background: RIBO is a selective inhibitor of CDK4/6 with demonstrated
juvant, non-blinded, randomized phase III trial (EudraCT 2018-003749- efficacy and is well tolerated when combined with ET in pre-/peri- and
40). It investigates whether HR+/HER2- intermediate-risk patients (pts) postmenopausal women with HR+, HER2– advanced breast cancer. Given
(about 20 % of HR+/HER2- early breast cancer, EBC) identified during these findings and considering the role of CDK4/6–Rb–E2F pathway dys-
screening (OncotypeDX and 3-week endocrine therapy (ET)) derive additional regulation in ET resistance, there is a rationale for evaluating whether RIBO +
benefit from 2-years of the CDK4/6 inhibitor ribociclib plus ET compared to ET prevents, or delays acquired resistance to ET in the adjuvant setting, to
chemotherapy (CT) (followed by adjuvant ET). Co-primary endpoints are improve invasive disease-free survival (iDFS). Methods: The phase 3 mul-
disease-free and distant disease-free survival. Methods: Starting Q1 2019 ticenter, randomized, open-label NATALEE trial will evaluate the efficacy
(enrollment 36 months, 80 sites), 5600 pts will be screened and 1670 and safety of RIBO + ET as adjuvant treatment in patients with HR+, HER2–
randomized in a 3:2 ratio (1002 to ribociclib + ET; 668 to standard CT followed EBC. Eligible women (any menopausal status) and men aged $ 18 years will
by ET). Pre-/postmenopausal pts with histologically confirmed invasive HR+/ be randomized to RIBO 400 mg/day (3 weeks on/1 week off) + ET or ET
HER2- EBC at clinically enhanced risk (cT2-4 or Ki67 . 20 % or G3 or cN+) alone. In both arms, ET will comprise daily continuous letrozole 2.5 mg/day
are eligible if they fulfill the ADAPT intermediate-risk group criteria: Re- or anastrozole 1 mg/day; men and premenopausal women will also receive
currence Score (RS) # 25 and poor endocrine response or RS . 25 and good goserelin 3.6 mg once every 28 days. Treatment with RIBO will last 36 months
endocrine response in p/cN0-1 pts or RS # 25 with good endocrine response in whereas treatment with ET (in both arms) will last 60 months. Patients must
c/pN2-3 pts. Endocrine responsiveness is determined by Ki67 response (drop have had American Joint Committee on Cancer (8th ed.) Anatomic Stage II
to # 10 %) after 3-week ET. Treatment duration is 2 years for the ribociclib + (either N0 with grade 2-3 and/or Ki67 $ 20% or N1) or III EBC, with an initial
ET (premenopausal: AI + GnRH) arm and 16-24 weeks for the CT arm; diagnosis # 18 months prior to randomization, and completed chemotherapy
treatment can be given in the neoadjuvant or adjuvant setting. 5-year follow-up and radiotherapy (if indicated). Patients receiving standard (neo)adjuvant ET
consists of standard adjuvant ET. Patient reported outcomes (ePROs) are are eligible only if this treatment was initiated within 12 months of randomi-
collected using CANKADO; ECG monitoring is performed using a novel zation. Key exclusion criteria include previous CDK4/6 inhibitor treatment and
CANKADO-based methodology. For translational analyses, tumor tissue will clinically significant, uncontrolled heart disease and/or cardiac repolarization
be collected at baseline (prior to ET), after 3-weeks ET (+/- 1w). Additional abnormality. The primary endpoint is iDFS using STEEP (Standardized Defi-
samples are required if residual tumor is diagnosed in case of neoadjuvant nitions for Efficacy End Points) criteria as assessed by the investigator; sec-
treatment and at time of recurrence. Exploratory tissue biomarker research ondary endpoints include recurrence-free survival, distant DFS, overall survival,
will be conducted to assess alterations of molecular markers (e. g., ESR1, patient-reported outcomes, and RIBO pharmacokinetics. Safety and tolerability
PIK3CA, CCND1, CDKN2A, RB1). Circulating DNA and tumor cells from will also be evaluated. Estimated enrollment is 4000 patients from 425
blood samples will be used to assess mutations, gene expression, etc. sites in 21 countries. Recruitment is ongoing. Clinical trial information:
Clinical trial information: 2018-003749-40. NCT03701334.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Breast Cancer—Local/Regional/Adjuvant 33s

TPS598 Poster Session (Board #89a), Sun, 8:00 AM-11:00 AM TPS599 Poster Session (Board #89b), Sun, 8:00 AM-11:00 AM
ALEXANDRA/IMpassion030: A phase III study of standard adjuvant chemotherapy ABC trial (A011502): A randomized phase III double-blinded placebo
with or without atezolizumab in early-stage triple-negative breast cancer. First controlled trial of aspirin as adjuvant therapy breast cancer. First Author:
Author: Heather L. McArthur, Cedars-Sinai Medical Center, Los Angeles, CA Wendy Y. Chen, Dana-Farber Cancer Institute, Boston, MA
Background: Early stage triple negative breast cancer (TNBC) is associated Background: In-vitro and in-vivo evidence suggest that aspirin may have an
with a high risk of distant relapse. Because TNBC does not currently have anti-tumor effect. Multiple epidemiologic studies have reported improved
specific targeted agents approved for use in the early setting it is treated breast cancer survival among regular aspirin users compared to non-users.
primarily with chemotherapy. TNBC may be more immunogenic than other Pooled data from randomized trials of aspirin for cardiovascular disease have
subtypes of breast cancer and promising clinical activity has been reported also reported a decreased risk of metastatic cancer among aspirin users.
with the anti–PD-L1 antibody, atezolizumab, in Phase 1/1b metastatic However, the exact benefits and risks for breast cancer survivors need to be
TNBC trials. Furthermore, the randomized phase 3 IMpassion130 study confirmed in a randomized controlled trial. Even if the clinical effect were
demonstrated enhanced anti-tumor activity when atezolizumab was co- modest, the global impact would be substantial since aspirin is inexpensive
administered with chemotherapy in the first line metastatic setting, with and widely available. Methods: The primary objective is to compare the
benefit mainly observed in PD-L1+ cohort. ALEXANDRA/IMpassion030 will effect of 300 mg daily aspirin versus placebo upon invasive disease-free
evaluate the efficacy and safety of atezolizumab in combination with survival (iDFS) in high risk HER2 negative breast cancer patients. Secondary
standard anthracycline/taxane adjuvant chemotherapy in early TNBC pa- objectives include effects on overall survival, cardiovascular disease, tox-
tients. Methods: ALEXANDRA/IMpassion030 is a global, prospective, ran- icity, and adherence. A biospecimen repository will be created for correlative
domized, open-label, phase 3 trial investigating the efficacy, safety and analyses including tumor collection at baseline and blood and urine samples
pharmacokinetic profile of adjuvant atezolizumab plus standard chemo- and questionnaires assessing lifestyle factors associated with inflammation
therapy versus chemotherapy alone in early TNBC. In total, 2300 patients (pain, sleep, stress, and depression) at baseline and 2 years. Study design:
with operable stage II or III TNBC, confirmed by central pathology review, will Subjects will be randomized (1:1) to aspirin 300 mg vs placebo daily for 5
be randomized. Patients are stratified by type of surgery, nodal status, and years in a double-blinded fashion. Stratification factors include hormone
centrally assessed PD-L1 status. Adjuvant treatment will consist of weekly receptor (HR) status (positive vs negative), body mass index ( , or $ 30 kg/
paclitaxel 80 mg/m2 for 12 weeks followed by dose dense anthracycline m2), and stage (II vs III). Subjects will be followed every 6 six months while
(epirubicin 90 mg/m2 or doxorubicin 60 mg/m2) and cyclophosphamide on study drug, then annually for 10 years. Accrual goal is 2936 patients to
600 mg/m2 for 4 doses every 2 weeks or the same chemotherapy regimen (T- reach 381 iDFS events. We have 80% power to detect HR 0.75 assuming 5-
EC/AC) given concomitantly with atezolizumab 840 mg every 2 weeks fol- year iDFS on placebo of 77%. Eligibility: Eligible subjects include patients
lowed by maintenance atezolizumab 1200 mg every 3 weeks until com- aged 18-70 diagnosed with a primary invasive HER2 negative breast cancer.
pletion of 1 year of atezolizumab. The primary end-point is invasive disease- If HR positive, tumors need to be node positive and diagnosed within the past
free survival (iDFS) and secondary end-points include iDFS by PD-L1 and 10 years. If HR negative, tumors can be node positive or T2-4N0 within
lymph node status, overall survival, safety, patient functioning and health 18 months of diagnosis. Subjects who are currently anticoagulated or those
related quality of life (HRQoL). Tumor tissue and blood samples will be with a prior history of GI bleeding, atrial fibrillation, or myocardial infarction
collected for biomarker research. The first patient was enrolled on August are excluded. Subjects who regularly use aspirin (defined as $ 5 days per
2nd 2018, and approximately 430 sites are expected to be opened globally week) need to stop 30 days prior to enrollment. Updated accrual numbers
in 30 countries. Clinical trial information: NCT03498716. will be given at the time of presentation. Clinical trial information:
NCT02927249.

TPS600 Poster Session (Board #90a), Sun, 8:00 AM-11:00 AM TPS601 Poster Session (Board #90b), Sun, 8:00 AM-11:00 AM
NRG Oncology/NSABP B-51/RTOG 1304: Phase III trial to determine if KEYNOTE-756: Randomized, double-blind, phase 3 study of pembrolizumab
chest wall and regional nodal radiotherapy (CWRNRT) post mastectomy (Mx) vs placebo combined with neoadjuvant chemotherapy and adjuvant endocrine
or the addition of RNRT to whole breast RT post breast-conserving surgery therapy for high-risk, early-stage estrogen receptor–positive, human epidermal
(BCS) reduces invasive breast cancer recurrence-free interval (IBCR-FI) in growth factor receptor 2–negative (ER+/HER22) breast cancer. First Author:
patients (pts) with pathologically positive axillary (PPAx) nodes who are ypN0 Fatima Cardoso, Champalimaud Clinical Center/Champalimaud Foundation,
after neoadjuvant chemotherapy (NC). First Author: Eleftherios P. Mamounas, Breast Unit, Lisbon, Portugal
NRG Oncology, and Orlando Health UF Cancer Center, Orlando, FL
Background: Although ER+/HER22 breast cancer (BC) has better overall
Background: This phase III post-NC trial evaluates if CWRNRT post-Mx or prognosis than other subtypes, a high-risk subpopulation is characterized by
whole breast irradiation (WBI) with RNRT after BCS significantly reduces the high-grade tumors, decreased sensitivity to endocrine therapy (ET), higher
IBCR-FI rate in pts with PPAx nodes that are pathologically negative after NC. responsiveness to chemotherapy (CT), and worse prognosis. Based on prior
Secondary aims are OS, LRR-FI, DR-FI, DFS-DCIS, second primary cancer, studies, increased pathological complete response (pCR) rates after neo-
and comparison of RT effect on cosmesis in reconstructed Mx pts. Correlative adjuvant CT may have a substantial impact for patients with high-risk, early-
science examines RT effect by tumor subtype, molecular outcome predictors stage HR+/HER22 BC. KEYNOTE-756 (ClinicalTrials.gov, NCT03725059)
for residual disease, and predictors for the degree of reduction in loco- is a global, randomized, double-blind, phase 3 study of pembrolizumab (vs
regional recurrence. Methods: Clinical T1-3, N1 IBC PPAx nodes (FNA or placebo) + CT as neoadjuvant treatment followed by pembrolizumab (vs
core needle biopsy) pts complete $8 weeks of NC (anthracycline and/or placebo) + ET as adjuvant treatment for patients with high-risk, early-stage
taxane). HER2+ pts receive anti-HER2 therapy. Following NC, BCS or Mx, ER+/HER22 BC. Methods: Patients with T1c-2 cN1-2 (tumor size $2 cm)
sentinel node biopsy ($2 nodes) and/or Ax dissection with histologically or T3-4 cN0-2 grade 3, invasive, ductal ER+/HER22 BC will be stratified by
negative nodes is performed. ER/PR and HER-2neu status before NC is lymph node involvement (positive vs negative), tumor PD-L1 status (positive
required. Pts may receive appropriate adjuvant systemic therapy. Radiation [CPS$1] vs negative [CPS , 1]), ER positivity (ER+ $10% vs ER+ , 10%),
credentialing with a facility questionnaire/case benchmark is required. and anthracycline dosing schedule (every 3 weeks [Q3W] vs Q2W), then
Random assignment for Mx pts is to no CWRNRT or CWRNRT and for BCS pts randomized 1:1 to neoadjuvant treatment with pembrolizumab 200 mg Q3W
to WBI or WBI+RNRT. Statistics: 1,636 pts are to be enrolled over 5 yrs or placebo combined with paclitaxel (80 mg/m2 Q1W) for 4 cycles followed
(definitive analysis at 7.5 yrs). Study is powered at 80% to test that RT by doxorubicin (60 mg/m2) or epirubicin (100 mg/m2), each with cyclo-
reduces the annual hazard rate of events for IBCR-FI by 35% for an absolute phosphamide (600 mg/m2) Q2/3W for 4 cycles. After definitive surgery
risk reduction of 4.6% (5-yr cumulative rate). Intent-to-treat analysis with 3 (6 radiation therapy, as indicated), patients will receive adjuvant treatment
interim analyses (43, 86, and 129 events) and a 4th/final analysis at 172 of pembrolizumab (200 mg Q3W) or placebo for 9 more administrations
events. Pt-reported outcomes focusing on RT effect will be provided by 736 pts combined with ET, which can be given for up to 10 years. There will be no
before random assignment and at 3, 6, 12, and 24 mos. Accrual as of 2-7-19 is crossover between treatment arms when moving from neoadjuvant to ad-
1,164 (71.1%). Contacts: Protocol: CTSU member website https:// juvant treatment. Dual primary endpoints are pCR rate (ypT0/Tis ypN0) and
www.ctsu.org. Questions: NRG Oncology Pgh Clin Coord Dpt: 1-800-477- event-free survival (EFS). Secondary endpoints include ypT0/Tis and ypT0
7227 or [email protected]. Pt entry: OPEN at https://open.ctsu.org or the OPEN ypN0 pCR rates in all patients and all 3 pCR definitions in those with PD-L1+
tab on CTSU member website. Support: U10 CA-2166; -180868, -180822; tumors, EFS in patients with PD-L1+ tumors, overall survival, safety, and
189867; Elekta NCT01872975 Clinical trial information: NCT01872975. health-related quality of life. Interim analyses are planned. Enrollment is
currently ongoing. Clinical trial information: NCT03725059.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
34s Breast Cancer—Local/Regional/Adjuvant

TPS602 Poster Session (Board #91a), Sun, 8:00 AM-11:00 AM TPS603 Poster Session (Board #91b), Sun, 8:00 AM-11:00 AM
Cctg MA.39 tailor RT: A randomized trial of regional radiotherapy in biomarker The international collaboration of active surveillance trials for low-risk DCIS
low-risk node-positive breast cancer (NCT03488693). First Author: Wendy R. (LORIS, LORD, COMET, LORETTA). First Author: Chizuko Kanbayashi,
Parulekar, Canadian Cancer Trials Group, Kingston, ON, Canada Niigata Cancer Center Hospital, Niigata, Japan
Background: Biomarker low risk, ER positive (+), HER2 negative (-) breast Background: Retrospective data suggest breast cancer-specific survival
cancer with low burden nodal involvement may be associated with good rates with versus without surgery in patients with low-grade ductal carcinoma
outcomes (Woodward 2016, Mamounas 2017). There is conflicting data in situ (DCIS) are similar. Some DCIS patients have a low likelihood of
regarding the efficacy of regional radiotherapy after breast conserving sur- progression to invasive cancer, but predicting who is at risk has not been
gery (BCS) or mastectomy in these patients (Kyndi 2008, Whelan 2015, established. Thus, treatment with a well-balanced risk / benefit ratio has not
Poortmans 2015, Liu 2015). Our hypothesis is that the risk of recurrence in been achieved. Four active surveillance clinical trials for low risk DCIS have
patients with biomarker low risk, ER+, Her2- breast cancer and involvement commenced in the United Kingdom (LORIS), Europe (LORD), United States
of 1-3 lymph nodes where regional RT is omitted will not be inferior to the risk (COMET), and Japan (LORETTA). We aim to examine the effectiveness &
of recurrence in patients treated with regional RT. Methods: MA39 is a safety of active surveillance compared with surgical based treatment approaches
Canadian Cancer Trials Group led, NCTN sponsored, randomized phase III for low-risk DCIS patients. Methods: Non surgical approaches are of the two
study comparing breast cancer recurrence free interval (BCRFI) in patients types; active surveillance (AS) alone and AS + endocrine therapy (ET). In the
with ER+, Her2-, LN 1-3+ breast cancer that is low risk as defined by Oncotype randomized trials LORIS and LORD, the study arms are AS only, but while ET is
Dx Recurrence Score , 18. Secondary objectives include a comparison of an option in COMET, ET is mandatory in the single arm trial LORETTA. COMET
DFS, breast cancer mortality, OS, locoregional and distant recurrence free and LORETTA have broader inclusion criteria as compared to LORIS and LORD.
intervals, toxicity, arm volume and mobility measurements, patient reported In COMET, comedo necrosis is eligible. In LORETTA, findings other than cal-
outcomes and cost effectiveness. Key eligibility criteria include: age $ 40 cification on mammography (MMG) are also eligible (e.g. low echo area on breast
years; BCS or mastectomy with axillary dissection and 1-3 positive axillary ultrasound). Leaders of the four trials hold regular meetings to foster international
nodes; BCS and SLNB alone and 1-2 positive axillary nodes; mastectomy and DCIS trials collaboration to share information. LORIS Clinical trial information:
SLNB alone and only 1 positive axillary node; planned endocrine therapy $ 5 ISRCTN27544579, LORD Clinical trial information: NCT02492607, COMET
years; adjuvant chemotherapy allowed. Statistical design: The primary analysis Clinical trial information: NCT02926911, LORETTA Clinical trial information:
will be a test of non-inferiority (NI) in the intention to treat population. If the UMIN000028298 [JCOG1505]. Clinical trial information: UMIN000028298,
upper bound of a one-sided 95% interval for the hazard ratio for BCRFI is NCT02492607, NCT02926911, ISRCTN27544579.
, 1.4, NI will be declared. Using a one-sided a of 0.05 and a power of 87%, it
is anticipated that 278 events are required. With an expected 5 years of accrual LORIS LORD COMET LORETTA
and 4.5 years of follow-up, 2140 patients are needed for the final sample size. Phase III III III III*
Conduct to Date: Study activation May 30 2018. Participation as of February Study AS AS AS+ET (choice) AS+ET
2019: Registrations 64 Randomizations 26. CIRB approval for continuation Screening MMG MMG MMG MMG,US, MRI**
of MA.39 was received on January 11 2019. Clinical trial information: Nuclear grade 1 or 2 1 1 or 2 1 or 2
Comedo necrosis No No Eligible No
NCT03488693. ER N/S N/S Positive Positive
HER2 N/S N/S Negative*** Negative
Size N/S Any size Any size # 2.5 cm
Patients/Target 113/932 25/1240 182/1200 27/340
AS: Active surveillance, ET: Endocrine therapy, N/S: Not stipulated in study
protocol, *: single arm confirmatory trial, **: breast US and MRI, ***: if tested.

TPS604 Poster Session (Board #92a), Sun, 8:00 AM-11:00 AM TPS605 Poster Session (Board #92b), Sun, 8:00 AM-11:00 AM
Phase II trial of nivolumab with chemotherapy as neoadjuvant treatment in NSABP B-59/GBG 96-GeparDouze: A randomized double-blind phase III
inflammatory breast cancer. First Author: Maryann J. Kwa, New York University clinical trial of neoadjuvant chemotherapy (NAC) with atezolizumab or
Cancer Institute, New York, NY placebo in patients (pts) with triple-negative breast cancer (TNBC) followed
by adjuvant atezolizumab or placebo. First Author: Charles E. Geyer, NSABP
Background: Inflammatory breast cancer (IBC) is the most aggressive form of
Foundation and Virginia Commonwealth University Massey Cancer Center,
breast cancer with poor prognosis and is often resistant to neoadjuvant
Richmond, VA
chemotherapy with risk of early recurrence and systemic spread of disease.
PD-L1 expression in IBC is frequent (Bertucci et al. Oncotarget 2015), and Background: TNBC is associated with higher percentages of pathological
blockade of the PD-1/PD-L1 axis with checkpoint inhibitors has emerged as a complete response (pCR) to neoadjuvant chemotherapy (NAC), and women
promising treatment to enhance anti-tumor immunity and clinical response. with a pCR have a favorable prognosis. However, Liedtke (2008) and Loibl
We hypothesize that PD-1 blockade with nivolumab in combination with (2017) found that women with residual disease have a substantially higher
neoadjuvant (primary) chemotherapy will increase the rate of pathologic risk of recurrence than women with other subtypes of breast cancer. Ad-
complete response (pCR) and reduce risk of recurrence in patients with IBC. ditionally, Adams (2017) and Schmid (2017) found that therapeutic
Methods: This is a single-arm open-label multicenter phase II study of blockade of PD-L1 binding by atezolizumab has resulted in relevant anti-
nivolumab with neoadjuvant chemotherapy in patients with non-metastatic tumor efficacy. Methods: Design: This is a phase III, double blind, placebo-
IBC (n = 52) (ClinicalTrials.gov: NCT03742986). All breast cancer subtypes control trial evaluating neoadjuvant atezolizumab with NAC followed by
(based on ER/PR/HER2) will be allowed. Patients will receive nivolumab adjuvant atezolizumab in TNBC. Pts are stratified by region (North America;
360 mg IV on day 1 (21-day cycle) for four cycles in addition to standard Europe), tumor size (1.1-3.0cm; .3.0cm), AC/EC schedule (q2w; q3w), nodal
chemotherapy. Cohort 1 (patients with triple negative breast cancer or status (positive; negative), and PD-L1 status (positive; negative or indeter-
hormone receptor-positive (HR)/HER2-negative IBC) will receive nivolumab minate) then randomized 1:1 to receive atezolizumab/placebo 1200 mg IV
in combination with paclitaxel followed by doxorubicin and cyclophospha- every 3 wks concurrently with both sequential regimens of weekly paclitaxel 80
mide (AC). Cohort 2 (patients with HER2-positive IBC) will receive nivolu- mg/m2 IV for 12 doses with every 3-wk carboplatin AUC of 5 IV for 4 doses
mab in combination with a taxane (docetaxel or paclitaxel), trastuzumab, followed by AC/EC every 2-3 wks (per investigator discretion) for 4 cycles.
and pertuzumab followed by AC. All patients will then undergo mastectomy Following surgery, pts resume atezolizumab/placebo 1200 mg IV every 3 wks
followed by radiation. The primary study objective is pCR rate (ypT0/Tis as adjuvant therapy for 6 months. Radiotherapy based on local standards is co-
ypN0). Secondary objectives will be safety, tolerability and invasive administered with atezolizumab/placebo. Eligibility criteria: Centrally
recurrence-free interval. Association of correlative biomarkers with pCR confirmed ER-neg, PR-neg, HER2-neg invasive breast cancer by ASCO/
and sensitivity or resistance to therapy with the combination of nivolumab CAP guidelines. Primary tumor must be stage T2 or T3 if cN0 or cN1 with
and chemotherapy will be evaluated. Analyses will include mutational and negative biopsy or T1c, T2, or T3 if cN1 with positive biopsy or cN2 or cN3.
neoantigen load, tumor-infiltrating lymphocytes (TILs) by histopatholog- LVEF .55% and no significant cardiac history. Statistical methods: Co-
ical assessment, T-cell receptor (TCR) by immunosequencing, and im- primary endpoints are event-free survival (EFS) and pCR breast/nodes.
mune gene profiles in the tumor. PD-L1 expression in tumor tissue is not Secondary endpoints include pCR breast, overall survival, distant disease-
required for enrollment but will be assessed as a predictive marker. Clinical free survival, safety and toxicity. Trial is an academic collaboration be-
trial information: NCT03742986. tween NSABP and GBG with support from Genentech/Roche. Support:
Genentech/Roche. Clinical trial information: NCT03281954.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Breast Cancer—Metastatic 35s

1000 Oral Abstract Session, Tue, 9:45 AM-12:45 PM 1001 Oral Abstract Session, Tue, 9:45 AM-12:45 PM
SOPHIA primary analysis: A phase 3 (P3) study of margetuximab (M) + Pyrotinib combined with capecitabine in women with HER2+ metastatic
chemotherapy (C) versus trastuzumab (T) + C in patients (pts) with HER2+ breast cancer previously treated with trastuzumab and taxanes: A randomized
metastatic (met) breast cancer (MBC) after prior anti-HER2 therapies (Tx). phase III study. First Author: Zefei Jiang, The Fifth Medical Center of Chinese
First Author: Hope S. Rugo, University of California San Francisco Comprehensive PLA General Hospital, Beijing, China
Cancer Center, San Francisco, CA
Background: Pyrotinib, an irreversible pan-ErbB receptor tyrosine kinase
Background: Pretreated HER2+ MBC lacks a defined standard of care, al- inhibitor, showed promising anti-tumour activity and acceptable tolerability
though T is commonly used. M has similar HER2 binding and anti- in patients with HER2+ metastatic breast cancer (MBC) in phase 1/2 trials.
proliferative effects as T. By contrast, M’s Fc region is engineered to increase Methods: This double-blinded, multicentre, randomised phase 3 trial was
affinity for both alleles of the activating Fc receptor (FcR), CD16A, and conducted in Chinese patients with HER2+ MBC previously treated with
decrease affinity for the inhibitory FcR, CD32B. The low affinity CD16A- taxanes and trastuzumab. Patients were randomly assigned (2:1) to receive
158F allele (~85% of population) has been associated with diminished 400 mg pyrotinib or placebo orally once daily for 21-day cycles in combination
clinical response to T. In a Phase 1 trial, M demonstrated acceptable safety, with capecitabine (1000 mg/m2 orally twice daily on days 1–14). The primary
anti-tumor activity, and evidence of HER2-specific antibody and T-cell endpoint (IRC-assessed progression free survival [PFS]) was assessed in pa-
responses. Methods: SOPHIA (NCT02492711), a randomized, open-label tients who received $1 dose of study treatment. Patients whose disease
P3 trial, enrolled pts with HER2+ MBC after pertuzumab and 1–3 lines of progressed on placebo plus capecitabine received subsequent single agent
prior Tx for MBC. Pts were randomized 1:1 to M (15 mg/kg IV q3w + C) or pyrotinib. Results: Between July, 2016 and November, 2017, 279 patients
T (6 [8 for loading dose] mg/kg IV q3w + C), stratified by met sites (#2, . 2), were randomised to pyrotinib plus capecitabine (n = 185) or placebo plus
lines of Tx for met disease (#2, . 2), and C choice (standard dose cape- capecitabine (n = 94) arms. The median PFS was 11.1 months (95% CI 9.66,
citabine, eribulin, gemcitabine, or vinorelbine). Primary endpoints are 16.53) in the pyrotinib plus capecitabine arm and 4.1 months (95% CI 2.79,
central blinded PFS and OS, assessed sequentially using the stratified log- 4.17) in the placebo plus capecitabine arm. seventy-one patients in placebo
rank test. Objective response rate (ORR) was a secondary endpoint. 257 PFS plus capecitabine arm received subsequent pyrotinib, showing single-agent
events were required to provide 90% power to show PFS superiority at response rate of 38.0% (95%CI 26.7%, 49.3%) and median PFS of 5.5 months
2-sided a = 0.05. Results: Intent-to-treat analysis (536 pts: M 266; T 270) (95% CI 4.07, 6.90). The most frequent ($5%) treatment-related $ grade 3
occurred after 265 PFS events. M prolonged PFS over T (median 5.8 vs 4.9 adverse events were diarrhoea (30.8% vs 12.8% ) and hand-foot syndrome
mo, hazard ratio [HR], 0.76; 95% CI, 0.59–0.98; P= 0.033). Treatment (15.7% vs 5.3%). Conclusions: In women with HER2+ MBC previously treated
effects were more pronounced in pts with CD16A genotypes containing a with taxanes and trastuzumab, pyrotinib plus capecitabine yield statistically sig-
158F allele (median PFS 6.9 vs 5.1 mo, HR, 0.68; 95% CI, 0.52–0.90; nificant better PFS. Pyrotinib monotherapy showed anti-tumour activity. Clinical trial
P= 0.005). In 524 pts with baseline measurable disease (M 262; T 262), information: NCT02973737.
ORR was higher with M (22%; 95% CI, 17.3-27.7%) vs T (16%; 95% CI
IRC Investigator
11.8-21.0%). Safety profiles were comparable in 529 pts who received
Pyrotinib Placebo Pyrotinib Placebo
study therapy. Grade $3 AEs and serious AEs occurred in 138 (52%) and 39 +capecitabine +capecitabine +capecitabine +capecitabine
(15%) vs 128 (48%) and 46 (17%) pts on M vs T, respectively. PFS data （N= 185） （N= 94） （N= 185） （N= 94）
cutoff: 10/10/18. Conclusions: In combination with chemotherapy in pre- Median PFS, months 11.1 4.1 10.9 4.1
treated HER2+ MBC, M improves PFS over T with comparable safety. CD16A (95% CI)
genotyping suggests a differential benefit in patients with a 158F allele. OS (9.66, 16.53) (2.79, 4.17) (8.31, 12.42) (3.45, 4.24)
HR (95% CI) 0.18 (0.13, 0.26) 0.24 (0.18, 0.33)
data are maturing. Clinical trial information: NCT02492711. P value P , 0.001 P , 0.001
ORR, n (%) 127 (68.6%) 15 (16.0%) 133 (71.9%) 15 (16.0%)
(95% CI) (61.4%, 75.3%) (9.2%, 25.0%) (64.8%, 78.2%) (9.2%, 25.0%)

1002 Oral Abstract Session, Tue, 9:45 AM-12:45 PM 1003 Oral Abstract Session, Tue, 9:45 AM-12:45 PM
Neratinib + capecitabine versus lapatinib + capecitabine in patients with IMpassion130: updated overall survival (OS) from a global, randomized,
HER2+ metastatic breast cancer previously treated with $ 2 HER2-directed double-blind, placebo-controlled, Phase III study of atezolizumab (atezo) +
regimens: Findings from the multinational, randomized, phase III NALA trial. nab-paclitaxel (nP) in previously untreated locally advanced or metastatic
First Author: Cristina Saura, Vall d’Hebron University Hospital, Barcelona, triple-negative breast cancer (mTNBC). First Author: Peter Schmid, Barts
Spain Cancer Institute, Queen Mary University of London, London, United Kingdom
Background: NALA (ClinicalTrials.gov NCT01808573) is a multinational, Background: IMpassion130 evaluated atezo (anti–PD-L1) + nP vs placebo +
randomized, open-label, phase III trial of neratinib (an irreversible pan-HER nP in 1L mTNBC. The primary PFS analysis found that atezo + nP signif-
tyrosine kinase inhibitor [TKI]) + capecitabine (N+C) vs lapatinib (a re- icantly improved PFS in intent-to-treat (ITT) and PD-L1+ pts vs placebo + nP,
versible dual TKI) + capecitabine (L+C) in patients with stage IV HER2+ with efficacy driven by the PD-L1+ population. At that time, the 1st interim
metastatic breast cancer (MBC) who had received $2 prior HER2-directed OS analysis was conducted (Schmid, NEJM 2018). Here we report the 2nd
regimens for MBC. Methods: Patients were randomized 1:1 to N (240 mg qd interim OS analysis. Methods: Eligible pts had histologically documented
po) + C (750 mg/m2 bid po) or L (1250 mg qd po) + C (1000 mg/m2 bid po). locally advanced or mTNBC, ECOG PS 0-1 and tumor tissue for PD-L1
Co-primary endpoints were centrally assessed progression-free survival testing. Pts were randomized 1:1 to IV atezo 840 mg or placebo on d1 and
(PFS) and overall survival (OS). Secondary endpoints were investigator- d15 + nP 100 mg/m2 on d1, d8 and d15 of each 28-d cycle until progression
assessed PFS; objective response rate (ORR); duration of response (DoR); (stratification factors: prior taxanes, liver metastases, PD-L1 on tumor-
clinical benefit rate (CBR); time to intervention for symptomatic metastatic infiltrating immune cells [IC]). RECIST 1.1 PFS (in ITT and PD-L1+ pts)
central nervous system (CNS) disease; safety; and patient-reported health and OS (tested in ITT and, if significant, PD-L1+ pts) were co-primary
outcomes. Results: 621 patients were randomized (307 to N+C; 314 to endpoints. Results: OS data are shown (Table). As of data cutoff (Jan 2,
L+C). The risk of disease progression or death was reduced by 24% with N+C 2019), 9% of pts in the atezo + nP arm and 3% in the placebo + nP arm were
vs L+C (HR = 0.76; 95% CI 0.63–0.93; p = 0.006); 6- and 12-month PFS still on treatment. Statistical significance was not demonstrated in ITT pts,
rates were 47.2% vs 37.8% and 28.8% vs 14.8% for N+C vs L+C, respectively. but a 7.0-month improvement in median OS was observed in PD-L1+ pts
OS rates at 6 and 12 months were 90.2% vs 87.5% and 72.5% vs 66.7% for with atezo + nP (25.0 mo) vs placebo + nP (18.0 mo; HR, 0.71 [95% CI:
N+C vs L+C, respectively (HR = 0.88; 95% CI 0.72–1.07; p = 0.2086). ORR in 0.54, 0.93]). A 4.5-mo safety update (Schneeweiss, ASCO 2019, sub-
patients with measurable disease at screening was improved with N+C vs L+C mitted) showed that atezo + nP remained tolerable. Conclusions: The 2nd
(32.8% vs 26.7%; p = 0.1201), as was CBR (44.5% vs 35.6%; p = 0.0328) IMpassion130 interim OS analysis was consistent with the 1st analysis,
and DoR (HR = 0.50; 95% CI 0.33–0.74; p = 0.0004). Time to intervention confirming clinically meaningful OS benefit with atezo + nP in previously
for symptomatic CNS disease (overall cumulative incidence 22.8% vs 29.2%; untreated PD-L1+ mTNBC. Clinical trial information: NCT02425891.
p = 0.043) was delayed with N+C vs L+C. Treatment-emergent adverse events
Atezo + nP Placebo + nP
(TEAEs) were similar between arms, but there was a higher rate of grade 3
diarrhea with N+C vs L+C (24.4% vs 12.5%). TEAEs leading to neratinib/ ITT population, events/pts, n/n (%) 255/451 (57%) 279/451 (62%)
HR (95% CI); log-rank P 0.86 (0.72, 1.02); 0.078a ―
lapatinib discontinuation were lower with neratinib (10.9%) than with lapatinib Median OS (95% CI), mo 21.0 (19.0, 22.6) 18.7 (16.9, 20.3)
(14.5%). Conclusions: N+C significantly improved PFS with a trend towards 2-year OS (95% CI), % 42 (37, 47) 39 (34, 44)
improved OS vs L+C. N+C also resulted in a delayed time to intervention for Median follow-up duration, mo 18.5 17.5
symptomatic CNS disease. Tolerability was similar between the two arms, with PD-L1+ population,b events/pts, n/n (%) 94/185 (51%) 110/184 (60%)
HR (95% CI) 0.71 (0.54, 0.93) ―
no new safety signals observed. Clinical trial information: NCT01808573. Median OS (95% CI), mo 25.0 (19.6, 30.7) 18.0 (13.6, 20.1)
2-year OS (95% CI), % 51 (43, 59) 37 (29, 45)
HRs estimated per stratified Cox model. a Not significant. b PD-L1 on IC $ 1% (VENTANA
SP142 IHC assay).

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
36s Breast Cancer—Metastatic

1004 Oral Abstract Session, Tue, 9:45 AM-12:45 PM 1005 Oral Abstract Session, Tue, 9:45 AM-12:45 PM
Randomized phase II study of eribulin mesylate (E) with or without pembrolizumab Capivasertib (AZD5363) plus fulvestrant versus placebo plus fulvestrant after
(P) for hormone receptor-positive (HR+) metastatic breast cancer (MBC). First relapse or progression on an aromatase inhibitor in metastatic ER-positive
Author: Sara M. Tolaney, Dana-Farber Cancer Institute, Boston, MA breast cancer (FAKTION): A randomized, double-blind, placebo-controlled,
phase II trial. First Author: Robert Hugh Jones, Velindre Cancer Centre and
Background: Studies of checkpoint inhibitor monotherapy show only mod-
Cardiff University, Cardiff, United Kingdom
est activity in HR+ MBC. We report data from the first randomized study
comparing E plus P versus E alone in HR+/HER2- MBC. Methods: Eligible Background: The PI3K/AKT signalling pathway is frequently activated in
patients (pts) had HR+/HER2- MBC, $2 lines of hormonal therapies and 0-2 patients (pts) with estrogen receptor (ER) positive breast cancer (ER+BC) and
lines of chemotherapy for MBC. Pts were randomized 1:1 to E 1.4mg/m2 has been implicated in endocrine therapy resistance. Capivasertib (AZD5363)
intravenously (IV) on d1 and d8 with P 200 mg/m2 IV on d1 of a 21-day cycle is a highly-selective, oral, small molecule AKT inhibitor. The FAKTION trial
(Arm A) or E alone (Arm B). At time of progression, pts in arm B could investigated the addition of capivasertib to fulvestrant for postmenopausal
crossover and receive P alone. Primary endpoint was progression-free sur- women with ER+ and HER2 negative BC after relapse or disease progression on
vival (PFS). Key secondary endpoints were: objective response rate (ORR) an aromatase inhibitor (AI). Methods: FAKTION is an investigator-led, double-
and overall survival (OS). Exploratory analyses assessed the association blind, placebo-controlled, randomised phase II trial. Patients were recruited
between PFS and PD-L1 status, tumor-infiltrating lymphocytes (TILs), neutrophil- from 21 UK sites and randomly assigned (1:1) to fulvestrant 500mg (day 1 and
lymphocyte ratio (NLR), tumor mutation burden (TMB), and genomic alterations 15 of cycle 1 and day 1 only of subsequent 28 day cycles) with either cap-
by next generation sequencing on archival tissue. Results: 88 pts initiated pro- ivasertib 400mg bd or placebo (4 days on/3 days off starting C1D15) until
tocol therapy; the median age was 58, median prior lines of chemotherapy 1, prior disease progression, unacceptable toxicity or withdrawal of consent. Allocation
lines of hormonal therapy 2. Median follow-up was 6.3 months. Median PFS and was balanced by minimisation according to PIK3CA mutation and PTEN ex-
ORR were not different between Arms A and B (PFS 4.1 vs 4.2 months p = 0.38; pression status, measurable/non-measurable disease, and primary/secondary
ORR 25% and 34% respectively (p = 0.49). 14 patients initiated crossover endocrine resistance. The primary endpoint was progression-free survival
treatment with pembrolizumab; 1 patient experienced a PR (ORR 7%). All- (PFS). The trial had 90% power to detect a hazard ratio of 0.65 at the one-
cause AEs occurred in 100% of pts (G3-4, 54.6%) including 2 treatment sided 20% significance level. Secondary endpoints included overall survival
related deaths on Arm A, both from known AEs attributed to both drugs. PD-L1 (OS), objective response and clinical benefit rates, safety and the effect of
assay was performed in 65 pts: 24 (36.9%) had PD-L1 positive ( . 1% with PI3K/Akt pathway activation on PFS. Results: Between Mar 2015 and Mar
22C3, centrally tested) tumors. PD-L1 status, TILs, NLR, TMB, and genomic 2018, 140 pts were randomised to fulvestrant + capivasertib (n = 69) or
alterations were not associated with PFS (Table). Updated data, including OS fulvestrant + placebo (n = 71). In the Intention-to-treat analysis, after 112
and genomic results, will be presented. Conclusions: Among pts with HR+/ events, median PFS was 10.3 months (m) for capivasertib compared to 4.8m
HER2- MBC, the combination of E and P was not associated with longer PFS than for placebo (Hazard Ratio (HR) 0.57; 95% CI: 0.39 to 0.84; one-sided p =
E alone in the ITT or PD-L1+ population, though the PD-L1+ subgroup had very 0.0017; two-sided 0.0035). Fifty-two deaths were reported. Median OS was
limited power to assess P benefit. Clinical trial information: NCT03051659. 26.0m for capivasertib compared to 20.0m for placebo, with a survival dif-
ference starting to emerge after 12m (HR = 0.59; 95% CI: 0.34 to 1.05; two-
median PFS, months (95% CI) P value
sided p = 0.071). Toxicity data and subgroup analyses including relative
PD-L1 positive (n = 24) P+ E 4.1 (1.8 – 8.5) 0.73 capivasertib benefit by PI3K/Akt pathway alteration will be presented at the
E 4.2 (2.8 – 5.9)
TIL > 10% (n = 11) P+ E 3.2 (0.4-NA) 0.70 conference. Conclusions: The trial met its primary endpoint. Addition of cap-
E 4.1 (2.3-NA) ivasertib to fulvestrant for patients with endocrine resistant advanced breast
NLR > 4 (n = 46) P+ E 5.2 (2.0 – 6.4) 0.23 cancer resulted in significantly longer PFS and an improvement in OS. The
E 4.1 (2.1 – 4.6)
TMB > 6 (n = 28) P+ E 4.1 (2.1-5.8) 0.70 FAKTION results warrant further investigation of capivasertib for the treatment of
E 3.9 (1.6-6.2) ER positive breast cancer. Clinical trial information: NCT01992952.

1007 Oral Abstract Session, Tue, 9:45 AM-12:45 PM LBA1008 Oral Abstract Session, Tue, 9:45 AM-12:45 PM
A randomized phase II study of palbociclib plus exemestane with GNRH agonist Phase III MONALEESA-7 trial of premenopausal patients with HR+/HER22
versus capecitabine in premenopausal women with hormone receptor-positive advanced breast cancer (ABC) treated with endocrine therapy 6 ribociclib:
metastatic breast cancer (KCSG-BR 15-10, NCT02592746). First Author: Overall survival (OS) results. First Author: Sara A. Hurvitz, UCLA Jonsson
Yeon Hee Park, Samsung Medical Center, Seoul, South Korea Comprehensive Cancer Center, Los Angeles, CA
Background: Endocrine treatment is preferred recommendation by clinical
guidelines in premenopausal as well as postmenopausal women with hormone
receptor(HR)-positive, HER2-negative metastatic breast cancer(MBC). In real-
world clinical practice, however, substantial numbers of patients are treated
with chemotherapy in earlier lines based on endocrine resistance and/or on
physician’s concern of worse prognosis associated with aggressive tumor
behavior and younger age. In terms of the chemotherapy regimens, capeci-
tabine seems one of the most popular options. The purpose of this phase II
study is to assess the safety and the clinical anti-tumor activity of exemestane
plus GNRH agonist in combination with palbociclib versus capecitabine in
premenopausal HR-positive MBC patients. Methods: This is a prospective,
two-arm, randomized, multi-center open-label phase II study of the Korean
Cancer Study Group. Patients were allowed with previous 1 line of che- The full, final text of this abstract will be available at
motherapy for MBC. De Novo metastatic patients should have been treated
with tamoxifen before enrollment. Patients were randomized to chemotherapy abstracts.asco.org at 7:30 a.m. ET on Saturday, June 1.
(capecitabine 1250 ㎎/㎡twice a day from day 1 to 14 every 3 weeks) or en- Onsite at the Meeting, this abstract will be printed in the
docrine therapy combination (exemestane 25 mg for 28 days and palbociclib Monday edition of ASCO Daily News.
125 mg for 21 days every 4 weeks with GNRH agoinst). Primary endpoint was
Progression-Free Survival (PFS). Results: Among 189 patients enrolled between
2016 and 2018 from 14 centers, 184 patients were randomly assigned to
chemotherapy (n = 92) or endocrine therapy with palbociclib (n = 92). Median
age was 44 (range 28-58). De Novo MBC was found equally in both arm (30%).
During median 14 months of follow-up, median PFS was superior in endocrine
with palbociclib than in capecitabine arm [19.0 vs. 11.3 months, p = 0.0493 by
log-rank test; Hazard Ratio (HR) 0.643 (0.415-0.999), p = 0.0493]. Ap-
proximately half of the patients (51%) were treatment naı̈ve in the advanced
setting (49% for palbociclib vs. 51% for capecitabine). Grade III or more
hematologic toxicities were more common in palbociclib than in capecitabine
with statistical significance (60.9% vs. 19.2%, p , 0.0001). Diarrhea (11% vs.
38%) and Hand-Foot syndromes (1% vs. 76%) were more common in capeci-
tabine arm. Conclusions: Exemestane plus palbociclib with ovarian suppression
showed clinical benefit in terms of PFS compared with capecitabine in patients
with premenopausal ER-positive MBC. Clinical trial information: NCT02592746.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Breast Cancer—Metastatic 37s

1009 Clinical Science Symposium, Sat, 3:00 PM-4:30 PM 1010 Clinical Science Symposium, Sat, 3:00 PM-4:30 PM
Molecular profiling of ER+ metastatic breast cancers to reveal association Genomic markers of early progression on fulvestrant with or without palbociclib
of genomic alterations with acquired resistance to CDK4/6 inhibitors. First for ER+ advanced breast cancer in the PALOMA-3 trial. First Author: Ben
Author: Pedram Razavi, Memorial Sloan Kettering Cancer Center, New York, O’Leary, Royal Marsden Hospital, The Institute of Cancer Research, London,
NY United Kingdom
Background: Genomic profiling of ER+ metastatic breast cancer (MBC) has Background: Markers of early progression on endocrine therapy in combi-
revealed highly prevalent genomic alterations (e.g. ESR1, NF1, ERBB2) nation with CDK4/6 inhibitors remain limited despite the key role of these
associated with exposure to antiestrogen therapy and endocrine resistance. combinations in treating of ER+ advanced breast cancer (ABC). We investi-
It is not known whether any such acquired genomic alterations are observed gated genomic aberrations in patients treated with fulvestrant, with and
after exposure to now standard CDK4/6 inhibitors (CDK4/6i). Methods: To without palbociclib, with a circulating tumor DNA analysis of baseline plasma
identify genomic alterations associated with acquired resistance to CDK4/6i + in the PALOMA-3 trial. Methods: PALOMA-3 was a phase III trial that ran-
antiestrogen combinations, we prospectively performed tumor and matched domized 521 patients with ER+/HER2- ABC 2:1 to palbociclib plus fulvestrant
normal sequencing on 1059 ER+ breast cancers from 845 MBC patients (P+F) or placebo plus fulvestrant (F). Using baseline plasma samples, somatic
collected prior to (n = 838) or post-treatment with CDK4/6i (n = 221), in- mutations were assessed with a 17 gene panel. Copy number aberrations
cluding 110 pre- and post-treatment pairs. We performed gene enrichment (CNA) were characterized with a 14 gene panel including ~800 SNPs in 8
analyses to identify the oncogenic mutations and copy number alterations that commonly altered regions for estimation of tumor purity, the percentage of
were more frequent in post-CDK4/6i samples compared to CDK4/6i-naı̈ve plasma DNA that was derived from tumor cells. Results for mutations and
samples and further compared these results to those of post-hormone alone CNAs were available in 310 pts (203 P+F, 107 F) and were associated with
therapy datasets. Results: The post-CDK4/6i samples were collected following clinical characteristics and progression free survival (PFS) using univariable
exposure to CDK4/6i plus aromatase inhibitors (51%), plus fulvestrant (28%), and multivariable Cox proportional hazards models, including tumor purity and
or multiple/other (21%). Along with alterations previously associated with treatment as variables. Results: In the multivariable analysis of the whole
resistance to hormonal therapy alone, our analysis identified multiple genes to cohort, higher baseline tumor purity in plasma was associated with worse PFS
be specifically enriched in the post-CDK4/6i tumors. Among these, loss-of- (HR 1.20, 95% CI 1.09 – 1.32, p = 0.0001, HR per 10% increase in purity).
function alterations in RB1 were significantly enriched in the post-CDK4/6i Baseline TP53 mutation was also associated with shorter PFS (HR 1.84, 95%
compared to CDK4/6i-naı̈ve samples (7.9% vs. 2.7%, p-value = 1.5e-5). The CI 1.27 – 2.65, p = 0.0011), as was baseline FGFR1 amplification (HR 2.91,
majority of the RB1 mutations in the post-CDK4/6i tumors had loss of het- 95%CI 1.61 – 5.25, p = 0.0004). PIK3CA and ESR1 mutations had no sig-
erozygosity (LOH), while LOH was uncommon in the CDK4/6i-naive tumors. nificant association with PFS in the multivariable model. Palbociclib treatment
Additionally, multiple alterations in effectors of PI3K/AKT signaling (excluding effect was comparable with the overall trial result (HR 0.43, 95%CI 0.32 - 0.57,
PIK3CA), cell cycle (such as CDKN2A loss) and in Hippo signaling were more p , 0.0001). TP53 mutations were significantly associated with visceral (q =
frequent in the post-CDK4/6i tumors. Conclusions: This large clinico-genomic 0.046) and soft tissue/LN metastases (q = 0.042) and the number of disease
analysis of the post- and pre-CDK4/6i MBCs reveals multiple genomic lesions sites (q = 0.0086) after correction for multiple testing. Conclusions: TP53
to be enriched after exposure to CDK4/6i, highlighting therapy-related ge- mutation, FGFR1 amplification, and tumor purity in plasma identified patients at
nomic evolution as a recurrent phenomenon and identifying unique genomic risk of early progression In PALOMA-3. If validated these results could inform
subsets that have strong potential to alter the benefit of subsequent lines of future clinical trials of CDK4/6 inhibitors combinations.
therapies.

1012 Clinical Science Symposium, Sat, 3:00 PM-4:30 PM 1013 Poster Discussion Session; Displayed in Poster Session (Board #94),
Genomic correlates of response to adjuvant trastuzumab (H) and pertuzumab Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
(P) in HER2+ breast cancer (BC): Biomarker analysis of the APHINITY trial. Sun, 11:15 AM-12:45 PM
First Author: Ian E. Krop, Dana-Farber Cancer Institute, Boston, MA Phase II COLET study: Atezolizumab (A) + cobimetinib (C) + paclitaxel (P)/
nab-paclitaxel (nP) as first-line (1L) treatment (tx) for patients (pts) with
Background: We conducted a comprehensive genomic and immune-marker
locally advanced or metastatic triple-negative breast cancer (mTNBC). First
based analysis to identify prognostic and predictive biomarkers beyond
Author: Adam Brufsky, University of Pittsburgh Medical Center, Division of
clinical parameters (eg nodal status) in the setting of HP-based therapy.
Hematology Oncology, Pittsburgh, PA
Methods: APHINITY, a phase III study, randomized 4805 pt with HER2+ BC
to adjuvant chemo/H, plus P or placebo. Tumor Infiltrating Lymphocytes Background: COLET showed that the addition of C (MEK1/2 inhibitor) to P resulted in
(TILs) were analysed using published methods. A nested case-control design an increased ORR (38%; Brufsky, SABCS 2017); IMpassion130 demonstrated
with 1023 pt samples (3 controls matched to 1 distant relapse) underwent clinical benefit with the combination of PD-L1 inhibitor A and nP as 1L tx for pts with
DNA (targeted) and RNA sequencing. Inverse probability weighting with mTNBC (Schmid, N Engl J Med, 2018). We investigated the efficacy and safety of A +
Kaplan-Meier estimator weights, and models adjusted for treatment, hor- C + P/nP in pts with mTNBC, as this combination may target multiple cancer immune
mone receptor, nodal status, age and chemotherapy type were used. Results escape mechanisms simultaneously. Methods: In the multi-stage, multi-cohort
Phase II COLET study, pts with histologically confirmed mTNBC were random-
are reported descriptively with 95% CIs. Interaction p-values are unadjusted
ized 1:1 to receive 1L tx with A 840 mg IV (d1, d15) + C 60 mg qd (d3-d23) +
for multiplicity. Results: DNA-seq, RNA-seq and TIL analyses were successful
P 80 mg/m2 IV (d1, d8, d15; cohort 2) or + nP 100 mg/m2 (d1, d8, d15; cohort 3) in
in 940/1023 (92%), 974/1023 (95%), and 4313/4804 (90%) samples, 28-day cycles until progression or toxicity. The primary endpoint (EP) was confirmed
respectively. Prognosis (arms pooled): PI3K/PTEN/AKT alterations (HR 1.35; ORR per investigator-assessed RECIST 1.1. Additional EPs were DOR, PFS, OS,
CI 1.01-1.79), MYC (HR 1.61; CI 1.16-2.23) or ZNF703 amplification (HR safety and exploratory efficacy by PD-L1 status. Results: As of 10 Aug 2018 (6.5-mo
1.62; CI 1.07-2.47) suggest poorer prognosis while TOP2A amplification (HR median follow-up), 63 and 62 pts were evaluable for efficacy and safety, re-
0.49; CI, 0.32-0.74) suggests better prognosis independent of anthracycline spectively. In cohorts 2 and 3, 21 pts (66%) and 20 pts (65%) had received neo/
use. Higher mRNA expression of an immune signature (IFNG, PD-L1, CXCL9 adjuvant taxane tx, 9 pts (28%) and 6 pts (19%) had a disease-free interval of #12
(HR 0.68; CI 0.52-0.89)) and TILs (HR 0.91; CI 0.86-0.96) also suggest better mo, respectively. All pts had $1 AE; 69% and 70% had Gr 3-5 AEs and 47% and
outcomes. Predictive effects: Table shows the HR for H+P benefit versus H for 43% had serious AEs in cohorts 2 and 3, respectively. Efficacy data for all pts and by
selected markers. PAM50 subtype did not predict benefit. Conclusions: In this PD-L1 expression on tumor-infiltrating immune cells (IC $1%; PD-L1+) are
comprehensive biomarker analysis, higher levels of immune markers and HER2 summarized in the Table. Conclusions: ORRs were similar between the A + C + P arm
appeared to be associated with better prognosis and greater H+P benefit. Clinical and A + C + nP arm. Numerically higher ORR and PFS were observed in pts with PD-L1+
trial information: NCT01358877. disease. The combination’s safety profile was consistent with the known individual safety
profiles, and A did not increase toxicity. Clinical trial information: NCT02322814.
Biomarker + Biomarker - P
A+C+P A + C + nP
N HR (95% CI) N HR (95% CI) Interaction
All Ptsa PD-L1+ PD-L12 All Ptsb,c PD-L1+ PD-L12
th
TILs > 75 %ile 988 0.35 (0.19-0.65) 3325 0.95 (0.75-1.20) 0.003 (n=32) (n=16) (n=9) (n=31) (n=15) (n=11)
IFNG > 75th %ile 240 0.45 (0.24-0.83) 730 0.96 (0.71-1.30) 0.029 Confirmed ORR, n (%) 11 (34) 7 (44) 1 (11) 9 (29) 5 (33) 3 (27)
th
CXCL9 > 75 %ile 240 0.49 (0.27-0.89) 730 0.95 (0.70- 1.28) 0.054 CR 2 (6) 1 (6) 0 0 0 0
PD-L1 > 75th %ile 240 0.52 (0.28-0.99) 730 0.91 (0.68-1.23) 0.121 PR
SD, n (%)
9 (28)
11 (34)
6 (38)
3 (19)
1 (11)
5 (56)
9 (29)
16 (52)
5 (33)
9 (60)
3 (27)
5 (45)
HER2 CN > = 6 4297 0.75 (0.60-0.93) 507 1.41 (0.80-2.47) 0.039 PD, n (%) 10 (31) 6 (38) 3 (33) 3 (10) 0 2 (18)
HER2 IHC3+ 4367 0.76 (0.61-0.94) 437 1.40 (0.74-2.66) 0.075 6-mo PFS rate, % 40.5 55.6 20.0 50.1 55.3 46.0
6-mo OS rate, % 84.1 85.6 75.0 90.0 86.7 90.1
PI3K/PTEN/AKT alt 312 1.20 (0.76-1.89) 627 0.74 (0.53-1.05) 0.102
a b c
7 pts and 5 pts with unknown PD-L1 status. 3 pts not evaluable.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
38s Breast Cancer—Metastatic

1014 Poster Discussion Session; Displayed in Poster Session (Board #95), 1015 Poster Discussion Session; Displayed in Poster Session (Board #96),
Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Sun, 11:15 AM-12:45 PM Sun, 11:15 AM-12:45 PM
Pembrolizumab (P) in patients (pts) with metastatic breast cancer (MBC) Pembrolizumab (pembro) with paclitaxel (taxol) or capecitabine (cape) as
with high tumor mutational burden (HTMB): Results from the Targeted Agent early treatment of metastatic triple-negative breast cancer (mTNBC). First
and Profiling Utilization Registry (TAPUR) Study. First Author: Ajjai Shivaram Author: David B. Page, Earle A. Chiles Research Institute at the Robert W.
Alva, University of Michigan, Ann Arbor, MI Franz Cancer Center, Portland, OR
Background: TAPUR is a phase II basket study evaluating the anti-tumor activity of Background: Atezolizumab (anti-PD-L1) plus nab-paclitaxel was shown to im-
commercially available targeted agents in pts with advanced cancers with specific prove outcomes in mTBNC in a phase III clinical trial. Subjects were required to
genomic alterations. P is an immune checkpoint inhibitor and HTMB is an emerging be . 12 months from curative-intent therapy in this trial. It remains unknown
predictive biomarker for checkpoint inhibitor therapy. Results in a cohort of MBC pts whether non-taxane chemo + anti-PD-1/L1 will be beneficial in mTNBC, or
with HTMB treated with P are reported. Methods: Eligible pts had advanced cancer, whether this approach is effective in rapidly-progressing patients ( , 12 mo
no standard treatment options, ECOG PS 0-1, measurable disease and acceptable from curative-intent therapy). Methods: mTNBC patients were enrolled in a
organ function. Genomic testing was performed using commercially available tests phase Ib study of anti-PD-1 (pembro, 200mg IV q3w) plus physician’s choice
selected by sites. Pts matched to P had HTMB defined as $9 mutations/megabase chemo (cape: n = 14, 2000mg BID, 7d on/7d off; or taxol: n = 14, 80mg/m2
(Muts/Mb) by a FoundationOne test (n=20) or approved by the TAPUR Molecular
q1w). Primary/secondary objectives were to evaluate safety/tolerability (pri-
Tumor Board for other tests (n=8). A Simon two-stage design was used to test a null
mary) and RECIST1.1 response (w12). The exploratory objective was to explore
rate of 15% vs. 35% (power = 0.85; a = 0.10). If $2 of 10 pts in stage 1 have
disease control (DC) (objective response (OR) or stable disease at 16 weeks (wks) for differences in immunomodulation according to chemo choice. Mixed ef-
(SD16+)), an additional 18 pts are enrolled. If $7 of 28 pts have DC, the drug is fects models were employed to compare the longitudinal effects of chemo on
considered worthy of further study. Secondary endpoints are progression-free peripheral immune cells (flow cytometry) and T-cell diversity (Immunoseq
survival (PFS), overall survival (OS) and safety. Results: Twenty-eight female assay). Results: Enrollment of the trial is complete (n = 28), with 100% of
MBC pts were enrolled from October 2016 to July 2018. Pts received P at 2 mg/kg evaluable patients tolerating therapy (n = 22) as of 2/1/2019. Cape ORR was
(n=8) or 200 mg (n=20) IV over 30 minutes, every 3 wks. HTMB ranged from 9 to 37 43% (5 PR, 1 CR, 2 SD) with median PFS = 155d. Taxol ORR was 25% (1 CR,
Muts/Mb. Demographics and outcomes are summarized in Table (N=28). No re- 1 PR, 3 SD) with median PFS = 99d. Subjects enrolled , 12 months from
lationship was observed between #Muts/Mb and PFS or OS. Two grade 3 AEs curative-intent therapy had numerically lower response (ORR = 27%, 1 CR, 2
(weight loss and hypoalbuminemia) and 1 grade 2 SAE (urinary tract infection) were PR, 3 SD) than subjects without rapid progression (ORR = 45%, 1 CR, 4 PR, 2
reported as at least possibly related to P. Conclusions: P demonstrated anti-tumor SD). No significant differences in immunomodulation were observed according
activity in heavily pre-treated MBC pts with HTMB. Additional study of P is war- to chemo type, however both cape & taxol were associated with declines in T-cell
ranted in MBC pts with HTMB. Clinical trial information: NCT02693535. quantity (CD4 p , .02, CD8 p , .04) and Immunoseq T-cell fraction over time.
Muts/Mb, median (range) 13 (9, 37) Conclusions: Pembro plus cape or taxol is safe with encouraging efficacy,
DC rate, % (OR or SD16+) (90% CI) 37% (24%, 46%) however activity may be lower in the setting of rapid progression following
OR rate, % (CR or PR) (95% CI) 21% (8%, 41%) curative-intent chemo. Cape+pembro efficacy is favorable with no measurable
Median PFS, wks (95% CI) 10.6 (7.7, 21.1)
Median OS, wks (95% CI) 31.6 (11.9, inf) differences in immunomodulation, and therefore cape may be preferred as a
Drug-related AEs, grades 3-4 (% of pts)
Drug-related SAEs, grades 3-4 (% of pts)
7%
4%
chemo backbone in selected patients. Both cape and taxol are associated with
Median age, yrs (range) 63 (36, 78) iatrogenic declines in T-cell quantity, which may explain the observed dropoff in
ECOG Performance Status, %
0 36% anti-PD-1/L1 activity in later lines. Clinical trial information: NCT02734290.
1 64%
Prior systemic regimens, %
2 7%
‡3 93%

1016 Poster Discussion Session; Displayed in Poster Session (Board #97), 1017 Poster Discussion Session; Displayed in Poster Session (Board #98),
Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Sun, 11:15 AM-12:45 PM Sun, 11:15 AM-12:45 PM
Triplet therapy (continuous ribociclib, everolimus, exemestane) in HR+/HER22 A phase II study of abemaciclib in patients (pts) with brain metastases (BM)
advanced breast cancer postprogression on a CDK4/6 inhibitor (TRINITI-1): secondary to HR+, HER2- metastatic breast cancer (MBC). First Author:
Efficacy, safety, and biomarker results. First Author: Aditya Bardia, Carey K. Anders, Duke Cancer Institute, Durham, NC
Massachusetts General Hospital Cancer Center, Boston, MA
Background: Abemaciclib is a selective CDK4 & 6 inhibitor approved to treat
Background: The combination of CDK4/6 inhibitor (CDK4/6i) + endocrine HR+, HER2- MBC pts on a continuous dosing schedule as monotherapy or in
therapy (ET) provides consistent improvement in PFS and response rates combination with endocrine therapy (ET). Clinical data demonstrate abe-
compared with single-agent ET as first- or subsequent-line therapy in HR+, maciclib penetrates the blood brain barrier resulting in comparable con-
HER22 advanced breast cancer (ABC), but the optimal regimen postCDK4/6i centrations in tissues and plasma. Methods: JPBO is a Simon 2-stage trial
progression, including the role of continued CDK 4/6 blockade, is unclear. evaluating abemaciclib in 6 pt cohorts with BM secondary to HR+ MBC,
Methods: TRINITI-1 is a Phase I/II, open-label trial (NCT02732119) of triplet non-small cell lung cancer, or melanoma. Here, we report on HR+, HER2-
therapy: ribociclib (RIB; CDK4/6i) + everolimus (EVE; mTORi) + exemestane MBC pts. Eligible pts had $1 new or not previously irradiated measurable
(EXE; ET) in men or postmenopausal women with HR+, HER22 ABC that BM $10mm or a progressive previously irradiated BM. Pts receiving ET at
progressed on prior CDK4/6i and up to 3 lines of therapy ($ 1 ET and # 1 the time of enrollment were permitted to continue the same ET provided that
chemotherapy regimen). Phase I determined RP2D; Phase II assessed efficacy/ extracranial (EC) disease was stable $3 months and the CNS progression
safety of RIB 300 or 200 mg + EVE 2.5 or 5 mg + EXE 25 mg/day. Here we occurred on the ET. Abemaciclib was orally administered 200mg BID.
present the first results in the entire patient population who received this triplet Primary endpoint was objective intracranial response rate (OIRR; [CR+PR])
regimen and the correlation of biomarkers with outcomes. Results: As of Oc- based on Neuro-Oncology BM response assessment criteria (RANO-BM).
tober 24, 2018, 95 patients were evaluable (ET refractory and postCDK4/6i) in Secondary endpoints included intracranial clinical benefit rate, PFS, and
Phases I (n = 17) and II (n = 78). Continuous RIB + EVE + EXE demonstrated safety. Results: 58 HR+, HER2- MBC pts were enrolled and 52 pts were
clinical benefit at week 24 in 39 patients (41.1%), exceeding the predefined evaluable. Pts had a median of 4 prior systemic therapies, 75% of pts had
primary end point threshold (. 10%). ORR was 8.4% by investigator as- prior chemotherapies (0-6, median of 2), and 71% of pts had prior ET (0-4,
sessment, median PFS was 5.7 months, and 1-year PFS was 33%. AEs were median of 1), in the metastatic setting. 50% of pts had prior whole brain
consistent with known safety profile of RIB, EVE, and EXE. Most common AEs radiotherapy, 39% stereotactic radiosurgery, and 8% surgical resection of
were neutropenia (all grades, 41.7%; grade 3/4, 31.3%), stomatitis (41.7%; BM. Median time from radiation to study enrollment was 9.4 months. Out of
3.1%), and fatigue (35.4%; 1.0%). No grade 3/4 QTc prolongation was noted. the 52 evaluable patients, 3 pts had a confirmed intracranial response (6%
ctDNA genotyping revealed patients with certain tumor alterations, eg ESR1, OIRR), and 38% of pts showed a decrease in the sum of their intracranial target
had shorter median PFS vs wild-type: 3.5 vs 6.9 mo (HR 1.76, 95% CI lesions. Intracranial clinical benefit rate (CR+PR+SD persisting for $ 6 months)
1.01–3.05). Additional genomic results, including PIK3CA, will be presented. was 25%. Median PFS was 4.4 months (95% CI, 2.6-5.5). Safety and tolerability
Conclusions: TRINITI-1 met its primary efficacy end point and is the first trial to were similar to previous reports for abemaciclib. Conclusions: Abemaciclib
demonstrate clinical benefit and tolerability of continuous triplet therapy with demonstrated intracranial clinical benefit in heavily pretreated HR+, HER2-
ET + mTORi + CDK4/6i in patients with ET-refractory HR+, HER22 ABC MBC pts with BM in this study. Further evaluations are ongoing to identify ABC
postCDK4/6i progression. Tumor genomic profile might impact the clinical patients with BM who might benefit most from abemaciclib. Clinical trial in-
outcome with triplet therapy and warrants additional research to guide rational formation: NCT02308020.
therapy selection. Clinical trial information: NCT02732119.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Breast Cancer—Metastatic 39s

1018 Poster Discussion Session; Displayed in Poster Session (Board #99), 1019 Poster Discussion Session; Displayed in Poster Session (Board #100),
Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Sun, 11:15 AM-12:45 PM Sun, 11:15 AM-12:45 PM
In-depth gene expression analysis of premenopausal patients with HR+/HER22 Characterization of mutational processes in ER+ metastatic breast cancer.
advanced breast cancer (ABC) treated with ribociclib-containing therapy in the First Author: Jorge Buendia-Buendia, Broad Institute of MIT and Harvard,
Phase III MONALEESA-7 trial. First Author: Yen-Shen Lu, National Taiwan Cambridge, MA
University Hospital, Taipei, Taiwan
Background: Diverse mutational processes are active at different stages of
Background: The Phase III MONALEESA-7 study (NCT02278120) is the first cancer development and cause the mutational burden in cancer cells. These
dedicated trial of endocrine therapy (ET) 6 a cyclin-dependent kinase 4/6 processes leave distinctive signatures, characterized by varying frequencies of
(CDK4/6) inhibitor in premenopausal patients (pts) with hormone receptor– nucleotide substitutions and their flanking base-pairs. Mutational signatures
positive (HR+)/human epidermal growth factor receptor 2–negative (HER22) found in early stage breast cancer have been described, but less is known about
ABC. The study demonstrated that the addition of ribociclib (RIB) to a non- the mutational processes in metastatic breast cancer (MBC). Methods: We
steroidal aromatase inhibitor (NSAI) or tamoxifen (TAM) + goserelin (GOS) performed whole exome sequencing (WES) on 209 metastatic tumor biopsies
significantly extended progression-free survival (PFS; hazard ratio [HR] 0.55; from patients with ER+ MBC. In a subset of patients, we obtained and performed
Tripathy D, et al. Lancet Oncol. 2018). Here we present a gene expression analysis WES on the matched primary biopsy (n = 62). Decomposition of mutational
of baseline tumor mRNA from MONALEESA-7. Methods: Premenopausal pts with signatures was done by determining the linear combination of 30 previously
HR+/HER22 ABC were treated with RIB or placebo (PBO) + GOS with either an reported mutational signatures (Alexandrov et al., 2013) that most accurately
reconstructs the mutational profile of each tumor sample. For comparison,
NSAI (letrozole or anastrozole) or TAM. Baseline archival tumor samples from 360
signature decomposition was also performed on 695 treatment-naı̈ve primary
of 672 intent-to-treat (ITT) pts were evaluated for gene expression (RIB n = 185;
ER+ breast cancers form The Cancer Genome Atlas (TCGA). Results: Cohort-
PBO n = 175) using a customized NanoString nCounter® GX 800-gene panel
level comparison of signature activity between our cohort of MBC biopsies
containing relevant breast cancer, CDK, and proliferation pathway–related genes. versus the primary ER+ breast cancers from TCGA demonstrated enrichment of
Pt subgroups were classified as having low or high mRNA expression using median Signature 13 in MBC samples (q = 0.002), attributed to activity of the APOBEC
expression as the cutoff. Results: PFS benefit in the biomarker-assessed group family of cytidine deaminases and implicated in resistance to endocrine
was similar to that in the ITT population. A trend toward a more pronounced therapies. Activity of Signature 22, associated to activity of the nucleotide-
PFS benefit with RIB was observed in pts with high vs low expression of CCND1 excision repair mechanism, was depleted in MBC tumors relative to TCGA ER+
(HR 0.38 vs 0.67, respectively), IGF1R (HR 0.33 vs 0.77), and ERBB3 (HR breast tumors (q = 0.02). Decomposition of mutational signatures at the clone level
0.33 vs 0.76). The PFS benefit seen with RIB also trended to be greater in pts in the subset of patients with both metastatic and matched primary samples allowed
with low vs high expression of CCNE1 (HR 0.38 vs 0.65, respectively) and MYC us to infer signatures acquired in the metastatic setting. Among mutations in ac-
(HR 0.37 vs 0.69). The PFS benefit with RIB was similar in pts with high vs low quired clones, we found enrichment of APOBEC signatures (Sig. 2, q = 0.018; Sig.
expression of FGFR1 (HR 0.45 vs 0.61, respectively), ESR1 (HR 0.57 vs 13, q = 0.003). Conversely, activity of Signature 1, attributed to the normal aging
0.57), and tumor proliferation genes, such as MKI67 (HR 0.50 vs 0.51). process, was enriched in mutations shared between primary and MBC samples (q =
Conclusions: This is the first gene expression analysis of a large set of pre- 0.004). This suggests that mutations developed early in tumor evolution may be
menopausal pts with ABC. The benefit with RIB was generally consistent driven primarily by natural processes, while later mutations may reflect selective
across gene expression subgroups, although the magnitude varied in certain pressures imposed by therapies and cellular defects acquired during disease pro-
subsets. This analysis suggests that there may be unique resistance mecha- gression. Conclusions: The profile of somatic mutations in MBC differs from early
nisms to ET 6 CDK4/6 inhibitors in premenopausal pts with ABC. Clinical trial stage breast cancer. APOBEC-related mutations arising late during tumor evolution
information: NCT02278120. are one possible mechanism for this difference. Further studying the associations
between clinical features and therapies to mutational processes could inform design
of clinical trials and management of MBC.

1020 Poster Discussion Session; Displayed in Poster Session (Board #101), 1021 Poster Discussion Session; Displayed in Poster Session (Board #102),
Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Sun, 11:15 AM-12:45 PM Sun, 11:15 AM-12:45 PM
End-of-study analysis from the phase III, randomized, double-blind, placebo Biosimilar trastuzumab-dkst monotherapy versus trastuzumab monotherapy
(Pla)-controlled CLEOPATRA study of first-line (1L) pertuzumab (P), trastuzumab after combination therapy: Final overall survival (OS) from the phase III
(H), and docetaxel (D) in patients (pts) with HER2-positive metastatic HERITAGE Trial. First Author: Cornelius F. Waller, University of Freiburg
breast cancer (MBC). First Author: Sandra M. Swain, Georgetown University Medical Center, Freiburg, Germany
Medical Center, Lombardi Comprehensive Cancer Center, Washington, DC
Background: The multicenter, double-blind, randomized, parallel-group, phase
Background: Progression-free and overall survival (PFS and OS) were signif- 3 Heritage trial (NCT02472964) evaluated efficacy and safety of trastuzumab-
icantly improved with 1L P + H + D v Pla + H + D in 808 pts with HER2-positive dkst (Ogivri), a trastuzumab biosimilar, vs trastuzumab, plus taxane as first-line
MBC in CLEOPATRA (NCT00567190). OS was increased by an unprece- therapy for patients with HER2+ metastatic breast cancer. Overall response
dented 15.7 mo (median 56.5 mo with P + H + D v 40.8 mo with Pla + H + D; at week (wk) 24 and progression-free survival (PFS) at wk 48 have been
HR 0.68; 95% CI 0.56, 0.84; p , .001) with a median follow-up of 50 mo reported (Rugo et al, JAMA 2017; ASCO 2018). Methods: Eligible patients
[Swain et al. NEJM 2015]). Here we report the end-of-study analysis with a were randomized 1:1 to trastuzumab-dkst or trastuzumab, plus taxane. After
median follow-up of 99 mo (max 120 mo). Methods: In this descriptive 24 wks, patients with responding/stable disease continued monotherapy
analysis, OS was compared between arms using the log-rank test, stratified by per randomization. Safety and OS during maintenance, cumulative through
prior treatment status and geographic region. The Kaplan–Meier approach 36 months of follow-up from last patient on study, are described. Results: 500
was used to estimate median OS, and a stratified Cox proportional hazards patients were randomized; 343 received monotherapy after 24 wks
model was used to estimate the HR and 95% CIs. Subgroup analyses of OS (trastuzumab-dkst, n = 179; trastuzumab, n = 164). 128 patients dis-
were performed for stratification factors and other baseline characteristics. continued monotherapy (trastuzumab-dkst, n = 63; trastuzumab, n = 65);
Results: Clinical cutoff was Nov 23, 2018. Since Jul 2012, 50 pts crossed mean time to discontinuation was 19 months in both groups. Treatment-
from the Pla to the P arm. These pts are counted in the Pla arm for efficacy emergent adverse events (TEAEs) during monotherapy were similar for
analyses and up to the first dose of P for safety analyses. The OS HR was 0.69 trastuzumab-dkst (69%) and trastuzumab (73%); most were low grade and
(95% CI 0.58, 0.82), favoring P + H + D. Median OS was 57.1 mo in the P arm serious TEAE rates were 6% in both groups. Cumulative rates of TEAEs of
(402 pts) and 40.8 mo in the Pla arm (406 pts; D 16.3 mo). The 8-year special interest were similar for hypersensitivity, pulmonary, and cardiac
landmark OS rates were 37% and 23%, respectively. The OS benefit in events (trastuzumab-dkst, 23%, 16%, and 5%; trastuzumab, 24%, 13%,
predefined subgroups, including in pts previously treated with H in the (neo) and 5%). Incidences of left ventricular ejection fraction (LVEF) , 50% $1
adjuvant setting (88 pts, HR 0.86; 95% CI 0.51, 1.43), remained consistent time postbaseline (trastuzumab-dkst, 5%; trastuzumab, 3%) and LVEF
with the overall result and previous reports. The overall safety profile of P + H + D , 50% postbaseline and $10% reduction (trastuzumab-dkst, 4%; tras-
was consistent with the known P safety profile. There was only one new serious tuzumab, 3%) were low. At 36 months, 169 patients had received further
adverse event suggestive of congestive heart failure (onset ~77 mo on treatment lines of therapy, with similar distribution of HER2 treatments, endocrine
in the P arm, resolution in 34 days, pt continued on study medication) and one therapies, and chemotherapies. Final median PFS was 11.1 months in
new symptomatic left ventricular systolic dysfunction (onset ~46 mo after both groups. Median duration of response was 9.9 and 9.8 months and
crossing to the P arm, resolution in 34 days) since the previous analysis. median OS was 35.0 and 30.2 months for trastuzumab-dkst and trastu-
Conclusions: The OS improvement with 1L P + H + D v Pla + H + D for pts with zumab, respectively: unstratified hazard ratio, 0.9 (95% CI, 0.70-1.17).
HER2-positive MBC was maintained after an additional 4 years of long-term Conclusions: Long-term safety data with similar median OS compared with
follow-up, as were the safety and cardiac safety profiles. Clinical trial in- originator trastuzumab further support the safety and efficacy profile of
formation: NCT00567190. trastuzumab-dkst. Clinical trial information: NCT02472964.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
40s Breast Cancer—Metastatic

1022 Poster Discussion Session; Displayed in Poster Session (Board #103), 1023 Poster Discussion Session; Displayed in Poster Session (Board #104),
Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Sun, 11:15 AM-12:45 PM Sun, 11:15 AM-12:45 PM
Genomic landscape of de novo stage IV breast cancer. First Author: Immunotherapy predictive biomarkers in metastatic breast cancer (MBC).
Ana Christina Garrido-Castro, Dana-Farber Cancer Institute, Boston, MA First Author: Jeffrey S. Ross, SUNY Upstate Medical University, Syracuse, NY
Background: Genomic profiling of primary and recurrent metastatic breast cancer Background: We queried whether comprehensive genomic profiling (CGP) of mBC subtypes
(rMBC) has revealed potential resistance mechanisms to therapy. In contrast, de could identify biomarkers that have been linked to responsiveness to immunotherapy (IO)
novo stage IV breast cancer (DNIV) represents an opportunity to elucidate metastatic treatments. Methods: DNA was extracted from 3,871mBC: 1,388 ER+/HER2-, 1,969 HER2
amplified (amp) and 643 TNBC. CGP was performed using a hybrid-capture assay. Tumor
drivers in the absence of treatment selection. Methods: Targeted NGS (Oncopanel, mutational burden (TMB) was determined on 1.1 Mbp of sequenced DNA and microsatellite
OP) using multiplexed copy number variation (CNV) and mutation (mut) detection instability (MSI) was determined on 114 loci in 564 cases. PD-L1 was determined by IHC (Dako
across the full coding regions of 300 genes and selected intronic regions of 35 genes 22C3). Results: Patient ages were similar. Genomic alterations (GA)/tumor were similar ranging
was performed on either primary or metastatic samples collected in patients (pts) from 5.9 to 7.3. MTOR pathway targets were common; lowest in TNBC. CDH1 and ESR1 GA most
with DNIV or rMBC. Mut/CNV in primary and metastatic tumors were compared per frequent in ER+/HER2- cases. ERBB2 short variant (SV) mut were most frequent in ER+/HER2-
subtype between DNIV and rMBC using Fisher´s exact test (significant p,0.05). and HER2 amp and not seen in TNBC. Other kinase targets were uncommon except for FGFR1 GA
FDR were computed (q,0.25). Results: Between 8/2013-9/2016, of 929 pts who in ER+/HER2-. BRCA1/2 GA least frequent in HER2 amp. AR was amplified in 1% of all mBC.
Markers of potential IO benefit: CD274 (PD-L1) amp (1-3%), BRAF GA (1-4%), TMB of $10
underwent OP testing 212 presented with DNIV; 136 HR+/HER2- (64%); 35 HR+/ mut/Mb (8-12%), TMB of $20 mut/Mb (2-3%), MSI-High (0.1-0.4%), PBRM1 GA (1%) and low
HER2+ (17%); 25 TNBC (12%); 16 HR-/HER2+ (8%). In 168 (79%) pts, the (1-10%) or high (0-3%) PD-L1 staining. Potential markers of resistance: inactivating GA in
primary was tested; 44 had a metastatic site tested. Comparison of primary HR+/ STK11 (1-2%) and MDM2 amplification (3-6%). Examples of mBC patients responding to IO
HER2- tumors showed that DNIV pts were more likely to harbor mut in CDKN1B, therapies will be presented. Conclusions: In addition to guiding targeted therapy selection, CGP
SETD2 and PMS2 and less likely to have TP53 mut than rMBC (Table). Metastases shows potential to identify GA linked to response and resistance to IO in mBC. The demonstrations
in HR+/HER2- DNIV (n=29) had higher mut in CDH1, PTCH1 and CTNNB1 and of clinical benefit of immunotherapy in mBC supports the need for the development of biomarkers
fewer CCND1 amplification (amp) than rMBC (n=121), albeit these findings lost used to guide the use of ICPI drugs for these patients.
significance after FDR correction. DNIV primary TNBC (n=19) was significantly ER+/HER2- HER2 Amp TNBC
enriched for CIITA mut (26% vs. 0%; q=0.046) and MYB amp (21% vs. 0%, Cases 1,259 1,969 643
q=0.098) compared to rMBC (n=101). TP53 mut, amp in RAD21, MYC, MYB, Age (range in years)
GA/tumor
55 (23-89)
6.1
55 (20-89)
7.3
53 (20-85)
5.9
PTK2 and EGFR, and deletions in CDKN2A/2B and MAP2K4 significantly predicted MTOR GA PIK3CA PIK3CA 38% PIK3CA 19%
PTEN PTEN 5% PTEN 15%
poorer overall survival in DNIV. Conclusions: DNIV primary and metastatic tumors NF1 NF1 7% NF1 8%
have distinct genomic profiles compared to rMBC. Alterations in genes involved CDH1 GA
ESR1 GA
7%
16%
4%
6%
3%
0.5%
in epigenetic modulation (KMT2D, SETD2) and epithelial-mesenchymal BRCA1/2 GA 3%/6% 2%/3% 7%/3%
HER2 amp 0% 100% 0%
transition (CDH1, PTCH1, CTNNB1) are more prevalent in HR+/HER2- HER2 SV 3% 7% 0%
DNIV. DNIV TNBC is enriched for CIITA mut, described to promote im- Other Kinase targets FGFR1 18%
FGFR2 2%
FGFR1 11%
FGFR2 1%
FGFR1 8%
FGFR2 4%
mune escape via reduced MHC class II expression. If validated, these findings EGFR 2% EGFR 3% EGFR 4%
KIT 1% KIT 2% KIT 2%
may provide insight into mechanisms underlying metastatic potential. MET 0.4% MET 1% MET 1%
BRAF 2% BRAF 1% BRAF 4%
Primary HR+/HER 2- tumors rMBC (N=232) DNIV (N=86) AR amp 1% 1% 1%
MSI High 0.2% 0.1% 0.4%
MUT N % N % p value CD274 (PD-L1) amp 1% 1% 3%
TMB > 10 mut/Mb 8% 12% 9%
SETD2 3 1.3 8 9.3 0.001 TMB > 20 mut/Mb 3% 2% 2%
CDKN1B 4 1.7 6 7.0 0.027 PD-L1 IHC low/high 1%/0% 5%/.5% 10%/3%
PBRM1 GA 1% 1% 1%
PMS2 2 0.9 4 4.7 0.048 STK11 GA 1% 1% 2%
KMT2D 13 5.6 11 12.8 0.052 MDM2 amp 6% 5% 3%
TP53 76 32.8 18 20.9 0.052

1024 Poster Discussion Session; Displayed in Poster Session (Board #105), 1025 Poster Session (Board #106), Sun, 8:00 AM-11:00 AM
Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Effect of HER2/neu 655 polymorphism on trastuzumab-induced cardiotoxicity
Sun, 11:15 AM-12:45 PM in HER2-positive breast cancer patients. First Author: Isabel Blancas, Hospital
mTORC1 activation assessed in metastatic sample to predict outcome in Universitario San Cecilio, Granada, Spain
patients with metastatic breast cancer treated with everolimus-exemestan:
Background: HER2 overexpression in breast cancer is associated with a poor
Results from the SAFIRTOR study. First Author: Thomas Denis Bachelot,
outcome, high risk of metastasis and a reduced overall survival. The intro-
GINECO-Centre Léon Bérard, Lyon, France
duction of Trastuzumab in the treatment scheme improved the prognosis of
Background: Using samples from TAMRAD study (Treilleux, Ann Oncol, these patients. Nevertheless, around 20% of patients develop cardiotoxicity.
2015), we previously reported that p4EBP1, a downstream protein of mTOR, The purpose of this study is to analyze the association of the HER2 Ile655Val
was associated with higher benefit to everolimus (eve). SAFIRTOR study was A . G polymorphism with trastuzumab-induced cardiotoxicity and with sur-
designed to validate clinical utility of this biomarker. Methods: Patients (pts) vival. Methods: The study included 93 patients recruited from San Cecilio
with ER+, HER2 negative, AI resistant MBC were prospectively included University Hospital in Granada (Spain) who were treated intravenously with
(NCT02444390). All pts had a biopsy of a metastatic site and were then Trastuzumab. The cardiotoxicity was diagnosed during the treatment follow-up
treated with standard eve + exemestane (exe) combination. The primary end considering a decrease of the left ventricular ejection fraction (LVEF) from
point was to validate that p4EBP1 expression is associated with longer PFS baseline or clinical manifestation of congestive heart failure. HER2 655 A . G
in patients treated with eve. 120 evaluable pts were needed for the pre was genotyping using TaqMan SNP technology. Results: Genotype frequen-
planed statistical analysis. All samples were collected and processed in a cies of HER2/neu 655 met Hardy-Weinberg equilibrium (p= 0.363). We did
standardized procedure in order to allow phophoproteins IHC staining. In not find differences in baseline LVEF in patients who developed cardiotoxicity
addition to p4EBP1, we explored prognostic value of pS6K, pAkt, PTEN and vs those who did not. However, in cardiotoxicity group, the symptomatic
LKB1, together with genomic alterations assessed by NGS and CGH arrays. patients had a baseline LVEF significantly lower than non-symptomatic
Results: 150 pts were included, 30 pts had no adequate sample, and further patients (57.7 vs 66.1, p , 0.028). Logistic regression analysis adjusted
13 had missing clinical data, 107 were evaluable for primary objective. by hormonal status and anthracycline treatment showed significant dif-
Median age was 62, they had previously progressed on AI treatment, either in ferences between AG and AA (OR = 3.00, CI95% 1.07-8.41, P= 0.037) or
the adjuvant (22 pts) or the metastatic setting (83 pts). 20 were considered AG and AA+GG (OR = 3.21, CI95% 1.15-8.96, P= 0.026). However, we
as primary hormone resistant, 87 as secondary resistant. The median Allread did not find association between Her2/neu Ile655Val polymorphism and
score for p4EBP1 was 5.5 (range: 0-6.5). Analysis of the primary endpoint disease-free survival or global survival. Conclusions: HER2 655 A . G
showed that p4EBP1 staining above the median is associated with a longer polymorphism is significantly associated with higher risk of trastuzumab-
PFS on eve+exe. (median PFS: 9.3 months, 95CI 6.3-13.1 for high p4EBP1 induced cardiotoxicity but is not associated to a differential survival rate.
versus 5.8 months, 95CI 3.7-7.8 for low p4EBP1, p = 0.02). Prognostic
value of high pEBP1 remained significant when assessed in a multivariate
analysis along classical clinico-biological prognostic factors for MBC (HR
0.57, 95%CI 0.38-0.88, p = 0.01). In this AI resistant population, the tumor
of 42 (46%), 33 (35%) and 5 (5.3%) pts carried an activating mutation for
ESR1, PIK3CA and AKT1, respectively. None of these mutational statuses
were correlated to outcome. Conclusions: This prospective study validates
p4EBP1 expression analysis to select patients most likely to benefit from
everolimus + exemestane. Clinical trial information: NCT02444390.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Breast Cancer—Metastatic 41s

1026 Poster Session (Board #107), Sun, 8:00 AM-11:00 AM 1028 Poster Session (Board #109), Sun, 8:00 AM-11:00 AM
A phase III, randomized, double-blind, placebo (Pla)-controlled study of Trastuzumab emtansine (T-DM1) and ribociclib, an oral inhibitor of cyclin
pertuzumab (P) + trastuzumab (H) + docetaxel (D) versus Pla + H+ D in dependent kinase 4 and 6 (CDK 4/6), for patients with metastatic HER2-
previously untreated HER2-positive locally recurrent/metastatic breast cancer positive breast cancer. First Author: Laura Spring, Massachusetts General
(LR/MBC) (PUFFIN). First Author: Binghe Xu, National Cancer Center/National Hospital Cancer Center, Boston, MA
Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of
Background: Current therapeutic efforts in the management of HER2+ MBC
Medical Sciences, and Peking Union Medical College, Beijing, China
focus on targeting the HER receptor family, however co-inhibition of targets
Background: In CLEOPATRA (NCT00567190), adding P to H + D significantly im- downstream of the HER2 pathway, such as cyclin D-CDK 4/6, could enhance
proved progression-free and overall survival (PFS/OS) v Pla + H + D in patients (pts) with therapeutic efficacy. We conducted a phase 1b study of the CDK4/6 inhibitor
previously untreated HER2-positive LR/MBC. PUFFIN (NCT02896855) is a China ribociclib and T-DM1 in patients (pts) with HER2+ MBC. The primary ob-
bridging study; the objective being to assess consistency of efficacy with CLEOPATRA. jective was to determine the recommended phase 2 dose (RP2D) of ribo-
Methods: Pts with previously untreated HER2-positive LR/MBC were randomized 1:1 to P +
ciclib plus T-DM1. Secondary objectives included safety, PK assessments,
H + D or Pla + H + D, stratified by visceral v non-visceral disease and hormone receptor
status. The primary endpoint was investigator-assessed PFS. Secondary endpoints and efficacy. Methods: Pts with HER2+ MBC who previously received at
included objective response rate (ORR in pts with measurable baseline disease), OS, least 1 taxane and trastuzumab-containing regimen in any setting were
and safety. The target sample size (240) was determined based on the consistency eligible. Pts with previous CDK4/6 inhibitor exposure, QTcF . 450msec, or
threshold for PFS, defined as hazard ratio (HR) , 0.81, which maintains $ 50% of the unstable brain metastases were excluded. Ribociclib was given orally for
risk reduction determined in CLEOPATRA (HR 0.62). Results: Two hundred forty- 2 weeks of a 21-day cycle (days 8-21), with T-DM1 given at 3.6 mg/kg every
three pts were randomized. Baseline/disease characteristics and prior therapies were 3 weeks on day 1. A standard 3+3 dose escalation design was used to
generally balanced between arms. For PFS, the HR was 0.69 (95% CI 0.49, 0.99) in the evaluate various doses of ribociclib in combination with T-DM1 to determine
ITT population. No cases of heart failure or symptomatic left ventricular ejection fraction the RP2D. Results: From 5/2016 – 10/2018, 10 pts (8/10 ER+) were
decline were reported. Efficacy/safety are shown in the table. Conclusions: PUFFIN met enrolled with a median age of 53 (38-72) and median of 1 (0-2) prior
its primary endpoint. Overall, efficacy data were consistent with CLEOPATRA (ITT therapies for MBC. During dose-escalation, pts received doses of 300 mg
population and Asian subgroup). Safety was also consistent and in line with the known P
(n = 3), 400 mg (n = 3), 500 mg (n = 3), and 600 mg (n = 1). No DLTs were
safety profile, with no new or unexpected signals reported. PUFFIN adds to the totality of
data with P in previously untreated HER2-positive LR/MBC, and supports the favorable observed. The most common treatment related grade 3 or higher AEs were
benefit–risk profile of P in Chinese pts. Clinical trial information: NCT02896855. neutropenia (50%), infection (20%), anemia (10%), and thrombocytopenia
(10%). 4/10 pts had dose reductions due to toxicity. The average con-
P+H+D Pla + H + D centration of ribociclib at steady state was similar at each dose level, ranging
n = 122 n = 121
from 273 to 413 ng/mL. Among 9 evaluable pts, the ORR was 33% (3/9) and
Median follow-up, mo 13.7 13.1
Median PFS, mo (95% CI) 14.5 (12.5, 18.6) 12.4 (10.4, 12.7)
the other 6 pts had stable disease. After a median follow-up time of
D, mo 2.1 10.9 months, the median PFS was 12.5 months (95% CI [10.5. 20.9]).
HR (95% CI) 0.69 (0.49, 0.99) Biomarker results will be presented at the meeting. Conclusions: Co-targeting
n = 105 n = 97
ORR, n (%) 83 (79.0) 67 (69.1) HER2 and CDK4/6 with the combination of ribociclib with T-DM1 was well
D, % (95% CI) 9.98 (–2.65, 22.60) tolerated with evidence of clinical activity. Based on PK analysis and dose
n = 122 n = 120 reductions, 400 mg is the RP2D. Further evaluation is warranted for patients
Safety, pts (%)
All deaths 12 (9.8) 13 (10.8) with HER2+ MBC. Clinical trial information: NCT02657343.
Grade ‡3 AE 86 (70.5) 83 (69.2)
AE leading to withdrawal from P/Pla 6 (4.9) 2 (1.7)
SAE 24 (19.7) 23 (19.2)
(S)AE, (serious) adverse event.

1029 Poster Session (Board #110), Sun, 8:00 AM-11:00 AM 1030 Poster Session (Board #111), Sun, 8:00 AM-11:00 AM
Tucatinib, palbociclib, and letrozole in HR+/HER2+ metastatic breast High frequency of HER2-specific immunity observed in patients (pts) with
cancer: Report of phase IB safety cohort. First Author: Elena Shagisultanova, HER2+ cancers treated with margetuximab (M), an Fc-enhanced anti-HER2
University of Colorado Cancer Center, Aurora, CO monoclonal antibody (mAb). First Author: Jeffrey L. Nordstrom, MacroGenics
Inc, Rockville, MD
Background: Based on preclinical synergy, we are conducting a phase IB/II
clinical trial of tucatinib (HER2 small molecule inhibitor, T), palbociclib (CDK4/6 Background: Previous studies have shown that 44-71% of trastuzumab (T)-
inhibitor, P) and letrozole (aromatase inhibitor, L) in HR+/HER2+ metastatic treated pts develop HER2-specific immunity (Clin Cancer Res 2007, 13:
breast cancer (MBC). Methods: Post-menopausal, or pre-menopausal women on 5133; Cancer Res 2016, 76:3702; Breast Cancer Res 2018, 20:52). M is
ovarian suppression, with prior $2 HER2 inhibitors at any time of disease; $1 an Fc-engineered mAb that shares similar HER2 binding and anti-
HER2 inhibitor for MBC or front line bone/soft tissue only disease are eligible proliferative activity as T. The Fc region of M has been engineered for in-
provided #2 endocrine agents for MBC. Prior CDK4/6 or HER2 small molecule creased affinity to the activating FcgRIIIA (CD16A) and lower binding to the
inhibitors are not allowed. Treatment entails T 300 mg BID, P 125 mg/day inhibitory FcgRIIB (CD32B), attributes that may enhance the mAb’s im-
21 days on, 7 days off, and L 2.5 mg/day. Safety was assessed using CTCAE mune function, such as antigen presentation. Methods: HER2+ cancer pts
v.4.03 with standard definitions for dose limiting toxicity (DLT). Dose reductions who progressed on prior therapy received M (0.1-6 mg/kg QW for 3 of every
of T and / or P for DLTs, and discontinuation of either P or L for toxicity were 4 weeks [N = 34]; or 10-18 mg/kg Q3W [N = 32]) in phase 1 trial
allowed at any time during the study. Safety thresholds were set as DLTs in #7/ NCT01148849. PBMC and plasma were collected pre-dose and 50 days
20 pts for T, #15/20 pts for P, or #14/20 pts attributable to both T and P. post-dose for 46 pts and . 4 years for 3 pts on long-term treatment. Re-
Results: Phase Ib enrolled 20 pts from 11/16/17 to 12/5/18. The median age is sponse to HER2 or control antigens (Ag) was assessed by IFNg ELISpot and
53y (22-70y), median number of prior lines of MBC therapy is 2 (0-5). 70% of antibody (Ab) ELISA. In 14 pts, T-cell antigen receptor (TCR) repertoire was
pts have visceral disease and 45% CNS disease. Prior treatment includes 100% assessed by immunosequencing. Results: Following M treatment, mean
of pts with trastuzumab and pertuzumab, and 45% of pts with prior TDM-1. One frequencies of IFNg-producing T cells specific for intra- or extracellular
pt required dose reduction of T; 9 (45%) had dose reduction and 2 (10%) fragments of HER2 increased by 2.5 to 6-fold (p , 0.0027, paired t test).
discontinued P for DLTs. One pt discontinued L. Safety boundaries were not Most (95%) of subjects responded to $2 of 6 (median = 5) HER2 Ag. Mean
crossed. The most common grade (G) $3 toxicities were neutropenia (G3 HER2-specific Ab concentration increased by 19-54% (p , 0.0001), with
55%, G4 15%), diarrhea (G3 20%), infections (G3 20%), thrombocytopenia all subjects responding to $2 (median = 5) of the 6 Ag. A small 1.6-fold
(G3 10%) and mucositis (G3 10%). The frequency and type of toxicities were increase in IFNg response to control CMV/EBV/Flu (but not tetanus or cyclin
consistent with those previously reported for each single agent. PK analysis D1) peptides was observed; no increase in Ab response to control Ag was
showed no interaction between P and T. As of 1/4/19, 14 pts (70%) remain on noted. Subsets of HER2-specific T-cell and Ab responses persisted during
study (5 pts for $6 months) and 6 pts removed for progression. No withdrawals long-term treatment. Median TCR clonality increased by 54% (p = 0.003),
for toxicity and no deaths on study. The longest time on study (ongoing) is with an average of 125 unique clones expanding, while overall TCR diversity
10 months for pts without CNS disease, and 6 months for pts with CNS remained unchanged (p = 0.19). Conclusions: Treatment of HER2+ cancer
disease. Conclusions: T, P, L combination showed an acceptable safety profile with M was associated with enhanced HER2-specific T-cell and Ab re-
and encouraging antitumor activity. RP2D of T is 300mg PO BID in combi- sponses together with increased TCR clonality, indicative of a more focused
nation with full doses of P and L. Enrollment in phase II cohort is ongoing. T-cell repertoire. The high frequency of HER2-specific immunity in M-
Clinical trial information: NCT03054363. treated patients ( . 95%) is consistent with its enhanced Fc region con-
tributing to linkage of innate and adaptive immune responses.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
42s Breast Cancer—Metastatic

1031 Poster Session (Board #112), Sun, 8:00 AM-11:00 AM 1032 Poster Session (Board #113), Sun, 8:00 AM-11:00 AM
Elevated serum activin A and PD-L1 and survival in the CCTG MA.31 phase Disparities in treatment patterns and overall survival (OS) in hormone
III trial (trastuzumab vs. lapatinib) in first-line HER2+ metastatic breast receptor-positive HER2+ metastatic breast cancer (MBC): A National Cancer
cancer. First Author: Prashanth Reddy Moku, Penn State Hershey College of Database Analysis. First Author: Abby Statler, Cleveland Clinic Taussig
Medicine, Hershey, PA Cancer Institute, Cleveland, OH
Background: In MA.31 the trastuzumab-taxane combination led to longer PFS Background: Determinants of variation in therapy utilization and OS are unclear for
than lapatinib-taxane in HER2+ metastatic breast cancer (MBC). In MA.31 we patients diagnosed with hormone receptor+, human epidermal growth factor 2
previously reported the prognostic/predictive utility of pretreatment serum PD- positive (HER2+) MBC. This study aimed to identify if there are disparities in first-
L1 (SABCS 2018, PD3-10) and serum activin A (SABCS 2016, P6-07-06) line treatment patterns and outcomes for this subpopulation of MBC patients.
separately; here we evaluate them combined. Methods: MA.31 accrued 652 Methods: We analyzed MBC patients included in the National Cancer Database
centrally and/or locally-identified HER2-positive patients, and pretreatment diagnosed with estrogen receptor-positive (ER+) and/or progesterone receptor-
serum was available for 382 patients (184 in trastuzumab arm, 198 in lapatinib positive (PR+) and HER2+ disease (i.e. triple positive) treated with endocrine
arm). The ELLA immunoassay platform (ProteinSimple, San Jose, CA) was used therapy or chemotherapy between 2010 and 2015. Analyses describe the dis-
tribution of treatments administered in the first-line setting. Kaplan-Meier method
to quantitate serum PD-L1, and ELISA for activin A (R&D Systems, Minneapolis,
was used to estimate distributions of OS, which were compared among patient
MN). Results: In correlation analysis, pretreatment serum PD-L1 was moderately
cohorts using the log-rank test. Results: Of the 6215 patients diagnosed with triple
correlated with serum activin A (r = 0.21, p = 0.004). In univariate analysis for positive MBC, the majority were 50-70 years old (n=3414 [55%]), female (n=
OS, the combination of higher serum PD-L1 and higher serum activin A (median 6122 [98%]), and white (n=4478 [72%]). Four distinct treatment patters were
cutpoints) (vs. both low) was significant for shorter OS in the trastuzumab arm identified; hormonal therapy was the most common (n= 2289 [37%]), followed by
(HR 6.62, p=0.0005) and in the lapatinib arm (HR 3.25, p=0.0003)(table). In hormonal therapy + anit-HER2 (n=1471 [24%]), chemotherapy (1280 [20%]),
multivariate analysis for OS (17 covariates included), elevated serum activin A/ and chemotherapy + anit-HER2 (n=1175 (19%)). Significant differences in de-
PD-L1 combination remained the most significant independent covariate in the mographic, socioeconomic, and disease characteristics were identified across
trastuzumab arm (HR 12.40, p=0.001), and in the lapatinib arm (HR 5.2, groups. Disparities in OS were also observed; the unadjusted 5-year OS was
p=0.0001). Conclusions: In the CCTG MA.31 trial, elevated pretreatment serum substantially lower among older patients, African Americans (AA), those with
activin A (TGF-B superfamily) and PD-L1 was associated with a shorter OS to government insurance, and lower income (Table). Conclusions: This is the first
HER2-targeted treatment. Multiple mechanisms, including immune evasion, study to report disparities in treatment patterns and OS among real-world triple
may decrease the effectiveness of HER2-targeted therapy. Elevated serum positive MBC patients. Further investigation is required to determine if there are
activin A and PD-L1 may identify HER2-positive MBC patients who would benefit independent causal associations between poor prognosis and the identified de-
from inhibitors of the HER2, PD-1, and activin A pathways. mographic and socioeconomic characteristics.
Univariate analysis of OS of serum activin A and PD-L1 combined. Factor Level Median 5 year OS P-value

TRAS arm1 LAPT arm2 Age ,50

50-70
0.55
0.39
, 0.0001

.70 0.20
Serum biomarker (ActA & PD-L1) HR CI (95%) # Pts HR CI (95%) # Pts Sex Female 0.40 0.02
Male 0.24
both high vs both low 6.62 2.27 - 19.32 54 3.25 1.65 - 6.42 62 Race White (Non-Hispanic) 0.40 ,0.0001
ActA high / PD-L1 low vs both low 4.15 1.34 - 12.87 39 2.62 1.26 - 5.44 37 AA 0.31
Asian 0.52
ActA low / PD-L1 high vs both low 3.16 0.95 - 10.51 36 0.93 0.36 - 2.35 34 Hispanic/Latinos 0.53
both low* 1 * 55 1 * 65 Others 0.68
Insurance Government 0.28 ,0.0001
Not Insured 0.40
* Referent cohort Private 0.52
1
p = 0.0012 Income ,38K 0.28 ,0.0001
2 38-68 K 0.39
p = 0.0003 .68 K 0.48

1033 Poster Session (Board #114), Sun, 8:00 AM-11:00 AM 1034 Poster Session (Board #115), Sun, 8:00 AM-11:00 AM
Real-world treatment patterns and outcomes in ER+/PR+/HER2+ metastatic Clinical characteristics of patients with no evidence of disease (NED) versus
breast cancer (MBC) patients: A National Cancer Database analysis. First residual disease (RES) to anti-HER2 therapy in metastatic breast cancer
Author: Abby Statler, Cleveland Clinic Taussig Cancer Institute, Cleveland, OH (MBC): A multi-institutional analysis. First Author: Zachary William Neil
Veitch, Princess Margaret Cancer Centre, Toronto, ON, Canada
Background: Treatment patterns and clinical outcomes are unclear for MBC
patients diagnosed with estrogen receptor-positive (ER+) and/or progesterone Background: Anti-HER2 therapy has improved survival in HER2+ MBC. Yet,
receptor-positive (PR+) human epidermal growth factor 2 positive (HER2+) large patient cohorts with no evidence of disease (NED) with long-term follow
disease (i.e. triple positive). This study aimed to: 1) examine the utilization of up are incompletely described in the literature. We evaluated the clinical
first-line therapy among ER+/PR+/HER2+ MBC patients and 2) compare characteristics of patients with HER2+ MBC and prolonged response to anti-
overall survival (OS) between the identified regimens. Methods: We analyzed HER2 therapy, exhibiting NED vs Residual disease (RES). Methods: Patients
triple positive MBC patients included in the National Cancer Database who treated with chemotherapy plus trastuzumab (CT) from 2005-2013, or taxane
were treated with endocrine therapy (ET) or chemotherapy (chemo) between plus trastuzumab-pertuzumab (TTP), or trastuzumab-emtansine (TDM1) from
2010 and 2015. Kaplan-Meier method was used to estimate distributions of 2012-2016 for HER2+ MBC at Princess Margaret Cancer Centre in Toronto,
OS, which were compared among patient cohorts using the log-rank test. Ontario or in Alberta, Canada were included. Duration on anti-HER2 therapy
Results: A total of 6,234 patients were included in the analysis; of these, 3770 (without switch) was collected. Patients with median duration of response
(60%) received ET and 2464 (40%) received chemo. Of those with complete (MDR; months) 2x higher than phase II/III trials for each regimen (CT = 18.2;
survival data, there was no difference in median OS between patients treated TTP = 40.4; TDM1 = 25.2) were included to select for prolonged response.
with chemo vs. ET; however, those who received anti-HER2 therapy had Clinical features (ie: stage at diagnosis, survival, etc) and oncologist/radiologist
significantly better OS than those who did not (median OS 49.4 vs. reported best response were collected. Responses were grouped as NED
41.0 months, p , 0.0001). Median OS stratified by ET or chemo with and (including sclerotic bone metastases) or RES. Clinical variables were evaluated
without anti-HER2 further supported these findings, revealing the addition of by Chi-square and survival by Kaplan-Meier (log-rank). Results: 2403 patients
anit-HER2 therapy to chemotherapy and ET resulted in superior median OS (CT = 1830; TTP = 394; TDM1 = 179) were evaluated. After cut-off, 119
(Table). Conclusions: This is the first study to evaluate treatment utilization patients (5%) were included, with NED in 41% (49/119) and RES in 59% (70/
and OS among real-world triple positive MBC patients treated with ET or 119). More women aged , 50 vs $50 (p = 0.015) had NED (61%) than RES
chemo. Our results suggest the majority of patients in the United States are (39%). No breast surgery (curative or metastatic) showed higher (p = 0.015)
treated with first-line ET; furthermore, the reported OS outcomes support the rates of NED (49%) relative to RES (27%). More women having NED (92%)
consideration of ET plus anti-HER2 therapy as a first-line treatment option for than RES (37%) were still alive (p , 0.0001) at data cut-off, with improved
ER+/PR+/HER2+ MBC. Prospective trials evaluating de-escalation of systemic mPFS (years; p , 0.0001) for NED (11.2, 95%CI: 11.2-NR) vs RES (3.0, 2.7-
therapy in this subgroup of patients and future research to identify biomarkers 3.6), and mOS (years; p , 0.0001) for NED (NR) vs RES (3.4, 95%CI: 2.8-
to determine which patients can avoid chemotherapy are warranted. 4.2). Significance persisted with TDM1 patients excluded. Treatment type (p =
0.053) and number of organs with metastases (p = 0.067) were of borderline
Median OS Months significance. Conclusions: Patients with NED have improved survival com-
Group Total N Number of Deaths (95%CI) P-value
pared to RES. Younger age and avoiding breast surgery correlates with NED.
Chemo 1151 632 38.4 ( 33.8 , 43.0 ) , 0.0001 Evaluation of genomic factors influencing NED are planned.
Chemo + Anti-Her2 726 239 46.8 ( 44.9 , NA )
ET 1995 1042 42.3 ( 39.5 , 45.6 )
ET + Anti-Her2 936 245 56.0 ( 45.1 , NA )

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Breast Cancer—Metastatic 43s

1035 Poster Session (Board #116), Sun, 8:00 AM-11:00 AM 1036 Poster Session (Board #117), Sun, 8:00 AM-11:00 AM
Safety, efficacy, and biomarker analysis of pyrotinib in combination with Association of HER2 alterations and ESR1 mutations in cell-free DNA
capecitabine in HER2-positive metastatic breast cancer patients: A phase I (cfDNA) with circulating tumor cells (CTCs), multiple metastasis, and
clinical trial. First Author: Fei Ma, National Cancer Center/National Clinical prognosis in stage III/IV breast cancer (BCa). First Author: Qiang Zhang,
Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Northwestern University, Department of Medicine, Division of Hematology/
Sciences and Peking Union Medical College, Beijing, China Oncology, Lurie Cancer Center, Chicago, IL
Background: Cross-signaling in the ErbB family is an important mechanism Background: The monitoring of CTCs and cfDNA in metastatic BCa showed
in Trastuzumab resistance. Pyrotinib is an irreversible pan-ErbB inhibitor ability to predict treatment resistance and survival. Here we report a highly
targeting EGFR/HER1, HER2 and HER4, which may offer the potential for significant correlation between HER2 alterations and ESR1 mutations of
improved efficacy to block HER2 signaling in trastuzumab-resistant breast cfDNA with CTCs and prognosis of BCa, which may help to predict disease
cancer. This phase I study assessed the safety, tolerability, maximum- recurrence and treatment benefit. Methods: A total of 85 blood samples
tolerated dose (MTD), pharmacokinetics, antitumor activity and predictive (7.5ml/each) were collected from 85 patients with stage III/IV BCa who
biomarkers of pyrotinib in combination with capecitabine in patients with received treatments at Northwestern RHLCCC. CTC enumeration was per-
HER2-positive metastatic breast cancer (MBC). Methods: Patients received formed in FDA approved CELLTRACKS ANALYZERII System (Menarini).
oral pyrotinib 160 mg, 240 mg, 320 mg, or 400 mg once daily continually Plasma cfDNA was analyzed using Guardant360 NGS-based assay including a
plus capecitabine 1000mg/m2 twice daily on days 1 to 14 of a 21-day cycle. 73-gene panel for genomic alterations or mutations. We previously reported
Pharmacokinetic blood samples were collected predose on day 1 and day 14 cut-off of 5.7% (2018 ASCO) was used to dichotomize the prognostic value of
of treatment. Next-generation sequencing (NGS) was performed on circu- cfDNA percentage. Kruskal-Wallis test was used for statistics. Results: Of the
lating tumor DNA (ctDNA) to probe for predictive biomarkers of this com- 85 whole samples analyzed, there were 72 samples and 67 samples without
bination. Results: A total of 28 patients were enrolled. All 28 (100%) patients ESR1 mutations (ESR1-) and HER2 alterations (HER2-) respectively, and
experienced at least one treatment-related Adverse Events (AE), which were there are 13 samples and 18 samples that had ESR1 mutations (ESR1+) and
predominantly grade 1 or 2. Grade 3 treatment-related AE occurred in 12 (42.9%) HER2 alterations (HER2+, 10 amplified, 7 mutated, 1 for both) respectively.
patients; anemia (14.3%) and diarrhea (10.7%) were the most common grade CTC positive ($5) were detected in 13/57 ESR1-HER2- samples (Group 1) and
3 AEs. Three (10.7%) patients discontinued capecitabine administration due to 5/15 ESR1-HER2+ samples (Group 2), 7 /10 ESR1+ HER2- samples (Group
AEs. The overall response rate (ORR) was 78.6% (95% CI: 59.0% to 91.7%), and 3), 3/3 ESR1+ HER2+ samples (Group 4). The median CTCs number/sample in
the disease control rate (complete response+ partial response+ stable disease) was Group 3 (15 CTCs) and Group 4 (12 CTCs) were significantly higher than Group
96.4% (95% CI: 81.7% to 99.9%). The median progression-free survival (PFS) 1 (0 CTC) and Group 2 (2 CTCs) (P = 0.0020). There were a significant higher
was 22.1 months (95% CI: 9.0 to 26.2 months). ORR was 70.6% (12/17) in average metastasis sites in Group 3 (3 sites) and Group 4 (3 sites) in compared
trastuzumab-pretreated patients and 90.9% (10/11) in trastuzumab-naive to Group 1 (2 sites) and Group 2 (1 site) (P = 0.0035). Furthermore, patients in
patients. NGS analysis of all genetic alterations of HER2 bypass signaling Group 4 (ESR1+HER2+) has the worst prognosis in compared to other groups
pathway, PI3K/Akt/mTOR pathway and TP53 in baseline blood samples (P = 0.0151) on overall survival. Conclusions: Both ESR1 mutations and
suggested that concomitant (two or more) genetic alterations were signifi- HER2 alterations in cfDNA contribute to CTCs detection and disease me-
cantly associated with poorer PFS compared to none or one genetic alteration tastasis sites independently, when ESR1 mutations plays a major role. The
(median, 15.8 vs. 26.2 months, p = 0.006). Conclusions: Pyrotinib in synergy of ESR1 mutation and HER2 alteration expands the predictive role of
combination with capecitabine are well-tolerated and demonstrate prom- liquid biopsy tests monitoring the metastatic prognosis and endocrine re-
ising antitumor activity in HER2-positive MBC patients. Clinical trial in- sistance for clinical decision-making.
formation: NCT02361112.

1037 Poster Session (Board #118), Sun, 8:00 AM-11:00 AM 1038 Poster Session (Board #119), Sun, 8:00 AM-11:00 AM
Comparison of outcomes in a population-based cohort of women with Alpelisib (ALP) with fulvestrant (FUL) in patients (pts) with PIK3CA-mutated
metastatic breast cancer receiving anti-HER2 therapy with clinical trial hormone receptor-positive (HR+), human epidermal growth factor receptor-
outcomes. First Author: Inna Y. Gong, Department of Medicine, Univer- 2-negative (HER2-) advanced breast cancer (ABC): Primary or secondary
sity of Toronto, Toronto, ON, Canada resistance to prior endocrine therapy (ET) in the SOLAR-1 trial. First Author:
Dejan Juric, Massachusetts General Hospital Cancer Center, Harvard
Background: Little data exist for comparing cardiac safety and survival out-
Medical School, Boston, MA
comes of anti-HER2 therapy with concurrent trastuzumab (T) and pertuzumab
(P) or ado-T emtansine (TDM1) in metastatic breast cancer (MBC) patients Background: A contributor to ETR, phosphatidylinositol 3-kinase (PI3K) pathway
enrolled in randomized clinical trial (RCT) vs those in the real world. Furthermore, hyperactivation can result from mutations to PIK3CA; ~40% of pts with HR+,
whether older patients have worse outcomes is unknown. Methods: This was a HER2– ABC exhibit tumors with this mutation. Use of the oral a-specific PI3K
retrospective population-based cohort of all women with MBC treated with inhibitor ALP + FUL significantly improved progression-free survival (PFS) in
concurrent T with P or TDM1 in Ontario (between 2012 and 2017), identified pts with a PIK3CA mutation (HR 0.65; 95% CI, 0.50-0.85; P,0.001) in
from New Drug Funding Program and linked to Ontario Cancer Registry and other SOLAR-1, which included both ET sensitive (ETS) and ETR pts (Table). ETS
administrative datasets. Outcomes were incident heart failure (HF, defined as pts were later excluded by a protocol amendment. ETR was further defined as
hospitalization or emergency room visit for HF) and overall survival (OS). RCT primary (1R) or secondary (2R) per ESMO criteria in both 1L and 2L pts. This
data were obtained from digitizing survival curves as per established methods and subgroup analysis evaluated pts with a PIK3CA mutation based on tx line and
compared with cohort OS data using log-rank test. Age based comparison of endocrine status. Methods: SOLAR-1 was a phase 3, randomized, double-
outcomes was conducted for women $ 65 years old vs younger. Results: Our blind study of ALP 300 mg QD or PBO Q28d + FUL 500 mg Q28d + C1d15 in
cohort composed of 833 (28% . 64 years old), and 397 (28% . 64 years old) men and postmenopausal women with HR+, HER2– ABC whose disease pro-
women treated with P and TDM1, respectively, of which 46 and 30 had baseline gressed on/after an aromatase inhibitor. PFS was estimated by Kaplan-Meier
HF, respectively. 49% and 99.5% of women received T prior to P and TDM1, method and median PFS (mPFS) presented by tx arm. A stratified Cox pro-
respectively. Incident HF following P or TDM1 initiation was low (P 26 women, portional hazards model estimated HR and 2-sided 95% CI. Results: Of 341 pts
TDM1 8 women; Table). HF events was not more in women $ 65 years old in the PIK3CA mutant cohort, 39 (11%) were ETS; 302 (89%) were ETR. mPFS
compared to women , 65 treated with P (16 vs. 10, p = 0.23). Unadjusted OS in the ALP vs PBO arms was 22.1 vs 19.1 mo (HR 0.87; 95% CI, 0.35-2.17) for
was significantly worse than RCT OS (Table; P HR 1.67, 95% CI 1.37-2.03, p , ETS pts and 9.4 vs 4.2 mo (HR 0.64; 95% CI, 0.48-0.84) for ETR pts. For ETR
pts, mPFS for 1L (n=138) was 9.0 vs 4.7 mo (HR 0.69; 95% CI, 0.46-1.05) and
0.0001; TDM1 HR 2.80, 95% CI 2.27-3.44, p , 0.0001). Older women had
for 2L (n=161) was 10.9 vs 3.7 mo (HR 0.61; 95% CI, 0.42-0.89).
worse OS than younger women for P (HR 1.54, 95% CI 1.22-1.96, p = 0.0003),
Conclusions: In SOLAR-1, mPFS was improved with ALP + FUL vs PBO + FUL
but not for TDM1 (HR 1.08, 95% CI 0.81-1.43, p = 0.62). Conclusions: HF
across ETR pts in 1L and 2L. Representation of ETS pts was low in SOLAR-1,
incidence during P or TDM1 therapy in this real world cohort was relatively low.
which included more ETR pts. Analysis of the PI3K pathway in ETS pts is
Survival in this cohort was significantly worse compared to RCT, particularly for warranted in future studies. Clinical trial information: NCT02437318.
older women, suggesting importance of evaluating effectiveness in an unselected
patient population to facilitate informed decision-making based on real-world Tx Line Endocrine Status Resistance Resistance Definition n
risks and survival outcomes. 1L ETS – Relapse $12 mo after end of adj ET 39
ETR 1R Relapse ,24 mo on adj ET 25
P TDM1 2R 24 mo on adj ET #relapse ,12 mo after end of adj ET 113
CLEOPATRA (n = 407) Cohort (n = 833) EMILIA (n = 495) Cohort (n = 397) 2L ETR 1R Progression ,6 mo on ET for ABC 19
2R Progression $6 mo on ET for ABC 132
HF, n (%) 27 (6.6) 26 (3.1) 8 (1.7) 1 (0.25)
Follow-up, mos 49.5 28.2 18.6 23.5 Of 302 ETR pts, 10 pts in 2L could not be assigned to 1R/2R per ESMO criteria; 3 pts could not be
Median OS, mos 56.5 39.2 30.9 15.4
Survival at 1 year, % 94 (92-97) 79 (76-82) 85 (82-89) 53 (48-58)
assigned to 1L/2L and subsequently to 1R/2R.
Survival at 3 years, % 68 (63-73) 53 (49-57) 41 (29-52) 22 (16-28)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
44s Breast Cancer—Metastatic

1039 Poster Session (Board #120), Sun, 8:00 AM-11:00 AM 1040 Poster Session (Board #121), Sun, 8:00 AM-11:00 AM
Patient-reported outcomes (PROs) in patients (pts) with PIK3CA-mutated Alpelisib (ALP) + endocrine therapy (ET) in patients (pts) with PIK3CA-
hormone receptor-positive (HR+), human epidermal growth factor receptor- mutated hormone receptor-positive (HR+), human epidermal growth factor-
2–negative (HER2–) advanced breast cancer (ABC) from SOLAR-1. First 2-negative (HER2-) advanced breast cancer (ABC): First interim BYLieve
Author: Ingrid A. Mayer, Vanderbilt-Ingram Cancer Center, Nashville, TN study results. First Author: Hope S. Rugo, University of California San
Francisco Comprehensive Cancer Center, San Francisco, CA
Background: Approximately 40% of pts with HR+, HER2– ABC have tumors
with a PIK3CA mutation, resulting in phosphatidylinositol 3-kinase (PI3K) Background: In the phase 3 SOLAR-1 study, ALP + fulvestrant (FUL) im-
pathway hyperactivation. Use of the oral a-specific PI3K inhibitor alpelisib proved PFS in pts with HR+, HER2– ABC with a PIK3CA mutation overall and
(ALP) + fulvestrant (FUL) in SOLAR-1 significantly improved (vs placebo in the small group of pts with prior cyclin-dependent kinase 4/6 inhibitor
[PBO] + FUL) both progression-free survival (PFS) (median 11.0 vs 5.7 mo, (CDKi) use. We report interim data from the BYLieve study in pts with
respectively; HR 0.65; 95% CI, 0.50-0.85; P, 0.001) and objective re- PIK3CA-mutated ABC and prior CDKi exposure. Methods: BYLieve is an
sponse rate (ORR) (measurable disease: 36% vs 16%; P, 0.001) in the ongoing, phase 2, open-label, non-comparative study of ALP 300 mg QD +
PIK3CA mutant cohort. In addition to primary efficacy and safety measures, ET in men and women with PIK3CA-mutated HR+, HER2– ABC whose
PROs offer valuable insight into therapeutic benefit by measuring whether disease progressed on/after CDKi + ET. Pts are permitted #2 prior anticancer
quality of life (QoL) is maintained during treatment. Methods: Postmenopausal therapies and #1 prior chemotherapy regimen for ABC. Pts with prior CDKi
women or men with HR+, HER2– ABC whose disease progressed on/after an and AI (FUL cohort) receive ALP and FUL 500 mg Q28d + C1d15 IM. Pts
aromatase inhibitor were randomized to receive ALP 300 mg once daily or with prior CDKi and FUL (LET cohort) receive ALP and letrozole (LET) 2.5 mg
PBO, + FUL 500 mg every 28 days + Cycle 1, Day 15. Secondary objectives PO QD. In this preplanned interim analysis, conducted after $20 pts in FUL
included PROs using the EORTC QLQ-C30, EQ-5D-5L, and BPI-SF scales. had $6 mo of follow-up, descriptive data are reported for preliminary safety
PROs were collected at screening, every 8 wk for 18 mo then every 12 wk and efficacy in the FUL and LET cohorts. Results: At data cutoff, 64 and
thereafter, at end of treatment, and during follow-up for efficacy. Linear mixed 36 pts were enrolled in the FUL and LET cohorts, respectively; 39 pts (FUL,
effects models were used to assess score changes from baseline. Time to 10% n = 21; LET, n = 18) have safety and efficacy data with $6 mo follow-up and
deterioration (TTD), an established measure of clinically meaningful change in are reported here. Data on 100 pts enrolled at the time of data cutoff will be
QoL, was compared between the treatment arms’ survival distribution using presented. In the 39 pts with $6 mo follow-up, median ALP duration was
Kaplan-Meier methodology. Results: At baseline, 93% of pts in the PIK3CA 5.3 and 5.5 mo in FUL and LET, respectively; median duration of FUL and
mutant cohort (n = 341) completed questionnaires; $75% completed them LET was 5.6 mo. Median relative ALP dose intensity was 93% (FUL) and
post-baseline. Adjusted mean changes from baseline in EORTC global health 87% (LET). Most common grade $3 adverse events were hyperglycemia
status/QoL scores were , 10% for all visits through wk 96 for both arms, with a (38.1% (FUL) and 27.8% (LET)) and rash (4.8% (FUL) and 27.8% (LET)).
mean difference between arms of , 3% for all visits. There was no difference Only 2 pts (5%; 1 pt per cohort) discontinued due to an AE. In pts with
between arms in TTD in global health/QoL status (HR 1.03; 95% CI, 0.72- centrally confirmed PIK3CA mutation (n = 20 (FUL); n = 17 (LET)), ORR was
1.48). Analysis of TTD in EORTC physical, social, and emotional functioning 20% (FUL) and 18% (LET), CBR was 40% (FUL) and 35% (LET). Efficacy
scores revealed no meaningful differences between arms. Conclusions: In and safety data for the 100 enrolled pts will be presented at the meeting.
addition to significantly improving PFS and ORR, overall QoL was maintained Conclusions: Pending further readout of the ongoing BYLieve trial, safety
in pts treated with ALP + FUL. Clinical trial information: NCT02437318. and tolerability of ALP and hormonal therapy in pts with prior CDKi are
consistent with those of SOLAR-1; discontinuation due to toxicity was rare.
NCT03056755. Clinical trial information: NCT02437318.

1041 Poster Session (Board #122), Sun, 8:00 AM-11:00 AM 1042 Poster Session (Board #123), Sun, 8:00 AM-11:00 AM
Interim results from the full population of the phase 3b CompLEEment-1 Next-generation sequencing (NGS) results among hormone receptor-positive
study of ribociclib (RIBO) plus letrozole (LET) in the treatment of HR+/HER2– (HR+), human epidermal growth factor receptor 2-negative (HER2-) metastatic
advanced breast cancer (ABC). First Author: Michelino DeLaurentiis, National breast cancer (MBC) patients treated with a CDK4 & 6 inhibitor: A retrospective
Cancer Institute “Fondazione Pascale,” Department of Breast and Thoracic observational study based on real-world data. First Author: Erika Paige Hamilton,
Oncology, Naples, Italy Sarah Cannon Research Institute and Tennessee Oncology, Nashville, TN
Background: RIBO, an oral, selective inhibitor of CDK4/6 (CDK4/6i), is Background: Few real-world studies have characterized the frequency of
approved for use in combination with endocrine therapy (ET) in women with genomic alterations of MBC tumors. Data characterizing alterations before
hormone receptor-positive (HR+), human epidermal growth factor receptor and after treatment containing a CDK4 & 6 inhibitor (CDK4 & 6i) are similarly
2-negative (HER2–) ABC in multiple countries worldwide. Here we report limited. We explore the genomic landscape of HR+/HER2- MBC tumors from
interim safety and efficacy results from CompLEEment-1, a phase 3b trial patients (pts) treated with a CDK4 & 6i in order to characterize potential
evaluating RIBO+LET in an expanded patient (pt) population and the largest mechanisms underlying sensitivity and resistance. Methods: NGS results of
CDK4/6i trial in ABC to date. Methods: Pts with HR+, HER2– ABC, #1 line tumor and liquid biopsies obtained from 130 pts with estrogen receptor (ER+)/
of prior chemotherapy (CT), and no prior ET for ABC received RIBO+LET. progesterone receptor (PR+)/HER2- MBC between Jan 2008 to Sept 2016
Treatment regimens and study endpoints have been reported previously (De were analyzed. All pts received therapy containing a CDK4 & 6i for MBC
Laurentiis, et al. ASCO 2018. Poster 1056). Results: Overall, 3,246 pts, at a community cancer network and had NGS results available before
who received $1 dose of study treatment, were evaluated (cut-off date, and/or after exposure to CDK4 & 6i. Samples were classified as sensitive (n =
August 8, 2018). Median duration of RIBO exposure was 8.1 months (min, 69; duration of therapy $6 mo) or resistant (n = 61; duration of therapy , 6
0.0; max, 22.4). Demographic and baseline characteristics indicated a mo). The frequency of genomic alterations with likely or known significance
diverse population including men (1.2%), premenopausal women (22.2%), including short variants, indels, copy number variants, and fusions were
and patients aged $70 years (19.5%). Pts were well represented in terms of characterized. Results: Alterations in 215 unique genes were identified from
age, race, and disease history; 5.9% of pts received prior CT for ABC. The the NGS results; PIK3CA, TP53, ESR1, CCND1, and FGFR1 were the most
only non-hematologic any-cause grade $3 AEs $5% were increased alanine frequently altered genes. Select alterations in ESR1 (n = 21 vs 9) and RAD21
(7.3%) and aspartate (5.3%) aminotransferase. Treatment-related AEs (any (n = 5 vs 0) were more frequent after exposure to CDK4 & 6i. In NGS obtained
grade) led to discontinuation in 11.4% of pts. Of the 51 (1.6%) on-treatment before exposure to CDK4 & 6i, alterations in select genes including RB1,
deaths, 26 were due to study indication and 25 to other reasons. The median MDM2, AURKA, and MYC were more frequent in the resistant samples,
time to progression was not estimable (NE) (95% confidence interval [CI], whereas ARID1A alterations were more frequent in sensitive samples. Of the
17.1-NE). Overall response rate was 20.5% (95% CI, 19.1%-21.9%) and 6 pts with paired NGS samples pre- and post-CDK4 & 6i treatment, alter-
clinical benefit rate was 66.1% (95% CI, 64.4%-67.7%). Consistent mean ations in MYC, CDKN2A, PIK3CA, BRCA1, or RB1 were acquired in 3 pts.
change from baseline in Functional Assessment of Cancer Therapy – Breast Conclusions: Based on real-world data, this study describes the genomic
Cancer questionnaire scores indicated that pts maintained their quality of landscape of ER+/PR+/HER2- MBC tumors from pts treated with CDK4 & 6i
life throughout treatment. Conclusions: This interim analysis demonstrates and identifies potential mechanisms underlying sensitivity and resistance to
the safety, tolerability, and efficacy of RIBO+LET in a large, diverse cohort of this new class of drugs. Further evaluation in larger datasets is warranted.
pts with HR+, HER2– ABC who had not previously received ET for ABC. Data inclusive of other ER/PR subtypes will be presented.
Safety results were consistent with those observed in RIBO pivotal studies
and no new safety signals were observed. Clinical trial information: NCT02941926.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Breast Cancer—Metastatic 45s

1043 Poster Session (Board #124), Sun, 8:00 AM-11:00 AM 1044 Poster Session (Board #125), Sun, 8:00 AM-11:00 AM
A GINECO randomized phase II assessing addition of an aromatase inhibitor Efficacy and safety of talazoparib (TALA) or physician’s choice of therapy
to oral vinorelbine in pretreated metastatic breast cancer patients. First (PCT) in United States patients (pts) with HER2- germline BRCA1/2-mutated
Author: Pierre Heudel, GINECO-Centre Léon Bérard, Lyon, France (gBRCAm) locally advanced/metastatic breast cancer (LA/MBC) in the EMBRACA
study. First Author: Sami Diab, University of Colorado Cancer Center, Aurora, CO
Background: For ER+/HER2- metastatic breast cancer (mBC), efficacy of
endocrine therapy + chemotherapy combination remain an open question. Background: TALA is a poly(ADP-ribose) polymerase (PARP) inhibitor approved in the US
We hypothesized that continuing ER targeted therapy after progression in for HER2- gBRCAm LA/MBC. Approval was based on results from the Phase 3 EMBRACA
combination with chemotherapy may improve disease control. The objective trial comparing efficacy/safety of TALA (1 mg/d) to PCT (capecitabine, eribulin, gemci-
tabine, vinorelbine) in HER2- gBRCAm LA/MBC pts. This analysis describes outcomes in
of the CHEOPS trial was to assess the benefit of adding aromatase inhibitor US pts included in the pivotal study. Methods: Clinical findings from US pts enrolled in
(AI) to metronomic chemotherapy,oral vinorelbine, 50mg/3 time a week (OV) EMBRACA were analyzed. Pt characteristics, progression-free survival (PFS), objective
for AI pre-treated, ER+/HER2- mBC patients. Methods: Eligible patients had response rate (ORR), clinical benefit rate (CBR), and safety/adverse events (AEs) were
to have progressed on endocrine therapy and one or two lines of chemo- among the parameters assessed. Results: Of 431 randomized pts, 156 pts (36%) were from
therapy. They were randomized between vinorelbine (OV) and vinorelbine + the US (TALA: 99; PCT: 57). Pt characteristics were balanced, although a higher per-
AI (OV+AI). Primary end point was progression-free survival (PFS). To show centage in the TALA arm had more poor prognostic features (eg, triple-negative breast
an increase of median PFS (from 3.5 to 5.5 month, HR 0.636), with alpha = cancer, disease-free interval , 12 mo, and more disease sites). TALA improved PFS, ORR,
CBR, and duration of response (DOR) vs PCT (Table). 22% of pts in the TALA arm had a
5% and power = 80%, 130 evaluable patients were needed. Results: 121
continued objective response at month 12 vs 0 pts in the PCT arm. The most common AEs
patients were Included (OV = 61; OV+AI = 60). Median age was 68 (range: in the TALA arm included anemia, neutropenia, thrombocytopenia, fatigue, nausea, alo-
49-87), Median time from metastatic diagnosis was 3.2 years (range 0 - pecia, and headache; hematologic grade 3/4 AEs occurred more often than nonhematologic
16.9). 109 patients (90%) had visceral metastases. They all had previously AEs. Conclusions: In US pts with HER2- gBRCAm LA/MBC, TALA demonstrated significant
received an AI and had been treated with one line (N = 66, 54.5%), or 2 lines improvements in outcomes vs PCT with a manageable safety profile. Clinical trial information:
(N = 55, 45.5%) of chemotherapy. Median PFS was increased from NCT01945775.
2.3 months with OV to 3.7 months with OV+AI, but this difference was not TALA PCT
significant (HR 0.73 [95 % CI 0.50-1.06], log-rank test: P = 0.09) 81 (n = 99) (n = 57)
patients (67%) had at least one adverse event (AE) of grade $ 3 (40 (66%) PFS, median, mo (95% CI) 9.0 (7.0-12.9) 5.8 (4.2-6.7)
for OV vs 41 (68%) for OV+AI). The most common grade $ 3 AE were: GT HR (95% CI) 0.5 (0.3-0.8)
ORR, n (%) 51 (63.0) 11 (24.4)
gammas (23%), neutropenia (18%), arterial hypertension and lymphopenia OR (95% CI) 5.5 (2.4-16.1)
(17%). The occurrence of 3 toxic deaths (OV = 1; OV+AI = 2) secondary to CBR, % (95% CI) 68.7 33.3
febrile aplasia motivated the early cessation of this clinical trial. 9 patients (5 (58.6-77.6) (21.4-47.1)
OR (95% CI) 4.7 (2.2-10.6)
OV (10%) and 4 OV+AI (8%) presented an objective complete or partial DOR, median, mo (95% CI) 5.0 (4.1-6.4) 3.1 (1.4-5.6)
response. Conclusions: The addition of AI to OV over OV alone in AI resistant HR (95% CI) 0.4 (0.2-1.1)
Any grade hematologic AEs, n (%)
mBC was associated with a non-significant improvement of PFS, but both Anemia 49 (49.5) 11 (25.6)
PFS are lower than expected. Metronomic OV schedule, at 50 mg three Neutropenia 22 (22.2) 13 (30.2)
Thrombocytopenia 19 (19.2) 3 (7.0)
times a week, requires close biological monitoring. The question of hormonal Any grade nonhematologic AEs, n %
treatment and chemotherapy combination remains open. Clinical trial in- Fatigue 59 (59.6) 21 (48.8)
formation: EudraCT Number: 2015-000401-39. Nausea 47 (47.5) 22 (51.2)
Alopecia 34 (34.3) 10 (23.3)
Headache 32 (32.3) 11 (25.6)
Treatment discontinuation, n/N*, (%) 6/99 (6.1) 6/43 (14.0)
*Safety population

1045 Poster Session (Board #126), Sun, 8:00 AM-11:00 AM 1046 Poster Session (Board #127), Sun, 8:00 AM-11:00 AM
A phase Ib/II trial of lenvatinib (len) and letrozole (let) incorporating Endocrine-based targeted combination versus endocrine therapy alone as
pharmacodynamics studies in postmenopausal women with hormone receptor first-line treatment in elderly patients with hormone receptor-positive advanced
positive (HR+) locally advanced/metastatic breast cancer (LABC/MBC). First breast cancer: Meta-analysis of phase II and III randomized clinical trials. First
Author: Joline Si Jing Lim, National University Cancer Institute, Singapore, Author: Claudia Omarini, University Hospital of Modena, Modena, Italy
Singapore
Background: Combined endocrine approaches have been widely investigated
Background: Endocrine blockade (EB) is standard of care for patients (pts) with as first-line treatment in hormone receptors positive metastatic breast cancer.
HR+ LABC/MBC. RET over-expression (RET+) occurs in up to 75% of HR+ breast In particular, multiple randomized trials showed that the addiction of CDK
cancers and is a postulated mechanism of endocrine resistance. Preclinical studies (cyclin-dependent kinase) 4/6 inhibitors to endocrine therapy (ET) increase
show cross talk between RET and estrogen receptor, and at least additive treatment progression free survival (PFS). Elderly patients (aged $65 years) are under
(Tx) effect of Len+EB. Methods: We performed a phase Ib trial (3+3 dose esca- represented in most of the clinical studies. Moreover, due to the multi-
lation) to study safety, tolerability, pharmacodynamics and efficacy of Len+Let. morbidity and the major toxicity associated with the targeted agents, the
Both drugs were given as continuous daily dosing with 2 weeks (wks) of Len alone, combination strategy in that subgroup is widely discussed. The present meta-
followed by Len+Let for 12 wks then surgery (LABC), or till disease progression (PD) analysis aimed to understand the role of the new endocrine approaches in
(MBC). Serial tumor biopsies (n = 15) were done at baseline, after Len alone, 4 wks women aged $65 years. Methods: This meta-analysis included first line phase
post Len+Let, and at surgery [LABC] / upon PD [MBC]. Results: 16 pts were treated II/III randomized published trials comparing (ET) to the experimental strategy.
(4 LABC, 12 MBC); Among MBC pts, median lines of prior Tx was 3 (range 0-10); Trials with no data about hazard ratios (HR) for PFS in the subgroup of patients
84.6%, 66.7%, and 58.3% had prior EB, EB+CDK4/6 inhibitor (i), and chemo- aged $65 years were excluded. The heterogeneity of the data was evaluated by
therapy (CT) respectively. At dose level (DL) 1, 2/4 pts had dose-limiting toxicities Chi-square Q test and I2 statistic. Results: 8 studies were included in the
(DLT). There was no DLT at DL-1, but 6/6 pts needed dose reductions (DR), with 4/6 analysis. 4 trials (Paloma1/TRIO-18, Paloma2, Monaleesa2, Monarch3) in-
DR within 6 wks of Len+Let (3 G3 hypertension [HTN], 1 G3 wound pain), deeming
vestigated the role of CDK 4/6 inhibitors, 2 trials (SWOG and FACT) analysed
DL-1 intolerable. At DL-2, 0/6 pts had DLT; this was declared recommended phase
the combination of Fulvestrant plus Aromatase Inhibitors, while other two trials
2 dose (RP2D). Most frequent G3 toxicities (tox) were HTN (6/16), proteinuria (2/
explored the association of ET with Bevacizumab (LEA) and Temsirolimus
16) and palmar-plantar erythrodysesthesia (PPE) (2/16), with no G4/5 tox. Len+Let
(HORIZON), respectively. Overall, the meta-analysis showed a PFS advantage
was active with 93.8% overall disease control rate (DCR) (50.0% partial response
[PR], 43.8% stable disease [SD]). Among MBCts (8/12 had prior EB+CDK4/6i), for the experimental arms [HR 0.77, p 0.016] with a significant high/moderate
DCR $12 wks was 91.7%; 1 pt had sustained PR for 48 wks and 1 ongoing PR heterogeneity [I2 65.46%, p 0.005]. The 4 studies adding CDK4/6 inhibitors
at 40 wks. 9/16 pts had RET+ tumors on immunohistochemistry at baseline, and to ET showed a significant improvement in PFS compared to ET alone. No
66.7% showed down-regulation with Tx (RECIST: 4 PR, 2 SD). Conclusions: Len+ significant advantages for the addition of anti-angiogenic agents or Fulvestrant
Let showed significant anti-tumor activity, even in pts who failed prior CT or EB+ to ET have been found in elderly population subgroup. Conclusions: The novel
CDK4/6i. RP2D of 14mg Len and 2.5mg Let is tolerated with efficacy; dose ex- experimental combo-strategies in the first line setting showed an improvement
pansion is currently underway. Clinical trial information: NCT02562118. in PFS in the subgroup of elderly patients. Adding CDK4/6 inhibitors to ET
significantly prolongs PFS as compared to ET alone. The magnitude of PFS
RECIST Best benefit due to addition of CDK4/6 inhibitors to ET is age-independent.
Response
DL Let dose (mg) Len dose (mg) N DLT Pts with DR PR SD PD
1 2.5 20 4 G3 PPE 3/4 3 1 0
G3 proteinuria
-1 2.5 16 6 - 6/6 4 1 1
-2 2.5 14 6 - 2/6 1 5 0

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
46s Breast Cancer—Metastatic

1047 Poster Session (Board #128), Sun, 8:00 AM-11:00 AM 1048 Poster Session (Board #129), Sun, 8:00 AM-11:00 AM
A phase II study of pembrolizumab in combination with palliative radiotherapy Germline BRCA1and BRCA2 mutations in patients with HER2-negative
(RT) for hormone receptor-positive (HR+) metastatic breast cancer (MBC). metastatic breast cancer (mBC) treated with first-line chemotherapy: Data
First Author: Romualdo Barroso-Sousa, Hospital Sı́rio-Libanês, Brası́lia, Brazil from the German PRAEGNANT registry. First Author: Peter A. Fasching,
University Hospital Erlangen, Erlangen, Germany
Background: RT is frequently used for palliation in MBC. In animal models
its use has been reported to induce distant (abscopal) tumor responses when Background: Germline BRCA1/2 (gBRCA) mutations (mt) are some of the
combined with immune checkpoint inhibitors. Here, we report the safety and few actionable alterations in mBC patients. The PARP inhibitors olaparib/
efficacy of palliative RT plus pembrolizumab in a phase II single-arm study in talazoparib are more effective than chemotherapy (ctx) in patients with a
patients (pts) with HR+/HER2- MBC. Methods: Eligible pts had HR+/HER2- gBRCAmt and HER2 negative(-) mBC. In mBC patients the TNT-study
MBC, ECOG PS ,2, indication for palliative RT, and $1 measurable lesion suggested a better progression-free survival (PFS) for gBRCA-mt compared
outside of the RT field; there was no limit on prior lines of therapy. A total RT to patients with a gBRCA1/2 wildtype (wt) when treated with platinum and
dose of 20 Gray was delivered over 5 daily fractions. Pembrolizumab was not with a taxane. Otherwise little is known about the prognostic effect of
given at 200 mg IV 2-7 days before day 1 of RT, then every 3 weeks until gBRCA1/2mt in mBC patients. Methods: PRAEGNANT (NCT02338167)
disease progression. The primary endpoint was objective response rate (ORR) is a prospective mBC registry with a focus on molecular biomarkers. Patients
outside the field of radiation by RECIST v1.1. Using the Simons “optimal” were eligible for this analysis if their mBC was HER2- and treated with ctx for
method, if $ 1/8 pts responded during the first stage, 19 more would be the first time (referred to as first-line ctx). Hormone receptor (HR) positive
enrolled. If $ 3/27 responded, the null hypothesis (ORR=3%) would be patients had to have all hormone therapies exhausted. Mutation frequencies
rejected in favor of a 20% ORR. Predefined secondary endpoints included and their association with patient and tumor characteristics were analyzed.
progression free survival (PFS) and toxicity. Analyses associating PD-L1 Multivariable Cox regression models were built with commonly established
expression, tumor-infiltrating lymphocytes (TIL), and neutrophil/lymphocyte prognostic factors and gBRCA mutation status as predictors of PFS and overall
ratio (NLR) with outcomes were exploratory. Results: Eight women were survival (OS) from first-line ctx. Results: Out of 2932 PRAEGNANT patients,
enrolled into the first stage of the trial; no objective responses were seen, and 576 were HER2- and received first-line ctx. Of those 529 patients with gBRCA
the study was closed to further accrual. The median age was 59y (37-68y), 6 genotype results and follow up information could be analyzed. 24 patients (4.5%)
(75%) had ECOG PS 1, all had bone and 5 (63%) had liver metastases. The had a gBRCAmt (11 BRCA1, 13 BRCA2). Mutation rate in HR positive patients
median number of prior cytotoxic therapies for MBC was 2 (range 0 to 8). was 3.9% (17/432) and 7.2% (7/97) in HR negative patients. Most patients
While one patient had a PR by RECIST criteria, this patient experienced received ctx either as the first treatment in the metastatic setting or after one line of
concurrent clinical progression. Two pts had SD , 16 weeks and 5 had PD as hormone therapy (n=382; 72.2%). Multivariable Cox regression models showed
best response. The median PFS was 1.4 months (95% CI 0.4 – 2.1). All-cause an adjusted hazard ratio for gBRCAmt vs. gBRCAwt patients of 0.70 (95% CI:
adverse events occurred in 87.5% of pts (G3-4, 12.5%). TIL were available for 0.43-1.15) for PFS and of 0.41 (95% CI: 0.18-0.93) for OS. Most frequent ctx
6 pts: 4 had #10%, and 2 . 10%. Among 5 pts with PD-L1 status available, 2 treatments were taxane (52%) or capecitabine based (21%). Additionally, the
were positive. Six pts had NLR . 4. Conclusions: Pembrolizumab combined prevalence of somatic BRCA1/2 mutations in this population will be presented.
with RT was well-tolerated, and no unexpected adverse events were observed; Conclusions: In this HER2- mBC population under ctx gBRCA mutation rates were
however, clinical benefit of the combination was not demonstrated in this within the expected range of about 5%. Within the analyzed population patients
heavily pretreated HR+ population. Clinical trial information: NCT03051672. with a gBRCA mutation seemed to have a better OS than patients without a
mutation. PFS results pointed in the same direction without statistical signifi-
cance. However, with only 24 mutations replication of these results in additional
cohorts is warranted. Clinical trial information: NCT02338167.

1049 Poster Session (Board #130), Sun, 8:00 AM-11:00 AM 1050 Poster Session (Board #131), Sun, 8:00 AM-11:00 AM
Clinical significance of circulating tumor cells (CTCs) in hormone receptor- Exploratory analysis of the effect of taselisib on downstream pathway
positive (HR+) metastatic breast cancer (MBC) patients (pts) receiving modulation and correlation with tumor response in ER-positive/HER2-
letrozole (Let) or Let plus bevacizumab (Bev): CALGB 40503 (Alliance). negative early-stage breast cancer from the LORELEI trial. First Author:
First Author: Mark Jesus Mendoza Magbanua, University of California San Paolo Nuciforo, Molecular Oncology Group, Vall d’Hebron Institute of On-
Francisco Helen Diller Family Comprehensive Cancer Center, San Francisco, CA cology (VHIO), Barcelona, Spain
Background: CALGB 40503 randomized HR+ MBC postmenopausal pts to Let alone Background: Taselisib (T) is an oral, potent, selective inhibitor of Class I PI3-
or Let+Bev as first-line therapy. Adding Bev to Let prolonged progression-free survival kinase with enhanced activity against PIK3CA mutant cancer cells. Results
(PFS) but not overall survival (OS) (Dickler JCO 2016). We performed a correlative from the LORELEI trial have demonstrated a significant improvement in ORR
study to assess prognostic and predictive value of CTCs in this population. (objective response rate) by centrally assessed magnetic resonance imaging in
Methods: Blood was collected prior to initiation of treatment. CTCs were enumerated
using US FDA-cleared CellSearch assay; samples with $5 CTCs per 7.5 mLs of blood
all randomized patients as well as in the PIK3CA mutant (MT) cohort treated
were considered CTC-positive (CTC+). Correlation of CTCs with PFS and OS was with neoadjuvant T plus letrozole (L) compared to placebo (P) plus L. Here we
assessed using Cox regression analysis. Median follow-up was 39 months (mo). present the results of exploratory analyses of selected pathway-related
Results: Of 343 pts treated, 294 had CTC data and were included in this analysis. phosphoproteins. Methods: Baseline (BL) and week3 (W3) tumor biopsies
Original study results that showed improved PFS (HR = 0.75; 95% CI: 0.59-0.96) but were obtained from 334 patients enrolled in the trial. Phosphoproteins
not OS (HR = 0.87; 95% CI: 0.65-1.18) in pts receiving Let+Bev compared to Let (pAKT, pPRAS40 and pS6) were analyzed by IHC. BL levels as well as
were recapitulated in this subset. In multivariable analysis, CTC+ pts (31%) had changes from BL to W3 were correlated with response assessed either by
significantly reduced PFS (HR = 1.49; 95% CI: 1.12-1.97) and OS (HR = 2.08; 95% ORR or cell cycle arrest (Ki67 at W3 , 2.7%). Results: In the overall pop-
CI: 1.49-2.93) compared to CTC- pts. Moreover, CTC+ pts who did not receive Bev had ulation, BL phosphoproteins levels were similar between the T and P arms.
worse PFS (HR = 2.31; 95% CI: 1.54-3.47) and OS (HR = 2.64; 95% CI: 1.59-4.40)
Higher pAKT (p , 0.001) and pPRAS40 (p = 0.004) levels were observed in
(Table). CTC+ pts who received Bev had numerically longer median PFS (18.0 vs.
7.0 mo) and OS (33.6 vs. 27.1 mo) compared to CTC+ pts with no Bev; however, tests
MT vs wild-type (WT), whereas the opposite result was found for pS6 (p =
for interaction between CTC status and Bev (yes vs. no) were not statistically sig- 0.03). Treatment-induced absolute changes of phosphoproteins adjusted for
nificant for PFS (p=0.70) or OS (p=0.84). Conclusions: CTCs were highly prognostic in BL levels were not significantly different between the T and P arms in the
this study involving addition of Bev to first-line Let in postmenopausal HR+ MBC. Further overall population, except for pPRAS40 with higher decrease in the T arm (p =
research to determine the potential predictive value of CTCs in the setting of both 0.014). After stratification for PIK3CA genotype, a significantly greater de-
metastatic disease and early breast cancer is warranted. Support: U10CA180821, crease in expression levels was observed for pPRAS40 (p , 0.001) and pS6
U10CA180882; Genentech; https://acknowledgments.alliancefound.org; NCT00601900. (p = 0.020) in MT tumors treated with T. The treatment effects were not
Survival in HR+ MBC pts receiving Let or Let+Bev stratified by CTC status. Clinical trial significantly different in the WT population. A trend for an association between
information: NCT00601900. decrease in pS6 levels at W3 and improved ORR was observed in the MT (p =
Median survival in Adjusted Adjusted Likelihood-Ratio 0.08) and T (p = 0.09) subgroups. The magnitude of pS6 suppression at W3
Total Events mo (95% CI) Hazard Ratio (95% CI) p-value
was higher in tumors achieving a cell cycle arrest in the MT/T subgroup (biserial
PFS
CTC-: Let+Bev 108 70 18.4 (15.0-23.5) 1.0
0.0012 correlation = -0.473). Conclusions: Exploratory analyses of phosphoproteins
CTC-: Let 94 74 14.7 (11.4-18.9) 1.44 (1.02-2.02) showed bioactivity of taselisib as indicated by downstream pathway
CTC+: Let+Bev 46 42 18.0 (13.6-23.7) 1.44 (0.98-2.13) suppression. Translational research aiming to integrate these results with
CTC+: Let 46 38 7.0 (2.8-10.9) 2.31(1.54-3.47)
OS 0.0003 additional exploratory biomarkers data is currently ongoing. Clinical trial
CTC-: Let+Bev 108 35 49.1 (42.4-NE) 1.0
CTC-: Let 94 44 45.0 (40.1-50.1) 1.29 (0.82-2.03)
information: NCT02273973.
CTC+: Let+Bev 46 34 33.6 (26.6-40.0) 2.20 (1.37-3.55)
CTC+: Let 46 28 27.1 (20.6-36.1) 2.64 (1.59-4.40)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Breast Cancer—Metastatic 47s

1051 Poster Session (Board #132), Sun, 8:00 AM-11:00 AM 1052 Poster Session (Board #133), Sun, 8:00 AM-11:00 AM
PIPA: A phase Ib study of b-isoform sparing phosphatidylinositol 3-kinase Molecular characterization and monitoring of patient ctDNA in phase I study
(PI3K) inhibitor taselisib (T) plus palbociclib (P) and fulvestrant (FUL) in of H3B-6545 in ER+ MBC. First Author: Vicki Rimkunas, H3 Biomedicine,
PIK3CA-mutant (mt) ER-positive and taselisib (T) plus palbociclib (P) in Cambridge, MA
PIK3CA-mutant (mt) ER-negative advanced breast cancer. First Author:
Background: Because of lack of effective treatment in endocrine resistant
Javier Pascual, Royal Marsden Hospital and The Institute of Cancer Research,
metastatic breast cancer (MBC), we developed H3B-6545, a novel selective
London, United Kingdom
ERa covalent antagonist, capable of irreversibly inactivating both wild-type
Background: PI3K and CDK4/6 inhibitors synergise in ER+ve and –ve and mutant ERa. The aims of this study are to 1) characterize hotspot
PIK3CA-mt breast cancer (BC) models. Escalation phase of this study set mutation profiles in heavily pretreated MBC and correlate ESR1, PIK3CA
recommended phase 2 dose (RP2D) at P 125mg on a 3/1 scheme plus T 2mg and AKT1 mutations in plasma vs tumor tissue 2) determine if mutations in
daily (Lim, ASCO 2017). Here we present the results of expansion cohorts for ESR1 or PIK3CA predict response to H3B-6545 and 3) evaluate if longitudinal
PIK3CA-mt BC patients (pts). Methods: The primary objective was to assess tracking of ctDNA correlates with response to H3B-6545. Methods: Fresh
the confirmed objective response rate (ORR) of the P + T + FUL triplet in pts plasma samples were collected at baseline (predose), cycle 1 day 15 (C1D15),
with measurable PIK3CA-mt ER+ve HER2-ve advanced BC, with up to two C2D1, C3D1 and every 8 weeks thereafter with a final sample collection at
prior lines of chemotherapy for advanced disease. PIK3CA mutation was disease progression. At baseline, BEAMing digital PCR was used to evaluate
assessed in tissue or plasma DNA analysis. Exploratory objectives included hotspot mutations in ESR1, PIK3CA and AKT1. Patient specific ctDNA
assessment of efficacy of P + T in a cohort of pts with PIK3CA-mt advanced mutations were subsequently monitored by ddPCR. Baseline tumor biopsies
ER-ve BC. Safety is reported overall for 44 patients including an additional were subjected to a targeted Next Generation Sequencing (NGS) panel to
cohort of 7 PIK3CA-unknown ER+ve BC pts treated with P + T + letrozole identify hotspot mutations. Results: In 77% of patients (30/39), mutations
(LET). For the P + T + FUL triplet a Simon minmax design was used, with were detected at baseline by the BEAMing assay and of those, 21/39, 16/39
6 responses in 25 patients required to declare efficacy. Results: We and 3/39 had mutations in ESR1, PIK3CA and AKT1, respectively. 20%
recruited 24 assessable patients with PIK3CA-mt ER+ve HER2-ve advanced (9/39) of patients exhibited co-mutations in ESR1 and PIK3CA. In 60% (9/15)
BC, median age 57 (42-74), median 3 (1-9) prior therapy lines for advance of patients, DNA mutations identified by the plasma BEAMing assay were also
disease, with 24 (100%) receiving prior endocrine therapy and 23 (96%) detected in the tumor biopsy whereas; DNA mutations found in tissue were also
prior aromatase inhibitor. ORR was 33% (8/24, 95% CI 16-55%), with detected in plasma in 86% (12/14) of cases. Serial ctDNA monitoring revealed
median progression free survival (PFS) 7.9m (95% CI 5.6-11.8). For the 11 that in patients with confirmed partial responses (3/3), ctDNA levels were
assessable PIK3CA-mt ER-ve pts (8 HER2-ve, 3 HER2+ve) receiving P + T, undetectable by C2D1. In contrast, ctDNA levels increased from baseline in
ORR was 0% (0/11), clinical benefit rate (CBR) 27% (3/11) and median PFS 3/4 patients with progressive disease. Exploration of ctDNA ratios (day 15/
4.3m (95% CI 1.8-6.1). Most common AEs across all cohorts were neu- baseline and day 30/baseline) and correlations of PIK3CA and ESR1 muta-
tropenia (80%), fatigue (50%), mucositis (50%) and thrombocytopenia tions with response to H3B-6545 will be presented. Conclusions: ctDNA is a
(30%). Most common grade 3/4 AEs were neutropenia (57%) and rash reliable sample type for assessing ESR1, PIK3CA, and AKT1 mutations in
(11%). Translational research is ongoing. Conclusions: The triplet of P + T + MBC, overcoming the challenges of obtaining biopsies in the metastatic
FUL has promising efficacy in pre-treated PIK3CA-mt ER+ve advanced BC. setting. In addition, ctDNA dynamics may be a useful tool to monitor the
A subset of patients with PIK3CA-mt ER-ve advanced BC had clinical benefit efficacy of H3B-6545. Clinical trial information: NCT03250676.
from P + T. The combination of P + T +/- FUL/LET was well tolerated with
anticipated AEs. Clinical trial information: NCT02389842.

1053 Poster Session (Board #134), Sun, 8:00 AM-11:00 AM 1054 Poster Session (Board #135), Sun, 8:00 AM-11:00 AM
A large retrospective analysis of CDK 4/6 inhibitor retreatment in ER+ Dose-escalation study of SAR439859, an oral selective estrogen receptor
metastatic breast cancer (MBC). First Author: Carlos Henrique dos Anjos, (ER) degrader (SERD), in postmenopausal women with ER+/HER2- metastatic
Memorial Sloan Kettering Cancer Center, New York, NY breast cancer (mBC). First Author: Aditya Bardia, Massachusetts General
Hospital Cancer Center, Harvard Medical School, Boston, MA
Background: Sequential retreatment with endocrine therapy (ET) has been
the clinical paradigm for ER+ MBC due to persistent dependence on hormone Background: SERDs result in ER competitive antagonism and degradation
signaling. Recently CDK4/6i + ET have improved PFS and are routinely uti- and can block signaling in ER-dependent tumors resistant to other endocrine
lized in the first/second- line setting. Whether this paradigm of sequential therapies. This study investigates SAR439859, a potent oral SERD, +/-
retreatment holds for CDK 4/6i is unknown. To evaluate the potential benefit of palbociclib in ER+/HER2- mBC. Here are preliminary results, as of 28 Nov
CDK4/6i re-treatment we conducted this retrospective analysis. Methods: We 2018, for single-agent SAR439859 dose escalation. Methods: Part A of this
identified ER+/HER2- MBC pts treated with $ 2 lines of CDK4/6i at our Phase 1/2 study (NCT03284957; TED14856) assessed SAR439859 dose
institution between 2015-2018. We categorized pts based on reason for escalation (dose range: 20–600 mg once daily [QD]; 3 + 3 design) in
discontinuation of their first CDK4/6i: cohort 1 – switch to alternate CDK4/6i postmenopausal women with ER+/HER2- mBC treated for $ 6 months with
due to toxicity; cohort 2 – retreatment with same CDK4/6i beyond pro- prior endocrine therapy and # 3 chemotherapies in the advanced setting.
gression with change of ET and cohort 3- switch to alternate CDK4/6i as Endpoints: dose-limiting toxicities (DLTs); maximum tolerated dose (MTD);
monotherapy or with same or another ET. We analyzed pt demographics, safety; pharmacokinetics (PK); tumor response (RECIST 1.1); pharmacody-
imaging reports and time to subsequent therapy (TTST) for every CDK4/6i namic (PD) inhibition of ER occupancy (18FES-PET scan). Results: Patients
line for each cohort. If a pt received . 2 lines of CDK4/6i, then that pt was (pts; n = 16) had a median age of 59.5 years (range 40–79), ECOG perfor-
evaluated for every CDK4/6i exposure. Results: 135 pts received $ 2 lines of mance status of 0 (62.5%) or 1 (37.5%) and a median of three prior anticancer
CDK4/6i treatment (Tx). Cohorts 1, 2 and 3 had 23, 43 pts and 84 pts therapies (range 1–8) in the advanced setting (endocrine therapy n = 16;
respectively. In Cohort 2, 95% of pts received 2 subsequent CDK 4/6i + ET chemo/targeted therapy n = 13). All pts had $ 1 treatment emergent adverse
Tx; 56% had the second CDK4/6i in second-line met setting. TTST1 (1st event (mostly grade 1–2); most frequent were asthenia/fatigue (43.8%), hot
CDK4/6i Tx) was 9.6m (95% CI 4.9 - 11 m), TTST2 (second CDK4/6i Tx) flushes (37.5%), nausea (37.5%), diarrhea (31.3%), constipation (31.3%),
was 4.5m (95% CI 3.3 – 7.6 m) and 35% had TTST2 $ 24 weeks. For Cohort and decreased appetite (31.3%). There were no DLTs at any of the five dose
3, 48% were retreated with a different CDK 4/6i in $ fifth-line. 51% re- levels (maximum administered dose: 600 mg QD); MTD was not reached.
ceived 2 subsequent CDK 4/6i Tx with 18% in second-line met setting. In 18FES-PET scans, signal inhibition . 87% occurred with plasma
TTST1 was 9.6 m (95% CI 5.9 – 12 m), TTST2 was 4.4 m (95% CI 3.8 – 5.9 concentrations . 100 ng/mL. There was a dose proportional increase of
m) and 29% had TTST2 $ 24 weeks. In cohort 3, 29% (n=24) pts had PD as exposure up to 400 mg after repeated QD doses. Average Ctrough was
best response at the time of first CDK4/6i exposure but 29% (7/24) reached after repeated 400 mg QD allowing 90% of 18FES-PET signal
achieved a radiologic response to their second CDK4/6i Tx. Pts had tu- inhibition. One pt (6.3%) had confirmed partial response (150 mg QD); eight
mor sequencing using MSK-IMPACT which will be correlated with TTST. (50%) had stable disease (SD) including three (18.8%) long-term SD
Conclusions: This large single institution retrospective analysis suggests ($ 24 weeks); seven (43.8%) had progressive disease. Conclusions: SAR439859
that retreatment with a CDK4/6i regimen should be evaluated in pro- had a favorable safety profile, high ER occupancy and encouraging antitumor
spective trials. Additionally, despite PD as best response with the first activity (to be confirmed in dose expansion) in pretreated pts with ER+/HER2-
CDK4/6i (palbociclib/ribociclib) regimen, a subset of pts had radiologic mBC. With no DLTs and MTD, 400 mg QD was selected for expansion cohorts
response to a subsequent abemaciclib-containing regimen, which is an based on safety, PD and PK data. Funding: Sanofi. Clinical trial information:
hypothesis generating observation. NCT03284957.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
48s Breast Cancer—Metastatic

1055 Poster Session (Board #136), Sun, 8:00 AM-11:00 AM 1056 Poster Session (Board #137), Sun, 8:00 AM-11:00 AM
Real-world evidence of male breast cancer (BC) patients treated with Lasofoxifene as a potential treatment for ER+ metastatic breast cancer. First
palbociclib (PAL) in combination with endocrine therapy (ET). First Author: Author: Muriel Laine, University of Chicago, Chicago, IL
Cynthia Huang Bartlett, Pfizer Inc, Collegeville, PA
Background: Estrogen receptor positive (ER+) metastatic breast cancers
Background: The rarity of BC in men limits the feasibility of randomized (MBC) that express constitutively active somatic ESR1 mutations at Y537S
clinical studies in this population. Treatment guidelines recommend that and D538G allow tumors to progress in the presence of approved endocrine
men with BC be treated similarly to postmenopausal women. PAL, a cyclin- therapies. For patients with ER+ MBC, fulvestrant is the first line of treat-
dependent kinase 4/6 inhibitor, is used in men with metastatic BC (mBC) in ment. Palbociclib or other CDK4/6 inhibitors are now being included.
real-world clinical practice, presenting an opportunity to utilize real-world Preliminary studies show that lasofoxifene, a selective ERa modulator
evidence to enable healthcare providers to assess novel agents in this space. (SERM), was effective in reducing tumor growth in an endocrine resistant
Methods: Two parallel approaches were taken. In the first approach, pharmacy xenograph model expressing ERa mutations Y537S or D538G. Additionally,
and medical claims data from IQVIA Inc were retrospectively analyzed to lasofoxifene more effectively inhibited the development of liver and lung
describe the treatment patterns and duration of PAL + ET (aromatase inhibitor metastases than fulvestrant. Lasofoxifene is currently under evaluation in a
or fulvestrant) compared to ET in men with mBC. The second approach was a phase 2 study. Because certain combinations of a hormonal agents like
retrospective analysis of data derived from electronic health records in the fulvestrant improved efficacy, we investigated the combination of palbociclib
Flatiron Health database to understand real-world clinical response to PAL + and lasofoxifene as a potential therapeutic for mutant ESR1 MBC. We hy-
ET vs ET alone. Median duration of treatment (mDOT) was estimated by the pothesized that this combination should improve outcome and compared it to a
Kaplan-Meier method. Results: Between Feb 2015 and Apr 2017, 12.9% combination with fulvestrant. Methods: We first determined the optimal dose
(147/1139 [IQVIA dataset]) of men receiving treatment for mBC were pre- of lasofoxifene in an intraductal (MIND) xenograph model of MCF-7 cells that
scribed PAL + ET for any line of therapy. The mDOT in the first-line setting was express active ERa Y537S and D538G. Subsequently, we performed combi-
numerically longer in the PAL cohort (n=37) compared with the non-PAL nation studies with lasofoxifene (10mg/kg 5/week SQ) +/- palbociclib
cohort (n=214; 8.5 vs 4.3 mo, respectively). In particular, mDOT in the first- (100mg/kg gavage, 5/week) or fulvestrant (5mg/mouse/week, SQ) +/-
line setting was longer with PAL + letrozole (LET; n=26) than with LET alone palbociclib. Results: Lasofoxifene alone was significantly more effective
(n=63; 9.4 vs 3.0 mo, respectively). In the Flatiron Health dataset between than fulvestrant at inhibiting the metastasis of both MCF7 Y537S and
Feb 2015 and July 2017, the real-world maximum response rate in the PAL + D538G tumors to the lungs and liver. Lasofoxifene + palbociclib was more
ET cohort across all lines of therapy in the mBC setting (n=12) was 33.3% effective than fulvestrant + palbociclib at reducing primary tumor growth;
(2 complete responses [CR], 2 partial responses [PR]) vs 12.5% (0 CR, 1 PR) both combinations demonstrated an increased response. Lasofoxifene +
for the ET alone cohort (n=8). Conclusions: The real-world data sources used in palbociclib was more effective at inhibiting liver metastasis than either
this study support that men with mBC derive clinical benefit from the addition drug alone and was more effective than fulvestrant + palbociclib at re-
of PAL to ET. Given the challenges of conducting randomized clinical trials in ducing metastasis to the liver and lung. Structural studies showed that
men with mBC, noninterventional, real-world evidence data appear to be useful lasofoxifene effectively disrupts the active conformation of the ERa Y537S
to delineate the benefit of such therapies in this setting. Funding: Pfizer. ligand-binding domain. Conclusions: These results demonstrate that
lasofoxifene, in combination with CDK4/6 inhibitors like palbociclib, has
promise for treating endocrine therapy resistant ER+ MBC patients whose
tumors express activating ESR1 mutations, more effectively than either
drug alone.

1057 Poster Session (Board #138), Sun, 8:00 AM-11:00 AM 1058 Poster Session (Board #139), Sun, 8:00 AM-11:00 AM
A multicenter analysis of abemaciclib after progression on palbociclib in Everolimus and exemestane for the treatment of metastatic hormone
patients (pts) with hormone receptor-positive (HR+)/HER2- metastatic breast receptor-positive breast cancer patients previously treated with CDK4/6
cancer (MBC). First Author: Seth Andrew Wander, Massachusetts General inhibitor based therapies. First Author: Madeline Cook, Oregon Health
Hospital Cancer Center, Boston, MA and Sciences University, Portland, OR
Background: Cyclin-dependent kinase 4/6 inhibitors (CDK4/6i) are widely Background: The combination of everolimus (EVE) and exemestane (EXE) is
used for pts with HR+/HER2- MBC. The MONARCH-1 trial of abemaciclib approved as second line endocrine therapy for metastatic hormone receptor
monotherapy in pre-treated pts demonstrated a median progression free positive breast cancer (mHRBC) patients who progressed on non-steroidal
survival (PFS) of 6.0 months, leading to approval as monotherapy in a CDK4/ aromatase inhibitor (NSAI) therapy based on the BOLERO-2 trial. However,
6i-naı̈ve population. There are no data on abemaciclib in HR+/HER2- MBC none of the patients in BOLERO-2 received prior CDK4/6 inhibitors, which
after progressive disease (PD) with CDK4/6i. Methods: We evaluated clinical have since become standard of care for metastatic HRBC. As such, the
outcomes in pts with HR+/HER2- MBC who received abemaciclib following clinical benefit of EVE + EXE in mHRBC patients previously treated with
PD on prior palbociclib or ribociclib at 4 US academic centers. We con- CDK4/6 inhibitors remains unknown. Methods: We reviewed patients $18yo
ducted genomic analysis utilizing next-generation sequencing of tissue with mHRBC treated with EVE + EXE following NSAI alone or NSAI + CDK
samples and blood (cell-free/cfDNA) when available. Results: From 2/2015 4/6 inhibitor at our institution between 2012-2018. Data collected included
through 1/2019, 58 pts with HR+/HER2- MBC received abemaciclib fol- patient and tumor characteristics, therapies in the metastatic setting,
lowing PD on prior palbociclib. 20 pts (34%) received sequential courses of special interest adverse events, and clinical outcomes. The primary objective
therapy, while 38 pts (66%) had at least one intervening non-CDK4/6i was comparing PFS for EVE + EXE therapy between patients who received
regimen. 14 pts (24%) received abemaciclib monotherapy and 44 pts (76%) prior CDK4/6 inhibitor therapy and those who did not. Secondary endpoints
received it in combination with an antiestrogen, including fulvestrant (52%), included overall survival (OS). Patient features were summarized with de-
an aromatase inhibitor (22%), and tamoxifen (2%). 22 pts (38%) required scriptive statistics and time-to-event measures were estimated using the
dose reduction, while 7 (12%) discontinued due to toxicity. At data cutoff Kaplan-Meier method. Differences between groups were tested with Fisher’s
(1/23/2019), 20 pts (34%) had early PD (duration , 90 days), while 21 pts exact, Kruskal-Wallis, or log-rank test. Results: Thirty-three patients were
(36%) had treatment duration exceeding 6 months, including 10 who re- included in the study; 17 had prior CDK4/6 inhibitor therapy and 16 did
main on treatment at interim analysis (range 181-413 days). The median not. Subjects that took EVE + EXE for , 28 days were excluded. Patient
PFS was 5.8 months (95%CI 3.4 – 8.0). Preliminary analysis of cfDNA characteristics, including prior therapies and sites of metastatic disease,
revealed RB1 and FGFR1 alterations in pts with PD on abemaciclib. Ad- were not significantly different. There was no significant difference in PFS
ditional analyses with mature clinical data and genomic sequencing will be (median 5.7 vs 4.7 months, p = 0.890) or OS (median 17.8 vs 11.4 months,
provided at the meeting. Conclusions: This is the first multi-center experi- p = 0.177) between patients who received prior CDK4/6 inhibitors and those
ence to demonstrate that a substantial proportion of pts continue to derive who did not, respectively. Steroid mouthwash use was associated with a
clinical benefit with abemaciclib after prior CDK4/6i, highlighting the po- reduced incidence of stomatitis. Conclusions: EVE + EXE is well tolerated
tential for its use following CDK4/6 blockade. A second subset had early and shows similar efficacy in metastatic HRBC patients who received CDK4/
progression, suggesting cross-resistance to CDK4/6i via common pathways. 6 inhibitor therapy and those who did not. There was a non-significant trend
Future effort should be directed towards validating potential biomarkers towards improved OS in the CDK4/6 inhibitor group that needs to be further
to guide optimal utilization of continued CDK4/6 blockade in HR+/HER2- evaluated in larger patient cohorts.
MBC.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Breast Cancer—Metastatic 49s

1059 Poster Session (Board #140), Sun, 8:00 AM-11:00 AM 1060 Poster Session (Board #141), Sun, 8:00 AM-11:00 AM
Phase I dose escalation of H3B-6545, a first-in-class highly Selective ERa Association of drug-related polymorphisms with palbociclib-related neutropenia:
Covalent Antagonist (SERCA), in women with ER-positive, HER2-negative Pharmacogenetic analysis of PALOMA-2/-3 (P2/3). First Author: Hiroji Iwata,
breast cancer (HR+ BC). First Author: Erika Paige Hamilton, Tennessee Aichi Cancer Center Hospital, Nagoya, Japan
Oncology, PLLC and Sarah Cannon Research Institute, Nashville, TN
Background: The most common PAL treatment related adverse event is
Background: H3B-6545 inactivates both wild-type and mutant ERa by neutropenia. ABCB1 and ERCC1 variants are associated with increased
targeting cysteine 530 and enforcing a unique antagonist conformation. chemotherapy drug exposure and CYP3A7*1C may be associated with reduced
Methods: Women with locally advanced or metastatic HR+ BC are treated exposure. Pharmacogenetic analyses of these variants in patients (pts) from
(tx) with H3B-6545 administered once daily orally over a 28 day cycle after P2/3 may reveal associations between single nucleotide polymorphisms (SNPs)
progression on at least one hormonal therapy and at least one additional and early occurrence of grade 3/4 (G3/4) neutropenia. Methods: ABCB1
therapy/regimen. Dose escalation uses a 3+3 design with the option to (rs1045642, rs1128503), ERCC1 (rs3212986, rs11615), and CYP3A7*1C
backfill previously cleared doses and allows for intrapatient dose escalation. (rs45446698) variants were analyzed in germline DNA from pts with HR+/
This phase 1 explores the safety, pharmacokinetics and pharmacodynamics HER2– advanced breast cancer from P2 (n=584) and P3 (n=442) by TaqMan
of H3B-6545 in women with HR+ BC to identify the recommended Phase 2 assay. Association between variants and incidence of G3/4 low absolute
dose. Results: As of 10-Dec-2018, 32 pts have been tx with H3B-6545 at neutrophil count (ANC) at Cycle 1 Day 15 (C1D15) was assessed with the exact
doses of 100 to 450 mg/day; 97% had prior tx with a CDK4/6 inhibitor and Cochran-Armitage trend test. Odds ratios (OR) and 95% confidence intervals
56% had received $3 lines of prior anti-cancer therapy. No dose-limiting (CI) were estimated by logistic regression. Results: In total, 652 PAL-treated
toxicities and only one Grade 3 treatment related adverse event (TRAE) have pts had SNP, race, and C1D15 ANC data. Minor allele frequencies (MAF),
been observed (lymphocyte count decrease). The most common ($10%) incidence rates, and relative risk of G3/4 C1D15 ANC for ABCB1 and ERCC1
TRAEs include asymptomatic sinus bradycardia, diarrhea, nausea, fatigue, variants are given in the Table. CYP3A7*1C was only found in non-Asians (MAF
anemia, decreased appetite, and hot flush. H3B-6545 was rapidly absorbed 6%). Conclusions: This is the first comprehensive assessment of pharmaco-
with a tmax of 2-4 h. Plasma concentration increased with dose from 100 to genetic data from P2/3. ABCB1 and ERCC1 SNP allele frequencies differ be-
450 mg, and was similar on C1D1 and C1D15. Consistent with the H3B- tween Asians and non-Asians. Despite combining P2/3 data, we lacked power to
6545 mechanism of action and preclinical data, H3B-6545 inhibits ER detect moderate associations; further investigation of these SNPs with G3/4
target gene expression and shows a 50% decrease in Ki67 levels across all C1D15 ANC is warranted. Pfizer Clinical trial information: NCT01740427,
dose levels post-tx. ESR1 (60%) and PIK3CA (34%) mutations were de- NCT01942135.
tected in plasma at baseline and changes in mutant allele frequencies show Asian Non-Asian2
correlation in response to tx. Stable disease was observed in 15 pts (47%) (n=122) (n=530)
and 34% of pts completed at least 6 months of tx. Partial responses (PRs) 67 (54.9) 123 (23.2)
G3/4 C1D15 ANC, n (%)
were observed in 3 pts: 1 pt (mutant) received 2 prior lines of therapy and SNP Alleles MAF1 (%) OR (95% CI) Ptrend MAF (%) OR (95% CI) Ptrend
2 pts (1 mutant and 1 wild-type) received .5 prior lines of therapy including ABCB1_rs1128503 T.C 35 TC 1.29 (0.60-2.76) 0.28 58 TC 0.61 (0.37-1.03) 0.03
fulvestrant and capecitabine; all 3 pts received a prior CDK4/6 inhibitor. CC 2.00 (0.60-6.66) CC 0.51 (0.29-0.90)
Conclusions: H3B-6545 has been well-tolerated up to the 450 mg dose level ABCB1_rs1045642 C.T 38 CT 0.60 (0.27-1.31)
TT 0.78 (0.26-2.33)
0.43 52 CT 1.41 (0.83-2.42)
TT 1.48 (0.81-2.72)
0.23

with early signs of single-agent anti-tumor activity in a post CDK4/6 setting. ERCC1_rs11615 G.A 29 GA 0.75 (0.35-1.58) 0.19 60 GA 1.10 (0.60-2.02) 0.09
Dose escalation continues in pts with advanced HR+ BC. Clinical trial in- ERCC1_rs3212986 C.A 27
AA 0.33 (0.08-1.46)
CA 1.60 (0.76-3.38) 0.11 26
AA 1.58 (0.85-2.93)
CA 0.90 (0.59-1.38) 0.37
formation: NCT03250676. AA 3.19 (0.60-17.0) AA 0.68 (0.29-1.60)
1
Minor allele in Asians. 2Non-Asians are predominantly self-reported (94%) White Caucasians.

1061 Poster Session (Board #142), Sun, 8:00 AM-11:00 AM 1062 Poster Session (Board #143), Sun, 8:00 AM-11:00 AM
RIBECCA: A phase IIIb, multicenter, open label study for women with The clinical utility of strict laboratory monitoring of CDK 4/6 inhibitors in
estrogen receptor-positive locally advanced or metastatic breast cancer metastatic breast cancer patients. First Author: Urvi Patel, EUH, Decatur,
treated with ribociclib (LEE011) in combination with letrozole—Results GA
of the second interim analysis. First Author: Arnd Nusch, Onkologische
Background: Cyclin-dependent kinase 4/6 inhibitors (CDK 4/6i) are widely
Praxis Velbert, Velbert, Germany
used in the treatment of hormone receptor positive, metastatic breast cancer
Background: RIBECCA is a national, multi-center, open-label, single-arm phase (MBC). CDK 4/6is are well-tolerated and highly effective, but can cause
IIIb trial assessing the efficacy and safety of ribociclib in combination with neutropenia. Therefore, the FDA mandates assessment of the absolute neu-
letrozole in a patient population similar to the populations of MONALEESA-2, -3 trophil count (ANC) every 2 weeks for the first 2 cycles of therapy and monthly
and -7. Here we present the results of the second preplanned interim analysis. thereafter. Strict ANC monitoring is optimal in theory, but presents a challenge
Methods: Main inclusion criteria allowed enrollment of men or women with in the real world, impacting quality of life for patients (pts) and adding financial
metastatic or locally advanced breast cancer not amenable to curative treatment burden to the healthcare system. This study aims to evaluate whether strict
by surgery or radiotherapy, and histological or cytological confirmation of ER+, monitoring of ANC significantly impacts treatment decisions, pt safety, and
HER2- breast cancer, irrespective of their menopausal status. The primary disease outcomes. Methods: A retrospective chart review of 160 pts with MBC
objective is to assess the clinical benefit rate (CBR) after 6 months. Secondary was conducted at the Winship Cancer Institute in pts prescribed CDK 4/6is
objectives include: progression free survival (PFS), overall survival (OS), safety from Feb 2015 to Jan 2019. The pt’s ANC at C1D1, C1D14, C2D1, C2D14
and changes in quality of life. Results: The cut-off date for this second interim were recorded along with various pt outcomes. Pts were divided into strict
analysis was 12 months after the last patient was enrolled in the pretreated and monitoring (SM) or relaxed monitoring (RM) groups. SM is defined as pts who
premenopausal cohort. Here we report a preliminary analysis on safety, efficacy received ANC testing within 72h of C1D14, C2D1, and C2D14. RM was
and quality of life. We describe the baseline characteristics, safety data and the defined as not meeting those criteria. Results: Palbociclib was used in 152
clinical benefit rate (CBR) at 24 weeks, of patients (pts) with at least 24 weeks pts; abemaciclib was used in 8. Average age was 58. Study population was
follow up (n = 411). The median observation time for all patients in this analysis 55% Caucasian; 38% African American; 6% Asian. Pts had an average of 3
was 8.1 months. Baseline characteristics: of 411 pts, 409 were female and 2 lines of prior therapy and average of 9 cycles on a CDK 4/6i. 71% of pts
male. Median age: 64 yrs; 46 pts pre-or perimenopausal, 363 postmenopausal; suffered neutropenia (40% grade 3, 4% grade 4, 2.5% febrile neutropenia).
ECOG 0-1: 97.0%; median time since first recurrence: 1.6 months; 72.5 % pts PFS and OS did not statistically differ between groups (Table), but OS data is
had bone metastases (40.6% bone only), 30.7% liver, 27.3% lung and 29.2% immature. Conclusions: This study analyzed a more diverse and heavily pre-
other metastases. Median relative dose intensity was 0.905 for ribociclib and 1 treated MBC population than previous CDK 4/6 studies. The prevalence of
for letrozole. The most common treatment emergent AEs (all grades) were neutropenia (grade 3/4) aligns with results described in the PALOMA trial. PFS
neutropenia and/or neutrophil count decreased (54.5%), nausea (40.6%), fa- was higher in the RM group than in the SM group, suggesting strict monitoring
tigue (36.5%), alopecia (32.6%), leukopenia or WBC decreased (29%), for neutropenia does not improve PFS. If ANC monitoring parameters could be
nasopharyngitis (23.1%), diarrhea (21.4%), ALT increased (20.7%), AST in- safely relaxed for CDK 4/6i, it could improve quality of life and decrease fi-
creased (19%). The CBR by week 24 was 67.6%. Preliminary information on nancial toxicity across the system. This is a topic worthy of further study.
quality of life will be provided. Conclusions: The results of the second interim Neutropenia Grade 3 Neutro- Grade 4 Neutro- Febrile Neutro-
analysis in this additional patient population are in line with data published from N mPFS mOS (%) penia (%) penia (%) penia (%)
the pivotal phase III studies MONALEESA-2, MONALEESA-3 and MONALEESA- SM 53 15 months (95% CI 36 months 83.1 48.5 4.6 4.6
7. Clinical trial information: NCT03096847. RM
0.2-37.7)
72 27.5 months (95%
(22.8-NR)
NR 66.3 35.7 1.2 1.1
CI14-69.1)
P-value 0.198 0.056 0.021 0.115 0.319 0.310

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
50s Breast Cancer—Metastatic

1063 Poster Session (Board #144), Sun, 8:00 AM-11:00 AM 1064 Poster Session (Board #145), Sun, 8:00 AM-11:00 AM
Effect of metformin on PARP inhibitors-induced epithelial-mesenchymal AL101 mediated tumor inhibition in notch-altered TNBC PDX models. First
transition and PD-L1 expression in triple-negative breast cancer. First Author: Author: Esther Channah Broner, Johns Hopkins Medical Center, Baltimore,
Ye Han, China Medical University Affiliated Shengjing Hospital, Shenyang, MD
China
Background: The Notch pathway is activated during mammary gland de-
Background: Poly (ADP-ribose) polymerase (PARP) inhibitors have emerged velopment and has been implicated as a key driver in breast cancer. There is
as promising targeted therapies for BRCA-mutated cancers by blocking an urgent need to identify new therapeutic strategies for triple-negative
repair of DNA double-strand breaks. However, resistance to PARP inhibitors breast cancer (TNBC), a sub-type associated with poor prognosis and no
have been described in some patients lowering overall response rates. The available targeted therapies. Notch gain of function (GOF) genetic alter-
mechanisms underlying PARP inhibitor (PARPi) resistance are an area of ations are potential tumor drivers found in ~10% of TNBC. This motivated
active investigation. Methods: PARPi adaptive resistant clones (MDA-MB- the development of Notch inhibitors, including AL101 a pan-Notch, gamma
468, MDA-MB-231, HCC1806) were generated in triple-negative breast secretase inhibitor (J Clin Oncol 36, 2018 abstract 2515). AL101 is cur-
cancer cell lines. Through morphologic observation and functional analysis, rently being evaluated in Adenoid Cystic Carcinoma patients with activating
we evaluated epithelial-mesenchymal transition (EMT) and changes in Notch mutations (NCT03691207, ACCURACY trial). Here, we aim to test
immune checkpoint programmed death-ligand 1 (PD-L1). We also down- the activity of AL101 in TNBC patient derived xenograft (PDX) models with
regulated the expression of PD-L1 by shRNA to study the role of PD-L1 in Notch activating genetic alterations. Methods: Gene expression cluster
PARPi resistance. We evaluated the immunology sensitivity to cytotoxic analysis was performed for 38 TNBC PDX tumors using a list of 21 Notch
T cell upon PD-L1 change using a murine ex-vivo CD8+ T cell killing assay target genes. Seven tumors, bearing a “Notch-on” signature, were enriched
and a comparison of total killing cells percentage per well. Results: We with mutated/fusion (M/F) Notch genes and clustered separately from all other
demonstrated that inhibition of PARP enhances EMT, which induces tumors. Of 9 models selected for study, 4 had a Notch-on signature and were
phosphorylation of Akt at S473. This in turn upregulates the expression of expected to respond to AL101. Tumors were implanted into female athymic nude
PD-L1 by 2-3 fold in triple-negative breast cancer cells. In addition, mice. Once tumors reached an average size of 150-300 mm3, mice (n = 5/group)
PARPi–induced EMT occurred independent of PD-L1 upregulation in triple- were randomized to Vehicle or AL101 treatment arms (3 mg/kg, PO, 4on/3off)
negative breast cancer cells. Metformin administration (10mM) was found to until tumors reached 1500 mm3 or day 60. Results: As measured by tumor
reverse EMT by blocking the p-Akt S473 axis through activation of AMPK, growth inhibition (TGI), AL101 was more potent in tumors with a putative Notch-
resulting in downregulation of PD-L1 expression and sensitizing PARPi- on signature. Within these 4 models, M/F genes were present in Notch1-NRR
resistant cancer cells to T cell killing. Conclusions: In summary, we iden- GOF (103% TGI p = 0.0004); Notch2-fusion (62%TGI p = 0.036); Notch3-
tified that induction of EMT is a new mechanism for PARP inhibitor re- fusion (75% TGI p = 0.032); or Notch4-fusion (147% TGI p , 0.00001).
sistance. Metformin was able to reverse EMT and therefore a combination of Tumors lacking the Notch signature did not respond significantly to AL101: WT
metformin and PARP inhibitors may be a promising therapeutic strategy to Notch (43% TGI p = 0.0104; 64% TGI p = 0.13); Notch1 with a predicted loss of
increase the efficacy of PARP inhibitors and tumor sensitivity to T cells. function mutation (12% TGI p = 0.53), Notch1 Variant of Unknown Significance
(VUS) (30% TGI p = 0.44), Notch2 VUS (41% TGI p = 0.44). Conclusions: We
demonstrate that in TNBC PDX models, the presence of a Notch-on signature and
Notch GOF mutations/fusions correlates with potent response to AL101. These
data support the clinical development of AL101 as a targeted therapy for TNBC
with Notch GOF alterations.

1065 Poster Session (Board #146), Sun, 8:00 AM-11:00 AM 1066 Poster Session (Board #147), Sun, 8:00 AM-11:00 AM
Association of SOX9 expression with sensitivity to CDK7 inhibition and Efficacy and safety of anti-PD-1 antibody SHR-1210 combined with apatinib
overall survival in triple-negative breast cancer. First Author: Xiaoxiang in patients with advanced triple-negative breast cancer. First Author: Jieqiong
Guan, Jiangsu Province Hospital, The First Affiliated Hospital of Nanjing Liu, Guangdong Provincial Key Laboratory of Malignant Tumor Epigenetics and
Medical University, Nanjing, China Gene Regulation, Department of General Surgery, Sun Yat-Sen Memorial
Hospital, Sun Yat-Sen University, Guangzhou, China
Background: Triple-negative breast cancer (TNBC) has been shown to dis-
play cyclin-dependent kinase 7 (CDK7)-dependent transcriptional addition Background: PD-1/PD-L1 blockade monotherapy delivered positive outcomes
and is sensitive to a CDK7 inhibitor THZ1. As tumor heterogeneity is well in early-phase trials in advanced triple-negative breast cancer (TNBC). However,
known, we sought to identify predictive markers of sensitivity to CDK7 in- the highest objective response rate (ORR) is 18.5%. IMPassion130 trial has
hibition in TNBC. Methods: Published mRNA microarray data was analyzed demonstrated that only PD-L1+ TNBC had overall survival benefit from anti-PD-
to identify genes associated with sensitivity of TNBC cells to the CDK7 L1 antibody combined with chemotherapy. Preclinical studies found that
inhibitor THZ1. Cell survival assay after THZ1 treatment was performed antiangiogenic therapy could sensitize anti-PD-1 treatment via inducing PD-L1
using Cell Counting Kit-8. Expression of SOX9 was examined by Western blot expression and increasing CTLs infiltration in tumor microenvironment. Thus, we
in cell lines and IHC in breast cancer tissue microarray. Patient survival designed a phase II, open-label trial (NCT03394287) of SHR-1210 (anti-PD-1
analysis was done using the Kaplan-Meier method. Results: Analysis of antibody) in combination with apatinib (VEGFR2 inhibitor) in patients with
published mRNA microarray data showed that SOX9 was overexpressed in advanced TNBC. Methods: 20-58 advanced TNBC patients whose systemic
the MDA-468 and BT-549 TNBC cell lines, which were sensitive to the CDK7 therapy lines in the metastatic setting was less than 3, will be randomly (1:1)
inhibitor THZ1 and showed decreased SOX9 expression with THZ1 treat- enrolled from Sun Yat-sen Memorial Hospital to receive either SHR-1210
ment. We treated 10 TNBC cell lines with various concentrations of THZ1 200mg Q2W plus apatinib 250mg, continuous dosing (d1-d14), or SHR-
and found that these cell lines displayed a wide spectrum of sensitivity to 1210 200mg Q2W plus apatinib 250mg, intermittent dosing (d1-d7), until
THZ1, with an IC50 ranging from 20-30 nM in sensitive lines (MDA-468 and progression or unacceptable toxicities. Primary endpoint was ORR. Secondary
MDA-231) to ~300 nM in insensitive lines (BT-20 and MDA-157). There end points included PFS, DCR, TTR, DoR, CBR, one year-OS and toxicity.
was a strong correlation of SOX9 expression by Western blot and sensitivity to Results: Until Jan 30, 2019, 34 patients were enrolled, 10 in the intermittent
THZ1 in these cell lines (Pearson correlation, r = 0.82, P= 0.01). IHC of two dosing arm, and 24 in the continuous dosing arm. 26 (76.5%) patients had prior
tissue microarrays containing 278 breast cancer samples showed that SOX9 systemic therapy in the metastatic setting. At the cutoff date (Jan 30, 2019),
expression was significantly higher in TNBC (n = 59) compared to hormone 28 patients were evaluable for ORR as per RECIST. The ORR was 47.4% (9 of
positive breast cancer (n = 173; P, 0.01). Among the 59 TNBC cases, 19) in the continuous dosing cohort, and no confirmed objective response was
43 had high SOX9 expression and 16 had low SOX9 expression. Patients with found in the intermittent dosing arm. The DCR was 68.4% in the continuous
high or low SOX expression had similar grade, T stage and nodal involvement dosing arm, whereas it was 44.4% in the other arm. The median PFS was
(All P. 0.20). Finally, with standard of care treatment, TNBC patients with 2 months in the intermittent dosing cohort, while it was not reached in the
high SOX9 expression had a significantly worse overall survival (OS) compared continuous dosing cohort. The most common adverse events (AEs) was fatigue
to patients with low SOX9 expression (2 year OS rate 100% vs 92.9%, 5 year (65.0%), hand-foot syndrome (63.3%), and elevated aspartate aminotrans-
OS rate 93.8% vs 83.3%, log rank P= 0.03). Conclusions: SOX9 expression ferase and/or alanine aminotransferase (73.3%). There were no treatment-
level in TNBC cell lines was associated with sensitivity to the CDK7 inhibitor related deaths. Another 5 patients will be enrolled into the continuous dosing
THZ1. TNBC patients with high expression of SOX9 had worse OS, and may arm because of its favorable response. Conclusions: Anti-PD-1 antibody SHR-
benefit from CDK7 inhibition as a novel therapy. 1210 combined with apatinib demonstrates favorable clinical activity and a
manageable safety profile in patients with advanced TNBC. Clinical trial in-
formation: NCT03394287.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Breast Cancer—Metastatic 51s

1067 Poster Session (Board #148), Sun, 8:00 AM-11:00 AM 1068 Poster Session (Board #149), Sun, 8:00 AM-11:00 AM
Patient-reported outcomes (PROs) from the phase III IMpassion130 trial of IMpassion130: Expanded safety analysis from a P3 study of atezolizumab (A) +
atezolizumab (atezo) plus nabpaclitaxel (nP) in metastatic triple-negative nab-paclitaxel (nP) in patients (pts) with treatment (tx)-naı̈ve, locally advanced
breast cancer (mTNBC). First Author: Sylvia Adams, New York University or metastatic triple-negative breast cancer (mTNBC). First Author: Andreas
Cancer Institute, New York, NY Schneeweiss, National Center for Tumor Diseases, Heidelberg University
Hospital and German Cancer Research Center, Heidelberg, Germany
Background: In the IMpassion130 study in 1L mTNBC (N = 902), PFS with
atezo + nP was significantly better than with placebo (P) + nP in ITT (HR, Background: IMpassion130 showed PFS benefit with A + nP vs placebo (P) + nP as
0.80) and PD-L1 IC+ (HR, 0.62) patients (pts). Clinically meaningful OS 1L tx for mTNBC in the ITT and PD-L1 IC+ pts. We report expanded safety data with
improvement (HR, 0.62) was also seen in PD-L1+ pts. PROs were used to 4.5-mo longer follow-up (FU), focusing on adverse events of special interest (AESI),
document pt perspectives on overall clinical benefit of atezo + nP. potentially immune related. Methods: Pts with unresectable locally advanced or
Methods: Pts received either atezo 840 mg or P q2w + nP 100 mg/m2 on mTNBC received nP 100 mg/m2 IV (d1, 8 and 15 of a 28-d cycle) + A 840 mg IV
days 1, 8 and 15 of each 28-day cycle until disease progression or in- q2w or P until PD or toxicity. Safety was a secondary endpoint. Results: With 15.6
tolerance. Pts completed the EORTC QLC-C30 and breast cancer module mo of median FU, of 453 pts with A+nP and 437 pts with P+nP, 49% and 43% had
(QLQ-BR23) on day 1 of each cycle, at end of treatment (Tx) and q4w during Gr 3/4, 1% and ,1% had Gr 5, 23% and 19% had serious AEs, and 58% and 42%
had AESI, respectively. Most AESI ($86%; either arm) were Gr 1/2. 14% (A+nP) vs
follow-up for 1 y. Time to deterioration (TTD; first $ 10-point decrease from
6% (P+nP) received systemic corticosteroids within 30 d of AESI onset. The only
baseline [BL] held for 2 cycles) in HRQoL was a pre-defined secondary
any-Gr AESI differing with A+nP vs P+nP were rash (34% vs 26%), hypo- (18% vs
endpoint. Exploratory endpoints included TTD in function, and mean and 5%) and hyperthyroidism (5% vs 1%) and pneumonitis (4% vs ,1%). The leading
mean change from BL scores (changes $ 10 considered clinically mean- cause of withdrawal was peripheral neuropathy, with Gr 3 affecting 6% (A+nP) vs 3%
ingful) in HRQoL, function and disease/Tx-related symptoms. Results: BL (P+nP). AESI median time to onset (TTO) was consistent with A monotherapy trials.
completion was 92% (QLQ-C30) and 89% (QLQ-BR23) and remained Conclusions: A+nP had a tolerable safety profile, with no meaningful changes since
. 80% through Cycle 20 in both ITT and PD-L1 IC+ pts. No differences in the primary data cut. No cumulative toxicities or new or late-onset safety signals were
median TTD in HRQoL (ITT: HR, 0.97 [95% CI: 0.80, 1.18]; PD-L1 IC+: HR, seen with longer FU. Clinical trial information: NCT02425891.
0.94 [95% CI: 0.69, 1.28]), physical function (ITT: HR, 1.04 [95% CI: Clinically Relevant AESI Summary.
0.86, 1.26]; PD-L1 IC+: HR, 1.02 [95% CI: 0.76, 1.37]) or role function P+nP A+nP
(ITT: HR, 1.01 [95% CI: 0.83, 1.22]; PD-L1 IC+: HR, 0.77 [95% CI: 0.57, (n = 437) (n = 453)

1.04]) were observed between arms in either population. Mean scores at BL Resolved
AESI,
First AESI
Any-Gr AESI
AESI, n (%) AESI, n (%) (any Gr)
for HRQoL (ITT: 66.0 [atezo + nP] vs 64.3 [P + nP]; PD-L1 IC+: 67.5 vs n/n (%) Leading to A
Any Gr Any Gr Any Median TTO, Median Duration, Withdrawal,
65.0), physical function (ITT: 80.4 vs 79.2; PD-L1 IC+: 82.8 vs 79.4) and Medical Concepta Gr 3/4 Gr 3/4 Gr mo (range) mo (range) n (%)
role function (ITT: 72.7 vs 71.0; PD-L1 IC+: 73.7 vs 71.7) were similar Hypothyroidism 20 0 80 0 38/80 3.8 13.0 0
between arms and throughout the course of Tx. In both arms, HRQoL, Hyperthyroidism
(5)
6 0
(18)
21 1
(48)
16/21
(0.5-24.4)
3.8
(0.1-26.9*)
1.9 0
physical and role function, and Tx symptoms (fatigue, diarrhea, nausea, (1) (5) (,1) (76) (1.1-18.4) (0.2-14.7*)
Pneumonitis 1 0 16 2 12/16 4.7 3.2 1
vomiting) were stable during Tx, with no clinically meaningful changes seen (,1) (4) (,1) (75) (0.9-19.9) (0.5-13.1*) (,1)
until pts discontinued Tx. Conclusions: PRO data suggest that atezo + nP Rash 115
(26)
2
(,1)
155
(34)
4
(,1)
126/155
(81)
1.3
(0.0-25.3)
1.0
(0.0-27.3*)
1
(,1)
was tolerable and similar to nP alone in maintaining HRQoL and day-to-day Hepatitis 7 1 11 7 6/11 4.6 2.0 2
(2) (,1) (2) (2) (55) (0.9-28.3) (0.3-12.0*) (,1)
function relative to BL. This confirms atezo + nP had clinical benefit without Colitis 3 1 5 1 4/5 6.7 0.8 1
compromising HRQoL, physical and role function, or worsening Tx symptoms Adrenal insufficiency
(1)
0
(,1)
0
(1)
4
(,1)
1
(80)
3/4
(3.0-15.6)
4.9
(0.6-12.8*)
5.5
(,1)
1
vs P + nP in 1L mTNBC. Clinical trial information: NCT02425891. (,1) (,1) (75) (4.6-12.9) (1.1-13.6*) (,1)
a
*Censored. Grouped MedDRA preferred terms.

1069 Poster Session (Board #150), Sun, 8:00 AM-11:00 AM 1070 Poster Session (Board #151), Sun, 8:00 AM-11:00 AM
A phase II clinical trial of pembrolizumab and selective androgen receptor Randomized, optimal dose-finding, phase II study of tri-weekly nab-paclitaxel
modulator GTx-024 in patients with advanced androgen receptor-positive in patients with metastatic breast cancer (ABROAD). First Author: Fumikata
triple-negative breast cancer. First Author: Jin Sun Lee-Bitar, City of Hope Hara, Cancer Institute Hospital of JFCR, Koto, Tokyo, Japan
National Medical Center, Duarte, CA
Background: Although nab-paclitaxel (nab-PTX) has shown superior efficacy
Background: Androgen receptor (AR) targeting therapy has shown single compared to conventional paclitaxel in metastatic breast cancer (MBC),
agent activity in triple negative breast cancer (TNBC). GTx-024, a non- chemotherapy induced peripheral neuropathy (CIPN) was more frequently
steroidal selective androgen receptor modulator (SARM), demonstrated observed in nab-PTX. In a single arm Phase 2 trail (CA002-0LD), low dose
preclinical and clinical activity in AR+ breast cancer. The current study is nab-PTX (175mg/m2) every 3 weeks (q3w) demonstrated a good objective
designed to test the safety and efficacy of GTx-024 and pembrolizumab in response rate (39.5%) without grade 3 or higher CIPN. Herein, we con-
patients with AR+ metastatic TNBC (mTNBC). Methods: This is an open- ducted multicenter randomized controlled study to evaluate optimal dose
label phase 2 study for AR+ mTNBC. Eligible participants receive pem- of nab-PTX comparing lower dose (LD or MD) to standard dose (SD).
brolizumab 200mg IV every 3 weeks in combination with GTx-024 18mg po Methods: This study compared three different doses of q3w nab-PTX (SD:
daily. Key eligibility criteria include patients with AR+ ( . 10%, 1+ by IHC); 260 mg/m2 vs. MD: 220 mg/m2 vs. LD: 180 mg/m2) in patients with HER2
mTNBC; ECOG 0-1; measurable disease per RECIST 1.1. Patients are ex- negative metastatic breast cancer. Primary endpoint was progression-free
cluded if they had prior checkpoint inhibitors or AR targeted agents. The survival (PFS). Grade 3/4 neuropathy rates in the three doses are estimated
primary objective is to evaluate the tolerability of GTx-024 and pem- by the logistic regression. Optimal dose was selected by 2 step selection. At
brolizumab, and determine the response rate. Results: Seventeen patients first, if hazard ratio (HR) for PFS was less than 0.75 or more than 1.33, the
were enrolled in the study. One patient was ineligible due to previously inferior dose was dropped. Then, if estimated incidence rate of grade 3/4
undiagnosed brain metastases. Ten of 16 patients had visceral metastasis neurotoxicity exceed10%, that dose was also dropped. This trial is registered
(lung or liver), and 15% of patients had received $ 3 previous lines of therapy with the University Hospital Medical Information Network (UMIN), Japan
for mTNBC. Among 16 patients evaluable for response, 2 patients achieved a (protocol ID C000012429). Results: In this study, 141 patients were
best response of partial response (PR), 2 patient had stable disease (SD, 18 randomly assigned to SD (n = 47), MD (n = 46) or LD (n = 48). Median PFS
and 19 weeks ), 11 patients had progressive disease (PD), and 1 patient is was 6.66 vs 7.34 vs 6.82 months, respectively. HR was 0.73 (95% con-
too early for restaging imaging. Durable response was found in 1 patient. fidence interval (CI): 0.42-1.28) in MD vs SD. SD was dropped due to in-
Grade 3 toxicities include 1 diarrhea and 1 dry skin. Grade 2 adverse events feriority to MD. HR was 0.77 (95%CI 0.47-1.28) in LD vs SD, and 0.96
include 3 elevated liver function, 1 adrenal insufficiency, 1 hyperthyroidism, 1 (95%CI 0.56-1.66) in LD vs MD. LD and MD were carried over to next step
palpitation, 1 diarrhea, 1 hyperhydrosis, 1 hot flashes and 1 headache. Three due to equivalence. Overall survival was not different among all dose arms.
patients had dose delay and two patients had dose reduction. Conclusions: AR Rate of dose reduction by treatment course was significantly higher in SD
targeted therapy GTx-024 combined with pembrolizumab is well tolerated with arm. Estimated incidence of grade 3/4 neurotoxicity rate was 29.5% in SD,
clinical activity. Clinical trial information: NCT02971761. 14.0% in MD and 5.9% in LD. Final selected dose was LD 180mg/m2. HR-
QOL results will be presented. Conclusions: Low dose nab-PTX at 180 mg/
m2/3 weeks could be an optimal dose with good clinical efficacy and tol-
erability for patients with MBC. Clinical trial information: C000012429.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
52s Breast Cancer—Metastatic

1071 Poster Session (Board #152), Sun, 8:00 AM-11:00 AM 1072 Poster Session (Board #153), Sun, 8:00 AM-11:00 AM
Outcomes of talazoparib (TALA) versus physician’s choice of chemotherapy A phase Ib trial of the cyclin-dependent kinase inhibitor dinaciclib (dina) in
(PCT) in patients (pts) with advanced breast cancer (ABC) and a germline combination with pembrolizumab (P) in patients with advanced triple-
BRCA (gBRCA) mutation by line of chemotherapy (CT) in the EMBRACA trial. negative breast cancer (TNBC). First Author: Amy Jo Chien, University of
First Author: Johannes Ettl, Klinikum Rechts der Isar, Technische Universität California San Francisco Helen Diller Family Comprehensive Cancer Center,
München, Munich, Germany San Francisco, CA
Background: The PARP inhibitor TALA was approved in the US for treatment of Background: Increased expression of the MYC transcriptional oncogene is
gBRCA-mutated ABC based in part on the EMBRACA study. Understanding found in 70% of TNBC and is associated with poor prognosis. In MYC-
the outcomes of EMBRACA pts relative to prior CT is a current unmet need. overexpressing TN xenografts, CDK1 inhibition with dina results in synthetic
Methods: EMBRACA was a randomized Phase 3 trial comparing TALA 1 mg lethality, and attenuates distant metastasis. In syngeneic models, the
daily vs PCT (capecitabine, eribulin, gemcitabine, vinorelbine) in gBRCA- combination of dina with anti-PD1 therapy is synergistic and increases
mutated ABC. Clinical outcomes were assessed by line of prior CT for ABC in immune cell tumor infiltration and activation. Methods: Patients (pts) with
intent-to-treat (ITT), triple-negative breast cancer (TNBC), and hormone advanced TNBC were enrolled. Pts received intravenous dina day 1 and 8 in
receptor-positive (HR+) breast cancer cohorts. Results: 431 pts were ran- combination with fixed dose P 200 mg once every 21 days. Dina was dose
domized (ITT; TALA 287; PCT: 144). TALA was generally more effective than escalated using a toxicity probability interval design targeting a dose limiting
PCT across efficacy endpoints regardless of line of CT (Table). For the ITT toxicity (DLT) rate of 25%. Evaluable disease and pre-treatment metastatic
population, TALA improved progression-free survival (PFS) and objective response biopsies were required. After 17 pts were enrolled, eligibility was amended to
rate (ORR) vs PCT for each line of CT assessed. Other prespecified subgroups require #2 lines of prior chemotherapy, and LDH #1.5x normal. Results: 22
(TNBC and HR+) will be presented. Conclusions: In pts with gBRCA-mutated pts were enrolled (median age 50, median 2 prior lines of therapy, 10 pts
ABC, TALA demonstrated improvements in clinical outcomes compared with PCT (45%) had disease which was previously ER+). Dina was dose escalated from
regardless of prior lines of CT. Clinical trial information: NCT01945775. 12 to 33 mg/m2 in 4 cohorts. No DLTs were observed. Of the first 5 patients
in the 33 mg/m2 cohort, 2 pts required a dose reduction (1 pt due to re-
TALA vs PCT 0L CT 1L CT ‡2L CT
current grade (G) 3 fever and G2 chills; 1 pt due to recurrent G2 diarrhea,
ITT
N 111 vs 54 107 vs 54 69 vs 36
chills, fatigue), and 1 pt discontinued due to G3 hemolytic anemia. 4 ad-
PFS, median, mo 9.8 vs 8.7 8.1 vs 4.6 5.8 vs 4.2 ditional pts were treated at 33 mg/m2 without DLT or dose reduction. G $3
Hazard ratio (HR) (95% CI) 0.57 (0.34-0.95) 0.51 (0.33-0.80) 0.56 (0.34-0.95) adverse events (AEs) in all pts included neutropenia (ntp) (36.3%), febrile
ORR, N (%) 66 (80) vs 15 (37) 45 (57) vs 8 (20) 26 (46) vs 8 (24)
Odds ratio (95% CI) 6.9 (2.7-16.8) 5.1 (2.0-14.2) 2.7 (0.9-7.8) ntp (13.6%), and fatigue (13.6%). Most common all-grade AEs included
TNBC fatigue (73%), diarrhea (59%), nausea (50%), ntp (45%), mucositis (32%),
N 52 vs 26 50 vs 21 28 vs 13 anorexia (23%). Immune-related AEs included sinusitis (3 pts), hemolytic
PFS, median, mo 7.3 vs 5.5 5.4 vs 3.5 4.3 vs 1.5
HR (95% CI) 0.67 (0.35-1.27) 0.58 (0.29-1.12) 0.46 (0.21-1.03) anemia (1 pt), pneumonitis (1pt), and rash (2 pts). Of 21 pts evaluable for
ORR, N (%) 31 (78) vs 4 (20) 21 (55) vs 2 (13) 11 (46) vs 0 response, 1 pt had complete response (4.8%), 2 pts had partial response
Odds ratio (95% CI) 12.1 (2.8-56.8) 6.5 (1.3-70.9) Not estimable (NE) (9.5%), and 5 pts had stable disease (23.8%). Dose expansion and cor-
(2.2-NE)
HR+ relative biomarker studies are ongoing. Conclusions: The recommended
N 59 vs 28 57 vs 33 41 vs 23 phase 2 dose of dinaciclib given in combination with pembrolizumab is 33
PFS, median, mo 12.2 vs 8.9 9.0 vs 5.9 7.6 vs 5.6
HR (95% CI) 0.41 (0.17-0.97) 0.43 (0.22-0.81) 0.60 (0.3-1.2) mg/m2 on D1, 8 of 21-day cycle. Toxicities are generally manageable with
ORR, N (%) 35 (81) vs 11 (52) 24 (59) vs 6 (24) 15 (46) vs 8 (40) dose reduction and dose delay. Clinical trial information: NCT01676753.
Odds ratio (95% CI) 3.9 (1.0-14.3) 3.8 (1.1-13.6) 1.3 (0.35-4.7)

1073 Poster Session (Board #154), Sun, 8:00 AM-11:00 AM 1074 Poster Session (Board #155), Sun, 8:00 AM-11:00 AM
Effector T-cell cytolytic activity modules derived from CD3+ single cells from Brain metastasis in patients with hereditary BRCA-mutated invasive breast
human primary triple-negative breast cancer (TNBC) in multiple solid tumors cancer. First Author: Haven Garber, MDACC, Houston, TX
to predict response to immune checkpoint blockade therapy (ICB). First
Background: In the past six months, two poly(ADP-ribose) polymerase
Author: Chaitanya Ramanuj Acharya, Duke University Medical Center,
(PARP) inhibitors have been approved for the treatment of patients with
Durham, NC
metastatic breast cancer and germline pathogenic variants in BRCA1 or
Background: The prognostic and predictive value of tumor infiltrating lym- BRCA2 (gBRCA). In addition, recent data from the IMpassion 130 trial lends
phocytes for ICB has been recognized in a variety of tumor types, including support for the role of immunotherapy in a subset of patients with triple
TNBC. Nonetheless, our understanding of the mechanistic aspects of T cell negative breast cancer (TNBC). Approximately 60% of patients with gBRCA1
activation remains incomplete. We hypothesize that a specific effector phe- have TNBC. There is evidence that both PARP inhibitors and checkpoint
notype of T cell cytolytic activity (ECA) is a consistent feature of epithelial inhibitors cross the blood-brain barrier and both classes of agents have
tumors, possibly varying by tumor types with a range of inflammatory features. entered clinical trials for patients with brain metastases in other tumor types.
Methods: We evaluated 6,311 purified CD3+ single cells from human primary We studied the clinical course of breast cancer patients with gBRCA and
TNBC and computed sample set enrichment scores of a set of previously brain metastasis at our institution to inform clinical trial design for this group
published immune metagenes. Following unsupervised clustering of the en- of patients with poor outcomes. Methods: Patients with stage I to III invasive
richment scores of the entire single cell population, two subgroups of cells with breast cancer, gBRCA, and eventual development of brain metastasis were
highest and lowest average enrichment score of T cell cytolytic activity formed identified from clinical databases. Data analyzed included breast cancer
a basis for detecting functional gene expression modules. Spectral de- subtype and stage at diagnosis, treatment, time to distant recurrence and to
composition and Jackstraw analysis estimated eight modules with overlapping discovery of brain metastasis, and overall survival from time of development
sets of genes. Each gene expression module was then used to train a Random of brain metastasis. Results: Patients in our cohort (n = 24, to date) were
Forest classifier of ECA phenotype. Results: We discovered that our module- diagnosed at a young age (median age 39) and primarily had TNBC (21/24,
derived classifiers were prognostic not only in TNBC samples obtained from both 87.5%) with infiltrating ductal carcinoma histology. Nineteen patients
TCGA (N = 150) and METABRIC (N = 320) datasets but also in 14 other tumor had gBRCA1 and 4 patients had gBRCA2. All but 1 patient received
types encompassing 6,000 samples. For example, patient samples from TCGA anthracycline-based chemotherapy in the neoadjuvant/adjuvant setting.
dataset predicted to be in group ECA ‘High’ have better progression-free Median time to distant metastasis was 2 years (range: 0.8 – 15) and the brain
survival (p-value: 0.0098l; HR: 0.30) and better overall survival (p-value: was the first site of recurrence in 5 of 24 (21%) patients. Median time from
0.0066; HR: 0.17). In both breast datasets, gene within the classifier are diagnosis to development of brain metastasis was 2.6 years (range: 1.2 – 19)
relatively under-expressed in ER+ tumors as opposed to HER2+ and TNBC and most patients (18/25, 72%) had multiple brain metastases discovered
(p-value , 2.2e-16). In a dataset of normal, pure DCIS and mixed DCIS on the initial brain MRI. Median overall survival (OS) was 3.7 years (range:
(GSE26304;N = 114), the same genes were relatively under-expressed in 1.8 – 24) and median OS from the time of brain metastasis was 7 months
DCIS samples relative to invasive tumors (p-value , 2.2e-16). Additionally, (range: 1 month – 13 years). Conclusions: Breast cancer patients with
in a pre-therapy tumor dataset of fifty-one advanced melanoma patients germline pathogenic variants in BRCA1/2 who develop brain metastasis
treated with Nivolumab, who previously either progressed on ipilimumab or have a dismal prognosis. These patients may benefit from an agent with
were ipilimumab-naı̈ve, our module-derived classifier was able to classify intracranial activity at the time of first distant recurrence.
responders and non-responders with 77% accuracy (p-value = 0.02) and was
associated with progression-free survival (p-value = 0.03; HR: 0.28).
Conclusions: Our study highlights one important application of single-cell
genomics in our understanding of immune microenvironment and poten-
tially identify new immunotherapy targets.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Breast Cancer—Metastatic 53s

1075 Poster Session (Board #156), Sun, 8:00 AM-11:00 AM 1076 Poster Session (Board #157), Sun, 8:00 AM-11:00 AM
Determinants of concordance in clinically relevant genes (CRG) from synchronously A phase II, single-arm study of apatinib and oral etoposide in pretreated
acquired tumor biopsies (tBx) and ctDNA in metastatic breast cancer (MBC). metastatic HER2-negative breast cancer. First Author: Nanlin Hu, Chinese
First Author: Mafalda Oliveira, Medical Oncology Department, Breast Cancer Academy of Medical Sciences, Cancer Hospital, Beijing, China
Group, Vall d’Hebron University Hospital, Vall d’Hebron Institute of Oncology
Background: There is no standard treatment strategy for patients with locally
(VHIO), Barcelona, Spain
advanced or metastatic breast cancer suffering progression after one prior
Background: NGS in ctDNA from MBC is feasible and results may be informative chemotherapy with metastasis setting. Apatinib is a tyrosine kinase inhibitor
for patients’ management, especially in non-luminal tumors (Oliveira et al, ASCO targeting vascular endothelial growth factor receptor 2(VEGFR-2). Etoposide
2018). We aimed to study the determinants of concordance in CRG in a cohort of is a highly active chemo-drug in the treatment of advanced breast cancer,
60 MBC patients undergoing tBx and ctDNA collection. Methods: MiSeq both as a single agent or in combination regimens, and is well tolerated,
Amplicon-based NGS (59 cancer-related genes) was performed in one single with a low incidence of severe toxicity. This study is performed to assessed
metastatic lesion per patient and compared with liquid biopsies taken at the same the efficacy and safety of apatinib and oral etoposide in patients with HER2
time point at disease progression to prior treatment. The concordance in CRG negative locally advanced or metastatic breast cancer for whom at least one
(PIK3CA, AKT1, ERBB2, ESR1, PTEN, BRAF, FGFR1, HRAS, KRAS, and lines of prior chemotherapy had failed. Methods: This open-label, single arm
PIK3R1) in tBx vs ctDNA was determined at patient and at mutation (mut) level study enrolled patients with HER2-negative breast cancer, pretreated with
and correlated with mutant allele fraction (MAF), total disease volume (TDV), and anthracycline, taxanes, and who failed in the metastatic setting at least one
clinical characteristics. True positive in plasma (TPP): patient with a mut detected prior chemotherapy regimens and at least one endocrine drug for hormone
both in ctDNA and tBx. TDV was defined as all metastasis volume assessed by CT
receptor-positive patients. Apatinib was administered 425/500mg daily
scan (excluding sclerotic bone metastasis), and analyzed by an experienced radi-
according to patients ECOG(Eastern Cooperative Oncology Group) status,
ologist using the 3DSlicer semiautomatic segmentation tool (TDV = pixel size x
oral etoposide was administered 50mg/m2 for first 10 days in a 21-days
number of pixels). Results: Concordance in CRG at patient and mut level was 72%
and 55%, respectively. Concordance for ERBB2 (1/1; 100%) and PIK3CA (17/22;
cycle. The primary end point of this study was progression free survival (PFS).
77%) was higher than for ESR1 (8/20; 40%) and AKT1 (2/6; 33%). ctDNA failed to Secondary end points included objective response rate (ORR), disease
detect 14 mut present in tBx (ESR1 n = 5, PIK3CA n = 5, AKT1 n = 3, BRAF n = 1). control rate (DCR), overall survival (OS), and toxicity. The treatment duration
Concordance was 100% for non-luminal and 60% for luminal cases (P = 0.01). In is until disease progression or intolerability of apatinib or oral etoposide.
univariate analysis, concordance was not associated with MAF in tBx (P = 0.15), Results: 20 eligible patients were enrolled in this, open-label, single arm
TDV (p = 0.86), number of prior lines of therapy (P = 0.57), number of metastatic study and received apatinib and oral etoposide with a median age of 54 years
sites (P = 0.56) or presence of visceral metastasis (P = 1.0). In patients with old(range 36 to 66 years). Median follow-up time was 11months. 20 patients
PIK3CA mut (N = 22), those with TPP had a numerically higher TDV than those were eligible for efficacy analysis. Median PFS was 5.6 months (95%
where a PIK3CA mut was not detected in ctDNA (20.9cm3 vs 5.1cm3, P = 0.28). confidence interval (CI), 4.01 m – 8.42 m). ORR was 20% (4/20). DCR was
Across all patients, in the multivariate logistic model adjusted for other factors, 70% (14/20). Median OS was 11.2 months (95% CI, 9.6 m – 14.95 m). The
TDV was a determinant of TPP (OR 1.02, 95%CI 1.0-1.06; P = 0.059). For each most common grade 3/4 treatment-related AEs were hypertension (30%),
increase of 1cm3 in TDV, there was a 2% increase in the probability of detecting a and proteinuria (5%), nausea (5%). 35%(7/20) patients had dose reduction
mut in ctDNA. Conclusions: Our results suggest that liquid biopsy testing for the because of adverse events, after that all adverse events can return to less
detection of actionable CRG is clinically valid in MBC, although its yield depends than 2 grade. Conclusions: Apatinib with oral etoposide exhibited objective
on several factors – tumor subtype, analyzed gene, and possibly tumor volume – efficacy in pretreated, metastatic HER2-negative breast cancer with man-
that reflect both tumor heterogeneity and tumor shedding rate. Due to the po- ageable toxicity. Prospective studies enrolling more patients are needed.
tential clinical implications, the observation that mutation detection in ctDNA may Clinical trial information: NCT03535961.
correlate with tumor volume merits further study in a larger dataset.

1077 Poster Session (Board #158), Sun, 8:00 AM-11:00 AM 1078 Poster Session (Board #159), Sun, 8:00 AM-11:00 AM
First results of a prospective registry in unresectable locally advanced or First-line (1L) ribociclib (RIB) plus letrozole (LET) for postmenopausal women
metastatic breast cancer patients: GEICAM/2014-03 (RegistEM). First Author: with hormone receptor-positive (HR+), HER2- advanced breast cancer (ABC):
Carlos Jara, Hospital Universitario Fundación Alcorcón, Medical Oncology MONALEESA-2 long-term safety results. First Author: Joyce O’Shaughnessy,
Department, GEICAM, Spanish Breast Cancer Group, Madrid, Spain Baylor University Medical Center, Texas Oncology, US Oncology, Dallas, TX
Background: In Spain there is limited prospective data for unresectable Background: In the phase III MONALEESA-2 study (NCT01958021), 1L
locally advanced breast cancer (ULABC) or metastatic breast cancer (MBC) treatment (Tx) with RIB (cyclin-dependent kinase 4/6 inhibitor) plus LET
patients (pts) treated as per clinical practice. RegistEM study will provide significantly prolonged PFS vs placebo (PBO) + LET in patients (pts) with HR+/
epidemiological, pathological and clinical data, including treatments given HER2– ABC (HR 0.56; P,0.0001). In the primary publication (data cutoff,
for different disease stages. Understanding the real distribution of the Jan 2016), the median duration of exposure was 13.0 mo (RIB + LET) and
different BC subtypes is the primary objective. Methods: This is a non- 12.4 mo (PBO + LET), and the most common all-grade adverse events (AEs) in
interventional cohort study enrolling approximately 1,400 pts with advanced the RIB + LET arm were neutropenia (74.3%), nausea (51.5%), infections
disease diagnosed from January 2016 to December 2018, either after re- (50.3%), fatigue (36.5%), and diarrhea (35.0%; Hortobagyi GN, et al. N Engl
currence or as first diagnosis, in 38 Spanish sites. Biological samples J Med. 2016;375:1738-48). Here we present long-term safety data from
(primary tumor, metastatic lesions, blood) are currently being collected. In MONALEESA-2. Methods: Pts received either RIB + LET (n=334) or PBO +
this first analysis, we include 489 pts who met study criteria before October LET (n=330). AEs were graded per CTCAE v4.03. Results: As of data cutoff
31, 2017. All data are described in two subgroups: on the most recent tumor (Oct 2018), 79 (24%) pts (RIB + LET) and 39 (12%) pts (PBO + LET) remained
lesion or on the primary breast tumor. Results: At first diagnosis, 67.9%, on Tx. The median duration of exposure was 20.2 mo (range, 0-54) for RIB + LET
31.5% and 0.6% of pts had early BC (EBC), MBC and ULABC, respectively. and 14.1 mo for PBO + LET, and 30% and 17%, respectively, had $ 36 mo of
In the total analysis population, median age at diagnosis of advanced disease exposure. Neutropenia was the most common exposure-adjusted any-grade AE of
was 59.6 years, most of pts were white (98.2%), female (99.4%) and special interest (AESI) in the RIB + LET arm (Table). AEs were the most common
postmenopausal (70%). Family history of BC and ovarian cancer was re- reason for RIB dose reductions (56.6%) and interruptions (73.4%). Discontin-
ported in 5.7% pts. In ~390 pts BC clinical subtypes distribution was lu- uations due to AEs occurred in 19.2% and 4.2% of pts in the RIB + LET and PBO +
minal B(HER2-)-like (~55%), luminal B(HER2+)-like (~16%), luminal A- LET arms, respectively, and the most common AEs leading to discontinuation in
like or triple negative (TN) (~10% each) and HER2 enriched-like (~8%). the RIB + LET arm were increased ALT (4.5%), GI disorders (3.3%), and in-
Median time to recurrence (years) in EBC pts was: luminal A-like 5.8, lu- creased AST (2.7%). Prolonged QT led to dose interruption in 0.9% of patients
minal B(HER2-)-like 5.1, luminal B(HER2+)-like 3.9, HER2 enriched-like (no discontinuations), and the additional follow-up did not reveal any sudden
2.7 and TN 1.7. Bone (59%), visceral (58%) and lymph node (27%) lesions deaths related to prolonged QT interval beyond the 1 identified in the primary
were the most frequent metastatic locations. The two most frequent ther- analysis. Conclusions: AEs occurring with 1L RIB + LET in postmenopausal pts
apies in first line consisted in: endocrine therapy (ET) (47%) and ET+bio- with HR+/HER2– ABC were manageable and the safety profile was comparable
logical therapy (BT) (29%) for luminal A-like; ET (32%) and ET+BT (32%) to that in the primary report. Clinical trial information: NCT01958021.
for luminal B(HER2-)-like; chemotherapy (CT)+ET+BT (43%) and CT+BT RIB + LET PBO + LET
(24%) for luminal B(HER2+)-like; CT+BT (68%) and CT (16%) for HER2 Exposure-adjusted AESI, rate per 100 pt Tx years (n = 334) (n = 330)

enriched-like; CT (59%) and CT+BT (34%) for TN. Conclusions: These first Neutropenia 164.8 4.1
Infections 67.4 60.5
data confirm that luminal B (HER2-)-like subtype is the most predominant in Leukopenia 29.8 4.3
MBC. Respiratory disorders
Hepatobiliary toxicity
27.8
18.8
31.0
11.6
Anemia 14.0 4.6
Renal toxicity 6.9 2.4
Thrombocytopenia 5.9 0.6
QT interval prolonged 4.7 2.6

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
54s Breast Cancer—Metastatic

1079 Poster Session (Board #160), Sun, 8:00 AM-11:00 AM 1080 Poster Session (Board #161), Sun, 8:00 AM-11:00 AM
Tumor subtype and other prognostic factors in breast cancer patients with Phase II clinical trial using anti-CD3 x anti-HER2 bispecific antibody armed
brain metastases: The updated graded prognostic assessment (Breast-GPA). activated T cells (HER2 BATs) for HER2-negative (0-2+) metastatic breast
First Author: Paul W. Sperduto, University of Minnesota Gamma Knife cancer. First Author: Lawrence G. Lum, University of Virginia, Charlottesville,
Center, Minneapolis, MN VA
Background: Brain metastases (BM) are a common and fatal complication of Background: This study presents a phase II cell therapy trial in 32 women with
breast cancer but survival varies widely based on various prognostic factors metastatic or locally advanced HER2- breast cancer (BrCa) who received in-
(PF). Hence, patient counseling and therapeutic decisions should be in- fusions of anti-CD3 x anti-HER2 bispecific antibody armed activated T cells
dividualized. We previously published a prognostic index (Breast GPA) based (BATs). This phase II study (NCT 01022138) was conducted to determine if
on cohort A (1985-2007, n = 642), updated it with tumor subtype in cohort BATs infusions could improve time to progression (TTP) and overall survival (OS),
B (1993-2010, n = 400) and are now updating it with a larger contemporary as well as to confirm the toxicity profile of BATs. Methods: The phase II included
cohort (C). Methods: A multi-institutional (19) multi-national (3) retro- 32 patients with a median of 4 lines of therapy (7 TNBC and 25 HR/PR+ HER2-
spective database of 2473 breast cancer patients with BM diagnosed from patients) with an average age of 52.5 years (range 28-75 years). Twenty-one
1/1/2006-12/31/2017 was created and compared to our prior cohorts. patients had $3 lines of prior therapy and 11 patients had 1-2 lines. Peripheral
Demographic, clinical, molecular factors, tumor subtype and treatment were blood mononuclear cells (PBMC) were stimulated with anti-CD3 antibody and
correlated with survival. Kaplan-Meier survival estimates were calculated expanded in IL-2, armed with HER2Bi, and aliquoted for the clinical trial.
and compared with log-rank tests. Results: The median survival (MS) for Patients received oncologist’s choice of chemotherapy (4 cycles/4 months)
cohorts A, B and C improved over time [12, 14 and 16 mo, respectively followed by 3 infusions of BATs given once per week for 3 weeks and a boost
( , 0.01)] despite the subtype distribution becoming less favorable: Luminal given 12 weeks after the 3rd infusion. Results: Fifteen of 32 (ORR of 46.8%)
B (ER/PR/HER2+) decreased from 26% to 21%; HER2 (HER2+/ER/PR-) who had received any cells had stable disease (SD) at 1 month after the last
decreased from 31% to 17%, Luminal A (ER/PR+/HER2-) increased from infusion, and 8 of 15 (25%) had SD . 4 months. For patients who completed 3
20% to 31%; Basal (ER/PR/HER2-) was unchanged at 24%.MS by subtype or 4 infusions (17-83 x 109 BATs), 8 of 31 patients had TTP . 4 months. One
improved from 21 to 27 mo in Luminal B, 18 to 25 mo in HER2, 10 to 14 mo patient completed 2 infusions (17 x 109 BATs). There were no dose limiting
in Luminal A and 6 to 9 mo in Basal tumors. The number of BM was 1 in toxicities (DLTs). Tumor markers decreased in 13 of 23 (56.5%) patients with
35%, #4 in 67% and . 10 in 18%. PF significant for survival were tumor evaluable markers. The median OS was 13.8, 16.5, and 12.4 months for all, ER/
subtype, age, KPS, number of BM and extracranial metastases (ECM) (all PR+, and TNBC, respectively. OS for all patients with chemosensitive (chemoS)
, 0.01). Surprisingly, Hispanic women (7%) showed improved survival and chemoresistant (chemoR) disease was 14.6 and 8.6 months (NS), re-
(p , 0.01). BRCA1 was mutated in 57/533 (11%) and those patients spectively. OS for chemoS and chemoR disease in HER2- ER/PR+ patients was
showed a trend (0.16) toward improved survival. Treatment patterns have 16.5 and 8.6 months (NS), respectively. OS for chemoS and chemoR disease in
changed: the use of whole brain radiation therapy decreased from 71% to TNBC patients was 12.4 and 22.6 months, respectively (NS). The median TTP
67% to 47% in cohorts A, B and C, respectively. Conclusions: Despite the for all, HER2- ER/PR+, and TNBC patients was 2.7, 2.9, and 1.4 months,
shift to less favorable tumor subtypes, MS has improvedbut varies widely by respectively. Increases in serum IL-2 and IL-12 were associated with BATs
diagnosis-specific PF. Compared to prior cohorts, number of BM and ECM infusions. Conclusions: Targeting HER2- tumors was safe. There were trends
were identified as new PF. Ethnic, genetic and treatment differences be- toward improved survival in patients who were HER2-/ER/PR+ TNBC, patients
tween the eras are apparent. The updated Breast GPA, based on these data, who were chemoS, was associated with increased TTP and OS in all groups, and
and the correlation between BRCA1 and tumor subtype will be presented. was associated with decreased tumor markers in those who received 4 infusions.
Immune studies showed evidence for induction of adaptive immunity directed at
breast cancer antigens. Targeting metastatic HER2- BrCa with BATs shows
promise. Clinical trial information: NCT 01022138.

1082 Poster Session (Board #163), Sun, 8:00 AM-11:00 AM 1083 Poster Session (Board #164), Sun, 8:00 AM-11:00 AM
ARID1a as a marker of prognosis and increased sensitivity to CDK4/6, mTOR Evaluation of oral S-1 as a first-line chemotherapy for metastatic HER2-
1/2 and Src homology region 2 phosphatase (SHP 1/2) inhibitors in breast negative breast cancer: An analysis of two randomized phase III studies
cancer (BC). First Author: Veronica Mariotti, Moffitt Cancer Center, Tampa, (SELECT BC-CONFIRM and SELECT BC). First Author: Reiki Nishimura,
FL Breast Center, Kumamoto Shinto General Hospital, Kumamoto, Japan
Background: ARID1a (AT Rich Interactive Domain 1A) is part of the SWI/ Background: Anthracycline regimens and taxane have been first line che-
SNF complex, which regulates gene transcription, and is believed to be a motherapeutic options for HER2 negative metastatic breast cancer. In a
tumor suppressor gene. Low ARID1a expression has been associated with previous phase III trial (SELECT BC), non-inferiority of S-1 was demon-
poor prognosis in BC. The aim of this study was to explore the clinical strated in terms of overall survival (OS). The SELECT BC-CONFIRM study
significance of ARID1a mutation and expression loss, and its potential as a was designed to confirm the results of the SELECT BC study and to combine
therapeutic target in BC. Methods: We analyzed publicly available genomic the two randomized studies. Methods: Patients (n = 618) in the first trial
databases to study the clinical implication of ARID1a mutations and gene were randomly assigned (1:1) to the S-1 group or the taxane group. Patients
expression in BC. Results: ARID1a was mutated in ~5-7 % of BCs within (n = 230) in the second trial (SELECT BC-CONFIRM) were randomly
TCGA/METABRIC/MSK (5511 samples), but did not show differences in assigned to the anthracycline group or the S-1 group. Treatment continued
frequency between histology, grade, or estrogen receptor (ER)/HER2 re- until tumor progression, unacceptable toxic effects, or completion of six
ceptor status. MSK metastatic tissue samples had higher incidence of courses in the standard regimen group and four courses in the S-1 group. The
ARID1a mutation compared to primary tumor samples (7.6% vs 4.4%, x2 primary endpoint was OS and secondary endpoints were progression-free
P = 0.0073). Analysis of ARID1a in KMPLOT showed that lower gene ex- survival (PFS), time to treatment failure, adverse events, HRQOL and cost-
pression was associated with worse relapse-free survival and overall survival effectiveness. A pooled analysis of the two studies was predefined to confirm
across all BCs, but the difference was primarily in molecularly classified the results of the SELECT BC study. Results: 1. The HR for the anthracycline
luminal A tumors. Mutations in ARID1a did not show an association with group was 1.09 [95%CI 0.80-1.48] in SELECT BC-CONFIRM, and the
outcomes in TCGA/METABRIC/MSK datasets. Pathway analysis of ARID1a estimated predictive posterior probability that the HR does not exceed the
showed it is involved in regulating ER ligand driven signaling and interacts threshold 1.333 was 90.27%. 2. Median OS was 32.7 months (S-1 group)
with targets regulated by CDK4 and mTOR activity. CancerRxgene drug and 36.3 months (standard treatment group). S-1 was not inferior to
sensitivity analyses on BC cell lines revealed that ARID1a mutated BC cell standard treatment in terms of OS (p non-inferiority = 0.0062). Median PFS
lines were significantly more sensitive to palbociclib, SHP1/2, and mTOR1/2 was 11.2 months (S-1 group) and 11.2 months (standard treatment group).
inhibitors compared to ARID1a wildtype cell lines. Conclusions: Reduced 3. Treatment was discontinued due to adverse events (i.e., neutropenia,
activity of ARID1a in luminal BC cells may negatively affect prognosis by febrile neutropenia, fatigue and edema) in 5.7% in the S-1 group and 6.6%
altering ER signaling leading to activation of druggable resistance mecha- in the standard treatment group 4. The EORTC QLQ-C30 questionnaire
nisms, particularly in metastatic tissue. Loss of function ARID1a mutations (global health status) revealed that there was no difference between the S-1
may sensitize cancer cells to CDK4/6, mTOR1/2, and SHP1/2 inhibitors and anthracycline groups (p = 0.257), but there was a significant difference
in vitro. Further research in ARID1a mutated ER+ BCs using combinations of between the S-1 and taxane groups (p = 0.0039). Conclusions: S-1 is not
these inhibitors is warranted. inferior to taxane or anthracycline with respect to OS as a first-line treatment
for MBC. S-1 should be considered a new option as a first-line chemotherapy
for HER2-negative metastatic breast cancer patients. Clinical trial in-
formation: UMIN000005449.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Breast Cancer—Metastatic 55s

1084 Poster Session (Board #165), Sun, 8:00 AM-11:00 AM 1085 Poster Session (Board #166), Sun, 8:00 AM-11:00 AM
Oral paclitaxel in the treatment of metastatic breast cancer (MBC) patients. Reversion and non-reversion mechanisms of resistance (MoR) to PARP
First Author: Ming-Shen Dai, Tri-Service General Hospital, Taipei, Taiwan inhibitor (PARPi) or platinum chemotherapy (chemotx) in patients (pts) with
BRCA1/2-mutant metastatic breast cancer (MBC). First Author: Adrienne Gropper
Background: Intravenous (IV) paclitaxel is an effective treatment for breast
Waks, Dana-Farber Cancer Institute, Boston, MA
cancer. Oral administration paclitaxel is preferable to IV regarding mini-
mizing IV injections, anaphylactic reactions to cremaphor, steroid pre- Background: Little is known about MoR to PARPi and platinum chemotx in MBC
medications, hospital visits, and relevant costs. However, paclitaxel has pts with BRCA1/2 mutations. Biomarkers predictive of response/resistance have
poor oral absorption due to active excretion by P-glycoprotein (P-gp) in the not been identified, but could have clinical utility. Methods: We obtained 8
intestinal cells. Oraxol (Athenex, USA) is an oral paclitaxel and HM30181, a BRCA-mutant metastatic tumor biopsies from MBC pts with acquired resistance
novel oral inhibitor of intestinal P-gp which enables the oral administration of to DNA-damaging tx (PARPi/platinum) on a prospective tissue collection pro-
paclitaxel. We report the final results of a pharmacokinetics (PK) study, tocol. In 7/8 patients, we also obtained pre-tx biopsies. Whole exome sequencing
including clinical response and tolerability of Oraxol in treatment of met- (WES) was performed on each tumor and on germline DNA from blood. We
astatic breast cancer patients. Methods: Multicenter, single-arm, open- performed immunohistochemical (IHC) staining for RAD51 foci for functional
label, PK study of Oraxol (HM30181A at 15mg, plus oral paclitaxel assessment of intact homologous recombination (HR). Results: 4/7 pts with
205mg/m2) administered orally for 3 consecutive days weekly for up to complete WES analysis acquired a somatic reversion mutation likely to result in
16 weeks. Paclitaxel PK was assessed at week-1 and week-4. Tumor Re- functional BRCA1/2 protein in the post-tx tumor specimen after platinum (2 pts)
sponse was measured at weeks 8 and 16 using RECIST criteria 1.1. Toxicity or PARPi (2 pts; Table). 4/7 pts had plausible non-reversion MoR identified by
was assessed using CTCAE v4.03. Results: Twenty-eight MBC patient were WES, including alterations in genes involved in replication fork protection and
studied with a mean age of 56.6 years (range: 38 - 79 yrs). 26 patients had DNA end resection. As expected, in all pts with genomic reversion, RAD51 foci
failed mutiple previous chemotherapies. There were 11 (42.3%) partial were acquired in the post-resistance tumor, consistent with reconstitution of HR.
response, 12 (46.2%) stable disease and 3 (11.5%) progressive disease in In 2 pts without reversion, presence of RAD51 foci post-resistance was mixed.
26 evaluable patients. Three patients had treatment-related SAEs (grade $3 Reversion mutations occurred both with and without other alterations that could
neutropenia) and all patients recovered completely. PK results showed that possibly lead to fork protection, suggesting . 1 MoR could occur in the same
the AUC of oral paclitaxel at week-1 was reproducible at week-4 (3050 to tumor. 3 pts whose tumors demonstrated RAD51 foci post-resistance were
3594 ng-hr/mL). Conclusions: Oral paclitaxel showed very encouraging anti- later re-exposed to DNA-damaging tx, to which all had intrinsic resistance.
cancer activity in MBC patients who failed previous chemotherapies with Conclusions: BRCA1/2 reversion was identified as a MoR in the majority of pts.
acceptable toxicity. Weekly oral paclitaxel can achieve paclitaxel exposure WES identified potential novel MoR in fork protection and end resection genes.
similar to that of weekly IV paclitaxel (80mg/m2) reported previously. PK of The presence of RAD51 foci by IHC was consistent with BRCA protein functional
oral paclitaxel is reproducible. Clinical trial information: NCT03165955. status from genomic data and predicted response to later DNA-damaging tx,
suggesting RAD51 IHC may be a clinically useful biomarker.
BRCA reversion post-DNA Possible non-reversion MoR from RAD51 foci post-DNA
Pt damaging tx WES (gene) damaging tx
292 No Yes (many) Unknown
303 No No No
318 Yes, No Yes
339 Yes Yes (KMT2B) Yes
349 Yes Yes (KMT2C) Yes
359 No Yes (MRE11) Yes
510 Yes No Yes
565 Pending Pending Yes

1086 Poster Session (Board #167), Sun, 8:00 AM-11:00 AM 1087 Poster Session (Board #168), Sun, 8:00 AM-11:00 AM
Hyperprogressive disease in advanced triple-negative breast cancer (aTNBC) Association of secondary somatic mutations in BRCA1/2 with clinical
treated with immunotherapy (IO). First Author: Tira Jing Ying Tan, Division of resistance to PARP inhibitors and chemotherapy. First Author: Lingjun Zhu,
Medical Oncology, National Cancer Centre Singapore, Singapore, Singapore The First Affiliated Hospital of Nanjing Medical University, Jiangsu Province,
China
Background: Hyperprogression of disease (HPD), a rapid acceleration of tumor
growth rate (TGR) has been reported with IO in other tumor types. Here, we Background: Somatic reversion mutations in either BRCA1/2 has been re-
explore HPD in aTNBC. Methods: A retrospective chart review identified ported to lead to the resistance of platinum-based chemotherapy or PARPi.
aTNBC patients who consented for IO clinical trials at Princess Margaret In this study we try to analyze the secondary somatic mutations in BRCA1/2
Cancer Centre between June 2013 and June 2018. Demographic data, in patients with germline mutations. Methods: Using gene-panel target-
medical history, details of trial enrolment and RECIST 1.1 response to study captore next generation sequencing, we analyzed the secondary somatic mu-
treatment were recorded. Patients with RECIST 1.1 measurable disease on CT tations from 86 patients with BRCA1/2 germline mutations. Results: Eighty-six
scans or physical examination before trial entry, at trial baseline and at cases with BRCA1/2 gremline mutations were identified. Secondary somatic
protocol-defined interval following IO start were evaluable for TGR as defined mutations restoring BRCA1/2 were identified in 7 patients, including 2 breast
by Champiat et al. Clin Cancer Res 2017. HPD defined as a $2-fold increase in cancer, 3 ovarian cancer, 1 prostate cancer and 1cholangiocarcinoma patient.
TGR between baseline and on-trial restaging assessment. Univariable logistic For these seven patients, five had been treated with platinum-based chemo-
regression used to identify variables [age, co-morbidity index, prognostic index, therapy without PARPi and the other two (patient 1 and 2) with PARPi (olaparib).
performance status, distant disease free interval (dDFI), lactate dehydrogenase, Patient 1 and 2 both received targeting therapy of PARP inhibitor olaparib after
no. of metastatic sites, visceral disease and no. of prior treatment lines] asso- the germline BRCA1/2 mutation was detected. About six months later, plasma
ciated with HPD. Overall survival (OS) curves were estimated with the Kaplan- ctDNA was sequenced. Result showed that the germline mutations remained and
Meier method and compared by the log-rank test. Results: 99 patients with additional larger deletions was detected. These secondary somatic mutations are
aTNBC consented for 15 IO clinical trials, 60% IO monotherapy, 22% not predicted to significantly affect the BRCA1/2 protein, and are likely to cause
chemotherapy+/-IO and 18% IO combinations. Median age 52 (range 25-78), resistance to platinum-based chemotherapy or PARPi therapy by restoring
median no. of lines of prior systemic therapy for advanced disease 1 (range 0-8). BRCA1/2 ORF and DNA repair function. Conclusions: Secondary somatic mu-
15% had de-novo metastatic disease, 58% recurred after a dDFI of , 3 years tations that restore BRCA1/2 in carcinomas with germline BRCA1/2 mutations
and 25% after a dDFI of . 3 years. 61% had , 3 metastatic disease sites, and predict resistance to platinum-based chemotherapy and PARP inhibitors, some
71% had metastases involving the viscera. 66 received IO treatment, 40 patients strategies to reverse this type of drug resistance need further investigation.
(20 monotherapy, 7 IO combination, 13 chemotherapy+/-IO) were evaluable for Patient Disease Gene Germline Mutation Secondary somatic mutations
TGR. Median TGR pre-IO was 74.3 (range -17 – 1680) and post-IO was 2.5 1 Breast cancer BRCA2 c.176delC p.P59_N72delinsQTYLKLHKGNH
(-71.4 – 223). 4 patients (10%) met criteria for HPD. All 4 treated with （p.P59Qfs*21） p.P59_P65del
2 Ovarian cancer BRCA1 c.3968_3971del p.A1308_M1324delinsGSFLDWFFQT
monotherapy PD1 inhibitor and received at least 2 further lines of therapy post- （p.Q1323Rfs*12）
3 cholangiocarcinoma BRCA2 c.5645C . A p.S1882D
trial; 1 patient treated with IO as first-line therapy, 3 in the second or later lines. （p.S1882*）
There was no significant difference in the overall OS of patients with HPD and 4 Ovarian cancer BRCA1 c.981_982delAT
（p.C328*fs*1）
p.S308_C328delinsQGANITDGLEVRKR

patients who did not meet definition for HPD HR 0.89, (95% CI: 0.26-3.01; p = 5 Ovarian cancer BRCA1 c.4801A . T p.K1601_V1602delinsFF
（p.K1601*）
0.41). Univariable analysis did not identify factors associated with HPD. 6 Prostate cancer BRCA2 c.5895delT p.H1966_E2002delinsISQSHLQILVGFLAQQVENLSRYQMLHYKTQ
Conclusions: HPD was observed in 10% of aTNBC treated on IO clinical trials. 7 Ovarian cancer BRCA1
（p.H1966Ifs*38）
c.3704delA p.N1215_V1234delinsS
HPD was not associated with worse survival outcomes or known prognostic （p.N1235Tfs*29）

factors in our analysis.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
56s Breast Cancer—Metastatic

1088 Poster Session (Board #169), Sun, 8:00 AM-11:00 AM 1089 Poster Session (Board #170), Sun, 8:00 AM-11:00 AM
Differences in breast cancer outcomes amongst Black United States-born Detection of subthreshold microsatellite instability in breast cancer: An
and Caribbean-born immigrants. First Author: Priscila Barreto Coelho, ongoing investigation. First Author: Joshua Z. Drago, Memorial Sloan Kettering
Jackson Memorial Hospital/University of Miami, Miami, FL Cancer Center, New York, NY
Background: The Black population in the US constitutes of 4 million im- Background: Microsatellite instability (MSI) and mismatch repair deficiency
migrants, with 50% from the Caribbean. It has been shown that breast (MMRd) can occur sporadically in solid tumors across histologic subtype and
cancer is responsible for 14%-30% of cancer deaths in the Caribbean; this is predict clinical response to immune checkpoint blockade (ICB). MSIsensor
up to two times higher than the USA. Methods: Retrospective cohort of 1369 uses next generation sequencing to report the percentage of unstable
self-identified Black women with breast cancer. Data was obtained from microsatellites in a given tumor sample as a score. While scores of $10
Jackson Memorial Health Systems and University of Miami Health System and , 3 have excellent sensitivity/specificity for MSI/MMRd, intermediate
Tumor Registry. Individual-level data from 1132 cases was used to estimate scores of 3-10 are of unknown clinical significance, and may include pa-
hazard rations (HRs) of women born in the Caribbean (CB) or in the USA tients with MSI-like genomic signatures who would be overlooked by his-
(USB) using Cox proportional hazards regression analysis for overall survival. torical assays. Methods: MSIsensor testing was performed on primary or
Median follow-up was 115 months (interquartile range, 91.9-138.1 months) metastatic tumor specimens from 2,371 patients with breast cancer. Scores
per participant. Results: Data from 622 (54.9%) USB women and 507 of 3-10 were considered MSI-Indeterminate (MSI-I). Concurrent somatic
(45%) CB women diagnosed with breast cancer between 2006-2017. 90% mutations were determined using the MSK-IMPACT assay. Results: Of the
(n = 1232) of the cohort is of non-Hispanic ethnicity. Caribbean immigrants 2,371 patients tested, 147 were found to have MSI-I tumors. Ninety-nine
from Haiti (18.3%), Jamaica (6.5%), Bahamas (3.1%), Cuba and Dominica (67.3%) were metastatic specimens, and 48 (32.7%) were primary spec-
Republic (2.8% each), Trinidad and Tobago (1%) and other nationalities from imens. Of the MSI-I patients, 64 (43.5%) were estrogen receptor positive,
the Organization of Eastern Caribbean States were included, mean age 55.7 57 (38.8%) were triple negative, and 24 (16.3%) were HER-2 amplified.
[95% CI, 54.7-56.8]; USB mean age 57.6 [95% CI, 56.4-58.7] (P = 0.02). Somatic TP53 mutations were detected in 109 (74.1%) MSI-I patients,
Compared to USB, CB had lower BMI at diagnosis 29.6 [95% CI, 28.9-30.3] including 54 (94.7%) triple negative patients. Fourteen MSI-I patients were
versus 30.9 [95% CI, 30.1-31.7, P = 0.015]. Compared to CB patients, USB found to have somatic mutations in MMR genes (MLH1, MSH2, MSH6,
patients had more ER- [31.4% vs 39.1 %, P = 0.018] and triple negative breast PMS2 or EPCAM), but no germline mutations were found. Of the 99 patients
cancers [19.6% vs 27.9%, P = 0.003]. Compared to USB patients, CB pre- with MSI-I metastatic disease, 15 received ICB as part of their clinical care,
sented at more advanced stage, III and IV [44.2% vs 35.2%], p = 0.016. In spite exhibiting a median progression-free-survival (PFS) of 3 months. Of the
of higher advanced stage at diagnoses, CB patients had a better breast cancer 4 patients who demonstrated a PFS of . 6 months on ICB, 2 were found to
overall survival [HR = 0.75; 95%CI, 0.59-0.96; P = 0.024]. Black Hispanic have somatic mutations in MMR genes. Conclusions: We introduce a sub-
patients had a better overall survival [HR = 0.51; 95%CI, 0.28-0.93; p = 0.028] population of breast cancer patients whose tumors exhibit genomic signs of
compared to non-Hispanic Blacks. Compared to Hispanic Caribbean, non- microsatellite instability without meeting the classic criteria for MSI/MMRd
Hispanic Caribbean had a worse overall survival [HR = 1.98; 95%CI, 1.00- as previously defined in colon and endometrial cancers. A subset of these
3.94; P = 0.048]. The distribution of patients treated at the private cancer center patients may exhibit the properties of MSI/MMRd at the epigenetic and
and the safety net hospital were the same, differences in outcomes observed are protein-expression levels, and thus potentially benefit from ICB.
due to intrinsic differences. Conclusions: This is the largest analysis to date of
self-identified Black breast cancer patients in the context of nativity, race, ethnic
identity and overall survival with clinico-pathologic characteristics. CB immi-
grants diagnosed with breast cancer have a better overall survival than US born
Black patients. This finding suggests that within the African diaspora in the USA,
additional factors beyond race contribute to the outcomes.

1090 Poster Session (Board #171), Sun, 8:00 AM-11:00 AM 1091 Poster Session (Board #172), Sun, 8:00 AM-11:00 AM
Gaps in metastatic breast cancer patient knowledge and understanding in TAKTIC: A prospective, multicenter, uncontrolled, phase IB/II study of
Mexico. First Author: Cynthia Villarreal-Garza, Depto. de Tumores Mamarios LY2780301 (LY) in combination with weekly paclitaxel (wP) in HER2-
e Investigación, Instituto Nacional de Cancerologı́a (INCan), Mexico City, negative locally advanced (LA) or metastatic breast cancer (MBC) patients.
DF, Mexico First Author: Cecile Vicier, Aix-Marseille Univ, CNRS,INSERM, Institut-Paoli-
Calmettes, Department of Medical Oncology,CRCM, Marseille, France, Marseille,
Background: Among patients with metastatic breast cancer (MBC), there is
France
paucity of data regarding knowledge and understanding about disease stage
and goals of treatment. The limited evidence shows that up to 40% of MBC Background: Phosphatidylinositol-3-kinase (PI3K)/AKT/mammalian target
patients ignore their stage and reports inadequate educational resources and of rapamycin (mTOR)-pathway is frequently activated in HER2-negative
support. Aim: To assess MBC patients’ knowledge of their disease and breast cancer and may play a role in taxane-resistance. LY is a dual inhibitor
satisfaction with the medical information received. Methods: 185 patients of p70 S6 kinase and AKT. TAKTIC study aimed to determine the recommended
diagnosed with MBC who attended follow-up medical appointments at phase II dose (RP2D) of the combination of LY with wP (phase Ib) and to estimate
INCan from June to December 2018 were asked to complete a survey to overall response rate (ORR) of this regimen (phase II) in HER2-negative LA or
assess their knowledge on MBC curability, subtype, topics reviewed with the MBC patients, both in the overall patient population and in patients with acti-
oncologist, need of additional information regarding MBC and satisfaction vation of PI3K/AKT pathway (PI3KAKT+). Methods: HER2-negative inoperable
with the quality of the information they received. Clinical data was obtained LA or MBC patients (pts), with (phase Ib) or without (phase II) previous
from the medical records. Descriptive statistics were applied, and the as- cytotoxic treatment for advanced disease were eligible. Oral LY (400 or
sociation between qualitative and quantitative variables was assessed with 500mg) was administered daily in combination with intravenous wP (70 or
the Chi square test and the Mann Whitney U test, respectively. This study was 80mg/m2). A modified CRM using an adaptive Bayesian model guided the
supported by the SPARC mBC Challenge. Results: Only 52% of patients dose escalation of both agents. PI3KAKT+ was defined as activating mutation
knew that their disease is not curable, while 31% were not sure and 17% of PIK3CA and/or AKT by targeted NGS or homozygous loss of PTEN by array
thought it was curable. 82% ignored their MBC subtype and 81% reported to comparative genomic hybridization (aCGH) or loss of PTEN by immunohis-
want more information about their cancer. 64% scored their satisfaction with tochemistry, as evaluated on available fresh tumor tissue. Results: A total of
the provided information $9 (out of 10), and 74% stated that their most 12 and 35 patients (pts) were included in the phase Ib and II, respectively. In
helpful source of information was the medical staff. A statistically significant phase Ib, only 1 dose-limiting toxicity (confusion) was observed at the last dose
association was found between the perception of the impossibility of curing level (LY, 500 + wP, 80), which was determined as RP2D. Main drug-related
MBC and better patient satisfaction (p = 0.038). Likewise, better patients’ adverse events (AE) in phase Ib were skin toxicity (92% of pts, G3-4 in 33%),
satisfaction was associated with older age (p = 0.002), longer time since and paresthesia (50% of pts, G3-4 in 8%). In the phase II study, ORR was 62.9 %
initial diagnosis (p = 0.018) and since diagnosis of MBC (p = 0.014). [44.9,78.5] including 1 CR and 21 PR in the overall population and 55.6 %
Conclusions: Our results dramatically illustrate the lack of understanding [30.8,78.5] in PI3KAKT+ pts (10 PR in 18 pts). Median progression-free
patients have regarding their cancer, even when reporting high satisfaction survival was 12.4 months [7.9,17.9] and 6-month clinical benefit rate was
with the received information, and identify a critical need for improvement of 82.9% [66.4,93.4]. AEs in phase II were similar to phase I part, except that
healthcare providers’ communication skills. Directed efforts should be 17% of pts experienced pneumonia (G3-4 in 9%). Conclusions: Combining
implemented to enhance patients’ comprehension to promote their partici- LY and wP in HER2-negative LA or MBC was feasible with preliminary
pation in decision-making processes, adherence and, ultimately, outcomes. evidences of efficacy, independently of PI3K/AKT activation. Clinical trial
information: NCT01980277.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Breast Cancer—Metastatic 57s

1092 Poster Session (Board #173), Sun, 8:00 AM-11:00 AM 1093 Poster Session (Board #174), Sun, 8:00 AM-11:00 AM
RUBY: A phase II study testing rucaparib in germline (g) BRCA wild-type Survival outcomes of premenopausal patients diagnosed with invasive
patients presenting metastatic breast cancer (mBC) with homologous recombination lobular carcinoma. First Author: Hee Jeong Kim, Department of Surgery,
deficiency (HRD). First Author: Anne Patsouris, Institute of West Cancerology University of Ulsan College of Medicine and ASAN Medical Center, Seoul,
Paul Papin, Angers, France South Korea
Background: PARP-inhibitors improve PFS in mBC patients (pts) harboring a Background: The survival outcomes of postmenopausal women with invasive
gBRCA mutation (mut). However, unlike ovarian cancer, there is no evidence lobular carcinoma (ILC) are similar to invasive ductal carcinoma (IDC) when
until now that this class of agents has efficacy in gBRCA wild-type (WT) pts. treated with aromatase inhibitor, but inferior when treated with tamoxifen.
In RUBY, we evaluated rucaparib in gBRCA WT mBC pts, and whose tumor We sought to investigate the survival outcomes of premenopausal ILC when
present with HRD as assessed by genome-wide Loss of Heterozygosity (LOH) compared to IDC using a population-based analysis. Methods: We used
score. Methods: 713 gBRCA WT women with HER2- mBC, initiating a first Surveillance, Epidemiology, and End Results data (SEER) between 1990 to
metastatic chemo, were screened for high ($18%) genomic tumor LOH 2015 to identify premenopausal patients (defined as , 50 years old), di-
generated from a SNP array on metastatic sample. Eligible pts with a high agnosed with stage I to III IDC or ILC. Breast Cancer Specific Survival (BCSS)
LOH score or somatic (s)BRCA mut and $1 prior chemo regimen were was assessed using log-rank test and piecewisecox models. Annual hazard of
proposed to enter RUBY and receive oral rucaparib 600 mg BID continuously BCCS for hormone receptor positive (ER+) ILC and ER+ IDC were calculated
in 28-day cycles until disease progression. The primary endpoint was from year 0 to 20 and defined as the proportion of patients with a BCSS event
clinical benefit rate (CBR), defined by complete (CR) and partial response (PR) during a 1-year interval. Results: The study includes a total of 170,352 pts
or stable disease (SD) $16 weeks. We used a Simon’s two-stage design diagnosed with either IDC (n = 158,733) and ILC (n = 11,619). 71% of IDC
(p0=20%; p1=40%), responses in $4/17 pts were expected to move to and 95% of ILC pts were ER+. Median age was 44 years old and median
second step, and $11/37 pts to be considered of clinical interest (a=10% and follow up was 90 months(IQR 40-151 months). Survival analysis revealed a
power of 90%). Whole genome sequencing (WGS) was performed retro- significant time-dependent effect of histology for BCSS (p , .0001). When
spectively to further assess potential biomarkers of PARP inhibitor response. compared to IDC, ILC pts had better BCSS in the first 10 years after diagnosis
Results: Tumors from 221 (31%) pts were LOH high. 41 pts were enrolled, (HR 0.73, 95% CI 0.68-0.78), but worse BCSS outcome after year 10 (HR
including 4 pts with sBRCA mut. Median prior metastatic chemo lines was 1.80, 95% CI 1.59- 2.03). Similar results were observed when adjusting for
2 (1-5), 17 pts had TNBC at diagnosis. As of 14 Jan 2019, 16 pts were alive, ER status, histologic grade and stage on multivariate analysis. Among pts
5 are still on treatment. The median number of cycles was 2 (1 -20), and 37/40 # 35 years old(n = 371 ILC; 18086 IDC) survival analysis revealed a non-
patients were evaluable for CBR. 5 pts (13.5%) demonstrated clinical benefit significant trend towards inferior outcome for ILC compared to IDC throughout
(1 CR [LOH high], 3 PR [2 LOH high, 1 sBRCA2] and 1 SD.31 weeks the whole follow-up period (HR = 1.2 95% CI 0.96-1.52). Annual hazard of
[sBRCA1]). 19 pts had grade 3-4 toxicities. 3 pts discontinued due to toxicity. BCCS events showed a peak at year 5 for both IDC and ILC. In the subset of
4/5 responders pts had their tumor profiled by WGS: preliminary analyses showed IDC, we noticed a decreasing hazard of BCSS from years 6 through 20. By
that 4 pts presented high large scale state transitions, and 3 presented a somatic contrast, in the subset of ILC, we observed higher frequencies of BCSS from
biallelic loss of function in HR-related genes. The fifth responder harbored a mut years 6-20 when compared to IDC (p , 0.0001). Conclusions: In this
on gPALB2 and sBRCA2 at inclusion. Conclusions: In this study, rucaparib population-based analysis, premenopausal ILC had worse BCSS estimates
demonstrates antitumor activity in a subset of gBRCA WT mBC pts whose tumor when compared to premenopausal IDC. This is explained by a higher incidence
has high LOH. Final analyses of WGS will provide insights about HRD signatures of late events in the subset of ILC when compared to IDC.
and drivers alterations associated with response. Clinical trial information:
NCT02505048.

1094 Poster Session (Board #175), Sun, 8:00 AM-11:00 AM 1095 Poster Session (Board #176), Sun, 8:00 AM-11:00 AM
Development and validation of novel microenvironment-based immune Clinical characteristics and outcome of metaplastic breast cancer: A retrospective
molecular subtypes of breast cancer: Implications for immunotherapy. tertiary care center experience. First Author: Bicky Thapa, Department of
First Author: Yunfang Yu, Guangdong Provincial Key Laboratory of Malignant Medicine, Cleveland Clinic, Cleveland, OH
Tumor Epigenetics and Gene Regulation, Department of Oncology, Sun Yat-sen
Background: Metaplastic breast cancer (MBC) is a rare neoplasm which
Memorial Hospital, Sun Yat-sen University, Guangzhou, China
accounts for less than 1% of all breast cancers. MBC is associated with worse
Background: Breast cancer treatment with immunotherapy can improve prognosis and there is a paucity of literature on management. We evaluated
clinical benefits, but the majority of patients did not respond to the treat- the clinical characteristic and outcomes of MBC patients at our institution.
ment. To understand tumor–immune interactions in breast cancer, we Methods: After IRB approval, 136 patients diagnosed with MBC were
identified novel microenvironment-based immune molecular subtypes. reviewed from the Cleveland Clinic tumor registry from 2000-2017. Patients
Methods: A training cohort of 1,394 breast cancer patients from the Mo- were evaluated for overall survival (OS) and progression free survival (PFS)
lecular Taxonomy of Breast Cancer International Consortium profiled by RNA using univariable analysis. Time to event variables were estimated by
and DNA sequencing data were analyzed to calculate immune-related gene Kaplan-Meier method. Results: A total of 136 pathologically proven MBC
biomarkers and to assign prognostic categories using LASSO Cox regression patients were included in the study. Median age at diagnosis was 60 years
model. Additionally, 969 patients from The Cancer Genome Atlas data set (27-92). Eighty two percent (n = 112) had nuclear grade III, 7% (n = 10) had
was used as an independent validation cohort. We further compared tumor high grade dysplasia, 2% (n = 3) had nuclear grade I, and 4% (n = 5) had
mutation burden (TMB) and cytolytic activity (CYT) levels between different nuclear grade II; 60% (n = 82) patients were diagnosed at stage II, 21% (n =
immune molecular subtypes. Results: Using the LASSO model, we estab- 28) at stage I, 14% (n = 19) at stage III, and 5% (n = 7) at stage IV. Estrogen
lished an immune molecular classifier based on following 5 features: IFN-g receptor, progesterone receptor and Her2 expression were positive in 16%
signature, ICOSLG, TNFRSF14, Mast.cells.resting, and T.cells. CD4.me- (n = 22), 9% (n = 12), and 10% (n = 14) respectively. Only 37% (n = 50)
mory.resting. Then we found that it contained two distinct microenvironment- patient had lumpectomy, 18% (n = 25) received hormonal therapy, 56% (n =
based subtypes (immune class and non-immune class), characterized by 76) received radiation, 51% (n = 70) received anthracycline chemotherapy
significant differences in overall survival in the training cohort (hazard ratio and 26% (n = 36) received non-anthracycline chemotherapy; 37% (n = 50)
[HR] 0.71; 95% confidence interval [CI] 0.61 to 0.81; P , 0.001) and in the had chemotherapy after 4 weeks of surgery and 35% (n = 48) patients had
validation cohort (HR 0.34; 95% CI 0.22 to 0.54; P , 0.001). We found an chemotherapy within 4 weeks of surgery. On univariable analysis, the 5-year
inverse association between immune gene expression and TMB levels (r = OS for stage III was 30% (14% - 64%), hazard ration (HR) of 4.53 (95% CI,
0.096, P , 0.001). Immune class subtype patients with good prognosis had 1.71 - 12.01) (p = 0.002), for stage IV HR of 43.26 (95% CI, 12.34 -
significantly lower TMB and higher CYT than did non-immune class subtype 151.64) (p = 0.001); chemotherapy within 4 weeks of surgery was asso-
patients with poor prognosis (all, P , 0.05). The clinical use of the immune ciated with a higher risk of death, HR of 0.30 (95% CI, 0.12 - 0.74) (p =
molecular subtypes showed a closer association with survival than did IFN-g 0.009). Hormonal therapy, radiation therapy, surgery and type of chemo-
signature or PD-L1 expression (all, P , 0.05). The robustness of the immune therapy was not associated with significant change in OS and PFS. In our
molecular subtypes was confirmed in the validation cohort. Conclusions: We cohort, 2-year OS was 79 % (73 % - 87 %) and 5-year OS was 69 % (61 % -
revealed novel immune molecular subtypes, which represented better utility in 77 %); 2-year PFS was 61 % (52 % - 70 %) and 5-year PFS was 72 % (65 % -
predicting breast cancer patients’ survival compared with IFN-g signature or 81 %). Conclusions: Stage of MBC and chemotherapy within 4 weeks of
PD-L1, and could be an important guide for precision immunotherapy. surgery was associated with statistically significant OS and PFS on uni-
variable analysis. Randomized clinical trials are warranted to improve out-
comes in MBC patients.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
58s Breast Cancer—Metastatic

1096 Poster Session (Board #177), Sun, 8:00 AM-11:00 AM 1097 Poster Session (Board #178), Sun, 8:00 AM-11:00 AM
A phase II study of pembrolizumab and capecitabine for triple-negative (TN) Measuring on-treatment genome-wide tumor copy number alterations in cell-free
and hormone receptor-positive, HER2-negative endocrine-refractory metastatic DNA (cfDNA) in plasma is highly prognostic in metastatic breast cancer. First
breast cancer (MBC). First Author: Ami N. Shah, Northwestern University, Author: Adriana Aguilar, Segal Cancer Centre/Jewish General Hospital and
Chicago, IL McGill University, Montreal, QC, Canada
Background: Response rates to single agent immune checkpoint blockade in Background: The clinical management of metastatic breast cancer depends
unselected MBC are low; however, they may be augmented when combined on the measurement of tumor response to successive drugs by serial imaging
with chemotherapy. Methods: We conducted a single-arm, phase II study of and changes in blood tumor markers, which remain the standard of care
patients with TN or endocrine-refractory MBC who were candidates for despite poor sensitivity and specificity. Highly sensitive and specific cfDNA
capecitabine. Patients were treated with pembrolizumab 200 mg IV day 1 secreted from the tumor can detect the changes in tumor-specific aberra-
and capecitabine 1000 mg PO BID days 1-14 of a 21-day cycle. The primary tions that have been shown to be associated with patient response in the
endpoint was progression free survival (PFS) and secondary endpoints were metastatic setting. However, most approaches require prior sequencing of
overall response rate (ORR), safety and tolerability. The study had 80% the tumor to target specific known mutations. Methods: Using low coverage
power to detect a 2 month (mo) improvement in mPFS with the addition of genomic sequencing, a genomic instability number (GIN) was measured in
pembrolizumab over historic controls treated with capecitabine alone. cfDNA based on the detection of genome-wide tumor-specific DNA copy
Results: Thirty patients, 16 with TN and 14 endocrine-refractory MBC, were number alterations for 27 patients with metastatic breast cancer. The GIN
enrolled from 2017-18. Patients had a median age of 51 years and value and its variation from baseline before treatment, as well as within
received a median of 1 prior line of therapy for MBC. Of 29 evaluable pa- 10 days and 3 weeks after start of therapy were compared with tumor re-
tients, the mPFS was 4.1 mo (95% CI, 2.3-8.2 mo) and median overall sponse, progression free survival (PFS) and overall survival (OS) of the
survival was 15.4 mo (95% CI, 8.2-16.6 mo). ORR was 14% (n = 4), stable patients. Patients were followed for a median of 22 months and we used a
disease (SD) was 41% (n = 12), and clinical benefit rate (CBR = PR + SD . 6 previously published GIN threshold at 170 for high vs low GIN values. Se-
mo) was 28% (n = 8). The ORR and CBR were similar between disease quencing was performed blinded to the clinical results. Results: Baseline GIN
subtypes (ORR 13% and 14%, CBR 25% and 29% for TN and endocrine values were not associated with tumor response at 3or 6 months, but showed a
refractory, respectively). The 1-year PFS rate was 20.7% and 3 pts have trend towards lower OS with higher GIN (p = 0.12). GIN values fell by an
ongoing responses. The most common adverse events were low grade and average of 28% in responders (stable disease or response) and 23% in those
consistent with those seen in MBC patients receiving capecitabine, in- with progression (p = 0.85), but remained lower at 3 weeks only in the re-
cluding hand-foot syndrome, gastrointestinal symptoms, fatigue, and sponders. High GIN values within 10 days and 3 weeks were associated with
cytopenias. Toxicities at least possibly from pembrolizumab included grade markedly worse OS (p = 0.014 and p = 0.009 respectively) and those at
3 or 4 liver test abnormalities (7%), rash (7%), and diarrhea (3%), as well as 3 weeks with worse PFS (p = 0.017). Hence the median survival of patients
grade 5 hepatic failure in a pt with liver metastases. Conclusions: Compared with high GIN at 10 days or 3 weeks was 12 months vs not reached for those
to historical controls, pembrolizumab with capecitabine did not improve PFS with low GIN. The percentage drop of GIN at 10 days but not at 3 weeks was
in this biomarker unselected cohort; however, some patients had prolonged significantly associated with PFS (p = 0.016). Conclusions: These results
disease control. Future studies of chemo-immunotherapy in MBC with liver demonstrate that GIN values of cfDNA measured at early on-treatment time
metastases require close safety monitoring. Clinical trial information: points can predict PFS and OS with a high degree of accuracy. These findings
NCT03044730. deserve further study in a larger cohort but hold the promise of early prediction
of clinical outcomes in a tumor-independent genome-wide approach.

1098 Poster Session (Board #179), Sun, 8:00 AM-11:00 AM 1099 Poster Session (Board #180), Sun, 8:00 AM-11:00 AM
Evaluation of pathologic and genomic characteristics in male breast cancer Patients with metastatic breast cancer enrolled in phase I clinical trials:
(MBC) patients. First Author: Damien Mikael Hansra, Oncology and Radiation Clinical outcomes and cohort trends. First Author: Jennifer Weiss, University
Associates/Mercy Research Institute, Miami, FL of Colorado, Aurora, CO
Background: MBC is a rare entity comprising less than 1% of breast cancers Background: Phase I clinical trials have traditionally enrolled patients with
[Siegel RL 2017]. Due to the low incidence of MBC, information about the advanced solid tumors and many providers perceive the likelihood of clinical
genomic landscape of MBC is lacking. Here we describe detailed pathologic and benefit as low. The purpose of this study was to evaluate clinical outcomes
genomic characteristics of MBC patients. Methods: IRB approval was obtained for patients with metastatic breast cancer enrolled on Phase I clinical trials
for a retrospective analysis of archived pathology on patients treated at Cancer and explore differences in outcomes for patients enrolled in all-comer versus
Treatment Centers of America. Comprehensive genomic profiling of tumors was breast cancer-specific cohorts. Methods: We performed a retrospective chart
derived from Foundation One next generation sequencing. Clinical information review of patients with metastatic breast cancer enrolled in Phase I clinical
was derived from retrospective chart review. Inclusions: adult males with breast trials at the University of Colorado Cancer Center from 2012-2018. We
cancer with stage IV metastatic disease. Exclusions: Females, stage 0-IIIC dis- included trials with Phase I and/or Phase Ib in the title. Studies or cohorts
ease, missing genomic and pathology information. Results of clinical, pathology enrolling patients with $ 3 tumor types were considered all-comer and those
and genomic data were summarized. Results: 10 patients met study criteria. with enrollment restricted to breast cancer or a breast cancer subtype were
Median age: 56 yrs., range 39-61 yrs. Race: 5/10 (50%) Caucasian, 5/10 (50%) considered breast cancer-specific. Results: A total of 208 patients were
African American. Number of prior treatment regimens: mean = 2.6 (range 0-6). enrolled in Phase I clinical trials, 168 in breast cancer-specific cohorts and
Intrinsic subgroup: hormone receptor (HR)+/HER2- 7/10 (70%), HR+/HER2+ 2/ 40 in all-comer trials. Patients on average were 56.9 years old (range 31-79),
10 (20%), HR-/HER2+ 1/10 (10%), triple negative breast cancer (TNBC) 0/10 98.6% (205/208) female, 1.4% (3/208) male, 57.2% ER+/Her2-, 30.1%
(0%). Tumor histology: invasive ductal carcinoma 10/10 (100%). Histology ER-/Her2- and 11.1% Her2+. Patients received on average 2.1(range 0-10)
grade: poorly differentiated (diff.) 4/10 (40%), moderate diff. 4/10 (40%), well prior lines of chemotherapy in the metastatic setting. Patients enrolled on
diff. 1/10 (10%), unknown 1/10 (10%). Biopsy site: primary tumor 4/10 (40%), Phase I clinical trials remained on study without progression on average for
metastatic site 4/10 (40%), liquid biopsy 1/10 (10%). Most frequent genomic 138 days (CI 95%, 112.64 to 163.91). Patients enrolled on breast cancer-
alterations: PIK3CA 5/9 (56%), CCND1 4/9 (44%), ZNF703 3/9 (33%), FGF4 3/ specific studies remained on study for 152 days (CI 95%, 120.66 to
10 (33%), GATA3 3/9 (33%) FGF 19 3/9 (33%), FGF3 3/9 (33%) Alterations in 182.56) compared to 82 days (CI 95%, 59.43 to 105.13) for those enrolled
FGF seen 14/38 (36.8%) of total genomic alterations. Most frequent alterations by on all-comer trials, p, 0.05. Patients went off study for disease progression
primary tumor: ZNF703 3/4 (75%), FGF3 2/4 (50%), FGFR1 2/4 (50%), CCND1 (83.17%), adverse events (7.69%), and other (9.14%), including with-
2/4 (50%), FGF19 2/4 (50%), FGF4 2/4 (50%); by metastatic site: PIK3CA 4/5 drawal of consent. Conclusions: Patients with metastatic breast cancer
(80%), GATA 3 2/5 (40%), CCND1 2/5 (40%), FGF3 1/5 (20%), FGF19 1/5 previously treated with multiple lines of chemotherapy in the metastatic
(20%), FGF4 1/5 (20%). Genomic alteration by histologic subgroup; HR+HER2- setting enrolled in Phase I clinical trials received clinical benefit from
PIK3CA 3/6 (50%), CCND1 3/6 (50%), FGF4 3/6 (50%), ZNF703 3/6 (50%), treatment that is favorable compared to historical controls of late-line
FGF19 3/6 (50%), FGF3 3/6 (50%); HR+/HER2+: PIK3CA 2/2 (100%), CCND1 chemotherapy. The majority of patients were treated on breast cancer-
1/2 (50%), GATA3 1/2 (50%), MLL3 1/2 (50%); TNBC: BRCA2 1/1 (100%), specific cohorts consistent with trends in Phase I trial design including
BARD1 1/1 (100%). Microsatellite status was stable in 6/6 (100%) of patients. more tumor specific cohorts.
Conclusions: MBC patients display a heterogeneous variety of complex genomic
alterations. Mutations in FGF genes were most commonly observed. Other
common alterations seen in this series include PIK3CA, CCND1, ZNF703, and
GATA3. Furthermore, the genomic profile of primary tumor site differed from the
genomic profile of the metastatic site.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Breast Cancer—Metastatic 59s

TPS1100 Poster Session (Board #181a), Sun, 8:00 AM-11:00 AM TPS1101 Poster Session (Board #181b), Sun, 8:00 AM-11:00 AM
A phase III study comparing trastuzumab emtansine with trastuzumab, pertuzumab, B-PRECISE-01 Study: A phase Ib trial of MEN1611, a PI3K Inhibitor,
and docetaxel in elderly patients with advanced stage HER2-positive breast combined with trastuzumab 6 fulvestrant for the treatment of HER2-positive
cancer: (JCOG1607 HERB TEA study). First Author: Akihiko Shimomura, advanced or metastatic breast cancer. First Author: Martine J. Piccart-
Department of Breast and Medical Oncology, National Cancer Center Hospital, Gebhart, Institut Jules Bordet, Université Libre de Bruxelles, Brussels,
Tokyo, Japan Belgium
Background: Systemic chemotherapy with anti-HER2 therapy is the stan- Background: MEN1611 is a potent, selective class I inhibitor of PI3K, a key
dard of care for HER2-positive advanced breast cancer. Patient outcomes enzyme in the transduction of various extracellular growth factor signals
have improved remarkably with the use of novel anti-HER2 drugs, including essential for cell survival and apoptosis. The discovery in human cancers of
trastuzumab (H), pertuzumab (P), and trastuzumab emtansine (T-DM1). The frequent PIK3CA mutations, that have been linked to a worse outcome in
combination treatment comprising H, P, and docetaxel (D) (HPD) is highly advanced HER2-positive breast cancer, makes PI3K an attractive thera-
recommended as the 1st-line treatment for patients with HER2-positive peutic target. Preclinical and clinical evidences support the development of
advanced breast cancer. In contrast, for elderly patients over 65 years of age, MEN1611 in combination with other agents in the context of solid tumors.
this regimen seems to be intolerable mentally and physically, and impairs Methods: B-PRECISE-01 is an open-label, multicenter, phase Ib dose es-
their quality of life. A new standard treatment with less toxicity and non- calation study in patients with PIK3CA mutant tumors, HER2-positive ad-
inferior efficacy for elderly patients is needed. Methods: We have planned a vanced or metastatic breast cancer which has progressed after at least 2 lines
randomized, multicenter, open-label, phase III trial to confirm the non- of anti-HER2 based therapy. PIK3CA mutations are assessed centrally by real-
inferiority of T-DM1 compared to HPD in terms of overall survival (OS) in time PCR assay in DNA derived from archived tumor samples. MEN1611 will
elderly patients with HER2-positive advanced breast cancer. The eligibility be administered orally BID for continuous 28-day cycles until disease pro-
criteria are as follows: 1) histologically proven metastatic breast cancer, 2) gression, in combination with weekly IV infusions of trastuzumab. In addition,
age 65-74 years with a performance status (PS) score 0-2, or 75-79 years HR-positive postmenopausal patients will also be treated with fulvestrant. After
with a PS score 0-1, 3) HER2 overexpression or amplification confirmed in the completion of the dose escalation phase (Step 1), the study will continue in
primary or metastatic tissues, and 4) no anti-HER2 therapy with chemo- an expansion cohort (Step 2) testing the Recommended Phase 2 dose (RP2D)
therapy for breast cancer, excluding (neo) adjuvant therapy. Patients are in a total of 15 patients in each of the treatment groups. The primary study
randomized to receive either HPD (H 6 mg/kg, P 420 mg/body, and D 60 mg/ objective is to assess combination safety and select RP2D. Secondary ob-
m2) or T-DM1 3.6 mg/kg every 3 weeks. The dose up of D (75 mg/m2) after jectives include assessment of pharmacokinetics and pharmacodynamics,
the second cycle is defined based on the data regarding safety during the first preliminary clinical activity of MEN1611 in combination with trastuzumab +/-
cycle. The primary endpoint is OS. The secondary endpoints are progression- fulvestrant, and correlation with PIK3CA mutations and other relevant cancer
free survival, response rate, adverse events, breast cancer-related death, and genes mutational status. Adverse events will be graded according to NCI
deterioration of activities of daily living. The trial is designed to achieve 70% CTCAE v4.03. Responses will be evaluated according to RECIST v1.1. Study
power to confirm non-inferiority of T-DM1 to HPD at a one-sided alpha of 5% variables will be presented by dose-cohort and overall using appropriate de-
with a non-inferiority margin of 1.3 in terms of hazard ratio. With a median OS scriptive statistics. The enrollment began in July 2018 at European sites and
of 30 months in both arms, 6 years of accrual, and 5 years of follow up, the US sites will shortly participate; up to date the first dose cohort level has been
planned sample size is 330. The trial began in January 2018 and nineteen achieved. Clinical trial information: NCT03767335.
patients have been enrolled as of January 2019. Clinical trial information:
UMIN000030783.

TPS1102 Poster Session (Board #182a), Sun, 8:00 AM-11:00 AM TPS1103 Poster Session (Board #182b), Sun, 8:00 AM-11:00 AM
A phase III, multicenter, randomized, open label trial of [fam-] trastuzumab XENERA-1: A phase II trial of xentuzumab (Xe) in combination with everolimus
deruxtecan (DS-8201a) versus investigator’s choice in HER2-low breast (Ev) and exemestane (Ex) in patients with hormone receptor-positive (HR+)/
cancer. First Author: Shanu Modi, Memorial Sloan Kettering Cancer Center, human epidermal growth factor receptor 2-negative (HER2-) metastatic breast
New York, NY cancer (mBC) and non-visceral involvement. First Author: Peter Schmid,
Centre for Experimental Cancer Medicine, Barts Cancer Institute, Queen Mary
Background: Human epidermal growth factor receptor 2 (HER2)-targeted
University of London, London, United Kingdom
therapies have greatly improved survival in HER2+ breast cancer (BC).
However, there are many more BC patients with HER2-low expression Background: The mTOR inhibitor Ev, combined with Ex, is a mainstay in the
(immunohistochemistry [IHC] 1+ or 2+/ in situ hybridization-negative), for treatment of post-menopausal women with advanced HR+/HER2- BC.
whom no HER2-targeted therapies are approved. [Fam-] trastuzumab der- However, the activity of Ev is limited by counter-regulatory feedback mech-
uxtecan (T-DXd; formerly DS-8201a) is an antibody-drug conjugate with a anisms in cancer cells, involving reactivation of insulin-like growth factor (IGF)/
humanized HER2 antibody, peptide-based cleavable linker, and a topo- mTOR survival signaling. Combining Ev with the humanized IGF-1 and IGF-2
isomerase I inhibitor payload. It has a drug-to-antibody ratio of ~8 and the ligand-blocking antibody Xe abrogates this feedback, thus intensifying in-
membrane permeability of the cleaved payload enables a cytotoxic bystander hibition of tumor growth; this leads to a pronounced effect in patients with non-
effect. In an ongoing phase 1 trial, T-DXd demonstrated an objective re- visceral (e.g., bone and lymph node) metastases. The phase II XENERA-1 trial
sponse rate (ORR) of 44.2% (19/43) with a manageable safety profile in evaluates the combination of Xe with Ev and Ex in post-menopausal women
heavily pretreated, advanced HER2-low BC (Oct 2018 data cutoff; Modi with HR+/HER2- BC. Methods: XENERA-1 (NCT03659136) is a double-
et al, SABCS 2018). Methods: DESTINY-Breast04 is a randomized, phase blind, placebo-controlled, randomized study to assess the efficacy and
3, 2-arm, open-label, multicenter study comparing efficacy and safety of safety of Xe in combination with Ev and Ex, in post-menopausal women with
T-DXd vs physician’s choice in HER2-low, unresectable and/or metastatic HR+/HER2- locally advanced/mBC and non-visceral disease. The population
BC. Approximately 540 subjects will be randomized 2:1 to T-DXd (5.4 mg/kg comprises post-menopausal mBC patients who have progressed on #1 pre-
intravenous every 3 weeks) or physician’s choice (capecitabine, eribulin, vious line of a non-steroidal aromatase inhibitor, with or without a CDK 4/6
gemcitabine, paclitaxel, or nab-paclitaxel). Randomization is stratified by inhibitor, who may have received fulvestrant. Other inclusion criteria are: an
HER2 IHC status, number of prior treatments, and prior CDK4/6 inhibitor Eastern Cooperative Oncology Group Performance Status of 0 or 1; adequate
therapy/hormone receptor (HR) status. HER2 IHC will be assessed and organ function; and non-visceral disease (absence of brain, liver, lung, peri-
confirmed by central testing based on archival samples. The primary efficacy toneal or pleural metastases). Patients are randomized (1:1) to receive Xe
endpoint is progression-free survival (PFS) per mRECIST v1.1 determined by (1000 mg/week, iv) or placebo (weekly, iv), in combination with Ev (10 mg/day)
blinded independent central review. Secondary efficacy endpoints are and Ex (25 mg/day). Treatment will continue until disease progression, un-
investigator-assessed PFS, overall survival, confirmed ORR, and duration of acceptable toxicity or other reasons. The primary endpoint is progression-free
response. Safety endpoints include serious adverse events (AEs), treatment- survival according to RECIST 1.1 criteria, by independent review. Secondary
emergent AEs, and AEs of special interest (interstitial lung disease/ endpoints are: overall survival; disease control; duration of disease control;
pneumonitis, cardiotoxicity, and infusion-related reactions). Primary PFS objective response; and time to progression of pain/intensification of palliation.
analysis will occur when 318 events are observed in HR+ subjects; providing Safety and pharmacokinetic endpoints, and exploratory biomarkers will also be
90% power to detect a hazard ratio of 0.68 with 1-sided a of 0.025. En- evaluated. The first patient was enrolled in January 2019; target enrollment is
rollment began in Dec 2018. Clinical trial information: NCT03734029. 80 patients in 12 countries. Clinical trial information: NCT03659136.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
60s Breast Cancer—Metastatic

TPS1104 Poster Session (Board #183a), Sun, 8:00 AM-11:00 AM TPS1105 Poster Session (Board #183b), Sun, 8:00 AM-11:00 AM
EMERALD: A randomized, open label, phase III trial to evaluate the efficacy Phase 1/2 dose-escalation and expansion study investigating SAR439859
and safety of elacestrant (RAD1901) versus investigator’s choice (IC) of +/- palbociclib in postmenopausal women with estrogen receptor-positive
endocrine therapy (ET) for ER+/HER2- advanced breast cancer (BC) follow- (ER+)/HER2- metastatic breast cancer. First Author: Aditya Bardia,
ing CDK4/6 inhibitor (CDK4/6i) therapy. First Author: Aditya Bardia, Massachusetts General Hospital Cancer Center, Harvard Medical School,
Massachusetts General Hospital Cancer Center, Harvard Medical School, Boston, MA
Boston, MA
Background: Endocrine therapy +/- cyclin-dependent kinase 4/6 inhibitors,
Background: Estrogen receptor-positive (ER+) BC comprises ~70% of all BC such as palbociclib, is the standard of care for ER+/HER2- breast cancer.
and advanced/metastatic ER+ disease (mBC) remains a major clinical Tumors often become resistant to this combination but retain ER signaling
challenge. The addition of CDK4/6i to ET has improved progression-free dependence, allowing for sequential ER-directed therapy. Unlike other
survival (PFS); however, novel treatments are needed after progression. classes of endocrine therapy with one mode of action, selective ER degraders
Putative mechanisms of endocrine resistance, such as ESR1 mutations (SERDs) block signaling by both ER competitive antagonism and degradation,
(mESR1), also indicate the need for additional therapies. Elacestrant, an targeting resistance settings that other treatments cannot. SAR439859 is a
oral selective estrogen receptor degrader (SERD), demonstrated anti-tumor potent, oral SERD with improved preclinical efficacy and pharmaceutical
activity in preclinical models of ER+ BC, including models resistant to CDK4/ properties vs other SERDs. This study investigates SAR439859 +/- palbociclib
6i and models with mESR1. An interim evaluation of a phase 1 trial in postmenopausal women with ER+/HER2- metastatic breast cancer.
(NCT02338349) of elacestrant in heavily pretreated patients (pts) with Methods: This prospective, open-label, non-randomized Phase 1/2 study
mBC, demonstrated an overall response rate (ORR) of 27% with a PFS of 5.4 (NCT03284957; TED14856) assesses SAR439859 single agent at dose
mo (Bardia, SABCS, 2017). Responses were seen in pts with prior CDK4/6i levels increasing from 20 mg/day up to the maximum administered dose (Part
and with wild-type (WT) or mESR1. Methods: This is a multicenter, in- A) followed by cohort expansion at the recommended dose (RD; Part B). The
ternational, randomized, open-label, active-controlled phase 3 trial for post- study will also assess two dose levels of SAR439859, in combination with
menopausal women or men with mBC. Pts must have received 1-2 prior lines palbociclib 125 mg/day (Days 1–21 in 28-day cycles; Part C) followed by
of ET, #1 line of chemotherapy for mBC, and have documented progression cohort expansion (Part D). Postmenopausal women with ER+/HER2- meta-
on a CDK4/6i. Pts with measurable disease (RECIST v1.1) or bone-only static breast cancer, who received $ 6 months of prior endocrine therapy, are
disease are eligible. Pts are randomized 1:1 to elacestrant (400 mg orally eligible. Patients were permitted to have received # 3 (Part A) or # 1 (Parts
daily) or IC of fulvestrant or an aromatase inhibitor. Stratification factors B–D) prior chemotherapies for metastatic disease. Exclusion criteria include
include ESR1 mutation status (detected by ctDNA), prior fulvestrant Eastern Cooperative Oncology Group performance status $ 2, concomitant
treatment and presence of visceral disease. The primary endpoints are PFS illness (including those related to HIV or hepatitis and other cancers # 3 years)
by blinded independent review committee (IRC) in pts with mESR1 and in all and factors potentially affecting absorption of SAR439859 or palbociclib.
pts (WT or mESR1). Secondary endpoints include: overall survival; PFS by Study endpoints include assessment of dose-limiting toxicities, determination
IRC in WT, PFS by investigator review; ORR, duration of response, and of maximum tolerated dose and RD in dose escalation (Parts A and C), and
clinical benefit rate; safety; pharmacokinetics; and quality of life. Approx- objective response rate according to RECIST v1.1 in dose expansion (Parts B
imately 466 pts will be enrolled to detect 340 PFS events in all pts and D). 18FES-PET scan between Days 11–15 in Part A will assess ER
(power $90%, hazard ratio (HR) = 0.667) and 160 PFS events in the availability. Safety, pharmacokinetics and response were evaluated for Parts
mESR1 subset (power $80%, HR = 0.610), overall a level at 2-sided 5% A–D. Recruitment and screening are ongoing (Part A n = 16; B n = 18; C n = 2;
using the Hochberg procedure. The EMERALD study is open for enrollment. D n = 0). Funding: Sanofi. Clinical trial information: NCT03284957.
Clinical trial information: NCT03778931.

TPS1106 Poster Session (Board #184a), Sun, 8:00 AM-11:00 AM TPS1107 Poster Session (Board #184b), Sun, 8:00 AM-11:00 AM
A phase II study of dual immune checkpoint blockade (ICB) plus androgen CONTESSA: A multinational, multicenter, randomized, phase III registration
receptor (AR) blockade to enhance thymic T-cell production and immuno- study of tesetaxel plus a reduced dose of capecitabine in patients (pts) with
therapy response in metastatic breast cancer (MBC). First Author: David B. HER2-, hormone receptor + (HR+) locally advanced or metastatic breast
Page, Earle A. Chiles Research Institute at the Robert W. Franz Cancer cancer (LA/MBC) who have previously received a taxane. First Author: Joyce
Center, Portland, OR O’Shaughnessy, Baylor University Medical Center, Texas Oncology, US
Oncology, Dallas, TX
Background: ICB (atezolizumab, anti-PD-L1) is known to improve survival
when added to chemo, however only in PD-L1-positive, triple-negative MBC. Background: Chemotherapy treatments with robust efficacy that preserve
ICB is less effective in hormone receptor positive (HR+) MBC, or when quality of life are needed. Tesetaxel (T) is a novel, oral taxane that has
administered following palliative chemo. Novel approaches are required to potential advantages over currently available taxanes, including: oral ad-
broaden clinical benefit of ICB, particularly in PD-L1-negative, HR+, or ministration with a low pill burden and Q3W dosing regimen; no observed
chemo-experienced MBC. Dual ICB with anti-PD-1 (nivolumab) and anti- hypersensitivity reactions; preclinical evidence of CNS penetration; and
CTLA-4 (ipilimumab) is associated with enhanced activity in melanoma improved activity against chemotherapy-resistant tumors. Over 600 pts have
other malignancies, but has not been explored extensively in MBC. Androgen been treated with T in clinical studies. T had robust monotherapy activity in a
receptor (AR) blockade, in addition to known direct cytostatic effects in AR- Phase 2 study in 38 pts with HER2-, HR+ MBC who received T Q3W, with a
expressing MBCs (50% of TNBC, . 75% of HR+ MBC), may also modulate confirmed ORR per RECIST 1.1 of 45% and median PFS of 5.4 mo. The
immune response. AR blockade has been shown experimentally to stimulate confirmed ORR in taxane-pretreated pts was 45%. Preclinical and clinical
thymic production of naı̈ve T-cell clones, which in turn can facilitate de novo studies suggest that reducing the dose of capecitabine (C) in combination
anti-tumor immune responses. Concurrent ICB can enhance the activity of with a taxane may result in reduced toxicity without reduction in efficacy.
these T-cell clones by interfering with PD-1-mediated peripheral tolerance. Preclinical data also suggest that T may penetrate the brain at clinically
This combination approach is promising in MBC in light of known AR relevant concentrations. CONTESSA investigates T plus a reduced dose of C
positivity, and the routine use of lymphodepleting chemo regimens in the as an all-oral regimen in HER2-, HR+ LA/MBC, with revised eligibility criteria
curative-intent setting. Methods: This is a phase II trial of dual immune to allow inclusion of pts with CNS metastases. Methods: CONTESSA is a
checkpoint blockade (nivolumab 240mg IV q2w; ipilimumab 1mg/kg IV 600-pt, multinational, multicenter, randomized (1:1), Phase 3 registration
q6w) plus AR blockade (bicalutamide, 150mg PO daily, dose reduction study comparing T (27 mg/m2 on Day 1 of a 21-day cycle) plus a reduced
allowed) in triple-negative MBC (cohort A: AR-positive [ . 1% by IHC]; dose of C (1,650 mg/m2/day on Days 1-14 of a 21-day cycle) to the approved
cohort B: AR-negative) or HR+ MBC (cohort C) in subjects who received 0/1 dose of C alone (2,500 mg/m2/day on Days 1-14 of a 21-day cycle) in pts with
prior chemotherapies in the non-curative setting. Objectives include 24- HER2-, HR+ LA/MBC previously treated with a taxane in the (neo)adjuvant
week clinical benefit rate by iRECIST (primary), safety (CTCAE v4.0), and setting. The protocol was newly amended to allow pts with known CNS
other response measures (RECIST1.1, PFS, OS). Efficacy for each cohort is metastases. The primary endpoint is PFS assessed by an Independent
defined as . 20% improvement in response over historical control (30% per Radiologic Review Committee (IRC). CONTESSA is 90% powered to detect a
EMBRACE clinical trial) employing a Simon 2-stage design to minimize 42% improvement in PFS (HR = 0.71). Secondary endpoints are OS, ORR,
futility (n = 46/cohort, stage I n = 15). Thymic generation of T-cells will be and disease control rate. Enrollment began in Dec 2017. Following review in
measured via quantitative deep sequencing of T-cell receptors (TcR, Jan 2019, the Independent Data Monitoring Committee recommended that
ImmunoSEQ assay) and TcR excision circles (TRECs), as well as real-time the Study continue as planned. Clinical trial information: NCT03326674.
flow cytometry using surrogate cell surface markers of recent thymic emi-
gration. Enrollment has commenced, sites: Earle A. Chiles Research In-
stitute (Portland, OR), Memorial Sloan Kettering Cancer Center (New York,
NY). Clinical trial information: NCT03650894.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Breast Cancer—Metastatic 61s

TPS1108 Poster Session (Board #185a), Sun, 8:00 AM-11:00 AM TPS1109 Poster Session (Board #185b), Sun, 8:00 AM-11:00 AM
Randomized phase II trial of venetoclax + fulvestrant versus fulvestrant in A multicentered randomized phase II comparison of single-agent carboplatin
estrogen receptor+, HER2– locally advanced or metastatic breast cancer versus the combination of carboplatin and everolimus for the treatment of
following recurrence or progression during or after a CDK4/6 inhibitor: advanced triple-negative breast cancer. First Author: Jami Aya Fukui,
VERONICA. First Author: Geoffrey J Lindeman, Department of Medical University of Hawaii Cancer Center, Honolulu, HI
Oncology, Peter MacCallum Cancer Centre, Melbourne, Australia
Background: Triple negative breast cancer (TNBC) is an aggressive disease
Background: CDK4/6 inhibitors (CDK4/6is) administered with endocrine with unmet clinical needs. Women with TNBC tend to be younger and
therapy have demonstrated improvements in progression-free survival (PFS) demonstrate early recurrence, higher histological grade, higher rate of vis-
for estrogen receptor (ER)+ advanced breast cancer (BC), but resistance ceral metastasis and increased mortality rates when compared to hormone
occurs, and new options are needed in the post-CDK4/6i setting. BCL2 is an positive breast cancer. Prognosis for metastatic TNBC is especially poor. Due
estrogen-responsive anti-apoptotic molecule overexpressed in 75% of BCs. to lack of targeted therapies, there is no standard treatment of choice for
The BCL2 inhibitor venetoclax (Ven) has shown improved outcomes and triple negative breast cancer and chemotherapy remains the accepted
tolerability in hematological malignancies such as chronic lymphocytic leu- standard. Many chemotherapeutic agents have been reported to have
kemia, and has been investigated in BC. A phase 1b study of Ven + tamoxifen clinical activity either as single agent or in combination. Seventy percent of
demonstrated safety and an efficacy signal in ER+, BCL2+ metastatic BC breast cancers with BRCA-1 germline mutations are triple negative, which
(mBC). Preclinical data for Ven + fulvestrant (Ful) have also shown synergy. suggests a shared carcinogenic pathway between them. In preoperative and
Based on these proof-of-principle data, the current study evaluates safety and metastatic settings, both TNBC and BRCA-1 associated breast cancers are
efficacy of Ven + Ful vs Ful in women with ER+, HER2– locally advanced (LA)/ particularly sensitive to DNA cross-linking agents such as platinum com-
mBC progressing after first- or second-line of prior therapy for metastatic pounds due to defective DNA repair by homologous recombination. The
disease, including $8 wks of a CDK4/6i. Methods: VERONICA is a global, recent TNT trial showed in patients with triple negative metastatic or re-
randomized, phase 2, multicenter, open-label study. Eligible patients (pts) are current locally advanced breast cancer with BRCA1/2 mutations, carboplatin
aged $18 yrs with confirmed ER+, HER2–, inoperable LA/mBC, $1 mea- resulted in a significantly higher overall response rate versus docetaxel (68%
surable lesion, tissue evaluable for BCL2, and ECOG performance status 0–1. versus 33.3%; p=0.03). Triple negative breast cancers are associated with a
Prior Ful or Ven, or prior chemotherapy for LA/mBC are prohibited. Stratified by high frequency of PTEN loss, which leads to mTOR activation. Moreover, it
BCL2 expression (low vs high) and number of prior lines of mBC therapy (1 vs has been reported that mTOR activation may confer resistance to platinum
2), pts are randomized 1:1 to Ven 800 mg PO daily + Ful 500 mg IM (cycle agents such as cisplatin, a phenomenon that may be reversible by the
1 days 1 and 15, and day 1 of each subsequent 28-day cycle) vs Ful 500 mg IM addition of an mTOR inhibitor such as everolimus. There are reports of
alone. Treatment continues until disease progression or intolerable toxicity. synergy between mTOR inhibitors and platinum compounds in pre-clinical
Primary endpoint is clinical benefit rate defined as complete/partial response + and clinical data. Methods: We have opened a multi-centered randomized
stable disease for $24 wks from randomization. Secondary efficacy endpoints phase II trial comparing carboplatin AUC 4 q 3 weeks vs carboplatin AUC 4 q
include PFS, objective response rate, duration of response, and overall sur- 3 weeks combined with daily 5 mg everolimus. 41 of planned 72 patients
vival. Safety, pharmacokinetic, biomarker (e.g. BCL2 and PI3K expression) from the Mount Sinai Health System have been enrolled and are randomized
and patient-reported outcome analyses will also be conducted. Currently, 21 of in a 2:1 allocation. The primary objective is to compare progression-free
the planned 100 pts have been enrolled; enrollment is ongoing. Clinical trial survival in patients treated with carboplatin+everolimus to patients treated
information: NCT03584009. with carboplatin alone. Patients may have had up to 3 prior regimens for
metastatic disease. Exploratory biomarker assessment is being done to
identify markers of response. Clinical trial information: NCT02531932.

TPS1110 Poster Session (Board #186a), Sun, 8:00 AM-11:00 AM TPS1111 Poster Session (Board #186b), Sun, 8:00 AM-11:00 AM
SGNLVA-002: Single-arm, open label phase Ib/II study of ladiratuzumab A phase II trial of olaparib and durvalumab in metastatic BRCA wild type
vedotin (LV) in combination with pembrolizumab for first-line treatment of triple-negative breast cancer. First Author: Zahi Ibrahim Mitri, Oregon Health
patients with unresectable locally advanced or metastatic triple-negative & Science University, Portland, OR
breast cancer. First Author: Hyo S. Han, H. Lee Moffitt Cancer Center &
Background: There is an urgent need to develop novel non chemotherapy
Research Institute, Tampa, FL
treatments for metastatic triple negative breast cancer (mTNBC) patients
Background: There are currently no curative treatments for patients with who otherwise have a poor prognosis. Immune checkpoint blockade (ICB)
metastatic triple-negative breast cancer (mTNBC), and prognosis for this and PARP inhibitors (PARPi) have independently shown promise for the
disease is very poor. Emerging treatment combinations of anti-programmed treatment of mTNBC, and the combination has shown early benefit in the
death ligand 1 (PD-L1) agents with chemotherapy have shown promise in MEDIOLA and TOPACIO trials. This trial looks to 1) evaluate the efficacy of
mTNBC. SGN-LIV1A, or ladiratuzumab vedotin (LV), is a novel investigational the combination of the PARPi olaparib and the PD-L1 inhibitor durvalumab,
humanized IgG1 antibody-drug conjugate (ADC) directed against LIV-1, which and 2) perform extensive multi-omics including protein based image ana-
is highly expressed in breast cancer cells. LV mediates delivery of monomethyl lytics (multiplex IHC, cyclic immunofluorescence) on serial biopsies to
auristatin E (MMAE), which drives antitumor activity through cytotoxic cell identify predictive biomarkers and resistance mechanisms. Methods: Trial
killing and induces immunogenic cell death (ICD). Preliminary results from an Design: This is a single-arm phase II study to assess the efficacy of the
ongoing phase 1 study of LV monotherapy has shown LV to be well tolerated combination of olaparib and durvalumab in BRCA-wildtype mTNBC. mTNBC
and to have encouraging antitumor activity in patients with mTNBC. Com- participants will undergo a pre-treatment biopsy, then will start a 4 week
bining LV and pembrolizumab may result in complementary, as well as syn- induction treatment with olaparib (300 mg PO BID). At the end of 4 weeks of
ergistic, activity through LV-induced ICD that creates a microenvironment single agent therapy, participants will undergo a repeat on-treatment biopsy,
favorable for enhanced anti-PD-L1 activity. Methods: This single-arm, open- following which durvalumab (1500 mg IV every 4 weeks) will be added to
label, phase 1b/2 study evaluates the safety and antitumor activity of LV in olaparib. Participants will also be offered an optional biopsy on progression.
combination with pembrolizumab as first-line therapy for patients with Endpoints: The primary endpoint of this study is overall response rate (ORR)
unresectable locally advanced or mTNBC (NCT03310957, 2017-002289- to olaparib and durvalumab therapy. Secondary efficacy endpoints include
35). Patients must have measureable disease per RECIST v1.1, an ECOG score clinical benefit rate, duration of response, progression-free, and overall
of 0 or 1, and no prior cytotoxic or anti-PD-L1 treatment for advanced disease. survival. The incidence and severity of on-treatment adverse events will be
This study has 2 parts that are enrolling sequentially: a dose-finding phase that collected per CTCAE 5.0. Statistical Methods: 28 participants are planned
starts at LV 2.5 mg/kg + pembrolizumab 200 mg intravenously every three for enrollment to this study. A 2-stage analysis will be performed using a
weeks, and a dose expansion phase. The primary objectives are to evaluate the Simon 2-stage Minimax design. The null (ICB alone) and alternative (ICB +
safety/tolerability and objective response rate of LV + pembrolizumab, and PARPi) hypotheses are: H0: p = 0.15 and Ha: p = 0.35. For the primary
identify the recommended phase 2 dose of LV. The secondary objectives are to endpoint, a total sample size of 28 participants will achieve 80% power to
assess duration of response, disease control rate, progression-free survival, and detect the ORR difference of 0.20 with one-sided type I error =0.05. The trial
overall survival. Additional objectives include assessing PD-L1 and LIV-1 will be terminated in stage I if 2 or less out of the first 15 participants
expression-response relationship. Study enrollment is ongoing in the US respond. If the trial goes on to the stage II, a total of 28 participants will be
and EU. Clinical trial information: NCT03310957. studied. If the total number responding is less than or equal to 7, the
combination is rejected. Current Enrollment: The study was activated on 1/7/
2019. To date, 3 out of 15 patients have been accrued to stage I of the study
Clinical trial information: NCT03801369.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
62s Breast Cancer—Metastatic

TPS1112 Poster Session (Board #187a), Sun, 8:00 AM-11:00 AM TPS1113 Poster Session (Board #187b), Sun, 8:00 AM-11:00 AM
VIOLETTE: A randomized phase II study to assess the DNA damage response Dora: A randomized phase II multicenter maintenance study of olaparib
inhibitors AZD6738 or AZD1775 in combination with olaparib (Ola) versus alone or olaparib in combination with durvalumab in platinum responsive
Ola monotherapy in patients (pts) with metastatic, triple-negative breast advanced triple-negative breast cancer (aTNBC). First Author: Sarah Sammons,
cancer (TNBC). First Author: Andrew Tutt, King’s College London School of Duke Cancer Institute, Durham, NC
Medicine, London, United Kingdom
Background: PARP inhibition (PARPi) with olaparib is approved in HER2-
Background: TNBC comprises »15% of invasive breast cancer cases and negative germline BRCA mutant (gBRCAm) metastatic breast cancer.
alterations in BRCA1/2 are associated with »5% of all BCs. Ola (a poly ADP- Maintenance PARPi in relapsed platinum-sensitive ovarian cancer improves
ribose polymerase inhibitor [PARPi]) is approved for treating pts with HER2- median PFS regardless of gBRCA mutation status. Preclinical work has
negative metastatic BC with a germline BRCA mutation (gBRCAm), dem- shown that platinum response strongly correlates with olaparib response in
onstrating an improvement in progression-free survival (PFS). Alterations in breast cancer models; hence, maintenance therapy trials are underway in
other non-BRCA1/2 homologous recombination repair (HRR)-related genes aTNBC. PARPi modulates immune responses and enhances immunoge-
(non-BRCA HRRm) may also confer sensitivity to Ola therapy in pts with nicity in many preclinical models. We hypothesize that olaparib either alone
TNBC. Ola, AZD1775 (a WEE1 checkpoint inhibitor) and AZD6738 (an or in combination with the PD-L1 inhibitor durvalumab will have clinical
ataxia telangiectasia and Rad3-related protein inhibitor) target DNA damage efficacy as maintenance therapy in aTNBC subjects who have responded to
repair and cell cycle regulation. Preclinical studies in TNBC models show platinum-based chemotherapy. Methods: DORA is a randomized, international,
synergistic antitumor effects of Ola+AZD1775 and Ola+AZD6738, vs Ola multicenter, phase II study designed to explore the efficacy of olaparib or olaparib
monotherapy supporting the clinical evaluation of these combinations. in combination with durvalumab as maintenance therapy in platinum-sensitive
Methods: VIOLETTE is a global, multicenter, open-label, phase II study aTNBC. 60 subjects will be enrolled following a minimum of 3 cycles of
(NCT03330847) randomising 1:1:1 450 pts with advanced TNBC to 3 treatment with platinum-based (cisplatin or carboplatin) chemotherapy
treatment arms: 1) Ola 200 mg bid daily + AZD1775 150 mg bid D1-3, D8- as a single agent or combination therapy in the first or second-line setting.
10 q21, 2) Ola 300 mg bid daily + AZD6738 160 mg qd D1-7 q28, or 3) Ola Subjects deriving clinical benefit (CR / PR / SD) from platinum-based
300 mg bid daily q28. All pts will be stratified by prior platinum exposure. therapy will be eligible and randomized in a 1:1 ratio. Patients in arm 1 will
Each treatment arm of »150 pts will be comprised of 3 biomarker strata of receive olaparib orally 300mg BID continuously and in arm 2 will receive
»50 pts each (A: BRCAm; B: non-BRCA HRRm; C: non-HRRm). Centralized olaparib orally 300mg BID continuously in combination with durvalumab
tumor molecular testing will be deployed to detect mutation(s) in 15 HRR 1500mg IV every 4 weeks. Assessment of tumor response will be done
genes. Eligible pts will have received 1-2 prior lines of chemotherapy for every 8 weeks. Primary endpoint: progression-free survival. Secondary
metastatic disease, including an anthracycline or taxane. Exclusion criteria endpoints: overall survival, clinical benefit rate, safety. Correlative ana-
include prior PARPi therapy. The primary endpoint is PFS (each combination lyses: pre-treatment archival/fresh biopsy samples are mandated. Post-
vs Ola alone) assessed by blinded, independent central review (RECIST treatment tissue biopsy is requested. Serial ctDNA will be collected at
v1.1). Secondary endpoints are objective response rate, duration of re- baseline, staging, and progression to correlate with response and track
sponse, change in tumor size, and overall survival for comparisons between emerging genomic alterations in a platinum sensitive cohort under the
combinations and for each combination vs Ola alone; drug exposure; and pressure of PARP inhibition. Whole exome DNA sequencing, IHC for PDL-1
safety and tolerability. The first prespecified futility analysis in Stratum C has and TILs will be performed on tissue samples. ClincalTrials.gov Identifier:
met the recruitment target and will be assessed by unblinded review April NCT03167619. (Moore K, et al "SOLO-1: Phase III trial of maintenance
2019. Clinical trial information: NCT03330847. olaparib following platinum-based chemotherapy in newly diagnosed
patients with advanced ovarian cancer and a BRCA1/2 mutation" ESMO
2018; Abstract LBA7-PR). Clinical trial information: NCT03167619.

TPS1114 Poster Session (Board #188a), Sun, 8:00 AM-11:00 AM TPS1115 Poster Session (Board #188b), Sun, 8:00 AM-11:00 AM
A randomized phase II study of nab-paclitaxel + durvalumab + neoantigen Nimbus: A phase II study of nivolumab plus ipilimumab in metastatic
vaccine versus nab-paclitaxel + durvalumab in metastatic triple-negative hypermutated HER2-negative breast cancer. First Author: Romualdo
breast cancer (mTNBC). First Author: Leonel Fernando Hernandez-Aya, Barroso-Sousa, Hospital Sı́rio-Libanês, Brası́lia, Brazil
Division of Medical Oncology, Department of Medicine, Washington University
Background: A previous study from our group showed that approximately 9%
School of Medicine, St. Louis, MO
of metastatic breast cancer (MBC) is hypermutated, defined as a tumor
Background: mTNBC is associated with poor outcomes and lacks targeted mutational burden (TMB) $10 Mutations/Megabase (Mut/Mb). The aim of
therapies. Immune modulation with PD-1/L1 inhibitors are emerging as this study is to evaluate if patients with hypermutated HER2-negative MBC
effective anticancer therapies. In mTNBC, atezolizumab (anti-PD-L1) plus benefit from the combination of nivolumab plus ipilimumab. Methods: This
nab-paclitaxel demonstrated an improvement in PFS compared to nab- is an open-label, single-arm, multicenter, phase 2 study assessing the efficacy
paclitaxel alone. Cancer vaccines targeting neoantigens may enhance the of nivolumab 3 mg/Kg intravenously (IV) every 14 days plus Ipilimumab 1 mg/
activity of immune checkpoint inhibition (ICI). Neoantigens are targets for Kg IV every 6 weeks in subjects with hypermutated metastatic HER2-negative
CD8 T-cells following ICI. T-cell responses to neoantigens are high in affinity breast cancer. Patients with measurable HER2-negative MBC, TMB $10 Mut/
and are not limited by central mechanisms of self-tolerance. Next-generation Mb assessed by a cancer-gene panel evaluating . 300 genes and performed
sequencing and epitope prediction algorithms are used to identify/prioritize in a CLIA-certified laboratory, and 0-3 prior lines of chemotherapy in the
neoantigens for vaccine design and development. Preclinical studies have advanced setting are eligible. The primary objective is overall response rate
shown that neoantigen vaccines are well tolerated and may be synergistic according to RECIST 1.1. Secondary objectives include the safety and
with anti-PD-1/L1 therapy. Methods: Eligible mTNBC patients are ran- tolerability of the combination, progression-free survival, and overall sur-
domized to either Arm-1 (nab-paclitaxel + durvalumab + neoantigen vac- vival. The study will follow a two-stage design. In the first stage 14 patients
cine) or Arm-2 (nab-paclitaxel + durvalumab). Initially, all participants are will be enrolled. If there is at least one patient with objective response,
treated with a run-in of gemcitabine + carboplatin (18-weeks; Part A). During accrual will continue to the second stage where an additional 16 patients will
this time sequencing and neoantigen vaccine production is performed. be enrolled. If there are at least 4 patients with an objective response among
Subsequently, patients are treated with nab-paclitaxel + durvalumab + neo- the 30 patients, the regimen will be considered worthy of further study. If the
antigen vaccine vs. nab-paclitaxel + durvalumab (Part B). The neoantigen true response rate is 5%, the chance the regimen is declared worthy of
vaccine is given subcutaneously. Participants in Arm-1 receive vaccinations on further study is less than 5%. If the true response rate is 25%, the chance
Days 1, 4, 8, 15, 22, 50 and 78. Durvalumab is administered at 1500 mg IV that the regimen is declared worthy of further study is . 90%. Tumor bi-
every 4 weeks. Nab-paclitaxel is administered at 100 mg/m2 IV on Days 1, 8, opsies, peripheral blood, and stool collection are mandatory and will be
and 15 of each 28-day cycle. Key eligibility criteria include patients with newly obtained at baseline, on treatment (end of cycle 1), and at disease pro-
diagnosed mTNBC; measurable disease; and tumor accessible for biopsy. The gression and will be assessed for potential biomarkers of treatment response.
primary endpoint is PFS defined as time from the initiation of Part B to The trial was activated in February 2019, and accrual should be completed
progression or death. Secondary endpoints include safety, objective response in 18 months. Clinical trial information: NCT03789110.
rate, clinical benefit rate and OS. The exploratory endpoints include evaluating
the immune response induced by the neoantigen vaccine, investigating bio-
markers of response including TILs, PD-L1, and immune signature by gene
expression, and mutational landscape. This trial is currently recruiting patients
(NCT03606967).

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Breast Cancer—Metastatic 63s

TPS1116 Poster Session (Board #189a), Sun, 8:00 AM-11:00 AM TPS1117 Poster Session (Board #189b), Sun, 8:00 AM-11:00 AM
KX-ORAX-001: An open label, randomized, multicenter, phase III registrational NRG-BR002: A phase IIR/III trial of standard of care therapy with or without
study to determine the safety, tolerability, and tumor response of oraxol stereotactic body radiotherapy (SBRT) and/or surgical ablation for newly
(HM30181A + oral paclitaxel) and its comparability to IV paclitaxel in patients oligometastatic breast cancer (NCT02364557). First Author: Steven J.
with metastatic breast cancer (MBC). First Author: Gerardo Antonio Chmura, The University of Chicago Medicine, Chicago, IL
Umanzor Funez, DEMEDICA, San Pedro Sula, Honduras
Background: This is a randomized Phase II/III trial to evaluate if stereotactic
Background: Paclitaxel is used in multiple cancer types including the treat- body radiotherapy (SBRT) and/or surgical resection (SR) of all metastatic
ment of breast cancer after failure of combination chemotherapy for metastatic sites in newly oligo-metastatic breast cancer who have received up to
disease or relapse within six months of adjuvant chemotherapy. Due to poor 12 months of first line systemic therapy without progression will significantly
oral absorption, paclitaxel is administered IV and is associated with extra improve median progression free survival (PFS). If this aim is met the trial
administration costs, inconvenience, burden for the patient, and hypersen- continues as a phase III to evaluate if SBRT/SR improves 5 year overall
sitivity reactions to the solubilizing agent Cremophor EL. HM30181A is an survival. Secondary aims include local control in the metastatic site, new
intestinal P-gp pump blocker that, when administered orally with paclitaxel distant metastatic rate, and technical quality. Translational primary end-
capsules (Oraxol), enhances paclitaxel absorption. Interim data from a phase I point is to determine whether , 5 CTCs is an independent prognostic marker
crossover study showed Oraxol (15-mg HM30181A + 205 mg/m2) 3 days/week for improved PFS and OS. Methods: Women with pathologically confirmed
had similar AUC0-‘ to IV paclitaxel 80 mg/m2 over 1 hour (Oraxol 5078 ng*hr/ metastatic breast cancer to # 4 sites who have been diagnosed within
mL vs. IV paclitaxel 5652 ng*hr/mL), however the Cmax was substantially 365 days with metastatic disease and the primary tumor site disease is
lower with Oraxol which may result in lower incidence of neuropathy. In a phase controlled. CNS metastases are ineligible. ER/PR and HER-2 neu status is
II single arm study of Oraxol in MBC 45.8% and 41.7% of subjects had PR or required. Site radiation credentialing with a facility questionnaire and pre-
SD according to RECIST 1.1. Methods: KX-ORAX-001 is a, multi-national treatment review of first case is required. Randomization is to standard
open-label, randomized (2:1 Oraxol to IV paclitaxel) phase III registration study systemic therapy with local radiotherapy/ surgery for palliation when nec-
comparing Oraxol (15-mg HM30181A + paclitaxel 205 mg/m2 daily x 3 days essary versus ablative therapy of all metastases with SBRT and/or SR. For the
QW) to IV paclitaxel 175 mg/m2 over 3 hours every third week, in female phase IIR portion to detect a signal for improved median PFS from 10.5 months
patients with histologically- or cytologically-confirmed MBC for whom treat- to 19 months with 95% power and a 1-sided alpha of 0.15 and accounting for
ment with IV paclitaxel monotherapy has been recommended. The study is ineligible/lost patients, 128 patients will be required. For the Phase III, an
powered to demonstrate the superiority of Oraxol on confirmed tumor response additional 232 patients will be required to definitively determine if ablative
rate vs IV Paclitaxel 175mg/m2 q3weeks. The primary efficacy endpoint is therapy improves 5-year overall survival from 28% to 42.5% (HR=0.67), with
confirmed tumor response according to a blinded central radiologist using 85% power and a one-sided type I error of 0.025. For the translational research
RECIST 1.1 criteria. The study is designed to enroll approximately 360 assuming a two-sided probability of type I error of 0.05, the number of patients
evaluable subjects. Two interim DSMB reviews were conducted at approxi- accrued in the Phase II-R and Phase III portions will provide sufficient power of
mately 90 and 180 evaluable subjects for safety, futility, and efficacy. En- at least 91% and 93% to detect whether , 5 CTC’s is prognostic for PFS and
rollment was initiated Dec 2015 at 45 sites in Central and South America and OS, respectively. Present accrual (1-31-2019): 105. Contact Information:
enrollment has been completed. At final analysis, if there is an approximate Protocol: CTSU member web site https://www.ctsu.org. Enrollment: OPEN at
difference of $10% favoring Oraxol, a p-value of 0.045 [2-tailed] will be https://open.ctsu.org. Support: This project is supported by NRG Oncology
achieved. Secondary endpoints are PFS and OS. Clinical trial information: grants U10CA180868 and U10CA180822 from the National Cancer Institute
NCT02594371. (NCI). Translational science is supported by the Ludwig Foundation for Cancer
Research. Clinical trial information: NCT02364557.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
64s Cancer Prevention, Hereditary Genetics, and Epidemiology

1500 Oral Abstract Session, Mon, 9:45 AM-12:45 PM 1501 Oral Abstract Session, Mon, 9:45 AM-12:45 PM
Effect modifiers in a randomized phase III trial of low-dose tamoxifen in Impact of a breast cancer (BC) polygenic risk score (PRS) on the decision to
breast preinvasive disease. First Author: Andrea De Censi, Division of take preventive endocrine therapy (ET): The Genetic Risk Estimate (GENRE)
Medical Oncology, E.O. Galliera Hospital, Genoa, Italy trial. First Author: Julian Oliver Kim, Dept of Radiation Oncology, CancerCare
Manitoba, Winnipeg, MB, Canada
Background: Low-dose tamoxifen (babytam) at 5 mg/day for 3 years de-
creases local or contralateral recurrence by 52% in women with hormone Background: Despite BC risk reduction of 50-65% by preventive endocrine
sensitive breast pre-invasive neoplasia after surgery (DeCensi et al JCO therapy (ET), very few at-risk women choose to take them. A woman’s
2019). Here we report the results of exploratory analyses to assess whether perceived BC risk correlates with uptake of ET. A PRS comprised of 77 BC
the benefit of babytam varies among subgroups of patients defined by in- genetic susceptibility loci (Single Nucleotide Polymorphisms (SNP) im-
dividual characteristics. Methods: Post-hoc subgroup analyses were per- proves the accuracy of risk prediction for BC. We examined the impact of the
formed according to a mixed approach based on the test for interaction and addition of individualized PRS BC risk prediction to standard risk calculator
biological plausibility. Incidence of invasive breast cancer or DCIS was the estimates on intent to take BC prevention medication. Methods: Eligible
primary endpoint. HRs were estimated using Cox proportional-hazards women had $5% 10 yr BC Tyrer-Cuzick risk (IBIS) or 5 year Gail score $3%,
modeling. Results: Age at menopause, smoking status and Ki-67 with no history of BC or hereditary BC syndrome. Standard BC risk estimates
exhibited a significant interaction with treatment. Specifically, the effect (IBIS or Gail) were incorporated into the counselling on BC preventive ET. A
of babytam was greater in women aged . 50y (n = 293, HR = 0.27, 95%CI: self-reported questionnaire at baseline quantified intention to take ET and
0.10-0.73) than in women aged #50y (n = 207, HR = 0.86, 0.35-2.07), explored factors associated with this decision. Blood samples were obtained
p-interaction = .09. Never smokers (n = 307) had a greater benefit than and genotyped for 77 SNPs, individualized PRS were calculated then in-
former (n = 68) or current smokers (n = 97): HR = 0.28, 0.11-0.70 vs HR = corporated into IBIS and Gail predictions for 5 yr, 10 yr, & lifetime BC risk.
0.57, 0.09-3.45 vs HR = 1.51, 0.41-5.64, respectively (p = .05). Tumors At a second visit, PRS risk & prevention recommendations were revisited.
with Ki-67 above the median level of 10% (n = 133) had a greater effect Post visit questionnaires assessed change in intent to take ET. Multivariable
(HR = 0.27, 0.09-0.81) than Ki-67 #10% (n = 145, HR = 1.58, 0.45-5.60, linear regression was performed to assess impact of baseline variables on
p = .04). Weaker statistical interactions (p . .1) were also found for waist change in intent to take medication. Results: From 2016 to 2017, 151
circumference and hot flashes (HF) at baseline. Women with waist women in Canada & USA were enrolled, median age: 56.1 (range 36-76.4),
circumference $89 cm (metabolic syndrome, n = 208) had a greater effect 35.6% were premenopausal, 98.7% were Caucasian. Median 5yr, 10yr, &
(HR = 0.22, 0.07-0.78) than women , 89 cm (n = 228, HR = 0.61, 0.25- lifetime IBIS risk estimates were 3.8% (2.0-11.5), 7.9% (5.0-23.1), and
1.46). Compared with placebo and no HF, babytam effect was stronger in 25.3% (5.5 to 92.2). PRS increased BC risk estimates in 84 (55.6%) and
women with HF (HR = 0.13, 0.02-0.96) than in women on babytam and no reduced BC risk estimates in 67 (44%) women. After PRS risk counselling,
HF (HR = 0.50, 0.24-1.03) or placebo and HF (HR = 0.72, 0.31-1.69, log- intention to take ET significantly changed (p,0.001): 41.9% of those with
rank p-trend = .004). Additional subgroups according to obesity, family history increased PRS were more inclined, and 46.7% of women with decreased
of breast or ovarian cancer, alcohol use, extent of surgery, radiotherapy for PRS were less inclined to take ET. On multivariable regression, increase in
DCIS, ER and HER2 expression, positive margins and treatment compliance PRS (p,0.0001) and less concern about ET side effects (p,0.0001) were
showed no significant heterogeneity of treatment. Conclusions: Exploratory associated with greater intent to take ET. Conclusions: In high risk women,
analyses showed a trend to a higher effect of babytam in women aged 50 or PRS significantly changed BC risk estimates & intent to take preventive ET.
older, never smokers, women with hot flashes or abdominal obesity and Further assessments of the impact of PRS scores on compliance with ET are
tumors with Ki-67 above 10%. Our results provide insight into the efficacy of warranted. Clinical trial information: NCT02517593.
babytam towards a personalized preventive approach. Clinical trial in-
formation: NCT01357772.

1502 Oral Abstract Session, Mon, 9:45 AM-12:45 PM 1503 Oral Abstract Session, Mon, 9:45 AM-12:45 PM
Outcomes after adjuvant radiotherapy in breast cancer patients with and Risk of subsequent cancer diagnosis in patients treated with 3D conformal,
without germline mutations: A large, single-institutional experience. First intensity modulated, or proton beam radiation therapy. First Author:
Author: Bhavana Sree Vangara Chapman, The University of Texas MD Michael H. Xiang, Department of Radiation Oncology, Stanford University,
Anderson Cancer Center, Houston, TX Stanford, CA
Background: Women with germline mutations in DNA repair pathways are at Background: Approximately half of cancer patients receive radiation therapy
an increased risk of developing breast cancer. We posit that tumors arising in (RT). Modern RT modalities (3D conformal [3DCRT], intensity modulated
these patients may be more sensitive to radiotherapy and, therefore, patients [IMRT], proton beam [PBRT]) have been theorized to pose different risks of
may experience improved locoregional control and survival outcomes following second cancers, but the relationship between RT modality and subsequent
adjuvant radiotherapy as compared to patients without DNA repair pathway cancers has been unclear due to their rarity. Methods: Pediatric and adult
mutations. Methods: We evaluated the records of 2,221 women with stage patients with a first cancer diagnosis who received 3DCRT, IMRT, or PBRT
0-III de novo primary breast cancer treated with surgery and adjuvant radio- were identified in the National Cancer Database. To analyze a more uniform
therapy who all underwent genetic testing at our institution from 1993 to population, cases were required to be non-metastatic and have at least
2018. Mutations were categorized as pathogenic variant, variant of unknown 2 years of follow-up time. Ten cancer types were included: head/neck, upper
significance (VUS), or negative. The Kaplan-Meier method was used to esti- gastrointestinal (GI), lower GI, gynecological, lymphoma, non-small cell
mate the locoregional recurrence rate (LRR), rate of distant metastasis (DM), lung, prostate, breast, bone/soft tissue, and brain/central nervous system.
disease-free survival (DSS), and overall survival (OS) from the time of surgery. Diagnosis of a subsequent cancer was determined using a variable denoting
Results: The median age at diagnosis was 45 years (range 19-84). Median the sequence of malignant neoplasms over the lifetime of the patient. The
follow-up time was 7 years (95% confidence interval 6.6-7.4). Among 1,960 risk of subsequent cancer diagnosis was modeled using multivariable logistic
patients with evaluable radiation records, 752 (38.4%) received breast only regression adjusting for age, follow-up time, cancer type, RT dose, che-
radiation, 12 (0.6%) received chest wall only radiation, and 1,196 (61.0%) motherapy, and other factors. Propensity score matching was additionally
received breast/chest wall and regional nodal radiation. A total of 255 (11.4%) used to balance baseline characteristics. Results: In total, 430,866 patients
and 162 (7.3%) patients had a pathogenic variant mutation and a VUS only, were included (33.4% 3DCRT, 65.1% IMRT, 1.5% PBRT) with median
respectively. Pathogenic variant and VUS in BRCA1/2 mutations were detected follow-up of 5.0 years and total follow-up period of 2.35 million person-
in 216 (9.7%) and 82 (3.7%) patients, respectively. Perturbations in ATM, years. In the comparison of IMRT relative to 3DCRT, there was no difference
CHEK2, MLH, MSH2/6, MUTYH, PALB2, RAD50/51, and/or TP53 were in the risk of subsequent cancer diagnosis (adjusted odds ratio [OR] 1.01;
detected in 71% (85/119) of patients who tested positive for a non-BRCA1/2 95% confidence interval [CI] 0.98-1.03; p = 0.62). In contrast, recipients of
pathogenic variant or VUS. On univariate analysis, there was no significant PBRT had significantly lower risk of subsequent cancer diagnosis relative to
association between BRCA1/2 mutation status or any genetic mutation and IMRT (adjusted OR 0.31; 95% CI 0.26-0.37; p , 0.0001). The benefit
rate of LRR or DM, DSS, or OS (p . 0.10 for all). Clinicopathological features associated with PBRT persisted in sensitivity analyses that excluded patients
including advanced stage and lymphovascular invasion were associated with with prostate cancer (71.6% of the PBRT cases), receipt of chemotherapy,
higher cumulative incidence of LRR and DM as well as shorter DFS and OS and/or follow-up time less than 5 years. Conclusions: Risk of subsequent
(p , 0.01 for all). Conclusions: Herein we report on the largest cohort of cancer diagnosis was similar between IMRT and 3DCRT and significantly
women with breast cancer treated with adjuvant radiotherapy at a single in- lower for PBRT. PBRT may be preferred in situations where avoidance of
stitution who have undergone germline testing. Our findings suggest that the second cancers is paramount, such as pediatrics and young adults.
overall prognosis of breast cancer treated with adjuvant radiotherapy in patients
with germline BRCA1/2 or other genetic predisposition is similar to patients
with sporadic breast cancer. Further investigation to evaluate acute or late
toxicities and secondary cancers as a result of radiotherapy is warranted.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Cancer Prevention, Hereditary Genetics, and Epidemiology 65s

1504 Oral Abstract Session, Mon, 9:45 AM-12:45 PM 1505 Oral Abstract Session, Mon, 9:45 AM-12:45 PM
Breast radiotherapy among ATM-mutation carriers. First Author: Leslie A. Relative risks of prostate cancer associated with different family cancer
Modlin, Memorial Sloan Kettering Cancer Center, New York, NY histories. First Author: Jennifer Lynn Beebe-Dimmer, Wayne State University
School of Medicine, Karmanos Cancer Institute, Detroit, MI
Background: Syndromes of DNA repair deficiency may confer both cancer
predisposition and increased sensitivity to DNA damaging agents, such as Background: Recently developed clinical guidelines have suggested that
ionizing radiation. Whereas homozygous deficiency of ATM causes the ataxia men in families with Hereditary Prostate Cancer (HPC), Hereditary Breast
telangiectasia syndrome, heterozygous ATM mutation carriers exhibit in- and Ovarian Cancer (HBOC), and Lynch Syndrome (LS) be referred for
creased rates of breast, pancreas and prostate cancers. ATM repairs DNA consideration of genetic testing, especially in the setting of aggressive
double-strand breaks; consequently, mutation carriers may exhibit excessive disease. However, while a family history (FH) of the same disease among
radiotherapeutic (RT) toxicity. We evaluated the tolerability of adjuvant close relatives is an established risk factor for prostate cancer (PC), a direct
breast radiation in ATM mutation carriers. Methods: We identified 167 ATM comparison of risk associated with specific FH, and particularly with respect
mutation carriers presenting to our institution with breast cancer; of these, to known familial cancer syndromes, in a single population is needed.
91 received RT and records were reviewed for RT-related morbidity. Tox- Methods: The Utah Population Database was used to identify 569,320 men,
icities were graded per CTCAE v5. Associations of clinicopathologic features 40+ years with a pedigree that included at least three consecutive gener-
with toxicity were evaluated by multivariate regression. Results: Of 91 ATM ations. Each man was evaluated for FH of FPC, HPC, HBOC and LS, as well as
mutation carriers receiving breast RT, 31% (n = 28) harbored a pathogenic their own PC status. PC cases (N=34,889) were identified from both the
mutation whereas 69% (n = 63) harbored variants of uncertain significance SEER Utah Cancer Registry and death certificates and classified into one or
(VUS). 71% (n = 65) underwent lumpectomy and adjuvant whole-breast RT; more subtypes: early-onset (EO [age of diagnosis ,60 years]), lethal, and/or
29% (n = 26) had mastectomy and PMRT. Nine patients underwent bilateral aggressive (Gleason Grade $7, metastatic, or lethal). Relative risks (RR)
RT for a total of 100 RT courses. 86% of RT courses comprised whole-breast/ associated with each PC subtype, adjusted for important covariates, were
chest-wall tangents; 14% were VMAT or protons. Median tangent dose was calculated in STATA using a modified Poisson regression with robust error
50Gy; 62% included an additional boost (median 10Gy) and 48% used a variances to obtain corresponding confidence intervals (CIs) for each FH
bolus (median thickness 0.3cm). Lymph nodes were treated in 43% (n = definition. Results: A FH of HPC conveyed the greatest relative risk for all PC
39). At last on-treatment visit, 31% had grade 2 dermatitis, 4% had other subtypes (RR=2.30; 95% CI 2.21-2.39), followed by HBOC and LS. Fur-
grade 2 events (fatigue, seroma, decreased range of motion, or pain), and 1% thermore, the strongest risks associated with FH were generally observed for
had grade 3 dermatitis. At 1 year, 13% had grade $2 toxicity: grade 2a EO disease. No differences in risk by degree of FH were observed for either
lymphedema (n = 3), and grade 2 (n = 1) or 3 (n = 2) capsular contracture. At lethal or aggressive disease. Conclusions: In this large population-based
last follow-up, 4 PMRT patients had capsular contracture (3 with grade 2, 1 family database, the risk of PC was shown to vary by cancer FH and was most
with grade 3); 1 patient had grade 2b lymphedema. Overall, no patients had strongly associated with EO disease. These results are critically valuable in
significant (grade $2) telangiectasias, fibrosis, or fat necrosis; no grade 4 or understanding and targeting high-risk populations that would benefit from
5 toxicities were seen. Neither pathogenic ATM mutations nor VUS were genetic screening and enhanced surveillance.
associated with acute or late RT toxicities. Conclusions: We found no evi-
Any PC EO PC Lethal PC Aggressive PC
dence of excess breast radiation toxicity among ATM mutation carriers, Family History RR (95% CI) RR (95% CI) RR (95% CI) RR (95% CI)
either pathogenic or VUS. Breast conserving therapy can be safely con-
HPC 2.30 (2.21-2.39) 3.88 (3.3-4.57) 2.21 (1.95-2.49) 2.31 (2.16-2.46)
sidered in this population. HBOC 1.46 (1.43-1.5) 2.05 (1.86-2.26) 1.39 (1.29-1.49) 1.47 (1.41-1.52)
LS 1.15 (1.12-1.19) 1.34 (1.18-1.53) 1.07 (0.98-1.17) 1.14 (1.09-1.2)

1506 Oral Abstract Session, Mon, 9:45 AM-12:45 PM 1507 Poster Discussion Session; Displayed in Poster Session (Board #1),
Ten-fold increase in genetic testing in pancreatic and metastatic prostate Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
cancer with implementation of point of care (POC) testing. First Author: Mon, 4:30 PM-6:00 PM
Heather Symecko, Basser Center for BRCA, University of Pennsylvania, Clonal hematopoiesis of indeterminate potential (CHIP) mutations in solid
Philadelphia, PA tumor malignancies. First Author: Madison Conces, Cleveland Clinic,
Cleveland, OH
Background: Germline genetic testing (GT) for cancer susceptibility is rec-
ommended for pancreatic and advanced prostate cancer patients, due to Background: CHIP predisposes to a higher risk of developing hematological
potential implications for targeted therapies and risk assessment of family malignancies and cardiac events. Multiple germline mutations have been
members. Traditional cancer GT programs may create barriers for certain recognized as contributing to CHIP, most notably ASXL1, DNMT3A, and
patient populations. To more effectively integrate testing into standard on- TET2. The frequency of CHIP mutations in solid tumor malignancies (STM)
cology care POC GT was introduced in early 2018 in a joint protocol is unknown. We report the frequency and incidence of CHIP mutations in
with Memorial Sloan Kettering Cancer Center. Here we report pre and post adult patients (pts) with STM. Methods: Data from 880 pts with STM who
POC referral and testing numbers at the University of Pennsylvania. underwent next generation sequencing (NGS) at Foundation One from 2013-
Methods: Patients with metastatic prostate or pancreatic cancer were ascer- 2017 was collected. This excluded two pts with known primary hemato-
tained through their GU/GI oncologist onto an IRB approved protocol and logical malignancies who were removed. A list of CHIP mutations using
shown an educational video about GT by research staff who obtained informed NCCN guidelines as well as numerous original research articles was created,
consent and facilitated biospecimen collection. Genetic counselors returned and tabulation of pathogenic or likely pathogenic mutations (ASXL1,
results and provided post-test counseling by phone. To evaluate the impact of DNMT3A, TET2, JAK2, SF3B1, TP53, GNAS, N/KRAS) was performed. To
this model on the uptake of GT services, the number of patients who were date, only the lung, breast, and colorectal cancer pts have been annotated
referred to and proceeded with GT was compared before and after study ini- (N = 372) since these cancers have an overall higher incidence and prev-
tiation. Results: In 2017, 77 patients were referred to genetics of which 45 alence in society. Results: Annotation of lung cancer pts (155/880), breast
underwent genetic counseling and testing. Twenty-nine (38%) did not com- (118/880), and colorectal cancer pts (99/880) is collected and represents
plete genetic counseling or testing, and 3 later underwent testing through the about 40% of all pts. At least one CHIP mutation was present in 44.5% (69/
POC study. Since the study launched in 2018, 407 patients were referred and 155) lung cancer pts, 32.2% (38/118) in breast cancer pts, and 7.1% (7/
underwent testing through the study. This represents a ten-fold increase in 99) in colorectal cancer pts. Most common mutations found were TP53 and
patients who underwent GT. Conclusions: Comparing uptake of GT services KRAS at 29.6% (110/372) and 28.0% (104/372), respectively. Mutations
before and after study initiation suggests that a POC model with abbreviated in genes not known to be somatic drivers for solid tumor malignancies,
pre-test education and post-test genetic counseling by phone is a possible particularly SF3B1, DNMT3A, and JAK2, were found at very low frequencies
solution to barriers of traditional genetic counseling, increasing physician 0.8% (3/372), 0.5% (2/372), and 0.3% (1/372), respectively. Notably,
referrals and uptake of testing by patients. This approach allows for more timely ASXL1 and TET2 mutations were not encountered in any pts. Conclusions: In
access to genetic information that may impact treatment strategies and tumor NGS testing, multiple CHIP mutations were noted to be present within the
medical management of family members. Clinical trial information: pending. cohort of lung, breast, and colorectal cancers. There is a need to further un-
Prostate 2017 Prostate 2018 Pancreas 2017 Pancreas 2018
derstand clinical consequences of CHIP mutations incidentally found in pts
with STM given known clinical implications of CHIP. We will report on clinical
Male (N) 29 165 19 128
Female (N) - - 29 114 data (comorbidities), response/non-response to therapy, and identify specific
6SD 59 6 8.0 6 9.2 67.5 6 11.0 6 10.2 molecular patterns of mutations to further understand the role of CHIP in pts
Positive (N/total tested) 3/10 12/147 3/35 17/223 with STM.
VUS (N/total tested) 2/10 22/147 10/35 44/223

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
66s Cancer Prevention, Hereditary Genetics, and Epidemiology

1508 Poster Discussion Session; Displayed in Poster Session (Board #2), 1509 Poster Discussion Session; Displayed in Poster Session (Board #3),
Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Mon, 4:30 PM-6:00 PM Mon, 4:30 PM-6:00 PM
Inherited DNA repair and cell cycle gene defects in chronic lymphocytic Effect of germline ATM mutations on clonal hematopoiesis. First Author:
leukemia. First Author: Nicholas S. Moore, Dana-Farber Cancer Institute, Thomas Paul Slavin, City of Hope, Duarte, CA
Boston, MA
Background: Clonal hematopoiesis (CH) in myeloid related-genes is asso-
Background: Chronic lymphocytic leukemia (CLL) is among the most heri- ciated with development of primary and secondary leukemia and athero-
table cancers, with 60% of disease risk genetically determined. However, sclerotic disease, as well as, decreased overall survival. Identification of
most of the genetic heritability of CLL remains unexplained. Previously, we factors beyond age and cytotoxic exposures that predispose to CH may be
identified ATM as the first CLL risk gene. Here, we leverage a deep-learning- useful to both recognize individuals at increased risk for CH and to better
based germline variant calling algorithm to explore germline mutational understand how CH develops. We have previously shown that germline
enrichment in DNA repair and cell cycle genes in CLL. Methods: A two-stage mutations in the DNA repair gene ATM may predispose to CH. We hy-
case-control analysis was conducted using gene-based mutational enrich- pothesized here that heterozygous ATM germline mutation carriers would
ment analysis of 50 established cancer predisposition DNA repair and cell have higher rates of CH in myeloid genes compared to controls.
cycle genes. In the discovery phase, a total of 285 Spanish patients and Methods: Germline DNA samples from 34 heterozygous ATM germline
5,608 ancestry-matched controls were evaluated. In the validation stage, an mutation carriers (cases) and 22 controls without ATM germline mutations
independent cohort of 514 European patients and 27,173 ancestry- were sequenced on an Illumina 2500 using a custom 79-gene-myeloid-CH-
matched controls were analyzed. An FDR correction was applied to both coding-exon-amplicon-based Qiaseq panel. Read depth averaged 130x.
datasets and genes with a q-value , 0.2 in both cohorts were considered Pathogenic and likely pathogenic CH variants (PV) above an allele fraction of
significant. Results: Our joint analysis of 799 CLL patients from 2 geneti- 2% were used for analyses. Cases and controls were compared using a rank-
cally distinct cohorts and 32,781 ancestry-matched cancer-free controls sum test. Results: Cases had a higher median age (56 years, range 30-82)
identified ATM and CHEK2 as significantly enriched in both CLL datasets. than controls (48 years, range 5-72). Cases and controls were similar in solid
First, our analysis recaptured the previously reported finding of ATM variant tumor cancer history and known exposure to cancer cytotoxic therapy;
enrichment in CLL patients. Carriers of pathogenic ATM mutations in our 73.5% vs 86.4%, and 18.1 vs 20.6%, respectively. The number of CH PV
cohorts (n = 9 patients, discovery: 1.05%, validation: 1.17%) were 2.8–3.7 was similarly associated with age in both cases and controls (cor = 0.31, p =
times more likely to develop CLL compared to cancer-free individuals 0.01). Cases displayed more CH PVs than controls (total 62 vs 3 PVs, median
(discovery: OR = 2.8, 95%CI = 0.7–9.0, q-value = 0.181; validation: OR = 2 PVs vs 0, p = 10-6). Of note, cases frequently had a concomitant second
3.7, 95%CI = 1.6–8.3, q-value = 0.0454). In addition, our analysis (n = 10; 29% of cases) or third (n = 4; 11.8% of cases) unique ATM CH PV,
identified 21 CLL patients carrying pathogenic CHEK2 alterations (dis- whereas no ATM CH PVs were seen in controls. Even after excluding ATM CH
covery: 1.40%, validation: 3.31%), making CLL patients 4.4-8.0 times PVs, CH PVs were more frequent in cases (p = 0.00003). After ATM CH PVs,
more likely to carry such alterations compared to controls (discovery: OR = the most frequent CH PVs in cases were in NF1 (5 PVs), BCORL1 (4 PVs),
8.0, 95%CI = 2.3–27.0, q-value = 0.026; validation: OR = 4.4, 95%CI = and DMNT3A (4 PVs). Conclusions: Our study supports ATM as a strong
2.5–7.3, q-value , 0.001). Conclusions: Our analysis of genetically distinct predisposition locus for myeloid gene CH. CH in ATM germline mutation
CLL cohorts, using a high-sensitivity variant calling algorithm, supports carriers frequently involved unique low allele fraction PVs in ATM, suggesting
CHEK2 as a potentially novel CLL predisposition gene that may explain a ATM germline PVs are driving production of likely bi-allelic ATM inactivation
portion of the missing monogenic heritability of CLL. In addition, this study in white blood cells, or complete ATM loss. Complete ATM loss may be a
highlights the DNA repair and cell cycle regulation pathways as potential nidus particularly for lymphocytic leukemia, as bi-allelic ATM inactivation
drivers of CLL susceptibility. is a frequent somatic finding.

1510 Poster Discussion Session; Displayed in Poster Session (Board #4), 1511 Poster Discussion Session; Displayed in Poster Session (Board #5),
Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Mon, 4:30 PM-6:00 PM Mon, 4:30 PM-6:00 PM
Antiviral therapy to reduce hepatocellular carcinoma recurrence in patients Characterization and clinical outcomes of mismatch repair deficient (dMMR)
with low HBV-DNA levels: A randomized controlled trial. First Author: Gang small bowel adenocarcinoma (SBA). First Author: Alicia Latham, Memorial
Huang, Eastern Hepatobiliary Surgery Hospital, National Innovation Alliance Sloan Kettering Cancer Center, New York, NY
for Hepatitis and Liver Cancer, Shanghai, China
Background: SBA is a rare cancer known to be associated with Lynch
Background: Despite antiviral treatment has been shown to reduce hepato- syndrome (LS). The prevalence, tumor characteristics, and clinical course of
cellular carcinoma (HCC) recurrence after curative treatment for hepatitis B SBAs in the setting of LS is not well understood. We sought to characterize
virus (HBV)-related HCC in patients with high preoperative HBV-DNA levels, SBA according to mismatch repair and germline mutation status, comparing
it is still unclear whether antiviral therapy is useful in reducing recurrence in clinical outcomes. Methods: A retrospective review of SBAs at a single in-
patients with low preoperative HBV-DNA levels. Methods: In this randomized stitution identified 74 SBAs that were assessed for dMMR either via im-
controlled trial, 200 patients who underwent curative resection for HCC with munohistochemical staining (IHC) or microsatellite instability status (MSI)
low baseline HBV-DNA levels were ran- domly assigned to receive preemptive using NGS. Germline DNA was analyzed for mutations in LS-associated
antiviral therapy or not. The primary endpoints were recurrence-free survival. mismatch repair genes (MLH1, MSH2, MSH6, PMS2, EPCAM) and when
This study was censored on March 31, 2015 when all surviving patients available, clinical records were reviewed. Results: Of 74 individuals with
had a minimum follow-up of 60 months. The analysis was done on an SBA, 28.4% (21/74) of tumors exhibited dMMR and/or high-frequency MSI
intention-to-treat basis. Results: The baseline clinical, laboratory, and tumor (MSI-H). The overall prevalence of LS in SBA was 9.5% (7/74), with all LS
characteristics of the 2 groups were comparable. The 1-, 3-, and 5-year patients having dMMR/MSI-H tumors. 33.3% (7/21) of dMMR/MSI-H SBA
recurrence-free survival rates for the antiviral group and the control group patients had LS. The distribution of germline mutations among LS patients
were 85.9%, 55.2%, and 52.0% and 80.6%, 40.9%, and 32.3%, re- was 57.1% (4/7) in MLH1, 28.6% (2/7) in MSH2, and 14.3% (1/7) in
spectively. The corresponding overall survival rates for the 2 groups were PMS2. One patient with an dMMR/MSI-H tumor was found to have a high-
94.0%, 75.7%, and 64.1% and 90.0%, 62.4%, and 43.7%, respectively. penetrance APC mutation, diagnostic of familial adenomatous polyposis
The recurrence-free survival and overall survival for the antiviral group were (FAP). In the 38 patients with available clinical follow-up, median age of
significantly better than the control group (P 1/4 0.016, P 1/4 0.004, re- onset was similar in the dMMR/MSI-H vs the pMMR/MSS group (62 vs 57,
spectively). After adjusting for confounding prognostic factors in a Cox p = 0.8). The prevalence of synchronous/metachronous cancers was 5.9%
model, the relative risks of recurrence and death for antiviral treatment were (1/17) in the pMMR/MSS group and 38% (8/21) in the dMMR/MSI-H group
0.601 [95% confidence interval (CI), 0.409– 0.884; P 1/4 0.010] and (p = 0.02), with 62.5% (5/8) of these in LS (p = 0.001; synchronous/
0.509 (95% CI, 0.333–0.778; P 1/4 0.002), respectively. Antiviral therapy metachronous in LS (5/7) vs. non-LS (4/31)). In the pMMR/MSS group,
was an independent protective factor of late tumor recur- rence (hazard ratio 58.8% (10/17) of patients presented with metastatic disease at diagnosis,
[HR] 1/4 0.316, 95% CI 0.157 – 0.637; P 1/4 0.001) but not of early tumor compared to 19% (4/21) in the dMMR/MSI-H group (p = 0.01). In dMMR/
recurrence (HR 1/4 0.782, 95% CI, 0.493–1.240; P 1/4 0.296). MSI-H SBA patients with early-stage disease, 11.8% (2/17) recurred,
Conclusions: In patients with low preoperative HBV-DNA levels, antiviral compared to 71.4% (5/7) in the pMMR/MSS group (p = 0.009).
therapy significantly reduced HCC recurrence after R0 hepatic resection. Conclusions: This preliminary evaluation suggests that SBA exhibiting
Clinical trial information: ChiCTR-IPR-15006587. dMMR/MSI-H status is more closely associated with early-stage disease and
lower recurrence rates, similar to prior observations in colon cancer. LS was
found in 9.5% of all SBA and in 33.3% of dMMR/MSI-H tumors, suggesting
that germline assessment for LS in SBA is warranted.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Cancer Prevention, Hereditary Genetics, and Epidemiology 67s

1512 Poster Discussion Session; Displayed in Poster Session (Board #6), 1513 Poster Discussion Session; Displayed in Poster Session (Board #7),
Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Mon, 4:30 PM-6:00 PM Mon, 4:30 PM-6:00 PM
New onset diabetes as a predictive factor of focal lesions in the pancreas in a Prevalence and penetrance of breast cancer-associated mutations identified
high-risk screening program. First Author: Maria Fernanda Montiel, University by multiple-gene sequencing in the Women’s Health Initiative. First Author:
of Texas MD Anderson Cancer Center, Houston, TX Allison W. Kurian, Stanford School of Medicine, Stanford, CA
Background: Genetic evolution studies have suggested the existence of a Background: Next-generation sequencing enables rapid analysis of many
window of opportunity to improve clinical outcomes by intercepting pre- inherited cancer susceptibility genes. Little is known about the prevalence
malignant lesions. This study reports the outcomes of Pancreatic Cancer and penetrance of pathogenic variants (PVs) in such genes among post-
(PC) surveillance in a high risk (HR) cohort followed at The University of Texas menopausal women with breast cancer, who comprise the majority of all
MD Anderson Cancer Center (MDA) between 2014 and 2018. Methods: The breast cancer patients. Methods: The Women’s Health Initiative enrolled
MDA PC High-Risk Clinic (MDA-PCHRC) performs surveillance based on risk post-menopausal women from 1993-1998. We conducted a nested case-
stratification. This study reports 54 months of surveillance. The patients were control study using banked DNA samples of 2,195 women who subsequently
stratified based on PC family history, personal history of other cancers, and developed invasive breast cancer (cases) and 2,322 cancer-free controls.
germline mutations (BRCA1/2, PALB2, STK11, CDKN2A, TP53) with high Sequenced genes were APC, ATM, BARD1, BMPR1A, BRCA1, BRCA2,
susceptibility to PC. Low risk patients were not offered surveillance while BRIP1, CDH1, CDK4, CDKN2A (p16INK4a and p14ARF), CHEK2, EPCAM,
patients in the moderate or high-risk category were enrolled into a program GREM1, MLH1, MSH2, MSH6, MUTYH, NBN, PALB2, PMS2, POLD1,
which included annual screening with blood markers (CA 19-9, fasting glucose, POLE, PTEN, RAD51C, RAD51D, SMAD4, STK11, and TP53. PV were
HbA1C, amylase, and lipase) and magnetic resonance imaging/ chol-
defined using American College of Medical Genetics criteria. PV prevalence
angiopancreatography (MRI/MRCP). High risk patients also had a baseline
is reported as proportions and penetrance as the odds ratio (OR) and 95%
endoscopic ultrasound (EUS). We evaluated the yield of premalignant lesions
confidence interval (CI) of PV versus none among cases versus controls.
detection and our main goal was to detect predictive factors of focal pancreatic
lesions and to validate our risk stratification strategy. Results: A total of 206
Results: Among cases, the median age at diagnosis was 73 years; 66% were
patients have been referred to our clinic during this time period. From this group, White, 18% Black, 6% Hispanic, 6% Asian and 4% other. The prevalence of
126 (61%) patients completed at least one cycle of baseline surveillance, for the PVs in any gene was significantly higher in cases (6.61%, 95% CI 5.57-
purposes of the analysis we only focus in the high risk (n=71) and moderate risk 7.65%) versus controls (4.09%, 95% CI 3.29-4.90%). The prevalence of
group (n=38). We have identified de novo pancreatic focal lesions in 22 patients, BRCA1/2 PVs was 1.2% in cases and 0.22% in controls. Among cases, the
20 from the high risk group (28%) and 2 from the moderate group (5%). Those prevalence of PVs in other breast cancer-risk genes was 2.3% (ATM, CDH1,
lesions included 7 patients with simple cysts, 9 with side-branch IPMN, 3 with BARD1, BRIP1, CHEK2, NBN, and PALB2 collectively), two-fold higher
main duct IPMN, 1 with pseudocyst, 1 with mucinous cyst and 1 with a solid than PVs in BRCA1/2. Prevalence of BRCA1/2 PVs decreased with age
nodule (pancreatic neuroendocrine tumor). We compared demographic in- among cases, while prevalence of ATM, CHEK2 and PALB2 PVs did not.
formation (age, gender, and ethnicity) as well as family and personal medical Statistically significant ORs for breast cancer penetrance were observed for
history between patients with focal pancreatic lesions vs negative or diffuse BRCA1 (5.43, 95% CI 1.19-51.52), BRCA2 (4.71, 95% CI 1.84-15.08),
findings. We found that new onset diabetes was significantly correlated with BARD1 (9.78, 95% CI 1.04-1295.87) and PALB2 (6.30, 95% CI 1.93-
presence of focal pancreatic lesions 5 (22%) of patients with focal lesions versus 31.94). Conclusions: Approximately 7% of women diagnosed with post-
patients without non focal lesion 2 (2%) (P=0.003). Conclusions: Screening at menopausal breast cancer carry a PV in a cancer susceptibility gene. In
the MDA-PCHRC detect pancreatic premalignant lesions in 20% of the patients contrast to studies of younger breast cancer patients, PVs in other breast
in our cohort. We validated our risk stratification methodology and found that cancer-related genes were two times more common than in BRCA1/2.
new-onset diabetes is predictive of pancreatic lesions, thus suggesting that this Results may guide genetic testing of women with post-menopausal breast
factor could be an important biomarker of focal lesions in a HR population. cancer.

1514 Poster Discussion Session; Displayed in Poster Session (Board #8), 1515 Poster Discussion Session; Displayed in Poster Session (Board #9),
Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Mon, 4:30 PM-6:00 PM Mon, 4:30 PM-6:00 PM
Racial and ethnic differences in the results of multigene panel testing of Adequacy of self-reported family history in electronic health record for
inherited cancer predisposition genes in breast cancer patients. First Author: genetic risk assessment for Lynch syndrome. First Author: Mala Pande,
Siddhartha Yadav, Mayo Clinic, Rochester, MN The University of Texas MD Anderson Cancer Center, Houston, TX
Background: The prevalence of germline mutations in non-white patients with Background: Self-reported family cancer history (FCH) is one of the key
breast cancer and the germline genetic drivers of breast cancer risk in non- indicators of hereditary cancer risk. Studies have shown that accurate FCH
white populations are largely unknown. Methods: The study population in- documentation by healthcare providers is suboptimal, but data regarding
cluded 77,900 women with breast cancer (Non-Hispanic white: 57,003; patient-provided FCH are limited. We evaluated the quality of FCH as entered
Black: 6,722; Asian: 4,183; Hispanic: 5,194; Ashkenazi-Jewish: 4,798) who by the patient into the electronic health record (EHR) to determine its
underwent germline multigene panel testing of cancer predisposition genes adequacy for Lynch syndrome (LS) risk assessment. Methods: At our tertiary
from March 2012 to December 2016. The prevalence of predisposition gene referral cancer center, FCH is self-reported via an online questionnaire sent
mutations in racial and ethnic populations relative to non-Hispanic Whites was prior to appointment, which is reviewed/updated by clinic nurse during
assessed while accounting for age at diagnosis of breast cancer, family history initial visit and then imported into EHR review of systems. Records of all new
of breast and ovarian cancer, and estrogen receptor status of breast tumors. patients from September 2016 to August 2017 were retrospectively
Associations between mutations in each gene and breast cancer risk were reviewed and analyzed. FCH quality was estimated by calculating rates of
evaluated using reference controls. Results: The overall frequency of patho- reporting of 3 FCH variables required for PREMM5, a risk-prediction model
genic mutations in known breast cancer predisposition genes was 9.1% for for LS. Parameters required for the model were sex, age, personal history of
non-Hispanic Whites, 9.8% for African Americans, 10.2% for Hispanics, cancer, and for FCH, degree of kinship (first/second degree), cancer site/
7.6% for Ashkenazi-Jewish, and 7.5% for Asians. BRCA1 mutations were type, and age at diagnosis. Results: Of 47,647 unique patients, 47.5%
enriched (p , 0.05) and CHEK2 mutations were under-represented in all reported FCH for 1 or more relative (46.1% were first degree, 64.8% second
racial and ethnic populations relative to non-Hispanic Whites. BRCA2 and degree, 3.0% other, and 2.4% missing). A cancer type/site was specified for
BARD1 mutations were enriched in African Americans and Hispanics relative 88.8% reporting FCH. Age at diagnosis was listed for 21.7% of the relatives’
to non-Hispanic Whites, whereas PALB2 and RAD51C mutations were cancers. Overall, only 20.9% provided all 3 FCH data elements required for
enriched in Hispanics. Among genes with mutation counts large enough for running PREMM5 (9.9% of the total sample, n=4738). Fewer men (9.5%)
assessment, mutations in BARD1, BRCA1, BRCA2, PALB2 and TP53 were than women (28.1%) provided all 3 FCH elements. Furthermore, 46.7% of
significantly associated with clinically relevant increased risks (odds ratio breast cancer patients, 21.9% of gastrointestinal cancer patients, 47.2%
(OR) . 2) of breast cancer across all ethnicities and races. Rates of variants of and 23.1% of patients seen for cancer prevention screening and endoscopy
uncertain significance were highest among Asians (29%), followed by blacks respectively, reported 3 FCH elements. Lower rates were observed for other
(27%), Hispanics (21%), non-Hispanic whites (16%) and Ashkenazi-Jews cancers. Conclusions: Patient self-reported FCH is suboptimal for estima-
(14%). Conclusions: While there is some similarity across ethnic groups, tion of LS risk and genetic counseling referral. Future steps to optimize
substantial heterogeneity exists in the prevalence of mutations in breast online patient-facing FCH collection to enable routine automated risk-
cancer predisposition genes across major racial and ethnic groups in the US assessment in an essentially provider-free setting may include, patient
population. These findings contribute to our understanding of breast cancer education regarding importance of FCH, EHR prompts for FCH completion,
risk and have significant implications for genetic testing, screening, and and implementation of algorithms in EHRs using FCH to identify patients at
management of patients with an inherited predisposition to breast cancer, risk for hereditary cancer predisposition syndromes.
with a need for continued analysis with increased cohort size in ethnic minority
groups.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
68s Cancer Prevention, Hereditary Genetics, and Epidemiology

1516 Poster Discussion Session; Displayed in Poster Session (Board #10), 1517 Poster Discussion Session; Displayed in Poster Session (Board #11),
Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Mon, 4:30 PM-6:00 PM Mon, 4:30 PM-6:00 PM
Genome-wide association study using whole-genome sequencing to Gliomas in the context of Li-Fraumeni syndrome: An international cohort.
identify a novel locus associated with cardiomyopathy risk in adult survivors First Author: Orli Michaeli, Hospital for Sick Children, Toronto, ON, Canada
of childhood cancer: Utility of a two-stage analytic approach. First Author:
Background: Li-Fraumeni syndrome (LFS) is a cancer predisposition syn-
Yadav Sapkota, St. Jude Children’s Research Hospital, Memphis, TN
drome associated with germline mutation in the TP53 tumor suppressor
Background: Survivors of childhood cancer are at increased risk of gene. As a result of increased awareness and surveillance imaging, more
treatment-related cardiomyopathy, found to be modified by genetic factors. asymptomatic low-grade brain lesions are being identified, raising important
To further investigate genetic risks of cardiomyopathy, we utilized whole- questions regarding the management of those patients. Sporadic low-grade
genome sequencing (WGS) in a clinically phenotyped cohort of long-term gliomas (LGG) in the pediatric age rarely transform to malignant lesions,
survivors of pediatric cancer. Methods: Utilizing a novel 2-stage analytic whereas the prognosis of high-grade gliomas (HGG) is grim in all age groups.
approach, we first performed association analysis for ejection fraction (EF) Although HGG is a hallmark of LFS, little is known of the natural history of
using WGS data in European-descent childhood cancer survivors from the St. these lesions in this syndrome. Methods: For this multi-institutional retro-
Jude Lifetime Cohort (SJLIFE). EF was analyzed as a continuous variable to spective study, anonymized clinicopathologic data from TP53 mutation
increase statistical power for genetic discovery. Common variants (minor carriers with gliomas were collected and analysed. Results: Our cohort in-
allele frequency (MAF) . 0.05) were analyzed using linear regression, cluded 61 patients, of whom 71% (n = 45) were children or young adults
adjusting for age at diagnosis, sex, age at follow-up, doses of anthracycline (age , 25 years). 39% of patients with known family history of cancer had a
and average radiation dose to the heart, and eigenvectors. Rare/low- close relative with a brain tumor. Of 31 patients with low grade lesions at
frequency variants were aggregated by different functional annotations presentation, 83% (n = 26) were identified through surveillance. Five-year
and agnostic 4-kb sliding windows, testing jointly using Burden/SKAT test. progression free survival (PFS) for these patients was 48%, though two
In the second stage, only the variant showing genome-wide significance with patients progressed later. Furthermore, at 5 years 25% of these patients had
EF was tested for its association with cardiomyopathy risk. Results: Among biopsy proven malignant transformation to HGG. This “transformation free
the 2,015 SJLIFE survivors with WGS data, a locus on 6p21.2 near survival” rate did not plateau, as at 7 years 56% of patients transformed. When
KCNK17achieved genome-wide significance with EF (rs2815063; MAF = considering death from a brain tumor, the 5- and 10- year overall survival (OS)
0.13; per allele beta = -0.016; P = 2.10´10-8), which replicated in 320 for the LGG group was 100% and 83%, respectively. Additional 3 patients
SJLIFE African survivors (MAF = 0.48; beta = -0.015; P = 0.004). In SJLIFE succumbed to other LFS related malignancies. For the HGG group, consisting
Europeans, 282 had a CTCAE Grade 2-5 cardiomyopathy. rs2815063 was of 30 patients, the 5 year OS was 35% (median follow-up 19.5 months),
significantly associated with increased risk of cardiomyopathy [per allele comparing favorably with the sporadic HGG population as reported in the
odds ratio (OR) = 1.38; P = 0.02], which replicated in 3,957 European survivors literature. Almost all of these patients presented with clinical symptoms.
from the Childhood Cancer Survivor Study (163 CTCAE Grade 3-5 self-reported Notably, 12 (40%) of them had a prior malignancy. Conclusions: Our analysis
cases; per allele OR = 1.39; P = 0.038). rs2815063 alters DNA binding motif of suggests that the risk of transformation of LGG in the setting of LFS is high and
EWSR1-FLI1, whose expression was found to lead to cardiomyopathy and death warrants ongoing surveillance. Interestingly, there are a considerable number
due to chronic cardiac failure in mice. Conclusions: Using a 2-stage approach, of long- term survivors in our HGG group, although the median follow up is still
we report a novel locus for cardiomyopathy in childhood cancer survivors, which short. Further study to examine potential genotype- phenotype correlations in
warrants additional work to gain mechanistic insights. germline TP53 mutation carriers will inform strategies to identify those pa-
tients at highest risk of glioma progression.

1518 Poster Session (Board #12), Mon, 1:15 PM-4:15 PM 1519 Poster Session (Board #13), Mon, 1:15 PM-4:15 PM
Germline mutations and onset of lung adenocarcinoma in smokers and Gaining insights into the DICER1 syndrome: An early report from the Italian
nonsmokers. First Author: Karen L. Reckamp, City of Hope Comprehen- DICER1 registry. First Author: Arcangela De Nicolo, Cancer Genomics
sive Cancer Center, Duarte, CA Program, Veneto Institute of Oncology IOV-IRCCS, Padua, Italy
Background: Eligibility for lung cancer screening is based largely on pack- Background: DICER1 is a key endoribonuclease in the microRNA pathway
years of smoking, missing many cases. To propose additional groups for that modulates gene expression. Germline loss of function variants in
screening, this observational study evaluated whether germline mutations DICER1, first found in pleuropulmonary blastoma, have been subsequently
associated with cancer risk accelerate onset of lung adenocarcinoma (LA) in linked to a variety of cancerous (and non) conditions referred to as DICER1
ever- and never-smokers. Methods: Patients with LA and family history of syndrome. In 2018, the Italian Society of Human Genetics launched an
cancer were recruited from our oncology clinic and the Clinical Cancer initiative aimed at establishing a national registry of DICER1 germline se-
Genomics Community Research Network. With consent, blood samples were quence variants. Methods: Centers involved in genetic testing for cancer
screened by large multi-gene panel for 4 categories of germline mutation predisposition were asked to report any identified DICER1 germline variants
[lung cancer-associated genes (TP53, EGFR); BRCA2; other genes in and related clinical information. Five University and/or research institutes
Fanconi anemia (FA) pathway; other DNA repair genes]. Accelerated failure- filled-in the electronic survey. Informed consent was obtained from patients
time models of age at LA diagnosis, adjusted for sex, ethnicity, and packs per or their legal guardians prior to DNA testing by NGS and/or Sanger se-
day, were constructed for never-smokers and ever-smokers. Statistical quencing. Results: Six DICER1 sequence variants were identified in 11
significance, at p,0.05 limited the False Discovery Rate to 5% across 8 individuals. Three missense variants are secondary results of NGS panels for
hypotheses. Results: In never-smokers with LA (n=104), mutated BRCA2, cancer predisposition and lack definitive categorization in online databases.
TP53 or EGFR were associated with younger age at diagnosis, while mutation in Three previously unreported variants are predicted to be protein truncating and,
other FA or DNA repair gene was not. In ever-smokers with LA (n=65), mutated hence, likely pathogenic. Of these, DICER1 c.4844delA p.(Lys1615Argfs*5)
BRCA2 and other FA gene were associated with younger age at diagnosis, while and c.4886C . G p.(Ser1629*) result from ad hoc testing offered to probands
other mutation categories were not (Table). Conclusions: Regardless of smoking based on a history of early onset follicular thyroid carcinoma and botryoid-type
history, BRCA2 mutation carriers experience accelerated onset of LA, as do embryonal rhabdomyosarcoma of the cervix and of pleuropulmonary blastoma
never-smokers carrying TP53 or EGFR mutation and ever-smokers with mu- 2nd type, respectively. DICER1 c.4643T . A p.(Leu1548*), instead, results
tation in FA gene other than BRCA2. With the exception of TP53 carriers (who from whole exome sequencing in two siblings with malignant melanoma who
merit whole body MRI), lung cancer screening with low-dose computed to- tested non informative for alterations in the CDKN2A and CDK4 melanoma
mography, starting earlier in adulthood than usual, may be warranted for in- predisposing genes. Further investigation unearthed thyroid disease in the
dividuals with germline mutations in these genes. Age at Diagnosis of Lung family and identified two other young carrier individuals, one unaffected and
Adenocarcinoma, by Germline Mutation and Smoking History, Adjusted for Sex, one thyroidectomized due to multinodular goiter. A DICER1 somatic hot spot
Ethnicity, and Packs per Day. sequence variant was detected in goiter specimens. Conclusions: Via the newly
Mean (95% CI)
established national registry we uncovered novel DICER1 germline sequence
Age at Diagnosis in Mean (95% CI) variants and uncommon genotype-phenotype associations. Our joint effort will
Germline Mutation N Never-Smokers p N Age at Diagnosis in Ever-Smokers p help us to refine our knowledge of the rare DICER1 syndrome, to inform research
EGFR or TP53 5 50.3 (39.6-63.9) 0.03 2 70.4 (56.4-88.0) 0.52 studies, and to improve testing and clinical management strategies.
BRCA2 2 43.0 (30.8-60.1) 0.01 4 60.2 (50.2-72.2) 0.001
Fanconi nemia Gene 3 59.1 (44.6-78.5) 0.82 6 63.3 (54.1-74.0) 0.002
other than BRCA2 3 70.3 (52.6-94.1) 0.18 4 74.8 (62.8-89.1) 0.94
Other DNA Repair Gene 91 60.6 (56.3-65.3) --- 49 74.5 (70.5-78.7) ---
Negative/Other Panel Gene

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Cancer Prevention, Hereditary Genetics, and Epidemiology 69s

1520 Poster Session (Board #14), Mon, 1:15 PM-4:15 PM 1521 Poster Session (Board #15), Mon, 1:15 PM-4:15 PM
Genetic identification and characterization of Lynch syndrome in a multi-ethnic Prevalence and clinical implications of mismatch repair (MMR) deficiency in
biobank. First Author: Rachel Rosenblum, Department of Medicine, Icahn unselected non-serous epithelial ovarian cancer (EOC) patients (pts). First
School of Medicine at Mount Sinai, New York, NY Author: Mariana Scaranti, The Royal Marsden NHS Foundation Trust and
The Institute of Cancer Research, London, United Kingdom
Background: Lynch syndrome (LS), caused by germline pathogenic variants
in mismatch repair (MMR) genes, results in increased risk of colorectal, Background: It has previously been reported that 6-7% of clear cell (CC) and
endometrial, and other cancers. LS has a prevalence of ~1 in 440 in European endometrioid (E) ovarian cancers are MMR deficient (MMRd). The prevalence of
ancestry populations; prevalence data in other populations are limited. We MMRd in other histological subtypes and correlation with germline (g) MMR
identified and characterized carriers of pathogenic MMR gene variants in the (Lynch Syndrome) mutations in unselected non-serous EOC pts is less clear.
multi-ethnic BioMe Biobank in New York City. Methods: Exome sequence MMRd in solid tumors has been associated with enhanced response to immu-
data from ~31,000 BioMe participants were evaluated for known (per ClinVar) notherapy, hence knowledge of MMR status has therapeutic and familial impli-
and predicted (loss-of-function) pathogenic variants in MMR genes. Pop- cations. We aimed to study the prevalence and implications of MMRd and gMMRd
in unselected pts with non-serous EOC. Methods: Routine immunohistochemistry
ulation groups were defined by genetic ancestry. Participant questionnaires
(IHC) was performed for the MMR proteins MLH1, MSH2, MSH6 and PMS2 in all
and electronic health records (EHRs) of carriers were reviewed for personal or non-serous EOC pts from June 2016; retrospective MMR IHC testing in pts in
family history of malignancy. Results: We identified 48 carriers of 33 distinct follow-up was performed. Pts with MMRd tumors were referred for gMMR testing.
pathogenic variants in PMS2 (48%), MLH1 (27%), MSH6 (15%), and MSH2 Results: We analyzed 66 unselected pts with non-serous EOC. Median age was
(10%), for an estimated prevalence of ~1/640 in the BioMe Biobank. 56.4 years (yrs). The majority had E ovarian cancer (54.5%) followed by CC
Prevalence was higher among individuals of Non-Jewish European (N = 14; 1/ (25.8%), mixed histology (12.1%), mucinous (4.5%) and mullerian (3%) sub-
400) and African (N = 14; 1/490) ancestries, compared to Puerto Rican types. Endometriosis was noted in 45.5% of pts, and 75% were FIGO stage I and II
(N = 8; 1/640), Ashkenazi Jewish (N = 6; 1/690), and other/mixed (N = 6) at diagnosis. Seventeen pts (25.8%) had concurrent endometrial cancer, all
ancestries. Carriers had a median age of 56 (range 27 to 77) years and were 50% Grade I. On IHC, 15.2% were MMRd: 5 E, 2 CC, 2 mixed and 1 mullerian-type. Of
female. Overall rate of malignancy among carriers was 38%, with the lowest these, 3 pts (30%) had gMMR mutation, 2/3 did not meet the Revised Bethesda
rate in PMS2 (26%) and the highest rate in MSH6 (57%) variant carriers. We criteria for testing. A lower average body mass index (Kg/m2) was noted in MMRd
found a high prevalence of endometrial cancer (21% of female carriers) and a 25.9 versus 30.1 in MMR proficient (MMRp). Median age at diagnosis was 53.5
lower prevalence of colorectal cancer (4% of all carriers). Only 2 carriers (4%) yrs in the MMRd and 57.7 yrs in MMRp. A higher frequency of concurrent en-
had a diagnosis of LS in their EHRs, and only 1 carrier met Amsterdam di- dometrial cancer was observed on the MMRd group (60%) versus (20%) on MMRp
agnostic criteria for LS. Conclusions: These data show that ~0.15% of par- (p = 0.007). No statistically significant difference in overall survival or disease-
ticipants in a multi-ethnic biobank are carriers of pathogenic MMR gene free survival was observed between the MMRd and MMRp population.
Conclusions: Our study has shown a higher prevalence of somatic MMRd in non-
variants and suggest that the prevalence is higher in European and lower in
serous EOC (15.2%) than in previously published literature with a significant
non-European ancestry populations. Notably, most carriers do not have a proportion found to carry gMMR mutations (4.5%). These interim findings support
clinical diagnosis of LS and do not meet diagnostic criteria for LS. Carriers the role of universal MMR IHC testing in non-serous EOC regardless of family
demonstrate variable rates of cancer, which may contribute to under-diagnosis history.
of LS. Genomic screening for pathogenic MMR variants may lead to earlier
diagnosis of LS and improved outcomes. Patterns of MMR loss IHC (n) gMMR mutation found
MSH2/MSH6 2 1
MSH2
MLH1/PMS2 4 1
MLH1
MSH6 2 1
MLH1
PMS2 2 0

1522 Poster Session (Board #16), Mon, 1:15 PM-4:15 PM 1523 Poster Session (Board #17), Mon, 1:15 PM-4:15 PM
Risk of cancer in first-degree relatives of childhood cancer patients: A linked Rare tumor with matched germline whole exome sequencing to identify
longitudinal population-based registry study. First Author: Laura-Maria somatic and inherited variants of clinical significance. First Author: Bryce
Madanat-Harjuoja, Radcliffe Institute for Advanced Studies, Harvard University, Perkins, University of Arizona, Department of Medicine, Tucson, AZ
Boston, MA
Background: Rare cancer incidence is defined as ,6/100,000 cases. These
Background: Population based data on risk of cancer in relatives of child- cases are challenging due to delays in diagnosis, late stage at diagnosis, lack
hood cancer patients are sparse. Using linked population-based registries, of standard of care and poor outcomes. We present molecular profiling of 28
we set out to evaluate risk of early onset cancer in first-degree relatives of rare tumor cases analyzed by whole exome sequencing (WES) seen in the
childhood cancer patients. Methods: We queried the Finnish Cancer Reg- Early Phase Therapeutics Program. Methods: We performed WES on 15 rare
istry and ascertained a cohort of 9135 individuals diagnosed with at least tumors with matched germline DNA, as well as tumor only on 2 additional
one cancer under the age of 21 years between 1970 and 2012. We then went cases. We sequenced a panel of 500 candidate cancer genes in 6 tumor/
on to identify a total of 58,211 unique first- and second-degree relatives by normal and 6 tumor only cases. Copy number alterations (CNAs) were
linking to the Central Population Registry. Relatives were then linked back to assessed for the tumor-normal WES cases with SNVs and small indels for all
the annually updated Finnish Cancer Registry to identify cancer diagnoses in cases. Caris (AZ) (tumor only) and liquid biopsy (Gaurdant360, CA) were also
siblings, offspring and parents of childhood cancer patients, restricting to conducted in these patients for comparison. Results: Rare tumors affected
cancers occurring under the age of 40. Risk of cancer in relatives of the index 18 different tissues of the body with 1 case affecting tissue typical of
case was estimated using standardized incidence ratios (SIRs) comparing common cancers. The rare tumor group contained 25% pathogenic or likely
cancer age and period specific incidence in relatives to that of the general pathogenic germline variants compared to 7.1% for common tumors. This
population. Results: A total of 288 cancers were diagnosed in relatives increased rate of inherited pathogenic variants met statistical significance:
during the 900,907 years of follow-up, while 266 cancers were expected. Fisher exact test (p = 0.0118). A total of 69 sequence variants and 8 CNAs
The overall risk of cancer in siblings of childhood cancer patients was el- were identified, 37 were actionable. For all but 3 patients, there was at least
evated (SIR 1.18 95% CI 1.00-1.39). 144 of the childhood cancer patients one variant associated with an actionable outcome given off-label use or
were identified as having a sibling additional to index case with a diagnosis of clinical trial participation. There was excellent concordance with results
cancer at age , 40; 44 of these 144 also had a parent with early onset cancer. from commercial sequencing; however, our tumor/germline sequencing
The risk of early onset cancer was elevated in offspring overall (SIR 1.79 95% identified additional germline variants of clinical significance, all were loss-
CI 1.05-2.81) and in offspring of retinoblastoma, malignant bone tumor and of-function variants in tumor suppressors. Tumor vs. normal analysis showed
neuroblastoma patients. Siblings of lymphoma patients were at elevated risk that the majority of the commercially reported VUS were in fact germline
of early cancer, and the mothers of 11 of 27 sibling pairs (lymphoma + VUS. In one case, a reported pathogenic variant we found to be an inherited
cancer , 40 yo) also had cancer at age , 40. Conclusions: Linked registries pathogenic germline variant. 15 patients received at least 1 line of therapy
allow family history of cancer to be evaluated across multiple relatives and to indicated by genomic sequencing. For these patients the PFS ratio when
be longitudinally updated. Results are generally reassuring with regard to risk compared to the most recent line of therapy prior to targeted therapy was
of cancer in relatives of childhood cancer patients. Elevated risk in relatives of 2.02, with 10/15 (67%) patients having a PFS ratio of .1.0, and 6/15
retinoblastoma and malignant bone tumor patients are in line with the known (40%) of patients having PFS ratio .1.3. Conclusions: We demonstrate the
cancer syndromes associated with these tumor types, and lymphoma and importance of tumor-normal analysis, especially in the context of rare tu-
neuroblastoma families need further analysis. mors. Rare tumor patients may disproportionately benefit from tumor vs.
normal WES at diagnosis to improve PFS within targeted therapy trials,
guided by genomics.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
70s Cancer Prevention, Hereditary Genetics, and Epidemiology

1524 Poster Session (Board #18), Mon, 1:15 PM-4:15 PM 1525 Poster Session (Board #19), Mon, 1:15 PM-4:15 PM
Tumor mutation burden and PD-L1 expression in SDH/FH mutated solid Preventive surgery after multiplex genetic panel testing (MGPT). First Au-
tumors. First Author: Leylah Drusbosky, NantHealth, Culver City, CA thor: Gregory Idos, University of Southern California Norris Comprehensive
Cancer Center, Los Angeles, CA
Background: Succinate Dehydrogenases and Fumarate Hydratase (SDH/FH)
deficient tumors are characterized by succinate/fumarate accumulation and Background: Guidelines recommend consideration of prophylactic surgery
resultant pesudohypoxia that drives malignant transformation. This state of for patients with a germline pathogenic variant in some cancer predisposition
pseudohypoxia leads to dysregulation of PD-1 receptor-ligand signaling. In genes. We assessed surgery utilization in a prospective, multi-institutional
this study, we explored tumor mutation burden (TMB), gene expression of cohort study of MGPT. Methods: 2000 patients had MGPT and completed
PD-L1, and expression of other immune checkpoint- associated genes in a questionnaires at 3, 6, and 12 months. Patients reported surgical utilization
diverse cohort of human tumors harboring SDH A, B, C, D and FH mutations. and indication (treatment or prevention). Surgery utilization was assessed
Methods: Retrospective analysis was performed on whole exome sequencing according to cancer history and MGPT test results: Positive, pathogenic
(WES; ~150x coverage) and whole transcriptomic RNAseq (~200x106 reads variant; VUS, variant of uncertain significance; Negative, benign variants.
per tumor) data from NantHealth to identify tumors harboring SDHx and/or Results: Overall, 12.9% (198/1537) of patients reported surgery after
FH mutations. WES was performed on tumor tissue and matched normal MGPT (median follow-up 13 months). Only 31.3% (62/198) of patients
for each patient to assess TMB. TMB was measured by counting all somatic- specified that their surgery was preventive. Preventive surgery utilization was
specific non-synonymous exonic mutations, with . 200 mutations quali- significantly higher among patients who tested positive (n=30, 14.9%)
fied as TMB-high. Immune checkpoint therapy-related gene expression was compared to those testing negative (n=20, 2.3%, p,0.001) or VUS (n=12,
evaluated for PDL1, CTLA4, IDO, LAG3, FOXP3, PDL2, TIGIT, TIM3 and 2.2%, p,0.001). Preventive surgery was very low among patients testing
OX40. Results: Among tumor samples from 3377 patients analyzed, 42 negative or VUS who had no personal history of cancer in the relevant organ
patients were found to harbor potentially-pathogenic & pathogenic muta- (Table). For example, mastectomy was not reported among any patients
tions in the SDHA, B, C, D and FH genes. The most common tumor types with testing negative or VUS who had no personal history of breast cancer (Table).
SDH/FH mutations were lung (n = 7), breast (n = 6), and colon cancer (n = 6). Conclusions: More than one year after MGPT, prophylactic surgery use was
Our analysis revealed that TMB was positively correlated with the presence of low among patients with VUS or negative results, especially among those
SDH/FH mutations (p , 0.001). High PD-L1 expression was also signifi- with no personal history of cancer at the relevant site. Surgery utilization.
cantly correlated with the presence of SDH/FH mutation (p , 0.05), while
Positive VUS Negative
CTLA4, IDO, LAG3, FOXP3, and OX40 expression was significantly higher in
SDH/FH mutated samples (p , 0.05). Conclusions: We report for the first Surgery Preventive Surgery Preventive Surgery Preventive
N N (%) N (%) N N (%) N (%) N N (%) N (%)
time an association between increased TMB and increased PD-L1 expres-
sion in a variety of SDH/FH mutated tumors. These key parameters, imply Personal History of Relevant Cancer*
Mastectomy 55 23 15 (27.3) 179 52 11 (6.1) 284 80 17(6.0)
that a higher TMB may drive the evolutionary pressure to select clones with a (41.8) (29.1) (28.2)
PDL1 high phenotype. This observation supports a potential therapeutic role Oophorectomy 5 3 (60.0) 0 1 1 (100) 0 14 2 (14.3) 0
for inhibition of PD-1/PD-L1 pathway in these tumors. Hysterectomy 2 2 (100) 0 6 1 (16.7) 0 2 2 (100) 0
No Relevant Personal Cancer History**
Mastectomy 132 5 (3.8) 5 (3.8) 312 2 (0.6) 0 531 1 (0.2) 0
Oophorectomy 144 20 15 (10.4) 436 6 (1.4) 1 (0.2) 713 9 (1.3) 2(0.3)
(13.9)
Hysterectomy 144 7 (4.9) 3 (2.1) 404 2 (0.5) 0 686 9 (1.3) 1(0.1)
NOTE: N values are the number of survey responders in each test result and cancer history
category *Mastectomy, Breast Cancer; Oophorectomy, Ovarian Cancer; Hysterectomy,
Ovarian Cancer, Uterine Cancer **Individuals may have a history of other cancer(s)

1526 Poster Session (Board #20), Mon, 1:15 PM-4:15 PM 1527 Poster Session (Board #21), Mon, 1:15 PM-4:15 PM
Preimplantation genetic diagnosis: What do BRCA mutation carriers think? Prospective Registry of Multiplex Testing (PROMPT): Follow-up. First Author:
First Author: Olivia R Khouri, NYU School of Medicine, New York, NY Jamie Brower, Basser Center for BRCA, University of Pennsylvania, Philadelphia,
PA
Background: The use of pre-implantation genetic diagnosis (PGD) to select
against BRCA mutated embryos for in-vitro fertilization (IVF), introduces Background: Prospective Registry of Multiplex Testing (PROMPT) is an
complex choices for patients with pathogenic BRCA mutations. We sought to online registry for individuals who have had multiplex panel testing for
describe the uptake of and attitudes toward this technology in this patient cancer susceptibility. The main objective of this registry is to ascertain
population. Methods: We conducted a prospective survey study at a single families to allow penetrance calculations for mutations in less well char-
institution in New York City affiliated with both a Cancer Center and Fertility acterized genes. Methods: Since September 2014, health care providers
Center, to assess attitudes and utilization of PGD. Cancer Center staff and commercial laboratories have given PROMPT information with the test
distributed surveys to patients with known BRCA mutations between April and results to eligible individuals. Participants with pathogenic mutations or
August 2018. Survey participation was voluntary and anonymous. Survey data variants of unknown significance in cancer susceptibility genes have self-
were analyzed using descriptive statistics and two-tailed t tests. Results: 80 enrolled in PROMPT. We have sent annual follow-up surveys to participants
survey responses were collected. A majority of the patient population iden- to capture updated personal and family histories, additional genetic testing
tified as Caucasian (87.5%, 70) and Jewish (52.5%, 42). The survey was in the family, and changes in cancer screening patterns. Results: To date,
distributed to all age groups; however 81% (65) were between 26 and 45 years 5800 participants have enrolled in PROMPT and 2321 follow-up surveys are
of age. 63.8% (51) had heard of PGD prior to completing the survey, while completed. Of those eligible for follow-up, 1980 (44%) have completed at
36.3% (29) had not. Only 40% of respondents (32) felt sufficiently educated least one survey. 134 (5.8%) reported a new cancer diagnosis since
regarding PGD. 35% (28) patients met with an REI, of whom 6.3% (5) utilized baseline. Breast cancer (37, 27.6%) and non-Melanoma skin cancer (35,
IVF with PGD, and 21.3% (17) plan to use IVF with PGD in the future. 11.3% 26.2%) were the most commonly reported new diagnoses. 5.7% of par-
(9) wish they had known about this technology prior to starting a family, 20% ticipants with an ATM variant, 5.2% of those with a BRCA2 variant, and
(16) would not have used PGD had they known about it prior to childbearing. 4.3% of those with a CHEK2 variant and at least one follow-up survey have
Reasons respondents were unlikely to pursue PGD included: cost (38.8%, reported a new primary cancer diagnosis (excluding non-Melanoma skin
31), completed childbearing (25%, 20), medical risk (18.8%, 15), and cancer). In addition, 70 (3%) participants reported a cancer recurrence
ethical concerns (16.2%, 13). Patients gave cost estimates for PGD ranging since time of last follow-up. Bone (9, 12.9%) and liver (8, 11.4%) were the
from $500 - $120,000. There was no statistically significant correlation most common recurrence sites. 12.7% of those with a TP53 variant, 7.1% of
between likelihood of pursuing PGD and parity (p = 0.45), religion (p = 0.78), those with a BARD1 variant, and 3.3% of those with an ATM variant and at
education level (p = 0.13), number of family members affected by BRCA least one follow-up survey have reported a recurrence. 59 (2.5%) partici-
mutations (p = 0.20) or by cancer (p = 0.11). Conclusions: Overall, a small pants had additional clinical genetic testing, and 555 (24%) participants
number of patients with pathogenic BRCA mutations utilize PGD. A minority of reported clinical genetic testing of family members. Conclusions: PROMPT
survey respondents felt adequately educated about PGD. Reported barriers to collects updated health information from participants to achieve the primary
uptake were varied, and there was a wide range of cost estimates reported. Our goal of determining cancer incidence among carriers of different mutations.
results suggest that increased patient education regarding PGD in BRCA Moreover, PROMPT has increased participant contacts through use of an-
mutation carriers is warranted. nual follow-up surveys and gene-specific surveys with subsets of partici-
pants. Ongoing efforts include obtaining pathology reports of new and
recurrent cancer diagnoses on study.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Cancer Prevention, Hereditary Genetics, and Epidemiology 71s

1528 Poster Session (Board #22), Mon, 1:15 PM-4:15 PM 1529 Poster Session (Board #23), Mon, 1:15 PM-4:15 PM
Germline variants in urothelial carcinoma: Analysis of pathogenic and likely Prognosis value of a genetic score based on germline genetic variants in a
pathogenic variants in 645 subjects. First Author: Sarah Abou Alaiwi, prospective cohort of early triple-negative breast cancer patients. First
Dana–Farber Cancer Institute, Boston, MA Author: Elsa Curtit, University Hospital - Medical Oncology Department,
Besançon, France
Background: While small studies have supported a genetic cancer pre-
disposition among subjects with urothelial carcinoma (UC), systematic Background: Triple-negative breast cancers (TNBC) are a heterogeneous group of
germline evaluation of this population is lacking. Here, we report the tumors with poor outcome. In this study, the association between germline genetic
prevalence of germline variants among subjects with UC from multiple variants and invasive disease-free survival (iDFS) was analyzed in TNBC patients.
centers completing panel-based testing at a large, commercial laboratory. Methods: A genome wide-association study (GWAS) aimed to identify variants
Methods: 1149 UC subjects underwent germline testing of 1 to 126 genes (single nucleotide polymorphisms – SNPs) associated with prognosis in 1121
using massively parallel sequencing with customized capture bait-sets to patients with TNBC in the SIGNAL prospective cohort. Associations between gene
analyze exonic regions, flanking intronic sequences, and copy number al- variants and iDFS were assessed in univariate Cox regression models. Variants
were combined in a score to identify risk categories. A prognostic model based on
terations. Pathogenic (P) and likely pathogenic (LP) were confirmed using
breast cancer stage and genetic variants was estimated using a multivariate Cox
orthogonal technology in accordance with Invitae standard operating regression. Interaction between stage and genetic score was tested. Discrimi-
practices. Analysis was limited to 645 subjects who completed testing of a nation of the model was assessed by the Harrell’s C statistic and internal validity by
shared set of 42 genes. P/LP variants including single nucleotide variants/ bootstrap method. Results: The characteristics of the 1121 patients were rep-
indels/ copy number variants are reported. De-identified personal and family resentative of a population with early TNBC. Four SNPs on chromosomes 9 and 2
cancer histories were evaluated. Fisher’s Exact test and the Mann-Whitney were found significantly associated to iDFS in univariate Cox models. Homozygous
test were used to analyze categorical and continuous variables respectively. status for the most frequent allele was associated with poorer iDFS for two SNPs
Results: Among the 645 UC subjects with 42-gene testing for any indication, and this status was present in 50% and 57% of the population. For the two other
median age at testing was 60 years (6-88) and 326 (51%) were female. P/LP SNPs, the most frequent allele was associated with more favorable iDFS. Three
variants were identified in 21 (50%) of the 42 genes in 98 (15%) of subjects, prognostic categories were derived from the genetic score. The following table
including Lynch syndrome genes (n = 26 [4%]), BRCA1/2 (n = 16 [2.5%]), presents the results from the multivariate Cox model including genetic score and
CHEK2 (n = 15 [2.3%]), and heterozygous MUTYH (n = 12 [1.9%]). Among disease stage. Clinical trial information: RECF1098. Conclusions: In a pro-
18 DNA damage repair (DDR) genes assessed, 90 P/LP variants were de- spective cohort of 1121 patients with early TNBC, 4 genetic variants (SNPs) were
tected in 88 subjects (12.2%). There was no significant association between associated with iDFS. A score involving SNPs provided similar prognostic in-
presence of a DDR gene variant and age at diagnosis, gender or reported dications as breast cancer stages. A search assessing the function and the role of
family history of UC in a first degree relative (n = 48). Among subjects the involved genes is ongoing.
with documented history of UC only without other cancers (n = 195), 24 n (events) HR 95%CI Bootstrap 95% CI pvalue
(12.3%) had P/ LP variants, of which 23 (11.8%) were in a DDR gene. 1017* (204)
Conclusions: Germline P/LP variants were identified in 15% of UC subjects stage I 347 (40) 1 , 0,0001
most of which (92%) were in DDR genes, including 27% in Lynch syndrome II 484 (90) 1,64 1,13-2,38 1,00-2,83
III 177 (74) 4,86 3,31-7,15 3,03-8,7
genes. PARP and T-cell checkpoint inhibitors may warrant evaluation in score low 652 (93) 1 , 0,0001
subjects with germline DDR mutations. Further validation in unselected UC medium 311 (90) 2,29 1,71-3,06 1,52-3,37
pts is warranted to propose examining germline P/LP variants in all UC high 45 (21) 4,58 2,85-7,37 2,15-9,37
patients. * missing data: 104 / C-Harrell at 0.71 (95% CI 0.67-0.75) No significant interaction was found
between stage and genetic score (p = 0.5184).

1530 Poster Session (Board #24), Mon, 1:15 PM-4:15 PM 1531 Poster Session (Board #25), Mon, 1:15 PM-4:15 PM
Test of InheRET, an online tool to facilitate NCCN Guideline compliant The PHACT Study: Population Health and Cancer Testing. First Author:
referrals for cancer genetic counseling. First Author: Lynn McCain, Mallika Sachdev Dhawan, University of California San Francisco, San
University of Michigan, Ann Arbor, MI Francisco, CA
Background: Identifying the ~60 million unaffected persons in the US at risk Background: There is considerable uncertainty on cancer risk and recom-
for inherited cancers has the potential to reduce their cancer risk by up to mendations for genetic testing in various populations. The purpose of this
95%. However, most of these individuals are not identified currently be- study is to test types and frequencies of cancer risk mutations in large,
cause of multifactorial deficits in the 3-generation pedigree collection in unaffected multiethnic populations, as well as feasibility and acceptance of
clinical settings. Methods: Here we evaluated the impact InheRET, an online general population germline testing. Methods: After consent, germline ge-
family history gathering and risk assessment reporting tool, has on facili- netic testing via Color Genomics 30-gene Cancer Risk panel with personal
tating National Comprehensive Cancer Network (NCCN) Guideline-compliant and family cancer history assessment was offered to 500 participants re-
referrals for cancer genetic counseling/genetic evaluation by decreasing and/ siding in the San Francisco Bay Area. Participants were older than 21 and
or removing the barriers of 1) in-clinic 3-generation family history collection, without a known family history of a cancer risk mutation. Recruitment oc-
and 2) interpretation of the family and personal history in light of current curred at random and through various, non-cancer related community
NCCN Guidelines. Patients enrolled from primary care and specialty clinics events. Genetic counseling through a genetic counselor at UCSF was offered
completed the family health history from a web-enabled devices using to all participants. Post-participation surveys were sent out to those who
InheRET Inherited Risk Evaluation Tool. Results: Of 255 enrolled patients, completed testing with 195 responses to assess attitudes towards genetic
78.4% completed the history form and, of these, 86.5% completed the testing. Results: 500 participants completed testing; 35 were found to
feedback survey. 39.2% of primary care and 79.9% of specialty cancer have a cancer risk mutation (7.0%). The majority of these have been in
genetics patients were found to be at increased risk. Patients with # HS moderate risk cancer mutations including CHEK2 (1.0%), APC (0.8%),
education ranked InheRET at 4.7/5.0 for Understandability and 3.33/5.0 for MUTYH (1.4%) and NBN (0.4%); higher risk mutations were found in
Ease of Use. Following the addition of clarifying directions, they ranked BRCA1 (0.4%), BRCA2 (0.6%), BRIP1 (0.4%), PMS2 (0.2%) and PALB2
InheRET at 5.0 and 4.0, respectively. Pts. ,60 ranked InheRET between 3.5 (0.2%). Data via self-reporting and SNP testing via the Color Genomics
and 3.8 on Ease of Use, while those 60-69 and 70+ ranked it 3.0 and 2.7, platform was collected to contextualize the racial, ethnic, and geographic
respectively, potentially reflective of lower technical skills. Age did not impact association of these results. Rates of VUS (variants of uncertain significance)
Understandability. In the cancer genetics clinic, 86% of pts. completed the differed among the various ethnic groups (p = 0.027) with the lowest rates in
online survey, in ~ 1 week, compared to paper forms, which required re- Europeans vs. all other ethnicities (16.6% vs. 29%, p = 0.001). 37 (19.8%)
minders, and .4 weeks to obtain. Conclusions: In this pilot, InheRET out of 187 respondents reported that their genetic testing influenced testing
accelerated appropriate referrals and efficient utilization of genetic in family members. In those 37 families, 8 additional people were positive for
counseling services; improvements to navigation are being implemented genetic mutations. Cost was the biggest barrier to pursuing prior genetic
prior to dissemination. testing according to most individuals (20% of respondents); others were too
busy (12%) or did not feel genetic testing was medically recommended
(9%). Conclusions: The PHACT study demonstrates that large population
screening of cancer risk mutations is feasible. The number of participants
found to have mutations was greater than expected although the majority of
these were in moderate risk genes. We found high rates of VUS in non-
Europeans, which points to the complexity of population-based genetic
testing.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
72s Cancer Prevention, Hereditary Genetics, and Epidemiology

1532 Poster Session (Board #26), Mon, 1:15 PM-4:15 PM 1533 Poster Session (Board #27), Mon, 1:15 PM-4:15 PM
Characterization of Lynch syndrome (LS) associated cancers in patients with Safety and efficacy of aspirin for primary prevention of cancer: A meta-
immune dysfunction. First Author: Shahla Bari, Moffit Cancer Center, analysis of randomized controlled trials. First Author: Tarek Haykal, Hurley
Tampa, FL Medical Center/Michigan State University, Flint, MI
Background: LS is caused by a germline mutation in one of several DNA Background: In the United States, cancer is the second leading cause of
mismatch repair (MMR) genes: MLH1, MSH2, MSH6 or PMS2. In- mortality, and millions more battle cancer worldwide. As such, primary
appropriate immune responses as seen in chronic inflammatory conditions prevention of cancer is a major interest globally. Aspirin has been studied
as well as immunodeficiency states confer increased risk of developing as a primary prevention method for multiple diseases, mainly cardiovascular
cancer. This aim of this study was to evaluate the effect of immune dys- disease and various forms of cancer. The role of aspirin as a primary pre-
function on the characteristics of LS associated cancers. Methods: This vention of cancer is still controversial and may be more beneficial in certain
was a retrospective analysis of LS patients and carriers at two institutions cancers over others. With rapidly surfacing large randomized controlled trials
listed above. We evaluated mutational profiles, immune status, age of onset (RCTs) studying this subject, we aimed to evaluate the efficacy and safety of
of first and subsequent cancers in this cohort. Results: 106 patients with aspirin as a primary prophylaxis for cancer. Methods: A comprehensive
mutations consistent with LS were included. 72 patients had at least one electronic database search was conducted for all RCTs that compared aspirin
cancer while 34 were carriers. 44% patients were Caucasian female, 18% versus placebo for the prevention of any type of disease, and where cancer
white males, 14% African American males, 11% African American females incidence or mortality was reported. The primary outcome was cancer-
and 10% Hispanic females. Colon cancer (CRC) was the most common related mortality. Secondary outcomes were cancer incidence, all-cause
cancer (44%) and PMS2 was the most common mutation, noted in 35 mortality, major bleeding, any bleeding and gastrointestinal (GI) bleeding.
patients (33%). Of the 72 patients with LS associated cancer, 18 patients We calculated risk ratios (RRs) and 95% confidence intervals (CIs) using a
were either immunosuppressed or had an autoimmune condition. Of the 10 random-effects model at the longest follow-up period. Results: We included
patients who had an autoimmune condition ,7 had multiple cancers. Of the 9 16 RCTs with 104,018total patients, mean age of 60.51 years, mean follow-
patients who were immunosuppressed, 5 had multiple cancers. Out of a total up of 5.48 years, and a male percentage of 38.72%. We found that aspirin
of 18 out of 72 patients who had multiple cancers, 12 (66%) had either an was not associated with a significant reduction of cancer-related mortality
autoimmune condition or were immunosuppressed. CRC was the index compared with placebo (RR 0.99; 95% CI: 0.87-1.12; P = 0.85: I2= 41%).
cancer in 42% and breast in 33% of patients with multiple cancers. Patients Compared with placebo, aspirin was not associated with significant re-
with MSH2 were most likely to have an immune related condition (32%) and duction of all-cause mortality (RR 0.97; 95% CI: 0.92-1.02; P = 0.19; I2=
accounted for 41% of patients with multiple cancers. The median age of first 13%) or cancer incidence (RR: 0.98; 95% CI: 0.92-1.04; P = 0.43; I2=
cancer in this group was 46 years while it was 48.5 years in the population 16%). However, aspirin treatment was associated with significantly in-
without immune dysfunction (p = 0.2). There was a high prevalence of breast creased risks of any bleeding (RR 1.63; 95% CI: 1.31-2.03; P , 0.01),
cancer (24%) as a LS associated cancer in our study population. 66% of the major bleeding (RR 1.41; 95% CI: 1.26-1.57; P , 0.01), and GI bleeding
patients with PMS2 mutation had breast cancer with a median age of onset of (RR 1.85; 95% CI: 1.38-2.48; P , 0.01) compared with placebo.
48 years (62 years for sporadic cancer). Conclusions: Our study is the first to Conclusions: Our study did not find any significant reductions in cancer-
look at the effect of immune dysfunction in LS patients. Immune dysfunction related mortality or cancer incidence when compared with placebo. Our
was associated with a higher rate of multiple cancers and was more com- study also highlights the dangers of aspirin for primary prevention of cancer
monly associated with the MSH2 mutation. It also highlights importance of as aspirin was found to cause higher rates of bleeding (any bleeding, major
aggressive screening for breast cancer in LS patients (especially with PMS2 bleeding, and GI bleeding) compared to placebo at the longest follow-up
mutation). period with no significant benefit in cancer primary prevention.

1534 Poster Session (Board #28), Mon, 1:15 PM-4:15 PM 1535 Poster Session (Board #29), Mon, 1:15 PM-4:15 PM
Role of vitamin D supplementation for primary prevention of cancer: Meta-analysis Opportunistic off-target cancer screening in organized programs: The EDIFICE
of randomized controlled trials. First Author: Varun Samji, Hurley Medical Center/ 6 survey. First Author: Jean F. Morere, Paul Brousse Hospital, Villejuif, France
Michigan State University, Flint, MI
Background: The efficacy and benefit/risk ratio of organized nationwide
Background: In the United States cancer is the second leading cause of cancer screening programs rely on the age range of eligible average-risk
mortality, as such, primary prevention of cancer is a major public health populations. We studied the characteristics of off-target populations who
concern. Vitamin D supplementation has been studied as a primary pre- underwent opportunistic screening for colorectal (CRC), breast (BC) or
vention method for multiple diseases including cardiovascular disease, cervical (CC) cancer, #5y prior to the recommended age. Methods: The
osteoporosis, diabetes mellitus and cancer. The role of aspirin as primary French nationwide observational survey, EDIFICE 6, was conducted online
prevention of cancer is still controversial. With fast emergence of large (June 26-July 28, 2017) on a core sample of 12 046 individuals (18-69y).
randomized controlled trials (RCTs) in that regards, we aimed to evaluate the Representativeness was ensured by quota sampling on age, sex, profession,
efficacy of Vitamin D supplementation as primary prophylaxis for cancer. and stratification by geographical area/type of urban district. Opportunistic
Methods: A comprehensive electronic database search was conducted for all screening for BC (in 533 women, age 45-49y), CRC (in 1331 individuals,
RCTs where comparison of Vitamin D supplementation versus placebo for the age 45-49y) or CC (in 633 women, age 20-24y) was assessed in terms of
prevention of any type of disease with at least 3 years of Vitamin D sup- smoking status (current, former/never smoker), marital status (single, living
plementation was used and where cancer incidence or mortality was re- with a partner), type of residential area (urban, rural), having a close relative
ported. The primary outcome was cancer-related mortality and cancer with cancer, social vulnerability (EPICES score), and self-reporting own
incidence. We calculated risk ratios (RRs) and 95% confidence intervals cancer risk (higher, identical/lower than average). Results: In the off-target
(CIs) using a random-effects model at the longest follow-up. Results: We populations, screening rates were 78% for BC (N = 418, mammogram), 13%
included 10 RCTs with 79,055 total patients, mean age of 68.07 years, a for CRC (N = 172, fecal test or colonoscopy) and 42% for CC (N = 264,
female percentage of 78.02% and a minimum follow-up of 4 years and more. cervical Pap smear test). Premature BC screening rates were significantly
Vitamin D was associated with significant reduction of cancer-related higher (P , 0.05) in non-vulnerable than in vulnerable individuals (84% vs
mortality compared with placebo (RR 0.87; 95% CI: 0.79-0.96; P = 69%), and among those self-reporting their own BC risk as higher than
0.05: I2= 0%). Compared with placebo, Vitamin D was not associated with average (84% vs 76% reporting own BC risk as identical/lower than average).
significant reduction of cancer incidence (RR: 0.96; 95% CI: 0.86-1.07; Premature CC screening rates were correlated with: smoking status (66% in
P = 0.46; I2= 31%). Conclusions: Our study highlights that the use of Vitamin current smokers vs 35% in former/never smokers), and marital status (63%
D supplementation for primary prevention of cancer is important as it does in those living with a partner vs 34% single). Lastly, factors correlated with
decrease cancer-related mortality once cancer is diagnosed, however it has no premature CRC screening were: type of residential area (urban, 15% vs rural,
role or effect on cancer incidence. 8%), and believing own risk of CRC to be higher than average (27% vs 8% of
those who self-reported their own CRC risk as identical/lower than average).
Conclusions: This analysis reveals several factors related to premature
screening for BC, CRC and CC, provides clear insight into off-target cancer
screening uptake profiles, and hints at new strategies to ensure the optimal
risk/benefit ratio of screening practices.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Cancer Prevention, Hereditary Genetics, and Epidemiology 73s

1536 Poster Session (Board #30), Mon, 1:15 PM-4:15 PM 1537 Poster Session (Board #31), Mon, 1:15 PM-4:15 PM
Management of high, moderate, and low penetrance ovarian cancer susceptibility Implementation of a low-dose computerized tomography (LDCT) lung cancer
gene mutations: An assessment of current practice patterns. First Author: screening program (LCSP) across a large integrated health system. First
Catherine H. Watson, Duke University, Durham, NC Author: Marija Bjegovich-Weidman, Aurora Health Care/Aurora Cancer Care,
Milwaukee, WI
Background: Limited information exists regarding appropriate risk-reduction
strategies for women with moderate and low penetrance ovarian cancer Background: The LDCT LCSP was launched as a critical component of our
(OVCA) susceptibility mutations. We sought to assess current practice Cancer Program to support tobacco cessation efforts and increase early
patterns for women with these genetic changes through a survey of members detection. Initially it was offered as a self-referral low cost screening. The
of the Society of Gynecologic Oncology (SGO). Methods: All full SGO program was expanded when the Affordable Care Act and Center for
members were e-mailed a survey consisting of two vignettes: 1) a 35-year-old Medicare/Medicaid Services covered it as a preventative services benefit in
premenopausal woman who desires pregnancy; 2) a 55-year-old post- January 2015. Methods: 9 LDCT LCSP locations were implemented be-
menopausal woman with multiple comorbidities. Each vignette contained tween 2014-September 2016. Program data are submitted to the American
sub-scenarios in which the patient had either a BRCA1 (RR = 30-60), College of Radiology Lung Cancer Data Registry since 2016. In 2017, a Best
RAD51C (RR = 5.0) or ATM (RR 1.5-2.0) OVCA susceptibility mutation. Practice Alert was created within our electronic health record (EHR) to alert
Respondents were queried about their preferred management approach. the primary care clinician if his/her patient met criteria for a LDCT. Each of
Chi-square test was used for statistical analysis. Results: 193 (15%) of the sites managed their own programs up until September 2018 when a
1284 SGO members responded. 58% were in academic practice. For the dedicated team (Team) of two nurses and one data support specialist was
premenopausal woman, 52%, 13% and 6% would perform an RRSO prior to justified. The Team focus is to increase awareness of the LDCT LCSP and
age 40 in the setting of a BRCA1, RAD51C and ATM mutation respectively. criteria for eligibility, improve tobacco history taking and pack year docu-
47%, 68% and 9% would perform RRSO at 40-50; and 0%, 9% and 23% mentation in the EHR, increase smoking cessation counseling and referral,
would perform RRSO at menopause, depending on the gene mutated. For the and facilitate presentation of all Lung RADS category 4 cases for review at
postmenopausal woman with comorbidities, 98%, 85% and 42% would one of our two Multidisciplinary Lung Cancer Case Conferences. Stan-
proceed with RRSO in the setting of a BRCA1, RAD51C and ATM mutation dardized management of key incidental findings was developed for coronary
respectively; 2%, 8% and 39% would observe (with/without screening); and artery calcification, non-lung masses, thoracic aortic aneurysm, and critical
0%, 7% and 19% would do further research prior to proceeding. Distribution pulmonary conditions. To date, we have not examined the impact of the
of responses for carriers of RAD51C and ATM mutations were different from LDCT LCSP on smoking cessation rates. All 9 program sites have been
BRCA1 in both vignettes [p , 0.001]. Conclusions: Respondents were more named a Screening Center of Excellence by the Lung Cancer Alliance.
likely to perform RRSO earlier and more frequently in the setting of a BRCA1 Results: In 2016, 1849 LDCT Screenings were performed, 4701 (154%
mutation compared to either a RAD51C or ATM mutation. However, there increase) in 2017 and 7154 (52.5% increase) in 2018. Cancer Detection
was lack of consensus in management of the moderate and low penetrance rates were 1.3% in 2016, 1.8% in 2017 and 1.3% for January-June 2018.
OVCA susceptibility mutations. These patterns likely reflect the limited Cancer registry data reports a 9% increase in Stage 0, 1, 2A and a 7.2%
information available regarding the timing and magnitude of risk associated decrease in Stage IV at time of diagnosis from 2014-2017. Conclusions: The
with these genes. Given the immediate and long-term morbidity of oopho- implementation of a LDCT LCSP has increased the percentage of patients
rectomy, more data regarding age-specific risks and appropriate risk- diagnosed at an earlier stage of lung cancer. With standardized management
reduction strategies for moderate and low penetrance OVCA susceptibility of key incidental findings, we anticipate improvement in early detection and
mutations is needed. management of cardiac and pulmonary diseases.

1538 Poster Session (Board #32), Mon, 1:15 PM-4:15 PM 1539 Poster Session (Board #33), Mon, 1:15 PM-4:15 PM
Analysis of healthy breast tissue from Komen Tissue Bank shows distinct Low-fat dietary pattern and breast cancer mortality by metabolic syndrome
histology, decreased proliferation, and lower periductal collagen deposition degree: Secondary analyses of the Women’s Health Initiative (WHI) Dietary
in women with prolonged history of breastfeeding. First Author: Mustafa Modification randomized trial. First Author: Kathy Pan, Los Angeles Biomedical
Basree, University of Pikeville Kentucky College of Osteopathic Medicine, Research Institute at Harbor-UCLA Medical Center, Torrance, CA
Pikeville, KY
Background: The WHI Diet Modification (DM) trial randomized 48,835
Background: Epidemiological studies have shown that prolonged breastfeed- postmenopausal women with no prior breast cancer to a low-fat dietary in-
ing is associated with a reduced risk of developing triple negative/basal-like tervention or comparison group. After 16.1 years follow-up, the intervention was
breast cancer (TN/BLBC). We have modeled abrupt involution (AI) following associated with an 18% reduction in risk of death after breast cancer (P =0.01),
short breastfeeding and gradual involution (GI) of the mammary gland that with greater reduction (29%) in those with waist circumference$88 cm (J Clin
occurs over time upon prolonged breastfeeding in wild-type FVB/N mice and Oncol 2017). To extend these findings, we examined the influence of the dietary
discovered prominent histological and molecular changes in the AI glands over intervention on breast cancer mortality in subgroups defined by number of
time. Further, we demonstrated that breast tissue from healthy women who metabolic syndrome (MS) components with 19.6 years median cumulative
breastfed ,6 months showed enrichment in stem-cell and cell renewal follow-up. Methods: WHI DM has been previously described. Four MS com-
pathways. Here, we corroborate those studies using normal human breast tissue ponents were determined at entry: 1) waist circumference$ 88 cm, 2) high
obtained from Susan G. Komen for the Cure Tissue Bank (KTB). Methods: FFPE blood pressure or anti-hypertensive use, 3) high cholesterol history and 4)
breast tissue sections obtained from KTB (Protocol #2017CO184). Donors were diabetes history, with women categorized as having 0 (n=10,639), 1-2
parous women, aged 18 to 45, without history of breast cancer and for whom (n=30,948), or 3-4 (n=4,246) MS components. Forest plots of hazard ratios
breastfeeding history was available. H&E sections and TDLU, the primary an- (HRs) were generated with P-values for interaction between randomized group
atomical source of most breast cancers, of women who breastfed for $6 months assignment and number of MS components. Results: Women with 3-4 MS
(GI, n=49) vs. those who breastfed for #3 months (AI, n=20) were evaluated by a components were more likely to be Black, be obese (BMI $30), and have
blinded pathologist. Masson Trichrome stain was used to measure collagen diabetes (all P , 0.001). Breast cancers in women with 3-4 MS components
deposition. Ki67 immunohistochemistry was utilized to determine proliferation. were less likely to be local stage (P = 0.005) or well differentiated (P = 0.03).
Statistical significance was measured using Fisher’s exact t-test and two-sample The magnitude of reduction in deaths from breast cancer in the dietary in-
t-test with a p-value of ,0.05. Results: H&E analysis revealed that breast tissue tervention vs comparison group increased as the number of MS components
obtained from women in the AI cohort exhibited histological features of in- increased (interaction P = 0.01). Hazard ratios (HR) and 95% confidence
flammation (p-value= 0.025). Using Ki67 IHC (AI, n=15; GI, n=32) and Masson intervals (CI) for death from breast cancer for intervention vs comparison groups
Trichome stain (AI, n=3; GI, n=4), sections in the AI cohort showed 2-fold for women with 0 MS components was 1.09 95% CI, 0.63-1.87, with risk low
increase in proliferation of lobular epithelium (p-value= 0.048) and 1.4-fold in both randomization groups (0.028% and 0.026%, respectively); for women
increase in periductal collagen deposition (p-value= 0.027) when compared to with 1-2 MS components, HR 0.80 95% CI 0.62-1.02; and for women with
GI cohort. Age, race, and BMI were not statistically different between AI and 3-4 MS components, HR 0.31 95% CI, 0.14-0.69, with risk in the intervention
GI cohorts. Conclusions: Breast tissue from parous women who breastfed group reduced to 0.026%. Conclusions: Adoption of a low-fat dietary pattern
#3 months is histologically different than tissue of women with $6 months had a greater effect on reducing deaths from breast cancer in women with more
history of breastfeeding. We are currently staining more breast tissue samples MS components, suggesting that this is a high risk group more likely to benefit
obtained from KTB. Experiments are underway to assess the long-term effect of from the dietary intervention. Clinical trial information: NCT00000611.
breastfeeding on breast epithelial cell hierarchy and biomarkers of inflammation.
Understanding this mechanistic link will help in developing prevention strate-
gies, particularly for African-American women who have lower prevalence of
breastfeeding and higher incidence of TN/BLBC.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
74s Cancer Prevention, Hereditary Genetics, and Epidemiology

1540 Poster Session (Board #34), Mon, 1:15 PM-4:15 PM 1541 Poster Session (Board #35), Mon, 1:15 PM-4:15 PM
Whole exome sequencing of germline DNA of individuals presenting Risk of colon adenoma in patients with breast cancer. First Author: Yinghong
extreme phenotypes of high and low risk to develop tobacco-induced lung Wang, The University of Texas MD Anderson Cancer Center, Houston, TX
adenocarcinoma (LUAD) according to KRAS status. First Author: Jose Luis
Background: In our clinical practice at a tertiary cancer center, we have ob-
Perez-Gracia, Department of Medical Oncology, Clinica Universidad de
served increased adenoma detection rate (ADR) in patients with breast cancer.
Navarra, Pamplona, Spain
Here, we describe ADR in patients with breast cancer to define the appropriate
Background: Individual susceptibility to carcinogens may depend on the genetic timing to initiate colonoscopy screening in these patients. Methods: We
background. We characterized the constitutional exome of individuals presenting conducted a retrospective study of patients with breast cancer who underwent a
extreme phenotypes of high sensitivity and resistance to develop tobacco in- colonoscopy after their diagnosis of breast cancer between 2000 and 2017. A
duced LUAD, correlating the results to KRAS status. Methods: From an iden- control group (n = 3295) comprised patients without any type of cancer who
tification cohort (n=3,631) we selected 100 caucasian heavy smokers that either underwent their first screening colonoscopy between 2008 and 2017 was used
developed LUAD at early age (cancer cohort, n=50) or did not develop LUAD or in the logistic regression. Results: Of the 62,820 patients who had a diagnosis
other tumors at advanced age (cancer free cohort, n=50). We sequenced their of breast cancer, 3304 were included. The mean age was 59 years. Regarding
germline DNA with the Agilent Human Exome Capture v5 (21,522 genes, ADR, 1803 patients (55%) had adenomas. High-grade dysplasia was evident
357,999 exons). Using logistic regression we selected the most significant in 28% of polyps and invasive adenocarcinoma in 172 (5%). The median time
variants between both cohorts and correlated them with KRAS mutation status of from breast cancer diagnosis to adenoma detection was 3 years (IQR 1-6). The
LUAD patients. Results: mean ages for the cancer and cancer free cohorts were ADR was 21% in patients younger than 40 years (n=63), 39% in patients
50 (range 34-55) and 78 years (72-90). Mean tobacco consumptions were 44 between 40 and 50 years (n=314), 54% in patients between 50 and 60 years
(range 6-72) and 55 pack-years (20-124). Median coverage was 96% at .10X; (n=1420), and 60% in patients older than 60 years (n=1507). ADR in patients
median depth was 97X. Table shows the most significant variants. rs7240666 younger than 50 years of age who do not have a family history of colorectal
(ALPK2) achieved top significance (p=8.14x10-5, OR 0.18). rs78898229 cancer or a body mass index (BMI) higher than 30 kg/m2 was 26%. A sub-
(ANKRD36C) and rs74866537 (PTPN4) were predominantly represented in sequent colonoscopy was performed in 831 patients who had colonic adenoma
patients with KRAS+ tumors, OR: 16 (3.3-78) and 11.9 (3.3-43); and in the initial colonoscopy. The ADR was 40% in patients who had a repeat
rs12426243 (CCDC41) in KRAS- tumors (OR: 13 (3-53). Conclusions: Our colonoscopy within 3 years, 50% within 3-5 years, and 53% . 5 years.
study characterizes for the first time the genotype of individuals presenting
Multivariate logistic regression analyses revealed an increased risk of colon
extreme phenotypes of high and low risk to develop tobacco-induced LUAD
adenoma with older age, male sex, higher BMI, and personal history of breast
according to KRAS status. Our results warrants further study to assess their value
cancer (P,0.05). Conclusions: In patients with breast cancer, ADR was higher
to screen these clinically relevant phenotypes; and to identify mechanisms of
than that of patients without history of cancer. Notably, breast cancer was an
high susceptibility and resistance to carcinogens.
independent risk factor for colon adenoma. In patients who are younger than 40
GENE SNP cDNA Protein
ExAC
frequency
Global
p-value
p- KRAS negative
(n=14)
p- KRAS positive
(n=14)
years of age, screening colonoscopy should be considered within five years of
ADAM15 rs11264303 c.1647A.C p.Thr549= 0.41 2.09E-04 0.001472 0.01256
breast cancer diagnosis. Multivariate logistic regression: risk factors of
PPP3R1 rs11343435 c.220+ Non coding, 0.27 4.43E-04 0.004176 0.016 adenoma.
82delA intron
ANKRD36 rs78898229 c.1194- Non coding, 0.067 2.66E-04 0.03077 0.0006357 Adenoma
63C.T intron
PTPN4 rs74866537 c.588- Non coding, 0.35 4.28E-04 0.09431 0.0001459 Characteristic OR (95% CI) P
85dupT intron
RP11- rs2230760 c.3627G.A p.Pro1209= 0.13 4.42E-04 0.001072 0.04709 Age 1.0 (1.0-1.0) , 0.01
1105G2.3
CCDC41 rs12426243 c.1212A.G p.Arg404= 0.11 4.21E-04 0.0002652 0.1428
Male sex 1.5 (1.2-1.8) , 0.01
LPCAT4 rs2279686 c.884+ Non coding, 0.46 3.58E-04 0.001522 0.05459 BMI 1.0 (1.0-1.0) 0.01
56G.A intron Family history 1.1 (0.9-1.2) 0.38
ALPK2 rs7240666 c.6469A.G p.Ile2157Val 0.87 8.14E-05 0.03009 0.01272 Smoking 1.0 (0.9-1.2) 0.53
FRG1B rs62196372 c.*17C.A 3´UTR 0.49 1.43E-04 0.008004 0.02879 Breast cancer 2.8 (2.4-3.2) , 0.01

1542 Poster Session (Board #36), Mon, 1:15 PM-4:15 PM 1543 Poster Session (Board #37), Mon, 1:15 PM-4:15 PM
The impact of smoking cessation on breast cancer patients’ survival. First Effectiveness of low-dose CT scan for lung cancer screening in the community
Author: Mazen Jizzini, The University of Texas MD Anderson Cancer Center, setting. First Author: Jason Aboudi Mouabbi, Ascension St John Hospital and
Houston, TX Medical Center, Detroit, MI
Background: Breast cancer remains to be one of the highest causes of cancer Background: Lung cancer is the leading cause of cancer death in the United
mortality amongst females globally, second only to lung cancer. Smoking is States (US) and worldwide. Chest X-ray (CXR) is ineffective in reducing lung
strongly associated with increased all-cause mortality, including breast cancer mortality. National Lung Cancer Screening Trail (NLST) reported 20%
cancer related death. It has also been shown to have a negative influence on reduction in mortality with the use of low-dose computed tomography (LDCT) scan
long-term survival after successful breast cancer treatment. Prior studies to screen high risk individuals. Therefore, major organizations including US
have shown that smoking cessation may lead to improved prognosis and Preventive Services Task Force has adopted LDCT for lung cancer screening in
better outcomes. Methods: This is a retrospective cohort study of breast high risk populations. However, The generalizability of this approach in com-
munity setting is yet to be confirmed. Our objective is to assess the ability of LDCT
cancer patients who were identified as smokers, some of who were referred to
in detection of lung nodules and lung cancer in the community setting and
the tobacco treatment program (TTP) located at MD Anderson Cancer Center. compare the results to those reported in the NLST. Methods: Charts of subjects
TTP includes careful patient screening, motivational counseling, and phar- who underwent LDCT screening between 2013 and 2016 at SJHMC were ret-
macotherapy. We complemented the original data collected by conducting in- rospectively reviewed. Demographic data, the results of the LDCT scans, in-
depth chart reviews to extract data including patient demographics, date of terventions performed, complications of procedures and pathology findings were
diagnosis, stage of cancer, smoking status, duration of abstinence and dates of collected. All cancer cases found by LDCT and the stage of cancers were
follow-up or death. We then examined associations between smoking status documented. The results of our study were statistically compared to the results of
and survival status using multinomial regression models adjusting for bio- both arms of the NLST (CT and CXR arms). Since CXR is ineffective for lung cancer
markers of disease and personal characteristics. Results: Among all breast screening, CXR arm serves equivalently to no screening. Results: The baseline
cancer patients (N = 31069), we identified those who are smokers (n = 2320) characteristics of the subjects are significantly different between this study and
by matching the TTP database with smoking status from our institutional NLST. LDCT in our study detected significantly higher positive findings. There are
electronic health records. Of those, 740 patients were referred to TTP. Amongst more cancers detected in this study compared to NLST CT and CXR arms, which
these, 242 patients quit smoking and remained abstinent at the 9 month could reflect higher incidence of cancer in this community or higher proportion of
follow-up. Compared with non-abstainers, those who quit were more likely to be current smokers in our study. In this study, LDCT detected cancers at higher stages
alive with no evidence of disease during the observation time (RR = 1.62, p = compared to that of the NLST CT arm but similar stages to NLST CXR arm. This may
0.045). When analyzed at different stages, the RR went from 1.35 (p = 0.42) to indicate that LDCT when performed in the community is less effective in detecting
cancer at early stages. Conclusions: The community population have different
2.77 (p = 0.34) for stages 3 and 1, respectively. Although the strength of this
characteristics compared those enrolled in clinical trials. This may limit the gen-
relationship varied among disease stage, the direction of the relationship re- eralizability of the results. Population-based studies are needed to confirm the
main consistent. Conclusions: Our data shows that smoking cessation is as- results of the NLST.
sociated with improved survival status amongst breast cancer survivors across
all stages. Comprehensive smoking cessation services may improve survivor- This study NLST CT arm (B) NLST CXR arm (C)
Variable (A) (N = 672) (N = 26,309) (N = 26,035) p-value
ship when started as early as the time of diagnosis. Further analysis of the
association between smoking cessation and other associated medical out- Positive result – no (%)
No 413 (61.5) 19,118 (72.7) 23,648 (90.8) A vs B , 0.0001
comes will be conducted to further determine the specific impact of cessation Yes 259 (38.5) 7191 (27.3) 2387 (9.2) A vs C , 0.0001
programs. Lung cancer –no (%) N = 672 N = 26,309 N = 26,035 A vs B , 0.0001
Yes 18 (2.7) 279 (1.1) 193 (0.7) A vs C , 0.0001
No 654 (97.3) 26,030 (98.9) 25,842 (99.3)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Cancer Prevention, Hereditary Genetics, and Epidemiology 75s

1544 Poster Session (Board #38), Mon, 1:15 PM-4:15 PM 1545 Poster Session (Board #39), Mon, 1:15 PM-4:15 PM
Improving risk assessment of obesity-associated breast cancer. First Author: Prognostic significance of blood-based cancer detection in plasma cell-free
Neil M. Iyengar, Memorial Sloan Kettering Cancer Center, New York, NY DNA (cfDNA): Evaluating risk of overdiagnosis. First Author: Geoffrey R.
Oxnard, Dana-Farber Cancer Institute, Boston, MA
Background: Elevated body mass index (BMI) is associated with increased
risk of estrogen receptor (ER)-positive postmenopausal breast cancer. The Background: Screening tests for early cancer detection are often criticized
risk is also elevated in women with a normal BMI but excess body fat. These due to risk of overdiagnosis—detection of good prognosis cancers which may
risks may be driven by breast white adipose tissue inflammation (WATi), not require immediate treatment. We recently reported development of
which is associated with elevated aromatase levels and systemic metabolic cfDNA sequencing approaches for cancer detection; longitudinal follow-up
dysfunction (e.g. hyperinsulinemia). We hypothesized that body fat as- (F/U) data were utilized here to evaluate prognostic significance of cancer
sessment is superior to BMI for detecting the pathophysiology that promotes detection using cfDNA. Methods: Plasma cfDNA samples were subjected to
obesity-related breast cancer, particularly among normal BMI women. whole-genome bisulfite sequencing (WGBS, 30X) as part of a previously-
Methods: Non-tumorous breast tissue was collected from women undergoing reported Circulating Cell-free Genome Atlas (CCGA; NCT02889978) sub-
mastectomy for breast cancer treatment or prevention. Breast WATi was study. This exploratory analysis evaluated the overall survival (OS) of training
detected by the presence of crown-like structures in the breast, which are and test set participants (pts) with cancer (20 cancer types, any stage I-IV).
composed of a dead/dying adipocyte surrounded by CD68+ macrophages. Combining train and test set pts, univariate and multivariate analyses (Cox
Body composition was measured prior to mastectomy via dual energy X-ray proportional hazards) assessed OS association with WGBS result (cancer
absorptiometry. Exercise behavior was also assessed prior to surgery using detected vs not detected, set at 98% specificity), clinical stage (IV vs I-III),
the Godin Leisure Time Exercise Questionnaire. Associations among cate- diagnostic method (symptom- vs screen-detected), sex, age, and histologic
gorical variables were examined using X2 or Fisher’s exact test. Relationships grade. Results: Of 827 pts from the training set with F/U (median 12.2 mo),
between continuous variables were examined using the Spearman corre- 334 (40.4%) had WGBS-detected cancer. Among 127 (15.4%) pts with
lation. Results: From April 5, 2016 to August 31, 2018, 100 patients were cancer that died during F/U, cancer was detected in 104 (81.9%). Results
enrolled; median age 49 (range 29 to 82) years. Breast WATi was present in were similar in the test set. In univariate analyses all variables were associated
56/100 (56%) women and was associated with elevated BMI and body fat with prognosis, including WGBS result (HR 7.7 p,0.001). In multivariate
levels, breast adipocyte hypertrophy, postmenopausal status, metabolic analyses accounting for other covariates, the three variables that most sig-
syndrome and decreased physical activity (P , 0.05). Among 39 women nificantly remained prognostic were WGBS (HR 3.0, p,0.001), clinical stage
with normal BMI, breast WATi was present in 14 (36%) and was associated (HR 3.3, p,0.001), and diagnostic method (HR 3.0, p,0.001). Validation of
with elevated body fat levels, breast adipocyte hypertrophy, dyslipidemia, these findings is ongoing in an independent cohort of ~5,000 cancer pts from
and decreased physical activity (P , 0.05). There was no statistically sig- CCGA using an optimized assay; updated performance results will be reported.
nificant association between BMI and breast WATi in the normal BMI group. Conclusions: Cancers detected using WGBS of cfDNA had a worse prognosis
Menopausal status and total fat mass had greater sensitivity and specificity than cancers not detected. WGBS cancer detection carried comparable
for the detection of breast WATi compared to a BMI-based model (AUC prognostic significance as clinical stage. By preferentially detecting higher risk
0.843 vs. 0.779, respectively). Conclusions: Measurement of body fat is cancers, cancer detection using plasma cfDNA may avoid some of the over-
superior to BMI for predicting breast inflammation, which has been shown to diagnosis that has been seen with some existing cancer screening methods.
promote obesity-related breast cancer. Clinical trial information: NCT02889978.

1546 Poster Session (Board #40), Mon, 1:15 PM-4:15 PM 1547 Poster Session (Board #41), Mon, 1:15 PM-4:15 PM
A comparison of risk factors for cigarette and e-cigarette use in the United Comparison of risk-reducing surgery in women with BRCA and non-BRCA
States adult population. First Author: Leo Chen, University of British Co- ovarian cancer susceptibility genes. First Author: Zachary Phillip Schwartz,
lumbia, Vancouver, BC, Canada Cedars Sinai Medical Center, Los Angeles, CA
Background: The US CDC and public health agencies have reported alarming increases Background: Risk reducing gynecologic surgery (RRSO) is standard of care
in e-cigarette (ecig) use among youth, even as cigarette (cig) use among youth decline. In for women with BRCA mutations. The optimal management for women with
this study, other risk factors for cig and ecig use are compared. Methods: This study used non-BRCA ovarian cancer susceptibility mutations remains unclear. We
data from the Health Information National Trends Survey 5 Cycle 1 survey, conducted in
2017. Univariate survey-weighted logistic regression analyzed responses as a nationally
sought to characterize the practice patterns for these women at our two
representative US population. Results: Inverted trends included being 35 or older (cig: institutions. Methods: Women with germline ovarian cancer susceptibility
OR=1.22, p,0.01; ecig: OR=0.79, p,0.01), being a student (cig: OR=0.77, p,0.01; genes who had a RRSO were identified from 1/2000-1/2019 in an IRB ap-
ecig: OR=1.24, p,0.01) or retired (cig: OR=1.09, p,0.01; ecig: OR=0.89, p,0.01) proved study. All patients were asymptomatic with no suspicion for malignancy
compared to being employed, and being single (cig: OR=0.92, p,0.03; ecig: OR=1.18, at time of RRSO. Clinico-pathologic characteristics were extracted from the
p,0.01). Having considered quitting smoking was not significantly associated with ecig medical records. Continuous variables were analyzed with Kruskal-Wallis and
use. Conclusions: Segments of the US adult population educated with anti-tobacco categorical variables analyzed with chi square and t-tests. Results: 152
campaigns may remain at increased risk for ecig use.
BRCA1, 95 BRCA2, and 63 Non-BRCA mutation carriers were identified—50
Survey-Weighted Univariate Logistic Regression Analysis of Cigarette and eCigs Use. Lynch (22 MLH1, 13 MSH2, 13 MSH6, 2 PMS2) and 13 Other (6 BRIP1, 2
cig ecig
OR P-Value OR P-Value RAD51C, 5 RAD51D). There was no difference between age at testing, age at
Age 18-34 R
RRSO, and interval between testing and RRSO between groups. Genetic
35-39 1.20 0.01 0.88 0.17 counseling was higher in Non-BRCA patients. Family history of ovarian cancer
40-44 1.18 ,0.01 0.82 ,0.01
45+ 1.22 ,0.01 0.79 ,0.01 was more common in women with BRCA1 and Other germline mutations
Gender Male
Female
R
0.93 ,0.01 1.00 0.92
compared to BRCA2 and Lynch. Family and personal history of breast cancer
Considered Quitting No R was high in all groups except Lynch carriers. Prophylactic mastectomy was
Yes 1.00 0.95 1.11 0.18
Harm of ecig vs cig Just as R seen mostly in BRCA mutation carriers. Concomitant hysterectomy was per-
Much less 1.25 ,0.01 1.44 ,0.01 formed in the majority of women (BRCA1 59%, BRCA2 57%, and Other 62%),
Less 1.01 0.70 1.21 ,0.01
More 1.07 0.29 1.06 0.32 with the highest frequency in Lynch carriers (86%, p,.01). Occult cancer was
Much more 1.10 0.17 1.08 0.22
Occupation Status Employed R
only seen in BRCA mutation carriers: BRCA1 (7%), BRCA2 (2%), Lynch (0%),
Unemployed 1.05 0.43 1.17 0.02 Other (0%). Conclusions: In this cohort, women with Non-BRCA mutations are
Homemaker 1.01 0.91 0.99 0.81
Student 0.77 ,0.01 1.24 ,0.01 managed similarly to women with BRCA mutations. We observed no occult
Retired
Disabled
1.09
1.22
,0.01
,0.01
0.89
1.08
,0.01
0.03
cancers in Non-BRCA patients. The optimal role of surgery as a risk reducing
Marital Status Married R strategy in this group requires further study.
Divorced 1.07 0.02 1.06 0.02
Widowed 1.04 0.16 0.93 ,0.01
Separated 1.12 0.17 1.10 0.18 BRCA1 BRCA2 Lynch Other p=
Single 0.92 0.03 1.18 ,0.01
Education No HS R Age at Genetic Testing (Median, Yr) 42 46 47 46 0.39
HS or Equal 0.96 0.30 0.88 0.03 Age at RRSO (Median, Yr) 46 47 47 47 0.42
Post-HS 0.99 0.81 0.85 0.01
Some College 0.98 0.67 1.02 0.70 Interval from Test to RRSO (Median, Yr) 0.51 0.64 0.35 0.58 0.72
College Grad 0.81 ,0.01 0.91 0.10 Documented Genetic Counseling (%) 31 32 100 85 ,.01
Post-Grad 0.80 ,0.01 0.82 ,0.01 Fam Hx of Ovarian Cancer (%) 55 29 22 69 ,.01
Race White R
Black 0.91 0.02 0.91 ,0.01 Fam Hx of Breast Cancer (%) 82 71 48 62 ,.01
Other 0.83 ,0.01 0.95 0.19 Personal Hx of Breast Cancer (%) 41 43 12 31 ,.01
Children in Household No R
Yes 0.96 0.03 1.00 0.88
Prophylactic Mastectomy (%) 68 58 0 8 ,.01

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
76s Cancer Prevention, Hereditary Genetics, and Epidemiology

1548 Poster Session (Board #42), Mon, 1:15 PM-4:15 PM 1549 Poster Session (Board #43), Mon, 1:15 PM-4:15 PM
Association between breast cancer mortality-to-incidence ratios and state Contralateral breast cancer risk according to first breast cancer characteristics
health disparities in the United States. First Author: Yu-Che Lee, University among United States women from 1992 to 2015. First Author: Cody Ramin,
of Miami, Miami, FL Division of Cancer Epidemiology and Genetics, National Cancer Institute,
National Institutes of Health, Bethesda, MD
Background: Breast cancer is the most commonly diagnosed cancer and second
leading cause of cancer deaths among women in the United States. The cancer Background: After recent advances in breast cancer treatment and increasing
mortality-to-incidence ratio (MIR) provides a population-based indicator of cancer uptake of contralateral prophylactic mastectomies, estimates of contralateral
survival and has been established previously to evaluate healthcare variations among breast cancer (CBC) risk by year of diagnosis and other patient characteristics
different countries. We aim to evaluate the association, which has not been in-
are needed to help inform decision making. Methods: We estimated CBC risk in
vestigated before, between MIR of breast cancer and state-level health disparities in
the United States. Methods: We used United States Cancer Statistics (USCS) da-
399,032 1-year survivors of a first primary breast cancer (stage I-III) in the US
tabase to calculate 6-year average of MIRs for breast cancer from 2010 to 2015. Surveillance, Epidemiology, and End Results Database (1992-2015). CBC
America’s Health Rankings (AHR) is a platform using weighted measures in 5 was defined as an invasive second breast cancer diagnosed in the opposite
different categories (Behaviors, Community & Environment, Policy, Clinical Care and breast 12+ months after the first breast cancer diagnosis. We estimated
Outcomes) to determine annual state health rankings. Six-year average (2010-2015) of standardized incidence ratios (SIRs) and 5-year cumulative incidence of CBC
health uninsured rate by state was obtained from the U.S. Census Bureau and 5-year by calendar period, age, breast cancer subtype, and receipt of hormonal
average (2010-2014) of health spending per capita by state was obtained from Centers therapy for the initial breast cancer. SIRs were calculated as the observed
for Medicare & Medicaid Services. The correlations between breast cancer MIRs and number of CBCs among survivors compared to the expected number of first
state health variables were calculated by linear regression analyses. Results: From breast cancers in the general population. Cumulative incidence was estimated
2010 through 2015, 1,390,357 females were diagnosed with breast cancer and in women without contralateral prophylactic mastectomies and accounted for
246,671 females died from breast cancer in the United States. The 6-year average of competing risks. Results: Among 399,032 breast cancer survivors, 11,365
age-adjusted incidence rate, mortality rate and MIRs were 124.2 6 1.3 per 100,000
population, 21.1 6 0.6 per 100,000 and 0.170 6 0.007, respectively. Among fifty
cases of CBC were diagnosed through 2015. Risk of CBC was elevated over the
states we included for analyses, Hawaii had the lowest MIR (0.116 6 0.014) and entire study period (SIR = 2.23, 95% CI = 2.19-2.27). SIRs for CBC declined
Nevada had the highest MIR (0.204 6 0.004). AHR showed Hawaii had the highest over calendar period and this decreasing trend was observed irrespective of age,
health ranking (No. 1) whereas Louisiana had the lowest health ranking (No. 50) in estrogen receptor (ER) status, and hormonal therapy. Survivors had an overall
2015. In our analysis, states with better health rankings, lower health uninsured rates 5-year cumulative incidence of CBC of 1.49% (95% CI = 1.44%-1.54%),
and higher health spending per capita were significantly correlated with lower MIRs which decreased over time to 1.31% (95% CI = 1.23%-1.41%) in 2008-
(R2 = 0.695, 0.453 and 0.253, respectively; all P , 0.001). Conclusions: The dif- 2014. For recent diagnoses, the 5-year cumulative incidence of CBC was
ference of MIRs for breast cancer was strongly associated with state health diversities. higher after ER-negative (1.80%, 95% CI = 1.55%-2.07%) and triple negative
These findings suggest that MIR of breast cancer can be an applicable measure to tumors (1.98%, 95% CI = 1.52%-2.55%), and lowest for women who received
evaluate and reflect the state-level health disparities in the United States. hormonal therapy (1.01%, 95% CI = 0.90%-1.13%). Conclusions: Although
2010- CBC risk is declining in the US from 1992-2015, survivors have approximately
2015 2015 2010-2015 Health 2010-2014 Health Spending Per twice the risk of an incident breast cancer (in the contralateral breast) com-
State AHR MIRs Uninsured Percentage Capita, Dollars per Year
pared to the general population. The 5-year cumulative risk of CBC is highest
Hawaii 1 0.116 6.3 6816.0 after ER-negative/triple negative tumors highlighting the need for medical
Vermont 2 0.146 6.2 9354.6
Massachusetts 3 0.137 3.7 10068.0 surveillance and targeted interventions among these patients.
Arkansas 48 0.192 14.7 6827.6
Mississippi 49 0.203 16.2 7170.4
Louisiana 50 0.193 15.9 7398.6

1550 Poster Session (Board #44), Mon, 1:15 PM-4:15 PM 1551 Poster Session (Board #45), Mon, 1:15 PM-4:15 PM
Epidemiology and survival in patients with urethral clear cell carcinoma. Association of geographic clustering of cutaneous T-cell lymphoma in the state
First Author: Mausam Patel, University of Arkansas for Medical Sciences, of Georgia with environmental exposure to benzene and trichloroethylene. First
Little Rock, AR Author: Pamela Blair Allen, Emory University, Winship Cancer Institute,
Atlanta, GA
Background: Clear cell carcinoma (CCC) of the urethra is a very rare his-
tologic variant of urethral adenocarcinoma. The majority of studies have been Background: Geographic clustering of CTCL has been recently reported in
case reports and case series with no large population based studies. A large registries, but its association with environment factors is unknown.
retrospective analysis was performed with the National Cancer Institute’s Benzene and trichloroethylene (TCE) are two common carcinogenic envi-
(NCI) Surveillance, Epidemiology, and End Results (SEER) database to ronmental toxins associated with hematological cancers. We investigated
establish the epidemiology of urethral CCC and determine the clinical factors associations between geographic clustering of CTCL incidence in the state of
associated with survival. Methods: All cases of clear cell carcinoma of the Georgia with benzene and TCE exposure. Methods: We obtained county-level
urethra diagnosed from January 1, 1973 to December 31, 2014 were extracted incidence of CTCL within Georgia from the Georgia Cancer Registry between
from SEER. Age at diagnosis, sex, marital status, race, grade, stage, surgery, 1999-2015. To account for the demographic structure in each county,
radiation, chemotherapy, overall survival (OS), disease specific survival (DSS), standardized incidence ratios (SIR) were calculated by dividing the observed
and survival months were extracted for analysis. Descriptive statistics were number of cases of CTCL in Georgia by the expected number of cases using
calculated for all variables. Univariable analysis to assess for differences in national incidence rates by age, sex, and race. Using spatial analyses, we
survival with respect to covariates was performed using the log rank test. assessed for population-adjusted county-level clustering of SIRs. We also
Multivariable analysis was performed with Cox proportional hazards regression recorded county-level exposure concentration of benzene and TCE between
models to determine the predictive performance of covariates with respect to 1996-2014 from the EPA’s National Air Toxics Assessment database. Linear
OS and DSS, reported as hazard ratio [HR] with 95% CIs. Comparisons were regression analyses on CTCL incidence were performed comparing SIRs to
considered statistically significant at P , 0.05. Results: Sixty one cases were exposure levels of benzene and TCE by county. Results: Our analyses dem-
extracted for analysis. The mean 6 SD was 63.0 6 13.9 years. Fifty eight onstrated significant geographic clustering of CTCL in Georgia (Moran’s I sta-
(95.1%) patients were female with 53 (86.8%) locoregional cases at pre- tistic 0.0991, p-value = 0.022). Local spatial tests revealed several statistically
sentation. There were 50 (82%), 18 (29.5%), and 12 (23.0%) patients who significant hot spots throughout Georgia, particularly around Atlanta. This
underwent surgery, radiation, and chemotherapy, respectively. On univariable clustering was strongly correlated with benzene (R2 0.0824, p-value 0.0006)
analysis, the following covariates were associated with both OS and DSS: age, and TCE (R2 0.0614, p-value 0.0016) exposure concentration. Among the four
stage, and surgery (all p , 0.001; log rank test). On multivariable analysis, most populous counties in Georgia (Cobb, Dekalb, Fulton, and Gwinnett) CTCL
surgery was a predictor for improved OS and DSS (HR, 0.178; 95% CI [0.068; incidence was 1.7 to 2.7 times higher than the average county, and benzene and
0.464]) and HR, 0.166; 95% CI [0.196; 2.132], respectively). Additionally, TCE exposure concentration was 3.0 to 6.3 times higher. Conclusions: These
both increasing age and distant disease were associated with worse OS and results demonstrate non-random geographic clustering of CTCL incidence in
DSS. Neither radiation nor chemotherapy were significantly associated with OS Georgia. This is the first analysis to correlate geographic clustering of CTCL with
or DSS. Conclusions: Surgery improves OS and DSS in patients with urethral environmental toxic exposures, demonstrating a statistically significant corre-
CCC. While neither radiation nor chemo were significantly associated with lation between environmental exposure to benzene and TCE and CTCL in-
survival, additional studies are necessary to determine how these therapeutic cidence within Georgia.
interventions may impact prognosis. SIR [Benzene] [TCE]
Georgia Average 0.65 0.71 0.06
County
Cobb 1.14 2.53 0.35
Dekalb 1.75 2.70 0.16
Fulton 1.34 2.74 0.20
Gwinnett 1.11 2.34 0.17

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Cancer Prevention, Hereditary Genetics, and Epidemiology 77s

1552 Poster Session (Board #46), Mon, 1:15 PM-4:15 PM 1553 Poster Session (Board #47), Mon, 1:15 PM-4:15 PM
Insulin resistance measured by the triglyceride-glucose index and risk of Insurance disparity in the United States cancer survivors’ smoking rates: A
obesity-related cancers: An epidemiological investigation in more than trend study from NHIS 2008-2017. First Author: Yannan Zhao, Fudan
500,000 individuals. First Author: Josef Fritz, University of Colorado, University Shanghai Cancer Center, Shanghai, China
Boulder, CO
Background: Smoking rates have been decreasing in the U.S over the last
Background: The role of insulin resistance as a mediator in the association of decade. Smoking cessation is a critical part of cancer treatment and sur-
body mass index (BMI) with site-specific cancer risk has, to our knowledge, vivorship care. However, little is known about the trend of smoking rates in
never been systematically quantified. We aimed to determine to what extent U.S. cancer survivors and how it varied by individuals’ insurance coverages.
insulin resistance measured as the logarithmized triglyceride glucose product Methods: We conducted a retrospective study to evaluate the temporal trend
(TyG index) mediates the effect of BMI on risk of obesity-related cancers. of smoking rates using the National Health Interview Survey from 2008
Methods: A total of 510,471 individuals from six European cohorts with a through 2017. Adult cancer survivors (n = 20122) were included in the
mean age of 43.1 years were included in the study. We fitted Cox models, analysis. The outcomes were self-reported current smoking behavior. In-
adjusted for relevant confounders, to investigate associations of TyG index with surance coverage was categorized into any private (age #65), other coverage
ten common obesity-related cancer sites, and quantified the proportion of (age #65), uninsured (age #65), Medicare + any private (age . 65), and
the effect of BMI mediated through TyG index. Results: During a median other coverage (age . 65). We combined every two years data to improve
follow-up of 17.2 years, 16 052 individuals developed obesity-related statistical power in the subgroup analysis. Weighted analyses were per-
cancers. TyG index was associated with the risk of cancers of the kidney formed with SAS 9.4 to account for the complex design. Results: The
(hazard ratio (HR) per one standard deviation increase 1.13, 95% con- smoking rates in cancer survivors decreased from 18.4% in 2008 to 12.5%
fidence interval: 1.07-1.20), liver (1.13, 1.04-1.23), pancreas (1.12, in 2017. However, the smoking rates varied remarkably by insurance status
1.06-1.19), colon (1.07, 1.03-1.10), and rectum (1.09, 1.04-1.14). (p , 0.001). There was a decreasing trend of smoking rates in participants
Substantial proportions of the effect of BMI were mediated by TyG index for with any private (age #65) (17.3% in 2008/2009 to 12.0% in 2016/
cancers of the pancreas (42%), rectum (34%), and colon (20%); smaller 2017), Medicare + any private (age . 65) (7.5% in 2008/2009 to 5.9% in
proportions for kidney (15%) and liver (11%); none for endometrium, ovary 2016/2017), and other coverage (age . 65) (13.2% in 2008/2009 to
and breast (postmenopausal). Conclusions: In this pooled cohort study 9.2% in 2016/2017) whereas the current smoking rates remains high in
including more than 500,000 individuals, insulin resistance measured as cancer survivors with other coverage (age #65) (40.1% in 2008/2009 to
the logarithmized triglyceride glucose product significantly mediated the 34.4% in 2016/2017) and uninsured (age #65) (43.4% in 2008/2009 to
effect of overweight and obesity on risk of cancers of the kidney, liver, 43,1% in 2016/2017). Conclusions: Cancer survivors report less smoking
pancreas, colon, and rectum. In contrast, insulin resistance did not me- behaviors over the last decade which is similar to the general population.
diate the risk for cancers of the endometrium, ovary and breast. Our results However, the smoking rate remains dangerously high in non-elderly cancer
confirm a promoting role of insulin resistance in the pathogenesis of survivors without any private insurance.
gastrointestinal cancers. Although often claimed, insulin resistance does
not appear to connect excess body weight with cancers of the female re-
productive organs.

1554 Poster Session (Board #48), Mon, 1:15 PM-4:15 PM 1555 Poster Session (Board #49), Mon, 1:15 PM-4:15 PM
Comorbidity and racial differences in risk of mortality of men with breast Burden of thrombocytopenia in adult cancer patients receiving chemotherapy.
cancer. First Author: Carol Parise, Sutter Institute for Medical Research, First Author: Gerald A. Soff, Memorial Sloan Kettering Cancer Center, New
Sacramento, CA York, NY
Background: Black men with breast cancer have more concomitant disease Background: Thrombocytopenia is a common toxicity of chemotherapy, yet
and worse survival than white men. Less is known about concomitant disease there are limited data on its occurrence in routine clinical practice.
and survival in Hispanic and Asian/Pacific Islander (API) men with breast Methods: Using structured patient-level data from the Flatiron Health EHR-
cancer. The purpose of this study was to compare differences in survival and derived database, we assessed risk (3-month cumulative incidence) of
risk of mortality of white, black, Hispanic, and Asian/Pacific Islander (API) chemotherapy-induced thrombocytopenia (CIT) in adult patients (2012-
men with breast cancer with increasing comorbidity. Methods: We identified 2017) based on platelet counts, overall and by each grade of CIT, cancer
1,497 cases of first primary male invasive breast cancer from the California type, and chemotherapy regimen (Table); and the co-occurrence of other
Cancer Registry 2000-2015 with a documented Charlson Comorbidity Index hematologic abnormalities. Results: Of 15,521 solid tumor patients who
(CCI). The CCI is a weighted index based on the presence of certain comorbid initiated chemotherapy, 13% had evidence of CIT within 3 months (platelet
conditions following a cancer diagnosis and weighted by the severity of these count ,100x109/L), 4% had grade 3 (25 to ,50x109/L) and 2% had grade
conditions. A score of 0 indicates no significant comorbidity and scores of 2 4 (,25x109/L) CIT. Of the solid tumors examined, incidence was highest in
or more are interpreted as a high comorbidity burden. Bivariate associations melanoma patients. In hematologic malignancies (N = 2,537), 3-month risk
between race and AJCC stage, tumor grade, estrogen receptor (ER) status, was even higher with nearly 30%, 16%, and 12% having any grade, grade 3
human epidermal growth factor 2 (HER2), and socioeconomic status (SES) and 4 CIT, respectively; and the greatest risk being in multiple myeloma
were compared using the x 2 Test of Independence. Kaplan Meier Survival patients. Anthracycline-based regimens were associated with the highest
analysis was used to compare unadjusted survival among the races. Cox risk of CIT (7% grade 3; 4% grade 4), followed by gemcitabine- and
Regression was used to assess risk of mortality for each race when adjusted platinum-based regimens. Anemia often accompanied first evidence of CIT
for factors that had a statistically significant (p , 0.10) bivariate association (49%); isolated thrombocytopenia occurred in 15%. Conclusions: This
with race/ethnicity. Analyses were conducted within each level of the CCI (0, study provides a current snapshot of CIT risk in a large sample of adult
1, and 2 or more). Results: Among men with a CCI of 0 or 2, blacks had worse patients undergoing chemotherapy in routine clinical practice, highlighting
unadjusted survival than whites. There were no differences in survival for patients at highest risk for CIT and underscoring the complexity of managing
men with a CCI of 1. Stage, SES, ER, and type of surgery all had statistically cancer treatment.
significant bivariate associations with race/ethnicity. For men with a CCI of 0,
Select chemotherapies and 3-month Cumulative Incidence %
Hispanics (HR = 0.367; 95% CI = 0.167, 0.801) and APIs (HR = 0.422; cancer types, N (95% CI): Overall Grade 3 Grade 4
95% CI = 0.189, 0.941) had a reduced risk of mortality when compared with Gemcitabine, 1,662 15.8 (14.3, 17.4) 5.5 (4.5, 6.7) 2.5 (1.9, 3.2)
whites. Black men had the same risk of mortality as white men. There were no Platinum, 8,051 13.6 (13.0. 14.2) 4.3 (3.9, 4.7) 2.1 (1.8, 2.5)
differences in risk of mortality by race for men with a CCI of 1 or 2. Anthracycline, 1,972 16.4 (15.0, 18.0) 7.1 (6.0, 8.4) 3.6 (2.9, 4.4)
Taxane, 2,116 6.6 (5.7, 7.7) 2.0 (1.5, 2.5) 0.8 (0.5, 1.3)
Conclusions: Black men with breast cancer and no comorbidity have the Multiple Myeloma, 644 37.3 (34.2, 40.6) 23.5 (20.8, 18.8 (16.4,
same risk of mortality as white men while Hispanic and API men have lower 26.6) 21.5)
risk of mortality. There are no racial disparities in adjusted risk of mortality in Non-Hodgkin’s Lymphoma, 24.4 (22.2, 26.7) 13.2 (11.5, 9.1 (7.5,
1,321 15.0) 11.0)
men with breast cancer with any concomitant disease. Melanoma, 58 21.4 (14.7, 31.2) 13.3 (6.8, 5.0 (1.8,
26.0) 14.1)
All Hematologic Malignancies, 28.2 (26.5, 30.0) 16.3 (14.8, 12.4 (11.2,
2,537 17.9) 13.8)
All Solid Tumors, 15,521 12.8 (12.3, 13.4) 4.2 (3.9, 4.6) 1.9 (1.7, 2.1)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
78s Cancer Prevention, Hereditary Genetics, and Epidemiology

1556 Poster Session (Board #50), Mon, 1:15 PM-4:15 PM 1557 Poster Session (Board #51), Mon, 1:15 PM-4:15 PM
Cardiometabolic risk factors and survival after cancer in the women’s health Breast cancer screening adherence at multiple timepoints over eight years
initiative. First Author: Michael S. Simon, Barbara Ann Karmanos Cancer among women in a familial cohort. First Author: Marcy L. Schaeffer, Johns
Institute, Wayne State University, Detroit, MI Hopkins Bloomberg School of Public Health, Baltimore, MD
Background: Features associated with metabolic syndrome have been con- Background: Since 2007, U.S. guidelines recommend cancer-free women
nected to both risk and poorer outcomes for certain cancers. Methods: We used with $20% lifetime breast cancer (BC) risk undergo BC screening with
data on 12,107 women enrolled in the Women’s Health Initiative diagnosed mammogram and breast MRI. There is limited long-term data on BC screening
prospectively with either local or regional stage cancer to evaluate the asso- adherence among young, high-risk women. To address this knowledge gap, we
ciation between cardiometabolic risk factors identified at study entry, (elevated examined utilization of multiple BC screening modalities over time.
waist circumference (WC), hypertension, high cholesterol, and presence of type Methods: Eligible women were $ 30 years old, had no history of BC/ovarian
2 diabetes), with death from either cancer, cardiovascular disease (CVD), or cancer, an intact breast, are enrolled in the Breast and Ovarian Surveillance
other causes. Cancer sites included those previously linked in the published Service (BOSS) Cohort, and visited the Johns Hopkins Cancer Genetics
literature to metabolic syndrome and cancer risk: breast, colorectal, endo- Clinic for risk assessment within 2 months of cohort enrollment (N = 374).
metrial, non-Hodgkin’s lymphoma, kidney, pancreatic, ovarian, stomach and All screening was self-reported at baseline, 4, and 8 years. A subset has
liver. Multiple imputation methods were used to account for missing data been validated. We categorized women by BC risk (Tyrer-Cuzick version)
(6.9%). Cox proportional hazards models were used to estimate hazard ratios obtained at the clinic. We examined frequency of screening over follow-up,
(HR) and 95% confidence intervals (CI), adjusted for other significant pre- and defined adherence to annual mammography and breast MRI based on
dictors of survival. Results: After a median follow-up of 10.0 years, there were age- and risk-based guidelines. We modeled the association between BC
3,612 total deaths with 1,547 (43 %) due to cancer. Most participants risk and adherence at 4 and 8 years using logistic regression. Results: At
(60.3%) had at least 2 cardiometabolic risk factors, and 5.8% had 3 or 4. baseline, the median age was 47 years, 31% had lifetime risk , 20%, and
Having 3-4 risk factors, was associated with significantly higher mortality due 69% had risk $20%. Frequency of mammography and clinical breast exam
to cancer (HR: 1.39, 95% CI: 1.12, 1.71), CVD (HR: 2.62, 95% CI: 1.91, over follow-up was . 60%, while frequency of breast MRI and breast
3.59), and other causes (HR: 1.97, 95% CI: 1.57-2.47) compared with no risk ultrasound was , 40%. Twenty-five percent of high-risk women at 4 years
factors. High WC was associated with higher mortality due to cancer and other and 40% at 8 years did not report any mammography, breast MRI, or breast
causes, and history of diabetes, and hypertension were associated with ultrasound. At 4 years, high-risk women were 85% less adherent [multi-
higher mortality due to CVD, and other causes. Conclusions: Among women variable adjusted OR = 0.15; 95%CI = 0.07, 0.34] to BC screening
diagnosed with early or regional stage cancer, the presence of 3-4 car- guidelines compared to women with a risk of , 20% due to low uptake of
diometabolic risk factors which is consistent with a diagnosis of metabolic breast MRI, while at 8 years, high-risk women were also less adherent to
syndrome, are significantly associated with death due to cancer, CVD, and mammography [multivariable adjusted OR = 0.42; 95%CI = 0.18, 0.95].
other causes. Attention to primary prevention focused on weight control, We observed similar associations among women at high-risk at 5- and 10-
physical activity and diet after cancer diagnosis and treatment, can have an years. Adherence at 4 years predicted adherence at 8 years. Interestingly,
important positive influence on survivorship after cancer. women who did not uptake MRI complied with other health screening
including for colorectal cancer. Conclusions: High-risk women were not
adherent to risk-appropriate BC screening, and adherence did not improve
over time. Low adherence appears specific to BC screening. New ap-
proaches to BC screening are urgently needed for this high-risk group.

1558 Poster Session (Board #52), Mon, 1:15 PM-4:15 PM 1559 Poster Session (Board #53), Mon, 1:15 PM-4:15 PM
Characteristics of cancer patients with a history of solid organ transplantation: Predicting future cancer incidence by age and gender. First Author:
Analysis of a patient cohort reporting to a large cancer center. First Author: Wesley B. Garner, University of Tennessee Health Science Center, Memphis,
Susen Burock, Charité Comprehensive Cancer Center, Berlin, Germany TN
Background: Solid organ transplant recipients have a 2-4-fold elevated risk Background: Cancer remains a substantial and unique burden on society.
of developing cancer compared to the general population due to immuno- While the impact of changing demographics on cancer incidence has pre-
suppressive therapy. Cancer patients with a history of organ transplantation viously been characterized (Smith et al, JCO, 2009), this has not been done
(TPx) may or may not have a cancer attributed to immunosuppression. Here with updated population data. Our objective was to update projections on the
we report on a cohort of cancer patients of the Charité, diagnosed with the number of new cancer diagnoses in the United States by age and gender
first occurrence of a solid cancer in 2010-2014 and a history of TPx. through 2040. Methods: Population-based cancer incidence data were
Methods: The cancer registry database of the Charité Comprehensive Cancer obtained using SEER 18 delay-adjusted data. Population estimates were
Center was queried for patients diagnosed with the first occurrence of solid made by age, race, and gender using the 2010 US Census data population
cancer between 2010 and 2014 and a previous solid organ transplantation. projections to calculate future cancer incidence rates. Trends in age- ad-
General tumor and patient characteristics as well as outcome were analyzed. justed incidence rates for 23 cancer types were calculated as previously
Results: A total of 226 patients (151 male, 75 female) were identified and described (Edwards et al, Cancer, 2014). Results: From 2020 to 2040 the
included in the analysis. Transplanted organs included kidney in 148 pa- projected total cancer incidence will increase by almost 30% from 1.86
tients, liver in 63, and heart, lung or a combination of both in 15. The median million to 2.4 million. This increase is due to the projected increase in
age at transplantation and initial tumor diagnosis was 54.7 years and 63.4 population growth, particularly in older individuals. The population of older
years respectively. Median interval between organ transplantation and di- adults will represent a growing proportion of total cancer diagnoses. Spe-
agnosis of the cancer was 8.1 years, range 0 - 39 years. 60.6% of cancers cifically, patients $65 years old will make up 69% of all new cancer di-
developed . 10 years after TPx. The interval between TPx and development agnoses, while 13% of new diagnoses will be in patients $85 years old by
of cancer inversely correlated to patient age at TPx, with a median of 17.9 2040 (see Table). Cancer diagnoses in females are projected to rise 27%,
years (age group 18-34), 13.0 years (35-49), 6.7 years, (50-64), and 5.3 while male cancer diagnoses are projected to increase by 32% from 2020 to
years, (65+ years). The majority of cancers occurred within the first 10 years 2040. The incidence rates for lung, colorectal, and prostate cancer are
after TPx, however especially for the age group 35-49 cancer diagnoses expected to decline, while those for thyroid, liver, melanoma and myeloma
peaked 20-25 years after transplantation. The most common cancer types are expected to increase. Conclusions: The landscape of cancer care will
where non-melanoma skin cancer (34%), followed by kidney (15%), lung continue to change over the next several decades. The burden of disease will
(13%) prostate (8.61% of all male patients) and colorectal (4.87%). Median remain substantial and will continue to disproportionately affect older
survival of all patients after tumor diagnosis was 4.6 years, with 1.0 years for adults. The growing proportion of older cancer patients and changes in site-
lung cancer, 4.6 years for colorectal cancer, 5.7 years for prostate cancer specific cancer incidence rates remain of particular interest. These pro-
respectively. The median OS was not yet reached for non-melanoma skin jections should help guide future health policy and research priorities.
cancer and kidney cancer. Conclusions: Our data support the necessity of Projected No. of Cancer Pts. From 2020 to 2040 by Age and Gender (in Thousands).
long- term follow up and cancer screening in patients after organ trans- Age 65 and Age 85 and
plantation beyond the commonly practiced 10 years post transplantation All Women Men Older Older
screening. Cancer site and year No. %* No %* No %* No. %* No. %*
All
2020 1,866 - 907 - 959 - 1,117 - 146 -
2030 2,210 18 1,057 17 1,153 20 1,482 33 199 36
2040 2,415 29 1,150 27 1,265 32 1,664 49 316 116

Incidence = Rate/100,000 *Percent change from 2020

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Cancer Prevention, Hereditary Genetics, and Epidemiology 79s

1560 Poster Session (Board #54), Mon, 1:15 PM-4:15 PM 1561 Poster Session (Board #55), Mon, 1:15 PM-4:15 PM
Health insurance literacy, financial hardship and financial sacrifices among Incidental atypical hyperplasia/LCIS in mammoplasty specimens and
cancer survivors in the United States. First Author: Jingxuan Zhao, American subsequent risk of breast cancer. First Author: Francisco Acevedo, Pontificia
Cancer Society, Atlanta, GA Universidad Católica de Chile, Santiago, Chile
Background: Rising costs of cancer care have imposed substantial financial Background: Proliferative breast lesions with atypia (atypical hyperplasia
burden on cancer survivors. To date, little is known about the associations and lobular carcinoma in-situ (LCIS)) increase the risk of breast cancer (BC).
between potentially modifiable patient characteristics, including health Most cases are diagnosed in the context of an abnormal mammogram. Little
insurance literacy (HIL), on financial burden among cancer survivors. This is known about BC risk for patients with these lesions who are asymptomatic.
study aimed to evaluate the associations between HIL and financial hardship Mammoplasty specimens allow us to study breast tissue in asymptomatic
and financial sacrifices among adult cancer survivors in the United States. healthy women. We previously published the rate of atypia in the largest
Methods: We identified 914 adult cancer survivors from the 2016 Medical reported mammoplasty cohort. The aim of this study is to examine the risk
Expenditure Panel Survey Experiences with Cancer Questionnaire. HIL of BC in the atypia cohort. Methods: Breast pathology reports were retro-
was measured based on the question “Did you ever have a problem un- spectively reviewed for evidence of atypical ductal hyperplasia (ADH),
derstanding health insurance or medical bills related to your cancer, its atypical lobular hyperplasia (ALH) or LCIS in bilateral reduction mammo-
treatment, or the lasting effects of that treatment?” Medical financial plasty specimens from five institutions within a single healthcare system
hardship was measured in three domains—1) material (e.g. problems paying between 1990 to 2017. Patients with prior or concurrent BC or prior atypia
medical bills); 2) psychological (e.g. worry about large medical bills); and 3) were excluded. Data was extracted from electronic medical records using
behavioral (e.g. delay or forego healthcare because of cost). Financial natural language processing and manual review to assess subsequent risk of
sacrifices were based on questions related to changes in spending on va- BC. Results: From our mammoplasty cohort of 4771 patients, 295 patients
cation or leisure activities. We used multivariable logistic regression mod- were found to have atypia (6.2%) at baseline. 40 of these patients were lost
eling to separately evaluate the associations between HIL problems and 1) to follow-up and excluded from the study. For the remaining 255 patients,
financial hardship and 2) financial sacrifices. Results: 18.9% cancer sur- 13 had severe ADH bordering on ductal carcinoma in situ, 52 had LCIS, 119
vivors aged 18-64 years and 14.6% survivors $65 years reported HIL had ALH, and 71 had ADH at baseline. The median age at baseline was
problems. Regardless of age groups, cancer survivors with HIL problems 52.1 (range 17.9 – 74.3). With a median follow-up of 7.7 years, of the 255
were more likely to report any material (OR =3.2; 95% CI:1.9-5.2) or patients 9 patients developed BC (8 invasive carcinomas, 1 ductal carci-
psychological (OR=7.2; 95% CI: 4.1-12.7) financial hardship than those noma in situ). 81.3% of the cohort did not receive chemoprevention. Only
without the problems, as well as more likely to delay or forgo multiple medical one patient out of the nine who developed BC received chemoprevention.
care due to cost, including prescription medicine (OR=3.6; 95% CI: 1.8- The risk of developing BC among women with atypia at baseline was 0.5%,
7.1), specialist visit (OR=2.6; 95% CI: 1.2-5.8), and follow-up care 2.9% and 4.1%, at 3, 5 and 10 years respectively. Conclusions: Patients
(OR=2.1, 95% CI 1.2-4.0). Higher likelihood of reporting all measures of with asymptomatic atypias found in reduction mammoplasty specimens
financial sacrifices were observed among those with HIL problems in both appear to be at lower risk of developing BC than those diagnosed with atypia
age groups (all p,0.05). Conclusions: Cancer survivors with HIL problems in the context of an abnormal mammogram. These results may provide
were more likely to report financial hardship and financial sacrifices than guidance on how to manage this group of patients related to future screening
those without the problems. Improving HIL may help mitigate financial and/or chemoprevention.
hardship.

1562 Poster Session (Board #56), Mon, 1:15 PM-4:15 PM 1563 Poster Session (Board #57), Mon, 1:15 PM-4:15 PM
Does ethnicity affect the relationship between body mass index (BMI) and A real-world evidence study of CDK4/6 inhibitor treatment patterns and
overall survival (OS) in non-small cell lung cancer (NSCLC)? A pooled outcomes in metastatic breast cancer by gBRCA mutation status. First
analysis of 17,326 International Lung Cancer Consortium (ILCCO) patients Author: Kimmie McLaurin, AstraZeneca Pharmaceuticals, LP, Gaithersburg,
(pts). First Author: Aline Fusco Fares, Princess Margaret Hospital, Toronto, MD
ON, Canada
Background: Limited data exist on the real-world treatment patterns and ef-
Background: In Caucasian-predominant populations, overweight or obese fectiveness of cyclin-dependent kinases 4 and 6 (CDK4/6) inhibitors in the
NSCLC pts (BMI$25-kg/m2) have better prognosis while underweight germline BRCA (gBRCA) mutated breast cancer population. Methods: Adults
(BMI#18.5-kg/m2) pts have a worse prognosis. A large pooled sample allowed with human epidermal growth factor receptor 2 negative (HER2-), hormone
us to evaluate the role of ethnicity in this BMI-NSCLC OS relationship. receptor positive (HR+) metastatic breast cancer (mBC) treated with a CDK4/6
Methods: Using individual data, survival analysis was performed on 15 ILCCO inhibitor from 01Jan2013–31Jan2018 were retrospectively selected from the
studies, assessing the interactions between ethnicity and BMI on overall Flatiron Health Oncology electronic medical record database. Patients with
survival (OS). Adjusted Hazard Ratios (aHR) from Cox models and adjusted known gBRCA status were classified as having gBRCA mutated (gBRCAm) or wild
penalized smoothing spline plots were generated. Results: Among 13416 type (gBRCAwt) disease. Time to first subsequent therapy or death (TFST) and
(77%) Caucasian, 2975 (17%) Asian, and 935 (5%) Black NSCLC pts an- overall survival (OS) were calculated from start of the earliest line of therapy with a
alyzed, we confirmed that for all pts, being underweight at NSCLC diagnosis CDK4/6 inhibitor. Kaplan-Meier (KM) medians were estimated, and TFST and OS
was associated with worse OS (aHR=1.68, CI 1.5-1.8, p,0.001), while compared between gBRCA groups with Cox models stratified by line of therapy
overweight/obese pts had improved survival (aHR 0.89, CI 0.8-0.9, p,0.001), and adjusting for demographic and clinical characteristics that modified hazard
when compared to pts with normal BMI. In general, Black pts had poorer OS ratios (HRs) for gBRCA status by . 10%. Results: Of 2968 HER2- HR+ patients
than Caucasian pts (aHR 1.26 CI 1.1-1.4, p,0.001). However, the BMI-OS with mBC receiving a CDK4/6 inhibitor, gBRCA status was known for 859
relationship differed according to ethnicity (BMI-ethnicity interaction, (28.9%). Patients with gBRCAm and gBRCAwt received letrozole plus palbo-
p=0.009): Caucasian underweight pts had poorer OS (aHR 1.67, CI 1.5-1.8, ciclib (42.4 and 39.8%, respectively), fulvestrant plus palbociclib (32.9 and
P,0.001) while overweight/obese pts had improved OS (aHR 0.89, CI 0.8- 30.7%), or other CDK4/6 regimens (24.7 and 29.5%) across all lines. The
gBRCAm group had a non-significant, shorter TFST than gBRCAwt (stratified HR
0.9, P,0.001). In Asian pts, these two associations were aHR 1.15, CI 0.8-
1.24; 95% CI 0.96–1.59). OS was significantly shorter in gBRCAm than
1.5, P=0.33, and aHR 0.91, CI 0.7-1.1, P=0.32, respectively. In Black pts,
gBRCAwt patients (stratified HR 1.50; 95% CI 1.06–2.14). Conclusions: The
these two associations were aHR 1.06, CI 0.7-1.5, p=0.73 and 0.75, CI
results of this real-world study suggest that treatment outcomes with CDK4/6
0.6-0.8 p,0.001, respectively. For overweight/obese patients, as BMI
inhibitors may be poorer in patients with gBRCAm compared with gBRCAwt
rises from 25-kg/m2 through 45-kg/m2, the prognosis worsens for Cau- disease. These findings indicate a higher unmet need among patients with
casian pts, remains stable for Asian pts, but improves for Black pts. gBRCAm, potentially requiring alternative treatment options.
Conclusions: InCaucasian pts, being underweight had a greater negative
impact on OS than for Asian or Black pts, while being obese had a greater gBRCAm
Characteristic or Outcome (N = 85; 9.9%) gBRCAwt (N = 774; 90.1%)
beneficial impact in Blacks than in other ethnic groups. These ethnic
differences likely reflect genetically-informed muscle/adipose tissue dis- Mean (SD) age, years 53 (13.4) 58 (12.0)
Line of earliest CDK4/6 use
tributions, where Black pts may have less sarcopenic obesity than other First, % 42.4 37.9
ethnicities. In future prognostic studies, BMI relationships must account TFST, months* 11 (6–18) 14 (12–15)
for ethnic differences. Second, % 31.8 32.7
TFST, months* 10 (6–11) 10 (8–12)
Third and higher, % 25.9 29.5
TFST, months* 6 (3–11) 7 (5–9)
*Data are KM median (95% confidence interval [CI])

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
80s Cancer Prevention, Hereditary Genetics, and Epidemiology

1564 Poster Session (Board #58), Mon, 1:15 PM-4:15 PM 1565 Poster Session (Board #59), Mon, 1:15 PM-4:15 PM
Risk of anxiety and depression in breast cancer survivors compared to women Risk of breast implant-associated anaplastic large cell lymphoma (BIA-ALCL)
who have never had cancer: A population-based cohort study in the United in a cohort of 3,546 women prospectively followed after receiving textured
Kingdom. First Author: Helena Carreira, London School of Hygiene and breast implants. First Author: Paola Ghione, Memorial Sloan Kettering Cancer
Tropical Medicine, London, United Kingdom Center, New York, NY
Background: Breast cancer survivors are the largest group of cancer survivors Background: Breast implant-associated anaplastic large-cell lymphoma
in the United Kingdom (UK). Having had a breast cancer diagnosis may (BIA-ALCL) is a rare subtype of T-cell lymphoma, developing in in the fluid or
adversely affect the patient’s mental health. We aimed to estimate the long- capsule surrounding breast implants, primarily or exclusively in those with
term risk of anxiety and depression in women with history of breast cancer textured surfaces. Several prior series have estimated the risk of BIA- ALCL at
compared to those who have never had cancer. Methods: We conducted a 1/6920 - 1/3800 women in retrospectively defined cohorts (from diagnosed
matched population-based cohort study, using data from the Clinical cases within national or pathology databases), approximating the population
Practice Research Datalink (CPRD) GOLD primary care database. The ex- at risk from sales records or other estimates (Sirinvasa 2017; Loch-Wilkinson
posed cohort included all adult women diagnosed with breast cancer be- 2017; de Boer 2018). Methods: A prospective cohort study was conducted
tween 1987 and 2018; the unexposed group included women with no in the population that underwent breast reconstruction by a single surgeon at
cancer history, matched to exposed women in a 4:1 ratio on primary care Memorial Sloan Kettering Cancer Center (MSKCC) from April 1993 to De-
practice and age. Cox regression models stratified on matched set were used cember 2017. Patients had long-term follow-up, and events related to
to estimate hazard ratios of the association between breast cancer survi- implants were prospectively recorded. We identified all cases of BIA-ALCL by
vorship and anxiety and depression. Results: 59,972 women (mean 62 cross-checking data from internal clinical records, pathology records, and
years; standard deviation (SD) 14.0) had history of breast cancer. The outside reports. Incidence rate per person-years and cumulative incidence
median follow-up time was 3.0 years (SD 4.4), which amounted to 256,186 when accounting for competing risk were calculated. 134 women who re-
person-years under observation. The comparison group included 240,387 ceived smooth-surface implants were excluded from the analysis, since
women followed up over 3.5 years (SD 4.5) (1,163,819 person-years). The these implants have not been associated with BIA-ALCL. Results: From
incidence of anxiety in breast cancer survivors was 0.08 (95% confidence 1993 to 2017, 3546 patients underwent 6023 breast reconstructions using
interval (95%) 0.07-0.08) per 1000 person-years, and the incidence of textured surface implants. All reconstructions were performed by a single
depression was 70 (95%CI 68-71) per 1000 person-years. The risks of both surgeon (PGC) on patients enrolled in this study. To identify BIA-ALCL
depression and anxiety were raised in breast cancer survivors compared with occurrence, clinical and pathological data were assessed from a prospective
controls, and this appeared to be driven by the first 3 years following di- database. Median follow-up was 7 years (range, 3 days - 24.7 years). Eight
agnosis (Table). Conclusions: Breast cancer survivors in the UK had sig- women developed ALCL after a median exposure of 11.2 years (range, 8.3-
nificantly higher risk anxiety and depression diagnosed in primary care for 15.8 years). Overall risk of BIA-ALCL in this cohort was 0.294 cases per
three years following diagnosis than women who never had cancer. Risk of 1000 person-years (1/443 women). Conclusions: This study, evaluating the
anxiety and depression in breast cancer survivors compared to women who risk of women with textured breast implants from a prospective database with
did not have cancer by time since diagnosis. long-term follow-up, demonstrated that the incidence rate of BIA-ALCL may
Anxiety Depression
be higher than previously reported. These results can help inform implant
Time since diagnosis (years) HR 95% CI HR 95%CI
choice for women undergoing breast reconstruction.
0-1 1.81 1.73-1.89 1.30 1.26-1.34
1-3 1.24 1.15-1.29 1.26 1.16-1.35
3-5 1.04 0.95-1.14 1.00 0.91-1.10
>5 1.01 0.96-1.07 0.99 0.95-1.03
Overall 1.30 1.27-1.34 1.19 1.17-1.21

HR – hazard ratio; 95%CI – 95% confidence interval

1566 Poster Session (Board #60), Mon, 1:15 PM-4:15 PM 1567 Poster Session (Board #61), Mon, 1:15 PM-4:15 PM
Factors associated with breast cancer mortality-per-incident case in low-to- Contemporary changes in localized and metastatic prostate cancer incidence
middle income countries (LMICs). First Author: Anees B. Chagpar, Yale Cancer by geographic area following decreased PSA screening. First Author: Daniel X.
Center, New Haven, CT Yang, Department of Therapeutic Radiology, Yale University School of Med-
icine, New Haven, CT
Background: Reducing breast cancer mortality is a key healthcare challenge
worldwide. We sought to determine the impact of macro-socioeconomic Background: In the setting of decreased PSA screening, the incidence of
factors on breast cancer age-standardized incidence (ASI) and mortality-per- metastatic prostate cancer has been increasing in the United States. This
incident case (MPI) in LMICs. Methods: Data regarding breast cancer ASI was chronologically proceeded by decreasing localized prostate cancer in-
and mortality in 78 LMICs were obtained from IARC/WHO. MPI was defined cidence. While decreased detection of localized disease is hypothesized to
as the age standardized mortality divided by the ASI. Country-level socio- increase likelihood of metastatic disease at diagnosis, it is unclear whether
economic data were obtained from World Bank, UN Development Project, the two are geographically connected. Methods: Prostate cancer incidence
and WHO data sources. Results: In 2018, the median ASI for breast cancer was obtained from the of Surveillance, Epidemiology, and End Results
was 26.5 (range; 5-67.3) per 100,000 population in LMICs, with a median (SEER) database for men 70 years or older. SEER Summary Stage 2000 was
MPI of 50.6% (range 27-70%). ASI and MPI were inversely correlated used to classify localized (local) and metastatic (distant) prostate cancers.
(Spearman rho = -0.236, p = 0.044). Factors associated with ASI and MPI Changes in incidence rates were calculated by health services areas (HSA),
are shown in the table below. We found no factor could discriminate between which each represents a relatively self-contained region of hospital care. We
the highest and lowest quartile in terms of ASI. However, all (except health chose a priori to examine most recent years 2012-2015 for changes in
expenditure as a % of GDP) were significantly different between the highest metastatic disease, and proceeding years 2008-2011 for changes in lo-
and lowest quartile in terms of MPI. Conclusions: Results suggest consid- calized disease. Population-weighted linear regression that was robust to
erable variation in terms of breast cancer MPI within LMICs. Improved rates outliers was performed. Results: A total of over 66,600 cases of localized
are seen with increasing GDP, literacy, contraceptive use, and provision of and 6,400 cases of metastatic prostate cancer from 200 HSAs were in-
doctors and mammography, but overall % GDP expended on public health cluded for analysis. From 2008 to 2011, localized incidence decreased from
does not seem to significantly influence breast cancer MPI. 613.6 to 534.2 per 100,000 men overall, and for each HSA on average
ASI MPI MPI Quartile
decreased by 30.3 per 100,000 men for each year. From 2012 to 2015,
metastatic incidence increased from 54.7 to 62.1 per 100,000 men overall,
p-
Factor Rho value Rho p-value Lowest Highest p-value and for each HSA on average increased by 2.1 per 100,000 men for each
year. Linear regression between HSA-level changes in localized and meta-
Mammography machines per pop. (2014) 0.114 0.444 -0.351 0.016 33.3 8.6 < 0.001
% adult women with BMI‡30 (2016) 0.306 0.009 -0.406 < 0.001 25.7 11.8 0.010 static disease revealed a correlation coefficient of -0.023 (SE = 0.017, p =
Income inequality (GINI coefficient); 2017 0.097 0.419 -0.784 < 0.001 0.69 0.44 < 0.001 0.16, 95% CI -0.056 to 0.009), representing lack of a statistically sig-
GDP per capita (2017) 0.278 0.022 -0.672 < 0.001 2640 670 < 0.001
Avg fertility rate per woman (2000-2005) -0.125 0.305 0.547 < 0.001 3.2 6.0 < 0.001 nificant relationship between decreases in localized disease and later in-
Contraceptive prevalence rate (%, 1995- 0.092 0.477 -0.698 < 0.001 60 9 < 0.001 creases in metastatic disease within each health services region.
2005) Conclusions: Despite concerns of increasing metastatic prostate cancer
Doctors per 100,000 pop (1993) 0.112 0.402 -0.582 < 0.001 59.0 5.5 < 0.001
Adult literacy rate (2004) 0.091 0.449 -0.548 < 0.001 79.8 27.2 < 0.001 incidence coinciding with decreases in PSA screening and localized cancer
Public health expenditure (% GDP; 2001) -0.053 0.659 -0.114 0.337 2.85 2.45 0.462 incidence, we do not observe a statistically significant geographic and
temporal relationship between metastatic and localized disease at the HSA
level. Our study is limited by short lead time and thus this trend warrants
continued surveillance.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Cancer Prevention, Hereditary Genetics, and Epidemiology 81s

1568 Poster Session (Board #62), Mon, 1:15 PM-4:15 PM 1569 Poster Session (Board #63), Mon, 1:15 PM-4:15 PM
Adverse effects of early bilateral oophorectomy on body composition: Results “Early impact” of the lung cancer screening in United States population in
from a nationally representative sample of United States women. First Author: the SEER Registries. First Author: Isabel M Emmerick, University of Massachusetts,
Pritesh S Karia, Department of Epidemiology, Johns Hopkins Bloomberg Worcester, MA
School of Public Health, Baltimore, MD
Background: The Lung Cancer Screening Trial (NLST) demonstrated im-
Background: Prior studies suggest that bilateral oophorectomy (BO), a com- proved overall survival (OS) and lung cancer specific survival (LCSS), likely
mon cancer prevention strategy, may be associated with adiposity. However, due to finding early-stage NSCLC. Our study investigates the impact of the
the impact of BO on lean mass, a potential marker of healthy aging, and whole- NLST publication in 2011 on the lung cancer outcomes in the general US
body composition is not known. Declines in lean mass have been linked to Population by assessing the incidence rates, ratio of early/late stage, and
physical disability and mortality. We examined the association between BO and lung cancer mortality in the years immediately prior to and following this
total and regional distribution of fat and lean mass in a cross-sectional study. publication. Methods: Rate sessions from the SEER18 database were
Methods: The study population included women 35-70 years who underwent accessed during the years 2008-2015. We analyzed overall lung cancer
dual-energy x-ray absorptiometry (DXA) scans at enrollment as part of the incidence and mortality rates. The ratio of early/late stage was obtained by
National Health and Nutrition Examination Survey 1999-2006 (N = 3,764). dividing the number of stage I and II cases by the number of stage III and IV
Multinomial logistic regression models were used to examine the relationship diagnosed by year. We investigate changes in level and trend using inter-
between prior BO and tertiles of fat and lean mass. Models were adjusted for rupted time series in STATA12, considering 2011 as the intervention. In
age, race, education, BMI at age 25, physical activity, smoking, alcohol use, addition, we performed a T-test for averages ratios comparing the years
parity, oral contraceptive use and hormone replacement therapy use. 2007-2010 to the years 2012-2015 for the entire lung cancer population
Results: Women with prior BO , 45 years (n = 346) had 2.9-times higher odds and for subgroups by median family, ethnicity, Sex, Age and SEER Registry.
than women without BO (n = 3,212) of being in the highest compared to the Results: Although the overall lung cancer rates remained stable during the
lowest tertile of total fat mass (OR, 2.91; 95% CI, 1.93-4.38) and 2.7-times study period, a significant increase in the ratio of early/late stage was ob-
higher odds of being in the lowest compared to the highest tertile of total lean served following the release of NLST for the overall lung cancer population
mass (OR, 2.67; 95% CI, 1.81-3.95). Results were similar when stratified by (p=0.006) and for the screening age group (p= 0.014). The effects of ratio of
age at enrollment ( , 45, 45-54, and $55). Similarly, among women with early/late stage as noted in the overall group persisted for all patient sub-
normal BMI at enrollment, those with prior BO , 45 years (n = 74) had higher groups, except for patients associated with a median income ,$40,000, for
odds of being in the highest tertile of total fat mass (OR, 9.88, 95% CI, 2.21- those there were white, and for the following regions Detroit Metro, Iowa,
44.00) and the lowest tertile of total lean mass (OR, 10.09; 95% CI, 2.72- Greater and Rural Georgia and Louisiana where no association was found
37.46). These differences in body composition were most pronounced in the between the release of the NLST changes in the ratios of early detection even
trunk region. No difference was observed in women with BO $45 years more, in some cases there was a decrease in late stage detection. There was
compared to women without BO. Conclusions: Women with a history of early no impact on lung cancer mortality in the general lung cancer population or
BO experience significant changes in body composition, including increased in any patient subgroups. Conclusions: Since the publication of the NLST in
fat mass and decreased lean mass, even while maintaining a normal BMI. If 2011, there has been no impact on lung cancer mortality or incidence of lung
validated in future prospective studies, our results suggest that a compre- cancer in the general US population. However, favorable increase in the
hensive evaluation of body composition may be warranted in young women who proportion of early stage lung cancers, depending upon median family in-
undergo BO. come, race and location. We expected a greater impact of lung cancer
screening after 2015 since CT-screening for lung cancer was adopted by
CMS and other insurances during that year.

1570 Poster Session (Board #64), Mon, 1:15 PM-4:15 PM 1571 Poster Session (Board #65), Mon, 1:15 PM-4:15 PM
Sex-specific mortality trends in adolescents and young adults with cancer Racial disparities in cancer survival: A trend analysis based on SEER data
from 2007 to 2016. First Author: Allison Close, UPMC Children’s Hospital of (1973-2010). First Author: Seyed Navid Alavi, Howard University Hospital,
Pittsburgh, Pittsburgh, PA Washington, DC
Background: Adolescents and young adults (AYA) aged 15-39 years make up Background: African Americans have higher incidence of cancer and lower
approximately 70,000 new oncology cases in the USA. Historically, mortality survival rates compared to other ethnicities. We studied the 5-year relative
from cancer has smaller incremental improvements in AYA patients when survival between black and white races for the most common cancers in the
compared to children and older adults, and not much is known about current United States. Methods: Data was obtained from the SEER database, the
sex-specific trends. We assessed overall and sex specific AYA mortality for largest population-based cancer database including 28% of US population.
the last 10 years (2007-2016). Methods: Trends in age-adjusted mortality Data containing 5-year relative survival from the patients who were di-
rates per 100,000 (1972-2016) were obtained from the CDC’s National agnosed from 1973 to 2010. We included data for cancers of colorectal,
Center for Health Statistics (NCHS). Average annual percent changes lung, prostate, breast and melanoma, the most common cancers in the
(AAPCs) in relative survival were analyzed using NCI’s JPSurv webtool and United States. Results: For colorectal cancer average 5-year relative survival
mortality AAPCs were quantified using Joinpoint regression analysis. from 1973 to 2010 is 59.9% for whites and 51.5% for blacks. Same results
Results: Overall declines in mortality are similar in AYA men and women from for lung cancer are 14.6% for whites and 12.2% for blacks, for breast cancer
1972-2016, with 54% and 51% decline, respectively. In the most recent 10 is 84.5% for whites and 71.6% for blacks, for prostate cancer is 86.9% for
years of data (2007 to 2016), combined sex AYA mortality AAPC’s declined whites and 80.5% for blacks, and for melanoma is 87.9% for whites and
by about 0.8% per year, slightly slower than declines in children ,15 years 66.4% for blacks. The average black to white 5-year relative survival ratio is
(1.3% per year) and adults ages .40 years (1.5% per year). Among AYA 0.86, 0.84, 0.85, 0.92, and 0.76 for cancers of colorectal, lung, breast,
males there have been 10 year AAPC mortality declines in leukemia (-1.8%), prostate, and melanoma, respectively. This ratio has decreased from 0.89 to
Hodgkin lymphoma (HL) (-5.1%), Non-Hodgkin lymphoma (NHL) (-4.1%) 0.86 and from 0.87 to 0.81 for colorectal and lung cancer, respectively and
and melanoma of the skin (-3.4%). For AYA females, 10 year AAPC mortality for cancers of breast, prostate and melanoma it has increased from 0.85
declines occurred in leukemia (-1.9%), ovarian (-1.5%), HL (-10%), NHL to 0.87, from 0.88 to 0.89, and from 0.73 to 0.80 respectively.
(-4.9%) and melanoma (-2.8%). These declines have been offset by stable or Conclusions: Our analysis shows that for colorectal and lung cancer the
increasing mortality rates for several common AYA cancers, including co- survival rate difference between blacks and whites has increased over 4
lorectal cancer (CRC) (1.1%) and bone and joint cancer (0.6%) in AYA decades but for cancers of breast, prostate and melanoma this ratio has
males. AYA females have experienced mortality increases for CRC (0.6%), decreased. Better understating of the factors contributing to racial differ-
bone and joint cancer (0.5%) and uterine corpus cancer (2.8%). ences in cancer survival has potential applicability in policymaking for a
Conclusions: In general, mortality rates for both AYA men and women have better and equal health care delivery.
declined over the past 10 years due to decreased mortality in hematologic
malignancies and melanoma. Despite overall improvement, tumor cate-
gories in both AYA males and female such as CRC, bone and joint cancer, and
uterine corpus cancers show increasing mortality. These diseases require
specific investigations by both pediatric and adult researchers.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
82s Cancer Prevention, Hereditary Genetics, and Epidemiology

1572 Poster Session (Board #66), Mon, 1:15 PM-4:15 PM 1573 Poster Session (Board #67), Mon, 1:15 PM-4:15 PM
Genomic profiling of pulmonary lymphoepithelioma-like carcinoma (PLELC). Relationship of liver cancer with LRP1B or TP53 mutation and tumor
First Author: Chengzhi Zhou, State Key Laboratory of Respiratory Disease, mutation burden and survival. First Author: Longrong Wang, Liver Surgery
National Clinical Research Center for Respiratory Disease, Guangzhou Institute Department, Shanghai Cancer Center, Shanghai, China
of Respiratory Health, the First Affiliated Hospital of Guangzhou Medical
Background: Liver cancer (LC), one of the most common causes of cancer-
University, Guangzhou, China
related deaths, is a serious medical problem due to the limited treatment
Background: PLELC, a rare and distinct type of primary lung cancer, is options available for this disease. Immune checkpoint inhibitors (ICIs) that
characterized by Epstein-Barr virus (EBV) infection. Histologically, it re- are proved to be beneficial in the treatment of advanced LC. Similar to
sembles undifferentiated nasopharyngeal carcinomas (NPC). Only a few observations in other cancers, these ICIs as monotherapy may benefit only a
hundred cases have been reported since its discovery. Due to the extreme fraction of HCC patients. Tumor mutation burden (TMB) is an important
rareness, its genomic landscape remains elusive. Methods: Tissue samples predictor for efficacy of ICIs in several solid tumors. However, the research on
of 27 PLELC patients (13 males and 14 females) with various stages (Ib to IV) exploring the association between TMB and mutant genes with high fre-
were subjected to targeted sequencing using a panel consisting of 520 cancer- quency of liver cancer is limited. Besides, previous research on prognostic
related genes, spanning 1.6Mb of human genome. Results: Collectively, we factors primarily focus on pathological features, probing the effects of ge-
identified 184 somatic mutations spanning 109 genes, including 107 SNVs, 12 netic variations are urgently needed to study. Methods: Whole-exome se-
insertions or deletions (INDELs) and 65 copy-number amplifications (CNAs). quencing data of 377 liver tumors from the Cancer Genome Altas (TCGA) and
Approximately, 50% of patients had CNAs. One patient had no mutation de- next generation sequencing (NGS) data of 655 liver tumors from Chinese
tected from this panel. Except for 2 patients, 1 with HER2 amplification and clinical dataset were analyzed to explore the associated between LRP1B or
another with KRAS mutation, no other classic NSCLC driver genes were de- TP53 gene alteration and TMB. TMB was defined as total number of somatic
tected. The most frequently mutated genes were CCND1, TP53, DAXX and non-synonymous mutations in coding region. Results: In total, 9.3% (35/
NFkBIA, occurring in 30%, 26%, 22% and 22% of patients, respectively. 377) of hepatic tumors in TCGA and 7.8% (51/655) in clinical cohort
Interestingly, 78% (21/27) patients had mutations in epigenetic regulators. Of harboring LRP1B mutation. LRP1B mutation was significantly associated
the 184 mutations identified, 51 occurred in epigenetics-related genes. Pathway with higher TMB in both TCGA cohort (P = 0.0003) and clinical cohort (P =
analysis also revealed an enrichment of genes participating in chromatin 0.0005). The frequency of TP53 mutation was 28.1% (106/377) in TCGA
remodeling and organization. Next, we compared the genomic profile of PLELC cohort, however, TP53 mutation represented a greater rate of 54.5% (357/
with lung adenocarcinoma and EBV positive NPC. The frequency of TP53 655) in Chinese clinical cohort. TP53 mutation was also associated with
mutations was significantly higher in lung adenocarcinoma (68% vs 26%, p = higher TMB in the two cohort (P = 0.0005 for the TCGA cohort and P =
0.021). Comparing to NPC, PLELC had significantly more mutations in epi- 0.0010 for the clinical cohort). In addition, prognosis analysis was per-
genetic regulators. TMB analysis revealed a median TMB of 1.6/Mb, significantly formed on patients in TCGA cohort. LRP1B statue resulted in significantly
lowered than lung adenocarcinomas (p , 0.01). We also assessed PD- shorter overall survival (OS, median, 20.9 vs 61.7 months; HR, 2.22; P =
L1expression and revealed that 67% had an overexpression of PD-L1. Inter- 0.0012), and a decreasing trend on progression-free survival (PFS) without
estingly,TP53-mutant patients were more likely to associated low PD-L1 significant difference (median, 8.7 vs 16.6 months; HR, 1.28; P = 0.2839).
expression (p , 0.01). Conclusions: In this study, we elucidated a distinct TP53 mutation was an independent risk factors affecting both OS (HR 1.58,
genomic landscape associated with PLELC with no classic NSCLC driver mu- P = 0.0109) and PFS (HR 1.59, P = 0.0027), respectively. Conclusions: The
tation but an enrichment of mutations in epigenetic regulators. The observation results indicated that LRP1B or TP53 mutation was a poor prognostic factor
of high expression of PD-L1 and lack of canonical druggable driver mutation in LC, and patients harboring any of these two gene mutations might easily
raises the potential of immunocheckpoint blockade therapy for PLELC. benefit from the immunotherapy.

1574 Poster Session (Board #68), Mon, 1:15 PM-4:15 PM 1575 Poster Session (Board #69), Mon, 1:15 PM-4:15 PM
A pan-cancer analysis of microsatellite instability as a predictor of Lynch Characterization of genomic alterations in Chinese LCNEC and SCLC via
syndrome in Chinese population. First Author: Pinzhu Huang, Department of comprehensive genomic profiling. First Author: Lin Wu, Department of
Colorectal Surgery, The Sixth Affiliated Hospital of Sun Yat-Sen University, Thoracic Medicine, The Affiliated Cancer Hospital of Xiangya School of
Guangzhou, China Medicine, Central South University (Hunan Cancer Hospital), Changsha,
China
Background: Microsatellite instability (MSI) and mismatch repair deficiency
(dMMR) are primarily tested in colorectal (CRC) and endometrial cancer (EC) Background: LCNEC and SCLC are aggressive neuroendocrine carcinomas
to aid Lynch Syndrome (LS) diagnosis. A pan-cancer study presented at with overlap in clinical, histopathologic, morphologic and genomic features.
2018 ASCO, however, revealed that up to 50% of LS patients had tumors not Differential molecular features between the two subtypes have not been well
typically associated with LS, suggesting that patients with an MSI-high (MSI- elucidated, contributing the uncertainty for optimal clinical strategy for each
H) phenotype should proceed to germline testing regardless of tumor type or subtype. Here we interrogated the genomic characteristics in LCNEC as
family history. We thus set out to examine this potentially practice changing compared to SCLC along with their histologically related subtypes: carci-
notion in Chinese population. Methods: MSI status and germline mutations noids and atypical carcinoids via comprehensive genomic profiling.
in MLH1, MSH2, MSH6 and PMS2 genes were determined using a targeted Methods: FFPE samples from 31 LCNECs, 35 SCLCs, 14 carcinoids and 22
next generation sequencing panel covering 100 MSI loci as well as MMR atypical carcinoids were sequenced in a CLIA-certified sequencing labo-
genes. Tumor mutation burden (TMB) levels were calculated for LS patients ratory using 520-cancer-related gene panel, with an average sequencing
with MSI-H and MSS tumors, and intergroup differences were assessed using depth of 13853. Results: Comparative mutational analysis revealed that
Mann Whitney U test or Fisher’s exact test. Results: Of 6,288 advanced tumors both LCNEC and SCLC sub-cohorts displayed higher rate of TP53 alterations
spanning . 27 cancers, 0.8% (48/6,288) were EC, 21.6% (1,362/6,288) than that of carcinoid (p , 0.001, p , 0.001). SCLC patients harbored more
were CRC, and 77.6% (4,878/6,288) were other malignancies. 3.6% (224/ RB1 and PIK3CA mutations than LCNECs (p , 0.001, p = 0.014) and
6,288) of the samples were found to be MSI-H and 3.5% (217/6,288) har- carcinoids (p , 0.001, p = 0.018). In addition, mutation frequencies of
bored MMR mutations (somatic and germline). Germline mutations indicative LRP1B, FAT1, PRKDC, NOTCH1, SPTA1, EPHA3 and KEAP1 in SCLC were
of LS were identified in 0.1% (8/6,064) of the MSS group and 17% (38/224) significantly higher than that in carcinoid. Mutations in TP53 and RB1
of the MSI-H group (p , 0.001). In contrast with the 2018 ASCO report, up to occurred concurrently in 83% (29/35) SCLC patients, whereas in only
63.8% of the 224 MSI-H tumors were CRC/EC, and only 8.9% (3/38) of the LS 32.3% (10/31) LCNECs. We further investigated the distribution of mu-
patients had MSI-H non-CRC/EC tumors (1 ovarian clear cell, 1 small bowel, tations across KEGG pathways and found that mutation frequencies in both
and 1 gastric cancer). LS patients with non-CRC/EC tumors were more likely to HIF-1 and Notch signaling pathways were lower in LCNEC than SCLC (p =
be MSS compared to those with CRC/EC (70.0% vs 2.7%, p , 0.001). Al- 0.032, p = 0.025). Copy number variation (CNV) analysis revealed that
terations in MLH1/MSH2 were present in 78.3% (36/46) of the LS patients, LCNEC and SCLC had comparable CNVs which were significantly higher than
and they demonstrated significantly better correlation with a MSI-H phenotype carcinoid (p , 0.001, p , 0.001) and atypical carcinoid (p = 0.010, p =
than MSH6/PMS2 alterations (94.4% vs. 40.0%, p = 0.0005). Additionally, in 0.028). TMB analysis also revealed a comparable TMB status of LCNEC
line with previous reports showing co-ocurrence of MSI-H and high TMB in (12.7/Mb) and SCLC (11.9/Mb), and relatively lower TMB in both carcinoid
gastrointestinal cancers, the LS patients with MSS tumors had a significantly (2.4/Mb, p , 0.001, p , 0.001) and atypical carcinoid (5.6/Mb, p = 0.003,
lower median TMB compared with the MSI-H population (4.6 muts/Mb vs. p = 0.009) than LCNEC and SCLC. Conclusions: We demonstrated the
91.8 muts/Mb, p , 0.001). Conclusions: Our study showed that in Chinese differential genomic characteristics in the four subtypes of neuroendocrine
population, CRC/EC still predominated among LS-associated cancers, while carcinomas. Compared with SCLC, LCNEC has lower mutation frequencies
non-CRC/EC LS patients were more likely to present with an MSS phenotype. in RB1, PIK3CA, as well as HIF-1 and Notch signaling pathways. In addition,
The value of MSI-H/dMMR as a predictor of LS across different tumor types LCNEC and SCLC had comparable CNV and TMB status, which significantly
warrant further investigation. higher than that of carcinoids and atypical carcinoid.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Cancer Prevention, Hereditary Genetics, and Epidemiology 83s

1576 Poster Session (Board #70), Mon, 1:15 PM-4:15 PM 1577 Poster Session (Board #71), Mon, 1:15 PM-4:15 PM
Blood-based genomic profiling of circulating tumor DNA from patients with Outcomes associated with rapid genetic testing for BRCA1 and BRCA2 at
advanced biliary tract cancer. First Author: Junying Wang, Zhongda Hospital, time of breast cancer diagnosis. First Author: Kelly A. Metcalfe, University of
Southeast University, Nanjing, China Toronto, Toronto, ON, Canada
Background: Biliary tract cancer (BTC) is a highly lethal malignancy as di- Background: Bilateral mastectomy improves survival for women with BRCA-
agnosis occurring at late stages and marginally sensitive to chemotherapy. associated breast cancer. Most women do not know their BRCA status at the
Increasing evidence indicates targeted therapeutics may provide new hope for time of BC diagnosis when making surgical decisions. Rapid genetic testing
improving clinical response in BTC, hence better comprehending the genomic (RGT) allows a woman to have genetic test results prior to treatment decision
profile is particularly important. However, tissue of BTC is highly wide tumor making, but it is unclear if RGT has an impact on treatment choices. It is also
heterogeneity and often inadequate for molecular characterization, a proper unclear if there are psychosocial implications associated with genetic testing
method is urgently needed. Circulating tumor DNA (ctDNA) is an emerging at the time of breast cancer diagnosis. The objective of the current study was
technology for detecting actionable alterations, and may be regarded as a to assess the impact of RGT at the time of BC diagnosis on surgical decision-
reliable tool to reveal genomic signature. Methods: Next-generation se- making and psychosocial functioning. Methods: Eligible women were re-
quencing (NGS) targeted 150 cancer-related genes was used to detect blood- ferred from participating surgeons at BC diagnosis. Women completed
based ctDNA from 154 Chinese patients with BTC. The mean sequencing baseline questionnaires assessing treatment preferences, cancer related
depth was more than 30003. Somatic genomic alternations (GA) including distress, anxiety, and depression. All participants received in-person pre-test
single nucleotide variation (SNV), copy number variation (CNV) and fusion were genetic counselling and genetic test results were given within 10 days.
analyzed and compared with an internal tissue genomic database (545 Chi- Participants completed surveys at 1 week and 1 year post-genetic testing to
nese patients with BTC) tested by NGS and TCGA database (n = 227) tested by assess treatments and psychosocial functioning. Results: 1010 women
the whole exome sequencing (WES). Allele frequency (AF) represented the consented to participate and 60 (5.9%) were identified with a BRCA mu-
percentage of mutant allele reads relative to total allele reads (mutant plus wild tation. 15% of those identified with a BRCA mutation did not meet provincial
type). Maximum somatic allele frequency (MSAF) was defined as the maximum eligibility criteria for genetic testing, and 20% were eligible prior to a breast
AF (0.1% , AF , 35%) of all the somatic alterations identified per sample. cancer diagnosis but had not received testing. Mean levels of cancer-related
Results: Among ctDNA database, at least one GA was found in 95% (147/154) distress, anxiety and depression declined significantly from baseline to
of samples (a median of 4 GA per patient). The median MSAF across all cases 1 year for all women (all p , .05), and there were no differences at any time
was 6.47% (range, 0%-34.8%). Pathologic type (P , 0.001) and sex (P , point between those with and without a BRCA mutation. Of those identified
0.001) were significantly related with MSAF, respectively. Frequencies of SNV with a BRCA mutation, 67.3% reported that their surgery choice changed.
in commonly mutated genes from ctDNA were similar to those observed among 73.7% of BRCA carriers had a bilateral mastectomy, compared to 20.2% of
tissue samples, like TP53 (35.1% vs 40.4%) and KRAS (20.1% vs 22.6%), BRCA negative (p , 0.001). Most women used genetic testing results for
however, a little lower in TCGA database (TP53 24.2%; KRAS 10.1%). Be- surgical decision making (96.1% of BRCA positive and 86.4% for negative).
sides, the consistency of CNV detected from ctDNA and tissue was relatively Conclusions: Rapid genetic testing for BRCA1 and BRCA2 at the time of BC
poor, and tumor heterogeneity might be in charge of this phenomenon. Among diagnosis does not have a negative impact on psychosocial functioning.
the highly frequent mutations (AF . 5%) in ctDNA, 45% of genes was There are no differences in cancer-related distress, anxiety or depression
considered as druggable targets, such as EGFR/RAS/RAF pathway and AKT/ between women who receive a positive result compare to a negative genetic
mTOR/PI3K pathway. Conclusions: These findings demonstrated that ctDNA test result. Furthermore, surgical choice changed for many women identified
tested by NGS was feasible in revealing genomic profiles and identifying with a BRCA mutation, with the majority electing for bilateral mastectomy.
potential therapeutic targets. Noninvasive ctDNA could be used as a com-
plementary approach to tissue testing in patients with metastatic BTC.

1578 Poster Session (Board #72), Mon, 1:15 PM-4:15 PM 1579 Poster Session (Board #73), Mon, 1:15 PM-4:15 PM
Early results from the BRCA Founder Outreach (BFOR) Study: Population Disparities in pretest genetic counseling among a population-based sample
genetic screening using a medical model. First Author: Kelly Morgan, of young breast cancer patients. First Author: Sonya Reid-Lawrence,
Memorial Sloan Kettering Cancer Center, New York, NY Vanderbilt University Medical Center, Nahsville, TN
Background: Barriers to population screening for BRCA mutations include Background: Per national practice guidelines, pre-test genetic counseling
access, availability of counseling, and readiness of care providers to participate (GC) through a board-certified or credentialed genetics health professional
in this process. The BRCA Founder OutReach (BFOR) study evaluates a digital (GHP) is recommended when testing for hereditary cancer. We sought to
approach to genetic testing of a defined population using a medical model and compare differences in rates of pre-test GC among young breast cancer (BC)
risk-adapted follow-up. Methods: The BFOR study (Bforstudy.com) includes patients tested with or without GHP involvement across three racial groups
web-based enrollment open to individuals in four US cities who are age 25 or (Black, Hispanic and non-Hispanic white (NHW)). Methods: A population-
older and have at least one grandparent of Ashkenazi Jewish (AJ) ancestry. based sample of Black, Hispanic and NHW women diagnosed with invasive
Participants receive web-based education, provide consent, complete ques- BC # age 50 from 2009 to 2012 were recruited through the Florida State
tionnaires, and note their preference for receiving results either from their Cancer Registry. Participants were asked to complete a baseline ques-
primary care provider (PCP) or BFOR staff. BRCA AJ founder mutation results tionnaire and release medical records for verification of clinical information
are disclosed by (e)mail or phone, depending on need for additional counseling/ and genetic testing. We compared the rates of pre-test GC across racial
genetic testing. Participants will be surveyed by email for up to 5 years; a subset groups in women tested with or without GHP involvement using Analysis of
of PCPs is also being surveyed. Results: From March 2018 to January 2019, Variance. Multivariate logistic regression analysis was also conducted to
2562 participants enrolled: 78% female; , 30 years old, 8%; 30-50 years, adjust for potential confounders. Results: Of 1618 participants, 828 had
39%; . 50 years, 53%. At enrollment, 33% requested disclosure of results by genetic testing based on medical records and/or self-reported on their
PCP. Among 847 PCPs invited to disclose results, 45% accepted, 50% de- questionnaire. There were 170 (20.5%) with GHP involvement (either
clined and 5% have yet to respond. 69 (3.2%) participants tested positive for a through consultation and/or test ordering) and the remaining 658 women
BRCA founder mutation, of whom 8 (12%) had no significant family history. (79.5%) had no documentation of GHP involvement. Among patients tested
2087 participants tested negative, of whom 6% reported a known family without GHP involvement, rates of pre-test GC were significantly lower
mutation, 38% reported a family history of breast/ovarian cancer, and 56% no among Black women (34.8%) compared to Hispanics (80%) and NHW
such history. The most common reason for study participation was referral by a (78.7%) (p , 0.001). In contrast, among those with GHP involvement, rates
friend. One individual with a distant history of breast cancer tested positive for a of pre-test GC were similar among Black (89.7%), Hispanic (81.1%) and
BRCA2 mutation and underwent risk reducing surgery that identified an early NHW (84.6%) (p = 0.89). Conclusions: Our results suggest that among
stage fallopian tube carcinoma. Her daughter then tested positive and un- young breast cancer patients tested for hereditary cancer without GHP in-
derwent prophylactic surgeries. Conclusions: Population screening of in- volvement, Blacks were significantly less likely to receive pre-test GC,
dividuals at higher risk for cancer-predisposing mutations is feasible and compared to the other two groups. In contrast, rates of pre-test GC among
identifies individuals who would not have been tested using clinical criteria. those with GHP involvement were similar across all groups. These results
Preliminary findings reveal challenges for engaging PCPs and at-risk in- suggest a disparity in receipt of pre-test GC (which is standard of care per
dividuals, particularly men. Ongoing follow-up and a second phase of the study national guidelines) among Blacks tested without GHP involvement. These
will address these barriers to testing. Clinical trial information: NCT03351803. findings are concerning given the need to offer guideline-adherent care to all
patients receiving hereditary cancer testing.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
84s Cancer Prevention, Hereditary Genetics, and Epidemiology

1580 Poster Session (Board #74), Mon, 1:15 PM-4:15 PM 1581 Poster Session (Board #75), Mon, 1:15 PM-4:15 PM
Tumor sequencing with germline genetic testing: Identification of patients Implications of incidental germline findings identified in the context of
with hereditary cancer and precision treatment eligibility. First Author: clinical whole exome sequencing (WES) for guiding cancer therapy. First
Scott T. Michalski, Invitae, San Francisco, CA Author: Bryan P. Schneider, Indiana University School of Medicine, Indi-
anapolis, IN
Background: Cancer is a fundamentally genetic disease, as such, somatic
and germline mutation analysis is used in the comprehensive assessment of Background: The identification of incidental germline mutations in the
patients with cancer. Studies report that approximately 10% of patient’s context of next-generation sequencing to guide therapy for cancer patients
tumors have clinically significant variants known to predispose to hereditary has become an unintended consequence of advancing technologies. While
cancer, with medical implications for both patients and family members. We not the primary goal, additional information buried within the genomic data
retrospectively reviewed a series of patients where providers suspected a generated could be important for family members to better understand
somatic variant also existed in the germline and followed up with clinical inherited disease risks and take action. While targeted gene panels are
germline genetic testing. We report the rate of concordance between predominantly used in practice, comprehensive WES is becoming in-
germline and somatic results and their clinical impact. Methods: Our study creasingly more common. The broader use of WES increases the complexity
used de-identified data from 1043 consecutive patients who underwent and scope of potential cancer and non-cancer pathogenic germline variation.
somatic genetic testing followed by germline testing with NGS-based he- Methods: 582 adult metastatic cancer patients from the Indiana University
reditary cancer gene panels. Results: Somatic results most frequently Health Precision Genomics Program were sequenced with tumor and
prompting germline testing included variants in BRCA2 (290), BRCA1 germline CLIA-based WES. The primary intent was for guiding cancer
(174), TP53 (158), ATM (70), MLH1 (65), APC (65), PMS2 (61), MSH6 therapy. Germline variant call files were mined for pathogenic/likely path-
(58), PTEN (54) and CDH1 (42). In 364/1043 cases (35%) the variant was ogenic (P/LP) variants using the ClinVar database. Clinvar classifications
detected as likely pathogenic or pathogenic (LP/P) in the germline. Genes were narrowed only to high quality submitters. Results: We identified 1,454
confirmed as germline variants in 60-100% of cases included: FANCA, P/LP variants in 209 genes. 92% of patients harbored at least one variant. 72
AXIN2, RAD50, MUTYH, BLM, PALB2, CHEK2, FANCD2, MITF, SDHB. P/LP variants were identified in 19 cancer predisposition genes with BRCA2
Variants in: FH, BRCA2, RET, ATM, SDHA, BRIP1, MSH2, BRCA1, BAP1, being the most common. 63 patients (10.8%) carried a P/LP variant in a
EGFR and RAD51D confirmed in the germline in 25-60%. Variants were gene that would be recommended by the ACMG to be reported due to clinical
rarely germline for TP53 (3%), APC (3%), PTEN (2%) and none in CDKN2A, actionability with the most common being ATP7B (N = 17), BRCA2 (n = 13),
NF1 and STK11. In 24 (2%) cases a LP/P germline variant was detected but MUTYH (n = 8), and BRCA1 (n = 5). We observed a preponderance of
not reported in the tumor. Conclusions: Approximately 1/3 of patients sus- patients who carried autosomal recessive non-cancer pathogenic mutations
pected to have hereditary risk after tumor testing had LP/P germline variants. of varying penetrance. Notable mutated genes included CFTR (cystic fi-
Notably, some genes had a high probability of variants occurring in the brosis, n = 16), BTD (biotinidase deficiency, n = 43), and CBS (homo-
germline, while others were primarily seen in tumors. Interestingly, 6% of the cystinuria, n = 90). Of 209 mutated genes, 173 genes had mutations present
germline variants were not included on the somatic report due to technical in less than 1% of our population, demonstrating significant genetic het-
and gene content differences in either assays or due to differences of clinical erogeneity. Conclusions: The majority of patients undergoing clinical cancer
classification between somatic and germline testing. Adding germline re- WES harbor pathogenic germline variation. Identification of clinically ac-
sults to somatic testing may inform options for precision treatment, pre- tionable germline findings will create additional burden on oncology clinics
vention, or early detection of, secondary malignancies and guide genetic as broader WES becomes commonplace.
counseling of family members.

1582 Poster Session (Board #76), Mon, 1:15 PM-4:15 PM 1583 Poster Session (Board #77), Mon, 1:15 PM-4:15 PM
Unselected germline screening in pancreatic adenocarcinoma yields high Comprehensive germline multigene panel testing changes clinical care for
rates of pathogenic and likely pathogenic variants (PV) in hereditary cancer patients with breast cancer: Untapped clinical utility and PARP inhibitor trial
susceptibility genes. First Author: Carol Cremin, Hereditary Cancer Program, eligibility. First Author: Peter D. Beitsch, Dallas Surgical Group, Dallas, TX
BC Cancer, Vancouver, BC, Canada
Background: HBOC testing guidelines were established to identify patients
Background: Recent literature cites a germline mutation rate of 3.9-15.1% with clinically actionable variants and limit economic burden. We report the
in patients diagnosed with pancreatic ductal adenocarcinoma (PDAC) impact of germline results on health outcome based on clinical decision
depending on breadth of genes tested, pre-selection of high-risk history and making and treatment interventions, regardless of guidelines, in a multi-
population substructure. True incidence of germline mutations in PDAC is center registry. Methods: 20 community-based and academic participated
unknown in unselected population. Methods: All patients (pts) diagnosed in an IRB approved registry. Patients with breast cancer were tested with an
with PDAC in the province of British Columbia, Canada and referred to the 80-gene panel; clinical information was collected. Results: Data on 912
Hereditary Cancer Program, were eligible to undergo 30 gene Color saliva kit patients has been analyzed to date. 50.5% met NCCN criteria; 49.5% did
testing under a research protocol, or clinical multigene testing if they met not. Pathogenic/likely pathogenic (P/LP) germline mutations were found in
existing local criteria regardless of whether they signed on to the protocol. 8.65% of patients. Of all patients with P/LP findings, 85% had variants in
Any healthcare provider or patients themselves could refer. Results: 243 pts cancer-risk genes with established management recommendations and 80%
were referred between August 2016 and October 2018 but 25.1% (61) had germline variants conferring eligibility for clinical trials and precision
declined and 9.1% (22) died before testing. Of the 141 pts who consented to medicine-based cancer treatments, such as PARP inhibitors. For 62% of
research protocol and completed germline testing, median age was 64 patients with P/LP germline mutations, clinicians reported results impacted
(46.1-81.0), 68.8% were European, 1.4% Ashkenazi Jewish heritage, 42% patients’ health outcome. And for 69%, results impacted the health outcome
male, 61% non-smoker, 24.1% had personal history of a second cancer and of patients’ relatives. There was no significant association between BRCAPRO
39% had metastatic disease. Baseline characteristics were similar between scores and patients having a P/LP finding, whether in BRCA1/2 alone (p =
the PV positive and uninformative group. 20/25 PV were in known PDAC 0.42) or for any cancer gene (p = 0.57). Physician reported impact on patient
susceptibility genes with cascade screening implications (ATM (9), BRCA2 outcome associated significantly with the presence of P/LP germline findings
(4), BRCA2/ATM (1), CDKN2A (4), and MSH2 (2)). Excluding the 3 PV (p , 0.00001). There was no significant difference in the clinician reported
identified through carrier testing (1) and prior research identification (2), the clinical utility of variants of uncertain significance (VUS) compared to negative
rate of PV in unselected, unrelated PDAC cohort is 22/138 (15.9%). Uti- results (p = 0.467). Conclusions: Comprehensive panel testing of breast
lizing previous NCCN criteria for BRCA1/BRCA2 testing or for familial cancer patients impacts physician assessed patient outcomes and informs
pancreatic cancer did not appear to select for patients with higher risk of PV changes in surgical treatment strategy, medical therapies and proactive
positive (12/65 (18.4%) PV positive rate versus 13/76 (17.1%) in those that screening. The data suggest that BRCAPRO calculators are poor predictors of
didn’t meet criteria). Previous criteria would have missed 52% (13/25) PV in germline presence of P/LP findings. Physicians in this study demonstrate the
ATM (6), BRCA2/ATM (1), BRCA2 (2), MSH2 (2), NBN (1), CHEK2 (1). To ability to discern the clinically actionable value of P/LP mutations from non-
date, a third of families with PV identified have accessed cascade testing in actionable VUS. Multigene panels impact breast cancer patient care by
38 relatives. Conclusions: Given the high incidence of 15.9% PV in he- identifying precision medicine treatment interventions, and guiding long-term
reditary cancer susceptibility genes, our data support recommendations for medical management and preventive surveillance for patients and family
universal germline genetic testing of PDAC pts. members.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Cancer Prevention, Hereditary Genetics, and Epidemiology 85s

1584 Poster Session (Board #78), Mon, 1:15 PM-4:15 PM 1585 Poster Session (Board #79), Mon, 1:15 PM-4:15 PM
Germline mutation profile among Hispanic women with epithelial ovarian Genomic testing and treatment landscape in patients with advanced non-small
cancer (EOC). First Author: Yanin Chavarri Guerra, Instituto Nacional de cell lung cancer (aNSCLC) using real-world data from community oncology
Ciencias Medicas y Nutricion Salvador Zubiran, Mexico City, Mexico practices. First Author: Hinco Jasper Gierman, Integra Connect, West Palm
Beach, FL
Background: Hospital-based studies have reported a 15% prevalence of
BRCA1/BRCA2(BRCA) mutations, with a slightly higher yield of other pre- Background: While aNSCLC is a leading cause of US cancer deaths, targeted
disposition genes on multigene panel testing (MGPT) among women with therapies and immune checkpoint inhibitors (ICPi) have emerged as im-
EOC, and National Comprehensive Cancer Network guidelines recommend portant new treatment options for these pts NCCN guidelines recommend
genetic cancer risk assessment for women with EOC. However, there is testing of eight genes in aNSCLC patients at diagnosis. Targetable alter-
limited data about the genetic epidemiology of EOC among under- ations (TA) in four genes, EGFR, ALK, ROS1, and BRAF, are associated with
represented populations, such as Hispanics. Consequently, we deter- FDA-approved therapies. The labels for ICPis indicate that pts with TAs in
mined the germline mutation profile of Hispanics with EOC, and compared EGFR and ALK are not candidates for first line treatment with ICPi.
them with non-Hispanics. Methods: We included all women with a personal Methods: The Integra Connect database, which includes electronic medical
history of EOC from the U.S. and Latin America (LatAm; Mexico, Colombia, record (EMR) and claims data on approximately 600,000 cancer patients, was
and Peru), enrolled in the Clinical Cancer Genomics Community Research queried across five community oncology practices (289 oncologists) to identify
Network registry. We assessed the prevalence of pathogenic variants (PV) in aNSCLC patients (stage 3B or 4) treated since January 2017. Manual review of
BRCA1/BRCA2(BRCA) and other genes, contrasting the germline mutation charts was done to abstract tumor type/stage, drug regimens, and evidence of
profile between Hispanics living in LatAm, U.S. Hispanics, women of somatic genetic testing. A Wilcoxon rank sum test was used to test difference in
Ashkenazi Jewish (AJ) ancestry in the US, and other U.S. non-Hispanics. time to results (TTR) for blood- vs tissue-based tests. Results: A total of 1,203
Results: Among 1186 women with EOC (209 from LatAm, 254 U.S. His- aNSCLC patients were identified. Testing rates varied from 54% for EGFR to
panics l, 78 AJ, and 645 other non-Hispanic), 262 (22%) had a PV in 22% for all 4 genes (table). 163 patients had a TA in EGFR, ALK, ROS1 or
BRCAgenes. Hispanics from LatAm and the U.S. had a similar frequency of BRAF, and 55% of these pts did not receive targeted therapy. 84 pts with TA’s
BRCAmutations to AJ (30.6%, 29.9%, and 38.4%, respectively; p = 0.14); in EGFR or ALK had no evidence of progression on targeted therapy, yet 31
while non-Hispanics showed a significantly lower frequency of BRCA- (37%) received an ICPi; 24% had the TA test result prior to ICPi use and 13%
mutations (14.2%, p = 0.03). The most frequently mutated gene was received the TA result after starting ICPi. Median TTR for blood-based somatic
BRCA1(n = 197, 74.6%), followed by BRCA2(n = 67, 25.3%). Among tests was shorter than tissue-based tests (4 vs 14 days, p-value= 3.5-e07).
BRCA-negative cases (n = 924), 59% (n = 545) were evaluated by MGPT and Conclusions: Our analysis in the community oncology setting for aNSCLC pts
PVs were identified in 2.9% [6 Hispanics (1.2%), 3 AJ (3.8%) and 26 Non- finds evidence of underutilization of genomic testing, underutilization of
Hispanics (4%)]), of which 66% (n = 23) were in mismatch repair genes targeted therapies, and ICPi use outside of label. Further research is needed to
(MSH2, MLH1, MSH6, PMS2), and 34% (n = 12) in other EOC-associated identify strategies to increase testing in aNSCLC pts to provide physicians with
genes (BRIP1,NBN,PALB2, RAD51C, and RAD51D). Clinically actionable the information needed to make optimal treatment decisions.
PVs in ATM (n = 4; 0.3% ) and CHEK2 (n = 6; 0.5% ) were also observed. Patients with TA in EGFR, ALK, ROS1, Patients with TA in in EGFR, ALK
Conclusions: Hispanics with EOC have an elevated frequency of PV, similar BRAF (n=163) without TKI progression (n = 84)
Patients tested %
to that of classic founder populations such as AJ, and significantly higher Gene (n= 1,203) On targeted therapy No targeted therapy No ICPi TA prior to ICPi TA after ICPi
than other non-Hispanics. This is partially explained by a high prevalence of All 4 Genes (T4) 22% 45% 55% 63% 24% 13%
recurrent LatAm-specific PV, highlighting the importance of conducting EGFR 54%
ALK 51%
genetic studies in underrepresented populations. There was modest in- ROS1 43%
cremental benefit of MGPT. BRAF 29%

1586 Poster Session (Board #80), Mon, 1:15 PM-4:15 PM TPS1587 Poster Session (Board #81a), Mon, 1:15 PM-4:15 PM
Incidence of second primary neoplasms among cancer survivors in the A phase II study of PD-1 inhibition for the prevention of colon adenomas in
United States, 2000 through 2015. First Author: Eric Adjei Boakye, patients with Lynch syndrome and a history of partial colectomy. First
Saint Louis University Center for Health Outcomes Research, St. Louis, MO Author: Joanne M. Jeter, The Ohio State University, Columbus, OH
Background: The number of cancer survivors in the United States is projected Background: Colon cancers and adenomas that are associated with Lynch
to exceed 20 million by 2024. Survivors are at risk of developing a second syndrome (LS) often display microsatellite instability (MSI), a characteristic
primary neoplasm (SPN) – a leading cause of survivor death. We described the that is associated with increased response to treatment with PD-1 inhibitors.
risk of developing a SPN among survivors of common cancers (smoking-related Because LS patients with a history of colon cancer are at increased risk of
vs non-smoking-related) in the United States. Methods: We identified patients having a second primary colon cancer or high-risk adenoma, preventive
aged $18 years who were diagnosed with a primary cancer from the 10 sites measures are of particular interest in this population. We hypothesize that a
with highest survival rates and stratified as smoking-related (urinary bladder, maintenance schedule of nivolumab can be safely administered to LS pa-
kidney & renal pelvis, uterine cervix, oral cavity & pharynx, and colon & rectum) tients with a history of treated colon cancer with remaining colon at risk in
and non-smoking related (prostate, thyroid, breast, corpus & uterus, and order to decrease the incidence of adenomas, advanced adenomas and
non-Hodgkin lymphoma) from Surveillance, Epidemiology, and End Results second primary colon cancers. Methods: OSU 17198 is a phase II multi-
(2000-2015). SPN was defined as the first subsequent primary cancer center, single-arm study of nivolumab in patients with germline MLH1 or
occurring $2 months after first cancer diagnosis. Excess SPN risk was MSH2 mutations and a history of hemicolectomy for colon cancer at least
quantified using standardized incidence ratios (SIRs) stratified by sex. one year prior to study entry. Subjects must have completed any adjuvant
Results: A cohort of 2,908,349 patients (50.1% female) was identified and therapy at least 6 months prior to study participation and may not have
260,267 (9.0%) developed SPN (7.6% of females and 10.3% of males). All received prior therapy with a PD-1 inhibitor. Nivolumab is given at 240mg IV
index cancer sites were associated with a significant increase in SPN risk for every 3 months for two years, and colonoscopies will be performed prior to
females and males (except prostate cancer). Index smoking-related cancers study entry, after the fourth dose, after the eighth dose, and one year after the
(SIR range 1.20 – 2.17 for females and 1.12 – 1.91 for males) had higher eighth dose. Subjects will be monitored for auto-immune adverse effects.
increased risk of SPN than non-smoking-related cancers (SIR range 1.08 – The primary endpoint is incidence of adenomas at three years, and secondary
1.39 for females and 0.55 – 1.38 for males) relative to the general population. endpoints are safety, incidence of advanced adenomas, and incidence of
Conclusions: Nearly 10% of cancer survivors developed an SPN, and those colon and non-colon cancers at three years. Approximately 104 subjects will
with smoking-related cancers had higher risk. Given the increasing number of be enrolled to obtain 94 evaluable subjects. This study is currently open for
cancer survivors and importance of SPN as a cause of cancer death, these enrollment at the Ohio State University and at various stages of activation at
findings can improve secondary prevention and surveillance guidelines. seven additional sites in the United States. Enrollment of this study is
Men Women
anticipated to be completed in 2020, and data collection is anticipated to be
Index cancer Observed SPN SIR (95% CI) Index cancer Observed SPN SIR (95% CI)
complete in 2023. This study has undergone safety review by the FDA and
Smoking-Related Smoking-Related
the Ohio State University Institutional Review Board. Clinical trial in-
Urinary bladder 25,211 1.77 (1.75, 1.79) Urinary bladder 5,283 1.65 (1.60, 1.69) formation: NCT03631641.
Kidney & renal pelvis 9,461 1.53 (1.50, 1.56) Kidney & renal pelvis 4,448 1.50 (1.45, 1.54)
Oral cavity and pharynx 9,077 1.91 (1.87, 1.95) Uterine cervix 2,488 1.50 (1.44, 1.56)
Colon & rectum 21,375 1.12 (1.11, 1.14) Oral cavity and pharynx 3,444 2.17 (2.10, 2.24)
Non-Smoking-Related Colon & rectum 15,825 1.20 (1.18, 1.22)
Prostate 71,866 0.65 (0.65, 0.66) Non-Smoking-Related
Thyroid 1,722 1.23 (1.18, 1.29) Thyroid 3,730 1.12 (1.08, 1.16)
Non-Hodgkin lymphoma 10,011 1.38 (1.36, 1.41) Breast 56,725 1.17 (1.16, 1.18)
Corpus & uterus, NOS 10,131 1.08 (1.06, 1.10)
Non-Hodgkin lymphoma 7,239 1.39 (1.36, 1.42)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
86s Cancer Prevention, Hereditary Genetics, and Epidemiology

TPS1588 Poster Session (Board #81b), Mon, 1:15 PM-4:15 PM TPS1589 Poster Session (Board #82a), Mon, 1:15 PM-4:15 PM
Phase I trial of endoxifen gel versus placebo gel in women undergoing breast Omega-3 fatty acids and ERPR(-) and HER2/neu(+/-) breast cancer
surgery. First Author: Katrina Alber, Northwestern University, Chicago, IL prevention. First Author: Chidimma Kalu, City of Hope, Duarte, CA
Background: Despite large Phase III clinical trials that have established the Background: The most aggressive breast cancer subtypes tend to be estrogen
success of selective estrogen receptor modulators (SERMs) for breast cancer and progesterone receptor negative (ERPR(-)) and human epidermal growth
prevention and therapy of duct carcinoma in situ (DCIS), acceptance by factor receptor type 2 positive (HER2(+)). Women with these breast cancer
women likely to benefit has been low, primarily because of toxicity related to subtypes (triple negative, ERPR(-)HER2(+)) tend to experience worse
systemic exposure. Local drug delivery to the breast in gel form is an at- clinical outcomes and have a relatively higher risk of recurrence. Our pre-
tractive alternative since low systemic levels could minimize toxicity. vious research demonstrated that dietary omega-3 (n-3) fatty acids can
Endoxifen (ENX) is an active metabolite of tamoxifen, that has unique ac- significantly inhibit ERPR(-) HER2(+) tumorigenesis in MMTV-HER2/neu
tivity compared with 4-hydroxytamoxifen (4-OHT). It is smaller and more transgenic mice fed fish oil vs corn oil-based diets. The fish oil diet group
polar than 4-OHT making it potentially more suitable for transdermal de- developed 30% fewer breast tumors, had lower Ki67 expression, and ex-
livery. The NCI PREVENT program has developed ENX transdermal alcoholic perienced less mammary atypia relative to the corn oil diet group. Other
gel products. Methods: We are conducting a randomized, double-blinded, studies involving diet, nutrition and breast cancer point to the potentially
placebo-controlled, Phase I trial to establish the dermal tolerability and protective effect of an anti-inflammatory diet on the risk of developing ER(-)
safety of endoxifen (ENX) gel. 38 women planning unilateral or bilateral and HER2(+) breast cancer, further supporting the evidence that the ERPR(-),
mastectomy will be enrolled across 3 institutions in 3 cohorts: (a) ENX gel HER2(6) subtypes may be highly responsive to this bioactive nutrient.
10mg (N = 8) vs. placebo gel (N = 4) daily; (b) ENX gel 20mg (N = 8) vs. Methods: This is a double-blinded, randomized clinical trial of high dose
placebo gel (N = 4) daily; (c) the maximum tolerated dose (N = 8) with last (~5.4 g EPA+DHA) vs low dose (~0.9 g EPA+DHA in fatty acid mix of the
dose 72 hours prior to surgery. Treatment duration will be 4 6 1 weeks. All typical American diet) of n-3 fatty acids in breast cancer survivors of ERPR(-),
participants will be evaluated for toxicity and skin tolerability. Secondary HER2(6) breast cancer. Eligible participants will take 5 capsules/day for
endpoints include TAM metabolite measurements in breast tissue and 12 months, with cellular samples of breast epithelial and/or adipose tissue
plasma; serum hormone concentrations, serum estrogenic response, obtained by fine needle aspirations of the contralateral breast. The study
changes in coagulation parameters, gene expression changes reflective of aims to determine whether n-3 fatty acid supplementation will modify fatty
therapeutic effects, and experienced symptoms. 65 potential participants acid metabolite content in breast adipose tissue, modulate cytomorphology
have been pre-screened for eligibility. 33 were ineligible prior to contact, and/or cell proliferation in breast epithelial cells, affect DNA methylation
most commonly due to the use of neoadjuvant chemotherapy. Of 32 po- patterns, and modulate pro- vs anti-inflammatory gene expression patterns in
tential participants who have been eligible to be contacted, 21 did not breast adipose tissue. Correlative aims will evaluate possible associations
consent for screening, most commonly because they were too overwhelmed between factors such as breast adipose tissue, red cell membrane fatty acid
with their recent diagnosis. 7 have consented, 4 are pending consent, and 6 profiles, BMI, and reported dietary intake. Sample size of 40 participants per
have started study intervention. No adverse events have been reported to arm was calculated to provide at least 80% power to detect a statistically
date. This pre-surgical trial testing transdermal ENX for breast cancer significant difference for each primary endpoint. This study focuses on women
prevention is accruing as projected. The results will establish the skin safety survivors of high risk breast cancer subtypes, specifically triple negative or
of this agent, provide data on skin permeability, and the duration of drug ERPR(-)HER2(+) disease, who are currently without long term adjuvant op-
retention in the breast. Clinical trial information: NCT03317405. tions. Eligibility criteria include prior diagnosis of ERPR(-) stage 0 to III breast
cancer, #5 years from completion of standard therapy.The study was closed to
accrual in November 2018; less than 9 months of follow-up remain for active
study participants. Clinical trial information: NCT02295059.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Central Nervous System Tumors 87s

2000 Oral Abstract Session, Mon, 1:15 PM-4:15 PM 2001 Oral Abstract Session, Mon, 1:15 PM-4:15 PM
Second interim and first molecular analysis of the EORTC randomized phase Randomized phase IIb clinical trial of continuation or non-continuation with
III intergroup CATNON trial on concurrent and adjuvant temozolomide in six cycles of temozolomide after the first six cycles of standard first-line
anaplastic glioma without 1p/19q codeletion. First Author: Martin J. treatment in patients with glioblastoma: A Spanish research group in neuro-
Van Den Bent, Erasmus MC Cancer Centre, Rotterdam, Netherlands oncology (GEINO) trial. First Author: Carmen Balana, Institut Catala
Oncologia Badalona, Hospital Germans Trias I Pujol, Badalona/Barcelona,
Background: The 1st interim analysis of the CATNON trial showed benefit from
Spain
adjuvant (adj) temozolomide (TMZ) on overall survival (OS) but remained in-
conclusive about concurrent (conc) TMZ. A 2nd interim analysis was planned Background: The GEINO-14-01 trial (NCT02209948) investigated the role
after 356 events. Methods: The 2x2 factorial design phase III CATNON trial of extending temozolomide (TMZ) for 6 cycles after the standard 6 cycles
randomized 751 adult patients with newly diagnosed non-codeleted anaplastic to improve 6m-PFS, SLP and OS in newly diagnosed glioblastoma (GBM)
glioma to either 59.4 Gy radiotherapy (RT) alone; the same RT with concTMZ; patients (p). Methods: Between 08/2014 and 11/2018, 166 p were screened
the same RT and 12 cycles of adjTMZ or the same RT with both concTMZ and and 159 randomized to extend (80p) or not (79p) TMZ treatment for 6 cycles
adjTMZ (doi: 10.1016/S0140-6736(17)31442-3). MGMT promoter meth- after proving stable disease in the MRI performed before inclusion. Central-
ylation (MGMTmeth) status was re-assessed with the Infinium Methylation ized review of histology and determination of MGMT status, if not previously
EPIC Beadchip using the MGMT_STP27 model. Isocitrate dehydrogenase 1 available, were performed before randomizing patients. Two criteria of strat-
and 2 (IDH) mutation (mt) status was assessed with glioma targeted Agilent ification were used: MGMT status and presence/absence of residual disease on
SureSelect baits sequence using an Illumina HiSeq2500 Rapid PE100. the basal MRI (defined as a residual enhancement larger than 1cm in one). The
Results: With a median follow-up of 56 months and 356 events, the hazard primary endpoint was differences in 6mPFS, secondary endpoints were dif-
ratio (HR) for OS adjusted for stratification factors after concTMZ was 0.968 ferences in PFS, OS, toxicity, between arms and per stratification factors.
(99.1% CI 0.73, 1.28). 5-year OS was 50.2% with and 52.7% without Results: Median age was 60.3 (range 29-83), 97p (61%) were methylated,
concTMZ (95% CI [44.4, 55.7] and [46.9, 58.1]). An IDHmt was found in 335 basal MRI showed residual disease in 57p (35.8%). After a median follow up of
of 480 assessed cases (70%). Median OS was 19 mo (95% CI 16.3, 22.3) in 14.0 months, with 121 p(76.1%) already progressed and 81p (50.9%) already
IDHwt tumors and 116 mo (95% CI 82.0, 116.6) in IDHmt tumors. HR for OS dead, median PFS is presented. Median (m) PFS is 8.0 months (95%CI: 5.7-
after concTMZ in patients with known IDH status. Clinical trial information: 10.2). There is no difference in mPFS between arms (adjusted HR = 0.98,
NCT00626990. IDHmt was predictive of benefit from adjTMZ (IDHmt HR: 95% CI: 0.82-1.18, P = 0.907). Methylated tumors had longer mPFS
0.41, 95% CI 0.27, 0.64; IDHwt: HR 1.05, 95% CI 0.73, 1.52; interaction (HR=0.57, 95% CI: 0.39-0.83, P=0.004) irrespectively to the study treat-
test p = 0.001). In IDHmt patients that received adjTMZ, the HR for OS after ment. Conclusions: There is not apparent benefit of continuing TMZ treatment
concTMZ was 0.71 (95% CI 0.35, 1.42, p=0.32). MGMTmeth was found in for more than 6 cycles. Data will be actualized for the congress.Supported by a
288 of 410 assessed cases (70%), interaction test for concTMZ (p = 0.092) Grant of the ISCIII: PI13/01751. Clinical trial information: NCT02209948.
and adjTMZ (p = 0.166) did not reach statistical significance. Conclusions: In
the entire study cohort, concTMZ did not increase OS. However, in IDHmt
tumors a trend towards benefit of concTMZ is present. AdjTMZ increased OS in
IDHmt but not in IDHwt tumors. The ongoing molecular analyses and further
follow-up will allow full assessment of efficacy in the molecular subgroups.
Patients n events HR [95% CI] interaction test
IDH
wt 145 120 1.27 [0.89, 1.82] p = 0.19 p = 0.016
mt 335 92 0.67 [0.44, 1.03], p = 0.06

2002 Oral Abstract Session, Mon, 1:15 PM-4:15 PM 2003 Oral Abstract Session, Mon, 1:15 PM-4:15 PM
Updated predictive analysis of the WHO-defined molecular subgroups of A phase I, open label, perioperative study of AG-120 and AG-881 in
low-grade gliomas within the high-risk treatment arms of NRG Oncology/ recurrent IDH1 mutant, low-grade glioma: Results from cohort 1. First
RTOG 9802. First Author: Erica Hlavin Bell, The Ohio State University, Author: Ingo K. Mellinghoff, Memorial Sloan Kettering Cancer Center, New
Columbus, OH York, NY
Background: This study sought to update the predictive significance of the Background: AG-120 (ivosidenib [IVO]) is a first-in-class oral inhibitor of
three WHO-defined molecular glioma subgroups (IDHwt, IDHmt/noncodel, mutant isocitrate dehydrogenase 1 (mIDH1) evaluated in 66 glioma patients
and IDHmt/codel) in the subset of specimens available for analysis in NRG (pts) in an ongoing phase 1 study. AG-881 (vorasidenib [VOR]) is an oral,
Oncology/RTOG 9802, a phase III trial of high-risk low-grade gliomas (LGGs) potent, brain-penetrant inhibitor of mIDH1/2 evaluated in 52 glioma pts in
treated with radiation (RT) with and without PCV after biopsy/surgical re- an ongoing phase 1 study. In an orthotopic glioma model, IVO and VOR
section. Notably, this is the first phase III study to evaluate the predictive reduced 2-hydroxyglutarate (2-HG) by 85% and 98%, respectively, despite
value of the WHO subgroups in LGGs using prospectively-collected, well- different brain:plasma ratios (,0.04 vs 1.33). Methods: Primary endpoint:
annotated long-term overall survival data, in a post-hoc analysis. Methods: IDH1/ brain tumor 2-HG concentration with IVO or VOR treatment in mIDH1 low-
2 mutation status was determined by immunohistochemistry and/or next- grade glioma. Pts with recurrent non-enhancing WHO-2016 Grade (Gr) 2 or 3
generation sequencing. 1p/19q status was determined by Oncoscan and/or mIDH1-R132H oligodendroglioma or astrocytoma undergoing craniotomy
450K methylation data. Treatment effects on overall survival (OS) and were randomized 2:2:1 to IVO 500mg QD, VOR 50mg QD, or no treatment for
progression-free survival (PFS) by marker status were determined by the Cox 4 wks preoperatively in Cohort 1. Post-operatively, pts continued to receive
proportional hazard model and tested using the log-rank test in a secondary IVO or VOR and control pts were randomized 1:1 to IVO or VOR. Tumors were
and exploratory analysis. Results: Of all the randomized eligible high-risk G2 assessed for mIDH1 status, cellularity, 2-HG, and drug concentration. Treated
patients (N = 251) in NRG Oncology/RTOG 9802, 106(42%) patients had samples were compared to control pts and mIDH1 and wild type (WT) banked
tissue available with sufficient quality DNA for profiling. Of these, 80(75%) reference (ref) samples. Plasma and CSF 2-HG were assessed. Pts with non-
were IDHmut; 43(41%) were IDHmut/non-co-deleted, 37(35%) were IDH- evaluable tissue were replaced. Results: As of 29 Nov 2018, 26 pts (17M, 9F;
mut/co-deleted, and 26(24%) were IDHwt. Upon univariate analyses, no 25 Gr 2, 1 Gr 3) were randomized preoperatively (IVO 10, VOR 11, control 5),
significant difference in either PFS or OS was observed with the addition of 25 received drug (IVO 12, VOR 13). At the data cut, 19 tumors were analyzed
PCV in the IDHwt subgroup. Both the IDHmut/non-co-deleted and IDHmut/co- with 16 evaluable. Common (.10%) TEAEs (all grade 1/2): diarrhea (36%),
deleted subgroups were significantly correlated with longer PFS (HR = 0.32; hypocalcemia and constipation (each 20%), anemia, hyperglycemia, pruritus,
p = 0.003; HR = 0.13; p , 0.001) and OS (HR = 0.38; p = 0.013; HR = 0.21; headache and nausea (each 16%), and hypokalemia and fatigue (each 12%).
p = 0.029) in the RT plus PCV arm, respectively. Conclusions: Our analyses Mean brain:plasma ratio: 0.16 for IVO, 2.4 for VOR. Tumor 2-HG results are
suggest that both IDHmut/non-co-deleted and IDHmut/co-deleted subgroups shown in Table. Updated data from Cohort 1 will be presented. Conclusions: In
received benefit from treatment with PCV although sample size is limited and Cohort 1 of this phase 1 perioperative study, IVO and VOR were CNS penetrant
analyses are post-hoc. Our results also support the notion that IDHwt high-risk and lowered 2-HG compared to untreated samples. Cohort 2 is open and will
LGG patients do not benefit from the addition of PCV to RT. Funding: evaluate IVO 250mg BID and VOR 10mg QD. Brain tumor 2-HG concentration.
U10CA180868, U10CA180822, and U24CA196067. Also, R01CA108633, Clinical trial information: NCT03343197.
R01CA169368, RC2CA148190, U10CA180850-01, BTFC, OSU-CCC (all to GEO-mean (range) tumor Mean % change (95% CI)
AC). Clinical trial information: NCT00003375. 2-HG, mg/mL 2-HG vs. controls
IVO pts (n=6) 10 (2.2,104) -93% (74, 98)
VOR pts (n=6) 6.8 (3.9, 10) -95% (82, 99)
Control mIDH1 141 (4.8, 909)
(n=65 [4 pts, 61 ref])
WT ref (n=15) 2.7 (0.46,12)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
88s Central Nervous System Tumors

2004 Oral Abstract Session, Mon, 1:15 PM-4:15 PM 2005 Oral Abstract Session, Mon, 1:15 PM-4:15 PM
Phase I study of a brain penetrant mutant IDH1 inhibitor DS-1001b in Efficacy and safety of selinexor in recurrent glioblastoma. First Author:
patients with recurrent or progressive IDH1 mutant gliomas. First Author: Andrew B. Lassman, Columbia University Irving Medical Center, New York,
Atsushi Natsume, Nagoya University School of Medicine, Nagoya, Japan NY
Background: WHO grade II/III gliomas frequently harbor isocitrate de- Background: New treatment modalities are needed for recurrent glioblastoma
hydrogenase 1 (IDH1) mutations, resulting in intratumoral accumulation of (rGBM). Selinexor (SEL) is a novel, oral selective inhibitor of nuclear export which
oncometabolite 2-hydroxyglutarate (2-HG) and subsequent clonal expan- forces nuclear retention of tumor suppressor proteins including p53 and p27,
sion. DS-1001b is an oral selective inhibitor of mutant IDH1 R132X that was leading to apoptosis. We previously reported interim results showing tolerability,
designed to penetrate the blood-brain barrier. Methods: In this first-in- preliminary efficacy, and blood-brain barrier penetration in a surgical cohort (N =
human, multicenter, phase I study (NCT03030066), eligible patients 8). We now report updated results following completion of accrual to non-surgical
(pts) with recurrent/progressive IDH1 mutant glioma received DS-1001b cohorts (N = 68). Methods: This is an open-label, multicenter, phase 2 study of
SEL monotherapy. Patients (pts) not undergoing surgery for measurable rGBM (per
twice daily (bid), continuous. A modified continual reassessment method
RANO) were enrolled in one of 3 arms encompassing different dosing schedules.
was used for dose escalation. RANO and RANO-LGG criteria were used to Treatment was continuous, although cycles were defined as 28 days and response
assess tumor response. Pts who planned to undergo salvage surgery after was assessed every other cycle by MRI. Prior treatment with radiotherapy and
developing progressive disease (PD) and who provided informed consent temozolomide was required and prior bevacizumab was exclusionary. The primary
received DS-1001b treatment until surgery. Tumor samples were also ob- endpoint was 6-month progression free survival (6mPFS) rate, calculated by the
tained from those pts to measure the free form of DS-1001b and 2-HG levels. Kaplan Meier method. Results: A total of 76 pts were enrolled. Median age was 56
Results: Between Jan 2017 and Oct 2018, DS-1001b (125-1400 mg bid) years (range 21-78). Median number of prior treatments was 2 (range 1-7). At the
had administered for 45 pts (median age 44 yrs, prior radiation therapy end of the 6 cycles, 30.2% patients on 80 mg QW were free from progression. The
100%, prior chemotherapy 82%), and 17 pts were continuing treatment. 6mPFS rate on 80 mg QW was 15.1%. Best RANO-defined responses (assessed
Maximum tolerated dose (MTD) was not reached. Most AEs were Gr 1-2. Gr 3 locally) among 26 evaluable pts on 80 mg QW included 1 complete response, 2
AEs were observed in 42.2% of pts. No Gr 4 or 5 AEs or serious drug-related partial responses, 7 stable disease, and 16 with progressive disease. Median
AEs were reported. One dose limiting toxicity was Gr 3 white blood cell count duration of response was 10.8 months. The most common related adverse events
decreased (1000 mg bid). Common AEs ( . 20%) were skin hyperpig- in pts on ~85 mg BIW/60 mg BIW/80 mg QW were nausea (42%/64%/60%),
mentation, diarrhea, pruritus, nausea, rash, and headache. Of 29 evaluable leukopenia (38%/7%/43%), fatigue (71%/71%/43%), neutropenia (29%/14%/
pts with contrast enhancing gliomas, one, three and 10 achieved complete 33%), decreased appetite (46%/71%/27%), and thrombocytopenia (67%/29%/
23%). Conclusions: SEL demonstrated efficacy, with durable responses and
response, partial response and stable disease (SD), respectively. Of
disease stabilization in rGBM. Based on the favorable efficacy and safety profile,
evaluable nine pts with contrast non-enhancing gliomas, two achieved minor
SEL at a dose of 80 mg QW is recommended for further development in rGBM.
response and seven achieved SD. Peak plasma concentration (Cmax) and Clinical trial information: NCT01986348.
area under the curve (AUC) increased dose-dependently. The brain/plasma
ratio of free form of DS-1001b ranged 0.19‒0.77 in 3 pts. Conclusions: DS- 50 mg/m2 (~ 85 mg) BIW 60 mg BIW 80 mg QW
1001b was well tolerated up to 1400 mg bid with favorable brain distri- Parameter (N = 24) (N = 14) (N = 30)
bution, and MTD was not reached. Recurrent/progressive IDH1 mutant 6mPFS rate (%) 9.7 (2.7, 35.4) 11.4 (1.9, 67.9) 15.1 (6.1, 37.1)
glioma pts responded to treatment. Investigation is ongoing to determine the Overall Response Rate (%) 8.3 (1.0, 27.0) 7.1 (0.2, 33.9) 10.0 (2.1, 26.5)
Median Overall Survival (months) 9.0 (4.9, 16.4) 8.5 (7.8, NE) 9.4 (7.0, NE)
recommended Phase II dose. Clinical trial information: NCT03030066.

2006 Oral Abstract Session, Mon, 1:15 PM-4:15 PM 2007 Oral Abstract Session, Mon, 1:15 PM-4:15 PM
Activity of larotrectinib in TRK fusion cancer patients with brain metastases Trabectedin for recurrent WHO grade II or III meningioma: A randomized
or primary central nervous system tumors. First Author: Alexander E. Drilon, phase II study of the EORTC Brain Tumor Group (EORTC-1320-BTG). First
Memorial Sloan Kettering Cancer Center; Weill Cornell Medical College, New Author: Matthias Preusser, Medical University of Vienna, Comprehensive
York, NY Cancer Center, Vienna, Austria
Background: TRK fusions are oncogenic drivers of a variety of cancers, many Background: EORTC-1320-BTG investigated the activity, safety and quality
of which can involve the central nervous system (CNS). Larotrectinib is an of life of therapy with the tetrahydroisoquinoline alkaloid trabectedin (Yon-
FDA-approved selective TRK inhibitor for the treatment of TRK fusion cancer delis) in patients with recurrent higher-grade meningiomas. Trabectedin was
(Drilon et al, NEJM 2018). While larotrectinib has been shown to cross the originally derived from the Caribbean sea squirt, Ecteinascidia turbinata, and
blood–brain barrier (Ziegler et al, Br J Cancer 2018), its clinical activity in a currently is manufactured by total synthesis. Methods: Adult patients with
series of TRK fusion cancers with primary or metastatic intracranial disease has histological diagnosis of WHO grade II or III meningioma and radiologically
not been described. Methods: Patients (pts) with non-primary CNS solid tu- documented progression after maximal feasible surgery and radiotherapy were
mors with brain metastases, or primary CNS tumors harboring a TRK fusion randomly assigned in a 2:1 ratio to receive intravenous trabectedin (1.5 mg/m2
treated with larotrectinib in 2 clinical trials (NCT02637687, NCT02576431) every three weeks) or local standard of care (LOC). The primary endpoint
were identified. Larotrectinib was administered until disease progression (PD), was progression-free survival (PFS). Results: Within 22.1 months, we
withdrawal, or unacceptable toxicity. Disease status was investigator-assessed randomized a total of 90 patients (n=29 in LOC arm, n=61 in trabectedin arm)
(RANO and RECIST). Data cutoff: July 30, 2018. Results: 14 pts were iden- in 35 institutions and nine countries. In the LOC arm, the following treatments
tified: 5 non-primary CNS solid tumors (3 lung cancer, 2 thyroid cancer; fusion were administered: hydroxyurea (n=11), bevacizumab (n=9), none (n=4),
type: 2 ETV6-NTRK3, 2 SQSTM1-NTRK3, 1 EPS15-NTRK1; age range 25–79 chemotherapy (n=3), somatostatin analogue (n=1), combined chemotherapy
y) and 9 primary CNS tumors (3 glioma, 2 glioblastoma, 1 astrocytoma, 3 NOS; and somatostatin analogue (n=1). With 71 PFS events, median PFS was
fusion type: 3 BCR-NTRK2, 2 KANK-NTRK2, 1 each of AFAP1-NTRK1, 4.17 months in the LOC and 2.43 months in the trabectedin arm (hazard ratio
AGTPBP1-NTRK2, ETV6-NTRK3, SPECC1L-NTRK2; age range 2–79 y). In the [HR] for progression, 1.42; 80% CI, 1.00-2.03; p=0.204) with a PFS-6 rate of
5 pts with non-primary CNS tumors, the best objective response to therapy was 29.1% (95% CI, 11.9%-48.8%) in the LOC and 21.1% (95% CI, 11.3%-
PR in 3 (60%, 1 pending confirmation), SD in 1 (20%), and not evaluable (NE) 32.9%) in the trabectedin arm. Median OS was 10.61 months in the LOC and
in 1 (20%). Duration of response ranged from 9+ to 13 mo. In the 9 pts with 11.37 months in the trabectedin arm (HR for death, 0.98; 95% CI, 0.54-1.76;
primary CNS tumors, disease control was achieved in all evaluable pts (primary p=0.94).Grade 3 to 5 adverse events occurred in 44.4% (18.5% related, 4
PD not observed; 1 pt required dose increase). The best objective response to serious adverse events, 0 lethal events) of the patients in the LOC and 59%
therapy was PR in 1 (11%; pending confirmation, 255% tumor shrinkage, (32.8% related, 57 serious adverse events and 2 toxic deaths) of patient in the
ongoing at 3.7 mo), SD in 7 (78%; tumor shrinkage range 21% to 224% for trabectedin arm. Conclusions: In this first prospective randomized trial per-
pts with measurable disease, 5 had SD . 4 mo), and NE in 1 (11%). Duration of formed in recurrent grade II or III meningioma, trabectedin did not improve
treatment ranged from 2.8–9.2+ mo. Conclusions: Larotrectinib is active in pts PFS and OS and was associated with significantly higher toxicity as compared
with TRK fusion cancers with intracranial disease. Confirmed responses and to LOC treatment. The data collected in this study may serve as benchmark for
durable disease control were seen in metastatic disease and primary CNS future clinical trials in this setting. Clinical trial information: NCT02234050.
tumors of various histologies. These results further support expanded testing for
TRK fusions across all cancers, including primary CNS tumors. Clinical trial
information: NCT02637687 and NCT02576431.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Central Nervous System Tumors 89s

2008 Oral Abstract Session, Mon, 1:15 PM-4:15 PM 2009 Poster Discussion Session; Displayed in Poster Session (Board #198),
A phase I study of convection-enhanced delivery of 124I-8H9 radio-labeled Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
monoclonal antibody in children with diffuse intrinsic pontine glioma: An Sun, 4:30 PM-6:00 PM
update with dose-response assessment. First Author: Mark M. Souweidane, NRG Oncology CC001: A phase III trial of hippocampal avoidance (HA) in
Memorial Sloan-Kettering Cancer Center, New York, NY addition to whole-brain radiotherapy (WBRT) plus memantine to preserve
neurocognitive function (NCF) in patients with brain metastases (BM). First
Background: Diffuse intrinsic pontine glioma (DIPG) represents one of the
Author: Vinai Gondi, Northwestern Medicine Cancer Center Warrenville and
most deadly central nervous system tumors of childhood with a median
Northwestern Medicine Proton Center, Warrenville, IL
survival of less than 12 months. Convection-enhanced delivery (CED) has
been recently hypothesized as a means for efficiently distributing thera- Background: NRG CC001, a phase III trial of WBRT plus memantine
peutic agents within the brain stem. We conducted this study to evaluate (WBRT+M) with or without HA, sought to evaluate the neuro-protective
CED in children with DIPG. Methods: We performed a standard phase I dose effects of lowering the hippocampal radiation dose. Methods: Patients (pts)
escalation study in patients with non-progressive DIPG 4 to 14 weeks post- with BM were stratified by RPA class and prior radiosurgery/surgery and
completion of radiation therapy. Seven dose levels of a single injection of randomized to WBRT+M or HA-WBRT+M (30Gy/10 fractions). Standardized
124
I-8H9 (Omburtamab) (range 0.25 to 4.0 mCi) were studied. Results: 37 NCF tests were performed at baseline, 2, 4, 6, and 12 months (mos). The
children were treated with 34 evaluable for primary and secondary end- primary endpoint was NCF failure, defined as decline using the reliable
points. The median age at enrollment was 6.8 years old (range 3.2 - 17.9). change index on Hopkins Verbal Learning Test-Revised, Trail Making Test, or
There was no dose limiting toxicity (DLT). Among adverse events that were at Controlled Oral Word Association. Cumulative incidence estimated NCF
least possibly related to the treatment, there were no grade 4 or 5 events, and failure (death without NCF failure was competing risk); between-arms dif-
only 4 reversible grade 3 events in 4 patients (2 hemiparesis, 1 skin infection ferences tested using Gray’s test. Deterioration at each collection time point
and 1 anxiety). Estimations of distribution volumes based on T2-weighted was tested using a chi-square test. Patient-reported symptoms were
imaging were dose dependent and ranged from 1.5 to 20.8 cm3, and for dose assessed using the MD Anderson Symptom Inventory Brain Tumor module
level 7, 10.5 - 19.0 cm3. The mean volume of distribution/volume of infusion and analyzed using mixed effects models and t-tests. Results: From 7/2016
ratio (Vd/Vi) was 3.4 61.1, and for dose level 7, 3.5 6 1.0. The mean lesion to 3/2018, 518 pts were randomized. Median follow-up was 7.9 mos.
absorbed dose was 33.3 6 25.9 Gy, and for dose level 7, 50.1 6 22.9 Gy. HA-WBRT+M was associated with lower NCF failure risk (adjusted hazard
The mean ratio of lesion-to-whole body absorbed dose was 910. The mean ratio (HR) = 0.74, p = 0.02) due to lower risk of deterioration in executive
volume of distribution/tumor volume ratio on dose level 7 was 82.5%, but the function at 4 mos (p = 0.01) and encoding (p = 0.049) and consolidation (p =
mean tumor overlap was 40.5%. No death occurred as a result of the 0.02) at 6 mos. Age#61 predicted lower NCF failure risk (HR = 0.60, p = 0.0002);
treatment. Median survival was 15.3 months (n = 29, 95% CI 12.7 - 17.4). non-significant test for interaction indicated independent effects of HA and age.
Median follow-up time of the 5 surviving patients is 27.2 months (range 11.5 - Patient-reported fatigue (p = 0.036), difficulty speaking (p = 0.049) and problems
72.4). Overall survival rate at 12 months was 64.7% (22/34, 4 alive), and remembering things (p = 0.013) at 6 mos favored the HA-WBRT+M arm.
overall survival rate at 24 months 14.7% (5/34, 3 alive). Conclusions: CED Imputation models accounting for missing data also favored the HA-WBRT+
in the brain stem of children with DIPG who were previously irradiated is a M arm for patient-reported cognition (p = 0.011) and symptom interference
safe therapeutic strategy. An infusion volume of 4,000 mcl appears to be a (p = 0.008) at 6 mos. Treatment arms did not differ in toxicity, overall survival,
reasonable single dose for a target distribution volume but enhanced tumor or intracranial progression. Conclusions: HA during WBRT+M for BM better
coverage is likely needed. There seems to be a survival benefit using this preserves NCF and patient-reported symptoms, while achieving similar in-
therapeutic strategy and outcomes might be dependent on dosimetry and tracranial control and survival. Supported by grants UG1CA189867 (NCORP),
distribution patterns. Clinical trial information: NCT01502917. U10CA180868 (NRG Oncology Operations), DCP from the National Cancer
Institute. Clinical trial information: NCT02360215.

2011 Poster Discussion Session; Displayed in Poster Session (Board #200), 2012 Poster Discussion Session; Displayed in Poster Session (Board #201),
Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Sun, 4:30 PM-6:00 PM Sun, 4:30 PM-6:00 PM
A randomized phase II trial of veliparib (V), radiotherapy (RT) and temozolomide Glioblastoma evolution pattern under surgery and radio(chemo)therapy
(TMZ) in patients (pts) with unmethylated MGMT (uMGMT) glioblastoma (RCHT) to identify novel methylome based glioma subtypes. First Author:
(GBM). First Author: Mustafa Khasraw, Royal North Shore Hospital/ University of Maximilian Knoll, Departments of Radiation Oncology, Neurology, Neuro-
Sydney, St Leonards, Australia surgery, Heidelberg University Hospital, National Center for Tumor Disease
(NCT), UKHD and German Cancer Research Center (DKFZ), German Cancer
Background: TMZ offers minimal benefit in uMGMT GBM pts. V is synergistic
Consortium (DKTK), Core-Center Heidelberg, Heidelberg, Germany
with both RT and TMZ in preclinical models, safe when combined with
either RT or TMZ clinically, but the triplet (V+RT+TMZ) is poorly tolerated. Background: Identification of isocitrate dehydrogenase mutations (IDHm) and
This study examined a novel approach to patients with uMGMT GBM. glioma CpG island methylator phenotype (G-CIMP) as well as methylation of O6-
Methods: VERTU is a randomized Phase 2 trial comparing Arm A (Standard methylguanine DNA methyltransferase (MGMT) promotor has substantially im-
of care) = RT (60Gy/30 fractions) + TMZ (75mg/m2 daily) followed by TMZ proved stratification of glioma patients into prognostic subgroups. In extension of
(150–200mg/m2D 1–5) every 28 days for 6 cycles vs Arm B (experimental static pre-therapy diagnostic, we sought to investigate the impact of glioblastoma
evolution under selection pressure of standard therapy on methylome level.
arm) = RT (60Gy/30 fractions) + V (200mg PO BID) followed by TMZ
Methods: For the training cohort (T), methylome (450k Illumina) data of paired
(150–200mg/m2D 1–5) + V (40mg bid, D 1–7) every 28 days for 6 cycles in pts samples from 50 patients with glioblastoma (GBM, 11 G-CIMP+) were analyzed,
with newly diagnosed centrally determined uMGMT GBM. The study aims to i.e., primary (P) and at the time point of recurrence (R, re-surgery) after standard
randomize 120 pts (2:1 to the experimental arm). The primary endpoint was therapy at NCT. For 39 pairs matching RNASeq data was analyzed. Validation
6 months progression free survival (6mPFS) with multiple secondary and tertiary cohorts consisted of Heidelberg (V1) total n = 650, GBM (n:585, 8 G-CIMP+),
endpoints. Evaluation of feasibility and safety was planned after completion of grade III (n:65, CIMP+ 65), Austrian GBM (V2, n = 499, 36 IDHm, pyrosequencing
RT in the first 60 pts (Stage 1). (ANZCTR #ACTRN12615000407594). Tumor data) and the TCGA (V3) Lower-grade-glioma cohort (LGG, n = 477, grade III n:
tissue and serial bloods were collected for translational research. Results: 125 247, 178 G-CIMP+, grade II n: 228, 206 G-CIMP+). Results: Limited number of
pts were randomized (41 Arm A, 84 Arm B). Mean (range) age 58 (22–78) consensus differentially methylated probes (DMP) were found across all P vs. R
years, 70% male, 61% ECOG 0, 86% macroscopic resection, 14% biopsy. At samples (nCpG = 411 CpGs, FDR , 0.05). In contrast, heterogeneity in GBM
the time of analysis (cut-off date: 04/Feb/2019), median follow up was evolution was found by similarity analysis of delta-methylome data of 50 PR pairs
16.5 months, 76 pts had died. 6mPFS (95% CI, Kaplan-Meier estimate) was resulting in two distinct clinical subgroups and one “intermediate” group. In-
37% (22–52) in Arm A and 53% (41–63) in Arm B, and median PFS was 4.4m triguingly, n = 114.652 DMP (FDR , 0.05) was found by comparing the evolu-
tionary “poor” (n = 15) vs. “good” (n = 13) GBM phenotypes. A random forest
(95% CI 4.0–6.0) for Arm A and 6.2m (95% CI 4.9–7.1) for Arm B (HR = 0.81,
classifier was built to identify the evolutionary subgroups in P samples. The per-
95%CI 0.54–1.21). 50% of pts in Arm A and 53% in Arm B experienced $ G3
formance of “good” prognosis classifier was in T cohort HR: 0.54 [0.30-0.97], p =
adverse events (AEs). The most common G 3/4 AEs were decreased platelets, 0.04; V1: 0.57 [0.43-0.76], p , 0.001, V2: 0.62 [0.47-0.82], p , 0.001, LGG:
seizures, hyperglycemia and diarrhea (each 5%) in Arm A and decreased 0.16 [0.08-0.32], p , 0.001. In “good” prognosis group (T), neither G-CIMP+ (n =
platelets (13%) and seizures (11%) in Arm B. Conclusions: In this multicenter, 3) nor MGMT-STP27 (oddsratio, OR: 0.56, p = 0.47) was enriched. MGMT-STP27
randomized study, the experimental therapy was feasible and well tolerated. The OR was 0.47 (V1, p = 0.47) or 1.28 (V2, p = 0.45), respectively. The evolutionary
observed 6mPFS appeared longer in Arm B, but at the time of submitting the subgroups remain prognostic independent of GCIMP status in LGG (V3). “Poor”
abstract, this result did not meet the prespecified primary endpoint. More glioma are enriched for RTKI/II methylome subtypes, and contain less frequently
mature results will be presented at the annual meeting. QoL in VERTU is re- the mesenchymal subtype. Bevacizumab treatment showed a survival benefits only
ported separately. Central MR review, biomarker analyses, including DNA repair in “poor” subtype (V1+2). Conclusions: Discovery of a methylome based classifier
and methylation signature analyses are ongoing. Clinical trial information: of glioma evolution informs on “good” and “poor” prognosis subtypes and may have
ACTRN12615000407594. ramification for stratifying patients for therapy such as e.g., antiangiogenesis.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
90s Central Nervous System Tumors

2013 Poster Discussion Session; Displayed in Poster Session (Board #202), 2014 Poster Discussion Session; Displayed in Poster Session (Board #203),
Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Sun, 4:30 PM-6:00 PM Sun, 4:30 PM-6:00 PM
Impact of predictive impact of MGMT promoter methylation in malignant Genomic characterization of lung tumors and metastatic (Met) sites in advanced
astrocytomas depends on the methylation subgroup. First Author: Wolfgang (Adv) NSCLC. First Author: Melinda D Willard, Eli Lilly and Company,
Wick, National Center for Tumor Diseases (NCT), UKHD and German Cancer Indianapolis, IN
Research Center (DKFZ), Heidelberg, Germany
Background: Molecular alterations (MA) found in brain (Br) mets of NSCLC
Background: O6-methylguanine DNA-methyl transferase (MGMT) status is pts can differ from primary and/or other met sites, which may explain why
predictive for alkylating chemotherapy in most series, but there are non- therapies targeting primary tumors are less effective at preventing and
benefitting subgroups. Despite multiple attempts, MGMT has not been un- treating intracranial disease. We analyzed the frequency of known driver
ambiguously established as a predictive biomarker for patients with malignant genes in adv NSCLC pts and the association with overall survival (OS).
gliomas. Further, these tumors are to be better classified according to global Methods: This retrospective observational study identified pts from the
methylation profiles. Methods: Long-term efficacy data of the NOA-08 trial Flatiron- Foundation Medicine NSCLC Clinico-Genomic Database who were
(NCT01502241) that compared efficacy and safety of radiotherapy (RT, n= diagnosed with adv NSCLC from 1 Jan 2011 to 31 Oct 2017 and had tumor
176) to temozolomide (TMZ, n= 193) in patients . 65 years with anaplastic tissue analyzed at any time following initial diagnosis via targeted DNA
astrocytoma (AA) or GB as well as genome-wide DNA methylation patterns and sequencing by FoundationOne. Descriptive statistics summarized MA from
copy number variations assessed by methylation arrays in a biomarker subset lung and met sites (Br and non-brain [NB]). OS was measured from adv
(n= 104) and an independent cohort (n= 380) have been used to assess the diagnosis to death or last activity date (censored). Multivariable Cox pro-
interaction between MGMT status and methylation subgroups. Results: In the portional hazard regression model was used for time-to-event analysis.
long-term update of NOA-08 patients with MGMT methylated tumors had Results: Of 3257 pts, data were available from lung (n = 1621), Br (n = 180),
longer OS and EFS when treated with TMZ (18.4 [13.9-24.4] months and 8.5 and NB sites (n = 377): liver (n = 167), bone (n = 124), adrenal (n = 63), and
[6.9-13.3] months) versus RT (9.6 [6.4-13.7] months and 4.8 [4.3-6.2] spine (n = 23). Median age at adv diagnosis was 66.2 yrs. TP53(63.3%),
months, HR 0.44 [0.27-0.70], p , 0.001 for OS and 0.46 [0.29-0.73], p = KRAS(28.8%), EGFR(15.6%),STK11 (13.5%), and CDKN2A(8.5%) were
0.001 for EFS). These data compared favorably with recently published data frequently mutated genes in lung samples. Genes for Br vs NB sites included
from patients treated with chemoradiation (Perry et al. NEJM 2017). Im- TP53(70.6%; 64.7%), KRAS(36.1%; 26.5%), EGFR (9.4%; 18.8%),
portantly, only patients with glioblastomas of the methylation class receptor STK11 (18.9%; 12.7%), and CDKN2A(6.1%; 10.1). KEAP1alterations were
tyrosine kinase II (RTKII) and mesenchymal but not RTK I demonstrated the also present in 10% (Br), 7.4% (NB), and 6% of lung samples. In treated pts,
predictive impact of MGMT in the NOA and the independent validation cohort. lack of alterations in select genes (STK11, TP53, KEAP1) was associated with
Conclusions: MGMT promoter methylation as a strong but methylation longer OS, whereas lack of other alterations (ARID1A, EGFR, ALK, ROS1) was
subclass-dependent predictive biomarker for the use of alkylating chemo- associated with a shorter OS (p , 0.05). Patients with select mutations co-
therapy in malignant gliomas. The data call for embedding of MGMT tests into occurring with KRAS had higher risk of death compared to those with KRAS
global methylation analyses for all patients with malignant gliomas potentially only (p , 0.05). Conclusions: Based on pts with NSCLC whose tumor tissue
treated with alkylating chemotherapy. underwent DNA sequencing, the most frequently altered genes in lung and Br
samples included TP53,KRAS, EGFR, STK11,andCDKN2A, with some being
significantly associated with OS. Prognosis of NSCLC pts depends on clinical,
demographic, and genomic factors and should be carefully considered to
optimize clinical outcome.

2015 Poster Discussion Session; Displayed in Poster Session (Board #204), 2016 Poster Discussion Session; Displayed in Poster Session (Board #205),
Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Sun, 4:30 PM-6:00 PM Sun, 4:30 PM-6:00 PM
Circulating tumor DNA analysis (ctDNA) for genomic testing in NSCLC SurVaxM with standard therapy in newly diagnosed glioblastoma: Phase II
patients with isolated CNS progression. First Author: Mihaela Aldea, trial update. First Author: Manmeet Singh Ahluwalia, Burkhardt Brain Tumor
Medical Oncology Department, Gustave Roussy, Villejuif, France NeuroOncology Center, Neurological Institute, Taussig Center Institute,
Cleveland Clinic, Cleveland, OH
Background: Genomic DNA profiles are mandatory in advanced, treatment
naive non-small cell lung cancer (NSCLC) patients (pts) and strongly rec- Background: SVN53-67/M57-KLH (SurVaxM) is a novel cancer vaccine
ommended at progression (PD) on personalized treatment. In pts with PD designed to stimulate an immune response targeting the tumor-specific
limited to central nervous system (CNS), tissue biopsy is difficult and the antigen survivin. A multi-center, single-arm phase 2 clinical trial of SurVaxM
performance of ctDNA is unknown. Methods: Clinical, molecular, imaging in survivin-positive newly diagnosed glioblastoma (nGBM, NCT02455557)
data of NSCLC pts included in 2 prospective studies from 01.2016 to is now fully enrolled and data updated. Methods: Patients (n = 63) with
11.2018 at Gustave Roussy were collected. Inclusion criteria were: stage IV nGBM were enrolled at 5 US cancer centers and followed for safety, 6-month
disease and any known tissue genomic alteration (GA) EGFR, ALK, BRAF, progression-free survival (PFS6), 12-month overall survival (OS12) and
KRAS, HER2, ROS1, MET, PIK3CA, TP53. Plasma ctDNA collected at immunologic response. All patients underwent craniotomy with near-total
baseline/PD were analyzed by next-generation sequencing (NGS-InVis- resection ( , 1 cm3 residual contrast enhancement), TMZ chemoradiation,
ionFirst™-Lung) in 3 groups: pts with isolated CNS (iCNS), extra-CNS only adjuvant TMZ and SurVaxM. Patients received 4 doses of SurVaxM (500
(noCNS) or both combined (cCNS) disease. iCNS was defined as any PD to mcg) in Montanide with sargramostim (100 mcg) biweekly, followed by
CNS, while stable/absent extra-CNS metastases (mts). ctDNA was considered maintenance SurVaxM with adjuvants every 12 weeks until tumor pro-
positive if $1 GA was found. ctDNA in cerebrospinal fluid (CSF) were also gression. Immunogenicity of SurVaxM was assessed by detection of survivin-
collected. Results: Out of 245 pts with $1 ctDNA: 56 had iCNS (66 samples), specific antibody (IgG) and CD8+ T-cell levels. Results: Median age was 60
97 noCNS (127 samples) and 92 cCNS (107 samples). In this cohort, 60% yrs (range, 20-82), 53% methylated MGMT, 46% unmethylated MGMT
were female, median age 60 years, 47% smokers; 92% had adenocarcinoma. (1 N/A) and 60% were male. Survivin expression ranged from 1-40%
The median number of mts sites was 3 in noCNS/cCNS groups. Proportions of (median 12%) by immunohistochemistry. Median time to first immunization
tissue GA at baseline were (iCNS vs noCNS/cCNS): EGFR (50% vs 44%), ALK was 3.0 mo (1.9-4.0 mo) from diagnosis. There have been no RLT or grade $
(30% vs 11%), BRAF (4% vs 12%), KRAS (5% vs 15%), HER2 (2% vs 5%), 3 SAE attributable to SurVaxM. The most common AE was grade 1-2 in-
ROS1 (5% vs 4%). Tyrosine kinase inhibitors were used in 73% iCNS vs 61% jection site reactions. OS12 was 86% from first immunization and 93.4%
noCNS/cCNS. Local brain treatments were performed in 43% (n = 24) vs 32% from diagnosis. OS12 for meMGMT was 93.1% and unMGMT was 78% from
(n = 29) and leptomeningeal mts (LM) detected in 34% (n = 19) vs 8% (n = 9), first immunization. Median time to tumor progression (mPFS) was
in iCNS and cCNS, respectively. CtDNA was positive (+) in 52% in iCNS vs 13.9 months from diagnosis. Median OS has not yet been reached. SurVaxM
84% in noCNS and 92% in cCNS (p , 0.0001). In iCNS, there was a non- produced an increase in survivin-specific IgG titre from pre-vaccine baseline
significantly higher proportion of + ctDNA in pts with LM vs only brain disease to $ 1:10,000 in 67% of pts and $ 1:100,000 in 27%. CD8+ T cell re-
(59 vs 48%, P = 0.44). 12/56 pts of iCNS group had serial ctDNA, being sponses were observed. Anti-survivin IgG and OS were correlated.
collected also at time of cCNS. In 25% of cases, a negative ctDNA at time of Conclusions: SurVaxM immunotherapy generated encouraging efficacy and
iCNS shifted to + at time of cCNS. In 12 iCNS pts, ctDNA was + in 6 (50%) immunogenicity in nGBM and has minimal toxicity. A randomized, pro-
plasma and in 10 (83%) paired CSF (p = 0.193). Conclusions: Detection of spective trial of SurVaxM in nGBM is planned. Clinical trial information:
GA by plasma ctDNA is lower in NSCLC pts with isolated CNS PD. Alternative NCT024455557.
strategies (as CSF analysis) should be explored.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Central Nervous System Tumors 91s

2017 Poster Discussion Session; Displayed in Poster Session (Board #206), 2018 Poster Discussion Session; Displayed in Poster Session (Board #207),
Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Sun, 4:30 PM-6:00 PM Sun, 4:30 PM-6:00 PM
Updated phase I trial of anti-LAG-3 or anti-CD137 alone and in combination Quantitative radiographic analysis of phase II and III trials in recurrent
with anti-PD-1 in patients with recurrent GBM. First Author: Michael Lim, glioblastoma treated with VB-111 with or without bevacizumab or bevaci-
The Johns Hopkins Hospital, Baltimore, MD zumab monotherapy. First Author: Benjamin M. Ellingson, University of
California Los Angeles, Los Angeles, CA
Background: Preclinical GBM data targeting the checkpoint molecules Lag-
3 and CD137 have shown promising anti-tumor immune response with Background: VB-111 is a non-replicating adenovirus carrying a pro-apoptotic
resultant improved survival when combined with anti-PD-1. Here we report transgene for TNFR1/Fas under the control of a modified murine promoter to pre-
our experience from a multi-arm safety study in patients with recurrent GBM proendothelin 1. The transgene is expressed only in angiogenic endothelial cells,
treated with anti-Lag-3 and anti-CD137. Methods: The Adult Brain Tumor and therefore VB-111 results in targeted apoptosis of neovascular vessels. The
Consortium (ABTC) 1501 trial is a phase I, open label, multicenter, multi- current study characterizes the quantitative radiographic results and impact on
arm dose-finding/safety study of anti-LAG-3 (BMS-986016) or anti-CD137 OS in phase 2 and 3 trials of recurrent glioblastoma (GBM) patients treated with
(BMS-663513) alone and in combination with anti-PD-1 in patients at first VB-111 with or without bevacizumab (BV) or BV monotherapy. Methods: MRI
recurrence of GBM. The primary objective is to define MTD for the mono and data from a phase 2 (NCT01260506) and randomized, double arm, controlled
combinational treatment. The major secondary objective is to explore for a phase 3 (GLOBE; NCT02511405) trial of VB-111 in recurrent GBM were used in
signal in efficacy. The key inclusion criteria are adults, first recurrence of current study: Arm A) VB-111 monotherapy until progression followed by com-
GBM following RT+TMZ, TLC$1000/ul, KPS$ 60%, stable corticosteroid bination VB-111 and bevacizumab (BV) (“Primed Combination”; Phase 2; N =
regimen, measurable disease, and written informed consent. Sequential 24); Arm B) VB-111 in combination with BV (“Unprimed Combination”; Phase 3;
allocation was used for the treatment assignment at starting dose of 80mg for N = 124) and Arm C) BV monotherapy (“Control”; Phase 3; N = 120). Contrast
anti-LAG-3 and 8mg for anti-CD137. Anti-PD-1was given at a flat dose of enhanced T1-weighted digital subtraction was used to quantify tumor volume at
240 mg in the combination treatment arms. The 3+3 design is used for the all time points. Results: Baseline tumor volume was not significantly different
dose finding with a target DLT rate , 33%. Results: to date 44 patients were between patient cohorts (Kruskal-Wallis; P = 0.1482; median~20mL). Contin-
enrolled into the trial with median age at 57, median KPS at 90. Median uous measures of baseline tumor volume were prognostic for OS in all treatment
treatment cycle was 3 and 39% tumors were MGMT methylated. The highest groups when controlling for therapy and age (Cox, P , 0.001, HR = 1.02). In
patients with small tumors ( , 25mL), the “primed combination” cohort (Arm A)
safe dose for Anti-LAG-3 alone is 800 mg without a DLT. The safe dose for
from the phase 2 trial had a significant OS advantage compared to both upfront
anti-CD137 alone arm is 8mg with 1 DLT, and 2 grade 3 elevated serum ALT
combination of VB-111 and BV (Arm B; P = 0.0094, HR = 0.5328; median OS =
at end of cycle 2. Combination arms of Anti-LAG-3 +anti-PD-1 (160 mg/
7mo vs. 15mo) as well as BV alone (Arm C; P = 0.0248, HR = 0.5776; median
240mg as the highest dose combination) had one DLT (hypertension) and no
OS = 8.5mo vs. 15mo). Patients with a radiographic response ( . 65% reduction)
toxicities were seen in the combination arm of Anti-CD137+Anti-PD-1 had a significant survival difference from non-responders when controlling for age,
(3 mg/240 mg). mOS was 14 months for anti-CD137 alone, 8 months for baseline tumor volume, and treatment arm (P = 0.0014, HR = 0.5822). Notably,
Anti-Lag-3, and 7 months for Anti-Lag-3 + Anti-PD-1. Correlative data will be in responders to VB-111 monotherapy or combination therapy after priming with
discussed. Conclusions: The trial is ongoing. The RP2D is 800mg for anti- VB-111 exhibited characteristic, expansive areas of necrosis in areas of initial
LAG-3 as a monotherapy and 8mg for anti-CD137. For the combination disease. Conclusions: Small recurrent tumors have a significant OS advantage
arms, 160 mg of Anti-LAG-3 and 240 mg of anti-PD-1 and 3 mg of anti- when “priming” with VB-111 monotherapy prior to combination VB-111 and BV
CD137 and 240 mg antiPD-1 were the RP2D. Clinical trial information: at recurrence. Patients responding to VB-111 exhibit specific imaging charac-
NCT02658981. teristics related to the drug mechanism of action. Clinical trial information:
NCT02511405; NCT01260506.

2019 Poster Discussion Session; Displayed in Poster Session (Board #208), 2020 Poster Discussion Session; Displayed in Poster Session (Board #209),
Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Sun, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Sun, 4:30 PM-6:00 PM Sun, 4:30 PM-6:00 PM
First-in-human phase I trial of the combination of two adenoviral vectors Evaluation of controlled IL-12 in combination with a PD-1 inhibitor in
expressing HSV1-TK and FLT3L for the treatment of newly diagnosed subjects with recurrent glioblastoma. First Author: E. Antonio Chiocca,
resectable malignant glioma: Initial results from the therapeutic reprogramming Brigham and Women’s Hospital, Boston, MA
of the brain immune system. First Author: Pedro R. Lowenstein, Univ of
Background: Ad-RTS-hIL-12 (Ad) is a novel gene therapy candidate con-
Michigan Medical School, Ann Arbor, MI
ditionally expressing IL-12 under the control of veledimex (V) acting via the
Background: This is the initial report on a first in human Phase I dose es- proprietary RheoSwitch Therapeutic System (RTS) gene switch with a
calation trial of the combination of two adenoviral vectors expressing HSV1- therapeutic window. Intratumoral Ad + oral V monotherapy (Phase 1 study,
TK or Flt3L for the treatment of newly diagnosed, resectable malignant NCT02026271) resulted in a new sustained intra-tumor influx of activated
gliomas. The absence of functional dendritic cells from the brain precludes cytotoxic T cells, consistent with an immune-mediated anti-tumor effect
anti-brain tumor immune responses. We combined tumor cytotoxicity (Ad- improving median overall survival (mOS) of subjects with recurrent glio-
HSV1TK) with recruitment of dendritic cells to the brain (Ad-Flt3L) to induce blastoma (rGBM). This correlated with an increased circulating CD8+/FoxP3+
an effective anti-tumor immune response. This strategy induced an effi- T cell ratio (“cytoindex”), an emerging biomarker for mOS. PD-1 expression on
cacious, cytotoxic CD8 and CD4 T-dependent immune response in many infiltrating T cells at biopsy after Ad+V, supports combining controlled IL-12
animal models of glioma. This immune response also generated anti-tumor with a PD-1 inhibitor to further augment T-cell-mediated anti-tumor effects.
memory, and the capacity for neoantigen recognition. Methods: The trial was The rationale is also supported by increased OS (100% combo vs 63% for
approved by FDA and all institutional cttees. Treatment was administered Ad+V vs 40% for anti-PD-1) in mice bearing GL-261 glioma. Methods: An
intraoperatively following complete glioma resection in newly diagnosed ongoing open label, dose-escalation Phase 1 trial (NCT03636477) is eval-
tumors. The trial consisted of vector dose escalation, starting at 1x10^9 i.u., uating safety and tolerability of local, controlled IL-12 with nivolumab (nivo) in
and increasing to 1x10^11 i.u. of each vector. Dose escalation proceeded by adult subjects with rGBM. Ad was administered by single intratumoral in-
increasing the vector dose through a total of 6 combinations administered to 6 jection (2 x 1011 viral particles, Day 0) plus V (10-20 mg) PO QD x 15 with nivo
cohorts of 3 patients each. Two cycles of 14 days each of valacyclovir were (1-3mg/kg) IV on Days -7, 15, then Q2W. Results: Safety data revealed a
administered to activate HSV1-TK cytotoxicity. Cycle 1 starts on Day 1-3 post similar profile as Ad +V monotherapy. Adverse reactions (ARs) during follow-on
surgery for 14 days, and Cycle 2 on Week 8-12. Standard radiation, i.e., 60 Gy nivo dosing were consistent with anti-PD-1 reports. ARs were manageable and
in 2 Gy fractions over 6 weeks, with concurrent temozolomide, was followed by reversible with no synergistic toxicities. Nivo alone did not alter peripheral IL-
cyclic temozolomide. Results: Examination of tumor samples at primary re- 12 levels (median baseline (before anti-PD-1) 0.9 pg/mL; Day 0 1 pg/mL)
section and first recurrence show an increase in the infiltration of inflammatory increasing to 5.5 pg/mL on Day 3. Nivo alone increased peripheral T cells
cells. The experimental treatment was well tolerated. At this time the MTD has (CD3+CD8+ median baseline 23%; Day 0 26%) and Ad+V elevated peripheral
not been reached. There were approx. 248 AEs, and 26 SAEs; these have not CD3+CD8+ to 31% at Day 14. Nivo alone decreased regulatory T cells (FoxP3
been linked to treatment. At this time the MTD has not been reached. A baseline 1.5% vs Day 0 0.8%). Ad+V decreased these to 0.3% (Day 14).
secondary outcome is overall survival. Preliminary analysis of partial data may Combination therapy improved the cytoindex (baseline 15; Day 0 29; Day 14
suggest that the combined viral vector therapy may provide a clinically sig- 80). Conclusions: Controlled IL-12 production using Ad + V with nivo is a
nificant survival. Conclusions: Our results show for the first time that rational combination with initial data consistent with immune-mediated anti-
reprogramming of the host’s brain immune system to recognize gliomas tumor effects with a favorable safety profile, warranting continued in-
reveals a new approach for the treatment of highly malignant brain tumors. vestigation in rGBM. Clinical trial information: NCT03636477.
Clinical trial information: NCT01811992.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
92s Central Nervous System Tumors

2021 Poster Session (Board #210), Sun, 8:00 AM-11:00 AM 2022 Poster Session (Board #211), Sun, 8:00 AM-11:00 AM
Full enrollment results from an extended phase I, multicenter, open label Longitudinal analysis of quality of life following treatment with asunercept
study of marizomib (MRZ) with temozolomide (TMZ) and radiotherapy (RT) plus reirradiation versus reirradiation in progressive glioblastoma patients.
in newly diagnosed glioblastoma (GBM). First Author: Warren P. Mason, First Author: Wolfgang Wick, National Center for Tumor Diseases (NCT),
Princess Margaret Cancer Centre, University Health Network, Toronto, ON, UKHD and German Cancer Research Center (DKFZ), Heidelberg, Germany
Canada
Background: Palliation of symptoms and the maintenance of quality of life (QoL)
Background: Proteasome inhibition sensitizes glioma cells to TMZ and RT, are important goals in cancer treatment.1,2 Beyond progression free survival
providing a novel therapeutic strategy for GBM. MRZ, an irreversible, brain- (PFS) and overall survival (OS), health related QoL was one of the secondary
penetrant, pan-proteasome inhibitor with anti-glioma activity was combined endpoints in the asunercept plus irradiation Phase II trial (NCT01071837) in
with standard TMZ/RT → TMZ in newly diagnosed GBM (NCT02903069), to recurrent glioblastoma.3 Current analysis presents time to deterioration (TtD) of
determine the recommended dose (RD). The primary endpoint of this expanded QoL using data from this study. Methods: Data from patients (pts) with a baseline
phase 1 trial was toxicity, with secondary endpoint of OS. Methods: Patients and $1 post-baseline QoL assessment were included in this analysis. TtD was
were enrolled in separate cohorts (TMZ/RT+MRZ→TMZ+MRZ, N=15; TMZ/ defined as the time from randomization to the first deterioration in the EORTC
QLQ-C15, PAL EORTC QLQ-BN20 and Medical Research Council (MRC)-
RT→TMZ+MRZ, N=18) in dose-escalation (3+3 design), followed by dose-
Neurological status. Deterioration was defined as a decrease of $10 points
expansion (N=20) with TMZ/RT+MRZ at RD → TMZ+MRZ at RD. A separate
from baseline in the QLQ-C15 PAL overall QoL and functioning scales, an in-
cohort received TMZ/RT→TMZ+MRZ at RD with Tumor Treating Fields (Optune, crease of $10 points from baseline in the QLQ-C15 PAL fatigue scale and the
N=13). MRZ was infused IV (10 min at 0.55, 0.7, 0.8, and 1.0 mg/m2) on Days QLQ-BN20 total sum of score, and a rating of “Worse” in the MRC-Neurological
1, 8, 15, 29, 36 (42-day TMZ/RT+MRZ cycle) and Days 1, 8, 15 (28-day TMZ+ status. Pts without a deterioration were censored at the last QoL assessment.
MRZ cycle). Results: 66 patients treated; median age 58 years, 68% male, Kaplan-Meier estimates were used to describe TtD and both treatment groups
50% receiving corticosteroid at baseline, 52% unmethylated MGMT. Dose- compared using the logrank test. The relationship between progression of disease
limiting toxicities (DLTs) in dose-escalation cohorts: 1 (fatigue) at 0.7 mg/m2 (PD) and QoL deterioration has been investigated. Results: Compared to reirra-
MRZ, 5 (ataxia/diarrhea; ataxia/confusion; myocardial infarction, delirium/ diation alone, treatment with asunercept + reirradiation was associated with sig-
ataxia; ataxia/fatigue) in 1.0 mg/m2 cohorts. MRZ demonstrated a steep dose- nificant improvement of TtD (P#0.01; Table). PD was a key driver for QoL
response with treatment-emergent adverse events (TEAEs)/DLTs pre- deterioration and the median TtD was comparable with PFS in favour of the asu-
dominately CNS AEs (Grade $3 TEAEs in 12 of 12 patients at 1.0 mg/m2 vs nercept treatment arm. Conclusions: Treatment with asunercept plus irradiation
22 of 41 patients at #0.8 mg/m2); the RD for MRZ was determined to be significantly prolongs TtD and maintains QoL, versus reirradiation alone in
0.8 mg/m2. Most common TEAEs (all grades): fatigue, nausea (both 70%), progressive glioblastoma patients. Clinical trial information: NCT01071837.
hallucination (54%), vomiting (53%), headache (47%), confusional state References: 1. NCCN. Central Nervous System Cancers. Version 2018.2; 2. Stupp R
(33%), ataxia, constipation, muscular weakness (all 29%). Conclusions: CNS et al. Ann Oncol 2014;25(S3):iii93-iii101; 3. Wick et al. Clin Cancer Res
TEAEs were short-lasting, reversible and ameliorated by early dose reductions 2014;20:6304–13.
(29% patients dose-reduced), allowing patients to remain on treatment. For Asunercept + RT RT
patients receiving MRZ with TMZ/RT→TMZ (N=35), the median OS was N Median TtD, days N Median TtD, days P value
14.8 months (17 deaths, median follow-up 14.3 months), and 7 patients QLQ-CL15 PAL
remain active (Cycles 11-23). The MRZ RD + TMZ/Optune combination was Overall QoL 49 166 21 107 0.0099
Physical functioning 53 183 22 89 0.0069
tolerated, with 4 of 13 patients treated on this arm remaining active. An Emotional functioning 50 NR* 21 117 0.3002
Fatigue 50 98 21 88 0.5956
international Phase 3 trial (EORTC 1709-BTG/CCTG CE.8, NCT03345095) is QLQ-BN20 Total score 52 NR* 22 139 0.5419
ongoing. Clinical trial information: NCT02903069. MRC Neurological status 57 166 25 103 0.0319

* = Not reached

2023 Poster Session (Board #212), Sun, 8:00 AM-11:00 AM 2024 Poster Session (Board #213), Sun, 8:00 AM-11:00 AM
DGM1 may serve as a novel genetic biomarker of response to enzastaurin in Barriers to accrual and enrollment in brain tumor trials. First Author:
glioblastoma. First Author: Nicholas A. Butowski, University of California, Eudocia Quant Lee, Dana-Farber Cancer Institute, Boston, MA
San Francisco, CA
Background: A major impediment to improving neuro-oncology outcomes is
Background: Despite countless clinical trials being conducted, little has poor clinical trial accrual. Methods: We convened a multi-stakeholder group
changed over the last decade in the chemotherapies available for glioblastoma including Society for Neuro-Oncology, Response Assessment in Neuro-
(GBM) with survival remaining poor. Meaningful advances in treating this Oncology, patient advocacy groups, clinical trial cooperative groups, and
deadly malignancy may rely on precision medicine. We discovered a novel other partners to determine how we can improve trial accrual. Results: We
pharmacogenomic biomarker for enzastaurin (enz) in treating lymphoma describe selected factors contributing to poor trial accrual and possible
(lymph). We evaluated if this biomarker can be used to predict enz response in solutions. Conclusions: We will implement strategies with the intent to
GBM. Methods: Biomarker discovery was performed by a genome-wide screen double trial accrual over the next 5 years.
using DNA extracted from blood samples from a ph 3 enz lymph trial and
confirmed in an independent ph 2 enz lymph trial. The biomarker was then Challenges Potential solutions
evaluated for its predictability in GBM using the archived DNA samples from a Patient and Community Factors Engagement with patient advocacy
prior ph 1/2 enz GBM trial. Results: A novel biomarker, Denovo Genomic Limited patient awareness of trial op- groups. Improved online search
Marker 1 (DGM1), a germline polymorphism on chromosome 8, was found to portunities. Patient misconceptions tools, smart phone apps, or patient
about research study involvement. navigators.
be highly correlated with response to enz in the two lymph trials. Using DNA Disparities Unconscious bias training. Strengthen
extracted from blood of pts from the single-arm ph 1/2 study of newly di- Unconscious bias. Lack of diversity in pipeline of underrepresented minority
agnosed GBM receiving enz added to radiation and temozolomide (tmz), we oncology workforce. candidates.
found median OS for DGM1+ pts treated with enz was 18 mon vs 12.8 mon for Physician and Provider Factors Reinforce routine discussion of clinical
DGM1- pts, HR (95% CI) 0.68 (0.25, 1.81), p = 0.12. In addition, we found Failure to discuss clinical trials as an trials in addition to existing standard
pts in the GBM study receiving a mean daily dose of enz $ 245 mg had an OS option with patients. Failure to refer to therapies. Enhance incentives for
of 19.8 mon vs 14.9 mon for pts receiving a mean daily dose of , 245 mg [HR trials outside one’s own institution. patient enrollment in clinical trials or
referrals to centers for trials; make
(95% CI) 0.55 (0.34, 0.90)]; enz 500 mg/day was used in the lymph studies. most trials available more widely.
Conclusions: These data are supportive of DGM1 as a potentially predictive Clinical Trial Factors Minimize clinic visits.
biomarker for enz response in both lymph and GBM. There is an ongoing Patient/caregiver hardships due to Limit inclusion/exclusion to criteria
biomarker-driven pivotal ph 3 study in lymph at 500 mg/day, and DGM1 in frequent study center visits. critically relevant to study primary
GBM will be further evaluated in a planned randomized ph 2b study in newly Excessively stringent eligibility crite- endpoint.
diagnosed GBM with 500 mg/day of enz in combination with tmz. ria. Trial arms with limited equipoise. Include patients with primary and met
astatic brain tumors in early phase
oncology clinical trials. Weighted or
center-based randomization.
Site and Organizational Factors Effective leadership of multidisciplinary
Requirement for specialized person- team and organization culture to pro
nel, training, infrastructure, resources. mote accrual.
Limited support resources at centers. Greater partnership between academic
Travelling distances. and community oncology centers.
Better support for trial patients and
availability of trials in patients’ geo
graphic region.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Central Nervous System Tumors 93s

2025 Poster Session (Board #214), Sun, 8:00 AM-11:00 AM 2026 Poster Session (Board #215), Sun, 8:00 AM-11:00 AM
Phase 1/2a study of once daily oral BAL101553, a novel tumor checkpoint Plasma cell-free circulating tumor DNA (ctDNA) detection in longitudinally
controller (TCC), in adult patients with progressive or recurrent glioblastoma followed glioblastoma patients using TERT promoter mutation-specific droplet
(GBM) or high-grade glioma. First Author: Juanita Suzanne Lopez, Drug digital PCR assays. First Author: Christine Cordova, National Institutes of
Development Unit-The Institute of Cancer Research and The Royal Marsden Health, Bethesda, MD
NHS Foundation Trust, Sutton, United Kingdom
Background: There is a critical need for more specific and less invasive
Background: BAL101553 (prodrug of BAL27862) is a novel TCC that diagnostic and pharmacodynamic biomarkers in glioblastoma (GBM) pa-
promotes tumor cell death by modulating the spindle assembly checkpoint. tients (pts). Previously, we detected TERT promoter hotspot mutations
BAL27862 is a lipophilic, small molecule shown in rodents to penetrate the (C228T and C250T) in the ctDNA of IDH wildtype (IDHwt) TERT promoter
brain (brain/plasma ratio around unity), with promising antitumor activity in mutant GBM pts with 100% specificity using mutation-specific droplet
orthotopic preclinical GBM models as monotherapy or in combination with digital PCR (ddPCR) assays. Here, we explored the dynamics and clinical
radiotherapy (RT) with or without temozolomide. In this ongoing study associations of mutant TERT ctDNA levels in GBM pts undergoing therapy.
(NCT02490800, CDI-CS-002), daily oral BAL101553 was initially exam- Methods: We examined 14 pts with suspected IDHwt GBM based on pre-
ined in solid-tumor patients, with an MTD of 16 mg/d and DLTs of G4 operative MRI. Plasma was isolated and frozen from ~15 mL whole blood
hyponatremia and G2 hallucinations (Lopez 2018, JCO 36, 2018, suppl. samples collected pre- and post-op, at end of radiation (RT), and 1, 3, and
A2530). Subsequently the study was expanded by including a separate 6 m after end of RT. TERT promoter mutations were identified in FFPE tumor
cohort of patients with progressive or recurrent GBM or high-grade glioma samples using ddPCR assays for C228T/C250T. Plasma samples were
(Ingles Garces 2017, JCO 35, 2018, suppl. TPS2601). Methods: Patients analyzed using ddPCR assays specific for the corresponding tumor mutation.
with histologically-confirmed GBM or high-grade glioma, with progressive or The validated thresholds for positive detection were 1.5 (C228T) and 1.7
recurrent disease after prior RT with/without chemotherapy, received once- copies/mL (C250T). Results: 13/14 (92.9%) IDHwt tumors had TERT
daily oral BAL101553 (28-day cycles) in a 3+3 dose-escalation design to mutations (7 C228T and 6 C250T). Six of these 13 (46%) pts had positive
determine the maximum tolerated dose (MTD). Adverse events were plasma TERT ctDNA preop (4 C228T, 2 C250T). The mean cross sectional
assessed by CTCAE v4.03 grade (G), and tumor response by RANO every two area of enhancing disease at presentation for positive or negative preop
cycles. Pharmacokinetics (PK) were evaluated on Day 1 of Cycles 1 and 2. mutant ctDNA was similar. All 4 pts with multiple contrast enhancing lesions
Results: In the ongoing study, 23 pts (13M/10F; median age 50 y), median had positive preop mutant ctDNA. 2 pts who were negative initially de-
(min–max) number of prior regimens = 2 (1–5), received doses of 8, 15, 20, veloped detectable mutant ctDNA preceding progression. 3/4 pts with
25 or 30 mg oral BAL101553 once daily. One DLT of reversible G2 de- equivocal radiographic pseudoprogression had ctDNA dynamics that cor-
pression and fatigue occurred at 20 mg. Both mean Cmax and AUC increased related with eventual clinical outcome. One patient with unresectable GBM
with dose between 8 and 30 mg. The PK exposure in GBM patients was lower had declining mutant ctDNA in later collections during clinical stability.
than for solid tumor patients, in particular at 20 and 25 mg. At 25 mg/d (n = Conclusions: We detected plasma TERT ctDNA in 46% of TERT mutant
3), one patient with IDH-mutated GBM had a partial response (63% area GBM pts before surgery, and in 100% of pts with multiple contrast en-
reduction per RANO) and continues on study . 8 months, and another hancing lesions. TERT mutant ctDNA levels correlated with pseudoprog-
patient had stable disease for 5 months. At 15–20 mg/d, stable disease was ression or true disease progression and predicted progression before MRI.
observed in 3/10 patients. Conclusions: The current data in patients with These data suggest that larger studies to test circulating cell-free TERT
GBM or high-grade glioma suggest that BAL101553 is well tolerated at dose mutation as a diagnostic and pharmacodynamic biomarker in GBM are
levels above the MTD established in patients with advanced solid tumors, warranted.
and shows indications of clinical activity. Clinical trial information:
02490800.

2027 Poster Session (Board #216), Sun, 8:00 AM-11:00 AM 2028 Poster Session (Board #217), Sun, 8:00 AM-11:00 AM
Safety and activity of a first-in-class oral HIF2-alpha inhibitor, PT2385, in Survival outcomes in glioma patients with noncanonical IDH mutations:
patients with first recurrent glioblastoma (GBM). First Author: Roy E. Strowd, Beyond diagnostic improvements. First Author: Enrico Franceschi,
Wake Forest School of Medicine, Winston-Salem, NC Department of Medical Oncology, Bellaria Hospital, Azienda USL - IRCCS
Institute of Neurological Sciences, Bologna, Italy
Background: Hypoxia inducible factor 2-alpha (HIF2a) mediates cellular
responses to hypoxia and is overexpressed in GBM. PT2385 is an oral HIF2a Background: According to the 2016 WHO classification of Central Nervous
inhibitor with in vivo activity against GBM. Methods: A two-stage single-arm System tumors, the assessment of exon 4 mutations in IDH1 or IDH2 genes
open-label phase II study of adults with first recurrent GBM following is an essential step in the characterization of gliomas. The R132H mutation
chemoradiation with measurable disease was conducted through the Adult is the most frequent alteration in IDH1 gene, however other non-canonical
Brain Tumor Consortium. PT2385 was administered at the phase II dose IDH mutations have been identified. The aim of this study was to evaluate the
(800 mg b.i.d.). The primary outcome was objective radiographic response prognostic role of IDH non-canonical mutations. Methods: We analyzed our
(CR+PR); secondary outcomes were safety and survival. Exploratory ob- institutional data warehouse for all consecutive patients (pts) with newly
jectives included PK (day 15 Cmin), PD, and pH-weighted amine-CEST MRI diagnosed, histologically proven grade II – IV IDH mutant gliomas. IDH
to quantify tumor acidity at baseline and explore associations with drug sequencing was performed using the 454 GS-Junior next generation se-
response. Stage 1 enrolled 24 patients with early stoppage for #1 response. quencer (NGS) (Roche Diagnostic, Mannheim, Germany). All analyses were
Results: Of the 24 patients, mean age was 61611 years, median KPS 80, performed on DNA from formalin fixed and paraffin embedded (FFPE)
MGMT promoter methylated in 46%. PT2385 was well tolerated. Grade $3 specimens. Results: The analysis included 493 pts with IDH mutations. We
drug-related AEs were hypoxia (n = 2), anemia (1), hyperglycemia (1), found 279 (56.6%) grade 2, 173 grade 3 (35.1%) gliomas, and 41 (8.3%)
hyponatremia (2) and lymphopenia (2). No objective radiographic responses IDH mutant glioblastoma. Canonical IDH1 R132H mutation was found in
were observed; median PFS was 1.8 months (95%CI 1.6-3.1). Drug ex- 428 pts (86.8%). The remaining pts showed IDH2 (3.9%) or IDH 1 non-
posure varied widely (Table) and did not differ by corticosteroid use (p = canonical mutations (mainly R132C, R132G, R132S – 9.3%). Median
0.12), antiepileptics (p = 0.09), or sex (p = 0.37). Patients with high follow-up time was 80.5 months. Pts with non-canonical mutations showed a
systemic exposure had significantly longer PFS (6.7 vs 1.8 months, 0.009). younger median age (32 vs 39 years, p , 0.001). Other clinical charac-
Non-enhancing infiltrative disease with high acidity gave rise to recurrence. teristics and treatments were similar across IDH groups. Median survival was
Baseline acidity correlated significantly with treatment duration (R2= 0.49, 145 months (95%CI: 137.7 - 152.9) and 198.6 (95%CI 155.2– 242.1) in
p = 0.017). Conclusions: Drug exposure to PT2385 was variable. Signals of patients with IDH R132H and non-canonical mutations, respectively (p =
activity were observed in GBM patients with high systemic exposure and 0.013). In multivariate analysis grading (p , 0.001), extent of surgery (p ,
acidic (e.g. hypoxic) lesions on baseline imaging. A second-generation 0.001), 1p19q codeletion (p = 0.003) and presence of non-canonical
HIF2a inhibitor is being studied. Clinical trial information: NCT03216499. mutations (p = 0.022) showed a significant role for improved survival.
Conclusions: Detecting non-canonical IDH1 mutations is essential for di-
Day 15 Cmin (ng/ < 300 (n = 11, 300£Cmin < 1000 (n = ‡1000 (n = 3, P- agnosis and for prognosis in patients with gliomas. Differential enzymatic
mL) 55%) 6, 30%) 15%) Value
activity of non-canonical IDH1 mutations, resulting in different levels 2-
mPFS (months, 1.8 (0.9-3.1) 1.8 (1.6-5.2) 6.7 (5.4-11.3) 0.009 hydroxyglutaratecould be the reason of improved survival.
95%CI)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
94s Central Nervous System Tumors

2029 Poster Session (Board #218), Sun, 8:00 AM-11:00 AM 2030 Poster Session (Board #219), Sun, 8:00 AM-11:00 AM
Glioblastoma gene expression subtypes and correlation with clinical, molecular Study of tumor infiltrating immune CELLS and vasculature in human
and immunohistochemical characteristics in a homogenously treated cohort: gliomas: Differences in IDH1/2 mutant versus IDH1/2WT tumors. First
GLIOCAT project. First Author: Estela Pineda, Medical Oncology, Hospital Author: Maria Cruz Martin Soberón, Hospital Universitario 12 de Octubre,
Clinic Barcelona, Barcelona, Spain Madrid, Spain
Background: Glioblastoma (GBM) gene expression subtypes have been Background: Gliomas harboring mutations in IDH1/2 show a higher overall
described in last years, data in homogeneously treated patients is lacking. survival time than “wild type” (wt) tumors. Although the clinical aspects are
Methods: Clinical, molecular and immunohistochemistry (IHC) analysis well described, little is known about the underlying mechanisms by which
from patients with newly diagnosed GBM homogeneously treated with these mutations generate such a difference in the clinical course. Our group
standard radiochemotherapy were studied. Samples were classified based has recently described that IDH1/2 mutations induce a distinct vascular
on the expression profiles into three different subtypes (classical, mesen- phenotype in the tumors, with less blood-brain barrier (BBB) leakage than the
chymal, proneural) using Support Vector Machine (SVM), the K-nearest IDH1/2 wt gliomas (In Press, DOI:10.1101/541326). Methods: Prospective
neighbor (K-NN) and the single sample Gene Set Enrichment Analysis study analyzing a cohort of 20 patients with primary gliomas resected in one
(ssGSEA) classification algorithms provided by GlioVis web application. institution. Samples were obtained in the first surgery and 12 IDHmut and 8
Results: GLIOCAT Project recruited 432 patients from 6 catalan institutions, IDHwt gliomas were included. Immune infiltration was analysed by flow
all of whom received standard first-line treatment (2004 -2015). Best cytometry and vasculature by inmunohistochemistry. For molecular biology
paraffin tissue samples were selected for RNAseq and reliable data were studies, western blots were performed with Mini-PROTEAN system. Proteins
obtained from 124. 82 cases (66%) were classified into the same subtype by were visible by enhanced chemoluminescence. Results: We show that the
all three classification algorithms. SVM and ssGEA algorithms obtain more immune component also differentiates these two pathologies. There is sig-
similar results (87%). No differences in clinical variables were found be- nificantly less immune infiltration in IDH1/2 mutant gliomas. Within the CD45
tween the 3 GBM subtypes. Proneural subtype was enriched with IDH1 subset, IDH1/2 mutant gliomas have a reduced proportion of T lymphocytes
mutated and G-CIMP positive tumors. Mesenchymal subtype (SVM) was with a different T cell exhaustion profile and an increased proportion of CD11b+
enriched in unmethylated MGMT tumors (p = 0.008), and classical (SVM) in cells in comparison to IDH1/2 wt cases. Myeloid compartment distribution is
methylated MGMT tumors (p = 0.008). Long survivors ( . 30 months) were also different in these two types of tumors, showing an augmented proportion
rarely classified as mesenchymal (0-7.5%) and were more frequently of the M2 (CD206+) and the neutrophil subsets in IDH1/2 wt gliomas.
classified as Proneural (23.1-26.). Clinical (age, resection, KPS) and mo- Moreover, a higher proportion of CD45 PDL1+ was present in the IDH1/2 wt
lecular (IDH1, MGMT) known prognostic factors were confirmed in this serie. tumors samples. The analysis of the vasculature showed an increase density
Overall, no differences in prognosis were observed between 3 subtypes, but a and the lumen size of the vessels of the IDH1/2 wt compared to the IDH1/2
trend to worse survival in mesenchymal was observed in K-NN (9.6 vs 15 ). mutant gliomas which correlate with changes in the immune profile. The
Mesenchymal subtype presented less expression of Olig2 (p , 0.001) and biochemical analysis showed that there is an increment in EGFR and PDGFR
SOX2 (p = 0.003) by IHC, but more YLK-40 expression (p = 0.023, SVM). On activity in the IDH wt gliomas that is related with more vascular aberrations
the other hand, classical subtype expressed more Nestin (p = 0.004) com- and higher CD45 infiltrate. This suggests that EGFR and PDGFR are the key
pared to the other subtypes (K-NN). Conclusions: In our study we have not regulators of the tumor microenvironment. Conclusions: To understand the
found correlation between glioblastoma expression subtype and outcome. matching between the immune infiltration and vasculogenesis is relevant for
This large serie provides reproducible data regarding clinical-molecular- interpreting data coming from the clinical trials with checkpoints inhibitors.
immunohistochemistry features of glioblastoma genetic subtypes. At the time abstract submission survival analysis is not yet available due to
the short time of follow-up but in May 2019, the number of expected events
for analysis will be reached.

2031 Poster Session (Board #220), Sun, 8:00 AM-11:00 AM 2032 Poster Session (Board #221), Sun, 8:00 AM-11:00 AM
Phase I trial of TG02 plus dose-dense or metronomic temozolomide for Phase II study to evaluate safety and efficacy of MEDI4736 (durvalumab) +
recurrent anaplastic astrocytoma and glioblastoma in adults. First Author: radiotherapy in patients with newly diagnosed unmethylated MGMT glioblastoma
Jing Wu, NCI, Bethesda, MD (new unmeth GBM). First Author: David A. Reardon, Dana-Farber Cancer Institute
and Harvard Medical School, Boston, MA
Background: Therapies targeting multiple survival pathways simultaneously
may be more effective for high-grade gliomas, a disease highly resistant to Background: Durvalumab (durva), a human IgG1 monoclonal Ab against PD-
treatment. Our preclinical studies have shown potent anti-glioma effects of L1, is FDA-approved for selected patients with bladder and non-small cell
TG02 and synergy with temozolomide (TMZ) through modulation of tran- lung cancers. PD-L1 is expressed by some GBM tumors, while GBM in-
scription and cellular metabolism. A phase I/II trial was launched to test the filtrating T lymphocytes often express PD-1. Radiation induced cell death
combination of TG02 and TMZ in recurrent malignant gliomas and herein we releases tumor antigens and could potentiate anti-PD-(L)1 therapy.
report the phase I results. Methods: Adults with recurrent high-grade as- Methods: This ongoing Phase 2 open-label study (NCT02336165) evaluates
trocytoma, KPS $ 60, normal organ function, # 2 prior relapses were the safety and efficacy of durva (10 mg/kg every 2 weeks) in 5 GBM cohorts.
enrolled. The primary endpoint was dose limiting toxicity (DLT) from the start Results are presented for Cohort A, which evaluates durva + standard ra-
of the combined treatment to 4 weeks after in each arm. Bayesian optimal diotherapy (RT, 60 Gy over 30 fractions) followed by durva monotherapy in
interval (BOIN) design was employed to determine the maximum tolerated patients with new unmeth GBM after maximum safe resection. The primary
dose (MTD) with the target DLT rate of 35% and the toxicity profile of the efficacy endpoint for Cohort A is overall survival at 12 months (OS12);
combination of TG02 (starting dose 200mg orally on days 1, 12, 15, and 26) secondary endpoints include safety/tolerability, tumor response rate, and
and TMZ, either as a dose-dense (DD; 125mg/m2/d, 7on/7off, Arm 1) or progression-free survival (PFS). Historical benchmarks of median OS and
metronomic (MN; 50mg/m2/d, Arm 2) dosing schedule on a 28-day cycle. OS12 for patients with new unmeth GBM following standard therapy are
Results: Forty patients were enrolled; 38 were evaluable; 70% male; overall 12.7 months and 50%, respectively (EORTC 26981-22981/NCIC CE.3).
median age 50.7; median KPS 90. Of 18 evaluable patients in Arm 1 (DD Results: Median follow-up of 40 enrolled patients is 24.5 months (data
TMZ), at TG02 dose level 200mg, 1/6 had a DLT: Gr3 diarrhea. At TG02 dose cutoff = 05 Nov 2018). Baseline characteristics: male, 70%; median age,
level 250mg, 3/12 had DLTs: Gr4 neutropenia for over 5 days, Gr3 elevated 57.0 [22 to 77] years; ECOG PS0, 60.0%; ECOG PS1, 40.0%; measurable
ALT, and Gr3 fatigue. Of 20 evaluable patients in Arm 2 (MN TMZ), at TG02 disease, 80.0%; and dexamethasone use, 32.5%. Treatment-related ad-
dose level 200mg, 1/6 had a DLT: recurrent Gr3 neutropenia. At TG02 dose verse events with maximum CTCAE grade $ 3 occurred in 14 (35.0%)
level 250mg, 5/12 had a DLT: Gr3 elevated ALT, Gr3 fatigue, and Gr4 patients; the most common were asymptomatic increased lipase (n = 6) and
neutropenia. At TG02 dose level 300mg,1 out of 2 had a DLT: Gr4 febrile increased amylase (n = 2). Twenty-four of 40 patients were alive at
neutropenia, Gr4 elevated ALT, Gr4 elevated AST, which resulted in hos- 12 months (Kaplan-Meier for OS12, 60.0% [90% CI: 46.1, 71.4]). Median
pitalization. Therefore, the TG02 dose level of 250mg was declared as the OS was 15.1 (95% CI: 12.0, 18.4) months. As of 05 Nov 2018, 8 (20%)
MTD in both Arm 1 and Arm 2. Conclusions: The combination of TG02 at the patients remain alive, with ongoing survival ranging from 15.7 to
MTD of 250mg with DD or MN TMZ has a tolerable toxicity profile. Cohort 34.9 months. Tumor immunocorrelative and systemic studies are pending.
expansion continues at the MTD in both arms to conduct pharmacokinetics Conclusions: This is the first study report of anti-PD-L1 for new GBM. Durva
and pharmacogenetics to better elucidate the toxicity profile. Objective was well tolerated when combined with RT and seemed to have efficacy
responses have been observed, suggesting activity of this regimen and among patients with new unmeth GBM. Further studies may be warranted.
supporting continued investigation with the phase II randomized compo- Clinical trial information: NCT02336165.
nent. Clinical trial information: NCT02942264.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Central Nervous System Tumors 95s

2033 Poster Session (Board #222), Sun, 8:00 AM-11:00 AM 2034 Poster Session (Board #223), Sun, 8:00 AM-11:00 AM
A TITE-CRM phase I/II study of disulfiram and copper with concurrent GLIAVAX: A stratified phase II clinical trial of avelumab and axitinib in
radiation therapy and temozolomide for newly diagnosed glioblastoma. patients with recurrent glioblastoma. First Author: Bart Neyns, Universitair
First Author: Jiayi Huang, Washington University School of Medicine in Ziekenhuis Brussel, Brussels, Belgium
St. Louis, St. Louis, MO
Background: Patients (pts) with recurrent glioblastoma (rGB) have a poor
Background: Disulfiram (DSF) has shown promising activity against glio- prognosis, and no treatment option demonstrated to improve survival in a
blastoma in preclinical studies and is more effective when combined with randomized trial. Axitinib (AXI), an oral VEGFR 1-3 inhibitor has demon-
copper (Cu). Our previous phase I study established the maximum tolerated strated single agent activity in rGB and reduces the need for corticosteroids
dose (MTD) of DSF when combined with adjuvant temozolomide (TMZ). This (CS). Avelumab (AVE) is a fully human anti-PD-L1 IgG1 antibody with
phase I/II study aims to establish the MTD when disulfiram and copper are clinical activity in various tumor types. Combination of AXI and AVE may
combined with concurrent radiation therapy (RT) and TMZ for newly di- improve the outcome of pts with rGB. Methods: This open-label, dual-strata,
agnosed glioblastoma and to explore preliminary efficacy. Methods: Eligible single-center phase 2 clinical trial investigated the activity of AXI plus AVE in
patients were treated with standard RT and TMZ plus escalating doses of DSF adult pts with rGB following prior surgery, RT and temozolomide. Pts were
(250 mg - 375 mg PO QD) and Cu (2 mg PO TID), followed by adjuvant TMZ stratified according to their baseline use of CS. Pts without baseline need for
plus DSF (500 mg/day) and Cu. The time-to-event continual reassessment CS initiated treatment with AXI (5 mg oral BID) plus AVE (10 mg/kg IV Q2W)
method (TITE-CRM) was used to continuously estimate the probability of (cohort-1). Pts in need of CS initiated AXI as a monotherapy; AVE could be
dose-limiting toxicity (DLT) and to assign patients to doses with an estimated added to AXI after 6 wks if the CS dose could be tapered to a physiologic dose
DLT probability of approximately 20% with a margin of 5%. Tumor mutations level or less (cohort-2). Six-month-PFS served as the primary endpoint
were evaluated with next-generation sequencing for all patients. (with a prespecified threshold of $ 50% for cohort-1) according to Fleming
Results: Eighteen glioblastoma patients were treated with the study therapy: one-stage design. Results: Between Jun 2017 and Aug 2018, 54 pts (27 per
8 with DSF of 250 mg/day and 10 with 375 mg/day. Three DLTs were cohort) were enrolled (med age 55 y [range 19-75]; 63% male; 91% WHO
observed: 1 with 250 mg/day (grade 2 urinary incontinence and ataxia), and PS 0-1). All pts in cohort-1 and 16 pts (59%) in cohort-2 received at least 1
2 with 375 mg/day (both grade 3 elevated liver enzymes). DSF had an dose of AVE. The 6-month-PFS was 18% (95% CI 4-33) in both cohorts. At
estimated DLT probability of 10% (95% CI: 3-29%) at 250 mg/day, and the time of analysis, 2 pts were progression-free and continuing study
21% (95% CI: 7-42%) at 375 mg/day. After a median follow-up of treatment. Median OS in cohort-1 and -2 was respectively 26 wks (95% CI
12.3 months, 1-year progression-free survival (PFS) was 57%, and 1-year 21-32) and 18 wks (95% CI 14-22). No clear relation was found between
overall survival (OS) was 69%. There was no significant difference in PFS/OS baseline cognitive functioning (Cogstate subtests) and PFS/OS. The best
when stratified by DSF doses, surgical extent, or MGMT methylation status. overall response rate (iRANO) was 41% and 26% respectively for pts in
However, glioblastomas with IDH1 (n = 6), BRAF (n = 2), or NF1 (n = 1) cohort-1 and -2. The most frequent all-grade treatment-related adverse
mutations had significantly better PFS and OS than those without the events (TRAE) were dysphonia (67%), lymphopenia (50%), diarrhea (48%),
mutations: 1-year PFS: 100% vs 22%, respectively, p = 0.001; 1-year OS: hypertension (48%), and fatigue (46%). The incidence of grade 3-4 TRAE
100% vs 42%, respectively, p = 0.006. Conclusions: The MTD of DSF with was 30%; there were no grade 5 AE. Conclusions: The combination of AVE
RT/TMZ/Cu for glioblastoma is 375 mg/day, and the recommended phase II plus AXI is sufficiently well tolerated but did not meet the threshold for
dose is 250 mg/day. Although confirmation with larger sample size is activity justifying further investigation in an unselected population of pa-
needed, the combination demonstrates promising preliminary efficacy for tients with rGB. Clinical trial information: NCT03291314.
the subset of glioblastoma with IDH1, BRAF, and NF1 mutations. Clinical
trial information: NCT02715609.

2035 Poster Session (Board #224), Sun, 8:00 AM-11:00 AM 2036 Poster Session (Board #225), Sun, 8:00 AM-11:00 AM
Clinical characteristics, treatment (Tx) patterns, and overall survival (OS) in Effect of grade on survival in IDH-mutant grade II and grade III gliomas. First
advanced (Adv) NSCLC patients (Pts) with and without brain metastases Author: Giuseppe Lamberti, Department of Medical Oncology S.Orsola
(BM). First Author: Emily Nash Nash Smyth, Eli Lilly and Company, Indianapolis, Malpighi Hospital Bologna, Bologna, Italy
IN
Background: The 2016 WHO classification dramatically changed the di-
Background: BM in NSCLC pts are associated with significant morbidity and agnosis of gliomas. Diffuse gliomas are classified according to the presence
mortality. This analysis describes the frequency and timing of BM devel- of IDH-mutation (IDH-mut) and the deletion of both 1p and 19q chromo-
opment, pt characteristics, systemic txs, and OS in NSCLC pts with and some arms (1p/19q codel). Now debate is whether grade still has an in-
without BM. Methods: This retrospective observational study identified pts dependent prognostic value. The aim of this study was to find out if grade is a
from the Flatiron-Foundation Medicine NSCLC Clinico-Genomic Database prognostic factor independently of molecular status. Methods: We analyzed
diagnosed from 1 Jan 2011 to 31 Oct 2017 with adv NSCLC and a tumor our institutional data warehouse for all consecutive patients (pts) with newly
sample analyzed via FoundationOne. Tx pattern data were summarized by diagnosed, histologically proven Grade II or Grade III IDH-mut gliomas. IDH
period (1 Jan 2011-1 Mar 2015; 2 Mar 2015-31 Dec 2017), therapy class 1/2 assessment by polymerase chain reaction (PCR)or immunohistochem-
(eg, anti-VEGF and EGFR, platinum-based), and BM occurrence. Descriptive istry (IHC) was accepted. Next Generation Sequencing (NGS) forIDH1(exon
statistics were used to summarize data; Chi-square and t-tests assessed 4) and IDH2(exon 4) was performed on all specimens wild-type for the IDH.
statistically significant differences. OS was measured by site of met (BM only Results: The analysis included all the 399 pts who had a grade II (n = 250,
vs no-BM only vs BM and no-BM) via K-M methods from adv diagnosis until 62.7%) or grade III (n = 149, 37.3%). Median follow-up time was 105.3 months.
death or last activity date (censored). Results: Of 3257 pts, 1018/3257 After surgery, 72 pts (18.0%) received RT alone, 44 (11.0%) received CT alone,
(31.3%) had BM during follow-up; 726/1018 (71.3%) presented with BM 135 (33.8%) received both RT and CT, and 142 (35.6%) follow-up without any
within 30 days of adv diagnosis. The median age at adv diagnosis was 66.2 treatment. Median survival was 148.1 months. In multivariate analysis Grade
yrs. Relative to pts without BM, BM pts were younger, more likely to be (HR = 0.342, 95%CI: 0.221 – 0.531; P , 0.001) and 1p/19q codeletion (HR =
female, of Asian descent, have stage IV disease, $2 met sites (including BM) 0.440, 95%CI: 0.290 – 0.668; P , 0.001) were independently associated with a
at initial presentation, $3 met sites (including BM) during follow-up, and lower risk for death. The difference in survival remained when adjusted for his-
non-squamous histology (all p , 0.01). Approximately 78% (n = 2534) were tological subtype. Residual disease after surgery or biopsy negatively affected
treated with $1 systemic tx; platinum-based chemo-combinations were the survival (HR 2.151, 95%CI 1.375 – 3.367, P = 0.001). Post-surgical
most common 1st line tx, regardless of BM status. Increased use of PD-1/L1 treatment with RT + adjuvant CT improves survival in respect to follow-up
tx was seen in 1st, 2nd, and 3rdline during the latter vs earlier period. No and other treatments (HR: 0.316, 95%CI 0.156 – 0.641, P = 0.001).
statistically significant difference in OS was observed in pts with BM only Conclusions: Grade still affects survival in IDH mutant Grade II and III gli-
(17.1 mos; 95% CI 12.5-29.9), no-BM only (21 mos; 95% CI 19.4-22.8), or omas. This effect was independent onmolecular features, surgical extension
BM and no-BM (20.4 mos; 95% CI 18.9-23.3) (log rank p = 0.3027). and post-surgical treatments. Clinical management of gliomas should con-
Conclusions: In met NSCLC pts with a tumor sample that was molecularly tinue to take into account grade as well as molecular characteristics.
profiled, OS was comparable, regardless of site(s) of disease; additional
multivariate analyses including molecular profiles are needed. BM screening
at initial diagnosis is important given the frequency in NSCLC. Future studies
should assess whether the shift in systemic tx patterns impact the devel-
opment and clinical outcomes.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
96s Central Nervous System Tumors

2037 Poster Session (Board #226), Sun, 8:00 AM-11:00 AM 2038 Poster Session (Board #227), Sun, 8:00 AM-11:00 AM
Adjuvant chemotherapy to improve survival in average-risk adult medulloblas- Phase II trial of palbociclib in recurrent RB-positive anaplastic oligoden-
toma patients: Long-term results. First Author: Giuseppe Lamberti, Department droglioma: A Spanish group for research in neurooncology (GEINO) trial. First
of Medical Oncology S.Orsola Malpighi Hospital Bologna, Bologna, Italy Author: Juan Manuel Sepulveda-Sanchez, Hospital Universitario 12 de
Octubre, Madrid, Spain
Background: Medulloblastoma is extremely rare in adults and, therefore, it is
difficult to accrual patients in clinical trials. Radical surgery and radio- Background: The pRB-dependent cell cycle checkpoint is altered in the vast
therapy (RT) provide a significant control of disease. Nevertheless, about majority of anaplastic oligodendrogliomas (AO), either by homozygous de-
25% of average-risk patients have a relapse and die because of disease letion or by hypermethylation of CDKN2A and/or CDKN2B, or by amplifi-
progression. The role of chemotherapy (CT) after standard RT for average-risk cation and/or overexpression of CDK4. Palbociclib is an oral inhibitor of
adult patients remains controversial. Methods: We analyzed 48 average-risk CDK4 and 6 that has already been shown to be highly active in breast cancer.
patients according to Chang classification diagnosed from 1988 to 2016. Methods: We conducted a multicenter, open-label, phase II trial evaluating
Median age was 29 years (range 16-61), M/F ratio was 26 (54.2%)/22 efficacy and safety of Palbociclib in patients with AO that progressed to
(45.8%). Fifteen patients had classic medulloblastoma (31.3%), 15 pa- radiotherapy and more than one chemotherapy regimen containing Temo-
tients had desmoplastic medulloblastoma (31.3%), 5 patients had extensive zolomide and/or Lomustine. Inclusion criteria included: histologically and
nodularity (10.4%) and 2 patients had large cells/anaplastic histology molecularly confirmed grade III oligodendroglioma (WHO 2016 classifica-
(4.2%). The patients were homogeneously distributed in the two groups: 24 tion, IDH1/2 mutation and 1p/19 codeletion were mandatory), recurrence
(50%) received adjuvant RT alone and 24 (50%) received RT + CT that after radiotherapy and 1 or 2 chemotherapy regimens and conserved RB
consisted in a platinum-etoposide based combination. Results: After a protein expression by immunohistochemistry (IHC). Patients were treated
median follow-up of 12.5 years, CT increases progression-free survival rate at with Palbociclib 125 mg/daily 3 weeks on/1off. The primary objective of the
15 years (PFS-15 82.3 6 8.0% in RT-CT group vs. 38.5% 6 13.0% in RT study was progression-free survival at 6 months (6M-PFS). Results: Between
group p = 0.05) and overall survival rate at 15 years (OS-15 89.3% 7.2% vs. October 2015 and September 2018, 34 patients were enrolled across ten
52.0% 13.1%, p = 0.02). Among patients receiving CT, the reported hospitals. The study was stopped early secondary to lack of efficacy, with
grade $ 3 adverse events were: 9 cases of neutropenia; 6 cases of G3 74% of evaluable patients progressing within 6 months. Number of patients
neutropenia (25%) and 3 cases of G4 neutropenia (13%), 1 case of G3 alive and free from progression at 6 months after the enrollment was 9 (26%)
thrombocytopenia (4%) and 2 cases of G3 nausea (8%). Conclusions: Our out of the first 34 patients, below the minimum number required (18 out of
study with a long follow up period suggests that adding adjuvant chemo- 40) to consider Palbociclib as an active drug in this population. With a
therapy to RT might improve PFS and OS in average-risk adult medullo- median follow-up of 11.2 months, the median PFS was 3 months (95% CI:
blastoma patients. 2.5-3.5 months). Median overall survival (OS) was 23.1 months (95% CI:
17.2-25 months). There were no partial or complete responses and only 11
patients (32%) achieved stable disease as best response. Palbociclib was
well tolerated with neutropenia (Grade 3 or 4: 40%) and thrombocytopenia
(Grade 3 or 4: 15%) as the most common adverse effects (AEs). Both AEs
had no significant impact since there were no episodes of febrile neutropenia
or bleeding. Conclusions: Despite the good tolerance and drug exposure,
Palbociclib monotherapy did not show favorable activity in recurrent AO.
Clinical trial information: NCT02530320.

2039 Poster Session (Board #228), Sun, 8:00 AM-11:00 AM 2040 Poster Session (Board #229), Sun, 8:00 AM-11:00 AM
MDNA55: A locally administered IL4 guided toxin as a targeted treatment Cancer differentiation analysis technology as a novel technology for cerebral
for recurrent glioblastoma. First Author: Dina Randazzo, Duke University cancer screening. First Author: Hongmei Tao, AnPac Bio-Medical Science
Medical Center, Durham, NC and Technology Co., LTD, Shanghai, China
Background: IL4 receptor (IL4R) is frequently and intensely expressed on a Background: While the current cancer screening methods mostly failed to
variety of human cancers and is associated with poor survival outcomes. detect cerebral cancer, a novel, promising technology named cancer dif-
Determining the role of the IL4R biomarker in glioblastoma (GBM) will be ferentiation analysis (CDA) technology has been developed to measure novel
important for treatment with targeted therapies such as the IL4 fusion toxin bio-physical properties to obtain valuable multi-level and multi-parameter
MDNA55. Methods: A classification for IL4Ra expression in GBM tissues by information including protein, cellular and molecular level information.
H-Score was developed using a validated immunohistochemistry-based ap- Initial results showed that CDA technology is capable of detecting cerebral
proach. MDNA55-05 is an open-label study of MDNA55 administered cancer with a high degree of sensitivity and specificity. Methods: In this
intratumorally via convection enhanced delivery in recurrent GBM. Levels of study, samples from 78 cerebral cancer patients and 321 healthy individuals
IL4Ra expression were assessed retrospectively in 24 subjects in the clinical were measured. Peripheral blood of each individual was drawn in EDTA
trial and were correlated with GBM history, imaging responses and survival tubes. One class of bio-physical property in blood samples was utilized for
outcomes following treatment with MDNA55 to explore clinical validation. CDA tests. CDA data were conducted using SPSS, and the results were
Results: Range, linearity, specificity and sensitivity testing using a rabbit shown in table. Results: The average CDA values of cerebral cancer and
polyclonal antibody to IL4Ra were performed using normal cortex (negative control groups were 52.30 and 33.38 (rel. units) respectively. The results
control) and a panel of normal human tissues and GBM cases from tissue indicated that cerebral cancer could be significantly distinguished from the
banks. A total of 41 GBM samples were screened and grouped by reactiv- control (p , 0.001). Area under ROC curve (AUC) was 0.980, and sensitivity
ity thresholds: H-Scores $50 were observed in 95% of cases (39/41), and specificity was 92.3% and 96.6% respectively. Conclusions: Initial
H-Scores $200 were observed in 51% of cases (21/41), and H-Scores $250 results showed that CDA technology could effectively distinguish cerebral
were observed in 24% of cases (10/41). GBM tissues obtained at initial diagnosis cancer from healthy individuals. As a novel bio-physical based cancer de-
from subjects enrolled in the trial show that moderate/high IL4R expression tection approach with multi-level and multi-parameter expressions, CDA
(H-Score . 75) was associated with shorter time to first relapse when compared to could be a potential candidate for cerebral cancer screening. Results from
subjects with low IL4R expression (H-Score # 75) (10.3 mos vs. 16.7 mos, Statistical Analysis of CDA.
respectively) after upfront standard-of-care treatment, consistent with published Average Median SD of
findings that IL4R expression is associated with more aggressive disease. Re- CDA Gender Age Average Median CDA CDA CDA AUC
markable decreases in tumor size seen in some subjects following MDNA55 Data (Male Range Age Age (rel. (rel. (rel. (rel.
Group Set %) (year) (year) (year) units) units) units) units) Sensitivity Specificity
treatment were associated only with moderate/high IL4R expression and survival
rate at 12 months in this group was also improved (OS12 = 55%) compared to Control 321 64 30 - 53 53 33.30 33.38 5.81 / / /
86
subjects with low IL4R expression (OS12 = 30%). Conclusions: Treatment options Cerebral 78 65 30 - 55 57 52.30 51.72 8.06 0.980 92.3% 96.6%
for patients with recurrent GBM are very limited and positive outcomes remain rare. cancer 83
Targeting therapies such as MDNA55 by IL4R status may improve patient out-
comes and help guide patient selection strategies for future clinical studies.
Clinical trial information: NCT02858895.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Central Nervous System Tumors 97s

2041 Poster Session (Board #230), Sun, 8:00 AM-11:00 AM 2042 Poster Session (Board #231), Sun, 8:00 AM-11:00 AM
A gene signature of response to radiotherapy in patients with grade II-III Health-related quality of life (HRQL) in VERTU: A randomized phase II trial
oligodendrogliomas. First Author: Elizabeth Moyal, Institut Claudius of veliparib (V), radiotherapy (RT), and temozolomide (TMZ) for newly
Regaud, IUCT-O, Toulouse, France diagnosed MGMT unmethylated (uMGMT) glioblastoma (GBM). First Author:
Hao-Wen Sim, The Kinghorn Cancer Centre, St Vincent’s Hospital Sydney,
Background: Grade II and III Oligodendroglioma associate mutations of
Sydney, Australia
isocitrate deshydrogenase 1 or 2 genes and the whole-arm chromosomal loss
of 1p and 19q and have a better prognosis than other gliomas. However, even Background: The VERTU trial (ANZCTR #ACTRN12615000407594)
if the preferred treatment consists of a combination of radiotherapy (RT) and compared Arm A (standard of care) = RT (60Gy/30 fractions) + TMZ (75mg/
chemotherapy, some patients will less respond to this treatment and will m2 daily) followed by TMZ (150–200mg/m2 D1–5) every 28 days for 6 cycles
relapse faster , in part because of an heterogeneity in the response to RT. In vs Arm B (experimental arm) = RT (60Gy/30 fractions) + V (200mg PO BID)
the aim to identify factors of response to RT, we analyzed clinical and followed by TMZ (150–200mg/m2 D1–5) + V (40mg PO BID, D1–7) every
molecular data of patients with grade II-III oligodendroglioma exclusively 28 days for 6 cycles in pts with newly diagnosed centrally determined
treated with RT in the POLA cohort. Methods: Gene expression profiles on uMGMT GBM. To ensure that veliparib was not associated with clinical
Affymetrix expression arrays of patients from the POLA cohort with co- detriment, serial HRQL assessments were performed for comparison as a
deleted 1p/19q grade II/II gliomas treated by exclusive RT were used to secondary objective. Methods: Pts completed the EORTC quality of life core
identify a gene expression set predictive of radiation sensitivity. The primary questionnaire (QLQ-C30) and brain cancer module (BN20) every 4 weeks (w)
endpoint was the progression free survival (PFS), defined as the time from (baseline: w0; concurrent: w4,8; adjuvant: w10,14,18,22,26,30). Based
treatment start until progression or death. A supervised approach with pe- on relevance to GBM patients, 5 HRQL scales (global health [GH], physical
nalized regression was applied to select most informative predictors, and functioning [PF], social functioning [SF], motor dysfunction [MD] and
then a risk score was created based on the linear predictor given by the communication deficit [CD]) were pre-selected for primary analysis. Maxi-
multivariable model. Results: Forty-five patients corresponded to the study mum change from baseline score (clinically relevant deterioration/
criteria, with a median age at diagnosis of 45 (range 23- 64). The supervised improvement defined as $10-point change) during the progression-free
approach allowed identifying a three-gene prognostic set including Sem- period, and deterioration-free survival (time to deterioration/progression/
aphorin -3C (SEMA3C), Neuronal Pentraxin 2 (NPTX2 ) and the Metabo- death) were evaluated. Results: Patient characteristics were well-matched
tropic Glutamate Receptor 5 (GRM5), involved in proliferation, migration and (Arm A: N = 41, median age = 62, male = 68%, ECOG 0 = 66%, macroscopic
inflammation. The risk score associated to these three genes was statistically resection = 88%; Arm B: N = 84, median age = 60, male = 70%, ECOG 0 =
associated to PFS (HR = 2.72, p = 0.00005) and remains significant when 65%, macroscopic resection = 86%). Almost all completed at least one
adjusted on clinical covariates age at diagnosis, necrosis, endothelial pro- HRQL assessment (98%). HRQL assessments during the progression-free
liferation and type of surgery (complete, partial or subtotal surgery) (HRadj = period were completed in 87% (Arm A) and 90% (Arm B) of cases. For Arm A
2.36, p = 0.001). Conclusions: We report an independent three genes vs B, the proportion of patients who experienced a deterioration in GH (59%
SEMA3C-NPTX2-GRM5 risk score signature of response to radiotherapy in vs 64%, p = 0.69), PF (53% vs 53%, p . 0.99), SF (46% vs 53%, p = 0.56),
patients with oligodendroglioma, which highlights the heterogeneous re- MD (63% vs 58%, p = 0.70) and CD (45% vs 46%, p . 0.99) were similar.
sponse in this reputed good prognosis population. This signature could help in Deterioration-free survival was not statistically different for any HRQL item.
determining the adapted treatment as well as potential new targets to address. Conclusions: The addition of veliparib to standard of care for newly di-
agnosed uMGMT GBM does not appear to compromise HRQL. This would
support the primary efficacy analysis of the VERTU trial. Clinical trial in-
formation: ACTRN12615000407594.

2043 Poster Session (Board #232), Sun, 8:00 AM-11:00 AM 2044 Poster Session (Board #233), Sun, 8:00 AM-11:00 AM
Pembrolizumab (Pem) in recurrent high-grade glioma (HGG) patients (PTS) A preliminary comprehensive molecular-based nomogram for individualized
with mismatch repair deficiency (MMRd): An observational study. First estimation of survival in patients with newly diagnosed glioblastoma utilizing
Author: Giuseppe Lombardi, Department of Oncology, Oncology 1, Veneto global microRNA expression data. First Author: Denise Fabian, Ohio State
Institute of Oncology IOV-IRCCS, Padua, Italy University, Columbus, OH
Background: Pem, an immune checkpoint inhibitor, demonstrated to be Background: Glioblastoma (GBM) is the most aggressive and common pri-
active in various neoplasms with MMRd. No data exists about its efficacy in mary brain tumor. Nomograms are prediction models that help form in-
MMRd glioma PTS. Methods: MMRd HGG relapsed after receiving RT and dividualized risk scores for cancer patients, which are valuable for treatment
CT were treated with Pem. MMR status was analyzed by immunohisto- decision-making. The aim of this study is to create a refined nomogram by
chemistry, including the MLH1, MSH2, MSH6, and PMS2 markers. MMRd including novel molecular variables beyond MGMT promoter methylation.
was defined as presence of a weak (wMMRd) or absent (aMMRd) signal for at Methods: Clinical data and miRNA expression data were obtained from 226
least one MMR protein. Other inclusion criteria were: ECOG PS 0-2, histo- newly diagnosed GBM patients. Clinical data included age at diagnosis, sex,
logically confirmed glioma, dexamethasone #4 mg. Pem was administrated at Karnofsky performance status (KPS), extent of resection, O6-methylguanine-
200 mg every 3 weeks until disease progression or unacceptable toxicity. DNA methyltransferase (MGMT) promoter methylation status, IDH mutation
Tumor response was evaluated by brain MRI every 10 weeks according to the status and overall survival. Due to low representation of less than 13 cases
RANO criteria. OS and PFS were evaluated by Kaplan-Meier curves. each, IDH mutant glioblastomas and patients submitted to biopsy-only were
Results: among 167 glioma PTS, we found 22 MMRd gliomas. 12 PTS were excluded. Total RNA was isolated from formalin-fixed paraffin-embedded
treated with Pem: 8 wMMRd and 4 aMMRd. According to Bethesda criteria, all (FFPE) tissues; miRNA expression was subsequently measured using the
PTS had microsatellite stability. Tumor histologies included 5 anaplastic NanoString human miRNA v3a assay. A Cox regression model was developed
astrocytoma, 1 anaplastic oligodendroglioma, 6 glioblastoma (GBM). MSH2 using glmnet R package with the elastic net penalty while adjusting for known
deficiency was found in 6 cases , MSH6 deficiency in 9 cases, PMS2 and prognostic factors. A dichotomized genomic score was created by finding the
MLH1 deficiency in 2 cases. Median number of prior lines of chemotherapy optimal cutpoint (maximum association with survival) of the linear combi-
was 1 (range 1-5). Stable disease (SD) was reported in 4 PTS (33%); 8 PTS nation of the selected. A nomogram was generated using known clinical
showed progressive disease (PD). PTS with anaplastic gliomas showed a prognostic factors, specifically age, sex, KPS, and MGMT status along with the
statistically significant association with SD (p=0.03, OR=3); all GBM PTS dichotomized genomic score. Results: Four novel miRNAs were found to
reported PD; status of MMRd (weak/absent), IDH (mutated/wild-type), MSH2 significantly correlate with overall survival and were used to create the di-
and MLH6 (deficient/proficient) were not associated with SD. Median follow chotomized miRNA genomic score (GS). This score split the cohort into a poor
up was 14.7 ms. OS was 5.6 ms (95% CI 0.1-13.8), PFS 2.4 ms (95% CI 1.8- performing group (GS_high) and a better performing group (GS_low) (p =
2.9). OS was 2.8 ms and 5.6 ms (p=0.9), PFS was 1.8 ms and 3.1 ms (p=0.5) 0.0031). A final nomogram was created using the Cox proportional hazards
in PTS with wMMRd and aMMRd, respectively. PTS reporting SD and PD had model (Figure 1). Factors that correlated with improved survival included
PFS of 7.4 ms (95% CI 4.6-10.2) and 1.8 ms (95% CI 0.2-3.4), p=0.002; OS younger age, KPS . 70, MGMT methylation and a low genomic score.
was “not reached” and 2.8 ms in PTS having SD vs PD (p=0.04), respectively. Conclusions: This study is a proof of concept demonstrating that integration of
Grade $3 adverse events were reported in 8% of PTS. Conclusions: a subgroup molecular variables beyond MGMT methylation improve existing nomograms
of recurrent MMRd HGG might benefit from Pem, especially anaplastic gli- to provide individualized information about patient prognosis. Future di-
omas. There was a trend for a longer PFS and OS in PTS with aMMRd. The rections include a more comprehensive analysis, including proteomic and
enrollment and analyses for identifying additional molecular predictive factors methylation data, and subsequent validation in an external cohort. Finally,
are ongoing. network analysis integrating molecular signatures of poor performers will help
identify therapeutic targets.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
98s Central Nervous System Tumors

2045 Poster Session (Board #234), Sun, 8:00 AM-11:00 AM 2046 Poster Session (Board #235), Sun, 8:00 AM-11:00 AM
Health-related quality of life (HRQoL) evaluation in the REGOMA trial: A Clinical efficacy of tumor-treating fields for newly diagnosed glioblastoma.
randomized, phase II clinical trial analyzing regorafenib activity in relapsed First Author: Yang Liu, University of Rochester, Rochester, NY
glioblastoma patients. First Author: Giuseppe Lombardi, Department of
Background: Recent clinical trials have shown that adding tumor treating
Oncology, Oncology 1, Veneto Institute of Oncology IOV-IRCCS, Padua, Italy
fields (TTF) to the Stupp protocol (SP) has increased survival after glioblastoma
Background: REGOMA trial showed that regorafenib (REG) significantly improved (GBM) diagnosis. However, whether this regimen improves population-based
OS and PFS in relapsed glioblastoma (GBM) patients (pts) with respect to survival for patients with GBM remains unknown. Methods: We retrospectively
lomustine (LOM). REG showed a different toxicity profile compared to LOM. Here, identified adult patients with newly diagnosed GBM treated at our institution
we report final results of the HRQoL assessment, a secondary end point. from January 2000 to July 2017 (n = 438, median age: 63 years).We grouped
Methods: HRQoL was measured using the European Organization for Research patients into three time periods for comparison: 2000-2004 (group 1, prior to
and Treatment of Cancer (EORTC) core questionnaire (QLQ-C30) and brain SP), 2005-2013 (group 2, SP) and 2014-2017 (group 3, adding TTF to SP).
module (QLQ-BN20) administered before any MRI assessments, every 8 weeks The Kaplan-Meier method was used to estimate survival. Statistical analysis
(+/- 2 weeks) until disease progression. To evaluate treatment impact on HRQoL,
included unadjusted group comparisons by Chi-square and Log-rank tests and
questionnaires at progression were excluded. Mixed-effect linear models were
fitted for each of the HRQOL domain to examine the change over progression-free
adjusted group comparisons using logistic and Cox models. Results: Thirty-
time within and between arms. The models included the time of questionnaire seven percent (43/117) of patients with GBM in group 3 received TTF with SP
assessment, the treatment group and their interaction, as fixed effects, and a therapy; when compared to those who received SP only, these patients had
compound symmetry covariance structure for the random effects. Differences of at significant improvements in 6-month and 1-year overall survival (OS) rates
least 10 points were classified as a clinically meaningful change. To correct for (100.0% vs. 82.4%, p , 0.01; 86.0% vs. 66.2%, p , 0.05, respectively)
multiple comparisons and to avoid type I error, the level of significance was set at (unadjusted for prognostic factors including sex, age, KPS and extent of
P = 0.01 (2-sided). Results: Of 119 randomized pts, 117 participated in the HRQoL resection) and an increased trend of median OS (479.0 vs. 448 days, p =
evaluation, and 114 had a baseline assessment (n = 56 REG; n = 58 LOM). No 0.269). However, after adjusting for those prognostic factors, we didn’t find a
statistically significant differences were observed in any generic or cancer specific statistically better survival for patients treated with TTF (OR: 6.156, p =
domain during treatment in the REG and LOM arms, or between the two arms, except 0.097, OR: 2.102, p = 0.185, respectively). Furthermore, multivariate Cox
for the appetite loss scale which was significantly worse in PTS treated with REG proportion hazards model after adjusting for those prognostic factors showed
(Global mean 14.7 (SD = 28.6) vs 7.6 (SD = 16.0); p = 0.0081). The proportion of no significant survival benefits for patients treated with TTF and SP compared
pts with a clinically meaningful worsening for appetite loss was not statistically to those treated with SP only (HR = 0.797, p = 0.648). In addition, we didn’t
different between the two arms (9 out of 24 and 0 out of 13 in the REG and LOM arm, find significant increases of 6-month, 1-year survival rates and median OS
respectively; p = 0.0146). Conclusions: In the REGOMA trial, HRQoL did not change for patients in group 3 when compared to those in group 2 who had seen
during REG treatment. Pts treated with REG and LOM reported no significant dif-
increased trends of survival trends when compared to those in group 1.
ference in HRQoL. Clinical trial information: NCT02926222.
Conclusions: Although adding TTF to SP appeared to benefit patients with
T0 T1 T2 p- value GBM, this effect might be due to selection bias, e.g., TTF was offered to those
Global Health Status 63.0 (21.3) 60.7 (20.2) 54.2 (23.3) 0.2 patients with better prognostic factors. Ascertaining the long-term benefits of
Role Functioning 73.2 (30.3) 71.3 (29.1) 63.1 (34.1) 0.07 TTF requires further investigation.
Cognitive Functioning 78.0 (26.2) 81.3 (25.6) 75 (21.4) 0.8
Emotional Functioning 74.0 (23.3) 72 (24.6) 76.8 (16.4) 0.4
Social Functioning 78.9 (26.5) 76 (25) 76.2 (29) 0.8
Appetite Loss 8.9 (19.6) 18.7 (32) 31 (44.3) 0.002
Motor Dysfunction 17.1 (21.3) 17.8 (24.4) 19.8 (29.3) 0.3
Some HRQoL items during REG treatment

2047 Poster Session (Board #236), Sun, 8:00 AM-11:00 AM 2048 Poster Session (Board #237), Sun, 8:00 AM-11:00 AM
EGFR amplification predicted selective sensitivity to PARP inhibitors with Interim results of a phase I/IIa trial of a therapeutic CMV vaccine against
high PARP-DNA trapping potential in human GBM. First Author: W. K. Alfred recurrent glioblastoma (GBM). First Author: Andrew B. Lassman, Columbia
Yung, The University of Texas MD Anderson Cancer Center, Department of University Irving Medical Center, New York, NY
Neuro-Oncology, Houston, TX
Background: Cytomegalovirus (CMV) antigens have been reported in over
Background: Poly-ADP-ribose polymerase (PARP) is an enzyme critical for 90% of GBM tumors. CD4+and CD8+T cells are most frequently directed
regulating a variety of DNA damage repair mechanisms such as BER/SSBR, against the gB and pp65 antigens, respectively, and are immunogenic
and PARP inhibitors have been shown to have single agent activity in breast targets in a CMV-based GBM vaccine. Methods: We have initiated a phase I/
and ovarian cancer patients with BRCA 1/2 mutations. However, PARP IIa clinical trial for patients with recurrent GBM using gB/pp65 enveloped
inhibitor such as veliparib has limited single agent activity in GBM and virus-like particles (eVLPs) formulated with GM-CSF and administered in-
identifying markers predicting sensitivity is critical to select individuals or tradermally. Subjects are vaccinated monthly until tumor progression, with
certain groups of patients for PARP inhibitor therapy. Methods: Potency and immunomonitoring performed 2 weeks after each vaccination and MRI
selectivity of PARP inhibitors were analyzed in a panel of glioma stem cells exams every 6 weeks. In phase I, eligible patients were age 18-70 with
(GSCs) with varying genetic background. In vivo anti-tumor activity was Karnofsky Performance Status at least 70, normal end-organ function, on
evaluated in xenograft models. Results: In this study, we report that PARP stable or decreasing corticosteroids of at most 4mg dexamethasone (or
inhibitor, talazoparib, showed strong single-agent cytotoxicity and remark- equivalent), with recurrent GBM following any standard initial therapy and
able selective activity in glioma stem cells (GSCs). This single agent activity any number of recurrences. The primary endpoint was safety/tolerability and
was strongly correlated with EGFR amplification. GSCs with EGFR ampli- secondarily to assess immunogenicity. Three vaccine doses (0.4mg, 2mg,
fication (which occurs in about 45% of GBMs) showed higher oxidative base and 10mg pp65) were evaluated with 6 subjects in each cohort and DSMB
damage, DNA breaks, and genomic instability than non-amplified GSCs. To safety review of the first 3 subjects in each cohort prior to enrolling additional
sustain the elevated basal oxidative stress, EGFR-amplified GSCs had in- subjects. Results: The DSMB identified no DLTs or safety concerns with any
creased basal expression of DNA repair proteins. As a result of blocked DNA of the doses. Grade 2, 3 or 4 AEs occurred in 66%, 22% and 11% of par-
damage repair by talazoparib treatment, DNA damage accumulated and lead ticipants, respectively, but were not related to vaccine administration. Twelve
to increased PARP-DNA complexes, which was then trapped by talazoparib men and 6 women were enrolled with a median age 54 (range 39-66). Prior
and resulted in high toxicity. The PARP-DNA trapping function of PARPi is therapies included radiotherapy, temozolomide, and nivolumab. Immuno-
essential as olaparib and veliparib, two PARP inhibitors with weak DNA- logical analyses demonstrate robust boosting of CMV-specific antibody titers
PARP trapping potential did not show sensitivity in GSCs. In contrast, and T cell responses against both gB and pp65 antigens in some but not all
Pamiparib, another PARP inhibitor with similar PARP-DNA trapping ability subjects, across all dose cohorts. Boosting of IFN-gsecreting T cells (measured
to that of talazoparib, showed selective sensitivity in EGFR-amplified GSC. by ELISPOT) exceeded the assay threshold for several subjects. Stable disease
Conclusions: Our data showed that EGFR amplified GSCs with higher basal by MRI of 3 months or greater has been observed in 2 subjects in the high dose
DNA damage exhibited therapeutic vulnerability to PARP inhibitors with high cohort and 1 subject in the low dose cohort and may correlate with vaccine
PARP-DNA trapping ability, and that EGFR amplification is a potential response. Conclusions: The phase IIa extension phase of the trial planned to
selection or predictive biomarker for PARP inhibitor therapy in GBM. begin in Q2 2019 is designed to explore efficacy in an additional 10 subjects that
will receive the optimal vaccine dose and includes the additional requirements of
unifocal, measurable enhancing tumor 1-3 cm across at first recurrence and no
prior immunotherapy. Clinical trial information: NCT03382977.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Central Nervous System Tumors 99s

2049 Poster Session (Board #238), Sun, 8:00 AM-11:00 AM 2050 Poster Session (Board #239), Sun, 8:00 AM-11:00 AM
Evaluating the capacity of connectome analysis to predict survival in high-grade Stratified monotherapy approach according to MGMT methylation status in
astrocytoma. First Author: Rebecca A. Harrison, The University of Texas MD elderly patients with glioblastoma. First Author: Mitsuaki Shirahata, Saitama
Anderson Cancer Center, Houston, TX Medical University International Medical Center, Saitama, Japan
Background: While factors such as age, histology and tumor molecular var- Background: The elderly patients with glioblastoma have an extremely poor
iants (e.g. IDH status) contribute to prognosis in patients with high grade prognosis. As they often have some degree of age-related vulnerability, it is
astrocytoma (HGA), there remains a wide variability in patient survival out- especially important to minimize a risk of treatment-related adverse events
comes. The connectome, or brain network organization, incorporates biologic, by optimizing treatment intensity for this population. We conducted phaseⅡ
molecular and environmental processes providing a uniquely parsimonious clinical trial to investigate the efficacy of stratified monotherapy approach
summary of key prognostic factors. This study compared the capacity of according to O6-methylguanine-DNA methyltransferase (MGMT) methyla-
machine learning (ML) models based on baseline connectomics and clinical tion status in elderly patients with glioblastoma. Methods: Patients aged 70
variables to predict patient survival in HGA. Methods: Patients with a new years or older with Karnofsky performance status (KPS) of at least 60 were
diagnosis of HGA and a presurgical 3D, T1-weighted MRI available were eligible for this study. MGMT methylation status was quantitatively assessed
retrospectively identified. Individual patient connectomes were derived from by pyrosequencing based on the average methylation ratio of 16 CpG sites in
MRI with 90 cortical/subcortical features. Presurgical clinical features in- the MGMT gene promoter. The patients with highly methylated MGMT
cluded age, gender, histology, tumor grade and IDH status. Three ML algo- promoter defined as an average methylation ratio with 30% or higher were
rithms were implemented: extreme learning machine with Buckley–James treated with temozolomide (TMZ) monotherapy (standard 5/28 regimen),
estimator (ELMBJ), random survival forest (RSF) with logrank splitting and while the others with low or intermediate levels of MGMT promoter methylation
RSF with concordance index (CI) splitting. For each algorithm, we used a 60/ were treated with radiation therapy (40Gy/15fr) alone. Results: Between April
40 training/testing split with 50 iterations and CI as the performance metric. 2013 and December 2017, 70 patients were enrolled in this study. Median
We tested three models: 1) connectome only, 2) clinical only, and 3) con- age was 78 years (70-91) and median KPS was 60 (60-100). Of 70 patients,
nectome plus clinical variables. Results: Of patients identified (n = 105), 66 19 patients with highly methylated MGMT promoter received TMZ mono-
had glioblastoma and 39 had anaplastic astrocytoma. Thirty-eight harbored therapy, while the remaining 51 patients were treated with radiation therapy.
IDH mutation. Median overall survival was 27.43 months (SD 39.57). Median progression-free survival (PFS) and median overall survival (OS) were
Connectome-only models showed better prediction performance compared to 7.5 and 17.4 months in the TMZ group, respectively. Median PFS and median
clinical-only models across all algorithms. ELMBJ showed the best perfor- OS were 4.6 and 10.4 months in the radiotherapy group, respectively.
mance (connectome median CI = 0.522, clinical CI = 0.201). Connectome Conclusions: For elderly glioblastoma patients with highly methylated MGMT
models also performed as well as combined models (e.g. median CI = 0.523 promoter, TMZ monotherapy could be a treatment option. Clinical trial in-
for ELMBJ). Conclusions: This study demonstrates the potential of a con- formation: UMIN000012172.
nectome model to predict survival of patients with HGA. Replication in a larger
sample is required to validate these results and refine ML models including
examination of additional clinical features. If successful, use of a simple T1
MRI could provide additional variables to augment existing prognostic pre-
diction, especially in scenarios where tumor genotyping is not available.

2051 Poster Session (Board #240), Sun, 8:00 AM-11:00 AM 2052 Poster Session (Board #241), Sun, 8:00 AM-11:00 AM
The timing of chemoradiotherapy after surgical resection and its impact on Correlation of systemic and local inflammation with survival prognosis in
overall survival in glioblastoma. First Author: Robert H. Press, Department of glioma patients. First Author: Pegah Mir Seyed Nazari, Clinical Division of
Radiation Oncology, Winship Cancer Institute of Emory University, Atlanta, Haematology and Haemostaseology, Department of Medicine I, Comprehensive
GA Cancer Center, Medical University of Vienna, Vienna, Austria
Background: Prior studies examining time to initiate chemoradiotherapy Background: Immune modulating therapies have been a long withstanding
(CRT) after surgical resection (S) in glioblastoma (GBM) have not provided treatment approach in glioma. However, gliomas are characterized by a
clear consensus on its clinical impact. We sought to evaluate the effect that particular absence of tumor infiltrating lymphocytes in the local tumor mi-
differential timing of adjuvant therapy may have on overall survival (OS). croenvironment. We aimed to gain insight on the distinct patterns of in-
Methods: With the National Cancer Database (NCDB), patients (pts) with flammation associated with survival prognosis in glioma. Methods: Patients
GBM who underwent S and adjuvant CRT from 2004-2013 were analyzed. were recruited at time of glioma diagnosis or progression in the prospective
Analysis was performed for the entire cohort as well as by Radiation Therapy observational Vienna Cancer and Thrombosis Study (CATS). A single blood
Oncology Group (RTOG) recursive partitioning analysis (RPA) classes (i.e. I, draw was performed at study inclusion. PD-L1 expression in the tumor tissue
II, and III). Time from S to CRT was grouped weekly (i.e. 0-1, 1-2, 2-3, 3-4, was investigated via immunohistochemistry. Optimal cut-off according to ROC
4-5, 5-6, 6-7, 7-8, and . 8 weeks). Pts were excluded if they died within the curve was used to assess cut off values for survival analysis. Results: 193
first 8 weeks to account for immortal time bias. Kaplan-Meier analysis, log- patients with glioma (75.6% glioblastoma (WHO grade IV), 19.7% anaplastic
rank testing, and multivariate (MVA) Cox proportional hazards regression glioma (WHO grade III), and 4.7% diffuse glioma (WHO grade II)) were in-
were performed with OS as the primary outcome. Results: A total of 30,414 cluded. 40/193 (20.7%) glioma had an IDH1 mutation. Membranous PDL1
pts were included for analysis. RPA class I, II, and III contained 903, 4,347, expression in the tumor tissue was observed in 20/193 (10.4%) patients. 1/20
and 25,164 pts, respectively. The most common time to initiate CRT was patient presented with PD-L1 expression and IDH1 mutation (p = 0.082). PD-
week 4-5 (n = 7389), and this group served as reference for survival analysis. L1 significantly correlated with increased monocyte count (median: 0.657 vs.
On MVA, weeks 0-1 (hazard ratio [HR] 1.18, 95% confidence interval [CI] 0.450 [G/L], p = 0.008), higher C-reactive protein (CRP) (0.43 vs. 0.1 [mg/
1.02-1.35), 1-2 (HR 1.24, CI 1.17-1.32), and 2-3 (HR 1.11, CI 1.07-1.15) dL], p = 0.005) and higher fibrinogen (379 vs. 303 [mg/dL], p = 0.001).
demonstrated worse OS (all p , 0.03). For RPA class I pts, week 1-2 (HR Presence of IDH1 mutation significantly correlated with increased platelet
2.07, CI 1.08-3.95) was associated with worse OS (p = 0.028). For RPA count (303 vs. 232 [G/L], p = 0.001) and lower Neutrophil/Lymphocyte (N/L)
class II pts, weeks 1-2 (HR 1.34, CI 1.14-1.57), 2-3 (HR 1.18, CI 1.07- ratio (3.34 vs. 5.13, p = 0.016). Higher lymphocyte count ( . 1.484 [G/L],
1.31), and 3-4 (HR 1.10, CI 1.0-1.21) were associated with worse OS (all log-rank: p = 0.011), higher platelet count ( . 245.5 [G/L], p = 0.0001), as
p , 0.05). For RPA class III pts, weeks 0-1 (HR 1.18, CI 1.02-1.38), 1-2 well as decreased N/L ratio ( , 5.13, p = 0.001) were significantly associated
(HR 1.22, CI 1.14-1.3), and 2-3 (HR 1.09, CI 1.05-1.14) were associated with increased survival prognosis. Conclusions: PD-L1 expression in tumor
with worse OS (all p , 0.03). No time point after week 5 was associated with tissue was associated with markers of systemic inflammation in glioma pa-
change in OS for the overall cohort or any RPA class subgroup. tients. Systemic inflammation markers furthermore predicted improved sur-
Conclusions: These data provide insight into the optimal timing of CRT in vival. Immune modulating therapy approaches might be a promising approach
GBM and describe RPA-class specific outcomes. In general, OS was neg- in subgroups of glioma associated with increased baseline interaction of
atively impacted if CRT started less than 3 weeks from S. Waiting up to immune system and glioma.
8 weeks, however, was not detrimental to OS and suggests delaying CRT
beyond week 4-5 should be considered if clinically indicated without undue
concern.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
100s Central Nervous System Tumors

2053 Poster Session (Board #242), Sun, 8:00 AM-11:00 AM 2054 Poster Session (Board #243), Sun, 8:00 AM-11:00 AM
Evaluation of controlled IL-12 as monotherapy in subjects with recurrent Decision making in surveillance of high-grade gliomas using perfusion MRI
GBM. First Author: Rimas Vincas Lukas, Northwestern University, Chicago, as adjunct to conventional MRI and artificial intelligence. First Author:
IL Sotirios Bisdas, University College London, London, United Kingdom
Background: Interleukin-12 (IL-12), a master regulator of the immune Background: Surveillance of High-Grade Gliomas (HGGs) remains a major
system, results in anti-tumor responses in preclinical models, but safe use challenge in clinical neurooncology. Histopathological validation is not an
requires tightly controlled production. It was conditionally produced in Ph1 option during the course of disease and imaging surveillance suffers from
“main” study (NCT02026271) in subjects with recurrent glioblastoma ambiguous features of both disease progression and treatment related
(rGBM) using a replication-incompetent adenovirus modified to express IL- changes. This study aimed to differentiate between Pseudoprogression (PsP)
12 under transcriptional control of the proprietary RheoSwitch Therapeutic and Progressive Disease (PD) using an artificial intelligence (support vector
System (Ad-RTS-hIL-12, Ad) regulated by dose of veledimex (V). Mono- machine - SVM) classification algorithm. Methods: Two groups of patients
therapy resulted in sustained intra-tumor influx of activated cytotoxic T cells, with histologically proven HGGs were analysed, a group with a single time
consistent with immune-mediated anti-tumor effect, improving overall point DSC perfusion MRI (45 patients) and a group with multiple time point
survival (OS). This correlated with increased circulating CD8+/FoxP3+ T-cell DSC perfusion MRI (19 patients). Both groups included conventional MRI
ratio (“cytoindex”), an emerging biomarker of OS. While widely used with studies prior and after each perfusion MRI. This study design aimed to
neurosurgery, dexamethasone (dex) blunts response to immunotherapies, replicate decision making in clinical practice including multiple previous
nevertheless median mOS of subjects who received 20mg V of 12.7 mo studies for each patient. SVM training was performed with all available MRI
(n=15) at 13.1 mo follow-up. However, subanalysis (n=6) showed low-dose studies for each group and classification was based on different feature
dex (total #20 mg) during V dosing improved mOS (17.8 mo). We report a 36 datasets from a single or multiple (subtracted features) time points. Clas-
subject substudy in rGBM with limited dex, total rGBM treated (n=70+). sification accuracy comparisons were performed by calculating prediction
Methods: Ongoing Phase 1 substudy (NCT03679754) assesses safety and error rates for different feature datasets and different time point analyses.
tolerability of local, inducible IL-12 by single intratumoral injection of Ad (2 x Results: Our results indicate that the addition of multiple time point perfusion
1011 viral particles) + V (20 mg PO QD x15 doses Days 0-14) in subjects not MRI combined with structural (conventional with gadolinium-enhanced se-
receiving dex 4 wks prior to Ad. Results: As of 03Jan19, the majority of new quences) MRI features results in optimal classification performance (median
subjects received low-dose dex (total #20mg Days 0-14). The initial impact error rate: 0.016, lowest value dispersion). Subtracted feature datasets im-
of dex on mOS will be reported. As in the main study, Ad+V 20 mg re- proved classification performance, more prominently when the final and first
spectively increased (median) serum IL-12 and downstream IFN-g from Days perfusion studies were included in the analysis. On the contrary, in the single
0-3: 0.8 to 8.8 pg/mL and 0 to 8.6 pg/mL. Between Days 0-14, there was net time point group analysis, structural feature-based classification performed
increase in cytoindex (from 20 to 46). The safety profile was similar to the best (median error rate: 0.012). Conclusions: Validation of our results with a
main study with the main adverse reaction (AR) being mild to moderate larger patient cohort may have significant clinical importance in optimising
cytokine release syndrome (CRS) characterized by flu-like symptoms. No imaging surveillance and clinical decision making for patients with HGG.
grade 4 CRS was noted; all ARs were manageable and reversable upon
holding V. Conclusions: Local, controlled IL-12 production using the Ad + V
platform in subjects with rGBM safely activates the immune system and
when dex is limited, appears to further improve mOS, which warrants
continued investigation. Clinical trial information: NCT03679754.

2055 Poster Session (Board #244), Sun, 8:00 AM-11:00 AM 2056 Poster Session (Board #245), Sun, 8:00 AM-11:00 AM
Analysis of the EF-14 phase III trial reveals that tumor treating fields alter Molecular genetic, host-derived and clinical determinants of long-term
progression patterns in glioblastoma. First Author: Suriya A. Jeyapalan, survival in glioblastoma: First results from the ETERNITY study (EORTC
Rhode Island Hospital, Brown University, Newton, MA 1419). First Author: Michael Weller, Laboratory of Molecular Neuro-
Oncology, Department of Neurology, and Neuroscience Center Zurich,
Background: The EF-14 [NCT00916409] trial showed that addition of al-
University Hospital and University of Zurich, Zurich, Switzerland
ternating electric fields (Tumor Treating Fields, TTFields) to Temozolomide
(TMZ) resulted in improved survival in newly diagnosed Glioblastoma (GBM) Background: Glioblastoma represents the most aggressive primary brain tu-
patients with supratentorial tumors treated compared to TMZ alone. TTFields mor in adults, and less than 5% of patients survive 5 years from diagnosis.
delivery is planned to optimize dose at the tumor bed, leading to the hy- Factors influencing this long-term survival are poorly understood. Methods: In
pothesis that TTFields treated patients are more likely to exhibit distal pro- cooperation with the European Organisation for Research and Treatment of
gressions, including progression to the infratentorial brain where TTFields Cancer (EORTC) in Brussels, Belgium, more than 20 clinical sites in the US,
dose is minimal when targeting the supratentorium. Here we present analysis Europe and Australia have registered patients with centrally confirmed glio-
of the EF-14 trial testing this hypothesis. Methods: Patients on treatment for blastoma who survived $ 5 years, collecting clinical data including therapy
more than two months who had an MRI that exhibited progression were in- and quality of life-related factors, as well as biospecimens allowing to analyse
cluded in the study (treatment: N=280/466, control: N=122/229). Regions of molecular and immunological parameters. Results: At the cut-off of December
enhancing tumor, necrosis and resection were contoured on T1 contrast MRIs 31, 2018, 392 patients were registered, of which 232 had glioblastoma
acquired at baseline and at the date of first progression. New lesions at confirmed by central pathology review; 59 dropped out due to histology other
progression were classified as distal if they appeared outside of a Proximal than glioblastoma. Glioblastomas were isocitrate dehydrogenase (IDH)-wildtype
Boundary Zone (PBZ) of 20 mm surrounding the lesions identified in the in 70.7% and had a positive O6-methylguanine DNA methyltransferase (MGMT)
baseline MRI. The rate of occurrence of distal progressions in the TTFields- promotor methylation status in 75.9%. Median age at diagnosis was 52 years
treated arm was compared to the rate observed in the control arm. Patients (range: 21-77 years). There was enrichment for patients with gross total re-
with (distal) infratentorial progression were identified. Results: Distal pro- section. Further analyses are ongoing. Conclusions: In a comprehensive effort,
gressions were more common in the treatment arm (49/280 (18%) vs. 10/122 the consortium funded by the US Brain Tumor Funders’ Collaborative char-
(8%) P,0.02; chi-squared). Infratentorial progression were observed in 4% acterizes factors modulating long-term survival in glioblastoma in a unique large
(10 patients) of the treatment arm vs. 0 patients in the control (P,0.002 t- patient cohort. Clinical trial information: NCT 03770468.
test). Distal lesions at progression were more distant from the original lesion in
the TTFields treated arm (58.57 + 28.12 mm vs 46.61 + 20.48 mm,
P,0.02; Wilcoxon rank sum test. The relative tumor growth rates in TTFields
treated patients were significantly slower than those observed in the control
arm (0.036+ 0.126 ml/day vs. 0.036+ 0.183 ml/day P,0.03; t-test).
Conclusions: This analysis indicates that adding TTFields to TMZ could im-
pact GBM growth patterns. The results suggest that TTFields increases local
control of tumor growth, emphasizing the need for adaptive treatment after
progression to control progressing disease. Clinical trial information:
NCT00916409.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Central Nervous System Tumors 101s

2057 Poster Session (Board #246), Sun, 8:00 AM-11:00 AM 2058 Poster Session (Board #247), Sun, 8:00 AM-11:00 AM
Efficacy of re-irradiation with carbon ions (RiCi) in patients with recurrent Detection of targetable somatic alterations in glioblastoma (GBM) and
high-grade glioma (rHGG) compared to the standard re-irradiation with clinical impact. First Author: Michael Fusco, Moffitt Cancer Center,
photons (RiP): The reference multicenter cohort of the German Cancer Tampa, FL
Consortium Radiation Oncology Group (DKTK-ROG). First Author: Maximilian
Background: In GBM, molecular markers are utilized to establish an in-
Knoll, Departments of Radiation Oncology, Neurology, Neurosurgery, Heidelberg
tegrated diagnosis as described in the WHO 2016 guidelines and identify
University Hospital, National Center for Tumor Disease (NCT), UKHD and
patients (pts) with molecular targets amenable to therapeutic intervention.
German Cancer Research Center (DKFZ), German Cancer Consortium (DKTK),
Herein we review our experience at Moffitt Cancer Center. Methods: A ret-
Core-Center Heidelberg, Heidelberg, Germany
rospective chart review between 4/1/2013 and 11/1/2018 was performed to
Background: Local recurrence after surgery and radio(chemo)therapy collect demographic, clinical, disease, treatment and outcome variables on
remains a major obstacle in curative treatment of patients with HGG. 163 unique pts with GBM whose tumors underwent comprehensive genomic
Eradication of radioresistant glioma subpopulations (hypoxic- and stem cell profiling by FoundationOne or CDx testing. Genomic data was analyzed for
like cells) together with formation of an antiangiogenic and immuno- recurrent alterations and tumor mutational burden (TMB). Results: Median
permissive glioma niche are among beneficial radiobiological effects re- age was 58 years (range 19 to 85). 13% were IDH1 or 2-mutated. Among the
cently attributed to carbon ion irradiation in preclinical models. The impact 141 IDH-wild type (wt) pts, TERT promoter mutations occurred in 83% and
of this novel therapy in management of rHGG patients remains elusive. CDKN2A/B co-deletion in 65%. O6-methylguanine-DNA methyltransferase
Methods: 197 patients with rHGG (grade III: 71, IV: 126) received RiCi (MGMT) promoter methylation was seen in 33%. A median of 5 clinically
between Nov 2009 and Feb 2018 at HIT with a median dose of 42GyRBE in relevant alterations were identified per tumor sample (range, 2 to 34) and a
14 fractions. In DKTK-ROG multicenter cohort n:565 rHGG patients (grade median of 2 mutations (range, 0 to 6) were found to be actionable after review
III: 63, IV: 479) underwent RiP between 1997-2016 with a median dose of by our molecular tumor board. The most commonly actionable alterations were
36 Gy in 14 fractions. Median follow up was 34.2 months (M) for RiCi and found in EGFR, BRAF and genes associated with homologous recombination
7.1 M for RiP (DKTK) cohort. All three prognostic scores validated in DKTK- deficiency (HRD) (see table). Four pts were treated with EGFR-targeted
ROG cohort were evaluated and re-irradiation risk score (RRRS) considering therapy, one pt with an HRD alteration received a PARP inhibitor (progres-
initial grade, Karnofsky Performance Score and age at re-irradiation was sion free survival [PFS] of 34 weeks), and two pts with BRAF V600E received
utilized for stratification and matched comparisons. Results: Median PFS dabrafenib/trametinib combination therapy (treatment ongoing at 14 weeks
was 5.08 [4.26-5.87] M (grade III: 6.79, grade IV: 3.64) after RiCi, data was and 38 months, respectively). Median TMB (n = 118) was 4 (range, 0 to 371
not available for RiP. Median OS was 10.52 [9.28-12.66] M (grade III: 28.2, Muts/Mb). One pt who received 44 months of temozolomide exposure had a
grade IV: 8.53) after RiCi compared to 7.93 [7.15-8.79] M (grade III: 10.89, hypermutated tumor (371 Muts/Mb) and was treated on trial with pem-
grade IV: 7.93) after RiP. Among the three prognostic scores evaluated, brolizumab, but progressed after 2 months. Conclusions: Though limited in
RRRS most robustly correlated with OS. RiCi was associated with HR of 0.52 patients with GBM, clinically actionable alterations are found in a small subset
([0.49-0.72], p = 0.0000002), and HR 0.66 ([0.51-0.85], p = 0.001) for and can translate into meaningful clinical benefit.
RRRS matched analysis. Conclusions: Carbon ions demonstrated activity in IDH-Mut IDH-Wt
rHGG. This effect is most prominent in grade III while grade IV patients may Alterations n (%) n (%)
further benefit from innovative multimodal strategies. Based on these en- H3F3A (K27M, G35R) 0 (0%) 5 (4%)
couraging results prospective randomized trials utilizing RRRS for stratifi- HRD 6 (27%) 15 (11%)
Amp of chromosome 4q (e.g. Genes - KDR, KIT, PDGFRA) 2 (9%) 12 (9%)
cation are recommended. EGFR amplification (Median copy #: 57, range 11 to 166) 1 (5%) 61 (43%)
EGFR vIII and other EGFR 1 (5%) 52 (37%)
BRAF 0 (0%) 5 (4%)
FYN-ROS1 rearrangement 0 (0%) 1 ( < 1%)
NTRK1 (duplication exons 10-17) 0 (0%) 1 ( < 1%)

2059 Poster Session (Board #248), Sun, 8:00 AM-11:00 AM 2060 Poster Session (Board #249), Sun, 8:00 AM-11:00 AM
Carbon ion reirradiaton for patients with malignant gliomas: Toxicity and first Oncolytic polio/rhinovirus recombinant (PVSRIPO) against WHO grade IV
results of the prospective dose-escalation phase I/II CINDERELLA trial. First malignant glioma (MG): Experience with retreatment of survivors from the
Author: Stephanie E Combs, German Cancer Consortium (DKTK) Core Center phase I trial. First Author: Annick Desjardins, Duke University Medical
Heidelberg and DKTK Partner Site Munich (TUM), Munich, Germany Center, Durham, NC
Background: The prospective phase I/II CINDERELLA trial investigates Background: We completed a study evaluating a single intratumoral delivery
toxicity and effectiveness of a dose escalated reirradiation with carbon ions of PVSRIPO in recurrent WHO grade IV MG patients (N Engl J Med. 2018 Jul
in patients with recurrent gliomas. Methods: Following a dose escalating 12;379(2):150-161). Some patients who originally benefitted from the
protocol, 52 patients with WHO°II-IV gliomas were irradiated with carbon infusion of PVSRIPO demonstrated tumor recurrence, and we hypothesized
ions with doses of 3 Gy (RBE) in 10 – 16 fractions in 7 dose leels. Median age that retreatment could trigger an immune recall effect, further extending
was 42 years (range: 28 - 69) with 19 female and 33 male participants. their survival. We now report the impact of second and third intratumoral
Forty-one patients were diagnosed with WHO°III/IV gliomas and 11 patients reinfusion of PVSRIPO in patients treated in the original dose finding study.
with WHO°II gliomas. At the time of reirradiation, all patients showed Methods: Eligible patients were adults with recurrent supratentorial WHO
contrast-enhancing recurrences. MRI-based treatment planning encom- grade IV MG who were experiencing disease recurrence after having
passed the contrast enhancing lesion (GTV) with additional safety margins of benefitted from the first infusion of PVSRIPO. Additional eligibility criteria
5 mm (CTV) and 3 mm (PTV). Clinical follow-up visits including contrast- included: solitary tumor 1-5.5cm in diameter; $4 weeks after chemo-
enhanced MRI were scheduled every two months. We used RANO-criteria for therapy, bevacizumab or study drug; adequate organ function; KPS$70%;
diagnosis of progression. Survival rates were analyzed with Kaplan-Meier and positive anti-polio titer. One patient each was retreated at 1 x 107
estimator. Relevant prognostic factors were determined with log rank-test, TCID50 and 1 x 1010 TCID50, and three patients were retreated on the
and toxicity was classified according to CTCAE v4.0. Results: Median time identified phase 2 dose of 5 x 107 TCID50. Results: As of 2/09/2019, five
between first irradiation and reirradiation was 9 months (range: 7 – 228). patients have received a second intratumoral dose of PVSRIPO on study, one
PTV size was 12 – 310 ml. During follow-up $°3 toxicities were not observed. of which received a total of 3 doses. The patients who received two infusions
Follow-up MRI suggested radiation necrosis in 4 patients. Median overall of PVSRIPO were retreated 72 months, 43 months, 34 months, and
survival was 352 days and was not influenced by age, gender or radiation dose. 6 months after the first infusion. One additional patient received a second
A significant trend for improved survival rates was seen in patients with small infusion of PVSRIPO 60 months after the first infusion and a third infusion of
target volumes (480 days [PTV , 75ml] vs. 322 days [PTV . 75ml], p = 0.06) PVSRIPO 78 months after the first infusion. All patients demonstrated soap
and initial low grade histology (497 days [WHO °II] vs. 322 days [WHO°III/IV], bubble degeneration on imaging, and two patients demonstrated tumor
p = 0.069). During follow-up, 45 patients had local progression, while clinical contraction. No grade 3 or higher adverse events related to PVSRIPO were
deterioration was not seen. Median local progression-free survival was 138 days. observed after retreatment. Three of these patients remain alive more than
Twenty-eight patients received chemo-/immunotherapy after reirradiation. Of 81, 80 and 52 months following the first PVSRIPO infusion and more than 9,
those, 14 patients were treated with bevacizumab. Progression after reirra- 20 and 18 months after the second infusion, respectively. Two patients died
diation did not influence overall survival significantly. Conclusions: Carbon ion 63 months and 20 months after the first infusion of PVSRIPO and 19.6 and
re-irradiation with 10-16 fractions of 3 Gy for patients with recurrent gliomas is 14 months after the second, respectively. The patient treated 3 times re-
safe; no dose limiting toxicities were observed. Median overall survival with ceived the third infusion more than 2 months ago. Conclusions: Intratumoral
approximately one year is high in comparison to other treatment options. It reinfusion of PVSRIPO via CED is safe, and encouraging efficacy results have
remains unclear if RANO-criteria is securing the diagnosis of therapy failure been observed. Clinical trial information: NCT01491893.
after carbon ion reirradiation. Further randomized controlled trials must be
awaited to evaluate the effectiveness of reirradiation of carbon ions compared to
other treatment options. Clinical trial information: NCT01166308.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
102s Central Nervous System Tumors

2061 Poster Session (Board #250), Sun, 8:00 AM-11:00 AM 2062 Poster Session (Board #251), Sun, 8:00 AM-11:00 AM
Are patients with oligodendroglioma at higher risk for radiation neurotoxicity? A phase I study of MGMT-P140K transfected hematopoetic progenitor cells
First Author: Haroon Ahmad, University of Virginia, Charlottesville, VA (HPC) combined with TMZ/O6BG dose escalation for newly diagnosed,
MGMT unmethylated glioblastoma: Tolerance and evidence of survival
Background: Symptomatic radiation neurotoxicity (RN), manifesting on MRI
benefit. First Author: Andrew E. Sloan, University Hospital Case Medical
as focal necrosis and/or T2 signal abnormality, is a dreaded complication of
Center, Cleveland, OH
radiation therapy (RT). While RT is standard of care for anaplastic gliomas,
the long-term benefit vs risk profile in low-grade gliomas is not well defined. Background: GBM is the most common malignant brain tumor with a median
Patients with oligodendroglioma carry a better overall survival than those survival of 15 months despite surgery and radio-chemotherapy. The most
with astrocytoma. Anecdotally, they are more prone to experience RN than important mechanism of TMZ resistance is the O6-methylguanine-DNA meth-
astrocytomas, as suggested by Acharya et al in 2017. We hypothesized that, yltransferase (MGMT) gene which repairs temozolamide-induced DNA methyl-
independent of grade, oligodendrogliomas have a higher incidence of RN as ation. The MGMT inhibitor O6-benzylguanine (BG) demonstrated efficacy in
compared to astrocytomas. Methods: We reviewed the records of 628 pa- depleting MGMT and maximizing tumor response in early phase clinical trials.
tients with WHO grade II and III gliomas from our institution. Study pop- However, MGMT expression is also low in hematopoietic cells, so this approach
ulation comprised 326 patients with: standard fractionated RT, pathology led to unacceptable bone marrow toxicity and thus has been abandoned. We
confirmation by a neuropathologist, and follow up of at least 2 years after hypothesized that chemoprotection of hematopoietic HPC with an MGMT mutant
diagnosis. RN was defined as either histologically confirmed by pathology or (MGMT-P140K) characterized by normal methyltransferase activity, coupled
requiring intervention for clinically presumed RN (bevacizumab or high-dose with low affinity for BG would maximize anti-tumor response while enabling
steroids.) A separate category included patients with dramatic cognitive patients to tolerate TMZ & BG dose escalation with minimal toxicity. A phase I
trial was performed to test this hypothesis. Methods: 10 adults with newly di-
decline with increased T2 signal abnormality, in the absence or tumor
agnosed MGMT unmethylated, IDH-1 WT, GBM underwent standard surgery and
progression. Results: There were 131 patients with oligodendroglioma,
radiation, followed by transplantation with autologous CD34+ HPC engineered to
based upon 1p/19q co-deletion (105 cases) or histology in the absence of
express MGMT-P140K using a lentiviral vector. We tested tolerance and efficacy
molecular testing (26 cases). The remaining 195 patients had astrocytoma of three different paradigms for conditioning bone marrow and re-infusion of
with intact 1p/19q, isocitrate dehydrogenase (IDH) wild-type, or diagnosed HPC. To assess chemo-protection, patients’ blood counts and transgene marking
histologically absent molecular testing. The incidence of RN were 18.3% were monitored during and after treatment, as was toxicity, response, and
and 8.2% for oligodendroglioma and astrocytoma, respectively (p = progression-free and overall survival. Results: Treatment was moderately toxic
0.0063). An additional four patients with oligodendroglioma and two with with 3/10 patients suffering grade 3-4 hematologic toxicity; no high grade non-
astrocytoma had significant cognitive deterioration with increased T2 signal hematologic toxicity was observed . Viral transduction rates ranged from 3-75%
abnormality, without tumor progression. Conclusions: The greater than two- and were clearly improved in Arm III utilizing BCNU conditioning and intra-patient
fold increase in RN incidence for oligodendrogliomas is significant and dose escalation of TMZ/O6GB. In patients tolerating 3 cycles or more, P140K-
suggests patients with oligodendrogliomas may be more at risk to develop MGMT gene markings in peripheral blood and bone marrow cells increased 3-26-
RN. Therefore, in patients with oligodendroglioma, the consideration of fold with only mild (Grade 2-3) mylosuppression consistent with chemo-protection
fractionated RT needs to be weighed against the increased potential for RN. as hypothesized. Median PFS and OS was 22 and 31 months respectfully, and
Analysis of baseline imaging and patient characteristics variables that three patients in Arm III are healthy and progression free at 36-39 months. OS
correlate with development of RN are ongoing and will be presented at the exceeded RPA predicted survival by 3.3-fold suggesting clinical benefit. Viral in-
meeting. sertion site analysis demonstrate lack of clonal dominance. Conclusions: P140K-
MGMT transfected HPC enables TMZ/ BG dose escalation with acceptable toxicity
and increased survival in a small cohort of selected patients. A phase II study is
ongoing. Clinical trial information: NCT01269424.

2063 Poster Session (Board #252), Sun, 8:00 AM-11:00 AM 2064 Poster Session (Board #253), Sun, 8:00 AM-11:00 AM
Retrospective review for outcomes of IDH mutant, 1p/19q co-deleted Prospective phase II trial in patients with first relapse of high-grade astro-
gliomas based on initial treatment. First Author: Christopher Ray Trevino, cytoma using TVB-2640 in combination with bevacizumab versus bevacizumab
MD Anderson Cancer Center, Houston, TX alone. First Author: Brandon Konkel, Cancer Therapy and Research Center at
UT Health Science Center, San Antonio, TX
Background: Treatment for oligodendroglioma in the molecular era is based
on trials that used histologic criteria alone and has evolved over time to Background: Recurrent glioblastoma (rGBM) following chemoradiation is
include observation and a combination of radiation (RT) and chemotherapy. associated with a poor prognosis. While bevacizumab is the most common
In particular, treatment for high-risk patients is variable including sequential salvage therapy, responses remain brief and without an associated survival
RT followed by temozolomide (TMZ) or procarbazine, lomustine, and vin- benefit. Resistance may involve overexpression of Fatty Acid Synthase
cristine (PCV) or chemoradiation with TMZ. This is a single institutional (FASN). Our institution is conducting a phase 2 study of bevacizumab
review of outcomes of molecular IDH mutant, 1p19q co-deleted oligo- with FASN inhibitor TVB-2640 in patients with GBM in first relapse.
dendroglioma based on initial treatment. Methods: Using both Palantir Methods: This is a prospective, phase 2 study of bevacizumab with TVB-
Foundry and direct record review, we stratified adult patients with grade II/III 2640 in patients with GBM in first relapse. Primary end point is progression
gliomas diagnosed from 1997-2017 harboring an IDH mutation and 1p/19q free survival (PFS). Inclusion criteria are: age $ 18, ECOG 0 to 2, GBM
co-deletion based on initial treatment and excluded patients without mo- progression following standard combined modality treatment. Randomiza-
lecular testing. Overall survival (OS) and progression free survival (PFS) were tion into two arms for the first 28 days is included for exploratory biochemical
calculated by Kaplan-Meier analysis. Results: 187 patients were initially analysis: patients in arm 1 receive bevacizumab every 2 weeks in combi-
managed as: 55 (31%) observation, 16 (9%) RT alone, 51 (29%) che- nation with TVB-2640; those in arm 2 receive bevacizumab alone every
motherapy alone, 39 (22%) sequential RT and chemotherapy, and 17 (10%) 2 weeks. MR-Spectroscopy (MRS) and serum sampling for exosome analysis
chemoradiation. For all patients, a subtotal (STR) or gross total resection are obtained on patients at day 1 and 28 of first cycle. Starting on cycle 2 day
(GTR) was associated with prolonged OS vs biopsy (NR vs 123 mos, p = 1, all patients converge to a single arm and continue to receive bevacizumab
0.007). Age , 40 did not correlate with OS (p = 0.87). In the chemotherapy in combination with TVB-2640. Results: We have enrolled 24 patients to
group, although PCV demonstrated improved PFS vs TMZ (157 vs 45 mos, date; 23 have started treatment. Of those 23 patients, 10 have died, 4 have
p = 0.004) there was no difference in OS, only 3 patients received PCV. For progressed but are still alive, 2 withdrew, and 7 are still active on trial. The
patients treated sequentially with RT and PCV or TMZ vs chemoradiation, PFS6 is and OS9 are both currently 50%, which compares favorably with
there were no differences in OS (NR vs 140 mos; p = 0.87) or PFS (49 vs 45 historical controls. There have been no reports of grade 4 or 5 treatment-
mos, p = 0.52), though sequential treatment trended toward prolonged related AEs (of note, 2 deaths were thought definitely unrelated to treatment,
survival. Conclusions: This data supports both STR and GTR as prognostic including 1 case of intracerebral hemorrhage, and 1 case of sepsis). There
for overall survival. Despite the prolonged PFS observed with patients treated have been two cases of grade 3 hand-foot syndrome thought definitely related
with PCV alone which did not translate to improved OS, the small PCV sample to treatment. Updated results will include PFS, response, and biomarker
size limits the strength of this evidence. We did not find significant survival analysis (exosome, MRS). Conclusions: The combination of TVB2640 with
differences between sequential RT and chemotherapy vs concurrent, but bevacizumab appears be well tolerated. PFS6 and OS9 are both currently
interestingly, there was a trend toward improved OS in sequentially treated 50%. The study has completed accrual with final data expected later in 2019.
patients. Clinical trial information: NCT03032484.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Central Nervous System Tumors 103s

2065 Poster Session (Board #254), Sun, 8:00 AM-11:00 AM 2066 Poster Session (Board #255), Sun, 8:00 AM-11:00 AM
Phase I/II study of depatuxizumab mafodotin (ABT-414) monotherapy or Functional profile and clinical significance of major tumor infiltrating lymphocyte
combination with temozolomide in Japanese patients with/without EGFR- subsets in lung cancer-associated brain metastases. First Author: Benjamin Y.
amplified recurrent glioblastoma. First Author: Yoshitaka Narita, National Lu, Yale School of Medicine, New Haven, CT
Cancer Center Hospital, Chuo-ku, Tokyo, Japan
Background: Despite the biological and clinical implications, the immune
Background: The poor prognosis of glioblastoma (GBM; WHO grade IV) composition and functional characteristics of adaptive immune cells in brain
results from a high rate of disease recurrence and lack of effective treatment metastases (BrM) are poorly understood. Using multiplexed quantitative
options. Depatuxizumab mafodotin (depatux-m, ABT-414) is comprised of immunofluorescence (QIF), this study evaluates the level and functional
an EGFR-directed antibody, depatuxizumab (depatux, ABT-806), conju- profile of major T-cell subsets in primary lung tumors, BrM, and extracranial
gated to the microtubule toxin monomethyl auristatin F (MMAF, mafodotin). metastases (ECM) from lung cancers. Methods: A tissue microarray was
Once bounded with tumor cells, depatux-m is internalized and releases the constructed from formalin-fixed, paraffin-embedded tumor samples of 94
cytotoxin, resulting in cell death. Here, we report safety, pharmacokinetic lung cancer patients treated at Yale Cancer Center between 2002-2013.
(PK) and efficacy in an ongoing phase 1/2 study of Japanese patients Multiplexed QIF was used to evaluate the cases with a panel containing
with/without EGFR-amplified recurrent GBM (rGBM). Methods: M13-714 phenotype markers for major T-cell subsets (CD3, CD4, CD8 and FOXP3),
(INTELLANCE-J, NCT02590263) is a non-randomized, phase 1/2 study in and cell-localized activation and proliferation (granzyme-B and Ki-67).
Japanese patients. Phase 1 assessed tolerability and PK where the dose Signal for each marker was measured in marker-selected tissue compart-
escalation of depatux-m was from 0.5 to 1.25 mg/kg/Q2W at day 1 and 15 ments using the Automated Quantitative Analysis (AQUA) platform. Asso-
during 28-day cycle until progression disease (PD) or intolerable toxicity. ciations between markers and major clinicopathologic variables were
Phase 2 assessed efficacy and safety of depatux-m in EGFR-amplified, rGBM studied. Results: In total, 40 primary lung tumors, 63 BrM, and 15 ECM
and patients received 1.0 mg/kg of depatux-m on day 1 and 15 + 150 mg/m2 were analyzed, including paired samples from 22 patients. Lung cancer
temozolomide (TMZ) on days 1-5 during each 28-day cycle until PD or in- histology included adenocarcinoma 62.5%, squamous cell carcinoma
tolerable toxicity. Results: As of 10 Jan 2019, 38 patients (WHO grade $3) 11.5%, small cell 9.4%, and other non-small cell 16.7%. BrM had both
were enrolled (9 in phase 1, 29 in phase 2). There was no dose limiting toxicity significantly lower levels of CD3+ T-cells (p, 0.0001) and T-cell granzyme B
in phase 1. The recommended phase 2 dose was 1.25 mg/kg where the most (p= 0.0188) compared with primary lung tumors. No significant differences
common adverse events (AEs) were punctate keratitis in 21 patients (72%); were observed in T-cell Ki-67 levels across tissues. FOXP3 measured in
lymphopenia in 14 patients (45%), thrombocytopenia in 13 patients (41%). CD4+ T-cells were significantly lower in BrM compared with primary ma-
Grade 3/4 AEs included thrombocytopenia and lymphopenia in 20 patients lignancies (p= 0.0002) and ECM (p= 0.0404). Among patients with BrM,
(69%). Ocular AEs were reported in 27 patients (93%) including punctate higher levels of CD3+ T-cells in BrM were associated with longer overall
keratitis (72%). PK results (31 patients) in both phases were similar to those survival. Conclusions: Lung cancer-associated BrM have lower T-cell in-
of non-Japanese result. Progression Free Survival (PFS) of 27 patients in filtration, cytolytic function, and regulatory T-cells than primary lesions.
phase 2 for 12 and 6 months were 8% and 27.5% respectively. The median These results indicate lower adaptive anti-tumor responses in BrM and
PFS was 4 months. The overall survival (OS) for 24, 12 and 6 months were suggest the presence of a tolerogenic microenvironment in the brain.
28%, 62.5% and 93% respectively. The median OS was 15.5 months. Overcoming this could be used to design optimal treatment strategies.
Conclusions: Preliminary safety, PK and efficacy in Japanese patients with/
without EGFR-amplified, rGBM suggests depatux-m was tolerated and showed
encouraging anti-GBM effects. Clinical trial information: NCT02590263.

2068 Poster Session (Board #257), Sun, 8:00 AM-11:00 AM 2069 Poster Session (Board #258), Sun, 8:00 AM-11:00 AM
Stereotactic radiosurgery for resected brain metastases: Does the surgical Insight into the brain metastasis journey: Initial survey results from patients
corridor need to be treated? First Author: Siyu Shi, Stanford Cancer Institute, and caregivers. First Author: Nicole Willmarth, American Brain Tumor As-
Palo Alto, CA sociation, Chicago, IL
Background: Post-operative stereotactic radiosurgery (SRS) is the standard Background: Brain metastases (BM) are the most common central nervous
of care for resected brain metastases, but SRS techniques are not stan- system tumors in the US. Though the exact incidence is unknown, BM are
dardized. Although expert consensus guidelines recommend that the sur- estimated to occur in up to 10-20% of all cancers. Despite the high fre-
gical corridor leading to the resection cavity be included in the SRS plan, this quency, there is little systematic knowledge about how BM are typically
statement is not based on data. We analyzed the patterns of failure and diagnosed and treated. The American Brain Tumor Association (ABTA) seeks
toxicity with post-resection SRS, with the hypothesis that the corridor needs to understand the BM journey: symptoms, diagnosis, treatment, and end of
not be targeted with SRS. Methods: In this IRB-approved retrospective life, through a survey of BM patients and caregivers. Methods: Two surveys
review, from 428 lesions treated from 2005-2018 with post-resection SRS, were developed by the ABTA with vendor, PSB Research, after careful lit-
58 lesions (57 patients) had evaluable data and a ‘deep’ tumor with a erature review. The surveys were reviewed by a panel of clinicians who treat
surgical corridor defined as $ 1.0 cm from the surface pre-operatively. SRS BM patients. Online survey research was conducted between 8/13-9/16/18,
targeted the surgical corridor, defined as the surgical tract uninvolved by with one survey for adults with BM (N = 237) and another for caregivers (N =
tumor on pre-operative imaging, in 33(57%). Brain failure was defined as 211). Respondents came from PSB’s panels and ABTA collaborators:
local (LF) if within the surgical cavity involved with tumor pre-resection, LUNGevity, Melanoma Research Foundation and the Kidney Cancer Asso-
corridor (CF) if within the surgical tract leading to the surgical cavity, distant ciation. Results: Ninety percent of patients, and a similar number of care-
(DF) if a new parenchymal tumor, or leptomeningeal (LMD) for new nodular/ givers, were surprised by the diagnosis, with only 20% of patients knowing
classical leptomeningeal enhancement. The cumulative incidences of about BM before diagnosis. Most caregivers were the adult child of a patient.
failure and adverse radiation effect (ARE) were analyzed with death and The impact of the diagnosis was primarily emotional. Top concerns after
whole brain radiation therapy as competing risks. Results: The median diagnosis, for both patients and caregivers, were likelihood of treatment
follow-up was 14 months. Not targeting surgical corridor was associated with success and impact on quality of life. Although a majority of patients were
prior SRS/resection for other brain metastases (23% vs. 0%, p=0.01), happy with the quality of information given, they stated a need to receive a
deeper tumors (median 2.1 cm vs. 1.4 cm, p,0.01), and systemic treatment greater quantity of information about treatment success and options. Only
within 3 months (p =0.01), but not other factors (p.0.10). The 12-month 30% of patients were referred to a patient advocacy organization. When re-
failure rates, if surgical corridor was not treated vs. treated, respectively, ferred, information on treatment success rates and options was most sought.
were: CF 8% (1-24%) vs. 0% (p=0.12), LF 4% (0-17%) vs. 13% (4-27%) Conclusions: Direct patient and caregiver feedback provides valuable insight
(p=0.32), LMD 40% (19-61%) vs. 10% (2-23%) (p=0.03), DF 65% (43- towards understanding the BM journey and resources needed to support
81%) vs. 35% (19-52%) (p=0.02), and ARE 8% (1-22%) vs. 13% (4-28%) patients and caregivers. A subsequent survey among oncologists and other
(p=0.35). After adjusting for systemic treatment, differences were not clinicians, planned for spring of 2019, will add to these findings.
statistically significant (p.0.05). Conclusions: Omitting the surgical cor-
ridor in post-operative SRS for resected brain metastases was not associated
with statistically significant differences in recurrences or adverse radiation
effect. Surgical corridor does not need to be included in all post-resection
SRS.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
104s Central Nervous System Tumors

2070 Poster Session (Board #259), Sun, 8:00 AM-11:00 AM TPS2071 Poster Session (Board #260a), Sun, 8:00 AM-11:00 AM
A phase II trial of bevacizumab in patients with recurrent solid tumor brain Novel anti-EGFRvIII bispecific T cell engager (BiTE) antibody construct in
metastases who have failed whole brain radiation therapy (WBRT). First glioblastoma (GBM): Trial in progress of AMG 596 in patients with recurrent
Author: Priya Kumthekar, Northwestern Memorial Hospital, Chicago, IL or newly diagnosed disease. First Author: Mark Rosenthal, Peter MacCallum
Cancer Centre, Melbourne, Australia
Background: Brain metastases (BM) are the most common intracranial
malignancy with overall a poor prognosis estimated at approximately Background: GBM is the most aggressive primary brain tumor in adults and is
4 months from time of initial diagnosis for treated patients, and even lower extremely difficult to treat. Patients with GBM tend to progress rapidly within
after failing WBRT after which treatment options have been limited and weeks or months. Median overall survival is only 12–15 months despite
outcomes poor. Methods: This is an open label phase 2 study where patients aggressive treatment, and less than 5% of patients survive 5 years. GBM also
who have previously failed WBRT received bevacizumab at a dose of 10 mg/ severely impacts quality of life and cognitive function. Approximately 50% of
kg intravenously every two weeks until CNS disease progression with one GBM tumors test positive for amplification or mutation of the epidermal
cycle being defined as 4 weeks. The primary endpoint was objective ra- growth factor receptor (EGFR), the most common of which is the EGFRvIII
diographic tumor response as defined by modified Response Assessment in gain-of-function mutation. AMG 596 is a bispecific T cell engager (BiTE®)
Neuro-oncology (RANO) criteria. Secondary endpoints included progression antibody construct designed to crosslink and engage CD3-positive T cells to
free survival (PFS) at 6 months, time to progression, time to response, EGFRvIII-positive tumor cells, inducing tumor cell lysis and T cell pro-
duration of response, overall survival (OS), quality of life (QOL) as measured liferation. A clinical trial is being conducted for this novel immunotherapy
by the FACT-G and FACT-Br and safety. Results: A total of 27 patients were agent in patients with EGFRvIII-positive GBM. Methods: NCT03296696 is a
consented and registered to study of which 24 were evaluable for ORR (3 phase 1, first-in-human, open-label, sequential dose-escalation and dose-
came off study prior to first follow up MRI brain). Medianage was 53 (range expansion study evaluating the safety, tolerability, and pharmacokinetics
27-73), median number of cycles was 5.5 (range 1-20) with a median follow and pharmacodynamics (PK/PD) of AMG 596 in patients with EGFRvIII-
up of 8.7 months (range 2.4-47.9mo). Of the 24evaluable patients, there positive GBM. AMG 596 is administered via continuous intravenous in-
were 6 Partial response, 16 stable disease and 2progressive disease. The fusion. The study is expected to enroll approximately 82 patients total and
6 month PFS: 46% (95% CI: 25% - 67%) and median PFS was 5.3 months. comprises two groups (Group 1: patients with recurrent GBM; Group 2:
Median OS was 9.5 months (95% confidence interval 6.3m – 15.0m). For patients with newly diagnosed GBM in the maintenance treatment phase
the patients who completed sequential QOL assessments, there was no after standard of care). Key inclusion criteria include: male or female; $ 18
significant decline in QOL but there was a nonsignificant improvement in the years of age; with pathologically documented and diagnosed grade IV GBM;
FACT-Br scores. Overall, treatment was well tolerated with 3grade 3 adverse Eastern Cooperative Oncology Group performance status # 1; life expectancy
events seen: hypertension (n = 3), headache (n = 1) and thrombotic event $ 3 months per study investigator; and acceptable renal, hematological, and
(n = 1). Conclusions: For this WBRT failure BM population, we were able to hepatic function. The primary endpoint evaluates the safety and tolerability of
show a 25% disease response to bevacizumab therapy along with good drug AMG 596 via collection of treatment-emergent adverse events. Additional
tolerability and no noted central nervous system bleeding. Improved survival endpoints include objective response rate per modified Response Assessment
as compared to historical controls was seen 9.5 m. Of the 24 evaluable in Neuro-Oncology Criteria and PK/PD analyses of AMG 596 in serum. The study
patients, 81% (22/24) experienced clinical benefit defined as stable disease began enrolling patients in April 2018 and enrollment is ongoing. For more
or better. Bevacizumab therapy could be a viable option for solid tumor BM information, please contact Amgen Medical Information: [email protected]
patients who experience progression following WBRT, however a larger trial com. Clinical trial information: NCT03296696.
is required to confirm this data. Clinical trial information: NCT01898130.

TPS2072 Poster Session (Board #260b), Sun, 8:00 AM-11:00 AM TPS2073 Poster Session (Board #261a), Sun, 8:00 AM-11:00 AM
EORTC 1709/CCTG CE.8: A phase III trial of marizomib in combination A phase I/II study of nivolumab plus or minus ipilimumab in combination
with standard temozolomide-based radiochemotherapy versus standard with multifraction stereotactic radiosurgery for recurrent high-grade
temozolomide-based radiochemotherapy alone in patients with newly diagnosed radiation-relapsed meningioma. First Author: Jiayi Huang, Washington
glioblastoma. First Author: Patrick Roth, University Hospital Zurich, Zurich, University School of Medicine in St. Louis, St. Louis, MO
Switzerland
Background: There is currently lack of effective therapy for meningiomas
Background: The standard treatment for patients with newly diagnosed that have relapsed despite surgery and radiation therapy (RT). Reirradiation
glioblastoma comprises maximum safe surgery, radiotherapy (RT), and have been used in selected cases, but the long-term clinical outcomes
concomitant and up to six cycles of maintenance temozolomide (TMZ) remained poor, especially for high-grade meningiomas. Preclinical data have
chemotherapy (TMZ/RT→TMZ). Despite this intense therapy, the prognosis suggested synergy between hypofractionated radiosurgery with immune
remains poor and there is an urgent need to develop new therapeutic options. checkpoint inhibitors such as PD1 and CTLA4 inhibitors. The purpose of this
Marizomib is a novel, irreversible and brain-penetrant pan-proteasome in- study is to evaluate feasibility and preliminary clinical efficacy of combining
hibitor. Following its successful assessment in phase I trials in patients with reirradiation using hypofractionated radiosurgery with concurrent nivolumab
newly diagnosed as well as recurrent glioblastoma, marizomib is now being (PD1 inhibitor) plus or minus ipilimumab (CTLA4 inhibitor) for recurrent
investigated in a phase III trial. Methods: EORTC 1709/CCTG CE.8 is a high-grade meningiomas. Methods: During the phase I portion, eligible
multicenter, randomized, controlled, open label phase III superiority trial. patients will be treated according to treatment-escalation schema following
Eligibility criteria include histologically confirmed newly diagnosed glio- the modified 3+3 design (Table). The maximum tolerated combination
blastoma and a performance status $70. Approximately a total of 750 (MTC) will be the regimen at which #1/6 patients experience dose-limiting
patients will be enrolled and randomized 1:1. Stratification factors include toxicity within 8 weeks of the start of study therapy. During the phase II
institution, age, Karnofsky performance status and extent of surgery. The portion, a total of 24 evaluable patients will be enrolled at the MTC using
primary objective of this study is to compare overall survival in patients Simon’s MiniMax two-stage design. Key eligibility criteria include patients
receiving marizomib in addition to standard of care (TMZ/RT→TMZ) with with recurrent grade II-III meningiomas after prior RT; age $ 18 years; ECOG
patients receiving standard treatment only. The testing strategy specifies the score # 2; measurable disease but # 5 cm (or 20 cm3); prior radiation
determination of this objective in both the intent-to-treat population and the dose # 70 Gy with at least 6 months interval; normal organ function; no
subgroup of patients with tumors harboring an unmethylated MGMT pro- active autoimmunity. The primary endpoints are to determine the MTC
moter. Secondary endpoints include progression-free survival, safety, (phase I) and the objective response rate of the MTC (phase II). Secondary
neurocognitive function and quality of life. The study is accompanied by a endpoints include safety, duration of response, progression-free survival,
translational research program. The study will be opened at 50 EORTC sites overall survival. Exploratory endpoints include developing an immune or
in Europe and done as an intergroup collaboration with the Canadian Cancer molecular signature for predicting treatment response and resistance. The
Trials Group (CCTG) with 25 sites in Canada and additional sites in the US. trial is actively enrolling and funded by the National Cancer Institute Ex-
Patient enrolment started in June 2018 and as of January 29, 2019, a total perimental Therapeutics Clinical Trials Network (NCI-ETCTN). Treatment
of 85 patients have been randomized. An update on the enrolment status Escalation Schema. Clinical trial information: NCT03604978.
will be provided at the ASCO conference. Clinical trial information:
NCT03345095. Dose
Treatment Level Nivolumab Ipilimumab Radiosurgery
Cohort A 480 mg IV Q4W -- 8 Gy x 3 QOD
Cohort B 3 mg/kg IV Q2W 1 mg/kg IV Q6W 8 Gy x 3 QOD

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Central Nervous System Tumors 105s

TPS2074 Poster Session (Board #261b), Sun, 8:00 AM-11:00 AM TPS2075 Poster Session (Board #262a), Sun, 8:00 AM-11:00 AM
Phase I study of PD-L1 inhibition with avelumab and laser interstitial thermal A phase Ib trial of CB-839 (telaglenastat) in combination with radiation
therapy in patients with recurrent glioblastoma. First Author: Adilia Hormigo, therapy and temozolomide in patients with IDH-mutated diffuse astrocytoma
Icahn School of Medicine at Mount Sinai, New York, NY and anaplastic astrocytoma (NCT03528642). First Author: Sani Haider
Kizilbash, Mayo Clinic, Rochester, MN
Background: Glioblastoma (GBM) the most frequent malignant brain tumor
in the adult has a dismal prognosis and limited treatment options. Current Background: IDH mutant astrocytomas express high levels of 2-hydroxy-
advances have highlighted how tumors and specifically GBM evade the glutarate (2-HG), an oncogenic metabolite which drives gliomagenesis.
immune system by exploiting the mechanisms of tolerance and inducing Excess 2-HG inhibits branched chain amino acid transaminase, which
local and systemic immunosuppression. Another hurdle in the treatment of catalyzes glutamate synthesis from branched chain amino acids. This defect
GBM is the blood-brain barrier (BBB). Recent work suggests that MRI-guided causes these tumors to become more reliant on glutaminase for glutamate
laser interstitial thermal therapy (LITT) can increase the permeability of the biosynthesis from glutamine. CB-839 (telaglenastat) is a novel glutaminase
BBB and may have an abscopal effect. Therefore, utilizing MRI-guided inhibitor which is well tolerated in humans. McBrayer et al have recently
LITT, a potential immunogenic cell death-inducing procedure that disrupts demonstrated that CB-839 further depletes intracellular glutamate and
the BBB and makes Avelumab a PD-L1 monoclonal antibody being more glutathione in IDH mutant glioma cells, and enhances RT (radiation therapy)
accessible to GBM tumors, seem a valid approach for immunomodulation efficacy in an orthotopic glioma model. Methods: NCI #10218 is a CTEP
and successful implementation of a combined regimen to treat brain cancer. supported phase I clinical trial investigating the safety and tolerability of CB-
Methods: This is a prospective non-randomized open label to characterize 839 when combined with RT/TMZ (temozolomide) in patients with pre-
the tolerability and safety profile of Avelumab in combination with LITT in viously untreated IDH mutant grade II/III astrocytoma. Patients with grade II
patients with recurrent glioblastoma who were treated with radiation therapy and III astrocytomas will be treated with 50.4 and 59.4 Gy of RT, re-
with concurrent Temozolomide chemotherapy at diagnosis, and whose tumor spectively, with standard doses of concurrent TMZ. CB-839 will also be
at recurrence measures less then 3 cm3. Avelumab is administered within a administered orally concurrently with RT, with doses escalated in cohorts
week after real-time MRI-guided LITT therapy and every 2 weeks thereafter. based on a standard 3+3 design. After defining the maximum tolerated dose
On part A patients are treated with intravenous Avelumab alone and on part B (MTD) of CB-839, an expansion cohort will evaluate the pre- and post-CB-
patients receive Avelumab in combination with MRI-guided LITT. Part A 839 therapy metabolome of patients with IDH mutant astrocytoma. En-
completed enrollment without DLT. Enrollment on part B began in October rollment to this cohort will require measurable disease and patients will
2018. A Simon minimax two-stage design is being used for efficacy. Toxicity additionally be treated with a 7 day run-in of CB-839 at MTD prior to RT. The
will be scored using the NCI-CTCAE 4.03 criteria. Blood samples and tumor effect of CB-839 on the metabolome will be studied in both plasma (LC/MS/
tissue will be collected for correlative studies. Quantification of the changes MS) and tumor (magnetic resonance spectroscopy), along with PK to confirm
in inflammatory and immunosuppressive profiles across time points for pa- adequacy of systemic exposure. Preliminary data on neurocognitive end-
tients receiving treatment with Avelumab will be obtained. This information points will also be acquired. NCI #10218 is currently activated for enroll-
will instruct future immunotherapy approaches to treat GBM and the rational ment to cohort 1. Clinical trial information: NCT03528642.
for those combinations. Clinical trial information: NCT03341806.

TPS2076 Poster Session (Board #262b), Sun, 8:00 AM-11:00 AM

Oral DNA vaccination targeting VEGFR-2 combined with anti-PD-L1 avelumab
in patients with progressive glioblastoma, a phase I/II study: NCT03750071.
First Author: Wolfgang Wick, National Center for Tumor Diseases (NCT), UKHD
and German Cancer Research Center (DKFZ), Heidelberg, Germany
Background: The vaccine (VXM01) is a VEGFR-2 coding DNA vaccine,
using a Salmonella Ty21a carrier for oral application. VEGFR-2 is over-
expressed in glioblastoma and serves as a promising target for VEGFR-2
primed T cells with the potential to alter tumor angiogenesis and/or eliminate
VEGFR-2 expressing tumor cells. VXM01 was well tolerated in a previous
phase I/II study involving 14 patients with progressive glioblastoma multi-
forme. Immunological correlates of vaccination and anti-tumor immunity in
the blood and in the tumor were detected. At least one objective clinical
response was attributed to vaccine monotherapy, with one more PR achieved
in combination with nivolumab. Prolonged overall survival was associated
with peripheral immune responses against VEGFR-2, J Clin Oncol 36, 2018
(suppl; abstr 2017). A combination study with the anti PD-L1 checkpoint
inhibitor monoclonal antibody avelumab is currently underway. Methods: A
multicentre, open-label phase I/II study (EudraCT.gov no. 2017-003076-
31), will enrol 30 patients with progressive glioblastoma, previously treated
with temozolomide/radiotherapy. The primary objective is to evaluate safety
and tolerability of the vaccine in combination with avelumab. In a 1+2 safety
run in, two cohorts of non-reoperable patients will be vaccinated with one of
2 doses of the oral vaccine (106 or 107 CFU) with concurrent intravenous
avelumab. After safety evaluation and recommendation of study continua-
tion by the Data Safety Monitoring Board, 18 non-re-operable and 6 re-
operable patients will be treated with 107 CFU of the vaccine + avelumab.
Vaccinations for all patients will be on day 1, 3, 5, and 7, followed by 4-
weekly boosts until progression. Avelumab 800 mg will be administered
every two weeks until progression. The enrolment of cohort 1 started with
inclusion of the 1st patient in November 2018. The end of study is week 60.
Follow up visits will be on months 1, 3, 6, 12 and 24. Objective response rate
(ORR), clinical response using iRano criteria, immunological correlates
before and after treatment using ELISpot, FACS, TCR-sequencing, IF, and
IHC laboratory methods. Clinical trial information: NCT03750071.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
106s Developmental Immunotherapy and Tumor Immunobiology

2500 Oral Abstract Session, Sun, 8:00 AM-11:00 AM 2501 Oral Abstract Session, Sun, 8:00 AM-11:00 AM
Phase I study of pembrolizumab in people with HIV and cancer. First Author: Phase II study of durvalumab (MEDI4736) in cancer patients HIV-1-infected.
Thomas S. Uldrick, HIV/AIDS Malignancy Branch, CCR, NCI, Bethesda, MD First Author: Marı́a González-Cao, Instituto Oncológico Dr. Rosell, Barcelona,
Spain
Background: People with HIV have been excluded from immuno-oncology
(IO) studies. Anti- PD-1/PD-L1 therapies are approved for a growing number Background: Durvalumab (MEDI4736), a programmed cell death-ligand 1
of cancers. We evaluated pembrolizumab (pembro) in people with HIV and (PD-L1) blocking antibody, is currently approved for treatment of several
cancer. Methods: CITN-12 is a multicenter phase 1 trial. Key eligibility: cancer types. As HIV-1-infected (HIV+) patients have been excluded from
advanced cancer; ECOG #1; CD4 $100 cells/mL; $4 weeks antiretroviral cancer clinical trials, there are no data on the safety of durvalumab in this
therapy (ART), HIV viral load (VL) , 200 copies/mL. Exclusion: uncontrolled population. Methods: DURVAST (NCT03094286) is a multicenter, open-
HBV/HCV, autoimmune disease. Participants (pts) accrued into CD4 based label, phase 2 clinical trial evaluating the safety and feasibility of durvalu-
cohorts (C): C1 100-199; C2 200-350; C3 . 350 CD4 cells/mL. Pembro mab treatment at the recommended dose of 1500 mg Q4W in HIV+ cancer
200 mg IV administered Q3W for up to 35 doses. Adverse events (AE) patients with solid tumors. Secondary endpoints include analyses of antitu-
evaluated by CTCAE. Immune related AE $ grade (Gr) 2 were events of moral activity in terms of objective response rate and duration of response
clinical interest (irECI). Clinical benefit (tumor shrinkage or stable disease (DOR). An associated translational sub-study includes the assessment of
[SD] $24 weeks) was estimated. Data were locked for safety analysis and antiviral activity and the interaction of chronic viral infection with anti-cancer
publication once C2 and C3 completed accrual and all pts completed $2 response and drug tolerance. Results: Twenty HIV+ individuals with advanced
cycles. Accrual continued for 6 C1 pts and a new phase 1b Kaposi sarcoma solid tumors were enrolled (Table). All participants maintained their standard-
(KS) cohort (C4). Results: 30 pts, characteristics: C1 (6), C2 and 3 (12 of-care antiretroviral therapy. Basal plasma viremia was undetectable and CD4+
each), median (med) age 57 years (range 39-77), 28 men, 2 women, 60% T-cell count was over 200/mm3. There were no durvalumab-related serious
White, 30% Black, 10% Hispanic. Med CD4 285 cells/mL (132 - 966). adverse events. Only 8 patients (40%) presented drug-related adverse events
Cancers: KS (6), non-Hodgkin lymphoma (NHL) (5), non-AIDS defining (19) – (all grade 1-2) including diarrhea (15%), rash (15%), nausea (15%) and as-
most common: anal (6) and squamous cell skin (3). Prior radiation (19), med thenia (10%). Best response includes: partial response in 5 (25%) (4 NSCLC
prior systemic therapies 2 (0-8). Safety observed over 183 cycles, med 5 (1- and 1 anal cancer), stable disease in 4 (20%) (3 NSCLC and 1 melanoma) and
32). Treatment emergent AE $ possibly attributed to pembro mostly Gr 1-2, progression disease in 11 (55%) patients. At data cut-off, 8 patients (40%)
with 20% of pts having Gr 3. irECI: hypothyroidism (6), elevated AST/ALT (1), remained on therapy for a median of 10.5 months (range: 6-19 m). Median DOR
pneumonitis (3), rash (2), musculoskeletal (1).1 KS pt developed KSHV- has not been reached (range 1m to 19 m+). Plasma viremia remained sup-
associated multicentric Castleman disease (KSHV-MCD) and died of the AE. pressed during the study suggesting no viral reactivation upon durvalumab
HIV was controlled and increasing CD4 counts were observed. Best response: treatment. Conclusions: DURVAST study demonstrates durvalumab safety in
complete (lung, 1), partial (NHL, 2), SD $24 weeks (KS, 2), SD , 24 weeks HIV+ cancer patients and suggests an excellent tolerance profile. Understanding
(13), progressive disease (10), not evaluable (2). Conclusions: Pembro has how chronic viral infection could contribute to a better tolerance towards immune
acceptable safety in cancer pts with HIV on ART and . 100 CD4 cells/mL, checkpoint inhibitors will open a new way for the development of safer anti-cancer
similar to patients without HIV. Anti-PD-1 may unmask KSHV-MCD and such immunotherapy strategies. Clinical trial information: NCT03094286.
KSHV-viremic patients should be excluded. Clinical benefit was noted in Total (n) 20
several tumor types. Anti-PD1 is appropriate for FDA-approved indications in Age Median (Range) 54 (30-73)
this population. Patients with HIV meeting appropriate eligibility criteria Sex Male (%) 16 (80%)
NSCLC 14 (70%)
should be included in IO studies. Clinical trial information: NCT02595866. Melanoma 2 (10%)
Anal cancer 2 (10%)
Bladder cancer 1 (5%)
SCLC 1 (5%)
1st line 8 (40%)
2nd – 4th line 12 (60%)

2502 Oral Abstract Session, Sun, 8:00 AM-11:00 AM 2503 Oral Abstract Session, Sun, 8:00 AM-11:00 AM
A phase II study of pembrolizumab for HPV-associated papilloma patients A phase II, randomized, double-blind, placebo-controlled trial evaluating
with laryngeal, tracheal, and/or pulmonary involvement. First Author: Sara I. efficacy and safety of namodenoson (CF102), an A3 adenosine receptor
Pai, Massachusetts General Hospital Cancer Center, Boston, MA agonist (A3AR), as a second-line treatment in patients with Child-Pugh B
(CPB) advanced hepatocellular carcinoma (HCC). First Author: Salomon M.
Background: Recurrent respiratory papillomatosis (RRP) is caused by human
Stemmer, Davidoff Cancer Center, Rabin Medical Center-Beilinson Hospital,
papillomavirus (HPV) types 6 & 11. RRP proliferates in the respiratory tract
Petah Tikva, Israel
impacting breathing, swallowing, and voice and carries a 1-4% risk of ma-
lignant transformation. There is no curative therapy for RRP. Given the Background: There is no established primary treatment for patients with
tolerized host immune response against HPV, in part through upregulation of advanced HCC and severe liver dysfunction (Child-Pugh B class; CPB), thus
the PD1:PDL1 axis, the safety and efficacy of pembrolizumab (pembro) as a this representing a clear unmet need. Namodenoson, an A3AR agonist,
novel treatment for this benign tumor patient (pt) population was evaluated showed promising preliminary results in this population in an open label phase
in a phase II clinical trial. Methods: RRP pts . 12 years of age were treated 1/2 clinical study (NCT00790218), with median overall survival (OS) of
with pembro 200mg every 3 weeks. Adjuvant surgical debridement of RRP 8.1 months. We present the results of a double blind, randomized phase 2,
was permitted for airway obstruction but not dysphonia. Primary endpoints placebo-controlled study (NCT02128958), assessing the efficacy and safety
were best overall response (ORR) (measured by endoscopic lesional burden) of namodenoson as a second-line therapy of patients with advanced HCC and
and safety. Greater than 5 pts with disease in response out of 21 (assuming . 1 CPB class. Methods: Patients were randomized 2:1 to BID namodenoson
of first n = 11 with disease in response) provided 86% power to distinguish (25 mg; n = 50) or placebo (n = 28) in 15 centers globally. Primary endpoint
between a 15% and a 38% ORR (one-sided 8% binomial test). Serial biopsies was OS and secondary endpoints were safety, progression-free survival (PFS),
(up to 8 biopsies/patient over the 24 months of treatment) to identify bio- objective response (OR) and disease control rate (DCR). Assessment of OS and
markers of response and mechanisms of immune resistance (PD-L1 expres- PFS was done by log rank test at a one final analysis when 75 deaths had
sion, mutations in HLA class I antigen presentation machinery, and tumor occurred. Response was assessed by RECIST (local investigator) and mRE-
mutational burden) are underway. Results: Accrual is complete (n = 21 pts CIST (central review). Results: The study did not meet the primary end point,
accrued between May 2016 and Jan 2019). Median age (range) was 45 (19- with median OS 4.1 months (mo) for namodenoson vs. 4.3 mo for placebo
68), 57% (12/21) were male and 67% (14/21) were white. As of February 1, (HR: 0.82). Pre-planned subgroup analysis of Child-Pugh 7 patients (n=56;
2019, ORR is 43% (9/21) (.95 two-stage CI: 22%-66%) (4 of 11 with namodenoson=34, placebo=21) showed median survival 6.8mo vs 4.3 mo
juvenile-onset RRP and 5 of 10 with adult-onset RRP disease responded). No [HR: 0.77 (95% CI 0.49-1.40)]. Similarly, for this subgroup of patients PFS
complete responses have been observed. Fatigue was the most frequent was 3.5 mo vs 1.9 (HR=0.87). In terms of objective response, 3/34 patients
treatment related adverse event (TRAEs); Grade 3 TRAEs included uveitis and assessed achieved OR (9%) with namodenoson vs 0% for placebo. Namo-
hypophysitis, both were reversible upon pembro discontinuation. The frequency denoson was generally well-tolerated, with no treated patients being with-
of surgical interventions was reduced in all pts undergoing surgery for airway drawn for toxicity and no cases of treatment-related deaths. The most
palliation prior to study entry. Conclusions: Pembro reduces the need for routine common adverse event (.10%) were anemia, abdominal pain, ascites,
surgical interventions based on the response rates being achieved. Further study nausea, asthenia, fatigue, peripheral edema, and increased AST. Treatment-
of pembro +/- other agents is warranted to achieve and sustain complete re- related grade 3 toxicities accounted for anemia, fatigue and hyponatremia.
sponses in this population. Clinical trial information: NCT02632344. Conclusions: Namodenoson has demonstrated favorable clinical safety
profile in patients with advanced HCC and severe liver dysfunction. Although
the primary end-point was not met, the subgroup analysis showed a positive
signal of efficacy for OS in patients with Child-Pugh 7. Both safety and
efficacy results warrant testing this drug in a phase III trial. Clinical trial
information: NCT02128958.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Immunotherapy and Tumor Immunobiology 107s

2504 Oral Abstract Session, Sun, 8:00 AM-11:00 AM 2505 Oral Abstract Session, Sun, 8:00 AM-11:00 AM
Results from a first-in-man, open label, safety and tolerability trial of CAN04 Immunobiology, preliminary safety, and efficacy of CPI-006, an anti-CD73
(nidanilimab), a fully humanized monoclonal antibody against the novel antibody with immune modulating activity, in a phase 1 trial in advanced
antitumor target, IL1RAP, in patients with solid tumor malignancies. First cancers. First Author: Jason J. Luke, University of Chicago Comprehensive
Author: Ahmad Awada, Institut Jules Bordet, Brussels, Belgium Cancer Center, Chicago, IL
Background: Interleukin 1 receptor Accessory protein, IL1RAP, is expressed Background: CPI-006 inhibits CD73, a nucelotidase that converts AMP to
in several solid tumors, both on cancer cells and tumor-associated in- adenosine and functions as a lymphocyte adhesion molecule. CPI-006 is a
flammatory cells. CAN04 is a first-in-class fully humanized monoclonal humanized IgG1 FcgR binding-deficient antibody that binds to CD73+ T and
antibody targeting IL1RAP blocking IL-1 alpha and beta signaling and B lymphocytes leading to activation of B cells and expression of CD69. This
triggering antibody dependent cellular cytotoxicity. Methods: The primary study investigates the immunobiology, safety, and efficacy of CPI-006
objective was to assess safety and tolerability of weekly CAN04 in order to monotherapy and in combination with CPI-444, an adenosine A2A re-
define the Recommended Phase 2 Dose (RP2D). Patients (pts) with relapsed ceptor (A2AR) antagonist (NCT03454451). Methods: Patients with re-
or refractory non-small cell lung cancer (NSCLC), pancreatic ductal ade- lapsed solid tumors were treated in a 3 + 3 escalation study with 1, 3, 6 or 12
nocarcinoma (PDAC), breast or colorectal cancer were included using a 3+3 mg/kg CPI-006 (Q3w IV infusion) monotherapy or in combination with CPI-
dose escalation design. Key eligibility criteria were ECOG #1, normal organ 444 (100 mg, PO, BID). Flow cytometry was performed on blood samples
function and no bleeding disorder/coagulopathy. Tumor responses were for lymphocyte subset analysis and receptor occupancy. Results: 17 patients
evaluated according to irRC every 8 weeks. PK and biomarkers were analyzed were enrolled; 11 monotherapy and 6 combination. CPI-006 was associated
in serum. Results: 22 pts were enrolled across 5 cohorts (1-10 mg/kg). with Grade 1 infusion reactions occuring within 30 minutes of the first
Demography: mean age 62 yrs (39-81); gender 14 M and 8 F; median infusion and were eliminated by premedication with non-steroidals. No DLTs
number of prior lines of therapy 3 (range 1-11). AEs occurred mainly fol- with monotherapy or combination therapy were seen. Receptor occupancy
lowing the first dose and the most common AEs were: infusion related re- on peripheral lymphocytes was maintained for the full dosing interval at 12
action (IRR) (41%), pyrexia (27%), fatigue (23%), chills (23%) and nausea mg/kg. Pharmacodynamic effects suggesting immune modulation were
(23%). AE grade 3 or 4: one IRR, one neutropenia/leukopenia and one observed within 1 hr of infusion at all dose levels and included a decrease in
hypokalemia, all of them grade 3. Serum CRP and IL-6 were reduced after peripheral blood CD73pos B cells (mean reduction 86%, p , 0.05), in-
two weeks of treatment. There were linear increases of AUC and Cmax (1-10 creased CD73neg CD4 T cells (mean increase 37%, p , 0.01), and de-
mg/kg) and CAN04 exposure at 10mg/kg was above levels associated with creased CD8 T cells (mean reduction 20%, p , 0.01) compared to baseline.
signs of efficacy in preclinical models. In pts receiving at least one dose of Overall, CD4:CD8 ratios were increased. Tumor regression was observed in a
CAN04, 9/20 (45%) had SD by irRC (7/20 had SD by RECIST 1.1) at 8 weeks prostate cancer patient after 5 cycles of monotherapy at 6 mg/kg; peripheral
follow up. Two pts, one with NSCLC and one with PDAC, had SD for 6 and B cells partially returned by the second cycle and reached a new homeostatic
4 months (latter still on therapy). Conclusions: CAN04 demonstrated a level through subsequent cycles. No change in serum immunoglobulins were
manageable safety profile and a RP2D of 10 mg/kg has been established. observed. Conclusions: CPI-006 induces a rapid lymphocyte redistribution,
The dose expansion phase of the trial will evaluate CAN04 as monotherapy as including a transient reduction of circulating CD73pos B cells suggesting
well as in combination with relevant chemotherapy regimens in NSCLC and redistribution to lymphoid tissues, and an increased CD4:CD8 ratio, con-
PDAC in separate arms. Clinical trial information: NCT03267316. sistent with increased TH effector/memory cells in the blood. The treatment
has been well-tolerated, and there is early evidence of anti-tumor activity of
CPI-006 monotherapy. Clinical trial information: NCT03454451.

2506 Oral Abstract Session, Sun, 8:00 AM-11:00 AM 2507 Oral Abstract Session, Sun, 8:00 AM-11:00 AM
Determination of the recommended phase II dose of birinapant in combination Phase Ib study of MIW815 (ADU-S100) in combination with spartalizumab
with pembrolizumab: Results from the dose-escalation phase of BPT-201. (PDR001) in patients (pts) with advanced/metastatic solid tumors or lymphomas.
First Author: Russell J. Schilder, Thomas Jefferson University, Philadelphia, First Author: Funda Meric-Bernstam, Department of Investigational Cancer
PA Therapeutics, Division of Cancer Medicine, MD Anderson Cancer Center, Houston,
TX
Background: Birinapant is a bivalent SMAC mimetic targeting cIAP1. Syn-
ergistic effects of combining birinapant with immune checkpoint inhibitors Background: MIW815 (ADU-S100) is a novel synthetic cyclic dinucleotide
have been demonstrated in preclinical models. Based on these observations, a that activates the STimulator of INterferon Genes (STING) pathway impacting
clinical trial with birinapant and pembrolizumab was initiated (NCT02587962). tumor cells, tumor microenvironment, vasculature, tumor-associated fibro-
Methods: Patients $18 years with advanced solid tumors without further blasts, and priming APC and CD8+ T cells. Spartalizumab is a humanized
suitable standard therapeutic options were eligible for inclusion. Birinapant IgG4 mAb that blocks the binding of PD-1 to PD-L1/2. Preclinical data support
(5.6-22 mg/m2) was administered IV on day 1 and 8 in addition to pem- synergistic antitumor effects when MIW815 (ADU-S100) is combined with
brolizumab 200 mg on day 1 in a 21-day cycle until disease progression using checkpoint inhibitors. Methods: In this Phase Ib dose escalation study, pts
standard 3+3 dose-escalation. The primary objective was to determine the with advanced/metastatic solid tumors or lymphoma received MIW815 (ADU-
safety and tolerability of the recommended phase 2 dose (RP2D) of birinapant S100) (intratumoral injections [50–800 mg] either weekly [3 weeks on/1 week
in combination with pembrolizumab. Secondary and exploratory objectives off] or Q4W) and spartalizumab (400 mg IV Q4W). Injected and non-injected
included antitumor activity assessed by RECIST 1.1 and iRECIST, pharma- tumor biopsies were obtained at baseline and on treatment. Primary objectives
cokinetics and assessment of biomarkers including serum cytokines, cIAP1, are to determine safety and identify a dose/schedule for future studies.
PD-L1 expression and tumor infiltrating lymphocytes. Results: Nineteen Preliminary activity, pharmacokinetics (PK), and pharmacodynamics (PD) are
patients were enrolled at 4 dose levels of 5.6 (n = 3), 11 (n = 3), 17 (n = 6) and also being explored. Results: As of Jan 11, 2019, 66 pts (median age: 61 y)
22 (n = 7) mg/m2. Most common tumors were pancreatic (n = 5), colorectal with various solid tumors or lymphomas have been treated. Treatment was
(n = 4), ovarian (n = 3) and sarcoma (n = 3). Median prior therapies were discontinued in 49 pts (74%) due to disease progression (n = 28), pt/
4 (0-12). The most common AE related to any of the study drugs was rash physician decision (n = 18), AE (n = 2), or death (n = 1). No DLTs were
occurring in 3 patients. Ten patients had 17 SAE’s of which only one (sto- reported during the first cycle at any dose level. Most common ($5 pts)
matitis) was judged related to birinapant. Increased ALT/AST (G3/G2) leading treatment-related AEs (TRAEs) were injection site pain (12%), pyrexia
to missed day 8 dose constituted a DLT at 22 mg/m2. Grade 2 lipase increases (11%), and diarrhea (9%). Grade 3/4 TRAEs (in $2 pts) were increased AST
were seen in 2 patients. No cases of Bell’s palsy were detected. ORR by and ALT (3% each). Serious TRAEs were pyrexia (3%), increased amylase,
RECIST 1.1 was 5.6% (n = 1) in 18 evaluable patients. The responding increased lipase, diarrhea, fatigue, hyperthyroidism, partial seizures, dyspnea,
patient had microsatellite stable colorectal carcinoma (MSS-CRC)) and and pneumonitis (all 2%). Partial responses in pts with PD-1–naive TNBC and
remains on therapy 13+ months after first dose. By iRECIST, ORR was PD-1–relapsed/refractory melanoma have been observed. MIW815 (ADU-
11.1%. CBR (PR+SD) by RECIST was 22.2%. The exposure to birinapant S100) plasma exposure generally increased in a dose-dependent manner
generally increased with dose. The RP2D was determined to be 22 mg/m2. with a rapid terminal half-life. Response data, PK and PD analyses will be
Conclusions: Birinapant and pembrolizumab is a safe and tolerable com- presented. Conclusions: Thus far, MIW815 (ADU-S100) + spartalizumab has
bination that has shown encouraging signals of efficacy. A phase 2 study demonstrated antitumor activity in PD-1–naive TNBC and PD-1–relapsed/
evaluating efficacy of this combination in MSS-CRC is ongoing. Clinical trial refractory melanoma. The combination is well tolerated, with no DLTs reported
information: NCT02587962. to date. The MTD has not been reached and dose escalation is ongoing.
Clinical trial information: NCT03172936.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
108s Developmental Immunotherapy and Tumor Immunobiology

2508 Oral Abstract Session, Sun, 8:00 AM-11:00 AM 2509 Clinical Science Symposium, Tue, 8:00 AM-9:30 AM
First-in-human study of REGN3767 (R3767), a human LAG-3 monoclonal Phase I trial of Claudin 18.2-specific chimeric antigen receptor T cells for
antibody (mAb), 6 cemiplimab in patients (pts) with advanced malignancies. advanced gastric and pancreatic adenocarcinoma. First Author: Xianbao
First Author: Kyriakos P. Papadopoulos, South Texas Accelerated Research Zhan, Department of Oncology, Changhai Hospital, Second Military Medical
Therapeutics, San Antonio, TX University, Shanghai, China
Background: We present initial safety, pharmacokinetics (PK), and efficacy Background: As a promising approach for some cancers, chimeric antigen
from the dose escalation study of R3767, alone (mono) or in combination receptor T cell therapy has limited success in solid tumors. Claudin18.2
with cemiplimab (REGN2810), a PD-1 mAb (combo), in pts with advanced (CLDN 18.2) is a stomach-specific isoform of Claudin-18, and highly
malignancies (NCT03005782). Methods: Pts who had progressed on prior expressed in gastric and pancreatic adenocarcinoma, the advanced form of
therapy(ies) and/or for whom no therapy with clinical benefit was available both of which have urgent unmet medical needs. We previously developed
were enrolled; most pts had received no prior anti-PD-1/PD-L1. Pts received and demonstrated ability of CLDN 18.2-specific CAR (CAR-CLDN18.2)
R3767 1, 3, 10, or 20 mg/kg every 3 weeks (Q3W) 6 cemiplimab 3 mg/kg or T cells to eradicate CLDN 18.2-positive gastric cancer xenografts without
350 mg Q3W IV for #51 weeks. Crossover from mono to combo was allowed obvious on-target off-tumor toxicity (Huang J. JNCI 2018). Methods: In this
at progression. R3767 PK were evaluated. Tumor measurements were single-arm, open-label, first-in-human phase I pilot study (NCT03159819)
performed Q6W for the first 24 weeks and subsequently Q9W. Data cut-off to investigate the safety and explore the efficacy of the autologous CAR-
date was Aug 25, 2018. Results: Mono: 27 pts (median age: 66 yr; ECOG CLDN18.2 T cells, patients with confirmed CLDN 18.2 positive advanced
PS: 0 [n=4], 1 [n=23]) were treated. There were no dose-limiting toxicities gastric or pancreatic adenocarcinoma aged 18 to 70 years received 1 or more
(DLTs). The most common treatment-emergent adverse event (TEAE) was cycles of CAR-CLDN18.2 T cell infusion(s) after lymphodepletion pretreat-
nausea (22.2%). Grade $3 immune-related adverse events (irAEs) of in- ment (fludarabine and cyclophosphamide, with or without nab-paclitaxel)
creased alanine and aspartate aminotransferases (each 3.7%) were re- until disease progression or presence of intolerable toxicity. Adverse Event
ported. By investigator-assessment (per RECIST 1.1; INV), best response (AE) grade categorization is according to CTCAE 4.0, and tumor response
was stable disease in 11 pts. Combo: 42 pts (median age: 60 yr; ECOG PS: 0 was assessed per RECIST 1.1. Results: As of November 30th, 2018, 12
[n=15], 1 [n=27]) were treated. One pt treated with R3767 3 mg/kg Q3W + subjects with metastatic adenocarcinoma (7 gastric and 5 pancreatic) were
cemiplimab 3 mg/kg Q3W experienced DLT of grade 4 elevated blood treated with 1–5 cycles (total of 0.5 - 55 X 108) of CAR-positive T cells
creatine phosphokinase, associated with grade 3 myasthenia syndrome and infusions. There were no serious adverse events, treatment-related death or
grade 1 elevated troponin. The most common TEAEs were fatigue (33.3%) severe neurotoxicity occurred in the study. No grade 4 AEs except for
and nausea (21.4%). Grade 3 irAE of hypothyroidism (2.4%) was also re- decreased lymphocytes, neutrophils and white blood cells. All cytokine release
ported. By INV, 2 (both small cell lung cancer) combo pts and 2 (endometrial syndromes observed were grade 1 or 2. Among the 11 evaluable subjects, 1
cancer and cutaneous squamous cell carcinoma) of 12 additional pts who achieved a complete response (gastric adenocarcinoma), 3 had partial re-
crossed over from mono to combo had partial responses. PK: R3767 con- sponses (2 gastric adenocarcinomas and 1 pancreatic adenocarcinoma), 5
centrations in serum increased in a dose-dependent manner and were had stable disease and 2 had progression of disease. The total objective
unaffected by combo. Conclusions: The safety profile of R3767 6 cemi- response rate was 33.3%, with median PFS of 130 days estimated using
plimab was generally tolerable; PK was linear. Early efficacy signals were Kaplan-Meier method [95% CI (38, 230)]. Conclusions: This clinical study
detected despite the difficult-to-treat pt population. Biomarker studies are indicated that CAR-CLDN18.2 T cell therapy were safe and well tolerated and
ongoing. R3767 20 mg/kg or 1600 mg fixed dose equivalent Q3W as mono may have promising therapeutic efficacy in patients with advanced gastric and
and combo were selected for further evaluation. Clinical trial information: pancreatic adenocarcinoma. Clinical trial information: NCT03159819.
NCT03005782.

2510 Clinical Science Symposium, Tue, 8:00 AM-9:30 AM 2511 Clinical Science Symposium, Tue, 8:00 AM-9:30 AM
Results of a phase I study of bispecific anti-CD19, anti-CD20 chimeric Regional delivery of mesothelin-targeted CAR T cells for pleural cancers:
antigen receptor (CAR) modified T cells for relapsed, refractory, non-Hodgkin Safety and preliminary efficacy in combination with anti-PD-1 agent. First
lymphoma. First Author: Nirav Niranjan Shah, Medical College of Wisconsin, Author: Prasad S. Adusumilli, Memorial Sloan-Kettering Cancer Center, New
Milwaukee, WI York, NY
Background: Anti-CD19 CAR-T cell therapy is a breakthrough treatment (tx) Background: We conducted a phase I dose escalation trial of first-in-human
for patients (pts) with relapsed/refractory (R/R) B-cell non-Hodgkin lym- autologous chimeric antigen receptor (CAR) T-cell immunotherapy targeting
phoma (NHL). Despite impressive outcomes, non-response and relapse with mesothelin (MSLN), a cell-surface antigen that is highly expressed in pleural
CD19 negative disease remain challenges. Through dual B-cell antigen cancers- malignant pleural mesothelioma (MPM) and metastatic lung and
targeting of CD20 and CD19, with a first-in-human bispecific lentiviral CAR- breast cancers. Methods: A single dose of CD28-costimulated MSLN CAR
T cell (LV20.19CAR), we aim to improve response rates while limiting CD19 T cells with the I-caspase-9 safety gene was administered intrapleurally in
negative relapse. Methods: Pts were treated on a Phase 1 dose escalation + patients with MSLN-expressing pleural tumors. Following a 3+3 design,
expansion trial (NCT03019055) to demonstrate safety of a 41BB/CD3z patients were treated in dose escalating cohorts (dose range 3E5 to 1E7 CAR
LV20.19CAR T cell for adults with R/R B-cell NHL. Safety was assessed by T cells/kg) following IV cyclophosphamide lymphodepletion (first 3 patients
incidence of dose limiting toxicities (DLTs) within 28 days post-infusion. did not receive cyclophosphamide). A subset of MPM patients received
Starting dose was 2.5 x 10^5 cells/kg with a target dose of 2.5 x 10^6 cells/ subsequent anti-PD-1 therapy, off-protocol, which we have shown to prolong
kg. All pts received fludarabine+cyclophosphamide for lymphodepletion. CAR T-cell functional persistence in preclinical models. Results: Twenty
Results: 11 pts have completed tx to date. 9 pts in dose escalation and 2 pts patients (18 MPM, 1 lung cancer, 1 breast cancer) were treated (prior lines of
in expansion phase. Median age was 54 years (46-67) and histology in- therapy 1–8, 35% received $3 lines of therapy). No CAR T-cell–related
cluded DLBCL = 5 pts, MCL = 4 pts, and CLL = 2 pts. In dose escalation, toxicities higher than grade 1 were observed. Intense monitoring for on-
3 pts were treated at 2.5 x 10^5 cells/kg, 3 pts at 7.5 x 10^5 cells/kg, and target, off-tumor toxicity by clinical (chest or abdominal pain), radiological
3 pts at 2.5 x 10^6 cells/kg with no DLTs. As a result, 2.5 x 10^6 cells/kg (CT/PET or echocardiogram for pericardial effusion, ascites), laboratory
was selected for expansion. In terms of safety, 6 pts developed Grade 1-2 (troponin elevation), and EKG evaluation found no evidence of toxicity.
cytokine release syndrome (CRS) and 3 pts had Grade 1-2 neurotoxicity Fourteen MPM patients received subsequent anti-PD1 therapy (1–21 cycles,
(NTX). No patient had grade 3-4 CRS or NTX and none required ICU level pretreatment tumor PD-L1 , 10% in all patients except one), with 1 patient
care. 4 pts required 1-2 doses of tocilizumab for CRS. The day 28 overall developing grade 3 pneumonitis that responded to steroid treatment. CAR
response rate (ORR) for all pts was 82% (6/11 = complete response (CR) and T cells were detected in the peripheral blood of 13 of 14 patients (1-
3/11 = partial response). All CR pts remain in remission, the longest . 1 39 weeks). At data cut-off date (Jan 31, 2019), among 14 MPM patients that
year. All progressing pts underwent repeat biopsy, and all retained either received combination therapy (follow-up 13-77 weeks, median 31 weeks),
CD19 or CD20 positivity. Additional pts are being enrolled in the expansion best responses included 2 patients with complete metabolic response on
phase and updated data will be presented. Conclusions: Phase 1 results PET (62 and 39 weeks ongoing); 5 partial responses and 4 stable disease by
from the LV20.19 CAR T clinical trial demonstrate that infusion of 2.5 x investigator assessment. Conclusions: Intrapleurally administered MSLN-
10^6 cells/kg is safe for further investigation with no DLTs among treated targeted CAR T cells were safe. Encouraging antitumor activity of MSLN-
pts. Down-regulation of target antigens was not identified as a mechanism of targeted CAR T-cell therapy was observed when combined with anti-PD1
resistance in progressing pts. With limited toxicity and encouraging ORR, therapy and shows promise for future development of this approach. Clinical
dual targeted LV20.19CAR T cells merits further investigation. Clinical trial trial information: NCT02414269.
information: NCT03019055.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Immunotherapy and Tumor Immunobiology 109s

2512 Poster Discussion Session; Displayed in Poster Session (Board #156), 2513 Poster Discussion Session; Displayed in Poster Session (Board #157),
Sat, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Sat, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Sat, 1:15 PM-2:45 PM Sat, 1:15 PM-2:45 PM
Ipilimumab versus placebo after complete resection of stage III melanoma: CX-072, a PD-L1 Probody therapeutic, as monotherapy in patients with
Long-term follow-up results the EORTC 18071 double-blind phase 3 ran- advanced solid tumors: Preliminary results of PROCLAIM-CX-072. First
domized trial. First Author: Alexander M. M. Eggermont, Gustave Roussy Author: Aung Naing, MD Anderson Cancer Center, Houston, TX
Cancer Centre and University Paris-Saclay, Paris, France
Background: Anti-programmed cell death ligand 1 (PD-L1) immunotherapies
Background: Since 2015, ipilimumab (Ipi) is an approved treatment for stage have improved survival in many cancers, but immune-related adverse events
III melanoma based on a significantly (P=0.0013) prolonged recurrence-free (irAEs) have been observed, especially in combination therapy. CX-072 is an
survival (RFS) (Eggermont et al, Lancet Oncology, 2015). At a median follow- investigational Probody therapeutic directed against PD-L1, designed to
up of 5.3 years, RFS (HR=0.76) and distant metastasis-free survival (DMFS) be preferentially activated in the tumor microenvironment and to reduce
(HR=0.76), assessed by an IRC, and overall survival (OS) (HR=0.72) were irAEs. Methods: This is an ongoing phase 1/2a study (PROCLAIM-CX-072;
prolonged in the Ipi group as compared to the placebo (Pbo) group (Eggermont NCT03013491) evaluating CX-072 in patients (pts) with metastatic or
et al, NEJM, 2016), despite a 53.3% (Ipi) vs 4.6% (Pbo) treatment dis- unresectable solid tumors with no further standard curative treatment options
continuation rate due to adverse events. Methods: In this randomized double- and with no prior anti-PD1, PD-L1 or anti-CTLA4 treatment. Pts were un-
blind trial, eligible patients (pts) included those $18 yrs of age who underwent selected for PD-L1 expression. We report preliminary results from expansion
complete resection of stage III cutaneous melanoma (excluding lymph cohorts in anal squamous cell carcinoma (SCCA), cutaneous squamous cell
node metastasis #1 mm or in-transit metastasis). 951 pts were randomized carcinoma (cSCC), small bowel adenocarcinoma (SBA), triple-negative breast
(stratified by stage and region) 1:1 to Ipi 10 mg/kg (n=475) or placebo cancer with skin lesions (TNBC), or undifferentiated pleomorphic sarcoma
(Pbo, n=476) q3w for 4 doses, then every 3 mos for up to 3 yrs until com- (UPS). Pts were treated with CX-072 monotherapy 10 mg/kg intravenously
pletion, disease recurrence, or unacceptable toxicity. Here, we report the every 14 days. Results: As of 30 Nov 2018, 51 pts received CX-072 10 mg/kg
comparison between the Ipi and Pbo groups regarding the long-term efficacy monotherapy: SCCA (n = 9), cSCC (n = 5), SBA (n = 9), TNBC (n = 9), and UPS
outcomes using the local investigator assessments. Results: Overall, 20%/ (n = 19). Median age was 63 y (range, 32-80), 67% were female, and pts had a
44%/36% of pts had AJCC-7 stage IIIA/IIIB/IIIC, 42% ulcerated primary, and median of 3 prior regimens (range, 1-12). Median treatment duration was 1.8 mo
58% macroscopic lymph node involvement. Median follow-up was 6.9 yrs. (range, 0.3-14.7). A grade 3/4 treatment-related adverse event (AE) was ob-
The RFS, DMFS and OS benefit observed in the Ipi group was long-lasting served in 1 pt (2%; gr 3 generalized rash). No grade 3/4 irAEs were observed. Two
(almost 10% difference at 7 years) and consistent across subgroups: no pts discontinued treatment due to AEs: nausea (pt with SCCA) and sepsis (pt
significant predictive factors could be detected. Conclusions: In this phase III with SBA), neither treatment-related. Partial responses (confirmed and un-
trial, Ipi, administered at 10 mg/kg, as adjuvant therapy provided, at a 6.9 yr confirmed) were observed in cSCC (n = 1), TNBC (n = 2), and UPS (n = 1)
median follow-up, a sustained improvement in the RFS, DMFS, and OS long- (Table). Conclusions: CX-072 10 mg/kg monotherapy demonstrated anticancer
term results in patients with high-risk stage III melanoma. Clinical trial in- activity in heavily pretreated pts with advanced solid tumors, including responses
formation: NCT00636168. in cSCC, TNBC with skin lesions, and UPS, with a safety profile that compares
RFS DMFS OS
favorably to historical controls. Clinical trial information: NCT03013491.
Ipi Pbo Ipi Pbo Ipi Pbo Best Available Response SCCA cSCC SBA TNBC UPS
No. of events 273 323 247 292 173 223 (efficacy evaluable* pts), n (n = 3) (n = 3) (n = 4) (n = 2) (n = 16)
5-year rate 43.9% 32.5% 49.9% 39.8% 65.2% 54.1%
7-year rate 39.2% 30.9% 44.5% 36.9% 60.0% 51.3% Confirmed or unconfirmed partial response 0 1 0 2 1
Median (yrs) 2.7 1.5 5.0 2.4 NR 7.8 Stable disease 1 1 2 0 3
HR (95% CI)† 0.75 (0.63-0.88) 0.76 (0.64-0.90) 0.73 (0.60-0.89)
Log-rank p-value† 0.0004 0.0018 0.0021 Progressive disease 2 1 2 0 12
HR: Hazard Ratio provided by the Cox model; CI: Confidence interval; NR: not reached; †stratified by stage provided at randomization. *Required postbaseline assessment.

2514 Poster Discussion Session; Displayed in Poster Session (Board #158), 2515 Poster Discussion Session; Displayed in Poster Session (Board #159),
Sat, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Sat, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Sat, 1:15 PM-2:45 PM Sat, 1:15 PM-2:45 PM
A phase I study of ALX148, a CD47 blocker, in combination with established Distinct immunogenomic properties of melanomas with stable disease as
anticancer antibodies in patients with advanced malignancy. First Author: best response to immune checkpoint blockade (ICB). First Author: Natalie
Laura Quan Man Chow, Division of Medical Oncology, University of Wash- Vokes, Dana-Farber Cancer Institute, Boston, MA
ington, Seattle, WA
Background: ICB has improved survival in melanoma. Patients with stable
Background: CD47 is a myeloid checkpoint upregulated by tumor cells to disease (SD) as best treatment response represent an intermediate response
evade the host’s immune response. ALX148 (A) is a fusion protein comprised phenotype whose biology has been incompletely characterized. Methods: Whole
of a high affinity CD47 blocker linked to an inactive human immunoglobulin exome and transcriptome sequencing from pre-treatment tumors in melanoma
Fc region. ALX148 is well tolerated in combination with pembrolizumab (P) or patients treated with ICB (anti-CTLA-4 and/or anti-PD-1) were assembled
trastuzumab (T) with no maximum tolerated dose (MTD) identified (ASCO and uniformly analyzed (WES n = 293; WES+RNA-seq n = 159). RECIST
2018 #3068, SITC 2018 #P335). Safety and antitumor activity of ALX148 (v1.1) was used to determine complete or partial response (CR/PR; n = 94),
(10 mg/kg QW) in combination with T or P are reported in patients (pts) SD (n = 42), or progressive disease (PD; n = 157). Gene set enrichment
including those with anti-HER2 or checkpoint inhibitor (CPI) relapsed/ analysis (GSEA) was performed on 50 “hallmark” gene sets to identify
refractory diseases. Methods: Patients with malignancy including non-small pathways differentially expressed in patients with SD. CIBERSORT was
cell lung cancer (NSCLC: CPI resistant/refractory or PD-L1 tumor proportion used to infer relative proportions of 22 immune cell types in each sample.
score ,50%) and head and neck squamous cell carcinoma (HNSCC: pro- Mutation antigenicity was determined by calculating patient-specific
gressed on platinum therapy) received A+P. Patients with HER2 malignancy mutation affinity for MHC class I peptides. Results: GSEA identified en-
including gastric/gastroesophageal junction (GEJ) cancer (progressed on T + richment of multiple immune-related gene sets in SD tumors, including
fluoropyrimidine-based therapy) received A+T. Safety, response, pharmaco- TNF-a signaling and interferon-ɣ response (FDR q , 0.1, SD vs CR/PR and
kinetic (PK), and pharmacodynamic (PD) markers were assessed. Preliminary SD vs PD). SD tumors had higher HLA and antigen presentation pathway
data from fully enrolled cohorts are reported as of 20 Jan 2019 (safety)/28 Jan expression, and increased cytolytic T cell activity compared to CR/PR and
2019 (efficacy). Results: Seventy-nine pts received A+P (All, n=50; NSCLC, PD. CIBERSORT analysis identified higher total immune infiltrate in SD
n=23; HNSCC, n=20) or A+T (All, n=29; Gastric/GEJ, n=23) as of data cutoff. patients compared to CR/PR and PD (Mann-Whitney U p = 0.03 and p ,
Forty-seven pts reported mostly low grade treatment related adverse events. The 0.001, respectively) but not in patients with CR/PR vs PD (p = 0.124).
most common were fatigue (11%), AST increase (9%), ALT increase (8%), However, checkpoint expression, including PD-1, PD-L1, and LAG3, was
anemia (8%), and platelets decreased (6%). In select tumor histologies, an- also higher in SD patients. Mutation load did not differ between SD and CR/
ticancer activity was observed in initial response-evaluable pts [NSCLC (n=23) PR or PD patients (SD median 2.87 vs CR/PR median 7.98, Mann-Whitney
1PR, 8SD; HNSCC (n=17) 3PR, 4SD; and Gastric/GEJ (n=21) 4PR, 6SD]. U p = 0.104; PD median 3.42, p = 0.210). However, SD patients had more
Preliminary results indicate favorable ALX148 PK and CD47 target occupancy antigenic passenger mutations (SD vs CR/PR, p = 0.001; vs PD, p ,
profiles, and positive effects on tumor infiltrating immune cells. Results will be 0.001); there was no difference in antigenicity of driver mutations.
updated at presentation. Conclusions: ALX148 demonstrates excellent toler- Conclusions: Pre-treatment melanomas from patients with SD contain
ability with favorable PK/PD characteristics to date. Objective responses were more antigenic passenger mutations and demonstrate a global increase in
observed in patients with late line NSCLC, HNSCC, and Gastric/GEJ, including immune signaling. This may describe a subset of patients with pre-existing
disease relapsed/refractory to prior CPI and HER2-targeted therapies. Clinical dysfunctional immune response that is minimally responsive to ICB.
trial information: NCT03013218. Further characterization of the tumor-immune interaction in these patients
may inform improved interventions.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
110s Developmental Immunotherapy and Tumor Immunobiology

2516 Poster Discussion Session; Displayed in Poster Session (Board #160), 2517 Poster Discussion Session; Displayed in Poster Session (Board #161),
Sat, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Sat, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Sat, 1:15 PM-2:45 PM Sat, 1:15 PM-2:45 PM
Analysis of early mortality in randomized clinical trials evaluating anti-PD-1/ Prognostic and predictive value of an immune-related adverse event among
PD-L1 antibodies: A systematic analysis by the United States Food and Drug stage III melanoma patients included in the EORTC 1325/KEYNOTE-054
Administration (FDA). First Author: Flora Mulkey, U.S. Food and Drug pembrolizumab versus placebo trial. First Author: Alexander M. M. Egger-
Administration, Silver Spring, MD mont, Gustave Roussy Cancer Centre and University Paris-Saclay, Paris,
France
Background: Many studies exhibit what seems to be dis-proportionately
higher early mortality (EM) in anti-PD-1/PD-L1 containing arms (IO) when Background: Several studies suggested that patients (pts) with an immune-
compared to active control arms (AC), resulting in early crossing of the related adverse event (irAE) during immunotherapy have better outcomes
Kaplan-Meier overall survival curves. We examine if EM with the use of IO is than those without. It remains uncertain whether these observations can be
specific to certain demographic and disease characteristics. Methods: Data explained by guarantee-time bias or the role of irAE as an indicator of drug
from 16 randomized AC trials submitted to FDA containing 6055 IO and activity. Here, we investigated the association between irAEs and recurrence-
3604 AC patients in HNSCC, Melanoma, NSCLC, RCC, and Urothelial free survival (RFS) in the double-blind EORTC 1325/KEYNOTE-054 trial that
Carcinoma were evaluated for signs of EM. Study-specific and pooled compared pembrolizumab and placebo in high-risk stage III melanoma pts.
piecewise hazard ratios (HRs) were used to quantify EM from 0 to 60 and . Methods: Eligible pts included adults with complete resection of cutaneous
60 days. Additionally, HRs up to 60 days were used to assess the extent melanoma metastatic to lymph node(s), classified as stage IIIA (lymph node
specific subgroups account for EM. Results: Piecewise HRs comparing OS metastasis . 1 mm), IIIB or IIIC (without in-transit metastasis) and with no
between IO and AC changed direction, . 1 to , 1 in 11 trials; melanoma (5/ active autoimmune disease that required systemic treatment in past 2 years.
6), NSCLC (3/7), HNSCC (1/1), RCC (1/1), and urothelial cancer (1/1). When We used a Cox model adjusted for sex, age, and stage with a time-varying
pooled, NSCLC studies retained this EM pattern, although attenuated, with covariate taking a value zero before the irAE onset and a value one afterwards
HR (95% CI) of 1.12 (0.91, 1.38) #60 days and 0.66 (0.61, 0.72) after to estimate the association between the occurrence of irAEs and RFS.
60 days. This was not observed in the pooled melanoma studies: 0.88 (0.63, Results: Consistent with the main analysis in the ITT population (n = 1019,
1.24) # 60 days and 0.59 (0.53, 0.67) after 60 days. EM in both arms was Eggermont et al, NEJM, 2018), RFS was longer in the pembrolizumab than in
associated with poor ECOG performance status (PS), increased LDH, and the placebo arm (HR = 0.56, 98.4% CI: 0.43-074) among pts who started the
high tumor burden. Comparing EM patients in the IO and AC arms, a larger treatment (n = 1011). The incidence of irAE on study was 37.3% in the
proportion were female in the melanoma trials (41% vs. 28%), a smaller pembrolizumab (n = 509) and 9.0% in the placebo arm (n = 502) and, in each
proportion had squamous histology in the NSCLC trials (32% vs. 41%), and a treatment group, it was similar in males and females. The occurrence of an
larger proportion were PD-L1 negative (56% vs. 36% melanoma; 60% vs. irAE was significantly associated with a longer RFS in the pembrolizumab arm
43% NSCLC). Analysis of the pooled melanoma studies suggests PD-L1 (HR = 0.61, 95% CI: 0.39-0.95, P = 0.03). This was true for both males and
negative melanoma patients with high baseline tumor burden and PS females. However, in the placebo arm, no association was observed (HR =
played a role in EM with HR before 60 days of 1.49 (0.75, 2.97). However, 1.39, 95% CI: 0.83-2.32, P = 0.21). Compared to the placebo arm, the
these results were not reproducible in NSCLC. Conclusions: Potential risk reduction in the hazard of recurrence or death in the pembrolizumab arm was
factors for EM were assessed in individual and pooled trials. While several greater (P = 0.028) after an onset of an irAE (HR = 0.37, 95% CI: 0.24-0.57)
factors—negative PD-1/PD-L1 status and high ECOG, LDH and tumor than without/before an irAE (HR = 0.61, 95% CI: 0.49-0.77). Conclusions: In
burden—seem to play a role in EM, these high-risk subgroups do not fully the EORTC 1325/KEYNOTE-054 study conducted in high-risk stage III
explain the EM patterns observed in the IO treated patients. melanoma pts, the occurrence of an irAE was strongly associated with a longer
RFS in those treated with pembrolizumab, but not with placebo. Clinical trial
information: NCT02362594.

2518 Poster Discussion Session; Displayed in Poster Session (Board #162), 2520 Poster Discussion Session; Displayed in Poster Session (Board #164),
Sat, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Sat, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Sat, 1:15 PM-2:45 PM Sat, 1:15 PM-2:45 PM
Safety and efficacy of cryopreserved autologous tumor infiltrating lymphocyte Impact of bridging chemotherapy on clinical outcome of CD19 CAR T therapy
therapy (LN-144, lifileucel) in advanced metastatic melanoma patients who in adult ALL. First Author: Karlo Perica, Memorial Sloan Kettering, New York,
progressed on multiple prior therapies including anti-PD-1. First Author: Amod NY
Sarnaik, H. Lee Moffitt Cancer Center, Tampa, FL
Background: Autologous chimeric antigen receptor (CAR) T cell therapy has
Background: Treatment options are limited for patients with advanced mel- shown to be effective in CD19+ relapsed or refractory (R/R) B-ALL but
anoma who have progressed on checkpoint inhibitors and targeted therapies requires a 2-4 week period of cell processing and manufacture. During this
such as BRAF/MEK inhibitors (if BRAF-V600E mutated). Adoptive cell “bridging period,” patients are vulnerable to disease progression and
therapy utilizing tumor-infiltrating lymphocytes (TIL) has shown antitumor complications. We sought to characterize bridging strategies in our pub-
efficacy with durable long-term responses in heavily pretreated melanoma lished study of CD19 CAR T cell therapy in adults with R/R ALL (Park et al.,
patients. Safety and efficacy of lifileucel (LN-144), a centrally manufactured NEJM, 2018). Methods: We performed a retrospective review of adult pa-
autologous TIL therapy are presented. Methods: C-144-01 is a global Phase 2 tients with R/R ALL treated with 19-28z CAR T therapy at MSKCC. Bridging
open-label, multicenter study of the efficacy and safety of lifileucel in patients therapy was defined as any therapy given from trial enrollment to cell infusion
with unresectable metastatic melanoma. We report on Cohort 2 (N = 55) and classified as either high intensity (remission-inducing or myelosup-
patients who received cryopreserved lifileucel. Tumors resected at local in- pressive regimens, eg hyper-CVAD or high-dose cytarabine based regimens)
stitutions were processed in central GMP facilities for TIL production in a 22- or low intensity (maintenance and/or less myelosuppressive regimens, eg
day process. Final TIL infusion product was cryopreserved and shipped to POMP, Blinatumomab, TKI). Results: Of 53 patients who received CAR
sites. Patients received one week of cyclophosphamide/fludarabine pre- T cell infusion, 19 were bridged with a high intensity regimen and 34 with a
conditioning lymphodepletion, a single lifileucel infusion, followed by up to 6 low intensity regimen. There was no difference in number of prior therapies,
doses of IL-2. Results: In 55 patients with Stage IIIC/IV unresectable mel- pre-bridging chemotherapy disease burden, and prior transplant status
anoma, 3.1 mean prior therapies (anti-PD1 100%; anti-CTLA-4 80%; BRAF/ between groups. High intensity therapy was associated with a higher rate of
MEK inhibitor 24%), high baseline tumor burden (110 mm mean target lesion Gr3-4 infectious complications during the bridging period (78% vs 32%, p ,
sum of diameters), ORR was 38% (2 CR, 18 PR, 1 uPR). Of 21 responders, 4 0.002 by Fisher’s Exact) but not with response to bridging or CAR T cell
have progressed to date with median follow up of 7.4 months. Overall disease therapy, relapse free survival, post-CAR Gr3-4 cytokine release syndrome
control was 76%. Improved responses in some patients were observed with (CRS) or neurotoxicity (NT). Patients in both groups who converted from
longer follow up. Most (54) patients progressed on prior anti-PD1 and those morphologic to molecular disease during bridging (n=9) had a decreased rate
with PD-L1 negative status (TPS , 5%) were among responders. Mean cells of eventual severe CRS (0% vs 41%, p=0.01) or NT (0% vs 55%, p ,0.01)
infused was 28 x109. Median IL-2 doses administered was 6.0. Adverse compared to patients with persistent morphologic disease. In all patients
events resolved to baseline, 2 weeks post TIL infusion, a potentially important enrolled on trial (n=83), use of high compared to low intensity bridging was
benefit of one-time TIL therapy. Conclusions: Lifileucel treatment results in not associated with higher rates of successful CAR T cell infusion (63% vs
38% ORR in heavily pretreated metastatic melanoma patients with high 79%, p.0.05) or a combined endpoint of CAR T cell infusion or alternative
baseline disease burden who received prior anti-PD1 and BRAF/MEK inhibitor if therapy including transplant (80% vs 86%, p. 0.05). Conclusions: High
BRAF mutated. Based on these data, a new Cohort 4 in C-144-01 has been initiated intensity bridging therapy is associated with a high risk of infectious com-
to support lifileucel registration. Clinical trial information: NCT02360579. plications without a clear benefit in outcome in R/R ALL receiving CD19 CAR
T cells. Selection of bridging regimen therefore requires consideration of
previous treatments and patient status to maximize the efficacy and safety.
Clinical trial information: NCT01044069.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Immunotherapy and Tumor Immunobiology 111s

2521 Poster Discussion Session; Displayed in Poster Session (Board #165), 2522 Poster Discussion Session; Displayed in Poster Session (Board #166),
Sat, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Sat, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Sat, 1:15 PM-2:45 PM Sat, 1:15 PM-2:45 PM
First-in-human first-in-class phase I trial of murlentamab, an anti-Mullerian- Regorafenib plus nivolumab in patients with advanced gastric (GC) or
hormone receptor II (AMHRII) monoclonal antibody acting through tumor- colorectal cancer (CRC): An open-label, dose-finding, and dose-expansion
associated macrophage (TAM) engagement, as single agent and in combination phase 1b trial (REGONIVO, EPOC1603). First Author: Shota Fukuoka,
with carboplatin (C) and paclitaxel (P) in AMHRII-expressing advanced/ Department of Gastroenterology and Gastrointestinal Oncology, National
metastatic gynecological cancer patients (pts). First Author: Alexandra Cancer Center Hospital East, Kashiwa, Japan
Leary, Gustave-Roussy Cancer Campus, Villejuif, and Groupe d’Investiga-
Background: Immune suppressive cells such as regulatory T cells (Tregs) or
teurs Nationaux pour l’Etude des Cancers Ovariens, France
tumor-associated macrophages (TAMs) may contribute to resistance to anti-
Background: Membranous expression of AMHRII is found in ~70% of gy- PD-1/PD-L1 inhibitors (A-PD1). Regorafenib, a potent inhibitor of angiogenic
necological tumors. Murlentamab (M) binds with high affinity both AMHRII and oncogenic kinases, reduced TAMs in tumor models. The combination of
(at cell membrane) and CD16 (on macrophage, via its low fucose Fc). M re- regorafenib plus A-PD1 exhibited superior tumor growth suppression com-
programs TAMs, restoring their antitumoral functions (phagocytosis) resulting in pared to either treatment alone in murine models. Methods: In this study, we
cytotoxic T cell reactivation. Methods: Pts with advanced/metastatic AMHRII- enrolled patients (pts) with previously treated, advanced GC or CRC. The pts
expressing ovarian, cervical or endometrial cancer with measurable disease and received regorafenib plus nivolumab in a dose-finding phase to estimate the
performance status # 1 received M as single agent (SA) in 8 dose escalating and maximum tolerated dose (MTD). Additional pts were enrolled in a dose-
2 expansion cohorts. Combination with CP was studied in 2 escalating cohorts. expansion phase to further establish the safety and determine the pre-
Safety, recommended dose determination, antitumor activity, pharmacody- liminary efficacy. Regorafenib of 80 to 160 mg was administered once daily
namics (PD) effects (circulating immune cells and tumor microenvironment for 21 on 7 days off with intravenous nivolumab 3 mg/kg every 2 weeks. The
(TME) from paired biopsies) were assessed. Results: 68 heavily pretreated primary endpoint was dose-limiting toxicity (DLT) during cycle one (4 weeks)
(median 4 prior lines) pts received M for 0.5 to 11 months (mo) (59 pts M SA to estimate the MTD and the recommended dose. Results: Fifty pts were
and 9 pts M + CP). No dose limiting toxicity was reported. Most common toxicity enrolled (25 GC; 25 CRC) until October 2018. The median prior treatment line
was G1-2 asthenia (29 %). Eight pts (12%) had G $ 3 reversible toxicities was 3 (range 2-8). During dose-escalation, 3 DLTs were observed with
(asthenia, nausea/vomiting, anorexia, arthralgia). No antidrug antibody was regorafenib 160 mg, including grade (G) 3 maculopapular rash, mucositis and
detected. One partial response (PR) was achieved with M SA in a granulosa pt. proteinuria, while there was no DLT with 80 or 120 mg. In the dose expansion
In CP combination, 4/9 pts (44%) responded to treatment (1 Complete Re- cohort with regorafenib 120 mg, the dose was reduced to 80 mg owing to
sponse and 3 PRs). Overall, 22/67 (33%) pts were progression-free at 4 mos. frequent G3 skin toxicities. Grade $ 3 treatment related adverse events
Among 17 pts treated $ 6 mos, 6/9 (67%) granulosa pts with M SA and 4/5 occurred in 17 pts; the common events ( . 5%) being rash (14%), palmar-
(80%) endometrium and cervix with CP combination had a longer PFS than plantar erythrodysesthesia (10%), and proteinuria (8%). Objective tumor
under previous regimen. PD blood assessment of 25 pts treated with M SA response was observed in 19 pts (38%) including 11 MSS GC, 7 MSS CRC and
showed an increase in classical monocytes, and T cells and neutrophils acti- 1 MSI-H CRC for response rates of 44% in GC and 29% in MSS CRC. Three of
vation. Changes in TME under M will be presented. Conclusions: Murlentamab the 7 A-PD1 pretreated GC pts achieved a partial response. The pre- and post-
was very well tolerated, demonstrated immune PD effects and showed hints of treatment tumor samples showed a reduction of FoxP3hiCD45RA-Tregs
antitumor activity. These results together with its innovative immunological fraction at the tumor response. Conclusions: The combination of regorafenib
mode of action support development of M in AMHRII-expressing cancers, in 80mg plus nivolumab had a manageable safety profiles and encouraging anti-
combination with chemotherapy or other immune oncology drugs. Clinical trial tumor activity in MSS GC and CRC pts, which warrants further investigations
information: NCT02978755. in a larger cohort. Updated biomarker analysis will be presented. Clinical trial
information: NCT03406871.

2523 Poster Discussion Session; Displayed in Poster Session (Board #167), 2524 Poster Session (Board #168), Sat, 8:00 AM-11:00 AM
Sat, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Targeting MHC-linked wild type p53 with TCR mimic single chain diabody
Sat, 1:15 PM-2:45 PM for cancer immunotherapy. First Author: Suman Paul, Ludwig Center and
A phase I multicenter study to assess the safety, tolerability, and immuno- Howard Hughes Medical Institute, Sidney Kimmel Comprehensive Cancer
genicity of mRNA-4157 alone in patients with resected solid tumors and in Center, Johns Hopkins University School of Medicine, Baltimore, MD
combination with pembrolizumab in patients with unresectable solid
Background: Increased tumor suppressor protein p53 expression is observed
tumors. First Author: Howard A. Burris, Sarah Cannon Research Institute,
in a wide range of human cancers. As a result there is intense interest in
Nashville, TN
targeting p53 for cancer therapy. Intracellular p53 is inaccessible to
Background: T-cell targeting of mutation-derived epitopes (neoantigens) has therapeutic antibodies that bind cell surface proteins. However, intracellular
been demonstrated to drive anti-tumor responses. Immunizing patients against proteins including p53 are degraded into peptides that are presented on cell
such neoantigens in combination with a checkpoint inhibitor (CPI) may elicit surface in association with HLA class I molecules. Thus p53 peptide-HLA
greater anti-tumor responses than CPI alone. Mutations are rarely shared be- (p53-HLA) complexes can be antibody targets. Methods: Using phage
tween patients, thus requiring a personalized approach to vaccine design. display we identified a novel anti-p53-HLA single chain variable fragment
Methods: A phase I dose escalation study of mRNA-4157 as adjuvant mono- (scFv) clone-43 that recognizes a wild-type p53 10-mer epitope bound to
therapy in patients with resected solid tumors (melanoma, bladder carcinoma, HLA-A*2402. By coupling our clone-43 scFv with an anti-CD3 scFv, we
HPV negative HNSCC, NSCLC, SCLC, MSI-High, or TMB High cancers) and in generated a single chain diabody (scDb) designed to activate T-cells against
combination with pembrolizumab in patients with advanced or metastatic cancer p53-expressing target cells. Results: In-vitro co-culture of clone-43 scDb
is being conducted to evaluate safety. mRNA-4157 is a lipid encapsulated
with donor human T-cells and p53 expressing SIG-M5 cancer cells results in
personalized vaccine encoding multiple neoantigens selected using a proprietary
SIG-M5 cell killing and concomitant T-cell interferon gamma (IFNg) release.
algorithm designed to induce neoantigen specific T cells and associated anti-
tumor responses. Patients may receive up to 9 cycles (Q3W) of mRNA-4157 by
In contrast, similar co-culture with SIG-M5 p53-knock out (KO) cells showed
intramuscular injection (0.04 – 1 mg). In the combination arm, pembrolizumab no cell killing and minimal IFNg release demonstrating specificity of clone-
(200 mg) is administered for two cycles prior to combination with mRNA-4157; 43 to p53 expressing cells. Additionally, in-vivo growth of p53 expressing
patients may continue pembrolizumab after completion of 9 cycles of combination SW480 cancer cell xenografts in NSG mice was completely terminated by
therapy. Primary end points include safety, tolerability, and recommended phase clone-43 scDb injections. A major concern for wild-type p53 epitope tar-
2 dose. Results: 33 patients received mRNA-4157; 13 as monotherapy and 20 in geting is potential on-target off-tumor effect on non-cancerous tissue. We
combination with pembrolizumab. No DLTs were reported, and treatment related observed significant in-vitro clone-43 scDb mediated killing of human HLA-
AEs have generally been of low grade and reversible, and no drug related SAEs or A*24:02 peripheral blood mononuclear cells. To better evaluate effect of
AEs $ grade 3 have been observed. Of the 13 patients on adjuvant monotherapy clone-43 scDb on non-neoplastic human cells, we engrafted HLA-A*24:02
(3 melanoma, 8 NSCLC, 2 MSI-High), 12 patients remain disease free on study, human CD34+ hematopoietic stem cells into NSG mice to generate a hu-
median follow-up of 8 months. 20 patients have been treated in combination manized mouse model with circulating mature human CD45+ cells. Clone-
(1 TMB-high, 4 bladder, 2 HNSCC, 1 melanoma, 7 NSCLC, 2 SCLC, 3 MSI-high), 43 scDb treatment resulted in selective depletion of circulating human cells
12 had progressed on prior CPI, 16 have been restaged and there are 1 CR (on while the same cells persisted in mice treated with unrelated control scDb.
pembrolizumab prior to vaccination), 2 PR, 5 SD for at least 5 combination cycles, Conclusions: Our observation that immune targeting of wild-type p53 epi-
5 PD, 2 iuPD, and 1 patient is non-evaluable for response but remains on study. tope results in significant off-tumor hematopoietic cell death is contrary to
Neoantigen specific T cell responses have been detected by IFN-g ELISpot from previously published reports and carries important implications for future
PBMCs. Conclusions: mRNA-4157 is safe and well tolerated at all dose levels anti-p53 antibody and vaccine design for cancer immunotherapy.
tested. Clinical responses have been observed in combination with pem-
brolizumab and neoantigen-specific T cells have been induced, supporting
the advancement of mRNA-4157 to phase 2. Clinical trial information:
NCT03313778.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
112s Developmental Immunotherapy and Tumor Immunobiology

2525 Poster Session (Board #169), Sat, 8:00 AM-11:00 AM 2526 Poster Session (Board #170), Sat, 8:00 AM-11:00 AM
First-in-human, dose-escalation, phase (ph) I trial to evaluate safety of anti-Axl A phase Ia/Ib trial of the anti-PD-L1 human monoclonal antibody (mAb),
antibody-drug conjugate (ADC) enapotamab vedotin (EnaV) in solid tumors. CS1001, in patients (pts) with advanced solid tumors or lymphomas. First
First Author: Malaka Ameratunga, The Alfred Hospital, Melbourne, Australia Author: Lin Shen, Beijing Cancer Hospital, Beijing, China
Background: Axl, a transmembrane receptor tyrosine kinase, is aberrantly Background: CS1001 is the first full-length, fully human anti-PD-L1 mAb
overexpressed in various human cancers and associated with poor prognosis developed by the OMT transgenic rat platform, which mirrors natural IgG4
and treatment resistance. EnaV, a novel ADC of anti-Axl human IgG1 and human antibody with expected PK profiles, and may potentially reduce the risk
monomethyl auristatin E, demonstrated potent anti-tumor activity in xe- of immunogenicity and toxicity in pts. This first-in-human Phase Ia/Ib study of
nograft models. Methods: In a ph1 trial (NCT02988817), patients (pts) with CS1001 was conducted to evaluate the safety, tolerability, PK profile, and anti-
relapsed/refractory cancer received single agent EnaV, 0.3–2.8 mg/kg once tumor activity of CS1001 in pts with advanced solid tumors or lymphomas.
every 3 wks (1Q3W) or 0.45–1.4 mg/kg 3 times over 4 wks (3Q4W). End- Methods: Pts with advanced solid tumors or lymphomas were enrolled in the
points included dose-limiting toxicities (DLTs), adverse events (AEs) and dose escalation Phase Ia, receiving CS1001, Q3W, IV, at escalating doses from
pharmacokinetics (PK). DLTs were classed as hematological (e.g. Grade [G] 3, to 10, 20, 40 mg/kg and 1200 mg. Dose escalation was aided by a 3+3 dose
3/4 febrile neutropenia; G4 neutropenia or anemia) or non-hematological escalation scheme. DLT was evaluated within 3 weeks after the initial dose. Pts
(e.g. severe skin toxicities; G3/4 neuropathy or infusion reactions; $G3 with various tumor types were enrolled in the dose expansion Phase Ib to assess
treatment-related AEs in first treatment cycle). Upon determining maximum anti-tumor activity and safety, including NSCLC, esophageal carcinoma, GC,
tolerated dose (MTD) per arm and recommended ph2 dose (RP2D), ph2a HCC, cholangiocarcinoma, etc. Safety was assessed by monitoring AEs and the
(dose expansion) will enroll #297 pre-treated pts with advanced/metastatic associated grades per NCI CTCAE v4.03, tumor assessed per RECIST v 1.1
cancer in 7 cohorts. Results: 47 pts with NSCLC (n=8), melanoma (n=9), (solid tumors) or Lugano 2014 (lymphomas). Results: As of 30 Nov 2018, 29
ovarian (n=22), cervical (n=3) and endometrial (n=5) cancer enrolled in ph1 pts, median age of 53 (23-75) yrs, were enrolled in Phase Ia, 3 mg/kg (N = 3);
(1Q3W n=32; 3Q4W n=15). Most pts were female (87%), White (94%) and 10 mg/kg (4); 20 mg/kg (3); 40 mg/kg (3) and 1200 mg flat dose (16). A total of
aged ,65 y (66%). MTD was 2.2 mg/kg in 1Q3W arm and 1.0 mg/kg in 20 pts discontinued treatment due to disease progression (14), death (2),
3Q4W arm; RP2D was 2.2 mg/kg 1Q3W. EnaV median elimination half-life: withdrawal by pts (2) and AEs (2; Grade [G] 4 hepatic function abnormal and
0.9–2.2 d across doses/schedules. In 47 enrolled pts, there were 6 DLTs G3 pulmonary tuberculosis, both were not related to treatment). 9 pts remain
(Table). Most common AEs (any G; $40% pts) were fatigue (64%), nausea on treatment. Median treatment duration was 126 (21-408+) days. No DLTs
(57%), constipation (57%), diarrhea (47%), vomiting (45%) and decreased were observed. 27 of 29 pts had TRAEs with the most frequent TRAEs in-
appetite (43%). 3 pts (1Q3W arm) had partial response (1 NSCLC [2.2 mg/kg cluding anaemia (14), proteinuria (13) and blood bilirubin increased (8). G3
dose]; 2 ovarian [1.5 and 2.4 mg/kg dose levels]). Conclusions: The RP2D of TRAEs include anaemia (2) and platelet count decreased (1). SAEs were
single agent EnaV in pre-treated pts with solid tumors was 2.2 mg/kg 1Q3W. reported in 6 pts and they were TRAEs. Three G4 AEs were reported: anaemia
EnaV had encouraging preliminary anti-tumor activity and will be evaluated (1), hypokalaemia (1) and hepatic function abnormal (1), they were not TRAEs
in 7 ph2a expansion cohorts to further assess safety, tolerability, PK, anti- as determined by the investigators. irAEs occurred in 7 pts (24%). Among the
tumor activity and Axl expression. Funding: Genmab A/S. Clinical trial in- 29 evaluable pts, 7 pts had PR and 8 had SD, mDoR was not reached. In Phase
formation: NCT02988817. Ib, 97 pts were enrolled, with 65 pts on treatment and 32 pts discontinued
DLT Dose, mg/kg (n)
from treatment. The most frequent reason on the discontinuation was disease
1Q3W
progression (21). Phase Ib enrollment is still ongoing. Conclusions: CS1001 is
Constipation 2.0 (1); 2.2 (1) well tolerated without DLT across tested dose levels. Evidence of anti-tumor
Vomiting 2.2 (1)
g-glutamyltransferase increase 2.4 (1) activities was observed. Currently, 1200 mg flat dose Q3W is being explored in
3Q4W
Febrile neutropenia 1.2 (1)
various tumor types in Phase Ib, and safety and efficacy results will be dis-
Diarrhea 1.2 (1) played in the presentation. Clinical trial information: NCT03312842.

2527 Poster Session (Board #171), Sat, 8:00 AM-11:00 AM 2528 Poster Session (Board #172), Sat, 8:00 AM-11:00 AM
A phase I study to assess safety, pharmacokinetics (PK), and pharmacody- Ramucirumab (Ram) and durvalumab (Durva) treatment of metastatic non-
namics (PD) of JNJ-64457107, a CD40 agonistic monoclonal antibody, in small cell lung cancer (NSCLC), gastric/gastroesophageal junction (G/GEJ)
patients (pts) with advanced solid tumors. First Author: Emiliano Calvo, adenocarcinoma, and hepatocellular carcinoma (HCC) following progression
START Madrid, Centro Integral Oncológico Clara Campal, Madrid, Spain on systemic treatment(s). First Author: Yung-Jue Bang, Seoul National
University College of Medicine, Seoul, South Korea
Background: JNJ-64457107 (JNJ-107) is an agonistic human monoclonal
(IgG1) antibody targeting CD40, a novel target for anti-tumor immunotherapy Background: A Phase 1b study (NCT02572687) was conducted to examine the
with a central role in adaptive and innate immunity. Methods: JNJ-107 was combined effects of Ram (anti VEGFR2) and Durva (anti PD-L1). Methods: Patients
administered intravenously Q2W in treatment cycles of 28 days. Dose esca- (pts) with previously-treated, advanced NSCLC (Cohort [CH] A), G/GEJ ade-
lation was pursued with (w) and without (wo) pre-infusion steroids for mitigation nocarcinoma (CH B), HCC (CH C), ECOG PS 0-1, and no prior Ram or IO
of infusion related reactions (IRRs). Dose-limiting toxicity (DLT), safety, PK, PD therapy, received Ram (10 mg/kg) + Durva (1125 mg) Q3W (CH A) or Ram (8
and antitumor activity were evaluated. Results: 95 pts of age 18-80 years mg/kg) + Durva (750 mg) Q2W (CH B, C). Primary objective was to assess
(median 59) were enrolled in 7 cohorts (n = 50, 75mg/kg – 2000 mg/kg) w/ safety; efficacy was also examined. PD-L1 expression of tumor cells (TC) +/-
steroids and 5 cohorts (n = 45, 75 mg/kg – 1200 mg/kg) wo/steroids and re- immune cells (IC) in pretreatment tumor biopsies were assessed using SP263
ceived 1-26 (median 3) cycles of JNJ-107. Two DLTs occurred: Grade (G) 3 immunohistochemistry. “High” PD-L1 is $25% TC for NSCLC and $25% TC
headache lasting 5 days at 1200 mg/kg w/steroids and G3 ALT/AST + G2 or IC for G/GEJ, HCC. Results: CH A, B and C enrolled pts with ECOG PS 1 (%)
bilirubin increase at 1200 mg/kg wo/steroids. Most frequent adverse events of 43, 66, 68; and average of 2, 2, 1 prior regimens, respectively. The most
($20%) were pyrexia (41%), fatigue (39%), pruritus (39%), headache (26%), common grade 3/4 treatment-emergent adverse events (AE) are hypertension
chills (26%), nausea (22%) and rash (21%). IRRs were reported in 51% of pts (HTN) (14.3, 17.2, 17.9%), anemia (3.6, 24.1, 21.4%), and fatigue (10.7,
(G1-G2 50%, G3 1%). Most commonly reported IRRs ($10%) were pruritus 10.3, 10.7%). Grade 3/4 AEs of special interest ( . 5% total pts) for Ram:
(31%), rash (15%), chills (13%) and flushing (12%). Cetirizine and Mon- HTN, bleeding events (3.6, 10.3, 10.7%), Venous thromboembolic events (0,
telukast premedication during dose escalation wo/steroids significantly re- 10.3, 7.1%); for Durva: increase in lipase (10.7, 3.4, 10.6%) and AST (3.6,
duced IRRs. Preliminary PK of JNJ-107 suggest target mediated drug 3.4, 17.9 %). Data from CH B,C suggest a trend toward increased efficacy in
disposition with rapid decline in serum concentrations (half-life of ~13h at 600 pts with high PD-L1 expressing tumors. Conclusions: Ram + Durva generated
mg/kg and ~ 24h at doses $1200 mg/kg). Dose proportional increase in Cmax no unexpected toxicities and demonstrated antitumor activity. Results in pts
and AUC(last) was observed at doses $1200 mg/kg and more than dose with high PD-L1 HCC and G/GEJ cancer warrant further evaluation. Clinical
proportional increase was seen at lower doses. PD of JNJ-107 demonstrated trial information: NCT02572687.
dose-independent reduction in peripheral blood B cells recovering in a dose-
(A) (B) (B) (B) (C) (C) (C)
dependent manner. MCP-1, IP-10, MIP-1a, and MIP-1b cytokines showed (A) NSCLC NSCLC (A) NSCLC G/GEJ G/GEJ G/GEJ HCC HCC HCC
high-level responses with cytokine/chemokine secretion kinetics consistent, Total High Low Total High Low Total High Low
PD-L1 subgroup N = 28 n=5 n = 17 N = 29 n = 14 n = 12 N = 28 n = 11 n = 15
but not confirmatory, for activation of antigen presenting cells. IFNg, TNF, and
ORR, n (%) 3 (11) 1 (20) 2 (12) 6 (21) 5 (36) 0 3 (11) 2 (18) 0
IL-8 showed moderate responses reflecting downstream T cell activation. Other Disease control rate 16 (57) 3 (60) 9 (53) 16 (55) 10 (71) 4 (33) 17 (61) 8 (73) 7 (47)
tested cytokines were at low levels. A partial response lasting 9.2 months (ORR+SD rate), n
(%)
was observed in 1 pt with renal cell cancer. 13 (14%) and 10 (11%) pts Median duration of - (16.6, -) 15.4 (3.25, - (5.6, -)
showed stable disease lasting $3 months and $ 6 months, respectively. response, mo (95%
CI)
-)

Conclusions: The CD40 agonist JNJ-107 has a manageable safety profile Median PFS, 2.7 4.1 2.6 2.6 5.5 1.5 4.4 5.6 2.8
mo (95% CI) (1.6, 5.8) (1.4, -) (1.3, (1.5, 7.1) (1.8, (1.4, (1.6,5.7) (1.5, -) (0.7,
with favorable PK and PD properties. Future clinical development will 5.8) 6.8) 2.6) 5.5)
require combination with either chemotherapy, or other immunotherapies Median OS, 11.0 16.4 7.5 12.4 14.8 5.5 10.7 16.5 5.7
mo (95% CI) (6.2, (11.0, -) (3.7, (5.5, 16.9) (7.2, -) (3.3, (5.1, (5.1, (1.9,
such as antitumor vaccines or checkpoint inhibitors. Clinical trial information: 15.2) 13.2) 16.2) 18.4) 18.4) 14.4)
NCT02829099.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Immunotherapy and Tumor Immunobiology 113s

2529 Poster Session (Board #173), Sat, 8:00 AM-11:00 AM 2530 Poster Session (Board #174), Sat, 8:00 AM-11:00 AM
Phase 1a/1b study of first-in-class B7-H4 antibody, FPA150, as monotherapy Tumor responses and early onset cytokine release syndrome in synovial
in patients with advanced solid tumors. First Author: Jasgit C. Sachdev, sarcoma patients treated with a novel affinity-enhanced NY-ESO-1-targeting
Scottsdale Healthcare, Paradise Valley, AZ TCR-redirected T cell transfer. First Author: Mikiya Ishihara, Mie University,
Tsu, Japan
Background: B7-H4, a transmembrane protein of the B7 family, is a negative
regulator of T cell function, expressed at high levels on several cancers, Background: Adoptive transfer of TCR-redirected T cells has been reported to
including approximately 50% of breast, ovarian and endometrial cancers. exhibit efficacy in some of melanoma and sarcoma patients. However, there
FPA150 is a fully human antibody against B7-H4 that blocks inhibition of have not been well known about cytokine release syndrome (CRS) or its
T cell activity and has enhanced antibody-dependent cell-mediated cyto- relations to tumor responses. This study evaluates clinical responses in
toxicity. It is the first therapeutic molecule targeting B7-H4 to enter the association with the cell kinetics and CRSs after transfer of high-affinity NY-
clinic. We report preliminary results from an ongoing phase 1a/1b study of ESO-1 TCR-gene transduced T cells in NY-ESO-1-expressiong cancer pa-
FPA150 in advanced solid tumors. Methods: Phase 1a included dose es- tients (NCT02366546). Methods: We developed a novel-type affinity-
calation in which B7-H4-unselected patients with advanced solid tumors enhanced NY-ESO-1-specific TCR and an originally-developed retrovirus
were treated with FPA150 at doses between 0.01 to 20 mg/kg every three vector that encodes siRNA to silence endogenous TCR creation. The NY-
weeks (Q3W) in an accelerated titration followed by 3+3 design and a ESO-1/TCR sequence is mutated for high affinity with replacements of G50A
separate dose exploration cohort in which B7-H4+ (H-score$100) patients and A51E in CDR2 region. This is a first-in-man clinical trial of the novel NY-
were treated at doses of 3 or 10 mg/kg Q3W with mandatory pre- and on- ESO-1-specfic TCR-T cell transfer to evaluate the safety, in vivo cell kinetics
treatment biopsies. Results: As of 12/31/2018, 24 patients with a median of and clinical responses. It was designed as a cell-dose escalation from 5 x108
3 prior therapies were treated with FPA150, 6 of whom were in the B7-H4+ to 5 x109 cells. NY-ESO-1-expressing refractory cancer patients were en-
dose exploration cohort. Seven patients from dose escalation were also rolled, with 3+3 cohort design. Cyclophosphamide (1,500mg/m2) were
retrospectively identified as B7-H4+. Most patients received FPA150 at 3 administered prior to the TCR-T cell transfer as pre-conditioning. Results: 9
mg/kg (n=8) or 10 mg/kg (n=6). Median number of doses was 3 (range 1-11). patients were treated with the NY-ESO-1/TCR-T cell transfer. The TCR-
No dose-limiting toxicities or treatment-related serious adverse events were T cells expanded in peripheral blood with a dose-dependent manner, as-
reported, and there were no treatment-related AEs (TRAEs) leading to sociated with rapid proliferation within 5 days after the cell transfer. 3
discontinuation of FPA150. Most TRAEs were Grade 1-2, with diarrhea and patients receiving 5x109 cells developed early-onset CRSs, with elevations
fatigue most common (16.7%). Grade 3 TRAE hypertension occurred in 1 of serum IL-6, IFN-g. The CRSs developed on day1 or 2 after the cell transfer.
patient. FPA150 displayed approximately dose-proportional exposure at They were well managed with tocilizumab treatment. 3 synovial sarcoma
doses $0.3 mg/kg with half-life of 1-2 weeks. Conclusions: FPA150 patients exhibited tumor shrinkages of partial responses, and they all had
monotherapy demonstrated a favorable safety profile and evaluation of anti- high-expression of NY-ESO-1 in the tumor samples, namely, 75% or more.
tumor activity is ongoing. 20 mg/kg Q3W was selected as the recommended Exploratory analysis revealed that multiple chemotactic cytokines in-
dose. Phase 1b enrollment of FPA150 monotherapy in patients with B7-H4+ cluding CCL2 and CCL7, and IL-3 increased in the serum from the patients
breast, ovarian and endometrial cancer began in February 2019. We will with CRS. The proportions of effector-memory phenotype T cells in the
present updated safety, PK, and preliminary biomarker and efficacy data. infused cell-product were significantly associated with CRS development.
Clinical trial information: NCT03514121. Conclusions: The affinity-enhanced NY-ESO-1/TCR-T cell transfer exhibited
early-onset CRS in association with in vivo cell proliferation and sequential
tumor responses in the patients with high-NY-ESO-1-expressing synovial
sarcoma. Clinical trial information: NCT02366546.

2532 Poster Session (Board #176), Sat, 8:00 AM-11:00 AM 2533 Poster Session (Board #177), Sat, 8:00 AM-11:00 AM
Correlation of circulating EBV-targeted cytotoxic T lymphocyte precursors Phase I study of OKT3 x hu3F8 bispecific antibody (GD2Bi) armed T cells
(EBV-CTLp) and clinical response following tabelecleucel (tab-cel) infusion (GD2BATs) in GD2-positive tumors. First Author: Maxim Yankelevich,
in patients with EBV-driven disease. First Author: Blake T. Aftab, Atara Childrens Hosp of Michigan/Wayne State Univ, Detroit, MI
Biotherapeutics, Thousand Oaks, CA
Background: With the proven success of anti-GD2 monoclonal antibodies in
Background: EBV is implicated in a variety of diseases. Tab-cel is an inves- eradicating minimal residual disease in neuroblastoma (NB), exploiting
tigational off-the-shelf, allogeneic T-cell immunotherapy utilizing endogenous antibody based anti-GD2 in T cell mediated strategies has potential to
T cell receptors targeting EBV antigens. We hypothesized the clinical activity of combat higher disease burden and improve patient outcome. We hypoth-
tab-cel is mediated by expansion and persistence of EBV-specific T cells. esized that arming of ex vivo expanded and activated, autologous, blood
Therefore, we quantified circulating EBV- CTLp after tab-cel administration derived T cells (ATC) with chemically heteroconjugated GD2Bi should re-
and examined the correlation between expansion and clinical response. direct them to target NB. In vitro, ATC coated (armed) with 50 ng/106 cells of
Methods: Samples from 10 patients with EBV+ post-transplant lymphoproli- GD2Bi exhibited specific killing of NB and osteosarcoma (OS) cell lines.
ferative disease (PTLD) and other EBV-associated diseases enrolled in an Methods: In this phase I study (NCT02173093), patients with GD2-positive
multicenter expanded access protocol (EAP) study (NCT02822495) were tumors received 8, biweekly infusions of GD2BATs + daily low-dose IL-2 and
analyzed. To evaluate CTLp frequencies, limited dilution analysis was per- biweekly granulocyte-macrophage colony stimulating factor (GM-CSF). The
formed on samples taken at baseline and day 34 post first tab-cel dose (end study followed the standard 3+3 design with dose levels of 40, 80, and 160 x
cycle 1). The day 34 persistence of circulating EBV-CTLp from best overall 106 GD2BATs/kg/infusion. Results: Twelve patients (NB = 7, OS = 3,
response to initial tab-cel product was tested using the two-tailed Mann- Desmoplastic Small Round Cell Tumor = 2) were enrolled from 11/2013 to
Whitney test. Changes in inflammatory cytokines were also measured. 12/2017 and 9 completed therapy. Adequate ATCs could not be grown in
Results: Responders represented in this sampling (n=6; 2 PR and 4 CR) one patient and two patients did not complete 8 infusions because of rapid
showed a median 5.8-fold increase in circulating CTLp between baseline disease progression. Infusions were given in outpatient settings. All patients
and day 34 (range: 0.8 to 133-fold). Five of 6 responders showed an in- developed a mild, dose-independent and manageable form of cytokine re-
crease in EBV-CTLp at day 34 of $ 3.8-fold while 1 pt showed no change in lease syndrome with grades 2-3 fevers/chills, headaches and occasional
CTLp (0.8-fold change). In contrast, the 4 non-responders (3 SD; 1 PD) hypotension for up to 48 hours after infusion. No patients developed sig-
showed a median 0.3-fold decrease in EBV-CTLp from baseline (range: 1.2 nificant pain. Maximum tolerated dose was not reached. Evidence of activity
to 0.02-fold; ns). Cumulative analyses revealed a statistically significant was seen in several patients including one patient with OS who had a PET
correlation between the fold-change of circulating CTLp at day 34 and response, one patient with NB who had complete bone marrow response (this
clinical response (p=0.038) which did not appear to correlate with the type patient had remained progression free for 2.5 years after completion of
of the EBV-associated disease. Inflammatory cytokines showed no infusions), and another NB patient who had a minor response on MIBG scan.
meaningful change from baseline. The safety profile remains consistent Four patients with NB are currently alive after additional therapies at 12, 14,
with previously reported data. Conclusions: These data support the cor- 18, and 47 months post BAT infusions. Conclusions: Autologous T cells from
relation of clinical activity of tab-cel with the expansion and persistence of heavily pretreated patients could be expanded ex vivo to large numbers,
EBV-specific T-cells at day 34 post-treatment, as well as the use of cir- armed with GD2Bi, cryopreserved and thawed for safe IV administration up to
culating CTLp as a biomarker for response in clinical studies. Clinical trial total dose of 1.28x109/kg. Ongoing phase II arm of the trial will focus on
information: NCT02822495. evaluation of clinical activity of GD2BATs in patients with NB. Clinical trial
information: NCT02173093.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
114s Developmental Immunotherapy and Tumor Immunobiology

2534 Poster Session (Board #178), Sat, 8:00 AM-11:00 AM 2535 Poster Session (Board #179), Sat, 8:00 AM-11:00 AM
Efficacy and safety of CAR19/22 T-cell “cocktail” therapy in patients with The phase I clinical study of CART targeting BCMA with humanized alpaca-
refractory/relapsed B-cell non-Hodgkin lymphoma. First Author: Liang derived single-domain antibody as antigen recognition domain. First Author:
Huang, Tongji Hospital, Tongji Medical College, Huazhong University of Lu Han, Department of Immunology, Affiliated Cancer Hospital of
Science and Technology, Wuhan, China Zhengzhou University and Henan Cancer Hospital, Zhengzhou, China
Background: Antigen escape relapse has emerged as a major challenge for long- Background: Several phase I clinical trials already shown Chimeric antigen
term disease control post CD19-directed therapies, to which dual-targeting of receptor T cells (CART) targeting BCMA has the promised effects to treat the
CD19 and CD22 has been proposed as a potential solution. Methods: Between relapsed/refractory (RR) multiple myeloma (MM), RRMM. We developed
Mar 2016 and Jan 2018, we conducted a pilot study (ChiCTR-OPN-16008526) CART cells (CART-BCMA) using one single-domain antibody as recognition
in 38 patients (pts), who had refractory/relapsed B-cell non-Hodgkin lymphoma domain. The anti-BCMA single-domain antibody was derived from the al-
(B-NHL), to evaluate the efficacy and safety of sequential infusion of anti-CD19 paca, and humanized with the affinity of 1.14nM. The CART-BCMA use the
and anti-CD22, two single-specific, third-generation CAR19/22 T-cell “cock- 4-1BB and CD3z intracellular regions as T cell activation domain.
tail”. The cutoff date for data collection was Apr 30, 2018. Results: At a Methods: A phase I, single arm clinical study was conducted to assess safety
minimum follow-up of 3 months (mos), 26 of the 36 evaluable pts achieved an and efficacy of CART-BCMA. The enrolled RRMM patients had received
overall response (ORR), including 18 with a complete response (CR) and 8 average 10 lines of prior treatment, no matter BCMA expression level on
with a partial response (PR). The ORR at mo 3 was consistent in different plasma cells. Patients were subjected to a lymphodepleting regimen with Cy
subgroups, irrespective of pathologic subtypes, cell of origin, cytogenetic or (300-600 mg/m2, d-5, -4) and Flu (30 mg/m2, d-5 to d-3) before CART
genomic aberrations. At the data cutoff, 15 of the 18 pts who had a CR at mo 3
infusion at the dose of 2-103106 CAR+ cells/kg. The efficacy was assessed
maintained their responses, 2 of 8 pts who had a PR within 3 mos continued to
based on the IMWG Criteria, and the toxicity was graded by CTCAE 4.02.
have a CR without additional therapies. Collectively, the best ORR was 83.3%,
with a best CR rate of 55.5% and a best PR rate of 27.8%. With a median follow-
Results: As of December 31, 2018, 16 patients were infused with autolo-
up of 5.3 mos (range, 0.4 to 16.2), the median PFS was 5.8 mos, and the median gous CART-BCMA cells, and had at least 1 month of follow-up. Many patients
OS was not reached (NR). Pts received therapy at first relapse had better PFS than have M protein in serum, but haven’t the high percent of plasma cells in bone
those who received therapy at the time with primary refractory diseases or at marrow, which are difficult to be treated by CART cells because the tumor
multiple relapses. Notably, pts who achieved an overall response at mo 3 (R3m) cells are aggregated, not diffused in bone marrow. 3 patiens were diagnosed
had significantly extended PFS and OS when compared with pts who did not. with extramedullary diseases, were evaluated as PR at D28 (tumor SPD
Repeated biopsy and IHC was conducted in 3 of the 13 pts. However, loss of decreasing ＞50%). 13 patients haven’t extramedullary diseases, at D28,
CD19 or CD22 was not detected. Of the 9 pts with IgH/MYC translocation, with a ORR is 84.6% (11/13); At 10 weeks, 7 patients were evaluated, ORR is
median follow-up of 10.1 mos, the median PFS and median OS were NR. At data 100% (sCR/CR 42.8%, VGPR 14.3% , PR 42.8% ); 5 patients reached
cutoff, 7 pts who had achieved R3m maintained their responses, including all the 16 weeks, 1 relapsed, 4 kept remission. The Pt3 and Pt6 shows the CRS
4 pts with double-hit lymphoma. However, of the 10 pts with del(17p) or TP53 grade 3 or 4, other patients shows the grade 0-2 CRS, the CRS is man-
mutation, with a median follow-up of 5.3 mos (range, 2.7 to 14.5), the median PFS ageable. Conclusions: Our result demonstrates the promising efficacy
was 3.6 mos and the median OS was 9.9 mos. All pts experienced reversible CRS, compared with other reported results of CART targeting BCMA, and supports
with 21.1% were of high-grade. Neurotoxicity developed in 13.2% pts and were all further development of this anti-RRMM cellular immunotherapy. Clinical
low-grade. Conclusions: Our results indicated that sequential infusion of CAR19/ trial information: NCT03661554.
22 T-cell is efficient and safe for pts with B-NHL. Dual antigen targeting is a
promising approach to circumvent antigen loss relapse after CAR T-cell therapy. The
impact of genetic subtypes and clinical parameters further underscores the critical
importance of personalized immunotherapies. Clinical trial information: ChiCTR-
OPN-16008526.

2536 Poster Session (Board #180), Sat, 8:00 AM-11:00 AM 2537 Poster Session (Board #181), Sat, 8:00 AM-11:00 AM
Ligand-inducible, prostate stem cell antigen (PSCA)-directed GoCAR-T cells Effect of minimal lymphodepletion prior to ACT with TBI-1301, NY-ESO-1
in advanced solid tumors: Preliminary results with cyclophosphamide (Cy) 6 specific gene-engineered TCR-T cells, on clinical responses and CRS. First
fludarabine (Flu) lymphodepletion (LD). First Author: Carlos Roberto Author: Marcus O. Butler, Princess Margaret Cancer Centre, University
Becerra, Baylor University Medical Center, Dallas, TX Health Network, University of Toronto, Toronto, ON, Canada
Background: Cell-surface protein PSCA is upregulated in many solid tumors Background: Adoptive transfer of T cell receptor (TCR) gene-engineered
and correlates with disease stage. BPX-601, an autologous T-cell product T cells can induce durable anti-cancer responses. Post-infusion cytokine
expressing a PSCA-CD3z CAR and a rimiducid (Rim)-inducible MyD88/ release syndrome (CRS) has been associated with clinical utility. Pre-
CD40 co-activation switch to augment T-cell proliferation and persistence, is infusion lymphodepletion (LD) may influence CRS, graft persistence, and
designed to have enhanced efficacy in solid tumors vs traditional CARs. This clinical responses. While the optimal LD regimen is not yet defined, most
ongoing first-in-human study assesses safety, biologic, and clinical activity include both cyclophosphamide (CY) and fludarabine (FLU). TBI-1301 is a
of BPX-601+Rim in PSCA+ cancers. Updated results, including those from novel gene therapy produced by engineering autologous lymphocytes to
patients (pts) who underwent LD with Flu/Cy, are presented. Methods: BP- express an NY-ESO-1-specific TCR using a retrovirus vector that encodes
012 is a 2-part, open-label trial. Part 1 is a 3+3 dose escalation of BPX-601 siRNA to silence endogenous TCR. Since less intensive LD may be sufficient
(1.25–5.0x106 cells/kg; Day [D] 0) given prior to a single, fixed Rim dose with the use of this novel vector, we are conducting a study where patients are
(0.4 mg/kg; D7) in pts with previously treated PSCA+ metastatic pancreatic, treated with TBI-1301 following LD with CY alone. Methods: Eligibility
gastric, or prostate cancers with measurable disease. Results: As of Jan-22- includes informed consent, HLA-A*02:01 or A*02:06 haplotype, and NY-
2019, 15 pts have received BPX-6016Rim. Two pts at the highest cell dose ESO-1 expression by immunohistochemistry. Eligible patients undergo
received Flu/Cy for LD on D25 to D23 before BPX-601; LD after Flu/Cy was harvest of PBMC which are then processed locally to generate engineered
96.6% and 84.3%. Thirteen pts received Cy alone on D23; in these pts, LD TBI-1301 cells. The study design is to infuse 5x109 cells (day 0) to patients
ranged from 0–68.6%. Rapid cell expansion by D4 was observed in all pts following LD with CY (750 mg/m2 on day -3 and -2). Endpoints include
with peak vector copy number 8.3-fold higher with Flu/Cy (n = 2) vs Cy LD safety, efficacy, and biological correlates for persistence of NY-ESO-1-
(n = 13). Serum IP-10, IL-6 and TNFa increased . 2-fold from baseline in $1 specific T cells post infusion. Results: Thus far, 9 patients have been
pt in all Rim cohorts, with 3- to 20-fold Rim-dependent cell expansion in 6 treated, and 8 have received the target dose. To date, 8 patients are
pts. No CRS or DLTs were reported. After Rim, one Flu/Cy pt experienced a evaluable for response and toxicity, and no DLTs have been observed.
serious Grade 2 AE (encephalopathy) related to BPX-601+Rim that resolved Despite LD with CY alone, all 4 patients with synovial sarcoma and 1 with
with IV steroids; despite time-matched nonserious Grade 1 pyrexia, the pt melanoma experienced clinical and laboratory evidence of grade 1-2 CRS
had no other CRS symptoms. After BPX-601+Rim and $1 scan, best re- with increased CRP, ferritin, and IL-6 levels. CRS resolved spontaneously in
sponses were 8 SD and 3 PD (1 non-evaluable). With a median follow-up of all but one patient who required tocilizumab due to grade 2 nausea/vomiting.
9.8 wks, time to next treatment (tx) after BPX-601 ranged from 2.7–22.1 Two subjects experienced grade 3 tumor-associated pain. Other treatment-
wks (n = 8) and ongoing tx-free intervals range from 9.1–30.1 wks (n = 4). associated grade 3 or 4 toxicities included neutropenia and hypo-
Conclusions: BPX-601+Rim was well-tolerated with manageable safety and phosphatemia. Best overall response by RECIST is as follows: 2 partial
early evidence of enhanced CAR T-cell expansion and prolonged persistence responses, 5 stable disease, and 1 progressive disease. Biomarker analysis
after Flu/Cy vs Cy. Additional pts will undergo Flu/Cy LD prior to BPX-601 demonstrates persistence of transferred TBI-1301 cells, . 100 days in some
with single- and repeat-dose Rim. Clinical trial information: NCT02744287. patients. Conclusions: TBI-1301 appears to be safe and to possess anti-
tumor activity. Despite LD with CY alone, grade 1-2 CRS is induced. Ad-
ditional cohorts to this study will examine the role of repeat infusions to
enhance anti-tumor activity. Clinical trial information: NCT02869217.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Immunotherapy and Tumor Immunobiology 115s

2538 Poster Session (Board #182), Sat, 8:00 AM-11:00 AM 2539 Poster Session (Board #183), Sat, 8:00 AM-11:00 AM
Safety and efficacy of adoptive cell transfer using autologous tumor Phase 1 trial of anti-CD19 chimeric antigen receptor T (CAR-T) cells with
infiltrating lymphocytes (LN-145) for treatment of recurrent, metastatic, tumor necrosis alfa receptor superfamily 19 (TNFRSF19) transmembrane
or persistent cervical carcinoma. First Author: Amir A. Jazaeri, The University domain. First Author: Paolo Fabrizio Caimi, Adult Hematologic Malignancies
of Texas - MD Anderson Cancer Center, Houston, TX and Stem Cell Transplant Program, University Hospitals Seidman Cancer
Center, Cleveland, OH
Background: There is a high unmet medical need for effective treatments for
patients with recurrent, metastatic, or persistent cervical cancer. Most Background: AntiCD19 CAR-T cells have shown encouraging anti-lymphoma
patients are young and survival rates are poor. ORR for second line therapies activity. Decreasing the time from apheresis to CAR-T infusion can make this
is between 4 and 14% for chemotherapy and recently approved immuno- therapy available to pts with rapid progression. We present the interim results
therapy. Adoptive cell transfer using tumor infiltrating lymphocytes (TIL) of a phase I clinical trial using on-site CAR-T manufacture. Methods: Adult
have demonstrated durable responses in some patients with recurrent pts with r/r CD19+ B cell lymphomas who failed $ 2 lines of therapy were
cervical cancer thus offering the potential for long-term disease control. enrolled. Autologous T cells were transduced with a lentiviral vector (Len-
Methods: Study C-145-04 is an ongoing, open-label, multicenter Phase 2 tigen Technology, Inc,LTG1563) encoding an antiCD19 binding motif, CD8
clinical trial evaluating the safety and efficacy of LN-145 TIL therapy in linker and TNFRSF19 transmembrane region, and 4-lBB/CD3z domains.
patients with advanced cervical cancer who have undergone at least one prior GMP-compliant manufacture was done using CliniMACS Prodigy, in a
line of chemotherapy. Prior checkpoint inhibitor therapy is an exclusion 12-day culture. Dose levels were 0.5, 1 and 2 x 106 CAR-T cells/kg.
criterion. The primary endpoint is ORR per RECIST 1.1; secondary endpoints Lymphodepletion was done with cyclophosphamide (60mg/kg x 1) and
include duration of response (DOR), disease control rate (DCR), and LN-145 fludarabine (25mg/m2/d x 3). Results: 7 pts (4 women, 3 men) were en-
safety. Tumors surgically harvested at local institutions are shipped to rolled. Median age was 60y [range 43-69]. Diagnoses were DLBCL (n = 3)
central GMP facilities for TIL generation in a 22-day manufacturing process. PMBCL, follicular lymphoma (FL), transformed FL, and transformed lym-
Final LN-145 TIL product is cryopreserved and shipped to sites. Patients phoplasmacytic lymphoma; with a median of 4 previous treatments. Six pts
receive one week of preconditioning lymphodepletion (cyclophosphamide, had symptomatic refractory disease. CAR-T cell product manufacture was
fludarabine), a single LN-145 infusion, followed by up to 6 doses of IL-2 successful in all pts. Mean transduction rate was 44% [range 29-57]. CAR-
(600,000 IU/kg). Results: As of 4 Feb 2019, 27 efficacy-c patients have T cell doses were 0.5 x 106/kg (n = 3) and 1 x 106/kg (n = 4). Median apheresis
received Gen 2 of LN-145, with a mean age of 47 years and 2.6 mean prior to infusion time was 13 days [range 13–20], 5 products were infused fresh.
lines of therapy. Preliminary efficacy results: ORR was 44% (1 CR, 9 PR, 2 CAR-T persistence based on vector sequence, peaked in peripheral blood
uPR), DCR was 89% at 3.5-month median study follow-up with 11/12 MNCs between days 14-21. Five pts are evaluable for safety. CRS grade 1 - 2
patients maintaining their response. Improved responses were observed in 4 (Lee) occurred in 4 pts; with 3 requiring treatment. Grade 4 CRES (CARTOX-
patients with longer follow-up. Mean TIL cells infused was 28x109. Median 10) occurred in 1 pt, with resolution after corticosteroids; considered a DLT as
IL-2 doses administered was 6.0. The adverse event profile was generally it lasted more than 72 hours. No treatment-related mortality has occurred. 4/5
consistent with the underlying advanced disease and the profile of the evaluable pts have achieved complete response. One pt did not respond and
lymphodepletion and IL-2 regimens. Conclusions: LN-145 results in 44% died. After a median follow up 3 months, all responding pts are alive and 1
ORR in previously treated cervical cancer patients with acceptable safety relapsed 6 mo after treatment. Conclusions: Second generation antiCD19
and efficacy profile. LN-145 offers patients a viable therapeutic option CAR-T cells with TNFRS19 transmembrane domain have clinical activity
warranting further investigation. Clinical trial information: NCT03108495. against refractory NHL. Short manufacture time achieved by local CAR-T cell
manufacture with the CliniMACS Prodigy enables treatment of a very high risk
NHL population. Clinical trial information: NCT03434769.

2540 Poster Session (Board #184), Sat, 8:00 AM-11:00 AM 2541 Poster Session (Board #185), Sat, 8:00 AM-11:00 AM
Comprehensive report of anti-CD19 chimeric antigen receptor T cells (CAR- Clonal expansion of tumor infiltrating lymphocytes (TILs) in the peripheral
T) associated non-relapse mortality (CART-NRM) from FAERS. First Author: blood of metastatic melanoma patients is significantly associated with
Kartik Anand, Houston Methodist Cancer Center, Houston, TX response to CTLA4 blockade-based immunotherapy. First Author: Arjun
Background: CAR-T cells targeting CD19 positive B-cells have improved outcomes for
Khunger, Cleveland Clinic, Cleveland, OH
relapsed/refractory non-Hodgkin lymphoma (NHL) and B-cell acute lymphoblastic Background: Patients with metastatic melanoma were treated on a clinical
leukemia (B-ALL). CAR-T emergent toxicities for FDA approved therapies leading to non- trial with tremelimumab and High Dose Interferon-Alfa (HDI) (Tarhini. J Clin
progression related death have been reported in the pivotal studies. However, they are
Oncol. 2012). We previously reported that patients who achieved disease
underreported and there remains a need to obtain a comprehensive report of NRM
emergent with anti-CD19 CAR-T. Methods: We retrospectively searched FDA adverse control and clinical response had significantly greater T-cell clonality (p =
events reporting system (FAERS) for all adverse events (AE) related to “Tisagenlecleucel(T)” 0.0008) and T-cell fraction (p = 0.044) respectively in their pretreatment
and “Axicabtagene ciloleucel(AC)” reported from 2013-2018. FAERS contains AEs tumor biopsy samples (Tarhini. J Clin Oncol. 2017). In this study, we further
from clinical trials and standard of care patients. All cases with the outcome of death characterize T-cell repertoire clonality and clonal expansion in the peripheral
were analyzed. Results: Total numbers of anti-CD19 CAR-T pts reported were 636, out blood at different time points to evaluate the association between repertoire
of which 288 cases received “T” and 348 received “AC”. Out of total 129 total deaths, features and clinical response. Methods: Patients received tremelimumab
95 died due to non-disease progression. Patient characteristics are summarized in 15 mg/kg I.V. every 12 weeks and HDI was given concurrently. Responses
Table. CART-NRM for entire cohort was 15%; 21% for “T” and 10% for “AC”. Major
toxicities reported include CRS, hematological, cardiovascular, neurological and in-
were assessed by RECIST as complete (CR) or partial (PR), stable disease
fectious. Difference in mortality is likely related to different patient population, di- (SD) or progression (PD). Peripheral blood mononuclear cells (PBMCs) from
agnoses and the CAR-T construct. Conclusions: CART-NRM remains considerably high treated patients (N = 33) were obtained at baseline, day 29, and day 85
at 15%. Our analysis highlights the major toxicities and informs the potential oppor- (following tremelimumab-HDI treatment); tumor samples at baseline were
tunities for interventions to reduce mortality. We will present updated data with also obtained (N = 18). The T-cell receptor beta chain (TCRB) repertoire of
comparative analysis of published clinical studies at the upcoming ASCO Meeting in PBMCs and tumor samples was immunosequenced using the immunoSEQ
Chicago. assay (Adaptive Biotechnologies), and repertoire clonality was assessed at
Analysis of cases of anti-CD19 CAR-T NRM. baseline, day 29, and day 85. Differential abundance analysis was used to
Total Deaths (n=95) Tisangenlecleucel (n=61) Axicabtagene ciloleucel (n=34) detect and quantify peripheral clonal expansion pre- versus post-treatment
Median age
Number of patients £18 years
21 years(3-78 years; n=57)
26(median-age 7.5 years)
64 years(13-71 years; n=21)
1(13 years)
and identify the subset of peripheral clones also detected in the tumor
Number of patients >18 years 31(median-age 48 years) 20(median-age 64 years) repertoire. The Morisita Index of repertoire similarity was also calculated to
Unknown age of patients 4 13
Sex T AC compare global repertoire changes between pre- and post-treatment PBMC
Male
Female
33
25
12
17
samples. Results: T-cell repertoire turnover, as measured by the Morisita
Unknown 3 5 Index, showed a trend towards responders (CR/PR) having greater turnover
Indications T AC
ALL 31 2 (lower Morisita Index) post-treatment than non-responders (SD/PD). Simi-
NHL
Chronic lymphocytic leukemia
13
2
23
0
larly, the total number of clones expanding in the peripheral repertoire varied
Unknown 15 9 over time within an individual (p = 0.034) but was not significantly affected
Non Relapse Mortality* T AC
Cytokine release syndrome(CRS) 14(23%) 18(53%) by response to therapy (p = 0.275) or by on-treatment time point (p = 0.768).
Hematological 28(46%) 9(26%)
Cardiovascular 29(47%) 14(41%) When the analysis was restricted to peripherally expanded clones that were
Neurological
Renal
28(46%)
25(41%)
19(56%)
3(9%)
also found in the tumor repertoire, responders had significantly more TILs
Gastrointestinal 17(28%) 9(26%) expanded in the periphery at day 29 than non-responders (p = 0.036).
Respiratory 20(32%) 8(23%)
Infectious 34(56%) 15(44%) Conclusions: Our analysis of the peripheral T-cell repertoire following
Hepatic
Median time to AEs
4(6%)
7 days
3(9%)
5 days
treatment showed that detection of TILs in early peripheral clonal expansion
correlates with response to therapy.
*Overlapping toxicities reported

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
116s Developmental Immunotherapy and Tumor Immunobiology

2542 Poster Session (Board #186), Sat, 8:00 AM-11:00 AM 2543 Poster Session (Board #187), Sat, 8:00 AM-11:00 AM
Bespoke circulating tumor DNA (ctDNA) analysis as a predictive biomarker Evolution of the myeloid-derived suppressor cells in advanced breast cancer
in solid tumor patients (pts) treated with single-agent pembrolizumab (P). and comparative analysis with a healthy population cohort. First Author:
First Author: Marco Adelmo James Iafolla, Princess Margaret Cancer Centre, Natalia Palazón-Carrión, Virgen Macarena University Hospital, Seville,
Toronto, ON, Canada Spain
Background: Limited data exist in the clonal dynamics of serial ctDNA as a Background: High levels of myeloid-derived suppressor cells (MDSCs) seem a
predictive biomarker in advanced solid tumor pts receiving immune negative prognostic factor in advanced breast cancer (ABC) patients (pts).
checkpoint blockade. Methods: Pts with mixed solid tumors received single Preclinical studies suggest an immunomodulatory effect of some classical
agent P (anti-PD-1) 200 mg IV Q3wks in the investigator-initiated phase II anti-tumor agents through alteration of MDSCs homeostasis. We analyzed the
INSPIRE trial (NCT02644369). ctDNA was assayed at baseline (B) and start association of MDSCs and clinical evolution of ABC pts, taking into account
of cycle 3 (C3) using a pt-specific amplicon-based NGS assay (Signatera™). the systemic treatment (tx) modulation of MDSCs levels in pts from two studies
Samples were considered ctDNA positive if $2 of 16 pt-specific targets met (“A”: GEICAM/2015-04 PANGEA-BREAST, NCT03025880 “Efficacy and
the qualifying confidence score threshold. Results: Results of 70 pts are Safety of Pembrolizumab and Gemcitabine in HER2-negative ABC”, and “B”:
presented. Demographics: male 46%; median age=60 yrs (range 21–82); PI-0502-2014 “Peripheral blood analyses of immune response induced by 1st
head and neck (20%), triple negative breast (14%) and ovarian (14%) line tx of ABC according to clinical guidelines”). Methods: MDSCs (CD33+
cancers comprised the major malignancies. Median no. of P cycles=4 (range CD11b+) levels were determined by flow-cytometry in peripheral blood
2–35); follow up was 14m (range 2–29); RECIST responses: CR 2.9% (n=2), samples at three time points (basal, at cycles 3 and 6) from: 39 HER2-
PR 17% (n=12), CBR (CR+PR+SD$6 cycles) 31% (n=22), RECIST/clinical negative heavily pretreated pts from study “A”, 43 non-pretreated pts (all
PD (n=43/10; 65%/15%). Median PFS=3.3m and median OS=17.8m. 68/ subtypes) from study “B” and 20 women from a healthy cohort (HC), with no
70 pts had ctDNA detected at baseline (median=16/16 variants) demon- cancer diagnosis. MDSCs levels from the different cohorts were compared and
strating 97% sensitivity. Table shows correlation of DctDNA (ctDNAB correlated with pts with Clinical Benefit (CB: partial/complete response +
compared to ctDNAC3) with clinical efficacy parameters, whereas ctDNAB disease stabilization) vs pts with Progressive Disease (PD). Results: Tx re-
values did not reach statistical significance. Conclusions: A strong corre- sponse was assessed in 33 pts (85%) from study “A” and 39 pts (91%) from
lation exists between DctDNA with OS, PFS, CBR and ORR with P, sug- study “B”. CB was observed in 11 pts (28%) from study “A” and in 34 (79%)
gesting it is a potential predictive biomarker in pts with mixed solid tumors. from study “B” while PD was observed in 22 pts (56%) from study “A” and in 5
Clinical trial information: NCT02644369. (12%) from study “B”. Basal MDSCs levels were significantly higher in ABC
ORR CBR PFS OS
pts (studies “A”+”B”) than in HC (15.95 vs 0.81 cells/ml, p = 0.009). At cycle
Endpoint
N = 68* N = 68*
Endpoint
N = 69* N = 69* 6, MDSCs were considerably lower in pts with CB vs DP (2.90 vs 13.75 cells/
CR/PR/ ml, p , 0.001). This decrease was more pronounced in study “B” than in study
SD‡6 SD<6 ↓ from ↑ from ↓ from ↑ from
CR/PR SD/PD cycles cycles/PD baseline baseline baseline baseline “A” pts (p , 0.001 vs p = 0.074, respectively), probably due to differences
Subgroup N = 13 N = 55 N = 20 N = 48 Subgroup N = 31 N = 38 N = 31 N = 38 in number of events, tumor subtypes and tx between both studies.
DctDNA Median = Median = Median = Median = Results based Median = Median = Median = Median = Conclusions: Our results suggest that ABC pts show alterations in MDSCs and
(% -91.5% 31.5% -75% 49% on DctDNA 5.6m 2.9m 24.0m 9.5m
change) Range = Range = Range = Range = 6m = 6m = 6m = 6m = that their decrease along tx may have a positive predictive value, highlighting
-100% to -98.7% to -100% to -94.6% to
18.1% 2,458.1% 96.1% 2,458.1%
48.4% 10.5% 90.3%
12m =
73.7%
12m =
the importance that immune-competent status may play in the evolution of
79.9% 46.7% ABC. MDSCs may represent a target for therapeutic purposes in ABC.
P-value P , 0.001 P , 0.001 Adjusted HR 0.49 (0.28–0.84) 0.38 (0.18–0.80)
(95% CI)^ P = 0.01 P = 0.01

^Adjustment on cohorts; *2 pts were excluded from ORR/CBR analyses as baseline ctDNA = 0; 1 of these 2 pts (with PR)
excluded from PFS/OS analyses as C3 ctDNA = 0.

2544 Poster Session (Board #188), Sat, 8:00 AM-11:00 AM 2545 Poster Session (Board #189), Sat, 8:00 AM-11:00 AM
Development of a baseline prognostic cytokine signature that correlates CD4+ T-cell immunity predicts long-lasting antitumor immunity after PD-1
with nivolumab (NIVO) clearance (CL): Translational pharmacokinetic/ blockade therapy. First Author: Kyoichi Kaira, Division of Respiratory
pharmacodynamic (PK/PD) analysis in patients with renal cell carcinoma Medicine, Saitama Medical University International Medical Center, Hidaka,
(RCC). First Author: Rui Wang, Bristol-Myers Squibb, Princeton, NJ Japan
Background: CL of checkpoint inhibitors has been identified as a predictive Background: Patients treated with programmed cell death 1 (PD-1)-blockade
covariate of overall survival (OS) in several tumors. Determination of CL therapy fall into 3 distinct subgroups: non-responders presenting early disease
requires post-treatment samples, which negates its utility as a baseline progression, long survivors who achieve durable disease control, and the
prognostic biomarker. This study aims to identify a baseline composite remaining short-term responders. We reported that the prediction formula
cytokine signature that correlates with NIVO CL in patients with RCC using comprised of the percentages of CD62L-downregulated (CD62Llow) and CD25+
translational PK/PD analysis. Methods: Peripheral serum PK (NIVO CL) and FOXP3+CD4+T cells in the peripheral blood predicted non-responders of non-
serum biomarkers to assess PD (Myriad Rules-Based Medicine customized small cell lung cancer patients (n = 50) scheduled to receive anti-PD-1-
inflammatory cytokine panel) were analyzed from 985 patients with RCC antibody (nivolumab) therapy in the 2017 ASCO meeting. In this study, we
enrolled in 3 clinical trials. CheckMate (CM) 009 (NCT01358721) and CM included 171 patients with NSCLC who were scheduled for nivolumab
025 (NCT01668784) of NIVO (n = 481) were used for model development treatment after obtaining written informed consent. Peripheral blood mono-
(training dataset). CM 010 (NCT01354431) of NIVO and a cohort treated nuclear cells (PBMC) were examined before and after Nivolumab therapy up to
with everolimus in CM 025 were included in model application (test dataset; 2 years to investigate the differences between long survivors and short-term
n = 504). PK/PD analyses were conducted using a machine learning al- responders. Methods: The patients received Nivolumab at a dose of 3 mg per
gorithm with performance assessed by receiver operating characteristic kilogram of body weight every 2 weeks. Tumor response was assessed with the
(ROC) curve and accuracy by confusion matrix. Results: The model selected use of the Response Evaluation Criteria in Solid Tumors (RECIST), version 1.1,
the top-10 baseline inflammatory cytokines to form a composite cytokine at week 8 and every 8 weeks thereafter. PBMCs were analyzed with a 18-color
signature, which predicted NIVO CL (high vs low) that was significantly microfluorometer, LSR Fortessa and a masscytometer, CyTOF. Results: The
associated with OS (p , 0.001) across all 3 studies (training and test responder-type patient group whose prediction formula values were greater
datasets). The same cytokine features were associated with OS of everolimus than 192 showed significantly longer PFS (P, 0.0001) and OS (P, 0.0001).
(p , 0.01), suggesting the potential prognostic nature of the composite The long survivors who consisted of tail plateau of PFS exhibited significantly
signature. The PK/PD analysis provided a robust description of the asso- more CD62LlowCD4+T cells than the short-term responders as pre-existing
ciation between selected cytokines and CL (ROC = 0.71). Identified cyto- immunity. The remaining responders kept significantly higher percentages of
kines (eg, serum C-reactive protein known to reflect immune cell CD62LlowCD4+T cells (P= 0.0088) and prediction formula values (P= 0.017)
modulation) have been shown individually to be associated with RCC than the patients with acquired resistance. Conclusions: The pre-existing CD4+T
prognosis. A multivariable approach resulting in tumor-specific composite cell balance between primed effector and regulatory T cells correlated with
signatures may provide more accurate prognostic value. Conclusions: The anti-PD-1 therapy response. Further, CD62Llowcell-dominant CD4+T cell
baseline composite cytokine signature could serve as a clinically useful immunity was required to maintain durable antitumor reactivity induced by
biomarker for patients with RCC, pending further evaluation, as it may anti-PD-1 antibody therapy. These results have important clinical implication,
provide improved prognostic accuracy for long-term clinical outcome as they support anti-PD-1 therapy provision to all potentially responding pa-
compared to individual cytokines. tients and pave the way for new treatment strategies for patients with distinct
CD4+T cell immune statuses. Clinical trial information: UMIN000020719.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Immunotherapy and Tumor Immunobiology 117s

2546 Poster Session (Board #190), Sat, 8:00 AM-11:00 AM 2547 Poster Session (Board #191), Sat, 8:00 AM-11:00 AM
Molecular circulating tumor DNA response to identify long-term survival in ctDNA analysis for personalization of consolidation immunotherapy in localized
patients receiving immunotherapy with initial radiologic stable disease. First non-small cell lung cancer. First Author: Everett J Moding, Stanford University
Author: Matthew David Hellmann, Thoracic Oncology Service, Department School of Medicine, Stanford, CA
of Medicine, Memorial Sloan Kettering Cancer Center and Weill Cornell
Background: Detection of molecular residual disease via circulating tumor
Medical College, New York, NY
DNA (ctDNA) analysis after chemoradiation (CRT) in localized non-small cell
Background: Early on-treatment changes in ctDNA may identify responders lung cancer (NSCLC) predicts risk of relapse. We explored the hypotheses
to immunotherapy and complement radiologic assessment of benefit. Here that (1) patients with undetectable ctDNA after CRT may not require con-
we investigate how early changes in ctDNA associate with long-term survival solidation immunotherapy (CI) and (2) ctDNA analysis could monitor the
following treatment with immunotherapy, and if differential patterns in effectiveness of CI in patients with residual ctDNA after CRT. Methods: We
molecular ctDNA response (MCR) among patients with radiologic stable applied CAPP-Seq ctDNA analysis to 88 plasma and matched leukocyte
disease (SD) at first on-treatment scan could identify patients deriving samples collected pre-CRT, post-CRT but pre-CI, and mid-CI in 22 patients
benefit from treatment. Methods: Paired pre- and on-treatment (week 6-8) with Stage IIB-IIIB NSCLC treated with CRT followed by CI. Identification of
plasma samples from 3 cohorts of patients treated with durvalumab (D) +/- patient-specific tumor variants was performed using tumor tissue or pre-
tremelimumab (D+T) were evaluated (NCT01693562, NCT02087423, treatment plasma, and ctDNA was quantified using a tumor mutation-
NCT02261220). CtDNA was profiled with the 73-gene Guardant 360 assay. informed bioinformatic strategy. Freedom from progression (FFP) defined
Nonsynonymous variants were summarized per patient to calculate variant radiographically by RECIST 1.1 criteria was compared in patients with
allelic frequency changes (dVAF) and on-treatment variant allele frequency ctDNA detected or not detected at pre-CI and mid-CI landmarks.
(pVAF). A combination of dVAF and pVAF was used to define MCR. Results: Median follow up from the start of CRT was 11 months. ctDNA
Results: The reduction of ctDNA (dVAF,0) and undetectable on-treatment detection was associated with inferior rates of FFP when compared to pa-
ctDNA (pVAF=0) were each associated with improved OS and PFS. An optimal tients with ctDNA not detected both pre-CI (12-month 33% vs. 76%, P =
threshold for MCR was determined from one cohort, then applied to the other 0.015, HR 7.51, 95% CI 1.47-38.24) and mid-CI (12-month 0% vs. 86%,
cohorts. MCR associated with significantly improved PFS and OS across all P , 0.0001, HR 123.3, 95% CI 16.21-937.8). In patients with un-
three cohorts (Table). MCR was then applied to a pooled subgroup of patients detectable ctDNA after CRT, FFP was similar to a historical cohort of patients
with initial radiologic SD from all three cohorts (n=78). Patients with radio- with undetectable ctDNA after CRT alone (12-month 88% vs. 87%, P =
logic SD and MCR were significantly more likely than those without MCR to 0.56, HR 0.55, 95% CI 0.07-4.18), suggesting that such patients may not
achieve radiologic CR or PR (pooled Odds ratio 12.7, p,0.001), had improved benefit from CI. All patients with detectable ctDNA pre-CI in whom ctDNA
PFS (stratified pooled HR 0.36, p,0.001), and improved OS (stratified pooled increased mid-CI developed progressive disease. Finally, in 2 patients with
HR 0.38, p=0.005). Conclusions: MCR is an early on-treatment tool that may ctDNA detected after CRT, CI led to elimination of ctDNA at the mid-CI
identify patients with improved long-term survival and patients with radiologic timepoint. One of these patients developed an isolated local recurrence
SD who derive clinical benefit from immunotherapy. MCR may be a supportive 22 months after CRT and the other patient is currently disease free at
endpoint in prospective clinical trials. MCR and survival benefit. 11 months, suggesting clinical benefit from CI. Conclusions: Our results
suggest that ctDNA analysis may allow personalization and response
NCT01693562 NCT02087423 NCT02261220 monitoring of CI following CRT for NSCLC. Validation in more patients
Study D (n=72) D (n=71) D+T (n=36)
followed by prospective testing in clinical trials will be required to establish
MCR, n (%) 34 (47) 28 (39) 16 (44) clinical utility of such an approach.
PFS HR (95% CI) 0.28 (0.16,0.50)⁂ 0.40 (0.22,0.73)** 0.12 (0.05,0.32)⁂
OS HR (95% CI) 0.25 (0.13,0.49)⁂ 0.34 (0.14,0.83)* 0.15 (0.05,0.44)⁂
*,0.05, **,0.01, ⁂,0.001

2548 Poster Session (Board #192), Sat, 8:00 AM-11:00 AM 2549 Poster Session (Board #193), Sat, 8:00 AM-11:00 AM
Phase 1 study of LY3022855, a colony-stimulating factor-1 receptor (CSF- Safety and immunobiological activity of guadecitabine sequenced with
1R) inhibitor, in patients with metastatic breast cancer (MBC) or metastatic ipilimumab in metastatic melanoma patients: The phase Ib NIBIT-M4
castration-resistant prostate cancer (MCRPC). First Author: Karen A. Autio, study. First Author: Anna Maria Di Giacomo, Medical Oncology and Im-
Memorial Sloan Kettering Cancer Center, New York, NY munotherapy, University Hospital of Siena, Siena, Italy
Background: Tumor-associated macrophages (TAM) correlate with increased Background: DNA hypomethylating agents show broad immuno-modulatory
invasiveness, growth, and immunosuppression. Activation of CSF-1R results activity in neoplastic cells, and may improve the effectiveness of cancer
in proliferation, differentiation, and migration of monocytes/macrophages. immunotherapies. The phase 1b NIBIT-M4 trial investigated a previously
CSF-1R inhibition with LY3022855 (LY), a human immunoglobulin G sub- unexplored therapeutic strategy using the next-generation DNA hypo-
class 1 (IgG1) monoclonal antibody (mAB), may have favorable anti-tumor methylating agent guadecitabine sequenced with ipilimumab for the
effects. We evaluated the safety and clinical response of LY monotherapy. treatment of advanced melanoma. Methods: Patients with unresectable
Methods: Patients (pts) with advanced refractory MBC and MCRPC received Stage III/IV melanoma received escalating doses of guadecitabine 30, 45 or
LY intravenously in 6-week cycles in cohorts: A) 1.25 mg/kg every 2 weeks 60 mg/m2 subcutaneously on Days 1–5 every three weeks, and ipilimumab
[Q2W]; B) 1.0 mg/kg on Weeks 1, 2, 4, and 5; C) 100 mg once weekly; D) 3 mg/kg intravenously on Day 1 every three weeks, starting one week after
100 mg Q2W. MCRPC pts were enrolled in cohorts A and B; MBC pts were guadecitabine, for four cycles. Primary endpoints were the safety, tolerability
enrolled in all cohorts. Anti-tumor activity was assessed using RECIST v1.1 by and maximum tolerated dose of treatment; secondary endpoints included
radiological imaging every 6 weeks. Results: Thirty-four pts (22 MBC; immune-related disease control and objective response. Genome-wide meth-
12 MCRPC) received $1 dose of LY. Median age was 57.0 years (range: ylation, RNA sequencing, and immunohistochemistry analyses were performed
32.0–81.0) for MBC pts and 72.5 years (range: 58.0–84.0) for MCRPC pts. on tumor samples collected at baseline, W4 and W12. (NCT02608437).
Baseline Eastern Cooperative Oncology Group performance status was 0 (n = Results: 19 patients were treated and evaluable for safety and efficacy. The
13, 38.2%), 1 (n = 18, 52.9%), or 2 (n = 3, 8.8%). MBC pts were hormone most common treatment-related adverse events of any grade were myelotoxicity
receptor (HR) positive (n = 20), HR negative (n = 1), or unknown (n = 1); 3 (n = 17; 89%) and immune-related adverse events (n = 12; 63%). Grade 3 or 4
MBC pts received concurrent hormone therapy. Common treatment-related myelotoxicity occurred in 15 (79%) patients. There were no dose limiting
adverse events of any grade were fatigue (38.2%), decreased appetite toxicities. Rates of immune-related disease control and objective response were
(26.5%), nausea (26.5%), increased lipase (23.5%), and increased creatine 8/19 (42%) and 5/19 (26%), respectively. Exploratory analyses of tumour
phosphokinase (20.6%). No complete or partial response was observed. samples (n = 8) showed that median CpG site methylation at Week 4 (74.5%)
Stable disease (SD) was observed in 5/22 MBC pts (duration 82–302 days) and Week 12 (75.5%) was significantly lower (p , 0.05) than at baseline
and 3/7 evaluable MCPRC pts (duration 50–124 days). Two MBC pts (9%; (80.3%), with a median of 2454 (Week 4) and 4131 (Week 12) differentially
Cohort A) had durable SD . 9 months and 1 pt had palpable reduction in a expressed genes identified compared to baseline; among the 136 pathways
nontarget neck mass. Circulating CSF1 and IL-34 increased at Day 8 sug- significantly modulated by treatment, the most frequently activated were
gestive of target engagement. Pharmacokinetics of LY were consistent with immune-related. Tumour immune contexture analysis (n = 11) demonstrated
other IgG1 mAbs. Conclusions: LY3022855 was well tolerated and showed up-regulation of Human Leukocyte Antigen (HLA) class I molecules on mel-
evidence of target engagement. Clinically meaningful SD . 9 months was anoma cells, and an increase in CD8+, PD-1+ T cells and in CD20+ B cells in
observed in 2 MBC pts. Tumor biomarker analyses are underway. Clinical trial post-treatment tumour core specimens. Conclusions: Sequential guadecitabine
information: NCT02265536. and ipilimumab is safe and tolerable in patients with metastatic melanoma, and
has promising immunological and anti-tumour activity. Clinical trial in-
formation: NCT02608437.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
118s Developmental Immunotherapy and Tumor Immunobiology

2550 Poster Session (Board #194), Sat, 8:00 AM-11:00 AM 2551 Poster Session (Board #195), Sat, 8:00 AM-11:00 AM
An open label, multicenter phase II study combining imprime PGG (PGG) Community and academic partnerships: Moving a new generation of clinical
with pembrolizumab (P) in previously treated metastatic triple-negative trials in NCI Community Oncology Research Program (NCORP) into community
breast cancer (mTNBC). First Author: Steven O’Day, John Wayne Cancer oncology practices. First Author: Worta J. McCaskill-Stevens, National
Institute, Santa Monica, CA Cancer Institute, Rockville, MD
Background: Checkpoint inhibitor (CPI) monotherapy shows limited clinical Background: NCORP is a model program that bridges academic and com-
response in previously treated mTNBC patients (pts) (Table). Agents are munity oncology practices and research. Over the past decade, community
needed that extend this benefit to more mTNBC pts. PGG is a novel, IV cancer investigators have adopted new technology, encountered new
administered PAMP that, in pts with 20ug/ml anti-beta glucan antibody treatment sequalae, and faced rising cost of care with its financial toxicity
(ABA+), activates innate immune cells. Preclinically, PGG reprograms myeloid imposed upon individuals seeking care. Opportunities are abundant for
cells to repolarize the immunosuppressive tumor microenvironment & en- community investigators to assess feasibility and uptake of research ad-
hance antigen presentation, driving T cell activation- the mechanistic basis vances into community practice settings, yet these opportunities are met
to explore PGG + P in mTNBC patients. Methods: 44 mTNBC pts ( 1 line of with the challenges of dynamic changes in types of organizations delivering
chemotherapy [Tx] for metastatic disease, ABA) received PGG (4 mpk IV cancer care and diversity of populations within their catchment areas. Little
weekly) + P (200 mg IV q3w) until PD or intolerable toxicity. 1° endpoints were information is shared about how and to what extent the health environment
ORR by RECIST v1.1 & safety. 2° endpoints included OS & DCR. CT scans (q6 influences this partnership and the implementation of a broad cancer re-
wks) were reviewed locally. Tumor biopsies (pre & 6 wks on Tx) & blood samples search portfolio. Methods: This abstract reports on the continued interest
were assessed for PGG-mediated immune activation. Results: Table shows and participation of community oncologists in research which is demonstrated
IMPRIME 1 clinical response data (Keynote086, PCD4989g shown for con- by 987 practices with over 4000 investigators in NCORP. Since 2014, over
text). Confirmed response was also evident in pts with liver or visceral me- 30,000 individuals enrolled in symptom management, screening, surveil-
tastases, high LDH. 10 IMPRIME 1 pts were originally ER/PR+, received lance, quality of life, and treatment trials. An additional 4500 patients and
hormonal Tx and progressed to TNBC. Of these, 5 are confirmed PR, 4 SD (3 still clinicians have enrolled in care delivery studies. Results: NCORP has been
on Tx), 1 PD. No unexpected safety signals were observed. Conclusions: These central in evaluating the most effective strategies for investigators to effec-
are the first clinical data to suggest that PGG provides added clinical benefit for tively communicate to patients the science of genomically-driven trials. It has
pts with previously treated mTNBC and support further development of PGG + P also provided ways of bringing the pediatric and AYA patients access to the
for mTNBC. Clinical trial information: NCT02981303. most up-to-date treatment strategies and new therapies in their community.
This creates the least disruption on family structure/dynamics, diminished
IMPRIME 1 Keynote 086a PCD4989gb
% (N= 44) % (N=170) % (N=94)
traveling requirements/costs, and reduced the financial burden. NCORP
promotes involvement of treating oncologists in research activities. This also
ORR 13.6 (6) 5.3 6.4 improves care for patients not enrolled in clinical trials. Therefore, NCORP
SD 40.9 (18) 18 13
PD 40.9 (18) 60.6 64 serves as a laboratory to determine the most effective strategies for co-
DCR - CR+PR+SD ‡ 24wks 22.7 (10) 7.6 10 management of cancer patients and survivors. Conclusions: Several ques-
Median OS in months 13.7 9.0 7.3 tions however remain to be addressed using this clinical trial model. These
% OS 6/ 12 months 83.0/ 65.5 69.7/ 39.8 60/ 37 include: how to continue to reduce disparities in cancer care and clinical trial
a
Adams 2018, bEmens 2019 In peripheral blood, Tx increased activated (HLA-DR/ participation; and, what are the best strategies for fostering implementation of
CD86+) monocyte & dendritic cell subsets as well as CD8 T cells (Ki67/HLA-DR/PD1+), cancer care models in community practice.
particularly in responsive pts. All tumor biopsy pairs showed heavy infiltration by
activated myeloid (PDL1+) and CD8 T cells (Ki67/granzyme B+) after Tx. These data
support the mechanistic basis for PGG-based combination with P. NCT02981303
sponsored by Biothera in collaboration with Merck & Co., Inc.

2552 Poster Session (Board #196), Sat, 8:00 AM-11:00 AM 2553 Poster Session (Board #197), Sat, 8:00 AM-11:00 AM
Re-evaluating eligibility criteria: Analysis of factors leading to nonpartici- Phase II study of spartalizumab (PDR001) and LAG525 in advanced solid
pation and outcomes of patients (pt) with advanced cancer who signed tumors and hematologic malignancies. First Author: Nataliya Volodymyrivna
consent but were not treated in early-phase immunotherapy (IO) trials. First Uboha, University of Wisconsin, Carbone Cancer Center, Madison, WI
Author: Roberto Carmagnani Pestana, University of Texas MD Anderson
Background: Spartalizumab and LAG525 are monoclonal antibodies tar-
Cancer Center, Houston, TX
geting PD-1 and LAG-3, respectively. Dual blockade of PD-1 and LAG-3 has
Background: Eligibility criteria protect the safety of trial pt and delineate the shown synergistic antitumor activity in preclinical models. Here we describe
study population. Excessively restrictive criteria, however, can negatively preliminary efficacy of spartalizumab + LAG525 across seven tumor types.
impact accrual and prevent access to beneficial investigational treatments. Methods: Phase II, open-label, parallel-cohort study was conducted in pts
Recently, ASCO issued a statement on the need to broaden eligibility criteria with solid or hematologic malignances relapsed and/or refractory to
and make trials more representative. We aim to characterize the factors standard-of-care therapies. Prior immunotherapy was prohibited. LAG525
leading to non-participation and the outcomes of pt who signed consent for (400 mg) and spartalizumab (300 mg) were dosed intravenously every
phase I IO trials but ultimately did not receive any therapy on that trial. 3 weeks. Primary endpoint was clinical benefit rate at 24 weeks (CBR24),
Methods: We identified 696 consecutive pt w/ advanced cancer who con- assessed using a Bayesian hierarchical model-based futility analysis. Pos-
sented to participate on IO phase I trials from 10/2015-12/2017, and terior probability that clinical benefit exceeds historical control (Pr) was
collected pt characteristics as well as clinical outcomes, and compared estimated to determine futility at interim against prespecified thresholds.
participants (P) to non-participants (NP). Results: Among the 696 pt who Results: As of January 7 2019, 76 pts received spartalizumab + LAG525; 72
initially consented to participate on IO phase I trials, 178 (25.6%) were pts were eligible for analysis (Table). The Pr cut-off for all arms was . 0.70.
never treated. Median age was 60 in both groups, and there were no dif- Hence, neuroendocrine tumors (NET), small cell lung cancer (SCLC) and
ferences regarding median number of metastatic sites (n = 2 vs 2 ) or sex diffuse large B-cell lymphoma (DLBCL) cohorts all met the expansion criteria
distribution (F 53% vs 54%); NP had received less lines of therapy (median = with Pr of 0.971, 0.975 and 0.804 respectively. CBR24 were as follows;
3 vs 4, p = 0.016). Reasons for non-participation were: 48 (26%) alternate NET: 0.86, SCLC: 0.27, DLBCL: 0.43. Prostate, sarcoma and ovarian co-
therapy (for 18, geography was the main reason), 29 (16%) clinical pro- horts did not meet the expansion criterion (Pr . 0.70) but were not declared
gression/ decline in PS, 14 (8%) did not have enough biopsy tissue, 13 (7%) futile; enrollment was paused pending results of next analysis. Gastro-
new lab abnormality, 11 (6%) new brain mets, 63 (35%) had other reasons esophageal (GE) cancer cohort was stopped for futility due to Pr (0.071)
(death, concurrent medications, financial factors). Median time from sig- below the futility threshold. Conclusions: Spartalizumab and LAG525
nature of consent to final exclusion of trial was 19 days (0-82). 54 of NP showed promising activity in NET, SCLC and DLBCL that met expansion
eventually enrolled in other trial, including 29 in immunotherapy trial. criteria. The GE cohort was declared futile. Remaining cohorts are paused
Median overall survival (OS) was significantly lower for NP vs P (median 6.9 pending further analysis. Clinical trial information: NCT03365791.
vs 18.0, HR 0.5; p , .0001). Conclusions: One quarter of patients who Pts treated Pts with
signed consent for early-phase immunotherapy trials were unable to start on ‡24 weeks clinical benefit Observed Posterior Enrollment
study. NP had significantly decreased OS. Detailed examination of these Cohort (n) (n) CBR24 probability recommendation

reasons can lead to recognition of modifiable factors and streamline the SCLC 15 4 0.27 0.975 Expansion criteria met
GE cancer 12 2 0.17 0.071 Stopped for futility
pretrial period, to guarantee this vulnerable population has maximal access Prostate 11 5 0.46 0.432 Expansion threshold not
to start therapy on study. met
Sarcoma 10 4 0.40 0.629 Expansion threshold not
met
Ovarian 10 2 0.20 0.420 Expansion threshold not
met
NET 7 6 0.86 0.971 Expansion criteria met
DLBCL 7 3 0.43 0.804 Expansion criteria met

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Immunotherapy and Tumor Immunobiology 119s

2554 Poster Session (Board #198), Sat, 8:00 AM-11:00 AM 2555 Poster Session (Board #199), Sat, 8:00 AM-11:00 AM
Preliminary safety, efficacy, and pharmacokinetics (PK) results of KN046 Assessment of the Fanconi anemia repair pathway as a predictor of clinical
(bispecific anti-PD-L1/CTLA4) from a first-in-human study in subjects with activity of pembrolizumab (PEM). First Author: Miguel Angel Villalona-Calero,
advanced solid tumors. First Author: Jermaine Coward, Icon Cancer Care, Miami Cancer Institute Baptist Health South Florida, Miami, FL
Brisbane, Australia
Background: Given the activity of immune checkpoint inhibitors (ICI) in
Background: KN046 is a novel bispecific antibody that blocks both PD-L1 mismatch repair deficient tumors, we evaluated if homologous recombination
interaction with PD1 and CTLA-4 interaction with CD80/CD86. KN046 has a repair deficiency associates with solid tumor response to ICI. Methods: We
wild type IgG1 Fc portion that preserves intact effector functions, such as conducted a phase 2 trial (NCT03274661) of PEM in metastatic solid tumor
depletion of Tregs in tumor microenvironments. This first-in-human study patients progressing on standard of care and for whom PEM had no FDA
evaluated the safety, tolerability, PK and preliminary efficacy of KN046 in approved indication. We evaluated a triple stain (FANCD2foci/DAPI/Ki67)
subjects with advanced solid tumors. Methods: This traditional “3+3” dose- immunofluorescence functional assay of the Fanconi Anemia pathway (FATSI)
escalation design study enrolled patients (pts) with advanced unresectable in treated patients’ archived tumors as a correlative biomarker. Patients with
or metastatic solid tumors refractory or intolerant to standard therapies. microsatellite unstable tumors were not eligible. The primary objective was
Previous treatment from PD1 or PD-L1 immune checkpoint inhibitors was objective response rate (iORR, CR+PR) by Immune Response Criteria, with the
allowed. KN046 was administered intravenously Q2W. Dose limit toxicity hypothesis that patients with FATSI negative tumors will have better clinical
(DLT) evaluation period is 28 days. The planned dose levels (DL) were 0.3, 1, outcome. Secondary objectives were progression free survival (PFS), 6 months
3, 5 and 10 mg/kg. Efficacy evaluation was performed by RECIST 1.1 every PFS and survival. PEM was given every 3 weeks and computed tomography
8 weeks. Results: As of Dec 13, 2018, 10 pts had been enrolled (0.3 mg/kg, scans were performed every 6 weeks. We utilized a two-stage phase II trial
n = 1; 1 mg/kg, n = 3; 3 mg/kg, n = 3; and 5 mg/kg, n = 3). Median duration of design to detect an iORR $ 20% in the whole population tested vs. the null
treatment was 8 (range: 2-24) weeks. 1 DLT was observed at 5 mg/kg dose (a hypothesis that the true iORR #5%. If $ 2 of the first 20 evaluable patients
grade 3 immune-related hepatitis without elevation in total bilirubin; re- had an objective response the trial proceeded to full accrual of 39 evaluable
versible in two weeks). The most common ($30%) treatment-emergent AEs patients. Outcomes were evaluated according to FATSI staining. Results: 42
(TEAE) were Fatigue, Diarrhea, Nausea, Vomiting. Six immune-related patients (40 evaluable) (35F,7M; median age 62[36-83]) enrolled. Median #
TEAEs (Abdominal pain lower, Arthralgia, Hepatic function abnormal, Hy- of prior regimens was 2[1-7]. Primary Dx included ovarian/fallopian (13),
perthyroidism, Nausea and Transaminitis) were observed in 3 pts. One pt endometrial (10), colorectal (3), cervix (2), pancreatic(2), vaginal (2) and 1
with NSCLC from 3 mg/kg cohort had confirmed completed response. Two each of various others. No unexpected toxicities occurred. Response evalu-
pts (TNBC and nivolumab refractory RCC) from 1 mg/kg cohort had shown ation showed 2 CR, 5 PR, 11 SD, 22 PD and 2 NE (iORR 18%). FATSI tumor
long-term stable disease ( . 12 weeks). Faster clearance of KN046 was analyses results are available in 34 patients; 25 FATSI positive, 9 negative. 2
observed at lower dose might be due to target-mediated clearance. T1/2 is PR, 8 SD, 14 PD, 1 NE occurred among the FATSI (+) (iORR 8%) and 2 CR, 2
approximately 7~9 days at doses of 3 mg/kg and above when saturation PR, 2 SD, 3 PD among the FATSI (-) patients (iORR 44%). mPFS and 6m-PFS
occurs. Conclusions: Single agent KN046 has an acceptable safety profile were 54 days and 12% (3/25) in FATSI (+), versus 248 days and 56% (5/9) in
and is in line with previously reported safety data from other immune FATSI (-) patients; p = 0.017. Conclusions: PEM has meaningful antitumor
checkpoint inhibitors. Preliminary efficacy results are promising. PK data activity in non MSI-high malignancies with no current FDA approved in-
from initial 4 cohorts support Q2W schedule. The study is currently ongoing dications. Evaluation of FATSI as a biomarker supports a biomarker selected
at dose level of 5 mg/kg Q2W. Clinical trial information: NCT03529526. population approach. Clinical trial information: NCT03274661.

2556 Poster Session (Board #200), Sat, 8:00 AM-11:00 AM 2557 Poster Session (Board #201), Sat, 8:00 AM-11:00 AM
Preliminary results with tislelizumab, an investigational anti-PD-1 antibody, Association between MDM2/MDM4 amplification and PD-1/PD-L1 inhibitors-
in Chinese patients with nasopharyngeal cancer (NPC). First Author: Siyang related hyperprogressive disease: A pan-cancer analysis. First Author: Wei-
Wang, The Fifth Affiliated Hospital, Sun Yat-sen University, Guangzhou, qiang Ju, Department of Organ Transplantation, The First Affiliated Hospital,
China Sun Yat-sen University, Guangzhou, China
Background: Epidemiology of NPC is characterized by a unique geographic Background: Immune checkpoint inhibitors have demonstrated a clear
distribution, with China having one of the highest incidence rates of NPC survival benefit in various tumor types. However, accelerated disease pro-
worldwide. Tislelizumab is an investigational monoclonal antibody with high gression, documented as hyperprogressive disease (HPD), was reported in a
affinity and specificity for PD-1. Tislelizumab was engineered to minimize subset of patients treated with PD-1/PD-L1 inhibitors. Until now, the
binding to FcɤR on macrophages in order to abrogate antibody-dependent mechanisms underlying HPD have not been elucidated. Previous studies
phagocytosis, a mechanism of T-cell clearance and potential resistance to anti- have demonstrated that MDM2/MDM4 amplification were associated with
PD-1 therapy. Previous reports from this phase 1/2 study (CTR20160872) have HPD. In the present study, we evaluated the relationship between MDM2/
shown that single-agent tislelizumab was generally well tolerated and dem- MDM4 amplification and HPD. Methods: We reviewed extensive clinical
onstrated preliminary antitumor activity in Chinese patients (pts) with advanced trials of PD-1/PD-L1 inhibitors in advanced solid tumor patients updated to
solid tumors. In the dose-verification part of this study, the recommended dose January 2019, and estimated the incidence of HPD, which was defined as
was established as 200 mg IV Q3W. Here we present preliminary results from time-to-treatment failure (TTF) , 2 months, and . 50% increase in tumor
the NPC cohort of this study. Methods: Chinese pts with advanced or meta- burden compared with pre-immunotherapy imaging in this study. The
static, histologically or cytologically confirmed WHO type II-III NPC were en- proportions of MDM2/MDM4 amplification across different cancer types
rolled in the indication-expansion phase of this study. Enrolled pts received were obtained from The Cancer Genome Atlas (TCGA) and our own database
tislelizumab 200 mg IV Q3W until unacceptable toxicity, consent withdrawal, or respectively. Then we plotted the incidence of HPD and the corresponding
no evidence of continued clinical benefit. Antitumor activity (per RECIST v1.1) proportion of MDM2/MDM4 amplification across various cancer types in
and safety/tolerability (per NCI-CTCAE v4.03) were assessed. Results: As of 11 TCGA. Results: Overall, 19 published clinical trials of 1318 patients treated
May 2018, 20 NPC pts (median age 49 yr [range 35–61]) were enrolled. Most with PD-1/PD-L1 inhibitors were included for analysis, covering 12 types of
pts were male (85%) and non-smokers (65%). All pts received prior radio- solid cancers. The incidences of HPD among these studies were ranging from
therapy; 19 pts (95%) received $1 line of systemic treatment and the median 1.58% in renal clear cell carcinoma to 24.3% in sarcoma. Correspondingly,
number of prior lines of systemic treatment was 2 (range 0–10). At the cut-off the proportions of MDM2/MDM4 amplification for these cancer types in
date, 15 pts remain on treatment and the median study follow-up was 5.5 mo TCGA were 0.74% in renal clear cell carcinoma to 20.38% in sarcoma. In
(range 0.46–9.0). Of 15 response-evaluable pts, 3 achieved a confirmed partial our database, in total, 60 patients with MDM2/MDM4 amplification were
response (PR) and 9 achieved stable disease; 1 patient had an unconfirmed PR. identified in 2931 patients with the highest proportion of MDM2/MDM4
Seven patients experienced $1 treatment-related AE (TRAE); hypothyroidism amplification in sarcoma (22 of 152, 14.5%). A significant correlation was
(n = 3) was the only TRAE that occurred in $2 pts. No grade $3 TRAEs or detected between the incidence of HPD and the corresponding proportion of
serious AEs were reported. Furthermore, no AEs led to either treatment in- MDM2/MDM4 amplification in TCGA across various cancer types (P ,
terruption or discontinuation. Conclusions: Tislelizumab was generally well 0.001, R2 = 0.67). Conclusions: Our results suggest that MDM2/MDM4
tolerated and demonstrated antitumor activity in previously treated pts with amplification may be associated with rapid disease progression in patients
advanced NPC. Clinical trial information: CTR20160872. receiving PD-1/PD-L1 inhibitors among various tumor types. The exact
mechanisms underlying HPD are needed to be further evaluated.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
120s Developmental Immunotherapy and Tumor Immunobiology

2558 Poster Session (Board #202), Sat, 8:00 AM-11:00 AM 2559 Poster Session (Board #203), Sat, 8:00 AM-11:00 AM
Antitumor activity and safety of MK-1308 (anti-CTLA-4) plus pembrolizu- Family history of cancer as surrogate predictor for immunotherapy with anti-
mab (pembro) in patients (pts) with non-small cell lung cancer (NSCLC): PD-1/PD-L1 immune checkpoint inhibitors: The FAMI-L1 study. First Author:
Updated interim results from a phase I study. First Author: Ruth Perets, Alessio Cortellini, Medical Oncology, St Salvatore Hospital, Department of
Rambam Medical Center, Technion - Israel Institute of Technology, Haifa, Biotechnological and Applied Clinical Sciences, University of L’Aquila,
Israel L’aquila, Italy
Background: An ongoing multicenter, open-label, phase 1 study of the Background: In the preliminary analysis of the FAMI-L1 study, we found a
anti–CTLA-4 antibody MK-1308 in combination with pembro in advanced significant association between family history of cancer (FHC) and better
solid tumors (NCT03179436) revealed a manageable safety profile and clinical outcomes with anti-PD1/PD-L1 inhibitors. Methods: We retro-
promising efficacy in pts with first-line (1L) advanced NSCLC. Data from a spectively evaluated advanced cancer patients treated with single agents
larger sample size and longer follow-up are presented. Methods: In dose PD1/PD-L1 inhibitors. Patients were categorized as follow: FHC-high (in
escalation (DE), pts with advanced solid tumors received MK-1308 by IV case of at least one cancer diagnoses in both straight and collateral family
administration at 25, 75, or 200 mg Q3W 31 cycle then in combination with line), FHC-low (in case of a cancer diagnoses in only one family line) and
pembro 200 mg Q3W 34 cycles followed by pembro monotherapy (up to 35 FHC-negative. FHC was collected till the second degree of relatedness.
cycles). In dose confirmation (DC), pts with 1L advanced NSCLC received Results: Between September 2013 and May 2018, 772 consecutive pa-
MK-1308 at 25 or 75 mg—Q3W or Q6W—plus pembro 200 mg Q3W (up to tients were evaluated. Median age was 68 years; male/female ratio was 521/
35 cycles). Safety (all treated pts), efficacy (subset of 1L NSCLC pts), 251. Primary tumors were: NSCLC (58.3%), melanoma (22.1%), renal cell
pharmacokinetics (PK, all treated pts), and PD-L1 tumor expression (subset carcinoma (16.6%) and others (3%). 114 patients (14.9%) had ECOG-PS $
of 1L NSCLC pts) were analyzed. Results: 213 pts were treated (DE, n=39; 2. 341 patients (44.3) were FHC-positive: 268 of them (34.75) were FHC-
DC, n=174). All pts were included in the safety analyses (median follow-up, low while 74 (9.6%) were FHC-high. FHC-high patients had a significantly
8 months); 113 pts from DC were included in the efficacy analyses (median higher incidence of irAEs compared to FHC-negative (55.4% vs 35.6%; p =
follow-up, 8 months). PK showed a dose-dependent increase in MK-1308 0.0012) and to FHC-low (41.4%; p = 0.0323). No significant differences
exposure. Neither target dose-limiting toxicity ($10%) nor maximum tol- were found in terms of ORR among subgroups (data not shown). At median
erated dose were reached for MK-1308 plus pembro; however, toxicity in- follow-up of 15.8 months, median PFS was 9.1 months (95%CI: 8.1-10.4;
creased with increasing MK-1308 dose and shorter dosing intervals. 452 events) and median OS was 19.7 months (95%CI: 15.7-24.4; 436
Treatment-related adverse events grade $3 occurred at the lowest rates censored). No significant differences were found regarding PFS (data not
at 25 mg Q3W in DE (0%) and 25 mg Q6W in DC (25%) and at the highest shown). Median OS of FHC-high patients was 31.6 months (95%CI: 26.2-
rates at 200 mg Q3W in DE (75%) and 75 mg Q3W in DC (50%). Efficacy 31.6; 50 censored patients), which was significantly longer than 18.2 months
was observed at all MK-1308 dose levels and intervals: confirmed ORR per (95%CI: 14.7-21.3; 229 censored patients) of FHC-negative patients (HR =
RECIST 1.1 by central review in 1L advanced NSCLC was 39% at 25 mg 0.60 [95%CI: 0.39–0.92), p = 0.0213). No significant differences in terms of
Q3W, 33% at 25 mg Q6W, 22% at 75 mg Q6W, and 25% at 75 mg Q3W; OS were found between FHC-high/low patients (data not shown). After
6-month PFS and OS rates are 67% and 89% for the 25 mg Q6W arm. adjusting for primary tumor, sex, treatment-line, number of metastatic sites
There was a 25% ORR in PD-L1–negative 1L advanced NSCLC pts. and ECOG-PS, FHC-high was confirmed an independent predictor of longer
Conclusions: MK-1308 plus pembro was generally well tolerated with no OS compared to FHC-negative (HR: 0.57 [95%CI: 0.37-0.88], p = 0.0098).
unexpected toxicity and conferred encouraging antitumor activity in 1L Conclusions: FHC-high seems to be an independent predictor for longer OS in
advanced NSCLC pts. Efficacy, safety, and PK data suggest that 25 mg given cancer patients treated with anti-PD-1/PD-L1. DNA damage and response
Q6W is the recommended phase 2 dose for MK-1308 in combination with (DDR) genes alterations may underlie that results.
pembro. Clinical trial information: NCT03179436.

2560 Poster Session (Board #204), Sat, 8:00 AM-11:00 AM 2561 Poster Session (Board #205), Sat, 8:00 AM-11:00 AM
IOLite: Multipart, phase 1b, dose-finding study of the PD-1 inhibitor dostarlimab Fatal adverse events associated with pembrolizumab in cancer patients: A
in combination with the PARP inhibitor niraparib 6 bevacizumab (bev), or with meta-analysis. First Author: Gol Minoo Golshani, Stony Brook University
platinum-based chemotherapy 6 bev for advanced cancer. First Author: Hospital, Stony Brook, NY
Nashat Y. Gabrail, Gabrail Cancer Center, Canton, OH
Background: Pembrolizumab, an immune checkpoint inhibitor (ICI) against
Background: Novel combinations of drugs may overcome resistance in pa- programmed cell death-1(PD-1) protein has emerged as an effective treat-
tients (pts) with solid tumors who had progressed on standard therapy. ment for many cancers. Although better tolerated than chemotherapy, it has
Methods: IOLite (NCT03307785) is a multicenter, open label, multipart unique immune related adverse event and little is known about its risk of fatal
study to determine dosing and evaluate safety and efficacy of dostarlimab in adverse events (FAE). Therefore, we conducted a meta-analysis of clinical
combination with approved therapies in pts with advanced solid tumors. Pts trials to determine the incidence and risk of fatal adverse events with pem-
were enrolled in each part based on tumor histology, prior treatment (tx) brolizumab. Methods: A systematic search for phase I-III clinical trials of
history, and physician preference. Primary endpoint is dose-limiting toxic- pembrolizumab was conducted using databases from PUBMED, and abstracts
ities (DLTs) deemed as tx-related per investigator, and safety and tolerability presented at the American Society of Clinical Oncology (ASCO) conferences
of the combination. Tumor responses were assessed per RECIST v1.1. until October 2018. Eligible studies included prospective clinical trials of
Results: Parts A-D (see Table) are fully enrolled. One complete response was pembrolizumab with available data on FAE. Data on FAE was extracted from
reported in part B (endometrial); confirmed partial responses in part A each study and pooled for calculations. Incidence, relative risk (RR) and 95%
(ovarian, small cell lung [SCLC], gastrointestinal stromal [GIST]); part B confidence intervals (CI) were calculated by employing fixed or random-effects
(breast [2], bladder, SCLC); part C (prostate, fallopian tube), and part D models. Results: A total of 11 clinical trials of pembrolizumab, with 3713
(endometrial, non-SCLC). Stable disease was reported in part A (colorectal, patients were included for analysis. The overall incidence of FAE with
prostate [2], breast, GIST, gastric); part B (SCLC, squamous cell, head and pembrolizumab was 1.2% (95% CI: 0.5-2.8%).The incidence of FAE sig-
neck, prostate); part C (pancreatic, ovarian, GIST, breast [2], liver, endo- nificantly varied among different tumor types (P=0.02), ranging from 0.2% in
metrial); part D (ovarian, head & neck, cholangiocarcinoma). At data cutoff, melanoma to 3.1% in breast cancer.The incidence of FAE was significantly
24 pts remain on treatment. PK’s of dostarlimab and niraparib (nir) were not higher (P,0.001) with chemotherapy plus pembrolizumab (7.0%, 95%CI:
altered by any of the combination agents tested. Conclusions: Dostarlimab is 4.9-10%) as compared to pembrolizumab alone (0.7%, 95% CI: 0.4-1.2,
well tolerated in combination with nir 6 bev, or carbo-pac 6 bev. Preliminary p=,0.001). There was no significant difference in the risk of FAEs when
efficacy data show responses in various histologies. No new safety signals pembrolizumab was compared with chemotherapy with RR=1.24 (95% CI:
were identified. Clinical trial information: NCT03307785. 0.8-1.89, P=0.31). Conclusions: Pembrolizumab is similar to chemotherapy
in the risk of fatal adverse events in cancer patients. Combination of pem-
Part Dose N DLT Grade
brolizumab with chemotherapy increased the risk of FAE in comparison with
A Dostarlimab + nir 200 mg 16 Mucosal inflammation (n = 1) 3
Hypertension (n = 1) 3
pembrolizumab alone. Further studies are needed to identify risk factors.
Dostarlimab + nir 300 mg 6 None Incidence and relative risk of FAE with pembrolizumab by tumor type and with combination chemotherapy.
B Dostarlimab + carbo-pac 14 Aspartate aminotransferase increased (reversible) 3 Number of studies Incidence% (95% CI) RR (95% CI) P value
(n = 1)
C Dostarlimab + nir 200 mg + bev 6 Carotid artery intimal tear (n = 1) 3 Type of treatment.
Pembrolizumab alone 8 0.7 (0.4-1.2) 0.69 (0.3-1.5) 0.3
Dostarlimab + nir 300 mg + bev 7 Neutropenia (n = 1) 4 Pembrolizumab & Chemo 3 7 (4.9-10) 1.58 (0.9-2.5) 0.07
D Dostarlimab + carbo-pac + bev 6 None Tumor type.
NSCLC 6 2 (08-5.2) 1.27 (0.8-1.9) 0.9
Dosing: Dostarlimab: 500 mg IV Q3W34, then 1000 mg Q6W; Urothelial 2 0.8 (0.2-3.9) 0.94 (0.2-3.7) 0.9
Niraparib: 200 or 300 mg QD; Melanoma 2 0.2 (0-1.3) 1.5 (0.1-36.8) 0.8
Breast 1 3.1 (0.4-19.1) N/A* N/A*
Carboplatin: area under curve 5 or 6 Q3W; Overall 11 1.2 (0.5-2.8) 1.24 (0.8-1.9) 0.3
Paclitaxel: 175 mg/m2Q3W; Bev: 15 mg/kg Q3W
*No control arm available

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Immunotherapy and Tumor Immunobiology 121s

2562 Poster Session (Board #206), Sat, 8:00 AM-11:00 AM 2563 Poster Session (Board #207), Sat, 8:00 AM-11:00 AM
Open-label, multicenter, phase I study to assess safety and tolerability of Early incidence of immune-related adverse events (irAEs) predicts efficacy in
adavosertib plus durvalumab in patients with advanced solid tumors. First patients (pts) with solid tumors treated with immune-checkpoint inhibitors
Author: Manish R. Patel, Florida Cancer Specialists and Sarah Cannon (ICIs). First Author: Chris Morehouse, MedImmune, Gaithersburg, MD
Research Institute, Sarasota, FL
Background: Treatment with ICIs can manifest immune-related adverse
Background: Adavosertib (AZD1775; A) is a highly selective inhibitor of WEE1. events (irAEs), which have correlated with clinical outcomes in certain tu-
This Phase I study (NCT02617277) investigated a range of doses and mors. However, timing of these events and how early irAEs correlate with
schedules for oral A plus IV durvalumab (DV), a human monoclonal antibody outcomes is unclear. We assessed whether early occurring irAEs could
targeting PD-L1, to determine the maximum tolerated dose (MTD) and rec- predict survival in pts treated with durvalumab (D), an anti-PDL1 and
ommended Phase II dose (RP2D) in patients (pts) with advanced solid tumors. combined with tremelimumab (D+T), an anti-CTLA4 in two clinical stud-
Methods: Four 28-day schedules (Sch) were evaluated with pts receiving DV ies. Methods: Two phase 2 non-randomized clinical trials evaluating D
1500 mg on day (d) 1 of each schedule (Table). Patients continued treatment if (D4190C00001 (1108), N=756; available data per internal data re-use
they showed clinical benefit in the absence of any discontinuation criteria. Pts policy) or D+T (D4190C00010 (C10), N=327; expansion and ICI naı̈ve
received A monotherapy for PK analysis prior to the start of combination therapy cohorts) in multiple solid tumor types were analyzed. Prevalence of pts
in Sch B, C (d –7 to –5) and D (d –9 to –5). MTD was determined using a 3+3 experiencing irAEs, regardless of grade was 30% and 59% in studies 1108
dose-escalation cohort design. Predefined dose-limiting toxicities (DLTs) were and C10, respectively, with most frequent including dermatitis/rash (25%),
evaluated during the first cycle of study treatment. Results: 54 pts received A thyroid (15%), diarrhea/colitis (14%), and pancreatic enzyme elevation
(most common primary tumor sites: colon, 19%; lung, 13%; breast, 11%). The
(5%). Overall survival (OS) was correlated with irAE timing prior to 6, 8, 12,
most common grade $3 AEs were fatigue (15%), diarrhea (11%) and nausea
16, 20 and 24 weeks following D or D+T treatment. Kaplan Meier and log-
(9%). DLTs were nausea (n = 2) and diarrhea (n = 1). 7 pts (13%) had A-related
SAEs, including reversible and confounded drug-induced liver injury (Sch B
rank analyses were used. Results: In both studies, pts who experienced at
125 mg and Sch C; 1 each). Disease control rate (DCR) for the total cohort was least one irAE by study completion had improved median OS (mOS, 1108:
36%. Preliminary PK at 150 mg BID suggests adequate coverage for cell kill 23.1 mos [18.2, 26.9]; C10: 16.3 mos [12.5, 31.4]) relative to those who
activity and no drug–drug interaction. Conclusions: The MTD/RP2D was A experienced none (1108: 6.3 mos [5.4, 7.3]; C10 4.6 mos [3.3, 6.1]).
150 mg BID (3 d on, 4 d off; treatment d 15–17, 22–24) with DV 1500 mg (d 1 Median time (weeks) to first and second irAE occurred earlier in C10
Q4W); safety profile was considered acceptable. Preliminary evidence of an- compared to 1108, 3.9 vs. 5.6 and 6.9 vs. 10.1, respectively. When as-
titumor activity was observed. Clinical trial information: NCT02617277. sociating timing of irAEs with survival, there was a significant differential in
mOS at each time interval evaluated between pts with at least one irAE and
A dose (days DLTs, Grade ‡3 AEs, n A discontinuations due to A dose reductions due to DCR,
Sch on/off) A treatment days N n (%) (%) AE, n (%) AE, n (%) n (%) those with none, with differentiation at 6 weeks and maximal survival benefit
A 125 mg BID 1–5, 15–19 6 2 4 (67) 1 (17) 1 (17) 1 at 24 weeks following treatment with D or D+T (Table). Conclusions: Early
B
(5/9)
125 mg BID 15–17, 22–24 7*
(33)
0 3 (43) 0 0
(17)
3
occurrence of irAEs may be predictive of survival benefit in pts treated with D
(3/4) (50)* or D+T. OS by 1+ or no irAE(s) occurring up to 6 or 24 weeks post treatment.
150 mg BID 15–17, 22–24 12 0 7 (58) 0 1 (8) 5
(3/4) (42)
175 mg BID 15–17, 22–24 7 0 5 (71) 1 (14) 2 (29) 4 No irAE 1+ irAE
(3/4) (57)
C 125 mg BID 8–10, 15–17, 7† 1 6 (86) 1 (14) 0 1 Study Time Interval (weeks) N mOS mos (95% CI) N mOS mos (95% CI)
(3/4) 22–24 (17)† (14)
D 200 mg QD (5/
2)
15–19, 22–26 6 0 3 (50) 0 0 2
(33)
1108 ,6 631 8.6 (7,9.8) 125 15.2 (9.5,20.2)
250 mg QD (5/ 15–19, 22–26 6 0 4 (67) 1 (17) 1 (17) 3 ,24 550 6.7 (5.9,8.3) 206 20.2 (15.4,24.3)
2)
300 mg QD (5/ 15–19, 22–26 3 0 2 (67) 0 1 (33)
(50)
0
C10 ,6 197 7.3 (5.6,9.8) 130 13.1 (10.8,22.9)
2) ,24 139 5 (3.6,6.5) 188 15 (11.4,31.4)
*Only 6 pts received DV; †DLT evaluable in 6 pts only

2564 Poster Session (Board #208), Sat, 8:00 AM-11:00 AM 2565 Poster Session (Board #209), Sat, 8:00 AM-11:00 AM
Severe immune-related adverse events in anti-PD-1-treated patients are Association between past medical history (PMH) of autoimmune events and
clustered into distinct subtypes by peripheral blood T-cell profiles. First adverse events of special interest (AESI). First Author: Jamie Renee Brewer,
Author: Kyung Hwan Kim, Graduate School of Medical Science and Engi- U.S. Food and Drug Administration, Silver Spring, MD
neering, Korea Advanced Institute of Science and Technology, Daejeon,
Background: PD-1/L1 inhibitor therapy has become the standard of care for
South Korea
many advanced solid tumors. A notable limitation of PD-1/L1 inhibitor
Background: Although anti-programmed death-1 (PD-1) treatment has therapy is the concern for AESI that are likely to be immune related. It is
shown remarkable anti-tumor efficacy, immune-related adverse events unclear whether a history of autoimmune events is a predisposing risk
(irAEs) develop with heterogeneous clinical manifestations. Immunological making these adverse events more likely. We aimed to evaluate the asso-
understanding of irAEs is currently limited. In the present study, we analyzed ciation between autoimmune-associated PMH and AESI on PD-1/L1 in-
peripheral blood T cells obtained from cancer patients who received anti-PD- hibitors. Methods: We pooled data across seven pivotal trials for PD-1/L1
1 treatment to determine the immunological characteristics of severe irAEs. inhibitors in urothelial cancer (UC) identifying patients with AESI submitted
Methods: This study included 31 patients with refractorythymic epithelial tumor from 2016 - 2017. AESI were determined using a pooled preferred term list
(TET) who were enrolled in a phase II trial of pembrolizumab (NCT02607631) for each trial. Information collected from trials included PMH, AESI events
and 60 patients with metastatic non-small cell lung cancer (NSCLC) who re- and toxicity grade. Results: In total, 1747 immunotherapy treated patients
ceived pembrolizumab or nivolumab. T-cell profiling was performed by multi- were identified, with 1068 (61%) having an AESI and 277 (26%) having an
color flow cytometry using peripheral blood obtained immediately before autoimmune-related PMH. The most common autoimmune-associated
treatment and 7 days after the first dose of anti-PD-1 antibodies. Results: Severe events in the PMH were thyroid disorders (64%), asthma (23%), atopic
irAEs ($ grade 3) occurred in 7 TET patients (22.6%) and 6 NSCLC patients dermatitis/eczema (6%), irritable bowel syndrome (5%), and psoriasis (4%).
(10.0%). Patients with severe irAEs exhibited a significantly lower fold increase AESI occurred in 68% of patients with an autoimmune-related PMH and
in the frequency of effector regulatory T (eTreg) cells after anti-PD-1 treatment, 60% of patients without an autoimmune-related PMH. The most common
higher ratio of T helper-17 (Th17) and T helper-1 cells at baseline, and higher AESI in patients with an autoimmune-related PMH were diarrhea (35%),
percentage of Ki-67+cells among PD-1+CD8+T cells post-treatment. In rash (27%), renal disorder (27%), and pruritis (23%). The most common
clustering analysis, patients with severe irAEs were grouped into four AESI in the general trial population were diarrhea (28%), pruritis (26%),
distinct subtypes: Th17-related, TNF-related,Treg-related, and CD8- rash (25%), increased creatinine (14%), and elevated AST and ALT (10%
related. Conclusions: Severe irAEs after anti-PD-1 treatment were clus- and 9%). The majority of events were Grade 1-2 (87% in patients with an
tered into four immunological subtypes, indicating that development of autoimmune-related PMH and 84% in the general trial population).
severe irAEs is not attributed to a single mechanism. Further investigations Conclusions: Pooled clinical trial data shows a slight numeric increase in
in larger cohorts are needed to validate our current findings. AESIs in patients with autoimmune-associated PMH. Limitations include
potential lack of consistency of PMH documentation and adverse event
reporting. There did not appear to be a pattern of association between PMH
and type of AESI event. Grades of AESI events in the population with
autoimmune-associated PMH were similar to the general trial population.
This suggests that PD-1/L1 inhibitors may be safely administered to patients
with UC and a PMH of some autoimmune-associated events. Further ex-
ploration is needed.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
122s Developmental Immunotherapy and Tumor Immunobiology

2566 Poster Session (Board #210), Sat, 8:00 AM-11:00 AM 2567 Poster Session (Board #211), Sat, 8:00 AM-11:00 AM
Characteristics of patients receiving immune checkpoint inhibitors (ICI) in Understanding contribution and independence of multiple biomarkers for
ASCO’s CancerLinQ. First Author: Wendy S. Rubinstein, American Society of predicting response to atezolizumab. First Author: Parantu K. Shah, Global
Clinical Oncology’s (ASCO) CancerLinQ, Alexandria, VA Development, EMD Serono Research & Development Inc., Billerica, MA
Background: ICI’s have demonstrated significant clinical benefit since the Background: No biomarker satisfactorily predict response to anti-PD-L1
first FDA approval in 2011 of ipilimumab for metastatic melanoma. Five therapies. Biomarker studies suffer from small sample size, presence of
additional ICI therapies have since been approved across several indications. disease subtypes, and lack of simultaneous measurement of multiple bio-
The objectives of this study were to describe the clinical and demographic markers. The IMvigor210 dataset (Mariathasan et al., Nature 2018) provides
features of patients receiving ICI treatment along with utilization patterns in baseline measurements for multiple biomarkers of response to atezolizumab
real-world settings. Methods: We conducted a retrospective, observational (n range: 105-298) coupled with genomewide RNAseq profiles. We ex-
cohort study using statistically de-identified data from January 2011 to amined predictive performance of individual biomarkers and combined
November 2018 in CancerLinQ, ASCO’s real-world oncology database, information from multiple biomarkers to measure changes in predictive
which now contains EHR data from 49 diverse oncology practices in the U.S. performance. Methods: We built classification models (PR/CR vs. PD/SD)
Adult patients diagnosed with any cancer type who received $1 dose of an using genes and gene sets that provide information on pathways (mSigDB),
ICI (see Table) and had $2 clinical visits were eligible for inclusion. Patients immune components (xCell, Cibersort), and predictors of response (IMPRES,
were excluded if they received an ICI prior to its first FDA approval date to Immunophenoscore, and TIDE). Prognostic features were removed based on
avoid inclusion of clinical trial patients. Descriptive statistics were used to survival association in TCGA. All experiments were done with repeated five-
examine treatment patterns and clinical characteristics of patients receiving fold double cross validation. Predictions from the gene sets model were used
ICIs. Results: This analysis included 12,712 patients who received an ICI. as a single biomarker. PD-L1 expression by IHC in tumor core and immune
Median patient age was 67.4 years [IQR 59.3, 75.3]; 58% were male. White cells, tumor mutation burden(TMB), neo-antigen burden (NB), location of
race made up the highest percent (83%) of ICI patients, followed by Black metastatic disease, immune phenotype and genomic subtypes were then
race (9%) and Other (8%). The most common primary cancers at the start of systematically merged with the gene set based model. Results: NB was the
treatment were lung cancer (36%), melanoma (8%), urothelial cancer (2%) best predictor of response (AUC 0.77), while a model combining NB, TMB,
and renal cell carcinoma (2%). Of the 8,444 patients with known disease ECOG and expression signatures was marginally better (AUC 0.81) with a
stage, 5,446 (64%) had Stage IV cancer. Breakdown of ICI treatment patterns chance of over fitting. Chi-square tests for independence suggested that
can be found in the accompanying table. Uptake of ICIs was the most rapid for examined biomarkers do not provide independent information explaining
nivolumab, which had the highest use (49%), followed by pembrolizumab for lack of increase in AUC. Signatures for TP53 mutations, M1 macrophages,
rapid adoption and use (30%). Conclusions: This analysis gives insights into CD8+ T effector cell and DNA repair, among others, were present frequently
patient characteristics and real-world treatment patterns for ICIs. ICIs were in classification using gene expression information (AUC 0.71), suggesting
used most widely in males, lung cancer patients and patients with advanced their independent contributions to response. Adding gene expression in-
disease. These baseline characteristics inform our analyses of ICI use in formation to NB didn’t improve AUC for response but provided better survival
patients with autoimmune disease, also reported herein. stratification. Conclusions: Integration of examined biomarkers with ma-
ICI Medication N (%)
chine learning did not improve response prediction significantly. We are now
examining sizes of subgroups defined by combination of low NB/TMB with
Atezolizumab 644 (5)
Avelumab 22 (0) these biomarkers.
Combination ICI 619 (5)
Durvalumab 116 (1)
Ipilimumab 1303 (10)
Nivolumab 6219 (49)
Pembrolizumab 3789 (30)

2568 Poster Session (Board #212), Sat, 8:00 AM-11:00 AM 2569 Poster Session (Board #213), Sat, 8:00 AM-11:00 AM
Cardiovascular complications of immune checkpoint inhibitor therapy. First Organ dysfunction (dys) and clinical outcomes in patients (pts) treated with
Author: Samip R. Master, Louisiana State University Health Sciences immune checkpoint inhibitors (ICIs). First Author: QI LIU, US Food and Drug
Center, Shreveport, LA Administration, Silver Spring, MD
Background: Cardiac toxicity has largely been underestimated toxicity of Background: ICIs (anti-PD-L1/PD-1/CTLA-4) are approved in multiple can-
checkpoint inhibitors. There have been several cases of myocarditis and cers. The impact of organ dys on the pharmacokinetics of ICIs is known, but
fatal heart failure reported in patients treated with checkpoint inhibitors. We associated clinical outcomes are not well characterized. We compared real-
did a retrospective analysis of data of adverse effects of drugs that has been world (rw) clinical outcomes in ICI-treated pts by liver and renal function.
made available to public by the FDA. Methods: The FDA has made the data Methods: This retrospective study used longitudinal, patient-level data from
on adverse effects of various treatments available to general public through community practices in the Flatiron Health electronic-health record (EHR)-
the FDA Adverse Events Reports System (FAERS) public dashboard. We derived database. We included pts diagnosed with advanced cancers (NSCLC,
investigated the cardiac toxicities of various immune check point inhibitor renal cell, melanoma, gastric/esophageal, or head and neck) on or after 1/1/
therapies available at FDERS for the years 2017-2018. Results: The 2011, treated with an ICI with follow-up through 12/31/2018 and with
reviewed the reported side effects of pembrolizumab, nivolumab, atezoli- baseline liver or renal function results in the EHR #30 days prior to ICI start.
zumab, avelumab, durvalumab and ipilimumab from FDA data. A total of Organ function was stratified as normal, mild, moderate, or severe dys based
36,848 toxicities from immunotherapies were reported. Out of that, on NCI CTCAE. We computed unadjusted median estimates for rw time to
2316(6.2 %) were cardio toxicities and 816 were fatal. The most common treatment discontinuation (rwTTD) for any reason and overall survival (OS)
cardiac complications were as follows: myocarditis (15%), atrial fibrillation across baseline groups using the Kaplan-Meier method. Results: Of 15,979
(13%), pericardial disease including pericardial effusion (13%), cardiac pts, we identified 12,978/12,840 pts with evaluable renal/liver function,
failure (17%) and coronary artery disease (19%). Approximately 50%, 43%, respectively; median follow-up was 5.1 mos and median age was 69.0 yrs
40% 22% and 15 % of cases with myocarditis, ischemic heart disease, (IQR: 61.0, 76.0) for both. Most pts had NSCLC (69.4/69.0%), were men
cardiac failure, atrial fibrillation and pericardia disease were fatal. (60.1/60.0%), white (73.5/73.6%), and diagnosed at stage IV (58.7%/
Conclusions: Out of the reported cases of adverse reaction to check point 58.6%). Most ICI was given in 1st-line (42.3/42.1%) (outcomes in Table).
inhibitor, 6.2% were cardio toxicities. 35% of cardio toxicities were fatal. Conclusions: Pts with categorically worse baseline liver function had pro-
Half of the cases who developed myocarditis died. There was no statistical gressively worse on-treatment outcomes, including shorter OS, which differed
difference in rate of cardiotoxicities caused by PD1, PDL1 or CTLA 4 from trends in renal dys. Whether baseline dys is prognostic or predictive of ICI
inhibitors. outcomes should be further investigated in addition to reasons for discon-
tinuation. Clinical outcomes (unadjusted median times, mos [95% CI]) by
organ function.
Normal Mild Moderate Severe
RENAL, n (%) 10,996 (84.7) 1,593 (12.3) 324 (2.5) 65 (0.5)
OS 10.0 11.4 10.0 8.2
(9.6, 10.4) (10.6, 12.4) (7.9, 11.8) (7.4, 14.8)
rwTTD 3.4 (3.2, 3.4) 3.7 (3.3, 4.1) 3.4 (3.1, 4.2) 3.4 (2.1, 7.3)
LIVER, n (%) 11,116 (86.6) 1,495 (11.6) 159 (1.2) 70 (0.5)
OS 10.8 6.8 (5.8, 7.6) 3.3 (2.0, 5.8) 2.2 (1.3, 4.3)
(10.4, 11.2)
rwTTD 3.6 (3.4, 3.7) 2.3 (2.3, 2.8) 1.8 (0.9, 2.3) 0.9 (0.5, 1.4)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Immunotherapy and Tumor Immunobiology 123s

2570 Poster Session (Board #214), Sat, 8:00 AM-11:00 AM 2571 Poster Session (Board #215), Sat, 8:00 AM-11:00 AM
A phase II clinical trial of ipilimumab/nivolumab combination immunotherapy Outcomes after early initiation of nonsteroidal immunosuppressive therapy
in patients with rare upper gastrointestinal, neuroendocrine, and gynecological in patients with immune checkpoint inhibitor-induced colitis. First Author:
malignancies. First Author: Oliver Klein, Medical Oncology Unit, Austin Health, Yinghong Wang, The University of Texas MD Anderson Cancer Center,
Heidelberg, Australia Houston, TX
Background: Patients (pts) with rare cancers represent an unmet medical Background: Current treatment guidelines for immune-mediated colitis
need and have an inferior overall survival compared to patients with more (IMC) recommend 4 to 6 weeks of steroids as first-line therapy, followed by
common malignancies. Due to their low frequency, no therapies, including nonsteroidal immunosuppressive therapy (NSIST) (infliximab or vedolizu-
immunotherapies, have systematically been investigated in this population. mab) in patients who do not respond to steroids. We assessed the effect of
Ipilimumab (ipi)/Nivolumab (nivo) combination treatment has demonstrated early NSIST introduction and number of NSIST infusions on clinical out-
significant clinical activity in pts with advanced melanoma and renal cell comes. Methods: We performed a retrospective review of patients with IMC
carcinoma and response rates with this regimen are higher compared to who received NSIST between January and December 2018. Logistic re-
single agent anti-PD-1 therapy. This phase II study assessed the efficacy and gression analyses were used to assess associations between clinical features
safety of ipi/nivo in rare cancer pts. Methods: 60 pts with advanced rare and outcomes of IMC (Table). Results: Of the 1,459 patients who received
upper gastrointestinal (GI), neuroendocrine (NE) and gynaecological (GY) immune checkpoint inhibitor, 179 developed IMC of any grade; 84 of them
malignancies were enrolled in 3 cohorts. Patients received nivo 3mg/kg and received NSIST. Of the 84 patients who received NSIST, 79% were males
ipi 1mg/kg every 3 weeks for four doses, followed by nivo 3mg/kg every 2 weeks. with mean age of 60. Compared with patients who received NSIST .10 days
Treatment continued for up to 96 weeks, or until disease progression or the after IMC onset, patients who received early NSIST (#10 days) required
development of unacceptable toxicity. Response (RECIST 1.1) was assessed fewer hospitalizations (P=0.03), experienced steroid taper failure less fre-
every 12 weeks. The primary endpoint was clinical benefit rate (CBR), CR, PR quently (P=0.03), had fewer steroid tapering attempts (P,0.01), had a
and SD. Exploratory endpoints include correlation of efficacy with relevant shorter course of steroid treatment (P=0.09), and had a shorter duration of
biomarkers including PDL1 status and tumour mutation burden. Results: 42 symptoms (P,0.01). Risk factors of IMC recurrence after weaning off
pts have so far undergone restaging, 11 pts clinically progressed prior to their steroids included: 1) needing multiple hospitalizations (P,0.01), 2) ex-
first restaging scan. 50 pts have received prior therapy (1-5 lines). Objective periencing steroid taper failure after NSIST (P=0.02), 3) receiving infliximab
responses have been observed in a range of different malignancies. Clinical trial rather than vedolizumab (P=0.02), 4) receiving fewer than three infusions of
information: NCT02923934. Grade 3/4 immune related adverse events were NSIST (P=0.02), 5) having higher fecal calprotectin levels after NSIST
detected in 31% of pts. The results of correlative biomarker studies will be (P=0.01), and 6) receiving a longer course of steroids (P=0.02), hospital-
presented at the meeting. Conclusions: Ipi/Nivo combination treatment has ization (P,0.01) and IMC symptoms (P,0.01). Unsuccessful weaning
efficacy in a wide range of advanced rare malignancies. Immune related toxicity is from steroids after NSIST was associated with high IMC grades (P,0.01);
in keeping with previously reported clinical trials using the same dosing regimen. multiple hospitalizations (P,0.01); steroid-resistant IMC (P,0.01); long
# pts 53 GI (16) NE (20) GY (17)
interval from IMC to NSIST initiation (P=0.01); and long duration of ste-
roids (P,0.01), IMC symptoms (P,0.01), and hospitalization (P,0.01).
ORR 32% 31% 25% 41%
3 pts Cholangioca 3 pts Atypical bronchial carcinoid 1 pt Ovarian clear cell ca Conclusions: NSIST should be introduced early in the disease course of IMC
2 pts Gallbladder ca 1 pt Pancreatic NET 1 pt Ovarian germ cell tumor instead of waiting until failure of steroid therapy or steroid taper. Patients
1 pt Adrenocortical ca 1 pt Uterine clear cell ca
1 pt Uterine carcinosarcoma who received three or more infusions of NSIST had more favorable clinical
1 pt Uterine leiomyosarcoma outcomes.
1 pt Vaginal SCC
1 pt low grade serous ovarian ca
SD 36% 19% 55% 29%
CBR 68% 50% 80% 70%
PD 32% 50% 20% 30%

2572 Poster Session (Board #216), Sat, 8:00 AM-11:00 AM 2573 Poster Session (Board #217), Sat, 8:00 AM-11:00 AM
Immunomodulation by HDAC inhibition: Results from a phase Ib study with Efficacy and a novel clinicopathologic-genomic nomogram of atezolizumab
vorinostat and pembrolizumab in metastatic urothelial, renal, and prostate in advanced non-small cell lung cancer (POPLAR and OAK): A combined
carcinoma patients. First Author: Roberto Pili, Indiana University School of analysis of two multicenter, randomized, phase II/III trials. First Author:
Medicine, Indianapolis, IN Yunfang Yu, Guangdong Provincial Key Laboratory of Malignant Tumor
Epigenetics and Gene Regulation, Department of Oncology, Sun Yat-sen
Background: Immunosuppressive factors such as regulatory T cells (Tregs)
Memorial Hospital, Sun Yat-sen University, Guangzhou, China
and myeloid-derived suppressive cells (MDSCs) limit the efficacy of im-
munotherapies. Histone deacetylase (HDAC) inhibitors have been shown to Background: Atezolizumab, a programmed death ligand 1 (PD-L1) inhibitor,
have immunomodulatory effects. We have previously reported that HDAC prolonged overall survival (OS) compared with docetaxel among patients with
inhibitors have synergistic antitumor effects in combination with PD-1 in- previously treated advanced non–small-cell lung cancer (NSCLC) in two in-
hibition in tumor models by inhibiting the function of Tregs and MDSCs. dependent multicentre, randomized trials (POPLAR and OAK). We conducted a
Thus, we conducted a Phase Ib clinical study with the HDAC inhibitor combined analysis of the two trials to evaluate its efficacy and genomic bio-
vorinostat and the PD-1 inhibitor pembrolizumab in patients (pts) with markers, and to further developed a novel predictive clinicopathologic-genomic
metastatic urothelial, renal and prostate carcinoma. Methods: The primary nomogram of immunotherapy in NSCLC. Methods: Patients (N = 1,137) with
objective was to evaluate the safety and tolerability of this combination stage IIIB/IV NSCLC and disease progression after previously platinum-based
strategy. The phase I portion consisted of two dose levels of vorinostat (100 chemotherapy were randomly assigned (1:1) to receive atezolizumab (1200
and 200 mg, PO daily 2 weeks ON and one week OFF) and a fixed, standard mg/kg every 3 weeks) or docetaxel (75 mg/m2 every 3 weeks). The primary endpoint
dose of pembrolizumab (200 mg IV every 21 days). Patients were assigned to was OS. We applied a two-stage meta-analysis of pooled individual patient data
three cohorts: Cohort A (previously treated, anti-PD1/PD-L1 naı̈ve urothelial in the intention-to-treat population. In OAK trial, patients treated with atezoli-
and renal cancer pts = 15), Cohort B (previously treated, anti-PD1/PD-L1 zumab were randomly assigned (1:1) to the training group or the validation group
resistant urothelial and renal cancer pts = 14), and Cohort C (prostate cancer to develop a predictive clinicopathologic-genomic nomogram of immuno-
pts = 14). Results: Dose levels 1 (4 enrolled, 3 evaluable) and 2 (4 enrolled, therapy. POPLAR and OAK were registered with ClinicalTrials.gov, numbers
3 evaluable) were completed without DLTs and 200 mg was the Phase II NCT02008227 and NCT01903993. Results: In the pooled analysis, the median
recommended dose for vorinostat. The most common resolved grade 3/4 overall survival was 13.49 months (95% confidence interval [CI], 11.95 to
15.22) with atezolizumab versus 9.66 months (95% CI, 8.73 to 10.70) with
toxicities were acute kidney injury (n = 1), anemia (n = 1), diarrhea (n = 1),
docetaxel. The risk of death was 28% lower with atezolizumab than with docetaxel
and hypothyroidism (n = 1) in the dose expansion cohorts. We have enrolled
(hazard ratio [HR], 0.72; 95% CI, 0.62 to 0.83; P , 0.001). The race, sex, tumor
43 pts (37 evaluable) in the dose expansion cohorts. For Cohort A, B, and C
histology, eastern cooperative oncology group performance status, PD-L1 ex-
the median PFS were 2.8 months, 5.2 months, and 3.5 months. Two PR
pression, and especially pretreatment mutation (TP53, DNMT3A and KEAP1)
were observed including the dose escalation phase. Two PCA pts have were significantly associated with OS, and were used for the development of the
achieved undetectable PSA. We have performed several correlative studies predictive nomogram. The clinical use of the nomogram showed a closer asso-
including flow cytometry and gene expression analysis on peripheral blood ciation with 3-year OS than the blood-based tumor mutational burden (bTMB) or
mononuclear cells, PDL-1 staining and PSMA PET scans in a subset of pts. PD-L1 expression alone (nomogram, AUC = 0.818; bTMB, AUC = 0.701; PD-L1,
Conclusions: The results from this phase Ib suggest that the combination of AUC = 0.526) among NSCLC patients who had received atezolizumab. The
vorinostat and pembrolizumab is relatively well tolerated and may be active superior predictability of the nomogram was further confirmed in the validation
in a subset of immune checkpoint resistant UC/RCC pts and immune and entire OAK cohorts. Conclusions: Among patients with advanced, previously
checkpoint naı̈ve PCA pts. Clinical trial information: NCT02619253. treated NSCLC, OS was significantly better with atezolizumab than with doce-
taxel. Furthermore, we constructed a novel and powerful clinicopathologic-
genomic nomogram for personalized immunotherapy options.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
124s Developmental Immunotherapy and Tumor Immunobiology

2574 Poster Session (Board #218), Sat, 8:00 AM-11:00 AM 2575 Poster Session (Board #219), Sat, 8:00 AM-11:00 AM
Neoadjuvant presurgical PD-1 inhibition in oral cavity squamous cell Feasibility and toxicity of neoadjuvant nivolumab with or without ipilimumab
carcinoma. First Author: Joshua Dean Horton, Medical University of South prior to extensive (salvage) surgery in patients with advanced head and neck
Carolina, Charleston, SC cancer (the IMCISION trial, NCT03003637). First Author: Charlotte L. Zuur,
Netherlands Cancer Institute (NKI-AVL), Amsterdam, Netherlands
Background: Oral cavity squamous cell carcinoma (OCSCC) is a highly
prevalent surgically-treated subset of head and neck cancer with frequent Background: surgery w/wo adjuvant radiotherapy (RT) for (recurrent) ad-
recurrence and poor survival. Immunotherapy has demonstrated efficacy in vanced head and neck squamous cell carcinoma (HNSCC) results in 30-
recurrent/metastatic head and neck cancer, but has not been validated in the 50% 5-year OS, indicating the need for novel treatment options. In recurrent
neoadjuvant presurgical setting. Methods: A Simon two stage design was metastatic HNSCC nivolumab nearly tripled the 2-year OS. Aiming at im-
used in this single-arm, Phase II clinical trial with a preplanned analysis after proving clinical outcome in advanced HNSCC in a curative setting, we tested
completion of stage one. The first stage included 9 patients with stage II-IVA the feasibility of nivolumab 6 ipilimumab neoadjuvant to (salvage) surgery
OCSCC who received 3-4 biweekly doses of 3mg/kg Nivolumab (anti- w/wo RT. Methods: investigator-initiated phase-IB/II trial to assess feasi-
programmed death 1 [PD-1]) followed by definitive surgical resection for bility of neoadjuvant nivolumab monotherapy (240 mg in week 1&3: arm-A)
cure. The primary endpoint was overall response rate to treatment. Sec- or in combination with ipilimumab (1 mg/kg in week 1: arm-B) before surgery
ondary endpoints were safety and feasibility. Results: Presurgical Nivolumab (# week 5) w/wo RT for advanced HNSCC. Results: 12 patients were in-
therapy resulted in an overall response rate of 44% (95% CI: 14-79%) with cluded (3+3 design, both arms) in phase-IB of this study; 7/12 (58%)
four patients having .30% reduction in tumor size consistent with partial patients had pre-existent moderate-to-severe comorbidities (ACE-27). All
response. An additional patient had stable disease while the remaining four patients were HPV negative. All patients received surgery as planned (25-
patients progressed through treatment. Neoadjuvant Nivolumab was not 33 days after start of immunotherapy) with no unexpected wound healing
associated with delays in definitive surgical treatment. There were no grade problems. In both groups, 4 patients (67%) experienced immune-related
3-4 adverse events and no treatment interruptions. At median follow up of toxicity: grade 1-2 (n = 4) and grade 3-4 (n = 1; colitis) in arm-A; grade 1-2
10 months (2-16), there were 4 recurrences in 3 patients and one death. (n = 5) and grade 3-4 (n = 2; colitis and elevated liver enzymes) in arm B.
Objective response by RECIST 1.1 criteria on interval imaging predicated Immune-related toxicity was managed with prednisone (n = 2) and infliximab
eventual pathologic response in 100% of patients. Conclusions: Neoadjuvant (n = 1). There was 1/6 (12.5%) pathological response in arm-A (1 near
presurgical PD-1 blockade is associated with encouraging response rate and complete response, nCR) and 3/6 (50%) in arm-B (1x partial response and
demonstrates feasibility and safety for OCSCC. Clinical trial information: 2x nCR). No patients with nCR had a recurrence at follow-up (median
NCT03021993. 10 months). Preliminary data (mutational load will be added) show increased
H7-B3 gene expression in non-responders before treatment, and increased
endothelial cell and NK cell gene expression in responders post-treatment.
Overall, in these 12 patients, neoadjuvant ipilimumab + nivolumab resulted
in a significant increase in immune-related gene expression when compared to
nivolumab only, irrespective of treatment response. Conclusions: neoadjuvant
ipilimumab + nivolumab can safely be administered prior to major surgery for
advanced HNSCC. Efficacy is promising and will be further evaluated in the
phase-II trial continuation. Clinical trial information: NCT03003637.

2576 Poster Session (Board #220), Sat, 8:00 AM-11:00 AM 2577 Poster Session (Board #221), Sat, 8:00 AM-11:00 AM
Effect of exonic microsatellite instability of B2M on the predictability of MSI/ Immune-mediated colitis after resumption of immune checkpoint inhibitor
dMMR for immunotherapy. First Author: Jia Wei, The Comprehensive Cancer therapy. First Author: Hamzah Abu-Sbeih, The University of Texas MD
Centre of Drum Tower Hospital, Medical School of Nanjing University & Anderson Cancer Center, Houston, TX
Clinical Cancer Institute of Nanjing University, Nanjing, China
Background: Immune checkpoint inhibitor (ICI) therapy is often suspended
Background: Microsatellite instability/mismatch repair deficiency (MSI/ because of immune-mediated diarrhea and colitis (IMDC). We examined the
dMMR) has been approved as a biomarker for immune checkpoint in- recurrence rate and risk factors for IMDC after ICI resumption. Methods: This
hibitor therapy (ICI) . However, the deficiency of mismatch repair system also retrospective multicenter study examined patients who resumed ICI therapy
caused wide-spreading insertion/deletion (indel) mutations in the exonic after improvement of IMDC between 1/2010 and 11/2018. Univariate and
microsatellite sites of critical immunotherapy related genes such as b-2- multivariate logistic regression analyses assessed the association of clinical
macroglobulin (B2M) whose product is critical to antigen presentation. covariates and IMDC recurrence. Results: Of the 167 patients in our analysis,
Cases of aquired resistance to ICI treatment caused by the inactivation of 32 resumed an anti-CTLA-4 agent and 135 an anti-PD-1/L1 agent. The
B2M have been reported. Therefore, we investigated the mutational profile of median duration from IMDC to restart of ICI treatment was 49 days (IQR, 23-
B2M across the MSI/dMMR high prevalent cancers such as colorectal cancer 136). IMDC recurred in 57 (34%) patients overall (44% of those resuming an
(CRC), gastric cancer (GC) and endometrial cancer (EC). Methods: FFPE anti–CTLA-4 and 32% resuming an anti–PD-1/L1 agent); 47 of these patients
tumor samples from 37 CRC, 46 GC and 25 EC patients with matched (82%) required immunosuppressive therapy for recurrent IMDC (Table). The
normal tissues were collected for next-generation sequencing (NGS)-based median duration from ICI resumption to IMDC recurrence was 53 days (IQR
450 genes panel assay. Genomic alterations including single base sub- 22-138). On multivariate logistic regression, patients who received anti-PD-1/
stitution, short and long insertions/deletions, copy number variations, and L1 therapy at initial IMDC had a higher risk of IMDC recurrence (odds ratio
gene rearrangement and tumor mutational burden were assessed. Immu- [OR], 3.45, 95%CI, 1.59-7.69; P,0.01). Risk of IMDC recurrence was
nohistochemistry (IHC) with antibody against B2M were performed on higher for patients who required immunosuppression for initial IMDC (OR,
available samples to estimate the expression and localization of these 3.22; 95%CI, 1.08-9.62; P=0.02) or had longer duration of IMDC symptoms
proteins. Results: Exonic microsatellite sites of B2M gene have been found in the initial episode (OR, 1.01; 95%CI, 1.00-1.03; P=0.03). Risk of IMDC
instable in 51% (19/37) of CRC, 22% (10/46) of GC and 8% (2/25) of EC. recurrence was lower for those who resumed anti–PD-1/L1 therapy than for
MSI caused small indels at B2M coding region leads to reading frame shift those who resumed anti–CTLA-4 therapy (OR, 0.30; 95%CI, 0.11-0.81;
and the production of nonfunctional truncated proteins. Biallelic frameshift P=0.02). Conclusions: One-third of patients who resumed ICI treatment after
mutations, causing non-functional proteins, were found in 47% (9/19) of IMDC experienced recurrent IMDC. IMDC recurrence was less frequent after
CRC and 30% (3/10) of GC patients carrying Indels in B2M gene. IHC assays resumption of anti–PD-1/L1 than after anti–CTLA-4. Characteristics of re-
showed the impaired expression of B2M proteins due to frameshift muta- current IMDC based on resumed ICI therapy.
tions. In addition, NGS test confirmed concurrent mutation in B2M for a Anti–CTLA-4 Anti–PD-1/L1
MSI-H CRC patient with primary resistance to forth-line anti-PD1 treatment. Characteristic N = 32 (%) N = 135 (%) P

Conclusions: The extensive mutations in the coding region of genes caused Recurrence of symptoms 14 (44) 43 (32) 0.30
Time from ICI resumption to IMDC recurrence, days (IQR) 26 (2-43) 79 (27-141) 0.02
by the dMMR render the cancer cell sensitive to ICI. However, the con- Treatment of recurrence 0.31
comitant variants in exonic microsatellite sites of B2M gene may compro- Symptomatic only
Steroid
3 (9)
8 (25)
8 (6)
31 (23)
mise the predictability of MSI/dMMR as an ICI biomarker. Biallelic mutation Infliximab/vedolizumab add-on 3 (9) 4 (3)
Grade of recurrent diarrheaa 0.50
of B2M may cause primary resistance to ICI while the mono-allelic mutation 1 2 (6) 9 (7)
2 11 (34) 29 (22)
of B2M may gain acquired resistance. Therefore, B2M status could be 3-4 1 (3) 5 (4)
considered when using MSI/dMMR as a biomarker for ICI. Grade of recurrent colitisa
1 7 (22) 23 (17)
0.39

2-3 6 (19) 20 (15)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Immunotherapy and Tumor Immunobiology 125s

2578 Poster Session (Board #222), Sat, 8:00 AM-11:00 AM 2579 Poster Session (Board #223), Sat, 8:00 AM-11:00 AM
Tumor mutational burden (TMB) and response rates to immune checkpoint MEDIPLEX: A phase 1 study of durvalumab (D) combined with pexidartinib
inhibitors (ICIs) targeting PD-1, CTLA-4, and combination. First Author: (P) in patients (pts) with advanced pancreatic ductal adenocarcinoma
Arsen Osipov, Johns Hopkins University School of Medicine, Department of (PDAC) and colorectal cancer (CRC). First Author: Philippe Alexandre
Oncology, Balimore, MD Cassier, Centre Léon-Bérard, Lyon, France
Background: ICIs targeting PD-1/L1 and/or CTLA-4 have activity against Background: Targeting tumor associated macrophages is an emerging
many different cancers. We and others have previously shown that a higher strategy to increase the responsiveness of PDAC and CRC to anti-PD(L)1.
TMB, a surrogate for an increased number of expressed tumor neoantigens, Pexidartinib (P) is an orally active, small-molecule kinase inhibitor that
is an important biomarker for response to anti-PD-1/L1 monotherapy. targets the colony-stimulating factor-1 receptor (CSF1R) on macrophages.
Whether the relationship between the TMB and response to ICIs extends Methods: Adult pts with advanced/ metastatic PDAC or CRC were treated
beyond anti-PD-1/L1 is unknown. Methods: We identified 30 major solid with a fixed dose of D (1500mg q4w, IV) and ascending doses of P (400,
tumor types for which TMB has been described using a genomic profiling 600, 800 and 1000mg/d, orally). Dose escalation was conducted according
assay performed by Foundation Medicine. We conducted searches of to a Likelihood Continual Reassessment Method with a 28-day window to
MEDLINE (from Jan 1, 2010 to Jan 20, 2019), as well as abstracts pre- evaluate dose-limiting toxicity (DLT), a stopping rule advised dose escalation
sented at ASCO, ESMO, AACR Annual Meetings 2010-2018 to identify the termination in case of a high probability ( . 90%) for the next 6 pts to be
objective response rate (ORR) for anti-PD-1/L1, anti-CTLA-4 and combi- assigned to the same dose. Following the determination of RP2D, 14 pts with
nation anti-PD-1/L1 plus anti-CTLA-4, in each of these cancer types. We PDAC and 14 pts with CRC who consented to serial tumor biopsies were
pooled the response data from the largest published studies that evaluated enrolled in expansion cohorts to assess preliminary anti-tumor activity and
the ORR. We excluded studies that; enrolled , 10 evaluable patients, in- biomarkers. Results: 19 pts (12M, 7F, median age, 56 y [range, 43-76y])
vestigated ICI therapies in combination with other agents, and studies that were enrolled in 4 dose escalation cohorts (P 400, 600 and 800mg/d: 3 pts
selected patients based on immune-related biomarkers. Across tumor types, each and P 1000mg/d: 10 pts). Pharmacokinetic analysis showed dose-
median TMB was compared to ORR utilizing the coefficient of determination dependent increase in the exposure of P from 400 to 1000 mg. Two DLTs
(r2) derived from simple linear regressions. Results: TMB is strongly asso- (AST/ALT elevations including one with bilirubin increase) were seen at dose
ciated with response to anti-PD-1/L1 monotherapy (n = 8798, r2= 0.4704, level P 1000mg/d. The most frequent ( . 2pts) related (to either D, P or both)
p , 0.001), and combination anti-PD-1/L1 plus anti-CTLA-4 (n = 2280,r2= AEs were: fatigue, maculopapular rash/pruritus/dry skin, hair color changes,
0.4082, p = 0.004). Available ORR data were more limited with CTLA-4 anorexia, edema (periorbital, limbs or face), AST/ALT increases, bilirubin
monotherapy and the relationship between ORR and TMB did not meet increases, nausea, vomiting and diarrhea. The most frequent ($2pts)
statistical significance (n = 1377, r2= 0.2606, p = 0.1086). The additional Grade $ 3 AEs related to P were: AST/ALT increase, ALP increase, neutrophil
ORR benefit of adding a CTLA-4 inhibitor to anti-PD-1/PDL1 therapy in- or white blood cell count decrease, and fatigue. Clinical benefit rate at
creased with increasing TMB. In tumor types with a lower TMB ( , 10 2 months was 21% (4 SD /19pts), 2 pts with MSI-H CRC had SD for more than
mutations/MB), combined ICI therapy led to an average improvement of 6 months including 1 pt still receiving single agent D after . 18 months. The
5.5% in ORR over PD-1/L1 monotherapy, versus 21.8 % ORR improvement RP2D for the combination was P 800mg/d + D 1500mg q4w. Enrolment in the
in high TMB tumors ($10 mutations/MB). Conclusions: A strong relation- expansion cohorts was completed in January 2019. Conclusions: Toxicity was
ship exists between TMB and clinical activity of both PD-1/L1 monotherapy consistent with the expected profiles of the individual drugs and no un-
and combination ICIs with PD-1/L1 plus CTLA-4. The clinical benefit of expected events were seen with the combination. Updated data will be
adding anti-CTLA-4 to anti-PD-1/L1 is greatest in high TMB tumors and presented at the meeting. Clinical trial information: NCT02777710.
limited in low TMB tumors.

2580 Poster Session (Board #224), Sat, 8:00 AM-11:00 AM 2581 Poster Session (Board #225), Sat, 8:00 AM-11:00 AM
Patient survival with immune checkpoint inhibitors and targeted agents in Using machine learning algorithms to predict response and toxicity to
phase 1 trials: A propensity score weighted analysis. First Author: Itziar immune checkpoint inhibitors (ICIs) in melanoma patients. First Author:
Gardeazabal, Vall d’Hebron University Hospital and Institute of Oncology Paul Johannet, NYU School of Medicine, New York, NY
(VHIO), Barcelona, Spain
Background: There is growing interest in optimizing patient selection for
Background: There have been important changes in early drug development treatment with immune checkpoint inhibitors (ICIs). We postulate that
units with an unprecedented increase of immune-oncology (IO) trials. phenotypic features present in metastatic melanoma tissue reflect the bi-
Currently at the Vall d’Hebron Institute Oncology (VHIO) close to 50% of our ology of tumor cells, immune cells, and stromal tissue, and hence can
Phase 1 trials (Ph1t) portfolio includes IO drugs, while from 2011 to 2015 provide predictive information about tumor behavior. Here, we test the
more than 80% of our trials assessed targeted agents (TA). We wanted to hypothesis that machine learning algorithms can be trained to predict the
investigate whether this swift had a positive impact on patient (pts) outcome. likelihood of response and/or toxicity to ICIs. Methods: We examined 124
Methods: We performed a retrospective analysis of the pts treated with IO stage III/IV melanoma patients who received anti-CTLA-4 (n = 81), anti-PD-1
and TA at VHIO Ph1t Unit from Jun’11 to May’18. Only patients treated with (n = 25), or combination (n = 18) therapy as first line. The tissue analyzed
IO in $ 2nd line were included (and without an approved IO therapy as per was resected before treatment with ICIs. In total, 340 H&E slides were
standard-of-care) and those with TA classified as tiers II-III-IV by the ESMO digitized and annotated for three regions of interest: tumor, lymphocytes,
scale for clinical actionability of molecular targets ESCAT (which also and stroma. The slides were then partitioned into training (n = 285), vali-
represents unapproved indications). The aim of this study was to compare dation (n = 26), and test (n = 29) sets. Slides were tiled (299x299 pixels) at
overall survival (OS) for the two cohorts. Given the non-randomized nature of 20X magnification. We trained a deep convolutional neural network (DCNN)
the study a propensity score weighting (PSW) was used to control for se- to automatically segment the images into each of the three regions and then
lection bias in treatment effect estimation. Results: Out of 545 eligible pts, deconstruct images into their component features to detect non-obvious
281 (51.5%) received TA and 264 (48.5%) IO, with unadjusted median OS patterns with objectivity and reproducibility. We then trained the DCNN for
(mOS) of 7.7 months (m) and 9.2m, respectively. In univariate analysis, two classifications: 1) complete/partial response versus progression of
OS was associated with tumor type, number of previous treatment lines, regimen disease (POD), and 2) severe versus no immune-related adverse events
(monotherapy vs combination), and clinical-laboratory prognostic factors (irAEs). Predictive accuracy was estimated by area under the curve (AUC) of
(Vioscore: albumin , 3.5 g/dl; LDH . upper limit of normal; neutrophil/ receiver operating characteristics (ROC). Results: The DCNN identified
[leukocytes minus neutrophils] ratio (dNLR) . 3; more than 2 sites of me- tumor within LN with AUC 0.987 and within ST with AUC 0.943. Prediction
tastasis; and presence of liver metastasis) (p , 0.05). After adjusting for these of POD based on ST-only always performed better than prediction based on
factors in a PSW model, the IO group showed statistically significant longer OS LN-only (AUC 0.84 compared to 0.61, respectively). The DCNN had an av-
with HR = 0.75 (CI95% 0.65 – 0.86, p , 0.0001). The In a stratified analysis erage AUC 0.69 when analyzing only tumor regions from both LN and ST data
by tumor type we found no significant heterogeneity in the relative benefit of IO sets and AUC 0.68 when analyzing tumor and lymphocyte regions. Severe
over TA. Conclusions: In real world data from our Ph1t population, treatment irAEs were predicted with limited accuracy (AUC 0.53). Conclusions: Our
with IO was associated with longer OS than treatment with TA, even after results support the potential application of machine learning on pre-treatment
adjusting for known prognostic factors and treatment selection biases. These histologic slides to predict response to ICIs. It also revealed their limited value
results suggest that the likelihood of patient benefit with IO therapies in Ph1t in predicting toxicity. We are currently investigating whether the predictive
is increasing. capability of the algorithm can be further improved by incorporating additional
immunologic biomarkers.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
126s Developmental Immunotherapy and Tumor Immunobiology

2582 Poster Session (Board #226), Sat, 8:00 AM-11:00 AM 2584 Poster Session (Board #228), Sat, 8:00 AM-11:00 AM
High prevalence of IBD-associated genetic variants in patients (pts) with Overcoming genetically based resistance mechanisms to PD-1 blockade.
immune checkpoint inhibitor (ICI) enteritis/colitis. First Author: Shilpa First Author: Davis Yuri Torrejon, University of California Los Angeles
Grover, Brigham and Women Hospital, Boston, MA Jonsson Comprehensive Cancer Center, Los Angeles, CA
Background: Colitis is a frequent toxicity of ICI therapy but there is paucity of Background: Mechanism-based strategies to overcome resistance to anti-PD1
data on risk factors. Specific serological markers and genetic polymorphisms therapy are urgently needed. Using CRISPR/Cas9 genome editing tools, we
have been associated with inflammatory bowel disease (IBD) (ulcerative developed acquired resistant models through JAK1/2 and B2M loss of function
colitis, Crohn’s disease). However, the prevalence of these markers in pts (LoF) mutations in human melanoma cell lines and in the murine MC38 colon
with ICI colitis is unknown. We performed a pilot study to determine the carcinoma, known for high mutational load and good response to anti-PD-1.
prevalence of IBD-associated genetic and serologic biomarkers in pts with We hypothesized that the downstream activation of the IFN-receptor pathway
ICI colitis. Methods: Cancer pts with histologically confirmed ICI enteritis/ or the activation of natural killer (NK) cells would overcome this resistance.
colitis and no history of IBD underwent commercial IBD panel testing. The Methods: We studied signaling changes in four human cell lines (parental and
panel included 4 genetic markers (ATG16L1, NXK2–3, ECM1, STAT3), 8 LoFs) exposed to IFN-gamma using RNAseq. In addition, we analyzed the in-
serological markers (anti-A4-Fla2, anti-A4-FlaX, anti-CBir1, anti-OmpC, vivo antitumor activity in MC38 variants with anti-PD1 and characterized the
ASCAIgA, ASCA-IgG, pANCA, ANCA), and 5 inflammatory markers (vas- tumor microenvironment using mass cytometry (CyTOF). Finally, we tested
cular endothelial growth factor [VEGF], intracellular adhesion molecule 1 strategies to overcome resistance mechanisms with SD-101 (TLR-9 agonist) and
[ICAM-1], vascular cell adhesion molecule 1 [VCAM-1], C-reactive protein, bempegaldesleukin (NKTR-214, CD-122 biased agonist) with the extent of CD8
serum amyloid A [SAA]). Clinical testing on serum samples was performed by and NK1.1 depletion. Results: RNAseq differential gene expression analysis
Prometheus Laboratories (San Diego, CA). Results: Of 15 cancer pts with showed that the IFN-gamma induced increased expression of antigen presenting
biopsy confirmed ICI colitis, 10 (67%) were homozygous for 1 or more of 4 machinery, IFN-gamma signaling and chemokines (CXCL9/10) was lost in JAK1/
genetic markers. The remaining 5 pts were all heterozygous for two or more of 2-LoF human melanoma cell lines. The significant antitumor activity of anti-PD-1
the genetic markers. One or more serologic markers associated with IBD were against MC38 parental cell line was lost in JAK1/2 and B2M LoF sublines, and
CyTOF analysis revealed that anti-PD-1 therapy was unable to increase tumor CD8+
elevated in 7/15 (47%) pts. Serum reactivity was noted for ASCA-IgA (1/15,
T-effectors in these LoF tumors. The intratumoral administration of SD-101
7%), ASCA-IgG (1/15, 7%), anti-OmpC (3/15, 20%), anti-CBR IgG (2/15,
(50 mg/injection q4dx3wks) was able to overcome local resistance even in non-
13%), anti-A4-FlaX (1/15, 7%), and ANCA (2/15, 13%). One or more in-
injected sites in JAK1/2 and IFNAR-type-I LoF tumors, and systemic adminis-
flammatory markers were elevated in 13/15 (88%) pts. Elevations in VEGF,
tration of bempegaldesleukin (0.8 mg/kg, q9dx2, i.v.) was able to overcome
VCAM-1, ICAM-1, and SAA were noted in 2 (13%), 8 (53%), 8 (53%), and resistance in B2M LoF with significantly increased survival (Table). Depletion
11 (73%) pts, respectively. Only 6 (40%) pts had elevations in CRP levels studies showed complete abrogation of anti-tumor response with anti-NK1.1 in
despite the presence of active inflammation on biopsy. The IBD panel was JAK1 LoF and B2M LoF, and partial abrogation with anti-NK1.1 or anti-CD8 in
reported as being consistent with Crohn’s disease in 2 pts, ulcerative co- JAK2 LoF tumors. Conclusions: Even in the extreme setting of genetic resistance
litis in 1 pt and inconclusive for type but consistent with IBD in 1 pt. to PD-1 blockade by JAK1/2 LoF, resistance can be overcome by SD-101, a TLR9
Conclusions: In this pilot study, all patients with ICI colitis, were either agonist, while resistance of B2M LoF can be overcome by bempegaldesleukin
homozygous or heterozygous for two or more high risk IBD alleles. If vali- (NKTR-214), a CD-122 biased agonist. Our findings support the testing of these
dated, such testing may prospectively identify pts at risk for developing ICI rational mechanistic strategies in patients with a-PD1 resistance.
colitis.
JAK1 JAK2 B2M

Iso aPD1 SD101 aPD1+SD101 Iso aPD1 SD101 aPD1+SD101 Iso aPD1 NKTR-214 a-PD1+NKTR-214

n 5 5 5 5 5 5 5 5 5 5 5 5
Median survival (days) 27 27 64 94 21 21 32 36 19 21 90 100
p (Log-rank test) ns *** *** ns *** *** ns *** ***
***p , 0.0005

2585 Poster Session (Board #229), Sat, 8:00 AM-11:00 AM 2586 Poster Session (Board #230), Sat, 8:00 AM-11:00 AM
Measuring the long-term “tail of curve” survival benefits in oncology trials: A Genetic determinants of adverse events in cancer patients receiving immune
comparison of the ASCO Value Framework and the ESMO Magnitude of checkpoint inhibitors. First Author: Noha Abdel-Wahab, Section of Rheu-
Clinical Benefit Scale. First Author: Louis Everest, Odette Cancer Centre, matology and Clinical Immunology, Department of General Internal Medi-
Sunnybrook Health Sciences Centre, University of Toronto, Toronto, ON, cine, The University of Texas MD Anderson Cancer Center, Houston, TX
Canada
Background: Immune checkpoint inhibitors (ICI) can cause profound
Background: Recently, anti-cancer agents have generated excitement due to immune-related adverse events (irAEs). The host genetic background is
their capacity to preserve long-term survival in some patients, represented likely to play a role in irAEs susceptibility as the phenotype of toxicity varies
by a “tail of the survival curve”. However, as traditional measures of clinical among patients (pts) and many pts do not develop toxicity despite continued
benefit may not accurately capture long-term survival, amendments to inhibition. Methods: Genotyping was performed for 89 melanoma pts who
various valuation frameworks have been proposed to capture this benefit. The received ICI using the Infinium Multi-Ethnic Global-8 v1.0 Bead Chip. The
purpose of this study was to determine how frequently immune checkpoint genotype data was extracted using PLINK (v1.90b3.34) and processed for
inhibitor vs. non-immune checkpoint inhibitor anti-cancer agents, displayed quality control including on call rate (.99%), genotyping rate (.95%),
trends of long-term survival, as defined by the American Society of Clinical minor allele frequency (no less than 5%), and Hardy-Weinberg Equilibrium
Oncology Value Framework (ASCO-VF) and European Society of Medical (p,1x10-6). The pairwise genetic distance was calculated using identity-by-
Oncology Magnitude of Clinical Benefit Scale (ESMO-MCBS), as well as to state (IBS matrix) implemented in the genome option of PLINK, and the
analyze the degree of agreement between ASCO and ESMO frameworks. population-structure-based clustering was carried out using IBS matrix,
Methods: Anti-cancer agents from phase II or III randomized controlled trials pairwise population concordance test (p,1x10-3) and phenotype distri-
(RCTs) cited for clinical efficacy evidence in drug approval by the Food and bution for all pts, resulting in 7 structure groups. In the analytical stage,
Drug Administration (FDA) between January 2011 and March 2018 were 602,463 variants in autosomal chromosomes were included for the asso-
identified. Data required for ASCO-VF and ESMO-MCBS were extracted. ciation test. The test was performed using additive genetic model with exact
Difference in how often long-term survival bonuses were awarded were cal- logistic regression, adjusted for age, sex, and population cluster. Results: 44
culated in all RCTs, as well as immune checkpoint inhibitor and non-immune pts had arthritis, colitis, hypohysitis, thyroiditis, or multiple irAEs (cases),
checkpoint inhibitor RCTs individually. Cohen’s Kappa statistic was calculated and 45 did not have irAEs after a minimum of one year of treatment
to evaluate agreement between ASCO-VF and ESMO-MCBS. Results: 100 (controls); median age was 64 (23-92) years; 71% were male. A total of 30
RCTs were analyzed. RCTs were awarded ASCO-VF version 2 (v2) “tail of the variants/single nucleotide polymorphisms (SNPs) had p-value smaller than
curve” bonuses more often than ESMO-MCBS version 1.1 (v1.1) “immuno- 10^-5 level were identified. The top variants/SNPs are listed in table.
therapy-triggered” long-term plateau adjustments (45% vs. 2.6%). Comparing Conclusions: Genetic variants associated with irAEs were identified. Addi-
to non-immune checkpoint inhibitor RCTs, immune checkpoint inhibitor RCTs tional larger studies are needed to validate these findings, and to establish
were not more likely to receive ASCO-VF v2 bonuses/ESMO-MCBS v1.1 ad- their potential functional relevance.
justments (p = 0.32/ p = 0.40). Long-term survival agreement between the two Top variants/SNPs from the association testing with irAEs.
frameworks was poor (kappa: 0.01; p = 0.50). Conclusions: The ASCO-VF v2 CHR SNP BP P Gene
and ESMO-MCBS v1.1 may require additional refinement in order to accu- 3 rs11711517 65,112,627 2.39E-06
rately capture the benefit of long-term survival or immune checkpoint inhibitor 5 rs11743438 170,218,232 5.56E-06 GABRP
5 rs11743735 170,213,256 8.34E-06 GABRP
and non-immune checkpoint inhibitor agents may not preserve substantially 20 JHU_20.57183980 58,608,925 8.85E-06 DSC2
different long-term survival populations. 15 JHU_15.93602126 93,058,898 8.89E-06
15 JHU_15.93604000 93,060,772 8.89E-06
18 rs470753 65,315,670 1.65E-05
15 rs4778080 93,060,296 1.92E-05
12 rs2117997 29,143,428 2.44E-05 ANKRD42

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Immunotherapy and Tumor Immunobiology 127s

2587 Poster Session (Board #231), Sat, 8:00 AM-11:00 AM 2588 Poster Session (Board #232), Sat, 8:00 AM-11:00 AM
Real-world outcomes of underrepresented patient populations treated with Intra and perinodular CT delta radiomic features associated with early
immune checkpoint inhibitors (ICIs): African American descent, poor ECOG response to predict overall survival (OS) in immunotherapy-treated non-small
performance status, and chronic viral infections. First Author: Neil J. Shah, cell lung cancer (NSCLC): A multisite multi-agent study. First Author: Prateek
Lombardi Comprehensive Cancer Center, Medstar Georgetown University Prasanna, Case Western Reserve University, Cleveland, OH
Hospital, Washington, DC
Background: None of the current biomarkers for predicting response to
Background: ICIs have now become standard of care treatment for multiple checkpoint inhibitors (ICIT) for advanced NSCLC are associated with long-
malignancies. However, patients (pts) who are African American decent term benefits, such as improved OS. In this multi-agent (nivolumab,
(AA), have a poor ECOG performance status (PS) or chronic viral infections pembrolizumab, or atezolizumab) multi-site study (Cleveland Clinic, Univ. of
[human immunodeficiency virus (HIV), hepatitis B (HBV) and hepatitis C Pennsylvania), we demonstrate that changes in computer-extracted textural
(HCV)] were underrepresented in early clinical trials with ICIs and outcome patterns, from within and 30mm outside the nodules, between baseline and
data in these pt populations is not well reported. Methods: We performed a post-treatment CT following ICIT correlate with RECIST-derived responses,
retrospective analysis of pts treated with ICIs (anti-PD(L)-1, anti-CTLA-4, or and are prognostic of OS. Methods: CT scans from 139 NSCLC patients both
combination ICIs) across five MedStar Health hospitals from January 2011 to pre-, and post 2-3 cycles of ICIT were acquired from 2 sites. Patients with
April 2018. Investigator-assessed best responses were noted. CTCAE v4.03 objective response/stable disease per RECIST v1.1 were defined as ‘responders’,
was used to capture immune-related adverse events (irAEs). Results: We and those with progressive disease were ‘non-responders’. The cohort was di-
identified 765 pts treated with 829 unique ICIs therapies across different vided into a discovery (D1 = 50) and two validation sets (D2 = 62, D3 = 27). 454
malignancies. A total of 203 AA pts, 178 pts with a pre-treatment ECOG intranodular texture (IT) features, and 7426 perinodular features (PT) were
PS $2, 21pts with HIV, and 50 pts with HBV/HCV were noted. Any grade and extracted from the temporalscans, Relative differences were computed to yield a
grade $ 3 irAEs in the HIV cohort were 24% and 10% with an ORR of 29%. set of ‘delta-radiomic’ descriptors. In D1, 8 features that evolved the most
Any grade and grade $ 3 irAEs in HBV/HCV were 50% and 26% with an ORR between baseline and post-treatment CT, and performed the best in identifying
of 21%. No viral reactivation or changes in pts anti-viral medications were responders, were determined. These were then used with a Linear Discriminant
noted during ICIs treatment. The ORR in AA pts was 35%. Any grade and Analysis classifier to identify the responders from the non-responders. We then
grade $ 3 irAEs in the AA cohort were 27% and 8%, respectively. The ORR in computed a radiomic risk score (RRS) system and tested its prognostic ability in
pts with ECOG PS $2 was 14%. Any grade and grade $ 3 irAEs in this cohort assessing differences in OS. Results: A combination of 5 IT, 3 PT delta radiomic
were 20% and 4%. Similar trends were seen in the subset of patients with features yielded an AUC of 0.88 6 0.08 in D1 and a corresponding AUC = 0.85
NSCLC treated with anti-PD(L)1 monotherapy (Table). Outcomes of NSCLC and 0.81 in D2 and D3, respectively. Multivariate survival metrics are shown in
pts treated with anti-PD(L)-1 monotherapy. Conclusions: ICI therapy was not Table. Conclusions: Delta-radiomic features, both from inside and outside the
associated with any new safety signal in the above underrepresented pop- nodules, could be used to identify patients likely to derive clinical benefit from
ulations. Prospective studies are needed to validate this data. ICIT (eg: OS) beyond anatomic response.
Cohorts (N) ORR (N) Any grade irAEs (N) Grade ‡3 irAEs (N) HR, p-val
Entire cohort (232) 21% (44/214*) 31% (66) 10% (22) Variable D1 D2 D3
African American (102) 19% (18/94*) 22% (22) 5% (5)
Gender (M vs F) 0.95 (0.48 - 1.87), 0.84 (0.28 – 2.53), 0.89 (0.31 – 2.6),
Caucasian (112) 21% (22/104*) 37% (41) 13% (14)
.87 .76 .83
ECOG PS 0-1(163) 23% (34/148*) 34% (55) 10% (17)
Smoking (Y vs N) 1.14 (0.49 - 2.68), 2.32 (0.6 – 8.9), 2.29 (0.67 – 7.8),
ECOG PS ‡2 (75) 16% (10/64*) 21% (16) 7% (5)
.75 .22 .3
*Response evaluable patients. RRS 3.07 (1.74 - 5.42), 2.75 (1.48 – 3.2), 2.05 (1.28 – 3.25),
.0001 .004 .003

2589 Poster Session (Board #233), Sat, 8:00 AM-11:00 AM 2590 Poster Session (Board #234), Sat, 8:00 AM-11:00 AM
Tumor mutation burden analysis in a 5,660 cancer patient cohort reveals Association of the imbalance between early and late differentiated intra-tumor
cancer type-specific mechanisms for high mutation burden. First Author: CD4 T cells with mutational burden in non-small cell lung cancer. First Author:
Xiaodong Jiao, Department of Medical Oncology，Changzheng Hospital, Ehsan Ghorani, Cancer Immunology Unit, University College London Cancer
Second Military Medical University, Shanghai, China Institute, London, United Kingdom
Background: Tumor mutation burden (TMB), calculated by whole-exome Background: CD4 T helper cells are key orchestrators of immunity in states of
sequencing (WES) or large NGS panels, has an important association with persistent antigen exposure. In chronic viral infection, loss of immune
immunotherapy responses. Elucidating the underlying biological mecha- control is associated with CD4 differentiation skewing (CD4ds) resulting from
nisms of high TMB might help develop more precise and effective means for decline of early progenitors and gain in abundance of exhausted and ter-
TMB and immunotherapy response prediction. Meanwhile, the landscape minally differentiated subsets. Here, we set out to identify whether a similar
of TMB across different cancer types and its association with other mo- process occurs within the tumour microenvironment, contributing to immune
lecular features have not been well investigated in large cohorts in China. dysfunction. Methods: Multiregional samples of tumour and non-tumour
Methods: Cancer patients whose fresh tissue (n = 1556), formalin-fixed, lung tissue from patients with untreated, surgically resected non-
paraffin-embed (FFPE) specimen (n = 1794), and pleural fluid (n = 84) were small cell lung cancer (NSCLC) within the first 100 recruited to the
profiled using 295- or 520-gene NGS panel. The association of the TMB prospective lung TRACERx study were analysed by high dimensional flow
status with a series of molecular features and biological pathways was in- cytometry of tumour infiltrating lymphocytes (TILs) and paired bulk tumour
terrogated using bootstrapping. Results: TMB, measured by 295- or 520- exome and RNA sequencing. We additionally reanalysed publically available
cancer-related gene panels, were correlated with WES TMB based on in silico single T cell and bulk tumour RNA sequencing from patients with NSCLC.
simulation in the TCGA cohort. We compared the TMB landscape across 11 Results: Unsupervised clustering and dimension reduction revealed a het-
cancer type groups and found the highest average TMB in lung squamous erogenous landscape of CD4 TILs, with evidence of CD4ds in association with
cell carcinoma, whereas the lowest TMB was established in sarcoma. High tumour mutational burden. Loss of PD1-CCR7+ T central memory enriched
microsatellite instability, DNA damage response deficiency, and homolo- early differentiated cells was accompanied by gain in abundance of PD1+
gous recombination repair deficiency indicated significantly higher TMB. populations with exhausted (CD57-ICOShiCTLA4hi) and terminally differ-
The independent predictive power for TMB of twenty-six biological pathways entiated effector (CD57+Eomes+) phenotypes. Further characterisation of
was tested in 10 cancer groups. FoxO signaling pathway most commonly these populations by single cell RNA sequencing revealed differential ex-
correlated with low-TMB; significant association was identified in four pression of key genes involved in transcriptional regulation, co-inhibition and
cancer groups. In contrast, no pathway was significantly correlated with co-stimulatory pathways. A validated gene signature of CD4ds was associated
high-TMB in more than two cancer groups. Overall, we discovered that the with worse outcomes in TRACERx and TCGA cohorts. Conclusions: Our
underlying pathways which may be the main drivers of TMB status varied findings reveal remodelling of the CD4 differentiation landscape in asso-
greatly and sometimes had an opposite association with TMB across dif- ciation with tumour genomic characteristics, underscoring how mutational
ferent cancer types. Moreover, we developed a 14- and 22-gene signature for burden in the context of chronic stimulation may lead to loss of immune
TMB prediction for LUAD and LUSC, respectively, with only 10 genes shared fitness and elucidating key regulatory pathways in potentially clinically
by both signatures, indicating a histology-specific mechanism for driving relevant CD4 subsets.
high-TMB in lung cancer. Conclusions: The findings extended the knowl-
edge of the underlying biological mechanisms for high TMB and might be
helpful for developing more precise and accessible TMB assessment panels
and algorithms in more cancer types.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
128s Developmental Immunotherapy and Tumor Immunobiology

2591 Poster Session (Board #235), Sat, 8:00 AM-11:00 AM 2592 Poster Session (Board #236), Sat, 8:00 AM-11:00 AM
Evidence for selective silencing of MHC-binding neoepitopes to avoid Can serum IL-6 levels predict sarcopenia and poor outcome in relapsed/
immune surveillance. First Author: Rahul Parulkar, Nantomics LLC, refractory gynecologic cancer patients? First Author: Tomoyuki Yoshikawa,
Santa Cruz, CA Department of Clinical Oncology, National Defense Medical College Hospital,
Tokorozawa, Japan
Background: Overall response rates to immune checkpoint inhibition (ICI)
are , 50% even in TMB-high patients (e.g. Checkmate-227), suggesting Background: Cancer cachexia occurs in more than half of cancer patients
other mechanisms of immune escape exist beyond expressing checkpoints. and can be the primary cause of death for at least 20% of all patients. Cancer
At least 18% of somatic-specific exonic DNA variants are not expressed into cachexia also lowers quality of life in cancer survivors due to a severe loss of
mRNA (Rabizadeh, 2018), yet the selection criteria for which variants to skeletal muscle mass. Although a multitude of cytokines have been impli-
silence remains unclear. We sought to determine if immunogenicity of cated in facilitating a cachectic state, the correlation of serum IL-6 and
variants factors into their suppression. Methods: Somatic-specific single cancer-induced muscle wasting in gynecologic cancer patients has not been
nucleotide variants (SNVs) were identified from paired tumor/normal whole- elucidated. Methods: The correlation between serum level of IL-6 and
exome sequencing (WES), and annotated as expressed if observed in . = 2 skeletal muscle volume in the patients with gynecologic cancers that re-
RNAseq reads. MHC1 binding affinity for 9-mer neoepitope peptides ceived multiple lines of therapy was retrospectively evaluated. We used the
resulting from said SNVs were predicted using NetMHC within presented psoas muscle index [PMI (cm2/m2)], the psoas major muscle area at the fifth
HLA-types. Cases with . 200 non-synonymous exonic mutations were lumber level divided by the height squared, measured using digital axial CT
designated as TMB-high in accordance with Rizvi et al, 2015. Tumor im- images, for the value of skeletal muscle volume. The level of IL-6 cut-off for
mune activity was inferred by RNAseq expression of 6 checkpoint/TME elevation was defined as more than 12.0 pg/mL. The comparison of the
markers, as well as by estimating immune infiltration using RNAseq survival distributions from the day of IL-6 measuaments was made using a
deconvolution of immune genesets (Bindea et al 2013). Significant asso- log-rank test. Results: A total of 74 cases were assessed for the serum IL-6
ciations between TMB, neoantigen-load, expressed neoepitope binding and PMI: 32 cases with Mullerian cancers, 24 cases with endometrial
affinities, and immune activity were analyzed. Results: Within a clinical cancers, 13 cases with cervical cancers, and 5 patients with others. The
database of 1,363 cases with T/N/R sequencing, a total of 147,015 po- group with elevated IL-6 were associated with the lower PMI (t-test, p =
tential neoepitopes were identified. A small but significant enrichment was 0.0154). The patients with IL-6 elevation had significantly worse survival
observed for silencing neoepitopes that are predicted to bind MHC1 (OR = compared with those with normal IL-6 (1y-OS; 31% vs. 80%, p , 0.0001).
1.22, p = 2.4e-78 one-sided Fishers test). The silencing rate was similar In 28 patients with more than two-point measurements of IL-6, the patients
between the 17% of patients with high TMB vs others, but was increased in with the decrease to the level of IL-6 cut-off had favorable survival compared
35% of all patients with high inferred immune infiltration (N = 490, OR = with those without the decrease (6m-OS; 100% vs. 52%, p = 0.0069).
1.30, p = 1.8e-31). A further silencing enrichment was observed in 19% of Conclusions: Serum level of IL-6 could be a sentinel biomarker for cancer-
all patients displaying high immune activity but low PDL1 expression (N = induced sarcopenia in the patients with gynecologic cancers. Additionally,
263, OR = 1.44, p = 4.0e-45). Conclusions: We observe significant pref- IL-6 could be a biomarker to determine further continuation of aggres-
erential silencing of MHC binding neoepitopes. Specifically, when tumor sive chemotherapy for the patients that had received multiple lines of
infiltrating immune cells are activated, silencing neoepitopes may be an chemotherapy.
alternative to checkpoint expression for avoiding an immune cascade. Pa-
tients with TILs and silenced neoepitopes may benefit from epigenetic
priming therapy prior to ICI therapy.

2593 Poster Session (Board #237), Sat, 8:00 AM-11:00 AM 2594 Poster Session (Board #238), Sat, 8:00 AM-11:00 AM
Association of an inflammatory gene signature with CD8 expression by CD8+ T cells in tumor parenchyma and stroma by image analysis (IA) and
immunohistochemistry (IHC) in multiple tumor types. First Author: Peter M gene expression profiling (GEP): Potential biomarkers for immuno-oncology
Szabo, Bristol-Myers Squibb, Princeton, NJ (I-O) therapy. First Author: Peter M Szabo, Bristol-Myers Squibb, Princeton,
NJ
Background: A multiparameter tumor inflammation assay based on gene
expression profiling (TIA-GEP) can extend the utility of IHC to interrogate the Background: Distribution patterns of CD8+ T cells within the tumor mi-
tumor microenvironment (TME). Using CD8 expression assessed by IHC croenvironment (TME) can be assessed by IA, which may reflect underlying
(CD8-IHC) as a surrogate for inflammation, statistical modelling was used to tumor biology and serve as a potential biomarker to assess the utility of I-O
develop a specific gene signature on the TIA-GEP panel to predict CD8-IHC. therapy. These patterns are variable and may be classified as immune desert
The correlation between TIA-GEP and CD8-IHC and the prevalence of in- (minimal infiltrate), excluded (T cells confined to tumor stroma or to the
flammation were explored across multiple tumor types. Methods: Levels of invasive margin), or inflamed (T cells diffusely infiltrating tumor parenchyma
inflammation were measured by CD8-IHC and TIA-GEP on 1778 procured and stroma). We hypothesized that association of a GEP signature with
samples across 12 tumor types. Quality control metrics involved sample input abundance of parenchymal and stromal T-cell infiltrates may identify bio-
quality, technical errors, and inter-run variability. Generalized linear models markers of response or resistance to I-O therapy. To test this, we applied an
were used to identify an inflammation score that predicts the CD8-IHC score in AI-powered IA platform to quantify CD8+ T cells by geographical location
melanoma and SCCHN tissue. The predictive accuracy of this signature was and used GEP to define both CD8 abundance and associated geographic
also examined in 10 additional tumor types. Results: Assessment of TME localization to tumor parenchyma and stroma. Methods: We performed an
inflammation by CD8-IHC was consistent with that observed by TIA-GEP in analysis using a tumor inflammatory GEP assay and CD8 immunohisto-
multiple tumor types. The range of inflammation varied across different tumor chemistry on procured specimens (335 melanoma, 391 SCCHN). Digitized
types, with relatively lower inflammation range and scores in SCLC, ovarian, slides were used to train a convolutional neural network to quantify the
and prostate cancers, and higher values in NSCLC, melanoma, SCCHN, and number of CD8+ T cells in stroma, tumor parenchyma, parenchyma-stromal
gastric cancers. R2 x 100 values reflecting percent variation in CD8-IHC interface, and invasive margin. Generalized constrained regression models
associated with TIA-GEP ranged from 62.4% to 79.2% (P , 0.0001) for all were used to predict GEP signatures specifically for stromal and paren-
tumor types except prostate cancer (32.5%). Low correlation in prostate chymal CD8+ T cells. Results: Parenchymal and stromal GEP scores were
cancer may be a result of low prevalence of inflammation by CD8-IHC. Es- highly concordant with CD8+ infiltrate geography (adj-r2: 0.67, 0.65, re-
timated linear regression slopes between CD8-IHC and TIA-GEP ranged from spectively; P # 0.01). Little overlap existed between gene sets associated
0.74 in SCLC to 1.27 in gastric cancer. Conclusions: The results suggest that with parenchymal and stromal CD8 T-cell geographies. CSF1R and NEC-
the inflammation signature is a robust potential diagnostic tool predicting TIN2 gene expression was observed to correlate inversely with parenchymal
inflammation in the TME. The inflammation signature not only correlates with localization and directly with stromal CD8+ T-cell abundance. Conclusions: GEP
CD8-IHC for multiple tumor types, but also leverages the alternative benefits signatures can be identified that are concordant with various CD8+ T-cell
associated with TIA-GEP, which include information related to tumor localization patterns in melanoma and SCCHN, demonstrating that GEP-IA
inflammation-associated biomarkers and flexibility in exploring the value of can be developed to identify the immune status of interest in the TME. The
other genomic signatures. specific genes identified have potential to elucidate mechanisms of re-
sistance and/or inform I-O targets that can be further evaluated in relation to
clinical significance in future studies.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Immunotherapy and Tumor Immunobiology 129s

2595 Poster Session (Board #239), Sat, 8:00 AM-11:00 AM 2596 Poster Session (Board #240), Sat, 8:00 AM-11:00 AM
Evaluation of tumor microenvironment and biomarkers of immune check- Dynamic changes of neutrophil-to-lymphocyte ratio (NLR), platelet-to-
point inhibitor (ICI) response in metastatic renal cell carcinoma (mRCC). lymphocyte ratio (PLR), and lactate dehydrogenase (LDH) during treatment
First Author: Jason Zhu, Department of Medicine, Duke University School of with immune checkpoint inhibitors (ICIs) in non-small cell lung cancer (NSCLC).
Medicine, Durham, NC First Author: Alessandro Russo, Medical Oncology Unit-A.O. Papardo, Messina
and Department of Human Pathology University of Messina, Messina, Italy
Background: ICIs are now standard of care for mRCC; however, there are few
biomarkers to predict ICI response. Recent data from atezolizumab/bevacizumab Background: ICIs have revolutionized the therapeutic landscape of NSCLC.
trials in mRCC suggest tumors with high Thigh
eff /PD-L1+ are more likely to respond However, with the exception of PD-L1 expression, predictive biomarkers are
to ICI. Here, we use two gene panels as well as other inflammation markers in lacking. The aim of this study was to evaluate the dynamic changes of some
the tumor microenvironment to correlate with ICI responses. Methods: This markers of inflammation over time and the outcome of NSCLC patients (pts)
multicenter study evaluated 86 patients (pts) with mRCC treated with ICIs. treated with nivolumab (N) or pembrolizumab (P). Methods: All consecutive
FFPE tumor samples were evaluated by RNA sequencing for Teff status. Two NSCLC pts treated with N or P between Aug. 2015-Dec. 2018 were ana-
gene panels were analyzed: a Teff Gene Panel (CD8, CD27, IFNG, GZMA, lyzed. Laboratory results were collected at baseline, 6 weeks (6-wk), and
GZMB, PRF1, EOMES, CXCL9, CXCL10, CXCL11, CD274, CTLA4, FOXP3, 12 weeks (12-wk) and correlated with the outcome. NLR and PLR were
TIGIT, IDO1, PSMB9, TAP1) and a 5-Gene panel (FOXP3, CCR4, KLRK1, defined as absolute neutrophil and platelet count divided by lymphocyte
ITK, and TIGIT) based on the gene expression pattern of tumors in our cohort. count, respectively. NLR $5, PLR $200, and LDH levels $ upper normal
Objective response rates (ORRs, defined as CRs and PRs) were correlated limit were considered high. Overall survival (OS) was defined as time from ICI
with PD-L1 status (positivity was defined as $1% TPS based on Dako 22C3 start to death and Progression Free Survival (PFS) as time from treatment
IHC assay), and TMB (0-10, 10-20, $20 mut/Mb), and tumor inflammation start to progression disease or death for any cause. OS and PFS curves were
(high CD8 expression compared to a large reference population). Best re- estimated using the Kaplan–Meier method and compared with the log-rank
sponses to ICI was determined by an expert radiologist using RECIST 1.1 test. Results: We included 71 consecutive NSCLC pts treated with either N
criteria. Inflamed tumor status, Teff gene panel, 5-gene panel, PD-L1 status, (75%) or P (25%). Baseline characteristics: median age 69 years (range 46-
and TMB were associated with ORR and tested using a chi-squared test with 80), sex male 76%, squamous histology in 39%. PD-L1 expression (39/71):
Yates’s continuity correction. Results: ORR was 50% (4/8) for PD-L1 , 1% in 20%, 1-49% in 45%, and $50% in 35%. NLR $5 was associated
positive pts and 14% (9/65) for PD-L1 negative pts (p = 0.042). The with lower PFS and OS, with an increased predictive value over time (p =0.01
majority of tumors (95%, 82/86) had TMB , 10 mut/mb. 43 pts (50%) were and p =0.009 at baseline; p =0.007 and p ,0.001 at 6-wk; p ,0.001 and
classified as Thigh low
eff and 43 pts were classified as Teff . ORR was 23% (10/43) p ,0.001 at 12-wk, respectively). PLR $200 at baseline and 12-wk was
in the Thigh low
eff cohort and 12% (5/43) in the Teff cohort (p = 0.256). ORR was significantly associated with shorter OS (p =0.05 and p =0.004, re-
31% (14/45) in the 5-Gene high cohort and 2% (1/41) in the 5-Gene low spectively), but no in terms of PFS at all the three time points. Finally,
cohort (p = 0.001). Conclusions: TMB and tumor inflammation based on LDH $UNL at baseline was associated with shorter PFS and OS (p =0.02
CD8 did not reliably predict for objective responses in this study of mRCC pts and p =0.03), as well as a reduction of LDH levels at 12-wk compared with
treated with ICIs. Gene expression signatures provide a more comprehensive baseline values (p =0.006 and p =0.004). Conclusions: Baseline evaluation
evaluation of the tumor microenvironment and may lead to better predictive of NLR, PLR and LDH levels is significantly associated with outcome in
biomarkers for ICI response than individual biomarkers such as PD-L1, TMB, NSCLC treated with single agent ICIs. Moreover, dynamic changes of LDH
or CD8 expression. levels at 12 weeks significantly predicted outcome. These easy to determine
parameters may have a place in the selection process of pts candidate for
immunotherapy.

2598 Poster Session (Board #242), Sat, 8:00 AM-11:00 AM 2599 Poster Session (Board #243), Sat, 8:00 AM-11:00 AM
Application of artificial intelligence to predict a new class of novel synthetic A dose-finding study of the SMAC mimetic Debio 1143 when given in
lethal targets. First Author: Spyro Mousses, Systems Oncology, Scottsdale, combination with avelumab to patients with advanced solid malignancies.
AZ First Author: Rosalyn A. Juergens, Juravinski Cancer Centre, McMaster
University, Hamilton, ON, Canada
Background: Synthetic lethal targets are proteins that are contextually
vulnerable. Inhibitors of PARP1, for example, selectively produce a lethal Background: Second mitochondria-derived activator of caspase (SMAC)
phenotype in the context of cancer cells which have lost BRCA1 or BRCA2 mimetics regulate apoptosis and modulate NFkB signaling which drives the
function. As a high mutation rate is a hallmark of many cancers, targeting expression of genes involved in immune and inflammatory responses. In
synthetic lethal interactions to selectively inhibit cancer cells with altered patient (pt) tumors, Debio 1143 increased PD-1/PD-L1 expression and
genetic backgrounds may increase the specificity and efficacy of thera- tumor infiltrating lymphocytes. In pre-clinical models, it synergizes in vitro
peutics. Recently, clinical trials have targeted synthetic lethal pairs such as and in vivo with PD1/PD-L1 checkpoint inhibitors (CPIs). Methods: In a
EGFR and BRAF, TP53 and BCL2, and PTEN and CHD1. Previous attempts phase I study, using a mCRM model, avelumab (10 mg/kg i.v. on D1&15
to identify synthetic lethal targets have relied on empirical results from q4w) was combined with escalating doses of Debio 1143 (100 mg/d to 250
published studies of biological pathways perturbed in cancer cells. De- mg/d orally, D1-10 & D15-24 q4w) to define the RP2D. Consenting adult pts
veloping strategies to rapidly identify synthetic lethals by combining multiple with advanced solid tumors, normal organ function, and PS-ECOG = 0-1
experimental and computational approaches would result in a new class of were eligible provided none received prior CPI. Dose-limiting toxicities
potential cancer drug targets beyond the existing efforts that rely on single (DLTs), efficacy, safety, PK, PD and biomarkers were assessed. Results: As
experimental or computational methods alone. Methods: Here we present of DEC’18, 16 pts were treated; M/F: 8/8; ECOG = 0 in 6 (38%); median
Expansive AI, an artificial intelligence augmented knowledge network that age = 58 (28-79); 5 pts had NSCLC, 2 MPM, 2 ovarian and 7 had other tumors
enables rapid hypothesis generation for accelerated discovery research. (n = 1 each). Common AEs were: nausea (69%); fatigue (62%); vomiting
Using a purpose-built, hypergraph database of massive, integrated genomic (50%); cough, dyspnea, myalgia (44% each); diarrhea, anorexia (38%
and biomedical data, we can query all synthetic lethals and their component each); pruritus and constipation (31% each). These were generally grade 1-
genes, as well as a wealth of data related to these genes. The database of 2, occasionally grade 3. One pt had a DLT at 250 mg/d dose: a grade 3 AST/
biological data includes 11,000+ cancer genomes from TCGA, prior ALT increase. No treatment-related AEs grade 4 or higher occurred. No dose-
knowledge resources such as gene ontology and pathway resources, and relationships for laboratory abnormalities were observed, except for ALT/AST
experimental data including chemical and protein interaction and patent increases, which at 200 mg/d were all grade 1 and asymptomatic. Maximal
data. The hypergraph’s architecture allows for linking and nesting data, tolerated dose was not reached and there were no dose reductions. In 15
enabling efficient extraction of biologically-relevant features. Results: Using evaluable pts, 1 PR (NSCLC) and 5 SD (RECIST v1.1) were observed. Tumor
these features, a neural network classified 540 new candidate pairs that shrinkage . 15% was seen in 2 other NSCLC pts. PK showed high inter-
have previously not been reported. The candidate pairs were filtered to patient variability and dose-proportional increase. TNFa and IFNg peaked in
include only known oncogenes and least-studied genes. This produced a list plasma following Debio 1143 dose on D1 after 8 hrs, and on D17/22, in a
of gene pairs which may represent the most novel class of synthetic lethal dose-proportional manner. Four pts developed anti-avelumab antibodies.
target candidates identified to date. Conclusions: We highlight the results of Conclusions: Debio 1143 at 200 mg/d can be safely combined with ave-
this AI-based approach and discuss validation efforts of the predicted in- lumab. Toxicity was predictable and mild. Clinical activity was observed in
teractions in specific cancer contexts. NSCLC pts. PK was linear; no drug interaction was suspected. PD and
biomarker analysis is ongoing. Expansion at this RP2D is ongoing in NSCLC.
Clinical trial information: NCT03270176.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
130s Developmental Immunotherapy and Tumor Immunobiology

2600 Poster Session (Board #244), Sat, 8:00 AM-11:00 AM 2601 Poster Session (Board #245), Sat, 8:00 AM-11:00 AM
First-in-human (FIH) trial evaluating immune activation and safety of PIN- 2 Preliminary results from CLASSICAL-Lung, a phase 1b/2 study of pepinemab
administered intravenously to patients with advanced solid tumors. First (VX15/2503) in combination with avelumab in advanced NSCLC. First
Author: Colin Bier, PIN Pharma, Inc., New York, NY Author: Michael Rahman Shafique, Department of Thoracic Oncology,
Moffitt Cancer Center and Research Institute, Tampa, FL
Background: Innate immunity is an integral component necessary to initiate
systemic anti-tumor immunity in patients with progressive malignancy. PIN- Background: Rational combination therapies are needed to overcome re-
2 is a novel immunomodulating agent derivatized from a transactivator (Tat) sistance mechanisms in NSCLC. Pepinemab is an IgG4 humanized
protein that stimulates innate immunity in vivo by promoting differentiation monoclonal antibody targeting semaphorin 4D (SEMA4D, CD100). In vivo
of peripheral blood monocytes into activated APC’s linking the innate and preclinical models demonstrated antibody blockade of SEMA4D promoted
adaptive immune systems resulting in endogenous T-cell priming against a immune infiltration and reduced function and recruitment of immunosup-
multitude of tumor associated antigens unique to the individual. Methods: A pressive myeloid cells within the tumor. Importantly, preclinical combina-
FIH clinical trial was conducted in patients (pts) with extensively pretreated tions of anti-SEMA4D with various immunotherapies enhanced T cell activity
solid tumors to evaluate the pharmacodynamics (PD) and safety of PIN-2. and tumor regression. The CLASSICAL-Lung clinical trial tests the combi-
8 pts (2 men), mean age 62.7 (67.9) years, and a median of 4.5 prior nation of pepinemab with avelumab to couple immune activation via
treatment lines, were enrolled in 2 Australian centers. 2 pts received 1 cycle checkpoint inhibition with beneficial modifications of the immune micro-
of treatment and 6 received 2 cycles. Pts were given 300 mg of PIN-2 IV 3 environment via pepinemab. Methods: This ongoing phase 1b/2, open label,
times/wk for 2 wks followed by a 1 wk rest period. A 2nd cycle of treatment single arm, first-in-human combination study is designed to evaluate the
was offered based on pt and investigator preference. Plasma was collected at safety, tolerability and efficacy of pepinemab in combination with avelumab
6 and 24H post-infusion 1 to evaluate immune activation. Th1 cytokines in 62 patients (pts) with advanced (stage IIIB/IV) NSCLC (NCT03268057).
(TNF-a, IFN-g, IL-12, CSF-2) were analyzed to assess PD activity. Results: A The trial is split into dose escalation (n = 12) and dose expansion (n = 50)
significant increase in TNF-a was seen 6H following PIN-2 infusion (p = phases and includes 2 cohorts; 1) pts who are immunotherapy naı̈ve, and 2)
0.0142), demonstrating rapid onset of immune activation. There were no pts whose tumors progressed during or following immunotherapy (IO failure).
clear changes in the other parameters evaluated. PIN-2 was rapidly cleared Pts in the dose escalation cohorts received ascending doses of pepinemab
from plasma, with mean T1/2 = 24.0 (8.07) min, Tmax = 1.06 (0.665) min, i.v. (5, 10, 20 mg/kg, Q2W) in combination with avelumab i.v. (10mg/kg,

·
Cmax= 77,500 (61,600) pg/mL, and AUC0-inf = 690,000 (493,000)
pg min/mL. 3 pts discontinued treatment, 2 for adverse events (1 gr 2
infusion reaction, 1 unrelated SAE of abdominal pain) and 1 for disease
Q2W). Results: Dose escalation is complete and the RP2D was selected as
10mg/kg pepinemab, Q2W. No pts experienced a TRAE leading to study
discontinuation or death. The most frequent related AEs were grades 1 or 2
progression on day 12. Treatment related AEs were grade 1 and 2, and readily fatigue, pyrexia, or chills; no grade 3 AEs occurred in more than one subject.
managed. There was a single unrelated gr 3 event (anemia) and no AEs . gr3. One DLT, a grade 3 pulmonary embolism occurred in the 10mg/kg pepi-
2 pts developed anti-drug antibodies; however, these did not result in changes nemab cohort, and resolved without reoccurrence. The disease control rate
in the immuno-PD profile. Conclusions: PIN-2 was generally well tolerated for pts treated . 2 months is 90% (19/21), and, at this early stage, a PR
with an acceptable safety profile. PIN-2 caused an early increase in TNF-a, with a 49% reduction in target lesion was observed in at least 1 of 8 pts in the
consistent with PD activity predicted by preclinical data. Further study alone IO failure cohort. Updated data from the dose expansion phase will be
and in combination with other agents in pts with advanced solid tumors is presented. Conclusions: Preliminary data suggest the combination is well
warranted. Clinical trial information: ACTRN12617001597381. tolerated and shows initial signals of antitumor activity. Dose escalation is
complete and the expansion phase is ongoing. Clinical trial information:
NCT03268057.

2602 Poster Session (Board #246), Sat, 8:00 AM-11:00 AM 2603 Poster Session (Board #247), Sat, 8:00 AM-11:00 AM
Safety profile of INT230-6, a novel intratumoral (IT) formulation, during THOR-707: Using synthetic biology to reprogram the therapeutic activity of
injections into a variety of refractory deep and superficial tumors with evidence interleukin-2 (IL-2). First Author: Marcos E Milla, Synthorx Inc. Research &
of tumor regression and immune activation. First Author: Anthony B. Development, La Jolla, CA
El-Khoueiry, University of Southern California Norris Comprehensive
Background: Recombinant interleukin-2 (rIL-2; aldesleukin) is an approved
Cancer Center, Los Angeles, CA
immunotherapy in melanoma and renal cell carcinoma based on complete
Background: INT230-6 is comprised of cisplatin (CIS), vinblastine (VIN) durable remissions. The anti-neoplastic properties of IL-2 are mediated by
and an amphiphilic penetration enhancer which facilitates dispersion interactions with the beta-gamma chain (IL-2Rbg) on naı̈ve CD8+ T cells,
throughout tumors and diffusion into cancer cells. In preclinical experi- which lead to their expansion and differentiation into T effector and T
ments, INT230-6 led to necrosis and recruitment of immune cells with high memory cells directed against the tumor. However, the widespread use of IL-
rates of complete responses of injected and bystander tumors. This abstract 2 in oncology is limited by interaction with the high affinity IL-2 receptor
highlights the safety and early pharmacodynamic activity of this approach. alpha chain (IL-2R⍺) on regulatory CD4+ T cells (Tregs), which leads to
Methods: Patients with solid tumors that progressed on all standard treat- immunosuppression, and on innate lymphoid cells in the vascular endo-
ments were enrolled. Dose escalation occurred by increasing number of thelium, which leads to eosinophilic recruitment and activation, and the
tumors injected, loading per tumor, and total dose. INT230-6 was injected sometimes fatal complication of vascular leak syndrome (VLS). A rIL-2
once every 2 weeks in multiple lesions for 5 sessions. Patients were mon- biased toward IL-2bg affinity with no IL-2R⍺interaction could fill unmet
itored for safety and tolerability weekly. Pharmacokinetic(Pk) samples and needs in oncology. Methods: Using a synthetic biology platform, we have
peripheral blood were collected for flow cytometry and circulating cytokines. engineered THOR-707, a rIL-2 that contains a novel amino acid encoded in
Pre and on study biopsies are ongoing. Results: 28 patients (14 unique the IL-2 gene via a new DNA base pair (X-Y). The novel amino acid serves as a
cancer types) receiving a median of 3 prior treatments were enrolled. Doses hook for site specific pegylation that extends half-life, blocks IL-2R⍺ en-
from 0.3 ml up to-80 ml of INT230-6 were given into single lesions with gagement and binds to the IL-2Rbg. Results: In non-human primates, THOR-
some patients receiving a total of 120 mL ( = 9.7mg VIN exceeding the IV VIN 707 can be dosed to maximize the level of cytotoxic CD8+ T lymphocytes
dose) without significant systemic absorption or typical cytotoxic adverse without elevation of VLS-inducing eosinophils. In murine tumor models,
events. Pk analysis suggests that systemic exposure of VIN or CIS is ~10% of THOR-707 induced the expansion of peripheral and intratumoral CD8+ T cells
injected. No DLT’s or drug-related SAE’s reported. The most frequent ad- without expansion of suppressive Tregs. Single-agent dose-dependent anti-
verse event was grade 1 or 2 pain at injected site. Superficial tumors showed tumor efficacy was observed in two syngeneic mouse models. In combination
signs of response including flattening, areas of necrosis and ulceration. with a PD-1 inhibitor, survival of tumor-bearing mice was longer than either
Tumor reduction, apparent in in both injected and bystander tumors, may agent as monotherapy. Efficacy in tumor models was durable, suggesting
indicate an abscopal effect. An increase . 30% in CD8 T-cells was seen in activation of CD8+ memory T cell populations. Conclusions: THOR-707 is a
the blood of 3/9 evaluable patients. Conclusions: INT230-6 was safe and reprogrammed, site-directed, singly-pegylated rIL-2 that changes the phar-
well tolerated in . 100 injections (28 patients) with encouraging activity and macologic profile of IL-2, potentially providing a favorable risk-benefit profile.
pharmacodynamic effects in advanced refractory tumors. Additional analysis First-in-human studies are expected to begin this year evaluating THOR-707
of immune cells from on study biopsies will be presented. A new cohort will as monotherapy and in combination with a PD-1 inhibitor. Based on pre-
evaluate combination with an anti-PD1 antibody to understand if local tumor clinical evidence to-date, THOR-707 may potentially address existing and
destruction can increase systemic antigen load, increase immune cell emerging unmet needs across multiple solid tumors.
recognition and initiate a systemic immune response. Clinical trial in-
formation: NCT 03058289.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Immunotherapy and Tumor Immunobiology 131s

2604 Poster Session (Board #248), Sat, 8:00 AM-11:00 AM 2605 Poster Session (Board #249), Sat, 8:00 AM-11:00 AM
AB928, a novel dual adenosine receptor antagonist, combined with chemotherapy 2D and 3D thermally bioprinted human MCF-7 breast cancer cells: A promising
or AB122 (anti-PD-1) in patients (pts) with advanced tumors: Preliminary results model for drug discovery. First Author: Aleli Campbell, University of Texas
from ongoing phase I studies. First Author: John D. Powderly, Carolina at El Paso, El Paso, TX
BioOncology Institute, Huntersville, NC
Background: Breast cancer (BC) is the second leading cause of cancer death
Background: AB928, a selective, small-molecule A2aR/A2bR antagonist, po- following lung cancer. Bioprinting, the use of computer aided process to
tently blocks the immunosuppressive effects of high adenosine concentrations print biological living and non-living material to create patterns in 2D or 3D
in the tumor microenvironment. Preclinically, combining adenosine receptor structures, is a novel technique that has been proposed to be used to develop
inhibition with either chemotherapy or anti-PD-1 resulted in greater tumor tissue engineered solutions for a wide array of clinical applications, e.g., skin
control, suggesting AB928 may have additive activity when paired with either of grafting. We investigate here if bioprinted breast cancer cells show some of
these agents in cancer pts. Methods: Three dose-escalation (3+3 design) the hallmarks of cancer tissues, and thus may represent good in vitro models
studies are assessing the safety, pharmacokinetics (PK), pharmacodynamics, for drug discovery. Methods: For this study, MCF-7 BC cells were cultured,
and clinical activity of increasing doses of AB928 (75, 150, 200 mg orally once stained, counted and turned into a bioink solution by suspending in
daily) in combination with: standard pegylated liposomal doxorubicin in triple- phosphate buffered saline solution. The cells were bioprinted over a 96-well
negative breast cancer (TNBC) and ovarian cancer (OC); standard mFOLFOX6 plate and pre-incubated for 18 hours in DMEM and RPMI media with 10%
in gastroesophageal cancer (GEC) and colorectal cancer (CRC); and AB122 Fetal Bovine Serum and Charcoal Stripped Serum, respectively. After 18
(240 mg every 2 weeks) in various advanced tumors. Following identification hours of incubation the media was supplemented with Tamoxifen at 5mM,
of the recommended phase 2 dose of AB928 in combination with che- 10mM, 50mM, 90mM and 110mM concentrations. Cytotoxicity was mea-
motherapy or AB122 in dose escalation, the following tumor cohorts may be sured 24 hours post-treatment using a differential nuclear staining assay and
expanded (15-40 pts/cohort) to further test the combinations: TNBC and an INCell 2000 bioimager system. Results: Bioprinted cells exposed to high
OC, GEC and CRC, and renal cell carcinoma. Results: As of 01Feb2019, 9 concentrations of Tamoxifen (90 mM and 110mM) exhibited a viability of
pts were treated across the 3 studies, and time on treatment ranged from 1- 8.2% and 10.8%, respectively. Whereas viability of manually seeded cells at
182 days (table). Overall, AB928 combination therapy was well tolerated. those concentrations was 0.11% and 0.05%. Viability of negative and
Two pts underwent post-baseline disease assessment; both had stable positive controls was at 7.6% and 97.0% for the bioprinted samples and
disease. Preliminary data indicate that AB928 PK and adenosine receptor for the normally seeded cells was 4.9% and 98.8% respectively.
coverage in cancer pts are similar to what was previously assessed in healthy Conclusions: In our study, we have established a novel 2D/3D breast tumor
volunteers. AB122 PK and PD-1 coverage are equally unaffected by AB928 co- model applying bioprinting technology for drug discovery. The higher cell
administration. Updated data, including biomarker data, will be presented at viability of MCF-7’s at high concentrations of Tamoxifen could be attributed
the meeting. Conclusions: Early results showed a favorable safety profile of to the hormesis effect and activation of chaperone proteins, e.g., HSP70 and
AB928 combination therapy. All 3 studies are actively recruiting pts. Clinical HSP90, possibly caused by bioprinting. We hypothesize that bioprinted
trial information: NCT03719326; NCT03720678; NCT03629756. MCF-7 cells also show increased levels of chaperone proteins, which may in a
Study in TNBC & OC Study in GEC & CRC Study in various tumors way mimic their in vivo behavior. In this novel in vitro 2D/3D model, the
(NCT03719326) (NCT03720678) (NCT03629756) bioprinted cells show a more biological relevant behavior than normally
Pts treated 1 2 6 cultured cells. Insights into the cell behavior after bioprinting may elucidate
Days on treatment 49 1-7 7-182
Safety, n how to build improved in vitro models for BC research.
Related ‡G3 AE 0 0 0
SAE 0 1 2
Related SAE 0 0 0
Dose-limiting toxicity 0 0 0
Best response, n
Stable disease Not yet assessed Not yet assessed 2

2606 Poster Session (Board #250), Sat, 8:00 AM-11:00 AM 2607 Poster Session (Board #251), Sat, 8:00 AM-11:00 AM
Balixafortide (a CXCR4 antagonist) + eribulin in HER2-negative metastatic Immune profiling of tumor-infiltrating T cells using mass cytometry. First
breast cancer (MBC): Survival outcomes of the phase I trial. First Author: Author: David Roumanes, Immunoscape, Cambridge, MA
Peter A. Kaufman, Breast Oncology, Division of Hematology/Oncology,
Background: Immunotherapy recent successes have opened new avenues
Burlington, VT
for the treatment of cancer and the presence of tumor-specific CD8+ T cells
Background: Balixafortide (B) is a potent antagonist of the chemokine re- in tumor-bearing individuals offer a promising therapeutic target. However,
ceptor CXCR4. Preclinical evidence suggests that disrupting CXCR4 de- the detection and profiling of such T cells are challenging due to the need to
pendent pathways prevents development of breast cancer metastases, detect rare antigen-specific T cell subpopulations in patient samples that are
enhances the cytotoxic effect of chemotherapy and immunotherapy, and limited in size thus making it difficult to exploit these parameters for pre-
counteracts tumor cell evasion of the immune system. Encouraging safety and dictive signatures of clinical response. Moreover, the identification and
efficacy data were published recently from the ongoing Phase 1 trial in- analysis of neoantigen-specific CD8+ T-cells in tumor-bearing individuals is
vestigating B + eribulin (E) in patients with HER2 negative MBC (Pernas S. challenging due to the small pool of such cells. Methods: In order to identify
et al. Lancet Oncol. 2018; 19: 812224). The objective response rate, median therapy-relevant tumor antigens and to facilitate a concurrent in-depth
progression free survival and median overall survival (OS) for the expanded characterization of cells directed towards these targets, immunoSCAPE
cohort (EC) and the overall efficacy population (OEP) were 37.5% and 29.6%, leverages the high-dimensional immune profiling capabilities of cytometry
6.2 months and 4.5 months, and 18 months and 16.8 months, respectively. by time of flight (CyTOF) combined with a unique technology allowing the
Here we report the 18 and 24 months landmark OS data from this trial. identification rare antigen-specific T-cell subsets. Results: We applied this
Methods: This trial enrolled 56 patients with HER2-negative, CXCR4-positive technology to patient tumor-infiltrating lymphocytes from human cancer
MBC, previously treated with 123 chemotherapy regimens for MBC. A 3+3 samples and tumor-derived neoantigens recognized by T-cells were iden-
dose escalation design was used, followed by an EC. All cohorts received E on tified and characterized. Interestingly, the majority of patient-derived tumor
days 2 and 9, and B on days 123 and 8210 of 21 day cycles. The association infiltrates consisted of tumor-unrelated T-cells characterized by a diverse
between various baseline biomarkers and treatment outcomes including OS is phenotype. Strikingly, the expression of CD39 was absent from these by-
currently being investigated in a multivariate analysis (MVA). Results: Landmark stander cells, suggesting that CD39 could be a useful biomarker for the
survival data for the trial are shown in the table. Clinical trial information: identification of putative tumor-reactive T cells. Conclusions: Simultaneous
NCT01837095. Conclusions: Landmark 18 months and 24 months OS data are immune profiling revealed that tumor-unrelated, bystander CD8+ T-cells are
consistent with the positive trend of all efficacy read-outs observed in this study phenotypically different in human tumor infiltrates and identified CD39 as a
and safety information is consistent with what was previously reported. Although putative marker of neoantigen-specific T-cells. By providing insights into the
inter-trial comparisons should be interpreted with caution, these survival rates, nature, frequency and phenotype of antigen-specific T-cells, immuno-
especially for the EC, are higher than those reported for eribulin monotherapy in SCAPE’s unique target discovery and high-dimensional immune profiling
similar MBC populations. These promising results suggest that B + E could platform is a valuable tool for the development of novel diagnostic and
potentially provide a new treatment option in heavily pre-treated patients with therapeutic strategies in immunotherapy.
HER2 negative MBC and this is currently being investigated in a pivotal, ran-
domized trial.
Landmark (months) OS for EC (95% CI) OS for OEP (95% CI)
18 50% (29.1–67.8) 42.4% (28.9–55.2)
24 33.3% (15.9–51.9) 25% (14.3–37.3)
The results from the MVA will be presented at the meeting.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
132s Developmental Immunotherapy and Tumor Immunobiology

2608 Poster Session (Board #252), Sat, 8:00 AM-11:00 AM 2609 Poster Session (Board #253), Sat, 8:00 AM-11:00 AM
Phase I study of KN035, the first subcutaneously administered, novel fusion Phase I safety and pharmacokinetic study of KN035, the first subcutaneously
anti-PD-L1 antibody in patients with advanced solid tumors in China. First administered, novel fusion anti-PD-L1 antibody in Japanese patients with
Author: Jian-Ming Xu, 307 Hospital of PLA, Beijing, China advanced solid tumors. First Author: Toshio Shimizu, National Cancer Center
Hospital (NCCH), Tokyo, Japan
Background: KN035 is a novel fusion protein of humanized anti-PD-L1
single domain antibody and human IgG1 Fc fragment, formulated for Background: KN035 is a novel fusion protein of humanized anti-PD-L1 single
subcutaneous (SC) injection. Methods: The escalation phase followed a domain antibody and human IgG1 Fc, formulated for subcutaneous (SC)
modified 3+3 design with a 28-day DLT evaluation period and 8 dose levels injection. A phase I safety and pharmacokinetic (PK) study was conducted in
were planned at 0.1, 0.3, 1.0, 2.5, 5 and 10 mg/kg SC weekly. One patient Japanese patients. Methods: Patients with advanced solid tumors were
each was enrolled at 0.1 and 0.3 mg/kg dose levels. Additional dose levels treated with KN035 SC once every-7-days (QW) or once every-14-days (Q2W)
followed traditional 3+3 design. Response was assessed per RECIST 1.1 schedules with the dose limiting toxicities (DLT) evaluation period of 28 days.
every 12 weeks. Results: As of 11/2/2018, 17 patients were enrolled in the For the QW schedule, the starting dose was 1 mg/kg (n=3) with escalations to
escalation phase (urothelial carcinoma (n=2), hepatic cell carcinoma (n=2), 2.5 (n=4), and 5 (n=3) mg/kg. For the Q2W schedule, 6 patients were planned
intrahepatic cholangiocarcinoma (n=2), thymic carcinoma (n=2), colorectal at the dose levels of 2.5 and 5 mg/kg. Results: No DLT was observed up to the
cancer£¨n=2£©£¬renal cell carcinoma (RCC, n=3), Squamous-cell lung highest dose level of 5 mg/kg QW. No maximum tolerated dose (MTD) was
carcinoma (n=1) and ovarian cancer (n=1)). The majority of subjects had reached. Among evaluable treated subjects (n=14), there were two confirmed
advanced disease stage, stage IV (15/17) and stage III (2/17). A total of 7 partial responses. Preliminary PK analysis suggested that after SC adminis-
subjects received radiotherapy, 16 subjects received surgery, and 13 tration, KN035 was slowly absorbed (Tmax ~ 4 d) and the mean residual time
subjects received systematic anti-cancer therapies from previous treatment. (MRT) was 21 days. Apparent clearance (CL/F) and volume of distribution (Vz/
None had received prior checkpoint inhibitor treatment. Planned maximum F) were on average 0.58 L/day and 11 L, respectively. Plasma levels generally
dose of 10 mg/kg was reached (n=3) without DLT occurred. There was only decreased mono-exponentially with an average terminal elimination half time
one Grade 3 drug related Treatment Emergent Adverse Event (TEAE) oc- around 13 days after reaching the peak concentration post SC administration.
curred at 0.3 mg/kg dose level, which was immune related dermatitis and Exposures of KN035 increased approximately proportionally with dose.
resolved later. All other drug related TEAEs were either Grade 1 or 2, with the Trough concentrations were maintained above 15 mg/mL post administration
most common events as elevated ALT (5/17) and elevated AST (4/17). of 5 mg/kg Q2W. No apparent exposure-body weight relationship was ob-
Among all enrolled subjects, three subjects had confirmed PR, including one served. Conclusions: KN035 exhibits a favorable safety profile in patients
RCC subject at 2.5 mg/kg and one Intrahepatic cholangiocarcinoma subject with advanced malignancies and preliminary results demonstrate encouraging
at 5 mg/kg, and one cholangiocarcinoma subject at 10 mg/kg. Conclusions: anti-tumor activity. Based on PK data from the Q2W schedule, a fixed dose
KN035 exhibits a favorable safety profile and promising preliminary anti- with less frequent dosing schedule of every 3 or 4 weeks is presently being
tumor activity in patients with advanced malignancies. Clinical trial in- evaluated. Clinical trial information: NCT03248843.
formation: NCT03101488.

2610 Poster Session (Board #254), Sat, 8:00 AM-11:00 AM 2611 Poster Session (Board #255), Sat, 8:00 AM-11:00 AM
Immunological impact of canerpaturev (C-REV, formerly HF10), an oncolytic A phase 1/2a study of GEN-009, a neoantigen vaccine based on autologous
viral immunotherapy, with or without ipilimumab (Ipi) for advanced solid peptide immune responses. First Author: Roger B. Cohen, University of
tumor patients (pts). First Author: Takayuki Nakayama, Department of Ex- Pennsylvania Perelman School of Medicine, Philadelphia, PA
perimental Therapeutics, National Cancer Center Hospital, Tokyo, Japan
Background: Tumor-specific neoantigens provide individualized targets for
Background: C-REV, an oncolytic, spontaneous mutant of Herpes Simplex immunotherapy. In silico selection methods are sub-optimal at predicting
Virus type 1 (HSV-1), is a cancer immunotherapy agent that combine direct immunogenic targets, missing up to 70% of true neoantigens. ATLAS is a
tumor cell killing with immune modulation. A phase I study for solid tumors powerful tool that screens all candidate neoantigens for pre-existing patient-
with cutaneous and/or superficial lesions treated with C-REV monotherapy specific CD4 or CD8 responses in an HLA agnostic assessment. ATLAS also
and a phase II study for unresectable or metastatic melanoma treated with C- identifies inhibitory peptides that may suppress tumor immunity and ac-
REV and Ipi combination therapy were conducted. Immune status of cancer celerate tumor progression. The GEN-009 vaccine contains stimulatory but
pts before and after administration of C-REV with/without Ipi has been no inhibitory peptide antigens. Methods: GEN-009-101 is a first-in-human
unclear. Methods: A phase I study (n = 6) included solid tumor pts with phase 1/2a study testing platform feasibility, safety, immunogenicity and
cutaneous and/or superficial lesions treated with C-REV monotherapy (1 x 106 clinical activity in selected solid tumors. After next-generation tumor se-
and 1 x 107 TCID50/mL/dose; 4 injections q2-4wk). In phase II study (n = 28), quencing and cytokine-based ATLAS assessment using autologous T cells
C-REV (1 x 107 TCID50/mL/dose; 4 injections q1wk; then up to 15 injections and APCs, up to 20 stimulatory synthetic long peptides are used in each
q3wk) was injected into each tumor for advanced melanoma pts. Four Ipi personalized vaccine. GEN-009 is administered with poly-ICLC on weeks 0,
infusions (3 mg/kg) were administered at q3wk. Immune-monitoring was 3, 6, 12 and 24. Part A, a safety and immunogenicity pilot, has completed
conducted before and after treatment in tumor microenvironment usingpaired target enrollment of patients without evidence of disease to receive GEN-
biopsy samples by multiplex immunohistochemistry (mIHC) and in peripheral 009; Part B has 5 tumor-specific cohorts of up to 15 pts naı̈ve to PD-1
blood by flow cytometry (FCM). Results: In the phase I study, significant blockade who will receive GEN-009 with a SOC immunotherapy; Part C: up to
infiltrations of CD8+and CD4+ T cells were observed at tumor local site sta- 15 pts refractory to PD-1 inhibitors will receive GEN-009 monotherapy.
tistically in three pts (60%) among five pts. In the phase II study, FCM of Results: GEN-009 has been successfully generated for patients. Repeated
peripheral blood (n = 10) showed that the responders (irSD, n = 7, 70%) tend dosing has been well tolerated with mild local discomfort and no DLT. ATLAS
to express the higher levels of ICOS on CD4+ T cells as a pharmacodynamic screening results below show notable interpatient variability; one subject
biomarker of ipi monotherapy reported previously (Ng Tang D, et al. Cancer had only CD4 neoantigens, one had only CD8, another had a strong CD8 bias,
Immunol Res. 2013) and lower levels of PD-L1 on monocyte after two months and one patient had prominent inhibitory peptides. Conclusions: GEN-009
of treatment. Moreover, mIHC analysis of paired tumor biopsy samples (n = is a neoantigen vaccine that personalizes tumor specific targets and the
11) revealed that five pts (45%) among 11 pts were confirmed persistent individual patient’s capacity to respond. Immunogenicity data will assess
infection of C-REV at the injected site by qPCR. Disease control rate of pts with CD4 and CD8 T cell responses to each vaccine neoantigen. Clinical trial
the virus DNA detected on Days 85/169 was higher than that without it (100% information: NCT03633110.
[n = 5, irPR; 1, irSD; 4] vs. 33% [n = 6, irSD; 2, irPD; 4]). Furthermore, median Somatic
OS of pts with or without the DNA detected was 342 or 251 days respectively. mutations/ Stimulatory Inhibitory
Conclusions: Our results suggest C-REV injection in the tumor local site have Pt Tumor type Mb neoantigens antigens CD4 CD8

potential to enhance systemic immune response of Ipi. Clinical trial infor- 1 NSCLC 1.25 6 0 p
2 Bladder 3.15 16 4 p
mation: NCT03153085. 3 Melanoma 28.69 199 41 p p
4 Bladder 3.53 18 1 p
5 NSCLC 3.56 16 9 p

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Immunotherapy and Tumor Immunobiology 133s

2612 Poster Session (Board #256), Sat, 8:00 AM-11:00 AM 2613 Poster Session (Board #257), Sat, 8:00 AM-11:00 AM
A donor-dependent in vivo model for single agent and drug combination Interim analysis of a phase II study of nivolumab combined with ipilimumab
cytokine release syndrome safety evaluation. First Author: James G. Keck, in patients with pediatric solid tumors in adulthood (GETHI021). First
The Jackson Laboratory, Sacramento, CA Author: Xabier Mielgo, Hospital Universitario Fundación Alcorcón, Alcorcón,
Spain
Background: Although antibodies and CART cells therapies have been
successfully used for cancer therapy, they can have lethal adverse effects Background: Solid pediatric tumors that appear in adulthood are a het-
such as cytokine release syndrome (CRS). The animal models and in vitro erogeneous group characterized by a low incidence, lack of standard ther-
human PBMC assays presently in use can’t reliably predict the CRS in apeutic options and reduced survival. We have designed the first phase II
patients. A predictive marker for identifying patients at risk for developing clinical trial of nivolumab and ipilimumab in this setting, Here, we present
CRS upfront would improve the safety of immune-oncology drug develop- the results of the first cohort with 30 evaluable patients. Methods: This is a
ment. Methods: We have developed a rapid, sensitive and reproducible multicenter, open-label, single arm Phase II study conducted in 15 centers
in vivo humanized mouse model for quantitating CRS. The NSG mouse and of the Spanish Group for Rare Cancer (GETHI). We aimed to evaluate efficacy
its derivatives are engrafted with human PBMCs. On day 6 we induced and safety of the combination of nivolumab and ipilimumab in adult patients
cytokines release with pembrolizumab, avelumab, atezolizumab, ipilimu- ( 18 years) with locally advanced or metastatic childhood malignancies that
mab, anti-CD28, ATG and OKT3 in single dose; as well as combination have progressed or are not candidates to standard therapy. Treatment consisted
treatments involving pembrolizumab, lenalidomide, ATG and anti-CD28. on nivolumab 3 mg/kg IV q2w + ipilimumab 1 mg/kg IV q6w for 6 months or until
Furthermore, we compared our method versus the in vitro PBMC assay. The progression/unacceptable toxicity, for a maximum of 24 months. Primary
cytokine levels were also compared to the dose response. Results: There are endpoint was overall response rate (ORR) according to RECIST v1.1 criteria. We
about 10-15% CD45+ human cells on day 5 of engraftment; and among of used a Simon optimal two-stage design, with a first stage including first 30
them, there were approximately 70% CD3 T cells and 25% CD56 NK cells. evaluable patients. Results: 20 patients were male and median age was 43
All tested cytokines, human IFN-g, IL-2, IL-4, IL-6, IL-10 and TNF were (range 20-75). Most frequent histologies were medulloblastoma (4) neuro-
upregulated after 2 and 6 hours of OKT3, ATG, anti-CD28, pembrolizumab, blastoma (4) and Ewing family tumors (3). 90% had received prior systemic
avelumab and atezolizumab drug treatment. Mouse’s rectal temperatures therapy with 37% presenting progressive disease as best response. Median
dropped from 37-38 °C to about 36 °C at 6 hours’ time point in the treated previous treatment lines were 3 (range 1-9). 27 patients were PS0-1, and 3
groups. There is various cytokines release levels, low to high response in PS2. 6 patients have been treated for $6 months . Only one discontinued for
different donors with anti-CD28 treatment. All donors showed high response adverse events. With a median follow up of 4,3 months (range 0,4-11,3), 1
to OKT3. The cytokine release levels were consistent with a dose response or patient has achieved a deep partial response (PR) (3,6%), 10 stable disease
variable PBMC engraftment. The cytokine levels were also higher in some (SD) (35,7%) and 17 progressive disease (PD) (60,7%). 2 patients died before
drug combination studies such as pembrolizumab combined with lenali- radiologic evaluation. Clinical benefit rate (CR+PR+SD) was 39,3%. Median
domide or ATG; anti-CD28 combined with ATG. Our in vivo method was able progression free survival (PFS) was 1,8 months (95% CI 1,3-2,3), with a 3-
to determine CRS missed in the in vitro testing method. Conclusions: We months-PFS of 32,7% and 6-months-PFS of 20%. Median overall survival (OS)
have developed a rapid, sensitive and reproducible novel in vivo PBMC was 6,8 months (95% CI 3,3-10,2). 12 (40%) patients presented adverse
humanized mouse model that is able to differentiate human PBMC donors events (AE) of any grade and 6 (20%) experienced a grade AE deemed as
based on individual safety response to single agent and combination ther- possibly related to treatment. Conclusions: The combination of nivolumab and
apeutics of immune checkpoint inhibitors and possibly CAR-T therapy. This ipilimumab showed significant clinical benefit in this population with little
assay could be employed in future drug development. therapeutic options. One case of metastatic esthesioneuroblastoma, achieved a
dramatic tumor response and represents the first patient with this extremely rare
histology treated with immunotherapy. Clinical trial information: EudraCT
2016-003946-99.

2614 Poster Session (Board #258), Sat, 8:00 AM-11:00 AM 2615 Poster Session (Board #259), Sat, 8:00 AM-11:00 AM
Gut microbiota and clinical outcomes treated with nivolumab in Chinese Response to immune checkpoint inhibition and survival in BRCA-associated
non-small cell lung cancer. First Author: Shun Lu, Department of Oncology, recurrent ovarian cancer. First Author: Ying L Liu, Memorial Sloan-Kettering
Shanghai Lung Cancer Center, Shanghai Chest Hospital, Shanghai JiaoTong Cancer Center - Fellowship (GME Office), New York, NY
University, Shanghai, China
Background: Alterations in the DNA mismatch repair pathway increase
Background: Gut microbiome affecting responses to immune checkpoint susceptibility to immune checkpoint inhibition (ICI). Tumors with BRCA-
inhibitors (ICIs) against non-small cell lung cancer (NSCLC) has been in- related DNA repair defects may have increased antigenicity, which could
vestigated in western population. However, considering genetic variation, drive response to ICI. Responses to ICI in ovarian cancer (OC) have been
this phenomenon remains in vague in east-Asian NSCLC population. The modest. We seek to evaluate the association of BRCA mutations with re-
study is designed to explore the relationships between gut microbiome and sponse to ICI and survival in recurrent OC. Methods: A single-center, ret-
clinical outcomes treated with anti-PD-1 blockade in Chinese patients. rospective review identified 103 women with recurrent OC and known BRCA
Methods: 37 NSCLC patients received the treatment of Nivolumab were mutation status (90 germline and 33 somatic testing) who received ICI
enrolled in the study from the clinical trials CheckMate870 (NCT03195491). between 1/2013-7/2018 (98 on study). Clinical characteristics and duration
Fecal samples were collected at the starting point, every time point performing of ICI (Long . = 24 vs. Short , 24 weeks) were compared by BRCA status.
clinical evaluation and that with disease progression. 16s sequencing was Kaplan Meier survival analysis was used to calculate progression-free (PFS)
applied to assess the gut microbiota characteristics. Peripheral immune profiles and overall survival (OS) from start of ICI, and CoxPH models/logrank test were
were determined by multi-color flow cytometry in parallel. Results: When used to assess survival differences by BRCA status. Results: Deleterious
subgrouping patients into responders (R) and non-responders (NR) groups germline (g) or somatic (s) BRCA 1/2 mutations were present in 29 (28%)
according to the clinical response assessed by RECIST1.1, patients in R group women (12 gBRCA1, 9 gBRCA2, 3 sBRCA1, 5 sBRCA2, 1 gBRCA1/sBRCA1,
harbored higher diversity of gut microbiome at the starting point with consistent 3 gBRCA2/sBRCA2, and 1 gBRCA2/sBRCA1). Patients (pts) with BRCA
composition along the treatment. Analyzing progression-free survival (PFS) mutations had more lines of treatment prior to ICI (median 5 vs. 4, p = 0.03)
according to RECIST 1.1, patients with higher microbiome diversity had sig- and a longer time from diagnosis to ICI (median 54 vs. 38.5 months (mo), p =
nificantly prolonged PFS when compared to those with low diversity. Compo- 0.01), but there were no significant differences in other variables including
sitional difference was observed between two groups as well with the histology (86% high grade serous), stage at diagnosis (96% Stage III/IV), and
enrichment of Alistipes putredinis, Bifidobacterium logum, Prevotella corpri in platinum status (83% resistant), p . 0.05. Four pts (15%) with BRCA
R group whereas Ruminococcus_unclassified in NR group. Analysis of systemic mutations had long duration of ICI as compared with 20 pts (27%) in those
immune responses using multi-color flow cytometry revealed that patients with a without mutations, p = 0.20. Median PFS was 2.2 mo (95% CI 1.7-2.7) in
high abundance of microbiome diversity in the gut had more frequencies of those with BRCA mutations and 3.4 mo (95% CI 2.1-4.0) in those without
memory CD8+ T cell subset in the periphery in response to anti-PD-1 therapy. mutations, HR 1.22 (95% CI 0.78-1.91, p = 0.38). At a median follow-up of
Conclusions: Our results report the strong correlation between the gut micro- 23.3 mo, median OS was 21.3 mo (95% CI 13.7-31.8) in those with BRCA
biome diversity and the responses to anti–PD-1 immunotherapy in Chinese mutations and 19.8 mo (95% CI 13.8-25.3) in those without. This was not
NSCLC patients regardless of genetic variation between Western and Chinese significantly different, HR 1.00 (95% CI 0.54-1.87, p = 0.99), after ad-
population. Patients with a favorable gut microbiome (such as high diversity) justment for prior lines and time from diagnosis to ICI. Conclusions: In our
have enhanced immune responses mediated by effector T cell function in the study of heavily pretreated OC pts receiving ICI, BRCA 1/2 mutations were not
periphery. These findings thus provide important implications for the prediction associated with improved response or survival. These findings should be
and the evaluation of anti-PD-1 immunotherapy against NSCLC. validated in larger studies.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
134s Developmental Immunotherapy and Tumor Immunobiology

2616 Poster Session (Board #260), Sat, 8:00 AM-11:00 AM 2617 Poster Session (Board #261), Sat, 8:00 AM-11:00 AM
Interleukin-6 is potential target to de-couple checkpoint inhibitor-induced Frameshift mutations (Fsindel) complement tumor mutation burden (TMB)
colitis from antitumor immunity. First Author: Daniel H. Johnson, The University in predicting survival after immune checkpoint inhibitors (ICI) in a pancancer
of Texas MD Anderson Cancer Center, Houston, TX analysis. First Author: Vaia Florou, University of Miami Miller School of
Medicine, Sylvester Comprehensive Cancer Center, Miami, FL
Background: A deep understanding of the immunobiology of checkpoint
inhibitor (CPI) induced immune related toxicities, such as immune related Background: ICI benefit certain patients (pts) with various malignancies and
enterocolitis (irEC), and how these compare to the immune signatures in discovering biomarkers for response is an active research field. Recently,
tumors could lead to the development of strategies that de-couple auto- higher TMB (top 20% in each histology) based on nonsynonymous single
immunity from anti-tumor immunity. Methods: Total RNA from patient- nucleotide variants from MSK-IMPACT was shown to correlate with superior
matched irEC and normal colon FFPE tissue from patients [n = 12] survival in a pancancer cohort. Fsindels may generate more immunogenic
receiving CPIs were profiled with the 770 gene NanoString nCounter neoantigens and robust T cell infiltrates, thus predicting better responses to
PanCancer Immune Profiling Panel (NanoPCIP). The mean fold change in ICI. We previously demonstrated the clinical implication of fsindel in lung
gene expression from normal vs. irEC inflamed colonic tissue and baseline cancer pts on ICI. However, its value in other solid cancers has not been
vs. on-treatment tumor samples from patients responding or non-responding evaluated. Methods: Comprehensive genomic profiling (CGP) of the tumors
to ipilimumab based therapy were analyzed. C57BL/6 mice with B16.BL6 was performed by FoundationOne to derive fsindel and TMB as previously
melanoma tumors were treated with systemic anti-IL-6 + anti-CTLA-4 vs. described. Pts with advanced solid cancers who received ICI and had CGP
anti-CTLA4 alone vs. placebo and tumor size was measured. Results: In available were included. We categorized pts into two groups; 0 fsindel (FS-)
patients with irEC, the highest significantly upregulated differentially and more than 1 fsindel (FS+). Also, they were categorized into TMB high
expressed gene (DEG) in inflamed colon tissue encoded for IL-6 (Fold (top 20%) and TMB low (bottom 80%) within their own histology. Pro-
change +24.1). None of the significant and highest upregulated DEGs in the gression free survival (PFS) and overall survival (OS) were compared.
colitis, including IL-6, were significantly upregulated in responding tumors. Results: One hundred thirty-one pts excluding lung cancer were included.
Interestingly, IL-6 was also the highest upregulated DEG in non-responding There were 11 histology groups: 14 soft tissue sarcomas, 19 GU, 23 GI, 23
tumors numerically. When comparing mean fold changes across these an- skin, 10 HEENT, 10 RCC, 9 GYO, 6 pancreas, 5 mucosal melanoma, 4
alyses, the gene with the largest difference in upregulatation between colitis breast, and 8 others. 74 pts received pembrolizumab, 25 nivolumab, 29
and responding tumors was IL-6; the other highest upregulated genes in ipilimumab/nivolumab, and 3 atezolizumab. All pts had metastatic disease,
colitis encoded for neutrophil and monocyte chemotactic molecules. In our mean age was 61 years and 55 (42%) were women. Among the 131 pts, 74
mouse models, the addition of IL-6 blockade to anti-CTLA-4 therapy sig- were FS- and 57 FS+. The presence of fsindel (FS+) was significantly
nificantly improved tumor shrinkage compared to anti-CTLA-4 alone. correlated with overall response (p = 0.032) and clinical benefit rates (p =
Conclusions: Our data demonstrates that IL-6-mediated inflammation may 0.025). TMB-high did not show any significant difference in PFS (p = 0.1) or
be more prevalent in irEC and tumors not responding to CPIs than in tumors OS (p = 0.28) when compared to the TMB low. However, in a combined
responding, and blocking IL-6 enhances CPI anti-melanoma activity. Tar- model of TMB and fsindel, TMB high and FS+ patients had significantly
geting IL-6 may ameliorate irEC without hindering anti-tumor immunity. better PFS compared to patients who had either TMB high or FS+ or neither
(TMB low and FS-) (p = 0.021). Conclusions: Combined model of TMB high
and fsindel (+) correlated with superior PFS in advanced solid cancer pts on
ICI, in concordance with previous report for lung cancer. Validation in a larger
cohort is underway.

2618 Poster Session (Board #262), Sat, 8:00 AM-11:00 AM 2619 Poster Session (Board #263), Sat, 8:00 AM-11:00 AM
Pan-tumor prognostic value of multiple immune protein expressions. First Phase I trial of the combination of the heat shock protein-90 inhibitor
Author: Tiphaine Lambert, Drug Development Department (DITEP), Villejuif, onalespib (AT13387) and the cyclin-dependent kinase inhibitor AT7519M
France in patients with advanced solid tumors. First Author: Khanh Tu Do, Dana-
Farber Cancer Institute/Brigham and Women’s Hospital, Boston, MA
Background: Using multiple immune-checkpoint proteins (ICP) screening in
clinical routine could improve the evaluation of patients’ prognosis and Background: The 90kDa heat shock protein (HSP90) participates in the
ultimately tailor their treatment choice. We have evaluated this hypothesis in folding, stabilization, activation, and proteolytic turnover of aberrant proteins
the context of early drug clinical trials. Methods: Patients included in that contribute to the growth and survival of cancer cells. HSP90 inhibition
MOSCATO-02 trial had refractory cancers and were candidate for phase 1 leads to degradation of these aberrant proteins through the ubiquitin-
study. They were proposed to have a biopsy on an accessible tumor site for proteosome pathway, allowing for simultaneous targeting of multiple path-
the analysis of four proteins by immunohistochemistry (IHC) and RNAseq: ways. Inhibition of HSP90 alone stimulates a compensatory upregulation of
PD-L1, CD3, CD8 and FOXP3. Quantification of IHC staining was separated HSP70. The transcriptional induction of HSP70 has been linked to the
between intratumoral, intersitial and stromal by semi-quantitative method. activity of CDK9. Combined inhibition of onalespib-mediated HSP90 in-
Their relations to prognosis have been evaluated by survival Random Forest hibition and AT7519-mediated CDK9 inhibition has demonstrated syner-
and compared to classical prognosis clinical variables, such as age and RMH gistic anti-tumor activity in preclinical models at NOAEL doses, justifying
score (calculated by the number of metastatic sites, lactate dehydrogenase a Phase 1 study. Methods: We conducted an open-label phase 1 study
(LDH) and serum albumin). Results: From April 2016 to September 2017, following a 3+3 trial design. Patients received a 1-week lead-in of onalespib
228 patients included in MOSCATO-02 had a successful biopsy procedure alone (C0), followed by onalespib/AT7519M on days 1,4,8, and 11 of a 21-
with available IHC expression analysis. The main tumor subtypes were day cycle. Pharmacokinetic samples were obtained on C0D1 after onalespib
gastro-intestinal, urological, head and neck, breast and lung. RNAseq an- alone, and on C1D1 and D11 with the combination. HSP70 protein levels
alyzes were performed for two thirds of the patients (N=170). Median overall were analyzed in PBMC and plasma samples collected at baseline, after
survival was 8.1 months (CI95% 7.79 – 10, 65). We found that, in a cohort of onalespib alone, and after the combination, in order to demonstrate AT7519-
phase I patients, RMH score was the most important variable used to es- mediated suppression of HSP70 expression. Patients enrolled to the ex-
timate prognosis. Prognosis value of immune proteins were considerably pansion phase underwent optional paired tumor biopsies for assessment
inferior compared to clinical criteria. Among those proteins, the percentage of proof-of-mechanism demonstration of modulation of client proteins.
of PD-L1 low score (1+) and average staining intensity of CD3 were the most Results: Twenty-eight patients have been treated, 10 of whom were enrolled
valuables for prognosis evaluation. Variables with very few importance to to the expansion cohort with optional tumor biopsies. The MTD is DL2:
prognosis estimation were CD8 and FOXP3 IHC scores, biopsy site and onalespib 80 mg/m2 IV + AT7519M 21 mg/m2 IV on days 1,4,8, and 11 of a
cancer types, subsequent treatments by immunotherapies or targeted 21-day cycle. At DL3, DLTs included Grade 3 troponin elevation and
therapies. Conclusions: In this cohort of patients with refractory cancers, the mucositis. Drug-related adverse events occurring in $ 30% of patients in-
RMH score is confirmed as highly prognosis. Immune proteins could be used clude diarrhea, fatigue, mucositis, nausea, and vomiting, consistent with
as a support to guide patient’s selection but does not constitute effective known toxicities of these agents. Two patients with colorectal and endometrial
prognosis criteria. cancer, respectively, remained on study for 10 cycles with SD as the best
response. Modulation of HSP70 were demonstrated in patient plasma
samples. Conclusions: The combination of onalespib and AT7519 is toler-
able, although the doses of both agents were below the monotherapy MTDs.
Prolonged disease stabilizations were observed. Pharmacokinetic and
pharmacodynamic analyses are ongoing, including assessment of HSP70
expression in plasma and tumor. Clinical trial information: NCT02503709.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Immunotherapy and Tumor Immunobiology 135s

2620 Poster Session (Board #264), Sat, 8:00 AM-11:00 AM 2621 Poster Session (Board #265), Sat, 8:00 AM-11:00 AM
Non-small cell lung cancer (NSCLC) next generation sequencing (NGS) Overestimation of tumor mutational burden (TMB) using algorithms
using the Oncomine Comprehensive Assay (OCA) v3: Integrating expanded compared to germline subtraction. First Author: Kaushal Parikh, Mayo
genomic sequencing into the Canadian publicly funded health care model. Clinic, Rochester, MN
First Author: Kirstin Perdrizet, University of Toronto, Toronto, ON, Canada
Background: TMB is an emerging predictor of survival with immunotherapy.
Background: Standard of care (SOC) molecular diagnostics for stage IV TMB is determined by taking the difference between somatic and germline
NSCLC patients in Ontario, Canada includes publicly reimbursed EGFR/ datasets when tumor-normal pairs are available. In the case of commonly
ALK, and selected BRAF and ROS-1 testing. Other genomic alterations are utilized tumor-only sequencing, additional steps are needed to estimate the
not tested routinely; however, enhanced molecular testing may broaden somatic alterations. Computational tools have been developed that de-
treatment options for patients. This study evaluated costs, identified ac- termine germline contribution based on sample copy state, purity estimates
tionable targets, and determined clinical trial eligibility as a result of using and occurrence of the variant in population databases. Given the potential
the OCAv3 NGS in stage IV NSCLC patients. Methods: In a prospective study bias of population datasets, we hypothesized that tumor-only filtering ap-
of stage IV NSCLC out-patients at Princess Margaret Cancer Centre (Toronto) proaches may overestimate the actual TMB. Methods: We assessed the TMB
without EGFR/ALK/KRAS/BRAF alteration (unless failure of prior targeted from 50 tumors in 10 diseases including all missense, indels, and frameshift
therapy), diagnostic samples were tested by OCAv3 (ThermoFisher; 161 variants with an allelic fraction (AF) $5% and Coverage $100X within the
genes: hotspots, fusions, and copy number variations). Primary endpoints tumor. Tumor-only TMB was evaluated against the gold standard of matched
were incremental actionable targets and clinical trial opportunities as a germline subtracted TMB at three levels. Level 1 removed all the tumor-only
result of broader OCAv3 testing. Secondary endpoints include feasibility and variants with AF in the non-TCGA ExAC database $1%. Level 2 removed all
cost from the Canadian public healthcare perspective, and treatment out- variants observed in population databases simulating a naive approach of
comes. Results: Of 65 enrolled patients (Feb 2018-Jan 2019; 40 (62%) removing germline variation. Level 3 used an internal tumor-only pipeline for
completed/14 (21%) screen fail/ 11 (17%) pending), median age of calculating TMB. Results: There were significantly higher estimates of TMB
completed cohort was 65, 60% (N = 24) female, never/light smokers 68% with Level 1, Level 2 and Level 3 tumor-only filtering approaches than that
(N = 27), Asian 38% (N = 15), previously treated 33% (N = 13). Actionable determined by germline subtraction, resulting in significant bias. Whereas
targets beyond SOC were identified in 33% (N = 13): ERBB2 (N = 8), there was no correlation between TMB estimates and tumor-germline TMB for
BRAFV600 (N = 3), NRG fusion (N = 1), MET exon 14 (N = 1). New clinical Level 1 filtering, there were improvements in correlations for Level 2 and Level
trial options were identified in 70%. Failure of NGS was secondary to in- 3. Conclusions: The tumor-only approaches that filter variants in population
sufficient tissue [91% (N = 10) of screen failures; usually due to tissue databases overestimate TMB compared to that determined by germline
exhaustion from prior SOC molecular testing]. Incremental costs per case subtraction. Despite improved correlations with more stringent filtering ap-
beyond EGFR/ALK are estimated at $540 CAD. If ROS-1 and BRAF testing proaches, these falsely elevated estimates may result in the inappropriate
were publicly reimbursed at current rates, the incremental profiling cost with categorization of tumor specimens and negatively impact clinical trial results
OCAv3 would be $90 CAD per case. Conclusions: Although a key barrier to and patient outcomes.
implementation is lack of funding for NGS in the Canadian publicly funded
Level 1 Level 2 Level 3
system, the OCAv3 consolidates genomic testing, identifies additional ac-
tionable targets, and substantially increases clinical trial eligibility for pa- Difference above germline subtraction (mean, 42.2 4.67 1.62
SD) (613.4) (62.25) (61.69)
tients at a small incremental cost. Clinical trial information: NCT03558165. P value of paired t test p = 8.12e-28 p = 6.74e-15 p = 0.0003
95% limits of agreement (Bland-Altman) 16 to 68 0.26 to 9.07 -1.7 to 4.9
Correlation coefficient r= 0.09 r= 0.62 r= 0.69
Pearson correlation p value p = 0.52 p = 1.23e-06 p = 2.71e-08

2622 Poster Session (Board #266), Sat, 8:00 AM-11:00 AM 2623 Poster Session (Board #267), Sat, 8:00 AM-11:00 AM
Breast cancer with insertion or deletion exhibits the immunogenic phenotype. Baseline tumor-immune signatures associated with response to bempegaldesleukin
First Author: Yongmei Yin, The First Affiliated Hospital of Nanjing Medical (NKTR-214) and nivolumab. First Author: Michael E. Hurwitz, Yale School
University, Nanjing, China of Medicine, New Haven, CT
Background: Although immunotherapy has been proven to be effective in a Background: PIVOT-02 is an ongoing phase 1/2 study of bempegaldesleukin
wide range of malignant tumors, breast cancer remains to be one of poorly (NKTR-214), a CD122-preferential IL-2 pathway agonist, plus nivolumab in
immunogenic tumors and only a small subset of patients with breast cancer patients with advanced solid tumors. Bempegaldesleukin (NKTR-214) in-
achieved benefits from immunotherapy. Therefore, identification of better creases proliferative tumor infiltrating lymphocytes (TIL) and cell surface
predictive biomarkers to guide patient selection is highly desirable. It has PD-1 on immune cells and PD-L1 on tumor cells, demonstrating potential
been reported that insertion or deletion (indels) could create a novel open synergy with anti-PD-1 therapy. Pre-treatment tumor biopsies from meta-
reading frame and generate more neoantigen which may mediate response to static 1L melanoma (MEL) and urothelial carcinoma (UC) patients were
immunotherpay. However, the pattern of indels mutations in breast cancer is analyzed to correlate baseline immune phenotype to response. Methods: Pre-
still unclear. Methods: Whole-exome sequencing data, RNA-Seq data and treatment TIL (CD8+ T cells/mm2 and %CD3+ by IHC; 29 MEL; 22 UC) were
clinical data of 1096 breast tumors from The Cancer Genome Altas (TCGA) measured and divided into high and low groups based on median values. PD-
database were used to analyze the pattern of indels in breast cancer. Next L1 (% PD-L1 on tumor cells by IHC [28-8 PharmDx]; 33 MEL; 23 UC) was
generation sequencing (NGS) data of 81 metastatic breast tumors from scored negative (,1%) or positive ($1%). Interferon gamma gene score
clinical dataset were also used to validate the indels mutation pattern in (IFNG; 11 MEL) was scored as high or low based on median p value of ,0.1 for
different molecular subtype. Results: 81.7% (895/1096) of breast tumors in 15 genes (EdgeSeq). High and low groups were correlated with responses per
TCGA dataset harbored at least one indels mutation. Hormonal receptor (HR) RECIST 1.1. Results: Baseline demographics and prognostic factors were
negative tumors were associated with higher burden of indels mutations than balanced in the biomarker subgroups. Response rates for response evaluable
HR positive tumors in both TCGA dataset (P = 0.05) and NGS-clinical MEL and UC were 53% (SITC 2018) and 48% (ASCO-GU 2019), respectively.
dataset (P = 0.003). Indels were significantly correlated with higher TMB In MEL, median values of CD3-TIL and CD8-TIL were 19% and 203 cells/
and neoantigen level in TCGA cohort (P , 0.0001). In addition, tumors with mm2, respectively. Response rate correlations were 67% and 20% with IFNG
at least eight indels mutations (cut off as 80% percentile) exhibited even high and low, 79% and 29% with CD3-TIL high and low, 79% and 33% with
higher TMB (P , 0.0001) and neoantigen (P , 0.0001) level. Among 45 CD8-TIL high and low, and 68% and 43% with PD-L1 positive and negative.
immune related genes, the mRNA expression of 22 genes were significantly Most importantly, responses were observed in patients with the least favorable
higher in tumors with indels mutations, such as LAG3, IL18, IL6, CTLA4 and tumor microenvironment, characterized as both PD-L1 negative and TIL low,
PDCD1. Indels group also showed a high levels of genome instability in terms with responses of 17% (1/6 CD8-TIL), and 25% (2/8 CD3-TIL), respectively.
of HRD-LOH (P = 0.004), NtAI (P = 0.000), wGII (P = 0.001) and LST (P = Similar correlative trends were observed in UC, with 50% (4/8 CD8-TIL) and
0.014). Conclusions: Breast tumors with indels mutations exhibited the 38% (3/8 CD3-TIL) responses in patients with least favorable microenvi-
immunogenic phenotype. Further studies are warranted to investigate the ronment. Conclusions: The biomarker program included in PIVOT-02 iden-
potential value of indels as a predictive biomarker for immunotherapy in tified baseline immune signatures correlated with response for MEL and UC.
breast cancer. The response rates observed in both the favorable and unfavorable tumor
microenvironments indicate the potential of this combination and support its
broad development. Clinical trial information: NCT02983045.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
136s Developmental Immunotherapy and Tumor Immunobiology

2624 Poster Session (Board #268), Sat, 8:00 AM-11:00 AM 2625 Poster Session (Board #269), Sat, 8:00 AM-11:00 AM
TMB standardization by alignment to reference standards: Phase II of the Tissue immune response in epithelial ovarian carcinoma. First Author: Jon
Friends of Cancer Research TMB Harmonization Project. First Author: Henriksen, Department of Oncology, Vejle Hospital, Institute of Regional
Diana M. Merino, Friends of Cancer Research, Washington, DC Health Research, University of Southern Denmark, Vejle, Denmark
Background: Tumor mutational burden (TMB) is a predictive biomarker of Background: Epithelial ovarian cancer (EOC) is a highly malignant disease
response to immune checkpoint inhibitors across multiple cancers. In Phase 1 with a fatal outcome for most patients. During recent years immunological
of the Friends of Cancer Research Harmonization Project, we demonstrated a mechanisms have proven important in relation to the treatment and prog-
robust correlation between TMB estimated using targeted next-generation nosis of cancer, but within EOC the knowledge is still sparse. Understanding
sequencing (NGS) gene panels and whole exome sequencing (WES) applied to the importance of immune markers to the prognosis of ovarian cancer is
MC3-TCGA data. These findings demonstrated variability in TMB estimates essential for the future treatment of EOC. The aim of the present study was to
across different panels. Phase 2 evaluates sustainable TMB reference stan- investigate the prognostic impact of intratumoral PDL-1 expression, T cells,
dard materials for TMB alignment to assess this variability. The goal of this neutrophil granulocytes (NG) and Natural Killer (NK) cells in a population
effort is to establish best practices for estimating TMB in order to improve con- based cohort. Methods: All patients diagnosed with ovarian cancer in
sistency across panels, for the sake of optimizing clinical application and facili- Denmark in 2005 were included in the study. Immunohistochemical
tating integration of datasets generated from multiple assays. Methods: Fifteen staining was performed on tumor tissue from 412 patients. Antibodies for
laboratories with targeted panels at different stages of development participated. PD-L1, T cells (CD8), NG (CD66b), and NK cells (CD57) were used. Cell
We identified a set of reference standards consisting of 10 well-characterized densities were analyzed using a digital image analysis method. The primary
human-derived lung and breast tumor-normal matched cell lines. WES was endpoint was overall survival (OS). Results: In high grade serous carcinoma
performed using a uniform bioinformatics pipeline agreed upon by all team (HGSC) the median OS in patients with a high level of tumor infiltrating
members (WES-TMB). Each laboratory used their own sequencing and bio- T cells was 37 vs 25 months in patients with a low level(p = 0.0008).
informatics pipelines (tumor-only and tumor-normal) to estimate TMB Multivariate analysis showed a hazard ratio (HR) of 0.72 (p = 0.020). The
according to genes represented in their respective panels (panel-TMB). The median OS in patients with a high level of tumor infiltrating NK cells was 45
association between WES-TMB and each panel-TMB was investigated using vs 29 months in patients with a low level (p = 0.0310). Multivariate analysis
regression analyses. Bias (relative to WES-TMB) and variability in TMB es- showed a HR of 0.67 (p = 0.041). PD-L1 and NG had no statistically
timates across panels were rigorously assessed. All analyses were blinded. significant impact on OS. Only T cells showed prognostic significance across
Results: The set of reference standards spanned a clinically meaningful TMB histological subtypes with a HR of 0.72 (p = 0.007) in favor of a high density
range (4.3 to 31.4 mut/Mb). Preliminary data from 12 laboratories shows a of T cells. Conclusions: The present population based study demonstrated
good correlation between panel-TMB and WES-TMB in this empirical analysis. prognostic importance of tumor infiltrating T cells and NK cells in HGSC.
Across panels, regression R2 values range 0.77-0.96 with slopes ranging Neither PD-L1 nor NG held prognostic significance.
0.60-1.26. Calibration analyses that seek to minimize variability of TMB
estimates across panels using the established set of reference standards are
ongoing, as well as investigating cancer type dependence on the relationship
between panel-TMB vs. WES-TMB, which will be available at the time of
presentation. Conclusions: Preliminary findings demonstrate feasibility of
using sustainable reference control cell lines to standardize and align esti-
mation of TMB across different targeted NGS assays. Future studies aim to
validate reference standard material as a reliable alignment tool by using
formalin-fixed paraffin-embedded human tumor samples.

2626 Poster Session (Board #270), Sat, 8:00 AM-11:00 AM 2627 Poster Session (Board #271), Sat, 8:00 AM-11:00 AM
Tumor mutational burden (TMB) profile of K-RAS/TP-53 co-mutation in Impact of prior chemotherapy or radiation therapy on tumor mutation burden
metastatic non-small cell lung cancer (m-NSCLC). First Author: Abdul Rafeh in NSCLC. First Author: Sushma Jonna, Georgetown University Medical
Naqash, East Carolina University/Vidant Cancer Center, Greenville, NC Center, Washington, DC
Background: Early data suggests that co-occurring genetic events define Background: Higher non-synonymous tumor mutation burden (TMB) in non-
biological heterogeneity in K-RAS mutant NSCLC, with K-RAS/ TP-53 (KP) small cell lung cancer (NSCLC) is associated with a higher likelihood of
co-mutated subset having potential therapeutic vulnerabilities to immune response to checkpoint inhibitors. Tissue samples subject to TMB analysis
checkpoint blockade (ICB). To explore the immunological basis for these may be obtained after exposure to cytotoxic chemotherapy or radiation
findings, we evaluated the immune biomarker profile (TMB/PD-L1) in KP therapy – both of which introduce somatic mutations in DNA and can in-
mutant m-NSCLC using a large next-generation sequencing (NGS) dataset. fluence the number of identified mutations. The role of TMB as a potential
Methods: Caris life sciences NGS dataset consisting of 1317 m-NSCLC predictive marker for immunotherapy is evolving, and the impact of prior
tissue samples from 2016-18 was queried. PD-L1pos was defined as $ 1% therapy on TMB could influence interpretation. Methods: Eligible cases were
staining using 22c3 Dako assay. TMB was measured by counting all somatic from patients with confirmed NSCLC, available clinical annotation and
non-synonymous missense mutations using targeted NGS (592 genes). tumor molecular profiling including TMB analysis at a CLIA-certified ge-
TMB-high (H) was defined as $ 10 mutations/Megabase (mut/Mb). P-values nomics laboratory (Caris Life Sciences, Phoenix, AZ) using the Illumina
were calculated using Chi-square and Mann-Whitney test. Results: K-RAS NextSeq platform. TMB was calculated using only missense mutations that
mutations were identified in 28.7% (378/1317). Within this K-RAS mutant had not been previously reported as germline alterations. Treatment history
group, KP subset constituted 49.4% (187/378), remaining were K-RAS was obtained for each patient under an IRB approved protocol to determine
mutated/ TP-53 wild type (K-Pwt). 72.2 % (135/187) of KP had PD-L1pos whether patients had had received chemotherapy or radiation therapy in the
with 51.9% (97/187) having PD-L1 $ 50%. KP had higher median TMB vs. year prior to collection of the tissue subject to TMB analysis. Data analysis
K-Pwt (14.5 vs. 9.0 mut/Mb, p,0.001) and higher % of TMB-H vs. K-Pwt was performed using the chi-square test of deviance to evaluate whether
(79.9 vs. 45.1%, p,0.001; Table). Even in the PD-L1neg group, KP had TMB was statistically significantly different between groups, correcting for
higher % of TMB-H vs. K-Pwt (86.5 vs. 41.5%, p,0.001). K-RAS or TP-53 smoking status. Results: Out of 1,118 patients identified, 459 cases met all
exon-subtypes had no difference in median TMB or % of TMB-H. Across eligibility criteria and were evaluated. 76 patients (17%) received either
metastatic sites, brain tissue had the highest % of KP subset (38.3%, 68/ chemotherapy or radiation prior to tissue collection. Samples acquired prior to
187) followed by bone (28.9%, 54/187). Within KP subset, brain tissue had any therapy had a median TMB of 10 mut/Mb vs. 11 mut/Mb in samples
higher median TMB vs. bone (16 vs. 11 mut/Mb, p,0.01) as well as greater acquired after any therapy. After adjusting for smoking, there was no significant
% of TMB-H vs. bone (86.5 vs. 68.5%, p=0.01). Conclusions: This is the difference in TMB between these cohorts (p = 0.41). Secondary pair wise
largest dataset to date highlighting the unique immune profile of KP mutant analysis showed no statistically significant difference in TMB from chemo-
m-NSCLC. Our results show that KP subset has a significantly higher TMB therapy-naı̈ve and chemotherapy-treated samples (p = 0.28). The same was
than K-Pwt, especially in the PD-L1neg subgroup. Metastatic site-specific true for radiation (p = 0.75). Collection of clinical data is ongoing and further
variations in TMB were also observed for the KP subset. These findings could analysis, including additional cases will be presented. Conclusions: Though
have therapeutic implications in guiding patient selection for ICB and merit cytotoxic chemotherapy and radiation therapy can introduce somatic mutations,
prospective investigation. prior exposure to either was not associated with a significant difference in TMB.
Variable KP vs. K-Pwt p
TMB high (%) 79.9 vs. 45.1 ,0.001
Median TMB (mut/Mb) 14.5 vs. 9.0 ,0.001

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Immunotherapy and Tumor Immunobiology 137s

2628 Poster Session (Board #272), Sat, 8:00 AM-11:00 AM 2629 Poster Session (Board #273), Sat, 8:00 AM-11:00 AM
The consensus Immunoscore adapted to biopsies in patients with locally A multiplex immunofluorescence assay to assess immune checkpoint
advanced rectal cancer: Potential clinical significance for a “Watch and inhibitor-targeted CD8 activation and tumor colocalization in FFPE tissues.
Wait” strategy. First Author: Carine El Sissy, Hopital Européen Georges First Author: Tony Navas, Clinical Pharmacodynamics Biomarker Program,
Pompidou, Paris, France Applied/Developmental Directorate, Frederick National Laboratory for Cancer
Research, Frederick, MD
Background: We investigated whether an adaptation to rectal biopsies of the
recently validated consensus Immunoscore, could predict the response to Background: Immune checkpoint inhibitors promote antitumor immune
neoadjuvant treatment and delineate clinical responders that could benefit responses by enhancing T-cell activity. Measuring the pharmacodynamic
from a “Watch and Wait” (W&W) strategy with acceptable outcomes. effects of these drugs is challenging, as it requires assessing both immune
Methods: Initial biopsies from 273 patients with locally advanced rectal cell and cancer cell populations. To evaluate T cell activation in tumor tissue
cancer (LARC) treated by neoadjuvant chemoradiotherapy (nCRT) followed from patient biopsies, we developed a robust multiplexed immunofluores-
by Total Mesorectal Excision (TME), were immunostained for CD3+ and cence assay. Methods: Our assay uses novel oligo-conjugated antibodies
cytotoxic CD8+ T cells and quantified by digital pathology to determine the (Ultivue) for simultaneous quantitation of TCR activation (phospho-
Immunoscore within pre-treatment Biopsy (ISB). Expression level of 44 CD3zeta), immune checkpoint signaling via PD-1 (p-SHP1/p-SHP2), and
immune related genes post-neoadjuvant treatment was investigated by the net stimulation/inhibition resulting from the integration of these two
Nanostring technology (n = 64 patients). Results were correlated with re- pathways in CD8 cells (p-ZAP70), while also providing the proximity of CD8
sponse to neoadjuvant treatment, disease free survival (DFS) and time to cells to tumor tissues, identified by b-catenin. The method was clinically
recurrence (TTR). Prognostic performance of ISB was finally assessed in 73 validated using custom tissue microarrays (TMA) containing tumor biopsies
LARC treated by W&W strategy. Results: ISB Low, Intermediate and High of 3 different histologies (CRC, NSCLC, and breast). Results: From a total of
were respectively observed in 23.3, 50.4 and 26.3 % of the cohort. ISB was 192 tumor core biopsies, 20/64 NSCLC, 9/64 CRC, and 3/65 breast TMA cores
positively and significantly correlated with the response to nCRT, as eval- were found to have a significant number of CD8+ tumor infiltrating lymphocytes
uated by Dworak classification (P = .0034), ypTNM (P = .0003), down- (TILs) at baseline ( . 50 cells in the examined section). In 18 of the 20 NSCLC
staging (P = .0014), and neoadjuvant rectal (NAR) score, (P , .0001). ISB cores, $50% of CD8 cells both inside and outside of the tumor were activated
status was also positively associated with the degree of local immune ac- (CD3z-pY142+). In 6/9 CRC cores, $50% of CD8+ cells inside tumor tissues
tivation post-neoadjuvant treatment. ISB High patients were at low risk of were activated, and in 4/9 CRC cores, $50% of CD8+ cells in stroma were
relapse, with 5-year DFS rates of 81.1 % (CI, 71.3-92.1 %) as compared to activated. In 2/3 breast tumor cores, 90% of CD8+ cells inside tumor tissues
57.8 % (CI, 45.9-72.9 %) in ISB low patients. In multivariate analysis, ISB were activated; in the remaining core, 90% of CD8+ cells in stroma were
was the only significant parameter at presentation associated with DFS (High activated. Interestingly, all 192 cores had minimal to no expression of activated
vs Low: P = .001). Among W&W patients, significant difference was observed Zap70 (pY493) in CD8+ cells. Conclusions: Depending on tumor histology,
for TTR according to ISB status (High vs Low: P = .025). Conclusions: ISB baseline biopsy samples may contain variable numbers of activated CD8+ TILs
could provide a reliable estimate of the response to nCRT and risk of re- (CD3z-pY142+), which may reside inside or outside of tumor regions and
currence in LARC patients’ treated by TME or W&W strategy. express very low levels of Zap70-pY493. Anti-PD-1 therapy is predicted to
enhance T-cell cytotoxic activity, as demonstrated by an increased number of
TILs and elevated Zap70-pY493 expression. This assay is being used for
pharmacodynamic evaluations in ongoing immunotherapy clinical trials.
Funded by NCI Contract No HHSN261200800001E.

2630 Poster Session (Board #274), Sat, 8:00 AM-11:00 AM 2631 Poster Session (Board #275), Sat, 8:00 AM-11:00 AM
Tumor mutational burden (TMB) and PD-L1 expression as predictors of Quantitative MHC II protein expression levels in tumor epithelium to predict
response to immunotherapy (IO) in NSCLC. First Author: Emily Castellanos, response to the PD1 inhibitor pembrolizumab in the I-SPY 2 Trial. First
Flatiron Health, New York, NY Author: Julia Dianne Wulfkuhle, George Mason Univ, Columbia, MD
Background: PD-L1 expression and TMB, as a proxy for neoantigen burden, Background: Response to immune checkpoint inhibitors has been associ-
have been correlated with response to IO in advanced NSCLC (aNSCLC) ated with immune activation and mutational load within a tumor. Previous
clinical trials, but their combined utility is unclear. We assessed TMB and results in other tumors have implicated MHC II protein tumor cell expression
PD-L1 as predictors of response in aNSCLC patients (pts) after IO mono- as a response predictor to immune checkpoint inhibitors. In the I-SPY 2
therapy in a real-world setting. Methods: Pts had a diagnosis of aNSCLC, TRIAL, the anti-PD1 therapeutic antibody pembrolizumab (P) was available
comprehensive genomic profiling of 186-315 genes/1.1 megabase (Mb), to HER2-negative subtypes and graduated in both the HR+/HER2- and
PD-L1 testing of pre-IO specimens, and were treated in the Flatiron Health TNBC signatures. Pre-specified biomarker analysis was performed to test
network (1/2011 - 6/2018). Clinical characteristics and real-world tumor tumor MHC II expression as a predictor of response to P in the I-SPY 2 TRIAL
response (rwTR) were obtained via technology-enabled abstraction of cli- based on its central role in tumor antigen presentation. Methods: 156 pa-
nician notes and radiology/pathology reports, and linked to genomic data in tients (P: 67, controls: 89) had RPPA and pCR data. RPPA-based quanti-
the Flatiron Health-Foundation Medicine Clinico-Genomic Database. A tative data for pan-MHC II protein isotypes HLA-DR/DP/DQ/DX and HLA-DR
general additive model examined the predictive value of TMB (as continuous protein isotype was obtained from LCM-enriched tumor epithelium, and
measure) and PD-L1 level on rwTR. A reduced PD-L1-only model was protein levels were assessed for association with pCR in the P and control
compared to the full model using Akaike Information Criterion (AIC). rwTR arms separately using the Wilcoxon Rank Sum test (p , 0.05). Analysis was
predictions at representative TMB and PD-L1 levels were calculated. also performed in the HR+ and HR- subgroups. Markers were analyzed
Results: Of 426 pts, PD-L1 expression was high ($50%) in 140, low (1- individually; p-values are descriptive and were not corrected for multiple
49%) in 123, and negative (,1%) in 163. Median TMB was 9.6 mut/Mb comparisons. Results: Across all P- treated patients, the HLA class II
(IQR 4.4 - 14.8) overall, 11.3 in responders and 8.7 in non-responders. TMB molecules –DR and -DR/DP/DQ/DX had a positive association with response
did not correlate with PD-L1 level (Kruskal-Wallis p=0.29). The TMB + PD- to P (p = 0.014 and p = 0.001). Expression of HLA-DR/DP/DQ/DX also had a
L1 model had superior prediction of rwTR than the PD-L1 model, as assessed positive association with response to P in HR+ tumors. Neither of these
by lower AIC score. In the combined model, higher TMB and PD-L1 levels associations were seen in the control arm samples. Conclusions: The ob-
were each associated with higher rwTR likelihood (Table). Predicted rwTR servation of elevation of MHC II protein expression in HER2- responding
probability, % (95% CI), by TMB and PD-L1 in line 1. Conclusions: TMB and patients treated with P suggests that activation of antigen peptide exchange
PD-L1 expression are independent markers that, when combined, have facilitated by these molecules in T and B cells may enhance response to P
increased predictive power for response to IO. High TMB + low/neg PD-L1 treatment.
behaved similarly to low TMB + high PD-L1, and high TMB + high PD-L1
predicted the highest rwTR. Investigation of these biomarkers as comple-
mentary predictors of progression and overall survival is ongoing.
PD-L1 level Negative Low High
TMB (mut/Mb)
1 21 (11 - 38) 13 (4 - 39) 43 (29 - 59)
5 25 (15 - 39) 22 (10 - 42) 47 (35 - 60)
10 30 (19 - 44) 39 (24 - 56) 52 (41 - 62)
15 36 (23 - 51) 53 (36 - 70) 56 (44 - 67)
20 41 (25 - 59) 58 (38 - 77) 60 (46 - 73)
30 51 (28 - 73) 51 (21 - 81) 69 (46 - 85)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
138s Developmental Immunotherapy and Tumor Immunobiology

2632 Poster Session (Board #276), Sat, 8:00 AM-11:00 AM 2633 Poster Session (Board #277), Sat, 8:00 AM-11:00 AM
Evaluation of TMB estimates for the prediction of response to immune Association of genetic variations within the T-cell costimulatory LIGHT gene
checkpoint blockage. First Author: Jan Budczies, Institute of Pathology, with outcome in stage II and III colon cancer. First Author: Martin D. Berger,
University of Heidelberg, Heidelberg, Germany Division of Medical Oncology, USC Norris Comprehensive Cancer Center,
Keck School of Medicine, Los Angeles, CA
Background: Tumor mutational burden (TMB) is an emerging biomarker in
immuno-oncology (IO). Panel sequencing (PS) is a promising approach for its Background: T-cell activation plays a key role in maintaining an effective host
implementation in clinical practice. Methods: We analyzed TMB as predictor immunity and antitumor control. Targeting costimulatory immune checkpoint
of IO response in a multi-cancer cohort published by Miao et. al. The per- proteins can lead to increased antitumor immunity. We hypothesize, that
formance of three large panels (Illumina TSO500, Qiagen TMB [QIAseq] and variations in genes encoding for T-cell activation molecules may predict
Oncomine TMB [OTMB]) and two small panels (Illumina TST170 and outcome in stage II and III colon cancer patients. Methods: The impact of 4
Oncomine Comprehensive Assay [OCAv3]) was compared to WES by in silico functional single nucleotide polymorphisms (SNPs) within the LIGHT, ICOS,
simulations. Separation of responders (PR/CR) from non-responders (PD) CD80 and GITR genes on time to recurrence and overall survival was evaluated
was analyzed in the multi-cancer cohort (n = 193), in the lung cancer (n = in 209 patients with stage II and III colon cancer. Genomic DNA was extracted
36) and in the melanoma (n = 125) subcohort. We also simulated PS in the from formalin-fixed paraffin embedded tissue and the SNPs were analyzed by
TCGA pan-cancer cohort. Results: In lung cancer, TMB was strongly pre- PCR-based direct sequencing. Results: Baseline characteristics were as fol-
dictive for IO response (area under ROC curve [AUC] 0.78-0.94). WES lows: median age = 70y (19-91); female/male ratio = 41.6% / 58.4%; 111
performed (borderline)-significantly better than PS for all five panels patients had stage II, and 98 stage III colon cancer. The LIGHT rs3760746
(OCAv3: p = 0.011, TST170: p = 0.01, QIAseq: p = 0.048, OTMB: p = SNP showed significant association with recurrence rate in the overall pop-
0.063, TSO500: p = 0.11). For the cut-point of 199 mutations, mis- ulation. Patients harboring any G allele had a lower 3-years recurrence rate
classification rates compared to WES (16.7%) were borderline-significantly compared to those with a A/A genotype (16% vs 30%) in both univariate (HR
higher for the small panels OCAv3 (33.3%, p = 0.087) and TST170 (36.1%, 0.53, 95% Confidence intervall (CI) 0.29-0.96, p = 0.033) and multivariate
p = 0.054), but not for the large panels. In melanoma, TMB was moderately analyses (HR 0.52, 95% CI 0.29-0.95, p = 0.034). This trend was most
predictive (AUC 0.58-0.63) and WES performed (borderline)-significantly evident among patients with stage III colon cancer. Here again, G allele
better than the OCAv3, TST170, QIAseq and TSO500 panels. In the multi- carriers had both a lower 3-years recurrence and a longer 5-years overall
cancer cohort, WES did not perform better than PS. TBM estimates from PS survival rate compared to those having a A/A genotype (21% vs 40% and 77%
include an inherent fuzziness originating from restriction to a limited part of vs 43%) in both univariate (HR 0.42, 95% CI 0.19-0.90, p = 0.021 and HR
the coding sequence. Based on a random mutation model, we derived a 0.51, 95% CI 0.25-1.03, p = 0.046) and multivariate analyses (HR 0.43,
mathematical formula for the coefficient of variation (CV) of TMB: The CV 95% CI 0.20-0.93, p = 0.033 and HR 0.30, 95% CI 0.14-0.65, p = 0.002).
decreases inversely proportional with both the square root of the TMB level Conclusions: Our results provide the first evidence that polymorphisms within
and with the square root of the panel size. We showed that the mathematical the T-cell costimulatory LIGHT gene might serve as prognostic markers in
law is valid for real-word mutation data. Conclusions: Small panels (size , 1 patients with stage II and III colon cancer. Targeting LIGHT might be a
Mpb) performed imprecise in diagnostic TMB estimation. Even using the promising approach to further optimize treatment options and to improve
largest commercially available panels it can be challenging to capture the outcome of colon cancer patients in the adjuvant setting.
full predictive information of TMB. The detrimental effect of small panel size
can be addressed by using larger panels, but halving the CV of TMB ne-
cessitates quadruplication of the panel size.

2634 Poster Session (Board #278), Sat, 8:00 AM-11:00 AM 2635 Poster Session (Board #279), Sat, 8:00 AM-11:00 AM
Comparison between whole exome sequencing (WES) and single nucleotide Electrostatic human leukocyte antigen-neoantigen interactions and durable
polymorphism (SNP)-based tumor mutation burden analysis. First Author: benefit in non-small cell lung cancer patients treated with immunotherapy.
Kirsten Timms, Myriad Genetics, Inc., Salt Lake City, UT First Author: Amy Lauren Cummings, David Geffen School of Medicine at
UCLA, Los Angeles, CA
Background: Immune checkpoint inhibitors (ICI) block proteins which en-
able cancer cells to evade the immune system. Recent studies have shown Background: Human leukocyte antigen (HLA) binding relies on energy from
that the higher the tumor mutation burden (TMB) the greater the likelihood of the interaction of B-pocket residues with anchor amino acids (AA). Among
response to ICI therapy. Analysis of TMB has focused on WES of paired tumor HLA class I supertypes, only HLA-B has distinct electrostatic B-pocket
and normal samples. This study tests the feasibility of measuring TMB from a specificities, and of 7 HLA-B supertypes, B08, B27, and B44 feature
SNP-based resequencing assay (myChoice HRD Plus). Methods: WES and binding pockets with preferences for charged AAs (Lund Immunogen).
myChoice HRD Plus were performed on matched tumor and normal DNA Whether electrostatic interactions in HLA-neoepitope binding would identify
from 44 breast and 12 colon tumors. myChoice HRD Plus combines ho- superior neoantigens and associate with survival in NSCLC patients treated
mologous recombination deficiency analysis with resequencing of 108 with immunotherapy was unknown. Methods: Forty patients with advanced
genes and microsatellite instability analysis. WES-based TMB was calcu- NSCLC treated with single agent pembrolizumab on a clinical trial with at
lated by identifying all variants in paired samples, and subtracting germline least 5 years follow-up underwent paired tumor-normal whole-exome se-
variants. Two SNP-based TMB (SbTMB) methods were utilized to calculate quencing (WES) with Illumina HiSeq 2000/3000. HLA typing used normal
TMB. The first used germline subtraction similar to the WES-based method. (germline) WES from peripheral blood mononuclear cells analyzed with
The second utilized an algorithm which removed background germline BWA-ALN and Athlates software (Liu Nuc Acids Res); supertype was de-
variants. Median sequence length to calculate TMB was 9.7 Mb for WES, 4.6 termined by 2008 criteria (Sidney BMC Immunol). Tumor nonsynonymous
Mb for SbTMB (germline subtraction), and 1.9 Mb for SbTMB (algorithm). coding mutations were identified with GATK v3.8, annotated with Ensembl-
Results: Correlation coefficients for WES vs. SbTMB (germline subtraction) VEP, and passed through pVAC-Seq using a NetMHC 4.0 algorithm to
and SbTMB (algorithm) were 0.895 and 0.908, respectively. The two identify potential neoepitopes 9 AAs in length (Hundal Genome Med).
SbTMB methods had a correlation coefficient of 0.834. SbTMB measures of Neoepitopes were characterized based on mutant AA charge (D/E negative,
TMB were generally higher than WES-based TMB with a mean increase in H/K/R positive) and position. High affinity neoepitopes (HAN) were defined
score of 1.6 variants/Mb for SbTMB (germline subtraction; p = 4.6x10-6) and as those an with IC50 , 50 nM with wildtype IC50 . 50 nM (Ghorani Annals
1.5 variants/MB for SbTMB (algorithm; p = 1.2x10-5). No significant dif- Oncol) and a mutation to a known B-pocket supermotif (K in position 3 or 5
ference in magnitude of TMB score between the SbTMB measures was for B08, R in position 2 for B27, E in position 2 for B44) (Lund Immunogen).
observed (0.04 variants/Mb; p = 0.88). Conclusions: SNP-based methods Progression-free survival (PFS) was compared with logrank tests and pro-
for calculating TMB produced highly concordant scores compared to WES- portional hazards (JMPv14, Cary, NC). Results: Of the 40 patients, 29
based methods. SbTMB assays produced elevated TMB scores, consistent (72.5%) had at least one B08, B27, or B44 allele. One or more supertype-
with selective pressure against mutations in coding regions of genes, matched HAN were found in 10 of the 29 (34.5%), including 6/7 with PFS .
necessitating a higher score threshold for when using a SbTMB assay. This 2 years, 3 of whom continue on therapy beyond 5 years. Median PFS in those
SbTMB analysis expands the utility of myChoice HRD Plus, and provides a with HAN was 26.7 months (m) vs 4.3 m in those without (HR 0.34, 95% CI
method for calculation of TMB without sequencing a germline comparator. 0.11-0.88, p = 0.024). Conclusions: Electrostatic charge may serve as a
mechanism for enhanced binding affinity in HLA-B supertypes with a
preference for charged AA in their B-pockets. Identification of favorable
HLA-matched neoepitopes may identify distinct prognostic groups and
potentially durable responders to immunotherapy in NSCLC.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Immunotherapy and Tumor Immunobiology 139s

2636 Poster Session (Board #280), Sat, 8:00 AM-11:00 AM 2637 Poster Session (Board #281), Sat, 8:00 AM-11:00 AM
Effect of subcutaneous multipeptide active antigen-specific immunotherapy Survival by stage and tumor measurability in metastatic melanoma patients
at lymph nodes and tumor sites on clinical outcomes in progressive tumors. treated with autologous dendritic cell tumor cell vaccines. First Author:
First Author: Juan Pablo Marquez-Manriquez, CICS USA, Seattle, WA Robert O. Dillman, AIVITA Biomedical, Irvine, CA
Background: The tumor and lymph nodes (LN) microenvironment clinically Background: Survival of cancer patients is greatly affected by stage and
speaking are mainly unexplored and there are limited clinical studies of tumor burden. The purpose of this study was to determine survival for
immunotherapeutic approaches manipulating those in progressive cancer melanoma patients who were treated with patient specific vaccines in the
patients. The tumor and LN microenvironment offers an opportunity to treat context of prospective clinical trials, by cohorts defined by stage and tumor
locally with multi-peptide immunotherapy in challenging clinical situations measurability. Methods: Metastatic melanoma patients were treated with
where we can have still a potential safe and effective therapeutic approach autologous dendritic cells loaded with antigens from irradiated cells from
using the concept treat locally treat systemically. Methods: N = 11 patients short-term autologous tumor cell lines (DCV). All patients had a metastatic
were enrolled after the local IRB ethic committee approved the pilot clinical melanoma lesion surgically resected, from which a tumor cell line was
study number CICS/SS/0035. Previous identified targets such as FAP, b-2- established. Irradiated tumor cells (ITC) were incubated with autologous
microglobulin, Fascin, RCAS1/EBAG9, Bcl-2, survivin, Sox2, Ape-1, Valosin dendritic cells (DC) to produce the DCV, which were injected s.c. in 500
containing-protein (VCP) and EGFR were analyzed by IHC, Th1 and CD8 micrograms GM-CSF weekly x 3 weeks, then monthly for 5 months. Data was
long-peptides were predicted, immune assays using PBMCs including tumor pooled for DCV-treated patients enrolled in either of two phase II trials: one
microenvironment were performed to detect Granzyme B by ELISPOT, single-arm (NCT00948480), one randomized (NCT00436930). Patients
naturally processed epitopes by T-cell expansion and cytokines. Patients were assigned to one of three cohorts based on their most advanced stage of
were treated subcutaneously (S.C.) eight times in the axillary and inguinal disease prior to treatment, and whether they had measurable disease at the
LN area one week apart. Afterwards we treated S.C. as well but now in the time of treatment. Survival was determined per Kaplan and Meier.
areas with tumor activity according with the CT or PET every week 10 times. Results: The final therapeutic products consisted of autologous DC with non-
Initial DTH and at the end of the treatment was performed. Pathological, phagocytosed ITC making up 0% to 20% of cells in the final product. There
clinical and immunological correlations were made using multivariate were 45 men and 27 women. Median age was 52 years (range 17 to 83).
analysis. Results: Despite progressive disease 100% of the patients Tumor sources were 37-lymph node, 20-viscera, and 15-soft tissue. No
responded to the treatment. 80% had CR and 20% pseudo-progression and patients were lost to follow up; all surviving patients were followed 5 years.
then CR. 100% of the patients increased the levels of Granzyme B against Toxicity was minimal. Median overall survival (OS) for all 72 patients was
Bcl-2 (p = 0.001), VCP (p = 0.0001), Ape-1 (p = 0.005) and RCAS1 (P = 49.4 mos; 5-year OS 46%. There was no correlation between survival and the
0.0001) and this correlated with the scans post-treatment. The patients number of DC or ITC in the first three injections. Patients with recurrent stage
showed more CD8 infiltration in the DTH test at the tumor site (p = 0.002) 3 disease that had not recurred (n=18) had a 72% 5-year OS; patients with
than in the non-tumor site (p = 0.01). The number of metastatic lesions in non-measurable stage 4 (n=30) had a 53% 5-year OS. Patients with
lungs (p = 0.004) and hepatic tissue (p = 0.001) disappeared and correlated measurable stage 4 (n=18) had received an average of four prior therapies.
with increased levels of IL-12 production in vivo. Conclusions: Treating They had a median OS of 18.5 months, and 2-year OS of 46%.
immunologically the LN and the microenvironment of the areas with tumor Conclusions: This patient-specific DCV was associated with encouraging
activity is a feasible and safe approach to limit systemic toxicity and improve survival in all three clinical subsets. Because of its mechanism of action and
the clinical outcomes in patients with progressive cancer including bone absence of toxicity, it should be evaluated further. Clinical trial information:
sarcoma, epithelial ovarian cancer, PDAC, pediatric sarcomas and TNBC. NCT00948480, NCT00436930.

2639 Poster Session (Board #283), Sat, 8:00 AM-11:00 AM 2640 Poster Session (Board #284), Sat, 8:00 AM-11:00 AM
Preliminary results of a phase I clinical trial using an autologous dendritic Phase I trial of a modified vaccinia ankara (MVA) priming vaccine followed
cell cancer vaccine targeting HER2 in patients with metastatic cancer or by a fowlpox virus (FPV) boosting vaccine modified to express brachyury and
operated high-risk bladder cancer (NCT01730118). First Author: Hoyoung M. costimulatory molecules in advanced solid tumors. First Author: Julie Marie
Maeng, National Cancer Institute, Bethesda, MD Collins, National Cancer Institute, Bethesda, MD
Background: We developed a HER2 targeting autologous dendritic cell (DC) Background: Brachyury, a transcription factor, plays an integral role in
vaccine transduced with an adenovirus expressing the extracellular and epithelial-to-mesenchymal transition, metastasis, poor prognosis, and re-
transmembrane domains of HER2 (AdHER2). In mice, the homologous vac- sistance to chemotherapy. It is expressed in many tumor types, and rare in
cine cured virtually all mice with established or metastatic tumors. Protection normal tissue, making it an ideal immunologic target. BN-Brachyury com-
was dependent on antibodies against HER2 that inhibited phosphorylation, prises heterologous vaccination with recombinant MVA priming followed by
but was ADCC independent. We translated these findings into a clinical trial. FPV boosting, each encoding transgenes for brachyury and three cos-
Methods: This is an open-label, phase I study in patients with 1) metastatic timulatory molecules (B7-1, ICAM1, and LFA-3). Heterologous prime boost
cancer that progressed after $ 1 standard therapies, or 2) history of high risk approach is intended to optimize immunogenicity, as previously observed.
bladder cancer with definitive treatment, whose tumor is HER2 immunohis- Methods: Pts with metastatic solid tumors were treated with 2 monthly doses
tochemistry (IHC) score $ 1+ or FISH HER2/CEP17 ratio $ 1.8. Part 1 of the of MVA-brachyury SC at the previously tested dose, 2.2 x 109 infectious units
study enrolled patients naı̈ve to HER2-directed therapies and Part 2 enrolled (IU), followed by FPV-brachyury SC, 1 x 109 IU, for 6 monthly doses and then
patients who progressed with $ 1 anti-HER2 therapy. Results: In Part 1, the every 3 months for up to 2 years. The primary objective was to determine
lowest dose level (5E+6 viable DCs, N=7, 2 inevaluable) showed no benefit. At safety and tolerability and establish the RP2D. Immune assays were con-
the second and third dose level (10E+6 and 20E+6; N=7 and N=4; 0 and 1 ducted to evaluate immunogenicity. Results: In 10 pts (3 chordoma, 6 GI, 1
inevaluable in each), 1 CR (ovarian), 1 PR (stomach), and 3 SD (1 ovarian papillary thyroid), no dose-limiting toxicities or serious treatment-related
carcinosarcoma and 2 colon) were observed. Two bladder cancer patients who adverse events (TRAEs) were observed. The only Grade 3 TRAE was sedation
received vaccine as an adjuvant did not recur for +24 and +36 month each. In associated with fever, which resolved spontaneously and did not recur with
Part 2 (N=6, 2 inevaluable), 1 male breast cancer patient showed SD. Response subsequent cycles. All other TRAEs were Grade 1 or 2; the most common was
assessed by Modified Immune Related Response Criteria is summarized in the
injection-site reaction in all patients. Five pts have had stable disease for . 24
Table. Injection-site reactions occurred in all patients and were self-limited.
wks (per RECIST v1.1) and remain on treatment. One pt with chordoma, for
Echo, EKG and troponin follow up to 2 years showed no cardiac toxicity. Dose-
which BN-Brachyury was granted orphan drug designation, has had a 13.2%
expansion cohort (40E+6) is enrolling. Conclusions: We have translated a cancer
reduction in tumor size. As previously demonstrated, brachyury-specific T cell
vaccine from mice to a clinical trial. Preliminary results of a phase I trial of an
autologous AdHER2 DC vaccine show potential clinical benefit in select patients
responses were observed, as were responses against cascade antigens (non-
with HER2 expressing tumors with no cardiac toxicity. Clinical trial information: encoded antigens) CEA and MUC-1. Conclusions: Heterologous MVA- and
NCT01730118. FPV-brachyury is well tolerated and induced immune responses to brachyury
and cascade antigens, suggesting induction of immunologically relevant tu-
Part I
Part II
mor cell destruction. These data have informed combining BN-Brachyury with
10E+6
20E+6 20E+6
checkpoint inhibition (NCT03493945) and radiation (NCT03595228) to
Primary/HER2 IHC Response Duration (wk) Primary/HER2 IHC Response Duration (wk) Primary/HER2 IHC Response Duration (wk)
evaluate potential for synergetic activity in selected populations. Clinical trial
Colon/3 PD - Colon/1 PD - Breast/2 SD 24
Stomach/3 PR(-50%)* 16 Ovary/3 CR* 89** Breast/3 PD - information: NCT03349983.
Colon/2 SD 16 Ovary/1 SD(-20%) 48 Breast/3 PD -
Colon/1 SD 16 Breast/3 PD -
Ovary/2 PD -
Colon/3 PD -
Ovary/2 PD -

* Peptide array showed antibody response . x3 against HER2 peptides in select responders. ** Recurred with HER2 IHC 0 in a metastatic
lesion suggesting immune escape.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
140s Developmental Immunotherapy and Tumor Immunobiology

2641 Poster Session (Board #285), Sat, 8:00 AM-11:00 AM 2642 Poster Session (Board #286), Sat, 8:00 AM-11:00 AM
A phase I study (E011-MEL) of a TriMix-based mRNA immunotherapy (ECI-006) Final results of a phase I study evaluating INVAC-1, a novel DNA vaccine
in resected melanoma patients: Analysis of safety and immunogenicity. expressing an inactive form of human telomerase reverse transcriptase
First Author: ANA Maria Arance Fernandez, Department of Oncology and (hTERT) in patients with advanced solid tumors. First Author: Luis Teixeira,
Haematology, Papa Giovanni XXIII Cancer Center Hospital, Barcelona, Sénopôle Saint Louis, Service d’Oncologie Médicale, Hôpital Saint-Louis,
Spain APHP, Paris Diderot University, Paris, France
Background: ECI-006 is a combination of TriMix (mRNAs encoding for Background: INVAC-1 is an optimized DNA plasmid encoding an inactive
dendritic cell [DC] activating molecules [CD40L, CD70 and caTLR4]), and form of human Telomerase Reverse Transcriptase (hTERT), a universal
mRNAs encoding for melanoma-specific tumor-associated antigens (TAAs): tumor antigen expressed in most of human tumors with little or no expression
tyrosinase, gp100, MAGE-A3, MAGE-C2, and PRAME. DCs transfected ex in somatic cells. We report here the final results of a First-In-Human Phase I
vivo with TriMix and TAAs mRNAs showed significant clinical activity in study evaluating INVAC-1 as a single agent in patients (pts) with advanced
combination with ipilimumab in metastatic melanoma without increasing solid tumors, ended in June 2018. Methods: A two center Phase I trial
toxicity. This study aims to assess the safety and immunogenicity of ECI-006 evaluated INVAC-1 given monthly for a minimum of 3 cycles and up to 9
vaccine administered intranodally (i.n.) in an adjuvant setting for patients cycles by intradermal injection followed by electroporation (n = 20) or using a
with resected melanoma. Methods: Twenty patients who underwent re- needle-free injection system (n = 6). Primary objectives included safety,
section of stage IIc/III/IV cutaneous melanoma received 5 administrations of tolerability and dose limiting toxicities to identify the maximum tolerated
ECI-006 (either 600 mg or 1800 mg [n = 10, each]) injected i.n. on Day 1 and dose and recommended phase 2 dose. Secondary objectives included im-
after 2, 4, 6 and 14 weeks. Treatment-emergent adverse events (TEAEs) mune response (assessed by IFN-g Elispot) and anti-tumor activity. Immuno-
were graded using CTCAE version 4.0.3. Blood samples for immune mon- monitoring included detection of autoantibodies, lymphocyte phenotyping
itoring (ELISPOT and intracellular cytokine staining [ICS]) were collected and inflammatory cytokine levels in blood. Anti-tumor activity was evaluated
pre-dose and at weeks 4, 7, 14 and 15. Results: Nineteen patients com- through RECIST 1.1 adapted to immune response, and plasma circulating
pleted the treatment. One patient in the low dose group discontinued the tumor DNA (ctDNA). Results: 26 pts with refractory/progressive tumors were
study after 4 doses due to disease relapse. Administration of ECI-006 was enrolled and treated with 3 escalating doses of 100, 400 and 800 mg. 15 pts
well tolerated. No serious adverse events or TEAEs Grade 3 or higher were experienced stable disease according to RECIST. For 11 of them, the
reported. Of all TEAEs, myalgia and fatigue were the most reported in 3 treatment was extended, up to 9 months. INVAC-1 was well tolerated with no
(15%) and 5 (25%) patients, respectively. ELISPOT and ICS were performed dose-limiting toxicities. No significant biological signs of autoimmunity were
on T cells pre-stimulated in vitro for 10-12 days, using a previously in-house observed. No significant modification in inflammatory plasma cytokines
validated protocol. Vaccine-induced immune responses according to pre- levels was observed after INVAC-1 administration. INVAC-1 triggered de
defined criteria were detected in 4/10 and 3/9 patients treated with the low novo or enhanced pre-existing CD4/CD8 specific anti-hTERT response in
and high dose, respectively. Samples from these patients are currently being 63% of pts. This specific anti-hTERT immune response was enhanced ex
subjected to T-cell receptor repertoire analysis. Conclusions: Among pa- vivo by adding the immune checkpoint inhibitor nivolumab. ctDNA was
tients undergoing resection of stage IIc/III/IV melanoma, i.n. administration evaluated in 17 pts. We observed a ctDNA decrease in 6 cases, a stable level
of ECI-006 at 600 or 1800 mg was generally well tolerated. ECI-006 in 5 cases and an increase in 6 cases. Conclusions: Results indicate that
demonstrated to be immunogenic in a proportion of patients. These re- INVAC-1 was well tolerated and immunogenic at the doses and schedule
sults warrant further development of ECI-006 in combination with anti-PD-1 tested. Disease stabilization was obtained for the majority of pts (58%)
therapy in melanoma patients. Clinical trial information: NCT03394937. according to RECIST criteria or ctDNA levels. Clinical trial information:
NCT02301754.

2643 Poster Session (Board #287), Sat, 8:00 AM-11:00 AM TPS2644 Poster Session (Board #288a), Sat, 8:00 AM-11:00 AM
Immunogenicity and tolerability of personalized mRNA vaccine mRNA- A first-in-human study of KY1044, a fully human anti-ICOS IgG1 antibody as
4650 encoding defined neoantigens expressed by the autologous cancer. monotherapy and in combination with atezolizumab in patients with selected
First Author: Gal Cafri, National Cancer Institute Surgery Branch, Bethesda, advanced malignancies. First Author: Sonia Quaratino, Kymab Ltd, Cam-
MD bridge, United Kingdom
Background: Therapeutic vaccination against cancer has proven very Background: The Inducible T-cell costimulator (ICOS/CD278) is related to
challenging with little clinical benefit. Vaccines against non-viral tumors the CD28 superfamily and is induced upon T cell activation. There is a
have mainly targeted differentiation antigens, cancer testis antigens, and hierarchical order of ICOS expression level, in which highly immunosup-
over-expressed antigens. However, negative selection in the thymus against pressive TReg (CD4+Foxp3+) present in the tumor microenvironment (TME)
these normal non-mutated antigens severely limits the ability to generate are at the top level of expression and CD8 TEff cells are at the bottom level. In
high avidity anti-cancer T-cells. The importance of neoantigens to each addition, ICOS expression on TRegs is higher in the TME than in the blood or
patient’s unique cancer as targets for immunotherapy has been extensively spleen. Methods: KY1044, a fully human anti-ICOS IgG1 kappa monoclonal
studied, by our group and others. It is now clear that neoantigen-specific antibody, selectively binds to ICOS with high affinity (which is maintained at
T-cells are present in most cancers and these neoantigens derived from intratumoral acidic pH) and has a dual mechanism of action: it promotes the
somatic mutations offer a specific and highly immunogenic target for per- preferential depletion of intratumoral ICOSHigh TRegs resulting in an increase
sonalized vaccination. We developed a process to identify immunogenic in the TEff:TReg ratio in the TME; and it stimulates ICOSLow TEff cells. Pre-
T-cell epitopes derived from tumor-specific mutations using tumor- clinical data demonstrate that KY1044 monotherapy or in combination with
infiltrating lymphocytes. Methods: We combined, for the first time, validated anti-PD-L1 is associated with immune cell activation and anti-tumor re-
defined neoantigens, predicted neoepitopes and mutations in driver genes sponse. In order to validate the dual mechanism of action in vivo, studies in
into a single mRNA concatemer (mRNA-4650) to vaccinate patients with mice and cynomolgus monkeys were conducted. Firstly, KY1044 injected
metastatic common epithelial cancers. We are conducting a phase I/II trial in i.v. at doses up to 100 mg/kg weekly were well tolerated in non-human
patients with metastatic melanoma, gastrointestinal, or genitourinary can- primates. In addition, pharmacodynamic studies in mice and cynomolgus
cers with at least one lesion that is resectable. Patients must have an ECOG monkey confirm preferential depletion of ICOSHigh cells. KY1044-CT01 is an
status of #1 with adequate organ and bone marrow function. Patients are open-labelled first in human Phase I/II study assessing the safety, tolera-
vaccinated intramuscularly at two-week intervals for four cycles, and dosing bility, PK, PD and anti-tumor activity of KY1044 administered every 3 weeks
may be repeated for a second course of vaccination. Key primary endpoints (Q3W) as an i.v. single agent infusion and in combination with atezolizumab
are safety, tolerability and the development of T-cell reactivity as well as (1200 mg, Q3W IV) in adult patients with advanced/metastatic malignan-
objective response rate. Results: mRNA-4650 is safe at all dose levels cies. The primary endpoint of the Phase I dose escalations, designed as
studied to date with no reported DLTs or drug related SAEs. Neoantigen sequential but overlapping arms of KY1044 monotherapy and combination
specific CD8 and CD4 T cells responses against neoepitopes included in with atezolizumab, is safety and tolerability. Secondary endpoints are the
the vaccine have been observed and no tumor regressions were seen. characterization of pharmacokinetic, pharmacodynamic and efficacy pro-
Conclusions: Our data show that such a personalized mRNA vaccination is files in all patients. In the Phase II part, the primary endpoint is overall
feasible and can elicit neoantigen specific T cell responses. Combination of response rate (ORR), and the measure of clinical efficacy will be confirmed
vaccines with checkpoint inhibitors or adoptive T cell therapy can open the as per RECIST 1.1. and immune-related (ir)RECIST. An intensive biomarker
possibility to develop effective immunotherapies for patients with the plan is integral to the study design to underpin the phenotypic and molecular
common epithelial cancers. Clinical trial information: NCT03480152. changes in both peripheral blood and tumor. Clinical trial information:
NCT03829501.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Immunotherapy and Tumor Immunobiology 141s

TPS2645 Poster Session (Board #288b), Sat, 8:00 AM-11:00 AM TPS2646 Poster Session (Board #289a), Sat, 8:00 AM-11:00 AM
Phase 1 with expansion cohorts in a study of NEO-201 in adults with chemo- A phase I/Ib multicenter study to evaluate the humanized anti-CD73 antibody,
resistant solid tumors. First Author: Maria Pia Morelli, NCI, Bethesda, MD CPI-006, as a single agent, in combination with CPI-444, and in combination
with pembrolizumab in adult patients with advanced cancers. First Author:
Background: NEO-201 is a novel humanized IgG1 monoclonal antibody
Mehrdad Mobasher, Corvus Pharmaceuticals Inc, Burlingame, CA
(mAb) generated against the Hollinshead allogenic colorectal cancer vaccine
platform. Briefly, tumor-associated antigens (TAA) derived from tumor Background: CD73 expression is elevated in tumors and contributes to in-
membrane fractions pooled from colorectal cancer surgical specimens were creasing levels of immunosuppressive adenosine in the tumor microenvi-
screened for delayed-type hypersensitivity and evaluated in clinical trials. ronment. CD73 knockout mice exhibit reduced tumor growth and resistance
The original vaccine was used to generate monoclonal antibodies, one of to experimental metastasis. Inhibition of CD73 activity with an anti-CD73
which is NEO-201. In preclinical data generated in our laboratory, we have antibody blocks adenosine production, shown to inhibit tumor growth in
demonstrated that NEO-201 exert anti-tumor activity by natural killer (NK)- syngeneic models. CPI-006 is a humanized IgG1 FcgR binding-deficient
mediated antibody-dependent cytotoxicity (ADCC) against several tumor anti-CD73 antibody now being investigated in this Phase 1/1b multicenter,
type including colorectal and pancreatic cancer models (Fantini, et al. open label trial as single agent (SA) or combination with CPI-444, an oral,
2018). We have identified NEO-201 antigen as a glycosylated form of small molecule, selective A2aR antagonist or in combination with pem-
CEACAM-5 and -6, which is expressed by tumor tissue but is not present in brolizumab, an anti-PD1 indicated for the treatment of patients across a
the surrounding healthy tissue (David, et al. 2018). This could result in a number of malignancies (NCT03454451). Methods: Up to 462 subjects will
specific anti-tumor activity without significant normal tissue toxicity. Nev- be enrolled at approximately 35 sites in the US, Canada and Australia.
ertheless, toxicity was further assessed in non-human primates and transient Eligible patients must have: non-small cell lung, renal cell carcinoma,
neutropenia was the only adverse event observed. Based on this data we urothelial bladder, cervical, colorectal, ovarian, pancreatic, prostate, head
designed a first in human phase I trial to evaluate the safety, maximum- and neck, triple-negative breast, endometrial, select sarcomas and non-
tolerated dose (MTD), pharmacokinetics (PK) and pharmacodynamics (PD) Hodgkin lymphoma malignancies relapsed, refractory or intolerant to 1 to 5
of the humanized monoclonal antibody NEO-201. Methods: This is a first-in- standard therapies; aged $ 18 yo; adequate organ function and measurable
human phase 1 study with expansion cohort to determine the maximum disease. The objectives of the study are 1) evaluate the safety and tolerability
tolerated dose (MTD) and recommended phase II dose (RP2D) of NEO-201 of SA CPI-006, in combination with CPI-444 and in combination with
in adults with advanced solid tumors that have high likelihood pof expression pembrolizumab, 2) evaluate the pharmacokinetics of each regimen and 3)
NEO201 antigen and have progressed to standard of treatments and have a identify potential biomarker signals predictive of response. Study design in
PS0-2 ECOG. Study design is a classic Fibonacci (3+3) dose escalation, table. Study Design. Clinical trial information: NCT03454451.
with a cohort expansion at the MTD. NEO-201 is administered intravenously Dose escalation.
every two weeks, and four different dose levels will be explored (DL1 = 1mg/
kg, DL2 = 2mg/kg, DL3 = 4mg/kg and DL4 = 6mg/kg). No intra-patient dose 3+ 3 Design: 1, 3, 6, 12, 18, 24 mg/kg
escalation is allowed. Patients will be evaluated for safety every two weeks, CPI-006 SA CPI-006 + CPI-444 CPI-006 +
with weekly laboratory testing, according to CTCAEv4.0. and with a DLT pembrolizumab
window of 28 days (cycle 1). Response will be assessed every 8 weeks (2 Dose Expansion Stage 1
(N=11 per cohort)
cycles of treatment) according to RECISTv1.1. Additionally, biological
CPI-006 SA CPI-006 + CPI-444 CPI-006 +
samples will be collected to understand NEO-201 pharmacokinetic, the pembrolizumab
effect on the immune process and their correlation with treatment toxicity NSCLC RCC Others NSCLC RCC Others NSCLC RCC Others
and response. As of February 2019 we have completed enrollment in the first If 1 or more responses observed in a disease cohort, proceed to Stage 2
DL and are evaluating for DLT. Clinical trial information: NCT03476681. Dose Expansion Stage 2
(N=17 per cohort)

TPS2647 Poster Session (Board #289b), Sat, 8:00 AM-11:00 AM TPS2648 Poster Session (Board #290a), Sat, 8:00 AM-11:00 AM
A phase I clinical trial using armored GPC3 CAR T cells for children with A phase II study of autologous tumor infiltrating lymphocytes (TIL, LN-144/
relapsed/refractory liver tumors. First Author: David Henry Michael Steffin, LN-145) in patients with solid tumors. First Author: Jason Alan Chesney,
Baylor College of Medicine, Houston, TX James Graham Brown Cancer Center, University of Louisville, Louisville, KY
Background: CAR T therapies have been successful against hematologic Background: Adoptive cell therapy (ACT) with tumor infiltrating lymphocytes
malignancies, but have benefited only a handful of patients with solid (TIL) has demonstrated durable complete responses in immunogenic tumors
cancers. Glypican 3 (GPC3) is an attractive immunotherapeutic target due to with high mutational burden in metastatic melanoma patients who had not
its preferential expression on multiple pediatric and adult solid cancers and received prior immune checkpoint inhibitors (ICI); CR rate 24%. Pem-
lack of expression on non-malignant tissues. GPC3-CAR T cells were tested brolizumab is an approved agent for the treatment of metastatic melanoma
preclinically and inclusion of the 4-1BB costimulatory endodomain with IL- and head & neck cancers, among others. Further, ICI have been reported to
15 and IL-21 co-expression enabled CAR T cells to expand and persist the potentially enhance the efficacy of TIL therapy. One aim of this study is to
most in vitro and in vivo and led to robust antitumor activity in vivo. We are improve the efficacy response for early line patients by combining TIL with
now testing GPC3-CAR T cells with IL15 and IL-21 for the first time in anti-PD-1 in ICI-naı̈ve patients with metastatic melanoma (Cohort 1) and
children with relapsed/refractory liver tumors. Methods: In this Phase 1 trial head & neck cancers (Cohorts 2). In Cohort 3, TIL therapy alone is offered to
(GAP, NCT02932956), we are evaluating patients in 3 cohorts: 1) GPC3- NSCLC patients who have received prior systemic therapy, including ICI.
CAR alone; 2) GPC3-CAR and IL-15; 3) GPC3-CAR with IL-15 and IL-21. We Methods: IOV-COM-202 is a prospective, Phase 2 multicenter, open-label
will 1) define the safety and establish the Recommended Phase 2 Dose study in which 36 patients (12 per cohort) are to be enrolled in one of three
(RP2D) of GPC3-CAR T cells co-expressing IL-15 and IL-21; 2) determine cohorts; Cohorts 1 and 2: TIL therapy in combination with pembrolizumab, or
persistence and anti-tumor activity of GPC3-CAR T cells; 3) examine changes Cohort 3: TIL therapy alone. Patients will have tumors resected at local
in gene and protein expression in the tumor microenvironment associated with centers and shipped to a central GMP facility to undergo a 22-day
potential immune escape mechanisms. Inclusion criteria are the following: manufacturing process that yields cryopreserved infusion product (LN-144/
age #18; histology proven, GPC3-positive tumor; life expectancy.12 weeks; LN-145) that is shipped back to treating center. All patients receive TIL
Child-Pugh-Turcotte score,7; serum AST,5 times ULN; total bilirubin,3 therapy consisting of 1 week of preconditioning cyclophosphamide/
times ULN for age; INR #1.7; absolute neutrophil count.500/ml; platelet fludarabine, followed by a single infusion of LN-144/LN-145 (Day 0) and
count.20,000/ml; Hgb$9.0 g/dl. Toxicity will be monitored using the up to 6 doses of IL-2 (600,000 IU/kg). Patients in Cohorts 1 and 2 also
Common Terminology Criteria of Adverse Events v4. The RP2D will be de- receive pembrolizumab on Day -1 and then Q3W for up to 2 years or until
termined by the standard 3+3 dose escalation method using 5 dose levels. disease progression or acceptable toxicity. Co-primary endpoints for each
Persistence will be quantified using RT-PCR and flow cytometry. Antitumor cohort are objective response rate (ORR) per RECIST 1.1, and safety
activity will be defined by 3D imaging using RECIST 1.1 criteria and the (grade $ 3 TEAE). Eligibility criteria: Cohorts 1 (melanoma) and 2 (head &
immune-related response criteria. Immune-escape will be examined using neck): patients must not have received prior ICI (eg, anti-PD-1, anti-CTLA-4)
single cell RNA sequencing and imaging of paraffin-embedded tissues using and may have received up to 3 lines of prior systemic therapy, Cohort 3
codetection by indexing to evaluate candidate proteins. Data will be analyzed (NSCLC): patients must have received 1-3 prior lines of systemic therapy
via descriptive statistics. Cohort 1 of this study is now open for enrollment. including ICI. After tumor resection for TIL manufacturing, patients must
Clinical trial information: NCT02932956. have additional measurable disease for assessment per RECIST 1.1. Ade-
quate bone marrow/organ function and ECOG PS of 0 or 1 is required.
Clinical trial information: NCT03645928.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
142s Developmental Immunotherapy and Tumor Immunobiology

TPS2649 Poster Session (Board #290b), Sat, 8:00 AM-11:00 AM TPS2650 Poster Session (Board #291a), Sat, 8:00 AM-11:00 AM
Intravenous administration of ALKS 4230 as monotherapy and in combination Adoptive transfer of tumor-infiltrating lymphocytes in patients with sarcomas,
with pembrolizumab in a phase I study of patients with advanced solid tumors. ovarian, and pancreatic cancers. First Author: Rodabe Navroze Amaria, The
First Author: Ulka N. Vaishampayan, Wayne State University, Detroit, MI University of Texas - MD Anderson Cancer Center, Houston, TX
Background: ALKS 4230 is a fusion protein of circularly permuted IL-2 and Background: Adoptive cell therapy (ACT) with tumor-infiltrating lymphocytes
IL-2 Receptor (IL-2R) a designed to selectively bind the intermediate- (TIL) has a long history of efficacy in metastatic melanoma, and is being
affinity (ia) IL-2R, comprised of IL-2Rb and gc, for activation of CD8+ increasingly considered across other solid tumors. Preclinical data generated
T cells and NK cells, which drive antitumor immune responses. In contrast, at MD Anderson Cancer Center has demonstrated the ability to grow TIL
unmodified IL-2 activates high-affinity (ha) IL-2R, driving the expansion of from a variety of tumor types including various types of sarcomas, ovarian and
immunosuppressive CD4+ regulatory T cells (Tregs) at concentrations below pancreas cancers. We are testing the efficacy of TIL across multiple tumor
those that activate iaIL-2R expressing cells. Binding of IL-2 to haIL-2R on types using two different manufacturing protocols. Methods: We are con-
endothelial cells may contribute to capillary leak syndrome seen with high- ducting two ongoing investigator initiated basket TIL therapy trials. The first
dose IL-2. Thus, selective activation of the iaIL-2R by ALKS 4230 has the (NCT03449108) includes cohorts with poorly differentiated soft tissue and
potential to enhance tumor killing and improve tolerability. ALKS 4230 has bone sarcomas, osteosarcoma, and platinum resistant ovarian cancer. The
previously been shown to improve antitumor activity relative to IL-2 in murine TIL product used in this trial is an investigational cell product (LN-145,
models. In this clinical study, ALKS 4230 will be assessed as monotherapy Iovance Biotherapeutics, Inc.). The second trial (NCT03610490) includes
and in combination with anti-PD-1 therapy. Methods: ALKS 4230 is being cohorts of osteosarcoma, platinum resistant ovarian cancer, and pancreatic
studied in adults with advanced solid tumors in a phase I first-in-human trial cancer (who have progressed on, or received maximal benefit from, front-line
designed primarily to assess the safety of ALKS 4230 alone and with therapy). For this trial, TIL are manufactured at MD Anderson Cancer Center
pembrolizumab. The study will also determine a monotherapy recommended using a protocol that includes the use of urelumab (an agonistic anti-CD137
phase 2 dose (RP2D) and characterize pharmacokinetics, pharmacody- antibody) combined with T cell receptor activation during TIL expansion. In
namics (PD), immunogenicity, and evidence of anti-tumor activity. It will be both trials eligible subjects undergo tumor harvest using a surgical excisional
conducted in 3 parts: monotherapy dose escalation (Part A), monotherapy biopsy of the tumor for TIL manufacturing, receive a modified cyclophos-
dose expansion at the RP2D (Part B), and combination therapy with phamide and fludarabaine lymphodepletion regimen and up to six doses of
pembrolizumab (Part C). ALKS 4230 is administered as a 30 minute IV IL-2 (600,000 IU/kg) following TIL infusion. No intervening therapy is
infusion once daily for five days in each 14 or 21 day cycle. Part A is in- allowed between tumor harvest and initiation of lymphodepletion. The
patient. Eligibility requires ECOG PS 0-1 and adequate bone marrow, liver primary endpoint for each cohort is ORR as assessed by investigators using
and kidney function. Part B will enroll 21 patients each in renal cell car- RECIST 1.1 criteria. The Simon’s two stage design is used to monitor the
cinoma and melanoma cohorts. Part C will enroll up to 79 patients total into 3 efficacy of each cohort independently. In the first stage, 10 patients will be
cohorts based on tumor type and prior anti-PD-1 therapy; a 4th cohort will treated per cohort. If there is no confirmed response in these 10 evaluable
enroll patients from Part A or B who received at least 4 cycles of ALKS 4230 patients, the cohort will be terminated. If the cohort moves forward to Stage
or experienced disease progression on monotherapy. The primary PD end- II, an additional 8 patients will be treated leading to a total of 18 patients.
point is change from baseline in CD8+ T, NK, and Treg cell counts. In- Three or more responders out of 18 treated patients for the cohort will be
flammatory cytokine levels will also be measured. Parts A and C are currently considered clinically relevant to justify further investigation. Enrollment is
enrolling. Clinical trial information: NCT02799095. ongoing in all cohorts in both trials. An accrual update will be provided at the
annual meeting. Clinical trial information: NCT03449108, NCT03610490.

TPS2651 Poster Session (Board #291b), Sat, 8:00 AM-11:00 AM TPS2652 Poster Session (Board #292a), Sat, 8:00 AM-11:00 AM
The "INSIGHT" Trial: Two new strata of an explorative, open-labeled phase I A first-in-human phase I study of FS118, an anti-LAG-3/PD-L1 bispecific
study evaluating the feasibility and safety of subcutaneous IMP321 injections antibody in patients with solid tumors that have progressed on prior PD-1/PD-L1
(LAG-3Ig fusion protein, eftilagimod alpha) combined with either standard-of- therapy. First Author: Timothy A Yap, The University of Texas MD Anderson
care drug therapy or PD-L1 inhibition (avelumab) in advanced-stage solid Cancer Center, Houston, TX
tumor entities. First Author: Daniel Wilhelm Mueller, University Cancer Center
Background:PD-1/PD-L1 checkpoint inhibitors demonstrated remarkable
Frankfurt, Institut für Klinisch-Onkologische Forschung and IKF Klinische
anti-tumor activity, but only a minority of patients achieve full clinical benefit
Krebsforschung GmbH am Krankenhaus Nordwest, Frankfurt, Germany
with deep and durable responses. Translational studies suggest resistance to
Background: The two new strata of the INSIGHT trial evaluate feasibility and cancer immunotherapy can be mediated by additional immune checkpoints
safety of s.c. injections of IMP321 (eftilagimod alpha) in combination with e.g. lymphocyte-activation gene 3 (LAG-3). The combination of LAG-3 and
either SOC first/second-line drug therapy (Stratum C) or in combination with PD-1 mAbs synergistically improved anti-tumor response in murine models
an PD-L1 inhibitor (avelumab; Stratum D) in advanced stage solid tumors as and early clinical trials. In a small cohort, TIL LAG-3 expression enriched for
well as to generate first efficacy data. This proof-of-concept data could build responsiveness in PD-1/PD-L1 relapsed/refractory patients. FS118 is a novel
the basis for further clinical studies exploring the therapeutic potential of bispecific antibody incorporating a LAG-3 binding Fc-region into a PD-L1-
combinations of active immunotherapy using IMP321 with SOC drug specific IgG1 antibody to potentially deliver superior anti-tumor efficacy
therapies or immunotherapies targeting the PD-1/PD-L1 axis in various solid while limiting immunotherapy-related adverse effects by dual targeting.
tumor entities. IMP321 is a MHC class II agonist that activates antigen- Methods: The FiH study (NCT03440437) is being conducted in adult pa-
presenting cells (primary target cells) and then CD8 T cells (secondary target tients with solid tumors who failed prior PD-1/PD-L1 treatment. Primary
cells). Activation of the dendritic cell network and subsequent T cell re- objectives of the study are to determine safety, PK and the maximum
cruitment at the tumor site with IMP321 may lead to enhanced anti-tumor tolerated/recommended Phase 2 dose of FS118. Secondary objectives in-
CD8 T cell responses. Thus, especially combinations with PD-1/PD-L1 in- clude preliminary evidence of efficacy, immunogenicity, PD profile and
hibitors might display interesting effects by activating immune cells and exposure/response correlation. 50 subjects from four study sites in the USA
disabling immune inhibitory mechanisms at the same time. Methods: This are planned to enrol in dose escalation initiated with accelerated titration (5
is a prospective investigator initiated phase I trial consisting of four strata. single subject cohorts) followed by 3+3 design and expansion cohorts.
New stratum C: Patients with solid tumors treated with SOC chemo- or FS118 is administered weekly IV in 21-day treatment cycles until pro-
targeted therapy in first or second line receive concomitant s.c. IMP321 gression, unacceptable toxicity, withdrawal, or death. Patients are followed
injections. This combination is aimed to enhance the immune response for safety, overall survival and initiation of subsequent therapy. DLT
against tumor cells compared to chemo-/targeted SOC therapy alone. New clearance, dose escalation and cohort expansion (to further characterise
stratum D: Patients will receive avelumab i.v. q2w along with s.c. IMP321 safety, PK/PD or clinical efficacy) are supervised by a safety review com-
injections. This combination is aimed to enhance efficacy by combining mittee (SRC). Translational studies assess PD-L1/LAG-3 receptor occu-
IMP321’s activating effects on immune cells with the release of immune pancy, soluble PD-L1/LAG-3 levels and the correlation of FS118 exposure
inhibitory effects caused by interruption of the PD-1/PD-L1 axis. It is with selected PD markers of target engagement and response. Translational
planned to enroll 20 patients in Stratum C and 12 patients in stratum D. endpoints include TIL analysis, transcriptomic profiles and target expression
Main efficacy endpoint is the overall response rate (RECIST 1.1). Overall analyses on tumor tissues. Cohorts 1 through 6 have been completed, en-
recruitment has started; currently (Feb 2019) 14 patients have been en- rollment in cohort 7 began December 2018. Clinical trial information:
rolled. EudraCT: 2016-002309-20. Clinical trial information: NCT03252938. NCT03440437.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Immunotherapy and Tumor Immunobiology 143s

TPS2653 Poster Session (Board #292b), Sat, 8:00 AM-11:00 AM TPS2654 Poster Session (Board #293a), Sat, 8:00 AM-11:00 AM
A first-in-human, multicenter, open label, phase I study in patients with A phase Ia/b study of TIM-3/PD-L1 bispecific antibody in patients with
advanced and/or refractory solid malignancies to evaluate the safety of advanced solid tumors. First Author: Matthew David Hellmann, Memorial
intravenously administered ATOR-1015. First Author: Jeffrey Yachnin, Sloan Kettering Cancer Center and Weill Cornell Medical College, New York,
Karolinska Institutet, Stockholm, Sweden NY
Background: ATOR-1015 is a human bispecific IgG1 antibody targeting Background: Programmed cell death 1 immune checkpoint inhibitors (anti-
cytotoxic T-lymphocyte associated protein 4 (CTLA-4) and the tumor ne- PD-1, anti-PD-L1) have demonstrated clinical benefit in a subset of patients
crosis factor receptor superfamily member 4, OX40 (also known as CD134). with manageable safety across a variety of tumor types. T-cell immuno-
Both in vitro and in vivo, ATOR-1015 induces activation of cytotoxic T cells globulin and mucin-domain-containing molecule-3 (TIM-3) can be co-
and depletion of regulatory T cells (1). In syngeneic tumor models, using expressed with PD-1 on exhausted T-cells and may be upregulated in tu-
human OX40 transgenic mice cross-reacting with both targets, ATOR-1015 mors refractory to anti-PD-1 therapy (Koyama et al. 2016). Pre-clinical
is demonstrated to localize to the tumor. Further, the effects of ATOR-1015 studies demonstrated that blockade of both PD-1 and TIM-3 improved
are shown to occur in the tumor area and not in the spleen (1). Treatment survival of tumor-bearing mice compared to blocking anti-PD-1 only
with ATOR-1015 also reduces tumor growth and improves survival in several (Koyama et al. 2016). LY3415244 is a TIM-3/PD-L1 bispecific antibody
tumor models in mice, including bladder, colon and pancreatic cancer (1). that has the ability to target and inhibit both TIM-3 and PD-L1 and the
The non-clinical safety profile and the pharmacokinetics were established in potential to overcome primary and acquired anti-PD-(L)1 resistance by a novel
cynomolgus monkeys and the data were used for the dosing schedule. mechanism to bridge TIM-3- and PD-L1-expressing cells. Methods: Study
Methods: This is a multicenter, open-label, dose escalation study enrolling JZDA is a multicenter, nonrandomized, open-label, Phase 1a/1b study of
patients with advanced and/or refractory solid malignancies (NCT03782467). LY3415244 in patients with advanced solid tumors. In Phase 1a, subjects
The primary objective of the study is to determine the maximum tolerated dose with any tumor type who are either PD-(L)1 inhibitor-naı̈ve or exposed are
(MTD) or the recommended phase 2 dose (RP2D) and to establish the safety eligible. In Phase 1b, expansion cohorts are planned in subjects with PD-
profile of ATOR-1015. ATOR-1015 is administered intravenously biweekly (L)1-experienced NSCLC, urothelial carcinoma, and melanoma. Patients
as a single agent until confirmed progressive disease, unacceptable toxicity or with malignant mesothelioma are not required to have received prior anti-PD-
withdrawal of consent. The study will start with single patient cohorts until (L)1 therapy. The primary objective is to assess safety and tolerability of
grade 2 toxicities are observed, thereafter the study follows a modified 3+3 LY3415244 and identify the recommended Phase 2 dose (RP2D) in Phase 1a
design. At the MTD/RP2D, or a lower dose, up to 20 patients are planned for (dose escalation). Safety and tolerability of the RP2D will be assessed in Phase
additional safety and efficacy evaluation. Study enrollment was initiated 1b (dose expansion). The secondary objectives are to assess the pharmaco-
in January 2019. A total of up to 53 patients are estimated to be enrolled in kinetics of LY3415244 in Phase 1a/1b and assess early antitumor activity of
the study. (1) Månsson Kvarnhammar et al. Journal for ImmunoTherapy of LY3415244 in Phase 1b cohorts. Pre- and on-treatment biopsies will be
Cancer 2018; 6(Suppl 1):115. Abstract P683. Clinical trial information: obtained to explore potential biomarkers of response. During Phase 1a, dose
NCT03782467. escalation cohorts will proceed via a modified toxicity probability interval-2
(mTPI-2) design with a 1-cycle (28-day) dose-limiting toxicity (DLT) obser-
vation period. LY3415244 will be dosed intravenously every 2 weeks. Data
from Phase 1a will determine the RP2D, which will be used for all cohorts in
Phase 1b. The study is currently open to enrollment. Clinical trial information:
NCT03752177.

TPS2655 Poster Session (Board #293b), Sat, 8:00 AM-11:00 AM TPS2656 Poster Session (Board #294a), Sat, 8:00 AM-11:00 AM
A phase III trial-in-progress comparing tislelizumab plus chemotherapy with A randomized, placebo-controlled, phase III trial-in-progress to evaluate the
placebo plus chemotherapy as first-line therapy in patients with locally efficacy and safety of tislelizumab plus chemotherapy as first-line treatment
advanced unresectable or metastatic gastric or gastroesophageal junction for unresectable, locally advanced recurrent/metastatic esophageal squamous
(G/GEJ) adenocarcinoma. First Author: Rui-hua Xu, Sun Yat-sen University cell carcinoma (ESCC). First Author: Jian-Ming Xu, The Fifth Medical Center,
Cancer Center, Guangzhou, China People’s Liberation Army General Hospital, Beijing, China
Background: In patients (pts) with locally advanced or metastatic G/GEJ Background: ESCC remains the predominant histological subtype of, and
cancer, fluoropyrimidine- and platinum (plt)-based combination chemo- accounts for most deaths from, esophageal cancer. PD-1 inhibition has
therapy is first-line standard of care. Despite improvement in chemotherapy demonstrated antitumor activity and was generally well tolerated in patients
regimens, outcomes are poor and survival remains low. Tislelizumab, an (pts) with advanced unresectable or metastatic ESCC. Tislelizumab, an
investigational anti-PD-1 antibody, was engineered to minimize binding of FcgR investigational anti-PD-1 antibody, was engineered to minimize binding to
on macrophages in order to abrogate antibody-dependent phagocytosis, a FcgR on macrophages to abrogate antibody-dependent phagocytosis, a
mechanism of T-cell clearance and potential resistance to anti-PD-1 therapy. mechanism of T-cell clearance and potential resistance to anti-PD-1 ther-
Previous reports suggested tislelizumab, as a single agent and in combination apy. Results from early phase clinical studies suggest single-agent tisleli-
with chemotherapy, was generally well tolerated and had antitumor activity in zumab was generally well tolerated and had antitumor activity in pts with
pts with advanced solid tumors, including G/GEJ cancer. Methods: This global, solid tumors, including ESCC. Methods: This phase 3, randomized, placebo-
double-blind, randomized, phase 3 study (NCT03777657) is designed to controlled, double-blind study (NCT03783442) is designed to evaluate the
compare plt/fluoropyrimidine + tislelizumab versus plt/fluoropyrimidine + efficacy and safety of tislelizumab plus chemotherapy as first-line treatment
placebo as first-line therapy for pts with locally advanced or metastatic G/GEJ of unresectable, locally advanced recurrent or metastatic ESCC. Adult pts
cancer. Approximately 720 pts from 160 centers will be randomized 1:1 to with histologically confirmed ESCC that had metastatic disease either at first
receive tislelizumab (200 mg IV Q3W) or placebo (IV Q3W) in combination with diagnosis or with a $6 month treatment-free interval will be eligible. Ad-
chemotherapy. Oxaliplatin (130 mg/m² IV Q3W) plus capecitabine (1000 ditional eligibility criteria include measurable/evaluable disease, ECOG
mg/m2 orally twice daily for 2 weeks) or cisplatin (80 mg/m² IV Q3W) plus performance score #1, and no prior anti-PD-(L)-1, PD-L2, or other first-line
5-fluorouracil (800 mg/m2/day IV on Days 1–5 Q3W) will be used as backbone therapy or palliative radiation treatment #4 weeks before treatment. Ap-
chemotherapy on an individual basis. Chemotherapy will be administered for up proximately 480 pts will be randomized 1:1 to receive investigator-chosen
to 6 cycles; capecitabine maintenance therapy is optional for pts who received chemotherapy (ICC) plus tislelizumab 200 mg IV Q3W or ICC plus placebo.
capecitabine and oxaliplatin. PD-L1 expression will be assessed using the Chemotherapy options include: platinum (cisplatin 60–80 mg/m2 or oxa-
VENTANA PD-L1 (SP263) assay. Progression-free survival and overall survival liplatin 130 mg/m2 IV Q3W) plus 5-FU 750–800 mg/m2 IV daily for 5 days
are primary endpoints in the intent-to-treat and PD-L1-positive analysis sets of Q3W; or platinum plus capecitabine 1000 mg/m2 orally BID for 14 days
the study. Secondary endpoints include overall response rate, duration of re- Q3W; or platinum + paclitaxel 175 mg/m2 IV Q3W. Progression-free survival
sponse, quality-of-life outcomes, and the safety/tolerability profile of combi- and overall survival are coprimary endpoints; secondary endpoints include
nation therapy. Exploratory endpoints include disease control rate, time to objective response rate, duration of response, and health-related quality-of-
response, and an analysis of potential predictive biomarkers including, but not life. Safety will be assessed by monitoring adverse events, physical exam-
limited to, PD-L1 expression. Clinical trial information: NCT03777657. inations, vital signs, and electrocardiograms. This study is actively enrolling.
Clinical trial information: NCT03783442.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
144s Developmental Immunotherapy and Tumor Immunobiology

TPS2657 Poster Session (Board #294b), Sat, 8:00 AM-11:00 AM TPS2658 Poster Session (Board #295a), Sat, 8:00 AM-11:00 AM
A phase I study evaluating COM701 in patients with advanced solid tumors. SWOG 1609 (DART): A phase II basket trial of dual anti-CTLA-4 and anti-PD-
First Author: Drew W. Rasco, South Texas Accelerated Research Therapeutics 1 blockade in rare tumors. First Author: Sandip Pravin Patel, University of
(START), San Antonio, TX California, San Diego, Moores Cancer Center, San Diego, CA
Background: There is a high unmet medical need for the treatment (tx) of Background: Immune checkpoint blockade, in particular anti-CTLA-4 and
patients (pt) who are refractory to or relapse following tx with checkpoint anti-PD-1-directed approaches, have improved outcomes in various tumor
inhibitors. Newer checkpoint therapies with novel mechanisms of action that types. However, little is known about the efficacy of these agents in advanced
can activate T cells and demonstrate antitumor activity in this pre-tx pt rare solid tumors. We sought to investigate the activity of ipilimumab and
population are urgently needed. COM701 is a novel first-in-class humanized nivolumab in previously unstudied rare solid tumors, with planned biomarker
IgG4 monoclonal antibody that binds with high affinity to PVRIG (poliovirus evaluation pending including whole exome sequencing, RNAseq, and
receptor related immunoglobulin domain containing) blocking its interaction multiplex immune profiling via the NCI CIMACs. Methods: We performed a
with its ligand, PVRL2. Both PVRIG and PVRL2 are part of the DNAM axis as prospective, open-label, multicenter phase II clinical trial of ipilimumab
are TIGIT and PD1. Inhibition of PVRIG leads to enhanced activation of T and (1mg/kg iv q6weeks) plus nivolumab (240mg iv q2weeks) across 37 cohorts
NK cells, and PVRIG results in tumor growth inhibition in mouse tumor of rare tumors. Eligible patients had incurable rare cancer, defined histo-
models. We hypothesize that COM701 will demonstrate antitumor activity in logically with an incidence of less than 6 in 100,000 per year, and did not
pts who are checkpoint inhibitor pre-tx. Methods: NCT03667716 is an have an approved or standard therapy available that had been shown to
ongoing open-label first-in-human phase 1 study in pts with advanced solid prolong overall survival. Patients were required to be 18 years of age or older,
tumors. The initial part of this study (Arm A) will evaluate escalating doses of have a Zubrod performance status of 0-2, with absolute neutrophil count $
COM701 monotherapy IV Q3 weekly with single pt cohorts for the initial 4 1,000/mcL, platelets $ 75,000/mcL, hemoglobin $ 8 g/dL, creatinine
and then 3+3 design. Key Inclusion Criteria: Age $18 yrs, histologically clearance $ 50 mL/min, total bilirubin # 2.0 x institutional upper limit of
confirmed locally advanced/ metastatic solid malignancy and has exhausted normal (IULN), AST and ALT # 3.0 x IULN, TSH or free T4 serum # IULN,
available standard therapy, ECOG 0-1, prior anti-PD-1, anti-PD-L1, anti- and normal adrenocorticotropic hormone (ACTH) # IULN. The primary
CTLA-4, OX-40, CD137 permissible. Key Exclusion Criteria: Active auto- endpoint was overall response rate (ORR) by RECIST v1.1 (complete (CR)
immune disease requiring systemic therapy in the last 2 years, symptomatic and partial responses (PR)); secondary endpoints included progression-free
interstitial or inflammatory lung disease, untx or symptomatic central ner- (PFS) and, overall survival (OS), stable disease (SD) $ 6 months, and
vous system metastases. Primary objectives are safety and tolerability of toxicity. The primary objective of this Phase II trial was to evaluate the overall
COM701 as measured by the incidence of adverse events (AEs) and dose- response rate (ORR, confirmed complete and partial responses [CR and PR])
limiting toxicities (21-day DLT window), pharmacokinetics of COM701, and by RECIST v1.1. Our objective was to distinguish between a true ORR 15%
to identify the maximum tolerated dose and/or the recommended dose for (null hypothesis) versus 30% (alternative hypothesis). A Simon’s two-stage
expansion. Secondary objectives are to characterize the immunogenicity and design was used, which required an analysis on the first 6 eligible patients
preliminary antitumor activity of COM701. Statistical Considerations: AEs who received therapy. If 1 or more of the 6 patients had a response (con-
graded as per CTCAE v4.03, responses as per RECIST v1.1. The analyses of firmed CR or PR), an additional 10 patients were to be accrued. The study
all study objectives will be descriptive and hypothesis generating. No DLTs was activated on 1/13/17 with the first patient treated on 3/1/17. The trial is
have been observed in the single pt cohorts. Assessment of pts enrolled into currently open at 862 sites across the NCTN (with 352 sites having enrolled
cohort 5 is ongoing at the time of this submission. Clinical trial information: patients) and 554 patients enrolled to date. Clinical trial information:
NCT03667716. NCT02834013.

TPS2659 Poster Session (Board #295b), Sat, 8:00 AM-11:00 AM TPS2660 Poster Session (Board #296a), Sat, 8:00 AM-11:00 AM
Pembrolizumab in MMR-proficient metastatic colorectal cancer pharmaco- JAVELIN BRCA/ATM: A phase 2 trial of avelumab (anti–PD-L1) plus talazoparib
logically primed to trigger dynamic hypermutation status: The ARETHUSA (PARP inhibitor) in patients with advanced solid tumors with a BRCA1/2 or
trial. First Author: Salvatore Siena, Niguarda Cancer Center, ASST Grande ATM defect. First Author: David Michael Hyman, Memorial Sloan Kettering
Ospedale Metropolitano Niguarda, Milan, Italy Cancer Center, New York, NY
Background: Metastatic colorectal cancer (CRC) harbouring genetic defects Background: Defects in DNA damage response genes, including BRCA1/2
in the mismatch-repair pathway (MMRd) presents with a high tumor mu- and ATM, confer sensitivity to PARP inhibitors. Talazoparib is a potent, oral
tational burden (TMB), and is highly sensitive to anti–programmed cell death PARP inhibitor with a dual mechanism of action (PARP enzyme inhibition
protein 1 (PD-1) immune checkpoint inhibitors. We recently showed in and PARP trapping). Avelumab is a human anti–PD-L1 IgG1 monoclonal
preclinical models that the pharmacological treatment with temozolomide antibody with a wild-type Fc region that has shown clinical activity in multiple
(TMZ) can induce the inactivation of MMR genes, and consequently the tumor types. Preclinical and early clinical data suggest that combining a PARP
increase of TMB and immunogenic neoantigens, thus suggesting that TMZ inhibitor with an immune checkpoint inhibitor may provide improved activity.
could be used to prime MMR proficient (MMRp) tumors for response to Methods: JAVELIN BRCA/ATM (NCT03565991) is an ongoing, open-label,
checkpoint inhibitors. Accordingly, mCRC patients recruited in previous multicenter, phase 2 trial assessing the combination of avelumab and tala-
clinical trials where TMZ was administered, acquired alterations of MMR zoparib. Enrollment of »200 patients with a histologically confirmed locally
genes upon treatment and showed remarkable increase in TMB at disease advanced or metastatic solid tumor that has progressed on . 1 line of
progression (PD). We thus designed the ARETHUSA clinical trial to test standard-of-care treatment for locally advanced or metastatic disease and
whether a priming course with TMZ in patients can sensitize mCRC to the has a germline or somatic defect in BRCA1 or 2 (cohort 1) or ATM (cohort 2)
anti-PD1 inhibitor pembrolizumab. Methods: Arethusa (NCT03519412) is a genes is planned. Patients with concomitant defects in . 1 gene (BRCA1,
2-cohorts, phase II trial consisting of three different phases. In the BRCA2, or ATM) will be enrolled in cohort 1. Exclusion criteria include prior
SCREENING, 348 mCRC RAS-mutated patients will be tested for MMR treatment with an immune checkpoint or PARP inhibitor, prior treatment with
status. MMRd patients will proceed directly to TRIAL for immediate pem- any other anticancer or radiation therapy within 2 weeks prior to enrollment, a
brolizumab treatment (expected 14). MMR-proficient (MMRp) patients will known history of an immune-mediated or autoimmune condition, and known
be further tested for expression of O6-methylguanine-DNA methyltransferase symptomatic brain metastases requiring steroids. The primary endpoint is
(MGMT) by immunohistochemistry and by promoter methylation analysis. confirmed objective response by blinded independent central review
IHC-negative, promoter methylation-positive MMRp patients (expected 67) according to RECIST v1.1 and according to the Prostate Cancer Working
will enter in the PRIMING phase and will be treated with TMZ until PD. TMB Group 3 (PCWG3) for patients with metastatic castration-resistant prostate
will then be assessed on tumor biopsies at resistance. Those patients that will cancer (mCRPC). Secondary endpoints include safety; investigator-assessed
have . 20 mutations/megabase will proceed to TRIAL (expected 20) and will confirmed objective response, time to tumor response, duration of response,
be treated with pembrolizumab. Overall response rate (primary outcome), and progression-free survival (per RECIST v1.1 and per PCWG3 for patients
Progression Free, and Overall Survival, and treatment related toxicities with mCRPC); overall survival; pharmacokinetic parameters; and potential
(secondary outcomes) in MMRp pembrolizumab-treated patients will be predictive biomarkers. The study is currently enrolling patients at centers in
estimated., while the MMRd cohort will be used for comparison. Tissue the United States and Europe. Clinical trial information: NCT03565991.
biopsies, longitudinal blood and stool collection will be used for discovery of
predictive molecular biomarkers and assessment of tumor evolution. Clinical
trial information: NCT03519412.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Immunotherapy and Tumor Immunobiology 145s

TPS2661 Poster Session (Board #296b), Sat, 8:00 AM-11:00 AM TPS2662 Poster Session (Board #297a), Sat, 8:00 AM-11:00 AM
A phase I, first-in-human, open label, dose-escalation and cohort expansion The IRX-2 regimen combined with nivolumab in recurrent/metastatic solid
study of MGD019, a bispecific DART protein binding PD-1 and CTLA-4 in tumors: A phase 1b study to evaluate the safety, determine recommended
patients with unresectable or metastatic neoplasms. First Author: Jason J. phase 2 dose (RP2D), and investigate the biologic and clinical activity. First
Luke, University of Chicago Comprehensive Cancer Center, Chicago, IL Author: Rohit K. Jain, H. Lee Moffitt Cancer Center & Research Institute,
Tampa, FL
Background: Immune checkpoint molecules, including CTLA-4 and PD-1,
attenuate the duration and strength of adaptive immune responses to limit Background: IRX-2 is a biologic immunotherapeutic containing a mixture of
immune-mediated tissue damage. Tumors may inhibit cellular immune cytokines including IL-2, IL-1b IL-6, IL-8, TNFa, GM-CSF, and IFN-g. It is
activation by expressing ligands that bind checkpoint molecules and inhibit derived from stimulating human PBMCs with phytohemagglutinin. Pre-
T-cell function in the tumor microenvironment. Blockade of these inhibitory clinical studies have shown that IRX-2 enhances dendritic cell maturation,
pathways is the primary mechanism of action of several novel cancer im- T cell activation, and NK cell stimulation. In a Phase 2 trial of head and neck
munotherapy agents. Combined blockade of PD-1 and CTLA-4 with two squamous cell carcinoma (HNSCC) patients, IRX-2 increased immune ac-
checkpoint inhibitors, ipilimumab and nivolumab, increases antitumor tivation in the tumor microenvironment and was correlated with greater
activity beyond either single agent alone in patients with metastatic mel- progression free survival and overall survival. Moreover, it has been shown
anoma or other malignancies. MGD019, a novel bispecific molecule that co- that the degree of lymphocyte infiltration is an important prognostic factor for
engages and coordinately inhibits both PD-1 and CTLA-4 signaling, was treatment with anti- programmed cell death protein 1 (PD-1) monoclonal
developed to potentially improve antitumor activity and/or safety relative to antibodies. These data provide a compelling rationale for incorporating IRX-
the monoclonal antibody combination. MGD019 is an Fc-bearing tetravalent 2 regimen into anti-PD-1 treatment strategies to increase the level
DART molecule (bivalent for each antigen) that can independently block of lymphocyte infiltration in the TME and improve immune response.
either checkpoint molecule, with preferential co-blockade in cells co- Methods: This is a phase Ib trial exploring safety and tolerability of IRX-2
expressing both molecules demonstrated in vitro. It is hypothesized that regimen in combination with nivolumab. Patients with recurrent or meta-
MGD019 might be clinically active in either checkpoint naı̈ve or checkpoint static renal cell carcinoma, urothelial carcinoma, non-small cell lung cancer,
experienced patients after prior PD-1/PD-L1 inhibitors. Methods: This Phase HNSCC and melanoma are eligible. Patients who have received prior anti-
1 study will characterize safety, dose limiting toxicities, and maximum PD-1/PD-L1 antibodies are eligible. IRX-2 regimen consists of cyclophos-
tolerated dose (MTD)/maximum administered dose (MAD) of MGD019. Dose phamide 300mg/m2 (Day 1) and subcutaneous IRX-2 injections in 2 or 4
Escalation will enroll patients with advanced solid tumors of any histology in lymph node-bearing regions (based on dose level) for 10 days every
sequential escalating doses in cohorts of 3 to 9 patients in a 3+3+3 design. 3 months. Two dose levels of IRX-2 regimen will be studied. Nivolumab is
Once the MTD/MAD is reached, a Cohort Expansion phase will characterize administered at 240 mg Q2W. Once the RP2D is determined, an additional
safety and initial antitumor activity per RECIST v1.1 and irRECIST in patients 88 pts will be enrolled in an expansion phase, for a planned total enrollment
with specific tumor types anticipated to be sensitive to dual checkpoint of approximately 100 pts. Dose expansion phase will include cohorts of the 5
blockade. Additional endpoints include pharmacokinetics; immunogenicity; different diseases. Each cohort will include two groups: 1) anti-PD-1/PD-L1
impact of MGD019 on various measures of immune-regulatory effects in antibody naı̈ve tumors, and 2) progressed during or after anti-PD-1/PD-L1
peripheral blood and biopsy specimens; and relationship between antitumor antibodies. The primary study objective is to determine safety and tolerability
activity and gene profiles, tumor mutational burden, and PD-1, PD-L1, and of combination therapy. Secondary objectives are to evaluate the objective
CTLA-4 expression on tumor cells and immune cell infiltrates within biopsy response rate, progression-free survival, and overall survival. Study Progress:
specimens. Patients will be followed for survival approximately every 3 months Study is actively accruing. Clinical trial information: NCT03758781.
for 2 years. Clinical trial information: NCT03761017.

TPS2663 Poster Session (Board #297b), Sat, 8:00 AM-11:00 AM TPS2664 Poster Session (Board #298a), Sat, 8:00 AM-11:00 AM
A phase Ib/IIa study of rucaparib (PARP inhibitor) combined with nivolumab Feasibility study of microbial ecosystem therapeutics (MET-4) to evaluate
in metastatic castrate-resistant prostate cancer and advanced/recurrent effects of fecal microbiome in patients on immunotherapy (MET4-IO). First
endometrial cancer. First Author: Raanan Alter, University of Chicago, Author: Tira Jing Ying Tan, Princess Margaret Cancer Centre, Toronto, ON,
Chicago, IL Canada
Background: Immune checkpoint blockade (ICB) antibodies have made a Background: Differences in microbiome diversity and composition in immune
major impact in a wide range of cancers. However, only subsets of patients checkpoint inhibitor (ICI)-responders vs non-responders have been demon-
across all malignancies benefit from ICB. In particular, metastatic castrate- strated. Transplantation of responder feces in mouse models recapitulated the
resistant prostate cancer (mCRPC) and advanced endometrial cancers (EC) ICI-responsive phenotype. MET-4 is an oral alternative to fecal transplant
have shown very limited responses to ICB. The central hypothesis of this trial consisting a well-defined mixture of intestinal bacteria isolated from healthy
is that the combination of PARP inhibitor (rucaparib) with PD-1 inhibitor donor stool sample. We hypothesize that co-administration of MET-4 with ICI
(nivolumab) will enhance ICB efficacy in mCRPC and mEC patients. Given is safe and results in alterations of the gut microbiota. Methods: Three cohorts
that PTEN loss has also been associated with poor response to ICB, a (n = 65) of subjects with any advanced solid tumor type treated with mon-
secondary hypothesis of this study is that the combination therapy will have otherapy anti- PD1/PD-L1 antibody outside of a therapeutic clinical trial will
differing efficacy based on the PTEN mutation status of the tumor. be enrolled. Group A: safety cohort of 5 subjects already on ICI will receive
Methods: This is an investigator-initiated Phase 1b/IIa clinical trial of MET-4 in addition to standard of care (SOC) ICI. If , 2 subjects report adverse
rucaparib and nivolumab singly and in combination, in mCRPC and mEC events of CTCAE grade $3 within 4 weeks at least possibly related to MET-4,
patients. Patients are randomized to one of three arms – rucaparib, nivo- this dose will be declared safe and groups B/C may start enrolling. Group B (n =
lumab, or both drugs in combination for 4 weeks. Metastatic biopsy samples 40): subjects with advanced solid tumors starting on ICI, randomized 3:1 to
are collected at baseline and after 4 weeks on treatment, after which all arms MET-4 plus SOC vs. SOC. Group C (n = 20): subjects with advanced solid
will switch to combination therapy. The primary objective is to assess fea- tumors already on ICI with first unconfirmed disease progression randomized
sibility of the combination, and to elucidate changes in immune infiltrates by 1:1 to MET-4 plus ICI continuation vs. continuing ICI. Serial stool samples will
Nanostring RNA sequencing, multiplex immunofluorescence, 3D mapping, be collected for taxonomic composition, diversity, metagenomics content and
IHC, and flow cytometry. Secondary objectives are to assess clinical re- MET-4 species abundance. We anticipate the following analyses: 16S rRNA
sponse, and correlate changes in TME with PTEN status. We have currently sequencing, shotgun metagenomics sequencing, qPCR, Nanostring nucleic
enrolled 4 patients to the study, and collected pre- and 4 week on-treatment acid detection and metabolomics profiling. Serial blood sampling for flow
biopsies. This study presents an opportunity for in-depth TME analysis that cytometry/CyTOF. Immune microenvironment of tumor specimen will be
will enable the delineation of the effects of PARP inhibition singly and in examined using immunohistochemistry. Other major inclusion criteria: will-
combination with PD-1 blockade, on immune subsets within the TME. The ingness to provide correlative samples, RECIST v1.1 measurable disease and
correlative analyses will also lead to the discovery of novel biomarkers of ECOG 0-2. Subjects unable to swallow oral medications are excluded. For the
response/resistance, and suggest additional immunooncology combinations primary objective of cumulative relative abundance and changes of ICI-
for specific molecular subsets of prostate and endometrial cancers. Clinical responsiveness associated species between baseline and day 12 MET-4,
trial information: NCT03572478. assuming a change of 0.5 standard deviation (SD) of microbial alpha di-
versity, our study will have $84% power to identify a significant difference
given a significance level at 0.05 in group B. Assuming a change of 0.9 SD of
microbial alpha diversity, we will have $83% power to identify a significant
difference in group C. Response rates and progression free survival will be
assessed per RECIST v1.1 and compared with historical data. Clinical trial
information: NCT03686202.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
146s Developmental Immunotherapy and Tumor Immunobiology

TPS2665 Poster Session (Board #298b), Sat, 8:00 AM-11:00 AM TPS2666 Poster Session (Board #299a), Sat, 8:00 AM-11:00 AM
Phase 1/1b multicenter trial of TPST-1120, a peroxisome proliferator- An open label, multicenter, phase 1b/2 study of rebastinib (DCC-2036) in
activated receptor alpha (PPARa) antagonist as a single agent (SA) or in combination with carboplatin to assess safety, tolerability, and pharmaco-
combination in patients with advanced solid tumors. First Author: Ginna kinetics in patients with advanced or metastatic solid tumors. First Author:
Laport, Tempest Therapeutics, San Francisco, CA Anthony W. Tolcher, NextOncology, San Antonio, TX
Background: Tumor cells initially favor glucose metabolism via aerobic Background: Rebastinib is a potent, orally administered, kinase switch
glycolysis. As tumors rapidly proliferate and metastasize, glucose stores are control inhibitor selectively targeting the tunica interna endothelial cell
depleted and facilitated by a hypoxic tumor microenvironment (TME) and kinase (TIE2). TIE2 is primarily expressed in endothelial cells and has
metabolic reprogramming shifts intracellular metabolism(IcM) towards fatty critical roles in angiogenesis. In addition, TIE2 is highly expressed in a
acid oxidation (FAO). Fatty acids support metabolism of suppressive immune subset of macrophages, TIE2-expressing macrophages (TEMs), which are
cells in the TME in addition to tumor growth. PPARa is a ligand-activated known to have proangiogenic, pro-metastatic, and immunosuppressive
nuclear transcription factor which regulates lipid metabolism and FAO. properties. Accumulating evidence suggests that chemotherapies, such as
TPST-1120 is a first in class, oral selective PPARa antagonist that blocks carboplatin, increase the recruitment and activity of pro-tumoral TEMs,
transcription of PPARa target genes leading to an intracellular metabolism leading to chemotherapy resistance. Taken together, investigation of
shift from FAO to glycolysis. Reduction of fatty acids in the TME leads to rebastinib in combination with a chemotherapy such as carboplatin, one of
direct killing of tumor cells dependent on FAO, skews macrophages from the most commonly used agents across different tumor types, is warranted in
immune suppressive M2 phenotype to an effector M1 phenotype and fa- advanced solid tumors. Methods: This study is an open-label, Phase 1b/2,
cilitates the cytotoxicity of immune effector cells. TPST-1120 also restores multicenter study in patients with advanced or metastatic solid tumors. The
thrombospondin-1, a known natural inhibitor of angiogenesis, to homeo- study has two parts: the first part is the 3+3 dose escalation phase designed
static levels within the TME. TPST-1120 has an IC50 of 0.04 nM with . 35 to evaluate the safety, tolerability and pharmacokinetics of 50 mg and
fold selectivity over other PPAR isoforms. Preclinical studies in multiple 100 mg rebastinib twice daily in combination with carboplatin of AUC 5 or 6
tumor models show efficacy of TPST-1120 as a SA and in combination(combo) administered once every three weeks to determine the recommended phase
with an anti-PD1 monoclonal antibody (mAb) and chemotherapy. 2 dose (RP2D). Patients who have exhausted all therapies and for whom
Methods: We have initiated a phase 1/1b multicenter, open label Dose carboplatin is considered appropriate treatment will be enrolled. The second
Escalation (DEs) and Dose Expansion (DEx) trial to evaluate TPST-1120 as a part is the dose expansion phase with three cohorts: previously treated breast
SA and in combo with nivolumab, an anti-PD1 mAb; docetaxel, a chemo- cancer, recurrent, platinum-sensitive ovarian cancer, and malignant me-
therapeutic agent and cetuximab, an anti-EGFR mAb. Objectives: 1) eval- sothelioma to evaluate the safety, tolerability, and efficacy of the RP2D. A
uate safety and tolerability of continuous dosing of TPST-1120 2) identify Simon’s two-stage design will be used in the second part and initially up to
a recommended phase 2 dose (RP2D) and 3) evaluate efficacy. Eligibility: 18 patients will be enrolled into each cohort. If more than 4 responses are
1) patients with select advanced solid tumors who have failed 1 and up to observed, then the cohort will be expanded up to 33 patients. This trial is
5 prior therapies. This phase 1/1b adaptive design has 4 DEs arms, 1 SA arm expected to enroll up to 117 patients in total, with approximately 18 patients
and 3 combination arms in which TPST-1120 is combined with nivolumab, in the first part and up to 99 patients in the second part. The study is
docetaxel or cetuximab. The RP2D of TPST-1120 to proceed to DEx will be currently open only in the US. Clinical trial information: NCT03717415.
determined by safety and biomarkers during DEs. The DEx arms have 8
histology-specific cohorts, 4 SA arms and 4 combo arms and will follow a
2-stage expansion design. Biomarker analyses include gene expression
profiling of PPARa-associated genes, tumor markers of immune modulation
and serum lipid profiling. The total sample size is up to 338 pts. This trial is
accruing at U.S sites. Clinical trial information: NCT03829436.

TPS2667 Poster Session (Board #299b), Sat, 8:00 AM-11:00 AM TPS2668 Poster Session (Board #300a), Sat, 8:00 AM-11:00 AM
A multicenter, phase II study of soluble LAG-3 (Eftilagimod alpha) in A multicenter, open label, first-in-human study of an oncolytic viral vector
combination with pembrolizumab (TACTI-002) in patients with advanced expressing an agonistic anti-CD40 antibody (NG-350A) in patients with
non-small cell lung cancer (NSCLC) or head and neck squamous cell epithelial tumors (FORTITUDE). First Author: Aung Naing, University of
carcinoma (HNSCC). First Author: Julio Antonio Peguero, Oncology Con- Texas MD Anderson Cancer Center, Houston, TX
sultants PA, Department of Research, Houston, TX
Background: NG-350A is a transgene modified variant of the oncolytic
Background: Eftilagimod alpha (efti, IMP321) is a recombinant LAG-3Ig platform virus enadenotucirev (EnAd) which expresses a fully human agonist
fusion protein that binds to MHC class II and mediates antigen-presenting anti-cluster of differentiation 40 (anti-CD40) antibody. The principal ad-
cell (APC) activation followed by CD8 T-cell activation. Pembrolizumab vantage of encoding anti-CD40 within an oncolytic virus is the ability to
binds to the PD-1 receptor, blocking both immune-suppressing ligands, PD- potentially achieve very high levels within the tumor coupled with direct
L1 and PD-L2, from interacting with PD-1 to help restore effector T-cell cytotoxicity due to viral lysis and stimulation of the immune-system. NG-
responses. The rationale to combine efti and pembrolizumab comes from 350A infects and selectively replicates in tumor cells. The anti-CD40 an-
their complementary mechanisms of action. Efti activates APCs and lead to tibodies are expected to activate the patient’s own dendritic cells, macro-
an increase in activated T cells which effect potentially reduces the number phages and B-cells to drive CD4+ and CD8+ T-cell immuno-inflammatory
of non-responders to pembrolizumab. Combining an APC activator like efti to responses and immune mediated tumor cell killing. EnAd is a tumor-
pembrolizumab is therefore fundamentally different from many other trials selective chimeric Ad11/Ad3 group B oncolytic adenovirus developed us-
combining two checkpoint inhibitors like an anti-LAG-3 mAb with an anti- ing directed evolution. Phase I clinical studies have identified a well-
PD-1 mAb. In a previous phase I study (NCT 02676869) in metastatic tolerated systemic dose and regimen for EnAd monotherapy. EnAd
melanoma the combination was found to be safe and well tolerable with shows a high level of selective replication and cell killing for a broad range of
encouraging signs of clinical activity. Methods: In the course of this mul- carcinoma cell lines (of epithelial origin) with little replication in normal and
ticenter, open label, Phase II study, patients will be recruited for each of non-carcinoma cells. Methods: This first in human study will evaluate the
three indications: A: 1st line, PD-X (PD-1 or PD-L1) naı̈ve non-small cell lung safety, tolerability and preliminary efficacy of NG-350A together with virus
cancer (NSCLC); B: 2nd line, PD-X refractory NSCLC; C: 2nd line PD-X naive kinetics, immunogenicity and other pharmacodynamic effects to elucidate
head and neck squamous cell carcinoma (HNSCC). The study is designed the mechanism of action of NG-350A in patients with advanced or meta-
according to Simon’s optimal two-stage design, with objective response rate static epithelial tumours. In the dose escalation phase up to 33 patients
acc to iRECIST as primary endpoint. Secondary endpoints include pro- evaluable for dose-limiting toxicity will receive NG-350A by IV infusion on
gression free survival and overall survival. In case there are more responses Day 1, 3 and 5 at 6 US sites. The first IV cohort in the dose-escalation phase
achieved than a predefined threshold (each part counted separately) in pts will utilize the conventional ‘3+3’ design; thereafter dose recommendations
recruited in the initial stage (n = 58), additional pts (51) will be recruited in will be based on a continual reassessment method. Following determination
stage 2. Efti will be administered for a maximum of 18 cycles (1 cycle = of the recommended phase 2 dose up to 20 patients will be treated in a dose-
3 weeks) as 30 mg subcutaneous injection every 2 weeks for the first 8 and expansion cohort. In a parallel cohort, up to 12 patients will receive a single
every 3 weeks for the 10 following cycles. Pembrolizumab (200 mg in- dose of NG-350A by intratumoural (IT) injection on Day 1 for direct delivery
travenous infusion every 3 weeks) is administered in parallel for up to 35 of high viral titres to tumor. Up to six patients are planned to undergo surgical
cycles. Patients are followed up for progression and survival. Clinical trial resection of a tumor lesion to optimize translational research. Clinical
information: 03625323. conduct of the study was initiated in February 2019.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Immunotherapy and Tumor Immunobiology 147s

TPS2669 Poster Session (Board #300b), Sat, 8:00 AM-11:00 AM TPS2670 Poster Session (Board #301a), Sat, 8:00 AM-11:00 AM
Phase 1 trial of TLR9 agonist lefitolimod in combination with CTLA-4 A phase I/II study of live biotherapeutic MRx0518 in combination with
checkpoint inhibitor ipilimumab in advanced tumors. First Author: Matthew pembrolizumab in patients who have progressed on prior anti-PD-1 therapy.
Reilley, University of Virginia, Charlottesville, VA First Author: Shubham Pant, University of Texas MD Anderson Cancer
Center, Houston, TX
Background: Drugs targeting pathogen associated molecular patterns are an
attractive strategy to stimulate the immune system. Toll-like receptors (TLR) Background: The gut microbiome has emerged as a new therapeutic target to
have generated significant interest as an effective means of stimulating the augment the efficacy of immune checkpoint blockade. MRx0518 is a novel,
immune system that result in the killing of tumor cells. TLR-9 agonists can gut microbiome-derived, oral live biotherapeutic, designed to induce a broad
function to promote an early immune response and are an appealing partner immunostimulatory response to re-engage PD-1 inhibitor activity. Preclinical
for combination with checkpoint blockade to improve immune activation. studies showed that MRx0518 reduced tumour growth in models of kidney,
Lefitolimod (MGN1703) is a covalently closed dumbbell-shaped DNA lung and breast cancer. MRx0518 increased CD4 and CD8 T cell and NK cell
molecule that functions as a TLR-9 agonist. We developed a clinical trial infiltration into the tumour and decreased Tregs. Upregulation of tumour
combining lefitolimod with ipilimumab (anti-CTLA4) in patients with ad- TLR5 was observed and linked to the bacterial flagellin moiety, which was
vanced malignancies. In the dose-escalation phase 19 patients were en- shown to strongly induce NFkB, cytokine responses and IFNg+ CD4 and CD8
rolled and no DLTs were encountered. Safety data and maximum planned T cells. The study, one of the first oncology trials conducted with live bio-
dose level of lefitolimod at 120mg weekly and ipilimumab 3mg/kg every therapeutics, is a single center, open label, safety and preliminary efficacy
3 weeks was previously presented. Adverse events related to the combination study of MRx0518 in combination with pembrolizumab in patients with solid
included fatigue, appetite loss, rash, and anemia. Methods: This trial tumors who have progressed on PD-1 inhibitors. Methods: Trial consists of 2
(NCT02668770) was designed to evaluate the safety profile and maximum parts. In Part A, 12 patients receive pembrolizumab 200 mg every 3 weeks
tolerated dose of lefitolimod with ipilimumab. A 3+3 trial design was used to plus 1 capsule (bid) of MRx0518 with a DLT period of 1 cycle (21 days). In
establish safety of the combination at each dose level and guide the decision Part B, up to 30 patients per cohort (NSCLC, Urothelial, Renal and Mela-
to escalate dose. Lefitolimod is administered via subcutaneous (SC) in- noma) will receive pembrolizumab 200 mg every 3 weeks plus 1 capsule
jection weekly while ipilimumab is given at 3mg/kg intravenous on day 8 of (bid) of MRx0518 for up to 35 cycles or until disease progression per RECIST
each 3 week cycle. Lefitolimod starting dose was 15mg SC weekly with 3 1.1. The primary end points are safety and tolerability of MRx0518 in
dose level escalations up to 120mg SC weekly. Patients receive treatment for combination with pembrolizumab (Parts A and B) and clinical benefit of
4 cycles (total 12 weeks) with the combination, and those with stable disease MRx0518 in combination with pembrolizumab (Part B). Secondary end
or response were eligible to remain on lefitolimod therapy for up to 1 year. points are objective response rate, duration of response, disease control rate,
Eligible patients have a metastatic or unresectable solid tumor refractory to and progression-free survival. Exploratory end points include biomarkers of
standard therapies, ECOG # 2, and normal organ and bone marrow function. treatment effect, effect on microbiota and overall survival. Recruitment is
Patients are allowed to have received prior checkpoint blockade agents. ongoing. Clinical trial information: NCT03637803.
Enrollment in expansion cohorts in ongoing. To better understand relevant
immunologic changes associated with treatment, paired pre- and post-
treatment biopsies of target lesions and peripheral blood collection dur-
ing treatment is required for target expansion cohort patient populations. In
addition to evaluating target patient populations at the combination dose
established during escalation, an expansion cohort for patients with cuta-
neous metastases involves combination treatment with intratumoral delivery
of lefitolimod. Clinical trial information: NCT02668770.

TPS2671 Poster Session (Board #301b), Sat, 8:00 AM-11:00 AM

An open label, multicenter, phase I/II study of RP1 as a single agent and in
combination with PD1 blockade in patients with solid tumors. First Author:
Mark R. Middleton, Churchill Hospital, Oxford, United Kingdom
Background: RP1 is an attenuated oncolytic HSV-1 that expresses a fuso-
genic glycoprotein from gibbon ape leukemia virus (GALV-GP R-) and GM-
CSF. RP1 induces potent GALV-GP R- enhanced immunogenic cell death
and host anti-tumor immunity in murine tumor models and increases PD-L1
expression. This clinical trial (NCT03767348) was designed to test the
hypotheses that RP1 is safe when given alone and together with nivolumab
(phase 1) and has efficacy together with nivolumab in four tumor types
(phase 2). Methods: The primary goals of this clinical trial in a total of ~150
patients are to define the safety profile of RP1 alone and together with
nivolumab, determine the recommended phase 2 dose (phase 1), and then
in four phase 2 cohorts, to determine objective response rate in patients with
melanoma, non-melanoma skin cancer, urothelial carcinoma and MSI-H
solid tumors. Secondary objectives include duration of response, CR rate,
PFS, viral shedding, and immune biomarker analysis. Patients with ad-
vanced cancer who failed prior therapy were eligible for the phase I com-
ponent. In Phase 2 patients with histologic diagnoses of the four tumor types
(N=30 for each) and who meet safety criteria for nivolumab treatment are
eligible. Prior treatment with checkpoint blockade is not allowed except for
the melanoma cohort. In the phase 1 portion patients are treated by intra-
patient dose escalation of virus (range, 104 - 108 PFU) by intratumoral
injection every two weeks for 5 total doses followed by 12 patients dosed 8
times at the RP2D in combination with nivolumab. Phase 1 patients were
divided into two groups based on presence of clinically accessible lesions
amenable to direct injection or those with visceral/deep lesions requiring
image guidance for injection. In the phase 2 portion patients will receive the
RP2D for eight injections and nivolumab will be given starting with the
second RP1 injection. For the phase 1 portion, a modified 3+3 dose es-
calation design is used to assess safety and in the phase 2 portion, statistical
analysis will be performed using a two-stage three-outcome optimum design
with objective responses determined by RECIST criteria. As of February 11,
2019, 27 patients have been enrolled. Clinical trial information:
NCT03767348.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
148s Developmental Therapeutics and Tumor Biology (Nonimmuno)

3000 Oral Abstract Session, Mon, 8:00 AM-11:00 AM 3001 Oral Abstract Session, Mon, 8:00 AM-11:00 AM
Belvarafenib, a novel pan-RAF inhibitor, in solid tumor patients harboring A phase I dose escalation (DE) study of ERK inhibitor, LY3214996, in
BRAF, KRAS, or NRAS mutations: Phase I study. First Author: Tae Won Kim, advanced (adv) cancer (CA) patients (pts). First Author: Shubham Pant,
Department of Oncology, Asan Medical Center, University of Ulsan College of University of Texas MD Anderson Cancer Center, Houston, TX
Medicine, Seoul, South Korea
Background: LY3214996 is a selective and potent ERK1/2 inhibitor that has
Background: Belvarafenib (HM95573/GDC-5573) is an oral type II pan-RAF demonstrated tumor growth inhibition in several pre-clinical tumor models
kinase inhibitor which demonstrates selective anti-tumor activity in several with BRAF, RAS, or MAP2K1 mutations. This is the first-in-human Phase 1
non-clinical cancer models and in cancer patients with RAS- or RAF- mutation. Study of LY3214996 in adv CA pts. Methods: The goals of this DE study were
Here we present overall safety and efficacy findings of two phase 1 studies, to determine a recommended Phase 2 dose (RP2D), safety, pharmacokinetic
consisting of dose escalation and dose expansion stages. Methods: Patients (PK), and preliminary efficacy of LY3214996 (NCT02857270; I8S-MC-
with advanced solid cancers harboring documented RAS- or RAF- mutation JUAB; Eli Lilly & Co.). Pts with adv CA, $18 yrs of age, ECOG #1, and with
were enrolled. In the dose escalation study, patients were treated with Bel- adequate organ function were eligible. Pharmacodynamic (PD) biomarkers
varafenib at a starting dose of 50 mg once daily (QD) to 800 mg BID to assess including pRSK were evaluated in blood and paired tumor tissue. The DE
safety and tolerability and identify the recommended dose (RD). Dose esca- phase evaluated PO doses using the Bayesian model-based toxicity band
lation was guided based on pharmacokinetic data and used a traditional 3+3 method. Results: A total of 51 pts with median age of 62 yrs (range: 21-81)
design. The dose expansion study was comprised of 6 cohorts (according to the received at least 1 dose of LY3214996 with a median of 3 cycles (range: 1-
type of tumor and RAS- or RAF gene mutation) and patients received the RD of 12). Most pts had a mutation in RAS (N = 33) or BRAF (N = 16) and had a
Belvarafenib. The primary objective was to explore anti-tumor activity (per median of 4 prior lines of treatment. The DLTs observed in the study include
RECIST 1.1) and pharmacodynamic effects. Results: The dose escalation grade (G) 3 cough and fatigue, G3 dehydration, increased creatinine (Cr), G3
study included 72 patients in 9 dose cohorts (cut-off date of 18 Jan 2017). increased CPK, G3 rash . 7 days, and 1 pt with renal failure. TRAEs to
Dose dependent increase in exposures observed up to 650 mg BID. The most LY3214996 occurring in $10% of pts included nausea, vomiting, diarrhea,
common treatment-emergent adverse events that occurred in more than 20% dermatitis acneiform, fatigue, pruritus, and blurred vision. LY3214996
of patients were rash, dermatitis acneiform and pyrexia. A total of 4 DLTs exposures increased with dose. Tumor regression was observed in 7 pts with
(different kinds of rashes) were reported and included 2 DLTs at the 800 mg BRAF/non-BRAF mutant CA including 5 pts who failed prior IO/MAPK in-
BID level. Therefore, 650 mg BID was considered the MTD and 450 mg BID hibitors. Four pts achieved stable disease (2 BRAF, 1 RAS and 1 CRAF
was identified as the RD for Belvarafenib. There were 7 partial responses (3 mutation) that lasted . 4 mos. Up to 100% pRSK decrease from baseline in
confirmed PRs) from 200 mg QD to 800 mg BID in NRAS-mutant melanoma, tumor was observed. Conclusions: LY3214996 had an acceptable safety
BRAF-mutant melanoma, KRAS-mutant sarcoma, and BRAF-mutant GIST. profile, favorable PK, and potent tumor PD inhibition at RP2D. This supports
Four of nine patients with NRAS-mutant melanoma had a PR (ORR 44%). The further exploration of LY3214996 as monotherapy and in combination in CA
dose expansion study included 63 patients in 5 indication-specific and basket pts with activating MAPK pathway alterations. Clinical trial information:
cohorts administered with 450 mg BID Belvarafenib (cut-off date of 6 Oct NCT02857270.
2018). No new safety signal was detected. There were two PRs each in patients
with NRAS-mutant melanoma (2/9), BRAF-mutant melanoma (2/6) and
BRAF-mutant CRC (2/7), respectively. Conclusions: Belvarafenib was well
tolerated and exhibited anti-tumor activity in patients with advanced solid
tumors harboring RAS or RAF mutations. Belvarafenib is being further in-
vestigated in combination with the MEK inhibitor cobimetinib. Clinical trial
information: NCT02405065, NCT03118817.

3002 Oral Abstract Session, Mon, 8:00 AM-11:00 AM 3003 Oral Abstract Session, Mon, 8:00 AM-11:00 AM
Dabrafenib and trametinib in patients with tumors with BRAF V600E/K Phase 1 study evaluating the safety, tolerability, pharmacokinetics (PK), and
mutations: Results from the molecular analysis for therapy choice (MATCH) efficacy of AMG 510, a novel small molecule KRASG12C inhibitor, in
Arm H. First Author: April K.S. Salama, Duke University, Durham, NC advanced solid tumors. First Author: Marwan Fakih, The Judy and Bernard
Briskin Center for Clinical Research, City of Hope, Duarte, CA
Background: The NCI-MATCH precision medicine trial assigns patients (pts)
with solid tumors, lymphomas, or multiple myeloma with progression on prior Background: The KRASG12C mutation is found in approximately 13% of lung
treatment to a targeted therapy based on genetic alterations identified in pre- adenocarcinomas and 1–3% of other solid tumors, but there is no approved
treatment biopsies. Arm H (EAY131-H) evaluated the combination of the therapy that targets this mutation. AMG 510 is a novel small molecule that
BRAF inhibitor (inh) dabrafenib (DAB), and the MEK inh, trametinib (TRM), specifically and irreversibly inhibits KRASG12C by locking it in an inactive
in pts with BRAF V600E/K mutations. Methods: Pts with melanoma, thyroid, GDP-bound state. Methods: This phase 1, first-in-human, open-label,
or colorectal cancer were excluded. Pts with NSCLC were excluded after the multicenter study (NCT03600883) is evaluating the safety, tolerability,
U.S. Food and Drug Administration (FDA) approved DAB/TRM for this in- PK, and efficacy of AMG 510 in adult patients (pts) with locally-advanced or
dication. Pts received DAB 150 mg po BID and TRM 2 mg PO daily on 28 day metastatic KRASG12C mutant solid tumors. The primary endpoint is safety;
cycles until disease progression or intolerable toxicity; restaging was per- key secondary endpoints include PK, ORR (assessed every 6 weeks [wks]),
formed every 2 cycles. The primary endpoint was objective response rate DOR, and PFS. Key inclusion criteria: KRASG12C mutation identified through
(ORR); secondary endpoints included progression-free survival (PFS), 6- DNA sequencing, measurable or evaluable disease, ECOG PS #2, life
month PFS, and overall survival (OS). Results: A total of 35 pts were enrolled expectancy .3 months (mo). Key exclusion criteria: active brain metastases,
from 1/2016-2/2018; 2 were ineligible (CrCl below criteria; labs out of myocardial infarction within 6 mo. A dose exploration will determine the
window). Over 17 distinct tumor histologies were represented. 58% of pts maximum tolerated dose (MTD) or recommended phase 2 dose (RP2D). A
were female, median age was 63 (range 21-85), 94% were Caucasian, and dose expansion will enroll pts with NSCLC, CRC, and other advanced solid
48% of pts had received at least 3 prior therapies (range 1- 8). The confirmed tumors carrying the KRASG12C mutation. AMG 510 will be given PO until
ORR was 33.3% (90% CI 19.9%, 49.1%), with a median duration of re- disease progression, intolerance, or withdrawal of consent. Results: 22 pts
sponse (DoR) of 12 months (mon). Varied histologies had a DoR of . 12 (8 men, 14 women; median age 55.5 y) were enrolled in the first 3 dose
mon: histiocytic sarcoma, cholangiocarcinoma and mixed adenoneur- cohorts. Tumor types: 6 NSCLC, 15 CRC, 1 other. Most pts (n=17) had $3
oendocrine carcinoma of unknown primary, among others. Median PFS was prior lines of treatment (tx). Median tx duration was 28 d (range: 8–134).
9.4 mon; the 6 mon PFS rate was 70.6% (90% CI 58.2%-85.5%), and an 5 pts reported 10 treatment-related AEs (grade 1, n=9; grade 2, n=1); there
additional 10 pts had a PFS . 5.5 mon. Median OS has not been reached. At were no DLTs. Tumor response was evaluated in 9 pts (4 with $2 assess-
the time of data cutoff (12/2018) 11 pts continue on treatment. Adverse ments); 13 pts have not reached their first assessment.1 pt had a PR (NSCLC
events (AE) were comparable to previously reported profiles of DAB/TRM; no at wks 6 and 12, tx ongoing), 6 pts had SD (4 CRC and 2 NSCLC; median tx
new AEs were identified. The most frequent grade 3 AEs were fatigue, duration 9.7 wks [range: 6.3–19.1], tx ongoing), 2 pts had PD. 20 pts are
neutropenia, hyponatremia, hypophosphatemia, and urinary tract infection; continuing to receive AMG 510. A second PR (NSCLC at wk 6, tx ongoing)
there was 1 grade 4 sepsis; no grade 5 AEs. Conclusions: In this pre-treated, was reported after data cutoff. Conclusions: AMG 510 has been well tol-
mixed histology cohort, DAB and TRM showed promising activity outside of erated at the dose levels tested and has shown antitumor activity when
currently approved FDA indications warranting further investigations. Cor- administered as monotherapy to patients with advanced KRASG12C mutant
relative analyses are planned. Clinical trial information: NCT02465060. solid tumors. MTD has not been determined, and enrollment into the dose
exploration is ongoing. Clinical trial information: NCT03600883.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Therapeutics and Tumor Biology (Nonimmuno) 149s

3004 Oral Abstract Session, Mon, 8:00 AM-11:00 AM 3005 Oral Abstract Session, Mon, 8:00 AM-11:00 AM
First-in-human phase 1 study of ABBV-085, an antibody-drug conjugate Single agent ONC201 in adult recurrent H3 K27M-mutant glioma. First
(ADC) targeting LRRC15, in sarcomas and other advanced solid tumors. First Author: Isabel Arrillaga, Massachusetts General Hospital, Boston, MA
Author: George D. Demetri, Dana-Farber Cancer Institute and Ludwig Center
Background: H3 K27M-mutant glioma is associated with a poor prognosis
at Harvard Medical School, Boston, MA
and there is no effective therapy following radiation. We report the clinical
Background: ABBV-085 is an ADC (conjugated to monomethyl auristatin E, experience with single agent ONC201, the first small molecule DRD2 an-
drug:antibody ratio of 2:1) directed against leucine-rich repeat containing tagonist in oncology, in adults with recurrent H3 K27M-mutant glioma.
15 (LRRC15), a type 1 transmembrane protein highly expressed on the Methods: Twenty-nine adult patients with recurrent H3 K27M-mutant gli-
surface of sarcomas and cancer-associated fibroblasts in stroma of many oma have been treated with single agent ONC201 as of January 20, 2019:
other cancers. ABBV-085 induced antitumor activity in both in vitro and 19 patients on NCT03295396; 8 patients on NCT02525692; 2 patients on
xenograft models of sarcoma. This phase 1, first-in-human, 2-part study compassionate use protocols under the Sponsor’s IND. Median age was 57
assessed the safety/tolerability of ABBV-085 in patients (pts) with advanced years old (range: 17-74), median prior lines of therapy was 2 (range: 1-4) and
solid tumors (NCT02565758). Methods: Eligible pts ($18 yr; advanced all patients received prior radiation (median 8.5 months from radiation
solid tumors) received ABBV-085 intravenously in a 3+3 dose-escalation completion to ONC201 initiation). ONC201 was orally administered at
(DE) design; 0.3- to 4.8-mg/kg doses every 2 wk (8 cohorts). Pharmaco- 625 mg weekly, except for one patient dosed once every 3 weeks. Results: As
kinetics (PK) were assessed in cycle 1 and cycle 3. Results: As of Dec 2018, of February 5, 2019, 13 of 29 patients remain on-trial within median follow
78 pts were enrolled in monotherapy DE and dose-expansion (EXP) cohorts up of 6.5 months (range: 0.6-33.6), 8 patients are alive but off-trial with
(#2.7 mg/kg, n = 21; 3.6 mg/kg, n = 45; 4.2 mg/kg, n = 6; 4.8 mg/kg, n = 6); median follow up of 2.4 months (range: 0.2-9), and 8 patients have expired.
median age: 58 yr (range 21–84); median treatment (Tx) duration: 6.2 wk Nine of 29 patients (31%) remain progression-free on ONC201 with a
(range 0.3–54.4). Overall, 77 (98.7%) pts reported $1 Tx-emergent adverse median follow up of 6.5 months (range 0.6-33.6). No dose-limiting toxicities
events (TEAEs). Fatigue (48.7%) was most common; 19 (24.4%) pts re- or treatment discontinuations due to toxicity occurred. Three patients have
ported grade 1/2 blurred vision (reversible on study discontinuation). experienced durable partial responses by RANO (4.3-28.5 months). In
Grade $3 TEAEs were reported in 56 (71.8%) pts; anemia (14.1%) was the addition, one patient experienced complete regression that continues for
most common. Dose-limiting toxicities occurred at 3.6 mg/kg (n = 1; . 14 months of all , 1 cm tumor lesions that are not measurable by RANO.
anemia), 4.2 mg/kg (n = 1; hypertriglyceridemia), and 4.8 mg/kg (n = 2; ileus Furthermore, 10 patients had a best response of stable disease by RANO,
and nausea); 3.6 mg/kg was chosen as the recommended phase 1b dose 12 patients experienced progressive disease, and 3 patients are not yet
(RP1bD). PK exhibited dose-proportional increase in the area under the evaluable. Among the patients with a best response of stable disease by
curve after single-dose administration; half-life was 2.84 days at the RP1bD. RANO, one patient had . 50% tumor regression in the basal ganglia that did
Of the 27 sarcoma pts (DE [n = 8]/EXP [n = 19] cohorts; undifferentiated not qualify as a partial response by RANO due to a new lesion on a con-
pleomorphic sarcoma [n = 10], osteosarcoma [n = 10], and other sarcomas firmatory scan. Another patient with stable disease by RANO has had 37%
[n = 7]) treated at the RP1bD, 4 (14.8%) had confirmed partial response tumor regression so far in the brainstem and remains on-treatment for
(PR; 2 [7.4%] unconfirmed), 8 (29.6%) had stable disease, 11 (40.7%) had 6 months. All tumor regressions remain durable to date and some were
progressive disease; 2 (7.4%) were not evaluable. The median duration of associated with improvements in disease-associated neurological symp-
response (confirmed responders) was 7.6 mo (95% CI: 5.6–9.2). Updated toms. Conclusions: Single agent ONC201 is well tolerated and clinically
safety and efficacy data will be reported. Conclusions: ABBV-085 was well active in adult recurrent H3 K27M-mutant glioma patients. Clinical trial
tolerated with durable PR observed in pts with advanced sarcomas. Clinical information: NCT03295396; NCT02525692.
trial information: NCT02565758.

3006 Oral Abstract Session, Mon, 8:00 AM-11:00 AM 3007 Oral Abstract Session, Mon, 8:00 AM-11:00 AM
Talazoparib beyond BRCA: A phase II trial of talazoparib monotherapy in First-in-human trial of the oral ataxia telangiectasia and Rad3-related (ATR)
BRCA1 and BRCA2 wild-type patients with advanced HER2-negative breast inhibitor BAY 1895344 in patients (pts) with advanced solid tumors. First
cancer or other solid tumors with a mutation in homologous recombination Author: Johann S. De Bono, Royal Marsden NHS Foundation Trust and The
(HR) pathway genes. First Author: Joshua James Gruber, Stanford University Institute of Cancer Research, London, United Kingdom
School of Medicine, Stanford, CA
Background: The ATR kinase is a key regulator of the DNA damage response
Background: Talazoparib, a PARP inhibitor, is active in germline BRCA1/2 (DDR) machinery, activated by DNA damage and replication stress. BAY
mutant advanced HER2-negative breast cancer, but its activity beyond 1895344 is a novel, potent, and selective ATR inhibitor with anti-tumor
BRCA1/2 is unknown. We conducted a single institution phase II trial to activity in preclinical models with DDR defects. Methods: Pts with advanced
evaluate talazoparib in patients (pts) with advanced HER2-negative breast metastatic solid tumors resistant or refractory to standard treatment, with
cancer or other solid tumors with a germline (g) or somatic (s) alteration in HR and without DDR defects, received BAY 1895344 BID, 3 days (d) on/4 d off
pathway genes not including BRCA1/2. Methods: Eligible pts had mea- continuously in 3-weekly cycles. Results: As of December 20, 2018, 18 pts
surable disease, lacked a germline or somatic mutation in BRCA1/2, re- with colorectal (4), breast (3), prostate (2), and ovarian (2) cancers were
ceived at least one prior therapy for advanced HER2-negative breast cancer enrolled across 6 cohorts (5 mg, 10 mg, 20 mg, 40 mg, 60 mg, and 80 mg
or other solid tumor and had a HR pathway gene mutation: PALB2, CHEK2, BID). Median prior lines of treatment was 5. No dose-limiting toxicities
ATM, NBN, BARD1, BRIP1, RAD50, RAD51C, RAD51D, MRE11, ATR, (DLTs) were reported in the 5-40 mg cohorts. 2/3 pts had DLTs in the 80 mg
PTEN, FANCA, FANCC, FANCD2, FANCE, FANCF, FANCG, FANCL. Pts with cohort (grade [G] 4 neutropenia, G4 neutropenia and G4 thrombocytopenia)
no progression on or within 8 weeks of their last platinum dose were eligible. and 2/7 had DLTs in the 60 mg cohort (G4 neutropenia, G2 fatigue). 40 mg
Pts were treated with talazoparib 1 mg po daily until disease progression. BID 3 on/4 off was defined as the maximum tolerated dose. Most common
Response was assessed every 8 +/- 1 weeks. If 2 or more responses were treatment-emergent adverse events included anemia, neutropenia, nausea,
observed in 10 pts in stage I, the study would proceed to stage II and enroll and fatigue. Pharmacokinetics appeared dose proportional. Pharmacody-
10 additional pts. The null hypothesis of a # 5% objective response rate namic analyses showed modulation of pH2AX and/or pKAP1 in paired tumor
would be rejected if at least 3 of 20 respond. Results: Twenty pts were biopsies at exposures associated with preclinical anti-tumor activity. In 13
enrolled; 13 breast cancer (12 HR+/HER2-, 1 TNBC) and 7 non-breast pts with and without DDR defects treated at dose levels $40 mg BID, the
cancer (pancreas, colon, uterine, testicular, parotid salivary). Median age objective response rate was 30.7%, including 2/2 pts at 40 mg (appendix
was 54 years. Of 12 response evaluable pts with breast cancer, 3 had a and urothelial cancer), 1/8 pts at 60 mg (breast), and 1/3 pts at 80 mg
RECIST response (ORR = 25%, 2 gPALB2, 1 gCHEK2/gFANCA/sPTEN) and (endometrial). All responders had ATM protein loss of expression and/or ATM
3 additional pts (gPALB2, sATR, sPTEN) had SD $ 6 months (CBR = 50%). mutation; median treatment duration was 347 d (range 293-364 d). A
No responses were seen in non-breast tumors; 2 (gCHEK2 testicular, gATM BRCA1-mutant, olaparib-resistant ovarian cancer pt (60 mg) had a CA125
colon) had SD $ 6 months. Talazoparib was well tolerated; 5 patients re- response and stable disease .10 months. 41 additional pts have been
quired dose reduction for hematologic toxicity. Results of tumor HR deficiency enrolled in ongoing expansion cohorts in cancers with DDR defects (prostate,
status assessment from metastatic biopsies and serial ctDNA profiling will breast, gynecologic, colorectal) or ATM protein loss (all comers) with re-
be presented. Conclusions: In this proof-of-concept phase II study, single sponses observed. Conclusions: The ATR inhibitor BAY 1895344 is toler-
agent talazoparib demonstrated activity in HER2-negative advanced breast ated at biologically active doses with anti-tumor activity against cancers with
cancer pts with a HR pathway mutation beyond BRCA1/2. Further evaluation certain DDR defects, including ATM protein loss. Clinical trial information:
of talazoparib in this population is warranted. Clinical trial information: NCT03188965.
NCT02401347.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
150s Developmental Therapeutics and Tumor Biology (Nonimmuno)

3008 Oral Abstract Session, Mon, 8:00 AM-11:00 AM 3009 Poster Discussion Session; Displayed in Poster Session (Board #1),
Phase 1/2 trial of FF-10502-01, a pyrimidine antimetabolite, in patients Sat, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
with advanced cholangiocarcinoma and solid tumors. First Author: Filip Sat, 3:00 PM-4:30 PM
Janku, The University of Texas MD Anderson Cancer Center, Houston, TX A phase I study of mirvetuximab soravtansine (IMGN853) and gemcitabine
(G) in patients with FOLR1-positive recurrent epithelial ovarian (EOC),
Background: FF10502 is a synthetic pyrimidine nucleoside similar to gemci-
endometrial cancer (EC), or triple-negative breast cancer (TNBC). First
tabine (gem) with a sulfur in the pentose ring. FF10502 is a more potent in-
Author: Mihaela C. Cristea, City of Hope, Duarte, CA
hibitor of DNA polymerase Beta than gem with activity in gem resistant patient
(pt) derived xenograft models. FF10502 is avidly taken up into DNA and has Background: IMGN853 is an antibody-drug conjugate targeting the folate
greater activity against quiescent cells than gem. Methods: Pts . 18 years old receptor alpha (FOLR1), linked to maytansinoid, DM4. IMGN853 has
with advanced disease who had progressed on standard of care were enrolled into promising single agent activity in FOLR1+ EOC. The recommended phase 2
9 dose levels to determine maximum tolerated dose (MTD) and dose limiting dose (RP2D) is 6 mg/kg, based on adjusted ideal body weight (AIBW) IV every 3
toxicities (DLTs) and subsequently into two expansion cohorts: biliary or solid wks. This study evaluates IMGN853 and G. Methods: Patients (pts) with
tumors (ST). FF10502 at doses of 8 to 135 mg/m2 was administered iv on days FOLR1+ tumors including: platinum resistant EOC with #4 prior chemo-
1, 8, 15 of a 28-day cycle until progressive disease or toxicity. PK/PD evaluations therapy (CT) regimens, EC with #2 CT and TNBC #4 CT are eligible. FOLR1 +
were performed on all pts. Response was assessed by RECIST 1.1. Results: 76 is defined as $25% of tumor staining (all tumors) $2+ intensity (EOC, EC)
pts were treated; 35 pts in dose escalation, including 7 cholangiocarcinoma pts. and $1+ (TNBC). A standard 3+3 design combines IMGN853 and G. EOC pts
MTD was 90 mg/m2. DLTs included 2 pts with hypotension at 135mg/m2 (G3 undergo one research biopsy to assess intratumoral vs. circulating DM4 level
and G4) and 1 pt each with G3 fatigue and G2 rash at 100mg/m2. In expansion, and biopsy vs. archival tissue FOLR1 expression. Dose-limiting toxicity (DLT) is
19 cholangiocarcinoma, 3 gallbladder and 19 other pts (13 pancreatic, 2 assessed during cycle 1. Responses as per RECIST 1.1 are assessed at 12 wks.
urothelial, and 1 each ovarian, prostate, NSCLC, SCCHN each) were treated. 1 pt and adverse events (AEs) are evaluated by CTCAE v4.0. Results: From 10/
with prior rituximab for ITP developed PML. G3 treatment-related low platelets 2017 to 1/2019 a total of 15 pts. were treated (3 additional pts have con-
occurred in 3 pts at 90mg/m2 after cycle 1. There were 5 partial responses sented on dose level [DL] 4):10 EOC, 3 EC and 2 TNBC. One pt. on DL1 had
(PRs), including 4 pts who had progressed on prior gemcitabine: 3 of 26 pts with grade (G) 4 thrombocytopenia (PLT) DLT. Three pts were inevaluable for DLT
cholangiocarcinoma, 1 urothelial carcinoma and 1 chondroblastic osteosar- and were replaced: 1 pt. at DL1 with G4 neutropenia without fever of unknown
coma. 7 cholangiocarcinoma pts stayed on therapy for $6 months. FF10502 duration due to delayed follow up blood work, 1 pt. at DL2 and 1 pt. at DL3 due
incorporation intoperipheral blood cellular DNA was seen, andbiomarker- to incomplete cycle 1. No DLTs were observed on DL2-3. Day 8 cycle 1 dose
analysisdata to identify pts with higher potential for clinical response will be modifications were required in 3 of 4 patients on DL3 (for mucositis [1 pt.], and
presented. Conclusions: FF10502 is well tolerated in pts with advanced cancers PLT [2 pts]). We are now enrolling at the RP2D for both agents, and MTD will be
refractory to standard therapies. Early signals of efficacy warranting further determined prior to May 2019. Conclusions: IMGN853 in combination with G
exploration were seen in heavily pretreated cholangiocarcinoma pts (median: 4 is achievable at clinically relevant doses and the recommended Phase 2 dose
prior therapies). Patient selection based on differential effects of FF10502 on and MTD will be reported. Support: NCCN grant; with support from Immu-
DNA polymerases will be explored. Clinical trial information: NCT02661542. noGen Corp and Cancer Center Support Grant P30CA033572. Clinical trial
information: NCT02996825.
IMGN853 Gemcitabine
DL (mg/kg, AIBW) D1q3W) (mg/m2, D1D8q3W) Evaluable/Treated DLTs Responses
1 5 mg/kg IV, day 1 600 mg/m2 IV, d1,8 6/7 1 G4 PLT 3 PR (2 EOC+1 EC)
2 6 mg/kg IV, day 1 600 mg/m2 IV, d1, 8 3/4 None 1 PR (1 TNBC)
3 6 mg/kg IV day 1 800 mg/m2 IV, d1, 8 3/4 None 2 PR (2 EOC)
4 6 mg/kg IV, day 1 1000 mg/m2 IV, d1,8 8 3 consented - -

3010 Poster Discussion Session; Displayed in Poster Session (Board #2), 3011 Poster Discussion Session; Displayed in Poster Session (Board #3),
Sat, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Sat, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Sat, 3:00 PM-4:30 PM Sat, 3:00 PM-4:30 PM
Phase 1 dose escalation study of XMT-1536, a novel NaPi2b-targeting antibody- Results of the phase 1b study of ABBV-399 (telisotuzumab vedotin; teliso-v)
drug conjugate (ADC), in patients (pts) with solid tumors likely to express NaPi2b. in combination with erlotinib in patients with c-Met+ non-small cell lung
First Author: Anthony W. Tolcher, NextOncology, San Antonio, TX cancer by EGFR mutation status. First Author: D. Ross Camidge, University
of Colorado Cancer Center, Aurora, CO
Background: XMT-1536 is a Dolaflexin ADC targeting the sodium-phosphate
cotransporter NaPi2b, expressed in ovarian, non-squamous lung, papillary Background: Telisotuzumab vedotin (ABBV-399; teliso-v [T]) is a c-Met–
thyroid, endometrial, papillary renal and salivary duct cancers. Methods: In targeted antibody and MMAE drug conjugate. Activity of T was shown in late-line
this ongoing Phase 1 study, pts with solid tumors likely to express NaPi2b, who c-Met+ non-small cell lung cancer (NSCLC) irrespective of EGFR mutation (M+)
progressed on standard therapy, are treated with intravenous XMT-1536 status. We present mature data from the T+ erlotinib (E) cohort of a phase 1b
using a 3+3 design with a modified Fibonacci escalation. NaPi2b expres- study (NCT02099058) by EGFR M+ status. Methods: T was administered at 2.4
sion by IHC is being examined retrospectively in archived tumors. Primary mg/kg (dose-escalation phase) or 2.7 mg/kg IV Q3W, and E at 150 mg PO QD/
objectives in dose escalation are safety and tolerability and determination of prior tolerated dose in adult patients (pts) with advanced NSCLC. Efficacy-
maximum tolerated dose (MTD) and recommended Phase 2 dose (RP2D). evaluable pts were c-Met+ (central lab IHC H-score $150 or local lab MET
(ClinicalTrials.gov NCT03319628). Results: As of Jan. 28, 2019, 36 pts (22 amplification/Ex 14 skipping) and had $1 postbaseline scan or discontinued
ovarian, 7 endometrial, 4 NSCLC, 3 other) have received treatment with XMT- study. EGFR M+ was defined as del19 or L858R by local lab. PK was assessed.
1536. Treatment was initially given every 3 weeks (q3w); 20 pts were treated in Results: As of Dec 2018, 42 NSCLC pts received T+E; 37 were c-MET+ (36
dose cohorts from 3 to 40 mg/m2. There was one DLT of reversible AST el- evaluable; 35 H-score$150, 1 MET amplified). Median age was 65 years, 25 pts
evation at 40 mg/m2. The dosing interval was then changed to every 4 weeks (69%) had ECOG PS 1, 29 (81%) were EGFR M+ (97% had prior EGFR TKI,
(q4w), and dose escalation was restarted at 20 mg/m2. There was one DLT of 55% 3rd-generation TKI, 69% TKI as last prior therapy, and 62% platinum
reversible AST elevation at 30 mg/m2 on the q4w schedule. Further followup doublet). All-grade (Gr; $20%) adverse events (AEs) were dermatitis acneiform
and dose escalation are ongoing. The most common ($10% of patients) (38%), diarrhea (36%), peripheral motor/sensory neuropathy (52%; 7% Gr 3),
treatment-related adverse events (TRAEs) have been nausea, fatigue, head- dyspnea, fatigue, hypoalbuminemia (31% each), decreased appetite, nausea
ache, increased AST, anorexia, increased alkaline phosphatase, fever, in- (24% each), asthenia, vomiting (21% each). Gr $3 ($10%) AE: pulmonary
creased GGT, myalgia, and vomiting. Grade 3 TRAEs were reversible AST embolism (14%). PK of T+E was similar to single-agent T. The table presents
increases in 3 patients and increased GGT, decreased lymphocytes, and efficacy data. Conclusions: These data suggest acceptable safety and promising
systolic congestive heart failure in 1 patient each. Treatment-related serious activity of T+E and support further study in EGFR M+ c-Met+ NSCLC pts for
AEs of fever and systolic congestive heart failure occurred in 1 patient each. whom frontline EGFR TKI failed. Clinical trial information: NCT02099058.
Among patients dosed at 20 mg/m2 or higher who had restaging scans (n=20), EGFR M+ EGFR non-M+
there were 2 PR, in ovarian cancer pts at 30 mg/m2 q3w and 20 mg/m2 q4w, (n = 29) (n = 7)*
and 11 SD, with disease control maintained for up to 24 weeks. Patient-level Objective response rate†, % (95% CI) 34.5 (17.9, 54.3) 28.6 (3.7, 71.0)
Complete response, n 1 0
results for NaPi2b expression will be presented. The systemic exposure of Median duration of response, mo (95% CI) Not reached (NR) NR
total payload showed approximately dose-proportional increase. Plasma (2.8, not estimable [NE]) (NR, NE)
Median progression-free survival (PFS), mo (95% CI) NR 5.9
concentration of free drug payload and its active metabolite were low. (2.8, NE) (1.2, NE)
Conclusions: XMT-1536 has been well-tolerated up to the 30 mg/m2 dose level Median follow-up, mo
6-mo PFS rate (95% CI)
4
0.51 (0.30, 0.69)
6
0.43 (0.10, 0.73)
with early signs of anti-tumor activity. Dose escalation continues in pts with Median treatment duration, wk (range)
T 15.4 (3.1–45.1) 18.1 (3.1–39.1)
advanced solid tumors likely to express NaPi2b. Clinical trial information: E 22.9 (3.1–110.4) 12.0 (4.6–42.0)
NCT03319628.
*EGFR wild-type, mutations other than del19 or L858R, and unknown mutation status. †RECIST version 1.1.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Therapeutics and Tumor Biology (Nonimmuno) 151s

3012 Poster Discussion Session; Displayed in Poster Session (Board #4), 3013 Poster Discussion Session; Displayed in Poster Session (Board #5),
Sat, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Sat, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Sat, 3:00 PM-4:30 PM Sat, 3:00 PM-4:30 PM
A phase 0 first-in-human study using NU-0129: A gold base spherical Phase 1, first-in-human study of TRAIL receptor agonist fusion protein
nucleic acid (SNA) nanoconjugate targeting BCL2L12 in recurrent glioblas- ABBV-621. First Author: Mark J. Ratain, University of Chicago, Chicago, IL
toma patients. First Author: Priya Kumthekar, Northwestern Memorial
Background: ABBV-621 is a potent tumor-necrosis factor-related apoptosis-inducing
Hospital, Chicago, IL ligand (TRAIL) receptor agonist fusion protein that induces apoptotic cell death,
Background: Glioblastoma is a difficult to treat tumor with therapeutics limited particularly in DR4/5 expressing tumor models. Methods: Patients (pts) with previously
by their ability to cross the blood brain barrier. SNAs, i.e., gold nanoparticle cores treated solid tumors and ECOG 0–2 were administered ABBV-621 (2.5–15 mg/kg IV)
covalently conjugated with a corona of densely packed, highly oriented siRNA on day (D) 1 (dose level [DL] 1) or D1D8 (DL2 and beyond) of each 21-day cycle. Dose
oligonucleotides targeted to the GBM oncogene BCL2L12, represent a novel escalation (DE) was guided by a Bayesian continual reassessment method. In addition
class of blood-brain and blood-tumor barrier-permeable nanomedicinal conju- to PK studies, blood-based PD markers of apoptosis (M30, M65) and drug binding were
assessed. Results: As of 14 December 2018, 57 pts were enrolled in the DE portion, of
gates, for suppressing gene expression in the tumors of GBM patients.
which 30% had pancreatic, 23% colorectal cancer, and 47% other tumor types; 13
Methods: This is a single-arm, open-label, “window of opportunity” phase 0 were KRAS mutant. Median age was 61 yrs. 60% were male; pts had a median of 4
first-in-human trial to determine the safety and bioavailability of a novel prior regimens (range 1–10). Pts per DL: 2.5 (5 on D1, 16 on D1D8), 3.75 (12), 5 (6),
nanotherapeutic compound, NU-0129. Enrolled patients were treated with 6.5 (6), 8.5 (4), 11 (4), and 15 mg/kg (4). Median duration of ABBV-621 exposure was
intravenous NU-0129 at the dose of 0.04mg/kg. This treatment dosing was 2 cycles (range 1–11). Seven pts had dose-limiting toxicities: respiratory failure (5 mg/
considered microdosing defined as 1/50ththe NOAEL (no observed adverse kg; Grade 5, the only treatment-related death), blood bilirubin increased (3.75, 6.5 mg/
event level) from non-human primate studies. Treatment was followed by kg), nausea (3.75 mg/kg), fatigue (3.75 mg/kg), increased ALT (2.5, 3.75, 6.5, 15 mg/
tumor resection 8-48 hours later. Primary outcome patient safety and toxicity kg), and increased AST (6.5 mg/kg). Summary of AEs is shown in Table. Clinical trial
was monitored weekly for 3 weeks post-infusion. Secondary objectives in- information: NCT03082209. A partial response (duration 20 weeks) was observed in a
cluded biodistribution of NU0129 in tissue, evaluation of pharmacokinetics of pt with pancreatic cancer (2.5 mg/kg D1D8). 27 pts had stable disease (6 pts for
NU0129 and the feasibility of NU0129 administration. Exploratory objectives . 12 weeks). ABBV-621 PK was linear (mean 6 SD clearance was 1.79 mL/h/kg 6
included Bcl2L12 expression and post treatment apoptotic markers as well as 0.44) with a terminal half-life of 36.7 6 5.55 h (n = 49). ABBV-621 bound to decoy
progression free survival and overall survival rates. Results: 8 patients were receptors on neutrophils for up to 168 h; the duration of binding was dose-dependent.
enrolled, treated and subsequently underwent surgical resection. No signifi- M30 and M65 increased at 8, 24, and 48 h following ABBV-621, but effect was
cant treatment related toxicities were seen. Severe ( . grade 3) adverse events independent of dose. Conclusions: ABBV-621 shows evidence of antitumor activity
were observed in two patients: hypophosphatemia (one grade 3, one grade 4) and effect on blood-based markers of apoptosis, with acceptable toxicity (MTD not
reached). NCT03082209.
and one patient with grade 3 lymphopenia, all were considered as “possibly
related” by treating oncologists. In 6 of the 8 patients sufficient tumor tissue AEs attributed to ABBV-621 by investigator, n (%) AEs in ‡5 pts Grade 3/4 AEs Serious AEs
was available for analysis of gold accumulation by ICP-MS (inductively coupled Increased ALT 11 (19) 7 (12) 2 (4)
Increased AST 10 (18) 5 (9) 2 (4)
plasma-mass spectrometry), and gold accumulation was seen in the tumor Nausea 10 (18) 1 (2) 0
tissue of all 6 of these patients. Conclusions: Macrodosing of the nano- Diarrhea 7 (12) 1 (2) 0
Stomatitis 7 (12) 0 0
therapeutic NU-0129 was well tolerated in glioblastoma patients with no Pyrexia 6 (11) 0 0
unexpected adverse effects and showed initial evidence of crossing blood brain Bilirubin increased 5 (9) 2 (4) 3 (5)
Decreased appetite 5 (9) 0 0
barrier. Immunohistochemistry for Bcl2L12 expression, apoptotic markers, Fatigue 5 (9) 1 (2) 0
and PK studies are pending. The demonstration of gold nanoparticles in the Vomiting 5 (9) 0 0
tumor tissue validates this approach for drug delivery. Clinical trial information:
NCT03020017.

3014 Poster Discussion Session; Displayed in Poster Session (Board #6), 3015 Poster Discussion Session; Displayed in Poster Session (Board #7),
Sat, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Sat, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Sat, 3:00 PM-4:30 PM Sat, 3:00 PM-4:30 PM
Phase I trial of IACS-010759 (IACS), a potent, selective inhibitor of complex Phase I study of CC-90010 in patients with advanced solid tumors and
I of the mitochondrial electron transport chain, in patients (pts) with relapsed/refractory non-Hodgkin lymphoma (R/R NHL). First Author: Victor
advanced solid tumors. First Author: Timothy A Yap, The University of Moreno, START Madrid - FJD, Hospital Universitario Fundación Jimenez
Texas MD Anderson Cancer Center, Houston, TX Diaz, Madrid, Spain
Background: A subset of tumors possess genetic or microenvironmental al- Background: Bromodomain and extra-terminal (BET) proteins are epigenetic
terations that render cells dependent on mitochondria oxidative phosphorylation readers that control expression of genes involved in cell growth and onco-
(OXPHOS) for survival. IACS, a potent oral selective inhibitor of mitochondrial genesis. CC-90010 is an oral, potent and reversible BET inhibitor that showed
complex I, showed robust responses in multiple preclinical tumor models, promising activity in lymphoma and solid tumor cell lines and reduced tumor
providing strong rationale for clinical testing. Methods: Pts with advanced growth in xenograft models. Methods: CC-90010-ST-001 (NCT03220347;
cancers received IACS in increasing dose levels (DL) using 3+3 dose escalation. 2015-004371-79) is a phase I, first-in-human study of CC-90010 in patients
7-day QD induction of IACS was followed by maintenance weekly (QW) or twice (pts) with advanced solid tumors and R/R NHL. Three schedules and 11 dose
weekly (BIW) dosing. Phamacokinetics (PK), lactate and pH were assessed levels were evaluated (Table). Primary objectives were to determine safety,
serially. Paired tumor biopsies were assessed for pharmacodynamic and pre- maximum-tolerated dose and/or recommended phase II dose (RP2D). Sec-
dictive biomarkers. Results: 18 pts were treated; M/F 16/2; ECOG PS 0/1: 3/15. ondary objectives were the identification of early activity signals, pharmaco-
Mean age 49 (23-69) yrs. Tumors comprised advanced colorectal (n = 4), kinetics and pharmacodynamics (PD). Results: As of 10 Dec 2018, 69 pts
castration resistant prostate cancer (CRPC) (n = 3), pancreatic (n = 2), other were enrolled, 67 with solid tumors and 2 with R/R NHL. Data shown are from
cancers (n = 9). DL1: 2mg QD 7 days induction/0.5mg QW maintenance (n = 3); all pts (N = 69). The median age was 57 y (range, 21–80), 38 (55%) were
DL2: 2.5mg QD 7 days/1mg QW (n = 3); DL3: 3mg QD 7 days/3mg QW (n = 3); male, and the median number of prior systemic anticancer regimens was 3
DL4: 2.5mg QD 7 days/2.5mg BIW (n = 4); DL5: 2mg QD 7 days/2mg BIW (n = (range, 1–9). The RP2Ds were dose cohorts 3A and 4B. Dose-limiting toxicities
5). IACS was well tolerated with 12 (67%) pts reporting G1-2 IACS related (n = 6) occurred in dose cohorts 3A, 3C, and 4B. Grade 3/4 treatment-related
toxicities, such as raised lactate (n = 10), nausea (n = 8), fatigue (n = 7), adverse events (TRAEs) occurred in 17 pts (25%), most commonly ($2 pts)
vomiting (n = 5), myalgia (n = 4) and peripheral neuropathy (n = 4). 1 pt in DL3 thrombocytopenia (7%), platelet count decreased (4%), fatigue (3%), and
and 2 pts in DL4 had $G3 IACS related toxicities, such as nausea (n = 2), increased alanine aminotransferase (3%). No deaths from toxicity occurred.
vomiting (n = 1), raised lactate (n = 1), dehydration (n = 1), visual changes (n = Two pts (endometrial carcinoma and astrocytoma) had a partial response (PR);
1), and peripheral neuropathy (n = 1). Raised lactate was not associated with 1 occurred after the data cutoff. Seven pts had prolonged stable disease (SD)
acidosis. DL5 is now being expanded to assess the maximum tolerated dose . 9 mo. Exposures and PD marker regulation increased with dose in each
(MTD). PK showed good oral bioavailability, with long T1/2 and low intrapatient dosing schedule; terminal half-life was ~ 73 h. Conclusions: Most of the TRAEs
variability. Cmax = 14nM on Day 7 at the end of DL5 induction phase, con- observed were mild or moderate in severity, reversible, and manageable by
firming biologically active doses. 7 pts had best response of RECIST stable dose adjustments and/or supportive care. Promising ongoing anticancer ac-
disease. A pt with heavily pretreated CRPC achieved RECIST partial response tivity with prolonged SD and PRs were observed. The preliminary clinical data
with resolution of CRPC related pain. Conclusions: IACS is well tolerated with provide the rationale for dose expansion of CC-90010 in pts with selected
preliminary evidence of antitumor activity. MTD expansions include CRPC, advanced malignancies. Clinical trial information: NCT03220347.
TNBC, pancreatic cancer and molecularly selected (ENO1 loss; SMARCA4 1 2 3A 3B 3C 4A 4B 4C 5A 5B 5C
mutation) tumor cohorts. Clinical trial information: NCT03291938. Dose Level: (n = 7) (n = 7) (n = 4) (n = 6) (n = 6) (n = 7) (n = 7) (n = 7) (n = 6) (n = 6) (n = 6)
Dose, mg 15 15 25 30 15 40 45 25 30 55 35
Schedule, days on/off 3/4 3/11 3/11 4/24 2/5 3/11 4/24 2/5 3/11 4/24 2/5

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
152s Developmental Therapeutics and Tumor Biology (Nonimmuno)

3016 Poster Discussion Session; Displayed in Poster Session (Board #8), 3017 Poster Discussion Session; Displayed in Poster Session (Board #9),
Sat, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Sat, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Sat, 3:00 PM-4:30 PM Sat, 3:00 PM-4:30 PM
Design and development of the molecular analysis for Therapy Choice (NCI- Efficacy of entrectinib in patients (pts) with solid tumors and central nervous
MATCH) Designated Laboratory Network. First Author: James V. Tricoli, system (CNS) metastases: Integrated analysis from three clinical trials. First
National Cancer Institute, Rockville, MD Author: Salvatore Siena, Niguarda Cancer Center, Grande Ospedale Met-
ropolitano Niguarda, and Department of Oncology and Hemato-Oncology,
Background: NCI-MATCH is a precision medicine trial that assigns treatment
Università degli Studi di Milano, Milan, Italy
to refractory cancer patients by tumor mutation profile rather than by histology.
After screening fresh tumor biopsies from nearly 6000 patients many treat- Background: Entrectinib potently inhibits kinases encoded by the NTRK and ROS1 genes.
ment arms did not meet accrual due to the low prevalence of the eligible It achieves therapeutic levels in the CNS with antitumor activity in intracranial tumor models.
variants. NCI MATCH developed an approach to identify patients for the We report integrated data (31 May 2018 cut-off) from 3 Phase 1/2 entrectinib trials (ALKA-
372-001, EudraCT 2012-000148-88; STARTRK-1, NCT02097810; STARTRK-2,
remaining arms utilizing a network of academic and commercial CLIA-certified NCT02568267) for a large cohort of adults with ROS1 fusion-positive NSCLC (ROS1+)
labs that perform NGS assays as routine care at MATCH participating sites. or NTRK fusion-positive solid tumors (NTRK+), with/without baseline CNS metastases.
Methods: Candidate labs were recruited through a notice in the Federal Methods: Pts had locally advanced/metastatic NTRK+ or ROS1+ tumors by nucleic acid-
Register and posted on the NCI and ECOG ACRIN web sites. Twenty-seven labs based confirmation. Baseline CNS metastases were identified by CT/MRI. Tumor assessments
(17 academic/10 commercial) submitted applications. After acceptance each were at wk 4, then every 8 wk by blinded independent central review (RECIST v1.1). Primary
lab analyzed a common set of 10 DNAs extracted from 8 cell lines and 2 endpoints: ORR, DOR. Secondary endpoints: CBR, PFS, OS, intracranial efficacy and safety.
clinical samples for concordance with the central NCI-MATCH NGS assay. Results: Most pts had $1 prior therapy; 33% had baseline CNS metastases. Outcomes for the
ROS1+ NSCLC (n = 53) and NTRK+ solid tumors (n = 54; 24% sarcoma, 18% NSCLC)
Results: For the 17 labs with concordance results, a median of 8 (range 2 – 58)
efficacy evaluable data sets are shown (table). Entrectinib was tolerable with a manageable
copy number variants (CNVs) were evaluated by the NGS assay of each DL, with safety profile; most treatment-related AEs were grade 1–2. Conclusions: Entrectinib induced
the number evaluated depending on each lab’s clinical assay panel content. clinically meaningful durable responses in pts with ROS1+ NSCLC or NTRK+ solid tumors with
CNV concordance between central and DL assays, as measured by positive or without CNS disease. Clinical trial information: NCT02097810; NCT02568267.
percent agreement (PPA), averaged 98.7% (range 87.5% - 100%) with the
ROS1+ (N = 53) NTRK+ (N = 54)
central assay as referent and 94.1% (range 77.8% – 100%) with the DL assay
Baseline CNS disease status
as referent. For single nucleotide variants (SNVs) and Insertion/deletions
No (n = 30) Yes (n = 23) No (n = 42) Yes (n = 12)
(Indels) combined, a median of 19 variants (range 11 – 26) were evaluated
by each DL for concordance. PPA between central and DL assays averaged ORR, % (95% CI)
Complete response, n (%)
80.0 (61.4, 92.3)
3 (10.0)
73.9 (51.6, 89.8) 59.5 (43.3, 74.4) 50.0 (21.1, 78.9)
0 4 (9.5) 0
98.0% (range 87.5% – 100%) and 98.6% (range 90.0% – 100%) with Partial response, n (%) 21 (70.0) 17 (73.9) 21 (50.0) 6 (50.0)
DOR* 24.6 12.6 12.9 NE
central and DL assay as referents, respectively. Strong correlations were ob- (11.4, 34.8) (6.5, NE) (7.1, NE) (4.2, NE)
served between central and DL assays for both CNVs (median r = 0.93; 0.33 – CBR, % (95% CI) 80.0 73.9 61.9 75.0
(61.4, 92.3) (51.6, 89.8) (45.6, 76.4) (42.8, 94.5)
1.00) and SNV/Indels (median r = 0.98; 0.67 – 0.99). Conclusions: Our PFS* 26.3 13.6 12.0 7.7
results suggest that different NGS assay platforms using diverse strategies for (15.7, 36.6) (4.5, NE) (8.7, 15.7) (4.7, NE)
OS* NE NE 20.9 14.3
target enrichment and data analysis may still achieve high concordance if pre- (10.5, NE) (16.8, NE) (51, NE)
analytical variables are minimized and the common genomic regions in- Intracranial NE NE
ORR, % (95% CI) 55.0 (31.5, 76.9)a 54.5 (23.4, 83.3)b
terrogated by each assay are well-understood. The designated lab network DOR* 12.9 (5.6, NE) NE (5.0, NE)
allows for a wider search for rare variants in tumors and provides a model for PFS* 7.7 (3.8, 19.3) 14.3 (5.1, NE)

conducting future clinical trials. Clinical trial information: NCT02465060. *Median, months (95% CI). ORR, objective response rate; DOR, duration of response; CBR, clinical benefit rate;
PFS, progression-free survival; OS, overall survival; NE, not evaluable
a
n = 20; bn = 11

3018 Poster Discussion Session; Displayed in Poster Session (Board #10), 3019 Poster Discussion Session; Displayed in Poster Session (Board #11),
Sat, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Sat, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Sat, 3:00 PM-4:30 PM Sat, 3:00 PM-4:30 PM
Genome-wide cell-free DNA fragmentation profiling for early cancer Multimodality liquid biopsy for early monitoring and outcome prediction in first-
detection. First Author: Alessandro Leal, Johns Hopkins University School line metastatic HER2-negative breast cancer: Final results of the prospective
of Medicine, Sidney Kimmel Comprehensive Cancer Center, Baltimore, MD cohort from the French Breast Cancer InterGroup Unicancer (UCBG)— COMET
study. First Author: Jean-Yves Pierga, Institut Curie, Paris, France
Background: Analyses of cell-free DNA (cfDNA) in the blood provide a non-
invasive diagnostic avenue for patients with cancer. However, cfDNA analyses Background: Circulating Tumor Cells (CTC) are independent markers of
have largely focused on targeted sequencing of specific genes, and the char- progression-free survival (PFS) and overall survival (OS) in patients (pts) with
acteristics of the origins and molecular features of cfDNA are poorly understood. metastatic breast cancer (MBC). Monitoring circulating tumor DNA (ctDNA) can
We developed an ultrasensitive approach that allows simultaneous examination detect mutation associated with resistance to treatment and its variations reflect
of a large number of abnormalities in cfDNA through genome-wide analysis of changes in tumor burden. We prospectively monitored CTC, Circulating Endo-
fragmentation patterns. Methods: We used a machine learning model to ex- thelial Cells (CEC), serum markers and ctDNA during first line chemotherapy for
amined cfDNA fragmentation profiles of 236 patients with largely localized MBC. Methods: The French cohort COMET is a prospective study including first
breast, colorectal, lung, ovarian, pancreatic, gastric, or bile duct cancer and 245 line HER2 negative pts receiving weekly paclitaxel and bevacizumab . Blood
healthy individuals. Estimation of performance was determined by ten-fold cross samples were obtained at baseline (BL) and before the second cycle of chemo-
validation repeated ten times. Results: cfDNA profiles of healthy individuals therapy (C2).We present here the final planned analysis. Results: From 09/2012 to
reflected nucleosomal patterns of white blood cells, while patients with cancer 11/2014, 286 patients were included: 198 for ctDNA, 251 for CEC and 283 for
had altered fragmentation patterns. The degree of abnormality in fragmentation CTC. Median age was 56 years and 23% of pts had triple negative BC. At baseline,
profiles during therapy closely matched levels of mutant allele fractions in cfDNA 71% of pts had $1 detectable CTC per 7.5 ml of blood (median 4 CTC, range 1-
as determined using ultra-deep targeted sequencing. The sensitivity of detection 30,000). With a threshold of $5 CTC, 49% of pts were positive at baseline and
ranged from 57% to . 99% among the seven cancer types at 98% specificity, 22% at C2. For ctDNA, out of the first 196 pts analyzed, 147 had at least one
with an overall AUC of 0.94. Fragmentation profiles could be used to identify the somatic mutation (SNV) detected in plasma (75%). The average number of
tissue of origin of the cancers to a limited number of sites in 75% of cases. mutations per pt was 2.4 (range 1 to 9). Most commonly mutated genes were TP53
Combining our approach with mutation-based cfDNA analyses detected 91% of and GATA3. ESR1 was mutated in 10.6% of the pts and restricted to the ER+
subgroup. PIK3CA was mutated in 23.2% of the pts. Median Allelic Frequency was
cancer patients. Conclusions: This effort is the first study to demonstrate
9.1% . Only 68 pts (36%) had detectable ctDNA at C2. At baseline, CTC and
genome-wide cell-free DNA fragmentation abnormalities in patients with cancer.
ctDNA levels were correlated (r = 0.40, p , 0.0001). Despite no complete overlap,
Results of these analyses highlight important properties of cfDNA and provide a
24 pts (12%) had no CTC nor ctDNA detected at baseline. Median follow-up was
facile approach for screening, early detection, and monitoring of human cancer.
53 months and median OS was 32 months. Detectable CTC and ctDNA at baseline
and at C2 were significantly associated with decreased PFS and OS. CEC and
serum markers level had no prognostic value. At multivariate analysis, triple
negative status, detectable ctDNA at C2, CTC $5 at C2 and grade 3 on primary
tumor were independent prognostic factors. Conclusions: This is the largest
prospective cohort assessing the respective prognostic values of early CTC and
ctDNA changes in homogenously treated first line MBC pts. Early decrease of CTC
and ctDNA after one cycle of chemotherapy are independent predictive markers of
favorable outcome, with a stronger value for ctDNA compared to CTC. Clinical
utility of early ctDNA variations monitoring and changes in mutation profile remain
to be demonstrated. Clinical trial information: NCT01745757.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Therapeutics and Tumor Biology (Nonimmuno) 153s

3020 Poster Discussion Session; Displayed in Poster Session (Board #12), 3021 Poster Session (Board #13), Sat, 8:00 AM-11:00 AM
Sat, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, A phase I study of pazopanib with weekly paclitaxel and carboplatin in
Sat, 3:00 PM-4:30 PM advanced solid tumors. First Author: Nancy Chan, Rutgers Cancer Institute
Circulating androgen receptor (AR) gene amplification and resistance to of New Jersey, New Brunswick, NJ
177Lu-PSMA-617 in patients (pts) with metastatic castration-resistant
Background: Pazopanib (pazo) is an oral tyrosine kinase inhibitor of VEGFR,
prostate cancer (mCRPC): Results of a phase II clinical trial. First Author:
PDGFR and c-Kit. It is a weak inhibitor of CYP3A4 and CYP2C8 and may
Ugo De Giorgi, Istituto Scientifico Romagnolo per lo Studio e la Cura dei
decrease paclitaxel (P) clearance. Daily pazo with P and carboplatin (C) every
Tumori (IRST) IRCCS, Meldola, Italy
21 days was not feasible on a previous study. We hypothesized that pazo dosed
Background: Plasma AR gain is associated with poor prognosis in mCRPC pts intermittently and on a different day from P and C may be tolerable. We sought to
treated with abiraterone/enzalutamide, however these pts could benefit from determine the maximum tolerated dose (MTD), dose-limiting toxicity (DLT), and
docetaxel (Conteduca et al, Eur Urol 2019). In a phase 2 clinical trial with 177Lu- pharmacokinetics (PK) of pazo with weekly P and C. Methods: Using a 3+3
PSMA-617 in mCRPC pts who progressed after standard survival-prolonging standard design, a schedule of P 60-80 mg/m2 and C AUC2 on days 1, 8, and
treatments, we aimed to determine if plasma AR gene status enable early as- 15 with pazo 400-800 mg on days 2-5, 9-12, and 16-26 on a 28-day cycle was
sessment of 177Lu-PSMA-617 activity for mCRPC. Methods: Between April evaluated. Pazo alone could be continued if P and C were omitted due to
2017 and November 2018, 43 mCRPC pts were treated with 177Lu-PSMA-617 maximal benefit or toxicity. PK was collected during cycles 1 and 2. Results: 34
in a phase 2 study. Pts younger than 75 years and not heavily pretreated received patients (pts) were treated over 6 dose levels (Table). Mean age 57 (37-79).
5.5 GBq of 177Lu-PSMA-617, while other pts received 4.2 GBq per cycle, for a Tumor types: breast (22), lung (3) and other (9); 27 had prior platinum. Delay in
total of 4-6 cycles, q8 weeks. We determined AR copy number by droplet digital starting cycle 2 due to grade 3 neutropenia was a DLT at dose level 2 and 5. Pts
polymerase chain reaction (ddPCR) on pretreatment plasma samples. We eval- on 5A missed dosing during C1 and C2 due to neutropenia and required
uated associations between plasma AR amplification and PSA response ($50% subsequent growth factor, and this was deemed unlikely to be sustainable long-
PSA decline from baseline) and imaging response/progression (as measured by term. All grade toxicities included anemia (62%), neutropenia (59%), and
bone scan, CT, and PSMA PET/CT). Logistic regression was used to estimate the thrombocytopenia (56%). Protocol-defined MTD was not determined. PK
odds ratio (OR) and 95% confidence intervals (95% CI) in order to evaluate the analysis showed a dose proportional increase in pazo concentration, consistent
independent relevance of AR status and pts without PSA response and those with with previous reports. Pazo did not alter the PK of C. Cmax of P was higher C2D1
early progressive disease defined as treatment interruption occurring within vs C1D1; mean Cmax ratio between C2D1:C1D1 was 1.63 (95% CI:1.29-1.96).
4 months of the start of 177Lu-PSMA-617. Results: Forty of 43 pts (median age: There were 11 objective responses (3 CRs, 8 PRs). Five breast pts were on pazo
72 years, range 54-86) were fully evaluable for this analysis. A PSA response was alone for a median of 9 cycles (2-52) with CR (2), PR (2) and SD (1); a
reported in 15 (37.5%) of the 40 pts, 3 of 15 (20%) with AR gene gain, and 12 of squamous cell of unknown primary in CR received 22 cycles. Clinical trial
25 (48%) with no gain (P = 0.080). Early progressive disease was observed in 17 information: NCT01407562. Conclusions: PK confirm that pazo is a weak
(42.5%) of the 40 pts, 12 of 15 (80%) with AR gene gain and 5 of 25 (20%) with inhibitor of CYP3A4 and CYP2C8. Myelosuppression was a major adverse event
no gain (P = 0.0002). The OR for pts without PSA response (decline , 50%) at all dose levels. MTD was not determined. Antitumor activity was achieved
having AR gain was 3.69, 95% CI 0.83-16.36, p = 0.085. The OR for pts with with this alternate combination schedule and sustained responses from se-
early PD having AR gain was 16.00, 95% CI 3.23-79.27, p = 0.0007. The quential pazo monotherapy was observed.
evaluation of germline alterations in DNA damage repair (DDR) genes is ongoing Dose Level Paclitaxel Carboplatin Pazopanib No. of Pts No. of Pts with DLT
(i.e., BRCA2, BRCA1, ATM). Conclusions: Plasma AR status assessment using
1 60 mg/m2 AUC 2 400 mg 3
ddPCR identifies mCRPC resistant to 177Lu-PSMA-617. These data suggest 2 60 mg/m2 AUC 2 600 mg 9 1
potential better activity of 177Lu-PSMA-617 in earlier phases of prostate cancer. 3 70 mg/m2 AUC 2 600 mg 7
Prospective evaluation of treatment decision making based on plasma AR status is 4 70 mg/m2 AUC 2 800 mg 4
5 80 mg/m2 AUC 2 800 mg 8 1
warranted. Clinical trial information: NCT03454750. 5A 80 mg/m2 AUC 2 600 mg 3

3022 Poster Session (Board #14), Sat, 8:00 AM-11:00 AM 3023 Poster Session (Board #15), Sat, 8:00 AM-11:00 AM
Exceptional responders to abexinostat (ABX) plus pazopanib (PAZ) in Phase 1a study results investigating the safety and preliminary efficacy of
pretreated renal cell carcinoma (RCC) and other solid tumors: Long-term ABL001 (NOV1501), a bispecific antibody targeting VEGF and DLL4 in
follow-up of a phase 1b study. First Author: Rahul Raj Aggarwal, UCSF Helen metastatic gastrointestinal (GI) cancer. First Author: Jeeyun Lee, Samsung Medical
Diller Family Comprehensive Cancer Center, San Francisco, CA Center, Sungkyunkwan University School of Medicine, Seoul, South Korea
Background: We previously reported the initial phase 1b study results of Background: Antiangiogenic therapy has been a successful clinical strategy
PAZ + ABX, a potent pan-HDAC inhibitor, demonstrating acceptable toxicity for the treatment of various cancer types. To date, all approved antiangiogenic
profile and encouraging anti-tumor activity (Aggarwal et al. JCO 2017). We drugs primarily inhibit the VEGF/VEGFR pathway. Delta-like ligand 4 (DLL4)
report the long-term follow up of exceptional responders and additional has been identified as a potential drug target in VEGF-independent angio-
correlative analyses associated with clinical outcomes. Methods: Key effi- genesis. A dual blockade of both VEGF and DLL4 could be a promising strategy
cacy endpoints included objective response rate and duration of response. to overcome anti-VEGF therapy resistance. ABL001 (NOV1501) has been
Peripheral blood histone acetylation, HDAC expression, and plasma VEGF developed as a bispecific antibody to bind and inhibit both DLL4 and VEGF
levels were analyzed and associated with clinical outcomes. Results: 51 pts thereby significantly suppressing tumor angiogenesis. Methods: In a classical
(RCC subset; N = 22) were enrolled between June 2012 and October 2015. 3+3 dose-escalation design, ABL001 was administered IV at doses ranging
10 pts (20%) had experienced disease progression on prior PAZ; 59% had from 0.3, 1, 2.5, 5, and 7.5 mg/kg biweekly (NCT03292783: the next doses of
received any prior VEGF-targeting therapy. 9 evaluable pts (18%) (N = 6 ABL001 are 10 and 12.5 mg/kg). After the first administration of ABL001 in
RCC; 2 thyroid; 1 mesothelioma) achieved partial tumor response (PR), of each cohort, DLT (dose limiting toxicity) was observed for 3 weeks. Tumor
which 6 had prior progression on VEGF-targeting therapy. 7/10 (70%) of pts assessments were performed every 6 weeks and cardiac assessments were
with prior disease progression on PAZ monotherapy had reduction in tumor performed every cycle. Results: From 2017 November to February 2019, 18
burden on study. The median duration of response was 9.1 months (range patients were enrolled on this trial. All patients were heavily pre-treated with at
1.2 to 70+), and clinical benefit rate (PR or stable disease . 6 months) was least 3 prior lines of chemotherapy. All patients in cohort 4 and 5 were either
33%. Five treatment-refractory pts achieved durable PRs lasting for . 2 metastatic colorectal cancer or gastric cancer. Of the 5 cohorts, there was no
years duration, and one previously PAZ-refractory patient with RCC remains DLT observed during dose escalation. In addition, there was no maximum
on treatment with ongoing PR for . 6 years. Higher HDAC2 expression was tolerated dose identified up to 7.5 mg/kg dose. The most common treatment-
associated with prolonged progression-free survival (median PFS 5.9 vs. related adverse events (AEs) (including all dose levels and all grades) occurred
3.5 months, log-rank p = 0.02). Induction of histone acetylation on ABX were hypertension, anorexia, general weakness, headache and anemia. Pre-
lead-in treatment was associated with subsequent time to progression (p = liminary results of pharmacokinetic (PK) analysis demonstrated slightly shorter
0.002). On-treatment plasma VEGF levels were inversely correlated with mean half-life than conventional monoclonal tantibodies due to the bispecific
PBMC histone acetylation (p = 0.02). Conclusions: Markedly durable re- nature of the ABL-001. In addition, preliminary pharmacodynamic (PD)
sponses with PAZ + ABX are achievable, including in pts with PAZ- and biomarker analysis using PBMC and plasma samples showed engagement of
VEGF-refractory RCC and other solid tumor malignancies. Host factors in- both VEGF/VEGFR and DLL4/Notch1 pathway modulation after ABL001
cluding HDAC expression and acetylation status may identify those most administration. One gastric cancer patient at 7.5 mg/kg achieved unconfirmed
likely to benefit. A randomized phase 3 study is underway of PAZ + ABX as a partial response at the time of this writing. Conclusions: ABL001 therapy has
first- or second-line therapy in pts with locally advanced or metastatic RCC been well tolerated up to 7.5 mg/kg with no significant treatment related
(RENAVIV; NCT03592472). Clinical trial information: NCT01543763. adverse events and showed preliminary anti-tumor activity in heavily pre-
treated cancer patients. After completion of this ongoing phase 1a study,
phase 1b/2a study is planned in combination of ABL001 with chemotherapy or
anti-PD-1 antibody. Clinical trial information: 03292783.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
154s Developmental Therapeutics and Tumor Biology (Nonimmuno)

3024 Poster Session (Board #16), Sat, 8:00 AM-11:00 AM 3025 Poster Session (Board #17), Sat, 8:00 AM-11:00 AM
The dynamic detection of drug area under curve (AUC) guides clinical usage A polymorphism within the mismatch repair gene predicts prognosis and
of docetaxel in solid tumors. First Author: Yan Zhang, Internal Medicine- adjuvant chemotherapy benefit in gastric cancer patients. First Author:
Oncology, Jiangsu Cancer Hospital, The Affiliated Cancer Hospital of Deqiang Wang, Department of Medical Oncology, Cancer Therapy Center,
Nanjing Medical University, Nanjing, China Affiliated Hospital of Jiangsu University, Zhenjiang, China
Background: The dosage of most chemotherapy drugs were performed based Background: Radical surgery with subsequent adjuvant chemotherapy was
on the patients’ body surface area (BSA), including docetaxel (DTX). Pre- effective treatment for early-stage gastric cancer (GC) patients. Unfortu-
vious studies showed that this conventional administration of DTX might nately, after optimal multimodality therapy, up to 30% to 40% of patients
cause adverse event, such as neutropenia, and neutropenia was proved to undergoing resection will relapse within 5 years. There are no validated
associate with DTX area under curve (AUC). This study was designed to prognostic and predictive biomarkers for GC patients who receive adjuvant
evaluate the effect of dose-administration of DTX based on dynamic de- chemotherapy, and current patient selection is based mainly on post-
tection of DTX AUC on clinical outcomes. Methods: A total of 209 patients operative pathological staging. Defective mismatch repair (MMR) or micro-
with DTX chemotherapy (one cycle every 3 weeks) were enrolled, and all satellite instability (MSI) may affect GC outcome. Polymorphisms of MMR
patients received 2-6 cycles of treatment. In the first cycle, dosage of DTX genes with a low-penetrant effect can cause heterogeneous MMR capability
based on BSA was administrated in all study population. From the second among individuals. It is not known about the impact of these polymorphisms on
cycle, one group patients (control group) received DTX according to tradi- GC outcome. Methods: The polymorphisms rs1800734 in MLH1, rs2303428
tional BSA and the other group patients (experimental group) on the basis of and rs3732183 in MSH2, rs735943 in EXO1, and rs11797 in TREX1 were
dynamic detection of DTX AUC. The primary outcome was incidence rate of selected and analyzed in independent discovery and validation sets that in-
neutropenia and the second outcome was disease control rate (DCR). cluded 167 and 593 patients, respectively. MSI was determined. Results: In
Results: Patients with grade 3 or higher neutropenia from the fourth to sixth the discovery set, both the rs2303428 TC+CC and the rs11797 GA+AA ge-
cycle of DTX chemotherapy were significantly lower in the experimental notypes significantly correlated with poor overall survival (OS; P , 0.05). In the
group compared with the control group (P= 0.039, 0.012, and 0.001, validation set, we confirmed the prognostic association for the rs2303428
respectively). In the experimental group, compared with the first cycle, and TC+CC genotype (P = 0.036) but not for the rs11797 GA+AA genotype
the number of patients falling within the therapeutic window increased by (P = 0.737). Furthermore, the prognostic role of the rs2303428 TC+CC ge-
27.19% in the sixth cycle after dose adjustment according to the AUC value notype was observed in non-cardia (P = 0.005) but not in cardia GC (P = 0.934).
of previous cycle. The DCR in the experimental and control group is 85.32% The multivariate model showed that the rs2303428 TC+CC genotype was an
and 72.00%, respectively (P= 0.018). Conclusions: The administration independent predictor for OS in non-cardia patients (HR = 1.54; 95% CI: 1.02-
method based on dynamic detection of AUC of DTX could significantly 2.32; P = 0.040). Moreover, fluoropyrimidines-based adjuvant chemotherapy
reduce incidence rate of neutropenia and received a higher DCR, but the significantly improved OS (HR = 0.29; 95% CI: 0.15-0.58; P , 0.001) for non-
result needed to be confirmed in further studies. cardia patients with the rs2303428 TC+CC genotype but not for those with the
rs2303428 TT genotype. The rs2303428 genotypes were not associated with
MSI frequency. Conclusions: The rs2303428 TC+CC genotype may predict
prognosis and adjuvant chemotherapy benefit in non-cardia GC patients in-
dependent of MSI. To our knowledge, our study is the first to report the
prognostic and predictive roles of MMR genotype in GC. Although prospective
validation is necessary, our findings have the potential to improve patient se-
lection for adjuvant chemotherapy and spare large numbers of GC patients’
unnecessary therapy.

3026 Poster Session (Board #18), Sat, 8:00 AM-11:00 AM 3027 Poster Session (Board #19), Sat, 8:00 AM-11:00 AM
A phase I dose-finding and pharmacokinetics study of CPC634 (nanoparticle Phase I trial of chloroquine (CQ)/hydroxychloroquine (HCQ) in combination
entrapped docetaxel) in patients with advanced solid tumors. First Author: with carboplatin-gemcitabine (CG) in patients with advanced solid tumors.
Florence Atrafi, Erasmus MC Cancer Institute, Rotterdam, Netherlands First Author: Nagla Fawzy Abdel Karim, The University of Cincinnati, Cin-
cinnati, OH
Background: CPC634 is a novel product with docetaxel temporarily entrapped
within stabilized CriPec nanoparticles. We performed the first-in-human study Background: Autophagy is a catabolic process triggered in cells during periods
with CPC634 (NCT02442531). Methods: Patients ($18 years) received of stress to enable their survival. Established tumors utilize autophagy to survive
CPC634 intravenously either 3-weekly (Q3W) (part 1, 15-100 mg/m2), 2- periods of metabolic or hypoxic stress. Inhibition of early stage autophagy can
weekly (Q2W) (part 2, 45 mg/m2) or Q3W with dexamethasone premedication rescue cancer cells, while inhibition of late stage autophagy will lead to cell death
(part 3) following a 3+3 design. Primary objectives were to assess safety, due to accumulation of damaged organelles. The antimalarial drugs CQ and HCQ
establish the maximum tolerated dose (MTD), recommended phase 2 dose inhibit late phase autophagy. The goal of our study is to assess the safety,
(RP2D), and to evaluate the pharmacokinetic (PK) profile of CPC634. tolerability and activity of combining CQ/HCQ with CG in advanced solid tumor
Results: Thirty-three patients (part 1; n = 24, part 2; n = 3, part 3; n = 6) were patients who either progressed on other therapies or in whom CG is a therapeutic
treated. Skin toxicity was dose limiting at doses . 60 mg/m2 in part 1, and at a option. Methods: This single institution phase 1 dose-escalation study was
45 mg/m2 dose in part 2. Skin toxicity was cumulative but resolved after designed to evaluate the maximum tolerated dose (MTD) of CQ, later substituted
ceasing treatment. The MTD in part 1 was set at 70 mg/m2. In part 3, the with HCQ, in combination with CG in patients with previously treated advanced
60 mg/m2 was explored which resulted in improved skin tolerability even after solid tumors. Secondary objectives were to determine ORR, PFS and OS. A
repeated administrations without dose limiting toxicities. The RP2D was starting dose of 50 mg of CQ/HCQ was used in conjunction with CG, and in-
therefore set at 60 mg/m2 with dexamethasone premedication. Grade $3 creased in increments of 50 mg in each dose cohort. Grade 3 or greater toxicity
adverse events (CTCAE version 4.03) were skin toxicity (21%), fatigue that is treatment-related, and was not self-limited, or controlled in less than
(8%), neutropenia (6%), peripheral sensory (8%) and motor neuropathy (4%), 7 days was considered dose limiting toxicity (DLT). Results: Twenty-three pa-
stomatitis (4%), infections (4%) and hypomagnesemia (3%). Alopecia grade 1 tients were enrolled with a median follow up of 6 months. HCQ 100 mg was
was reported in 15% of patients. CPC634 exhibited a dose-proportional PK found to be the MTD in combination with CG with $Grade 3 thrombocytopenia
profile. One partial response and sixteen cases of stable disease (RECIST 1.1) and/or neutropenia as dose-limiting. Median OS was 11 months, and the 1- and
were confirmed in part 1 and in part 3 as best response. Conclusions: CPC634 3- year overall survival rates were 30% and 7%, respectively. Median progression
could be administered safely but showed cumulative, though reversible free survival was 5 months and the 6-, 12-, and 18-months progression-free
skin toxicity at high doses. The RP2D was set at 60 mg/m2 Q3W with survivals were 48%, 21% and 14%, respectively (Table). Conclusions: The MTD
dexamethasone premedication. Additional studies assessing the intratumoral identified for CQ/HCQ was lower than previously reported with concomitant use
exposure to CPC634 (NCT0371243) and a phase II efficacy study of CPC634 of chemotherapeutic regimes, likely due to the myelosuppressive nature of CG.
in patients with platinum resistant ovarian cancer (NCT03742713) is currently Clinical trial information: NCT02071537.
ongoing. Clinical trial information: NCT02442531.
Outcome Number of patients (N=) Percentage (%)
Response Rate (RR)PR 1 5
SD 15 68
PD 6 27
Disease control Rate(DCR)
>6 months 48
>12 months 21
>18 months 14

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Therapeutics and Tumor Biology (Nonimmuno) 155s

3028 Poster Session (Board #20), Sat, 8:00 AM-11:00 AM 3029 Poster Session (Board #21), Sat, 8:00 AM-11:00 AM
Prospective cohort study of the impact of hospital-wide dihydropyrimidine PDX validation of a 3D microtumor platform. First Author: Ellen Sampson,
dehydrogenase (DPYD) genotype testing for fluoropyrimidine-based chemo- SageMedic Corp, Redwood City, CA
therapy on adverse events and hospital costs. First Author: Theodore John
Title: Patient-derived xenograft validation of a 3D microtumor platform
Wigle, University of Western Ontario, London, ON, Canada
Background: Patient-derived xenograft (PDX) mouse models are thought to
Background: Fluoropyrimidines remain integral components of modern most closely reflect the biology of a patient’s cancer. Unfortunately, growing
chemotherapy for solid tumors, and their toxicities can be reduced by sufficient tumor in a PDX model takes several months and more often than not,
pretreatment DPYD genotyping. Our main objective was to demonstrate the the tumor fails to grows at all. The SAGE Direct Platform, an in-vitro model, can
feasibility of implementing a hospital-wide pretreatment DPYD testing create hundreds of live microtumors from virtually every patient’s viable biopsy
service based on the CPIC 2013 guideline on fluoropyrimidines and DPYD. and test a panel of clinically relevant drugs within no more than 1 week. Thus,
Methods: We enrolled participants prior to planned fluoropyrimidine treat- concordance of results from a PDX model with results of the SAGE Direct
ment as well as those who had experienced adverse events (AEs) after Platform would support a rational for the platform to be potentially useful to
initiation of therapy, from December 1, 2013 to November 30, 2018. The predict tumor response in cancer patients. Methods: A bladder cancer from a
patients tested pretreatment were analyzed as a prospective cohort to assess 77 year old female was used to establish a PDX model. Mice were divided into
AEs within 90 days of fluoropyrimidine initiation and associated hospital three groups receiving either saline (control), cisplatin, or gemcitabine in-
cost. The primary outcome was the rate of severe global fluoropyrimidine- traperitoneal on the days 1, 8, and 13, and tumor growth was observed. One
related toxicity in the pretreatment cohort (grade$3, CTCAE v.4.0.3). tumor sample was used to create 3D microtumors and those were tested using
Results: Of 1362 patients genotyped for DPYD within the study period 1041 the same drugs. Results: Tumor growth (exceeding 1,000 mm3) was similar
were enrolled pretreatment and included in the primary analysis. The median after cisplatin compared to control (4.8 vs. 3.7 weeks). After gemcitabine
age was 65 years (19-90), 57% male, 51% 5-FU, and 49% capecitabine. tumors initially shrank and only started growing a couple of weeks after the end
Dose reductions were recommended for 21 DPYD variant carriers who were of treatment so that 1,000 mm3 was only reached after 10.2 weeks (p,0.001
detected pretreatment. There was no significant difference in the primary compared to cisplatin and control). In the SAGE Direct Platform the EC50 of
outcome between DPYD variant (29%) and wild type (18%) patients cisplatin was 97.3 mM and thus two orders of magnitudes higher than the
(Fisher’s exact test p = 0.25). Costs associated with ER visits and hospi- EC50 of gemcitabine, which was 0.7 mM. Conclusions: Both the PDX model
talizations at our tertiary care centre were $1,268 (89-8,562) (Median and the SAGE Direct Platform have shown this bladder cancer to be virtually
(IQR)) and $2,961 (341-13,567) for DPYD variant (n = 4) and wild-type (n = resistant to cisplatin while very sensitive to gemcitabine. The next steps of
99) patients respectively. Post-AE genotyping (n = 70) found five DPYD these preliminary data could be to repeat this experimental design with other
variant patients; all experienced grade$3 toxicity, costs were $15,825 tumors and/or to start an observational cohort study in patients correlating the
(10,962-25,310), and one poor metabolizer died due to complications. SAGE Direct Platform results to patient outcomes.
Targeted next generation exome sequencing of DPYD wild-type patients who
experienced severe AEs identified five potentially deleterious genetic vari-
ants in ABC efflux transporters. Conclusions: Pretreatment DPYD genotype
guided dosing of fluorouracil and capecitabine is feasible and benefits
patients, health care providers, and hospitals. Our data supports adoption of
pretreatment DPYD genotyping as a standard of care.

3030 Poster Session (Board #22), Sat, 8:00 AM-11:00 AM 3031 Poster Session (Board #23), Sat, 8:00 AM-11:00 AM
Anticancer activity in patients with advanced ovarian and biliary tract Phase I study of the Aurora A kinase (AurA) inhibitor TAS-119 with paclitaxel
cancers treated with NUC-1031 and a platinum agent. First Author: (P) in advanced solid tumors. First Author: Dana Backlund Cardin,
Sarah Patricia Blagden, University of Oxford, Oxford, United Kingdom Vanderbilt-Ingram Cancer Center, Nashville, TN
Background: The inhibition of cellular nucleotide metabolism to promote Background: AurA is a key regulator of cell division, including mitotic spindle
apoptosis is a key principle of cancer therapy. This, in combination with assembly. However, elevated levels of AurA have been reported to abrogate
platinum-induced DNA-damage, is key to promoting anti-cancer activity in a the mitotic spindle checkpoint activated by taxanes leading to treatment
variety of tumors, including ovarian, biliary tract, lung, breast and bladder. resistance. Preclinical studies of TAS-119 + P showed enhanced antitumor
NUC-1031, a phosphoramidate transformation of gemcitabine is designed activity and suggested an optimal timing window of the combination. Study
to overcome resistance mechanisms that limit the efficacy of this nucleoside objective was to assess the safety of TAS-119 + P in adult patients (pts) with
analog. NUC-1031 has shown broad clinical activity across multiple solid advanced solid tumors. Methods: Dose escalation used a 3+3 design to
tumors as both a single agent and in combination with platinum agents. We determine the maximum tolerated dose (MTD); 7 dose levels (DLs) were
show potential synergism between NUC-1031 and a platinum agent in explored, starting with 25 mg TAS-119 BID dosed 4 days/week (d/wk) and
advanced ovarian (OC) and biliary tract (BTC) cancers. Methods: PRO-002 weekly 90 mg/m2 P for 3 weeks of a 4 week cycle (DL1). Plasma samples were
was a phase Ib study; 25 patients (pts) with recurrent OC who had exhausted collected during cycle 1 to evaluate pharmacokinetics. In expansion, pts with
all other therapy options received NUC-1031 + carboplatin. 17 pts were advanced breast/ovarian cancers were treated at the MTD. Results: Dose
considered platinum resistant (10) or platinum refractory (7). ABC-08 is a escalation enrolled 26 pts with various cancers, the majority being pancreas,
phase Ib study, 14 pts with advanced BTC treated in the first-line setting with colon, and ovarian; 2 pts were not evaluable for DLT assessment and replaced.
NUC-1031 + cisplatin. Results: In PRO-002, strong efficacy signals were A DLT (neutropenia and elevated AST) was observed in 2/3 pts in DL1. Dosing
observed in non-platinum-responsive patients. Of the 17 response-evaluable was modified to 25 mg TAS-119 BID 2 d/wk and P 70 mg/m2 (DL2) and no
platinum-resistant or refractory pts, 5 partial responses (PRs) and 11 stable DLTs were observed. Zero DLTs were observed in the next 4 doses: 70 mg/m2
diseases (SDs) were achieved, resulting in an ORR of 29% and a DCR of P + TAS-119 2 d/wk at 50 mg BID (DL3) or 75 mg BID (DL4), or 80 mg/m2 P +
94%. NUC-1031 + carboplatin was well-tolerated with no unexpected AEs; TAS-119 at 75 mg BID 2 d/wk (DL5) or 75 mg BID 3 d/wk (DL6). TAS-119 was
DLTs were myelosuppression and fatigue. Encouraging response rates were escalated to 100 mg BID 3 d/wk (DL7) and 3 DLTs (n=1, elevated ALT; n=1,
also observed in ABC-08 compared to historical standard of care (ABC-02). diarrhea and mucositis) occurred in 2/3 pts. Three additional pts were then
One CR (7%), 6 PRs (43%) and 1 SD (7%) were observed, resulting in an enrolled at DL6 to confirm the MTD. One breast and 2 ovarian pts were enrolled
ORR of 50%. NUC-1031 + cisplatin was well-tolerated, with no unexpected in expansion before the trial was suspended by the company. Toxicities ob-
AEs or DLTs. Complementary in vitro evidence suggests that the beneficial served in $30% of pts were diarrhea, nausea, and fatigue (most #Gr2).
interaction occurs whereby platinum treatment sensitizes cells to NUC- Plasma TAS-119 exposure increased dose-proportionally in 2 d/wk and 3 d/wk
1031. Conclusions: Increasing evidence suggests that NUC-1031 in schedules; no impact of TAS-119 on PK of paclitaxel was found. The dis-
combination with a platinum agent may have synergistic properties, leading ease control rate was 59% (17/29); 4 of these pts had a partial response.
to enhanced anti-cancer activity. In both OC and BTC, durable tumor Conclusions: The combination had a manageable toxicity profile at the MTD of
shrinkage was observed. This was particularly encouraging in a platinum 80 mg/m2 P + TAS-119 at 75 mg BID 3 d/wk. Preliminary clinical activity was
resistant/refractory OC population. Future studies utilizing both NUC-1031 seen in 59% of pts, including tumor responses in a majority (4/7) of pts with
plus a platinum agent will further elucidate the potential of this therapeutic ovarian/fallopian tube cancers. Clinical trial information: NCT02134067.
combination.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
156s Developmental Therapeutics and Tumor Biology (Nonimmuno)

3032 Poster Session (Board #24), Sat, 8:00 AM-11:00 AM 3033 Poster Session (Board #25), Sat, 8:00 AM-11:00 AM
A phase I study of the oral administration of irinotecan in combination with Risk of QTc interval prolongation among cancer patients treated with tyrosine
the potent P-glycoprotein (P-gp) inhibitor HM30181A. First Author: Antonio kinase inhibitors. First Author: Anan Abdelmoti Abu Rmilah, Mayo Clinic,
Jimeno, University of Colorado, Aurora, CO Rochester, MN
Background: Irinotecan is a prodrug of the potent topoisomerase inhibitor Background: QTc interval prolongation can lead to life-threatening compli-
SN-38. In animals, oral administration of irinotecan with the selective min- cations such as torsade de pointes (TdP), ventricular tachycardia (VT), and
imally absorbed P-gp inhibitor HM30181A increased the bioavailability of sudden cardiac death (SCD). It can occur with various tyrosine kinase in-
irinotecan. Oral administration of irinotecan may also increase the conversion hibitors (TKIs) but comparative analyses on the incidence and complication
to SN-38. Objectives: To determine the MTD and DLT of orally administered rates are scarce. We thus conducted a comprehensive analysis of TKI use and
irinotecan in combination with HM30181A 15 mg on day 1 of a 21-day cycle. QTc prolongation in clinical practice. Methods: We retrospectively reviewed
Additional objectives include determining the recommended phase 2 dose and the electronic medical records of all cancer patients who were treated with TKI
the PK of irinotecan and SN-38. Methods: This was a phase 1 dose escalation between 01/2005 and 12/2018 at our institution. QTc prolongation was
study enrolling cohorts of 3-6 patients with advanced malignancies. Patients defined as a QTc $ 450 ms or 460 ms among male or female patients, re-
had Hb $9 gm/dL, ANC $ 1.5x109/L, platelets $ 100x109/L, adequate hepatic spectively. For each type of TKIs, we determined the administration rate and
and renal function, ECOG 0-1 and were not homozygous for UGT1A1*28. incidence of QTc interval prolongation. We also studied the frequency of QTc
Patients were administered HM30181A 15 mg and oral irinotecan 20, 40, 80, prolongation $ 500 ms, rate of increase of the QTc interval by $ 60 ms, and
120, 160, 200, 240, 280 and 320mg/m2. Results: Thirty male and female the development of complications (VT, TdP and SCD). Results: In the present
patients, mean age 60.9 (range 33-78) were enrolled into this ongoing study. study, we analyzed the data of 685 cancer patients (431 male and 254 fe-
The most common cancers were ovarian (6), colorectal (4), breast (4), endo- male), including 299 patients with RCC, 188 with chronic leukemia, 55 with
metrial (3), and pancreatic (3). The median number of cycles administered was acute leukemia, 65 with thyroid cancer, 48 with lung cancer and 39 with GIST.
3 (range 1-9). Treatment-related Grade 3-4 AEs were experienced by 12 (40%) These patients received 902 TKI administrations and QTc prolongation was
subjects. The most common were nausea 7 (23%), vomiting 6 (20%) and reported in 1/3 of these (289 administrations). The highest frequency was seen
abdominal pain 3 (10%). Treatment-related SAEs were experienced by 6 (20%) with imatinib, nilotinib and dasatinib (30, 40 and 50%). Among cases of QTc
patients (nausea or vomiting in 4 subjects). DLTs occurred in 2 patients at the prolongation, a QTc interval $ 500 ms was documented in 53 (18.3%) and
320 mg/m2 dose level (neutropenia and C. Difficile diarrhea) and additional QTc progression $ 60 ms in 72 (25%). Complications were found in 14 cases
patients are being enrolled at the 280mg/m2 dose level to define the MTD. Acute (5%) including VT in 9, TdP in 2 and SCD in 3 administrations. Conclusions: The
cholinergic diarrhea has not been observed. The best response was stable current findings suggest that TKI therapy leads to QTc prolongation in 1/3 of
disease in 9/21 evaluable patients. PK at the three highest dose levels is patients on average and most commonly with the Bcr-Abl TKIs, imatinib,
summarized below. Conclusions: Oral administration of HM30181A in combi- nilotinib and dasatinib. While SCD is rare (1%) it can still evolve and in 5% of all
nation with irinotecan tablets results in pharmacologically active concentrations QTc prolongations with TKIs are potentially life-threatening. These data support
of SN-38. Confirmation of the MTD when dosed on a 21-day cycle is ongoing. recommendations for serial ECGs in cancer patients undergoing TKI therapy.
Phase 2 studies are being planned. Clinical trial information: NTC02250157. Total Prolonged QTc QTc ‡ 500 QTc progression ‡ 60 VT SCD TdP
Irinotecan SN38 Imatinib 165 54 13 10 2
Dose Nilotinib 75 33 8 19
(mg/m2) tmax Cmax AUC24h tmax Cmax AUC24h t1/2 Dasatinib 115 58 10 16 2 0 1
N=3 (h) (ng/mL) (ng∙h/mL) t1/2 (h) (h) (ng/mL) (ng∙h/mL) (h) Sunitinib 134 31 1 2 1 1
Pazopanib 165 36 5 6 2 1
240 3.0 798 6.960 11.1 3.0 28.3 264 20.0 Axinitinb 45 9 3 3
280 2.0 763 6,040 12.2 1.5 37.6 293 18.5 Sorafenib 41 14 2 2
Cabozantinib 33 9 1 2 1 1 1
320 3.0 1,055 11,297 18.7 2.0 38.5 300 20.4 Others 129 45 10 12 1 0 0

3034 Poster Session (Board #26), Sat, 8:00 AM-11:00 AM 3035 Poster Session (Board #27), Sat, 8:00 AM-11:00 AM
Phase I study of DFP-11207, a novel oral 5-FU with enhanced PK and Associations of insulin-like growth factor binding proteins and adiponectin
improved tolerability, in patients with solid tumors. First Author: Jaffer A. with disease progression and mortality in metastatic colorectal cancer:
Ajani, The University of Texas MD Anderson Cancer Center, Houston, TX Results from CALGB/SWOG 80405 (Alliance). First Author: Brendan John
Guercio, Brigham & Women’s Hospital, Boston, MA
Background: DFP-11207 combines a 5-fluorouracil (5-FU) pro-drug with a
reversible dihydropyrimidine dehydrogenase inhibitor and a potent inhibitor of Background: Energy balance-associated biomarkers such as insulin-like growth
orotate phosphoribosyl transferase resulting in enhanced pharmacological factors (IGFs) and IGF-binding proteins (IGFBPs) have been associated with risk
activity of 5-FU with decreased gastrointestinal and myelosuppressive toxicity. and prognosis of various malignancies. Their relationship to disease progression
Methods: Patients with advanced solid tumors were treated in this single arm, and mortality in metastatic colorectal cancer (mCRC) requires further study.
dose escalation study. Accelerated dose escalation using single pt cohorts was Methods: In a prospective cohort study, baseline plasma IGFBP-3, IGFBP-7, C-
followed until drug-related Gr2 adverse events occurred; then a 3+3 design peptide, IGF-I, and adiponectin were measured at trial registration among 1,086
was followed to determine maximum tolerated dose (MTD). Pts were dosed patients participating in a National Cancer Institute-sponsored clinical trial of
daily in 28-day cycles until intolerable toxicity or disease progression. PK first-line therapy for mCRC. We used Cox proportional hazards regression to
sampling was performed for DFP-11207 metabolites on all pts and a 6 pt adjust for confounders and examine associations of biomarkers with overall (OS)
cohort received DFP-11207 in fed and fasted states to determine drug bio- and progression-free survival (PFS). Results: Higher plasma IGFBP-3 was as-
availability. Results: Primary tumors among the 23 enrolled pts included sociated with longer OS (adjusted Ptrend , .001) and PFS (adjusted Ptrend =
esophageal, colorectal, gastric, pancreatic and gallbladder. Seventeen pts .003). Compared to patients in the lowest IGFBP-3 quintile, patients in the
were treated at 8 dose levels of oral DFP-11207 administered daily, ranging highest quintile experienced an adjusted HR for all-cause mortality of 0.58
from 40 to 440 mg/m2/d. At 440 mg/m2/d one pt experienced drug-related Gr3 (95% CI 0.42 to 0.78) and for disease progression or mortality of 0.60 (95% CI
dehydration and mucosal inflammation, and Gr4 febrile neutropenia; a second 0.45 to 0.82). Higher plasma IGFBP-7 was associated with shorter OS (adjusted
pt experienced Gr4 febrile neutropenia. One DLT of Gr3 vomiting occurred Ptrend , .001) and PFS (adjusted Ptrend = .02). Compared to patients in the
among 6 pts treated at 330 mg/m2/d dosed q12hrs confirming the MTD. Six lowest IGFBP-7 quintile, patients in the highest quintile experienced an ad-
pts were administered 300 mg DFP-11207 q12 hrs in a fed or fasted state to justed HR for all-cause mortality of 1.52 (95% CI 1.24 to 1.88) and for disease
determine drug bioavailability. Among all pts, the most frequently reported progression or mortality of 1.28 (95% CI 1.05 to 1.57). C-peptide and IGF-I were
drug-related TEAEs were fatigue (47.8%), nausea (47.8%), decreased appetite not significantly associated with patient outcomes (adjusted Ptrend = .73 and .30
(39.1%), diarrhea (26.1%), vomiting (21.7%), anemia (13.0%), dysgeusia for OS). Adiponectin was not associated with OS; there was a U shaped as-
(13.0%), mucosal inflammation (13.0%) and palmar-plantar erythrodysaesthesia sociation between adiponectin and PFS, wherein low and high values were
syndrome (13.0%). PK analyses of pts treated at 330 and 440 mg/m2/d indicate associated with shorter PFS (Pnon-linear trend = .03). Conclusions: In patients with
5-FU concentrations of ~20 ng/mL throughout the dose cycle. There was no mCRC, high plasma IGFBP-3 and low IGFBP-7 were associated with reduced
substantial difference in DFP-11207 bioavailability among pts in a fed or fasted risk of disease progression and mortality. These data suggest that energy-balance
state. Out of 21 efficacy evaluable pts, 7 pts had stable disease (33.3%), of which associated biomarkers may offer prognostic and biologic insights into mCRC.
2 had prolonged stable disease of . 6 months. No pts achieved CR or PR. Support: U10CA180821, U10CA180882, BMS, Genentech, Pfizer, Sanofi;
Conclusions: DFP-11207 at the dose of 330 mg/m2/d q12hrs is well-tolerated in https://acknowledgments.alliancefound.org.
pts with solid tumors with mild myelosuppressive and gastrointestinal side effects,
and results in circulating 5-FU levels conducive to an anti-tumor effect. DFP-
11207 can be explored as monotherapy or substitute for 5-FU, capecitabine or S-
1 in combination treatment regimens. Clinical trial information: NCT02171221.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Therapeutics and Tumor Biology (Nonimmuno) 157s

3036 Poster Session (Board #28), Sat, 8:00 AM-11:00 AM 3037 Poster Session (Board #29), Sat, 8:00 AM-11:00 AM
Predictive and prognostic values of circulating tumor DNA (ctDNA) clearance Progastrin, a novel ubiquitous cancer blood biomarker for early detection and
in osimertinib treated advanced non-small cell lung cancer cohort. First Au- monitoring. First Author: Benoit You, Institut de Cancérologie des Hospices
thor: Yong Song, Nanjing General Hospital of Nanjing Military Command Civils de Lyon (IC-HCL), CITOHL, EMR UCBL/HCL 3738, Lyon, GINECO &
(NGH)- Jinling Hospital, Nanjing, China GINEGEPS, France, Lyon, France
Background: Although growth advantage of certain clones would ultimately Background: The successes of recent publications on “multi-tumor” circu-
translate into a clinically visible disease progression, radiological imaging lating markers highlight the relevance of novel universal diagnostic cancer
does not reflect clonal evolution at the molecular level. CtDNA, validated as a serum biomarkers. Since the Wnt/ß-catenin/Tcf4 pathway, activated in many
tool for mutation detection in lung cancer, reflects dynamic molecular tumors, induces the GAST Gene encoding progastrin synthesis, we hypothe-
changes. Here, we evaluated the potential of ctDNA in monitoring molecular sized that progastrin, easily measurable in the blood, might be a “multi-tumor”
changes and predicting clinical outcomes of EGFR T790M-positive osi- diagnostic biomarker. Methods: Progastrin levels were measured in the blood
mertinib treated NSCLC pts. Methods: This prospective multicenter study, samples of 1319 patients with 12 different cancer origins, and compared to
enrolled 72 T790M positive osimertinib-treated advanced NSCLC pts who those of 557 asymptomatic 18-75 years old blood donors. Moreover the
progressed on prior EGFR-TKI to evaluate the potential of ctDNA in monitoring, longitudinal kinetics of progastrin concentrations were serially assessed during
is part of the ongoing ASTRIS study (NCT02474355). Longitudinal plasma treatments in 168 patients with ovarian cancers enrolled in the randomized
samples, collected from 52 pts, were subjected to sequencing using a panel CHIVA trial (NCT01583322, GINECO), 191 patients with peritoneal involvement
consisting of 168 lung cancer-related genes. Results: Genomic profile prior to from gastro-intestinal cancers enrolled in BIG-RENAPE trial (NCT03787056),
the initiation of osimertinib revealed that mutations participating in cell cycle and in 95 HCC patients. The progastrin was measured using an ELISA test
(14 pts, p = 0.004) and P53 pathways (43 pts, p = 0.032) were associated developed by ECS Progastrin (Prilly, Switzerland). Results: Compared to healthy
with shorter OS (p53 was excluded from analysis due to high mutation fre- blood donors, progastrin was found at higher concentrations in the plasma of
quency). Interestingly, pts with undetectable ctDNA at first follow-up (within cancer patients: median 4.47 vs 0.20 pM, P , 0.0001; diagnostic discriminative
50 d, n = 41) were correlated with longer PFS (p = 0.009) and OS (p = 0.022). power, ROC analysis AUC = 0.86 (95% CI, 0.83-0.89; P , 0.0001). Progastrin
With a median follow-up of 168 d (ranged from 40 - 550 d), 32 pts experienced levels were found elevated in all cancer groups, regardless of disease stages, and of
radiological disease progression. Among them, 11 (34%) experienced mo- pathology origins: ROC AUCs ranged from 0.71 to 0.93, all P , 0.0001 (Table).
lecular progression reflected by emergency of new mutation or increased allelic The longitudinal progastrin changes during treatments, suggest relationships to
frequency of existing mutation prior to radiological progression, with an average tumor burden, and potential monitoring value. Conclusions: Progastrin is a novel
leading time of 74 days. Pts with molecular PD prior to radiological PD were ubiquitous cancer biomarker, easily detectable in the blood using an affordable
more likely to harbor any gene copy number amplification (CNA, p = 0.035) ELISA test (CancerRead Lab test(R)). It may change the future paradigms about
and p53 (p = 0.023) mutations at radiological PD. In addition, pts with CNA screening (in particular for populations at higher or lower risks of cancer), cancer
at radiological PD had shorter PFS (p = 0.002) and OS (p = 0.052). diagnostic & monitoring.
Conclusions: This clinical trial study demonstrates that ctDNA clearance at
Oesophagus/Gastric Colorectal Pancreas HCC
first follow-up can serve as a predictive and a prognostic marker for pts un- Gastro-Intestinal 0.95 [0.92-0.96] 0.84 [0.80-0.89] 0.93 [0.88-0.98] 0.85 [0.79-0.90]
dergoing osimertinib treatment. Furthermore, it revealed the potential of
Genito-Urinary Prostate Kidney
ctDNA in early detection of disease progression, preceding imaging modalities 0.82 [0.77-0.87] 0.92 [0.89-0.95]
with an average lead time of 74 days. Gynecology Breast Ovarian Uterus
0.81 [0.75-0.87] 0.83 [0.79-0.88] 0.81 [0.75-0.88]
Others Lung Melanoma Brain
0.88 [0.82-0.93] 0.83 [0.75-0.91] 0.71 [0.64-0.79]

3038 Poster Session (Board #30), Sat, 8:00 AM-11:00 AM 3039 Poster Session (Board #31), Sat, 8:00 AM-11:00 AM
Can the enumeration of circulating tumor cells (CTCs) and the characterization Development and analytical validation of a 523-gene clinical assay for cell-free
of circulating tumor DNA (ctDNA) provide insight into organ tropism in met- DNA. First Author: Robin Harrington, Molecular Characterization Laboratory,
astatic breast cancer (MBC)? First Author: Lorenzo Gerratana, Department of Frederick National Laboratory for Cancer Research, Frederick, MD
Medicine-Hematology and Oncology, Feinberg School of Medicine, North-
Background: Liquid biopsies are emerging as a powerful complement to tumor
western University; Department of Medicine (DAME), University of Udine,
biopsies for the clinical management of cancer patients. A large gene panel
Chicago, IL
with robust analytical performance that accurately assesses variants, tumor
Background: Liquid biopsy provides real-time data about prognosis and ac- mutational burden (TMB), and microsatellite instability in plasma would be of
tionable mutations in MBC. The aim of this study was to explore the combination high value for immunotherapy studies, monitoring minimal residual disease
of ctDNA analysis and CTCs enumeration in estimating target organs more and early cancer detection. To this end, we have completed the initial vali-
susceptible to MBC involvement. Methods: This retrospective study analyzed 85 dation for the cell-free DNA (cfDNA) assay, TruSight Oncology 500 (TSO500),
MBC patients (pts) characterized for both CTCs and ctDNA at baseline. CTCs which interrogates the full coding region of 523 genes plus selected intronic
were isolated through the CellSearch kit (Menarini Silicon Biosystems, PA), regions for fusion detection in 23 driver genes. Methods: Cell-free DNA was
while ctDNA was analyzed using the Guardant360 NGS-based assay (Guardant extracted from plasma collected from Streck or EDTA blood tubes and quan-
Health, CA). Pts with $ 5 CTC/7.5 ml of blood were defined as Stage IV ag- titated to achieve an assay input of $10 ng. Libraries were constructed using
gressive as previously reported (Cristofanilli et al 2019). Statistical associations unique molecular identifiers (UMIs) and duplex barcodes for error correction,
were explored through uni- and multivariate logistic regression and Fisher’s exact then enriched by target capture and sequenced on a NovaSeq 6000. Healthy
test. Results: 37% of pts were diagnosed with hormone receptor positive donor (HD) specificity assessment used matched white blood cell results to filter
(HRpos) MBC, 26% with HER2-positive MBC and 37% with triple negative MBC germline and clonal hematopoiesis variants. Contrived specimens were used to
(TNBC), 28 pts (33%) were defined as stage IV aggressive. The most observed evaluate sensitivity. Single nucleotide variants (SNVs) (n = 36), insertion/
metastatic sites were bone (37%), lymph nodes (29%), lung (27%) and liver deletions (indels) (n = 19), copy number variants (CNVs) (n = 6), and fu-
(25%). In multivariate analysis, IBC and ESR1 mutations were the only sig- sions (n = 5) were tested in 2 multi-site replicates. Results: Sensitivity of
nificant factors associated with liver metastases (respectively, OR 0.12, P = detection at 0.5% variant allele fraction (VAF) was . 95% and . 97% for SNVs
0.038 and OR 24.01, P = 0.019), while no associations were found with respect and indels, respectively. All expected CNVs were identified at the targeted
to lung localizations. Intriguingly, all HRpos MBC pts with ESR1 mutations had threshold of $1.3X change and showed strong correlation with matched digital
bone metastases (P = 0.022), while IBC and Stage IV aggressive were in- PCR results. All fusions were identified at $0.4% VAF. Specificity in HD was
dependently associated with bone metastases (respectively OR 0.10, P = 0.006 . 99.99%. In 22 temporally matched tumor and blood samples from late-stage
and OR 19.92, P = 0.003). FGFR1 and NF1 were associated with lymph node patients, 58% of all reportable mutations in tumor were identified in cfDNA.
localizations (OR 3.68, P = 0.046, OR 4.39, P = 0.031, respectively), while Preliminary TMB analysis identified one TMB high case with POLE p.P286R
CDK6 and TP53 alterations were associated with serosal involvement (OR observed in both tissue and cfDNA. Conclusions: In this initial validation study
14.34, P = 0.029, OR 0.08, P = 0.031, respectively). Notably, TNBC and IBC the TSO500 cfDNA assay exhibited high sensitivity and specificity consistent
were both associated with soft tissue spreading (respectively OR 3.7, P = 0.011, with requirements for clinical applications. Ongoing studies will further evaluate
OR 2.79, P = 0.018). Conclusions: These results suggest that ctDNA and CTCs TSO500 as a complement or potential alternative to tissue biopsy for the ge-
enumeration could give useful insights on MBC organotropism, suggesting a nomic profiling of cancer patients.
possible role for future monitoring strategies that dynamically focus on high-risk
organs defined by tumor biology.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
158s Developmental Therapeutics and Tumor Biology (Nonimmuno)

3040 Poster Session (Board #32), Sat, 8:00 AM-11:00 AM 3041 Poster Session (Board #33), Sat, 8:00 AM-11:00 AM
Dynamic monitoring of circulating tumor DNA next-generation gene sequencing Whole-genome cell-free DNA (cfDNA) changes as a dynamic blood-based
as a predictive biomarker of response and progression-free survival after biomarker for early response assessment of advanced tumors. First Author:
pembrolizumab monotherapy in patients with advanced NSCLC. First Andrew A. Davis, Robert H. Lurie Comprehensive Cancer Center, Feinberg
Author: Charu Aggarwal, University of Pennsylvania, Philadelphia, PA School of Medicine, Northwestern University, Chicago, IL
Background: Circulating tumor DNA next generation sequencing (ctDNA Background: Liquid biopsies have potential clinical utility as dynamic bio-
NGS) is increasingly being used to detect mutations (MT) in patients (pts) markers for treatment response. We analyzed serial changes in whole-genome
with metastatic (m) NSCLC. Limited data exist on the correlation of baseline (WG) cfDNA to identify patients with disease progression prior to routine
ctDNA NGS profile and serial ctDNA NGS monitoring to response to im- imaging. Methods: We prospectively collected clinical data and blood from 69
munotherapy. Methods: We conducted a prospective study in pts with advanced cancer patients (28 lung, 25 breast, 16 other). Blood was collected
mNSCLC receiving pembrolizumab monotherapy. Plasma was collected at at baseline prior to initiation of a new treatment and at one or two additional
weeks 0 (T0), 9 (T1), and 18 (T2). ctDNA NGS was performed using a 73 timepoints (median 21 and 42 days). We isolated plasma cfDNA and prepared
gene panel. Number of MTs and variant allelic fraction (VAF) were de- sequencing libraries for WG sequencing or WG bisulfite sequencing (median
termined at baseline, and serially; change in mean VAF was calculated depth 20X). We quantified changes in the fraction of tumor-derived cfDNA over
between T1-T0, and T2-T0. Response rate (RR) was assessed using RECIST the initial course of treatment to predict progression vs. no progression.
1.1. Correlations were made for pt characteristics, RR, progression free Treatment response at first post-treatment imaging was determined by RECIST
survival (PFS), and overall survival (OS). Results: We analyzed 95 samples 1.1 and clinical assessment. Study endpoints were agreement with first post-
from 33 pts, 21 female, median age 69 (range 51-89 years), smokers (n = treatment imaging and progression-free survival (PFS) by cfDNA prediction.
29), adenocarcinoma (n = 23), 25 pts enrolled at initial diagnosis, majority Results: Median age of patients was 70 and 59% were female. Patients were
had high PD-L1 $50% (n = 29, 88%). At T0, 32 pts had detectable MT, treated with the following therapies: chemotherapy (37), immunotherapy (17),
median number of MTs was 4 (range 0-21), (non-synonymous MT = 3), most endocrine (9), or targeted therapy (6). Patients with predicted progression by
common MT was TP53 (n = 21). Confirmed PR was 27% (n = 9), clinical cfDNA (14), indicated by an increase in tumor fraction at either post-treatment
benefit rate (SD+PR) was 64% (n = 21), and 2 pts were not evaluable for blood collection, had shorter PFS (median 63 days) compared to patients
response. Smokers were more likely to have higher number of MT at T0 (4 vs. without an increase (N = 55; median 255 days), with hazard ratio of 10.3 (95%
1 p = 0.003); there was no correlation with smoking and overall RR (p = confidence interval 4.6-23.4, log-rank P = 1x10-11). Positive predictive value
0.17). RR was not related to number of MT at T0, p = 0.37. A decrease in was 100% for disease progression and negative predictive value was 78%.
ctDNA VAF was seen in 6/9 pts with PR (mean VAF change range -0.11 to These findings were consistent in subset analyses of patients with lung (log-
-0.001); 2/5 pts with PD showed an increase in mean VAF while 3 showed rank P = 2x10-5), breast (log-rank P = 3x10-4), and those treated with im-
decrease. At median follow up of 9.26 months (mos), median PFS and OS munotherapy (log-rank P = 5x10-6). Conclusions: Our results show the ability
were 7.4 and 10.5 mos, respectively. Median PFS was longer for pts with a to detect early disease progression with high fidelity using WG cfDNA prior to
decrease in ctDNA VAF at both T1-T0 (8.9 vs. 5 mos, p = 0.02) and T2-T0 first imaging. These findings were consistent across multiple tumor types and
(9.1 vs. 5.5 mos, p = 0.006). OS and additional biomarker analyses in- treatments, including immunotherapy patients. Once validated, this dynamic,
cluding correlation of response to a 2mb ctDNA plasma-based NGS panel predictive, blood-based biomarker could aid in clinical decision making for
will be reported at the meeting. Conclusions: Our results demonstrate that it early treatment change as a novel and cost-effective approach.
is feasible to serially monitor plasma NGS, decline in mean ctDNA VAF
correlates with radiographic response and PFS on immunotherapy with
pembrolizumab.

3042 Poster Session (Board #34), Sat, 8:00 AM-11:00 AM 3043 Poster Session (Board #35), Sat, 8:00 AM-11:00 AM
A prospective study tracking longitudinal changes in genome-wide cell-free NMR-metabolite-resonance signature to predict HR+ breast cancer patient
DNA (cfDNA) methylation to identify early nonresponders to cancer response to CDK4/6 inhibitors. First Author: Bo Zhang, Olaris Therapeutics,
treatment. First Author: Andrew A. Davis, Robert H. Lurie Comprehensive Cambridge, MA
Cancer Center, Feinberg School of Medicine, Northwestern University,
Background: Advanced ER+ breast cancer patients have reported pro-
Chicago, IL
longation of stable disease when treated with a CDK4/6 inhibitor as a
Background: Methylation is an epigenetic modification linked to cancer monotherapy or in combination with endocrine treatment. However ~20% of
pathogenesis. The aim was to determine if changes in cfDNA methylation patients are intrinsically resistant and all patients eventually acquire re-
patterns before and after initiation of treatment could predict non-response sistance to these therapies. There is a critical need to identify biomarkers
to treatment prior to routine imaging and clinical follow-up. Methods: We that accurately predict response and resistance to CDK4/6 inhibitors.
prospectively collected clinical data and blood from 28 patients with ER-positivity, luminal patterns of gene expression, Rb function, overexpression
metastatic malignancies (13 lung, 11 breast, 4 other). Blood was drawn prior of cyclin D1, cyclin E, CDK6 and low levels of p16 are biomarkers that do not
to start of a new treatment, after first cycle (median 30 days), and/or second accurately match clinical outcomes. Methods: We performed a retrospective
cycle (median 57 days). We performed whole-genome (WG) bisulfite se- study analyzing plasma-based metabolites from a baseline (pre-dose) and
quencing (median depth 18X) on plasma cfDNA to determine methylation ~2 months post treatment of 21 women with estrogen-receptor-positive (ER+)
levels. By tracking how methylation levels deviate from unaffected in- metastatic breast cancer treated with CDK4/6 inhibitors. Results: By corre-
dividuals, from baseline to subsequent timepoints, we classified patients as lating the metabolite expression profiles to clinical outcomes we were able to
either progressors (greater deviance) or non-progressors. Treatment response identify a metabolic signature that could differentiate the CDK4/6 responders
at first follow-up imaging (FUI) was determined by RECIST 1.1. Study and resistant patients with a predictive accuracy of . 90%. Further we were
endpoints were agreement with first FUI and progression-free survival (PFS) able to identify independent signatures predictive of response for individual
by cfDNA classification. Results: The cohort consisted of 68% females and CDK4/6 inhibitors palbociclib and ribociclib. Conclusions: The results of this
the median age was 70. Main treatment regimens were chemo- (N = 12), study could lead to a paradigm shift in the administration of CDK4/6 inhibitors
immuno- (6), endocrine (5), or targeted-therapy (5). PFS was significantly wherein prior to treatment and during treatment patient plasma is screened to
shorter (log-rank p = 8 x 10-7) in patients classified as progressors by cfDNA determine whether that individual patient is responsive or resistant to a CDK4/
(N = 8; median: 62 days) compared to non-progressors (N = 20, median: 6 inhibitor.
263 days). For patients classified as progressors, the cfDNA assay preceded
imaging and clinical evaluation by a median of 34 days. 7 out of 8 patients
classified as cfDNA progressors were later confirmed to progress at first
follow-up evaluation (88% positive predictive value). The one patient who
was classified as progressor based on cfDNA was stable based on FUI (day 93
of treatment) but was later confirmed as progression on day 128 by FUI. For
the remaining patients, 18 of 20 did not progress (90% negative predictive
value). Thus, sensitivity for the assay for identifying progression was 78%
and specificity was 95%. Conclusions: Our results show that WG cfDNA
methylation change is a novel signature with potential to identify patients
whose treatment regimen is ineffective prior to imaging.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Therapeutics and Tumor Biology (Nonimmuno) 159s

3044 Poster Session (Board #36), Sat, 8:00 AM-11:00 AM 3045 Poster Session (Board #37), Sat, 8:00 AM-11:00 AM
Mastocheck: Notable plasma protein biomarker for diagnosis of breast Circulating bacterial DNA as a tool towards noninvasive biomarkers for
cancer in the real clinical practice by using multiple reaction monitoring- colorectal adenocarcinoma and adenoma. First Author: Ke-Feng Ding,
based mass spectrometry. First Author: Yumi Kim, Department of Surgery, Department of Surgical Oncology, Second Affiliated Hospital, and the Key
Seoul National University College of Medicine, Seoul, South Korea Laboratory of Cancer Prevention and Intervention, China National Ministry of
Education, Zhejiang University College of Medicine, Hangzhou, China
Background: Breast cancer is the most frequently diagnosed cancer and the
most leading cause of cancer-related deaths among women worldwide. Background: The gut microbiota is closely associated with the progression of
Although screening mammography is available, there is an ongoing interest colorectal neoplasia. While most metagenomics studies utilized fecal
in improved early detection and prognosis. And also, serum tumor marker samples, circulating bacteria DNA in colorectal adenocarcinoma (ADC) or
levels, such as CA 15-3 and others, may reflect disease progression and adenoma (ADM) patients remain unexplored. This study aimed to charac-
recurrence, they have not proven to be sensitive for early disease detection. terize microbiota DNA in plasma samples and build a machine-learning
Research investigating biomarkers for early detection, prognosis and the model for ADC and ADM early detection. Methods: In this proof-of-concept
prediction of treatment responses in breast cancer is rapidly expanding. study, we performed whole genome sequencing (~30X) of plasma samples
However, no validated biomarker currently exists for use in routine clinical from 25 ADC patients, 10 ADM patients, and 22 healthy controls (HC).
practice, and breast cancer detection and management remains dependent Significant biomarkers were identified in the discovery cohort (12 ADC and
on invasive procedures We aimed to develop biomarker for diagnosis of 11 HC) and built into a random-forest model which was tested in the val-
breast cancer in the real clinical practice by using proteomics technology. idation cohort (13 ADC and 11 HC). These biomarkers were further examined
Methods: Based on our previous studies, we performed verification and in ADM and tested for abundance difference with ADC and HP. Results: In
validation of 124 candidate proteins by using proteomics approach. Among the discovery cohort, 111 species had increased relative abundance in ADC
these 124 candidate proteins, the three proteins (neural cell adhesion compared to HC and 165 species had decreased relative abundance. Al-
molecule L1-like protein, apolipoprotein C-1, carbonic anhydrase-1) with teration in several species such as Flavobacterium and Ruminococcus
highest statistical significance were selected. We created the performance torques were consistent with previously published results in faecal and gut
algorithm of the 3-protein diagnostic model to predict of the breast cancer. microbiome samples. The random forest-recursive feature elimination model
We performed several experiments for establishment and validation of cut- selected 28 significant species from the discovery cohort (mean AUC = 0.98,
off value. Furthermore we conducted test for acquisition of sample stability repeated 2-fold cross-validation) and yielded an AUC of 1 in the validation
and more experiments to achieve the reproducibility and level of evidence, cohort. Interestingly, most biomarker species in ADM patients, with abun-
compared with other cancers (colon, thyroid, ovary, pancreas and lung dance intermediate between ADC and HC, were distinguished from HC. To
cancer) and established effect of anesthesia. Results: Total 1226 samples further test the clinical utility of this model, sequencing data were randomly
(532 patients of breast cancer, 562 healthy women and 100 sample of other down-sampled to 1X and the model performance remained robust (AUC = 1)
cancers) was analyzed. The sensitivity, specificity and accuracy from con- at distinguishing ADC and ADM from HC. Conclusions: This study is the first
firmation experiment was 71.58%, 85.25% and AUC 0.8323, respectively. effort to characterize circulating bacteria DNA in patients with ADC and
The result of comparison with other cancers, there are no statistical sig- ADM. Our findings revealed significant difference in relative abundance of
nificant difference and no relevance with effects of anesthesia. With these several bacterial species between ADC, ADM and HC. A predictive model
results, we recently got permission it to use for in vitro diagnostic use from constructed with selected microbial features accurately distinguished ADC
Korea Food and Drug Administration. Conclusions: In this study, we and ADM from HC. Circulating bacteria biomarkers represent potential non-
developed a plasma protein biomarker that may help to diagnosis of breast invasive tools for early diagnosis of colorectal neoplasia.
cancer in the real clinical practice. By using MRM approach, the 3-protein
biomarker was validated in an independent cohort with acceptable accuracy
for early diagnosis of breast cancer.

3046 Poster Session (Board #38), Sat, 8:00 AM-11:00 AM 3047 Poster Session (Board #39), Sat, 8:00 AM-11:00 AM
Comprehensive genomic profiling of circulating cell-free DNA (cfDNA) Tumor specific DNA in bronchial lavage as a new diagnostic tool in lung
distinguishes focal amplification (amp) from aneuploidy among MET amps cancer. First Author: Caroline Brenner Thomsen, Department of Oncology,
in diverse advanced cancer types. First Author: Yuichi Kumaki, Tokyo Vejle Hospital, Vejle, Denmark
Medical and Dental University, Tokyo, Japan
Background: A considerable fraction of lung cancer patients raise diagnostic
Background: MET amps can occur from focal gene copy number gain (e.g. challenges requiring invasive procedures with a certain risk of complications.
MET-driven) or gain of chromosome 7 (e.g. aneuploidy); however, the Therefore, new diagnostic tools are of major interest. Aberrant methylation of
contribution of each to MET amp is not well established. MET inhibitor- the HOXA9 gene occurs in almost all malignant lung tumors and HOXA9
sensitive lung cancers harboring high-level MET amp have been reported in methylated DNA (meth-ctDNA) is shed into the circulation. The present
the absence of other sensitizing MET alterations (alts), e.g. exon 14 skip- study aimed at a prospective investigation of the possible diagnostic value of
ping, particularly among those with higher MET to chromosome 7 ratios. HOXA9 meth-ctDNA in bronchial lavage (BL). Methods: Patients enrolled
Methods: 3,114 samples from 2,902 Asian patients with advanced solid were referred from the general practitioner suspecting lung cancer. The
tumors were tested with a comprehensive cfDNA NGS panel (Guardant360) diagnostic package according to national guidelines includes chest and
between Oct 2015-Dec 2018. This 70-73 gene assay evaluates single abdominal CT scan, bronchoscopy, relevant blood tests, and histopatho-
nucleotide variants (SNV), selected insertion-deletions (indels), fusions, and logical or cytological verification. Twelve ml liquid was collected at bron-
copy number gains. Focal amp was determined bioinformatically as having choscopy for analysis of meth-ctDNA based on ddPCR technology according
statistically higher copy number relative to other genes, such as BRAF, or to our published method. The analysis was performed blinded to the clinical
CDK6, in the same chromosome arm. Results: MET alts associated with data and compared to the final diagnosis. Results: Eighty-nine patients were
aberrant signaling were found in 223 pts (7.7%) with 18 different cancer consecutively included from the 1 November 2018 to 31 January 2019.
types, most commonly lung (128/1,678), colorectal (36/349), and prostate Fifty-six patients (62.9%) were diagnosed with lung cancer and 33 (37.1%)
(11/48). Among 223 pts, 189 pts (84.8%) had amps, 38(17.0%) had exon with a variety of benign diseases. Meth-ctDNA was found in 42/56 of the
14 skipping, and 8 (3.6%) had activating SNVs. 39.7% of MET amp was patients with a malignant tumor, sensitivity = 75.0% (95%CI=61.6-85.6%),
focal but differed by cancer type; highest prevalence was in gastroesoph- whereas 31/33 of the patients without cancer were negative, specificity =
ageal (80%) and lowest in prostate cancers (9%). Samples with focal MET 93.9% (95%CI= 79.8-99.3%). Table summarizes the results. The false
amp had higher plasma copy number compared to those with non-focal MET negative samples were mainly from patients with peripheral tumors. The two
amp (mean 5.8 vs. 2.5; p , 0.0001) and lower total number of alts per false positive patients included one patient with Cryptogenic Organizing
sample (8.8 vs. 11; p = 0.0122). Focal MET amp was more common than Pneumonia and one with unspecific nodule. Conclusions: The presence of
non-focal MET amp among 419 EGFR mutated samples (6.9% vs. 3.8%, p = meth-ctDNA in BL has a high sensitivity and specificity. If validated, the
0.05) suggesting focal MET amp may be biologically more relevant as a analysis represents a valuable adjunct in the diagnosis of lung cancer. Po-
mechanism of EGFR TKI resistance. Conclusions: This is the first study to tentially, it could save the patients from numerous examinations with potential
use cfDNA to examine focal vs. non-focal MET amp. Focal MET amp harmful risks and ensure a fast diagnosis. The relation between meth-ctDNA
accounted for ~40% of all MET amps, was found in 2.6% of pts with diverse and final lung cancer diagnosis (N= 89).
cancers, was associated with higher plasma copy number, and found in a Final lung cancer diagnosis.
higher proportion of EGFR mutated lung cancer samples. The ability to
differentiate may be clinically relevant given higher MET to chromosome 7 Presence of
meth-ctDNA Yes No Total
ratios have been associated with improved therapeutic response.
Positive 42 2 44
Negative 14 31 45
Total 56 33 89

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
160s Developmental Therapeutics and Tumor Biology (Nonimmuno)

3048 Poster Session (Board #40), Sat, 8:00 AM-11:00 AM 3049 Poster Session (Board #41), Sat, 8:00 AM-11:00 AM
Polymorphisms in the dopamine (DA) signaling to predict outcome in Genome-wide cell-free DNA (cfDNA) methylation signatures and effect on
patients (pts) with metastatic colorectal cancer (mCRC): Data from TRIBE, tissue of origin (TOO) performance. First Author: Minetta C. Liu, Mayo Clinic,
MAVERICC, and FIRE-3 phase III trials. First Author: Francesca Battaglin, Rochester, MN
Division of Medical Oncology, USC Norris Comprehensive Cancer Center,
Background: For multi-cancer detection using cfDNA, TOO determination is
Keck School of Medicine, Los Angeles, CA
critical to enable safe and efficient diagnostic follow-up. Previous array-based
Background: Strong evidence supports the critical role of the gut-brain axis in studies captured , 2% of genomic CpGs. Here, we report genome-wide
modulating gastrointestinal function and homeostasis. Available data suggest fragment-level methylation patterns across 811 cancer cell methylomes rep-
an involvement of the dopaminergic pathway in CRC dynamics. DA could resenting 21 tumor types (97% of SEER cancer incidence), and define effects
inhibit proliferation and migration of tumor endothelial cells and enhanced 5- of this methylation database on TOO prediction within a machine learning
fluorouracil efficacy in CRC preclinical models. Hence, we hypothesized that framework. Methods: Genomic DNA from 655 formalin-fixed paraffin-embedded
genetic variants in DA signaling may predict treatment outcomes in mCRC pts. (FFPE) tumor tissues and 156 isolated cells from tumors was subjected to a
Methods: The impact on outcome of 22 selected single nucleotide poly- prototype 30x whole-genome bisulfite sequencing (WGBS) assay, as previ-
morphisms (SNPs) in 9 genes of the DA signaling pathway (DRD1, DRD2, ously reported in the Circulating Cell-free Genome Atlas (CCGA) study
DRD3, DRD4, DRD5, TAAR1, SLC6A3, SLC18A2, PPP1R1B) was analyzed (NCT02889978). Two independent TOO models, one with and one without the
on a total of 884 pts enrolled in three independent randomized first-line trials: methylation database, were fitted on training samples; each was used to predict
TRIBE (n = 324), MAVERICC (n = 324), and FIRE-3 (n = 236). Genomic DNA on the test set. A WGBS classifier was used to detect cancer at 98% specificity;
from blood samples of pts was genotyped through the OncoArray, a custom reported TOO results reflect percent agreement between predicted and true TOO
array manufactured by Illumina. A meta-analysis approach using the META- among those detected cancers (166 cases: 81 stage I-III, 69 stage IV, 16 non-
SOFT software was used to quantify SNPs prognostic effects and heteroge- informative). Results: Genome-wide methylation data generated from this da-
neities across treatment arms. P values were adjusted for multiple testing using tabase allowed fragment-level analysis and coverage of ~30 million CpGs across
the false discovery rate (FDR) method. Results: Overall, DRD3 rs3732790, the genome (~60-fold greater than array-based approaches). Incorrect TOO
rs9817063 and rs2134655 showed a significant nominal p value (P) in assignments decreased by 35% (20% to 13%) after incorporating methylation
association with tumor response (TR) across trials (P= 0.032, P= 0.021, P= database information into TOO classification. Improvement was observed across
0.027, respectively). TAAR1 rs8192620 showed an association with both all cancer types and was consistent in early-stage cancers (stage I-III). Respective
progression free survival (PFS) (P= 0.01) and overall survival (OS) (P= 0.033), performances in breast cancer (n = 23) were 87% vs 96%; in lung cancer (n =
similar to DA transporter SLC6A3 rs6347 (P= 0.016 and P= 0.002, re- 32) were 85% vs 88%; in hepatobiliary (n = 10) were 70% vs 90%; and in
spectively). SLC6A3 rs6347 association with OS remained significant after pancreatic cancer (n = 17) were 94% vs 100%. Results using an optimized
FDR (PFDR= 0.045). Subgroup analyses showed a significant association with approach informed by these results in a large cohort of CCGA participants will be
PFS for DRD1 rs267410 and SLC6A3 rs2652510 in females (PFDR= 0.056), reported. Conclusions: Incorporating data from a large methylation database
and between SLC6A3 rs6347 and OS (PFDR= 0.041) and SLC6A3 rs6876890 improved TOO performance in multiple cancer types. This supports feasibility of
and TR (PFDR= 0.05) in KRAS wild type. Conclusions: Our results suggest that this methylation-based approach as an early cancer detection test across cancer
SNPs in DA signaling may have a prognostic value in mCRC pts receiving first- types. Clinical trial information: NCT02889978.
line treatment. Upon validation, these findings may provide novel insight on
the role of DA signaling in CRC and possibly contribute to open novel ther-
apeutic perspectives.

3050 Poster Session (Board #42), Sat, 8:00 AM-11:00 AM 3051 Poster Session (Board #43), Sat, 8:00 AM-11:00 AM
CTC versus biopsy tissue sequencing: A concordance analysis of genomic Identification and validation of a serum microRNA panel for detection of
copy number profile from mCRPC patients (pts). First Author: Howard I. early-stage breast cancer. First Author: Ruiyang Zou, MiRXES Pte Ltd,
Scher, Memorial Sloan Kettering Cancer Center, New York, NY Singapore, Singapore
Background: MSK-IMPACT (Integrated Mutation Profiling of Actionable Cancer Background: The implementation of mammogram-based screening has sig-
Targets), is a high throughput, targeted-DNA-sequencing panel for somatic nificantly improved the early detection of breast cancer in the west. However,
mutations created by the Department of Pathology at Memorial Sloan Kettering the use of screening mammography is less prevalent in Asia partly due to social
Cancer Center (MSK). The MSK-IMPACT is FDA approved for tumor tissue and cultural reasons. The aim of this study was to determine if a serum
profiling to guide treatment selection. Recognizing access to tumor tissue in microRNA (miRNA) panel could be used as biomarkers to assist in the early
many cancers is difficult and may harbor inter & intra lesional heterogeneity, we detection of breast cancer. Methods: We conducted a multi-center, multi-
sought to evaluate concordance of sequencing single CTCs vs. paired biopsy ethnic study to identify and validate miRNA biomarkers for the early detection
analyzed by MSK-IMPACT, to assess CTC vs. tumor clonality, and their re- of breast cancer. A total of 1070 subjects including 550 breast cancer cases
lationship to outcomes. Methods: 148 biopsy samples from 138 mCRPC pts (predominantly stage 1 and 2) and 520 matched controls from 6 independent
were submitted for IMPACT analysis and a blood draw within 30-day prior to sources were included in this study. Among these, there were 768 American
initiation of a new line of treatment. Blood samples were sent to Epic Sciences and European subjects recruited by biobanks and 302 Singaporean Asian
for CTC detection. The detected CTCs underwent single cell low pass whole Subjects recruited at the National Cancer Centre Singapore and the National
genome sequencing for copy number variation (CNV). For each set of matched University Hospital. The study was conducted in 3 phases. First, 119 European
samples, DNA copy number profiles from IMPACT and CTC sequencing were Caucasian serum samples (Discovery Cohort) were interrogated to identify
compared for similarity. Results: Of the 114 successfully sequenced samples differentially expressed miRNAs between early-stage breast cancer cases and
58 were from lymph nodes, 23 from bone, 25 from liver or lung, and 8 from matched controls, among 520 circulating miRNA candidates by quantitative
other soft tissue. Of the 111 patients with CTCs analyzed, a total of 1073 CTCs RT-PCR using MiRXES assays. The remaining 951 subjects from 5 in-
were sequenced (range 1-27, median = 4 per pt). Of patients with both IMPACT dependent sources were assigned into two groups for biomarker optimization/
& CTC, .80% pts had multiple subclones with distinguishable CNV. Con- algorithm development (Optimization Cohort, n = 451) and validation (Vali-
cordance of genomics clones was found in 63%, and discordant in 37% cases. dation Cohort, n = 500). Results: Among the 520 circulating miRNAs mea-
The clonal concordance between tissue biopsy and CTC was higher when the sured, 241 were quantified in absolute copy numbers for 119 subjects in the
biopsy was obtained from bone marrow or a visceral site (71% concordance) Discovery Cohort. Thirty-two candidate miRNAs were identified. These
than from a lymph node (53% concordance). Conclusions: Single CTC se- miRNAs consistently showed differential expression between cancer and
quencing is often concordant to metastatic tissue, but unique CTC clones and control subjects in the Optimization Cohort. A multi-variant panel of 8 miRNAs
the presence of multi-clonal disease highlight the potential to be underrepre- and an algorithm was developed using the combined Discovery and Optimi-
sented through tissue biopsy, especially when taken from the lymph node. zation Cohorts, with an AUC of . 0.90. When validated in the independent
Validation Cohort, the panel demonstrated an AUC of . 0.85. Conclusions: We
Cohort (n=148) Frequency of Observation developed and validated a serum miRNA panel that is both sensitive (~70%)
IMPACT result 77% (114/148) and specific (~90%) in detecting early-stage breast cancer.
CTC result 75% (111/148)
Both CTC & IMPACT 61% (90/148)
IMPACT only 16% (24/148)
CTC only 14% (21/148)
No results 9% (13/148)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Therapeutics and Tumor Biology (Nonimmuno) 161s

3052 Poster Session (Board #44), Sat, 8:00 AM-11:00 AM 3053 Poster Session (Board #45), Sat, 8:00 AM-11:00 AM
Cell-free methylated DNA (cfMeDNA) immunoprecipitation and high Training and validation study for sequential monitoring of CAMLs in circu-
throughput sequencing technology (cfMeDIP-seq) in patients with clear cell lation to predict ongoing progression in lung cancer patients undergoing
renal cell carcinoma (ccRCC). First Author: Pier Nuzzo, Dana-Farber Cancer definitive radiotherapy. First Author: Daniel Adams, Creatv MicroTech, Inc.,
Institute, Boston, MA Monmouth Junction, NJ
Background: CfmeDNA is a promising biomarker for non-invasive assessment Background: Cancer Associated Macrophage-Like cells (CAMLs) are a re-
of solid tumors: i) MeDNA is tissue- and tumor-specific ii) cfDNA methylation cently described circulating stromal cell common in the peripheral blood of
changes are stable unlike DNA alterations iii) ‘methylation target size’ is larger cancer patients that are prognostic for progressive disease. Further, it has
than identifying specific genomic alterations and, therefore, more sensitive. been shown that changes in CAML size (i.e. enlargement above 50mm) can
CfMeDIP-seq is a sensitive assay for genome-wide bisulfite-free cfMeDNA predict progression free survival (PFS) in thoracic cancers (e.g. lung). We
profiling, that requires 1-10 ng input DNA. We tested the feasibility of cfMeDIP- enrolled 104 unresectable non-small cell lung cancer (NSCLC) patients,
seq to detect ccRCC across TNM stages. Methods: We evaluated plasma cfDNA with an initial training set review of 54 patients, to determine if change in
collected prior to nephrectomy in 46 pts with ccRCC: 25 stage I, 7 stage II, 6 CAML size after radiation therapy was predictive PFS. Methods: A 2 year
stage III, 8 stage IV. cfMeDIP-seq involves four steps: 1) cfDNA end-repair, A- single blind prospective study was undertaken to test the relationship of
tailing, and adapter ligation 2) cfMeDNA immunoprecipitation and enrichment $50mm CAMLs to PFS based on imaging in lung patients before and after
using an Ab targeting 5-methylcytosine (quality control by qPCR to ensure ,1% induction of chemo radiation, or radiation therapy. To achieve a 2-tailed
of unMeDNA and .99% reaction specificity) 3) adapter-mediated PCR to 90% power (a = 0.05) we recruited a training set of 54 patients and vali-
amplify cfMeDNA 4) high-throughput NGS for cfMeDNA data. A previously- dation set of 50 patients all with pathologically confirmed unresectable
derived model (Shen et al, Nature, 2018) was used to classify pts as having NSCLC: Stage I (n = 14), Stage II (n = 16), Stage III (n = 61) & Stage IV (n =
ccRCC or not based on cfMeDNA. cfMeDIP-seq paired end data was reduced to 13). Baseline (BL) blood samples were taken prior to start of therapy & a 2nd
300 bp windows of the genome that map to CpG islands, shores, shelves, and blood sample (T1) was taken after completion of radiotherapy (~30 days).
FANTOM5 enhancers; a classifier was then built using the top 1,000 most Blood was filtered by CellSieve filtration and CAMLs quantified. Analysis by
variable fragments between pts with ccRCC and cancer-free controls. Statistical CAML size of , 49 mm or $50 mm was used to evaluate PFS hazard ratios
comparisons were performed in the R statistical environment, with the caret (HRs) by censored univariate & multivariate analysis. Results: CAMLs were
package being used for classifier construction and evaluation. Results: The found in 95% of samples averaging 2.7 CAMLs/7.5mL sample at BL, with
average amount of cfDNA isolated from 1 ml of ccRCC plasma was CAMLs $50 mm having reduced PFS (HR = 2.2, 95%CI1.3-3.8, p = 0.003).
19.8639.8 ng/mL [1.95-260]. Greater than 99% specificity of reaction At T1, 18 patients had increased CAML size $50 mm with PFS (HR = 4.6,
and ,1% of unMeDNA was achieved in 46/46 samples (100%). The 95%CI2.5-8.3, p , 0.001). In total, $50 mm CAMLs at BL was 76%
previously-derived classifier of ccRCC correctly predicted 46/46 pts (100%) as accurate at predicting progression within 24 months while $50 mm CAMLs
having ccRCC. Across 3 rounds of 5-fold cross-validation, the classifier per- at T1 was 83% accurate at predicting progression. Conclusions: In unre-
formed with a Cohen’s Kappa of 0.93. Conclusions: CfMeDIP-seq is a non- sectable NSCLC patients, enlargement of CAMLs during treatment is an
invasive, cost-effective, and sensitive assay to detect cancer-specific cfmeDNA indicator active progression. We identify that a single $50 mm CAML after
in ccRCC pts prior to nephrectomy. With further validation, cfmeDNA may induction of radiotherapy, in our training set and confirmed in our validation
detect minimal residual disease after nephrectomy for ‘precision’ adjuvant set, is an indicator of poor prognosis. We suggest that changes in CAML size
therapy. during therapy may indicate the efficacy of therapy and could potentially
help shape subsequent therapeutic decisions.

3054 Poster Session (Board #46), Sat, 8:00 AM-11:00 AM 3055 Poster Session (Board #47), Sat, 8:00 AM-11:00 AM
Associations between baseline serum biomarker levels and cachexia/ Efficacy of tyrosine kinase inhibitors (TKIs) based on the ALK resistance
precachexia in pretreated non-small cell lung cancer (NSCLC) patients. mutations on amplicon-based liquid biopsy in ALK positive non-small cell
First Author: Gabriela C. Lobato, Rush University Medical Center, Chicago, IL lung cancer (NSCLC) patients (pts). First Author: Laura Mezquita, Medical
Oncology Department, Gustave Roussy, Villejuif, France
Background: We previously reported associations of pretreatment serum
biomarkers with clinical outcomes in a cohort of advanced NSCLC patients Background: Acquired ALK resistance mutations (mut.) are the main mech-
that progressed on front-line therapy. This study aims to elucidate mech- anism of tyrosine kinase inhibitor (TKI) resistance (30-50%). While next-
anisms underlying cancer cachexia/ pre-cachexia by evaluating relationships generation TKIs are more active on mut. than earlier TKIs, compound ALK
between baseline serum biomarker values and sequential changes in body resistance are associated with failure to next-generation TKIs. We evaluated
weight, body mass index (BMI), and neutrophil/lymphocyte ratio (NLR) in the clinical utility of detecting ALK resistance mutations in blood to predict TKI
NSCLC patients. Methods: We used Luminex immunobead assays to survey efficacy. Methods: ALK positive advanced NSCLC pts were prospectively
101 protein biomarkers in sera from advanced NSCLC (n = 138) collected enrolled between Oct. 2015 and Aug. 2018 in 8 French institutions. Pro-
prior to their salvage regimen. Serial parameters associated with cancer ca- spective samples were collected; ctDNA was analyzed by amplicon-based
chexia included body weight, BMI, and NLR. Outcome variables (progression- Inivata InVisionFirst-Lung. Results: A total of 101 pts with advanced ALK
free survival (PFS) and overall survival (OS)) were extracted with full IRB- positive NSCLC were enrolled and 328 samples collected. In samples col-
approval. Biomarkers were evaluated as continuous variables with the cachexia lected at TKI failure (N=74), we detected 9 single and 7 complex ($2) ALK
surrogates using Pearson correlations, whereas associations of PFS and OS resistance mut. (22%), associated with EML4-ALK variant 3 (38%) vs. variant
were accomplished with the Cox PH test. Results: High baseline values of BMI 2 (13%) vs. variant 1 (none); 30% had other somatic mut. (mainly TP53 and
and low baseline NLR were associated with both OS and PFS (each p , 0.05), KRAS, PI3KCA, MET, etc.). No mutations were detected in 48% of samples
though weight failed to reach significance. PFS and OS were similarly asso- (ctDNAneg). ALK mut. were more frequent after 2nd/3rd generation TKI (43%
ciated with percent changes (relative to baseline) in weight (p , 0.01), BMI post-lorlatinib (7), 29% post-2nd gen. (31), 11% post-crizotinib (36)).
(p , 0.01), and NLR (p , 0.001). Thirteen biomarkers were found to be ALKG1202R was the most common, as single (n=3) or complex mut. (n=4).
associated (p , 0.05) with baseline BMI values, including positive correlations The median overall survival (mOS) was 100.4 mo. (95% CI 41.9-158.9) and
with leptin, sol.VEGFR2, and c-peptide and inverse correlations with adipo- the median progression free-survival (mPFS) to subsequent line was 2.8 mo.
nectin, ferritin, ghrelin, IGFBP-1 and IL-8; fifteen biomarkers were associated (0.7-4.9). Patients with ctDNAneg had mOS of 105 mo. (39.3-172.1) vs.
with baseline NLR (all p , 0.05), including positive correlations with visfatin, 58.5 mo. (33.1-84.0) if $1 ALK mut. vs. 44.1 mo. (20.0-68.2) if others
insulin, and serum amyloid A and inverse correlations with IGF-II. Fifteen (P=0.001). Pts with the complex ALK mut. had worse OS compared to singles
biomarkers were found to be associated (p , 0.05) in common with percent ALK mut. (mOS 26.9 mo. vs. 58.5 mo., P=0.001); ALK complex mut. were
weight and BMI changes, including positive correlations with IGFBP-3 and associated with poor efficacy to subsequent therapy (PFS ,3 mo. in 57%; no
inverse correlations with insulin, FGF-2, TNF-alpha, and resistin. Only pro- cases with PFS .6 mo.) vs. single mut., with longer PFS (PFS .6 mo. in 56%).
lactin and placental growth factor were found to be associated (p , 0.05) with Detectable ALKG1202R mut. were associated with shorter median OS (58.3
percent change in NLR. Conclusions: A series of circulating protein biomarkers mo.; 7.9-109.1) vs. overall population; 86% of cases developed rapid PD
primarily connected with metabolic regulation and systemic inflammation/ (PFS ,3mo.) to subsequent therapy with only one durable response to lor-
acute phase response were found to be associated with cachexia/ pre-cachexia latinib (PFS .6mo.). Conclusions: The absence of ctDNA mutations at TKI
in NSCLC patients. Additional cohorts are currently being tested to verify these failure was associated with prolonged OS, whereas complex ALK mutations at
findings. TKI failure may predict resistance to subsequent therapy. Larger and spe-
cifically designed studies should be performed to validate these findings.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
162s Developmental Therapeutics and Tumor Biology (Nonimmuno)

3056 Poster Session (Board #48), Sat, 8:00 AM-11:00 AM 3057 Poster Session (Board #49), Sat, 8:00 AM-11:00 AM
Baseline cfDNA characteristics and evolution of cfDNA profile during Analytical validation of a tumor-agnostic integrated multianalyte circulating
treatment with selective FGFR inhibitor TAS-120. First Author: Tyler J. tumor DNA (ctDNA) assay in early-stage cancer. First Author: Anna Hartwig,
Moss, The University of Texas MD Anderson Cancer Center, Houston, TX Guardant Health, Inc., Redwood City, CA
Background: There is an increasing role for cfDNA in monitoring response and Background: ctDNA sequencing has been rapidly adopted for the identification
mechanisms of resistance. We performed cfDNA analysis in a subset of pa- of targetable somatic alterations (alts) in patients with advanced cancers.
tients enrolled on a Phase I trial with an irreversible, selective FGFR1-4 in- However, early stage disease detection has been hindered by low levels of
hibitor, TAS-120. Methods: 58 plasma samples from 17 patients (13 with ctDNA in circulation and the presence of confounding non-tumor-related so-
cholangiocarcinoma) were analyzed on a 73-gene, next-generation sequencing matic alts. We developed and validated a ctDNA assay that combines somatic
panel. Selected patients(pts) had longitudinal samples. Results: At least one and epigenomic signals to detect early stage tumors without tumor tissue or
alteration was detected in 46 cfDNA samples, in 16 (94%) of 17 pts – a pt with white blood cells (WBC). Methods: Using a single input sample, our assay
GBM had no alterations detected. 14 pts had alterations in FGFR2/3 by integrates the sensitive detection of genomic alts with quantification of epi-
genomic testing of archival tumor samples, comprising 20 total alterations (18 genomic signals associated with cancer. Non-tumor alts (e.g., clonal hema-
unique). 10 of 20 FGFR2/3 alterations were also detected by cfDNA testing: 4/ topoiesis of indeterminate potential; CHIP) are excluded using a newly
5 SNVs, 1/2 amplifications, 5/13 fusions. Three pts had FGFR/FGF alterations developed bioinformatic classifier. To assess analytical sensitivity, specificity,
not included (thus not detected) in the cfDNA panel: 2 with FGF ligand and positive and negative reproducibility, we tested 337 clinical and con-
amplification, and one FGFR4 mutation. 6 pts (35%) had PR, 5 (29%) had SD trived samples. Results: Clinical specificity was determined using 80 plasma
and 6 (35%) PD as a best response to TAS-120. Four pts had prior FGFRi: 2 samples from 50-75 year old presumptive cancer-free donors, and resulted in a
had a PR, 1 SD, and 1 PD on TAS-120. Baseline cfDNA mutations became single false positive (99% specificity). Analytical sensitivity (limit of detection)
undetectable during treatment in 4/6 pts with PR. 4 of 6 PD pts had other was established using a dilution series of 4 different late stage CRC pts tested in
driver mutations at baseline including mutations in PIK3CA, KRAS, IDH1, triplicate at a clinically relevant DNA input (30 ng) across multiple batches.
BRCA2, or amplifications in PIK3CA, PDGFR. 9 pts with cfDNA available at 100% sensitivity was maintained even at the lowest tested level (estimated
progression after SD/PR: 3 had acquired FGFR2 mutations (one each of 0.1% tumor level). Positive/negative reproducibility was assessed by testing
V564L, V564F, or N549K). Two also acquired alterations in other candidate triplicates of diluted late stage samples and age-matched healthy donors, re-
resistance genes (PTEN and MAP2K1). Another pt had low variant allele spectively, across different cfDNA inputs, and multiple reagent lots. Both
frequency (VAF) NRAS G12D and BRAF A694T pretreatment and had SD. At positive and negative calls were 100% concordant across all replicates. In-
progression, cfDNA revealed an increase in NRAS VAF and mutations acquired dependent estimation of tumor levels from epigenomic or genomic signals
in the MAPK pathway. One pt with prior FGFRi acquired FGFR2 V564I and produced highly concordant results (correlation r-value: 0.82, p-value: 3e-16).
V564K detected by cfDNA prior to initiation of TAS-120, and had a PR on TAS- Conclusions: We designed and validated a highly specific ctDNA assay that
120. There was a drop in FGFR2 V564I VAF with response that subsequently integrates both genomic and epigenomic signals to allow for accurate and
increased with progression. The patient also acquired a FGFR2 V564L mu- quantitative tumor level detection in early stages of the disease without re-
tation at progression. Conclusions: FGFR alterations can be detected by quiring tumor tissue or WBC.
cfDNA. cfDNA may detect potential resistance mechanisms, including PI3K or
MAPK pathway alterations and acquired FGFR2 mutations. Patients with
gatekeeper mutations in cfDNA at baseline may still respond to TAS-120.
Further study is needed to determine the impact of FGFR2 mutations and co-
alterations on TAS-120 sensitivity.

3058 Poster Session (Board #50), Sat, 8:00 AM-11:00 AM 3059 Poster Session (Board #51), Sat, 8:00 AM-11:00 AM
Changes in DNA hydroxymethylation for the detection of multiple cancers in Pooled analysis of phase I dose-escalation and dose cohort expansion studies
plasma cell-free DNA. First Author: Anna Bergamaschi, Bluestar Genomics, of IMP4297, a novel PARP inhibitor, in Chinese and Australian patients with
San Diego, CA advanced solid tumors. First Author: Junning Cao, Fudan University Shanghai
Cancer Center, Shanghai, China
Background: Methylation and hydroxymethylation of cytosines enable the
epigenomic regulation of gene suppression and activation. 5-hydroxymethyl- Background: Poly (ADP-ribose) polymerase (PARP) enzymes play critical roles
cytosine (5hmC) is globally decreased in tumor tissue. However, genome-wide in DNA damage detection and repair. IMP4297 is a novel, potent PARP1/2
analysis using precise 5hmC labelling techniques reveals more nuanced inhibitor (IC50 6.27nM/1.57nM) and has demonstrated to be 20-fold more
changes upon tumorigenesis and raises the possibility that this loss could be potent than Olaparib in anticancer animal models. Two phase I studies were
exploited for developing a cancer biomarker. This suggests that 5hmC profiles performed to evaluate and characterize the tolerability and safety, pharma-
might enable discrete classification of not only tumor tissue but also of tumor cokinetics, and antitumor activity of single agent IMP4297 in Chinese and
cell-free DNA (cfDNA). We sought to identify genome-wide 5hmC changes in Australian patients with advanced ovarian, breast, prostate and other solid
plasma based cfDNA from cancer patients representing multiple disease types, tumors. Methods: Dose escalation used a 3+3 design with a modified Fibo-
stages and clinical characteristics in comparison with non-cancer patients. nacci escalation. Dose cohort expansion was planned after efficacy was ob-
Methods: cfDNA was isolated from plasma, enriched for the 5hmC fraction served at the lowest dose level. Patients received IMP4297 monotherapy orally
using chemical labelling, sequenced, and aligned to the genome to determine once a day until disease progression or unacceptable toxicity. Results: As of
5hmC counts per genomic feature. Regularized regression models were Jan 12, 2019, 56 patients, including 23 BRCA mutation carriers (BRCA+),
constructed to classify cancer samples (age matched or corrected for smoking had been enrolled at 2-100 mg dose level. No DLT was observed. In these two
status) on non-overlapping training (80% of all samples) and test sample sets studies, the most frequent treatment-related adverse events (TRAEs) were
(20% of all samples). Results: 226 non-cancer patients and 278 cancers leukopenia (20%), followed by anemia (18%), nausea (18%) and thrombo-
across four cancer types (breast, colorectal, lung-squamous and pancreas) cytopenia (14%). The majority of TRAEs were grade 1 or 2. Grade 3 TRAEs
were included in this study, where more than 60% of cancer samples were occurred in five patients (anemia, n=2; vomiting, n=1; thrombocytopenia,
early stage disease (I or II). Upon comparison with non-cancer samples, 5hmC n=1; elevated AST, n=1). Only one patient had a dose reduction due to grade 3
peaks have reduced enrichment in exons in breast, colorectal and lung cancer thrombocytopenia. No serious TRAEs were observed. In 15 BRCA+ patients
but not in pancreatic cancer. Further, 5hmC peaks in pancreas show different who had measurable lesions, the ORR was 33% and the DCR was 80%. There
patterns of enrichment in 3’UTR, translational termination sites, promoters were 4 BRCA+, platinum-sensitive ovarian cancer patients with an ORR of
and LTR. Overall 5hmC signal density was reduced in late stage cancers across 75% and a DCR of 100%. One patient with somatic BRCA mutated urothelial
all four diseases. The ability to classify non-cancer versus cancer patients was carcinoma showed a 76% decrease in tumor size. Conclusions: IMP4297 has
evaluated via cross-validation of out of fold prediction in the training set with been well-tolerated with significant anti-tumor activity. The 100 mg daily dose
AUC . 0.84 for all four cancer types. Further, test set sensitivity across all four was selected as the RP2D based on safety, pharmacokinetics and clinical
cancer types was found to be . 66% with 98% specificity. Conclusions: These activity, and will be further characterized in dose expansion and phase II studies.
findings suggest that 5hmC changes in plasma cfDNA enable classification of Tumor response to treatment (RECIST 1.1) in patients with measurable lesions.
early stages of breast, colorectal, lung-squamous and pancreas cancer and are Clinical trial information: NCT03508011 and NCT03507543.
promising biomarkers for disease detection. BRCA+ (n) PR (n) %
20mg 4 2 50%
60mg 4 1 25%
80mg 3 1 33%
100mg 2 1 50%
platinum-sensitive ovarian cancer patients 4 3 75%

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Therapeutics and Tumor Biology (Nonimmuno) 163s

3060 Poster Session (Board #52), Sat, 8:00 AM-11:00 AM 3061 Poster Session (Board #53), Sat, 8:00 AM-11:00 AM
Is the optimal biological dose of oncologic molecular-targeted therapies also A phase I multiple-dose escalation study to assess the safety, tolerability,
clinically effective? First Author: Pauline Corbaux, Université de Lyon, and pharmacokinetics of VEGF-receptor inhibitor telatinib (EOC315) in
Université Claude Bernard Lyon 1, Faculté de Médecine Lyon-Sud, Lyon, Chinese patients with advanced solid tumors. First Author: Hongming
France Pan, Sir Run Run Shaw Hospital, Zhejiang University School of Medicine,
Hangzhou, China
Background: The determination of the optimal biological dose (OBD) de-
fined as the lowest safe dose associated with biological efficacy, appears to Background: Telatinib (EOC315) is a highly selective inhibitor of VEGFR/PDGFR
be a promising endpoint for defining the recommended phase 2 trial dose (VEGFR 1-3, PDGFR-b, and c-Kit tyrosine kinases). This phase I study was to assess
(RP2D) of novel oncologic targeted therapies in early-phase clinical trials. the safety, tolerability, and pharmacokinetics (PK) of Telatinib in Chinese patients
However, the clinical relevance of OBD is still unknown. We conducted a with advanced solid tumors. Methods: Telatinib was administered to Chinese pa-
review to assess if the OBDs of molecular targeted therapies defined during tients with advanced refractory solid tumors as a single agent in 3+3 dose escalation
early phase trials were useful during subsequent drug development and for design, starting from 600mg and escalated to 900mg and 1200mg, given orally
twice daily. The PK profile, safety, and tolerability were evaluated per protocol.
oncologic drug approvals. Methods: A systematic review was conducted to
Efficacy was evaluated with RECIST 1.1 criteria every 6 weeks. Results: A total of 15
identify all the molecular targeted therapies approved by FDA in solid and
subjects（6 colorectal cancer, 4 lung cancer, 1 head and neck cancer, 1 melanoma,
hematological malignancies, and for which early phase trials defined OBDs. 1 thymic carcinoma, 1 esophageal carcinoma,1 peritoneal carcinoma）were en-
The publications of efficacy trials leading to the first FDA approvals were rolled per protocol between July 2017 and August 2018, and 13 subjects received at
reviewed, as were the FDA approved doses and dosing schedules, which were least second line therapies before enrollment. Telatinib was well tolerated in the
compared to OBDs found in the early phase trials. Results: OBDs were re- three dose arms. No dose limiting toxicities (DLTs) occurred during the dose es-
ported in the early phase trial articles of 39.5% (32/81) FDA approved calation phase. CTC grade 3 AEs observed include hypertension (46.7%, 7/15),
targeted therapies from 1999 to 2017 (19 small molecules and 13 fatigue (6.7%, 1/15), transaminase elevation (6.7%, 1/15), hand-foot syndrome
monoclonal antibodies (mAbs)). The maximum tolerated doses (MTD) had (6.7%, 1/15), oral mucositis (6.7%, 1/15), neutropenia (6.7%, 1/15), urobilinogen
not been reached for 59.4% (19/32) of these drugs. When both MTD and elevation (6.7%, 1/15), left ventricular systolic dysfunction/decreased ejection
OBD had been defined, OBD were lower than MTD in 84.6% of cases, and fraction (6.7%, 1/15). No CTC grade 4 AE were observed. There were 2 drug related
equal for the others. The OBDs were chosen as the RP2Ds for 56.3% of the SAEs（hospitalization due to high blood pressure. The PK profile of Telatinib
molecules. In that case, the final FDA approved doses were consistent with (EOC315) at 600, 900, 1200 mg in Chinese patient cohorts is summarized in Table.
the OBDs for 83.3% of the drugs. These percentages did not differ in be- For 12 evaluable patients, DCR was 58.3%. For all patients, mPFS was 15 weeks
tween small molecules and mAbs. OBDs mainly relied on indirect effects on (3.3-34.3w). Conclusions: This study demonstrated the safety and tolerability of
the involved signaling pathway elements for small molecules (11/19, Telatinib (EOC315) in a multiple dose escalation design at 600, 900, and 1200 mg
PO bid in Chinese patients with advanced refractory solid tumor. Telatinib AUC
57.9%), and on involved receptor occupancies for mAbs (6/13, 46.2%).
increased dose-proportionally from 600 mg to 900 mg bid, where 900 mg Telatinib
In total, 23.5% of all FDA approved molecular targeted therapies were
bid is the maximum feasible and recommended dose for future studies in Chinese
approved at their OBDs. Conclusions: Although still poorly investigated, OBD patients with advanced tumors. Clinical trial information: NCT03175497.
may represent a relevant complementary endpoint in in early phase trials of
Telatinib multiple does pharmacokinetic parameters （D14）.
novel anti-cancer targeted therapies, as OBDs are selected as the final FDA
Dose AUC0-12h Vz/F CLss/F
approved doses in 83.3 % of cases when chosen as the RP2Ds, which is (mg) t1/2 (h) (h*ng/mL) (L) (L/h) RAUC0-12h
much higher than the previously reported 58.0 % when MTDs are chosen as 600（N=3） 6.2062.36 5309.3663661.96 1259.216828.99 146.25672.37 1.6361.13
the RP2Ds (Fontes-Jardim et al. JNCI 2015). 900（N=3+6） 3.9061.81 9708.0262687.77 543.426145.27 99.00627.58 1.1760.74
1200（N=3） 7.98 7243.4462776.60 2723.15 181.06661.05 0.7860.39

3062 Poster Session (Board #54), Sat, 8:00 AM-11:00 AM 3063 Poster Session (Board #55), Sat, 8:00 AM-11:00 AM
Modular phase I/II clinical trial evaluating the selective MET-kinase inhibitor First-in-human phase I and pharmacological study of TAS-119, a selective
OMO-1 in patients with advanced malignancies: Safety and proof of Aurora A (AurA) kinase inhibitor, in patients (pts) with advanced solid tumors.
mechanism. First Author: Martijn P. Lolkema, University Medical Center First Author: Debbie Robbrecht, Erasmus MC, Rotterdam, Netherlands
Utrecht, Utrecht, Netherlands
Background: AurA is a serine threonine kinase regulating cell division and cell
Background: MET kinase is a therapeutic target in a range of cancer indications; cycle progression and has a role in carcinogenesis. This clinical trial in-
it is a primary oncogenic driver and a mechanism of therapy resistance. OMO-1 vestigated safety, pharmacokinetics and -dynamics and antitumor activity of
is a highly potent, selective oral inhibitor of MET kinase and Organic Cation the selective oral AurA kinase inhibitor TAS-119. Methods: Pts with advanced
Transporter 2 (OCT2). Methods: This study assesses the safety, tolerability, solid tumors were enrolled into 6 dose escalation cohorts (70-300 mg BID
pharmacokinetics (PK) and preliminary activity of OMO-1 in patients (pts) 4 days on/3 days off; every 3 out of 4 weeks; or the same schedule in a
with advanced malignancies (NCT03138083). Module 1 data, evaluating continuous weekly schedule). In the expansion phase (intermittent schedule),
ascending doses of OMO-1 monotherapy, are reported here. Results: As of pts with small-cell lung cancer (SCLC), breast cancer, or MYC-amplified/B-
January 16, 2019, 34 pts were enrolled at 5 twice-daily (BD) dose levels of OMO- catenin mutated (MT) tumors were enrolled, and pts with other solid tumors
1: 100, 200, 250, 350, and 400 mg, including 10 with MET gene amplified or in a basket cohort. Results: Overall, 34 pts were enrolled to the escalation
mutated tumours. OMO-1 was generally well tolerated between 100 - 250 mg (median age 67 years; 45.3% . 2 prior therapies); DLT was observed in 5
BD; pts were in the study for an average of 94 days (range: 15-291 days) and 20/ (16.1%) of 31 DLT evaluable pts; 1/10 at 150 mg, 1/6 at 200 mg, 1/5 at
34 pts discontinued due to disease progression. Most frequently-reported AEs 250 mg, and 2/2 at 300 mg BID (fatigue, nausea, dry eyes, corneal epithelial
were nausea (17/34), vomiting (14/34) and fatigue (14/34), mainly G1-2. microcysts). The maximum tolerated dose (MTD) was 250 mg BID and rec-
Notably, no peripheral oedema, cardiovascular events or non-malignancy re- ommended Phase 2 dose (RP2D) was 200 mg BID. The most frequent
lated LFT abnormalities were observed. A total of 36 SAEs were reported: 17 in treatment-emergent adverse events were diarrhea (28.3%), eye disorders
11 subjects were considered related to OMO-1, and included nausea (3/17), (27%), fatigue (22.9%), and decreased appetite (14.8%). Grade 3 ocular
vomiting (4/17), chills, diarrhoea, influenza-like illness (2/17), increased blood toxicity were corneal epithelial microcysts in 1 pt (300 mg cohort) and punctate
bilirubin, blood creatinine (3/17) and neutrophil count, and sepsis. A dose of keratitis (expansion breast cancer cohort) in 1 pt. Toxicity grade $ 3 in $ 10% of
250 mg BD was determined as the recommended Phase 2 dose (RP2D); pts were diarrhea (escalation part only), and increased lipase. Plasma exposure
doses $350mg BD were not in keeping with optimum long-term dosing: at was dose-proportional and accumulation ratio was low. Pharmacodynamic data
400 mg BD, 2/3 subjects experienced influenza-like illness (G2 and G3) and at demonstrated target inhibition. Overall, 40 pts were enrolled to multiple ex-
350 mg BD 2/5 subjects had G2 fatigue and nausea/vomiting. OMO-1 has a half- pansions (10 SCLC, 9 breast cancer, 13 MYC-amp/B-cat MT tumors, 8 other;
life of 2.5-3 hrs and plasma exposure is dose-proportional without accumulation. median age 60 years; 72.5% . 2 prior therapies). Median delivered relative
Elevated creatinine was observed across all dose levels, consistent with OCT2 dose intensity was 89.1% (47.9% - 100%). Stable disease was reported in
inhibition. IHC analysis on paired tumour biopsies from a MET-mutated NSCLC 37.8% of patients but no complete or partial responses. Conclusions: TAS-119
pt dosed at 200 mg BD showed near-complete inhibition of phosphorylated MET, demonstrated favorable safety and tolerability. Low-grade eye toxicity was a
without affecting total MET. Conclusions: OMO-1 has a favourable safety profile dose-dependent toxicity. Preliminary anti-tumor activity of monotherapy TAS-
at a RP2D of 250mg BD. Expansion cohorts for MET mutated/amplified tumour 119 is limited. A Phase 1 trial combining TAS-119 with paclitaxel was con-
types are enrolling. Clinical trial information: NCT03138083. ducted in parallel. Clinical trial information: NCT02448589.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
164s Developmental Therapeutics and Tumor Biology (Nonimmuno)

3064 Poster Session (Board #56), Sat, 8:00 AM-11:00 AM 3065 Poster Session (Board #57), Sat, 8:00 AM-11:00 AM
Rethinking about the dose limiting toxicities (DLTs): They can be equivocal! Effectiveness of ASCO’s adverse event reporting decision aid: Results from
First Author: Wei Zhong, Pfizer Inc, Cambridge, MA an interventional study. First Author: Kathryn Finch Mileham, Levine Cancer
Institute/Atrium Health, Charlotte, NC
Background: The primary goal of oncology phase I trials is to determine the
maximum tolerated dose (MTD) in a small number of patients. Within the Background: Investigators often send adverse event (AE) reports to sponsors
standard design methodology currently employed there remains many chal- that are incorrectly categorized as serious or attributed to the investigational
lenges and one pertains to the variability around defining dose limiting tox- drug. Such errors contribute to a high volume of uninformative IND safety
icities (DLTs). The pre-specified DLT criteria may not well reflect the clinician’s reports that sponsors submit to FDA and all participating investigators,
practical experience with some adverse events, while others may actually be straining stakeholder resources and impeding the detection of valid safety
related to the disease status more than the investigational drug and could be signals. To improve the quality of AE reporting, ASCO developed and tested a
misclassified. DLT misclassification seen in testing a small sample size from Decision Aid Tool (DAT). Methods: An interventional study with a cross-over
each dose group could generate a large bias on dose finding and the MTD design was conducted. Physician investigators and research staff were ran-
estimation. Methods: To mitigate the risk of dichotomizing and misclassifying domized to receive case studies. Cases were assessed for seriousness and
DLTs, we proposed a strategy that introduced the new concept of “equivocal” attribution, first unassisted and then with the DAT. Participants completed a
DLT or AE. A novel dose escalation approach is applied to increase the var- feedback survey. Effectiveness of reporting and attribution were assessed using
iability associated with less interpretable AEs so that the model recommen- logistic regression. Results are reported as odds ratios (OR) with 95% confi-
dations are more weighted towards the unequivocal AEs/DLTs. To evaluate this dence intervals (CI). Results: Most of the 29 participants reported that the DAT
novel approach, we established a framework incorporating two types of sys- was helpful (93%), improved their decision-making time (69%) and confidence
tematic measurement errors on DLT misclassification, one for the mis- in reporting (83%), and that they would use it in practice (83%). The DAT did
classified DLT that is not related to the drug treatment while the other for the not significantly affect accuracy of determining seriousness (OR, 0.87; 95% CI:
non-DLT AE that should be considered severe and relevant to dose finding. In 0.31, 2.46) but it did significantly increase accuracy of attributing a serious
our simulation studies, the Bayesian logistic regression model (BLRM) was AE to a drug (OR, 3.60; 95% CI: 1.15, 11.4). Conclusions: The DAT shows
used to guide dose escalation in simulated trials to compare the novel promise as a method to reduce errors in attribution of AEs, which may help to
weighting approach with the traditional approach. A few numerical examples ensure the detection of valid safety signals. Many participants were experienced
were also included for method illustration. Results: For different types of clinical trialists, and the DAT may show greater utility as an educational tool for
measurement errors, simulation studies showed that the weighting approach novice investigators, research staff, and students.
could successfully improve the trial performance, with higher chance of finding
the correct MTD and treating more patients at the MTD level. Conclusions: The
DLT weighting strategy provides a flexible but powerful tool that may in-
corporate the clinician’s valuable experience on some specific DLTs/AEs and
improve MTD estimation in oncology phase I dose-escalation trials.

3066 Poster Session (Board #58), Sat, 8:00 AM-11:00 AM 3067 Poster Session (Board #59), Sat, 8:00 AM-11:00 AM
The i3+3 design for phase I clinical trials. First Author: Yuan Ji, The Uni- Safety and tolerability of veliparib, an oral PARP inhibitor, and M6620 (VX-
versity of Chicago, Chicago, IL 970), an ATR inhibitor, in combination with cisplatin in patients with
refractory solid tumors. First Author: Arjun Mittra, Division of Cancer
Background: Other than the 3+3 design, new model-based statistical designs
Treatment and Diagnosis, National Cancer Institute, Bethesda, MD
like the mTPI design (Ji and Wang, 2013, JCO) are alternative choices for
oncology dose-finding trials, including immune oncology dose-finding trials Background: M6620 (M), a potent ATR inhibitor, has synergistic activity
(Atkins et al., 2018, Lancet Oncology). One major criticism of the 3+3 design is with cisplatin (C) in multiple preclinical models, resulting in DNA damage
that it is based on simple rules, does not depend on statistical models for in- and antitumor activity. We hypothesize that inhibition of both homologous
ference, and leads to unsafe and unreliable operating characteristics. However, recombination and base excision repair through the combination of M6620
the rule-based nature allows 3+3 to be easily understood and implemented in and veliparib (V, a potent inhibitor of PARP1/2) would result in accumulation
practice, making it practically attractive and friendly. Can friendly rule-based of lethal double stranded breaks induced by cisplatin and increased anti-
designs achieve great performance seen in model-based designs? For four tumor activity and initiated a phase-1 dose escalation trial of this combi-
decades, the answer has been NO. Methods: We propose a new rule-based nation in patients (pts) with advanced solid tumors (NCT02723864).
design called i3+3, where the letter "i" represents the word "interval". The i3+3 Methods: This is a standard 3+3 dose escalation design with 21-day cy-
design is based on simple but more clever rules that account for the variabilities cles. M is given IV day 2 (D2) and D9; V orally twice daily D1-3 and D8-10; C
in the observed data. In short, the i3+3 design simply asks clinicians to compare IV D1 (and D8 from dose level 3 [DL3] onwards) at 40 mg/m2 (with option of
observed toxicity rates with a prespecified toxicity interval, and make dose holding after cycle 6). Primary objectives: safety; tolerability; maximum
escalation decisions according to three simple rules. No sophisticated modeling tolerated dose (MTD). Secondary objectives: pharmacodynamic (PD) bio-
is needed and the entire design is transparent to clinicians. Results: We compare markers; antitumor activity. Dose-limiting toxicity (DLT) evaluated during
the operating characteristics for the proposed i3+3 design with other popular cycle 1, response using RECIST 1.1. Results: Thirty-seven patients enrolled,
phase I designs by simulation. The i3+3 design is far superior than the 3+3 median 5 lines of prior therapy (Range 1-12). MTD: V 200 mg, M 210 mg/m2,
design in trial safety and the ability to identify the true MTD. Compared with C 40 mg/m2 (DL6). DLT: grade (gr) 4 thrombocytopenia (DL4), hypo-
model-based phase I designs, i3+3 also demonstrates comparable perfor- phosphatemia (not resolved in 24 hrs., DL3), infusion reaction (DL7). Common
mances. In other words, the i3+3 design possesses both simplicity and gr 3/4 toxicities: anemia (41%), thrombocytopenia (30%), neutropenia (27%),
transparency of the rule-based approaches, and the superior operating char- hyponatremia (11%) and hypophosphatemia (8%). Of 34 pts evaluable for
acteristics seen in model-based approaches. An online R Shiny tool is provided to response: 2 partial response (PR) (6%, 1 confirmed & 1 transient), 22 stable
illustrate the i3+3 design, although in practice it requires no software to design or disease (SD) (65%, median 4 cycles, range 2-11). Of 24 pts with PR/SD,
conduct a dose-finding trial using the design. Conclusions: The i3+3 design 22 had prior platinum chemotherapy. After July 2018, V was held for gr $2
could be a practice-altering method for the clinical community. It may increase anemia until improved to gr 1, followed by dose reduction. Of 5 pts treated in this
the safety and efficiency of dose finding trials. period, 3 required V held. PD analysis of circulating tumor cells is underway to
elucidate biomarkers of DNA damage and apoptosis. Conclusions: This com-
bination of M6620, veliparib and cisplatin is safe, with activity seen in pts
having received prior platinum. The most common toxicity was anemia, which
prevented adequate delivery of veliparib. While MTD was established in a heavily
pretreated population, consideration should be given to continued dose esca-
lation in pts who have received fewer prior lines of therapy. Funded in part
by NCI Contract No. HHSN261200800001E. Clinical trial information:
NCT02723864.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Therapeutics and Tumor Biology (Nonimmuno) 165s

3068 Poster Session (Board #60), Sat, 8:00 AM-11:00 AM 3069 Poster Session (Board #61), Sat, 8:00 AM-11:00 AM
Image-guided surgery for tumor agnostic detection of solid tumors using the Radiomics features to identify distinct subtypes of triple-negative breast
pH-activated micellar imaging agent ONM-100. First Author: Floris Jan cancers. First Author: Haruka Itakura, Stanford Univ Medcl Ctr, Stanford, CA
Voskuil, UMCG, Groningen, Netherlands
Background: We sought to gain new insight into triple-negative breast cancer
Background: ONM-100, a micelle-based polymer imaging agent conjugated (TNBC), an aggressive, clinically distinct subgroup of breast cancers, by
to indocyanine green (ICG) and with an exquisitely pH-sensitive binary applying a sequence of computational approaches to tumor segmentation,
activation mechanism, may be used for tumor detection. ONM-100 micelles three-dimensional anatomic characterization, and tumor subtyping. We
dissociate in acidic environments resulting in activation of the fluorescent extracted algorithmically-derived quantitative imaging (radiomics) features
ICG tag. As nearly all solid cancer types are acidotic, ONM-100 has the from each TNBC lesion in breast magnetic resonance imaging (MRI) to identify
potential to act as a broadly indicated tumor agnostic imaging agent. This underlying subtypes. Methods: We evaluated tumors on pre-treatment, post-
first-in-human study investigates the safety and feasibility of ONM-100 as a contrast MRI from 90 patients with non-metastatic TNBC. We employed active
tumor agnostic imaging agent for intra-operative fluorescent imaging of contour segmentation and semi-automated identification of tumor regions-of-
various solid tumors. Methods: ONM-100 was IV administered 2468h prior interest. We extracted 900 radiomics features from each segmented tumor
to surgery in a dose escalation scheme (0.1-1.2mg/kg). Patients with his- using an algorithm that characterizes the size, shape, texture, and edge
topathologically confirmed breast cancer (BC), head and neck squamous cell sharpness of tumors at the voxel level. We applied k-means consensus
carcinoma (HNSCC), colorectal cancer (CRC) and esophageal cancer (EC) clustering, a statistical tool for unsupervised discovery, and performed 1000
were included. Blood was drawn to assess safety and pharmacokinetic data. bootstraps with resampling on the feature vectors to examine all resulting
Intra-operative fluorescence images were collected before and after tumor clusters from k=2 to 10. Based on two diagnostic metrics of consensus sta-
excision. Post-excision fluorescence images were obtained from serially bility, we selected the optimum cluster number. We performed Significance
sliced specimens and correlated with standard histopathological assess- Analysis of Microarrays to identify statistically significant radiomics features for
ment. Results: 30 patients (11 BC, 13 HNSCC, 3 EC, 3 CRC) were enrolled. each cluster. Results: We identified three distinct image-based clusters in 117
No ONM-100 related serious adverse events were observed and the agent tumors from 90 TNBC patients (multifocal lesions in n=13). Cluster 1 (n=97)
was well-tolerated. A strong and sharply demarcated fluorescent signal was was distinguished by 330 radiomics features (False Discovery Rate
observed in all patients with vital tumor tissue (median CNR ranging 1.85- [FDR] ,5%) and Cluster 2 (n=13) by 85 features (FDR,5%), whereas Cluster
14.05) which correlated with tumor on final histopathology. HNSCC and 3 (n=7) was not significantly associated with features. Clinical characteristics
superficially located BC as well as peritoneal metastasis could be clearly did not differ across the three clusters, with mean age (49.1611.7) and
visualized in vivo during surgery. In four patients (BC and HNSCC), peri- clinical stage distributions (stage I: 20.7%, II: 55.4%, III: 23.9%) for the
operatively, tumors otherwise unnoticed by the surgeons were detected on cohort mirroring those of individual clusters. Among those who received
the margin or wound bed using fluorescence imaging. Additionally, two BC neoadjuvant therapy, we observed pathologic complete response in 50% (23
tumor lesions were detected that were missed by conventional pre-operative of 46, 95% CI, 0.36-0.64) of patients in Cluster 1, 83% (5 of 6, 95% CI, 0.54-
imaging and pathological assessment. Conclusions: ONM-100 appears to be 1.0) in Cluster 2, and 0% (0 of 3) in Cluster 3. Conclusions: Radiomics
safe and enables fluorescent visualization of tumors both in vivo and ex vivo. features providing voxel-level characteristics of tumor morphology differenti-
The first-in-human data demonstrate the feasibility for potential use of ONM- ated TNBC into three distinct subtypes. These subtypes, defined by radiomics
100 for image guided surgery, margin assessment and detection of occult biomarkers, may be associated with clinical response to neoadjuvant therapy.
disease. Clinical trial information: NTR 7085.

3070 Poster Session (Board #62), Sat, 8:00 AM-11:00 AM 3071 Poster Session (Board #63), Sat, 8:00 AM-11:00 AM
[18F] Fluciclatide PET as a biomarker of clinical response to combination A functional measurement of MAPK pathway activation to predict response
therapy of pazopanib and paclitaxel in patients with platinum-resistant or to MEK inhibitors in RAS-mutated patients. First Author: Shumei Kato,
platinum-refractory advanced ovarian cancer: Results of a phase Ib study. University of California San Diego, La Jolla, CA
First Author: Rohini Sharma, Imperial College London, London, United
Background: MEK inhibitors can be used to treat patients with mutations
Kingdom
that affect the MAPK pathway. Several MEK inhibitors are currently FDA-
Background: Angiogenesis has been shown to be a driver of platinum re- approved and effectively treat BRAF-mutated tumors, but RAS-mutated
sistance in ovarian cancer. We assessed the effect of combination pazopanib cancers are considered more resistant. However, it is unclear how the many
and paclitaxel followed by maintenance pazopanib in patients with platinum distinct RAS variants impact the MAPK pathway and are affected by MEK
resistant/refractory ovarian cancer. Integrins avb3 and avb5 are both up- inhibitors. We hypothesized that the level of MAPK pathway activation in-
regulated in tumour-associated vasculature. [18F]Fluciclatide is a novel PET duced by different RAS mutations may predict response to MEK inhibition.
tracer that has high affinity for integrins avb3/5, and was used to assess the Methods: Thirteen RAS mutations from 34 patients treated with MEK in-
anti-angiogenic effect of pazopanib. Methods: We conducted an open-label, hibitors at UCSD were synthesized, expressed in a HeLa-derived cell line and
phase Ib study in patients with platinum resistant/refractory ovarian cancer. analyzed in vitro using a functional mutational analysis assay based on
Patients received 1 week of single agent pazopanib (800mg daily) followed assessing downstream reporters in order to measure the activity of these
by combination therapy with weekly paclitaxel 80mg/m2. Following comple- mutations on the MAPK pathway. Each mutation received an activity score
tion of 18 weeks of therapy, patients continued with single agent pazopanib based on known oncogenic RAS mutation. Results: The most common type
until disease progression. Dynamic [18F]Fluciclatide-PET imaging was con- of cancer was colorectal cancer (N = 13). All patients received the MEK
ducted at baseline and after 1 week of pazopanib. Response (RECIST 1.1), inhibitor, trametinib, based therapy. Patients were stratified into two groups:
toxicities and survival outcomes were recorded. Circulating markers of angio- above an activity score of 1 (14 pts) or below it (20 pts). Median progression-
genesis were assessed with therapy. Results: Fourteen patients were included in free survival (PFS) after MEK inhibitor treatment correlated with higher
the intention-to-treat analysis. Complete and partial response was seen in 7 MAPK activity score (9 vs 3 months; P = 0.041). Conclusions: Using a novel
patients (54%). Median progression free survival (PFS) was 7.97 months, and functional assay methodology for characterization of MAPK activation, we
overall survival (OS) was 18.5 months. A reduction in [18F]fluciclatide uptake show that various RAS mutations activate the MAPK pathway to different
was observed following 1 week of pazopanib, and the reduction in uptake was levels. Higher activity is associated with longer PFS after MEK inhibitor
predictive of long PFS. Elevated baseline circulating angiopoietin and FGF were treatment, suggesting that the relationship between signal transduction
predictive of greater reduction in SUV60,mean following pazopanib. Kinetic strength and clinical relevance merits additional exploration.
modelling indicated a reduction in K1 and Ki following pazopanib indicating
reduced radiotracer delivery and retention. Conclusions: Combination therapy
followed by maintenance pazopanib is effective and tolerable in patients with
platinum resistant/refractory ovarian cancer. We have shown that [18F]
fluciclatide-PET uptake parameters alter with pazopanib therapy indicating
an anti-angiogenic response. Clinical trial information: NCT01608009.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
166s Developmental Therapeutics and Tumor Biology (Nonimmuno)

3072 Poster Session (Board #64), Sat, 8:00 AM-11:00 AM 3073 Poster Session (Board #65), Sat, 8:00 AM-11:00 AM
Clinical impact of tissue of origin testing and mutation profiling in the No-cost next generation sequencing of advanced cancer patients within the
Solving Unknown Primary Cancer (SUPER) national prospective study: Strata Precision Oncology Network supports clinical trial enrollment. First
Experience of the first two years. First Author: Linda R. Mileshkin, Peter Author: Marc Ryan Matrana, Ochsner Clinic Foundation, New Orleans, LA
MacCallum Cancer Centre, Melbourne, Australia
Background: Widespread integration of systematized next generation se-
Background: Cancer of unknown primary (CUP) has a poor prognosis with a quencing (NGS)-based precision oncology is hindered by numerous barriers.
median survival of less than 12 months. SUPER is a prospective cohort study Hence, we developed the Strata trial (NCT03061305), a screening protocol
designed to create a national biobank of patients (pts) with no confirmed to determine the impact of scaled precision oncology. Methods: We imple-
primary site following diagnostic work-up. Tumor and blood samples undergo mented no-cost NGS on formalin fixed paraffin embedded (FFPE) clinical
mutational profiling for actionable mutations using the 386 gene PeterMac samples for all patients with advanced tumors, a common portfolio of partnered
Comprehensive Cancer Panel (CCP) plus CUPGuide, a microarray gene- therapeutic clinical trials, and robust infrastructure development across the
expression site-of-origin assay. We aimed to determine the clinical impact Strata Precision Oncology Network. Results: Across the network of 17 centers,
of CUPGuide and CCP profiling. Methods: 172 pts were enrolled between specimens from 8673/9222 (94%) patients were successfully tested in the
2013-2015. Baseline demographics, treatments, investigations and clinico- Strata CLIA/CAP/NCI-MATCH accredited laboratory using comprehensive
pathological characteristics were collected over 12 months. Clinicians com- amplicon-based DNA and RNA NGS. Patients were tested with one of three
pleted clinical management questionnaires before and after receiving results. StrataNGS test versions; the most recent panel assesses all classes of actionable
Results: Molecular analysis was performed for 124/172 (72.1%) pts with alterations (mutations, copy number alterations, gene fusions, microsatellite
sufficient DNA and/or RNA. CUPGuide was completed for 97/124 (78.2%); instability, tumor mutation burden and PD-L1 expression). Median surface area
primary site predictions were made in 84/97 patients (86.6%). The most of received FFPE tumor samples was 25mm2 (interquartile range 9-95mm2),
common primary site predictions were lung, gastric, ovary and breast. CUP- and the median turnaround time from sample receipt to report was 6 business
Guide predictions resulted in a change in management in 10/84 (12%) of days. 2577 (27.9%) patients had highly actionable alterations, defined as
cases and confirmed current management already commenced by the clinician alterations associated with within-cancer type FDA approved or NCCN guideline
in 53/84 (63%). Mutation profiling was completed in 103/124 (83.1%) pts recommended therapies (1072 patients), NCI-MATCH trial arms (1467 pa-
with actionable mutations found in 11 pts, 4 of whom received subsequent tients), Strata-partnered therapeutic trials (327 patients), or specific alteration-
targeted therapy. Testing was considered to have a clinical impact in 70/120 matched FDA approved therapies in patients with cancers of unknown pri-
cases (58%): either resulting in a change in treatment (n = 14), diagnosis of a mary (71 patients). Of the 1467 patients matched to an NCI-MATCH trial arm,
pathogenic germline finding (n = 8) or a moderate/high confidence tissue of 15 enrolled. Of the 327 patients matched to one of nine Strata-partnered
origin prediction (n = 58). There were two deaths prior to the availability of the clinical trials, 77 (24%) were screen failures, while 250 (76%) have either
CUPGuide results and eleven deaths prior to availability of the CCP results. enrolled or are being actively followed for enrollment upon progression.
Conclusions: Molecular analysis for CUP pts has clinical impact in the majority Conclusions: Through streamlined consent methods, electronic medical record
of cases. Timeliness of return of results, drug access and insufficient tissue for queries, and high throughput laboratory testing at no cost to patients, we
testing are barriers to greater impact that need to be addressed to improve the demonstrate that scaled precision oncology is feasible across a diverse network
care of pts affected by CUP. of healthcare systems when paired with access to relevant clinical trials. Clinical
trial information: NCT03061305.

3074 Poster Session (Board #66), Sat, 8:00 AM-11:00 AM 3075 Poster Session (Board #67), Sat, 8:00 AM-11:00 AM
Genomic analysis of metastatic solid tumors in veterans: Findings from the Comparison of an automated cartridge-based system for mRNA assessment
VHA National Precision Oncology Program. First Author: Pradeep Poonnen, with central immunohistochemistry in the neoadjuvant GeparX trial. First
Duke University Health System/Durham VA Medical Center, Durham, NC Author: Carsten Denkert, Institute of Pathology, Charité-Universitätsmedizin,
Berlin, Germany
Background: Scalable next generation sequencing (NGS) technologies have
enabled incorporation of precision oncology into clinical practice, informing Background: Hormone receptors, HER2 and Ki-67 are prognostic values
treatment decisions based on tumor genomics. The Veterans Health Ad- typically determined for breast cancer (BC) outcome and prediction of
ministration (VHA) is the largest integrated healthcare system in the U.S., therapy response. A RT-qPCR based system, the Xpert Breast Cancer
serving a higher percentage of rural patients (36%) than the national average STRAT4, can be used to classify BC tissues regarding their hormone receptor
(14%). To implement and standardize the practice of precision oncology status, HER2 and proliferation via Ki-67. We compared mRNA expression
across a diverse healthcare system, the VHA established the National Precision analysis of ER, PR, HER2, and Ki-67 by this automated in-vitro diagnostic
Oncology Program (NPOP). Methods: Tumor or peripheral blood specimens platform (GeneXpert) (GX) with central immunohistochemistry (IHC) in a
were collected from Veterans with advanced solid tumors who were eligible for large clinical trial cohort. Methods: BC patients from the prospective GBG
treatment with targeted or immunotherapeutic drugs. Specimens were se- neoadjuvant trial GeparX (NCT02682693) (still recruiting) were included in
quenced using cancer gene panels at two commercial laboratories. Annotated this biomarker project. We used formalin-fixed paraffin embedded (FFPE)
results were generated by the vendors and independently using IBM Watson for pretherapeutical core biopsies with a tumor content . 10%. One 4 mm FFPE
Genomics. Levels of evidence treatment recommendations were based upon tissue section was first processed with the Xpert FFPE Lysis Kit, the sample
OncoKB criteria. Results: Between July 2016 and June 2018, 3713 samples lysate was placed in the STRAT4 cartridge system and then tested on the GX
were collected from 72 facilities; the sequencing success rate was 86%. The system in which the purification, amplification and real-time detection took
majority of samples came from males with lung, prostate and colorectal place within two hours automatically. Results: A total of 503 (99%) of the
cancers. Thirty-four percent of samples submitted were from rural patients. 509 samples had a valid measurement of all four genes. 258 samples
The most commonly mutated genes included TP53, ATM and KRAS. Over (51.3%) of the cohort were classified in central pathology as ER positive,
70% of samples sequenced had at least one actionable mutation, and clinical 196 (39%) as PR positive and 78 (15.5%) as HER2-positive, and 421
trials were the recommended option in over 50%. The most frequent therapies samples (83.7%) were Ki-67-high ( . 20%). The simple kappa coefficient
prescribed in response to NGS testing were immune checkpoint inhibitors, was for ER = 0.7938, PR = 0.6540, HER2 = 0.8172 and Ki-67 = 0.3655.
EGFR kinase inhibitors and PARP inhibitors. Interestingly, prostate can- This indicates, that the measurements for ER, PR and HER2 showed a high
cers among Veterans had a higher frequency of mutations in genes associated correlation between both methods, whereas the measurement of Ki-67 does
with a neuroendocrine phenotype compared with the general population. not. The accuracy between the STRAT4 and IHC was 89.7% for ER, 83.3%
Conclusions: Implementation of precision oncology into clinical practice is for PR, 94.6% for HER2 and 86.7% for Ki-67. According to molecular
feasible across the diverse VHA system, including rural community sites. subgroups, highest accuracy regarding Ki-67, was determined in TNBC
Veterans have unique occupational exposures that might inform underlying (96.2%; luminal: 81.1%; HER2-positive: 76.9%). Conclusions: Our results
causes of distinct mutational signatures identified here. Our results highlight show a high concordance between standardized central IHC and automated
the importance of increasing the availability of clinical trials for Veterans. mRNA expression analysis for the most important BC biomarkers ER, PR and
HER2. For the proliferation marker Ki-67, the concordance is slightly lower.
The STRAT4 assay might offer additional option to conventional methods for
BC biomarker assessment, in particular in settings where IHC is not feasible.
To determine the clinical validity, additional outcome analyses are neces-
sary. Clinical trial information: NCT02682693.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Therapeutics and Tumor Biology (Nonimmuno) 167s

3076 Poster Session (Board #68), Sat, 8:00 AM-11:00 AM 3077 Poster Session (Board #69), Sat, 8:00 AM-11:00 AM
Utility of somatic mutation panel testing in patients with advanced cancer Proteomic profile of high-risk luminal A early breast cancers. First Author:
receiving treatment in an Irish teaching hospital. First Author: Hadia Khan, Nawale Hajjaji, Centre Oscar Lambret, Lille, France
Bon Secours Hospital, Cork, Ireland
Background: Breast cancer is a heterogeneous disease with a wide range of
Background: Tumor testing for potentially actionable somatic mutations via com- outcomes. Among the intrinsic breast cancer subtypes, luminal A tumors are
mercially available panel tests has entered routine clinical practice in many countries. considered to have a favorable prognosis. However, molecular studies char-
In Ireland the cost of these tests is not covered by insurance companies and must be acterizing the genomic landscape of luminal A tumors revealed a molecular
paid for by patients. Use of these tests is sporadic and depends on patient and heterogeneity within this subtype, which also translated to variability in sur-
clinician factors (including ability to pay). Existing data suggest that such testing vival. A better understanding of the biology of this tumor subgroup is therefore
results in a direct impact on patient therapy in a minority of patients only. We reviewed needed to determine the appropriate therapeutic strategy. The aims of the
the impact of somatic mutation testing on treatment selection and outcomes in study were to determine the frequency of high-risk luminal A tumors in a real
patients attending a medical oncology service in a teaching hospital in Ireland.
life cohort of early breast cancers and provide a proteomic characterization of
Methods: We performed a retrospective study of patients who had commercial panel
testing performed as part of routine oncology care. All patients opportunistically tested this subgroup using a mass spectrometry approach. Methods: 222 early breast
between 2013 and 2018 were included. Patients having focused molecular tests for cancer patients with hormone receptor positive and HER2 negative tumors
approved therapies (e.g. RAS mutations in colon cancer, EGFR and ALK mutations in treated at our institution had a PAM50-based genomic assay Prosigna
non-small cell lung cancer) were excluded.We reviewed medical records to assess the to estimate their risk of recurrence. This assay assigned each tumor sample
frequency and utility of mutations detected, the impact of testing on next and to an intrinsic molecular subtype of breast cancer. Luminal A and B tumors
subsequent lines of therapy, and the effectiveness of therapy. Results: 74 panel tests were analyzed with MALDI mass spectrometry imaging combined with
were performed in 71 patients. 39 tests (53%) detected mutations, of which 21 microproteomics, a spatially-resolved on-tissue shotgun proteomic technology,
(28%) were potentially actionable. 36 patients (51%) had further treatment after to determine the proteomic profiles of both cancer cells and stroma.
testing was performed. 9 tests (12%) led to test-based treatment. The mean duration Results: Among the 129 luminal A breast cancers identified in our cohort, 67
of test-based treatment was 34 days (range 1-90 days). No patients had benefit from (51%) had a risk of distant recurrence of 10% or more (32% had a 10% to 15%
test based treatment, defined as tumour response or disease stabilisation on restaging risk, and 19% a risk greater than 15%). High-risk luminal A tumors had a
scans. 23 patients died within 90 days of panel tests being requested. Among patients distinctive proteomic profile compared to low-risk luminal A or to luminal B
starting and completing a subsequent line of therapy after testing, the mean duration tumors. Overexpression of the methionine biosynthesis pathway was the main
of therapy with test-based treatment was 39 days (range 6-90) and for standard of differential protein expression observed in cancer cells and stroma of high-risk
care treatment was 56 days (range 1-262 days). Conclusions: While testing for tumor-
luminal A. Inflammation mediated by chemokine and cytokine signaling pathway
specific somatic mutations with proven predictive benefit is very useful, somatic
and integrin signaling were also overexpressed in high risk luminal A compared to
mutation panel testing for non standard of care genetic alterations is not of utility in
this real world setting. Its role in Ireland should be limited to identification of suitable luminal B. In the stroma of luminal B tumors, EGR signaling, Ras and FGF
early phase clinical trials. Discussions of panel testing should include frank discussion pathways and angiogenesis were overexpressed compared to high-risk luminal A
of expected benefits, and should also address factors such as patient ability to travel tumors. Conclusions: Real life data showed a significant proportion of high-risk
for clinical trials. luminal A breast cancers. MALDI mass spectrometry proteomics revealed dis-
tinctive tumor and microenvironment profiles in this breast cancer subgroup.
Number Percent
Total tests 74 100%
Mutation detected 39 53%
Potentially actionable 21 28%
Test-based treatment 9 12%

3078 Poster Session (Board #70), Sat, 8:00 AM-11:00 AM 3079 Poster Session (Board #71), Sat, 8:00 AM-11:00 AM
MET kinase domain rearrangements across 10 cancer types. First Author: An artificial intelligence approach to variant calling of ALK resistance
Jun Zhao, Beijing Cancer Hospital, Beijing, China mutations. First Author: Jochen K Lennerz, Massachusetts General Hos-
pital, Boston, MA
Background: MET is a transmembrane receptor tyrosine kinase and deregulated
in many kinds of tumors by mutation, rearrangement and amplification. Since Background: ALK tyrosine kinase inhibitors (TKIs) are effective in treating
the first constitutively active MET rearrangement (TPR-MET) was discovered, advanced anaplastic lymphoma kinase (ALK) fusion-positive non-small-cell
many other MET rearrangements have been identified in various tumor types. lung cancers (NSCLC), and specific ALK variants are associated with the
However, the frequency and characteristic of MET rearrangement in Chinese development of resistance to specific TKIs. Humans struggle to harness the full
cancer patients is still unclear. Methods: Targeted sequencing using 1021-gene potential of the highly complex next-generation sequencing bioinformatics
panel or 59-gene panel was performed on 3952 tissue samples and 5100 blood- pipeline output. As a consequence, the decision to report a variant remains
based ctDNA samples from 9052 unique patients across 10 cancer types. All difficult, and we considered the discrete nature of the data and the binary
MET exons were sequenced, but MET intronic breakpoints were not specifically decision (report vs. not-report) as an ideal setting to apply an artificial in-
baited. Results: 24 (0.27%) MET kinase domain rearrangements (KDRE) were telligence (AI) approach for variant reporting. Methods: We assessed di-
identified in 9052 patients. Specifically, 0.25% (16/6284) in non-small cell agnostic performance of an AI model in calling ALK-resistance mutations in n =
lung cancer (NSCLC), 0.69% (2/290) in gastric adenocarcinoma, 0.32% (2/ 50 consecutive ALK fusion positive patients who relapsed on TKI-therapy and
897) in colorectal cancer, 0.33% (2/610) in breast cancer, 0.3% (1/330) in underwent repeat biopsy at MGH. The random forest model was derived from
hepatocellular carcinoma and 0.69% (1/145) in ovarian cancer, none in 139 independent datasets (training and validation) capturing the reporting decision
pancreatic cancer, 136 thyroid cancer, 111 renal cell carcinoma and 110 on . 36,000 variants with ~500 features per variant resulting in a matrix of
esophageal squamous cell carcinoma. Among all of the MET KDRE, 17 were . 18 million data points. The model output is a contiguous prediction score
fusions with 5’ identified partner, 3 were kinase domain duplication (KDD) and 4 from 0 (not report) to 1 (report) and a visual drill-down functionality allows
were probable fusions with unidentified partner. The most common 5 ’partner exploration of the underlying features that contributed to the decision.
gene was CAPZA2, followed by CD47 and TES. In the MET KDRE cases in Results: Examination of n = 76 tests from n = 50 patients with a total of n =
NSCLC, 56.25% (9/16) did not found any clinical actionable variants referring 130 reported variants (and = 115 not reported variants) included a total of n =
to the NCCN guideline. In addition, MET amplification, EGFR L858R or exon 19 31 ALK point mutations: p.1156(n = 2), p.1171(n = 8), p.1174(n = 2),
deletion and KRAS mutation co-occurred in 25% (4/16), 18.75% (3/16) and p.1180(n = 2), p.1196(n = 1), p.1198(n = 1), p.1202(n = 8), p.1203(n = 1),
12.5% (2/16) of NSCLC MET KDRE cases respectively. Conclusions: Our re- p.1204(n = 1), p.1206(n = 1), p.1269(n = 4). Setting a screening thresh-
sults, for the first time, illustrate the MET KDRE across 10 cancer types among old of the model at . 10% for reporting showed only one false-negative
Chinese population and might provide some novel targets to develop new (p.Ile1171Asn) variant and 96.7% sensitivity. The average model score for
therapies for patients with MET KDRE. MET KDRE across different tumor types. ALK variants was 0.664 (range: 0.08–0.98; median 0.8) and did not show
Total Sam- Fusion with identified 5’ KDD Probable fusion with uniden-
significant differences from other reported variants (0.636; 0–1; 0.7; t-test
Tumor Types ples (n) partner (n) (n) tified partner (n) 5’ partner
0.66). The model would have called n = 18 of the non-reported control variants
Non-Small Cell
Lung Cancer
6284 12 2 2 CAPZA2(2); CD47(2); TES; CAV1; ITGA9; HLA-DRB1; TFEC;
CTTNBP2; ANK1; STARD3NL (average 0.07; range , 0.001–0.64; P , 0.0001) and was 84% specific.
Colorectal Cancer
Breast Cancer
897
610
1
1
1
0
0
1
ST7
TBX15 Review of the drill-down function identified prior call frequency, allelic ratio,
Hepatocellular
Carcinoma
330 1 0 0 TPM1
and predicted transcript consequences as common model features. Impor-
Gastric 290 2 0 0 CAPZA2; TES
Adenocarcinoma tantly, the model is currently agnostic to the medical literature and does not
Ovarian Cancer 145 0 0 1 　-
take clinical parameters (e.g. TKI type) into account, which may further im-
prove performance. Conclusions: Applying artificial intelligence to large dis-
crete datasets is one approach to help identify clinically relevant variants in the
setting of ALK resistance in ALK-fusion positive NSCLC.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
168s Developmental Therapeutics and Tumor Biology (Nonimmuno)

3080 Poster Session (Board #72), Sat, 8:00 AM-11:00 AM 3081 Poster Session (Board #73), Sat, 8:00 AM-11:00 AM
Whole transcriptome sequencing in metastatic cancer: A review of expres- Transcriptome-based cancer type prediction for tumors of unknown origin.
sion outliers in 113 metastatic breast cancer patients. First Author: Nathalie First Author: Jack Michuda, Tempus, Chicago, IL
LeVasseur, BC Cancer Agency, Vancouver, BC, Canada
Background: Tumors of unknown origin occur in approximately 5% of newly
Background: The genomic profiling of breast cancers has led to a greater diagnosed cancers and are difficult to treat without establishing the tissue
understanding of the mutational landscape of metastatic breast cancer type from which they derive. Establishing tumor origin guides standard of
(MBC) with potential therapeutic implications. Despite these advances, care treatment for several NCCN targeted therapy guidelines. Leveraging
there is a paucity of data regarding the additive value and relevance of gene tissue specificity in gene expression profiles, classification models based on
expression across histological and molecular subtypes, which represents the RNA expression offer a promising approach to identify the likely primary
majority of informative and actionable findings identified in the BC Cancer cancer site in tumors of unknown origin. Methods: In this study, we
personalized oncogenomics program (POG). Methods: Informative findings developed a transcriptome-based cancer type classifier trained on over
with potential clinical application from whole genome sequencing (WGS) 10,000 tissue samples annotated by pathologists and sequenced for RNA
and whole transcriptome sequencing (WTS) in MBC patients between 2012- expression to identify conserved patterns of expression characteristic of 30
2018 were reviewed. Variants observed in pathway genes of potential clinical tumor types across primary and metastatic tissue sites. The classifier
relevance, as defined by a curated list of genes, were examined across probabilistically ranks cancer of origin. Results: Overall, the accuracy of the
histological subtypes. High and low expression outliers relative to TCGA most probable cancer prediction was 85%, 88% within primary tumors and
breast cases, defined as expression greater than 98th percentile and FC . 2 77% within metastatic tumors. The top three cancers types with the highest
compared to Illumina breast dataset and lower than 25th percentile and accuracy were colorectal (accuracy in metastatic: 93%, accuracy in primary
FC , -2 compared to Illumina breast dataset, respectively, were then an- tumors: 99%), breast (95%, 96%) and lung (87%, 94%). Classifier per-
alyzed to establish how many outliers were observed in pathways of potential formance was lower in low-purity metastatic tumors where the surrounding
clinical relevance. Results: A total of 113 cases were included. WGS normal tissue obscures the tumor transcriptional profile, though the clas-
revealed that TP53 was the most frequent single nucleotide variant (SNV) in sifier still achieves 71% accuracy on metastatic tumors with less than 50%
triple negative breast cancer (23/30, 77%), whereas PIK3CA (37/78, 47%), purity. Conclusions: We present a novel method to probabilistically predict
PTEN (11/78, 14%) and ESR1 (19/78, 24%) were most frequent in ER tumor type for cancers of unknown origin using RNA-Seq. Our method
positive cases and CDKN2A (2/18, 11%) in HER2 positive cases. Across all achieves robust classification that is applicable to primary and metastatic
subtypes, the mTOR and cell cycle pathways were found to have the highest tumors and demonstrates the value of utilizing RNA-Seq to aid cancer di-
frequency of SNVs, with the identification of 86 and 71 variants, re- agnosis and treatment decisions.
spectively. Expression data for 113 RNA-sequenced patients revealed a high
frequency of expression outliers in the mTOR pathway (26 high expression
and 424 low expression outlier genes) and cell cycle pathways (35 high
expression and 331 low expression outlier genes), but also in the WNT
pathway (96 high expression and 490 lower expression outlier genes) and
NOTCH pathway (84 high expression and 564 low expression outlier genes).
Conclusions: Frequently identified SNVs across histological subtypes were
correlated with expression outliers in pathways of clinical relevance in breast
cancer. Additional informative findings, in pathways of potential clinical
relevance not historically targeted in breast cancer, were identified with
WTS. The clinical utility of these findings warrants further study.

3082 Poster Session (Board #74), Sat, 8:00 AM-11:00 AM 3083 Poster Session (Board #75), Sat, 8:00 AM-11:00 AM
Observational, multicenter, prospective study to assess the impact on Machine learning algorithm analysis using a commercial 592-gene NGS
patients’ outcome of a systematic screening of oncogenic drivers in ad- panel to accurately predict tumor lineage for carcinoma of unknown primary
vanced cancer: The GETHI XX-16 study. First Author: Carmen Beato, (CUP). First Author: Jim Abraham, Caris Life Sciences, Phoenix, AZ
Hospital Virgen de la Macarena, Sevilla, Spain
Background: The diagnosis of a malignancy is typically informed by clinical
Background: Identification of “agnostic” genetic drivers in cancer is fore- presentation and tumor tissue features including cell morphology, immuno-
seen as a major step forward in precision medicine.Unfortunately, “off label” histochemistry, cytogenetics, and molecular markers. However, in approxi-
use of targeted therapies is not widely available and many oncogenic al- mately 5-10% of cancers, ambiguity is high enough that no tissue of origin can
teration do not present the same behaviour accross all tumor types.We aimed be determined and the specimen is labeled as a Cancer of Occult\Unknown
to analyze the real impact on patients management of the implementation Primary (CUP). Lack of reliable classification of a tumor poses a significant
of a systematic screening of genetic alterations in centers of the Spanish treatment dilemma for the oncologist leading to inappropriate and/or delayed
Group for Rare Cancer (GETHI). Methods: We designed an observational, treatment. Methods: 40,000 tumor patients with NGS data were used to
prospective and multicenter study to molecularly characterize any adult construct a multiple parameter lineage-specific classification system using an
patient with advanced cancer.Formalin fixed paraffin-embedded samples advanced machine learning approach. The dataset for each classifier was split
were studied by TrkA-C,ROS1 and ALK immunohistochemistry followed 50% for training and the other 50% for testing. The training task for each
by RT-PCR when positive to confirm gene fusions. Additonally, the Next classifier was to identify the cases that were similar to the cases it was trained
Generation Sequencing paltform ArcherFusion Plex (able to detect point on against a backdrop of randomly selected cases of other histological origins.
mutations and rearrangements in 53 cancer related genes) was imple- Results: Tumor lineage classifiers predicted the correct classifications where
mented.Clinical data regarding treatment administered and outcome, were the primary site was known with accuracies ranging between 85% and 95%.
collected from patients identified as harboring drugable alterations. When applied to CUP cases (n = 500), an unequivocal result could be obtained
Results: Up to 26 hospitals all over the country got involved in the study. 341 100% of the time. Conclusions: Lineage predictors can render a histologic
tumoral tissues, representing 41 different histologies were collected. Mo- diagnosis to CUP cases that can inform treatment and potentially improve
lecular studies could be performed in 292 samples that led to the identi- outcomes.
fication of 33 patients as harboring somatic oncogenic mutations. 21 were
considered druggable and 5 got targeted therapy directed against the al-
teration identified (three glioblastoma patients with EGFR mutations re-
ceived erlotinib, one prostate cancer with a BRAF fusion received trametinib
and one lung cancer with ALK translocation, previously deemed as negative
by standard screening, received crizotinib). One of the glioblastoma patients
achieved a long lasting stabilitation and both the prostate and lung tumors
presented dramatic partial responses. Conclusions: Though only few cases
harboring drugable alteratons got specif treatment, 50% achieved a
meanignful benefit. A wide access to molecular screening and targeted drugs
could improve the outcome of cancer patients.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Therapeutics and Tumor Biology (Nonimmuno) 169s

3084 Poster Session (Board #76), Sat, 8:00 AM-11:00 AM 3085 Poster Session (Board #77), Sat, 8:00 AM-11:00 AM
Determining clinical relevance of genomic heterogeneity in an ethnically First-in-human study of AZD5153, a small molecule inhibitor of bromodo-
diverse cohort of newly diagnosed patients with breast cancer. First Author: main protein 4 (BRD4), in patients (pts) with relapsed/refractory (RR)
Padma Sheila Rajagopal, The University of Chicago, Chicago, IL malignant solid tumor and lymphoma: Preliminary data. First Author:
Judy Sing-Zan Wang, Florida Cancer Specialists and Sarah Cannon Research
Background: The trajectory from early breast cancer to distant metastasis
Institute, Sarasota, FL
has not been precisely characterized. We are building an ethnically diverse,
longitudinal cohort of prospectively ascertained breast cancer patients with Background: BRD4 is a bromodomain and extraterminal (BET) protein that
integrated genomic, transcriptomic, epidemiological and clinical data, with regulates oncogenic programs by modifying gene transcription and additional
the goal of identifying biomarkers that can improve on clinical predictors. mechanisms. AZD5153 is a novel, reversible BRD4 inhibitor with bivalent
Methods: Our goal is 500 histologically confirmed invasive cases with a mechanism of action and enhanced antitumor activity in preclinical models.
minimum follow-up of 5 years. To date, Tempus Labs, Inc. has completed xT This phase 1, multicenter, dose escalation study (NCT03205176) assesses
assays (595-gene panel DNA-seq and full-transcriptome RNA-seq) on 127 AZD5153’s safety, pharmacokinetics (PK), and pharmacodynamics (PD). We
cases with matched tumor-normal samples. Clinical information was obtained report here preliminary, unvalidated data from AZD5153 monotherapy in pts
from electronic health records and our cancer registry. Results: Median age of with RR solid tumor, including lymphoma. Methods: Adult pts received oral
diagnosis was 51, with 47% African-Americans. 73% had stage 2 or 3 disease AZD5153 QD/BID to determine the MTD. During dose escalation, a continual
(2 patients were stage 4). 24.4% had TNBC; 30% had HER2+ and 62.2% reassessment model was used to estimate toxicity and all final decisions were
HR+ status. Somatic alterations were identified in 560 genes with most made by the Safety Review Committee. PK and PD were characterized using
common mutations in TP53 (56%), MCL1 (35%), PIK3CA (30%) and ERBB2 standard methods. Results: As of 1 Nov 2018, 28 pts (78.6% female, median
(17%). Somatic mutations in BRCA1 (10%) and BRCA2 (5%) were associated age 66.5 y) were treated in 7 cohorts: 2 mg QD (3 pts), 5 mg QD (3 pts), 10 mg
with increased tumor mutation burden (TMB). 2 patients were MSI high; 6 QD (3 pts), 10 mg BID (5 pts), 15 mg BID (4 pts), 20 mg BID (7 pts), and
were equivocal. After a median follow-up of 6.5 years, 46 patients died and 38 30 mg QD (3 pts). Treatment was ongoing in 8 pts at data cut-off. Safety
had recurrent disease. With adjustment for clinical factors, TMB showed a findings showed 50% of pts experienced treatment-related AEs. 25% of pts
slight nonsignificant negative association with recurrence-free survival (HR: experienced treatment-related Grade $3 AEs, which were thrombocytopenia
0.97, 95% CI: 0.91-1.00, p = 0.31). TP53 was associated with recurrence- and fatigue (7.1% each), and anemia, diarrhea, and platelet count decreased
free survival (HR: 1.76, 95% CI 0.94-3.29; p = 0.08) as was MSI (HR: 0.24, (3.6% each). SAEs were observed in 25% of pts; none of the SAEs was at-
95% CI: 0.06-1.06, p = 0.06). Conclusions: These data support the impor- tributable to AZD5153 alone. Dose-limiting toxicities of thrombocytopenia (1
tance of integrating tumor sequencing in a diverse cohort with full clinical pt) and diarrhea with herpetic rash leading to discontinuation (1 pt) occurred at
annotation to assess the diversity of actionable genomic alterations at the time 20 mg BID. 53.6% of pts discontinued due to disease progression. Total
of primary diagnosis to develop interventions for prevention of metastases. median treatment duration was 1.3 mo (range up to 8.9 mos). Dose pro-
portional increase in Cmax and AUC were observed across the dose range tested.
Tmax ranged from 0.5 to 3 h and t1/2 was 6 h. Dose-dependent changes in
expression of target genes (eg, HEXIM1, HIST2H2BF, CD274, and CCR2) and
platelet counts were observed in the peripheral blood. Conclusions: AZD5153
monotherapy is safe and tolerated at doses up to 30 mg QD and 15 mg BID.
Linear increase in PK was observed. Additional safety and efficacy updates will
be reported at the annual meeting. Clinical trial information: NCT03205176.

3086 Poster Session (Board #78), Sat, 8:00 AM-11:00 AM 3087 Poster Session (Board #79), Sat, 8:00 AM-11:00 AM
c-AMP/MAPK dysregulation and its impact on survival and response to PIPA: A phase Ib study of selective ß-isoform sparing phosphatidylinositol
immunotherapy in advanced melanomas. First Author: Rami Al-Rohil, 3-kinase (PI3K) inhibitor taselisib (T) plus palbociclib (P) in patients (pts)
Duke University Medical Center, Durham, NC with advanced solid cancers—Safety, tolerability, pharmacokinetic (PK),
Background: Immunotherapies blocking the interaction of CTLA-4 or Programmed Death 1 (PD-1) with their and pharmacodynamic (PD) analysis of the doublet combination. First
ligands are the standard of care for advanced MEL although many pts fail to benefit from immunotherapy. Author: Juanita Suzanne Lopez, Drug Development Unit-The Institute of
Herein, we seek to identify epigenetic and genetic signatures associated with lack of response in patients Cancer Research and The Royal Marsden NHS Foundation Trust, Sutton,
treated with immunotherapy. Methods: We performed whole exome sequencing of 580 cancer-related genes
and whole-genome DNA methylation arrays targeting . 850k CpG sites covering promoters, enhancers, and
United Kingdom
transcription factor sites in 28 MEL samples from patients treated with PD-1 +/- CTLA-4 inhibitors. Findings
were correlated with collected clinical history. Results: Findings are summarized in the table. Unsupervised
Background: Oncogenic hyperactivation of the PI3K pathway is common in
clustering and multi-parametric analysis showed a distinct methylation signature independent of age, sex, multiple cancers, with preclinical data showing that CDK4/6 inhibitors
stage, site of metastasis, or type of treatment (adj. p,0.01). Pathway analysis identified c-AMP/MAPK and sensitise PIK3CA mutant cancers to PI3K inhibitors. We report the activity of
PI3K-Akt signaling pathways (adj. p=2.10E-05 and 8.19E-06, respectively) enrichment in non-responders.
This coincided with significant increase of mutational events in c-AMP/MAPK and PI3K-Akt pathways (p=
the P+T in solid tumors with PI3K pathway activation, along with the PD
0.0001) including deleterious events affecting PDE4DIP (p=0.0002), a negative regulator of mTORC1, in non- biomarker analysis. Methods: We previously reported the dose escalation
responders. C-AMP/MAPK/PI3K genomic alterations were associated with a worse OS and PFS but not worse phase identifying 125mg P given 3-weeks-on, 1-week-off in combination
response rate (p=0.04, p=0.01, and p=0.20). PDE4DIP deleterious events were associated with decreased with T 2mg as the recommended phase 2 dose (R2PD, Lim, ASCO 2017). We
response rate, worse OS and PFS (p=0.002; p=0.002;p=0.00003). Conclusions: Convergent epigenetic
dysregulation of cAMP/MAPK signaling and inactivation of PDE4DIP is associated with worse outcome and lack report the results in solid tumors with confirmed activating mutations (mts)
of response to immunotherapy, respectively. in the PI3K pathway, from dose escalation and expansion, with no prior
Feature Number P- value
exposure to CDK4/6 or PI3K pathway inhibitors. PD studies include analyses
Gender of platelet-rich plasma (PRP) and paired tumour biopsies. Results: 20 pts
Responders
Male 9 (64%)
p=0.98
(median age 61, range 34-72) were treated at the doublet RP2D, M/F 7/13,
Female
Non-responders
5 (36%) with a median 4 prior treatments (range 2-11). Tumour types included
Male
Female
9 (64%)
5 (36%)
colorectal, breast, lung, endometrial,oligodendroglioma and head and neck
Metastatic Stage
Responders
p=0.24 cancers. Durable disease control occurred in 3 patients with ER+ advanced
(M1a)
(M1b)
1 (6%)
6 (36%)
breast cancer with responses lasting .6 months including 1 pt with a
(M1c)
Non-responders
7 (58%) H1047R PIK3CAmt with an ongoing RECIST PR.36 cycles, 2 pts with
(M1a) 4 (36%) PIK3CAmt colorectal cancer had RECIST SD for .5 months, and 1 patient
(M1b) 3 (21%)
(M1c) 7 (43%) with a PIK3CGmt anaplastic oligodendroglioma had clinical and radiological
Median OS/c-AMP/MAPK/PI3K pathway dysregulation p=0.04
Present 48 months benefit lasting 5.5 months. Treatment was well tolerated with predictable
95% CI (0.24;0.83)
Absent 90 months G1-2 adverse events (AEs). G3 toxicities of neutropenia (n=6), thrombo-
95% CI (0.21;0.90)
Median PFS/c-AMP/MAPK/PI3K pathway dysregulation p=0.01 cytopenia (2), rash (2), mucositis (1) and raised transaminases (1 each) were
Present 40 months
95% CI (0.40;0.80) all transient with no G4/5 AEs. Significant decreases in tumour pRb, and
Absent 95 months
95% CI (0.13;0.0.90) pAKT and pGSK3ß in PRP, confirmed modulation of both CDK4/6 and PI3K
PDE4Pdeleterious events /Response rate
Present 4 (29%)
p=0.0021
pathways at R2PD. Conclusions: Doublet P+T is well tolerated at the
Absent
Median OS/PDE4P deleterious events
10(71%)
p= 0.002
combination RP2D, with PD evidence of PI3K and CDK4/6 modulation in
Present 39 months
95% CI (0.27;0.97)
both plasma and tumor. Promising preliminary anti-tumor activity is seen in a
Absent 90 months
95% CI (0.22;0.95)
mixed histology cohort selected for activating PIK3 mutations. Clinical trial
Median PFS/PDE4P deleterious events
Present 9 months
p=0.00003 information: NCT02389842.
95% CI (0.31-0.95)
Absent 30 months
95% CI (0.36-0.90)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
170s Developmental Therapeutics and Tumor Biology (Nonimmuno)

3088 Poster Session (Board #80), Sat, 8:00 AM-11:00 AM 3089 Poster Session (Board #81), Sat, 8:00 AM-11:00 AM
Distinct radiological patterns of drug-induced pneumonitis (R-DIP) in early- Phase 1 of ABTL0812, a proautophagic drug, in combination with paclitaxel
phase clinical trials and predictive factors affecting outcome: A 10-year and carboplatin at first-line in advanced endometrial cancer and squamous
systematic review from the Royal Marsden Hospital Phase I Drug Develop- cell lung carcinoma. First Author: Lorena Farinas-Madrid, Vall d’Hebron
ment Unit experience. First Author: Angelika Terbuch, Royal Marsden University Hospital Institute of Oncology (VHIO), Barcelona, Spain
Hospital and The Institute of Cancer Research, Sutton, United Kingdom
Background: ABTL0812 is a novel anti-cancer agent that induces a strong
Background: We studied clinical and radiological parameters influencing DIP autophagy-mediated cell death by a dual mechanism. It inhibits the Akt/
in patients (pts) participating in phase I clinical trials, aiming to investigate mTOR axis by upregulating TRIB3, an endogenous Akt inhibitor, and in-
predictive factors affecting DIP, in particular those affecting outcome. duces reticular (ER)-stress. Preclinical data in squamous non-small cell lung
Methods: 2439 consecutive stage IV cancer pts on phase I clinical trials from carcinoma (Sq-NSCLC) and endometrial cancer (EC) has indicated drug
2007 to 2017 were identified. Pts with respiratory symptoms or abnormal lung efficacy as a single agent and potentiation of chemotherapy. Methods: A
imaging were reviewed in detail, with longitudinal analysis of imaging by an phase 1 clinical study was designed where ABTL0812 was administered
experienced radiologist. R-DIP was categorized according to internationally orally in combination with 175 mg/m2 paclitaxel/carboplatin AUC5 D1 every
recognized criteria. Results: 60 pts developed R-DIP (overall incidence 2.5%); 3 weeks (P/C), and posterior ABTL0812 as a maintenance therapy until
most frequent in pts receiving drug conjugates (31.1%) followed by targeted disease progression or unacceptable toxicity. The study included first-line
therapies (8.3%). Hypersensitivity pneumonitis was most common (33.3%) patients (pts) with advanced Sq-NSCLC or advanced/recurrent EC. The
followed by non-specific interstitial pneumonitis (30%) and cryptogenic design included a 3+3 de-escalation trial followed by an expansion cohort,
organising pneumonitis (26.7%). 45% pts who developed R-DIP were clini- where the starting dose of ABTL0812 was 1300 mg tid and if at least 2 pts
cally asymptomatic. The number of affected lobes (OR 1.47, 95% CI: 1.19- experienced a DLT, the dose level would be de-escalated to 1000 mg tid.
1.81, p , 0.001) and the pattern of R-DIP (OR 5.83 for ARDS, 95% CI: 0.38- Safety and tolerability were the primary endpoints and preliminary efficacy
90.26, p = 0.002) were significantly associated with a higher CTCAE according to RECIST v1.1 criteria and pharmacodynamic biomarkers (TRIB3
pneumonitis grading. 23% pts (14/60) had investigational medicinal product and CHOP an ER-stress biomarker, by qPCR in whole blood) were the
(IMP) temporarily discontinued or had a dose reduction while 42% pts (25/60) secondary endpoints. Results: 16 EC and 5 Sq-NSCLC pts were enrolled.
had IMP permanently discontinued. 48% pts were treated with steroids. The One DLT, a grade 4 neutropenia, appeared in the first cohort of 6 pts and no
number of affected lobes, pattern of R-DIP and steroid therapy did not in- de-escalation was applied. Fourteen pts were included in an expansion
fluence an improvement in R-DIP (p = 0.65, 0.27 and 0.23 respectively). cohort with the same dose level (1300 mg tid), and 1 DLT, a grade 3 febrile
Continuation of treatment resulted in worsening of DIP in 42.9% of cases. The neutropenia, was observed. Therefore, the dose of 1300 mg tid was selected
only predictive factor for an improvement in DIP was an interruption of as RP2D in combination with P/C. Most frequent grades 2-4 AEs were
treatment (OR 0.05, 95% CI: 0.01-0.35, p = 0.01). 14 pts were retreated neutrophil count decreased (n = 9, 43%), nausea (n = 5, 24%), fatigue (n =
with a reoccurrence of R-DIP in 4 pts (28.6%). Conclusions: R-DIP from novel 4, 19%), followed by anemia, vomiting, dyspepsia, platelet count decreased,
agents in early phase clinical trials presents in varied radiological patterns, with arthralgia and neurotoxicity (n = 2, 10% each). Seventeen pts (13 EC and 4
findings often preceding clinical symptoms. Treatment interruption leads to Sq-NSCLC) who completed at least two treatment cycles were evaluable for
improvement of DIP and should be considered early. Close clinical and ra- efficacy; 1 CR (EC), 8 PR (7 EC and 1 Sq-NSCLC), 7 SD (5 EC and 2 Sq-
diological surveillance is recommended should IMP be restarted. NSCLC) and 1 PD (Sq-NSCLC) were observed. Pharmacodynamic bio-
markers showed increased TRIB3 and CHOP levels. Conclusions: The
combination of ABTL0812+P/C was safe and tolerated, efficacy signals were
observed, and biomarker modulation confirmed drug activity. The triple
combination is currently being evaluated in both indications in a Phase 2
study. Clinical trial information: NCT03366480.

3090 Poster Session (Board #82), Sat, 8:00 AM-11:00 AM 3091 Poster Session (Board #83), Sat, 8:00 AM-11:00 AM
Design, engineering, and characterization of a novel long-acting (Pegylated) A phase Ib study of prexasertib, a checkpoint kinase (CHK1) inhibitor, and
single isomer human arginase for arginine depriving anticancer treatment. LY3023414, a dual inhibitor of class I phosphatidylinositol 3-kinase (PI3K)
First Author: Kuo-Ming Yu, Athenex, Inc., Hong Kong, China and the mammalian target of rapamycin (mTOR) in patients with advanced
solid tumors. First Author: David S. Hong, The University of Texas MD
Background: Arginine deprivation therapy is an attractive strategy to treat
Anderson Cancer Center, Houston, TX
arginine-auxotrophic cancers with deficient expression of argininosuccinate
synthetase, argininosuccinate lyase or ornithine transcarbamylase. We have Background: Prexasertib inhibits CHK1, a kinase involved in DNA repair and
designed and engineered a novel human arginase with single site pegylation replication. LY3023414 inhibits PI3K/mTOR signaling, implicated in the
exerting excellent preclinical pharmacologic profile to serve as a new class of development of malignant disease. Prexasertib + LY3023414 has resulted
therapy. Methods: Human arginase has three cysteines (at position 45, 168, in enhanced antitumor activity in triple negative breast cancer (TNBC)
303) and none of them is in or close to the active site. Two cysteines were in vitro models. Methods: This Phase 1b study in patients (pts) with solid
mutated to serines, leaving the only cysteine at 45 for the simple and cost- tumors assessed escalating doses of prexasertib (60-105 mg/m2 IV every
effective synthesis of a single isoform of pegylated human arginase. Different 14 days [q14d]) and LY3023414 (100-200 mg orally twice daily [BID]).
forms of PEG moieties were evaluated for the selection of a drug candidate Dose escalation ceased once the maximum tolerated dose of each mono-
(PT01), followed by extensive characterization. Results: Converting Cys at therapy was reached. An initial expansion cohort (Arm E) explored prexasertib
168 and 303 to serine impacted least on enzymatic activity (with cobalt 105 mg/m2 q14d + LY3023414 200 mg BID. Subsequent expansion cohorts
cation). Pegylation with different sizes and shapes showed that 20 and evaluated prexasertib 105 mg/m2 q14d + LY3023414 150 mg BID in pts with
40 kDa (linear and branched) had similar PK/PD profile without damaging solid tumors with PIK3CA mutations (Arm E2) or TNBC (Arm E3). Results: Fifty
enzymatic activity. Therefore, arginase modified with a linear 20 kDa PEG pts were enrolled (escalation: n = 13; Arm E: n = 9; Arm E2: n = 15; Arm E3: n =
was chosen as the candidate. A single 0.4 mg/kg IV dose of PT01 in rats 13). No dose-limiting toxicities (DLTs) were observed during escalation
induced 4 days of near complete plasma arginine depletion, while 6–7 days however DLT-equivalent toxicities were observed in 2 pts in Arm E (anemia,
of depletion between 1.2 and 2 mg/kg. Plasma arginine levels were re- neutropenia, thrombocytopenia, oral mucositis, abdominal pain, fatigue). Due
versible. First-order clearance of both plasma PT01 concentration and to toxicity, a reduced dose of LY3023414 (150 mg BID) was assessed in Arm
activity suggested a terminal half-life of about 20 hours. In vitro assay E2/E3. In the 28 patients treated in Arms E2/E3, common treatment-related
showed very potent cytotoxicity at sub-nM level against various cell lines of adverse events (any grade; grade $3) were: leukopenia/neutropenia (82%;
breast, prostate, and pancreas in origins. In two mouse cancer models (hard- 79%), thrombocytopenia (46%; 36%), nausea (46%; 0%), stomatitis (39%,
to-cure pancreas and castration-resistant prostate), weekly infusion at 5 and 4%), vomiting (36%; 0%), and anemia (29%; 18%). Febrile neutropenia was
10 mg/kg induced significant tumor growth inhibition of 44-67%. All mice reported in 25% of pts. Dose reductions in Arm E2/E3 were common. In
experienced dose-dependent but rapidly reversible weight loss following escalation, 2 pts achieved a partial response (PR) and 3 pts achieved stable
each weekly dose. Conclusions: A novel single isoform of pegylated human disease (SD). In Arm E, 78% of pts achieved SD. Of the pts evaluable at the
arginase was created, showing excellent enzymatic activity, PK/PD profiles, time of data transfer, PRs were achieved in 1 pt with an unknown primary (Arm
and cytotoxicity in vitro. Mouse xenograft models showed good tumor growth E2) and 2 pts with TNBC (Arm E3). Each agent’s pharmacokinetic profile was
inhibition activity with tolerable toxicity as manifested on transient weight consistent with prior monotherapy data. Conclusions: Prexasertib +
loss during therapy. LY3023414 showed preliminary efficacy in heavily pretreated pts with solid
tumors but was associated with toxicity, suggesting supportive care may be
required. Clinical trial information: NCT02124148.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Therapeutics and Tumor Biology (Nonimmuno) 171s

3092 Poster Session (Board #84), Sat, 8:00 AM-11:00 AM 3093 Poster Session (Board #85), Sat, 8:00 AM-11:00 AM
Preclinical testing of ultra-rapid FLASH total abdominal irradiation demon- First-in-human imaging of nanoparticle entrapped docetaxel (CPC634) in
strates survival benefit and decreased gastrointestinal toxicity compared to patients with advanced solid tumors using 89Zr-Df-CPC634 PET/CT. First
conventional external beam radiation. First Author: Karen Levy, Stanford Author: Iris H.C. Miedema, Amsterdam UMC, Vrije Universiteit Amsterdam,
University, Stanford, CA Department of Medical Oncology, Cancer Center Amsterdam, Amsterdam,
Netherlands
Background: Total abdominal irradiation (TAI) is not widely used in the
treatment of ovarian cancer due to high abdominopelvic toxicity. Ultra-rapid Background: CPC634 is a nanoparticle entrapping docetaxel designed to
FLASH irradiation has been shown to spare the lung, skin and brain from improve tumor accumulation and tolerability compared to conventionally
radiation toxicity in preclinical models. Conventional radiotherapy delivers a administered docetaxel by taking advantage of the presumed enhanced
dose-rate of 3-4 Gy/minute, while our small animal FLASH system uses a permeability and retention (EPR) effect. In vivo imaging with zirconium-89
linear accelerator to generate a dose-rate of 200 Gy/second. Our data (89Zr)-desferal (Df)-CPC634 will provide valuable information on its bio-
demonstrates that use of FLASH-TAI in a preclinical model protects against distribution and will quantify tumor retention. Methods: Patients with solid
death from irradiation and confers gastrointestinal (GI) protection when tumors not amenable to standard therapy received 37 MBq, 0.1-2mg of 89Zr-
compared to conventional external beam irradiation. Ongoing studies are Df-CPC634 tracer and whole body PET/CT scans were obtained at 2, 24 and
evaluating the potential for tumor control in an ID8 mouse model of ovarian 96h post-injection (p.i.). Patients were administered CPC634 (60mg/m2)
cancer. Methods: Female C57BL/6 mice received TAI using FLASH and two weeks later followed by a second tracer injection and scans at 24 and
conventional (CONV) radiation at increasing doses: 8.5 Gy, 10.5 Gy and 12 96h p.i. Biodistribution was quantified by delineating organs of interest and
Gy. Unirradiated controls and irradiated cohorts were analyzed at 5-days and calculating mean %ID/kg. Visual tumor retention was defined as focal uptake
12 months post-irradiation. Normal tissue toxicity was determined by in tumor lesions exceeding local background and quantified as standardized
measuring total body weights, stool counts, laboratory analysis, histological uptake peak values (SUVpeak) in volumes of interest. Results: Five patients
analysis, immunohistochemistry, immunofluorescence microscopy and were included. Biodistribution of 89Zr-Df-CPC634 showed significant re-
survival. Results: Solid stool production was preserved in FLASH mice, tention in healthy liver, and spleen compared to lung (respectively 2.54,
whereas a 50-63% decrease was observed in the CONV cohorts. Histology 1.61 and 0.56 mean %ID/kg at 96h p.i.), supporting apparent opsonization
demonstrated that FLASH preserves small intestinal architecture. TUNEL of nanoparticles in cells of the reticuloendothelial system. Visual retention
analysis demonstrated an increase in apoptosis throughout the small in- was observed in 16/37 evaluable tumor lesions with the highest intensity at
testine of only the CONV cohort. Exploratory necropsy of all surviving cohorts 96h p.i, compatible with the assumed EPR effect. Tumor retention showed
at 12 months post-irradiation was notable for secondary transmural proximal intra- and interpatient heterogeneity, with a mean %ID/kg of 3.43 [1.14-
duodenal adenocarcinomas within the radiation field in 25% of only the aged 9.32]. Pre-administering unlabeled CPC634 did not change the mean tumor
CONV cohorts. There was no laboratory evidence of long-term hematopoietic, retention of 89Zr-Df-CPC634 (at 96h p.i. mean 3.50 %ID/kg [1.64-9.97]),
liver or kidney toxicity at 12 months. Survival analysis was notable for death however, four additional lesions were visible in comparison to tracer only.
of all 12 Gy CONV mice by 9 days post-irradiation whereas 75% of the 12 Gy Conclusions: The biodistribution of 89Zr-Df-CPC634 was consistent with a
FLASH mice were alive at 11 months. Conclusions: FLASH protects against prolonged exposure of nanoparticle containing docetaxel. 89Zr-Df-CPC634
death from TAI, improves small intestine epithelial integrity following TAI, showed high retention in tumors confirming the EPR effect of these
protects against radiation-induced apoptosis and may protect from sec- nanoparticle in humans, and supporting their further development for tumor
ondary gastrointestinal tumors in the radiation field. Our discovery that targeting of therapeutic agents. A Phase II efficacy study in platinum re-
FLASH is a safe strategy to deliver effective doses of total abdominal ra- sistant ovarian cancer (NTC03742713) is currently ongoing. Clinical trial
diation potentially identifies a new opportunity to utilize FLASH-TAI for information: NCT03712423.
treatment of ovarian peritoneal metastases.

3094 Poster Session (Board #86), Sat, 8:00 AM-11:00 AM 3095 Poster Session (Board #87), Sat, 8:00 AM-11:00 AM
A first-in-human phase I/II trial of SRA737 (a Chk1 Inhibitor) in subjects with A phase I/II first-in-human trial of oral SRA737 (a Chk1 inhibitor) given in
advanced cancer. First Author: Elizabeth Ruth Plummer, Northern Centre for combination with low-dose gemcitabine in subjects with advanced cancer.
Cancer Care, Newcastle-upon-Tyne, United Kingdom First Author: Udai Banerji, Drug Development Unit-The Institute of Cancer
Research and The Royal Marsden NHS Foundation Trust, Sutton, United
Background: SRA737 is a potent, highly selective and orally-bioavailable
Kingdom
inhibitor of checkpoint kinase 1 (Chk1). SRA737-01 was designed to in-
vestigate the safety and tolerability of continuous, daily dosing with SRA737 Background: SRA737 is a potent, highly selective and orally-bioavailable
and to evaluate preliminary efficacy in expansion cohorts of prospectively- inhibitor of checkpoint kinase 1 (Chk1). SRA737-02 was designed to in-
selected genetically-defined subjects with advanced tumors. Methods: The vestigate the safety, tolerability and preliminary activity of SRA737 in a novel
escalation phase employed an accelerated titration design starting at 20 mg combination with sub-therapeutic doses of gemcitabine (low dose gemcita-
administered orally in 28-day cycles. Incremental 100% dose escalations in bine; LDG) utilized to potentiate SRA737’s activity by induction of replication
single-subject cohorts were followed by a rolling-6 design once SRA737- stress in subjects with advanced solid tumors. Methods: Phase 1 dose es-
related $ Grade 2 toxicity was observed during Cycle 1. The expansion phase calation investigated cohorts of 3 to 6 subjects receiving escalating doses of
enrolled subjects prospectively selected by next-generation sequencing SRA737 for 2 days after LDG administration on days 1, 8, 15 of 28-day cycles.
with: high grade serous ovarian, colorectal, metastatic castration-resistant Phase 2 expansion cohorts explored the hypothesis that LDG strongly syn-
prostate, non-small cell lung, and head and neck cancers. Results: In es- ergizes with SRA737 in subjects with genetically-defined tumors hypothesized
calation, 18 subjects received SRA737 in 9 dose level cohorts, from 20 to to be sensitive to Chk1 inhibition: urothelial, high grade serous ovarian, small
1300 mg QD; median treatment duration 62.5 days (range 1 to 226). Of cell lung, soft tissue sarcoma, and cervical or anogenital cancers. Results: A
these subjects, 3 experienced dose limiting toxicity (DLT; inability to receive total of 55 subjects received SRA737 in 13 dose escalation cohorts at doses of
75% of the planned dose); 2 at 1300 mg QD due to gastrointestinal in- 40 to 600 mg SRA737 combined with LDG doses of 50 to 300 mg/m2. No
tolerability and 1 at 500 mg BID due to thrombocytopenia. The maximum protocol-defined dose limiting toxicities (DLTs) have been observed. The
tolerated dose (MTD) was established at 1000 mg QD or 500 mg BID. The pharmacokinetic profile of SRA737 revealed an AUC0-24 and Cmax of
Cmax and AUC0-24 at 1000 mg QD were 2391 ng/mL and 26795 ng∙h/mL 3550 ng∙h/mL and 548 ng/mL at 150 mg SRA737. At this dose, the Cmin (52
respectively and the Cmin (411 ng/mL) exceeded that determined in pre- ng/mL) exceeded that determined in preclinical models to be effective. En-
clinical models to be effective. Doses $ 300 mg QD also exceeded this level. rollment into expansion cohorts was initiated at 500 mg SRA737 plus 100 mg/
Of 462 subjects prospectively screened for genetic alterations associated m2 LDG with intra-patient dose escalation permitted to 250 mg/m2 LDG.
with Chk1 sensitivity, 93 were enrolled in expansion across all tumor types. Approximately 80 subjects were planned and 82 have been treated. Me-
Overall, the most commonly reported treatment-emergent adverse events dian treatment duration was 51 days (range 1 to 358). The most common
were diarrhea (70%), nausea (64%), vomiting (51%), and fatigue (47%); the treatment-emergent adverse events were nausea (53%), vomiting (45%),
majority were of mild to moderate severity. Conclusions: In this first-in- fatigue (40%), diarrhea (38%), and anemia (28%); the majority were of mild to
human trial of SRA737 monotherapy, the MTD was 1000 mg/day and based moderate severity. Proof-of-concept clinical activity has been seen in tumor
on overall tolerability and PK, the recommended Phase 2 dose is 800 mg/ types such as anal, cervical, and rectal. Conclusions: The combination of LDG
day. The successful enrollment of prospectively-selected genetically- and SRA737 has been well tolerated. This first-in-human clinical study
defined subjects will allow response data to be correlated with genomic provides proof-of-concept that sub-therapeutic LDG effectively potentiates
profiles hypothesized to confer sensitivity to Chk1 inhibition. Clinical trial SRA737. This novel replication stress-targeted therapy warrants further
information: NCT02797964. evaluation in genetically pre-defined solid tumors. Clinical trial information:
NCT02797977.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
172s Developmental Therapeutics and Tumor Biology (Nonimmuno)

3096 Poster Session (Board #88), Sat, 8:00 AM-11:00 AM 3097 Poster Session (Board #89), Sat, 8:00 AM-11:00 AM
Nanoparticle entrapped docetaxel (CPC634) enhances intratumoral doce- A phase I study of the APE1 protein inhibitor APX3330 in patients with
taxel exposure compared to conventional docetaxel (Cd) in patients with advanced solid tumors. First Author: Safi Shahda, Indiana University Melvin
solid tumors. First Author: Florence Atrafi, Erasmus MC Cancer Institute, and Bren Simon Cancer Center, Indianapolis, IN
Rotterdam, Netherlands
Background: APX3330 is an orally administered anti-cancer, anti-CIPN agent
Background: Failure or resistance to chemotherapy may be caused by sub- targeting the APE1 protein. APE1 maintains NFkB, STAT3, AP-1 and HIF-1a
therapeutic intratumoral drug levels. Nanomedicine aim to improve intra- in a reduced form, acting as a regulator of transcription factors. A dual function
tumoral drug exposure. CPC634 is a nanoparticle entrapping docetaxel. We protein, APE1 also plays a role in protecting against oxidative DNA damage in
hypothesized that CPC634 increases intratumoral docetaxel exposure. neurons. APX3330 is a highly selective inhibitor of APE1 redox function in
Methods: In this randomized cross-over study we assessed intratumoral and tumors that enhances the neuronal protection function of APE1. Methods: We
plasma pharmacokinetics (PK) of docetaxel after intravenous administration report on study NCT03375086 evaluating APX3330 in patients with incurable
of CPC634 and conventional docetaxel (Cd). The study was powered to malignancies. Eligibility required adequate organ function, PS 0-2 and tumors
identify an 25% increase in intratumoral docetaxel exposure of CPC634 not amenable to curative therapy. 1° and 2° objectives included determining
relative to Cd. Patients ($18 years) were randomized to receive 75 mg/m2 the recommended phase 2 dose (RP2D), the safety and PK/PD profiles of
CPC634 in cycle 1 and Cd in cycle 2 or vice versa. After drug administration, APX3330 and reporting any RECIST anti-tumor activity. Patients received
patients underwent tumor biopsies during both cycles. Total docetaxel was APX3330 b.i.d, in 21-day cycles. AE evaluation included 1 pt/cohort until the
determined for both drugs and released docetaxel for CPC634 in tumor occurrence of $ G2 toxicity at which time the study proceeded in a 3+3 design.
tissue and in plasma with a validated LC-MS/MS method. PK data were Additional patient were also recruited in cohorts in order to attain PK/PD and
analyzed with mixed model analysis. Results: Sixteen evaluable patients biopsy samples. Results: Between 2/18 and 8/18, 19 subjects (13M, 6F) with
were included. Intratumoral PK revealed a 323% (95% CI: 148,621) higher median age of 69 y started therapy. Dose (mg/d) escalation and number of
total docetaxel (p , 0.001) for CPC634. Released docetaxel for CPC634 patients treated (n) per each cohort proceeded as follows: 240 mg (1), 360 (4),
was comparable to total docetaxel levels for Cd (95% CI: -35-,67) (p = 0.43). 480 (2), 600 (6) and 720 (6). APX3330 was well tolerated at dose levels from
Plasma released docetaxel for CPC634 exhibited an 89% (95% CI: 86, 91) 240-600 mg/d. The most frequent treatment-related adverse event (all grades)
lower (p , 0.001) peak plasma concentration (Cmax) and 81% (95% CI: 46, was G1 fatigue. A G3 rash occurred in two subjects at the 720 mg level defining
125) higher (p , 0.001) area under the curve (AUC) relative to Cd. 600 mg/d as the RP2D for further development. Six subjects had disease
Conclusions: CPC634 resulted in higher intratumoral total docetaxel and stabilization for $ 4 cycles, and of these, four subjects with the following
comparable released docetaxel levels relative to total docetaxel for Cd. diagnosis, RECIST response and days on study included: (CRC, PR, 356),
CPC634 had a favorable plasma PK profile with a lower Cmax and prolonged (Endometrial, SD, 316), (Melanoma, SD, 245), (Prostate, SD, 246). Final
higher systemic exposure relative to Cd. These results indicate that CPC634 PK and PD data, including proteomic, transcriptome, APE1 serum levels
could improve intratumoral docetaxel exposure compared to Cd. Additional and CTC analyses are pending and will be reported at the conference.
studies assessing the intratumoral exposure to CPC634 (NCT0371243) Conclusions: APX3330 is an orally administered inhibitor of APE1. This phase I
and a phase II efficacy study of CPC634 in patients with platinum resistant study identified 600 mg PO daily as the RP2D for further development. RECIST
ovarian cancer (NCT03742713) are currently ongoing. evaluation identified signs of clinical activity in this un-selected population of
patients with advanced cancer. PD analyses indicate APX3330 mediated tar-
geting of the APE1 protein. Clinical trial information: NCT03375086.

3098 Poster Session (Board #90), Sat, 8:00 AM-11:00 AM 3099 Poster Session (Board #91), Sat, 8:00 AM-11:00 AM
A phase I, open label, multicenter dose escalation study of AZD2811 nano- FGFR2: A pan-genomic target. First Author: Russell Madison, Foundation
particle in patients with advanced solid tumors. First Author: Melissa Lynne Medicine, Inc., Cambridge, MA
Johnson, Sarah Cannon Research Institute, Nashville, TN
Background: FGFR2 genomic alterations (GA) have been described in a
Background: Aurora kinase B (AURKB) represents a potential target for variety of solid tumors and emerged as biomarkers for investigational agents
therapy in solid and hematological malignancies. AURKB inhibitor AZD1152 undergoing clinical trials. Methods: 201,766 primarily relapsed/refractory
(barasertib) was previously investigated in solid tumor pts in a phase I setting. malignancies were evaluated with a hybrid-capture based sequencing assay
AZD2811-nanoparticle (np) is a novel, encapsulated slow release AURKB Tumor mutational burden (TMB) was determined on 0.8-1.1 Mbp of se-
inhibitor offering several advantages over AZD1152 (Ashton S et al., Sci Transl quenced DNA and reported as mut/Mb. Microsatellite instability (MSI) was
Med 2016). We report the completed dose-escalation safety, pharmacoki- determined on 114 loci. PD-L1 expression was determined by IHC (Dako
netics (PK), preliminary activity and defined maximum tolerated dose (MTD) 22C3 antibody). Results: FGFR2 GA were detected in 2,993 (1.5%) cases
of AZD2811-np in pts with advanced solid tumors (NCT02579226). featuring short variant (SV) mut (42%), copy number changes (27%),
Methods: Adult pts with advanced solid tumors received AZD2811-np IV on rearrangements/fusions (28%) and multiple GA (3%). The most frequent SV
Day 1 (D1) and 4 (D4) Q4 week (wk) in six cohorts 15-200 mg/infusion without GA were S252W, N549K, C382R, P253R, Y375C, K659E and R664W. A
the use of g-csf in cycle 1. D1 Q4wk and Q3wk schedules were investigated up small cohort (2%) of tumors featured the V564I and V564L GA that are
to 600 mg/infusion (including cohorts with mandatory g-csf prophylaxis on day associated with resistance to TKI drugs. The FGFR2-altered cases were 69%
8). A standard 3+3 design was used. PK was assessed in cycle 1. Results: 50 female/31% male with median age of 61 yrs. Most frequent GA in FGFR2
pts were recruited into 12 cohorts. D1, D4 Q4wk schedule: 24 pts (15, 25,38, altered cancers: TP53 (47%), PIK3CA (22%), PTEN (20%), ARID1A (18%),
50, 100 mg/infusion (n=3/cohort), 200 mg/infusion (n=9)). All cohorts were CDKN2A/2B (18/14%) and MYC (12%). FGFR2 SVs most common in
tolerated. Transient grade 4 neutropenia was observed in 7/9 pts at 200 mg/ endometrial, breast carcinomas (ca) and CUP. FGFR2 amplification most
infusion, including 1 DLT (gr4 . 7 days) D1 Q4wk: 200 mg(n=3) was tol- common in breast, gastroesophageal and lung ca. FGFR rearrangement/
erated. D1 Q3wk: 23 pts were evaluated (200/400 mg (n=3,7), and 400/600/ fusions most common in cholangioca (37%), CUP (15%), pancreatobiliary
500 mg with mandatory g-csf (n=3/5/6)). 400 mg without g-csf was not (12%) and breast ca (6%). The FGFR2-BICC1 was the most frequent fusion
tolerated (1 gr3 mucosal inflammation & 1 gr4 neutropenia . 7 days). 600 mg followed by fusions with TACC2, AHCYL1, CCDC6, VCL, and KIAA1217.
with g-csf was not tolerated (gr3 febrile neutropenia & gr3 fatigue). 25/50 pts MSI-High status present in 6.8% of evaluable FGFR2 altered cases (63% in
experienced AE $gr 3 (21 considered AZD2811-np-related, 19 neutropenia- endometrial ca). Median TMB was 3.5 mut/Mb with 21.8% featuring $ 10
related, no deaths within-DLT period). AZD2811-np caused transient gr1/2 mut/Mb and 12.0% featuring $ 20 mut/Mb. Only 63% of MSI-High FGFR2
fatigue, nausea, diarrhoea and mucosal inflammation. AZD2811 total blood mut tumors had TMB $ 20 mut/Mb. 12.7% FGFR2 mut+ had . 1% PD-L1
PK appears dose proportional with a t1/2 of 30-50 hours irrespective of staining with 3.4% . 50% staining. 29% of PD-L1 IHC+ cases in NSCLC.
schedule. Released AZD2811 concentrations ~1% of total. 14 pts (28%) had FGFR mut ca’s responding to anti-FGFR2 therapies will be presented.
disease stabilisation. 1 prostate ca. pt had a confirmed partial response (PR) Conclusions: FGFR2 GA are most frequent in cholangioca, breast, GI tract,
(continued tx to 451 days). Conclusions: The MTD for AZD2811-np is 500 mg lung ca and CUP, with enrichment of FGFR2 fusions in biliary tract ca. Cases
D1 Q3wk. AZD2811-np is now being investigated in a small cell lung cancer with FGFR2 GA typically do not feature other kinase driver GA and are
expansion. Clinical trial information: NCT02579226. associated with mut in the MTOR/PIK3CA/AKT pathways. Finally, in contrast
with RTK driver GA in EGFR (5.7%) and ERBB2 (7.9%), at 12.0%, across all
tumor types, FGFR2 mut cancers may have higher frequency of TMB $ 20
mut/Mb suggesting potential immunotherapy responsiveness.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Therapeutics and Tumor Biology (Nonimmuno) 173s

3100 Poster Session (Board #92), Sat, 8:00 AM-11:00 AM 3101 Poster Session (Board #93), Sat, 8:00 AM-11:00 AM
A randomized phase 0 trial of the mitochondrial inhibitor ME344 or placebo Patient-derived organoid (PDO), a new personalized therapy selection tool for
added to the antiangiogenic (Aa) bevacizumab in early HER2-negative breast prompt clinical decision making in metastatic gastrointestinal (mGI) cancer
cancer (E-HERNEBC). First Author: Miguel Quintela-Fandino, CNIO-Spanish patients. First Author: Hao Xie, Mayo Clinic, Rochester, MN
National Cancer Research Center, Madrid, Spain
Background: PDO is a promising translational tool that recapitulates the
Background: We have shown that when Aas induce vascular normalization biology and drug response of donor cancer patient. However, an unmet need
(VN), tumors escape upregulating mitochondrial metabolism. Mitochondrial is to have PDO drug-screening data available for treatment decision making
inhibition with ME344 induced synergy with various Aas. We also found that in clinic. We conducted a pilot study to determine whether PDO testing
VN could be traced by showing a 10% decrease in tumor FDG-PET SUV from results will be available at critical treatment decision points in metastatic GI
day (d) 0 to d8 of Aa. We studied the activity of adding ME344 or placebo to cancer patients. Methods: Metastatic GI cancer patients undergoing core-
Bev (Ki67 decrease) in E-HERNEBC in a phase 0 randomized trial. As a needle biopsy were eligible. Tumor cells isolated from #4 fresh biopsy
secondary objective we measured the activity of the combination in patients tissues were grown in a Matrigel-based culture. PDO response to anti-cancer
(Pts) showing VN according to FDG-PET. Methods: Untreated E-HERNEBC Pts drugs were evaluated; and when available, correlated with donors’ clinical
with T . 1cm, any N, M0 underwent a baseline FDG-PET (d1) and received a response to the same agent(s). PDO response was defined as IC50 , 0.1 3
single dose of Bev (15mg/kg) prior to randomization (1:1) to arm A (FDG-PET published Cmax of the drug clinically; stable as IC50 between 0.1 to 10 3
on d8 followed by ME344 10 mg/kg IV on d8, d15 and d21) or Arm B (FDG- Cmax. Radiographic response was per RECIST criteria. Results: We enrolled
PET on d8 followed by placebo on d8, d15 and d21). Tumors were biopsied on 27 refractory metastatic GI cancer patients (9 colorectal [CRC], 9 pancreas,
d0 and 28. A 40 Pts sample size was powered to detect a 30% relative and 9 biliary tract). Median lines of therapy were 4, 2, and 2; the success rate
difference between arms in digitally acquired Ki67 decrease from d0 to d28 of organoid establishment was 89%, 44%, and 55%, respectively. The
(alpha 0.05, beta 0.2). Results: Arm A: 20 Pts; Arm B: 21 Pts. Baseline median time from biopsy to availability of drug-testing data was 64 days
characteristics were in arm A vs B: age 58.4(41.5-75.3) vs 53.6(39-82.8); (range: 24 to 93 days). The median time from biopsy to next CT re-staging in
T1(30%)/T2(60%)/T3(10%) vs T1(52%)/T2(48%)/T3(0%); N0(80%)/ donors was 64 days. The established PDOs shared histological and genomic
N1(20%) vs N0(81%)/N1(19%); ER+(75%)/TNBC(25%) vs ER+(71.4%)/ features with donor clinical tissue. PDO and clinical responses to the same
TNBC(28.6%); Ki67 31.6% (3.6%-70%) vs 25.2% (1.2% - 81.5%). PET- agent(s) were correlated in 2 CRC donors including (1) BRAFV600E-mutated
SUV decreased . 10% from d0 to d8 in 6/20 (arm A) and 6/21 (arm B) Pts. PDO responded to vemurafenib + panitumumab, as did the donor who had
Two G3 adverse events (blood pressure) were reported (1/arm) and deemed partial response (PDO drug-testing data were available 55 days post-biopsy,
related to Bev. Results of the primary endpoint: table. Conclusions: ME344 23 days prior to restaging scan); (2) KRAS/FGF-dual amplified PDO had
showed significant biologic activity, enhancing the effect in Ki67 decrease stable disease status to regorafenib, as did the biopsied lesion from the donor
vs placebo when added to Bev in E-HERNEBC. The activity was greater in (73 days post-biopsy, 5 days post-scan). Conclusions: We showed the
TNBC. A trend for greater activity in patients experiencing VN according to feasibility of completing PDO drug sensitivity testing in metastatic GI cancer
FDG-PET was observed. Clinical trial information: NCT02806817. patients within a short time that could impact clinical decision making,
All Patients Arm A Arm B P value
particularly in CRC. PDO drug response showed correlation with clinical
response. With further refinement, PDO can be a powerful tool for per-
Absolute Ki67 change Day 1 to Day 28 -13.3 -1.1 0.0107
Relative Ki67 change Day 1 to Day 28 -23.9% +186% , 0.01 sonalizing cancer therapy in metastatic GI cancer patients.
Relative Ki67 change D1 to D28 in patients with VN by FDG-PET -33.4% +11.8% 0.09
Relative Ki67 change D1 to D28 in patients without VN by FDG- -18.8% +168% NS
PET
Patients by receptor Arm A - TNBC Arm A – HR+
Absolute Ki67 change Day 1 to Day 28 -39.8 -5.5 , 0.01
Relative Ki67 change Day 1 to Day 28 -73.6% -8.7% , 0.01

3102 Poster Session (Board #94), Sat, 8:00 AM-11:00 AM 3103 Poster Session (Board #95), Sat, 8:00 AM-11:00 AM
Molecular biology and treatment strategies for non-V600 BRAF-mutant The Circulating Cell-free Genome Atlas (CCGA) Study: Size selection of cell-
NSCLC. First Author: Marcelo Vailati Negrao, Department of Thoracic / free DNA (cfDNA) fragments. First Author: Darya Filippova, GRAIL, Inc.,
Head and Neck Medical Oncology - The University of Texas MD Anderson Menlo Park, CA
Cancer Center, Houston, TX
Background: Detection of somatic copy number aberrations in individuals with
Background: BRAF alterations (alts) account for ~4% of non-small cell lung cancer via cfDNA whole-genome sequencing (WGS) is challenging at low
cancers (NSCLC) with 50% being non-V600 alts. Because these alts are tumor fractions. Given that tumor-derived cfDNA fragments are shorter than
functionally heterogeneous and have a poorly characterized genomic those from healthy tissues, this exploratory analysis evaluated the potential
landscape, determining appropriate treatment strategies is a challenge. effect of size selection on the ability to detect cancer. Methods: CCGA WGS
Methods: The Guardant360 clinical database was queried for NSCLC pa- libraries were in silico and in vitro size selected to estimate the change in tumor
tients (pts) with BRAF alts. Alts were categorized by clonality, type and class fraction by tumor types (breast, lung, and colorectal [CRC]) and stage (I-III vs
(1 and 2: BRAF monomer and dimer signaling; 3: requires co-occurring IV). In silico analyses used clinically evaluable training set samples with WGS
upstream RAS-mediated signaling). Functionality and drug screen assays assay results (n = 1422: 560 non-cancer [NC], 862 cancer [C] stages I-IV);
were performed in Ba/F3 cells. Pts with non-V600 mutations were analyzed classification (cancer/non-cancer) performance was estimated using frag-
for sensitivity to MEK +/- BRAF inhibitors (M+Bi). Results: 306 unique ments within the 90-150 bp range. In vitro analyses used a subset of samples
BRAF alts were identified and the majority were observed once (233/306; (n = 93: 28 NC, 65 C stages I-IV), including C cases sampled within a range of
76%). Amplifications (806/1663; 48.5%) and missense alts (795/1663; tumor fractions; tumor fraction was also measured at each progressive removal
47.8%) were the most common occurrences. Missense alts were pre- of maximum-length fragments (intervals of 10 bp: 150 bp down to 50 bp).
dominantly clonal (58%), and of known functionality (428/795; 54%). All Results: In silico and in vitro analyses, respectively, resulted in median
class 1-2 alts were activating in Ba/F3 cells, while class 3 alts were found to 2.0060.58-fold (at 6.9162.64X depth) and 2.0060.52-fold (at 2364.45X
have variable functionality (activating: 4/9). Functionality was correlated depth) increases, in overall tumor fraction (compared to non-size-selected 36X
with clonality as demonstrated by class 1-3 alts having higher clonality depth). This was consistent across tumor types (in silico: 1.7860.73 breast,
compared to variants of unknown significance (VUS) (1: 56%; 2: 54%; 3: 2.0060.58 CRC, 2.0060.41 lung; in vitro: 2.0060.82 breast, 2.5160.52
45%; VUS: 38%; P,0.01). Drug screens for G469V and L597R alts showed CRC, 2.5360.94 lung) and stages (in silico: 2.0060.74 I-III, 1.7860.52 IV;
resistance to first generation BRAF inhibitors (IC50 $100nM), but sensi- in vitro: 2.0060.55 I-III, 1.6860.29 IV). Tumor fraction increased with initial
tivity to M+Bi (IC50 0.02-36nM). Growth inhibition was more pronounced fragment length titrations, but not following size selection to shorter lengths
for dabrafenib + trametinib (D+T) (IC50 ,0.1nM) compared to encorafenib + ( , 140 bp). Classifier trained on in silico size-selected data had increased
binimetinib (IC50 8-35nM) and vemurafenib + cobimetinib (IC50 2-36nM). sensitivity at 98% specificity compared to those trained on non-size-selected
BRAF D594G mutation (class 3) was not activating in Ba/F3 cells. Three pts data (p , 1e-5). Conclusions: In silico and in vitro size selection consistently
with non-V600 alts were treated with M+Bi. G469V and D594G had rapid increased tumor fraction across cancer types and stages, and this increase was
disease progression (PFS 2 and 4 mos respectively), while pt with L597R has maximized by tuning the length range of size selection. Relative to full-depth
ongoing partial response (PFS 8+ mos). Conclusions: BRAF alts show corre- data, classification performance improved significantly. These data suggest
lation between clonality and functionality, which provides important clinical that size selection targeting cfDNA under 140 bp may enhance cfDNA-based
information given the numerous VUS in the BRAF non-V600 setting. Drug cancer detection. Clinical trial information: NCT02889978.
screens reveal non-V600 alts may be sensitive to M+Bi and suggest D+T is the
most active combination. Clinical data supports that some non-V600 BRAF
mutations may be sensitive to M+Bi.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
174s Developmental Therapeutics and Tumor Biology (Nonimmuno)

3104 Poster Session (Board #96), Sat, 8:00 AM-11:00 AM 3105 Poster Session (Board #97), Sat, 8:00 AM-11:00 AM
First-in-human phase I trial of anti-hepatocyte growth factor (HGF) antibody Targeting the p300/CBP interactome through blockage of the CH1-domain
(YYB101) in refractory solid tumor patients: Integrative pathologic-genomic triggers tumor regression in AR-positive and AR-negative xenograft models of
analysis and the final results. First Author: Jeeyun Lee, Samsung Medical castration-resistant prostate cancer (CRPC). First Author: Valentino Cattori,
Center, Sungkyunkwan University School of Medicine, Seoul, South Korea Inthera Bioscience AG, Waedenswil, Switzerland
Background: It has been demonstrated in vivo that HGF-MET signaling axis is a Background: In advanced prostate cancer, the paralog transcription co-
key molecular determinant in tumor invasion and there is a significant as- activators p300/CBP are often highly expressed and have been associated
sociation in HGF expression and mesenchymal phenotype in addition to im- with disease progression and poor prognosis. While several inhibitors of the
mune cell recruitment. We have developed a HGF neutralizing humanized bromo- and histone acetyltransferase domains of p300/CBP have been de-
monoclonal antibody antibody, YYB-101. The aim of this study was to de- scribed, past efforts to develop drug-like ligands of other regions of this at-
termine the maximum tolerate dose (MTD), safety, pharmacokinetics (PK), and tractive target have been unsuccessful. Methods: A rationally designed small
pharmacodynamics (PD) of YYB101, in patients with refractory solid tumors. molecule modulator of the CH1-domain of p300/CBP was tested in a panel of
Methods: YYB101 was administered intravenously at once every 2 weeks prostate cancer cell lines, followed by cell cycle analysis and beta-galactosidase
doses of 0.3, 1, 3, 5, 10, 20, 30 mg/kg, according to a 3+3 dose escalation staining. Inhibition of the p300-dependent androgen receptor (AR) related
design. Enrolled patients were planned to receive YYB101 until disease transcriptional response was determined in a luciferase reporter assay and by
progression or intolerable toxicity. The escalation and expansion cohorts qPCR analysis of expression of downstream genes like prostate-specific antigen
(20mg/kg) were completed. Pre-planned biomarker analysis was performed in (PSA), transmembrane protease-serine 2 (TMPRSS2) and prostein (SLC45A3).
parallel. Results: 39 heavily pre-treated refractory cancer patients were en- In vivo effects were evaluated in cell line- and patient-derived xenograft
rolled and received YYB101. No DLT was observed. YYB101 demonstrated models of CRPC. Results: Selective blockage of the CH1 domain of p300/
dose proportional PK up to the dose of 30 mg/kg. No patients discontinued CBP results in sustainable anti-proliferative effects in AR-positive and AR-
treatment because of adverse events. Based on PK analysis and toxicity data, negative prostate cancer cells inducing apoptosis and/or senescence.
the recommended dose was determined as 20 mg/kg. Of 39 evaluation pa- Transcriptome and gene expression analyses revealed the downregulation
tients, there was 1 confirmed partial response for . +14months (2.5%, N = 1; of various drivers of cell cycle progression as well as decreased expression
1 (of 2) sebaceous carcinoma) and 17 stable disease as best response (43.5%, of hormone-induced, AR-regulated genes. In enzalutamide-resistant xe-
N = 17; 7 (of 13) CRC, 3 (of 4) melanoma, 1 (of 2) sebaceous carcinoma, 1 (of nograft models of CRPC, oral administration of the compound triggered
3) gastric, 1 (of 1) basal cell carcinoma, 2 (of 10) ovarian cancer, 1 (of 1) HCC, tumor regression/eradication at well tolerated doses. Serum PSA levels
1 (of 1) lung cancer). Of note, 1 sebaceous carcinoma patient who have failed were strongly decreased in treated animals. Conclusions: Simultaneous
to $2+ lines of chemotherapy, have been responding to YYB for 14 months. inhibition of both, AR-signaling and downregulation of p300/CBP activity,
The MET and HGF expressions by immunohistochemistry (IHC) were evaluated may cause profound and long lasting antitumoral effects in patients with
in 19 and 17 tumor specimens, respectively. Neither protein expressions were advanced prostate cancer. Future clinical investigation of this novel oral
significant predictors for treatment response to anti-HGF antibody. However, small molecule agent is warranted.
we have observed significant reduction in HGF in responders to YYB. Two long-
term responders had mesenchymal signature in RNA sequencing.
Conclusions: YYB101 has a favorable safety profile in patients with refractory
solid tumors and a dose-proportional PK. Efficacy data are encouraging and
phase II combination therapy with YYB101 is planned to be open in metastatic
CRC patients as salvage treatment. The predictive power of mesenchymal
signature in YYB responders will be defined prospectively. Clinical trial in-
formation: 02499224.

3106 Poster Session (Board #98), Sat, 8:00 AM-11:00 AM 3107 Poster Session (Board #99), Sat, 8:00 AM-11:00 AM
The landscape of RET alterations from 56,970 adult patients with cancer: Host transcriptomic signatures associated with dysbiosis in a preclinical
Clinical implications. First Author: Alexander Andreev-Drakhlin, MDA, model of lung cancer. First Author: Mariam El-Ashmawy, NYU Langone
Houston, TX Medical Center, New York, NY
Background: Activating receptor-tyrosine kinase rearranged during trans- Background: Enrichment of the lower airway microbiota with oral com-
fection (RET) mutations and fusions have been recognized as potent drivers mensals has been associated with transcriptomic changes affecting several
of oncogenesis. Recent identification of highly potent and selective RET inflammatory pathways associated with non-small cell lung cancer (NSCLC)
inhibitors holds great promise in the management of RET-dependent tumors. development and progression. Using a mouse model of NSCLC, we evaluated
Here we present a comprehensive analysis of RET alterations in pan-cancer the effects of lower airway dysbiosis on tumor progression and host tran-
adult malignancies. Methods: We analyzed 59,347 samples from 56,970 scriptomics. Methods: Preclinical model of lung cancer was constructed by
patients available from AACR Project GENIE (Cancer Discov. 2017) data- introducing luminescence-tagged Kras mutated cells into C57/B6 mice,
base for the prevalence of RET fusions, mutations, and copy number al- causing lung cancer to develop. Lower airway dysbiosis was induced by
terations in diverse cancer types. Results: A total of 1414 RET alterations weekly intratracheal challenge with either PBS or Veillonella parvula in wild
were detected, including 91 fusions (6.4%), 1166 missense mutations type and lung cancer mice. Experiments were repeated twice to evaluate for
(82.5%), 136 truncating mutations (9.6%), and 21 in-frame mutations survival as well as lower airway host response using flow cytometry and RNA
(1.5%). RET fusions were observed in 0.15% of tumor samples and were sequencing (HiSeq). Sequence data was processed using a validated mouse
most commonly identified in non-small cell lung cancer, thyroid cancer, gene expression signature matrix with cibersort from https://cibersort.
colorectal cancer, prostate cancer, and gastric cancer (62.6%, 18.6%, stanford.edu and DESeq using FDR correction. Results: In wild type mice,
5.5%, 4.4%, 3.3% of identified RET fusions, respectively). RET fusions lower airway dysbiosis with Veillonella did not affect the survival, weight gain
were significantly co-altered with MAPK3/ERK1 (p=0.045), SETD2 or airway lumen diameter. Among lung cancer mice, dysbiosis led to in-
(p=1.36E-07), and EIF4E (p=0.045), while there was a negative association creased mortality, weight loss, and tumor burden. Multiple transcriptomic
between RET fusions and EGFR (p=0.009634), TP53 (p=0.02267), and signatures were identified among the dysbiosis groups (both in WT and lung
KRAS (p=2.53E-05) alterations. Most common RET gene upstream partners cancer mice). Unsupervised hierarchical clustering of immune cell profiles
were KIF5B, CCDC6, and NCOA4 (42.9%, 24.2%, 7.7% of identified RET using cibersort on whole transcriptome showed near perfect separation
fusions, respectively). RET missense mutations were found in 2.0% of tumor between the four experimental conditions. Amongst the most differentially
samples; 136 (11.7%) of identified missense mutations, including 8 RET enriched immune cell subsets, we identified that lung dysbiosis upregulates
gatekeeper V804M/L mutations, were characterized as likely oncogenic, 12 genes annotated to Th1 and Th2 cells (p , 0.01, q , 0.2). Using flow
(1.0%) as likely benign, and 1018 (87.3%) as variants of unknown sig- cytometry, we identified that PD-1, IL-17, and ROR-gamma are differentially
nificance using OncoKB database. RET amplifications occurred in 1.5% of expressed in CD4+ cells in dysbiosis conditions, and these patterns are
tested samples. Conclusions: While RET fusions represent extremely rare consistent in whole RNA transcriptome. Conclusions: Transcriptomic sig-
events in multiple cancers, RET missense mutations occur in 2% of ma- natures reveal immune profiles associated with dysbiosis, an experimental
lignancies. Most RET missense variants are described as variants of unknown condition associated with worse outcomes in lung cancer. This investigation
significance, limiting the impact of precision oncology for the majority of provides novel insights into how disruption of the lower airway microbiome
patients with RET alterations. Further functional characterization of RET may contribute to the pathogenesis of NSCLC.
variants is warranted. MAPK pathway co-alterations in patents with RET
fusions may present a strategy for future therapeutic combinations.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Therapeutics and Tumor Biology (Nonimmuno) 175s

3108 Poster Session (Board #100), Sat, 8:00 AM-11:00 AM 3109 Poster Session (Board #101), Sat, 8:00 AM-11:00 AM
Prediction of olaparib sensitivity for variants of unknown significance in Cell-specific upregulation of lung “cancer signature genes” in the small
homologous repair genes. First Author: Sandy Chevrier, Research Platform in airway epithelium of asymptomatic smokers. First Author: Mahboubeh
Biological Oncology, Center GF Leclerc, Dijon, France Rostami, Weill Cornell Medical College, New York, NY
Background: the recent use of PARP inhibitors in clinical practice gives Background: Most lung cancers are derived from the small airway (6th -23rd
very interesting outcome for ovary tumors with BRCA1 or BRCA2 mutation generations) epithelium (SAE). While a pathogenic mutation in a driver gene
but also in other tumors with homologous repair deficiency. Nevertheless, is critical to transform a SAE cell to a malignant cell, single cell analysis of
no hotspot mutations are present, consequently, more than 85% of ob- lung cancers has demonstrated that individual cancer cells have transcrip-
served variants have unknown significance, blocking the use of PARP tional alterations in many “cancer signature genes” (Lambrechts D et al,
inhibitor. Methods: Exome analysis was performed on a cohort of 27 Nature Med 2018; 24:1277) that are not driver genes, but likely support the
patients treated with olaparib. After bioinformatics analyses, variant in- malignant state. Based on the concept that dysregulation of many of these
terpretation was performed by interrogating different databases. For genes facilitate the malignant state, we hypothesized that, with the stress of
variants of unknown significance (VUS), PROVEAN software and allelic smoking, some of these cancer-supporting transcriptional modifications occur
frequency normalized with tumor cellular content were used to classify long before the random hit with a driver mutation, providing a soil for the driver
VUS as potentially benign or potentially deleterious. Results: Among the mutation. Methods: To assess this hypothesis, we applied drop-seq single cell
27 patients analyzed, 16 harbored already classified variants (3 benign transcriptome analysis to assess SAE recovered by fiberoptic bronchoscopy
and 13 pathogen variants) and 11 had VUS. The first Progression Free and brushing of n = 3 healthy non-smokers and n = 3 healthy smokers. Using
Survival (PFS) analysis showed that benign variants did not respond to unsupervised clustering, 11 cell types were identified including major SAE
olaparib with a median survival of 62 days, whereas pathogenic variants cell types (basal, intermediate, club, mucous and ciliated), rare epithelial
had a median of 109 days. Surprisingly, VUS had a median of 136 days, cells (ionocyte, neuroendocrine and undefined NCLhi) and rare immune/
suggesting that some of them could be classified as potentially delete- inflammatory cell (Tcell, mast, antigen presenting). Results: Comparison of
rious. On the subset of 11 patients with VUS, we applied PROVEAN smokers and nonsmokers showed that smoking significantly altered the
prediction classifying 5 variants as benign and 6 variants as deleterious, transcriptome (n = 426 genes upregulated, n = 572 genes downregulated) in
with a median PFS of 54 days and 140 days (p=0.3235), respectively. different SAE cell types; 21% -56% of smoking-upregulated genes in the
With the second prediction, based on variant allelic frequency, we ob- major SAE cell population are identified as “cancer signature genes”, where a
tained PFS of 73.5 months for benign variants and 210 days for dele- number have cell-specific smoking regulated patterns within individual cell
terious ones (p=0.29). By combining both predictions, we classified as types including HSPB1 in mucous cells, ADH7 in ciliated, LAMB3 in basal and
benign, VUS predicted benign with both predictions, and as deleterious, intermediate cells, MIF intermediate and club cells, ALDH3A1 in all main SAE
VUS predicted as deleterious with at least one prediction. Consequently, cell types. Conclusions: These data support the concept that cigarette
we perfectly discriminated benign from deleterious variants with a median smoking, long before the development of cancer, reprograms specific SAE cells
PFS of 36 days for predicted benign and 177 days for predicted delete- to provide the biological soil to support driver genes to function.
rious (p=0.0084). From all patients, PFS were significantly different
(p=0.0003) between benign (n=6, 56 days) and deleterious variants
(n=21, 140 months). Conclusions: Our work tends to show that VUS of
homologous repair genes could be predicted as benign or deleterious, and
could increase the number of patients eligible for a treatment by PARP
inhibitors. The number of patients needs to be increased in order to
validate our prediction algorithm.

3110 Poster Session (Board #102), Sat, 8:00 AM-11:00 AM 3111 Poster Session (Board #103), Sat, 8:00 AM-11:00 AM
Clinical, pathological and genetic predictors of patient-derived xenograft Integrative analyses of signaling and DNA damage repair pathways in
(PDX) engraftment in EGFR-mutated lung adenocarcinoma (LUAD). First patient-derived xenograft (PDX) models from NCI’s patient-derived models
Author: Sebastiao N. Martins-Filho, University Health Network, University of repository (PDMR). First Author: Biswajit Das, Molecular Characterization
Toronto, Toronto, ON, Canada Laboratory, Frederick National Laboratory for Cancer Research, Leidos
Biomedical Research, Inc., Frederick, MD
Background: PDX are useful preclinical models to study drug response and
resistance. Different specimen types have been used to generate PDX Background: Patient-derived xenografts (PDXs) are increasingly being used
models including histological (surgery and CT-guided biopsy) and cytological in translational cancer research for preclinical drug efficacy studies. The
preparations (EBUS and pleural effusions). We hypothesize that engraftment National Cancer Institute (NCI) has developed a Patient-Derived Models
is not stochastic and is affected by many factors including sample type and Repository (NCI PDMR; pdmr.cancer.gov) of PDXs with clinical annotation,
tumor pathological and molecular properties. To improve sample selection proteomics, and comprehensive genomic datasets to facilitate these studies.
and cost-effectiveness of PDX experiments, we investigated clinical, his- Here, we present an integrative genomic, transcriptomic, and proteomic
tological and genetic correlates of engraftment in EGFR-mutated LUAD. analysis of critical signaling and DNA damage repair pathways in these PDX
Methods: We assessed PDX engraftment from 96 surgical resections, 13 CT- models, which represent 9 common and multiple rare tumor histologies.
guided biopsies, 21 EBUS and 14 pleural effusions of EGFR-mutated LUAD. Methods: 304 PDX models from 294 patients were established from various
Sixty-five samples, including 6 engrafted (XG) and 54 non-engrafted (noXG) solid tumor histologies from patients with primary or metastatic cancer.
were evaluated by exome sequencing. Results: Engraftment was successful Whole Exome Sequencing, RNA-Seq and Reverse Phase Protein Array
in 9/96 (9%) surgical resections, 6/13 (46%) CT-guided biopsies, and 0/35 (RPPA) were performed on 2-9 PDXs per model across multiple passages. An
cytological samples. Biopsies taken at time of treatment failure (compared to integrative workflow was applied on multiple data sets to detect pathway
treatment naive biopsies) correlated with greater engraftment (p=0.007, activation. Results: We profiled 10 signaling and 5 DNA repair pathways in
AUC = 0.68). Multivariable regression analysis of clinical variables at the the PDMR dataset. We observed that: (i) a large fraction (40%) of PDX
time of sampling identified advanced (vs early) stage (p = 0.003) and models have at least 1 targetable mutation in the RTK/RAS and/or PIK3CA
histological (vs cytological) preparations (p , 0.001) as the strongest pathways; (ii) 131 models (45%) have putative driver and oncogenic mu-
predictors of engraftment (AUC = 0.79). Among tumor histologic features, tations and copy number variants (CNVs) in the WNT, TGFRb , NRF2 and
solid (vs lepidic, acinar and papillary) pattern was associated with greater NOTCH pathways. In addition, 17% of PDX models have targetable muta-
engraftment (p , 0.001). Presence of EGFR-T790M (p = 0.004) and tions in DNA damage repair pathways and 20 PDMR models have a DNA
TP53 (p = 0.009) mutations were associated with greater engraftment; all mismatch repair defect (MSI-H). We confirmed activation of the signaling
XG samples carried TP53 mutations. EGFR-Ex19del (p = 0.076) showed a pathways in a subset of PDX models by pathway enrichment analysis on gene
trend towards engraftment whereas EGFR-L858R (p = 0.086) trended expression data from RNASeq and phosphoprotein-specific antibody
towards non-engraftment. Conclusions: Advanced stage, post-therapy binding data from RPPA. Activation of DNA repair processes was confirmed
tumors, T790M+ and TP53+ EGFR-mutated LUAD samples obtained by enrichment of relevant mutational signatures and loss of heterozygosity in
for histological processing are more likely to engraft as PDXs. Despite low these PDX models. Conclusions: Genomic analysis of NCI PDMR models
engraftment rates, these models are useful to study novel therapeutic revealed that a large fraction have clinically relevant somatic alterations in
strategy and elucidation of resistance mechanisms. key signaling and DNA damage repair pathways. Further integrative analyses
with matched transcriptomic and proteomic profiles confirmed pathway
activation in a subset of these models, which may prioritize them for pre-
clinical drug studies.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
176s Developmental Therapeutics and Tumor Biology (Nonimmuno)

3112 Poster Session (Board #104), Sat, 8:00 AM-11:00 AM 3113 Poster Session (Board #105), Sat, 8:00 AM-11:00 AM
Genomic somatic alterations of human epidermal growth factor-2 (HER2) A randomized, multicenter clinical trial to determine the efficacy and safety
gene: A pan-cancer analysis. First Author: Xinhua Zhu, Nanjing Drum Tower of pegfilgrastim (GEMA BIOTECH) compared to pegfilgrastim (Roche) for
Hospital,The Affiliated Hospital of Nanjing University Medical School, prevention of chemotherapy induced neutropenia in patients with breast
Nanjing, China cancer. First Author: Martin Eduardo Richardet, Sanatorio Acongacua,
Cordoba, Argentina
Background: Human epidermal growth receptor 2 (HER2) is a well-known
oncogenic drive gene with multiple targeted therapeutic options. In this study, Background: Peg-Neutropine, GEMA BIOTECH SAU biosimilar Peg-Filgrastim, is
we aim to assess the landscape of HER2 alterations in solid tumors and evaluate the first Peg-Filgrastim approved in LATAM for prevention of febrile neutropenia in
the feasibility of circulating tumor DNA (ctDNA) tested by next-generation patients treated with myelosuppressive chemotherapy. Methods: Study pop-
sequence (NGS) as a tool to detect HER2 alterations. Methods: Alterations of ulation: women with stage 2, 3 or 4 of breast cancer scheduled to receive 4 or 6
HER2 by NGS (Illumina NextSeq 500) were queried in 3D Medicines database. cycles of chemotherapy (with Taxane) at 3 weeks interval. Stratification was
The mean depth of tissue and circulating tumor DNA (ctDNA) test was 500X based on breast cancer stage. Study drug was administered subcutaneously
and 5000X, respectively. 11,013 patients tested using tumor tissue and 6,970 in a 6 mg dose. The study was blind to the assessors. The primary endpoint was
patients tested using ctDNA were included in this analysis. Results: Of 11,013 Duration of Severe Neutropenia (DSN, Absolute Neutrophil Count-ANC , 500/
patients tested using tumor tissue, any HER2 and known or likely deleterious mm3) in the first cycle of chemotherapy. Secondary endpoints were incidence
HER2 mutations were identified in 739 (6.7%) and 531 (4.8%) patients, of severe neutropenia (SN), other efficacy measures, and incidence of ADRs.
respectively. Of 531 patients who carried known or likely deleterious HER2 The non-inferiority margin for DSN was estimated in less than 1 day. Results: A
mutations, 263 (49.5%) had HER2 amplification and 259 (48.8%) had single total of 120 subjects were randomized 1:1, 58 were treated with Peg-
nucleotide variations (SNVs). Across all tumor types, breast cancer was found to Neutropine and 62 with Peg-Filgrastim (Roche). Efficacy: SN was de-
have the highest frequency of HER2 amplification (14.9%, 48/323), followed veloped in 52/283 (18,4%) cycles with Peg-Neutropine in 27 patients and 48/
by gastric cancer (6.6%, 31/470) and biliary tract cancer (5.8%, 33/571). 297 (16,2%) cycles with Peg-Filgrastim (Roche) in 20 patients (p=0,4836).
Moreover, 11% (8/73) of duodenal cancer, 4.5% (7/154) of urothelial cancer, In the first cycle, 16 patients with Peg-Neutropine and 11 patients with Peg-
3.8% (18/470) of gastric cancer, 3.1% (142/4555) of lung cancer, 2.9% (17/ Filgrastim (Roche) developed SN. In per protocol analysis mean DNS in the
571) of biliary tract cancer, 2.8% (44/1562) of colorectal cancer and 2.7% (9/ first cycle was 0,78 6 1,53 days for Peg-Neutropine group and 0,5361,25 for
323) of breast cancer carried known or likely deleterious HER2 SNVs. Of 6970 Peg-Filgrastim (Roche) group (95% IC for the difference -0,26; 0,76). Per ITT
patients tested using ctDNA, any HER2 and known or likely deleterious HER2 analysis the mean DSN was 0,9061,79 for Peg- Neutropine group and
mutations were identified in 592 (8.5%) and 277 (4.0%) patients, respectively. 0,5061,21 for Peg-Filgrastim (Roche) group, (95% IC for the difference
In the ctDNA cohort, 15.7% (36/230) of breast cancer and 3.1% (5/161) of -0,15; 0,95). For all the efficacy secondary endpoints the differences were not
biliary tract cancer carried HER2 amplification. However, 11.6% (20/173) of statistically significant. Safety: 7 ADRs were developed by 3 subjects with Peg-
gastric cancer had HER2 amplification tested by ctDNA which was higher than Neutropine and 31 ADRs were developed by 10 subjects with Peg-Filgrastim
that tested using tissue. Furthermore, 5.6% (13/230) of breast cancer, 4.5% (Roche). The most common reaction was myalgia, and other ADRs were ar-
(2/44) of urothelial cancer, 3.4% (6/173), 2.5% of biliary tract cancer and thralgia, asthenia, bone pain and acid sensitive syndrome. Conclusions: Based
2.0% (94/4586) lung cancer harbored known or likely deleterious HER2 SNVs on the non-inferiority margin established we conclude that Peg-Neutropine is
in ctDNA cohort. Conclusions: HER2 alterations existed across tumor types and biosimilar to Peg-Filgrastim (Roche). Clinical trial information: NCT03404752.
the landscape of genomic alterations in HER2 gene varied according to different
type of tumor. In addition, ctDNA can be used as a potential tool to detect HER2
alterations.

3114 Poster Session (Board #106), Sat, 8:00 AM-11:00 AM 3115 Poster Session (Board #107), Sat, 8:00 AM-11:00 AM
Development and clinical validation of Lantern Pharma’s AI engine: A modeling and simulation study of less frequent dosing of nivolumab
Response algorithm for drug positioning and rescue (RADR). First Author: 480 mg. First Author: Cody J. Peer, National Cancer Institute, Bethesda, MD
Umesh Kathad, Lantern Pharma, Dallas, TX
Background: Nivolumab was originally approved at 3 mg/kg q2w. However,
Background: The Response Algorithm for Drug positioning and Rescue there is abundant evidence that doses as low as 0.1 mg/kg q2w are effective,
(RADR) technology is Lantern Pharma’s proprietary Artificial Intelligence and a randomized trial in RCC demonstrated equivalence across a dose range
(Al)-based machine learning approach for biomarker identification and of 0.3-10 mg/kg q3w. Modeling and simulation have been used to amend the
patient stratification. RADR is a combination of three automated modules labeled dosage to 240 mg q2w or 480 mg q4w, with the latter yielding an
working sequentially to generate drug- and tumor type-specific gene sig- estimated steady-state trough concentration (Ctrough) of ~50 ug/mL. Given
natures predictive of response. Methods: RADR integrates genomics, drug the high cost of nivolumab and the lack of a dose-response relationship, we
sensitivity and systems biology inputs with supervised machine learning hypothesized that less frequent dosing of 480 mg would maintain thera-
strategies and generates gene expression-based responder/ non-responder peutic serum concentrations. The objective of this study was to use modeling
profiles for specific tumor indications with high accuracy, in addition to and simulation to develop alternative dosing strategies. Methods: A simu-
identification of new correlations of genetic biomarkers with drug activity. lation model was built from a published population pharmacokinetic model,
Pre-treatment patient gene expression profiles along with corresponding incorporating time-dependent clearance. Various alternative dosing schedules
treatment outcomes were used as algorithm inputs. Model training was were simulated, beginning with the third dose (doses 1 and 2 were 480 mg at
typically performed using an initial set of genes derived from cancer cell line wk 1 and 5). We conservatively chose 4.5 mg/mL as the target concentration
data when available, and further applied to patient data for model tuning, (TC), slightly above the mean simulated Ctrough at 0.3 mg/kg q3w (4.1 ug/mL),
cross-validation and final gene signature development. Model testing and although even lower levels are likely efficacious. The simulated dose schedules
performance computation were carried out on patient records held out as were q8w, q10w, q12w and q14w, beginning with the third dose. Simulations
blinded datasets. Response prediction accuracy and sensitivity were among were performed on 50 simulated patients, with each simulation replicated 5
the model performance metrics calculated. Results: On average, RADR times. Results: The simulated Ctrough following doses 2-4 are presented in the
achieved a response prediction accuracy of 80% during clinical validation. table below. Dosing q12w should maintain TC in . 70% of patients, and q14w
We present retrospective analyses performed as part of RADR validation dosing should achieve TC in . 55% of patients. Conclusions: Modeling and
using more than 10 independent datasets of patients from selected cancer simulation provide evidence that nivolumab can be effectively dosed q8-14w
types treated with approved drugs including chemotherapy, targeted therapy (after the first 2 doses), resulting in a potential 70% cost savings. As
and immunotherapy agents. For an instance, the application of the RADR responding patients generally have a 35-45% decrease in clearance over the
program to a Paclitaxel trial in breast cancer patients could have potentially first 6 months of treatment, even less frequent dosing may be required for
reduced the number of patients in the treatment arm from 92 unselected subsequent doses. Randomized trials of this interventional pharmacoeconomic
patients to 24 biomarker-selected patients to produce the same number of strategy are indicated. Similar opportunities may exist for other checkpoint
responders. Also, we cite published evidence correlating genes from RADR inhibitors.
derived biomarkers with increased Paclitaxel sensitivity in breast cancer.
Conclusions: The value of RADR platform architecture is derived from its Dose q8w q10w q12w q14w
validation through the analysis of about ~17 million oncology-specific 2 19.9 (18.2 – 21.7) 13.4 (12.0 – 14.9) 9.0 (7.9 – 10.3) 6.1 (5.4 – 6.9)
clinical data points, and ~1000 patient records. By implementing unique 3 18.5 (16.8 – 20.3) 11.6 (10.4 – 12.9) 7.4 (6.5 – 8.4) 4.8 (4.2 – 5.4)
biological, statistical and machine learning workflows, Lantern Pharma’s 4 18.3 (16.6 – 20.1) 11.3 (10.1 – 12.7) 7.2 (6.3 – 8.2) 4.7 (4.0 – 5.4)
RADR technology is capable of deriving robust biomarker panels for pre- *values represent geo mean (95% CI)
selecting true responders for recruitment into clinical trials which may
improve the success rate of oncology drug approvals.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Therapeutics and Tumor Biology (Nonimmuno) 177s

3116 Poster Session (Board #108), Sat, 8:00 AM-11:00 AM 3117 Poster Session (Board #109), Sat, 8:00 AM-11:00 AM
Correlation between NDRG1 gene polymorphism and neuropathy (N) in Concomitant intake of abiraterone and food to increase pharmacokinetic
metastatic breast cancer (MBC) patients (pts) enrolled in the PAINTER study exposure: Real-life data from a therapeutic drug monitoring program. First
(Polymorphism And INcidence of Toxicity in ERibulin treatment). First Author: Author: Stefanie L. Groenland, The Netherlands Cancer Institute - Antoni van
Nicla Maria La Verde, Department of Oncology ASST Fatebenefratelli Sacco Leeuwenhoek, Amsterdam, Netherlands
P.O. Fatebenefratelli, Milan, Italy
Background: Abiraterone acetate is registered for the treatment of metastatic
Background: MBC is an incurable disease and therefore treatment focuses castration resistant prostate cancer. Pharmacokinetic (PK) exposure has
mainly on prolonging pts survival and improving quality of life. Eribulin (E) is a been linked to efficacy, since patients with Cmin $ 8.4 ng/mL have a sig-
microtubule inhibitor that increased overall survival in pretreated pts. E pe- nificantly longer progression free survival compared to patients with a Cmin
ripheral N is reported in 13.9-35% of cases. PAINTER main objective was to below this threshold (7.4 vs. 12.2 months, p = 0.044) (Carton, 2017). At the
survey tolerability of E in real life in MBC, while secondary endpoints were to recommended fixed dose of 1000 mg QD administered in a modified fasting
investigate the relationships between specific genetic polymorphisms and in- state, 35% of patients do not reach this efficacy threshold (Carton, 2017),
cidence and severity of peripheral N. Methods: This is a multicenter, inter- providing a strong rationale for therapeutic drug monitoring (TDM). Since a
ventional, single-arm, phase IV study, that enrolled pts who received E after clinically relevant food effect has been established, concomitant intake of
taxanes and antracyclines (dose 1.4 mg/m2 day 1, 8 every 21 days). PAINTER abiraterone and food could offer a cost-neutral solution in case of low ex-
study follow-up is still ongoing. Genomic DNA was isolated from whole blood posure (Chi, 2015). This study aims to evaluate whether PK-guided abir-
samples (Maxwell whole blood DNA kit. Promega). 15 SNPs (Single Nucleotide aterone dosing is feasible and results in an increased proportion of patients
Polymorphisms) were genotyped by Taqman specific assays. For SNPs analysis, with concentrations above the target. Methods: Patients starting regular
we selected pts with avaliable clinical data and who completed E treatment. N treatment with abiraterone were included. PK sampling occurred 4, 8 and
was evaluated by medical examination. The associations between peripheral N 12 weeks after start of treatment, and every 12 weeks thereafter. Abiraterone
(any grade) and the selected polymorphisms were evaluated with Fisher exact concentrations were measured and Cmin was calculated. In case of Cmin
test. Results: From May 2014 to June 2018, 180 pts were enrolled in the , 8.4 ng/mL and acceptable toxicity, a PK-guided intervention was advised.
PAINTER study from 20 Italian hospitals and 135 were analysed for the present As a first step, concomitant intake of abiraterone and a light meal or a snack
report. Pts and tumor characteristics were as follow: median age 62 years (31- was advised. Results: In total, 35 patients were included, of which 18
85), ductal carcinoma 78.5%, visceral disease 70.4%, luminal type 62.6%, patients (51%) had at least one Cmin , 8.4 ng/mL. These patients were
Her2 positive 20.3%, triple negative 17.1%, previous median treatment lines advised to take abiraterone concomitantly with food, after which Cmin in-
for MBC 5 (0-18), previous N reported in 17.8% of pts (sensory 87.5%, motor creased significantly from 5.6 (47%) ng/mL [mean (CV%)] to 40.6 (110%)
12.5%). N (all grades) were reported in 33.4% of patients (G3-G4: 3%). Among ng/mL (p = 0.006) without additional toxicities. This intervention led to
the selected SNPs, one allelic variant (rs2233335 G/G versus G/T or T/T) in adequate exposure in 15 patients (83%). Seventeen patients had all Cmin
NDRG1 gene had a statistically significant association with N (p 0.0010). levels $ 8.4 ng/mL, in these patients mean Cmin was 31.5 (65%) ng/mL.
Conclusions: The data reported demonstrate for the first time that the allelic Conclusions: TDM of abiraterone was applied in clinical practice and proved
variant rs2233335 (G/T and T/T) in NDRG1 gene correlates with E induced N. to be feasible. Concomitant intake with food resulted into a significant
These data, if corroborated, will allow a tailored treatment with E. Clinical trial increase in Cmin and offers a cost-neutral opportunity to optimize treatment
information: NCT02864030. for patients with low PK exposure. Up to 100 patients will be included to
evaluate the effect of PK-guided abiraterone dosing on treatment efficacy.
rs2233335 G1-2-3 Neurotoxicity G0 G1 G2 G3
Allelic variant n (%) n (%) n (%) n (%) n (%) Clinical trial information: NL6695.
G/G 4/24 (16.7) 20 (22.2) 1 (3.7) 1 (7.1) 2 (50.0)
G/T 14/60 (23.3) 46 (51.1) 10 (37.0) 3 (21.4) 1 (25.0)
T/T 27/51 (52.94) 24 (26.7) 16 (59.3) 10 (71.4) 1 (25.0)

3118 Poster Session (Board #110), Sat, 8:00 AM-11:00 AM 3119 Poster Session (Board #111), Sat, 8:00 AM-11:00 AM
A phase I dose-escalation study of two cycles carboplatin-olaparib followed Boosting pazopanib exposure by splitting intake moments: A prospective
by olaparib monotherapy in patients with advanced cancer. First Author: pharmacokinetic study in cancer patients. First Author: Stefanie L.
Jill J.J. Geenen, Netherlands Cancer Institute, Amsterdam, Netherlands Groenland, The Netherlands Cancer Institute - Antoni van Leeuwenhoek,
Amsterdam, Netherlands
Background: The PARP-inhibitor olaparib has single-agent activity in BRCA
mutated breast and ovarian cancer. Preclinical studies show synergistic Background: Pazopanib is approved for the treatment of renal cell carcinoma
effects when combining PARP-inhibitors and platinum drugs in BRCA1/2 (RCC) and soft tissue sarcoma (STS). Due to high (40-70%) interpatient
mutated cancer cell models. A formulation change from olaparib capsules to variability in pharmacokinetic (PK) exposure, 16-20% of patients do not reach
tablets initiated a new dose finding study of olaparib tablets BID continu- the 20 mg/L exposure threshold related to prolonged progression free survival
ously with carboplatin. Methods: Patients were included in a 3+3 dose- (20 weeks versus 52 weeks, respectively) with the currently used fixed dose of
escalation schedule in the following dose-levels: olaparib 25mg BID and 800 mg QD (Suttle, 2014; Verheijen, 2017). PK simulations showed that, due
carboplatin AUC 3 d1/d22, olaparib 25mg BID and carboplatin AUC 4 d1/ to non-linear absorption of pazopanib, splitting the intake (400 mg BID) leads
d22, olaparib 50mg BID and carboplatin AUC4 d1/d22, olaparib 75mg and to an increase in Cmin and AUC0-24h of 75% and 59%, respectively (Yu, 2017).
carboplatin AUC 4 d1/d22 and olaparib 100mg BID and carboplatin AUC 4 This study aimed to show whether switching patients from an 800 mg QD to a
d1/d22. After two cycles patients continued olaparib 300mg BID as mon- 400 mg BID dose schedule will lead to a significant increase in PK exposure.
otherapy. Primary objective was to assess the Maximum Tolerable Dose Methods: We performed a prospective PK trial (NL6137) in which PK sam-
(MTD). Secondary objectives were to investigate the preliminary response pling at the 800 mg QD dose schedule occurred at day 1, after which the intake
rate, pharmacodynamics and systemic exposure. Results: In total 24 pa- moments were split into 400 mg BID during one week, followed by PK
tients were included with breast cancer (n = 18), ovarian cancer (n = 3), sampling at day 8. Paired samples t-tests were used to assess differences in
melanoma (n = 1), colorectal cancer (n = 1) and esophageal cancer (n = 1). Cmin, Cmax and AUC0-24h between these two dose schedules. To detect an
Nineteen out of 24 patients had a germline BRCA mutation (79%). Most increase in PK exposure of 50% (2-sided a = 0.05 and b = 0.20), 10 evaluable
common AEs were nausea (46%), fatigue (33%) and platelet count decrease patients were needed. Results: Eleven patients (6 RCC and 5 STS) have been
(33%). The majority of AEs (83%) were grade 1/2 in severity. Because two included, of whom ten were evaluable for PK analyses. Using the 800 mg QD
dose-limiting toxicities (consisting of $ 7 days dose delay of cycle 2 or dose schedule mean Cmin, Cmax and AUC0-24h were 26.7 mg/L (coefficient of
missing $ 5 doses of olaparib due to hematologic toxicity) occurred in dose- variation (CV%) 44.7), 46.1 mg/L (CV% 37.1) and 809 mg h/L (CV% 42.1),
level 4, dose-level 3 (olaparib 75mg and carboplatin AUC 4; n = 6 patients) respectively. Switching to 400 mg BID resulted in an increase of both Cmin and
was determined to be the MTD. Fourteen out of 24 patients (56%) had a AUC0-24h to 40.7 mg/L (CV% 37.0, p = 0.013) and 1059 mg h/L (CV% 33.1,
partial response as best response, according to RECIST 1.1. Systemic ex- p = 0.068), respectively, while Cmax did not significantly change (56.5 mg/L,
posure of the olaparib tablet formulation appeared comparable to the pre- CV% 33.2, p = 0.185). One patient (9%) experienced grade 3 diarrhea after
vious capsule formulation with an olaparib tablet AUC0-14 of 16.3 mg/ml*h at splitting intake moments, leading to treatment interruption. This strategy could
MTD. PARP activity in PBMCs was decreased by 98.7% 6 0.14% at day potentially save up to 2000 USD/patient/month compared to conventional
eight compared to day one for dose-level 3. Conclusions: Olaparib tablets dose increments. Conclusions: This study demonstrates that boosting pazo-
75mg BID and carboplatin AUC 4 for two cycles preceding olaparib mon- panib exposure by splitting intake moments leads to a significant increase in
otherapy is a feasible and tolerable treatment schedule with encouraging Cmin, of 52%, with acceptable tolerability. Therefore, this new dose schedule
clinical antitumor activity. Clinical trial information: NCT02418624. offers a safe and cost-neutral opportunity to optimize treatment for patients
with low PK exposure. Clinical trial information: NL6137.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
178s Developmental Therapeutics and Tumor Biology (Nonimmuno)

3120 Poster Session (Board #112), Sat, 8:00 AM-11:00 AM 3121 Poster Session (Board #113), Sat, 8:00 AM-11:00 AM
Long-term follow-up of pharmacokinetics (PK) and immunogenicity of the Accumulation of active metabolite M-2 predicts overall survival (OS) of
anti–PD-1 antibodies nivolumab (Nivo) and pembrolizumab (Pembro) in chemorefractory metastatic colorectal cancer patients treated with regor-
real-world practice. First Author: Masahide Fukudo, Department of Hospital afenib (REGO). First Author: Benoit Rousseau, Oncology Department,
Pharmacy and Pharmacology, Asahikawa Medical University, Asahikawa, Japan Hopital Henri Mondor, APHP, Creteil, France
Background: The PD-1 blockers Nivo and Pembro are widely used to treat Background: TEXCAN, a prospective phase II GERCOR study of treatment with
patients (pts) with various types of cancer, but their PK and immunogenicity REGO in chemorefractory metastatic colorectal cancer (mCRC) patients
have not been adequately characterized in clinical practice. Here we report the (NCT02699073) included a prospective pharmacokinetic (PK) ancillary study
first long-term follow-up of PK and anti-drug antibodies (ADAs) of Nivo and aiming to investigate correlations between OS and concentrations (C) of REGO
Pembro, correlated with efficacy and safety. Methods: We included 147 pts and its active metabolites M-2 and M-5. Methods: 55 patients were included,
receiving Nivo (n = 98) or Pembro (n = 49) between May 2016 and Jan 2019. with the same inclusion/exclusion criteria as CORRECT (NCT01103323), and
Plasma samples were longitudinally collected before each infusion and after treated orally with 160 mg REGO daily for 3 weeks on and 1 week off. 34
discontinuation for as long as samples were obtainable. Drug concentrations patients had PK samples at C1D15 and 26 at C2D15 for Cmin. REGO, M-2
were measured by ELISA (LLOQ: 0.0125 mg/mL), and ADAs were evaluated by and M-5 Cmin were measured by LC-MS/MS. PK analyses studied the link
bridging ELISA. Results: Median (range) follow-up was 6.0 (0.1-38.7) mo, and between OS and PK parameters: Cmin of REGO, M-2 and M-5 at C1 and
1718 samples were analyzed. ADAs were confirmed at baseline or at last accumulation of pharmacological active metabolites between C1 and C2,
sample for both Nivo (2 [2.2%] and 4 [4.5%] pts, respectively) and Pembro (2 assessed by the C2/C1 ratio of M-2 or M-5 Cmin concentrations. Results: REGO,
[4.2%] and 3 [6.7%] pts). Of the 4 baseline ADA-positive pts, 3 experienced M-2 and M-5 Cmin [median (Q1-Q3)] were respectively 1.99 (1.03-2.73), 1.44
drug-induced fever after initial infusion. Pts developing ADAs at last sample (0.89-2.49) and 1.61 (0.79-2.37) mg/L at C1D15 and 1.90 (1.10-2.76), 1.29
had earlier progression than ADA-negative pts (median PFS: 46 vs. 119 days, (0.77-2.24) and 1.17 (0.45-2.42) mg/L at C2D15. C2/C1 M-2 ratio and M-5
log-rank P = 0.0827). Persistent drug exposure until ~1 y beyond discon- ratio medians were 0.82 (0.50-1.78) and 0.75 (0.41-1.93), respectively.
tinuation was observed for both drugs. In 1 Nivo-treated pt with delayed Univariate analyses showed a major OS benefit in patients with C2/C1 M-2
adrenal insufficiency 8.6 mo after discontinuation, Nivo was still detectable ratio $median vs , median (12.6 vs 4.0 months respectively, hazard ratio =
(0.2 mg/mL). In 71 and 41 efficacy-evaluable pts receiving Nivo and Pembro, 0.35, 95% confidence interval 0.14-0.86, p-value = 0.023) but not for C2/C1
respectively, mean trough levels in the early period (~cycle 6) were significantly M-5 ratio $median. Multivariate analyses, including the CORRECT REGO-
higher in pts achieving response than in pts with progressive disease at first SCORE groups, showedan independent 66% reduction in death risk in the group
assessment (Nivo: 43.5 vs. 31.0 mg/mL, P = 0.0107; Pembro: 30.6 vs. 22.1 of patients with C2/C1 M-2 ratio $median. The C2/C1 M-2 ratio correlated with
mg/mL, P = 0.0174). Conclusions: Our findings provide insight into the eti- C1 REGO+M-2+M-5 (Csum) (0.53, p-value = 0.006). Restricted Cubic spline
ological mechanism of late-onset adverse events associated with PD-1 analysis showed an increased OS benefit as the C2/C1 M-2 ratio rises and when
blockers. Moreover, ADA may potentially influence clinical outcomes, and it C1 Csum ranged between 2.5 and 5.5 mg/L. PK parameters were not associated
may be possible to optimize dose in certain pts with lower drug exposure for with toxicities. Conclusions: M-2 accumulation between C1 and C2 is in-
improved efficacy, warranting further investigation. Clinical trial information: dependently associated with improved OS in mCRC patients treated by REGO.
UMIN000033036. M-2 accumulates and OS is favorable when C1 REGO+M-2+M-5 sum ranged
Nivo (n = 98) Pembro (n = 49)
between 2.5 and 5.5 mg/L. These results may lead to develop individual REGO
dosage modification strategies based on PK monitoring. Clinical trial in-
Cancer type, n
NSCLC/MPM 21/2 37/0 formation: NCT02699073.
SCCHN 21 0
Gastric cancer 21 0
RCC/ Urothelial carcinoma 18/0 0/10
MEL 15 2

3122 Poster Session (Board #114), Sat, 8:00 AM-11:00 AM 3123 Poster Session (Board #115), Sat, 8:00 AM-11:00 AM
Larotrectinib efficacy and safety in adult TRK fusion cancer patients. First Phase 1b study of selinexor, a first-in-class selective inhibitor of nuclear
Author: David S. Hong, The University of Texas MD Anderson Cancer Center, export (SINE) compound, in combination with doxorubicin in patients (pts)
Houston, TX with locally advanced or metastatic soft tissue sarcoma (STS). First Author:
Eoghan Ruadh Malone, 3601, Dublin, Ireland
Background: A broad range of pediatric and adult malignancies harbor TRK
fusions involving the NTRK1, NTRK2, and NTRK3 genes. The highly- Background: Selinexor is a first-in-class SINE compound with single-agent
selective TRK inhibitor, larotrectinib, has previously shown a high overall activity in STS. We undertook this study to determine the safety, tolerability
response rate (ORR) and a favorable safety profile in patients (pts) with TRK and efficacy of selinexor in combination with doxorubicin in pts with in-
fusion cancer. To better delineate efficacy in adults, as pediatric pts have a curable STS. Methods: This phase 1b study was conducted using a bayesian
particularly high ORR, here we report updated efficacy and safety data from model (modified toxicity probability index). Patients with locally advanced or
the adult subset of pts with TRK fusion cancer treated with larotrectinib. metastatic STS received selinexor at either 60 or 80mg weekly PO plus
Methods: Adult pts (aged 18 or older) with TRK fusion cancer detected doxorubicin (75mg/m2 IV q21 days, max 6 cycles). Pts with stable disease
by local testing in 2 larotrectinib clinical trials (NCT02122913 and (SD) or better (per RECIST 1.1 criteria) after 6 cycles of combination
NCT02576431) were analyzed. Larotrectinib was administered 100 mg PO treatment received selinexor monotherapy until disease progression or un-
BID until disease progression, withdrawal, or unacceptable toxicity. Disease acceptable toxicity. Disease assessments were made with standard imaging
status was assessed by both investigator (INV) and independent assessment after every 2 cycles. Results: 24 pts (19F/5M, ECOG 0/1: 12/12, median age
(IRC) using RECIST v1.1. Results: As of July 30, 2018, 83 adults (median 58.5 years [range 34-74]) were enrolled. Disease subtypes included leio-
age: 57 y, range 20–80 y) with TRK fusion cancer had been treated. Cancer myosarcoma (n = 6), malignant peripheral nerve sheath tumor (n = 3) and
types included salivary gland (23%) and thyroid cancer (19%), soft tissue other sarcomas (n = 15). Three pts at 60mg selinexor and 21 pts at 80mg
sarcoma (14%), lung cancer (13%), colon cancer and melanoma (7% each), selinexor were treated. The most common G3 drug related adverse events
GIST (5%), and bone sarcoma, cholangiocarcinoma, and appendiceal, were hematological, neutropenia n = 13 (54%), anemia n = 6 (25%). There
breast, and pancreas cancer (#2% each). TRK fusions involved NTRK1 were 4 dose-limiting toxicities (2 febrile neutropenia, 1 vomiting and 1
(40%), NTRK2 (2%), and NTRK3 (57%). 77% of pts had received prior unresolved fatigue) all at the 80mg dose level, but does not satisfy criteria for
systemic therapy (median lines: 2, range 0–10). In 74 pts evaluable per INV, maximum tolerated dose. Two patients had clinically significant and relevant
the ORR was 76% with 9% CR, 57% confirmed PR, 9% PR pending drop in ejection fraction, presenting with cardiac symptoms. Of the 24
confirmation, 12% SD, 11% PD, and 1% not determined; 9 pts were non- evaluable pts 4 (17%) had a partial response, 16 (67%) had SD as best
evaluable (NE) due to lack of post-baseline assessment. In 65 pts evaluable response and SD . 16 weeks was seen in 13 pts (54%). PK analysis of
per IRC, the ORR was 68% with 17% CR, 51% PR, 15% SD, 12% PD, and selinexor did not demonstrate changes compared to single agent profile. The
5% NE. With a median follow up of 17.2 and 17.5 mo per INV and IRC, estimated median PFS and OS are 5.5 (95% CI:4.1-7.0) and 9.4 (6.6-13.8)
respectively, the median duration of response had not been reached (ranges months. Conclusions: Our initial data demonstrate that the combination of
identical: 1.9+ to 38.7+ months). At data cutoff, 63% remained on selinexor at 80mg with doxorubicin is tolerable and is associated with
treatment; 30% had discontinued due to disease progression. Adverse clinical benefit. Longer term follow up of available patients will be needed to
events were mostly grade 1–2. Conclusions: Larotrectinib demonstrated understand toxicity profile. Clinical trial information: NCT03042819.
robust tumor-agnostic efficacy and a favorable safety profile in adult pts with
TRK fusion cancer. These results support testing for TRK fusion cancer in pts
with advanced solid tumors, regardless of site of primary diagnosis. Clinical
trial information: NCT02122913 and NCT02576431.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Therapeutics and Tumor Biology (Nonimmuno) 179s

3124 Poster Session (Board #116), Sat, 8:00 AM-11:00 AM 3125 Poster Session (Board #117), Sat, 8:00 AM-11:00 AM
A phase I study of a novel MDM2-P53 antagonist APG-115 in Chinese A phase I study of a novel IAP inhibitor APG-1387 as a monotherapy or in
patients with advanced soft tissue sarcomas. First Author: Xing Zhang, combination with pembrolizumab in treatments of patients with advanced
Melanoma and Sarcoma Medical Oncology Unit, Sun Yat-Sen University solid tumors. First Author: Drew W. Rasco, South Texas Accelerated Re-
Cancer Center, Guangzhou, China search Therapeutics (START), San Antonio, TX
Background: APG-115 is a novel and orally active small-molecule MDM2 Background: APG-1387 is a novel, bivalent small molecule IAP (inhibitor of
inhibitor. APG-115 alone or in combination with chemotherapeutic, targeted apoptosis proteins) inhibitor. It has shown strong antitumor activities in
or IO agents have shown potent antitumor activities in multiple human xe- multiple human xenograft cancer models. APG-1387 also acts as host immune
nograft tumor models and human cancer patient derived xenograft (PDX) modulator, supporting the notion that APG-1387 in combination with anti-
models. Methods: The patients with advanced solid tumors were enrolled in PD1 antibody for cancer therapy. Methods: This study consists of two parts
this study in China (CTR20170975). The study objectives were to assess (NCT03386526). Part 1 is the dose escalation study of APG-1387 including a
safety, pharmacokinetics (PK), pharmacodynamics (PD) and antitumor activity mPC (metastatic pancreatic cancer) cohort expansion. Part 2 is the dose
of APG-115. The patients received APG-115 (ranging 100–200 mg) orally escalation and cohort expansion study of APG-1387 in combination with
QOD for first 21 days of a 28-day-cycle, until disease progression. Antitumor pembrolizumab. APG-1387 is administrated IV for 30 minutes once weekly
response assessment was performed every 8 weeks per RECIST v1.1. Archived in a 21-day-cycle. Pembrolizumab is administrated at 200mg IV on day1 of a
tumor tissues were collected for analyses of MDM2 and TP53 before treat- 21-day-cycle. The study objectives are to assess the safety, pharmacokinetics
ment. Results: As cut-off on Jan 4 2019, total 13 patients (9 soft tissue (PK), pharmacodynamics (PD), and efficacy (assessed per RECIST v1.1 every
sarcomas (STSs), 2 adenoid cystic carcinomas (ACCs) and 2 osteosarcomas) 6 weeks). Results: Through Jan 4, 2019, total 23 patients had been treated in
were treated in 3 cohorts of APG-115 (100mg, 150mg, 200mg). The median 4 cohorts of APG-1387 (20mg, 30mg, 45mg, 60mg). Two DLTs were observed
number of prior systemic anticancer therapies was 2 (range 0-4). Two DLTs at 60mg including lipase increase and Bell’s palsy, MTD was determined
were observed in one patient at 200mg including thrombocytopenia and febrile as 45mg. Nineteen of 23 patients experienced at least 1 TEAE. The most
neutropenia. The most common TEAEs ($50% of pts) included: anemia, common TEAEs were nausea (21.7%), fatigue (17.4%), decreased appetite
thrombocytopenia, vomiting, hypercholesterolaemia, and leukopenia. SAEs (13.0%), and abdominal pain (13.0%). Three grade 3 TEAEs including el-
occurred in 7 patients (54%), four of which were treatment related. The most evated bilirubin, lipase increase, and shortness of breath were documented at
common Grade 3 or 4 TRAEs were anemia (38.5%), thrombocytopenia 45mg and 60mg. Three out of six mPC patients (one at 60mg, two at 45mg)
(38.5%), leukopenia (30.8%), and neutropenia (23.1%). One partial response achieved SD, one of them at 45mg has been treated . 5 cycles with confirmed
was observed in a liposarcoma patient with MDM2-amplification and TP53- SD (-18%). Two patients, who were treated with APG-1387 at 20mg in
wild type at the 150mg cohort, 5 patients (3 STSs, 2 ACCs) had SD as the best combination with pembrolizumab, had no DLT observed during the first cycle.
overall response. PK analyses indicated an approximately dose proportional Preliminary PK data of APG-1387 showed a dose proportionality in exposure
increase in Cmax and AUC0-t following a single or multiple oral administration (Cmax and AUC) over the dose range of 20-60 mg. Conclusions: APG-1387 was
across dose levels. Preliminary PD data showed that serum MIC-1 increase was well tolerated and had manageable adverse events. The potential effects of
exposure dependent within the dose range tested. Conclusions: Preliminary APG-1387 alone or in combination with pembrolizumab deserve further ex-
data suggested that APG-115 had promising anti-tumor activity in treatment of ploration in patients with advanced solid tumors, especially in the mPC pa-
patients with MDM2-amplification and TP53-WT liposarcoma. Safety profile tients. Clinical trial information: NCT03386526.
and PD effect were consistent with other MDM2 inhibitors. Dosing regimen
optimization are ongoing. Clinical trial information: CTR20170975.

3126 Poster Session (Board #118), Sat, 8:00 AM-11:00 AM 3127 Poster Session (Board #119), Sat, 8:00 AM-11:00 AM
A phase I study of a novel MDM2 antagonist APG-115 in patients with A dose escalation pharmacokinetic (PK) and pharmacodynamic (PD) study
advanced solid tumors. First Author: Drew W. Rasco, South Texas of mTORC1/2 inhibitor XP-105 (BI 860585) as monotherapy and in com-
Accelerated Research Therapeutics (START), San Antonio, TX bination with exemestane or paclitaxel in patients (pts) with advanced solid
tumors. First Author: Filippo G. De Braud, Fondazione IRCCS Istituto
Background: APG-115 is a potent and orally active small-molecule MDM2
Nazionale dei Tumori, Milan, Italy
protein inhibitor. Binding to MDM2 protein, APG-115 restores p53 tumor
suppressive function via induction of apoptosis in tumor cells retaining wild- Background: Resistance to mTORC1 inhibition may develop through feed-
type p53. In addition, enhanced antitumor activity was demonstrated in the back loop leading to upregulation of mTORC2. XP-105, also known as BI
syngeneic tumor models after APG-115 combined with PD-1 blockade. 860585, is a potent dual mTORC1/2 inhibitor designed to overcome such
Methods: This Phase I study (APG-115-US-001) was designed to enroll the resistance. This Phase 1 trial (NCT01938846) was performed to determine
patients with advanced solid tumors in US (NCT02935907). Study ob- the MTD and activity of XP-105 alone or in combination with exemestane or
jectives included to assess safety, dose limited toxicity (DLT), pharmaco- paclitaxel in pts with advanced solid tumors. Methods: A 3+3 escalation
kinetics (PK), pharmacodynamics (PD), and antitumor activity (assessed design was used; Pts received XP-105 (5–300 mg/day) monotherapy or
every 8 weeks per RECIST v1.1). The patients received APG-115 orally every (40–220 mg/day) combined with fixed-dose exemestane 25 mg/day, or
other day (QOD) at the designated dose (ranging from 10 to 300 mg) for first 80–160 mg/day combined with paclitaxel 60 or 80 mg/m2/week. A reduction
21 days of a 28-day cycle, until disease progression. Results: Up until Jan 4 of pAKT/total AKT ratio was used as a PD marker of target inhibition.
2019, total 28 patients were treated with APG-115 at various doses (one Results: 90 pts were treated (41 with monotherapy, 25 and 24 in combi-
patient at 10mg, 20mg and 50mg, respectively; 14 patients at 100mg, 6 nation with exemestane, or paclitaxel respectively). XP-105 MTD was de-
patients at 200mg, and 5 patients at 300mg). The median number of prior fined as 220 mg daily for monotherapy, and 160mg daily with exemestane
systemic anticancer therapies was 4 (range 0-15). The DLTs were observed 25 mg/d or paclitaxel 80 mg/m2/week. In the monotherapy arm, stable
during cycle 1, including one grade 2 thrombocytopenia at 200mg, one grade disease (SD) was reported in 8 pts (20%), with a median duration of
3 thrombocytopenia at 300mg, and one grade 3 fatigue at 100mg and 300mg 11 months. In the exemestane combination arm, 4 (16%) partial responses
respectively. The most common AEs (reported in $10% of pts) included: (PR) were reported. In the paclitaxel combination arm, 1 complete response
fatigue, nausea, vomiting, diarrhea, decreased appetite, dehydration, neu- (CR) and 4 PRs were reported (OR rate 21%). Disease control rate (CR/PR/
trophil count decreased, white blood cell count decreased, pain in extremity, SD) was 20%, 28%, and 58% in the monotherapy, XP-105/exemestane, and
thrombocytopenia. The most common Grade 3 or 4 treatment related AEs were XP-105/paclitaxel arms, respectively. A sustained reduction in pAKT/total
fatigue (10.7%), and thrombocytopenia (10.7%). Six patients had stable AKT to , 50% of baseline levels was observed with XP-105 $120mg daily.
disease (SD) after two cycle treatments, two of them are continuing in this Overall, XP-105 was well tolerated; in the XP-105/paclitaxel combination
study. PK analyses indicated that exposure (Cmax and AUC) generally increases the most frequent drug-related AEs were diarrhea and fatigue (58.3% each),
with the increase of dose level from 20 mg to 300 mg. Conclusions: APG-115 hyperglycaemia (54.2%), anaemia (50%). Grade $3 AEs were hyper-
was well tolerated and had manageable adverse events. The MTD/RP2D of APG- glycaemia, fatigue, diarrhea, anaemia, leukopenia. No PK interaction was
115 monotherapy with oral administration, QOD for 21 days of a 28-day cycle for observed. Conclusions: The MTD for XP-105 monotherapy and in combi-
treatment of patients with advanced solid tumors was determined as nation with exemestane or paclitaxel was defined as 220 mg and 160mg
100 mg. Further evaluation of APG-115 in combination with pembrolizumab once daily, respectively. Combination regimens showed higher activity as
in patients with advanced solid tumors is ongoing. Clinical trial information: compared to monotherapy with durable OR in about 20% of pts. The ob-
NCT02935907. served safety profile of XP-105 compared favorably to those reported from
other mTOR inhibitors. Clinical trial information: NCT01938846.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
180s Developmental Therapeutics and Tumor Biology (Nonimmuno)

3128 Poster Session (Board #120), Sat, 8:00 AM-11:00 AM 3129 Poster Session (Board #121), Sat, 8:00 AM-11:00 AM
Cytoplasmic cyclin E independently predicts recurrence in older patients Clinicopathologic characteristics of NRG1 fusion-positive cancers: A single-
with primary breast cancer. First Author: Simon Johnston, University of institution study. First Author: Alison M. Schram, Memorial Sloan Kettering
Nottingham, Nottingham, United Kingdom Cancer Center, New York, NY
Background: Primary breast cancer in the older ( . 70 years) population has Background: NRG1 rearrangements are oncogenic drivers across several tu-
distinct biological characteristics associated with favourable outcome, such mor types. Chimeric proteins encoded by NRG1 fusions activate HER3,
as higher rate of estrogen receptor (ER) positivity. Due to comorbidities, older resulting in heterodimerization with HER2 and activation of downstream
patients with primary breast cancer are more likely to die of non-breast signaling. Preclinical and preliminary clinical data suggest that targeting
cancer-related causes compared to their younger counterparts. Biomarkers HER3 may be an effective treatment strategy for patients with NRG1 fusion-
that may influence treatment strategy therefore require interpretation in the positive tumors. We aimed to describe the clinical and genomic characteristics
specific biological and clinical context of older women. Cyclin E regulates of patients identified at our institution with NRG1 fusions. Methods: We
cell cycle transition from G1 to S phase, and its deregulation is implicated in analyzed results from prospective targeted exome and/or RNA sequencing
breast cancer pathogenesis. Tumour-specific isoforms of cyclin E localise to performed at Memorial Sloan Kettering between 2014-2018 involving
the cytoplasm. Expression of cytoplasmic cyclin E (c-cyclin E) is linked with . 30,000 samples. NRG1 fusion-positive tumors were identified and these
poor clinical outcome. We now present multivariate analysis of breast cases were manually reviewed. Results: NRG1 fusions were detected in 24
cancer-specific survival (BCSS) by c-cyclin E and clinical markers of disease patients. Cancer types included lung (N = 9), pancreas (N = 7), breast (N = 5),
biology from a cohort of older women. The primary outcome, BCSS, excludes prostate (N = 1), gallbladder (N = 1), diffuse large B-cell lymphoma (N = 1),
deaths from competing causes and is used as a surrogate for tumour biology. and cancer of unknown primary (N = 1). 6/9 lung cancers had mucinous
Methods: Between 1973 and 2010, 813 older women underwent initial differentiation. The majority of patients were Caucasian (N = 17), half were
surgery for early breast cancer and were followed up in a dedicated clinic in female (N = 12) and ages ranged from 24-82 years-old. Targeted exome
Nottingham. Excised tumours from 517 of these patients were successfully sequencing identified the fusion in 10/23 cases tested, including 2 not
incorporated into a tissue microarray (TMA). Expression of c-cyclin E was confirmed by RNA. The remaining 14 were detected using RNA. Fusion
assessed by IHC using an assay developed at MDACC, along with a panel of partners included CD74 (N = 6), SLC3A2 (N = 2), SDC4 (N = 2), ATP1B1 (N =
24 biomarkers. Of these, ER, progesterone receptor (PR), human epidermal 2), FOXA1 (N = 1), SLCA4 (N = 1), ROCK1 (N = 1), TNKS (N = 1), CCND1 (N =
growth factor 2 (HER2) and Ki67 are in current clinical use and are analysed 1), PAK1 (N = 1), STAU3 (N = 1), RAD51 (N = 1), CD44 (N = 1), NCOR1 (N =
alongside c-cyclin E. Grade was assessed from the primary tumour. Multi- 1), RBPMS (N = 1), and WHSC1L1 (N = 1). Pancreas cancers were KRAS wild-
variate analysis of BCSS was performed by Cox proportional hazard test. type and lung cancers had no co-occurring alterations in ALK, ROS1, EGFR,
Results: In multivariate analysis alongside markers of disease biology cur- RET, MET, RAS, or RAF. All tumors were microsatellite stable. A durable
rently used in the clinic (ER, PR, HER2, Ki67 and grade), c-cyclin E is the response was achieved with anti-HER3 antibody therapy (GSK2849330) in a
only factor that independently predicts BCSS in this cohort of older women patient with a CD74-NRG1-rearranged invasive mucinous adenocarcinoma
(HR 5.0, 95% CI 2.1 – 12.0; p, 0.001). Conclusions: In the older pop- (previously reported). Four patients treated with an irreversible small molecule
ulation with primary breast cancer, c-cyclin E expression is the only in- HER2 inhibitor (afatinib) did not respond to treatment, suggesting direct
dependent biological marker of BCSS. Patients with low c-cyclin E targeting of HER3 may be superior to HER2 inhibition in patients with NRG1
expression are unlikely to die of breast cancer. These data have potential to fusion-positive tumors. Conclusions: NRG1 fusions occur in several tumor
influence treatment strategy in older patients. For example, patients with types and may be amenable to targeting with HER3-directed therapy.
ER+, c-cyclin E negative disease plus multiple co-morbidities may be
suitable for primary endocrine therapy. This hypothesis warrants prospective
clinical evaluation.

3130 Poster Session (Board #122), Sat, 8:00 AM-11:00 AM 3131 Poster Session (Board #123), Sat, 8:00 AM-11:00 AM
Molecular differences between lymph nodes (LNs) and distant metastases Measuring phospho-MET by multiplex immunofluorescence to aid in selection
(mets) in colorectal cancer (CRC). First Author: Alberto Puccini, USC Keck of patients with MET activation in tumors. First Author: Tony Navas, Clinical
School of Medicine, Los Angeles, CA Pharmacodynamics Biomarker Program, Applied/Developmental Directorate,
Frederick National Laboratory for Cancer Research, Frederick, MD
Background: LNs mets are thought to occur before distant mets. However,
lymphatic and distant mets arise from independent subclones of the primary Background: Currently, patient selection criteria for clinical testing of MET
tumor, suggesting that LNs are not essential intermediaries for distant mets. inhibitors are limited. Robust studies selecting patients based on MET protein
We aimed to comprehensively characterize the molecular profile of LN mets expression, MET gene amplification, or mutations have not met their efficacy
and to explore the differences between LN vs distant mets and primary goals. Development of microscopy-based assays to quantify levels of phospho-
tumors. Methods: Tumor samples from primary CRCs, LNs, and distant mets MET (pMET) in tumors has been hampered by poor antibody specificity. Here,
were analyzed using NGS (MiSeq on 47 genes, NextSeq on 592 genes), we present the development and validation of a robust, highly specific multiplex
immunohistochemistry. Tumor mutational burden (TMB) was calculated immunofluorescence assay (IFA) that measures pY1235-MET and total MET in
based on somatic nonsynonymous missense mutations, and microsatellite tumor tissue. Methods: This assay utilizes antibodies to pY1235-MET (NCI-
instability (MSI) was evaluated by NGS of known MSI loci. Results: In total, 23111), total MET (D1C2), and plasma membrane (PM) marker Na+/K+-ATPase,
11871 tumors samples were examined, comprising primaries (N = 5862), each conjugated to a different Alexa Fluor dye. We used tumor tissue from
distant (N = 5605) and LNs mets (N = 404). The most frequently mutated crizotinib-treated SNU5 xenograft models to demonstrate pY1235-MET assay
genes in LNs were TP53 (72%), APC (61%), KRAS (39%), ARID1A (20%), fitness-for-purpose and cross-platform assay concordance with our validated
and PIK3CA (12%). LNs showed a higher mean TMB (13 mut/MB) vs distant pMET ELISA. In addition, this IFA was validated by phospho-peptide competition
mets (9, P , .0001). TMB-high (17mut/MB) was more frequent in primaries using custom tissue microarrays (TMA) derived from patients with colorectal
and LNs vs distant mets (9.5% and 8.8% vs 4.2%, P , .001 and P = .001, carcinoma (CRC). Finally, we developed quantitative algorithms to assess
respectively), as well as MSI-H (8.8% and 6.9% vs 3.7%, P , .001 and P = pY1235 MET levels in the plasma membrane and nucleus using PM and DAPI
.017, respectively). TMB-high is significantly higher in LNs vs distant mets masks, respectively. Patient-derived xenograft models (PDX) were obtained from
and primaries (P , .0001), independent of MSI-H status. Analyzing distant NCI’s Patient-Derived Models Repository (www.pdmr.cancer.gov). Results: The
mets by location, LNs showed higher TMB compared to lung, liver and prevalence of high pY1235-MET expression in CRC patient specimens was
peritoneum mets (P , .0001). Overall, LNs showed significantly different greater than expected; of the 64 TMA cores evaluated, 29 (45%) and 19 (29%)
rates of mutations in APC, KRAS, PI3KCA, KDM6A, and BRIP1 (P , .01 for had high pY1235-MET and total MET levels, respectively, as defined by mean
all comparisons) vs primaries; while presenting a distinct molecular profile marker area of $ 30 mm2/cell. To address the potential utility of pY1235-MET
compared to distant mets (TP53 72 vs 67%; KRAS 39 vs 50%; RNF43 7 vs as a diagnostic biomarker, we examined 15 CRC PDX models by pMET ELISA
4%; ATM 5 vs 3%; KDM6A 4 vs 1%; BRCA2 4 vs 2%; MSH6 3 vs 2%; and IFA. Two CRC tumor models were positive for pY1235-MET expression in
PTCH1 4 vs 1%; BRCA1 2 vs 1%; GNAS 2 vs 5%; P , .05 for all com- both assays. The pY1235-MET IFA results and gene expression data were used to
parisons). Our cohort of 30 paired samples confirmed the molecular het- select PDX models for ongoing preclinical trials of potent MET inhibitors.
erogeneity between primaries, LNs, and distant mets. Conclusions: This is Conclusions: This novel pY1235-MET IFA will enable clinicians to address the
the largest study to investigate the molecular differences between LNs mets, utility of activated MET as a biomarker for patient selection and/or prediction of
distant mets and primary tumors in CRC patients. Our data support the response in clinical trials of MET inhibitors. Funded by NCI Contract No.
hypothesis that lymphatic and distants mets harbor different mutation HHSN261200800001E.
profiles which suggests that they may arise from distinct subclones.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Therapeutics and Tumor Biology (Nonimmuno) 181s

3132 Poster Session (Board #124), Sat, 8:00 AM-11:00 AM 3133 Poster Session (Board #125), Sat, 8:00 AM-11:00 AM
Actionable coalterations in breast tumors with pathogenic mutations in the HER family protein expression and activation predicts response to combi-
homologous recombination DNA damage repair pathway. First Author: nation T-DM1/pertuzumab in HER2+ patients in the I-SPY 2 TRIAL. First
Arielle Lutterman Heeke, Levine Cancer Institute, Atrium Health, Charlotte, NC Author: Julia Dianne Wulfkuhle, George Mason Univ, Columbia, MD
Background: Homologous recombination (HR) deficient breast tumors may have Background: T-DM1 (T), a conjugate of the anti-HER2 therapeutic antibody
genomic alterations that suggest responsiveness to targeted therapies other than trastuzumab and the microtubule assembly inhibitor emtansine, was ad-
PARP inhibitors. Methods: Comprehensive molecular profiles of 4,647 breast ministered in combination with pertuzumab (P), an anti-HER2 therapeutic
tumors performed at Caris Life Sciences using 592-gene NGS (average read antibody, to HER2+ breast cancer patients in the I-SPY 2 TRIAL, and
depth 500X) were reviewed to identify somatic pathogenic mutations in HR graduated in all HER2+ subtypes. Pre-specified biomarker analysis was
genes ARID1A, ATM, ATRX, BAP1, BLM, BRCA1/2, BRIP1, CHEK1/2, FANCA/ performed to identify candidate biomarkers associated with pCR within the
C/D2/E/F/G/L, KMT2D, MRE11, NBN, RAD50/51/51B & PALB2, as well 41 HER family and cell proliferation pathways in patients treated with T+P. We
markers associated with treatment response. Results: Overall, 17.9% of breast hypothesized that quantitative measurement and activation of HER2 and
tumors have HR mutations (HR-MT, 831/4647). HR-MT is seen most in HER2– activation of its major dimerization partner, EGFR would predict response to
disease [hormone receptor (hr)+/HER2– (18.3%, n=2183), TNBC (18.2%, T+P. Methods: In the T+P treatment arm, 49 had RPPA and pCR data. 40
n=1568), hr–/HER2+ (12.9%, n=217)]. Mean TMB is higher for HR-MT tumors RPPA endpoints including 14 total/phospho-proteins in the HER family were
across subtypes (9.2 mut/Mb vs 7.6 WT, p=,0.0001) & independent of MS assessed for association with pCR using logistic regression (likelihood ratio
status. HR-MT hr–/HER2+ tumors are more likely to have PD-L1 overexpression test; p , 0.05). Analysis was also performed adjusting for HR status and
(25% vs 13.1% hr–/HER2+ WT, p=0.10), whereas MSI is more prevalent in HR- within HR subsets. Markers were analyzed individually; multiple comparison
MT HER2– (hr+/HER2– 2.3%, TNBC 1.4%, HER2+ 0%). Mutations in chro- correction (Benjamini-Hochberg) was applied to all p-values. Our statistics
matin remodeling genes (*) are more common in HR-MT. Additional co- are descriptive and do not adjust for multiplicities of other biomarkers
alterations are outlined in the Table. Conclusions: In breast cancer, HR-MT is outside this study. Results: Of the endpoints tested, only quantitative total
associated with HER2– disease & markers of response to immunotherapy. HER2 expression, phospho-HER2 (Y1248 and Y877), phospho-EGFR
Clinical trials combining HRD targeted agents & immunotherapy are underway & (Y1173 and Y1068), and phospho-SHC Y317 had a positive association
could be enriched through comprehensive molecular profiling. Mutations were with response in the population as a whole, and in a model adjusting for HR
identified in both HR-MT & HR WT tumors that suggest other targets for status (BH p , 0.05). In HR subset analysis, these 5 analytes had un-
treatment. corrected p , 0.05 regardless of HR subtype but only survived p-value
correction in HR+ tumors. Conclusions: Quantitative measurement of HER2
HR-MT (%, +/total) HR WT (%, +/total) p-value
protein positively associates with response to T+P in patients already
PIK3CA 26.4 (217/823) 30.3 (1150/3792) 0.02 identified as HER2+ by central IHC and FISH testing. Activation of HER2
tumor PD-L1 (IHC) 13.2 (104/788) 11 (405/3677) 0.08
ESR1 8.2 (67/820) 7.9 (298/3790) 0.77 and its dimerization partner, EGFR, also associate with response to T+P in
RB1 4.9 (37/754) 4.4 (150/3442) 0.51 HR+ patients. While our results need to be validated in larger prospective
AKT1 2.1 (17/817) 3.7 (140/3789) 0.02 trials, they indicate that new approaches to measure more quantitatively the
ERBB2 2.8 (23/831) 2.6 (100/3812) 0.81
ARID2* 1.3 (10/796) 0.5 (20/3671) 0.03 amount and activation state of HER2 and activated EGFR may more ef-
JAK1 1.1 (9/796) 0.2 (7/3657) 0 fectively identify patients that respond to HER2 targeted therapies than
Pathogenic mutation frequency # 1%: AR, BRAF, CCND1, CDKN2A, EGFR, ERBB3, IDH1, HER2 IHC and FISH alone.
IDH2, JAK2, KIT, MET, MTOR, RET, SMARCB1*, SMARCE1*, SMARCA4*, SS18L1*
No mutations: ATR, AURKA/B, BCL7A*, BCL11A/B*, CDK4/6, ERBB4, NTRK1/2/3, PBRM1*,
POLE
By IHC (HR-MT vs WT): AR 52.3 (415/794) vs 54.9 (2014/3669) [p=0.18], EGFR 28.6 (4/14) vs
32.4 (36/111) [p=0.77], cMET 11.1 (1/9) vs 5.5 (5/91) [p=0.50]

3134 Poster Session (Board #126), Sat, 8:00 AM-11:00 AM 3135 Poster Session (Board #127), Sat, 8:00 AM-11:00 AM
Pharmacodynamics of PI3K, MEK, and AKT inhibitors through isoform- Machine learning methods with salivary metabolomics for breast cancer
specific measurements of MEK, ERK, AKT, and ribosomal protein S6 in detection. First Author: Takeshi Murata, Department of breast surgery,
needle biopsies. First Author: William Herrick, Leidos Biomedical Research, National Cancer Center Hospital, Tokyo, Japan
Inc., Frederick, MD
Background: Saliva is non-invasively accessible and informative biological fluid
Background: The success of drugs targeting the MAPK or PI3K pathways in which has high potential for the early diagnosis of various diseases. The aim of
cancer has been slowed by insufficient pathway suppression or onset of this study is to develop machine learning methods and to explore new salivary
resistance through rebound or cross-pathway activation. Isoform-specific biomarkers to discriminate breast cancer patients from healthy controls.
measurement of key signaling molecules and a pathway convergence marker Methods: We conducted a comprehensive metabolite analysis of saliva samples
ribosomal protein S6 (RPS6), may provide valuable pharmacodynamic (PD) obtained from 101 patients with invasive carcinoma (IC), 23 patients with
evidence to assess drug effectiveness. Methods: We developed Luminex ductal carcinoma in situ (DCIS) and 42 healthy controls, using capillary
multiplex assays to measure total and phosphorylated forms of RPS6, ERK1, electrophoresis and liquid chromatography with mass spectrometry to quantify
ERK2, MEK1, MEK2, AKT1, AKT2, and AKT3 from a single tumor biopsy. hundreds of hydrophilic metabolites. Saliva samples were collected under 9h
Fit-for-purpose validation was performed with three drugs currently in fasting and were split into training and validation data. Conventional statistical
clinical trials. A SW620 (KRAS G12V) model was tested with selumetinib analyses and artificial intelligence-based methods were used to access the
(NSC741078, 20 mg/kg), MK2206 (NSC756656, 24 mg/kg), and two discrimination abilities of the quantified metabolite. Multiple logistic regression
PTEN-null models (PC3 & HCC70) with AZD8186 (NSC777572, 25 & 50 (MLR) model and an alternative decision tree (ADTree)-based machine learning
mg/kg). Tumors were collected at multiple timepoints after single and re- methods were used. The generalization abilities of these mathematical models
peated dosing, processed by a validated method (PMID 27001313) and were validated in various computational tests, such as cross-validation and
data analyzed relative to vehicle control (n = 4-5/grp). Results: The multi- resampling methods. Results: Among quantified 260 metabolites, amino acids
plexed assays demonstrated satisfactory analytical performance. Treatment and polyamines showed significantly elevated in saliva from breast cancer
of the SW620 model with a single dose of selumetinib suppressed pERK1/2 patients, e.g. spermine showed the highest area under the receiver operating
( . 90%), MEK1/2 (50%) and pRPS6 (40-60%) up to 4 h (last time point characteristic curves (AUC) to discriminate IC from C; 0.766 (95% confidence
tested) which was reversed by 24 h. AKT inhibitor, MK2206, suppressed interval [CI]; 0.671 – 0.840, P , 0.0001). These metabolites showed no
pAKT1/2 (75%) and pAKT3 (50%) 2 h post-dose with reversal at 24 h but significant difference between C and DICS, i.e., these metabolites were elevated
did not significantly suppress pRPS6, implying limited downstream mod- only in the samples of IC. The MLR yielded higher AUC to discriminate IC from
ulation. Dosing selumetinib for 21 days had no added effect on pERK C; 0.790 (95% CI; 0.699 – 0.859, P , 0.0001). The ADTree with ensemble
isoforms. In both PTEN-null models, AZD8186 suppressed all pAKT iso- approach showed the best AUC; 0.912 (95% CI; 0.838 – 0.961, P , 0.0001).
forms by 40-75% up to 4 h post-dose. The pAKT remained suppressed at 7 h In the comparison of these metabolites in the analysis of each subtype, seven
in HCC70 but both pAKT and pRPS6 rebounded by 7 h in PC3. The basal metabolites were significantly different between Luminal A-like and Luminal B-
pERK1/2 levels were low and unaffected by AZD8186 in PC3 and HCC70. like while, but few metabolites were significantly different among the other
Conclusions: We established the fitness of PD assays to provide critical subtypes. Conclusions: These data indicated the combination of salivary
evidence regarding the duration of on-target effects, timeline of biomarker metabolomic profiles including polyamines showed potential ability to
reversal and effectiveness of pathway suppression for three distinct classes screening breast cancer in a non-invasive way.
of drugs. The PD assays are ready for an ongoing clinical trial of AZD8186
and to support clinical development of FGFR inhibitors. Funded by NCI
Contract HHSN261200800001E.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
182s Developmental Therapeutics and Tumor Biology (Nonimmuno)

3136 Poster Session (Board #128), Sat, 8:00 AM-11:00 AM 3137 Poster Session (Board #129), Sat, 8:00 AM-11:00 AM
The Cancer Molecular Screening and Therapeutics Program (MoST): Action- The whole genome landscape of adult metastatic sarcoma. First Author:
able mutation frequencies in a population with rare and less common Eric Yang Zhao, Canada’s Michael Smith Genome Sciences Centre, Vancouver,
cancers. First Author: Subotheni Thavaneswaran, Garvan Institute of BC, Canada
Medical Research, University of New South Wales (Faculty of Medicine),
Background: Metastatic sarcomas represent a heterogeneous, difficult to treat
Darlinghurst, NSW, Australia
family of cancers with poor median overall survival of 18 months. While global
Background: Personalizing therapy will arguably have no greater impact than on sequencing initiatives have catalogued genomic variation among primary
patients (pts) with rare (, 6 per 100,000 population) or less common cancers sarcomas, metastatic sarcoma is less well understood. Genome-guided tar-
(6-12/100,000). MoST combines a molecular screening platform and geted therapy has made promising advances but is difficult to study in sar-
biomarker-driven treatments for pts with advanced cancer, with a particular comas due to heterogeneity and low prevalence. Whole genome and
focus on rare and less common cancers (RLC). Methods: Molecular screening transcriptome analysis (WGTA) can help elucidate sarcoma oncogenesis,
was performed using in-house and commercial panels on archival tumor tissue. metastasis, and potential therapeutic targets. Methods: Using whole genome
A Molecular Tumor Board by consensus reported on pathogenic variants with (80X) and transcriptome (200M read) sequencing of 43 metastatic sarcomas
potential therapeutic actionability. Tiers of actionability were defined as: Tier across 19 subtypes, we analyzed structural variants (SV), copy-number variants
1–eligible for a MoST substudy; Tier 2–clinical evidence of efficacy in any (CNV), mutation signatures, gene expression, and the immune microenvi-
cancer type, Tier 3—preclinical evidence. The clinical and molecular char- ronment. All prior treatments were retrieved through chart review. Results: 17
acteristics of the first 1,000 pts are presented here. Results: Pts were recruited patients (40%) attempted WGTA-informed therapy, of which 8 (47%) were
from Sept 2016 to Dec 2018. A report was issued in 94% of cases in a median classified as responders. Metastatic sarcomas demonstrated recurrent CNVs,
of 7.7 weeks from consent. In 6%, there was insufficient tissue. The median age with 17p11-p12 amplification in 42% of cases. Some recurrent expression
at cancer diagnosis was 35 years (range 4-85 years), and 49% were male. Pts outliers were associated with potential targets (e.g. MYOCD, PMP22, COPS3)
had a median of 2 lines of prior systemic therapy (0-11), and a median baseline while others (e.g. ADORA2B) have not been previously observed in sarcoma.
ECOG performance status of 0 (range 0-3). 82% of pts had RLCs. A total of Discovery of oncogenic fusions refined diagnoses in two cases with atypical
2642 pathogenic variants were reported, of which 1144 (43%) were deemed histology. Clustering by mutation signatures distinguished histological sub-
therapeutically actionable. 651(57%) of actionable variants (AVs) occurred in types, and two signatures were novel in sarcoma: (1) a strong base excision
RLC (Table). Most commonly, AVs were found in the cell cycle, homologous repair signature associated with NTHL1 loss and (2) a cisplatin-associated
recombination repair (HR) and fibroblast growth factor (FGF) pathways. signature exclusive to platinum-treated cases. Frequent homologous re-
559(66%) of pts had at least one AV identified, 30% tier 1, 63% tier 2 and 6% combination deficiency was observed and was associated with response to
tier 3, including 66% of RLC. In 30% of cases, a tumor mutational burden .11 ifosfamide in three leiomyosarcomas. Of four immunotherapy-treated cases,
mutations/ megabase was reported. Conclusions: Here we report a high fre- the only responder demonstrated outlier CIBERSORT immune infiltration
quency of AVs in RLC, providing a rational basis for assessing the potential of score, which did not correlate with PD-L1 expression. Conclusions: This is the
personalized therapy in a population with a historically unmet need for effective first in-depth WGTA of metastatic sarcoma. We found recurrent and potentially
treatment. Clinical trial information: ACTRN12616000908437. targetable CNVs, expression outliers, mutation signatures, and immune
Variant counts by gene. markers. Our results suggest that clinical translation is promising using ac-
Cell
cycle HR
PTCH1/ HER2/ KIT/
SMO FGF PIK3CA MTOR EGFR MMR 3
AKT, AR,
PDGFR BRAF AR RET ALK ROS1 NF1 NTRK MET BCL2 EZH2, etc.
tionable insights obtained through WGTA.
Tier 1 2 3
RLC 242 180 6 78 51 33 37 31 40 25 20 0 4 7 4 30 2 13 9 88
(n=
688)
Common 48 58 2 26 22 11 7 10 15 10 3 1 3 1 4 24
(n=154)

3138 Poster Session (Board #130), Sat, 8:00 AM-11:00 AM 3139 Poster Session (Board #131), Sat, 8:00 AM-11:00 AM
Mutation and treatment profiles of patients with concurrent EGFR and ALK A prognostic 10-miRNA risk score (10-miRNA RS) in predicting neoadjuvant
actionable mutations (muts). First Author: Jun Zhao, Beijing University chemotherapy sensitivity of luminal breast cancer. First Author: Chang Gong,
School of Oncology Beijing Institute for Cancer Research, Beijing, China Breast Tumor Center, Sun Yat-sen Memorial Hospital, Sun Yat-sen Uni-
versity, Guangzhou, China
Background: Actionable muts in EGFR and ALK define two molecular sub-
types sensitive to EGFR-TKIs and ALK-TKIs, respectively. Although generally Background: The 10-miRNA risk score is a prognostic 10-gene expression
mutually exclusive, they did co-exist in some cases. However, when and how do signature specifically developed in luminal breast cancer associated with
they co-exist are not well understood. Methods: Pts with concurrent actionable relapse-free survival. Since high-risk patients identified by10-miRNA RS
muts in ALK and EGFR were selected from our database. Their mutation had worse prognosis but better outcome with chemotherapy than low-risk
profiles and treatment histories were analyzed. PFS was estimated using patients (Gong C et al, EBioMedicine. 2016), this model may facilitate
Kaplan-Meier method. Results: Among 341 ALK-positive (ALK-pos) and 3804 personalized therapy-decision making for luminal breast cancer patients.
EGFR-positive (EGFR-pos) pts, 9 (2.6% of ALK-pos, 0.2% of EGFR-pos) had Therefore, we seek to validate whether high-risk group are more sensitive to
concurrent EGFR and ALK actionable muts, including 3 EX19Indel + EML4- chemotherapy than low-risk group by assessing the predictive value of 10-
ALK, 2 EX19Indel + STRN-ALK, 2 L858R + L1152R, 1 L858R + EML4-ALK, miRNA RS for pathological complete response (pCR) in patients receiving
and 1 G719C + S768I + STRN-ALK. All 9 pts had lung cancer. One pt with neoadjuvant chemotherapy (NAC). Methods: The 10-miRNA gene expres-
EX19Indel + EML4-ALK was treatment naı̈ve. The other 8 pts have taken $ 1 sion and clinicopathological data were prospectively gathered from 251
EGFR-TKIs. The mPFS of these pts on first-generation EGFR-TKIs was 22 mo pretreated biopsy-diagnosed luminal breast cancer patients from 4 breast
(95% CI: 11 - NR). Except for 1 pt who progressed on Gefitinib and sub- cancer centers. Formalin-fixed paraffin-embedded tissues from basal line
sequently on Osimertinib had a T790M+C797G, the other 7 EGFR-TKI re- biopsy were used for the detection of 10-miRNA expression to calculate the
sistance pts had no common known resistance muts. 3 pts ordered NGS tests RS. The correlation between pCR and the 10-miRNA RS classification were
before taking EGFR-TKIs. None of them had ALK muts at that time. Later, 1 pt identified. Results: In this prospective, multicenter study, the overall pCR
(19Indel) gained an STRN-ALK after 15 mo on Osimertinib, 1 pt (L858R) rate was 13.6% (34/251). The 10-miRNA RS of the pCR group was sig-
gained an EML4-ALK after 5 mo on Gefitinib, and 1 pt (L858R) gained an nificantly higher than the non-pCR group (P = 0.015). Fifty-one percent of
L1152R after 10 mo on Afatinib. Therefore, ALK muts were likely developed as patients were classified as low-risk according to the 10-miRNA RS classi-
resistance mechanisms during EGFR-TKIs therapies in these 3 pts. Un- fication and 49% as high-risk with a RS cut-off point of 2.144. The 10-
fortunately, with no information on ALK status before EGFR-TKI therapies, we miRNA RS classification was associated with a pCR rate of 9.4% in the low-
can not tell if the ALK muts were also developed during and conferred re- risk group and 17.8% in the high-risk group (P = 0.041). The correlation
sistance to EGFR-TKI therapies in the other 5 pts. Both STRN-ALK and ALK between the pCR and the 10-miRNA RS classification was significant in
L1152R were recorded 4 times in our database, and they concurred with EGFR subgroup analysis stratified by molecular subtypes (8% vs. 13.2% in luminal
actionable muts in 3 and 2 of the 4 records, respectively. Conclusions: ALK B1; 14.7% vs. 30.1% in luminal B2; no pCR was observed in all 13 luminal A
and EGFR actionable muts concurred at a relatively low frequency in our pts. In subtype). In multivariate analysis, the 10-miRNA RS remained significantly
some cases, ALK muts were developed during EGFR-TKI therapies. Developed associated with pCR and independent from subtype, ki67 and other clini-
either together or sequentially, some combinations of EGFR and ALK muts, copathological characteristics. Conclusions: 10-miRNA RS clearly defined
such as L858R with L1152R and EX19Indel with ALK fusion, may form more that high-risk patients are more sensitive to chemotherapy which leads to a
easily or may be preferable than other combinations for the development or higher pCR rate in NAC patients. Thus, 10-miRNA RS is not only a prognostic
evoluation of tumors. factor but an effective method in determining whether a patient would
undergo surgery or receive NAC prior to surgery. Clinical trial information:
ChiCTR-DDD-17013651.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Therapeutics and Tumor Biology (Nonimmuno) 183s

3140 Poster Session (Board #132), Sat, 8:00 AM-11:00 AM 3141 Poster Session (Board #133), Sat, 8:00 AM-11:00 AM
Prediction of biomarker status, diagnosis and outcome from histology slides MET exon 14 skipping analogs: Rare but potentially clinically actionable.
using deep learning-based hypothesis free feature extraction. First Author: First Author: Rebecca Feldman, Caris Life Sciences, Phoenix, AZ
Eldad Klaiman, Roche Pharma Reseach and Early Development, Roche
Background: The DpY motif within the exon 14 juxtamembrane domain of the
Innovation Center Munich, Penzberg, Germany
MET receptor gene is critical for Cbl-mediated negative regulation. Splicing
Background: Recently, histological pattern signatures obtained from diagnostic alterations that delete this residue, known as exon 14 skipping mutations
H&E images have been found to predict mutation, biomarker status or outcome. (ex14sk mt), lead to prolonged MET protein stability and oncogenic signaling.
We report here on a novel deep learning based framework designed to identify Specific mt at the Y1021 (aka 1003) residue are thought to lead to similar
and extract predictive histological signatures. We have applied this framework in effects as ex14sk, but due to their rarity, their role in NSCLC is unknown. We
3 experiments, predicting specifically the microsatellite status (MSS) of co- sought to identify and characterize non-ex14sk mt that include/surround
lorectal cancer (CRC), breast cancer (BC) micrometastasis in Lymph nodes Y1021. Methods: Retrospective review of molecular profiles for non-ex14sk
(LN) and Pathologic Complete Response (pCR) in BC diagnostic biopsies. mt that include/surround the DpY motif (Y1021) in MET. Two NGS platforms
Methods: Our deep learning based algorithm was trained on histology images at were included: MiSeq (2014-2017; n=2865) and NextSeq (2017-2019;
20X magnification. Algorithms were trained for binary classification for each of n=6084). Immunohistochemistry (IHC) of cMET (SP44) and co-occurring al-
the three cohorts. We used 75% of the images for training and test our algorithm terations (EGFR, KRAS, ALK, ROS) were also reviewed. Results: Of 8,949
on the remaining 25% of the images. Cohort details are as follows: MSS for CRC: NSCLC patients with successful NGS of MET gene by either platform, 13 cases
94 patients’ H&E stained tissue images from the Roche internal CRC80 dataset or 0.2% were identified to have an alteration within the amino acids of interest.
(MSS n =24; MSI n = 70) were used. BC LN: 270 patients’ H&E stained tissue Eleven cases included substitutions at Y1021 (5 phenylalanine, 4 histidine and
images from the CAMELYON16 dataset ( LN(+) n = 110 ; LN(-), n =160) were 3 arginine) and the remaining two cases included small insertion-deletions
used. pCR for BC: 225 patients’ H&E stained tissue images from the Tryphaena p.E1017_Y1021delinsH and p.D1020_Y1021delinsV, the latter was later
Study BO22280, neoadjuvant, Trastuzumab/Pertuzumab chemotherapy com- excluded as it co-harbored an ex14sk mt. Conclusions: Similar to patients with
bination trial. (pCR=111, non-pCR n=114). Results: We report and assess al- ex14sk mt, substitutions and small indels at Y1021 exhibit Clinicopathological
gorithm performance on each of the cohorts by Area Under the Curve (AUC). features such as previous smoking history and older age, mutual exclusivity with
Prediction of MSS in the CRC80 status yielded AUC 0.9. Prediction of LN oncogene drivers and MET protein overexpression. The rarity of these analogous
invasion on CAMELYON16 dataset yielded AUC 0.85. Prediction of pCR on the ex14sk mt suggests deletions of exon 14 provide cellular advantages beyond
Tryphaena cohort yielded an AUC of 0.8. Conclusions: We present a new ap- Cbl-mediated ubiquitinylation of MET. Although rare, the impact of these mt on
proach to generate predictive signatures based on conventional diagnostic H&E efficacy of Met-directed therapy deserves further exploration.
images and a novel machine learning framework. The CRC80 and CAMELYON16 Case Smoking Hx Sex/Age Histology Stage protein change cMET IHC Staining Pattern
cohorts served as a confidence building experiments with predictive features well
1 M/82 adc p.Y1021F 2+ 90%
known by clinicians and visually confirmed. The predictive algorithm for pCR in 2 Former F/84 adc III “ 2+ 10%
the Tryphaena cohort yielded both response prediction and the high predictive 3 F/85 adc IV “
4 F/80 adc IV “ 3+ 100% membranous/cytoplasmic
value FOVs. These included tissue patterns which have not until now been 5 M/83 other IV “ 2+ 100% membranous/cytoplasmic
considered to influence on the prediction of pCR. 6 F/71 adc p.Y1021H 2+ 5%
7 F/77 adc “ 3+ 100%
8 Former M/79 mixed IV “ 3+ 100% membranous
9 Former M/81 adc IV “
10 Former M/83 adc IV p.Y1021N 3+ 100%
11 Former F/69 adc IB “ 1+ 100% cytoplasmic
12 Never F/84 adc p.E1017_Y1021delinsH 2+ 80% membranous

3142 Poster Session (Board #134), Sat, 8:00 AM-11:00 AM 3143 Poster Session (Board #135), Sat, 8:00 AM-11:00 AM
Effects of immune architecture on response to adjuvant capecitabine in Vascular endothelial growth factor A (VEGF-A) amplification and long-term
triple-negative breast cancer (FinXX trial). First Author: Saranya Chumsri, response to ramucirumab (ram) in metastatic gastric cancer (mGC): The
Mayo Clinic, Jacksonville, FL VERA study. First Author: Alessandra Raimondi, Medical Oncology De-
partment, Fondazione IRCCS Istituto Nazionale dei Tumori, Milan, Italy
Background: Recent studies have demonstrated a benefit of adjuvant cape-
citabine, particularly in triple negative breast cancer (TNBC) patients with Background: The anti-VEGFR-2 monoclonal antibody ram, alone or with
residual disease after neoadjuvant chemotherapy. However, biomarkers to paclitaxel, is a cornerstone of second-line treatment of mGC. Even if about
predict which patients are more likely to benefit from capecitabine are needed. half patients do not benefit from ram, no predictive biomarkers have been
Methods: The nanoString Breast Cancer 360 (BC360) and PanCancer identified so far. In TCGA, VEGF-A amplification was found in 7% of cases,
Immunoncology (IO360) panels were used to quantify mRNA expression in almost exclusively in chromosomal instability subtype. We hypothesize that
TNBC samples in the FinXX trial. FinXX is a phase III trial which randomized VEGF-A amplification in tumor cells could lead to autocrine/paracrine
high risk patients to receive either 3 cycles of docetaxel followed by 3 cycles of stimulation of tumor growth beside angiogenesis, potentially identifying a
cyclophosphamide, epirubicin, and fluorouracil (Arm A: T+CEF) vs. 3 cycles of patients’ subgroup with exceptional responses to ram. Methods: VERA was a
docetaxel plus capecitabine followed by 3 cycles of cyclophosphamide, epi- multicentric, prospective study based on a translational hypothesis. mGC
rubicin, and capecitabine (Arm B: TX+CEX). Gene signature scores were patients were included according to the following criteria: 1) complete (CR)
analyzed using prespecified algorithms developed by nanoString. Digital or partial response (PR) to single-agent ram; 2) .6 months PFS to single-
Spatial Profiling was carried out using GeoMX platform. Results: A total of 111 agent ram; 3) .10 months PFS to paclitaxel+ram. According to a Fleming
TNBC patients in FinXX trial were included (57 in Arm A and 54 in Arm B) with single-stage design, hypothesizing a prevalence of VEGF-A amplification of
10.2 years median follow up. There were 7 cancer- and immune-related gene 1% and 15% among all-comers and exceptional responders, 20 exceptional
signatures identified by BC360 and IO360 panels that were significantly responders were required to reject the null hypothesis of low prevalence of
associated with improved recurrent free survival favoring an addition of VEGF-A amplification, with alpha- and beta- errors of 0.05 and 0.10, re-
capecitabine. These include cytotoxic cell signature (HR 0.37, 95%CI 0.15- spectively. VEGF-A amplification (defined as .10% tumor cells with $10
0.92, p 0.03), endothelial signature (HR 0.18, 95%CI 0.04-0.83, p 0.03), VEGF-A copies, variably sized signal clusters or a ratio of VEGF-A gene to
mast cell signature (HR 0.43, 95%CI 0.21-0.88, p 0.02), PDL2 gene (HR centromere of $2) was centrally assessed through fluorescent in situ hy-
0.29, 95%CI 0.09-0.99, p 0.05), immunoproteasome (HR 0.34, 95%CI bridization on pre-treatment FFPE tumor tissue. Results: At 7 Italian Cen-
0.13-0.89, p 0.02), exhausted CD8 (HR 0.29, 95%CI 0.09-0.97, p 0.04), ters, we included 20 patients satisfying the 1st (n=1), 2nd (n=2), or 3rd
and PD1 (HR 0.44, 95%CI 0.20-1.02, p 0.05). Conclusions: Analysis of RNA (n=17) criterion. Clinical-pathological features were: M/F, 11/9; median age
abundance signatures strongly suggests that there are important immune 63 years; gastric/GEJ, 17/3; intestinal/diffuse, 14/6, HER2+/HER2-, 4/16.
features that are associated with benefit from capecitabine in TNBC. However, Median PFS and overall survival to ram-based treatment were 15.6 and
analysis of RNA extracted from whole tumor sections lacks spatial discrimi- 25.7 months, with best response: CR/PR/SD, 0/10/10. VERA met its primary
nation. We anticipate that a more detailed, spatially-defined analysis of protein endpoint, revealing 3/20 (15%) tumors with VEGF-A amplification (1 case
abundance, using the novel NanoString GeoMX platform, will provide more presenting big clusters, 1 small clusters and 1 with .10% tumor cells
insights and define specific immune features associated with improved out- with $10 VEGF-A copies). Conclusions: Validation analyses of first- and
come. Additional results of GeoMX will be reported at the meeting. Clinical trial second-line randomized trials could confirm VEGF-A amplification as a
information: NCT00114816. biomarker of long-term response to ram-based treatment in mGC patients,
advancing treatment personalization.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
184s Developmental Therapeutics and Tumor Biology (Nonimmuno)

3144 Poster Session (Board #136), Sat, 8:00 AM-11:00 AM 3145 Poster Session (Board #137), Sat, 8:00 AM-11:00 AM
Reassignment of HER2 status for subgroups of breast cancer according to BRCA1 genetic variant to predict survival in metastatic colorectal cancer
the 2018 updated American Society of Clinical Oncology and College of (mCRC) patients (pts) treated with FOLFIRI/bevacizumab (bev): Results
American Pathologists guidelines: The impact of combined immunohisto- from phase III TRIBE and FIRE-3 trials. First Author: Madiha Naseem,
chemistry (IHC) and fluorescence in situ hybridization (FISH) reflex testing Division of Medical Oncology, USC Norris Comprehensive Cancer Center,
in a large national reference laboratory. First Author: Katherine Geiersbach, Keck School of Medicine, Los Angeles, CA
Mayo Clinic, Rochester, MN
Background: BRCA muts in CRC are associated with a higher tumor mutation
Background: Updated ASCO/CAP Guidelines for HER2 testing in breast burden irrespective of microsatellite instability, which highlights the possibility
cancer have been most impactful on the resolution of certain challenging of using PARP-inhibitors(i) in CRC in the future. Early phase studies have
groups of FISH results. We review the change in assignment of HER2 status shown that combination of PARP-i with oxaliplatin or irinotecan enhances
in a large series of breast cancers referred to a large national reference lab- tumor lysis in CRC. In this study, we investigated the influence of mutations in
oratory for FISH testing since the introduction of the 2018 updated guidelines. the Homologous Repair Pathway genes on survival outcomes among mCRC pts
Methods: Patient samples submitted to the Mayo Clinic Cytogenetics Labo- treated with oxaliplatin or irinotecan-based regimens. Methods: The impact of
ratory (N = 2208) were analyzed by FISH. Samples with Group 2, Group 3, or selected SNPs within 4 genes (BRCA1, BRCA2, RAD51, BARD1) on OS/PFS
Group 4 FISH results were reflexed to immunohistochemistry (IHC) in our was analyzed through the OncoArray, a custom array manufactured by Illu-
central laboratory; FISH slides for those cases with equivocal 2+ IHC results mina, on genomic DNA from blood samples of 431 pts enrolled in 2 ran-
were re-scored in the regions of invasive cancer showing more intense mem- domized trials. TRIBE FOLFIRI/bev arm (n = 215, mPFS/OS: 9.7/26.2 mo)
branous staining. A subset of 202 samples with Group 4 FISH results were also served as discovery cohort, FIRE-3 FOLFIRI/bev arm (n = 107, mPFS/OS:
reflexed to the previously employed reflex FISH assay (HER2/D17S122), and 11.5/31.4 mo) as validation and TRIBE FOLFOX/bev arm (n = 109, mPFS/OS:
these were also re-analyzed according to the new reflex IHC/FISH process. 10.8/26 mo) as control. Results: Significant associations were found among
Results: 382 of 2208 breast cancer samples tested (17.3%) had FISH results carriers of BRCA1 rs8176318 SNP, where C . A base change is known to
categorized as Group 2 (N = 17, 0.8%), Group 3 (N = 34, 1.5%), or Group 4 (N = reduce BRCA1 expression among CRC cells. In the discovery cohort, pts with A/
331, 15%) and required reflex IHC testing, and of those, 75% were 2+ equivocal A had shorter OS (22.4 vs 27.3 mo, P = .009) and PFS (7.5 vs 10.5 mo, P =
and required targeted re-analysis of the FISH slide according to the 2018 .0006) compared to carriers of any C allele in both univariate and multivariate
updated guidelines. Re-analysis of the FISH slide resulted in switching between analysis. Same results were observed in pts with left-sided CRCs (PFS-7.5 vs
Groups 1-5 in 19.4% of cases, but HER2 status was changed by FISH re-scoring 11 mo, P = .005; OS- 25.6 vs 32.3, P = .034) and among males (PFS- 7.5 vs
in only 7.7% of cases re-scored (1.0% of all samples), generally due to only minor 10.3 mo, P = .008; OS- 25.7 vs 31.3 mo, P = .008) in both uni and mul-
shifts in HER2 copy number and HER2/control ratios between the initial and tivariate analysis. These results were also seen in the validation cohort: A/A
IHC-guided reflex FISH scores. In the subset of 202 cases tested by both reflex carriers in left-sided CRCs had poor OS (26.1 vs 36.0 mo, P = .027) and PFS
methods, the previously employed HER2/D17S122 reflex probe set was positive (9.5 vs 11.7 mo, P = .002. Males with A/A genotype also had poor OS (24.7 vs
in 123 cases (60.9%), whereas reflex IHC/FISH was positive in only 10 cases 32.5 mo, P = .028) and PFS 96.9 vs 12.2 mo; P = .0002). In the control
(7.9%). Including positive reflex IHC (0.4%) and positive reflex FISH results cohort, A/A genotype carriers had poor tumor response in overall (P = .011) and
(2.1%), the overall assignment of positive HER2 status on our series of 2208 left-sided disease (P = .034). These outcomes were independent of KRAS
cases was 11.5%. Conclusions: Overall rates of HER2 positive FISH results have mutation status. No significant relationship was observed among females with
declined under the most recent ASCO/CAP guideline update as a consequence of mCRC. Conclusions: This is the first study to report that BRCA1 mut influence
new recommendations for reflex testing for Groups 2-4. This change is largely due survival outcomes among mCRC pts, particularly among males and those with
to reassignment of Group 2 and Group 4 results as negative in the absence of left-sided disease. Prospective trials are warranted to assess the utility of
positive IHC. routine BRCA mut testing and the role of PARP-i in improving survival out-
comes in this pt population.

TPS3146 Poster Session (Board #138a), Sat, 8:00 AM-11:00 AM TPS3147 Poster Session (Board #138b), Sat, 8:00 AM-11:00 AM
First-time in-human study of VMD-928, an allosteric and irreversible TrkA A first-in-human study of, NUC-7738, a 3’dA phosphoramidate, in patients
selective inhibitor, in patients with solid tumors or lymphoma. First Author: with advanced solid tumors or lymphoma (NuTide 701). First Author:
Vincent Chung, City of Hope, Duarte, CA Hagen P Schwenzer, University of Oxford, Oxford, United Kingdom
Background: Tropomysin receptor kinase A (TrkA) is a protein encoded by the Background: Nucleoside analogs form the backbone therapy for both he-
NTRK1 gene. NTRK fusions involving the kinase domain are oncogenic for matological and solid malignancies. However, their clinical effectiveness is
multiple tumor types and larotrectinib was recently approved for advanced severely limited by key cellular resistance mechanisms linked to increased
solid tumors harboring NTRK gene fusions. Larotrectinib, an ATP-competitive, breakdown, impaired activation and transport. NUC-7738 is a phosphor-
reversible pan-TrkA/B/C inhibitor, has shown impressive response rates in amidate transformation of cordycepin (3’-deoxyadenosine; 3’-dA), a de-
patients harboring these fusions; however, resistance can develop due to ac- rivative of adenosine that was first isolated from Cordyceps sinensis. The
quired ATP-site mutations. This has been previously identified in other onco- cytotoxic effect of 3’-dA is largely attributed to intracellular generation of the
genic driver kinases such as ALK and EGFR treated with ATP-competitive triphosphate metabolite, 3’-dATP, terminating DNA and RNA synthesis.
kinase inhibitors. A newly approved allosteric ALK/EGFR inhibitor brigatinib was Although 3’-dA has shown potent anti-tumor activity in non-clinical studies,
able to clinically overcome acquired resistance of many ATP-competitive ALK/ it has not been successful in clinical studies mainly because of rapid en-
EGFR inhibitors (1). Also, irreversible EGFR inhibitors such as afatinib (ATP- zymatic degradation by adenosine deaminase. NUC-7738 is not a substrate
competitive) were active against tumors resistant to first-generation inhibitors for adenosine deaminase and has been designed to bypass the key resistance
(2), although their efficacy can be compromised by acquired ATP-site mutations pathways which have limited the clinical effectiveness of cordycepin.
(3). VMD-928 is the first oral small-molecule TrkA (NTRK1) selective inhibitor Methods: NuTide:701 is a two-part, first-in-human Phase I study in patients
with dual allosteric and irreversible mechanisms of action. It inhibits TrkA non- with advanced solid tumors and lymphoma who have exhausted all standard
competitively at an allosteric (non-ATP) site and has no resistance in vitro to treatment options. The primary objective is to determine the RP2D and
acquired ATP-site mutations such as G667C. VMD-928 in vitro has little or no schedule of NUC-7738. Secondary objectives include safety, PK/PD and
activity against 348 other kinases including TrkB (NTRK2) and TrkC (NTRK3). anti-tumor activity. Part 1, in patients with advanced solid tumors, will
We are conducting the first time in human phase 1 trial of oral VMD-928, a novel establish the RP2D and dose administration schedule of NUC-7738 for Part
allosteric and irreversible TrkA selective inhibitor. Methods: This is an open 2. Part 2 will further evaluate the selected RP2D and designated dosing
label, Phase 1 study investigating the safety, pharmacokinetics (PK) and schedule in an expansion cohort of patients with advanced solid tumors or
pharmacodynamics (PD) of oral VMD-928 in adults with advanced solid tumors lymphoma. The study initiated in Q1 2019. Clinical trial information:
or lymphoma (NCT03556228). In part 1 of the study, an accelerated titration NCT03829254.
scheme will be utilized to determine the recommended phase 2 dose and
evaluate PK / PD of VMD-928. In part 2, expansion cohorts including patients
with thymic, pancreatic, triple-negative breast carcinoma, or solid tumors with
TrkA alterations will be accrued to further evaluate safety and efficacy. Part 3 of
the study will characterize the biologically active dose. The study is open and
accruing patients at City of Hope. Clinical trial information: NCT03556228.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Therapeutics and Tumor Biology (Nonimmuno) 185s

TPS3148 Poster Session (Board #139a), Sat, 8:00 AM-11:00 AM TPS3149 Poster Session (Board #139b), Sat, 8:00 AM-11:00 AM
Trial in progress abstract phase I trial of 5-aza-4’-thio-2’-deoxycytidine (Aza- Update on the Drug Rediscovery Protocol: Expanded use of existing anti-
TdC) in patients with advanced solid tumors. First Author: M. Cecilia Monge B., cancer drugs in patients with a known molecular profile. First Author:
National Cancer Institute, National Institutes of Health, Bethesda, MD Jade Maxime van Berge Henegouwen, Leiden University Medical Center,
Department of Medical Oncology, Leiden, Netherlands
Background: The nucleoside analog 5-aza-4’-thio-2’-deoxycytidine (Aza-
TdC) inhibits DNA methyltransferase 1 (DNMT1), a methyltransferase in- Background: With the emergence of large-scale genetic tumor profiling and
volved in methylation-mediated silencing of tumor suppressor genes. At- the increasing availability of approved targeted therapies, precision medi-
tenuation of DNA methylation via DNMT1 inhibitors results in reactivation of cine has become crucial in cancer treatment. However, for many cancers the
silenced tumor suppressor genes and can lead to tumor growth arrest and relative contribution of either tumor type or genetic aberration to drug
apoptosis. The DNMT1 inhibitors decitabine and 5-azacytdine are currently sensitivity often remains unknown. Since drug access is generally limited to
FDA-approved for use in myelodysplastic syndromes and are also used in the on-label indication and outcome of off-label use is not systematically
patients with acute myeloid leukemia. Relative to these compounds, Aza- collected in clinical practice, innovative trials facilitating drug access,
TdC exhibits enhanced stability and incorporation into DNA and has shown whilst systematically analyzing treatment outcomes, are urgently needed.
improved preclinical antitumor activity in both leukemia and solid tumor Methods: The Drug Rediscovery Protocol (DRUP) is an ongoing, prospective,
xenograft models. This study seeks to evaluate the safety and maximum non-randomized, multi-drug, and pan-cancer trial, in which patients with
tolerated dose (MTD) of oral Aza-TdC in patients with advanced solid tumors. advanced cancer, who have exhausted all standard of care treatment op-
Secondary study objectives include assessing objective response by RECIST tions, are treated with either targeted or immunotherapy matched to their
1.1, pharmacokinetic (PK) analysis, and examining re-expression of tumor genetic tumor profile. All submitted patients are reviewed and enrolled in
suppressor genes inhibited by methylation in circulating tumor cells (CTCs). multiple parallel cohorts, preceded by a baseline tumor biopsy for whole
Methods: Patients are treated with Aza-TdC on days 1-5 and 8-12 of each genome sequencing to confirm previously identified variants and for ex-
21-day cycle. The study follows Simon accelerated titration design 3, with ploratory biomarker analyses. Each cohort is defined by a study drug, his-
100% dose increments and 1 patient per dose level. Accelerated titration tologic tumor type, and molecular tumor profile. Efficacy is analyzed per
will continue until 1 patient experiences a dose-limiting toxicity (DLT) or 2 cohort: 8 patients in stage I and 16 more in stage II if $ 1 response is
patients experience drug-related grade 2 toxicity at any dose level, after observed in the first stage. Primary endpoints include objective response
which, a 3 + 3 dose escalation design will be used. Blood samples are rate, stable disease at 16 weeks, and grade $3 adverse events. Since the
collected for PK and CTC analyses. An MTD expansion cohort is planned, in start of recruitment in September 2016, 870 patients have been submitted
which tumor biopsies will be collected for further pharmacodynamic as- for review and 365 patients (42%) have started treatment in one of 101
sessments. Patients included in this study must be $18 years old and have opened cohorts. Eight cohorts have graduated to the second stage, two
histologically documented solid tumors that have progressed on standard cohorts completed accrual in either their first or second stage, and one
therapy and for which there is no other standard therapy available. Dose level cohort was closed due to a registered indication. Twenty-two different study
3 has been completed without any DLTs; enrollment to dose level 4 began in treatments (i.e. immunotherapy, monoclonal antibodies, and PARP/small
February 2019. Funded by NCI Contract No. HHSN261200800001E. molecule inhibitors), provided by 11 different pharmaceutical companies,
Clinical trial information: NCT03366116. are currently available in DRUP. Data sharing with similar trials such
as TAPUR and CAPTUR enables to achieve completion of slow accruing
cohorts and affirm conclusions. Clinical trial information: NCT02925234.

TPS3150 Poster Session (Board #140a), Sat, 8:00 AM-11:00 AM TPS3151 Poster Session (Board #140b), Sat, 8:00 AM-11:00 AM
A phase I clinical study to evaluate the safety, tolerability, pharmacokinetics Basket of baskets (BoB): A modular, open label, phase II, multicenter study
(PK), and antitumor activity of FN-1501 monotherapy in patients with to evaluate targeted agents in molecularly selected populations with ad-
advanced solid tumors. First Author: Gary Edward Richardson, Monash vanced solid tumors. First Author: Irene Brana, Vall d’Hebron Institute of
University, Cabrini Hospital, Malvern, Australia Oncology, Barcelona, Spain
Background: Receptor tyrosine kinases (RTK), a group of transmembrane Background: Basket trials with targeted agents can show high response rates
proteins, are responsible for growth factor signaling transduction in normal for tumors with specific molecular profiles, granting extension of the label of
cellular functions. Abnormal RTK functions are associated with human some drugs. In other cases, study results were disappointing, likely due to the
tumorigenesis. FMS-like tyrosine kinase 3 (FLT3) belongs to the type III rarity of molecular alterations, limits in trial design and the difficulties in
receptor tyrosine kinase family and plays a well-established role in normal applying molecular tumor profiling in the clinical setting. Methods: Basket of
growth and differentiation of hematopoietic precursor cells. FLT3 mutations Baskets (BoB), NCT03767075, aims to bridge the gap between Academic
have been reported to occur in approximately 30% newly diagnosed AML Genomics and clinical applications (ready-to market multi-marker Companion
patients. The internal tandem duplications mutation (FLT3/ITD) is the major Diagnostics) by providing a sustainable and adaptable (to new technologies,
mutation and correlated with more aggressive progress and poor prognosis. markers, and therapeutic agents) platform for co-development of drug/
FN-1501 is an inhibitor of various tyrosine kinases such as cyclin-dependent companion diagnostic. BoB is a novel platform trial from Cancer Core
kinase 4/6(CDK4/6), platelet-derived growth factor receptor (PDGFR), KIT Europe, a recently established sustainable European network for innovative
protein, anaplastic lymphoma kinase (ALK) and RET protein, particularly cancer research. This protocol has two parts: (1) Part A includes a molecular
potent on FLT3. The preclinical data generated from biochemical, cell based profiling program for subjects with advanced solid tumors (iPROFILER), a
and animal in vivo studies suggest that FN-1501 as a single agent could offer variant annotation tool, and a molecular tumor board to select the most ap-
cancer patients clinical benefit by inhibiting multiple tyrosine kinases in- propriate treatment. It also enables testing/developing companion diagnostics
cluding FLT3, PDGFR, KIT, ALK, and RET. Methods: This is a Phase1, open linked with the therapeutic part (part B). (2) Part B includes iBASKET, a
label, multicenter, dose-escalation study that will evaluate the safety, modular multi-arm basket trial for subjects with tumors harboing selected
pharmacokinetics (PK), and preliminary efficacy of FN-1501 in up to 33 molecular alterations. Each module is focused on a certain molecular pathway
cancer patients with solid tumors. There is a dose escalation phase that will or on certain molecular alterations that may confer sensitivity to the study drug
be followed by an expansion cohort. The dose escalation phase utilizes a or study drug combination evaluated in that module/arm. The current version of
standard “3 + 3” design where doses of FN-1501 will be escalated up to the iBASKET (Module 1- Atezolizumab in genomically-selected patients) is open
Maximum-Tolerated Dose (MTD) or until the Recommended Phase 2 dose for enrollment for patients with advanced neoplasms bearing one of the fol-
(RP2D) is identified. Once the MTD or RP2D dose is identified, an expansion lowing alterations: Arm 1A: BRCA1 or BRCA2; Arm 1B: MLH1, MSH2, MSH6,
cohort including patients with hematologic malignancies will be enrolled to PMS2; Arm 1C: POLE, POLD1 mutations; Arm 1D: hypermutated tumors; Arm
further evaluate the safety and efficacy of FN-1501. Key exploratory ana- 1E: other mutations in DNA-repair genes; Arm 1F: PDL1 gene amplification.
lyses will include an evaluation of safety and efficacy and levels of expression All patients enrolled in Module 1 will receive single-agent atezolizumab.
and/or amplification of FLT3 mutations. As of February 8, 2019, cohorts 1 New Modules for genomically selected populations can be added through
and 2 have been completed without a dose limiting toxicity (DLT). A total of amendments. Our final aim is to achieve drug repurposing of treatments, co-
11 patients have been treated. Enrollment to cohort 3 is on-going. Clinical develop multi-marker companion diagnostics and a large database of
trial information: NCT03690154. knowledge in Precision Medicine. Clinical trial information: NCT03767075.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
186s Developmental Therapeutics and Tumor Biology (Nonimmuno)

TPS3152 Poster Session (Board #141a), Sat, 8:00 AM-11:00 AM TPS3153 Poster Session (Board #141b), Sat, 8:00 AM-11:00 AM
A phase I study of [225Ac]-FPI-1434 radioimmunotherapy in patients with TROPHY-U-01: A phase II open-label study of sacituzumab govitecan
IGF-1R expressing solid tumors. First Author: Rosalyn A. Juergens, (IMMU-132) in patients with advanced urothelial cancer after progression
Juravinski Cancer Centre, McMaster University, Hamilton, ON, Canada on platinum-based chemotherapy and/or anti-PD-1/PD-L1 checkpoint in-
hibitor therapy. First Author: Scott T. Tagawa, Sandra and Edward Meyer
Background: Type I insulin-like growth factor receptor (IGF-1R) is a trans-
Cancer Center, New York, NY
membrane protein which is overexpressed in solid tumors including non–
small cell lung, prostate, and breast cancers. [225Ac]-FPI-1434 is a radio- Background: Patients (pts) with advanced urothelial cancer (UC) who progress
immunoconjugate consisting of a humanized monoclonal antibody that binds to after checkpoint inhibitor (CPI) therapy (following failure of or ineligibility for
the external domain of IGF-1R, a proprietary bifunctional chelate, and an alpha- platinum-based chemotherapy) have limited options. Trop-2 is an epithelial
emitting radionuclide actinium-225 (Ac-225), which binds to the external cell surface antigen overexpressed in UC (Avellini. Oncotarget 2017). Saci-
domain of IGF-1R. Internalization of the conjugate and decay of Ac-225 causes tuzumab govitecan (SG) is an antibody-drug conjugate that targets Trop-2 and
tumor cell death primarily through double stranded DNA breaks. The indium- delivers the active metabolite SN38 of the topoisomerase I inhibitor irinotecan
111 analog, [111In]-FPI-1547, with the identical antibody and bifunctional to tumor cells (Starodub. Clin Cancer Res 2015). In a phase 1/2 trial, pts with
chelate is used for patient selection, in-vivo imaging, and quantification of IGF- advanced cancers received SG on days 1 and 8 of a 21-day cycle. In the UC
1R targets prior to therapy. Based on anti-tumor activity of [225Ac]-FPI-1434 in cohort, 45 evaluable pts received SG 10 mg/kg with a median of 2 (range 1–6)
preclinical models, favorable toxicology studies in cynomolgus monkeys, and prior therapies. Objective response rate (ORR) was 31%; median duration of
prior human experience with the unconjugated antibody, the first in human response was 12.9 mo. Grade $3 adverse events in $5% of pts were
clinical evaluation was initiated. Methods: This open-label multi-center phase I neutropenia/neutrophil count decreased (38%), anemia (13%), hypo-
study (NCT03746431) follows a modified 3+3 dose-escalation design to phosphatemia (11%), diarrhea (9%), fatigue (9%), and febrile neutropenia
characterize the safety profile, determine a maximum tolerated dose (MTD), (7%). Median progression-free survival (PFS) was 7.3 mo and overall sur-
evaluate dose-limiting toxicities (DLT), describe pharmacokinetics, derive ra- vival (OS) 16.3 mo (Tagawa 2019 ASCO Genitourinary Cancers Symposium).
diation dose estimates to normal organs, and evaluate the objective response These results warrant further investigation in a dedicated phase 2 trial.
rate of [225Ac]-FPI-1434 therapy in patients with IGF-1R expressing solid Methods: TROPHY-U-01 (NCT03547973) is a single-arm, global phase 2 trial
tumors. Eligibility requirements for therapy include: presence of at least one evaluating the antitumor activity of SG (10 mg/kg on days 1 and 8 of a 21-day
measurable lesion as determined by sufficient tumor uptake using SPECT/CT of cycle) in 140 pts with advanced UC and measurable disease. Patients are also
an imaging analog [111In]-FPI-1547; radiation dose estimates of the planned required to have an Eastern Cooperative Oncology Group Performance Status
therapeutic activity within prespecified limits; and adequate bone marrow score of 0 or 1 and creatinine clearance $30 mL/min. The pivotal cohort
reserves, hepatic, and renal function. Dose cohorts begin with 10 kBq [225Ac]- (Cohort 1: progression after both platinum chemotherapy and CPI) will enroll
FPI-1434 per kilogram (kg) patient weight and successively increase to 20, 40, 100 evaluable pts in a Simon 2-stage design with . 90% power accounting for
80, and 120 kBq/kg as a single intravenous injection per patient followed by an dropouts to exclude the null hypothesis or ORR , 12%; an exploratory cohort
8-week DLT evaluation period. This trial is currently enrolling patients. Clinical (Cohort 2: 40 pts) includes platinum-ineligible pts who progress after prior CPI.
trial information: NCT03746431. The primary objective is ORR per RECIST 1.1, assessed by central review.
Secondary objectives include response duration, PFS, and OS. Adverse events,
pharmacokinetics, and tissue correlates will also be assessed. Enrollment
began August 2018. Clinical trial information: NCT03547973.

TPS3154 Poster Session (Board #142a), Sat, 8:00 AM-11:00 AM TPS3155 Poster Session (Board #142b), Sat, 8:00 AM-11:00 AM
A phase II study for prostate cancer monitoring using 18F-DCFPyL and blood- Clinical Trial in Progress: The FLEX Big Data Platform explores new gene
based biomarkers. First Author: Emerson A. Lim, Columbia University- expression profiles and investigator-initiated protocols in early-stage breast
Herbert Irving Comprehensive Cancer Center, New York, NY cancer. First Author: Sarah Untch, Agendia, Irvine, CA
Background: Assessing treatment response in castrate resistant prostate Background: Genomic signatures are revolutionizing the definition, identifi-
cancer (CRPC), remains a challenge due to the limited sensitivity and cation, and treatment of breast cancer. To precisely stratify breast cancers into
specificity of existing imaging modalities. Understanding prostate cancer actionable subgroups, full genome expression data and matching clinical data
biology with tumor biopsies does not address the issue of tumor heteroge- must be aggregated into a large data set. Such a data set will accelerate
neity or cellular degradation during the decalcification process of bone research and discovery, especially for smaller patient subsets who are not as
biopsies. Next generation positron emission tomography (PET) imaging and widely represented within the current body of literature. Methods: FLEX is a
circulating biomarkers might provide additional insights on treatment re- multicenter, prospective, population-based, observational trial for patients
sponses and inform clinical decision-making earlier in therapy. 18F-DCFPyL with Stage I, II, and III breast cancer. All patients with stage I to III breast
(PyL) is a second-generation fluorinated PSMA PET tracer that has superior cancer who receive MammaPrint, with or without BluePrint on a primary breast
sensitivity and specificity to detect prostate cancer compared to standard tumor are eligible for enrollment. The study’s primary aim is to create a large
imaging. Its role in assessing tumor response to therapy has not been scale, population-based registry of full genome expression data matched with
evaluated. Circulating tumor DNA (ctDNA) in blood can provide tumor ge- clinical data to investigate new gene associations with prognostic and/or
nomic information, while exosomes in serum and urine may provide data on predictive value. Secondary objectives include utilizing the shared study in-
the proteomic landscape of tumors. Methods: We are conducting a pro- frastructure to examine and generate hypotheses for targeted subset analyses
spective study of 15 men with metastatic CRPC who are scheduled to start a and/or trials based on full genome expression data. The design of FLEX allows
new systemic therapy for their disease. Upon enrollment, subjects will have targeted sub-studies and sub-analyses to be added as appendices after the
baseline assessments with standard cross-sectional imaging, 99mTc bone initial baseline study is opened. Patients enrolled in the initial study are also
scan, and blood work. Standard scans will be performed every 8-12 weeks eligible for inclusion in sub-studies where they meet all criteria and additional
until progression of disease. PyL PET/CT scans and liquid biopsies (ctDNA consent is not required. Additional clinical data will be collected as specified in
and exosomes) will occur at baseline, 6 weeks after starting their new the appendix protocols. The FLEX collaborative platform allows participating
therapy, and at disease progression. Lesions seen on PET/CT images will be investigators the opportunity to author their own sub-study protocols, as ap-
identified by a certified reader. The maximum standardized uptake value proved by the FLEX Steering Committee of their peers. 13 sub-studies have
(SUVmax) will be measured and recorded in up to the five hottest lesions and already been identified and are under development. Eligibility: The study will
normalized to a background SUVmean measured in the liver, spleen, kidney, enroll a minimum of 10000 patients aged $18 years with histologically proven
mediastinum, and parotid glands. Changes in the normalized SUV from invasive stage I-III breast cancer who signed informed consent. Enrollment
baseline to the 6-week PyL PET scan will be correlated to PSA response by began April 2017 and 1506 patients have been enrolled. Clinical trial in-
the Pearson’s correlation coefficient. The Kaplan-Meier method will be used formation: NCT03053193.
to evaluate progression free survival and overall survival dichotomized by the
median value of SUV change. Blood for ctDNA and exosomes will be stored
for future analysis. The study is open with two patients enrolled at the time of
submission. One patient has completed his initial PyL PET/CT scan. Clinical
trial information: NCT03585114.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Therapeutics and Tumor Biology (Nonimmuno) 187s

TPS3156 Poster Session (Board #143a), Sat, 8:00 AM-11:00 AM TPS3157 Poster Session (Board #143b), Sat, 8:00 AM-11:00 AM
TiFFANY study: A multicenter phase II basket-type clinical trial to evaluate FUZE clinical trial: a phase 2 study of Debio 1347 in FGFR fusion-positive
efficacy and safety of pan-FGFR inhibitor TAS-120 for advanced solid advanced solid tumors irrespectively of tumor histology. First Author:
malignancies with FGFR alterations identified by circulating tumor DNA. David Michael Hyman, Memorial Sloan Kettering Cancer Center, New York, NY
First Author: Tomoko Jogo, Department of Gastroenterology and Gastroin-
Background: Dysregulation of fibroblast growth factor receptor (FGFR)
testinal Oncology, National Cancer Center Hospital East, Kashiwa, Japan
signaling by FGFR fusions is implicated in many cancers. Debio 1347 is a
Background: Approximately 7% of advanced solid malignancies have FGFR selective oral inhibitor of FGFR 1-3 tyrosine kinases. It exhibited high an-
gene alterations. However, standard treatment for FGFR-altered malignancies titumor activity in in vitro and in vivo tumor models with FGFR1-3 gene
has not been established. Moreover, circulating tumor DNA (ctDNA) analysis fusions. Preliminary data from an ongoing phase 1 trial show efficacy and
has a potential to accurately identify FGFR alterations by assessing spatial and tolerability in patients (pts) harboring FGFR 1-3 fusion irrespectively of
temporal intratumoral heterogeneity, which have shown to be associated with a tumor type. We present the design for a multicenter, basket, 2-stage,
poor prognosis and resistance to anti-cancer therapy. Methods: We are con- adaptive single arm Phase 2 trial investigating Debio 1347 in pts with solid
ducting an investigator-initiated multicenter phase II basket-type trial to in- tumors harboring FGFR1-3 fusion/rearrangement. Methods: Adults with
vestigate efficacy and safety of TAS-120, a highly selective covalent pan-FGFR locally advanced/unresectable or metastatic tumors with documented
inhibitor, for the patients with advanced solid malignancies with FGFR al- FGFR1-3 gene fusion/rearrangement who require systemic therapy and have
terations identified by ctDNA analysis as a part of the Nationwide Cancer progression after $1 prior standard treatment or have no satisfactory al-
Genome Screening Project (GOZILA study, UMIN000029315). Eligibility ternative treatment option are eligible. Three cohorts are included: biliary
criteria include histologically confirmed unresectable advanced or recurrent tract cancer (cohort 1), urothelial cancer (cohort 2) and all other solid tumors
solid tumors regardless of histology of origin; ECOG PS of 0 or 1; refractory or (cohort 3). Primary brain tumors are excluded. Other key exclusion criteria
intolerant to the standard therapies; and clonal FGFR alterations (FGFR1-3 include prior treatment with FGFR1-3 selective inhibitor; clinically signif-
gain-of-function mutations, FGFR1,2 amplifications and FGFR2,3 fusions) icant corneal/retinal disorder; history of calcium/phosphate homeostasis
identified by a 73-gene sequencing ctDNA panel (Guardant360). Enrolled disorder or systemic mineral imbalance with ectopic soft tissue calcification,
patients will receive TAS-120 20 mg once daily, orally, in a 21 day-cycle. The and symptomatic/unstable brain metastases , 1 month before enrollment.
primary endpoint is to clarify objective response rate (ORR) assessed by in- Genomic screening of tumor tissue is done at local or central laboratory with
vestigators per RECIST v1.1. The secondary endpoints are to evaluate progression- post-hoc central confirmation by RNA sequencing. Eligible pts will receive
free survival, duration of response, time to treatment failure, disease control rate, Debio 1347, 80 mg PO once daily in 28-day cycles until occurrence of
overall survival, ORR by central determination, and incidence of adverse events. progression or unacceptable toxicity. Primary Endpoint is objective response
Target sample size is determined as 26 to test the null hypothesis of ORR as 5% rate (ORR) based on independent central review using RECIST v.1.1. The
with one-sided alpha level of 2.5% and power of 80% to detect an expected value targeted sample size (N=125) will provide approximately 90% power to
of ORR as 25%. Furthermore, tumor tissue and ctDNA will be serially collected reject H0: ORR # 15% at an overall 5% significance level based on an
and analyzed to investigate the resistance mechanisms and provide clinically expected ORR of 30% in at least one of the cohorts. Secondary endpoints
meaningful biomarker which may be used for identifying and implementing are: duration of response, disease control rate, progression-free survival,
treatment changes. Clinical trial information: 194624. overall survival, safety, tolerability, and quality of life. An interim analysis for
futility and homogeneity will be performed after 27 evaluable pts. PK sparse
sampling is performed to assess exposure-response relationships with ef-
ficacy and safety. Biomarkers of response and resistance will be explored.
Accrual is opened in US, EU, Asia and Australia. Clinical trial information:
NCT03834220.

TPS3158 Poster Session (Board #144a), Sat, 8:00 AM-11:00 AM TPS3159 Poster Session (Board #144b), Sat, 8:00 AM-11:00 AM
A phase I open label study evaluating VT1021 in patients with advanced solid Express study: A trial in progress exploring the association between low level
tumors. First Author: Michael Cieslewicz, Vigeo Therapeutics, Cambridge, MA of genomic alteration and exceptional and unexpected response to targeted
therapies in patients with solid tumors. First Author: Olivia Le Saux, CH Lyon
Background: VT1021 is a cyclic pentapeptide that functions as a potent
Sud, Lyon, France
inducer of thrombospondin-1 (Tsp-1) expression in the tumor microenviron-
ment (TME). By triggering the production of Tsp-1, VT1021 reprograms the Background: Molecular targeted agents (MTA) resulted in breakthroughs in
TME from one that is immune-suppressive and tumor-promoting, to one that selected niches. It is often assumed that tumor regression is consecutive to an
activates the adaptive immune system and is tumor-inhibiting. Tsp-1 repro- oncogenic de-addiction effect. An emerging hypothesis suggests that genomic
grams the TME to: (i) induce apoptosis in tumor cells that express CD36 on instability may be associated with poor response to MTA. Indeed, the accu-
their cell surface, (ii) convert macrophages from M2 to M1 polarization, which mulation of defects in multiple oncogenes or tumor suppressor genes may
promotes phagocytosis and blunts immunosuppression and (iii) inhibit an- result in the activation of multiple oncogenic pathways. These multiple sig-
giogenesis. Preclinical studies have shown robust anti-tumor activities of naling would mechanically result in a limitation of the oncogenic de-addiction
VT1021 in animal models of ovarian, pancreatic and breast cancer, including process. Another hypothesis, suggests that tumor heterogeneity could also be
complete tumor regression and reprogramming of the immune TME. These associated with poor outcome under MTA. Such heterogeneity could also result
observations led to the initiation of the first-in-human study of VT1021. from the genomic instability, and be appraised by bioinformatic and functional
Methods: This study is a first-in-human, Phase 1, open-label, multicenter, approaches. In this study, we thought to investigate whether molecular profiles
dose escalation (Part 1) study with dose expansion (Part 2) in advanced solid reflecting a low level of genomic alterations in genes causally implicated in
tumors. The primary objectives are to assess the safety and tolerability of oncogenesis could be associated with an exceptional response (ER) to MTA.
VT1021, to assess dose-limiting toxicities (DLT), and to determine the Methods: This is an exploratory, multicenter, multicohort, prospective trial
maximum tolerated dose (MTD) and the recommended Phase 2 dose (RP2D). conducted in 264 adult patients, with advanced breast, lung, colorectal,
Secondary objectives include the evaluation of pharmacokinetics (PK) and ovarian, kidney cancers and melanoma, having presented an ER to an ap-
pharmacodynamic (PD) effects of VT1021 in tumor and tumor microenvi- proved MTA. ER is defined using the definition chosen by the NCI which
ronment, and assessment of preliminary antitumor activity. VT1021 is ad- combines the three criteria: - complete or partial response, - lasting . 6
ministered intravenously twice weekly. DLTs will be assessed in the first cycle months, - and not expected in . 10% of the patients in this drug – organ
(Days 1-28) of the dose escalation cohort and are defined as grade 3 adverse situation. The primary objective is to assess whether ER can be associated
events related to VT1021. In Part 1 of the study, 24-30 patients will be with a low level of genomic instability in the tumor. Low genomic instability is
enrolled to determine the MTD and RP2D for expansion. In Part 2 of the study, defined by the presence of less than the 5th quantile of genomic alterations
80-100 patients will be enrolled, grouped into cohorts based on disease (mutations, amplifications, deletions) to be expected in the given tumor type as
subtypes (ovarian, pancreatic, Triple-negative breast cancers, and glioblas- per TCGA database. For each tumor type, the null hypothesis H0: p = 0.05 will
toma). Blood samples and biopsy samples from patients will be collected to be tested, against the one-sided alternative hypothesis p . 0.05. For each of
assess PK properties and PD responses systemically as well as in the TME. No the 6 cohorts, a sample size of 44 patients is necessary to achieve 80% power
formal statistical hypothesis testing will be conducted in this study. This at p = 15 with a one-sided level 5% test. Patients presenting an ER will be
study is currently open for enrollment in the US. Clinical trial information: identified retrospectively, in a nationwide manner, then monthly reviewed and
NCT03364400. validated for inclusion by a panel of pathology experts. As of February 2019, 75
patients have been included. The identification of molecular traits associated
with ER might serve the development of predictive classifiers for precision
medicine. This study also represents a unique opportunity to better understand
cancer biology. Clinical trial information: NCT02701907.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
188s Developmental Therapeutics and Tumor Biology (Nonimmuno)

TPS3160 Poster Session (Board #145a), Sat, 8:00 AM-11:00 AM TPS3161 Poster Session (Board #145b), Sat, 8:00 AM-11:00 AM
SGNTV-001: Open label phase 2 study of tisotumab vedotin for locally A phase I/II multiple expansion cohort trial of MRTX849 in patients with
advanced or metastatic disease in solid tumors. First Author: David S. Hong, advanced solid tumors with KRAS G12C mutation. First Author: Kyriakos P.
The University of Texas MD Anderson Cancer Center, Houston, TX Papadopoulos, South Texas Accelerated Research Therapeutics, San
Antonio, TX
Background: Tisotumab vedotin (TV) is being developed for patients with
cervical cancer and other solid tumors known to express Tissue Factor (TF). Background: RAS proteins are part of the family of small GTPases which
Expression of TF on tumor cells has been associated with poor prognosis in regulate intracellular signaling pathways responsible for cell growth, migration,
several tumor types. Four tumor types were chosen for this study based on survival and differentiation. Oncogenic point mutations in RAS in codons 12,
unmet medical need and TF expression. TV is an antibody-drug conjugate 13, and 61 occur in up to one-third of all human cancers and result in
composed of a TF-targeted fully human monoclonal immunoglobulin G1 constitutive activation of RAS signaling, playing a key role in uncontrolled
conjugated via a protease-cleavable valine citrulline linker to the drug mon- cellular growth and malignant transformation. Mutant KRASG12C in particular
omethyl auristatin E (MMAE). TV-mediated delivery of MMAE drives antitumor comprises approximately 14% of lung adenocarcinoma and 4% of colon
activity through cytotoxic cell killing and has been shown to induce immu- adenocarcinoma, and less commonly in certain other types of cancer. For
nogenic cell death (ICD). In preclinical studies, TV treatment resulted in potent decades, KRAS was considered undruggable due to its high affinity for GTP/
and long-lasting tumor regression in TF-expressing xenograft models derived GDP and the lack of a clear binding pocket. Recent discoveries have en-
from a variety of solid cancers, including patient-derived xenograft models with abled the development of compounds, including MRTX849, that covalently
heterogeneous TF expression. TV has shown preliminary evidence of activity in bind to KRASG12C at the cysteine at residue 12, lock the protein in its inactive
heavily pretreated patients with many TF-expressing tumor types in the GDP-bound conformation, and inhibit KRAS-dependent signal transduction.
GEN701 phase 1 study. Methods: SGNTV-001 (innovaTV 207) is a global, MRTX849 is a potent, orally-available, mutation-selective small molecule
open label, multicenter phase 2 trial designed to assess the safety, tolerability, covalent inhibitor of KRASG12C. MRTX849 inhibits KRASG12C signaling in cell
and activity of TV for the treatment of solid tumors. (NCT03485209; EudraCT lines harboring this mutation, and results in tumor regression in a broad
2017-005076-26). Patients are currently enrolling in 4 cohorts: colorectal spectrum of KRASG12C animal models. Methods: This multi-center, Phase 1/2,
cancer (CRC), squamous non-small cell lung cancer (NSCLC), exocrine multiple expansion cohort trial evaluates the safety, pharmacokinetics (PK),
pancreatic adenocarcinoma, and squamous cell carcinoma of the head and metabolites, pharmacodynamics, and clinical activity of MRTX849 in patients
neck (SCCHN). Patients must have measureable disease per RECIST v1.1 that with advanced solid tumor malignancies with a KRAS (p.G12C) mutation. The
has progressed despite standard of care systemic treatment in the locally study starts with an evaluation of dose and regimen of MRTX849 using a
advanced or metastatic setting (up to 3 prior lines for CRC, 2 for squamous combination of the accelerated titration and modified toxicity probability in-
NSCLC and SCCHN, and 1 for pancreatic adenocarcinoma) and have an ECOG terval designs, with MRTX849 initially administered once daily in a continuous
score of 0 or 1. The primary endpoint is investigator-determined confirmed regimen expressed in 3-week cycles. As potentially viable regimens are
ORR as measured by RECIST v1.1. Secondary objectives include DOR, PFS, identified, Phase 1b expansion cohorts will be opened to provide greater safety
OS, DCR, and TTR, as well as safety and pharmacokinetic parameters. The first and PK data for determination of the recommended Phase 2 dose (RP2D) and
patient was enrolled in July 2018 and the study is currently enrolling across 16 regimen. In Phase 2, separate cohorts of patients by histological diagnosis,
sites in the US (as of January 2019) with additional sites opening in EU including non-small cell lung cancer, colorectal, and other solid tumors, will be
countries. Clinical trial information: NCT03485209. enrolled and evaluated for clinical activity using a predictive probability design.
The study is open for enrollment, and recruitment is ongoing. Clinical trial
information: NCT03785249.

TPS3162 Poster Session (Board #146a), Sat, 8:00 AM-11:00 AM TPS3163 Poster Session (Board #146b), Sat, 8:00 AM-11:00 AM
A phase I study of BOS172738 in patients with advanced solid tumors with A phase I dose-escalation and immune biomarker study of intravenous FF-
RET gene alterations including non-small cell lung cancer and medullary 10832, liposomal gemcitabine, in patients with advanced solid tumors. First
thyroid cancer. First Author: Patrick Schoffski, UZ Leuven, Leuven, Belgium Author: Erkut Hasan Borazanci, HonorHealth Research Institute, Scottsdale,
AZ
Background: RET gene alterations (mutations and fusions) leading to con-
stitutive kinase activity have been identified in various tumor types including Background: FF-10832 (832) is a liposomal formulation of gemcitabine
non-small cell lung cancer (NSCLC), medullary thyroid (MTC), colon, breast (GEM) that demonstrates a prolonged half-life and preferential uptake in
and ovarian cancer. The current generation of multi-kinase inhibitors ap- tumor vs normal tissues and marrow in pre-clinical models. Macrophage
proved for treatment of such tumors, do not selectively target RET and exhibit uptake has been shown in the tumor microenvironment (TME), with sub-
significant off-target activity especially against vascular endothelial growth sequent GEM release in tumor cells. This relative selectivity and anti-tumor
factor receptor 2 (VEGFR2), resulting in dose-limiting toxicities that prevent immunological changes observed in the TME may lead to decreased toxicity
the full inhibition of RET in those tumors. Recently, early clinical data from a and increased efficacy compared to GEM. Ferumoxytol (FMX) may be a
class of more selective RET inhibitors have shown promising results with a surrogate for nanoparticle penetration into tissue and is being examined as a
more favorable safety profile in patients with RET alterations. BOS172738 potential correlate for activity. Methods: This ongoing Phase 1, 3+3 dose-
is a novel RET inhibitor with nanomolar potency against RET and approx- escalation study of 832 will determine the safety profile, maximum tolerated
imately 300-fold selectivity against VEGFR2. This phase 1 study is assessing dose, dose-limiting toxicities (DLT) and recommended Phase 2 dose, and
the safety and tolerability of BOS172738 in patients with advanced solid will be followed by expansion. Enrollment of up to 60 patients (pts) with
tumors with RET alterations. Methods: NCT03780517 is a phase 1, open advanced solid tumors is planned. Pre-treatment (tx) FMX MRI scans are
label, multicenter, dose escalation trial to evaluate the safety, efficacy, performed, followed by 832 administration on Days 1 &15 of each 28-day
pharmacokinetics, and pharmacodynamics of BOS172738, an orally dosed cycle until disease progression or unacceptable toxicity. In addition to
RET kinase inhibitor, in patients with advanced solid tumors with RET gene standard biomarker and imaging evaluations, change in macrophage po-
alterations. RET gene alteration status will be assessed locally but confirmed larity, myeloid-derived suppressor cell (MDSC) and regulatory T cell pop-
centrally. The study is comprised of 2 parts: in Part A (dose escalation), ulations are being investigatedin peripheral blood and tumor tissue. Clinical
patients with advanced solid tumors with RET gene alterations will receive trial information: NCT03440450.
BOS172738 orally once daily in each 28-day cycle. Select patients in Part A
are eligible for intrapatient dose escalation. On establishing the recom-
mended phase 2 dose (RP2D), Part B (expansion) will enroll up to an ad-
ditional 60 patients to 1 of 3 tumor type-specific cohorts. The 3 expansion
cohorts will each consist of up to 20 advanced cancer patients with: 1) RET
gene-fusion NSCLC; 2) RET gene-mutant MTC; and 3) other RET gene-
altered advanced tumors or NSCLC/MTC with prior specific RET gene-
targeted therapy. Patients in expansion cohorts will receive BOS172738
daily at the RP2D until disease progression or other discontinuation criteria
have been met. The study is currently open to enrollment globally with the
first patient entered in 01/2019. Clinical trial information: NCT03780517.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Developmental Therapeutics and Tumor Biology (Nonimmuno) 189s

TPS3164 Poster Session (Board #147a), Sat, 8:00 AM-11:00 AM TPS3165 Poster Session (Board #147b), Sat, 8:00 AM-11:00 AM
Phase I study of procaspase activating compound -1 (PAC-1) in combination A window of opportunity trial of atorvastatin in p53-mutant and p53 wild type
with temozolomide (TMZ) for the treatment of recurrent malignant glioma. First malignancies. First Author: Mohammad Telfah, University of Kansas Cancer
Author: Martin Kelly Nicholas, University of Illinois at Chicago, Chicago, IL Center, Westwood, KS
Background: The caspase family of cysteine proteases play key roles in the Background: Mutations in p53 contribute to tumor progression. A rational
initiation and execution of apoptosis. The activation of procaspase-3 to approach is to destabilize mutant (m) p53. The group at the University of Kansas
caspase-3 is critical in both the intrinsic and extrinsic apoptotic cascades. Cancer Center screened compounds that suppress m p53 in a preclinical
Procaspase-3 levels are elevated in many cancers, including glioblastoma model. Luciferase-based reporter assay identified statins as suppressors of m
(GBM). As a result, caspase-3 levels are abnormally low in these tumors; thus p53 expression. In vitro validation assay demonstrated atorvastatin (A)
they avoid apoptosis. PAC-1 is a small molecule that directly activates suppressed m p53 level and cell growth selectively; and depletion of mevalonic
procaspase-3 and induces apoptosis of cancer cells. PAC-1 has activity acid lead to degradation of m p53. These effects were limited to mutations in
against a wide range of cancer cell lines, and in animal models of cancer. PAC- the conformation of p53, while wild-type and DNA contact mutations were not
1 crosses the blood brain barrier and has been shown to synergize with TMZ in as sensitive to statin-induced degradation of p53. M p53 xenograft model
both canine malignant glioma and meningioma that arise spontaneously. confirmed that A could suppress tumor growth at a concentration that can
Methods: This Phase I dose escalation study uses a modified- Fibonacci 3+3 decrease LDL level. The primary objective of this trial is to determine if A
design to determine the MTD of PAC-1 when combined with TMZ in patients decreases the level of conformational m p53. The secondary objective is to
with recurrent malignant gliomas: anaplastic astrocytoma (AA) and GBM (open assess the effects of A on Ki-67 and caspase-3 in conformational m p53 tumors.
to enrollment). Here, we focus on component 2 of the study. Primary ob- Methods: This is an open-label, window of opportunity pilot trial to see if A given
jectives: to establish MTD of PAC-1 when combined with a fixed dose of TMZ, for 1 to 4 weeks at a dose of 80 mg/day is sufficient to reduce the levels of
tolerability, and toxicity using CTCAE v.4. Secondary and correlative objec- conformational m p53 in the tumor tissues. Subjects with new diagnosis of
tives: pharmacokinetics, pharmacodynamics, preliminary anti-tumor activity malignancy with a planned surgical therapy, and subjects with previously
correlation with procaspase-3 expression in tumor tissue, radiographic re- treated AML, in between treatment regimens, are eligible. Tissues from solid
sponse using the Response Assessment in Neuro-Oncology (RANO) criteria, tumors, and bone marrow or peripheral blood samples from AML will be used to
and neurocognitive function using a validated test battery. Inclusion criteria: screen for m p53 by immunohistochemistry (IHC). Subjects will receive A at
diagnosis of recurrent high grade glioma (AA or GBM), ECOG PS 0-2, adequate 80 mg/day po for 1 to 4 weeks. Pharmacokinetics at pre-dose and 1-hour
organ function. Exclusion criteria: received prior cytotoxic therapy in the last 3- post-dose on Day 1 and on the day of surgery will be done. Mutational
6 weeks (duration based on prior therapy) or uncontrolled chronic illness. analysis using exome sequencing technique will be done on m p53. Using
Administration and design, Component 2: PAC-1, orally administered, is dosed IHC, the amount of p53 in pre-treatment and post-treatment samples will be
at 375-650 mg daily (up to 3 dose levels) on days 1-21 of each 28-day cycle. A measured and compared simultaneously. The levels of Ki67 and caspace-3
fixed dose of TMZ, (150 mg/m2), is administered orally, days 8 -12 of will be tested and compared between pre-treatment and post-treatment
each cycle. The study is currently enrolling patients for Component 2. Clinical samples in subjects with conformational m p53, between conformational
trial information: NCT02355535. and non-conformational m p53, and in wild-type p53 tumors. The trial is
actively enrolling subjects. The results of this trial will determine further in-
vestigations on the role of atorvastatin in tumors with p53 mutations in a
placebo-controlled, randomized trial. Clinical trial information: NCT03560882.

TPS3166 Poster Session (Board #148a), Sat, 8:00 AM-11:00 AM TPS3167 Poster Session (Board #148b), Sat, 8:00 AM-11:00 AM
A phase Ia/Ib, open label, multicenter, dose-escalation study of BI 907828 A CRUK first-in-human phase I trial of a CDC7 Inhibitor, LY3143921
(MDM2-p53 antagonist) in adult patients with advanced or metastatic solid hydrate, in patients with advanced solid tumors. First Author: Peter F.
tumors. First Author: Curtis Robert Chong, Memorial Sloan Kettering Cancer Gallagher, Queen’s University Belfast, Belfast, United Kingdom
Center, New York, NY
Background: CDC7 is a protein with key roles in DNA replication initiation,
Background: Inactivation of tumor protein 53 (TP53) is a central mechanism the intra-S-phase checkpoint and M-phase completion. CDC7 is over-
of tumors to promote survival and proliferation. The murine double minute 2 expressed in malignant compared to non-malignant cells, particularly
(MDM2) oncogene is the primary cellular negative regulator of TP53. Small those with TP53 mutations, making it an attractive therapeutic target.
molecule inhibitors of the MDM2-p53 interaction are currently being LY3143921 hydrate is an orally administered ATP-competitive CDC7 in-
evaluated in clinical trials as new anti-cancer drugs, and clinical data hibitor. Pre-clinical studies in colorectal cancer (CRC) and squamous non-
published to date support the rationale to target MDM2-amplified tumors. BI small cell lung cancer (sqNSCLC) demonstrate favourable anti-cancer ac-
907828 is a potent MDM2-p53 antagonist that has shown efficacy in mouse tivity, particularly in squamous NSCLC and in CRC with TP53 null and
models of human cancer. Methods: NCT03449381 is a Phase 1a/1b, open- missense mutations. We hypothesise that solid tumours mutated in TP53
label, multicenter, dose-escalation trial of BI 907828 in patients aged $18 will be sensitive to LY3143921 therapy. Methods: This is a first-in-human,
years with advanced/metastatic solid tumors. Primary objectives of the phase I trial of LY3143921 hydrate (LY3143921) monotherapy given twice
Phase 1a (dose-escalation) part are to determine: the maximum tolerated daily, continuously on a 21 day schedule until disease progression, patient
dose (MTD) of BI 907828 based on dose-limiting toxicities (DLTs) during the (pt) withdrawal or unacceptable toxicity (NCT03096054). Eligible pts have
first treatment cycle; the recommended dose for expansion (RDE); safety and histologically proven advanced/metastatic solid tumours for which no further
tolerability. Patients in Arm A will receive one dose of BI 907828 every standard therapy exists and WHO PS 0-1. Pts have regular clinical as-
21 days, and patients in Arm B one dose on Days 1 and 8, every 28 days. sessment and tumour imaging every 2 cycles. Phase Ia (dose escalation) is
Secondary objectives include pharmacokinetics (PK), pharmacodynamics recruiting in a 3+3 design following 3 initial single patient cohorts (starting
(PD) [GDF-15 induction in plasma], and preliminary anti-tumor activity. dose 30 mg OD), enriching for patients with malignancies associated with
Primary endpoints are the number of patients with DLTs during the first p53 mutations (CRC, sqNSCLC, high grade serous ovarian, squamous cell
treatment cycle, and the MTD. Phase 1a will include »50 evaluable patients oesophageal, squamous cell head & neck, urothelial, pancreatic and triple
with locally advanced or metastatic solid tumors with either a known TP53 negative breast cancer). Recruitment to cohort 6 (180 mg BD) is ongoing. On
wild type (wt) status or unknown TP53 status, regardless of MDM2 am- determination of the maximum tolerated dose (MTD) and recommended
plification status at the time of study entry. The main objectives of Phase 1b phase II dose and schedule (RP2D), 2 expansion cohorts (# 25 pts each) of
(expansion cohorts) are to assess efficacy, safety, and PK profiles at the RDE, patients with CRC and sqNSCLC will be evaluated. Primary objectives:
and to determine the recommended dose for Phase 2. The primary endpoint assess safety and tolerability of LY3143921, determine MTD and RP2D.
is objective response, where best response is determined according to Secondary objectives: evaluate preliminary efficacy and PK profile of
RECIST v1.1, or RANO criteria for patients with glioblastoma. Phase 1b will LY3143921. All pts will have archival tumour tissue retrospectively ana-
include up to 150 evaluable patients with TP53 wt tumors, assigned to four lysed, while patients in phase Ib will also have pre- and on-treatment tumour
different cohorts, including non-squamous NSCLC, soft tissue sarcoma, biopsies. Evaluation of potential predictive and pharmacodynamic bio-
glioblastoma, urothelial carcinoma, and solid tumors with brain metastases. markers including p53 mutations, phosphorylated MCM2, cyclin B1 and
As of 8th February 2019, 11 patients have been enrolled. Clinical trial molecular subgroups of target tumours will be included. Clinical trial in-
information: NCT03449381. formation: NCT03096054.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
190s Developmental Therapeutics and Tumor Biology (Nonimmuno)

TPS3168 Poster Session (Board #149a), Sat, 8:00 AM-11:00 AM TPS3169 Poster Session (Board #149b), Sat, 8:00 AM-11:00 AM
A phase I, first-in-human, dose-escalation and dose-expansion study of the Phase 1, open-label, dose-escalation study of M3814 + avelumab 6 radio-
multitarget kinase inhibitor TT-00420 in patients with advanced solid therapy (RT) in patients (pts) with advanced solid tumors. First Author:
tumors. First Author: Sarina Anne Piha-Paul, The University of Texas MD Johanna C. Bendell, Sarah Cannon Research Institute/Tennessee Oncology,
Anderson Cancer Center, Houston, TX Nashville, TN
Background: Multi-target kinase inhibitors have gained increasing attention in Background: DNA-dependent protein kinase (DNA-PK) regulates a key DNA
the past few years due to their capabilities of simultaneously targeting several damage response (DDR) pathway for double-strand break (DSB) repair. DNA-
hallmarks of cancer. Triple negative breast cancer (TNBC), the most aggressive PK inhibition augments DNA DSB damage generated by many antitumor
type of breast cancer, is a highly heterogeneous disease composed of several therapies, including RT. DNA damage and repair impact the interaction of
subtypes with distinct genomic profiles and activating pathways. TT-00420 tumors with the immune system; combining immune checkpoint inhibitors
is a multi-target kinase inhibitor that targets Aurora A/B, receptor tyrosine (CPIs) with RT + DDR-targeted agents may modulate the tumor immune
kinases (RTKs) involved in angiogenesis, and other kinases involved in tumor- microenvironment, enhancing responsiveness to CPIs. M3814 (small
associated inflammation and immune escape. Preclinical studies have molecule selective DNA-PK inhibitor) has demonstrated monotherapy ac-
established signs of efficacy for TT-00420 in TNBC. Targets inhibited by TT- tivity in several tumor cell lines, and M3814 + RT combined with avelumab
00420 are among the key dysregulated pathways directly involved in the (programmed death ligand 1 mAb) significantly delayed tumor growth vs
tumorigenesis of TNBC. TT-00420 is efficacious against most subtypes of either agent alone + RT in MC38 syngrafts. This study will evaluate the
TNBC cell lines. This anti-TNBC activity is confirmed in both cell line derived clinical utility of M3814 combined with avelumab 6 RT in pts with advanced
xenograft (CDX) and patient derived xenograft (PDX) TNBC model in vivo, in solid tumors. Methods: NCT03724890 is a 2-part first-in-man study in adult
which TT-00420 is active both as first-line and second-line treatment. TT- pts with advanced or metastatic solid tumors. Part A is enrolling pts with
00420 demonstrated good oral bioavailability and pharmacokinetic properties measurable/evaluable solid tumors (RECIST v1.1); Part B will enrol pts with
in mice, rats and dogs, and revealed mechanism-related but manageable primary or metastatic tumor(s) in the lung which is/are amenable to be irra-
toxicities. The IND approval of TT-00420 was granted by the FDA on Sept. 27, diated. In Part A, M3814 will be given orally twice daily. In Part B, M3814 +
2018. Methods: TT420X2101 is an open-label, first-in-human, multicenter, TRT will be given once daily, 5 days/wk for 2 wk. In both parts, pts will receive
phase I study including a dose escalation portion in adult patients with ad- avelumab iv once every 2 wk from Day 1 until disease progression/
vanced solid tumors, followed by dose expansion in two parallel cohorts, TNBC unacceptable toxicity. Part B will initiate once the Safety Monitoring Com-
cohort (N=22) and selected advanced tumor (SAT) cohort (N=22). Adverse mittee declares the first dose level of Part A to have acceptable safety/
events will be evaluated per CTCAE v5.0 criteria, and tumor responses will be tolerability. Primary objectives are to define the recommended Phase 2
evaluated per RECIST v1.1 criteria. Patients aged 18-75 with measurable dose (RP2D) of M3814 when combined with avelumab (Part A) and with
target lesion(s) at baseline and ECOG status of 0 or 1 will be enrolled. Patients avelumab + TRT (Part B) via dose-limiting toxicities (DLTs) occurring during
are treated with a single daily oral administration of TT-00420 continuously the first 3/4 (Part A/B) wk of treatment. Secondary objectives include safety/
for a 28-day cycle. Dose escalation is driven by Bayesian modeling with over- tolerability, pharmacokinetics, immunogenicity, preliminary antitumor activity
dose control. Approximately 66 patients are expected to be enrolled and (BOR, PFS, OS). Sample size for each part depends on the number of DLTs/
treated in this study. Primary endpoint is to evaluate dose limiting toxicity dose levels for M3814; dose escalation will be based on a Bayesian logistic
(DLT) and identify maximum tolerate dose (MTD), if feasible, or establish regression model with overdose control. Part A aims to include 6–24 pts (#4
recommended dose for Dose Expansion. Preliminary efficacy and PK profile dose levels), Part B 6–18 pts (#3 dose levels). Recruitment began in Dec
will be evaluated as well. Enrollment is currently ongoing. Clinical trial in- 2018. Clinical trial information: NCT03724890.
formation: NCT03654547.

TPS3170 Poster Session (Board #150a), Sat, 8:00 AM-11:00 AM TPS3171 Poster Session (Board #150b), Sat, 8:00 AM-11:00 AM
Leeomic: A comprehensive proteomic analysis towards discovery of Prospective observational study for treatment resistance-related gene
predictive patterns of protein expression to ribociclib sensitivity and screening using plasma circulating tumor DNA in the third generation EGFR
resistance—A compLEEment-1 Canadian correlative sub-study. First Au- TKI osimertinib therapy elucidator. First Author: Akihiro Tamiya, Department
thor: Stephen K. L. Chia, BC Cancer Agency, Vancouver, BC, Canada of Internal Medicine, National Hospital Organization Kinki-Chuo Chest
Medical Center, Sakai, Japan
Background: Despite developments in the treatment of advanced hormone
receptor positive (HR+), human epidermal growth factor receptor 2 negative Background: Osimertinib (Osi) is a third-generation epidermal growth factor
(HER2-) breast cancer, primary or acquired resistance eventually occurs in receptor (EGFR) – -tyrosine kinase inhibitor (TKI) that potently and selec-
all cases and there is still very limited understanding of the mechanisms of tively inhibits both EGFR-TKI sensitizing and EGFR T790M resistance
resistance to therapy. LEEOMIC is a sub-study of the main CompLEEment-1 mutations (mu). O showed superior efficacy compared with first generation
(N = 3255 patients enrolled, CLEE011A2404 v03) trial, an open-label, EGFR-TKIs in patients (pts) with previously untreated advanced NSCLC
phase 3b study evaluating ribociclib + letrozole as first-line therapy in an harboring sensitizing EGFR mu in the prior phase III trial. Therefore, Osi is
expanded advanced breast cancer patient population which recruited over one of the most important standard therapies in EGFR mu positive pts.
250 Canadian patients. The purpose of this Canadian correlative sample However, there are few reports about mechanisms of Osi resistance as the
collection study is to explore the mechanisms of response and resistance to first line (1L) EGFR-TKI. Understanding of 1L Osi resistance mechanisms is
ribociclib in combination with letrozole through proteomic and ctDNA essential to conduct future therapeutic strategy for EGFR mu NSCLC pts. We
analysis. Methods: The British Columbia Cancer Research Centre team planned the analysis of the resistant mechanism by the ultra-sensitive next-
developed a novel and optimized MS/MS platform called SP3-Clinical Tissue generation sequence (NGS) using the circulating tumor (ct) DNA to clarify a
Proteomics (SP3-CTP) to perform in-depth proteome profiling ( . 8,000 resistant mechanism of 1L Osi. Methods: We prospectively collect ctDNA
proteins) from formalin fixed paraffin embedded (FFPE) material (10-micron samples from EGFR mu positive NSCLC who will be treated with Osi as 1L.
section). SP3-CTP analysis of the proteome of the study patients who did not Planned enrollment is 180 cases, we estimate an analysis case at exacer-
achieve clinical benefit (primary resistance: progression within 3 months of bation to be 120 cases in a registration period and the observation period of
treatment) will be compared to the proteome of the sub-group of prolonged for each two years. ctDNA samples are collected at treatment initiation, 3
responders (time to progression of 22 months or more) in order to identify and 12 months after Osi treatment, and disease progression. We will analyze
biomarkers that can predict response or de-novo resistance to therapy. the mechanisms of 1L Osi by the ultra-sensitive NGS using ctDNA. The key
Archival tumor biopsies (primary or metastatic) collected from the study will inclusion criteria are histologically or cytologically proven NSCLC, EGFR mu
be submitted for proteomic analysis to identify proteomic expression levels positive, pts treated with Osi as 1L, and written informed consent. The key
that may serve as predictor of response. It is anticipated that over 150 exclusion criteria are patients who were deemed unsuitable for participation
samples will be collected. If available, blood samples taken at time of by an attending physician due to other conditions. The primary outcome is to
progression or end of treatment will also be analyzed for ctDNA for genetic clarify the incidence and ratio of Osi resistance by the ultra-sensitive NGS
profiling and to study if there is any correlation between genetic mutations method using ctDNA at treatment initiation and progression disease. And key
and response or resistance to therapy. Currently, both tissue and blood secondary endpoint is to examine how the detection quantity of the Osi
samples are being collected and no analysis has been conducted thus far. resistance association mu in ctDNA at treatment initiation influences pro-
Clinical trial information: NCT03613220. gressive disease. Clinical trial information: jRCTs031180051.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Colorectal) Cancer 191s

3500 Oral Abstract Session, Sat, 3:00 PM-6:00 PM 3501 Oral Abstract Session, Sat, 3:00 PM-6:00 PM
Three versus six months adjuvant FOLFOX or CAPOX for high risk stage II and Prospective pooled analysis of four randomized trials investigating duration
stage III colon cancer patients: The efficacy results of Hellenic Oncology of adjuvant (adj) oxaliplatin-based therapy (3 vs 6 months {m}) for patients (pts)
Research Group (HORG) participation to the International Duration Evaluation with high-risk stage II colorectal cancer (CC). First Author: Timothy Iveson,
of Adjuvant chemotherapy (IDEA) project. First Author: Ioannis Sougklakos, University Hospital Southampton NHS Foundation Trust, Southampton,
University of Heraklion, Heraklion, Greece United Kingdom
Background: The IDEA aimed to investigate whether a 3-month (3M) of Background: 6m of oxaliplatin-based treatment is an option as adj che-
oxaliplatin/fluoropyrimidines-based adjuvant chemotherapy (CT) is non- motherapy for patients with high risk stage II CC (T4, inadequate nodal
inferior to the 6-month (6M) in 3-year disease free survival (DFS) in stage harvest, poorly differentiated, obstruction, perforation or vascular/perineural
high risk stage II and in stage III colon cancer (CC). Methods: HORG-IDEA invasion). The IDEA collaboration showed shorter treatment duration to be
randomized patients between 3M and 6M of CT with FOLFOX4 or CAPOX appropriate for most pts with stage III colon cancer. The results of the 4 IDEA
with primary end point the 3 years DFS (3yDFS). Results: In total 1121 studies with stage II pts are presented here. Methods: A prospective, pre-
patients, 413 with high risk stage and 708 with stage IIICC, were ran- planned pooled analysis of high-risk stage II patients from 4 concurrently
domized between May 2009 and October 2015. The median follow-up was conducted randomized phase III trials (SCOT, TOSCA, ACHIEVE-2, HORG)
67 (38-126) months. There were 79 DFS events (43 in 3M and 38 in 6M was performed to evaluate non-inferiority (NI) of 3m compared with 6m (ref)
arm) in high risk stage II patients leading to 3yDFS rate of 82.7 and 83.4% of adj FOLFOX/CAPOX (regimen preselected, not randomized). The primary
for 3M and 6M, respectively (HR: 1.05; 95%CI: 0.68-1.63, p = 0.829). endpoint was disease-free survival (DFS), NI was to be declared if the 2-sided
Similarly, 214 DFS events (161 in 3M and 153 in 6M arm) has been 80% confidence interval (CI) for DFS hazard ratio (HR 3m v 6m) estimated
recorded in stage III patients, leading to a 3yDFS rate of 72.9% in the 3M vs. by a stratified Cox model was below 1.2. 542 DFS events were required to
74.1% in the 6M (HR = 1.06; 95%CI: 0.81–1.42, p = 0.622). For high risk provide 80% power to declare NI. NI was also examined within regimen, T4
stage II patients receiving FOLFOX4, 3yDFS rate was 76.7% in the 3M (Yes v No) and inadequate nodal harvest (Yes v No) as pre-planned subgroups.
vs.79.3% in the 6M (HR = 1.21; 95%CI: 0.54–2.70 p = 0.641). For high Results: The primary analysis included 3273 randomised pts of which 1254
risk stage II patients receiving CAPOX 3-year DFS rate was 85.4% in the 3M had FOLFOX and 2019 had CAPOX. There were 552 events and the median
vs. 83.8% in the 6M (HR = 0.99; 95%CI: 0.59–1.67 p = 0.968). For stage follow-up was 60.2 m. There was significantly less grade 3-5 toxicity with 3m
III CC patients receiving mFOLFOX6, 3-year DFS rate was 71.5% in the 3M treatment (p , .0001). The 5-year DFS rate was 80.7% and 84.0% for 3m and
vs.77.3% in the 6M (HR = 1.18; 95%CI: 0.74–1.86 p = 0.479). For stage III 6m treatment with an estimated DFS HR of 1.18 (80% CI:1.05-1.31, p for
CC patients receiving CAPOX 3-year DFS rate was 74.5% in the 3M vs. 74.7% NI = 0.404). For CAPOX the estimated HR was 1.02 (80% CI: 0.88-1.17, p for
in the 6M (HR = 0.99; 95%CI: 0.70–1.44 p = 0.991). Conclusions: The NI = 0.087) and for FOLFOX the estimated HR was 1.42 (80% CI: 1.19-1.70,
results of the HORG-IDEA study are in line with those of the global IDEA p for NI = 0.894). The test for interaction between duration and regimen was not
project, indicating that 3yDFS is depended on the administered adjuvant statistically significant (p = .174 adjusted for multiple testing) but was stronger
regimen, and the choice of regimen and duration should be personalized. than that for the other subgroups examined. Conclusions: In the overall
Clinical trial information: NCT01308086. population non-inferiority for 3m adj treatment in pts with high-risk stage II
CC was not shown. As with the stage III population the choice of adj regimen
appears important (although this did not reach statistical significance) with a
small difference in DFS between 3 and 6 m treatment if CAPOX is used.
Clinical trial information: ISRCTN59757862.

3502 Oral Abstract Session, Sat, 3:00 PM-6:00 PM 3503 Oral Abstract Session, Sat, 3:00 PM-6:00 PM
Re-evaluating disease-free survival (DFS) as an endpoint versus overall Association of colon cancer (CC) molecular signatures with prognosis and
survival (OS) in adjuvant colon cancer (CC) trials with chemotherapy +/- oxaliplatin prediction-benefit in the MOSAIC Trial (Multicenter International
biologics: An updated surrogacy analysis based on 18,886 patients (pts) Study of Oxaliplatin/5FU-LV in the Adjuvant Treatment of Colon Cancer). First
from the Accent database. First Author: Qian Shi, Department of Health Author: Katherine L. Pogue-Geile, NSABP/NRG Oncology, Pittsburgh, PA
Science Research, Mayo Clinic, Rochester, MN
Background: MOSAIC and C-07 showed that oxaliplatin (OX) added to 5-
Background: DFS with 3 years median follow-up (3yDFS) was validated as a fluorouracil plus leucovorin significantly improved disease free survival
surrogate for OS with 5 years median follow-up (5yOS) in adjuvant chemo- (DFS). However, OX is associated with neurotoxicity and the vast majority of
therapy CC trials prior. Recent data showed improved survival after recurrence patients do not receive OX-benefit; highlighting the importance of an OX-
and OS, over time, in pts who received adjuvant FOLFOX. Hence, re-evaluation benefit predictor. In C-07, colon tumors with a CRCA (Colorectal Cancer
of the association between DFS and OS, as well as the optimal follow-up of OS Assigner) stem-like subtype were associated with a poor prognosis and no
to aid its utility in future adjuvant trials is needed. Methods: Individual patient OX-benefit, but stage III patients with an enterocyte subtype did. We tested
data from 8 randomized adjuvant studies conducted from 1998-2009 were the association of CRCA and CMS molecular subtypes with prognosis and OX-
included; 3 trials tested anti-VEGF or anti-EGFR agents. Trial-level surrogacy benefit in stage III MOSAIC patients. Also, recombination proficiency scores
examining the correlation of treatment effect estimates (i.e. hazard ratios) of (RPS) were tested for the same associations. RPS scores quantify the ef-
3yDFS and 5y to 8yOS was evaluated using both linear regression (R2WLS) and ficiency of DNA-damage repair. Low RPS have been associated with inferior
Copula bivariate (R2Copula) models. For the R2, a value closer to 1 indicates a overall survival (OS) in non-small cell lung carcinoma patients treated with
stronger correlation. Prespecified criteria for surrogacy required either R2WLS or surgery alone but better OS for patients who received surgery plus che-
R2Copula $ 0.80 and neither , 0.7, with lower-bound 95% Confidence Interval motherapy. Methods: Gene expression profiles from 590 stage III pts with
(CI) . 0.60. The rank correlation coefficient (r) quantified the individual-level follow up were successfully profiled by a custom designed nCounter code set.
surrogacy. Results: Total of 18,886 pts were analyzed, with median age 60, CRCA subtypes were determined by a locked down algorithm based on a re-
54% male, 83% stage III, 59% . 12 nodes examined. Median follow-up for estimated centroid using 72 genes (Song et al 2016). CMS subtypes were
survival ranged from 5 to 10 years across trials. Trial level correlation between determined by modified single sample predictor (SSP) using 84 genes. RPS
3yDFS and OS remained strong (R2WLS $0.74; R2Copula $ 0.89) and increased scores were determined as previously described by Pitroda et al (2014).
as the median follow-up of OS extended longer (see table). Analyses limited Signature predictions were made while blinded to clinical outcome and
to stage III pts and/or trials tested biologics showed consistent results. signature performance was evaluated while blinded to gene expression.
Conclusions: 3yDFS remains a validated surrogate endpoint for 5yOS in ad- Results: The stem-like subtype was associated with a very poor prognosis
juvant trials in CC pts per prespecified criteria. The correlation was (Stem-like vs others HR=1.56, p,0.01) and no OX-benefit. CRCA and CMS
strengthened with more than 6 years of follow-up for OS. subtypes did not associate with OX-benefit. Using a median cut point, stage
III patients with low RPS scores received significant OX-benefit (HR=0.67,
N trials (N pts) R2WLS (95% CI) R2Copula (95% CI) r (95% CI) p=0.033, N=290) and patients with high RPS scores did not (HR=1.2,
5yOS 8 (18,886) .74 (.54, .94) .89 (.77, 1.0) .94 (.88, 1.0) p=0.32, N=300) with significant interaction p=0.025. Conclusions: The
6yOS 8 (18,886) .86 (.74, .98) .93 (.85, 1.0) .96 (.92, 1.0) observation in C-07, that the stem-like subtype has the poorest prognosis
6.5yOS 7 (15,700) .93 (.86, 1.0) .94 (.88, 1.0) .97 (.94, 1.0) and did not receive OX-benefit, was validated in MOSAIC, identifying pa-
7yOS 6 (13,088) .92 (.76, 1.0) .95 (.89, 1.0) .97 (.94, 1.0) tients who need new therapies. RPS scores may help to identify the subset of
8yOS 4 (8,435) .87 (.62, 1.0) .97 (.92, 1.0) .99 (.96, 1.0)
patients with OX-benefit in stage III CC, confirmation of this observation is
currently being investigated in C-07. Support: PA DoH; NSABP; Sanofi.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
192s Gastrointestinal (Colorectal) Cancer

3504 Oral Abstract Session, Sat, 3:00 PM-6:00 PM 3505 Oral Abstract Session, Sat, 3:00 PM-6:00 PM
FOxTROT: an international randomised controlled trial in 1052 patients NRG-GI002: A phase II clinical trial platform using total neoadjuvant
(pts) evaluating neoadjuvant chemotherapy (NAC) for colon cancer. First therapy (TNT) in locally advanced rectal cancer (LARC)—First experimental
Author: Matthew T. Seymour, National Institute for Health Research Clinical arm (EA) initial results. First Author: Thomas J. George, NRG Oncology, and
Research Network, Leeds, United Kingdom The University of Florida Health Cancer Center, Gainesville, FL
Background: NAC is well established in many solid tumours but has not Background: This NCTN multi-arm randomized phase II modular clinical trial
undergone large-scale evaluation in colon cancer. Methods: Pts had operable, platform utilizes TNT with parallel EAs in LARC. EAs are not intended for direct
non-obstructed colon cancer; CT-predicted stage T3-4, N0-2, M0, and were fit comparison, but rather to test a variety of hypotheses in a consistent high-risk
for FOLFOX and surgery. They were randomised 2:1 to the novel sequence (6 pt population with correlative biomarkers. Primary endpoint (EP) and available
wk FOLFOX NAC, then surgery, then 18 wk FOLFOX) or control (surgery then secondary EPs from the first EA using veliparib (a PARPi) are reported.
24 wk FOLFOX). RAS-wt pts allocated to the novel arm could optionally be sub- NCT02921256. Methods: Stage II/III pts with LARC (with any ONE of the
randomized 1:1 to 6 panitumumab (pan) during the NAC phase. Two "dealer’s following: distal location [cT3-4 #5cm from anal verge, any N]; bulky [any cT4
choices" allowed total chemo duration 12 wk instead of 24 (in older/low-risk or tumor within 3mm of mesorectal fascia]; high risk for metastatic disease
pts) and OxCap in place of FOLFOX (except in pts randomized 6 pan). Primary [cN2]; or not a sphincter-sparing surgery [SSS] candidate) were randomized to
endpoint is freedom from recurrent or persistent disease after 2 yrs, by ITT. neoadjuvant FOLFOX (x 4mo) → chemoRT (cape with 50.4Gy +/- veliparib
Secondary endpoints include safety, histological stage, completeness of re- 400mg PO BID) → surgery 8-12 wks later. Primary EP: 4 point reduction in
section, OS. Results: 1052 pts were randomised, Jun 2008-Dec 2016, at 85 Neoadjuvant Rectal Cancer (NAR) score with a one-sided a = 0.10 and 80%
centres in UK, Denmark and Sweden. Conclusions: NAC was well tolerated and power. NAR compared by linear model controlling for stratification and pos-
safe, with no increase in perioperative morbidity and a trend toward fewer sibly other factors. Secondary EPs: OS, DFS, toxicity, pCR, cCR, therapy
serious postoperative complications. Evidence of histological regression was completion, negative surgical margins, and SSS. Binary EPs compared by
seen in 59% pts after NAC, including some pCRs. This resulted in marked Fisher’s exact test. Reported p-values are two-sided. Results: From 10/2016 -
histological downstaging and a halving of the rate of incomplete resections. We 2/2018, 178 pts were randomized (88 control, 90 veliparib). Baseline
observed an improvement in 2-yr failure rate (HR=0.77), but this fell short of characteristics were balanced except for candidate for SSS at entry (39%
statistical significance (p=0.11). NAC for colon cancer improves surgical control, 61% veliparib). 140 pts were evaluable for NAR (72 control, 68
outcomes and can now be considered as a treatment option; longer follow-up veliparib). Mean NAR was 12.6 control (95% CI: 9.8–15.3) vs 13.7 for
and further trials are required to confirm the long-term benefits, refine its use veliparib (CI: 10.2–17.2). Controlling for stratification (p = 0.69) or stratifi-
and optimise case selection. Clinical trial information: 87163246. cation and candidate for SSS (p = 0.78), NAR difference was not significant.
pCR = 21.6% vs 33.8% (p = 0.14); cCR = 28.2% vs 33.3% (p = 0.60); and
Novel Control
SSS = 52.5% vs 59.3% (p = 0.43). Most common grade 3/4 AEs were diarrhea
Pts 698 354 and cytopenias. The EA had two deaths (cardiac arrest [FOLFOX]; enterocolitis
Received ‡1 cycle NAC 674 (97%) - [chemoRT]). Conclusions: Veliparib added to chemoRT as part of TNT was safe
Primary resection attempted 684 (98%) 349 (99%)
*Incomplete resection (R1, R2 or nil) 33 (5%) 35 (10%) p=0.001 and without unexpected short-term toxicities but failed to improve the NAR
*pCR 25 (4%) 0 p,0.0001 score. Support: U10CA180868, -180822; UG1-189867; U24-196067;
*pT stage: £2;3;4 16;64;20% 6;64;30% MH,0.0001 AbbVie. Clinical trial information: NCT02921256.
*pN stage: 0;1;2 60;25;15% 48;25;27% MH,0.0001
*Complication prolonging postop stay 79 (12%) 48 (14%) p=0.29
*Anastomotic leak 22 (3%) 20 (6%) p=0.08
2-yr failure (relapse/persistent dis.) 98 (14%) 62 (18%) HR=0.77 p=0.11
(items marked * provisional pending final data checks)

3507 Oral Abstract Session, Sat, 3:00 PM-6:00 PM 3508 Oral Abstract Session, Sat, 3:00 PM-6:00 PM
Randomized phase III study comparing FOLFOX + bevacizumab versus Updated results of TRIBE2, a phase III, randomized strategy study by GONO in
folfoxiri + bevacizumab (BEV) as 1st line treatment in patients with metastatic the first- and second-line treatment of unresectable mCRC. First Author: Chiara
colorectal cancer (mCRC) with $3 baseline circulating tumor cells (bCTCs). Cremolini, Department of Translational Research and New Technologies in
First Author: Javier Sastre, Hospital Clı́nico San Carlos, Instituto de Investigación Medicine and Surgery, Unit of Medical Oncology 2, Azienda Ospedaliera
Hospital Clinico San Carlos (IdISSC), Madrid/Spain, CIBERONC, Madrid, Spain Universitaria Pisana, Pisa, Italy
Background: FOLFOXIRI+BEV has demonstrated a survival benefit com- Background: In the phase III TRIBE study FOLFOXIRI/bev significantly improved
pared with FOLFIRI plus BEV (TRIBE Lancet Oncol 2015) in first-line Response Rate (RR), PFS and OS when compared with FOLFIRI/bev as initial
mCRC. Nevertheless, due to its safety profile, this schedule is not recom- treatment of mCRC. However, the actual advantage by the triplet could be lower when
mended for all pts. In addition, we have showed that the detection of $3 compared with a pre-planned sequential strategy of doublets (FOLFOX, FOLFIRI).
bCTCs is a poor prognostic factor for survival (MACRO The Oncologist 2012). TRIBE2 (NCT02339116) is a phase III trial in which unresectable mCRC pts were
The VISNU-1 trial compares FOLFOX + BEV vs FOLFOXIRI + BEV in pts with randomized 1:1 to FOLFOX/bev followed by FOLFIRI/bev after PD (arm A) or
FOLFOXIRI/bev followed by the reintroduction of the same regimen after PD (arm B).
mCRC and $3 bCTCs. Progression-free survival (PFS) is the primary end-
A pre-planned interim analysis showed a significant advantage for arm B in terms of
point. Secondary endpoints included overall response rate (ORR) and overall
PFS2, primary endpoint of the study, defined as the time from randomization to PD on
survival (OS). Methods: This is an open, multicentric, randomized phase III any treatment given after first PD or death (PD2). Methods: The study had 80% power
trial. Patients with mCRC younger than 70 years, ECOG 0-1 were randomized to detect a HR for PFS2 of 0.77 in favor of arm B with an overall 2-sided-a error of
to FOLFOX+BEV (arm A) or FOLFOXIRI+BEV (arm B), stratified per KRAS 0.05 (0.0131 and 0.0455 for the interim and final analyses, planned at 303 and 466
mutation (mutated vs WT) and number of involved organs (1 vs .1). PFS2 events, respectively). Secondary endpoints included RR, 1st-PFS, i.e. the time
Results: 349 pts were included in the ITT population; 177 in group A and from randomization to the first evidence of PD or death (PD1), 2nd-PFS, i.e. the time
172 in group B. Characteristics of the pts, molecular profiling and safety from PD1 to PD2, and OS. Results: From February 2015 to May 2017, 679 pts (arm
analysis have been previously presented at ASCO 2018 and showed that this A/B: 340/339) were enrolled in 58 Italian sites. Main pts’ characteristics were (arm A/
schedule had an acceptable toxicity profile. Efficacy analysis in the ITT B): right side 38%/38%, synchronous mets 89%/89%, RAS mutant 65%/63%,
population is shown in the table. Conclusions: In this population with very BRAF mutant 10%/10%. At a median follow up of 30.6 mos, 514 (arm A/B: 272/
bad prognosis, our study met its primary endpoint. Pts who received FOL- 242) PD2, 594 (arm A/B: 303/291) PD1 and 408 (arm A/B: 217/191) OS events
FOXIRI + Bev benefit for a statistically significative PFS and ORR. OS were collected. A significant advantage by upfront FOLFOXIRI/bev was confirmed in
showed a trend of benefit in the experimental arm. According to these re- terms of PFS2 (19.1 vs 16.4 mos, HR 0.74, 95%CI 0.62-0.88, p,0.001), RR (62%
sults, FOLFOXIRI-Bev could be considered an adequate treatment option for vs 50%, OR 1.61, 95%CI 1.19-2.18, p=0.002) and 1st-PFS (12.0 vs 9.8 mos, HR
pts with mCRC and $3 bCTCs. Clinical trial information: 2012-000846-37. 0.75, 95%CI 0.63-0.88, p,0.001). A significant OS benefit for pts in arm B was also
observed (27.6 vs 22.6 mos, HR: 0.81, 95%CI: 0.67-0.98, p=0.033). Out of 594
FOLFOX + Bev FOLFOXIRI + Bev pts with a PD1 event, 470 (79%, arm A/B: 251/219) received a treatment after PD. In
Variable n=177 n= 172 p value the per-protocol analysis (N=323), pts in arm B showed significantly longer 2nd-PFS
(6.5 vs 5.8 mos, HR 0.76, 95%CI 0.59-0.97, p=0.024). Conclusion: Upfront
PFS (median in m) 9.3 12.4 P=0.0004
ORR (%) 52.0 59.0 0.1685 FOLFOXIRI/bev followed by the pre-planned reintroduction of the same agents after
OS (median in m) 17.6 21.7 0.862 PD provided a statistically significant and clinically relevant PFS2 and OS benefit
when compared with the pre-planned sequential administration of FOLFOX/bev and
FOLFIRI/bev in unresectable mCRC patients. A median OS of 27.6 mos was reached
despite the high percentage of pts with poor prognostic features (RAS and BRAF
mutations, right side, synchronous mets). Clinical trial information: NCT02339116.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Colorectal) Cancer 193s

3509 Poster Discussion Session; Displayed in Poster Session (Board #1), 3510 Poster Discussion Session; Displayed in Poster Session (Board #2),
Mon, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Mon, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Mon, 11:30 AM-1:00 PM Mon, 11:30 AM-1:00 PM
A randomized phase II trial of second-line CAPTEM versus FOLFIRI in MGMT A multicenter randomized phase III trial of capecitabine with or without irino-
methylated, RAS mutated metastatic colorectal cancer (mCRC) patients. tecan driven by UGT1A1 in neoadjuvant chemoradiation of locally advanced
First Author: Filippo Pietrantonio, Fondazione IRCCS Istituto Nazionale dei rectal cancer (CinClare). First Author: Zhen Zhang, Department of Radiation
Tumori, Milan, Italy Oncology, Fudan University Shanghai Cancer Center, Shanghai, China
Background: Overall response rate (ORR) to temozolomide (TMZ) is ~10% in re- Background: Our phase I/II study identified irinotecan dose differentiated by
fractory mCRC pts with MGMT methylation detected by qualitative assays, e.g. UGT1A1 genotype in the neoadjuvant CRT and showed improved pCR. The
methylation-specific PCR (MSP). ORR to irinotecan/FOLFIRI in second-line trials was objective of this phase III study was to further investigate irinotecan combined
4-16%. The efficacy of TMZ may be improved by its combinatorial use in earlier lines with capecitabine-based chemoradiotherapy as preoperative treatment for
and molecular selection beyond MSP. Lack of MGMT expression by immunohisto- locally advanced rectal cancer. Methods: We underwent a prospective, ran-
chemistry (IHC) and high MGMT % methylation by MethylBEAMing (MB) are prog- domized, open-label, multicenter, phase 3 trial in China from Nov.2015 to
nostic for higher ORR/PFS in TMZ-treated pts. Methods: This multicenter, randomized Dec.2017. Eligible patients with clinical stage T3-4 and/or N+ rectal ade-
phase 2 trial investigated PFS superiority of second-line CAPTEM (Arm A: capecitabine
nocarcinoma were randomly allocated to two arms. The approach in control
750 mg/sqm bid days1-14/TMZ 75 mg/sqm bid days10-14q28 days) over FOLFIRI
(arm B) in RAS mutated mCRC pts with MGMT methylation centrally confirmed by arm (Arm A, n = 180) was pelvic radiation of 50 Gy/25 fractions with con-
MSP. Eligible pts: ECOG PS 0-1, measurable disease, failure of 1st-line oxaliplatin- current capecitabine 825 mg/m2 twice daily 5 days per week, followed by a
based tx (or relapse within 6 mos from oxaliplatin-based adjuvant tx). Randomi- cycle of XELOX two weeks after the end of CRT. The experimental arm (Arm B,
zation was stratified by time from the start of oxaliplatin-based therapy to n = 180) was pelvic radiation with capecitabine 625 mg/m2 twice daily 5 days
PD ( , /$9 months); prior bevacizumab (yes/no). A one-sided log-rank test with a per week and combined with weekly irinotecan. The irinotecan dose was used
sample size of 82 pts (41 per arm) achieved 90% power at a 5% significance level to based on UGT1A1 genotype of 80mg/m2 for UGT1A1*1*1 or 65mg/m2 for
detect mPFS increase from 2 to 4 mos. Secondary endpoints: safety, QoL, OS, ORR. UGT1A1*1*28 weekly, followed by a cycle of XELIRI. The primary endpoint is
Exploratory endpoints: predictive value of MGMT IHC/MB. Results: From Nov 2014 to pathological complete response (pCR). This trial was registered with Clin-
Feb 2019, 82 pts (arm A/B: 41/41) were enrolled in 18 Italian sites. Baseline icalTrials.gov, number NCT02605265. Results: Surgery was performed in
characteristics (arm A/B): males 44/56%, median age 70/67, ECOG PS 0 54/51%, 86.5% and 88.2% of patients in two groups, with 38.9% and 30.5% of
right-sidedness 37/39%, 1 metastatic site 44/34%, prior bevacizumab 68/66%, 1st- patients got abdominoperineal resection respectively. The pCR rate was
line PFS 9,4/10,2 months. At a median follow up of 26.6 mos, 70 PFS/46 OS events 17.5% in Arm A and 33.8% in Arm B (P = 0.001). Four and 6 patients
were collected. The mPFS was 3.6 vs 4.1 mos in arm A vs B (HR = 1.26;95%CI 0.78- maintained a complete clinical response status at least 12 months and were
2.02;p = 0.34) and mOS was 9.1 vs 14.2 mos (HR =1.08;95%CI 0.60-1.94;p = marked as cCR. The CR rate, including pCR and cCR, was 17.4% in Arm A and
0.79). ORR and DCR (arm A/B): 12/10% and 51/51%. Grade 3-4 adverse events: 15/ 33.1% in Arm B (P = 0.001). The most common grade 3-4 adverse events
44% (diarrhea 0/12%, stomatitis 0/7%, anemia 2/10%, neutropenia 2/22%,
during preoperative treatment were leucopenia (3.4% vs. 25.3%), neutropenia
thrombocytopenia 7/0%). Neither MGMT IHC nor MB status were prognostic.
MGMT IHC positive subgroup, arm A (n = 12) vs arm B (n = 22): mPFS, 2.0 vs 4.1 mos (1.7% vs. 19.7%) and diarrhea (1.7% vs. 13.5%) in two arms. The overall
(HR = 2.06;95%CI 0.96-4.45;p = 0.06), mOS, 6.4 vs 10.6 mos (p = 0.78), ORR (0% rate of surgical complications were not significantly different between arms
vs 14%) and DCR (25% vs 55%;OR = 0.28;95%CI 0.06-1.31;p = 0.11). In MGMT (11.0% vs. 14.6%). Conclusions: Adding irinotecan guided by UGT1A1 to
IHC negative subgroup, no PFS/OS/ORR/DCR differences were noted between the two capecitabine-based neoadjuvant chemoradiotherapy significantly increases
arms. P interaction IHCxArm: 0.171 for PFS, 0.917 for OS, 0.06 for DCR. Similar complete tumor response. The treatment toxicities were increased but toler-
accuracy was achieved by MB. Conclusions: The use of TMZ should be explored by able. This treatment can be as an option for ‘watch and wait’ approach. Clinical
phase 3 trials enrolling MGMT IHC-negative +/- high MGMT % methylated mCRC. trial information: NCT02605265.
Clinical trial information: NCT02414009.

3511 Poster Discussion Session; Displayed in Poster Session (Board #3), 3512 Poster Discussion Session; Displayed in Poster Session (Board #4),
Mon, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Mon, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Mon, 11:30 AM-1:00 PM Mon, 11:30 AM-1:00 PM
Final results and OS of the randomized phase II VOLFI trial (AIO- KRK0109): CCTG CO.26: Updated analysis and impact of plasma-detected microsatellite
mFOLFOXIRI + panitumumab versus FOLFOXIRI as first-line treatment in stability (MSS) and tumor mutation burden (TMB) in a phase II trial of
patients with RAS wild- type metastatic colorectal cancer (mCRC). First durvalumab (D) plus tremelimumab (T) and best supportive care (BSC) versus
Author: Michael Geissler, Klinikum Esslingen, Department of Internal BSC alone in patients (pts) with refractory metastatic colorectal carcinoma
Medicine, Oncology/Hematology, Gastroenterology, Esslingen, Germany (rmCRC). First Author: Eric Xueyu Chen, Princess Margaret Cancer Centre,
University Health Network, Toronto, ON, Canada
Background: This trial evaluated activity and safety of mFOLFOXIRI + panitumumab
vs FOLFOXIRI in ECOG 0-1, primarily non-resectable mCRC patients. The final Background: Targeting both PD-L1 and CTLA-4 may be synergistic immu-
primary endpoint was presented at ASCO and ESMO 2018. Now we report for the notherapy approaches. CO.26 evaluated if dual inhibition leads to improved pt
first time the final results regarding OS and PFS. Methods: Prospective 2:1 ran- survival vs BSC alone in rmCRC. Methods: rmCRC pts were randomized 2:1 to
domized, controlled, open label multi-center, phase II trial comparing mFOLFOXIRI D+T vs BSC. Treatment consisted of D (1500 mg) D1 q 28 days and T (75 mg)
(Ox 85 mg/m2, Iri 150 mg/m2, 5-FU 3000mg/m2 cont. 48h, LV 200 mg/m2) + D1 for first 4 cycles, and supportive measures. Primary endpoint was overall
Panitumumab 6 mg/KG (arm A) with FOLFOXIRI (Ox 85 mg/m2, Iri 165 mg/m2, 5-
survival (OS). Two-sided p , 0.10 was considered statistically significant. Cell-
FU 3200mg/m2 cont. 48h, LV 200 mg/m2; arm B), both arms q2w. Prospective
free (cf)DNA sequencing for MSI and TMB used GuardantOMNI panel and
strata were cohort 1: irresectable mCRC (n=65), and cohort 2: chance of secondary
resection of metastatic lesions (n=31). Primary endpoint was ORR, secondary
baseline plasma. Results: From 08/2016-06/2017, 180 pts were enrolled. Pt
endpoints were secondary resection rate (cohort 2), DCR, PFS, OS, toxicity, quality of characteristics were balanced between arms. At median follow-up of
life (QLQ-C30). Financially supported by an unrestricted grant from Amgen. 15.2 months (mos), median OS was 6.6 mos for D+T and 4.1 mos for BSC (p =
Results: A total of 96 patients were randomized (63 arm A, 33 arm B). ORR was 0.07; Hazard ratio (HR): 0.72, 90% confidence interval (CI): 0.54 – 0.97).
87.3% in arm A and 60.6% in arm B (p=0.0041, OR 4.47; 95%-CI 1.614- Progression free survival (PFS) was 1.8 mos vs 1.9 mos, respectively (HR 1.01,
12.376). Secondary resections of metastases in the ITT population were observed in 90% CI 0.76 – 1.34). Disease control rate (DCR) was 22.6% for D+T and
33$3% (arm A Pmab) versus 12$1% (arm B) (OR=3.63; 95%-CI 1.13–11.67, 6.6% for BSC (p = 0.006). cfDNA analysis was successful in 168/169 pts
p=0$029) and in cohort 2 in 75% (arm A Pmab) versus 36.4% (arm B) (OR=5.25; (99.4%). Two pts were MSI-high. In 166 MSS pts, OS HR was 0.66 (p=0.024;
95%-CI 1.07–25.8, p=0.05), respectively. Median PFS was similar in the study 90% CI 0.49-0.89). Excluding the MSI-H cases (TMB of 74.7 and 247.1 mts/
arms (9.7 mo in both arms, HR 1.071; 95%-CI 0.689-1.665, p=0.76). OS in the Mb), mean TMB was 20.4 6 16.3 mts/Mb (range: 0.96 – 114.0). In MSS
ITT population showed a strong trend in favour of the Pmab-containing arm A with a pts, a pre-specified cutpoint of 20 mts/Mb stratified pts into high and low TMB
median OS of 35.7 mo compared to 29.8 mo in arm B (HR: 0$67; 95%-CI 0.41- groups but was not predictive for OS , PFS, or DCR (interaction p-values . 0.7).
1.11, P=0$12). mOS of cohort 2 was 52.0 mo in arm A versus 41.7 mo in arm B (HR Using a minimum p-value approach, pts with TMB .28 mts/Mb (21% of MSS
0.413; 95%-CI 0.15-1.12, p=0.07). Further results regarding to sidedness and pts) had the greatest OS benefit (HR 0.34, 90% CI 0.18-0.63) for D+T
BRAF mutational status will be presented. Conclusions: The addition of Pmab to a
(interaction p = 0.07). High TMB was associated with a trend in worse
mFOLFOXIRI regimen in patients with RASwildtype metastatic colorectal cancer
prognosis for OS in the BSC arm using both 20 mts/Mb (HR 1.26, 90% CI
significantly improved objective response rate and the rate of secondary resection of
metastases. Although PFS was comparable, there was a strong trend towards im-
0.76-2.12) and 28 mts/Mb (HR 2.59 90% CI 1.46-4.62) cutpoints.
proved OS in the Pmab arm. Future studies are warranted to confirm this trend Conclusions: D+T significantly prolonged OS in pts with rmCRC. High TMB
towards improved overall survival with this regimen. Clinical trial information: may select a group of MSS pts who benefit from D+T. Plasma TMB appeared
NCT01328171. prognostic in the BSC arm. This is the first study showing combined PD-L1 and
CTLA-4 inhibition prolongs survival in pts with MSS rmCRC. Updated results
based on deaths in more than 90% of pts will be presented. Clinical trial
information: NCT02870920.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
194s Gastrointestinal (Colorectal) Cancer

3513 Poster Discussion Session; Displayed in Poster Session (Board #5), 3514 Poster Discussion Session; Displayed in Poster Session (Board #6),
Mon, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Mon, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Mon, 11:30 AM-1:00 PM Mon, 11:30 AM-1:00 PM
Validation of the Immunoscore prognostic value in stage III colon cancer A phase II study of ipilimumab and nivolumab with radiation in microsatellite
patients treated with oxaliplatin in the prospective IDEA France cohort study stable (MSS) metastatic colorectal adenocarcinoma (mCRC). First Author:
(PRODIGE-GERCOR). First Author: Franck Pages, INSERM, Laboratory of Aparna Raj Parikh, Massachusetts General Hospital, Boston, MA
Integrative Cancer Immunology, Equipe Labellisée Ligue Contre le Cancer,
Background: mCRC remains a lethal cancer and immunotherapy in MSS mCRC
Paris, France
has yet to show significant activity. In preclinical models, radiation induced
Background: The Immunoscore (IS), which has been shown to prognostically cellular damage may increase responsiveness to immunotherapy via the abscopal
classify Stage I-III colon cancer (CC) patients, was assessed in the IDEA effect, with evidence for synergy between radiation therapy (RT) and dual
France cohort study evaluating 3 versus 6 months of oxaliplatin-based ad- checkpoint blockade. In this study, we assessed dual blockade of CTLA-4 and PD-
juvant chemotherapy in Stage III CC patients. Methods: Densities of CD3+ 1 combined with RT as a strategy to stimulate an immune response for patients
and cytotoxic CD8+ T cells in the tumor and invasive margin of each patient with MSS mCRC. Methods: In this open-label, single arm phase-2 study, we
were quantified by digital pathology and converted to IS using pre-defined enrolled 40 MSS mCRC patients (pts). Eligible pts had histologically-confirmed
published cut-off. The performance of IS to predict disease-free survival MSS mCRC, ECOG PS 0-1, and progression on at least two lines of therapy.
(DFS) was assessed in the modified intention-to-treat population, in each Treatment (Tx) consisted of ipilimumab (1mg/kg q6 weeks), nivolumab (240 mg
study arm, and was adjusted with relevant clinical features in multivariable q2 weeks) and 3 fractions of 8 Gy of RT at cycle 2 every other day. Tx continued
until progression, discontinuation or withdrawal. The primary endpoint was
Cox models. Harrell’s C-statistics was used to investigate the IS performance.
Disease Control Rate (DCR), with radiological evaluations every 3 months. Ex-
Results: 1322 patients were included; 82 were excluded due to pre-analytical
ploratory endpoints included ORR, PFS, OS and safety. Response was defined as
non-conformity. IS was successfully analyzed in 1062 (85.6%) eligible pa-
disease control outside the radiation field. We obtained serial tumor biopsies pre-
tients. In a 2-category IS analysis, Low and (Int+High) IS were observed in tx, during checkpoint blockade alone (cycle 1) and 2 weeks after initiation of
n=599 (43.6%) and n=463 (56.4%) patients, respectively. IS was signifi- radiation. Intention-to-treat analysis (ITT) includes all pts receiving at least one
cantly correlated with T stage, T/N stage (T1-3 and N1 versus T4 and/or N2), dose of study agent. Results: 40 pts (median age 59 years (26-83), 58% male)
and microsatellite instability status. The study met its primary objective of enrolled and started treatment from 7/2017 to 12/2018. DCR was 17.5% (7/40)
validating that Low IS identifies patients with higher-risk of relapse or death with a 7.5% (3/40) ORR by ITT. Median duration of disease control was 77 days in
[HR=1.54; 95%CI 1.24-1.93, p=0.0001]. The 3-year DFS rates were the ITT; 252 days for those with disease control (n=7) based on the first re-staging
66.80% [95%CI 62.23-70.95] and 77.14% [95%CI 73.50-80.35] for Low scans at 3 months or censored (n=3) and 70.5 days for pts with PD (n=17) or who
IS and (Int+High) IS, respectively. In multivariable analysis, IS remained did not receive RT due to clinical progression (n=13). In the modified ITT (all pts
independently associated with DFS (p,0.0012) when combined with T/N receiving RT and restaged), N=24 pts, excluding 1 pt pending 1st scans post-RT,
stage. The addition of IS to the T/N stage significantly improved the model DCR was 29.2% (7/24) and ORR 12.5% (3/24). Median duration of disease
discrimination capacity [bootstrap C index mean difference, 0.022; 95%CI control in mITT was 77.5 days: 252 days for those with disease control and
0.005-0.04]. In addition, IS in 3 categories (Low, Int, High) and as a con- 77 days for those with PD. TRAEs were reported in 22/40 (55%). 20/40 (50%)
tinuous variable were also both significantly associated with DFS (all with grade $3 toxicities, with fatigue, nausea, vomiting, diarrhea, infusion-related
p,0.001). In univariable analysis, IS was also associated with DFS in reaction and dyspnea being the most common. 1(2%) pt died of respiratory failure
6 months arm (p,0.0001); a similar trend was observed in 3 months arm possibly related to tx. Conclusions: Dual blockade of CTLA-4 and PD-1 with RT is
(p=0.09). Conclusions: IS was confirmed as a prognostic factor of DFS in feasible and demonstrates durable activity in pts with MSS mCRC. There are 3 pts
Stage III CC patients in the prospective IDEA France cohort study. Clinical trial who have not completed RT or had their post-RT re-staging. We will report updated
information: NCT03422601. efficacy data and outcomes from correlative serial tumor biopsies upon trial
completion. Clinical trial information: NCT03104439.

3515 Poster Discussion Session; Displayed in Poster Session (Board #7), LBA3516 Poster Discussion Session; Displayed in Poster Session (Board #8),
Mon, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Mon, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Mon, 11:30 AM-1:00 PM Mon, 11:30 AM-1:00 PM
A randomized controlled trial of the conventional technique versus the Long-term survival after laparoscopic versus open resection for colorectal
no-touch isolation technique for primary tumor resection in patients with liver metastases. First Author: Åsmund Avdem Fretland, Oslo University
colon cancer: Primary analysis of Japan Clinical Oncology Group study Hospital, Oslo, Norway
JCOG1006. First Author: Yasumasa Takii, Department of Gastroenterolog-
ical Surgery, Niigata Cancer Center Hospital, Niigata, Japan
Background: The no-touch isolation technique (NTIT) aims to reduce cancer
cells flowing from the primary tumor site to the liver and other organs by first
ligation of blood vessels that feed the primary tumor. The efficacy of NTIT has
not been proved in previous studies. To acquire the answer of an unsolved
problem more than 60 years, we conducted a phase III trial to confirm the
superiority of NTIT in patients with cT3/T4 colon cancer. This is the primary
analysis of the disease-free survival (DFS) as the primary endpoint.
Methods: Eligibility criteria included histologically proven colon cancer;
tumor located in the cecum, ascending, transverse, descending, sigmoid or
rectosigmoid colon; clinical T3 or T4, N0-2, M0; patients age 20-80 years. The full, final text of this abstract will be available at
Patients were randomized preoperatively to either conventional technique
(CoT) arm or NTIT arm. Operation was performed in open surgery. Patients abstracts.asco.org at 7:30 a.m. ET on Monday, June 3.
with pathological stage III received adjuvant chemotherapy with capecita- Onsite at the Meeting, this abstract will be printed in the
bine. The primary endpoint was DFS. Planned sample size was 850 to Monday edition of ASCO Daily News.
detect a hazard ratio (HR) of 0.732 in DFS with one-sided alpha of 5% and
power of 80%. Results: A total of 853 patients were randomized (CoT: 427,
NTIT: 426) between January 2011 and November 2015. The 3-year DFS
were 77.3% and 76.2% in the CoT arm and NTIT arm,respectively. The HR was
1.029(95% CI 0.800-1.324); thus the superiority of NTIT was not confirmed
(p = 0.59). The 3-year overall survival (OS), 3-year recurrence-free survival
(RFS) and 3-year liver-recurrence-free survival (LRFS) are shown in the Table.
Conclusions: The superiority of NTIT to CoT was not confirmed. NTIT does not
improve the DFS or OS, RFS, LRFS in the patents with stage II and III colon
cancer. Clinical trial information: UMIN000004957.
CoT (95% CI) NTIT (95% CI) HR (95% CI)
3-year DFS 77.3% (73.1-81.0%) 76.2% (71.9-80.0%) 1.029 (0.800-1.324)
3-year OS 94.8% (92.2-96.5%) 93.4% (90.6-95.4%) 1.006 (0.674-1.501)
3-year RFS 81.8% (77.8-85.2%) 79.0% (74.8-82.6%) 1.220 (0.919-1.620)
3-year LRFS 89.5% (86.2-92.1%) 87.0% (83.4-89.8%) 1.226 (0.877-1.714)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Colorectal) Cancer 195s

3517 Poster Discussion Session; Displayed in Poster Session (Board #9), 3518 Poster Discussion Session; Displayed in Poster Session (Board #10),
Mon, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Mon, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Mon, 11:30 AM-1:00 PM Mon, 11:30 AM-1:00 PM
Repeated centralized MDT resectability assessment during first-line treatment A pooled analysis of multicenter cohort studies of post-surgery circulating
in 1086 Finnish metastatic colorectal cancer (mCRC) patients nationwide tumor DNA (ctDNA) in early stage colorectal cancer (CRC). First Author:
(prospective RAXO study). First Author: Pia J. Osterlund, Tampere University Jeanne Tie, The Walter and Eliza Hall Institute of Medical Research; Peter
and Central Hospital, Tampere, Finland MacCallum Cancer Centre, Melbourne, Australia
Background: Resection of oligometastatic mCRC is curable in up to half and is Background: Studies in multiple tumor types have demonstrated the prognostic
probably underused in real-life oncology practice. In 2011, a prospective na- impact of ctDNA analysis after curative intent surgery. Emerging data suggest that
tionwide centralized evaluation of resectability was initiated to assess upfront conversion of ctDNA from detectable to undetectable after adjuvant chemo-
and conversion to resectability. Methods: The RAXO-study included 1086 therapy may reflect treatment efficacy. Further review of existing study data could
mCRCs referred for oncological treatments at all 21 Finnish oncology units (40% increase the precision of ctDNA guided adjuvant therapy decision making.
of eligible mCRCs in Finland). The MDT at Helsinki tertiary referral center (TRC) Methods: We combined individual patient (pt) data from three independent
assessed resectability of liver, lung and other metastases upfront and twice prospective cohort studies that enrolled 485 pts with stage II or III CRC. Clinicians
during first-line therapy. Resectability evaluation was provided online and re- were blinded to ctDNA results. We evaluated pt outcomes over a 5-year follow-up
sections were performed according to clinical praxis, with major resections at period (median, 47.2 months). Plasma samples were collected 4 to 10 weeks after
TRCs. Results: Median age was 66.5 (range 24-90) yrs, 60% males, ECOG was surgery. Mutations in ctDNA were assayed using Safe-SeqS. Results: ctDNA was
0 in 295 (27%), 1 in 600 (55%) and 2-3 in 191 (18%). Tumor was right-sided detected after surgery in 59 (12%) pts overall (11.0%, 12.5% and 13.8% re-
in 310 (29%), left in 769 (72%) and multiple in 7 (.6%). In the MDT as- spectively for samples taken at 4-6, 6-8 and 8-10 weeks; P = 0.740). ctDNA
detection was associated with nodal status; 8.7%, 16.7% and 32.4% in N0, N1
sessment, 435 (40%) were considered resectable, 28% upfront and 12% after
and N2 disease (P , 0.001), but remained an independent adverse prognostic
conversion. Resections or local ablative therapy were performed in 396 (91% of
factor in multivariable analysis. ctDNA detection was associated with poor overall
resectable, R0-1 in 74% and R2 in 17%), with 368 liver-, 91 lung- and 96 other survival for pts treated (mortality ratio, 3.0; P = 0.026) or not treated with adjuvant
operations. One metastatic site (n = 407), 2 (n = 369) and 3-6 (n = 310) were or chemotherapy (mortality ratio, 5.17; P , 0.001). The median MAF (mutant allele
became resectable in 65, 33 and 15%, respectively, and R0-1 resected in 54, frequency) in pts with detectable ctDNA was 0.046%. For pts not treated with
30 and 6%, respectively (P , 0.001). In multivariate analysis age , 66.5 (OR adjuvant chemotherapy, 3 year recurrence free survival (RFS) was 9% in pts with a
1.51), ECOG 0 (OR 2.40); left-sided primary (OR 1.55), metachronous me- MAF . 0.046% vs 33% with a MAF # 0.046% (HR, 2.7; P = 0.032). For
tastases (OR 1.49), 1-2 metastatic sites (OR 1.86) increased R0-1 resection chemotherapy treated pts, 3 year RFS was 25% in pts with a MAF . 0.046% vs
rates; whereas primary/relapse in situ (OR 0.49), metastases in lung (OR 0.22), 70% with a MAF # 0.046 (HR, 3.1; P = 0.025). In 90 pts with recurrence, ctDNA
peritoneum (OR 0.30) or lymph nodes (OR 0.23) decreased R0-1 rates (P , had been detected post surgery in 3 of 20 (15%) with locoregional recurrence, 27
0.05). In univariate analyses BRAF wt (OR 0.34), normal CEA (OR 0.31) or of 60 (45%) with distant recurrence and 5 of 10 (50%) with both (P = 0.044).
normal Ca19-9 (OR 0.45) associated with R0-1 rates. Median OS was 2.4 yrs Conclusions: Where samples for ctDNA analysis were collected 4 to 10 weeks post
(CI95% 2.3-2.6) in all, 6.7 yrs in R0-1 (HR 0.15; CI95% 0.11-0.19), 2.8 yrs in surgery, sampling timing may not significantly impact detection rates. The
R2 (0.50; 0.36-0.68) and 1.7 yrs in non-resected (reference group). Median prognostic significance of ctDNA detection can be further stratified by MAF level,
PFS was 1.1 yrs (1.0-1.1) in all patients, 2.4 yrs in R0-1 (0.24; 0.20-0.28), 1.4 but MAF level may not impact adjuvant treatment benefit. ctDNA analysis is most
yrs in R2 (0.55; 0.42-0.71) and 0.8 yrs in non-resected patients. P , 0.001 for sensitive for detecting minimal residual disease at distant sites.
all comparisons. Conclusions: Repeated centralized MDTs in real-life mCRC
patients show high resectability (40%) and resection rates (36%), even if
multiple metastatic sites are present. Curative resection associates with good
survival. Clinical trial information: NCT01531621.

3519 Poster Discussion Session; Displayed in Poster Session (Board #11), 3520 Poster Discussion Session; Displayed in Poster Session (Board #12),
Mon, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Mon, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Mon, 11:30 AM-1:00 PM Mon, 11:30 AM-1:00 PM
Prognostic value of tumor deposits for disease free survival in patients with Relative contribution of clinical and molecular features to outcome within
stage III colon cancer: A post hoc analysis of IDEA France phase III trial low and high risk T and N groups in stage III colon cancer (CC). First Author:
(PRODIGE-GERCOR). First Author: Jean Francois Delattre, Department of Frank A. Sinicrope, Mayo Clinic, Rochester, MN
Medical Oncology, AP-HP, Hôpital Saint-Antoine, Paris, France
Background: Duration of adjuvant FOLFOX or CAPOX for stage III CC is being guided
Background: Tumor deposits (TDs) are isolated tumor foci in the pericolic, by pt stratification into low (T1-3N1) and high (T4 or N2) risk groups based on the
perirectal or mesocolic fat without residual lymph node (LN) tissue. TDs seem IDEA study. We determined the relative contributions of clinical and molecular
to impact the prognosis of stage III colon cancer (CC) patients (pts) but are only features for prediction of time-to-recurrence (TTR) and survival after recurrence
considered in TNM staging in the absence of LN metastases (LNM). We aimed (SAR) within each risk group. Methods: Stage III CC (N=5,430) from 2 trials of
at evaluating the prognosis value for disease free survival (DFS)of TDs in adjuvant FOLFOX 6 cetuximab with similar outcome by study arm [NCCTG N0147
(Alliance), PETACC-8] were used. Tumors were analyzed for mismatch repair (dMMR
International Duration Evaluation of Adjuvant Chemotherapy (IDEA) France
vs pMMR), mutations in KRAS (exon 2) and BRAFV600E. Median pt follow-up was
phase III study (NCT00958737) that compared 3 versus 6 months of adjuvant
83.4 months. Relative contributions to predicting outcome were assessed using x2
FOLFOX or CAPOX for stage III CC pts. Methods: All pathological reports of pts (Harrell’s rms) based on multivariable (MV) Cox models. Results: N (50.8%) and T
included in IDEA France trial were retrospectively analyzed. DFS according to (31.8%) stage were the top two contributors to prediction of TTR which supports risk
the presence or absence of TDs was evaluated using Kaplan-Meier estimator. grouping. High risk (n=2566) vs low risk (n=2774) pts had poorer TTR (HR 2.7, 95%
Multivariable Cox model analysis was performed to evaluate the association CI, 2.4-3.0) and SAR [HR 1.6 (1.4-1.9)], both p,.0001. TTR: KRAS contributed
between TDs and DFS. This analysis did not included immunohistochemical the most to predicting TTR among high (58.6%) and low (51.1%) risk pts (Table).
biomarkers. Results: Among the 2022 pts included in IDEA France study, Contribution of MMR (16%) to predicting TTR was limited to low risk pts. Contri-
1942 (96%) were analyzed. 80 pts were excluded: no pathological report (n = bution of BRAFV600E to TTR was nearly 3-fold increased in high vs low risk pts. SAR:
68), pts without treatment (n = 12). TDs were found in 184 pts (9.47%), of BRAFV600E contributed the most to predicting SAR, especially in high vs low risk pts
whom 74 with N1a/b (40%), 55 with N1c (30%) and 55 with N2 LN stage (2-fold increase). Tumor sidedness and performance status (PS) were key con-
(30%). All characteristics were similar according to the presence of TDs, tributors to SAR, but not TTR. MV associations: TTR: low risk, KRAS [HR 1.7 (1.4-
except for tumor/node (TN) stage (T4 and/or N2 are more frequent in pts with 2.3], MMR [HR 0.55 (.35-.87), gender (M/F) [HR 1.3 (1.0-1.5)], all p,.04]; high
TDs; p = .0046). The 3-year DFS rates were 65.59% [95% confidence interval risk: BRAF [HR 1.3 (1.1-1.7)], sidedness (R vs L) [HR 1.14 (1.0-1.3)], KRAS
(95%CI) 58.04-72.12] and 74.71% [95%CI 72.57-76.71] for pts with and [HR 1.4 (1.2-1.6)], all p,.04]. SAR: BRAF, sidedness, PS (all p,.05).
without TDs, respectively (p = 0.0079). In multivariable analysis, TDs were Conclusions: KRAS mutation was the strongest predictor of shorter TTR in both risk
groups whereas BRAFV600E was the primary driver of SAR, especially in high risk pts.
associated with higher risk of recurrence or death (hazard ratio (HR) = 1.36,
Support: U10CA180821, U10CA180882, U24CA196171; BMS, Pfizer, Sanofi.
95%CI 1.05-1.75, p = .0201), as well as T4 and/or N2 (HR = 2.21, 95%CI
NCT00079274.
1.03-1.59, p , .001), 3 months of adjuvant treatment (HR = 1.29, 95%CI
1.09-1.52, p = .0029), obstruction (HR = 1.28, 95%CI 1.03-1.59, p = % relative contributions.

.0233) and male (HR = 1.24, 95%CI 1.04-1.46, p = .0151). Adding TDs Low Risk (T1-3N1) High Risk (T4 or N2)

count to the LNM count, 35 out of 1454 N1a/b/c CC pts (2.4%) were TTR SAR TTR SAR
reclassified as N2 and experienced worse 3 years DFS than confirmed N1 CC BRAF 6.7% 26.1% 18.5% 54.5%
KRAS 58.6 2.7 51.1 4.9
pts (p = .0151). Conclusions: TD is an independent and valuable prognostic MMR 15.8 2.0 1.6 0.2
factor for DFSin stage III CC pts and should be considered whatever the LNM Sidedness .04 30.7 8.2 34.0
status. Gender 11.7 0.2 0.9 1.1
Age 0.2 8.2 4.1 0.4
PS 7.0 30.1 15.5 7.0

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
196s Gastrointestinal (Colorectal) Cancer

3521 Poster Session (Board #13), Mon, 8:00 AM-11:00 AM 3522 Poster Session (Board #14), Mon, 8:00 AM-11:00 AM
Nivolumab (NIVO) + low-dose ipilimumab (IPI) as first-line (1L) therapy in Recent changes in overall survival of real-life stage IV colorectal cancer
microsatellite instability-high/DNA mismatch repair deficient (MSI-H/dMMR) patients. First Author: Patricia Hamers, University Medical Center Utrecht,
metastatic colorectal cancer (mCRC): Clinical update. First Author: Heinz-- Utrecht, Netherlands
Josef Lenz, USC Norris Comprehensive Cancer Center, Los Angeles, CA
Background: In the past decade, the reported median overall survival (mOS) in
Background: In the phase 2 CheckMate 142 trial, NIVO + low-dose IPI phase 3 trials of metastatic colorectal cancer (mCRC) patients increased from
provided robust and durable clinical benefit and was well tolerated as 1L approximately 16 to over 36 months. However, only 2.5-20% of cancer patients
therapy for MSI-H/dMMR mCRC (Lenz et al. Ann Oncol 2018;29:LBA18). participate in clinical studies and these are often patients with favourable
Longer follow-up data will be presented. Methods: Patients with MSI-H/dMMR prognostic factors. Therefore, we explored for which proportion of real-life stage
mCRC and no prior treatment for metastatic disease received NIVO 3 mg/kg IV CRC patients OS improved in recent years. Methods: Nationwide population-
every 2 weeks + low-dose IPI 1 mg/kg every 6 weeks until disease progression or based data of all stage IV (synchronous metastatic) CRC patients diagnosed
discontinuation. The primary endpoint was investigator-assessed objective between 2008-2016 in the Netherlands who received local and/or systemic
response rate (ORR). Results: For all 45 patients (median follow-up was antitumor therapy were obtained from the Netherlands Cancer Registry (NCR).
13.8 months), ORR was 60% (95% CI 44.3–74.3). Responses were con- Initial treatment was registered in the NCR. Vital status was recorded until
sistent with the overall population across subgroups including age, Eastern January 31st 2018. OS per incidence year was determined for various per-
Cooperative Oncology Group (ECOG) performance status, prior adjuvant/ centiles, which represent the number of months after which 10-30-50-70-80-
neoadjuvant therapy, and mutation status (Table). Seven patients (16%) 90% of the patients had died. For some percentiles survival time exceeded
had grade 3–4 treatment-related adverse events (TRAEs); 3 (7%) had any follow-up duration. Results: The total study population comprised 21,047
grade TRAEs leading to discontinuation. Updated response, survival, and patients. mOS remained unchanged in the period 2008-2016 at around
safety data after a longer follow-up (median 19.9 months) will be presented. 15 months. OS of p10 and p30 increased by 1,5 months to 3,6 and 10,0
Conclusions: NIVO + low-dose IPI demonstrated robust and durable clinical respectively. For the ‘best’ (longest-living) patients (p70-p80) OS improved in
benefit and was well tolerated. Evaluated subgroups had responses consistent the period 2008-2016 by 4 and 6 months to 30 and 43 months, respectively.
with the overall population. NIVO + low-dose IPI may represent a new 1L Follow-up duration is insufficient to analyse change over time for the 10% ‘best’
treatment option for patients with MSI-H/dMMR mCRC. Clinical trial in- patients. mOS did not change for any of the treatment subgroups except for the
formation: NCT02060188. patients who received exclusively non-systemic therapy (e.g. metastasectomy,
ORR a,b
in overall patients and subgroups. radiotherapy) in which mOS improved by 11 months to almost 20 months.
Conclusions: mOS of real-life stage IV CRC patients has not improved since
NIVO + low-dose IPI 2008. We observed a clinically relevant survival improvement in only 30% of all
n/N (%)
treated patients and in all patients who received exclusively non-systemic
ORR (all patients)c 27/45 (60) therapy. These data illustrate the different outcomes between trial patients
Age, years ,65 $65 14/22 (64) 13/23 (57) and total patient population, which emphasizes that trial patients represent a
ECOG performance status 0 $1 13/25 (52) 14/20 (70)
selection of patients with favourable prognosis. Our results highlight the value of
Prior adjuvant/neoadjuvant therapy Yes No 12/19 (63) 15/26 (58)
Mutation status real-life data to determine efficacy of innovations in daily clinical practice.
BRAF/KRAS wild type 8/13 (62)
BRAF mutationc 12/17 (71)
KRAS mutation 5/10 (50)
Unknown 2/5 (40)
a
Investigator assessed. bMedian follow-up defined as time on study from first dose
to data cutoff, which was 13.8 months (range 9–19). cPreviously reported.

3523 Poster Session (Board #15), Mon, 8:00 AM-11:00 AM 3524 Poster Session (Board #16), Mon, 8:00 AM-11:00 AM
A regulatory program that promotes metastasis in colorectal cancer (CRC) The role of maintenance strategy in metastatic colorectal cancer (mCRC): A
through modulation of mRNA stability. First Author: Hani Goodarzi, UCSF, systematic review and meta-analysis. First Author: Mohamad Bassam
San Francisco, CA Sonbol, Mayo Clinic, Phoenix, AZ
Background: CRC progression accompanies dysregulations in pathways of gene Background: In mCRC, induction combination chemotherapy with targeted
expression control. Regulatory pathways that govern RNA decay have emerged as agents is considered the mainstay of treatment. This is typically followed by
key mechanisms coopted by cancer cells. Here, we have described a novel maintenance therapy vs. observation which had been examined in various
regulatory program that acts as a suppressor of metastasis in CRC. Methods: We trials. However, it remains unclear how best optimize maintenance strategy.
have developed a computational approach called PRADA that identifies master We aim to evaluate comparative effectiveness to support best maintenance
regulators of aberrant mRNA stability. By applying this tool to a compendium of strategy. Methods: We searched PubMed, Embase, and Cochrane CENTRAL
gene expression data collected from patient samples and colon cancer cell lines, for randomized controlled trials (RCT) evaluating different maintenance
we have identified a novel regulatory program involved in CRC metastasis. We strategies in previously untreated mCRC patients (pts): observation (obs),
have used xenograft models and genomic technologies to functionally dissect this bevacizumab (bev), fluoropyrimidine (FP), FP+bev, or continuing induction
pathway. We have also performed multivariate analysis in public datasets, as well regimen (CTX). Outcomes of interest included OS and PFS. The overall effect
as qPCR-based measurements in tumor samples to establish its clinical rele- was pooled using the DerSimonian random effects model. We conducted
vance. Results: PRADA identified the RNA-binding protein RBMS1 as a key network meta-analysis based on White’s multivariate meta-regression to pool
factor in CRC metastasis. RBMS1, which is silenced in highly metastatic cells, evidence from direct and indirect comparisons. Agents were ranked using
binds and stabilizes a large regulon of mRNAs. Silencing RBMS1 in established surface under the cumulative ranking (SUCRA) probabilities. Higher SUCRA
lines resulted in increased liver colonization in xenograft models and its over- scores correspond to greater efficacy. Results: Twelve trials at low risk of bias
expression reduced metastasis. We also identified the set of genes that are directly (5540 pts) were included. Network meta-analysis shows no benefit of CTX
bound and regulated by RBMS1 and function downstream. Loss of RBMS1 and over obs in terms of PFS (HR 0.7; 95% CI 0.46-1.09) and OS (HR 0.95;
its regulon provide a signature predictive of RFS in localized CRC (n = 574, HR = 95% CI 0.85-1.07). Compared to obs, maintenance therapy shows PFS
0.48, p = 0.038) even when controlled for known prognostic factors in a mul- benefit (HR 0.58; 95% CI 0.43-0.77) with only a trend in OS (HR 0.91; 95%
tivariate analysis (Table). We also observed a 4-fold reduction (P,1e-3) in CI 0.83-1.009). All maintenance strategies (FP, FP+bev, and bev) show
the expression of RBMS1 in stage IV tumors relative to earlier stage disease. significant improvement in PFS over obs. On SUCRA analysis, maintenance
Conclusions: In sum, we have discovered a previously unknown regulatory treatment (FP or FP+bev) has the highest likelihood of achieving better PFS
pathway of RNA stability that acts as a suppressor of metastasis in CRC which may (67.1% for FP and 99.8% for FP+bev) and OS (81.3% for FP and 73.2% for
inform new therapeutic targets among the RBMS1 regulon for adjuvant therapy. FP+bev). Conclusions: A maintenance strategy with at least a FP with or
Multivariate analysis of gene expression indicates that RBMS1 silencing is associated with poor relapse- without the addition of bevacizumab is preferred. However, given the lack of a
free survival in CRC patients.
clear OS benefit, obs is an acceptable alternative. Optimal maintenance
GSE39582 HR p-value
strategies should be dependent on factors including patient preferences, cost
RBMS1 low vs high 0.427 0.015 and toxicities.
Stage 2vs1 6.981 0.055
Stage 3vs1 11.528 0.016
Stage 4vs1 46.319 0.000
Strategy SUCRA % for OS SUCRA % for PFS
MSS vs MSI 2.805 0.021 Obs 0.3 0.3
BRAF mut vs WT 1.379 0.377
KRAS mut vs WT 1.282 0.197 Bev 32.6 36.5
CMS2 vs CMS1 0.754 0.477 FP 81.3 67.1
CMS3 vs CMS1 0.585 0.242 FP + Bev 73.2 99.8
CMS4 vs CMS1 1.325 0.470 CTX 46 46.3

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Colorectal) Cancer 197s

3525 Poster Session (Board #17), Mon, 8:00 AM-11:00 AM 3526 Poster Session (Board #18), Mon, 8:00 AM-11:00 AM
Prognosis of microsatellite instability and/or mismatch repair deficiency ROCKET: A randomized, multicenter phase 2 study of RRx-001 + irinotecan
stage III colon cancer patients after disease recurrence: Results of an accent versus regorafenib in 3rd/4th line colorectal cancer. First Author: Tony R. Reid,
meta-analysis of seven studies. First Author: Julien Taieb, Hôpital Européen University of California, San Diego, San Diego, CA
Georges-Pompidou, Sorbonne Paris Cite/Paris Descartes University, Paris,
Background: The purpose of this study was to assess the efficacy of combi-
France
nation therapy with RRx-001, an aerospace-derived macrophage repolarizing
Background: Microsatellite instable/deficient mismatch repair (MSI) meta- agent with epigenetic and vascular normalization properties, plus irinotecan
static colorectal cancers have been reported to be of poor prognosis. The vs. regorafenib plus irinotecan in 3rd/4th line metastatic colorectal cancer that
interaction between MSI and BRAFV600E mutation complicates the picture. previously received treatment with irinotecan. Methods: Between June 2014
Methods: Patients with resected stage III CC from 7 studies with disease and August 2015, 34 patients were enrolled. Therapy consisted of intravenous
recurrence and data available for MSI and BRAFV600E status were analyzed. administration of RRx-001 at 4 mg once weekly until progression followed by
The primary endpoint was survival after recurrence (SAR) to assess the intravenous infusion of irinotecan at 180 mg/m2 on day 1 in a 21-day cycle vs.
prognostic roles of MSI and BRAF V600E , respectively. Associations of 160 mg oral regorafenib daily for 3/4 weeks followed at progression, if ap-
markers with SAR were analyzed using Cox proportional hazards models plicable, by irinotecan 180 mg/m2 on day 1 in a 21-day cycle. Results: The
adjusted for clinicopathologic features (data collected 12/1998 to 11/2009). survival follow up for RRx-001 + irinotecan range was approximately
Results: Among 2630 patients with cancer recurrence (1491 men [56.7%], 14.5 months (mean 6.0 months, median 5.0 months, standard deviation
mean age, 58.5 [19-85] years), multivariable analysis revealed that patients 4.5 months). The survival follow up for regorafenib range was approximately
with MSI tumors (n = 220) had significantly better SAR (adjusted hazard ratio 10.8 months (mean 4.5 months, median 4.0 months, standard deviation
[aHR], 0.82; 95% CI, 0.69-0.98; P = .029) than patients with microsatellite 3.9 months). The median survival time was 8.6 months for RRx-001 + iri-
stable /proficient MMR (MSS) tumors (n = 1766). This was also observed when notecan and 4.7 months for regorafenib. The median progression free survival
looking at patients treated by the standard FOLFOX adjuvant regimen only (aHR, was 7.5 months for RRx-001 + irinotecan vs. 1.7 months for regorafenib. In
0.76; 0.58-1.00; P = .048). Same trends were observed when looking at MSI/ addition, the toxicity profile of RRx-001 + irinotecan was dramatically im-
dMMR patients outcome in BRAFV600E wild-type (aHR, 0.84; P = .10) and proved compared with regorafenib. The regorafenib arm was terminated for
mutant (aHR, 0.88; P = .43) subgroups separately, without reaching statistical futility at interim analysis, after which any enrolled patients only received RRx-
significance. As previously described poor SAR was observed in BRAFV600E 001 co-infused with an aliquot of autologous blood to evaluate the safety of
mutants vs wild type patients (n = 244; aHR, 2.06; 95% CI, 1.73-2.46; P , that method of administration. Conclusions: The results of this trial demon-
.0001) and this was also true in BRAFV600E mutants MSI/dMMR patients (n = strate improved efficacy and efficacy of RRx-001 + irinotecan compared with
77, aHR, 2.65 ; 95% CI, 1.67-4.21; p , .0001). Other factors associated with a regorafenib in patients with metastatic colorectal cancer. Clinical trial in-
poor SAR were : olderage, male gender, T4/N2, proximal primary tumor location, formation: NCT02096354.
poorly differentiated adenocarcinoma, and early recurrence (by 1y increase).
Conclusions: In stage III colon cancer patients recurring after adjuvant che-
motherapy and before the era of immuno-oncologic agents, MSI/dMMR was
associated with a better survival compared to MSS. BRAFV600E mutation seems
to be a poor prognostic factor for both MSI/dMMR and MSS/pMMR patients.

3527 Poster Session (Board #19), Mon, 8:00 AM-11:00 AM 3528 Poster Session (Board #20), Mon, 8:00 AM-11:00 AM
Apatinib monotherapy for chemotherapy-refractory metastatic colorectal Genomic alterations after EGFR blockade in patients with RAS wild-type
cancer: A multicenter, single-arm, prospective study. First Author: Fen metastatic colorectal cancer: Combined tissue and blood-based analysis
Wang, Department of Oncology, Peking University Shenzhen Hospital, from SCRUM-Japan GI-SCREEN and GOZILA. First Author: Yoshiaki
Shenzhen, China Nakamura, Department of Gastroenterology and Gastrointestinal Oncology,
National Cancer Center Hospital East, Kashiwa, Japan
Background: Apatinib is an oral highly-selective tyrosine kinase inhibitor (TKI)
that blocks vascular endothelial growth factor receptor 2 (VEGFR-2). This Background: Anti-EGFR therapy (tx) in RAS wild-type (wt) metastatic co-
exploratory study evaluated the efficacy and safety of apatinib monotherapy lorectal cancer (mCRC) induces resistance through acquired genomic alter-
in patients with chemotherapy-refractory metastatic colorectal cancer. ations. We aimed to define such alterations in Nationwide Cancer Genome
Methods: In this multicenter, single-arm, prospective study, 48 patients with Screening Project tissue and blood specimens, using next generation se-
metastatic colorectal cancer who had failed at least two lines standard che- quencing (NGS). Methods: Tumor specimens in patients (pts) with RAS wt
motherapies including fluorouracil, oxaliplatin and irinotecan were recruited mCRC were obtained from SCRUM-Japan GI-SCREEN (tissue) and GOZILA
from 14 centers in Guangdong, China. Apatinib at a 500mg dose was ad- (circulating tumor DNA [ctDNA] from blood). Genomic alterations were
ministered daily continuously. Each cycle was 4 weeks (28 days). The primary compared using the Oncomine Comprehensive Assay (SCRUM) and Guar-
endpoint was progression free survival (PFS). Secondary end points included dant360 (GOZILA), before anti-EGFR tx and after progression. Results: 373
overall survival (OS), objective response rate (ORR), disease control rate (DCR), total actionable alterations were identified in 71 pts with available matched
quality of life (QoL) and toxicity. Results: A total of 48 patients was enrolled in tissue and ctDNA; 255 (68%) were acquired after anti-EGFR tx progression.
the study from September 3, 2015 to June9, 2017. Four patients achieved a Frequently seen acquired oncogenic alterations included KRAS mutations
partial response, and 22 achieved stable disease, representing a response rate (27%) and amplifications (amps) of EGFR (41%), CDK6 (24%), BRAF (20%),
of 8.3% and a disease control rate of 60.4%. Median follow-up time was MYC (17%), MET (14%), PIK3CA (11%), FGFR1 (11%), and KRAS (10%).
10.3 months. Median progression-free survival (PFS) and overall survival (OS) Fusions of RET, ALK, and FGFR3 were newly acquired in 1-4%. Acquired
of evaluable patients (n=41) were 4.7 months (95% confidence interval [CI] alterations co-arose in multiple pathways, including the cell cycle, PI3K-AKT,
3.7-5.9) and 9.7 months (95% CI 5.9-13.6). The most common grade 3 or 4 and MAPK, although 29% of pts had none. Acquired mutations were less
adverse events (AE) were hypertension (12.5%), hand-foot syndrome (10.4%), frequently clonal versus primary mutations (p,0.0001), but clonal acquired
thrombocytopenia (10.4%), proteinuria (8.3%) and mucositis oral (6.3%). mutations were seen in several oncogenes, including EGFR, KRAS, and
Conclusions: Apatinib monotherapy shows promising efficacy and manage- PIK3CA. A subset of acquired KRAS, MET, CCND2, and EGFR amps had high
able toxicities in patients with chemotherapy-refractory metastatic colorectal (.7) adjusted plasma copy numbers (ApCN). Acquired ERBB2 amps were
cancer. Further phase 3 trial is warranted. Clinical trial information: identified in 3 pts (4%) with a median ApCN of 4, one of whom (ApCN=4.2),
ChiCTR1900020503. treated with dual HER2 blockade, progressed after 2 cycles. Conclusions: Our
integrated analysis revealed that anti-EGFR tx of pts with RAS WT mCRC led to
acquired genomic alterations in multiple oncogenic pathways. Although most
acquired alterations were subclonal, a subset of oncogenic alterations had
relatively high clonality and ApCN, suggesting potential targets for overcoming
acquired resistance to anti-EGFR tx. Early progression in a pt with an ApCN of
4.2 suggests low-level/subclonal acquired alterations may not be effective
treatment targets.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
198s Gastrointestinal (Colorectal) Cancer

3529 Poster Session (Board #21), Mon, 8:00 AM-11:00 AM 3530 Poster Session (Board #22), Mon, 8:00 AM-11:00 AM
Screening patients for fluoropyrimidine-related toxicity risk. First Author: Tumor dynamics with fluorouracil/folinic acid, irinotecan, and oxaliplatin
Olivier Capitain, Institut de Cancerologie de l’Ouest, Site Paul Papin, Angers, (FOLFOXIRI) plus panitumumab (pmab) or FOLFOXIRI alone as initial
France treatment of RAS wildtype metastatic colorectal cancer (mCRC): Central
radiologic review of VOLFI—A randomized, open label, phase-2 study (AIO
Background: Severe, sometimes fatal, toxicity can occur during the 1st or 2nd course
of chemotherapy using fluoropyrimidines (FPs), and poses a serious public health
KRK0109). First Author: Dominik Paul Modest, Department of Medicine III,
problem. FPs carry a 3-5% risk of grade $ 3 early toxicities and 0.2% risk of death University Hospital, LMU Munich, Munich, Germany
linked to Dihydropyrimidine Dehydrogenase (DPD) deficiency. Methods: Of 29,000 Background: The VOLFI trial demonstrated improved objective response rate
patients screened since July 2000, 472 were referred to us due to severe toxicity (ORR) with the addition of pmab to modified triplet chemotherapy with
during 1st round 5-FU, or because pre-screening was done too late. Toxicity eval- FOLFOXIRI in a 2:1 (63 patients FOLFOXIRI plus pmab; 33 patients FOL-
uation was performed according to the NCI scale of adverse reactions to cancer drugs FOXIRI) randomized, controlled, phase II trial in patients with untreated RAS
(0=none, 5=death). Patients were previously 5-FU naı̈ve, had different cancers, and
wildtype mCRC. Methods: Radiologic images from the study were centrally
received various protocols, eg. 5FU or Capecitabine; bolus6continuous or per os.
The reliability of the following 4 pre-treatment screening tests to predict grade$ 4 examined according to RECIST 1.1. We further assessed early tumor shrinkage
toxicity was assessed: 1) DPYD genotype mutation (*2A,*2B,*7, 13, HapB3) 2) (=ETS: 20% shrinkage of tumor diameter at first re-assessment) and depth of
Plasma uracil (U) level, 3) Plasma dihydrouracil/uracil ratio (UH2/U) 4) a multi- response (=DpR= maximum shrinkage of lesions defined as relation of smallest
parametric approach with genotyping, UH2/U ratio and key patient factors (age, sex, tumor diameter to baseline). Moreover, time to depth of response was cal-
etc.). McNemar’s test with Bonferroni correction was used for statistical analysis. culated (randomisation to depth of response image). Results: Images were
Results: Of the 472 referred patients, 169 had grade 4 or 5 toxicity, of which 41 died available for 88 of 96 patients (91.7%), 86 patients (89.6%) had at least one
from toxicity. 98 had one or plus DPYD mutation: 42(42.9%)*2A; 43(43.9%)*2B; follow-up image and were included in the central review. According to central
3(3%)*7; 4(4%)*13; 8(8.16%) HapB3; 1 was homozygous *2A. Data below review, objective response rates were 89.2% vs 66.7% with FOLFOXIRI plus
compare the 4 screening methods for predicting grade 4-5 toxicity. Conclusions: The pmab vs FOLFOXIRI alone (P=0.02). ETS was also significantly more frequent
multiparametric approach is statistically (p,0.0001) the most efficient in terms of (Fisher’s exact test; P=0.01)) and DPR (Wilcoxon test; P= 0.004) significantly
preventing grade 4 and 5 toxicity (death) due to 5-FU treatment. Around 290,000 greater with pmab as compared to chemotherapy alone. See table for details.
patients are treated with 5-FU per year in the USA. Assuming a 0.2% mortality rate Time to DpR was similar in the panitumumab- vs chemotherapy alone arm (3.9
due to toxicity, around 580 lives could be saved per year using the multiparametric (95% confidence interval 2.8-4.7) vs. 4.2 (95% CI 3.6-5.7) months, re-
pre-treatment test. spectively. P=0.63). Conclusions: In this central review, pmab significantly
Grade 4-5 tox DPYD Mutations Uracil >16ng/ml UH2/U <6 Multiparametric improves ORR, the rate of ETS and also DpR when added to a mFOLFOXIRI
regimen. Our findings underline the potential of this highly active regimen in
n 169 169 169 169
Sensitivity n (%) 68 (40,23%) 108 (63,9%) 133 (78,69%) 161 (95,27%) patients with RAS wildtype mCRC that need to achieve early and profound
False negative n (%) 101 (59,76%) 61 (36,1%) 36 (21,3%) 8 (4,73%) shrinkage of the tumor. Additional analysis including molecular subgroups and
tumor sidedness will be shown at the meeting. Clinical trial information:
Grade 5 tox DPYD Mutations Uracil >16ng/ml UH2/U <6 Multiparametric NCT01328171.
n 41 41 41 41
Sensitivity n (%) 20 (48,78%) 33 (80,5%) 37 (90,24%) 40 (97,56%) Central review population
False negative n (%) 21 (51,22%) 8 (19,5%) 4 (9,76%) 1 (2,44%)
mFOLFOXIRI plus panitumumab P-
(N=56) FOLFOXIRI (N=30) value
ETS 20% (%) 48 (85.7) 18 (60.0) 0.01
DpR median % -58.9 (-100.0 to 130.0) -40.9 (-83.8 to 0.004
(range) 25.9)

3531 Poster Session (Board #23), Mon, 8:00 AM-11:00 AM 3532 Poster Session (Board #24), Mon, 8:00 AM-11:00 AM
Metastasectomy and BRAF mutation: An analysis of survival outcome in Intratumoral CD3+ and CD8+ T-cell densities in patients with deficient DNA
metastatic colorectal cancer. First Author: Thiru Prasanna, University of mismatch repair (dMMR) metastatic colorectal cancer (mCRC) receiving
Canberra, Canberra, ACT, Australia programmed death-1 (PD-1) blockade. First Author: Sakti Chakrabarti, Mayo
Clinic, Rochester, MN
Background: Surgical resection of oligometastases improves survival in metastatic
colorectal cancer (mCRC). It is unclear whether such benefit is consistently observed Background: Colorectal cancer with dMMR display heterogeneity in the extent
for BRAF V600E mutant (MT) and wild type (WT) mCRC. We conducted a retro- of intratumoral T-cell infiltration which may explain their variable re-
spective analysis to explore the influence of BRAF mutation status on survival sponsiveness to PD-1 blockade. We examined the association of intratumoral
outcomes after metastasectomy. Methods: Data collected from two large prospective CD3+ and CD8+ T-cell densities (TCD) with objective response rate (ORR) and
population databases in Australia (Treatment of Recurrent and Advanced Colorectal response duration in patients with dMMR mCRCs receiving pembrolizumab
Cancer (TRACC) and South Australian cancer registry). Overall survival (OS) and (PEM). Methods: Record review was performed on 12 patients with dMMR
recurrence free survival (RFS) for BRAF MT and WT mCRC were evaluated by
mCRC treated with PEM (200 mg intravenously every 3 weeks) after failure of
Kaplan-Meier method and compared by log-rank test. Results: 513 patients who had
undergone metastasectomy were identified, 6% were BRAF MT. Median age 63. prior chemotherapy [median no. of regimens was 1 (range 1-4)] between 01/
Metastasectomy rate was lower in BRAF MT (13 v 27%). In BRAF WT, 4% un- 2015 and 12/2017. CD3+ and CD8+ TCDs were analyzed in the primary tumor
derwent resection of metastases (mets) in .1 organ at diagnosis and 5% had 3 or 4 core (CT) and at the invasive margin (IM) by immunohistochemistry and au-
metastasectomies versus none in BRAF MT. Median OS in BRAF MT v WT: 25.7 v tomated image analysis to determine density score (0 to 100) for each T-cell
48.5 months (HR 1.95; 1.18-3.22). In a multivariate model adjusting for variables subtype and compartment (Ventana Medical Systems, Inc.). Patients were
which were significant on univariate analysis, OS differences were not statistically categorized as 1) responders [CR (complete response) + PR (partial response)]
significant. Right primary tumor, intact primary, .1 metastatic sites at diagnosis, vs. non responders [SD (stable disease) + PD (progressive disease)] per RECIST
non R0 resection, peritoneal mets and synchronous mets were independent pre- version 1.1, and 2) by duration of response (, or . 12 months).
dictors of worse OS. Among 364 patients with RFS data there was no difference Results: Median follow-up post PEM was 19.5 (9-41) months. Responders
between BRAF MT and WT (16 v 19 months, p=0.09). Rate of downsizing was higher included 2 CR and 5 PR; non-responders included 4 SD and 1 PD. The ORR and
with triplet chemo than doublet +/- bevacizumab or doublet/EGFR in BRAF WT (50 v median time to response were 58.3% (7/12) and 12 weeks (range 9-40),
30%) as well as MT (33 v 11%). Conclusions: Median OS was . 2 years in BRAF MT respectively. CD3+ and CD8+ TCD scores were higher in responders vs. in non-
V600E after metastasectomy in this study consistent with an OS benefit. OS did not responders as well as in patients who had disease control for . 12 months;
differ after metastasectomy between BRAF MT and WT in a multivariate model. differences were greatest for CD8+ CT (Table). Conclusions: Among patients
Presence of BRAF MT should not impact patient selection for metastasectomy. treated with PEM, data suggest higher intratumoral CD3+ and CD8+ TCDs in
Patient characteristics. responders versus non-responders and in those with a longer duration of disease
BRAF MT n=31
(%)
BRAF WT n=483
(%) p
control. If confirmed, TCDs may potentially predict responsiveness to PD-1
blockade in dMMR mCRCs.
Resected primary 90 85 0.6
Synchronous met 57 51 0.5
MSI-H 13 3 0.02
Responders Non-responders Disease control Disease control
Right primary 70 28 0.001 TCD (n= 7) (n= 5) >12 mo., n= 8 <12 mo., n= 4
Rectal primary 10 31 0.001
Liver met at diagnosis 40 66 0.005 CD3+ IM, median(range) 54 (15-100) 46 (9-96) 59(9-96) 39 (18-51)
Lung met at diagnosis 13 26 0.19 CD3+ CT, median(range) 74 (34-98) 52 (7-100) 79 (8-98) 54 (34-83)
Peritoneal mets at diagnosis 60 31 0.002 CD8+ IM, median(range) 47 (13-89) 37 (11-83) 60 (11-89) 29 (15-54)
Complete resection 73 76 0.8 CD8+ CT, median(range) 88 (31-100) 37 (10-94) 88 (10-100) 36 (31-66)
1st metastasectomy 40 62 0.02
-Liver
-Lung 7 17 0.2
-Peritoneum 53 24 0.001

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Colorectal) Cancer 199s

3533 Poster Session (Board #25), Mon, 8:00 AM-11:00 AM 3534 Poster Session (Board #26), Mon, 8:00 AM-11:00 AM
RAS-amplified colorectal cancers: Microsatellite stability status, RAS/BRAF Impact of gender on the safety profile of chemotherapy plus bevacizumab in
mutations, and prediction of anti-EGFR resistance. First Author: mCRC: A pooled analysis of TRIBE and TRIBE2 studies. First Author: Gemma
Alexa Betzig Schrock, Foundation Medicine, Inc., Cambridge, MA Zucchelli, Department of Translational Research and New Technologies in
Medicine and Surgery, Unit of Medical Oncology 2, Azienda Ospedaliera
Background: Pre-clinical models implicate RAS amplifications (RASa) as a
Universitaria Pisana, Pisa, Italy
mechanism of resistance to anti-Epidermal Growth Factor Receptor (EGFR)
therapies. Yet, there is little guidance on the impact of RASa, as identified by Background: Based on retrospective experiences, gender seems to affect the
next generation sequencing (NGS), on anti-EGFR response. Methods: We safety profile of chemotherapy (CT), with a higher incidence of CT-related adverse
investigated the Foundation Medicine (FM) database for RASa in CRC and events (AEs) among females than males. Here we focus on the impact of gender on
characterized this population based on patient (pt) characteristics and other the toxicity of FOLFOXIRI/bevacizumab (bev) as compared with doublets (FOLFOX
concurrent genomic alterations. We subsequently investigated City of Hope or FOLFIRI)/bev in two randomized phase III studies by GONO: TRIBE and
(COH) mCRC molecular data set (using FoundationOne) and described the TRIBE2. Methods: The risk of experiencing CT-related AEs in males and females
was estimated in univariable analysis in the overall safety population and
RASa population characteristics and response to anti-EGFR. Results: FM
according to treatment arms (doublets/bev and FOLFOXIRI/bev). In order to assess
cohort included 21,315 CRC unique pt cases. 365 (1.5%) pts had RASa, of
the independent weight of gender on the risk of developing AEs, multivariable
which 123 (0.6%) had $ 20 copy number (CN) (RASa$20). The incidence of logistic regression models were built. Results: Among 1187 patients enrolled in
MSI-H, RAS and BRAF short variant mutations in the overall, RASa, and TRIBE and TRIBE2 studies, 1176 (684 males, 58%, and 492 females, 42%)
RASa$20 populations were (MSI-H: 5%, 0%, 0%), (RAS: 54%, 32%, 2%), were included in the safety population. Overall, women had a significantly higher
and (BRAF: 6%, 1%, 0%), respectively. Copy number variation inversely risk of CT-related AEs, in particular gastrointestinal and hematologic AEs, as-
correlated with likelihood of RAS mutation. COH cohort included 396 thenia and alopecia, independently of the treatment arm. The risk of CT-related
mCRC. 13 pts (3.3%) had RASa. The median RAS CN was 25 (7 - 79); 8/13 AEs was increased with FOLFOXIRI/bev vs doublets/bev independently of gender
pts had RASa$20. All but 3 COH pts with RASa had a RAS/BRAF-WT tumors. (p for interaction: 0.329). Notably, among women treated with FOLFOXIRI/bev
Pts with concurrent RAS mutations had relatively low RAS CN (7, 8, &13). 50% and 68% experienced any grade of vomiting and nausea, respectively.
Tumors with RASa were predominantly left-sided (12/13). 7/13 pts were Conclusions: Female mCRC patients have a higher risk to develop CT-related AEs.
treated with anti-EGFR therapy. All 7 pts were RAS/BRAF wild-type: 6 left- In women treated with FOLFOXIRI/bev the high incidence of nausea and vomiting
sided; 3 pts 1st/2nd lines (1 pt oxaliplatin-based and 2 pts irinotecan-based) may suggest the need for an intensification of the antiemetic prophylaxis.
and 4 pts chemo-refractory (all irinotecan-based). All 4 chemo-refractory pts Univariate Multivariate
had RASa$20and progressed on anti-EGFR on 1st post-treatment CT (PFS in
Grade Females (%) Males (%) OR p OR p
all pts # 3 mo). One 1st line pt had SD (PFS = 4 mo). One 2nd line pt had SD
(PFS = 7 mo) and one had PD as best response. Conclusions: NGS identifies Nausea All 65 54 1.57 , 0.01 1.55 , 0.01
$3 6 3 2.08 0.01 1.98 0.02
RASa and RASa$20in 1.5% and 0.6% of CRC, respectively. RASa$20 tumors Vomiting All 41 29 1.73 , 0.01 1.72 , 0.01
are MSS and, generally lack RAS/BRAF mutations, and predict for resistance $3 5 1 4.18 , 0.01 4.07 , 0.01
to anti-EGFR. This supports the potential relevance of 20-CN cut-point for Diarrhea All 65 61 1.16 0.24 1.13 0.34
$3 15 12 1.34 0.09 1.33 0.11
the exclusion of pts from anti-EGFR. These findings would benefit from Asthenia All 66 60 1.30 0.03 1.31 0.03
additional validation from retrospective analyses of completed prospective $3 12 8 1.62 0.02 1.65 0.01
anti-EGFR clinical trials. Alopecia All 14 10 1.55 0.02 1.56 0.02
Anemia All 57 49 1.33 0.02 1.31 0.03
$3 3 1 2.62 0.04 2.55 0.05
Neutropenia All 69 54 1.86 , 0.01 1.90 , 0.01
$3 44 30 1.86 , 0.01 1.90 , 0.01
Febrile Neutropenia All 8 5 1.60 0.06 1.67 0.04

3535 Poster Session (Board #27), Mon, 8:00 AM-11:00 AM 3536 Poster Session (Board #28), Mon, 8:00 AM-11:00 AM
Comprehensive transcriptome analysis reveals link between epigenetic Impact of age on safety and efficacy of first-line FOLFOXIRI/bevacizumab in
dysregulation, endogenous retrovirus expression and immunogenicity in mCRC: A pooled analysis of TRIBE and TRIBE2 studies. First Author:
metastatic colorectal carcinoma. First Author: Shehara Ramyalini Mendis, Federica Marmorino, Department of Translational Research and New
BC Cancer, Vancouver, BC, Canada Technologies in Medicine and Surgery, Unit of Medical Oncology 2, Azienda
Ospedaliera Universitaria Pisana, Pisa, Italy
Background: Endogenous retrovirus (ERV) elements represent genomic
footprints of ancestral retroviral infections within the human genome. Background: FOLFOXIRI/bevacizumab is a valuable upfront option in mCRC
Previous studies have demonstrated increases in ERV mRNA as a result of based on results of phase III TRIBE and TRIBE2 studies: 1187 pts aged
DNA hypomethylation, and ERV transcription has been associated with 18–70 years with ECOG performance status (PS) # 2 or between 71–75 years
increased immunogenicity in metastatic renal cell carcinoma. We performed with an ECOG PS of 0 were randomized to receive first-line FOLFOXIRI/
comprehensive bioinformatics analysis of ERV transcription in metastatic bevacizumab or a doublet (FOLFIRI in TRIBE and mFOLFOX6 in TRIBE2)/
colorectal carcinoma (mCRC), to identify novel links between ERV tran- bevacizumab. Here, we aimed at assessing the effect of the intensification of
scription, epigenetic dysregulation and immunogenicity in metastatic co- the upfront chemotherapy (triplet versus doublet) in terms of safety and ef-
lorectal carcinoma (mCRC). Methods: Tumour samples from 63 patients ficacy in pts aged , 70 versus 70-75. Methods: Subgroup analyses for ORR,
with mCRC were subjected to RNA sequencing as part of the Personalized PFS, G3/4 overall adverse events (AEs), chemo-related and bevacizumab-
OncoGenomics program (POG; NCT02155621) at BC Cancer. Patients were related AEs were performed according to baseline age. Results: 182 (15%) out
enrolled between 07/2012-07/2017. ERV transcription was quantified of 1187 pts were 70-75 years old (97 in the FOLFOXIRI/bevacizumab and 85
across 702,533 distinct loci. Tumors were classified ERV-hi if their total in the doublets/bevacizumab arms). The benefit provided by the intensification
ERV expression (RPKM) was greater than the mean across all samples. High of the upfront chemotherapy was independent of the age subgroup in terms of
antiviral gene expression tumors (AVG-hi) were designated as having a mean both ORR (p for interaction = 0.684) and PFS (p for interaction = 0.634). The
expression of IFIH1, DDX58, TLR3, TANK, TBKBP1, TBK1, IRF3 and IRF7 risk of overall and chemo-related G3/4 AEs was increased with the triplet
that was greater than the mean across all samples. All pairwise comparisons independently of age (p for interaction = 0.736 and 0.790), while no dif-
of gene expression were subjected to multiple hypothesis correction. ference in bevacizumab-related AEs was observed in both subgroups (p for
Results: Median age was 59 years, with 34 (54%) male and 1 tumor interaction = 0.566). In the overall population, as compared to younger pts,
microsatellite unstable. ERV-hi tumors showed increased expression of DNA those aged 70-75 were more susceptible to overall G3/4 AEs (70% vs 57%, p =
demethylators TET2 (q=0.0045) and TET3 (q,0.0001). Significant overlap 0.001). In the FOLFOXIRI/bevacizumab arm a higher incidence of G3/4 di-
existed between ERV-hi and AVG-hi tumors (18/27, p=0.016). Tumors both arrhea (27% vs 17%, p = 0.016) and febrile neutropenia (16% vs 6% p =
ERV-hi and AVG-hi trended towards increased PD-L1 expression (p=0.055) 0.001) and a lower incidence of all grade nausea (51% vs 65%, p = 0.009)
and showed a significant increase in survival compared to tumors with high and vomiting (26% vs 44% p = 0.001) were reported among elderly pts.
antiviral expression in the absence of high ERV transcription (p=0.0043). Conclusions: The activity and efficacy of FOLFOXIRI/bevacizumab are con-
Conclusions: Our results suggest DNA demethylation drives increased ERV firmed among selected pts between 70 and 75 years old, with a relative in-
transcription and ERV-associated immunogenicity in mCRC. Moreover, we crease in the risk of chemo-related AEs similar to that of younger pts. However,
provide novel insight into the impact of ERV transcription on the biology of elderly pts are more susceptible to experience AEs independently of the
mCRC, highlighting ERV transcription as a potential biomarker and target for treatment arm. Considering the increased incidence of febrile neutropenia and
precision immunotherapy. Clinical trial information: NCT02155621. diarrhea with FOLFOXIRI/bevacizumab, the use of G-CSF as primary pro-
phylaxis or an initial dose reduction of irinotecan and 5-fluorouracil might be
considered in this population.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
200s Gastrointestinal (Colorectal) Cancer

3537 Poster Session (Board #29), Mon, 8:00 AM-11:00 AM 3538 Poster Session (Board #30), Mon, 8:00 AM-11:00 AM
A randomized, double-blinded, placebo-controlled multicenter phase II trial A phase II study of lenvatinib in patients with metastatic colorectal cancer
of adjuvant immunotherapy with tecemotide (L-BLP25) after R0/R1 hepatic refractory to standard chemotherapy: LEMON study (NCCH1503). First
colorectal cancer metastasectomy (LICC): Final results. First Author: Carl Christoph Author: Hirokazu Shoji, Division of Gastrointestinal Medical Oncology,
Schimanski, Klinikum Darmstadt GmbH and Universitätsmedizin der Johannes National Cancer Center Hospital, Tokyo, Japan
Gutenberg-Universität Mainz, Darmstadt and Mainz, Germany
Background: Although regorafenib significantly improved overall survival
Background: Hepatic metastasectomy is the only potential curative treat- compared with placebo for patients with metastatic colorectal cancer (mCRC)
ment option for stage IV colorectal cancer (CRC) limited to liver metastases refractory to standard chemotherapies in the global phase III CORRECT trial,
(LM). After R0 resection of LM the high recurrence rate remains a major regorafenib is often discontinued due to toxicity. Lenvatinib is an inhibitor of
challenge. L-BLP25 is an antigen-specific cancer vaccine targeting mucin 1 VEGF receptors 1–3, FGF receptors 1–4, PDGF receptors a, RET, and KIT,
(MUC1). The LICC trial aimed to improve survival outcome in mCRC patients showing receptor inhibition profile and kinetics different from regorafenib. This
(pts) after R0/R1 LM resection. Methods: This LICC trial, a binational, phase II study was conducted to evaluate the efficacy and safety of lenvatinib
multicenter, double-blinded, placebo controlled phase II trial, included pts in patients with mCRC refractory to standard chemotherapies. Methods: mCRC
with stage IV LM limited CRC after resection of primary tumor and LM (R0/ patients with measurable lesions, PS 0-1, and refractory to standard che-
R1) within the last 8 weeks, ECOG 0/1 and adequate organ function. Pts motherapies including fluoropyrimidine, oxaliplatin, irinotecan, bevacizumab,
were 2:1 randomized to receive L-BLP25 or placebo. L-BLP25 930 mg was cetuximab or panitumumab (if RAS wild-type), and TAS-102 (trifluridine/
administered as 8 weekly subcutaneous doses followed by 6 week main- tipiracil) were eligible. Prior treatment with regorafenib was not allowed.
tenance intervals until recurrence or a maximum of 2 years. Cyclophos- Patients received lenvatinib orally at a dose of 24 mg once daily in 28-day
phamide 300 mg/m2 (CP) or matching saline (NS) was given intravenously cycles until unacceptable toxicity or disease progression. The primary endpoint
3 days prior to first L-BLP25/placebo. Co-primary endpoints were was disease control rate (DCR) assessed by independent central review.
recurrence-free survival (RFS) and 3-year overall survival (OS), secondary Secondary endpoints included safety, response rate, progression-free survival
endpoints were RFS and OS in subgroups with different MUC1 expression (PFS), and overall survival. The planned sample size was 30 patients to
and safety. Differences in RFS and OS were analyzed with exploratory log- expect a DCR of 60% with a threshold DCR of 35%, one-sided alpha of 5% and
rank tests on the intention-to-treat population. Results: Of 121 pts enrolled power of 80%. Results: Between October 2016 to January 2018, 30 patients
between Oct 2011 and Dec 2014, 79 pts received L-BLP25+CP, 42 pla- were enrolled. All had received prior chemotherapy; 14 (47%) and 16 (53%)
cebo+NS. Baseline characteristics were well balanced. Median age was 60 had received 3 or $ 4 prior lines for advanced disease, respectively. The
years. Median RFS was 6.1 (90% CI: 5.8-8.8) vs. 11.4 months (90% CI: median number of lenvatinib cycles was 4 (range 1-13). Two patients achieved
5.0-20.3) and estimated 3-year OS rate 69.1% vs. 79.1% for L-BLP25 and partial response and 19 patients had stable disease, resulting in a response
placebo, respectively. Two-factorial Cox regression models showed no im- rate of 6.7% and DCR of 70.0% (95% CI: 50.6-85.3%). Median PFS was
pact of MUC1 expression or treatment on RFS or OS. The most common L- 3.6 months (95% CI: 2.6-3.7). The most common grade $ 3 adverse events
BLP25-related grade 3/4 adverse events were diarrhea, anemia and back were hypertension (53%), elevated serum aspartate aminotransferase (13%),
pain. There was one death in the L-BLP25 arm due to Merkel cell carcinoma thrombocytopenia (10%), and anorexia (7%) without unexpected safety signals.
assessed by the investigator as being potentially related to vaccination. Conclusions: Lenvatinib showed promising antitumor activity with acceptable
Conclusions: The LICC trial failed to meet its primary endpoint of signifi- toxicity for patients with mCRC refractory to standard chemotherapies. Clinical
cantly improving RFS and OS with L-BLP25. MUC1 expression was not trial information: UMIN000023446.
associated with outcome. Clinical trial information: NCT01462513.

3539 Poster Session (Board #31), Mon, 8:00 AM-11:00 AM 3540 Poster Session (Board #32), Mon, 8:00 AM-11:00 AM
Evaluating gender as a predictive marker for response to bevacizuamb (Bev) Efficacy of retreatment with anti-EGFRs in mCRC is not predictable by
in metastatic colorectal carcinoma (mCRC): Pooled analysis of 3369 patients clinical factors related to prior lines of therapy: A multi-institutional analysis.
(pts) in the ARCAD database. First Author: Ofer Margalit, Sheba Medical First Author: Daniele Rossini, Department of Translational Research and
Center, Ramat Gan, Israel New Technologies in Medicine and Surgery, Unit of Medical Oncology 2,
Azienda Ospedaliera Universitaria Pisana, Pisa, Italy
Background: Previous studies suggest a possible gender-specific response to Bev
in mCRC, showing a benefit in males, while the effect in females is less significant. Background: Retrospective analyses and phase 2 studies suggest that ad-
Therefore, we evaluated response to Bev according to gender. Methods: Data from ministering an anti-EGFR in advanced lines may be effective in mCRC pts who
3369 mCRC patients enrolled on 4 first-line randomized trials testing Bev (2000- achieved benefit from a 1st-line anti-EGFR containing regimen. The identi-
2007) were pooled. Association between gender and progression-free survival fication of clinical features associated with benefit from anti-EGFR re-
(PFS)/overall survival (OS) was evaluated by stratified Cox regression model, ad- treatment (re-tx) in pts experiencing PD during 1st-line anti-EGFR (rechal-
justed for potential confounders. Predictive value was evaluated by interaction lenge) or after its interruption (reintroduction), is a major clinical need.
(inter.) effect between gender and treatment. In a pre-planned secondary analysis, Methods: A real-life data-base including a total of 5530 pts treated at 6
analyses were stratified using an age cut-point of 60 years (commonly used in breast
insitutions from December 2010 to October 2018 was queried. Pts retreated
cancer trials) to evaluate the possible role of menopausal-related effects.
with anti-EGFRs, with RAS/BRAF wild-type status on tissue samples, who had
Results: OS was not statistically different between males and females (median OS
[mOS], 18.8 vs. 17.6 months [mo], adjusted hazard ratio [HRadj], 0.93, 95%
received a 1st-line anti-EGFR-based tx with at least SD as best response, and at
confidence interval [CI], 0.84-1.03, p, 0.15) in the overall population. Bev was least one further line of therapy before anti-EGFR re-tx, were included. The
associated with an improved mOS in males and females, with a 2.3 and 0.6 mo association with RECIST response (RR), PFS and OS was investigated for the
benefit, respectively, as well as an improved PFS. There was no statistically sig- following variables: RR (PR or CR vs SD) and PFS during 1st-line; time from the
nificant interaction effect between gender and treatment (see table). Further last anti-EGFR administration to 1st-line PD (i.e. re-introduction vs rechal-
stratified by age (, vs. $ 60 years), Bev resulted in improved PFS and OS in both lenge); reason for anti-EGFR discontinuation in 1st-line (PD vs. other); number
genders, at all ages, except for the effect in young females which did not reach of anti-EGFR-free lines of therapy before re-tx; anti-EGFR free interval (time
statistical significance (see table). Conclusions: Our results confirmed the mOS between the last anti-EGFR administration in 1st-line and the time of re-tx);
benefit from addition of Bev to first-line chemotherapy in mCRC in both genders, primary tumor side; time from the diagnosis of metastatic disease to re-tx
although the benefit in females was , 1 month. For females under the age of 60, ($ vs. , 18 mos). Results: Data from 86 patients were retrieved, 56 (65%)
there are uncertainties for mOS benefit from addition of Bev and further evaluation and 30 (35%) received anti-EGFR rechallenge or reintroduction, respectively.
is needed. Median anti-EGFR free interval was 15.1 mos. The RR during re-tx was 19.8%,
p p
with a DCR of 46.5%. Median PFS and OS were 3.6 and 10.2 mos, re-
Age Gender Bev mPFS (mo) HRadj (95% CI) Inter. mOS (mo) HRadj (95% CI) Inter. spectively. No significant association of investigated features with RR and PFS
All F N 6.9 .08 17.3 .35
was observed. No differences in RR or PFS were observed among patients
Y 8.8 .77 (.68-.87) 17.9 .82 (.70-.96) receiving anti-EGFR re-tx as rechallenge or reintroduction (20.4% vs 23.1%,
M N 7.4 17.7 p = 0.99; median PFS: 3.49 vs 4.97 mos, p = 0.61). Patients with left-sided
Y 9.4 .67 (.61-.74) 20.0 .74 (.65-.85) tumors had longer OS (HR: 0.50, 95%CI: 0.26-0.93, p = 0.005).
< 60 F N 7.3 .07 18.3 .20
Y 8.8 .83 (.69-.99) 17.2 .85 (.67-1.08) Conclusions: Clinical factors that are generally believed to affect the efficacy of
M N 7.7 18.3 anti-EGFR re-tx are not confirmed in our series. Therefore, clinicians should
Y 10.3 .66 (.56-.77) 21.4 .71 (.56-.89) not rely on those characteristics in their decision-making on anti-EGFR re-tx,
‡ 60 F N 6.6 .42 15.9 .66
Y 8.8 .73 (.62-.86) 19.3 .80 (.64-.99) and adequate studies for implementing liquid biopsy in clinical practice are
M N 7.2 17.4 urgently needed.
Y 9.0 .68 (.60-.78) 19.0 .76 (.64-.90)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Colorectal) Cancer 201s

3541 Poster Session (Board #33), Mon, 8:00 AM-11:00 AM 3542 Poster Session (Board #34), Mon, 8:00 AM-11:00 AM
Effect of patient age on efficacy of FOLFIRI plus cetuximab vs bevacizumab in cfDNA for accurate determination of RAS and BRAF mutations using
1st-line treatment of metastatic colorectal cancer: An analysis of FIRE-3 (AIO OncoBEAM liquid biopsy in metastatic colorectal cancer patients: Results
KRK 0306). First Author: Volker Heinemann, University Hospital Munich, of the real-world multicentric ColoBEAM study. First Author: Alexandre
LMU Munich, Munich, Germany Harle, Institut de Cancérologie de Lorraine, Service de Biopathologie, CNRS
UMR 7039 CRAN Université de Lorraine, Nancy, France
Background: FIRE-3 compared 1st-line therapy with FOLFIRI plus either cetuximab
or bevacizumab in 592 KRAS exon 2 wt mCRC patients. The subgroup of extended Background: Determination of KRAS, NRAS (RAS) and BRAF mutations is a
RASwt patients consisted of 400 patients. The median age of patients treated in standard of care for the management of patients with metastatic colorectal
FIRE-3 was 64 years. Analyses of efficacy of doublet-chemotherapy in patients .70 cancer (mCRC). RAS mutations are well characterized resistance biomarkers to
years have shown inconsistent results. Methods: In this exploratory analysis, patients anti-EGFR antibodies and BRAF V600 mutations indicate poor prognosis.
were grouped into cohorts with 65 years and 70 years as a cut-off for age-related Tissue biopsy has traditionally been used to determine RAS and BRAF status,
analysis of the FIRE-3 study population. ORR was compared using Fisher´s exact but liquid biopsy analysis of circulating tumor DNA (ctDNA) has demonstrated
test, Survival analyses were done using Kaplan-Meier estimation and median survival
utility as a less invasive tool to expedite molecular testing results to the clinic.
times were compared using log-rank testing. Results: Within the RAS wt population,
patients older than 70 years had a significantly shorter OS when compared to
The ColoBEAM study reports the performance of plasma mutation testing in a
patients #70 years of age in both arms (p= 0.02 for cetuximab arm, p= 0.02 for real-life prospective series of 278 patients across 8 centers. Methods: Plasma
bevacizumab arm). Patients #65 years and #70 years had a significantly longer OS derived ctDNA was prepared from 20mL blood samples prospectively collected
when treated with FOLFIRI plus cetuximab compared to FOLFIRI plus bevacizumab from mCRC patients who had not received chemotherapy in the prior 15 days.
(p= 0.01 and 0.02 respectively). The OS benefit of cetuximab treated patients ctDNA was centrally assessed using OncoBEAM and results compared to those
compared to bevacizumab treated patients was consistent for patients #65 years obtained by routine analysis of tissue. Both tissue and blood samples with
(p=0.005) and patients #70 years (p= 0.009) of age. In patients older than 70 discrepant RAS results were blindly reassessed with OncoBEAM. Results: Of
years, however, comparable efficacy of bevacizumab and cetuximab was observed. 278 patients enrolled, 202 blood samples were available for OncoBEAM
This effect of age on outcome appears to be affected by sidedness. RAS wt pop- testing. RAS and BRAF V600E mutations were detected in tissue in 132/202
ulation (n=400). Conclusions: In the overall RAS wt population, younger patients (65.4%) and 4/198 (2.0%) patients, respectively. Analysis of the first ctDNA
have a significant OS benefit when treated with FOLFIRI plus cetuximab compared sample as compared to tissue DNA resulted in a kappa coefficient (k) of 0.52
to FOLFIRI plus bevacizumab, while this was not the case in patients older than 70 [0.41 – 0.63] and accuracy of 75.2% (65.1% sensitivity; 94.3% specificity).
years. OncoBEAM testing of a second sample resulted (k) of 0.66 [0.56 - 0.76] and
ORR PFS p# OS P# accuracy of 83.2% (77.3% sensitivity; 94.3% specificity). Of the 4 samples
£ 65 years N (%) p* (months) (HR) (months) (HR) with a BRAF V600E mutation in tumor tissue 2 were detected in blood. In the
FOLFIRI + Cet 104 75.6 0.08 11.2 0.42 33.1 0.01 subgroup of patients with liver metastasis (n=136), accuracy was 88.2%
FOLFIRI + Bev 105 63.0 10.2 1.10 24.8 0.68 (87.4% sensitivity; 90.2% specificity) for RAS and BRAF status with (k) of
£ 70 years N ORR P* PFS P# OS P#
(%) (months) (HR) (months) (HR) 0.73 [0.61 – 0.86]. In a subgroup of chemotherapy naı̈ve patients with liver
FOLFIRI + Cet 136 79.1 0.02 10.7 0.52 33.3 0.02 metastasis (n=49), accuracy was 91.8% (93.3% sensitivity; 89.5% speci-
FOLFIRI + Bev 150 65.2 10.5 1.10 27.5 0.73
> 70 years N ORR p* PFS p# OS P# ficity) for RAS and BRAF status with (k) of 0.83 [0.67 – 0.99].
(%) (months) (HR) (months) (HR) Conclusions: The results of the ColoBEAM study confirm plasma ctDNA as a
FOLFIRI + Cet 63 72.7 0.28 8.8 0.90 23.6 0.25
FOLFIRI + Bev 51 61.9 10.4 0.98 23.8 0.67
credible surrogate marker to tissue DNA for RAS and BRAF status assessment
and may be incorporated as a first-line theragnostic assessment. New testing
*= Fisher`s exact test p; #= Logrank Test p
on a second sample for wild-type status demonstrated 91.8% concordance
between blood and tissue. Clinical trial information: NCT02751177.

3543 Poster Session (Board #35), Mon, 8:00 AM-11:00 AM 3544 Poster Session (Board #36), Mon, 8:00 AM-11:00 AM
Development and validation of a prognostic score for overall survival integrating ctDNA as a potential prognostic marker for locally advanced rectal cancer
baseline metabolically active tumor volume measured by 18F-FDG PET/CT patients with ‘watch and wait’ approach. First Author: Lifeng Yang,
and clinical factors for metastatic colorectal cancer patients. First Author: Department of Radiation Oncology, Fudan University Shanghai Cancer
Erwin Woff, Nuclear Medicine Department, Institut Jules Bordet - Université Center, Shanghai, China
Libre de Bruxelles (ULB), Brussels, Belgium
Background: ‘Watch and Wait’ policy has currently led to growing interest for
Background: This study aimed to develop and validate a prognostic score in- organ-preservation after neoadjuvant chemoradiation (nCRT) to improve
tegrating baseline metabolically active tumor volume (MATV) and clinical quality of life. However, how to predict and select patients who may achieve
factors in metastatic colorectal cancer (mCRC) patients. Methods: The devel- clinical complete response is still an unsolved issue. We conducted a pilot
opment cohort included chemorefractory mCRC patients enrolled in two pro- study to evaluate the potential role of ctDNA as a biomarker to predict
spective multicenter non-randomized trials evaluating sorafenib/regorafenib as treatment outcome and improve risk stratification in locally advanced rectal
last line therapy. The validation cohort included mCRC patients from another cancer (LARC). Methods: In this study, we recruited 119 patients with LARC
center, treated with chemotherapy and bevacizumab as first line. Baseline receiving nCRT. 595 serial plasma samples were collected at d0, d15, d25
MATV was defined as the sum of metabolically active volumes of all target of radiotherapy as well before and 7 days post surgery. The level of ctDNA was
lesions identified on the baseline 18F-FDG PET/CT. MATV optimal cutoff for OS calculated by dynamic monitoring the mutant allele frequency of somatic
prediction was determined from the development cohort with Contal and mutations in plasma. Plasma and tissue samples were subjected to targeted-
O’Quigley’s method. MATV, age, gender, BMI, ECOG PS, years since diagnosis, NGS using a 422 cancer-related genes panel. We followed up patients with
and KRAS status were included in a multivariate analysis. A prognostic score to concomitant CT until disease progression or death. Results: Detected mu-
predict OS was developed from the development cohort using Cox proportional tation of TP53 and APC gene in pre-treatment samples was negatively
hazards model. Results: MATV and clinical factors were evaluable respectively correlated with patients’ response to nCRT. Alterations in homologous re-
in 155 and 122 patients of the development and validation cohorts. In uni- combination and adherens junction pathways were associated with a better
variate analysis, MATV with cutoff set at 100 cm³ identified two risk groups with response (P , 0.05). Detection of pre-treatment mutations in any time
different median OS (mOS) in both the development (4.5 vs 10.9 months, HR: points during nCRT was significantly (P = 0.03) decreased from TRG3 to
2.64; p , 0.001) and validation cohorts (20.9 vs 42.9 months, HR: 2.39; p , TRG0 group (33%, 29%, 22% and 4%, respectively); while detection of
0.001). A multivariate analysis identified four independent negative predictors acquired mutations showed an opposite trend (P = 0.04). A predictive model
of OS (high MATV, short time since diagnosis, poor PS, BMI , 25). Combining based on support vector machine was developed for prediction of pCR
these factors in a prognostic score for OS (best cutoff:-2) allowed to identify two achieving a mean AUC of 0.85 assessed by repeated cross validation.
risk groups with different mOS in the development (4.4 vs 13.4 months, HR: Further, detection of pre-treatment mutations after completion of nCRT was
3.67; p , 0.001) and validation cohorts (25 vs 63.8 months, HR: 2.5; p = significantly associated with worse disease-free survival (DFS) (P , 0.05).
0.001). Conclusions: In mCRC patients, the high prognostic value of baseline Through tracking clonal extinction, persistence and emergence, patients
MATV found in the development cohort was confirmed by external validation, were grouped into four evolutionary subtypes with distinct TRG and DFS.
independently of patients’ treatment. In both the development and validation Conclusions: Our data showed the prognostic value of ctDNA on DFS. Dy-
cohorts the prognostic score for OS allowed to identify two risk groups of mCRC namic monitoring of ctDNA can be used to predict TRG and prognosis in
patients with significantly different mOS. MATV and our prognostic score for OS LARC patients receiving nCRT. ctDNA sequencing depicts the evolutionary
should provide a firm basis for risk stratification, in clinical practice and re- trajectories of sensitive and resistant clones during nCRT in LARC. CtDNA
search trials. could potentially be used to guide patient selection for W&W strategy.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
202s Gastrointestinal (Colorectal) Cancer

3545 Poster Session (Board #37), Mon, 8:00 AM-11:00 AM 3547 Poster Session (Board #39), Mon, 8:00 AM-11:00 AM
Genetic variants in RNA binding protein (RBP) to predict outcome in Who can benefit from a liver surgery for metastatic colorectal cancer in the
metastatic colorectal cancer (mCRC): Data from FIRE-3, TRIBE, and MAV- era of modern chemotherapy? A post hoc analysis of the MIROX phase III
ERICC trials. First Author: Hiroyuki Arai, Chiba Cancer Center, Chibashi, trial. First Author: Samira Makhloufi, Medical Oncology Unit, Lille, France
Japan
Background: Despite improvement in colorectal liver metastasis (CLM)
Background: RNA binding proteins (RBPs) post-transcriptionally regulate treatment, survival after liver surgery remains highly variable. Several
gene expression by stabilizing or destabilizing target messenger RNA. Al- clinicopathologic prognostic factors have been reported whose validity in the
though alteration of RBPs affects many steps of cancer development, its era of modern chemotherapy remains to be defined. This study aimed to
clinical implication in mCRC remains unclear. Methods: We analyzed data analyze the prognostic factors associated with survival after CLM resection.
from mCRC patients (pts) enrolled in three first-line randomized trials (FIRE- Methods: Clinicopathologic data of patients included in the MIROX phase III
3, TRIBE, and MAVERICC). Genomic DNA from blood samples of pts was trial who underwent surgery for isolated CLMs were analyzed. The primary
genotyped through the OncoArray, a custom array manufactured by Illumina. endpoints were 5-year overall survival (OS) and disease-free survival (DFS).
Candidate 30 SNPs in 10 RBP genes (MSI1, MSI2, ELAVL1, RBM3, Univariate Cox analysis was performed to identify associations with OS and
LIN28A, LIN28B, IGF2BP1, IGF2BP2, IGF2BP3, ZFP36) were tested on DFS and select variables included in a multivariate model to determine their
association with progression-free survival (PFS) and overall survival (OS). To independent prognostic value. Results: A total of 181 patients were ana-
evaluate prognostic effects and heterogeneities across treatment arms, lyzed. The median follow-up period was 6.42 years (95% CI: 5.15-8.71),
meta-analysis approach using the METASOFT software was conducted. We and the 5-year OS and DFS rates were 67.1% and 35.4%, respectively. On
also tested interaction between each SNP and treatment within each trial, multivariate analysis, Fong’s clinical risk score (CRS) as a categorical var-
i.e. FIRE-3 cohort (FOLFIRI+cetuximab (Cet) vs FOLFIRI+bevacizumab iable (CRS 0-1 vs 2-3 vs 4-5, p = 0.036) and polymorphonuclear neutrophil
(Bev)) and MAVERICC cohort (FOLFIRI+Bev vs FOLFOX6+Bev). For multiple (PMN) count (.6000/mm3vs #6000/mm3, p = 0.006) before chemo-
testing, p values were adjusted by the false discovery rate (FDR) method. therapy were found to be independent prognostic factors for OS. However,
Results: A total of 884 pts’ SNPs data were available (FIRE-3: n = 236, only Fong’s CRS remained significantly associated with DFS (p = 0.027).
TRIBE: n = 324, and MAVERICC: n = 324). Meta-analysis combining three The final OS model was used to establish a nomogram that allows individual
trials showed RBM3 rs926152 (adjusted p = 0.045) and RBM3 rs2249585 OS estimations at 1, 3, 5, and 10 years. Conclusions: Fong’s CRS and PMN
(adjusted p = 0.016) were significantly prognostic for PFS. Whereas, in count were independently associated with poor OS after CLM resection.
terms of OS, only LIN28B rs314277 (adjusted p = 0.045) was significant, Fong’s CRS was also associated with DFS. The established prognostic no-
and RBM3 rs926152 (adjusted p = 0.057) and RBM3 rs2249585 (adjusted mogram could predict OS more accurately before CLM treatment.
p = 0.059) had a trend. Interaction test showed several SNPs were po-
tentially predictive (raw p , 0.05), although without any significance after
FDR adjustment: in FIRE-3 cohort, MSI2 rs1822381, RBM3 rs926152,
LIN28B rs221635, IGF2BP1 rs2969 for OS; in MAVERICC cohort, MSI1
rs1179442 and MSI2 rs3826301 for OS, ELAVL1 rs4804244 for PFS.
Conclusions: Our results indicate prognostic potential of SNPs in RBP
genes, such as RBM3 and LIN28B, in mCRC. However, we find no distinct
evidence that these SNPs can predict differential effect between Cet and
Bev, or between oxaliplatin- and irinotecan-based chemotherapy.

3548 Poster Session (Board #40), Mon, 8:00 AM-11:00 AM 3549 Poster Session (Board #41), Mon, 8:00 AM-11:00 AM
Trifluridine/tipiracil plus bevacizumab in elderly patients with previously Randomized phase II study on the influence of BRAF and PIK3CA mutations
untreated metastatic colorectal cancer (KSCC1602): A multicenter, phase II on the efficacy of FOLFIRI plus bevacizumab (Bev) or cetuximab (Cet), as
clinical trial. First Author: Eiji Oki, Department of Surgery and Science, first line therapy of patients (pts) with RAS wild-type metastatic colorectal
Kyushu University, Fukuoka, Japan carcinoma (mCRC) and ,3 baseline circulating tumor cells (bCTCs). First
Author: Enrique Aranda, IMIBIC, Reina Sofı́a Hospital, University of
Background: Elderly patients are often intolerable in combination of cyto-
Córdoba, CIBERONC, Instituto de Salud Carlos III/ Spain, Córdoba, Spain
toxic agents. Recently, Trifluridine/tipiracil plus bevacizumab has been
shown as good candidate for the vulnerable patients. We aimed to assess the Background: The outcome for mCRC has changed since the introduction of new
efficacy and safety of trifluridine/tipiracil plus bevacizumab in elderly pa- chemotherapy schedules and targeted therapies, however new predictive biomarkers
tients with metastatic colorectal cancer. Methods: Patients aged 70 years are needed. bCTCs and BRAF / PIK3CA mutations have been studied as a potential
and older with previously untreated, unresectable, metastatic colorectal predictive biomarkers. The primary endpoint was progression-free survival (PFS) in
cancer, who were not deemed to be candidates for oxaliplatin-based or pts WT KRAS and ,3 bCTCs, according to BRAF/PIK3CA status. Methods: This is
irinotecan-based chemotherapy regimens, were included in this trial. an open, multicentric, randomized phase II trial and included wildtype KRAS mCRC
Treatment consisted of trifluridine/tipiracil (35 mg/m2 orally a day on day 1-5 pts (RAS after approval of protocol amendment), younger than #70 with ,3 bCTCs,
ECOG 0-1 and available tissue for molecular analyses. Pts were stratified per number
and day 8-12) with bevacizumab (5mg/kg intravenously on day 1 and day
of metastatic organs involved (1 vs .1) and mutation status of BRAF and/or PIK3CA
15), given every 4 weeks until disease progression. The primary endpoint was (WT vs MUT) and randomized to group A (FOLFIRI+Bev) or group B (FOLFIRI+Cet).
progression-free survival (PFS). The secondary endpoints were response rate Results: 240 pts (196 WT and 44 MUT: 6 BRAF, 12 PIK3CA and 6 BRAF +
(ORR), Overall survival (OS) and adverse events (AEs). Results: Between PIK3CA) were included. General characteristics per mutation status (WT vs MUT):
2017 January and 2018 March, 39 patients were enrolled in this study. The Mean age (59 vs 61 years), gender (Male/Female 68/32 vs 70/30%), ECOG 0/1 (57/
characteristics of patients were male/female : 17/22, median age : 77.5 43 vs 66/34%), primary tumor unresected (48 vs 64%), RAS MUT in 12 pts (11 and
(range: 70-88) and PS: 0/1 : 31/8. The ORR of 37 patients were assessed by 1 pts, respectively), previous chemotherapy (12 vs 9%). Overall response rate (ORR)
the central review committee. The best response was 40.5(15/37) % (95% was 52 and 41% in the WT and MUT groups, respectively. PFS and overall survival
CI: 24.8-57.9), and the disease control rate was 86.5 (32/37) % (95% CI: (OS) are presented in the table. Conclusions: In the low risk mCRC pts according to
71.2–95.5). The AEs which were frequent as grade 3 or 4 were leucopenia bCTCs, BRAF and/or PIK3CA MUT have a negative impact in OS and a trend to worse
(71.8 %), neutropenia (48.7 %), anorexia (12.8 %), febrile neutropenia PFS in the ITT population. The impact of treatment is under evaluation and will be
(10.3 %) and fatigue (10.3 %). No treatment related death was reported. provided during the meeting. Clinical trial information: 2012-000840-90.
Conclusions: The combination of trifluridine/tipiracil plus bevacizumab is an ITT population FOLFIRI + Bev FOLFIRI + Cet
effective and well-tolerated regimen for elderly patients with metastatic N=240 N=126 N=114
colorectal cancer. Hematological adverse events were need for caution. The HR HR HR
primary endpoint of PFS will be presented in the end of this year. Clinical trial WT MUT (IC95%) WT MUT (IC95%) WT MUT (IC95%)
information: UMIN000025241. N=196 N=44 pvalue N=102 N=24 All pvalue N=94 N=20 All pvalue
PFS 12.7 9.1
1.136 12.9 9.3 12.5 0.992 12.5 8.5 11,5 1.452
m (0.739- (0.554- (0.765-
1.745) 1.776) 2.755)
p=0.562 p=0.978 p=0.253
OS 34,7 20,7 1.878 36.0 18.6 32,9 2.022 34.1 23.7 33.3 1.714
m (1.269- (1.194- (0.947-
2.779) 3.425) 3.102)
p=0.0016 p=0.008 p=0.078

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Colorectal) Cancer 203s

3550 Poster Session (Board #42), Mon, 8:00 AM-11:00 AM 3551 Poster Session (Board #43), Mon, 8:00 AM-11:00 AM
Bevacizumab (BV) maintenance (M) after first-line chemotherapy (CT) plus Modulation of autophagy: A phase II study of vorinostat (VOR) plus hydroxy-
BV for metastatic colorectal cancer (mCRC) patients (pts): A meta-analysis of chloroquine (HCQ) vs regorafenib (RGF) in chemo-refractory metastatic
individual pts data (IPD) from three phase III studies. First Author: Lisa colorectal cancer (mCRC). First Author: Sukeshi Patel Arora, UT Health
Salvatore, Oncologia Medica, Fondazione Policlinico Universitario “A. San Antonio Cancer Center, San Antonio, TX
Gemelli”, IRCCS, Roma, Italy
Background: Agents targeting the angiogenic pathway have been a cornerstone
Background: Although CAIRO3 and AIO KRK 0207 trials demonstrated the therapy in mCRC. In chemo-refractory mCRC, RGF, an oral multikinase inhibitor with
benefit of BV + fluoropyrimidine as a M regimen after induction CT + BV, the considerable angiogenic inhibition, has shown modest effects on survival. We re-
role of BV alone is not clear. Indeed, SAKK 41/06 and PRODIGE 9 trials ported that autophagy modulation using the autophagy inhibitor, HCQ, enhances the
failed to demonstrate the superiority of BV alone vs no M, while AIO KRK anti-cancer activity of the histone deacetylase inhibitor, VOR, via ubiquitinated
0207 showed the non-inferiority of BV alone vs combo M. Thus, in order to protein accumulation in CRC. A phase 1b study confirmed VOR/HCQ is active and
evaluate the magnitude of the eventual benefit of M with BV alone vs no M, an tolerated in refractory mCRC. We conducted a prospective randomized study to
evaluate efficacy of VOR/HCQ vs RGF in mCRC patients (NCT02316340) and
IPD meta-analysis was performed. Methods: Trials whereas mCRC pts were
report a planned interim analysis. Methods: Randomized, controlled trial of VOR
prospectively randomized to receive BV M or not were considered eligible. 400 mg and HCQ 600 mg PO daily vs RGF 160 mg PO daily (3 weeks on, 1 week off),
Primary end-points were PFS and OS, both from the start of induction and M. Q4weeks, in advanced CRC patients. Crossover was optional after first progression. A
Univariate and multivariate analyses for PFS and OS were performed, with total of 76 patients are planned. Primary endpoint: mPFS. Secondary endpoints:
the following variables: baseline ECOG PS; age ( . vs # 65 years); RAS and mOS; adverse events (NCI-CTCAEv3.0); PD analysis: 27-plex Human Cytokine
BRAF status; LDH and CEA baseline level; RR (PR or CR vs SD) during Array, NGS analysis (Guardant Health) on cell-free, ctDNA. Results: At interim
induction; induction CT (oxa- vs iri-based); resected primary tumor; primary analysis, n = 42 patients enrolled from 2/2015-10/2017: n = 20 VOR/HCQ (5
tumor side; synchronous vs metachronous; adjuvant treatment; number (N) crossed to RGF), n = 22 RGF (13 crossed to VOR/HCQ). 38 patients evaluable (at
of metastatic sites; liver-only disease. Results: IPD of 1,064 pts enrolled in least C1 completed). Median age 58.4, 40% NH vs 60% H. mPFS on first arm: 1.90
the PRODIGE 9, AIO KRK 0207 and SAKK 41/06 trials were collected. mo VOR/HCQ vs 4.35 mo RGF [p = 0.032, HR: 2.277]. mOS: 6.77 mo VOR/HCQ vs
Considering the different timing of randomization in PRODIGE 9 (at the start 7.23 mo RGF [p = 0.90, HR: 1.05]. Grade 3/4 AEs (see table). In both arms, there
of induction) vs AIO KRK 0207 and SAKK 41/06 (at the start of M), IPD of was trend towards decreased IL-1b, IL-2, IL-6, IL-10, TNFa, IFNg but an increase in
pts not progressed during induction and starting M phase entered the GM-CSF after treatment. Responders (4+ cycles) had lower baseline MaxMAF versus
analysis. 909 pts were included, 457 (50%) received BV M. Median PFS nonresponders for both arms. In responders, there was trend toward a decrease in
from induction start was 9.6 and 8.9 months in BV group vs no M group, MaxMAF at C2 and then increase at progression. Conclusions: VOR/HCQ did not
respectively (HR 0.78; 95%CI: 0.68-0.89; p , 0.0001). At the multivariate improve survival when compared to RGF. VOR/HCQ has a favorable safety profile, but
further planned subgroup analysis is pending to identify biomarkers of efficacy in
PFS analysis, BV M, resected primary tumor, N of metastatic sites and liver-
responders. Clinical trial information: NCT02316340.
only disease were significant. No difference in terms of OS between the 2
groups was observed. Conclusions: This is the first IPD meta-analysis in- Grade 3/4 AEs VOR/HCQ (n) RGF (n)
vestigating the role of BV alone M vs no M after first-line induction CT+BV in anemia 1 -
mCRC pts. Despite the significant PFS improvement in favor of BV M, the bilirubin increased - 1
absolute benefit appears limited, and without a clear clinical relevance. On diarrhea 2 -
these bases, a predictive nomogram to identify pts most likely to benefit from elevated transaminases - 2
hand foot syndrome - 2
BV M is under evaluation and will be presented during the Congress. hypertension - 1
hypokalemia - 1
fatigue 1 1
thrombocytopenia 3 -

3552 Poster Session (Board #44), Mon, 8:00 AM-11:00 AM 3554 Poster Session (Board #46), Mon, 8:00 AM-11:00 AM
Blood-based genomic profiling of cell-free DNA (cfDNA) to identify Clinical, pathological and prognostic features of rare BRAF mutations (MTs)
microsatellite instability (MSI-H), tumor mutational burden (TMB) and in metastatic colorectal cancer (mCRC): A bi-institutional retrospective
Wnt/B-Catenin pathway alterations in patients with gastrointestinal (GI) analysis (REBUS study). First Author: Brunella Di Stefano, Fondazione
tract cancers. First Author: James Isaacs, Duke University, Durham, NC Policlinico Universitario A. Gemelli-IRCCS-UOC Oncologia Medica, Rome,
Italy
Background: MSI-H cancers are responsive to immune checkpoint blockade
(ICB), but nearly half of all patients experience primary or early treatment Background: Recently, 3 classes of BRAF MTs have been described. BRAF
resistance. Activation of the WNT/B-Catenin pathway can lead to immune V600 MTs, which identify mCRC with poor prognosis and not benefitting from
exclusion and may drive resistance to ICB. Methods: 12 patients had stage anti-EGFR drugs, belong to class 1. Class 2 and 3 include BRAF non-V600
III (N = 1) or IV (N = 11) MSI-H GI tract (small bowl, colon, or rectal) cancers. MTs, which occur in about 1-2% mCRC and are associated to favourable
Blood samples were obtained after (N = 5) or during (N = 5) ICB. 2 patients prognosis and specific clinicopathologic features. Class 2 and 3 differ in kinase
did not receive ICB. Blood samples from 8 patients with microsatellite stable activity and sensitivity to anti-EGFR: class 2 are activated and RAS-
(MSS) metastatic colorectal cancer were included as controls. The Guardant independent MTs; class 3 are kinase-dead and sensitive to inhibition of
Health (Redwood City, CA) Omni 2.0 mb panel was used to analyze cfDNA. This study aims to retrospectively evaluate features and prognostic role of rare
We analyzed MSI-H status, TMB, and mutations within the WNT/B-Catenin BRAF non-V600 compared to BRAF V600E MTs in mCRC pts treated at 2
pathway, including APC, RNF43 and CTNNB1. Results: Of 12 patients with Italian Institutions. Methods: mCRC pts harboring BRAF MTs, assessed by
MSI-H GI cancers, 1 sample failed enrichment due to hemolysis. MSI-H was means of NGS, pyrosequencing or RT-PCR, treated between Jan-13 and Dec-
not detected in 2 patients with a history of MSI-H in tissue; however these 18 at 2 Italian Institutions, were retrospectively analyzed. Clinico-pathological
patients had a complete response to ICB at the time of blood collection. The and treatment characteristics and survival data were collected. Results: 55 pts
Omni panel identified MSI-H in the remaining 9 patients with MSI-H disease bearing BRAF MTs were identified. Of those, 46 (84%) harbored a V600E and
in tissue. Among 8 control patients with MSS disease in tissue, MSI-H was 9 (16%) a non-V600 MT. Within the non-V600 group, 3 MTs (K601E, G469A,
not detected. Median TMB (mutations/Mb) was greater for MSI-H specimens G469R) belonged to class 2, while 5 MTs (G466E, G466A, 2 D594G,
(109; range 30-807) than for MSS specimens (13; range 6-24). All 8 patients D594N), belonged to class 3. One pt harboured a T599I MT, whose kinase
with MSS GI cancers were identified to have APC mutations, and none were activity is unknown. Compared to BRAF V600E mCRC, BRAF non-V600 mCRC
found to have CTNNB1 or RNF43 mutations. Of 9 evaluable MSI-H GI cancers, were more frequently left-sided (p .017) and displayed a lower grade (p .045).
2 had APC mutations alone. The remaining 7 carried RNF43 mutations In addition, non-V600 mCRC pts had a lower tumor burden (involving one
(G659fs). All patients with RNF43 mutations were found to have disease metastatic site) (p .026) and underwent more frequently to resection of
progression while on ICB. Among these 7 patients with RNF43 mutations, 6 metastases with radical intent (77.7 vs 18%; p .000175). mOS was signif-
had additional mutations in APC or CTNNB1. Conclusions: Blood based ge- icantly longer in the non-V600 compared to the V600E group (61.3 vs 20.4 m;
nomic profiling can identify MSI-H cancers. Patients with MSI-H cancers HR 0.41, 95%CI 0.18-0.93; p .05). No difference in activity and efficacy of
resistant to ICB in this cohort have mutations in RNF43 as well as additional anti-EGFR agents was observed between class 2 and 3. Conclusions: Despite
mutations in APC or CTNNB1, suggesting that co-activation of the WNT/B- the small size of our retrospective analysis, the results were consistent with
Catenin pathway may be biologically important. Further study of the role of previous evidences. BRAF non-V600 MTs identified a subgroup of mCRC,
WNT/B-Catenin pathway activation in ICB resistance will be pursued using differing both in terms of clinicopathologic characteristics and prognosis from
tumor tissue from this cohort. BRAF V600 mCRC. Interestingly, the better prognostic features allowed more
frequently radical resection of metastases, positively impacting on survival.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
204s Gastrointestinal (Colorectal) Cancer

3555 Poster Session (Board #47), Mon, 8:00 AM-11:00 AM 3556 Poster Session (Board #48), Mon, 8:00 AM-11:00 AM
Randomized phase II trial of adjuvant hepatic arterial infusion (HAI) + systemic Chemotherapy rechallenge or reintroduction (CTr/r), regofenib (REG) and
FOLFIRI +/- panitumumab (Pmab) in patients with resected RAS wild type TAS-102 for metastatic pretreated colorectal cancer (mCRC) patients (pts):
colorectal cancer hepatic metastases (CRLM). First Author: Nancy E. Kemeny, A propensity score analysis of treatment beyond second-line (PROSERpINA
Memorial Sloan Kettering Cancer Center, New York, NY Study). First Author: Maria Alessandra Calegari, Oncologia Medica, Fon-
dazione Policlinico Universitario “A. Gemelli”, IRCCS, Roma, Italy
Background: HAI therapy has improved recurrence free (RF) survival in several
randomized trials after resection of CRLM. The purpose of this trial was to determine Background: The optimal treatment for mCRC beyond 2nd line is still
whether systemic Pmab added to adjuvant HAI + FOLFIRI in RAS WT pts increases questioned. Recently, REG and TAS-102 showed to improve survival com-
15 months RF survival (RFS) after liver resection. Secondary endpoints are eval- pared to BSC. While in real-world practice CTr/r is often considered in this
uation of overall survival, toxicity, and predictive biomarkers. Methods: RAS WT pts setting, supporting evidences are limited. In absence of studies comparing all
with resected liver mets were randomized to HAI + SYS (+/-) Pmab after stratification these strategies, we aimed to compare the prognostic performance of CTr/r,
by clinical risk score ($ 3 or , 3) and previous chemotherapy (Y/N). For a particular REG and TAS-102 in mCRC treated beyond 2nd line. Methods: mCRC pts
arm, if 24 or more patients (pts) were alive and RF at 15 months, the regimen in that
progressing after at least 2 lines of CT, treated with CTr/r, REG or TAS-102
arm would be considered worthy of further investigation. The initial dose of HAI
FUDR 0.12mg/Kg + dexamethasone was infused over the first 2 weeks of a 5-week
between Jan-10 and Jan-19 were considered eligible. The primary endpoint
cycle. Systemic chemotherapy was delivered on days 15 and 29 (irinotecan 125 was OS; secondary endpoints were PFS and RR. Cox’s proportional hazard
mg/m2, LV 400 mg/ m2, 5FU 1000 mg/m2 48-hour continuous infusion and +/- Pmab models for survivals were estimated. A propensity score (PS) adjustment for
6mg/kg). Patient characteristics were compared between arms using Fisher’s exact baseline characteristics was further accomplished for survival analysis.
test and Wilcoxon rank-sum test. Survival curves were estimated using the Kaplan- Results: The clinical data of 341 pts (CTr/r 133, REG 150, TAS-102 58) were
Meier method and compared by the log-rank test. Results: After randomization of 75 retrospectively collected. At multivariate analysis type of treatment, ECOG PS,
pts, the arm receiving + Pmab met the decision rule of having $ 24 patients alive and number of metastatic sites and treatment line independently correlated with
RF at 15 months. The two arms had similar pt characteristics and toxicity, with the OS (p , .001, p .001, p , .001 and p .029, respectively). The mOS was 18.5
exception of Pmab related rash (Table). The 15-month RFS is 79% and 67% in +/- (95% CI, 14.3–22.7), 6 (95% CI, 5.6–9.5) and 7.6 months (95%CI,
Pmab arms, respectively. With a median follow-up of 45 months, 3-year RFS is 65% 5.6–9.5), for CTr/r, REG and TAS-102 group, respectively (log-rank p ,
[CL 0.45-0.78] and 42% [CL 0.24-0.57], and 3-year survival is 96% and 90% in .0001). mOS was significantly longer for pts receiving CTr/r than for those
+/-Pmab arms, respectively. Conclusions: In this trial, the addition of Pmab to HAI treated with REG/TAS-102 (15.8 vs 7.1 months; adjusted HR 1.96, 95% CI
and SYS showed promising activity without increase in biliary toxicity and should be 1.44-2.66; p , .0001) at the PS analysis, adjusted for ECOG PS, number of
further investigated in a larger study. Predictive biomarkers will be presented. metastatic sites and treatment line; 2-yrs OS was 34% and 11.6% for CTr/r and
Clinical trial information: NCT01312857. REG/TAS-102, respectively. PFS was significantly longer for pts receiving CTr/r
ARM A ARM B than for those treated with REG/TAS-102 (5.5 vs 3.9 months; HR 1.45, 95%
HAI + SYS + Pmab HAI + SYS CI 1.11-1.91; p .006) at the PS analysis. Accordingly, RR was higher in pts
Patient Characteristics (n = 37) (n = 38)
receiving CTr/r compared to REG/TAS-102 (29.0 vs 1.5%; Chi-square p ,
Synchronous 81 % 76 %
‡ 3 metastases 41 % 40 % .00001). Conclusions: Our analysis, although underpowered, generates the
CRS ‡ 3 43 % 45 % hypothesis of a superiority of CTr/r in comparison to REG or TAS-102, in both
Post op CEA > 5 19 % 16 % efficacy and activity. Given the retrospective nature of our analysis, and the
Largest diameter ‡ 5 30 % 18 %
Toxicity (Grade 3/4) potential role of selection bias in treatment assignment, a prospective vali-
Diarrhea 19 % 11 % dation is mandatory.
Alk phos 8% 10 %
Stents 3% 3%
No significant difference in any factors

3557 Poster Session (Board #49), Mon, 8:00 AM-11:00 AM 3558 Poster Session (Board #50), Mon, 8:00 AM-11:00 AM
A phase I study of PolyPEPI1018 vaccine plus maintenance therapy in Exosomes as novel prognostic biomarker in potentially resectable colorectal
patients with metastatic colorectal cancer with a predictive biomarker cancer liver metastatic (CCLM) patients. First Author: Ina Valeria Zurlo,
(OBERTO). First Author: Joleen Marie Hubbard, Mayo Clinic, Rochester, MN Oncologia Medica, Università Cattolica del Sacro Cuore, Rome, Italy
Background: The goal of this study was to evaluate the safety, tolerability and Background: Target therapies and new surgical strategies deeply modify the
immunogenicity of a single dose of PolyPEPI1018 as an add-on to main- history of CCLM patients (pts). Several prognostic scoring systems have been
tenance therapy in subjects with metastatic colorectal cancer (mCRC). developed but no one is able to identify pts who should be excluded from a
PolyPEPI1018 is a peptide vaccine containing 12 unique epitopes derived potentially useless surgery. Currently research is committed in identifying
from 7 conserved cancer testis antigens (CTAs) frequently expressed in early biomarkers able to discern pts who could benefit from an aggressive
mCRC. These epitopes were designed to be Personal EPItopes (PEPIs), i.e. approach. Exosomes are arising as promising biomarkers in cancer. The aim
predicted by our novel PEPI test to bind to at least three autologous HLA of this pivotal study was to analyze the association among exosome levels
alleles and more likely to induce T-cell responses than epitopes presented during CCLM-pts treatment, clinical outcomes and the KRAS status.
by a single HLA. Methods: mCRC patients in the first line setting received the Methods: We enrolled 22 pts with CCLM candidate to preoperative che-
vaccine (dose: 0.2 mg/peptide) just after the transition to maintenance motherapy (pCT) and subsequent liver surgery. A blood sample was collected
therapy with a fluoropyrimidine and bevacizumab. Vaccine-specific T-cell before pCT, after surgery, monthly during follow-up and at progression (PD).
responses were first predicted by the PEPI test (using the patient’s complete Exosomes were isolated by ultracentrifugation and characterized by standard
HLA genotype and antigen expression rate) and then measured by ELISpot methods. Exosomes concentration was assessed by Bradford assay. We
after one cycle of vaccination. Results: Eleven patients were vaccinated with adopted ddPCR™ KRAS G12/G13 Screening Kit to evaluate the KRAS
PolyPEPI1018. The most common adverse events were transient skin re- status in exosomal DNA (e-DNA). Results: 22 CCLM pts received pCT and
actions (local inflammation at the site of the injections, e.g. erythema, underwent liver surgery: 5 major hepatectomies and 17 multiple liver re-
redness and itchiness) and flu-like syndrome. No grade 3 or higher adverse sections. Changes in exosomes plasma levels were found to correlate with
events related to the vaccine occurred. Initial analysis on 4 patients dem- each treatment step,resulting reduced after pCT and surgery and increased
onstrated that T-cell responses were elicited by 96% of vaccine peptides. at PD, respectively (p = 0.0026). Pts with higher baseline exosome levels
The overall percentage agreement between PEPI test-predicted and Elispot- experimented shorter PFS than those with lower levels (p = 0.0033 HR 0.2).
measured CD8+ T cell responses was 71%, consistent with our retrospective No association was found between exosome levels after liver surgery and
analysis on 64 vaccine clinical trials involving 1,790 patients. Two of these 4 disease free interval nor overall survival. KRAS status on e-DNA was eval-
patients had unexpected tumor size reduction. Based on these encouraging uated on 10 pts in baseline, in pCT, after surgery, and in PD samples. In 8 out
results, the trial was amended to administer 3 doses of PolyPEPI1018 given of 10 pts e-DNA displayed the same mutational status than the one detected
12 weeks apart. Conclusions: PolyPEPI1018 combined with maintenance on tumor DNA. Changes in e-DNA KRAS copies were found statistically
therapy was safe and well-tolerated in mCRC patients. Unprecedented significant in pCT vs surgery and pCT vs PD (p = 0.039; p = 0.04).
immune responses were induced after single dose, with broad CRC-specific Conclusions: Our study suggests a prognostic role of exosome levels in CCLM
T cell responses and high accuracy prediction of CD8+ T cell responses. This pts. Moreover, we showed that KRAS mutational status could be monitored
promising activity in mCRC patients led to a trial amendment to administer 3 during the post-surgery follow up by analyzing e-DNA. Overall, our data
doses of PolyPEPI1018 in combination with systemic therapy. Clinical trial confirm the potential role of exosomes in liquid biopsy tool to monitor
information: NCT03391232. molecular changes during the treatment of CCLM pts.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Colorectal) Cancer 205s

3559 Poster Session (Board #51), Mon, 8:00 AM-11:00 AM 3560 Poster Session (Board #52), Mon, 8:00 AM-11:00 AM
Gender and survival benefit from initial irinotecan in metastatic colorectal Predictive factors for early mortality after initiation of regorafenib or trifluridine/
cancer: Analysis of the XELAVIRI (AIOKRK0110) study. First Author: tipiracil in refractory metastatic colorectal cancer. First Author: Toshiki
Kathrin Heinrich, Department of Medicine III, University Hospital, Munich, Masuishi, Aichi Cancer Center Hospital, Nagoya, Japan
Germany
Background: Regorafenib (REG) and trifluridine/tipiracil (FTD/TPI) have
Background: XELAVIRI compared initial vs sequential irinotecan (iri) in been recognized as standard treatments for patients (pts) with refractory
combination with fluoropyrimidine (FP) plus bevacizumab (bev) in patients metastatic colorectal cancer (mCRC). Because these drugs have limits on
(pts) with mCRC, trial identification: NCT01249638. In the full analysis set efficacy benefit for some pts, we are necessary to select pts who may be
of the study, non-inferiority of time to failure of strategy (TFS) was not shown better not to receive REG or FTD/TPI. However, no reports are available on
(primary endpoint). Pts with RAS/BRAF wildtype (wt) tumors benefitted from how to predict pts with early mortality after initiation of these drugs.
initial iri. Methods: The study endpoints objective response rate (ORR), Methods: We retrospectively evaluated pts with mCRC who were registered
progression-free survival (PFS), time to failure of strategy (TFS) as well as in a multicenter observational study (the REGOTAS study). The main in-
overall survival (OS) were evaluated in female vs. male pts as well as mo- clusion criteria were ECOG PS of 0–2, refractory or intolerant to fluo-
lecular subgroups (i.e. RAS mutational status). Interaction of treatment and ropyrimidines, oxaliplatin, irinotecan, and anti-VEGF and anti-EGFR therapy
gender was tested by likelihood ratio tests. Results: Of 421 patients, 281/ (if KRAS wild type), and no prior use of REG or FTD/TPI. Predictive factors for
140 were male/female. In male patients, ORR was 33.6% without and early mortality (# 12 weeks from initiation of REG or FTD/TPI) were eval-
58.3% with initial iri (P , 0.001). PFS (HR: 0.54 (95%CI 0.42-0.69) P , uated by multivariate analysis for survival with all variables with P values
0.001) and OS (HR: 0.63 (95%CI 0.47-0.85), P = 0.002), were also of ,0.05 from the univariate analysis, using the Cox proportional hazards
significantly better with initial iri. In the subgroup analysis, this effect was model. In this analysis, the pts who lived at first 15 weeks were defined as
especially pronounced in pts with RAS/BRAF wt tumors. In female pts, ORR censored case. Results: A total of 523 pts (REG, 212; FTD/TPI, 311) were
was 43% in both arms, PFS was similar (HR: 1.09 (95%CI 0.76-1.55), P = eligible. Predictive factors for early mortality were without primary tumor
0.65) without and with initial iri. In OS, a strong trend for inferior outcome resection [adjusted hazard ratio (aHR), 1.56; p = 0.02], the low level of
with initial iri was seen (HR: 1.46 (95%CI 0.95-2.24), P = 0.08) that albumin (aHR, 2.31; p , 0.0001), the high level of CRP (aHR, 2.31; p ,
reached significance in the multivariate analysis (HR: 1.73 (95%CI 1.04- 0.0001), and short time from diagnosis of mCRC (aHR, 1.77; p = 0.002) by
2.86, P = 0.034). Female patients with RAS/BRAF wt tumors did not benefit multivariate analysis. The pts harboring all these poor factors were classified
from initial iri (HR 1.05 (95% CI 0.46-2.41), P = 0.903 for OS). Formal into the high (H) risk of early mortality group. Two groups of pts with H and
interaction of treatment and gender was seen for ORR (P = 0.018), PFS (P = non-H were identified, with 12-week mortality rate of 39 and 14%, with 14-
0.002) and OS (P = 0.001). There were some trends for more pronounced week mortality rate of 70 and 18%, and with median survival time (MST) of
toxicities in female pts treated with Irinotecan. Conclusions: This unplanned 2.9 and 7.8 months, respectively (HR of H/non-H, 4.02; p value , 0.0001).
exploratory analysis suggests that gender might interact with efficacy of In pts treated with REG, H and non-H groups had 12-week mortality rate of
initial iri when used in the context of FP and bev. While especially male RAS 38 and 16%, 14-week mortality rate of 79 and 18%, and MST of 2.8 and
wild-type patients derived a significant and clinically meaningful benefit 7.8 months, respectively. In pts treated with FTD/TPI, H and non-H groups
from initial use of iri, this was not observed in female patients with RAS wt had 12-week mortality rate of 41 and 13%, 14-week mortality rate of 63 and
tumors. Clinical trial information: NCT01249638. 18%, and MST of 3.2 and 7.7 months, respectively. Conclusions: Our
predictive model for early mortality after initiation of REG or FTD/TPI might
be useful for selecting pts who should not receive either these drugs.

3561 Poster Session (Board #53), Mon, 8:00 AM-11:00 AM 3562 Poster Session (Board #54), Mon, 8:00 AM-11:00 AM
Utilization and reach of the Fight Colorectal Cancer Late Stage MSS CRC Feasibility of a national expert panel to determine resectability in patients
Clinical Trial Finder. First Author: Reese Garcia, Fight Colorectal Cancer, with initially unresectable colorectal cancer liver metastases (CRLM). First
Springfield, MO Author: Karen Bolhuis, Amsterdam UMC, University of Amsterdam, The
Netherlands, Amsterdam, Netherlands
Background: Colorectal cancer (CRC) remains one of the most lethal cancer
killers worldwide. Recently, research has shown great strides in the treat- Background: Decision on optimal treatment strategy for CRLM remains
ment of MSI-H mCRC using immunotherapy, however, these treatments have complex because uniform (un)resectability criteria are lacking. We hy-
not been effective in MSS patients, who make up a majority of CRC cases. pothesize that the use of an expert panel can improve the identification of
Due to numerous barriers, clinical trial enrollment numbers remain as low as patients with potentially resectable CRLM. The Dutch Colorectal Cancer
9% of the eligible populations, despite the reliance of many late stage CRC Group (DCCG) Expert Panel was established in conjunction with the CAIRO5
patients on clinical trials for treatment. Perhaps greatest of these barriers is study (Huiskens J et al. BMC Cancer 2015), a multicenter, randomized,
the lack of meaningful patient-facing clinical trial matching, making ad- phase-3 trial, investigating optimal systemic induction treatment in patients
vances in MSS mCRC IO clinical research extremely slow. Methods: In May with initially unresectable CRLM. Here, we present the feasibility of this
2017, Fight Colorectal Cancer (FightCRC) launched its web-based trial panel. Methods: The DCCG Expert Panel consists of 13 liver surgeons and 4
finder, The Late Stage MSS Trial Finder (TF) with the late Dr. Tom Marsilije, a radiologists. Consensus was reached on predefined (un)resectability criteria
stage IV CRC patient and researcher, and Flatiron Health. The TF is a publicly at baseline. An online platform allowed resectability-assessment by 3 sur-
available immunotherapy-based repository of clinical trials. An algorithm geons in case of inter-surgeon agreement, and 5 surgeons if they disagreed.
automatically codes for a subset of trials from ClinicalTrials.gov to be CRLM were assessed as 1) resectable 2) potentially resectable, or 3) per-
uploaded into the tool, and trained FightCRC advocates follow a strategic manently unresectable. Patients with initially unresectable CRLM were
logic flow to prioritize trials of highest potential benefit and lowest risk for evaluated at baseline and subsequently every 2 months as long as CRLM
patients. Results: Between 30 and 100 trials are uploaded into the TF for were considered potentially resectable. Results: Overall, 397 panel evalu-
curation each week. A total of 378 trials have been indexed in the TF to date. ations in 183 patients were analyzed. Median time to panel conclusion was
In February 2019, a mobile application was introduced. From May 2017 to 7 days (IQR 5-11 days) and 204 (51%) evaluations showed inter-surgeon
January 2019, the tool has seen . 15,000 users, yielding 26,000 searches disagreement, with major disagreement (resectable versus permanently
in 105 countries; primarily from the United States, China, the United unresectable) in 24 (14%) and 12 (29%) evaluations after 2 and 4 months of
Kingdom, Canada, and France. On average, users navigate to 2.5 pages and systemic treatment. Ultimately, 84 (79%) patients with resectable CRLM
spend . 2.5 minutes per use. Providers are using this as a tool to find clinical underwent resection and 23 (27%) resections included portal vein embo-
trials and to discuss these options in real time. CRC patient feedback lization or 2-stage procedures. In resectable CRLM with inter-surgeon
confirmed the platform functionality. Conclusions: The Trial Finder is a agreement versus disagreement, R0 resection was achieved in 39 (75%)
unique tool for MSS mCRC patients pursuing clinical trials. The success of versus 28 (52%) patients, p = 0.013. Median time to recurrence was similar
the tool may be attributed to patient focused selection of therapies that show between resections with panel agreement versus disagreement, 8 versus
promise. The FightCRC late stage MSS CRC trial finder is being widely 6 months, p = 0.447. Conclusions: This study shows the feasibility of a
utilized, in diverse settings. With our patient curators and Medical Advisory national Liver Expert Panel for prospective resectability assessment of pa-
Board, FightCRC will improve the search features and outcome tracking with tients with initially unresectable CRLM. High inter-surgeon disagreement
user feedback. The goal for the TF is to address key barriers to patient entry supports the use of a panel. We aim to further validate the panel with
into clinical trials and promote patient-provider discussions to inform de- outcome parameters. Clinical trial information: NCT02162563.
cision making.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
206s Gastrointestinal (Colorectal) Cancer

3563 Poster Session (Board #55), Mon, 8:00 AM-11:00 AM 3564 Poster Session (Board #56), Mon, 8:00 AM-11:00 AM
Aggressiveness of care and overall survival in young metastatic colorectal Association between gene fusions and anti-EGFR resistance signature in
cancer patients. First Author: Madeleine Fish, Massachusetts General colorectal cancer. First Author: Thereasa A. Rich, Guardant Health, Red-
Hospital, Boston, MA wood City, CA
Background: Colorectal cancer (CRC) incidence in patients younger than 50 Background: Acquired resistance to anti-EGFR therapy in metastatic co-
years of age is steadily rising by 2% annually. Early-onset CRC usually lorectal cancer (mCRC) has been characterized by a circulating tumor DNA
presents with more aggressive features; however, data on prognosis are (ctDNA) signature including any sub-clonal RAS mutation, co-existing RAS
widely conflicting. Clinicians may hold an age-related bias in treating mutations, or co-existing EGFR mutations. Here, we investigated if fusions in
younger patients, but this proclivity and its effects have not been quantified. ctDNA are associated with this anti-EGFR signature for CRC patients (pts).
Methods: Patients with a history of metastatic CRC who consented to a Methods: 4289 advanced stage CRC pts underwent molecular profiling
departmental chart review protocol were collected between 2014 and 2018 using a plasma-based NGS assay that included FGFR2, FGFR3, RET, ALK,
at Massachusetts General Hospital. The cohort was divided into two groups NTRK1, and ROS1 fusions. Available clinical histories were reviewed.
based on age at initial diagnosis: , 50 and $50. Data were gathered on Correlations between fusions and clinicopathological features were mea-
treatments and clinicopathological features. A log-rank test compared sured with a Fischer exact test. Relative frequencies of genomic alterations
survival from the diagnosis of metastatic disease between age groups. The were compared between fusion-present vs -absent cases with an unpaired t-
distributions of clinicopathological features were compared using Wilcoxon test. Clonality for a given alteration was called for a relative variant allele
rank sum tests. Results: 464 metastatic CRC patients were identified. 155 frequency (rVAF) . 50 %, while subclonal was defined as , 50% rVAF.
patients (33%) were , 50 (median age 43, 49% female) and 309 patients Results: 44 unique fusions were detected in 40 (1.1%) of the 3808 pts with
(67%) were $50 (median age 61, 45% female). Sex did not significantly alterations present: RET (N = 16), FGFR3 (N = 12), ALK (N = 10), NTRK1
differ between the two groups (p = 0.45). Patients , 50 received more lines (N = 3), ROS1 (N = 2), and FGFR2 (N = 1). Relative to non-fusion variants
of therapy after metastatic diagnosis than patients $50 (mean 2.7 v. 2.2; p = detected, fusions were more likely to be subclonal (OR 8.2, p , 0.0001).
0.002). Younger patients also received more resections of distant metas- Mutations associated with a previously reported anti-EGFR resistance were
tases (mean 0.62 v. 0.48; p = 0.01). A higher rate of enrollment in clinical found in FGFR3 (7/12 pts), RET (7/15 pts) and ALK (5/10 pts). In fusion-
trials for patients , 50 approached significance (p = 0.06). Even so, present cases, co-existing RAS mutations were more likely to be subclonal
patients , 50 did not see a significant survival benefit over older patients (2/ than non-fusion cases (OR 14, p , 0.0001). EGFR mutations were more
5-year survival from metastatic diagnosis 77%/47% v. 73%/38%, p = 0.23). common in fusion present cases (OR 3.7, p = 0.0001) and predominantly
Patients , 50 had a lower proportion of right-sided tumors (p = 0.0002) and subclonal (97%). EGFR mutations aggregated to ectodomain sites (85%)
BRAF mutations (p = 0.0009). There was no difference in MSI status (p = previously linked to acquired anti-EGFR resistance. For 27 pts with available
0.28), RAS mutational status (p = 0.40), mucinous features (p = 0.53), or clinical histories, 21 (78%) received anti-EGFR treatment prior to ctDNA
signet ring features (p = 0.26). Conclusions: Overall survival in patients , 50 testing. Conclusions: Actionable fusions using a ctDNA NGS assay were
is similar to patients $50, despite patients , 50 receiving more aggressive predominantly subclonal and co-existed with subclonal RAS and EGFR
therapy. Further study is warranted to better understand these differences. mutations. These clinicopathologic features are consistent with a previously
Potential areas of interest include performance status, age-related treatment validated signature linked to resistance to anti-EGFR therapies in CRC. We
bias, and biological factors. hypothesize that fusions may arise as a previously undescribed mechanism
of acquired resistance to anti-EGFR therapies in CRC pts.

3565 Poster Session (Board #57), Mon, 8:00 AM-11:00 AM 3566 Poster Session (Board #58), Mon, 8:00 AM-11:00 AM
Survival according to mutations in BRAF, KRAS, or microsatellite instability Circulating tumor DNA dynamics, serial testing and evolution on treatment in
(MSI-H) after cytoreductive surgery and hyperthermic intraperitoneal 322 colorectal cancer patients. First Author: Pashtoon Murtaza Kasi, Mayo
chemotherapy (HIPEC) in patients with peritoneal metastases from colorectal Clinic, Jacksonville, FL
cancer. First Author: Stein G Larsen, Section for Surgical Oncology, Nor-
Background: According to the American Society of Clinical Oncology and
wegian Radium Hospital; Department of Gastroenterological Surgery, Oslo
College of American Pathologists (ASCO/CAP) joint review on circulating tumor
University Hospital, Oslo, Norway
DNA (ctDNA) issued in March 2018, widespread use of ctDNA assays in most
Background: Patients with metastatic colorectal cancer (mCRC) and mu- patients with advanced cancer is still an area of ongoing research. However,
tations in BRAF V600E (mutBRAF) or KRAS (mutKRAS) have a worse multiple studies thereafter published and/or presented support its use in
prognosis after liver or lung surgery/ablation, whereas the impact of patients with metastatic colorectal cancer (CRC). This has led to several in-
microsatellite instability (MSI-H) has not been well studied. Few patients stitutions adopting it as ‘clinical practice’. The aim of this study is to report on
with mutBRAF receive liver or lung surgery (1-4%), whereas mutBRAF is our institution’s adoption of ctDNA testing for every patient at the time of
present in 5-12% of mCRC trial patients and in up to 20% of the general diagnosis and/or time of progression. Methods: We report on results of 322
mCRC population. The frequency and prognostic role of mutBRAF, mutK- CRC patients with 607 ctDNA tests at our center from January 2017 to
RAS and MSI has not been well studied after cytoreductive surgery (CRS) and February 2019 using a commercially available platform (Guardant360).
hyperthermic intraperitoneal chemotherapy (HIPEC) for peritoneal metas- Results: Among 322 patients of our cohort, a total of 607 ctDNA tests were
tases from colorectal cancer. Methods: The Norwegian Radium Hospital is done (Table). 127 (39.4%) of these tests were serial analyses. In the CRC
the only center offering CRS and HIPEC in Norway. From 2004 to 2015 257 patients who had serial testing, at progression, mechanisms of resistance
patients with histology proven peritoneal metastasis from colorectal cancer, included acquisition of KRAS, NRAS, EGFR mutations; and HER2- and MET-
appendiceal cancer excluded, was consecutively enrolled. Molecular ana- amplifications. The subclonal mutations were noted to disappear when the
lyses of KRAS, BRAF and MSS/MSI in mutBRAF were done. Fourteen pa- selective inhibition was stopped. This was seen in patients on targeted
tients were excluded due to missing tumour blocks (7), unsuccessful therapies/biologics rather than chemotherapy. This was of value in treatment
analysis (4) and other malignant disease (1). Results: 180 of 243 patients modification, clinical trial selection and/or monitoring of disease progression in
obtained complete cytoreductive surgery and received HIPEC for 90 minutes these patients. Conclusions: While ctDNA testing may not be ready for pri-
with Mitomycin C (45-70mg). Median survival for the 180 patients was metime in all advanced cancers, it is increasingly being adopted in practice for
47 months and 5-year survival rate 40.1%. Median disease-free survival was especially metastatic CRC. Of particular value is the serial ctDNA testing in the
10 months. mutBRAF was found in 23.4% of cases, mutKRAS 35.1% and RAS/RAF wildtype subset and now BRAF V600E mutant CRC on anti-EGFR
double-wild type 41.5%. mutBRAF with MSS was found in 16.4%, mut- based therapies.
BRAF with MSI-H in 7.0%. 3-year disease free survival (DFS) and median
Total number of patients 322
overall survival (OS) was 38.9% and 59 months with mutBRAF with MSI-H,
significantly higher compared to 24.2% and 30 months in patients with Total number of tests 607
Number of serial analyses 127
double wild type, 13.2 % and 41 months in mutKRAS and 17.9% and RAS/RAF wild-type 214 (66.4%)
22 months in mutBRAF with MSS. Conclusions: A surprisingly high fre- Number of RAS mutations 83 (25.8%)
quency of mutBRAF was seen in mCRC patients after CRS and HIPEC for Number of V600E BRAF mutations 18 (5.6%)
Number of non-V600E BRAF mutations 7 (2.2%)
peritoneal metastatic disease. Patients with mutBRAF and MSI-H had a Number of HER2 amplifications 13 (4%)
significantly better DFS and OS after CRS and HIPEC. DFS for patients with Number of HER2 mutations 7 (2.2%)
mutBRAF and MSS was numerically lower but not statistically different from MSI-High* 3 (1%)
patients with mutKRAS or double wild type. *Of note, 25(7.8%) of CRC were dMMR/MSI-High in the cohort. CtDNA testing company started
reporting MSI-High later in 3rd quarter of 2018.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Colorectal) Cancer 207s

3567 Poster Session (Board #59), Mon, 8:00 AM-11:00 AM 3568 Poster Session (Board #60), Mon, 8:00 AM-11:00 AM
Young-onset colorectal cancer patients disproportionately diagnosed with Tumor and microenvironment evolution during chemotherapy combine with
advanced disease: Patient self-reported journey of symptom duration and bevacizumab in colorectal cancer liver metastasis. First Author: Wangjun
misdiagnosis. First Author: Ronit Yarden, Colorectal Cancer Alliance, Liao, Nanfang Hospital, Nanfang Medical University, Guangzhou, China
Washington, DC
Background: Identifying patterns of pathological tumor resistant to treat-
Background: Colorectal cancer (CRC) is the second leading cause of cancer- ment regimens and improving tumor’s chemotherapy sensitivity from lon-
related death among males and females in the US. Despite a decrease in gitudinal and multi-omics data integration are of paramount importance in
overall incidence and mortality, there has been an alarming increase of CRC order to provide the best chances of cure, especially, for colorectal cancer
diagnosis among young adults (20-49 years old). The Colorectal Cancer liver metastases. Methods: We elucidated genomic (ctDNA and whole exome
Alliance launched a comprehensive survey for young-onset CRC patients and sequencing), immunomic (DNA-based T cell receptor sequencing) and
survivors via social media to track the self-reported clinical, psychosocial, transcriptomic changes from liver metastases, peripheral blood and match
financial and quality of life experiences of this often overlooked, group. primary tumors that accompany the evolution of colorectal cancer liver
Methods: The survey was completed by 1195 living patients and survivors. metastases. In total, 197 histopathologically distinct areas of liver metas-
The majority of participants (57%) were diagnosed between the ages of 40 tases and 72 peripheral blood samples at multiple time points from 15
and 49, 33% of patients/survivors were diagnosed between the ages 30-39 patients with colorectal cancer were analysis in this study. Results: In
and about 10% were diagnosed before the age of 30. Only 8% of the re- responding patients, mutation load from plasma were reduced from baseline
spondents were diagnosed with Lynch syndrome although about 28% re- (P , 0.001), but changed slightly in tumor tissues (P = 0.351). Tran-
ported some family history. Results: Our survey revealed a higher proportion scriptomic analysis and immunohistochemistry revealed that increased
of the young-onset patients and survivors (71%), diagnosed with advanced infiltration of neutrophils and monocytes were associated with worse out-
stage tumors, compared with ACS report for overall CRC patients (60%). The comes and insensitive response to chemotherapy (Neutrophils: P = 0.003;
late stage diagnosis subjected young patients to aggressive therapies and a Monocytes: P = 0.032). Activation of fatty acid metabolism, JAK-STAT,
substantial decrease in quality of life including neuropathy, anxiety, clinical PPAR, insulin and TGF-b signaling pathway were observed in progressive
depression, and sexual dysfunctions. Most respondents (63%) waited 3- tumor. TCR diversity in response metastatic regions was significantly in-
12 months before visiting a doctor, with higher proportion of females waited creased compared with non-responder (P = 0.008). Combination of bev-
more than 12 months compared with males (22% vs. 15% p = 0.02). acizumab with chemotherapy facilitated T cell infiltrating to the tumor
Moreover, even when visited their doctors, most patients indicated that they microenvironment which might consequently benefit from checkpoint im-
were initially misdiagnosed. The majority of the respondents (67%) saw at munotherapy (P = 0.006). Conclusions: Our integrative and comparative
least 2 physicians, and some more than 4 physicians, prior to their diagnosis. genomic analysis provides a new paradigm for understanding the evolution
Patients that saw 3 or more physicians prior to diagnosis were more likely to and treatment resistance of colorectal cancer liver metastases, with im-
be diagnosed with advanced disease. Interestingly, half of the patients that plications for identifying ways to advance treatment regimen and monitoring
were seen by one physician also claimed they were initially misdiagnosed. treatment response of colorectal cancer liver metastases.
Conclusions: Our survey indicates that medical professionals and young
adults need to be aware of the increasing incidence of young-onset CRC, and
the importance of timely screening when signs and symptoms are present,
regardless of age. Yet, 50% of physicians did not explain to the patients’
family members about their elevated risk of the disease and their need for
screening.

3569 Poster Session (Board #61), Mon, 8:00 AM-11:00 AM 3570 Poster Session (Board #62), Mon, 8:00 AM-11:00 AM
Serial ctDNA analysis as a real-time indicator of neoadjuvant chemoradio- Anti-epidermal growth factor receptor therapy in combination with chemo-
therapy in rectal cancer. First Author: Jiaolin Zhou, Peking Union Medical radiotherapy for the treatment of locally advanced anal canal carcinoma:
College Hospital, Peking Union Medical College, Chinese Academy of Results of a phase II study with panitumumab (FFCD 0904). First Author:
Medical Sciences, Beijing, China Thomas Aparicio, Department of Gastroenterology, Saint Louis Hospital, Paris,
France
Background: Neoadjuvant chemoradiotherapy (nCRT) is nowadays the stan-
dard of care for the locally advanced rectal cancer (LARC). However, there is no Background: Standard treatment of anal squamous cell carcinoma is 5-
effective method to predict patients’ possible benefits from nCRT and monitor fluorouracil (5FU) and mitomycin C (MMC) based chemoradiotherapy (CRT).
the response to it. Methods: Patients with locally advanced middle and low This phase II study studied the tolerance and complete response (CR) rate at
rectal cancer of stage cT3-4N0M0 or cTanyN+M0 were enrolled from August 8 weeks of panitumumab (Pmab) combined with MMC-5FU based CRT.
2017 to July 2018. All patients received nCRT with long-term radiation Methods: Patients with locally advanced tumor without metastases (Stage T2,
plus fluorouracil based chemotherapy, followed by the radical surgery. Serial T3 or T4, whatever N stage; Stage N1-N3 whatever T stage) were treated with
plasma samples were collected pre-nCRT, during nCRT, and preoperatively two RT periods (45Gy in 5 weeks and a boost of 20Gy in 2 weeks) with
(8 weeks after the completion of nCRT). Somatic mutations were detected concomitant CT sessions of 5FU/MMC at RT weeks 1 and 5. Pmab was ad-
with next-generation sequencing using a 1021-gene panel with peripheral ministered on RT weeks 1, 3, 5 and 7 according to the doses defined by a
blood lymphocyte DNA as a germline control. Results: This prospective cohort previous phase 1.study (MMC: 10 mg/m² at J1 and J29; 5FU: 400 mg/m² from
study enrolled 61 patients with rectal cancer. The pathological complete J1 to J4 and from J29 to J32, Pmab: 3mg/kg). The expected rate of CR at
response (pCR) rate and the downstage rate was 31% (19/61) and 80% (49/ 8 weeks to continue in phase III was 80%. Results: Forty-five patients (male: 9
61), respectively. ctDNA was detectable in 77% (47/61), 18% (11/61) and (20%), female: 36 (80%); median age: 60.1 [41.5-81]) were enrolled in 15
13% (8/61) of blood samples obtained pre-nCRT, during nCRT and pre- French centers. All patients but one completed the CRT. Median duration of
operatively, respectively. No significant association was observed between pre- CRT was 52 days [30-76].Fourteen patients had a RT interruption because of
nCRT ctDNA status with any clinicopathological factors, including age, gender, toxicity. Most common related grade 3-4 toxicities observed were digestive
differentiation or tumor circumferential extent. Among the 8 patients with (51.1%), hematologic (lymphopenia: 73.4%; neutropenia: 11.1%), radiation
detectable ctDNA preoperatively, pathological tumor regression grade (TRG) of dermatitis (28.8%) and asthenia (11.1%). On patient died because of
CAP 2-3 were observed and hepatic metastasis was found in 4 patients within mesenteric ischemia during the CRT (total dose: 36 Gy). In ITT analysis, the CR
2 months. For patients with undetectable pre-operative ctDNA, a higher rate at 8 weeks after CRT was 66.7% [90%CI: 53.4-78.2]. Median follow-up
proportion archived pathological downstaging (85% vs 50%). The correlation was 16.2 months [14.1-18.2]. Overall survival, recurrence-free and
between preoperative ctDNA status and achievement of pathological down- colostomy-free survival at one year were 94.6% [95%CI: 75.8-98.7], 72.2%
stage was independent of age, gender or differentiation (p = 0.02). In addition, [95%CI: 55.0-83.7] and 78.2% [95%CI: 60.6 – 88.6] respectively. Six
preoperative ctDNA positivity was associated with the persistently involved (13%) patients had a colostomy with abdomino-perineal amputation due to a
lymph node (p = 0.02). However, neither pre-nCRT nor during-nCRT ctDNA tumour recurrence. Conclusions: Despite an acceptable tolerance, pan-
status was associated with pathological downstaging or persistently lymph itumumab in combination with CRT for locally advanced anal cancer failed to
node involvement. Conclusions: Detectable ctDNA after the completion of meet the expected CR rate to justify further clinical trials. Clinical trial in-
nCRT is a predicator of unsatisfactory curative effect of patients with LARC, formation: NCT01581840.
which might indicate novel treatment intensification studies. Clinical trial
information: NCT03042000.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
208s Gastrointestinal (Colorectal) Cancer

3571 Poster Session (Board #63), Mon, 8:00 AM-11:00 AM 3572 Poster Session (Board #64), Mon, 8:00 AM-11:00 AM
Outcomes of salvage surgery for anal squamous cell carcinoma: A systematic Factors affecting differential outcomes in the definitive treatment of anal
review and meta-analysis. First Author: David Patrick Cyr, Department of cancer between HIV+ and HIV- patients. First Author: Matthew Susko,
Surgery, University of Toronto, Toronto, ON, Canada University of California San Francisco Medical Center, San Francisco, CA
Background: Following primary chemoradiotherapy for squamous cell car- Background: Anal cancer is an uncommon malignancy with numerous factors
cinoma of the anal canal (A-SCCa), 10% of patients have persistent cancer, that influence treatment outcomes. Historically, HIV+ patients were restricted
and 30% develop local recurrence after initial complete response. Both from entering clinical trials, limiting data on their outcomes to small retro-
persistent and recurrent A-SCCa may be amenable to salvage surgery, which spective reports. This study seeks to understand the factors related to anal
typically involves intense health care resource utilization. Because only cancer outcomes, specifically the differences between HIV+ and HIV- patients.
small cohorts have been reported, we synthesized the evidence for salvage Methods: Inclusion criteria was non-metastatic anal squamous cell carcinoma
surgery to gain a comprehensive understanding of outcomes. Methods: We treated with a definitive course of chemotherapy and radiation between 2005
systematically searched MEDLINE, Embase, and Cochrane Library (until and 2018 at a single institution. Clinical data related to baseline character-
October 11, 2018) for studies reporting on persistent or recurrent A-SCCa istics, treatment parameters, and post-treatment follow-up were extracted for
treated with salvage surgery. Quality assessment was performed using the calculation of freedom from local recurrence (FFLR) and overall survival (OS).
Institute of Health Economics Quality Appraisal Checklist. Overall survival Univariate analysis (UVA) and multivariate analysis (MVA) were done using cox
(OS) and disease-free survival (DFS) were pooled using two approaches: proportional hazard model, and FFLR and OS were calculated using the
survival curve meta-analysis, and exact binomial likelihood random-effects Kaplan-Meier method. Results: During the study period, 111 patient initiated
model for survival probabilities. We used meta-regression, subgroup and definitive treatment for anal cancer. Median age was 56.7 years (IQR: 51.4-
sensitivity meta-analyses to explore sources of heterogeneity. Results: We 63.5), and 47% (N = 52) were HIV+. At median follow-up of 28 months, 12
identified 39 observational studies that included 1388 patients. Pooled 5- and 24-month FFLR was 84.1% and 78.2% respectively, with 24-month OS of
year OS was 45.5% (95% CI 40.6 to 49.9; 33 studies; 1308 patients), and 87.3%. MVA demonstrated significant association between FFLR and T-stage
did not differ in patients resected for recurrent (14 studies, n=259 patients) HR 4.02 (95% CI: 2.14-7.55) p , 0.001, elapsed treatment time (median of
vs. persistent (n=238 patients) disease. There was no association of OS with 50 days) 1.08 (95% CI: 1.04-1.12) p , 0.001, and diagnosis to treatment
study year, tumor size, or resection margin at the aggregate-level. Pooled 5- start (median time of 15 weeks) 1.05 (95% CI: 1.01-1.08) p = 0.005. Ad-
year DFS was 38.3% (95% CI 31.4-43.9; 14 studies; 554 patients). Pooled ditional analysis with log-rank test for FFLR demonstrated significant differ-
30-day complications were 65.3% (95% CI 50.2-77.9; 17 studies; 720 ence between patients taking , 50 days to complete treatment (p = 0.03),
patients): major complications 27.7% (95% CI 22.3-33.8), reoperations and , 15 weeks from diagnosis to treatment completion (p = 0.006). In HIV+
12.7% (95% CI 8.7-18.2), and mortality 1.7% (95% CI 1.1-2.6%). Pooled patients, post-treatment CD4 , 150 was significantly associated with worse
perineal complications were 32.7% (95% CI 25.0-41.4). Conclusions: Salvage OS on log-rank test (p = 0.016), with pretreatment CD4 values being non-
surgery offers 5-year OS of »45% and DFS of »40% for recurrent/persistent A- significant. Conclusions: This study represents the largest single institution
SCCa. Major complications and perineal wound complications are common, but report of HIV positive patients treated for anal cancer. No difference in local
postoperative mortality is rare. Presently, there are no reports of patient-reported recurrence or overall survival between HIV+ and HIV- patients was elucidated;
outcomes such as quality of life after salvage surgery for A-SCCa. Comparative however, HIV+ patients with lower post-treatment CD4 counts had worse OS.
effectiveness studies comparing surgery to other treatments are warranted. The most significant predictors of local recurrence were advanced T-stage,
increased time from diagnosis to treatment initiation, and prolonged treatment
time.

3573 Poster Session (Board #65), Mon, 8:00 AM-11:00 AM 3574 Poster Session (Board #66), Mon, 8:00 AM-11:00 AM
Chemoradiotherapy for anal cancer: A population-based study of treatment Combination of tissues analysis and immune infiltrate in localized colon
interruption, treatment completion, and associated outcomes. First Author: cancer using artificial intelligence in PETACC8 study. First Author: Cynthia
Michael J. Raphael, Department of Oncology, Queen’s University, Kingston, Reichling, CHU Dijon, Dijon, France
ON, Canada
Background: We used artificial intelligence to perform tissue classification
Background: Chemoradiotherapy (CRT) is the standard treatment for squamous and count CD3 and CD8 in each subclass and determined their role in
cell anal carcinoma (SCCA). Here, we describe CRT delivery in routine practice outcome prediction in PETACC8 cohort of stage III colon cancer treated with
and explore the association between treatment interruption, non-completion FOLFOX or FOLFOX plus cetuximab. Methods: We developed artificial in-
and outcomes. Methods: The Ontario Cancer Registry was used to identify all telligence aimed to detect tumor, healthy mucosa, stroma and immune cells
incident cases of SCCA treated with curative intent RT in Ontario, Canada on whole slide of CD3 and CD8 staining. The invasive margin (IM) was also
(2007-2015). Treatment interruption was defined a priori as . 7 days between automatically determined. Using a lasso algorithm, the software was able to
consecutive fractions. Completed CRT was defined as receiving $50 Gy and 30+ detect digital parameters within the tumor core (TC) which were related to
fractions of RT along with 2 concurrent doses of chemotherapy. Log-binomial patients’ outcome (variable called DGMate for DiGital tuMor pArameTErs).
regression models were used to estimate risk ratios (RR) between patient CD3 and CD8 lymphocytes density were also quantified automatically by the
characteristics and 1) failure to complete CRT and 2) salvage abdominoperineal software in TC and at IM. Associations with disease-free survival (DFS) were
resection (APR). Cox proportional hazard models were used to estimate hazards evaluated by multivariable Cox regression adjusting for age, T/N stage,
ratios (HR) between treatment interruption or non-completion and 1) cancer sidedness, KRAS/BRAF, DNA mismatch repair (MMR). Results: On 1220
specific survival (CSS) and 2) overall survival (OS). Results: We identified 1593 samples collected, data could be generated for 1018 patients. We observed
patients with SCCA; 73% (n = 1161) initiated curative intent RT. Median RT that a high IM stromal area and a high DGMate were associated with a poorer
dose and duration was 54 Gy (IQR, 50.4-67.8) and 46 days (IQR, 42-53), DFS [HR 5.65 (95% CI, 2.34, 13.67), p , 0.0001; HR 2.72 (95% IC, 1.92,
respectively. Treatment interruption . 7 days occurred in 23%. CRT was 3.85), p,0.001 respectively for the continuous variable]. A higher density of
completed by 59%. Factors associated with CRT non-completion were age CD3+ TC, CD3+ IM and CD8+ TC were significantly associated with a longer
. 70 vs. , 50 (RR 0.70, 95% CI: 0.59-0.93) and greater comorbidity (1+ vs. 0, DFS (HR 0.75 (95% IC, .66, .87), p,0.0001; HR 0.78 (95% IC, .68, .88),
RR 0.57, 95% CI: 0.39-0.85). Treatment interruption . 7 days appears to be p,0.0001; HR 0.83 (95% IC, .71, .96), p=0.01). All these immune
associated with salvage APR (RR 1.39, 95% CI: 0.91-2.12). In an exploratory variables were significantly correlated with each other. ANOVA test dem-
analysis, the association between treatment interruption . 10 days and salvage onstrated that CD3+ TC gave a similar prognostic value compared to the
APR reached statistical significance (RR 1.63, 95% CI, 1.06-2.53). Treatment classical CD3/CD8 immunoscore (p=0.44). The combination of IM stromal
interruption was not associated with inferior CSS (HR 0.87, 95% CI: 0.60- area, DGMate and CD3 outperformed the classical CD3/CD8 immunoscore
1.25) nor OS (HR 0.96, 95% CI: 0.76-1.23). Failure to complete CRT was not to estimate patients’ prognosis (C-index= 0.601 vs 0.578, p-value=0.04).
associated with higher rates of salvage APR nor inferior CSS, but was associated Adding this new variable to classical clinical prognostic parameters we
with inferior OS (HR 1.40, 95% CI: 1.13-1.72). Conclusions: In routine clinical generated a nomogram which predicted the risk of relapse of stage III colon
practice, treatment interruption and non-completion among patients with SCCA cancer with a stronger predictive value compared to clinical parameters or
are common. Quality improvement initiatives to optimize treatment continuity the immunoscore. Conclusions: We propose a new fully automated method
and completion are needed. The observed association between failure to of whole slide analysis using a software based on artificial intelligence which
complete CRT and OS is likely a reflection of confounding by indication, which classify tissue and determine tumor and immune parameters on one single
is highlighted by the lack of association between CRT completion and CSS. slide stained with CD3 antibody. This valuable strategy outperforms
immunoscore and clinical outcome prediction models.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Colorectal) Cancer 209s

3575 Poster Session (Board #67), Mon, 8:00 AM-11:00 AM 3576 Poster Session (Board #68), Mon, 8:00 AM-11:00 AM
Prognosis and chemosensitivity of non-V600E BRAF mutations in metastatic High incidence of advanced stage cancer and prolonged rectal bleeding
colorectal carcinoma (mCRC): An AGEO French multicenter retrospective history before diagnosis in young-onset patients with colorectal cancer. First
cohort. First Author: Aline Derosiere, Amiens University Hospital, Amiens, Author: Gurprataap Singh Sandhu, University of Pittsburgh Medical Center,
France Pittsburgh, PA
Background: BRAF mutations are present in 5-15% of mCRC. V600E BRAF Background: In contrast to the older population, the incidence of colorectal
mutations account for ~80% of cases and are mostly found in right-sided cancer (CRC) in younger patients (aged , 50 years) has been increasing in
tumors. Non-V600E BRAF mutations are rare (~2% of mCRC), mostly left- the last three decades. Younger patients tend to present with more advanced
sided. Although BRAF V600E mutations are associated with a dismal disease, thought to be in part related to lack of routine screening colo-
prognosis, some studies suggest that non-V600E BRAF mutations may be noscopies. The goal of this study was to examine characteristics of young-
associated with a favorable outcome. The chemosensitivity of non-V600E onset CRC and potentially identify factors that may aid in earlier diagnosis
BRAF-mutated mCRC has never been studied. Methods: From 2017 to and treatment. Methods: We collected data for patients available through the
2018, all consecutive patients (pts) with non-V600E BRAF-mutated mCRC University of Colorado Cancer Center Cancer Registry. Inclusion criteria
(next generation sequencing) treated in the participating centers were in- included: 1) Diagnosis of colon or rectal cancer between the years 2012-
cluded. Survival analyses were performed using Kaplan-Meier method and 2018 and 2) age at diagnosis of less than 50 years. Pertinent data including
LogRank test. Results: A total of 108 pts in 34 centers in France were baseline characteristics, clinical presentation, family history, pathology,
included between October 2017 and August 2018 (median age, 66 years molecular testing, staging, and treatment were collected. Results: 211
[range, 58-77]; ECOG performance status 0-1, 86%). The primary was patients with young-onset CRC were available for review. Mean age at di-
mostly left-sided (66%). Main metastatic sites were the liver (73%), lungs agnosis was 42.4 years and 55.5% were males. A total of 42.1% had rectal
(33%), lymph nodes (39%) and peritoneum (26%). D594 (34%), G469 cancer and a majority of the colon cancer diagnoses had left-sided tumors
(15%), K601 (11%), N581 (7%) and L597 (7%) were the most frequent (66%). Regarding clinical presentation, 52.2% presented with rectal
mutations. A concomitant RAS mutation was found in 22% of pts. Micro- bleeding prior to diagnosis. Of those who presented with rectal bleeding, the
satellite instability (MSI) was found in 3/67 pts (4.5%). First-line chemo- average time from the onset of bleeding to diagnosis was 271.17 days.
therapy (CTx) (n = 69) efficacy was (overall response rate/disease control 42.9% of young-onset CRC were stage IV at the time of initial diagnosis.
rate) 49%/77% (anti-EGFR-containing CTx [n = 20], 75%/85%; Evaluation of the pathology specimens showed that 89.6% were adeno-
antiangiogenic-containing CTx [n = 22], 55%/73%). Median overall sur- carcinomas and 63.5% were grade 2 or higher. At diagnosis, the mean BMI
vival (mOS) was 25.6 months (95% CI : 17.1-43.8) overall; it was was 26.6 and the mean CEA was 135.5. A total of 72.5% of young-onset
8.0 months with best supportive care alone (n = 10), 16.0 months with patients had a positive family history of any cancer. KRAS or NRAS mu-
palliative CTx alone (n = 63), and attained 105.1 months with curative- tations were present in 49.6% of patients, BRAF V600E mutations were
intent management of metastases (n = 35). mOS did not differ according to present in 3.8%, and 10.8% were MSI-H. Conclusions: Prolonged rectal
sidedness (p = 0.22), type of mutation (p = 0.52), or its functional impact on bleeding history prior to diagnosis was noted in a significant proportion of
BRAF (p = 0.19). Conclusions: Non-V600E BRAF-mutated mCRC retain young-onset patients with colorectal cancer. Patients and primary care
sensitivity to CTx + biologics and harbor a good prognosis (especially when physicians should be made aware of this finding in order to facilitate timely
amenable to curative-intent surgery), regardless of the type of mutation and referral for colonoscopy which may lead to earlier diagnosis, less advanced
its impact on BRAF kinase function. Contrarily to BRAF V600E mutations, disease at diagnosis, and improved outcomes.
non-V600E mutations may occur along with RAS mutations, but un-
commonly with MSI.

3577 Poster Session (Board #69), Mon, 8:00 AM-11:00 AM 3578 Poster Session (Board #70), Mon, 8:00 AM-11:00 AM
Association of TGF-b expression with intratumoral infiltration of cytotoxic Dissection of the lateral lymph nodes with short axis of $ 5 mm affects the
T lymphocytes in patients with microsatellite-stable colorectal cancer. First prognosis in rectal cancer patients: A central review of 738 lower rectal cancer
Author: Amir Mehrvarz Sarshekeh, University of Texas MD Anderson Cancer magnetic resonance images. First Author: Koya Hida, Kyoto University, Kyoto,
Center, Houston, TX Japan
Background: Transforming growth factor- b pathway (TGF-b) has an estab- Background: The usefulness of lateral lymph node dissection (LLND) for lower
lished role in promoting growth, invasion, metastasis as well as epithelial to rectal cancer has been discussed. Concerning relapse-free survival (RFS), in
mesenchymal (EMT) transition. Among 4 different described molecular the JCOG0212 trial, the non-inferiority of avoiding LLND for rectal cancer
subtypes of colorectal cancer (CRC), consensus molecular subtype 4 (CMS4) without LLN enlargement of $10 mm was not shown; however, the superiority
comprises up to 25% of CRC pts, distinguished by activation of this pathway, of LLND was also not demonstrated. Methods: To examine whether the dif-
and is associated with higher relapse rate and poor prognosis. Recently, it has ference in dissection effect could be demonstrated by the size of LLN,
also been proposed that TGF-b activation drives immune evasion in murine magnetic resonance imaging (MRI) for the central review of pre-registered
models, but these findings have not been clinically validated. Methods: Using patients with stages II/III lower rectal cancer below the peritoneal reflection
multi-gene RNA expression profiling, fresh-frozen paraffin-embedded samples was performed, and the prognosis of the patients were prospectively in-
of 35 patients with CRC were analyzed to determine TGF-b and EMT ex- vestigated as a cohort. The relationship between LLND and RFS was also
pression levels. Multiplexed IHC staining was performed on FFPE tumor blocks evaluated. Results: MR images of 738 cases (mean distance from anal verge:
by using the Opal 7-Color fIHC Kit and the stained slides were scanned by a 4.5 cm) were evaluated. Twenty patients with LLN enlargement of $10 mm
Vectra multispectral microscope (PerkinElmer) to measure infiltration of im- were excluded from the analysis. Overall, of 718 patients, 310 did not undergo
mune cells (i.e., T lymphocytes, cytotoxic T lymphocytes (CTL), T cell antigen- LLND (non-LLND group), whereas the other 408 underwent LLND (LLND
experienced, macrophages, etc.) in the tumor, stroma, and both components. group). The lymph node sizes of $5 mm were observed in 10.0% and 21.1% of
TGF-b and EMT expression levels – as continuous variables - were compared patients in the non-LLND and LLND groups, respectively. Although there was
with the infiltration of various immune cells using Spearman’s rank correlation some bias in the background characteristics of the two groups (non-LLND vs.
analysis. Results: Among 35 pts, 28 pts had non-CMS1/MSS CRC. TGF-b RNA LLND; median age 66 vs. 61 years; laparoscopic surgery 64.2% vs. 17.9%;
expression in the tumor microenvironment of these samples was inversely preoperative treatment 39.4% vs. 27.9%; circumferential resection margin
associated with the infiltration of CTL into the tumor (r=-0.43, p= 0.022). In evaluated by MRI 37.3% vs. 49.9%; and postoperative complications 33.2%
contrast, there was no association of TGF-b with non-cytotoxic T-cells or vs. 44.4%), almost no difference in the 5-year RFS was observed (63.4% vs.
macrophage infiltration. The tumor and stromal CTL infiltration differed 67.5%).The LLND group had significantly more favorable with $5 mm LLN
substantially by CMS (p=0.04, p=0.02, respectively) with tumor infiltration (non-LLND 51.6% vs. LLND 69.0% P = 0.023) than with , 5 mm LLN (non-
lowest in CMS4 (n=7). Consistent with this, EMT gene signature, which in- LLND 64.7%; LLND 67.0%). Conclusions: Oncological benefit of LLND was
cludes TGF-b expression, showed a similar inverse correlation with CTL in- suggested in patients with enlarged lateral nodes on MRI using a large cohort of
filtration (r=-0.48, p=0.009). Conclusions: TGF-b and EMT gene signatures clinical stage II/III lower rectal cancer.
have important roles in the exclusion of CTL in the tumor microenvironment of
CRC pts. Inhibiting TGF-b pathway can potentially increase the intratumoral
infiltration of CTL, which is a necessary (but not sufficient) step for immu-
notherapy response in MSS CRC. Clinical trials evaluating this hypothesis are
currently ongoing (NCT03436563).

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
210s Gastrointestinal (Colorectal) Cancer

3579 Poster Session (Board #71), Mon, 8:00 AM-11:00 AM 3580 Poster Session (Board #72), Mon, 8:00 AM-11:00 AM
Population-based screening for BRAF V600E in metastatic colorectal cancer Retrospective analysis of overall survival (OS) by subsequent therapy in
(mCRC) to reveal true prognosis. First Author: Jenny E. Chu, University of patients (pts) with RAS wild-type (wt) metastatic colorectal cancer (mCRC)
British Columbia, Vancouver, BC, Canada receiving irinotecan 6 cetuximab in the EPIC study. First Author: Alberto F.
Sobrero, IRCCS Ospedale San Martino IST, Genova, Italy
Background: BRAFV600E (BRAF) mutations (mts) portend poor prognosis in mCRC
and patients (pts) may die before ascertainment. Since 2014, Vancouver Coastal Background: The multicenter, open-label, randomized, phase 3 EPIC study
Health (VCH) has performed reflex hereditary screening of CRCs with BRAF and (EMR 062202-025) investigated cetuximab + irinotecan vs irinotecan as a
mismatch repair (MMR) immunohistochemistry (IHC). We evaluated this BRAF mt second-line (2L) therapy in pts with EGFR-detectable mCRC. A retrospective
population-based cohort (BRAFPOP) to establish the true prognosis of BRAF mts in analysis in the EPIC RAS wt population showed improved overall response rate
mCRC. Methods: We reviewed all mCRCs from VCH between 4/2014 and 5/2018 for (29.4% vs 5.0%) and progression-free survival (5.4 vs 2.6 mo) upon the ad-
BRAF by IHC (VE1 antibody). Overall survival (OS) from stage IV diagnosis was dition of cetuximab to irinotecan. Median OS (mOS) was similar in both
compared to mCRCs with next generation sequencing (NGS) determined BRAF mts
treatment arms, possibly due to differences in subsequent therapies. We
(BRAFNGS) from BC Cancer & MD Anderson. BRAFNGS OS did not differ by center (p =
0.77). Results: See table for BRAF cohort baseline characteristic comparison. present OS by post-study therapy in the RAS wt population. Methods: 1298
BRAFPOP pts had worse OS than BRAFNGS pts (HR 2.5, 95% CI 1.6 – 3.9, P , RAS-unselected pts were enrolled from May 2003 to February 2006. The
0.0001). Median OS for all BRAF mt pts was 17.9 mos. Both groups had worse OS primary endpoint was OS. RAS status was determined retrospectively in 2018
than wild type pts (P , 0.0001). 52 (81%) of BRAFPOP pts were referred to oncology, from existing DNA samples using BEAMing technology; wt status was defined as
40 (63%) received chemotherapy, and 12 (19%) had NGS BRAF testing. BRAFPOP having a sum of mutated RAS allele frequencies of #5%, with all relevant
pts who had NGS testing with BRAF mts had OS comparable to other BRAFNGS pts alleles being analyzable. Results: Among the 452 pts with RAS wt mCRC, 231
(P = 0.89) and better OS than BRAFPOP pts that never had NGS testing (HR 0.37, 95% received cetuximab + irinotecan and 221 received irinotecan. Baseline char-
CI 0.18-0.76, P = 0.030). Pts with BRAF mts and MMR deficiency (dMMR) (n = 40) acteristics were similar in both arms. OS data by post-study therapy are
had worse OS than MMR proficiency (pMMR, n = 202) (1.6, 95% CI 1.0-2.5, P = summarized in the Table. No new or unexpected safety signals were observed.
0.011). This was driven by BRAFPOP dMMR pts (HR 1.9, 95% CI 0.9-4.0, P = 0.036) Conclusions: Post-study cetuximab was associated with improved OS in both
as no difference was seen by MMR in BRAFNGS pts (HR 1.3, 95% CI 0.8-2.2, P = treatment arms compared with post-study therapy without cetuximab and no
0.30). Conclusions: Current estimates of prognosis for mCRC with BRAF mts likely subsequent therapy, suggesting that cetuximab-based therapy may be suitable
underestimate its impact due to referral bias for NGS testing. BRAF mts with dMMR as a standard treatment for pts with RAS wt mCRC in the rechallenge setting.
are associated with worse prognosis than pMMR. This appears driven by BRAFPOP pts. Study limitations include a potential bias due to the differences in proportion of
BRAF Wild subsequent therapies with and without cetuximab between arms (almost 50%
BRAFPOP BRAFNGS P (POP vs Type of pts in the irinotecan arm received post-study cetuximab) as well as the
(n = 64) (n = 255) NGS) (n = 1762) likelihood for pts who live longer to receive cetuximab in any subsequent therapy
Female (%) 36 (56) 129 (51) 0.42 757 (43) line.
dMMR (% of known) 15 (23) 25 (14) 0.083 43 (3)
Subsequent Subsequent
pMMR (% of known) 49 (77) 153 (86) 1394 (97) therapy with therapy without No subsequent
Median age at diagnosis 72 [56- 61 [51- , 0.0001 55 [46-63] cetuximab cetuximab therapy Total
[Interquartile Range] 84] 68]
Cetuximab + irinotecan
Histology: Adenocarcinoma (%) 55 (86) 186 (73) 0.031 1487 (84) n 26 98 107 231
Mucinous/Signet Ring (%) 9 (14) 69 (27) 275 (16) mOS (95% CI), mo 28.0 (26.55-NE) 13.8 (12.16-17.12) 8.2 (6.51-11.76) 12.3 (11.37-14.09)
Synchronous Metastases (% of 40 (63) 172 (80) 0.0032 1131 (64) Irinotecan
known) n 104 37 80 221
mOS (95% CI), mo 19.1 (13.21-21.45) 9.5 (8.18-17.91) 3.9 (3.12-5.88) 12.0 (9.36-14.92)
OS (months) 6.0 20.7 , 0.0001 39.8
NE, not estimable.

3581 Poster Session (Board #73), Mon, 8:00 AM-11:00 AM 3582 Poster Session (Board #74), Mon, 8:00 AM-11:00 AM
Association between pretreatment Fusobacterium nucleatum and cancer Germline multigene panel testing in colorectal cancer: Precision therapy and
pain at six months postsurgery in newly diagnosed colorectal cancer patients: clinical management implications. First Author: Edward D. Esplin, Invitae,
Results from the ColoCare Study. First Author: Anita Roselyn Peoples, San Francisco, CA
Huntsman Cancer Institute, University of Utah, Salt Lake City, UT
Background: Recent studies suggest that the prevalence of abnormalities in
Background: Pain is a prevalent, debilitating symptom in more than half of homologous recombination deficiency (HRD) genes and other cancer genes
cancer patients. Accumulating evidence suggests a bi-directional relationship not traditionally associated with colorectal cancer (CRC) may be more
between gut microbiota and pain, potentially via inflammation and oxidative common in patients with CRC than previously appreciated. Herein, we in-
stress. Fusobacterium nucleatum (Fn), a pro-inflammatory anaerobic bacte- vestigate the efficacy of comprehensive multigene panels in patients with
rium, is frequently detected in colorectal cancer (CRC) patients. To date, no CRC to identify candidates for precision therapies. Methods: DNA se-
study has identified a relationship between Fn and cancer pain in CRC pa- quencing and exon-level copy number analysis were performed in over 9000
tients. We investigated the associations between pre-treatment Fn and cancer patients (pts) referred because of personal history of colon cancer between
pain at 6 months post-surgery in CRC patients. Methods: We utilized pre- 2013 and 2018 at a commercial diagnostic laboratory. The genes requi-
surgery stool samples collected from 80 prospectively followed, newly di- sitioned varied but consistently included 14 genes on a hereditary CRC
agnosed CRC patients recruited from the German site of the international panel; the patient data were de-identified and further analyzed for all 83
ColoCare Study. Eligible patients were neo-adjuvant treatment naı̈ve and did genes on a large hereditary cancer syndrome panel under an IRB-approved
not use antibiotics for at least 1 month before stool collection. Fn DNA was protocol. Results: Pathogenic/likely pathogenic (P/LP) findings were iden-
assessed via quantitative real-time polymerase chain reaction. Patients were tified in 2101 of 9669 pts (21%), 1838 (19%) pts when MUTYH hetero-
median split into Fn-high (.17.27; n=40) or Fn-low (#17.27; n=40). Cancer zygotes are excluded. When restricted to five Lynch syndrome (LS) genes,
pain was assessed using the 2 pain symptom items from the European Or- only 9% of patients had a P/LP finding, which increased to 15% of patients
ganization for Research and Treatment of Cancer Quality of Life Questionnaire- when 19 guidelines-based CRC genes were assessed. 137 pts (1.4%) had
Core-30 (lower score=lower pain) at pre- and 6 months post-surgery. two or more P/LP variants. P/LP variants were in MLH1, MSH2, MSH6,
Results: Before surgery, 48% of all patients reported any pain. At 6 months PMS2, CHEK2, APC, MUTYH, BRCA2, ATM, BRCA1, PALB2, RAD50,
post-surgery, we observed a decrease in cancer pain by 33% for Fn-low, while BRIP1, TP53, EPCAM, among others, of which 1.4% were BRCA1/2.
there was an increase in cancer pain by 41% for Fn-high. After controlling for When a comprehensive multigene panel was utilized P/LP variants in
pre-surgery cancer pain in ANCOVA model, we observed significantly higher genes with known therapeutic implications, such as in HRD and mismatch
mean cancer pain at 6 months post-surgery in the Fn-high group vs. the Fn-low repair deficiency (MMRD), were detected in 1408 (14%) of patients, and
group (24.07 vs. 13.44; effect size, ES=0.45; p=0.04). These results were 1670 (17%) had P/LP variants in genes with established clinical man-
maintained even after controlling for age, sex, tumor stage and site, adjuvant agement guidelines. Conclusions: This study suggests that 1 in 5 patients
chemotherapy, BMI, physical activity, and any pain medications (29.11 vs. with CRC harbor actionable germline variants, up to one-half of which remain
16.55; ES=0.53; p=0.03). Conclusions: These findings suggest that high Fn undetected when only LS genes are tested. Comprehensive panel testing
is an independent predictor of cancer pain at 6 months post-surgery in co- identified candidates for precision treatment and established management
lorectal cancer patients. Further research is needed to confirm and understand recommendations, and have clinical implications for both pts and their at
the mechanisms of these results. Funding: NCI U01 CA206110. risk family members.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Colorectal) Cancer 211s

3583 Poster Session (Board #75), Mon, 8:00 AM-11:00 AM 3584 Poster Session (Board #76), Mon, 8:00 AM-11:00 AM
Is the predictive and prognostic impact of sporadic and familial micro- Association of MAPK signaling subtypes with prognostic benefit for bevaci-
satellite instable stage III colon cancer different? A pooled analysis of the zumab in left-sided metastatic colorectal cancer (mCRC) patients treated with
PETACC8 and NCCTG N0147 (Alliance) trials. First Author: Aziz Zaanan, FOLFIRI + cetuximab / bevacizumab (FIRE-3 trial). First Author: Arndt Stahler,
Hopital Européen Georges Pompidou, Paris, France Department of Medicine III, University Hospital, LMU Munich, Munich,
Germany
Background: The Microsatellite instability (MSI) or deficient mismatch re-
pair (dMMR) phenotype is usually taken as a single biological entity whereas Background: We investigated the role of the MAPK pathway by mRNA ex-
no data are available concerning prognosis and response to chemotherapy pression profiles in microarrays of the FIRE-3 trial as it was formerly as-
between sporadic and familial dMMR cases. Methods: Resected KRAS exon sociated with prognosis. Methods: 451 patients provided eligible mRNA
2 wild-type (WT) tumor stage III colon cancers (N = 4596) from patients (pts) material for subsequent analyses of the MAPK pathway (295 genes). Non-
randomly assigned to FOLFOX +/- cetuximab in two adjuvant large phase III negative matrix factorized (NMF) clustering for normalized mRNA microarray
trials were prospectively analyzed for MSI status and dMMR mechanism data (Almac Inc, Xcel Array) was performed for 2 to 6 ranks against ran-
(sporadic vs familial). Stratified Cox models were used to assess prognostic domized controls. Linear models with adjustment for multiple testing
and predictive values of dMMR mechanism by treatment arms, adjusting for showed differential gene expression between groups. Single sample gene set
age, gender, tumor grade, ECOG PS, pT/pN stage and primary tumor lo- enrichment analysis (ssGSEA) was used to compare differentially enriched
cation. Results: Among dMMR patients with complete data for dMMR hallmarks of cancer gene sets. Kaplan Meier method, log rank test and Cox
mechanism analysis (N = 354), there were 255 (72%) sporadic (BRAF regression analyses were performed to estimate overall (OS) and progression
mutated or WT with MLH1 methylation) and 99 (28%) familial (loss of MSH2 free survival (PFS) between MAPK subtypes. Results: NMF clustering built
or MSH6, or loss MLH1 with BRAF WT and unmethylated MLH1) cases. A two groups of MAPK mRNA expression (coph: 0.91, silh: 1.00) without
large proportion of dMMR sporadic cases were mutated for BRAF (n = 200; cohort-based bias in principal component analysis. Group MAPK1 (n = 238)
80%). In pts treated with FOLFOX, the disease-free survival (DFS) was not was significantly associated with CMS2 (66.4 %), group MAPK2 (n = 213)
statistically different by dMMR mechanism, while for pts treated with with CMS4 (67.6 %, p , 0.0001). 5.551 of 23.561 genes were significantly
FOLFOX + cetuximab, the sporadic cases did worse than familial cases (DFS; differentially expressed between MAPK subtypes. 49 cancer hallmark gene
adjusted (adj) HR, 2.69; 95% CI, 1.02-7.08; P= 0.04). Considering the sets were significantly differentially enriched in ssGSEA (MAPK1: myc
predictive value, a deleterious effect of adding cetuximab to FOLFOX was targets, DNA repair, cell cycle, PI3K-AKT-mTOR pathway upregulation;
observed in sporadic (DFS; adjHR, 1.68; 95% CI, 1.01-2.79; P= 0.04) but MAPK2: EMT-related signatures, TGFß pathway, angiogenesis upregulation
not in familial dMMR pts (interaction P value regarding treatment effect = among others). In overall analysis, MAPK1 showed slightly better outcome
0.03). Furthermore, a non-significant trend to a deleterious effect of adding than MAPK2 (OS: HR: 0.80, 95% CI: 0.65 – 0.99, p = 0.049; PFS: HR:
cetuximab to FOLFOX was observed in BRAF mutant (DFS; adjHR, 1.66; 0.81, 95% CI: 0.66 – 1.00, p = 0.05). However, MAPK1 was significantly
95% CI, 0.95-2.92; P= 0.07) but not in BRAF WT pts. Conclusions: The more favourable for bevacizumab treatment in OS (MAPK1: 30.8 m,
addition of cetuximab to FOLFOX was associated with reduced DFS in MAPK2: 19.4 m, HR: 0.56, 95% CI: 0.39 – 0.81, p = 0.002) and PFS
patients with sporadic dMMR cases. Further studies including the meth- (MAPK1: 11.7 m, MAPK2: 9.8 m, HR: 0.68, 95% CI: 0.48 – 0.98, p =
ylator phenotype (CIMP) analysis are needed to validate these results. 0.038) in left sided tumors, while no difference was seen for cetuximab
Clinical trial information: NCT00265811 and NCT00079274. treatment, RAS and BRAF status. Conclusions: mCRC subtypes by MAPK
mRNA expression might contain prognostic information for the treatment
with bevacizumab beyond mutational status in patients with left sided tu-
mors of the FIRE-3 trial.

3585 Poster Session (Board #77), Mon, 8:00 AM-11:00 AM 3586 Poster Session (Board #78), Mon, 8:00 AM-11:00 AM
Influence of WNT and DNA damage response pathway alterations on outcomes Immune-related gene signature in predicting prognosis of early-stage colorectal
in patients with unresectable metastatic colorectal cancer. First Author: cancer patients. First Author: Jia Ke, The Sixth Affiliated Hospital, Sun Yat-sen
Sebastian Mondaca, Department of Medicine, Memorial Sloan Kettering University, Guangzhou, China
Cancer Center, New York, NY
Background: Immune-related genes (IRGs) were found to be associated with
Background: We assembled a large series of consecutive patients with the prognosis of colorectal cancer (CRC) patients. The aim of this study was
unresectable metastatic colorectal cancer (mCRC) to identify genomic to evaluate the impact of IRGs in predicting prognosis of early-stage CRC
biomarkers of response and survival. Methods: Patients with unresectable patients. Methods: According to the CIT microarray data set, 309 early-stage
mCRC treated at Memorial Sloan Kettering with genomic tumor profiling CRC patients were selected for generation of immune-related gene signature
between 2014 and 2017 were included. Patients who underwent upfront (IRGS). 5 independent data sets included 1587 CRC patients with complete
metastasectomy or received neoadjuvant/conversion chemotherapy were prognostic information were divided into a training cohort (566 patients) and
excluded. Clinical information was retrieved from electronic medical re- two validation cohorts (624 patients in validation-1 and 397 patients in
cords, and we evaluated associations between genomic profiles with pro- meta-validation). Prognostic analysis were performed to test the predictive
gression free survival (PFS) on first-line chemotherapy and overall survival value of IRGS. Results: Of 309 early-stage CRC patients, a prognostic im-
(OS). Categorical data were analyzed by Fisher exact test and time-to-event mune signature included 23 immune-related genes was constructed. In the
data were analyzed by Cox proportional hazards models. Results: Of 1453 training cohort, when considering patients with early tumor stage (I or II),
mCRCs profiled in this period, 471 patients met the study criteria. Median IRGS significantly stratified patients into immune low- vs high-risk groups in
age was 59 years (range, 18 to 95), and 73% of patients were stage IV at terms of disease-free survival (HR = 5.03, 95%CI = 2.94-8.62, P , 0.001).
diagnosis. Most tumors (91%) were microsatellite stable (MSS). The most Similarly, higher IRGS was correlated with significantly worse prognosis of
frequent first-line regimen was FOLFOX +/- bevacizumab (66%). Among early-stage CRC patients in validation-1 (HR = 2.71, 95%CI = 1.44-5.08,
MSS patients treated with oxaliplatin-containing regimens (n = 305), 7% P = 0.001) and meta-validation cohort (HR = 3.10, 95%CI = 1.60-6.00, P ,
harbored alterations in genes associated with DNA damage response (DDR) 0.001). When compared with Oncotype DX, we found IRGS achieved an
(BRCA1, BRCA2, ATM, PALB2). DDR gene alterations were not associated improved survival correlation in the training cohort (mean C-index, 0.85 vs
with PFS (P = 0.94) nor were different quartiles of large-state transitions (P = 0.65) and the validation-1 cohort (mean C-index, 0.72 vs 0.61). After in-
0.54). Genomic alterations that significantly varied by duration of response tegrated with clinical characteristics, IRGS remained as an independent
included BRAF (16%, 10%, and 5% for PFS , 6 months, 6-12 months, prognostic factor after adjusting for T stage and TNM stage of tumor in
and . 12 months, respectively) and APC (62%, 74%, and 80% for PFS multivariate analysis (HR = 2.02, 95%CI = 1.61-2.53, P , 0.001). Fur-
, 6 months, 6-12 months, and . 12 months, respectively). APC mutation, thermore, IRGS stratified immune low-risk group patients with adjuvant
single or dual, was associated with significantly longer PFS (HR 0.67) and chemotherapy showed even worse disease-free survival when compared with
OS (HR 0.59) in multivariate analysis versus no WNT pathway alteration or those without adjuvant chemotherapy (HR = 5.66, 95%CI = 3.153-10.16,
alterations in other WNT pathway genes (RNF43, AXIN2, CTNNB1). P , 0.001 in the training cohort and HR = 3.21, 95%CI = 1.74-5.92, P ,
Conclusions: In unresectable mCRC patients, mutations in APC were as- 0.001 in the validation-1 cohort). IRGS identified immune high-risk group
sociated with better outcomes; absence of an APC alteration or the oc- obtained a significantly higher immune and stromal infiltration (P , 0.001).
currence of other WNT pathway alterations was associated with shorter Particularly, the percentages of Macrophages M2 and CD8+ T cells infiltration
survival. Somatic alterations in DDR genes were not associated with out- were significantly different between these two groups. Conclusions: The
comes in mCRC patients receiving oxaliplatin-containing regimen. proposed prognostic IRGS is a promising system for estimating DFS of co-
lorectal cancer patients, especially those in early-stage. Further studies are
needed to evaluate the clinical utility of this system in predicting prognosis of
colorectal cancer patients.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
212s Gastrointestinal (Colorectal) Cancer

3587 Poster Session (Board #79), Mon, 8:00 AM-11:00 AM 3588 Poster Session (Board #80), Mon, 8:00 AM-11:00 AM
Identifying anti-EGFR (EGFRi) response subgroups using evidence of ctDNA Molecular landscape of colorectal cancers harboring R-spondin fusions. First
selective pressure. First Author: Christine Megerdichian Parseghian, Author: Andreas Seeber, Department of Internal Medicine V (Hematology
University of Texas MD Anderson Cancer Center, Houston, TX and Oncology), Innsbruck, Austria
Background: Metastatic colorectal cancers (mCRC) that respond to EGFRi Background: Gene fusions involving R-spondin (RSPO) family members
display a robust circulating tumor DNA (ctDNA) signature that reflects se- have been shown to drive Wnt-dependent tumor initiation in colorectal
lective pressure and clonal evolution. Conversely, non-responding tumors do cancer (CRC). Therapies targeting Wnt pathway are being actively investi-
not exhibit this signature. On this basis, we developed a novel method that gated for tumors harboring RSPO2/3 fusions. Here we set out to characterize
defines EGFRi sensitivity with improved biological confidence with fewer the molecular features of CRC with and without RSPO fusions to gain insight
patients (pts), and does not rely on clinical trial outcomes where responses into potential rationale combination therapy strategies. Methods: Tumor
may be confounded by concurrent chemotherapy. We used this method to DNA sequencing of 592 genes (NextSeq, Illumina), RNA sequencing of 53
further elucidate the association of several features that have been previously gene fusions (ArcherDx FusionPlex) and immunohistochemistry for PD-L1 on
reported to be associated with EGFRi resistance, namely tumor sidedness, tumor cells (SP142) were tested on CRC tumors at Caris Life Sciences,
BRAF, PIK3CA, or ERBB2 (HER2) MTs, and the absence of APC/TP53MT. Phoenix, AZ. Molecular profiles of RSPO2/3 positive (pos) were compared
Methods: We analyzed 112 pts with baseline tissue based RASWT mCRC who with negative (neg) tumors, Fisher-Exact was used for comparative analysis.
had progressed following EGFRi, and with plasma samples available for Results: A total of 1356 CRC samples were analyzed. RSPO3 and RSPO2
ctDNA sequencing using a blood based NGS assay. Using our previously fusions were detected in 42 (3.1%) and 4 (0.3%) samples, respectively,
validated EGFRi exposure signature, we identified pts with evidence of including 5 fusion events not previously reported (e.g., IFNGR1-RSPO3). A
selective pressure. Results: Post EGFRi ctDNA found 37% and 33% of pts female predominance was seen in RSPO fusion pos vs. neg tumors (71.7% vs
with left sided and transverse tumors displayed evidence of selective 45.0%, p , 0.001); no association with age or tumor sidedness was seen.
pressure, respectively. 0 pts with right sided tumors displayed evidence of RSPO2/3 fusions were mutually exclusive of MSI-high (0 vs. 5%), ERBB2
selective pressure; p= 0.01. Similarly, BRAFV600EMT displayed no evi- alterations (0 vs. 1% mutation, 4% amplification) and other Wnt pathway
dence of selective pressure vs 30% of WT pts; in contrast, selective pressure activation drivers including APC (2 vs. 75%), CTNNB1 (0 vs. 1.4%) and
was evident in pts with PIK3CAMT, ERBB2MT and pts with absence of APC/ RNF43 (0 vs. 5.3%) mutations. Significantly higher BRAF (26 vs. 7%),
TP53MT (42% vs 28%, 67% vs 28%, 24% vs 43%, respectively for MT vs RAF1 (4.5 vs. 0.4%) and SMAD4 (30 vs. 11%) mutation rates were seen in
WT, p= NS for all). BRAF, PIK3CA, ERBB2, and APC/TP53 MT were present RSPO pos vs. neg tumors (p , 0.05). A universal co-activation of MAPK
in 4/117, 12/108, 3/118 and 30/91 pts, respectively. ctDNA shedding was pathway (KRAS, NRAS or BRAF) was seen with RSPO fusions. There was a
similar for all subgroups, as was time from previous EGFRi, indicating that significantly elevated PD-L1 expression in RSPO3 pos tumors (14%)
these factors were not confounders. Conclusions: Consistent with prior large compared to RSPO neg (6%, p = 0.04) and APC-mutated (5%, p = 0.02)
randomized studies, no pts with right sided tumors or BRAFMT had evidence tumors that are MSS. Conclusions: This is the largest series of CRC cases
of biologic benefit as assessed by presence of selective pressure. In harboring an RSPO rearrangement reported to date. Comprehensive mo-
contrast a number of pts with transverse tumors, ERBB2MT, PIK3CAMT or lecular analyses asserted the unique molecular landscape associated with
absence of APC/TP53MT had evidence of EGFR selective pressure, con- RSPO fusions in CRC and suggested potential combinatorial approach to
firming that these are not absolute predictors of EGFR resistance and target Wnt/MAPK pathway. The immune modulatory effects specific to
suggesting a subset of these pts were deriving benefit from EGFR inhibition. RSPO2/3 fusion revealed by PD-L1 expression suggest co-targeting Wnt
This biology based approach has the potential to more efficiently evaluate pathway with PD1/PDL1 inhibitors in RSPO pos tumors.
biomarkers of targeted therapy in the future without reliance on large ran-
domized datasets.

3589 Poster Session (Board #81), Mon, 8:00 AM-11:00 AM 3590 Poster Session (Board #82), Mon, 8:00 AM-11:00 AM
Performance comparison of the methylated BCAT1/IKZF1 ctDNA test (COL- Prediction model for detecting circulating tumor DNA (ctDNA) in metastatic
VERA) with the CEA assay for detection of recurrent colorectal cancer. First colorectal cancer (mCRC). First Author: Allan Andresson Lima Pereira,
Author: Erin L. Symonds, Flinders University, Adelaide, SA, Australia Hospital Sı́rio-Libanês, Brası́lia, Brazil
Background: Early detection of recurrent colorectal cancer (CRC) will im- Background: While tissue-based assays have yields above 90% in solid tu-
prove treatment options, but the current standard blood test of mors, there is less known about factors that influence the sensitivity of ctDNA
carcinoembryonic-antigen (CEA) has suboptimal sensitivity for recurrence. for detecting mutations. Methods:We retrospectively evaluated mCRC patients
This study compared performance of a quantitative circulating tumor DNA (pts) who had plasma-derived NGS utilizing a highly-sensitive targeted 68-73-
(ctDNA) assay for methylated BCAT1 and IKZF1 (COLVERA) with that of gene ctDNA assay. In a case-control design, pts with a known mutation on
CEA. Methods: 301 patients were monitored for recurrence after clearance tissue and radiologic evidence of metastatic disease but no detectable ctDNA
of primary CRC. Blood was collected at scheduled intervals and concen- mutation were matched 1:3 with randomly selected pts with detectable
trations of CEA and ctDNA were measured using the LIAISON CEA test mutations and compared according to clinical, laboratory, and radiologic
(Diasorin) and the COLVERA ctDNA test (Clinical Genomics). Surveillance characteristics. A prediction score for ctDNA detection was built using a binary
for recurrent disease was examined using regular CT scans. Sensitivity of logistic backward stepwise regression analysis and tested in two independent
each blood test for recurrence was assessed in the sample collected closest data sets from different institutions. Area under the curve (AUC) from receiver
to the time of imaging confirming recurrence status. Absence of recurrence operating characteristics curves (ROC) were used for internal and external
was defined as at least two consecutive clear CT scans. Receiver operator validation. Results: From 416 pts who met inclusion criteria, plasma-derived
characteristic (ROC) analyses were used to determine optimal positivity NGS did not find tumor mutations in 66 cases (15.9%); 198 pts with de-
threshold for Colvera. Results: 131 patients underwent satisfactory as- tectable alterations were selected as controls. After multivariate analysis, the
sessment for recurrence and had blood testing performed within 12 months detection of ctDNA was associated with increasing age (OR 1.05; 95%CI
of determining recurrence status (61.8% male, mean age 62.6 6.12.2(SD) 1.02-1.09; p = .001), presence of liver (OR 5.82; 95%CI 2.55-12.49; p ,
y). Of the 47 recurrence cases, 37 (74%) were distant. The areas under the .001) and lymph node metastases (OR 3.28; 95%CI 1.51-7.60; p = .004),
ROC curves were 0.7761 and 0.8188 for CEA and COLVERA, respectively archival TP53 mutations (OR = 2.88; 95%CI 1.37-6.17; p = .006). A key
(each p , 0.001). An optimal cut-off of 12.8pg/sample was determined for determinant was timing of collection relative to disease status: plasma col-
COLVERA and the standard 5ng/mL cut-off was selected for CEA. COLVERA lected in newly diagnosed metastatic disease or after evidence of progression
had a significantly higher sensitivity for detecting recurrence as compared to was substantially more likely to have detectable alterations (OR 9.24; 95%CI
CEA (68.1% vs 31.9%, p , 0.001) with a similar specificity (97.6% Vs 4.11-22.40; p , .001); The simplified prediction model performed well in
96.4%, p = 0.6547). A multivariate analysis determined COLVERA to be a internal (AUC = 0.88) and external validation (AUC = 0.95; 163 pts).
predictor of recurrence independent of CEA with positive COLVERA samples Conclusions: Our validated prediction model provides clinicians and re-
being 66.4 times (95%CI 14.0-315.8) more likely to have recurrence searchers with a tool to screen for patients in whom ctDNA testing can out-
confirmed within the study timeframe, whereas CEA was not a significant perform tissue-based testing in detecting genomic alterations.
predictor of recurrence (p = 0.228). Conclusions: These findings indicate
that COLVERA, reporting in quantitative mode, is a more sensitive test than
CEA. It provides a viable alternative for sensitive and early detection of
recurrent CRC. Clinical trial information: 12611000318987.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Colorectal) Cancer 213s

3591 Poster Session (Board #83), Mon, 8:00 AM-11:00 AM 3592 Poster Session (Board #84), Mon, 8:00 AM-11:00 AM
Preoperative platelet leukocyte ratio as an independent risk factor for Exploring the genetic basis of Lynch-like syndrome through paired germline
postoperative outcomes in colorectal cancers. First Author: Carrie E Ryan, and tumor exome sequencing. First Author: Jason Willis, The University of
University at Buffalo, Department of Surgery, Buffalo, NY Texas MD Anderson Cancer Center, Houston, TX
Background: Although preoperative platelet/leukocyte ratio (PLR) is a predictor of Background: Lynch-like syndrome (LLS) is characterized by a diagnosis of
postoperative outcomes in various neoplasms, data is lacking for colorectal cancer mismatch repair deficient (dMMR) malignancy where somatic bi-allelic mu-
(CRC). We hypothesized that elevated preoperative PLR would be an independent tations in canonical MMR pathway genes (MLH1, MSH2, MSH6, PMS2) have
risk factor for postoperative complications and increased 30-day mortality in patients been identified as the main cause. Yet, a substantial proportion of cases
with surgically resected CRC. Methods: Patients undergoing resections for CRC were remain unexplained by MMR somatic bi-allelic events or germline mutations.
identified from the American College of Surgeons National Surgical Quality Im- We hypothesize that LLS cases with young-onset cancers carry cryptic germline
provement Program (ACS NSQIP) from 2005 to 2016 dataset. Logistic regression alterations in other pathways. To explore this contribution, we performed
models for 30-day morbidity and mortality for PLR #28.4 and .28.4 (calculated
analyses of the germline and tumor mutation landscapes in LLS patients
from ROC analyses) were completed. Univariate log-rank test, and multivariate Cox
proportional hazards regression were used for time-to-event analyses and data are diagnosed with dMMR cancers. Methods: 18 probands with young-onset
presented as odds/hazard ratio with confidence interval. Results: 98,398 patients (age ,50 years) dMMR colorectal or uterine cancers were selected from a
were included in the study. Elevated PLR was an independent predictor of 30-day familial cancer registry. The absence of deleterious germline MMR mutation
morbidity and mortality. Patients with high PLR were more likely to have a wound and/or somatic MLH1inactivation was confirmed by standard clinical testing.
infection (1.022, 1.01-1.04), pneumonia (1.2, 1.17-1.23), require reintubation We performed whole-exome sequencing (Illumina HiSeq) of germline (pe-
(1.23, 1.19-127), prolonged ventilation (1.31, 1.26-1.36), have renal insufficiency ripheral blood) DNA. Variant calls, quality-control, allele-frequency filtering
(1.17, 1.11-1.22), and have a postoperative myocardial infarction (1.23, 1.9-127). (,1% in reference cohorts), and in silicoannotation were performed using the
PLR value . 28.4 was among independent predictors of mortality (1.67, GATK and polyphen/SIFT tools. Pathway analysis was performed using the
1.47–1.87), along with the ASA class, age, perioperative use of steroids, serum DAVID suite. For 16 of 18 patients, targeted exon sequencing of 408 cancer-
albumin and INR. (Table). Conclusions: Elevated PLR is a significant preoperative related genes was performed on paired tumor/normal tissue samples (Ion
risk factor for 30-day morbidity and mortality. PLR may be considered as a potential Torrent AmpliSeq) and analyzed with VarScan 2. Results: 237,055 rare
risk assessment tool that predicts postoperative outcome following CRC resections. germline variants were detected in our cohort. We enriched a subset of 758
Multivariate analysis. variants with putative frameshift (45.1%), stop gain or loss (25%), or splice
95.0% CI for site alteration (29.9%). Pathway analysis of genes altered by this subset
HR revealed excess events in DNA damage repair (e.g. ERCC5, POLM, POLN,
Regression Co-efficent p-value HR Lower Upper EXO5) and mRNA splicing (e.g. SCAF1, SRSF4) pathways. Preliminary
analysis of somatic mutations profiles shows frequent alteration of known
Age (years) 0.03 ,0.001 1.03 1.03 1.04
Partial Dependent Functional Status 0.49 ,0.001 1.63 1.39 1.91 drivers including APC(64%) and NOTCH1(36%). Conclusions: Our exploratory
Totally Dependent Functional Status 0.89 ,0.001 2.48 1.90 3.10 analysis provides novel evidence that LLS patients may harbor an excess of
Steroid use 0.51 ,0.001 1.68 1.36 2.05 deleterious germline mutations in DNA damage repair- and mRNA splicing-
Pre-operative serum albumin (g/dL) -0.92 ,0.0001 0.40 0.37 0.43
Pre-operative INR 0.32 ,0.001 1.38 1.26 1.52
related genes. Future studies will identify genes which are targeted by both
ASA class 3 1.12 0.115 3.06 0.76 12.31 germline and somatic mutation with the goal of nominating putative causal
ASA class 4 1.83 0.010 6.23 1.54 25.13 genes. Defining additional mechanisms of dMMR in LLS cancers may help to
ASA class 5 2.97 ,0.001 19.39 4.54 82.90 refine prevention strategies for (un)affected individuals.
PLR >28.4/PLR £ 28.4 0.51 ,0.001 1.66 1.47 1.87

3593 Poster Session (Board #85), Mon, 8:00 AM-11:00 AM 3594 Poster Session (Board #86), Mon, 8:00 AM-11:00 AM
Association of microRNA-21 with efficacy of cetuximab in RAS wild-type Comparison of HER2 overexpression with total Her2 mutation on resistance
patients in the FIRE-3 study (AIO KRK-0306) and microRNA-21’s influence of EGFR-targeted therapy in Ras wild-type mCRC patients. First Author:
on gene expression in the EGFR signaling pathway. First Author: Lisa Miller- Wenju Chang, Institute of General Surgery, Zhongshan Hospital, Fudan
Phillips, Department of Medicine III, University Hospital, LMU Munich, University, Shanghai, China
Munich, Germany
Background: Cetuximab has shown clinical benefit in patients with metastatic
Background: FIRE-3 compared first-line therapy with FOLFIRI plus cetux- colorectal cancer (mCRC) harboring wild-type Ras, however, only partial patients
imab (cet) or bevacizumab (bev) in KRAS exon 2 wild-type (wt) patients with respond to cetuximab treatment. The effect of human epidermal growth factor
metastatic colorectal cancer. Recent analyses showed mircoRNA-21 (miR- receptor 2 (HER2) protein overexpression and Her2 gene mutant on the efficacy of
21) expression level may be a predictive biomarker for anti-EGFR-therapy cetuximab treatment was not well elucidated in patients with Ras wild-type unre-
raising the question whether miR-21 influences gene expression in the EGFR sectable mCRC. Methods: From June 2008 to December 2014, we identified 216
patients with Ras wild-type unresectable liver-limited mCRC base on our previous
signaling pathway. Methods: Reverse-transcription quantitative polymerase
study (ClinicalTrials: NCT01564810.), whose Her2 gene mutation was analyzed by
chain reaction assay identified quantitative miR-21 expression. Median
next-generation sequencing for single nucleotide polymorphism (SNP) of Her2 gene
expression was defined as a threshold value to discriminate FIRE-3 pop- and HER2 protein overexpression was determined by immunohistochemistry and
ulation into miR-21 low and high groups. Differential gene expression based fluorescence in situ hybridization (FISH). Results: Of these 216 patients, 103 were
on additional mRNA microarray data (Almac Inc, Xcel Array) was calculated received cetuximab plus chemotherapy (cetuxima group) and 113 were received
by linear models adjusted for multiple testing followed by single sample gene chemotherapy alone (chemotherapy group). The total rate of HER2 overexpression
set enrichment analysis (ssGSEA) to compare differentially enriched hall- was 8.8%, including 9.7% in cetuximab group and 7.9% in chemotherapy group.
marks of cancer gene sets. Overall response rate (ORR) was compared HER2 overexpression caused impaired survival compared with HER2 non-
using Fisher´s exact test. Median progression-free (PFS) and overall sur- overexpression patients in cetuxima group, with a median progression-free sur-
vival (OS) were analyzed using Kaplan-Meier estimation and log-rank test. vival (PFS) of 4 months (95% CI 2.482-5.518) versus 10 months (95% CI 8.963-
Results: 333 RAS wt patients provided material for miR-21 expression 11.037; P, 0.0001), and a median overall survival (OS) of 15 months (95% CI 7.5-
analysis. In these patients, low miR-21 expression was associated with 22.2) versus 36 months (95% CI 31.4-40.5) (P, 0.0001). While, HER2 over-
higher ORR (80.0% vs. 57.9%; p = 0.005) and longer OS (35.8 months (mo) expression had no effect on treatment efficacy in chemotherapy group, when
vs. 25.9 mo; p = 0.005) when cet vs bev was added to FOLFIRI. High miR-21 compared with HER2 non-overexpression paitents, with a median PFS of 5 months
expression was associated with comparable ORR (74.6% vs. 64.0%; p = (95% CI 2.228-7.772) versus 5 months (95% CI 4.004-5.996; P= 0.615), and a
0.21) and OS (24.5 mo vs. 23.8 mo; p = 0.4). There was no significant median OS of 21 months (95% CI 6.975-35.025) versus 21 months (95% CI
difference in PFS in either group. By comparing miR-21 low and high groups 17.772-24.228; P= 0.629). Meanwhile, we observed 25.5% of total Her2 mutant
(24.2% in cetuximab group and 26.5% in cetuximab group), among of them 5
using normalized mRNA microarray data, 538 genes were found to be
patients are HER2 overexpression. Total Her2 mutation has no impact on survival
significantly differentially expressed in RAS wt patients after adjustment for compared with Her2 wild-type ones in neither cetuximab group nor chemotherapy
multiple testing. Including data from the two groups into ssGSEA yielded 23 group. In further bioinformatics analysis is underdoing, and which subgroup or type
hallmark of cancer gene sets that were significantly differentially enriched; of Her2 mutant potential affect cetuximab treatment need confirm. Conclusions: We
among them, KRAS-signaling showed higher enrichment in the miR-21 high show HER2 overexpression rather than total Her2 mutant contribute to resistance of
group (adjusted p = 2.09 E-13). Conclusions: MiR-21 expression level might cetuximab treatment in patients with mCRC harboring wild-type Ras. Next, the
be a predictive biomarker for anti-EGFR-therapy by modulating KRAS sig- subgroup mutant in total Her2 mutant needs further analysis to confirm their roles in
naling in FIRE-3 patients. survival after cetuximab treatment.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
214s Gastrointestinal (Colorectal) Cancer

3595 Poster Session (Board #87), Mon, 8:00 AM-11:00 AM 3596 Poster Session (Board #88), Mon, 8:00 AM-11:00 AM
Gene expression and genetic variants in Parkinson’s disease (PD) genes to Differences in pathology and mutational status among colorectal cancer (CRC)
predict outcome in metastatic colorectal cancer (mCRC): Data from FIRE-3 patients pre-, post-, and during screening age. First Author: Gustavo Dos Santos
phase III trial. First Author: Francesca Battaglin, Division of Medical On- Fernandes, Hospital Sirio-Libanês, Brasilia, Brazil
cology, USC Norris Comprehensive Cancer Center, Keck School of Medicine,
Background: Screening protocols for CRC are broadly recommended and effective in
Los Angeles, CA reducing mortality. However, populations from different age groups can harbor
Background: PD is one of the most common age-related neurodegenerative distinct pathological and molecular profiles that can also be influenced by screening
disorders. Large epidemiological studies have consistently reported a re- and polyp resection, especially in older ages. Methods: We retrospectively analyzed
duced risk of CRC in PD patients (pts), but the biology behind this evidence is tumors from stage IV CRC patients from a central pathology laboratory in Brazil that
unclear. The methylation status of SNCA, one of the causal PD genes, has is a reference for mutational profiling countrywide. Patients were classified into age
been identified as a tool for CRC screening and early diagnosis when de- groups as the following: pre-screening (PrSA; ,45yo), screening(SA; 45-75yo) and
post-screening age (PoSA; .75yo). Every tumor has been centrally reviewed by the
tected in stool samples, and alterations in core PD genes are prevalent across
pathologist. Groups were compared regarding clinicopathologic features and
human malignancies including CRC. Methods: The impact on outcome of 13 presence of RAS and BRAF mutations. Results: We included 1244 pts (164 PrSA,
SNPs within 6 core PD genes (SNCA, PRKN, UCHL1, PINK1, DJ-1, LRRK2) 919 SA and 161 PoSA). There were no difference among groups regarding side-
was analyzed in pts enrolled in the randomized FIRE-3 trial. Genomic DNA dness(p= .68)and KRASmutations (p=.0.97). Stage IV at diagnose (p =.001),
from blood samples of pts treated with first-line FOLFIRI-cetuximab (cet, n = presence of signet-ring cell component (p, .001) along with poorly differentiated
129) and FOLFIRI-bevacizumab (bev, n = 107) was genotyped through the tumors (p= .006) were most found on young patients, while BRAFand NRASmu-
OncoArray, a custom array manufactured by Illumina. Gene expression levels tations where significantly more common among PosSA (table). Conclusions: PosSA
were measured from 102 tumor samples of pts in the cet arm by HTG and PreSA CRCs seem to present a distinct profile from SA populations, including
EdgeSeq Oncology Biomarker Panel. Results: In the cet cohort, pts carrying differences in the molecular and pathologic aspects. This could impact the fre-
the G/G variants of SNCA rs356165 and rs2736990 had significantly quency of screening tests among different age groups.
shorter mOS (30 vs 41.1 mo) compared to any A genotype in both uni- and Age Group
multivariable analysis (adjusted P[Padj] = .047 and .042, respectively). <45y 45 - 75y >75 P-Value
LRKK2 rs3761863 T/T allele carriers showed shorter mPFS (9.5 vs 13.3
Histology - Signet ring
mo, Padj = .01), while rs11564148 any A carriers had longer mPFS (14.2 vs Signet-ring 26 (15.9%) 55 (6.0%) 9 (5.6%) ,0.001
10.2 mo, Padj = .01) compared to reference genotypes. LRKK2 rs11564148 Non-Signet ring 138 (84.1%) 864 (94.0%) 152 (94.4%)
Missing 0 0 0
any A carriers also showed longer mOS in multivariable analysis (43.7 vs Histologic grade, N (%)
33.2 mo, Padj = .044). Any C allele carriers of PINK1 rs1043424 showed Well
Moderately
22 (13.4%)
79 (48.2%)
144 (15.9%)
526 (58.1%)
29 (18.1%)
77 (48.1%)
0.006

longer mPFS in uni- and multivariable analysis (Padj , .001). No significant Poorly 63 (38.4%) 235 (26.0%) 54 (33.8%)
Missing 0 14 1
interaction was found with gender, tumor location and RAS status. These Stage at Diagnosis, N (%)
associations were not observed in bev arm. High SNCA expression was I 1 (0.6%) 14 (1.6%) 0 (0%) 0.001
II 1 (0.6%) 67 (7.6%) 10 (6.5%)
associated with worse mPFS (log2 . 7.89, 5.9 vs 11.2 mo) and mOS (log2 III 41 (25.3%) 208 (23.5%) 48 (31.4%)
. 7.68, 17.9 vs 31.1 mo) in FIRE-3 cet arm (P , .05). Conclusions: We IV 119 (73.5%) 595 (67.3%) 95 (62.1%)
Missing 2 35 8
provide the first evidence that gene expression and genetic variants in PD NRAS
genes may have a predictive value in mCRC pts receiving first-line Mutated 4 (2.5%) 25 (2.8%) 13 (8.2%) 0.006
Wild-type 154 (97.5%) 871 (97.2%) 146 (91.8%)
cetuximab-based treatment. Our findings open new perspectives on the Missing 6 23 2
role of PD genes in CRC biology warranting further investigation. BRAF
Mutated 7 (4.4%) 42 (4.7%) 18 (11.3%) 0.007
Wild-type 151 (95.6%) 854 (95.3%) 141 (88.7%)
Missing 6 23 2

3597 Poster Session (Board #89), Mon, 8:00 AM-11:00 AM 3598 Poster Session (Board #90), Mon, 8:00 AM-11:00 AM
The 55 STAR study: Prognostic and predictive value of the 55-gene classifier Urban/rural disparities in stage at presentation of colorectal cancer among
(55GC) in stage III colon cancer. First Author: Toshiaki Ishikawa, Tokyo young adults in the United States, 2007-2015. First Author: Victoria
Medical and Dental University, Tokyo, Japan Nguyen, Sidney Kimmel Cancer Center at Thomas Jefferson University,
Philadelphia, PA
Background: Patient prognosis can be predicted based on cancer subtypes
classified according to DNA microarray results. The most robust classification Background: The overall incidence rate of colorectal cancer (CRC) has de-
system involves the consensus molecular subtypes, which uses over 600 genes clined in recent decades but is rising in young adults (YA). Disparities also exist
for classification. To simplify this classification, we recently constructed a 55- in CRC presentation and geographical residence. We examined associations
gene classifier (55GC) to classify colon cancer (CC) into three subtypes with between urban/rural residence and CRC stage at presentation among YA 18-50
different recurrence rates: “microsatellite instability (MSI)-like,” “chromo- years old, using the Surveillance, Epidemiology, and End Results (SEER)
somal instability (CIN)-like,” and “stromal” colon cancers. The 55GC has been 1973-2015 registry. Methods: Retrospective cohort study using SEER pa-
reported to be a useful and reproducible grading system for stage II CC re- tients (pts) diagnosed with CRC between 2007-2015, aged 18-50. Urban/
currence risk stratification. This study aimed to explore the usefulness of 55GC rural status was defined at the county level as large metro, small metro, urban
for classifying stage III CC patients. Methods: We retrospectively identified non-metro, and rural. Pts were grouped by age: 18-30, 31-40, and 41-50 years
stage III CC patients aged 20-79 years who underwent curative surgery and old. Stage was defined as in-situ/localized and regional/distant. We used
received adjuvant chemotherapy with or without oxaliplatin (OX) between multivariable logistic regression to describe associations between urban/rural
2009 and 2012 from 15 institutions. Propensity score matching was used to status with stage at presentation, adjusting for tumor location, histology, grade,
adjust for the number of lymph node metastases, tumor location, sex, and age. and patient attributes (e.g. insurance status). Results: 27,198 CRC pts were
Results: Among 938 eligible patients, 203 and 201 cases involving adjuvant analyzed: 62.2% large metro, 27.4% small metro, 9.4% urban non-metro, and
chemotherapy with and without OX were selected, respectively, using pro- 67.1% regional/distant stage. In multivariable analysis, YA in urban non-metro
pensity score matching. Ninety-five cases each from groups were analyzed counties had lower odds of regional/distant stage at presentation compared to
after exclusion of cases involving low-quality specimens and those involving YA in large metro counties (OR = 0.87, 95% CI = 0.79-0.97). Associations
chemotherapy for , 3 months. The 5-year relapse-free survival (RFS) in between small metro and rural status with stage at presentation were not
patients receiving adjuvant chemotherapy with and without OX were 77.1% significant (OR = 0.94, 95% CI = 0.89-1.01 and OR = 0.84, 95% CI = 0.64-
and 73.7%, respectively. Classification of the stage III CC according to 55GC 1.10, respectively). YA with Medicaid or no insurance had higher odds of
and related 5-year RFS rates were as follows: stromal (N = 60), 66.6%; CIN- regional/distant CRC (OR = 1.28, 95% CI = 1.17-1.39 and OR = 1.34, 95%
like (N = 78), 80.5%; and MSI-like (N = 52), 78.4%. The HRs for 5-year RFS CI = 1.20-1.51, respectively) compared to YA privately insured. Other factors in
for adjuvant chemotherapy with and without OX in each subtype were as YA associated with higher odds of regional/distant CRC included signet-ring
follows: stromal, HR = 0.791 (95% CI = 0.329-1.901); CIN-like, HR = 1.241 histology, poorly-, moderately-, and undifferentiated grade, and younger age
(95% CI = 0.465-3.308); and MSI-like, HR = 0.495 (95% CI = 0.145- (18-30 and 31-40 years old). Conclusions: YA in urban non-metro areas had
1.692). Conclusions: The stromal subtype showed poor prognosis in stage III lower odds of regional/distant CRC at presentation compared to YA in large
as well as stage II patients. Oxaliplatin had a good additive effect in adjuvant metro areas. YA with Medicaid, no insurance, signet-ring histology, poorly-,
chemotherapy for MSI-like subtype. The 55GC is useful for predicting the moderately-, and undifferentiated grade, and younger age had higher odds of
efficacy of adjuvant chemotherapy for CC. regional/distant CRC at presentation. Further research is needed to explore
the etiology of these differences.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Colorectal) Cancer 215s

3599 Poster Session (Board #91), Mon, 8:00 AM-11:00 AM 3600 Poster Session (Board #92), Mon, 8:00 AM-11:00 AM
Methylated DNA markers (MDMs) in primary (pCRC) and metastatic Gene mutations of SWI/SNF complex and molecular profile in colorectal cancer.
colorectal cancers (mCRC). First Author: Hao Xie, Mayo Clinic, Rochester, MN First Author: Ryuma Tokunaga, Division of Medical Oncology, USC Norris
Comprehensive Cancer Center, Keck School of Medicine, Los Angeles, CA
Background: Existing tools for post-treatment CRC surveillance and moni-
toring are insensitive and expensive. MDMs are broadly informative for early Background: The SWI/SNF complex includes proteins produced by 29 genes that
detection of CRC but have not been extensively studied for disease moni- regulate chromatin structure remodeling through effects upon transcription, repli-
toring. As a first step, we sought to assess the concordance of novel CRC- cation, and repair. Understanding how SWI/SNF gene mutations interact to affect
associated MDMs in pCRC and mCRC. Methods: A panel of 14 MDMs cancer progression could lead to new treatment strategies. Methods: We analyzed
previously identified to be highly discriminant for pCRC was selected on the 7,370 colorectal cancer (CRC) samples with immunohistochemical stains (IHC) and
basis of high median fold-change of MDM levels relative to buffy coat (682 Next-Generation Sequencing (NGS) using a 592-gene panel to examine the asso-
(IQR: 132-19347)). Surgically resected pCRC and paired mCRC were ciation between gene mutations of the SWI/SNF complex (ARID1A, ARID2, PBRM1,
SMARCA4, SMARCB1, SMARCE1, BCL11A, BCL11B, BCL7A, SS18, and
identified from institutional pathology databases. Quantitative methylation-
SS18L1) and molecular features. Results: The overall mutation rate of the SWI/SNF
specific PCR was used to assay MDMs. 30 paired samples per metastatic complex genes was 11.3% (ARID1A: 7.7%, ARID2: 1.7%, SMARCA4: 1.3%,
subtype were calculated to be sufficient. MDM levels were compared using PBRM1: 1.2%, BCL11A: 1.0%, SMARCB1: 0.5%, BCL11B: 0.5%, SMARCE1:
two-sample and paired Wilcoxon rank sum tests. Results: 87 patients with 0.3%, SS18: 0.3%, BCL7A: 0.1%, SS18L1: 0.1%). When compared to tumors with
paired pCRC and mCRC including 57 synchronous and 30 metachronous SWI/SNF wild-type genes, those tumors with SWI/SNF gene mutations showed
metastases were included. 41/87 (47%) had neoadjuvant and 59/87 (68%) significantly higher rates of microsatellite instability (MSI)-high (40.9% vs 2.4%,
had adjuvant chemotherapy. All synchronous metastases were to liver. P , 0.001), tumor mutational burden (TMB)-high (.= 10mut/MB) (56.8% vs
Metachronous metastases were to liver in 19/30 (63%) and to lung in 11/30 21.6%, P , 0.001) and PD-L1 positivity (17.9% vs 5.5%, P , 0.001). Tumors with
(37%). The levels of 14 selected MDMs were remarkably similar between each gene mutant also had strong association with the immune profile (MSI-high,
paired pCRC and mCRC (Table). Individual MDM levels and the average level TMB-high, and PD-L1 positivity) (Table). Furthermore, even SWI/SNF gene mu-
of all MDMs combined were not significantly different (p. 0.0018 by tant samples with microsatellite stable status were significantly associated with
Bonferroni correction). Conclusions: MDM levels are highly concordant in TMB-high (28.2%, P , 0.001) and PD-L1 positivity (10.0%, P , 0.001).
pCRC and mCRC. Thus, MDMs discovered from pCRC should be further Conclusions: Gene mutations of the SWI/SNF complex exhibit findings that suggest
studied for non-invasive surveillance after surgical resection and monitoring that this subgroup of CRCs may have a higher likelihood of response to PD-1 and PD-
of treatment response in mCRC. L1 targeting monoclonal antibodies. If validated in other data sets, these findings
can be used to justify clinical trials with eligibility based upon the presence of
Median fold change (IQR) Median fold change (IQR) mutations within the SWI/SNF complex.
pCRC vs Syn- pCRC vs Syn- Association between gene mutations of SWI/SNF complex and immune profile.
chronous pCRC vs Meta- chronous pCRC vs Meta- Genes Mutation rate (%) MSI-H rate (%) P value TMB-H rate (%) P value PD-L1 positive rate (%) P value
MDM mCRC chronous mCRC MDM mCRC chronous mCRC SWI/SNF genes 11.3 40.9 <0.01 56.8 <0.01 17.9 <0.01
ARID1A 7.7 46.1 <0.01 60.3 <0.01 19.1 <0.01
VAV3 1.0 (0.2, 4.4) 1.2 (0.3, 4.1) ZNF625 1.0 (0.4, 2.8) 1.1 (0.3, 2.6) ARID2 1.7 37.2 <0.01 54.6 <0.01 19.5 <0.01
CHST2 0.9 (0.4, 2.5) 1.0 (0.2, 5.2) SFMBT2 1.0 (0.4, 2.2) 1.5 (1.0, 3.2) SMARCA4 1.3 64.6 <0.01 77.1 <0.01 27.5 <0.01
PBRM1 1.2 35.3 <0.01 64.7 <0.01 13.9 0.03
OPLAH 0.6 (0.3, 1.6) 1.1 (0.5, 2.2) LRRC4 0.7 (0.3, 2.0) 0.8 (0.2, 2.6) BCL11A 1.0 87.2 <0.01 95.8 <0.01 25.5 ,0.01
QKI 0.9 (0.5, 2.3) 1.1 (0.4, 1.8) DOCK10 0.9 (0.5, 2.2) 0.9 (0.4, 1.4) SMARCB1 0.5 51.4 <0.01 62.2 <0.01 13.9 0.15
PPP2R5C 1.2 (0.4, 4.1) 1.0 (0.3, 3.3) IKZF1 1.0 (0.3, 2.0) 1.0 (0.2, 3.1) BCL11B 0.5 75.0 <0.01 81.3 <0.01 15.6 0.09
SMARCE1 0.3 66.7 <0.01 83.3 <0.01 50.0 <0.01
ARHGEF4 0.8 (0.3, 1.9) 1.1 (0.5, 2.0) NDRG4 1.1 (0.3, 3.2) 1.0 (0.1, 3.7) SS18 0.3 33.3 <0.01 42.9 0.18 13.3 0.44
PDGFD 1.0 (0.4, 1.7) 1.0 (0.4, 1.7) BMP3 1.0 (0.3, 4.8) 0.7 (0.08, BCL7A 0.1 33.3 0.02 55.6 0.06 22.2 0.14
SS18L1 0.1 50.0 <0.01 50.0 0.20 0 0.35
1.1)
P value was based on Fisher’s exact test (vs. wild).

3601 Poster Session (Board #93), Mon, 8:00 AM-11:00 AM 3602 Poster Session (Board #94), Mon, 8:00 AM-11:00 AM
Prognostic stromal and immune response expression patterns in early-stage A plasma-only integrated genomic and epigenomic circulating tumor DNA
colorectal cancer to predict genes intrinsically expressed by tumor epithelial (ctDNA) assay to inform recurrence risk in colorectal cancer (CRC). First
cells. First Author: Pratyaksha Wirapati, Swiss Institute of Bioinformatics, Author: Aparna Raj Parikh, Massachusetts General Hospital, Boston, MA
Lausanne, Switzerland
Background: ctDNA identifies patients (pts) at high risk for disease re-
Background: High stromal content (CMS4) and immune infiltration (CMS1) currence post CRC resection (post-op). Current ctDNA residual disease
are important features of the tumor micro-environment (TME) in early stage detection approaches assess only genomic alterations (alts) and rely on
colorectal cancer, with poor and good RFS prognosis, respectively. However, tissue sequencing to identify tumor-derived alts. We evaluated a plasma-only
it remains unclear how these TME features are influenced by intrinsic tumor ctDNA assay to identify high risk pts. Methods: 72 CRC pts (surgery only =
epithelial expression profiles. We examined how the former can be predicted 42; adjuvant therapy (adj) = 30) had post-op and/or post-adj plasma samples
by the latter, and assessed their respective prognostic contributions. (3-4mL). Extracted cfDNA (median 27 ng) was analyzed using a single-
Methods: Eighteen public gene-expression datasets (n=3511) were used for sample NGS test validated in early stage CRC that integrates assessment of
discovery, and an adjuvant trial dataset (PETACC3 n=688) was used for genomic alts with epigenomic cancer signature (Guardant Health, CA). A
validation. Whole-transcriptome hierarchical clustering was applied to variant classifier was applied to differentiate tumor-derived from non-tumor
identify intrinsic versus extrinsic co-expression modules, corroborated by derived alts in a tumor tissue-uninformed approach. Results: In the surgery
single-cell profiling data. The epithelial gene subset were used to predict cohort, samples were collected a median of 31 days (d) post-op. 7/8 pts with
previously assigned CMS labels, producing new CRC classifier that we refer post-op ctDNA detected (ctDNA+) recurred (PPV 88%; median time to
to as epithelial-CMS (eCMS). Results: Despite lacking stromal and immune recurrence (mTTR) 248d). The recurrence-free pt has , 180d follow-up. 7/
response genes, eCMS predicted CMS with an accuracy of 82% [CI 78-85%] 34 pts without ctDNA detected (ctDNA-) recurred (NPV 79%; mTTR 333d).
in the test dataset. Not only were the RFS prognostic values retained (eCMS1 1/1 Stage 0-II ctDNA+ pt recurred (PPV 100%; TTR 440d) while 1/20
vs non-eCMS1: HR=0.59 [CI 0.35–0.95, p=0.029], and eCMS4 vs non-e- ctDNA- recurred (NPV 95%; TTR 440d). 27 pts in the adj cohort had
CMS4: HR=1.82 [CI 1.38–2.41, p=0.0003]), but they were also signifi- samples collected a median of 37d post-adj. 6/6 ctDNA+ pts recurred (PPV
cantly improved when compared to the original CMS classification (CMS1 vs 100%, mTTR 239d). 4/21 ctDNA- pts recurred (NPV 81%, mTTR 466d). 2/
non-CMS1: HR=0.76 [CI 0.49–1.16, p=0.2] and CMS4 vs non-CMS4: 2 ctDNA+ and 0/11 ctDNA- Stage III pts recurred (PPV, NPV 100%, mTTR
HR=1.43 [CI 1.07–1.91, p=0.017]. In a multivariable regression com- 420d). All 3 post-op ctDNA+/post-adj ctDNA+ (ctDNA persistence) pts
parison, eCMS was superior to CMS (p=0.001 vs p=0.36). Each eCMS group recurred. 1/2 post-op ctDNA+/post-adj ctDNA- (ctDNA clearance) pts is
was associated with specific combinations of up- and down-regulation of recurrence free (306d). 2 post-op ctDNA-/post-adj ctDNA+ pts recurred. In
CRC-relevant pathways, such as CDX2, map-kinase, mucin secretion, DNA the entire cohort, ctDNA+ after standard therapy completion had a re-
damage repair and cell proliferation. Conclusions: Stromal and immune currence PPV 93%, NPV 80%, HR 11.29 (p , 0.0001). Conclusions: In
infiltration patterns in the TME are strongly associated with tumor epithelial post-op CRC, ctDNA detection utilizing a tumor-uninformed integrated
expression profiles, as captured by eCMS. This finding sheds new light on the genomic and epigenomic assay has high recurrence PPV and NPV following
interplay between intrinsic molecular features of CRCs and their TME, and standard therapy completion. ctDNA identifies pts who may benefit from
illuminates new paths for potential combination therapies. post-op adj therapy or additional/modified post-adj therapy. These findings
demonstrate that ctDNA detection from a single post-op/post-adj plasma
sample stratifies high/low risk pts and informs therapy decision making.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
216s Gastrointestinal (Colorectal) Cancer

3603 Poster Session (Board #95), Mon, 8:00 AM-11:00 AM 3604 Poster Session (Board #96), Mon, 8:00 AM-11:00 AM
Gender differences in prognostic value of immune-related biomarkers Association of PIK3CA mutations (mut) with immune engagement and clinical
in colon cancer patients randomized to surgery or surgery and adjuvant benefit from immunotherapy in microsatellite stable (MSS) colorectal cancer
chemotherapy treatment. First Author: Lisa Villabona, Karolinska Institutet, (CRC) patients (pts). First Author: Maliha Nusrat, Department of Investigational
Stockholm, Sweden Cancer Therapeutics, The University of Texas MD Anderson Cancer Center,
Houston, TX
Background: HLA-A*02, a common allele in the Scandinavian population, is a
negative prognostic factor in epithelial ovarian cancer. It is a strong predictor of Background: PI3K pathway alterations, PIK3CA mut or PTEN loss, are known
patient outcome, only inferior to clinical staging. This prognostic trait in ep- to modulate anti-tumor immune microenvironment. MSS CRC pts respond
ithelial ovarian cancer is stronger by the presence of the gene compared with infrequently to immunotherapy, suggesting the presence of a rare MSS im-
the expression of its protein, MHC class I. Microsatellite instability (MSI) is munogenic subset. We investigated the immune repertoire and outcomes on
used as a biomarker for prognosis and is suggested an increased tumor mu- immunotherapy trials in PIK3CA mut MSS CRC pts. Methods: Immune in-
tational burden which can make the tumor more susceptible for T cell me- filtrates and checkpoints were evaluated using quantitative immuno-
diated immunotherapy. Our aim was to analyze the prognostic markers HLA- histochemistry (IHC) on primary CRC. Mutations were assessed by next
A*02 genotype, MHC class I on tumor cells, the CD8+ lymphocyte infiltration generation sequencing. PIK3CA mut neoepitopes and HLA allele affinities
and MSI status in colon cancer patients with randomized treatment. were predicted using NetMHC 4.0 Server. Outcomes of MSS CRC pts enrolled
Methods: Clinical information and primary tumors were collected from 520 in 7 immunotherapy trials were assessed. Clinical benefit (CB) was defined as
colon cancer patients and followed for overall survival for 120 months. Patients CR, PR or SD of 24 weeks. Time to progression (TTP) was calculated using
hade stage II and III colon cancer and were randomized to surgery alone or Kaplan-Meier Method. PIK3CA mut vs wild type (wt) groups were compared
surgery and adjuvant chemotherapy. HLA-A*02 genotype was determined by using Mann-Whitney U, Fisher’s exact, or Log-Rank tests as appropriate.
conventional PCR. MHC class I, MSI status and CD8+ lymphocyte infiltration Results: PI3K alterations were present in 14/40 MSS CRC pts in IHC cohort (7
were determined by immunohistochemistry. Results: Female patients with a PIK3CA mut, 33 wt; 7 PTEN loss, 33 intact). The center of PIK3CA mut MSS
stage III tumor and HLA-A*02 genotype had a better outcome if they had CRC had higher median densities of CD3+ cells [1112 (IQ range 865-1421) vs
received adjuvant chemotherapy instead of just surgery (p = 0.03), whereas 435 (300-744) cells/mm3; P=0.037] and CD8+ cells [554 (331-1200) vs
this was not the case for patients with other HLA-A genotypes or in the male 185 (60-473) cells/mm3; P=0.037] as compared to PIK3CA wt tumors.
patients where HLA-type did not correlate to outcome. MHC class I expression Intratumoral immune infiltrates did not differ by PTEN IHC staining in MSS
did not act as a prognostic factor, however the presence of CD8+ lymphocytes CRC. PD-L1 H-scores were also higher in PIK3CA mut MSS CRC [85 (34-114)
in the invasive margin and inside the tumor was a positive prognostic factor for vs 29 (11-60); P=0.01]. Several activating PIK3CA mut (E542K, E545K,
overall survival (p = 0.01), although only statistically significant in the male H1047R) were predicted to generate true neoepitopes with high binding af-
patients (p = 0.03). 21% patients had a tumor with MSI (23% of the female finity to common HLA types. Indeed, among MSS CRC pts enrolled in 7
and 19% of the male patients respectively). MSI tumors had a slightly better immunotherapy trials, half (4/8) of PIK3CA mut pts derived CB as compared to
outcome and this was irrespective of gender and HLA-type. Conclusions: The 3/35 (8.6%) PIK3CA wt pts (P=0.015). PIK3CA mut pts had trend towards
prognostic traits of HLA-A*02 appear in this colon cancer cohort to act dif- longer TTP (3.8 months in mut vs 2.1 months in wt; P=0.08). CB or TTP did not
ferently in male and female patients. Also CD8+ infiltration is different be- differ by colon sidedness, monotherapy / combination therapy, number of mut,
tween genders. These findings suggest that men and women may have two or mut in other key genes (APC, SMAD4, TP53, KRAS, NRAS or BRAF).
different immune responses to malignancy. Conclusions: PIK3CA mut MSS CRC are associated with increased cytotoxic
T cell infiltration, higher PD-L1 expression, and greater clinical benefit from
immunotherapy. Further investigation of immunotherapy outcomes in the
context of neoepitope-HLA allele interaction may help identify a subset of
PIK3CA mut MSS CRC pts who are likely to benefit from immunotherapy.

3605 Poster Session (Board #97), Mon, 8:00 AM-11:00 AM 3606 Poster Session (Board #98), Mon, 8:00 AM-11:00 AM
Differences in clinical outcomes in young-onset colorectal cancer based on Voltage: Investigator-initiated clinical trial of nivolumab monotherapy and
ethnicity in an NCI designated comprehensive cancer center. First Author: subsequent radical surgery following preoperative chemoradiotherapy in
Benjamin David Fangman, The University of Texas Southwestern, Dallas, TX patients with microsatellite stable locally advanced rectal cancer. First
Author: Takayuki Yoshino, National Cancer Center Hospital East, Kashiwa,
Background: Ethnic disparities can impact clinical outcomes of young-onset co-
lorectal cancer (CRC) patients. We aimed to determine if differences in outcomes
Japan
based on ethnicity exist in young–onset CRC treated at an NCI-designated com- Background: Chemoradiotherapy (CRT) with surgery (S) is standard for pa-
prehensive cancer center program. Methods: A retrospective chart review for stage tients (pts) with locally-advanced rectal cancer (LARC), and nivolumab (nivo) is
II – IV young-onset CRC patients #45 years old diagnosed between 04/2011 and active in microsatellite instability-high (MSI-H) metastatic colorectal cancer
11/2015. Patients had to undergo treatment at safety-net Parkland Hospital (PH) or at (mCRC). We studied nivo and radical S following CRT (50.4 Gy with cape-
the Simmons Comprehensive Cancer Center (SCCC) in Dallas, TX. Demographic citabine 1,650 mg/m2) in T3–4 NanyM0 LARC. Methods: Phase I included
data, dates of surgery, adjuvant chemotherapy, recurrence or death were obtained.
testing of a recommended phase II dosing schedule (RP2S). Efficacy and
Results: Of 123 patients that met inclusion criteria, 15 were excluded due to in-
complete information. Of the remaining 108 patients, 36 (33%) and 72 patients safety were studied in phase II pts and those given RP2S in phase I. In Cohort
(67%) were treated at SCCC and PH, respectively. Sixty (55%) were non-Hispanic vs A-1, for microsatellite stable (MSS) LARC pts, the primary endpoint was
48 (44.4%) Hispanic. There were more Stage IV patients at SCCC vs Parkland centrally confirmed pathologic complete response (pCR) rate using AJCC
(58.3% vs 30.6%, p , 0.01) but there was no difference regarding ethnicity. Also, tumor regression grading. The estimated required sample size assuming null
no significant difference was seen between non-Hispanic White (NHW), Hispanic, and alternative hypotheses pCR = 10% and 30% was 37 pts, with a 1-sided
and Black patients in median days to colectomy (1 vs 13 vs 0; p = .402) or adjuvant alpha of 5% and power of 90%. Cohort A-2 was exploratory and included a
chemotherapy (55.5 vs 53.0 vs 64.0 days, p = .820). Hispanic patients had sig- maximum of 5 MSI-H pts. Results: Nivo 240 mg q2 weeks x 5 cycles, following
nificantly better overall survival (OS) than Black or NHW patients (p = 0.025). The CRT but pre-S, was the RP2S. From 1/17 to 6/18, 37 pts were enrolled in
OS benefit was driven by improved 5-year OS in stage II/III Hispanic vs NHW vs Black Cohort A-1. Eleven pts (30%; 90% CI 18-44%) showed pCR (AJCC grade (gr)
patients (95% vs 62% vs 60%; p = 0.06). Multivariate Cox Regression analysis 0). Including the 3 pts (8%) graded AJCC 1, 14 (38%) had major pathologic
showed stage II/III (p , 0.001) and Hispanic ethnicity (p , 0.001) were in- responses. In addition, clinical CR was observed in 1 pt (3%) refusing S after
dependently associated with improved outcomes. Conclusions: In young-onset CRC nivo. Both MSI-H LARC Cohort A-2 pts showed pCRs. Immune-related severe
treated at an NCI-designated comprehensive cancer center, Hispanic ethnicity had adverse events were observed in 2 pts (gr 3 myasthenia and gr 2 interstitial
better OS than other ethnicities and this was largely due to better outcomes in stage nephritis); both fully recovered and had S. No treatment-related deaths were
II and III CRC. The causes for these ethnic differences in young-onset CRC patients observed. pCR rates of 60% (6/10) and 19% (5/27) (p = 0.038, Fisher exact
needs further exploration.
test) were seen in pts with tumor cells with PD-L1 $1% and , 1% IHC
Non-Hispanic White Hispanic Black P value staining, respectively, performed on biopsy samples taken pre-CRT. Rates of
N 41 (38.3%) 49 (45.8%) 18 (15.9%) 62% (8/13) and 10% (1/10) (p = 0.029) were seen in 23 pts with samples
Gender (Female)
Age (SD)
13 (31.7%)
39.51 (5.60)
22 (44.9%) 8 (47.1%) 0.369
38.14 (5.40) 38.71 (5.41) 0.268 analyzable by flow cytometry, according to CD8+ lymphocyte /regulatory T cell
Stage (CD8/Treg) ratios $2 and , 2, respectively. Conclusions: A promising pCR rate
II 7 (17%) 12 (25%) 4 (20%) 0.427
III 13 (32%) 20 (43%) 9 (45%) of 30%, with mild toxicity, was shown in MSS LARC pts treated with nivo plus
IV
Side of Cancer
21 (51%) 15 (32%) 7 (35%)
radical S. PD-L1 expression and elevated CD8/Treg ratio may be better pre-
Left 31 (79.5%) 29 (61.7%) 14 (73.7%) 0.187 dictors of nivo benefit, warranting further study in a larger cohort. Clinical trial
Right 8 (20.5%) 18 (38.3%) 5 (26.3%)
Median Days to Colectomy (25-75 Quartile) 1 (0-27) 13 (1-46) 0 (0-13) 0.402 information: NCT02948348.
Median Days to Adjuvant Chemotherapy (25-75 Quartile) 55.5 (36-73) 53 (45-77) 64 (47-74) 0.820
Median OS (months) 44 (17-71) Not Reached Not Reached 0.025

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Colorectal) Cancer 217s

3607 Poster Session (Board #99), Mon, 8:00 AM-11:00 AM 3608 Poster Session (Board #100), Mon, 8:00 AM-11:00 AM
The genomic landscape and its prognostic significance for stage III colorectal Effect of neoadjuvant treatment on locally advanced rectal mucinous adeno-
cancer in Japan: An ancillary study of JCOG0910 trial—JCOG1506A1. First carcinoma or signet-ring cell carcinoma: A post-hoc analysis from 3 prospective
Author: Dai Shida, Deaprtment of Colorectal Surgery, National Cancer Center studies. First Author: Jianwei Zhang, The Sixth Affiliated Hospital of Sun Yat-sen
Hospital, Tokyo, Japan University, Guangzhou, China
Background: To date, large-scale genomic sequencings of colorectal cancers Background: Mucinous adenocarcinoma and signet-ring cell carcinoma were un-
(CRC) have been reported mainly from Western countries. However, ethnic common in locally advanced rectal cancer. And it has been reported that both
diversities, differences by stage, and the prognostic impact of the genomic mucinous adenocarcinoma and signet-ring cell carcinoma showed poor response to
landscape in CRC remain poorly identified. Methods: The subjects were 534 standard neoadjuvant chemoradiotherapy. Here, we tried to compare the efficacy of
patients (pts) with stage III CRC from the JCOG0910 study—a randomized different neoadjuvant treatment regimen on locally advanced rectal mucinous
phase-III trial conducted in Japan on 1564 pts to assess the efficacy of S-1 adenocarcinoma or signet-ring cell carcinoma (M/S). Methods: We enrolled pa-
versus capecitabine as adjuvant chemotherapy. Targeted-capture sequencing tients with locally advanced rectal cancer from 3 prospective clinical trials
(NCT01211210, NCT02217020 and NCT02887313), including FOWARC study
of 171 potentially CRC-associated genes was performed on both normal tissue
(N = 309), mFOLFOXIRI neoadjuvant chemotherapy trial (N = 103) and the total
and tumor samples, and somatic single-nucleotide variants and insertion/ neoadjuvant treatment with FOLFOX and radiotherapy (N = 129). Among the 541
deletions were determined. Tumors with MSIsensor scores . 7 and ultra- patients, 41 (7.6%) patients were M/S and 500 were non-M/S. Totally, 7 M/S
mutated tumors with POLE mutations were grouped as hypermutated tumors. patients and 84 non-M/S patients received 5FU concurrent with radiotherapy (Group
Genes whose alterations were associated with recurrence-free survival (RFS) A), 20 M/S patients and 208 non-M/S underwent FOLFOX concurrent with radiation
were evaluated using multivariable Cox regression models. Results: Of the 534 or total neoadjuvant treatment (Group B), 11 M/S patients and 92 non-M/S patients
pts (right-sided: 184, left-sided: 350), 109 pts had recurrences or died during underwent mFOLFOXIRI neoadjuvant chemotherapy alone (Group C). Other 3 M/S
the study. Mutation frequencies were as follows: TP53, 75.3%; APC, 75.1%; patients and 116 non-M/S patients received FOLFOX neoadjuvant chemotherapy
KRAS, 43.6%; PIK3CA, 19.7%; FBXW7, 18.5%; SOX9, 11.8%; COL6A3, alone (Group D). Results: Among M/S patients, only 4 (9.7%) achieved pathologic
8.2%; NOTCH3, 4.5%; NRAS, 4.1%; and RNF43, 3.7%. Thirty-one tumors complete response (pCR), and 6 (14.6%) patients had tumor downstaging to ypstage
were hypermutated (5.8%) (right: 14.1%, left: 1.4%). None of the 49 genes 0-I. In group A, the pCR rate was 14.3% and 11.9% (p = 0.85), and the tumor
with mutation frequencies . 3% showed a significant association with RFS downstaging rate was 14.3% and 36.9% (p = 0.22) in M/S and non-M/S patients,
based on Bonferroni’s adjustment for multiple testing. The following modest respectively. In group B, the pCR rate was 15.0% and 34.6% (p = 0.07), and the
associations were observed: mutant KRAS [HR, 1.66; p=0.011] and mutant tumor downstaging rate was 25.0% and 60.1% (p = 0.002) in M/S and non-M/S
RNF43 [HR, 2.17; p=0.055] had poorer RFS, whereas mutant COL6A3 [HR, patients, respectively. However, in group C and group D with chemotherapy alone as
0.35; p=0.040] and mutant NOTCH3 [HR, 0.18; p=0.093] had better RFS. neoadjuvant treatment, no M/S patients showed pCR or tumor downstaging; while in
non-M/S patients higher tumor downstaging rate was observed. Conclusions: M/S
RFS tended to be better for hypermutated than for non-hypermutated tumors
showed resistance to neoadjuvant chemotherapy along regimens. Even with che-
[HR, 0.53; p=0.229]. Conclusions: The overall mutation spectrum of our stage moradiotherapy, M/S patients showed poorer response than that of non/M/S patients.
III CRC cohort was generally similar to that of the Cancer Genome Atlas (TCGA). Further study was warranted to explore the new regimen for M/S patients.
However, the mutation frequencies of TP53, SOX9, and FBXW7 were higher,
and the proportion of hypermutated tumors was lower. Multiple gene mutations Regimens Histology N pCR P ypStage 0-I P

seemed to impact RFS, indicating that tumor genomic profiling has a high 5FU+RT M/S 7 1 (14.3%) 0.85 1 (14.3%) 0.22
Non-M/S 84 10 (11.9%) 31 (36.9%)
potential to support precision medicine for pts with CRC. FOLFOX+RT M/S 20 3 (15.0%) 0.07 5 (25.0%) 0.002
Non-M/S 208 72 (34.6%) 125 (60.1%)
mFOLFOXIRI M/S 11 0 0.07 0 0.0001
Non-M/S 92 21 (22.8%) 44 (47.8%)
FOLFOX M/S 3 0 0.64 0 0.17
Non-M/S 116 8 (6.9%) 45 (38.8%)

3609 Poster Session (Board #101), Mon, 8:00 AM-11:00 AM 3610 Poster Session (Board #102), Mon, 8:00 AM-11:00 AM
Stage-based variation in the impact of colon cancer surveillance. First Preoperative chemosensitivity testing as predictor of treatment benefit in
Author: Lucy Gately, Walter and Eliza Hall Institute, Melbourne, Australia adjuvant stage III colon cancer: Preliminary analysis of the PePiTA study.
First Author: Alain Hendlisz, Medical Oncology Department, Institut Jules
Background: Multiple meta-analyses have demonstrated that routine sur-
Bordet - Université Libre de Bruxelles (ULB), Brussels, Belgium
veillance following curative intent colorectal cancer surgery improves overall
survival. This benefit is largely driven by early detection and curative intent Background: Although being the standard-of-care for stage III colon cancer,
resection of oligometastatic disease. Intuitively, any surveillance benefit no biomarkers can identify patients (pts) who benefit from adjuvant che-
should be proportional to recurrence risk, leading some to question the value of motherapy. In metastatic pts, the lack of metabolic response (mR) in FDG-
surveillance for stage I patients where recurrence rates are low. However, the PET/CT after 1 chemotherapy cycle predicts the absence of treatment
survival benefit of surveillance has not previously been reported by stage. benefit. The PePiTA study aims to evaluate if the absence of mR after 1 cycle
Methods: We explored data from a multi-site cohort of colorectal cancer pa- of pre-operative chemotherapy is predictive of recurrence in non-metastatic
tients (pts) diagnosed from 1 January 2001 to 31 December 2016. Pts were colon cancer pts. Herein, we report a preliminary analysis on surgical out-
followed according to standard protocols with a standardized comprehensive comes, adverse events and mR assessment after 1 cycle of pre-operative
outcome data captured prospectively. Pts with a rectal primary or metastatic chemotherapy after completing the study accrual objective. Methods: Colon
disease at presentation were excluded from the analysis. We examined the cancer pts eligible for curative resection and ECOG #1 received 1 cycle of
correlation of stage at diagnosis with tumor recurrence and subsequent out- mFOLFOX followed by surgery in this prospective, multicentre, non-
comes. Results: Of 3608 colon cancer pts, 690 (19%) had stage 1, 1580 randomized trial. FDG-PET/CT was performed at baseline and after 1 cy-
(44%) had stage 2, and 1338 (37%) had stage 3 disease. Median follow-up cle of mFOLFOX. Adjuvant mFOLFOX was administered for up to 12 cycles
was 7.8 years. Stage at diagnosis impacted recurrence rate (4% stage I vs 12% for stage III pts, whereas for stage II pts the decision to pursue adjuvant
stage II vs 28% stage III, p , .0001) but not median time to recurrence. chemotherapy was at investigator’s discretion. A decrease $15% in SUVmax
Recurrence patterns varied with stage (e.g. distant nodal disease 5% vs 7% vs after 1 cycle of chemotherapy was defined as mR (EORTC criteria) at central
16%, p = .003; liver metastases 90% vs 53% vs 42%, p = 0.001). In pts with review. Results: mR was assessable on the primary tumor in 204/240 pts
recurrence, resection of oligometastatic disease varied significantly by stage (85%). In 11 pts (4.6%), staging was modified by the baseline FDG-PET/CT,
(58% vs 42% vs 30%, p , .0001) as did post-resection 5 year survival (91% which detected metastatic disease or other tumors. Pre-operative mFOLFOX
vs 66% vs 43%, p , 0.001). In pts with recurrence treated with palliative was administered to 218 pts, of which 14 (6%) had a grade $3 adverse
intent, stage at diagnosis also impacted post-recurrence 5 year survival (11% event. Surgery was performed in 218 pts, with a median delay of 20 days (6
vs 7% vs 5%, p , 0.03). Conclusions: Colon cancer stage at diagnosis to 59) after chemotherapy. Surgical complications occurred in 28 (13%) pts,
substantially impacts the proportion of pts able to undergo curative intent however no deaths occurred. The median SUVmax decrease between
surgery for surveillance detected recurrent disease, potentially driven by stage baseline and 2nd FDG-PET/CT was 24%. A mR was observed in 65.2% of the
specific metastatic patterns. Stage at diagnosis also has a significant impact pts, whereas 34.8% showed no mR, including 3% who had progressive
on post-resection survival outcomes potentially driven by stage specific bi- disease. Conclusions: One cycle of pre-operative mFOLFOX followed by a mR
ology. Our data indicate a far greater survival impact of surveillance for stage I assessment has shown to be a feasible and safe strategy, raising interest on
colon cancer than would be anticipated based on recurrence rate alone. the potential of neoadjuvant chemotherapy in colon cancer. The early mR
assessment identified pts that may not benefit from chemotherapy and might
have a worse prognosis. The substantiation of this hypothesis is expected
with the study’s long-term results. Clinical trial information: NCT00994864.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
218s Gastrointestinal (Colorectal) Cancer

3611 Poster Session (Board #103), Mon, 8:00 AM-11:00 AM 3612 Poster Session (Board #104), Mon, 8:00 AM-11:00 AM
Long-term outcomes after high-dose chemoradiotherapy for non-surgical Simultaneous versus staged resection for synchronous colorectal cancer liver
management of distal rectal cancer. First Author: Edina Dizdarevic, Vejle metastases: A population-based cohort study. First Author: Pablo Emilio
Hospital, Vejle, Denmark Serrano Aybar, McMaster University, Hamilton, ON, Canada
Background: Surgery is standard treatment for rectal cancer, but neoadjuvant Background: Simultaneous resection of colorectal cancer primary and liver
chemoradiotherapy (CRT) may result in clinical complete response (cCR) in metastases is not performed routinely due to concerns about safety. We hy-
selected patients, allowing for non-surgical management (NSM). Prospective pothesized that simultaneous resection has steadily increased overtime and
studies of NSM strategies are sparse however, and long-term data on quality of that the outcomes are similar. Methods: Population-based cohort study of
life (QoL) are limited. We conducted a single-arm phase II trial of high-dose patients undergoing resection for synchronous (resection of the primary co-
CRT for NSM of distal rectal cancer; we report secondary long-term patient- lorectal cancer and liver metastases within six months) liver metastases from
reported outcomes (PROs), local regrowth and overall survival (OS) in patients 2006-2015 by linking administrative datasets in Ontario, Canada. Outcomes:
managed non-surgically. Methods: Fifty-one patients with resectable, T2 or post-operative complications, length of hospital stay, and overall survival.
T3, N0–N1, low adenocarcinoma received 65Gy (IMRT, brachytherapy boost) Survival for the staged group was measured from the last surgical resection to
and oral tegafur-uracil. Patients with cCR 6 weeks after treatment (clinical death and estimated using Kaplan Meier and compared with the log-rank test.
examination, MRI, biopsy) were referred for observation, and followed closely Cox proportional hazard models were used to calculate risks for death. We
with clinical examinations, endoscopies, PET-CTs, and PROs for 5 years. aimed to identify practice patterns, outcomes of simultaneous vs. staged
Overall colorectal cancer specific QoL and specific symptom scores were resections for these patients. Results: Of 2,738 patients undergoing colorectal
compared between timepoints using paired Wilcoxon tests. Local regrowth was and liver resection for colorectal cancer, 1,168 were synchronous, of which,
estimated using cumulative incidence; overall survival using Kaplan-Meier 442 underwent simultaneous resection. Rate of synchronous disease pre-
estimates. Results: Forty patients achieved cCR after treatment; 28 were in sentation increased on average by 3% per year (p = 0.02). Median length of
follow-up at 24m, 21 at 36m, 18 at 60m. Patients left the trial due to local stay was shorter (8 vs. 11 days, p , 0.001); rate of major liver resections were
tumor regrowth (n=12), distant metastases (n=3), new primary cancers (n=6) lower (17% vs. 65%, p , 0.001), and 90-day post-operative mortality was
and loss to follow-up (n=1). Average QoL score did not differ between baseline higher (6% vs. 1%) for simultaneous resections. Major postoperative com-
(median 11.1) and 24m (13.7), 48m (11.1,) or 60m (6.9). See Table for plications were higher in the simultaneous group (28% vs. 23%, p = 0.067),
individual scores; only rectal bleeding deteriorated from baseline (significantly mostly due to a higher reoperation rate (6% vs. 3%, p = 0.034). Median overall
worse at 24m). At median follow-up of 5.0 years, local regrowth rate and OS survival was worse with simultaneous resection (40 months, 95%CI 35-46 vs.
were 31% (95 CI 15%-47%) and 85% (95 CI 75%-97%), respectively. 78 months, 95%CI 59-86). Risks factors for worse survival were comorbidities,
Conclusions: Long term follow-up after NSM of early rectal cancer showed rurality, right-sided primary and simultaneous resection. There is selection bias
excellent general colorectal cancer QoL and local symptom scores. that favours survival in the staged group, as patients must have survived the
(NCT00952926). EORTC QLQ – CR 29. Proportion reporting ‘quite a bit’ or first operation and have stable disease in order to undergo the second oper-
‘very much’ on symptom scales. Clinical trial information: NCT00952926. ation. Conclusions: Simultaneous resection is associated with worse post-
Baseline (n=40) 24m (n=28) 48m (n=22) 60m (n=18)
operative outcomes. Considering selection bias, randomized studies would be
necessary to determine the role of simultaneous.
Daytime urinary frequency 27% 12% 18% 11%
Urinary incontinence 0% 0% 9% 0%
Rectal pain 0% 4% 4% 0%
Blood in stools 8% 27% 23% 11%
Faecal incontinence 3% 7% 0% 0%
Daytime bowel movement frequency 14% 17% 17% 12%

3613 Poster Session (Board #105), Mon, 8:00 AM-11:00 AM 3614 Poster Session (Board #106), Mon, 8:00 AM-11:00 AM
Pathway analysis of hypoxia-related factors in early colorectal cancer patients Clinical features and survival among patients with standard-onset versus
with poor prognosis. First Author: Yingxin Tan, The Sixth Affiliated Hospital early-onset colorectal cancer by age groups. First Author: Ana Acuna Villaorduna,
of Sun Yat-sen University, Guangzhou, China Montefiore Medical Center, Albert Einstein College of Medicine, Bronx, NY
Background: Colorectal cancer is one of the most common malignancies Background: Colorectal cancer (CRC) incidence is increasing in patients
with a high mortality rate. Patients with stage I and stage II colorectal cancer younger than 50 years old. Currently, there are discordant recommendations
have limited options for treatment. Hypoxia affects the activation and regarding CRC screening: while the American Cancer Society favors to start
regulation of colorectal cancer cells and participates in its invasion and at age 45, the National Comprehensive Cancer Network and the US Pre-
migration. However, there is lack of an accurate and non-invasive method for ventive Task Force suggest starting at age 50. This study is aimed to compare
assessing tumor hypoxia. The aim of this study was developing and the incidence, clinical characteristics and survival of patients diagnosed
validating a hypoxia gene signature for predicting the outcome in stage I/II with standard-onset CRC (SO) versus early-onset colorectal cancer by age-
colorectal cancer patients. At the same time , we hypothesized that analysis groups. Methods: Patients diagnosed with CRC at ages older than 35 were
of database of CIT microarray dataset could identify important biomarkers for identified using the SEER registry and categorized into four groups based on
stage I/II colorectal cancer patients. Methods: A total of 309 colorectal age at diagnosis. EO1 (35-39), EO2 (40-44), EO3 (45-49) and SO (.50)
cancer patients of early stage with complete clinical information were en- years, respectively. Incidence, clinical features and survival were compared
rolled for construction generation of hypoxia-related gene signature (HRGS) among groups. Results: 178 678 patients were identified. 9.2% were di-
based on the CIT microarray dataset. 1877 colorectal cancer patients with agnosed before 50 years. Of these, 1.4%, 2.8% and 5.1% were EO1, EO2
complete prognostic information in 5 independent datasets were divided and EO3; respectively. Patients with early-onset CRC (EO) had higher fre-
into a training cohort and two validation cohort (TCGA and meta-validation). quency of Hispanics (13.9% vs. 8.4%, p,0.01), stage IV (24.8% vs.
Prognostic analysis was assessed in these cohort to evaluate the predictive 17.3%, p,0.01), left-sided tumors (74.1% vs. 56.9%, p,0.01) and better
value of HRGS. Results: A model of prognostic HRGS containing 14 hypoxia- survival compared to SO. Among EO groups, the frequency of poor/anaplastic
related genes was developed. In training cohort and two validation cohorts, grade was inversely proportional to age; stage IV was similar between EO2
patients in hypoxia high-risk group satisfied by our HRGS had significant and EO3 and lower in EO1. Black race, grade and stage were predictors of
poor disease free survival compared with those in the in the low risk group mortality for all EO groups; laterality was a mortality predictor in EO2 and
(HR=4.35, 95% CI=2.30-8.23, P,0.001 in training cohort, HR=2.14, EO3. Conclusions: EO-CRC and SO-CRC have different pathological features
95% CI=1.09-4.21, P=0.024 in TCGA cohort, HR=1.91, 95% CI=1.08- that should be considered for CRC screening. Higher rates of stage IV disease
3.39, P=0.024 in meta-validation cohort). When compared with Oncotype are encountered in patients between 40-49 years old; hence early screening
DX, HRGS achieved an improved survival correlation in the training cohort should be considered. Given higher rates of left-sided tumors, sigmoidos-
(mean C-index, 0.80 vs 0.65, P,0.05) and the validation cohort (mean C- copy might be an adequate tool for most patients with EO-CRC.
index, 0.70 vs 0.61 in the TCGA cohort, mean C-index, 0.68 vs 0.73 in the EO1 EO2 EO3 SO p*
meta-validation cohort). Analysis of the data found that patients with low
Incidence (per 100 000) 8.8 16.5 30.1 146.8
survival rates have significant relationships with genes regulated by the cell Male (%) 50.2\ 51.4 53.9 51.8 ,0.01
cycle pathway, such as mTROC1, E2F, G2-M, mitotic, oxidative phos- NHW (%) 1.1 2.3 4.5 92.1 ,0.01
NHB (%) 1.6 3.7 6.4 88.3 ,0.01
phorylation, MYC, PI3K-AKT-mTOR (P,0.005). Conclusions: HRGS was a Hispanic (%) 2.7 4.3 7.4 85.5 ,0.01
satisfactory prognostic prediction model for early stage colorectal patients. Poorly/Anaplastic grade (%) 23.5 20 19 18.7 ,0.01
Hypoxia-related genes that regulate the cell cycle pathway were associated Sigmoid to anal (%) 65.2 66 67.1 49.9 0.14
Left-sided (%) 74.1 73.4 74.4 56.9 0.32
with prognosis in patients with stage I and stage II colorectal cancer. Further Stage IV (%) 17.3 26.6 24.7 24.3 ,0.01
researches are needed to assess the clinical effectiveness of the system and 5-year OS 64.5 66.6 65.9 54.9
the treatment options for biological targets. p* comparison between EO groups

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Colorectal) Cancer 219s

TPS3615 Poster Session (Board #107a), Mon, 8:00 AM-11:00 AM TPS3616 Poster Session (Board #107b), Mon, 8:00 AM-11:00 AM
POLEM: Avelumab plus fluoropyrimidine-based chemotherapy as adjuvant The Global POLAR program: Two pivotal placebo-controlled studies of calm-
treatment for stage III dMMR or POLE exonuclease domain mutant colon angafodipir used on top of modified FOLFOX6 to prevent chemotherapy-
cancer—A phase III randomized study. First Author: David Lau, Royal induced peripheral neuropathy (CIPN). First Author: Maryam B. Lustberg,
Marsden NHS Foundation Trust, Sutton, United Kingdom Ohio State University Comprehensive Cancer Center, Division of Medical On-
cology, Columbus, OH
Background: Colon cancer with deficient mismatch repair (dMMR) and
POLE mutations are characterised by a high tumor mutational burden (TMB) Background: Oxaliplatin (OXA), is approved in combination with 5-FU/FA (5-
and an immunogenic lymphocyte infiltrate. Approximately 12% of stage III fluorouracil/folinic acid; FOLFOX) for metastatic as well as in adjuvant colo-
colon cancer have dMMR. POLE mutations occur in 1% of colon cancers, but rectal cancer (CRC) treatment. CIPN is a common adverse event, after OXA
is enriched in patients ,50 years of age. Colon cancer with high TMB have treatment. The incidence of chronic CIPN is approximately 15% after a
demonstrated to be sensitive to immune checkpoint inhibition in the cumulative dose of 780 to 850 mg/m² and 50% after a cumulative dose of
metastatic setting. We are conducting a phase III randomised trial to de- 1170 mg/m². OXA induced neuropathy, results in greatly reduced nitrated
termine if the addition of the anti-PD-L1 antibody, avelumab following manganese superoxide dismutase (MnSOD) activity. Treatment with a su-
adjuvant chemotherapy can improve disease free survival (DFS) in patients peroxide dismutase mimetic, such as calmangafodipir (CAL), prevents and
with stage III colon cancer with dMMR or POLE mutations. Methods: We are reverses oxaliplatin-induced neuropathies. This has been demonstrated in
recruiting patients with curatively resected, stage III colon cancer which are the PLIANT study, with CAL (Glimelius et al. 2017). Methods: The POLAR
dMMR or have a centrally confirmed POLE exonuclease domain mutation. program is a Phase 3, multicenter, placebo (PLC)-controlled program of CAL
Eligible patients are randomised in a 1:1 ratio to standard fluoropyrimidine- to prevent CIPN, recruiting in US, Europe (B, D, ES, F,ES, I and UK) and Asia
based chemotherapy (CAPOX [capecitabine, oxaliplatin] for 12 weeks or (J, SK, TW and HK) and is described below; POLAR A Patients with CRC,
capecitabine for 24 weeks) or chemotherapy followed by avelumab (10mg/ treatment of patients with Stage III or high-risk Stage II who are indicated for
kg, 2 weekly for 24 weeks). Stratification is by chemotherapy received and adjuvant modified FOLFOX6 (mFOLFOX6) chemotherapy for up to 6 months,
MMR status. The primary endpoint is DFS. Secondary endpoints include randomized in a 1:1 ratio, each arm n = 140: A: CAL (5 mmol/kg) +
overall survival, toxicity, quality of life, and health resource use. Exploratory mFOLFOX6 chemotherapy B: PLC + mFOLFOX6 chemotherapy POLAR M
objectives will investigate circulating, tumor and stool based biomarkers of Patients with metastatic colorectal cancer (mCRC), who are indicated for
avelumab benefit. The 3-year DFS rate in the control arm is expected to be first-line mFOLFOX6 chemotherapy for at least 3 months, without any pre-
~75%. Avelumab is expected to improve the 3-year DFS rate by 12% (i.e. planned treatment breaks and will be randomized in a 1:1:1 ratio, each arm n =
87%). Target accrual is 402 patients which provides 80% power to detect a 140: A: CAL (2 mmol/kg) + mFOLFOX6 chemotherapy B: CAL (5 mmol/kg) +
hazard ratio of 0.48 for DFS at a two–sided alpha of 0.05. This trial is a mFOLFOX6 chemotherapy C: PLC + mFOLFOX6 chemotherapy Primary
national, multi-centre phase III trial and it is anticipated that approximately objective is to compare CAL vs PLC with respect to the proportion of patients
40 centres in the UK will participate. This study opened to recruitment in with moderate or severe chronic CIPN. The primary endpoint is; Patient
August 2018. Clinical trial information: NCT03827044. reported symptoms as proportion of patients scoring 3 or 4 in at least 1 of the
first 4 items of the FACT/GOG-NTX-13 (i.e., FACT/GOG-NTX-4), targeting
numbness, tingling or discomfort in hands and/or feet, assessed 9 months
after the first dose of chemotherapy. In addition to conventional safety
endpoints, Progressive Free Survival and Overall Survival are assessed in the
POLAR M study. In the POLAR A study Disease Free Survival is one addi-
tional safety endpoint assessed. Results are expected during second half
2020. Clinical trial information: NCT03654729.

TPS3617 Poster Session (Board #108a), Mon, 8:00 AM-11:00 AM TPS3618 Poster Session (Board #108b), Mon, 8:00 AM-11:00 AM
A phase IIb study of ramucirumab in combination with TAS102 versus Phase III randomized sequential open-label study to evaluate the efficacy of
TAS102 monotherapy in metastatic, chemotherapy refractory colorectal FOLFOX + panitumumab followed by FOLFIRI + bevacizumab (Sequence 1)
cancer patients: The RAMTAS trial of the German AIO (KRK-0316). First versus FOLFOX+ bevacizumab followed by FOLFIRI + panitumumab
Author: Stefan Kasper, University Hospital Essen, Essen, Germany (Sequence 2) in untreated patients with wild-type RAS metastatic, primary
left (L)-sided, unresectable colorectal cancer (CRC): The CR-SEQUENCE.
Background: Patients with metastatic colorectal cancer (mCRC) with pro-
First Author: Ramon Salazar, Oncobell Program IDIBELL Institut Català
gressive disease on/after or who are intolerant to fluoropyrimidines, oxali-
d’Oncologia Hospital Duran i Reynals, CIBERONC, Hospitalet, Spain
platin, irinotecan, anti-angiogenic and anti-EGFR therapies have limited
therapeutic options and a dismal prognosis, with a median survival below Background: Both anti-EGFR and anti-VEGF therapies have shown clinical
6 months. Recently, Trifluridin/Tipiracil (TAS102) significantly improved benefit when they are added in first and second-line in L-sided CRC. The
survival in patients with refractory mCRC and ramucirumab has been ap- conflicting results in anti-VEGF vs. anti-EGFR studies (FIRE-3, PEAK and
proved in combination with FOLFIRI for the treatment of patients with mCRC CALGB/SWOG 80405 studies) suggest that the sequence of targeted therapies
after prior FOLFOX/bevacizumab first line therapy. Previous studies on both added to FOLFOX or FOLFIRI regimens in first- and second-line treatment could
components provide a strong rationale to conduct a randomized study be an important factor in the overall survival (OS) of mCRC patients. Currently,
evaluating the efficacy and safety of ramucirumab in combination with there are no randomized data on the sequential use of an anti-EGFR followed by
TAS102 in patients with refractory mCRC to improve efficacy and prevent an anti-VEGF or vice versa. Therefore, the aim of this randomized clinical trial is
resistance. Methods: This is an interventional, randomized, open label, to compare the efficacy of two treatment sequences, panitumumab followed by
multicenter, phase IIb study in patients with advanced mCRC. Eligible bevacizumab versus bevacizumab followed by panitumumab in combination
patients will be randomized 1:1 and receive either ramucirumab and with FOLFOX chemotherapy in first-line and with FOLFIRI in second-line in
TAS102 (ramucirumab 8 mg/kg on d1+15, q4w and TAS102 35 mg/m² on patients with wild-type RAS, primary L-sided, metastatic colorectal cancer
d1-5 and d8-12, q4w) or TAS102 alone. Primary endpoint is overall survival (mCRC). Methods: A phase III, multicentre, open-label and randomized two-
as assessed by the Kaplan-Meier method, assuming a 6 months survival arm clinical trial. Untreated patients with wild-type RAS mCRC (determined
probability of 70% with ramucirumab in combination with TAS102 and 58% locally), primary L-sided and unresectable will be screened for this trial. Eligible
with TAS102 alone. Treatment groups are compared using the log-rank test. patients will be randomized 1:1 to receive first-line (1L) panitumumab plus
A total of 144 patients will be enrolled at 30 sites (1-sided alpha 0.10, power FOLFOX and then bevacizumab plus FOLFIRI as second-line (2L) treatment
0.80). Main secondary endpoints are overall response rate, disease control (Seq. 1) or bevacizumab plus FOLFOX as 1L and then panitumumab plus
rate, progression free survival and quality of life. In addition, a large com- FOLFIRI as 2L treatment (Seq. 2). Randomization will be stratified by number of
prehensive translational research program will be conducted to identify novel metastatic organs involved (1 vs . 1). Primary objective is the comparison of the
predictive and prognostic biomarkers. The study started in December 2018. progression free survival (PFS) rate at 35 months (m) of Seq 1 vs Seq. 2.
By February 2019, a total of 3 patients have been enrolled. Clinical trial Secondary objectives: PFS from randomization to 2nd progression or death, OS
information: NCT03520946. rate at 35 months and OS of Seq. 1 vs Seq. 2; PFS, objective response rate,
disease control rate, early tumour shrinkage, Depth of Response, duration and
time to response and safety in 1L treatment and in 2L treatment in each
Sequence arm. Exploratory objectives: impact of baseline biomarkers predictive
of the efficacy in each Sequence arm and the clinical impact of clonal dynamics
by longitudinal analysis of circulating tumour deoxyribonucleic acid (ctDNA) in
plasma. The trial is in progress; 28 of up to 370 planned patients have been
recruited at the end of January 2019 (first patient in 31 October 2018). Clinical
trial information: NCT03635021.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
220s Gastrointestinal (Colorectal) Cancer

TPS3619 Poster Session (Board #109a), Mon, 8:00 AM-11:00 AM TPS3620 Poster Session (Board #109b), Mon, 8:00 AM-11:00 AM
PRODIGE 53-UCGI 30 (SULTAN): A randomized phase II study comparing Phase II study of durvalumab following neoadjuvant chemoRT in stage II-IV
treatment intensification with hepatic arterial infusion chemotherapy plus rectal cancer: NSABP FR-2. First Author: Thomas J. George, NRG Oncology,
systemic chemotherapy to systemic chemotherapy alone in patients with and The University of Florida Health Cancer Center, Gainesville, FL
liver-only colorectal metastases considered still non resectable after at least
Background: Locally advanced rectal cancer remains a clinical challenge with
two months of systemic induction chemotherapy. First Author: Valerie Boige,
few improvements noted over the past few decades. Although immunotherapy
Digestive Oncology, Gustave Roussy, Villejuif, France
has no current clinical role in microsatellite stable (MSS) colorectal cancer,
Background: In the first-line setting, current combination chemotherapy (CT) preclinical models suggest that radiotherapy (RT) can enhance neoantigen
achieve high objective response rates (ORR) ranging from 40% to 80% and presentation, modulate the microenvironment, and improve the likelihood of
leads to complete resection rate (CRR) of colorectal cancer liver metastasis anti-tumor activity with checkpoint inhibitor use. This prospective phase II trial
(CRCLM) in 25% to 50% of patients with initially non-resectable CRLM. will test that hypothesis in addition to confirming the safety of this approach
However, when patients with CRLM are still not amenable to resection after using a “window-of-opportunity” study design with the anti-PD-L1 agent
induction CT, ORRs (~5-30%) and CRRs ( , 10%) obtained with second-line durvalumab (MEDI4736). Methods: This multi-center phase II trial is currently
systemic CT are much lower. Combining hepatic arterial infusion (HAI) and enrolling patients (pts) with rectal cancer who are undergoing standard NCCN
systemic (SYS) CT for unresectable CRLM lead to high ORR and the potential guideline-compliant neoadjuvant chemoradiotherapy (CRT). Eligibility in-
of cure in the second-line and further setting. Methods: This multicenter cludes pts with MSS stage II-IV rectal cancer with adequate organ function and
randomized phase II trial conducted in 20 centers in France plans to include pre-treatment diagnostic tumor available for profiling who are undergoing CRT
140 patients with CRLM still not amenable to a curative intent-treatment after with intent to proceed to surgical resection. Stage IV disease must be limited
at least two months of first-line induction SYS CT whatever the tumor response. such that the primary pelvic tumor requires definitive management. Standard
Patients are randomized (1:1) between intensified biweekly regimen com- ineligibility criteria include active infections, systemic steroid use, or other
bining HAI oxaliplatin (100 mg/m²) and SYS FOLFIRI (leucovorin 400 mg/m², conditions making immunotherapy use unsafe. Treatment includes durvalu-
irinotecan 180 mg/m², and 5-FU 2.4 g/m²over 48 h) plus cetuximab 500 mab (750mg IV infusion once every 2 wks) for 4 total doses beginning within 3-
mg/m² or bevacizumab 5 mg/kg (according to RAS status and prior response/ 7 days after CRT completion. Surgery must be within 8-12 wks of the final CRT
tolerance to SYS induction CT) or SYS CT alone according to current guide- dose. Primary endpoint is a demonstrated improvement in Neoadjuvant Rectal
lines. The primary endpoint is to evaluate and compare the curative-intent (R0- Cancer (NAR) score compared to historical controls targeting a 20% relative risk
R1) resection (and/or ablation) rate (CRR) of CRLM in both arms. A gain of 20% reduction in DFS and 3-4% absolute OS improvement. Secondary endpoints
in CRR is expected (30% in the experimental arm vs. 10% in the control arm; a include OS, DFS, toxicity, pCR, cCR, therapy completion, negative surgical
5% [two-sided]; b 20%). Stratification factors at randomization are prior margins, sphincter preservation, off-target “abscopal” effects for the subset of
adjuvant or SYS induction oxaliplatin-based CT, tumor response to SYS in- stage IV pts, and exploratory assessments of tumor infiltrating lymphocytes,
duction CT, and center. Secondary endpoints include progression-free (overall, circulating immunologic profiles, and molecular predictors of response. A
hepatic and extrahepatic) and overall survival, ORR, relative change in the sum safety run-in phase has completed as a precedent to full enrollment. Enroll-
of longest diameters of RECIST target lesions (DoR) and feasibility/tolerability. ment now continues to 47 total pts to achieve 41 surgically evaluable pts.
Since study start, 3 patients have been enrolled to date (planned accrual Support: AstraZeneca-Medimmune, NSABP Foundation. Clinical trial in-
period: 3 years). Clinical trial information: NCT03164655. formation: NCT03102047.

TPS3621 Poster Session (Board #110a), Mon, 8:00 AM-11:00 AM TPS3622 Poster Session (Board #110b), Mon, 8:00 AM-11:00 AM
Clinical verification of circulating tumor RNA (ctRNA) as novel pretreatment Phase II trial PD-L1/PD-1 blockade avelumab with chemoradiotherapy for
predictor and tool for quantitative monitoring of treatment response in locally advanced resectable T3B-4/N1-2 rectal cancer: The Ave-Rec trial.
metastatic colorectal cancer (mCRC): A biomarker study of the DEEPER First Author: Michael Michael, Division of Cancer Medicine, Peter MacCallum
trial. First Author: Yu Sunakawa, Department of Clinical Oncology, St. Cancer Centre, Melbourne, Australia
Marianna University School of Medicine, Kawasaki, Japan
Background: Standard neoadjuvant long course chemoradiotherapy (LCCRT)
Background: A randomized phase II trial, DEEPER (JACCRO CC-13) for locally advanced rectal cancer (LARC) results in a complete pathological
[NCT02515734], is on-going to evaluate FOLFOXIRI plus cetuximab response rate of 10-30%: but 20-40% of patients (pts) are non-responders,
(cet) vs. FOLFOXIRI plus bevacizumab (bev) in terms of depth of response as 10-15% have local recurrence. Tumoural immune infiltrates are predictive of
primary endpoint in 360 mCRC patients (pts) with RAS wild-type tumors, response. Preclinical studies show that radiotherapy (RT) via interferon sig-
PS0-1, and no previous chemotherapy. This is a head-to-head comparative naling is immuno-stimulatory, enhancing local/distant tumour cell death. RT
trial of 2 key monoclonal antibodies in mCRC treatment; therefore, it would also stimulates PDL1 production and the immunosuppressive activity of
be of interest to perform the biomarker study for developing novel predictors myeloid derived suppressor cells. Hence PDL1 inhibition may be required to
of cet or bev. The clinical utility of circulating tumor DNA (ctDNA) analysis enhance the immuno-stimulatory effects of RT. Hypothesis: In pts with re-
has been largely validated for monitoring during treatment and companion sectable LARC, the anti-PDL1 antibody Avelumab post LCCRT may enhance
diagnostics after chemotherapy. However, there are few published results the pathological/imaging response rates whilst potentially reducing local/
regarding ctRNA in cancer treatment. Use of ctRNA from liquid biopsies distant relapse rates. Methods: (1) Trial Design: Phase II single arm trial,
would enhance tumor profiling through the trending of actionable bio- across 6 Australian sites (2) Endpoints: (a) Primary; Pathological response rate
markers not found in ctDNA, and allow for patient monitoring by measuring post-LCCRT, as documented by central pathologist, (b) Secondary; MRI/FDG
dynamic changes in levels of gene expressions. Methods: This study will PET imaging responses at 8 weeks post LCCRT (pre-surgery). Toxicity. (c)
enroll pts with willing to undergo biopsies of both tissue and blood among Exploratory; Tumoural immune cell subsets/checkpoint expression (by mul-
participants of the DEEPER trial. The estimated number is 250. The main tiplex immunohistochemistry and in-vitro functional assays) and ctDNA
purpose is to find novel predictors for efficacy of cet or bev in mCRC using analysis at baseline and during treatment. Distant relapse-free survival and the
liquid biopsies, which are performed to obtain ctDNA and ctRNA at 6 time documentation of sites of relapse. (3) Sample size: An increase in the pro-
points: pre-treatment, 8 weeks after treatment start, beginning of mainte- portion of pathological complete responses by . 25% (from 10% to 35%) will
nance phase in FOLFOXIRI-regimen, progression, before and 8 weeks after be considered clinically important. Power = 90%, a = 0.05, 41 pts are
2nd-line treatment. The tissue samples collected before chemotherapy will required– an additional 4 pts to allow for drop-out. Total sample size = 45pts.
be used for analyzing intra-tumoral genetic alterations. Associations be- Treatment: All pts to receive standard LCCRT (50.4Gy RT plus 5FU [225mg/
tween analytes in blood/tissue and the clinical outcomes of each treatment m2/day/CI] or Capecitabine [825mg/m2 BID on RT days] over 5.5 weeks). Post
(with cet or bev) will be assessed using Fisher’s exact test, Kaplan-Meier LCCRT (prior to surgery), pts receive 4 cycles Avelumab (10mg/kg, q2 weeks).
curves, and log-rank tests in univariate analyses. We will evaluate on whether Surgical resection 10-12 weeks post LCCRT. Fresh tumour biopsy and ctDNA
transcriptomic analysis in ctRNA could predict treatment efficacy more sampling pre LCCRT, pre Cycle 1 Avelumab and at surgery. Response by FDG
accurately compared to genomic analysis in ctDNA, and verify the clinical PET and pelvic MRI pre surgery. Pts to be followed up for 2 years. Major
utility of ctRNA testing in mCRC treatment. Accrual will continue until the Inclusion Criteria: Pts with LARC, MRI stage T3b-4/N1-2/M0, planned for
DEEPER trial is completed. Clinical trial information: UMIN000018412. LCCRT followed by curative resection, tumoural lower border within 12cm from
the anal verge, measurable disease (RECIST1.1), ECOG 0-1, adequate organ
function and no contraindications to Avelumab therapy. Current Enrolment: 11
of the planned 45 patients enrolled. Clinical trial information: NCT03299660.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Colorectal) Cancer 221s

TPS3623 Poster Session (Board #111a), Mon, 8:00 AM-11:00 AM TPS3624 Poster Session (Board #111b), Mon, 8:00 AM-11:00 AM
Efficacy of aspirin for stage III colorectal cancer: A randomized double-blind Aspirin as adjuvant treatment for colorectal cancer: Rationale and progress
placebo-controlled trial (JCOG1503C, EPISODE-III trial). First Author: Atsuo of the Add-Aspirin trial. First Author: Ruth E Langley, Medical Research
Takashima, Division of Gastrointestinal Medical Oncology, National Cancer Council Clinical Trials Unit at University College London, London, United
Center Hospital, Tokyo, Japan Kingdom
Background: Adjuvant chemotherapy is the current standard treatment for Background: There is now a body of evidence indicating a potential role for
stage III colorectal cancer after curative resection. However, the prognosis of aspirin in colorectal cancer (CRC) prevention. In cardiovascular trials, effects
stage III colorectal cancer is still poor even after curative resection and ad- on incidence of cancer metastases and short-term mortality suggest further
juvant chemotherapy. Recently, several observational studies suggested possible roles in the treatment setting, supported by observational studies of
the anti-tumor effect of aspirin for advanced colorectal cancer. The main aspirin use after cancer diagnosis. In the prevention setting, aspirin use has
mechanism of the anti-tumor effect by aspirin may be to suppress cyclo- been limited by toxicity concerns, particularly of serious bleeding. In the
oxygenase activity in the arachidonic acid cascade and to inhibit the pro- adjuvant setting, benefits associated with reducing recurrence and subsequent
duction of prostaglandins involved in tumor growth. So far, aspirin showed a treatment may outweigh these risks. The Add-Aspirin trial will investigate this,
prolongation of survival for colorectal cancer in several retrospective studies. and will also consider possible mechanisms of action for aspirin effects, in-
However, in these studies, aspirin was given not to be evaluated the effect on cluding the impact of PIK3CA mutations, where there are currently several
prognosis of colorectal cancer in adjuvant setting but to prevent cardiovascular theories and conflicting data. Methods: Add-Aspirin (ISRCTN74358648) is
event. In addition, baseline patient characteristics were imbalanced between an international, phase III, double-blind, randomised, placebo-controlled trial
aspirin group and non-aspirin group and both dosage amount and dosing recruiting patients who have undergone surgery and relevant adjuvant treat-
period of aspirin were different among patients. Methods: We planned a ment for stage II or III CRC, as well as those with completely resected CRC liver
randomized double-blind placebo-controlled phase III trial commenced in metastases. Parallel randomised cohorts will address the question in breast,
Japan in March 2018 to confirm the superiority of aspirin in terms of disease- gastro-oesophageal and prostate cancer. Participants take aspirin 100mg daily
free survival (DFS) over placebo for stage III colorectal cancer patients after for an 8-week run-in, to assess adherence and toxicity, and those suitable to
curative resection. Patients receive aspirin (100 mg/day) or placebo for 3 years proceed are randomised (1:1:1) to aspirin 100mg, aspirin 300mg or placebo
with the standard adjuvant chemotherapy of mFOLFOX6, CAPOX or capeci- daily for at least 5 years. A number of measures – including blood pressure
tabine until relapse or unacceptable toxicities. The primary endpoint is DFS control and PPI use where relevant - are in place to reduce bleeding risk. The
and the secondary endpoints are overall survival, relapse-free survival, relative primary outcome is disease-free survival (target hazard ratio = 0.8, n = 2600 in
dose intensity, adverse events, and serious adverse events. We assumed the 3- 5 years) with a long term analysis of survival planned across the tumour groups.
year DFS of aspirin arm as 74% based on two previous trials conducted by Translational work includes a sub-study monitoring urinary thromboxane B2
JCOG and expected a 6% increase in the 3-year DFS with aspirin adding to as a marker of platelet activation in a subgroup (n = 500) to investigate
standard adjuvant chemotherapy after curative surgery. A total of 880 patients mechanisms of action. Add-Aspirin opened in 2015 and recruited 1505 CRC
will be accrued from 20 Japanese institutions within 3 years, and 47 patients patients during the first 3 years from 137 UK centres. 1282 (85%) proceeded
were enrolled as of Jan 31, 2019. Both aspirin and placebo are provided by to randomisation. A pre-planned feasibility analysis of run-in data (n = 2253
Bayer Yakuhin Ltd. This trial has been registered at Japan Registry of Clinical across all 4 tumour groups) provided reassuring data on safety, tolerability and
Trials as jRCTs031180009 (https://jrct.niph.go.jp/detail/589). Clinical trial adherence, and recruitment continues with centres in India and Republic of
information: jRCTs031180009. Ireland recently joining. Clinical trial information: 74358648.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
222s Gastrointestinal (Noncolorectal) Cancer

4000 Oral Abstract Session, Sun, 9:45 AM-12:45 PM 4001 Oral Abstract Session, Sun, 9:45 AM-12:45 PM
APACT: phase III, multicenter, international, open-label, randomized trial of ARTIST 2: Interim results of a phase III trial involving adjuvant chemotherapy
adjuvant nab-paclitaxel plus gemcitabine (nab-P/G) vs gemcitabine (G) for and/or chemoradiotherapy after D2-gastrectomy in stage II/III gastric cancer
surgically resected pancreatic adenocarcinoma. First Author: Margaret A. (GC). First Author: Se Hoon Park, Samsung Medical Center, Sungkyunkwan
Tempero, University of California, San Francisco, San Francisco, CA University School of Medicine, Seoul, South Korea
Background: In metastatic pancreatic cancer (PC), nab-P/G demonstrated Background: Adjuvant chemotherapy and/or chemoradiotherapy have been
significantly longer overall survival (OS) vs G. APACT assessed efficacy & the standard of care in GC for years, supported by randomized trials. We
safety of nab-P/G vs G in surgically resected PC. Methods: Treatment (tx)- compared the efficacy of different chemotherapy regimens and chemo-
naive patients (pts) with histologically confirmed PC, macroscopic complete radiotherapy in patients with D2-resected, stage II/III, node-positive GC.
resection, ECOG PS 0/1, & CA19-9 , 100 U/mL were eligible. Stratification Methods: From Feb 2013 through Nov 2018, we randomly assigned, in a 1:1:
factors: resection status (R0/R1), lymph node status (LN+/2), & geographic 1 ratio, patients with pathologically-staged II or III, node-positive, D2-resected
region. Tx was initiated # 12 wks postsurgery. Pts received nab-P 125 GC, to receive adjuvant S-1 (40-60 mg twice daily 4-weeks-on/2-weeks-off) for
mg/m2 + G 1000 mg/m2 or G 1000 mg/m2 on days 1, 8, 15 of six 28-day cycles. one year, S-1 (2-weeks-on/1-week-off) plus oxaliplatin 130 mg/m2 (SOX) for
Primary endpoint was disease-free survival (DFS) by independent reviewer six months, or SOX plus chemoradiotherapy 45 Gy (SOXRT). Randomization
(IR); IRs received baseline clinical data & scans. Secondary endpoints were was stratified according to the type of surgery (total or subtotal gastrectomy),
OS & safety. »438 DFS events were needed for 90% power to detect an HR stage (II or III), and Lauren histologic classification (diffuse or intestinal). The
for disease recurrence or death of 0.73 with nab-P/G vs G at a 2-sided primary endpoint was disease-free survival (DFS). A total of 900 patients had to
significance level of 0.05. Results: 866 pts were randomized. Median age be enrolled to demonstrate superiority of SOX or SOXRT to S-1 (hazard ratio
was 64 y (range, 34 - 86); most pts had ECOG PS 0 (60%), LN+ (72%), & R0 [HR] 0.667), with 90% power at a two-sided significance level of 5%.
(76%). 69% of pts completed 6 tx cycles (nab-P/G, 66%; G, 71%). Median Results: A total of 538 patients were included for this interim efficacy analysis.
follow up for OS was 38.5 mo. Median IR-assessed DFS (439 events) was Median age was 58 years, men constituted 65%, and stage II and III were 31%
19.4 mo (nab-P/G) vs 18.8 mo (G) (HR, 0.88; 95% CI, 0.729 - 1.063; and 69%, respectively. Baseline tumor and patient characteristics were
stratified log-rank P = 0.1824). Investigator-assessed DFS (571 events) was balanced between treatment arms. Adverse events were as anticipated in each
16.6 mo (nab-P/G) vs 13.7 mo (G) (HR, 0.82; 95% CI, 0.694 - 0.965; arm, generally well-tolerated and manageable. DFS in the control arm (S-1)
nominal P = 0.0168). Interim OS (427 events) was 40.5 mo (nab-P/G) vs were significantly shorter than in SOX and SOXRT arms (stratified HR for
36.2 mo (G) (HR, 0.82; 95% CI, 0.680 - 0.996; nominal P = 0.045). recurrence): S-1 vs. SOX, 0.617 (P = 0.016) and S-1 vs. SOXRT, 0.686 (P =
Grade $ 3 TEAEs were reported in 86% vs 68% of pts with nab-P/G vs G. The 0.057). The DFS at 3-years was found to be 65%, 78% and 73% in S-1, SOX
most common grade $ 3 hematologic & nonhematologic TEAEs with nab-P/ and SOXRT arms, respectively. No difference in DFS between SOX and SOXRT
G vs G were neutropenia (49% vs 43%) & fatigue (10% vs 3%). TEAEs led to was found (HR 0.910, P = 0.667). Based on the results after the observation
death in 2 pts in each arm. Conclusions: IR DFS with nab-P/G was not of 145 recurrence events at the cutoff date of Dec 27, 2018, the indepen-
significantly longer vs G; median DFS with G was longer than historical data. dent data monitoring committee considered the results sufficient to meet
DFS by investigator (sensitivity analysis) and interim OS were improved with the endpoint of the trial and recommended early stopping of the trial.
nab-P/G vs G (HR 0.82 for both). Adjuvant nab-P/G may be an option for pts Conclusions: In patients with curatively D2-resected, stage II/III, node-positive
who are ineligible for FOLFIRINOX. Additional OS follow-up may better GC, adjuvant SOX or SOXRT was effective in prolonging DFS, when compared
support nab-P/G as an option in the adjuvant setting. Clinical trial in- to S-1 monotherapy. Clinical trial information: NCT0176146.
formation: NCT01964430.

4002 Oral Abstract Session, Sun, 9:45 AM-12:45 PM 4003 Oral Abstract Session, Sun, 9:45 AM-12:45 PM
A multicenter randomized controlled trial to evaluate the efficacy of surgery ABC-06 | A randomised phase III, multi-centre, open-label study of active
vs. radiofrequency ablation for small hepatocellular carcinoma (SURF trial). symptom control (ASC) alone or ASC with oxaliplatin / 5-FU chemotherapy
First Author: Namiki Izumi, Musashino Red Cross Hospital, Tokyo, Japan (ASC+mFOLFOX) for patients (pts) with locally advanced / metastatic biliary
tract cancers (ABC) previously-treated with cisplatin/gemcitabine (CisGem)
Background: Surgery (SUR) and radiofrequency ablation (RFA) are both
chemotherapy. First Author: Angela Lamarca, Department of Medical On-
known to be effective therapy for treating patients with small oligonodular
cology, The Christie NHS Foundation Trust / Institute of Cancer Sciences,
hepatocellular carcinoma (HCC), however there is only insufficient evidence
University of Manchester, Manchester, United Kingdom
about which therapy is more preferred approach. This randomized controlled
trial was designed to prospectively compare the efficacy of SUR and RFA as Background: Level A evidence supports use of CisGem as first-line chemo-
the first approach to primary HCC. Methods: In this open-label trial un- therapy for ABC; no robust evidence is available for second-line chemotherapy.
dertaken at 49 hospital in Japan, we recruited patients having primary HCC Methods: Pts diagnosed with ABC with disease progression after prior CisGem
with tumor foci numbering less than 3, each measuring 3 cm or less, Child- were randomised (1:1) to either ASC+mFOLFOX or ASC. Randomisation was
Pugh score of 7 or less, ages between 20 and 79 year. Before randomization, stratified by serum albumin levels ( , 35 vs $35 g/L), platinum sensitivity
technical and liver functional feasibility for both treatment arms were (determined from first-line CisGem) and disease extent (locally advanced vs
confirmed by joint chart review by surgeons and hepatologists. Patients were metastatic). Pts with ECOG PS0-1, adequate haematological, renal and liver
then randomly assigned in a 1:1 ratio to undergo SUR or RFA, stratified by function, and adequate biliary drainage were eligible. Primary end-point was
age, infection of hepatitis-C virus, number of tumors, tumor size and in- overall survival (OS) (multivariable Cox regression adjusted for stratification
stitution. The primary endpoint was recurrence free survival (RFS) and factors); sample size: 162 pts delivering 148 events were required (80%
overall survival (OS). Results: Between April 2009 and August 2015, total power; 5% two-sided alpha) for a hypothesised hazard ratio (HR) of 0.63.
308 patients were enrolled to this trial. Because of ineligibility 15 patients Assumed median survival for ASC was 4 months. Results: 162 pts (81 in each
were excluded, therefore 145 patients underwent SUR and 148 patients arm) were randomised (27 March ‘14 - 04 Jan ‘18); median age 65 yrs (range
underwent RFA finally. There was no perioperative mortality. Under the 26-84); sex: 80 (49%) male, 82 (51%) female; primary site: intrahepatic 72
median follow-up of 5 years, the 3-year RFS of patients underwent SUR and (44%), extrahepatic 45 (28%), gallbladder 34 (21%) and ampullary 11 (7%).
RFA was 49.8%, 47.7%, respectively (hazard ration [HR] 0.96, 95% CI Baseline characteristics were balanced between arms except platinum sen-
0.72-1.28; p = 0.793). OS will be analyzed and published after two years. sitivity (ASC+mFOLFOX 27 pts (33%); ASC 34 pts (42%)). After 150 OS
Conclusions: SUR and RFA were both safe therapeutic approaches and events, the adjusted HR was 0.69 (95% CI 0.50-0.97; p = 0.031;
provided equally RFS for early stage HCC smaller than 3 cm. Clinical trial ASC+mFOLFOX vs ASC). Median OS (months (m)), 6m and 12m OS-rate (%)
information: UMIN000001795. were 6.2m, 50.6% and 25.9% for the ASC+mFOLFOX and 5.3m, 35.5%,
11.4% for the ASC arm, respectively. Grade 3/4 toxicities were reported in
48 (59%) and 32 (39%) pts in the ASC+mFOLFOX and ASC arm, re-
spectively; these were balanced between arms except for fatigue and
neutropenia (more frequent in ASC+mFOLFOX arm); data cleaning is on-
going. No chemotherapy-related deaths were reported. Conclusion: Survival
with ASC was greater than assumed; ASC+mFOLFOX improved OS after
progression to CisGem with a clinically meaningful increase in 6m and 12m
OS rate. ASC+mFOLFOX should become standard of care in second-line for
ABC. Clinical trial information: NCT01926236.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Noncolorectal) Cancer 223s

4004 Oral Abstract Session, Sun, 9:45 AM-12:45 PM 4005 Oral Abstract Session, Sun, 9:45 AM-12:45 PM
Results of KEYNOTE-240: phase 3 study of pembrolizumab (Pembro) vs best Prospective randomized phase II trial of pazopanib versus placebo in patients
supportive care (BSC) for second line therapy in advanced hepatocellular with progressive carcinoid tumors (CARC) (Alliance A021202). First Author:
carcinoma (HCC). First Author: Richard S. Finn, University of California, Los Emily K. Bergsland, University of California San Francisco, San Francisco, CA
Angeles, Los Angeles, CA
Background: Patients (pts) with progressive advanced well-differentiated
Background: Pembro received accelerated approval based on results of neuroendocrine tumors arising outside of the pancreas have limited sys-
KEYNOTE-224, a phase 2 trial in pts with advanced HCC in the second line temic treatment options. Pazopanib (PZ) is an oral multi-kinase inhibitor
setting. KEYNOTE-240 (NCT02702401) was a randomized, placebo (Pbo) with activity against VEGFR-2,-3, PDGFR-a, and b, and c-KIT, with initial
controlled, phase 3 study of Pembro vs BSC in pts with previously treated data suggesting efficacy in CARC. Methods: This was a multicenter, ran-
advanced HCC. Methods: Eligible pts had a radiographic or pathologic di- domized, double-blind, phase II study of PZ (800 mg/day) versus placebo
agnosis of HCC, radiographic progression on/intolerance to sorafenib, Child- (PL) in progressive CARC. Key eligibility: low-intermediate grade CARC,
Pugh A disease and ECOG PS 0-1. Pts were randomized 2:1 to receive Pembro radiologic progressive disease (PD) , 12 months (mo), and adequate end-
200 mg + BSC or Pbo + BSC IV every 3 wk, stratified by geographic region, organ function. Prior somatostatin analog (SSA) mandated for midgut tu-
macrovascular invasion and ɑ-fetoprotein levels for #35 cycles or until mors. Concurrent SSA allowed if previous PD on SSA documented. Primary
confirmed PD/unacceptable toxicity. Response was assessed every 6 wk per endpoint was progression-free survival (PFS), defined as time from ran-
RECIST v1.1 by central imaging review. Co-primary endpoints were OS and domization to PD by central review or death. Secondary endpoints included
PFS. Secondary endpoints included ORR, DOR and safety. Data cutoff was Jan overall survival (OS), objective response rate (ORR) and safety. The trial had
2 2019 for OS; Mar 26 2018 for PFS and ORR. Results: 413 patients were 85% power to detect a difference in median PFS of 14 v 9 mo (hazard ratio
randomized; 278 to Pembro and 135 to Pbo. After a median follow up of 13.8 [HR] 0.64) at one-sided alpha = 0.1. A stratified log-rank test based on the
mo, 10.1% of pts remained on Pembro and 3.0% on Pbo. Pembro improved intend-to-treat (ITT) principle was used. Unblinding and crossover were
OS (HR: 0.78; one sided p = 0.0238) and PFS (HR: 0.78; one sided p = allowed if PD confirmed by central review. Results: 171 (97 PZ, 74 PL) pts
0.0209) vs Pbo; these differences did not meet significance per the pre- were randomized between 6/2013-10/2015: median age 63; 56% female;
specified statistical plan. ORR was 16.9% (95% CI 12.7-21.8%) for Pembro 66% small bowel primary; 87% concurrent SSA. Median follow-up of 31 mo;
vs 2.2% (95% CI 0.5-6.4%) for Pbo (nominal one sided p = 0.00001); re- 112 (56 PZ, 56 PL) PFS events observed. 6 pts (4 PZ, 2 PL) remain on initial
sponses on Pembro were durable (median DOR: 13.8 mo [1.5-23.6+]). Off treatment. Median PFS was 11.6 and 8.5 mo in PZ and PL, respectively
study, new therapy use was 42% for Pembro and 47% for Pbo. The safety (HR = 0.53, 1-sided 90% upper confidence limit [UCL] 0.69, p = 0.0005)
profile including incidence of hepatitis and other immune mediated events was which crossed the pre-specified protocol efficacy boundary. 49 PL pts re-
generally consistent with that previously reported in Pembro studies; no cases ceived PZ after PD. Median OS was 41 and 42 mo in PZ and PL, respectively
of HBV/HCV flare were identified. Conclusions: Pembro reduced the risk of (HR = 1.13, 1-sided 90% UCL 1.51, p = 0.70). RR data will be presented.
death by 22% and improved PFS over Pbo in pts with advanced HCC, although Notable grade 3+ adverse events were (PZ v. PL %) hypertension (35 v. 8),
significance was not reached per prespecified statistical criteria. ORR in the fatigue (11 v. 4), ALT (10 v. 0), AST (10 v. 0), and diarrhea (7 v. 4).
Pembro arm was consistent with that of KEYNOTE-224. Subsequent anti- Conclusions: PZ compared to PL was associated with significant improve-
cancer therapy in the Pbo arm likely impacted the OS results. The safety profile ment in PFS in patients with progressive CARC. The results confirm that
was comparable to that established for Pembro monotherapy. These results are VEGF signaling pathway is a valid target for therapy in CARC. Support:
overall consistent with those of KEYNOTE-224 further supporting second line U10CA180821, U10CA180882 https://acknowledgments.alliancefound.org.
therapy with Pembro in HCC pts. Clinical trial information: NCT02702401. Clinical trial information: NCT01841736.

4006 Oral Abstract Session, Sun, 9:45 AM-12:45 PM LBA4007 Oral Abstract Session, Sun, 9:45 AM-12:45 PM
Optimizing chemotherapy for frail and elderly patients (pts) with advanced Pembrolizumab with or without chemotherapy versus chemotherapy for
gastroesophageal cancer (aGOAC): The GO2 phase III trial. First Author: advanced gastric or gastroesophageal junction (G/GEJ) adenocarcinoma:
Peter S Hall, University of Edinburgh, Edinburgh, United Kingdom The phase 3 KEYNOTE-062 study. First Author: Josep Tabernero, Vall
d’Hebron University Hospital and Institute of Oncology, Barcelona, Spain
Background: Many pts with aGOAC are elderly and/or frail. We previously
compared epirubin/ oxaliplatin/ capecitabine (EOCap) vs OCap vs Cap in a
pick-the-winner study and found OCap best. GO2 was designed to find the
optimum dose of OCap and to explore the use of an objective baseline geriatric
assessment to individualize doses for maximum Overall Treatment Utility
(OTU), a composite of clinical benefit, tolerability, QL and patient value.
Methods: Pts with aGOAC were eligible if unsuitable for full-dose EOCap due to
age or frailty, but fit for OCap; GFR $ 30, bili ,2x ULN. Baseline assessment
included global QL; symptoms; functional scales; comorbidity; frailty. Ran-
domization was 1:1:1 to dose Level A (Ox 130 mg/m2d1, Cap 625 mg/m2bd
d1-21, q21d), B (80% Level A doses) or C (60% Level A doses). Pts with GFR
30-50 ml/min or bili 1.5-2.0 xULN received 75% of the allocated dose of Cap.
At 9 wks, pts were scored for OTU. Continuation thereafter was based on
clinical judgement. Non-inferiority (vs A) was assessed using PFS censored at The full, final text of this abstract will be available at
12 months, with boundary HR 1.34 (based on discussion with pts and cli-
nicians), needing 284 PFS events per 2-way comparison. Baseline fitness was abstracts.asco.org at 7:30 a.m. ET on Saturday, June 1.
assessed as predictive of OTU, overall and by interaction with dose level. Onsite at the Meeting, this abstract will be printed in the
Results: 514 pts were randomised, 2014-17, at 61 UK centres. Clinical trial Sunday edition of ASCO Daily News.
information: 44687907. Non-inferiority of PFS is confirmed for Level B vs A
(HR 1.09, CI 0.89-1.32) and for Level C vs A (HR 1.10, CI 0.90-1.33). Level C
pts had less toxicity and better OTU outcomes than A or B. When analysed by
baseline age, frailty and PS, Level C produced the best OTU even in younger,
less frail and better PS patients; no group was identified who benefit more from
the higher dose levels. Conclusions: This is the largest RCT to date specifically
investigating frail and/or elderly aGOAC pts, and should guide future treatment.
The lowest dose tested was non-inferior in terms of PFS and produced less
toxicity and better overall treatment utility.
Level A Level B Level C
Pts (PFS events) 170 (142). 171 (147). 173 (149)
Median age 76 76 77
% PS ‡2 31 32 31
% any Frailty; % very Frail 86; 61 82; 56 76; 58
% any Gr ‡3 non-haem adverse event. 56 56 37
Median PFS mo 4.9 4.1 4.3
OTU (wk 9): % Good/intermed./poor 35/34/31 36/26/38 43/27/29
Median OS mo 7.5 6.7 7.6

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
224s Gastrointestinal (Noncolorectal) Cancer

4009 Poster Discussion Session; Displayed in Poster Session (Board #114), 4010 Poster Discussion Session; Displayed in Poster Session (Board #115),
Mon, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Mon, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Mon, 3:00 PM-4:30 PM Mon, 3:00 PM-4:30 PM
Efficacy and safety of pembrolizumab (pembro) alone or in combination with Pembrolizumab versus chemotherapy as second-line therapy for advanced
chemotherapy (chemo) in patients (pts) with advanced gastric or gastroesoph- esophageal cancer: Phase 3 KEYNOTE-181 study. First Author: Manish A.
ageal (G/GEJ) cancer: Long-term follow up from KEYNOTE-059. First Author: Shah, Weill Cornell Medicine, New York-Presbyterian Hospital, New York, NY
Zev A. Wainberg, David Geffen School of Medicine at UCLA, Los Angeles, CA
Background: The phase 3 KEYNOTE-181 study compared pembrolizumab (pem-
Background: Interim analysis of a global, phase 2 KEYNOTE-059 study bro) vs chemo as second-line therapy for patients (pts) with advanced/metastatic
(NCT02335411) reported manageable safety and promising antitumor squamous cell carcinoma (SCC) and adenocarcinoma (ACC) of the esophagus
activity for pembro alone or pembro + chemo in pts with G/GEJ cancer. Here (NCT02564263). Methods: Eligible pts were randomized 1:1 to pembro 200 mg
we report long-term efficacy and safety data of all cohorts. Methods: Pts with Q3W for up to 2 years or choice of paclitaxel, docetaxel, or irinotecan. Randomization
recurrent or metastatic G/GEJ adenocarcinoma were enrolled in 3 cohorts. was stratified by histology (SCC vs adenocarcinoma) and region (Asia vs rest of
world). Primary end points were OS in the SCC, PD-L1 combined positive score
Cohort 1 pts (PD-L1positive or negative) received pembro alone after $2
(CPS) $10, and the ITT. Secondary endpoints included PFS, ORR, safety; ex-
prior lines of therapy. Cohort 2 pts (PD-L1positive or negative) received
ploratory endpoints included health-related quality of life (HRQoL) in CPS $10.
pembro + cisplatin (80 mg/m2 day 1) + 5-fluorouracil (800 mg/m2 days 1-5 Results: 628 pts were randomized (401 with SCC; 222 with CPS $10). As of Oct.
Q3W) or capecitabine (in Japan only, 1000 mg/m2 twice daily) as first-line. 15, 2018, median follow-up was 7.1 mo (pembro) vs 6.9 mo (chemo). In CPS $10,
Cohort 3 pts (PD-L1positive, combined positive score $1% using the PD-L1 OS was superior with pembro vs chemo (median 9.3 vs 6.7 mo; HR 0.69; 95% CI
IHC 22C3 pharmDx assay) received pembro alone as first-line. All pts re- 0.52-0.93; P= 0.0074). In CPS $10 SCC, median OS was 10.3 mo vs 6.7 mo and
ceived pembro 200 mg Q3W for up to 2 years. End points included safety, in CPS $10 ACC, median OS was 6.3 mo vs 6.9 mo; 12-mo OS rates were higher
ORR, DOR, and OS. Results: At data cutoff (Aug 8, 2018), median (range) with pembro vs chemo (Table). In SCC, median OS was 8.2 mo vs 7.1 mo; HR 0.78;
follow-up was 6 (1-38), 14 (2-40), and 21 (2-36) months for cohorts 1 (n = 95% CI 0.63-0.96; P= 0.0095. In the ITT, median OS was 7.1 mo vs 7.1 mo; HR
259), 2 (n = 25), and 3 (n = 31), respectively. In cohort 1, confirmed ORR 0.89; 95% CI 0.75-1.05; P= 0.0560. Updated OS will be presented. Grade 3-5
(95% CI) was 11.6% (8-16) overall, 15.5% (10-22) in PD-L1positive, and drug-related AEs ($10% incidence in either arm) included decreased white blood
6.4% (3-13) in PD-L1negative tumors. In cohort 2, confirmed ORR was cells (0% vs 10%), decreased neutrophils (0.3% vs 10%). In CPS $10, HRQoL
60.0% (39-79) overall, 73.3% (45-92) in PD-L1positive, and 37.5% (9-76) improved with pembro vs chemo only for EQ-5D VAS (difference in LS mean change
in PD-L1negative tumors. In cohort 3, confirmed ORR was 25.8% (12-45). from baseline 5.57; 95% CI 0.58-10.56). Conclusions: Pembro significantly im-
Median (range) DOR in months was 16.1 (2-35+), 4.6 (3-37+), and not proved OS vs chemo as second-line therapy for advanced esophageal cancer with
reached (2.1-32.5+) in cohorts 1, 2, and 3, respectively. OS at 1 year/2 years PD-L1 CPS $10, with a more favorable safety profile and stable and similar QOL.
These data support pembro as a new second-line standard of care for esophageal
was 24.6%/12.5%, 52%/32%, and 63.6%/40.1% in cohorts 1, 2, and 3,
cancer with PD-L1 CPS $10. Clinical trial information: NCT02564263.
respectively. In cohorts 1, 2, and 3, grade 3-5 treatment-related adverse
event (TRAE) incidence was 46 (18%), 20 (80%), and 8 (26%) respectively. CPS ‡10

TRAEs led to discontinuation in 6 (2%) and 3 (12%) pts in cohorts 1 and 2, Total SCC ACC

respectively, and to death in 2 (1%) pts in cohort 1. No TRAEs led to Pembro Chemo Pembro Chemo Pembro Chemo
N = 107 N = 115 N = 85 N = 82 N = 22 N = 33
discontinuation or death in cohort 3. Conclusions: These updated results
12-mo OS, % 43 20 48 23 23 15
demonstrate manageable safety, durable clinically meaningful activity of Median PFS (95% CI), mo 2.6 3.0 3.2 2.3 2.1 3.7
pembro in heavily pretreated pts, and promising efficacy of first-line pembro 12-mo PFS, %
(2.1-4.1)
21
(2.1-3.7)
7
(2.1-4.4)
23
(2.1-3.4)
7
(1.9-3.5)
14
(2.0-5.7)
7
(alone or + chemo) in pts with advanced G/GEJ cancer. Clinical trial in- ORR, % 21.5 6.1 22 7 18 3
Median DOR (range), mo 9.3 7.7 9.3 7.7 Not reached 4.4
formation: NCT02335411. (2.1+ to 22.6+) (4.3 to 16.8+) (2.1+ to 18.8+) (4.3 to 16.8+) (6.5 to 22.6+) (4.4-4.4)

4011 Poster Discussion Session; Displayed in Poster Session (Board #116), 4012 Poster Discussion Session; Displayed in Poster Session (Board #117),
Mon, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Mon, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Mon, 3:00 PM-4:30 PM Mon, 3:00 PM-4:30 PM
First-line pembrolizumab (P), trastuzumab (T), capecitabine (C) and oxaliplatin Nivolumab (NIVO) + ipilimumab (IPI) combination therapy in patients (pts)
(O) in HER2-positive metastatic esophagogastric adenocarcinoma. First Author: with advanced hepatocellular carcinoma (aHCC): Results from CheckMate 040.
Yelena Yuriy Janjigian, Memorial Sloan Kettering Cancer Center, New York, NY First Author: Thomas Yau, The University at Hong Kong, Hong Kong, China
Background: Trastuzumab stimulates HER2-specific T cell responses and Background: NIVO monotherapy (mono) is approved for sorafenib (SOR)-treated pts
increases tumor PD-L1 expression, and anti-PD-1 antibody can help enhance with HCC based on data from CheckMate 040 (NCT01658878), which reported an
T cell-specific immunity of trastuzumab. We conducted a phase II trial of objective response rate (ORR) of 14% and median overall survival (mOS) of
pembrolizumab with chemotherapy/trastuzumab. Methods: Patients (pts) with 16 months (mo). This is the first report of efficacy and safety of the NIVO + IPI
previously untreated HER2 IHC 3+ or FISH+ tumors irrespective of PD-L1 combination in SOR-treated pts with aHCC. Methods: Pts were randomized to 3 arms:
status received intravenous P 200 mg flat dose, T 6 mg/kg (after 8 mg/kg load), [A] NIVO 1 mg/kg + IPI 3 mg/kg Q3W (4 doses) or [B] NIVO 3 mg/kg + IPI 1 mg/kg
Q3W (4 doses), each followed by NIVO 240 mg Q2W, or [C] NIVO 3 mg/kg Q2W + IPI 1
O 130 mg/m2 every 3 weeks and oral C 850 mg/m2 2 weeks on/1 week off. 22
mg/kg Q6W. Treatment continued until intolerable toxicity or disease progression.
pts received 1 cycle of induction P/T prior to initiation of chemotherapy. The
Primary endpoints included safety and tolerability. Secondary endpoints included
primary endpoint was 6-months PFS; with target accrual of 37 pts. Secondary ORR (BICR per RECIST v1.1), duration of response (DOR), disease control rate (DCR),
endpoints included safety, OS, ORR, and biomarker analysis. Results: Accrual and OS. Cutoff was 25 Sep 2018. Results: 148 SOR-treated pts were randomized.
completed and 100% of the 32 evaluable pts had tumor regression (ranging Minimum follow-up for OS from last pt randomization date to data cutoff was 24 mo.
from -20% to -100%). The RECIST 1.1 ORR was 87% (25 PR, 3 CRs), and 12 At baseline: 88% had vascular invasion or extrahepatic spread, 91% had BCLC stage
(52%) of pts that received induction P/T x 1 cycle showed reduction in target C, 84% discontinued SOR due to disease progression and 14% due to toxicity.
lesions. Median PFS was 11.3 months (mo), with 67% 6 mo PFS. Median Overall, ORR was 31% (7 had a complete response [CR]) with a median DOR of 17
follow up was only 6.6 mo. In pts with available material, 14/36 (40%) had PD- mo; DCR was 49% and 24-mo OS rate was 40%. Pts in arm A had a mOS of 23 mo
L1 CPS .1 and median TMB was 4.4 mut/Mb (0-10.6). There was no cor- and 4 pts had a CR. The table shows additional efficacy results by arm. Overall, NIVO +
relation between PD-L1 status and PFS or OS. ERBB2 amplification was IPI was well tolerated; 37% of pts had a grade 3–4 treatment-related adverse event
evident by tissue-NGS in 17/29 (61%) and ctDNA-NGS in 17/30 (58%) pre- (TRAE; most common: pruritus and rash); 5% had grade 3–4 TRAEs leading to
treatment, while the remaining pts were ERBB2- by NGS likely due to tumor discontinuation. Conclusions: NIVO + IPI led to clinically meaningful responses and
heterogeneity or low tumor content. CtDNA maxVAF decreased in 16/24 tested had an acceptable safety profile in SOR-treated pts, with an ORR twice that of NIVO
pts after 1 cycle of induction T/P alone. irAEs included interstitial nephritis Gr4 mono (31% and 14%, respectively). Pts in arm A had the most promising mOS of 23
(3%), transaminitis Gr3 (11%), Gr4 (3%), colitis Gr3 (3%). Conclusions: Most mo. Clinical trial information: NCT01658878.
pts (51%) remain on therapy, and so the primary endpoint should be reached [A] NIVO1/IPI3 [B] NIVO3/IPI1 [C] NIVO3 Q2/IPI1
by 6/19. Updated survival, correlative studies and will be presented. These Q3W (n = 50) Q3W (n = 49) Q6W (n = 49)
promising preliminary safety and efficacy results led to initiation of a definitive ORR, n (%) 16 (32) 15 (31) 15 (31)
phase III Keynote 811 trial. Clinical trial information: NCT02954536. Complete response 4 (8) 3 (6) 0
Partial response 12 (24) 12 (24) 15 (31)
Stable disease 9 (18) 5 (10) 9 (18)
Progressive disease 20 (40) 24 (49) 21 (43)
DCR, % (95% CI) 54 (39–68) 43 (29–58) 49 (34–64)
mOS, mo (95% CI) 23 (9–NA) 12 (8–15) 13 (7–33)
12-mo OS rate, % (95% CI) 61 (46–73) 56 (41–69) 51 (36–64)
24-mo OS rate, % (95% CI) 48 (34–61) 30 (18–44) 42 (28–56)
NA, not available.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Noncolorectal) Cancer 225s

4014 Poster Discussion Session; Displayed in Poster Session (Board #119), 4015 Poster Discussion Session; Displayed in Poster Session (Board #120),
Mon, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Mon, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Mon, 3:00 PM-4:30 PM Mon, 3:00 PM-4:30 PM
Randomized phase II study of second-line modified FOLFIRI with PARP Final report of a phase I/II study of veliparib (Vel) in combination with 5-FU
inhibitor ABT-888 (Veliparib) (NSC-737664) versus FOLFIRI in metastatic and oxaliplatin (FOLFOX) in patients (pts) with metastatic pancreatic cancer
pancreatic cancer (mPC): SWOG S1513. First Author: E. Gabriela Chiorean, (mPDAC). First Author: Michael J. Pishvaian, Georgetown University,
University of Washington and Fred Hutchinson Cancer Research Center, Washington, DC
Seattle, WA
Background: 17 – 25% of mPDACs harbor DNA damage response (DDR) mutations,
Background: PC is characterized by DNA Damage Repair (DDR) deficiencies, the presence of which can be predictive of a response to platinum and PARP
including in BRCA1/2, ATM, and FANC genes. Given preclinical synergism inhibitor-based therapy. The PARP inhibitor, Vel is a potent sensitizing agent for, and
between veliparib with irinotecan, safety and preliminary efficacy, we has been safely combined with DNA-damaging chemotherapies. Methods: We
designed a randomized phase II study of mFOLFIRI (no 5-FU bolus) + veliparib initiated a Phase I/II trial of Vel + FOLFOX in pts with mPDAC. Pts received standard
vs FOLFIRI alone for 2nd line mPC patients (pts). Methods: Eligible pts had mFOLFOX6 except without the 5FU bolus, Q2 weeks. For the Phase I portion, a 3+3
mPC, adequate organ function, ECOG PS 0-1, and 1 prior non-irinotecan dose escalation of Vel identified a recommended Phase II dose of 200mg orally BID,
days 1-7, Q2 weeks. For the Phase II portion, we enrolled two cohorts: 1) Untreated
systemic therapy.143 pts were to be randomized (1:1) to veliparib vs control.
pts; 2) Previously treated pts. Also, for Phase II, pts were pre-selected if they had
Primary endpoint was overall survival (OS). All pts had blood and tumor bi- either a pathogenic germline or somatic DDR mutation (e.g. BRCA1/2, PALB2,
opsies at baseline to assess germline and somatic BRCA1/2 mutations (in- ATM), and/or a family history suggestive of a breast or ovarian cancer syndrome
tegrated), and homologous recombination (HR) or DDR biomarkers (labelled FH+). Objective response rate (ORR) was the primary objective of the Phase
(exploratory). Results: 123 pts were accrued between 09/2016 to 12/2017, II cohorts; key secondary endpoints were median progression-free survival (PFS) and
and 108 were included in this analysis. 117 pts were biomarker evaluable: overall survival (OS). Results: Between 01-2011 and 12-2018, 64 pts received
109 blood/106 tumors. 11 cancers (9%) had HR deficiency (HRD), including treatment, 31 in Phase I, and 15 untreated and 18 previously treated in Phase II. The
4 germline (BRCA1, BRCA2, ATM) and 7 somatic mutations (BRCA2, PALB2, combination was well tolerated, with the main Grade 3/4 AEs being myelosup-
ATM, CDK12). Additional 24 cancers (20%) had germline (n = 11, e.g., FANC, pression (16%) and nausea/vomiting (6%). Of the 64 pts, 55% were male; median
BLM, SLX4, CHEK2) or somatic mutations (n = 13, e.g., FANC, BLM, POLD1, age was 64; 95% had an ECOG PS of 1; 78% were platinum-naı̈ve; 69% were FH+;
RIF1, MSH2, MSH6) in other DNA repair genes, not classified as HRD. A and 27% had a known DDR mutation. 57 pts were evaluable for response, and the
planned interim futility analysis at 35% of expected PFS events determined ORR, PFS, and OS for the different pt subgroups are detailed below. The Phase II
the veliparib arm was unlikely to be superior to control. Most common grade 3/ cohorts achieved the primary endpoint of an ORR $ 25%. Most notably, plat-naı̈ve,
4 treatment related toxicities were neutropenia (33% vs 20%), fatigue (19% vs FH+, and DDR mutation+ pts had an ORR of 58%. Conclusions: The combination of
4%), and nausea (11% vs 4%), for veliparib vs control. Treatment exposure Vel + FOLFOX is safe, well tolerated, and shows promising efficacy particularly in
was similar for veliparib vs control: median 4 cycles (range 1-31 vs 1-32). plat-naı̈ve pts who are FH+ and/or harbor DDR mutations. A randomized trial to
assess the contribution of Vel to the regimen is warranted. Clinical trial information:
Median OS was 5.1 vs 5.9 mos (HR 1.3, 95%CI 0.9-2.0, p = 0.21), and
NCT01489865.
median PFS was 2.1 vs 2.9 mos (HR 1.5, 95%CI 1.0-2.2, p = 0.05) for
veliparib vs control arms, respectively. Correlations of gene mutations and Category (n) ORR (%) mPFS (mos) mOS (mos)
signatures with efficacy outcomes will be presented. Conclusions: Nearly 30% All pts (57) 26 3.7 8.5
of mPC pts had DNA repair gene abnormalities, including 9% with HRD. Plat-Naı̈ve (43) 33 5.3 9.4
Prior Plat (14) 7 1.9 5.0
Veliparib increased toxicity and did not improve OS when added to mFOLFIRI FH+ (43) 30 4.3 10.1
in biomarker unselected pts. BRCA1/2 and DDR biomarkers will be correlated No FH (14) 14 3.5 5.5
DDR mutation+ (16) 50 7.2 11.1
with efficacy to inform patient selection for future PARP inhibitor clinical trials. No mutation (41) 17 3.5 6.8
Clinical trial information: NCT02890355. Plat-Naı̈ve, FH+, DDR+ (12) 58 8.7 11.8

4016 Poster Discussion Session; Displayed in Poster Session (Board #121), 4017 Oral Abstract Session, Sun, 9:45 AM-12:45 PM
Mon, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, S-1 plus oxaliplatin versus S-1 plus cisplatin as first-line treatment for
Mon, 3:00 PM-4:30 PM advanced diffuse-type or mixed-type gastric/gastroesophageal junction
Rivaroxaban thromboprohylaxis in ambulatory patients with pancreatic adenocarcinoma: A randomized, phase 3 trial. First Author: Rui-hua Xu,
cancer: Results from a prespecified subgroup analysis of the CASSINI study. Sun Yat-Sen University Cancer Center, Guangzhou, China
First Author: Saroj Vadhan-Raj, The University of Texas MD Anderson Cancer
Background: Diffuse-type or mixed-type gastric adenocarcinoma is associated
Center, Department of Sarcoma Medical Oncology, Section of Cytokines and
with poor prognosis, and more effective treatment is needed. In Asia, S-1 plus
Supportive Oncology, Houston, TX
cisplatin (SP) is the standard first-line chemotherapy regimen for advanced
Background: Rivaroxaban thromboprophylaxis has been shown to reduce venous gastric cancer. Nevertheless, some clinical data suggested that oxaliplatin-
thromboembolism (VTE) on-treatment in ambulatory cancer patients in a recent based chemotherapy might be more efficacious and more tolerant than
randomized trial. Pancreatic cancer patients are at substantial risk for VTE; value of cisplatin-based chemotherapy. Methods: This trial is a multicenter, random-
thromboprophylaxis has not been definitively established. Methods: CASSINI was a ized, parallel-group, open-label, phase 3 trial in China. Patients aged 18-75
double-blind placebo-controlled trial of cancer patients initiating a new regimen, at years, with PS 0-2, adequate organ function, histology confirmed, unresect-
high risk for VTE (Khorana score $2), randomized to rivaroxaban 10 mg daily or able, advanced diffuse-type or mixed-type gastric adenocarcinoma/GEJA were
placebo up to 180 days. Patients were stratified by presence or absence of pan-
randomized 1:1 to S-1 plus oxaliplatin group (SOX) (S-1: 40-60mg bid on d1-
creatic cancer. Patients had screening ultrasound and blood drawn at baseline and
14, q3w; oxaliplatin: 130 mg/m2 on d1, q3w) or SP group (S-1: 40-60mg bid
every 8 wks. Primary efficacy endpoint was a composite of symptomatic DVT,
asymptomatic proximal DVT, any PE and VTE-related death. Primary safety endpoint
on d1-14, q3w; cisplatin: 60 mg/m2 d1, q3w). The primary endpoint was
was International Society on Thrombosis and Hemostasis (ISTH)-defined major overall survival (OS) in the full analysis set (FAS). The secondary endpoints
bleeding. Results: Of 1080 patients enrolled, 49 (4.5%) failed screening due to were progression-free survival (PFS), time to treatment failure (TTF) and safety.
baseline VTE, with even higher rates [24/362 (6.6%)] in patients with pancreatic Results: Between Jul 2013 and Jul 2018, 576 patients were randomized and
cancer. Of 841 randomized patients, 273 (32.6%) had pancreatic cancer with 558 initiated treatment (279 patients/group). The median number of che-
median age 66 y; 57% male and 155/273 (57% in each arm) completing the motherapy cycles was four in each group. In the FAS, the SOX group showed
double-blind period. During intervention (on-treatment) period, 5/135 (3.7%) improved OS (13.0 vs. 11.8 months, HR = 0.764, 95% CI: 0.636-0.918),
pancreatic cancer patients in the rivaroxaban arm and 14/138 (10.1%) in placebo PFS (5.7 vs. 4.9 months, HR = 0.752, 95%CI: 0.632-0.895), and TTF (5.2
arm had primary endpoint events [HR 0.35; 95%CI (0.13, 0.97), p = 0.03; number vs. 4.7 months, HR = 0.763, 95%CI: 0.641-0.909) compared with the SP
needed to treat, NNT = 16]. Major bleeding was not increased, occurring in 2 (1.5%) group. In terms of grade$3 adverse events, SOX showed lower occurrences of
patients in rivaroxaban arm and 3 (2.3%) in placebo arm. Further benefit with neutropenia (10.0% vs. 22.9%), leukopenia (9.7% vs. 21.9%), anemia (4.3%
rivaroxaban was observed when including primary and secondary endpoints (arterial/ vs. 14.3%),vomiting (3.9% vs. 10.4%), nausea (2.2% vs. 10.4%), anorexia
visceral events): 6/135 (4%) events in rivaroxaban vs 17/138 (12%) in placebo [HR, (2.2% vs. 6.8%), and febrile neutropenia (2.5% vs. 6.8%) than SP (all P ,
0.34; 95%CI (0.14, 0.87), P = 0.02; NNT = 13]. Correlative biomarker studies 0.05). The occurrence of grade#2 sensory neuropathy(41.6% vs. 12.2%, P ,
demonstrated significant decline in D-dimer values over time (weeks 8 and 16) in
0.001)was higher with SOX than with SP. Conclusion: Compared with SP, SOX
patients without VTE randomized to rivaroxaban prophylaxis compared to placebo
was more effective and less toxic (except neurosensory toxicity ) in patients
(P , 0.01), supporting clinical findings. Conclusions: Rivaroxaban substantially
reduced VTE in pancreatic cancer patients during intervention period. Given no
with previously untreated advanced diffuse-type or mixed-type gastric
increase in major bleeding, our findings suggest benefit to rivaroxaban thrombo- adenocarcinoma/GEJA. Clinical trial information: NCT01824459.
prophylaxis in pancreatic cancer patients initiating systemic therapy. Clinical trial
information: NCT02555878.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
226s Gastrointestinal (Noncolorectal) Cancer

4018 Poster Discussion Session; Displayed in Poster Session (Board #123), 4019 Poster Discussion Session; Displayed in Poster Session (Board #124),
Mon, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Mon, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session,
Mon, 3:00 PM-4:30 PM Mon, 3:00 PM-4:30 PM
PET-directed combined modality therapy for gastroesophageal junction cancer: Adjuvant chemotherapy versus perioperative chemotherapy (CTx) for resectable
First results of the prospective MEMORI trial. First Author: Sylvie Lorenzen, gastric signet ring cell (SRC) gastric cancer: A multicenter, randomized phase II
Third Department of Internal Medicine (Hematology/Medical Oncology), Kli- study (PRODIGE 19). First Author: Clarisse Eveno, CHRU Lille, Lille, France
nikum rechts der Isar, Technische Universitat Munchen, Munich, Germany
Background: The incidence of SRC gastric cancers is markedly increasing in
Background: We evaluated a PET-guided treatment stratification for improve- Western countries. SRC cancers may harbor intrinsic resistance to che-
ment in obtaining negative surgical margins (R0) in resectable gastroesophageal motherapy (CTx) leaving many clinicians unsure of the benefits of delaying
junction (GEJ) adenocarcinoma. According to sequential 18F-FDG PET, only surgery to pursue a neoadjuvant approach. The primary objective of this
40–50% of patients (pts) respond to neoadjuvant chemotherapy (CTX). Early study was to assess whether upfront surgery plus adjuvant CTx would provide
PET non-responders (P-NR) after induction CTX might benefit from changing to enough survival benefit for study in a phase III trial when compared to
chemoradiation (CRT). Methods: 75 pts with resectable GEJ adenocarcinomas perioperative CTx. Methods: Patients with stage IB-III SRC gastric cancer
were enrolled in this interventional, prospective, non-randomized multicenter were randomly assigned to receive upfront surgery plus adjuvant CTx (epi-
trial. Pts underwent baseline 18F-FDG PET scan followed by 1 cycle of CTX rubicin, cisplatin and 5-fluorouracil [ECF regimen], 6 cycles; experimental
(physicians’ choice, e.g. EOX, XP, mFOLFOX6). PET was repeated at day 14-21 arm [SurgFirst]) or perioperative CTx (ECF, 3 cycles before and 3 cycles after
and responders (P-R), defined as $ 35% decrease in SUVmax from baseline, surgery; control arm [CTxFirst]). Randomization (1:1) was stratified by tumor
continued with CTX. P-NR switched to CRT (41.4 Gy/23 fractions with weekly stage, tumor location, performance status and center. The primary endpoint
carboplatin/paclitaxel). Pts underwent surgery 4-6 weeks post-CTX/CRT. Pri- was overall survival (OS) at 2 years (OS2; target (H1): OS2 . 26%).
mary objective was an improvement of R0 resection rates in P-NR above a Results: 83 eligible patients were included in 27 centers from 11/12 to 09/
proportion of 70% based on results from the MUNICON1/2 trials. Secondary 16 (median age, 61 years (range: 32-80 years); male, 59%; ECOG PS 0-1,
endpoints include disease-free survival (DFS), overall survival (OS), measured 99%). Results were (CTxFirst/SurgFirst): full completion of CTx, 87%/77%;
from randomization to death from any cause, and translational endpoints. surgical resection, 82.5%/90%; major postoperative complications (Clavien
Results: Between 12/2014 and 07/ 2018 160 pts with resectable GEJ ade- Dindo III-IV), 24%/23%; R0 resection rate, 88%/78%; OS2, 60%/53.5%;
nocarcinomas were prospectively screened with PET in three German university and median OS, 39/28 months (exploratory hazard ratio, 0.71 [95%CI:
centers. Overall, 85 pts (53%) could not be included due to previously un- 0.40-2.64]). Conclusions: This trial met its primary endpoint (OS2 . 26%
detectable metastases (40/25%), no or too low FDG uptake of the primary tumor in the experimental arm). With OS2 rates . 50%, both CTx modalities
(21/13%), other reasons (24/15%). 75 eligible pts were enrolled in the study deserve further evaluation in Phase III studies in stage IB-III SRC gastric
and 69 were evaluable. Based on PET criteria, 47 (68%) and 22 (32%) were P-R cancer. Clinical trial information: NCT01717924.
and P-NR, respectively. R0 resection rates were 94% (44/47) for P-R and 91%
(20/22) for P-NR. Pathologic complete remission (pCR; , 10% vital tumor
cells), was 33% (15/46) in P-R and 55% (12/22) in P-NR. With a median follow-
up time of 19 months (mo), estimated 18 mo DFS was 71%/61% for P-R/P-NR,
respectively. Observed median 18 mo OS was 95% for P-R and 75% for P-NR.
Conclusions: Alternative CRT for GEJ adenocarcinoma improved R0- and pCR
rates among pts who were P-NR after induction CTX. PET response was
prognostic for a prolonged OS and DFS. Clinical trial information: 2014-
000860-16.

4020 Poster Discussion Session; Displayed in Poster Session (Board #125), 4021 Poster Session (Board #126), Mon, 8:00 AM-11:00 AM
Mon, 8:00 AM-11:00 AM, Discussed in Poster Discussion Session, Tumor mutational burden identifies chemorefractory gastric cancer with over-
Mon, 3:00 PM-4:30 PM all survival advantage after receiving toripalimab, a PD-1 antibody. First Au-
Randomized phase III trial of laparoscopy-assisted versus open distal gastrec- thor: Rui-Hua Xu, Sun Yat-sen University Cancer Center, Guangzhou, China
tomy with nodal dissection for clinical stage IA/IB gastric cancer (JCOG0912).
Background: Tumor mutational burden (TMB) is correlated with enhanced
First Author: Hitoshi Katai, National Cancer Center Hospital, Tokyo, Japan
objective response rate (ORR) and progression-free survival for certain cancers
Background: The number of patients undergoing laparoscopy-assisted distal receiving immunotherapy. This study aimed to investigate the safety and
gastrectomy (LADG) has been increasing worldwide. Several retrospective activity of toripalimab, a humanized PD-1 antibody, in advanced gastric cancer
studies have demonstrated equivalent survival after LADG compared to open (AGC), and the efficacy predictive value of biomarkers including TMB and PD-
distal gastrectomy (ODG). However, no confirmatory randomized controlled L1. Methods: This study was a part of phase Ib/II trial evaluating the safety and
trials has been published in a peer review journal to evaluate the efficacy of activity of toripalimab as a single agent therapy or in combination with che-
LADG compared with ODG, ensuring strict surgical skill and quality control of motherapy in chemo-refractory or treatment-naı̈ve AGC, esophageal squamous
surgery. We conducted phase III study to confirm that LADG is not inferior to cell carcinoma, nasopharyngeal carcinoma and head and neck squamous cell
ODG in efficacy. Methods: Eligibility criteria included histologically proven carcinoma. This report focused on the chemo-refractory AGC cohort receiving
adenocarcinoma in the middle or lower third of the stomach; clinical stage I toripalimab (3 mg/Kg d1, Q2W) as a single agent therapy. Primary endpoint
tumor (T1N0, T1N1, T2(MP)N0). Patients were preoperatively randomized was ORR. Biomarkers including tumor PD-L1 expression, TMB, microsatellite
to ODG or LADG. LADG was performed by accredited surgeon. The extent of instability (MSI) and Epstein-Barr virus (EBV) infection status were evaluated
nodal dissection was decided according to Japanese gastric cancer treat- for their correlation with clinical efficacy as preplanned. Tumor PD-L1 ex-
ment guidelines. The primary endpoint is relapse-free survival (RFS) and the pression was assessed with the SP142 immunohistochemistry assay, and the
secondary endpoints are overall survival (OS), short-term clinical outcomes, other biomarkers were assessed with whole exome sequencing based on tumor
and postoperative quality of life. Planned sample size was 920 patients in samples. Results: There were 58 subjects included in this cohort. The ORR
total, which was determined with at least 80% power, a one-sided alpha of was 12.1% and the disease control rate was 39.7%. Only 1 subject was MSI-H
5%, and a non-inferiority margin for a hazard ratio of 1.54. Before the 1st and achieved partial response. One out of 4 EBV positive subjects achieved
interim analysis, the primary endpoint was amended from OS to RFS in 2015 partial response. Significant higher ORR was observed in subjects with positive
because the surrogacy of RFS for OS was demonstrated and the predicted PD-L1 expression (ORR 37.5%, 3/8) or TMB $12 Mutations/Mb (ORR
number of events for OS was smaller than expected. Results: A total of 921 33.3%, 4/8) than those with negative PD-L1 expression (ORR 8.5%) or TMB ,
patients were randomized (ODG 459, LADG 462) between Mar. 2010 and 12 Mutations/Mb (ORR 7.0%). The TMB-high subgroup showed significant
Nov. 2013. Among 921 patients, 912 patients (99%) underwent assigned superior OS than the TMB-low subgroup (HR = 0.48 [96% CI 0.24 to 0.96],
surgery. Conversion to ODG was needed for 16 patients (3.5%) in LADG arm p = 0.038), while PD-L1 expression status failed to differentiate OS.
mainly due to advanced disease. 5-year RFS was 94.0% (95% CI: 91.4- Conclusions: Toripalimab demonstrated promising anti-tumor activity in
95.9%) in ODG and 95.1% (92.7-96.8%) in LADG. LADG was non-inferior chemo-refractory AGC patients. TMB might serve as a better predictive marker
to ODG for RFS. (HR: 0.84 [90% CI: 0.56-1.27 ( , 1.54)], p for non- for OS than PD-L1 expression for chemo-refractory AGC patients receiving
inferiority = 0.008). 5-year OS was 95.2% (92.7-96.8) in ODG and 97.0% PD-1 blockade immunotherapy. Clinical trial information: NCT02915432.
(94.9-98.2) in LADG (HR: 0.83 [95% CI: 0.49-1.40]). Conclusions: The
non-inferiority of LADG to ODG in RFS was confirmed. LADG has been
established as one of the standard treatments for clinical stage I gastric
cancer. Clinical trial information: UMIN000003319.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Noncolorectal) Cancer 227s

4022 Poster Session (Board #127), Mon, 8:00 AM-11:00 AM 4023 Poster Session (Board #128), Mon, 8:00 AM-11:00 AM
Impact of age and sex on chemotherapy (CTx) efficacy, toxicity and survival FOLFIRI plus ramucirumab versus paclitaxel plus ramucirumab for patients
in early oesophagogastric (OG) cancer: A pooled analysis of 3265 patients from with advanced or metastatic adenocarcinoma of the stomach or gastro-
four large randomised trials (OE02, OE05, MAGIC & ST03). First Author: Avani esophageal junction as second-line therapy: Interim safety and efficacy
Athauda, The Royal Marsden Hospital, Surrey, United Kingdom results from the phase II RAMIRIS Study (AIO-STO-0415) of the German
Gastric Group at AIO. First Author: Sylvie Lorenzen, Klinikum Rechts der Isar
Background: No large scale randomised data exists evaluating the impact of
der TU München, Munich, Germany
age and sex in patients (pts) undergoing potentially curative surgery and CTx for
OG cancer. However, differences in age and sex may be contributing factors to Background: Ramucirumab as monotherapy and in combination with pacli-
variability in CTx dose-response and toxicity which could also impact survival. taxel is a proven second-line option for advanced gastroesophageal adeno-
Methods: Data from four prospective randomised controlled trials were pooled carcinoma (GEA). More and more patients (pts) are pretreated with docetaxel in
using a two-stage meta-analysis. For survival data, hazard ratios were calcu- the perioperative or first-line setting. For those pts, the benefit of a combination
lated for pts ,70 and $70 years and between males and females. Pts were of ramucirumab and paclitaxel is unclear, and physicians would choose an
allocated to receive neoadjuvant platinum and fluoropyrimidine +/- anthra- irinotecan-based regimen as second line treatment. This provides a rationale
cycline and bevacizumab. Mandard tumour regression grade (TRG) and for the evaluation of FOLFIRI + ramucirumab. Methods: This is a multicenter,
prevalence of $G3 toxicities were compared according to the same subgroups randomized, investigator initiated, phase II trial, planned to include 111 pts
using Chi-squared test. Results: 3265 pts were included for survival analysis with advanced GEA to receive 2:1 either FOLFIRI plus ramucirumab every two
(2668 (82%) M, 597 (18%) F; 2626 (80%) ,70, 639 (20%) $70). A weeks (Arm A) or paclitaxel (days 1, 8, 15 of a 28-day cycle) plus ramucirumab
significant improvement in disease specific survival (DSS) (HR 0.78; every two weeks (Arm B). Primary endpoint is 6-months OS rate. This abstract
p,0.001) and OS (HR 0.78; p,0.001) was observed in females vs males. displays interim results of safety and overall objective response (ORR) in
Although OS was worse in older vs younger pts (HR 1.15; p=0.01) no sig- docetaxel pre-treated group from up to 65 randomized pts. The results were
nificant difference in DSS was observed (HR 1.04; p=0.52). For those pts who needed to decide on conducting a subsequent phase III study. Results: 58 (A,
underwent resection following neoadjuvant CTx, older patients (19 vs 13%; 36; B, 22) pts were included in the safety analysis and 50 pts with tumor
p=0.01) and female patients (19% vs 13%, p=0.02) were more likely to assessment in the response analysis. Main $ grade 3 adverse events were
achieve more favourable Mandard TRG 1&2 scores. Older pts experienced respectively in arms A/B: neutropenia (20%/22%), fatigue (6%/0%), diarrhea
significantly more $G3 neutropaenia (30 vs 22%; p=0.004). Females ex- (8%/3%), and related SAEs (14% v 23%). Twenty-nine of 50 pts (58%) were
perienced significantly more $G3 nausea (12 vs 7%; p=0.006), vomiting (10 pre-treated with docetaxel. In these pts, ORR was 30% in Arm A (5/17) and 8%
vs 5%; p#0.001) and diarrhoea (9 vs 4%; p=0.001). Conclusions: This study (1/12) in Arm B. Disease control rate (DCR) was 65% and 50% for Arm A and B
represents the largest pooled analysis of age and sex differences on safety of respectively. Conclusions: The interim safety analysis of the RAMIRIS trial has
neoadjuvant CTx and survival in early OG cancer. Females had significantly demonstrated feasibility of the combination of FOLFIRI and ramucirumab.
improved survival while experiencing more GI toxicities. Older pts achieved Docetaxel pre-treated pts had higher ORR and DCR when ramucirumab is
comparable DSS and thus, dependent on fitness, should be offered the same combined with FOLFIRI, instead of paclitaxel. EudraCT: 2015-005171-24.
treatment paradigm as younger pts. Clinical trial information: NCT03081143.

4024 Poster Session (Board #129), Mon, 8:00 AM-11:00 AM 4025 Poster Session (Board #130), Mon, 8:00 AM-11:00 AM
Tumor mutation burden and immunogenicity in gastric cancer with HER2 Treatment patterns and outcomes in Chinese gastric cancer by HER2 status:
alterations. First Author: Bo Wei, The Third affiliated hospital, Sun Yat-Sen A non-interventional registry study (EVIDENCE). First Author: Shukui Qin,
University, Guangzhou, China PLA Cancer Center of Bayi Hospital Affiliated to Nanjing University of
Chinese Medicine, Nanjing, China
Background: Human epidermal growth receptor 2 (HER2) is considered as
an oncogenic driver gene in gastric cancer (GC). Immunotherapy has been Background: Gastric cancer (GC) is the second leading cause of cancer-related
proven to be effective in GC patients. Previous studies indicated that patients deaths in China. Trastuzumab (TRA) has been used to treat HER2+ metastatic
harboring driver mutations were considered as poor candidate for immu- gastric cancer (mGC) in China since 2012. However, real-world data on ef-
notherapy. But the efficacy of immunotherapy for HER2 positive GC has not fectiveness and safety in Chinese patients are limited. Methods: This pro-
been defined. We therefore analyzed the immunogenicity of HER2 alter- spective, multicenter (85 hospitals), real-world noninterventional registry
ations in GC. Methods: Genomic profiling of DNA from 448 GC was per- study evaluated the effectiveness and safety of TRA in five cohorts of Chinese
formed using next-generation sequencing on 381 cancer associated genes. GC patients with different HER2 statuses from April 2013 to June 2018.
The expression of PD-L1 protein was evaluated in 192 GC with the use of an Effectiveness analysis was conducted in three cohorts: Cohort I (HER2+ mGC
automated immunohistochemical assay (Ventana, SP263). Whole-exome with TRA), Cohort II (HER2+ mGC untreated with TRA) and Cohort IV (HER22
sequencing, copy number variations, RNA-seq and clinical data of 443 GC mGC untreated with TRA). Safety outcomes of TRA-related adverse events
from The Cancer Genome Altas (TCGA) were also analyzed to further evaluate (AEs) were analyzed in Cohort I. Results: Cohorts I, II and IV included 709
the immunogenicity of HER2 alterations. TMB was defined as number of patients (174, 113 and 422, respectively; mean age 57.8 years; 72% male);
somatic non-synonymous mutations in coding regions. HER2 amplification 64.9% of patients were ECOG 0–1, 93.7% had a primary GC tumor and 42.3%
in TCGA was defined as “2” derived from the copy-number analysis algo- were at stage T4. Progressive disease was the cause of death in 32.8%, 27.4%
rithms GISTIC. Results: HER2 alterations including amplification, missense and 29.9% in Cohorts I, II and IV, respectively. Respective mean duration of
and fusion were present in 19.2% (85/443) of TCGA cohort and 11.4% (51/ follow-up was 422.5, 287.5 and 277.5 days. Median overall survival (OS) was
448) of clinical cohort. 14.0% (62/443) of TCGA cohort and 6.0% (27/448) 22.3, 17.2 and 17.4 months, respectively. After excluding patients who had
of clinical cohort harbored HER2 amplification. Higher TMB was observed in surgery, the respective median OS was 19.9, 15.3, and 12.6 months. For the
MSS/MSI-L patients carrying any HER2 alterations in TCGA cohort (P = first-line treatment, the median OS in Cohort I was 22.1 months, and the
0.009). On the contrary, HER2 alterations did not show a higher neoantigen median progression free survival (PFS) was 8.2, 6.9 and 6.2 months in Cohorts
level and HER2 alterations were associated with decreased immunogenicity I, II and IV, respectively. Response rates (RR) for first-line treatment in Cohorts
in terms of immune-related gene mRNA expression and immune infiltrates. I, II and IV were 51.7%, 18.4% and 32.8%, respectively. After propensity
In clinical cohort, HER2 alterations was also significantly associated with score matching, OS, PFS and RR were all significantly better in Cohort I versus
higher TMB (P = 0.001). Meanwhile, a trend of shrinking proportion of PD- II (all P,0.05). The most common regimen, TRA+XELOX (capecitabine+
L1 expression was observed in HER2 alteration subgroup (2/23, 8.7%) than oxaliplatin), was estimated to have the longest median OS at 34.6 months.
HER2 wild-type subgroup (47/169, 27.8%, P = 0.071). Furthermore, HER2 Grade $3 AEs were reported in 33.9% (59/174) of patients in Cohort I; anemia
amplification had significant positive associations with HRD score (P = was the most common AE (12.1%). Conclusions: TRA improved OS and PFS in
0.0169) in TCGA cohort which indicated an increased degree of genome Chinese HER2+ mGC patients compared with chemotherapy alone and
instability. Conclusions: Although HER2 alterations in GC were associated was well tolerated and effective when combined with a range of other
with increased TMB, HER2 alterations exhibited poor immunogenicity. therapies in a real-world setting. Clinical trial information: NCT01839500.
These findings indicated that HER2 alterations might confer resistance to
immune monotherapy.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
228s Gastrointestinal (Noncolorectal) Cancer

4026 Poster Session (Board #131), Mon, 8:00 AM-11:00 AM 4027 Poster Session (Board #132), Mon, 8:00 AM-11:00 AM
A phase II study of S-1, oxaliplatin, and nab-paclitaxel, and itraconazole A phase II trial of preoperative chemoradiotherapy and pembrolizumab for
aimed at conversion surgery for advanced and recurrent gastric cancer. First locally advanced esophageal squamous cell carcinoma (ESCC). First Author:
Author: Yoshihiko Nakamoto, Meiwa Hospital, Nishinomiya, Japan Min Hee Hong, Division of Medical Oncology, Department of Internal
Medicine, Yonsei Cancer Center, Yonsei University College of Medicine,
Background: Preclinical and clinical studies demonstrated that itraconazole, a
Seoul, South Korea
common anti-fungal agent, has anticancer activity. The purpose of this study
was to evaluate the efficacy of the chemotherapy with itraconazole on unre- Background: Even though preoperative chemoradiotherapy (CRT) showed
sectable, metastatic, and recurrent gastric cancer. Methods: All patients were survival improvement in patients with resectable ESCC in a randomized trial
referred to our clinic with a clinical diagnosis of unresectable gastric cancer. The over upfront surgery, ESCC still has a dismal prognosis. With the potential
regimen consisted of 160 mg/m2 nab-paclitaxel IV on day 1, 100 mg/m2 benefit of combining PD-1 blockade to CRT, we conducted a phase II trial
oxaliplatin IV on day 1, 60 mg/m2 S-1 orally on days 1-7, and 400mg itra- which assessed the efficacy, feasibility, and safety of the combination of
conazole orally on days -1 to 3, repeated every 3 weeks. Conversion surgery was preoperative CRT and pembrolizumab (PEM) in ESCC. Methods: Patients (pts)
allowed. The primary endpoint was overall survival (OS). Results: Between with histologically confirmed ESCC (clinical stage Ib to III according to the
2015 and 2018. 23 patients were enrolled. Their median age was 68 years American Joint Committee on Cancer 7th staging system) were enrolled. Pts
(range 40-80 years); stomach/gastroesophageal junction: 21/2; Stage IIIA/IIIB/ received concurrent neoadjuvant chemotherapy (weekly paclitaxel and car-
IV: 2/1/20. Among 10 patients who had liver metastases, 2 had simultaneous boplatin), radiotherapy (44.1 Gy in 21 fractions), and PEM (every 3 week,
lung metastases. Nine patients had peritoneal dissemination. Five patients with 200 mg) during 5 weeks followed by surgery. After surgery, pts were treated
stage IV had recurrent disease after primary surgery followed by adjuvant S-1. with PEM during 2 years or until progression, unacceptable toxicity, death, or
The other 18 patients had no history of surgery or chemotherapy. Response rate pts’ refusal, which came first. The primary endpoint was pathologic complete
was 70% (CR/PR: 2/14). Among 12 patients (67%) who had conversion response (pCR) rate in the primary tumor and secondary endpoints were overall
surgery, R0 resection was conducted in 8 and no residual tumor was observed in survival (OS), disease-free survival (DFS), the incidence of adverse events, and
2. Among enrolled 23 patients, median OS was 22 months (95%CI: . etc. Results: In a total of 28 enrolled pts (median age 60), 26 pts received
12 months) and 1-year OS rate was 81.8% (95%CI: 46.7%―95.5%). Grade 3/ esophagectomy. Two pts did not undergo surgery due to death (hematemesis)
4 neutropenia in 5 (22%), no grade 3/4 thorombocytopenia, grade 2 peripheral and consent withdrawal. There were two in-hospital mortality cases after
sensory neuropathy in 6 (26%). Conclusions: The addition of itraconazole to surgery, which were resulted from acute lung injury. The pCR in primary tumor
chemotherapy showed promising efficacy with high conversion surgery rate and was achieved in 46.1% of pts who underwent resection (95% CI: 28.8 – 64.6).
with acceptable toxicities. Clinical trial information: UMIN000021340. With a median follow-up of 11.7 months, median OS was not reached. Six-
month and 12-month OS rates were 89.3% and 82.1%, respectively. There
was a trend toward better DFS in the pCR group (n = 12) compared with the
non-pCR group (n = 14) (HR = 0.33, p = 0.1). Most common treatment-related
adverse events were neutropenia (50.0%) and liver enzyme elevation (30.8%)
in the neoadjuvant and adjuvant period, respectively. Conclusions: The ad-
dition of PEM to preoperative CRT in ESCC demonstrated promising efficacy
with acceptable toxicity. Based on the results, further investigation is war-
ranted in a phase III clinical trial. The exploratory endpoints including bio-
markers analyses are ongoing. Clinical trial information: NCT02844075.

4028 Poster Session (Board #133), Mon, 8:00 AM-11:00 AM 4029 Poster Session (Board #134), Mon, 8:00 AM-11:00 AM
Perioperative chemotherapy alone versus preoperative chemoradiotherapy Clinicopathological features of Epstein–Barr virus associated gastric carci-
for locally advanced distal esophageal and gastroesophageal junction cancer: noma with submucosal invasion. First Author: Hiroki Osumi, Department of
A 10-year review of the British Columbia (BC) Cancer Registry. First Author: Gastroenterology, Cancer Institute Hospital, Japanese Foundation for
Shiru Lucy Liu, BC Cancer, Vancouver, BC, Canada Cancer Research, Tokyo, Japan
Background: The optimal treatment strategy for resectable cancer of the distal Background: The incidence of lymph node metastasis (LNM) in pathological
esophagus (ESOPH) and gastroesophageal junction (GEJ) remains contro- T1b (pT1b) gastric cancer (GC) is around 20% and the majority of them have
versial. This study evaluates patterns of practice in BC, rates of complete no LNM. The Cancer Genome Atlas Research Network proposed the concept of
surgical resection, and survival outcomes of patients treated with perioperative molecular phenotype classifying GC into 4 phenotypes including Epstein-Barr
chemotherapy alone (CA), per MAGIC or FLOT4 protocol, versus preoperative virus-CIMP (EBV). EBV positive gastric cancer (EBVGC) is associated with a low
chemoradiotherapy (CRT), per CROSS protocol. Methods: We undertook a prevalence of LNM; however, EBV status is not considered in the present
provincial analysis of initially resectable, locally advanced, cancer of the indication of endoscopic resection (ER). We aimed to clarify the implication of
ESOPH and/or GEJ who underwent surgery in BC, from 2008 to 2018. EBV status for ER of pT1b GC. Methods: Consecutive cases of pT1b GCs
Baseline patient, tumor, treatment, and clinical outcome data were collected treated with curative surgery between 2005 and 2014 were retrospectively
from the BC Cancer Registry. Kaplan-Meier survival and multivariate regression analyzed. Tissue microarray was made and EBV-encoded RNA in situ hy-
analyses were conducted. Results: Among 575 patients, 468 underwent bridization was performed for evaluation of EBV status. Clinicopathological
surgery and were included (Table). More surgeries were aborted intra- factors and LNM status were compared between EBVGC and non-EBVGC
operatively in the CA cohort compared to CRT (12% vs 2%, p,0.001). There groups. Results: Among the 1221 pT1b GCs that underwent gastrectomy with
was no difference in age, sex, or ECOG performance status among the cohorts, regional lymph node dissection, 898 pT1bGCs were eligible in this study.
and 83% were adenocarcinoma. While 82% of ESOPH involving GEJ (N = EBVGC accounted for 7.9% (71 of 898) cases. Compared to non-EBVGC,
251, 54%) is treated with CRT, only 53% of GEJ alone (N=217, 46%) is EBVGC was more frequent in males (p = 0.0055), the upper third region (p ,
treated with CRT (p,0.001). CRT is associated with a higher rate of complete 0.0001), showed elevated growth features (p = 0.0059), and was associated
or partial pathologic response compared to CA (59% vs 39%, p=0.002). R0 with a lower frequency of accompanying ulceration (p = 0.002), greater depth
resection rate was 90% and 94% in the CA and CRT cohort, respectively of submucosal invasion (p = 0.017), and lower frequency of lymphatic invasion
(p=0.383). There is no statistically significant difference in overall survival, (p , 0.0001). Frequency of LNM was significantly lower in EBVGC than in
with medians of 29.6 and 26.0 months for patients treated with CA and CRT, non-EBVGC (4.2% vs. 21.9%, p , 0.0001). In EBVGC, tumors without
respectively (p=0.723). Cancer-specific survival is also not significantly dif- lymphovascular invasion showed significantly lower frequency of LNM than
ferent (p=0.565). In the CA cohort, 37% of patients complete all 8 cycles of those with lymphovascular invasion (0 of 50, 0%; vs 3 of 21, 14.3%; p =
FLOT and 52% of patients complete all 6 cycles of MAGIC (p=0.396). 0.023). Histologically, 84.5% (60 of 71) of EBVGC included carcinomas with
Conclusions: Patients treated with CRT have higher rates of complete resection lymphoid stroma and/or lace pattern components. Conclusions: pT1b EBVGC
and pathologic response, but their survival is not significantly different is a convincing candidate for ER, regardless of risk factors other than lym-
compared to those treated with CA. phovascular invasion.
Total Surgery attempted Surgery completed Pathologic response rate
CRT 420 322 316 156
CA 155 146 128 47
Total 575 468 444 203

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Noncolorectal) Cancer 229s

4030 Poster Session (Board #135), Mon, 8:00 AM-11:00 AM 4031 Poster Session (Board #136), Mon, 8:00 AM-11:00 AM
A phase Ib study of IMU-131 HER2/neu peptide vaccine plus chemotherapy Camrelizumab combined with capecitabine and oxaliplatin followed by
in patients with HER2/neu overexpressing metastatic or advanced adeno- camrelizumab and apatinib as first-line therapy for advanced or metastatic
carcinoma of the stomach or gastroesophageal junction. First Author: Yee gastric or gastroesophageal junction cancer: Updated results from a multi-
Chao, Taipei Veterans General Hospital, Taipei, Taiwan center, open label phase II trial. First Author: Lin Shen, Beijing Cancer
Hospital, Beijing, China
Background: Gastric cancer is the 5th most common cancer and the 3rd
leading cause of cancer deaths. HER2/neu is overexpressed in 15% - 25% of Background: Capecitabine plus oxaliplatin (CAPOX) is one of the standard
patients with gastric cancer. Monoclonal antibodies against HER2/neu are first-line treatments for advanced or metastatic gastric cancer. Camrelizumab
effective but alternatives are needed due to cost and global availability. IMU- (SHR-1210, an anti–PD-1 antibody) shows promising anti-tumor activity in
131 is a B-cell peptide vaccine composed of a fusion of 3 epitopes from the patients (pts) with advanced or metastatic gastric or gastroesophageal junction
extracellular domain of HER2/neu conjugated to CRM197 with the adjuvant (G/GEJ) cancer. Camrelizumab combined with CAPOX for untreated G/GEJ
Montanide. Polyclonal antibodies against IMU-131 peptides elicit antitumor cancer was assessed as a part of an ongoing multicenter, open-label phase 2
activity in vitro and a phase I study demonstrated safety and immunogenicity in trial (cohort 1), and encouraging preliminary results were reported. Here, we
Her-2 +/++ breast cancer patients. Methods: IMU-131 was given to patients present the updated safety and efficacy data. Methods: In this cohort, systemic
with HER2/neu overexpressing gastric or gastroesophageal junction (GEJ) treatment naı̈ve pts with HER2– advanced or metastatic G/GEJ adenocarci-
adenocarcinoma in an international open-label Phase 1b dose escalation trial noma were given camrelizumab 200 mg on Day 1, capecitabine 1000 mg/m2
performed in 14 Asian and Eastern European sites assessing safety, tolera- bid on Days 1–14 and oxaliplatin 130 mg/m2 on Day 1 of each 21-day-cycle for
bility, and immunogenicity. Each patient received IMU-131 on Days 0, 14, 4 to 6 cycles followed by camrelizumab 200 mg every 3 weeks plus apatinib
and 35, accompanied by cisplatin and 5-fluorouracil or capecitabine every 375 mg qd until disease progression or intolerable toxicity. The primary
21 days. Results: 14 patients were enrolled with advanced stage IIIb or IV with endpoint was objective response rate. Results: At data cutoff (Jan 20, 2019),
10 HER2 overexpressing tumors (7 x HER2+++, 3 x HER2++ FISH positive) 43 of the 48 enrolled pts were evaluable. Partial response was observed in 28
and 4 HER2++ expressing tumors. Mean age was 57 yo (range of 21 - 79) with pts (65%), and 19 (44%) were confirmed. Stable disease in 14 pts and
ECOG scores of 0 or 1 in 7 patients each. There were 9 Asian and 5 Caucasian progressive disease in 10 pts were reported. Median estimates for duration of
patients with 5 females and 9 males. Dose levels were 0.1, 0.3 and 0.5 mg response and progression-free survival were not reached. Grade $3 treatment-
with 3, 6, and 5 patients receiving those dose levels each. 11 patients received related adverse events (TRAEs) occurred in 9 pts (21%), included neutropenia,
all 3 doses with 3 patients who received only 2 doses due to disease pro- diarrhea, rash and elevated ALT, whereas none of the TRAEs was fatal. Ten pts
gression and 2 patients received a dose on day 182. Of the 14 patients dosed without progression after 4–6 cycles of camrelizumab and CAPOX combination
11 were evaluable for tumor progression at day 56 and later. Of those patients, therapy all received camrelizumab plus apatinib as sequential therapy, and no
the best response was 1 CR, 4 PR,5 SD and 1 PD. In the 0.1 mg dose group the new safety signals were observed. Conclusions: The updated results confirmed
best response was 1 CR and 2 SD, with 2 PR, 2 SD and 1 PD in the 0.3 mg that camrelizumab plus CAPOX followed by camrelizumab plus apatinib was
group and 2 PR and 1 SD in the 0.5 mg group. In patients with HER2 well tolerated with noteworthy responses as first-line therapy in advanced or
overexpression there was 1 CR, 4 PR, 2 SD and 1 PD, and in patients with metastatic G/GEJ cancer pts. Expansion of this cohort in a phase 3 study are
HER2++ expression there was 3 SD. There were no SAEs related to IMU-131 under way. Clinical trial information: NCT03472365.
and 1 patient had a mild injection site reaction. Conclusions: IMU-131 is a
promising B-Cell vaccine against HER2. Further work in a controlled phase 2
trial is ongoing. Clinical trial information: NCT02795988.

4032 Poster Session (Board #137), Mon, 8:00 AM-11:00 AM 4033 Poster Session (Board #138), Mon, 8:00 AM-11:00 AM
Pembrolizumab in previously treated metastatic esophageal cancer: Longer Phase 2 study of camrelizumab (anti-PD-1 antibody) combined with apatinib
term follow-up from the phase 2 KEYNOTE-180 Study. First Author: Ken and chemotherapy for the first-line treatment of advanced esophageal
Kato, Department of Gastrointestinal Medical Oncology, National Cancer squamous cell carcinoma. First Author: Bo Zhang, Department of Medical
Center Hospital, Tokyo, Japan Oncology, National Cancer Center/National Clinical Research Center for
Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking
Background: In the phase 2, open-label, KEYNOTE-180 (NCT02559687)
Union Medical College, Beijing, China
study, after a median follow-up of 5.8 months, pembrolizumab (pembro)
provided antitumor activity with durable responses in pts with previously Background: Both anti-PD-1 antibodies and molecular antiangiogenic agents
treated, advanced/metastatic adenocarcinoma (EAC) including Siewert type 1 have shown promising anti-tumor activities in patients with advanced
adenocarcinoma of the gastroesophageal junction or squamous cell carcinoma esophageal cancer. We conducted this single-center phase 2 study to evaluate
(ESCC) of the esophagus. Here we present results of an additional 10 months the efficacy and safety of camrelizumab (anti-PD-1 antibody) plus apatinib
of follow-up. Methods: Eligible pts with metastatic esophageal cancer, $2 (VEGFR2-TKI) in combination with liposomal paclitaxel and nedaplatin in the
prior lines of therapy, and tumor samples evaluable for biomarker expression, first-line treatment of patients with esophageal squamous cell carcinoma
received pembro 200 mg Q3W for up to 2 years, or until disease progression, (ESCC). Methods: Patients with unresectable locally advanced or metastatic
unacceptable toxicity, or withdrawal. Tumor response was assessed Q9W ESCC received camrelizumab 200mg d1, liposomal paclitaxel 150mg/m2 d1,
(RECISTv1.1, central review). PD-L1+ pts had combined positive score $10 nedaplatin 50mg/m2 d1 and apatinib 250mg d1-14. Treatments were re-
using IHC (22C3 antibody). Primary endpoint was objective response rate peated every 14 days for up to 6-9 cycles, followed by maintenance therapy
(ORR). Secondary endpoints included safety, DOR, PFS, and OS. Results: Of with camrelizumab, apatinib, or both. The primary end point was progression-
121 pts enrolled, 63 (52%) had ESCC and 58 (48%) had PD-L1+ (combined free survival (PFS) in the intention-to-treat population. Secondary end points
positive score $10) tumors. As of July 30, 2018, median follow-up duration, included objective response rate (ORR), disease control rate (DCR), overall
from randomization to data cutoff, was 5.8 mo (range, 0.2 mo to 27.8+ mo). survival (OS) and safety. PD-L1 positivity, defined as a combined positive score
ORR (CR+PR) was 10% (95% CI, 5%-17%); 2 (2%) CR,10 (8%) PR, 25 (CPS) $1, was evaluated by immunohistochemistry (IHC). Results: Between
(21%) SD. Median DOR was not reached ([NR] range, 2.1 mo to 25.1+ mo). Aug 6th 2018 and Feb 6th 2019, a total of 29 patients were enrolled. The
Median PFS was 2 mo (95% CI, 1.9%-2.1%) with 9-mo PFS rate of 9%. median age was 62 years (43-70). Most patients were male (22/29, 75.9%)
Median OS was 5.8 mo (4.5-7.2) with 12 mo OS rate of 27%. In ESCC, ORR with metastatic disease (25/29, 86.2%). Response evaluation by independent
was 14% (95% CI, 7%-25%); 2 (3%) CR, 7 (11%) PR, with median DOR NR central review was available in 26 patients, with 19 achieving a best response
(range, 4.2 mo to 25.1+ mo). In EAC, ORR was 5% (95% CI, 1-14); 3 PR, with of PR, 6 with SD, and 1 with PD. The ORR and DCR were 73.1% (19/26) and
median DOR NR (range, 2.1 mo to 15.6+ mo). In PD-L1+ pts, ORR was 14% 96.2% (25/26), respectively. Data for PFS and OS were not matured. The most
(95% CI, 6%-25%); 1 (2%) CR, 7 (12%) PR with median DOR NR (range, 4.2 common grade 3/4 adverse events were leucopenia (21/29, 72.4%) and
mo to 25.1+ mo). In PD-L1- pts ORR was 6% (95% CI, 2%-16%); 1 (2%) CR, neutropenia (15/29, 51.7%). Two cases of treatment-related SAEs occurred,
3 (5%) PR; median DOR NR (range, 2.1 mo to 17.3+ mo). Overall, 19 (16%) both led to hospitalization: one patient developed grade 3 febrile neutropenia,
pts had treatment-related grade 3-5 AEs. Seven (6%) pts discontinued due to a grade 4 leucopenia and grade 3 anorexia; another patient developed grade 4
treatment-related AE. There was one treatment-related death from pneumo- toxic epidermal necrolysis. Conclusions: Camrelizumab plus apatinib in
nitis. Conclusions: Pembro continued to provide durable clinical benefit combination with liposomal paclitaxel and nedaplatin could be a new treat-
with a manageable safety profile for pts with heavily pretreated esophageal ment option for patients with unresectable locally advanced or metastatic
cancer, with conversions of PR to CR observed. Clinical trial information: ESCC. Clinical trial information: NCT03603756.
NCT02559687.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
230s Gastrointestinal (Noncolorectal) Cancer

4034 Poster Session (Board #139), Mon, 8:00 AM-11:00 AM 4035 Poster Session (Board #140), Mon, 8:00 AM-11:00 AM
Prognostic value of serum soluble programmed death-ligand 1 (sPDL1) and POF (paclitaxel plus FOLFOX) versus IP PAC (intraperitoneal paclitaxel plus
dynamics during chemotherapy in advanced gastric cancer patients. First FOLFOX) versus FOLFOX as a first-line treatment in advanced gastric cancer
Author: Woochan Park, Seoul National University Hospital, Seoul, South Korea (AGC): Update from a multicenter, randomized phase II trial, FNF-004 trial.
First Author: Rongbo Lin, Gastrointestinal Medical Oncology, Fujian Cancer
Background: The soluble form Programmed Death-Ligand 1(sPDL1) is sug-
Hospital, Fuzhou, China
gested to have immunosuppressive activity and under investigation as candi-
date biomarker for immuno-oncology drug development. In this study, we Background: The PFS with POF was statistically significantly improved and IP PAC
measured the serum sPDL1 at pre-and post-chemotherapy and evaluated its was trending to improve compared to FOLFOX in first-line setting in AGC were
prognostic implication and dynamics during chemotherapy in advanced gastric reported in 2019 ASCO-GI (abstract 6). Update and subgroup analysis were released
cancer (GC). Methods: We prospectively enrolled 68 GC patients who were herein. Methods: The patients with AGC were randomized to three groups. The POF
candidates for palliative standard 1st-line chemotherapy, and blood was serially or IP PAC was paclitaxel 135 mg/m2intravenously (POF) or paclitaxel 80 mg/m2
collected at pre-and post-one cycle of chemotherapy, at best response and intraperitoneally (IP PAC) followed by mFOLFOX6 omitted 5-Fu bolus. Every 14 days
repeated for all three regimens. Up to 9 cycles of treatment were administered,
disease progression. sPDL1 was measured using an enzyme-linked immuno-
followed by S-1 until disease progression. The primary endpoint was PFS.
sorbent assay. Response to chemotherapy, overall survival (OS), progression-
Results: Between Nov 2015 and May 2018, 89 pts (30 POF, 29 IP PAC, 30
free survival (PFS) and other prognostic factors including neutrophil-lymphocyte FOLFOX) were randomly allocated. PFS, OS and RR were seen in the table below.
ratio (NLR) were obtained. The cut-off values of sPDL1 levels and changes for Either POF or IP PAC was statistically significantly better than FOLFOX in PFS. In
survivals were found using C-statistics. Results: The median baseline sPDL1 subgroup with female, peritoneal metastasis, ascites, lymphadenopathy in perito-
was 0.8ng/mL(range, 0.06 - 6.06ng/mL). The median OS and PFS were neal cavity, number of organs involved . 2, POF was statistically significantly better
14.9 months (95% CI: 7.33-22.47) and 8.0 months (95% CI: 5.96-10.0), than FOLFOX in PFS. In subgroup with female, gastrium of primary tumor site,
respectively. sPDL1 and NLR showed a positive correlation. Patients with low peritoneal metastasis, ascites, no lymphadenopathy out of peritoneal cavity, IP PAC
levels of sPD-L1 at diagnosis ( , 1.92 ng/mL) showed a better OS and PFS than was statistically significantly better than FOLFOX in PFS. Intravenously docetaxel
the patients with a high sPDL1 (OS: 18.3 vs. 95 months, P = 0.057, PFS: 8.9 plus S-1 still saw response after IP PAC. Conclusions: either POF or IP PAC improved
vs. 6.0 months, P = 0.04). The baseline sPDL1 before treatment were higher in survival compared to FOLFOX, especially in patients with female or peritoneum
the PD group than in the SD and PR groups (mean:2.91, 1.17, 1.19, P = metastasis. Only POF, not IP PAC, improved response rate compared to FOLFOX.
0.019). Patients whose sPDL1 increased after 1st cycle of chemotherapy Clinical trial information: NCT02845908.
showed the tendency of worse PFS and OS. When disease progressed, P value (HR, 95%CI)
sPDL1 increased compared with baseline (mean:1.31, 1.45, P = 0.029). POF IP PAC FOLFOX
(n=30) (n=29) (n=30) POF vs FOLFOX IP PAC vs FOLFOX
Conclusions: sPDL1 at pre-chemotherapy confers the prognostic value for PFS
PFS (m, 95% CI) 6.148 6.214 4.405 0.042 0.027
and OS in GC patients under palliative 1st-line chemotherapy. The dynamics of (3.898 (4.191–8.237) (1.803–7.008) (0.603, 0.354- (0.510, 0.291-
sPDL1 during chemotherapy correlates with disease courses. -8.398) 1.026) 0.894)
OS (m, 95% CI) 9.534 10.882 6.641 0.180 0.094
(8.034- (8.839 - (4.788 - (0.679, 0.386- (0.598, 0.332-
11.034) 12.925) 8.494) 1.195) 1.077)
CR (n, %) 4 (13.3%) 2 (6.9%) 2 (6.7%)
PR (n, %) 13 (43.3%) 9 (31.0%) 9 (30.0%)
RR (n, %) 17 (56.7%) 11(37.9%) 11(36.6%) 0.121 0.920
SD (n, %) 9 (30.0%) 12 (41.4%) 12 (40%)
PD (n, %) 4 (13.3%) 6 (20.7%) 7 (23.3%)

4036 Poster Session (Board #141), Mon, 8:00 AM-11:00 AM 4037 Poster Session (Board #142), Mon, 8:00 AM-11:00 AM
Association of frequent amplification of chromosome 11q13 in esophageal Trifluridine/tipiracil (FTD/TPI) in patients (pts) aged $65 years with met-
squamous cell cancer with clinical benefit to immune check point blockade. astatic gastric/gastroesophageal junction cancer (mGC/mGEJC): Subgroup
First Author: Feng Wang, Sun Yat-sen University Cancer Center, Guangzhou, analysis from TAGS. First Author: Kohei Shitara, National Cancer Center
China Hospital East, Chiba, Japan
Background: Esophageal squamous cell carcinoma (ESCC) is the predominant Background: 60% of newly diagnosed GC pts are . 65 y of age, a proportion that is
histological subtype of esophageal cancer in South America and East Asian increasing. The global phase 3 study TAGS (NCT02500043) demonstrated the
countries and remains an unmet medical need worldwide. Previous studies have efficacy and safety of FTD/TPI in previously treated pts with mGC/mGEJC. Here we
shown the efficacy of programmed cell death 1 (PD-1) targeted therapy in a subset report results in the pt subgroup aged $65 in TAGS. Methods: Pts with mGC/mGEJC
of patients with metastatic ESCC. However, robust predictive biomarkers to PD-1 treated with $2 prior chemotherapy regimens were randomized (2:1) to receive FTD/
antibody-based immunotherapy remain undefined. Methods: Patients included TPI (35 mg/m2 BID on days 1–5 and 8–12 of each 28-day cycle) or placebo, plus best
in this analysis were part of multi-center, phase Ib/II trial (NCT02915432) supportive care. A preplanned efficacy/safety analysis was performed in pts aged $65 y.
evaluating the safety and activity of toripalimab, a humanized PD-1 antibody in Results: Of 507 randomized pts, 228 (45%) were aged $65 y (range 65–89). The pt
subset aged $65 y was similar to the overall population, except for a higher incidence
solid tumors. To identify molecular determinants of response, we performed whole
of moderate renal impairment in the elderly subgroup (31% vs 17%). For pts
exome sequencing (WES), messenger RNA sequencing and immunohisto-
aged $65 y, baseline characteristics were generally balanced across the treatment
chemistry on patients’ samples and evaluated genomic and transcriptional bio- groups, although more pts treated with FTD/TPI than with placebo had ECOG PS 1
markers, PD-L1 expression and tumor mutational burden (TMB) for correlation (69% vs 59%). FTD/TPI had an efficacy benefit in pts aged $65 y, and the FTD/TPI
with clinical efficacy. Results: Sixty advanced chemo-refractory ESCC patients safety profile was similar in this subgroup vs the overall population (table). Treatment-
were enrolled and 59 were treated with toripalimab. 94.9% (56/59) patients related deaths (one in each treatment group) did not occur in pts aged $65 y. No
experienced at least one treatment related adverse event after 16 months; mostly drug-related deaths associated with cardiotoxicity were reported in pts aged $65 y.
grade 1 or grade 2. Treatment-related grade 3 or higher AEs occurred in 30.5% Although dose modifications were used more often in this subgroup, there was no
(18/59) of subjects. By the data cutoff date, 11 (18.6%; 95%CI 9.7 to 30.9) increase in discontinuations vs the overall population. Conclusions: FTD/TPI was safe
patients achieved an objective response, while the disease control rate was 47.5% and effective in pts aged $65 y, who had a higher incidence of moderate renal
(95%CI 34.3 to 60.9). Copy number analysis identified 24 out of 50 (48%) impairment vs the overall population. Clinical trial information: NCT02500043.
patients with amplifications of chromosome 11q13 region, which was consistent
Overall population1 Age ‡65 y
with elevated mRNA expression of amplified genes, including CCND1(Cyclin D1)
FTD/TPI Placebo FTD/TPI Placebo
and fibroblast growth factor family members (FGF3/4/19). Patients without
11q13 amplification, had significantly better objective response rate (ORR ITT population, n 337 170 154 74
Median OS, mo 5.7 3.6 6.2 5.4
30.8% versus 4.2%, p= 0.024) and progression free survival (3.7 versus HR (95% CI) 0.69 (0.56–0.85) 0.73 (0.52–1.02)
2.0 months, HR = 0.47 [95%CI 0.24 to 0.91], p= 0.025) when compared Median PFS, mo 2.0 1.8 2.2 1.8
with 11q13 amplified individuals. In contrast, patients with high TMB HR (95% CI) 0.57 (0.47–0.70) 0.44 (0.32–0.61)
Safety population, n 335 168 153 72
($12 Mutations/Mb; 11/47, 23.4%) or positive PD-L1 expression (TC or IC 1%; Grade ‡3 AEs of any cause, %
19/57, 33.3%) showed no significant advantage in ORR or survival. Any 80 58 80 51
Most commona
Conclusions: Toripalimab has demonstrated a manageablesafety profile and Neutropeniab 34 0 40 0
promising anti-tumor activity in chemo-refractory ESSC patients. Genomic am- Anemiac 19 8 18 8
plification of 11q13 region may serve as a negative predictive marker for ad- Actions taken for any-cause/grade AEs, %
Dose modification 58 22 61 22
vanced ESSC patients receiving anti-PD-1 based immunotherapy. Further Treatment discontinuation 13 17 12 14
interrogation of putative resistance genes that lie within this region is under study. a b c
Occurring in $10% of pts in any group. Includes decreased neutrophil count. Includes decreased
Clinical trial information: NCT02915432. hemoglobin level. 1. Shitara K, et al. Lancet Oncol 2018.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Noncolorectal) Cancer 231s

4038 Poster Session (Board #143), Mon, 8:00 AM-11:00 AM 4039 Poster Session (Board #144), Mon, 8:00 AM-11:00 AM
Trifluridine/tipiracil (FTD/TPI) in patients (pts) with metastatic gastroesoph- Pooled safety analysis from phase 3 studies of trifluridine/tipiracil (FTD/TPI)
ageal junction cancer (mGEJC): Subgroup analysis from TAGS. First Author: in patients (pts) with metastatic gastric/gastroesophageal junction cancer
Wasat Mansoor, Christie NHS, Manchester, United Kingdom (mGC/mGEJC) and metastatic colorectal cancer (mCRC). First Author: Eric
Van Cutsem, University Hospitals Gasthuisberg, Leuven and KU Leuven,
Background: The incidence of GEJC is increasing in North America and Europe,
especially among white men. Many pts present with metastatic disease or relapse
Leuven, Belgium
locally or systemically after resection of early-stage disease. The global phase 3 study Background: FTD/TPI was approved in 2015 for pretreated pts with mCRC based on
TAGS (NCT02500043) demonstrated the efficacy and safety of FTD/TPI in pre- the phase 3 RECOURSE trial. FTD/TPI recently demonstrating significantly im-
viously treated pts with metastatic gastric cancer (mGC)/mGEJC. Here we report proved overall survival vs placebo in pretreated pts with mGC/mGEJC in the phase 3
results in the mGEJC subgroup from TAGS. Methods: Pts with mGC/mGEJC treated TAGS trial. Methods: We evaluated the pooled safety of FTD/TPI in TAGS and
with $2 prior chemotherapy regimens were randomized (2:1) to receive FTD/TPI RECOURSE in all pts who received $1 dose of FTD/TPI (safety population). Pts were
(35 mg/m2 BID on days 1–5 and 8–12 of each 28-day cycle) or placebo, plus best required to have ECOG PS 0/1 and to have received $2 previous chemotherapy lines.
supportive care. A preplanned efficacy and safety analysis was performed in pts with Results: FTD/TPI and placebo were administered to 335 and 168 pts, respectively,
mGEJC. Results: Of 507 randomized pts, 145 (29%) had GEJC as the sole primary in TAGS, and 533 and 265 pts in RECOURSE. Baseline characteristics were
disease site (FTD/TPI, 98/337; placebo, 47/170). Of pts with mGEJC, 85% were balanced across treatment groups and reflected the disease populations. In the
male and 83% were white (overall population, 73% and 70%). Baseline charac- pooled population, 66% of pts were men and 75% had received $3 prior systemic
teristics were generally balanced for pts with mGEJC across treatment groups, except treatments. The safety profile of FTD/TPI was comparable between studies (table). In
for fewer pts having prior gastrectomy (40% vs 55%) and more pts having TAGS and RECOURSE, the most common any-cause grade (gr) $3 AEs in FTD/TPI-
received $3 prior regimens (74% vs 66%) in the FTD/TPI group than in the placebo treated pts were neutropenia (34%; 35%), anemia (19%; 17%), and leukopenia
group. FTD/TPI had an efficacy benefit in pts with mGEJC, and the FTD/TPI safety (9%; 13%). Gr $3 febrile neutropenia occurred in 2% and 4% of pts and gr $3 GI
profile was similar in this subgroup and the overall population (table). AEs in 21% and 12%. Gr $3 cardiac AEs were reported in 1% of FTD/TPI-treated pts
Conclusions: FTD/TPI showed a manageable safety profile and efficacy benefit in pts (both studies), in contrast to results obtained with other third-line agents. Similar
with mGEJC in the TAGS trial, despite heavier pretreatment of the FTD/TPI than the proportions of FTD/TPI-treated pts in both studies had AEs leading to dosing delay,
placebo group. Clinical trial information: NCT02500043. dose reduction, or treatment discontinuation. Dosing delay was used more often than
Overall population1 mGEJC
dose reduction to manage AEs. TRAEs leading to death occurred in one FTD/TPI-
treated pt ( , 1%) in each trial. Conclusions: In a pooled analysis, FTD/TPI was well
FTD/TPI Placebo FTD/TPI Placebo
tolerated with a consistent safety profile in pts with mGC/mGEJC or mCRC. The most
ITT population, n 337 170 98 47 frequent AEs were hematologic and GI, which were managed with dosing delays/
Median OS, mo 5.7 3.6 4.8 3.5
HR (95% CI) 0.69 (0.56–0.85) 0.75 (0.50–1.11) dose reductions. Clinical trial information: NCT02500043; NCT01607957.
Median PFS, mo 2.0 1.8 1.9 1.8
TAGS RECOURSE
HR (95% CI) 0.57 (0.47–0.70) 0.60 (0.41–0.88) (mGC/mGEJC) (mCRC)
Safety population, n 335 168 97 46
Grade ‡3 AEs of any cause, % FTD/TPI Placebo FTD/TPI Placebo
Any 80 58 77 59 (n = 335) (n = 168) (n = 533) (n = 265)
Most commona Any gr, % Gr ‡3, % Any gr, % Gr ‡3, % Any gr, % Gr ‡3, % Any gr, % Gr ‡3, %
Neutropeniab 34 0 25 0
Anemiac 19 8 13 4 Any-cause AEs 97 80 93 58 98 69 93 52
Fatigue 7 6 10 0 TRAEs 81 53 57 13 86 49 55 10
Abdominal pain 4 9 4 15 Actions taken for any-cause AEs
AEs of any grade or cause, % Dosing delay 57 41 21 16 52 35 13 8
Leading to dosing modification 58 22 54 24 Dose reduction 11 7 1 1 14 12 1 1
Treatment 13 11 17 12 10 8 14 11
Leading to treatment discontinuation 13 17 9 11 discontinuation
a
Occurring in $10% of pts in any group. bIncludes decreased neutrophil count. cIncludes decreased TRAEs, treatment-related AEs.
hemoglobin level. 1. Shitara K, et al. Lancet Oncol 2018.

4040 Poster Session (Board #145), Mon, 8:00 AM-11:00 AM 4041 Poster Session (Board #146), Mon, 8:00 AM-11:00 AM
Homologous recombination deficiency as prognostic marker in metastatic Safety of neoadjuvant chemoradiation (CRT) in combination with avelumab
gastric cancer. First Author: Nuno Sousa, Instituto Portugues de Oncologia (A) in the treatment of resectable esophageal and gastroesophageal junction
do Porto (IPO Porto), Porto, Portugal (E/GEJ) cancer. First Author: Nataliya Volodymyrivna Uboha, University of
Wisconsin, Carbone Cancer Center, Madison, WI
Background: Gastric cancer is the 5th cancer diagnosis and 3rd cause of
cancer death worldwide. Metastatic gastric cancer (mGC) has a median Background: Neoadjuvant CRT followed by surgery is the standard of care
survival of 11 months. mGC is an heterogeneous disease and different biologic (SOC) for patients (pts) with stage II/III E/GEJ cancer. However, recurrence
characteristics may justify differential therapeutic opportunities. Tumors with rates are high. Immunotherapy has demonstrated promising activity in
homologous recombination (HR) deficiency may benefit with treatment with advanced E/GEJ cancer. This trial evaluates safety and efficacy of peri-
PARP inhibitors or immune checkpoint inhibitors. The purpose of this study operative A with CRT in resectable E/GEJ cancer. Methods: This is a 2-part
was to evaluate the prevalence and prognostic impact of altered expression phase I/II trial. Part 1 is a run-in phase with 6 pts for safety evaluation. Part 2
of HR proteins as surrogates for homologous recombination deficiency (HRD) will enroll additional 18 pts in the expansion cohort. Pts with E/GEJ cancer of
in mGC. Methods: Multicenter retrospective cohort of mGC treated with any histology receive CRT (41.4 Gy in 23 fractions) with carboplatin and
platinum-based chemotherapy. HRD defined as absence of at least one of the paclitaxel as per SOC. Three doses of A (10 mg/kg, q14 days) are admin-
following proteins: ATM, ATR, CHK2, RAD51, RAD52, BRCA1, BRCA2, istered starting on day 29 of treatment, to coincide with the last chemo-
MRE11 by immunohistochemistry. Survival time calculated as the difference therapy dose. Surgery is performed ~8 weeks after CRT completion. Pts
between first cycle of platinum-based chemotherapy and death or last ob- receive 6 doses of A after resection. Dose-limiting toxicity (DLT) evaluations
servation. Association between HRD and survival examined with log-rank test. are completed on the first post-operative clinic visit, 2-4 weeks post re-
Results: 440 patients included, of which 70% male, with mean age of 58 years section. Results: Between 6/2018 and 2/2019, 6 pts (all male, median
(SD: 11). 75% of patients had mGC at diagnosis, 43% had 2 or more organs age 62) enrolled in part 1: 6 adenocarcinoma (100%); 1 E, 3 Siewert 1, 2
involved and 63% were registered as ECOG 0 or 1 at the start of first line Siewert 2; 1 cT2N0, 2 cT3N0, 3 cT3N1. All pts underwent successful
chemotherapy. The most common histologic subtype was tubular adenocar- resection with negative surgical margins. 1/6 pts had R1 resection due to
cinoma (44%) followed by diffuse carcinoma (32%). HRD was noted in 196 tumor extension to inked adventitial surface without invasion of surrounding
(45%) cases (95%CI: 40%-49%). The most commonly altered proteins were structures. There were no unexpected surgical complications. At resection, 2
ATM (21%) and BRCA2 (18%). In HRD tumors, 99 (51%) had altered ex- pts had ypT0N0, 2 ypT1N0, 1 ypT2N0, and 1 ypT3N1 disease. Combination
pression of only one HR protein; 47 (24%) had altered expression of two HR of CRT and A had an acceptable toxicity profile. No DLTs were seen in the
proteins; the remaining cases had altered expression of 3 or more proteins. first 5 pts, so expansion cohort is open to enrollment. No grade $3 immune-
After a median follow up of 11 months, median survival for the cohort was related AEs were observed. Immune-related hypothyroidism was seen in 1
11 months (95% CI 9-12). HRD was associated with an improved survival, patient (grade 2). 6/6 pts had reversible grade 3 or 4 lymphopenia; 1/6 had
HR = 0.61 (95%CI: 0.48-0.78, p , 0.001), that remained significant after grade 3 neutropenia. Correlative studies are ongoing and will be presented at
adjustment for sex, age, performance status and disease status (HR = 0.63; the meeting. Conclusions: Perioperative CRT with A is well tolerated with no
95%CI: 0.48-0.82; p , 0.001). Conclusions: HRD phenotype was present in unexpected toxicities. Additional safety and correlative data will be pre-
45% of mGC cases and is associated with improved prognosis for mGC treated sented at the meeting. This study is actively enrolling pts to an expansion
with platinum-based first line chemotherapy. cohort at University of Wisconsin. Clinical trial information: NCT03490292.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
232s Gastrointestinal (Noncolorectal) Cancer

4042 Poster Session (Board #147), Mon, 8:00 AM-11:00 AM 4043 Poster Session (Board #148), Mon, 8:00 AM-11:00 AM
Efficacy and safety of sintilimab in combination with XELOX in first-line Analysis of symptoms and functional HRQoL scales in TAGS, a phase III trial
gastric or gastroesophageal junction carcinoma (GC/GEJC). First Author: of trifluridine/tipiracil (FTD/TPI) in metastatic gastric cancer (mGC). First
Nong Xu, The First Affiliated Hospital, Zhejiang University, Hangzhou, China Author: Maria Alsina, Vall d’Hebron Institute of Oncology, Barcelona, Spain
Background: Immune checkpoint inhibitors have shown clinical benefit in Background: The phase 3, randomized, double-blind, placebo-controlled
advanced GC/GEJC. This phase 1b study evaluates the efficacy and safety of study (TAGS) evaluated the efficacy and safety of FTD/TPI (35 mg/m² given
sintilimab, an anti-programmed cell death-1 antibody (PD-1 Ab) in com- orally twice a day on days 1–5 and 8–12 of a 28-day cycle) in mGC patients
bination with XELOX for GC/GEJC in first-line setting. Methods: This phase who had previously received$2 prior regimens for advanced disease and
1b study enrolled treatment-naı̈ve unresectable locally advanced or meta- demonstrated a clinically relevant and statistically significant benefit in OS
static GC/GEJC patients without HER2 amplification in cohort F. Patients and PFS with a predictable and manageable safety profile. HRQoL data and
received sintilimab 200mg IV q3w until disease progression, unacceptable association between QoL and time to ECOG status deterioration (2 or more) are
toxicity or death, in combination with XELOX regimen (oxaliplatin 130mg/m2 reported here. Methods: HRQoL was evaluated using EORTC QLQ-C30 and the
IV D1 and capecitabine 1000mg/m2 PO BID D1-14) for up to 6 cycles. The gastric-specific module (QLQ-STO22) questionnaires at baseline and at every
primary objective was to evaluate the efficacy of the combination per RECIST 4 weeks thereafter until treatment discontinuation. Prespecified key HRQoL
v1.1 and safety and tolerability. Results: Totally 20 patients were enrolled in were changes from baseline and time to deterioration. Changes $10 points
cohort F. As data cutoff (15 Jan 2019), median follow up was 5.8 months were deemed clinically relevant. A time-dependent Cox-regression analysis
(range, 2.4 to 12.5). The median dose of sintilimab was 6.5 (range, 4 to 12). was performed to evaluate the association of 10-point Global Health Status
The objective response rate (ORR) was 85.0% (95%CI, 62.1 to 96.8) and deterioration with worsening ECOG status. Results: Of 507 patients ran-
disease control rate (DCR) was 100.0% (95%CI, 83.2 to 100.0). Among 17 domized, 332/337 (98.5%) of FTD/TPI and 164/170 (96.5%) of placebo had
patient with BOR of PR, two patients achieved a complete response (CR) of baseline QoL data. Overall compliance was 84% for both questionnaires.
the target lesion. The median duration of response (DOR) and median Demographic and disease were generally balanced between the two groups;
progression free survival (PFS) had not been met. Three patients underwent QoL scores were also similar between groups. HRQoL was largely maintained
resection of primary tumor after achieving a BOR of partial response (N=2) during treatment in both arms for most items; mean changes from baseline
and stable disease (N=1). The incidence of treatment emergent adverse remained under the 10-point threshold. Clinically relevant changes from
events (TEAEs) was 85.0%. Treatment-related AEs (TRAEs) occurred in 14 baseline were observed only for pain relief at cycle 2 (favouring FTD/TPI); and
(70.0%) patients. The incidence of TRAE $ Grade 3 was 15%. AEs of improved role functioning at cycle 3 (favouring placebo). In a sensitivity
immune-related etiology, occurred in 6 patients (30.0%). There were no AEs analysis including death or progression as an event, FTD/TPI was associated
that resulted in death. As data cutoff, 12 patients were still in treatment and with a positive trend suggesting a reduced risk of QoL deterioration across all
8 had discontinued treatment and were under survival follow up. The bio- scales compared to placebo (HRs ranged from 0.57 to 0.74. A 10-point Global
marker analysis including PD-L1 expression in tumor specimen was ongoing. Health Status deterioration was associated with a worsening ECOG status (HR,
Conclusions: Sintilimab in combination with XELOX in first-line GC/GEJC 95% CI, 1.5, 1.2 to 1.86). Conclusions: During the treatment period, HRQoL
shows promising anti-tumor efficacy and a tolerable safety profile. The remained stable for most functional and symptom scales in both arms, sug-
further randomized, phase 3 study of Sintilimab in combination with XELOX gesting that HRQoL is largely maintained with FTD/TPI. Treatment with FTD/
in this setting is ongoing (NCT03745170). Clinical trial information: TPI was associated with a positive trend toward a lower risk of QoL deterioration
NCT02937116. than placebo across all scales. Changes in QoL were informative for patients
‘expected ECOG status. Clinical trial information: NCT02500043.

4044 Poster Session (Board #149), Mon, 8:00 AM-11:00 AM 4045 Poster Session (Board #150), Mon, 8:00 AM-11:00 AM
Recurrence risk evaluation in stage IB gastric cancer with TP53 codon 72 A phase II feasibility trial of neoadjuvant chemoradiotherapy combined with
polymorphism. First Author: Satoshi Nishizuka, Iwate Medical University atezolizumab for resectable esophageal adenocarcinoma: The PERFECT
Institute for Biomedical Sciences, Morioka, Japan trial. First Author: Tom van den Ende, Amsterdam UMC, University of
Amsterdam, Department of Medical Oncology, Cancer Center Amsterdam,
Background: Post-operative adjuvant chemotherapy is not currently indicated
Amsterdam, Netherlands
for Stage IB gastric cancer. However, about 10% of these patients experience
recurrence and metastasis. Our previous study on a panel of gastric cancer Background: The CROSS study demonstrated the superiority of neoadjuvant
cell lines indicated that TP53 codon 72 polymorphisms may affect the degree chemoradiotherapy (nCRT) over surgery alone (van Hagen et al. NEJM. 2012).
of biological malignancy. Hence, we hypothesized that the TP53 codon 72 However, for resectable esophageal adenocarcinoma (rEAC) 5y survival is only
polymorphisms may have been associated with post-operative survival without 43%. PD1/PDL1 checkpoint inhibitors have shown promising efficacy for
adjuvant chemotherapy. In this study, we investigated the risk of recurrence several cancer types, including esophageal cancer. To further improve out-
after treatment of Stage IB gastric cancer patients carrying the TP53 codon 72 comes in rEAC, we performed a phase II trial of nCRT combined with
polymorphism and attempted to identify a subpopulation that should receive atezolizumab, a PD-L1 inhibitor. Methods: Pts with rEAC received standard
post-operative adjuvant chemotherapy. Methods: Among 658 gastric cancer dose CROSS regimen (5 cycles of IV: carboplatin AUC2, paclitaxel 50 mg/m2
patients who received gastrectomy with curative-intent, 130 Stage IB patients and concurrent 23 fractions of 1.8 Gy on weekdays) with atezolizumab (5
were enrolled in the present study. The TP53 codon 72 polymorphisms of cycles: 1200 mg IV, 3 weekly). Primary endpoint was the percentage of pts
formalin-fixed paraffin-embedded cancer tissue sections were assessed by completing treatment with atezolizumab. Secondary endpoints included:
direct sequencing using originally designed primers. Overall survival rate (OS) toxicity, post-operative complications (Clavien-Dindo), Mandard score, R0
and relapse-free survival rate (RFS) were analyzed based on the status of TP53 resection rate, PFS and OS. In total 40 pts will be enrolled. Results: Since July
codon 72 polymorphism "Arg/Arg", "Arg/Pro" and "Pro/Pro". The hazard ratio for 2017, 39 pts have been enrolled (87% males, median age 63). Neoadjuvant
each subgroup was compared by TP53 codon 72 polymorphism. All interaction treatment was completed by 31 pts and is ongoing in 8 pts. All cycles/fractions
p values were calculated using the likelihood test. Results: Of the 125 patients of nCRT were administered in 29/31 pts; 26 pts completed all cycles of
for whom polymorphism analysis results were available, the 5- and 10-year OS atezolizumab, 24 pts finished complete neoadjuvant treatment. Reasons for
was 84.5% and 63.9%, respectively. The 5- and 10-year RFS was 82.2% and missing any cycle of chemotherapy/atezolizumab included: toxicity (6 pts, in 3/
64.3%, respectively. When the study cohort was divided into two groups 6 pts immune-related adverse events (irAE)) and progression (1 pt). Grade 3-4
according to polymorphism status (i.e., Arg/Arg and Arg/Pro vs. Pro/Pro), both toxicity was observed in 15/31 pts (6/31 irAEs of any grade) which did not delay
the OS (hazard ratio [HR], 1.968; 95% confidence interval [CI], 0.770-7.430, surgery. Thus far 23/31 pts were resected, 3 pts are planned for surgery, 3 pts
p = 0.045) and RFS (HR, 1.976; 95% CI, 0.778-7.515, p = 0.033) of the Pro/ had interval metastases preoperatively, 1 pt died during treatment (pulmonary
Pro group across the entire observation period were significantly lower than embolism), and 1 pt declined surgery. Clavien-Dindo grade 3-4 complications
those for the Arg/Arg and Arg/Pro group. The majority of recurrences in Pro/Pro were seen in 11/23 pts with no surgery related mortality. A pathological
occurred within three years from the operation. Conclusions: Among Stage IB complete response (pCR), Mandard 1 was seen in 9/23 (39%) pts. All patients
gastric cancer patients that underwent gastrectomy with curative-intent, post- underwent an R0 resection. Updated results will be presented at the meeting.
operative adjuvant chemotherapy may be considered immediately after surgery Conclusions: Based on data thus far, atezolizumab added to nCRT is feasible.
for patients carrying the TP53 codon 72 Pro/Pro polymorphism. A pCR was observed in 39% of patients, which is promising compared to 23%
in the CROSS study. Treatment is associated with irAE which are manageable.
Biomarker research will be performed on blood (circulating tumor DNA), tissue
(immune microenvironment) and feces (microbiome). Clinical trial informa-
tion: NCT03087864.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Noncolorectal) Cancer 233s

4046 Poster Session (Board #151), Mon, 8:00 AM-11:00 AM 4047 Poster Session (Board #152), Mon, 8:00 AM-11:00 AM
Total neoadjuvant chemo (ctx; TNT) for locally advanced gastric cancer (GC): Clinical and molecular factors predicting resistance to first-line (1L) FOLFOX
The Memorial Sloan Kettering Cancer Center experience. First Author: in patients (pts) with advanced esophagogastric cancer (EGA) and patterns
Megan Greally, Memorial Sloan Kettering Cancer Center, New York, NY of subsequent therapy. First Author: Megan Greally, Memorial Sloan Ket-
tering Cancer Center, New York, NY
Background: Peri-op chemo (ctx) and surgery is a standard in the treatment
of GC, based on the MAGIC (NEJM 2006; 355:11) and FLOT4 (J Clin Oncol Background: In the US, FOLFOX is the most widely used 1L treatment for
35:4004 [abstr]) studies. However, less than half of patients (pts) completed advanced EGA. We evaluated if FOLFOX resistance (FR) impacted sub-
ctx in the MAGIC and FLOT4 studies, mainly from issues delivering post-op sequent therapy and characterized the clinical and molecular factors that
therapy. We assessed safety and feasibility of TNT, where all ctx is given pre- predict resistance to 1L FOLFOX. Methods: We reviewed pts with advanced
op. Methods: We reviewed GC pts who received TNT or peri-op ctx and had Her2-negative EGA treated with 1L FOLFOX from Jan 2013 to Aug 2017.
surgery; decision for TNT was by physician preference, based on clinical or Response or stable disease (SD) at time of first restaging scan defined pts as
radiographic benefit to justify completing ctx pre-op. Pt characteristics were FOLFOX sensitive (FS); progression defined FR. Pt characteristics were
compared using Fisher’s exact and Wilcoxon Rank Sum tests. Post-op length compared using Fisher’s exact and Wilcoxon Rank-Sum tests. Outcomes
of stay (LOS) was calculated from date of surgery (DOS) to date of discharge were correlated with clinical variables and MSK-IMPACT data. Microsatellite
and surgical morbidity was determined using the Clavien-Dindo classifi- instable (MSI) pts were excluded from gene and pathway analysis. Overall
cation. Progression free survival (PFS) and overall survival (OS) were cal- survival (OS) was calculated from start of FOLFOX using Kaplan-Meier
culated from DOS using Kaplan-Meier methods and compared between methods. Landmark analysis (2 months [mo] after starting FOLFOX) was
groups using the log-rank test. Results: 120 pts were identified, median age used to compare OS between groups. Results: We identified 311 pts, me-
63, 62.5% male, 98% ECOG 0/1. 93 pts (77.5%) received peri-op ctx and dian age 62, 73% male, 82% ECOG 0/1. 246 pts (79%) were FS and 65
27 (22.5%) received TNT. In peri-op pts, 19%, 43% and 38% received (21%) were FR. FR pts had a higher number of metastatic sites, p = 0.001.
FLOT, platinum/fluropyrimidine (FP) and ECF/EOX respectively. In TNT pts, Median OS was 13.4 mo in FS pts vs 4.3 mo in FR pts (p , 0.001). At time of
56%, 37% and 7% received FLOT, platinum/FP and ECF/EOX respectively. analysis, 213 pts (172 FS pts and 41 FR pts) and 110 pts (90 FS pts and 20
57% had subtotal gastrectomy. Surgical outcomes were similar between FR pts) had received 2nd-line (2L) and 3rd-line (3L) chemo respectively. In
groups; median LOS was 6 and 7 days (p = 0.31) in peri-op and TNT pts pts who received 2L chemo (ctx), there was no difference in ctx duration
respectively. There was no significant difference in Clavien Dindo grade I-II between FS and FR pts (2.1 vs 1.4 mo, p = 0.67). However, in pts who
or III-IV morbidity between groups (p = 0.103). There were no deaths. TNT received 3L ctx, the duration was longer in FS vs FR pts (1.6 vs 0.5 months,
pts received higher proportions of planned treatment than peri-op ctx pts: p = 0.002). In IMPACT tested pts (n = 130), univariate analysis identified
90% vs. 60% FP (0.001); 85% vs. 41% platinum ( , 0.001); 100% vs. 9% EGFR (33.3% vs 10.7%, p = 0.032) and WNT pathway (26.7% vs 4.9%, p =
epirubicin (0.015) and 53% vs. 28% docetaxel (p = 0.169). At median 0.015) alterations more frequently in FR pts vs FS pts. 12 pts were MSI, all
follow-up of 19 months, median PFS and OS were not reached. There was no were FS (3 SD, 8 PR, 1 CR). Of 63 pts who had IMPACT germline testing, 7
significant difference in PFS (p = 0.089) or OS (p = 0.59) between groups. (11%) had germline mutations identified, including CDH1 (n = 2) and ATM
Conclusions: TNT appears safe with no increase in post-op LOS or surgical (n = 1), all were FS. No pts were enrolled on genotype-matched trials as a
morbidity observed. TNT pts had higher percentage drug delivery, suggesting result of IMPACT testing. Conclusions: From time of first restaging scan, FR
potential benefit for administering all ctx before surgery. While longer pts have inferior OS to FS pts. The duration of 3L ctx was longer in FS vs FR
survival follow-up is required, TNT may be considered in pts with locally pts. Higher number of metastatic sites and EGFR and WNT pathway al-
advanced GC who are candidates for ctx. terations were associated with FR; MSI was not. IMPACT did not facilitate
enrolment on genotype-matched studies.

4048 Poster Session (Board #153), Mon, 8:00 AM-11:00 AM 4049 Poster Session (Board #154), Mon, 8:00 AM-11:00 AM
Health-related quality of life (HRQoL) of pembrolizumab (pembro) versus Correlation between immune-related adverse events and prognosis in pa-
physician choice single-agent paclitaxel, docetaxel, or irinotecan in subjects tients with gastric cancer treated with nivolumab. First Author: Ken Masuda,
with advanced/metastatic adenocarcinoma (ACC) or squamous cell carci- National Cancer Center Hospital, Tokyo, Japan
noma (SCC) of the esophagus that has progressed after first-line standard
Background: Recent studies have shown that immune-related adverse
therapy (KEYNOTE-181). First Author: Antoine Adenis, Institut du Cancer de
events (irAEs) caused by immune checkpoint inhibitors were associated with
Montpellier, Montpellier, France
clinical benefit in patients with melanoma or lung cancer. In advanced
Background: KEYNOTE-181 (NCT02564263) is an open-label, randomized, gastric cancer (AGC) patients, there have been few reports about the cor-
phase 3 trial in ACC and SCC of the esophagus that evaluated IV pembro 200 mg relation between irAEs and efficacy of immune checkpoint inhibitors.
Q3W for up to 2 years vs investigator choice of single-agent paclitaxel/docetaxel/ Therefore, in this study, we retrospectively investigated the correlation
irinotecan (control). Pembro was superior to control for OS in patients with between irAEs and efficacy in AGC patients treated with nivolumab.
PD-L1 CPS $10 (N = 222; median 9.3 vs 6.7 months; P= 0.0074). Here Methods: The subjects of this study were AGC patients that had received
we present results of prespecified HRQoL analyses in this population. nivolumab monotherapy between January 2015 and August 2018. IrAEs
Methods: The EORTC QLQ-C30 and EORTC QLQ-OES18 were administered were defined as those AEs having a potential immunological basis that
at baseline; weeks 2, 3, 4, 6, 9, 12, 18; every 9 weeks up to 1 year/end of required close follow-up, or immunosuppressive therapy and/or endocrine
treatment; and 30-day safety follow-up visit. Data from patients receiving $1 therapy. We divided the patients who received nivolumab into two groups
dose of study treatment and completing $1 HRQoL assessment were analyzed. based on occurrence of irAEs; those with irAEs (irAE group) or those without
Least squares mean (LSM) score change from baseline to week 9, 95% CI, and (non-irAE group). We assessed the efficacy in both groups. Results: Of the 65
nominal P values were calculated. Time to deterioration (TTD) ($10-point AGC patients that received nivolumab monotherapy, 14 developed irAEs.
decline from baseline) was assessed by Kaplan-Meier method and Cox re- The median time to onset of irAEs was 30.5 days (range 3–407 days).
gression model. HRs, 95% CIs, and nominal P values are provided. Results: The Median follow-up period for survivors was 32 months (95% CI, 10.8 to
HRQoL population included 218 PD-L1 CPS $10 patients (107 pembro, 111 34.5). The median progression-free survival was 7.5 months (95% CI, 3.6 to
control). QLQ-C30 compliance at week 9 was 88.9% for pembro and 83.9% for 11.5) in the irAE group and 1.4 months (95% CI, 1.2 to 1.6) in the non-irAE
control. There was no significant difference in LSM between arms (3.68; 95% group (HR = 0.11, p , 0.001). The median overall survival was 16.8 months
CI –2.28, 9.64; P= 0.2248) in global health status (GHS)/QoL score. Week 9 (95% CI, 4.4 to not reached) in the irAE group and 3.2 months (95% CI, 2.2
QLQ-OES18 compliance was 88.4% for pembro and 83.3% for control. QLQ- to 4.1) in the non-irAE group (HR = 0.17, p , 0.001). Multivariate analysis
OES18 scores were not significantly different between arms. TTD for pain (HR demonstrated that high ALP level (HR = 2.88; 95% CI, 1.51 to 5.51) and
1.02; 95% CI 0.58, 1.81; P= 0.5282), reflux (HR 1.69; 95% CI 0.83, 3.47; absence of irAEs (HR = 3.06, 95% CI, 3.06 to 23.46 for yes vs. no) were
P= 0.9254), and dysphagia (HR 1.81; 95% CI 0.97, 3.37; P= 0.9693) sub- associated with a poor prognosis. The most frequent irAEs was diarrhea/
scales were not significantly different between arms. Conclusions: Over 9 weeks, colitis (n = 5). Grade 3 adverse events were observed in 6 patients; hy-
patients treated with pembro had stable GHS/QoL scores similar to those of perglycemia (n = 2), diarrhea/colitis (n = 1), adrenal insufficiency (n = 1),
patients treated with single-agent docetaxel/paclitaxel/irinotecan. Combined with increased aspartate aminotransferase increased (n = 1), peripheral motor
the superior OS and lower rate of treatment-related AEs seen with pembro, these neuropathy (n = 1). One of the 14 patients experienced the irAE after
data support clinically meaningful benefit of pembro in esophageal cancer pa- discontinuation of nivolumab due to progression of disease. There were no
tients with PD-L1 CPS $10. Clinical trial information: NCT02564263. grade 4 or 5 adverse events related to nivolumab. Conclusions: Development
of irAEs was associated with clinical benefit for AGC patients receiving
nivolumab monotherapy.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
234s Gastrointestinal (Noncolorectal) Cancer

4050 Poster Session (Board #155), Mon, 8:00 AM-11:00 AM 4051 Poster Session (Board #156), Mon, 8:00 AM-11:00 AM
Perioperative (P) UGT1A1 genotype guided irinotecan (iri) dosing ‘gFOLFIR- Best supportive care (BSC) with or without low-dose chemotherapy (chemo)
INOX’ for gastroesophageal adenocarcinoma (GEA). First Author: Daniel V.T. in frail elderly patients with advanced gastroesophageal cancer (aGOAC):
Catenacci, University of Chicago Medical Center and Biological Sciences, The uncertain randomization of the GO2 phase III trial. First Author: Daniel
Chicago, IL Swinson, St James, Leeds, United Kingdom
Background: Complete resection (R0) and pathologic response grade (PRG) correlate Background: Before 2000, trials comparing BSC +/- chemo for aGOAC showed
with long-term GEA outcome. FOLFIRINOX demonstrated efficacy in advanced GEA; overall survival (OS) benefit, but in predominantly fit patients (pts). We have
gFOLFIRINOX improved tolerability. We evaluated R0, PRG and tolerability in this revisited this question in a modern context, using low-dose chemo in a frail
pilot P study. Methods: Gastric body (GB) + esophagogastric (EGJ) GEA patients (pts) population, with comprehensive baseline health and frailty assessment.
with $T3Nx or TxN+ were enrolled & treated with 4 pre + 4 postoperative biweekly Methods: In the GO2 trial, elderly and/or frail aGOAC pts with a “certain”
cycles of gFOLFIRINOX (5-FU 2400mg/m2 over 46 hrs; oxaliplatin 85mg/m2; iri: indication for chemo were randomised between 3 chemo doses. In this GO2
180mg/m2 for UGT1A1 genotype 6/6, 135mg/m2 for 6/7, 90mg/m2 for 7/7) (+ substudy, pts with an “uncertain” indication for chemo were instead rando-
trastuzumab (T) 6mg/kg then 4mg/kg for HER2+) with prophylactic peg-filgastrim.
mised to BSC 6 the lowest dose chemo. Pts were eligible if clinician and pt
1°endpoint R0 resection required 36 pts to assess for a 90% R0 rate (intention to treat
(ITT)) with 90% power + 0.05 alpha; $30/36 R0 considered positive. Co-1°endpoint
agreed the indication for chemo was uncertain. There was no PS threshold, but
was PRG (Becker); 36 pts provided 85% power with 0.05 alpha for a complete (pCR eGFR $30 and bili , 2xULN were required. Baseline assessment included
G1a) rate of 16%. 2°endpoints were safety/toxicity, PET response, & R0/PRG by global QL, symptom & functional scales, frailty and comorbidity. Random-
tumor site, histologic subtype, HER2 status, & UGT1A1 genotype. We report efficacy isation was 1:1 to BSC alone, or with oxaliplatin 78 mg/m2 d1, capecitabine
and toxicity data from the neoadjuvant (Neo) portion of the study; postop data & 375 mg/m2 bd d1-21 (modified if eGFR 30-50 ml/min or bili 1.5-2.0 xULN),
survival outcomes will be presented at the meeting. Results: 4 sites enrolled 36 ITT pts q21d. QL was reassessed after 9 and 18 wks. The primary endpoint analysis
between 2/2014-8/2018; 75% male, median age 66 (range 27-85). All pts completed was OS, adjusted for baseline factors. The sample size for this exploratory sub-
all 4 cycles of Neo therapy: 10% had any dose reduction of iri (16%/0%/25% by study was not pre-set, but around 60 pts were anticipated. Results: 558 pts
genotype 6/6, 6/7, 7/7); any G3+ toxicity occurred in 35% of pts (32% 6/6, 29% 6/7, entered GO2 at 61 centres 2014-17, of whom only 45 pts (8%) at 21 centres
75% 7/7). G3+ toxicity in $5% of pts: diarrhea (17.5%; 6/6 21%, 6/7 11%, 7/7 entered this uncertain randomisation. This would provide 80% power at p =
25%), anemia (5%), vomiting (5%). Efficacy is shown in the Table. Of pts going to 0.05 (2-tailed) to detect an OS HR of 0.3. OS was shorter in pts with worse
surgery, both R1 resections were GB linitus. PRG1(a+b) was achieved in 36% of ITT baseline PS (p,0.01) or distant mets (p,0.05). OS was not significantly
pts, 46% of intestinal type histology. Conclusions: Neo gFOLFIRINOX was tolerable improved with chemo; however we cannot exclude HR .0.32. QL deteriorated
with surrogate efficacy comparable to FLOT. Clinical trial information: NCT02366819. less with BSC+chemo than with BSC alone. Conclusions: In this frail, poor PS
PET response population, we observed a small survival benefit with chemo but this did not
PRG ITT (%) >-35% SUV (%)
reach statistical significance. Clinicians should carefully consider BSC alone
gFOLFIRINOX +/-T Incidence (%) R0 ITT (%) 1a+b 1a 1b 2 3 *In evaluable pts
as a valid treatment option for aGOAC pts with poor PS and/or frailty. Clinical
ITT N = 36 89 36 8 28 22 42 24/27 (89)
Primary: trial information: 44687907.
EGJ 72 92 38 8 31 27 35 91
GB 28 80 30 10 20 10 60 75
Histology:
BSC alone BSC + chemo
Intestinal 67 92 46 13 33 17 38 90
Mixed/Diffuse 33 83 17 0 17 33 50 86 Pts (deaths) 22 (20) 23 (17).
HER2: Median age 78.5 79
+ve 17 100 50 17 33 33 17 100 % PS ‡2 68 57
-ve 83 87 33 7 27 20 47 86
Genotype:
% frail; % very frail 96; 68 91; 70
6/6 50 83 33 17 17 28 39 93 Mean baseline EQ5D QL (scale 0-1) 0.64 0.61
6/7 44 94 44 0 44 19 38 83 Median OS mo unadjusted 3.0 6.1
7/7 6 100 0 0 0 0 100 -- OS adjusted Cox model HR= 0.69 [95%CI: 0.32-1.48], p=0.34
PET Response*: 89 92 100 100 100 100 63
Mean QL @ 9wks 0.37 0.45

4052 Poster Session (Board #157), Mon, 8:00 AM-11:00 AM 4053 Poster Session (Board #158), Mon, 8:00 AM-11:00 AM
The quality of life in neoadjuvant versus adjuvant therapy of esophageal Molecular comparison between peritoneal metastases (PM) and primary
cancer treatment trial (QUINTETT). First Author: Richard Malthaner, London gastric (GC) and gastroesophageal junction (GEJ) cancer. First Author:
Health Sciences Centre, London, ON, Canada Matthew K Stein, West Cancer Center, U Tennessee, Memphis, TN
Background: We compared the health-related quality-of-life (HRQOL) of Background: PM from GC or GEJ portend a poor prognosis and molecular
standard neoadjuvant cisplatin and 5-FU chemotherapy plus radiotherapy (N) differences are ill defined. Methods: We compared genomic profiles of primary
followed by surgical resection to adjuvant cisplatin, 5-FU, and epirubicin (P) GC and GEJ with PM patients (pts) and other metastases (OM) sent to Caris
chemotherapy with concurrent extended volume radiotherapy (A) following Life Sciences. Testing comprised immunohistochemistry (IHC) including
surgical resection for resectable esophageal carcinoma. Methods: 96 patients programmed death ligand 1 (PD-L1) combined positive score (CPS), copy
with stage I to III resectable cancer of the esophagus were enrolled into a number alterations (CNA), 592-gene next-generation sequencing (NGS),
prospective randomized trial (NCT00907543) from April 2009 to November microsatellite instability (MSI) and tumor mutational burden (TMB).
2016. Patients were randomized into 2 groups: N (47 cases) and A (49 cases). Results: 1366 cases were identified: 1041 GC (707 P, 98 PM, 236 OM) and
The primary end point was HRQOL using the FACT-E at one year. The sec- 325 GEJ (248 P, 5 PM, 72 OM). PM were increased in GC versus GEJ (9% v.
ondary endpoints included other HRQOL measures, overall survival (OS), 2%, p , 0.0001). 91% GC and 93% GEJ were adenocarcinoma (AD); GC were
disease-free survival (DFS), and adverse events. Results: The median follow-up more likely signet ring (SR) histology versus GEJ (11% v. 3%, p , 0.0001) and
was 5.0 years [95% CI :4.6 to 5.5]. The majority of patients had adenocar- GC PM were more likely SR versus other OM or P (13% v. 12% v. 7%, p =
cinomas of the distal esophagus/gastroesophageal junction (80.9% vs 0.067). The mean age of PM pts (57 years) was younger than primary GC (63,
87.8%). The stage distribution was: I 9%; II 22%; III 58%; TxN0-1 10%. p = 0.002) and OM (61; p = 0.044). More PM GC pts were female than P or OM
Using an intention-to-treat analysis there was no significant difference in the (48% v. 35% v. 34%, p = 0.03). No molecular profiling differences were seen
FACT-E total scores between arms at one year (p = 0.759), with 35.5% vs. between GEJ and GC pts and they were combined for analysis; findings
41.2% respectively showing an increase of $ 15 points (a priori minimal from 1246 AD pts are shown below (see Table). OM (9%, p = 0.041) had
clinical difference) compared to pre-treatment (p = 0.638). The HRQOL was more CNA in CCNE1 than PM (2%, p = 0.041) or P (5%, p = 0.002).
temporarily significantly inferior at 2 months in the N arm for FACT-E, EORTC Conclusions: Compared to P and OM GC, PM pts were younger, more likely
OG25, and EQ-5D-3L in the dysphagia, reflux, pain, taste, and coughing female and had a higher incidence of SR histology. PD-L1, HER2 IHC and
domains (p , 0.05). There were no 30-day mortalities but 2.1% vs. 10.2% 90- ERBB2 CNA were reduced in PM versus P, suggesting novel therapeutic
day mortalities (p = 0.204). There were no significant differences in either 5- targets are needed.
year OS (37.9% vs 28.9%, p = 0.321) or DFS (34.0% vs 25.5%, p = 0.551.
p-value p-value
48.9% of patients required chemotherapy to be modified or stopped in the N Characteristic PM (N = 87) P (N = 893) (PM v. P) OM (N = 266) (OM v. P)
arm compared to 57.1% in the A arm (p = 0.421). 51.1% of patients were able
TMB ‡10 mutations/Mb (%) 18/87 (21) 220/885 (25) 0.388 65/261 (25) 0.988
to complete the prescribed N arm chemotherapy without modification com- PD-L1 CPS 1+ (%) 26/83 (31) 412/855 (48) 0.003 98/256 (38) 0.005
pared to only 14.3% in the A arm (p , 0.001). Chemotherapy related adverse PD-L1 CPS ‡10 (%) 10/83 (12) 120/855 (14) 0.617 35/256 (14) 0.883
events significantly more frequent in the neoadjuvant arm (p , 0.05). Surgery MSI-high (%) 3/86 (3) 72/889 (8) 0.125 10/266 (4) 0.016
No. genes mutated ‡1% (%) 46/592 (8) 43/592 (7) 0.635 42/592 (7) 0.874
related adverse events were significantly more frequent in the neoadjuvant arm CDH1-mutated (%) 11/86 (13) 53/883 (6) 0.016 18/265 (7) 0.640
(p , 0.05). Conclusions: Trimodality therapy is challenging for patients with MUTYH-mutated (%) 3/87 (3) 4/891 (0) 0.002 5/266 (2) 0.020
resectable esophageal cancer regardless if it is given before or after surgery. ERBB2-mutated (%) 2/87 (2) 28/893 (3) 0.665 2/266 (1) 0.032
HER2-positive IHC (%) 2/81 (2) 70/796 (9) 0.048 15/236 (6) 0.232
Less toxic protocols are needed. Clinical trial information: 00907543. ERBB2 CNA (%) 1/85 (1) 69/837 (8) 0.019 19/255 (7) 0.684

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Noncolorectal) Cancer 235s

4054 Poster Session (Board #159), Mon, 8:00 AM-11:00 AM 4055 Poster Session (Board #160), Mon, 8:00 AM-11:00 AM
Comparison of efficacy and safety of second-line palliative chemotherapy Safety results of a phase III randomized trial of comparison of three paclitaxel-
with paclitaxel plus raltitrexed and paclitaxel alone in patients with meta- based regimens concurrent with radiotherapy for patients with local advanced
static gastric adenocarcinoma: A randomized phase II trial. First Author: esophageal squamous cell carcinoma (ESO-Shanghai 2). First Author: Dashan
Xiaoying Zhao, Department of Medical Oncology, Fudan University Shanghai Ai, Fudan University Shanghai Cancer Center, Shanghai, China
Cancer Center; Department of Oncology, Shanghai Medical College, Fudan
Background: Paclitaxel (PTX) is effective in concurrent chemoradiation (CCR)
University, Shangai, China
against esophageal squamous cell carcinoma (ESCC) . Which regimen, among
Background: Paclitaxel is a microtubule stabilizing agent that has been the cisplatin (DDP) (TP), carboplatin (CBP) (TC) or 5-Fu (TF) in combination with
standard second line chemotherapy in the treatment of advanced gastric PTX concurrent with radiotherapy, provides best prognosis with minimum
cancer. This study was designed to find out the clinical outcome of paclitaxel adverse events (AEs) is still unknown. Methods: The study compared two pairs
plus raltitrexed regimen as second line treatment in MGC patients. Methods: In of regimens: TF vs. TP and TF vs. TC concurrent with radiotherapy. Patients
an open, randomized, multi centers phase II clinical trial , 148 patients were with histologically confirmed ESCC (clinical stage II, III or IVa) were ran-
randomly assigned and treated with either RP (raltitrexed 3 mg/m2 d1 and domized into the three groups. Patients in TP group were treated with 2 cycles
paclitaxel 135mg/m2 d1,3w) or P (paclitaxel 135mg/m2 d1,3w) as second- of CCR followed by 2 cycles of consolidation chemotherapy with TP (DDP
line palliative chemotherapy. The primary endpoint is PFS, secondary endpoint 25 mg/m2/d, d1-3, PTX 175 mg/m2, d1, q28d). Patients in TF group were
is ORR, OS and safety. Results: In 148 randomly assigned and treated patients treated with 6 cycles of TF (5-Fu 300 mg/m2, civ 96h, PTX 50 mg/m2, d1, qw)
(RP = 73; P = 75), the majority of patients were males (94 vs. 54). Progression in CCR followed by 2 cycles of TF (5-FU 1800 mg/m2, civ 72h, PTX 175 mg/
free survival has a tendency to be prolonged with RP versus P (2.7m vs. 1.7m, m2, d1, q28d) in consolidation chemotherapy. Patients in TC group were
p = 0.148). Overall survival also has a tendency to be prolonged with RP versus P treated with 6 cycles of TC (CBP AUC = 2, d1, PTX 50 mg/m2, d1, qw) in CCR
(10.2m vs. 6.1m, p = 0.140). Overall response rate was equal with RP versus P followed by 2 cycles of TC (CBP AUC = 5, d1, PTX 175 mg/m2 d1, q28d) in
(6.8% vs.4.0%, p = 0.72). DCR in the RP group was 56.2%, P group was consolidation chemotherapy. The radiotherapy dose in all groups was 61.2 Gy
36.0%. Grade 3 to 4 treatment-related adverse events occurred in 36.2% delivered in 34 fractions. The primary endpoint was overall survival and the
(RP) v 28.2% (P) of patients. Frequent grade 3 to 4 toxicities for RP v P were: secondary endpoints were progression-free survival and adverse events.
neutropenia (11.0% v 4.0%), anemia (1.4% v 4.0%), thrombocytopenia Results: Between July 2015 and January 2018, 321 ESCC patients in 11
(1.4% v 5.3%), and all grade peripheral neurotoxicity (12.3% v 17.3%),all centers were enrolled. TP group had a significant higher incidence of acute
grade elevated aminotransferase (27.4% v 14.1%). Subgroup analysis shows grade 3/4 neutropenia (59.7% vs. 16.8%(TF) or 32.4%(TC)), thrombocyto-
if the disease combined with ascites or peritoneal involved , OS of RP regimen penia (12.7% vs. 3.5%(TF) or 6.2%(TC)), anemia (6.4% vs. 4.4%(TF) or
is more longer (p = 0.05). Conclusions: Second-line palliative chemotherapy 4.4%(TC)), fatigue (10.0% vs. 0.9%(TF) or 0.9%(TC)) and vomiting (5.5% vs.
with paclitaxel plus raltitrexed provides a tendency to prolong PFS and OS, and 0%(TF) or 0.9%(TC)) than other two groups (P , 0.05). TF group had a
the patients with ascites or peritoneal involved may get benifits from combined significant higher incidence of grade 3/4/5 esophagitis (13.1% vs. 1.8%(TP)
chemotherapy, which needs to be confirmed by larger sample studies. Clinical or 5.3%(TC)) and pneumonitis (4.4% vs. 0%(TP) or 1.8%(TC)) than other two
trial information: NCT02072317. groups (P , 0.05). One patient in TF group died of acute pneumonitis. One
patient in TF group and one in TC group died of acute esophagitis.
Conclusions: TP and TF regimen showed different severe AEs in CCR in ESCC
patients and TC showed mild AEs. Clinical trial information: NCT02459457.

4056 Poster Session (Board #161), Mon, 8:00 AM-11:00 AM 4057 Poster Session (Board #162), Mon, 8:00 AM-11:00 AM
Phase I trial of hyperthermic intraperitoneal chemoperfusion (HIPEC) with A pilot study of neoadjuvant FOLFIRINOX followed by chemoradiation for
cisplatin, mitomycin, and paclitaxel in patients with gastric adenocarcinoma gastric and gastroesophageal cancer: Preliminary results. First Author:
and carcinomatosis or positive cytology. First Author: Brian D. Badgwell, Jennifer Yon-Li Wo, Massachusetts General Hospital, Boston, MA
University of Texas MD Anderson Cancer Center, Houston, TX
Background: We performed a single-arm pilot study of total neoadjuvant
Background: This phase I trial evaluated the safety and toxicity of laparoscopic approach including FOLFIRINOX and chemoradiation (CRT) with concurrent
hyperthermic intraperitoneal perfusion with chemotherapy (HIPEC) combining carboplatin/taxol (C/T) followed by surgery in patients with locally advanced
mitomycin, cisplatin, and paclitaxel for patients with gastric cancer metastatic gastric or gastroesophageal junction (GEJ) cancer. Methods: Patients were
to the peritoneum. Methods: A Bayesian optimal interval design was used to enrolled on a NCI sponsored, prospective, single arm study (NCT03279237).
identify the maximum tolerated dose (MTD) of escalating doses of paclitaxel Key eligibility criteria included: histologically confirmed T3/4 or lymph node
(starting dose of 20 mg/m2 to maximum dose of 60 mg/m2) in combination (LN) positive gastric or GEJ cancer, ECOG PS #1, age 18+, life expectancy .
with flat doses of mitomycin (30 mg) and cisplatin (200 mg) during laparo- 3 months. Exclusion criteria included: visceral metastases, prior chemo-
scopic HIPEC for patients with gastric adenocarcinoma metastatic to the therapy or RT, or prior targeted therapy. Extensive LN disease beyond the
peritoneum. The primary endpoint was MTD. Secondary endpoints included surgical field (supraclavicular or para-aortic) was permitted if deemed feasible
surgical complications and overall survival (OS). Results: A total of 27 patients to be encompassed within a RT field. Laparoscopy was not required. Pts were
were treated from November 2017 through November 2018. No dose-limiting treated with neoadjuvant FOLFIRINOX x 8, restaging, CRT (45 Gy for gastric,
toxicities were found. Treatment-related grade 1-2 side effects were leuko- 50.4 Gy for GEJ) with concurrent C/T, restaging, followed by surgical resection.
penia (11%), oral dysesthesia (4%), arthralgia (4%), and diarrhea (4%). Dose reductions were at discretion of the treating physician. The primary
Treatment-related grade 3-4 side effects included leukopenia (4%) and objective was to determine the rate of completion of FOLFIRINOX x 8 followed
neutropenia (4%). The MTD was 60 mg/m2. Clavien-Dindo surgical compli- by CRT delivered in the preoperative setting. Secondary endpoints included: 1)
cations were grade I (all representing electrolyte deficiencies requiring re- acute toxicity and 2) pathologic complete response (pCR). Results: From Oct
placement) in 96% of patients, grade II in 4%, grade III in 0%, grade IV in 0%, 2017 to June 2018, 25 pts were enrolled. Median age was 60 (range:30-76),
and grade V in 4%. At a median follow-up of 15 months, the median OS from 17 pts were male (68%). All pts started FOLFIRINOX; 23 (92%) pts completed
diagnosis of metastatic disease and the date of surgery has not been reached. all 8 planned cycles. Two pts did not complete the planned 8 cycles due to
One- and 2-year OS rates from the date of metastatic disease were 74% and metastatic progression. Rates of grade 3+ overall, gastrointestinal, and he-
58%, respectively. The 1-year OS rate from the date of HIPEC was 51%. matologic toxicities were 28%, 12%, and 28% respectively. Of the entire
Conclusions: Laparoscopic HIPEC with mitomycin, cisplatin, and paclitaxel cohort, 23 (92%) pts started chemoRT and 22 (88%) pts completed chemoRT
appears safe at intraperitoneal doses of 30 mg, 200 mg, and 60 mg/m2, (1 pt died during CRT due to pulseless electrical activity arrest). All 22 pts
respectively. Although electrolyte abnormalities are common, systemic toxicity (88%) who completed CRT went for surgical exploration, of whom 2 pts were
of this therapy is modest. Survival rates are promising, supporting further found with intraoperative metastases. Therefore, 20 (80%) pts underwent
research into intraperitoneal therapy for stage IV gastric cancer. Clinical trial surgical resection. At time of abstract, 1 pt’s pathology is in process; 7 pts
information: NCT03330028. had a pCR (37% in resected cohort, 28% in ITT cohort), all with R0 resection.
Conclusions: Total neoadjuvant FOLFIRINOX followed by CRT is feasible with
acceptable rates of treatment completion and grade 3+ toxicity. In our small
series, the rate of pCR is promising and a follow-up study is currently planned.
Clinical trial information: NCT03279237.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
236s Gastrointestinal (Noncolorectal) Cancer

4058 Poster Session (Board #163), Mon, 8:00 AM-11:00 AM 4059 Poster Session (Board #164), Mon, 8:00 AM-11:00 AM
Safety and efficacy of durvalumab following multimodality therapy for locally Prospective evaluation of metabolic intratumoral heterogeneity using 18F-
advanced esophageal and GEJ adenocarcinoma: Results from Big Ten Cancer FDG-PET-CT in patients with advanced gastric cancer receiving palliative
Research Consortium study. First Author: Hirva Mamdani, Karmanos Cancer chemotherapy. First Author: Shin Hye Yoo, Department of Internal Medicine,
Institute, Detroit, MI Seoul National University Hospital, Seoul, South Korea
Background: Concurrent chemoradiation (CRT) followed by esophagectomy Background: Metabolic intratumoral heterogeneity (ITH) gives important
is a standard of care for locally advanced esophageal (LA-EAC) and GEJ ad- information on treatment response and prognosis. However, temporal
enocarcinoma. Approximately 50% of patients (pts) experience disease relapse changes in metabolic ITH and their associations with treatment outcome
within the 1st yr after treatment(tx) completion. No adjuvant tx has been shown have not been reported yet in gastric cancer (GC). We aimed to evaluate the
to improve survival in these pts. Immune checkpoint inhibitors have activity in early changes in metabolic ITH and their predictive roles in advanced GC
metastatic PD-L1 positive EAC. Preclinical studies have shown upregulation of patients receiving palliative chemotherapy. Methods: Unresectable locally
PD-1/PD-L1 pathway with RT +/- chemotherapy. Methods: We conducted a advanced or metastatic GC patients were prospectively enrolled before the
phase II trial evaluating safety and efficacy of durvalumab (durva), a mono- first-line palliative chemotherapy and underwent 18F-fluorodeoxyglucose
clonal antibody against PD-L1, in pts with LA-EAC and GEJ adenocarcinoma positron emission tomography (18F-FDG-PET-CT) at baseline (T1) and at
who have viable tumor in surgical specimen after neoadjuvant CRT and the first response evaluation follow-up (T2). SUVs (Standardized uptake
R0 resection. Pts received durva 1500mg IV every 4 weeks for up to 1yr. values), volumetric parameters, and textural features including entropyHisto
Results: 24 pts were enrolled from 4/2016-1/2018 (median age: 60yrs (range, and contrastGLCM were extracted from the primary gastric tumor at T1, T2,
43-70). 18 received carbo/paclitaxel and 6 received cis/5-FU concurrently with and DT (T2-T1) was evaluated. Associations of these parameters with
radiation. Staging at diagnosis: T2N0 (n=3, 12.5%), T2N2 (n=3, treatment response, progression-free survival (PFS), and overall survival (OS)
12.5%),T3N0 (n=6, 25%), T3N1 (n=6, 25%), T3N2 (n=4, 17%), T3N3 (n=1, were analyzed. Results: 87 patients were analyzed. Of 86 evaluable pa-
4%), T3Nx (n=1, 4%).19 pts (79%) had positive lymph nodes (LNs) at the time tients, 44 obtained partial response, 33 stable disease, and 8 progressed.
of surgery following CRT. 12 pts completed 1yr of tx, 12 came off tx before 1yr The objective response rate was 51.8% (95% confidence interval [CI],
because of relapse(6), AEs(5), and consent withdrawal (1). Median number of 40.7% to 62.7%). The median PFS and OS were 7.3 months (95% CI, 5.4 to
tx cycles was 12.5 (range, 2-13). Most common AEs were fatigue (n=8, 33.3%) 8.2 months) and 11.5 months (95% CI, 10.1 to 14.3 months), respectively.
and nausea (n=6, 25%). 3pts (12.5%) developed grade 3 irAEs: pneumonitis From T1 to T2, metabolic ITH was significantly reduced (P , 0.01), and the
(1), hepatitis (1), colitis (1). At median follow up of 14.5 mo (range, 1.7-24mo), degree of decrease was greater in responders than in non-responders (P ,
17 are disease free (including 5 who came off tx before 1yr). 7pts (29%) have 0.01). By multiple Cox regression analyses adjusted for clinical variables,
relapsed (3 alive, 4 died). 6/7pts had distant relapse (lung, brain, bone, cervical low entropyHisto at T2 (P= 0.001), larger decreases in coefficient of variance
LNs) and 1 had locoregional relapse. 1-yr RFS and OS were 79.2% and 95.5%, (P= 0.003) and contrastGLCM (P= 0.017) were associated with better
respectively. 2-yr OS was 59.2%. RFS probability at 26 mo was 67.9%. Median PFS. Low SUVpeak at T2 (P= 0.001), larger decreases in coefficient of
survival after relapse was 11.1 mo (range, 0.1-11.3mo). Conclusions: Adjuvant variance (P= 0.032) and being a responder were associated with better OS.
durvalumab following trimodality therapy for LA-EAC and GEJ adenocarcinoma Conclusions: Early reduction in metabolic ITH is useful to predict response
was safe and feasible with improvement in 1-yr RFS to 79.2% compared to to palliative chemotherapy, PFS and OS in advanced GC patients.
historical rate of 50%. OS results are encouraging in this high risk pt pop-
ulation. Clinical trial information: NCT02639065.

4060 Poster Session (Board #165), Mon, 8:00 AM-11:00 AM 4061 Poster Session (Board #166), Mon, 8:00 AM-11:00 AM
Phase I study of fluzoparib, a PARP1 Inhibitor in combination with apatinib Phase 1 open label trial of intraperitoneal paclitaxel (IPP) in combination with
and paclitaxel in patients (pts) with advanced gastric and gastroesophageal intravenous cisplatin (C) and oral capecitabine (X) in patients with advanced
junction (GEJ) adenocarcinoma. First Author: Jian-Ming Xu, Department of gastric cancer and peritoneal metastases. First Author: Sina Vatandoust,
GI Oncology, the Fifth Medical Center, General Hospital of PLA, Beijing, Flinders Medical Centre, Flinders University, Adelaide, Australia
China
Background: IPP is a potential treatment option in patients with gastric cancer
Background: Fluzoparib (SHR3162) is an oral, selective PARP1 inhibitor. In with peritoneal metastasis. IPP with a dose of 20 mg/m2 is well tolerated in
our gastric cancer PDX model, fluzoparib + apatinib + paclitaxel demonstrated combination with S1; however, this has been achieved in a specific genetic pool
significant tumor growth inhibition as compared to apatinib alone, and flu- (Japanese population), and S1 is not available in some countries. We in-
zoparib + paclitaxel. In this phase I study, we hypothesized that the combi- vestigated the Maximum Tolerated Dose (MTD) of IPP in addition to a standard
nation of fluzoparib+apatinib+paclitaxel should be safe and active in pts with chemotherapy combination in an Australian population. Methods: Study
advanced gastric and GEJ cancer. Methods: Dose-escalation phase (P1) ex- population included synchronous or metachronous metastatic HER-2 non-
plored 4 dose levels of fluzoparib with a 3+3 design to identify a recommended amplified gastric adenocarcinoma with histologically/cytologically proven
phase II dose (RP2D) for further study. Pts received fluzoparib (20, 30, 40, 60 peritoneal involvement and adequate organ function. Intra-peritoneal catheter
mg/twice daily)+apatinib (250mg/day)+paclitaxel (60mg/m2, Day1, 8, 15). was placed surgically. 3 + 3 standard dose-escalation design was used. MTD
Dose-expansion phase (P2) was to assess safety and efficacy. Pts received was defined as the highest dose level at which # 33% of patients had Dose
RP2D of fluzoparib+apatinib+paclitaxel until progression or intolerant toxicity. Limiting Toxicity (DLT). DLT was defined within the first 3 cycles. Recom-
Treatment was repeated every 4 weeks. Pts had to have progressive disease mended Phase-2 Dose was defined as equal to the MTD, or cohort-3 dose if MTD
after standard platinum-based regimen treatment. Adverse events (AE), PK, was not reached. Treatment: maximum of six 21-day cycles of C (80mg/m2 IV
and response per RECIST 1.1 (every 8 wks in pts with measurable disease) day 1) + X (1000mg/m2 PO BD days 1-14) + IPP (days 1 and 8). IPP doses for
were assessed. Results: 39 pts (median age 58) have been treated in P1 and Cohort-1, 2 and 3 were 10, 20 and 30mg/m2 respectively. Primary endpoint
P2, including fluzoparib 20mg (n=4), 30mg (n=27; 6 pts in P1, 21 pts in P2), was the MTD of IPP. Secondary endpoints included safety, tolerability, overall
40mg (n=6), and 60mg (n=2). The median treatment duration for this study response rate, ascites response rate, progression free survival and overall
was 2.8 months. No DLTs were reported in 20mg cohort. One DLT occurred in survival. Results: 15 patients were recruited in 3 cohorts: 9 males (60%),
30 mg cohort (grade 3 [G3] hypophosphatemia), 1 DLTs (1 grade 4 [G4] febrile median age at study entry: 61y (range 32-82). All had synchronous metastatic
neutropenia) occurred and 1 G4 neutropenia occurred and recovered in 3 days disease and were chemo-naı̈ve. Cohort-1 expanded to 6 patients due to 1 DLT
in 40mg cohort, 2 DLTs (1 G4 neutropenia, 1 G4 febrile neutropenia) in 60mg (grade 3 diarrhea), cohort-2 included 3 patients (no DLT) and cohort-3 was
cohort. Therefore, 40 mg dose was deemed the MTD. There were no treatment- expanded to 6 patients as planned and 1 DLT occurred (febrile neutropenia).
related deaths on study. The most common AEs$G3 were neutropenia, febrile MTD was not reached and Recommended Phase 2 Dose was determined as
neutropenia, and hypertension. 1 treatment-related discontinuation was ob- 30mg/m2. 8 patients (53%) completed all 6 cycles of treatment. The last
served. Of 36 evaluable pts, 12 (30.0%) had confirmed partial response and patient on the study has completed 3 cycles and is expected to complete 6
13 had stable disease (36.1%). Median progression-free survival was cycles by April 2019. No grade 4 or 5 toxicity was recorded. Conclusions: MTD
4.9 months. PK analysis will be presented. Conclusions: The RP2D of com- of IPP was not reached. IPP is safe in combination with C + X and the Rec-
bination of fluzoparib + apatinib + paclitaxel is well tolerated and has activity in ommended Phase 2 Dose is 30 mg/m2. Survival data will be presented when
pts with advanced gastric and EGJ cancer who have failed to platinum-based available. Clinical trial information: ACTRN12614001063606.
regimen. Clinical trial information: NCT 03026881.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Noncolorectal) Cancer 237s

4062 Poster Session (Board #167), Mon, 8:00 AM-11:00 AM 4063 Poster Session (Board #168), Mon, 8:00 AM-11:00 AM
A digital pathology demonstration of an "immune hot" ICOS+/CD45RO+ Landscape of innate and adaptive immunity targets in oesophagogastric
immunephenotype and the impact on survival in patients with esophageal adenocarcinoma (OGA). First Author: Elizabeth Catherine Smyth, Cambridge
adenocarcinoma. First Author: Matthew Philip Humphries, Queens Uni- University Hospital NHS Foundation Trust, Cambridge, United Kingdom
versity Belfast, Belfast, United Kingdom
Background: Anti-PD-1 therapy modestly improves survival in chemorefractory
Background: Therapies targeting immune checkpoints are changing our OGA. Combining PD-1 blockade with novel checkpoint inhibitors, T-cell co-
understanding of the biology and treatment of cancer. Analysing the immune stimulatory molecules, or myeloid suppressors could enhance PD-1 inhibition.
landscape in esophageal adenocarcinoma (EA) may help future prognosti- Herein, we explore the landscape of known targetable immune markers in non-
cation and therapeutic decision-making. Methods: We assembled 310 EA Asian OGA. Methods: OGA patient biopsies were prospectively collected and
cases in a tissue microarray format with associated clinicopathological in- clinically annotated from 19 UK cancer centres (Oesophageal Cancer Clinical
formation, including a discovery cohort of 156 EA from Northern Ireland and Molecular Stratification – OCCAMs network). Genomic (WGS) and tran-
and a 154 EA validation cohort from Aberdeen. We carried out validated scriptomic (bulk-RNA seq) data were generated using Illumina and processed
immunohistochemistry (IHC), stained for range of adaptive immune (CD3, using a validated in-house pipeline (Frankell, Nat Genetics 2019). Gene
CD4, CD8 and CD45RO) and immune checkpoint biomarkers (ICOS and expression was computed in transcripts per kilobase. Using unsupervised
IDO-1). Slides were digitised and assessed using QuPath image analysis clustering patient clusters were selected according to the expression of gene
software program to quantify their expression and correlate them with targets for immune therapy - PD-L1, LAG3, TIM3, TIGIT, ICOS, CCR2, CCR5,
outcome. Results: In the discovery cohort we identified a group of patients CXCR4, and CSF1R. Immune cell infiltration was extrapolated using GSVA
highly expressing several immune biomarkers, conferring a significant gene set enrichment analysis. Results: RNAseq data were available for 251
positive survival advantage (p = 0.022). CD3, CD4, CD8, CD45RO, and ICOS patients; 96% had operable tumours (Stage I: 6%, II: 63%, III: 22%, IV: 4%).
were individually prognostic for better overall survival (Log rank p = 0.0003; In untreated patients (n = 156) 3 subgroups were identified: immune low (83,
p = 0.0292; p = 0.0015; p = 0.0008; p = 0.0051 and p = 0.0264 re- 53%) with low level expression of all 9 markers, immune high (14, 9%) with
spectively). Multivariate and correlation analysis identified a subgroup of high expression of all or majority of markers and intermediate (59, 38%) with
CD45RO+/ICOS+ patients with significantly improved overall survival (p = heterogenous marker expression. Clinicopathological variables (sex, age,
0.0002). The co-expression of CD45RO+/ICOS+ immunophenotype was smoking, tumour location (gastric/GEJ/esophagus) and tumour regression
investigated in the validation cohort and a confirmed survival advantage was grade) were similarly distributed across subgroups. In a cohort of 114 patients
seen (p = 0.042). Additionally, the Opal Multiplex IHC technology revealed with matched WGS and RNAseq data tumour mutation burden was not dif-
the much higher frequency of single-cell, dual labelling of CD45RO+/ICOS+ ferent between subgroups. In post-chemotherapy biopsies (n = 95) a similar
in immune hot cases. Conclusions: These data demonstrate the advantage of co-expression pattern was observed. Gene enrichment analysis supported
immune markers other than the traditional CD3/CD4/CD8 in EA prognos- infiltration by cells of innate and adaptive immune system in immune high
tication. The fact that one of these biomarkers is an immune checkpoint patients. Neoantigen results, phenotypic immunohistochemistry and opti-
inhibitor may have therapeutic implications. mised survival outcomes according to lymphoid and myeloid target expression
will be presented. Conclusions: High level co-expression of immune regulatory
targets in OGA patients may limit the efficacy of anti-PD-1 monotherapy.
Combination immune directed therapies may be required in this patient group.

4064 Poster Session (Board #169), Mon, 8:00 AM-11:00 AM 4065 Poster Session (Board #170), Mon, 8:00 AM-11:00 AM
Comprehensive molecular characterization of clinical response in ramucirumab- Prospective validation of a serum miRNA panel for early detection of gastric
treated gastric cancer patients: Phase II trial with integrated genomic profiling. cancer. First Author: Lihan Zhou, MIRXES PTE LTD, Singapore, Singapore
First Author: Seung Kim, Samsung Medical Center, Sungkyunkwan University
Background: High mortality from gastric cancer is related to the late man-
School of Medicine, Seoul, South Korea
ifestation of its symptoms. A blood-based non-invasive biomarker with the
Background: The absence of tumor cells enables an enriched stromal envi- ability to detect all stages of gastric cancer could significantly improve
ronment to generate RNA signatures and through assembling genes involved in patient outcomes. We aimed to develop a novel serum miRNA assay for
tumor stroma, four distinct stromal signatures that reflected biological pro- diagnosis of gastric cancer. Methods: We conducted a multi-center study
cesses such as signature vascular mature (VM), vascular mature/inflammatory involving 892 gastric cancer and control subjects from Singapore and Korea
(VMI), vascular immature/noninflammatory (VINI) and inflammatory alone (I) to develop a multi-target miRNA assay. Using RT-qPCR, we quantified the
depending on mature of vasculature. We hypothesized that these stromal expressions of 578 serum miRNAs and constructed a 12-miR biomarker
specific signatures may provide additional information to the pre-existing panel through multi-variant data analysis. The results were generated with
TCGA/ACRG subtypes when predicting response to anti-angiogenesis agent the use of a logistic-regression algorithm, with the value of 40 or more
such as ramucirumab. Methods: We conducted a single-center phase II trial in considered to be positive. We subsequently validated this multi-miR assay
which we treated 61 unselected patients with metastatic GC with ramucirumab in a large prospective cohort involving 4566 subjects and compared its
plus paclitaxel as second line therapy and performed pre-planned integrated performance with traditional markers such as H.Pylori and Pepsinogen. All
genomic profiling. Results: Sixty-two patients were enrolled in this study participants underwent gastroscopy independent of the assay results.
between May 2014 and June 2017. The cut-off date for treatment outcome Results: Of the 4566 subjects that underwent gastroscopy and histopath-
analysis was January 2, 2019, at which time response evaluations were ological examination in the prospective cohort, 125 were diagnosed with
available for 57 patients with a median follow-up of 30.2months. In an intent- gastric cancer. The 12-miR assay achieved an Area-Under-Curve (AUC) of
to-treat analysis cohort, there was no CR and 22 patients achieved confirmed 0.84, significantly outperforming (p-value , 0.01) that of H.Pylori (AUC of
PRs resulting in an ORR of 35.5% (95% CI: 23.6 – 47.4). The response rate to 0.64) and Pepsinogen (AUC of 0.62). The sensitivity of the miRNA assay in
ramucirumab was considerably enriched in VM/VMI group (29.2%) (P = detecting early (stage 0-2) and late (stage 3-4) stage gastric cancer was
0.0003) when compared to I ( , 10%) or VINI group ( , 10%). The strongest 82.6% (95% CI, 68.6% to 92.2%) and 88.4% (95% CI, 78.4% to 94.9%)
response defined by maximal response to ramucirumab was shown in GC respectively at a specificity of 70.0% (95% CI, 67.8% to 71.9%). In
patients with VM/VMI signatures. Of note, VM/VMI patients had prolonged comparison, H.Pylori showed a sensitivity of 80.4% at a specificity of 44.3%
duration of response to ramucirumab/paclitaxel demonstrating that these whereas the Pepsinogen showed sensitivity of 9.52% at a specificity of
patients not only respond to ramucirumab but also had durable response. 95.3%. Using the miRNA assay as a pre-screening tool could potentially
Conclusions: This is the first study to demonstrate a clinically robust corre- reduce number of endoscopy needed by 62% in detecting one case of gastric
lation between stromal-based signature and response to anti-angiogenesis cancer. Conclusions: Our serum miRNA panel is a useful, non-invasive
inhibitor in GC. Clinical trial information: 02628951. screening test for gastric cancer. It is cost-effective as it can reduce un-
necessary diagnostic endoscopy.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
238s Gastrointestinal (Noncolorectal) Cancer

4066 Poster Session (Board #171), Mon, 8:00 AM-11:00 AM 4067 Poster Session (Board #172), Mon, 8:00 AM-11:00 AM
Neoadjuvant epirubicyn, oxaliplatin, capecitabine and radiation therapy The impact of postoperative complications on survival outcomes in patients
(NEOX-RT) followed by surgery for locally advanced gastric cancer (LAGC): with cT3/4a gastric cancer. First Author: Masanori Tokunaga, National
A phase II multicentric study. First Author: Antonino De Paoli, Radiation Cancer Center Hospital East, Kashiwa, Japan
Oncology Dept - IRCCS CRO Aviano-National Cancer Institute, Aviano, Italy
Background: Recently, the negative impact of postoperative complications on
Background: This study evaluates the feasibility, safety and efficacy of a long-term survival outcomes has been reported in patients with gastric cancer.
trimodality treatment, with surgery postponed after neoadjuvant chemotherapy However, most are single center, retrospective studies with different defini-
(CT) and chemoradiotherapy (CRT), in LAGC. Methods: Patients (pts) with tions of postoperative complications. The objective of this study was to
cT3-4 and/or N+ LAGC were eligible. Staging included endoscopic ultrasound, evaluate the impact of postoperative complications on long-term outcomes
PET-CT and laparoscopy. Three cycles of EOX (Epirubicyn 50mg/m2,q21 using the data of a multicenter randomized controlled trial (JCOG1001).
days, Oxaliplatin 130mg/m2,q21 days, and Capecitabine 625mg/m2 bid, by Methods: This study included 1191 out of all 1204 patients enrolled in
continuous oral administration (c.a.), followed by IMRT with 45Gy/25 frs, JCOG1001 which was aimed to confirm the superiority of bursectomy for
concurrent Capecitabine 625mg/m2 bid c.a. and weekly Oxaliplatin 30mg/m2 patients with cT3/4a locally advanced gastric cancer. Complications were
for 5 wks, was planned. Early PET-CT was performed after the 2nd EOX cycle to graded by Clavien-Dindo classification. The relationships between the grade
assess response or disease progression. Restaging was repeated after CT and ($grade II or $grade III) or type (all or intraabdominal infectious (pancreatic
CRT. Surgery was planned 4-6 wks after CRT, 22 wks from the start of NEOX- fistula, anastomotic leakage, and intra-abdominal abscess.)) of complications
RT. Pathologic complete response (pCR) was the primary endpoint. and survival outcomes were evaluated. Results: The incidences of $grade II
Results: From November 2008 to March 2016, 51 pts (5 G-E Junction, 17 and $grade III all complications were 23.0% and 9.7%, and those of $grade
Cardia, 15 Corpus, 14 Antrum) entered the study. The NEOX-RT program was II and $grade III intraabdominal infectious complications were 13.4% and
completed in 46 pts (90%) who proceeded to surgery and are assessable. 6.9%, respectively. The hazard ratios for overall survival (OS) of patients
Grade 3-4 toxicity (NCI-CTC criteria v.3) occurred in 13/51 pts (25%) during with $grade II and $grade III all complications and those of patients
EOX, including 1 toxic death, and 9.5% CT cycles required dose modification, with $grade II and $grade III intraabdominal infectious complications were
resulting in a CT compliance of 90%. No pts had progression during CT. shown in Table. With whichever definition we adopted, postoperative com-
Persistent G2-G3 toxicity occurred in 32/46 pts (69%) during CRT. However, plications were significantly associated with OS in both univariable and
41/46 pts (89%) received the planned 45Gy with Capecitabine at dose $75% multivariable analysis. Conclusions: Postoperative complication was identified
and 4-5 cycles of weekly Oxaliplatin in 52% pts. Curative resection (R0) rate as an independent prognostic factor in patients with cT3/4a gastric cancer.
was 89%; 4 pts (8.7%) had peritoneal carcinomatosis at surgery done after a Hazard ratios for overall survival by univariable and multivariable Cox pro-
median of 23 wks. pCR was reported in 9/46 pts (19.6%). Major postop portional hazard model. Clinical trial information: UMIN000003688.
complications occurred in 5 pts (11%). At median f-up of 62 mos (23-109),
5-yr OS and DFS in all and pCR pts were 58%, 100% and 51%, 75%, re- Hazard ratio [95% CI]
spectively. Conclusions: This trimodality program was feasible and safe. Most Univariable Multivariable
pts completed the planned treatment. The pCR rate of 19.6% was remarkable All complications $CD II (vs. ,CD II) 1.47 [1.10- 1.42 [1.05-
and met the hypothesis of pCR = 20%. A high R0 rate was also reported and 1.96] 1.92]
delayed surgery didn’t increase complications. The notable survival rates are $CD III (vs. ,CD III) 1.72 [1.18- 1.72 [1.17-
available to be compared with ongoing phase III trials. Clinical trial in- 2.51] 2.52]
formation: 2008-002715-40. Intraabdominal $CD II (vs. ,CD II) 1.46 [1.03- 1.49 [1.04-
2.06] 2.15]
infectious $CD III (vs. ,CD III) 1.64 [1.04- 1.67 [1.06-
complication 2.56] 2.64]

4068 Poster Session (Board #173), Mon, 8:00 AM-11:00 AM 4069 Poster Session (Board #174), Mon, 8:00 AM-11:00 AM
Prognostic significance of sarcopenia in metastatic esophageal squamous Impact of adjuvant therapy in patients with a microscopically positive margin
cell carcinoma. First Author: Kirsty Taylor, Princess Margaret Cancer Centre, after resection for gastroesophageal cancer. First Author: Lucy Xiaolu Ma,
Toronto, ON, Canada Princess Margaret Cancer Centre, Toronto, ON, Canada
Background: Sarcopenia is defined as low skeletal muscle mass and represents Background: A microscopically positive (R1) resection margin following
a quantifiable marker of frailty. Disease related symptoms of anorexia, nausea resection for gastroesophageal (GE) cancer has been documented to be a
and dysphagia, in addition to reduced physical activity contribute to muscle poor prognostic factor. The optimal strategy and impact of different mo-
wasting in metastatic esophageal squamous cell cancer (MESCC) patients. This dalities of adjuvant treatment for an R1 resection margin remain unclear.
study set out to evaluate the prognostic utility of sarcopenia and its association Methods: A retrospective analysis was performed for patients (pts) with GE
with nutritional indices. Methods: MESCC patients (pts) with available ab- cancer treated at the Princess Margaret Cancer Centre from 2006-2016.
dominal CT imaging, attending Princess Margaret Cancer Centre between 2011 Electronic medical records of all pts with an R1 resection margin were
and 2016, were identified from the institutional database. Skeletal muscle reviewed. Kaplan-Meier and Cox proportional hazards methods were used to
index (SMI), normalized by height, was calculated at the third lumbar (L3) analyze recurrence free survival (RFS) and overall survival (OS) with stage
vertebra using SliceOMatic software. SMI cutoffs for sarcopenia were 34.4cm2/m2 and neoadjuvant treatment as covariates in the multivariate analysis.
in females and 45.4cm2/m2 in males based on previously established con- Results: We identified 78 GE cancer pts with an R1 resection. 11% had
sensus. Nutritional risk index (NRI) was calculated using weight and albumin neoadjuvant chemotherapy, 14% chemoradiation (CRT), 75% surgery
with malnutrition defined as , 97.5. Results: Of the 58 pts analyzed, 26 alone. 28% had involvement of the proximal margin, 13% distal, 56%
presented with de novo MESCC, median age was 64 (range 48-85), 30 pts were radial, 3% had multiple positive margins. By the American Joint Committee
ECOG PS #1 and 45% received systemic therapy. 93% of pts experienced on Cancer 7th edition classification, 88% had a pT3-4 tumour, 66% pN2-3
weight loss . 5% in the 3 months preceding diagnosis and median BMI was nodal involvement, 64% grade 3, 68% with lymphovascular invasion. 3%
20.4 (range 16.3-34.9). Twenty-four (41%) pts were sarcopenic (SP) with were pathological stage I, 21% stage II and 74% stage III. Adjuvant therapy
differences in BMI and NRI (p , 0.05) compared to non-sarcopenic (NSP) pts. was given in 46% of R1 pts (24% CRT, 18% chemotherapy alone, 3%
Median BMI in SP pts was 18.9 (16.3-25.6), 46% had a BMI , 18.5 and none radiation alone, 1% reoperation). Median RFS for all pts was 12.6 months
were obese (BMI $ 30). By NRI, 58% of SP pts were malnourished. Males (95% CI 10.3-17.2). Site of first recurrence was 71% distant, 16%
comprised 71% of SP pts (p = 0.03) but no difference from NSP MESCC pts was locoregional, 13% mixed. Median OS was 29.3 months (95% CI 22.9-50)
identified with age, race, ECOG PS or smoking status with univariate analysis. for all pts. The 5 year survival rate was 23% (95% CI 12%-43%). There was
Median overall survival (OS) was 6 months; 4.2 in SP pts and 6.2 in NSP pts. no significant difference in RFS (log-rank test p = 0.63, adjusted p = 0.14) or
Significant difference was identified with NRI (p = .0.009) but not sarcopenia OS (log-rank test p = 0.68, adjusted p = 0.65) regardless of adjuvant therapy.
(p = 0.247) or BMI (p = 0.393). With a multi-variate Cox model for NRI and Conclusions: Most pts with positive margins after resection for GE cancer
sarcopenia, including age, sex, race, and ECOG PS, only ECOG PS was a had advanced pathologic stage and prognosis was poor. Our study did not
significant predictor of mortality, HR for 2-3 vs 0-1 of 5.4 (2.5-11.9) p , find improved RFS or OS with adjuvant treatment and only one pt had
0.001. Conclusions: Sarcopenia at diagnosis was not associated with OS. NRI reresection. The main failure pattern was distant recurrence, suggesting that
was superior to BMI alone with respect to discriminating pt outcomes, however pts being considered for adjuvant RT should be carefully selected. Further
ECOG PS was the only measure significantly associated with survival. studies are required to determine factors to select pts with good prognosis
despite a positive margin, or those who may benefit from adjuvant treatment.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Noncolorectal) Cancer 239s

4070 Poster Session (Board #175), Mon, 8:00 AM-11:00 AM 4072 Poster Session (Board #177), Mon, 8:00 AM-11:00 AM
A landscape of circulating tumor DNA in esophageal adenocarcinoma and First-line avelumab + axitinib in patients with advanced hepatocellular
squamous cell carcinoma. First Author: Kabir Mody, Mayo Clinic, Jacksonville, carcinoma: Results from a phase 1b trial (VEGF Liver 100). First Author:
FL Masatoshi Kudo, Kindai University, Faculty of Medicine, Osaka, Japan
Background: Esophageal cancer (EC) is a lethal malignancy with limited Background: Combining an immune checkpoint inhibitor with a targeted anti-
treatment options. Genomic analyses have led to the elucidation of numerous angiogenic agent may leverage complementary mechanisms of action for treatment
dysregulated genes in esophageal adenocarcinoma (AC) and squamous cell of advanced/metastatic (a/m) hepatocellular carcinoma (HCC). Avelumab is a hu-
carcinoma (SCC), and the potential for advancement of targeted therapies in man anti–PD-L1 IgG1 antibody with clinical activity in various tumor types; axitinib
this disease. Data regarding circulating tumor DNA (ctDNA) plasma analysis in is a tyrosine kinase inhibitor selective for VEGF receptors 1/2/3. VEGF Liver 100
EC in real-world clinical practice is limited. Methods: We performed ctDNA (NCT03289533) is a phase 1b study evaluating safety and efficacy of avelumab +
next-generation sequencing (NGS) analysis in patients (pts) with EC (January axitinib in treatment-naive patients (pts) with HCC; interim results are reported here.
Methods: Eligible pts had confirmed a/m HCC, $1 measurable lesion, a fresh or
2015- February 2018). ctDNA analysis was performed using Guardant 360
archival tumor specimen, ECOG PS #1, and Child-Pugh class A. Pts received
(Guardant Health, CA) which detects single nucleotide variants and insertion/ avelumab 10 mg/kg IV Q2W + axitinib 5 mg orally BID until progression, un-
deletion mutations, and specific amplifications and fusions, in up to 73 acceptable toxicity, or withdrawal. Endpoints included safety and objective response
different genes. The mutant allele fraction (MAF) for detected alterations was (RECIST v1.1; modified [m] RECIST for HCC). Results: Interim assessment was
calculated relative to wild type in ctDNA. Therapeutic relevance was defined as performed after a minimum follow up of 6 months based on the released study data
alterations within OncoKB levels 1-3B and R1. Results: Among 450 pts, 487 set (clinical cut-off date: Aug 1, 2018). As of the cut-off date, 22 pts (median age:
total samples were analyzed (77% AC, 31% SCC). ctDNA NGS revealed at 68.5 y) were treated with avelumab (median: 20.0 wk) and axitinib (median: 19.9
least one genomic alteration (excluding variants of uncertain significance and wk). The most common grade 3 treatment-related adverse events (TRAEs) ($10% of
synonymous mutations) in 81% of pts (90% AC, 88% SCC). Median number of patients) were hypertension (50.0%) and hand-foot syndrome (22.7%); no grade 4/
alterations per AC patient was 4 [range, 1-59] and a median MAF of 0.84% 5 TRAEs were reported. Immune-related AEs (irAEs) ($10% of pts) were hypo-
(range, 0.02% - 83.7%); SCC was 5 [range, 1-26], with a median MAF of thyroidism (31.8%) and hyperthyroidism (13.6%). No grade $3 irAEs were re-
0.99% (range, 0.01% - 85.2%). The total number of unique alterations was ported; no pts discontinued treatment due to TRAEs or irAEs. Based on Waterfall plot
1,162. The most commonly altered genes in AC: TP53 (70%), KRAS (20%), calculations, tumor shrinkage was observed in 15 (68.2%) and 16 (72.7%) pts by
ERBB2 (18%), EGFR (16%), PIK3CA (16%); in SCC: TP53 (88%), PIK3CA RECIST and mRECIST, respectively. ORR was 13.6% (95% CI, 2.9%-34.9%) and
(24%), CCND1 (23%), KRAS (21%), EGFR 15%). Therapeutically relevant 31.8% (95% CI, 13.9%-54.9%) by RECIST and mRECIST, respectively. OS data
were immature at data cutoff. Conclusions: The preliminary safety of avelumab +
alterations will be described. Conclusions: ctDNA plasma profiling of pts with
axitinib in HCC is manageable and consistent with the known safety profiles of
EC is a feasible alternative and non-invasive method to gather comprehensive
avelumab and axitinib when administered as monotherapies. This study demon-
genomic data. Further large comparison studies to assess landscape of ge- strates antitumor activity of the combination in HCC. Follow-up is ongoing. Clinical
nomic alterations observed through ctDNA versus tissue-based assays, in trial information: NCT03289533.
addition to studies of targeted therapy outcomes based on ctDNA-detected
alterations, are needed. RECIST
N = 22
mRECIST
N = 22
Confirmed ORR, % (n)* 13.6 (3) 31.8 (7)
95% CI 2.9-34.9 13.9-54.9
Median PFS, mo* 5.5 3.8
95% CI 1.9-7.3 1.9-7.3
6-mo PFS rate, %* 35.1 30.9
95% CI 15.3-55.8 12.5-51.5

*per investigator assessment.

4073 Poster Session (Board #178), Mon, 8:00 AM-11:00 AM 4074 Poster Session (Board #179), Mon, 8:00 AM-11:00 AM
Ramucirumab (RAM) for sorafenib intolerant patients with hepatocellular A phase II study of anti–PD-1 antibody camrelizumab plus FOLFOX4 or
carcinoma (HCC) and elevated baseline alpha fetoprotein (AFP): Outcomes GEMOX systemic chemotherapy as first-line therapy for advanced hepato-
from two randomized phase 3 studies (REACH, REACH2). First Author: cellular carcinoma or biliary tract cancer. First Author: Shukui Qin, Jinling
Josep M Llovet, Icahn School of Medicine at Mount Sinai, New York, NY Hospital, Nanjing, China
Background: Oral multikinase inhibitors that have shown improvements in overall Background: Advanced hepatocellular carcinoma (HCC) and biliary tract
survival (OS) in HCC are associated with clinically important toxicities that commonly cancer (BTC) patients (pts) have very limited treatment options. Considering
require dose adjustment or discontinuation (D/C) due to intolerance. REACH and the immunogenic effects of oxaliplatin, combination of camrelizumab with
REACH-2 studied RAM in patients (pts) with HCC who progressed on or were in- oxaliplatin-based chemotherapy might bring a better clinical benefit.
tolerant to sorafenib (SOR), and REACH-2 only enrolled pts with baseline AFP $400 Methods: That was an ongoing single-arm, multicenter phase 2 trial. Ad-
ng/mL. In REACH-2 RAM treatment (trt) improved OS compared to placebo (P), vanced HCC or BTC pts naive to systemic treatment were given camreli-
supporting findings in REACH pts with baseline AFP $400 ng/mL. An exploratory
zumab (3 mg/kg i.v., every 2 weeks) plus typical FOLFOX4 (infusional
analysis of outcomes by reason for D/C of SOR was performed. Methods: Pts had
fluorouracil, leucovorin and oxaliplatin) or GEMOX (gemcitabine and oxa-
advanced HCC, Child-Pugh A, ECOG PS 0-1, and prior SOR. Pts were randomized to
RAM 8 mg/kg or P Q2W. A pooled independent pt data analysis (stratified by study) of
liplatin) regimen. Primary endpoints were confirmed objective response rate
REACH-2 and REACH pts (AFP $400 mg/mL) was performed. Results are re- (ORR) per RECIST v1.1 and safety per CTC AE 4.03. Results: From Apr 27,
ported by reason for SOR D/C (intolerance or disease progression). OS and PFS were 2017 to Oct 31, 2018, 34 Chinese HCC and 47 BTC pts were treated, in
evaluated using Kaplan-Meier method and Cox proportional hazard model. Objec- which 27 (79.4%) HCC and 17 (36.2%) BTC pts were HBV-infected. In the
tive response rate (ORR), disease control rate (DCR) and safety are reported. 34 evaluable HCC pts, confirmed ORR was 26.5% and disease control rate
Results: Baseline characteristics in the pooled population were generally balanced (DCR) was 79.4%. Median time to response (TTR) was 2.0 mo (range
between trt arms in each subgroup. Median durations of prior SOR were 2.5 mo for 1.5–5.7). Six of the 9 responses were still ongoing, and median duration of
SOR intolerant (n = 70) and 4.0 mo for SOR progressors (n = 472). Median OS (RAM response (DoR) was not reached (range 3.3–11.5+ mo). Median progression-
v P) was 10.2 v 6.7 mo for SOR intolerant and 8.0 v 4.7 mo for SOR progressors free survival (PFS) was 5.5 mo. At data cutoff, 61.7% BTC pts were still
(Table). Rates of D/C due to trt-related adverse events (AEs) (Table) (7% in each receiving study drug. In the 43 evaluable BTC pts, with a median duration of
subgroup), and Grade $3 AEs (most frequently hypertension) were consistent with exposure of 2.9 mo, confirmed ORR was 7.0% and DCR was 67.4%. Median
those observed in each study. Conclusions: Acknowledging limitations of sample TTR was 1.9 mo (range 1.8–2.1). Median DoR was 5.3 mo (range 3.7–7.0).
size, the RAM trt benefit in SOR intolerant pts was consistent with that in the ITT Median PFS was not reached yet. Median estimates for overall survival in
population. RAM was well tolerated in SOR intolerant pts with low rates of D/C due to both HCC and BTC were also not reached. Grade $3 treatment-related
related-AEs. Clinical trial information: NCT01140347, NCT02435433. adverse events (TRAEs) occurred in 85.3% of HCC and 57.4% of BTC pts,
SOR intolerant SOR progressors most commonly neutrophil count decreased (HCC: 55.9%; BTC: 29.8%),
Analysis Population N = 70 N = 472 white blood cell decreased (HCC: 38.2%; BTC: 21.3%), platelet count
(RAM v P) (RAM 42, P 28) (RAM 274, P 198)
decreased (HCC: 17.6%; BTC: 12.8%), and anaphylaxis (BTC: 19.1%).
OS median, mo 10.2 v 6.7 8.0 v 4.7
HR (95% CI) 0.59 (0.34, 1.02) 0.71 (0.58, 0.88) Only one BTC pt stopped treatment due to a TRAE (recurrent Grade 2 anemia
PFS median, mo 4.4 v 1.4 2.7 v 1.6 related to FOLFOX4). Grade $3 immune-related AEs occurred only in
HR (95% CI) 0.32 (0.19, 0.55) 0.64 (0.52, 0.79) 5.9% of HCC (lipase increased) and 3.8% of BTC pts (anaphylaxis).
ORR, % 12 v 0 4v1
DCR, % 79 v 21 53 v 39 Conclusions: Camrelizumab plus FOLFOX4 or GEMOX chemotherapy was
D/C due to related AEs any grade, % 12 v 0 9v4 tolerable and might offer a new promising choice for advanced HCC and BTC
pts. Clinical trial information: NCT03092895.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
240s Gastrointestinal (Noncolorectal) Cancer

4075 Poster Session (Board #180), Mon, 8:00 AM-11:00 AM 4076 Poster Session (Board #181), Mon, 8:00 AM-11:00 AM
Multicentric prospettive study of validation of angiogenesis-related gene IGF-Child-Pugh score as a predictor of treatment outcome in advanced
polymorphisms in hepatocellular carcinoma patients treated with sorafenib: hepatocellular carcinoma patients treated with sorafenib. First Author:
Interim analysis of INNOVATE study. First Author: Andrea Casadei Gardini, Yehia I. Abugabal, University of Texas MD Anderson Cancer Center,
IRST-IRCCS, Meldola, Italy Houston, TX
Background: In the ePHAS study we analyzed three eNOSpolymorphisms and Background: Our recent published studies concluded that Lower levels of
at univariate analysis, patients with eNOS-786-TTgenotype had significantly Insulin like growth factors-I (IGF-I) is correlated with shorter overall survival
shorter median Progression Free Survival (PFS) and Overall Survival (OS) (OS) in HCC, and IGF-CP scores assigned based on serum bilirubin, serum
compared to those with other genotypes. On the basis of these preliminary albumin level, prothrombin time, and plasma IGF-1 provides better prognostic
results, our aim is to validate in a prospective study this data in patients with stratification. Sorafenib is the first frontline drug approved for the treatment of
HCC treated with sorafenib. Methods: This is a prospective Italian multicenter CP class A patients with advanced HCC. CP class A is the standard criterion for
study, that includes 141 HCC patients receiving sorafenib. We analyzed eNOS- active therapy and trials entry in HCC. In this study we aimed at evaluating the
786and itwas analyzed by Real Time PCR in relation to the primary end point predictive ability of IGF-CP to sub-stratify old CP classes and better predict
(OS). Event-time distributions were estimated using the Kaplan-Meier method sorafenib outcomes. Methods: Total of101 patients were prospectively en-
and survival curves were compared using the log-rank test. Results: 141 HCC rolled from MD Anderson Cancer Center (MDACC). Blood sample were col-
patients (122 males and 19 females), prospectively treated with sorafenib lected and tested for IGF-I and IGF-CP was calculated into class A, B and C.
from May 2015 to September 2018 were included. Median age was 69 years Median OS and progression free survival (PFS) were analyzed, and log rank test
(range 28-88 years). 120 patients had Child-Pugh A and 21 had Child-Pugh was used to compare PFS and OS between subgroups of IGF-CTP score of
B7. 43 had BCLC-B and 98 patients had BCLC-C. Atunivariate analysis, we patients. Results: Among CP class, patients who were reclassified as IGF-CP
confirmed that eNOS-786 TT genotype were significantly associated with a (B) (Old A/new B) had significantly shorter OS in months (m) was 7.6m (95%
lower median OS than the other genotypes (8.8 vs 15.7 months, HR 1.69, CI= 5.23-26.51m ) and PFS of 2.99m (95% CI=2.53-5.26m) with
95% CI 1.02-2.83 p=0.0424). Following adjustment for clinical covariates (P,0.001) in both, as compared to patients’ who classified as class A by both
(age, gender, etiology, BCLC stage, serum a-FP level, MELD score), multi- scoring systems (AA), who had OS of 15.43m (95% CI=12.3-31.18m) and
variate analysis confirmed eNOS- 786 and BCLC stage as the independent- PFS of 4.97m (95% CI=3.26-7.2m), (P,0.001) in both. Conclusions: IGF-
sprognostic factors predicting OS (TTvsTC+CC; HR: 2.39, 95% CI 1.14-5.03 CTP score sub-stratified CP A class, and provided better prognostic stratifi-
p=0.0211; C vs B;2.23, 95% CI 1.44-4.77 p=0.039). Conclusions: Our cation and accuracy than CP score in predicting sorafenib survival outcomes in
prospective study confirms the prognostic role of eNOS-786 in advanced HCC HCC. This approach may lead to a paradigm shift in predicting efficacy and
patients treated with sorafenib. Clinical trial information: NCT02786342. toxicity of systemic HCC therapies and in stratifying patients for active therapy
and selection in HCC clinical trials.

4077 Poster Session (Board #182), Mon, 8:00 AM-11:00 AM 4078 Poster Session (Board #183), Mon, 8:00 AM-11:00 AM
Randomized clinical trial of transcatheter arterial chemoembolization plus DNA damage repair (DDR) gene alterations as a predictive biomarker for
radiofrequency ablation versus transcatheter arterial chemoembolization for response to platinum-containing chemotherapy in advanced biliary tract
hepatocellular carcinoma with intermediate stage (BCLC stage B) hepato- cancer (BTC). First Author: Heejung Chae, Asan Medical Center, Seoul,
cellular carcinoma beyond Milan criteria. First Author: Xin Yin, Liver Cancer South Korea
Institute & Zhong Shan Hospital, Fudan University, Shanghai, China
Background: Although many studies using whole-exome sequencing or tar-
Background: To determine treatment efficacy and safety of transarterial geted sequencing have reported the molecular profile of BTC, its clinical
chemoembolization (TACE) combined with radiofrequency ablation (RFA) implications remains unclear. In this study, we assessed a predictive role of
(hereafter, TACE+RFA) in patients with intermediate stage (BCLC stage B) DDR gene mutations in advanced BTC patients treated with platinum-
hepatocellular carcinoma (HCC) beyond Milan criteria. Methods: In this containing regimen. Methods: Eighty-eight patients with pathologically-
randomized clinical trial, 110 patients with intermediate stage HCC beyond confirmed BTC who received first-line gemcitabine-cisplatin combination
Milan criteria (single tumor with diameter 5-7cm, median; 3-5 multiple (n = 69) or fluoropyrimidine-oxaliplatin combination (n = 19) were included in
nodules with diameter less than 5cm) were included and randomly assigned to this analysis. Targeted exome sequencing was performed using Foundation
TACE+RFA group (n=55) and TACE group (n=55) at liver cancer institute, Medicine T7 assay or in-house OncoPanel AMC. Germline or somatic muta-
Zhongshan hospital. The primary endpoint was overall survival (OS). The tions in ATM, ATR, BAP1, BARD1, BRCA1, BRCA2, BRIP1, CHEK2,
secondary end point was progression-free survival (PFS) , time to progress FAM175A, GEN1, MLH1, MSH2, MSH6, MRE11A, NBN, PALB2, PMS2,
(TTP) and best objective response (BOR). Results: The median OS in TACE+RFA RAD50, RAD51, RAD51C, RAD51D, and XRCC2 were classified as DDR gene
and TACE group were 29 and 18 months, respectively. The median TTP and mutations. Data regarding baseline characteristics and treatment outcomes
BOR were 15.7 months and 69.1 % in TACE+RFA group and 12.4 months were retrospectively obtained from medical records. Results: The median age
and 40.0 % in TACE group (P=0.004). The 1-, 3-, and 4-year overall sur- was 62 years (range, 25-78), with male comprising 64.8% (n = 57). By primary
vivals for TACE+RFA group and TACE group were 97.2%, 67.9% and 59.4% tumor site, 21 patients with GBC (23.9%), 44 with ICC (50.0%) and 23 with
versus 84.0%, 46.7% and 37.3% , respectively (P = 0.008). The corre- ECC (26.1%) were included. Most patients received palliative chemotherapy
sponding PFS were 47.3%, 27.2% and 21.7% versus 35.6%,15.3% and for their initially metastatic (50.0%) or recurred (44.3%) disease; the rest
11.4% , respectively (P = 0.04).The incidences of major complications in 5.7% had locally advanced disease. The median PFS and OS of overall pa-
TACE+RFA group were comparable to those in TACE group (P=0.14). tients were 7.1 and 16.1 months, respectively with median follow-up duration
Conclusions: TACE+RFA was superior to TACE in improving tumor response of 20.2 months. DDR gene mutations were found in 63.5% of patients. BRCA2
and overall survival for patients with intermediate stage (BCLC stage B) (18.2%) was most frequently mutated, followed by ATM (13.6%), and ATR
hepatocellular carcinoma beyond Milan criteria. Clinical trial information: (8.0%). DDR gene mutations were significantly associated with prolonged PFS
NCT03636620. (presence vs. absence; median, 6.9 vs. 5.7 months; P = 0.013) and OS
(median, 21.0 vs. 13.3 months, P = 0.009). The impact of DDR gene mu-
tations remained significant in multivariate analyses for PFS that included
other prognostic factors (hazard ratio, 0.51; P = 0.009), but not for OS.
Conclusions: The presence of DDR gene mutations might be a promising
predictive biomarker for response to platinum-based chemotherapies in ad-
vanced BTC. Future investigation using novel agents targeting DDR gene
alteration in BTC are warranted.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Noncolorectal) Cancer 241s

4079 Poster Session (Board #184), Mon, 8:00 AM-11:00 AM 4080 Poster Session (Board #185), Mon, 8:00 AM-11:00 AM
Pembrolizumab (pembro) for advanced biliary adenocarcinoma: Results Comprehensive genomic profiling in FIGHT-202 reveals the landscape of
from the KEYNOTE-028 (KN028) and KEYNOTE-158 (KN158) basket actionable alterations in advanced cholangiocarcinoma. First Author: Ian M.
studies. First Author: Yung-Jue Bang, Seoul National University College Silverman, Incyte Corporation, Wilmington, DE
of Medicine, Seoul, South Korea
Background: Genomic studies of cholangiocarcinoma (CCA) have identified
Background: Antitumor activity with pembro, an anti–PD-1 antibody, has been actionable alterations in multiple genes including IDH1, IDH2, FGFR2 and
observed in patients (pts) with advanced/metastatic biliary tract cancers (BTC), BRAF, but no targeted therapies have been approved for this indication.
who have limited treatment options. We present follow-up data from pts with Pemigatinib (formerly INCB054828) is a selective FGFR1-3 inhibitor cur-
advanced BTC treated with pembro in the KN158 (NCT02628067; phase 2) rently being evaluated in multiple tumor types, including advanced CCA
and KN028 (NCT02054806; phase 1) studies. Methods: Eligible pts $18 y in harboring FGFR2 rearrangements. Comprehensive genomic profiling (CGP)
the KN158/KN028 BTC cohorts had histologically/cytologically confirmed was used to identify and enroll advanced CCA patients with FGFR2 rear-
incurable advanced BTC that progressed after/failed any number of prior rangements into FIGHT-202 (NCT02924376). Here we provide an overview
standard treatment regimens, measurable disease per RECIST v1.1, ECOG PS of the genomic landscape of advanced CCA and identify actionable alter-
of 0/1, and no prior immunotherapy. PD-L1–positivity (membranous PD-L1 ations. Methods: CGP was performed on tumor samples from 1104 patients
expression in $1% of tumor and associated inflammatory cells or positive with advanced CCA using FoundationOne, a broad-based genomic panel
staining in stroma) was required for eligibility in KN028, but not KN158. Pts which identifies mutations, rearrangements, and amplifications in 315
received pembro 200 mg Q3W (KN158) or 10 mg/kg Q2W (KN028) for up to cancer genes. Results: The most frequently altered genes in advanced
2 y. Radiographic imaging occurred Q9W for 12 mo (KN158) or Q8W for 6 mo CCA were TP53 (38.1%), CDKN2A/B (28.8%), KRAS (21.9%), ARID1A
(KN028) and Q12W thereafter. Primary efficacy endpoint in both studies was (15.7%), SMAD4 (11.3%), BAP1 (10.6%), IDH1 (10.5%), PBRM1
ORR by RECIST 1.1. Response assessed by independent central review is (10.0%), FGFR2 (9.4%), ERBB2 (7.6%), PIK3CA (7.0%), MDM2/FRS2
reported. Results: Median (range) follow-up was 7.5 (0.6–29.5) mo in the 104 (5.8%), and BRAF (4.7%). FGFR2:BAP1 was the most significantly co-
pts from KN158 and 6.5 (0.6–33.1) mo in the 24 pts from KN028 with BTC. occurring alteration pair (odds ratio = 8.5; q-value = 1.08 x 10-13, Fisher’s
All pts in KN028 and 61 in KN158 had PD-L1–positive tumors. No pt had exact test). 42.9% of patients had at least one alteration for which a targeted
MSI-H tumors (not assessed in KN028). In KN158, ORR was 5.8% (6/104, all agent has been either approved or is under investigation. 91 (8.2%) patients
PR [including 1 pt with PD-L1–negative tumor]; 95% CI, 2.1%–12.1%) and had FGFR2 rearrangements, involving 44 unique partner genes, 37 (84.1%)
median duration of response (DOR) was not reached (NR; range, 6.2 to 23.2+ of which were observed only once. The most prevalent FGFR2 rearrangement
mo). Median OS and PFS were 7.4 mo (95% CI, 5.5–9.6) and 2.0 mo (95% CI, partner, BICC1, occurred in only 28 (30.7%) FGFR2 rearrangement positive
1.9–2.1). 12-mo OS rate was 32.7%. In KN028, ORR was 13.0% (3/23, patients. FGFR2 activating point mutations were found in 13 (1.2%) pa-
all PR; 95% CI, 2.8%‒33.6%) and median DOR was NR (range, 21.5 to tients. Of 1,091 evaluable patients for microsatellite instability (MSI) or
29.4+ mo). Median OS and PFS were 6.2 mo (95% CI, 3.8‒10.3) and 1.8 mo tumor mutational burden (TMB), only 10 (0.9%) were MSI-H and 13 (1.2%)
(95% CI, 1.4‒3.7), respectively. 12-mo OS rate was 27.6%. Grade 3–5 had high TMB ($ 20 mutations/megabase). None of the MSI-H or TMB-High
treatment-related AEs occurred in 13.5% in KN158 (1 pt had grade 5 renal patients had FGFR2, IDH1 or IDH2 activating alterations. Conclusions: The
failure) and 16.7% of pts in KN028 (no grade 5). 18.3% in KN158 and high frequency (42.9%) of patients with actionable alterations and myriad
20.8% of pts in KN028 had an immune-mediated AE or infusion reaction. FGFR2 rearrangement partners strongly support the use of fusion partner-
Conclusions: Pembro provides durable antitumor activity, regardless of PD-L1 agnostic CGP in advanced CCA.
expression, and manageable toxicity in a subset of pts with advanced BTC.
Clinical trial information: NCT02054806 and NCT02628067.

4081 Poster Session (Board #186), Mon, 8:00 AM-11:00 AM 4082 Poster Session (Board #187), Mon, 8:00 AM-11:00 AM
CT-based radiogenomic signature to identify isocitrate dehydrogenase (IDH) Efficacy and safety of pembrolizumab in patients with PD-L1 positive
1/2 mutations in advanced intrahepatic cholangiocarcinoma. First Author: advanced biliary tract cancer (BTC): A prospective cohort study. First Au-
Tagwa Idris, The University of Texas MD Anderson Cancer Center, Houston, thor: Junho Kang, Department of Oncology, Asan Medical Center, University
TX of Ulsan College of Medicine, Seoul, South Korea
Background: IDH1/2 mutations have a high prevalence (20%) in intra- Background: For patients with advanced BTC, standard chemotherapy has
hepatic cholangiocarcinoma (iCCA) and can be associated with therapeutic limited benefit and no molecular targeted agents have been approved.
benefit from IDH inhibitors. Radiomics, a developing field within imaging, Pembrolizumab is an anti PD-1 immune checkpoint inhibitor which has
has shown its ability to discriminate between tumors of distinct genomic shown modest activity for advanced BTC patients in prior single-arm phase I/
profiles and mutational status. Methods: We developed a radiogenomic II studies. Considering the heterogeneity of BTC, more data are needed to
signature to robustly predict IDH1/2 mutation status (mutated versus wild- evaluate the clinical outcomes of pembrolizumab in unresectable or met-
type [WT]) in 22 patients with iCCA using the pretreatment CT scans. The astatic BTC. Methods: In this prospective cohort study, 39 patients with PD-
triphasic hepatic CT scan was used to segment the lesion. After semi- L1 positive BTC who received pembrolizumab in Asan Medical Center,
automatic segmentation of the tumor, the extracted volume of interest (VOI) Seoul, Korea were included (ClinicalTrials.gov identifier, NCT03695952).
was imported into our in-house radiomic pipeline and 610 radiomic features PD-L1 expression was assessed using immunohistochemistry and PD-L1
were extracted. The least absolute shrinkage and selection operator re- positive tumors were defined as the expression of PD-L1 in $ 1% of tumor
gression (LASSO) and minimum redundancy and maximum relevance cells. Pembrolizumab was given at a fixed dose of 200 mg intravenously,
(mRMR) were used for feature selection. Selected features were used to every 3 weeks. Results: The median age was 61 years old (range, 41-76) and
build a classification model for prediction of IDH1/2 mutation status 22 (56.4%) patients were male. Intrahepatic cholangiocarcinoma (CCA) was
(XGboost). The performance of the radiomics model was assessed using the most common type (n = 18, 46.2%), followed by gallbladder cancer (n =
leave-one-out cross-validation (LOOCV). Results: Of 22 patients, 16 patients 12, 30.8%) and extrahepatic CCA (n = 9, 23.1%). Most of the patients had
(male, 6; female, 10; average age, 55.5 years) had IDH1 (N = 14) or IDH2 distant metastasis (n = 37, 94.9%). Pembrolizumab was administered as
(N = 2) mutations and 6 patients (male, 4; female, 2; average age = 55.5 years) 2nd-, 3rd- and 4th or greater line chemotherapy in 18 (46.2%), 16 (41.0%)
had IDH1/2 WT.The CT-derived radiomic signature robustly predicted and 5 (12.8%) patients, respectively, and median 2 cycles (range 1-10) of
presence of IDH1/2 mutations versus WT with an area under the curve (AUC), pembrolizumab were given. In 36 patients whose response was assessable,
sensitivity and specificity of 98.4%, 83.3% and 93.8%, respectively (P = partial response (PR) and stable disease were achieved in 4 (11.1%) and 13
0.037) and in a subgroup analysis presence of IDH1 mutation versus WT (36.1%), respectively. In 19 (52.8%) patients, progressive disease was the
with an AUC, sensitivity and specificity of 98.2%, 83.3% and 92.8%, re- best response. In patients with PR, the median time to response was
spectively (P = 0.035). Conclusions: To our knowledge, this is the first study 2.1 months (95% confidence interval (CI), 0.4 – 3.9). With a median follow-
investigating the ability of radiogenomics as a potential method to predict up duration of 4.4 months (95% CI, 2.4 – 6.4), median progression-free
the IDH1/2 mutation status in iCCA patients. Our data suggest that radio- survival and overall survival was 1.5 months (95% CI, 0.4 – 2.6) and
genomic signature may correlate with IDH1/2 mutations and represent a 4.3 months (95% CI, 2.6 – 6.1), respectively. No grade 3/4 adverse events
promising non-invasive tool to stratify the patients based on molecular (AEs) were reported and grade 1/2 fatigue (n = 4, 10.3%) was the most
alterations. common AE. Conclusions: In PD-L1 positive BTC, pembrolizumab showed
modest efficacy with 11.1% of response rates although our patients were
heavily pretreated. Considering the limited therapeutic options and poor
survival for these patients, further evaluation of immunotherapy including
biomarker analysis is needed.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
242s Gastrointestinal (Noncolorectal) Cancer

4083 Poster Session (Board #188), Mon, 8:00 AM-11:00 AM 4084 Poster Session (Board #189), Mon, 8:00 AM-11:00 AM
Final analysis of phase II trial of regorafenib (REG) in refractory advanced Clinical and molecular features of patients with cholangiocarcinoma har-
biliary cancers (BC). First Author: Dae Won Kim, H. Lee Moffitt Cancer boring FGFR genetic alterations. First Author: Alice Boileve, Institut Gustave
Center and Research Institute, Tampa, FL Roussy, Villejuif, France
Background: While gemcitabine plus cisplatin has demonstrated significant Background: Genetic alterations (GAs) in the fibroblast growth factor re-
antitumor activity as 1st line therapy of BC, there is no effective treatment ceptor (FGFR) pathway are emerging as promising therapeutic targets in
after failure of gemcitabine-based therapy. REG is an oral multi-kinase CCA. The clinical and molecular features of patients (pts) with CCA harboring
inhibitor that targets angiogenesis, oncogenesis and cancer proliferation/ FGFR GAs are reported here. Methods: A retrospective chart review was
metastasis. We evaluated the efficacy of REG in BC. Methods: Patients (pts) performed in pts with CCA who were found to have an FGFR GA on tumor
with histologically proven BC who progressed on at least one line of systemic molecular profiling as part of routine care. Data on demographics, risk
therapy received REG 160 mg daily 21 days on 7 days off, in 28 day cycles. factors, pathology, systemic therapy, radiographical response, progression
The primary endpoint was 6-month (mo) overall survival (OS) and the free survival (PFS), and overall survival (OS) were collected. Results: Among
secondary endpoints were median OS, progression free survival (PFS) and 65 pts, the median age at diagnosis was 55 years old (range = 27-92), and
response rates (RR). Pre and post-treatment plasma were collected for 38 (58%) pts were female, 63 (97%) had intrahepatic CCA, and 5 (11%)
cytokine evaluation. Results: A total of 39 pts received at least 1 dose of had chronic HBV. At presentation, 37% of pts had resectable disease. Of 47
REG; 32 pts were evaluable for efficacy. Median age was 62 (range: 27-88) pts with a known CA 19-9 at the time of initial diagnosis, 21 (45%) had a
years and the primary sites of tumor were intrahepatic cholangiocarcinoma value , 35U/mL. FGFR2 fusions were the most common FGFR GA (78%),
(68.8%), extrahepatic (18.8%), and gallbladder (12.5%). Pts were con- followed by FGFR2 mutations (14%), FGFR3 mutations (4%), FGFR3 fusion
sidered evaluable for efficacy if patients received more than 1 cycle of REG. (2%) and FGFR1 amplification (2%). The most common fusion partners
For 32 evaluable pts, 6 mo OS was 52% with median PFS of 2.8 mo (95% CI: were BICC1 (20%), POC1B (6%), SORBS1 (6%), DBP (4%), and TACC2
1.1-4.5) and median OS of 7.9 mo (95% CI: 0-18.7). Median PFS and OS of (4%). The most common co-alterations were in ARID1A, CDKN2A/B, TP53,
the pts (n=20) failed 1 line of therapy were 3.7 mo (95% CI: 3.2-4.1) and BAP1, IDH1, HER2, BRCA2, and PTEN. The median lines of palliative
13.8 mo (95% CI: 1.8-25.8), respectively. Median PFS and OS of the pts systemic therapies received was 3 (range = 0-8), and 9/65 (14%) pts had .
(n=12) failed 2 lines were 1.8 mo (95% CI: 1.63-1.97) and 4.5 mo (95% CI: 1 FGFR inhibitor (FGFRi). For the 30 (46%) pts with FGFR2 fusions who
2.6-6.3), respectively. RR was 9.4% (2 PR and 1 unconfirmed PR) and DCR received gemcitabine/platinum as first line palliative systemic therapy, the
was 62.5%. Total 71.8% of grade 3/4 adverse events (AE) were observed, median PFS was 4.7 months (95% CI: 2.1-6.0). In the overall population,
and the most common AE were fatigue (56.4%) and hypertension (53.8%). the median OS from time of initial diagnosis was 35.8 months (95% CI:
Dose modification was required in 49% of the pts. Among the 23 cytokines 29.7-52.7). Among 46 pts who received an FGFRi on a clinical trial and
analyzed, elevated baseline VEGF-A was associated with good prognosis (HR had $ 1 follow-up scan, the overall response rate (ORR) by RECIST v1.1 in
0.62, p=0.01). Elevated baseline TIMP-1 (HR 1.79, p=0.04) and IL-6 (HR pts with FGFR2 fusions, was 35.8% (14/39) on their first FGFRi; ORR was
1.33, p=0.05) were associated with poor prognosis. REG treatment de- 16.7% (1/6) for pts with FGFR2 mutations. Conclusions: Pts with CCA
creased BMP-9, GP130, VEGF-R2 and VEGF-R3 and increased IL-6, PIGF, harboring FGFR GAs were found to have a high rate of normal CA 19-9 and
TIMP-1, VCAM-1 and VEGF-A significantly. Conclusions: The primary short median PFS on first line gemcitabine/platinum compared to historical
endpoint was met in this study. VEGF-A may be further evaluated as a controls but additional comparative studies are necessary to evaluate these
predictive biomarker for REG in BC. Further randomized trials are warranted findings.
to confirm the efficacy and the correlative data. Clinical trial information:
NCT02115542.

4085 Poster Session (Board #190), Mon, 8:00 AM-11:00 AM 4086 Poster Session (Board #191), Mon, 8:00 AM-11:00 AM
Frequency of BRCA mutation in biliary tract cancer and its correlation with Efficacy and safety of FOLFIRINOX in advanced biliary tract cancer after
tumor mutational burden (TMB) and microsatellite instability (MSI). First failure of gemcitabine plus cisplatin: A phase II trial. First Author: Ali
Author: Gilbert Spizzo, Experimental Oncology, Tyrolean Cancer Research Belkouz, Amsterdam UMC, University of Amsterdam, Amsterdam,
Institute, Innsbruck, Austria Netherlands
Background: Biliary tract cancers constitute ~3% of cancers worldwide with Background: Currently there is no established standard treatment after
incidence increasing, especially for intrahepatic cholangiocarcinoma (IHC). failure of gemcitabine plus cisplatin (GemCis) for advanced biliary tract
The prognosis of these tumors remains dismal and novel treatment strategies cancer (BTC). Based on the efficacy of FOLFIRINOX in advanced pancreatic
are needed to improve overall survival. BRCA mutations occur in biliary tract cancer, which has histological and prognostic similarities with BTC, a Phase
cancers but their frequency in distinct sites of biliary tract cancer is unknown. 2 study was conducted to determine whether FOLFIRINOX is effective and
Moreover, no data are available correlating BRCA mutation with immunogenic safe in BTC. Methods: Patients with BTC and an ECOG PS of 0/1 who had
markers such as TMB, MSI, or PD-L1 expression. Methods: Tumor samples disease progression or unacceptable adverse events (AEs) after at least 3
from 1288 primary biliary tract cancers, comprising IHC (n = 746), extra- cycles of GemCis were included. Patients received oxaliplatin 85 ml/m2,
hepatic cholangiocarcinoma (EHC) (n = 189), gallbladder (GBC) (n=353) were irinotecan 180 mg/m2, leucovorin 400 mg/m2, fluorouracil bolus at 400 mg/
profiled at Caris Life Sciences, Phoenix, AZ. Testing included NextGen SEQ m2, followed by fluorouracil continuous infusion of 2400 mg/m2 over 46-
(MiSeq on 47 genes, NextSeq on 592 genes) and PD-L1 IHC (SP142). TMB hour every 2 weeks. This phase 2 study was conducted according to the two-
was calculated based on somatic nonsynonymous missense mutations, and stage Simon’s Design. Stage 2 was activated if at least 1 objective response
MSI was evaluated by NGS of known MSI loci. Results: BRCA mutations were rate (ORR) or 2 stable diseases were observed among 10 patients in stage 1
detected in 3.6% (N = 46) of samples (BRCA1 0.6%, BRCA2 3%), no dif- and a maximum of 3 patients had severe AEs within the first 6 weeks of
ferences were seen based on the site of the tumor. In GBC and IHC BRCA2 treatment. If more than 4 patients required a dose reduction in stage 1, stage
mutations (4.0% and 2.7%) were more frequent than BRCA1 (0.3% and 0.4, 2 was initiated with a standard dose reduction (fluorouracil bolus was
p , 0.05) while in EHC, similar frequency was observed (BRCA1: 2.1%; omitted and irinotecan reduced to 140 mg/m2). Primary outcome was ORR
BRCA2: 2.6%). There was no significant association with gender or age. In per RECIST 1.1 and secondary outcomes were overall (OS), progression free
BRCA-mutant biliary tract cancer the most frequently mutated genes were survival (PFS), and safety profile. Results: Forty patients were screened and
TP53 (55.6%), ARID1A (52.2%) and KRAS (26.1%), KMT2D/C (20%, 13%) 30 patients were included between May 2016 and July 2018. Median age
and CDKN2A(13%). Overall, BRCA mutations were associated with a higher was 60 years and 63% of patients were males. In stage 1, 5 patients
rate of MSI-H (19.5% vs 1.7%, p = 0.001) and higher TMB in both MSI-H and required a dose reduction within the first 6 weeks due to AEs, leading to
MSS tumors (p,0.05). When investigated separately, BRCA association with initiation of stage 2 with modified FOLFIRINOX after inclusion of 10 pa-
elevated TMB was seen in IHC and EHC, but not in GBC. No correlation was tients. The partial response rate was 10% (3/30), disease control rate 67%
seen with PD-L1 expression. TP53, KMT2D/C, RB1, PTEN, KDM6A mutations and median OS and, PFS were 10.7 and 6.2 months, respectively. Most
and FGFR1 amplifications were significantly higher in BRCA mutated tumors common grade 3/4 adverse events include neutropenia (50%), anemia
(p , 0.05). Conclusions: BRCA mutations are found in a significant subgroup (17%), diarrhea (13%), thrombocytopenia (10%), and deviated liver
of biliary tract tumors and are associated with an immunogenic tumor profile. function tests (10%). Conclusions: This is the first Phase 2 study with
These data provide rationale for trials testing PARP inhibitors in combination modified FOLFIRINOX in BTC showing promising disease control rate, OS,
with immunotherapy and targeted therapies in patients with BRCA-mutant and PFS, with an acceptable safety profile. Modified FOLFIRINOX is cur-
biliary tract cancers that are MSS. rently tested as a first-line treatment for patients with BTC in the ongoing
randomized prospective Phase 2/3 AMEBICA trial (NCT02591030). Clinical
trial information: NCT02456714.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Noncolorectal) Cancer 243s

4087 Poster Session (Board #192), Mon, 8:00 AM-11:00 AM 4088 Poster Session (Board #193), Mon, 8:00 AM-11:00 AM
Profiling of 3,634 cholangiocarcinomas (CCA) to identify genomic alter- Association of adverse events (AEs) with efficacy outcomes for cabozantinib
ations (GA), tumor mutational burden (TMB), and genomic loss of hetero- (C) in patients (pts) with advanced hepatocellular carcinoma (aHCC) in the
zygosity (gLOH). First Author: Milind M. Javle, University of Texas MD phase III CELESTIAL trial. First Author: Ghassan K. Abou-Alfa, Memorial
Anderson Cancer Center, Houston, TX Sloan Kettering Cancer Center, New York, NY
Background: The management of CCA has evolved as targeted and immune Background: Class-specific AEs occurring with tyrosine kinase inhibitors
checkpoint inhibitor (ICPI) therapies have emerged. We used comprehensive have been associated with improved efficacy outcomes in several tumor
genomic profiling (CGP) to characterize the genomic alterations (GA) that have types including aHCC. In the phase 3 CELESTIAL trial (NCT01908426), C,
potential to personalize therapy for CCA. Methods: 3634 CCA underwent an inhibitor of VEGFR, MET, and AXL, improved overall survival (OS) and
hybrid capture based CGP on 0.8-1.1 Mb of the coding genome to identify GAs progression-free survival (PFS) vs placebo (P) in pts with previously treated
in exons and select introns in up to 404 genes, TMB, microsatellite status aHCC. Here, we retrospectively evaluate the association of palmar-plantar
(MSI) and % monoallelic genome (gLOH). PD-L1 expression was determined erythrodysaesthesia (PPE) and hypertension (HTN) with OS and PFS for C in
by IHC (Dako 22C3). Results: 52% of CCA were female with a median age of the CELESTIAL trial. Methods: 707 pts with aHCC were randomized 2:1 to
62 years (range 16 - . 89). The most common biopsy sites were liver (74%), receive 60 mg C or P once daily. Eligible pts had Child-Pugh score A, ECOG
lymph node (4%), bile duct (3.3%), and lung (2%). MSI-high was rare (1%), PS #1, must have received prior sorafenib, and could have received up to
118 and 47 cases had TMB . 10 and . 20 mut/mb respectively. Of the latter, two prior regimens of systemic therapy for HCC. OS and PFS with C were
51% (24/47) were MSI-H. PD-L1 amplification (AMP) was present in 0.27%. evaluated for pts with any grade PPE or grade $3 HTN within the first
Of 490 CCA tested, 43 (9%) were positive for PD-L1 expression. 11% of cases 8 weeks of study treatment. Results: Overall, 374 (80%) pts in the C arm and
had gLOH . 16%, only 2 cases had both TMB . 20 and gLOH . 16%. GA 179 (76%) pts in the P arm completed $8 weeks of treatment. In the first
were most common in TP53 (31%), CDKN2A (29%), KRAS (20%) and 8 weeks, 188 (40%) of C-treated pts developed any grade PPE vs 11 (5%) of
ARID1A (17%). Potentially targetable GAs included FGFR2 (11%, 85% fu- P-treated pts, and 61 (13%) of C-treated pts developed grade $3 HTN vs 3
sions), BRAF (5%, 50% V600E), ERBB2 (5%, 72% AMP), MET (2%, 90% (1%) of P-treated pts. Median OS with C was 14.4 mo for pts with any grade
AMP), EGFR (0.52%) and rarer ( , 0.5%) FGFR3, RET, FGFR1, ALK, and PPE vs 8.4 mo for pts without PPE (HR 0.59, 95% CI 0.47-0.74), and
ROS1 fusions. The FGFR2 fusions had 128 unique 3’ partner genes including median PFS with C was 6.5 mo vs 3.7 mo, respectively (HR 0.63, 95% CI
BICC1 (26%), CCDC6 (3.2%), AHCYL1 (2.6%) and KIAA1217 (2.6%). 0.51-0.78). Median OS with C was 16.1 mo for pts with grade $3 HTN vs
FGFR2 fusions occurred in a mutually exclusive fashion from high gLOH (p , 9.5 mo for pts without grade $3 HTN (HR 0.56, 95% CI 0.39-0.80), and
0.002), but not high TMB. GA in IDH1 (15%) were mutually exclusive of median PFS with C was 7.4 mo vs 4.4 mo, respectively (HR 0.59, 95% CI
FGFR2 fusions (p , 1e-13), but co-occurred with PBRM1 GA (23%, p , 1e- 0.43-0.82). Some imbalances in baseline characteristics were present. Pts
21), ARID1A (26% p , 1e-10). IDH1 GA had gLOH similar to the overall CCA with PPE had better ECOG PS (60% vs 47% ECOG 0), better liver function
population but were enriched for low TMB (p , 1e-3). Conclusions: Nearly (48% vs 34% ALBI grade 1), and less macrovascular invasion (24% vs 30%)
20% of CCA cases harbor targetable kinase GA, half of which were FGFR2 than those without. Likewise, pts with grade $3 HTN had better ECOG PS
fusions. Independently, an additional 10% (gLOH) and 1% (high TMB, MSI (61% vs 51% ECOG 0), better liver function (56% vs 37% ALBI grade 1),
and/or PD-L1 AMP) may benefit from PARP inhibitors and ICPI respectively. and less macrovascular invasion (20% vs 29%) than those without.
Independently, co-mutation of IDH1 and PBRM1/ARID1A defines a class of Conclusions: The development of PPE or grade $3 HTN with C was asso-
CCA that warrants further investigation for sensitivity to PARP inhibitors and ciated with prolonged OS and PFS in pts with previously treated aHCC al-
may serve as a paradigm for other tumors (ie. gliomas) with a similar co- though some imbalances in baseline characteristics between comparator
occurrence landscape. groups were present. Clinical trial information: NCT01908426.

4089 Poster Session (Board #194), Mon, 8:00 AM-11:00 AM 4090 Poster Session (Board #195), Mon, 8:00 AM-11:00 AM
Nab-paclitaxel plus S-1 as first line treatment for advanced or metastatic Cell-free junctional DNA fragment from hepatitis B virus integration in HCC
biliary tract adenocarcinoma: A phase 2 study. First Author: Yongkun Sun, for monitoring postresection recurrence and clonality. First Author: Ya-Chun
Department of Medical Oncology, Beijing Key Laboratory of Clinical Study on Wang, TCM Biotech International Corp., New Taipei City, Taiwan
Anticancer Molecular Targeted Drugs, National Cancer Center/Cancer
Background: About one-third of patients suffer tumor recurrence within the
Hospital, Chinese Academy of Medical Sciences and Peking Union Medical
first year after surgical resection of HCC. Early recurrence compromised their
College, Beijing, China
overall survival. Timing detection of HCC recurrence and its clonality is
Background: Gemcitabine plus cisplatin or S-1 can be used as first-line required to implement therapeutic trials appropriately. This study examined
treatment for advanced or metastatic biliary tract adenocarcinoma. Multiple the virus-host chimera DNA (vh-chimera DNA), generated from junctions of
phase 2 studies found that gemcitabine, oxaliplatin, capecitabine, S-1 were HBV integration in HCC chromosome and released into blood, as a potential
not superior to gemcitabine plus cisplatin. Nab-paclitaxel plus S-1 was circulating biomarker for this clinical setting. Methods: We established a
effective and well-tolerated in pancreatic cancer. Methods: Patients with capture-next generation sequencing (NGS) platform to identify the HBV
pathological confirmed advanced or metastatic biliary tract adenocarcinoma integrations in 50 resected HBV-related HCC. For individual HCC, the major
(gallbladder carcinoma, intrahepatic cholangiocarcinoma ICC, extrahepatic clonal HBV integration sites were chosen to design specific primers for
cholangiocarcinoma ECC) were treated with Nab-paclitaxel plus S-1(Nab- droplet digital PCR (ddPCR) to detect and quantify the vh-chimera DNA in
paclitaxel 120mg/m2, d1 and d8; S-1 80-120mg/d, d1-14; q21d). Patients plasma samples, collected either just before surgery or two months after
that received PR or SD (RECIST1.1)after 6 cycles were given S-1 mainte- surgery. Levels of vh-chimera DNA were then correlated with baseline HCC
nance treatment. The primary endpoint was ORR. The study used Simon’s size or recurrence in one-year follow up. Results: We succeeded in detecting
Two Stage design. Results: From March 2016 to September 2018, we HBV integrations in the HCC from 44 out of 50 HBV-related HCC patients
recruited 54 patients, with 27 males（50%）. The median age was 58(34- (88%). The copy number of vh-chimera DNA in plasma at surgery from 42
73yrs). As of Dec 31 2018, the median treatment cycle was 4(1-6 cycles). 51 patients correlated with tumor sizes, with the detection limit at 1.5-2 cm.
patients were evaluable for efficacy: PR 14(27.5%), SD 22 (DCR=PR+SD: Among the plasma collected 2 months after surgery, 26.2% of samples
70.6%), PD 15 (29.4%). The median PFS was 6 months, and the median OS contained the same HCC signature vh-chimera DNA as baseline plasma,
was 13.2 months. The response rate varied in different tumor location: indicating a possible residual tumor. Consistently, 81.8% of them suffered
gallbladder carcinoma 53.8% (7/13), ICC 18.2% (6/33), ECC 20% (1/ HCC recurrence within one year. The signature vh-DNA in the plasma
5).Common grade 3/4 AEs were: leucopenia 17 (31.5%), hyper- suggested the majority of recurrences coming from the original HCC clones,
bilirubinemia 5(9.3%), Mucositis 4 (7.4%)，neurotoxicity 2 (3.7%), diarrhea whereas 2 from de novo ones. Conclusions: This study supported vh-chimera
2 (3.7%), omit 1(1.9%), fatigue 1 (1.9%), thrombocytopenia 1 (1.9%), ALT DNA as a new circulation marker for detecting the existence of most HBV-
increase 1 (1.9%). Conclusions: Nab-paclitaxel plus S-1 as first line treat- related HCC. This new biomarker may complement AFP to help detect re-
ment for advanced or metastatic biliary tract adenocarcinoma was effective sidual or recurrent HCC and their clonality after curative therapies.
and well-tolerated, especially for gallbladder carcinoma (ORR 53.8%). This
regimen need further exploration. Clinical trial information: NCT03830606.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
244s Gastrointestinal (Noncolorectal) Cancer

4091 Poster Session (Board #196), Mon, 8:00 AM-11:00 AM 4092 Poster Session (Board #197), Mon, 8:00 AM-11:00 AM
Novel staging system using carbohydrate antigen (CA) 19-9 in extrahepatic SHR-1210 plus GEMOX as first line treatment in biliary tract cancer: Results
cholangiocarcinoma (ECCC) and its implications on overall survival (OS). from a single-arm exploratory study. First Author: Xiaofeng Chen, The First
First Author: Sri Harsha Tella, University of South Carolina, Columbia, SC Affiliated Hospital of Nanjing Medical University, Nanjing, China
Background: Optimal management of ECCA patients with elevated CA19-9 Background: SHR1210 is a humanized anti-programmed cell death receptor
remains undefined. We hypothesized CA 19-9 elevation as a marker of ag- 1 (PD-1) antibody. We conducted a single arm exploratory study to evaluate
gressive biology in ECCA and that inclusion of CA 19-9 in the staging may the efficacy and safety of SHR-1210 plus gemcitabine and oxaliplatin
improve OS discrimination. Methods: Patients with ECCA with CA 19-9 levels (GEMOX) as first line treatment in patients (pts) with biliary tract cancer
reported to the National Cancer Database (years 2004-2015) were included. (BTC). Methods: Pts received SHR-1210 (3mg/kg, total dose #200mg, ivd,
The patients were classified based on their CA19-9 levels and a new staging D1/2W) combined with gemcitabine (800 mg/m2, ivd, D1/2W) and oxali-
system was proposed. Based on the current knowledge, we considered 37 U/ml platin (85mg/m2, ivd, D2/2W). Combined chemotherapy lasted for no more
as our cut off. Kaplan Meier method was used to compare OS between the than 8-12 cycles. Once chemotherapy intolerance occurred or at end of 12-
groups. The net reclassification improvement (NRI) model was used to assess cycle combined chemotherapy, pts with stable disease or objective response
the predictive improvement in the proposed survival model. Results: A total of would continue to take SHR-1210 as single agent until disease progression
2100 patients met the inclusion criteria: 601 (32%) and 1436 (68%) had or intolerable toxicity. Response was assessed every 8 weeks. Results: From
normal and elevated CA19-9 levels, respectively. Rates of chemo (p=0.16) and February 2018 to Dec 15, 2018, 32 eligible pts were recruited, and 27 pts
radiation therapy (p=0.07) were similar between groups, but patients with who had been treated for more than 2 months were included in this analysis.
elevated CA19-9 were less likely to undergo resection. Resected patients with Median age was 64 (range 47-75) years. 16 pts were bile duct cancer, while
CA19-9 elevation had higher 30-day mortality (p=0.02) and lower median OS 11 pts were gallbladder cancer. 26 pts can be evaluated for efficacy. Twelve
(p,0.01). Patients with elevated CA 19-9 levels had decreased stage-specific pts achieved partial response (46.15%), 12 pts stable disease (46.15%),
survival in all stages (p,0.01). On adjusted analysis, CA19-9 elevation in- and 2 pts progressive disease. Pts with gallbladder cancer had the trend of
dependently predicted poor OS (HR: 1.67 [1.42-1.97]) with impact re- higher objective response (63.64% vs 33.33%, p = 0.23) than those with
sembling nodal metastasis, positive resection margin, lymphovascular cholangiocarcinoma. 19 pts had tissue sample for next generation se-
invasion, and non-receipt of surgery or chemotherapy (p,0.01). CA 19-9 quencing. Gallbladder cancer had the tendency of higher median tumor
included proposed staging system (Table) had a better OS discrimination over mutation burden (TMB) than cholangiocarcinoma (8.1mut/Mb vs 5.4mut/
AJCC 7th edition. The new staging system had a concordance of 60% as Mb, p=0.33). Pts with high TMB(.8.6 mut/Mb, based on geenseeq BTC
opposed to 58% for the AJCC staging, leading to an improvement of 2% database) had significantly higher objective response than low TMB (100%
(p,0.01). NRI of 46% (95% CI: 39-57) indicates that the new staging system vs 26%, p=0.0294). The most common grade $3 adverse events were
is substantially effective at re-classifying events at 12 months as compared to nausea (18.52%),increased GGT (gammaglutamyltransferase,18.52%)，
AJCC staging. Conclusions: Elevated CA19-9 was found to be an independent hypokalemia(18.52%) and fatigue (18.52%). Conclusions: SHR-1210 plus
risk factor for mortality in ECCA and its inclusion in the newly proposed staging GEMOX showed promising efficacy with tolerable adverse events for BTC pts.
system markedly improved OS discrimination. Gallbladder cancer pts seem to benefit more from this treatment. Tumor
New proposed staging system. mutation burden may be a predictive factor for immunotherapy. Clinical trial
New Stage information: NCT03486678.
Stage 1 AJCC Stage 1 or AJCC Stage 2
Stage 2 AJCC Stage 3 or AJCC Stage 1+r or AJCC Stage 2+r
Stage 3 AJCC stage 3+r
Stage 4 AJCC Stage 4
r= elevated CA 19-9 $ 38 U/ml.

4093 Poster Session (Board #198), Mon, 8:00 AM-11:00 AM 4094 Poster Session (Board #199), Mon, 8:00 AM-11:00 AM
Impact of tumor shrinkage pattern by biweekly triplet gemcitabine/cisplatin/s- The prognostic role of soluble transforming growth factor-b related with
1 for biliary tract cancers: Implication for neoadjuvant therapy (KHBO1401-1A soluble programmed death-ligand 1 in biliary tract cancer. First Author:
study). First Author: Shogo Kobayashi, Departments of Surgery Osaka Uni- Jin Won Kim, Seoul National University Bundang Hospital, Seoul, South
veristy, School of Medicine, Osaka, Japan Korea
Background: There have not been any new evidenced regimen for biliary Background: We previously reported that soluble programmed death-Ligand
tract cancers (BTC) after ABC-02 study, we conducted biweekly triplet 1 (sPD-L1) at pre-chemotherapy indicated the prognostic value for overall
gemcitabine/cisplatin/s-1 regimen (GCS) and compared with conven- survival (OS) and the dynamics of sPD-L1 during palliative chemotherapy
tional doublet gemcitabine/cisplatin regimen (GC) as phase III (KHBO1401) correlated with disease burden in biliary tract cancer (BTC). Transforming
study. Biweekly GCS was proved not only to prolong patients’ survival (HR growth factor (TGF) -b attenuates tumor response to PD1/PD-L1 inhibitors.
0.791 (90% C.I. 0.628-0.996), one-sided P = 0.046) but also to achieve Strategy of dual targeting of PD1/PD-L1 and TGF-b is now under in-
high response rate (42% versus 15%, P , 0.001) and good conversion rate vestigation. This study aimed to evaluate the association between soluble
(2.5% versus 0.0%), and would be good for neoadjuvant therapy. Herein, we TGF-b (sTGF-b) and sPD-L1, dynamics during chemotherapy and its
investigated tumor shrinkage pattern to explore possibilities of neoadjuvant prognostic role in BTC. Methods: Study population consisted of 90 BTC
therapy. Methods: Totally 246 patients were enrolled in multi-center phase patients treated with first line chemotherapy. Blood samples at pre-and post-
III KHBO1401 study between 2014 and 2016. Tumor shrinkage pattern chemotherapy and at disease progression (PD) were prospectively collected.
(best response, timing, response at 100 days (14 weeks, approx. 6 cycles in Plasma sTGF-b and sPD-L1 levels were measured by using an enzyme-linked
GCS and 4 cycles in GC), etc.) and survival were investigated in the patients immunosorbent assay. Results: The median progression free survival (PFS)
with measurable BTC (n = 183, 74%, 91 in GCS and 92 in GC) as sub- and OS of all patients was 6.9 months (m) (95% CI, 5.2-8.6) and 11.5 m
analysis. P , 0.05 was considered statistically significant. Results: Tumor (95% CI, 9.4-13.6). The best response was CR in 7 (7.8%), PR in 20
shrinkage pattern could be divided to 4 categories by the response at (22.2%), SD in 52 (57.8%), and PD in 11 patients (12.2%). The mean
100 days after enrollment; category A ( , -30% in size), B (-30% to 0%), C baseline sTGF-b and sPD-L1 were 16.4 ng/ml and 1.3 ng/ml. There was a
(0% to +20%), and D ( . +20%). GCS arm contained more category A & B positive association between sTGF-b and sPD-L1 in terms of baseline levels
(61 (67%) vs. 33 (36%), P , 0.0001). Each category predicted best re- and changes after chemotherapy (at pre-chemo, Pearson correlation =
sponse and overall survival (p , 0.0001). Timing for maximum tumor re- 0.578, p , 0.001; change after chemotherapy, Pearson correlation =
sponse were different among categories, category A achieved maximum 0.542, p , 0.001). Patients with higher pre-chemotherapy sPD-L1 ( . 1.3
tumor shrinkage at 165 +/- 76 days in GCS and 225 +/- 190 days in GC, ng/ml) showed worse OS (9.2 vs 16.2 m, p , 0.001). Both sPD-L1 (1.8 vs
category B at 139 +/- 78 versus 154 +/- 82 days, and category C and D did 1.0 ng/ml, p , 0.001) and sTGF-b (20.5 vs 11.6 ng/ml, p , 0.001) were
not achieve tumor shrinkage. Maximum tumor shrinkage in category A increased significantly at the time of PD compared with pre-chemotherapy.
was -53% in GCS versus -65% in GC (P = 0.0892), and 20% patients in GCS Regarding changes after chemotherapy, increased sTGF-b after chemo-
underwent tumor regrowth 154 +/- 143 days later. Conclusions: GCS pro- therapy (D . 3.2 ng/ml) had worse prognosis (PFS: 5.1 vs 7.3 m, p = 0.024;
vided faster and more tumor shrinkage with better survival in the comparison OS: 9.2 vs 12.3 m, p = 0.028). This prognostic value of change of sTGF-b
of GC, although it had 20% risk of re-growth after 6 cycles. after chemotherapy was also significant in multivariable analysis with other
clinical factors (PFS: HR = 1.78, p = 0.022; OS: HR = 1.86, p = 0.018).
Conclusions: In BTC, there is a positive association between sTGF-b and
sPD-L1 value in terms of baseline levels and changes after chemotherapy.
sTGF-b could be associated with the survival, particularly, increased value
after chemotherapy indicates worse prognosis.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Noncolorectal) Cancer 245s

4095 Poster Session (Board #200), Mon, 8:00 AM-11:00 AM 4096 Poster Session (Board #201), Mon, 8:00 AM-11:00 AM
A phase I study of H3B-6527 in hepatocellular carcinoma (HCC) or intrahepatic Non-invasive detection of acquired resistance to FGFR inhibition in patients
cholangiocarcinoma (ICC) patients (pts). First Author: Teresa Macarulla Mercade, with cholangiocarcinoma harboring FGFR2 alterations. First Author:
Vall d’Hebron University Hospital and Vall d’Hebron Institute of Oncology, Anna M. Varghese, Memorial Sloan Kettering Cancer Center, New York, NY
Barcelona, Spain
Background: FGFR2 alterations are present in 14% of cholangiocarcinomas
Background: FGF19 overexpression is hypothesized to hyperactivate FGFR4 (CCA) and are promising targets of investigational FGFR-directed therapies.
and its downstream signaling pathway leading to enhanced tumor growth in Cell-free DNA profiling has emerged as a non-invasive approach to monitor
HCC/ICC. Targeting FGFR4 may have therapeutic benefit in HCC/ICC with disease and longitudinally characterize tumor evolution. We describe the use of
altered FGF19 signaling. A phase 1 study (NCT02834780) was initiated to circulating tumor DNA (ctDNA) among patients (pts) with FGFR2-altered CCA
assess H3B-6527, an investigational highly selective covalent FGFR4 in- receiving FGFR-targeted therapy in the identification of acquired FGFR2
hibitor. Methods: Adult pts with advanced HCC or ICC, ECOG PS 0-1, well mutations (mut) at resistance. Methods: Serial blood samples were collected
compensated liver function, and who progressed after at least one prior from 8 pts with FGFR-altered CCA for ctDNA isolation and next generation
therapy, were administered H3B-6527 orally QD (once daily) on a 21-day sequencing. Plasma ctDNA collected at baseline and resistance to FGFR-
cycle following a 3+3 design. Patients in the dose escalation phase were targeted therapy were sequenced using a custom ultra-deep coverage cfDNA
treated regardless of FGF19 status. Adverse events (AEs), pharmacokinetics panel, MSK-ACCESS, incorporating dual index primers and unique molecular
(PK), and pharmacodynamics (PD) were assessed. Response was de- barcodes to enable background error suppression and high-sensitivity mut
termined by RECIST 1.1 or modified RECIST every 6 weeks. Results: As of detection. The assay was enhanced to include all protein-coding exons and
06-Jan-2019, 37 pts have been treated with H3B-6527 at doses of 300 to relevant introns of FGFR2. In 5/8 pts, genomic profiling of an initial tumor
1400 mg QD (23 pts in escalation; 14 in expansion). In dose escalation, a biopsy was performed. Results: 8 pts with FGFR2-altered CCA (7 gene fusions,
total of 17 patients with HCC, Child-Pugh A received prior systemic therapy 1 amplification) were treated with FGFR-targeted therapies. 7/8 pts exhibited
including 100% with prior TKI and 35% with prior IO. 12% had hepatitis B stable disease or partial response. 19 total acquired mut in FGFR2 were
virus and 47% had hepatitis C virus. H3B-6527 plasma levels increased with detected at resistance in 5/8 pts (between 1-9 unique mut identified in each
dose from 300 to 1000 mg QD and plateaued. H3B-6527 was rapidly sample). All mut were located in the kinase domain. Conclusions: Acquired
absorbed with a tmax of ~2-3 h and showed a terminal half-life of ~4-5 h, mut in FGFR2 are seen in pts who have developed resistance to targeted
following administration of 1000 mg (fasted). No dose-limiting toxicities therapy. CtDNA can be used to identify these mut at the time of acquired
or $ Grade 3 treatment-related AEs (TRAE) have been observed in esca- resistance. The multitude of FGFR2 mut observed within individual pts suggest
lation. Most common TRAEs ($ 10%) were diarrhea, nausea, and vomiting. heterogeneity and evolutionary convergence of resistance mechanisms. Our
Based on safety, PK, and PD, 1000 mg QD was the recommended phase 2 results illustrate the utility of ctDNA as a less invasive way to monitor for signs
dose. Durable stable disease and partial responses (PR) have been observed of resistance and to identify other potential targetable alterations.
on the once daily fasted schedule; 2 of 17 pts with HCC achieved PRs and an Pt Baseline FGFR2 Alteration FGFR2 Acquired Resistance Mutations
additional 7 with stable disease were on treatment for $ 5 months.
1 FGFR2-KIAA1217 N549K, L550F
Conclusions: H3B-6527 is well tolerated and demonstrates early signs of 2 FGFR2-BICC1
clinical activity. Dose expansion on QD schedule and exploration of BID 3 FGFR2 amplification V564L
(twice daily) schedule is ongoing. Clinical trial information: NCT02834780. 4 FGFR2-WAC M538L, M537I, N549H, N549T, N549K, V564I, E565A, D650Y, K659Q
5 FGFR2-VCL
6 FGFR2-NRAP
7 FGFR2-KIAA1217 N549K, N549D, V564L, E565A, L617V, Q746L
8 FGFR2-DDX21 K659M

4097 Poster Session (Board #202), Mon, 8:00 AM-11:00 AM 4098 Poster Session (Board #203), Mon, 8:00 AM-11:00 AM
A phase II study of nivolumab in patients with advanced refractory biliary Randomized, open-label, perioperative phase II study evaluating nivolumab
tract cancers (BTC). First Author: Richard D. Kim, H. Lee Moffitt Cancer alone or nivolumab plus ipilimumab in patients with resectable HCC. First
Center & Research Institute, Tampa, FL Author: Ahmed Omar Kaseb, The University of Texas MD Anderson Cancer
Center, Houston, TX
Background: Biliary tract cancers (BTC) are often typically diagnosed at an
advanced stage. There is no established second line option for patients with Background: In HCC, surgical resection is associated with high recurrence
advanced BTC who have failed one prior systemic therapy. The phase II study rate, and no effective neoadjuvant or adjuvant therapies currently exist. On
evaluated safety and efficacy of nivolumab, anti PD-1 antibody in refractory the basis of of previous reports on the efficacy and safety of anti–PD-1
BTC patients. Methods: Pts with histologically proven BTC who progressed on (nivolumab) and anti–CTLA-4 (ipilimumab) antibodies against HCC, we
at least one line but no more than three lines of systemic therapy received initiated a randomized pilot trial of perioperative immunotherapy for re-
nivolumab 240mg IV q2weeks for 16 weeks and then 480 mg IV every 4 weeks sectable HCC. Methods: This is a randomized, phase II pilot trial of nivo-
until disease progression or unacceptable toxicity. The primary endpoint of the lumab (Arm A) or nivolumab + ipilimumab (Arm B) as pre-operative
study was objective response rate (ORR) by RECIST 1.1 every 8 weeks. The treatment for patients (pt) with HCC who are eligible for surgical resection. Pt
Simon two staged design was used to assess ORR.18 patients were accrued are given nivolumab 240 mg every 2 weeks (wk) for a total of 6 wk. Pt in Arm
and if one response was seen, the plan was to accrue additional 34 patients. B are treated concurrently with ipilimumab 1 mg/kg every 6 wk. Surgical
Secondary endpoints were PFS, OS and safety profile. Results: At data cutoff resection occurs within 4 weeks after last cycle of therapy. Pt continue
(Jan 14, 2018), 54 patients with BTC (female: 50%, Median age: 65 years) adjuvant immunotherapy for up to 2 years after resection. Primary objective
were enrolled. The primary sites of tumor were intrahepatic chol- is the safety and tolerability of nivolumab +/- ipilimumab. Secondary ob-
angiocarcinoma (63%) extrahepatic (11%), and gallbladder (26%). 30 pts jectives include overall response rate, complete response rate and time to
(56%) failed 1 line of therapy and 24 (44%) failed more than one line of progression. Exploratory objectives include evaluating the pre- and post-
therapy. 45 pts (1 pt withdrew consent, 1pt just enrolled prior to data cutoff treatment immunological changes in tumor tissues and peripheral blood.
and 7 pts came off the study due to clinical progression) were evaluable for Results: 17 pt were enrolled at the time of interim analysis (8 in Arm A, 9 in
response rate. Out of 45 pts, 10 pts (22%) achieved PR (1 unconfirmed PR) Arm B) and 14 were evaluable. Most pt (53%) were 60-70yo, and males
and 17 pts (37.8%) achieved SD. DCR was 60%. All patients who responded (70%). 6 pt were HCV-positive and 4 had chronic hepatitis B. 14 pt pro-
were microsatellite stable. For evaluable 45 pts with median follow up of ceeded with resection as planned; surgery was aborted for 2 pt (1 for frozen
13.34 months, median PFS was 3.98 months (95% CI: 2.33-5.98) and the abdomen and 1 for development of contralateral liver nodule). One is still
median OS was 14.22 months (95% CI: 6.64-NA). 6 and 12month OS was receiving preoperative therapy. Pathologic complete response (pCR) was
71.4 and 52.3% and 6 and 12 month PFS was 35.2% and 24.1% re- observed in 4/14 evaluable pt – 2 in Arm A and 2 Arm B (29% pCR rate). 4 pt
spectively. Most common treatment related AEs (TRAE) was alkaline phos- in Arm B and 1 in Arm A experienced grade 3 or higher toxicity prior to
phatase increased (24.5%). Grade III/IV TRAEs were seen in 11 pts (20.4%); surgery. Conclusions: We report a pCR rate of 29% in an interim analysis of a
most common were hyponatremia (3 pts) and elevated alkaline phosphatase (2 phase II pilot trial of perioperative immunotherapy for resectable HCC.
pts). No treatment related AEs led to discontinuation of the study drug. Tissue Treatment was safe and surgical resection was not delayed. The study is
samples were collected in all pts with planned correlative studies underway ongoing and results may contribute to a paradigm shift in the perioperative
including the PDL 1 status. Conclusions: Nivolumab was well tolerated and treatment of HCC. Clinical trial information: NCT03222076.
has shown promising efficacy in refractory BTC including durable responses
lasting 2 years. Further randomized trial is warranted in refractory BTC. Clinical
trial information: NCT02829918.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
246s Gastrointestinal (Noncolorectal) Cancer

4099 Poster Session (Board #204), Mon, 8:00 AM-11:00 AM 4100 Poster Session (Board #205), Mon, 8:00 AM-11:00 AM
Clinical and prognostic significance of serum levels of fatty acid binding Predictors of poor outcome for transarterial chemoembolization (TACE) in
proteins in hepatocellular carcinoma (HCC). First Author: Yehia I. Abugabal, hepatocellular carcinoma (HCC). First Author: Petra Prins, Medstar George-
University of Texas MD Anderson Cancer Center, Houston, TX town University Hospital, Washington, DC
Background: Limited data are available about the prognostic effect of fatty Background: The use of TACE in select patients with BCLC stage B HCC has
acid binding proteins (FABP) in viral and non-viral-related hepatocellular been shown to improve survival. Despite this, it remains unclear which
carcinoma (HCC). Previous studies suggested that selected FABP could be a patients will benefit from repeated TACE versus switching to systemic
potential target markers for HCC chemotherapy response and may correlated therapy upon disease progression. The purpose of this study is to identify
with presence of cirrhosis and poor outcome. We aimed to test the asso- prognostic factors that predict poor outcomes in patients who receive TACE.
ciation between plasma levels of Liver (L)-FABP, Heart (H)-FABP, and Methods: In this single-institutional retrospective analysis, patients with
Adipose (A) FABP and HCC. Methods: we enrolled 767 HCC patients from unresectable HCC were treated with TACE between 2007-2016. Relevant
MD Anderson Cancer Center. Under IRB approval, baseline patients’ factors such as staging by BCLC stage B, Child-Pugh score, vascular invasion
characteristics were retrieved from medical records and blood samples were (VI), tumor thrombus (TT), AFP levels, and number of TACE treatments
collected and tested form plasma levels of L-, A-, H-, FABPs. Descriptive within six months from the initiation of TACE were analyzed using either
statistics were performed and the median values of FABPs among 200 Pearson’s chi-square test or the student’s t-test. The Kaplan-Meier method
normal controls (NC) were used as cutoff values of FABPs. Overall survival was used for survival analysis. Results: Patients (n = 176) underwent TACE;
(OS) was estimated by Kaplan Meier curve and log rank test. Results: FABPs 45% had stage I-II disease, 42% were BCLC stage B prior to TACE, 71%
were highly expressed in HCC cases than controls. Mean values (6SE) of were Child-Pugh A, 21% had extrahepatic spread, 34.7% had VI, and 26%
AFABP, HFABP, and LFABP were significantly higher in cases [25.6 (.7), had TT. The median number of TACE treatments was 2 (range, 1- 6). The
10.8 (.5), and 47.8 (1.9)] than controls [19.1 (.8), 7.7 (2), 22. 9 (.5)], P , median overall survival (mOS) was 43 months (m) (95% CI 31.3-54.7) and
.001. All FABPs were significantly associated with cirrhosis, higher Child mOS from start of TACE was 34m (95% CI 26.2-41.8). Elevated AFP
Pugh Score (CTP), advanced stage in Barcelona clinic liver cancer stage (.400) correlated with decreased mOS (25m vs. 35m, p=0.041). Similarly,
(BCLC), higher AFP levels, vascular invasion and thrombosis, and tumor the presence of TT correlated with poor outcomes (25m vs. 37m, p=0.015).
nodularity. Median OS (months) (95%CI) were significantly short in patients The mOS was also negatively impacted by having 3 or more TACE treatments
with higher level of AFABP, HFABP, and LFABP [9.3 (6.8-11.9), 9.4 (6.8- within a 6 m period (25m vs. 38m, p = 0.09). AFP .400, TT, and interval
11.9), and 11.1 (8.8-13.3)] as compared to patients with low levels [16.4 between TACE were all independent factors in this multivariate analysis,
(13.8-18.9), 16.4 (14.2-18.6), and 17.9 (14.9-20.9) respectively (P , resulting in a shorter mOS of approx. 2 years compared to 3 years in patients
.01). The significance was observed in non-viral related HCC for LFABP and without these negative prognostic factors. There was a strong association
HFABP, but not AFBABP. Conclusions: To the best of our knowledge, we with both elevated AFP and TT (Chi square p=0.009). Conclusions: Elevated
describe the largest study correlating FABPs levels with clinical and prog- AFP (.400), the presence of TT, and a need for 3 or more TACE treatments
nostic characteristics of HCC. Higher levels were associated with poor within 6 months appear to be independent predictors for shorter mOS in
survival. These findings suggest that LFABP and HFABP may be used as patients receiving TACE. Patients with these poor prognostic factors tend to
potential prognostic biomarkers for non-viral-related HCC. have more aggressive HCC, and earlier initiation of systemic therapy might
provide benefit to these patients. A larger study is needed for confirmation of
these findings.

4101 Poster Session (Board #206), Mon, 8:00 AM-11:00 AM 4102 Poster Session (Board #207), Mon, 8:00 AM-11:00 AM
An open label, single-arm, two-stage, multicenter, phase II study to evaluate A phase I study of oncolytic immunotherapy of metastatic neuroendocrine
the efficacy and safety of TLC388 as second-line treatment in subjects with tumors using intralesional rose bengal disodium: Cohort 1 results. First
poorly differentiated neuroendocrine carcinomas (TCOGT1Z14). First Au- Author: Timothy Jay Price, Queen Elizabeth Hospital, University of Adelaide,
thor: Ming-Huang Chen, Department of Oncology, Taipei Veterans General Adelaide, Australia
Hospital, Taipei, Taiwan
Background: Metastatic neuroendocrine neoplasms (mNEN’s) originating in
Background: Therapeutic options for metastatic poorly differentiated neuro- the gastrointestinal tract are frequently slow growing yet both symptom and
endocrine carcinoma (NEC) after prior platinum-based chemotherapy are disease control remain important. Treatment options include resection,
unknown. Camptothecin analogs, like topotecan and irinotecan, are approved chemoablation, systemic somatostatin analogues (SSA) or peptide receptor
chemotherapy in small cell lung cancer (SCLC). NEC is considered to have radionuclide therapy (PRRT), but additional options are needed and one such
similar biological behavior to SCLC. The aim of this study was to analyze the option is hepatic intralesional (IL) rose bengal disodium (PV-10), an oncolytic
efficacy of TLC388 (Lipotecan) Hydrochloride, which is a novel camptothecin immunotherapy under development for solid tumours. Methods: This phase 1
analog, in pretreated metastatic NEC patients. Methods: This single-arm, study is evaluating the safety, tolerability and reduction of biochemical
2-stage, phase 2 clinical trial was conducted at 4 community and aca- markers and symptoms resulting from percutaneous administration of PV-10 in
demic centers in Taiwan. Patients aged 20 years or older enrolled between July 12 subjects with progressive mNEN with hepatic lesions not amenable to
2015 to May 2018 had confirmed metastatic NEC with prior systemic therapy resection or other potentially curative therapy. Target lesion(s) must be 1.0 -
with etoposide plus cisplatin. Patients received intravenous 40 mg/m2 of 3.9 cm in longest diameter. In Cohort 1 (n = 6) subjects receive PV-10 to a
TLC388 on Days 1, 8 and 15 of a 28-day cycle until disease progression or single hepatic lesion per treatment cycle, and can receive PV-10 to additional
unacceptable toxic effects. Results: twenty-three patients with a median age of uninjected hepatic lesions $6 weeks after prior injection. Cohort 2 (n = 6)
61 (range, 44-73) years, including 18 men (78%), were enrolled. Patients subjects may receive injection of multiple lesions per treatment cycle. The
received a median of 2 (range, 0-6) treatment cycles. Among 20 evaluable primary endpoint is safety. Secondary endpoints include objective response
patients, three patients showed a stable disease and no patient a complete or rate (ORR) assessed by contrast enhanced CT and 68Ga-DOTATATE PET,
partial remission, resulting in a disease control rate of 15%. Median PFS biochemical response (CgA) and patient-reported outcome (EORTC QLQ-C30
was 1.8 (95% CI, 0.4-15) months and median OS was 4.3 (95% CI, 1.7-15) and GI.NET21). Results: Cohort 1 has fully enrolled, with 4 of 6 subjects male,
months. The most common treatment-related hematologic adverse events median age 65yrs, range 47-72. Primary sites were: small bowel 3, pancreas 2,
at grade 3 or higher were leukopenia (22.7%), anemia (31.8%), and caecal 1; grade: Gd1 = 5, Gd2 = 1. All patients received prior SSA and PRRT.
thrombocytopenia (18.2%), respectively. Conclusions: TLC388 shows mod- Median CgA was 645 (range 30-2819). To date 1 subject has received 4 PV-10
est antitumor activity in metastatic NEC. Clinical trial information: treatment cycles, 1 has received 2 cycles, and 4 have received a single cycle.
NCT02457273. Toxicity has been acceptable, including pain post procedure, carcinoid flare
and nausea. LFT’s have remained stable. Overall QOL score was stable for 5 of
6 subjects. ORR in injected lesions is 50% (progression in 1 subject), with
overall disease control of 84%. CgA response: 5 stable, 1 progression. One
subject with “carcinoid pellagra” had rash resolution. Response follow-up is
ongoing and additional efficacy and functional data will be presented.
Conclusions: Hepatic IL PV-10 elicited no safety concerns with encouraging
evidence of both local and systemic disease control. Enrolment to Cohort 2 is
underway. Clinical trial information: NCT02693067.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Noncolorectal) Cancer 247s

4103 Poster Session (Board #208), Mon, 8:00 AM-11:00 AM 4104 Poster Session (Board #209), Mon, 8:00 AM-11:00 AM
A phase II, open label, multicenter trial of avelumab in patients with advanced, Oxaliplatin and 5-fluorouracil (FOLFOX) in advanced well-differentiated
metastatic high-grade neuroendocrine carcinomas NEC G3 (WHO 2010) digestive neuroendocrine tumors: A multicenter national retrospective study
progressive after first-line chemotherapy (AVENEC). First Author: Christian from the French Group of Endocrine Tumors (GTE). First Author: Paul Girot,
Fottner, I. Medical Department, Mainz University Medical Center, Mainz, CHD Vendée La Roche Sur Yon, La Roche Sur Yon, France
Germany
Background: Oxaliplatin-based regimens have shown promising antitumor
Background: High grade Neuroendocrine Neoplasias (NEN) are rare tumors activity in digestive neuroendocrine tumors (NETs), however the available
with a poor prognosis and no established second line therapy when progressive data are limited. Our aim was to assess the tumor response and survival in a
after first line platinum-based chemotherapy resulting in a median overall large series of patients treated with oxaliplatin and 5-fluorouracil (FOLFOX)
survival (OS) of 5 months. This study aims to evaluate the efficacy and safety of for advanced digestive NETs.Oxaliplatin-based regimens have shown
the anti-programmed death ligand-1 (PD-L1) antibody Avelumab in patients promising antitumor activity in digestive neuroendocrine tumors (NETs),
(pts) with NEN G3 progressing after first-line chemotherapy. Methods: In a however the available data are limited. Our aim was to assess the tumor
multicenter, national, single-arm, open-label, phase II trial the efficacy and response and survival in a large series of patients treated with oxaliplatin and
safety of Avelumab was evaluated in patients with metastatic progressive 5-fluorouracil (FOLFOX) for advanced digestive NETs. Methods: All patients
Neuroendocrine Carcinomas (NEC G3) according to WHO 2010, excluding with advanced well-differentiated digestive NETs treated with at least 3
Merkel cell carcinoma and small cell lung cancer. Results: From 12/2017-11/ cycles of FOLFOX between 2004 and 2018 in 12 centers of the French GTE,
2018 a total of 29 pts (20 male, 69%), were enrolled (16 NEC G3 and 11 were retrospectively included. Best response according to the RECIST 1.1
moderately differentiated NETG3). Mean age was 59.2610.2 ys (range 33- criteria, progression-free survival (PFS) and overall survival (OS) were
75), median follow up 16.5 weeks (3-48). Median Ki67 was 60% (range 20- evaluated. The prognostic factors for PFS were investigated by multivariate
95%). Site of origin included pancreas (12), genito-urinary tract (4), stomach- analysis using a Cox proportional hazard model including variables with a p
esophagus (3), colo-rectum (3), lung (2), ear-nose-throat (2), papilla of Vater value # 0.20 in univariate analysis. Results: One hundred and forty-nine
(1). In an interim analysis the DCR (stable disease or partial remission patients were included. Primary tumor location was pancreas (n = 88), small
according to irRECIST) after 8 weeks was 32% (4 SD, 2 PR). In responders, intestine (n = 37), stomach (n = 7), rectum (n = 4) and unknown without lung
mean duration of disease control was 20 (613.8) weeks, with 4 pts. showing tumor at CT scan (n = 13). Partial response rate was of 31% for pancreatic
stable disease or partial remission $6 months. Median OS was 4.2 months NETs, 13% for small intestine NETs, 14% for gastric NETs, 25% for rectal
(range 1- .12). Treatment-related adverse events occurred in 11 of 29 pts NETs and 38% for unknown primary NETs. Median PFS were, respectively,
(38%) and were mainly mild to moderate (CTCAE-grade 1 [52%], 2 [44%] and 9, 9, 14, 4 and 6 months, and median OS were 30, 28, 31, 25 and
3 [4%]) and included fatigue (n=6; 20.6%), diarrhea (n=4; 13.7%), fever/ 15 months. Significant poor prognostic factors for PFS after FOLFOX in
chills after infusion (n=4; 13.7%), loss of appetite and nausea (n=4; 13.7%), digestive NETs were: progressive disease (HR = 2.5, p = 0.018), hepatic
skin rash (n=1; 3%), deterioration of preexisting psoriasis (n=1; 3%) and involvement . 50% (HR = 1.8, p = 0.009), prior targeted therapy (HR = 1.5,
abdominal pain (n=1; 3%). Conclusions: Immune checkpoint blockade with p = 0.048) and rectal primary tumor (HR = 4.2, p = 0.01). Among pancreatic
avelumab in pretreated high grade NEN shows relevant activity in a subset of NETs, the 9 insulinomas had a 22 months PFS versus 9 months for the others
patients with excellent tolerability. Clinical trial information: NCT03352934. (p = 0.025), and serum glucose normalization was obtained in 8 out of 9
cases. Conclusions: FOLFOX has a promising clinical activity for gastro-
enteropancreatic NETs, especially in insulinomas.

4105 Poster Session (Board #210), Mon, 8:00 AM-11:00 AM 4106 Poster Session (Board #211), Mon, 8:00 AM-11:00 AM
The SUNEVO (GETNE-1408) trial to evaluate the activity and safety of Final results of the TALENT trial (GETNE1509): a prospective multicohort
thecombination of sunitinib with evofosfamide (TH-302) in patients with phase II study of lenvatinib in patients (pts) with G1/G2 advanced pancreatic
G1/G2 metastatic pancreatic neuroendocrine tumours (pNETs) naı̈ve forsys- (panNETs) and gastrointestinal (giNETs) neuroendocrine tumors (NETs).
temic treatment: A phase II study of the Spanish Task Force Group for First Author: Jaume Capdevila, Medical Oncology Department, Vall d’Hebron
Neuroendocrine and Endocrine Tumors (GETNE). First Author: Enrique University Hospital; Vall d’Hebron Institute of Oncology (VHIO), Barcelona,
Grande, MD Anderson Cancer Center Madrid, Madrid, Spain Spain
Background: Angiogenesis plays an important role in tumorigenesis and Background: Approved systemic therapies for advanced NETs have showed
progression of pNETs. Evofosfamide (EVO) is a DNA alkylator prodrug that limited tumor shrinkage and no data of activity after progression to prior
selectively activates under hypoxia. Sunitinib as monotherapy shows a wide targeted agents (TA) is available. Lenvatinib, a potent VEGFR1-3 & FGFR1-4
range of responses (9-24%) in similar populations. We hypothesized that inhibitor may increase efficacy and revert primary and acquired resistance to
sunitinib-induced hypoxia might increase the cytotoxic activity of EVO in TA. We report the final results of the TALENT trial. Methods: Two in-
patients with metastatic pNETS and naı̈ve for systemic treatment other than dependent cohorts were included: panNETs and giNETs. All pts had baseline
somatostatin analogues (SSA). Methods: This is a phase-II, single-arm, and documented progression disease (PD) by RECIST. For panNETs, PD to TA
multicenter trial of EVO (340mg/m2 on days 8, 15 and 22 every 4 weeks) and was mandatory, regardless of prior therapy with somatostatin analogs (SSAs)
sunitinib (37.5mg/day continuously). Primary endpoint was Objective Re- or chemotherapy (CHT); and for giNETs, PD on SSAs. Pts were treated with
sponse Rate (ORR) by RECIST v1.1 assessed every 8 weeks. A Simon two- lenvatinib at 24 mg qd until PD or intolerable toxicity. The primary endpoint
stage optimal design was used, considering a minimum of 3 responses in the was overall response rate (ORR) by central radiology review. Progression-free
first 18 pts in order to start with the second stage (power = 0.80, alpha = 0.05). (PFS) and overall survival (OS) were assessed by investigator. With 55 pts per
Results: Between May/2015 and May/2018, 17 pts were included (median arm, our study was powered to identify an ORR $25% (90% power, 5%
age was 62.4 y.o). Prior SSA was reported in 7 (41.2%) pts and 8 (47.1%) a-error). Results: We recruited 111 pts: 55 panNETs and 56 giNETs (78%
had a Ki-67 . 10%. There were 2 responders (11.8%; n = 1 complete and n = from small intestine). Prior therapies were CHT 32%, SSAs 87%, everolimus
1 partial response); stable disease was observed in 76.5%. Median (range) 70% and sunitinib 30% for panNETs. ORR was 29%, 42.3% for panNETs
PFS and duration of response were 10.4 months (m) (2.9-17.9m) and, re- and 16.3% for giNETs. With a median follow-up of 19 m, PFS and OS for
spectively 24.4m (13.7-35.2m), respectively. Grade 3 or 4 adverse events panNETs were 15.5 m (95% CI 11.3-not reached (NR)) and 29.2 m (95% CI
occurred in 11 (64.7%) pts, being neutropenia (33.3%), fatigue (16.7%), 23.2-NR); and 15.4 m (95% CI 11.5-19.4) and NR for giNETs, respectively.
thrombocytopenia (11.1%), hand-foot syndrome (5.6%), and pancreatitis Pts who obtained a response by RECIST had a significantly better PFS
(5.6%) the most frequent. Toxicity led to treatment discontinuation in 5 compared with non-responders (NR vs 11.2 m in panNETs (p=0.004);
(38.5%) pts. Dose reductions were reported in 20% (sunitinib) and 100 % 37.2 m vs 14.9 m in giNETs (p=0.005). In the subgroup analyses, all pts
(EVO) of pts. Conclusions: Combination of sunitinib and EVO failed to dem- obtained the same benefit in PFS and ORR, including tumor grade, prior
onstrate activity in terms of tumor shrinkage as only two patients achieved therapies, hormone release, primary location and tumor burden. The most
response, therefore, second stage was not proceeded. While cross trial frequent G3/4 adverse events were hypertension (22%), fatigue (11%) and
comparisons are difficult, response rate of 12% with the combination was diarrhea (11%). Dose reductions/interruptions were needed in 91.8% with a
disappointing. Concerns over toxicity arose; translational analysis are un- median dose of 20 mg qd. Conclusions: To our knowledge, we report the
dergoing. Clinical trial information: NCT02402062. highest ORR by central radiology assessment with a TA in this setting.
Lenvatinib showed a promising PFS and OS in a pretreated population with
benefit across subgroups. Further development in advanced NETs is war-
ranted. Clinical trial information: NCT02678780.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
248s Gastrointestinal (Noncolorectal) Cancer

4107 Poster Session (Board #212), Mon, 8:00 AM-11:00 AM 4108 Poster Session (Board #213), Mon, 8:00 AM-11:00 AM
Molecular characterization of the tumour microenvironment in neuroendocrine Long-term survival and safety from a multi-center, open-label, pivotal phase
malignancy. First Author: David James Pinato, Imperial College London, 2 study of iobenguane I 131 in patients (Pts) with unresectable, locally
London, United Kingdom advanced or metastatic pheochromocytoma or paraganglioma (PPGL). First
Author: Richard B. Noto, Rhode Island Hospital, Providence, RI
Background: A comprehensive characterization of the tumour microenviron-
ment is lacking in neuroendocrine tumors (NETs), where immunotherapy is Background: PPGL, rare neuroendocrine tumors with a 5-yr survival rate as
undergoing efficacy testing. We investigated drivers of cancer-related immu- low as 12%, have a high unmet need for effective treatment options.
nosuppression across NETs of various sites and grade using multi-parameter AZEDRA, a high-specific-activity iodine-131 meta-iodobenzylguanidine
immunohistochemistry and targeted transcriptomics. Methods: Tissue micro- (HSA I-131-MIBG), is the first and only FDA- approved therapeutic radio-
arrays (n = 102) were stained for PD-L1 & 2, Indoleamine-deoxygenase-1 (IDO- pharmaceutical agent indicated for the treatment of adult and pediatric pts
1) and evaluated in relationship to functional characteristics of tumor-infiltrating with iobenguane scan positive, unresectable, locally advanced or metastatic
T-lymphocytes (TILs) and biomarkers of hypoxia/angiogenesis including VEGF- PPGL who require systemic anticancer therapy. Methods: Pts with advanced
A, Hif-1a and Carbonic Anhydrase-IX. PD-L1 expression was tested in circu- disease who were heavily pre-treated and were ineligible for curative surgery
lating tumour cell (CTCs, n = 12) to evaluate its relationship with metastatic or chemotherapy received a dosimetric dose followed by up to two thera-
dissemination. Results: PD-L1 expression was highest in lung NETs (n = 30, p = peutic doses (each at 296 MBq/kg to a max of 18.5 GBq). The primary
0.007), whereas PD-L2 was highest in pNETs (n = 53, p , 0.001) with no endpoint, defined as the proportion of pts with at least 50% reduction of all
correlation with grade, stage or biomarkers of hypoxia. Incubation of QGP-1 and antihypertensive medication(s) lasting $6 months, was met and previously
BON-1 NET cells in 1% O2 did not induce PD-L1 expression confirming reported. Updated secondary endpoints including overall survival (OS) and
transcriptional independence from hypoxia. PD-L1+ NETs (n = 26, 25%) had safety are reported. Results: A dosimetric dose of HSA I-131-MIBG was
frequent IDO-1 co-expression (p = 0.03), greater CD4+/FOXP3+ and CD8+/PD1+ administered to 74 pts. Of those, 68 pts received one therapeutic dose and
TILs (p , 0.001) and necrosis (p = 0.02). CD4+/FOXP3+ infiltrate was highest 50 received two doses of HSA I-131-MIBG. Clinical benefit rates (objective
PD-L1/IDO-1 co-expressing tumours (p = 0.006). Survival was predicted by tumor responses defined by RECIST 1.0 and stable disease) were observed
tumour grade (p , 0.001) and necrosis (p , 0.001) but not PD-L1, PD-L2 nor in 71.4% and 98.0% of pts receiving one and two therapeutic doses, re-
IDO-1. High-grade NETs had lower CD4+/FOXP3+ and CD8+/PD1+ TILs density spectively. As of Jan 25, 2019, median OS for all pts was 41.1 months (95%
(p , 0.001) and Nanostring immune-profiling revealed enrichment of CI 31.1, 91.2). Median OS was 17.5 months (95% CI 4.0, 31.5) and
macrophage-related transcripts in cases with poorer prognosis. We identified PD- 48.7 months (95% CI 33.2, 91.2) in pts receiving one and two doses,
L1(+) CTC subpopulations in 75% of evaluated patients (n = 12). respectively. A tail of survival was observed, with OS of 73.1% at 2 yrs and
Conclusions: PD-L1 expression correlates with T-cell exhaustion independent of 44.2% at 4 yrs. The most common ($50%) adverse events were nausea,
tumour hypoxia and is enhanced in a subpopulation of CTCs, suggesting its fatigue, and myelosuppression. Myelosuppressive events resolved within 4-8
relevance to the progression of NETs. These findings support a potential wks without requiring stem cell transplantation. Late radiation toxicity in-
therapeutic role for PD-L1/IDO-1 inhibitors in a subset of NETs. cluded 8 pts with secondary malignancies (myelodysplastic syndrome
(MDS), acute myeloid leukemia (AML), acute lymphoblastic leukemia (ALL),
colon cancer, and lung carcinoma) of which MDS, ALL and AML were
considered related to I-131 radiotherapy. Conclusions: Updated results from
this pivotal phase 2 study suggest that HSA I-131-MIBG is an efficacious
and safe treatment for advanced PPGL. Clinical trial information:
NCT00874614.

4109 Poster Session (Board #214), Mon, 8:00 AM-11:00 AM 4110 Poster Session (Board #215), Mon, 8:00 AM-11:00 AM
Impact of gender on multikinase inhibitors (MKIs) toxicity in patients (pts) Blood-based next-generation sequencing analysis of neuroendocrine
with advanced pancreatic and gastrointestinal neuroendocrine tumors tumors. First Author: Walid Labib Shaib, Winship Cancer Institute, Emory
(NETs): A pooled analysis of two phase II trials with pazopanib and University, Atlanta, GA
lenvatinib. First Author: Jorge Hernando-Cubero, Vall Hebron University
Background: Neuroendocrine tumors (NET) comprise around 2% of all
Hospital, Vall Hebron Institute of Oncology (VHIO), Bracelona, Spain
malignant tumors of the gastrointestinal system. The genomic landscape of
Background: Retrospective data in some cancer types suggested a possible NET has not been well studied. The aim of this study was to confirm the
different toxicity profile with chemotherapy and targeted therapies according feasibility of next generation sequencing (NGS) using ctDNA in NET and
to gender. However, data from prospective studies are still very limited, characterize common alterations in the genomic profile. Methods: Molecular
especially in infrequent tumors such as NETs. Methods: Pts with advanced alterations in 114 plasma samples from 114 patients with NET using
pancreatic and gastrointestinal NETs treated with pazopanib or lenvatinib in clinical-grade NGS of ctDNA (Guardant360Ò) across multiple institutions
the multicenter open-label phase II studies PAZONET and TALENT re- were evaluated. The test detects single nucleotide variants in 54-73 genes,
spectively, were included in the analysis. Both studies were performed by copy number amplifications, fusions, and indels in selected genes.
Spanish Task Force Group for Neuroendocrine Tumors (GETNE). All toxicity Results: A total of 114 NET patients were evaluated, of which 64 (56.1%)
grades with an incidence higher than 5% were considered for univariate were female. Mean age was 59.7 years with a range between 23-89 years.
review. Additionally, all grade 3-4 toxicities were analyzed separately. ctDNA NGS testing was performed on 114 plasma samples; 1 patient had
Results: 155 pts (47.7% female) with 1213 adverse events (AEs) (20% G3- testing performed twice. Genomic alterations were defined in 94 (n = 94/
4) divided in 121 categories were included. In female patients, liver toxicity, 114, 82.5%) samples with a total of 289 alterations identified after ex-
headache, pyrexia, nausea/vomiting, hair/skin disorders and dizziness were cluding variants of uncertain significance (VUSs) and synonymous muta-
significantly more common (table). The only toxicity with higher incidence in tions. Alterations were identified in at least one sample from 83 patients;
men was dysphonia (OR 0.42, 95% CI 0.2-0.9, p 0.02). There were no TP53 associated genes were most commonly altered (n = 83/289, 28.7%),
gender differences in grade 3-4 toxicities. Conclusions: We observed sig- followed by KRAS (n = 22, 7.6%), PI3CA (n = 15, 5.2%), CCNE1 (n = 15,
nificant differences in toxicity AEs by gender in two prospective phase II 5.2%), BRAF (n = 13, 4.5%), MYC (n = 12, 4.1%), ERBB2 (n = 11, 3.8%),
studies with MKIs in NETs patients. Potential different approach to manage APC (n = 10, 3.5%), EGFR (n = 10, 3.5%), MET (n = 10, 3.5%), PTEN (n =
toxicity may be adopted based on gender. 9, 3.1%), RB1 (n = 9, 3.1%), CDK6 (n = 7, 2.4%), AR (n = 5, 1.7%),
ARID1A (n = 5, 1.7%), FGFR1 (n = 5, 1.7%), and PDGFRA (n = 5, 1.7%).
Toxicity (all Women Difference Other genomic alterations of low frequency, but clinical relevance included:
grades) (%) Men (%) (%) Odds Ratio (95% CI) p
CDK4 (n = 4, 1.3%), NF1 (n = 4, 1.3%), RAF1 (n = 4, 1.3%), GNAS (n = 3,
Liver toxicity 64.9 41.9 23 2.97 (1.54-5.73) 0.001 1.0%), KIT (n = 3, 1.3%), BRCA2 (n = 2, 0.7%), CCND2 (n = 2, 0.7%),
Headache 32.4 17.6 14.9 2.5 (1.16-5.4) 0.01 CTNNB1 (n = 2, 0.7%), JAK2 (n = 2, 0.7%), NRAS (n = 2, 0.7%), SMAD4
Pyrexia 21.6 8.1 13.5 3.44 (1.26-9.36) 0.01 (n = 2, 0.7%), and TERT (n = 2, 0.7%). Alterations in AKT1, ALK, ATM,
Nausea/Vomiting 70.3 58.1 12.2 2.08 (1.07-4.05) 0.02
Hair disorders 23 10.8 12.2 2.72 (1.09-6.75) 0.03 BRCA1, CCND1, CDKN2A, FGFR2, MTOR, RHOA, SMO and STK11 were all
Skin disorders 56.8 44.6 12.2 1.9 (1.007-3.61) 0.04 reported once (n = 1, 0.3%). Conclusions: Evaluation of ctDNA is feasible
Dizziness 20.3 9.5 10.8 2.68 (1.02-7.02) 0.04 among individuals with NET. Liquid biopsies are not invasive and can
Dysphonia 17.6 36.5 18.9 0.42 (0.2-0.9) 0.02 provide personalized options for targeted therapies in NET patients.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Noncolorectal) Cancer 249s

4111 Poster Session (Board #216), Mon, 8:00 AM-11:00 AM 4112 Poster Session (Board #217), Mon, 8:00 AM-11:00 AM
Analysis of patient diaries in the NETTER-1 Study of 177Lu-DOTATATE Efficacy and safety of pembrolizumab in patients with advanced adrenocor-
versus high-dose octreotide in progressive midgut neuroendocrine tumors. tical carcinoma. First Author: Nitya Prabhakar Raj, Memorial Sloan Kettering
First Author: Jonathan R. Strosberg, Moffitt Cancer Center, Tampa, FL Cancer Center, New York, NY
Background: The primary statistical analysis for the NETTER-1 trial showed a Background: Adrenocortical carcinomas (ACC) are rare and aggressive.
clinically and statistically significant PFS benefit with 177Lu-DOTATATE vs. Treatment options are limited and marked by poor efficacy and substantial
high-dose octreotide. 177Lu-DOTATATE treatment was also correlated with a toxicity. In this phase II single-center study, the efficacy and safety of
significant delay in time to deterioration in HRQoL. In addition to HRQoL pembrolizumab was assessed in patients (pts) with advanced ACC.
questionnaires, patients were asked to record presence or absence of a range of Methods: Enrolled pts were aged $18 y with advanced ACC, ECOG # 1,
symptoms in a daily diary. Methods: A Mixed Model Repeated Measures available tumor samples for biomarker analysis. Pts received pembrolizumab
(MMRM) was used to analyze the change, compared to baseline, of the oc- 200 mg Q3W for 2 y or until disease progression, intolerable toxicity,
currence of abdominal Pain, diarrhea and cutaneous flushing as these physician/patient decision to stop treatment. Imaging was performed every 9
symptoms were regarded as the most relevant to judge the overall disease wks. Tumor PD-L1 positivity (modified proportion score $ 1% or presence of
status. For each visit (week = 0, 4, 8, etc.) the number of days with symptoms stromal interface) was evaluated. Primary endpoint was ORR (by RECIST
during the previous period was calculated. At baseline, the number of days with v1.1). Secondary endpoints included DOR, PFS, OS, safety. Somatic and
symptoms was counted over the previous 6 weeks, whereas the time frame germline next-generation sequencing was performed. Results: 39 pts were
between visits lasted 4 weeks. Results: The estimated number of days with treated. Median age 62 (range, 19-87), 28% ECOG 0, 72% received $ 1
symptoms declined significantly more in the 177Lu-dotatate arm compared to therapy. In available samples to date, 7/31 (23%) PD-L1+. At time of
the octreotide arm. The difference in change and the confidence intervals analysis, median follow-up among survivors was 17.8 mo (range, 5.4-34.7).
for the symptoms abdominal pain, diarrhea and flushing of skin are, re- ORR was 23.1% (95% CI, 11.1-39.3); 0 CR, 9PR. Seven pts (17.9%) had
spectively: -3.11 [-4.88; -1.34], -3.11 [-5.04; -1.18] and -1.98 [-3.88; SD as best response. Among the 9 PRs, median time to PR was 4.1 mo
-0.08]. Conclusions: Analysis of symptom diaries confirms that 177Lu- (range, 1.7-10.5) and median DOR was not reached (95% CI, 4.1-not
Dotatate can palliate clinically relevant symptoms when compared to high- reached). Three pts achieving PR have completed 2 y of treatment with
dose octreotide. ongoing response noted. Tumor PD-L1 status is currently available in 6 pts
with PR, 2/6 (33%) PD-L1+. Median PFS was 2.1 mo (95% CI, 2.0-10.7).
Median OS was 24.9 mo (95% CI, 4.2-not reached); 2-year OS rate was 50%
(95% CI, 36-69%). In the 34 tested tumors, germline testing identified 2 PR
pts with Lynch syndrome; the remaining 7 PRs were MSS. Median tumor
mutation burden for all PRs was 4.1 mutations/megabase (range, 0-31.5).
There was no significant relationship between somatic alterations and re-
sponse to treatment. Grade 3/4 treatment-related AEs occurred in 7/39
(17.9%) pts. Two pts discontinued therapy due to AEs; both pts achieved
PRs and continue to respond. All pts with PRs had LFT elevation $ grade 2.
Conclusions: Pembrolizumab demonstrated antitumor activity and was well
tolerated in advanced ACC. Durable responses were noted. Complete
evaluation of tumor PD-L1 and microsatellite status will be reported at the
meeting. Clinical trial information: NCT02673333.

4113 Poster Session (Board #218), Mon, 8:00 AM-11:00 AM 4114 Poster Session (Board #219), Mon, 8:00 AM-11:00 AM
Surgery and peptide receptor radionuclide therapy: An effective multimodal Clinical efficacy and toxicity data on phase I study of fosbretabulin in
approach for metastatic neuroendocrine tumors. First Author: Andreja combination with everolimus in neuroendocrine tumors. First Author: Aman
Frilling, Imperial College London, London, United Kingdom Chauhan, University of Kentucky, Division of Medical Oncology, Lexington, KY
Background: Neuroendocrine neoplasia (NEN) of the pancreas (PanNEN) or Background: Fosbretabulin, a synthetic, water-soluble, phosphorylated pro-
small bowel (SBNEN) frequently present with metastases at initial diagnosis, drug of the natural product combretastatin A4 (CA4P), initially isolated from
undermining the efficacy of surgical treatment. Peptide receptor radionuclide the bark of the South African bush willow, Combretum caffrum, is the lead
therapy (PRRT) with radiolabelled somatostatin analogues, 90Y-DOTATOC and compound in a class of agents termed vascular disrupting agents (VDAs).
177
Lu-DOTATATE, has been shown to achieve prolonged progression-free Everolimus, an mTOR inhibitor, is FDA approved for the management of well-
survival (PFS) and overall survival (OS) in a substantial number of non- differentiated NETs. A Phase I trial combining fosbretabulin and everolimus to
surgical patients with advanced NEN. Our aim was to prospectively de- determine the recommended Phase II trial dose (RP2D), safety data and early
termine the efficacy of a combination of radical loco-regional surgery and clinical efficacy in metastatic GEPNET patients was conducted. Methods: An
177Lu PRRT in patients with metastasised NEN. Methods: A set of inclusion investigator-initiated, single center, open-label, phase I study involving
criteria was defined (e.g. PanNEN or SBNEN, G1/G2 NEN, initial tumour GEPNETs incorporated partial order continual reassessment method (PO-
diagnosis, treatment naı̈ve patient, stage IV NEN, positivity on 68Ga DOTA- CRM) to define the dose escalation. The primary objective was to establish
TATE or DOTATOC PET/CT, eligibility for surgery and PRRT). Patients un- the maximum tolerated dose (MTD) of the combination of everolimus and
derwent PRRT within 3 months following surgery. Follow-up included fosbretabulin in NETs that have progressed after at least one prior regimen for
biochemistry and imaging. Outcome measures included 1-, 3-, and 5-year OS metastatic disease. Secondary objective included identifying the safety profile
and PFS from initial diagnosis. Results: Forty-one patients met eligibility of the combination using NCI CTCAE4 reporting criteria. Patients received
criteria and were included. There were 26 males (63.4%) and median age at daily oral everolimus (2.5 mg, 5 mg, 7.5 mg, and 10 mg). Fosbretabulin was
surgery was 58.8 years (range 32.1-78.3). All patients with SBNEN underwent administered IV 60 mg/m2 either q3 weekly or q weekly based on PO-CRM.
right hemicolectomy, terminal ileal resection and mesenteric lympadenec- Patients were treated for 12 weeks with all combinations. RECIST 1.1 was
tomy. In PanNEN patients either Whipple procedure or distal pancreatectomy used to evaluate radiological responses at 3 month. Results: Of the 17 patients
and peripancreatic lymphadenectomy were performed. The median number of enrolled, 16 completed the 12-week trial. One patient was not evaluable due to
PRRT cycles was 4 (range 2-6). Post-treatment mortality was 0%. Surgical noncompliance. No DLTs were observed at day 21. The highest dose of 10 mg
morbidity was 12% (all grade 1 according Clavien-Dindo) and transient grade 1 daily oral everolimus in combination with weekly 60mg/m2 IV fosbretabulin is
toxicity occurred post PRRT in 40%. There was no grade 3 toxicity. Median the RP2D. No grade 4 or 5 toxicities were noted. Grade 3 toxicities were seen in
follow-up was 5.48 years (range 0.53 – 11.98). Median PFS and OS were 3.33 5 patients; abdominal pain and hyperglycemia (not related to study drug),
years and 9.07 years, respectively. Progression-free survival (with 95% CI) was fatigue (possibly related), decreased lymphocyte count and anemia (related).
at 1-, 3-, and 5-years 80% (68.7-92.6), 60.9% (45.9-75.9) and 43.3% Several patients had delay in treatment due to grade 2 AE’s (GI symptoms,
(27.4-59.3), respectively. Overall survival (with 95% CI) at 1-, 3-, and 5-years rash, thrombocytopenia) and one patient was unable to complete treatment
was 97.6% (93-100), 97.6% (93-100), and 95% (87-100), respectively. due to pneumonitis. All evaluable patients except one had stable disease at
Conclusions: Radical loco-regional surgery for primary tumours combined with 3 months. One patient showed SD but non target lesion demonstrated PD. One
PRRT provides a novel, highly efficacious approach in metastasised NEN. patient had . 30% decrease in tumor size but overall sum of lesions showed
SD. A detailed table with all grade toxicities and waterfall plot of RR will be
presented at the meeting. Conclusions: Ten mg PO daily everolimus plus 60
mg/m2 fosbretabulin IV weekly is the RP2D. Early clinical data suggests
clinical activity and stable disease in all but one patient at 3 months. Clinical
trial information: NCT0301429.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
250s Gastrointestinal (Noncolorectal) Cancer

4115 Poster Session (Board #220), Mon, 8:00 AM-11:00 AM 4116 Poster Session (Board #221), Mon, 8:00 AM-11:00 AM
Antitumor efficacy of concurrent everolimus with hepatic transarterial bland Immune checkpoint inhibitors (ICIs) in gastrointestinal (GI) cancer: Immune-
embolization (evero-embo) in patients with metastatic well differentiated related adverse events (IRAEs) and efficacy. First Author: Satya Das, Vanderbilt
neuroendocrine tumor (NET). First Author: Lowell Brian Anthony, University University Medical Center, Nashville, TN
of Kentucky, Lexington, KY
Background: Despite the therapeutic promise of ICIs for patients (pts) with
Background: Hepatic transarterial embolization (HAE) is an effective loco- some advanced malignancies, they are FDA-approved for only a few GI
regional therapy for neuroendocrine tumor (NET) management. Systemic cancer pts. In NSCLC, melanoma and urothelial carcinoma, there is
targeted therapies, such as everolimus and sunitinib, are typically held 2- emerging data that pts who experience IRAEs while on ICIs have improved
4 weeks prior to and after procedures. The safety of concurrent use of ever- outcomes compared with pts who do not. This association in GI cancer pts
olimus with HAE has been previously reported (GI-ASCO). HAE induces anoxic has not been reported. Methods: We retrospectively analyzed outcomes for
injury while everolimus effects cell growth, proliferation and survival. Com- metastatic GI cancer pts receiving ICIs for FDA-approved indications (later-
bining these two modalities may result in clinical synthetic lethality effectively line MSI-H tumors, 2nd line hepatocellular carcinoma (HCC), 3rd line PD-L1+
debulking significant hepatic disease and/or delay progression. Historically gastric (GA)/gastroesophageal junction (GEJ) adenocarcinoma), at Van-
bland HAE has a median hepatic PFS of ~9 months. In this study, the clinical derbilt Ingram Cancer Center, Winship Cancer Institute and Stanford Cancer
efficacy of evero-embo is examined. Methods: A review of clinical and ra- Institute. Our primary aim was to compare progression-free survival (PFS)
diographic data was conducted for all sequential patients who underwent and overall survival (OS) between pts who did and did not experience IRAEs.
evero-embo between September 2016 and April 2018 at the University of Secondary aims were comparison of these outcomes within pts who expe-
Kentucky Markey Cancer Center. An independent radiologist performed rienced IRAEs, by initial IRAE severity (Grade (G)3/G4 vs G1/G2) (CTCAE
RECIST measurements. Patients were required to have had systemic ever- v5.0), time-to-onset (TTO) (# 6 weeks (w) vs . 6 w) and management
olimus for $ 1 month prior to embolization in order to be included in this study (steroids vs drug cessation vs observation). PFS and OS were determined by
and be on everolimus immediately post procedure. Patients with at least Kaplan-Meier (KM) analysis; KM comparisons were done by the logrank test.
12 months post procedure follow up were included for efficacy review. Results: Between 1/2015-12/2018 61 GI cancer pts with HCC (28), co-
Results: A total of 51 HAEs with concurrent systemic everolimus were per- lorectal cancer (27) and GA/GEJ cancer (6) were treated with ICIs; median
formed in 34 NET patients. Twenty one of 24 patients were noted to have a age was 63 years. The majority (59) received single-agent nivolumab or
partial response. Rest had stable disease. Hepatic progression was not ob- pembrolizumab while minority (2) received nivolumab and ipilimumab;
served. Twenty-one of the 34 patients have had 12 or more months of follow up median time on ICIs was 5.9 months (mos). Twenty-four pts experienced
post procedure (median of 17 months). None of these 21 patients have had initial IRAEs (6 G3/G4, 18 G1/G2); median TTO was 3.8 mos. Pts who
hepatic progression. Conclusions: Evero-embo results in a partial response rate experienced any IRAE had improvements in PFS and OS compared to those
of 62% and may have significant antitumor activity when compared to bland who did not (PFS: 32.4 mos (95% confidence interval (CI), 32.4-not
hepatic artery embolization in NET patients. With a median follow-up of 17 reached (NR)) vs 4.8 mos (95% CI, 2.9-8.7), p = 0.0001; OS: 32.4 mos
mos, hepatic progression has not occurred in any patient. Additional follow up (95% CI, 32.4-NR) vs 8.5 mos (95% CI, 6-NR), p = 0.0036). Among pts who
is necessary to compare the median hepatic PFS of evero-embo to the historical experienced IRAEs, PFS and OS differences between above-specified
drug-eluting bead HAE PFS. subgroups did not meet statistical significance. Conclusions: GI cancer
pts who experienced IRAEs while on ICIs had marked improvements in PFS
and OS compared to those who did not, suggesting the predictive potential
for IRAEs as a clinical biomarker in this population.

4117 Poster Session (Board #222), Mon, 8:00 AM-11:00 AM 4118 Poster Session (Board #223), Mon, 8:00 AM-11:00 AM
Updated results of a phase IIa study to evaluate the clinical efficacy and Efficacy and safety of lanreotide 120 mg in the treatment of clinical symptoms
safety of erdafitinib in Asian advanced cholangiocarcinoma (CCA) patients associated with inoperable malignant intestinal obstruction (IMIO): Results
with FGFR alterations. First Author: Joon Oh Park, Department of Medicine, from a phase II multicenter study. First Author: Lionel Duck, Clinique St-Pierre,
Samsung Medical Center, Sungkyunkwan University School of Medicine, Ottignies, Belgium
Seoul, South Korea
Background: Intestinal obstruction is a severe complication in patients (pts)
Background: Patients (pts) with advanced CCA who progressed on or after first with digestive or gynecological cancers. For inoperable pts, there is a need to
line chemotherapy have no approved treatment options. Fibroblast growth relieve symptoms and limit nasogastric tube (NGT) use. Previous studies have
factor receptor (FGFR) gene alterations are observed in many tumor types suggested the efficacy of somatostatin analogues in relieving obstruction-
including 14-17% in CCA. Erdafitinib, an orally bioavailable, selective pan- related symptoms such as nausea, vomiting and pain. Methods: This was a
FGFR kinase inhibitor, has shown clinical activity against solid tumors with single arm, prospective study (NCT02275338). Pts with IMIO received one
FGFR alterations. Methods: LUC2001 is an open-label, multicenter, Ph2a deep subcutaneous injection of LAN 120mg at day 0 (D0). Evaluations were
study in advanced CCA pts with FGFR alterations (FoundationOne), who performed on D7, 14 and 28. The primary endpoint was the proportion of
progressed after $ 1 prior treatment. The primary endpoint is objective re- responders before or at D7. Response was defined as #2 vomiting episodes/
sponse rate (ORR; RECIST 1.1). The secondary endpoints are disease control day (for pts without NGT at baseline) or no vomiting recurrence (after NGT
rate (DCR), progression free survival (PFS), duration of response (DOR), safety removal), during at least 3 consecutive days at any time point between the D0
and pharmacokinetics (PK). Disease is evaluated every 8 weeks until disease and D7. In line with the literature, a proportion of 30% responders was used as
progression (PD). Results: As of 3 Dec 2018, 222 CCA pts were molecularly reference for defining statistical significance. Responders at D28 were
screened; 34 had FGFR alterations, of whom 14 (8 FGFR2 fusion, 3 FGFR2 offered a second LAN 120 mg injection. Results: 52 pts with advanced GI or
mutation, 1 FGFR3 fusion, 2 FGFR3 mutation) were dosed 8 mg once daily ovarian malignancies were included in 15 Belgian sites. 17 pts without NGT
with up titration option. Median age was 51.5 years. 13/14 and 12/14 pts had and 35 with NGT. 21 pts received a second dose of LAN. Median age was 68.0
prior platinum or gemcitabine based therapy respectively, 7/14 pts got re- (59.5; 76.0) years. On D7 the proportion of responders in the ITT population
treated with platinum or gemcitabine based therapy, and 9/14 pts had $2 prior was 24/52 (46.2%), significantly greater than the reference proportion of 30%
lines of therapy. Median number of treatment cycles was 5.0 (range: 1; 22) and (one-sided binomial test: p = 0.006). Pts without NGT responded better (15/
treatment duration was 4.83 (range: 0.5; 20.3) months. In 12 evaluable pts, 17, 88.2%) than pts with NGT (9/35, 25.7%). Pts without ascites responded
there were 6 confirmed partial response (PR), 4 stable disease (SD) and 2 PD; better (57.7% vs 34.6%). Pts with NGT showed a steady trend for clinical
ORR (CR+PR) was 6/12 (50.0%), DCR (CR+PR+uCR+uPR+SD) was 10/12 improvement leading to sustainable responses of 45.7% on D14. Median time
(83.3%); median DOR was 6.83 months (95% CI: 3.65; 12.16); median PFS to response was 9 days for the overall population; 3 days for patients without
was 5.59 months (95% CI: 1.87, 13.67). In 10 evaluable FGFR2+ pts, ORR NGT vs 14 days for patients with NGT (p , 0.001). The most frequently
was 6/10 (60.0%); DCR was 10/10(100%); median PFS was 12.35 months reported AEs were GI disorders (in 34 pts). The most common events were
(95% CI: 3.15, 19.38). The most common TEAEs ( . 30%) were hyper- diarrhoea and abdominal pain. Conclusions: Our study is the first using long
phosphatemia, dry mouth, stomatitis, and dry skin. 9 pts had $ Grade 3 AEs (8 acting LAN 120mg in patients with IMIO and suggests an effect in controlling
Grade 3,1 Grade 5), of which 7 drug related. TEAE led to treatment 1 dis- clinical symptoms in pts with and without NGT at baseline. LAN 120 mg safety
continuation, 6 dose reductions and 1 death (not drug related). The results of profile was similar to that reported for the other indications. Clinical trial
PK and PK/PD relationship were consistent with other erdafitinib studies in information: NCT02275338.
different ethnic background pts. Conclusions: Asian advanced CCA pts with
FGFR alterations treated with erdafitinib had encouraging efficacy and ac-
ceptable safety profile similar to experience in other tumor types and pop-
ulations. Clinical trial information: NCT02699606.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Noncolorectal) Cancer 251s

4119 Poster Session (Board #224), Mon, 8:00 AM-11:00 AM 4120 Poster Session (Board #225), Mon, 8:00 AM-11:00 AM
Molecular profile of ampulla of vater carcinoma (AVC): A rare tumor type with Paclitaxel/carboplatin with or without cetuximab for treatment of carcinoma
meaningful molecular alterations. First Author: Jo~
ao Pinto, Oncology Division, with unknown primary (PACET-CUP): Results of a multi-center randomized
Hospital Beatriz Ângelo, Loures, Portugal phase II AIO trial. First Author: Alwin Krämer, Clinical Cooperation Unit
Molecular Hematology/Oncology, German Cancer Research Center (DKFZ)
Background: AVC is a rare type of cancer with dismal prognosis and limited
and Dept. of Internal Medicine V, University of Heidelberg, Heidelberg,
therapeutic options due to the lack of specific clinical trials. Two histologic
Germany
subtypes predominate, namely pancreatobiliary and intestinal. A variety of
molecular alterations have been described in AVC, but their clinical and ther- Background: Paclitaxel/carboplatin is one of the standard regimens for
apeutic implications have not been studied in detail. Methods: Retrospective empiric treatment of patients (pts) with carcinoma with unknown primary
cohort study of patients (pts) diagnosed with AVC treated in our institution from (CUP) when a specific therapy is not available. The EGFR antibody cetux-
2010 to 2018. We routinely performed Next Generation Sequencing in all AVC imab demonstrated efficacy in several cancer types. Methods: Pts with newly
tumors. Our main objectives were to describe the molecular profile of AVC and diagnosed non-resectable, undifferentiated or adeno-CUP were randomized
correlate with clinical outcomes. Results: Out of 26 pts with AVC, 13 pts were to 6 cycles paclitaxel 175 mg/m² and carboplatin AUC 5 every three weeks
male (50%), median age 65 (range 43-83), 7 pts (27%) had stage IV disease at (arm A) or the same chemotherapy plus cetuximab (400 mg/m², then
diagnosis. Histologic type was pancreatobiliary in 18 pts (69%), intestinal in 7 250 mg/m² weekly; arm B) at 13 German centres. CUP pts belonging to
pts (27%) and mixed in one case (4%). We identified KRAS mutations (mut) in favorable prognosis groups [i.e. women with isolated peritoneal carcino-
10 pts (7 pancreatobiliary, 2 intestinal, 1 mixed), TP53 mut in 6 pts (4/1/1), matosis or axillary lymph node metastases, men with retroperitoneal lymph
PIK3CA mut in 3 pts (3/0/0), ERBB2 mut in 3 pts (2/1/0), CTNNB1 mut in 3 pts nodes, pts with specific tumor entities according to histology / immuno-
(2/1/0). In pancreatobiliary we found single cases with RNF43, BRCA1 and histochemistry] were excluded from the trial. The primary endpoint was
CHEK2 mut; while in intestinal we found single cases with NRAS and BRAF mut. progression-free survival (PFS), secondary endpoints were response rate
One tumor of intestinal subtype had microsatellite instability (MSI). Three pts (RR) and overall survival (OS). Results: Between 03/2010 and 03/2017, 72
were included in phase I clinical trials, 2 of them with trials based on tumor pts were randomized to arm A and 78 pts to arm B. The median age was 61
profile (ERBB2 mut with pan-HER inhibitor and MSI with immunotherapy). years, 84 pts were male (40 and 44 pts in arm A and B). 58 pts had a
Median overall survival (OS) was 21 months for pts with stage I, II and III disease performance status (PS) of 0 (24 and 34 pts), 89 pts a PS of 1 (47 and 42 pts
(95% CI 12.37-not reached) and 13.2 months for stage IV disease at diagnosis in arm A and B), and one patient in arm A had a PS of 2. PFS and OS did not
(95% CI 5.73-not reached). In cox models, median OS was not dependent on differ between arms. The median PFS was 3.7 [95%CI: 3.0 - 4.4] and 4.6
KRAS or TP53 mutation status, or histological subtypes. Conclusions: AVC is a [95%CI: 2.9 - 6.2] months in arms A and B (HR 0.98 [95%CI: 0.70-1.37]),
rare type of cancer with two differentiated histological subtypes harboring unique the median OS 8.1 [95%CI: 6.5 - 9.8] and 7.4 [95%CI: 5.1 - 9.6] months in
molecular alterations that can be matched to investigational therapies. A broader arms A and B, respectively (HR 1.10 [95%CI: 0.77 - 1.56]). There was a
knowledge of the biology of these tumors is needed to improve patient outcomes. statistically non-significant trend towards a higher RR with cetuximab: 11
pts had a partial response in arm A (15% [95%CI: 7.9 - 26%]), and 17 pts in
arm B (22% [95%CI: 13%-33%], p = 0.30). Twenty-five pts in each arm had
stable disease, and 36 pts in each arm had progressive disease or were not
evaluable for response. Conclusions: This study represents one of the largest
trials in pts with CUP and demonstrates that randomized trials are feasible in
this disorder with high medical need. The outcome of CUP pts in the un-
favorable prognosis group was not improved by adding cetuximab to empiric
therapy paclitaxel/carboplatin. Clinical trial information: NCT00894569.

4121 Poster Session (Board #226), Mon, 8:00 AM-11:00 AM 4122 Poster Session (Board #227), Mon, 8:00 AM-11:00 AM
H3B-6527 clinical biomarker assay development and characterization of Variation in the surgical management of locally advanced pancreatic cancer.
HCC patient samples. First Author: Pavan Kumar, H3 Biomedicine, Inc., First Author: Bradley Norman Reames, University of Nebraska Medical
Cambridge, MA Center, Omaha, NE
Background: FGFR4/FGF19 signaling axis is a novel therapeutic target in Background: Recent reports suggest patients with locally advanced pan-
HCC. Multiple covalent FGFR4 inhibitors, including H3B-6527, are under creatic cancer (LAPC) may become candidates for curative resection fol-
clinical development. Preclinical efficacy studies in mice (including PDX) lowing neoadjuvant therapy, with encouraging survival outcomes. Yet the
have shown that FGF19 expression (FGF19+) is a predictive biomarker for optimal management approach for LAPC remains unclear. We sought to
FGFR4 inhibitor response. The mechanisms driving FGF19 expression in investigate surgeon preferences for the management of patients with LAPC.
HCC is largely unknown however, in some cases, focal amplification of Methods: An extensive electronic survey was systematically distributed by
ch11q13.3 containing FGF19 gene is thought to drive the FGF19 expres- email to an international cohort of pancreas surgeons. Data collected in-
sion. Consistent with the preclinical observations, clinical studies have also cluded surgeon practice characteristics, preferences for staging and man-
shown that FGF19+is a predictive biomarker for FGFR4 inhibitor response. agement, and 6 clinical vignettes (with detailed videos of post-neoadjuvant
However, these trials have also reported a large number of FGF19+patients arterial and venous imaging) to assess attitudes regarding eligibility for
failing to respond to FGFR4 inhibitors necessitating refinement of patient surgical exploration. Results: A total of 150 eligible responses were received
selection strategies. In an attempt to obtain deeper insights into the role of from 4 continents. Median duration in practice was 12 years (IQR 6-20) and
FGF19+as a predictive biomarker and potentially uncover additional bio- 75% respondents work in a university setting. Most (84%) are considered
markers that will enable improvement in patient selection strategies, we have high volume, 33% offer a minimally-invasive approach, and 48% offer
characterized a set of 258 HCC patient samples Methods: Samples were arterial resection in selected patients. A majority (70%) always recommend
acquired from biobanks and utilized to qualify clinical assays including neoadjuvant chemotherapy, and 62% prefer FOLFIRINOX. Preferences for
FGF19 copy number (FISH), mRNA expression (qRT-PCR), FGF19 protein duration of neoadjuvant therapy varied widely: 39% prefer $2 months, 41%
(IHC), and a focused NGS panel for assessing both mutations and copy prefer $4 months, and 11% prefer 6 months or more. Forty-one percent
number. A multiplexed protein and mRNA platform enabled assessment of frequently recommend neoadjuvant radiation, and 51% prefer standard
p-ERK and Ki67 (protein) and Cyp7A1 (mRNA) amongst other exploratory chemoradiotherapy. Age $80 years and CA 19-9 of $1000 U/mL were
PD biomarkers from two FFPE slides. Results: FGF19 positivity rates for IHC commonly considered contraindications to exploration. In 5 clinical vi-
and qRT-PCR were 18% (41/225) and 42% (87/209), respectively. The gnettes of LAPC, the proportion of respondents that would offer exploration
overall correlation was 60%, with 63% (22/235) IHC positive cases also following neoadjuvant varied extensively, from 15% to 54%. In a vignette of
being positive by qRT-PCR. For IHC+/qRT-PCR (-) cases, RNA quality may oligometastatic pancreatic liver metastases, 32% would offer exploration if a
have impacted assay sensitivity. 4% (9/244) of samples were positive for favorable biochemical and imaging response to therapy is observed.
FGF19 copy number. Among samples with FGF19 copy number gains, 22% Conclusions: In an international cohort of high volume pancreas surgeons,
(2/9) did not show positive FGF19 expression. Conclusions: Based on our there is substantial variation in attitudes regarding staging preferences and
data, FGF19 mRNA is a more inclusive selection strategy and offers an surgical management of LAPC. These results underscore the importance of
approach to further refine thresholds for efficacy as determined in the clinic. coordinated multi-disciplinary care, and suggest an evolving concept of
Multiplexed protein-mRNA assays were also validated and implemented to “resectability.” Patients and their oncologists should have a low threshold to
enable a more comprehensive clinical biomarker program. consider a second opinion for the surgical management of LAPC, if desired.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
252s Gastrointestinal (Noncolorectal) Cancer

4123 Poster Session (Board #228), Mon, 8:00 AM-11:00 AM 4124 Poster Session (Board #229), Mon, 8:00 AM-11:00 AM
Neoadjuvant FOLFIRINOX versus adjuvant gemcitabine in pancreatic A phase Ib dose-escalation and cohort-expansion study of safety and activity
cancer. First Author: Adam R Wolfe, Ohio State, Columbus, OH of the transforming growth factor (TGF) b receptor I kinase inhibitor galu-
nisertib plus the anti-PD-L1 antibody durvalumab in metastatic pancreatic
Background: In the metastatic or adjuvant setting for pancreatic cancer, the
cancer. First Author: Davide Melisi, University of Verona, Verona, Italy
combination chemotherapy of fluorouracil, leucovorin, irinotecan, and oxali-
platin (FOLFIRINOX) resulted in longer overall survival (OS) compared to Background: Pancreatic cancer (PC) is characterized by a highly immuno-
gemcitabine therapy. We conducted an institutional study to compare the suppressive microenvironment, and immune checkpoint inhibitors as mono-
efficacy of neoadjuvant modified FOLFIRINOX (neo-mFOLFIRINOX) to ad- therapy have been ineffective to date. TGFb is commonly viewed as a powerful
juvant gemcitabine (adj-gem) for pancreatic cancer patients who completed immunosuppressive cytokine, and inhibition of its signaling reverses this
resection. Methods: The study retrospectively enrolled patients from 2006 to suppression and activates adaptive immune responses. A combination of TGFb
2017 from Ohio State University. While patients who received adjuvant and PD-L1 inhibition may act synergistically to induce immune restoration and
gemcitabine were considered to be resectable upfront, patients who received to improve antitumor responses. This Phase 1b study (NCT02734160) eval-
neo-mFOLFIRINOX were either staged as borderline resectable (BR) or un- uated the combination of galunisertib plus durvalumab in recurrent or refractory
resectable (UR) by the institutional tumor board group. 111 patients received metastatic PC. Methods: Eligible patients (pts) were $18 years old, had ECOG
adj-gem (average cycles, 5.5) and 52 patients received neo-mFOLFIRINOX status #1, and had not received treatment with anti-PD-1, anti-PD-L1, or TGFb
(average cycles, 3.5). The survival rates were determined by the Kaplan-Meier R1 kinase inhibitors. The primary objective was to assess the safety and the
method and analyzed using Cox regression and log-rank test. Results: At a recommended dose of galunisertib given 14 days on/14 days off in combination
median follow up of 21.3 months, the median OS was 35.4 months in the neo- with durvalumab 1500 mg every 4 weeks. Four dose levels of galunisertib were
mFOLFIRINOX group and 21.8 months in the adj-gem group (hazard ratio, tested in the dose escalation portion of the study: 50 mg QD, 50 mg BID, 80 mg
0.56, 95% confidence interval (CI), 0.37-0.84 p = 0.005). The OS rate at 3 BID and 150 mg BID, followed by the cohort expansion portion of the study at
years was 46% in the neo-mFOLFIRINOX group and 22% in the adjuvant the recommended Phase 2 dose (RP2D). Secondary objectives included pre-
gemcitabine group (p = 0.001). The median disease free survival (DFS) was liminary assessment of activity by response rate, (RECIST v1.1), median PFS
18.6 months in the neo-mFOLFIRINOX group and 12.0 months in the adj-gem (mPFS), and OS (mOS). Results: 42 pts (25F/17M) were treated in the study
group (hazard ratio, 0.63, 95% CI, 0.43-0.93 p = 0.022). The DFS rate at 3 (median age 56.5 y; 71.4% had received $2 prior systemic regimens). There
years was 17% in the neo-mFOLFIRINOX group and 11% in the adj-gem group was no dose limiting toxicity and galunisertib 150 mg BID was chosen as the
(p = 0.02). On surgical pathological specimen review, the neo-mFOLFIRINOX RP2D. In the 32 pts treated at this dose, Grade $3 related AEs included AST
group had statistically (p , 0.05) lower tumor grade, lower rates of perineural and GGT elevations (2 pts each), and ALT and alkaline phosphatase elevations,
invasion and lymphovascular invasion, lower pathological T stage, lower and neutropenia (1 pt each). One partial response and 7/32 stable diseases
pathological N stage, and lower number of nodes positive compared to the adj- were observed (disease control rate 25%); mPFS was 1.9 months (95% CI: 1.5,
gem group. Frequencies of obtaining R0 resections were higher in the neo- 2.2) and mOS was NR (95% CI: 3.6, NR). Biomarker data will be presented at
mFOLFIRINOX versus adj-gem groups but not statistically different (51.9% vs the meeting. Conclusions: The combination of galunisertib plus durvalumab
40.4, p = 0.2). The average age and performance status were similar between had an acceptable tolerability and safety profile. The activity of this combination
the two groups. Conclusions: At our institution, BR and UR pancreatic cancer in second and third line PC patients warrants further consideration. Clinical trial
patients who received neo-mFOLFIRINOX and completed resection had longer information: NCT02734160.
OS, DFS, and more favorable pathological indicators compared to those pa-
tients who had upfront surgery and adjuvant gemcitabine. Randomized clinical
trials comparing neoadjuvant versus adjuvant FOLFIRINOX are needed to
validate these findings.

4125 Poster Session (Board #230), Mon, 8:00 AM-11:00 AM 4126 Poster Session (Board #231), Mon, 8:00 AM-11:00 AM
PanCO: An open-label, single-arm pilot study of phosphorus-32 (P-32; Onco- The effect of neoadjuvant chemotherapy with gemcitabine and S-1 for resect-
sil) microparticles in patients with unresectable locally advanced pancreatic able pancreatic cancer (randomized phase II/III trial; Prep-02/JSAP-05). First
adenocarcinoma (LAPC) in combination with FOLFIRINOX or gemcitabine + Author: Sohei Satoi, Department of Surgery, Kansai Medical University,
nab-paclitaxel (GNP) chemotherapies. First Author: Paul J. Ross, Guy’s and St Hirakata, Japan
Thomas’ NHS Foundation Trust, London, United Kingdom
Background: Despite recent progress of adjuvant chemotherapy for resected
Background: LAPC is associated with a poor prognosis. Current standard pancreatic ductal adenocarcinoma (PDAC), its survival remains limited. We
treatment is limited to chemotherapy or chemo-radiotherapy. P-32 Micro- conducted a randomized controlled trial to compare neoadjuvant chemo-
particles is a brachytherapy device that implants a predetermined dose of P- therapy (NAC) with upfront surgery (UP-S) for patients with resectable PDAC.
32 into pancreatic tumours via endoscopic ultrasound (EUS) guidance. This Methods: Patients with resectable PDAC, all confirmed cytologically or his-
reports the initial results of a pilot study in combination with chemotherapy. tologically were enrolled. Patients received 2 cycles of gemcitabine and S-1
Methods: Eligible patients were permitted to receive either GNP or FOL- regimen (GS) followed by surgery (NAC) or UP-S after randomization (1:1).
FIRINOX. P-32 was implanted at week 4 or 5. The dose of P-32 was cal- Patients in both arms received adjuvant chemotherapy using S-1 for 6 months
culated from tumour volume to deliver an absorbed dose of 100 Gy. Diffusion after surgical resection. The primary endpoint was overall survival (OS);
pattern of the P-32 suspension was assessed by EUS and bremsstrahlung secondary endpoints included adverse events, resection rate, recurrence-free
SPECT/CT imaging. Safety data was graded using CTCAE v4.0 criteria. survival, residual tumor status, nodal metastases, and tumor marker kinetics.
Response was assessed according to RECIST 1.1 with CT scans every Results: A total 362 patients were randomly assigned to NAC-GS (n=182) or
8 weeks and FDG-PET scans at baseline and week 12. Results: 50 patients UP-S (n=180) for 3 years (2013-16). The median OS was 36.7 months in
were enrolled (Intent-to-Treat population (ITT)) of which 42 were implanted NAC-GS and 26.6 months in UP-S; HR 0.72 (p=0.015, stratified log-rank test)
with the device (Per Protocol population (PP)). 10 received FOLFIRINOX and at 2.5 year after final enrollment. Crude resection rate for NAC and UP-S were
40 GNP. Median age was 65, 28 were male and all had a PS 0/1. 1070 77%, 72% respectively. There was no operative mortality in both groups.
adverse events (ITT) were reported; 153 (80% of patients) were $ Grade 3. Although G3/4 adverse events were observed frequently (73%) during NAC, no
The most common AEs of $ Grade 3 were haematological (39, 46%) and significant difference for both groups was observed for perioperative outcomes
gastrointestinal disorders (30, 34%). No serious device- or radiation-related including blood loss, operation time, R0 resection rate and post-operative
toxicities have been reported. PP Local Disease Control Rate at Week 16 was morbidity. Significant decrease of pathological nodal metastases in NAC was
90%; 95% CI: 77-97% and at Week 24 was 71%; 95% CI: 55-84%. Overall noted compared to those in UP-S by pathological evaluation for resected
Response Rate (ORR) was 31%; 95% CI: 18-47%. Median change in tumour patients(p,0.01). Although significant decrease of viable tumor cells was
volume from Baseline to Week 16 and to Week 24 was -38% (range +89% to observed in primary tumor after NAC compared to UP-S (p,0.01), Evans IIb or
-90%) and -27.5% (range +139% to -79%). Ten (24%) patients underwent more was found in only 14 % of resected patients in NAC. Hepatic recurrence
surgical resection following repeat staging. Eight patients had R0 margin. after surgery was significantly reduced in NAC (30.0%) compared to UP-S
Conclusions: The use of EUS-guided implantation of P-32 is feasible, with (47.5%) in observed period. Conclusions: The strategy of NAC showed
an acceptable safety profile in combination with first-line chemotherapy for significant longer survival compared to that of UP-S with acceptable fea-
LAPC patients. Encouraging OR and DCR are observed. Further follow-up to sibility. The effect of NAC might imply the control of subdiagnostic liver
inform results of local progression free survival and progression free survival metastases before surgery for resectable PDAC. Clinical trial information:
is warranted. Acknowledgement: Nab-paclitaxel was supported by Specialised UMIN000009634.
Therapeutics Australia Pty Ltd. Clinical trial information: NCT03003078.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Noncolorectal) Cancer 253s

4127 Poster Session (Board #232), Mon, 8:00 AM-11:00 AM 4128 Poster Session (Board #233), Mon, 8:00 AM-11:00 AM
Final results of JASPAC05: Phase II trial of neoadjuvant S-1 and concurrent NEONAX trial: Neoadjuvant plus adjuvant or only adjuvant nab-paclitaxel
radiotherapy followed by surgery in borderline resectable pancreatic cancer. plus gemcitabine for resectable pancreatic cancer, a phase II study of the
First Author: Shinichiro Takahashi, Department of Hepato-Biliary Pancreatic AIO pancreatic cancer group (AIO-PAK-0313)—Safety interim analysis.
Surgery, National Cancer Center Hospital East, Kashiwa, Japan First Author: Waldemar Uhl, Ruhr-University Bochum, St. Josef Hospital,
Bochum, Germany
Background: Borderline resectable pancreatic cancer (BRPC) is frequently
associated with positive surgical margins and a poor prognosis when treated with Background: Survival in pancreatic cancer (PDAC) is still poor even after cu-
upfront surgery. This study was designed to assess whether neoadjuvant che- ratively intended resection. Perioperative treatment approaches improve out-
moradiotherapy (CRT) with S-1 increases the R0 resection rate. Methods: This come in various tumor entities. Data on perioperative treatment in resectable
was a multicenter, single-arm, phase II study. Patients with BRPC received S-1 PDAC are limited and there is a debate whether neoadjuvant treatment might
(40 mg/m2 bid) and concurrent radiotherapy (50.4 Gy in 28 fractions) before impair subsequent surgery by adding perioperative morbidity or mortality.
surgery if they fulfilled any of the following: (1) bilateral impingement of the Methods: NEONAX is a randomized phase II study (planned 166 patients) of
superior mesenteric vein or portal vein; and (2) tumor contact #180° with the perioperative gemcitabine/nab-paclitaxel (Arm A: 2 pre- and 4 post-operative
superior mesenteric artery, common hepatic artery, or celiac axis. The primary cycles, Arm B: 6 cycles adjuvant) for patients with primarily resectable PDAC.
endpoint was the R0 resection rate in BRPC confirmed by central review. Primary objective is DFS at 18 months after randomization. Secondary objectives
Secondary endpoints were overall survival (OS), progression-free survival (PFS), are 3-year OS-rate and DFS-rate, progression during neoadjuvant therapy, R0/R1
response rate (RECISTv1.1), pathological response rate, surgical morbidity resection rate and QoL. Results: NEONAX was initiated in March 2015 in 26
(Clavien–Dindo classification), and toxicity (CTCAEv4.0). At least 40 patients centers for PDAC surgery in Germany. The data represent the safety interim
were required, with one-sided a = 0.05 and b = 0.05, with an expected and analysis (IA) of the first 48 patients. 25 patients were randomized to Arm A and
threshold value for the primary endpoint of 30% and 10%. Results: Fifty-two 23 to Arm B. Patients’ median age was 65.3 years (56.3% males, 43.8% fe-
patients were eligible, of whom 41 had BRPC by central review. CRT was males, 85.4% ECOG 0). Out of 25 patients in Arm A 20 patients (80%) un-
completed in 50 (96%) patients and was well tolerated. The rate of grade 3/4 derwent surgery, compared to 21 of 23 patients (91.3%) in Arm B with upfront
toxicity with CRT was 43%. The R0 resection rate was 52% (95% CI, surgery. Reasons for no resection were intraoperatively determined small liver
37.6%–66.0%) in 52 eligible patients and 63% (95% CI, 46.9%–77.9%) in metastases (2 cases, Arm A), withdrawal of informed consent (2 cases in each
arm) and 1 patient with uncontrolled cholestasis (arm A). Postoperative com-
41 patients with BRPC. The radiological response rate was 5.8%, while de-
plications occurred in 45% of arm A and 42.8% of arm B. (pancreatic fistula:
struction of . 50% of tumor cells was shown microscopically in 32% of patients.
15% in arm A and 9.5% in arm B, infections: 10% in arm A and 9.5% in arm B)
Postoperative grade III/IV adverse events were observed in 7.5% of operated
All resected patients were alive 60 days after surgery. At least 1 adverse event
patients. Among the 52 eligible patients, the 2-year OS rate, median OS, and
(AE) NCI-CTCAE $ grade 3 occurred in 60% of the perioperative and 39.1% of
median PFS were 51%, 25.8 mo, and 6.7 mo. Of the 41 patients with BRPC, the adjuvant treatment arm. Most common AEs were neutropenia (16.7%), fatigue
2-year OS rate, median OS, and median PFS were 58%, 30.8 mo, and 10.4 mo. (10.4%) and infections (10.4%). Conclusions: There was an increase in NCI-
Conclusions: S-1 and concurrent radiotherapy appear to be feasible and ef- CTCAE $ grade 3 events in the perioperative arm, but this was manageable and
fective at increasing the R0 resection rate with encouraging survival rates in did not result in increased peri- or postoperative mortality. 8% of patients in the
BRPC. A phase II/III trial evaluating this treatment is ongoing. Clinical trial perioperative arm did not get resected due metastases detectable during surgery,
information: NCT02459652. but not on preoperative imaging immediately prior to surgery. Therefore, it cannot
be determined whether these metastases were preexistent or developed during
neoadjuvant treatment. In conclusion, the first interim analysis of the NEONAX
trial shows that this protocol can be safely applied to patients with resectable
PDAC in a perioperative setting. Clinical trial information: NCT02047513.

4129 Poster Session (Board #234), Mon, 8:00 AM-11:00 AM 4130 Poster Session (Board #235), Mon, 8:00 AM-11:00 AM
Metformin use and pancreatic cancer survival in U.S. veterans with diabetes Relacorilant (RELA) with nab-paclitaxel (NP): Safety and activity in patients
mellitus: Are there racial differences? First Author: Adetunji T. Toriola, with pancreatic ductal adenocarcinoma (PDAC) and ovarian cancer (OvCA).
Washington University School of Medicine in St. Louis, St. Louis, MO First Author: Pamela N. Munster, University of California San Francisco, San
Francisco, CA
Background: Experimental and observational studies suggest that metformin
holds promise in improving survival among pancreatic cancer patients. How- Background: Glucocorticoid receptor (GR) pathway activation has been linked
ever, findings from prior observational studies have been questioned because with chemotherapy resistance (CTR). RELA (formerly CORT125134, Corcept
most did not control for immortal time bias, which can overestimate the survival Therapeutics), a potent selective GR modulator, in combination with paclitaxel
benefit of a drug. In addition, previous studies did not present data on African reduced CTR and enhanced activity against tumor growth in preclinical models
American patients. Thus, it is unknown if any survival advantage from metformin of solid tumors. Methods: Patients (pts) with advanced solid tumors, #3 prior
extends to African Americans. To address these limitations, we analyzed data lines of cytotoxic therapy, ECOG status 0-1, and adequate marrow function
from the U.S. Veterans Health Administration (VHA). Methods: A population- received RELA (100, 150, or 200mg) + NP (60, 80, or 100mg/m2). Once daily
based retrospective cohort study of 3,811 (N = 773 are African Americans) RELA was given either continuously (CON) or intermittently (INT) (day before,
pancreatic cancer patients with pre-existing diabetes mellitus diagnosed within of, and after NP). NP was dosed weekly for 3 of 4 weeks (wks) of a 28-day cycle.
the VHA between October 1, 1998 and December 30, 2010, and followed until Prior NP therapy was allowed. Results: 72 pts have been enrolled [mean age
December 2014. We calculated hazard ratios (HR) and 95% confidence in- 60 (range 18-81), mean number of prior therapies 3, prior taxane (TXN)
tervals (CI) using both the time-varying Cox proportional hazards regression treatment 54/72 (75%)]. 61 pts received $1 dose of RELA. Grade $3
model, which controls for immortal time bias, and conventional Cox model. AE $10% for CON: neutropenia (6/43, 14%); INT: neutropenia (6/18, 33%),
Analyses were adjusted for confounders. We also stratified analyses by race. anemia (2/18, 11%), and mucosal inflammation (2/18, 11%). Prophylactic G-
Further, we performed analyses among patients who were metformin naı̈ve (N = CSF became mandatory in later cohorts. Recommended Phase 2 Dose: RELA
1158) at the time of pancreatic cancer diagnosis (most representative of pa- 100mg-CON/150mg-INT + NP 80mg/m2 (exposures similar to NP 100mg/m2
tients enrolled in clinical trials). Results: Median survival was 4.5 months due to CYP3A4 inhibition by RELA). Disease control (DC) . 24 wks was noted
among metformin users versus 3.7 months among non-users. Metformin use in 5/27 (19%) PDAC pts: 3 PR, 2 SD (27-50 wks). 3 pts achieved benefit
was not associated with pancreatic cancer survival in analysis using the time- despite progression on prior TXN with time to progression (TTP) 1.9-3.6x
varying Cox model: HR = 1.05 (95% CI 0.92-1.14, P-value = 0.28). Results longer than prior TXN. 4/13 (31%) OvCA pts had DC . 24 wks: 1 CR, 1 PR, 2
were identical among non-Hispanic Whites and African Americans. In analysis SD (33-54+ wks). 1 pt had TTP 4.4x longer than prior TXN. 3 additional PRs
using conventional Cox model, metformin use was associated with an artificial were observed: acinar pancreatic cancer, TTP 31 wks (4.4x prior TXN); vulvar
survival benefit: HR = 0.89 (95% CI 0.83-0.98, P-value = 0.01). Among SCC HPV+, TTP 55 wks (3.9x prior TXN); cholangiocarcinoma, DC 29+ wks.
patients who were metformin naı̈ve at the time of pancreatic cancer diagnosis, Expression of GR-regulated genes involved in inflammation, apoptosis, and
metformin use was associated with improved survival in analysis using the time- CTR distinguished pts with DC from pts without DC, providing proof of
varying Cox model: HR = 0.77 (95% CI 0.61-0.98, P-value = 0.03). The HRs mechanism. Conclusions: RELA+NP resulted in durable disease control in pts
were 0.78 (95% CI 0.61-0.99, P-value = 0.04) among non-Hispanic Whites with metastatic PDAC, OvCA, and other solid tumors, including those that have
and 1.20 (95% CI 0.75-1.93, P-value = 0.45) among African American pa- progressed on prior TXN. TTP was often several-fold longer than previously
tients. Conclusions: We observed no associations between metformin use and achieved on TXN therapy. Toxicities are manageable with prophylaxis for
pancreatic cancer survival. Nevertheless, we noted improved survival (limited to neutropenia. Further evaluation in OvCA NCT03776812, PDAC, and others
non-Hispanic White patients) among patients who were metformin naı̈ve at the are planned. Clinical trial information: NCT02762981.
time of pancreatic cancer diagnosis, which requires conformation in other
studies.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
254s Gastrointestinal (Noncolorectal) Cancer

4131 Poster Session (Board #236), Mon, 8:00 AM-11:00 AM 4132 Poster Session (Board #237), Mon, 8:00 AM-11:00 AM
Elevated pretreatment serum IL-8 and PD-L1 and overall survival in a phase Homologous recombination deficiency (HRD): A biomarker for first-line (1L)
III randomized advanced pancreatic cancer clinical trial. First Author: Ashok platinum in advanced pancreatic ductal adenocarcinoma (PDAC). First
Maddukuri, Penn State Hershey College of Medicine, Hershey, PA Author: Wungki Park, Memorial Sloan Kettering Cancer Center, Department
of Medicine, Gastrointestinal Oncology, New York, NY
Background: We previously reported the prognostic and predictive utility of
pretreatment serum PD-L1 in the CCTG MA.31 serum bank (SABCS 2018, Background: HRD is an emerging biomarker for platinum therapy in PDAC. The
abstr PD3-10). IL-8 (CXCL8) is a pro-inflammatory cytokine that binds to clinical implications regarding differences in outcome between germline and
CXCR1 and CXCR2 and promotes tumor immune escape and progression. somatic HRD in advanced PDAC treated with 1L platinum is unexplored.
High serum IL-8 levels are associated with poor prognosis in many cancers, Methods: We evaluated overall survival (OS) for advanced PDAC (stage III/IV)
and have recently been reported to predict for reduced OS to nivolumab in based on their pathogenic germline (gHRD) and somatic HRD (sHRD) using
lung cancer and melanoma (ASCO 2018, abstr #3025). In this study, we integrated genomic profiling from MSK-IMPACT and 1L platinum use. HRD
retrospectively evaluated combined pretreatment serum IL-8 and PD-L1 on defined by pathogenic alterations from the following genes: BRCA1/2, PALB2,
overall survival (OS) from a phase III randomized pancreatic cancer trial of ARID1A/B/2, ATR, ATRX, ATM, BAP1, RAD50/51C/D, BRIP1, NBN, CHECK1/
first-line therapy (octreotide + 5-FU vs. 5-FU) that had reported no signif- 2, FANCA/C, CDK12, and MRE11. Results: Advanced PDAC patients (n=461)
icant OS difference between treatment arms. Methods: This study had 147 treated at MSK enrolled in a prospective database, were evaluated. Median
patients with serum available for this retrospective biomarker analysis from follow-up was 27.6 months (95% CI, 24.6-30.6). Both germline and somatic
an advanced pancreatic cancer phase III clinical trial.TheELLA immuno- profilings were available for n=350 (76%) but only somatic profiling was
assay platform (ProteinSimple, San Jose, CA) was utilized to quantitate available for n=111 (24%). We identified n=52 patients with gHRD (11.3%),
serum levels of IL-8 and PD-L1. Kaplan-Meier life table analysis was used to n=42 patients with sHRD (9.1%), and 48 patients with somatic VUS for HRD
correlate serum biomarkers with overall survival (OS). Results: In univariate genes. From all 461 patients, the OS was not different between 1L non-
analysis, pretreatment serum IL-8 was a significant biomarker as a con- platinum vs. 1L platinum groups (19 M vs. 19.3 M), regardleess of their HRD
tinuous variable (HR = 1.004; p = 0.012) and trended significant at the status. (Table) The OS was superior for gHRD vs. non-gHRD (28.7 M vs. 18.2
median cutpoint (HR = 1.379; p = 0.098) for OS, however serum PD-L1 was M), regardless of 1L treatment choice. However, similar significant OS supe-
not significant at any cutpoint. When serum PD-L1 and IL-8 levels were riority was neither observed in sHRD vs. non-sHRD, nor in VUS sHRD vs. non-
analyzed as combined biomarkers (median cutpoints), the serum IL-8 high / VUS sHRD. In a subgroup analysis of 1L platinum treated patients, the OS was
PD-L1 high cohort had a significantly shorter OS vs the serum IL-8 low / PD- superior in gHRD vs. non-gHRD (NR vs. 17.9 M); however, there was no OS
L1 low cohort (HR = 1.816; p = 0.017). Conclusions: In this phase III difference between sHRD and non-sHRD. Conclusions: In advanced PDAC
randomized clinical trial in advanced pancreatic cancer, pretreatment serum patients, only gHRD predicted better overall survival for first-line platinum
IL-8 was a significant biomarker for OS, but serum PD-L1 was not. Higher chemotherapy. These findings emphasize the importance of germline mutation
combined pretreatment serum levels of PD-L1 and IL-8 (both biomarkers testing of HRD in PDAC. Biomarker validation and functional definition of HRD
high vs. both low) were prognostic for reduced OS in this phase III pancreatic such as loss of heterozygosity analysis is underway.
cancer trial. Further study of circulating IL-8 and PD-L1 is warranted in All treatments (461) 1L non-platinum (184) 1L platinum (277)
pancreatic cancer for evaluation of targeted and investigational therapies,
(N) OS (M) HR p-value OS (M) HR p-value OS (M) HR p-value
including the immune checkpoint inhibitors and anti-IL8 therapy.
All pts (461) 19.0 - - 19.0 - - 19.3 0.95 0.676
No gHRD (409) 18.2 0.54 0.003 18.8 0.74 0.419 17.9 0.47 0.004
HRD (52) 27.7 24.3 NR
No sHRD (419) 18.3 0.92 0.670 18.2 0.68 0.252 18.3 1.12 0.666
sHRD (42) 21.5 23.8 20.6
Overall survivals for different HRD groups and 1L treatment groups.

4133 Poster Session (Board #238), Mon, 8:00 AM-11:00 AM 4134 Poster Session (Board #239), Mon, 8:00 AM-11:00 AM
Association of BRCA-mutant pancreatic cancer with high tumor mutational Hereditary cancer genetic testing among patients with pancreatic cancer.
burden (TMB) and higher PD-L1 expression. First Author: Andreas Seeber, First Author: Nassim Taherian, Myriad Genetics, Inc., Salt Lake City, UT
Department of Internal Medicine V (Hematology and Oncology), Innsbruck,
Background: Pancreatic cancer (PC) is typically diagnosed at a late, untreatable
Austria stage, with a 5-year survival rate of only ~8%. Genetic testing for individuals with PC
Background: In the U.S. 56,000 Americans are expected to be diagnosed with may aid in therapy decisions, as those with a germline or somatic pathogenic variant
pancreatic cancer in 2019. Prognosis in pancreatic cancer is poor. Therefore, (PV) in a DNA-repair gene may benefit from PARP inhibitors. In addition, germline
new treatment strategies are urgently needed to improve survival. BRCA1 and genetic testing for unaffected family members can identify high risk individuals who
BRCA2 mutations have been described to be the most common genetic may be appropriate for surveillance studies. We assessed the results of hereditary
cancer genetic testing among individuals with a personal history of PC and evaluated
mutations involved in familial pancreatic cancer. The optimal treatment
several possible risk factors. Methods: Individuals with PC who had germline testing
regimen to use in BRCA-mutant pancreatic cancer has still to be established.
for 25-29 cancer-susceptibility genes between September 2013 and November
Moreover, no data are available on association of BRCA mutation with immune- 2018 were included in this analysis (N = 1,676). Clinical characteristics were ob-
associated markerssuch as tumor mutational burden (TMB), microsatellite tained from provider-completed test request forms and included personal cancer
instability (MSI) or PD-L1 expression. Methods: Tumor samples of 2824 history (PHx), family cancer history (FHx), and age at diagnosis. Results: Overall,
patients with pancreatic ductal adenocarcinoma were analyzed for BRCA 12.6% (212/1676) of patients with PC carried a PV, most commonly in BRCA2
mutation by NGS and for other genes (MiSeq on 47 genes, NextSeq on 592 (3.8%), ATM (2.7%), and PALB2 (1.2%). PVs were more common in men and for
genes) at Caris Life Sciences, Phoenix, AZ. TMB was calculated based on individuals who had a PHx of additional cancer(s) (see Table). Age at PC diagnosis did
somatic nonsynonymous missense mutations, and MSI was evaluated by NGS not impact the positive PV rate. The PV positive rate was elevated among individuals
of known MSI loci. PD-L1 expression was evaluated using immunohisto- with PC and at least two relatives with PC (15.1%) and for individuals with a FHx of
chemistry. Results: In 4.4% (N = 124) of all pancreatic adenocarcinoma cancer at an early age (14.2%). The PV positive rate remained . 10% regardless of
samples BRCA mutations were detected. BRCA2 mutations were more nearly all other FHx characteristics evaluated, including the absence of any FHx (see
common: 3.1% (N = 89) vs 1.1% BRCA1 mutations (N = 35). BRCA mutations Table). Conclusions: In this cohort, a substantial proportion of individuals with a PHx
were associated with younger age (BRCA1: 61 yrs for mutated vs. 64 for wild- of PC carried PVs, regardless of age at diagnosis and personal and family cancer
type, p = 0.07; BRCA2: 61 yrs vs. 64, p = 0.002; both: p , 0.001). BRCA history.
mutations were associated with higher MSI-H frequency (4.8% vs. 1.2%, p = Risk Factor Total Positive PV Rate
0.002), elevated PD-L1 expression (22% vs. 11%, p , 0.001) and higher Cohort 1676 12.6%
Gender
TMB (mean 8.7 mut/MB vs. 6.5, p , 0.001); the differences remain sig- Men 542 15.3%
nificant in MSS tumors (p , 0.05). BRCA-mutant pancreatic carcinomas Women
PHx Cancer
1134 11.4%

showed a significantly lower mutation frequencies in TP53 (59% vs 73%, p = PC Only 958 11.3%
PC + Other Cancer(s) 718 14.5%
0.001), CKDN2A (13% vs 25%, p = 0.006), but higher frequencies in APC Age at PC diagnosis
(6.5% vs 2.2%), KMT2A (1.9% vs 0.2%), AMER1 (1.9 vs 0.5%) and SETD2 £50 years
> 50 years
265
1278
12.8%
13.1%
(3.7% vs 0.4%) mutations (p , 0.05 for all comparisons). Conclusions: BRCA FHx of PC
Any FHx of PC 544 12.7%
mutations are found in a significant subgroup of pancreatic ductal adeno- ‡2 relatives with PC 126 15.1%
carcinoma and these carcinomas are associated with an immunogenic tumor 1 relative with PC
No FHx of PC
418
1132
12.0%
12.6%
profile. These data suggest evaluating PARP inhibitors in combination with FHx of Any Cancer
FHx of Other Cancer(s) 1500 12.8%
immunotherapy in patients with BRCA-mutant pancreatic adenocarcinoma ‡2 relatives with Other Cancer(s) 1184 13.9%
1 relative with Other Cancer(s) 316 8.5%
especially in tumors that are MSS. No FHx 97 10.3%
Age at Diagnosis for FHx
Any Cancer Diagnosed £50 815 14.2%
Any Cancer Diagnosed > 50 651 10.8%

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Noncolorectal) Cancer 255s

4135 Poster Session (Board #240), Mon, 8:00 AM-11:00 AM 4136 Poster Session (Board #241), Mon, 8:00 AM-11:00 AM
Clinical and immune responses using anti-CD3 x anti-EGFR bispecific antibody Methylated circulating tumor DNA (Met-DNA) as an independent prognostic
armed T cells (BATs) for locally advanced or metastatic pancreatic cancer. First factor in metastatic pancreatic adenocarcinoma (mPAC) patients. First
Author: Lawrence G. Lum, University of Virginia, Charlottesville, VA Author: Daniel Pietrasz, Paul Brousse Hospital, Villejuif, France
Background: Conventional chemotherapy (chemo) for locally advanced Background: Circulating tumor DNA has emerged as prognostic biomarker in
pancreatic cancer (LAPC) and metastatic pancreatic cancer (MPC) has oncology. Many different genes can be mutated within a tumor, complicating
dismal responses and poor survival rates. Arming activated T cells (ATC) with procedures, even with highly sensitive next-generation sequencing (NGS).
anti-CD3 x anti-EGFR bispecific antibody (BATs) makes every ATC into an DNA methylation in promotor of specific genes is an early key epigenetic
EGFR-specific cytotoxic T cell that secretes cytokines, proliferates, and kills change during oncogenesis. Specific methylated genes could be a potential
tumor. Methods: We report on 5 phase I (P1) and 15 phase II (P2) patients. relevant cancer biomarker that may substitute for NGS panels. The aim of
In our phase I study, BATs were used to treat LAPC or MPC patients at this study was to assess the prognostic value of Met-DNA in mPAC.
Karmanos Cancer Institute (NCT0140874) in a dose escalation involving Methods: Prognostic value of Met-DNA was assessed in a prospective cohort
3 weekly infusions of 1, 2, and 4 x 1010 BATs/infusion, followed by a booster (PLAPAN) of mPAC (training cohort), correlated with NGS, then in two
infusion at 3 months (mos) for a total of up to 8 x 1010 BATs. No dose limiting prospective independent validation cohorts from two randomized phase II
toxicities were observed in the outpatient infusions. Fifteen patients treated trials (PRODIGE 35 and 37). Plasma samples were collected before che-
on a phase II (NCT02620865) at KCI and (NCT03269526) at University of motherapy on EDTA-coated tubes. Met-DNA was quantified using two
Virginia received biweekly infusions of 1010 BATs/infusion over 4 weeks for a specific markers of pancreatic DNA methylation by digital droplet PCR and
total of 8 x 1010 EGFR BATs. Results: Four patients had stable disease (SD) correlated with prospectively registered patient (pts) characteristics and
for 6.1, 6.5, 5.3, and 36 mos. Two patients had complete responses (CR) oncologic outcomes (progression free survival (PFS) and overall survival
when chemo was restarted after BATs. The median overall survival (OS) for (OS)). Results: 330 patients (pts) were enrolled. 60% (n = 58) of the 96 pts
17 evaluable patients (3 of 4 infusions in the P1 and all 8 infusions in the P2) of the training cohort had at least one Met-DNA marker. The correlation with
was 31 mos, and the median OS for all 20 patients (3 in the P2 who did not NGS assessment was R = 0.93 (Pearson; p , 0.001). 59.5% (n = 100/168)
complete 8 infusions) is 14.5 mos (95% CI, 7.5-45.2 mos). Patient and 59% (n = 39/66) of pts had detectable Met-DNA in the 2 validation
IT20104 had an apparent “pseudoprogression” after 3 BATs infusions, but cohorts. In the training cohort, Met-DNA was correlated with poor OS (HR =
achieved a CR after restarting capcitabine and is alive off therapy at 54 mos 1.82; 95%CI 1.07-2.42; p = 0.026). In validation cohorts, Met-DNA was a
(24 mos after stopping capecitabine). Immune evaluations on the P1 pa- prognostic factor of PFS (HR = 1.62; 95%CI 1.17-2.25, p = 004) and OS
tients show specific cytotoxicity to MiaPaCa-2 by peripheral blood mono- (HR = 1.79; 95%CI 1.28-2.49, p , 0.001) in PRODIGE 35, as in PRODIGE
nuclear cells (PBMC) increased from 21% to 31% 2 weeks after the 3rd 37: PFS HR = 1.79 (95%CI 1.07-2.99; p = 0.026) and OS HR = 2.08 (95%
infusion, and IFN-g EliSpots increased from , 20 to 1000 IFN-g EliSpots/ CI [1.18-3.68], p = 0.01), respectively. In multivariate analysis adjusted on
106 PBMC (p , 0.03). Patient IT 20121 (SD for 36 mos) increased IFN-g gender, age, CA19-9 . 40UI.mL, treatment arm, number of metastatic sites
EliSpots from 250 to 3200/106 PBMC after 8 infusions. Innate cytotoxicity and stratified on center, Met-DNA was independently associated with poor
responses in the P1 patients increased significantly after infusions (p , OS in both trials: HR = 1.81 (95%CI 1.10-2.98; p = 0.02) and HR = 3.62
0.04). Levels of IP-10 increased significantly (p , 0.04), and levels of IL-8 (95%CI: 1.32-9.93; p = 0.01). Conclusions: This study demonstrates that
decreased but not significantly (p , 0.07). Conclusions: Infusions of BATs Met-DNA is a strong independent prognostic factor in mPAC. These results
are safe and induce endogenous adaptive anti-tumor responses. Targeting argue for patient’s stratification on ctDNA status for further randomized
PC with BATs may stabilize disease, leading to improved OS, as well as trials. Clinical trial information: NCT02827201 and NCT02352337.
evidence that BATs infusions can induce anti-tumor activity and immuno-
sensitize tumors to subsequent chemo. Clinical trial information:
NCT014084,NCT03269526,NCT02620865.

4137 Poster Session (Board #242), Mon, 8:00 AM-11:00 AM 4138 Poster Session (Board #243), Mon, 8:00 AM-11:00 AM
SWOG S1505: Initial findings on eligibility and neoadjuvant chemotherapy Improved overall survival (OS) for advanced pancreatic cancer (PDAC)
experience with mFOLFIRINOX versus gemcitabine/nab-paclitaxel for re- patients (pts) enrolled in the Know Your Tumor (KYT) program whose tumors
sectable pancreatic adenocarcinoma. First Author: Davendra Sohal, harbored highly actionable molecular alterations and who received
Cleveland Clinic, Cleveland, OH molecularly-matched therapies (tx). First Author: Michael J. Pishvaian,
Georgetown University, Washington, DC
Background: Clinical outcomes after curative therapy of resectable pancreatic
ductal adenocarcinoma (PDA) remain suboptimal. For early control of systemic Background: Initial results from the KYT program demonstrated that 27% of
disease with aggressive perioperative chemotherapy (CTx), we conducted a PDACs harbor highly actionable molecular alterations (herein labelled
prospective trial in the National Clinical Trials Network (NCTN) setting. “actionable biomarkers”), defined as biomarkers that predict for a high
Methods: S1505 was a randomized phase II trial of periop (12 weeks pre-, response rate to appropriately targeted tx, in any cancer type. Within this
12 weeks post-op) CTx with either mFOLFIRINOX (5-fluorouracil, irinotecan, cohort, the median progression-free survival on molecularly-matched tx was
oxaliplatin – without bolus 5-FU and leucovorin; Arm 1), or gemcitabine/nab- 2 months longer than unmatched tx. Here, we present OS data emphasizing
paclitaxel (Arm 2). Eligibility required adult patients with ECOG PS 0 or 1, the 125 pts with “actionable biomarkers” who did or did not receive
confirmed tissue diagnosis of PDA, and resectable disease: no involvement of molecularly-matched tx. Methods: PanCAN and Perthera have coordinated
the celiac, common hepatic, or superior mesenteric arteries (and, if present, tumor molecular profiling through commercial labs (NGS/IHC panels) for
variants); , 180° interface between tumor and vessel wall, of the portal or PDAC pts since 2014. Results are reviewed by a molecular tumor board, and
superior mesenteric veins; patent portal vein/splenic vein confluence; no me- tx options are prioritized based on the actionable biomarkers, in the context
tastases. Primary outcome is 2-year overall survival (OS), using a “pick the of the pt’s tx history. Pts are followed longitudinally to track physician tx
winner” design; for 100 eligible patients, accrual up to 150 patients was choices and survival outcomes. Cox regression was used to assess differ-
planned, to account for cases deemed ineligible at central radiology review. ences in OS (measured from date of diagnosis until death). Results: Of 1053
Results: From 2015 to 2018, 147 patients were enrolled; 74 to Arm 1; 73 to pts who received a Perthera Report, 25% had “actionable biomarkers”. OS
Arm 2. At central radiology review, 42/147 (29%) were ineligible; of these, 15 analyses across 454 pts with adequate tx history are shown in the Table
(36%) had venous involvement $180°, 22 (52%) had arterial involvement, 28 below. Notably, pts with “actionable biomarkers” who received a
(67%) had distant disease. One patient had distal cholangiocarcinoma (in- molecularly-matched tx had a significantly increased OS compared to those
eligible); one withdrew consent after randomization. Eligible patients (n = 103) with “actionable biomarkers” but who did not receive molecularly-matched
had median age 64 years; males 58%; whites 89%; PS 0 64%. Of 103, 99 tx. Subgroup analyses related to tx history and specific molecular pathways
(96%) started and 86 (83%) completed preop CTx. There was one death due to that warrant further investigation will be discussed. Conclusions: When the
sepsis and 61 additional patients experienced grade 3/4 toxicities. To date, 76 ~25% of PDAC pts whose tumors harbored “actionable biomarkers” received
of 99 (77%) patients went to surgery and 72 (73%) underwent resection. molecularly-matched tx, they had a better OS. These findings support the
Conclusions: This is the first-ever NCTN study of periop CTx for resectable PDA. need to test all pts with PDAC, and just as importantly, to maximize access to
Accrual was brisk, establishing feasibility. Ineligible cases after central radi- molecularly-matched tx for appropriate pts, to achieve the best pt outcomes.
ology review highlight quality control and physician education imperatives for
neoadjuvant PDA trials. Preop CTx safety and resection rates are encouraging. # mOS
Cohort Patients (years) p-value (HR [95% CI])
Follow up for OS is ongoing. Clinical trial information: NCT02562716.
“Actionable biomarkers”, Matched Tx 25 2.58 --
vs. “Actionable biomarkers”, Un- 100 1.51 0.00700 (0.46 [0.26-
matched Tx 0.81])
vs. No “actionable biomarkers” 329 1.42 0.00125 (0.42 [0.25-
0.71])

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
256s Gastrointestinal (Noncolorectal) Cancer

4140 Poster Session (Board #245), Mon, 8:00 AM-11:00 AM TPS4141 Poster Session (Board #246a), Mon, 8:00 AM-11:00 AM
BIONADEGE: Genomic profiling of small bowel adenocarcinoma from the TENERGY: Multicenter phase II study of atezolizumab monotherapy follow-
NADEGE prospective cohort. First Author: Thomas Aparicio, Department of ing definitive chemoradiotherapy with 5-FU plus cisplatin in patients with
Gastroenterology, Saint Louis Hospital, Paris, France locally advanced esophageal squamous cell carcinoma. First Author:
Hideaki Bando, Aichi Cancer Center Hospital, Nagoya, Japan
Background: Small bowel adenocarcinoma (SBA) is a rare tumour. Large ge-
nomic analyses with prognostic assessments are lacking. Methods: BIONADEGE Background: The standard treatment for patients with unresectable locally
is an ancillary study of the NADEGE cohort that enrolled 347 patients (pts) with advanced esophageal squamous cell carcinoma (ESCC) is definitive chemo-
SBA from 2009 to 2012. Next generation sequencing investigates the presence radiotherapy (CRT) using 5-FU plus cisplatin. However, complete response (CR)
of 740 hot spot somatic mutations in 46 genes involved in carcinogenesis. The rates are only 11% to 25%, and median overall survival (OS) is 9 to 10 months.
MSI (MicroSatellite Instable) status was assessed using 5 microsatellites. The The improved therapeutic efficacy of combining immunotherapy with radiation
MMR (MisMatch Repair) status was assessed by immunochemistry (4 anti- has been gaining interest. Our basic research suggested that sequential
bodies). Results: A total of 196 tumour samples were collected and 125 pts had treatment with anti-PD-L1 agents soon after completion of CRT is the best
conclusive results for mutation analysis. The clinical and tumours characteristics combination. Twelve months of anti-PD-L1 antibody following platinum-based
were comparable in the NADEGE and BIONADEGE cohort except for metastatic CRT significantly improved progression-free survival (PFS) and OS in patients
stage at diagnostic underrepresented in the BIONADEGE cohort (17.7%) due to with locally advanced non-small cell lung cancer (Antonia SJ, et al. N Engl J
missing tumour sample. A predisposing disease was reported in 25 (20.0%) Med. 2018). Based on this background information, we have planned a phase II
cases (among them 14 Lynch syndromes and 7 Crohn diseases). The number of clinical trial to evaluate the safety and efficacy of atezolizumab monotherapy
mutation observed was 0 in 9.6% pts, only 1 in 32.0%, 2 in 26.4% and $3 in following definitive CRT in patients with locally advanced ESCC. Methods: The
32.0%. The most frequent genomic alteration were KRAS (44.0%), TP53 main inclusion criteria are unresectable locally advanced ESCC without distant
(38.4%), PIK3CA (20.0%), APC (18.4%), SMAD4 (14.4%) and ERBB2 metastasis, completion of treatment with 60 Gy of radiation plus two con-
(7.2%). Altogether, a genomic alteration was observed in 90.3% of tumour. comitant cycles of chemotherapy (cisplatin 70 mg/m2 on day 1 and 5-FU
KRAS mutation were more frequent in synchronous metastatic tumour than in 700 mg/m2 on days 1–4, every 28 days), and adequate organ function. Within
localized tumour (72.7% vs 38.2%, p = 0.003). There was no significant 4 weeks after CRT, patients will be registered in the study and started on
difference of mutation rate according to primary location for the most frequently 1200 mg of atezolizumab every three weeks until 12 months or disease pro-
altered gene. With caution to small sample, IDH1 mutation is more frequent and gression. The primary endpoint is the CR rate by the investigator’s assessment.
APC mutation never seen in Crohn disease. The rate of dMMR tumor was 38.6% Overall response rate, PFS, OS, treatment-related adverse events, and CR rate
in localized tumour and 0% in synchronous metastatic tumour. After a median by central assessment are secondary endpoints. A total of 50 patients will be
follow-up of 55 months (95%CI [44-63]), M0 stage, pN0, pT1-2 were associate enrolled, including 40 with primary locally advanced ESCC and 10 with
with better survival in univariate analysis. No significant prognostic value of postoperative loco regionally recurrent ESCC. As an exploratory biomarker study,
genomic alteration was associated with OS. dMMR status was associate with a biopsies from the primary site and blood collections will be performed at 3 time
better prognosis for OS in pts with MMR status determined by immunohisto- points (before CRT, after CRT, and four weeks after the start of atezolizumab).
chemistry (HR = 0.55 [0.29-1.01], p = 0.055). Conclusions: A high frequency of We will analyze the phenotype of immune-competent cells, neoantigens, tumor
targetable alteration is observed in SBA. There is several specificities according mutation burden, PD-L1 status, and Human Leukocyte Antigen haplotyping.
to predisposing disease. No association between genomic alteration and prog- Clinical trial information: UMIN000034373.
nostic was observed except a trend for a better prognosis associate with dMMR.

TPS4142 Poster Session (Board #246b), Mon, 8:00 AM-11:00 AM TPS4143 Poster Session (Board #247a), Mon, 8:00 AM-11:00 AM
Perioperative atezolizumab in combination with FLOT versus FLOT alone in CA224-060: A randomized, open label, phase II trial of relatlimab (anti-LAG-3)
patients with resectable esophagogastric adenocarcinoma: DANTE, a ran- and nivolumab with chemotherapy versus nivolumab with chemotherapy as
domized, open-label phase II trial of the German Gastric Group of the AIO first-line treatment in patients with gastric or gastroesophageal junction
and the SAKK. First Author: Salah-Eddin Al-Batran, Institute of Clinical adenocarcinoma. First Author: Kynan Feeney, St. John of God Hospital,
Research (IKF) at Krankenhaus Nordwest, UCT-University Cancer Center, Murdoch, Notre Dame University, Fremantle and Edith Cowan University,
Frankfurt, Germany Joondalup, WA, Australia
Background: Perioperative FLOT chemotherapy has become a standard of Background: Blockade of the immune checkpoint receptor programmed death-
care for locally advanced, resectable gastric cancer and adenocarcinoma of 1 (PD-1) has shown clinical benefit in multiple tumor types. Nivolumab
the GEJ. However, patient outcomes are still unsatisfactory and 5-year (anti–PD-1) has demonstrated a survival advantage versus (vs) placebo in
survival in T3-4 or nodal positive disease is still around 50%. Targeting the patients (pts) with advanced gastric cancer (GC) or gastroesophageal junction
PD-1/PD-L1 pathway has proven active in different cancers, including cancer (GEJC) (Kang YK et al. Lancet 2017;390:2461–71). Lymphocyte
esophagogastric cancer, and was associated with response rates in the 10- activation gene 3 (LAG-3) is an immune checkpoint molecule that negatively
15% range in unselected, heavily pre-treated gastric cancer patients. regulates effector T-cell function and is a marker of T-cell exhaustion. Pre-
Atezolizumab is a PD-L1 inhibitor with established efficacy and tolerability liminary data in melanoma suggest that combining nivolumab with relatlimab
profiles. This study evaluates atezolizumab in the perioperative treatment of (anti–LAG-3) could improve efficacy without substantially increasing toxicity vs
locally advanced, potentially resectable gastric or GEJ adenocarcinoma in nivolumab especially, but not exclusively, in LAG-3-expressing pts (Ascierto PA
combination with FLOT. Methods: This is a large, multinational, prospective, et al. Ann Oncol 2017;28(S5):LBA18). Furthermore, LAG-3 expression was as
multicenter, randomized, investigator-initiated, open label phase II trial. high as 33% in an analysis of solid tumors including GC (Edwards R et al. J
Patients with locally advanced, potentially resectable adenocarcinoma of the Immunother Cancer 2017;5(S3):P510). Study CA224-060 will assess the
stomach and GEJ ($cT2 and/or N-positive) without distant metastases are clinical efficacy and safety of relatlimab and nivolumab with chemotherapy for
enrolled. Eligibility status is centrally evaluated. Patients are randomized 1: first-line treatment of GC or GEJC. Methods: This is a randomized, open-label,
1 to 4 pre-operative 2-week cycles (8 weeks) of FLOT (Docetaxel 50 mg/m²; multicenter, phase 2 study of relatlimab and nivolumab with oxaliplatin-based
Oxaliplatin 85 mg/m²; Leucovorin 200 mg/m²; 5-FU 2600 mg/m²) followed chemotherapy vs nivolumab with oxaliplatin-based chemotherapy. Approxi-
by surgery and 4 additional cycles of FLOT plus atezolizumab at 840 mg mately 250 adult pts with untreated, locally advanced, unresectable or met-
every 2 weeks, followed by a total of 8 additional cycles of atezolizumab at astatic GC or GEJC will be enrolled. To be randomized, pts must have tumor
1200 mg every 3 weeks as monotherapy (arm A) or FLOT alone (arm B). tissue for analysis of biomarkers, LAG-3 status, and PD-L1 combined positive
Primary endpoint is time to disease progression or relapse after surgery (PFS/ score. Key exclusion criteria include HER2-positive status, untreated CNS
DFS) as assessed by the Kaplan-Meier-Method. The statistical design is metastases, or significant cardiovascular disease. The primary endpoint is
based on a target HR of 0.68, a power of 0.8, and a significance level of p, objective response rate (ORR) using RECIST v.1.1 by blinded independent
0.05 (1-sided log rank test). A total of 295 patients will be randomized. Main central review in LAG-3-expressing pts. Other endpoints include investigator-
secondary endpoints are rates of centrally assessed pathological regression assessed ORR, ORR in LAG-3-negative pts, duration of response, overall sur-
(rates of complete and nearly complete pathological regression), overall vival, progression-free survival, and safety and tolerability. Efficacy signals in
survival, R0 resection, and safety. Recruitment started in Sept 2018; by biomarker subgroups will be explored. Currently, 26 sites are activated with 15
February 2019, a total of 27 patients have been randomized. Clinical trial randomized pts. Clinical trial information: NCT03662659.
information: NCT03421288.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Noncolorectal) Cancer 257s

TPS4144 Poster Session (Board #247b), Mon, 8:00 AM-11:00 AM TPS4145 Poster Session (Board #248a), Mon, 8:00 AM-11:00 AM
Modified FOLFOX versus modified FOLFOX plus nivolumab and ipilimumab Phase II study of a telomerase-specific oncolytic adenovirus (OBP-301,
in patients with previously untreated advanced or metastatic adenocarci- Telomelysin) in combination with pembrolizumab in gastric and gastroesoph-
noma of the stomach or gastroesophageal junction: Moonlight, a randomized ageal junction adenocarcinoma. First Author: Uqba Khan, Weill Cornell
phase 2 trial of the German Gastric Group of the AIO. First Author: Medical College, New York, NY
Salah-Eddin Al-Batran, Institute of Clinical Research (IKF) at Krankenhaus
Background: Although checkpoint inhibitors (CPIs) can produce durable re-
Nordwest, UCT-University Cancer Center, Frankfurt, Germany
sponses in gastric cancer patients (pts) in the 3rd line setting, the response rate
Background: The majority of patients (pts) with gastroesophageal cancer is only 10-15%. Therefore, there is a huge unmet need to enhance the re-
present with inoperable or metastatic disease and there is a strong need for sponse rate of CPIs to provide benefit to wide range of pts. A novel concept in
efficient and tolerable first-line (1L) treatment. Oxaliplatin-based regimens immuno-oncology is the use of cancer specific oncolytic viral therapy. In
like FOLFOX have become one standard of care. However, median survival is addition to the specific killing of the tumor by the virus, these agents can
still below 12 months. Results from trials using nivolumab plus ipilimumab induce an immunogenic cell death in the tumor to augment the immune
treatment of subjects with advanced/metastatic GC and GEJ cancers dem- activation driven by PD-1 inhibition. OBP-301 is an oncolytic adenovirus
onstrated clinical activity, in pts whose tumors did or did not express PD-L1; in genetically modified to be able to selectively replicate in cancer cells by in-
addition, nivolumab alone and in combination with ipilimumab demonstrated troducing human telomerase reverse transcriptase (hTERT) promoter. Results
clinical benefits in various other tumor types. Based on this clinical experience, of a phase I study of OBP-301 in solid tumor pts demonstrated the safety and
the AIO-STO-0417 trial (Moonlight) has been designed to evaluate the efficacy of intra-tumoral injection of OBP-301. A pre-clinical study of the
combination of chemotherapy with two checkpoint inhibitors in first-line combination of OBP-301 with anti-PD-1 antibody has also shown significant
therapy of pts with gastroesophageal adenocarcinoma. Methods: This is a synergistic activity as well. Based on these encouraging pre-clinical and
prospective, multicenter, randomized, investigator-initiated phase II trial. Pts clinical data, we designed a phase II clinical trial to examine the safety and
with Her2-negative, inoperable, advanced or metastatic gastric or esoph- efficacy of combination of pembrolizumab and OBP-301 in the treatment of
agogastric junction cancer will be randomized 1:1 to 1L treatment with PD-L1 positive metastatic gastric/GEJ adenocarcinoma. Methods: This is a
FOLFOX (Oxaliplatin 85 mg/m²; Leucovorin 400 mg/m²; 5FU 400 mg/m² on multicenter, non-randomized phase II trial of OBP-301 with pembrolizumab in
d1 of each treatment cycle and 5FU 1200 mg/m² continuous infusion over metastatic gastric/GEJ adenocarcinoma that has progressed on at least 2 lines
24 hrs d1 and d2) every 2 weeks plus Nivolumab 240 mg every 2 weeks and of prior therapy. Eligibility criteria include PD-L1 positive tumors as defined
Ipilimumab 1mg/kg every 6 weeks (Arm A) or FOLFOX alone (Arm B). Primary by a combined positive score, performance status #1, and good end organ
endpoint of the trial is progression-free survival based on the ITT population. function. The primary endpoints are to examine objective response rate and
Main secondary endpoints are overall survival, objective response rate, Safety safety of OBP-301 with pembrolizumab. The secondary endpoints are to
and Quality of life (EORTC QLQ-C30). 118 pts (59 per arm) will be enrolled to examine disease control rate, duration of response, overall survival and pro-
provide 80% power for detecting an average HR of 0.68 using the log rank test gression free survival. Correlative studies are planned to identify biomarkers for
at a one-sided type I error of 10%. At the date of submission, (Feb 2019), 28 of response to combination therapy by using multiparameter flowcytometry,
planned 118 pts are randomized. Clinical trial information: NCT03647969. single-cell transcriptional profiling and immunohistochemistry. All eligible pts
will receive 1x1012Viral Particles/mL of OBP-301 administered every 2 weeks
for total of 4 injections, injected directly into tumor via upper endoscopy. Every
pt will also receive pembrolizumab 200 mg IV every 3 weeks for 2 years or until
progression. Pts will be enrolled in a Simon two stage design, with 18 pts in the
first stage. If 3 or more pts respond to the combination therapy, the study will
move forward to stage 2, with 19 more pts enrolled. The study is currently
enrolling pts.

TPS4146 Poster Session (Board #248b), Mon, 8:00 AM-11:00 AM TPS4147 Poster Session (Board #249a), Mon, 8:00 AM-11:00 AM
KEYNOTE-811 pembrolizumab plus trastuzumab and chemotherapy for A single-arm, open phase II clinical trial of anti-programmed death-1
HER2+ metastatic gastric or gastroesophageal junction cancer (mG/GEJC): antibody SHR-1210 combined with nimotuzumab as second-line treatment
A double-blind, randomized, placebo-controlled phase 3 study. First Author: of advanced esophageal squamous cell carcinoma. First Author: Feng Wang,
Yelena Yuriy Janjigian, Memorial Sloan Kettering Cancer Center, New York, NY Department of Oncology, The First Affiliated Hospital of Zhengzhou Uni-
versity, Zhengzhou, China
Background: Combination therapy with the anti-HER2 antibody trastuzu-
mab with fluoropyrimidine and platinum is the current standard for patients Background: Approximately 40% of patients (pts) with esophageal cancer are
with HER2+ mG/GEJc. We hypothesize that combination anti–PD-1 and diagnosed with advanced unresectable or metastatic disease; the 5-year
anti-HER2 therapy will result in T-cell activation, augment ADCC, and survival rate for advanced disease is 5%. No standard therapy is available in
potentiate antitumor immune response in HER2+ patients. This phase 2 China for Advanced Esophageal Squamous Cell Carcinoma(ESCC) patients
study in HER2+ mG/GEJc demonstrated the safety and preliminary efficacy progressed after first-line chemotherapy.Inhibition of programmed cell death
of trastuzumab/pembrolizumab/chemotherapy; the overall response rate was protein-1 (PD-1) has demonstrated promising antitumor activity and man-
87%, and the disease control rate was 100% Janjigian YY, ASCO GI 2019). ageable safety in pts with advanced unresectable or metastatic ESCC. SHR-
KEYNOTE 811, a global, multicenter, randomized, placebo-controlled, 1210, a humanized IgG4 monoclonal antibody, has high affinity and speci-
phase 3 study, is underway. Methods: Key eligibility criteria are age $18 ficity for PD-1 molecule. SHR-1210 was generally well tolerated and had
years; previously untreated unresectable or metastatic HER2+ (centrally preliminary antitumor effects in pts with solid tumors, including ESCC.
confirmed IHC 3+ or IHC 2+/ISH .2.0) G/GEJ adenocarcinoma; life Nimotuzumab, a humanized anti-epidermal growth factor receptor monoclonal
expectancy .6 months with RECIST v1.1 measurable disease; adequate antibody h-R3, has been shown to be effective and safe in the treatment of
organ function and performance status. Patients will be randomly assigned head and neck cancer,non-small cell lung cancer (NSCLC) and esophageal
1:1 to receive chemotherapy with pembrolizumab 200 mg IV flat dose or Cancer in several phase II studies.The purpose of this study is to observe and
placebo with trastuzumab 6 mg/kg (after 8 mg/kg load) Q3W up to 2 years or evaluate the efficacy and safety of anti-PD-1 antibody SHR-1210 combined
until intolerable toxicity or disease progression. Investigator choice che- with nimotuzumab as second-line therapy in patients with advanced ESCC.
motherapy will include day 1 cisplatin 80 mg/m2 IV and /5-fluorouracil Methods: Patients, age 18-75, with measurable tumor lesion, failed in or
800 mg/m2/day IV (days 1-5) or oxaliplatin 130 mg/m2 IV and capecitabine progression after 1st line chemotherapy, were enrolled in this study.Patients
1000 mg/m2 BID days 1-14 (Q3W). Primary end points are progression-free received SHR-1210 200 mg once every 2 weeks (Q2W) combined nimotu-
survival and overall survival. Secondary end points are objective response zumab 200 mg weekly until disease progression, death or unacceptable
rate, duration of response, and safety and tolerability. Adverse events are toxicity.Assessments included response by RECIST v1.1 every 6 wks and
graded per NCI CTCAE v4.0 and will be monitored for 30 or 90 days after safety (physical examination, vital signs, ECOG PS, laboratory tests).The
treatment. Patients will be followed up for survival. Planned enrollment is primary endpoint is the objective response rate (ORR),and the secondary end
approximately 692 patients. Clinical trial information: NCT03615326. points include the diseases control rate (DCR),duration of response
(DOR),progression-free survival (PFS),and overall survival(OS). Additionally,
we try to identify biomarker to predict efficacy of SHR-1210 and Nimotuzumab
with target capture sequencing and gene expression profile as exploratory
endpoints. Clinical trial information: NCT03766178.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
258s Gastrointestinal (Noncolorectal) Cancer

TPS4148 Poster Session (Board #249b), Mon, 8:00 AM-11:00 AM TPS4149 Poster Session (Board #250a), Mon, 8:00 AM-11:00 AM
RAP: A phase II trial with ramucirumab, avelumab, and paclitaxel as second line A phase II study of TAS-102 in combination with ramucirumab in advanced,
treatment in gastro-esophageal adenocarcinoma of the arbeitsgemeinschaft refractory gastric or gastroesophageal junction (GEJ) adenocarcinoma. First
internistische onkologie (AIO). First Author: Anica Högner, Charité–University Author: Rutika Mehta, Roswell Park Cancer Institute, Buffalo, NY
Medicine Berlin, Department of Haematology, Oncology and Tumor-
Background: Patients with advanced gastric cancer experience a 5 year
immunology, Berlin, Germany
survival rate ,10% even with multimodality therapy representing a clear
Background: Combination of ramucirumab and paclitaxel resembles the unmet need for improved treatment. In advanced gastric and gastroesophageal
standard treatment option in second line therapy with improvement of re- junction (GEJ) adenocarcinoma, ramucirumab (a monoclonal antibody against
sponse rate and overall survival (REGARD, RAINBOW). Response rates to PD- VEGFR2) has demonstrated clinical activity and has been approved as second
1/L1 blockade in gastro-esophageal cancer patients rank within 10–20%, line therapy in combination with paclitaxel with a response rate of 28% and
whereby PD-1/L1 blockade is reported to impressively extend survival rates absolute overall survival benefit of 8 weeks when compared to placebo.
in responders. Trials investigating either the synergistic effect of anti- However, there are many patients that cannot tolerate paclitaxel due to prior
angiogenesis and anti-PD-L1 or chemotherapy combined with anti-PD-L1 exposure to oxaliplatin causing neuropathy. Therefore, novel combinations
are promising. Based on these data we hypothesize benefit from combining with ramucirumab, is highly desirable. TAS-102 is an oral cytotoxic agent with
immunotherapy by checkpoint inhibition with VEGF-directed treatment and two active components; trifluridine (TFD) which inhibits tumor cell growth by
chemotherapy induced increase of immunogenicity of tumor cells. This study being incorporated into DNA during DNA synthesis and tipiracil (TPI) which
investigates the incorporation of PD-L1 blockade by avelumab in the second inhibits the metabolism of TFD, thereby prolonging its ability to exert effect.
line setting by combination with the actual best second-line chemotherapy TPI also inhibits platelet derived endothelial cell growth factor which plays a
regimen in metastatic gastric cancer patients (paclitaxel+ramucirumab). key role with VEGF in tumor angiogenesis. The combination of a cytotoxic agent
Methods: The RAP trial (AIO-STO-0218, registered at ClinicalTrials.gov) is a with an antiangiogenic agent has demonstrated a significant anticancer ac-
single arm multicenter phase II trial. A total of 59 patients with metastatic or tivity in multiple cancers. In a recent Phase III study, TAS-102 significantly
locally advanced gastric or gastro-esophageal junction adenocarcinoma, ECOG prolonged overall survival as compared to best supportive care in patients with
0–1, who progressed after having received first-line therapy with platinum and GEJ and gastric cancers that had received at least 2 prior lines of treatment. We
fluoropyrimidine doublet with or without anthracycline, docetaxel or trastu- hypothesize that a combination of TAS-102 and ramucirumab might increase
zumab within the last six months will receive avelumab and ramucirumab on efficacy without causing unmanageable toxicity. Methods: This is a single
day 1, 15 and paclitaxel on day 1, 8 and 15 of a 28-day cycle until disease institutional phase II single arm two-stage design trial using the combination of
progression (RECIST v1.1), intolerable toxicity, withdrawal of consent or at a TAS-102 and ramucirumab in advanced, refractory gastric or GEJ adeno-
maximum treatment of 1 year. The primary endpoint is the overall survival rate carcinoma. Eligible patients include those with histologically confirmed gastric
(OSR) at 6 months. Sample size calculation is based on a Simon 2-stage or GEJ adenocarcinoma that have received at least 1 prior line of treatment with
design with a one-sided alpha error of 10% and a power of 80%, an expected performance status 0 or 1 and preserved organ function. Ramucirumab will be
OSR at 6 months of $ 65% and a 0 hypothesis # 50%. Secondary endpoints administered 8mg/kg every 2 weeks and TAS-102 at doses of 35 mg/m2 twice
include OS, OSR at 12 months, PFS, safety and tolerability, duration of re- daily. Each cycle length will be 28 days. The primary endpoint is 6 month OS
sponse. Ethics commission approved the study protocol in January 2019. and secondary endpoints are safety, objective response rate and PFS. Fifteen
Updated patient accrual will be presented. Clinical trial information: AIO-STO- patients will be enrolled in the first stage. If $ 7 of the 15 are alive at 6 months,
0218. an additional 10 patients will be enrolled in the second phase. Enrollment is
currently ongoing. Clinical trial information: NCT03686488.

TPS4150 Poster Session (Board #250b), Mon, 8:00 AM-11:00 AM TPS4151 Poster Session (Board #251a), Mon, 8:00 AM-11:00 AM
Ramucirumab and irinotecan in patients with previously treated gastro- Assessment of ramucirumab plus paclitaxel as switch maintenance versus
esophageal adenocarcinoma. First Author: Haeseong Park, Washington continuation of first-line chemotherapy in patients (pts) with advanced
University School of Medicine, St. Louis, MO HER2-negative gastric or gastroesophageal junction cancers: The ARMANI
phase III trial. First Author: Federica Morano, Medical Oncology Department,
Background: Ramucirumab is used for treatment of metastatic gastro-
Fondazione IRCCS Istituto Nazionale dei Tumori, Milan, Italy
esophageal adenocarcinoma after disease progression on first-line chemo-
therapy. Superior survival outcome is expected when combined with Background: Platinum/fluoropyrimidine regimens are the backbone of first-
paclitaxel. However, many patients suffer from neuropathy after oxaliplatin- line therapy for advanced gastric cancer (AGC). The optimal duration of first-
containing first-line chemotherapy and are unable to tolerate paclitaxel. line therapy is still unknown and its continuation until disease progression
Irinotecan has shown survival benefit as a single agent or in combination with represents the standard. However this strategy is often associated with cu-
other agents, but has not been used in combination with ramucirumab for mulative toxicity and rapid development of drug resistance. Moreover, only
treatment with gastroesophageal cancer. We hypothesize that this combi- 40% of AGC pts are eligible for second-line treatment. This study aims at
nation regimen of irinotecan plus ramucirumab administered as second-line assessing whether switch maintenance to ramucirumab plus paclitaxel will
treatment will be well-tolerated with improved outcomes similar to paclitaxel extend the progression-free survival (PFS) of subjects with HER-2 negative
plus ramucirumab in patients with advanced gastroesophageal cancer. AGC who have not progressed after a first-line with a platinum/fluoropyrimidine
Circulating levels of angiogenic factors are correlatives of particular interest regimen. The hypothesis is that the early administration of an active, non-cross
in this study. Methods: This is a multi-institutional, single-arm phase II resistant regimen may delay disease progression and, consequently, improve
clinical trial of ramucirumab and irinotecan. Primary objective of the study is pts’ quality of life. This strategy may also rescue all those subjects that become
to determine the progression-free survival in patients treated with this ineligible for a second-line therapy due to the rapid clinical deterioration.
combination after disease progression on first-line chemotherapy. Sec- Methods: This is a randomized, open-label, multicenter, phase III trial. Eli-
ondary objectives are to determine other indices of efficacy including overall gibility criteria are: unresectable/metastatic HER-2 negative AGC or gastro-
survival, time to progression, objective response rate, and clinical benefit esophageal junction (GEJ) cancer; ECOG PS 0-1; measurable and/or evaluable
rate; and to evaluate toxicity and tolerability. Patients with confirmed di- disease by RECIST v1.1; no progression after 3 months of therapy with either
agnosis of gastroesophageal adenocarcinoma with measurable disease are FOLFOX4, mFOLFOX6 or XELOX . The primary endpoint is to compare PFS of
included. Patients are required of have disease progression during or within pts in ARM A (continuation of the same first-line therapy with oxaliplatin/
4 months of first line chemotherapy. Key exclusion criteria include squa- fluoropyrimidine) versus ARM B (switch maintenance to ramucirumab and
mous histology; prior irinotecan or ramucirumab use; active brain metas- placlitaxel). Secondary endpoints are: overall survival, time-to-treatment fail-
tases; or other contraindications to ramucirumab including recent history of ure, overall response rate, duration of response, percentage of pts receiving a
gastrointestinal bleeding or perforation, thromboembolic event, and un- second-line therapy per treatment arm, safety and quality of life. Exploratory
controlled hypertension. Patients receive ramucirumab 8mg/kg with irino- analyses to identify primary resistance and prognosis biomarkers are planned,
tecan 180mg/m2 IV every 14 days. We plan to enroll 40 patients which will including Next-Generation Sequencing (NGS) on archival tumor tissues. The
provide 85% power at a 0.05 significance level to detect a median pro- ARMANI study is sponsored by the Fondazione IRCCS Istituto Nazionale dei
gression free survival time of 4 months compared to historic control of Tumori and it is ongoing at 29 Italian centers with a planned population of 280
2.5 months. 25% of patient accrual is complete as of February 2019. pts. Clinical trial information: NCT02934464.
Clinical trial information: NCT03141034.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Noncolorectal) Cancer 259s

TPS4152 Poster Session (Board #251b), Mon, 8:00 AM-11:00 AM TPS4153 Poster Session (Board #252a), Mon, 8:00 AM-11:00 AM
Lenvatinib (len) plus pembrolizumab (pembro) for the first-line treatment of A multicenter phase II trial of rucaparib in combination with nivolumab as
patients (pts) with advanced hepatocellular carcinoma (HCC): Phase 3 maintenance therapy for patients with advanced biliary tract cancer. First
LEAP-002 study. First Author: Josep M Llovet, Icahn School of Medicine Author: Vaibhav Sahai, University of Michigan, Ann Arbor, MI
at Mount Sinai, New York, NY
Background: Patients (pts) with advanced biliary tract cancers (BTC) have a
Background: Len, an inhibitor of VEGF receptors 1-3, FGF receptors 1-4, poor prognosis with a median overall survival (OS) less than 12 months. Using
PDGF receptor a, RET, and KIT, is approved for first-line treatment of whole exome NGS, 26 (49%) pts in a 53 pt cohort had either DNA damage
unresectable HCC (uHCC) based on the open-label phase 3 REFLECT study in repair (DDR) pathway mutations (somatic and/or germline, n = 18), or iso-
which len showed noninferior overall survival (OS) and significantly improved citrate dehydrogenase 1 (IDH1) mutations (n = 8), and may have potentially
objective response rate (ORR), progression-free survival (PFS), and time-to- benefited from PARP inhibition. Further, disruption of the mutated DDR
progression (TTP) vs sorafenib. In the phase 2 KEYNOTE-224 study of pembro pathways with a PARP inhibitor may result in increased mutational burden and
(a PD-1 inhibitor) as second-line treatment of advanced HCC, pembro showed neoantigens leading to immunogenicity, thus providing the rationale for
meaningful clinical efficacy in pts previously treated with sorafenib, with combination with a PD-1 antibody. This phase 2 trial is designed to investigate
median PFS 4.9 mo, median OS 12.9 mo, and a manageable safety profile. In the role of a PARP inhibitor in combination with a PD-1 antibody in pts with
results from the phase 1b KEYNOTE-524 trial, len+pembro was well-tolerated, advanced BTC. Methods: Key eligibility criteria include histologically con-
with promising antitumor activity in pts with uHCC. LEAP-002 is a phase 3 firmed advanced, unresectable biliary adenocarcinoma (intra- or extra-hepatic,
study to evaluate the safety and efficacy of len+pembro vs len+placebo as first- and gallbladder) without progression after 4-6 months of 1st line platinum-
line therapy for advanced HCC. Methods: Eligible pts are $18 y and have HCC based systemic chemotherapy, measurable disease per RECIST v1.1, ECOG
confirmed by radiology, histology, or cytology; ECOG PS 0/1; BCLC stage C or PS 0-1, Child-Pugh A or B7, and absence of autoimmune disease or chronic
stage B disease not amenable to locoregional therapy or curative treatment steroid use. Primary objective is to evaluate progression-free survival (PFS) rate
approach; CP class A liver score within 7 days before study; and $1 mea- at 4 months. Secondary objectives include evaluation of objective response
surable lesion by RECIST v1.1. Pts with past or ongoing HCV infection and rate per immune related (ir)RECIST criteria, median PFS and OS, and safety in
those with controlled HBV are eligible. 750 pts will be randomized 1:1 to this patient population. Exploratory objectives include identification of pre-
receive len 12 mg (body weight [BW] $60 kg) or 8 mg (BW ,60 kg) orally once dictive biomarkers of response and mechanisms of resistance through serial
daily plus pembro 200 mg or placebo IV Q3W. Pembro and len will be ad- biopsies and blood collection (pre, on and post therapy), including sequential
ministered until disease progression or unacceptable toxicity, with a maximum whole exome/transcriptomic analysis with immune cell subset analysis.
35 cycles for pembro. Stratification will be by geographic region (Asia vs Japan Treatment includes rucaparib 600 mg PO BID on days 1-28 with nivolumab
and Western regions); macroscopic portal vein invasion or extrahepatic spread 240 mg on days 1, 15 Q4 weeks. In absence of disease progression, pts may
or both (yes or no); alpha fetoprotein #400 ng/mL vs .400 ng/mL; and ECOG continue therapy up to 2 years. Accrual goal is 32 evaluable pts. Using a null
PS 0/1. Primary end points are PFS per RECIST v1.1 by blinded independent hypothesis value of a 63% PFS rate at 4 months, and an 85% alternative
central review (BICR) and OS. Secondary end points are ORR, duration of hypothesis, this ongoing study has 80% power, with a one-sided alpha of 0.05
response, disease control rate, and TTP per RECIST v1.1 by BICR, efficacy per to identify treatment efficacy in the study arm. Clinical trial information:
modified RECIST, pharmacokinetics, and safety. Imaging assessments will be NCT03639935.
performed Q9W on study. AEs will be graded per CTCAE v4.0 and monitored up
to 90 days after last dose. Clinical trial information: NCT03713593.

TPS4154 Poster Session (Board #252b), Mon, 8:00 AM-11:00 AM TPS4155 Poster Session (Board #253a), Mon, 8:00 AM-11:00 AM
A multi-center phase Ib/II study of nal-irinotecan, 5-fluouracil and leuco- Infigratinib versus gemcitabine plus cisplatin multicenter, open-label, ran-
vorin in combination with nivolumab as second-line therapy for patients with domized, phase 3 study in patients with advanced cholangiocarcinoma with
advanced unresectable biliary tract cancer. First Author: Vaibhav Sahai, FGFR2 gene fusions/translocations: The PROOF trial. First Author: Milind M.
University of Michigan, Ann Arbor, MI Javle, MD Anderson Cancer Center, Houston, TX
Background: Patients (pts) with advanced biliary tract cancers (BTC) have poor Background: Cholangiocarcinoma is the most common biliary tract malig-
prognosis despite systemic chemotherapy, and treatment beyond first-line nancy with approximately 5,000–10,000 new cases annually in the USA.
platinum doublet remains investigational. The immunomodulatory properties The fibroblast growth factor receptor (FGFR) family plays an important role
of conventional cytotoxic therapy, particularly in regard to the upregulation of in cholangiocarcinoma, with FGFR2 gene fusions detected in about 15%
PD-L1 expression rendering tumor cells more sensitive to T cell-mediated lysis of patients with cholangiocarcinoma. Infigratinib is an ATP-competitive,
and neoantigen production, rapid emergence of chemotherapy resistance, and FGFR1–3-selective oral tyrosine kinase inhibitor. First-line treatment with
known modest efficacy of single agent PD-1 antibody in BTC provide a ra- chemotherapy offers only modest benefit and more effective treatment
tionale for combining chemotherapy and immunotherapy. This multi-center, options are needed. Based on preliminary response data of infigratinib in
phase Ib/II, single-arm study is designed to investigate the role of nal- relapsed/refractory cholangiocarcinoma with FGFR2 fusions/translocations
irinotecan, 5-FU and leucovorin in combination with nivolumab as second- (Phase 2 Study CBJG398X2204), the PROOF trial is evaluating infigratinib
line therapy in pts with advanced BTC. Methods: Key eligibility criteria include versus gemcitabine + cisplatin in front-line patients with advanced chol-
histologically confirmed advanced, unresectable biliary carcinoma (intra- or angiocarcinoma with FGFR2 gene fusions/translocations. Methods: Patients
extra-hepatic and gallbladder) with progression or intolerance of first-line with advanced/metastatic or inoperable cholangiocarcinoma are randomized
systemic therapy (excluding irinotecan and PD-1/PD-L1 antibody), measur- 1:1 to oral infigratinib once daily for 21 days of a 28-day treatment cycle
able disease per RECIST v1.1, ECOG PS 0-1, Child Pugh A or B7, and absence versus IV gemcitabine (1000 mg/m2) + cisplatin (25 mg/m2) on days 1 and 8
of autoimmune disease or chronic steroid use. Primary objective of the phase Ib of a 21-day cycle. Treatment will continue until confirmed progressive
portion is to determine the recommended phase 2 dose, and of the phase II disease by central review, intolerance, withdrawal of informed consent, or
portion is to evaluate the median progression-free survival. Secondary ob- death. After 8 cycles of gemcitabine + cisplatin, patients can continue
jectives include evaluation of objective response rate per immune related (ir) treatment if the investigator considers that they are deriving continued
RECIST, median OS and safety in this patient population. Exploratory ob- benefit. Patients on the gemcitabine + cisplatin arm who progress can cross-
jectives include identification of biomarker predictors of response and over to infigratinib. The primary endpoint is progression-free survival (PFS,
mechanisms of resistance through serial biopsies and blood collection (pre, on per RECIST v1.1 central review). Secondary endpoints include overall survival,
and post therapy), including sequential whole exome/transcriptomic analysis PFS (investigator determined), overall response rate, disease control rate,
and immune cell subset analysis (tissue and blood). Therapy includes nal- duration of response, and safety. Quality of life, PK and exploratory genetic
irinotecan 70 mg/m2, leucovorin 200 (dose level -1) or 400 mg/m2 (dose level alterations/biomarkers will also be measured. Current status: The study was
0), 5-fluouracil 2400 mg/m2 IV over 46 hours, and nivolumab 240 mg on day 1 initiated in February 2019 with planned enrollment of 350 patients with
every 2 weeks for 6 months. In the absence of disease progression, pts may confirmed FGFR2 gene fusions/translocations. Clinical trial information:
continue therapy for up to 2 years. Accrual goal is 30 evaluable pts. Using a null NCT03773302.
hypothesis value of median PFS of 2.9 months, and an alternative hypothesis
of 5.0 months, this ongoing study has . 80% power, with a two-sided alpha of
0.05 to identify treatment efficacy of study arm. Clinical trial information:
NCT03785873.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
260s Gastrointestinal (Noncolorectal) Cancer

TPS4156 Poster Session (Board #253b), Mon, 8:00 AM-11:00 AM TPS4157 Poster Session (Board #254a), Mon, 8:00 AM-11:00 AM
NUC-1031 in combination with cisplatin for first-line treatment of advanced Phase 3 (COSMIC-312) study of cabozantinib (C) in combination with
biliary tract cancer. First Author: Jennifer J. Knox, Princess Margaret Cancer atezolizumab (A) versus sorafenib (S) in patients (pts) with advanced
Centre, Toronto, ON, Canada hepatocellular carcinoma (aHCC) who have not received previous systemic
anticancer therapy. First Author: Robin Kate Kelley, UCSF Helen Diller
Background: Cisplatin and gemcitabine (CisGem) is the global standard of
Family Comprehensive Cancer Center, San Francisco, CA
care for 1st-line treatment of patients (pts) with advanced biliary tract cancer
(BTC). No agents have regulatory approval for this disease. CisGem achieves Background: C inhibits tyrosine kinases involved in tumor growth, angio-
an objective response rate (ORR) of 26% and median overall survival (OS) of genesis, and immune regulation, including MET, VEGFR, and TAM kinases
11.7 months (ABC-02). Key cancer resistance mechanisms limit gemci- (Tyro3, AXL, MER). C is approved for treatment of aHCC after prior S based
tabine efficacy. NUC-1031, a phosphoramidate transformation of gemci- on improved overall survival (OS) vs placebo in the phase 3 CELESTIAL trial
tabine, is designed to overcome resistance mechanisms associated with poor (Abou-Alfa NEJM 2018). Standard of care for first-line treatment of aHCC is
gemcitabine response. Promising signs of efficacy have been observed with tyrosine kinase inhibition with S or lenvatinib, and phase 3 trials of immune
single agent in a phase I study in solid tumors (Blagden et al 2018) and in the checkpoint inhibitors (ICIs) in first- and second- line aHCC are ongoing. C
phase Ib ABC-08 study of NUC-1031 + cisplatin 25 mg /m2 d1, d8 q 21 days may promote an immune-permissive tumor environment, which could en-
for the 1st-line treatment of advanced BTC. 14 pts have been enrolled across hance response to ICIs. C is being evaluated in combination with the anti-PD-
2 cohorts (NUC-1031: 625 mg/m2 and 725 mg/m2). In 11 pts evaluable for L1 antibody A in multiple tumor types including HCC in a phase 1 study; and
response ORR was 64% (1 CR, 6 PRs) and DCR was 73%. PFS/OS data is dose, preliminary clinical activity, and safety have been established in aRCC
maturing. The combination was very well-tolerated with no unexpected (Agarwal Ann Oncol 2018). A in combination with bevacizumab, an anti-
adverse events or dose-limiting toxicities. The RP2D in combination with VEGF antibody, has shown preliminary clinical activity in first-line aHCC
cisplatin is 725 mg/ m2. Safety, coupled with encouraging efficacy signal (Pishvaian Ann Oncol 2018). Here, we present the study design of a phase 3
has led to initiation of a global Phase III development program. Methods: A trial of C+A vs S in pts with aHCC who have not received prior systemic
Phase III, open-label, randomized head-to-head study of NUC-1031 + therapy. Methods: This international, randomized, open-label phase 3 trial
cisplatin versus CisGem for the 1st-line treatment of advanced BTC will (NCT03755791) is evaluating the efficacy and safety of C+A vs S as first-line
include pts $18 years with histologically- or cytologically-proven BTC (in- treatment for aHCC. Eligibility criteria include age $18 years, BCLC stage B
cluding cholangiocarcinoma, gallbladder, or ampullary cancer), that is not or C, Child-Pugh A, ECOG PS 0 or 1, and measurable disease per RECIST
resectable and who have had no prior systemic chemotherapy for locally 1.1. Patients are randomized 6:3:1 to an experimental arm of C (40 mg qd) +
advanced/metastatic disease. A total of 828 pts will be randomized (1:1) to A (1200 mg infusion q3w), a control arm of S (400 mg bid), and an ex-
either 725 mg/m2 NUC-1031 + 25 mg/m2 cisplatin or 1000 mg/m2 gem- ploratory arm of C monotherapy (60 mg qd). 640 pts are planned at ~200
citabine + 25 mg/m2 cisplatin, administered on Days 1 and 8 of a 21-day sites globally. Randomization is stratified by disease etiology (HBV [with or
cycle, respectively. Primary objectives are OS and ORR. Secondary objec- without HCV], HCV [without HBV], or other), region (Asia, other), and the
tives include further measurements of efficacy, safety, pharmacokinetics, presence of extrahepatic disease and/or macrovascular invasion (yes, no). OS
and patient-reported quality of life. The study will be conducted at ap- and progression-free survival are coprimary endpoints and objective re-
proximately 120 sites across North America, Europe and Asia Pacific sponse rate is a secondary endpoint. Additional endpoints include safety,
countries. Clinical trial information: NCT02351765. pharmacokinetics, and correlation of biomarker analyses with clinical out-
comes. Enrollment in COSMIC-312 is ongoing. Clinical trial information:
NCT03755791.

TPS4158 Poster Session (Board #254b), Mon, 8:00 AM-11:00 AM TPS4159 Poster Session (Board #255a), Mon, 8:00 AM-11:00 AM
NET-02: A multi-center, randomised, phase II trial of liposomal irinotecan Phase 2 trial of Lu-177-DOTATATE in inoperable pheochromocytoma/
(nal-IRI) and 5-fluorouracil (5-FU)/folinic acid or docetaxel as second-line paraganglioma. First Author: Frank Lin, National Institutes of Health,
therapy in patients (pts) with progressive poorly differentiated extra- National Cancer Institute, Bethesda, MD
pulmonary neuroendocrine carcinoma (PD-EP-NEC). First Author:
Background: Pheochromocytoma/paraganglioma (PHEO/PGL) is a rare
Mairead Geraldine McNamara, Institute of Cancer Sciences, University of
malignancy that arises from chromaffin cells of typically the adrenal medulla
Manchester, Medical Oncology Department, The Christie NHS Foundation
but can also be of extra-adrenal origin. These tumors produce excessive
Trust, Manchester, United Kingdom
catecholamines such as epinephrine and norepinephrine which causes
Background: The prognosis for pts with PD-EP-NEC is poor. First-line labile hypertension, tachycardia, and flushing. Lu-177-DOTATATE (Lu-
treatment for advanced disease is etoposide/platinum-based chemother- 177-dodecanetetraacetic acid-tyrosine-3-octreotate) is a radiolabeled so-
apy, analogous to that of high grade lung NEC, with no standard second-line matostatin analog that is FDA approved for somatostatin receptor-positive
treatment, and is an area of unmet need. Methods: This is a multi-centre, neuroendocrine tumors. It is being being investigated in PHEO/PGL, which
randomised, phase II trial of nal-IRI; 80mg/m2 intravenously (IV) over 90 overexpress somatostatin receptors. Amino acid solutions containing lysine/
mins, prior to 5-FU; 2400 mg/m2 infusion over 46 hrs and folinic acid, arginine (L/A) are routinely co-administered with Lu-177-DOTATATE for
Q14 days, or docetaxel; 75mg/m2 IV over 60 mins, Q21 days, as second-line renal radioprotection, although solutions containing other amino acids are
therapy in pts with progressive PD-EP-NEC (Ki-67 . 20%), with the overall also used. Methods: This is a prospective, single center, open label Phase 2
aim of selecting a treatment for continuation to a phase III trial. The standard study evaluating the efficacy of Lu-177-DOTATATE in PHEO/PGL. Ninety
arm is that used in high-grade lung NEC, of which docetaxel is a second-line patients will be enrolled, divided into two cohorts of 45 patients each (SDHx
therapy option (NCCN guidelines) and combination regimens such as mutation vs. apparent sporadic). Lu-177-DOTATATE is given at a fixed dose
Irinotecan/5-FU are a second-line therapy option currently used without trial of 200 mCi with a co-administration of L/A amino acid solution q8 weeks for
evidence for this subset of pts. Pts must have had prior treatment with first- 4 cycles. The primary endpoint is the progression-free-survival (PFS) rate at
line platinum-based chemotherapy, have documented disease progression 6 months. Secondary endpoints include response rate, overall survival, time
and have an ECOG performance status of #2. This study plans to recruit 102 to progression, quality of life measures, and examination of potential bio-
pts from 16 UK centres (over 37 mths). Primary endpoint is 6-mth markers such as biochemical profiles, Ga-68-DOTATATE PET, and F-18-
progression-free survival (PFS) rate; trial is designed to have an 80% chance FDG PET scans. Eligibility criteria include inoperable disease (including
of demonstrating that the one-sided 95% confidence interval of the 6 mth non-metastatic), histological confirmation of PHEO/PGL, evidence of dis-
PFS rate excludes 15%, if the true rate is at least 30%, where 30% is the ease progression by RECIST 1.1, ECOG performance status of 1 or better,
required level of efficacy, and a rate of , 15% would give grounds for re- and a positive Ga-68-DOTATATE PET scan. Exclusion criteria include prior
jection. If both treatment arms exceed the required level of efficacy to treatment with systemic radionuclide therapy such as I-131-MIBG, brain
warrant further evaluation in a phase III trial, treatment with the higher PFS parenchymal metastases, and standard organ dysfunction limitations. In-
rate at 6 mths will be selected. Secondary endpoints include overall survival, terim analysis using a Simon two-stage optimal design will be performed
objective response rate, toxicity, quality of life, serum neuron-specific separately for each cohort after enrollment of 18 patients. First patient
enolase. Exploratory endpoints include quantification of circulating tu- accrual to this ongoing study was in the Fall of 2017, and as of February
mour cells (CTCs), circulating tumour deoxyribonucleic acid (ctDNA) and 2019, fourteen patients have been accrued. Preliminary results will be
molecular profiling of CTCs, ctDNA and tumour tissue, and generation of reported at the completion of stage 1 for each cohort. Clinical trial in-
CTC-derived xenografts. This trial is open and has enrolled 6 pts at time of formation: NCT03206060.
submission. Clinical trial information: 10996604.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Noncolorectal) Cancer 261s

TPS4160 Poster Session (Board #255b), Mon, 8:00 AM-11:00 AM TPS4161 Poster Session (Board #256a), Mon, 8:00 AM-11:00 AM
A randomized noncomparative phase II study of maintenance therapy with A randomized phase II trial of niraparib plus either nivolumab or ipilimumab
multiepitope vaccine Tedopi (OSE2101) 6 nivolumab or FOLFIRI after in- in patients with advanced pancreatic cancer whose cancer has not pro-
duction chemotherapy (CT) with FOLFIRINOX in patients (Pts) with advanced gressed on platinum-based therapy. First Author: Kim Anna Reiss, University
pancreatic ductal adenocarcinoma (aPDAC) (TEDOPaM–PRODIGE 63 GER- of Pennsylvania Abramson Cancer Center, Philadelphia, PA
COR D17-01 study). First Author: Cindy Neuzillet, Medical Oncology De-
Background: The treatment paradigm for advanced pancreatic ductal ade-
partment, Curie Institute, Versailles Saint-Quentin University, Saint Cloud,
nocarcinoma (PDAC) typically involves ongoing chemotherapy until either
France
disease progression or clinical deterioration. A subset of patients with ad-
Background: FOLFIRINOX (5-fluorouracil [5FU], folinic acid [FA], irinotecan [Iri], vanced PDAC have exceptional responses to platinum-based chemotherapy.
and oxaliplatin [Ox]) is a standard 1st-line treatment in fit Pts with aPDAC. Anti-PD-1/ We hypothesized that durable platinum sensitivity in patients with advanced
PD-L1 as single agents have failed in PDAC so far and new combination immu- PDAC might be indicative of a DNA repair deficiency, and that these patients
notherapies are needed. Tedopi (OSE2101) is a multiple neoepitope vaccine re- may respond to a combination of niraparib, a PARP inhibitor, plus immune
stricted to HLA-A2 positive Pts, targeting 5 tumor-associated antigens (CEA, HER2, checkpoint blockade. Methods: We have enrolled 25 of 84 planned patients
MAGE2, MAGE3, TP53) that are frequently expressed in PDAC. This study aims to on study NCT 03404960. Eligibility criteria include inoperable PDAC and
assess the efficacy and safety of Tedopi alone and in combination with anti-PD-1
stability on platinum-based chemotherapy for $16 weeks without evidence
nivolumab, or FOLFIRI as maintenance therapy in Pts with aPDAC after FOLFIR-
of progressive disease. Patients who have progressed on platinum-based
INOX induction CT. Methods: TEDOPaM - PRODIGE 63 is a 3-arm, Fleming 2-stage,
open-label, randomized, non-comparative phase II study. 156 Pts with locally
treatment or who have received prior therapy with PARP inhibitors are ex-
advanced or metastatic, pathologically proven PDAC; ECOG performance status 0-1; cluded. Patients are randomized to receive oral niraparib 200mg PO daily
HLA-A2 genotype; controlled disease (objective response or stable disease) after 8 plus nivolumab 240mg IV every two weeks in continuous 28 day cycles or
cycles of (modified) FOLFIRINOX; and adequate organ functions, are randomized (1: oral niraparib 200mg PO daily plus ipilimumab 3mg/kg IV every three weeks
1:1, stratified on center, tumor stage, and best response to FOLFIRINOX) into 3 for four doses in continuous 21 day cycles. The primary endpoint is
arms: Clinical trial information: NCT03806309. In Arms B and C, FOLFIRI is progression-free survival at 6 months. Secondary endpoints include re-
reintroduced at disease progression or unacceptable toxicity. Primary endpoint: sponse rate, duration of response and overall survival. Paired biopsies are
overall survival rate at M12. Secondary endpoints: progression-free survival (CT-scan obtained, as well as serial blood collections for circulating tumor cells
Q8W), duration of disease control, safety, objective response rate, RECIST v1.1/ (CTCs), circulating tumor DNA (ctDNA) and peripheral blood mononuclear
iRECIST comparison, HRQoL (EORTC QLQ-C30), Q-TWiST. An interim analysis is cells (PBMCs). Correlative assays will include germline whole exome se-
planned after inclusion of 20 Pts in each arm. Translational research will be per- quencing and analyses of serially collected PBMCs, CTCs and ctDNA to
formed on tumor tissue (initial FFPE biopsy and optional re-biopsy at inclusion): identify genomic and immunologic innate and adaptive resistance mech-
RNAseq (cancer and stroma), mutation burden, MMR status, immune infiltrates; anisms. Clinical trial information: NCT 03404960.
and in blood (before and on-treatment): cytokine panel, PBMC phenotyping,
vaccine-antigen specific T-cells, TCR repertoire, and extracellular vesicles, to ex-
plore biomarkers and pharmacodynamics effects of Tedopi 6 nivolumab. Enrollment
(12th Feb 2019): 1. ClinicalTrials Registration: NCT03806309.
IV; FA 400 mg/m2, Iri 180 mg/m2, 5FU bolus
Arm A (reference): FOLFIRI (n = 52) 400 mg/m2 + continuous 2400 mg/m2/46h
Arm B (experimental): Subcutaneous injection on D1 Q3W for 6 doses then
Tedopi (n = 52) Q8W until month 12 [M12] then Q12W up to M24
Arm C (experimental): Tedopi + nivolumab 360 mg IV on D1 Q3W for
Tedopi + nivolumab (n = 52) 6 infusions then 480 mg Q4W up to M24

TPS4162 Poster Session (Board #256b), Mon, 8:00 AM-11:00 AM TPS4163 Poster Session (Board #257a), Mon, 8:00 AM-11:00 AM
Improving cascade genetic testing for families with inherited pancreatic Phase II multi-institutional study of nivolumab (Nivo), cabiralizumab (Cabira),
cancer (PDAC) risk: The genetic education, risk assessment and testing and stereotactic body radiotherapy (SBRT) for locally advanced unresectable
(GENERATE) study. First Author: Matthew B. Yurgelun, Dana-Farber Cancer pancreatic cancer (LAUPC). First Author: Deirdre Jill Cohen, Laura and Isaac
Institute, Boston, MA Perlmutter Cancer Center, NYU Langone Health, New York, NY
Background: 4-10% of PDAC patients harbor pathogenic germline variants in Background: Treatment of LAUPC most commonly involves chemotherapy +/- RT.
cancer susceptibility genes, including APC, ATM, BRCA1, BRCA2, CDKN2A, Patients(pts) who can be downstaged and undergo R0 resection have sig-
EPCAM, MLH1, MSH2, MSH6, PALB2, PMS2, STK11, and TP53. For nificant improvement in overall survival, but conventional chemoRT
families with such pathogenic variants, the greatest potential impact of converts , 10% of patients with LAUPC. If the effects of RT can be aug-
germline testing is to identify relatives with the same pathogenic variant mented then higher R0 resection rates may be achieved and improve sur-
(cascade testing), thereby providing the opportunity for early detection and vival. In pre-clinical models, RT leads to increased expression of M-CSF from
cancer interception of PDAC and other associated malignancies. Numerous pancreatic tumor cells and marked immune suppression within the tumor
factors limit cascade testing in real-world practice, including family dynamics, microenvironment via expansion of tumor associated macrophages (TAMs).
widespread geographic distribution of relatives, access to genetic services, and Concurrent blockade of M-CSF with RT reduces TAM infiltration, prevents
misconceptions about the importance of germline testing, such that the the generation of tumor promoting T cell populations, and increases the
preventive benefits of cascade testing are often not fully realized. The primary therapeutic effect of RT. RT also induces up-regulation of PD-L1 in TAMs,
aim of this study is to analyze two alternative strategies for cascade testing in another mode of immune suppression that can account for RT resistance in
families with inherited PDAC susceptibility. Methods: 1000 individuals (from LAUPC. (Seifert et al. 2016). These data suggest the efficacy of RT in LAUPC
approximately 200 families) with a confirmed pathogenic germline variant in is limited by its promotion of innate and adaptive immune suppression.
any of the above genes in a 1st/2nd degree relative and a 1st/2nd degree relative CSF1R blockade with Cabira combined with PD-1 blockade with Nivo may
with PDAC will be remotely enrolled through the study website (www.gener- enhance the efficacy of SBRT by reprogramming the TAM compartment in
atestudy.org) and randomized between two different methods of cascade tumors, thereby preventing an immune suppressive phenotype and aug-
testing (individuals with prior genetic testing will be ineligible): Arm 1 will menting T-cell mediated anti-tumor response. Methods: Single arm phase II
undergo pre-test genetic education with a pre-recorded video and live in- study designed to evaluate safety, tolerability, and surgical resection rate in
teractive session with a genetic counselor via a web-based telemedicine LAUPC pts treated with concurrent Nivo, Cabira, and SBRT. Exploratory
platform (Doxy.me), followed by germline testing through Color Genomics; Arm endpoints include immune changes within blood and tissue following
2 will undergo germline testing through Color Genomics without dedicated pre- treatment and correlation with clinical endpoints. Key eligibility: completion
test genetic education. Color Genomics will disclose results to study personnel of 2- 6 months standard induction chemotherapy, normal organ and marrow
and directly to participants in both arms. Participants in both arms will have the function, pre- and on-treatment biopsy, and PS # 1. Following initial biopsy
option of pursuing additional telephone-based genetic counseling through and placement of fiducials, Cabira 4mg/kg and Nivo 240mg are given D1 of
Color Genomics. The primary outcome will be uptake of cascade testing. every 14 day cycle. SBRT 6.6 Gy x 5 consecutive fractions starts D8. After 2
Secondary outcomes will include participant self-reported genetic knowledge, cycles, repeat biopsy and imaging is performed. Treatment with Cabira and
cancer worry, distress, decisional preparedness, familial communication, and Nivo continues every 2 weeks and imaging is done every 8 weeks, at which
screening uptake, which will be measured via longitudinal surveys. Enrollment time pt is assessed for surgical resection. If pt is downstaged, treatment is
will begin February, 2019. Clinical trial information: NCT03762590. discontinued and pt proceeds to surgery. Preliminary 6 pt safety cohort is
monitored for unacceptable toxicities. If , 3 unacceptable toxicities in the
first 6 subjects enrolled, then plan for expansion phase with 14 more pts. As
of abstract submission, 3 pts have been enrolled. Clinical trial information:
NCT03599362.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
262s Gastrointestinal (Noncolorectal) Cancer

TPS4164 Poster Session (Board #257b), Mon, 8:00 AM-11:00 AM TPS4165 Poster Session (Board #258a), Mon, 8:00 AM-11:00 AM
Adaptive Dose Escalation Trial of Stereotactic Body Radiation Therapy A phase I/II study of GSK3145095 alone and in combination with anticancer
(SBRT) in combination with GC4419 in pancreatic cancer. First Author: agents including pembrolizumab in adults with selected solid tumors. First
Jon Holmlund, Ascenta Therapeutics, Carlsbad, CA Author: Deirdre Jill Cohen, Laura and Isaac Perlmutter Cancer Center, NYU
Langone Health, New York, NY
Background: Local progression causes up to 30% of deaths from pancreatic
cancer (PC) and is also a significant source of morbidity. Stereotactic body Background: The immunosuppressive myeloid infiltrate characteristic of the
radiotherapy (SBRT) offers the potential for improved therapeutic index over tumor microenvironment in pancreatic cancer represents a major therapeutic
standard fractionation, but current regimens of 5-7 Gy/fraction x 5 are con- barrier in this disease. Modulation of this infiltrate may increase sensitivity to
strained by nearby organ tolerance and offer only palliation without improv- immune checkpoint blockade in this and other tumors with a similar phe-
ing survival. Safe dose escalation is necessary to improve SBRT efficacy. notype. The receptor interacting protein 1 (RIP1) is a serine/threonine kinase

· ·
GC4419, a superoxide dismutase mimetic, selectively converts superoxide
(O2 -) to hydrogen peroxide (H2O2) and oxygen. O2 -initiates normal tissue
damage due to RT. GC4419 is in a Phase 3 trial (NCT03689712) to reduce
that becomes active upon homeostatic disruptions. Bound to RIP3 and mixed
lineage kinase domain-like protein (MLKL), RIP1 kinase activity drives nec-
roptosis. However, RIP1 also signals in response to inflammatory stimuli in-
RT-induced oral mucositis in head and neck cancer, based on positive results dependently of its association with RIP3. A correlation between increased
in a randomized Phase 2 trial for that indication (Anderson, ASCO 2018). RIP1 protein expression and a worse prognosis has been reported in a variety of
GC4419 improved the survival of mice receiving 8.5 Gy x 5 to the upper solid tumors. Furthermore, in an unbiased screen RIP1 was identified as a top

·
abdomen. Cancer cells are less tolerant to elevated H2O2, and more tolerant to
elevated O2 -, than normal cells, and GC4419 demonstrated mechanism-
dependent synergy with high dose-fraction RT in a human tumor xenograft with
gene contributing to resistance to immunotherapy (Manguso 2017). In murine
models, RIP1 kinase activity has been reported to drive pancreatic onco-
genesis. Inhibition of RIP1 in the pancreatic TME leads to the replacement of
inducible expression of catalase (Sishc, AACR 2018). Thus, adding GC4419 tumor-permissive myeloid infiltrates with innate cells promoting an anti-tumor
to SBRT may increase both the efficacy and the safety of the latter. response by the adaptive immune system (Seifert 2016; Wang 2018) and
Methods: 48 patients with localized, unresectable PC without frank duodenal synergized with anti-PD-1 treatment. These data suggest that the small
invasion, who have received 3+ cycles of induction chemotherapy, are to be molecule RIP1 inhibitor GSK3145095 may have therapeutic potential in
randomized 1:1 to placebo or GC4419, 90 mg IV, prior to each of 5 con- multiple tumor types. Methods: This is a four-part phase 1/2 study designed to
secutive daily (M-F) SBRT fractions. A phase I/II Late Onset Efficacy/ Toxicity evaluate the safety, PK, PD, and preliminary activity of GSK3145095 given
tradeoff (LO-ET) based adaptive design adaptive model drives SBRT dose orally to participants with selected advanced or recurrent solid tumors. Part 1
escalation in each arm based on a dual endpoint (Gr 3-4 GI toxicity or death will be conducted in approximately 30 adults with pancreatic cancer with
;stable disease or better) by 90 days post SBRT. The planned dose levels are escalating doses of GSK3145095. Part 2 will combine escalating doses of
10, 11 and 12Gy x 5 fractions (BED10=100,112.5 and 132Gy, re- GSK3145095 with 200 mg pembrolizumab and may be conducted in a
spectively) as an integrated boost to the gross tumor volume (GTV). Primary broader population of selected solid tumors. Part 3 represents a cohort ex-
endpoint: Maximum tolerated dose of SBRT with GC4419 or placebo. pansion of Part 2. Part 4 may investigate the combination of additional an-
Exploratory endpoints include change in tumor radiographic resectability, ticancer agent(s) with one or more doses of GSK3145095 identified as safe in
correlative studies (ctDNA, exosomal DNA, tumor exome/transcriptome Part 1. References: Manguso RT. Nature. 2017;547(7664):413-418. Seifert
sequencing, immune profiling). Supported by Galera Therapeutics, Inc. L. Nature. 2016;532(7598):245-249. Wang W. Cancer Cell 2018; 34: 757-
Clinical trial information: NCT03340974. 774. Clinical trial information: NCT03681951.

TPS4166 Poster Session (Board #258b), Mon, 8:00 AM-11:00 AM TPS4167 Poster Session (Board #259a), Mon, 8:00 AM-11:00 AM
PRIMUS-002: A multicentre, open-label, phase II study examining FOLFOX A randomized, phase II clinical trial of preoperative stereotactic body
and nab-paclitaxel (FA) and nab-paclitaxel and gemcitabine (AG) as neo- radiation therapy versus conventionally fractionated chemoradiation for
adjuvant therapy for (borderline) resectable pancreatic cancer (PC), focusing resectable, borderline-resectable, or locally advanced type a pancreatic
on biomarker and liquid biopsy development. First Author: Derek B. Grose, adenocarcinoma. First Author: William Adrian Hall, Medical College of
Beatson West of Scotland Cancer Centre, Glasgow, United Kingdom Wisconsin, Milwaukee, WI
Background: There is increasing evidence suggesting benefit from a neo- Background: There is growing consensus for the use of neoadjuvant therapy in
adjuvant approach to PC. However, the optimal regimen is unclear and will patients with potentially operable pancreatic adenocarcinoma (PC). However, there
likely require a precision medicine approach, where patient and tumor attri- is not consensus on the type and duration of chemotherapy or radiation therapy (RT)
butes define therapy. Platinum-containing regimens have shown survival dose. Stereotactic body radiation therapy (SBRT) has gained popularity despite the
benefit for PC, with occasional exceptional responders, but biomarkers (BM) of absence of prospective data for its use in the preoperative setting. Furthermore,
response are not well defined and treatment decisions are often based on SBRT preoperatively has not been standardized. At present, there exists no ran-
domized data comparing preoperative SBRT with conventionally fractionated
patient performance status (PS) and co-morbidity. Tumors with defects in
concurrent chemo-RT. We designed this trial to examine differences between pre-op
BRCA1/2and other Fanconi Anemia genes show defective DNA damage re-
RT dose and fractionation schedules. Methods: This study is a prospective, ran-
sponse (DDR), conferring potential selective sensitivity to DNA-damaging domized, two-arm, phase II clinical trial. Eligible patients must have cytologically
agents (e.g. platinum) and newer targeted agents. We have shown that confirmed PC and be deemed suitable for surgical resection with resectable, bor-
DDR deficiency (DDRd) is present in up to 20% of PC. This study aims to derline resectable, or locally advanced type A disease, based on cross-sectional
exploit DDRd as a therapeutic vulnerability, with integrated analysis to define imaging. Before randomization patients are stratified by clinical node positivity,
candidate BM for FA and AG response. Methods: PRIMUS-002 will enroll neoadjuvant chemotherapy, and stage of disease. Patients are then randomized to
patients registered on the Precision-Panc Master Protocol who are molecularly either 50.4 Gy over 28 fractions with concurrent weekly Gemcitabine vs SBRT to a
profiled using the Precision-Panc Clinical Cancer Genome including a novel total dose of 25-35 Gy over 5 fractions. The primary endpoint of the study is
DDRd assay, and the transcriptome with longitudinal sampling (pre-, during, pathologic node positivity. We hypothesize that patients treated with neoadjuvant
and post-treatment). Patients receive either FA (nab-paclitaxel 150mg/ chemotherapy followed by conventionally fractionated chemo-RT will have a lower
m2IV,oxaliplatin 85mg/m2, folinic acid 350mg flat dose, fluorouracil infusion rate of pathologic node positivity as compared to those patients treated with neo-
2400mg/m2continuous IV infusion), orAG (nab-paclitaxel 125mg/m2, gem- adjuvant chemotherapy followed by SBRT. Secondary endpoints include patient
citabine 1000 mg/m2) for 3 months,based on patient age and PS.Following reported quality of life, local recurrence, primary tumor pathologic response, margin
initial safety analysis, chemoradiation may be introduced. The primary end- status, surgical complications, MR based treatment response, and overall survival.
point is disease progression (DP) during neoadjuvant therapy. The study is We anticipate a node positivity rate of 37% when using preoperative chemotherapy
followed by SBRT. We hypothesize that treatment with chemotherapy followed by
designed to detect a 20% difference in DP between the BM+ve (10%) and
conventionally fractionated chemo-RT will reduce the rate of node positivity to 17%.
BM –ve (30%) in patients treated with FA (90% power, 5% 1-sided level of
Using a one-sided Type I error rate of 0.1, approximately 88 total patients (44 per
statistical significance)., Exploratory translational endpoints include surrogate arm) provide an 80% power to detect the hypothesized difference in pathologic node
therapeutic response assessment using CA19.9, PET-CT SUV, DWI-MRI and positivity between the two arms. To address patient dropout, an additional 14
ctDNA. Current Enrolment: 2 patients enrolled to date: 1 to receive FA and 1 to patients (about 15%) will be enrolled for a total target accrual of 102 patients. The
AG treatment. Clinical trial information: ISRCTN34129115. trial opened in November 2018 and 8 of the planned 102 patients have been
enrolled. Clinical trial information: NCT03704662.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gastrointestinal (Noncolorectal) Cancer 263s

TPS4168 Poster Session (Board #259b), Mon, 8:00 AM-11:00 AM

Niraparib in metastatic pancreatic cancer after previous chemotherapy (NIRA-
PANC): A phase 2 trial. First Author: Anup Kasi, University of Kansas Cancer
Center, Westwood, KS
Background: Attempts to improve therapy for patients with pancreatic ade-
nocarcinoma with traditional chemotherapy have largely failed to meaningfully
improve survival. Therefore, there is a critical need for identification of specific
molecular changes that define prognosis and potentially guide therapy de-
cisions. Defective DNA damage response pathways in pancreatic cancer
represent a targeted opportunity for treatment. PARP inhibitors exert activity in
tumor cells that may not be effectively able to repair initially single-stranded and
cumulatively double-stranded DNA breaks and can have a heightened sus-
ceptibility in tumor cells over normal tissue. This concept is referred to as
synthetic lethality. Niraparib is an orally available, potent, highly selective
PARP-1 and -2 inhibitor. We are studying the efficacy of Niraparib in pancreatic
cancer patients that harbor DNA repair defects. Methods: This study is funded
by a research grant from TESARO. Pre-screening of patients to find biomarker
positive patients is funded by KU Cancer Center. This is a phase II open label
single arm trial in metastatic pancreatic cancer patients with germline or so-
matic mutations, either already known, or tested after consent to pre-screening
tumor tissue analysis in BRCA1/2, PALB2, ATM, NBN, ATR, BRIP1, IDH1/2,
RAD51, RAD51B/C/D, RAD54L, CDK12, BARD1, FAM175A, BAP1, CHEK1/
2, GEN1, MRE11A, XRCC2, SHFM1, FANCD2, FANCA, FANCC, FANCG,
RPA1, ARID1A. Patients are being treated with Niraparib 300mg or 200mg by
mouth daily for 28 days (1 cycle = 28 days) (200mg dose is for participants
whose baseline weight is , 77 kg [169.756 lbs] or baseline platelet count is ,
150,000 mL). The primary objective is to assess antitumor efficacy of niraparib
using Objective Response Rate per RECIST 1.1. Secondary objectives include
PFS, OS, DCR, DOR, and safety. Eligible patients received . 1 line of therapy,
no prior PARP inhibitor(s), have measurable disease, and ECOG PS 0-1. Accrual
target enrollment of 18 patients over a period of 24 months with a study duration
of 30 months. Correlative studies include assessment of pharmacokinetics,
circulating tumor cells and storing samples for future research. The trial is
currently enrolling. Clinical trial information: NCT03553004.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
264s Genitourinary (Nonprostate) Cancer

4500 Oral Abstract Session, Mon, 8:00 AM-11:00 AM 4501 Oral Abstract Session, Mon, 8:00 AM-11:00 AM
Pembrolizumab (pembro) plus axitinib (axi) versus sunitinib as first-line therapy A pilot randomized study evaluating nivolumab (nivo) or nivo + bevacizumab
for metastatic renal cell carcinoma (mRCC): Outcomes in the combined IMDC (bev) or nivo + ipilimumab (ipi) in patients with metastatic renal cell carcinoma
intermediate/poor risk and sarcomatoid subgroups of the phase 3 KEYNOTE- (MRCC) eligible for cytoreductive nephrectomy, metastasectomy or post-
426 study. First Author: Brian I. Rini, Cleveland Clinic Taussig Cancer Institute, treatment biopsy (Bx). First Author: Jianjun Gao, The University of Texas
Cleveland, OH MD Anderson Cancer Center, Houston, TX
Background: In KEYNOTE-426, pembro + axi significantly improved OS (HR Background: Cytoreductive surgery (CS) including cytoreductive nephrectomy
0.53, P , .0001), PFS (HR 0.69, P = .0001), and ORR (59.3% vs 35.7%, and metastasectomy provides clinical benefits to patients with MRCC. However,
P , .0001) vs sunitinib and had manageable toxicity as first-line therapy for CS has not been evaluated in the setting of immune checkpoint (IC) therapy. We
mRCC (NCT02853331). The pembro + axi benefit was observed across all performed a pre-surgical/biopsy trial to evaluate biological and clinical activity of
IMDC risk groups and regardless of PD-L1 expression. We present data for the nivo +/- (bev or ipi) in patients with MRCC. Methods: In this open-label, ran-
combined intermediate/poor risk group and for patients (pts) with sarcomatoid domized trial (NCT02210117), patients with MRCC w/o prior IC therapy and
features. Methods: 861 eligible pts with clear-cell mRCC, no prior systemic anti-VEGF therapy were randomized 2:3:2 to receive nivo (3mg/kg q2wks x3),
therapy for mRCC, and KPS $70 were randomized 1:1 to pembro 200 mg IV nivo + bev (10mg/kg q2wks x3) or nivo + ipi (1mg/kg q3wks x2), followed by CS
Q3W for a maximum of 35 cycles plus axi 5 mg orally BID (N = 432) or sunitinib or Bx, and then nivo maintenance therapy up to 2 yrs. Clinical response per
50 mg orally QD (4-wk on/2-wk off) (N = 429). Primary endpoints were OS and RECIST criteria was assessed at $12 wks. Pre- and post-treatment blood and
PFS (RECIST v1.1 by blinded, independent central review [BICR]). ORR tumors were obtained for correlative studies. Results: All of 105 patients have
(RECIST v1.1 by BICR) was the key secondary endpoint. The intermediate/ been accrued and 104 are evaluable for response. For all patients, best overall
poor risk group was prespecified; the sarcomatoid group was exploratory. HRs response (BOR = complete response [CR] + partial response [PR]) including
and their 95% CIs were calculated with a Cox proportional hazards model. surgery effect was 55% nivo, 44% nivo + bev, 43% nivo + ipi. Median PFS
None of the analyses were multiplicity-controlled. Results: 592 (68.8%) of all (95% confidence interval) was 14.5 months (5.5, not reached [NR]) nivo, 7.6
randomized pts were of IMDC intermediate/poor risk — 294 in the pembro + (4.8, 8.9) months nivo + bev, 7.5 (2.0, 12.4) months nivo + ipi. Overall survival
axi arm, 298 in the sunitinib arm. Pembro + axi improved OS (HR 0.52, 95% (OS) at one year was 86% nivo, 73% nivo + bev, 83% nivo + ipi. Grade 3 or
CI 0.37-0.74; 12-mo rate 87.3% vs 71.3%), PFS (HR 0.67, 95% CI 0.53- higher toxicities related to therapy were 38% for nivo, 42% for nivo + bev
0.85; median 12.6 vs 8.2 mo), and ORR (55.8% [95% CI 49.9-61.5] vs (including 18% hypertension), and 47% for nivo + ipi. For patients with CS,
29.5% [24.4-35.1]) in pts with intermediate/poor risk; CR rates were 4.8% BOR including surgery effect was: 86% nivo, 88% nivo + bev, and 69% nivo +
(95% CI 2.6-7.9) vs 0.7% (0.1-2.4). Of the 578 pts with known status, 105 ipi. Median PFS was 17.3 months (6.1, NR) nivo, 7.6 (5.7, 10.3) months nivo +
(18.2%) had sarcomatoid features — 51 in the pembro + axi arm, 54 in the bev, 8.9 (2.9, 23.0) months nivo + ipi. OS at one year was 100% nivo, 94%
sunitinib arm. Pembro + axi improved OS (HR 0.58, 95% CI 0.21-1.59; 12- nivo + bev, 92% nivo + ipi. Median OS has not yet reached with a median follow-
mo rate 83.4% vs 79.5%), PFS (HR 0.54, 95% CI 0.29-1.00; median not up of 24.6 months. Immune and gene profiling analyses demonstrate: 1) tumor
reached vs 8.4 mo), and ORR (58.8% [95% CI 44.2-72.4] vs 31.5% [19.5- infiltrating CD8 T cells correlate with clinical responses to nivo or nivo + bev, but
45.6]) in pts with sarcomatoid features; CR rates were 11.8% (95% CI 4.4- not to nivo + ipi; 2) tumor IFN pathway gene expression correlates with re-
23.9) vs 0% (0.0-6.6). Conclusions: Pembro + axi provides benefit in the sponses; and 3) PD-L1 status, tumor mutation or mutation burden, neoantigens
combined population of pts with IMDC intermediate or poor risk and in pts did not correlate with response. Conclusions: IC therapy plus CS is safe and
whose tumors had sarcomatoid features. The observed benefits were con- beneficial to patients with MRCC and therefore, warrants testing, along with a
sistent with those seen in the total population. Clinical trial information: few correlative biomarkers, in a larger phase 3 trial. Clinical trial information:
NCT02853331. NCT 02210117.

4502 Oral Abstract Session, Mon, 8:00 AM-11:00 AM 4503 Oral Abstract Session, Mon, 8:00 AM-11:00 AM
Randomized, double-blind phase III study of pazopanib versus placebo in CALGB 90601 (Alliance): Randomized, double-blind, placebo-controlled
patients with metastatic renal cell carcinoma who have no evidence of disease phase III trial comparing gemcitabine and cisplatin with bevacizumab or
following metastasectomy: A trial of the ECOG-ACRIN cancer research group placebo in patients with metastatic urothelial carcinoma. First Author:
(E2810). First Author: Leonard Joseph Appleman, UPMC Hillman Cancer Jonathan E. Rosenberg, Memorial Sloan Kettering Cancer Center, New York, NY
Center, Pittsburgh, PA
Background: The combination of gemcitabine (G) and cisplatin (C) is a
Background: Patients with no evidence of disease (NED) after metastasectomy standard therapy for metastatic urothelial carcinoma (mUC). Based on data
for metastatic renal cell carcinoma (mRCC) are at high risk of recurrence, but that angiogenesis plays a role in UC growth and progression, a randomized
no systemic therapy has been shown to benefit this population. Pazopanib is an placebo-controlled trial was performed. Methods: Patients mUC, no prior
inhibitor of VEGFR and other kinases that improves progression-free survival in chemotherapy for metastatic disease and .12 months from prior (neo)
patients with measurable RCC metastatic disease. We performed a random- adjuvant chemotherapy and ECOG PS 0-1 were randomized 1:1 to G
ized, double-blind, placebo-controlled multicenter study to test the hypothesis 1000 mg/m2 IV days 1 and 8 and C IV 70 mg/m2 day 1 with bevacizumab
that pazopanib would improve disease-free survival in patients with mRCC (GCB) 15 mg/kg IV or placebo (GCP) day 1 every 21 days. Randomization was
rendered NED after metastasectomy Methods: Patients with NED following stratified by the presence of visceral metastases and prior chemotherapy.
metastasectomy were randomized 1:1 to receive pazopanib starting at 800 mg The primary endpoint was overall survival (OS) defined as the time from
daily vs. placebo for 52 weeks. Patients were stratified by 1 vs. . 1 site of randomization to death or last follow-up (FU). Secondary endpoints included
resected disease, and by disease-free interval # vs. . 1 year. Clinical as- progression-free survival (PFS), objective response rate (ORR), and $ grade
sessment for toxicity and patient-reported outcomes were performed every 3 toxicity. With 445 deaths, the log-rank test had an 87% power to detect a
4 weeks, and restaging scans every 12 weeks. The study was designed to hazard ratio (HR) of 0.74 with a 2-sided a=0.05. The primary analysis was
observe a 42% improvement in disease-free survival (DFS) from 25% to 45% based on the stratified log-rank test adjusting on stratification factors. Al-
at 3 years. Results: From August 2012 to July 2017, 129 patients were liance Data Safety and Monitoring Board approved the final OS analysis
enrolled. The study was unblinded after 83 DFS events had been observed be performed at 420 events due to lower than expected event rates.
(92% information). The median follow-up from randomization was 30 months Results: 506 patients were randomly assigned (252 GCB, 254 GCP)
(range 0.4 – 66.5 months). The study did not meet the primary endpoint: stratified by the presence of visceral disease and prior chemotherapy for
hazard ratio (95% CI) for DFS was 0.85 (0.55, 1.31) p= 0.47 in favor of UC. The median FU for patients still alive was 46.2 months. Median OS
pazopanib. At the time of unblinding, 22/129 (17%) of subjects had died. The was 14.5 months for patients treated with GCB and 14.3 months for
HR for overall survival (OS) was 2.65 (1.02, 6.9) in favor of placebo (p= 0.05). patients treated with GCP with a HR of 0.87 (95%CI 0.72-1.06; 2-sided
Patient-reported outcomes and laboratory correlates will be reported sepa- Wald p=0.17). The HR for PFS was 0.77 (95%CI 0.63-0.93) in favor of
rately. Conclusions: 52 weeks of pazopanib did not improve DFS compared to GCB (p=0.0074). Grade 3 or greater adverse event rate was 83.5% with
blinded placebo in patients with mRCC who were NED after metastasectomy. GCB compared to 80.7% with GCP. Conclusions: The addition of bev-
There was a trend toward worse overall survival with pazopanib. Clinical trial acizumab to GC chemotherapy did not result in improved OS (primary
information: NCT01575548. endpoint) in patients with mUC but there was a PFS improvement. The
observed median OS of about 14 months is consistent with prior phase III trials
of cisplatin-based chemotherapy. Support: U10CA180821, U10CA180882,
U10CA180820, U10CA180853, U10CA180888, Genentech https://
acknowledgments.alliancefound.org. Clinical trial information:
NCT00942331.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Genitourinary (Nonprostate) Cancer 265s

4504 Oral Abstract Session, Mon, 8:00 AM-11:00 AM LBA4505 Oral Abstract Session, Mon, 8:00 AM-11:00 AM
Randomized double-blind phase II study of maintenance pembrolizumab EV-201: Results of enfortumab vedotin monotherapy for locally advanced or
versus placebo after first-line chemotherapy in patients (pts) with metastatic metastatic urothelial cancer previously treated with platinum and immune
urothelial cancer (mUC): HCRN GU14-182. First Author: Matt D. Galsky, checkpoint inhibitors. First Author: Daniel Peter Petrylak, Yale School of
Department of Medicine, Icahn School of Medicine at Mount Sinai, Tisch Medicine, New Haven, CT
Cancer Institute, New York, NY
Background: Platinum-based chemotherapy for 1st-line treatment of pts with meta-
static urothelial cancer (mUC) is typically administered for a fixed duration followed by
observation until recurrence. PD-1 blockade with pembro improves survival of pts with
mUC progressing despite platinum-based chemotherapy. We explored the potential
benefit of earlier use of PD-1 blockade using a "switch maintenance" approach.
Methods: Pts with mUC achieving at least stable disease after up to 8 cycles of 1st-line
platinum-based chemotherapy were enrolled. Pts were randomized 1:1 to pembro
200 mg IV q3 weeks versus placebo for up to 24 months; pts progressing on placebo
could cross over to pembro. Randomization was stratified based on pre-chemotherapy
visceral metastases (Y/N) and response to 1st-line chemotherapy (CR/PR vs. SD). The
primary objective was to determine the progression-free survival (PFS) as per irRECIST
among pts treated with pembro versus placebo. Results: Between 12/2015 and 11/
2018, 107 pts were randomized to placebo (n=52) versus pembro (n=55). The The full, final text of this abstract will be available at
baseline pt characteristics are shown in the Table. Pts randomized to placebo and
pembro received a median of 6 and 8 cycles, respectively. Excluding patients with abstracts.asco.org at 7:30 a.m. ET on Monday, June 3.
baseline CRs, the objective response rate was 12% (5/42) on placebo and 22% (10/46) Onsite at the Meeting, this abstract will be printed in the
on pembro. Grade 3-4 treatment emergent adverse events occurred in 48% of pts on
placebo and 56% on pembro. At a median follow-up of 14.7 months, 41 pts have died
Monday edition of ASCO Daily News.
and 26/52 pts randomized to placebo have crossed over to pembro. PFS was sig-
nificantly longer in patients randomized to pembro vs. placebo (Maximum Efficiency
Robust Test p=0.036; log-rank p = 0.038). The 18-month restricted mean progression-
free survival time was 5.6 months with placebo and 8.2 months with pembro
(p=0.023). Conclusions: Switch maintenance pembro may “deepen” responses
achieved with 1st-line chemotherapy. Switch maintenance pembro prolongs PFS in pts
with mUC completing 1st-line platinum-based chemotherapy. Clinical trial information:
NCT02500121.
Baseline Characteristics.
Placebo (N=52) Pembro (N=55)
Age, median (range) 65 (44-87) 68 (41-83)
Pre-chemo metastatic disease
Visceral metastases 32 (62%) 39 (71%)
No visceral metastases 20 (38%) 16 (29%)
Median # cycles of 1st line chemo 6 5
Response to 1st line chemotherapy
CR/PR 36 (69%) 40 (73%)
SD 16 (31%) 15 (27%)
1st line chemo
Carboplatin-based 11 (21%) 16 (29%)
Cisplatin-based 41 (79%) 39 (71%)

4506 Oral Abstract Session, Mon, 8:00 AM-11:00 AM 4507 Oral Abstract Session, Mon, 8:00 AM-11:00 AM
SWOG S1314: A randomized phase II study of co-expression extrapolation Randomized trial of adjuvant chemotherapy versus adjuvant radiation therapy
(COXEN) with neoadjuvant chemotherapy for localized, muscle-invasive for locally advanced bladder cancer after radical cystectomy. First Author:
bladder cancer. First Author: Thomas W. Flaig, Division of Medical Oncol- Mohamed S. Zaghloul, Children’s Cancer Hospital, Cairo, Egypt
ogy, School of Medicine, University of Colorado, Aurora, CO
Background: Some chemotherapy-naı̈ve patients with locally advanced
Background: Both dose-dense Methotrexate-Vinblastine-Adriamycin/doxorubicin- bladder cancer (LABC) after radical cystectomy (RC) are sufficiently de-
Cisplatin (ddMVAC) and Gemcitabine-Cisplatin (GC) are accepted neoadjuvant conditioned that they are not candidates for adjuvant chemotherapy or de-
regimens for muscle-invasive bladder cancer (BC). We investigated COXEN, a gene cline it, even though such treatment may be warranted. There is no clear
expression model, as a predictive biomarker. Methods: Eligibility included Stage alternative adjuvant therapy for these patients, who are usually observed. In
cT2-T4a N0 M0, urothelial BC (mixed histology allowed), $ 5 mm of viable tumor, this study, we compare post-op radiotherapy (PORT) vs. adjuvant chemo-
cisplatin eligible, with plan for cystectomy. 237 patients were randomized between therapy in a randomized clinical trial. We hypothesized that PORT can achieve
ddMVAC, given every 14 days for 4 cycles, and GC, given every 21 days for 4 cycles. comparable disease-free survival (DFS). Methods: A randomized phase III
The primary objective was to assess whether the pre-specified dichotomous trial was opened to compare PORT vs. sequential chemo+PORT after RC for
treatment-specific COXEN gene expression profile is prognostic of pT0 rate or #
LABC & accrued from 2002–2008 at the NCI in Cairo. In 2007, a third arm
pT1 at surgery, and to assess whether COXEN score is a predictive factor between
comparing adjuvant chemo was added. Herein, we report the results of PORT
regimens and response. Logistic regression was used to model response, adjusting
for stratification factors. Results: 167 patients were included; the ddMVAC/GC
vs. adjuvant chemo. Patients #70 y/o with $1 of the following factors ($pT3b/
arms had a median age of 65/64, PS = 0 in 80%/75%, Male proportion of 88%/ T4a, grade 3, or positive nodes) with negative margins after RC + pelvic node
79% and T2 stage of 87%/92%. All had at least 3 cycles of chemo and surgery/ dissection were eligible. Routine follow-up & pelvic CT q6 months were
progression within 100 days of last chemo. There were favorable COXEN ddMVAC performed. PORT included 3D conformal pelvic RT (45Gy/1.5Gy BID). Chemo
scores in 32% and GC score in 26%. The pT0 rates for ddMVAC and GC were 32% included gemcitabine/cisplatin x 4. Post-hoc non-inferiority exploratory
and 35%; the rates of # pT1 were 55% and 49%, respectively. Conclusion: The analysis was performed. Results: The PORT arm accrued 78; the chemo arm
COXEN scores were not significantly prognostic for response in their individual accrued 45. 51% had urothelial carcinoma; 49% had squamous cell
arms; The COXEN GC score was significant predictor for downstaging in pooled carcinoma/other. The two arms were well-balanced except for gender (p =
arms. There was no evidence of an interaction between COXEN score and regimen 0.06). Two-year outcomes & overall adjusted hazard ratios (HR) for PORT vs.
in predicting response. The prospective data and samples from this study will allow chemo alone were 54% vs. 47% (HR 0.65(95%CI 0.35-1.19, p = 0.16) for
for further development of COXEN and other predictive biomarkers. Clinical trial DFS; 92% vs. 69% (HR 0.28(95%CI 0.10-0.82), p = 0.02 for LRFS; 75% vs.
information: NCT02177695. 79% (HR 2.39(95%CI 0.94-6.09), p = 0.07) for DMFS; 61% vs. 60% (HR
0.94(95%CI 0.52-1.69), p = 0.83) for OS. Late grade $3 GI toxicity was
Logistic Regression of COXEN Score Factor
observed in 6 PORT patients (8%) & 1 chemo patient (2%). Based on our data,
N Odds Ratio ** 95% CI ** p-value** there is a greater than 90% probability that the true difference in 2 yr DFS is
GC score* for pT0 in GC arm 82 2.63 0.82, 8.36 0.10 less than 10%, the pre-specified non-inferiority margin. Conclusions: This
For downstaging 82 1.79 0.60, 5.34 0.30 randomized study demonstrates superior local control with PORT vs. adjuvant
MVAC score* for pT0 in MVAC arm 85 1.12 0.42, 2.95 0.82 chemo with no significant differences in DFS, DMFS or OS. Results suggest
For downstaging 85 0.92 0.37, 2.27 0.86
GC score* for downstaging in both arms 167 2.33 1.11, 4.89 0.02 that PORT could be an option for patients with LABC after RC who are
MVAC score* for downstaging in both arms 167 0.90 0.46, 1.75 0.76 medically unfit for adjuvant chemo or who decline it. Clinical trial information:
NCT01734798.
* favorable based on prespecified algorithm and dichotomous cut point
** adjusted for two strat factors – clinical stage at baseline (T2 vs T3, T4a), PS (0 vs 1)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
266s Genitourinary (Nonprostate) Cancer

4508 Oral Abstract Session, Mon, 8:00 AM-11:00 AM 4509 Poster Discussion Session; Displayed in Poster Session (Board #335),
Cytoreductive nephrectomy (CN) in metastatic renal cancer (mRCC): Update on Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Carmena trial with focus on intermediate IMDC-risk population. First Author: Mon, 4:30 PM-6:00 PM
Arnaud Mejean, Department of Urology, Hôpital Européen Georges-Pompidou - A phase II study of RC48-ADC in HER2-positive patients with locally
Paris Descartes University, Paris, France advanced or metastatic urothelial carcinoma. First Author: Xinan Sheng,
Peking University Cancer Hospital, Beijing, China
Background: Carmena was a randomized phase III trial, testing the benefit of
CN followed by sunitinib (arm A) vs sunitinib alone (arm B), with stratification Background: Urothelial carcinoma (UC) is the third largest cancer type with
by MSKCC risk groups in 450 mRCC patients. Based on this trial, CN is not HER2 positive cancer. RC48-ADC is a novel humanized anti-HER2 antibody-
anymore recommended in mRCC (NEJM, Mejean et al, 2018). However there drug conjugate (ADC). This study was to evaluate the activity of RC48-ADC in
are questions about which patients could still benefit from CN, especially in HER2-positive patients with locally advanced or metastatic urothelial carci-
intermediate risk group. In the present study, we investigated different sub- noma (mUC). Methods: This study is an open-label, multicenter, single-arm,
groups from the Carmena trial to answer these questions. Methods: Carmena non-randomized phase II study. Eligibility criteria include: histologically con-
trial was initially stratified according to MSKCC risk groups. For the purpose of firmed UC, HER2-positive (IHC 2+ or 3+), ECOG PS 0-1, treated with $1 prior
this analysis, we reclassified the patients based on IMDC risk groups. We also systemic treatment. The patients received RC48-ADC treatment alone (2 mg/kg
analyzed patients with one metastatic site vs more than one, as well as patients IV infusion, q2w) until disease progression, unacceptable toxicity, withdrawal,
with secondary nephrectomy in arm B. Overall survival (OS) was the primary or study termination. The primary endpoint was objective response rate (ORR).
endpoint. Results: With a updated median FU of 61.5 months (mo), the Progress-free survival (PFS), overall survival (OS), and safety was also assessed.
median OS by ITT analysis was 15.6 vs 19.8 mo in arm A and B respectively Results: Patient enrollment for this study was completed in November 2018. A
stratified on MSKCC (HR 0.933 ; 95% CI [0.76- 1.15]) / stratified on IMDC total of 43 patients were enrolled, with a median age of 64 years old. At
(HR 0.957 ; 95% CI [0.78- 1.18]). Using IMDC risk group factors, 58.6% baseline, most patients (37/43) had visceral metastasis. Fourteen (32.6%)
patients were intermediate and 41.4 % were poor risk. When looking at in- patients had received $ 2 lines treatment and 8 (18.6%) patients had prior
termediate risk group only, 48.1% had only one risk factor (interval between immune checkpoint inhibitor (CPI) therapy in second line treatment. The
diagnosis and treatment , 1y), with a median OS of 30.5 and 25.2 mo in arm objective response rate was 60.5% (95% CI: 44.4%, 75.0%) and the DCR was
A and B respectively (HR 1.24 [0.81 – 1.90]). By contrast, 51.9 % had two risk 90.7% (39/43). As of Jan 23, 2019, the median PFS for the overall study
factors (mostly low hemoglobin, high corrected calcium or neutrophils), with a population was not yet reached, and the median PFS was 7.8 months (95% CI:
median OS of 16.6 and 31.2 mo in arm A and B respectively (HR 0.61 [0.41 – 4.9, 10.7) for the 9 patients who started RC48-ADC prior to Jun 30, 2018. The
0.91] p = 0.015). Regarding number of metastatic sites, 33% had only one ORR was 70.6% (12/17) in patients with HER2 FISH+ or IHC3+. The ORR was
metastatic site. Median OS was 23.6 and 22.7 mo in arm A and B respectively 64.9% (24/37) in patients with visceral metastasis and was 70.0% (14/20) in
(HR 1.08 [0.75 – 1.57]. Finally, 40 patients had a secondary nephrectomy in liver metastasis patients especially. The ORR was 64.3% in patients post to $ 2
arm B, with median OS of 48.5 mo [CI 95%: 27.9-64.4] vs 15.7 mo [CI 95%: lines treatment and 75.0% in patients post to immunotherapy. Common
13.3-20.5] in patients who never had surgery. Conclusions: With longer FU, treatment-related AEs were leukopenia (51.2%), hypoesthesia (41.9%), alo-
Carmena trial confirms that CN is not superior to sunitinib alone in ITT pecia (41.9%), neutropenia (37.2%), fatigue (34.9%), ALT increase (32.6%),
population, both with MSKCC and IMDC risk groups. However for patients with and AST increase (32.6%); Most were Grade 1 or 2. Conclusions: RC48-ADC
only one IMDC risk factor, CN might be beneficial. Number of metastatic site is has demonstrated a clinically meaningful ORR of 60.5% in pretreated HER-2
not helpful to define good candidates for surgery. Finally, patients with sec- positive mUC patients including those who underwent failure to the immu-
ondary nephrectomy have very long OS, supporting this strategy. Clinical trial notherapy. Clinical trial information: NCT03507166.
information: NCT00930033.

4510 Poster Discussion Session; Displayed in Poster Session (Board #336), 4511 Poster Discussion Session; Displayed in Poster Session (Board #337),
Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Mon, 4:30 PM-6:00 PM Mon, 4:30 PM-6:00 PM
Infigratinib in upper tract urothelial carcinoma vs urothelial carcinoma of the FIERCE-22: Clinical activity of vofatamab (V) a FGFR3 selective inhibitor in
bladder and association with comprehensive genomic profiling/cell-free combination with pembrolizumab (P) in WT metastatic urothelial carcinoma,
DNA results. First Author: Nazli Dizman, City of Hope Comprehensive preliminary analysis. First Author: Arlene O. Siefker-Radtke, The University
Cancer Center, Duarte, CA of Texas MD Anderson Cancer Center, Houston, TX
Background: Infigratinib (BGJ398) is a potent and selective FGFR1–3 in- Background: Patients (pts) with mUC who have failed platinum-based che-
hibitor with significant activity in patients (pts) with advanced or metastatic motherapy have a poor prognosis. Reported response rates to immune
urothelial carcinoma (mUC) bearing FGFR3 alterations [Pal et al 2018]. Given checkpoint inhibitors (ICI) are approximately 20%. 20% of pts with mUC
the distinct biologic characteristics of upper tract UC (UTUC) and urothelial harbor FGFR3 mutations or fusions (M/F), which may result in lower sensitivity
carcinoma of the bladder (UCB), we sought to determine if infigratinib had to ICI. V (B-701) is a fully human monoclonal antibody against FGFR3 that
varying activity in these settings. Methods: Eligible pts had mUC with acti- blocks activation of both the wildtype and genetically activated receptor. This
vating FGFR3 mutations/fusions and prior platinum-based chemotherapy, is a Phase 1b/2 study designed to evaluate V monotherapy window followed in
unless contraindicated. Pts received infigratinib 125 mg orally daily (3 wks on/ combination of V with P(VP) (NCT03123055). Methods: This trial enrolled
1 wk off). Overall response rate (ORR: CR+PR) and disease control rate (DCR; mUC pts with failure to $ 1 prior line of chemotherapy or recurrence
CR+PR+SD) were characterized in UCB and UTUC pts. Comprehensive ge- # 12 months of (neo)adjuvant chemotherapy, measurable disease and
nomic profiling was performed on FFPE tissues in a CLIA-certified lab ECOG ,2. Treatment consisted of v at 25 mg/kg alone for 2 week monotherapy
(Foundation Medicine; Cambridge, MA). Blood was collected for cell-free (cf) window followed by combination with P 200 mg q3w.during the V window
DNA analysis using a 600-gene panel on an Illumina HiSeq 2500 sequencer. paired tumor biopsy were obtained. Efficacy was assessed by investigators
Results: 67 pts were enrolled; the majority (70.1%) had received $2 prior (RECIST 1.1). Primary objectives were safety and activity [ORR]). Results: 35
antineoplastic therapies. ORR was 25.4% and DCR was 64.2%. In the 8 pts pts have received treatment (Ph1b:8, WT:20, M/F+: 7). WT patients were
with UTUC, 1 CR and 3 PRs were observed (ORR 50%); the remainder had a unselected for PD-1 status, predominately male (55%) white (95%), all had
best response of SD (DCR 100%). UTUC pts were predominantly 2nd line received at least 1 line of prior chemo and 60% had Bellmunt scores of . 1.
(62.5%), with only 2 (25%) showing response to previous treatment. In pts The safety profile is consistent with previously reported data for P. TEAE
with UCB, 13 PRs were observed (ORR 22%), and 22 pts had a best response occurring in .20% of patients were nausea, anemia, diarrhea and fatigue. Six
of SD (DCR 59.3%). Notable differences in genomic alterations between WT patients (30%) had responses (4 confirmed responses, 1 unconfirmed),
cohorts included a higher frequency of FGFR3-TACC3 fusions (12.5% vs and an additional patient with an iRECIST response. Responses occurred at a
5.8%) and FGFR3 R248C mutations (50% vs 11.5%), and a lower frequency median of 3.5 months. At 4+ months of follow up, 13(65%) remain on
of FGFR3 S249C mutations (25% vs 59.6%) in UTUC vs UCB. Consistent with treatment @ a median of 8 cycles (range: 1-13). At 5+ months the median PFS
previous reports [Sfakianos et al 2016], UTUC pts had a differential frequency has not been reached. Conclusions: VP combination therapy is well tolerated
of alterations in HRAS, CDKN2B and ARID1A. Sufficient cfDNA yield was with an encouraging ORR and prolonged PFS in the WT cohort; greater than
obtained in UTUC and UCB pts and a comprehensive comparison of these data one would anticipate from P alone based upon historical data. We will be
will be presented. Conclusions: Differences in cumulative genomic profile reporting 9+ month preliminary PFS/OS and updated OOR/DOR data. Clinical
were observed between UCB and UTUC in this FGFR3-restricted experience, trial information: NCT03123055.
underscoring the distinct biology of these diseases. Results with infigratinib in
UTUC support a planned phase III adjuvant study predominantly in this
population. Clinical trial information: NCT01004224.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Genitourinary (Nonprostate) Cancer 267s

4512 Poster Discussion Session; Displayed in Poster Session (Board #338), 4513 Poster Discussion Session; Displayed in Poster Session (Board #339),
Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Mon, 4:30 PM-6:00 PM Mon, 4:30 PM-6:00 PM
Atezolizumab (atezo) + bevacizumab (bev) versus sunitinib (sun) in pts with CheckMate 214 post-hoc analyses of nivolumab plus ipilimumab or sunitinib
untreated metastatic renal cell carcinoma (mRCC) and sarcomatoid (sarc) in IMDC intermediate/poor-risk patients with previously untreated advanced
histology: IMmotion151 subgroup analysis. First Author: Brian I. Rini, renal cell carcinoma with sarcomatoid features. First Author: David F.
Cleveland Clinic Taussig Cancer Institute, Cleveland, OH McDermott, Beth Israel Deaconess Medical Center, Dana-Farber/Harvard
Cancer Center, Boston, MA
Background: In the Phase 3 IMmotion151 trial, atezo + bev showed improved PFS
vs sun in untreated mRCC pts expressing PD-L1. Here we report results of a pre- Background: Patients (pts) with advanced renal cell carcinoma with sarcomatoid
specified subgroup analysis in pts whose tumors have sarc histology, an independent features (sRCC) have poor prognosis and suboptimal outcomes with anti-VEGF tar-
predictor of poor survival. Methods: Pts were randomized to receive atezo 1200 mg geted therapy. Nivolumab plus ipilimumab (N+I) demonstrated superior objective
IV q3w + bev 15 mg/kg IV q3w or sun 50 mg po qd for 4w on, 2w off. Coprimary response rate (ORR) and overall survival (OS) vs sunitinib (S) in previously untreated
endpoints were reported previously (Motzer ASCO GU 2018). Secondary endpoints pts with International Metastatic RCC Database Consortium (IMDC) intermediate/poor
included INV-PFS and OS in sarc pts and are shown here with INV-ORR, safety, PRO (I/P)-risk, clear-cell, advanced RCC in the phase 3 CheckMate 214 trial. Methods: We
and biomarker data. Results: 142 randomized pts (16%) from IMmotion151 had performed a post-hoc exploratory analysis of N+I vs S in CheckMate 214 sRCC pts.
tumors with any component of sarc histology; mPFS was 8.3 vs 5.3 mo with atezo + The presence of sarcomatoid features was assessed by keyword search for “sarco-
bev vs sun and mOS was NR vs 15.0 mo, respectively (see Table for PD-L1+). ORR matoid” in pts with available local pathology reports accompanying pretreatment
was 49% vs 14% and CR rate was 10% vs 3% in the atezo + bev vs sun arms. Grade tumor samples. Results: 842 (77%) of 1096 intention-to-treat pts had local pa-
3-4 AEs occurred in 27 pts (40%) with atezo + bev and 34 (49%) with sun. Using the thology reports available, including 112 randomized pts with I/P-risk sRCC (N+I, n =
MDASI scale, sarc pts reported longer median time to deterioration (TTD) of 60; S, n = 52). Baseline characteristics of sRCC pts were balanced between arms.
symptom interference with daily activities with atezo + bev vs sun (11.3 vs 4.9 mo). Notably, 47% vs 53% of I/P-risk sRCC pts in the N+I and S arms had tumor PD-L1
Prevalence of AngiogenesisHigh gene expression (GE) signature subset was lower expression $1% at baseline, which was higher than in all I/P-risk pts (N+I, 26% vs S,
(34% vs 65%) and T-effectorHigh GE subset was higher (54% vs 40%) in sarc vs non- 29%). In descriptive analyses performed at a minimum follow-up of 30 months,
sarc tumors. PD-L1+ disease was more common in sarc vs non-sarc tumors (63% vs confirmed ORR and complete response rate per investigator (RECIST v1.1), OS, and
39%). Conclusions: mRCC pts with sarc histology had longer OS and PFS and a progression-free survival (PFS) per investigator were improved with N+I vs S in I/P-risk
higher ORR/CR rate when treated with atezo + bev vs sun, regardless of PD-L1 status. pts with sRCC (Table). No new safety signals were seen in sRCC pts. Conclusions: In
Biomarker data support a biological correlate for the increased responsiveness to this post-hoc descriptive subgroup analysis of CheckMate 214, N+I demonstrated
atezo + bev in sarc pts. Clinical trial information: NCT02420821. promising efficacy and prolonged survival vs S, with consistent safety, in previously
All Sarc PD-L1+ Sarc
untreated, I/P-risk, advanced clear-cell RCC with sarcomatoid features. Prospective
studies of N+I that include pts with sRCC are ongoing. Clinical trial information:
Atezo + Bev Sun Atezo + Bev Sun
n = 68 n = 74 Stratified HR n = 36 n = 50 Stratified HR NCT02231749.
mPFS, mo 8.3 5.3 0.52 8.6 5.6 0.45 Randomized I/P-risk Randomized I/P-risk
(5.4, 12.9) (3.3, 6.7) (0.34, 0.79) (3.9, 15.3) (3.3, 6.7) (0.26, 0.77)
mOS, mo NR 15.0 0.56 NR 15.0 0.53
sRCC pts sRCC pts P value/HR (95% CI)
(18.3, NR) (8.7, NR) (0.32, 0.96) (17.1, NR) (8.4, NR) (0.27, 1.06) N+I (N = 60) S (N = 52)
12-mo OS, % 69 60 — 71 61 —
(58, 81) (48, 71) (56, 86) (47, 75)
ORR, % (95% CI) 56.7 (43.2–69.4) 19.2 (9.6–32.5) P, 0.001
ORR, % 49 14 — 56 12 — Complete re- 18.3 0 –
(36, 61) (7, 23) (38, 72) (5, 24) sponse, %
CR, % 10 3 — 14 4 — Median PFS, mo 8.4 (5.2–24.0) 4.9 (4.0–7.0) HR (95% CI), 0.61 (0.38‒0.97);
TTD MDASI, mo 11.3 4.9 0.61 — — — P, 0.03
(6.3, 17.5) (3.5, 7.6) (0.33, 1.11) Median OS, mo 31.2 (23.0‒NE) 13.6 (7.7‒20.9) HR (95% CI), 0.55 (0.33‒0.90);
m, median NR, not reached Parentheses denote 95% CI Stratification factors: MSKCC risk score, liver mets, PD-L1
P, 0.0155
status NE, not estimable

4514 Poster Discussion Session; Displayed in Poster Session (Board #340), 4515 Poster Discussion Session; Displayed in Poster Session (Board #341),
Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Mon, 4:30 PM-6:00 PM Mon, 4:30 PM-6:00 PM
Efficacy of immune checkpoint inhibitors (ICI) and genomic characterization Patient-reported outcomes (PROs) in IMmotion150: Atezolizumab (atezo)
of sarcomatoid and/or rhabdoid (S/R) metastatic renal cell carcinoma (mRCC). alone or with bevacizumab (bev) versus sunitinib (sun) in first-line metastatic
First Author: Ziad Bakouny, Dana-Farber Cancer Institute, Boston, MA renal cell carcinoma (mRCC). First Author: Sumanta K. Pal, City of Hope
National Medical Center, Duarte, CA
Background: S/R mRCC are poorly characterized rapidly progressing tumors as-
sociated with poor prognosis. Although conventional therapies are less effective for Background: The phase 2 IMmotion150 study showed improved PFS with atezo +
these tumors, emerging data suggests that ICIs may be especially effective. Our aim bev vs sun in patients (pts) whose tumors were PD-L1+ and corroborated the
was to characterize the genomic alterations (GA) in S/R mRCC tumors and evaluate activity of atezo monotherapy in previously untreated mRCC. This study offers an
their response to ICIs. Methods: We retrospectively compared the activity of first- opportunity to evaluate PROs with immunotherapy and VEGF-directed therapy
line ICIs to non-ICI-based therapies for S/R mRCC patients (pts) treated at DFCI and alone and in combination. Methods: Pts randomized to atezo 1200 mg IV q3w
analyzed sequencing data from an NGS panel (275-447 genes) on a subset of these alone or with bev 15 mg/kg IV q3w, or sun 50 mg po qd for 4 wk on/2 wk off
patients (matched by histology to non-S/R mRCC). For S/R mRCC pts treated with completed the MD Anderson Symptom Inventory (MDASI) and Brief Fatigue
ICI vs non-ICI therapies, overall survival (OS) and time to treatment failure (TTF) Inventory (BFI) q3w of each 6-wk cycle until progression (RECIST 1.1). Time to
were compared by Cox regression and objective response rate (ORR) by logistic deterioration (TTD; first $ 2-point score increase over baseline [BL]) and change
regression. GA frequencies were compared by Fisher’s test and tumor mutational from BL (effect size [ES] $ 0.2 suggests a clinically important difference vs sun)
burden (TMB) by Mann Whitney U between S/R and non-S/R mRCC. Results were in MDASI symptom severity and interference and BFI fatigue severity and in-
considered statistically significant if p , 0.05 or q , 0.10. Results: 125 S/R mRCC terference scores are reported for all-comers. Results: Completion rates were
pts were included (88 S, 23 R, 14 S&R) among which 103 were clear cell and . 90% at BL and $ 80% at most visits across arms. BL PRO scores indicating
48 had sequencing data. GA in BAP1 were significantly more frequent in S/R vs mild symptoms and interference were similar across arms. Delayed TTD of
non-S/R (25% vs 4.3%; q = 0.096) while other GA had similar frequencies and symptom severity and interference with daily life was seen with atezo vs sun and
TMB (median [IQR]) was similar (7.2 [5.2-8.4] vs 6.8 [5.3-9.1] mut/Mb; p = 0.98). atezo + bev vs sun and was longest with atezo alone (Table). Pts reported milder
Median follow-up was 35.4 (95% CI = 24.9 – 46.0) months (m). On multivariable
symptoms and less interference during the first 6 cycles with atezo vs sun: ES
analysis, S/R mRCC pts treated with ICI had significantly better clinical outcomes
mean (range) was 0.36 (0.07-0.68) for core symptom severity and 0.36 (0.04-
(Table). Conclusions: Pts with S/R mRCC have a higher frequency of BAP1 GA and
0.83) for symptom interference. The 5 worst pt-reported symptoms during tx (dry
better outcomes on ICIs compared to non-ICI-based therapies. Future studies
mouth, fatigue, rash, drowsiness, lack of appetite) were all in the sun arm; all 16
should determine the molecular mechanisms underlying the improved response to
symptoms measured were milder with atezo vs sun. Conclusions: PROs suggest
ICIs in S/R mRCC.
that atezo alone or with bev maintained daily function with minimal symptom
ICI Non-ICI Univariable Multivariable* interference vs sun. Clinical activity, safety and PRO data for atezo support its
(N = 39) (N = 86; 90.7% TKI-based) (OR/HR) (adjusted OR/HR)
investigation in adjuvant tx of high-risk RCC pts (IMmotion010; NCT03024996).
ORR 16/37 (43.2%) 20/72 (27.8%) 1.98 (0.86 – 4.54) 3.02 (1.15 – 7.98)
(N = 109)
Clinical trial information: NCT01984242.
Median TTF 4.8 m (1.6 – 8.0) 4.9 m (3.6 – 6.2) 0.69 (0.45 – 1.05) 0.65 (0.41 – 1.01)
(111/125 events) TTD: HR (95% CI) vs Sun.
24m OS rate 48.8% (30.6 – 67.0) 29.7% (19.7 – 39.7) 0.58 (0.34 – 0.99) 0.52 (0.30 – 0.90) Atezo Atezo + Bev
(89/125 events)
MDASI core symptom severity 0.39 (0.22, 0.71) 0.74 (0.45, 1.20)
* Adjusted for IMDC risk group and histology (clear cell vs non-clear cell)
MDASI RCC symptom severity 0.22 (0.12, 0.41) 0.60 (0.38, 0.94)
MDASI symptom interference 0.36 (0.22, 0.58) 0.70 (0.47, 1.04)
BFI fatigue severity 0.48 (0.33, 0.70) 0.75 (0.53, 1.06)
BFI fatigue interference 0.38 (0.24, 0.60) 0.67 (0.46, 0.99)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
268s Genitourinary (Nonprostate) Cancer

4516 Poster Discussion Session; Displayed in Poster Session (Board #342), 4517 Poster Discussion Session; Displayed in Poster Session (Board #343),
Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Mon, 4:30 PM-6:00 PM Mon, 4:30 PM-6:00 PM
Active surveillance in metastatic renal cell carcinoma (mRCC): Results from Safety and efficacy of nivolumab plus ipilimumab (NIVO+IPI) in patients
the Canadian Kidney Cancer information system (CKCis). First Author: Igal with advanced renal cell carcinoma (aRCC) with brain metastases: Interim
Kushnir, Ottawa Hospital Cancer Centre, Ottawa, ON, Canada analysis of CheckMate 920. First Author: Hamid Emamekhoo, University of
Wisconsin School of Medicine and Public Health, Madison, WI
Background: Active surveillance (AS) is a commonly used strategy in pa-
tients (pts) with low tumor burden or slow growing disease. However, few Background: Previous clinical trials of patients (pts) with aRCC, including
studies have assessed AS for mRCC compared to immediate treatment. We CheckMate 214, have mostly excluded pts with brain metastases. However,
aimed to assess the outcomes and safety of AS in comparison to immediate antitumor activity in pts with brain metastases has been observed in pts with
systemic treatment for mRCC pts. Methods: Using CKCis, mRCC pts di- melanoma treated with NIVO 1 mg/kg + IPI 3mg/kg and pts with non-small
agnosed between January 1, 2011 and December 31, 2016 were identified. cell lung cancer treated with NIVO 240 mg + IPI 1mg/kg. CheckMate 920 is
AS strategy was defined as: (1) start of systemic therapy $6 months after an ongoing, phase 3b/4 clinical trial of NIVO + IPI treatment in pts with aRCC
diagnosis of mRCC; or (2) never receiving systemic therapy for mRCC with an with a high unmet medical need. Here, we present the safety and efficacy
overall survival (OS) $1 yr (OS $ 1 yr a surrogate to exclude pts not started interim results for the cohort of pts with brain metastases. Methods: Pts with
on treatment due to poor prognosis). Pts starting systemic treatment previously untreated aRCC of any histology, with asymptomatic brain me-
, 6 months after diagnosis of mRCC were defined as receiving immediate tastases (not on corticosteroids or receiving radiation), and Karnofsky per-
systemic treatment. OS and time until 1st line treatment failure (TTF) be- formance status $70% were assigned to treatment with NIVO 3 mg/kg + IPI
tween the two cohorts were compared. Results: A total of 863 pts met criteria 1 mg/kg every 3 weeks for 4 doses, followed by NIVO 480 mg every 4 weeks.
for AS (cohort A). Of these, 370 started treatment $ 6 months after their Pts were treated until disease progression, unacceptable toxicity, or for a
initial diagnosis (cohort A1) and 493 never received systemic treatment and maximum of 2 years. The primary endpoint was the incidence of high-grade
were alive for $1 year (cohort A2). 848 pts received immediate systemic immune-mediated adverse events (IMAEs). Key secondary endpoints in-
treatment (cohort B). Median age for pts in cohort A and B was 65.1 (19.0- cluded progression-free survival (PFS) and objective response rate (ORR) by
91.5) vs. 62.2 yrs (23.1-87.1) (p , 0.0001). Sex distribution was not RECIST v1.1 per investigator. Exploratory endpoints included additional
statistically different. Pts in cohort A had fewer sites of metastatic disease vs. safety analyses and overall survival (OS). Results: Overall, 28 patients were
cohort B ( , 0.0001) and 23% of pts in cohort A had metastasectomy vs. 5% enrolled in the brain metastases cohort. With a minimum follow-up of
in cohort B (P = , 0.0001). Five-year OS probability was significantly greater 6.47 months, grade 3-4 IMAEs within 100 days of last dose were reported in
for cohort A than for cohort B (70.2% vs. 32.1%; P , 0.0001). After 6 cases. The grade 3-4 IMAEs observed in $ 1 patient were diarrhea, colitis,
adjusting for IMDC risk criteria and age, both OS (HR 0.46, 0.38-0.56, P , diabetic ketoacidosis, immune-mediated hepatitis, hypophysitis, and rash of
0.0001) and TTF (HR 0.79, 0.69-0.92, P = 0.0021) were greater in cohort any type (n = 1 each). No treatment-related grade 5 IMAEs were reported.
A1 vs. B. For cohort A1 the median time on AS was 14.2 m (range 6 – 71). ORR by RECIST v1.1 per investigator in all treated subjects was 28.6%
Conclusions: Based on the largest analysis of AS in mRCC to date, our data (95% CI 13.2–48.7). Median PFS in all treated subjects was 9.0 months
suggest that a subset of pts may be safely observed without immediate (95% CI 2.9–not estimable [NE]). Median OS has not been reached (95% CI
initiation of systemic therapy. Prospective validation is required in the 13.1–NE). Conclusions: In pts with aRCC and brain metastases who are
contemporary immunotherapy era. often excluded from clinical trials, NIVO + IPI treatment showed a safety
profile consistent with previous reports of this dosing regimen, with en-
couraging antitumor activity. Clinical trial information: NCT02982954.

4518 Poster Discussion Session; Displayed in Poster Session (Board #344), 4519 Poster Discussion Session; Displayed in Poster Session (Board #345),
Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Mon, 4:30 PM-6:00 PM Mon, 4:30 PM-6:00 PM
Phase II study of nivolumab and ipilimumab for advanced bladder cancer of Clinical outcomes according to PD-L1 status and age in the prospective
variant histologies (BCVH). First Author: Bradley Alexander McGregor, Dana- international SAUL study of atezolizumab (atezo) for locally advanced or
Farber Cancer Institute, Boston, MA metastatic urothelial carcinoma (UC) or non-UC of the urinary tract. First
Author: Cora N. Sternberg, Weill Cornell Medicine, New York, NY
Background: Patients with BCVH have poor outcomes and data regarding the
management of this heterogeneous group of patients is limited. Nivolumab Background: Atezo, a monoclonal antibody targeting PD-L1, is an approved
and ipilimumab has demonstrated safety and efficacy in urothelial carci- therapy for locally advanced/metastatic UC based on IMvigor210 and
noma and other malignancies. In this multicenter, single arm, multi-cohort IMvigor211 phase II and III trials. The single-arm SAUL study (NCT02928406)
phase II trial we evaluate the efficacy of nivolumab and ipilimumab in with a broader patient (pt) population demonstrated median overall survival (OS)
patients with BCVH and other advanced rare genitourinary cancers (NCT of 8.7 months and a safety profile consistent with previous atezo trials.
03333616). Herein, we report the preliminary results of the fully accrued Methods: Pts with locally advanced/metastatic UC or non-UC of the urinary tract
BCVH cohort. Methods: Eligible patients had metastatic BCVH, ECOG received atezo 1200 mg every 3 weeks until disease progression or un-
performance status of 0-2 and were either untreated or had received any acceptable toxicity. Populations excluded from IMvigor211 (renal impairment,
number of lines of prior therapy excluding prior immunotherapy. Patients ECOG PS 2, treated asymptomatic CNS metastases, stable controlled auto-
underwent a baseline biopsy and blood collection for correlative studies and immune disease, concomitant steroids, HIV positive, non-UC) were eligible. The
received treatment with nivolumab 3 mg/kg and ipilimumab 1 mg/kg in- primary endpoint was safety; OS and overall response rate (ORR) were secondary
travenously every 3 weeks for 4 cycles with continued maintenance of endpoints. Predefined subgroup analyses included outcomes according to PD-
nivolumab 480 mg IV every 4 weeks. The primary endpoint was overall L1 status (VENTANA SP142) and age in the overall population (and the
response rate (ORR) by RECIST 1.1. Results: 19 BCVH patients were en- IMvigor211-like subgroup for PD-L1). Results: Between Nov 2016 and Mar
rolled at 4 institutions between 4/2018 and 1/2019: squamous cell (n = 6), 2018, 1004 pts were enrolled; 997 received atezo. Efficacy is summarized
small cell (n = 3), adenocarcinoma (n = 3), urachal (n = 5), plasmacytoid below. Incidences of grade $3 treatment-related adverse events were similar
(n = 1), and spindle cell (n = 1). 13 (68%) patients had received prior irrespective of PD-L1 status (overall IC 0/1 vs 2/3: 11% vs 16%; IMvigor211-
systemic therapy including platinum-based chemotherapy in 92% patients. like IC 0/1 vs 2/3: 11% vs 15%) or age ($65 y: 13%; $75 y: 12%; $80 y:
Median number of cycles of ipilimumab plus nivolumab received was 10%). Conclusions: OS and ORR appear more favorable in IC 2/3 vs IC 0/1
3 (range 1-8) and median follow-up was 3.6 (0.3-8.8) months. 13 patients subgroups (overall and in the IMvigor211-like population). Atezo was effective
had undergone at least one scan; ORR was 31% (4/13, 80%CI: 14-52%), and well tolerated across subgroups including elderly pts. Clinical trial in-
with partial responses seen in small cell carcinoma (n = 2), urachal (n = 1) formation: NCT02928406.
and a complete response in 1 patient with plasmacytoid carcinoma. 3 patients All pts, PD-L1 IMvigor211-like, PD-L1 All pts, age
(16%) developed treatment-related grade 3 toxicities with 1 (5%) grade IC 0/1 IC 2/3 IC 0/1 IC 2/3 ‡65 y ‡75 y ‡80 y
4 toxicity. Conclusions: Nivolumab and ipilmumab resulted in objective Endpoint (n=666) (n=268) (n=427) (n=176) (n=624)a (n=227)a (n=78)a
responses in a subset of patients with BCVH with manageable toxicities. Deaths, n (%) 388 (58) 132 (49) 235 (55) 82 (47) 335 (54) 128 (56) 44 (56)
Median OS, 7.9 11.6 9.0 14.5 8.5 8.3 8.3
Updated clinical and correlative data will be presented. This combina- months (6.8–9.1) (8.8–18.8) (7.8–10.4) (9.5–18.8) (7.5–10.9) (7.3–10.9) (5.4–11.2)
tion may warrant further investigation in patients with BCVH, which has (95% CI)
6-month OS rate, 57 (53–61) 67 (61–72) 61 (56–66) 72 (65–78) 60 (56–64) 61 (54–67) 59 (47–69)
substantial unmet needs. Clinical trial information: NCT 03333616. % (95% CI)
ORR, % (95% CI) 10 (8–13) 21 (16–26) 10 (7–13) 23 (17–30) 14 (12–17) 13 (9–18) 8 (3–16)
a
Subgroups not mutually exclusive

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Genitourinary (Nonprostate) Cancer 269s

4521 Poster Session (Board #347), Mon, 1:15 PM-4:15 PM 4522 Poster Session (Board #348), Mon, 1:15 PM-4:15 PM
Correlation of methylthioadenosine phosphorylase (MTAP) loss with response Interim analysis of ibrutinib plus paclitaxel for patients with advanced urothelial
to anti-folate therapy in urothelial bladder carcinoma (UBC). First Author: carcinoma previously treated with platinum-based chemotherapy. First Author:
Omar Alhalabi, Division of Cancer Medicine, The University of Texas MD Daniel E. Castellano, Medical Oncology Service, Hospital Universitario 12 de
Anderson Cancer Center, Houston, TX Octubre, Madrid, Spain
Background: The MTAP gene encodes an essential enzyme for the salvage Background: Patients (pts) with advanced urothelial carcinoma (aUC) who
pathway of nucleotide synthesis and is frequently deleted in UBC. Anti-folate progressed after platinum-based chemotherapy (PBCT) have a median overall
agents such as pemetrexed can effectively inhibit the de novo pathway of survival (OS) of ~6 mo on CT and 8-10 mo on novel immune checkpoint
nucleotide synthesis and as a result, create a synthetic lethality in MTAP de- inhibitors (ICIs). Treatment (tx) with ibrutinib (ibr), a first-in-class, once-daily
ficient UBC. We hypothesize that MTAP gene loss correlates with enhanced inhibitor of Bruton’s tyrosine kinase (BTK) with immunomodulatory properties,
response to pemetrexed in UBC. Methods: We investigated MTAP gene deletion plus paclitaxel (pac) may improve outcomes for pts with aUC. Methods: Pts
rates in the TCGA database and determined MTAP protein loss rates by im- with aUC who had 1-2 prior tx ($1 PBCT and/or ICIs) formed one cohort of this
munohistochemistry (IHC) using a UBC tissue microarray (TMA) from 151 phase 1b/2 study. Recommended phase 2 dose per phase 1b was ibr 840
patients (pts). We then performed in vitro and in vivo studies using MTAP mg/d + weekly pac 80 mg/m2 in 21-d cycles; starting ibr dose of 560 mg/d.
proficient and MTAP deficient bladder cancer cell lines. At the clinical level, we nanoString gene expression assay (NanoString Technologies) was used for
performed a retrospective analysis based on MTAP status of pts treated with preliminary assessment of ibr-targeted kinases in baseline tumor biopsies.
pemetrexed as 2nd line at our institution between 2014 and 2018. We are now Results: This interim analysis included the first 29 pts treated (ibr 840/
enrolling pts in a single-arm, open-label, phase II clinical trial (NCT02693717) 560 mg/d, n = 25/4; median age 67 y). 52% had 2 prior regimens (prior
with pemetrexed in pts with MTAP deficient UBC. Results: Per our TCGA and PD-L1 inhibitors, 100%). Median time on study/tx was 7.2/2.3 mo (max tx:
TMA IHC analyses, MTAP deficiency rate was 25.9% and 27.8%, respectively. 11.5 mo). Unconfirmed overall response rate (ORR) was 41% (complete/
MTAP deficient UBC cell lines were at least 40 times more sensitive to partial response: 10%/31%). Disease control rate was 62%. ORR was 71%
pemetrexed than MTAP proficient lines. Knockdown of the MTAP gene in- for 7 pts with only lymph node metastases; ORR was 36% for 14 pts with 1
creased apoptosis rate by pemetrexed from approximately 20% to 60%. Ad- prior tx and 47% for 15 pts with 2 prior tx. Median duration of response was
ditionally, pemetrexed significantly inhibited the growth of MTAP deficient or 4.2 mo (90% CI: 1.9-7.1). Median progression-free survival was 3.6 mo
knockdown xenograft tumors but not MTAP proficient tumors. Retrospective (90% CI: 1.6-5.4). Median OS was 14.7 mo (90% CI: 7.7-15.9). Follow-
analysis of 12 pts using RECIST criteria indicated that all 4 MTAP deficient UBC up is ongoing for durability/survival with 6 pts alive without progression,
pts responded to pemetrexed whereas only 1 of 8 (12.5%) MTAP proficient UBC including 4 still on ibr. 76% of pts had grade $3 adverse events (AEs); 4
pts responded. Of the 6 pts enrolled on the clinical trial, 3 (50%) had complete discontinued ibr + pac due to AEs and 1 had major hemorrhage. Pac
or partial response, 1 had stable disease, 1 was not evaluable and 1 had disease exposure was not impacted by ibr; pac geometric mean Cmax (2066 ng/mL)
progression. Combined analysis of the entire experience demonstrates a higher and AUC0-‘ (3813 ng$h/mL) were consistent with historical data. Pre-
response rate in MTAP deficient UBC (70%) as compared to MTAP proficient liminary tumor biopsy data showed a nonsignificant trend of higher BTK/
UBC (12.5%). Conclusions: Our preclinical and clinical data demonstrate that ITK/BMX expression in tx responders; analyses are ongoing.
MTAP loss in UBC leads to a state of synthetic lethality when treated with Conclusions: Ibr + pac shows promising results for ORR/OS in pts with aUC
pemetrexed and should be further investigated as a novel biomarker to predict previously treated with PBCT and ICIs. No unexpected safety signals were
response to anti-folate agents. seen; ibr + pac seemed tolerable. Follow-up of these and additional pts will
support the activity and safety of ibr + pac in this study. Clinical trial
information: NCT02599324.

4523 Poster Session (Board #349), Mon, 1:15 PM-4:15 PM 4524 Poster Session (Board #350), Mon, 1:15 PM-4:15 PM
Circulating cell-free DNA (cfDNA) levels and fragmentation pattern can Nivolumab monotherapy in patients with advanced platinum-resistant uro-
distinguish nonmuscle invasive (NMI) from muscle-invasive (MI) and met- thelial carcinoma: Efficacy and safety update from CheckMate 275. First
astatic (met) bladder cancer (BC). First Author: Jaleh Fallah, Cleveland Author: Arlene O. Siefker-Radtke, The University of Texas MD Anderson
Clinic Foundation, Cleveland, OH Cancer Center, Houston, TX
Background: Occult MI and met BC may be under-staged. Circulating cfDNA Background: In the open-label, single-arm, phase 2 CheckMate 275 trial,
may be a dynamic, low-cost and minimally invasive biomarker. We evaluated objective response rate (ORR) for patients (pts) with metastatic urothelial
correlations between total circulating cfDNA and presence of MIBC and met carcinoma (mUC) with nivolumab (NIVO) was 20.4% with minimum follow-up
BC. We hypothesized that the relative abundance of circulating low molecular of 21.3 mo. Here, we report updated efficacy and safety data with minimum
weight cfDNA would correlate with BC stage. Methods: Peripheral blood from follow-up of 33.7 mo. Methods: Pts with platinum-resistant locally advanced or
pts with BC was collected in Streck BCT tubes and processed to obtain cf metastatic urothelial carcinoma received NIVO 3 mg/kg until disease pro-
nucleic acid extracts. Total cfDNA quantity (ng/ml) was assessed by fluo- gression or unacceptable toxicity. The primary endpoint was ORR by blinded
rimetry. cfDNA fragment size was measured by Bioanalyzer DNA analysis. independent review committee (BIRC) by RECIST v1.1 (including duration of
Wilcoxon rank sum test and Fisher’s Exact test were used to compare cfDNA response [DOR]). Secondary endpoints included progression-free survival (PFS)
quantity and fragmentation pattern among pts with NMIBC, MIBC, met BC. by BIRC, overall survival (OS), and ORR per investigator. Efficacy was evaluated
Results: Blood was obtained from 58 pts with BC (20% women, 34% never in all treated pts and by tumor PD-L1 expression. Safety and PFS by investigator
smokers, median age 71 (29-89). There was no significant difference in cfDNA were exploratory endpoints. Results: ORR by BIRC was 20.7% (95% CI
between MIBC and met BC, however, it was significantly lower in pts with 16.1–26.1) including 18 (7%) complete responses (CR; with 1 additional CR
NMIBC vs MIBC and met BC (table). The concentration of low molecular since the last report; Table). ORR per investigator was similar (24.8%). Median
weight fragments (LMW-frags) (100 - 400) base pairs and the ratio of LMW- DOR by BIRC was 20.3 mo (95% CI 11.5–31.3). Of 56 pts with best overall
Frag to cfDNA were significantly different between pts with NMIBC and pts response (BOR) of CR or partial response (PR), 59% had a DOR $12 mo.
with MIBC or met BC (table). Using median values as the cutoff, there was a Median PFS (mPFS) was 1.9 mo per BIRC (95% CI 1.9–2.3; Table) and 2.0 mo
significantly higher proportion of pts with cfDNA . 7 ng/ml and LMW-frags per investigator (95% CI 1.9–2.5). Median OS (mOS) was 8.6 mo (95% CI
. 1.6 ng/mL, in MIBC & met BC vs NMIBC (p , 0.001). The % of pts with 6.1–11.3; Table). 12, 24, and 36-mo OS rates were 40%, 30%, and 22%.
LMW-frags to cfDNA . 30%, was significantly different among NMIBC, MIBC and While efficacy was numerically higher in pts with tumor PD-L1 expression $1%,
met BC groups: 16%, 53%, 78%, respectively (p , 0.001). Conclusions: This efficacy was observed in all pts (Table). Any-grade treatment-related adverse
exploratory study suggests that cfDNA levels may correlate with BC stage. events occurred in 69% of pts (grade 3–4, 25%), mostly (59%) within the first
Measuring the relative abundance of LMW-frags with the expected size of cf 3 mo of initiating therapy. Conclusions: With long-term follow-up from CheckMate
DNA can enhance the specificity of cfDNA analysis for distinction between 275, NIVO continues to provide durable antitumor activity in pts with mUC. No
MIBC and met BC. Further studies are needed to confirm findings and define new safety signals were noted. Clinical trial information: NCT02387996.
the optimal cut-points for optimal BC staging. All treated pts PD-L1 < 1% PD-L1 ‡1%
N = 270 N = 146 N = 124
NMIBC MIBC met BC P-value
(N 19) (N 15) (N 24) (NMIBC vs MIBC & met BC) ORR (95% CI) 20.7 (16.1–26.1) 16.4 (10.8–23.5) 25.8 (18.4–34.4)
BOR by BIRC, n (%)
Median cfDNA (ng/ml), 0.7 9.7 9.3 , 0.001 CR 18 (7) 6 (4) 12 (10)
(IQR) (0.4,1.2) (4.0,13.3) (5.9,15.3) PR 38 (14) 18 (12) 20 (16)
Median LMW-frags (ng/ml), 0.0 3.3 3.8 , 0.001 Stable disease 56 (21) 24 (16) 32 (26)
(IQR) (0.0,0.0) (0.0,8.4) (1.4,31.5) Progressive disease 111 (41) 71 (49) 40 (32)
% LMW-Frags to cfDNA, 0.0 34.4 50.3 , 0.001 mPFS by BIRC (95% CI), mo 1.9 (1.9–2.3) 1.9 (1.7–2.0) 3.5 (1.9–3.7)
(IQR) (0.0,0.0) (0.0,60.8) (31.2,79.3) mOS (95% CI), mo 8.6 (6.1–11.3) 6.0 (4.4–8.1) 11.9 (9.1–19.1)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
270s Genitourinary (Nonprostate) Cancer

4525 Poster Session (Board #351), Mon, 1:15 PM-4:15 PM 4526 Poster Session (Board #352), Mon, 1:15 PM-4:15 PM
Outcomes of patients (pts) with metastatic urothelial cancer (mUC) and poor Treatment (tx) characteristics of patients (pts) with locally advanced or
performance status (PS) receiving anti-PD(L)1 agents. First Author: Ali Raza metastatic urothelial cancer (mUC) receiving checkpoint inhibitor (CPI)
Khaki, University of Washington, Seattle, WA monotherapy in a US clinical practice. First Author: Alicia K. Morgans,
Northwestern University, Chicago, IL
Background: Anti-PD(L)1 immune checkpoint inhibitors (ICI) prolong overall
survival (OS) after platinum chemotherapy in mUC. However, clinical out- Background: Approval of anti–PD-L1/anti–PD-1 CPI agents has changed the
comes in pts with poor PS at time of ICI initiation are unknown. We hy- mUC tx landscape, but real-world (RW) tx patterns are not well described. Here,
pothesized that ICI initiation in pts with ECOG PS 2-3 would be associated we describe pt characteristics, time on tx (TOT), tx-cycle distribution, relative
with worse outcomes vs. pts with ECOG PS , 2, and impact death location. dose intensity (RDI) and subsequent tx for pts receiving atezolizumab (atezo),
Methods: A retrospective cohort study in 8 institutions identified pts with mUC nivolumab (nivo) or pembrolizumab (pembro) monotherapy. Methods: Pts
who received ICI. Demographic, clinicopathologic, treatment (tx) patterns, tx diagnosed with mUC who completed atezo, nivo or pembro in the first-line (1L)
response, and outcomes were collected. Primary endpoint: overall response or prior-platinum second-line and beyond (2L+) settings by April 30, 2018,
rate (ORR). Secondary endpoints: median (m) OS in pts receiving ICI as 1st and were identified from the US-based Flatiron Health electronic health record–
2nd line (1L, 2L); odds of dying in hospital (vs elsewhere) for pts receiving ICI derived database. TOT was defined as time from first to last CPI administration +
(vs no tx) within 30 days of death; and estimated drug cost for pts with ICI 1 cycle, tx cycles as number of CPI doses received during TOT and RDI as ratio
within 30 days of death based on average wholesale price. Unadjusted logistic of actual to planned dose per week to reflect any dose interruption.
regression was used to assess association between ORR and ECOG PS (2-3 Results: RW data from pts receiving atezo, nivo and pembro were analyzed
vs , 2) and wald test was used to compare mOS between ECOG PS (2-3 vs , (Table). Up to 38% of pts had ECOG PS . 1. Median TOT ranged from 2.1-2.8
2). Results: 194 consecutive pts (30% women, 41% never smokers, median mo, with overlapping 95% CIs; mean TOT ranged from 2.7-4.1 mo. Over 50%
age at diagnosis 69) treated with ICI for mUC were identified. Median number of pts had # 4 tx cycles. 21%-38% of pts did not have RDI within 95%-105%
of total tx lines was 2; all pts received $1 ICI line (6 pts received 2 ICI lines); of the labeled dose. Most common subsequent txs were platinum-based
97, 79, 17 and 7 pts received ICI in 1L, 2L, 3L and 4L, respectively; 26% pts chemotherapy combinations with gemcitabine or taxanes (post–1L CPI) and
with ICI in 1L and 2L had ECOG PS 2-3. ORR and mOS are shown in table. taxane monotherapy or other CPI monotherapy/combinations (post–2L+ CPI).
Among 106 pts who died, 96 had available death location; of those, 8% Conclusions: Here, we present the largest analysis of RW CPI use in mUC to
received ICI within 30 days of death. Starting ICI within 30 days of death (vs no date. Overall, this unadjusted descriptive analysis showed relative compara-
tx) was associated with higher odds of hospital death (OR 6.05, 95%CI 1.3- bility of pt and tx characteristics and TOT across CPI-treated groups. Insights
27.6). Estimated average ICI cost/pt within 30 days of death was $1400.58. into RW tx allow for an understanding of how clinical trial data translate to
Conclusions: Pts with ECOG PS 2-3 at time of ICI initiation had similar ORR vs broader pt populations, including those with ECOG PS . 1, and may be useful
ECOG PS , 2 but worse mOS. ICI initiation within 30 days from death was for practitioners.
associated with higher likelihood of hospital death. ICI may not circumvent the Atezo, 1La Atezo, 2L+ Nivo, 2L+ Pembro, 1La Pembro, 2L+
negative prognostic role of poor PS, so biomarker-based pt selection is critical.
n 79 174 55 34 27
Limitations include lack of adjustment for selection bias and other con- ‡ 65 years, % 86 84 69 88 78
founders at time of ICI initiation; data validation is ongoing. ECOG PS > 1, %b 38 19 6 33 20
Median TOT, mo (95% CI) 2.1 (0.7, 4.2) 2.8 (1.4, 4.9) 2.0 (1.2, 3.6) 2.8 (1.4, 3.8) 2.7 (1.4, 3.5)
ECOG PS 1L ORR 1L mOS (mo) 2L ORR 2L mOS (mo) Mean TOT, mo (SD) 2.9 (2.9) 4.0 (3.8) 2.7 (2.1) 2.9 (1.9) 2.9 (1.8)
> 4 tx cycles, % 35 43 49 41 30
<2 28% 14 23% 16 95%-105% RDI, % 76 79 62 68 70
2-3 27% 6 35% 5 a
Unknown cisplatin eligibility and prior (neo)adjuvant platinum use. b % based on pts with known ECOG
p value NS 0.002 NS , 0.001 PS.

4527 Poster Session (Board #353), Mon, 1:15 PM-4:15 PM 4528 Poster Session (Board #354), Mon, 1:15 PM-4:15 PM
Durability of complete response (CR) with atezolizumab (atezo) in locally Association of cell-free DNA (cfDNA) levels with myeloid-derived suppressor
advanced/metastatic urothelial carcinoma (mUC). First Author: Yohann cells (MDSC) levels in blood of patients (pts) with muscle invasive (MI) and
Loriot, Institut de Cancérologie Gustave Roussy, Villejuif, France metastatic (met) bladder cancer (BC). First Author: Jaleh Fallah, Cleveland
Clinic Foundation, Cleveland, OH
Background: Atezo (anti–PD-L1) has been shown to elicit CRs in a number of
mUC patients (pts) in clinical trials. We sought to describe the kinetics, Background: cfDNA can be detected in healthy individuals but higher con-
durability and outcomes associated with these CRs in Ph I (PCD) and II centrations are present in pts with cancer. MDSC are immature immuno-
(IMvigor210) atezo studies, each with long-term follow-up. Methods: In PCD suppressive cells that can be mobilized from bone marrow by tumor-related
(pre-treated mUC) and IMvigor210 (Cohort 1, cisplatin-ineligible untreated factors. Higher blood MDSC levels have been associated with worse outcomes
mUC; Cohort 2, platinum-treated mUC), pts received atezo per protocol in pts with solid tumors including BC. We assessed correlations between
(Petrylak JAMA Oncol 2018; Balar Lancet 2017; Rosenberg Lancet 2016). cfDNA and MDSC levels in pts with MIBC and met BC. Methods: Peripheral
This post hoc analysis descriptively assessed pt disposition, time to and duration blood from pts with MIBC and met BC was collected in Streck BCT tubes and
of RECIST 1.1 response and overall survival in pts with CR. Results: CR rates processed to obtain cf nucleic acid extracts. Total cfDNA was determined by
were 13%, 8% and 7% in PCD, IMvigor210 Cohort 1 and Cohort 2, respectively. fluorimetry. Cell-free DNA fragment size was measured by Bioanalyzer DNA
First response was PR in most pts with CR. Median CR duration was . 3 y in analysis; 100-400 bp fragments (mono- and di-nucleosomal fragments linked
PCD, not estimable (NE) in IMvigor210 Cohort 1 and . 2 y in Cohort 2 (Table). to granulocytic processing of apoptotic and necrotic tumor cells) were des-
At data cutoff, all but 2, 0 and 1 pts were alive, respectively; across studies, ignated low molecular weight (LMW-frags). The % of MDSC (CD33+/HLADR-)
$ 40% of pts with CR were on treatment. CR pts had a first response (PR/CR) by a and subtypes were measured. MDSC subtypes were defined as poly-
median of 3.5 cycles. Further pt characteristics and survival outcomes will be morphonuclear (PMN-MDSC: CD15+/CD14-), monocytic (M-MDSC: CD15-/
reported. Conclusions: Across Ph I/II atezo mUC studies, CRs appeared durable CD14+) and uncommitted (UNC-MDSC: CD15-/CD14-). Spearman’s corre-
(median duration . 2 y) despite small pt numbers. Most pts with CR were alive, lation test was used for analysis. Results: Blood from 37 pts (19% women,
with responses ongoing after long-term follow-up (median follow-up . 30 mo). 40% never smokers) with MIBC or met BC was collected: 15 (41%) with MIBC
Clinical trial information: NCT01375842, NCT02951767, NCT02108652. and 22 (59%) with met BC at time of collection. There was a significantly
IMvigor210 Cohort 1 IMvigor210 Cohort 2
positive correlation between total MDSC and cfDNA levels (r = 0.57, P =
PCD (n = 12)a (n = 10)b (n = 22)c 0.0003). Among MDSC subtypes, there was a significantly positive correlation
Median atezo duration (range), mo 33.6 (10.4- 21.5 (6.2-35.9) 32.6 (8.8-35.6) between PMN-MDSC and cfDNA levels (r = 0.61, P , 0.0001). The higher
44.4) level of LMW-frags was significantly but moderately associated with higher
Median time to first response (range), mod 1.4 (1.2-8.3) 2.2 (2.0-10.3) 2.0 (1.9-8.2)
Median doses to first response (range), n 2.0 (2-12) 3.5 (3-15) 3.0 (3-12)
total MDSC (r = 0.43, P 0.008) and PMN-MDSC (r = 0.41, P 0.01) levels.
Median time to CR (range), mo 8.1 (1.2-28.3) 6.2 (2.1-13.8) 4.2 (1.9-13.2) There was no significant correlation between cfDNA level and other MDSC
Median CR duration (95% CI), mo 37.5 (33.3, NE 28.4 (18.6, NE) subtypes. Conclusions: There was a positive correlation between total and
37.5)
Pts with ongoing CR, n (%)e 8 (67) 8 (80) 12 (55) PMN-MDSC with cfDNA levels in blood from pts with MIBC and met BC. That
Pts with CR as first response, n (%) 2 (17) 3 (30) 8 (36) may suggest a putative role for MDSC in mediating cfDNA release into the
Treatment discontinuations due to PD, n (%) 2 (17) 0 3 (14)
Treatment discontinuations due to non-PD, 2 (17) 6 (60) 3 (14) circulation, consistent with prior reports of granulocyte-mediated ctDNA
n (%)f processing. Further studies need to identify mechanisms and implications of
PD, progressive disease. Data cutoff (median follow-up, mo): a Dec 31, 2016 (40.8); b Jul 12, 2017 (31.7); our findings and potential correlation with clinical outcomes.
c
Jul 12, 2017 (33.1). d First response of CR or PR. e No PD/death. f Excluded 1 PCD pt who completed only
16 cycles per an earlier protocol.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Genitourinary (Nonprostate) Cancer 271s

4529 Poster Session (Board #355), Mon, 1:15 PM-4:15 PM 4530 Poster Session (Board #356), Mon, 1:15 PM-4:15 PM
Molecular biomarker analysis and survival in patients (pts) with advanced Pembrolizumab (pembro) for patients (pts) with high-risk (HR) non–muscle
urothelial cancer (UC) previously treated with chemotherapy. First Author: invasive bladder cancer (NMIBC) unresponsive to Bacillus Calmette-Guérin
Bernadett Szabados, Barts Cancer Centre, Queen Mary University of London, (BCG): Updated follow-up from KEYNOTE-057. First Author: Ronald De Wit,
London, United Kingdom Erasmus University Medical Center, Rotterdam, Netherlands
Background: The biomarkers PD-L1, FOXP3, and CD8 have been explored in Background: Upregulation of the PD-1 pathway has been observed in BCG-
pts with advanced UC who progressed after platinum-based chemotherapy unresponsive NMIBC, suggesting that pembro may be beneficial. Efficacy and
(CTx). However, their relevance earlier in the disease process is less well safety of pembro in pts with HR, BCG-unresponsive NMIBC was evaluated in
understood. Methods: The Phase 2/3 LaMB study (NCT00949455) compared the single-arm phase 2 KEYNOTE-057 study (NCT02625961); updated re-
maintenance lapatinib vs placebo after first-line (1L) platinum-based CTx in sults for pts with carcinoma in situ (CIS) with or without papillary tumors
pts with HER1/HER2-overexpressing stage IV advanced UC. Pre-CTx archival (cohort A) are reported. Methods: Pts with histologically confirmed HR, BCG-
samples from this study were retrospectively analyzed and included both unresponsive CIS with or without papillary tumors who received adequate BCG
randomized and screen failure pts. PD-L1 expression was assessed (VENTANA therapy and were unable/refused to undergo radical cystectomy received
SP263 Assay) and categorized as high ($25% of tumor cells [TC] and/or pembro 200 mg Q3W for 24 mo or until recurrence, progression, or un-
immune cells [IC]) or low/negative ( , 25% TC and IC). Overall survival (OS) acceptable toxicity. Pts who developed HR NMIBC or progressive disease
and progression-free survival (PFS) were estimated via Kaplan-Meier method; during treatment were required to discontinue. Key end points were complete
results were stratified by PD-L1 expression. The exploratory biomarkers CD8 response rate (CRR), duration of response, and safety. Results: 102 pts
and FOXP3 were also analyzed. The prognostic significance of the biomarkers (median age, 73 years; CIS alone, 63.7%; median number of prior BCG in-
was explored by multivariable Cox proportional hazards models and a bootstrap stillations, 12) had enrolled in cohort A as of enrollment cutoff. Median (range)
method for model selection. Results: Of 446 pts (232 randomized; 214 duration of follow-up was 15.8 mo (4.6-28.2); 3-mo CRR was 40.2% (95% CI,
screened), 243 (54.5%) were assessed for PD-L1 expression, with 61 (25.1%) 30.6-50.4) by central assessment. Among 41 pts who had CR at 3 mo, median
PD-L1 high and 158 (65.0%) PD-L1 low/negative. In PD-L1 high and low/ CR duration was 12.7 mo (range, 0+ to 20.5+ mo); 75.4% had a CR duration
negative pts, respectively, median OS (95% CI) was 12.0 (9.4–19.7) vs $6 mo; 52.6% had a CR duration $12 mo (Kaplan-Meier method); 24 pts
12.5 months (10.4–15.5); median PFS (95% CI) was 6.5 (3.5–8.8) vs (58.5%) maintained CR at last follow-up, and 15 (36.6%) experienced re-
5.0 months (4.3–6.3). PD-L1 expression was not associated with OS or PFS in current NMIBC after CR; at the time of analysis, none progressed to muscle-
univariate analysis or in a multivariate model for OS (hazard ratio [HR] for PD- invasive or metastatic disease. CRR was 44.6% for pts with CIS alone (n = 65),
L1 high vs low/negative 1.4 [95% CI, 0.8–2.3]). In a multivariate model for 41.7% for CIS with T1 tumors (n = 12), and 28.0% for CIS with high-grade Ta
PFS, PD-L1 expression improved accuracy of the model by 23% and was a tumors (n = 25). Treatment-related adverse events (AEs) occurred in 66
significant variable (HR, 2.1 [95% CI, 1.2–3.5]). Results of analyses of CD8 (64.7%) pts; most frequent ($10%) were pruritus (10.8%), diarrhea (10.8%),
and FOXP3 will also be reported. Conclusions: Overall, these data suggest a and fatigue (9.8%). Grade 3/4 treatment-related AEs occurred in 13 (12.7%)
lack of association between PD-L1 expression and survival in pts receiving 1L pts. Immune-mediated AEs occurred in 19 (18.6%) pts. Conclusions: Pembro
platinum-based CTx. Mechanisms underlying the potential association of PD- continued to show encouraging antitumor activity in pts with HR, BCG-
L1 expression with PFS remain unclear. CD8 and FoxP3 exploratory analyses unresponsive CIS with or without papillary tumors and a safety profile con-
may help to elucidate these results. Clinical trial information: NCT00949455. sistent with that of previous experience. Updated data using additional follow-up
will be presented. Clinical trial information: NCT02625961.

4531 Poster Session (Board #357), Mon, 1:15 PM-4:15 PM 4532 Poster Session (Board #358), Mon, 1:15 PM-4:15 PM
Impact of immune-related adverse events on survival in patients with meta- Increasing use of neoadjuvant chemotherapy (NAC) in muscle-invasive
stastic urothelial carcinoma treated with immune-checkpoint inhibitors. First bladder cancer (MIBC): Prognostic impact of non-standard of care (SOC)
Author: Rafael Morales-Barrera, Vall d’Hebron Institute, Barcelona, Spain regimens. First Author: Yaw A. Nyame, Department of Urology, University of
Washington Medical Center, Seattle, WA
Background: Immune-checkpoints inhibitors (ICIs) represents the standard
of care for platinum-pretreated advanced urothelial cancer patients (pts). By Background: Cisplatin-based NAC can prolong overall survival (OS) in patients
enhancing T-cell activation, a unique spectrum of inflammatory side effects (pts) with MIBC. Utilization of NAC has increased to about 20% of pts with
has emerged, also known as immune-related adverse events (irAEs). Data MIBC over the last decade. We evaluated NAC utilization with and without SOC
regarding the association between irAEs and pts outcomes are conflicting. cisplatin-based combination regimens and oncologic outcomes using registry
Here we conducted a retrospective analysis to investigate the association data. Methods: This is a population-based analysis of linked SEER-Medicare
between irAEs profile and disease outcome in metastastic urothelial car- data (2004-2011). We identified 4534 pts with MIBC (cT2-4N0-1) undergoing
cinoma (mUC) pts. Methods: Medical records from pts with mUC included in radical cystectomy (RC). Based on pharmacy records data, pts were stratified
clinical trials between July 2013 and June 2018 and treated with ICIs were into 3 groups: SOC, non-SOC, and immediate cystectomy (IC). We used de-
reviewed. Pts previously treated with platinum-based chemotherapy or scriptive statistics to compare groups, and multivariate logistic regression to
cisplatin ineligible pts who had not been previously treated with chemo- define factors associated with receiving SOC NAC. Competing risk bladder
therapy were included. Clinical responses were assessed as complete re- cancer-specific mortality (BCSM) incidence curves were generated and KM
sponse (CR), partial response (PR), stable disease (SD), and progressive analysis was used to assess OS from time of RC. The impact of NAC on OS was
disease (PD) according to RECIST v1.1. Adverse events were graded based evaluated with Cox regression analysis. Results: 694 (15.3%) pts received
CTCAE v4.03. Overall survival (OS) was calculated from the date of initiation NAC, increasing from 11% in 2004 to 24.8% in 2011, with 345 (50%) re-
of ICI to the date of death. X2 test was used to determine differences in rates. ceiving non-SOC, e.g. gemcitabine/carboplatin (49.3%), gemcitabine alone
OS was estimated using Kaplan-Method and long rank test was used to (21.2%), carboplatin alone (14.8%), cisplatin alone (8.4%), and methotrexate/
assess differences between groups. All analyses were performed using SPSS vinblastine/ adriamycin/carboplatin (0.8%). On logistic regression, increasing
v21. Results: From a total of 52 pts, 44 (84.6%) were treated with ICI age (OR 0.91, 95%CI 0.88 – 0.94, p , 0.0001), Hispanic/Latin ethnicity (OR
monotherapy and 8 (15.3%) in combination (anti-CTLA4 or targeted ther- 0.49, 95%CI 0.22 – 1.10, p = 0.08), and $moderate renal dysfunction (OR
apy). Median age was 65 years, 42 pts (80.8%) were male, 44 patients 0.20, 95%CI 0.08 – 0.51, p , 0.001) were associated with lower odds of SOC
(84.6%) had ECOG PS 0-1, 14 pts (26.9%) had liver metastasis. Overall NAC. Non-SOC NAC was associated with higher BCSM (competing risk) and
irAEs were observed in 30 pts (57.7%) and 10 pts (19.2%) developed grade lower OS (KM) vs. IC and SOC NAC. On multivariable analysis, non-SOC NAC
3/4 irAES. Most common grade 3/4 irAEs were diarrhea (6.6%), rash (6.6%) was associated with higher risk of BCSM (HR 1.35, 95%CI 1.06 – 1.72, p =
and hepatitis (6.6%). Disease control rate (CR [26%]+PR[33%]+SD[20%]) 0.01) and lower OS (HR 1.38, 95%CI 1.11 – 1.70, p = 0.003) vs. SOC NAC.
was higher for patients with irAEs compared to those patients who did not Conclusions: About 50% of pts receiving NAC were not treated with SOC
developed irAEs (CR [13.6%]+PR[0%]+SD[22.7%], this difference was regimens. Non-SOC NAC was associated with higher bladder cancer death risk.
statically significant (P = 0.002). Median OS was 11.23 mo (CI 95%, 3.76- This stresses the role of SOC NAC ideally in a multidisciplinary expert setting, as
18.70) for the overall cohort, while median OS was 21.91 mo for those well as the need for timely RC and neoadjuvant clinical trials, including
patients with irAEs compared to 6.47 mo in patients who did not developed cisplatin-ineligible pts.
irAEs (P = 0.004). Conclusions: In this analysis we found that the devel-
opment of irAEs was a strong predictor of improved OS in mUC patients
treated with ICI.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
272s Genitourinary (Nonprostate) Cancer

4533 Poster Session (Board #359), Mon, 1:15 PM-4:15 PM 4534 Poster Session (Board #360), Mon, 1:15 PM-4:15 PM
Adenocarcinoma (ACB), urothelial carcinoma (UCB) and squamous cell Gemcitabine plus carboplatin versus gemcitabine plus oxaliplatin in cisplatin
carcinoma (SCCB) of the bladder: A Comprehensive Genomic Profiling unfit patients with advanced urothelial carcinoma: A randomized phase II
(CGP) Study. First Author: Joseph Jacob, SUNY Upstate Medical Univer- study (COACH, KCSG GU10-16). First Author: Jae-Lyun Lee, Asan Medical
sity, Syracuse, NY Center and University of Ulsan College of Medicine, Seoul, South Korea
Background: We performed a CGP to compare the genomic alterations (GA) in ABC, UBC and Background: We investigated the activity and safety of first-line gemcitabine-
SCCB. Methods: 143 cases of ACB, 2,142 cases of UCB and 83 cases of SCCB were subjected to
oxaliplatin (GemOx) compared with gemcitabine-carboplatin (GCb) in
CGP using a hybrid-capture based assay. Tumor mutational burden (TMB) was determined on 1.1
Mbp of sequenced DNA and microsatellite instability (MSI) was determined on 114 loci. PD-L1 cisplatin-ineligible patients with advanced UCCC. Methods: Treatment naı̈ve,
expression was determined by IHC. Results: ABC patients were younger and more often female than cisplatin-ineligible patients with advanced UCCC were randomly assigned to
UBC and SCCB (P , 0.0001). UBC and SCCB had a higher GA/tumor than ABC (P = 0.01). Un- GemOx [gemcitabine 1,000 mg/m2, oxliplatin 100 mg/m2 on day 1 (D1) every
targetable GA were similar in all 3 groups involving TP53 and KRAS. APC GA were more frequent in 2 weeks] or GCb (gemcitabine 1,000 mg/m2 on D1 and D8, carboplatin AUCC
ABC whereas TERT, CDKN2A/B and DNA-repair genes (ARID1A and KDM6A) more frequently
altered in UBC and SCCB. Targetable MTOR pathway GA (PIK3CA, TSC1, PTEN) were more
4.5 on D1 every 3 weeks) stratified by ECOG performance status (PS) and
frequent in UBC and SCCB as were targetable kinase alterations (FGFR3 and ERBB2). The UBC visceral metastases. The primary endpoint was objective response rate (ORR),
and SCCB had a significantly higher TMB than ABC (P , 0.0001) including mean TMB and TMB . and secondary endpoints were progression-free survival (PFS) and overall
20 mut/Mb (P , 0.0001). CD274 (PD-L1) was amplified more frequently in SCCB than ACB or survival (OS). Results: Between January 2011 and Mar 2017, 80 patients
UBC (P , 0.0001). MSI high status was very uncommon in all tumor types. Conclusions: Deep were enrolled; 39 patients and 40 patients were allocated to GCb and GemOx
sequencing reveals that ABC features a widely different genomic profile from UBC and SCCB. UBC
has the highest frequencies of targetable kinase GA and high TMB. SCCB has the highest fre-
arm, respectively. Median age was 72 years (range 46-84) in GCb arm and
quencies of IO efficacy predicting biomarkers including mean TMB and PD-L1 amp. Nonetheless, 72.5 (55-85) in GemOx arm. ECOG PS was 2 in 41% (GCb arm) and 43%
ABC does feature potential kinase targets such as FGFR3 and ERBB2. (GemOx arm), and median GFR was 45 ml/min (interquartile range 36-56)
Top TMB TMB
in GCb and 47 ml/min (37-56) in GemOx arm. ORR were 48.7% and 55.0%
Tumor Type (M/F)
Age GA/ Un-targetable Top Targetable
MedianRange tumor GA GA MSI High
Mean
TMB
‡10 mut/
Mb
‡20 mut/
Mb
PD-L1 IHC
Positive
in GCb and GemOx, respectively. With a median follow-up duration of
ACB 143 cases 57/86 58 (24-83) 5.4 TP53 79% PIK3CA 10% 2/106 (2%) 2.4 14 (10%) 4 (3%) 2/11 (18%)
37.8 months, median PFS and OS in GCb and GemOx arm were 5.5 months
KRAS 30%
SMAD4 14%
ERBB2 8%
PTEN 4%
(95% CI, 4.8-6.2) vs. 4.4 months (95% CI, 2.7-6.1) and 9.1 months (95% CI,
MYC 13%
CDKN2A
MET 4%
EGFR 4%
5.2-13.0) vs. 11.0 months (95% CI, 6.9-15.0), respectively. Grade 3 leu-
13% kopenia, neutropenia and fatigue were significantly more common in GCb arm
TERT 12% CD274 0%
ARID1A 10% (26% vs. 3%, p=0.003; 33% vs. 10%, p=0.014; 15% vs. 3%, p=0.057)
APC 8%
UCB 2,142 1597/ 67 (19-88) 7.7 TERT 68% PIK3CA 21% 11/1661 9.9 697 243 75/243 (31%) while any grade neuropathy was more common in GemOx (8% vs. 60%,
cases 545 TP53 56% FGFR3 19% (1%) (32%) (11%)
CDKN2A ERBB2 16% p,0.001). Conclusions: GemOx has showed comparable efficacy with GCb
35%
CDKN2B TSC1 9% and favorable hematologic toxicity profile. GemOx may be used as a new option
38%
ARID1A 23% PTEN 5%
for UCCC patients who are not suitable for platinum-containing chemotherapy.
KDM6A 23%
KMT2D 22%
CD274 1% Clinical trial information: NCT01487915.
KMT2D 22%
APC 3%
SCCB 45/38 62 (31-88) 8.2 CDKN2A PIK3CA 43% 1/69 10.4 26 13 7/18
70%
83 cases TP53 70% PTEN 8% (1%) (31%) (16%) (39%)
TERT 62% FGFR3 7%
CDKN2B ERBB2 6%
37%
KMT2D 31% EGFR 6%
APC 5% CD274 5%

4535 Poster Session (Board #361), Mon, 1:15 PM-4:15 PM 4536 Poster Session (Board #362), Mon, 1:15 PM-4:15 PM
Squamous-cell carcinoma variant histology (SCC-VH) in muscle-invasive Defects in DNA repair genes and long-term survival in cisplatin-based neo-
bladder cancer (MIBC): A comprehensive clinical, genomic, and therapeutic adjuvant chemotherapy for muscle invasive bladder cancer (MIBC). First
assessment from multiple datasets. First Author: Marco Bandini, Vita-Salute Author: Benjamin Miron, Fox Chase Cancer Center, Philadelphia, PA
San Raffaele University, Milan, Italy
Background: Although cisplatin-based neoadjuvant chemotherapy (NAC)
Background: Pure or predominant SCC-VH is not uncommon in MIBC. Nev- has demonstrated an overall survival (OS) benefit in MIBC, only a subset of
ertheless, very few data are available about the efficacy of neoadjuvant che- patients have pathologic complete response (pT0) at cystectomy. ATM, RB1
motherapy (NAC). Here, we examined the outcomes after NAC, explored novel and FANCC mutations have shown correlation with pT0 to cisplatin-based
therapeutic targets, and propose new results in these patients (pts) by in- NAC, as previously published. We now report updated OS and disease
tegrating multiple datasets. Methods: Within RISC and San Raffaele databases specific survival (DSS) from two phase II trials using these gene alterations as
(1990-2018), we identified 2858 MIBC pts with urothelial cancer (UC, N = biomarkers. Methods: Patients with stage T2-T4 (N0 or N1) MIBC were
2229) or VH (N = 629) who received RC +/- NAC. Kaplan-Meier and Cox re- enrolled in phase II trials of dose-dense NAC with MVAC (methotrexate,
gression analyses compared cancer-specific survival (CSS) between SCC and vinblastine, adriamycin, and cisplatin; NCT01031420) or GC (gemcitabine
UC with NAC stratification. Logistic regression models tested the odds of and cisplatin; NCT01611662). Patients were treated with NAC with plan for
clinical-to-pathological downstaging (cT . pT). Foundation Medicine (FMI) curative cystectomy. DNA from pretreatment tumor tissue was sequenced for
dataset was queried for SCC-VH. 97 pts were assayed with hybrid-capture based coding exons of 287 cancer-related genes and analyzed for mutations.
comprehensive genomic profiling (CGP). Finally, we looked at the results from Survival in patients with one or more mutations in ATM, RB1, or FANCC
the PURE-01 study, that is now amended and enrolling pts with VH genes was compared to those without mutations. Results: Of 58 pts treated,
(NCT02736266). Results: Overall, 127 (4.4%) had predominant SCC-VH, 157 38% (22/58 pts) had relevant mutations in the combined group of MVAC
(5.5%) UC+SCC. Among the NAC-treated pts, SCC was the only VH (N = 44) (13/34 pts) and GC (9/24 pts) trials. At a median follow-up of 56 months and
significantly associated with worse CSS, (p , 0.001) and higher mortality (HR minimum follow up of 16 months, patients with mutations had statistically
2.10, p = 0.003) vs. UC. After NAC adjustment, SCC-VH showed lower rate of significantly greater OS (p = 0.0043) and DSS (p = 0.0015). Median OS/
downstaging (3.7 vs 9.3%, OR 0.4, p = 0.028) vs. UC. Similar negative trends DSS was not reached for patients with a mutation in any group. At 5 years
were confirmed in pN0 pts, where SCC exhibited worse CSS (p = 0.006) and post treatment, OS/DSS were greater in mutated vs non-mutated patients in
higher mortality (HR 5.15, p = 0.002). In the FMI cohort, the median tumor all groups (see table). Conclusions: Long-term follow up reveals that pre-
mutational burden (TMB) of SCC was 6.25 mut/mb (vs 6.9 mut/mb of 1984 viously reported improved responses to cisplatin-based NAC associated with
UC), 27% of pts having . 10 mut/mb and 14% . 20 mut/mb. Clinically mutations in ATM, RB1 and FANCC also confer a clinically meaningful and
relevant alterations occurred in PIK3CA (42%), CCND1 (15%), PTEN (9.3%), statistically significant survival benefit in these patients. These alterations
FGFR3 (9.3%), and ERBB2 (6.2%). In the PURE-01 study, 13/84 (15.5%) may be useful as predictive biomarkers to allow clinicians to prioritize pa-
SCC-VH pts received pembrolizumab before RC. PD-L1 combined positive score tients most likely to benefit from NAC prior to radical cystectomy.
was $10 in 11/13 pts; results yielded 4 pT0 (30.8%), 10 pT#1 (76.9%), and
no deaths (median FUP: 10.4 mo). Conclusions: We present a comprehensive 5-yr DSS
5-yr OS (mutant) 5-yr OS (non-mut) 5-yr DSS (mutant) (non-mut)
assessment of SCC-VH in MIBC. SCC represents the VH with the lowest activity
of NAC. While CGP revealed multiple opportunities for targeted therapy, the Combined 85% (60.4%, 46% (29.5%, 90% (64.8%, 49% (31.6%,
efficacy of neoadjuvant pembrolizumab in SCC is encouraging. 94.9%) 61.7%) 97.3%) 64.9%)
MVAC 85% (51.2%, 52% (28.7%, 85% (67.1%, 52% (28.7%,
95.9%) 70.4%) 100%) 70.4%)
GC 86% (33.4%, 40% (16.5%, 100% (100%, 47% (19.5%,
97.9%) 62.8%) 100%) 70.1%)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Genitourinary (Nonprostate) Cancer 273s

4537 Poster Session (Board #363), Mon, 1:15 PM-4:15 PM 4538 Poster Session (Board #364), Mon, 1:15 PM-4:15 PM
Bladder Cancer Multidisciplinary Clinic (BCMC) model: Impact on imaging, Proximity to oil refineries and risk of bladder cancer: A population-based
pathology and treatment recommendations. First Author: Brian Winters, analysis. First Author: Tamer Dafashy, The University of Texas Medical Branch
Department of Urology, University of Washington Medical Center, Seattle, at Galveston, Galveston, TX
WA
Background: Exposure to aromatic amines is a risk factor for bladder cancer.
Background: Despite guideline-based standard of care recommendations in BC Incidence rates according to proximity to oil refineries are largely unknown. We
and upper tract urothelial carcinoma (UTUC), treatment remains variable across sought to determine proximity of oil refineries and bladder cancer incidence in
US. Experts recommend focusing BC care in tertiary centers. We hypothesized the State of Texas which is home to the largest number of oil refineries in the
that a BCMC model, with expert central pathology and radiology review, may United States. Methods: We used the Texas Cancer Registry database to
result in changes in corresponding reports, and, thus, treatment recommen- identify patients diagnosed with bladder cancer from January 1, 2001 to
dations. Methods: Our BCMC clinic format includes simultaneous consultation December 31, 2014. The U.S. census data from 2010 was used to ascertain
with Urologic, Medical and Radiation Oncology, with real time expert genito- overall population size, age and sex distributions. Heat maps of the 28 active
urinary pathology and radiology review. We retrospectively assessed the con- oil refineries in Texas were developed. Incidence of bladder cancer were
cordance between outside (pre-BCMC) imaging & pathology review and BCMC compared according to proximity ( , 10 vs. $ 10 miles) to an oil refinery. Risk
review. Differences between pre- and post- BCMC recommendations on ratios were adjusted using a Poisson regression model. Results: A total of
management were also assessed; descriptive statistics were used. Results: We 45,517 incident bladder cancer cases were identified of which 5,501 cases
identified 233 BC/UTUC patients (pts) referred to BCMC. Complete radio- were within 10 miles of an oil refinery. In adjusted analyses, bladder cancer risk
graphic and pathologic data were available for 209 pts. Median age at time of was significantly greater among males vs. females (Relative Risk (RR) 3.41,
evaluation was 68 (27-93) and 85% were PS ECOG 0-1. After BCMC review of 95% Confidence Interval (CI), 3.33-3.50), and greater among people living
outside records, 112 (53.6%) imaging and/or pathology changes were noted, within 10 miles from an oil refinery than those living outside a 10-mile radius
with 57 (27%) pts upstaged. Overall, imaging interpretation was changed in from an oil refinery (RR 1.19, 95% CI, 1.08-1.31). Conclusions: People living
25% of cases, and 20% of pts were upstaged. BCMC pathology review resulted within 10 miles from oil refineries were at greater risk for bladder cancer.
in changes in 59 (28%) pts. Among those, 42 (71%) had histologic subtype Further research into exposure to oil refineries and bladder cancer incidence is
addition or change, 9 (15%) had LVI/CIS status change, and 2 (3.4%) had low warranted.
to high grade conversion. In terms of pathology staging, 7 (12%) were
downstaged, and 5 (8.5%) upstaged. Further diagnostic work-up was recom-
mended in 71/209 (34%) pts, resulting in upstaging in 11/71 (15.5%) of
cases. Pre- and post- BCMC-recommended treatment modality differed in 55/
209 (26%) pts, while a new treatment modality was added in 28/209 (13%)
pts. These recommendations were followed 91.4% of the time (191/209 pts).
Conclusions: BCMC initiation at our institution resulted in imaging and/or
pathology diagnostic changes in almost half of cases, with approximately a
quarter of pts being upstaged. Findings reveal the importance of expert radiology
and pathology review in BC. Further study is needed to confirm the proposed
benefits and impact of BCMC on treatment response and outcomes.

4540 Poster Session (Board #366), Mon, 1:15 PM-4:15 PM 4541 Poster Session (Board #367), Mon, 1:15 PM-4:15 PM
Towards the noninvasive identification of pathologic responders to neoadju- Treatment sequencing of anti-PD-1/PD-L1 and carboplatin (carbo)-based
vant pembrolizumab in muscle-invasive urothelial bladder cancer (MIBC). First chemotherapy (chemo) in cisplatin-ineligible patients (pts) with metastatic
Author: Giuseppina Calareso, Radiology Unit, Fondazione IRCCS Istituto urothelial cancer (mUC). First Author: Xiao X. Wei, Dana-Farber Cancer
Nazionale dei Tumori, Milan, Italy Institute, Boston, MA
Background: The possibility to predict the pathologic complete response (pT0) Background: Anti-PD-1/PD-L1 agents and carbo-based chemo are therapy
or pT#1 after neoadjuvant immunotherapy may deeply impact the manage- options in 1st-line (1L) setting for cisplatin-ineligible pts with mUC. However,
ment of MIBC and orient clinical trials. The PURE01 study (Necchi, JCO optimal sequencing is unclear. Methods: We conducted a multicenter retro-
2018) evaluated preoperative pembrolizumab before radical cystectomy (RC) spective analysis of cisplatin-ineligible pts with mUC treated with 1L PD-1/PD-
in T2-3bN0MO patients (pts). The trial was amended to increase the sample L1 monotherapy followed by carbo-based chemo (IO→Cb) or the reverse order
size. Methods: Pts were assessed with bladder multiparametric magnetic (Cb→IO) without intervening systemic therapy. Perioperative cisplatin-based
resonance imaging (mpMRI: T2-weighted imaging, diffusion-weighted im- chemo was allowed if completed . 1 year from 1L mUC therapy initiation. To
aging, dynamic contrast enhancement [CE]) before and after treatment (3 assess association between overall survival (OS) and therapy sequence, a
cycles every 3 weeks) prior to RC. All mpMRI were made with bladder multivariate analysis (MVA) was performed from initiation of 2L therapy, ad-
catheterization. We tested the following complete response (CR) definition: justed for treatment sequence, time interval between initiation of 1L and 2L
any measurable residual disease (T2W, morphology), but no residual restricted therapies, Hb ( , 10 vs $10 g/dl), ECOG PS (0-1 vs 2-3), and metastatic site
diffusion and no early CE signals. Exams were independently assessed by 2 (LN/soft tissue only vs non-liver vs liver). Results: 146 pts (IO→Cb n = 43,
senior radiologists and tested against the pT0/pT#1 endpoints. Logistic re- Cb→IO n = 103) were evaluable with median age 72, 76% men, 78% ECOG PS
gression models analyzed mpMRI and biomarker data on comprehensive 0-1, 17.8% with liver metastasis. Baseline factors were balanced except for
genomic profiling (CGP) with FoundationONE CDx assay and PD-L1 combined higher proportion of men in IO→Cb group (91% vs 70%, p = 0.01). Median time
positive score (CPS). Results: From 02/17 to 11/18, 84 pts were assessed interval between initiation of 1L and 2L therapy for IO→Cb and Cb→IO were
before and after treatment (168 total mpMRI) and had the pathological re- 15.6mo (4.8-78.1) and 23.0mo (2.1-103.3), respectively. Response rates are
sponse available from RC. Baseline CR (TURB effect) was observed in 14 summarized (Table). On MVA, treatment sequence was not associated with OS
(16.7%) pts. The mean overall percentage of agreement (OPA) across the (HR 1.05, p = 0.85). Site of metastasis was the only factor significantly as-
mpMRI data was 98.2%. The sensitivity of post-therapy (PT) CR for pT0 sociated with OS (p = 0.002). Conclusions: In our retrospective analysis of
response was 0.80 (95%CI: 0.61-0.92); the specificity for pT#1 response cisplatin-ineligible pts with mUC regardless of PD-L1 expression, anti-PD-1/PD-
was 0.94 (95%CI: 0.80-0.99). PT-CR was associated with both pT0/pT#1 L1 followed by carbo-based chemo or the reverse sequence appeared to confer
endpoints (p , 0.001) after adjusting for clinical T-stage and histology. comparable OS. The observed response rates and time interval between ini-
Baseline CPS$10 and TMB$20 mut/Mb were strongly associated with pT0 tiation of 1L and 2L therapy are likely contributed by pt selection, where all pts
response (OR: 4.46 and 14.9). mpMRI CR post-pembrolizumab significantly received 2L. Further investigation of the ‘PD-L1 high’ population is warranted,
improved the c-index of the CPS/TMB models: up to 0.81 and 0.87 for pT0/ given higher response rates with anti-PD-1/PD-L1 vs ‘PD-L1 low’ population.
pT#1 endpoints. Conclusions: In post-pembrolizumab MIBC, the proposed Ongoing phase III trials will help inform optimal sequencing.
definition of radiological CR through mpMRI is endowed with limited inter- IO→Cb (N = 43) Cb→IO (N = 103)
observer variability and high accuracy in predicting pT#1. CPS/TMB-selected
1L 2L 1L 2L
pts who achieve CR with mpMRI can be reliably spared any surgical treatment. Best overall response PD-1/L1 Carbo Carbo PD-1/L1
Clinical trial information: NCT02736266. Unknown 0 (0%) 8 (18.6%) 3 (2.9%) 15 (14.6%)
PD 28 (65.1%) 8 (18.6%) 32 (31.1%) 53 (51.5%)
SD 11 (25.6%) 8 (18.6%) 21 (20.4%) 13 (12.6%)
PR 4 (9.3%) 19 (44.2%) 43 (41.7%) 19 (18.4%)
CR 0 (0%) 0 (0%) 4 (3.9%) 3 (2.9%)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
274s Genitourinary (Nonprostate) Cancer

4542 Poster Session (Board #368), Mon, 1:15 PM-4:15 PM 4543 Poster Session (Board #369), Mon, 1:15 PM-4:15 PM
FGFR-altered, advanced urothelial carcinoma (UC) and response to chemo- Erdafitinib in high-risk patients (pts) with advanced urothelial carcinoma
therapy prior to receiving erdafitinib. First Author: Andrea Necchi, Fondazione (UC). First Author: Se Hoon Park, Samsung Medical Center, Sungkyunkwan
IRCCS Istituto Nazionale dei Tumori, Milan, Italy University School of Medicine, Seoul, South Korea
Background: FGFR-altered, advanced UC has predominantly a luminal 1 Background: The pan-FGFR inhibitor erdafitinib exhibited a robust objective
subtype, which is associated with lower response rates to immunotherapy response rate (ORR) and tolerability among pts with FGFR2/3-altered advanced
and possibly also to chemotherapy. Objective response rates (ORR) for first- UC in the BLC2001 (NCT02365597) phase 2 study (Siefker-Radtke ASCO
line cisplatin-based regimens, such as gemcitabine-cisplatin (gem/cis) and 2018 #4503). Here we report a post hoc subgroup analysis to explore efficacy
methotrexate-vinblastine-doxorubicin-cisplatin (MVAC), historically range among high-risk pts. Methods: The analysis included 99 BLC2001 pts who
between 45-60% and for gemcitabine-carboplatin (gem/carbo) 35-45%. received the optimized dose regimen of 8 mg/d continuous (pharmacodynam-
However, the ORR on chemotherapy for the ~20% of patients with FGFR- ically guided uptitrated to 9 mg/d per serum phosphate). Results for ORR,
altered tumors is unknown. Methods: BLC2001 (NCT02365597) is an on- duration of response (DOR), progression-free survival (PFS), and overall survival
going global open-label phase 2 study of the pan-FGFR inhibitor erdafitinib in (OS) were analyzed by select baseline variables, with high-risk defined as
patients with locally advanced or metastatic UC with specific FGFR2/3 gene age .75 y, ECOG PS 2, hemoglobin ,10 g/dL, visceral metastases, and 2 or 3
alterations. Patients who had received first-line (1L) or second-line (2L) Bellmunt risk factors. Results: Efficacy results (Table) show investigator-
chemotherapy for advanced UC were identified. Investigator-reported ORR assessed ORR .36% and median PFS .5 mo across all subgroups except
(complete + partial responses) and median time to progression (TTP) on these ECOG PS 2. OS data are immature but generally follow the trend of PFS. With the
pretreatments were analyzed. Results: Of 210 patients treated with erdafitinib exception of ECOG 2, there were no differences in G3/4 serious AE proportions by
in BLC2001, 191 had received prior systemic therapy including 184 and 83 subgroup. Conclusions: The post hoc subgroup findings support that erdafitinib
patients who had received 1L and 2L chemotherapy, respectively. ORR were generally provides comparable efficacy in high-risk pts with FGFR-altered ad-
29.3% (54/184; 95% CI 22.8%, 35.9%) to 1L chemotherapy and 24.1% vanced UC as the overall population. Clinical trial information: NCT02365597.
(20/83; 95% CI 14.9%, 33.3%) to 2L chemotherapy. 1L therapy consisted of ORR, %, n/N DOR, mo PFS, mo OS, mo
gem/cis in 94 patients, gem/carbo in 59 patients, and MVAC in 22 patients, (95% CI) (95% CI) (95% CI) (95% CI)
with ORR (95% CI) of 35.1% (25.5%, 44.8%), 25.4% (14.3%, 36.5%), and Primary analysis of all patients 40.4, 40/99 5.59 8.08 13.80
22.7% (5.2%, 40.2%), respectively. In the 2L setting, of 46 patients who had (30.7, 50.7) (4.24, 7.23) (5.65, 8.77) (9.82, NE)
received a regimen containing a taxane (paclitaxel or docetaxel) or vinflunine, 8 Age 38.6, 32/83 5.59 7.39 12.02
<75 y (28.1, 49.9) (4.21, 7.00) (5.55, 8.51) (8.97, NE)
patients (17.4%; 95% CI 6.4%, 28.3%) achieved an objective response. ‡75 y 50.0, 8/16 13.37 14.75 14.03
Median TTP was 7.16 mo (95% CI 6.18, 7.49) after 1L chemotherapy (7.6 mo (24.7, 75.3) (2.56, 13.37) (3.98, 14.75) (3.98, NE)
ECOG 42.4, 39/92 5.98 8.08 14.03
for gem/cis, 6.3 mo for gem/carbo, and 5.3 mo for MVAC) and 4.35 mo (95% 0-1 (32.1, 53.1) (4.24, 7.39) (5.65, 11.07) (10.71, NE)
CI 3.3, 5.5) after 2L chemotherapy (3.6 mo for taxane or vinflunine). 2 14.3, 1/7 2.79 4.17 5.13
(0.4, 57.9) (NE, NE) (NE, NE) (2.99, 8.02)
Conclusions: In this post-hoc analysis, the overall ORR to prior 1L chemo- Hb 53.3, 8/15 5.98 8.31 NE
therapy was lower, but within the range expected based on historical data. <10 g/dL (26.6, 78.7) (2.96, NE) (4.34, NE) (5.98, NE)
Further investigation into the response to chemotherapy in FGFR alteration ‡10 g/dL 38.1, 32/84 5.55 7.39 13.80
(27.7, 49.3) (4.21, 7.39) (5.62, 8.77) (9.82, NE)
positive patients is warranted and may be useful for the development of 1L Visceral metastases 38.5, 30/78 5.98 8.25 13.80
trials of combination therapy. Clinical trial information: NCT02365597. Y (27.7, 50.2) (4.21, 13.37) (5.62, 14.75) (8.02, NE)
N 47.6, 10/21 4.57 5.95 14.03
(25.7, 70.2) (2.96, NE) (4.34, NE) (10.71, NE)
Bellmunt risk factor 41.6, 32/77 5.55 7.39 14.03
0 or 1 (30.4, 53.4) (4.21, 7.39) (5.62, 8.77) (10.71, NE)
2 or 3 36.4, 8/22 5.98 8.31 10.32
(17.2, 59.3) (2.79, NE) (4.17, NE) (5.13, NE)

4544 Poster Session (Board #370), Mon, 1:15 PM-4:15 PM 4545 Poster Session (Board #371), Mon, 1:15 PM-4:15 PM
Prognostic value of sequential 18F-FDG + Na18F PET/CT (NaF+FDG PET) in Analysis of EGFR mutant urothelial carcinoma (UC) reveals distinct muta-
metastatic genitourinary (GU) cancer patients (pts) treated with cabozantinib/ tional landscape. First Author: Russell Madison, Foundation Medicine, Inc.,
nivolumab +/- ipilimumab (CaboNivoIpi). First Author: Nicholas Peter Verdini, Cambridge, MA
National Cancer Institute at the National Institutes of Health, Bethesda, MD
Background: Genomic alterations (GA) of EGFR, are well recognized as
Background: NaF+FDG PET imaging are used to assess soft tissue and bone druggable oncogenic drivers in NSCLC, but the druggable GA EGFR L858R
metastases. The prognostic value of NaF+FDG PET in GU cancer pts was and exon 19 deletion (ex19del), are rarely observed in genitourinary cancer.
assessed as a secondary endpoint within a phase I trial of combination We reviewed the genomic landscape of advanced upper tract and bladder
CaboNivoIpi. Methods: NaF+FDG PET scans were collected at baseline and UC (UTUC and BUC) to assess the frequencies of druggable EGFR GA.
cycle (C)3 day (D)1 for 50pts. Up to 50 lesions/patient were analyzed at Methods: Tissue specimens from patients with UTUC (407) and BUC (2,402)
baseline, up to 10 lesions/organ in case of extensive disease. Lesion number were assayed using hybrid capture-based comprehensive genomic profiling
and whole-body metabolic tumor volume (wMTV) were recorded at baseline, (CGP). Tumor mutational burden (TMB) was determined on up to 1.1 Mbp of
C3D1 and percent change for FDG and NaF scans. Whole-body total lesion sequenced DNA and microsatellite instability (MSI) was determined on 95 or
glycolysis (wTLG) and its percent change was obtained for FDG scans. Pa- 114 loci. Results: EGFR alterations (EGFRalt) were present in 17 UTUC and 93
rameters were evaluated with respect to OS using Kaplan-Meier and log-rank, BUC (4.2% and 3.9%). Age distribution between the two subgroups was
with quartiles, then refined to show strongest distinction, adjusting p-values to similar, but UTUC was more prevalent in female patients (47% v 29%). BUC
account for this exploration. Parameters with strongest OS association were had a higher median TMB (5.2mut/mb v 7.8 mut/mb; p = 0.046) and the
also analyzed for associations with OS in urothelial carcinoma (UC) pts. prevalence of MSI-H cases was not significantly different. TERT (55% v 71%)
Results: 50 pts, (UC (n = 20); others (renal cell, prostate, urachal/ and TP53 (59% v 74%) were the most frequently mutated genes in EGFRalt
adenocarcinoma, germ cell, penile, bladder squamous cell (n = 2-7 each)). UTUC and BUC. Within EGFRalt, amplifications were the most common al-
Median (m) overall survival (OS) was 23.9 months (mo) (95% CI: 13.7mo – terations in both UTUC and BUC (13/17, 76%; 57/93, 61%). Amplifications
NE) with 29.7mo m potential follow up and mOS of 24.7mo (95% CI: 13.7mo- were mutually exclusive from cases with EGFR short variants (SV) in BUC (0/
NE) for UC. For FDG in all pts, wMTV: baseline # vs . 51.6, mOS (NR vs 34, 0%), and co-occurred with EGFR SV in four UTUC cases (4/8 50%). The
10mo, p = .0006), C3D1 # vs . 85 mOS (25.9mo vs 5.1mo, p = .0001), majority of EGFR SV were EGFR exon 20 insertions (EGFRexon20), which made
percent change (0/increase vs decrease, mOS 14mo vs 25.9mo, p = .015); up a larger proportion of EGFRalt in UTUC than BUC (7/17, 41% v 13/93,
wTLG: baseline # vs . 178, mOS (NR vs 11.5mo, p = .011), C3D1 # vs . 14%; p = 0.01). Compared to other EGFRalt, EGFRexon20 trended towards
300, mOS (25.9mo vs 8.3mo, p , .0001), percent change decrease vs in- mutual exclusivity of GA in commonly altered UC genes: TP53 (UTUC
crease, mOS (25.9mo vs 14.0mo, p = .016); and lesion number: baseline # EGFRexon20 v EGFRalt other: 0% v 100%, p = 5.1E-5; BUC EGFRexon20 v
vs . 13, mOS (25.9mo vs 9.9mo, p = .0090), C3D1 # vs . 13 mOS (25.9mo EGFRalt other: 0% v 86%, p = 2.2E-7), PIK3CA (14% v 10%; 0% v 19%),
vs 9.9mo, p , .0001) significantly predicted OS. In UC pts, wMTV percent RAF1 (0% v 10%; 0% v 16%), or FGFR3 (0% v 10%; 0% v 6.3%) alterations.
change (0/increase vs decrease, mOS 14mo vs. 25.9mo, p = .057), wTLG Only 2.2% (2/93) of BUC EGFRalt were L858R mutations and none
percent change decrease vs increase, mOS (NR vs 8.4mo, p = .0015), and were ex19del (0/93), while neither mutation was detected in UTUC.
lesion number C3D1 # vs . 13 mOS (25.9mo vs 2.8mo, p = .022) signif- Conclusions: EGFRexon20 defines a subset of UC cases, and is the most
icantly predicted OS. NaF parameters failed to do so. Conclusions: FDG wMTV common non-amplification GA seen in EGFR in UC, with some enrichment in
and wTLG at baseline, C3D1, and percent change and lesion number at UTUC. Consideration should be given to developing a trial for EGFR exon20 UC
baseline and C3D1, predicted OS in GU cancer pts on CaboNivoIpi. Clinical patients given the recent investigational work on inhibitors with activity against
trial information: NCT02496208. EGFRexon20, such as poziotinib and TAK-788.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Genitourinary (Nonprostate) Cancer 275s

4546 Poster Session (Board #372), Mon, 1:15 PM-4:15 PM 4547 Poster Session (Board #373), Mon, 1:15 PM-4:15 PM
KEYNOTE-052: Phase 2 study evaluating first-line pembrolizumab (pembro) in Interim analysis of the fierce-21 phase 2 (P2) study of vofatamab (B-701), a
cisplatin-ineligible advanced urothelial cancer (UC)— Updated response and selective inhibitor of FGFR3, as salvage therapy in metastatic urothelial carci-
survival results. First Author: Peter H. O’Donnell, The University of Chicago, noma (mUC). First Author: Begona Mellado, Hospital Clinic de Barcelona,
Chicago, IL Barcelona, Spain
Background: Initial results of the phase 2 KEYNOTE-052 (NCT02335424) Background: Patients (pts) with mUC with FGFR3 mutations who have failed
study led to approval of pembro for cisplatin-ineligible patients (pts) with platinum-based chemotherapy have a poor prognosis. Their response to im-
advanced UC. Updated results representing follow-up of over 2 y since last pt mune checkpoint inhibitors appears diminished 10% or less compared to WT
enrolled are presented. Methods: Pts had confirmed advanced UC, were pts. 20% of mUC pts harbor FGFR3 mutations or fusions (M/F). Vofatamab is a
cisplatin-ineligible (ECOG PS 2, CrCl $30 to ˂60 mL/min, grade $2 fully human monoclonal antibody against FGFR3 that blocks activation of the
neuropathy/hearing loss, NYHA Class III heart failure), and received no prior wildtype and genetically activated receptor. FIERCE-21 is a Phase 1b/2 study
chemotherapy for metastatic disease. Pts received pembro 200 mg IV Q3W designed to evaluate vofatamab monotherapy (VFM) or in combination with
until progression, unacceptable toxicity, withdrawal, or 24 mo of therapy, docetaxel (VFD). Methods: The P2 expansion enrolled mUC pts with FGFR3
whichever occurred first. Primary end point was confirmed ORR (RECIST v1.1, M/F+ tumor (identified with FoundationONE CDx™), who failed $ 1 prior line
independent central review). Key secondary end points: duration of response of chemotherapy (including prior taxane for pts receiving VFM) or recurred
(DOR), overall survival (OS), and safety. Data cutoff was September 26, 2018. # 12 months of (neo)adjuvant chemotherapy. Pts had measurable disease and
Results: Among pts assessed (N = 370), median age was 74 y, 85% had ECOG # 1. Treatment consisted of vofatamab at 25 mg/kg alone and in
visceral disease, and 30% were PD-L1 positive (combined positive score combination with docetaxel at 75 mg/m2 q3w. Efficacy was assessed by in-
[CPS] $10). Median follow-up was 11.4 mo (range, 0.1-41.2) for all pts and vestigators (RECIST 1.1). Primary objectives were safety and objective
29.3 mo (range 7-41.2) for responders. Confirmed ORR was 29% (95% CI, response-rate (ORR). Results: In the P2, 21 pts each received VFM and VFD.
24-34): complete response, 9% (n = 33); partial response, 20% (n = 73). 57% of VFD pts had received at least 2, and 71% of VFM at least 3 prior lines of
Median DOR was 30.1 mo (95% CI, 18.1-not reached [NR]); 67% and 52% of therapy. Best response to prior therapy was PD for 67% of VFD and 38% of
pts had DOR $12 and $24 mo, respectively. Median OS was 11.3 mo (range VFM. The safety profile is consistent with previously reported data. TEAEs
9.7-13.1); 12- and 24-mo OS rates were 47% and 31%, respectively. In pts occurring in . 20% of pts were decreased appetite, diarrhea, pyrexia, as-
with CPS ˂10 (n = 251) and $10 (n = 110), respectively, confirmed ORR was thenia, anemia, dyspnea, and fatigue. Most common vofatamab-related TEAEs
20% (95%CI, 16-26) and 47% (95% CI, 38-57). Median DOR for pts with in . 10% of pts were asthenia, diarrhea, decreased appetite and rash; all were
CPS , 10 and $10 was 18.2 mo (95% CI, 9.7-NR) and NR (95% CI, 18.1- Grade 1 or 2. In VFM, only 1 pt had a grade 3 TEAE and no pt discontinued
NR); DOR $24 mo was 45% and 57%, respectively. Median OS for pts with treatment due to an AE. There were no cases of hyperphosphatemia, ocular or
CPS , 10 and $10 was 9.7 mo (95% CI, 7.6-11.5) and 18.5 mo (95% CI, nail toxicity; 1 pt reported grade 2 skin toxicity. For pts receiving VM, median
12.2-28.5); 24-mo OS rates were 24% and 47% respectively. Treatment- age was 70 yrs, ECOG 1 = 67%, Hgb , 10 g/dL 5%, liver metastases 19%.
related adverse events (AEs) occurred in 67% of pts. Most common were Responses have been seen in 7 pts to date including those receiving both VFM
fatigue and pruritus (18% each); 21% were grade $3, including 1 death and VFD. Conclusions: Vofatamab both alone and combined with D in a q3w
(myositis). Conclusions: With extended follow-up, pembro continued to elicit schedule are well tolerated with a low frequency of grade 3 TEAEs. Both VFM
clinically meaningful, durable antitumor activity in cisplatin-ineligible pts with and VFD have demonstrated efficacy in terms of ORR. PFS/OS and DOR data
advanced UC and was more pronounced in those with PD-L1 expression will be presented at 7+ months for VFD and 9+ months for VFM. Clinical trial
CPS $10. Pembro safety profile was as expected. Clinical trial information: information: NCT02401542.
NCT02335424.

4548 Poster Session (Board #374), Mon, 1:15 PM-4:15 PM 4549 Poster Session (Board #375), Mon, 1:15 PM-4:15 PM
Immune correlates of CD73 expression in patients with urothelial carcinoma An FDA analysis of the association between adverse events and outcome in
(UC). First Author: Edwin Lin, University of Utah/Huntsman Cancer Institute, patients with urothelial cancer receiving a programmed death protein 1 or
Salt Lake City, UT programmed death ligand 1 (anti-PD-1/L1) antibody. First Author: Chana
Weinstock, U. S. Food and Drug Administration, Silver Spring, MD
Background: Checkpoint inhibitors have improved outcomes in UC. However,
response rates are low and additional mechanisms of immune evasion need to Background: To assess the relationship between tumor response rate, overall
be ascertained. CD73 (encoded by NT5E) converts extracellular AMP to survival, and the development of related adverse events of special interest
adenosine, which exerts an immunosuppressive effect in the tumor micro- (AESIs) or related immune-mediated adverse events (imAEs) in patients with
environment by inhibiting infiltrating T and NK cells. Utilizing The Cancer urothelial cancer treated with anti-PD-1/L1 antibodies. Methods: We examined
Genome Atlas (TCGA) bladder cancer dataset, we evaluated correlations be- seven trials that led to drug approval and which included 1747 patients with
tween NT5E expression and the immune milieu in UC. Methods: RNA-seq data metastatic or locally advanced urothelial cancer treated with an anti-PD-1/L1
from 411 primary UC tumor samples were obtained from the TCGA. Patients antibody. Five trials enrolled patients who had received prior platinum-based
were split into low, intermediate, and high NT5E expression groups (# -1, -1 to therapy and two enrolled patients who were cisplatin-ineligible. The datasets
1 and $1 standard deviation from the overall mean). A tumor inflammation were searched for AESIs, related AESIs, imAEs, and related imAEs. The re-
signature (TIS) reflecting an inflamed tumor phenotype was calculated based lationship to study drug was determined by the Investigator. Immune-mediated
on the averaged expression of 18 previously validated genes (Ayers et al, adverse events were defined as AESIs treated with topical or systemic corti-
2017). NT5E expression was compared between tumors with high and low TIS costeroids. Results: In these exploratory analyses, a related AESI was reported
scores and among the TCGA molecular subtypes. Abundance of infiltrating in 64% of responding patients and in 34% of patients who did not respond to
immune cell subsets was estimated based on expression of previously iden- the anti-PD-1/L1 antibody while a related imAE occurred in 28% and 12% of
tified 782 immune metagenes and compared between NT5E expression patients who did and did not respond to study drug, respectively. In a responder
groups (Charoentong et al, 2017). The Mann-Whitney U test assessed sta- analysis, an increase in overall survival was seen in patients with related AESIs
tistical significance, and the Bonferroni correction was used to control for false compared to those with no related AESI [hazard ratio (HR) 0.42; 95% CI: 0.37,
discovery rate. Results: NT5E expression was significantly higher in tumors 0.49]. Fifty-seven percent of responding patients with a related AESI reported a
with a high TIS score compared to those with low TIS score (P,0.0001) and related AESI prior to documentation of response. Conclusions: Patients who
correlated with expression of other immune checkpoints such as PD-L1, IDO responded to treatment with an anti-PD-1/L1 antibody were more likely to
and LAG-3 (each P,0.01). Patients with basal/squamous subtype had the report a related AESI or related imAE. This relationship did not appear to be due
highest NT5E expression compared to luminal or neuronal subtypes. High to the increased duration of exposure in responding patients. Systemic corti-
NT5E expression was associated with increased infiltrating NK cells, neu- costeroid use did not appear to affect the duration of response.
trophils, Tregs and decreased Type 2 T helper cells. Conclusions: High ex-
pression of NT5E in UC patients with an inflamed tumor phenotype was
associated with an increase in infiltrating Tregs, and the basal/squamous
subtype. Our findings highlight a potential role of CD73-adenosine pathway
as a mechanism of immune evasion and a novel therapeutic target in UC.
Further studies to assess the clinical impact of NT5E expression on outcomes
in UC patients treated with immunotherapy are needed. AT and NA: equal
contribution.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
276s Genitourinary (Nonprostate) Cancer

4550 Poster Session (Board #376), Mon, 1:15 PM-4:15 PM 4551 Poster Session (Board #377), Mon, 1:15 PM-4:15 PM
PD-L1/PD-1 expression as a predictor of response to BCG in patients with high- A pilot presurgical study evaluating anti-PD-L1 durvalumab (durva) plus
risk non–muscle invasive bladder cancer. First Author: Mathieu Roumiguie, anti-CTLA-4 tremelimumab (treme) in patients (pts) with high-risk muscle-
Department of Urology Institut Universitaire du Cancer de Toulouse – Oncopole - invasive bladder carcinoma (MIBC) who are ineligible for cisplatin-based
Rangueil University, Toulouse, France neoadjuvant chemotherapy (NAC). First Author: Jianjun Gao, The University
of Texas MD Anderson Cancer Center, Houston, TX
Background: Intravesical BCG instillation (IBI) is the gold standard adjuvant
treatment after transurethral resection of the bladder in high risk non muscle Background: In MIBC pts, especially in those with high risk features including
invasive bladder cancer (HR-NMIBC). IBI induce a type of Th1 immune re- lymphovascular invasion, hydronephrosis, T3b disease, or variant histology,
sponse requiring a recruitment of cytotoxic cells. This response is down- cisplatin-based NAC followed by cystectomy improves overall survival as
regulated by the PD-1/PD-L1 checkpoint inhibitors through an inhibition of the compared with cystectomy alone. However, it is estimated that over 50% of pts
action of CD8+ T cells against their target. Our purpose was to assess whether with MIBC are ineligible for cisplatin-containing therapy. Therefore, we pro-
PD-1/PD-L1 expression was associated with IBI response in HR-NMIBC. pose this pre-surgical trial with durva + treme for this population of pts.
Methods: Histologically confirmed HR-NMBC from 5 academic French in- Methods: This is a single-arm, pre-surgical clinical trial with durva + treme in
stitutions which underwent maintenance IBI were retrospectively included. pts with localized, high-risk MIBC (cT2-T4a) who are ineligible for cisplatin-
The following data were collected: pathological stage, grade, concomitant based NAC due to decreased renal function, neuropathy, hearing loss, or heart
carcinoma in situ, number of lesions and size. From a paraffin embedded failure; or refuse cisplatin-based NAC (NCT02812420). Each patient receives
samples of initial resection, a unique dedicated uropathologist quantified durva (1500 mg) plus treme (75 mg) on weeks 1 and 5. Pts then undergo
immunochemistry expression of CD3, CD8, PD-L1 (antibody SP263/ SP142, surgery at week 9-11. Pre- and post-treatment blood and tumor samples are
E1L3N, 28 8) in both tumour cells and tumoral microenvironment. Univariate collected for correlative biological analyses. Results: Twenty eight of 35 pts
and multivariate analyses were performed using Cox proportional hazards have been enrolled on this trial. Twenty-one pts have completed cystectomy as
model. Results: Overall140 patients (median age 66.5 years, range:35-87; of 11/16/19. Of these 21 pts, 9 (43%) had pathologically complete response
sex ratio male vs female:6:1) were included. The distribution of NMIBC tumour (pCR) and two (10%) had pathologic T1N0 (pT1) disease (#pT1N0 rate =
for stage and grade was: Ta 37.2% (n = 52), T1 62.8% (n = 88) and high grade 52%). Fourteen of 21 (67%) had down-staging of disease. Of note, 10 of these
100% (n = 140) respectively. The median number of IBI which were delivered 21 pts had 3-D mass (T3) on exam under anesthesia or clinical T4a disease;
was 12 (range 7-36). The median length of follow-up was 54.24 mo (95% CI = 5 of these 10 pts (50%) had pCR and one (10%) had pT1 disease (#pT1N0
49.91-58.68). Overall, 25 patients (17.9%) had a recurrence/or progression. rate = 60%). Only 5 of 28 (17%) pts developed grade 3 immune related toxicity
The 72mo Disease free survival (DFS) rate was 81.11% (95% CI = 72.20- including hepatitis and amylase/lipase elevation, and two (7%) resulted in
87.41). Using univariate analysis, we found that Age (HR = 1.07; [1.02-1.12] surgery delay for . 30 days. Immune profiling with CyTOF analysis of baseline
p = 0.005), CD3/CD8 ratio (HR per 10 units = 3.43 [1.62-7.23] p = 0.014) peripheral blood indicates that pts with pCR have significantly lower frequency
and PD-L1(HR per 10 units = 1.65 [1.15-2.38] p = 0.046) were associated of a Th2 subset as compared to pts with up-staging of disease. In addition,
with DFS. In multivariate analysis, Age (HR = 1.07 [1.02-1.13] p = 0.009), gene expression profiling analysis of baseline tumor tissues demonstrates a
CD3/CD8 ratio (HR per 10 units = 1.96 [1.28-3.00] p = 0.02) and PD-L1 significantly less immunosuppressive microenvironment in pts with pCR as
expression in tumour cells (HR per 10 units = 3.38 [1.61-7.11] p = 0.01) compared to pts with up-staging of disease. Conclusions: Our data indicate
remained significantly associated with DFS. Conclusions: PD-L1 expression in that durva plus treme is an effective and safe neoadjuvant therapy for pts with
tumour cells and the T cells population in tumour microenvironment were both MIBC ineligible for cisplatin-based therapy. Therefore, neoadjuvant therapy
predictive factors of BCG response in HR-NMIBC. These results build a sci- with durva + treme and a number of potential biomarkers warrant testing in a
entific rationale for pharmacological intervention on a molecular target using larger phase 3 trial. Clinical trial information: NCT 02812420.
immuno-oncology.

4552 Poster Session (Board #378), Mon, 1:15 PM-4:15 PM 4553 Poster Session (Board #379), Mon, 1:15 PM-4:15 PM
5-factor prognostic model for survival of patients with metastatic urothelial Correlation of circulating tumor DNA (ctDNA), tissue-based genomic profiling
carcinoma receiving three different post-platinum PD-L1 inhibitors. First and clinical efficacy in the biomarker directed Ph1b trial in metastatic bladder
Author: Guru Sonpavde, Dana-Farber Cancer Institute, Boston, MA cancer (BISCAY). First Author: Danielle Carroll, Translational Science, On-
cology, IMED Biotech Unit, AstraZeneca, Cambridge, United Kingdom
Background: A prognostic model for overall survival (OS) of metastatic uro-
thelial carcinoma (mUC) was previously reported in the setting of post- Background: BISCAY is a biomarker-directed Ph1b multi-arm platform study
platinum atezolizumab (Pond GR, GU ASCO 2018). This model was lim- exploring the combination of targeted therapies with anti-PD-L1, Durvalumab,
ited by employing only atezolizumab treated patients (pts), small size of the in advanced urothelial cancer. Methods: Next generation sequencing (NGS) of
validation dataset and unclear applicability to other PD-1/L1 inhibitors. tumour tissue samples from . 380 patients(pts) was performed using the
Hence, we constructed a robust prognostic model utilizing the combined FoundationOne assay alongside IHC for PD-L1. ct DNA from pts enrolled in
atezolizumab cohort as the discovery dataset and used 2 separate validation trial modules at treatment initiation was profiled using the Guardant Health
datasets comprised of post-platinum avelumab or durvalumab treated pts. OMNI platform assessing a panel of 500 genes. For a subset of pts, serial
Methods: The discovery dataset consisted of pt level data from 2 phase I/II plasma samples were also analysed to monitor early signs of response vs.
trials (IMvigor210 and PCD4989g) evaluating atezolizumab (n = 405). Pts resistance and changes in ct DNA dynamics using a bespoke NGS panel of 10
enrolled on 2 separate phase I/II trials, EMR 100070-001 that evaluated post- genes. Results: To date 149 pts have been actively enrolled across 7 different
platinum avelumab (n = 242) and CD1108 that evaluated durvalumab (n = biomarker selected and unselected treatment modules. Across all screened pts
189) comprised the validation datasets. Cox regression analyses evaluated the the most prevalent genomic alterations in tumour tissue were TERT promoter
association of candidate prognostic factors with OS. Factors were dichoto- (65%), TP53 (59%), KMT2D 21%, KDM6A 21%, with the most common CNV
mized and laboratory values were normalized by logarithmic transformation. CDKN2A/B loss (32 %). All enrolled pts tested had detectable ctDNA in
Stepwise selection was employed to propose an optimal model using the plasma. Similar genomic alterations, both frequency and type, were detected
discovery dataset. Discrimination and calibration were assessed in the ave- in both plasma ctDNA and tumour tissue with high concordance for module
lumab and durvalumab datasets following the validation procedure by Royston specific biomarkers used for patient allocation (80% (8/10) for ATM, BRCA1
and Altman (2013). Results: The 5 factors included in the optimal prognostic and 2). Alterations in putative biomarkers predictive of response to anti-PD-L1,
model in the discovery dataset were ECOG-PS (1 vs. 0; HR 1.80; 95% CI such as HRR/MMR alterations and high bTMB levels ( . 20mut/Mb) were
[1.36-2.36]), presence/absence of liver metastasis (HR 1.55; 95% CI [1.20- observed in22% and 40% patient plasma samples, respectively. Correlations
2.00]), number of platelets (HR 2.22; 95% CI [1.54-3.18]), neutrophil- between biomarkers across modules treatment efficacy have been explored.
lymphocyte ratio (NLR; HR 1.94; 95% CI [1.57-2.40]) and lactate Conclusions: All pts with advanced bladder cancer enrolled on BISCAY who
dehydrogenase (LDH; HR 1.60; 95% CI [1.28-1.99]). There was robust were plasma profiled had detectable ctDNA; frequencies of genomic alter-
discrimination of survival between low, intermediate and high-risk groups ations (in both tumour tissue and plasma) were comparable to prior published
based on 0-1, 2-3 and 4 factors. The concordance of survival was 0.692 in the data sets. ctDNA may be an attractive alternative to tissue-based NGS, pro-
discovery and 0.671 and 0.775 in the avelumab and durvalumab validation viding comprehensive dynamic snapshots of genomic landscapes at the start
datasets, respectively. Acceptable or good calibration of expected 1-year and during therapy, and warrants further prospective investigation in trials.
survival rate was observed. Conclusions: A 5-factor prognostic model is
prognostic for survival across 3 different PD-L1 inhibitors (atezolizumab,
avelumab, durvalumab) in this large study totaling 836 pts overall in the setting
of post-platinum therapy for mUC. This model may assist in prognostic
stratification and interpreting nonrandomized trials of post-platinum PD1/L1
inhibitors.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Genitourinary (Nonprostate) Cancer 277s

4554 Poster Session (Board #380), Mon, 1:15 PM-4:15 PM 4555 Poster Session (Board #381), Mon, 1:15 PM-4:15 PM
Recombinant humanized anti-PD-1 monoclonal antibody toripalimab in pa- Circulating tumor cell (CTC) enumeration in patients (pts) with metastatic
tients with metastatic urothelial carcinoma: Preliminary results of an open- genitourinary (mGU) tumors treated in a phase I study of cabozantinib and
label phase II clinical study. First Author: Xinan Sheng, Peking University nivolumab (CaboNivo) +/- ipilimumab (CaboNivoIpi). First Author: Andrea B.
Cancer Hospital, Beijing, China Apolo, Genitourinary Malignancies Branch, Center for Cancer Research,
National Cancer Institute, National Institutes of Health, Bethesda, MD
Background: Patients with advanced metastatic urothelial carcinoma (UC)
who experience disease progression after standard therapy have limited Background: CTCs may serve as biomarkers for clinical outcomes in GU tumor
treatment options. Phase I studies of toripalimab in subjects with heavily pts. We examined the association between baseline CTC enumeration, CTC
pretreated metastatic UC have demonstrated an acceptable safety profile and heterogeneity, CTC morphologic subtypes, and progression-free-survival,
promising clinical activity. Here we report the preliminary safety and efficacy overall survival and response to therapy with combination CaboNivo or
result of toripalimab in a phase II clinical study in Chinese patients with CaboNivoIpi. Methods: 123 samples from 52 pts with mGU tumors treated
refractory/metastatic urothelial carcinoma. (Clinical trial ID: NCT03113266). with CaboNivo (38 pts) or CaboNivoIpi (14 pts) drawn at Baseline, Cycle (C) 2
Methods: Metastatic UC Patients will receive toripalimab, also known as Day (D) 1, and C3D1 were processed using the Epic Sciences platform. CTCs
JS001, 3 mg/kg Q2W until disease progression or unacceptable toxicity. All were defined as cytokeratin (CK)+, CD45-, distinct morphology, intact nu-
patients with measurable disease will be assessed for clinical response every cleus. PD-L1 expression was also assessed. Results: From 07/20/2016-09/
8 weeks according to RECISTv1.1. Tumor PD-L1 expression and tumor 01/2018, 52 pts [urothelial carcinoma (UC) N = 33; plasmacytoid N = 1; Clear
mutational burden will be measured for correlation with clinical response. cell renal cell carcinoma N = 4; bladder adenocarcinoma N = 8; bladder
Results: From May 2017 to February 10, 2019, 79 patients were enrolled from squamous cell carcinoma N = 2; bladder small cell N = 2; renal medullary N =
7 participating centers. The median age was 61 years with 57.5% male. By the 2] were treated. Median age was 61.5 years (range 20-82); 35 (67%) were
cut-off date of Jan 20, 2019, common treatment related AEs were mostly male. N = 37 (71%) had visceral involvement, N = 15 (29%) with liver in-
grade 1 or 2, including anemia, hyperglycemia, ALT increased, AST increased volvement, N = 11 (21%) with bone involvement. CTCs were found in the
and hypothyroidism. Among 65 evaluable patients, 2 complete responses, 18 peripheral blood of 26/40 (65%) pts at baseline. 1/40 pts (bladder adeno-
partial responses, and 13 stable diseases were observed, for an objective carcinoma) had PDL1+ CTCs at Baseline. Median CTC/mL at Baseline, C2D1,
response rate (ORR) of 30.8% and a disease control rate of 50.8%. 70% (14/ C3D1, were not significantly different. CTC counts of . 2 and . 4 at C2D1
20) responses were ongoing by the cut-off date. PD-L1 expression results were were potentially associated with shorter OS (p = 0.071 and p = 0.045 without
obtained from 56 subjects. PD-L1+ patients (n=16, 28.6%) had significant adjustment for multiple cutoffs for evaluation). PFS results exhibited similar
better ORR than PD-L1- patients (n=40), 62.5% versus 15.0% (p,0.01). trends. Unsupervised clustering of CTC images identified 5 main CTC sub-
Conclusions: Toripalimab has demonstrated encouraging clinical activity in types, of which the presence of one was significantly associated with shorter
chemo-refractory UC patients and a manageable safety profile. Toripalimab OS (p = 0.0014, not adjusted for multiple testing). Additionally, we observed a
elicited a favorable 62.5% ORR in PD-L1 positive patients, while PD-L1 trend towards patients with higher CTC heterogeneity at C2D1 having longer
negative patients also achieved a 15% ORR, including one complete response. OS, when adjusting for the risk associated with CTC enumeration (p = 0.084).
Patients will be continuously monitored for additional safety and efficacy Conclusions: CTC were detected in pts with mGU tumors treated with Cabo-
readouts (DOR, PFS and OS). Clinical trial information: NCT03113266. Nivo and CaboNivoIpi. CTC values were somewhat but not statistically lower in
responders vs. non-responders. On treatment lower CTCs and the absence of
aggressive CTC subtypes were associated with better clinical outcomes. On-
going analyses include single cell genomics, and analysis of T-cell populations.
Clinical trial information: NCT02496208.

4556 Poster Session (Board #382), Mon, 1:15 PM-4:15 PM 4557 Poster Session (Board #383), Mon, 1:15 PM-4:15 PM
Causes of patient ineligibility in clinical trials of metastatic urothelial cancer. Impact of antibiotic use on clinical outcomes in patients with urothelial
First Author: Gwynn Ison, U.S. Food and Drug Administration, Silver Spring, cancer receiving a programmed death protein 1 or programmed death ligand
MD 1 (anti-PD-1/L1) antibody. First Author: Chana Weinstock, U. S. Food and
Drug Administration, Silver Spring, MD
Background: Registrational trials of PD-1/PD-L1 inhibitor therapy help inform
clinicians and patients about the expected outcomes of patients receiving these Background: Previous data has suggested that patients treated with anti-PD-1/L1
drugs. However, the clinical trial population does not reflect all patients with the antibodies who receive antibiotics during their therapy might have dramatically
underlying disease. There have been ongoing efforts to expand eligibility criteria decreased progression-free and overall survival 1,2. This has clinical implications for
for clinical trial enrollment that are expected to mitigate some of these factors. management of patients with suspected bacterial infection while on treatment with
We reviewed four registrational trials of PD-1/PD-L1 inhibitor therapy to better these agents. We assessed the relationship between antibiotic use and tumor
understand why patients were deemed ineligible for trial enrollment and response rate, progression-free survival, and overall survival in a large dataset of
whether expanded eligibility criteria may have addressed the reasons for in- patients with urothelial cancer treated with anti-PD-1/L1 antibodies. Methods: We
eligibility. Methods: We reviewed four trials that led to approval of PD-1/PD-L1 examined seven trials that led to drug approval and which included 1747 patients
with metastatic or locally advanced urothelial cancer treated with an anti-PD-1/L1
inhibitor therapy. These trials screened 1931 and enrolled 1253 patients with
antibody. Five trials enrolled patients who had received prior platinum-based
metastatic or locally advanced urothelial cancer who had previously received
therapy and two enrolled patients who were cisplatin-ineligible. Six were single
platinum-based therapy. We examined the datasets to determine patient de- arm trials and one was a randomized controlled trial whose control arm is not
mographics and reasons for trial ineligibility. Results: There were no differences included in these analyses. Concomitant medication datasets were searched for
in the demographics characteristics of patients who were eligible or ineligible for systemic antibiotic used by each patient while on treatment. Results: Overall, 51%
study treatment. However, when compared to the SEER database the screened of patients (n=892) were exposed to antibiotics (ABX+) and 49% (n=855) were not
population was younger, and minorities were under-represented. Causes of exposed (ABX-). In these exploratory analyses, small numeric differences in OS,
patient ineligibility included: 1) Lack of Tumor Tissue/PD-L1 Low Tumor PFS, and ORR were seen in ABX+ vs. ABX- patients. Median OS was 9.23 vs.
Staining (23%), 2) Underlying Disease characteristics were not met (19%), 3) 9.86 months, median PFS was 105 vs 101 days, and ORR was 20% vs. 21% in
Laboratory Abnormalities (18%), 4) Excluded Co-morbid Conditions (16%), 5) ABX+ vs. ABX- patients, respectively. Conclusions: Patients who were treated with
Poor Performance Status (15%), 6) Refused Consent (10%), 7) Other (7%), antibiotics while on therapy with an anti-PD-1/L1 antibody for urothelial cancer had
and 8) Incorrect Prior Therapy (5%). Conclusions: Clinical trial accrual should similar outcomes to those who were not treated with antibiotics. Numeric differ-
be representative of all patients with the underlying disease. Changes in trial ences in outcomes were not significant and did not duplicate previous analysis
eligibility criteria such as less stringent requirements concerning tumor tissue demonstrating a median OS that was doubled in ABX- patients1. Our exploratory
(when possible), co-morbid conditions, laboratory abnormalities, and perfor- analyses do not appear to demonstrate a clear need for practitioners to avoid
mance status may improve patient accrual. These, when added together, antibiotic use in patients treated with PD-1/L1 agents for fear of significantly
formed the majority of the reasons for clinical trial ineligibility. impacting clinical outcomes. References: 1) Tinsley et. al., ASCO annual meeting
2018, abstract 3010 2) Routy et. al., Science 05 Jan 2018: Vol. 359, Issue 6371.
Antibiotics = Yes Antibiotics = No
N = 892 N = 855 Hazard Ratio
Overall Survival 9.23(7.62,10.78) 9.86(8.74,10.97) 1.06(95%CI: 0.94,1.21)
Median(95%CI) months
Progression Free Survival 105(86,113) 101(84,122) 1.06(95%CI: 0.94,1.2)
Median(95%CI) days
Overall response rate, % 20(17,22) 21(18,24) N/A
(confirmed, per independent review)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
278s Genitourinary (Nonprostate) Cancer

4558 Poster Session (Board #384), Mon, 1:15 PM-4:15 PM 4559 Poster Session (Board #385), Mon, 1:15 PM-4:15 PM
RNAseq and DNA whole-exome sequence analysis reveal novel response A pilot safety study of gemcitabine and cisplatin (GC) with atezolizumab (A)
signatures to IO treatment in muscle invasive bladder cancer (MIBC) as first-line therapy in patients (pts) with metastatic urothelial cancer (mUC).
patients. First Author: Gregory Mayhew, GeneCentric Therapeutics, Inc., First Author: Samuel Aaron Funt, Memorial Sloan Kettering Cancer Center,
Research Triangle Park, NC New York, NY
Background: Objective: To examine in a cohort of anti-PD-(L)1 immune Background: GC has a high overall response rate (ORR) but a high relapse
checkpoint inhibitors (ICP) treated urothelial cancer patients a strategy rate in pts with untreated mUC. Inhibition of programmed death-ligand 1
combining treatment outcomes with molecular alterations, pathways, and (PD-L1) with A can lead to long-term survival, but single-agent ORR is
immune/tumor microenvironment features to determine potential responder modest. We report the outcomes of GC+A in a cohort pts with mUC.
and rapid-progression signatures. Methods: De-identified clinical history Methods: This study was designed to assess the safety of GC + A in 10 pts
and treatment outcomes were collected on 109 MIBC patients treated with with untreated mUC prior to testing GC + A in a neoadjuvant study in pts with
ICP agents. Archived FFPE samples from these patients were obtained and muscle-invasive disease. The primary endpoint was safety as assessed by a
processed for mRNAseq, exome-seq, tumor mutation burden (TMB), predefined dose limiting toxicity (DLT) rate during the first cycle in the first 6
microsatellite instability (MSI) and mutation panel testing. Comprehensive pts. Total accrual goal was 10 pts to collect preliminary data on ORR and
tumor/immune profiling is being analyzed in the context of ICP treatments progression-free survival (PFS). RECIST 1.1 assessments were performed
and RECIST 1.1 outcomes. A 60 gene MIBC 4-typer expression subtyper and every 9 wks. Pts received 6 cycles of GC + A induction and then A main-
other response associated predictors are used to stratify and identify positive/ tenance every 3 wks. Results: No DLTs occurred during the first cycle in the
negative ICP response indicators. Results: 109 patients were identified first 6 pts. Grades 3-4 neutropenia and anemia occurred in 6/10 and 7/10
(median age 75, 64% male, 78% white, 17% black). 74% of patients had pts, respectively. Three pts required gemcitabine dose reductions for he-
received prior platinum-based chemotherapy, and 12% had received 2 or matologic toxicity and 2 pts had febrile neutropenia. One pt discontinued
more prior lines of therapy. At initiation of ICP, 28% of patients had cisplatin after 2 cycles for grade 3 hearing impairment but completed in-
hemoglobin , 10, 30% had liver metastases, and 59% had ECOG per- duction with gemcitabine and A. Only 1 pt discontinued study therapy due to
formance status . 0. Mutation analysis of the first 66 patients showed TP53 treatment-related adverse events (AEs), including A-related grade 4 en-
(n = 34, 52%), FGFR (n = 17, 26%), CDKN2A (n = 13, 20%) and RB1 (n = cephalopathy and grade 3 polyneuropathy. Three of 10 pts had visceral (liver
12, 18%) as the top alterations. No patients (0/8) with known pathogenic or bone) metastases. Of the 10 pts, 1 pt is completing induction but meets
mutations in FGFR3 (S249C and TACC3-fusion) responded to ICP. Of pa- initial criteria for partial response (PR), 8 pts had confirmed PR, and 1 pt had
tients with T2 staging prior to ICP (37/66), overall survival was markedly progressive disease (PD). Of 9 pts with confirmatory scans, the median PFS
shorter (2.7 years) in those possessing FGFR3 mutations (n = 6/37) com- was 10.6 months (95% CI 6.7, N/A). Of 8 pts with confirmed PR, 5
pared to that for FGFR3 WT patients (5.7 years, n = 31/37; p = 0.045). eventually had PD, 1 has just completed induction, 1 remains without PD at
Further analyses of molecular features relative to treatment outcomes are 25 months, and 1 had consolidation surgery with a pathologic complete
ongoing to characterize response signatures. Conclusions: Our preliminary response and remains disease-free at 21 months. Conclusions: This 10 pt
cohort of patients with pathogenic FGFR3 alterations showed 0% favorable study met its primary safety endpoint. The neoadjuvant study is ongoing
response to ICP. We are expanding on this observation with further com- (NCT02989584). Although there were a substantial number of grade 3-4
prehensive molecular analyses and retrospective treatments/outcomes data. toxicities, therapy was discontinued due to treatment-related AEs in only 1
We anticipate identifying expression signatures that reflect ICP patient pt. Immune correlative studies are ongoing. Clinical trial information:
responder/non-responder signatures that may aid in future therapy NCT02989584.
decisions.

4560 Poster Session (Board #386), Mon, 1:15 PM-4:15 PM 4561 Poster Session (Board #387), Mon, 1:15 PM-4:15 PM
Multi-omics profiling of upper-tract urothelial carcinomas. First Author: Preliminary phase 2 clinical results of IL-15RaFc superagonist N-803 with
Gabriel G. Malouf, Strasbourg University Hospital, Strasbourg, France BCG in BCG-unresponsive non-muscle invasive bladder cancer (NMIBC)
patients. First Author: Karim Chamie, Institute of Urologic Oncology (IUO),
Background: Upper-tract urothelial carcinomas (UTUC) may harbor similar
Department of Urology, David Geffen School of Medicine at UCLA, Los
genetic profile as compared to bladder cancer, although frequencies of
Angeles, CA
mutated genes have been shown to differ between them. However, to the
best of our knowledge, the epigenetic landscapes of UTUC and their as- Background: Patients with non-muscle-invasive bladder cancer (NMIBC) un-
sociation with genetic alterations and clinico-pathological tumor features responsive to BCG therapy have limited treatment options. N-803 (also known
remain unknown. Methods: We collected 40 UTUC samples (20 non-muscle as ALT-803) is an IL-15-based immunostimulatory protein complex (IL-
invasive (NMI) and 20 muscle-invasive (MI) and carried out whole-exome 15RaFc) that promotes proliferation and activation of natural killer (NK) cells
sequencing (n = 30), DNA methylation using Infinium EPIC arrays (n = 35) and CD8+ T cells, but not regulatory T cells. Phase Ib data in BCG-naı̈ve patients
and RNA sequencing (n = 20). Validation was performed on TCGA bladder with NMIBC demonstrate that intravesical administration of N-803 with BCG
cancer dataset. Results: We identified 3232 putative somatic mutations induced complete response in all patients, without recurrences for the study
with an average of 2.1+/-2.6 mutations per megabase. Significantly mutated duration of 24 months. Methods: An open-label, single-arm multicenter Phase
genes were FGFR3 (50%), KDM6A (27%), MLL2 (27%) and ARID1A/B 2 study of intravesical BCG plus N-803 in patients with BCG-unresponsive high-
(23%). No difference in term of genetic alterations were identified between grade NMIBC (NCT03022825) was opened. The study has two cohorts: Cohort
MI- and NMI- UTUC. Unsupervised hierarchical clustering using most A, patients with BCG-unresponsive carcinoma in situ (CIS) [with or without Ta or
variable DNA methylation probes uncovered two robust DNA methylation T1 disease] and Cohort B, patients with BCG-unresponsive high-grade Ta/T1
epi-clusters. Epi-cluster C1 (n = 23; 65.7%) displayed markedly higher DNA disease. All treated patients receive intravesical N-803 plus BCG, similar to a
methylation relative to Epi-cluster C2 (n = 12; 34.3%). Notably, all muscle- standard induction and maintenance treatment schedule. The primary endpoint
invasive samples were enriched in C1 (16/17, 94.1%); conversely, C2 was for Cohort A is incidence of complete response (CR) of CIS at any time, and the
enriched with non-muscle-invasive samples (p = 0.0009). Overall, 14.209 primary endpoint for Cohort B is disease-free rate at 12 months. Results: To
probes were significantly hypermethylated in C1 as compared to C2 epi- date, forty-six patients have enrolled in this phase 2 trial (Cohort A (CIS), n = 23,
cluster; Gene Set Enrichment Analysis (GSEA) demonstrated that those were Cohort B (Papillary), n = 23). Of eleven evaluable patients in Cohort A, nine
enriched for PRC2 targets (p = 6x10-65). Integrative analysis with tumor patients (82%) have a reported CR. In addition, seven out of nine (78%)
genetic landscape showed that C1 epi-cluster was enriched for mutations in patients in Cohort A demonstrated CR at their 6-month response assessment. Of
SWI/SNF complex as compared to C2 (p = 0.02). We then applied our epi- thirteen evaluable patients in Cohort B, ten patients (77%) showed no evidence
signature to bladder TCGA cohort and obtained two similar epi-clusters of recurrence at their 3-month response assessment; of these, none (0/8)
associated with patients overall survival (p = 0.035). Conclusions: Our study evaluated past 3 months have had disease recurrence. Three serious adverse
demonstrate for the first time that difference between MI- and NM- invasive events (AEs) have been reported (E coli infection, anemia, and bacteremia),
UTUC might be related to epigenetic rather than genetic alterations. This with no immune-related AEs. Conclusions: Nine out of eleven (82%) patients
might pave the way for testing epigenetic therapies in non-muscle invasive with BCG-unresponsive CIS of the bladder demonstrated a complete response.
tumors with the aim to prevent recurrence and distant metastasis. Ten out of thirteen patients with BCG-unresponsive papillary NMIBC show no
evidence of disease at first assessment. Intravesical N-803 plus BCG was well-
tolerated and no patients experienced immune-related AEs. Clinical trial in-
formation: NCT03022825.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Genitourinary (Nonprostate) Cancer 279s

4563 Poster Session (Board #389), Mon, 1:15 PM-4:15 PM 4564 Poster Session (Board #390), Mon, 1:15 PM-4:15 PM
Biomarkers of outcomes in a randomized phase II trial of first-line paclitaxel, A reduced pazopanib dose with food: Is it more patient-friendly and does it
ifosfamide, and cisplatin (TIP) versus bleomycin, etoposide, and cisplatin reduce drug costs? First Author: Floor Lubberman, Radboud University
(BEP) for intermediate- and poor-risk germ cell tumors (GCT). First Author: Medical Center, Department of Pharmacy, Nijmegen, Netherlands
Darren R. Feldman, Memorial Sloan Kettering Cancer Center, New York, NY
Background: Pazopanib has been licensed for advanced soft tissue sarcoma
Background: We previously reported no difference in favorable response rate (FRR) or and metastatic renal cell carcinoma in a fixed oral daily dose of 800mg taken
PFS for TIP vs BEP. Here we present results of a pre-planned analysis of biomarkers of fasted. We hypothesized that ingesting pazopanib with food may improve
outcome. Methods: HCG and AFP were drawn on days 1 and 15 of each cycle and patients’ comfort and reduce gastro-intestinal adverse events. Moreover, a food
rates of decline classified as satisfactory [S] or unsatisfactory [US] by MSK (Motzer intervention, resulting in a better absorption, can lead to a lower dose, which
JCO 2007) and GETUG (Fizazi Lancet Oncol 2014) methods. IHC for ERCC1, could significantly reduce treatment costs. Methods: Part 1 of the study was
RAD51, PARP1, HER-2, and p-AKT was performed on pre-treatment tumor samples.
performed to determine whether 600mg pazopanib taken with a continental
An H-score (0 – 300) was calculated for each stain (H = stain intensity [0 – 3] x %
positive cells [0-100]). H-score and marker decline category were correlated with FRR
breakfast was bioequivalent to 800mg pazopanib taken fasted. In part 2,
(PR + CR) and PFS. Patients (pts) who received disease-stabilizing chemotherapy differences in GI-toxicity and patient satisfaction were assessed by the cancer-
were excluded from marker analyses. Results: Of 91 pts, 80 did not receive disease- therapy-satisfaction-questionnaire after both intake regimens. Finally, pa-
stabilizing treatment with 79 having sufficient marker values for analysis by the MSK tient’s preference for either intake regimen was asked. Results: 16 patients
method and 75 by GETUG. By MSK, 49 had S decline vs 30 US; by GETUG, 34 S vs were included in the bioequivalence study. The geometric mean ratio (fed/
41 US. FRR and PFS were improved for pts with S vs US decline by both methods and fasted) of the area under the plasma concentration time curve was 1.10 (90%
remained significant by the MSK method when stratified by IGCCCG group (Table). CI 1.00-1.19), maximum peak concentration was 1.12 (90% CI 1.02-1.22)
IHC (n=77) quality was adequate in 71 to 73 pts (varied by stain) and was positive and pazopanib trough concentration was 1.10 (90% CI 1.02-1.18). In part 2,
(H .0) for PARP in 68/73, ERCC1 in 54/71, RAD51 in 54/73, p-AKT in 5/72, and 60 patients were included. No differences were seen in the occurrence of GI-
HER2 in 4/72. Only PARP1 was associated with outcome with worse PFS for the toxicities under both intake regimens. Patients seem to be more positive about
lowest expression tertile (H , 180; p=0.013). Conclusions: PARP1 expression and their feelings about side effects (72.3(95% CI 68.1-76.5) vs 68$2 (62.7-
tumor marker decline rates, particularly by MSK method, were significantly associated 73.6); p=.092) and satisfaction with therapy scores were higher (84.7(95% CI
with outcome to initial chemotherapy in int/poor risk GCT. Future trials incorporating
81.4-87.9) vs 81.9 (78.7-85.2); p= .059) when pazopanib was taken with
marker decline into treatment allocation and validating the prognostic effect of
food. 41 (68%) of the patients preferred the intake with continental breakfast.
PARP1 expression are warranted. Clinical trial information: NCT01873326.
Conclusions: Intake of 600mg pazopanib with food results in bioequivalent
MSK Method
S vs US
GETUG Method
S vs US
exposure and was preferred over a standard pazopanib dose without food.
Outcome by marker decline.* (n = 49 vs 30) (n = 34 vs 41) Moreover, with this simple food intervention a large cost reduction can
All patients be realized in patients treated with pazopanib. Clinical trial information:
FRR, % 94 vs 53 91 vs 66
(p,0.001) (p=0.01) NCT02138526.
2y PFS, % 90 vs 50 88 vs 60
(p,0.001) (p=0.02)
Intermediate-risk
FRR, % 100 vs 75 100 vs 85
(p=0.008) (p=0.07)
2y PFS, % 100 vs 63 100 vs 76
(p,0.001) (p=0.02)
Poor-risk
FRR, % 85 vs 45 77 vs 57
(p=0.006) (p=0.22)
2y PFS, % 76 vs 45 69 vs 53
(p=0.02) (p=0.59)

*log-rank p value used for 2y PFS comparisons.

4565 Poster Session (Board #391), Mon, 1:15 PM-4:15 PM 4566 Poster Session (Board #392), Mon, 1:15 PM-4:15 PM
Comparative genomic profiling from tumor tissue and circulating tumor DNA Clinical and economic outcomes associated with sequential treatment in patients
(ctDNA) in 111 patients (Pts) with metastatic clear cell renal cell carcinoma. with advanced renal cell carcinoma (aRCC). First Author: Meredith M. Regan, 1
First Author: Ritesh Kotecha, Memorial Sloan Kettering Cancer Center, New Beth Israel Deaconess Medical Center, Dana-Farber/Harvard Cancer Center,
York, NY Boston, MA
Background: Circulating tumor DNA (ctDNA) assessment is a non-invasive Background: Immuno-oncology therapies (IOs) and tyrosine kinase inhibitors
approach for genomic interrogation of solid tumors. As a novel tool, key (TKIs) are recommended for the treatment of aRCC. As new drugs and
benchmarks for applications in metastatic clear cell renal cell carcinoma combination regimens emerge, there is interest in gaining a deeper un-
(mccRCC) are yet to be determined. To understand the utility of ctDNA, we derstanding of optimal treatment sequencing. We aimed to assess clinical and
performed a large cohort analysis using a comparative genomics approach economic outcomes associated with treatment sequences for untreated aRCC
integrating matched primary tissue and ctDNA genomic data. Methods: Pts patients with IMDC intermediate/poor risk. Methods: A discrete event simu-
with prior tumor mutational profiles generated via next generation sequencing lation model was developed to estimate the total costs and survival (in life-
(NGS) from nephrectomy or metastatic specimens underwent single-time point years; LYs) over patients’ lifetimes when receiving sequential treatment with
plasma collection. Targeted NGS sequencing with MSK-IMPACT was per- nivolumab + ipilimumab (N+I), sunitinib (SUN), pazopanib (PAZ), or cabo-
formed on tumor and ctDNA with subsequent bi-directional cross genotyping zantinib (CAB) as first-line (1L) treatment, and nivolumab (NIVO), axitinib
using Waltz 2.0. All pts had matched germline comparison from peripheral (AXI), PAZ, CAB, or lenvatinib + everolimus (LEN+EVE) as second line (2L).
blood; clinical data was extracted from medical record. Liberal (1-2 reads) and Efficacy inputs were derived from the CheckMate 214 trial and a network meta-
stringent ($3 reads) filters were applied, with a cut-off of , 30% allele analysis based on available literature. Safety and cost data were obtained from
frequency to remove germline mutations. Results: 111 mccRCC pts, of whom literature and publicly available sources. Results: N+I initiating sequences
available IMDC-risk was favorable (35%), intermediate (60%), and poor-risk were estimated to provide longer survival in mean LYs and lower mean costs/LY
(5%) were included for analysis. The median time between tissue and ctDNA versus sequences with 1L TKIs (table). The estimates of incremental cost-
collection was 23 months (R: 1-177), and 96% of patients had undergone effectiveness ratio (ICER) for N+I initiating sequences with 2L TKI mono-
nephrectomy prior to ctDNA collection. In primary tissue sequencing, 64/111 therapy were well below the willingness-to-pay threshold of $50,000. Using 2L
(58%) from nephrectomy and 42/111 (42%) from metastatic sites, 569 LEN+EVE, compared with 2L monotherapies, provided an incremental survival
unique alterations were identified across the whole cohort, with a median of 4 gain but at costs/LY close to $100,000. Conclusions: Use of 1L N+I followed
mutations/pt (R:1-23). RCC-specific alterations included VHL (88%), PBRM1 by TKI monotherapy is estimated to provide longer survival while being more
(48%), SETD2 (34%), KDM5C (17%), TP53 (14%). Across the cohort, 176 cost-effective versus TKIs followed by IOs or sequences cycling TKIs, mainly
alterations were identified in ctDNA. With cross genotyping, ctDNA alterations driven by a longer time to 2L treatment and longer treatment-free survival with
concordant with primary tumors were detected in 20% (22/111 pts, 28 unique N+I. Clinical trials with head-to-head comparisons of treatment sequences
alterations) using stringent criteria with a median of 1 mutation/pt (R:1-2). would be necessary to validate the findings of the study.
Using liberal criteria, concordance with primary tumors was 59% (66/111 pts,
142 unique alterations) with a median of 2 mutations/pt (R: 1-8). 1L 2L LYs* Total costs* Costs/LY*
Conclusions: This large cohort study matching oncogenomics from tumor and N+I CAB/AXI/PAZ 4.8–6.8 $264,722–$355,174 $48,832–$55,700
ctDNA highlights complexities and challenges of applying liquid biopsy in CAB NIVO/AXI/PAZ 4.6–5.1 $343,874–$359,143 $70,340–$77,566
biomarker development in mccRCC. SUN NIVO/CAB/AXI 3.1–3.7 $221,111–$277,802 $61,274–$74,794
PAZ NIVO/CAB/AXI 3.1–3.7 $221,597–$278,288 $61,405–$74,924
* Range for the sequences with different 2L options.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
280s Genitourinary (Nonprostate) Cancer

4567 Poster Session (Board #393), Mon, 1:15 PM-4:15 PM 4568 Poster Session (Board #394), Mon, 1:15 PM-4:15 PM
Phase I/II study of axitinib (axi) and nivolumab (nivo) in patients with meta- Association of human endogenous retrovirus (hERV) expression with clinical
static renal cell carcinoma (mRCC). First Author: Matthew R. Zibelman, Fox efficacy of PD-1 blockade in metastatic clear cell renal cell carcinoma
Chase Cancer Center, Philadelphia, PA (mccRCC). First Author: Jean-Christophe Pignon, Brigham and Women’s
Hospital, Boston, MA
Background: Drug combinations targeting vascular endothelial growth factor
(VEGF) and the programmed death one (PD-1) pathway have demonstrated Background: hERV levels positively correlate with tumor immune infiltrate
encouraging efficacy in patients (pts) with mRCC. We are conducting a phase I/ and were recently shown to be associated with clinical benefit to PD-1/PD-L1
II trial combining the VEGF-targeting tyrosine kinase inhibitor (TKI) axi and the blockade in two small cohorts of patients (pts) with mccRCC (Smith C.C. et al
PD-1 inhibitor nivo in mRCC pts. Methods: The phase I portion of this and Panda A. et al; 2018). We tested whether hERV levels correlate with
investigator-initiated, multi-center trial enrolled pts with TKI-refractory mRCC efficacy of nivolumab in a prospective phase II study of pts with mccRCC
(any line) and no prior anti-PD-1. It used a 3+3 design, with axi dosing starting (Checkmate 010). Methods: Reverse transcribed RNA extracted from 99
at 3 mg orally BID, escalating to a target dose of5 mg BID. Nivo dosing was FFPE pretreatment tumors were analyzed by RT-qPCR to assess levels of
fixed at 240 mg IV Q2 weeks or 480 mg Q4 weeks. Eligible pts received axi pan-ERVE4, pan-ERV3.2, hERV4700 GAG or ENV, and the reference genes
alone for a one week induction, followed by combination dosing. There was a 18S and HPRT1. Normalized hERV levels were transformed as categorical
4 week period to assess for dose limiting toxicities (DLTs). The primary value (high or low) using population quartiles as cutoffs. For each cutoff,
endpoint for phase I was safety and establishment of a recommended phase II samples with non-quantifiable hERV levels for which the limit of quantifi-
dose (RP2D) for axi. Safety and early efficacy data for the phase I cohort are cation was above the tested cutoff could not be categorized and were ex-
presented. Results: Twelve pts received treatment during the phase I portion, cluded from analysis. Log rank test was used to test the association of hERV
with all evaluable for toxicity. Ten were 2nd line and 2 were 3rd line or beyond. levels with PFS/irPFS (RECISTv1.1/irRECIST) at each cutoff using Holm-
One DLT was noted in the first cohort of three pts (grade 3 autoimmune Bonferroni correction for Type I error control; adjusted P-values are reported.
disorder), thus that cohort was expanded. No further DLTs were reported at axi Fisher’s exact test was then used to explore the association with ORR/irORR
3 mg BID, or at further expansion to 5 mg BID. Grade 3 adverse events (AEs) at (RECISTv1.1/irRECIST). Results: Among the hERV studied, only hERV4700
least possibly related to one of the drugs included expected side effects HTN, ENV was significantly associated with PFS/irPFS. At the 25th percentile
hyperglycemia, and transaminitis. One patient received steroids for an immune cutoff, 45 pts had high levels of hERV4700 ENV and 24 pts had low levels of
related AE (grade 3 transaminitis) that resolved to grade 1 with steroids. No hERV4700 ENV. Median PFS and irPFS were significantly longer in the high-
grade 4-5 AEs occurred. Axi at 5 mg BID was chosen as the RP2D. With 10 pts hERV4700 ENV group [7.0 (95% CI: 2.2 - 10.2) and 8.5 (95% CI: 4.2 -
evaluable for efficacy, 4 pts had partial responses, 4 pts had stable disease, 14.1) months, respectively] versus the low-hERV4700 ENV group [2.6
and 2 had progressive disease. Updated efficacy data will be presented. (95% CI: 1.4 - 5.4) and 2.9 (95% CI: 1.4 - 5.7) months, respectively] (P =
Conclusions: The phase I portion of this phase I/II trial has demonstrated 0.010 for PFS and P = 0.028 for irPFS). At the same cutoff, ORR and irORR
expected safety and established 5 mg BID as the RP2D axi dose. This ongoing rates were significantly higher in the high-hERV4700 ENV group [35.6 (95%
VEGF TKI/PD1 trial incorporating two drugs already approved for mRCC pts has CI: 21.9 - 51.2) % for both ORR/irORR] versus the low-hERV4700 ENV
shown encouraging signs of efficacy and has now expanded to include group [12.5 (95% CI: 2.7 - 32.4) and 8.3 (95% CI: 1.0 - 27.0) %, re-
treatment naı̈ve pts in addition to TKI-refractory pts as part of phase II. Clinical spectively] (P = 0.036 for ORR and P = 0.012 for irORR).
trial information: NCT03172754. Conclusions: hERV4700 ENV levels may predict outcome on nivolumab in
mccRCC. Validation of our results and correlation of hERV levels with im-
mune markers in a controlled phase III trial (CheckMate 025) is ongoing.

4569 Poster Session (Board #395), Mon, 1:15 PM-4:15 PM 4570 Poster Session (Board #396), Mon, 1:15 PM-4:15 PM
KEYNOTE-427 cohort B: First-line pembrolizumab (pembro) monotherapy First-line pembrolizumab (pembro) monotherapy in advanced clear cell renal
for advanced non‒clear cell renal cell carcinoma (NCC-RCC). First Author: cell carcinoma (ccRCC): Updated results for KEYNOTE-427 cohort A. First
Jae-Lyun Lee, Asan Medical Center and University of Ulsan College of Author: Scott S. Tykodi, University of Washington and Fred Hutchinson
Medicine, Seoul, South Korea Cancer Research Center, Seattle, WA
Background: Efficacy of PD-1 inhibitors (or any therapy) in NCC-RCC has not Background: KEYNOTE-427 (NCT02853344) is an open-label, single-arm,
been established. KEYNOTE-427 (NCT02853344) is a single-arm, open-label, phase 2 study to evaluate efficacy and safety of first-line single-agent pembro, a
phase 2 study of pembro monotherapy in patients (pts) with advanced clear cell programmed death 1 (PD-1) inhibitor, in patients (pts) with ccRCC (cohort A)
RCC (cohort A) and NCC-RCC (cohort B). Cohort B results are presented. and non–clear cell RCC (cohort B). Updated follow up from cohort A are pre-
Methods: 165 pts with histologically confirmed NCC-RCC, no prior systemic sented. Methods: Pts with histologically confirmed ccRCC, measurable per
therapy, measurable disease (RECIST v1.1), and KPS $70% enrolled. Pts RECIST v1.1, and no prior systemic therapy were eligible. Pts received pembro
received pembro 200 mg IV Q3W for 35 cycles (~2 y) or until progressive 200 mg IV Q3W for 2 y or until confirmed progressive disease, unacceptable
disease (PD), unacceptable toxicity, or withdrawal. Primary end point: objective toxicity, or pt decision to withdraw. Primary end point was objective response rate
response rate (ORR) per RECIST v1.1 by blinded independent central review. (ORR; per RECIST v1.1 blinded independent central review). Additional end
Additional end points: duration of response (DOR), population description by points included duration of response (DOR), disease control rate (DCR),
sarcomatoid differentiation, histology and PD-L1 expression (combined positive progression-free survival (PFS), overall survival (OS), and safety. Results: 110
score [CPS] $1 for PD-L1+). Results: Histology was confirmed by a central pts enrolled; median (range) follow-up was 18.0 (2.5-22.7) mo. Median age
pathologist: papillary 72% (n = 118), chromophobe 13% (n = 21), unclassified (range) was 64 (29-87); 38.2%, 47.3%, and 14.5% had favorable, in-
16% (n = 26); 62% were PD-L1+. At analysis, 49 pts had died and 3 had termediate, and poor IMDC risk, respectively; 47.3% were PD-L1 positive.
withdrawn. At median follow-up of 11.1 mo (range, 0.9-21.3), 56% of pts Confirmed ORR was 36.4% with 3 (2.7%) CRs and 37 (33.6%) PRs. Median
discontinued pembro due to PD or clinical progression. Overall ORR was 24.8% DOR was not reached. Median PFS was 7.1 mo (95% CI, 5.6-11.0) and median
(95% CI, 18.5-32.2; 8 [4.8%] CR, 33 [20.0%] PR). Median DOR was not OS was not reached. Results by IMDC category are outlined in the table. By PD-
reached. For responding pts, 81.5% had a response $6 mo. 12-mo PFS and OS L1 status, confirmed ORR was 44.2% and 29.3% for positive and negative,
rates were 22.8% and 72.0%, respectively. ORR (95% CI) was 25.4% (17.9- respectively. By sarcomatoid differentiation (n=11), confirmed ORR was 63.6%.
34.3) with papillary, 9.5% (1.2-30.4) with chromophobe, and 34.6% (17.2- Treatment-related AEs occurred in 80.9%, with pruritus (28.2%) and fatigue
55.7) with unclassified NCC-RCC; for responding pts, 82.1%, 50.0%, and (28.2%) most commonly reported. One pt died of treatment-related pneumo-
87.5% had a response $6 mo, respectively. Median DOR was not reached in nitis. Conclusions: With a median 18-months’ follow up, first-line pembro
any group. ORR (95% CI) was 44.7% (28.6-61.7) for pts with sarcomatoid monotherapy continued to show antitumor activity in pts with ccRCC. Mean-
differentiation (n = 38). ORR (95% CI) was 33.3% (24.3-43.4) and 10.3% ingful responses were observed in pts with intermediate/poor IMDC risk, PD-L1
(3.9-21.2) with CPS$1 and CPS , 1, respectively. Grade 3-5 treatment- positive and sarcomatoid differentiated tumors. Safety profile was comparable to
related adverse events (TRAEs) occurred in 11% of pts. 6 pts died of AEs, 2 of previously reported. Clinical trial information: NCT02853344.
TRAEs (pneumonia and cardiac arrest). Conclusions: Single-agent pembro
IMDC
showed encouraging antitumor activity in NCC-RCC, especially with papillary Total Favorable IMDC Intermediate/Poor
or unclassified histology. Safety profile of pembro was as expected. Updated N=110 N=42 N=68
data with additional follow-up will be presented. Clinical trial information: ORR, % (95% CI) 36.4 (27.4-46.1) 31.0 (17.6-47.1) 39.7 (28.0-52.3)
NCT02853344. DCR, % (95% CI) 57.3 (47.5-66.7) 61.9 (45.6, 76.4) 54.4 (41.9-66.5)
DOR, median (range), mo Not reached 13.7 (4.2-18.0) Not reached
DOR ‡ 6 mo (responders), % 76.8 68.4 80.9

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Genitourinary (Nonprostate) Cancer 281s

4571 Poster Session (Board #397), Mon, 1:15 PM-4:15 PM 4572 Poster Session (Board #398), Mon, 1:15 PM-4:15 PM
Retrospective analysis of the safety and efficacy of immune checkpoint TIVO-3: Subgroup analysis of progression-free survival of tivozanib com-
inhibitors (CPI) among patients (pts) with pre-existing autoimmune disorders pared to sorafenib in subjects with refractory advanced renal cell carcinoma
(AD) and renal cell carcinoma (RCC) or urothelial carcinoma (UC). First (RCC). First Author: Camillo Porta, Department of Internal Medicine, Uni-
Author: Nieves Martinez Chanza, Jules Bordet Institute, Brussels, Belgium versity of Pavia and Division of Traslational Oncology, IRCCS Istituti Clinici
Scientifici Maugeri, Pavia, Italy
Background: RCC and UC pts with clinically significant AD were generally excluded
from CPI trials due to potential AD exacerbation. Thus, the safety and clinical activity Background: Tivozanib (T) is a biochemically potent and highly selective
of CPI in AD pts are not well characterized. Methods: We retrospectively collected data VEGF tyrosine kinase receptor inhibitor in clinical development in RCC. The
from RCC and UC pts with AD treated with CPI at 9 centers. Adverse events (AEs) were TIVO-3 trial in 3rd and 4th line subjects with metastatic (m) RCC showed a
assessed using CTCAEv5 criteria. Objective response rate (ORR) was assessed by median progression free survival (mPFS) of 5.6 months (mos) for T compared
RECIST principles. Overall survival (OS) was estimated by Kaplan Meier. Results: Of
to 3.9 mos for sorafenib (S) (p = 0.017, HR = 0.73). Methods: Subjects with
103 pts (57 RCC & 46 UC) with a broad spectrum of AD such as psoriasis (22%),
thyroiditis (20%), rheumatoid arthritis (13%), polymyalgia rheumatica (8%), in-
mRCC who failed 2 or 3 prior systemic regimens, one of which included a
flammatory bowel disease (6%), multiple sclerosis (3%) and lupus (3%), most re- VEGFR TKI other than S or T, were stratified by IMDC risk category and type
ceived CPI as 1st or 2nd line (77% RCC, 93% UC) and anti-PD-1/L1 monotherapy of prior therapy (two TKIs; TKI plus checkpoint; TKI + other) then ran-
(65% RCC, 98% UC). At CPI start, 36 had clinically active AD (all grade 1-2) and 4(11%) domized in a 1:1 ratio to T or S. The primary objective was to compare PFS by
requiring systemic immunosuppression. AD exacerbations occurred in 37% (n = 38); most blinded independent radiological review. Pre-specified subgroups included
frequent: arthritis (12% RCC, 24% UC), rash (11% RCC, 9% UC). New onset immune prior treatment, IMDC prognostic group, and demographic characteristics.
related (ir) AEs occurred in 36% (n = 37); most frequent: colitis (12% RCC, 4% UC), rash Results: Tivozanib demonstrated PFS benefit in all subgroups including
(11% RCC, 9% UC), hypothyroid (each 7%), nephritis (7% UC). Table details timing and men, women, patients over 65, and under 65. The hazard ratio was the same
management. Median followup was 12.5 (1-52) mos for RCC and 14.5 (1-53) mos for UC. for patients enrolled in North America and the EU. Patients with ECOG
Median time on CPI was 6 mos (1-36) RCC and 4 mos (0.5-40) UC. At data cutoff, 39 RCC performance status (PS) of 0 had a lower HR than patients with ECOG PS of
& 36 UC had discontinued CPI; 16% for toxicity. ORR was 31% for RCC and 35% for UC. 1. There was an increase in HR from IMDC favorable vs IMDC intermediate vs
1 yr-OS rate was 74% (95%CI 58-84) for RCC and 60% (95%CI 43-74) for UC. IMDC poor. PFS favored tivozanib in patients who had two prior lines of
Conclusions: AD exacerbations or new irAEs occurred in 37% and 36% respectively and
therapy, those treated with three prior lines, those with a prior checkpoint
were generally manageable. Only a minority required CPI cessation due to toxicity.
inhibitor, or with two prior VEGFR TKIs. Conclusions: Tivozanib improved
AD EXACERBATIONS NEW irAEs PFS vs. sorafenib across several subgroups in TIVO-3. Patients with favorable
RCC UC RCC UC and intermediate IMDC risk and ECOG PS 0 seemed to derive the most
N = 57 N = 46 N = 57 N = 46 benefit. Patients treated with a prior checkpoint inhibitor or two VEGFR-TKIs
N (%) 18 (32) 20 (43) 21 (37) 16 (35) had a longer PFS than patients treated on sorafenib. Clinical trial information:
38 (37) 37 (36)
Grade (n, %) NCT02627963.
1/2 13(23) 14(30) 14(25) 9(20)
3/4 2(4) 4(9) 7(12) 5(11) Subgroup HR Subgroup HR
Unkn 3(5) 2(4) - 2(4)
Med Time CPI to AE, days (range) 69 (25-315) 32 (3-368) 48 (2-305) 120 (12-443) Male 0.64 IMDC Favorable 0.46
Systemic steroid (n, %) 5 (28) 12 (60) 11 (52) 9 (56) Female 0.72 IMDC Intermediate 0.69
DMARDs (n, %) 1 (6) 2 (10) 0 (0) 0(0)
CPI stopped for irAE (n, %) Age £65 0.74 IMDC Poor 1.15
Temporary 3 (17) 5 (25) 8 (38) 6 (36) Age >65 0.59 Two prior VEGFR-TKIs 0.57
Permanent 2 (11) 4 (20) 3 (14) 3 (19) ECOG 0 0.54 Prior Checkpoint Ab 0.55
ECOG 1 0.87 3rd Line 0.69
North America 0.71 4th Line 0.64
EU 0.69 HR , 1 favors tivozanib

4573 Poster Session (Board #399), Mon, 1:15 PM-4:15 PM 4574 Poster Session (Board #400), Mon, 1:15 PM-4:15 PM
Adjuvant axitinib dose modification in renal cell carcinoma (RCC): Analysis Sequential treatment with pazopanib (PAZO) followed by nivolumab (NIVO) in
of the ATLAS study. First Author: David I. Quinn, USC Norris Comprehensive patients with advanced or metastatic renal cell carcinoma (mRCC): Third
Cancer Center, Los Angeles, CA interim results of the non-interventional study PAZOREAL. First Author: Martin
Boegemann, University of Muenster Medical Center, Münster, Germany
Background: The ATLAS trial compared axitinib vs placebo in patients (pts)
with locoregional RCC at risk of recurrence after nephrectomy. ATLAS was Background: Randomized clinical trials for the implementation of new
stopped due to futility at a pre-planned interim analysis; results showed no therapies include only a selection of patients that are later treated with these
difference in disease-free survival (DFS) between the treatment arms. We new options. Thus, real-world evidence is urgently needed not only to monitor
explored whether pts treated longer with axitinib achieved better outcomes the translation of treatment approaches into routine practice but also to
and the impact of axitinib dose reduction or increase on DFS – the trial improve cancer treatment and survivorship care on a broader scale.
required dose reduction for toxicity and allowed dose escalation in the event Methods: PAZOREAL is a prospective, multicenter, non-interventional study
of no or minimal toxicity. Methods: Pts in ATLAS received a maximum of 3 y to evaluate effectiveness [primary time on drug (TD)], tolerability, safety, and
of study treatment. A landmark analysis was conducted comparing pts quality of life (QoL) in patients (pts) with mRCC, treated with 1st L PAZO
treated with axitinib #1 y vs . 1 y. Pts who recurred or censored prior to 1 y followed by 2nd L NIVO or everolimus (EVE). Results: Between Dec. 2015
were excluded. An analysis of daily dose characteristics was undertaken to and Sep. 2017, 421 pts were enrolled and 402 pts started 1st L PAZO
compare patients whose dose was reduced and whose dose was increased to treatment (Tx), 127 and 5 pts received NIVO and EVE as 2nd L Tx, resp., 56
those with a stable dose of axitinib. Cox proportional hazard model was used entered follow-up. At time of data-cut (08 Nov 2018) median TD was
for DFS analysis. Toxicity analysis using a 90-day landmark was also con- 6.6 months (95%CI 6.0-7.9) for 1st L PAZO and 4.1 months (95%CI 3.2-
ducted. Results: Overall, 264 axitinib-treated pts were included in this 5.8) for 2nd L NIVO (all pts), 8.1 months (95% CI 6.6-9.5) for PAZO and 3.2
analysis. Of these 42 pts were treated for #1 y and 222 pts for . 1 y. Pts (2.7-6.5) for NIVO Tx for trial eligible pts (39.1% of 402 pts). Median TD for
remaining on axitinib . 1 y vs #1 y did not have different DFS (hazard ratio pts with or without prior nephrectomy was 7.6 vs 4.5 months, resp. The
[HR] = 0.572, 95% confidence interval [CI]: 0.247–1.327, P= 0.1874). Pts clinical benefit rate of 1st L PAZO was 58.2 % (95% CI 53.3-62.9) based on
with dose reduction had longer DFS than those with a stable dose (HR = investigator assessment. Median OS of PAZO was 29.5 months (95%CI
0.458, CI: 0.305–0.687, P= 0.0001). Pts with dose increase did not have 23.6-NA) for all pts, 28.2 months (95% CI 22.2-NA) for NIVO in 2nd L. The
DFS different to stable dose pts (HR = 1.936, CI: 0.937–3.997, P= most commonly reported AEs for PAZO Tx were diarrhea (35%), nausea
0.0685). No difference in DFS in pts experiencing grade $2 adverse events (20.3%) and fatigue (17.5%). Most common PAZO-related grade 3/4 ad-
(AEs) vs grade , 2 AEs within 90 d of start of treatment was observed (HR = verse events were hypertension (5%), hypertensive crisis (2.3%) and GGT
0.885, CI: 0.419–1.869, P= 0.7488). Pts experiencing grade $3 AEs increase (1.8%). QoL evaluated by EQ-5D-5L remained stable over different
within 90 d of start of treatment had shorter DFS compared to those that did Tx lines. Conclusions: The interim results of the PAZOREAL study confirm a
not (HR = 1.643, CI: 0.963–2.801, P= 0.0653). Conclusions: DFS did not favorable overall survival in pts with mRCC treated with 1st L PAZO in a real-
vary based on duration of axitinib treatment, however, pts with dose re- world setting, good benefit-risk profile of PAZO and sustained QoL monitored
ductions had longer DFS vs pts with stable dose or dose increases. This over several treatment lines. In Germany NIVO as 2nd L Tx is commonly
difference suggests that there is a relation between axitinib exposure and applied after 1st L Tx with PAZO.
DFS as seen with sunitinib in the advanced setting and pazopanib in the
PROTECT study.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
282s Genitourinary (Nonprostate) Cancer

4575 Poster Session (Board #401), Mon, 1:15 PM-4:15 PM 4576 Poster Session (Board #402), Mon, 1:15 PM-4:15 PM
Consistent efficacy of nivolumab plus ipilimumab across number of In- Systemic therapy for advanced clear cell renal cell carcinoma (ccRCC) after
ternational Metastatic Database Consortium (IMDC) risk factors in Check- progression on immune-oncology plus VEGF targeted therapy combinations
Mate 214. First Author: Bernard Escudier, Gustave Roussy, Villejuif, France (IO-VEGF). First Author: Yasser Ged, Memorial Sloan Kettering Cancer
Center, New York, NY
Background: The IMDC prognostic model was created based on anti-VEGF
treatments for advanced renal cell carcinoma (aRCC), and may not be relevant Background: IO-VEGF combinations are the backbone for current and future
for immunotherapy. Methods: In a post hoc analysis of CheckMate 214, we therapeutic developments in RCC with several IO-VEGF regimens reporting
compared efficacy with nivolumab + ipilimumab (N+I) vs sunitinib (S) by positive results in phase 3 trials. However, limited data exists on outcomes to
number of IMDC risk factors present. Results: Among 1096 intent-to-treat subsequent therapy in patients progressing on IO-VEGF regimens. Methods: A
(ITT) patients (pts) in both arms, 21%, 61%, and 18% had favorable, in- retrospective analysis was performed on patients with ccRCC at the Memorial
termediate (int), or poor-risk, respectively. Of int-risk pts, 58% had 1 factor Sloan Kettering Cancer Center and Cleveland Clinic Cancer Institute who
(most commonly ,1 y from diagnosis [Dx], 52%; Hb , LLN, 27%; or initiated systemic therapy post IO-VEGF regimens including combinations with
KPS #70%, 10%); and 42% had 2 factors (of these pts, the most common VEGFR tyrosine kinase inhibitors (IO-TKI) and combinations with anti-VEGF
combination of 2 factors was ,1 y from Dx and Hb , LLN, 59%). Of poor-risk monoclonal antibodies (IO-mAB). Patients treated on unreported clinical
pts, 58% had 3 factors, 29% had 4 factors, and few had 5 (10%) or 6 (3%) trials were excluded from the outcomes analysis. The primary objective was
factors. Due to small numbers, pts with 4–6 factors were pooled. At 30-mo to evaluate the overall survival (OS) post IO-VEGF. The secondary objectives
minimum follow-up, RECIST v1.1-confirmed objective response rate (ORR) included objective response rate (ORR) and progression-free survival (PFS)
and complete response (CR) rate per investigator remained consistently higher according to RECIST v1.1. Kaplan-Meier methods and the log-rank test were
with N+I vs S across pts with 1–4 factors, although with S, ORR decreased with used to evaluate time from start of systemic therapy post IO-VEGF to the event
increasing number of factors (Table). Improved progression-free survival (PFS) of interest. Results: Fifty-nine patients were treated after discontinuation of IO-
and overall survival (OS) were seen with N+I over S irrespective of the number VEGF regimens. Prior IO-VEGF regimens included IO-mAB (n = 35, 59%) and
of factors present, including in pts with only 1 risk factor (Table). IO-TKI (n = 24, 41%). IMDC scores at the start of next line of therapy were
Conclusions: N+I showed consistent efficacy across number of IMDC risk favorable in 20%, intermediate in 60% and poor in 20%. Next line of therapy
factors, while S decreased in efficacy with increasing number of factors. included VEGFR-TKI monotherapy (n = 45, 76%), VEGFR-TKI based com-
Efficacy of N+I was superior to S in all int- and poor risk pts. These CheckMate binations (n = 6, 10%), mTOR inhibitors (n = 3, 5%), and unreported clinical
214 results along with prior CheckMate 025 data showing consistent OS trials (n = 5, 9%). VEGFR-TKI containing regimens (n = 51) included
benefit with N monotherapy across IMDC risk categories show a need for cabozantinib (n = 22), axitinib (n = 17), lenvatinib/everolimus (n = 4),
improved prognostic models for immunotherapies in aRCC. Clinical trial in- pazopanib (n = 4), and others (n = 4). Median OS was 24.5 months (95% CI
formation: NCT02231749. 12-NE) with a 12 months OS rate of 63%. The ORR was 27% (14/51) and the
S, S, S, S, S,
median PFS was 6.8 months (95% CI 4.8-11). No difference in post IO-VEGF
N+I,n=125 n=124 N+I, n=189 n=172 N+I, n =125 n=141 N+I, n=55 n=47 N+I, n=35 n=38 OS was observed when comparing IO- TKI vs IO-mAB (log rank p = 0.7).
No. of IMDC risk 0 0 1 1 2 2 3 3 4–6 4–6 Conclusions: Post combination IO-VEGF treatment, most patients received
factors
PFS, HR (95% CI) 1.23 0.77 0.83 0.44 0.86 VEGFR-TKIs. In this setting, VEGFR-TKIs continue to show clinical activity
OS, HR (95% CI)
(0.90–1.69)
1.22
(0.60–0.99)
0.62
(0.62–1.12)
0.72
(0.27–0.73)
0.50
(0.50–1.49)
0.63
similar to historic experiences of patients post VEGF monotherapy.
(0.73–2.04) (0.44–0.87) (0.51–01.02) (0.31–0.82) (0.36–1.10)
ORR, % 39 50 42 38 42 26 44 17 40 16
CR rate, % 8 4 10 1 17 1 13 0 0 0

4577 Poster Session (Board #403), Mon, 1:15 PM-4:15 PM 4578 Poster Session (Board #404), Mon, 1:15 PM-4:15 PM
First-line (1L) immuno-oncology (IO) combination therapies in metastatic Deferred cytoreductive nephrectomy among patients with newly diagnosed
renal-cell carcinoma (mRCC): Results from the international mRCC database metastatic renal cell carcinoma treated initially with sunitinib. First Author:
consortium (IMDC). First Author: Shaan Dudani, Tom Baker Cancer Centre, Bimal Bhindi, University of Calgary, Calgary, AB, Canada
University of Calgary, Calgary, AB, Canada
Background: While the CARMENA trial prompts more caution with upfront
Background: In mRCC, ipilimumab and nivolumab (ipi-nivo) is a 1L treatment cytoreductive nephrectomy (CN) in patients with metastatic renal cell carci-
option. Recent data have also shown efficacy of 1L IO-VEGF (IOVE) inhibitor noma (mRCC), 17% of patients in the sunitinib alone arm underwent deferred
combinations. Comparative data between these two strategies are limited and the CN (dCN). Upfront systemic therapy has been proposed as a potential litmus
efficacy of subsequent therapies remains unknown. Methods: Using the IMDC test to identify patients suitable for CN, but data on outcomes are limited. We
dataset, patients (pts) treated with any 1L IOVE combination were compared to sought to characterize outcomes of dCN after upfront sunitinib relative to
those treated with ipi-nivo. Multivariable Cox regression analysis was performed to sunitinib alone. Methods: Patients with newly diagnosed mRCC receiving
control for imbalances in IMDC risk factors. Results: 188 pts received 1L IO upfront sunitinib were identified from the International mRCC Database
combination therapy: 113 treated with IOVE combinations and 75 with ipi-nivo. Consortium (IMDC) from 2006-2018. All CNs done after initial sunitinib were
Baseline characteristics and IMDC risk factors were comparable between groups. included, excluding CNs performed after sunitinib failure. The outcomes were
When comparing IOVE combinations vs ipi-nivo, 1L response rate (RR) was 33% vs overall survival (OS) and time to treatment failure (TTF). Kaplan Meier and
40% (p=0.39), time to treatment failure (TTF) was 14.3 (95% CI 9.2-16.1) vs multivariable Cox regression analyses were performed; dCN was analyzed as a
10.2 months (95% CI 6.7-15.1, p=0.23), and median overall survival (OS) was not time-varying covariate to account for immortal time bias. Results: The cohort
reached (NR) (95% CI 22.3-NR) vs NR (95% CI 35.1-NR, p=0.17). When ad-
included 708 patients of whom 53 (7.5%) underwent dCN at a median of
justed for IMDC risk factors, the hazard ratio (HR) for TTF was 0.71 (95% CI 0.46-
6.5 months (IQR 3.5,10.5) from diagnosis. Patients in the dCN group were
1.12, p=0.14) and the HR for death was 1.74 (95% CI 0.82-3.68, p=0.14).
more likely to have better Karnofsky performance status (KPS), intermediate
Second-line (2L) treatments were varied. In pts receiving subsequent VEGF-based
therapy, 2L RR was lower in the IOVE (n=20) versus ipi-nivo (n=20) cohort (15% vs IMDC risk, fewer metastatic sites, and response to upfront sunitinib (Table).
45%; p=0.04), though 2L TTF was not significantly different (3.7 vs 5.4 months, There were 604 deaths during a median follow-up of 63 months. Median OS
p=0.40, n=55). The use of IO post IOVE was uncommon and 3/5 pts had PD as best and TTF with dCN were 43.5 and 19.8 months vs. 9.4 and 4.3 months without,
response; 2/5 had PR/SD but their 1L IOVE exposure was short at ,3 months. respectively. Upon multivariable analysis, dCN remained significantly asso-
Conclusions: There does not appear to be a superior 1L IO combination strategy in ciated with OS (HR 0.45, 95%CI 0.31-0.65; p , 0.001) but not TTF (HR
mRCC, as lOVE combinations and ipi-nivo have comparable 1L RR, TTF and OS. 0.73, 95%CI 0.52-1.01; p = 0.056). Conclusions: Patients who received dCN
Most pts received VEGF-based therapy in the 2L. In this group, 2L RR was greater in were carefully selected and achieved long OS. With these benchmark out-
pts who received ipi-nivo, though there was no difference in 2L TTF. comes, optimal selection criteria need to be identified and confirmation of the
role of dCN in a clinical trial is warranted.
IO-VEGF (N=113) Ipi-Nivo (N=75)
IMDC Risk Groups Variable dCN No CN p-value
Favourable 29/92 (32%) 17/64 (27%)
Intermediate 49/92 (53%) 33/64 (52%) Age in years, median (IQR) 61 (53-67) 63 (55-70) 0.08
Poor 14/92 (15%) 14/64 (22%) KPS < 80, n (%) 9 (15) 256 (38) , 0.001
2L Treatments Poor IMDC risk (vs intermediate), n (%) 15 (25) 333 (50) , 0.001
Axitinib 5/34 2/30
Cabozantinib 9/34 2/30 > 1 Metastatic site, n (%) 29 (55) 518 (79) , 0.001
Lenvatinib + Everolimus 2/34 0/30 Best response, n (%)
Nivolumab 5/34 0/30 Stable disease 16 (26) 225 (33) , 0.001
Pazopanib 2/34 9/30
Sunitinib 9/34 15/30
Partial/complete response 28 (46) 110 (16)
Other 2/34 2/30 Progressive disease 12 (20) 218 (32)
Unknown 5 (8) 119 (18)
*All non-significant (p . 0.05).

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Genitourinary (Nonprostate) Cancer 283s

4579 Poster Session (Board #405), Mon, 1:15 PM-4:15 PM 4580 Poster Session (Board #406), Mon, 1:15 PM-4:15 PM
Meta-analysis of randomized clinical trials (RCT) for the adjuvant treatment Dynamic contrast-enhanced MRI to predict intratumoral molecular hetero-
of renal cell carcinoma (RCC) with vascular endothelial growth factor re- geneity in clear cell renal cell carcinoma. First Author: Durga Udayakumar,
ceptor tyrosine-kinase inhibitors (VEGFR TKIs). First Author: Daniel Vargas University of Texas Southwestern, Dallas, TX
Almeida, BP-A Beneficencia Portuguesa de S~ ao Paulo, S~ao Paulo, Brazil
Background: Mutation/inactivation of VHL in clear cell renal cell carcinoma
Background: Although surgery is the cornerstone in the treatment of most (ccRCC) leads to upregulation of hypoxia inducible factors (HIFs) and angio-
cases of localized kidney cancer, up to 30% of patients will experience disease genesis. However, ccRCC is characterized by high intra-tumor heterogeneity
recurrence at three years of follow-up. Three RCTs with VEGFR TKIs (ASSURE, (ITH). Random small samples such as those in percutaneous biopsies are likely
PROTECT and ATLAS) failed to demonstrate improvement in disease-free limited for characterization of molecular alterations in heterogeneous ccRCCs.
survival (DFS). Only S-TRAC trial showed a significant improvement in DFS, We hypothesize that whole-tumor dynamic contrast-enhanced (DCE) magnetic
and was approved by the Food and Drug Administration (FDA). However, the resonance imaging (MRI) is useful to noninvasively identify ITH in ccRCC.
matter remains controversial among genitourinary oncologists. Therefore, we Methods: This IRB-approved, prospective, HIPAA-compliant study, included
performed a meta-analysis to better evaluate the potential benefit of adjuvant 62 ccRCCs. 3T DCE MRI was obtained prior to nephrectomy. Surgical spec-
VEGFR TKIs after curative intent nephrectomy. Methods: Eligible studies were imens were sectioned to match MRI acquisition plane. 182 snap frozen samples
searched in PubMed databases and limited to phase 3 RCT published from (49 tumors) and adjacent uninvolved renal parenchyma (URP) were collected.
January 1996 to December 2018 of US FDA-approved VEGFR TKIs reporting RNA isolations, cDNA library preparation and mRNA sequencing were per-
on patients with RCC treated in the adjuvant setting. A summary hazard-ratio formed using standard protocols. RNA expression in 81 tumor samples were
(HR) of disease-free survival (DFS) was calculated using 95% CIs by random- correlated (Spearman ranked) with % enhancement in a region of interest (ROI)
effects or fixed-effects models on the basis of the heterogeneity of included drawn in the same location of the tumor on pre- and 3 different post-contrast
studies. Results: Four RCT (ASSURE, S-TRAC, PROTECT and ATLAS trials) DCE MRI phases. Gene function overrepresentation (OR) analyses were done on
were selected for analysis, including a total of 4,820 patients. A VEGFR TKI top positively and negatively correlated genes. False discovery rate (FDR) , 0.1
(sunitinib, sorafenib, pazopanib or axitinib) was administered in 2,737 pa- was considered statistically significant. Results: Principal component analysis
tients, and 2,083 received placebo. The summary DFS HR for the overall of . 20,000 genes indicated distinct gene expression in tumors from URP.
population was 0.89 (95% CI 0.79-1.00; p = 0.06). When including the report Unsupervised clustering showed enrichment of ccA samples (better prognosis)
of the ASSURE with the sub-group analysis with high-risk patient population (n compared to ccB samples (worse prognosis). Importantly, ccA and ccB samples
= 3,946), the summary HR for DFS was 0.84 (95% CI 0.75-0.95, p = coexisted in 25% of tumors. DCE-MRI % enhancement correlated with ex-
0,0044). No evidence of publication bias was found. Conclusions: This is the pression of . 300 genes (p , 0.003, FDR , 0.1). OR analyses placed an-
first meta-analysis including the four RCTs in RCC adjuvant setting. This meta- giogenic pathway gene processes and the immune/inflammatory response
analysis failed to demonstrate improvement in DFS for patients receiving a processes within the top 5 positively- and negatively-correlated gene functions,
VEGFR TKI after curative intent nephrectomy. A modest benefit in DFS was respectively. HIF2 target genes correlated positively with % enhancement.
observed in a selected sub-group of patients with higher risk for recurrence. Conclusions: DCE MRI detects specific molecular signatures and may help
There is no data regarding overall survival. overcome the challenges of ITH in ccRCC. Further research is needed to explore
the potential role of DCE MRI to assess response to antiangiogenic and immune-
based therapies.

4581 Poster Discussion Session; Displayed in Poster Session (Board #346), 4582 Poster Session (Board #408), Mon, 1:15 PM-4:15 PM
Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Prognostic significance of CD73 expression in localized renal cell carcinoma
Mon, 4:30 PM-6:00 PM (RCC). First Author: Abhishek Tripathi, University of Oklahoma, Stephenson
Comprehensive genomic profiling (CGP) of upper-tract (UTUC) and bladder Cancer Center, Oklahoma City, OK
(BUC) urothelial carcinoma reveals opportunities for therapeutic and bio-
Background: CD73 or ecto-5’-nucleotidase mediates the de-phosphorylation of
marker development. First Author: Andrea Necchi, Fondazione IRCCS
adenosine monophosphate to adenosine which promotes immunosuppression
Istituto Nazionale dei Tumori, Milan, Italy
in the tumor microenvironment. Its prognostic significance in localized RCC has
Background: To understand the genomic landscape and inform the thera- not been well characterized. Methods: We assessed CD73 protein expression
peutic development of UC, 2463 cases were analyzed by CGP for genomic using immunohistochemistry (Cell Signaling Technology, D7F9A, 1:25) on
alterations (GAs) and for genome wide signatures. Methods: 479 UTUC and tissue microarrays (TMAs) containing tumor tissue from patients with pT1-4,N0-
1984 BUC FFPE tissues (77%/70% primary [PT] and 23%/30% metastatic 1, M0 RCC who underwent nephrectomy.CD73 expression was quantified
tumors [MT] from unmatched patients [pts]) underwent hybrid-capture based using a combined score (CS: intensity x percentage of cells staining positive).
CGP to evaluate all classes of GA. Tumor mutational burden (TMB) was de- CD73 positivity was defined as any CD73 expression on tumor cells (CS . 0).
termined on 1.1 Mbp of sequenced DNA and microsatellite instability (MSI) Patients were categorized into CD73 negative (CS = 0), low ( , median CS) and
was determined on 114 loci. Targetable GA and signatures were assessed high ($ median CS) groups. Clinical data was collected retrospectively.Baseline
according to the ESMO Scale for Clinical Actionability of molecular Targets patient characteristics were compared between CD73 negative and positive
(ESCAT). Results: 48% of UC pts overall harbored $1 tier 1-2 GA suggesting (high + low) groups using the Cochran-Armitage trend test. Multivariable Cox
benefit from approved or investigational targeted therapies (TT). Additionally, regression evaluated associations of CD73 expression with disease-free and
17% had a tier 3 GA that provides a strong rationale for clinical trial con- overall survival (DFS, OS) after adjusting for other baseline prognostic variables.
sideration. Non-FGFR3 kinase fusions were observed in 1% of pts (0.6% Results: Of the 112 patients included, clear cell was the most common his-
UTUC v 1.1% BUC), including BRAF/RAF1 fusions in 0.5%. BRAF mut/ tology (70%) followed by chromophobe (12.5%), and papillary (12%) RCC.
fusions were observed in 2% (49/2463) of cases and were mutually exclusive CD73 expression (CS . 0) was noted in 22% (n = 25) of tumors and was
with FGFR3 GA (p = 0.002). In comparing UC from anatomic sites, there were associated with more advanced stage (T3-4/N+: 37.5% vs. 25%; p = 0.02)
no differences of TMB-H ($20 mut/mb)/MSI-H for PT and MT but UTUC was and a trend towards higher nuclear grade ($3: 48% vs. 35%; p = 0.07). Median
enriched for MSI-H (3.4%) relative to BUC (0.77%, p , 0.001, all TMB-H). follow-up from nephrectomy was 9.7 yrs. In multivariable analysis adjusting
Excluding MSI-H pts, UTUC has lower median TMB (4.35 mut/mb) than BUC for nuclear grade ( , 3 vs. $3) and stage (T1-2, N0 vs. T3-4/N+), tumors
(6.96 mut/mb). FGFR3 GA (26% v 19%, p , 0.05) and specifically short with high CD73 expression had significantly worse DFS and OS (Table).
variants (SV) (20% v 13%) were enriched in UTUC vs BUC. HRAS SV were also Conclusions: CD73 expression was found in 22% of RCC patients and was
enriched in UTUC vs BC (7.3% v 3.0%), attributed to an enrichment in renal associated with adverse pathologic features and poor prognosis independent of
pelvis UC (10.1%) v ureteral UC (1.8%, p , 0.05). RB1 GA were more tumor stage and histologic grade. Our results provide strong rationale for further
frequent in BUC vs UTUC (21% v 7.8% p , 0.001). Conclusions: Against a investigation of the CD73/adenosine pathway as a therapeutic target in RCC.
background of 50% actionability in UC with opportunities for immunotherapy, DFS OS
TT, or combinations thereof, the UTUC cohort is enriched for FGFR3 and CD73 expression N 5-year Adjusted HR p 10-year OS Adjusted HR p
HRAS SV relative to BUC, with the observation of HRAS mutations pre- DFS
dominantly in UC of the renal pelvis, that warrants further investigation into the
distinct modes of oncogenesis for UC as stratified by anatomic origin. These Negative 87 75% Reference 64% Reference
Low 12 50% 1.2 (0.5-3.0) 0.66 71% 0.8 (0.3-2.3) 0.65
results argue strongly for the routine incorporation of CGP prior to systemic High* 13 42% 2.7 (1.3-5.9) 0.01 22% 2.6 (1.2-5.8) 0.02
therapy initiation in metastatic UC.
*$ median CS.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
284s Genitourinary (Nonprostate) Cancer

4583 Poster Session (Board #409), Mon, 1:15 PM-4:15 PM 4584 Poster Session (Board #410), Mon, 1:15 PM-4:15 PM
Atezolizumab plus bevacizumab in non-clear cell renal cell carcinoma (NccRCC) Malignant pheochromocytoma (MP): A comprehensive genomic profiling
and clear cell renal cell carcinoma with sarcomatoid differentiation (ccRCCsd): (CGP) study. First Author: Gennady Bratslavsky, Urologic Oncology Branch,
Updated results of activity and predictive biomarkers from a phase II study. First National Cancer Institute at the National Institutes of Health, Bethesda, MD
Author: Ronan Flippot, Dana-Farber Cancer Institute, Boston, MA
Background: We CGP to characterize the genomic alterations (GA) in MP and
Background: NccRCC and ccRCCsd are aggressive tumors associated with to enable the search for potential therapy targets. Methods: From a series of
poor prognosis and response to therapy. Combination strategies co-targeting 201,766 consecutive clinical cases, 44 cases of clinically advanced MP
VEGF signaling and inhibitory immune checkpoints are highly active in clear- underwent CGP using a hybrid-capture based commercial assay to evaluate
cell renal cell carcinoma, but data is lacking in NccRCC and ccRCCsd. We all classes of GA. Tumor mutational burden (TMB) was determined on 1.1
conducted a multicenter, open-label, single arm phase II trial of atezolizumab Mbp of sequenced DNA and reported as mutations/Mb and microsatellite
plus bevacizumab in NccRCC and ccRCCsd. Methods: Patients with NccRCC instability (MSI) was determined on 114 loci. PD-L1 expression was de-
and ccRCCsd ( . 20% sarcomatoid differentiation), and ECOG performance termined by IHC (DAKO 22C3 antibody). Results: All patients had clinically
status of 0-2 were eligible. Prior systemic treatment was allowed with the advanced recurrent and/or metastatic disease. 23 patients were females and
exception of prior PD-1/PD-L1-directed therapy. Atezolizumab 1200mg and 21 patients were males. There were 34 (77%) of MP known to have origi-
bevacizumab 15mg/kg were administered every 3 weeks until progression, nated in the adrenal gland and 10 (23%) of the MP were sequenced from
unacceptable toxicity, or patient withdrawal. Primary endpoint was objective metastatic site where the exact primary site was unknown. The primary tumor
response rate (ORR) per RECIST 1.1. Exploratory biomarker analyses included was used for sequencing in 14 (32%) of the MP cases and a non-primary
PD-L1 expression on tumor (TC) and immune cells (IC), and spatial analysis of tumor metastatic site (liver, lung, bone, soft tissue, lymph node, kidney,
the immune infiltrate. Results: Sixty patients received at least 1 cycle of peritoneal cavity, and chest wall) in 30 (68%) of the MP cases. There were
treatment, among whom 56 were evaluable for response (17 ccRCCsd and 39 2.3 GA/tumor. The most frequent un-targetable GA were ATRX (25%), TP53
NccRCC). ORR was 34% in the overall population, 53% in ccRCCsd and 26% (21%), SDHB (13%), CTNNB1 (7%), VHL (7%), and CDKN2A/2B, PIK3R2,
in NccRCC. Median progression-free survival was 8.4 months (95%CI, 6.9- NOTCH2 and MEN1 (all 5%). The most frequent potentially targetable GA
16.5). Baseline tumor tissue was available for 36 patients. TC PD-L1 included RET (9%), NF1 (9%) and FGFR1 (5%). PBRM1 GA were found in
expression $1% was associated with improved ORR (9/14, 64%) com- 2% of MAP. Germline mutations in known cancer predisposition genes were
pared to patients with PD-L1 expression , 1% (4/20, 20%). Patients with TC predicted in 8 (18%) of cases involving SDHB (5 cases) and BRCA1, MEN1,
PD-L1 expression $1% who experienced progressive disease as best response and MSH2 (1 case each). The genomic signatures of primary MP were not
had shorter average distance between tumor cells and nearest neighboring significantly different from that obtained from sequencing of metastatic site
immune cells at baseline. Further analysis of the immune tumor microenvi- biopsies. 0 (0%) of 5 MP stained positively for PD-L1 expression. The mean
ronment on an expanded cohort, including IC PD-L1 expression and correlation TMB was 2.95 mutations/Mb, the median TMB was 2.4 mutations/Mb.
with clinical outcomes, is ongoing and will be updated. Conclusions: The There 2 (5%) of MP with TMB $ 10 mutations/MB and 0 (0%) with TMB $
combination of atezolizumab plus bevacizumab is active in NccRCC and 20 mutations/Mb. 0 (0%) of 33 MP evaluated for MSI had a MSI-High status.
ccRCCsd. Candidate predictive biomarkers include PD-L1 expression in TC Conclusions: Although the GA/tumor is relatively low for MP, CGP can reveal
and topological analysis of the immune infiltrate. Clinical trial information: important potential therapy targets including RET, NF1 and FGFR1. MP do
NCT02724878. not reveal strong potential for immunotherapies with low TMB, absence of
MSI-High status and low (2%) PBRM1 mutation frequencies.

4585 Poster Session (Board #411), Mon, 1:15 PM-4:15 PM TPS4586 Poster Session (Board #412a), Mon, 1:15 PM-4:15 PM
Metastatic penile (mPSCC), uterine cervical (mCSCC), and skin (mSSCC) A randomized phase II study of atezolizumab plus recombinant human IL-7
squamous cell carcinomas: A comparative genomic profiling (CGP) study. (CYT107) or atezolizumab alone in patients with locally advanced or met-
First Author: Joseph Jacob, SUNY Upstate Medical University, Syracuse, NY astatic urothelial carcinoma (mUC): A Cancer Immunotherapy Trials Net-
work Trial (CITN-14). First Author: Evan Y. Yu, University of Washington,
Background: We compared the genomic alteration (GA) profiles of mCSCC, mPSCC and
mSSCC to study impact on the targeted and immunotherapy options for the men and Seattle, WA
women suffering from these refractory cancers. Methods: 78 mPSCC, 604 mCSCC and Background: Atezolizumab is a regulatory-approved PD-L1 antagonistic anti-
338 mSSCC underwent CGP using a hybrid-capture based assay. Tumor mutational body for the post-platinum mUC setting. Responses to atezolizumab are highly
burden (TMB) was determined on 1.1 Mbp of sequenced DNA and and microsatellite
efficacious in a subset of patients, but suboptimal or absent in most patients. IL-
instability (MSI) was determined on 114 loci. Results: The HPV+/CDKN2A- status was
significantly more frequent in the mCSCC than mPSCC or mSSCC (P , 0.0001). The GA/
7 (CYT107) is a homeostatic growth factor that promotes proliferation, dif-
tumor frequencies were similar for mCSCC and mPSCC, but significantly higher in ferentiation, and survival of T lymphocytes. We recently demonstrated CYT107
mSSCC (P , 0.0001). TP53 mutations were more common in mSSCC (UV light ex- significantly increases peripheral absolute lymphocyte and T cell numbers in
posure) and mPSCC (likely due to loss of an original HPV+ status). TERT, NOTCH1 and metastatic castration-resistant prostate cancer patients when administered
FAT1 GA were more frequent in mPSCC and mSSCC whereas PIK3CA GA were more after sipuleucel-T. We hypothesize expansion of T cells by CYT107 may improve
common in mCSCC. MTOR pathway targets (GA in STK11, FBXW7 and PTEN) were responses to PD-L1 inhibition. To test this hypothesis, we designed a ran-
more common in mCSCC than mPSCC or mSSCC. MSI high status was extremely rare in domized trial (NCT03513952) in mUC comparing the combination of CYT107
all cases. TMB $ 10/20 mut/Mb frequencies were noteworthy in mPSCC and mCSCC but and atezolizumab to atezolizumab alone. Methods: Patients with ECOG PS #2
extraordinarily high in mSSCC. Examples of patients with mCSCC, mPSCC and mSSCC and RECIST v1.1 measurable mUC with disease recurrence after platinum-
responding to targeted and immunotherapies will be presented. Conclusions: Although based chemotherapy are eligible. A safety run-in of 6 patients with staggered
mCSCC, mPSCC and mSSCC share a variety of clinicopathologic features, the 3 tumor enrollment to atezolizumab plus CYT107 will be followed by randomization
types can be sharply differentiated on CGP. The TP53, CDKN2A and HPV status of the
tumor types differ significantly with HPV+ higher in the mCSCC group. There are op-
if ,2 patients experience a DLT. An additional 48 patients will then be ran-
portunities for targeted therapies in all groups predominantly identified in the MTOR domized 1:1 to atezolizumab 1200 mg IV q3wks with or without CYT107 10 ug/
pathway. The relatively high numbers of cases with significantly elevated TMB in all 3 kg IM qwk X 4, started 1 wk before atezolizumab. The primary endpoint is
tumor types suggest that immunotherapies would be beneficial in a large subset of RECIST v1.1 ORR, with H0 14.8% and HA 45%, one-sided a 0.10; power 88%.
patients. An interim futility analysis will be performed after 24 randomized patients have
mPSCC (78 cases) mCSCC (604 cases) mSSCC (338 cases)
their first disease assessment; cessation of the trial will occur if an O’Brien-
Fleming futility boundary of ,-0.0063 in the ORR scale is observed between
GA/case 5.8 4.9 9.6
TP53 58% 11% 86% the experimental and control arm. Secondary endpoints include clinical benefit
CDKN2A 47% 4% 55%
TERT 45% 16% 45% rate, PFS, DOR, OS, results by PD-L1 expression stratification, and safety.
FAT1
NOTCH1
33%
22%
7%
4%
34%
39%
Exploratory correlative evaluations of tumor-infiltrating immune cells, interferon g
PTCH1 3% 1% 9% expression, inflammatory gene expression, ELISPOT, T cell receptor se-
PIK3CA 19% 37% 9%
CD274 amp 2% 4% 2% quencing, serum metabolite levels, gut microbiome, and PK analyses will be
PTEN
FBXW7
5%
12%
14%
14%
4%
4%
performed. Current state: Trial accrual has begun and is anticipated to complete
STK11 6% 12% 3% around mid-2020. Clinical trial information: NCT03513952.
MDM2 1% 3% 3%
MSI-High 0% 2% 2%
TMB (median) 3.6 5.2 43.2
TMB ‡10/‡20 18%/8% 24%/7% 71%/63%
HPV+ 29% 69% 5%
TP53 Mutated and HPV+ 22% 7% 31%
TP53 Mutated and HPV- 73% 93% 89%

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Genitourinary (Nonprostate) Cancer 285s

TPS4587 Poster Session (Board #412b), Mon, 1:15 PM-4:15 PM TPS4588 Poster Session (Board #413a), Mon, 1:15 PM-4:15 PM
A phase 3 randomized study of neoadjuvant chemotherapy (NAC) alone or in DUTRENEO Trial: A phase II randomized trial of DUrvalumab and TREm-
combination with nivolumab (NIVO) 6 BMS-986205 in cisplatin-eligible elimumab as NEOadjuvant approach in muscle-invasive urothelial bladder
muscle invasive bladder cancer (MIBC). First Author: Guru Sonpavde, Dana- cancer (MIBC) patients prospectively selected by immune signature scores.
Farber Cancer Institute, Boston, MA First Author: Enrique Grande, MD Anderson Cancer Center Madrid, Madrid,
Spain
Background: Immuno-oncology (IO) therapies have revolutionized the treat-
ment (tx) of pts with advanced bladder cancer (advBC). For pts with cisplatin- Background: Cisplatin-based neoadjuvant chemotherapy (CT) followed by
eligible MIBC, the recommended tx regimen is cisplatin-based NAC prior to cystectomy improves overall survival in patients (pts) with MIBC. Immune
radical cystectomy (RC). However, since only » 30% of pts achieve a path- checkpoint inhibitors as single agents are approved in pts with advanced UC.
ologic complete response (pCR) translating to improved long-term outcomes Combination of both programmed cell death ligand-1 (PD-L1) and cytotoxic
with approved regimens, new therapies are needed. PD-L1 expression is as- T-lymphocyte-associated antigen-4 (CTLA-4) checkpoints might be syner-
sociated with aggressive BC and has been shown to increase in BC after NAC, gistic. Durvalumab (DU) is a selective, engineered, human IgG1 monoclonal
suggesting that the PD-1/PD-L1 axis is a valid therapeutic target. Additionally, antibody (mAb) that blocks PD-L1 binding to PD-1 and CD80. Tremelimumab
expression of indoleamine 2,3-dioxygenase (IDO) is higher in BC than in (TRE) is an anti-CTLA-4 mAb of the lgG2 isotype. Pembrolizumab and ate-
normal bladder tissue and is associated with advanced disease and poor zolizumab, have shown a promising activity (including significant pathologic
clinical outcome. BMS-986205, a selective, potent, once-daily oral IDO1 complete responses (pT0)) in MIBC pts candidate to cistectomy, with better
inhibitor that works early in the IDO1 pathway to reduce kynurenine pro- results in those pts with PD-L1 overexpression. It is expected that the dual
duction, has demonstrated clinical activity in combination with NIVO (anti– targeting of the immune system with an anti-PDL1 + antiCTLA4 such as DU
PD-1) in pts with IO tx–naive advBC who had $ 1 prior line of therapy (ORR, and TRE in the neoadjuvant setting may improve these outcomes. In addition,
37%). Taken together, these data provide a rationale for investigating NAC + the most precise selection of pts according to a molecular INF-gamma sig-
NIVO + BMS-986205 in MIBC. Here we describe a randomized, partially nature is intended to increase the efficacy. Methods: This is a prospective and
blinded, phase 3 study evaluating the efficacy and safety of NAC 6 NIVO 6 randomized phase II, open-label study conducted in urothelial MIBC pts di-
BMS-986205 followed by RC and continued IO tx in pts with MIBC agnosed of T2-T4 and/or N+ candidates to cystectomy, ECOG 0-1 and ade-
(NCT03661320). Methods: Pts aged $ 18 years with previously untreated quate organ function. Pts will be treated according to the score of a pro-
MIBC (clinical stage T2-T4a, N0, M0), creatinine clearance $ 50 mL/min, and inflammatory signature (PIS) determined with Nanostring technology. Pts
predominant UC histology who are eligible for cisplatin-based NAC and RC will with a low PIS will receive standard cisplatin-based neoadjuvant therapy (22
be enrolled. Pts with evidence of positive lymph node; metastatic BC; or prior pt). Pts with a high PIS will be randomized 1:1 to receive cisplatin-based
systemic therapy, radiotherapy, or surgery for BC other than TURBT are not neoadjuvant therapy (22 pt) or DU 1500 mg + TRE 75 mg every 4 weeks x 3
eligible. Pts will be randomized to receive NAC (gemcitabine/cisplatin; arm A), cycles (22 pt). If more than 8 responses (pT0) are observed in first 22 pts
NAC + NIVO + oral placebo (arm B), or NAC + NIVO + BMS-986205 (arm C) included in DU+TRE arm, 24 additional pts will be recruited in this arm.
followed by RC (all arms); arms B and C will receive continued IO tx. Primary Primary objective is to assess the antitumor activity of DU+TRE measured as
endpoints include pCR after neoadjuvant tx and event-free survival (arms C vs pT0 rate in pts with a positive PIS. Disease free survival and safety profile will
A; arms B vs A). Secondary endpoints are overall survival and safety. This global be also evaluated. Tissue, plasma and urine samples will be collected for
study in 28 countries began accrual in Nov 2018 and has a target enrollment of translational studies. Clinical trial information: NCT03472274.
1200 pts. Clinical trial information: NCT03661320.

TPS4589 Poster Session (Board #413b), Mon, 1:15 PM-4:15 PM TPS4590 Poster Session (Board #414a), Mon, 1:15 PM-4:15 PM
ALBAN: An open label, randomized, phase III trial, evaluating efficacy of A phase III, randomized, open label, multicenter, global study of first-line
atezolizumab in addition to one year BCG (bacillus Calmette-Guerin) bladder (1L) durvalumab in combination with standard of care (SOC) chemotherapy
instillation in BCG-naive patients with high-risk nonmuscle invasive bladder and durvalumab in combination with tremelimumab and SOC chemotherapy
cancer (AFU-GETUG 37). First Author: Morgan Roupret, Urology University versus SOC chemotherapy alone in patients with unresectable locally
Paris 6, Paris, France advanced or metastatic urothelial cancer (UC). First Author: Matt D. Galsky,
Department of Medicine, Icahn School of Medicine at Mount Sinai, Tisch
Background: The majority of patients with bladder cancer are diagnosed with
Cancer Institute, New York, NY
non-muscle invasive bladder cancer (NMIBC). Trans-urethral resection of
bladder tumor (TURBT) followed by Bacillus Calmette-Guerin (BCG) therapy is Background: Despite high response rates to 1L SoC for locally advanced or
the standard of care for most patients with high-risk NMIBC, although clinical metastatic UC chemotherapy (gemcitabine + cisplatin or gemcitabine + car-
outcomes remain poor. Data from preclinical and clinical studies suggest that boplatin for patients who are cisplatin-ineligible [poor performance status,
programmed death-ligand 1 PD-L1 inhibition may enhance antitumor activity impaired renal function, comorbidities]), most patients experience disease
of BCG therapy. Atezolizumab, an anti–PD-L1 monoclonal antibody, is ap- progression. Novel strategies such as combining chemotherapy and immuno-
proved in the United States and the European Union as monotherapy when therapy offer hope for improving clinical outcomes. Durvalumab is a selective,
patients are ineligible for cisplatin or after prior platinum therapy. This approval high affinity, engineered human IgG1 mAb that blocks PD-L1 binding to PD-1
was based on the results from phase II and III IMvigor210/211 trials. We have and CD80. Tremelimumab is a human IgG2 mAb directed against CTLA-4. The
designed ALBAN, a phase III trial to evaluate the efficacy and safety of ate- mechanisms of action of PD-1 and CTLA-4 are nonredundant, so targeting
zolizumab given in combination with BCG versus BCG alone in patients with both checkpoint pathways may have additive or synergistic efficacy compared
high-risk NMIBC. Methods: In 30 centers in France, ALBAN (NCT03799835) with monotherapy. Studies of platinum-based chemotherapy combined with
is enrolling patients with histologically documented NMIBC who have not checkpoint blockade in other tumor types have yielded improved efficacy with
received prior BCG therapy. Eligible patients have to have high-risk features acceptable safety and support exploration of this approach for 1L locally ad-
(T1 staging, high grade, or in situ carcinoma), ECOG PS 0-2, and a tissue vanced or metastatic UC. Methods: NILE (NCT03682068) is a randomized,
sample for PD-L1 testing (by Ventana SP142 immunohistochemistry assay). open-label, multicenter, global trial that will enroll approximately 1265 patients
Patients are randomized 1:1 to: (Arm A) BCG alone (six weekly instillations of with histologically or cytologically documented, unresectable, locally advanced,
BCG, followed by three weekly maintenance instillations at 3, 6, 12 months), or metastatic transitional cell carcinoma of the urothelium. Patients will
or (Arm B) atezolizumab (1200 mg; IV; q3w for up to 1 year) combined with be randomized (1:1:1) to durvalumab + SoC chemotherapy, durvalumab +
BCG given as in arm A. Treatment with atezolizumab and BCG are continued tremelimumab + SoC chemotherapy, or SoC chemotherapy as 1L therapy.
in the absence of unacceptable toxicity, for 1 year. The primary endpoint is Primary endpoints are progression-free survival using blinded independent
recurrence-free survival in the intent-to-treat population. Secondary efficacy central review assessments per RECIST 1.1 and overall survival (OS). Secondary
endpoints include overall survival, PFS, complete response, disease wors- endpoints include objective response rate, OS at 24 months, proportion of
ening, and quality of life. Safety, biomarkers, and other exploratory endpoints patients alive and progression free at 12 months, duration of response, disease
will also be evaluated. Enrollment began in December 2018 with a target of control rate, time from randomization to second progression, and health-related
614 pts. Clinical trial information: NCT03799835. quality of life. Safety, pharmacokinetics, immunogenicity, and biomarkers will
also be assessed. The study opened for enrollment in September 2018. 2019
American Society of Clinical Oncology, Inc. Reused with permission. This
abstract was accepted and previously presented at the 2019 Genitourinary
Cancers Symposium. Clinical trial information: NCT03682068.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
286s Genitourinary (Nonprostate) Cancer

TPS4591 Poster Session (Board #414b), Mon, 1:15 PM-4:15 PM TPS4592 Poster Session (Board #415a), Mon, 1:15 PM-4:15 PM
Bacillus Calmette-Guerin (BCG) with or without pembrolizumab (pembro) for A phase III, randomized, open label, multicenter, global study of efficacy and
high-risk (HR) nonmuscle invasive bladder cancer (NMIBC) that is persistent safety of durvalumab in combination with gemcitabine+cisplatin (G+C) for
or recurrent following BCG induction: Phase III KEYNOTE-676 study. First neoadjuvant treatment followed by durvalumab alone for adjuvant treatment
Author: Ashish M. Kamat, The University of Texas MD Anderson Cancer in muscle-invasive bladder cancer (MIBC) (NIAGARA). First Author: Thomas
Center, Houston, TX Powles, Barts Cancer Institute, Queen Mary University of London, Royal Free
NHS Trust, London, United Kingdom
Background: Intravesical instillation of BCG is standard of care for patients
(pts) with HR NMIBC. However, many pts have persistent/recurrent HR Background: Management of MIBC includes both surgery and systemic
NMIBC after BCG induction and are at increased risk for progression to therapy. Neoadjuvant, cisplatin-based combination chemotherapy has
muscle-invasive disease. Interim data from the phase 2 KEYNOTE-057 demonstrated improved pathologic complete response (pCR), event-free
study has shown that the PD-1 inhibitor pembro had promising efficacy survival (EFS), and OS compared with radical cystectomy alone. Many
in HR NMIBC as monotherapy. KEYNOTE-676 (NCT03711032) is a ran- patients still develop recurrence, including progression to metastasis. Novel
domized, comparator-controlled, phase 3 trial to evaluate efficacy and safety strategies such as combining chemotherapy and immunotherapy in a
of pembro plus BCG in pts with persistent/recurrent HR NMIBC after BCG neoadjuvant setting and consolidating response post cystectomy in the
induction therapy. Methods: Pts are randomly assigned 1:1 to continue BCG adjuvant setting may improve clinical outcomes. Durvalumab is a selective,
therapy alone or receive BCG plus pembro 200 mg every 3 weeks. Treatment high affinity, engineered human IgG1 mAb that blocks PD-L1 binding to PD-
is stratified by carcinoma in situ (CIS) histology (presence/absence), PD-L1 1 and CD80. PD-L1 inhibition with durvalumab, in combination with a
combined positive score ($10/˂10), and timing of NMIBC persistence/ standard neoadjuvant regimen (G+C), may improve immune-mediated an-
recurrence (0 to #6, ˃6 to #12, or ˃12 to #24 mo). Pts are eligible if titumor response and increase the rates of pathologic responses and long-
they are $18 years of age with histologically confirmed persistent/recurrent term survival. Methods: NIAGARA (NCT03732677) is a Phase 3, ran-
HR NMIBC of the bladder after adequate BCG induction therapy, have domized, open-label, multicenter, global study that will enroll ~1050 pa-
undergone cystoscopy/transurethral resection of bladder tumor within tients randomized (1:1) to durvalumab and G+C combination (Arm 1) or G+C
12 weeks before randomization, have no concurrent extravesical disease, (Arm 2) as neoadjuvant chemotherapy prior to radical cystectomy. Following
and have an ECOG PS score of 0-2. Responses are assessed by cystoscopy radical cystectomy and during adjuvant therapy, patients in Arm 1 will re-
and blinded independent central review of urine cytology and biopsy (as ceive durvalumab monotherapy for 8 cycles (8 months); patients in Arm 2
applicable) every 12 weeks for years 1-2 and every 24 weeks for years 3-5 will receive no adjuvant treatment. Patients with resectable MIBC (clinical
and by computed tomography urography every 18 months through year 5. stage T2N0M0-T4aN0M0) with transitional cell histology planning to
Treatment will continue with pembro for up to 2 years and BCG for 3 years or undergo a radical cystectomy will be included. Primary endpoints are pCR
until confirmed HR NMIBC persistence, recurrence, or disease progression, rates at time of cystectomy following neoadjuvant treatment and EFS.
unacceptable toxicity, or pt/physician decision to withdraw. Primary end Secondary and exploratory endpoints include proportion of patients who
point is complete response rate in pts with CIS. Secondary end points are achieve pathologic response ,P2 (stages Pa, P1, and carcinoma in situ) at
event-free survival (EFS), recurrence-free survival, overall survival, disease- time of cystectomy following neoadjuvant treatment, EFS at 24 months,
specific survival, time to cystectomy, 12-month EFS rate in all pts, duration metastasis-free survival, proportion of patients who undergo cystectomy, and
of response (DOR), 12-month DOR rate in pts with CR and safety and OS at 5 years. Safety, patient-reported outcomes, pharmacokinetics, im-
tolerability. Recruitment began in November 2018 and will continue munogenicity, and biomarkers will also be assessed. Enrollment opened in
until ~550 pts are enrolled. Clinical trial information: NCT03711032. Dec 2018. Clinical trial information: NCT03732677.

TPS4593 Poster Session (Board #415b), Mon, 1:15 PM-4:15 PM TPS4594 Poster Session (Board #416a), Mon, 1:15 PM-4:15 PM
EV-103: Enfortumab vedotin plus pembrolizumab and/or chemotherapy for PrE0807 phase Ib feasibility trial of neoadjuvant nivolumab (N)/lirilumab (L)
locally advanced or metastatic urothelial cancer. First Author: Christopher J. in cisplatin-ineligible muscle-invasive bladder cancer (BC). First Author:
Hoimes, University Hospitals Cleveland Medical Center/Case Western Petros Grivas, University of Washington, School of Medicine, Seattle, WA
Reserve University, Cleveland, OH
Background: Neoadjuvant cisplatin-based chemotherapy before radical
Background: PD-(L)1-targeted immune checkpoint inhibitors (CPIs), such as cystectomy (RC) improves outcomes but ~50% of patients (pts) are cisplatin-
pembrolizumab (P), are approved for patients (pts) with locally advanced or unfit. Anti-PD(L)1 agents can prolong overall survival (OS) in platinum-
metastatic urothelial cancer (la/mUC) who progress after platinum, are in- resistant advanced BC and have shown high pathologic complete response
eligible for first-line (1L) cisplatin (PD-L1 positive), or are ineligible for 1L rate (pCR) and safety as single agent in phase II trials in the neoadjuvant
platinum. The ORR for CPI in all treated pts is ~25%, and a combination setting. The combination of anti-PD-1 and anti-KIR agents is feasible and
approach may provide additional benefit. Enfortumab vedotin (EV), an in- very attractive based on complementary and non-overlapping roles in reg-
vestigational antibody-drug conjugate, delivers the microtubule-disrupting ulating adaptive and innate immune response as well as impacting the
agent monomethyl auristatin E to cells expressing Nectin-4, found in 97% function CD8+ T and NK-cells. Higher CD8+ T cell density (TCD) at RC tissue
of mUC pt samples (Petrylak ASCO 2017). In a phase 1 study (NCT02091999), correlates with longer OS. We hypothesize, that combining anti-PD1 (N) with
EV (1.25 mg/kg) was generally well tolerated with a confirmed ORR of 43% in anti-KIR (L) is safe and feasible as neoadjuvant therapy in cisplatin-unfit pts
112 mUC pts (Rosenberg ASCO-GU 2019). These encouraging results, with the and results in high CD8+ TCD at RC. Methods: Phase Ib multi-institutional
potential for an enhanced immune response, suggest that EV+P may improve trial evaluating 2 doses (4 weeks apart) of N alone or N+L in 2 cohorts; pts
response rates and extend response durability. Methods: EV-103, a phase 1b will be assigned sequentially to N (Cohort 1), and if there is no negative safety
trial for non–resectable la/mUC pts with no prior CPI, added an additional signal after the first 12 pts, subsequent pts will be assigned to N+L (Cohort 2).
4 cohorts (Parts 2 and 3) in an Oct 2018 amendment. It is now expected to Key eligibility: cT2-4aN0-1M0 stage, $20% tumor at TURBT, adequate
enroll ~159 pts. Dose escalation pts (EV+P, 1L or 2L) must be ineligible for 1L organ function, no autoimmune disease within 2 years, no concurrent in-
cisplatin-based chemotherapy or have disease progression during/following vasive upper urinary tract carcinoma or other active cancer. Primary end-
treatment with $1 platinum-containing regimen. Dose expansion (Part 1) point: safety based on CTCAE v5.0 measured as the rate of $G3 treatment
will evaluate EV+P in 1L (Cohort A) and 2L settings (Optional Cohort B). Part 2 related adverse events (AE). Key secondary endpoints: CD8+ TCD absolute
will evaluate 1L EV+cisplatin (Cohort D), 1L EV+carboplatin (Cohort E), and 1L and % change between TURBT and RC, % of pts who do not get RC within
or 2L EV+gemcitabine (Optional Cohort F). Part 3 (Cohort G) will evaluate 1L 6 weeks after neoadjuvant treatment due to treatment-related AE, % pCR,
EV+P+cisplatin or carboplatin, depending on pts cisplatin-eligibility. In all cohorts, recurrence-free survival, and evaluation of biomarkers in tumor tissue, blood,
pts receive EV on Days 1 and 8 of each 3-week cycle. In combination therapies, urine. Rates of $Grade 3 AE with neoadjuvant treatment will be reported
pts receive P, cisplatin, and carboplatin on Day 1 or gemcitabine on Days 1 and along with 90% exact binomial CI. In Cohort 1, maximum CI width is 0.51; in
8 of each cycle. The primary objective is to assess the safety/tolerability of EV+P Cohort 2, it is 0.36. Our hypothesis is that the change in CD8+ TCD between
and/or chemotherapy. Secondary objectives are establishing EV recommended TURBT and RC will be about 3 CD8+ T cells / 100 tumor cells within HPF. Up
dose for combination therapies, assessing ORR per RECIST for all cohorts and to 43 pts will be enrolled for 36 eligible, treated pts (12:N, 24:N+L). Cohort
per iRECIST for therapies with pembrolizumab, as well as assessing disease 1 and 2 have 81% and 98% power, respectively, to detect the hypothesized
control rate, duration of response, PFS, OS, PK, and biomarkers. The study difference with 1-sided type I error rate of 0.05. Trial is open to accrual in
opened Oct 2017. Clinical trial information: NCT03288545. US. Clinical trial information: NCT03532451.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Genitourinary (Nonprostate) Cancer 287s

TPS4595 Poster Session (Board #416b), Mon, 1:15 PM-4:15 PM TPS4596 Poster Session (Board #417a), Mon, 1:15 PM-4:15 PM
A phase III randomized open label study comparing bempegaldesleukin PDIGREE: An adaptive phase 3 trial of PD-inhibitor nivolumab and ipili-
(NKTR-214) plus nivolumab to sunitinib or cabozantinib (investigator’s choice) mumab (IPI-NIVO) with VEGF TKI cabozantinib (CABO) in metastatic un-
in patients with previously untreated advanced renal cell carcinoma. First treated renal cell cancer (Alliance A031704). First Author: Tian Zhang, Duke
Author: Nizar M. Tannir, University of Texas MD Anderson Cancer Center, University Medical Center, Durham, NC
Houston, TX
Background: First-line treatment of mRCC has rapidly changed to include
Background: Bempegaldesleukin (NKTR-214) is a CD122-preferential IL-2 IPI-NIVO or CABO, with clinical benefit of each based on the Checkmate 214
pathway agonist that stimulates proliferation and activation of tumor antigen- and CABOSUN trials. Combination immunotherapy with VEGF therapies have
specific CD8+ T cells and natural killer cells within the tumor microenviron- shown benefit in the JAVELIN 101 and KEYNOTE 426 trials over sunitinib. It is
ment and increases PD-1/PD-L1 expression. These properties make bempe- yet unclear which patients (pts) benefit most from combination immunotherapy-
galdesleukin (NKTR-214) a potentially promising agent for combination VEGF inhibitors, and the optimal sequence of drugs. Methods: In an adaptive,
therapy with checkpoint inhibitors that target and inhibit the PD-1/PD-L1 randomized, multicenter, phase 3 trial (Alliance A031704, PDIGREE), pts will
pathway. In phase 1 studies, NKTR-214 plus nivolumab demonstrated en- start treatment with induction IPI 1mg/kg and NIVO 3mg/kg intravenously (IV)
couraging objective response rates (ORR) in first-line renal cell carcinoma once every 3 weeks. Key inclusion criteria include clear cell mRCC, IMDC in-
(RCC) and an acceptable safety profile. Immunotherapy with NKTR-214 plus termediate or poor risk, Karnofsky performance status .70, and no prior
nivolumab may lead to greater clinical benefit than tyrosine kinase inhibitors treatments for mRCC. Based on 3-month radiographic assessment (after
(TKIs), standard-of-care agents, in this patient population. Methods: This completing IPI-NIVO combination), pts with complete responses (CR) will un-
multicenter, randomized, open-label phase 3 study (NCT03729245) will dergo maintenance NIVO 480mg IV every 4 weeks, pts with progression of
evaluate the efficacy and safety of bempegaldesleukin (NKTR-214) plus disease (PD) will switch to CABO 60mg oral daily, and pts with non-CR/non-PD
nivolumab compared with investigator’s choice of TKI (sunitinib or cabo- will be randomized to NIVO 480mg IV every 4 weeks versus NIVO 480mg IV
zantinib) in patients with previously untreated advanced or metastatic RCC every 4 weeks with CABO 40mg oral daily. Randomization will be stratified by
with clear cell component. Exclusion criteria include active brain metastasis IMDC risk criteria and presence of bone metastases. The primary endpoint of the
and autoimmune disease. Approximately 600 patients will be randomized in a study is overall survival (OS). We hypothesize that 3-year OS rate will improve to
1:1 ratio, stratified by PD-L1 status ($1% vs , 1% or indeterminate), In- 70% for NIVO-CABO compared to 60% for NIVO alone; to achieve 85% power
ternational Metastatic RCC Database Consortium prognostic score (1-2 [in- with a two-sided alpha of 0.05 and exponential distribution, 696 patients will be
termediate risk] vs 3-6 [poor risk]); and TKI (sunitinib or cabozantinib; randomized. Accounting for 30% patients with either CR or PD, and 5% dropout
cabozantinib percentage to be capped at 50%). Combination therapy will from toxicity, up to 1046 pts will be enrolled. Key secondary endpoints include
consist of bempegaldesleukin (NKTR-214) 0.006 mg/kg intravenously (IV) PFS, 12-month CR rate, ORR based on RECIST 1.1 and irRECIST criteria, and
every 3 weeks (Q3W) plus nivolumab 360 mg IV Q3W until progression or death toxicity profiles. Quality of life will be assessed based on the FKSI-19, PROMIS-
or maximum of 2 years. TKI therapy will consist of sunitinib 50 mg orally once fatigue, and EQ5D-5L questionnaires. Biomarkers associated with CR and
daily (QD) for 4 weeks followed by 2 weeks off or cabozantinib 60 mg orally QD. association of IL-6 with treatment benefit will be assessed. Other tissue-based
Primary objectives are ORR by blinded independent central radiology (BICR) and plasma-based biomarkers are planned. Enrollment will begin this
assessment and overall survival. Secondary objectives are progression-free year. Support from UG1CA189823, U24CA196171; https://acknowl-
survival by BICR, safety, predictive value of PD-L1 expression, and quality of edgments.alliancefound.org. Clinical trial information: NCT03793166.
life. Enrollment is ongoing. Clinical trial information: NCT03729245.

TPS4597 Poster Session (Board #417b), Mon, 1:15 PM-4:15 PM

PROSPER: A phase III randomized study comparing perioperative nivolumab
(nivo) versus observation in patients with renal cell carcinoma (RCC) un-
dergoing nephrectomy (ECOG-ACRIN 8143). First Author: Lauren Christine
Harshman, Dana-Farber Cancer Institute, Boston, MA
Background: The anti-PD-1 antibody nivo improves overall survival (OS) in
metastatic RCC and is well tolerated. There is no standard adjuvant (adjuv)
systemic therapy that increases OS over surgery alone for non-metastatic RCC.
Priming the immune system prior to surgery with anti-PD-1 has shown an OS
benefit compared to a pure adjuv approach in mouse solid tumor models.
Multiple ph 2 studies in bladder, lung and breast cancers have shown re-
markable pathologic responses with neoadjuvant (neoadj) PD-1 blockade. Two
ongoing ph 2 studies of perioperative nivo in M0 RCC patients are showing
preliminary feasibility and safety with no surgical delays (NCT02575222;
NCT02595918). PROSPER RCC (NCT03055013) aims to improve clinical
outcomes by priming the immune system prior to nephrectomy with neoadj
nivo and continued engagement with adjuv blockade in patients with high risk
RCC compared to surgery alone. Methods: This global, unblinded, phase 3
National Clinical Trials Network study is currently accruing patients with
clinical stage $T2 or TanyN+ RCC of any histology planned for nephrectomy.
Oligometastases are permitted if can be rendered NED. We amended the study
to enhance accrual and patient quality of life by changing nivo dosing to
480mg q4 wks and requiring baseline tumor biopsy only in the nivo arm. The
investigational arm receives 1 dose of nivo prior to surgery followed by 9 adjuv
doses. The control arm undergoes standard nephrectomy followed by obser-
vation. Randomized patients are stratified by clinical T stage, node positivity,
and M stage. Accrual of 805 patients provides 84.2% power to detect a 14.4%
absolute benefit in recurrence-free survival (RFS) at 5 years assuming the
ASSURE historical control of ~56% to 70% (HR = 0.70). The study is powered
to evaluate a significant increase in OS (HR 0.67). Critical perioperative
therapy considerations such as safety, feasibility, and quality of life endpoints
have been integrated. PROSPER RCC embeds a wealth of translational work
aimed at investigating the impact of the baseline immune milieu, the changes
induced by neoadjuvant anti-PD-1 priming, and how both may predict clinical
outcomes. Clinical trial information: NCT03055013.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
288s Genitourinary (Prostate) Cancer

5000 Oral Abstract Session, Fri, 2:45 PM-5:45 PM 5001 Oral Abstract Session, Fri, 2:45 PM-5:45 PM
Impact of darolutamide (DARO) on pain and quality of life (QoL) in patients Interest of short hormonotherapy (HT) associated with radiotherapy (RT) as
(Pts) with nonmetastatic castrate-resistant prostate cancer (nmCRPC). First salvage treatment for metastatic free survival (MFS) after radical prostatectomy
Author: Karim Fizazi, Institut Gustave Roussy, University of Paris-Sud, (RP): Update at 9 years of the GETUG-AFU 16 phase III randomized trial
Villejuif, France (NCT00423475). First Author: Christian Carrie, Leon Berard Center, Radio-
therapy Department, Lyon, France
Background: DARO is a structurally distinct androgen receptor antagonist for
which in vitro and phase 1/2 studies suggest low risk of adverse events (AEs) Background: RT is the standard salvage treatment after RP. The role of HT is not
and drug–drug interaction. In the ARAMIS study of DARO in nmCRPC, formally demonstrated to date. This trial assessed the efficacy of RT alone vs RT+
metastasis-free survival (MFS) was significantly prolonged vs placebo (PBO) HT in terms of progression-free survival (PFS), metastase-free survival (MFS)
(40.4 vs 18.4 mo; hazard ratio [HR] 0.41; 95% confidence interval [CI] and overall survival (OS) in patients with biological relapse (BR) after RP. After a
0.34–0.50; P , 0.001) and interim overall survival (OS) favored DARO (HR median follow-up (FU) duration of 5.3 years, we previously reported [Carrie C,
0.71; 95% CI 0.50–0.99; P = 0.045). Methods: 1509 pts were randomized Lancet Oncol 2016] a benefit in PFS (80% vs 62% PFS free at 5 years; p ,
2:1 to DARO 600 mg (two 300 mg tablets) twice daily (n = 955) or PBO (n = 0.0001) in the combined arm, whatever the risk subgroups. Methods: Patients
554) while continuing androgen deprivation therapy (ADT). Primary end- (pts) were randomized (1:1) to RT alone or RT+HT (goserelin, for 6 months). The
point was MFS. Secondary endpoints included OS and time to pain pro- randomization was stratified according to radiotherapy modality and risk group.
gression (assessed by Brief Pain Inventory Short Form). QoL was assessed by Low risk was defined as Gleason score , 8, surgical margins+, psa doubling
European Organisation for Research and Treatment of Cancer QoL Prostate time . 8 months and no seminal vesicle involment. Assuming a 45% 5-year
Cancer module (EORTC-QLQ-PR25) at baseline (BL) and every 16 wks until PFS of 45% in the RT arm, the trial required 369 pts per arm to detect an
end of treatment. Analysis of time to deterioration in EORTC-QLQ-PR25 improvement of 12% on PFS in RT+HT arm (90% power and 5% bilateral alpha
subscales, defined as first occurrence of a minimally important difference risk), possibly translating into a 10% gain in OS (75% to 85% with 80%
(half the standard deviation of BL value), used Kaplan–Meier estimators and power). Biological relapse (BR) was defined according to ASTRO-consensus.
stratified Cox proportional hazard models. Results: DARO significantly Results: At the time of data cutoff (March 2019), the median duration follow-up
delayed pain progression vs PBO (40.3 vs 25.4 mo; HR 0.65; 95% CI was 112 months. We confirm the benefit of RT+HT on PFS (HR = 0.54 [CI95% =
0.53–0.79; P , 0.001); this was maintained beyond end of study treatment. 0.43-0.68] ; p , 0.0001) whatever the risk subgroup (HR = 0.47 [CI95% =
Time to deterioration of EORTC-QLQ-PR25 outcomes showed statistically 0.28-0.80] and 0.56 [CI95% = 0.44-0.73] for low and high risk patients,
and clinically significant delays with DARO vs PBO for urinary symptoms respectively. Metastatic free survival (MFS) is significantly improved in the
(25.8 vs 14.8 mo; HR 0.64; 95% CI 0.54–0.76; P , 0.01). Time to de- combined arm (HR = 0.73 [CI95% = 0.54-0.98] ; p = 0.034) with 69%
terioration of hormonal treatment-related symptoms was comparable with [CI95% = 63-74] versus 75% [CI95% = 70-80] of MFS at 10 years for RT alone
DARO vs PBO (18.9 vs 18.4 mo; HR 1.06; 95% CI 0.88–1.27; P = 0.52). and RT+HT, respectively. Conclusions: Salvage radiotherapy combined with
DARO was well tolerated. Exposure-adjusted incidences (pts per 100 years’ short term HT significantly improved 10-years metastatic free survival compared
exposure) of AEs of interest were similar/lower with DARO vs PBO (fatigue/ with salvage radiotherapy alone. GETUG-16 considered in the context of
asthenic conditions [11.3 vs 11.1], hypertension [4.7 vs 5.1], hot flush [3.7 previously published results from RTOG-9601, confirm that this strategy can be
vs 4.1], fracture [3.0 vs 3.5], falls [2.7 vs 4.1], cognitive disorder [0.3 vs considered as the new standard for salvage treatment after radical prostatec-
0.2], and seizure [0.2 vs 0.2]). Conclusions: For nmCRPC pts, DARO pro- tomy. Clinical trial information: NCT00423475.
longs MFS, is well tolerated, maintains QoL, and delays worsening of pain
and disease-related symptoms compared with PBO. Clinical trial informa-
tion: NCT02200614.

5002 Oral Abstract Session, Fri, 2:45 PM-5:45 PM 5003 Oral Abstract Session, Fri, 2:45 PM-5:45 PM
PSMA heterogeneity and DNA repair defects in prostate cancer. First Author: Updated results from a randomized phase II study of cabazitaxel (CAB)
Alec Paschalis, The Institute of Cancer Research, London, United Kingdom versus abiraterone (ABI) or enzalutamide (ENZ) in poor prognosis metastatic
CRPC. First Author: Kim N. Chi, BC Cancer, Vancouver, BC, Canada
Background: Prostate-specific membrane antigen (PSMA) is a promising target
for theranostics in metastatic castration resistant prostate cancer (mCRPC). Background: The treatment for poor prognosis mCRPC includes taxanes and
Methods: Membranous PSMA (mPSMA) expression was immunohistochemi- androgen receptor (AR) targeted therapy, however the optimal treatment is
cally evaluated in castration sensitive (CSPC) (n = 38) and mCRPC (n = 60) undefined. Methods: Patients (pts) with poor prognosis (liver metastases, early
tissue biopsies, and associations with molecular aberrations (next-generation CRPC ( , 12 months from ADT start), and/or . 3 of 6 poor prognostic criteria
sequencing; NGS) and clinical outcome were determined. Results: mPSMA (Chi et al, Annals of Oncol, 2016)) were randomized to receive CAB (Arm A) or
expression was significantly higher (p = 0.005) in mCRPC biopsies (median H- AR targeted therapy (Arm B, ABI or ENZ by investigator choice) with cross over
score [interquartile range]; 55.0 [2.8-117.5]) compared to CSPC biopsies (17.5 at progression. Primary objective was to determine the clinical benefit rate
[0.0-60.0]). Furthermore, patients with higher mPSMA expression ( . median (CBR) (PSA decline $50% (PSA50), objective response (OR), or stable
H-score) at diagnosis had higher Gleason Grade (p = 0.04) and shorter OS (p = disease (SD) $ 12 weeks). Plasma was sampled serially for circulating tumour
0.006). Critically, 42% (16/38) of CSPC biopsies and 27% (16/60) of mCRPC DNA (ctDNA). Results: 95 pts were randomized (Arm A: 45, Arm B: 50): 18%
biopsies were completely negative for mPSMA expression. In addition, CSPC and had liver mets, 88% early CRPC and 30% had . 3 of 6 poor prognostic criteria.
mCRPC biopsies expressing mPSMA demonstrated marked intra-tumor het- 52% of pts had prior docetaxel, half for castration sensitive disease. Table
erogeneity in expression levels, commonly exhibiting areas without detectable summarizes 1st-line therapy outcomes. Baseline ctDNA fraction . 15%
PSMA (CSPC – 100%; mCRPC – 84%), while heterogeneous mPSMA ex- (median) was associated with shorter 1st-line progression-free survival (PFS)
pression between metastases from the same patient was also observed. Sub- (median 2.8 vs 8.4 m, HR = 2.54, P , 0.001) and overall survival (OS)
sequent genomic analysis showed that mCRPC patients with deleterious DNA (median 14.0 vs 38.7 m, HR = 2.64, P = 0.001). ctDNA alterations in AR,
damage repair (DDR) aberrations have higher (p = 0.016) mPSMA expression TP53, PI3K pathway, RB1 and DNA repair were detected in 53%, 45%, 31%,
(87.5 [25.0-247.5]) than those without these (20 [0.3-98.8]). Furthermore, 9 23%, and 21% of pts. Shorter PFS and OS were associated with AR gain (HR
of the 11 patients (82%) responding to PARP inhibition had a mPSMA H-Score 2.57 (95% CI 1.63-4.06); HR 3.59, (1.9-6.69), respectively) and TP53
above the median. The association between mPSMA expression and DDR defects (HR 2.62 (CI 1.65-4.15); HR 3.33 (CI 1.8-6.14), respectively). Pts
aberrations was validated in an independent cohort with known DDR aberrations. with concurrent defects in TP53 and RB1 had a trend for worse PFS/OS than
Tumors with DDR aberrations had significantly higher mPSMA (ATM 212.5 pts with TP53 defect alone. AR rearrangements predicted to disrupt the ligand
[136.3-300] p = 0.005; BRCA2 300 [165-300] p = , 0.001) than unselected binding domain were detected in 6% of pts and had a shorter PFS (HR = 2.60
mCRPC biopsies (55.0 [2.75-117.5]). Finally, analyses of 122 mCRPC biopsy (1.11 - 6.09)) with a trend for shorter OS (HR = 2.27 (0.89 - 5.81)).
transcriptomes confirmed a negative correlation between PSMA and BRCA2 Conclusions: In this poor prognosis cohort, 1st-line treatment with CAB had a
mRNA expression (p = 1.5x10-5). Conclusions: mPSMA expression in CSPC and higher clinical benefit rate than treatment with ABI/ENZA. Elevated ctDNA and
mCRPC exhibits marked intra- and inter-patient heterogeneity, limiting the genomic alterations in AR and TP53 were prognostic. Supported in part by
clinical utility of PSMA-targeted theranostics. We show for the first time that Sanofi. Clinical trial information: NCT02254785.
DDR gene aberrations associate with high mPSMA expression and may serve as
Arm A (CAB) Arm B (ABI/ENZ) HR (95% CI) P
predictive biomarkers for PSMA-targeted therapies.
CBR (%) 88 70 - 0.043
PSA50 (%) 61 62 - 1.00
OR (%) 23 17 - 0.72
SD > 12 weeks (%) 26 8 - 0.007
Median PFS (m) 5.8 3.1 0.89 (0.57-1.38) 0.59
Median OS (m) 37.0 15.5 0.57 (0.31-1.03) 0.06

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Genitourinary (Prostate) Cancer 289s

5004 Oral Abstract Session, Fri, 2:45 PM-5:45 PM 5005 Oral Abstract Session, Fri, 2:45 PM-5:45 PM
TAXOMET: A French prospective multicenter randomized controlled phase II TOPARP-B: A phase II randomized trial of the poly(ADP)-ribose polymerase
study comparing docetaxel plus metformin versus docetaxel plus placebo in (PARP) inhibitor olaparib for metastatic castration resistant prostate cancers
mCRPC. First Author: MARC Pujalte Martin, Centre Antoine Lacassagne, (mCRPC) with DNA damage repair (DDR) alterations. First Author: Joaquin
Nice, France Mateo, The Institute of Cancer Research & The Royal Marsden, London,
United Kingdom
Background: Docetaxel (DOCE) is a standard of care in metastatic castration-
resistant prostate cancer (mCRPC). Several retrospective cohort studies Background: We previously reported the antitumor activity of olaparib (400mg
suggest a decrease in PC incidence and mortality with metformin (MET). MET BID) against molecularly unselected mCRPC (TOPARP-A; Mateo et al NEJM
has also demonstrated anti-tumor activity in PC preclinical models, with in- 2015). We now report TOPARP-B, a phase II trial for patients with mCRPC
crease apoptosis when added to DOCE. The addition of MET could enhance preselected for putatively pathogenic DDR alterations. Methods: Patients with
DOCE efficacy in mCRPC patients (pts). Methods: TAXOMET is a phase II mCRPC progressing after $ 1 taxane chemotherapy underwent targeted se-
prospective multicentric randomized controlled trial. Non-diabetic mCRPC pts quencing of tumor biopsies and were deemed eligible when alterations
were assigned 1:1 to receive DOCE 75mg/m2 every 21 days + prednisone (P) (germline or somatic; mono- or bi-allelic) in any DDR gene were detected.
5 mg twice a day and either MET 850mg twice a day (arm A) or placebo (arm Patients were randomized 1:1 under a “pick-the-winner” design to 400mg or
B), up to 10 cycles. The primary end point was PSA response rate ($50% 300mg of olaparib BID, aiming to exclude #30% response rate (RR) in either
decrease). Main secondary endpoints included objective response rate (ORR, arm. The primary endpoint RR was defined as radiological response (RECIST
according to RECIST v1.1), clinical and biological progression-free survival 1.1) and/or PSA50% fall and/or CTC count conversion (Cellsearch; $5 to , 5),
(PFS), overall survival (OS), toxicity and quality of life (QoL). Comparisons confirmed after 4-weeks. Analyses of RR per gene alteration subgroup was pre-
between arm A and B were performed using Chi² test for qualitative data and planned. Secondary endpoints included progression-free survival (PFS), tol-
Log-rank test for survival data. Results: From January 2013 to December erability. Results: Overall, 98 patients (median age 67.6y) were randomized,
2015, 99 pts were randomized (50 pts in arm A and 49 pts in arm B) in 10 with 92 patients treated and evaluable for the primary endpoint (70 RECIST-
french centers, and 95 pts were evaluable. No difference was observed be- evaluable; 89 PSA50%-evaluable; 55 CTC-evaluable). All had progressed on
tween arm A and arm B in PSA-response rate (72% in both arms), ORR (28% in ADT; 99% were post-docetaxel, 90% post-abiraterone/enzalutamide, 38%
both arms), clinical or biological mPFS (7.3 months vs 5.8 months p = 0.848) post-cabazitaxel. The overall RR was 54% (95%CI 39-69%, meeting threshold
and mOS (24.2 months (95CI: 17.2 – 33.7) vs 19.7 months (95CI: 14.8 – for primary endpoint) in the 400mg cohort and 37% (95%CI 23-53%) in the
36.8), p = 0.53), respectively. There was no difference between arms in 300mg cohort. With a median follow-up of 17.6 months (mo), the overall
adverse events, except a trend for diarrhea to be more common with MET (70% median PFS (mPFS) was 5.4 mo. Subgroup analyses per altered gene identified
in arm A vs 50% in arm B, p = 0.072), but few grade 3-4 events. There was no indicated response rates for: BRCA1/2 of 80% (24/30; mPFS 8.1mo); PALB2
difference in QoL according to QLQ-C30 score between the two arms during 57% (4/7; mPFS 5.3mo); ATM 37% (7/19; mPFS 6.1mo); CDK12 25% (5/20;
the treatment period. Conclusions: This is the first prospective randomized mPFS 2.9mo); others [ATRX, CHEK1, CHEK2, FANCA, FANCF, FANCG,
controlled trial to evaluate the combination of MET with DOCE+P in mCRPC. FANCI, FANCM, RAD50, WRN] 20% (4/20; mPFS 2.8mo). The highest
The addition of MET has no meaningful clinical benefit in this setting. Clinical PSA50% response rates were observed in the BRCA1/2 (22/30; 73%) and
trial information: NCT01796028. PALB2 (4/6; 67%) subgroups. Conclusions: Olaparib has antitumor activity
against heavily pre-treated mCRPC with DDR gene defects, with BRCA1/2
aberrant tumors being most sensitive but with confirmed responses in patients
with other DDR alterations. Clinical trial information: NCT01682772.

5006 Oral Abstract Session, Fri, 2:45 PM-5:45 PM 5007 Oral Abstract Session, Fri, 2:45 PM-5:45 PM
First results from TITAN: A phase III double-blind, randomized study of Decreased fracture rate by mandating bone-protecting agents in the EORTC
apalutamide (APA) versus placebo (PBO) in patients (pts) with metastatic 1333/PEACE III trial comparing enzalutamide and Ra223 versus enzalutamide
castration-sensitive prostate cancer (mCSPC) receiving androgen deprivation alone: An interim safety analysis. First Author: Bertrand F. Tombal, Université
therapy (ADT). First Author: Kim N. Chi, BC Cancer, Vancouver, BC, Canada Catholique de Louvain, Brussels, Belgium
Background: TITAN was designed to determine whether APA, a selective next- Background: Skeletal fractures, pathological or not, are a frequent and
generation androgen receptor inhibitor, plus ADT improves radiographic underestimated side-effect of systemic treatment of metastatic castration re-
progression-free survival (rPFS) and overall survival (OS) compared with PBO sistant prostate cancer (mCRPC). The ERA223 trial (NCT02043678) was re-
plus ADT in pts with mCSPC. Methods: In this randomized, double-blind phase cently unblinded following the report of a significant increase in the fracture rates
3 study, pts with mCSPC regardless of extent of disease were randomized (1:1) when abiraterone is combined with Ra223. Hence, FDA and EMA advised
to APA (240 mg/d) or PBO, added to ADT, in 28-day cycles. Pts with prior against this combination. The question whether mandated use of bone protecting
treatment (tx) for localized disease or prior docetaxel for mCSPC were allowed. agents (BPA), zoledronic acid or denosumab, would have mitigated the fracture
All pts received continuous ADT. Dual primary end points were rPFS and OS. risk and whether this risk also exists in the enzalutamide/Ra223 combination is
Secondary end points were time to a) initiation of cytotoxic chemotherapy, b) presently unknown. Methods: The phase III EORTC-1333-GUCG/PEACEIII
pain progression, c) chronic opioid use, d) skeletal-related event. Time-to- (NCT02194842) trial compares enzalutamide vs. a combination of Ra223
event end points were estimated by Kaplan-Meier and Cox proportional hazards and enzalutamide in asymptomatic or mildly symptomatic mCRPC patients
methods. This first planned OS interim analysis took place after ~50% of (https://www.eortc.org/research_field/clinical-detail/1333/). After the unblinding
expected events. Results: 525 pts were randomized to APA and 527 to PBO. of ERA223, the trial was amended (v4.0, April 19, 2018) to mandate that all
Median age was 68 y; 8% had prior tx for localized disease; 11% had prior patients must start a BPA. We report the fracture rate in the safety population of
docetaxel. 63% and 37% had high- or low-volume disease, respectively. At 146 treated patients as of 28/01/2019. Results: Overall, 54.2% of the patients
median 22.6 mo follow-up, 66% APA and 46% PBO pts remained on tx. APA in the enza/Ra223 arm and 51.4% of the enza arm did not receive BPA; 18.0%
significantly improved rPFS (HR, 0.48; 95% CI, 0.39-0.60; p , 0.0001), in the enza/Ra223 arm and 27.0% in the enza arm did not use BPA at ran-
with a 52% reduction in risk of death or radiographic progression; benefit was domization, but started during protocol treatment according to the v4.0
observed across all subgroups analyzed. Median rPFS was not reached in the amendment. 27.8% and 21.6% respectively, received BPA as of randomization.
APA group and 22.1 mos in the PBO group. APA also significantly improved OS In total, 45.8% of enza/Ra223 patients and 48.6% of enza only patients receive
(HR, 0.67; 95% CI, 0.51-0.89; p = 0.0053), with a 33% reduction in risk of bone protection on treatment. The fracture rate is reported in the table.
death. Median OS was not reached in the APA or PBO group. Time to initiation Conclusions: There is a 13% risk of fracture with enzalutamide in asymptomatic
of cytotoxic chemotherapy was significantly improved with APA (HR, 0.39; mCRPC, in line with previous reports. This risk is significantly increased to 33%
95% CI, 0.27-0.56; p , 0.0001). Based on these results, the independent when Ra223 is added to enzalutamide. Strikingly, the risk is almost abolished by
data monitoring committee recommended unblinding to allow crossover of mandatory continuous administration of BPA starting at least 6 weeks before the
PBO pts to receive APA. Rates of grade 3/4 adverse events (AEs) (42% APA, first injection of Ra223, thus emphasizing the importance of treating mCRPC
41% PBO) were similar, and discontinuations due to AEs (8% APA, 5% PBO) patients with BPA. Clinical trial information: NCT02194842.
were low. Conclusions: In the TITAN study in pts with mCSPC, including pts
Enza/Ra223 Enza Enza/RA223 Enza
with high- and low-volume disease and prior docetaxel, addition of APA to ADT With BPA With BPA Without BPA Without BPA Total
significantly improved rPFS and OS, and the safety profile was tolerable. These Bone fracture n=33 n=36 n=39 n=38* n=146
results support the addition of APA to ADT for tx of pts with mCSPC. Clinical
N 1 0 13 5 19
trial information: NCT02489318. % 3% - 33% 13% 13%
95% CI 0-16% - 19-50% 4-28% 8-20%
(*)BPA started after fracture in 1 patient

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
290s Genitourinary (Prostate) Cancer

5008 Oral Abstract Session, Fri, 2:45 PM-5:45 PM 5009 Poster Discussion Session; Displayed in Poster Session (Board #121),
Alliance A031201: A phase III trial of enzalutamide (ENZ) versus enzalutamide, Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
abiraterone, and prednisone (ENZ/AAP) for metastatic castration resistant Sat, 4:30 PM-6:00 PM
prostate cancer (mCRPC). First Author: Michael J. Morris, Memorial Sloan Phase 1b/2 study of enzalutamide (ENZ) with LY3023414 (LY) or placebo (PL)
Kettering Cancer Center, New York, NY in patients (pts) with metastatic castration-resistant prostate cancer (mCRPC)
after progression on abiraterone. First Author: Christopher Sweeney, Lank
Background: Androgen receptor (AR) signaling is an important growth mech-
Center for Genitourinary Oncology, Dana-Farber Cancer Institute, Boston, MA
anism in mCRPC, providing the rationale for treatment with AR axis inhibitors
such as ENZ and AAP. Targeting AR with anti-androgens such as ENZ can result Background: Preclinical and phase 1 results suggest PI3K/mTOR pathway
in compensatory autocrine and paracrine androgenic stimulation. Therefore, inhibition may enhance androgen receptor inhibition. We report the results of a
using ENZ with the androgen biosynthesis inhibitor AAP to dampen these re- double-blind, placebo-controlled, randomized Phase 1b/2 study of ENZ6LY (a
sistance mechanisms could improve clinical outcomes relative to ENZ alone. dual PI3K/mTOR inhibitor) in pts with mCRPC who progressed on abiraterone.
Methods: Men with progressive mCRPC by Prostate Cancer Working Group 2 Methods: Phase 1b pts received single-agent LY 200 mg twice daily (BID) for 1
criteria were eligible. Prior treatment with taxanes for mCRPC and any prior wk prior to starting LY+ENZ. Phase 2 pts were randomized 1:1 to 160 mg daily
treatment with ENZ or AAP was exclusionary. Patients (pts) were randomized 1:1 ENZ with PL or 200 mg BID LY on a 28-d cycle. The primary objective was
to ENZ or ENZ/AAP at standard FDA-approved doses. Randomization was progression-free survival (PFS: serological, radiographic [rPFS], or death) by
stratified by prior chemotherapy and Halabi prognostic three risk groups. Cas- PCWG2 criteria. Secondary objectives were rPFS, safety, decline in PSA, and
trating therapy was maintained. The primary endpoint was overall survival (OS) PK. Exploratory biomarker analyses included outcomes by presence of androgen
defined as the date of randomization from date of death or last follow-up. The log- receptor variant 7 (AR-V7). 92 primary PFS events were needed for the study to
rank test had 90% power to detect a hazard ratio for OS of 0.77 with a one-sided have at least 80% power at one-sided alpha=0.20. Results: LY+ENZ was
type I error rate of 0.025. Secondary endpoints included radiographic pro- tolerable during Phase 1b with 1 dose limiting toxicity observed in 13 enrolled
gression free survival (rPFS) and on-treatment PSA declines. Exploratory end- pts. Mean LY exposures remained in an efficacious range despite a 30% average
points included imaging changes, and changes in serum biomarkers such as decrease when combined with ENZ. In Phase 2, 129 pts were randomized to
androgens, angiokines, and circulating microRNA and RNA. The primary LY+ENZ (N=65) and PL+ENZ (N=64) (Table). Median PCWG2-PFS was
analysis was based on the stratified log-rank test adjusting on the stratification 3.7 mos (LY+ENZ) vs 2.9 mos (PL+ENZ) (HR 0.66, 95% CI 0.43, 0.99; p-value
factors. Results: Between January 2014 and August 2016, 1311 men were 0.0208). Conclusions: Combination LY+ENZ had a clinically manageable
randomized: 657 to ENZ and 654 to ENZ/AAP. Groups were well balanced safety profile. The primary end-point of PCWG2-PFS was met and is supported
between arms, including stratification variables. 15.6% of pts were high risk, by a clinically meaningful delay in rPFS in AR-V7 negative pts. The biomarker
35.3% intermediate, and 48.1% low. Median OS was 33.6 mo (95% CI 30.5- data provide important insights to inform future development strategies. Clinical
36.4) and 32.7 mo (29.9-35.4) respectively, two-sided p = 0.53. Fifty percent trial information: NCT02407054.
PSA decline rate was 80% vs. 76.5%. Grade 3-5 adverse events (AE) (all at- LY+ENZ PL+ENZ Unstratified HR
tributions) were 55.6% and 68.8% respectively. Treatment discontinuation due N=65a N=64a (95% CI), P-value
to AEs occurred in 5% and 12%, pt withdrawal in 5% and 13%, and progression rPFS, median (mos) 7.5 5.3 0.68 (0.41, 1.14), 0.069
or death in 57% and 48% of pts respectively. Conclusions: Addition of abir- rPFS AR-V7 negative, median (mos) 13.2 (n=51) 5.3 (n=54) 0.52 (0.28, 0.95), 0.028
rPFS AR-V7 positive, median (mos) 5.5 (n=9) 3.6 (n=8) 0.99 (0.27, 3.63), 0.991
aterone acetate to enzalutamide did not prolong survival in men with mCRPC. >50% reduction in PSA, % 20 25
The combination resulted in more AEs than enzalutamide alone. Support: Adverse event (AE), All / Grade ‡3, % 98 / 48 98 / 42
U10CA180821, U10CA180882, U24CA196171; https://acknowledgments. Discontinuation due to AE, % 20 6
Fatigue b 29 / 23 / 11 28 / 20 / 3
alliancefound.org. Clinical trial information: NCT01949337. Nausea b 23 / 28 / 8 20 / 11 / 2
Diarrhea b 28 / 23 / 6 8/6/2
a
Number of pts, unless stated otherwise b Treatment-emergent AEs in .30% of all pts, Grade 1 / 2 / $3, %

5010 Poster Discussion Session; Displayed in Poster Session (Board #122), 5011 Poster Discussion Session; Displayed in Poster Session (Board #123),
Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Sat, 4:30 PM-6:00 PM Sat, 4:30 PM-6:00 PM
Pembrolizumab (pembro) plus enzalutamide (enza) in abiraterone (abi)- Copy number analysis to identify tumor suppressor genes associated with
pretreated patients (pts) with metastatic castrate resistant prostate cancer enzalutamide (Enza) resistance and poor prognosis in metastatic castration-
(mCRPC): Cohort C of the phase 1b/2 KEYNOTE-365 study. First Author: resistant prostate cancer (mCRPC) patients. First Author: Xiangnan Guan,
Peter C.C. Fong, Auckland City Hospital, Auckland, New Zealand Knight Cancer Institute, Oregon Health & Science University, Portland, OR
Background: Pembro has activity as monotherapy in pts with pretreated ad- Background: Although enza prolongs life in mCRPC pts, the development of
vanced mCRPC. A phase 2 study suggested that pembro + enza after pro- drug resistance and subsequent disease progression is nearly universal. Seeking
gression on enza may have clinical activity. Data from cohort C (pembro + enza) to clarify molecular mechanisms that underlie enza resistance, we analyzed
of KEYNOTE-365 (NCT02861573), a phase 1b/2 umbrella study to test whole genome sequencing (WGS) and RNA sequencing (seq) of tumors ob-
combinations in mCRPC, are presented. Methods: Pts who were unsuccessful tained from patients with enza-naive or -resistant mCRPC. Methods: One
with or became intolerant to $4 weeks of abi in the prechemotherapy mCRPC hundred and one men with mCRPC who underwent image-guided biopsy and
state, with either PSA or radiologic progression within 6 mo before screening subsequent WGS were included (n = 64 with enza-naive and n = 37 with enza-
were included. Pts received pembro 200 mg IV Q3W with enza 160 mg/day resistant mCRPC). The differential copy number alteration (CNA) events
orally. Primary end points were safety and PSA response rate (confirmed PSA enriched in enza-resistant vs. naı̈ve samples were determined, and the prog-
decrease $50%). Key secondary end points were investigator-assessed ORR nostic significance of differential CNAs was assessed. RNA-seq data were
(RECIST v1.1), disease control rate (DCR: CR+PR+SD $6 mo), time to PSA evaluated to confirm that CNAs correlated with changes in gene expression of
progression, rPFS, and OS. Results: 69 pts began treatment (median age, 69 relevant loci and to identify potentially druggable targets selectively activated in
y; visceral disease, 26%; measurable disease, 36%). Median (95% CI) follow- tumors with specific CNAs. Results: Copy number loss was more common than
up was 9 (7-13) mo. Efficacy is outlined in the table. Treatment-related AEs gain in enza-resistant tumors. Specifically, we identified 123 protein-coding
occurred in 63 (91%) pts; most frequent ($20%) were fatigue (30%), rash genes that were more commonly lost in enza-resistant samples—eight of which
(23%), and nausea (22%). Grade 3/4 treatment-related AEs occurred in 28 were previously described tumor suppressor genes. There was a strong con-
(41%) pts; most common was rash (10%); no deaths were from treatment- cordance of copy number loss and reduced mRNA expression of these genes.
related AEs. Conclusions: The pembro + enza combination showed sustained We identified one gene from this list of eight genes whose copy number loss was
activity in abi-pretreated chemotherapy-naive mCRPC. AEs were considered associated with poor overall survival (median overall survival from date of CRPC
tolerable for the treatment combination; incidence of rash resolved with was 19.1 months in tumors with gene loss vs. 42.0 months in intact tumors,
standard-of-care treatment. Clinical trial information: NCT02861573. hazard ratio 3.8 [1.46–9.8], log-rank p = 0.003). Finally, Master Regulator
PSA response, n/N (%) analysis determined that tumors with copy number loss of this poor prognosis
Total population 18/67 (27) gene had activation of several potentially targetable factors, including the ki-
Measurable disease 10/25 (40)
ORR, per RECSIT v1.1 pts w/measurable disease, n/N (%) 5/25 (20); 2 CR, 3 PR nases Akt and PLK1. Conclusions: Copy number loss of specific tumor sup-
DCR per RECIST v1.1, n/N (%)
Total population 23/69 (33)
pressor genes is associated with enza resistance in mCRPC patients. Previously
Measurable disease 8/25 (32) unappreciated molecular subsets of enza-resistant CRPC were identified, in-
Duration of response per RECIST v1.1, median (range), mo 8.3 (0.0+ to 13.0+) cluding one subset associated with poor clinical outcome.
DOR, ‡6 mo, n (%) 3 (75)
Time to confirmed PSA progression in pts with measurable disease, median 18 (15-48)
(95% CI), wk
No confirmed PSA progression at 27 wk in pts with measurable disease, % 50
rPFS, PCWG-modified RECIST, median (95% CI), mo 6 (4-8)
OS, median (95% CI), mo Not reached (12 to not reached)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Genitourinary (Prostate) Cancer 291s

5012 Poster Discussion Session; Displayed in Poster Session (Board #124), 5013 Poster Discussion Session; Displayed in Poster Session (Board #125),
Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Sat, 4:30 PM-6:00 PM Sat, 4:30 PM-6:00 PM
Diagnostic performance of 18F-DCFPyL in the OSPREY Trial: A prospective Association of noninvasive, radiographic measurement of prostate-specific
phase 2/3 multicenter study of 18F-DCFPyL PET/CT imaging in patients (Pts) membrane antigen (PSMA) expression with response to PSMA-targeted ra-
with known or suspected metastatic prostate cancer (mPC). First Author: dionuclide therapy (TRT). First Author: Panagiotis J. Vlachostergios, Division of
Michael J. Morris, Memorial Sloan Kettering Cancer Center, New York, NY Hematology & Medical Oncology, Weill Cornell Medical College & New York-
Presbyterian Hospital, New York, NY
Background: Accurate detection of prostate cancer is imperative to patient
management, yet standard imaging methods perform poorly in accurately Background: Prostate surface membrane antigen (PSMA) is usually overex-
detecting mPC. 18F-DCFPyL is a novel PET imaging agent that selectively pressed in PC and is enriched in castration-resistant tumors. PSMA-TRT is of
binds to prostate-specific membrane antigen, a recognized target for prostate increasing interest to the field. Many in the field of theranostics have assumed
cancer. OSPREY was a prospective, multicenter study in pts with either newly that PSMA uptake on imaging is a pre-requisite for response. We have conducted
diagnosed high-risk prostate cancer (cohort A), or known or suspected mPC a number of trials which have incorporated PSMA imaging, but have not selected
(cohort B). Here we focus on Cohort B. Methods: 117 men planned for biopsy patients for treatment based upon imaging results and performed an analysis
of recurrent or mPC received 18F-DCFPyL. Pts underwent image-guided bi- examining the relationship between imaging and response. Methods: Men with
opsy. Sensitivity, positive predictive value (PPV), and safety of 18F-DCFPyL mCRPC had either planar radiolabeled J591 imaging (111In-J591 and/or
PET/CT were the key endpoints for Cohort B. 18F-DCFPyL PET/CT scans were 177Lu-J591) or 68Ga-PSMA11 PET/CT. Visual scores were assigned based
evaluated by three independent, blinded central readers; and results were upon PSMA uptake in tumors compared to liver uptake and scored on a 0-4 scale.
compared to histopathology as the truth standard. Results: The sensitivity and Imaging scores were associated with PSA decline ($30%, $50%) using Cox
PPV of 18F-DCFPyL PET/CT as compared to histopathology ranged from 92.9- regression analysis. As several studies were dose-escalation in nature with
98.6% (lower bound of 95% CI: 84.0-91.6%) and 81.2-87.8%, respectively. prior demonstration of dose-response, we controlled for dose administered.
Diagnostic performance by anatomic location showed high sensitivity and Results: 216 men with metastatic CRPC, median PSA 72.45ng/dl, were treated
high PPV in all sites of disease (Table). Only two (1.7%) cohort B pts with PSMA-TRT as follows: 177Lu-J591 (n=136), 177Lu-PSMA-617 (n=38),
experienced $1 drug-related AE (dysgeusia and generalized rash), both were Lu-J591 + Lu-PSMA-617 combination (n=6), 225Ac-J591 (n=7), 90Y-J591
mild (Grade 1) in severity. Conclusions: 18F-DCFPyL PET/CT was well tolerated (n=29). 116 (53.7%) pts received low dose and 100 (46.3%) high dose as
and demonstrated high sensitivity and PPV in accurately detecting nodal, previously defined in the individual studies. 55 (25.5%) pts had low PSMA
bone, and visceral/soft tissue metastases. A positive 18F-DCFPyL PET/CT scan expression by imaging (VS 0-1) whereas 161 (74.5%) had high PSMA (VS 2-4).
is highly likely to represent pathologically proven distant disease, demon- High PSMA expression was associated with more frequent PSA decline
strating the potential of 18F-DCFPyL as a PET imaging agent to favorably ($30%: 39.2 vs 17.9% p=0.003; $50%: 27.5 vs 8.9% p=0.004). When
influence treatment planning. Clinical trial information: NCT02981368. controlling for dose level, this association remained significant for low ($30%:
25 vs 8.6% p=0.04) and high doses of radionuclide therapy ($50%: 34.9 vs
Sensitivity PPV
Range for 3 readers (%) Range for 3 readers (%) 9.5% p=0.02). 13 (6%) pts with no PSMA uptake (VS=0) had PSA declines.
Anatomic Location N Evaluable (lower bound of 95% CI (%)) (lower bound of 95% CI (%)) Conclusions: This is the first study to formally analyze response to PSMA-TRT
All tissues 92-93 92.9-98.6 (84.0-91.6) 81.2-87.8 (72.9-80.0) by PSMA imaging expression in an unselected patient population. The level of
Bone 43-44 90.3-96.9 (74.0-82.9) 78.9-82.3 (66.0-69.5) PSMA expression measured by imaging is associated with the chance of
Lymph Node 39 93.3-100 (77.6-86.0) 80.0-90.6 (66.8-75.0) response. However, a subset of patients without any significant PSMA uptake
Viscera/Soft Tissue 10 88.9-100 (54.0-65.5) 90.0-100 (57.0-62.0)
on imaging did demonstrate response to PSMA-TRT, indicating that imaging
cannot exclude all patients that might benefit. Clinical trial information:
NCT03545165, NCT03276572, NCT03042468, NCT02552394.

5014 Poster Discussion Session; Displayed in Poster Session (Board #126), 5015 Poster Discussion Session; Displayed in Poster Session (Board #127),
Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Sat, 4:30 PM-6:00 PM Sat, 4:30 PM-6:00 PM
Prospective head-to-head comparative phase 3 study between 18F-fluciclovine Cardiovascular morbidity in a randomized trial comparing GnRH-agonist and
and 68Ga-PSMA-11 PET/CT in patients with early biochemical recurrence of antagonist among patients with advanced prostate cancer. First Author:
prostate cancer. First Author: Jeremie Calais, UCLA, Los Angeles, CA David Margel, Rabin Medical Center, Petah Tikva, Israel
Background: This is a prospective single-center, single-arm, head-to-head Background: Androgen-deprivation therapy (ADT) used in prostate-cancer
phase 3 study of paired 18F-fluciclovine (FACBC) and 68Ga-PSMA-11 may increase risk of cardiovascular disease (CVD). Limited preclinical and
(PSMA) PET/CT scans for localizing early biochemical recurrence (BCR) of retrospective clinical data suggest that use of gonadotrophin-releasing
prostate cancer (PCa) after radical prostatectomy (RP) (NCT02940262). hormone (GnRH)-antagonist may be associated with lower risk of CVD
Methods: Fifty consecutive patients with BCR and prostate specific antigen compared to GnRH-agonist. Methods: We conducted a randomized open-
(PSA) levels ranging from $0.2 to #2.0 ng/mL without any prior salvage label study comparing the one year incidence of major cardiovascular and
therapy were included. All patients underwent FACBC and PSMA PET/CT cerebrovascular event (MACCE) in prostate-cancer patients with pre-existing
scans within #15 days. PET/CT scans were each interpreted by 3 independent CVD commencing on GnRH-agonists or antagonists. Patients were followed
blinded expert readers not involved in study design and data acquisition. every 3 months for the development of MACCE defined as either death,
Region consensus interpretation (T,N,M1a,M1b,M1c) was generated based on myocardial infarction (MI), cerebrovascular event (CVA), or percutaneous-
majority rule in cases of reader disagreement (2 vs 1). PET/CT scans were coronary intervention (PCI). Serum levels of N-terminal pro-B-type natri-
considered as positive if any region was rated as positive. Detection rates per- uretic peptide (NTproBNP) were analyzed at baseline, 3, 6 and 12-months.
patient and per-region served as primary study endpoint. Results: Median time Results: Eighty patients were enrolled (41 randomized to GnRH-antagonist,
interval between the 2 scans was 6 days (range 1-15). Median PSA level at the 39 to GnRH-agonist). Patients in both arms had similar age, baseline
time of imaging was 0.50 ng/ml (mean 0.63; range 0.2-2.0 ng/ml). The cardiovascular and prostate-cancer characteristics. During follow-up 15
detection rates were significantly lower with FACBC than with PSMA PET/CT patients developed a new cardiovascular event. Of these, nine patients
per-patient (26% vs 56%; p = 0.003) and per-region for pelvic nodes (N) (8% developed MACCE (two deaths, one MI, two CVAs, and four PCI). Twenty
vs 30%; p = 0.003) or any extra-pelvic lesions (M) (0% vs 16%; p = 0.008). percent (n = 8) of patients randomized to GnRH-agonists had a MACCE
Reader agreement for PSMA PET/CT image interpretations was significantly compared to 3% (n = 1) randomized to antagonists (log-rank p = 0.013). The
higher than for FACBC PET/CT (0.67 vs 0.20; p = 0.015). Conclusions: In absolute risk reduction for MACCE at 12 months using GnRH-antagonist was
patients with BCR and low serum PSA levels after RP, PSMA PET/CT dem- 18% (95%CI 5-31). Baseline levels of NTproBNP predicted events (AUC =
onstrates higher detection rates and superior reader agreement when com- 0.73 95%CI 0.54-0.91 p = 0.03) and increased over time only among
pared with FACBC PET/CT. Therefore, PSMA PET/CT should be the imaging patients with CV events. Conclusions: This is the first prospective study to
modality of choice in patients with early BCR. Clinical trial information: test cardiovascular outcome among prostate-cancer patients receiving ADT.
NCT03515577. We demonstrated that in patients with pre-existing CVD, GnRH-antagonists
was associated with development of fewer cardiovascular events compared
to GnRH-agonists. Clinical trial information: NCT02475057.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
292s Genitourinary (Prostate) Cancer

5016 Poster Discussion Session; Displayed in Poster Session (Board #128), 5017 Poster Discussion Session; Displayed in Poster Session (Board #129),
Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Sat, 4:30 PM-6:00 PM Sat, 4:30 PM-6:00 PM
Outcomes of men with recurrent M0 prostate cancer who defer androgen Final results from the randomized CABADOC trial: Patient preference
deprivation therapy until metastasis. First Author: Catherine Handy Marshall, between cabazitaxel and docetaxel for first-line chemotherapy in metastatic
Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins University, castrate-resistant prostate cancer (mCRPC). First Author: Giulia Baciarello,
Baltimore, MD Department of Cancer Medicine, Gustave Roussy Cancer Campus, Paris-Sud
University, France, Villejuif, France
Background: Optimal timing and criteria for implementation of androgen
deprivation therapy (ADT) in men with relapsed M0 prostate cancer (PCa) re- Background: Docetaxel and cabazitaxel represent now the standard of care in
mains undefined. Early ADT induces the non-metastatic castration resistant PCa men with mCRPC with similar efficacy reported in metastatic first-line setting
(nmCRPC) clinical state. FDA approved apalutamide (SPARTAN) and enzalu- in the FIRSTANA phase 3 trial. We assessed patients’ preference between the
tamide (PROSPER) for nmCRPC based on prolongation of metastasis free two taxanes. Methods: Patients with taxane-naı̈ve mCRPC were randomized
survival (MFS) which is now considered a valid endpoint for drug approval. in a 1:1 ratio to receive either docetaxel 75mg/m2/q3w x 4 followed by
Because overall survival (OS) in PCa is usually long and long-term ADT is as- cabazitaxel 25mg/m2/q3w x 4 (DO-CA), or the reverse sequence (CA-DO).
sociated with irreversible adverse events and high costs, we sought to evaluate Randomization was stratified based on prior next generation AR axis inhibitors
OS and other outcomes of men with relapsed PCa and ADT deferred until use. The primary endpoint was patient preference between taxanes, as
metastasis. Methods: Retrospective review of 2,636 men who had radical assessed by questionnaires in patients who had received at least one cycle of
prostatectomy (RP) between 1981-2017 and developed biochemically re- each taxane and who had not experienced a progression while on the first
current PCa from a single-institution. Patients who received ADT prior to me- taxane. Results: From June 2014 to October 2016, 195 men were randomized
tastasis were excluded. Kaplan-Meier survival estimates of MFS and OS were in 17 centers. After adjusting for the treatment period effect, more patients
defined from RP to event or censor. Multivariable Cox proportional hazards preferred cabazitaxel (43%) vs docetaxel (27%) (p , 0.004); 30% had no
regression was used to identify prognostic factors. Results: 1,686 men treated preference between taxanes. Fatigue, patient-defined quality of life, hair loss,
with deferred ADT until metastasis or censored metastasis-free were eligible. and pain were the most common factors influencing patient preference. Fe-
Medians: follow up 10 years (IQR5-16), age 60 years, PSADT 33 months, brile neutropenia was experienced by 5 (7.1%) men treated with cabazitaxel
Gleason , 7 (24%), Gleason 7 (55%), Gleason . 7 (21%). 688 (41%) received during the first period who received G-CSF and by 2 (7.1%) of those who did
salvage radiotherapy. Median MFS and OS were 21 and 22 years, respectively not. No febrile neutropenia was reported with docetaxel in both arms and with
(Table). In multivariable models, age (HR 1.06, 95% CI 1.04-1.1), Gleason , 8 cabazitaxel during the 2nd period, irrespectively of the use of G-CSF. The
vs $8 (HR 0.4; 0.3-0.5) , RP stage (organ confined vs not 0.6; 0.5-0.8), PSADT incidence of diarrhea during the first 3-month period was slightly reduced with
(0.995, 0.993-0.997) and salvage RT (0.88; 0.81, 0.96) were associated with G-CSF use in men receiving cabazitaxel (32.1% vs 24.3%) but not in those
OS. Conclusions: Deferred ADT in relapsing M0 patients is associated with long receiving docetaxel (23.8% vs 25%). The median progression-free survival was
OS measured from time of local treatment, comparable to OS with salvage ADT 9.81 in the DO-CA arm and 9.33 months in the CA-DO arm. The median overall
in contemporary experience. Drug approval trials in nmCRPC should focus on survival was also similar in the two groups (22.64 in the DO-CA arm and
patients at high risk for metastasis and death prior to ADT, and determine 20.73 months in the CA-DO arm. Conclusions: Although cabazitaxel and
standardized criteria for initiation of ADT. Prolongation of MFS in nmCRPC docetaxel have similar efficacy when used as first-line in mCRPC men, more
requires further validation and may not necessarily reflect a net OS benefit. patients prefer cabazitaxel. Clinical trial information: NCT02044354.
Overall PSADT < = 6 months PSADT < = 10 months
# who develop mets 489 (29%) 172 (61%) 260 (55%)
Median MFS, yrs (95% CI) 21 (20-24) 9 (8-10) 12 (10-13)
Median OS, yrs (95% CI) 22 (21-23) 13 (11-15) 15 (13-18)

5018 Poster Discussion Session; Displayed in Poster Session (Board #130), 5019 Poster Discussion Session; Displayed in Poster Session (Board #131),
Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Sat, 4:30 PM-6:00 PM Sat, 4:30 PM-6:00 PM
Genomic predictors of benefit of docetaxel (D) and next-generation hormonal Targeted next-generation sequencing (tNGS) of metastatic castrate-sensitive
therapy (NHT) in metastatic castration resistant prostate cancer (mCRPC). prostate cancer (M1 CSPC): A pilot molecular analysis in the STAMPEDE
First Author: Anis Hamid, Dana-Farber Cancer Institute, Boston, MA multi-center clinical trial. First Author: Clare Gilson, MRC Clinical Trials Unit at
UCL, London, United Kingdom
Background: Predictive genomic biomarkers in mCRPC remain elusive. Prior
studies suggest that tumor suppressor (TS) loss is prognostic, and may result in Background: The STAMPEDE trial recruits men with high risk prostate cancer
less benefit from NHT, but no impact on D efficacy. We assessed genomic commencing first line systemic therapy. In a pilot study to ascertain the
predictors of differential benefit of androgen receptor-targeted therapy and feasibility of tNGS and the prevalence of common genomic aberrations, we
chemotherapy for mCRPC. Methods: Patients with mCRPC and targeted exome tested a commercial clinically-accredited assay on tumor blocks and present
sequencing of biopsies obtained after metastatic diagnosis were identified data obtained in the largest cohort of treatment-naı̈ve M1 CSPC to date.
(n=109). Patients with pure small cell histology (n=6) were excluded. Time from Methods: Archival FFPE blocks were retrieved from trial participants and a
NHT or D start to clinical/radiographic progression (time to treatment failure, single block submitted for sequencing by a Foundation Medicine. Inc. tNGS
TTTF) was estimated by Kaplan-Meier method, with censoring at next therapy or assay that includes 395 genes. Results: We successfully obtained tNGS data
last follow-up for non-progressors. Results: 80.1% of patients had bone and/or on 115 (62%) of 186 patients enrolled between Nov-2011 and April-2017 at
lymph node-only metastases at mCRPC diagnosis. In total, 87/103 (84.5%) 15 UK participating centers. The median age was 70 years (IQR 44-85); 97%
and 61/103 (59.2%) received NHT and D for mCRPC, respectively. Median had de novo M1 disease and 83% Gleason score $8. We observed PTEN
overall survival was 4.5 years from first mCRPC. The frequency and predictive deficiency (34%) due to copy-number loss (25%) or mutation (9%); TP53
association of selected recurrently-altered genes are detailed in the table. PTEN mutation or loss (33%) and aberrations in PI3K signaling (16%), genes in-
alterations (alts) were associated with worse TTTF on NHT, but not D, and a volved in DNA repair (14%), Wnt signaling (14%) and cell cycle control (6%).
similar trend was observed with BRCA2. Biallelic RB1 loss was strongly pre- In total, these aberrations were observed in 76% of patients, with 35%
dictive, conferring significantly shorter TTTF on both NHT and D. A score based harboring two or more. No androgen receptor (AR) mutations were detected.
on presence of tumor PTEN alt (1) and/or biallelic RB1 alt (1) was predictive of Conclusions: The prevalence of PTEN deficiency is comparable with that
TTTF on NHT (median TTTF of score 0 vs 1 vs 2: 14.7 vs 12 vs 3.8 months; log observed in mCRPC consistent with this being a feature of metastatic disease.
rank p=0.003). Conclusions: The presence of single or compound PTEN and In contrast, AR mutations are not observed in this treatment-naı̈ve group. The
RB1 alts predict poorer outcomes with NHT for mCRPC. Chemotherapy may prevalence of DNA repair deficiency is less than observed in mCRPC but more
be a preferred therapeutic strategy for this patient population. than reported in prostatectomy cohorts. Although it is possible to use FFPE
Abiraterone/Enzalutamide Docetaxel
biopsies for tNGS, the test failure-rate poises challenges to evaluating treat-
N=86 N=61 ments in low prevalence biomarker-defined groups. These data will inform the
Median TTTF=12.2 mo Median TTTF=5.1 mo
design and conduct of biomarker-directed trials.
Gene N=103 HR p-val HR p-val
PTEN mo/bi 59.2% 1.68 0.029 0.83 0.52
PTEN bi 32% 1.25 0.39 0.96 0.89
TP53 mo/bi 65% 1.17 0.52 0.84 0.55
TP53 bi 47.6% 1.1 0.67 0.96 0.87
RB1 mo/bi 65% 1.29 0.28 1.04 0.88
RB1 bi 12.6% 2.86 0.003 3.17 0.007
BRCA2 11.7% 1.83 0.088 0.31 0.091
ATM 5.8% 0.84 0.7 0.52 0.37
AR amplification 30.1% 1.02 0.92 0.81 0.46
AR mutation 10.7% 0.36 0.017 1 0.99

Key: mo/bi, monoallelic or biallelic loss; bi, biallelic loss only.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Genitourinary (Prostate) Cancer 293s

5020 Poster Discussion Session; Displayed in Poster Session (Board #132), 5021 Poster Session (Board #133), Sat, 1:15 PM-4:15 PM
Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, PSA decline and objective response rates in White (W), Black (B), and Asian
Sat, 4:30 PM-6:00 PM men with metastatic castration-resistant prostate cancer (mCRPC). First
HSD3B1 and overall survival (OS) in men with low-volume (LV) metastatic Author: Susan Halabi, Duke University Medical Center, Durham, NC
prostate cancer (PCa) treated with androgen deprivation therapy (ADT) or
Background: We have shown previously in multivariable analysis that AA men
chemohormonal therapy in the CHAARTED Randomized trial. First Author:
had 19% lower risk of death than C men with metastatic castration resistant
Jason W.D. Hearn, University of Michigan, Ann Arbor, MI
prostate cancer (mCRPC) treated with a docetaxel (D) and prednisone (P)
Background: The HSD3B1(1245A . C) variant allele, whose frequency varies based regimen. The primary goal of this analysis was to compare $50%PSA
by race, encodes a missense sequence that stabilizes the rate-limiting enzyme decline and objective response rate (ORR) in C men, AA men, and Asian men
responsible for extragonadal androgen synthesis, thus enhancing intratumoral with mCRPC treated with a DP based regimen. Methods: Individual patient
dihydrotestosterone (DHT) synthesis. Multiple retrospective studies have data from 8,151 mCRPC men randomized on eight phase III trials to a D
found that men inheriting the HSD3B1(1245C) variant allele exhibit early containing regimen were combined. Race used in the analysis was based on
resistance to ADT. We sought to validate these findings with prospective data self-report. The endpoints were $50% PSA decline from baseline defined by
from the Chemohormonal Therapy versus Androgen Ablation Randomized Trial the PSA working group 2 and ORR (defined as complete or partial response).
for Extensive Disease in Prostate Cancer (CHAARTED). Methods: Men with The logistic regression model was employed to assess the prognostic impor-
newly metastatic PCa were randomized to receive either ADT plus docetaxel tance of race in predicting $50% PSA decline and ORR adjusting for
at a dose of 75 mg/m2 every 3 weeks for 6 cycles (arm A) or ADT alone (arm B). treatment arm, performance status, and site of metastases. Results: Of 8,151,
We determined germline HSD3B1 genotype in the subset of men with LV 7,687 (94%) patients had evaluable PSA data. Of 7,687 men, 6,535 (85%)
disease ( , 4 bone metastases, no visceral metastases). We analyzed freedom were W, 445 (6%) were B, 395 (5%) were Asian and 312 (4%) had race
from castration-resistant prostate cancer (CRPC) and OS according to unspecified. Men with race unspecified were excluded from the analysis,
HSD3B1 genotype using Cox and Kaplan-Meier methods. Results: 197 pa- leaving 7,375 men. Percentage of patients who experienced PSA decline from
tients with LV disease had blood samples available and were genotyped, in- baseline were: 64%, 58% and 62% in W, B and Asian men, respectively. In
cluding 97 in arm A and 100 in arm B. Docetaxel did not improve OS of LV multivariable analysis adjusting for risk factors, the pooled odds ratios for PSA
men. Of the 197 men, 47% were homozygous wild-type (WT), 43% were decline for AA vs. W were 0.9 (95% CI = 0.7-1.1, p = 0.16) and 1.0 (95% CI =
heterozygous, and 10% were homozygous variant. When all 197 men were 0.8-1.2, p = 0.92) for Asian vs. W. In 2,760 patients with measurable disease,
analyzed as one goup, the median time to CRPC was 39.7 mos. in homozygous the percentage of patients who had ORR were: 39%, 31% and 34% in W, B
WT men vs. 25.0 mos. in men with one or more copies of the variant allele (HR and Asian men, respectively. In multivariable analysis, the pooled odds ratios
1.27, 95% CI 0.89 to 1.82; p = 0.187). Although OS data are still maturing, at for ORR were 1.0 (95% CI = 0.7-1.4) for B vs. W and 0.9 (95% CI = 0.6-1.4)
52 months OS was 83% (95% CI 75% to 91%) in homozygous WT men vs. for Asian vs. W. Conclusions: There were no differences in PSA decline and
64% (95% CI 55% to 74%) in men with one or more variant alleles. There ORR outcomes in W, B, or Asian men with mCRPC enrolled on these phase III
was a suggestion that docetaxel delayed development of CRPC among men clinical trials with DP. Clinical trial information: NCT00110214.
with at least 1 variant allele (20.3 vs. 40.7 mos.; HR 0.66, 95% CI 0.40 to
1.04; p = 0.08). Benefit for men with high-volume disease was not evident.
Conclusions: Inheritance of the HSD3B1(1245C) allele that augments DHT
synthesis may be associated with lower OS in men treated with ADT with or
without docetaxel for LV newly metastatic PCa. Additional study is warranted in
patients with LV disease. Clinical trial information: NCT00309985.

5022 Poster Session (Board #134), Sat, 1:15 PM-4:15 PM 5023 Poster Session (Board #135), Sat, 1:15 PM-4:15 PM
External validation of a prognostic model for overall survival (OS) in men with Efficacy of apalutamide (APA) plus ongoing androgen deprivation therapy
metastatic castration-resistant prostate cancer (mCRPC). First Author: (ADT) in patients (pts) with nonmetastatic castration-resistant prostate
Susan Halabi, Duke University Medical Center, Durham, NC cancer (nmCRPC) and baseline (BL) comorbidities (CM). First Author:
Eric Jay Small, Helen Diller Family Comprehensive Cancer Center, University
Background: We have previously developed and externally validated a prognostic
model of OS in men with mCRPC treated with docetaxel (D), which included eight
of California San Francisco, San Francisco, CA
predictors: opioid analgesic use, ECOG performance status, albumin, disease site, Background: The addition of APA to ongoing ADT in pts with nmCRPC sig-
LDH, hemoglobin, PSA, and alkaline phosphatase. We have used this model to nificantly prolonged metastasis-free survival (MFS), time to symptomatic pro-
develop prognostic risk groups. We sought to externally validate this model in a gression (SymProg), and second progression-free survival (PFS2) in SPARTAN.
broader group of men with mCRPC and in specific subgroups (White, Black, Asian We assessed the impact of APA on these end points in pts with or without BL
patients, different age groups) and to validate the two and three prognostic risk CM. Methods: Using Cox proportional hazards models, treatment effect of APA
groups in this large dataset. Methods: Data from 5,790 mCRPC men randomized on
was evaluated in SPARTAN pts with CM at BL, stratifying by the presence of BL
5 phase III trials were utilized to validate the prognostic model of OS: D +/- zibo-
tentan (ENTHUSE), D +/- lenalidomide (MAINSAIL), D +/- dasatinib (READY), D+/- diabetes/hyperglycemia (D/H), cardiovascular disease (CVD), hypertension
custirsen (SYNERGY), and tasquinimod/placebo)). We applied the estimated pa- (HTN), and renal insufficiency (RI). Results: Of 1207 SPARTAN pts, 1062
rameters from the prognostic model to each of the five data sets and computed a risk (88%) had $ 1 BL CM, including 703/806 (87%) APA pts and 359/401 (90%)
score. We assessed the predictive performance of the model by computing the time- PBO pts. A total of 226 (19%), 398 (33%), 798 (66%), and 774 (64%) pts had
dependent area under the receiver operating characteristic curve (tAUC) and vali- D/H, CVD, HTN, and RI, respectively; 323 (27%), 412 (34%), 259 (21%), and
dated the two-risk (low, high) and three-risk prognostic risk groups (low, in- 68 (6%) pts had 1, 2, 3, and 4 CM, respectively. Incidence of CM was balanced
termediate, high) that were defined by the model. Results: The tAUC for the different between arms. Pts with CM were older than pts with no CM (median age, 75 vs
groups is presented in the table. Race, age, and treatment subsets had similar 69 yrs, APA; 74 vs 69 yrs, PBO). MFS, SymProg, and PFS2 benefit with APA
results. For the two prognostic risk groups, the median OS in the low and high groups was significant in all CM subgroups, except PFS2 for pts with D/H (Table) and
were 27.6 months (95% CI = 26.6-28.7) and 13.8 months (95% CI = 13.3-14.4). regardless of the number of CM. The incidence of any treatment-emergent AE
For the three prognostic risk group, median OS in the low, intermediate and high was balanced between pts with and without CM. AEs with APA were not affected
groups were 29.7 months (95% CI = 28.3-31.4), 19.0 month (95% CI = 18.3-20.4) by any CM. Clinical trial information: NCT01946204. Conclusions: The benefit
and 12.1 months (95% CI = 11.5-12.9), respectively. Conclusions: This prognostic of APA + ongoing ADT in pts with nmCRPC was maintained in pts with D/H,
model for OS in men with mCRPC has been validated in a larger dataset, yields CVD, HTN, and RI. The safety profile of APA was not affected by any CM.
similar results across race, age and treatment groups. The model is robust and can be
used to identify prognostic risk groups of patients for stratification and enrichment APA efficacy in SPARTAN pts with BL CM. APA vs PBO, HR (95% CI).
trials. Clinical trial information: NCT00626548. MFS SymProg PFS2
HR, 0.30 HR, 0.45 HR, 0.49
All pts (0.24-0.36) (0.32-0.63) (0.36-0.66)
tAUC (95% CI) (N = 1207)
All (n = 5,790) 0.74 (0.73-0.76) CM Yes No Yes No Yes No
Race Any CM 0.32 0.17 0.50 0.25 0.54 0.25
White (n = 4,858) 0.75 (0.73-0.76)
(0.26-0.40) (0.09-0.30) (0.34-0.73) (0.11-0.56) (0.39-0.74) (0.10-0.61)
Black (n = 193) 0.75 (0.69-0.83)
Asian (n = 396) 0.71 (0.66-0.77)
D/H 0.26 0.30 0.34 0.47 0.53 0.5
Age (0.16-0.43) (0.24-0.38) (0.14-0.87) (0.33-0.69) (0.23-1.23) (0.36-0.69)
< 65 (n = 1,718) 0.78 (0.76-0.81) CVD 0.31 0.29 0.51 0.42 0.58 0.44
65-74 (n = 2,625) 0.74 (0.73-0.76) (0.22-0.43) (0.22-0.38) (0.27-0.96) (0.28-0.64) (0.37-0.90) (0.30-0.67)
> = 75 (n = 1,425) 0.71 (0.69-0.74) HTN 0.33 0.24 0.6 0.28 0.52 0.41
Treatment (0.26-0.43) (0.17-0.34) (0.38-0.95) (0.16-0.50) (0.36-0.76) (0.24-0.71)
DP (n = 4,545) 0.73 (0.72-0.75) RI 0.36 0.21 0.55 0.35 0.56 0.33
Non-DP (n = 1,245) 0.79 (0.76-0.81) (0.27-0.46) (0.15-0.30) (0.34-0.87) (0.20-0.60) (0.39-0.81) (0.18-0.58)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
294s Genitourinary (Prostate) Cancer

5024 Poster Session (Board #136), Sat, 1:15 PM-4:15 PM 5025 Poster Session (Board #137), Sat, 1:15 PM-4:15 PM
Age-related efficacy and safety of apalutamide (APA) plus ongoing androgen Predictors of falls and fractures in patients (pts) with nonmetastatic castration-
deprivation therapy (ADT) in subgroups of patients (pts) with nonmetastatic resistant prostate cancer (nmCRPC) treated with apalutamide (APA) plus
castration-resistant prostate cancer (nmCRPC): Post hoc analysis of SPARTAN. ongoing androgen deprivation therapy (ADT). First Author: YaoYao Guan
First Author: Julie Nicole Graff, Knight Cancer Institute, Oregon Health & Pollock, Helen Diller Family Comprehensive Cancer Center, University of
Science University, Portland, OR California San Francisco, San Francisco, CA
Background: SPARTAN, a randomized phase 3 placebo (PBO)-controlled study in pts Background: SPARTAN, a phase 3 study of APA vs placebo (PBO) added to
with high-risk nmCRPC and PSA doubling time # 10 mo, showed that, compared with ongoing ADT in pts with nmCRPC, demonstrated that APA significantly pro-
PBO, addition of APA to ongoing ADT treatment (tx) prolonged metastasis-free survival longs metastasis-free survival, time to symptomatic progression, and second
(MFS) by . 2 y, reduced the risk of symptomatic progression by 55%, and increased progression free survival (Smith et al. NEJM 2018), with no decline in health-
second progression-free survival (PFS2), which is the time from randomization to related quality of life (Saad et al. Lancet Oncol 2018). SPARTAN pts who
disease progression on first subsequent anticancer tx, or death. The impact of APA in received APA, vs PBO, with ongoing ADT had higher rates of falls (15.6% vs
terms of benefit and safety profile was evaluated in pts aged , 65, 65-74, and $ 75 y.
9.0%) and fractures (11.7% vs 6.5%). An analysis was performed to identify
Methods: Pts with nmCRPC were randomized 2:1 to APA (240 mg QD) or PBO; ADT
clinical characteristics associated with falls and fractures in APA-treated
was continuous. APA effect was analyzed by Cox models and Kaplan-Meier methods
across age subgroups. Results: Baseline characteristics among age groups were SPARTAN pts. Methods: Of 1207 pts enrolled, 806 were randomized to
similar, although ECOG PS 1 vs 0 increased with age. MFS benefit with APA was APA. Univariate Cox proportional hazards model (UVA) assessed the associ-
highly significant for all age subgroups (Table). In pts $ 75 y, MFS risk with APA vs ation of 47 baseline clinical characteristics (demographics, comorbidities, and
PBO was reduced by 59%; MFS risk was reduced by 86% and 76% for pts , 65 and medication use, including bone-sparing agents) with time to fall or time to
65-74 y, respectively. Risk of PFS2 with APA vs PBO was reduced across all age fracture. Characteristics with p values , 0.10 were included in a multivariate
subgroups. PFS2 in pts , 65, 65-74, and $ 75 y: HR, 0.09 (p , 0.0001); HR, 0.56 Cox proportional hazards model (MVA) to determine independent factors as-
(p = 0.0343); HR, 0.59 (p = 0.0092), respectively. Risk of symptomatic progression sociated with these outcomes (p , 0.05). Results: Factors associated with
was lessened with APA vs PBO for all age subgroups (Table). There was a similar time to both fall and fracture on UVA (p , 0.10) included older age, low serum
increase in incidence of tx-emergent adverse events (TEAE) with age in both tx arms albumin, and poor ECOG performance status (PS). Additional factors asso-
that remained higher with APA. Incidence of grade 3/4 TEAE ($ 75 vs , 65 y): APA, ciated with time to fall were cerebrovascular accidents/transient ischemic
50% vs 37%; PBO, 37% vs 28%. Conclusions: Pts in all age subgroups with high-risk attacks, neuropathy, depression, a-blocker use, and antidepressant use. On
nmCRPC had significant improvement in MFS with APA + ongoing ADT. The safety MVA, older age, poor ECOG PS, history of neuropathy, and a-blocker use were
profile of APA was similar across age subgroups. Clinical trial information: independently associated with falls; older age and low serum albumin were
NCT01946204. independently associated with fractures (Table). Conclusions: At initiation of
MFS HR Symptomatic Progression APA added to ongoing ADT, nmCRPC pts with higher risk of falls and fractures
(95% CI) p Value Median MFS (95% CI) HR (95% CI) p Value
can be identified and are candidates for intervention to reduce the risk for these
< 65 y
PBO, n = 43 1.00a 7.3 (3.8-22.0) 1.00a
events. Clinical trial information: NCT01946204.
APA, n = 106 0.14 (0.07-0.27) , 0.0001 NR 0.29 (0.11-0.77) 0.0133
65– < 75 y Clinical characteristics associated with time to fall or fracture HR
PBO, n = 169 1.00a 14.6 (11.0-18.3) 1.00a in nmCRPC pts treated with APA (95% CI) p Value
APA, n = 307 0.24 (0.18-0.34) , 0.0001 NR (29.4-NE) 0.42 (0.26-0.68) 0.0005
‡ 75 y Time to fall
PBO, n = 189 1.00a 18.5 (16.2-22.8) 1.00a Age ‡ 75 vs < 75 y 1.8 (1.3-2.7) 0.0017
APA, n = 393 0.41 (0.31-0.56) , 0.0001 40.5 (NE) 0.59 (0.33-1.05) 0.072 ECOG PS at baseline, 1 vs 0 1.7 (1.2-2.5) 0.0055
History of neuropathy, Y vs N 2.3 (1.4-3.7) 0.0011
a
Reference Prestudy use of a-blocker, Y vs N 2.6 (1.4-5.1) 0.0042
NR, not reached Time to fracture
NE, not estimable Age ‡ 75 vs < 75 y 1.6 (1.0-2.5) 0.0321
Albumin: < median vs ‡ median (median = 44 g/L) 1.6 (1.1-2.5) 0.0276

5026 Poster Session (Board #138), Sat, 1:15 PM-4:15 PM 5027 Poster Session (Board #139), Sat, 1:15 PM-4:15 PM
Concurrent or layered treatment with radium-223 (Ra-223) and enzalutamide Pembrolizumab (pembro) plus olaparib in docetaxel-pretreated patients (pts)
(Enza) or abiraterone plus prednisone/prednisolone (Abi/pred): A retrospective with metastatic castrate-resistant prostate cancer (mCRPC): Cohort A of the
study of real-world clinical outcomes in patients (pts) with metastatic phase 1b/2 KEYNOTE-365 study. First Author: Evan Y. Yu, University of
castration-resistant prostate cancer (mCRPC). First Author: Neal D. Shore, Washington, Seattle, WA
Carolina Urologic Research Center, Myrtle Beach, SC
Background: Individual activity with pembro or olaparib has been observed in
Background: In clinical practice, Ra-223 is often combined with Enza or Abi/ mCRPC pts who progressed on second-generation hormone therapy (HT)
pred. ERA 223 (NCT02043678) showed increased fracture risk with con- and chemotherapy. Data from cohort A (pembro+olaparib) of KEYNOTE-365
current Ra-223+Abi/pred. We assessed real-world symptomatic skeletal events (NCT02861573), a phase 1b/2 umbrella study to test combinations in mCRPC,
(SSEs) and overall survival (OS) of pts with mCRPC who received concurrent or are presented. Methods: Pts with mCRPC who progressed within 6 mo before
layered Ra-223+Enza or Abi/pred. Methods: Patients with mCRPC treated with screening, were docetaxel-pretreated (up to 1 other chemotherapy permitted) for
Ra-223 in US cancer clinics from 1/01/2013 to 6/30/2017 were identified mCRPC and had #2 second-generation HTs were eligible. Pts received pembro
from a Flatiron prostate cancer registry of electronic health records. Treatment 200 mg IV Q3W+olaparib 400 mg orally twice daily. Primary end points: safety
initiation defined subgroups: concurrent (both started within 30 days) or layered and PSA response rate (confirmed PSA decrease $50%). Key secondary end
(1 started $30 days after the other). Baseline (BL) was the first dose of Ra-223. points: ORR per RECIST v1.1 (investigator review), disease control rate (DCR:
Descriptive analysis was performed for BL characteristics, SSEs, and OS CR+PR+SD $6 mo), time to PSA progression, composite response rate, rPFS,
(Kaplan–Meier). Results: Of 625 pts treated with Ra-223, 48% received Ra- and OS. Results: Median (95% CI) follow-up was 11 (6-15) mo. 41 pts initiated
223+Enza or Abi/pred. Layered treatment was more common (73%) than treatment (median age, 69 y; visceral disease, 42%; RECIST-measurable, 68%;
concurrent (27%). BL characteristics and clinical outcomes were summarized homologous recombination deficient detected, 0%). Efficacy is outlined in the
[Table]. Conclusions: In a real-world setting, Ra-223+Enza or Abi/pred treat- table. Treatment-related AEs occurred in 39 (95%) pts; most frequent ($30%)
ment was mainly layered. SSE rates with layered vs concurrent Ra-223+Abi/ were anemia (37%), fatigue (34%), and nausea (34%). Grade 3-5 treatment-
pred varied between subgroups; results must be treated cautiously given small related AEs occurred in 21 (51%) pts. There were 2 deaths; 1 was treatment-
pt numbers and a non-randomized study. The ongoing PEACE III trial is in- related (cause unknown). Conclusions: Pembro+olaparib had activity in pts with
vestigating concurrent Ra-223+Enza; a Phase III study (ESCALATE) exploring mCRPC who were molecularly unselected and were previously treated with
layered Ra-223+Enza is planned. docetaxel and second-generation HT. The observed safety profile for the
Concurrent Layered
combination is consistent with individual profiles of pembro and olaparib.
Ra-223+ Ra-223+ Concurrent Layered Clinical trial information: NCT02861573.
Enza Enza Ra-223 +Abi/pred Ra-223 +Abi/pred All pts
PSA Response
BL (n = 44) (n = 123) (n = 39) (n = 97) (N = 625)
Total population 5/41 (12)
Time from CRPC to BL (mo), median 5 14 3 10 11 Measurable disease 4/28 (14)
Prior therapy, n (%) 15 (34) 60 (49) NA NA 344 (55) ORR RECIST v1.1, pts w/measurable disease, n/N (%) 2/28 (7); 2 PRs
Abi/pred NA NA 12 (31) 31 (32) 335 (54) Responses ongoing, n 2
Enza 10 (23) 23 (19) 12 (31) 24 (25) 164 (26) DCR per RECIST v1.1, n/N (%) 12/41 (29)
Docetaxel Total population 9/28 (32)
Any concomitant BHA, n (%) 29 (66) 71 (58) 24 (62) 59 (61) 343 (55) Measurable disease
Prior SSE, n (%) 19 (43) 71 (58) 20 (51) 47 (48) 314 (50) Composite response rate, n/N (%) 6/41 (15)
Prior pathologic fractures, n (%) 8 (18) 22 (18) 4 (10) 13 (13) 110 (18) Time to confirmed PSA progression in pts with measurable disease, median (95% CI), 15 (9-27)
wk
Outcomes
No confirmed PSA progression at 27 wk in pts with measurable disease, % 34
Any SSE, n (%) 9 (20) 35 (28) 14 (36) 22 (23) 168 (27)
rPFS per PCWG-modified RECIST, median (95% CI), mo 5 (4-8)
Pathologic fractures, n (%) 4 (9) 15 (12) 7 (18) 8 (8) 61 (10) 6-mo rPFS rate, % 48
OS from mCPRC (mo), median 28.1 26.9 28.3 34.5 28.1 OS, median (95% CI), mo 14 (8-not reached)
6-mo OS rate, % 73

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Genitourinary (Prostate) Cancer 295s

5028 Poster Session (Board #140), Sat, 1:15 PM-4:15 PM 5029 Poster Session (Board #141), Sat, 1:15 PM-4:15 PM
RESIST-PC phase 2 trial: 177Lu-PSMA-617 radionuclide therapy for metastatic Pembrolizumab (pembro) plus docetaxel and prednisone in abiraterone (abi)
castrate-resistant prostate cancer. First Author: Jeremie Calais, UCLA, Los or enzalutamide (enza)-pretreated patients (pts) with metastatic castrate
Angeles, CA resistant prostate cancer (mCRPC): Cohort B of the phase 1b/2 KEYNOTE-
365 study. First Author: Christophe Massard, Gustave Roussy Cancer Campus
Background: This is an investigator-initiated open-label prospective bi-centric
and University Paris-Sud, Villejuif, France
single-arm phase 2 clinical trial (NCT03042312) of 177Lu-PSMA-617 ra-
dionuclide therapy in patients with progressive metastatic castrate-resistant Background: Pembro had activity as monotherapy in pretreated advanced
prostate cancer (mCRPC). Methods: Patients with progressive mCRPC (bio- mCRPC. Data are presented here from cohort B (pembro + docetaxel/
chemical, radiographic or clinical) after $1 novel androgen axis drug (NAAD), prednisone) of KEYNOTE-365 (NCT02861573), a phase 1b/2 umbrella
either chemotherapy (CTX) naı̈ve or post-CTX, with sufficient bone marrow study to test combinations in mCRPC. Methods: Pts who progressed on or
reserve and normal kidney function were eligible. All patients underwent a became intolerant to $4 wk of abi or enza in the prechemotherapy mCRPC
screening PSMA PET/CT to confirm target expression. Patients received up to 4 state and progressed within 6 mo before screening were eligible. Pts received
cycles of 177Lu-PSMA-617 every 861 weeks and were randomized into 2 pembro 200 mg IV with docetaxel 75 mg/m2 IV Q3W plus prednisone 5 mg
treatment activities groups (6.0 or 7.4 GBq). Kidney dosimetry was performed orally twice daily. The primary end points were safety and PSA response rate
for the first cycle. Efficacy was defined as serum PSA decline of $50% from (confirmed PSA decrease $50%). Key secondary end points were investigator-
baseline at 12 weeks and served as primary endpoint. Results: 64 patients determined ORR (RECIST v1.1), disease control rate (DCR: CR+PR+SD $6
(median PSA 75 ng/ml; range 0.5-2425) were included in the study. 20% mo), time to PSA progression, rPFS, and OS. Results: 72 pts (median age, 68
were CTX naı̈ve while 80% were post-CTX (1.9 CTX regimens on average, range y; visceral disease, 36%; measurable disease, 50%) began pembro + doce-
1-4). 45% completed 4 cycles of 177Lu-PSMA-617. Androgen deprivation taxel. Median (95% CI) follow-up was 10 (8-12) mo. Efficacy is outlined in the
therapy was given concomitantly in 83%, NAAD in 23% and immunotherapy in table. Treatment-related AEs occurred in 69 (96%) pts; most frequent
6%. PSA decline of $50% was observed in 23% of patients at 12 weeks and in ($30%) were alopecia (43%), fatigue (40%), and diarrhea (39%). Grade 3-5
38% of patients at any time (best PSA response). The median time to best PSA treatment-related AEs occurred in 27 (38%) pts, including 2 deaths from
response was 22 weeks (range 6-49 weeks). 16% had a PSA decline of $90% treatment-related AEs (pneumonitis). Most commonly reported immune-
and 59% had any PSA decline ( . 0%). Mild and transient (CTCAE grade 1-2) mediated AEs were infusion-related reactions (11%) and colitis (10%).
side effects included xerostomia (72%), nausea/vomiting (69%) and bowel Conclusions: Pembro + docetaxel/prednisone has activity in pts with mCRPC
movement disorders (45%). CTCAE grade 3 toxicity included nausea/vomiting who previously progressed on second-generation hormone therapy. AEs were
(6%), anemia (8%), leukopenia (5%), kidney failure (3%), thrombocytopenia considered mild for the treatment combination. Clinical trial information:
(3%), and neutropenia (3%). The mean kidney dose was 2.7 Gy for the first NCT02861573.
cycle (range 0.9-5.9) i.e. 0.4 Gy/GBq (range 0.15-0.9). There was no dif- PSA response, n/N (%)
ference between the efficacy and toxicity for the 6.0 GBq (n = 23) and 7.4 GBq Total population 22/71 (31)
Measurable disease 8/35 (23)
(n = 41) treatment arms. Conclusions: 177Lu-PSMA-617 radionuclide therapy ORR, per RECIST v1.1 pts w/measurable disease, n/N (%) 5/36 (14); 5 PRs
is well tolerated in patients with progressive mCRPC. PSA declined by $50% DCR per RECIST v1.1, n/N (%)
Total population 41/72 (57)
in 38% of patients. The best PSA response rate occurred after 3 cycles. Measurable disease 18/36 (50)
Updated data will be provided at the time of the conference. Clinical trial Duration of response per RECIST v1.1, median (range), mo 5 (4-8+)
Response ‡6 mo, n (%) 1 (30)
information: NCT03042312. Time to confirmed PSA progression in pts with measurable disease, median 24 (15-30)
(95% CI), wk
No confirmed PSA progression at 27 wk in pts with measurable disease, % 45
rPFS per PCWG-modified RECIST, median (95% CI), mo 8 (8-10)
OS, median (95% CI), mo Not reached (13-not
reached)

5030 Poster Session (Board #142), Sat, 1:15 PM-4:15 PM 5031 Poster Session (Board #143), Sat, 1:15 PM-4:15 PM
Targeting IGF-1/2 with xentuzumab (Xe) plus enzalutamide (En) in metastatic Genomic characteristics of deleterious BRCA1 and BRCA2 alterations and
castration-resistant prostate cancer (mCRPC) after progression on docetaxel associations with baseline clinical factors in patients with metastatic
chemotherapy (DCt) and abiraterone (Abi): Randomized phase II trial results. castration-resistant prostate cancer (mCRPC) enrolled in TRITON2. First
First Author: Syed A. Hussain, University of Sheffield, Academic Unit of Author: Wassim Abida, Memorial Sloan Kettering Cancer Center, New York, NY
Oncology, Department of Oncology and Metabolism, Sheffield, United
Background: The phase 2 TRITON2 study (NCT02952534) is evaluating the
Kingdom
PARP inhibitor rucaparib in patients with mCRPC who have a deleterious
Background: Insulin-like growth factor receptor-1 (IGF-1R) signaling activates germline or somatic alteration in BRCA (BRCA1 or BRCA2) or 1 of 13 other
the PI3K/AKT pathway and may lead to androgen receptor (AR) transactivation DNA damage repair genes. Here we present analyses of tumor genomics and
and progression to endocrine treatment resistance. Xe, an IGF-ligand- baseline clinical characteristics in mCRPC patients with a deleterious alter-
neutralizing antibody, binds to IGF-1 and IGF-2 and inhibits IGF-1R signal- ation in BRCA. Methods: Plasma (baseline) and tissue (archival or baseline)
ing. This multi-center randomized phase II trial (NCT02204072) evaluated samples from patients with a deleterious alteration in BRCA were analyzed
anti-tumor activity of Xe plus En in mCRPC. Methods: Men with histologically/ using Foundation Medicine next-generation sequencing assays. The alter-
cytologically confirmed mCRPC and progression after DCt+Abi were randomized ations and zygosity of the alterations that were detected, as well as the somatic/
to receive Xe 1000mg IV QW + En 160mg/day oral, or En alone (28-day cycles germline status from Color Genomics testing, were summarized. Associations
until progression or intolerable adverse events [AEs]). Primary endpoint: between genomic alterations, DNA yield, allele frequency, and baseline
progression-free survival by investigator assessment (PFS-IA); secondary: PFS clinical characteristics were investigated. Results: Results are shown in the
by central review (PFS-CR), overall survival (OS), AEs. Results: Overall, 43 Table for a cohort of 40 BRCA2 and 5 BRCA1 patients enrolled in TRITON2
patients were randomized per arm; 70% Caucasian and 29% Asian (median age (Abida W et al. Presented at ESMO 2018. Abst 793PD). A biallelic alteration
70 y; range 46–88). At baseline (BL) the two arms were generally well balanced, was observed in 21 of the 22 BRCA2 patients (95%) for whom alteration
although 33% v 47% were ECOG PS O, and 72% v 56% had a Gleason total zygosity could be determined. Among the 5 BRCA1 patients, 1 alteration was
score $8. By data cut-off (23 October 2017), 39/43 (Xe+En) and 38/43 monoallelic and 4 were of unknown zygosity. Co-occurring alterations in
patients (En) had discontinued, most due to disease progression. The median cancer-related or DNA damage repair genes were observed in many patients
PFS-IA was 7.4 m for Xe+En (95% CI: 3.5–8.7) and 6.2 m for En (3.5–11.1) with BRCA alterations. At baseline, cell-free DNA (cfDNA) yield correlated
[HR = 0.99 (0.56–1.73); p = 0.96]. The results were similar after adjusting for positively with the sum of target lesions (STL; P= 0.04), but not with prostate-
BL ECOG PS and Gleason score. The median PFS-CR was 3.6 m for Xe+En specific antigen (PSA) levels (P= 0.86). No correlation was observed between
(3.5–8.1) and 6.2 m for En (3.6–8.3) (HR = 1.22 [0.70–2.13]; p = 0.48). OS allele frequency of the BRCA alteration baseline STL (P= 0.68) or PSA levels
data are immature. For the two arms, prostate-specific antigen (PSA) response (P= 0.97). Conclusions: Patients with a BRCA mutation enrolled in TRITON2
rates were 21% and 19%; maximum decline in PSA: -20 v -9 mg/L; PSA change demonstrate a profile of genomic alterations consistent with that of prior
at week 12: 19% v 18%; maximum decline in circulating tumor cells (CTC): studies of patients with mCRPC. Plasma cfDNA profiling showed a correlation
-52% v -35%; and CTC response: 16% v 11%. The most frequently reported between baseline cfDNA yield and measurable tumor burden, but not baseline
AEs were: fatigue 67% v 49%; decreased appetite 56% v 54%; weight re- PSA. Clinical trial information: NCT02952534.
duction 37% v 12%; anemia 33% v 44%; back pain 30% v 37%. Nine patients n (%) BRCA1 (n = 5) BRCA2 (n = 40)
discontinued Xe due to AEs. Conclusions: Addition of Xe to En did not prolong
Alteration type
PFS in mCRPC compared with En alone. There were no notable differences in Frameshift alteration 2 (40%) 20 (50%)
PSA-related endpoints and CTC between arms. Clinical trial information: Homozygous loss 0 (0%) 12 (30%)
NCT02204072. Germline/somatic status
Germline 2 (40%) 13 (33%)
Somatic 3 (60%) 27 (67%)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
296s Genitourinary (Prostate) Cancer

5032 Poster Session (Board #144), Sat, 1:15 PM-4:15 PM 5033 Poster Session (Board #145), Sat, 1:15 PM-4:15 PM
Profiling of genomic alterations in MAPK/ERK signaling in a large cohort of Personalized peptide vaccination for castration-resistant prostate cancer
metastatic prostate cancer (mPC) patients. First Author: Edwin Lin, University progressing after docetaxel chemotherapy: A randomized, double-blind,
of Utah/Huntsman Cancer Institute, Salt Lake City, UT placebo-controlled, phase III trial. First Author: Masanori Noguchi,
Kurume University School of Medicine, Kurume, Japan
Background: All mPC patients eventually progress on current treatments and
prognosis remains poor. In vitro models and small patient cohorts have shown Background: To develop a new treatment modality, we conducted a phase III
that mPC progression is associated with increased MAPK/ERK signaling. randomized trial of personalized peptide vaccination (PPV) for human leu-
However, in clinical mPC, the genomic alterations that cause aberrant MAPK/ kocyte antigen (HLA) -A24 positive patients with castration-resistant prostate
ERK signaling and their frequency are poorly defined. We hypothesize that cancer (CRPC) who failed docetaxel chemotherapy. Methods: Patients were
profiling of genomic alterations in MAPK/ERK in a large cohort of heavily randomly assigned in a 2:1 ratio to receive PPV or placebo. Four of 12
pretreated progressive mPC patients will provide a robust measure of impor- warehouse peptides selected based on preexisting peptide-specific immu-
tance, and reveal recurrent patterns of alteration. Given the large number of noglobulin G levels or the corresponding placebo were subcutaneously injected
drugs that target MAPK/ERK, these may be incorporated into novel combi- 6 doses weekly followed the maximum of 30 doses bi-weekly until disease
natorial treatments for mPC. Methods: 2,679 plasma samples from 2,309 progression. The primary end point was overall survival (OS), and secondary
men with mPC were assessed by a validated ctDNA NGS panel that sequences end points were progression-free survival (PFS) and immune responses.
73 clinically relevant cancer genes (Guardant360, Redwood City, CA) and Results: From August 2013 to April 2016, 310 patients were randomly
profiles indels, amplifications, and fusions with high sensitivity and specificity. assigned (207 to PPV and 103 to placebo), and 306 patients were analyzed by
Genes were assigned to gene sets corresponding to biological pathways and the full analysis set (204 to PPV and 102 to placebo). Baseline characteristics
molecular functional classes using the REACTOME database, followed by were balanced between groups. Estimated median OS was 16.1 months (95%
calculation of summary statistics. Interdependencies between genetic alter- CI, 13 to 18.2) with PPV and 16.9 months (95% CI, 13.1 to 20.4) with
ations at inter- and intra-gene set levels were discovered by a Bayesian network placebo (HR, 1.04; 95%CI, 0.79 to 1.37; P = 0.77). Median PFS was also not
approach. Results: 56% of mPC samples harbored alterations in MAPK/ERK significantly different among them. Median Grade $ 3 adverse events were
signaling genes. These included receptor tyrosine kinases (RTKs), MAP kinase observed in 41% in both groups. The analysis of treatment arm effects among
cascade, and cell cycle control genes. Gene amplifications were the most various subgroups revealed a lower HR for OS in favor of the PPV arm in
frequent alterations in RTK, RAF, and cell cycle control genes, a novel finding. patients with a , 64% neutrophil proportion (HR, 0.55; 95%CI, 0.33 to
Bayesian network analysis revealed positive interdependencies between all 0.93), with a significant interaction test (P = 0.003). Conclusions: PPV did not
MAPK/ERK genes. In RTK genes, co-amplifications were especially frequent prolong either OS or PFS in HLA-A24 positive patients with CRPC progressing
between MET & EGFR and PDGFRA & KIT. Conclusions: In a large cohort of after docetaxel chemotherapy. Clinical trial information: 0000113088.
mPC patients, we show that MAPK/ERK gene alterations are present in over
half of mPC patients. RTKs, BRAF, and CDK4/6 amplifications are among the
most frequent events, display recurrent patterns of co-alteration, and are
targetable by existing drugs. Future work to assess the biological and clinical
significance of these recurrent patterns of alteration will pave the way for novel
combinatorial treatments.

5034 Poster Session (Board #146), Sat, 1:15 PM-4:15 PM 5035 Poster Session (Board #147), Sat, 1:15 PM-4:15 PM
Phase 1 study of pasotuxizumab (BAY 2010112), a PSMA-targeting Bispecific Overall survival (OS) of African-American (AA) and Caucasian (CAU) men
T cell Engager (BiTE) immunotherapy for metastatic castration-resistant prostate who received sipuleucel-T for metastatic castration-resistant prostate cancer
cancer (mCRPC). First Author: Horst-Dieter Hummel, Comprehensive Cancer (mCRPC): Final PROCEED analysis. First Author: A. Oliver Sartor, Tulane
Center Mainfranken, University Hospital Würzburg, Würzburg, Germany Medical School, New Orleans, LA
Background: mCRPC has a poor prognosis and immunotherapies are largely Background: Prostate cancer risk and mortality are higher in AAs versus
ineffective. PSMA is a promising therapeutic target in mCRPC, and paso- CAUs. Post-hoc analyses of pooled Phase 3 data (n = 737) suggested
tuxizumab is a PSMA x CD3 BiTE that mediates tumor cell killing. substantial OS benefit for AA men receiving sipuleucel-T (n = 33) vs placebo
Methods: NCT01723475 was a first-in-human, multicenter, dose-escalation (n = 10) (McLeod 2012). Compared with pooled placebo patients (n = 249),
study in patients (pts) with mCRPC refractory to standard therapy. Pts received number needed to treat for OS benefit at 3 years was 3 for AAs and 8 for all
pasotuxizumab as a continuous intravenous infusion in cohorts of 3–4 pts. sipuleucel-T-treated patients (n = 488) (Moses 2019). Herein we analyzed
Dose-escalation followed a continuous reassessment methodology design. The PROCEED (NCT01306890), a large real-world registry, in which all patients
primary objective was to determine safety and maximum tolerated dose (MTD); received sipuleucel-T. Methods: In PROCEED, 1902 mCRPC patients
secondary objectives included pharmacokinetics, biomarkers, and tumor re- received $1 sipuleucel-T infusion. OS of all AA (n = 221) and CAU (n =
sponse. Results: 16 pts were enrolled into 5 dosing cohorts (5 mg/d, n = 3; 10 1649) men were compared. Baseline prostate-specific antigen (PSA), the
mg/d, n = 4; 20 mg/d, n = 3; 40 mg/d, n = 4; 80 mg/d, n = 2). All pts had $1 most important prognostic variable for OS after sipuleucel-T (Schellhammer
AE of any grade; most common were fever (94%), chills (69%), and fa- 2013), substantially differed by race. Thus, OS for a PSA-matched cohort
tigue (50%). 13 pts (81%) had $1 AE of grade $3; most common were (n = 219 AA; n = 438 CAU) was compared and univariable/multivariable
decreased lymphocytes and infections (both 44%). No grade 5 AE occurred. A analyses were performed. Post-sipuleucel-T use of OS-prolonging anticancer
serious AE related to study drug was reported for 1 pt (fatigue, 20 mg/d). No interventions was also assessed. Results: After a median follow-up of 46.6
anti-drug antibodies were observed. Recruitment was stopped before MTD was mo, median OS was 35.2 (all sipuleucel-T-treated AAs) and 29.9 mo (all
reached to facilitate initiation of a new study sponsored by Amgen. Antitumor sipuleucel-T-treated CAUs): HR 0.81, 95% CI 0.68–0.97; P = 0.03. In the
activity as indicated by PSA serum level decline was dose dependent, with a PSA-matched cohort, median OS was 35.3 and 25.8 mo, respectively (HR
mean best PSA change per dosing cohort versus baseline of +0.74% (5 mg/d), 0.70, 95% CI 0.57–0.86; P , 0.001). Sipuleucel-T-treated AAs with lower
–17.9% (10 mg/d), –37.4% (20 mg/d), –42.5% (40 mg/d) and –54.9% (80 baseline PSA had markedly longer median OS vs sipuleucel-T-treated CAUs.
mg/d). PSA decreases of $50% occurred in 3 pts (n = 1 each in 20 mg/d, 40 Among those with # median baseline PSA (29.48 ng/ml), median OS was
mg/d, and 80 mg/d cohorts). One long-term PSA responder was treated for 54.3 mo (AA) vs. 33.4 (CAUs); HR 0.52, 95% CI 0.37–0.72; p , 0.001.
14 months (40 mg/d) and one for 19.4 months (80 mg/d). The latter pt Along with other known prognostic factors, AA race was independently as-
showed a complete regression of soft-tissue metastases and marked regression sociated with prolonged OS on detailed multivariable analyses (HR 0.60,
of bone metastases as assessed by PSMA-PET/CT, . 90% reduction in PSA 95% CI 0.48–0.74; p , 0.001) and confirmed on sensitivity analyses. Post-
and alkaline phosphatase, and a significant and durable improvement in sipuleucel-T life-prolonging anti-cancer therapies were balanced between
disease related symptoms. Conclusions: Pasotuxizumab had an acceptable groups. Conclusions: Sipuleucel-T-treated AAs had significantly improved
safety profile and dose-dependent clinical activity in mCRPC pts. There were OS vs sipuleucel-T-treated CAUs. This analysis marks the largest known
two long term responders in the dose escalation. This is the first clinical study racial difference in OS in response to any therapy for mCRPC, a finding with
showing that a BiTE immunotherapy can be efficacious in solid tumors. implications for both prostate cancer pathophysiology and cancer immu-
Clinical trial information: NCT01723475. notherapy. Clinical trial information: NCT01306890.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Genitourinary (Prostate) Cancer 297s

5036 Poster Session (Board #148), Sat, 1:15 PM-4:15 PM 5037 Poster Session (Board #149), Sat, 1:15 PM-4:15 PM
Interlesional response assessment with 18F-sodium fluoride (18F-NaF) PET/CT Randomized phase II trial of a DNA vaccine encoding prostatic acid
in men with chemotherapy-naive bone metastatic castration-resistant prostate phosphatase (pTVG-HP) versus GM-CSF adjuvant in patients with PSA-
cancer (mCRPC) treated with enzalutamide (ENZA). First Author: Christos recurrent prostate cancer. First Author: Glenn Liu, University of Wisconsin
Kyriakopoulos, University of Wisconsin Carbone Cancer Center, University of Carbone Cancer Center, Madison, WI
Wisconsin, Madison, WI
Background: Phase I/II studies of pTVG-HP showed safety and increases in
Background: 18F-NaF PET/CT provides spatial and quantitative information prostatic acid phosphatase (PAP)-specific T cells. Based on these, we
on osteoblastic activity in men with bone mCRPC; thus, it can be used to assess conducted a multi-center, randomized phase II trial using pTVG-HP with GM-
interlesional response heterogeneity, a critical clinical issue. Here we assess CSF adjuvant vs GM-CSF alone in patients with PSA-recurrent, non-metastatic,
the proportion of men treated with ENZA with responding lesions by 18F-NaF non-castrate, prostate cancer (stage M0). Methods: 99 patients with M0
PET/CT at the time of prostate-specific antigen (PSA), standard radiographic, prostate cancer (negative CT and bone scan) with PSA doubling time (DT) of ,
or clinical progression. Methods: Men with progressive mCRPC with $ 2 le- 12m were randomly assigned to pTVG-HP + GM-CSF vs GM-CSF. Treatments
sions on bone scintigraphy were enrolled and treated with ENZA 160 mg daily were every 2 wk x 6, and then quarterly for up to 24m. Primary endpoint was 2-
at 3 US sites. 18F-NaF PET/CT scans were obtained at baseline (PET1), week year MFS. Secondary endpoints included median MFS, changes in PSA DT,
13 (PET2), and at the time of PSA progression (increase of $ 25% and $ 2.0 and immune response to PAP antigen. Additional analysis included quanti-
ng/mL above nadir), standard radiographic or clinical progression, or at 2 years tative changes in bone lesion activity by NaF PET/CT. Results: Few grade 3/4
without progression (PET3) using Quantitative Total Bone Imaging (QTBI). The events were observed and were not statistically different between arms. 2-year
primary endpoint was the proportion of men with $ 1 responding bone lesion MFS between control and pTVG-HP was 42.3% and 41.8% (p = 0.97). Median
(defined as a lesion with a total NaF standardized uptake value [SUV] less than MFS was not different (18.3m control versus 18.9m pTVG-HP, HR = 1.6, p =
baseline) on PET3. Evaluable men had scans at PET1 and PET3. Results: A 0.13). Changes in PSA DT, and immunity to the PAP target, were not different
total of 23 men (median age, 72 years [range, 51-93]; median PSA, 20.5 ng/ between study arms. A pre-planned subset analysis showed median MFS was
mL [range, 3.9-133.6]) were enrolled. The study met its primary objective; 22 significantly longer in patients with rapid PSA DT ( , 3m) treated with pTVG-
of 22 (100%) evaluable men had $ 1 responding bone lesion on QTBI at HP (6.1m versus 12.0m, n = 21, HR = 4.4, p = 0.03). In patients with baseline
PET3. Total disease burden changed from a mean baseline SUV of 5700 metastases detected on NaF PET/CT, total activity (SUVtotal) increased 17%
(range, 507-22,850) to 5590 (range, 213-17,090) at PET2 and 6020 (range, with GM-CSF alone, but decreased 29% with pTVG-HP from baseline to month
118-16,650) at PET3. The proportion of progressive lesions increased 6 (n = 27, p = 0.07). Conclusions: This trial did not demonstrate an overall
from a mean 7.8% (range, 0-29) at PET2 to 9.4% (range, 0-32) at PET3. increase in 2-year MFS following treatment with pTVG-HP, but suggested a
Conclusions: Although PSA response with ENZA is high, many men experience a modest effect in patients with rapidly progressive disease. NaF PET/CT im-
mixed response to treatment. While overall functional disease burden improves aging suggests that pTVG-HP has a detectable effect on decreasing metastatic
during treatment, an eventual increase in global burden is seen at the time of activity in bone. A separate trial suggests that pTVG-HP is an immune acti-
progression as measured by 18F-NaF PET/CT. At the primary endpoint analysis vating agent that improves activity of PD-1 blockade. A trial using pTVG-HP in
the proportion of progressing lesions is low, supporting both the hypothesis that a combination with nivolumab in patients with M0 prostate cancer is currently
substantial number of lesions continue to benefit from treatment and the concept underway. Clinical trial information: NCT01341652.
of treating beyond progression and selectively targeting nonresponding lesions
while keeping patients on ENZA. Clinical trial information: NCT02384382.

5038 Poster Session (Board #150), Sat, 1:15 PM-4:15 PM 5039 Poster Session (Board #151), Sat, 1:15 PM-4:15 PM
Platinum-based chemotherapy in metastatic prostate cancer with alterations Circulating tumor DNA fraction (ctDNA) as a surrogate predictive biomarker
in DNA damage repair genes. First Author: Jose Mauricio Mota, Memorial in metastatic castration-resistant prostate cancer (mCRPC). First Author:
Sloan-Kettering Cancer Center, New York, NY Vincenza Conteduca, Istituto Scientifico Romagnolo per lo Studio e la Cura
dei Tumori IRCCS, Meldola, Italy
Background: Platinum-based chemotherapy has shown palliative and radio-
graphic benefit in small unselected studies of metastatic castration-resistant Background: Plasma ctDNA is a promising minimally invasive biomarker in
prostate cancer (mCRPC). Alterations in DNA damage repair genes (DDRmut), mCRPC. Pre-treatment high levels of ctDNA reflect poor prognosis (Romanel
which occur in ~25% of patients with mCRPC, may sensitize to platinum- et al, Sci Transl Med 2015; Annala et al, Cancer Discov 2018). However, the
based chemotherapy, and may aid in the selection of patients for this therapy. role of plasma ctDNA in prostate tumour monitoring is largely unexplored. We
We sought to evaluate the efficacy of platinum-based chemotherapy in aimed to determine if monitoring tumour response by quantifying ctDNA levels
DDRmut mCRPC. Methods: We performed a retrospective review of patients in plasma could enable early assessment of therapy efficacy for mCRPC.
with prostate cancer who underwent tumor genomic profiling and received Methods: Between January 2011 and June 2016, 132 sequential plasma
platinum-based chemotherapy. Deleterious alterations in a panel including samples from 54 mCRPC patients (pts) (30 pre- and 24 post-chemotherapy)
BRCA2, BRCA1, ATM, FANCA, CDK12 or PALB2 were classified as DDRmut. treated with abiraterone (abi) were collected. Targeted next-generation se-
Absence of deleterious alterations in those genes was classified as DDRwt. quencing was performed on the PGM Ion Torrent using a 316 or 318 Chip to
MSI-H cases were excluded from analysis. Electronic charts, PSA values, and account for 1000X expected coverage per target. We estimated the global
scans were reviewed to assess for outcomes. Results: From October 2013 to tumour content for each sequential plasma sample from study patients by using
July 2018, 109 patients with mCRPC received platinum-based chemotherapy. the approach developed in (Carreira et al, Sci Trasl Med 2014; Romanel et al,
64/109 had prior taxane progression and were PARP inhibitor (PARPi) naı̈ve at Sci Transl Med 2015), which extends the CLONET framework (Prandi et al,
the time of platinum-based chemotherapy. DDRmut was found in 16/64 Genome Biol 2014). Prostate Cancer Working Group -3 (PCWG3) criteria were
(25%) of patients (BRCA2, n = 6; ATM, n = 2; CDK12, n = 4; FANCA, n = used to assess clinical, biochemical (PSA) and radiographic (RAD) progression
4; PALB2, n = 1). Visceral metastasis occurred in 4/16 (25%) of DDRmut disease (PD). We considered ctDNA PD any increase of ctDNA from baseline
patients and in 22/48 (46%) of DDRwt patients. PSA50 responses were more value. Results: In our cohort of 54 pts (median age: 75 years, range 70-78), we
common among DDRmut (8/15 evaluable = 53%, 95% CI, 30-75%) than observed 17 (31.5%) PD, 14 (25.9%) stable disease, and 23 (42.6%) partial/
among DDRwt patients (5/42 evaluable = 12%, 95% CI, 5-25%). Time on complete response after the first 3 months (mo) abi therapy. The odds ratio (OR)
platinum-based chemotherapy tended to be longer in the DDRmut group for PD having any increase in ctDNA and a PSA decline , 50% at ~3-mo
(median 3.1 vs 1.8 months; HR 0.73, 95% CI 0.42-1.26). Of 8 DDRmut therapy was 10.83, 95% CI 2.55-45.95, P = 0.001, and 3.27, 95% CI 0.89-
patients (BRCA2, n = 6; BRCA1, n = 1; ATM, n = 2) who received platinum- 12.3, P = 0.074, respectively. In addition, we assessed all 3 types of median PD
based chemotherapy after progression on a PARPi, 3/7 evaluable patients time from starting abi treatment, suggesting the ability of ctDNA variation to
(43%) had RECIST response or stable disease, and 2/7 evaluable patients predict overall PD [RAD PD = 6.8 mo, PSA PD = 4.4 mo, and ctDNA PD = 3.0
(29%) had a PSA50 response. Of 4 patients with ATM deleterious mutations, mo, P = 0.008). An increase of ctDNA levels during the first 3-mo abi treatment
none had a radiographic or PSA50 response to platinum-based chemotherapy. was significantly associated with a long-term androgen deprivation therapy
Conclusions: Platinum-based chemotherapy showed activity in DDRmut (ADT) before plasma sample collection (previous ADT . 24 mo vs 12 .
mCRPC patients before and after PARPi treatment. previous ADT #23 mo vs , 12 mo: P = 0.036). Conclusions: In mCRPC, an
early change in ctDNA fraction may be considered as a predictive biomarker
playing a key role in individualized disease monitoring. Prospective evaluation of
treatment decisions based on ctDNA is now required.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
298s Genitourinary (Prostate) Cancer

5040 Poster Session (Board #152), Sat, 1:15 PM-4:15 PM 5041 Poster Session (Board #153), Sat, 1:15 PM-4:15 PM
A phase II multicenter biomarker trial to study the predictive value of TMPRSS2- Integrating evolutionary dynamics into treatment of metastatic castrate-resistant
ERG before enzalutamide treatment in chemo-naı̈ve metastatic castration- prostate cancer (mCRPC): Updated analysis of the adaptive abiraterone (abi)
resistant prostate cancer. First Author: Enrique Grande, MD Anderson Cancer study (NCT02415621). First Author: Jingsong Zhang, H. Lee Moffitt Cancer
Center Madrid, Madrid, Spain Center & Research Institute, Tampa, FL
Background: TMPRSS2-ERG fusion gene is a common driver of prostate Background: To achieve better prostate cancer control and to delay the
cancer. The PREMIERE study is a phase II, single arm open-label, multicentre, emergency of treatment resistance, we developed an evolutionary game theory
clinical trial designed to analyse the predictive/prognostic value of TMPRSS2- model using Lotka-Volterra equations with three competing prostate cancer
ERG in first-line chemo-naı̈ve mCRPC patients treated with enzalutamide. "species": T+, Tp, and T-. T+ prostate cancer cells depend on exogenous an-
Methods: We centrally evaluated TMPRSS2-ERG in diagnostic samples using drogen; Tp cells express CYP17A1, produce and depend on androgen; and T-
PCR, FISH and IHC for ERG. Among exploratory biomarkers we included cells are androgen-independent and abi-resistant. We applied this model to
plasma DNA, AR copy number by ddPCR and CTC by AdnaTest. PCWG2 guide the on and off treatment cycles with abi for mCRPC. At the first interim
criteria were used for outcome evaluation. We correlated TMPRSS2-ERGand analysis with 11 patients, this approach was shown to prolong the time to cancer
other exploratory biomarkers with mCRPC outcomes. Results: Ninety eight progression with less than 50% drug usage compared to the conventional
patients with median age 77 y (range 59-95), ECOG 0/1 (54/46%) with mts continuous Abi (Nat Commun. 2017). Here we present the updated data of this
located in bone (82%), LN (48%) and visceral (17%). With a median FU of phase 2 study. Methods: Men with asymptomatic or minimal symptomatic
37.3 months, PSA response was PSA50: 82% and PSA90: 53%; median PSA- mCRPC were enrolled after they achieved . 50% PSA reduction with abi as a
PFS was 13.7m (95%CI 10.2-19.0), Rad-PFS 26.7m (95%CI: 22.0-NA) and frontline therapy for mCRPC. The primary objective is feasibility and is measured
OS 37.5m (95%IC: 33.7-NA). TMPRSS2-ERG was detected in 32 pts (33%), by the percentage of abi responsive men who remain to be responsive to abi
AR gain in 11 pts and CTCs in 35 pts. No differences were observed based on (defined as . 50% decline of the pre Abi PSA) after completing 2 adaptive
TMPRSS2-ERG status for PSA response (PSA50: 81% vs 83%; p=0.8), PSA- treatment cycles. The secondary objective is to assess the clinical benefits by
PFS (median 12.8 vs 14.7m; HR 0.98; 95%CI 0.58-1.67; p=0.95), Rad-PFS comparing the radiographic progression free survival (rPFS) in men undergoing
(median 28.4 vs 26.4m; HR 1.02; 95% 0.53-1.96, p=0.95) or OS (median adaptive Abi therapy to the historical AA 302 trial. Results: At the data cut off in
36.9 vs 38.1m; HR 1.23; 95%CI 0.69-2.21, p=0.48). Plasma AR gain was Jan 2019, the study has completed enrollment for the non-African American
associated with worse PSA-PFS (median 4.2 vs 14.7 m; p,0.0001), Rad-PFS cohort. 15 enrolled men had . 11 months of follow up. All 15 men were off Abi
(median 3.6 vs 28.4m; p,0.0001) and OS (median 12.7 vs 38.1m; for at least 3 months before abi was restarted for PSA progression at cycle 1.
p,0.0001). Plasma DNA and CTCs were also associated with worse outcome. Seven out of the 15 men had completed at least 2 adaptive therapy cycles. Four
Multivariate analyses of exploratory biomarkers are included in the table. of the rest 8 men remained on study and have not reached cycle 2. Six men were
Conclusions: The fusion gene TMPRSS2-ERG is not predictive nor prognostic off study due to radiographic progression at month 11, 20.4, 30, 30.5, 38 and
on enzalutamide treatment in first-line chemo-naı̈ve mCRPC patients. Plasma 53 from their first dose of Abi. Compare to the 16.2 months median rPFS in the
AR gain and CTCs are strong independent biomarkers associated with adverse AA 302 trial, the median rPFS of the 15 men would be no less than 30 months
outcome. Multivariate analysis of exploratory biomarkers. Clinical trial in- (p = 0.0068, Fisher’s exact test). Their average usage of Abi was 49% of the
formation: NCT02288936. continuous Abi. Conclusions: Adaptive Abi therapy is feasible in men who
PSA-PFS Rad-PFS OS
responded to Abi as a frontline therapy for mCRPC. The updated data are
consistent with our initial finding that our adaptive therapy approach can prolong
HR 95%CI p-value HR 95%CI p-value HR 95%CI p-value
the time to cancer progression with less than 50% drug usage compared to the
AR gain 3.36 1.53- 7.41 0.0026 12.97 5.29-31.79 ,0.0001 8.07 3.60-18.07 ,0.0001 conventional continuous Abi. Clinical trial information: NCT02415621.
CTC 2.89 1.70- 4.91 0.0001 4.73 2.39-9.39 ,0.0001 3.92 2.16- 7.10 ,0.0001
cDNA High 1.19 0.73- 1.95 0.4815 1.69 0.90-3.14 0.1004 1.84 0.99- 3.36 0.0537

5042 Poster Session (Board #154), Sat, 1:15 PM-4:15 PM 5043 Poster Session (Board #155), Sat, 1:15 PM-4:15 PM
Impact of enzalutamide and sequential flutamide and enzalutamide therapy Genomic landscape of metastatic hormone sensitive prostate cancer (mHSPC)
for castration-resistant prostate cancer after bicalutamide-combined androgen vs. metastatic castration-refractory prostate cancer (mCRPC) by circulating
blockade therapy onoverall survival: A follow-up study of randomized phase 2 tumor DNA (ctDNA). First Author: Andrew W Hahn, University of Utah
trial (OCCU-CRPC study). First Author: Taro Iguchi, Department of Urology, Hunstman Cancer Institute, Salt Lake City, UT
Osaka City University Graduate School of Medicine, Osaka, Japan
Background: mCRPC carries a poor prognosis, and targeted therapies have
Background: In Asia, bicalutamide-combined androgen blockade (CAB) is widely had minimal success in mCRPC. Novel genomic targets could improve drug
used to treat metastatic or locally advanced prostate cancer. Enzalutamide (ENZA) development. To date, large ctDNA studies in metastatic prostate cancer
shows benefits in men with metastatic and nonmetastatic castration-resistant have been descriptive with limited or no clinical annotation. Herein, we
prostate cancer (CRPC), although flutamide (FLU) is still used as an alternative hypothesize that profiles of genomic alterations (GAs) in ctDNA not only
anti-androgen therapy after bicalutamide-CAB. Methods: The multicenter open- differ significantly between, but can also be used to predict mCRPC vs.
label phase 2 trial OCUU-CRPC (NCT02346578) randomized patients (1:1) with mHSPC. These findings could help identify new drug targets for mCRPC
CRPC after bicalutamide-CAB to ENZA (160 mg/day) or FLU (375 mg/day). Pa- treatment. Methods: Men with mHSPC or mCRPC who underwent NGS of
tients were stratified according to distant metastases. The primary endpoint was the ctDNA using G360 (Guardant Health Inc.) at the Huntsman Cancer Institute
prostate-specific antigen (PSA) response rate ($50% decrease) at 3 months. The were included. Men were classified as mCRPC or mHSPC (patients with
endpoints of this follow-up observational study include PSA progression-free current or no prior ADT). G360 detects somatic mutations in selected exons
survival of ENZA therapy (PSA-PFS-ENZA), time to treatment failure of ENZA of 73 genes, amplifications in 18 genes, and selected fusions in 6 genes.
therapy (TTF-ENZA), and overall survival (OS). Results: In total, 103 patients were Two-sided students t-test was used to compare the %cfDNA and total GAs.
randomized to ENZA (n = 52) and FLU (n = 51). Overall, 67% had distant me- The Chi squared test was used to compare the frequency of each GA.
tastases and 10% had prior radical therapy. Twenty-five (48%) and 38 (75%) Machine learning (ML) algorithms were trained on GAs and benchmarked by
patients, respectively, discontinued their assigned treatment because of pro- cross-validated performance. GAs contributing to mCRPC vs. mHSPC
gressive disease or adverse events (AEs) and were treated by standard of care. Of the classification were measured by ML feature importance (e.g. odds ratios,
38 patients that discontinued FLU, 34 (89%) received ENZA as subsequent regression coefficients). Results: Of the 259 men included, 119 men had
therapy. The median follow-up time was 14 months (ENZA, interquartile range mHSPC and 140 had mCRPC. Men with mCRPC had more GAs (4.5 vs.
(IQR): 9.5-24.8) and 17.2 months (FLU, IQR: 9.8-24.9). PSA-PFS-ENZA was
1.86, p,0.0001) and higher %cfDNA (9.56% vs. 5.02%, p=0.02). In
longer in patients not treated with FLU (hazard ratio (HR), 0.29; 95% confidence
mHSPC, there was no significant difference in the number of GAs or %cfDNA
interval (CI), 0.14 to 0.63; p , 0.001), but there was no significant difference in
between men on ADT and those who hadn’t yet started ADT. ML algorithms
TTF-ENZA (HR, 1.27; 95% CI, 0.73 to 2.19; p = 0.397)and OS (HR, 0.77; 95%
used GAs to predict mCRPC with 78.1% sensitivity, 64.0% specificity,
CI, 0.31 to 1.89; p = 0.567) between the two groups. The AEs of ENZA were
consistent with those observed in prior phase 3 trials. Conclusions: ENZA after
76.7% PPV, 65.1% NPV, and 70.3% overall accuracy. mCRPC was
bicalutamide-CAB resulted in greater PSA-PFS than sequential FLU and ENZA enriched with GAs in AR, ARID1A, BRAF, BRCA2, CCNE1, CTNNB1, EGFR,
therapy. No differences were observed in TTF-ENZA and OS between ENZA after FGFR1, KIT, MET, MYC, PDGFRB, PIK3CA, and TP53. Of note, many of
bicalutamide-CAB and sequential FLU and ENZA therapy. Clinical trial in- these genes are involved in MAP/ERK signaling. Conclusions: Men with
formation: NCT02346578. mCRPC have more GAs, higher %cfDNA, and enrichment of GAs in the MAP/
ERK pathway compared to men with mHSPC. The distinct GAs seen in
ENZA FLU mCRPC represent novel therapeutic targets, especially in the MAP/ERK
Endpoints Median (95% CI), mo Median (95% CI), mo HR (95% CI)
pathway. We also show that machine learning can differentiate mHSPC and
PSA-PFS-ENZA NR (11.9, NR) 7.9 (4.4, 9.6) 0.29* (0.14, 0.63)
TTF-ENZA 15.0 (9.3, NR) 21.7 (14.6, 27.1) 1.27 (0.73, 2.19) mCRPC based on GAs detected in ctDNA.
OS NR (NR, NR) 28.4 (26.4, NR) 0.77 (0.31, 1.89)

*p , 0.001; ENZA; NR, not reached.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Genitourinary (Prostate) Cancer 299s

5044 Poster Session (Board #156), Sat, 1:15 PM-4:15 PM 5045 Poster Session (Board #157), Sat, 1:15 PM-4:15 PM
Health-related quality of life (HRQoL) and pain progression with enzalutamide Interim results from a phase 2 study of olaparib (without ADT) in men with
(ENZ) in metastatic hormone-sensitive prostate cancer (mHSPC) from the biochemically-recurrent prostate cancer after prostatectomy, with integrated
ARCHES study. First Author: Arnulf Stenzl, Department of Urology, University biomarker analysis. First Author: Emmanuel S. Antonarakis, Johns Hopkins
Hospital, Eberhard Karls University of Tübingen, Tübingen, Germany Sidney Kimmel Comprehensive Cancer Center, Baltimore, MD
Background: The Phase 3 ARCHES trial (NCT02677896) evaluated the ef- Background: Pts with biochemically-recurrent (BCR) prostate cancer after local
ficacy and safety of ENZ + androgen deprivation therapy (ADT) vs placebo therapy with a PSA doubling time (PSADT) of #6 mo are likely to develop
(PBO) + ADT in 1150 men with mHSPC. Here we report patient-reported metastases and death from their disease. We hypothesized that the PARP
outcome (PRO) data using Functional Assessment of Cancer Therapy-Prostate inhibitor olaparib would be effective as a non-hormonal therapy for biomarker-
(FACT-P) and Brief Pain Inventory Short Form (BPI-SF). Methods: FACT-P and unselected pts with BCR after prostatectomy, the first study of PARP inhibition
BPI-SF were assessed at baseline (BL), week (wk) 13, and then every 12 wks in the hormone-sensitive setting. We report pre-specified interim results from
until disease progression. Longitudinal changes were assessed using mean the first stage of the trial. Methods: This investigator-initiated multicenter study
scores and mixed-model repeated measures; lower BPI-SF scores represent (NCT03047135) enrolled pts with non-metastatic BCR after prostatectomy,
less pain/interference; higher FACT-P scores represent better HRQoL. Time with a PSADT of #6 mo. PSA had to be $1.0 ng/mL, with T $150 ng/dL. Pts
from BL to first deterioration in PRO score was assessed by Kaplan-Meier received olaparib 300 mg twice daily (without ADT), until a doubling of their
estimates and Cox proportional hazards models. Clinically meaningful dif- PSA or metastatic progression. The primary endpoint was confirmed $50%
ference was defined by change from baseline $10 for FACT-P total and $2 for PSA decline (PSA50 response). Secondary endpoints included safety, minor
worst pain/severity. Results: PRO instrument completion rates were high PSA response (1-50% reduction), and PSA progression-free survival. Integrated
(88296%) up to wk 73. At BL, men in both arms were generally asymptomatic biomarker analyses included somatic DNA sequencing, RNA expression
and reported good HRQoL (FACT-P total: ENZ + ADT, 113.9; PBO + ADT, analysis and IHC for DNA damage markers, using prostatectomy specimens.
112.7) and low pain (worst pain [item 3]: ENZ + ADT, 1.80; PBO + ADT, 1.77). The study was designed to enroll a biomarker-unselected population using a
HRQoL and pain scores remained stable over time and there were no clinically staged-adaptive design, with up to 50 pts. In the first stage, $3 PSA50 re-
meaningful differences between groups in change from BL to wk 73. The sponses out of 20 pts were required to proceed to the second stage; otherwise
proportion of men with no change or improvement in PRO scores (67–88%) enrichment with biomarker-selected pts would occur. Results: Between 5/2017
was similar in both groups at all time points up to wk 73. There was no and 11/2018, 20 men (median age, 65) enrolled in the first stage, with a
significant difference between arms for time to deterioration in FACT-P total median follow-up of 6 (range, 3-16) mo. Median PSADT was 3.1 mo, and 55%
(HR 0.90 [95% CI] (0.74, 1.09); p = 0.2998). ENZ + ADT significantly had Gleason $8 cancers. Seven men (35%) had BRCA2/ATM mutations. Three
delayed time to pain progression for worst pain (HR 0.82 [0.69, 0.98]; p = men (15%) had PSA50 responses (all had BRCA2 muts – 2 had complete PSA
0.0322) and pain severity (HR 0.79 [0.65, 0.97]; p = 0.0209) vs PBO + ADT. responses), and 4 other men (20%) had minor PSA responses. Median PSA
Conclusions: Men with mHSPC were generally asymptomatic and had high progression-free survival was greater in men with vs. without BRCA2/ATM
levels of HRQoL and low levels of pain at BL, likely due to most men initiating muts (9 vs. 4 mo; P= 0.02). Common toxicities of olaparib included fa-
ADT several months prior to study entry. No clinically meaningful differences in tigue, nausea, anemia and leukopenia; 2 men required a dose reduction.
HRQoL were observed between ENZ and PBO. The prolongation in radio- Conclusions: Olaparib (without ADT) was tolerable and showed activity in
graphic progression-free survival observed with ENZ + ADT was accompanied hormone-sensitive BCR prostate cancer, especially in men with BRCA2 muts.
by a significantly prolonged time to progression of worst pain and pain severity Second-stage enrolment is ongoing. Clinical trial information: NCT03047135.
vs PBO + ADT. Clinical trial information: NCT02677896.

5046 Poster Session (Board #158), Sat, 1:15 PM-4:15 PM 5047 Poster Session (Board #159), Sat, 1:15 PM-4:15 PM
Abiraterone acetate plus prednisone (AA+P) without continuing LHRH-therapy Plasma cell free DNA (cfDNA) based somatic aberrations detected in treatment
in patients with metastatic chemotherapy: Naive castrations-resistant prostate naı̈ve metastatic hormone sensitive prostate cancer (mHSPC), hormonally
cancer—Results from the SPARE-trial (NCT02077634). First Author: ablated mHSPC and metastatic castration resistant prostate cancer (mCRPC)
Carsten Henning Ohlmann, Malteser Hospital Bonn, Bonn, Germany states and their impact on survival. First Author: Manish Kohli, H. Lee Moffitt
Cancer Center, Tampa, FL
Background: The value of continuation of androgen deprivation therapy (ADT) in
metastatic castration-resistant prostate cancer (CRPC) remains controversial and Background: We identified plasma cell free DNA (cfDNA) based copy number vari-
clear evidence is lacking. Especially upon treatment with the life-prolonging ations (CNV); single nucleotide variants (SNVs) & TMPRSS-ERG fusion in four sub
cytochrome P450 17-alpha-hydroxylase/C17,20 lyase (Cyp17)-inhibitor, abir- states of metastatic prostate cancer (mPCa) and determined the impact on survival.
aterone acetate (AA), which in combination with prednisone (P), has the ability to Two mHSPC cohorts included treatment naı̈ve, gp-1 and mHSPC patients (pts)
further suppress serum testosterone levels over ADT alone, continuation of ADT responding to chronic androgen ablation (AA) (gp-2). Two mCRPC cohorts included
seems to be negligible. Methods: The exploratory phase II trial randomized mHSPC pts with PSA relapse on chronic AA (gp-3) and a clinically progressive mCRPC
CRPC patients to receive continued ADT plus AA+P versus AA+P alone cohort post AA (gp-4). Methods: Enrollment of mPCa pts was performed from 2009-14
(NCT02077634), funded by Jansen-Cilag GmbH, Germany. The primary endpoint who were followed until 2018. Plasma from collected blood was used for extracting
was rate of rPFS at month 12, not powered for a direct comparison between cfDNA. NGS was performed using Illumina HiSeq X for a preselected target panel of
treatment arms. Secondary endpoints included PSA response rate, objective 129 genes including DNA damage repair genes. Statistical analyses of genomic ab-
response, time to PSA progression and safety. Results: Altogether, 67 patients errations were performed in R 3.5.1 and Cox proportional-hazard models were used for
were randomized between 08/2014 to 04/2017. Median testosterone-levels (T) survival analyses. Results: A total of 255 pts were enrolled with 215 having adequate
cfDNA that passed NGS QC. Median study follow up was 90.2 (Range:73;99) months.
remained far below castrate-levels throughout treatment in all patients. However,
The table highlights pts in each gp with CNV, SNV, fusion events. ARamplification was
in 6 patients (18%) from Arm B, T-levels increased above castrate levels within
higher in mCRPC gps3&4 (20/103 pts) compared to 3/106 pts in mHSPC gps1&2 (p ,
28 days after cessation of AA+P treatment. Median treatment duration is shorter
0.001) and was prognostic for poor survival in mCRPC (p , 0.001;HR:3.34; 95%CI:
in Arm A. Safety analysis is underway and results will be presented. 1.9-5.76). 54/103 pts in gp3&4 had SNVs in TP53 compared to 34/106 pts in
Conclusions: Results of this exploratory study suggest that treatment with AA+P mHSPC gps1&2 (P , 0.01). SNVs in APC, AR, CDK12 and BRIP1 were also increased
without ADT may be effective in patients with mCRPC and that ADT may not be in gps3&4 (p , 0.01). Gp1 mHSPC pts with SNVs in ATM/CHEK2 had shorter re-
necessary in patients receiving AA+P. In some patients, serum-testosterone levels sponse to AA (p , 0.001; HR:3.66; 95%: CI:1.81-7.39). Conclusions: Plasma cfDNA
may rise rapidly upon treatment discontinuation so that the levels should be based somatic aberrations are detected with increased prevalence in mHSPC to
monitored closely. Clinical trial information: NCT02077634. mCRPC states. The ease of specimen collection and the need for molecular profiling in
LHRH+AA+P (Arm A) AA+P (Arm B) mPCa increases its potential for clinical applications in pt care.
Patients (n) 34 33 Median Number of Number of SNV Number of Number of Number of pts
Median age (range) 74 (60-86) years 76 (60-86) years PSA (ng/ml) CNVs detected pts with events pts with TMPRSS-ERG with fusion
Median baseline 31.9 (0.17-313.2) 20.59 (1.97-1680) (range) (gain/loss) CNVs (%) detected SNVs (%) fusion events events (%)
PSA (range) ng/ml ng/ml Treatment naı̈ve 6.65 7 gain; 11 loss 4/75 (5.3) 277 62/75 8 8/75 (10.6)
PSA-decline ‡50% 23/34 (67.6%) 24/33 (72.7%) mHSPC-gp-1 (N = (0,1438) (82.6)
Median serum testosterone level at baseline 0.08 (0.029-4.210a) 0.06 (0.029-0.429) 75)
ng/ml ng/ml mHSPC pts on chronic 0.29 3 gain; 5 loss 5/31 131 27/31 (87) 2 2/31 (6)
Median serum testosterone level at end-of- 0.029 ng/ml 0.029 ng/ml AA-gp-2 (N = 31) (0, 280) (16.1)
treatment visit mHSPC on AA with 2.7 19 gain; 8 loss 8/49 304 47/49 4 4/49 (8.1)
Median treatment 266 (0-988) 420 (55-1180) HR 1.667 PSA relapse (0, 460) (16.3) (95.9)
duration (d) (p=0.197)b mHSPC-gp-3 (N =
Time to PSA progression (d) 288 (26-989) 336 (0-1181) HR 1.733 49)
(p=0.188)b Clinically defined 15.9 51 gain; 22 22/54 (40) 381 52/54 14 14/54 (25.9)
Rate of rPFS at month 12 0.90 0.78 p = 0.4368b mCRPC gp-4 (N = (0, 435) loss (96.2)
54)
a
one patient was included by investigator despite non-castrate levels of tesosterone bstudy was not powered
for a direct comparison between treatment arms

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
300s Genitourinary (Prostate) Cancer

5048 Poster Session (Board #160), Sat, 1:15 PM-4:15 PM 5049 Poster Session (Board #161), Sat, 1:15 PM-4:15 PM
ARCHES: Efficacy of androgen deprivation therapy (ADT) with enzalutamide Suppression of testosterone production using abiraterone acetate (AA) with or
(ENZA) or placebo (PBO) in metastatic hormone-sensitive prostate cancer without androgen deprivation therapy (ADT) in metastatic castration resistant
(mHSPC). First Author: Andrew J. Armstrong, Duke Cancer Institute Center prostate cancer (mCRPC). First Author: Gautam Gopalji Jha, University of
for Prostate and Urologic Cancers, Durham, NC Minnesota, Minneapolis, MN
Background: ENZA has demonstrated benefit in men with metastatic and Background: Patients with metastatic prostate cancer (mPC) are treated with
nonmetastatic castration-resistant prostate cancer (CRPC). ARCHES assessed lifelong ADT even after progression and every treatment for mPC has been
the efficacy of ENZA with ADT in men with mHSPC, including pre-specified tested only in the setting of androgen deprivation. Androgen synthesis in-
subgroups based on prior therapy. Methods: ARCHES, a multinational, hibitor AA which inhibits synthesis of androgen that later get converted to
double-blind, Phase 3 study (NCT02677896), randomized patients (pts) with testosterone has been studied and approved only alongside ADT similar to all
mHSPC 1:1 to ENZA (160 mg/day) + ADT or PBO + ADT, stratified by disease other agents that have no or limited activity on testosterone suppression. We
volume (CHAARTED criteria) and prior docetaxel (doce) use. Primary endpoint reviewed the ability of AA to sufficiently suppress testosterone levels as
was radiographic progression-free survival (rPFS; centrally assessed radio- compared to AA plus ADT and its potential impact on cost savings.
graphic progression or death within 24 weeks of treatment discontinuation). Methods: This retrospective study included consecutive patients with
Secondary endpoints included time to initiation of new antineoplastic therapy mCRPC treated with AA alone or in combination with ADT (in absence of
and overall survival (OS). Treatment continued until disease progression or orchiectomy) who had been followed with serial testosterone values on
unacceptable toxicity. Results: 1150 men were randomized to ENZA (n = 574) therapy. A cost analysis was performed to determine the cost avoidance by
or PBO (n = 576). Overall, 63% had high-volume disease, 18% had prior doce, omitting leuprolide injections while on AA. The cost avoidance was calcu-
and 91% had prior ADT or orchiectomy (orch). Median follow-up was 14.4 mo. lated by multiplying the total number of injections by the wholesale ac-
ENZA + ADT significantly improved rPFS (Table); significant improvements in quisition cost of $5252.86 for a three month leuprolide injection.
rPFS were also reported in prior treatment subgroups. Secondary endpoints Results: Of 57 patients included in the final analysis, 36 received AA plus ADT,
improved with ENZA + ADT (Table), with no significant impact in time to 10 received AA alone, and 11 started off with AA plus ADT before transitioning
deterioration in urinary symptoms. OS data are immature. Grade 3–4 adverse to AA alone. Testosterone levels were drawn 235 times. Testosterone was
events (AEs) were reported in 23.6% of ENZA pts vs. 24.7% of PBO pts with no undetectable (below , 2 ng/dl) in both arms, 134 of 152 in combination arm
unexpected AEs. Conclusions: ENZA + ADT significantly improved rPFS and and 86 of 99 in the AA alone arm. The median testosterone concentration when
other efficacy endpoints vs. PBO + ADT in men with mHSPC. Preliminary detectable was 3 ng/dL in AA alone and 3.5 ng/dL in AA plus ADT. None in the
safety analysis appears consistent with the safety profile of ENZA in previous combination arm and only one testosterone value in AA arm had testosterone .
CRPC clinical trials. Clinical trial information: NCT02677896. than 30 ng/dl. The mean duration of AA use in this study was close to one year,
Endpoint ENZA + ADT PBO + ADT HR (95% CI)
and the total duration of therapy was approximately 61 years which could result
in elimination of 244 leuprolide administrations and approximately $1.29
rPFS: overall, n, median, mo n = 574, NR n = 576, 19.4 0.39* (0.30, 0.50)
Prior doce n = 103, NR n = 102, 14.0 0.53 (0.31, 0.92) million in total cost savings. Conclusions: AA alone is able to effectively
No prior doce n = 471, NR n = 474, 19.4 0.36 (0.27, 0.48) suppress testosterone synthesis in patients with prostate cancer. ADT with
Prior ADT or orch n = 535, NR n = 515, 19.4 0.41 (0.31, 0.52)
No prior ADT or orch n = 39, NR n = 61, NR 0.20 (0.06, 0.66) GnRH agonist or antagonist can be safely withheld while on therapy with AA
Time to initiation of new antineoplastic n = 574, 30.2 n = 576, NR 0.28* (0.20, 0.40) and testosterone values followed to confirm adequate androgen suppression.
therapy, n, median, mo
Objective response rate,† % 83.1* 63.7 _ This has acquired new significance after studies in patients with hormone
sensitive disease where median duration of treatment was 33 months which
*p , 0.0001; †Of those with measurable disease at baseline; NR = not reached
could translate to an avoidable expense of $55.5 million for 960 patients in
‘STAMPEDE’ study and $34.5 million in ‘LATITUDE’ study from leuprolide
administration in combination arm.

5050 Poster Session (Board #162), Sat, 1:15 PM-4:15 PM 5051 Poster Session (Board #163), Sat, 1:15 PM-4:15 PM
A multicentric phase II randomized trial of docetaxel (D) plus enzalutamide Clinical and genomic hallmarks of low PSA secretors in metastatic castration-
(E) versus docetaxel (D) as first-line chemotherapy for patients (pts) with resistant prostate cancer (mCRPC). First Author: Gustavo Rubio Romero,
metastatic castration-resistant prostate cancer (mCRPC): CHEIRON study. University of California San Francisco, San Francisco, CA
First Author: Orazio Caffo, Santa Chiara Hospital, Trento, Italy
Background: Metastatic disease burden out of proportion to low serum PSA
Background: D and E demonstrated to be efficacious in the treatment of level is frequently used as a clinical surrogate for the diagnosis of treatment-
mCRPC pts. Due to different antitumor mechanism of action of these agents, it emergent small cell neuroendocrine prostate cancer (t-SCNC), although
could be postulated that their combination can improve disease control. many t-SCNC patients (pts) have normal or elevated PSA levels. The clinical
CHEIRON study tried to demonstrate the candidate efficacy of chemo-hormonal and genomic characteristics of mCRPC pts who are Low PSA Secretors have
combination D+E versus D in mCRPC first-line. Methods: Eligibility criteria not been previously described. Methods: Eligible mCRPC patients (pts)
included mCRPC diagnosis, ECOG PS # 2, adequate renal, hepatic and he- underwent image-guided needle biopsy. Formalin-fixed paraffin embedded
matological functions, no prior treatment for mCRPC. Pts were randomized to tissue was evaluated with targeted next-generation DNA sequencing. Fresh
receive D 75 mg/m2 IV d1 q3w plus prednisone 5 mg PO BID for 8 courses alone frozen tissue from the same metastatic tumor underwent RNA-seq. A vali-
or plus E 160 mg PO daily for 24 weeks. Stratification criteria were presence of dated AR transcriptional signature was applied. Low PSA Secretors were
pain and visceral metastases. The primary endpoint of the study was the rate of defined as pts with PSA , 5 ng/mL plus $ 6 metastases on conventional
pts without disease progression (according to PCWG2) at 6 mos after ran- imaging at the time of tumor biopsy. Clinical and genomic characteristics
domization. Results: Between 09/2014 and 10/2017, 246 pts (median age 70 were compared between low PSA Secretors and all other pts. Results: Of 89
years, range 44-88, pain reported by 54 pts, visceral metastases present in 50 evaluable pts, 9 (10%) were identified as Low PSA Secretors. There was no
pts) were randomized to DE (120) or D (126). The rate of pts without disease difference between Low PSA Secretors and all other pts in: serum PSA at
progression at 6 mos was significantly higher in DE arm compared to D arm diagnosis, frequency of Gleason $ 8 adenocarcinoma at diagnosis, and
(89.1% vs 72.8%; p = 0.002). Similarly, a higher proportion of DE pts serum level of LDH, alkaline phosphatase, or hemoglobin at the time of
achieved a PSA reduction $ 50% compared to the baseline values compared to biopsy. Lung and/or liver metastases were more common in low PSA se-
the D pts (92% vs 69%; p , 0.0001). No differences were observed in terms of cretors (67% vs. 33%, p = 0.04). There was no difference in serum level of
objective response rate. Major haematological toxicities consisted of grade 3-4 LDH, alkaline phosphatase, or hemoglobin. Tumor biopsies from Low PSA
neutropenia (13 pts DE – 11 pts D); febrile neutropenia was observed in 10 DE Secretors were more likely to fall within a previously defined t-SCNC tran-
pts and in 6 D pts. At a median follow-up of 24 mos, the median progression free scriptional cluster (80% vs. 6%, p , 0.001). RB1 loss or inactivating
survival was 10.1 mos and 9.1 mos in DE and D arm, respectively (p = 0.01). In mutations appeared to be enriched in Low PSA Secretors (40% vs. 12%, p =
DE arm the median overall survival was 33.7 mos compared to 29.6 mos of the 0.09); there was no difference in frequency of TP53 alterations between
standard arm (p NS). Conclusions: The present study was the first phase II subgroups. AR transcriptional signature scores were lower in the Low PSA
randomized trial, which tested the addition of a new generation hormone agent to Secretor group (median score -3.63 vs. 0.66, p , 0.001). Conclusions: Low
D compared to D alone. From this data, DE improved the 6-mo disease control serum PSA levels in relation to metastatic tumor burden may be a reliable
with a prolongation of PFS compared to the standard chemotherapy. Clinical trial surrogate for the detection of mCRPC that harbors the transcriptional and
information: NCT02453009. genomic hallmarks of t-SCNC. Validation studies are warranted.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Genitourinary (Prostate) Cancer 301s

5052 Poster Session (Board #164), Sat, 1:15 PM-4:15 PM 5053 Poster Session (Board #165), Sat, 1:15 PM-4:15 PM
Aggressive variant prostate cancer (AVPC) molecular signature in castration- Treatment sequence in elderly metastatic castration-resistant prostate cancer
sensitive, de novo metastatic prostate cancer (M1PCa). First Author: Ana (mCRPC) patients (pts) in a prospective cohort study. First Author: Maria Jose
Aparicio, The University of Texas MD Anderson Cancer Center, Houston, TX Mendez-Vidal, Reina Sofı́a University Hospital, Cordoba, Spain
Background: AVPC are a subset of prostate cancers (PC) that share clinical Background: Abiraterone (Abi), enzalutamide (Enza) and docetaxel (Doc) are
features with the small-cell prostate carcinomas, a rare morphological variant all valid first-line (1L) mCRPC treatment options. SIOG guidelines (Droz, Eur
with atypical and virulent behavior. They are estimated to represent 30% of Urol 2017) recommend that fit elderly pts should receive the same treatment
lethal PC and are characterized by a molecular signature of combined defects as younger patients. Evidence of the optimal treatment sequence in this
($2) in Tp53, RB1 and PTEN (AVPC_MS). The AVPC_MS is associated with patient subpopulation is lacking. Methods: We evaluated the outcome of
androgen indifference in preclinical models and predicts for benefit from the elderly ($ 75 years [yrs]) pts treated in the prospective PROREPAIR-B cohort
addition of carboplatin to cabazitaxel in men with castration resistant prostate study (NCT03075735). We assessed the impact of 1L treatment option (Doc
cancer (CRPC). The prevalence and significance of the AVPC_MS in castration vs Abi/Enza) on overall survival (OS) and progression-free survival (PFS) to
sensitive M1PC is unknown. Methods: In a Phase II trial 119 men with M1PC 1L-therapy following PCWG2criteria. Uni- (UV) and multivariable (MV) cox-
treated with 6 months of standard systemic therapy (SST) were randomized to regression models were used. MV model covariates included local therapy,
the addition of local therapy. PCa samples obtained at diagnosis and after Gleason Score, stage IV at diagnosis, visceral metastases, ALP ($ULN), LDH
6 months of SST were stained for markers including Tp53, RB1 and PTEN, and ($ULN), haemoglobin (Hb; #LNL), albumin (#LNL) and ECOG PS.
subject to whole genome sequencing. Tp53 was considered defective if Results: 419 pts were included in the study. Of these, 137 (32,7%) had
expressed in $ 10% of tumor cells, and RB1 and PTEN if in # 10%. Pro- age $ 75 yrs. 48 (35%) received docetaxel and 88 (64.2%) had Abi/Enza as
gression free survival (PFS) was estimated from SST start. Results: To date first-line therapy. Of the 121 pts that progressed on 1L-therapy, 30 (24.8%)
specimens from 38 men have been evaluated. Immunohistochemistry (IHC) did not receive 2L therapy. Choice of 2L-therapy was: Doc in 37 (30,6%),
results are shown below. The median PFS of men with AVPC_MSPOSITIVE vs Abi/Enza in 38 (31.4%), Cabazitaxel in 9 (7.4%) and Radium-223 in 7
AVPC_MSNEGATIVEbaseline prostate tumors was 9.3 vs 15 months (HR 1.09, (5.8%) pts. Pts treated with 1L-Doc had higher rates of visceral metastases
95%CI 0.42-2.87, P=0.852). Conclusions: The rate at which the AVPC (22.9% vs 5.7%; p=0.003), high ALP (68.8% vs 43.2%; p=0.004) and low
molecular signature is detected in castration sensitive M1PCa appears similar Hb (12.5% vs 3.4%). PFS to 1L-therapy was longer for Abi/Enza than for Doc
to that in CRPC. This may have therapeutic implications. Updated molecular treated pts (9.6 vs 8.3m; HR: 0.52; p=0.001). The pattern of disease
and outcome data will be presented upon acceptance. progression (PSA, radiographic, clinical) was similar in Doce and Abi/Enza
treated pts. No difference between pts treated with initial Abi/Enza vs Doc
Baseline Prostate 6month Prostate Baseline Me- 6month Metas-
Biopsies (n=38) Biopsies (n=38) tastases (n=9) tases (n=15) was observed in OS (28.2 vs 24.8m; HR:1.18; p=0.474). No significant OS
differences were observed in the MV model. Conclusions: No differences in
Tp53, n evaluable 34 35 6 15
Positive (‡10% positive 13 (38.2%) 7 (20.0%) 2 (33.3%) 5 (33%) OS were observed between treatment sequences starting with Doc vs Abi/
cells) Enza in pts $ 75 yrs. Pts treated with 1L-Doc had worse baseline prognostic
RB1, n evaluable 38 36 7 15 features. Age should not be considered as a factor for treatment choice in
Negative (£ 10% positive 5 (13.2%) 3 (8.3%) 1 (14.3%) 2 (13%)
cells) elderly mCRPC pts based on treatment outcome.
PTEN, n evaluable 32 27 6 15
Loss, homogeneous 11 (34.4%) 17 (63.0%) 3 (50.0%) 9 (60%)
Loss, heterogeneous 7 (21.9%) 3 (11.1%) 1 (16.7%) 3 (20%)
AVPC_MS, n evaluable 30 26 6 15
Positive (‡ 2 abnormal 8 (26.7%) 6 (23.1%) 1 (16.7%) 5 (33%)
Tp53, RB1 and PTEN)

5054 Poster Session (Board #166), Sat, 1:15 PM-4:15 PM 5055 Poster Session (Board #167), Sat, 1:15 PM-4:15 PM
Results from a phase Ib/II study of enzalutamide and metformin in men with Complex biologic heterogeneity of de novo hormone naı̈ve metastatic prostate
castration resistant prostate cancer (CRPC). First Author: Mamta Parikh, cancer (HNPCa): Comparison of early progressors and prolonged responders to
University of California Davis Comprehensive Cancer Center, Sacramento, CA initial systemic treatment. First Author: Brian Francis Chapin, The University of
Texas MD Anderson Cancer Center, Houston, TX
Background: Preclinical models have shown that autophagy is a cell survival
mechanism to overcome treatment-induced stress and facilitate progression Background: Given the absence of biologically based biomarkers current therapy
and resistance in CRPC. Metformin (met) demonstrates autophagy inhibition allocation strategies for men with mHNPCa are based on anatomic distribution and
and resensitizes CRPC tumors to enzalutamide (enza) as a combination in in volume of metastases. We sought to determine the strength of the association be-
vitro models. A Phase Ib/II trial was conducted to evaluate the combination. tween clinical predictors (including met volume) and outcomes in men with mHNPCa
Methods: Eligible patients (pts) had established CRPC, ECOG 0-2, adequate at extremes of response to initial systemic therapy (ST). Methods: Data from a
hematologic and organ function, and # 2 prior treatments for CRPC. Major population screened for a phase II trial of Best ST +/- local therapy in mHNPCa was
exclusion criteria included prior enza or met treatment, presence of brain analyzed (NCT01751438). Pts had to have sustained responses to ST ($6 mos) in
metastases, history of DM2 or seizures. An initial cohort of 3 pts were treated order to be randomized. Early Progressors (EP, castration resistant progression ,6
mos from start of ST per PCWG2) were not randomized. Prolonged Responders (PR)
with enza 160 mg PO daily and met 500 mg PO bid (DL1), with met dose
were defined as those with progression free survivals (PFS) . median for the 109
escalated to 1000 mg PO bid (DL2) in subsequent 3 pts if dose-limiting
randomized pts. Baseline demographics were compared between groups (Fisher’s
toxicity (DLT) was observed in #1 pt in DL 1, and cycles of 28 days. Once exact test for categorical and Wilcoxon rank-sum test for continuous variables) and
primary objective of maximum tolerated dose (MTD) was established, ad- Kaplan Meier estimates of OS determined. Immunohistochemistry and genomic
ditional pts were enrolled at MTD for a total of 24 evaluable pts. Secondary sequencing of untreated tumor biopsies are ongoing. Results: Of 208 screened pts,
objectives of the study included PSA response and safety. Results: From 8/ 27 (13%) were identified as EPs, with a median PFS of 5.9 mos (95%CI: 5.3-6.5).
24/16-12/31/18, a total of 3 pts were enrolled to DL1, with no DLTs ob- The median PFS for the randomized cohort was 16.8 mos (95%CI: 12.7-37.6). 55
served in that cohort, with an additional 21 enrolled to DL2; 12 remain on (50%) pts were identified as PRs, with median PFS not reached (NR) at a median
trial. Median age was 71, with 8/24 (33%) and 3/24 (12.5%) with prior follow up of 31 mos (range: 7.5-75). Patients’ demographics are below (Table),
docetaxel and abiraterone treatment, respectively. One DLT due to Grade 3 grouped by EP and PR cohorts. Median OS was 24.8 mos vs NR in the Eps compared
abdominal pain was observed at DL2 prior to establishing MTD. Grade 3 AEs to PRs (p,0.001). Conclusions: The EP group was enriched for high risk features and
observed in 1/24 (4%) of pts were diarrhea, fatigue, hypertension, lower GI had a worse outcome. However, 33% of men in the EP group had low volume disease
bleed, myalgia; 2/24 (8%) had abdominal pain. One pt had Grade 4 hal- and 22% in the PR group had high volume disease. This suggests that clinical volume
lucinations. Among pts who discontinued, 8 were discontinued for pro- is an inadequate predictor of biologic response and highlights the need for biologic
gression, 3 withdrew consent, and 1 experienced an SAE (seizure). Pts sub-typing of de novo mHNPCa.
received median of 11 cycles (1-31) of treatment. PSA response $ 50% was Early Progressor Prolonged Responder
observed in 19/24 (79%) pts. Conclusions: The combination of enzaluta- Demographic n=27 n=55 p-value
mide 160 mg PO daily and metformin 1000 mg PO bid is well-tolerated, and Age, yrs (range) 63 (53-72) 65 (41-83) 0.39
clinically active. Clinical trial information: NCT02339168. Median PSA (Diagnosis), ng/mL (range) 70 (4-8383) 34 (6-2000) 0.16
Local Symptoms (Diagnosis), n (%) 21 (78%) 33 (60%) 0.14
CHAARTED Criteria, n (%) <0.001
High 18 (67%) 12 (22%)
Low 9 (33.3%) 43 (78%)
Docetaxel, n(%) 15 (56%) 10 (18%) 0.001
T Stage, n (%) 0.01
T2 8 (31%) 29 (54%)
T3 12 (46%) 23 (43%)
T4 6 (23%) 2 (4%)
Node Positive, n (%) 20 (74%) 33 (60%) 0.23

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
302s Genitourinary (Prostate) Cancer

5056 Poster Session (Board #168), Sat, 1:15 PM-4:15 PM 5058 Poster Session (Board #170), Sat, 1:15 PM-4:15 PM
Impact of germline DNA-repair gene BRCA2 and CHEK2 mutations on time AR changes in circulating-tumor DNA (ctDNA) in patients with metastatic
to castration resistance in patients with metastatic hormone-naı̈ve prostate castration-resistant prostate cancer (mCRPC) treated with high-dose
cancer: A single center analysis. First Author: Vsevolod Borisovich Matveev, testosterone. First Author: Marcus W. Moses, Tulane University, New
N.N. Blokhin National Medical Research Center of Oncology, Moscow, Orleans, LA
Russian Federation
Background: High-dose testosterone (HDT) is active in mCRPC pts and may
Background: Germline mutations in DNA repair genes are common in patients allow successful re-sensitization to previously utilized androgen-axis targeted
with andvanced and metastatic prostate cancer (mPCa). Although BRCA mu- therapies. The relationship of genomic alterations in AR gene to HDT re-
tation carriers have worse outcomes than noncarriers when conventionally treated sponsiveness is unclear. Methods: Analysis of consecutive pts treated with $1
for local or locally advanced PCa, recent studies reported conflicting results dose of HDT (testosterone cypionate q 2-4 weeks n = 29; continuous gel n = 4).
regarding their aggressiveness in patients with advanced disease. This study Baseline characteristics, ctDNA data (Guardant360), and clinical outcomes
aimed to examine the impact of germline BRCA1/2 and CHEK2 mutations on were assessed. Presence of genomic AR alterations included amplifications
time to castration-resistance in patients with mPCa, receiving hormonal an- (amps) and mutations (muts); all muts had allele fraction $0.3%. PSA re-
drogen deprivation therapy (ADT). Methods: A total of 76 patients with hormone- sponse rates included PSA declines of . 30% or $50%. PSA-progression-free
naive mPCa treated with first line ADT by luteinizing hormone-releasing hormone survival (PSA-PFS) was defined as HDT start date to PSA $ 25% over baseline
analogue (LHRHa) between 2014 and 2017 were recruited. Median follow-up after a second confirmed PSA rise. Results: Between May 2016 and Feb
was 34,8 mo. We focused on age, volume of metastatic spread, histologic grade, 2018, 33 mCRPC pts had median age 73 (58-85), 39% Gleason 8-10, 100%
family history. All patients were genotyped for germline mutations in the BRCA1, bone mets, 24% nodes + bone, and median baseline PSA level 36.1 ng/mL
BRCA2 and CHEK2 genes by polymerase chain reaction real-time and the (0.04-1290). HDT was given post-median of 2 (1-10) CRPC therapies. 73%
Sanger sequencing. We used the standard definition of castration-resistance PCa (24/33) of pts previously received abiraterone (n = 14), enzalutamide (n = 4),
(CRPC). Median time to CRPC were estimated using the Kaplan-Meier method, or both sequentially (n = 6) prior to HDT for a median of 10.5 months (0.7-
generated curves were compared using the log-rank test. Cox regression analyses 56.8). Baseline ctDNA showed 42% AR alterations (amps = 8, muts = 4,
were used to assess the prognostic value of BRCA1/2 and CHEK2 mutations. both = 2); 33% TP53, and 6% DNA repair (ATM n = 1; BRCA2 n = 1). With
Results: Pathogenic and likely pathogenic germline mutations in the BRCA2 and median follow-up 4.4 months, HDT given for median of 4.2 months (95% CI,
CHEK2 gene were identified in 19 (25 %) patients. No cases of BRCA1 mu- 3.6-4.8); 29% had PSA $50% response and 45% PSA $30% response.
tations were detected. Median time to CRPC was significantly shorter in BRCA2 Median PSA-PFS is immature at 5.5 months (95% CI, 1.5-9.5); 14 pts still on
and CHEK2 mutation carriers (7.9 mo, 95 % confidence interval (CI) 2.6 – HDT treatment. Grade $3 AEs were observed in 6% of pts (G4 thrombocy-
13.3), than in non-carriers (48.7 mo, 95 % CI 31.1 – 68.3, p , 0,001). There topenia = 1; G4 asthenia = 1). For pts with baseline AR alterations and HDT
was no significant difference in median time to CRPC in BRCA2 (7.9 mo, 95 % treatment, repeated ctDNA assays (n = 7) showed that 100% had decreased
CI 0.0 - 16.3) and CHEK2 mutation carriers (6.1 mo, 95 CI 5.0 - 7.2, p = 0,448) AR alterations. No relationship between PSA response and baseline ctDNA AR
meanwhile both were shorter than in non-carriers (p = 0.002 and , 0,001). characteristics are discerned at this time. Conclusions: HDT was safe and
Multivariate analysis confirmed both BRCA2 (hazard ratio [HR]: 2.63; 95 CI active in a subset of mCRPC. Responses were clearly noted for men receiving
1.32-5.26, p = 0.006) and CHEK2 (HR 6.66, 95 CI 2.35-18.89, p , 0.001) continuous daily testosterone gels, thus continuously high testosterone levels
mutations as an independent prognostic factor for time to CRPC, particularly in are active in addition to injection-induced bipolar changes. Further un-
mPCa with low-volume metastatis spread (HR 3.09, 95 % CI 1.36–7.05, r = derstanding of the genomic alterations predicting responsiveness to HDT in
0.007 and HR 14.1, 95 % CI 3.65-54.4, p , 0.001). Conclusions: BRCA2 and mCRPC is required.
CHEK2 carriers had worse outcomes (shortened time to CRPC) than noncarriers
when conventionally treated for metastatic PCa by standard first-line hormone
treatment with LHRHa.

5059 Poster Session (Board #171), Sat, 1:15 PM-4:15 PM 5060 Poster Session (Board #172), Sat, 1:15 PM-4:15 PM
A circulating tumor cell specific RNA assay for assessment of androgen Computerized histomorphometric features of glandular architecture predict
receptor signaling inhibitor sensitivity in metastatic castration-resistant risk of biochemical recurrence following radical prostatectomy: A multisite
prostate cancer. First Author: Pai-Chi Teng, Samuel Oschin Comprehen- study. First Author: Patrick Leo, Case Western Reserve University, Department
sive Cancer Institute, Cedars-Sinai Medical Center, Los Angeles, CA of Biomedical Engineering, Cleveland, OH
Background: Our objective is to develop a circulating tumor cell (CTC)-RNA Background: Following a radical prostatectomy (RP) for prostate cancer, a
assay for characterizing clinically relevant RNA signatures for the assessment patient may experience biochemical recurrence (BCR), defined as two con-
of androgen receptor signaling inhibitors (ARSIs) sensitivity in metastatic secutive prostate specific antigen (PSA) readings . 0.2 ng/mL. BCR is cor-
castration-resistant prostate cancer (mCRPC) patients. Methods: We de- related with metastasis and disease specific survival. Extant molecular based
veloped NanoVelcro CTC-RNA Assay by combining Thermoresponsive(TR)- companion diagnostic tests for predicting risk of BCR and disease progression
NanoVelcro CTC purification system with NanoString nCounter platform tend to be expensive and tissue destructive. We sought to evaluate whether
for CTC purification and RNA analysis. Based on the well-validated, tissue- computer extracted features of glandular architecture from routine digitized
based Prostate Cancer Classification System (PCS), we selected the most H&E slides could predict post-RP BCR risk. Methods: RP specimens from 683
aggressive and ARSI-resistant subtype- the PCS1, for CTC analysis. We patients (184 with BCR, 499 without) with post-surgical PSA follow-up in-
applied a rigorous bioinformatic process to develop a CTC-PCS1 panel that is formation were gathered from six sites. Median non-BCR follow-up was 3.2
specific to PC CTCs. We validated NanoVelcro CTC-RNA Assay and CTC- years. A representative tumor area was annotated on the diagnostic H&E slide
PCS1 panel with PC cell lines to demonstrate sensitivity and specificity of the of each patient. 324 (131 BCR) patients from two sites formed the training
PCS1 Z score (the likelihood estimate of the PCS1 subtype) for identifying set. The other 359 (53 BCR) patients formed the validation set. Glands were
PCS1 subtype and ARSI resistance. We then selected 31 blood samples from segmented by a deep learning model. 216 features describing gland arrangement,
23 PC patients receiving ARSIs to test in our assay. The PCS1 Z score of each shape, and disorder were then extracted. An elastic net Cox proportional hazards
sample was computed and compared with ARSI treatment sensitivity. model was constructed from the training set using the top 10 stable features
Results: We established a 16-gene CTC-PCS1 panel that consists of CTC- identified via feature selection. Risk score thresholds were chosen on the training
specific RNA signatures. The validation studies using PC cell lines showed set to stratify patients into low-, medium-, or high-risk. Validation set results were
that the assay can detect the RNA transcripts with high sensitivity and evaluated by the log-rank test and hazard ratio. For the 172 (37 BCR) patients for
scalability in the range of 1-100 cells. We also showed that the genes in CTC- whom Gleason grade and preoperative PSA values were available, risk classifi-
PCS1 panel is highly expressed in PC cells. We further demonstrated that the cations were compared using Cox proportional hazards regression. Results: Nine
CTC-PCS1 panel is highly specific in identifying PCS1-like samples, and the of the top features were gland shape features and one was a gland arrangement
high PCS1 Z score is associated with ARSI resistance. In patient bloods, feature. The hazard ratio between the low- and high-risk groups on the validation
ARSI-resistant samples (ARSI-R, n=14) had significantly higher PCS1 Z set was 3.04 (p , 0.05). The histomorphometric classifier was predictive of BCR
scores as compared with ARSI-sensitive samples (ARSI-S, n=17) (Rank-sum (p , 0.05, hazard ratio = 1.63) independent of Gleason grade group and pre-
test, P=0.003). In 8 patients who were initially sensitive to ARSI (ARSI-S) operative PSA in multivariate testing. Conclusions: Computer extracted features of
and later developed resistance (ARSI-R), we found that the PCS1 Z score gland morphology can stratify post-RP patients by BCR risk. Our computerized
increased from the time of ARSI-S to the time of ARSI-R (Pairwise T-test, histomorphometric model could serve as a prognostic tool in the post-RP setting.
P=0.016). Conclusions: Using our new methodology, we developed a first- BCR survival information for the 359 patient validation set.

in-class CTC-RNA assay and demonstrated the feasibility of transforming Risk group Num. % BCR Median BCR year p vs. low-risk HR vs. low-risk

clinically-relevant tissue-based RNA profiling into CTC tests. This approach Low 189 11 1.3 -- --
Medium 104 13 1.8 0.3030 1.38 (0.74 – 2.58)
allows for detecting RNA expression relevant to clinical drug resistance in a High 66 30 1.0 0.0002 3.04 (1.48 – 6.23)
non-invasive fashion, which can facilitate patient-specific treatment se- Log-rank p-value and hazard ratio (95% confidence interval) are reported. The model separates low- and high-risk
lection and early detection of drug resistance- a goal in precision oncology. patients.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Genitourinary (Prostate) Cancer 303s

5061 Poster Session (Board #173), Sat, 1:15 PM-4:15 PM 5062 Poster Session (Board #174), Sat, 1:15 PM-4:15 PM
Risk of depression following prostate cancer work-up: A nationwide study. First Prevalence of pathogenic germline variants in DNA repair by race, age, and
Author: Anne Sofie Friberg, Danish Cancer Society Research Center, ethnicity in men with prostate cancer. First Author: Alexandria Mara,
Copenhagen, Denmark Department of Medicine, Duke University School of Medicine, Durham, NC
Background: Little is known about the psychological impact of undergoing Background: Patients with advanced prostate cancer (PC) frequently harbor
evaluation for prostate cancer (PCa). We investigated the risk of developing a pathogenic or likely pathogenic (P/LP) germline variants (GVs) in mismatch
depression following PCa work-up with benign and malignant findings, re- repair (MMR) and homologous repair (HR) enzymes which have clinical
spectively, compared with cancer-free men. Methods: A nationwide cohort of and treatment implications. However, whether the prevalence of such GVs
men who underwent prostate needle biopsies in Denmark from 1997–2011 differ by race, ethnicity, family history, or age is unknown in men with PC.
was identified through the Danish Prostate Cancer Registry. Primary outcome Methods: This is a retrospective analysis of germline DNA from men with PC in
was indication of moderate to severe depression defined as hospital contact for the United States, tested by Invitae. Baseline characteristics including self-
depression or first redemption of a prescribed antidepressant. For comparison, identified race, ethnicity, family history (FH), and age were recorded. Race and
we selected a minimum of five age-matched cancer-free men per man who had ethnicity were analyzed by 3 cohorts: non-Ashkenazi Caucasian Americans
undergone PCa specific diagnostic work-up. We excluded men with other (CA), non-Ashkenazi African Americans (AA), and Ashkenazi Jewish Americans
cancer, major psychiatric disorder or use of antidepressants up to three years (AJ). Chi-square testing was performed to identify significant differences
before study entry. Information on outcome and covariates (age, period, co- across these categories between the three cohorts, with respect to combined
habitation status, income quintile and comorbidity) were retrieved from Na- and individual pathogenic MMR (MSH2/6, MLH1, PMS2, and MUTYH) and
tional Danish registries. We illustrated the risk of depression by cumulative HR genes (BRCA1/2, ATM, CHEK2, RAD51D, and PALB2). Results: 3057
incidence functions. Data were analyzed using Cox models adjusted for men were included in the final analysis: 2248 (74%) men were CA, 229 (7%)
possible confounders. Results: We identified 54,766 men who underwent were AA, and 210 (7%) were AJ. Of these, 2665 (87%) men had a FH of PC
work-up including transrectal biopsies of the prostate, among these, 21,419 and 463 (15%) had a P/LP GV. In addition, 1068 (35%) were found to have a
biopsy sets were benign and 33,347 men were diagnosed with PCa. We found variant of uncertain significance, and 35 (1.6%) had the HOXB13 G84E
an increasing cumulative incidence of depression in all groups. However, men variant. There were no significant differences in the overall prevalence of MMR
diagnosed with PCa had a significantly higher risk throughout up to 18 years of (CA 1.5% vs AA 0.9% vs AJ 1.4%, p = 0.89) or HR genes (CA 7.8% vs AA 7.9%
follow-up. The adjusted hazard ratio (HR) of depression in men diagnosed with vs AJ 10.5%, p = 0.37) by race/ethnicity. With respect to individual genes, AJ
PCa was increased throughout follow-up with the highest risk in the two years had a higher prevalence of pathogenic BRCA1 alterations (AJ 3.3% vs CA
following diagnosis (HR 2.77, 95% CI 2.66–2.87). After undergoing biopsies, 0.8% vs AA 1.7%, p = 0.0034) and CHEK2 alterations (AJ 4.3% vs CA 2.8%
men with benign results had an increased risk of depression (HR 1.22, 95% CI vs AA 0.4%, p = 0.02). There were no significant differences in the prevalence
1.14–1.31) in the first two years compared with cancer-free men; hereafter, we of individual or specific classes of GVs between those with or without a self-
found no difference. Conclusions: We found an increased risk of depression in reported FH of prostate or breast/ovarian cancers. There was also no associ-
men following diagnostic work-up for PCa compared with a matched back- ation between prevalence of HR genes and age at germline testing (p = 0.40).
ground population. In men diagnosed with PCa, the risk remained increased Conclusions: This national study found that the overall prevalence of patho-
throughout the study period. Future studies are needed to further analyze the genic GVs in MMR and HR genes do not differ by race, ethnicity, or age at the
impact of stage and treatment modalities. time of testing, and suggests that all men with advanced prostate cancer
should be offered germline testing.

5063 Poster Session (Board #175), Sat, 1:15 PM-4:15 PM 5064 Poster Session (Board #176), Sat, 1:15 PM-4:15 PM
Evolving natural history of metastatic prostate cancer. First Author: Nellie Ductal (PDC), acinar (PAC) and neuroendocrine (PNC) carcinomas of the
Nafissi, Mayo Clinic Arizona, Phoenix, AZ prostate: A comparative comprehensive genomic profiling (CGP) study. First
Author: Gennady Bratslavsky, Urologic Oncology Branch, National Cancer
Background: The systemic therapies available to patients with metastatic
Institute at the National Institutes of Health, Bethesda, MD
prostate cancer (mPC) have improved dramatically over the past decade. Prior
to 2010, the only agents with a proven survival benefit for patients with Background: PDC, PAC and PNC are histologic subtypes of prostate cancer (PC). We
metastatic disease were androgen deprivation therapy and docetaxel. Since queried whether these subsets would share similar genomic alterations (GA) reflecting their
then, five new agents have been FDA approved and have proven survival benefit disease biology and clinical features. Methods: CGP was performed using a hybrid capture-
based assay on 61 PDC, 4,132 PAC and 217 PNC. Tumor mutational burden (TMB) was
in phase III trials. Anecdotal experience suggests that the increased available determined on 1.1 Mbp of sequenced DNA and microsatellite instability (MSI) was de-
lines of therapy have changed the profile of mPC to include a higher prevalence termined on 114 loci. Results: The age, GA per tumor and TP53 GA of PDC, PAC and PNC
of visceral metastases. Methods: A retrospective review of 474 patients with were similar (Table). RB1 GA were predominant in PNC. TMPRSS2:ERG fusions were most
prostate cancer who died in 2009 and in 2016 was performed. Patients with frequent in PNC, intermediate in PAC and lowest in PDC. AR GA were more often identified
metastatic disease who had imaging within 6 months of death were included. A in PAC than PDC or PNC whereas PTEN GA were most frequent in PDC than PAC or PNC.
total of 164 patients were eligible for analysis. Results: Mean age at death Targetable GA were identified in all 3 groups when focused on BRCA2 (PARP inhibitors) and
overall was 77.4 years (SD 9.5) and did not differ signifiantly by cohort. Overall PIK3CA (MTOR inhibitors). ATM GA (PARP inhibitors) were more common in PAC than PDC
or PNC. BRAF GA (BRAF/MEK inhibitors) were more frequent in PDC and PAC than PNC.
rates of visceral and distant metastases to include lung, liver, adrenal, brain,
CDK12 GA potentially associated with immunotherapy (IO) benefit were similar in PDC and
renal, spleen, and thyroid, were higher in patients who died in 2016 as PAC and low in PNC. Low frequencies of MSI-High and low median TMB levels were similar
compared to those who died in 2009 (40.0% and 26.1%, respectively, p = in all 3 groups. Conclusions: The pathologic features of PDC, PAC and PNC have been
0.07). Lung metastases were more prevalent in patients who died in 2016 classically maintained as representative of 3 different tumor types with potentially con-
versus in 2009 at 26.3% and 13.0%, respectively (p = 0.05). Patients who trasting histogenesis. In the current CGP based study, all 3 tumor types did not display
died in 2009 received a median of 3 (range 1-10) systemic treatments versus 4 significant differences in genomic signatures other than the high RB1 GA. CGP may reveal
(range 0-13) in those who died in 2016 (p = 0.005). Forty-four percent of biomarkers that could direct patients to targeted (PARP, MTOR and BRAF/MEK inhibitors)
patients who died in 2016 used five or more lines of systemic treatments or immunotherapies (CDK12 GA, MSI-High or high TMB status) especially in PDC and PAC.
compared to 26.1% of patients in 2009. Conclusions: The emergence of new Top
Age Median GA/ Un-targetable Top TMB TMB
systemic therapies for mPC is changing the natural history of the disease. Forty range tumor GA Targetable GA CDK12 GA MSI High Median TMB ‡ 10 mut/Mb ‡ 20 mut/Mb
percent of patients now develop visceral metastases compared to 26% in the PDC 61 cases 66 4.1 TP53 46% PTEN 48% 8.2% 1.9% 2.6 3.3% 1.6%
past. These changes will drive the need for new treatment approaches targeting (51-86) TMPRSS2 21% AR 11%
FAS 11% PIK3CA 11%
visceral metastases. RB1 10% BRCA2 9%
BRAF 7%
RET 4%
Site of selected metastases 2009, N = 69 2016, N = 95 P-value ATM 0%
PAC 4,132 cases 66 4.4 TP53 43% PTEN 32% 6.0% 2.6% 2.6 3.8% 0.7%
Bone 61 (88.4%) 88 (92.6%) 0.42 (34-88) TMPRSS2 32% AR 21%
MYC 12% BRCA2 10%
Lymph node 41 (59.4%) 60 (63.2%) 0.63 RB1 7% PIK3CA 7%
Lung 9 (13.0%) 25 (26.3%) 0.05 ATM 6%
BRAF 4%
Liver 9 (13.0%) 20 (21.1%) 0.22 PNC 67 5.0 TP53 64% PTEN 33% 1.8% 1.0% 3.5 8.8% 2.8%
Adrenal 6 (8.7%) 8 (8.4%) 0.99 217 cases
PNC (30-86) RB1 55% BRCA2 12%
Brain 1 (1.4%) 4 (4.2%) 0.40 217 cases
TMPRSS2 42% AR 11%
Renal 0 (0.0%) 2 (2.1%) 0.51 MYC 11% PIK3CA 5%
Spleen 0 (0.0%) 1 (1.1%) 0.99 ATM 3%
BRAF 1%
Thyroid 1 (1.4%) 0 (0.0%) 0.42

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
304s Genitourinary (Prostate) Cancer

5065 Poster Session (Board #177), Sat, 1:15 PM-4:15 PM 5066 Poster Session (Board #178), Sat, 1:15 PM-4:15 PM
SLFN11 expression (exp) in castration-resistant prostate cancer (CRPC) Association of plasma DNA repair deficient (DRD) status with clinical outcome
patients (pts) to predict response to platinum-based chemotherapy (PLT). of metastatic castration-resistant prostate cancer (mCRPC) patients (pts) treated
First Author: Vincenza Conteduca, Dana Farber Cancer Institute, Boston, MA with abiraterone acetate (AA) plus prednisone/dexamethasone (+P/D). First
Author: Dong Shen, Janssen Research and Development, Spring House, PA
Background: Schlafen family member 11 (SLFN11) sensitizes tumor cells to
DNA-damaging agents and has been investigated as a potential predictive Background: Somatic DRD deficiency in 15-30% of mCRPC pts have been
biomarker of response to PLT and PARP inhibitors, especially in small cell lung observed and inhibition of enzyme poly ADP ribose polymerase (PARP) could
cancer (Lok, CCR 2017; Pietanza, JCO 2018). We aimed to explore the prove beneficial. We aimed to define DRD status using plasma from pts treated
predictive/prognostic role of SLFN11 in PLT-treated CRPC. Methods: We on AA and evaluate associations with prospectively-collected outcome mea-
assessed tumor exp of SLFN11 in PLT-treated, metastatic CRPC pts by sures. Methods: Plasma DNA samples (128 baseline [BL], 134 cycle 2 day 1
RNAseq (N=27) and/or CTCs (N=20) (via the Epic Sciences platform). In [C2D1], 46 progression [PROG]) from chemotherapy-naı̈ve mCRPC pts in a
addition, tumor morphology for neuroendocrine (NE) features and genomic phase 2 study (NCT01867710) evaluating AA+P/D were subjected to cus-
status of select genes (ie, AR, TP53, RB1, BRCA2, BRCA1, ATM) by whole tom target-capture next-generation sequencing. DRD assay was optimized and
exome sequencing were evaluated. Statistical comparisons used Cox re- validated to detect pathogenic point mutations, small insertions/deletions, and
gression analysis and Kaplan Meier method for the association with overall/ copy number alternations (DRD+) in 8 DRD genes: BRCA1, BRCA2, FANCA,
radiographic progression free survival (OS/rPFS). A dose response curve with ATM, CHEK2, HDAC2, BRIP1, and PALB2. Analysis for genomic aberrations
PLT was performed in patient-derived organoids using Cell Title Glo according was a secondary exploratory objective. Associations with overall survival (OS),
to the manufacturer’s protocol. Results: 41 CRPC (including 20 CRPC-NE) progression-free survival (PFS), and radiographic PFS (rPFS) were assessed
treated with PLT monotherapy (N=3) or PLT combination therapy (N=38) using Cox regression models and Kaplan Meier analyses. Results: 11.7% of BL
between August 2013 and December 2017 were evaluated. Median age was and 17.4% of PROG were DRD+. Bi-allelic was observed in 73.3% of BL DRD+
67.1 years (range 51-91). Median number of prior therapies was 2 (range 1-7). samples. Shorter PFS was observed in BL DRD+ vs DRD- (5.3 vs 15.5 mo;
A longer median rPFS was observed in all SLFN11+ pts treated with PLT HR: 2.32; 95%CI:1.39-4.28; P , 0.002). Median PFS for BL DRD biallelic +
(regardless of histology, RB1, TP53, PTEN, or DNA repair status) compared to vs DRD biallelic- was 5.1 vs 15.4 mo (HR: 2.49; 95% CI: 1.23-4.38; P ,
SLFN11- [5.2 vs 2.3 months, HR 3.5, 95%CI 1.6-7.7, P,0.0001]. No 0.0095). For multivariate analysis using DRD+, ALP, and LDH as covariates,
association was reported for OS. On multivariable analysis (Table), SLFN11 DRD+ (HR: 2.1; 95% CI: 1.18-3.75; P , 0.012) and high ALP (HR: 1.66;
was an independent factor associated with rPFS. Organoids derived from 95% CI: 1.08-2.56; P , 0.021) were strongly associated with worse PFS.
patient tumors expressing SLFN11 showed sensitivity to PLT in vitro. Median OS for BL DRD+ vs DRD- was 28.8 vs 41.3 mo (HR: 1.67; 95%CI:
Conclusions: SLFN11 exp identifies both CRPC and CRPC-NE pts with a better 0.88-3.18; P = 0.116). Median rPFS for BL DRD+ vs DRD- was 16.2 vs 20.9
response to PLT. Patient-derived organoids expressing SLFN11 confirmed mo (HR: 1.64; 95%CI:0.83-3.21; P = 0.152). Of 39 Pts with BL, C2D1 and
increased sensitivity to PLT. Larger prospective evaluation of therapy decisions PROG samples, 3 were DRD+ (7.7%) at all 3 timepoints, 3 (7.7%) only at BL,
based on SLFN11 exp is now required. 3 (7.7%) only at PROG (bi-allelic), 2 (5.1%) had extra deletion at PROG.
rPFS OS
Conclusions: Patients with mCRPC harboring DRD+ have worse outcomes
HR (95% CI) P HR (95% CI) P with AA and represent a population with an unmet medical need.
SLFN11 exp (yes vs no) 0.1 (0.03-0.7) 0.017 0.7 (0.1-4.4) 0.7
CRPC-NE vs CRPC 1.7 (0.5-6.6) 0.4 0.1 (0.02-0.6) 0.01
Visceral metastasis (yes vs no) 0.5 (0.1-2.4) 0.4 0.3 (0.04-1.9) 0.2
PSA (continuous) 1.0 (1.0-1.0) 0.1 1.0 (1.0-1.0) 0.05
Serum CgA (continuous) 1.0 (0.99-1.0) 0.7 1.0 (0.99-1.0) 0.8
AR alteration (alt) (yes vs no) 0.4 (0.1-1.7) 0.2 0.02 (0.01-0.2) 0.01
TP53 alt (yes vs no) 0.6 (0.1-3.8) 0.6 0.1 (0.1-0.8) 0.03
RB1 alt (yes vs no) 1.2 (0.2-7.9) 0.9 8.2 (0.7-100) 0.1
BRCA2/BRCA1/ATM alt (yes vs no) 0.7 (0.2-2.9) 0.7 0.3 (0.1-1.6) 0.2

5067 Poster Session (Board #179), Sat, 1:15 PM-4:15 PM 5068 Poster Session (Board #180), Sat, 1:15 PM-4:15 PM
Clinical outcomes and genomic analysis in patients with very high-risk WNT activating pathway mutations confer resistance to first-line antiandrogen
clinically localized prostate cancer treated by radical prostatectomy. First therapy in castration-resistant prostate cancer (CRPC). First Author: Pedro
Author: Pedro Isaacsson Velho, Sidney Kimmel Comprehensive Cancer Center Isaacsson Velho, Sidney Kimmel Comprehensive Cancer Center, Johns
at Johns Hopkins, Baltimore, MD Hopkins Hospital, Baltimore, MD
Background: Very high risk prostate cancer is associated with poor response to local Background: Wnt signaling is a cellular pathway responsible for embryogenesis
and systemic treatments, however few cases have been molecularly profiled. We and neoplasm. When activated, in about 10-20% of mCRPC, the WNT signaling
studied patients with primary Gleason pattern 5 disease (PG5; Gleason scores 5+5 or pathway may cause androgen-independent growth and consequently anti-
5+4) after undergoing radical prostatectomy (RP). Methods: Consecutive PG5 cases androgen resistance. To further characterize the clinical features and biology of
were retrospectively studied. Clinical-pathologic features, tumor somatic se- this subtype of mCRPC, we studied mutations in genes of this pathway, in-
quencing (UW Oncoplex), germline sequencing and PTEN, TP53 and ERG status by cluding APC, CTNNB1, RNF43, ZNRF3 and RSPO2, in patients who received
immunohistochemistry were correlated with biochemical relapse (BCR), metastasis-
treatment at Johns Hopkins Hospital (JHH). Methods: Patients with CRPC who
free survival (MFS), time to castration-resistance (TCR) and overall survival (OS).
received first-line antiandrogen (abiraterone or enzalutamide) therapy for CRPC
Kaplan–Meier curves were used in time-to-event data. Univariate and multivariable
(MVA) Cox proportional-hazards models (95% confidence intervals (CI Clinical-
at JHH were retrospectively studied. Pathological staging, extension of primary
pathological variables were selected for the multivariate analysis using stepwise disease, tumor somatic genomic data by Next Generation Sequencing (NGS)
method and were used to adjust the effect of each mutation. Results: Of 60 patients were correlated with clinical outcomes such as time to PSA progression, PSA
with available tissue and follow-up, 90% had non-organ confined disease, 58% had response and overall survival (OS). Cox regression was used to test associations
seminal vesicle invasion (SVI), 27% had lymph node involvement, and 15% had between variables and clinical outcomes and Kaplan-Meier method was used
ductal or intraductal histologic features. Overall, 43% developed metastasis with a for time-to-event data. Results: Of 137 pts who received first-line antiandrogen
TCR of 12 months. On multivariable analysis of clinical-pathologic variables, only therapy for CRPC, 10.9% (15 pts) had NGS with at least one activating WNT
ductal/intraductal histology (HR = 4.43; 95% CI: 1.76-11.15; p = 0.002) and SVI pathway alteration, including mutations in APC (6 pts), CTNNB1 (8 pts) and
(HR = 5.14; 95% CI: 1.83-14.47 p = 0.002) were associated with metastasis. Of RNF43 (3 pts). Patients with WNT mutations had fewer T3/T4 tumors (30.8%
patients with somatic sequencing data, 22% (11/49) had homologous re- vs. 51.4%, p = 0.24), but were balanced in other aspects. The median time to
combination (HR) and 12% (6/49) had mismatch repair (MMR) gene alterations. In PSA progression in WNT activated patients was 6.5 months vs. 9.6 months in
addition, 33% (16/49) had TP53 mutation and 51% (29/57) had PTEN loss. Only WNT negative patients (HR 2.3, 95% CI 1.3 - 4.1, p = 0.003). The presence of
TP53 mutation and PTEN loss were associated with metastasis on univariable WNT mutations (HR 2.3, 95% CI 1.2 - 4.3, p = 0.007) and previous che-
analysis and no genomic alterations remained significant in multivariable analyses. motherapy (HR 1.8, 95% CI 1.2 – 2.7, p = 0.003) were independently as-
Conclusions: Patients with PG5 have poor outcomes and more than one third have sociated with shorter time to PSA progression. PSA50 response was numerically
mutations in DNA repair genes. The unusually high incidence of actionable HR and
worse in WNT activated patients (53% vs 75%, p = 0.12). The OS in patients
MMR alterations provides the rationale for the design of adjuvant clinical trials in this
population.
with WNT mutations was 23.6 months vs 27.7 months in patients without WNT
mutations (HR 2.2, 95% CI 1.1 - 4.5, p = 0.01). Conclusions: Patients with
Type of Abnormality N Percentage WNT pathway mutations may have worse outcomes to first-line abi/enza
Homologous Recombination Pathway Mutation (n = 49) 11 22.4% compared to patients without these aberrations. Time to PSA progression
Mismatch Repair Deficiency (n = 49) 6 12.2% and OS were inferior in WNT activated population. Our data reinforce the
TP53 Mutation (n = 49) 16 32.7% necessity to develop effective WNT pathway inhibitors to test in clinical trials for
FOXA1 Mutation (n = 49) 6 12.2% WNT activated patients.
SPOP Mutation (n = 49) 4 8.2%
WNT pathway Mutation (n = 49) 4 8.2%
ERG Expression by IHC (n = 57) 22 38.6%
PTEN Loss by IHC (n = 57) 29 50.9%

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Genitourinary (Prostate) Cancer 305s

5069 Poster Session (Board #181), Sat, 1:15 PM-4:15 PM 5070 Poster Session (Board #182), Sat, 1:15 PM-4:15 PM
ATM loss in primary prostate cancer: Analysis of .1000 cases using a validated Cell-free DNA as a biomarker for taxane treatment in advanced prostate
clinical-grade immunohistochemistry (IHC) assay. First Author: Emmanuel S. cancer. First Author: Semini Sumanasuriya, Royal Marsden NHS Foundation
Antonarakis, Johns Hopkins Sidney Kimmel Comprehensive Cancer Center, Trust and The Institute of Cancer Research, London, United Kingdom
Baltimore, MD
Background: Antitumor efficacy of chemotherapeutic drugs docetaxel and
Background: ATM is a protein kinase acting as the main signal transducer of cabazitaxel against hormone-sensitive and castration-resistant prostate cancer
double-strand DNA break repair, in addition to mediating other cellular (CRPC) is well recognized. The identification and validation of minimally
functions. Germline and somatic pathogenic mutations in ATM occur in a invasive biomarkers of taxane response remain of paramount importance.
significant fraction of prostate cancers, and targeted therapies for ATM- Methods: Serial pre-, during and post-treatment plasma samples were col-
deficient tumors (e.g. ATR inhibitors) appear promising. Because DNA lected prospectively from two large Phase III clinical trials, FIRSTANA
sequencing assays frequently cannot distinguish mono-allelic from bi-allelic (NCT01308567) and PROSELICA (NCT01308580). Chemotherapy-naı̈ve
ATM alterations, a clinical-grade protein IHC assay for ATM loss is needed to patients (pts) were treated with docetaxel (75 mg/m2) or cabazitaxel (20 or
select men for these trials and better characterize ATM-deficient tumors. 25 mg/m2) in FIRSTANA, and pts previously treated with docetaxel received
Methods: We validated an automated dichotomously-scored IHC assay to cabazitaxel (20 or 25 mg/m2) as second-line chemotherapy in PROSELICA.
detect ATM protein loss in primary prostate cancer using prostate cancer cell Plasma cell-free DNA (cfDNA) was extracted, low-pass whole genome se-
lines with and without bi-allelic ATM inactivation and 49 high-grade (primary quencing (lp-WGS) libraries were prepared using the QIAGEN QiaSeq FX DNA
Gleason pattern 5) prostate tumors with known ATM genomic status. We then library kit, and samples sequenced on the Illumina NovaSeq 6000. Targeted
examined the frequency of ATM loss among 23 tumors with pathogenic next-generation sequencing (NGS) on a custom-made AmpliSeq panel of 30
germline ATM mutations, as well as . 1000 additional primary prostate genes, pre-selected for putative roles in taxane resistance, was also performed.
carcinomas using tissue microarrays (TMA). Results: ATM loss by IHC was lp-WGS copy number (CN) profiles were generated using ichorCNA (v0.1.0)
found in 14% (7/49) of primary Gleason pattern 5 (5+4 = 9 and 5+5 = 10) and targeted NGS variant calls derived using IonReporter software.
tumors. Of these, all cases with adequate tumor content and DNA yield had Results: Overall, lp-WGS data was generated from 265 samples (99 pts; 51
underlying pathogenic ATM mutations. Of the remaining 42 cases without treated on the FIRSTANA study, 48 treated on the PROSELICA study), ac-
ATM protein loss, none had ATM alterations. Among men with pathogenic quired at three time-points (pre-, during and at progression). Average cfDNA
germline ATM mutations, 74% (17/23) had ATM loss by IHC. Of these, 76% input was 10 ng and mean coverage achieved was ~2X (SD 1.2X). We observed
(13/17) had homogeneous loss of ATM protein in all tumor cells within a changes in lp-WGS tumor purity over time as a result of therapy; samples from
dominant tumor nodule, suggesting that ATM loss was an early clonal event. non-responding pts exhibited significantly higher tumor purity values post-
On TMA analysis, 90% (944/1044) of tumors were evaluable for ATM status treatment compared with samples from responding pts (p = 0.02). Targeted
by IHC. Among these, ATM loss was seen in 3.3% (31/944), and was NGS results were available from 294 pts (153 from FIRSTANA, 141 from
significantly more common in tumors with Gleason scores 9-10 (20/198; PROSELICA), and changes in tumor purity were also associated with treatment
10.1%) than in those of all other Gleason grades (11/746; 1.5%) (P, response (p = 0.04). CN frequency of key CRPC genes was similar to previously
0.0001). Conclusions: Validated ATM IHC is a sensitive assay for detecting reported datasets; several aberrant loci associated with response to taxane
underlying genomic ATM alterations. ATM protein loss appears to be an early therapy. Conclusions: cfDNA lp-WGS may have clinical utility in the man-
event occurring in the majority of tumors with underlying germline patho- agement of lethal prostate cancer. Funding: Sanofi. Clinical trial information:
genic ATM mutations, and is significantly enriched in high-grade prostate NCT01308580, NCT01308567.
cancers (especially Gleason grades 9-10).

5071 Poster Session (Board #183), Sat, 1:15 PM-4:15 PM 5072 Poster Session (Board #184), Sat, 1:15 PM-4:15 PM
Implications of the United States Preventive Services Task Force (USPSTF) Identification, incidence and clinical outcomes of patients (pts) with
recommendations on prostate cancer (PCa) stage migration. First Author: hypermutated prostate cancer (PC). First Author: Simon Yuen Fai Fu,
Iris Yeong- Fung Sheng, Cleveland Clinic, Cleveland, OH Auckland City Hospital, Auckland, New Zealand
Background: Prostate specific antigen (PSA) screening has been controversial, Background: A small proportion of metastatic PC exhibit outlier somatic
given unrefined screening guidelines leading to overdiagnosis and overtreatment mutation (mut) rates exceeding the average of 4.4 mut/Mb. The incidence,
of “indolent” PCa. In 2008, the USPSTF recommended against PSA screening clinical course and treatment response of pts with hypermutation (HM) is
for men aged $75 and in 2012 broadened this recommendation to include all poorly characterised. Methods: We performed targeted sequencing from a
men. The impact of these changes is unstudied. We hypothesize that these panel of PC genes using plasma cell-free DNA samples collected from
screening changes could delay the diagnosis of advanced PCa. Methods: The metastatic castration-resistant prostate cancer (mCRPC) pts and calculated
Surveillance, Epidemiology and End Results Program (SEER) was used to somatic mutation burden. HM samples were additionally subjected to whole
identify men (age 55-69) diagnosed with PCa between 2004-2015. PCa stage exome sequencing to determine trinucleotide mutational signatures and
was categorized as nodal (N1M0) and metastatic (NxM1). Trend analysis was microsatellite instability (MSI). Clinical data was retrospectively collected
stratified based on year 2004-2008 (group 1), 2009-2012 (group 2), and 2012- and compared to a control cohort of 199 mCRPC pts. Results: 671 samples
2015 (group 3). Using group 2 as a reference, multivariable logistic regression from 434 pts had ctDNA . 2% and were evaluable. 32 samples from 24 pts
was used to identify predictors for N1M0 and NxM1 in each group. Results: From had . 11 mut/Mb and fell above the 95th percentile for mutation burden
2004-2015, there were 603,323 eligible men diagnosed with PCa (group 1: with a median mutation burden of 34 mut/Mb. 11 pts had deleterious
262,240 men, group 2: 210,045 men, group 3: 131,038 men). In group 1, mutations or homozygous deletions in mismatch repair (MMR) genes and 4
1.4% had N1M0 and 2.8% had NxM1. In group 2, 1.6% had N1M0 and 3.7% further pts had evidence of MMR deficiency (MMRd) from mutational sig-
had NxM1. In group 3, 1.4% had N1M0, and 6.1% had NxM1. The adjusted natures and MSI status. The remaining 9 pts had either BRCA2 mutations
odds ratio (AOR) of N1M0 was 0.78 (95%CI 0.74-0.82; p,0.0001) in group 1 (n = 4), Kataegis (localized hypermutation, n = 3), or undefined causes for
and 1.71 (95%CI 1.63-1.80; p,0.0001) in group 3. Similar AOR trends were HM (n = 2). The incidence of MMRd was 3.5% (15/434), and germline
seen in NxM1 (group 1, 0.71; 95%CI 0.68-0.73, p, 0.0001 vs. group 3, 1.70;
MMRd was 0.2% (1/434). For MMRd pts with available clinical data (10/15)
95% CI 1.63-1.75, p,0.0001). (Table) Subset analysis of non-eligible pa-
at diagnosis, the median age was 73.6 y, 70% had Gleason score $8, and
tients (age .70 and ,55) showed a similar stage migration. Conclusions: With
50% presented with M1 disease. Comparing the MMRd with the control
each USPSTF recommendation, there have been significantly more diagnoses of
cohort, median time from ADT to CRPC was 9.1 m (95% CI 6.9–11.4) vs.
advanced PCa; suggesting stage migration. The sequelae of having advanced
PCa include more aggressive treatments, increased financial burden, and re-
18.2 m (95% CI 15.1–21.3), p = 0.001; median time from CRPC to death
duced quality of life. Future population studies are warranted to investigate was 13.1 m (95% CI 0.3–25.9) vs. 40.1 m (95% CI 32.4–47.8), p , 0.001.
whether the updated 2018 USPSTF recommendation now encapsulates the best Conclusions: HM and MMRd can be identified using liquid biopsy and could
target population. help to select pts for immunotherapy.
Regression results for N1M0 and NxM1.
N1M0 NxM1
AOR* (95% CI) P-value AOR* (95% CI) P-value
Year of diagnosis
2004-2008 0.78 (0.74-0.82) ,0.0001 0.71 (0.68-0.73) ,0.0001
2009-2012 Ref Ref
2013-2015 1.71 (1.63-1.80) ,0.0001 1.70 (1.63-1.75) ,0.0001
*Adjusted for age, race, income, and education.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
306s Genitourinary (Prostate) Cancer

5073 Poster Session (Board #185), Sat, 1:15 PM-4:15 PM 5074 Poster Session (Board #186), Sat, 1:15 PM-4:15 PM
Multi-gene hereditary cancer testing, family history and prognosis in men The rate of tumor growth, g, as a biomarker for overall survival (OS) in prostate
with prostate cancer. First Author: Panagiotis J. Vlachostergios, Division of cancer (PC) in clinical trials as well as in real-world data from the Veterans
Hematology & Medical Oncology, Weill Cornell Medical College & New York- Administration Medical Centers (VAMCs). First Author: Harshraj Leuva,
Presbyterian Hospital, New York, NY James J Peters VAMC, Bronx, NY
Background: Multi-gene hereditary cancer testing is common in prostate Background: Novel assessments of efficacy are needed to improve determi-
cancer (PC). We assessed the frequency of pathogenic mutations (mt) and nation of treatment outcomes in clinical trials and in real-world settings.
examined associations with family history (FH), cancer recurrence, and Methods: Cancer treatments usually lead to concurrent regression and growth
overall survival (OS). Methods: Men with clinically localized or metastatic PC of the drug-sensitive and drug-resistant fractions of a tumor, respectively. We
consented to germline DNA testing using a validated panel of 30 genes have exploited novel methods of analysis that assess these two simultaneous
associated with elevated risk for common cancers (Color Genomics). Chi- processes and have estimated rates of tumor growth (g) and regression (d) in
square test and Fisher’s exact test were used to compare pathogenic mt and over 30,000 patients (pts) with diverse tumors. Results: In prostate cancer
clinical characteristics. The Kaplan-Meier method was used to evaluate the (PC) we have analyzed both clinical trial and real-world data from Veterans.
associations of mt status with PSA recurrence and OS. Results: 315 men Using clinical trial data from 6819 pts enrolled in 15 treatment arms we have
(median age 69, range 38-89) were included; 140 (44.4%) with localized established separately and by combining all the data that g correlates highly
and 175 (55.6%) with metastatic PC. 215 (68.3%) reported a FH of any (p,0.0001) with overall survival (OS) – slower g associated with better OS. In
cancer in one and 101 (32.1%) in $ 2 first-degree relatives. Genomic PC, abiraterone (ABI) and docetaxel (DOC) are superior to placebo, prednisone
testing detected 12.1% of pts possessed a pathogenic mt (39 mt in 38 pts): and mitoxantrone. ABI (median g =0.0017) is superior to DOC (g=0.0021) in
BRCA2 (n = 12, 3.8%), CHEK2 (n = 10, 3.2%), APC (n = 7, 2.2%), MUTYH first line (p=0.0013); and ABI in 2nd line (g=0.0034) is inferior to ABI in 1st
(n = 3, 0.9%), ATM (n = 2, 0.6%), and 1 each (0.3%) with NBN, PALB2, line (g=0.0017; p,0.0001). Finally, using combined clinical trial data as a
RAD51D, MSH2, and PMS2. Variants of uncertain significance were de- benchmark we could assess the efficacy of novel therapies in as few as 30-40
tected in 52 (16.5%). Men with FH of prostate, breast, ovarian, pancreatic patients. Amongst 7457 Veterans, the median g on a taxane (g=0.0022) was
cancer or/and melanoma (182/315, 58%), were more likely to have BRCA2 similar to that from clinical trials (g=0.0012). Although only 258 Veterans
pathogenic mt (P = 0.049). Median age of diagnosis was 55 with BRCA2 received cabazitaxel (CAB), g values for CAB (g=0.0018) and DOC (g=0.0023)
versus 62 in non-carriers (P = 0.01). No association was found between FH were indistinguishable (p=0.3) consistent with their identical mechanism of
of Lynch-associated cancers (colon, endometrial, gastric) and mismatch action. Finally, outcomes with DOC in African American (AA) (g=0.00212) and
repair gene variants. The prevalence of pathogenic mt was significantly Caucasian (g=0.00205) Veterans were indistinguishable (p=0.9) and com-
higher with visceral metastases (P = 0.001) but did not differ between parable across all VAMCs. Conclusions: The rate of tumor growth, g, is an
localized and advanced PC pts (P = 0.602). In men with clinically localized excellent biomarker for OS both in clinical trials and in real-world settings. g
PC at diagnosis (n = 226, 71.7%), pathogenic mt were associated with a allows comparisons between trials and for large trial data sets to be used as
shorter time to recurrence (30 vs 53 months, P = 0.004). Men with advanced benchmarks of efficacy. Real-world outcomes in the VAMCs are similar to those
PC (n = 175) and germline APC or MUTYH mt had worse median OS (44 vs in clinical trials. In the egalitarian VAMCs DOC efficacy in PC is comparable in
318 months, P = 0.004). Conclusions: Our findings confirm the incidence of AA and Caucasian Veterans -- indicating inferior outcomes reported in AAs are
men with non-PC FH associated with DNA repair gene mt and support testing likely due to differential health care access, not differences in biology.
in early stage pts. Further, our data support the prognostic significance of
germline genetic alterations which deserves study in a larger cohort of pts.

5075 Poster Session (Board #187), Sat, 1:15 PM-4:15 PM 5076 Poster Session (Board #188), Sat, 1:15 PM-4:15 PM
Examination of the additive value of CTC biomarkers of heterogeneity (Het) Clinical and safety outcomes of TALAPRO-2: A two-part phase III study of
and chromosomal instability to nuclear-localized (nl) AR-V7+ CTCs in talazoparib (TALA) in combination with enzalutamide (ENZA) in metastatic
prediction of poor outcomes to androgen receptor signaling inhibitor (ARSi) castration-resistant prostate cancer (mCRPC). First Author: Neeraj Agarwal,
in metastatic castration resistant prostate cancer (mCRPC). First Author: University of Utah Huntsman Cancer Institute, Salt Lake City, UT
Howard I. Scher, Memorial Sloan Kettering Cancer Center, New York, NY
Background: TALA is a dual-mechanism PARP inhibitor that inhibits PARP
Background: Prediction of ARSi benefit in mCRPC is an unmet medical need. Recently, the catalytic activity and traps PARP on DNA. ENZA is a novel hormonal therapy
Epic Sciences CTC based nl AR-V7 test validated as a predictive biomarker in two multi- approved to treat castration resistant prostate cancer. TALA + ENZA may im-
center validation studies and has received Medicare coverage for use in mCRPC. While the nl prove clinical outcomes for men with mCRPC. However, TALA, is a substrate for
AR-V7 biomarker is highly specific to resistance and predictive of improved response with
taxane Rx, it is a measure of just one mechanism of resistance to ARSis. CTC Het measured
efflux drug transporters P-gp and BCRP. Prior to the initiation of TALAPRO-2
by the Shannon Index and CTC chromosomal instability measured by predicted number of part 1, the in vivo effect of ENZA on exposure of P-gp and BCRP substrates, such
Large Scale Transitions (pLST) have both been associated with poor OS to ARSis in previous as TALA, had not been evaluated. Methods: TALAPRO-2 part 1 was designed to
analysis. Here we investigate the relationship of Het and pLST to nlAR-V7 in order to assess determine TALA starting dose based on safety and pharmacokinetics (PK)
multi-clonal resistance and determine if these biomarkers can provide added sensitivity in the evaluation of TALA + ENZA. Pts were $18 yrs of age, had ECOG PS #1, with no
nlAR-V7 negative patient population. Methods: 275 blood samples from 2nd+ line mCRPC prior systemic treatment for mCRPC. The starting dose of TALA in the first 13 pts
patients prior to treatment with ARSi (n=148) or taxanes (n=137) were obtained between was 1 mg once daily (QD) + ENZA 160 mg QD (1 mg QD cohort). Based on safety
2012 and 2017 from 3 clinical centers. Detectable CTCs in each blood sample were assayed
review of prespecified target safety events and PK data, TALA dose was reduced
for nlAR-V7, Het, and pLST using the Epic Sciences platform. Biomarkers were analyzed in
context of each other and outcomes including clinical co-variates. Results: 94% of samples to 0.5 mg QD; additional pts were treated with a starting dose of TALA 0.5 mg
had detectable CTCs, 84% were evaluable for Het analysis (. 2 CTCs), and 76% were QD + ENZA 160 mg QD (0.5 mg QD cohort). Results: 19 pts were enrolled in
evaluable for pLST (. 3 CTCs). Conclusions: Addition of CTC Het (Shannon Index) and CTC part 1 (1 mg QD cohort, 13; 0.5 mg QD cohort, 6). The median (range) age was
chromosomal instability (pLST) biomarkers to nlAR-V7 identifies an additional 15% of 71 yrs (52-82). As of the analysis cutoff date, the median treatment duration
mCRPC pts (38% of total) that are predicted to have poor survival to AR signaling inhibitors. was 25 and 11 wks for the 1 mg QD and 0.5 mg QD cohorts, respectively.
Prevalence/co-occurrence of biomarkers, Pre 2nd+ Line (N = 275). Treatment-emergent adverse events (TEAEs) occurred in 19 pts. The most
common TEAE, anemia, occurred in 76.9% and 33.3% of pts in the 1 mg QD
All (275) AR-V7+ (62) AR-V7- (213) and 0.5 mg QD cohorts, respectively. TEAEs that led to TALA dose reduction
AR-V7+ 62 (23%) 62 (100%) 0 (0%) occurred in 6 pts (46.2%) and 0 pts in the 1 mg QD and 0.5 mg QD cohorts,
pLST+ 76 (28%) 43 (69%) 33 (16%)
Het+ 82 (30%) 35 (57% ) 47 (22%) respectively. In the 1 mg QD cohort, target safety events were reported for 7 pts
pLST or Het+ 104 (38%) 45 (73%) 59 (28%) (53.8%) vs 0 in the 0.5 mg QD cohort. 92% and 100% of pts had a 50%
decline from baseline in PSA in the 1 mg QD and 0.5 mg cohorts, respectively,
demonstrating preliminary anti-tumor activity. PK data showed that ENZA
Prediction of ARSi resistance by single and combined use of CTC nlAR-V7, Heterogeneity, and pLST prior to increased TALA exposure and that TALA 0.5 mg QD + ENZA maintained similar
therapy initiation, Pre 2nd+ Line (N = 148).
TALA exposure to that achieved with 1 mg QD monotherapy. Conclusions: TALA
Univariate HR (OS) p-value Multivariate HR (OS) p-value 0.5 mg QD + ENZA 160 mg QD had a manageable safety profile in pts with
V7+ 2.7 ,0.001 2.7 ,0.001 mCRPC and will be the starting dose for the randomized portion of TALAPRO-2.
pLST+ 3.3 ,0.001 2.5 ,0.001 Clinical trial information: NCT03395197.
Het+ 2.1* ,0.001 1.35** 0.021
pLST+ in V7- 2.8 ,0.001 2.3 0.008
Het+ in V7- 2.3* ,0.001 1.2** 0.15

*SI . 1.5, **continuous

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Genitourinary (Prostate) Cancer 307s

5077 Poster Session (Board #189), Sat, 1:15 PM-4:15 PM 5078 Poster Session (Board #190), Sat, 1:15 PM-4:15 PM
Assessment of 2,000 patients presenting to a multidisciplinary prostate Randomized phase II trial of presurgical androgen deprivation therapy (ADT)
cancer clinic in the United Kingdom. First Author: Pandora Rudd, St. with or without axitinib in prostate cancer (PCa) presenting with lymph node
Bartholomew’s Hospital, London, United Kingdom (LN) metastasis. First Author: Amado J. Zurita, The University of Texas MD
Anderson Cancer Center, Houston, TX
Background: Multidisciplinary clinics (MDCs) involving both oncologists and
urologists are recommended for managing radical prostate cancer patients. Background: Strategies integrating androgen targeted and locoregional thera-
The effectiveness of MDCs in arriving at best treatment decisions is unknown. pies are being increasingly used in PCa with regional spread, but are rarely
We analysed patient characteristics and management decisions over 8 years curative. Angiogenesis inhibitors delay progression in castration resistant PCa
in a MDC at Bart’s Hospital, London. Methods: Clinical data were collected in and may synergize with ADT through endothelial cell apoptosis in hormone naı̈ve
real time and analysed retrospectively, including demographics, tumour stage patients (pts). We hypothesized that the frontline combination of ADT with the
and grade, D’amico risk group, treatment choice and first clinician seen. We potent VEGF inhibitor Axi would improve PCa control relative to ADT alone and
compared variables in 1000 consecutive patients presenting between 2011- allow for meaningful time off systemic therapy after surgical consolidation.
2015 (cohort A) to 1000 patients presenting 2016-18 (cohort B) to investigate Methods: Pts with either clinically detected LN+ (TxN1M0 or TxNxM1a) or very
trends over time. Results: 2000 patients were included, age 65.2 6 8.6 years high risk for LN PCa were treated with ADT for 2 mos and then randomized 2:1 to
and 65.9 6 9.1 years (p=0.08), with presenting PSA 9.0 (6.3-14.4) and 9.2 respectively add open label Axi (5 mg PO bid) vs. continue ADT alone for 4 mos
(6.4-15.0) ng/ml (p=0.36), in cohort A and B respectively. Disease severity until surgery. Those responding with PSA #5 ng/mL were offered prostatectomy
and initial treatment decision are shown in the table. In low risk disease, 126 and extended pelvic lymphadenectomy (RP). ADT +/- Axi was withheld post-
(75%) patients had active surveillance in cohort A, and 158 (90%) in cohort B operatively and PSA measured q3 mos until 1 y. Primary objective: proportion of
(p=0.0003). In high risk disease, 202 (59%) patients had radiotherapy pts progression free (FFP) 12 mos after RP, defined as PSA #1.0 ng/mL and no
compared to 194 (50%) in cohort B (p=0.011). In cohort B, 127 (39%) radiation or ADT, aiming to detect a 35% difference favoring ADT+Axi.
patients seeing oncology first had radiotherapy compared to 143 (25%) pa- Results: 72 pts completed accrual. We report on the 54 pts with LN+ disease:
tients who saw urology first (p,0.0001). 76 (23%) and 154 (27%) patients median age 62 y (range 42-76), pretreatment PSA 22.9 ng/mL (range 3.6 -
had surgery, that saw oncology and urology first, respectively (p=0.11). 404.4), 38 N1 / 16 M1a. Table shows presurgical therapy outcomes. Path
Conclusions: In 2000 patients presenting to a prostate MDC over 8 years, responses in the prostate were similar between arms, but in 5 ADT+Axi vs. 0 ADT
active surveillance in low risk disease increased, radiotherapy in high risk LN+ pts there was no residual nodal disease. Testosterone recovery: 24/26
disease reduced, and the proportion undergoing surgery was unchanged. The ADT+Axi and 9/9 ADT RP pts by 6 mos. 21/36 ADT+Axi and 15/18 ADT pts have
initial clinician seen influenced treatment choice; having both specialists in failed; 1 y FFP estimates 48.0% (SE 8.6%) and 16.7% (SE 8.3%), respectively
the same consultation may improve consistency of treatment decisions. (p = 0.02). 1 y undetectable PSA 9 pts (6 ADT+Axi). No grade 4 toxicities or
Disease severity and treatment choice before and after 2016. unexpected side-effects were observed. Conclusions: 1 year after RP, ADT+Axi
Cohort A (2011-2015) Cohort B (2016-2018)
resulted in proportionally greater number of LN+ PCa pts off treatment and
N=1000 N=1000 p-value progression free than ADT alone. Tissue analysis is evaluating predictors of
Disease at presentation Low 167 (17) 175 (18) 0.68 benefit to further develop angiogenesis inhibition as part of combination strat-
(n; %) Intermediate 442 (44) 349 (35) <0.0001 egies for hormone naı̈ve PCa. Clinical trial information: NCT01409200.
High 341 (34) 390 (39) 0.03
Nodal 18 (2) 44 (4) 0.001 ADT+Axitinib ADT
Metastatic 32 (3) 42 (4) 0.29 n = 36 n = 18
Treatment Radiotherapy 360 (36) 298 (30) 0.004 RP on protocol 26 (72.2%) 9 (50.0%)
Active Surveillance 281 (28) 295 (29) 0.52 PSA > 5 RP off protocol 6 (16.7%) 1 (5.5%)
Surgery 259 (26) 260 (26) 0.99 Clinical progression 3 (8.3%) 2 (11.1%)
Other 100 (10) 147 (15) <0.0001 Withdrew early 1 (2.8%) 5 (27.8%)
Palliative cystoRP 0 (0%) 1 (5.6%)

5079 Poster Session (Board #191), Sat, 1:15 PM-4:15 PM 5080 Poster Session (Board #192), Sat, 1:15 PM-4:15 PM
CALGB 90203 (Alliance): Radical prostatectomy (RP) with or without neo- Evaluation of the patient-reported outcomes common terminology criteria for
adjuvant chemohormonal therapy (CHT) in men with clinically localized, adverse events (PRO-CTCAE) with abiraterone acetate plus prednisone
high-risk prostate cancer (CLHRPC). First Author: James Andrew Eastham, (AAP), degarelix (D), or the combination in men with biochemically recurrent
Memorial Sloan Kettering Cancer Center, New York, NY prostate cancer (BCRPC). First Author: Karen A. Autio, Memorial Sloan
Background: Neoadjuvant CHT followed by RP did not increase 3-year biochemical
Kettering Cancer Center, New York, NY
progression free-survival (bPFS) compared to RP alone in men with CLHRPC. However, Background: Patient-reported symptoms using the PRO-CTCAE provide insights
there is evidence that bPFS and overall survival over time was improved. In the current into the patient experience with care. Earlier use of AAP (an androgen bio-
analysis we assessed whether CHT followed by RP improved pathological specimen synthesis inhibitor plus prednisone) with androgen deprivation therapy in cas-
features compared to RP alone. Methods: CALGB 90203 (Alliance) is a Phase III study
which randomly assigned, in a 1:1 fashion, men with CLHRPC [biopsy Gleason Grade
tration sensitive disease may lead to increased symptoms. We previously
Group (GGG) 4 or 5 or Kattan pre-op nomogram bPFS , 60%] to RP alone or RP plus reported a randomized phase 2 trial of intermittent AAP, D, or AAP+D in BCRPC
neoadjuvant CHT [androgen deprivation plus docetaxel (75 mg/m2 every 3 weeks for 6 (NCT01751451) and now share the PRO-CTCAE results. Methods: Men were
cycles)]. We conducted an exploratory analysis comparing histologic findings, de- randomized 1:1:1 to AAP, D, or AAP+D for 8 months, then entered follow up with
termined at the treating center, in the RP specimens of men receiving CHT plus RP and PSA, testosterone, and safety monitoring. PRO-CTCAE was elicited from patients
men treated with RP alone. We used the Chi-square test, with P-values adjusted by the monthly for hot flashes (HF), fatigue, arthralgias, myalgias, anxiety, depression,
Holm method for multiple comparisons. Results: A total of 788 men (median age, 62; sexual function, plus overall QOL. Changes from baseline to end of treatment
range: 32-83 years) were randomized, with 738 ultimately undergoing RP. There was no were compared between groups. AUCs were calculated for each item as a
difference in pathologic GGG (Table). Men treated with neoadjuvant CHT had a lower measure of symptom severity over time. Results: 110 men were included.
pathologic T-stage and lower likelihood of having seminal vesicle invasion (SVI), positive
Compliance with PRO-CTCAE reporting from baseline to EOT was 93%. HF did
pelvic lymph nodes, or positive surgical margins (SM) (Table). Conclusions: Most
pathologic features in the RP specimen were improved in men receiving neoadjuvant
not differ between AAP+D and D, but were increased relative to AAP (all p ,
CHT compared to RP alone. The relationship between pathologic changes and the 0.05). These differences were consistent when HF were measured as an AUC (all
development of metastasis and survival require further analysis. RP pathologic out- p , 0.01). Fatigue severity did not differ between groups however men receiving
comes. Summary statistics are calculated for the number of patients with non-missing AAP reported a small worsening in activity interference from fatigue as compared
data for each characteristic. Clinical trial information: NCT00430183. to AAP+D (p , 0.05). Overall QOL scores were high and did not differ with
Neoadjuvant CHT + RP RP alone
AAP+D relative to AAP or D. Conclusions: Collection of PRO-CTCAE was feasible
Number (%) Number (%) Adjusted P-value and did not demonstrate differences in fatigue, HF, or QOL between AAP+D and
GGG 0.10 D. Comparisons of PRO-CTCAE to matched clinician-reported AEs, and changes
1 5 (2) 4 (1)
2 36 (15) 55 (16) in PRO-CTCAE with testosterone recovery during follow up are planned. Clinical
3
4
47 (19)
41 (17)
98 (28)
45 (13)
trial information: NCT01751451.
5 114 (47) 148 (42)
Pathologic T-stage ,0.001
T1 / T2 145 (41) 83 (23)
T3 211 (59) 274 (75)
T4 2 (1) 9 (2)
SVI 0.05
Yes 116 (32) 151 (41)
No 249 (68) 215 (59)
Pathologic Nodal-stage 0.05
N0 280 (80) 250 (70)
N1 68 (19) 97 (27)
NX 4 (1) 9 (3)
SM- Overall , 0.001
Positive 56 (18) 126 (45)
Negative 247 (82) 155 (55)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
308s Genitourinary (Prostate) Cancer

5081 Poster Session (Board #193), Sat, 1:15 PM-4:15 PM 5082 Poster Session (Board #194), Sat, 1:15 PM-4:15 PM
Targeting backdoor androgen synthesis through AKR1C3 inhibition: A Contemporary trends in treatment for low-risk prostate cancer. First Author:
presurgical hormonal ablative trial in high risk localized prostate cancer Jason Frankel, Virginia Mason Medical Center, Seattle, WA
(PC). First Author: Laura Graham, University of Washington, Seattle, WA
Background: Active surveillance (AS) is the preferred management of low risk
Background: Studies have shown that localized PCs may resist neoadjuvant (LR) prostate cancer (PCa) to decrease morbidity associated with over-
androgen receptor (AR)-targeted therapies as a result of persistent intra- treatment. We hypothesized management trends would show AS increasing
prostatic androgens, likely arising through upregulation of steroidogenic and initial definitive treatment decreasing for men with LR PCa. This study
enzymes. Therefore, we sought to evaluate clinical effects of combinatorial describes contemporary trends in initial treatment for LR PCa using the Na-
AR-targeted therapy, including indomethacin (Indo) to inhibit the ste- tional Cancer Database (NCDB), which includes ~50% of new PCa diagnoses
roidogenic enzyme AKR1C3, in men with high risk PC undergoing radical in the United States per year. Methods: NCDB was queried for all patients with
prostatectomy (RP). Methods: This was an open label, single-site, Phase II LR localized PCa (2010-2015), defined as non-metastatic PCa with PSA,10
neoadjuvant trial in men with localized high to very-high risk PC, as defined ng/mL, Gleason #6 and clinical T1-T2a primary tumor. Staging was assigned
by NCCN criteria. Patients received 12 weeks of neoadjuvant apalutamide using 2018 American Joint Committee on Cancer guidelines. Patients refusing
(Apa), abiraterone (Abi) plus prednisone, degarelix, and Indo at their re- treatment, unable to be treated due to comorbidity or with unknown treatment
spective FDA-approved doses followed by RP. The primary objective was to status were excluded. Men were grouped based on NCDB codes for initial
determine the pathologic complete response (pCR) rate. Secondary objec- treatment including prostatectomy (PR), radiation (RT), AS and no treatment.
tives included assessing for minimal residual disease (MRD) (i.e. #0.25 cm3 Those receiving initial definitive treatment were compared to those who were
tumor volume corrected for cellularity), measuring intraprostatic androgens not using logistic regression. Multivariable analysis was conducted for factors
and assessing molecular features associated with drug resistance. Twenty possibly correlating with management choice. Results: Of the 645,932 pa-
evaluable patients provided 91% power (one-sided alpha = 7.5%) to detect a tients identified, 134,540 met inclusion criteria. Of these patients, 114,183
difference in pCR rate of 5% (H0) vs. 25% (H1). Results: Twenty-two pa- (84.9%) underwent PR or RT compared to 20,357 (15.1%) that did not. When
tients enrolled and 20 were evaluable for the primary endpoint (1 patient examined over time, 92.4% of patients underwent initial PR or RT in 2010 vs.
came off to pursue stereotactic radiosurgery; 1 was removed after developing 73.9% in 2015 (p,0.001). 90% of patients seen at community cancer
grade 2 transaminitis). At baseline, the median PSA was 10.1 ng/mL (4.4- centers underwent definitive initial treatment compared to 78% of those seen
159.4), 4 (20%) patients had Gleason grade group (GG) 4 disease and 16 at academic centers (p,0.001). Conclusions: In this national, hospital-based
had GG 5 disease. At RP, 1 (5%) patient had a pCR, 6 (30%) had MRD, 18 cohort overall rates of definitive initial treatment for LR PCa remain high,
(90%) had ypT3 disease and 7 (35%) had lymph node (LN) metastases. though seem to be decreasing over time. Overtreatment of LR PCa remains a
Treatment was generally well tolerated and adverse events were consistent concern and illustrates the relatively slow incorporation of guideline recom-
with each individual drug’s known safety profile. Additional follow up data mendations into clinical practice at the national level.
and correlative work will be presented at the meeting. Conclusions: In our 2010 2011 2012 2013 2014 2015
cohort of men with high-risk PC, pCR rates remained low even with com- (N=28,336) (N=29,051) (N=21,568) (N=20,044) (N=17,839) (N=17,1702) p-value for trend
binatorial AR-directed therapy. Ongoing pharmacodynamic studies aimed at Initial PR or RT 26,189 26,078 18,573 16,303 13,966 13,074 ,0.001
(92.42%) (89.77%) (86.11%) (81.34%) (78.29%) (73.86%)
determining if Indo effectively inhibited AKR1C3 will provide important AS or No Treatment 2,147 2,973 2,995 3,741 3,873 4,628
insights regarding the utility of targeting this steroidogenic enzyme. Clinical (7.58%) (10.23%) (13.89%) (18.66%) (21.71%) (26.14%)
trial information: NCT02849990.

5083 Poster Session (Board #195), Sat, 1:15 PM-4:15 PM 5084 Poster Session (Board #196), Sat, 1:15 PM-4:15 PM
Human prostate cancer immune phenotypes after androgen deprivation IMRT pelvic radiotherapy with simultaneous integrated boost in high-risk
therapy. First Author: Matthew Dallos, Columbia University, New York, NY prostate cancer: Results after 10 years. First Author: Christian Ekanger,
Kreftavdelingen, Haukeland Universitetssjukehus, Bergen, Norway, Bergen,
Background: The prostate tumor microenvironment (TME) is generally non-
Norway
inflamed. However, we previously showed in pre-clinical models that an-
drogen deprivation therapy (ADT) induces a complex immune cell infiltrate. Background: To report 10 year results after image guided intensity-modulated
Whether ADT similarly promotes inflammation in patients remains unclear. radiotherapy (IMRT) with hypofractionated simultaneous integrated boost
Therefore, we collected specimens from patients with localized prostate (SIB) in high-risk prostate cancer. Methods: Between 2007 and 2009, 97
cancer treated with neoadjuvant degarelix and hypothesized that we could patients with an estimated risk of lymph node metastases above 15% (Roach
decipher key changes in the immune TME after ADT utilizing a novel cancer equation) were prospectively included in a phase II study. Patients were treated
systems biology approach. Methods: We identified a cohort of patients with 2-2.7 Gy to the prostate, vesicula seminalis and elective pelvic field in 25
treated with neoadjuvant degarelix (240mg SQ) as part of two clinical trials fractions over 5 weeks with androgen deprivation therapy for 2 years. Toxicity
(NCT01696877, NCT01542021). Patients were treated with degarelix was scored according to RTOG criteria and biochemical free survival (BFS)
either 4 days (N = 13), 7 days (N = 17) or 14 days (N = 8) prior to radical using the Phoenix definition. Patients were divided into three groups; very high-
prostatectomy. RNA was extracted from FFPE tissue and analyzed by risk patients (VHR) according to NCCN 2015 criteria (n=50), high-risk patients
RNAseq. To deconvolute fractional contributions of different immune cell (HR) (n=32), and patients with N+ disease and/or pretreatment s-PSA $100
subsets we performed CIBERSORT and Uniform Manifold Approximation (n=15). Differences were examined using Kaplan Meier estimates with log rank
and Projection (UMAP) analysis. Treatment groups were compared to a test. Results: Ten year BFS in the entire cohort was 63%. Metastasis-free
cohort of untreated matched controls (N = 37). Results: Degarelix induced a survival (MFS) was 77% and prostate-cancer-specific-survival (PCSS) 88%.
complex immune cell infiltrate in human primary prostate tumors with an Overall survival (OS) was 69% and local failure rate was 11%. VHR vs. HR
increase in both pro- and anti-inflammatory cells subsets and changes in subgroups had significant different BFS; 58% vs 84% (p=0.01) respectively.
expression of immune checkpoints compared to untreated matched con- MFS and PCSS in the VHR group compared to the HR group was 78% vs 91%
trols. Degarelix therapy also significant changed associated gene signatures (p=0.108) and 86% vs 97% (p=0.157) respectively. Patients with N+ and/or
within the lymphoid and myeloid compartments over time. Conclusions: ADT PSA.100 had worse outcome compared to the HR/VHR groups, but not all
leads to profound immune remodeling within the prostate TME. Our analysis had treatment failure. BFS was 33% vs 68% (p=0.001), MFS 47% vs 83%
of human primary prostate tumors supports the hypothesis that the optimal (p=0.000) and PCSS 73 % vs 90% (p=0.04), respectively. Patients who
time for immunologic intervention is the peri-castration period. These data reached a PSA nadir value below 0.1 (n=80) had significant better outcomes,
also suggest that combinatorial immunotherapy strategies that target par- with PCSS 93% vs 65% (p= 0.001) and BFS 74% vs 12% (p=0.000), re-
ticular immune cell subsets will likely be required to successfully promote spectively. Acute gr 2 GI and GU toxicity was observed in 27% and 40%, gr 3
robust anti-tumor immune responses in prostate cancer. GI and GU toxicity in 1% and 3%. Late gr 2 GI and GU toxicity at 3 years
appeared in 3% and 4% with no gr 3 toxicity. Conclusions: High-risk prostate
cancer patients treated with IMRT with SIB obtained favorable outcomes with
few serious side effects. There were significant better results in the HR versus
the VHR group, both better than the N+/PSA$100 group.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Genitourinary (Prostate) Cancer 309s

5085 Poster Session (Board #197), Sat, 1:15 PM-4:15 PM TPS5086 Poster Session (Board #198a), Sat, 1:15 PM-4:15 PM
Is there a role for testosterone replacement therapy in reducing biochemical Pamiparib, an investigational PARP inhibitor, in patients with metastatic
recurrence following radical prostatectomy? First Author: T Edward A, castration-resistant prostate cancer (mCRPC) and a circulating tumor cell
University of California, Irvine, Orange, CA (CTC) homologous recombination deficiency (HRD) phenotype or BRCA
defects: A trial in progress. First Author: Simon Chowdhury, Guy’s, King’s
Background: Historically, the use of testosterone replacement therapy (TRT)
and St. Thomas’ Hospitals, and Sarah Cannon Research Institute, London,
has not been recommended in men with a history of prostate cancer (PC).
United Kingdom
However, low testosterone levels are significantly associated with metabolic
complications, decreased sexual function, and (more recently) high-grade Background: Men with mCRPC who have a BRCA1/2 mutation (BRCA1/2mut)
PC. In 2009, in hopes of improving sexual function outcomes in men fol- or mutations in other genes resulting in HRD have a poor prognosis. A novel
lowing radical prostatectomy (RP), we began treating low-risk patients with liquid biopsy test (EPIC Sciences) identifies CTCs with an HRD phenotype.
TRT. The current study examines the impact of TRT on biochemical re- Preliminary studies showed that these men may respond to treatment with a
currence (BCR). Methods: Between December 2009 and June 2018, a PARP inhibitor. Pamiparib, an investigational PARP1/2 inhibitor, has shown
cohort of 850 patients underwent RP by a single surgeon. 152 (18%) men brain penetration and potent PARP–DNA complex trapping in nonclinical
were postoperatively placed on TRT for recovery of sexual function. All data studies. In early phase clinical studies (NCT02361723; NCT03333915),
was prospectively collected and retrospectively analyzed. TRT patients were pamiparib was generally well tolerated and showed preliminary antitumor
proportionately matched to 419 control patients by pathologic Gleason activity; 60 mg orally twice daily (BID) was established as the recommended
Grade Group (GGG) and stage. Univariate and multivariate comparisons were investigational dose. Methods: This open-label, global, phase 2 study
used to compare rates and time to BCR (two consecutive PSAs $ 0.2 ng/dl); (NCT03712930) evaluates the antitumor activity and safety/tolerability of
Cox regression modeling was used to generate a survival function at the mean pamiparib in mCRPC patients (pts) with CTC-HRD, assessed by the CTC-HRD
of covariates. Results: There were no statistically significant differences in assay, or deleterious germline/somatic mutations in BRCA1/2. Patients must
preoperative PSA, age, prostate weight, pathologic GGG and stage between have progressed on/after $1 androgen receptor-targeted therapy, received $1
the control versus TRT groups. Median follow-up time was 3 years in both taxane-based therapy, and have prostate-specific antigen (PSA) progression
groups. 7/152 (4.6%) and 39/419 (9.3%) patients experienced BCR in the per PCWG3 criteria. Four cohorts of pts will receive pamiparib 60 mg BID in 28-
TRT versus control groups, respectively (unadjusted, p=0.068). In adjusted day cycles. Cohort 1 will include ~50 pts with CTC-HRD+ +/- BRCA1/2mut
time to-analysis, TRT was an independent predictor of recurrence-free mCRPC with measurable metastatic disease; Cohort 2 will include ~30 pts with
survival, after controlling for GGG, p-stage, preoperative FT and PSA. A CTC-HRD+ +/- BRCA1/2mut mCRPC with bone-only disease; Cohorts 3 & 4 will
patient on TRT was approximately 53% less likely to experience a BCR (OR: include ~20 pts with CTC-HRD-/unk + BRCA1/2mut mCRPC with measurable
0.534, 95%CI: 0.288-0.993). Conclusions: After accounting for pathologic metastatic disease (Cohort 3), or bone-only disease (Cohort 4). Disease status
GGG, stage, and other significant covariates, the use of TRT independently will be assessed every 8 wks for 24 wks, then every 12 wks; PSA levels will be
reduced recurrence post-RP. These results suggest the need for a multi- tested every 4 wks. Co-primary endpoints are radiographic ORR assessed by
center randomized control trial. IRC (pts with measurable disease) and confirmed PSA response rate per
PCWG3 criteria (pts +/- measurable disease). Secondary endpoints include
ORR, time to PSA response/progression, duration of PSA response, time to
symptomatic skeletal event, radiographic progression-free survival, overall
survival, and safety. As of 05 December 2018, this study is actively enrolling.
Clinical trial information: NCT03712930.

TPS5087 Poster Session (Board #198b), Sat, 1:15 PM-4:15 PM TPS5088 Poster Session (Board #199a), Sat, 1:15 PM-4:15 PM
A phase Ib/II study of niraparib combination therapies for the treatment of Prostate cancer with oligometastatic relapse: Combining stereotactic ablative
metastatic castration-resistant prostate cancer (NCT03431350). First Au- radiotherapy and durvalumab, a randomized phase II trial (POSTCARD -
thor: Sumit Kumar Subudhi, The University of Texas MD Anderson Cancer GETUG-P13). First Author: Stéphane Supiot, Institut de Cancérologie de
Center, Houston, TX l’Ouest René Gauducheau, Nantes Saint Herblain, France
Background: Assessing multiple therapies in a single clinical trial can facilitate Background: Following Stereotactic Body RadioTherapy (SBRT) targeting
the rapid identification of new agents for the treatment of patients with met- prostate cancer (PCa) oligometastasis, a tumor response assessed by PSA
astatic castration-resistant prostate cancer (mCRPC). Niraparib (Nirap) is a decrease is observed in a majority of case. To increase this tumor response,
highly selective PARP inhibitor, with potent activity against PARP-1 and PARP-2 immune targeting agents can be combined with SBRT. We hypothesize that
deoxyribonucleic acid (DNA)-repair polymerases. PARP inhibition may be es- the anti-PDL1 agent Durvalumab will enhance immune response following
pecially lethal in tumor cells with genetic DNA damage response deficits (DRD). SBRT targeting oligometastatic lesions. In this multicenter randomized
Based on promising preclinical and clinical data, this study is designed as a phase II trial, we therefore propose to assess the comparative efficacy of
master protocol with nirap as a backbone therapy. Combination 1 assesses the SBRT with or without Durvalumab (MEDI4736) in oligometastatic recurrent
safety and efficacy of nirap plus JNJ-63723283 (JNJ-283), an anti-PD-1 hormone sensitive prostate cancer patients. Methods: Patients with PCa
monoclonal antibody. Combination 2 assesses nirap plus abiraterone acetate were eligible if they had a biochemical recurrence following treatment with
and prednisone (AA-P). Methods: This multicenter, global, open-label study is curative intent, with a maximum of 5 bone or lymph node metastases, seen
currently open at 18 sites in 5 countries of the planned XX sites, and is enrolling only on FCH-PET CT or Ga-PSMA PET CT, not seen on conventional imaging
patients with mCRPC who have progressed on $1 androgen-receptor targeted assessments (bone scan or thorax, abdomen and pelvis CT scan). Main
therapy for mCRPC. Enrollment at time of abstract submission was 25 for exclusion criteria are serum testosterone level , 8.5 nmol/ml, PSA doubling
combination 1. When combined with AA-P, the RP2D has been determined to be time less than 6 months, lung, brain, liver or other visceral metastases,
nirap 200 mg. The recommended phase-2 dose (RP2D) of nirap plus JNJ-283 any prior immune therapy. Patients were randomly assigned (2:1) to either
was determined in Part 1 based on the incidence of specified adverse events and SBRT + Durvalumab, or SBRT alone, with a stratification by investigation
PK data to be 480 mg every 4 weeks. For Part 2 of the study, patients are center, and number of metastases (1 vs 2-5). Durvalumab (1500 mg/cycle)
assigned to receive oral niraparib daily plus JNJ-283 infusions once every four will be started one month prior to SBRT to be able to evaluate PSA and
weeks until disease progression, unacceptable toxicity, death, study termination. immune response to the drug. It will be combined with SBRT (3x 9 or 3 x 11 Gy)
Part 2 is described in the table. Clinical trial information: NCT03431350. and then given adjuvantly until progression with a maximum of 12 months.
Combination 1: Nirap + JNJ283 Combination 2: Nirap + AA-P
Ninety-six patients will be recruited in 15 centers over a recruitment period of
2.5 years. The primary endpoint is two-year Progression-free survival. Secondary
Inclusion DRD positive or DRD negative DRD positive
criteria objectives include quality of life, androgen deprivation therapy free survival,
1 or 2 prior lines of androgen-receptor targeted 1 prior line of androgen-receptor targeted prostate cancer specific survival, overall survival, time to first symptomatic
therapy given for mCRPC therapy given for mCRPC
Expected Approximately 60 (n = 30 DRD positive and 30 n = 80 DRD positive event, acute and late toxicity. Patients in both groups will be observed for PSA
enrollment DRD negative) progression every 3 months after random assignment, with confirming imaging
Part 2 Objective response rate (ORR) Composite response rate (RR) defined as 1 of
the following by PCWG3: at PSA progression. Clinical trial information: NCT03795207.
Primary end Incidence and severity of AEs Objective response
point
Safety Circulating tumor cell response
assessment PSA decline of $50%
Incidence and severity of AEs

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
310s Genitourinary (Prostate) Cancer

TPS5089 Poster Session (Board #199b), Sat, 1:15 PM-4:15 PM TPS5090 Poster Session (Board #200a), Sat, 1:15 PM-4:15 PM
An open label phase I/IIa study to evaluate the safety and efficacy of CCS1477 Nivolumab and ipilimumab treatment in prostate cancer with an immunogenic
as monotherapy and in combination in patients with advanced solid/metastatic signature (NEPTUNES). First Author: Yien Ning Sophia Wong, Cancer Im-
tumors. First Author: Johann S. De Bono, Royal Marsden NHS Foundation munology Unit, University College London Cancer Institute, London, United
Trust and The Institute of Cancer Research, London, United Kingdom Kingdom
Background: CCS1477 is a potent, selective and orally bioavailable inhibitor Background: Responses to checkpoint inhibitor (CPI) monotherapy in patients with
of the bromodomain of p300 and CBP, two homologous and critical co- metastatic castration resistant prostate cancer (mCRPC) have been limited. This is in
activators of the androgen receptor (AR) and its variant forms, including part attributed to low tumour mutational burden (TMB) and low tumour
mutated, amplified and spliced AR, as well as c-Myc. CCS1477 represents a infiltrating lymphocytes (TILs). Previously ~20% patients with prostate cancer have
new therapeutic option for prostate cancer patients who have progressed after demonstrated high TILs or TMB1. We hypothesize that patients with higher TMB due to
failure of anti-androgen therapy and in combination with anti-androgens such mismatch repair deficiency (dMMR) or defective DNA damage response (dDDR) and
as enzalutamide or abiraterone. Methods: This is a Ph I/IIa study to determine patients with high TILs are more likely to respond to combination CPI with anti PD-1
and anti CTLA-4 therapy. Methods: NEPTUNES is a single arm phase II trial designed
the maximum tolerated dose (MTD) and/or recommended Phase II dose and
to assess the efficacy of nivolumab and ipilimumab in biomarker selected patients with
schedule(s) of CCS1477 and investigate clinical activity of CCS1477 mon- mCRPC that have progressed following $1 line of therapy. The immunogenic signature
otherapy and CCS1477 in combination with abiraterone or enzalutamide in (ImS) biomarker is defined by $1 of the following: 1) dMMR by immunohistochemistry
patients with metastatic castration resistant prostate cancer (mCRPC). The (IHC); 2) dDDR detected by the UW-OncoPlex sequencing assay and; 3) high TILs on
trial aims to enrol approximately 150 patients and is currently recruiting in the multiplexed IHC. The UW-OncoPlex assay detects mutations in .260 genes and
UK with plans to open additional sites in the USA (NCT03568656). Key provides an estimation of TMB. Assuming an ImS+ rate of 20%, we aim to pre-screen
inclusion criteria (for the mCRPC) require previous treatment with abiraterone 175 patients in order to enrol 35 patients into the main study. The primary endpoint is
and/or enzalutamide, taxane as well as evidence of disease progression composite response rate (CRR), achieved if $1 of the following criteria are satisfied: 1)
(PCWG-3 guidelines). Single dose and steady state pharmacokinetics will be radiological response by RECIST 1.1; 2) PSA response $50%; 3) conversion of
determined along with changes in plasma PSA, LDH and ALKP and in cir- circulating tumour cells (CTC) count from $5 cells at baseline to ,5 cells at week 9.
culating tumour cell number. Anti-tumour activity will be determined by The treatment will be deemed ineffective if the CRR is ,20%. Nivolumab (1 mg/kg)
standard imaging according to PCWG-3 guidelines. Paired tumour biopsies for and ipilimumab (3 mg/kg) is dosed every three weeks for up to 4 times, followed by a
biomarker assessment are being collected. Cohort 1 of the monotherapy dose- 480mg flat dose of nivolumab every 4 weeks for up to one year. Baseline biopsies are
escalation (rolling 6 design; 3-6 patients/cohort) has completed. Enrolment to mandated and paired biopsy at week 12 is encouraged. The secondary endpoints
cohort 2 began in January 2019. Dose finding in combination (CCS1477 + include safety, overall survival, and radiological and PSA progression free survival.
abiraterone; CCS1477 + enzalutamide) will be open once monotherapy dose Exploratory biological markers including TMB, mutational profiles, change in TILs and
liquid biomarkers will be correlated with the primary clinical endpoint. Since opening
escalation completes. Following definition of a recommended phase 2 dose
in February 2018, 126 patients have been pre-screened with 25 ImS+. To date, 9/25
and schedule for monotherapy and in combination, three expansion arms in ImS+ patients have been enrolled into the main study. The trial is ongoing, with patient
patients with mCRPC will be opened in parallel (25 patients/arm); CCS1477 accrual expected to complete by late 2019. References: 1Linch, M., Goh, G., Hiley, C.,
monotherapy; CCS1477 + abiraterone; CCS1477 + enzalutamide. A further Shanmugabavan, Y., McGranahan, N., Rowan, A., . . . Swanton, C. (2017). Intra-
expansion in patients with advanced solid tumours with a mutation in p300 or tumoural evolutionary landscape of high-risk prostate cancer: the PROGENY study of
CBP will also be opened. Clinical trial information: NCT03568656. genomic and immune parameters. Ann Oncol, 28(10), 2472-2480. doi:10.1093/
annonc/mdx355. Clinical trial information: NCT03061539.

TPS5091 Poster Session (Board #200b), Sat, 1:15 PM-4:15 PM TPS5092 Poster Session (Board #201a), Sat, 1:15 PM-4:15 PM
IMPACT: Immunotherapy in patients with metastatic cancers and CDK12 TALAPRO-2: Part 2 (P2) of the placebo-controlled phase 3 study of talazoparib
mutations. First Author: Melissa Andrea Reimers, University of Michigan, (TALA) with enzalutamide (ENZA) in metastatic castration-resistant prostate
Ann Arbor, MI cancer (mCRPC). First Author: Neeraj Agarwal, University of Utah Huntsman
Cancer Institute, Salt Lake City, UT
Background: Tumors with biallelic CDK12 loss have been identified as a
distinct subtype in metastatic castration resistant prostate cancer (mCRPC) Background: ENZA is approved to treat men with CRPC. TALA is a poly(ADP-
and other cancer types. The CDK12 biallelic loss mCRPC genomic signature, ribose) polymerase (PARP) inhibitor that inhibits PARP1/PARP2 and traps
distinct from homologous recombination deficient (HRD) and ETS fusion PARP on DNA, preventing DNA damage repair (DDR), and causing cell death in
signatures, is characterized by excessive tandem duplications, genomic in- BRCA1/2-mutated cells. TALA is approved in the US to treat germline BRCA1/
stability, gene fusion-caused putative neoantigens, and increased tumor T cell 2-mutated HER2- locally advanced/metastatic breast cancer. A combination
infiltration. Early clinical experience with anti-PD-1 immunotherapy in CDK12 of TALA with ENZA in mCRPC may improve clinical outcomes. TALAPRO-2
loss mCRPC patients (pts) is notable for deep and sustained PSA as well as (NCT03395197) is a 2-part study to evaluate the efficacy, safety, pharma-
radiographic responses. We hypothesize that CDK12 biallelic loss is a potential cokinetics and (patient) pt-reported outcomes of the combination treatment.
biomarker of immune checkpoint immunotherapy (ICI) efficacy in mCRPC and The focus here is on P2 of TALAPRO-2. Methods: Approximately 860 pts are
other cancers. Methods: IMPACT (NCT03570619) is a multi-center, open planned to be enrolled in P2 from multinational sites. Pts are aged $18 years,
label, phase 2 study of pts with metastatic cancers that harbor CDK12 biallelic have asymptomatic/mildly symptomatic mCRPC, Eastern Cooperative On-
loss. mCRPC pts will be enrolled in cohort A (n = 25) in a Mini-Max Simon Two- cology Group performance status #1, no brain metastases, and have not
Stage design, and all other pts in single-stage cohort B (n = 15). All pts will received taxanes/novel hormonal therapy (NHT). P2 is a randomized double-
receive induction therapy with nivolumab 3 mg/kg IV and ipilimumab 1 mg/kg blind study that will evaluate safety, efficacy, and pt-reported outcomes of
IV q3 weeks for up to 4 cycles, followed by maintenance nivolumab at 480 mg TALA (0.5 mg QD) + ENZA (160 mg QD) vs placebo + ENZA (160 mg QD). Pts
IV q4 weeks (up to 52 weeks in total). Eligible pts must have identified biallelic will be randomized to 1 of 2 treatment groups: TALA + ENZA, or matching
CDK12 loss on any CLIA/CAP approved next generation sequencing assay placebo + ENZA. Randomization will be stratified by prior treatment with NHT
and a histologic diagnosis of metastatic prostate adenocarcinoma or other for castration-sensitive prostate cancer (CSPC) or prior treatment with taxane-
metastatic carcinoma. No prior ICI is allowed. The primary endpoint is the based chemotherapy for CSPC (yes/no) and DDR mutation status (deficient vs.
overall response rate (ORR) in cohort A per PCWG3 criteria. An ORR of 30% is nondeficient/unknown). The primary endpoint is radiographic progression-free
targeted in cohort A. Secondary endpoints include safety, secondary efficacy survival (rPFS), defined as time to progression in soft tissue per RECIST v1.1 or
measures, quality of life, and survival measures. Exploratory objectives include in bone per PCWG3 criteria or death. The key secondary endpoint is overall
tumor whole exome analysis and changes in immune profiles with therapy. survival (OS). Efficacy will be assessed by radiography every 8 weeks up to
Comprehensive and serial monitoring of peripheral blood immune cell pop- week 25 and every 8-12 weeks thereafter. The analyses of rPFS will be
ulations will be performed via T cell clonal diversity assessment and multi- compared between TALA in combination with ENZA and placebo in combi-
parametric flow cytometry. Changes in myeloid and lymphoid populations will nation with ENZA by using a 1-sided stratified log-rank test. OS will be
be assessed from whole blood. Polarization and effector function of T cells and evaluated separately in the all comers and the DDR-deficient populations. Pt
activation of antigen presenting cells will be further characterized from isolated recruitment is ongoing. Results: n/a. Conclusions: n/a. Clinical trial in-
peripheral blood mononuclear cells. Study accrual is ongoing. Clinical trial formation: NCT03395197.
information: NCT03570619.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Genitourinary (Prostate) Cancer 311s

TPS5093 Poster Session (Board #201b), Sat, 1:15 PM-4:15 PM TPS5094 Poster Session (Board #202a), Sat, 1:15 PM-4:15 PM
A phase III, multicenter study to assess the diagnostic performance and Cabazitaxel with abiraterone versus abiraterone alone randomized trial for
clinical impact of 18F-DCFPyL PET/CT in men with suspected recurrence of extensive disease following docetaxel: the CHAARTED 2 Trial: A trial of the
prostate cancer (CONDOR). First Author: Michael J. Morris, Memorial Sloan ECOG-ACRIN Cancer Research Group (EA8153). First Author: Christos
Kettering Cancer Center, New York, NY Kyriakopoulos, University of Wisconsin Carbone Cancer Center, Madison, WI
Background: Early and accurate detection of recurrent or metastatic prostate Background: The E3805 CHAARTED trial showed a significant survival benefit
cancer remains an unmet diagnostic need for patient management. While from early treatment with docetaxel in patients (pts) with high-volume meta-
agents for positron emission tomography (PET), such as 11C-choline and 18F- static hormone-sensitive prostate cancer (HSPC). However, eventually all these
fluciclovine, have emerged as options for imaging recurrent prostate cancer, patient will develop castration-resistant disease (CRPC) and it is currently
these agents are not specific for the disease. 18F-DCFPyL is a novel, low- unknown what is the optimal treatment in that setting. We hypothesize that
molecular weight, PET radiopharmaceutical that binds selectively to prostate- treatment with a second round of chemohormonal therapy –this time in the
specific membrane antigen with high affinity. In prior studies, 18F-DCFPyL CRPC setting- will improve outcomes. Methods: This is a prospective ran-
PET/CT has shown reliable diagnostic performance in detecting metastatic or domized phase II open label trial led by ECOG-ACRIN. Pts (N = 210) with
recurrent prostate cancer (Rowe Mol Imaging Biol 2016 18:411-19; Gorin J metastatic CRPC who have previously received docetaxel for HSPC will be
Urol 2018 1999:126-32). Methods: CONDOR is a phase 3, multicenter, randomized to 2 treatment arms (1:1): abiraterone/prednisone plus cabazitaxel
open-label study designed to assess the diagnostic performance and clinical for up to 6 cycles versus abiraterone/prednisone alone. Stratification factors
impact of 18F-DCFPyL PET/CT in men with suspected recurrent or metastatic include ECOG performance status, time from initiation of ADT to development
prostate cancer. Approximately 200 patients are planned to be enrolled across of CRPC and visceral versus bone only metastases. The primary objective is
15 centers in the United States and Canada. Eligible patients $18 years of age to evaluate the efficacy of abiraterone/prednisone plus cabazitaxel versus
must have histologically confirmed prostate adenocarcinoma, have rising PSA abiraterone/prednisone alone. The primary endpoint of the trial is progression-
after definitive therapy, and negative or equivocal conventional imaging. A free survival (PFS), defined clinically or radiographically. Secondary endpoints
single 9 mCi (333 MBq) dose of 18F-DCFPyL is administered, followed by include PSA response, radiographic response, time to PSA progression, overall
whole body PET/CT scan 1 hour later. The primary objective is to assess the survival and safety. Exploratory objectives include response assessment based
correct localization rate (percentage of patients with a one-to-one corre- on AR-V7 status as well as by NaF PET/CT imaging. Key inclusion criteria
spondence between localization of at least one lesion identified on 18F- include: metastatic CRPC, prior treatment with at least 3 cycles of docetaxel for
DCFPyL PET/CT and the composite truth standard, defined as either evalu- metastatic HSPC, no prior treatment for CRPC (radium-223 or sipuleucel-T
able histopathology, informative correlative imaging, or PSA response after excluded), ECOG performance status of 0-2 and peripheral neuropathy of less
radiation therapy). Additional study objectives include safety and tolerability of than 2. The study was activated on 02/08/2018. As of 01/21/2019, 28/210
18
F-DCFPyL, impact on intended treatment plans, detection rates and PPV of pts have been enrolled. Support: CA180820, CA180794, CA180799,
18
F-DCFPyL by region, and detection rates by baseline PSA. 18F-DCFPyL PET/ CA180802; and Genzyme Corporation, a subsidiary of Sanofi S.A. Clinical
CT results are centrally reviewed by independent readers blinded to all clinical trial information: NCT03419234.
and other imaging information. As of February 9, 2019, a total of 36 patients
have been dosed in the study. Clinical trial information: NCT03739684.

TPS5095 Poster Session (Board #202b), Sat, 1:15 PM-4:15 PM TPS5096 Poster Session (Board #203a), Sat, 1:15 PM-4:15 PM
High dose testosterone in men with metastatic castrate-resistant prostate Focal radiation for oligometastatic castration-resistant prostate cancer (FORCE):
cancer (mCRPC) and homologous recombination deficiency (HRD). First A phase II randomized trial. First Author: Zachery Reichert, University of
Author: Julia Chen, Kinghorn Cancer Centre, Sydney, NSW, Australia Michigan, Ann Arbor, MI
Background: mCRPC progresses via adaptive mechanisms that allow on- Background: Metastasis Directed Therapy (MDT) is the focus of recently
going androgen receptor (AR) signalling despite castrate levels of androgens. completed and ongoing clinical trials. For biochemically recurrent prostate
Bipolar androgen therapy (BAT), cycling between supraphysiologic and cancer, MDT delays the initiation of androgen deprivation, but is noncurative
subphysiologic serum testosterone levels, aims to exploit these adaptations as . 75% of patient’s progress within 3 years. In metastatic castration-
to induce tumor regression. Extensive clinical data demonstrate the safety resistant prostate cancer (mCRPC), focal therapies are common for palliation
and efficacy of BAT in men with asymptomatic mCRPC. However, de novo of pain, but not routinely recommended for controlling systemic disease.
resistance is still common and predictive biomarkers to refine patient se- Whole genome sequencing has demonstrated that mCRPC metastases seed
lection are lacking. Pre-clinical data suggest that the induction of double- new metastases and patients with a greater burden disease develop more
stranded DNA (dsDNA) breaks by BAT may be crucial to its mechanism of rapid resistance to next-generation hormone therapies. Therefore, MDT to all
action. DNA repair defects, such as HRD, are particularly relevant in CRPC visible disease may be beneficial in oligometastatic CRPC. Methods: This
patients. We hypothesize that CRPC patients with DNA repair deficits such is a prospective, randomized phase 2 clinical trial with a primary endpoint of
as HRD, may be particularly responsive to BAT. Methods: This is a phase II progression-free (radiographic + clinical) survival at 18 months. Secondary
prospective single arm interventional trial (NCT03522064). Up to 30 pa- endpoints include objective disease PFS, PSA PFS, PSA response rate, non-
tients will be recruited based on a Simon two-stage design with a power of irradiated distant MFS and PROs. The utility of 68Ga-PSMA PET/CT will be
90% to detect an increase in response rate from 20% to 40%. Key inclusion explored and sequential analysis of circulating tumor DNA/circulating tumor
criteria include i) asymptomatic or minimally symptomatic mCRPC, ii) rising cells will be done. Sites of disease are defined by all available imaging
PSA despite a castrate serum testosterone and iii) HRD on germline, tumor (minimum of CT or MRI, NM bone scan, but potentially molecular imaging).
and/or circulating tumor DNA (ctDNA) analysis. Key exclusion criteria in- Oligometastatic CRPC is defined as between 1 and #5 treatment sites.
clude i) ADT , 1 year, ii) disease extent/sites that would cause significant Eligible patients with progressive oligometastatic CRPC may have prior
risk if tumor flare occurs (e.g.: brain) and iii) significant cardiac disease. sipuleucel-T but no therapy for mCRPC. Docetaxel or abiraterone within the
Previous PARP inhibitor therapy will be permitted in a subset. Participants hormone sensitive phase is allowed. Patients will be randomized 1:1
will receive IM testosterone enanthate 500mg q4w in combination with (stratified by 1-2 vs 3-5 oligometastases and whether molecular imaging in
ongoing LHRH antagonist/agonist or orchidectomy. The primary endpoint is mCRPC is available) to physician’s choice vs. physician’s choice + MDT.
PSA response rate defined as PSA reduction $50% from baseline. Sec- MDT may be delivered via multiple radiation oncology techniques to cyto-
ondary endpoints include time to PSA progression, quality of life, radiologic toxic doses. If the prostate itself has not been treated (on MDT arm), it will be
response and safety and tolerability. Exploratory endpoints include changes treated and count as 1 site. Statistical analysis will provide objective PFS by
in ctDNA and tumoral DNA alterations from baseline to progression. Accrual arm with associated 90% confidence intervals. The study is recruiting 72
is ongoing. Clinical trial information: NCT03522064. randomized patients at the University of Michigan Rogel Cancer Center and
soon the Ann Arbor Veteran’s Administration Hospital. The trial is funded by
the Prostate Cancer Foundation and a University of Michigan Cancer Center
Trial Support Award. Clinical trial information: NCT03556904.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
312s Genitourinary (Prostate) Cancer

TPS5097 Poster Session (Board #203b), Sat, 1:15 PM-4:15 PM TPS5098 Poster Session (Board #204a), Sat, 1:15 PM-4:15 PM
A phase II Salvage Trial of AR Inhibition with ADT and Apalutamide with Cognitive effects of androgen receptor (AR) directed therapies for advanced
Radiation therapy followed by docetaxel in men with PSA recurrent prostate cancer of the prostate (COGCaP). First Author: Julie Van, Vanderbilt University
cancer (PC) after radical prostatectomy (STARTAR). First Author: Tian Zhang, Medical Center, Nashville, TN
Duke University Medical Center, Durham, NC
Background: Androgen deprivation therapy (ADT) is the cornerstone of
Background: Androgen deprivation combined with salvage external beam treatment for prostate cancer (CaP). However, the relationship between ADT
radiation therapy (RT) have improved survival for patients (pts) with non- and the development of cognitive dysfunction in men with CaP is contro-
metastatic hormone naı̈ve PC and PSA recurrence after radical prostatec- versial. Past studies had various methodological limitations, including the
tomy (RP). Our recent STREAM trial showed addition of enzalutamide to RT inconsistency and insensitivity of measures used for cognitive testing.
and ADT had a 3-year progression free survival (PFS) of 53%. Adding ef- Methods: COGCaP is a multi-site, prospective observational study of cog-
fective PC treatments in this setting may further improve 3-year PFS. nitive function and patient reported outcomes in men with CaP treated with
Methods: STARTAR is an investigator-initiated phase 2 trial for salvage ADT and androgen receptor (AR) directed therapies such as enzalutamide or
treatment of biochemically recurrent PC following prostatectomy. Key in- abiraterone acetate (AA) conducted across four U.S. sites. Patients with
clusion criteria include histologic prostate adenocarcinoma, either Gleason metastatic castration-resistant or hormone sensitive CaP starting abiraterone
7 with T3/positive margin/1-4 positive lymph nodes or Gleason 8-10 disease, (N=50), or non-metastatic or metastatic castration-resistant CaP starting
PSA relapse within 4 years of prostatectomy (minimum PSA 0.2 ng/mL to enzalutamide (N=50) undergo cognitive and patient reported outcome as-
maximum PSA 4 ng/mL). Treatment involves ADT with apalutamide for sessments at baseline, 3, 6, and 12 months. The primary endpoint compares
9 months, continue with with prostate bed +/- nodal RT at month 3, followed mean change in cognitive function between groups at 3 months using
by 6 cycles of docetaxel 75mg/m2 IV every 3 weeks for 6 cycles. The primary CANTAB, a computer-based measure of cognitive function. This design
endpoint of the study is 3-year PFS. With a one-sided alpha of 0.05 to achieves a power of 80% to detect a between-group difference in mean
improve 3-year PFS from 50% to 75%, we will have 92% power by enrolling cognitive composite score of , 1 standard deviation and a two-sided type I
42 pts (including 10% dropout rate) based on the binomial test. Key sec- error rate of 0.05. Functional MRI (fMRI) images will be assessed between
ondary endpoints include 1, 2, and 3-year PSA recurrence rates with tes- baseline and 3 months to assess structural and functional changes in the
tosterone recovery, PSA PFS, PSA nadir, time to testosterone recovery, and brain as secondary endpoints. Study patients will be incorporated into a
safety of combination therapy. Quality of life will be assessed by EPIC model of genetic risk of cognitive dysfunction in order to identify populations
questionnaire. As of February 2019, we have enrolled and treated 12 pts in that are especially vulnerable to cognitive change when undergoing an-
this PCCTC trial. Accrual to the STARTAR trial is ongoing (NCT03311555). drogen deprivation. This ongoing study will define a reproducible method-
Clinical trial information: NCT03311555. ology for cognitive assessments via computer based tests that can be
standardized and disseminated within multi-site trials of men with CaP.
Clinical trial information: NCT03016741.

TPS5099 Poster Session (Board #204b), Sat, 1:15 PM-4:15 PM TPS5100 Poster Session (Board #205a), Sat, 1:15 PM-4:15 PM
VISION: An international, prospective, open-label, multicenter, randomized PROTEUS: A randomized, double-blind, placebo (PBO)-controlled, phase 3
phase 3 study of 177Lu-PSMA-617 in the treatment of patients with progressive trial of apalutamide (APA) plus androgen deprivation therapy (ADT) versus
PSMA-positive metastatic castration-resistant prostate cancer (mCRPC). First PBO plus ADT prior to radical prostatectomy (RP) in patients with localized
Author: A. Oliver Sartor, Tulane Medical School, New Orleans, LA high-risk or locally advanced prostate cancer (PC). First Author: Mary-Ellen
Taplin, Lank Center for Genitourinary Oncology, Dana-Farber Cancer Institute,
Background: The novel therapeutic drug 177Lu-PSMA-617 is a prostate
Boston, MA
specific membrane antigen (PSMA) targeting agent to deliver radionuclide
therapy for the treatment of pts with metastatic castration resistant prostate Background: Patients (pts) with localized high-risk PC experience disease
cancer. Based on preclinical data that demonstrated high PSMA binding progression rates of approximately 50% after RP (Kane et al. J Urol. 2007). With
affinity & compound internalization, prolonged tumor uptake, rapid kidney the approval of next-generation androgen receptor inhibitors, neoadjuvant
clearance, & high tumor-to-background ratio, 177Lu-PSMA-617 proceeded studies have shown that 6 months of androgen blockade may improve local
into clinical development. Preliminary clinical evidence indicates 177Lu- disease control at the time of RP (McKay et al. Prostate Cancer Prostatic Dis.
PSMA-617 may demonstrate clinical benefit in pts with mCRPC in a set- 2017; Taplin et al. JCO. 2014). The purpose of this study is to determine if
ting where pts had no recommended standard of care. This Phase 3 study will treatment with APA plus ADT before and after RP in pts with localized high-risk or
assess the efficacy of 177Lu-PSMA-617 in patients with progressive PSMA- locally advanced PC results in an improvement in pathologic complete response
positive mCRPC by measuring overall survival (OS) and radiographic pro- (pCR) rate and metastasis-free survival (MFS) compared with PBO plus ADT.
gression free survival (rPFS) in a randomized, prospective, open-label trial. Methods: This international multicenter trial is enrolling pts with localized high-
Methods: The primary objective of this study is to compare the 2 alternative risk or locally advanced PC who are candidates for RP. Eligibility criteria: Any
endpoints of rPFS & OS in pts with progressive PSMA-positive mCRPC who Gleason score (GS) $ 4 + 3 with $ 6 positive systematic biopsies (SB); any GS
receive 177Lu-PSMA-617 in addition to best supportive/standard of care vs pts $ 4 + 3 with $ 3 SB and prostate-specific antigen (PSA) $ 20 ng/mL; GS
treated with best supportive/best standard of care alone. Eligibility criteria are: $ 9 in $ 1 SB or targeted biopsies (TB); or $ 2 SB or TB with continuous GS $ 8,
PSMA expressing tumor; prior exposure to a taxane and novel androgen axis each with $ 80% involvement. Stratification: GS (7 or $ 8), cN0 or N1, and
drug. Pts will be randomized in a 2:1 ratio in favor of the investigational arm region (North America, Europe, or rest of world). Randomization: 1:1 to APA
with stratification factors of LDH, liver disease, ECOG score, and use of NAAD (240 mg) plus ADT (LHRHa) or PBO plus ADT. Pts will receive 6 treatment
at time of randomization as a standard of care. Under the alternative hy- cycles, followed by RP, followed by an additional 6 cycles. Dual primary end
pothesis, median OS on active is assumed to be 13.7 mo for a HR of 0.7306 points: pCR rate (to be assessed by blinded independent central pathology
and rPFS on the active is assumed to be 6 mo for a HR of 0.67. Planned review) and MFS (to be assessed by blinded independent central radiology
enrollment for this study is 750 patients. Enrollment began in June 2018 and review [BICR]). Secondary end points: PSA-free survival and progression-
continues; the IDMC last reviewed the trial for safety in January 2019 and free survival. Imaging with CT or MRI and bone scan will be conducted at
suggested that the trial continue as planned. Clinical trial information: baseline and then every 6 months following biochemical failure until
NCT03511664. documented distant metastasis by BICR, or death. Approximately 1500 pts
will be enrolled globally over 3.0 years in 240 sites in 19 countries. An
independent data monitoring committee is commissioned to review trial
data. Clinical trial information: NCT03767244.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Genitourinary (Prostate) Cancer 313s

TPS5101 Poster Session (Board #205b), Sat, 1:15 PM-4:15 PM

Randomized prospective phase 3 trial of 68Ga-PSMA-11 PET/CT molecular
imaging for prostate cancer salvage radiotherapy planning [PSMA-SRT].
First Author: Jeremie Calais, UCLA, Los Angeles, CA
Background: Salvage radiotherapy (SRT) for prostate cancer (PCa) re-
currence after prostatectomy offers long-term biochemical control in about
50–60% of patients. SRT is commonly initiated in patients with serum PSA
levels , 1 ng/mL, a threshold at which standard-of-care imaging is in-
sensitive for detecting recurrence. As such, SRT target volumes are usually
drawn in the absence of radiographically visible disease. 68Ga-PSMA-11
(PSMA) PET/CT molecular imaging is highly sensitive and may offer ana-
tomic localization of PCa biochemical recurrence. However, it is unclear if
incorporation of PSMA PET/CT imaging into the planning of SRT could
improve its likelihood of success. The purpose of this trial is to evaluate the
success rate of SRT for recurrence of PCa after prostatectomy with and
without planning based on PSMA PET/CT. Methods: We will randomize 193
patients to proceed with standard SRT (control arm 1, n = 90) or undergo a
PSMA PET/CT scan (free of charge for patients) prior to SRT planning
(investigational arm 2, n = 103). The primary endpoint is the success rate of
SRT measured as biochemical progression-free survival (BPFS) after initi-
ation of SRT. Biochemical progression is defined by PSA $ 0.2 ng/mL and
rising. The randomization ratio of 1:1.13 is based on the assumption that
approximately 13% of subjects randomized to Arm 2 will not be treated with
SRT because of PSMA-positive extra-pelvic metastases. These patients will
not be included in the primary endpoint analysis but will still be followed. The
choice of treating the prostate bed alone vs prostate bed and pelvic lymph
nodes, with or without androgen deprivation therapy (ADT), is selected by the
treating radiation oncologist. The radiation oncologist may change the ra-
diation plan depending on the findings of the PSMA PET/CT scan. Any other
imaging is allowed for SRT planning in both arms if done per routine care.
Patients will be followed until either one of the following conditions occur: 5
years after the date of initiation of randomization, biochemical progression,
diagnosis of metastatic disease, initiation of any additional salvage therapy,
death. Discussion: This is the first randomized phase 3 prospective trial
designed to determine whether PSMA PET/CT molecular imaging can im-
prove outcomes in patients with PCa early BCR following radical prosta-
tectomy. Clinical trial information: NCT03582774.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
314s Gynecologic Cancer

5500 Oral Abstract Session, Mon, 1:15 PM-4:15 PM 5501 Oral Abstract Session, Mon, 1:15 PM-4:15 PM
A randomized phase 3 trial of paclitaxel (P) plus carboplatin (C) versus paclitaxel Activity of durvalumab in advanced endometrial cancer (AEC) according to
plus ifosfamide (I) in chemotherapy-naive patients with stage I-IV, persistent or mismatch repair (MMR) status: The phase II PHAEDRA trial (ANZGOG1601).
recurrent carcinosarcoma of the uterus or ovary: An NRG Oncology trial. First First Author: Yoland Catherine Antill, Cabrini Health, Malvern, Australia
Author: Matthew A. Powell, Washington University School of Medicine in St.
Background: Deficient DNA mismatch repair (dMMR) occurs in approximately
Louis, St. Louis, MO
15% of AEC and is associated with a high tumour mutation burden. Expression
Background: Gynecologic carcinosarcomas (CS) are rare yet aggressive epi- of PD-1 and PD-L1 has been reported in up to 90% of ECs, including those with
thelial malignancies for which optimal therapy is debated. PI was shown to be proficient DNA mismatch repair (pMMR). We report here preliminary results of
superior to I. PC demonstrated compelling phase 2 activity with improved safety PHAEDRA, a single-arm phase 2 trial designed to determine the activity of
and convenience. Methods: Main inclusion: $18 y; chemotherapy naı̈ve stage I- single-agent durvalumab, an antibody to PD-L1, in 2 cohorts of women with
IVB or recurrent uterine (U) or ovarian (O) CS. Treatment randomised 1:1 to PC AEC. Methods: Participants (pts) had pMMR AEC progressing after 1-3 lines of
(P 175mg/m2 with C: AUC 6 or 5 if prior RT on D1) or PI (P: 135 mg/m2; I 1.6 g/ chemotherapy, or dMMR AEC progressing after 0-3 lines of chemotherapy, and
m2 D1-3; G-CSF support with dose escalation & de-escalation based on nadir were treated with durvalumab 1500mg IV Q4W. The primary endpoint was
counts) q21days for 6-10 cycles. Quality of life (QOL) (FACT-En TOI) and objective tumour response (OTR = complete response [CR] or partial response
neurotoxicity (FACT/GOG-Ntx subscale) administered at 4 timepoints. A strat- [PR] by iRECIST). Secondary endpoints included disease control at 16 weeks
ified log-rank test compared primary endpoint of overall survival (OS) from entry (DC16w = CR, PR, or stable disease at 16 weeks [SD16w]), immune-related
between treatment groups for non-inferiority (NI) of PC to PI. With 264 events, adverse events (irAEs), PD-L1, germline mutations and MLH1 methylation.
power was 80% for a null hazard ratio of 1.2 against a 13% greater death rate on Other secondary endpoints include: OTR and DC by RECIST1.1, other AE,
PI when type I error is limited to 5% for a one-tail test. NCT00954174. PFS, OS & quality of life will be reported later. Results: 71 pts with AEC were
Results: 637 pts accrued with a median follow-up of 61 months. The primary (U, recruited from Feb 2017 to Sep 2018: 35 dMMR and 36 pMMR. Median
n = 536) and secondary (O, n = 101) cohorts are analyzed separately and follow-up were 8.3 vs 14.8 months in dMMR vs pMMR pts. Median age: 67
included 449 and 90 pts eligible pts, respectively. For the U cohort:PC and PI (range 36-81); ECOG PS: 0-1 in 68, and 2 in 3. Pathology: endometrioid in
were randomly assigned to 228 and 221 eligible pts. Stage distribution: I (40%), 94% and 58%; serous in 0% and 31%; grade: high in 42% and 83% (dMMR
II (6%), III (31%), IV (15%) and recurrent (8%). The study met its primary and pMMR respectively). Durvalumab was the 1st, 2nd and subsequent line of
objective withPC not inferior to PI (intention-to-treat analysis;Median OS 37 vs. non-hormonal therapy in 15, 14, and 6 pts with dMMR and 0, 21, and 15 pts
29 mo, HR = 0.87; 90% CI = 0.70 to 1.075; p , 0.01 for NI, p . 0.1 for with pMMR. Among dMMR pts, the OTR rate was 40% (14/35, 95% CI 26-
superiority (S)).PFS (median on PC 16mo vs PI 12mo; HR = 0.73; p = , 0.01 56), with 4 CR and 10 PR; 7 others had SD 16w for a DC16w rate of 60% (21/
for NI, p , 0.01 for S). Toxicity (grade 1/2/3/4/5: PC 1/8/40/48/2%; PI 1/32/39/ 35, 95% CI 44-74). OTR rate was 40% as 1st line, 43% as 2nd line, and 33%
25/1%). Most of increase toxicity for PC was hematologic with G-CSF rarely used as subsequent line treatment. Among pMMR pts, the OTR rate was 1/36 (3%,
(N = 6). Confusion and genitourinary hemorrhage were significantly worse with 95% CI 1-14) with 1 PR; 6 others had SD16w for a DC16w rate of 19% (7/36;
PI. Both groups had decline in QOL and neurotoxicity scores. Similar trends were 95% CI 10-35). IrAEs occurred in 14 pts: hyperthyroidism in 6, hypothy-
noted for the O cohort (OS: PC 30mo vs PI 25mo; and PFS: 15 mo vs 10 roidism in 6, pneumonitis in 1 and hepatitis in 1. Conclusions: Durvalumab
respectively). Conclusions: PC was not inferior to PI for OS with longer PFS and monotherapy showed promising activity and safety in AEC with dMMR regardless
similar QOL and neurotoxicity. These results establish a new standard regimen of prior lines of chemotherapy, but there was limited evidence of activity in AEC
for women with CS. Clinical trial information: NCT00954174. with pMMR. Clinical trial information: ACTRN12617000106336.

5502 Oral Abstract Session, Mon, 1:15 PM-4:15 PM 5503 Oral Abstract Session, Mon, 1:15 PM-4:15 PM
Phase 2, two-group, two-stage study of avelumab in patients (pts) with Results from neoadjuvant chemotherapy followed by surgery compared to
microsatellite stable (MSS), microsatellite instable (MSI), and polymerase chemoradiation for stage Ib2-IIb cervical cancer, EORTC 55994. First Author:
epsilon (POLE) mutated recurrent/persistent endometrial cancer (EC). First Gemma Kenter, Center Gynaecological Cancer Amsterdam, Amsterdam,
Author: Panagiotis A. Konstantinopoulos, Dana-Farber Cancer Institute, Netherlands
Boston, MA
Background: Conflicting evidence on the value of neoadjuvant chemother-
Background: This non-randomized phase 2 study evaluated the PD-L1 in- apy followed by surgery compared to concomitant chemoradiation in Stage
hibitor avelumab in two cohorts of EC: i) MSI/POLE cohort including ECs with IB2-IIB cervical carcinoma led to this multinational multicenter trial. As the
immunohistochemical (IHC) loss of expression of at least one of the mis- trial is approaching completion of its follow-up, preliminary results are
match repair (MMR) proteins and/or documented mutation in the exo- presented. Methods: Between May 2002 and June 2014 a total of 620
nuclease domain of POLE and ii) MSS cohort including ECs with normal IHC patients with FIGO stage Ib2-IIb were randomized between neoadjuvant
expression of all MMR proteins. Methods: Eligibility criteria included chemotherapy followed by surgery (NACTS, arm 1, N=311) with standard
measurable disease, unlimited prior therapies, and any EC histology. Co- concomitant chemoradiotherapy (CCRT, arm 2, N=309) . In arm 1, radical
primary endpoints were confirmed objective response (OR) and progression- hysterectomy was required within 6 weeks after completion of cisplatin-
free survival rate at 6 months (PFS6). Avelumab 10 mg/kg IV was given every based chemotherapy with a cumulative minimum of 225mg/m2, in arm 2,
2 weeks until progression or unacceptable toxicity. In the 1st stage, 16 pts radiation consisted of 45-50 Gy plus boost concurrent with weekly cisplatin
were enrolled in each cohort; if there were $2 ORs or $2 PFS6 responses, chemotherapy (40 mg/m2 per week). Primary endpoint was 5-yrs overall
accrual would continue to the 2nd stage with enrollment of 19 additional pts. survival (OS). Results: Median follow-up time was 8.2 years ( 95% CI = 7.8
Overall, if there are $4 ORs or $8 PFS6 responses, avelumab would be yrs – 8.6 yrs)) and similar between both arms. A total of 191 deaths (31%)
considered worthy of further study in each cohort. Results: As of 12/2018, occurred. Age, stage and histological cell type were balanced in both arms.
33 pts were enrolled. The MSS cohort was closed at the 1st stage due to Protocol treatment was completed in 459 (74%) patients (71% for NACTS;
futility; of 16 pts in the MSS cohort, only 1 pt exhibited an OR and PFS6 82% for CCRT). In arm 1 238 (76%) patients underwent surgery. Main
response [ORR and PFS6 rate 6.25% (95% CI 0.16%-30.2%)]. Conversely, reasons for not having surgery as per protocol, were toxicity (25/74, 34%),
the MSI/POLE cohort reached the primary endpoint of 4 ORs after accrual of progressive disease (18/74, 24%) and insufficient response to NACT (12/
only 17 pts. Two pts in the MSI/POLE cohort did not initiate protocol therapy 74, 16%). Additional radiotherapy was given to 113 patients (36.3%) in arm
and were excluded from all analyses. Of 15 pts in the MSI/POLE cohort, 4 pts 1; additional surgery performed in 9 patients (2.9%) in arm 2. Short term
exhibited OR [1CR+3PRs, OR rate (ORR) 26.7% (95% CI 7.8%-55.1%)] severe adverse events ($G3) occurred more frequently in arm 1 than in arm 2
and 6 pts (including the 4 pts with OR) exhibited PFS6 responses [PFS6 rate (35% vs 21%, p , 0.001). The 5 year OS was 72% in arm 1 and 76% in arm
40.0% (95% CI 16.3%-66.7%)], 4 ongoing and 3 approaching 2 yrs. 2 (not statistically significant, difference = 4.0% (95%CI: -4% - 12%); HR
Twenty-two pts (71%) reported treatment related toxicities, 6 patients 0.87, 95%CI: 0.65-0.15, p=0.332). Conclusions: These preliminary results
(19%) G3 toxicities; there were no treatment-related G4 and G5 toxicities. In revealed no difference in 5-year OS between NACTS and CCRT, indicating
the MSI/POLE cohort, 5 of 6 PFS6 responses were observed in pts with $3 that quality of life and long term toxicity are important to decide optimal
lines of prior therapy (p = 0.011) and in tumors who were PD-L1 negative by treatment. The final results will be available by April 2019, including long-
IHC. Further correlative work will be reported at the meeting. Conclusions: In term toxicity and treatment effect across prognostic factors. Clinical trial
EC pts stratified by MSI/POLE status, MSI vs MSS status appears to be information: NCT00039338.
correlated with avelumab response even in PD-L1 negative tumors. Re-
sponses in the MSI/POLE cohort were more frequent in more heavily pre-
treated patients, a finding that warrants further investigation. Clinical trial
information: NCT02912572.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gynecologic Cancer 315s

5504 Oral Abstract Session, Mon, 1:15 PM-4:15 PM 5505 Oral Abstract Session, Mon, 1:15 PM-4:15 PM
Recurrence rates in cervical cancer patients treated with abdominal versus Combination of niraparib and bevacizumab versus niraparib alone as treat-
minimally invasive radical hysterectomy: A multi-institutional analysis of ment of recurrent platinum-sensitive ovarian cancer: A randomized con-
700 cases. First Author: Shitanshu Uppal, University of Michigan, Ann trolled chemotherapy-free study—NSGO-AVANOVA2/ENGOT-OV24. First
Arbor, MI Author: Mansoor Raza Mirza, NSGO and Rigshospitalet Copenhagen Uni-
versity Hospital, Copenhagen, Denmark
Background: Compare outcomes between open and minimally invasive radical
hysterectomy. Methods: Retrospective multi-institutional review of patients Background: Standard treatment of platinum-sensitive recurrent ovarian cancer
undergoing radical hysterectomy for stage IA1, IA2 and IB1 squamous, adeno- (PSROC) is platinum based combination chemotherapy 6 bevacizumab.
or adeno-squamous carcinoma between 01/01/2010 - 12/31/2017. However, this treatment modality is hardly curative, and is associated with
Results: From 704 cases that met the inclusion criteria, 185 (26.3%) significant toxicity. Both bevacizumab (BEV) and PARP inhibitors (PARPi) have
underwent open and 519 (73.7%) underwent minimally invasive surgery demonstrated efficacy in PSROC. There is preclinical evidence of enhanced
(MIS). Women treated with open surgery were older, had larger tumors on activity of the combination. This is the proof-of-concept randomized trial of
preoperative assessment as well as on final pathology assessment, had PARPi-BEV combination against PARPi monotherapy as treatment in PSROC,
higher proportion of patients with IB1 stage and adjuvant therapy. Patients regardless of number of previous lines of therapies. Methods: In this randomized,
undergoing open surgery had longer median follow-up compared to MIS (44 open-label, phase 2 study, women with measurable/evaluable, high-grade se-
vs. 30.3 months, p , 0.001). The two groups were similar in regard to race rous or endometrioid PSROC were randomized to niraparib 300mg once daily or
distribution, body mass index, comorbidities and preoperative histology. the combination of niraparib 300mg once daily and BEV 15mg/kg IV every
There were 13/185 (7%) recurrences and 10/185 (5.4%) deaths in the open 3 weeks until disease progression (1:1 randomization). The primary endpoint
compared to 42/519 (8.1%) recurrences and 26/519 (5%) deaths in MIS was progression-free survival (PFS). Stratification was according to homologous
(p = n.s for both). However, on multivariate analysis, after controlling for race, recombination-deficiency(HRD) status (MyChoice HRD) and chemotherapy-
comorbidities, preoperative tumor size, histology, grade and smoking status, free-interval (CFI)(6-12months (mo) vs. .12mo). First-line maintenance bev-
MIS had higher odds of recurrence (OR 2.24, 95% CI 1.04 - 4.87, p = 0.04). acizumab was permitted. Results: Of 97 enrolled patients, 48 were randomized
On a second model, in addition to prior mentioned factors, we included to niraparib monotherapy and 49 to the chemotherapy-free combination. The
lymphovascular space invasion, receipt of adjuvant therapy and vaginal combined treatment significantly improved PFS compared to niraparib alone:
margin status. Undergoing MIS remained associated with higher odds of median 11.9 vs. 5.5 mo; hazard ratio (HR) adjusted for stratification factors
recurrence (OR 2.37, 95% CI 1.1 - 5.1, p = 0.031). On sub-group analysis of 0.35; 95% confidence interval (CI),[0.21 to 0.57]; P,0.001. Pre-planned
cases with preoperative tumor size less than equal to 2 cm, there were 5/121 exploratory subgroup analyses: patients with HRD-positive tumors (n=54) HR
(4.1%) recurrence in open and 25/415 (6%) recurrences in MIS group (p = 0.36 (CI, 0.18-0.69); HRD-negative disease (n=43) HR, 0.47 (CI, 0.24-0.95);
0.34). Multivariate analysis did not show a higher rate of recurrence in MIS gBRCAmut patients (n=34) HR 0.53 (CI, 0.23-1.21); non-gBRCAmut patients
arm in this subgroup. In 26 cases of MIS where no vaginal manipulator was (n=63) HR 0.33; CI, 0.18-0.61); CFI of 6 to 12 mo (n=38) HR, 0.29 (CI, 0.14 to
used, no recurrences were noted. In comparison 19/270 (7%) recurrences 0.62); CFI of $12 mo (n=59) HR, 0.42; (CI, 0.22 to 0.80). There was no
were noted in intra-uterine manipulator (V-care/Zumi/Rumi) and 22/210 difference in treatment-emergent grade 3-4 adverse events except for the rate of
(11%) in vaginal manipulators (EEA sizer/Colpo Probe) groups (p = 0.119). hypertension (26.5% vs. 0%) and neutropenia (12.2% vs. 2.1%). Patient-
Conclusions: In this large retrospective analysis, patients undergoing MIS for reported outcomes measured using EORTC QLQ-C30 and OV28 were similar
early stage cervical cancer had higher odds of recurrence. In patients with 2 cm for both treatment arms. Conclusions: Both niraparib alone and the combina-
or less tumor on preoperative assessment, recurrence rates were similar between tion had meaningful activity in PSROC. Compared to niraparib alone, the
the two groups. Role of manipulator in increasing recurrence should be further chemotherapy-free regimen of niraparib and BEV significantly improved PFS in
studied in this patient population. women with PSROC,regardless of HRD status and duration of CFI. Clinical trial
information: NCT02354131.

5506 Oral Abstract Session, Mon, 1:15 PM-4:15 PM 5507 Oral Abstract Session, Mon, 1:15 PM-4:15 PM
Olaparib monotherapy versus (vs) chemotherapy for germline BRCA-mutated Randomized phase II CLIO study on olaparib monotherapy versus chemotherapy
(gBRCAm) platinum-sensitive relapsed ovarian cancer (PSR OC) patients in platinum-resistant ovarian cancer. First Author: Adriaan Vanderstichele,
(pts): Phase III SOLO3 trial. First Author: Richard T. Penson, Massachusetts BGOG & Department of Gynaecology and Obstetrics, University Hospitals
General Hospital, Boston, MA Leuven, Leuven Cancer Institute, Leuven, Belgium
Background: Data from a randomized Phase II trial (NCT00628251) of olaparib Background: The CLIO trial (NCT02822157) evaluated olaparib single-agent
(capsules, 200 or 400 mg bid, n=32 per arm) vs pegylated liposomal doxo- therapy versus standard of care chemotherapy in platinum-resistant (recurrence
rubicin (PLD, n=33) in gBRCAm OC pts with recurrence #12 months after prior within 6 months after last platin) ovarian cancer (PROC). Methods: Eligible
platinum therapy indicated efficacy for olaparib (Kaye et al. JCO 2012). patients with measurable disease and $1 prior line of chemotherapy were
However, the efficacy of PLD was higher than previously reported in this setting. randomized 2:1 to Olaparib (OLA) monotherapy (300 mg tablets, BID) or
We led a confirmatory Phase III, open-label study of olaparib vs non-platinum physician’s choice chemotherapy (CT; PLD 40 mg/m2 q 4 wks; Topotecan
chemotherapy in gBRCAm PSR OC pts (NCT02282020). Methods: Pts were 1.25 mg/m2 day 1—5 q 3 wks; Paclitaxel 80 mg/m2 day 1, 8,15 q 3 wks;
randomized (2:1) to olaparib tablets (300 mg bid) or chemotherapy treatment of Gemcitabine 1000 mg/m2 day 1, 8 and 15 q 4 wks). Primary endpoint was
physician’s choice (TPC) (paclitaxel [P; 80 mg/m2 on day 1 (D1), D8, D15, D22 objective overall response (ORR) per RECIST v1.1. Germline BRCA status was
every 4 weeks (q4w)], topotecan [T; 4 mg/m2 D1, D8, D15 q4w], gemcitabine available for all patients. Disease control rate (DCR) was defined as response for
[G; 1000 mg/m2 D1, D8, D15 q4w] or PLD [50 mg/m2 D1 q4w]) until pro- at least 12 wks. Results: 100 patients with PROC were randomized 2:1 to OLA
gression, stratified by: TPC, prior lines of chemotherapy (2–3 vs $4) and (N = 67) or CT (N = 33). Median prior lines of treatment was 3 (range: 1—8).
platinum-free interval (6–12 vs .12 months). Primary endpoint: ORR (blinded ORR (unconfirmed) was 18% (12/67) for OLA and 6% (2/33) for CT. ORR for
independent central review [BICR]). Secondary endpoints included PFS and OLA was 38% (5/13) in gBRCAm and 13% (7/54) in gBRCAwt patients. Of note,
safety. Results: 266 gBRCAm PSR OC pts were randomized (olaparib, n=178; 2 patients with gBRIP1 mutation had no response under OLA. DCR was 35.8%
TPC, n=88 [PLD, n=47; P, n=20; G, n=13; T, n=8]); 12 in the TPC arm withdrew (24/67) for OLA and 42% (14/33) for CT. DCR under OLA in gBRCAm was 62%
before receiving study treatment. 223 pts (84%) had baseline BICR measurable (8/13) compared to 30% (16/54) in gBRCAwt disease. The median duration of
disease (olaparib, n=151; TPC, n=72). ORR was 72% with olaparib vs 51% with response (DOR) and the median progression-free survival (PFS) was similar:
TPC (OR 2.53, 95% CI 1.40–4.58; P=0.002). HR for PFS by BICR was 0.62 5.4 months vs 4.5 months (DOR) and 2.9 vs 3.4 months (PFS) for OLA and CT,
(95% CI 0.43–0.91; P=0.013; median 13.4 vs 9.2 months [olaparib vs TPC]) respectively. Grade $3 treatment-related AEs occurred in 60% vs 52% for OLA
and by investigator assessment was 0.49 (95% CI 0.35–0.70; P,0.001; and CT, respectively. Somatic HRR mutation analysis is ongoing and will be
median 13.2 vs 8.5 months, respectively). Most common adverse events (AEs) presented. Conclusions: Olaparib monotherapy showed a favorable response
with olaparib were nausea (65% vs 34% [TPC]) and anemia (50% vs 25%) and rate in PROC compared with chemotherapy also in gBRCAwt patients. Analysis
with TPC were palmar-plantar erythrodysesthesia (PPE; 36% vs 1% [olaparib]) of clinical endpoints in relation to HRR is ongoing and will be presented. Clinical
and nausea. Most common grade $3 AEs in either arm were anemia (21% trial information: NCT02822157.
[olaparib] vs 0 [TPC]), PPE (0 vs 12%) and neutropenia (6% vs 11%). For
olaparib vs TPC, serious AEs were reported by 24% vs 18% and AEs led to
treatment discontinuation in 7% vs 20%. Conclusions: Pts with gBRCAm PSR
OC receiving olaparib monotherapy had a significant, clinically relevant im-
provement in ORR and PFS vs TPC, with no new safety signals. Clinical trial
information: NCT02282020.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
316s Gynecologic Cancer

5508 Oral Abstract Session, Mon, 1:15 PM-4:15 PM 5509 Clinical Science Symposium, Sun, 8:00 AM-9:30 AM
EWOC-1: A randomized trial to evaluate the feasibility of three different first- Sex hormone, insulin, and insulin-like growth factor signaling in recurrence of
line chemotherapy regimens for vulnerable elderly women with ovarian cancer high stage endometrial cancer: Results from the NRG Oncology/Gynecologic
(OC): A GCIG-ENGOT-GINECO study. First Author: Claire Falandry, GINECO- Oncology Group 210 trial. First Author: Gloria S. Huang, Yale School of
Centre Hospitalier Lyon Sud, Pierre-Benite, France Medicine, Yale Cancer Center, New Haven, CT
Background: The Geriatric Vulnerability Score (GVS) combining albumin, Background: Sex hormone and insulin/insulin-like growth factor (IGF) axis
lymphocyte count, ADL, IADL and HADS scores has been reported (Falandry C signaling pathways play an important role in endometrial cancer development
Ann Oncol 2013) to identify vulnerable elderly OC patients (pts) as those with a but their role in endometrial cancer recurrence is unknown. In this study GOG-
GVS$3. For such pts, Carboplatin (Cb) monotherapy or weekly Cb plus 8015 we evaluated these pathways in a prospective cohort of patients di-
paclitaxel (Pa) are often proposed as an alternative to Cb-Pa given every agnosed with the most common type of endometrial cancer, endometrioid
3 weeks. Methods: Pts $70 yrs with first line FIGO stage III/IV epithelial OC adenocarcinoma. Methods: Stage II-IV endometrioid endometrial adenocar-
were screened for GVS. Those with GVS$3 were randomized to receive either cinoma patients (N = 816) enrolled in the GOG-210 study with pre-treatment
arm A: Cb AUC5-6 + Pa 175mg/m², d1q3week or arm B: Cb AUC5-6 specimens were tested for tumor mRNA and protein expression levels of IGF1,
d1q3week or arm C:weekly Cb AUC2 + Pa 60mg/m² d1-d8-d15 q4week. IGF2, IGF binding proteins (IGFBP)-1and -3, the insulin (IR) and IGF-I re-
Primary endpoint is treatment feasibility defined as the ability to complete 6 ceptors (IGF1R), and phosphorylated (activated) IR/IGF1R as well as estrogen
chemotherapy courses without disease progression, early treatment stopping (ER) and progesterone receptors (PR) using quantitative PCR and immuno-
due to unacceptable toxicity or death. Inclusion of 240 pts was planned. histochemistry (IHC). Serum concentrations of insulin, IGF-I, IGFBP-3, es-
Results: Among 444 screened pts, 120 were randomized from 12/2013 to 04/ tradiol, estrone and sex hormone binding globulin were measured using
2017 (armA = B = C = 40). Pts characteristics were well balanced between ELISAs. Hazard ratios (HR) and 95% confidence intervals (CI) for risk of
arms A-B-C respectively: median age (79-82-80 yrs), FIGO stage IV (32-37- recurrence were obtained from multivariable Cox proportional hazard’s models
27%), primary surgery (65-72-70%), absence of macroscopic residuals (CC-0) with adjustment for age, stage and grade. Results: Recurrence occurred in 280
(7-5-7%), ECOG$2 (50-50-47%). Feasibility per protocol for arms A-B-C is (34%) cases during a mean of 5.4 years of follow-up. ER-positivity (HR 0.67,
65%, 47% and 60% (p = 0.15). Main reasons for treatment arrest are 95% CI 0.47-0.95), IR-positivity (HR 0.53, 95% CI 0.29-0.98) and serum
treatment toxicity (A:20%; B:15%; C:22.5%; p = 0.771) and disease pro- IGF-I levels (highest versus lowest quartile, HR 0.66, 95% CI 0.47-0.92) were
gression (A:7.5%; B:30%; C:2%; p = 0.004). Median PFS for arm A-B-C are inversely associated with recurrence risk. Conversely, circulating estradiol
12.5 mos (95%CI 10.3-15.3), 4.8 (3.8-15.3) and 8.3 (6.6-15.3), re- (highest versus lowest tertile, HR 1.55, 95% CI 1.02-2.36) and insulin (per 10
spectively (p , 0.001) and median OS for arm A-B-C is not reached (NR) (21, uU/ml, HR 1.52, 95% CI 1.12-2.06) and phosphorylated IGF1R/pIR ex-
NR), 7.4 (5.3-NR) and 17.3 (10.8-NR), respectively (p = 0.001). At the pre- pression (HR 1.40, 95% CI 1.02-1.92) were associated with increased risk of
planned intermediate analysis, the IDMC recommended to prematurely close recurrence. Conclusions: We identified novel sex hormone and insulin/IGF axis
the study as survival in armB was found significantly worse and the number of tissue and circulating biomarkers of recurrence in a prospective study of high
potential pts required to find a significant difference between both Cb-Pa stage endometrioid endometrial cancer. Circulating insulin and estradiol, and
regimens (arms A&C) was out of reach. Conclusions: Compared to 3-weekly tissue phosphorylated (activated) IGR1R/IR were independently associated
and weekly Cb-Pa regimens, Cb single agent was reported to be less active with recurrence. These findings support prioritizing studies to establish their
with significant worse survival outcome in vulnerable elderly pts. In this clinical utility as prognostic biomarkers and to investigate new strategies that
population Cb-Pa combination remains a standard. Clinical trial information: target these pathways for prevention and treatment of endometrial cancer
NCT02001272. recurrence.

5510 Clinical Science Symposium, Sun, 8:00 AM-9:30 AM 5511 Clinical Science Symposium, Sun, 8:00 AM-9:30 AM
Results of a phase 2 trial of ribociclib and letrozole in patients with either A phase II randomized study of avelumab plus entinostat versus avelumab
relapsed estrogen receptor (ER)-positive ovarian cancers or relapsed ER-positive plus placebo in patients (pts) with advanced epithelial ovarian cancer (EOC).
endometrial cancers. First Author: Gerardo Colon-Otero, Mayo Clinic, Jack- First Author: Karen Anne Cadoo, Memorial Sloan Kettering Cancer Center,
sonville, FL New York, NY
Background: Single agent aromatase inhibitor (AI) therapy is associated with Background: Preclinical evidence suggests that combining avelumab (A), a
limited clinical activity in ovarian cancer (OC) and endometrial cancers (EC). AI human anti-PD-L1 monoclonal antibody, and entinostat (E), a class I selective
therapy was associated with a progression free survival (PFS) at 12 weeks of histone deacetylase (HDAC) inhibitor, may increase tumor immunogenicity
only 20% in relapsed OC (Bowman et al, 2002) and a median PFS of 1 month and responsivity to checkpoint inhibition. This study evaluated whether A+E
in relapsed EC (Rose et al, 2000). In Estrogen Receptor (ER) positive met- would lead to improved progression free survival (PFS) vs A in pts with ad-
astatic breast cancer, clinical studies had shown a significant prolongation of vanced EOC. Methods: Pts with EOC which had progressed or recurred after
PFS with the addition of the cyclin kinase 4/6 inhibitor ribociclib to AI 1st-line platinum-based chemotherapy and received 3- 6 lines of therapy
(Hortobagyi et al, 2016). Here, we report the results of a phase 2 clinical trial of were randomized 2:1 to receive A (10 mg/kg IV Q2W) plus E (5 mg PO QW) or A
the combination of ribociclib and letrozole in patients with relapsed ER positive plus placebo (P). Treatment continued until disease progression (PD) or un-
OC or EC. Objectives: Primary endpoint was the proportion of patients with acceptable toxicity. The primary endpoint was PFS (investigator-assessed,
relapsed ER positive OC or EC alive and progression-free after 12 weeks of RECIST 1.1), stratifying on the presence/absence of bulky disease
therapy (PFS12) with the combination of ribociclib given at a dose of 400 mg (tumor $ 50 mm) and platinum-refractory disease. The hypothesis was that
orally daily and letrozole 2.5 mg orally daily. A PFS of 45% was considered a the combination would reduce the hazard of PD or death by 43%,
favorable result based on the data referenced above from Bowman et al. representing a 75% improvement in median PFS. 97 events (from 120 pts)
Methods: Eligibility criteria included patients with relapsed ER positive OC or provided 90% power with 1-sided significance level of 0.10. Secondary
EC, with measurable disease, not previously treated with ribociclib or AIs. endpoints included ORR, duration and time to response, toxicity, clinical
Xenografts were created from CT guided tumor biopsies at baseline to assess benefit rate, and OS. Results: 126 pts were enrolled, median age 63 yrs
feasibility. Results: A total of 40 patients were enrolled (20 with OC and 20 (range 43-82), median 4 prior lines, 83% serous histology. Median PFS was
with EC) ) with a median age of 61 years (range: 30-82) and 64.5 (range: 52- 1.64 and 1.51 mos for A+E and A+P, respectively (p = 0.31; HR 0.90, 95%
75) in the OC and EC groups respectively. Among the OC patients, 17 had high CI: 0.58-1.39). No significant differences in ORR (6% vs 5%), or OS (NE vs
grade serous carcinomas and 3 had low grade serous carcinomas. 11 EC 11.3 mos) were observed. 4 pts (3%) had clear cell EOC, with no responses
patients had endometrioid cancers (3 with grade 1 tumors) and 9 had high observed. The incidence of related adverse events (AEs) was higher in the
grade serous tumors. Ten out of 20 OC patients and 11/20 EC patients were A+E arm compared to A+P (any grade: 93% vs 78%, Grade 3/4: 41% vs 10%),
alive and progression-free at 12 weeks (PFS12 of 50 and 55%, respectively). and the most frequent ($20%) related AEs with A+E were fatigue (46%),
The most common grade 3 or higher adverse events (occurring in at least 5 pts) nausea (31%), diarrhea (26%), anemia (26%), and chills (20%). Grade 3/4
were leukopenia (18%), lymphopenia (18%), neutropenia (13%), and fatigue related AEs occurring in $5% with A+E were fatigue (9%), and neutropenia
(13%). 34 tumor biopsies were suitable for injection into mice and 44% (8%). 47% of pts in A+E arm required E dose holds/reductions. Discontin-
engrafted. ER expression persisted through multiple passages in mice. Two of uations due to AEs were similar between arms (21% vs 17.5% for A+E and
three EC PDX models exhibited improved PFS with letrozole/ribociclib com- A+P, respectively), as was duration of study therapy (median 4 and 5 cycles
pared to letrozole alone. Conclusions: The combination of ribociclib and started). Conclusions: In pts with heavily pretreated EOC, median PFS
letrozole is associated with a promising 50% and 55% PFS12 in patients with was not prolonged when E was added to A compared to A alone and
ER positive relapsed OC or EC respectively. Creation of xenograft tumor models the combination resulted in greater toxicity. Clinical trial information:
from CT guided biopsies of OC and EC tumors was feasible. Clinical trial NCT02915523.
information: NCT02657928.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gynecologic Cancer 317s

5512 Poster Discussion Session; Displayed in Poster Session (Board #335), 5513 Poster Discussion Session; Displayed in Poster Session (Board #336),
Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Sat, 4:30 PM-6:00 PM Sat, 4:30 PM-6:00 PM
Impact of adding nintedanib to neoadjuvant chemotherapy (NACT) for Adavosertib with chemotherapy (CT) in patients (pts) with platinum-resistant
advanced epithelial ovarian cancer (EOC) patients: The CHIVA double- ovarian cancer (PPROC): An open label, four-arm, phase II study. First
blind randomized phase II GINECO study. First Author: Gwenael Ferron, Author: Kathleen N. Moore, Stephenson Cancer Center at the University of
GINECO and Institut Claudius Regaud, Toulouse, France Oklahoma HSC and Sarah Cannon Research Institute, Oklahoma City, OK
Background: Nintedanib, an oral inhibitor of VEGF-FGF-PDGF receptors, has Background: Adavosertib (AZD1775; A), a highly selective WEE1 inhibitor, demonstrated activity
been shown to prolong progression-free survival (PFS) when added to adjuvant and tolerability in combination with carboplatin (C) in primary PROC. This study (NCT02272790)
assessed the objective response rate (ORR) and safety of A in PROC. Methods: Pts with recurrent
chemotherapy after primary surgery (duBois A, Lancet Oncol 2015). CHIVA trial RECIST v1.1 measurable PROC received A with C, gemcitabine (G), weekly paclitaxel (P), or
explored the role of nintedanib in combination with NACT. Methods: Patients pegylated liposomal doxorubicin (PLD) in 3- (C) or 4-week (G, P, PLD) cycles (Table). Tumor
(pts) with FIGO stage IIIC-IV chemotherapy-naive AEOC considered as unre- assessments were performed every 2 cycles until disease progression. Primary objective: ORR;
sectable after laparoscopic evaluation were randomized (2:1) to be treated with 3 other objectives: disease control rate (DCR), progression-free survival (PFS) and safety. Results: In
to 4 cycles (cy) of carboplatin (AUC 5 mg/mL/min) and paclitaxel (175 mg/m²) the 94 pts treated (median treatment duration 3 months; range 0–16 months), outcomes were
greatest with A (weeks [W]1–3) + C (Table), with ORR of 67% and median PFS (mPFS) of
(CP) before interval debulking surgery (IDS) followed by 2 to 3 cy of CP for a total
10.1 months for this cohort. Most common grade $3 treatment-emergent adverse events (TEAEs)
of 6 cy, plus either 200 mg of Nin (armA) or placebo (armB) twice daily on days are shown in the Table, with hematologic toxicity most notable with A (W1–3) + C. TEAEs led to A
2–21 q3week at cy 1&2, 5&6 and maintenance therapy for up to 2 years. The dose interruptions, reductions and discontinuations in 63%, 30% and 13% of the whole cohort,
primary endpoint was PFS. Results: Between Jan. 2013 and May 2015, 188 pts respectively. A possible positive relationship between CCNE1 amplification and response warrants
were included (124 arm A, 64 arm B) with a median Peritoneal Cancer Index of further investigation. Conclusions: A shows preliminary efficacy when combined with CT. Pts
22 (range 19-27). Pts characteristics were well balanced between both arms. receiving A (W1–3) + C showed greatest benefit. The increased but not unexpected hematologic
toxicity is a challenge and could be further studied to optimize the dose schedule and supportive
Median PFS was 14.4 mos (95%CI 12.2-15.4) and 16.8 (13.3-21.4) in arm A medications. Clinical trial information: NCT02272790.
and B respectively (HR:1.50, p=0.02). Median OS was 37.7 mos (29.8-41.0)
and 44.1 (32.7-not reached) in arm A and B respectively (HR:1.54, p=0.053). 225 mg BID 225 mg
Arm A was associated with more toxicity compared to arm B respectively (Grade 175 mg (2.5 Ds, BID (2.5 225 mg BID 175 mg BID 225 mg BID
daily (2 Ds, W1–3) + Ds, W1) + (2.5 Ds, (2.5 Ds, (2.5 Ds,
3&4 adverse events: 92 versus 71%), with increased early treatment discon- Cohort W1–3) + G† P*,‡ C§ W1–3) + C§ W1) + PLDk W1) + PLDk
tinuation before the 3rd cy (14.5 vs 6.2%) & CP dose reduction (12% vs 0%). Pts, n 9 38 23 12 6 6
Pts in Arm A reported inferior RECIST ORR to pre-IDS therapy compared to Arm ORR, n (%) 1 (11) 7 (18) 7 (30) 8 (67) 2 (33) 1 (17)
B (35.1 vs 55.9%). IDS was performed significantly less frequently in arm A DCR, n (%) 3 (33) 25 (66) 19 (83) 12 (100) 3 (50) 5 (83)
(58.1%) vs arm B (76.6%). However among pts who underwent IDS, complete mPFS, months 1.7 5.3 4.2 10.1 2.7 NC
TEAE grade ‡3,¶
surgical cytoreduction rate (76%) and peri/postoperative complication rate n (%)
(11.2%) were similar in both arms. Conclusions: The addition of nintedanib to Neutropenia/ 7 (78) 20 (53) 5 (22) 9 (75) 0 2 (33)
NACT increases toxicity and compromise chemotherapy efficacy leading to a neutropenia
decreased
reduced rate of IDS and worse PFS and OS for advanced EOC patient. Clinical Thrombocytopenia/ 3 (33) 4 (11) 12 (52) 9 (75) 0 1 (17)
trial information: 2011-006288-23. platelets
decreased
Anemia/Hb 1 (11) 8 (21) 13 (57) 6 (50) 0 0
decreased
*Data from an interim analysis; †800 mg/m2 IV D1, 8 and 15; ‡80 mg/m2 D1, 8 and 15; §AUC5
D1; k40 mg/m2 D1; ¶Affecting $20% of all pts. D, day

5514 Poster Discussion Session; Displayed in Poster Session (Board #337), 5515 Poster Discussion Session; Displayed in Poster Session (Board #338),
Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Sat, 4:30 PM-6:00 PM Sat, 4:30 PM-6:00 PM
Dose-dense early postoperative intraperitoneal chemotherapy in ovarian The prognosis of patients with recurrent or persistent ovarian clear cell
cancer: Randomized, phase II trial. First Author: Rongyu Zang, Ovarian carcinoma: Results from a randomized phase III study (JGOG3017/GCIG).
Cancer Program, Division of Gynecology Oncology, Department of Obstetrics First Author: Eiji Kondo, Department of Obstetrics and Gynecology, Mie
and Gynecology, Zhongshan Hospital, Fudan University; Department of University School of Medicine, Mie, Japan
Oncology, Shanghai Medical College, Fudan University, Shanghai, China
Background: There are currently no concise data on prognosis in patients with
Background: Dose-dense early postoperative intraperitoneal chemo (DD-EPIC) had been carried recurrent or persistent clear cell carcinoma (CCC). The JGOG3017/GCIG was a
out in advanced ovarian cancer (OC) pts in China over the past three decades but it was not randomized phase III trial to compare paclitaxel plus carboplatin (TC) to CPT-
proved by a prospective study. This trial was designed to confirm the benefit of DD-EPIC in
delaying progression and improving survival. Methods: In a multicenter, phase 2 trial, pts with
11 plus cisplatin (CPT-P) for stage I to IV CCC patients who have been di-
FIGO IIIC-IV OC and optimal debulking surgery (residual disease #1cm) were randomly allo- agnosed by central pathological review (CPR). A total of 619 patients were
cated to receive 4 doses of weekly DD-EPIC with cisplatin (50mg/m2) and etoposide (100mg/ eligible for efficacy analyses. The aim of this study was to investigate prognosis
m2) followed by 6 cycles of intravenous (iv) chemo with carboplatin and taxane every 3 weeks of these patients with recurrent or persistent CCC. Methods: We estimated
(DD-EPIC group), or standard iv chemo alone (iv group). (ClinicalTrials.gov, NCT01669226). median and restricted mean survival time (RMST) of post-progression survival
Results: Between 2009 and 2015, 218 pts were randomized, of whom 215 initiated treatment
(PPS) of patients with recurrent or persistent CCC by platinum sensitivity,
(106 to DD-EPIC and 109 to iv; for efficacy analyses). Totally, 36 pts (16$7%) were received
neoadjuvant chemo. With a median of 61$9 mos follow-up, 122 pts died (54 in DD-EPIC and 68 treatment arm, crossover chemotherapy, primary stage, residual tumor, per-
in iv group). Remarkable OS benefit of DD-EPIC was recorded (67$5 mos for DD-EPIC vs. 46$3 formance status and ethnicity. PPS rates at 6, 12, 18 and 24-month were also
mos for iv; HR 0$70, 95% CI 0$49-1$00, P=0$047). Pts in DD-EPIC had a significantly in- calculated. Results: Among the 619 patients, the recurrence rate of stages was
creased median PFS compared with those in iv group (21$7 vs. 16$8 mos; HR 0$64, 95% CI as follows: 6.3% (6/96) in stage IA/IB, 14.6% (46/315) in stage IC, and
0$47-0$86, P=0$003). Median TFST was 25$1 vs. 18$0 mos in favor of DD-EPIC (HR 0$62, 54.8% (114/208) in stage II- Ⅳ. The recurrence rate of surgical situations was
95%CI 0$46-0$83, P=0$002). Similar findings were detected in TSST (42$2 vs. 29$3 mos; HR
0$66, 95%CI 0$47-0$94, P=0$019). Grade 3 and 4 Leucopenia (53$8% vs. 35$2%), anemia
as follows: 19.4% (106/544) in complete surgery, 75% (27/36) in optimal,
(23$6% vs. 5$6%) and gastrointestinal events (10$4% vs. 1$9%) were more common in DD- and 84.6% (33/39) in suboptimal. Overall, 166 of 619 patients had recurrent
EPIC (P=0$006, P,0$001 and P=0$010, respectively). Ninety-one pts were detected by disease. The median PPS were 14.0 months (95% confidence interval [CI],
gBRCA testing, with 25$3% of cases carrying deleterious BRCAm, but PFS and OS benefit were 12.6 – 17.9) for all patients, 13.5 months (95% CI, 11.4 – 19.6) in the TC
observed in patients with BRCA-wild type (HR 0$46 and 0$55, 95%CI 0$27-0$81 and 0$27- group (n=77) and 14.4 months (95% CI, 11.0 - 18.8) in the CPT-T group
1$11, respectively). Conclusions: DD-EPIC with a higher completion rate and acceptable (n=89), with no significant difference between the two groups (hazard ratio,
treatment burden was associated with longer OS than standard iv alone. Owing to the benefit of
relatively long-term OS, DD-EPIC may be considered as a valuable option for OC, particularly in 1.02; 95% CI, 0.71 - 1.47, log-rank P = 0.898). The RMST of PPS for all
developing countries and BRCA-wild type pts. Clinical trial information: NCT01669226. patients was 14.6 months (95% CI, 13.3 - 15.8). Median PPS for patients with
platinum-resistant (44.6%, n=74) and platinum-sensitive (53.0%, n=88)
Median OS (mos)
disease were 10.9 months (95% CI, 8.9 - 13.3) and 18.8 months (95% CI,
Events/Pts HR (95% CI) DD-EPIC IV
15.0 - 28.7) (HR, 1.88; 95% CI, 1.30-2.72, p,0.001), respectively.
All 113/215 0.70 (0.49-1.00) 67.5 (57.0-78.1) 46.3 (35.1-57.5) Conclusions: The median PPS of patients with platinum-resistant recurrence
Age
<56yrs 62/107 0.54 (0.32-0.89) 76.3 (62.2-90.4) 48.4 (32.6-64.2) was significantly shorter than that of patients with platinum-sensitive re-
Histology currence. Data on PPS in patients with recurrent CCC that will be the basis of a
Serous 111/200 0.69 (0.47-1.00) 70.2 (58.9-81.5) 46.3 (34.2-58.3)
Residual disease future clinical trial in such patients were obtained.
0.1- 1 cm 85/137 0.62 (0.40-0.96) 62.8 (44.8-80.9) 39.0 (27.8-50.1)
gBRCA mut
No 33/68 0.55 (0.27-1.11) 75.2 (60.1-90.4) 52.9 (36.6-69.1)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
318s Gynecologic Cancer

5516 Poster Discussion Session; Displayed in Poster Session (Board #339), 5517 Poster Discussion Session; Displayed in Poster Session (Board #340),
Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Sat, 4:30 PM-6:00 PM Sat, 4:30 PM-6:00 PM
Surveillance in stage I MOGCTs (malignant ovarian germ cell tumors): A A comparison of adjuvant therapy approaches for patients with early-stage
MITO prospective study (multicenter Italian trials in ovarian cancer). First uterine serous carcinoma. First Author: Katherine Kurnit, The University of
Author: Alice Bergamini, Università Vita-salute San Raffaele, Milan, Italy Texas MD Anderson Cancer Center, Houston, TX
Background: The standard of treatment of stage I MOGCTs is surgery followed Background: Uterine serous carcinoma is a less common subtype of endo-
by BEP (bleomycin + etoposide + cisplatin) chemotherapy, except for stage IA metrial cancer that is associated with poorer survival. The optimal post-operative
dysgerminoma (D) and IAG1 immature teratoma (IT). Surveillance has adjuvant treatment strategy for these patients remains uncertain. Methods: This
emerged as a possible option to avoid adjuvant chemotherapy in IB-C1 D, IA-C multi-institutional, retrospective cohort study evaluated patients with early stage
G2 – G3 IT, and in stage IA mixed and yolk sac tumors (YST), after com- uterine serous carcinoma. Patients with FIGO Stage IA-II disease after surgery,
prehensive surgical staging (CSS) with negative postoperative markers. The whose tumors had serous or mixed serous/non-serous histology were included.
aim of this study was to analyze oncological outcome of stage I MOGCT patients Patients with carcinosarcoma were excluded. Clinical data were abstracted from
included in the MITO9 study. Methods: MITO9 was a prospective observa- local medical records. Summary statistics, Fisher’s exact, and Kruskal-Wallis
tional study analyzing data collected between 2013 and 2018. 41 patients tests were used to analyze demographic and clinical characteristics. Univariable
with stage I conservatively treated MOGCTs were included. Three groups were and multivariable analyses were performed for recurrence-free survival (RFS)
identified: group A. IA D and IAG1 IT candidate to surveillance according to and overall survival (OS). Results: 634 patients were included. 77% of patients
guidelines; group B. stages IB-C1 D, stage IA-C G2-G3 IT, stage IA mixed and had Stage IA disease, 42% showed no myometrial invasion. The majority had
YST were consulted about the option of close surveillance vs adjuvant che- pure serous histology (72%) and LVSI (76%). Adjuvant treatment varied: 12%
motherapy in case of CSS; group C. all other patients receiving BEP. received no adjuvant therapy, 7% had chemotherapy alone, 51% had cuff
Results: Median age was 25.6 years (range 14-40). Median follow up was brachytherapy, 12% had cuff brachytherapy with chemotherapy (cuff/chemo),
36,4 months. Group A included 12 patients, 5 IA G1 IT and 7 IA D. Group B and 19% underwent pelvic radiation (EBRT). Complete RFS and OS data were
included 24 patients. Of these, 2 out of 5 patients (40%) were positive at available for 607 and 609 patients, respectively, and the median follow-up time
restaging and were excluded from surveillance protocol. Seven of the 22 was 58 months. As compared with patients who received no adjuvant therapy,
remaining patients (31.8%) received chemotherapy, while 15 (68.1%) were patients who received cuff or cuff/chemo had improved RFS (cuff: HR 0.70, p =
enrolled in the surveillance protocol. Out of these 15 patients, 4 were stage IC 0.02; cuff/chemo HR 0.53, p = 0.01) and OS (cuff HR 0.56, p = 0.001; cuff/
D (one IC1, one IC2 and two IC3), 2 were mixed stage IA with YST tumor, 9 chemo HR 0.48, p = 0.01). In a direct comparison, patients with cuff/chemo
were G3 IT (four IA, three IC2, one IC3 and one IB). The 7 patients receiving had better RFS and OS than those with chemotherapy alone (RFS HR 0.52, p =
chemotherapy were: 1 dysgerminoma IC2, 2 YST IA, 3 IT G3 (one IA and one 0.03; OS HR 0.50, p = 0.05). There were no differences in RFS or OS for women
IC2) and 1 mixed IA tumour. Group C included 5 patients, three IC YST and two who received chemotherapy alone or EBRT. Improved survival with cuff and cuff/
mixed IC2 with YST. Survival of these patients was 100%, while disease free chemo persisted on multivariable analyses (included age, stage, LVSI, adjuvant
survival was 97.5%. Only one patient in C Group, a stage IA G3 IT treated with therapy type); additionally, EBRT was also associated with improved OS. In
adjuvant BEP, relapsed as mature teratoma. None of the patients in the analyses limited to patients without myometrial invasion, patients with cuff or
surveillance protocol experienced relapse. Conclusions: These data suggest cuff/chemo had improved RFS and OS compared with observation alone.
that close surveillance could be an alternative option to avoid adjuvant che- Conclusions: The use of adjuvant cuff brachytherapy with and without che-
motherapy in properly staged IB-C1 D, stage IA G2 – G3 IT, stage IA mixed and motherapy was associated with improved RFS and OS in patients with early stage
YST. These findings deserve further confirmation in an international co- uterine serous carcinoma.
operative setting.

5518 Poster Discussion Session; Displayed in Poster Session (Board #341), 5519 Poster Discussion Session; Displayed in Poster Session (Board #342),
Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Sat, 4:30 PM-6:00 PM Sat, 4:30 PM-6:00 PM
A randomized double-blind placebo-controlled phase II trial comparing Pembrolizumab with low dose carboplatin for recurrent platinum resistant
gemcitabine monotherapy to gemcitabine in combination with adavosertib ovarian, fallopian tube, and primary peritoneal cancer-interim results. First
in women with recurrent, platinum resistant epithelial ovarian cancer: A trial Author: John B. Liao, University of Washington, Seattle, WA
of the Princess Margaret, California, Chicago and Mayo Phase II Consortia.
Background: Pembrolizumab has shown activity in advanced recurrent ovarian
First Author: Stephanie Lheureux, University Health Network, Princess
cancer (AOC) with an 8% response rate and median progression-free survival
Margaret Cancer Centre, Toronto, ON, Canada
(PFS) of 2.1 months reported in KEYNOTE-100. Because platinum chemo-
Background: Platinum resistant ovarian cancer (OC) remains a therapeutic chal- therapies also induce T cell proliferation and enhance tumor cell recognition
lenge. High grade serous OC (HGSOC) harbors TP53 mutations leading to increased through PD-1/PD-L, we assessed the safety and activity of pembrolizumab with
dependency on S- and G2-phase checkpoints. Wee1 inhibition with Adavosertib (AZD carboplatin in platinum resistant AOC. Methods: Key eligibility criteria for this
1775) (A) induces G2 checkpoint escape. Gemcitabine (G) is an antimetabolite Phase 1/2 single arm trial were platinum resistant AOC, fallopian tube, or
therapy and blocks the progression of cells through the G1/S phase. We hypothesized peritoneal cancer, progression after subsequent systemic therapy, and ECOG PS
that combining G+A would be synergistic and overcome resistance. Methods: We 0-1. Pembrolizumab 200mg was given on Day 1 and carboplatin AUC 2 on Day 8
conducted a multicentre double-blind 2:1 randomized phase 2 trial to assess the and 15 of a 3 week cycle until progression, unacceptable toxicity, or consent
progression free survival (PFS) in women with recurrent platinum-resistant/refractory withdrawal. Imaging was done before cycles 4 and 8, then every 3 months and
HGSOC receiving G+A or G+placebo (P) (NCT02151292). Eligibility required
unconfirmed objective response assessed by blinded independent review per
measurable disease and feasibility of paired tumor biopsies; no limitation in prior lines
RECIST 1.1. Adverse events (AEs) were reported per Common Terminology for
of therapy. Non HGSOC histologic subtypes were enrolled in a separate non-
Adverse Events v5.0. PD-L1 expression was assessed by immunohistochemistry.
randomized exploratory cohort. A/P was given orally at 175mg QD on D1-2, D8-9
and D15-16 with G 1000mg/m² IV D1, D8 and D15 in a 28-day cycle until pro- Results: 27 patients (median age: 64) had received a median of 5 (range: 2-9)
gression or unacceptable AE. Tumor staging was scheduled every 8 weeks. TP53 prior lines of systemic therapy, which included bevacizumab in 74% of patients.
mutations were analyzed on archival tissue with Sanger sequencing, TAm-Seq and The most common treatment related (TR) AEs were lymphopenia (18%) and
IHC. TP53 mutation will be also assessed in circulating tumor DNA (ctDNA). Whole anemia (9%). The majority of TR AEs were grade 1 or 2 (93%). 6% of AEs were
exome and RNA sequencing were performed on paired tumor tissues. Results: 124 grade 3 with lymphopenia the most common. Two grade 4 AEs were neutropenia
patients (pts) with median of 3 prior lines of therapy (range 1-10) from 12 centres and lymphopenia. Of 23 patients evaluable for best objective response, 13.0%
across Canada and US were enrolled between Sep 2014 to May 2018, with 99 pts (95% CI, 2.7-33.6) had partial response (PR), 65.2% (95% CI, 42.7-83.6) had
randomized (65 in Arm G+A and 34 in G+P). 5 pts were ineligible; 64 pts have died. stable disease (SD), and 21.7% (95% CI, 7.4-43.7) had progression. 7 of the 23
The median follow-up was 14.3 months. Main related AE was hematologic toxicity evaluable patients (30.4%) had archival tumor with modified percent
(Anemia G$3: 31% in G+A vs 18% in G+P; Thrombocytopenia G$3: 31% vs 6%; scoring $5 for PD-L1 and all achieved PR (3/7, 42.8%) or SD (4/7, 57.2%).
Neutropenia G$3: 62% vs 30%). PFS was significantly improved from 3.0 to Overall median PFS was 4.6 months (95% CI, 2.7-6.2). Rate of PFS at 6 months
4.6 months (HR 0.56 (95%CI: 0.35-0.90, p=0.015 Log rank). There was a sig- was 40.4% (95% CI, 25.5-65.5). Median follow-up is 6.2 months and PFS is
nificant improvement in overall survival (OS) from 7.2 to 11.5 months (HR 0.56 based on current data, but 8 patients remain on study and estimates will be
(95%CI: 0.34-0.92, P=0.022). Partial response by RECIST 1.1 was observed in 13 updated. Conclusions: Pembrolizumab with low dose carboplatin was well
(21%) and 1 (3%) pts for Arms G+A and G+P, respectively (p=0.02). From the 25 pts tolerated and showed activity in heavily pretreated platinum resistant AOC.
in the exploratory cohort, 3 (12%) partial responses were observed. Final results will Survival and biomarker analyses are ongoing. Clinical trial information:
be reported at the meeting. Conclusion: Addition of adavosertib to gemcitabine in NCT03029598.
women with platinum resistant/refractory OC improved response rate, PFS and OS
with manageable toxicity. Clinical trial information: NCT02151292.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gynecologic Cancer 319s

5520 Poster Discussion Session; Displayed in Poster Session (Board #343), 5521 Poster Discussion Session; Displayed in Poster Session (Board #344),
Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Sat, 4:30 PM-6:00 PM Sat, 4:30 PM-6:00 PM
Mirvetuximab soravtansine, a folate receptor alpha (FRa)-targeting antibody- Evolve: A post PARP inhibitor clinical translational phase II trial of cediranib-
drug conjugate (ADC), in combination with bevacizumab in patients (pts) with olaparib in ovarian cancer—A Princess Margaret Consortium – GCIG Phase II
platinum-resistant ovarian cancer: Final findings from the FORWARD II study. Trial. First Author: Stephanie Lheureux, University Health Network, Princess
First Author: David M. O’Malley, The Ohio State University, Columbus, OH Margaret Cancer Centre, Toronto, ON, Canada
Background: Mirvetuximab soravtansine is an ADC comprising a FRa-binding Background: PARP inhibitors (PARPi) are approved therapies in high grade
antibody, cleavable linker, and the maytansinoid DM4, a potent tubulin- serous ovarian cancer (HGSOC). There are few studies after PARPi progression
targeting agent. As part of the Phase 1b FORWARD II trial (NCT02606305), and correlation with dynamic changes in resistance. We hypothesized that PARPi
the combination of mirvetuximab soravtansine with bevacizumab (BEV) was resistance could be overcome by adding an anti-angiogenic. Methods: We report
evaluated in pts with FRa-positive, platinum-resistant ovarian cancer (re- the first phase 2 trial assessing the combination of olaparib and cediranib after
currence within 6 months after last platinum). Methods: Pts received mirve- PARPi failure in HGSOC. This investigator initiated study included three cohorts
tuximab soravtansine (6 mg/kg; adjusted ideal body weight) and BEV (15 mg/kg) of 10 evaluable patients (pts): i) platinum sensitive post PARPi (PS), ii) platinum
on Day 1 of a 21-day cycle. Responses were assessed according to RECIST 1.1 resistant post PARPi (PR) and iii) exploratory cohort of pts re-challenged with
and adverse events (AEs) evaluated by CTCAE v4.03. Results: In total, 66 pts chemotherapy post PARPi progression (PE) (NCT02681237). The primary
received combination dosing at this level: 11 during escalation and 55 in ex- objective was to determine objective response rate by RECIST v1.1 and pro-
pansion. The median age was 63 years, pts received a median of 3 prior lines of gression free survival (PFS) at 16 weeks. Secondary objectives were to evaluate
systemic therapy (range 1-8), and 62% had received prior therapy with BEV. The safety, PFS, overall survival (OS) and mechanisms of PARPi resistance. Pts who
most common AEs were diarrhea (58%), nausea (50%), and blurred vision had radiographic progression on any PARPi were eligible. Archival tumor at initial
(48%), and were primarily low grade (# grade 2). Serious AEs were largely diagnosis and baseline tumor biopsy at PARPi progression were mandatory. Pts
gastrointestinal in nature, with small intestinal obstruction the most frequent received olaparib tablets 150mg BID with cediranib 20mg QD until progression
individual event (4 pts, 6%). Objective responses were seen in 27 pts for a or unacceptable toxicity. CT scans were performed every 8 weeks. Whole exome
confirmed overall response rate (ORR) of 41% (95% CI, 29, 54), median and RNA sequencing were performed on paired tumors tissues. Results: Thirty-
progression-free survival (mPFS) interval of 7.1 months (95% CI, 4.9, 9.5), and four pts were enrolled. BRCA1/2 mutations were found in 9/11 PS, 8/10 PR and
median duration of response (mDOR) of 8.6 months (95% CI, 4.9, 14.9). In a 7/13 PE pts. By RECIST1.1, four partial responses were observed (2 in PR and 2
subset analysis of pts (n = 16) who were bevacizumab-naı̈ve, had 1-2 prior in PE cohorts) and 18 stable disease. The 162week PFS was 54.5%
therapies, and medium/high FRa levels (i.e., $ 50% of cells with at least (31.8293.6) in PS, 50% (26.9292.9) in PR and 36% (15.6282.8) in PE,
moderate staining intensity) the ORR was 56% (95% CI, 30, 80), mPFS respectively. OS at 1 year was 81.8% (61.92100) in PS, 64.8% (39.32100) in
9.9 months (95% CI, 4.1, 15.9), and mDOR 12 months (95% CI, 6.0, 14.9). PR and 39.1% (14.72100) in PE. Main related adverse events were anemia,
Conclusions: The combination of mirvetuximab soravtansine with BEV exhibits hypertension, diarrhea and fatigue, grade 3 , 10%. Molecular analyses iden-
favorable tolerability in pts with platinum-resistant ovarian cancer, characterized tified different mechanisms of PARPi resistance in ~77% of evaluable pts with
by a manageable side-effect profile. The encouraging efficacy compares fa- matched pre-post PARPi progression biopsies such as reversion mutations in
vorably to reported outcomes for BEV and chemotherapy seen in similar patient BRCA1/2 and other homologous repair (HR) genes; BRCA, HR and MDR
populations. These data support continued exploration of the combination in upregulation, CCNE amplification and RIG-I like receptor downregulation.
ovarian cancer. Clinical trial information: NCT02606305. Conclusions: Treatment with olaparib-cediranib after PARPi failure was feasible
and met the predefined bar for efficacy in each cohort. This is the largest clinical
trial prospectively evaluating PARPi failure and correlating tissue genomic
mechanisms of resistance. Clinical trial information: NCT02681237.

5522 Poster Discussion Session; Displayed in Poster Session (Board #345), 5523 Poster Discussion Session; Displayed in Poster Session (Board #346),
Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Sat, 4:30 PM-6:00 PM Sat, 4:30 PM-6:00 PM
Exploratory analysis of the effect of maintenance rucaparib on postprogression A phase I study of veliparib incorporated into front-line platinum based
outcomes in patients (pts) with platinum-sensitive recurrent ovarian carcinoma cheotherpy and bevacizumab in epithelial ovarian cancer (NCT00989651):
(OC) and updated safety data from the phase 3 study ARIEL3. First Author: A GOG/nrg trial. First Author: Deborah Kay Armstrong, The Sidney Kimmel
Robert L. Coleman, The University of Texas MD Anderson Cancer Center, Comprehensive Cancer Center at Johns Hopkins University, Baltimore, MD
Houston, TX
Background: Veliparib, a poly-(ADP-ribose)-polymerase inhibitor, increases
Background: In ARIEL3, rucaparib maintenance treatment significantly improved anti-tumor activity when combined with platinum chemotherapy and has
progression-free survival (PFS) vs placebo in all predefined, nested cohorts: BRCA monotherapy activity in BRCA deficient tumors. This study was done to de-
mutation; BRCA mutation + wild-type BRCA/high loss of heterozygosity (LOH); and termine the recommended phase II dose (RP2D) of veliparib in combination
intent-to-treat (ITT) population. Methods: Pts were randomized 2:1 to receive oral with front line treatment for epithelial ovarian cancer (EOC). Methods: Eligible
rucaparib 600 mg BID or placebo. Exploratory endpoints of time to first subsequent patients had newly diagnosed, stage II-IV EOC. Six regimens were evaluated, 3
therapy (TFST), time to investigator-assessed PFS on the subsequent line of treatment or variations of chemo delivery with either continuous (D1-21) or intermittent
death (PFS2), and time to second subsequent therapy (TSST) were assessed in the
(days-2-5) veliparib BID. Chemo included 1: IV q3week carboplatin (C) (AUC
predefined cohorts. Results: Exploratory efficacy endpoint data are given in the Table. As
6) and paclitaxel(T) (175mg/m2); 2, IV q3week C (AUC 6) and weekly T(80mg/
of Dec 31, 2017, the most common treatment-emergent adverse events (TEAEs) of any
grade (rucaparib vs placebo) were nausea (75.8% vs 36.5%), asthenia/fatigue (70.7% vs
m2); and 3, IV T (135mg/m2, day 1), IP cisplatin (75mg/m2, day 1 or 2) and IP
44.4%), dysgeusia (39.8% vs 6.9%), and anemia/decreased hemoglobin (39.0% vs T (60mg/m2, day 8). Bevacizumab 15mg/kg started cycle 2 and continued as
5.3%). The most common grade $3 TEAEs were anemia/decreased hemoglobin (21.5% monotherapy cycles 7-22. A 3+3 dose escalation design evaluated dose-
vs 0.5%) and alanine/aspartate aminotransferase increase (10.2% vs 0.0%). limiting toxicities (DLTs) in cycles 1 and 2. Once , 2/6 patients experienced a
Conclusions: Rucaparib significantly improved the clinically meaningful endpoints TFST, DLT, that dose level was expanded to evaluate feasibility over 4 cycles.
PFS2, and TSST vs placebo in all predefined cohorts of pts with platinum-sensitive, Results: The study accrued 424 treated patients. For regimen 1, continuous
recurrent OC. The updated safety profile was consistent with prior reports. Clinical trial (Reg1c) the maximum tolerated dose (MTD) was 250mg veliparib BID but the
information: NCT01968213. feasible dose was found to be 150mg BID. For regimen 1, intermittent (Reg1i)
BRCA mutant or
the MTD and feasible dose were 400 and 250mg BID respectively. For Reg2c
BRCA mutant BRCA wild type/LOH high ITT the MTD and feasible dose were the same at 150mg BID. For Reg2i the MTD and
Placebo Placebo Placebo feasible dose were 250 and 150mg BID respectively. For Reg3c the MTD and
Rucaparib (n=130) (n=66) Rucaparib (n=236) (n=118) Rucaparib (n=375) (n=189) feasible dose are both 150mg BID and for Reg3i, the MTD was 400mg BID
TFST and the feasible dose felt to be 300mg BID. Median PFS by residual disease
Median, mo 19.0 7.2 16.4 7.6 12.5 7.4
HR (95% CI); P value 0.29 (0.20–0.42); 0.40 (0.30–0.52); 0.43 (0.35–0.53); and BRCA status is: (Positive residual disease) 14.6, 19.1 and 16.9 months
P,0.0001 P,0.0001 P,0.0001 for BRCA+, BRCAwt and BRCA ukn respectively. For no gross residual disease
PFS2 26.1 17.9 24.7 17.9 21.1 16.5
Median, mo 0.44 (0.29–0.69); 0.57 (0.41–0.79); 0.62 (0.48–0.79); the PFS is NR, 34.2 and 24.5 months respectively. Conclusions: Given the
HR (95% CI); P value P=0.0003 P=0.0006 P=0.0001 difficulty with toxicity not defined as a DLT, the RP2D for all regimens is
TSST NR 19.4 26.5 19.4 22.2 18.6
Median, mo 0.49 (0.31–0.78); 0.58 (0.41–0.82); 0.70 (0.54–0.91); veliparib 150mg BID. This data informed the dose that moved into the phase
HR (95% CI); P value P=0.0024 P=0.0018 P=0.0064 III trial GOG 3005/Velia: NCT02470585. Velia also incorporated maintenance
Visit cutoff Apr 15, 2017 (date of unblinding for primary efficacy analyses). veliparib instead of maintenance bevacizumab among all high grade serous
HRs estimated with a Cox proportional hazards model. patients (BRCA+ and wt). These results will determine utilization of veliparib in
CI, confidence interval; HR, hazard ratio; NR, not reached.
this space. Clinical trial information: NCT00989651.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
320s Gynecologic Cancer

5524 Poster Session (Board #347), Sat, 1:15 PM-4:15 PM 5525 Poster Session (Board #348), Sat, 1:15 PM-4:15 PM
Determination of eligibility criteria for salvage hysterectomy after definitive Human papillomavirus genotype and prognosis of invasive cervical cancer: A
radiotherapy/concurrent chemoradiotherapy for residual cervical disease. nationwide cohort study. First Author: Jiayao Lei, Karolinska Institutet,
First Author: Munetaka Takekuma, Department of Gynecology, Shizuoka Stockholm, Sweden
Cancer Center, Shizuoka, Japan
Background: The role of human papillomavirus (HPV) in development from
Background: Patients with persistent cervical cancer after definitive oncogenic infection to invasive cervical cancer (ICC) has been well estab-
radiotherapy/concurrent chemoradiotherapy (RT/CCRT) have a poor prog- lished. However, the association of HPV genotypes and prognosis of ICC
nosis. Salvage hysterectomy (HT) is potentially curative, but eligibility cri- is controversial. Methods: We identified all ICC diagnosed in Sweden
teria therefor have not been determined. Methods: Part 1) Retrospective during the years 2002-2011 (4254 confirmed cases after clinical and
review of patients with persistent cervical cancer treated with definitive RT/ histo-pathological review), requested all archival formalin-fixed, paraffin-
CCRT at 35 institutions of the Japanese Clinical Oncology Group (JCOG) embedded blocks and subjected them to comprehensive HPV genotyping.
from 2005–2014. Differences between a salvage HT group and a systemic Twenty out of twenty-five archives agreed to the study, contributing a total of
chemotherapy (CT) group after definitive RT/CCRT for residual tumor were 2845 confirmed cases with valid HPV results. Cases were followed up from
evaluated. Clinical variables influencing a salvage HT treatment decision date of cancer diagnosis to 31 December, 2015, migration from Sweden, or
were evaluated using logistic regression analysis. Part 2) Questionnaire- death; whichever occurred first. Five-year relative survival ratios (RSRs) were
based survey conducted by JCOG gynecologic oncologists assessing treat- calculated and excess hazard ratios (EHRs) with 95% confidence intervals
ment choice for patients with residual cervical disease after definitive RT/ (CIs) were estimated using Poisson regression. Results: HPV was detected in
CCRT. Patients with residual cervical tumor before, during and after de- 2365 tumors (83.1% of all cases). The five-year RSR by tumor HPV status
finitive RT/CCRT were surveyed for 86 conditions and appropriate candi- was 0.54 (HPV negative), 0.76 (HPV16 positive), 0.73 (HPV18 positive),
dates for salvage HT were evaluated using heat map analysis. Results: Part 0.72 (other high-risk HPV positive) and 0.56 (low-risk HPV positive) com-
1) We identified 298 patients who underwent salvage HT or systemic CT. pared to the age-matched general female population. Compared to cases
Median overall survival was 3.8 and 0.9 year in the HT and CT groups, with HPV-negative tumor, a significantly lower excess mortality was seen if
respectively (HR 0.4341, 95% CI 0.336-0.559, p , 0.01). FIGO stage and the tumor was positive for HPV16 (EHR:0.54, 95% CI 0.44-0.65), other
lymph node metastasis at initial treatment, performance status (PS) at di- high-risk HPV (EHR:0.47, 95% CI 0.37-0.60), and low-risk HPV (EHR:
agnosis of residual cervical tumor and parametrial invasion of residual 0.48, 95% CI 0.32-0.74), after adjustment for age, time since cancer
cervical tumor significantly influenced a salvage HT treatment decision. Part diagnosis, International Federation of Gynecology and Obstetrics (FIGO)
2) Heat map analysis showed that surveyed variables segregated into 3 stage, educational level and histology. However, the mortality among women
groups: i) in favor of salvage HT, ii) in favor of systemic CT, and iii) either. with HPV18 positive tumors were not statistically significantly different from
Conditions such as FIGO stage IB-IIB, PS of 0-1, residual tumor , 4 cm, no cases with HPV-negative tumors. In women with a single HPV infection of
parametrial invasion and no residual lymph node metastasis were included in either HPV16 or HPV18, those with HPV18-positive tumors had 56% (EHR:
group i, in favor of salvage HT. Conclusions: Eligibility criteria could be 1.56, 95% CI: 1.13-1.97) higher excess mortality compared to women with
determined based on the results of the current study, and a prospective HPV16-positive tumors. Conclusions: HPV genotype in cervical cancer tu-
clinical trial evaluating the survival benefit of salvage HT for residual cervical mor is associated with prognosis of ICC. Single HPV18 positivity indicated a
tumor after definitive RT/CCRT is being planned by JCOG. poorer prognosis than single HPV16 positivity. This could add information of
value beyond the established clinical prognostic factors for women di-
agnosed with ICC.

5526 Poster Session (Board #349), Sat, 1:15 PM-4:15 PM 5527 Poster Session (Board #350), Sat, 1:15 PM-4:15 PM
Clinically significant discrepancy between clinical and pathologic stage of Feasibility of visual inspection with acetic acid (VIA) screening for cervical
early-operable cervical cancer. First Author: Jennifer Gibbs, Department of cancer in Tanzania with emphasis on special populations. First Author:
Gynecologic Oncology, SUNY Downstate Medical Center, Brooklyn, NY Justine Chinn, University of California, Irvine School of Medicine, Irvine, CA
Background: The cornerstone of the management of cervical cancer (CC) Background: Following the report that VIA screening reduced cervical cancer
traditionally relies on clinical examination assessment (CE) of tumor size mortality by 31% in India (ASCO LBA2 2013; Shastri SS, et al JNCI 2014),
(TS) and local extension of disease. Previous reports demonstrate poor the W.H.O. endorsed VIA guidelines for Africa, where the global disease
accuracy of CE, with the most common discrepancy being failure to identify burden is highest. In Tanzania, cervical cancer is a major source of morbidity
parametrial involvement (PI). The goal of this study is to determine the and mortality, with nearly 10,000 new cases and 7,000 deaths annually.
accuracy of CE in comparison to final pathology (FP) in early operable CC. Due to lack of resources, therapies are limited and patient outcomes are
Methods: This is a multi-center retrospective review of patients with early CC further confounded by the relatively high prevalence of concurrent HIV
(FIGO stage IB1, IIA1). Data on age, race, histology, stage, CE findings, FP infection. We report on the feasibility of VIA screening in Tanzania with
report and receipt of adjuvant radiation therapy (RT) were collected. CE emphasis on unique populations. Methods: Our two 5-day VIA screen-and-
findings included TS, PI and vaginal involvement (VI). CE of TS, PI, and VI treat workshops in Buzuruga and Sangabuye Health Centres in Mwanza,
were compared to FP. Subanalysis was also conducted based on TS ( , or $ Tanzania were approved by the University of California, Irvine IRB and local
2cm) and location of tumor (exophytic vs endophytic). Analysis was per- health authorities. Participants were recruited from surrounding commu-
formed using paired-T and Cohen’s Kappa tests. Results: Final analysis nities and offered free cervical VIA screening, cryotherapy when indicated,
included 135 patients. Mean age was 52.6 years. The majority of patients and HIV rapid testing. Acetowhite lesions and/or abnormal vascular markings
had squamous cell carcinoma (72.6%). Overall, there was a significant were VIA+. Chi-square and Fisher exact tests were performed with statistical
difference between CE of TS compared to FP; mean error of 1.22 cm (p , significance assigned at 0.05. Results: During July 2018, 825 of 917
0.0001). In those with tumors $ 2cm the mean error was 1.28 cm (p , registered participants underwent VIA screening and 25.1% (n=207) were
0.0001). No significant discrepancy was observed in tumors , 2 cm (mean VIA+. 147 VIA+ non-pregnant women received same day cryotherapy and 15
error: 1.10cm; p = 0.5). CE of TS of endophytic tumors was poor (mean error (1.8%) with lesions suspicious for cancer were referred to Bugando Medical
1.68cm; p = 0.004) compared to exophytic tumors (mean error: 1.12 cm; Center. In the subanalysis of 64 HIV+ patients (23 diagnosed at the
p = 0.693). There was no significant difference in the identification of VI workshops, 41 with prior diagnosis on ART), HIV infection was not associated
between CE and FP (3.7% vs 8.89%; p = 0.067). No patients with PI on CE with VIA positivity (p=0.497). Additionally, a non-significant trend of higher
were included in this analysis. However, 14.07% of patients were found to VIA+ screens among newly diagnosed untreated HIV patients (27.7%) vs
have PI on FP (p , 0.0001). There was no difference in the accuracy CE of patients with known HIV on ART (17.5%) was observed (p=0.556).
TS between non-obese ( , 30 kg/m2) and obese patients ($30 kg/m2), mean Conclusions: VIA screening for cervical cancer, while feasible in Tanzania,
error 1.13 and 1.3, respectively (p = 0.061). As a results of FP, 55 patients will require follow-up and repetitive screening. Although cervical cancer is an
(40.7%) received adjuvant RT and 38 patients (28.14%) were upstaged AIDS-defining illness, lack of correlation between HIV infection and VIA-
from IB1 to IB2. Of these 38 patients, 36 (94.7%) went on to receive positivity may reflect the availability of W.H.O.-subsidized ART in sub-
adjuvant RT. Conclusions: CE of TS and PI is inaccurate, especially in Saharan Africa to attenuate HPV-mediated neoplastic transformation, as
tumors $ 2cm and endophytic tumors. This suggests imaging should be previously reported by others. Further study of this phenomenon is
strongly encouraged, particularly in the setting of the updated FIGO 2018 warranted.
staging system and recent debate over surgical approach.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gynecologic Cancer 321s

5528 Poster Session (Board #351), Sat, 1:15 PM-4:15 PM 5529 Poster Session (Board #352), Sat, 1:15 PM-4:15 PM
Comparative benefit of interstitial needles in addition to intracavitary Prognostic significance of the number of pelvic lymph-nodes resection in
applicators in the treatment of locally advanced cervical cancer. First Author: patients with cervical adenocarcinoma: An analysis from JGOG 1070S
Dorothy Chilambe Lombe, BC Cancer Agency/University of British Columbia, study. First Author: Yuji Ikeda, Nihon University, Tokyo, Japan
Kelowna, BC, Canada
Background: Although resection of more than 20 lymph-nodes is considered
Background: Cervical cancer is the leading cause of cancer mortality of to be adequate in pelvic lymphadenectomy for cervical cancer patients,
women in Low and Middle Income Countries (LMIC). Interstitial needles (IN) prognostic significance of the number of resected lymph-nodes (PLN-num)
have improved outcomes but the resources required in comparison to is clinically still unknown. Methods: This nationwide multicenter retro-
intracavitary brachytherapy (IC) alone has impeded uptake in endemic re- spective study (JGOG 1070S) examined consecutive 693 patients with
gions. We conducted a retrospective review of the utilisation of IN in the clinical stage IB-IIB cervical cancer who underwent radical hysterectomy
management of locally advanced cervical cancer and simulated 2D planning including pelvic and/or para-aortic lymphadenectomy between 2008-2009
by loading the applicators using standard Manchester loading (ML) to ex- at 87 institutions of the Japanese Gynecologic Oncology Group. Maximum
plore the magnitude of benefit that interstitial needles provide. Methods: 72 number of enrollments from one institution was limited 10 or less to min-
brachytherapy plans of 18 patients who had undergone treatment using imize the inter-institutional bias. Correlation between PLN-num and prog-
tandem and ring and had interstitial brachytherapy between 04/2016 and nosis was analyzed using Cox hazard model with considering histological
10/2018 were reviewed. ML plans prescribed to point A were generated to subtypes. Results: Of 473 eligible cases in this study, the average PLN-num
represent a 2D scenario but the known HR-CTV was taken into consideration per a case in each institution was positively correlated with the number of
and its dosimetric outcomes were compared to those of the 3D based plans. total cases treated in each institution per year. (R = 0.42, P = 0.012).
Results: The median tumour volume was 23 cm3. IN was used in 82 % of the Patients with high PLN-num showed favorable progression free survival
insertions. The median number of IN was 2 (range 0 – 6) with median (PFS) (P = 0.12). Focusing on adeno and adeno-squamous carcinomas,
percentage of IN dwell time 6.6 % (range 0.68 – 38.5). V100 was excellent significantly improved PFS was shown in high PLN-num cases (P = 0.012),
98.2% for ML 97.3% for 3D IN and 98.7% for 3D non-IN plans. The median although no significance was found in squamous cell carcinoma (P = 0.754).
HRCTV D90 was 8.5 Gy/fraction (cumulative EQD210101.4 Gy) for ML plans Multivariate analysis in adeno and adeno-squamous cases showed PLN-num
and 8.0 Gy/fraction (cumulative EQD210 91.4 Gy) for 3D plans. The ML as an independent prognostic factor (HR; 0.46, 95%CI; 0.24–0.84, P =
plans failed to meet the OAR goals except for the rectum, which was op- 0.026) along with disease stage and adjuvant therapeutics. Subset analysis
timally distanced by the rectal paddle. The median bladder, sigmoid and of adeno and adeno-squamous cases without adjuvant therapeutics showed
small bowel doses were 24% above the recommended constraint in the significant improvement of survival in high PLN-num group (P = 0.019).
individual plans and 15% cumulative EQD2. A statistically significant re- Conclusions: In our study, PLN-num in lymphadenectomy for patients with
lationship was found between the number of needles utilised, tumour volume cervical adeno or adeno-squamous carcinoma was clarified to be a signif-
(p , 0.001) and coverage (p = 0.006) but not delivered dose (p , 0.068). icant prognostic factor. Systematic total lymphadenectomy is recommended
Conclusions: 2D brachytherapy can provide adequate dose coverage for for these patients to obtain a favorable prognosis.
most tumours but IN provide a benefit in reducing the doses to OARs in a
significant number of patients. This justifies investment in resources for
uptake of interstitial needles to increase access to optimal treatment of
cervical cancer for women in LMIC. This research was made possible an
ASCO Conquer Cancer Foundation grant.

5530 Poster Session (Board #353), Sat, 1:15 PM-4:15 PM 5531 Poster Session (Board #354), Sat, 1:15 PM-4:15 PM
Association of detection of aflatoxin in plasma of Kenyan women with Cervical cancer harboring a Rb1 mutation may sensitize to cisplatin via
increased detection of oncogenic HPV. First Author: Jianjun Zhiang, PI3K/AKT pathway by regulating apoptosis. First Author: Peng Luo, Zhujiang
Indiana University School of Medicine, Indianapolis, IN Hospital of Southern Medical University, Guangzhou, China
Background: Cervical cancer is the leading cause of cancer-related deaths Background: Cervical cancer is one of the most common malignant tumors in
among women living in Africa. Only a small proportion of HPV-infected women women and major causes of cancer death in women. Although concurrent
develop cervical cancer and other cofactors may increase a woman’s risk of chemoradiotherapy has improved the treatment of cervical cancer, due to the
developing cervical cancer. Aflatoxin, a potent carcinogen and immunosup- heterogeneity of the tumor to chemotherapy or radiation therapy, especially
pressive agent, is produced by fungi that contaminate corn and other staple the varied response to chemotherapy, the patients respond differently to the
foods in sub-Saharan Africa. Women who ingest aflatoxin may be more likely to same therapeutic regimen, furthermore some patients get rapid progression.
have persistent infections with oncogenic HPV type. Methods: Demographics, Therefore, we attempted to analyze the potential relationship between ge-
behavioral data, plasma, and cervical swabs were collected from HIV- nomic mutation and chemotherapy response and survival in cervical cancer
uninfected women 18 and 45 years of age who presented for cervical can- patients. Methods: Clinical information and sequencing data of Cervical
cer screening at Moi Referral and Teaching Hospital (Eldoret, Kenya) and had Squamous Cell Carcinoma and Endocervical Adenocarcinoma (CESC) in
normal VIA examination. HPV testing was performed on cervical swabs using Genomic Data Commons (GDC) data portal were obtained through TCGA-
the Roche Linear Array Assay. Aflatoxin-albumin adduct (AFB1-lys) was de- biolinks. Only patients with stage IIB-IV CESC who received cisplatin only
tected and quantified in plasma. The association of plasma AFB1-lys detection and were included. Cisplatin sensitivity data and sequencing data of CESC
and concentration and the detection of HPV was examined. Results: Sufficient cell lines were obtained from the Genomics of Drug Sensitivity in Cancer
plasma was available from 88 HIV-uninfected women and was transported to (GDSC). Cox regression analysis, Kaplan-Meier survival analysis, differential
the U.S. for aflatoxin testing. Valid HPV testing results were available for 86 of analysis of gene expression and functional enrichment were used to explore
these women (mean age 34.0 years); 49 women (57.0%) had detectable the role of different mutations in survival and cisplatin sensitivity. Results: A
AFB1-lys and 37 (43.0%) had no detection. Substantial variation existed in total of 48 patients with stage IIB-IV CESC were enrolled. 77 genes with
plasma AFB1-lys concentrations among the 49 women (range 0.02 to 0.21 pg/ mutation frequency ＞10% were included in final analysis . Multivariate
mL). Detection of AFB1-lys was not associated with age, and other behavioral analysis showed that the mutation of Rb1 was an independent predictor of
factors such as number of lifetime partners, marital status and age at first sex. overall survival (OS) (HR = 0.07, 95%CI 0.01-0.73, P = 0.026). Patients
AFB1-lys detection was associated with detection of A9 HPV types (HPV 16, with mutant Rb1 had better overall survival (134.2 versus 86.8 months)
31, 33, 35, 52, and 58) as a group in cervical swabs (p = 0.029) as well as A9 compared with patients with wild-type Rb1. The half maximal inhibitory
types excluding HPV 16 (p = 0.020), but not with individual A9 types, A7 HPV concentration (IC50) of the mutant Rb1 cell line to cisplatin was signifi-
types (such as HPV 18), or low-risk HPV types. A concentration dependent cantly lower than that of the wild-type cell line (3.49 versus 10.15mM, P =
association of AFB1-lys was seen with detection of A9 HPV types as a group 0.038). A total of 332 differently expressed genes (DEGs) were identified
(p = 0.009), non-HPV 16 A9 types (p = 0.005), and HPV 52 (p = 0.042), but not and the KEGG pathway enrichment analysis showed that most DEGs were
with the A7 HPV types. Conclusions: AFB1-lys was detected in 57% of HIV- enriched in the PI3K/AKT pathway (P = 0.015). Conclusions: We found that
uninfected Kenyan women without cervical dysplasia. AFB1-lys-positive Rb1 mutation was an independent survival predictor in stage IIB-IV CESC,
women were more likely than AFB1-lys-negative women to have oncogenic and CESC patients with Rb1 mutation may be more sensitive to cisplatin.
HPV A9 types detected. Higher plasma AFB1-lys concentrations were asso-
ciated with increased likelihood of oncogenic HPV A9 type detection. Further
studies are needed to determine if chronic exposure to aflatoxin interacts with
HPV infection (and possibly HIV co-infection) to modulate the risk of cervical
cancer in women in Kenya and other developing countries.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
322s Gynecologic Cancer

5532 Poster Session (Board #355), Sat, 1:15 PM-4:15 PM 5533 Poster Session (Board #356), Sat, 1:15 PM-4:15 PM
Safety and efficacy of nimotuzumab and concurrent intensity-modulated PD-L1 expression, DNA mismatch repair genes, and HPV types in cervical
radiation therapy and chemotherapy for locally advanced cervical squamous squamous cell carcinoma. First Author: Tahmineh Haidary, SUNY Down-
cell cancer. First Author: Ang Qu, Peking University Third Hospital, Beijing, state Medical Center, Brooklyn, NY
China
Background: Previous studies observed low PD-L1 expression in cervical
Background: To evaluate the safety and efficacy of nimotuzumab plus con- cancer, and that high PD-L1 expression is associated with deficient DNA
current intensity-modulated radiation therapy (IMRT) and chemotherapy for the mismatch repair genes and poor prognosis. No study has investigated the PD-
treatment of locally advanced cervical squamous cell cancer (LACSCC). L1 expression differences among HPV types in cervical cancer. We aim to
Methods: From December 2013 to March 2017, 31 patients with stage (FIGO determine PD-L1 expression in our patient population and its clinical sig-
2009) IB2-IVA cervical squamous cell cancer were enrolled in this single-arm nificance. Methods: A total of 198 patients with SCC were identified and 60
clinical trial at an academic medical center and received concurrent chemo- had evaluable tumor specimens and clinical data. Immunohistochemistry on
radiotherapy plus nimotuzumab. All patients underwent at least 1 year of follow- MMR genes and PD-L1 expression using tissue microarrays was performed.
up. The prescription radiation dose was 50.4 Gy/28 F on the pelvic field with or HPV analysis for 15 high-risk types was done by multiplex PCR and gel
without extended-field radiation. An additional 30-36 Gy to Point A was de- electrophoresis. Correlation between PD-L1 expression, MMR status, HPV
livered with high-dose-rate techniques. Cisplatin 40 mg/m2 and nimotuzumab types and other clinical parameters was analyzed using x2 or Fisher’s exact
200 mg were infused intravenously once weekly during radiotherapy. The main test. Cumulative 5-year survival was analyzed by Kaplan-Meier curves and log
and secondary outcome measures were toxicity evaluated using CTCAE 4.0., and rank test. Results: Of the 60 patients, 90% were black, 55% between ages of
the short-term outcome evaluated by RECIST 1.1. Results: The median follow- 30-55 and 40% older than 55. 33 patients had poorly, 20 moderately, and 7
up duration was 29.7 months (13.3-61.2 months). All patients received external well-differentiated tumors. At 5-year follow up, 11 had recurrence, 14 patients
beam radiotherapy, brachytherapy, and nimotuzumab six times. Twenty-seven died of disease and 7 lost to follow up. 93.3% of tumors had positive staining
patients received six cycles of chemotherapy while four received only 4-5 cycles. for PD-L1 with 56.7% showing high expression. Patients aged 30-55 showed a
There was no life-threatening toxicity. The incidence of acute grade 3 bone higher rate of PD-L1 expression. Majority had intact MMR (91.7%). High-risk
marrow depression was 51.6% (16/31) and grade 3 gastrointestinal tract re- HPV DNA was extracted in 41 specimens (68.3%), 11 of these were infected
action was 9.7% (3/31). The incidence of late toxicities was 22.6% (7/31), and with $2 or more types, identifying 6 high-risk types not included in vaccine.
these included vaginal-rectal fistula, intestinal obstruction, rectal hemorrhage, Correlation analysis showed that there is significant correlation between age
hematuria, and vaginal stenosis. Complete response was achieved in 30 cases and PD-L1 expression on tumor cells (p = 0.047). No correlation between
(96.8%). The 1-year disease-free survival (DFS), local progression-free survival MMR status and PD-L1 expression. Mean disease free survival (DFS) was 43.7,
(LPFS), and overall survival (OS) rates were 87.1%, 90.3%, and 100%, re- 36.5, and 6.3 months for high, low and negative PD-L1 expression, re-
spectively. The corresponding 3-year values were 74.8%, 90.3%, and 86.7%. spectively (p = 0.002). In patients with stage I-II disease, mean DFS was 52.2,
Conclusions: Nimotuzumab plus concurrent IMRT and chemotherapy may 42.6 and 7.6 months for high, low and negative PD-L1 expression, respectively
represent a well-tolerated and effective treatment regimen in patients with (p = 0.001). Similar direct relationship is observed with overall survival. Cases
LACSCC. with mixed HPV infection are associated with late clinical stage at diagnosis
(p = 0.037) and lower PD-L1 expression. Conclusions: Contrary to previous
studies, our results showed remarkably high rate of PD-L1 expression in
cervical SCC, which is associated with longer DFS and OS, especially in early
stage tumors. Mixed HPV infection is associated with late clinical stage at
diagnosis and lower PD-L1 expression.

5534 Poster Session (Board #357), Sat, 1:15 PM-4:15 PM 5535 Poster Session (Board #358), Sat, 1:15 PM-4:15 PM
Comparison of definitive cervical cancer management with concurrent che- Recurrence and survival after robotic-assisted radical hysterectomy (RRH)
motherapy and radiation between two centers with variable resources and for early-stage cervical cancer (CC): Experience may matter. First Author:
opportunities for improved treatment delivery. First Author: Francis Adumata Christine K. Fitzsimmons, AdventHealth Cancer Institute, Orlando, FL
Asamoah, National Radiotherapy Oncology and Nuclear Medicine Center,
Background: The phase III LACC Trial found that minimally invasive surgery
Accra, Ghana
(MIS) / radical hysterectomy was inferior to open radical hysterectomy (ORH)
Background: Cervical cancer remains a global health challenge particularly in with reduced disease-free survival (86% v 96.5%) and a higher disease-
low to middle income countries with under resourced healthcare systems. We specific death rate (DSDR) (4.4% v 0.6%). We evaluated our experience with
present the experiences of two centers practicing in variable resource envi- attention to the learning curve. Methods: Patients (pts) with early-stage CC
ronments to determine predictors of improved radiochemotherapy outcomes. (4/2007-12/2017) who underwent RRH with a uterine manipulator were
Methods: This retrospective review describes baseline demographic and evaluated in a contemporaneously maintained database. First 10 learning
clinicopathologic characteristics of cervical cancer patients treated with curve cases per surgeon (Group A) were compared to all subsequent cases
concurrent chemotherapy and radiation between 2014 and 2017 at the (Group B). Inclusion criteria mirrored the LACC trial: . one-year follow-up,
National Radiotherapy Oncology and Nuclear Medicine Center (NRONMC) in adenocarcinoma, adenosquamous, or squamous carcinoma, stage IA2 or
Korle Bu Teaching Hospital, Accra, Ghana and Moffitt Cancer Center, Tampa, IB1 using FIGO 2014 guidelines, and pathologic tumor size (TS) of 4 cm or
Florida, USA. Results: Ghanaian patients presented at an older median age less. Atypical histologies and lesions . 4 cm were excluded. Study pa-
(56 vs. 49 years, p , 0.001), with predominantly stage IIIB disease (43% vs. rameters assessed included recurrence free survival (RFS), DSDR, and
16%, p , 0.001) and squamous cell histology (89% vs. 79%, p , 0.001). pattern of recurrence. Results: 144 RRH pts were identified and 90 met
Median treatment duration was longer for Ghanaian patients (58 vs. 52 days, inclusion criteria with mean age of 45.6614.3 years. Exclusions included
p , 0.001). Ghanaian patients were less likely to receive concurrent che- stage 1A1 without LVSI (n = 13), atypical histology (n = 10), lost to follow-up
motherapy (68% vs. 100%, p , 0.001) and interstitial brachytherapy im- (n = 13), and occult stage 1B2 (n = 18). 40 pts met Group A and 50 met
plants (0 vs 19%, p , 0.001). No Ghanaian patients received a radiation boost Group B criteria. Median follow-up was 61634.3 months (A = 71.5, B =
to pelvic or paraortic lymph nodes (p , 0.001). Ghanaian patients had lower 52.5). The 5-year RFS was 92% (95 CI 64%) and the DSDR 5.5% (n = 5).
local control (64% vs. 93%, p , 0.001) and overall survival (82% vs. 95%, p = There were 7 (7.8%) recurrences with median time to recurrence of 1268.3
0.02) at 24 months, respectively. For stages IB, IIA, IIB, IIIB, 24 month local mos. Recurrence in Group A (n = 6, 15%) exceeded Group B (n = 1, 2%), p=
control rates for NRONMC vs. Moffitt patients were (60% vs. 93%; p = 0.05), 0.025. DSDR was 10% Group A v 2% B (p= 0.184). The 4.5 yr RFS was
(89% vs. 100%; p = 0.35), (91% vs. 91%; p = 0.89), (53% vs. 91%; p = 84.8% (95 CI 67%) in Group A v 98% (95 CI 63%) in Group B. There were
0.02) and 24 month OS rates were (85% vs. 100%; p = 0.06), (100% vs. no differences in risk factors for recurrence between A & B (TS . 2 cm, LN
100%; p = 0.48), (85% vs. 96%; p = 0.2), (73% vs. 91%; p = 0.24), re- (+), adjuvant therapy (AT), and LVSI p. 0.05), except (+) vaginal margin
spectively. Treatment duration . 55 days predicted poorer overall survival on status (A = 10% v B = 0%, p= 0.034). Three recurrences involved carci-
multivariable analysis (MVA). Stage $III disease predicted poorer local control nomatosis, which may be insufflation related. All recurrent cases had TS
on MVA. Conclusions: Significant differences were noted in treatment and . 2 cm and 5 received AT. Conclusions: In this study, recurrence of disease in
disease characteristics between the two centers. Feasible improvements for early-stage CC clustered in the first 10 cases per surgeon and occurred in
patients treated at NRONMC include removing financial barriers to chemo- TS . 2 cm. This data suggests a possible learning curve effect and argues
therapy access, improving radiotherapy delivery capacity to reduce treatment against a uterine manipulator cause. Carcinomatosis may be insufflation
delays, and screening programs to reduce advanced disease presentation. related, unique to MIS, and deserves further study.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gynecologic Cancer 323s

5536 Poster Session (Board #359), Sat, 1:15 PM-4:15 PM 5537 Poster Session (Board #360), Sat, 1:15 PM-4:15 PM
Omission of adjuvant therapy in stage I clear cell ovarian cancer: Review of A randomized, double-blind, placebo-controlled phase Ib/II study of
the British Columbia (BC) cancer experience. First Author: Shiru Lucy Liu, ralimetinib, a p38 MAPK inhibitor, plus gemcitabine (G) and carboplatin
BC Cancer, Vancouver, BC, Canada (C) versus GC for women with recurrent platinum-sensitive ovarian cancer.
First Author: Ignace Vergote, BGOG and University Hospitals Leuven, Leuven
Background: Standard guidelines recommend adjuvant chemotherapy for
Cancer Institute, Leuven, Belgium
stage I clear cell ovarian cancer (CCOC), despite data demonstrating ex-
cellent outcomes. Since 2012, the BC Cancer provincial treatment guide- Background: p38 mitogen-activated protein kinase (MAPK) regulates cytokine
lines for surgically staged stage IA/B and IC1 (defined by intraoperative production in the tumor microenvironment and enables therapeutic resistance
rupture only) CCOC has been to offer observation only. We reviewed the of cancer cells. Ralimetinib (R) is a selective small-molecule inhibitor of p38a
clinical outcomes of stage I CCOC patients since policy implementation. and p38b MAPKs. Methods: Main inclusion criteria: $18 y; recurrent
Methods: A retrospective, population-based cohort study of all stage I CCOC platinum-sensitive epithelial ovarian, fallopian tube, or primary peritoneal,
patients operated on between April 2012 and December 2017 was con- cancer after first-line treatment. Phase (Ph)1b was to determine the recom-
ducted. Patient, tumor, surgical and clinical outcome data were collected. mended Ph2 dose (RP2D) of R administered 12-hourly (Q12H) on Days 1-10
Survival analysis was conducted using Kaplan-Meier methods. Results: 78 (21-day cycle [Q21D]) in combination with gemcitabine (G: 1000 mg/m2 on
patients with stage I disease were identified (see Table). Among stage IC1 Days 3 and 10) and carboplatin (C: AUC 4 on Day 3) for 6 cycles. In Ph2,
patients, 9 received adjuvant therapy despite provincial policy, 6 of which patients (pts) were randomized double-blind, 1:1 to RP2D R+GC or placebo
were due to sharp dissection. 40 patients with stages IA/B and IC1, who (P)+GC, for 6 cycles, followed by R 300 mg Q12H or P on Days 1-14, Q28D
underwent post-operative observation, were included in the analysis. Median until disease progression. The stratified log-rank test compared
duration of follow-up was 36 months. Median age at diagnosis was 55 years progression-free survival (PFS; primary endpoint) between treatment groups
and .50% patients had a Charlson Comorbidity Index of 0 (N= 26) and an in Ph2, at a 1-sided a level of 0.2. ClinicalTrials.gov, NCT01663857.
Eastern Cooperative Oncology Group performance status of 0 (N=28) prior to Results: 118 pts received $1 dose of R or P (safety population); 8 in Ph1b and
diagnosis. Lymph node dissection was not performed in 20 patients. All 16 110 in Ph2 (R+GC N = 58; P+GC N = 52). The RP2D for R in combination with
cases tested immunohistochemically for mismatch repair were intact, and 2 GC was 200 mg Q12H. The study met its primary objective (median PFS:
of 6 cases with tumour genomic sequencing had an AURKA aberration. R+GC 10.3 mo vs P+GC 7.9 mo; HR = 0.773, 2-sided p = 0.246). The
There were 4 recurrences (10%), 3 of which were metastatic. 5-year disease- secondary objectives of median overall survival (R+GC 29.2 mo vs P+GC 25.1
free survival is 90%, and 5-year overall survival is 95% for stage IA/B and mo; HR = 0.827, p = 0.469) or overall response rate (R+GC 46.6% vs P+GC
90% for stage IC1 (p=0.645). In comparison, 5-year overall survival for stage 46.2%; p = 0.967) were not statistically significant, and 32.4% vs 25.0% of
IC2 (surface involvement) and IC1 with sharp dissection (all received pts had normalized CA125 at the end of cycle 6. Most pts (safety population)
chemotherapy) is 82% and for stage IC3 (positive washings) is 23% experienced $1 Grade 3/4 treatment-emergent adverse event (TEAE: R+GC
(p,0.001). Conclusions: Outcomes of patients with stage I A/B and C1 63/66 [95.5%]; P+GC 48/52 [92.3%]). Decreased neutrophil count (60.6%
CCOC remain excellent. Adjuvant therapy can be safely omitted, with low vs 76.9%), platelet count (43.9% vs 38.5%), and white blood cell count,
recurrence rates and survival over 90% at 5 years. Consideration of disease (30.3% vs 26.9%), anemia (22.7% vs 25.0%), and increased alanine ami-
substage is valuable in predicting the clinical outcomes of stage I CCOC. notransferase (ALT) (19.7% vs 3.8%) were the Grade 3/4 TEAEs in $10% of
pts in the R+GC and P+GC arms, respectively. Conclusions: Addition of
Stage IA IB IC1 IC2 IC3 ralimetinib to GC resulted in modest improvements in PFS. Grade 3/4 elevated
N 25 2 22 9 20 ALT was more common in the ralimetinib arm. Clinical trial information:
NCT01663857.

5538 Poster Session (Board #361), Sat, 1:15 PM-4:15 PM 5539 Poster Session (Board #362), Sat, 1:15 PM-4:15 PM
Efficacy and safety of tivozanib in recurrent, platinum-resistant ovarian, Adverse events (AEs) with maintenance olaparib in newly diagnosed patients
fallopian tube or primary peritoneal cancer. First Author: Wendy M Swetzig, (pts) with advanced ovarian cancer (OC) and a BRCA mutation (BRCAm):
Northwestern University Feinberg School of Medicine, Chicago, IL Phase III SOLO1 trial. First Author: Nicoletta Colombo, University of Milan-
Bicocca and Istituto Europeo di Oncologia, Milan, Italy
Background: Tivozanib is a potent, selective pan-vascular endothelial growth
factor (VEGF) receptor tyrosine kinase inhibitor with a long half-life. This Background: In SOLO1 (NCT01844986), maintenance olaparib provided a
study assessed its activity in patients with recurrent, platinum-resistant substantial progression-free survival benefit vs placebo in newly diagnosed
ovarian cancer (OC), fallopian tube cancer (FTC) or primary peritoneal cancer pts with advanced OC, a BRCAm and clinical complete or partial response to
(PPC). Methods: This open-label phase II study used a Simon’s two-stage platinum therapy (HR 0.30; 95% CI 0.23–0.41) and was well tolerated
design. Eligible patients had recurrent, platinum-resistant OC, FTC or PPC; (Moore et al. NEJM 2018). We analysed the most common AEs and he-
ECOG PS of 0-1; normal end organ function; and measurable or detectable matologic AEs in SOLO1. Methods: Pts received olaparib tablets 300 mg
disease. There was no limit on the number of prior regimens. Treatment twice daily or placebo until progression unless they had no evidence of
consisted of tivozanib 1.5 mg orally once daily (3 weeks on/1 week off). The disease at 2 years, in which case treatment stopped. AEs were graded using
primary endpoint was response rate. Secondary endpoints were progression- CTCAE v4.0. Results: Of 391 pts randomized, 390 (olaparib, 260; placebo,
free survival (PFS), overall survival (OS), and toxicity assessment. If 1 partial 130) were treated and included in the safety analysis. Median treatment
response (PR) was observed in stage I [n = 12], enrollment proceeded to duration was approximately 25 months for olaparib vs 14 for placebo.
stage II. The null hypothesis was rejected for $ 4 responses in 30 patients. Median time to first onset of the most common AEs (nausea, vomiting,
Results: Thirty-one patients were enrolled, and 30 were treated. Twenty- fatigue/asthenia, anemia) and neutropenia and thrombocytopenia was
three had OC [76.67%], 5 FTC [16.67%] and 2 PPC [6.67%]. Twenty-six , 3 months; the first event lasted a median of , 2 months, apart from fatigue/
had measurable [86.67%] and 4 detectable disease [13.37%]. The median asthenia, which lasted a median of , 4 months (Table). AEs were usually
age was 60, and median number of prior regimens was 4 [range 1-9]. Four managed with supportive therapy and/or dose modification; few pts dis-
PRs [13.33%] were recorded. Twelve patients had stable disease (SD) continued. Conclusions: AEs in newly diagnosed pts with advanced OC
[40%]. The clinical benefit rate (PR + SD) was 53%. Seven patients treated with olaparib usually occurred early and were manageable, with few
[23.33%] survived progression-free for . 6 mos. One patient continued discontinuations. Clinical trial information: NCT01844986.
treatment for . 2 yrs. The median PFS was 4 mos [range 1-25] and median Fatigue/ Neutro- Thrombo-
OS was 8 mos [range 1-39]. There were no treatment-related deaths. Grade Nausea Vomiting asthenia Anemia* penia* cytopenia*
3-4 related toxicities were hypertension [8], fatigue [3], fistula [2], hypo- O P O P O P O P O P O P
natremia [2], intestinal perforation, obstruction, stroke, proteinuria, hypo- Pts with AE, n (%) 201 49 104 19 165 54 101 13 60 15 29 5
(77) (38) (40) (15) (63) (42) (39) (10) (23) (12) (11) (4)
magnesemia, hypoalbuminemia, portal hypertension, nausea and anemia [1 Median time to first event, 0.13 0.69 1.46 1.94 0.72 1.54 1.94 1.81 1.77 0.49 2.83 7.39
each]. Frequent grade 1-2 related toxicities included fatigue [19], hyper- months
Median duration of first event,† 1.41 0.43 0.07 0.03 3.48 2.30 1.87 1.64 0.76 0.49 0.95 0.49
tension [13], anorexia [12], arthralgia [11], diarrhea [11], weight loss [10], months
Supportive therapy, n (%) 117 15 28 3 11 0 72 4 11 2 2 1
hoarseness [8], headache [8] and nausea [7]. Exploratory analyses in tumor (45) (12) (11) (2) (4) (28) (3) (4) (2) (1) (1)
samples are ongoing. Conclusions: Tivozanib is active in patients with re- Dose interruption, n (%) 35 0 25 3 20 1 58 1 30 5 6 0
(13) (10) (2) (8) (1) (22) (1) (12) (4) (2)
current OC, FTC or PPC, without substantial toxicity, supporting its further Dose reduction, n (%) 10 0 0 0 15 1 44 1 10 1 4 0
development. Clinical trial information: NCT01853644. (4) (6) (1) (17) (1) (4) (1) (2)
Discontinuation, n (%) 6 1 2 0 6 1 6 0 1 0 1 0
(2) (1) (1) (2) (1) (2) ( , 1) ( , 1)
Grade 3–4 AE, n (%) 2 0 1 1 10 2 56 2 22 6 2 2
(1) ( , 1) (1) (4) (2) (22) (2) (8) (5) (1) (2)

*Grouped term; †AEs with no end date censored at end of safety follow-up or data cut-off, as applicable. O, olaparib;
P, placebo

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
324s Gynecologic Cancer

5540 Poster Session (Board #363), Sat, 1:15 PM-4:15 PM 5541 Poster Session (Board #364), Sat, 1:15 PM-4:15 PM
Results of the VENUS study: Bevacizumab efficacy and safety in platinum- Maintenance olaparib after platinum-based chemotherapy in patients (pts)
sensitive recurrent ovarian cancer (OC)—A real-life ambispective study. First with newly diagnosed advanced ovarian cancer (OC) and a BRCA mutation
Author: Isabelle Laure Ray-Coquard, Centre Léon-Bérard, Lyon, France (BRCAm): Efficacy by surgical and tumor status in the Phase III SOLO1 trial.
First Author: Cara Amanda Mathews, Women & Infants Hospital, Providence,
Background: The VENUS study reports on the efficacy/safety of bevacizumab (Bev)
in patients (pts) treated in the real-life setting. Methods: In this multicentric ob-
RI
servational ambispective VENUS study, all Pts were naive of any antiVEGF and Background: In SOLO1 (NCT01844986), maintenance olaparib significantly
received Bev +/- chemotherapy. Pts were followed until progression or death, for a improved progression-free survival (PFS) vs placebo (HR 0.30; 95% CI
maximum of 3 years since Bev initiation. De novo side effects were defined as 0.23–0.41; Moore et al. N Engl J Med 2018) in pts with newly diagnosed
symptoms for which patients were naı̈ve at baseline. Results: 148 OC pts were advanced OC and a BRCAm. This analysis evaluates olaparib efficacy by timing
included (27 centres), 10 excluded and 8 were lost of follow-up. 52 were retro- of surgery, presence of residual tumor following surgery and response status after
spective. Median age 64 years (55-70). 84.1% were advanced. Median duration of
completion of chemotherapy in SOLO1. Methods: Pts underwent cytoreductive
Bev was 8.6 months, min 1 max 36 months. Initial Bev dose was 15 mg/kg Q3W for
65.3%, 10.0 for 22.5%, 7.5 for 10.2% and 5.0 for 2%. 2 pts presented with surgery and were in clinical complete response (CR) or partial response (PR) after
thrombotic micro-angiopathy (1.4%). Before Bev, hypertension (HTN) was present platinum-based chemotherapy. Pts were stratified by response and received
in 28.9%; proteinuria in 11.3%. Incidence of de novo HTN was 25%. 43 pts olaparib tablets 300 mg twice daily or placebo. Investigator-assessed PFS and
(31.2%) experienced de novo Grade 1-2 Pu, for a total of 56 events, no grade 3-4 objective response were assessed using modified RECIST v1.1. Results: 260 pts
was observed. A total of 12 Grade 4 events occurred: 9 neutropenia and 3 were randomized to olaparib and 131 to placebo; one pt did not receive placebo.
thrombopenia. Mean overall survival (OS) and progression free survival (PFS) were Median follow-up was 41 months in both arms. 63% and 35% of pts underwent
30.0 and 13.3 months, respectively. Conclusions: 1) 1/3 of pts were treated at low upfront and interval surgery, 21% and 76% had residual and no residual
doses in this real-life study; 2) safety of Bev in real-life was manageable and as macroscopic disease after surgery, and 74% and 26% entered the study in
expected, 3) OS and PFS were consistent with those reported in the OCEANS study: clinical CR and PR (based on electronic case report form [eCRF] data). PFS was
PFS 12.4 and OS 33.6 months but lower than in the GOG-0213 study: PFS 13.8 significantly improved regardless of the timing of surgery, residual disease status
and OS 42.6 months. De novo events recorded during follow-up. after surgery or response after platinum-based chemotherapy (Table). In pts with
Event % of pts (n) Grade 3 (n events) Grade 4 (n events) baseline radiologic evidence of disease (n=80; eCRF), the objective response
Asthenia 65.2 (90) 2 0 rate was 43% for olaparib (CR, 28%) and 23% for placebo (CR, 12%).
Nausea 39.9 (55) 2 0 Conclusions: Maintenance olaparib improved outcomes compared with placebo
Epistaxis 35.5 (49) 0 0
Proteinuria 31.2 (43) 0 0 in pts with newly diagnosed advanced OC and a BRCAm, regardless of surgical or
Neutropenia
Constipation
29.0 (40)
26.1 (36)
18
0
9
0
tumor status. Clinical trial information: NCT01844986.
Diarrhoea 25.4 (35) 0 0
Anemia 24.6 (34) 10 0 Median PFS, months
Hematuria 22.5 (31) NA NA HR (95% CI)
Abdominal pain 21.0 (29) NA NA O P O vs P
Cephalgia 20.3 (28) 0 0
Vomiting 19.6 (27) 3 0 Upfront surgery NR 15.3 0.31 (0.21–0.46)
Dyspnea 18.8 (26) 1 0 Interval surgery 33.6 9.8 0.37 (0.24–0.58)
Thrombopenia 16.7 (23) 2 3
Edema (lower limbs) 13.8 (19) NA NA Residual disease 29.4 11.3 0.44 (0.25–0.77)
Anorexia 13.0 (18) 0 0 No residual disease NR 15.3 0.33 (0.23–0.46)
Arthralgia 10.9 (15) 0 0 Clinical CR*‡ NR 15.3 0.34 (0.24–0.47)
Thromboembolic event 9.4 (13) NA NA
Back pain 8.7 (12) NA NA Clinical PR†‡ 30.9 8.4 0.31 (0.18–0.52)
*NCI-CTCAE 4.03; G=Grade; de novo = pt with no event at inclusion and who presented an event during f/u; NA: Not *NED and a normal CA-125 level. †$30% decrease in tumor volume or NED after
available, not graded. Other adverse events with all grade incidence ,5% are not reported in this abstract. VENUS was chemotherapy but an abnormal CA-125 level. ‡By eCRF. NED, no radiologic evidence of
supported by an unrestricted educational grant from Roche France.
disease; NR, not reached; O, olaparib; P, placebo

5542 Poster Session (Board #365), Sat, 1:15 PM-4:15 PM 5543 Poster Session (Board #366), Sat, 1:15 PM-4:15 PM
Ofranergene obadenovec (VB-111) in platinum resistant ovarian cancer: with Circulating tumor DNA analysis to reveal genomic profiles for epithelial ovarian
an immunotherapeutic effect. First Author: Yael Chava Cohen, Tel Aviv cancer. First Author: Yang Xiang, Peking Union Medical College Hospital,
Sourasky Medical Center and Sakler Faculty of Medicine, Tel Aviv University, Beijing, China
Tel Aviv, Israel
Background: Circulating tumor DNA (ctDNA) analysis in epithelial ovarian
Background: VB-111 is a targeted anti-cancer gene therapy with a dual cancer (EOC) was previously reported, however with limited samples or limited
mechanism: anti angiogenic/vascular disruption and induction of an anti-tumor genes. Here, we reported an analysis of ctDNA in EOC cohort using targeted
directed immune response. We report final results of a phase I/II study of VB-111 sequencing with a 1021-gene panel. Methods: Patients with EOC were en-
in combination with paclitaxel in patients with platinum-resistant ovarian cancer. rolled, and treatment-naı̈ve tumor tissues and blood samples were collected.
Methods: Study NCT01711970 was a prospective, open label, dose escalating We utilized a 1021-gene NGS panel in matched tissue DNA and ctDNA to
study assessing combination treatment of VB-111 Q8W and weekly Paclitaxel. In identify somatic mutations with white blood cell DNA as a germline control.
the phase I part of the study patients were treated with escalating doses of Results: Mutations were identified in all of the 65 tissues and in 53 (81.5%)
intravenous VB-111 and Paclitaxel. In phase 2 patients were treated with ctDNA. The median ctDNA mutation allelic frequency was 2.5%, ranging from
therapeutic doses of VB-111 1x1013 Viral Particles and paclitaxel 80mg/m2. 0.1% to 36.2%. A median of 66.7% (12.5%-100.0%) of tissue derived
Assessments included safety, overall survival (OS), PFS, tumor response (CA- mutations were observed in ctDNA. Besides, there were 91 ctDNA private
125 and RECIST) and histopathology. Results: 21 patients with recurrent mutations, including TP53 gene mutations. The most frequently mutated
platinum-resistant ovarian cancer were enrolled and treated in 2 US sites. genes were TP53 (55.4%), PIK3CA (13.8%) and ARID1A (12.3%) in ctDNA
Patients received a mean of 2.3 61.8 repeat doses of VB-111. 17/21 received analysis, which were consistent with tissue analysis (60.0%, 26.2% and
the therapeutic dose. Median age was 65 (41-79) with a median of 3 (1-4) prior 20.0% of tissues with TP53, PIK3CA and ARID1A mutations, respectively).
lines of therapy. Half of the subjects were Platinum refractory, and half were Mutations of TP53 (37/42) in high-grade serous ovarian carcinoma (HGSOC),
previously treated with antiangiogenics. No DLTs were observed. VB-111 was PIK3CA (10/11) and ARID1A (8/11) in ovarian clear cell carcinoma, BRAF (4/
well tolerated and was associated with generally mild flu-like symptoms. In the 5) in low-grade serous ovarian carcinoma and PIK3CA (3/5), ARID1A (2/5) and
therapeutic dose cohort, a 58% CA-125 GCIG response rate was seen in PTEN (2/5) in endometrioid carcinoma were observed as the most commonly
evaluable patients including durable responses, and responses in patients with
genetic aberrations in ctDNA in different sub-types of EOC, which located in
platinum refractory disease and post anti-angiogenic failure. The median OS was
different signal pathways and suggested different pathogenesis. In total,
498 days in patients treated with Therapeutic Dose compared to 173 days in
90.5% (38/42) of HGSOC were ctDNA positive, comparing with 65.2% (15/
Sub-therapueutic dose (p = 0.028). Tumor Specimens taken after treatment
23) of other EOC subtypes (p = 0.012). In addition, 56.5% (13/23) of stage
demonstrated tumor infiltrated with cytotoxic CD8 T-cells and regions of apo-
ptotic cancer cells. Conclusions: Treatment with VB-111 in combination with
I~II EOC were ctDNA positive, comparing with 94.7% (36/38) of stage III (p =
weekly Paclitaxel was safe and well tolerated. Favorable tumor responses and 0.002). No association between ctDNA positivity and other clinic charac-
overall survival outcomes were associated with induction of an immunothera- teristics was observed, including pathological differentiation, CA125, lesion
peutic effect manifested as tumor infiltration with CD-8 T cells. Encouraging density (solid vs. cystic-solid and cystic). Multivariable analysis suggested
results are the basis for further exploration in the ongoing, placebo controlled, FIGO stage III (p = 0.008) as an independent predictor of ctDNA detection.
pivotal OVAL study. Clinical trial information: NCT01711970. Conclusions: In summary, genomic characterization of EOC may offer insights
into tumorigenesis and identify potential therapeutic targets in this disease.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gynecologic Cancer 325s

5544 Poster Session (Board #367), Sat, 1:15 PM-4:15 PM 5545 Poster Session (Board #368), Sat, 1:15 PM-4:15 PM
First-in-human (FIH) phase 1 (Ph1) study of MORAb-202 in patients (pts) SOLO1 versus SOLO2: Cost-effectiveness of olaparib as maintenance ther-
with advanced folate receptor alpha (FRA)-positive solid tumors. First Au- apy for newly diagnosed and platinum-sensitive recurrent ovarian carcinoma
thor: Toshio Shimizu, Department of Experimental Therapeutics, National among women with germline BRCA mutations (gBRCAmut). First Author:
Cancer Center Hospital, Tokyo, Japan Juliet Elizabeth Wolford, The Division of Gynecologic Oncology at the
University of California, Irvine, Orange, CA
Background: MORAb-202 is an antibody drug conjugate consisting of far-
letuzumab (a humanized monoclonal antibody that binds to FRA) paired Background: With the December 19, 2018 regulatory approval by the US
with a cathepsin B-cleavable linker to eribulin mesylate (a microtubule FDA of olaparib tablets as maintenance therapy for women with deleterious
dynamics inhibitor). We report preliminary results from a FIH Ph1 study of or suspected deleterious germline or somatic BRCAmut advanced ovarian
MORAb-202 in pts with FRA-positive solid tumors. Methods: This open- carcinoma, it becomes important to clarify the role of PARP inhibitors in this
label, ongoing, FIH study evaluated the safety, tolerability, pharmacoki- disease. We evaluated cost-effectiveness of olaparib in the upfront (SOLO1)
netics, pharmacodynamics, maximum tolerated dose, and/or the recom- versus the recurrent maintenance setting (SOLO2). Methods: Data were
mended dose of MORAb-202 (Part 1: Dose finding part with accelerated obtained from SOLO1, the phase 3 placebo-controlled randomized upfront
modified toxicity probability interval design; Part 2: Expansion part). Eligible maintenance study among gBRCAmut patients [median PFS greater than
pts had FRA-positive solid tumors who failed standard therapy and an ECOG 49.8 vs 13.8m: HR 0.30; 95% CI, 0.23-0.41; p , 0.001, NCT01844986]
PS of #1. MORAb-202 was administered by intravenous injection once and SOLO2, the phase 3 placebo-controlled randomized maintenance study
every 3 weeks and dose-limiting toxicities (DLTs) were assessed during the among gBRCAmut patients with platinum-sensitive recurrence and at least
first 21-day cycle. Efficacy endpoints were assessed with RECIST v1.1 by two prior lines of therapy [median PFS 19.1 vs 5.5m: HR 0.30; 95% CI,
investigator assessment. Results: As of Nov 16, 2018, 16 pts with con- 0.22-0.41; p , 0.0001, NCT01874353]. Investigator-assessed median
firmed FRA-positive tumors were enrolled and treated with MORAb-202 PFS and toxicity data from the trials were incorporated in a Markov model
across 4 dose levels in Part 1 (0.3mg/kg: n = 3 [2 endometrial and 1 ovarian], which transitioned patients through response, hematologic complications,
0.45mg/kg: n = 3 [3 ovarian], 0.68mg/kg: n = 3 [1 NSCLC, 1 ovarian, and 1 non-hematologic complications, progression, and death. Using TreeAge Pro
TNBC], 0.9mg/kg: n = 7 [4 ovarian, 1 endometrial, 1 NSCLC, and 1 TNBC]); 2015, the costs of pre-treatment testing (eg. gBRCAmut), medications, and
all completed . 1 cycle. One pt in the 0.9mg/kg cohort experienced DLTs of management of adverse effects were analyzed. Incremental cost-
alanine aminotransferase increased (grade 3) and gamma-glutamyl trans- effectiveness ratios (ICERs) per month of life gained and individual PFS-
ferase increased (grade 3). Treatment-emergent adverse events (TEAEs) life year saved (PFS-LYS) were also calculated and compared. Results: In
occurred in 15 pts (93.8%). The most common TEAEs were leukopenia and SOLO1, cost prior to progression was 1.7x that of SOLO2 ($937,440 vs
neutropenia (50% each). The objective response rate based on RECIST v1.1 $564,451). With the extended, estimated median PFS of at least 49.8m for
was 37.5% (6/16 pts) in Part 1 with 1 complete response (ovarian) at 0.9mg/ SOLO1 and 19.1m for SOLO2, upfront maintenance therapy was more cost-
kg and 5 partial responses including 2 pts (both ovarian) at 0.9mg/kg, 1 pt effective. SOLO 1 was associated with $312,480 PF-LYS per individual
(endometrial) at 0.3mg/kg, and 2 pts (1 TNBC and 1 NSCLC) at 0.68mg/kg. patient, while SOLO2 demonstrated $498,045 PF-LYS. Maintenance ola-
The disease control rate was 75% (12/16 pts). Exposure to MORAb-202 was parib was found to be more cost-effective in the 1st-line setting, with an ICER
dose proportional across the dose range investigated. Conclusions: MORAb- of $12,149 per month of life gained when compared directly to SOLO2.
202 escalation to 0.9mg/kg was manageable with encouraging initial an- Conclusions: Although the higher cost associated with olaparib in SOLO1
titumor activity in pts with FRA-positive solid tumors. Clinical trial in- reflects the longer time patients stay on drug due to extended PFS, the ICER
formation: NCT03386942. supports early use in the disease course as first-line maintenance therapy
among women with gBRCAmut advanced ovarian carcinoma.

5546 Poster Session (Board #369), Sat, 1:15 PM-4:15 PM 5547 Poster Session (Board #370), Sat, 1:15 PM-4:15 PM
Modeled CA-125 kinetics during neoadjuvant chemotherapy for predicting Recurrence of ovarian cancer in BRCAwt patients without maintenance
the likelihood of optimal interval debulking surgery in ovarian cancer therapy: Real-world evidence. First Author: Melinda Louie-Gao, TESARO,
patients: Data from CHIVA trial (a GINECO study). First Author: Patrick Inc., Waltham, MA
Robelin, Hospices Civils de Lyon, Pierre-Bénite, France
Background: Development of platinum resistance is a major clinical chal-
Background: A pre-operative predictive biomarker of CC0 interval debulking lenge in ovarian cancer (OC) treatment. In the phase 3 ENGOT-OV16/NOVA
surgery (IDS) likelihood would be helpful. The modeled CA125 elimination trial of the poly(ADP-ribose) polymerase inhibitor (PARPi) niraparib, 55% of
rate constant KELIM predicts OS in 1st line setting (You et al. Clin Cancer BRCA wild-type (BRCAwt) patients (pts) receiving placebo developed
Res 2019). The predictive/prognostic values of KELIM regarding CC scores platinum resistance after their last platinum-based therapy (ie, progressive
at IDS, and survivals, during neo-adjuvant chemotherapy were assessed. disease within 6 months of their last chemotherapy [CT] dose). Niraparib, a
Methods: The data of the CHIVA randomized phase II trial, comparing PARPi approved for the maintenance treatment of adult pts with recurrent
carboplatin-paclitaxel +/- nintedanib before IDS (NCT01583322), were OC following platinum-based CT, significantly prolongs progression-free
used. A semi-mechanistic model was built to describe CA125 longitudinal survival (PFS). This real-world data analysis investigated the risk of plati-
kinetics during the first 100 treatment days. The relationships between num eligibility loss for BRCAwt pts who did not receive maintenance therapy
KELIM and IDS CC scores, PFS & OS, were assessed with other major after platinum treatment. Methods: This retrospective study identified
prognostic factors (grade, histology, GCIG CA125 response, FIGO stage, and 5,535 pts with OC from January 2011–October 2018 using data from
arm) using multivariate logistic regression (logit), C-index & survival tests. Flatiron, a longitudinal, demographically and geographically diverse data-
Results: The longitudinal kinetics of 529 CA125 values, assessed every base derived from records of . 265 cancer clinics and . 2 million US cancer
3 weeks during neo-adj chemotherapy, were modeled in 133 patients (out of pts. BRCAwt pts who had received $2 lines of platinum-based CT, had
188). KELIM (as a continuous covariate) was the only significant predictive disease progression $6 months after their previous line of therapy, and had
factor of CC0 IDS likelihood using multivariate analyses (OR = 12.37, 95% no maintenance therapy (PARPi, bevacizumab, or CT agents) after their
CI [4.32-39.67]). CC0 IDS probability can be estimated with patient KE- current treatment were included. Kaplan-Meier analysis was used to esti-
LIM: $ 90 % if standardized KELIM $ 0.12. Non-parametric survival mate the probability of pts initiating next treatment or death, whichever
models confirmed the independent predictive values of KELIM categorized occurred first, within 6 months. Results: Of 5,535 pts diagnosed with OC,
by terciles regarding PFS & OS (Table). The parametric model linking KELIM 147 BRCAwt pts met the inclusion/exclusion criteria of this analysis (similar
(as a continuous covariate) with OS allows to predict the patient survivals to ENGOT-OV16/NOVA placebo arm). An estimated 56% of pts received the
(months) based on their estimated KELIM (HR = 0.20, [0.10-0.39]). next treatment or died within 6 months after their last platinum-based
Conclusions: The prognostic & predictive values of the modeled CA125 therapy. Median time to next therapy or death was 5.1 months (95%
KELIM are also confirmed regarding CC0 IDS likelihood, PFS and OS with confidence interval, 3.1–7.2). Conclusions: Our real-world data analysis
neo-adjuvant chemotherapy. Patient KELIM is calculable online, based on shows that 56% of BRCAwt pts who received platinum-based treatment
observed CA125 values, on http://www.biomarker-kinetics.org/. Clinical trial without maintenance therapy had recurrent OC within 6 months, classifying
information: 2011-006288-23. them as platinum resistant. Use of maintenance treatment options, such as
niraparib, has been shown to significantly prolong PFS after platinum-based
KELIM terciles PFS months PFS HR, [95% CI] OS months OS HR, [95% CI]
CT and may be beneficial in extending the platinum-free interval, enabling
Lower [min,0.037] 8.8 Ref 22.1 Ref pts to remain eligible for further platinum therapy.
Intermediate 11.2 0.46 [0.29- 44.6 0.38 [0.22-
(0.037,0.0684] 0.74] 0.68]
Upper (0.0684,max] 19.1 0.30 [0.19- 48.7 0.31 [0.18-
0.49] 0.55]

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
326s Gynecologic Cancer

5548 Poster Session (Board #371), Sat, 1:15 PM-4:15 PM 5549 Poster Session (Board #372), Sat, 1:15 PM-4:15 PM
Risk factors for progression or death in ovarian cancer patients who completed How long have we got? The accuracy of physicians’ estimates and scenarios for
first-line platinum treatment. First Author: Shannon Neville Westin, The survival time in 898 women with recurrent ovarian cancer (ROC). First Author:
University of Texas MD Anderson Cancer Center, Houston, TX Felicia Roncolato, Macarthur Cancer Therapy Centre, Sydney, Australia
Background: Limited real-world information is available in ovarian cancer Background: Predicting, formulating, and communicating prognosis in women
(OC) regarding prognostic factors for disease progression or death after initial with ROC is difficult. Best-case, worst-case, and typical scenarios for survival
treatment. Here, we assessed potential prognostic risk factors in OC patients time based on simple multiples of an individual’s expected survival time (EST)
(pts) who completed first-line (1L) platinum-based chemotherapy (CT) using estimated by their oncologist have proven accurate and useful in a range of
real-world data. Methods: This retrospective study identified 5535 pts di- advanced cancers. We sought the accuracy and prognostic significance of such
agnosed with OC from January 2011–October 2018 from the Flatiron estimates in the GCIG Symptom Benefit Study: a multinational, prospective
database, a longitudinal, demographically and geographically diverse da- cohort study of women with ROC (platinum resistant and potentially plati-
tabase derived from health records from . 265 cancer clinics and . 2 num sensitive ROC who have had more than 2 lines of chemotherapy).
million US cancer pts. Stage III/IV OC pts who completed 1L platinum-based Methods: Oncologists estimated EST at baseline for each woman they recruited
CT after primary debulking or interval debulking surgery were included. Pts to the GCIG Symptom Benefit Study in 11 countries. We hypothesised a priori
who received a poly(ADP-ribose) polymerase inhibitor (PARPi) in 1L treat- that oncologists’ estimates of EST would be unbiased (equal proportions
ment or as maintenance therapy after 1L treatment were excluded. Cox [approximately 50%] of women living longer versus shorter than their EST),
proportional hazards model was used to assess the association between imprecise ( , 33% living within 0.75 to 1.33 times their EST), and provide
baseline factors (neoadjuvant CT, disease stage, residual disease, BRCA accurate scenarios for survival time (approximately 10% dying within 1/4 of their
status, ECOG, age, platelet count, hemoglobin, and neutrophil count) and EST, 10% living longer than 3 times their EST, and 50% living from half to
time to next treatment (TTNT; a proxy for progression-free survival) or overall double their EST). We also hypothesised that oncologists’ estimates of EST
survival (OS) in these pts. Results: 1064 of 5535 pts were eligible per our would be independently significant predictors of survival in a multivariable Cox
inclusion/exclusion criteria. Neoadjuvant treatment, stage of disease, re- model adjusting for prognostic factors established in previous studies.
sidual disease after surgery, and BRCA mutation (BRCAmut) status were Results: Oncologists’ individualised estimates of EST in 898 women with ROC
significant prognostic factors for either TTNT or OS. Neoadjuvant chemo- were unbiased (55% of women lived longer than their EST) and imprecise (23%
therapy pts had a shorter TTNT (hazard ration [HR] = 1.37; P= .001) and OS lived within 0.75 to 1.33 times their EST). Scenarios for survival time based on
(HR = 1.64; P= .0002) than pts who underwent primary surgery after oncologists’ estimates of EST were remarkably accurate: 7% of women died
adjusting for other covariates. Stage IV pts had a shorter TTNT (HR = 1.26; within 1/4 of their EST, 13% lived longer than 3 times their EST, and 53% lived
P= .01) and OS (HR = 1.24; P= .09) than stage III pts. OS was also worse in from half to double their EST. The median EST was 12 months (range 3-70),
pts with vs without residual disease (HR = 1.27; P= .04) and worse in and median observed was 12.7 months. Oncologists’ estimates of EST were
BRCAwt than BRCAmut pts (HR = 1.37; P= .10). Conclusions: In this independently significant predictors of overall survival (HR 0.96, CI 0.94-0.98,
retrospective analysis of a real-world data set, BRCAwt status was associated p , 0.0001) in Cox models accounting for previously established prognostic
with higher risk of death. Receipt of neoadjuvant CT, higher stage of disease factors. Conclusions: Oncologists’ estimates of EST were unbiased, imprecise,
at diagnosis, or presence of residual disease after surgery were also asso- and independently significant predictors of survival time. Best-case, worst-case
ciated with a shorter TTNT or higher risk of death. These real-world data and typical scenarios based on simple multiples of EST were remarkably ac-
confirm previously identified prognostic factors. curate, and provide a useful approach for predicting, formulating, and
explaining prognosis in women with recurrent ovarian cancer. Clinical trial
information: ACTRN: 12607000603415.

5550 Poster Session (Board #373), Sat, 1:15 PM-4:15 PM 5551 Poster Session (Board #374), Sat, 1:15 PM-4:15 PM
Randomized phase III trial comparing pegylated liposomal doxorubicin Efficacy of maintenance olaparib for newly diagnosed, advanced ovarian
(PLD) 50 mg/m2 and 40 mg/m2 in patients with platinum-resistant Müllerian cancer patients (pts) by BRCA1 or BRCA2 mutation in the phase III SOLO1 trial.
carcinoma (JGOG3018). First Author: Akira Yabuno, Department of Gyne- First Author: Michael Friedlander, Prince of Wales Clinical School, University of
cologic Oncology, Saitama Medical University International Medical Center, New South Wales, and Royal Hospital for Women, Sydney, Australia
Hidaka, Japan
Background: In SOLO1 (NCT01844986), maintenance olaparib resulted in a
Background: The standard dose of single-agent pegylated liposomal doxo- significant improvement in progression-free survival (PFS) for newly diagnosed,
rubicin (PLD) is 50 mg/m2 every 4 weeks, but 40 mg/m2 has recently been BRCA1- and/or BRCA2-mutated, advanced ovarian cancer pts compared with
used in clinical practice, though there is no evidence available to support its placebo (HR 0.30, 95% CI 0.23–0.41; median not reached vs 13.8 months; Moore
use. Methods: A Phase III, randomized, multicenter, non-inferiority study et al. N Engl J Med 2018). We investigated PFS in SOLO1 for the subgroups of pts
comparing progression-free survival (PFS) of patients with platinum- with BRCA1 mutations (BRCA1m) or BRCA2 mutations (BRCA2m). Methods: All
resistant Mullerian carcinoma (epithelial ovarian, fallopian tube, or pri- pts were in clinical complete or partial response to platinum-based chemotherapy
mary peritoneal carcinoma) treated with an experimental arm (40 mg/m2 and were randomized to maintenance olaparib (300 mg twice daily; tablets) or
PLD) versus a standard arm (50 mg/m2 PLD) until 10 courses, disease placebo. After 2 years, pts with no evidence of disease discontinued study treat-
progression, or unacceptable toxicity was conducted. Eligible patients had ment, but pts with evidence of disease could continue study treatment. PFS by
# 2 prior lines. Stratification was by performance status (PS) and PFS of BRCAm was a predefined analysis. BRCAm were identified by central germline
prior chemotherapy (,3 months versus $3 months). The primary endpoint (Myriad or BGI) or local testing; Foundation Medicine testing confirmed tumor
BRCAm. Results: Median follow-up for PFS was ~41 months in the olaparib and
was PFS, and secondary endpoints were overall survival (OS), toxicity profile,
placebo arms. Of 391 randomized pts, 282 had BRCA1m (72%), 106 had
clinical response, and tolerability. The target total number of patients was
BRCA2m (27%) and three (1%) had both (Table). Two pts in the olaparib arm had
412. Results: The trial was closed due to accrual futility as patient re-
somatic BRCAm (one BRCA1m, one BRCA2m); all others had germline BRCAm. At
cruitment was slow , with 272 patients randomized to the experimental arm the primary data cut-off, 155 pts in the BRCA1-mutated group (55%), 43 in the
(n=137) and the standard arm (n=135). The final analysis was performed BRCA2-mutated group (41%) and none in the BRCA1/2-mutated group had disease
with 234 deaths and 269 events for PFS. Median patient age was 62 years; progression. The percentage of BRCA1-mutated pts who received olaparib and were
58% of patients had a treatment-free interval less than 3 months, and 81% progression-free at 1, 2 and 3 years was 86%, 69% and 53% (vs 52%, 36% and
of patients had PS 0. In the experimental versus standard arm, median PFS 26% receiving placebo) and for BRCA2-mutated pts was 92%, 85% and 80% (vs
was 4.0 months versus 4.0 months (HR 1.065, 95.8%CI: 0.830-1.366), 50%, 32% and 29%, respectively). Conclusions: Significant PFS benefit with
and median OS was 14.0 months versus 14.0 months (HR 1.078, 95%CI: olaparib versus placebo was demonstrated for all pts, regardless of whether they had
0.831-1.397). Adverse events $Grade 2 including oral cavity mucositis BRCA1m or BRCA2m. Statistical tests were not used to compare BRCA1- and
were more frequent in the standard arm than in the experimental arm (26.7% BRCA2-mutated pts, but those with BRCA2m appeared to receive greater benefit
vs. 13.5%, respectively; p=0.0089), but there was no difference in $Grade from maintenance olaparib than those with BRCA1m. Clinical trial information:
2 hand-foot-skin reactions (19.8% vs. 15.0%, respectively; p=0.333). NCT01844986.
Conclusions: The non-inferiority of PFS with the reduced dosing schedule PFS by BRCA1m and/or BRCA2m.
was not confirmed because the trial was closed prematurely, but PFS and OS BRCA1m BRCA2m BRCA1/2m
were similar. These results suggest a reduction of the standard dose of PLD
Olaparib Placebo Olaparib Placebo Olaparib Placebo
because of the low rate of oral mucositis in patients with platinum-resistant n=191 n=91 n=66 n=40 n=3 n=0
ovarian cancer treated with the lower dose regimen. Clinical trial information: Median PFS, months 41.4 13.8 NR 13.8 NR –
UMIN000003130. HR (95% CI) 0.41 (0.30–0.56) 0.20 (0.10–0.37) NC
NC, not calculable; NR, not reached

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gynecologic Cancer 327s

5552 Poster Session (Board #375), Sat, 1:15 PM-4:15 PM 5553 Poster Session (Board #376), Sat, 1:15 PM-4:15 PM
Delays from neoadjuvant chemotherapy to interval debulking surgery and Molecular stratification of endometrioid ovarian carcinomas. First Author:
survival in ovarian cancer. First Author: Ying L Liu, Memorial Sloan-Kettering Barbara Stanley, Nicola Murray Centre for Ovarian Cancer Research,
Cancer Center - Fellowship (GME Office), New York, NY Edinburgh, United Kingdom
Background: Prolonged time from primary surgery to chemotherapy is asso- Background: Endometrioid ovarian carcinomas (EC) have been historically
ciated with worse survival in ovarian cancer (OC); however, the impact of under-investigated. We sought to determine the molecular landscape
prolonged time from neoadjuvant chemotherapy (NACT) to interval debulking of contemporarily defined EC using whole exome sequencing (WES).
surgery (IDS) is unknown. Given increasing utilization of NACT, we seek to Methods: Tumours diagnosed as EC between May 1980 and December 2013
evaluate the role of delays from NACT to IDS (TIDS) on survival. Methods: At a were identified through the Edinburgh Ovarian Cancer Database. Pathology
single center, we prospectively identified 224 women with newly diagnosed review was performed according to WHO 2014. Other pathologies including
stage III/IV OC given NACT from 7/1/15 to 12/1/17. Clinical characteristics WT1 positive HGSOC were excluded. A selected cohort underwent WES and
were abstracted by two independent reviewers. Delays in TIDS were defined as were analysed for single nucleotide and copy number variants across a panel of
time from last preoperative carboplatin to IDS . 6 weeks. Fisher’s exact/ genes previously reported as mutated in endometrial, ovarian or pan cancer
Wilcoxon rank sum tests were used to compare clinical characteristics by delay studies. Tissue microarrays were stained for ER, PR and AR. Multivariable
in TIDS. Kaplan Meier method was used to estimate progression-free (PFS) and analysis for disease specific survival (DSS) was performed. Results: 125 tu-
overall survival (OS) from date of IDS. Log-rank test/multivariate CoxPH models mours were included. 61 tumours of all grades underwent WES. 5 molecular
were used to examine differences by delay groups, adjusting for covariates. groups were identified based on the presence of TP53 mutations (45.9%); or
Results: Of the 224 women, 159 underwent IDS, and 34 (21%) experienced an EClike profile (one or more mutations in ARID1A (41.0%); CTNNB1
TIDS delays. These women were older (median 68 vs. 65 years, p = 0.05) and (31.1%), PTEN (24.6%) or PIK3CA (23.0%)). Tumours with no mutations in
had more preoperative NACT cycles (median 6 vs. 4, p = 0.003). Patients with EClike genes were termed ECnull. Each group demonstrated differential DSS
delays in TIDS also had a longer interval from pathological diagnosis to start of (table). Some ECnull:TP53mut tumours also displayed mutations in KRAS, APC
NACT (TNACT), median 22 vs. 17 days, p = 0.01, and interval from IDS to and mismatch repair genes. Unsupervised clustering analysis based on the top
postoperative chemotherapy (TPOC), median 37 vs. 30 days, p = 0.01; how- 100 differentially mutated genes across the dataset validated these groups.
ever, neither TNACT nor TPOC predicted survival, p . 0.05. On univariate Somatic copy number alterations were significantly higher in the TP53mut
analysis, delays in TIDS were significantly associated with worse OS (HR 2.4 groups than the TP53wt groups (P,0.0001), and were identified across EClike
95% CI 1.2-4.8, p = 0.01); however, this was attenuated in multivariate genes in the ECnull groups. A PR histoscore of .150, but not ER or AR, was
models (HR 1.66 95% CI 0.8-3.4, p = 0.17), adjusting for age, stage and more frequent in the TP53wt group compared to the TP53mut groups
complete gross resection (CGR). On univariate analysis, delays in TIDS were not (P=0.003). TP53mut status (P=0.0182) and PR histoscore #150 (P=0.0115)
associated with PFS (HR 1.55 95% CI 0.97-2.5, p = 0.062), and in multi- were independently associated with DSS. Conclusions: EC is a heterogeneous
variate models, increase in number of preoperative NACT cycles (p = 0.005) disease comprising 5 molecular subgroups each demonstrating differential
and lack of CGR (p , 0.001) were the only variables predictive of worse PFS. clinical outcome. TP53 mutations and PR histoscore #150 are independently
Conclusions: Delays in TIDS are associated with OS, but not after adjustment associated with poor prognosis.
for age, stage and CGR, suggesting a need to maximize cytoreduction re-
gardless of delays in NACT. The role of preoperative NACT cycles on survival Group 5 year DSS Statistics
should be further studied. EClike:ARID1Awt TP53wt 100% HR=0.14[0.02-1.11], P=0.0623
EClike:ARID1AmutTP53wt 77% HR=0.24[0.08-0.67], P=0.0063
EClike:TP53mut 60% HR=0.48[0.14-1.69], P=0.2551
ECnull:TP53wt 80% HR=0.38[0.09-1.66], P=0.1977
ECnull:TP53mut 30% reference

5554 Poster Session (Board #377), Sat, 1:15 PM-4:15 PM 5555 Poster Session (Board #378), Sat, 1:15 PM-4:15 PM
Olaparib maintenance therapy in patients (pts) with a BRCA1 and/or BRCA2 Menopausal symptoms in epithelial ovarian cancer survivors: The GINECO
mutation (BRCAm) and newly diagnosed advanced ovarian cancer (OC): VIVROVAIRE2 study. First Author: Florence Joly, Centre François Baclesse,
SOLO1 China cohort. First Author: Lingying Wu, Cancer Hospital, Chinese CHU Côte de Nacre, Univ. UniCaen, Caen, France
Academy of Medical Sciences, Beijing, China
Background: We have previously shown that Epithelial Ovarian Cancer and
Background: SOLO1 (NCT01844986) is a randomized, double-blind, Phase III its treatments have significant negative effects on Quality of Life (QoL) and
trial evaluating the efficacy and safety of the PARP inhibitor, olaparib, as long term fatigue. The aim of the present multicentric VIVROVAIRE2 study
maintenance monotherapy in newly diagnosed advanced OC pts with a BRCAm. was to report the main menopausal, VasoMotor Symptoms (VMS) of Epi-
A separate pt cohort evaluated the efficacy and safety of olaparib in Chinese pts thelial Ovarian Cancer survivors (EOCS). Methods: 166 patients of the 322
in this setting. Methods: The China cohort of SOLO1 planned to enroll ~53 EOCS without relapse $3 years after first line of treatment accepted to
newly diagnosed OC pts who had completed first-line platinum-based che- participate to a gynecological consultation carried out by a gynecologist,
motherapy and were in clinical complete or partial response. This sample size including a questionnaire related to menopausal symptoms, a clinical ex-
provided around a 90% chance to observe an HR , 1, assuming a true HR = amination, and an osteodensitometry. VMS (hot flashes and/or night sweats)
0.62. Pts were randomized 2:1 to olaparib (300 mg bid; tablet) vs placebo. The were described according to natural menopause (NM) or surgically induced
primary endpoint was investigator-assessed progression-free survival (PFS; menopause (SIM). QoL and Fatigue were measured with the validated
modified RECIST v1.1). Sensitivity analysis of PFS was performed by blinded questionnaires FACT-G and FACT Fatigue. Results: Median age was 62 years
independent central review (BICR). Results: All 64 randomized pts received [20-83], FIGO stage III/IV (48%) and , 10% BRCA1&2 mutated. Histo-
study treatment (olaparib, n = 44; placebo n = 20). Median follow-up was logical subtypes were: high grade serous 31%, low grade serous 23%,
~30 months in both arms. Median PFS was not reached in the olaparib arm and endometrioid G2-3 (14%) endometriod G1( 4%), clear-cell 21%, mucinous
was 9.3 months in the placebo arm (Table). The most common AEs in the 5%. All EOCS had surgery, 97% of patients received platinum and taxane
olaparib group were nausea (n = 28, 63.6%), anemia (n = 25, 56.8%) and chemotherapy, median delay from treatment was 5 years [3-24] and 59
vomiting (n = 18, 40.9%). Grade $3 AEs occurred in 56.8% of olaparib pts vs (36%) had SIM. 14% of EOCS had osteoporosis; this rate was similar to the
30.0% of placebo pts; the most common grade $3 AE was anemia (n = 16, general population. 52% of patients (85) reported either hot flashes (47%)
36.4%). Olaparib dose interruptions, reductions and discontinuations occurred or night sweats (32%). 72% with SIM had VMS compared to 41% with NM
in 56.8%, 27.3% and 6.8% of pts, respectively (vs in 30.0%, 10% and 0% of (p , 0.001). VMS were not associated with poor global QoL or fatigue. At the
pts in the placebo arm). Conclusions: In the China cohort of SOLO1, a clinically time of the survey, only 8 (5 SIM & 3 NM) EOCS received hormone re-
relevant improvement in investigator-assessed PFS was observed in newly di- placement therapy (HRT). Among the 85 EOCS with VMS, 80 (94%) (38 SIM
agnosed OC pts receiving olaparib maintenance therapy. Olaparib treatment led (93%) 42 NM (95%)) did not benefit from HRT after cancer treatment.
to a 54% reduction in risk of progression or death vs placebo. The safety results Among 80 EOCS with VMS and no HRT, 25 (66%) with SIM and 34 (81%) of
were consistent with the known profile of olaparib in Chinese pts. Clinical trial NM had high grade serous, endometriod G2-3, clear cell and mucinous
information: NCT01844986. histology. Conclusions: Vasomotor symptoms are frequently reported by
PFS events, Median PFS, EOCS, particularly among surgically induced menopause patients. A ma-
n months
HR (95% CI; P value)
jority of EOCS with these symptoms might have beneficiated from hormone
Olaparib Placebo Olaparib Placebo Full analysis set replacement therapy.
(n = 44) (n = 20) (n = 44) (n = 20) (n = 64)
PFS, investigator-assessed 18 13 NR 9.3 0.46 (0.23, 0.97; 0.0320)
(48.4% maturity)
PFS, BICR (39.1% maturity) 13 12 NR 9.3 0.39 (0.17, 0.86; 0.0168)
CI, confidence interval; HR, hazard ratio; NR, not reached.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
328s Gynecologic Cancer

5556 Poster Session (Board #379), Sat, 1:15 PM-4:15 PM 5557 Poster Session (Board #380), Sat, 1:15 PM-4:15 PM
Cost-effectiveness analysis of laparoscopic disease assessment in ovarian Bevacizumab beyond progression: Impact of subsequent bevacizumab
cancer. First Author: Ross Harrison, University of Texas MD Anderson Cancer retreatment in patients with ovarian, fallopian tube, and peritoneal cancer
Center, Houston, TX after progression. First Author: Rebecca Ann Previs, Division of Gynecologic
Oncology, Duke Cancer Institute, Duke University Medical Center, Durham, NC
Background: Laparoscopic assessment of disease resectability can be useful
for treatment planning for patients [pts] with advanced ovarian cancer [OC] but Background: The Food and Drug Administration approved the use of bev-
may be associated with added cost. Methods: We performed a cost- acizumab for treatment of recurrent epithelial ovarian, fallopian tube, and
effectiveness analysis from a payer perspective to compare (1) a conven- primary peritoneal carcinoma (OC) in combination with chemotherapy. This
tional strategy, where standard new pt evaluation was used to assign pts to study evaluates whether patients immediately retreated with bevacizumab
either primary cytoreduction [PCS] or neoadjuvant chemotherapy with interval derive benefit after progressing on a bevacizumab-containing regimen.
cytoreduction [NACT], and (2) an alternative approach, where pts considered Methods: This multi-institutional, retrospective study compared patients with
candidates for PCS would undergo laparoscopy to evaluate disease re- high grade non-mucinous epithelial OC who received bevacizumab followed
sectability using a validated scoring system, who were then triaged to either directly by another bevacizumab-containing treatment regimen to patients who
PCS or NACT based on this evaluation. Diagnostic work-up, surgical and received bevacizumab followed by a regimen that did not contain bevacizumab
adjuvant treatment, perioperative complications, and progression-free survival (or received no further treatment). All patient retreated with bevacizumab had
[PFS] were included in the model. We derived model parameters from the stable or progressive disease on prior bevacizumab containing regimen.
literature and our institution’s experience with laparoscopic triage. Utility Progression free survival (PFS) and overall survival (OS) were estimated using
estimates for health states related to primary treatment were assessed pro- Kaplan and Meier product-limit estimator and modeled via Cox proportional
spectively and taken from the literature. Costs were estimated using Medicare hazards regression. PFS was measured from the date of first bevacizumab
reimbursement. Effectiveness was defined in quality-adjusted progression- treatment to the date of first progression, date of death or date of last clinic
free life years [QPFLYs]. We performed multiple sensitivity analyses. visit. OS was measured from the date of first bevacizumab treatment after
Results: Under baseline model parameters, the expected cost of treating one progression to the date of death or date of last contact/clinic visit. Statistical
pt under the conventional and alternative strategies was $26,539 and significance was defined at the 0.05 level. Results: 275 patients received
$26,653, respectively. The expected quality-adjusted progression-free sur- bevacizumab, of which 226 were evaluable; 102 received sequential treat-
vival for pts in the conventional and alternative strategies was 0.70 and 0.94 ment with bevacizumab and 124 received a bevacizumab containing regimen
QPFLYs, respectively. The calculated incremental cost-effectiveness was followed by a non-bevacizumab containing regimen at the time of progression.
$473.97 per QPFLY saved. The alternative strategy became cost saving if pts There was no significant difference between tumor grade, stage, or BRCA
found to have resectable disease by laparoscopy underwent cytoreduction mutation. Median follow-up for all subjects was 17 months (range: 1.2-
during the same procedure. The conventional strategy may be preferred if PCS 138.2 months). Median PFS was 10.21 months (95%CI: 8.05 - 11.79) and
increased PFS over NACT by $5 months. Conclusions: For newly-diagnosed median OS was 22.14 months (95%CI: 17.1 – 27.4). Median PFS for patients
advanced stage OC pts, laparoscopic assessment of disease resectability prior who received bevacizumab without retreatment was 5.1 months (95%CI: 4.3 –
to PCS was a cost-effective strategy. A conventional strategy may be preferred if 6.3) and 17.6 months (95%CI: 14.3 – 21.3) for patients who received se-
PCS produced substantially longer PFS. Sensitivity analysis suggests the quential bevacizumab retreatment (p , 0.001). Median OS for patients who
benefit of utilizing laparoscopic triage is influenced by mitigation of serious received bevacizumab without retreatment was 12.9 months (95%CI: 9.3 –
perioperative morbidity and associated costs. 16.7) and 30.1 months (95%CI: 26.1 – 35.4) for patients who received
sequential bevacizumab retreatment (p , 0.001). Conclusions: Our study
shows OC patients treated with bevacizumab-containing regimens sequentially
at the time of progression have significantly prolonged survival outcomes
compared to those patients who received no re-treatment with bevacizumab.

5558 Poster Session (Board #381), Sat, 1:15 PM-4:15 PM 5559 Poster Session (Board #382), Sat, 1:15 PM-4:15 PM
Patient preferences for maintenance PARP therapy in ovarian cancer Survival and clinical outcomes of ovarian cancer patients enrolled in phase I
treatment. First Author: Laura Jean Havrilesky, Division of Gynecologic On- clinical trials. First Author: Bradley Corr, University of Colorado, Aurora, CO
cology, Duke Cancer Institute, Duke University Medical Center, Durham, NC
Background: Ovarian cancer patients who enroll in Phase I clinical trials are
Background: Maintenance therapy with PARP inhibitors has become preva- typically platinum resistant, heavily pretreated patients with a poor prog-
lent in treating ovarian cancer. However, the preferences of women with nosis. Historically, clinical benefit of Phase I trials in this patient population
ovarian cancer regarding the risks, side effects and benefits afforded by has been uncertain. We assessed prognostic factors and survival in women
maintenance therapies are largely unstudied. Methods: A discrete choice with recurrent, previously treated ovarian cancer who enrolled in Phase I
experiment was designed to elicit the preferences of women with ovarian clinical trials. Methods: We performed a retrospective analysis of all ovarian
cancer regarding 6 attributes (levels presented in parentheses) relevant to the cancer patients who were treated on Phase I clinical trials from 2008 through
decision for maintenance PARP inhibitor therapy versus surveillance: (1) 2018 at the University of Colorado Cancer Center. Patient characteristics,
overall survival (36, 38, 42 months); (2) progression-free survival (15, 17, treatment-related toxicities and survival data were assessed. Descriptive
21 months); (3) nausea (none, mild, moderate); (4) fatigue (none, mild, statistics and Cox proportional hazards models were utilized to identify risk
moderate); (5) probability of death from myelodysplastic syndrome/acute factors associated with survival time. Results: A total of 132 individual
myelogenous leukemia (MDS/AML) (0%, 1%, 5%, 10%); and (6) monthly patients were treated on Phase I clinical trials. Patients had a median age of
out-of-pocket cost ($0, $50, $500, $1,000). Educational material was 59 years (range 33-88) with a median of 5.5 (range 1-13) previous che-
provided, with embedded questions to test respondents’ understanding. motherapy lines. 53/132 (40%) of patients were treated on multiple Phase I
Participants chose between 2 variable clinical scenarios and a constant trials with a median of 1 (range 0-5) prior Phase 1 clinical trial enrollments.
scenario representing a treatment break, with multiple iterations. Random- All patients had an ECOG performance status of 0 or 1. Overall response rate
parameters logit regression was applied to model participants’ choices as a (defined as complete or partial response) was 9% and disease control rate
function of attribute levels. Results: Of 150 women with ovarian cancer (defined as complete or partial response or stable disease as best response)
recruited, 95 were eligible and completed the survey. The mean age was 62, was 33%. Median overall survival (OS) was 11.5 months (95% CI: 9.3-
48% had recurrent ovarian cancer, and 13% were currently taking a PARP. 13.7). Two patients died on trial due to progression of disease while no
Participants always significantly preferred better clinical outcomes to worse patients died due to treatment-related toxicity. In multivariate analysis,
(except between 0% and 1% risk of MDS/AML), preferred low out-of-pocket independent risk factors predicting shorter survival were elevated CA-125
cost, and disliked the idea of a treatment break. Participants valued overall (HR 2.8; 95% CI: 1.6-5.2) and albumin , 3.5 g/dL (HR 2.5; 95% CI: 1.65-
survival most (average importance weight 25 out of 100 total), followed by 3.79). BMI . 25 predicted longer survival (HR 0.65; 95% CI: 0.44-0.96).
monthly out-of-pocket cost (24), risk of death from MDS/AML (18), nausea Conclusions: Phase I clinical trials for heavily pretreated ovarian cancer
(15), PFS (10) and fatigue (8). On average, participants would tolerate a 2% patients are safe by a standard of no patients experiencing toxicity-related
additional risk of MDS/AML in exchange for 2 additional months of PFS and deaths in our study. They are clinically efficacious with patients experiencing
7% additional risk of MDS/AML in exchange for 6 additional months of PFS. OS of 11.5 months, which is comparable to existing approved therapies.
Conclusions: Women with ovarian cancer are willing to accept the side effects Elevated CA-125 and low albumin levels predict shorter survival, while
of PARP maintenance therapy and a higher than clinically observed risk of BMI . 25 predicts longer survival. Phase I clinical trial options should be
MDS/AML in exchange for clinically observed levels of improvement in PFS. considered for all heavily pretreated ovarian cancer patients if available to
them.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gynecologic Cancer 329s

5560 Poster Session (Board #383), Sat, 1:15 PM-4:15 PM 5561 Poster Session (Board #384), Sat, 1:15 PM-4:15 PM
Effect of estrogen and progesterone receptor expression on progression-free Impact of BRCA mutation status and time to platinum resistance on patients
and overall survival outcomes in low-grade serous ovarian cancer. First Author: with advanced ovarian cancer. First Author: Alexandra Tyulyandina, Federal
Marta Llaurado Fernandez, University of British Columbia, Vancouver, BC, State Budgetary Institution «N.N. Blokhin National Medical Research
Canada Center of Oncology» pf the Ministry of Health of the Russian Federation (N.N.
Blokhin NMRCO), Moscow, Russian Federation
Background: Research on ER/PR receptor function in low-grade serous ovarian
cancer (LGSC) and the determinants of response to treatment are lacking. A Background: The influence of germline BRCA1/2 mutations (gBRCAmt) on
recent study (Sehouli et al.,2018) described ER/PR immunohistochemistry ovarian cancer patients (pts) long-term survival remains controversial.
(IHC) cut-points that distinguished PFS. Thus, we report on a group of patients Methods: 228 pts with serous and endometrial ovarian cancer stage Ic-IV were
with ER/PR expression by IHC in tumor samples of patients with LGSC and enrolled in the retrospective study. Next-generation sequencing testing of
used this information to evaluate survival outcomes. Methods: Clinical in- BRCA1/2 in blood was employed. Progression-free survival (PFS), overall survival
formation and FFPE sections were obtained from the Canadian Ovarian Ex- (OS) and time to platinum resistance (TPR) were analyzed. TPR was defined as
perimental Unified Resource (COEUR). Tissue microarray (TMA) sections were time from first line chemotherapy to registration of platinum resistance relapse.
stained for ER/PR using standard IHC techniques (MK). 50 stage 3 and 5 stage Results: The rate of pathogenic gBRCAmt was defined in 29.4% (67/228) pts.
4 patients were analyzed. ER/PR expression was scored using a simple scoring There was no any significant difference between BRCA1/2 mutation carries and
system ( , 1% cells staining, 1-50%, and $ 50%) and Allred scoring. We non-carries in both PFS (18.3 and 16.7 months, p = 0.27, HR 0.79, 95%CI
compared Kaplan-Meier (KM) survival (PFS and OS) curves using Log rank 0.52-1.20) and OS (71.9 and 79.1 months, p = 0.69, HR 0.88, 95%CI 0.46-
testing and Cox regression was used to model predictive/prognostic factors. A 1.68). However, TPR was significantly longer in pts with gBRCAmt than in
p-value of 0.05 was considered significant. Results: The mean age of the germline BRCA wild type (gBRCAwt) pts (51.4 and 34.4 months, p = 0.05, HR
population was 49.5 years (SD;13.7). Ninety percent of patients were treated 0.60, 95% CI 0.36-0.98). Pts with gBRCAmt had poor prognosis after regis-
by surgery followed by platinum-based chemotherapy (PBC). Simple scoring tration of platinum resistance. gBRCAwt pts had longer survival than gBRCAmt
did not discriminate outcomes as well for ER levels. PR Allred score ( , 2, vs after platinum-resistance relapse: 33.7 and 16.9 months respectively (p = 0.05;
2- , 6 vs $6) clearly discriminated KM curves for PFS (p = 0.036) and OS (p = HR 1.85, 95%CI 1.02-4.08). Conclusions: Our finding provided possible ex-
0.01). For Allred ER score ( , 7 vs.7- , 8 vs 8) did not distinguish PFS (p = planation of equal survival of pts with or without BRCA1/2 mutations. Long-term
0.4) but notably most patients received PBC after surgery. ER Allred score sensitivity to platinum-based chemotherapy allowed pts with gBRCA1/2mt to
significantly distinguished OS (p = 0.008). Significant factors on Cox re- control the disease for a long period of time. However the non-platinum regimens
gression for PFS were residuum (p = 0.008;95%CI:1.2-3.1) and PR (p = 0.05; had less efficacy in pts with gBRCAmt than gBRCAwt after platinum resistance.
95%CI:0.39-0.99), whereas for OS ER(p = 0.01:95%CI:0.2-0.8) and re-
siduum (p = 0.04;95%CI:1-2.8). Conclusions: ER/PR expression by Allred
scoring was associated with PFS and OS. Patients will benefit from much
needed research on ER/PR prediction/prognosis in LGSC. This work can inform
clinical trials selection/stratification and patient selection for endocrine
treatment.

5562 Poster Session (Board #385), Sat, 1:15 PM-4:15 PM LBA5563 Poster Session (Board #386), Sat, 1:15 PM-4:15 PM
Real life efficacy and safety data of bevacizumab-based front line treatment Impact of the Affordable Care Act on early-stage diagnosis and treatment for
in advance or metastatic ovarian cancer patients: Focus on patients with women with ovarian cancer. First Author: Anna Jo Smith, Johns Hopkins
malignant ascites—A phase IV study. First Author: Michail Nikolaou, Department of Gynecology and Obstetrics, Baltimore, MD
Hellenic Oncology Research Group (HORG), Athens, Greece
Background: The standard of care for Epithelial Ovarian cancer (EOC) is the
combination of a taxane plus a platinum compound (TC) whereas the addition
of bevacizumab (bev) to this regimen (TC-bev) has been shown to improve the
PFS. Patients (pts) with ascites have more aggressive disease and less overall
survival. The aim of the study was to evaluate the safety of the TC/bev regimen
in the real life clinical practice. Methods: A multi-center observational study,
approved by the ethics committees of the participating centers, including 314
pts with stage III/IV EOC, was conducted (11.2011-06.2014) in Greece. Two
independent cohorts, with similar clinico-pathologic characteristics, were
treated with front-line TC (n = 109) or TC/bev (n = 205) according to the
physician’s choice. 83 (40.5%) and 40 (36.7%) in the TC/bev and TC groups
presented with ascites. Results: Disease control was achieved in 90.7% and in
78.9% of patients treated with TC/bev and TC, respectively (p = 0.003). Pts The full, final text of this abstract will be available at
with ascites treated with TC/bev experienced a better overall response rate
(ORR) (68.7% Vs 55%) and less progression disease (PD) compared to pa- abstracts.asco.org at 7:30 a.m. ET on Saturday, June 1.
tients receiving TC (13.2% Vs 30.8%). The median PFS in all pts was 21.5mo Onsite at the Meeting, this abstract will be printed in the
and 12.4mo (p , 0.001) and median PFS in ascites pts was 18.1mo and Saturday edition of ASCO Daily News.
10.3mo in the TC/bev and TC cohort , respectively (p , 0.001). The median
OS was not reached in the TC/bev group and it was 36.9mo in the TC group,
(p = 0.059) while in the ascites pts also has not reached and it is 22.5m, re-
spectively (p = 0.023). The 3 year survival rate in all pts was 59.4% and 50.4%
and in ascites pts was 55.3% and 30% in the TC/bev and TC respectively.
Neutropenia was the most common grade 3/4 adverse event in 16.6% and
9.1% in TC/bev- and TC- treated patients (p = 0.072) with no other adverse
events . 5%. Conclusions: These real life data demonstrate that the com-
bination of TC/bev represents an active and well tolerated regimen offering
survival benefit in patients with stage III/IV EOC and especially in patients with
ascites. Additional larger prospective studies are required to confirm these
observations. Clinical trial information: NCT01982500.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
330s Gynecologic Cancer

5564 Poster Session (Board #387), Sat, 1:15 PM-4:15 PM 5565 Poster Session (Board #388), Sat, 1:15 PM-4:15 PM
Bevacizumab or PARP-inhibitors maintenance therapy for platinum-sensitive Rethinking breast cancer surveillance in women with BRCA-associated
(PS) recurrent ovarian cancer (rOC)? A network meta-analysis (NMA). First ovarian cancer in the post-solo trial era. First Author: Catherine S. John,
Author: Michele Bartoletti, Department of Medicine (DAME), University of Cedars Sinai Medical Center, Los Angeles, CA
Udine; Dipartimento di Oncologia Medica, Centro di Riferimento Oncologico di
Background: Patients with BRCA mutations are at increased risk of developing
Aviano (CRO), IRCCS, Udine, Italy
both breast (BC) and epithelial ovarian cancer (EOC). Optimal breast cancer
Background: Patients (pts) experiencing a PS rOC are generally re-exposed to a surveillance guidelines for BRCA mutation carriers following EOC has not been
platinum based-chemotherapy (CT). In this setting, the addition of a targeted defined due to high risk of EOC recurrence. The recent SOLO-1 trial
agent like bevacizumab (BEV) or PARP inhibitors (PARPi) as concomitant and/ demonstrated a survival benefit of olaparib maintenance therapy for newly
or maintenance therapy has shown to improve progression free survival (PFS). diagnosed women with advanced stage EOC. Olaparib reduced the risk of
In the absence of direct comparison in randomized trials (RCTs), we have disease-progression or death by 70% compared to placebo with a median
performed a NMA to evaluate differences in terms of efficacy between BEV and progression-free survival (PFS) of 36 months. Methods: An IRB-approved,
PARPi in pts with PS rOC, according to BRCA status. Methods: We searched multi-institutional study retrospective chart review was performed. Patients
PubMed, Embase and Medline for RCTs involving pts with PS rOC treated with had BRCA-associated EOC diagnosed between 1990-2015 without a history of
BEV (n = 3, 1563 pts) or PARPi (n = 5, 1839 pts). Only trials with PFS as prior BC or mastectomy. All women received combination chemotherapy for
primary endpoint were included. Trials in first line setting were excluded. EOC. The observed breast cancer free survival was adjusted to reflect the
Analyses have been done pooling pts who had received PARPi in three groups, enhanced 3-year PFS observed in olaparib-treated women from the SOLO-1
according to the available data on BRCA genes status: all comers (AC), BRCA trial. Kaplan-Meier survival curves were performed. Results: 191 patients with
mutated pts (BRCAm) and BRCA wild-type pts (BRCAwt). A frequentist ap- BRCA-associated EOC were included (135 BRCA1, 55 BRCA2, 1 BRCA1 and
proach has been used with R statistical software. To rank the effect size of BRCA2). Median age was 53 years. Most women had advanced stage, high-
treatments, surface under the cumulative ranking value (SUCRA) has been grade serous EOC (75%). The median overall survival was 7.7 years for BRCA
applied. Results: In AC pts, PARPi improved PFS compared to BEV (hazard 1, and 9.7 years for BRCA2 mutation carriers. Annual mammography and MRI
ratio [HR] = 0.70, 95% CI 0.54-0.91, test of heterogeneity [I2] = 40.5%). In were performed in 43% and 34% of women, respectively, with a median of 4
BRCAm pts the gain in PFS for PARPi was even higher compared to BEV (HR = mammograms and 3 MRI per patient. 16 women (8.3%) were diagnosed with
0.46, 95% CI 0.36-0.59, I2= 17.2%). In BRCAwt pts the benefit of PARPi BC over a median follow up of 80 months: 7 (44%) DCIS and 9 (56%) invasive
over BEV was not statistically significant (HR = 0.87, 95% CI 0.63-1.20, I2= ductal carcinoma. 14 (88%) women had early stage (0-2) BC. 28 (15%) of
35.7%) but PARPi had the highest likelihood of being ranked as the best women had risk-reducing mastectomy performed an average of 2.1 years
treatment in terms of efficacy according to SUCRA (90% and 60%, re- following their EOC diagnosis. The incidence of BC increased from 5.6% to
spectively for PARPi and BEV). Hazard ratio for PFS between PARPi, BEV and 11% at 5- and 10-years post EOC, and in the predicted model with olaparib,
CT in the three cohorts are reported in the table. Conclusions: According to from 10% to 17% at 5- and 10-years, assuming olaparib does not impact
indirect comparisons, PARPi performed the best for the treatment of PS rOC, breast cancer incidence. Conclusions: The risk of metachronous BC following
especially in BRCAm pts who had not previously received PARPi. BEV could be BRCA-associated EOC increases over time. In the post SOLO trial era, BC
still an option in BRCAwt pts. surveillance strategies in women with EOC should be optimized to reflect
improved outcome.
Treatments AC BRCAm BRCAwt
Incidence of Breast Cancer
PARPi vs BEV 0.70 (0.54-0.91) 0.46 (0.36-0.59) 0.87 (0.63-1.20)
PARPi vs CT 0.38 (0.31-0.47) 0.25 (0.21-0.31) 0.48 (0.36-0.63) Time since EOC Diagnosis PRIOR to SOLO-1 Predicted POST SOLO-1
BEV vs CT 0.55 (0.31-0.47) 0.55 (0.48-0.63) 0.55 (0.47-0.64) 2-year 2% 5.6%
5-year 5.6% 9%
10-year 11% 17%

5566 Poster Session (Board #389), Sat, 1:15 PM-4:15 PM 5567 Poster Session (Board #390), Sat, 1:15 PM-4:15 PM
Real-life data: Women with recurrent platinum-sensitive ovarian cancer and Everolimus plus letrozole treatment of recurrent gynecologic cancers. First
BCARGEM treatment. First Author: Frederique C. van der Scheun, De- Author: Kenneth David Miller, University of Maryland Marlene & Stewart
Department of Medical Oncology, Cancer Center University Medical Center Greenebaum Comprehensive Cancer Center, Baltimore, MD
Utrecht, University of Utrecht, Utrecht, Netherlands
Background: Hormonal therapy has limited activity in gynecologic (Gyn)
Background: Ovarian cancer is still the most mortal gynaecological cancer in the cancer treatment. mTOR inhibitors plus aromatase inhibitors (AI’s) improve
world. Even after achieving a good clinical response after initial treatment, the the response rate and response duration in breast cancer patients. We studied
cancer will relapse in 80% of the patients. Treatments that improves progression- this combination in heavily pre-treated women having estrogen receptor
free survival (PFS) are necessary. In 2012, the OCEANS trial showed an im- positive (ER+) Gyn cancers. Methods: This phase II study combines everolimus
proved PFS (12.4 months) for patients with platinum-sensitive recurrent ovarian, and letrozole for ER+ Gyn cancers with disease progression following primary
primary peritoneal, or fallopian cancer (ROC) treated with carboplatin, gemci- and salvage chemotherapy. 19 patients participated (Ovary-10, Endometrium-
tabine plus the monoclonal antibody bevacuzimab followed by maintenance 7, and Primary Peritoneal Cancer-2). The mean age was 64, prior lines of
therapy of bevacuzimab (BV) till progression (BCARGEM). Based on this trial, therapy ranged from 2-7, and median time from diagnosis to study entry was
bevacizumab was incorporated in the second line treatment protocol of various 67 months (m) (range 10-348m). Results: There were no complete re-
European countries for patients with platinum-sensitive ROC. However, there is sponders, but 7 of 19 (37%) patients treated had clinical benefit, with 1 PR
no real life information yet available on the effectivity and tolerance of this one and 6 with stable disease. In responding patients, the earliest time to best
randomized clinical trial (RCT) in clinical practise. Therefore, the aim of this response was 2m and the median time to progression was 5m (range 5-40+m).
study was to assess what the real time efficacy and tolerance of treatment with The mean number of treatment cycles was 11. Toxicities: The most common
BCARGEM was in patients with platinum-sensitive ROC. Methods: All patients adverse events were: hyperglycemia, rash, stomatitis, fatigue, and anemia. 7
with platinum-sensitive ROC and treated with BCARGEM in the UMC Utrecht patients required dose reductions, and 2 discontinued study drugs due to
Cancer Center were retrospectively selected for analysis. All data such as baseline pneumonitis. The therapy was generally well tolerated in both women , 65
information, assessments and adverse events during cycles were obtained from and . 65 years old. Toxicities were less common with reduced everolimus
medical records. The primary outcome was PFS; secondary outcomes were doses. Conclusions: Recurrent Gyn cancers become refractory to chemo-
relative dose intensity (RDI), adverse events (AE) leading to dose modifications,
therapy. We tested the combination of everolimus plus letrozole in heavily
and overall survival (OS). Results: Overall, the median PFS of the 39 patients
pretreated patients with ER + recurrent disease. 1 patient had a PR and 6 had
with platinum-sensitive ROC and treated with BCARGEM was 9.4 months (95%
stable disease with a range of 5-40+m. This oral regimen was generally well
CI 4.6 - 14.2). The median OS was 20.9 months (95% CI 15.0-26.7). The
tolerated and allowed time without IV chemotherapy, while providing clinical
average RDI for BCARGEM was 67%. None of the patients reached a RDI of
100%. In most patients (69%) neutropenia grade $ 3 led to dose modifications
benefit to patients with refractory ER+ Gyn cancers.
or discontinuance of BCARGEM. After treatment with BCARGEM, 34 patients
started treatment with BV till progression. The median number of cycles of BV
was 5. In 26 of the cases the BV was ended due to progression. Conclusions: In
our retrospective study, the PFS of real life patients treated with BCARGEM is
lower compared to the study patients in the OCEANS trial; 9.4 versus
12.4 months. In addition, the RDI of BCARGEM is very low reflecting the many
AEs leading to dose modifications or discontinuation. Therefore, this study
underlines the fact that results of RCTs with strict in- and exclusion criteria do not
represent actual outcomes in clinical care, because the selected study patient
does not match the patient we face in real time.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gynecologic Cancer 331s

5568 Poster Session (Board #391), Sat, 1:15 PM-4:15 PM 5569 Poster Session (Board #392), Sat, 1:15 PM-4:15 PM
Elucidation of PARP inhibitor activity in BRCAwt recurrent ovarian cancer by Multi-parametric FDG PET/MRI as an early predictor of response to neo-
hrr mutational gene profile analysis. First Author: Mansoor Raza Mirza, Nordic adjuvant chemotherapy in patients with epithelial ovarian cancer. First
Society of Gynecologic Oncology (NSGO) and Rigshospitalet University Hos- Author: Melissa Kristen Frey, Weill Cornell Medical College, New York, NY
pital, Copenhagen, Denmark
Background: For patients with ovarian cancer undergoing neoadjuvant che-
Background: Niraparib is an oral, selective poly(ADP-ribose) polymerase in- motherapy, the effectiveness of treatment is not evaluable by conventional
hibitor (PARPi) approved for maintenance treatment of BRCA mutated methods until all or much of the treatment has been given. The purpose of this
(BRCAmut) and BRCA wild-type (BRCAwt) recurrent ovarian cancer patients study is to investigate the performance of FDG PET, dynamic contrast-
(pts) who are in response to platinum-based chemotherapy. In the non- enhanced (DCE) and intra-voxel incoherent motion (IVIM) MR as early pre-
germline BRCA mutated (non-gBRCAmut) cohort of the ENGOT-OV16/ dictors of treatment response. Methods: Subjects with a new diagnosis of
NOVA trial, clinical benefit with niraparib vs placebo was seen in pts re- epithelial ovarian cancer underwent 3 cycles of standardized chemotherapy
gardless of their Myriad myChoice HRD test status (BRCAmut and homologous followed by cytoreduction. FDG PET/MR including DCE and IVIM was per-
recombination deficiency [HRD] score), with a hazard ratio (HR) of 0.38 in formed at baseline (T0), after cycle 1 (T1) and after cycle 3 (T2) of chemo-
HRD-positive (HRDpos) and 0.58 in HRD-negative (HRDneg) pts. To de- therapy. Final responses were categorized at T2 by RECIST 1.1. Image
termine if treatment benefit in HRDneg pts may result from mutations in other volumes at T1 were analyzed as predictors of final response. Parametric images
homologous recombination repair (HRR) genes, we examined the relationship of molecular diffusion restriction (D), tissue perfusion (D*), vascular volume
between progression-free survival and other HRR gene mutations in the NOVA fraction (F), blood- . interstitium constant of transfer (Ktrans), interstitum- .
non-gBRCAmut cohort. Methods: A retrospective, exploratory biomarker plasma constant of transfer (Kep), extravascular/extracellular volume % (Ve)
analysis was conducted using all available tumor samples from 331 pts en- and plasma volume % (Ve) were investigated along with routine measures of
rolled in the NOVA non-gBRCAmut cohort. Mutation status of HRR genes was SUV and ADC. Results: Nine subjects were enrolled, 8 were responders by
evaluated using a 43-gene NGS assay (Myriad Genetics), including BRCA1/2 RECIST at T2 and one had stable disease. At T0 the mean, min, and max
and 16 additional HRR genes. Results: In this exploratory analysis of the NOVA SUVmax of dominant tumor deposits was 11.5, 6.3, 19.0, respectively. Mean,
non-gBRCAmut cohort, niraparib demonstrated clinical benefit in pts with min, and max values were 1.0, 0.75 and 1.63 for ADCmean and 0.62, 0.30,
somatic BRCA mutation (HR, 0.27) and in BRCAwt pts (HR, 0.47). In ad- 0.96 for ADCmin. At T1, ADCmean increased in 8 subjects by +0.22% (s.d. +/-
dition, BRCAwt pts with other HRR gene mutations also derived benefit from 13%) and decreased by -3% in one subject. ADCmin increased in 8 subjects by
niraparib (HR, 0.31), as did BRCAwt/HRRwt pts (HR, 0.49). When BRCAwt/ +21% (s.d. +/-11%) and decreased by -23% in one subject. D increased for 8
HRRwt pts were categorized by HRD score, clinical benefit was also observed subjects (average +29% s.d. +/- 13%) and decreased by -10% in one. D*, F,
in both HRDpos and HRDneg pts, with HRs of 0.33 and 0.60, respectively. Kep, Ktrans, Ve and Vp had no recognizable pattern. At T2, SUVmax, SUVmin,
These results suggest that, although these biomarkers have good positive and ADCmean maintained their change direction across all subjects with
predictive value, they are not good negative predictors for niraparib benefit in measurable lesions. The only subject with a complete response at T2 had the
this indication. Conclusions: This retrospective, exploratory analysis of the highest ADCmin and ADCmean change at +45% after one cycle of chemo-
ENGOT-OV16/NOVA non-gBRCAmut cohort suggests that although pts with therapy (T1). The subject with stable disease at T2 had no significant dif-
somatic BRCA mutation and other HRR mutations benefit from niraparib ference in changes amongst all metrics. Conclusions: FDG PET/MR SUVmax
treatment, clinical benefit is also seen in HRDneg pts without HRR mutations, and ADCmean values obtained after one cycle of neoadjuvant chemotherapy
perhaps related to other genomic, epigenetic, or functional alterations within were consistently associated with partial anatomical treatment responses after
ovarian tumors yet to be defined. three cycles. Molecular diffusion restriction also was reliably associated with
treatment response. Future studies evaluating FDG PET/MR in platinum-
resistant patients may allow for early discontinuation of ineffective and
toxic treatment.

5570 Poster Session (Board #393), Sat, 1:15 PM-4:15 PM 5571 Poster Session (Board #394), Sat, 1:15 PM-4:15 PM
Correlation of surgeon radiology assessment with laparoscopic scoring in Comprehensive genomic analysis of mucinous ovarian cancer reveals unique
patients with advanced-stage ovarian cancer. First Author: Nicole D. therapeutic vulnerabilities. First Author: Dane Anthony Cheasley, Peter
Fleming, The University of Texas MD Anderson Cancer Center, Houston, TX MacCallum Cancer Centre, North Melbourne, Australia
Background: To determine the correlation between surgeon radiology assessment Background: Mucinous ovarian carcinoma (MOC) is a rare subtype of epithelial
and laparoscopic scoring in patients with newly diagnosed advanced stage ovarian ovarian cancer that responds poorly to ovarian chemotherapies and has an
cancer. Methods: Following IRB approval, 14 gynecologic oncologists from a unknown etiology. It is diagnostically challenging and can be confused with
single institution performed a blinded review of radiology imaging from 20 patients metastases from gastro-intestinal tract primaries. The GAMuT study is a multi-
with advanced stage ovarian cancer. All patients previously underwent laparo- national effort to understand molecular drivers and cell of origin of this rare
scopic scoring assessment to determine primary resectability at tumor reductive tumour, including identification of a genetic progression model and novel
surgery (TRS) using a validated scoring method from April 2013 to December therapeutic options. Methods: We performed RNAseq (n = 67), exome se-
2017. The patients with predictive index value (PIV) scores , 8 were offered quencing (n = 61), SNP arrays (n = 67) and whole genome sequencing (n = 5)
primary surgery and those with a score $8 received neoadjuvant chemotherapy on MOC and precursor lesions. A subset of ~500 genes was further evaluated
(NACT). Surgeons viewed contrasted CT imaging reports and images from all by targeted sequencing, including 129 MOC, 23 borderline mucinous tumours
patients in a blinded fashion and recorded PIV scores using the same validated (non-invasive) and 23 extra-ovarian mucinous metastases. Immunohisto-
scoring method. Linear mixed models were conducted to calculate the correlation chemistry data was collected for CK7, CK20, ER, PAX8, p53 and HER2 (n =
between radiology and laparoscopic score for each surgeon and as a group. Once 162-256). Extensive pathology review was performed and associated clinical
the model was fit, the inter-class correlation (ICC) and 95% confidence interval data obtained. Results: Comparison with TCGA and other data sets showed
was calculated. Results: Radiology review was performed on 20 patients with that MOC are distinct from mucinous tumours from other organs, including
advanced stage ovarian cancer who underwent laparoscopic scoring assessment. colorectal, appendiceal and gastric cancers. Our data supports a clear genetic
Most patients had stage IIIC disease (85%) and median laparoscopic score was 9 progression model from benign and borderline precursors to both low- and
(range 0-14). Surgeon faculty rank included Assistant Professor (n = 5), Associate high-grade MOC. TP53 mutation, ERBB2 amplification and increasing copy
Professor (n = 4), and Professor (n = 5). Median surgeon experience during the
number changes were key events associated with progression to invasive
study period with laparoscopic assessment was 13 cases (range 1-28) and TRS
disease, including a novel amplicon on 9p13. Copy number aberration burden
was 22.5 cases (range 2-48). The kappa inter-rater agreement was -0.017 (95%
was significantly associated with poor survival. We identified several recurrent
CI 0.023 to -0.005) indicating low inter-rater agreement between radiology review
mutational events suggesting utility of an existing targeted therapy, including
and actual laparoscopic score. The ICC in this model was 0.06 (0.02-0.21)
indicating that surgeons do not score the same across all the images. When using a
ERBB2 amplification (26%), ERBB3 mutation (4%) and BRAF mutation
clinical cutoff of PIV of 8, the probability of agreement between radiology and (9%). MOC could be included in clinical trials for novel agents targeting TP53
actual laparoscopic score was 0.56 (95% CI: 0.49-0.73). Number of laparoscopic missense mutation (46%), RNF43 mutation (12%), PIK3CA mutation (8%)
cases, TRS cases, or faculty rank was not significantly associated with agreement. and KRAS/NRAS mutations (66%). Other frequent events included CDKN2A
Conclusions: Surgeon radiology review did not correlate highly with actual lap- inactivation (57%), ARID1A mutation (9%) and TP53 inactivating mutations
aroscopic scoring assessment findings in patients with advanced stage ovarian (15%). Conclusions: MOC of any grade can derive from a primary ovarian
cancer. 44% of patients in our study may have been inadequately triaged by tumour precursor, and is distinct from extra-ovarian metastases. MOC is ge-
radiology review alone, which may have led to suboptimal TRS. Our study netically diverse and advanced disease should be assessed for targetable
highlights the utility of laparoscopic scoring assessment to determine resectability mutations which may provide novel therapeutic options.
over radiology assessment alone in ovarian cancer.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
332s Gynecologic Cancer

5572 Poster Session (Board #395), Sat, 1:15 PM-4:15 PM 5573 Poster Session (Board #396), Sat, 1:15 PM-4:15 PM
BRCA tumor test in ovarian cancers: The changing role of molecular pathology T cell trafficking within the peritoneal tumor environment and ovarian cancer
in the era of PARP inhibitor (PARPi) therapy. First Author: Caterina Fumagalli, progression. First Author: Sharina Palencia Desai, UNM Comprehensive
Division of Pathology, European Institute of Oncology IRCCS, Milan, Italy Cancer Center, Albuquerque, NM
Background: The PARPi Olaparib has been approved in maintenance setting Background: Ovarian cancer is associated with high mortality due to de-
of recurrent platinum-sensitive and BRCA mutated, ovarian cancer (OC) pa- tection at late stages with widespread peritoneal metastases at diagnosis in a
tients. However, according to the striking results of SOLO 1 trial, it could be majority of patients. Ovarian cancer recurrences are primarily found in the
shortly proposed even to newly diagnosed BRCA mutated OC women. We report peritoneal cavity, and peritoneal disease is the primary cause of morbidity
the results of the implementation of tumor BRCA test in diagnostic setting and mortality in this disease. This pattern of tumor engraftment and re-
within the frame of institutional workflow for the management of OC patients. currence indicates that the peritoneal tumor environment is distinct from
Methods: 223 women with OC were consecutively referred over 25 months to other niches. Our data indicate that selective peritoneal dissemination is
tumor BRCA test. The test was requested by gynecologic oncologist at the immunologically mediated, based on functional differences in peritoneal
diagnosis of non-mucinous and non-borderline OC, upon discussion on the and systemic T cells during ovarian cancer progression. This is consistent
implications of the test result and written consent collected for each patient. with data in other solid tumors demonstrating that tumor immunity is driven
Formalin-fixed paraffin embedded (FFPE) specimens of OC were tested using by regional lymphocyte populations. However, less is known about the
the automatized “Oncomine BRCA Research Assay” Next Generation Se- mechanisms that establish a permissive immune environment in the peri-
quencing (NGS) panel. The tumor BRCA test was also performed on 5 ar- toneal cavity. We hypothesize that pathways regulating T cell recruitment
chetypal cytological samples from ascites. Results: All the cases were and retention in the peritoneal cavity (PC) are co-opted by ovarian cancer
considered adequate for the NGS analysis according to the tumor cell content cells to enable intraperitoneal cancer dissemination and recurrence.
(more than 10%) and the DNA yield extracted (more than 10 ng). The tumor Methods: We have developed a novel model that uses direct in vivo labeling
BRCA test had a successful rate of 99.1% . The median Turn-Around Time of peritoneal cells in an established immune-competent high grade serous
(TAT) was 17 calendar days, from 33 days of the first trimester to 14 days of the murine cancer model. This functional approach enables us to identify T cell
last trimester of this analysis. Overall BRCA1 or BRCA2 pathogenic (P)/likely subsets retained in the PC with tumor engraftment and progression.
pathogenic (LP) mutations were found in 62 (28.1%) cases and variants of Results: We identified high expression of CD49d (a4 integrin) as the most
uncertain significance (VUS) in 25 (11.3%) cases, including 3 cases with a prevalent cell surface marker on T cells retained in the peritoneal cavity,
BRCA1 P variant and a concurrent BRCA2 VUS alteration. In detail, 47 P/LP consistent with prior published data in healthy mice and people. We
variants and 16 VUS were identified in BRCA1 whereas 15 P/LP mutations and demonstrated a functional role for CD49dhi in T cell retention by showing
9 VUS occurred in BRCA2. Complete concordance in tumor BRCA test results preferential binding to VCAM. A role for tumor cells mediating this in-
were seen between ascites cytological samples and matched tumors. teraction was observed based on enhanced binding affinity in vitro with
Conclusions: The tumor BRCA test could be implemented in routine diagnostic tumor monolayers. The importance of this mechanism is supported by high
setting, at diagnosis of non-mucinous and non-borderline OC. The test could be VCAM expression in multiple murine and human ovarian cancer cell lines,
performed on FFPE specimens, had an high successful rate and a TAT and T cell localization to VCAM rich areas within ovarian cancer tumors in
compatible with clinical needs. The promising data on cytological samples will vivo. Conclusions: CD49d not only defines a lymphocyte subset with a
be confirmed in larger series. significant role in tumor immunity but presents itself as an important po-
tential therapeutic target to modulate T cell trafficking.

5574 Poster Session (Board #397), Sat, 1:15 PM-4:15 PM 5576 Poster Session (Board #399), Sat, 1:15 PM-4:15 PM
The Circulating Cell-free Genome Atlas (CCGA) Study: Follow-up (F/U) on DPX-Survivac and intermittent low-dose cyclophosphamide (CPA) with or
non-cancer participants with cancer-like cell-free DNA signals. First Author: without epacadostat (E) in the treatment of subjects with advanced recurrent
Allen Lee Cohn, Rocky Mountain Cancer Center, US Oncology, Denver, CO epithelial ovarian cancer (DeCidE1 trial): T cell responses and tumor infiltration
correlate with tumor regression. First Author: Janos Laszlo Tanyi, University of
Background: A noninvasive cell-free DNA (cfDNA)-based cancer detection
Pennsylvania, Philadlephia, PA
assay offers the hope of a blood test that might reduce morbidity and mortality
of cancers, particularly those without recommended screening tests (eg, some Background: DPX-Survivac is a novel T cell activating therapy designed to
gynecologic cancers). CCGA (NCT02889978) is a prospective, multi-center, elicit an effective immune response against recurrent ovarian cancers that
longitudinal, case-control study evaluating models for discriminating cancer express the survivin protein. The survivin specific T cells induced by DPX-
versus non-cancer. Here, we report F/U of control participants (pts) who Survivac can infiltrate the tumors and are associated with clinical responses. It
demonstrated a cancer-signal in CCGA. Methods: Clinically evaluable samples is likely that achieving an anti-tumor effect requires a favorable ratio of T cells
(N = 2508) from pts enrolled without a cancer diagnosis (dx; NC) and to tumor cells. Epacadostat (E) is an IDO1 enzyme inhibitor that may enhance
treatment-naive pts with newly diagnosed cancer (C) were divided into training effector T cell proliferation, shifting the tumor microenvironment (TME) away
(n = 1564; 580 NC, 984 C) and test (n = 944; 368 NC, 576 C) sets. from an immunosuppressive state toward one supporting productive immune
Classification performance (cancer/non-cancer) was assessed via 3 prototype response. Methods: Recurrent ovarian cancer patients with advanced and
assays: whole-genome bisulfite (WGBS), whole-genome (WGS), and targeted metastatic progressive disease were treated with DPX-Survivac, intermittent
(507 gene) sequencing. Notable outlier NC pts were identified with cancer-like low dose CPA with or without E. In the Phase 1b, 53 subjects were enrolled to
scores in either $2 assay classification results or by the presence of known receive DPX-Survivac, low dose CPA and E BID. In the Phase 2, 12 subjects
cancer drivers with $1 assay classification result suggesting cancer. All pts are were randomized to receive DPX-Survivac and low dose CPA with or without E.
currently in F/U in accordance with study protocol (to date: 80% with . 10 mo The data on immunological responses, biomarkers, and clinical responses
and 15% with . 22 mo F/U). Results: Among training and test sets, 8 ( , 1%) were analyzed in relation to the baseline sum of target lesions per RECIST 1.1.
NC pts were identified with a cancer-like signal. To-date, 2 have been di- Results: The study showed that DPX-Survivac and intermittent low dose CPA
agnosed with a gynecologic malignancy: 1 stage IIIc clear cell endometrial with or without E can generate strong T cell responses. The infiltration of
carcinoma and 1 stage IIIc ovarian cancer, 3 and 2 months (mo) post- tumors with survivin-specific T cells correlates with the observed tumor re-
enrollment [PE], respectively. Among C pts in the study, sensitivity (at 98% gression. The sum of target tumor measurements at baseline by RECIST 1.1
specificity; WGBS) in these cancer types was: uterine/endometrial: 11% (n = correlated with observed clinical benefits. In the group of 15 patients with the
27 train) and 22% (n = 9 test); ovarian: 82% (n = 17) and 71% (n = 7). In baseline sum of target lesions less than 5 cm, all subjects have shown clinical
addition, a third NC pt was diagnosed with a stage IV lung cancer 15 mo PE. benefits. Four of these subjects reached partial response and remained without
Conclusions: This cfDNA-based assay detected a cancer-like signal that progression over a prolonged period. Conclusions: The treatment studied leads
anticipated a clinical presentation of cancer in undiagnosed pts as early as to strong survivin-specific T cell responses. Infiltration of tumors by survivin-
15 months prior to the actual dx. High specificity ( . 99%) requires ac- specific T cells correlated with clinical benefit in treated subjects. A predictive
counting for undiagnosed cancers in study design and analysis. Together, model based on tumor size to improve response to DPX-Survivac in recurrent
these data suggest that this prototype assay may have high performance ovarian cancer is being prospectively explored. Clinical trial information:
detecting a variety of gynecological and other cancers. Clinical trial in- NCT02785250.
formation: NCT02889978.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gynecologic Cancer 333s

5577 Poster Session (Board #400), Sat, 1:15 PM-4:15 PM 5578 Poster Session (Board #401), Sat, 1:15 PM-4:15 PM
Association of Ki67 expression levels and therapy outcome in low-grade serous Real-world bevacizumab utilization and outcomes in first-line ovarian
ovarian cancer. First Author: J.P. P. Grabowski, Charité Campus Virchow- cancer. First Author: Matthew J. Monberg, Merck & Co., Inc., Kenilworth, NJ
Klinikum, Berlin, Germany
Background: Bevacizumab (B) is approved in combination with carboplatin and
Background: Low-grade serous ovarian cancers (LGSOC) characterize differ- paclitaxel, followed by B monotherapy, for the treatment of advanced ovarian cancer
ent clinical pattern and lower chemotherapy responsiveness. The expression (OC) following surgery. We sought to describe B utilization and outcomes of B in first-
level of Ki67 is associated with prognosis differences in this patient group. line (1L) OC within the US and EU. Methods: This cross-sectional study included
However, Ki67 has not been evaluated as prognostic marker and a predictor of patients who were actively receiving treatment for OC. Data were collected at a single
therapy outcome until now. Methods: Patients with LGSOC and Ki67 ex- time point from 2496 patient forms between December 2017 and March 2018 from
340 oncologists/ gynecologists across the US, France, Germany, Italy, Spain, and
pression results were identified in institutional database. Receiver-operator
the UK. Patients were platinum sensitive if progression was noted . 6 months after
characteristics (ROC) curve analysis was performed to find cut off values of
frontline platinum therapy and resistant if the interval was 0-6 months. This analysis
Ki67% to discriminate patients with residual tumor mass after surgery from included patients who received 1L chemotherapy (chemo) with no maintenance or
maximal debulked patients, and platinum sensitive patients from platinum 1L chemo with bevacizumab maintenance. Results: B was used in combination with
resistant patients. Odd ratios (OR) and 95% confidence intervals (95% CI) chemo at 1L and as 1L maintenance monotherapy in 11% total study patients.
were calculated using univariate and multivariate logistic regression analysis. Those receiving 1L + B were more likely to have Stage IV disease, have good
Two-sided tests p , 0.05 and are considered statistically significant at a 95% performance status (PS) at diagnosis, and receive BRCA testing than patients re-
confidence interval. The statistical analysis was performed with the IBM SPSS ceiving chemo only. Treatment response, platinum sensitivity, and activities of daily
Statistics 25.0. Results: A total of 68 patients with LGSOC were included. All living are shown in the Table. Results did not vary by BRCA status. Conclusions: This
patients underwent surgery and 15 (22.1%) patients had residual mass study highlights differences in patient characteristics and outcomes between pa-
( . 0 mm) after cytoreduction. Sixty-one (89.7%) patients received platinum tients receiving/ who received 1L chemo only and those receiving/ who received B,
based first-line chemotherapy. Forty-three patients revealed a recurrence however, this study was not designed to formally compare 1L treatment options.
$6 months and eleven , 6 months. Patients with Ki67 , 3.6% had sig-
nificantly higher therapy-free interval (TFI$6 months), (OR = 13.9, 95%CI 1L Chemo Only Treatment + 1L Chemo + B + B
1.62-118.40, p = 0.016). In the multivariate analysis of TFI over 6 months Total No 1L maintenance Maintenance
including CA 125, age at diagnosis, peritoneal carcinomatosis and ascites N who initiated 1L 1498 1232 266
Mean age, yrs 64 64 61
( . 500ml) Ki67 , 3.6% remained significant (OR = 17.6, 95%CI 1.56- Stage IV, % 65 63 76
197.52, p = 0.020). Moreover, Ki67% . 3.6% were associated with higher ECOG PS 0-1 at diagnosis, % 81 78 92
risk of residual mass after surgery (OR = 6.75, 95%CI 1.39-32.87, p = BRCA tested, % 52 47 71
BRCA positive of those screened, % 24 25 22
0.018). Conclusions: It is the first study showing association between Some decrease in activities of daily living, % 73 73 71
Ki67% expression and duration of TFI to platinum-based chemotherapy as N who completed 1L 945 742 203
Complete response, partial response or stable 83 79 96
well as outcome of the surgery in LGSOC. Further prospective trials should be disease at completion of 1L, %
planned to develop predictive models in this patients. Platinum sensitive, % 40 35 58
Platinum resistant, % 14 13 15
Platinum refractory, % 14 17 4
Received 2L platinum, % 61 58 72
Received 2L nonplatinum, % 39 42 28
Mean 1L completed treatment duration, months 4.9 5.1 (434) 4.4 (200)
(N) (634)
Mean 1L completed maintenance duration, 8.3 NA (NA) 8.3 (149)
months (N) (149)
Mean time off treatment between 1-2 years of 6.3 8.4 (142) 2.7 (79)
diagnosis, months (N) (221)

5579 Poster Session (Board #402), Sat, 1:15 PM-4:15 PM 5580 Poster Session (Board #403), Sat, 1:15 PM-4:15 PM
Real-world data analysis of ovarian cancer (OC) maintenance utilization Clinical outcome of sequential chemotherapy after immune checkpoint
among maintenance eligible patients. First Author: David Garofalo, Integra inhibitors in advanced ovarian cancer. First Author: Luisa Bonilla, Princess
Connect, West Palm Beach, FL Margaret Cancer Centre, University Health Network, Toronto, ON, Canada
Background: Approximately 1% of US women will be diagnosed with epi- Background: Immunomodulation through check point inhibition is an im-
thelial OC during their lifetime. OC patients who achieve a response to portant treatment strategy in many cancers. In ovarian cancer (OC) response
platinum-based chemotherapy may benefit from maintenance therapy, with rates with immune checkpoint inhibitors (ICI) alone are around 10%. Che-
the goal of inducing a lasting remission or extending the time interval before motherapy antitumoural effect is driven by cytotoxicity and immunomodulatory
progression without any deleterious impact on quality of life1. This analysis, effect. ICI treatment reduces tumour induced immune-tolerance improving
based on real world data sourced from US community oncology practices, immunocompetence, essential for chemotherapy effect. We chose to in-
was designed to assess the current utilization of maintenance therapy among vestigate clinical outcomes of chemotherapy post ICI in women with OC.
maintenance eligible patients. Methods: This analysis utilized the Integra Methods: The Tumor Immunotherapy Program (TIP) database at the Princess
Data Exchange (DTX) database, a deidentified data source from community Margaret Cancer Centre identified patients with OC treated with chemotherapy
oncology practice systems (EMR, practice management, paid claims). This after ICI from 2011 to 2018. Evaluation of clinical outcomes including re-
retrospective study included 3,629 OC patients with at least two visits sponse rate (RR), progression free survival (PFS) and overall survival (OS) was
between 7/16/16 and 4/16/18. 398 patients who completed 2nd line or later assessed for pre ICI, ICI and post ICI. Results: 40 women with OC were treated
platinum-based chemotherapy for 4-9 cycles and/or had a complete/partial with chemotherapy after ICI. 90% had high grade serous histology, 7.5%
response between 1/1/17 and 7/31/18 were included. Potential mainte- carcinosarcoma and 2.5% low grade serous. Median number of pre ICI
nance therapy options were monotherapy of PARP inhibitors, bevacizumab, treatment lines was 3 (1-8) and 2 (1-6) in the post ICI setting. Median time of
and non-platinum-chemotherapy agents. Rate of maintenance therapy after ICI initiation from diagnosis was 3 years. At ICI all patients had PS 0-1 and
platinum-based treatment was assessed. Results: Our real-world analysis treated in clinical trials. 2% of the patients had platinum refractory disease,
found that 49% of 398 maintenance eligible patients received maintenance 88% had platinum resistant disease and 10% platinum sensitive disease.
therapy at least once following response to 2nd line or later platinum che- 50% were treated with ICI single agent, 16% were treated with ICI combined
motherapy. Among those that received maintenance, 46% received PARPi, with chemotherapy, 14% ICI combo and 17% ICI in combination with other
28% bevacizumab, and 26% non-platinum chemotherapy. Further, 56% of agents. Patients were treated for a median of 3 cycles (1-26). 8% experienced
women with BRCA mutations received maintenance treatment, compared PR, 18% SD, no CR were seen. 67% of the patients discontinued treatment
with 49% of women without BRCA mutations. Conclusions: Though there due to PD, 25% due to toxicity. Last treatment in pre ICI RR was 35%. First
are several options available, 51% of OC women studied who could po- treatment in post ICI RR was 30%. RR for each treatment used in post ICI was
tentially benefit from maintenance treatment did not receive maintenance. 9% for liposomal doxorubicin, 25% for single agent platinum, 29% for weekly
Only 56% of BRCA mutation carriers were targeted for maintenance in the paclitaxel and 67% for chemotherapy with bevacizumab. Median PFS in the
real world. Among patients that receive maintenance therapy following 2nd last pre ICI treatment was 6.5m and 5m in the first post ICI treatment. Median
line or later platinum chemotherapy 46% received a PARPi based regimen. PFS and OS for all the population was 53m and 54m respectively.
1) Quality of life in patients with recurrent ovarian cancer treated with Conclusions: ICI are associated with modest activity in OC, planned clinical
niraparib versus placebo (ENGOT-OV16/NOVA): results from a double-blind, trials exploring systematic sequential therapy integrating ICI, targeted agents
phase 3, randomized controlled trial. Lancet Oncol. 2018 Aug;19(8). and chemotherapy are needed.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
334s Gynecologic Cancer

5581 Poster Session (Board #404), Sat, 1:15 PM-4:15 PM 5582 Poster Session (Board #405), Sat, 1:15 PM-4:15 PM
Tumor stroma proportion to predict platinum chemoresistance in primary A phase II trial of durvalumab with or without tremelimumab in patients with
ovarian carcinomas: A prospective study. First Author: Emil Lou, University persistent or recurrent endometrial carcinoma and endometrial carcinosarcoma.
of Minnesota, Hopkins, MN First Author: Maria M Rubinstein, Memorial Sloan Kettering Cancer Center, New
York, NY
Background: Platinum chemotherapy resistance occurs in approximately 25%
of patients with ovarian carcinoma and represents a major barrier to effective Background: Monoclonal antibodies Durvalumab (D) and Tremelimumab (T)
care of this patient population. To date there are no effective nor validate inhibit binding of programmed cell death ligand 1 (PDL1) to PD1 and inhibit
predictive biomarkers of chemoresistance of ovarian carcinomas. We activation of cytotoxic T-lymphocyte-associated protein 4 (CTLA4), respec-
performed a prospective trial designed to enroll patients with ovarian masses tively, resulting in improved tumor immunosurveillance. There is rationale to
suspicious for ovarian cancer, with the goal of identifying tumor-based pre- study D and DT based on recent genomic and tumor microenvironment
dictive biomarkers of platinum resistance. Methods: 60 women were enrolled evaluation of endometrial cancer (EC). Methods: Eligible patients (pts) were
on the study. Tumor specimens were collected from 49 of these women with randomized to D or DT. Pts received D 1500 mg intravenously (IV) every
newly diagnosed pelvic masses, of which 29 were found to have histopatho- 4 weeks (wks). DT therapy pts received D 1500 mg IV every 4 wks and T 75 mg
logically proven primary ovarian carcinoma. Of these primary malignant cases, IV every 4 wks for 4 cycles, followed by D 1500 mg IV every 4 wks until
24 had specimens accessible for assessment of tumor-stroma proportion and progression or unacceptable toxicities. Pts were stratified by histology with 10
data available regarding chemosensitive vs chemoresistance status via review of carcinosarcoma or MSI-H EC pts per arm. Efficacy assessments were every 8
the medical record using a UMN IRB-approved protocol. Tumor slices were wks and treatment related adverse events (TRAEs) were assessed per CTCAE
stained with H&E and also for antibodies against two microRNAs (29b and v.4.03. The primary endpoint was overall response rate (ORR) by RECIST v1.1.
199a) differentially expressed in ovarian cancer cell lines. Tumor-stroma Descriptive statistics and 90% one sided CI are reported. Progression free
proportions were assessed by two experienced pathologists blinded to che- survival (PFS) rate at 24 wks (PFS24wks) was estimated by Kaplan Meier
moresistance status, with ,50% stroma scored as low proportion, .50% method. Results: At planned interim analysis, 56 pts were enrolled (28 per
scored as high proportion. Results: The average age of assessed patients with arm). 15 pts: carcinosarcoma, 15 pts: endometrioid (3: Gr1), 14 pts: serous,
malignant tumors was 62. 87.5% had high-grade epithelial carcinomas. and 12 pts: other histology. 5(9%) pts: MSI-H, 48(86%) pts: microsatellite
Baseline median CA-125 was 416 (range 32-2782). 80% of ovarian cancer stable (MSS), 3(5%): unknown. 2 pts were excluded due to early death. 27 pts
patients with chemoresistance had tumor stroma proportions .50%; 73.7% of per arm were evaluable for efficacy. In the D arm: 1 pt had complete response
cancer patients with chemosensitive tumors had proportions ,50% (p-value: (CR)(MSS) and 3 pts had a partial response (PR) (2:MSS, 1:MSI-H) with an
0.047). Expression of miR29b or 199a did not significantly correlate with ORR of 14.8% (CI: 6.6-100%). The median PFS was 7.6 wks and PFS24wks
chemoresistance. Conclusions: Tumor-stroma proportion is a useful predictive was 13.3% (CI 6.1-100%). Median duration of response (DOR) was 16 wks in
biomarker of platinum chemoresistance. If validated in larger datasets, it would the D arm. In the DT arm, 2 pts achieved CR (1:MSI-H, 1:MSS) and 1 had PR
be a relatively inexpensive and helpful tool for tailoring treatment strategies and (MSS). The ORR was 11.1% (CI: 4.2-100%). Median PFS was 8.1 wks,
clinical decision-making in women with ovarian cancer. PFS24wks was 18.5% (CI 10.1-100%) and DOR was 8 wks. Grade 3 TRAEs
occurred in 2 (7%) pts in D and 9 (32%) pts in DT. Grade 4 TRAEs occurred 1
(4%) pt in D and 3 (11%) pts in DT. 2 pts discontinued due to a TRAE. Most
common TRAEs in total were fatigue (23%), diarrhea (20%), nausea (14%),
vomiting (13%) and pruritis (11%). Conclusions: D and DT show modest
activity in EC. No new safety signals were identified. Second stage accrual is
ongoing. Clinical trial information: NCT03015129.

5583 Poster Session (Board #406), Sat, 1:15 PM-4:15 PM 5584 Poster Session (Board #407), Sat, 1:15 PM-4:15 PM
Association of total hysterectomy with survival among newly diagnosed A phase II, open labeled, single-arm study of dose-dense paclitaxel plus
uterine cancer patients with distant organ metastasis. First Author: Yue- carboplatin in advanced or recurrent uterine corpus cancer: KCOGG1303
feng Wang, University of Tennessee Health Sciences Center, Memphis, TN study. First Author: Kensuke Hori, Department of Obstetrics and Gynecology,
Kansai Rosai Hospital, Amagasaki, Japan
Background: There is growing evidence that definitive local therapies (surgery
or radiotherapy) may increase patient’s survival for some types of metastatic Background: We studied the effectiveness and safety of dose-dense paclitaxel
cancers. However, the role of total abdominal hysterectomy (TAH) for newly plus carboplatin in advanced or recurrent uterine corpus cancer. Methods: The
diagnosed uterine cancer with distant organ metastasis has not been estab- patient eligibility criteria were women aged 20–75 years with histologically
lished. The objective of this study is to determine the potential overall survival confirmed uterine corpus ; FIGO stage III who had residual tumors, FIGO stage
(OS) benefit associated with TAH for distant metastatic uterine cancer. IV, and recurrence after first-line radical treatment, or second-line chemo-
Methods: The National Cancer Database was analyzed to evaluate OS for newly therapy or radiotherapy. They received paclitaxel (80 mg/m2, days 1, 8, 15) +
diagnosed uterine cancer patients with metastasis to brain, lung, liver, bone or carboplatin (area under the curve 5, day 1 every 3 weeks). The primary
distant lymph node, treated with chemotherapy with or without TAH. Those endpoint was the response rate (RR). The secondary endpoints were feasibility,
without treatment, treated with definitive pelvic radiotherapy, or without progression-free survival, overall survival, and adverse effects. The threshold
baseline variables were excluded. OS was analyzed using the Kaplan-Meier RR was set to 40%. The expected RR of this treatment was set to 60%; the
method, log-rank test, Cox proportional hazards models, and propensity score- number of necessary cases calculated with a type I error of 5% and power of
matched analyses. In order to control the selection biases, we performed 80% was 44. Considering the existence of dropped cases, we set the target
Landmark analysis, and survival analysis by the sequence of chemotherapy and number of cases in this study to 48. Results: Forty-eight patients were reg-
TAH. Separate survival analysis was performed for patients who received istered, and 45 were eligible to receive the treatment. The median age of the
chemotherapy plus definitive pelvic radiotherapy (RT) or chemotherapy plus patients was 61 years (43–76). Twenty-two patients had recurrence; the others
TAH and definitive pelvic RT. Results: From 2010 to 2014, 1,809 uterine had primary advanced corpus cancer. On histology, there were 10 cases of
cancer patients with distant organ metastasis received chemotherapy alone serous carcinoma, 3 cases of endometrioid carcinoma G3, 2 cases of carci-
and 1,388 patients received chemotherapy plus TAH. At a median follow-up of nosarcoma, and 2 cases of clear cell carcinoma. Twenty-eight patients (62%)
13.4 months, addition of TAH to chemotherapy was associated with im- could receive 6 or more cycles of chemotherapy. The RR (complete, 13 cases;
proved survival on univariate (HR 0.57; P , 0.001) and multivariate analy- partial, 20 cases) was 73.3% (60.7–86.0 95% confidence interval).
sis (HR 0.59; P , 0.001) compared to chemotherapy alone. Propensity Conclusions: Dose-dense paclitaxel plus carboplatin was safe and effective for
score-matched analysis demonstrated superior median survival (19.8 vs advanced or recurrent uterine corpus cancer. Clinical trial information:
11.0 months) and 2-year OS (44% vs 28%) with TAH (multivariate HR 0.59; R000019874 UMIN000017138.
P , 0.001). Landmark analyses limited to long-term survivors of $0.5, $1,
and $2 years showed improved OS with TAH in all subsets (all P , 0.05). The
benefit of TAH was present among not only those involving one metastatic site
(HR 0.59; P , 0.001), but also those involving multiple metastatic sites (HR
0.60; P , 0.001). Separate survival analyses showed chemotherapy plus
definitive pelvic RT or chemotherapy plus TAH and RT were both superior to
chemotherapy alone. Conclusions: In this large contemporary analysis, uterine
cancer patients with distant organ metastasis receiving TAH and chemo-
therapy had substantial longer survival than patients treated with chemo-
therapy alone. Prospective trials evaluating TAH for metastatic uterine cancer
are warranted.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gynecologic Cancer 335s

5585 Poster Session (Board #408), Sat, 1:15 PM-4:15 PM 5586 Poster Session (Board #409), Sat, 1:15 PM-4:15 PM
Lynch-like syndrome in endometrial cancer: Features of a growing population. Mismatch repair deficiency as a predictor of adjuvant radiotherapy response
First Author: Sushmita Gordhandas, NewYork-Presbyterian/Weill Cornell in endometrioid endometrial carcinoma. First Author: Stefan Kommoss,
Medical Center, New York, NY Department of Women’s Health, Tuebingen University Hospital, Tuebingen,
Germany
Background: Current guidelines recommend screening all endometrial cancers (EC) and
colorectal cancers (CRC) for defects in DNA mismatch repair (MMR). Tumor screening Background: Adjuvant radiotherapy improves progression-free survival in
combined with germline genetic testing can categorize patients into three groups: intact- intermediate and high-risk endometrial cancer. However, so far there is no
MMR, Lynch syndrome (LS), and Lynch-like syndrome (LLS). Our objective was to evidence of improved overall or disease-specific survival after adjuvant ra-
describe features of the growing population of patients with LLS in EC and compare to
diotherapy. There is accumulating evidence that MMR proteins are involved
existing CRC literature. Methods: A systematic search of databases between 1990-2018
identified studies of EC patients with tumor testing (MMR immunohistochemistry or
in DNA repair following radiotherapy. We investigated the predictive value of
microsatellite instability) and germline assessment for LS. Data on clinicopathologic MMR status in terms of survival benefit after adjuvant radiotherapy in pa-
features was abstracted when available. Associations between LS, LLS, and intact-MMR tients with stage IB/II, grade 3 endometrioid endometrial cancer (EEC).
were analyzed using descriptive statistics. Results: The comprehensive search produced Methods: A retrospective multicenter cohort study was performed to com-
3,427 publications; 29 met inclusion criteria. Abstracted data and features of each group pare patients with histopathologically confirmed stage IB/II grade 3 EEC with
are presented in the table. Conclusions: In EC, LLS closely resembles LS with younger age and without adjuvant radiotherapy. Patients were classified according to the
at diagnosis, more advanced stage and higher grade as compared to patients with intact- Proactive Molecular Risk Classifier for Endometrial Cancer (ProMisE)
MMR. LLS in EC is similar to intact-MMR in regard to histology, and family history of LS- identifying ECs as either MMR-deficient, POLE, p53abn or p53wt. Multi-
associated cancer. The CRC literature is limited, but reports LS and LLS have similar variable Cox regression analysis explored associations between patient
stage, grade and histology. In CRC, LS and LLS are diagnosed at a younger age, and are characteristics, adjuvant treatment and outcome. Results: A total of 128
more likely to have family history of LS-associated cancers than intact-MMR. Features of patients were analyzed, including 57 patients (43.0%) with MMR-deficient
EC with intact-MMR, LLS, and LS.
EECs. Baseline characteristics were comparable, except a higher proportion
MMR-deficient
(N= 1047) P
of MMR-deficient EECs were stage II (36.8% vs. 15.5%, p = 0.006). Eighty-
Intact-MMR LLS LS
two patients (64.1%) received adjuvant radiotherapy (external beam [n =
(N=4608) (N=688) (N=212) Intact-MMR vs. LLS Intact-MMR vs. LS LLS vs. LS 55], vaginal brachytherapy [n = 27]). In multivariate analysis, adjuvant
Mean age- years (range) 62 (39-65) 53 (34-65) 51 (44-58) ,0.01 ,0.01 0.08 radiotherapy was independently associated with improved disease-specific
Mean BMI- kg/m2 (range) 36 (34-37) 35 (31-35) 28 (24-30) ,0.01 ,0.01 ,0.01
Grade- n (%) ,0.01 ,0.01 0.89 survival in patients with MMR-deficient EECs (hazard ratio 0.19, 95%-CI
1 471 (58) 31 (39) 39 (41)
2 175 (22) 25 (32) 27 (28) 0.05 - 0.77), but not in patients with MMR-proficient EECs (hazard ratio
3
Histology- n (%)
165 (20) 23 (29) 29 (31)
0.06 ,0.01 0.06 0.92, 95%-CI 0.37 - 2.31). Conclusions: Adjuvant radiotherapy improved
Adenosquamous
Carcinosarcoma
4 (1)
14 (1)
1 (1)
0 (0)
5 (4)
0 (0)
disease-specific survival in patients with MMR-deficient EECs, but not in
Clear cell
Endometrioid
13 (1)
979 (86)
2 (1)
128 (92)
4 (3)
107 (80)
those with MMR-proficient EECs. This study demonstrates the predictive
Mixed 43 (4) 7 (5) 13 (10) ability of MMR IHC to identify women who likely have increased benefit from
Serous 76 (7) 1 (1) 4 (3)
Stage- n (%) ,0.01 ,0.01 0.76 radiotherapy.
I 608 (81) 130 (70) 56 (67)
II 49 (7) 19 (10) 10 (12)
III 69 (9) 36 (19) 17 (21)
IV 22 (3) 2 (1) 0 (0)
Family History*- n (%)
Yes 28 (19) 19 (13) 88 (54) 0.13 ,0.01 ,0.01
No 118 (81) 129 (87) 74 (46)

*Defined as Amsterdam and/or Bethesda criteria, per study

5587 Poster Session (Board #410), Sat, 1:15 PM-4:15 PM 5588 Poster Session (Board #411), Sat, 1:15 PM-4:15 PM
Impact of non-compliance with guidelines in early type 1 endometrial Genomic biomarkers of recurrence in low-grade, early-stage endometrial
cancers management, study from FRANCOGYN group. First Author: adenocarcinoma. First Author: Katie Lee Hwang, Harvard Radiation Oncology
Hélène Costaz, Centre GF Leclerc, Dijon, France Program, Boston, MA
Background: To standardize surgical practices, ESMO-ESGO-ESTRO con- Background: Endometrial cancer is the most common gynecologic malignancy
sensus conference published in 2016 new guidelines on the management of in developed countries with over 60,000 new cases diagnosed in the United
endometrial cancer. The main objective of this study was to evaluate the States each year. Adjuvant therapy is often omitted for low-risk, early-stage
impact of non-compliance with current surgical guidelines on disease-free disease (FIGO stage IA, grade 1) but 1 in 20 women suffer recurrence after
survival and overall survival. Methods: 852 patients with presumptive stage I surgery alone. Hence, there is an important need for biomarkers of recurrence
and II type 1 endometrial cancer were included in a multicenter retrospective in this population to guide therapeutic management. Methods: We retro-
study, conducted between January 2000 and November 2015. The main spectively analyzed 74 patients with FIGO stage 1A, grade 1 endometrial
objective of this study was to evaluate the impact of non-compliance with endometrioid adenocarcinoma treated at our institution with hysterectomy
current surgical recommendations on overall survival and disease-free alone between 2009-2016. All patients had targeted genomic assessment of
survival. Results: Our study shows that 34.3% of patients (n = 292) did their tumors (OncoPanel; somatic mutations, copy number variations and
not benefit from optimal surgical treatment. These patients did not have a structural variants across 300 cancer genes). The primary outcome of interest
lombo-aortic lymphadenectomy (LAL) and were at high risk of recurrence. was freedom from recurrence (FFR). Outcomes were compared by the logrank
There is a significant difference in disease-free survival in favor of patients test and survival estimates calculated by Kaplan-Meier method. Results: We
undergoing surgery according to the recommendations, (Hazard Ratio (HR): identified 14 patients who recurred at a median time of 23.6 months after
0.37 (Confidence interval (95% CI): 0.26-0.54), p , 0.001). In multivariate surgery and 60 patients without recurrence at a median follow-up of
analysis, optimal surgical procedure performance is an independent factor 38.9 months. Age (median 57 years; log-rank p = 0.91) and BMI (median 31
for disease-free survival with HR at 2.04 (95% CI: 1.14-3.68), p = 0.01. kg/m2; log-rank p = 0.21) were not associated with risk of recurrence. The
There is a significant difference in overall survival in favor of patients un- median somatic mutation count in the cohort was 8. Patients with more than 8
dergoing surgery according to the recommendations, (HR: 0.31 (95% CI): somatic mutations had a significantly higher risk of recurrence (3-year FFR:
0.19-0.49), p , 0.001. In multivariate analysis, there is a trend toward 74% vs 90%; log-rank p = 0.004). At the level of individual genes, there were
significance with HR: 2.24 (95% CI: 1-5.05), p = 0.05. Older patients, four genes that were significantly associated with recurrence: CTNNB1 (p =
patients with a larger BMI, patients with no indication of LAL at the pre- 0.046), RHPN2 (p = 0.020), SF1 (p = 0.044), SQSTM1 (p = 0.034). Patients
operative ESMO classification, and no node involvement in are factors with a mutation in one or more of these four genes had a significantly higher
contributing to the decision of not to perform a LAL: p , 0.001, p = 0.03, risk of recurrence (3-year FFR: 62% vs 93%; log-rank p = 0.0004).
p , 0.001 and p , 0.001 respectively. Conclusions: This study shows that Conclusions: We have identified overall somatic mutation burden and muta-
patients with early type 1 endometrial cancer have improved recurrence-free tions in a subset of four genes (CTNNB1, RHPN2, SF1, SQSTM1) as de-
survival and a statistical trend for an increased overall survival when rec- termined by a validated 300-gene panel used in routine clinical practice as
ommended surgery is performed. Despite the current context of therapeutic prognostic biomarkers for patients with low-risk, early-stage endometrial
de-escalation, we must strive to achieve the recommended optimal surgery, endometrioid adenocarcinoma. These patients may benefit from the addition
even if it requires secondary surgical revision, to avoid underestimation of of adjuvant therapy. Validation with larger cohorts and prospective studies is
patients with a poorer prognosis. To improve endometrial cancers man- warranted.
agement, amelioration of the preoperative assessment by increasing the
sensitivity of emboli detection should be considered.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
336s Gynecologic Cancer

5589 Poster Session (Board #412), Sat, 1:15 PM-4:15 PM 5590 Poster Session (Board #413), Sat, 1:15 PM-4:15 PM
The prognostic significance of white adipose tissue inflammation in advanced- Identifying a potential biomarker for anti-PD-1 immunotherapy in patients
stage, high-grade, and serous endometrial cancers. First Author: Vance with advanced stage, surgically-resectable endometrial cancer. First Author:
Broach, Memorial Sloan Kettering Cancer Center, New York, NY Katherine Cynthia Fuh, Washington University, St. Louis, MO
Background: Obesity is associated with worse outcomes in endometrial Background: The FDA approval of pembrolizumab for patients with MSI-H or
cancer, but the underlying mechanisms are poorly understood. In other dMMR tumors has led to the treatment of a select cohort of endometrial cancer
obesity-related cancers, white adipose tissue inflammation (WATi) is an in- (EC) patients. We sought to ascertain tumor immune modulatory effects in the
dependent predictor of shortened cancer-specific survival. We hypothesized front-line setting for advanced stage III/IV EC patients regardless of MSI-H or
that WATi occurs in patients with endometrial cancers and is a prognostic dMMR. The primary objective was to determine the safety of preoperative and
marker of shortened survival. Methods: We conducted a retrospective cohort maintenance pembrolizumab. The secondary objective was to examine
study in which patients with stage III or IV grade 3 endometrioid (G3) or serous pembrolizumab-induced changes in peripheral immune effector phenotype in
endometrial cancer were included. Eligible subjects had archived omental order to identify potential biomarkers of clinical response. Methods: In an open
and/or peri-nodal adipose tissue available. WATi was detected by the presence label, single-arm Phase I trial, 8 EC patients were treated with 2 doses of
of dead/dying adipocytes surrounded by CD68+ macrophages forming a crown- preoperative pembrolizumab IV prior to surgery followed by chemotherapy and
like structure (CLS). Clinicopathologic data were abstracted from medical 4 doses of pembrolizumab IV. As an initial study, pre- and post-treatment (on
records. For association with WATi, Wilcoxon rank sum test was used for the day of surgery) peripheral blood was collected from 3 patients as well as a
continuous variables, Fisher’s exact test for categorical variables. Log rank test healthy control and processed for high-dimensional single-cell mass cytometry
was used to assess the association of WATi and survival. Results: A total of 95 (CyTOF) using an optimized antibody panel. Results: Six of 8 patients com-
patients who underwent debulking surgery from 2001–2017 were included pleted the treatment. One patient had rapid cancer progression and another
(median age, 67 years; range, 33-86 years). Of these, 51 (54%) had WATi. The had an exacerbation of comorbidities. Peripheral blood from 3 patients with
presence of WATi was unaffected by race, tumor histology or stage. Patients pathological response were then immunoprofiled using CyTOF. Data analysis
with WATi had a higher median body mass index (BMI) than those without revealed that the frequencies of CD8+ T cells, B cells and CD56hiCD16- NK
WATi (32.17 and 27.33 kg/m2, respectively; P , 0.001) and were more likely cells were lower, whereas the frequency of CD14+CD16-HLA-DRhi classical
to be obese (P = 0.01). Patients with the most severe WATi (n = 20) had shorter monocytes was higher in the cancer patients compared to controls. Cancer
progression-free survival (PFS) and a trend suggesting shorter overall survival patients had lower frequencies of circulating CD4+ and CD8+ naı̈ve T cells but
(OS) than those patients with less severe or no WATi (n = 75) (median PFS 15.8 higher frequencies of effector CD8+ and CD4+ T cells. Notably, the median
vs 59.2 months, respectively, P = 0.001; median OS 33.9 vs 59.4 months, expression of Granzyme B in CD8+ and CD4+ T cells was higher and median
respectively, P = 0.059). Conclusions: Visceral adipose inflammation is expression of signal regulatory protein (SIRP), CD172a-b on monocytes was
prevalent in obese patients with advanced G3 and serous endometrial cancer. lower for cancer patients compared to control. The frequencies of NK and
Severe inflammation was associated with significantly worse PFS. myeloid cells expressing CD137(4-1BB), PD-1+NK cells, and PD-L1+DCs
were greater in post-compared to pre-pembrolizumab. Conclusions: This is the
first trial to evaluate the use of neoadjuvant pembrolizumab in advanced stage
EC patients. Here, we present peripheral immune correlative data and show an
increase in markers of activation in patients with pathologic responses to
pembrolizumab. Additional data from this ongoing study will help us to identify
candidate predictive biomarkers. Clinical trial information: NCT02630823.

5591 Poster Session (Board #414), Sat, 1:15 PM-4:15 PM 5592 Poster Session (Board #415), Sat, 1:15 PM-4:15 PM
Trends of endometrial cancer incidence from 2000 to 2015 in the United p53 and p16 expression profiles reveal three prognostically relevant sub-
States. First Author: V V Pavan Kedar Mukthinuthalapati, John Stroger groups in vulvar cancer: A TMA based study by the AGO-CaRE-translational
Hospital of Cook County, Chicago, IL study group. First Author: Linn Lena Woelber, AGO & Department of Gy-
necology and Gynecologic Oncology, University Medical Center Hamburg-
Background: Recent studies have shown that obesity related cancers are increasing in
incidence in the US as the rates of obesity rise and some cancers, like colorectal cancer, Eppendorf, Germany, Hamburg, Germany
are occurring in younger age groups. We studied trends in incidence of endometrial cancer Background: Currently, there are two major pathways for tumorigenesis
(EC), one of the obesity related cancers, in a population wide analysis. Methods: We of vulvar squamous cell carcinoma (VSCC) – an HPV-dependent with p16
analyzed data from all cases of EC between 2000 and 2015 from 18 US cancer registries
overexpression as a surrogate for HPV-associated transformation and an HPV-
using the National Cancer Institute’s Surveillance, Epidemiology and End Results Pro-
gram. SEER*Stat was used to query the database for annual percent changes (APC),
independent route linked to lichen sclerosus, characterized by p53 mutation.
incidence ratios and percent change in incidence across different age groups, years of A possible correlation of HPV dependency with a favourable prognosis has been
diagnosis, histologic subtypes, grade and race. We also studied the reported rates and proposed. Methods: The AGO CaRE-1 study is a retrospective survey of pts with
trends of obesity in the US. Results: APC of age-adjusted EC incidence between 2000 primary VSCC FIGO stage $1B (UICC-TNM version 6) treated at 29 gyne-
and 2015 was +0.9% (95% confidence interval (CI) 1.1-0.6, p value,0.05). Incidence cologic cancer centers in Germany 1998-2008 (n = 1,618). For this CaRE-
of EC rose from 17.8 per 100,000 to 19.7 per 100,000 during the same duration. APC for translational sub-study available FFPE tissue was collected centrally (n = 648).
EC incidence for age groups 20-39 and .40 were +3.2% (p-value ,0.05) and +0.8% (p A tissue micro array (TMA) was constructed; p16 and p53 expression was
value ,0.05), respectively. For the age-group 20-39, endometrioid EC was the only determined by immunohistochemistry (IHC). HPV status and subtype were
histologic subtype that rose in incidence, with an APC of +5.5% and absolute percentage analyzed by PCR. Results: p16 IHC was interpretable in 550 TMA spots and
change of 156%. The APC of EC in 20-39 age group was more for whites (3.5%, p- considered positive in 166/550 (30.2%). HPV DNA was detected in 78.4% of
value,0.05) and Asians (2.2%, p-value,0.05) than blacks (1.8, p-value ,0.05). CDC the p16+ tumors, with HPV 16 being the most common subtype (88.3%). Pts
reported an increase in obesity rates in adults from 30.5% in 2000 to 37.7% in 2014.
Table shows trends of EC incidence in age groups 20-39 and .40 years across various
with p16+ tumors were younger at diagnosis (63 vs. 70 yrs for p16- tumors; p =
histologic subtypes. (Abbreviations: S significant, NS not significant, NC non-calculable). 0 , 0.01) and showed lower rates of lymph-node involvement (29.0% vs.
Conclusions: Endometrial cancer, especially of endometrioid histology, is increasing in 39.7%; p = 0.021). p53 IHC was interpretable in 597 spots, 187/597
incidence and is occurring more often in the younger population. The concomitant rise in (31.3%) were considered positive. Pts with p53+ tumors were older at first
obesity rates during the same period point towards a possible causality of the increased in diagnosis (71 vs. 66 yrs; p = 0.001 for p53- tumors) and showed lymph-node
incidence of EC. Population based strategies are needed to decrease the trends in obesity involvement more often (43.3% vs. 31.1%; p = 0.007). There was a relevant
so as to decrease the risk of endometrial cancer in younger women. number of tumors with neither p16 nor p53 overexpression (221/535); while
20-39 >40 co-expression of p53 and p16 was rare (12/535). For survival analyses, three
Age group
%age change from APC Statistical %age change from APC (95% Statistical
groups were defined: p53+ (n = 163), p16+/p53- (n = 151) and p16-/p53-
Histology 2000 to 2015 (95% CI) significance 2000 to 2015 CI) significance (n = 221). 2-y-disease-free (DFS) and overall survival (OS) rates were sig-
Endometrioid EC 156.3 5.5 (4.5- S 56.1 2.6 (2.2-3) S nificantly different between the groups: DFS: p53+ 47.0%; p16-/p53- 53%
6.6)
Serous NC NC NS 819.7 17.4 (16.4- S and p16+/p53- 65.5% (p , 0.001); OS: 70.4%, 72.6% and 82.7% (p =
Clear NC NC NS 14
18.5)
0.8 (-0.5- NS
0.003), respectively. Adjustment for age and nodal status showed consistent
2.1) p16 and p53 effects regarding DFS. Conclusions: p16 overexpression is as-
Mucinous -62.4 NC NS -60.5 -6 (-7.2-- S
4.7) sociated with an improved prognosis in VSCC while p53 positivity is linked to
Carcinosarcoma 170.651 NC NS 133.5 6.3 (5.2- S
7.4) an adverse outcome. Our data provide evidence of a clinically relevant third
Undifferentiated
carcinoma
NC NC NS 276.7 9.1 (5.4 -
13)
S subgroup of VSCC with a p53-/p16- phenotype showing an intermediate
Mesenchymal 295.9 NC NS 62.3 3.5 (1.7 - S prognosis that needs to be further characterized.
tumors 5.4)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gynecologic Cancer 337s

5593 Poster Session (Board #416), Sat, 1:15 PM-4:15 PM TPS5594 Poster Session (Board #417a), Sat, 1:15 PM-4:15 PM
Use of nivolumab as salvage therapy in heavily pretreated patients with BEATcc (ENGOT-Cx10/GEICO 68-C/GOG3030/JGOG1084): A randomized,
gynecologic malignancies. First Author: Heather Williams, Medical College open label, phase III study of cisplatin and paclitaxel chemotherapy with
of Georgia at Augusta University, Augusta, GA bevacizumab (CTx plus B) with or without atezolizumab (Atz) as first-line
treatment for metastatic, persistent, or recurrent (m/r) carcinoma of the
Background: There are limited effective treatments for gynecologic cancer
cervix (CCx). First Author: Ana Oaknin, Vall d’Hebron University Hospital,
patients who have been previously treated with multiple lines of chemotherapy.
Vall d’Hebron Institute of Oncology (VHIO), Barcelona, Spain
Immune checkpoint inhibitor (ICI) therapy has demonstrated significant ac-
tivity in certain cancers but has been inconclusive in most gynecologic ma- Background: The combination of CTx plus B is first line treatment for most
lignancies. The objective of this study was to determine the impact of salvage patients (pts) with m/r CCx not amenable for local therapy based on GOG240
ICI therapy in heavily pretreated gynecologic oncology patients. Methods: An results. GOG240 regimen showed an improvement in overall survival (OS)
IRB approved retrospective study was performed of women with gynecologic compared to CTx alone: 16.8 vs. 13.3 months (HR 0.77, 95% CI 0.62–0.95,
cancer treated with nivolumab on a compassionate use program between p = 0.007). However, further improvement in first line therapy outcomes is an
October 2015 and January 2018. Patient demographics, disease character- unmet need. Immune-checkpoint inhibitors are breakthrough therapies in
istics, pathology and treatment history were collected. Survival probabilities several tumor types, and their development in CCx is supported by a strong
were calculated. Results: Twenty-eight women were identified. Median age at scientific rationale. Human papillomavirus infection (HPV) causes more than
start of treatment was 63 years with a median of 4 prior lines of chemotherapy. 90% of CCx cases. PD-L1 is a HPV biomarker and is found frequently up-
Median ECOG status was 2. Disease site was evenly distributed among uterus, regulated in CCx. Nivolumab and pembrolizumab (Pb) (anti-PD-1 antibodies)
ovary and cervix. 67.9% of patients completed 3 or more cycles of treatment. have shown response rates of 26.3% and 14.3%, respectively, in pretreated m/r
Median PFS of all patients was only 2.6 months however when comparing CCx. This has led to the recent FDA approval of Pb in pretreated PD-L1+ m/r
patients who received 2-3 cycles (n = 13) with those who received 4 or more CCx. The BEATcc trial (NCT03556839) evaluates the addition of the anti-PD-
(n = 9), median PFS was statistically significant 2.4 months vs 6.4 months (p = L1 agent Atz to GOG240 regimen as first line treatment for m/r CCx, following
0.0005). When looking at treatment response, 7 patients had partial response/ the synergistic rationale between anti-VEGF agents and PD-1/PD-L1 blockade.
stable disease after 3 cycles (25%). Median PFS of the 7 “responders” was Methods: Eligible pts: m/r CC with adequate organ function. Pts will be ran-
6.6 months vs 2.5 months of the non-responders (p , 0.001). Only 1 of 9 domized 1:1 to either Arm A (control): C 50 mg/m2 + Tx 175mg/m2 + B 15 mg/
patients with uterine cancer had a disease response and that patient’s tumor kg (CTx plus B) i.v. D1 Q3W or Arm B (experimental): CTx plus B + Atz 1200 mg
was MSI high. Five patients had low grade serous ovarian cancer. Four of them i.v. D1 Q3W. Stratification factors: prior chemo-radiation, histology and Che-
had a treatment response with a median PFS of 6.1 months (range 3.8 – motherapy backbone (CTx vs carboplatin-Tx). Treatment is planned until dis-
25 months). Adverse events were experienced by 68% of patients; most ease progression, unacceptable toxicity or withdrawal of consent. Pts with a
commonly being fatigue (46.4%), arthralgia (25%), and anemia (21.4%). complete response after $6 cycles or those with unacceptable CTx toxicity may
Only 1 patient experienced a grade 3-4 event (a diffuse maculopapular rash). be allowed to continue only on biologics therapy. An Independent Data Mon-
Conclusions: In patients with heavily pretreated gynecologic malignancies with itoring Committee will analyze the safety of the first 12 pts in the experimental
suboptimal performance status, immune checkpoint inhibitor therapy may arm completing 2 treatment cycles. The primary endpoint is OS. The study
prolong survival without significant toxicity. Also, there may be a role for ICI in started enrolling in October 2018 and will enroll approximately 404 pts across
patients with historically chemo resistant low grade serous ovarian cancers. Europe, Japan, and the US. Clinical trial information: NCT03556839.

TPS5595 Poster Session (Board #417b), Sat, 1:15 PM-4:15 PM TPS5596 Poster Session (Board #418a), Sat, 1:15 PM-4:15 PM
KEYNOTE-826: A phase 3, randomized, double-blind, placebo-controlled Trial in progress: Phase II study of stereotactic body radiation therapy and
study of pembrolizumab plus chemotherapy for first-line treatment of persistent, atezolizumab in the management of recurrent, persistent, or metastatic
recurrent, or metastatic cervical cancer. First Author: Ronnie Shapira-Frommer, cervical cancer. First Author: Kamran A. Ahmed, H. Lee Moffitt Cancer
Sheba Medical Center, Ramat-Gan, Israel Center and Research Institute, Tampa, FL
Background: Cervical cancer arises in the setting of persistent infection with Background: There is no consistent recommendation for management of
high-risk human papillomavirus subtypes. Many patients with early-stage and metastatic cervical cancer beyond first line therapy with chemotherapy and
locally advanced carcinoma can be salvaged with radical surgery and che- bevacizumab. Pembrolizumab is now approved for PD-L1 positive or MSI-H/
moradiation, respectively. However, women with recurrent/metastatic disease dMMR metastatic or recurrent cervical cancer. Numerous pre-clinical
represent a poor prognostic group with high unmet clinical needs. In- studies have provided evidence to combine radiation therapy with im-
corporation of anti-angiogenesis therapy has emerged as a therapeutic option, mune checkpoint inhibition to improve response rates. The evidence is
but the survival benefit of 3.7 months over chemotherapy (CT) alone is modest strongest for short course, hypofractionated radiation regimens. We hy-
(Tewari et al. NEJM 2014). Because viral tumor antigen-specific T cells reside pothesize treatment with atezolizumab with hypofractionated radiation
predominantly in programmed cell death 1–expressing T-cell compartments, therapy will improve objective response rate (ORR) compared with atezo-
checkpoint inhibition may unleash a diverse antitumor T-cell response. Based lizumab alone in patients with recurrent, persistent, or metastatic cervical
on the 14.3% objective response in KEYNOTE-158, the US FDA granted cancer. Methods: The study is designed as a prospective, single arm,
accelerated approval to pembrolizumab (pembro) in June 2018 for second-line nonrandomized, open-label, phase II trial of stereotactic body radiation
therapy and beyond. Methods: KEYNOTE-826 is a phase 3, randomized, therapy (SBRT) with 24 Gy in 3 fractions to patients with $ 2 metastatic sites
double-blind, placebo-controlled, multinational trial of CT with pembro or with followed 1 week later by atezolizumab (1200 mg IV every 3 weeks) for
placebo for first-line treatment of recurrent, persistent, or metastatic cervical patients with recurrent, persistent, or metastatic cervical cancer. Dose re-
cancer. Patients not previously treated with CT for recurrence who are not ductions will not be allowed. The primary objective of the study is to evaluate
amenable to curative treatment will be randomized 1:1 to CT + pembro 200 mg the ORR by Immune-Modified Response Evaluation Criteria in Solid Tumors
or placebo every 3 weeks. The CT regimen (paclitaxel 175 mg/m2 + cisplatin (irRECIST) criteria following SBRT and atezolizumab. Secondary endpoints
50 mg/m2 or carboplatin AUC 5, with or without bevacizumab 15 mg/kg) will include progression free survival, overall survival, local control, and adverse
be selected by the investigator before randomization. Stratification factors events. Correlative aims include assessing blood and tissue biomarkers (i.e.
include metastasis status at diagnosis, bevacizumab use (yes/no), and tumor PD-L1, mutation burden, TCR repertoire etc.) for association with clinical
PD-L1 status (combined positive score ,1, 1 to ,10, or $10). Treatment will benefit. A total of 26 patients will be enrolled. An interim analysis will be
continue until disease progression, unacceptable toxicity, or voluntary patient performed to assess efficacy after 13 patients become evaluable.This study
withdrawal for up to 35 cycles (~2 years). Primary endpoints are progression- is open with 2 patients enrolled at the time of submission. Clinical trial
free survival (PFS) per RECIST v1.1 assessed by blinded independent central information: NCT03614949.
review and overall survival. Secondary endpoints are objective response, du-
ration of response, 12-month PFS, patient-reported quality of life, and safety.
Enrollment is ongoing. Clinical trial information: NCT03635567.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
338s Gynecologic Cancer

TPS5597 Poster Session (Board #418b), Sat, 1:15 PM-4:15 PM TPS5598 Poster Session (Board #419a), Sat, 1:15 PM-4:15 PM
CALLA: Efficacy and safety of durvalumab with and following concurrent DUO-O: A randomized phase III trial of durvalumab (durva) in combination
chemoradiotherapy (CCRT) versus CCRT alone in women with locally with chemotherapy and bevacizumab (bev), followed by maintenance durva,
advanced cervical cancer: A phase III, randomized, double-blind, multicenter bev and olaparib (olap), in newly diagnosed advanced ovarian cancer
study. First Author: Bradley J. Monk, Arizona Oncology (US Oncology Network), patients. First Author: Philipp Harter, Kliniken Essen-Mitte, Evangelische
University of Arizona College of Medicine, Creighton University School of Huyssens-Stiftung/Knappschaft GmbH, Essen, Germany
Medicine at St. Joseph’s Hospital, Phoenix, AZ
Background: Ovarian cancer (OC) is the leading cause of death from gyne-
Background: CCRT is the standard of care for locally advanced cervical cancer. cologic cancers in US women. Despite high response rates to first-line
CCRT with PD-1/PD-L1 pathway blockade may promote a more immunogenic treatment, ~70% of patients (pts) relapse within 3 years and then remain
environment through increased phagocytosis, cell death, and antigen pre- largely incurable. First-line treatment needs to be improved to achieve long-
sentation, leading to enhanced immune-mediated tumor surveillance. This term remission in pts and improve the cure rate. The Phase III SOLO1 trial
Phase 3, randomized, multicenter, international, double-blind, placebo- showed a meaningful clinical benefit for olap maintenance therapy in newly
controlled study is designed to determine the efficacy and safety of durvalumab diagnosed OC pts with a BRCA mutation (Moore et al N Engl J Med 2018).
with and following CCRT vs. CCRT alone in women with locally advanced Preliminary data suggest that combining a PD-L1 inhibitor, anti-angiogenic
cervical cancer (NCT03830866). Methods: The study will enroll immuno- and PARP inhibitor (triplet therapy) may achieve a synergistic antitumor effect.
therapy-naı̈ve adult patients (pts) with histologically confirmed cervical ade- The DUO-O study (NCT03737643) evaluates the efficacy and safety of
nocarcinoma or cervical squamous or adenosquamous carcinoma (FIGO treatment combinations involving standard-of-care platinum-based chemo-
Stages IB2-IIB with node [N] positive and IIIA-IVA with any N) and no prior therapy (chemo), VEGF inhibitor bev, anti-PD-L1 antibody durva and PARP
definitive surgical, radiation, or systemic therapy for cervical cancer. Ap- inhibitor olap, in women with newly diagnosed advanced OC. Methods: Eligible
proximately 714 pts will be randomized 1:1 to receive either durvalumab pts for this double-blind, randomized, Phase III study must have newly di-
(1500 mg intravenously [IV]) or placebo every 4 weeks for 96 weeks. All pts will agnosed, advanced, high-grade epithelial OC and either have completed
receive CCRT to the pelvis or pelvis + para-aortic radiotherapy field (45 Gy), primary surgery or plan to have interval debulking surgery. Depending on their
followed by image-guided brachytherapy with cisplatin (40 mg/m2) IV or tumor BRCA mutation (tBRCAm) status (determined by central test), pts will
carboplatin (AUC2) IV once weekly for 5 weeks (6th dose optional). Ran- join one of two independent cohorts. Pts in the non-tBRCAm cohort (n~906)
domization will be stratified by disease stage (FIGO Stage , III and N positive, will be randomized (1:1:1) before cycle 2 to: a) chemo + bev + placebo (for 6
FIGO Stage $III and N negative, or FIGO Stage $III and N positive) and region cycles) followed by bev (15 mg/kg [total 15 months]) + placebo maintenance
(US, Canada, EU, South Korea, and Japan vs. rest of the world). The primary treatment (IV and tablets); b) chemo + bev + durva (6 cycles) followed by bev +
endpoint is progression-free survival (assessed by investigator per RECIST v1.1 durva (1120 mg q3w [total 15 months]) + placebo (tablets) maintenance
or histopathologic confirmation of local tumor progression). Secondary end- treatment; or c) chemo + bev + durva (6 cycles) followed by bev + durva + olap
points are overall survival, objective response and complete response (CR) (300 mg bd tablets [24 months]) maintenance treatment. Pts in the open-label
rates, duration of response in pts with CR, incidence of local or distant disease tBRCAm cohort (n~150) will receive 6 cycles of chemo + durva followed by
progression or secondary malignancy, disease-related symptoms, and health- durva + olap maintenance therapy, with optional use of bev. The primary
related quality of life (EORTC QLQ-C30 and EORTC QLQ-CX24). Pharma- endpoint of progression-free survival will be assessed by modified RECIST 1.1.
cokinetics, immunogenicity, and safety of durvalumab will also be assessed. Pt Key secondary endpoints include overall survival, overall response rate and
enrollment is ongoing. Clinical trial information: NCT03830866. duration of response. Enrollment began in January 2019. Clinical trial in-
formation: NCT03737643.

TPS5599 Poster Session (Board #419b), Sat, 1:15 PM-4:15 PM TPS5600 Poster Session (Board #420a), Sat, 1:15 PM-4:15 PM
ENGOT-Ov41/GEICO-69-O/ANITA trial: A phase III randomized, double- ENGOT-OV44/FIRST study: A randomized, double-blind, adaptive, phase III
blinded trial of platinum-based chemotherapy (CT) with or without atezoli- study of platinum-based therapy with dostarlimab (TSR-042) + niraparib
zumab (ATZ) followed by niraparib maintenance with or without ATZ in versus standard-of-care (SOC) platinum-based therapy as first-line treatment
patients with recurrent ovarian, tubal or peritoneal cancer (OC) and platinum of stage 3/4 non-mucinous epithelial ovarian cancer (OC). First Author:
treatment-free interval (TFIp) .6 months. First Author: Antonio González- Anne-Claire Hardy-Bessard, CARIO-HPCA and Cooperative Gynecological
Martı́n, GEICO and Clı́nica Universidad de Navarra, Madrid, Spain Cancer Research Group (GINECO), Plerin, France
Background: Platinum-based CT is the treatment of choice for OC patients Background: Despite surgery and SOC therapy (paclitaxel and carboplatin 6
(pts) with TFIp . 6 months suitable for platinum. Niraparib is an oral PARP bevacizumab[bev]), 5-year survival rates remain low for patients (pts) with
inhibitor that significantly prolongs the progression-free survival (PFS) of OC FIGO stage 3/4 OC. Niraparib (ZEJULA) is the first selective poly(ADP-ribose)
pts when given as maintenance therapy after a response to platinum- polymerase inhibitor (PARPi) approved in the US and Europe for maintenance
rechallenge (both in gBRCA and non-gBRCA mutated pts). Atz is a hu- treatment in pts with recurrent OC regardless of BRCAmut status. Preclinical
manized monoclonal antibody targeting PD-L1 that showed activity in data suggest synergy with PARPi + anti-PD-1 blockade. Niraparib + pem-
heavily pretreated OC pts. Combination of anti-PD-L1 and CT may improve brolizumab has shown clinical efficacy in pts with platinum-resistant or
the activity of the anti-PD-L1 antibody by increasing the amount of antigens secondary refractory OC regardless of biomarker status. Dostarlimab is an
released after immunogenic cell death induced by CT. Combination of PARPi anti–PD-1 humanized monoclonal with clinical activity as monotherapy in
and anti-PD-L1/PD-1 has shown synergy in preclinical models and promising early phase trials. The primary objective of the currently enrolling FIRST trial is
clinical activity in certain indications. Methods: ANITA (NCT03598270) is a to compare PFS (per RECIST v1.1) in pts treated with SOC + dostarlimab +
phase III, randomized (1:1), double-blinded, multi-center study to assess niraparib to SOC. Methods: Eligible pts (up to 912) are FIGO stage 3 (with
the efficacy of the addition of Atz to platinum-based doublet CT followed by residual disease, CC0 high risk, or planned neoadjuvant therapy) or stage 4,
maintenance niraparib in patients with recurrent high grade serous or non-mucinous epithelial OC and ECOG score , 2. After 1 cycle of SOC, pts are
endometrioid OC with a TFIp . 6 months. Approximately 414 patients with stratified by concurrent bev use, BRCAmut/HRR status, and disease burden
ECOG 0-1, known BRCA status, at least one measurable lesion and # 2 prior then randomized as 1:1:2 to 1 of 3 arms (Table). An innovative feature of
lines will be randomized to placebo of Atz (Arm A) or Atz (Arm B) in com- ENGOT-OV44/FIRST (NCT03602859; EUDRACT 2018-000413-20) is the
bination with one of three possible standard platinum-based CT regimens pre-planned adaptive study design to adapt the control arm to the evolving
(investigator’s choice) followed by maintenance niraparib in combination SOCs in OC, allowing pts in the control arm to receive up to date SOC. These
with placebo or Atz, according to randomization, if experiencing response or adaptations will occur when practice-changing data are released. Following
stable disease by RECIST after CT. Stratification factors: 1) Platinum-based publication of SOLO1 results, BRCAmut pts will only be randomized to arm 2 or
regimen (paclitaxel-carboplatin vs gemcitabine-carboplatin vs PLD- 3 to ensure they receive niraparib. Further adaptations may be incorporated as
carboplatin); 2) Platinum-free interval (6-12 vs . 12 months); and 3) new data become available, leading to stop randomization in arm 1 or 2 of pts
BRCA status (mutated vs non-mutated). Dose of Atz is 1200 mg q3 weeks or based on their biomarker status. Clinical trial information: NCT03602859.
840 mg q2 weeks depending on the platinum-based regimen selected.
Niraparib initial dose (300 vs 200 mg) is decided based on body weight and Initial Randomization Scheme
platelet counts after CT according to RADAR analysis. Primary endpoint is Treatment period Arm 1* Arm 2 Arm 3
PFS based on investigator assessment by RECIST v1.1. Clinical trial in- Chemotherapy SOC + IV placebo SOC + IV placebo SOC + IV dostarlimab
formation: NCT03598270. Maintenance up to Oral placebo + IV Oral niraparib + IV Oral niraparib + IV
3 years placebo placebo dostarlimab
*No BRCAmut pts will be randomized to arm 1 following SOLO1 results.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gynecologic Cancer 339s

TPS5601 Poster Session (Board #420b), Sat, 1:15 PM-4:15 PM TPS5602 Poster Session (Board #421a), Sat, 1:15 PM-4:15 PM
AGO-OVAR 2.29 (ENGOT-ov34): Atezolizumab in combination with bevaci- Phase 2 trial of tisotumab vedotin in platinum-resistant ovarian cancer
zumab and chemotherapy versus bevacizumab and chemotherapy in recurrent (innovaTV 208). First Author: Haider Mahdi, The Cleveland Clinic, Cleve-
ovarian cancer (ROC). First Author: Frederik Marme, AGO & National Center for land, OH
Tumor Disease/Department of Gynecology, University of Heidelberg, Heidel-
Background: Ovarian cancer (OC) is the most lethal gynecologic cancer, ac-
berg, Germany
counting for »185,000 deaths worldwide in 2018. Most patients (pts) initially
Background: A standard non-platinum based treatment option in patients with respond to platinum-based chemotherapy (chemo), but more than 50% of pts
relapsed ovarian cancer is bevacizumab in combination with paclitaxel or recur. Pts who recur in #6 months have platinum-resistant OC (PROC), which
pegylated liposomal doxorubicin, but responses are still short-lived. Checkpoint- is associated with poor prognosis. Standard therapy for PROC includes chemo 6
inhibitors as single agent have limited activity in ovarian cancer. However, the bevacizumab (bev). However, many pts receive single-agent chemo, which
role of the checkpoint-inhibitor like atezolizumab, in addition to chemotherapy demonstrates limited response and survival (»12% ORR, 3-4 mo PFS, »12 mo
and bevacizumab in ovarian cancer is so far undefined. Methods: AGO-OVAR OS). Therefore, there is an urgent need for novel therapeutic strategies. Tissue
2.29 is a randomized (1:1), double blinded, phase III trial evaluating the ef- factor (TF) is a novel oncogenic target expressed in OC. Tisotumab vedotin (TV)
ficacy and safety of atezolizumab plus bevacizumab and chemotherapy (weekly is a first-in-class antibody drug conjugate comprising a TF-targeted fully
paclitaxel or pegylated liposomal doxorubicin) compared with placebo plus human monoclonal antibody conjugated to the microtubule-disrupting agent
bevacizumab and chemotherapy in patients with recurrent ovarian-, fallopian monomethyl auristatin E. TV has shown encouraging antitumor activity and a
tube, or primary peritoneal cancer with 1st or 2nd relapse within 6 months after manageable safety profile in PROC in the multicohort phase 1/2 innovaTV 201
platinum-based chemotherapy or 3rd relapse. A tumor biopsy available at study study. innovaTV 208 is a multicenter, open-label, phase 2 trial with a safety
entry for PD-L1 testing is mandatory. Patients are treated with chemotherapy run-in phase for a dose-dense regimen (DDR) evaluating the efficacy and safety
plus bevacizumab +/- atezolizumab/placebo until progression or prohibitive of TV in pts with PROC. Methods: innovaTV 208 will enroll »142 adult pts with
toxicity. Co-primary endpoints are overall survival and progression-free survival. platinum-resistant epithelial ovarian, primary peritoneal, or fallopian tube
It is planned to randomize 664 patients. A safety interim analysis will be done cancer; measurable disease by RECIST v1.1; and ECOG score 0-1. Eligible pts
when 24 patients have been randomized and completed at least cycle 1. As of must have received bev-containing treatment for OC. Pts with platinum-
1st February 2019, 24 patients have been randomized. Clinical trial in- refractory disease, increased risk of bleeding, active ocular surface disease,
formation: NCT03353831. or grade . 1 peripheral neuropathy will be excluded. A safety run-in phase for
the DDR will be performed in up to 12 pts who received #5 prior treatment
regimens for PROC. In the DDR, TV will be given at previously decided lower
doses IV 3Q4W for the same dose intensity as the standard 1Q3W dose; the
primary endpoint is incidence of DLTs. In phase 2, pts who received #1 prior
cytotoxic chemo regimen for PROC will be randomized to receive TV admin-
istered as IV 1Q3W or as IV 3Q4W, if shown to be tolerable. The primary
endpoint for phase 2 is confirmed ORR by RECIST v1.1. Secondary end-
points include DOR, time to response, DCR, CA-125 response rate by GCIG
criteria, PFS, OS, pharmacokinetics, and safety. Clinical trial information:
NCT03657043.

TPS5603 Poster Session (Board #421b), Sat, 1:15 PM-4:15 PM TPS5604 Poster Session (Board #422a), Sat, 1:15 PM-4:15 PM
ENGOT-OV43/KEYLYNK-001: A phase III, randomized, double-blind, active- Phase I (safety assessment) of durvalumab (MEDI4736) with focal sensitizing
and placebo-controlled study of pembrolizumab plus chemotherapy with radiotherapy in platinum resistant ovarian, primary peritoneal or fallopian tube
olaparib maintenance for first-line treatment of BRCA-nonmutated advanced epithelial carcinoma. First Author: Anna Tinker, British Columbia Cancer
epithelial ovarian cancer. First Author: Ignace Vergote, BGOG and University Agency, Vancouver, BC, Canada
Hospitals Leuven, Leuven Cancer Institute, Leuven, Belgium
Background: Radiation (RT) of malignant neoplasms can induce immunogenic
Background: There is a significant unmet need to develop new regimens for tumor cell death, alter the tumor micro-environment and enhance recruitment of
BRCA1/2-nonmutated advanced ovarian cancer (OC). The PARP inhibitor anti-tumor T cells. Co-administration of focal RT and an immune checkpoint-
olaparib is approved for women with platinum-sensitive, recurrent OC regardless inhibiting agent may overcome immune-suppressive signals and potentiate sys-
of BRCA1/2 status and, more recently, for newly diagnosed women with BRCA- temic responses. A phase I study is underway to assess the safety of RT combined
mutated OC. In the TOPACIO/KEYNOTE-162 study, the combination of the PD- with PD-L1 inhibition in patients with recurrent epithelial ovarian/fallopian tube/
peritoneal carcinomas (OV). Methods: Women with platinum resistant epithelial OV,
1–blocking antibody pembrolizumab (pembro) and niraparib demonstrated
ECOG PS 0/1 and # 2 lines of treatment in the platinum-resistant setting are el-
efficacy in platinum-resistant relapsed OCirrespective of BRCA1/2 status.
igible. 1 lesion evaluable by RECIST criteria (v 1.1) and 2 additional lesions suitable
ENGOT-OV43/KEYLYNK-001 (ClinicalTrials.gov, NCT03740165) is a phase for RT (minimal treatment volume 4cc) are required. Pre- and on-treatment biopsies
3, randomized, double-blind, active- and placebo-controlled study of pembro for correlative studies are mandatory. The primary objectives are to assess the safety
plus paclitaxel-carboplatin chemotherapy (CT) followed by olaparib mainte- and tolerability of the focal RT combined with the immune checkpoint inhibitor,
nance for first-line treatment of patients with BRCA1/2-nonmutated advanced durvalumab (D), as defined by dose-limiting toxicities (DLTs), and to define the
epithelial OC (EOC). Methods: Patients with stage III or IV BRCA-nonmutated maximum tolerated RT dose and treatment schedule. The secondary objectives are
EOC, primary peritoneal cancer, or fallopian tube cancer will be stratified by to evaluate the clinical activity of focal RT and D [RECIST (v 1.1), GCIG CA-125, and
surgery status (no residual tumor after primary debulking surgery [PDS], residual immune-related response criteria], progression free survival and overall survival. D
tumor after PDS, or planned interval debulking), bevacizumab use, and PD-L1 1500 mg delivered intravenously every 28 days = one cycle. RT to the 2 selected
status (combined positive score , 10 or $10). After one lead-in cycle of CT, target lesions is delivered 24-36 hours prior to the infusion of D. The RT starting
patients will be randomized 1:1:1 to receive: CT + pembro followed by olaparib dose-level is 24Gy (in 4 fractions) per lesion (given Days -1, 1 and 28 of Cycle 1 and
maintenance; CT + pembro followed by placebo; or CT + placebo followed by Day 1 of Cycle 2). A 3+3+3 design will permit more extensive exploration of toxicity if
placebo. The CT regimen will be administered for 5 cycles, and pembro 200 mg DLTs are observed (Table). Investigator assessed DLTs are defined by CTCAE v. 4.03
Q3W will be administered for 35 infusions. Olaparib 300 mg BID maintenance and include the following: any grade $3 adverse event suspected to be related to D or
therapy will start after the end of CT as concomitant treatment with pembro until RT (necrosis or recall reactions at previously irradiated sites, RT induced bowel
discontinuation or for 2 years if the patient has a complete response. Bev- perforation, any unexpected grade 3 or greater toxicity at the site of RT), any
grade $2 allergic or autoimmune event that involves vital organ function, and any
acizumab use is permitted at investigator’s discretion and determined pre-
other grade 3 allergic or autoimmune events that do not resolve to grade 1 before the
randomization. Primary endpoints are investigator-assessed progression-free next scheduled dose of D. Treatment may continue up to 12 months. Enrollment
survival (PFS) per RECIST 1.1 criteria and overall survival. Key secondary began August 2018. To date, no DLTs have been observed at dose level 1 (n= 3) and
endpoints are PFS per RECIST 1.1 assessed by blinded independent central enrollment is ongoing. Clinical trial information: NCT03283943.
review, PFS after next-line treatment, and safety. Enrollment is currently on-
going. Clinical trial information: NCT03740165. Durvalumab dose Total Radiation dose given per
Dose Level (q 28 days) target lesion (2 targets per patient) N
1 (starting dose level) 1500 mg 24 Gy/4 fractions 3+3+3
2 1500 mg 32 Gy/4 fractions 3+3+3

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
340s Gynecologic Cancer

TPS5605 Poster Session (Board #422b), Sat, 1:15 PM-4:15 PM TPS5606 Poster Session (Board #423a), Sat, 1:15 PM-4:15 PM
EUDARIO/ENGOTov-48: A European multicenter randomised phase II trial Phase Ib clinical investigation of intraperitoneal ipilimumab and nivolumab
on the combination of the HSP90 inhibitor ganetespib with carboplatin in patients with peritoneal carcinomatosis due to gynecologic malignancy.
followed by maintenance treatment with niraparib (+/- ganetespib) com- First Author: Emily Hinchcliff, The University of Texas MD Anderson Cancer
pared to platinum-based combination-chemotherapy followed by niraparib Center, Houston, TX
in relapsed platinum-sensitive ovarian cancer patients. First Author: Nicole
Background: The peritoneal cavity is a frequent site of metastasis and re-
Concin, Medical University of Innsbruck, Innsbruck, Austria
currence for gynecologic malignancy, including approximately 80% of ep-
Background: Carboplatin (C) mainly acts by forming interstrand crosslinks ithelial ovarian cancer (EOC) that presents with peritoneal involvement.
(ICL) within the DNA double helix, which can only be removed by the Fanconi These observations have led to the use of intraperitoneal (IP) route of ad-
Anemia (FA) pathway. HSP90 inhibitors destabilise a number of HSP90 ministration for traditional cytotoxic chemotherapy. IP immunotherapy is a
client proteins, such as those governing the FA DNA repair pathway and the recognized but under explored area of clinical investigation with many
G2/M checkpoint (e.g. Chk1 and Wee1). Kramer et al. (Cell Death Differ, potential advantages. Indeed, IP administered antibodies in both animals
2017) showed that the HSP90 inhibitor Ganetespib (G) virtually eliminates a and human subjects are associated with absent or much lower peripheral
functional FA DNA repair complex, therewith preventing the repair of DNA blood concentrations. In addition to higher local and lower systemic ex-
ICL in vitro and vivo. In parallel, G abrogated Chk1 and Wee1 expression and posure, other theoretical advantages include preferential binding to in-
circumvented a G2/M arrest. Consequently, cells with unrepaired DNA traperitoneal and intratumoral immune cells, and absorption through the
damage rushed into mitosis, which resulted in massive tumour cell death. draining lymphatics of the peritoneal cavity. These pelvic and peri-aortic
Furthermore, HSP90 inhibition has been shown to reduce the amount of lymph nodes represent the most relevant lymphoid organs and as such may
BRCA1 in the cell therewith broadening sensitivity towards PARPi and be the ideal site for T cell activation and trafficking back to the peritoneal
preventing acquired PARP resistance. Our trial approach is tested in ovarian tumor. Methods: The trial (NCT03508570) is a single-institution phase Ib
carcinomas with a mutant p53 background. EUDARIO (EUDRACT 017- trial to determine the recommended phase II dosing (RP2D) of IP admin-
004058-40) is funded by the European Commission (FP7 project GAN- istration of nivolumab in combination with ipilimumab. For the purpose of
NET53). Methods: Eligible patients have relapsed platinum-sensitive dose finding, the assessment period for dose limiting toxicity (DLT) is
ovarian cancer, no limits in prior lines, high-grade (but clear cell) histol- 12 weeks. The trial starts with a safety lead-in to confirm the safety of IP
ogy or carcinosarcoma, disease measurable or evaluable according to nivolumab before combining it with ipilimumab. A maximum sample size of
RECIST 1.1. Patients are randomised into 3 treatment arms (1:1:1), a) 12 will be used to find the RP2D for nivolumab, up to 24 patients for the
control arm: C+Gemcitabine or C+Paclitaxel (q3w, 6 cycles, investigator`s combination, and a planned expansion will be carried out such that at least
choice) followed by Niraparib, b+c) 2 experimental arms: C (AUC5, d1) + G 12 EOC patients are treated at RP2D of the intraperitoneal combination
(150mg/m2, d1) q3w 6 cycles followed by either Niraparib alone (arm b) or strategy. The secondary objectives are to describe the pharmacokinetics and
by Niraparib+G (arm c; G at 100mg/m2 weekly, limited to 9 months). toxicities, and to estimate the clinical benefit rate for the expansion cohort.
Niraparib (200/300mg/day) is given in case of SD, PR or CR after platinum- Translational objectives include description of immunologic and biologic
based treatment until disease progression. The main analysis will combine changes in serial blood and IP fluid collections as well as pre and on-
both experimental arms b+c and jointly compare them against arm a using treatment biopsies. Eligibility criteria include recurrent or progressive
log-rank test. Primary endpoint is PFS, secondary endpoints are PFS2, TFST, biopsy-confirmed platinum resistant EOC or other gynecologic cancer with
TSST, safety, ORR, PRO, OS. The first patient was dosed in January 2019. measurable peritoneal disease, and no exposure to prior treatment with
Clinical trial information: NCT03783949. checkpoint inhibition. Enrollment began in January of 2019 with 3 subjects
enrolled to date. Accrual update will be provided at the annual meeting.
Clinical trial information: NCT03508570.

TPS5607 Poster Session (Board #423b), Sat, 1:15 PM-4:15 PM TPS5608 Poster Session (Board #424a), Sat, 1:15 PM-4:15 PM
A phase 3 trial evaluating efficacy and safety of lenvatinib in combination with AtTEnd/ENGOT-en7: A multicenter phase III double-blind randomized con-
pembrolizumab in patients with advanced endometrial cancer. First Author: trolled trial of atezolizumab in combination with paclitaxel and carboplatin in
Vicky Makker, Memorial Sloan Kettering Cancer Center, New York, NY women with advanced/recurrent endometrial cancer. First Author: Nicoletta
Colombo, University of Milan-Bicocca and Istituto Europeo di Oncologia,
Background: Lenvatinib (LEN) is a multikinase inhibitor of vascular endo-
Milan, Italy
thelial growth factor receptors 1–3, fibroblast growth factor receptors 1–4,
platelet-derived growth factor receptor a, RET, and KIT. Pembrolizumab Background: Prognosis of advanced/recurrent endometrial cancer (EC) is
(PEMBRO) is a monoclonal antibody targeting programmed cell death receptor poor with median survival of 12-15 months for patients with measurable
1 (PD-1). Preliminary analyses of a phase 1b/2 study of LEN + PEMBRO disease. Treatment options are limited, with primary management being
showed promising antitumor activity and a manageable safety profile in ad- chemotherapy with carboplatin and paclitaxel. EC is known to be one of the
vanced endometrial cancer (EC). Methods: A multicenter, randomized, open- tumor types with highest mutational load. Ultra- and hyper-mutated EC,
label, phase 3 study (KEYNOTE-775/E7080-G000-309; clinicaltrials.gov which harbor POLE and mismatch repair gene defects respectively, have
NCT03517449) will evaluate efficacy and safety of LEN + PEMBRO vs shown peri-tumoral T cell infiltration and high expression of PD-1 and PD-L1
treatment of physician’s choice (TPC) in patients with advanced EC. Patients proteins, suggesting that immune regulation may enhance specific T cell
must be aged $ 18 years, have advanced EC that progressed after 1 prior responses and result in improved anti-tumour immunity. Preliminary data in
platinum-based therapy, have measurable disease per RECIST v1.1, and an EC patients have shown tumour control activity of the PD-L1 targeting agent
Eastern Cooperative Oncology Group performance status # 1. Patients must atezolizumab. Methods: 550 patients with newly diagnosed, advanced stage
have mismatch repair (MMR) status confirmed by central laboratory via im- III/IV or recurrent EC will be accrued during a period of 24 months with a 1:2
munohistochemistry on archived or fresh tumor biopsy. ~780 patients (~120 randomization ratio into two arms: i. control group receiving standard
MMR-deficient; ~660 MMR-proficient) will be randomized to receive LEN chemotherapy plus placebo IV every 21 days up to 6/8 cycles followed by
20 mg orally once daily and PEMBRO 200 mg intravenously (IV) every 3 weeks placebo until progression; ii. experimental group receiving standard che-
(Q3W) or TPC. Patients will be randomized first according to MMR status; motherapy plus 1200 mg atezolizumab IV every 21 days up to 6/8 cycles
MMR-proficient patients will be further stratified by ECOG PS, geographic followed by atezolizumab until progression. Standard chemotherapy will
region, and prior history of pelvic radiation. TPC is either doxorubicin 60 mg/m2 consist of 175 mg/m2 paclitaxel plus AUC5/6 carboplatin. Stratification
by IV Q3W or paclitaxel 80 mg/m2 by 1-hour IV infusion weekly (3 weeks on/1 factors are: histology, disease stage, microsatellite status, country of ex-
week off). The dual primary endpoints are progression-free survival (PFS; per perimental site. The study is planned to demonstrate a survival increase and
RECIST v1.1 by blinded independent central review) and overall survival (OS). is equally powered for PFS. Secondary endpoints include ORR, duration of
The PFS analysis will occur at the planned interim analysis (~363 OS events in response, PFS2, quality of life, adverse events and compliance. The study is
MMR-proficient patients; ~524 PFS events), and the study will have 99% sponsored by MaNGO group and will involve sites from ENGOT and GCIG
power to detect a hazard ratio (HR) of 0.55 with a 1-sided 0.0005 significance networks across Europe, Japan, Australia and New Zealand. Currently, the
level. A final OS analysis will occur at 518 OS events, when the study will have trial is open in Italy and in Switzerland where a total of 6 patients have been
90% power to detect a HR of 0.75 with a 1-sided 0.0245 significance level. enrolled. Clinical trial information: NCT03603184.
Secondary endpoints include objective response rate, health-related quality of
life, safety and tolerability, and pharmacokinetics. Clinical trial information:
NCT03517449.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Gynecologic Cancer 341s

TPS5609 Poster Session (Board #424b), Sat, 1:15 PM-4:15 PM

NRG GY012: A randomized phase II study comparing single-agent olaparib,
single agent cediranib, and the combination of cediranib/olaparib in women
with recurrent, persistent or metastatic endometrial cancer. First Author:
Helen Mackay, Princess Margaret Cancer Centre, Toronto, ON, Canada
Background: The Cancer Genome Atlas and others identified genomic events
suggesting that endometrial cancer (EC) should be susceptible to DNA repair
inhibition. Mutations in classical homologous recombination genes occur in
22% of EC, ARID1A 41% and PTEN loss occurs in 55% of EC. Data from pre-
clinical models suggest poly ADP-ribose polymerase (PARP) inhibitors alone
or in combination may be an effective therapeutic strategy in EC (Hansen
2016). Combinations of angiogenic inhibitors and PARP inhibitors have
demonstrated synergistic effects and have been well tolerated in other tumor
types. This study has been designed to compare 2 experimental arms ex-
ploring DNA repair inhibition versus cediranib alone which has previously
shown promising activity in GOG 229J (Bender 2015). Methods: This is a
multicenter randomized three arm study for patients with recurrent, met-
astatic or persistent EC. Patients are randomized 1:1:1 to cediranib PO
30 mg OD; olaparib 300 mg PO BID or the combination of cediranib 20 mg
PO OD with olaparib 300 mg PO BID. All treatment cycles are 28 days.
Primary endpoint is progression free survival (PFS). The study is powered to
detect an increase in median PFS from 3.6 (based on cediranib alone) to
7.2 months with 90% power, using a one-sided test with a = 0.05 per
comparison. Forty patients will be enrolled per arm, with an interim futility
analysis planned. Eligibility includes endometroid, serous, and mixed his-
tology EC; at least 1 prior line of chemotherapy (no more than 2 lines for
metastatic disease), prior endocrine or immunotherapy is allowed; ECOG
PS # 2; adequate hepatic, bone marrow, coagulation and renal function.
Archival tumor tissue and blood samples are being collected for translational
studies. The study is open across the NRG network; 24 patients are enrolled
to date . Amendments are planned to include additional arms investigating
combination strategies targeting DNA repair and angiogenesis. Clinical trial
information: 03660826.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
342s Head and Neck Cancer

6000 Oral Abstract Session, Fri, 2:45 PM-5:45 PM 6001 Oral Abstract Session, Fri, 2:45 PM-5:45 PM
Protocol-specified final analysis of the phase 3 KEYNOTE-048 trial of Ado-trastuzumab emtansine in patients with HER2 amplified salivary gland
pembrolizumab (pembro) as first-line therapy for recurrent/metastatic head cancers (SGCs): Results from a phase II basket trial. First Author: Bob T. Li,
and neck squamous cell carcinoma (R/M HNSCC). First Author: Danny Memorial Sloan Kettering Cancer Center, New York, NY
Rischin, Peter MacCallum Cancer Centre, Melbourne, Australia
Background: SGCs are rare tumors with no approved therapy for metastatic
Background: KEYNOTE-048 is a phase 3 study of P or P + chemo (C) vs disease. HER2 amplification occurs in 8% among all SGC histologies, and 25-
EXTREME (E) as 1L therapy for R/M HNSCC (NCT02358031). At the second 33% of the aggressive salivary duct carcinoma (SDC) histologic subtype. We
interim analysis (IA2), P significantly improved OS in the PD-L1 combined hypothesized that ado-trastuzumab emtansine, a HER2 targeted antibody drug
positive score (CPS) $20 and $1 populations and had noninferior OS in the conjugate, may be clinically active in these patients. Methods: A cohort of
total population with favorable safety; P+C significantly improved OS in the patients with HER2 amplified SGCs were enrolled into a multi-histology basket
total population with comparable safety. We present the protocol-specified trial of ado-trastuzumab emtansine, treated at 3.6mg/kg IV every 3 weeks. The
final results. Methods: 882 pts with locally incurable R/M HNSCC and no primary endpoint was overall response rate (ORR) by RECIST v1.1 or PERCIST.
prior systemic therapy in the R/M setting who provided a tumor sample for A Simon two-stage optimal design was applied with type I error rate under
PD-L1 testing were randomized to P 200 mg Q3W for 24 mo (n = 301), P for 2.7%, power of 89%, H0 10%, H1 40%; H0 will be rejected if 6 or more
24 mo + 6 cycles of C (cisplatin 100 mg/m2 or carboplatin AUC 5 Q3W + responses are observed in 24 patients. Other endpoints include duration of
5-FU 1000 mg/m2/d for 4 d Q3W) (n = 281), or E (cetuximab 400 mg/m2 response (DOR), progression-free survival (PFS), and toxicity. HER2 ampli-
loading/250 mg/m2 QW + 6 cycles of chemo) (n = 300). OS superiority was fication was identified by next generation sequencing (NGS), and tumors were
tested sequentially for P+C vs E in the CPS $20 population, then the CPS subsequently tested by fluorescence in situ hybridization (FISH) and immu-
$1 population, and for P vs E in the total population (superiority thresholds: nohistochemistry (IHC). Fluorescence lifetime imaging microscopy - Förster
one-sided P = .0023, .0026, and .0059, respectively). Data cutoff was 25 resonance energy transfer (FLIM-FRET) assessed the propensity for HER2-
Feb 2019 (~25 mo after last pt randomized). Results: P+C significantly HER3 heterodimerization, which leads to receptor internalization. Results: 10
improved OS vs E in the CPS $20 (HR 0.60, 95% CI 0.45-0.82, P = .0004; patients with HER2 amplified SGCs were treated. The median age was 65
median 14.7 vs 11.0 mo) and CPS $1 (HR 0.65, 95% CI 0.53-0.80, P , (range 36-90 years), 90% were male. The median lines of prior systemic
.0001; median 13.6 vs 10.4 mo) populations. HR (95% CI) for PFS was therapy was 2 (range 0-3). ORR was 90% (9/10, 95% CI 56-100%) including
0.76 (0.58-1.01) for CPS $20 and 0.84 (0.69-1.02) for CPS $1. ORR (P+C 5 complete responses after prior trastuzumab, pertuzumab and anti-androgen
vs E) was 42.9% vs 38.2% for CPS $20 and 36.4% vs 35.7% for CPS $1; therapy. After a median follow up period of 12 months (range 4-20 months),
median DOR was 7.1 vs 4.2 mo and 6.7 vs 4.3 mo, respectively. P did median DOR (range 2-19+) and median PFS (95% CI 4–22+ months) were not
not significantly improve OS vs E in the total population (HR 0.83, 95% CI reached. Toxicities included grade 1 or 2 infusion reaction, thrombocytopenia
0.70-0.99, P = .0199; median 11.5 vs 10.7 mo). HR (95% CI) for PFS was and transaminitis; there were no treatment related deaths. HER2 amplification
1.29 (1.09-1.53). ORR (P vs E) was 16.9% vs 36.0%; median DOR was by NGS (fold change 2.8 to 22.8) correlated with HER2/CEP17$2 by FISH (8/
22.6 vs 4.5 mo. All-cause gr 3-5 AE rates were 54.7% for P, 85.1% for P+C, 8 tested) or IHC3+ (10/10 tested). FLIM-FRET tested positive in 3/3.
and 83.3% for E. Conclusion: Overall, KEYNOTE-048 showed that compared Conclusions: Ado-trastuzumab emtansine is highly efficacious in patients
with E, P+C had superior OS in the PD-L1 CPS $20, CPS $1, and total with HER2 amplified SGCs as identified by NGS. This study has met its primary
populations with comparable safety and P had superior OS in the CPS $20 endpoint, and cohort expansion is warranted to confirm these results. Clinical
and $1 populations, noninferior OS in the total population, and favorable trial information: NCT02675829.
safety. These results support pembrolizumab and pembrolizumab + plati-
num + 5-FU as new 1L standards of care for R/M HNSCC. Clinical trial
information: NCT02358031.

6002 Oral Abstract Session, Fri, 2:45 PM-5:45 PM 6003 Oral Abstract Session, Fri, 2:45 PM-5:45 PM
TPExtreme randomized trial: TPEx versus Extreme regimen in 1st line recurrent/ Gemcitabine and cisplatin (GP) induction chemotherapy (IC) plus concurrent
metastatic head and neck squamous cell carcinoma (R/M HNSCC). First Author: chemoradiotherapy (CCRT) versus CCRT alone in locoregionally advanced
Joel Guigay, Department of Medical Oncology, Antoine Lacassagne Compre- nasopharyngeal carcinoma (NPC): A phase 3, multicenter, randomized
hensive Cancer Centre, FHU OncoAge, Université Côte d’Azur, Nice, France controlled trial. First Author: Jun Ma, Sun Yat-sen University Cancer Center,
Guangzhou, China
Background: After promising results from the GORTEC TPEx phase II trial, the
role of taxane instead of 5FU in 1st-line R/M HNSCC chemotherapy (CT) Background: GP regimen has been established as the standard first-line
remained to be confirmed by comparing TPEx to the reference EXTREME treatment option for patients with recurrent/metastatic NPC. However, its ef-
regimen. Methods: Randomized (1:1), open-label trial. Main inclusion criteria ficacy in locoregionally advanced disease remains unclear. Methods: Patients
were R/M HNSCC not suitable for locoregional treatment, age 18-70 years, with previously untreated, non-metastatic stage III-IVB (except T3-4N0M0,
PS ,2, creatinine clearance .60ml/min, prior cisplatin ,300 mg/m². Ref- AJCC 7th) NPC, aged 18–64 years without severe comorbidities were eligible.
erence EXTREME regimen (arm A: 6 cycles every 3 weeks (Q3W) of 5FU– They were randomly assigned (1:1) to receive GP IC (gemcitabine 1 g/m2on days
cisplatin-cetuximab (cetux) followed by weekly cetux maintenance) was 1 & 8, cisplatin 80 mg/m2 on day 1, q3w for 3 cycles) plus CCRT (cisplatin
compared to TPEx regimen (arm B: 4 cycles Q3W of docetaxel 75mg/m²– 100 mg/m2, q3w for 3 cycles, concurrently with intensity-modulated radio-
cisplatin 75mg/m²- cetux 250mg/m² with mandatory G-CSF support followed therapy) or CCRT alone. The primary endpoint was failure-free survival (FFS). The
by every 2W cetux 500mg/m² maintenance). The primary endpoint was calculated sample size was 238 per group, with an 80% power (two-sided a
Overall Survival (OS). To detect a hazard ratio (HR) of 0.72 (median OS in- 0.05) to detect a treatment failure hazard ratio (HR) of 0.52. Results: From Dec
crease from 10.1 to 14.0 months (mo) with 88% power, 2-sided significance 2013 to Sep 2016, 480 patients from 12 centers were randomly assigned to
level of 0.05, 374 deaths were required. 540 patients (pts) were planned to IC+CCRT (n = 242) or CCRT alone (n = 238) group. Baseline characteristics
enroll. Results: 539 pts were enrolled in 37 mo. Median age was 60 years, were well balanced. After a median follow-up of 39 months, 3-year FFS was
93% were smokers, 40% had oropharyngeal tumor (p16 or HPV DNA was done 85.8% in the IC+CCRT group and 77.2% in the CCRT alone group (intention-to-
in 85%, positive in 28%). In arm A, 44% of pts received all CT cycles vs 72% in treat population; HR 0.53, 95% confidence interval 0.34–0.81; P = 0.003). In
arm B. Delays in administration were more frequent in arm A (27% vs 10%). GP+CCRT group, 239 patients started GP IC and 231 (96.7%) completed all
Cisplatin was more frequently switched to carboplatin in arm A (34% vs 9%). three cycles. The most common $grade 3 adverse events (AE) in IC+CCRT and
Toxicity was lower in arm B: 34% pts had grade $4 adverse events during CT in CCRT group were mucositis (28.9% vs. 32.1%), neutropenia (28.0% vs.
arm B vs 50% in arm A (p,0.001). Less pts in arm A started maintenance than 10.5%) and leukopenia (26.4% vs. 20.3%). Conclusions: Adding GP IC to CCRT
in arm B (53% vs 73%). At time of analysis, the median follow-up duration was significantly improved FFS in locoregionally advanced NPC and is well tolerated
30 mo and 406 pts had died. OS was not significantly different between arms: with favorable toxicity profile. Clinical trial information: NCT01872962.
HR=0.87 (95%CI: 0.71-1.05), p=0.15. Median OS was 13.4 mo in arm A vs
IC+CCRT (%) CCRT (%) P value
14.5 in arm B. 2-year OS rate was 21.0% in arm A vs 28.6% in arm B.
Conclusions: This large randomized trial confirmed the encouraging survival Intention-to-treat population n = 242 n = 238
results of the TPEx regimen observed in the first phase II. OS in both arms was 3-y failure-free survival 85.8 77.2 0.003
higher than observed in previous randomized CT or immunotherapy combi- 3-y overall survival 94.9 90.7 0.02
3-y distant metastasis-free survival 91.6 85.9 0.03
nation trials. Despite lack of significant OS increase, taxane based TPEx 3-y locoregional failure-free survival 92.5 92.1 0.75
regimen appears to be a new option in 1st line R/M HNSCC, with a shorter time Safety population n = 239 n = 237
on CT and significantly lower toxicity than the EXTREME regimen. Clinical trial Completed radiotherapy 100.0 99.2 0.25
information: NCT02268695. Received concurrent cisplatin ‡ 200 mg/m2 80.6 95.8 , 0.001
‡ grade 3 AEs during IC 38.9
‡ grade 3 AEs during CCRT 65.3 55.3 0.03

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Head and Neck Cancer 343s

6004 Oral Abstract Session, Fri, 2:45 PM-5:45 PM 6006 Oral Abstract Session, Fri, 2:45 PM-5:45 PM
Induction chemotherapy followed by concurrent chemoradiotherapy versus A phase II randomized trial for early-stage squamous cell carcinoma of the
concurrent chemoradiotherapy alone in locoregionally advanced nasopharyn- oropharynx: Radiotherapy versus trans-oral robotic surgery (ORATOR). First
geal carcinoma: Long-term results of a phase 3 multicenter randomized Author: Anthony Charles Nichols, Western University and London Health
controlled trial. First Author: Ming-Yuan Chen, Sun Yat-Sen University Can- Sciences Centre, London, ON, Canada
cer Center, Guangzhou, China
Background: The incidence of OPSCC has risen rapidly, due to an epidemic of
Background: Initial 3-year results from our clinical trial in locoregionally ad- human papillomavirus (HPV) infection. Radiation therapy (RT) has historically
vanced nasopharyngeal carcinoma (NPC) patients showed that induction been the standard treatment, but transoral robotic surgery (TORS) has sur-
chemotherapy (IC) with cisplatin and fluorouracil (PF) resulted in improved passed RT in the US as the most common approach, based on assumptions
disease-free survival (DFS) with a marginally significant effect on distant of reduced toxicity or improved quality of life (QOL). No randomized trials
metastasis-free survival (DMFS), but the effect of IC on locoregional relapse- have previously compared these treatments. Methods: The ORATOR trial
free survival (LRRFS) and overall survival (OS) did not differ significantly. Here, (NCT01590355) enrolled patients with T1-T2 N0-2(#4 cm) OPSCC ame-
we present 5-year follow-up results. Methods: Our trial was a randomized, nable to TORS. We randomly assigned patients, stratified by p16 status, to RT
open-label phase 3 trial comparing IC followed by concurrent chemo- (70 Gy/35 fractions, with chemotherapy if N1-2) vs. TORS (6 adjuvant
radiotherapy (CCRT) versus CCRT alone in patients with stage III-IVB (except [chemo]RT based on pathology). The primary endpoint was a definitive
T3N0-1) NPC. The IC followed by CCRT group received cisplatin (80 mg/m² comparison of swallowing QOL at 1-year using the MD Anderson Dysphagia
d1) and fluorouracil (800 mg/m² d1-5) every three weeks for two cycles before Inventory (MDADI), powered to detect a 10-point improvement (a clinically-
CCRT. Both groups were treated with 80 mg/m² cisplatin every three weeks meaningful change [CMC]) in the TORS arm. Secondary endpoints included
concurrently with radiotherapy. The primary endpoints were DFS and DMFS. adverse events (AEs), other QOL outcomes [including EORTC scales, the Voice
We did efficacy analyses in the 476 randomized patients (intention-to-treat Handicap Index-10, Neck Dissection Impairment Index, and Patient Neuro-
population). Results: After a median follow-up of 82.6 months, the 5-year DFS toxicity Questionnaire], overall- and progression-free survival (OS, PFS). All
rate was 73.4% (95% confidence interval (CI) 67.7-79.1) in the IC followed by analyses were pre-specified and intention-to-treat. Results: Between 2012
CCRT group and 63.1% (95% CI 56.8-69.4) in the CCRT alone group (P = and 2017, 68 patients were randomized (n = 34 in each arm), in Canada and
0.005). The 5-year DMFS rate was also significantly higher in the IC followed Australia. Median age was 59 years; 87% were male. Primary tumor sites were
by CCRT group (82.8%, 95% CI 77.9-87.7) than in the CCRT alone group palatine tonsil (74%) or base of tongue (26%). Arms were well-balanced for
(73.1%, 95% CI 67.2-79.0, P = 0.013). Our updated analysis revealed an OS baseline factors, including p16 status (88% in each arm). Median follow-up
benefit of IC: the 5-year OS rate was 80.8% in the IC followed by CCRT group was 27 months. MDADI scores at 1-year were statistically superior in the RT
versus 76.8% in the CCRT alone group (P = 0.045). There were no significant arm (mean 6 SD: 86.9 6 11.4 vs. 80.1 6 13.0 in the TORS arm; p = 0.042),
differences in the rate of grade 3–4 late adverse events during follow-up but not meeting the definition of a CMC. For the other QOL metrics, outcomes
between the two groups. Conclusions: IC followed by CCRT provides long-term were similar at 1-year. Feeding tube rates at 1-year were 3% (n = 1) vs. 0%
DFS, DMFS, and OS benefits compared with CCRT alone in locoregionally respectively. Rates of treatment-related grade $2 AEs were similar (91% vs.
advanced NPC and, therefore, can be recommended for these patients. 100%, p = 0.24), with more neutropenia, constipation and tinnitus in the RT
Clinical trial information: NCT00705627. arm and more trismus in the TORS arm (all p , 0.05). There was one TORS
bleeding-related death. OS and PFS were similar. Conclusions: RT had su-
perior swallowing QOL scores at 1 year compared to TORS, but the difference
was not a CMC. Toxicities differed between the arms. This study provides the
first level 1 evidence to inform patients of the QOL impact of both approaches.
Clinical trial information: NCT01590355.

6007 Oral Abstract Session, Fri, 2:45 PM-5:45 PM 6008 Oral Abstract Session, Fri, 2:45 PM-5:45 PM
Neck dissections based on sentinel lymph node navigation versus elective neck Checkpoint inhibitors assessment in oropharynx cancer (CIAO): Safety and
dissections in early oral cancers: A randomized, multicenter, non-inferiority interim results. First Author: Renata Ferrarotto, The University of Texas MD
trial. First Author: Yasuhisa Hasegawa, Asahi University Hospital, Gifu, Japan Anderson Cancer Center, Houston, TX
Background: The objective of the study is to evaluate the non-Inferiority of Background: Anti-PD-1/PD-L1 are active in metastatic oropharynx squamous
survival, the superiority of postoperative disability, and the complication of cell carcinoma (OPC). Durvalumab (durva) and tremelimumab (tremi) target
the neck in neck dissections based on sentinel lymph node navigation in respectively PD-L1 and CTLA-4, which in combination may be synergistic.
early oral cancer patients, compared with standard selective neck dissec- Here we report the safety and interim results of durva vs. durva+tremi prior to
tions. Methods: This study was a randomized, multicenter, non-inferiority surgery in a window of opportunity trial in OPC. Methods: Pts were randomized
trial at 16 institutions in Japan. Eligibility criteria included histologically 1:1 to durva 1500 mg or durva 1500 mg + tremi 75 mg IV Q4W x 2 cycles. The
confirmed squamous cell carcinoma in the oral cavity; clinical categories T1 primary objective was to quantify pre- and post-treatment differences in CD8+
and T2, N0M0 by UICC TNM classification 7th edition, clinical depth of tumor infiltrating lymphocytes for the two arms. Secondary objectives included
invasion (DOI) of T1 was over 4mm (defined as late T1); previously untreated; safety, toxicity, ORR by RECIST, fraction of patients undergoing surgery at 8
age at least 18 years; and written informed consent. We randomly assigned wks, and percentage viable tumor cells in the surgical specimen. Serial pre-
patients (1:1) to receive either sentinel lymph node biopsy (SNB) or standard and post-treatment blood and tumor specimens were collected for ongoing
selective neck dissections (ND) with stratification of T category (lateT1 vs correlative analyses. Results: 28 pts enrolled: median age 64y, 27 (96%)
T2) and subsite (tongue vs others). The primary endpoint was 3-year overall male, 19 (68%) newly diagnosed, most (63%) at stage IVA (AJCC 7th Ed), 9
survival with a non-inferiority margin of 12%. Sentinel nodes (SNs) were (32%) had locoregional recurrence, 24 (86%) p16 positive, and 22 (79%)
detected using radioisotope method and examined with multislice frozen had # 10 PPY smoking history. Median follow-up was 7.6 months. The most
section analysis intraoperatively, following HE and cytokeratin stain for a common AEs were fatigue (36%), leukopenia/lymphopenia (25%), trans-
final postoperative diagnosis. Patients with positive SNs had neck dissec- aminitis (25%), and rash (21%). Grade 3 AEs occurred in 4 (14%) pts: 2
tions in a one-stage or back up procedure. Results: Between November 2011 elevated lipase, 1 diarrhea, and 1 hepatitis, all were manageable. There were
and January 2016, 271 patients were enrolled and randomized to SNB no grade .3 AEs. ORR was 43%: 50% had SD (including 29% tumor
group (134 patients) and ND group (137 patients) with a median follow-up of shrinkage in 1 pt). Treatment effect in the surgical specimen was observed in
37 months (IQR 36-39). Pathological positive nodal status was 34% 19 (79%) of 24 evaluable pts; 2 pts had major pathologic response (# 10%
(46/132) in SNB group and 26% (34/133) in ND group (Chi-Square p = viable tumor) at the primary site. Efficacy was equivalent in both arms. The 2
0.10). 3-year overall survival in SNB group was 89% (95%CI 82-93%), pts with PD and 1 pt with SD were switched to chemotherapy after durva +/-
which was non-inferior to that in ND group (86%, 95%CI 79-91%). 3-year tremi before resection; interestingly, each achieved a pCR in the primary. Most
relapse-free survival was 80% (95%CI 72-86%) in SNB group and 81% pts (57%) didn’t receive radiotherapy after surgery. There was a statistically
(95%CI 73-87%) in ND group. Arm abduction of postoperative 1 and significant association between ORR and treatment effect (p=0.014). The
3 months in ND group was disturbed significantly compared with SNB group. median percentage of viable tumor in the primary was 37.5% in pts with PR,
Conclusions: SNB navigated ND could replace elective ND without survival and 82.5% in SD (p=0.003). Conclusions: Durva +/- tremi prior to surgery was
disadvantage and reduce postoperative disability of the neck in patients with well tolerated in OPC pts. Activity is encouraging with treatment effect seen in
early oral cancer. Clinical trial information: 000006510. 79% of pts. The primary endpoint and complete efficacy data will be pre-
sented. Clinical trial information: NCT03144778.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
344s Head and Neck Cancer

6009 Clinical Science Symposium, Sun, 8:00 AM-9:30 AM 6010 Clinical Science Symposium, Sun, 8:00 AM-9:30 AM
Genomic and transcriptomic landscape of oral pre-cancers (OPCs) and risk of Evolutionary action score of TP53 analysis in pathologically high-risk HPV-
oral cancer (OC). First Author: William Nassib William, The University of negative head and neck cancer from a phase II clinical trial: NRG Oncology
Texas MD Anderson Cancer Center, Houston, TX RTOG 0234. First Author: Chieko Michikawa, Department of Head and Neck
Surgery, The University of Texas MD Anderson Cancer Center, Houston, TX
Background: The molecular landscape of OPCs and its association with neo-
plastic progression is largely unknown. We report the results of high throughput Background: An evolutionary action scoring algorithm (EAp53) based on
DNA/RNA profiling of OPCs from pts in the Erlotinib Prevention of Oral Cancer phylogenetic sequence variations and speciation stratifies head and neck
trial (EPOC), with long-term prospective follow-up. Methods: We performed next squamous cell carcinoma (HNSCC) patients bearing TP53 missense mutations
generation sequencing of 201 cancer genes (MD Anderson T200 platform) in as high-risk (high, EAp53$75), associated with poor outcomes, or low-risk
170 OPCs from EPOC, and RNA profiling using HTG EdgeSeq Oncology (low), with similar outcomes as TP53 wild-type (wt), and has been validated
Biomarker Panel containing 2,560 transcripts in a subset of 141 OPCs. 73 pts as a reliable prognostic marker. This study is designed to further validate prior
developed invasive OC during a median follow up of 7.3 years, from whom 23 findings that EAp53 is a prognostic marker for locally advanced HNSCC
paired OCs were profiled to characterize the evolutionary trajectory from OPCs to patients, and assess its predictive value for treatment outcomes to adjuvant
OCs. OPC molecular features were correlated with OC-free survival. Results were bio-chemoradiotherapy. Methods: Eighty one resection specimens from pa-
compared with TCGA invasive OC DNA/RNA profiles and an independent set of tients treated surgically for stage III or IV human papillomavirus-negative
86 OPCs with RNA data. Results: Similar to TCGA, C . T was the predominant (HPV(-)) HNSCC with high-risk pathologic features, who received either Arm 1)
substitution. The top mutated genes in OPCs were TP53 (29%), CDKN2A radiotherapy(RT)+cetuximab(CTX)+cisplatin or Arm 2) RT+CTX+docetaxel, as
(15%), NOTCH1 (11%) and PIK3CA (7%), which were also frequently mutated adjuvant treatment in a phase II randomized clinical trial (RTOG 0234) un-
(albeit at higher rates) in OCs from EPOC or TCGA. There was a progressive derwent TP53 targeted sequencing, and EAp53 scoring. The EAp53 scores
increase of tumor mutation burden (TMB, P , 0.05) and frequency of high-risk were correlated with clinical outcomes. Due to limited sample sizes, patients
TP53 mutations (P = 0.02) from hyperplasia, to dysplasia, to invasive OCs (P , were combined into 2 EAp53 groups: wt/low and high/other. Results: At
0.05). Median TMB was higher in OPCs from pts who developed OC (2.45 mut/ median follow-up of 10 years, there was a significant interaction between
Mb) vs those who did not (1.22 mut/Mb) (P , 0.01). Pts with TP53 mutated treatment and EAp53 group for overall survival (OS) (p = 0.008), disease-free
OPCs had shorter OC-free survival compared to TP53 wild-type (HR 1.81, 95% survival (DFS) (p = 0.05) and distant metastasis (DM) (p = 0.004). Within arm
CI 1.13-2.90, P = 0.01). A prognostic score was derived from a Cox regression 2, high/other showed worse OS [HR 4.69 (1.52-14.50)], DFS [HR 2.69 (1.16-
model which identified 12 mRNA transcripts associated with OC risk (HR 4.72, 6.21)], and had higher DM [HR 11.71 (1.50-91.68)] than wt/low. Within arm
95% CI 2.51-8.86, P , 0.01), and which was validated in the independent set 1, there was no significant difference by EAp53 in OS, DFS and DM. Within the
of 86 OPCs (HR 2.68, P , 0.01). This score was also associated with shorter wt/low group, arm 2 had better OS [HR 0.11 (0.03-0.36)], DFS [HR 0.24
overall survival when applied to invasive OCs from TCGA pts (HR 2.72, P , (0.09-0.61)], and DM [HR 0.04 (0.01-0.31)] than arm 1 but this was not
0.01). Conclusions: This is the first large-scale cohort of OPC pts with long- found in high/other. Conclusions: High/other EAp53 scores were associated
term, prospective follow up and comprehensive RNA/DNA profiling. We with worse survival for patients in arm 2. Arm 2 is associated with better
demonstrated an association between TMB, TP53 mutations, a 12-gene RNA survival than arm 1 for patients with wt/low EAp53. This benefit appears to be
signature score in OPCs, and OC risk. This study may provide a framework for largely driven by a reduction in DM. Further validation is required to determine
similar efforts of pre-cancer molecular profiling in the oral cavity and other sites, whether EAp53 can be used for personalized post-operative treatment de-
such as the PreCancer Atlas of the NCI. cisions in HPV(-) HNSCC.

6011 Clinical Science Symposium, Sun, 8:00 AM-9:30 AM 6012 Poster Discussion Session; Displayed in Poster Session (Board #1),
PIK3CA mutation as a prognostic factor in HPV-associated oropharynx Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
cancer. First Author: Brian T. Beaty, University of North Carolina, Chapel Sat, 4:30 PM-6:00 PM
Hill, NC EAGLE: A phase 3, randomized, open-label study of durvalumab (D) with or
without tremelimumab (T) in patients (pts) with recurrent or metastatic head
Background: PIK3CA is the most frequently mutated gene in HPV-
and neck squamous cell carcinoma (R/M HNSCC). First Author: Lisa F. Licitra,
associated oropharyngeal SCC (OPSCC), with a prevalence of 20-30%.
Head and Neck Cancer Medical Oncology 3 Unit, Fondazione IRCCS Istituto
While PIK3CA mutations have been associated with adverse outcomes in
Nazionale dei Tumori and University of Milan, Milan, Italy
cervical cancer, their prognostic significance in HPV-associated OPSCC
remains unknown. We sought to elucidate the significance of PIK3CA Background: EAGLE is a phase 3 study evaluating efficacy of D (anti-PD-L1 mAb)
mutations in a prospective cohort of HPV-associated OPSCC patients treated monotherapy and D+T (anti-CTLA-4 mAb) vs standard of care (SOC) in pts with R/M
with definitive chemoradiation (CRT). Methods: Seventy-eight patients with HNSCC who progressed following platinum-based therapy (NCT02369874).
HPV-associated OPSCC were prospectively enrolled on three protocols: LCCC Methods: Pts were randomized 1:1:1 to D 10 mg/kg IV every 2 weeks (Q2W), D+T (D
1121 (NCT03161821) or two phase II clinical trials of de-intensified CRT 20 mg/kg IV Q4W + T 1 mg/kg IV Q4W for 4 doses, then D 10 mg/kg IV Q2W), or SOC
(NCT02281955 / NCT03077243). De-intensified regimen was 60 Gy IMRT (investigator’s choice: cetuximab, taxane, methotrexate, or fluoropyrimidine-based
regimen). The primary endpoint was overall survival (OS) with dual primary objectives
with concurrent cisplatin (30mg/m2). Next-generation sequencing of tumor
of D+T vs SOC and D vs SOC. Additional endpoints included objective response rate
samples was performed using a targeted panel-based assay (UNCSeq),
(ORR), duration of response (DoR), and adverse events (AEs). Results: 240 pts were
including over 200 genes. We estimated disease-free survival (DFS) using randomized to D, 247 to D+T and 249 to SOC. An imbalance for Eastern Cooperative
the Kaplan-Meier method and compared treatment groups with two-sided Oncology Group performance status (ECOG PS) was seen in favor of the SOC arm (D,
log-rank test (Medcalc). Results: Sequencing was performed in 78 patients PS 0 = 26%, PS 1 = 74%; D+T, PS 0 = 26%, PS 1 = 74%; SOC, PS 0 = 32%, PS 1 =
with a median follow-up of 24 months. Seventy-five patients received 60Gy; 68%). The risk of death was not statistically significantly different for D compared
three patients received 70Gy. Ten patients had disease recurrence (2 re- with SOC (HR: 0.88; 95% CI: 0.72–1.08; P = 0.20) or D+T vs SOC (HR: 1.04; 95%
gional only, 5 distant only, 3 regional and distant). Thirty-eight of 78 patients CI: 0.85–1.26; P = 0.76). Efficacy data are provided in the table. Treatment-related
had at least one mutation identified (17 PIK3CA, 4 PTEN, 3 KRAS, 3 AEs Grade $3 were reported in 10.1% of pts (regardless of causality Grade $3 AEs
FBXW7, 3 FGFR3, 2 TP53, 2 prothrombin 20210, 1 NRAS, 1 BRCA1, 1 were 41.4%) in the D arm, 16.3% (51.2%) for D+T, and 24.2% (44.2%) for SOC.
factor V Leiden, 1 FLT3, 1 RAD50, 1 PIK3R1). The most common site of Following treatment, 2% of pts in D, 5% in D+T and 15% in SOC received im-
PIK3CA mutation was the helical domain (E545K – 8/17, E542K – 2/17). munotherapy. Conclusions: D and D+T did not demonstrate a statistically significant
Despite similar T/N staging and tobacco pack years, patients with WT- improvement in OS compared to standard chemotherapy in pts with R/M HNSCC.
PIK3CA had significantly higher DFS (93%) compared with 65% for pa- Median OS and ORR of D arm were similar to other studies with checkpoint in-
tients with PIK3CA mutations (p = 0.0009). Patients with mutations other hibitors. The SOC arm outperformed what has been seen for SOC arms in previous
than PIK3CA also had improved DFS relative to those with PIK3CA muta- studies; subsequent immunotherapy may have confounded the OS analyses. The
safety profile for D and D + T in R/M HNSCC is consistent with previous trials. Clinical
tions (96% vs. 65%; p = 0.0147). Conclusions: PIK3CA mutation is as-
trial information: NCT02369874.
sociated with worse DFS in a prospective cohort of newly diagnosed HPV-
associated OPSCC patients treated with definitive CRT. These findings D (n = 240) D+T (n = 247) SOC (n = 249)
suggest that patients with PIK3CA mutations may not be suitable for de- Median OS, mo (95% CI) 7.6 (6.1–9.8) 6.5 (5.5–8.2) 8.3 (7.3–9.2)
intensified therapy and investigation of novel treatment strategies may be Survival rate, % (95% CI)
appropriate. 12 mo 37.0 (30.9–43.1) 30.4 (24.7–36.3) 30.5 (24.7–36.4)
24 mo 18.4 (13.3–24.1) 13.3 (8.9–18.6) 10.3 (5.7–16.5)
ORR, % (95% CI) 17.9 (13.3–23.3) 18.2 (13.6–23.6) 17.3 (12.8–22.6)
DOR, mo 12.9 7.4 3.7

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Head and Neck Cancer 345s

6013 Poster Discussion Session; Displayed in Poster Session (Board #2), 6014 Poster Discussion Session; Displayed in Poster Session (Board #3),
Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Sat, 4:30 PM-6:00 PM Sat, 4:30 PM-6:00 PM
Palbociclib plus cetuximab versus placebo plus cetuximab in platinum-resistant, Results of a phase 2a, multicenter, open-label, study of RM-1929 photo-
cetuximab-naive, HPV-unrelated head and neck cancer: A double-blind immunotherapy (PIT) in patients with locoregional, recurrent head and neck
randomized phase II trial (PALATINUS). First Author: Douglas Adkins, squamous cell carcinoma (rHNSCC). First Author: David M. Cognetti, Thomas
Division of Medical Oncology and Alvin J. Siteman Cancer Center, Jefferson University Hospital, Philadelphia, PA
Washington University School of Medicine, St. Louis, MO
Background: Patients with rHNSCC who have failed standard of care have poor
Background: Cetuximab monotherapy results in a median overall survival (OS) prognoses and limited therapeutic options. In this study, final results are
of approximately 6 months (mo) in platinum-resistant recurrent/metastatic head reported of a phase 2a trial of photoimmunotherapy (PIT) with a targeted drug
and neck squamous cell carcinoma (HNSCC). HNSCC unrelated to human RM-1929, consisting of the EGFR-directed antibody cetuximab conjugated
papillomavirus (HPV) is driven by hyperactivation of the CDK4/6 and cyclin D1 to a photoactivatable dye (IRDye 700DX). Binding of the antibody-dye con-
(CD1) regulatory complex, resulting in cell cycle progression and tumor growth, jugate to cancer cells followed by photoactivation with nonthermal red light
suggesting that CDK4/6 inhibition can be a rational therapeutic strategy in this induces selective and rapid necrosis of the cancer cells, with minimal damage
setting. Palbociclib (PAL) is a selective CDK4/6 inhibitor that may reverse to surrounding tissue. Methods: A phase 2a, multicenter, open-label, study of
cetuximab resistance by countering the actions of deregulated CD1. PAL plus RM-1929 PIT in patients with locoregional, rHNSCC who could not be sat-
an epidermal growth factor receptor inhibitor synergistically reduced cell via- isfactorily treated with surgery, radiation, or platinum chemotherapy was
bility of HPV-unrelated HNSCC cell lines. In a single-arm, multicenter trial of conducted to evaluate the safety and efficacy of the drug, RM-1929. For each
platinum-resistant, cetuximab-naive, HPV-unrelated HNSCC, PAL in combi- treatment, nonthermal red light (690 nm) was applied to the tumors 24 hours
nation with cetuximab resulted in a median OS of 9.5 mo. Methods: In a double- post IV infusion of the drug. Surface illumination was administered for su-
blind randomized phase II trial, patients (pts) with platinum-resistant, cetux- perficial tumors and interstitial illumination via intratumoral placement of fiber
imab-naı̈ve, HPV-unrelated HNSCC were treated with cetuximab plus either optic diffusers for deep tumors. Therapeutic response was assessed using CT
PAL (arm A) or placebo (arm B). Pts were stratified by performance status (PS) RECIST 1.1 by an independent blinded radiologist. Results: Thirty rHNSCC
and prior immunotherapy (IT). 120 pts were required for 1:1 randomization to patients were enrolled. There were no dose-limiting toxicities and one Grade 1
have $ 80% power to detect a hazard ratio (HR) of 0.6 (corresponding to a photosensitivity reaction. Most reported AEs were mild to moderate in severity
median OS of 10 mo in arm A and 6 mo in arm B) using a 1-sided log-rank test with 96.7% (29/30) of patients with Grade 1 and 83.3% (25/30) with Grade 2,
P=0.10). Key secondary endpoints included progression-free survival (PFS), respectively. There were 13 (43.3%) patients who had at least one SAE. 86%
adverse events (AEs), and p16 status. Results: Pts (n=125) were randomized (19/22) of SAEs were deemed unlikely related to treatment, including all 3
(arm A, 65; arm B, 60). PS and prior IT were balanced between the arms. fatal SAEs. Three SAEs were reported to be possibly/probably related to
Median (95% CI) follow-up for OS was 15.9 (15.0–19.4) mo. Median OS was treatment (site/oral pain, tumor hemorrhage, and airway obstruction). ORR was
9.7 (7.3–13.9) mo in arm A and 7.8 (6.7–10.6) mo in Arm B (stratified by PS: 50% (15/30) with 16.7% (5/30) CR and 86.7% (26/30) DCR. Median PFS
HR=0.82 [95% CI, 0.54–1.25], P=0.18). Median PFS was 3.9 mo in arm A and OS results will be forthcoming. Conclusions: These data indicate that RM-
and 4.6 mo in arm B (stratified by PS: HR=1.00 [0.7–1.5], P=0.5). Hema- 1929 PIT treatment was generally well tolerated with majority of AEs as mild to
tologic AEs were more common in arm A. Only 11 pts (9%) received IT after moderate in severity. Preliminary data showed favorable response rates in a
being treated on the trial. Conclusions: Among pts with platinum-resistant, heavily pre-treated population. A global phase 3 clinical trial is currently
HPV-unrelated HNSCC, PAL plus cetuximab resulted in a trend of prolonga- underway. Clinical trial information: NCT02422979.
tion of median OS compared with cetuximab. Clinical trial information:
NCT02499120.

6015 Poster Discussion Session; Displayed in Poster Session (Board #4), 6016 Poster Discussion Session; Displayed in Poster Session (Board #5),
Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Sat, 4:30 PM-6:00 PM Sat, 4:30 PM-6:00 PM
Primary analysis of phase 2 results of cemiplimab, a human monoclonal anti- Discordant treatment response in primary tumors and lymph node metastases
PD-1, in patients (pts) with locally advanced cutaneous squamous cell carci- after four weeks of preoperative PD-1 blockade in head and neck squamous
noma (laCSCC). First Author: Michael Robert Migden, Departments of Der- cell carcinoma (HNSCC). First Author: Adam Luginbuhl, Thomas Jefferson
matology and Head and Neck Surgery, University of Texas, MD Anderson Cancer University Hospital, Department of Otolaryngology, Philadelphia, PA
Center, Houston, TX
Background: Discordant radiographic responses are described in other tumor
Background: Cemiplimab (REGN2810) produced substantial antitumor ac- types in response to immunotherapy with response at some anatomic sites and
tivity with durable responses in Phase 1 CSCC expansion cohorts and Phase 2 progression in others. Here we determined the frequency of discordant
metastatic (m) CSCC cohort. We now present the primary analysis of the Phase 2 treatment effects (TE) in HNSCC patients treated with immunotherapy in the
laCSCC cohort (NCT02760498; data cutoff date: Oct 10, 2018). Methods: Pts context of a neoadjuvant trial. Methods: 23 Patients with resectable primary
with laCSCC received cemiplimab 3 mg/kg IV every 2 weeks (Q2W). Tumor HNSCC were 1:1 randomized to receive nivolumab (240 mg IV Q 2 weeks x 2)
measurements were performed Q8W. The primary objective was to evaluate or nivolumab and tadalafil 10 mg daily. Surgery was performed 4 weeks after
objective response rate (ORR; complete response [CR] + partial response [PR]) the first nivolumab infusion. Resection specimens were graded histopatho-
according to independent central review (per RECIST 1.1 for scans; modified logically by two pathologists. Areas exhibiting TE (defined by fibrosis with
WHO criteria for photos). Results: 78 pts were enrolled (59 M/ 19 F; median age: chronic inflammation, foamy macrophage reaction and multinucleated giant
74 years; ECOG PS: 0 in 38 pts, 1 in 40 pts; primary CSCC site: head/neck in cells) were expressed relative to the total tumor area. This was assessed in the
79.5%; prior systemic therapy: 15.4%; prior radiotherapy: 55.1%). Median primary tumor and all lymph nodes (LN). Each primary lesion and individual LN
duration of follow-up was 9.3 months (range: 0.8–27.9). ORR by central review was defined as a) no response 0%TE, b) minimal response 1-19%TE, c)
was 43.6% (95% CI: 32.4–55.3; 10 CRs and 24 PRs); investigator-assessed response 20-99% or d) complete response 100%. Concordance was defined if
(INV) ORR was 52.6% (95% CI: 40.9–64.0; 13 CRs and 28 PRs). Median primary lesion and LNs were in the same ordinal data set. Results: 11/23
duration of response (DOR) has not been reached. The longest DOR at data cut- (48%) of subjects experienced concordant TE in the primary tumor and LNs.
off was 24.2 months and was still ongoing. Durable disease control rate (stable Within this cohort, 3 patients had a complete pathologic response both at the
disease or response for $16 weeks) was 62.8% (95% CI: 51.1–73.5). Median primary site and LNs. In contrast, 12/23 patients (52%) revealed discordant
observed time to response was 1.9 months (range: 1.8–8.8). Median TE between the primary tumor sites (average of 17% TE) and involved LNs
progression-free and overall survival have not been reached. Tumor PD-L1 status (average of 62% TE), (p= 0.018; signed rank test). Interestingly, in the dis-
is available for 48/78 pts, tumor mutational burden analysis (from targeted cordant group, TE effects in LNs were invariably greater than in primary lesions.
exome panel) is ongoing for $40/78 pts; response correlation analyses are In 5 of 11 patients with multiple involved LN, the TE varied between nodes.
planned. The most common treatment-emergent adverse events (AEs; all This included patients with adjacent LNs demonstrating 0% and 100% TE in
grades, Grade $3) were fatigue (42.3%, 1.3%), diarrhea and pruritus (both the same level. Systemic and local immune parameters as they relate to
26.9%, 0%), and nausea (21.8%, 0%). INV grade $3 immune-related AEs concordant and discordant TEs in individual patients will be presented
occurred in 10.3% of pts. One pt died due to an unknown cause that was including a type 1 immune bias. Conclusions: Early histologic evaluation of TE
assessed as treatment-related. Conclusions: Cemiplimab 3 mg/kg Q2W showed in patients with HNSCC receiving immunotherapy demonstrate a wide variety
substantial antitumor activity, durable responses, and acceptable safety profile of response between the primary tumor and LNs. Further investigations will
in pts with laCSCC. These data strongly support the recent FDA approval of lend insight into complex interactions of cancer cells with the microenvi-
cemiplimab-rwlc for pts with mCSCC or laCSCC who are not candidates for ronment. Clinical trial information: NCT03238365.
curative surgery or curative radiation. Clinical trial information: NCT02760498.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
346s Head and Neck Cancer

6017 Poster Discussion Session; Displayed in Poster Session (Board #6), 6018 Poster Discussion Session; Displayed in Poster Session (Board #7),
Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Sat, 4:30 PM-6:00 PM Sat, 4:30 PM-6:00 PM
Recombinant humanized anti-PD-1 monoclonal antibody (JS001) in patients Activity and tolerability of BLU-667, a highly potent and selective RET
with refractory/metastatic nasopharyngeal carcinoma: Interim results of an inhibitor, in patients with advanced RET-altered thyroid cancers. First Au-
open-label phase II clinical study. First Author: Fenghua Wang, Sun Yat-sen thor: Matthew H. Taylor, Oregon Health & Science University, Portland, OR
University Cancer Center, Guangzhou, China
Background: RET alterations are targetable oncogenic drivers in ~90% of
Background: Metastatic nasopharyngeal cancer (NPC) patients progressed advanced medullary thyroid cancer (MTC) and 20% of papillary thyroid cancer
afterstandardtherapy have limited treatment options. Toripalimab, also known (PTC), yet no selective RET inhibitors are approved. BLU-667 is an in-
as JS001, a humanized IgG4 antibody specific for human PD-1, has been vestigational highly potent and selective RET inhibitor targeting oncogenic
approved for 2nd line treatment of metastatic melanoma in China. Here we RET alterations including those that confer resistance to multikinase inhibitors
report the results from a phase IIstudy in metastatic NPC patients treated with (MKIs). We provide an update on the expanded experience of BLU-667 in RET-
toripalimab.(Clinical trial ID: NCT02915432). Methods: This multi-center, altered thyroid cancer from the registration-enabling ARROW study
open-label, phase II registration study is designed to evaluate the safety and (NCT03037385). Methods: ARROW is a global DE (30-600 mg daily [QD or
efficacy of toripalimab in metastaticNPC patients who have failed systemic BID]) and dose expansion (DX; 400 mg QD) study in pts with advanced solid
treatment. Toripalimabis given at 3 mg/kg IV Q2W until disease progression or tumors. Primary objectives are response rate (ORR; RECIST 1.1) and safety.
intolerable toxicity.Tumor PD-L1 expression, plasma EBV DNA level and other Results: As of 19Dec2018, 60 pts with RET-mutated MTC (M918T [37],
biomarkerswill be correlatedwith clinical response. Results: Enrollment of C634R/S/W [8], V804M [4], other/pending [11]) and 5 pts with RET-fusion+
190chemo-refractory metastatic NPCpatients was completed by Feb 2019 PTC (NCOA4 [3], CCDC6 [2]) received BLU-667 (37 DE, 28 DX). 58% had
from 17 participating centers. The median age was 46 years, with 89.5% prior MKI therapy. Among 49 response-evaluable MTC pts, ORR is 47% (95%
patients received at least 2 lines of prior systemic therapies. Treatment related CI: 33, 62; 2 complete and 21 partial responses (PR); 4 PR pending con-
adverse events (TRAE)occurred in 92% patients, which were mostly grade 1 or firmation; 25 stable disease; 1 progressive disease). 96% (22/23) of
2.Common TRAE includedanemia, hypothyroidism, AST increased, pro- responding pts continue treatment; 15 with response duration $ 6 months. 2/
teinuria, pyrexia, cough, constipation, ALT increased, hypoalbuminemia and 4 evaluable PTC pts had PR; all 5 enrolled PTC pts continue treatment at 8-
pruritus.Grade 3 or higherTRAEoccurred in 25% patients.By the cut-off date of 11 months. Responses in MTC occur regardless of MKI resistance (prior
Jan 7 2019, among 135 evaluable patients, 3 complete responses, 31 partial cabozantinib/vandetanib: ORR 46% (12/26)) or RET genotype (PR in 2/3
responses and 40 stable diseaseswere observed for an objective response rate evaluable pts with V804M). Disease control rate in MTC pts is 98%. Rapid
(ORR) of 25.2% and a disease control rate of 54.8%. PD-L1 expression results plasma clearance of RET variants and marked reduction in CEA and calcitonin
were obtained from 125 patients and 45.6% (57/125) were PD-L1+.PD-L1+ is observed. Treatment-related toxicity in MTC/PTC pts, generally low-grade
patientsachieved slightly higherORR than PD-L1- patients, 29.8% versus and reversible (28% had grade 3 events, no grade 4/5 events, no events re-
22.1%. In addition, an average drop of 47-fold plasma EBV DNA copy number quiring discontinuation), includes decreased WBC (23%), increased AST
was observed in responding patients, which typically proceeded the radio- (17%), increased ALT, blood creatinine, and phosphate, hypertension, and
graphic identification of clinical benefit. Conclusions: Toripalimab has decreased neutrophils (all 15%). Conclusions: BLU-667 demonstrates po-
demonstrated a manageable safety profileand encouraging clinical activity in tent, durable and broad antitumor activity and is well tolerated in MTC/PTC pts
the largest check-point blockade study in NPC to date. A change in plasma EBV regardless of MKI resistance and may significantly improve outcomes for pts
DNA copy number might serve as a predictive marker for favorable clinical with RET-altered thyroid cancers. Enrollment of the expansion is ongoing with
response. Patients will be continuously monitored for additional safety and registrational intent. Clinical trial information: NCT03037385.
survival readouts. Clinical trial information: NCT02915432.

6019 Poster Discussion Session; Displayed in Poster Session (Board #8), 6020 Poster Discussion Session; Displayed in Poster Session (Board #9),
Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Sat, 4:30 PM-6:00 PM Sat, 4:30 PM-6:00 PM
Anlotinib treatment in locally advanced or metastatic medullary thyroid carci- Alliance A091404: A phase II study of enzalutamide (NSC# 766085) for
noma: A multicenter, randomized, double-blind, placebo-controlled phase IIB patients with androgen receptor-positive salivary cancers. First Author:
trial. First Author: Dapeng Li, Tianjin Medical University Cancer Institute and Alan Loh Ho, Memorial Sloan Kettering Cancer Center, New York, NY
Hospital, Tianjin, China
Background: A subset of salivary gland cancer (SGCs) express the androgen
Background: Anlotinib (AL3818) is a novel multi-target TKI, inhibiting tumor receptor (AR). This phase II trial evaluated the anti-androgen enzalumatide
angiogenesis and proliferative signaling. Our previous single-arm phase 2 (Astellas) for patients with AR+ SGCs. Methods: Locally advanced/unresectable
ALTN/MTC trial (NCT01874873) has demonstrated that anlotinib has a or metastatic AR+ SGCs were enrolled (AR status was centrally confirmed). Prior
durable antitumor activity with a manageable adverse event profile in locally therapy with AR-targeted drugs was allowed. Enzalutamide 160 mg orally once
advanced or metastatic medullary thyroid carcinoma (MTC). Here we report daily was given (1 cycle= 28 days). The primary endpoint was confirmed re-
results of the phase IIB trial (ALTER01031, NCT02586350) of anlotinib for sponse (RR) according to RECIST v1.1 within the first 8 cycles. Secondary
locally advanced or metastatic MTC with a larger samples. Methods: Between endpoints were progression-free survival (PFS), overall survival (OS), and safety.
September 2015 and September 2018, 91 patients were enrolled in China. A Simon-Optimal two-stage design was used to detect a 20% RR (vs. 5%)
Eligible patients have diagnosed as phase IV MTC with relapsed and mea- (alpha = 5%; beta = 90%). . 1 response(s) in the first 21 would trigger accrual
surable disease and without antiangiogenetic target therapy. The patients were to 41; . 4 responses would be considered promising. Results: 46 eligible
randomly assigned in a 2:1 ratio to receive anlotinib or a matched placebo patients (pts) were enrolled (40 M, 6 F; median age 65) in 22 months. In the
(12 mg QD from day 1 to 14 of a 21-day cycle). Patients who have been first 21 pts, we initially had 2 confirmed PRs allowing for full study accrual,
diagnosed with disease progression by the Independent Imaging Committee though one was later changed to stable disease. Among the 46 pts, 7 had PR
could be unblinded and crossed to the treatment group if the patient previous as best response, though only 2 were confirmed within the first 8 cycles (4%
treated by placebo. The primary endpoint was progression-free survival (PFS). (95% CI: 0.5-15%). Two other PRs did not count towards the primary endpoint
Results: 91 patients were randomized 62 to anlotinib arm and 29 to placebo due to 1) development beyond 8 cycles (cycles 12-18) and 2) a confirmatory
arm. Until the data cutoff date (1 Feb 2019), median PFS was 20.67 months scan completed ,4 weeks apart. The other 3 pts with unconfirmed PR de-
(95%CI, 14.03-34.63) in anlotinib arm vs 11.07 (95%CI, 5.82-14.32) veloped progression of disease (PD) after the first PR scan. 24 pts had stable
months in placebo arm (HR 0.53, p = 0.0289). The OS data were not suf- disease; 15 pts PD as best response. Among 11 pts previously treated with
ficiently mature for analysis. Considerable improvement in ORR was observed AR-targeted therapy, best responses were 1 confirmed PR, 7 SD, 3 PD. With a
over the two arms (48.39% vs 3.45%, p , 0.0001). The adverse events (AEs) median follow-up of 11.7 months (mo), OS at 12 mo was 66% (95% CI:
were 100% in anlotinib arm and 89.66% in placebo arm. The most common 52-83%), PFS at 12 mo was 24% (95% CI: 14-42%), and median PFS was
AEs in anlotinib arm were hand-foot syndrome, hypertension, hyper- 5.5 mo (95% CI: 3.7-7.3). Conclusions: This is the first prospective trial
triglyceridemia and diarrhea. Conclusion: ALTER01031 met its primary evaluating an antiandrogen alone for AR+ SGCs. The failure to meet the
endpoint of PFS shows that anlotinib treatment is effective and well tolerated. protocol-defined measure of success was due in part to the lack of durability of
The safety profile was consistent and no new adverse events were identified. initial responses. The clinical activity observed suggests the AR-dependence of
These data potentially extend the role of anlotinib monotherapy as a new AR+ SGCs, even among those previously treated with other hormonal thera-
therapy strategy for MTC patients. Clinical trial information: NCT02586350. pies. Support: U10CA180821, U10CA180882; Astellas; https://acknowl-
edgments.alliancefound.org. Clinical trial information: NCT02749903.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Head and Neck Cancer 347s

6021 Poster Discussion Session; Displayed in Poster Session (Board #10), 6022 Poster Discussion Session; Displayed in Poster Session (Board #11),
Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Sat, 4:30 PM-6:00 PM Sat, 4:30 PM-6:00 PM
Longitudinal circulating Epstein–Barr virus DNA response to induction che- Mature results of the LCCC1413 phase II trial of de-intensified chemoradio-
motherapy and chemo-radiotherapy to identify biological phenotypes in EBV- therapy for HPV-associated oropharyngeal squamous cell carcinoma. First
associated nasopharynx of head and neck cancer. First Author: Jia-Wei Lv, Sun Author: Bhishamjit S. Chera, University of North Carolina at Chapel Hill,
Yat-sen University Cancer Center, Guangzhou, China Chapel Hill, NC
Background: Liquid biopsies have the utility for detecting minimal residual Background: To report the mature results from a prospective phase II clinical
disease in several cancers, but its clinical utility for real-time treatment ad- trial of highly de-intensified chemoradiotherapy (CRT) for patients with HPV-
aptation remains limited. We adopted Epstein-barr virus (EBV)-associated associated oropharyngeal squamous cell carcinoma (OPSCC). Methods: The
nasopharyngeal carcinoma (NPC) as a model to investigate its potential. We major inclusion criteria were: 1) T0-T3, N0-N2, M0, 2) p16 positive, and 3)
characterize longitudinal response of circulating EBV DNA to induction minimal/remote smoking history. Treatment was limited to 60 Gy intensity
chemotherapy (IC) and concurrent chemoradiotherapy (CCRT) in locally ad- modulated radiotherapy with concurrent weekly intravenous cisplatin 30 mg/
vanced NPC (LA-NPC), and investigate the association of complete biological m2(second choice was cetuximab). Patients with T0-T2 N0-1 disease did not
response (cBR, undetectable cfEBV DNA) during treatment to prognoses. receive chemotherapy. All patients had a 10 to 12-week post-treatment PET/
Methods: The medical records of 673 LA-NPC cases with serial EBV DNA CT to determine need for planned neck dissection. The primary study endpoint
measurements (pre-treatment, after each IC cycle, post-CCRT) were extracted. was 2 year progression free survival (PFS). Secondary endpoint measures
Cox regression and landmark analysis were used for survival analyses. include 2 year local control (LC), regional control (RC), distant metastasis free
Results: Four distinct phenotypes were identified based on their longitudinal survival (DMFS), cause specific survival (CSS) and overall survival (OS), and
cfEBV DNA response: 1) Early responders (200/673[29.7%]) achieved cBR patient reported symptoms (PRO-CTCAE) and quality of life (EORTC QLQ-C30
post-IC1; 2) Intermediate responders (332/673[49.3%]) included patients & H&N35). Results: 114 patients were enrolled (median f/u of 28.8 months,
with cBR post-IC2-4, and cBR post-CCRT after two IC cycles or following a range 2.6 to 51.4) with 84 having a minimum follow-up of 2 years. Smoking
temporary bounce (detectable reading following initial cBR); 3) Late re- status was as follows: 47% never, 33% # 10 pack years, and 19% . 10 pack
sponders (75/673[11.2%]) achieved cBR only post-CCRT after 3-4 IC cycles; years. Post-treatment PET/CT complete response rate was 93% at the primary
4) Treatment-resistant (66/673[9.8%]) patients demonstrated non-cBR post- site and 80% in the neck. Eleven patients had planned neck dissection with 4
IC+CCRT. These phenotypes were significantly correlated with prognoses, having pathological residual disease. Two year LC, RC, DMFS, PFS, CSS, and
adjusted for pre-treatment EBV DNA load, N-category and chemotherapy OS were the following: 96%, 99%, 91%, 88%, 97%, and 95%. Neither
intensity (AHRDFS = 3.46[2.01–6.25], intermediate; 7.50[4.24–14.77], late; smoking status nor receipt of cetuximab correlated with recurrence. Four
17.33[10.06–33.38], treatment-resistant vs. early responders, Pall , 0.01). patients with recurrent disease had PIK3CA mutations. Thirty four percent of
Interestingly, intermediate and late responders without cBR post-IC2 had patients required a feeding tube (none permanent) for a median of 10.5 weeks.
inferior survival despite more IC (HRDFS = 1.83[1.17–2.84], . 2 vs. .2 IC Mean pre- and 2-year post-treatment EORTC QOL scores were: Global 79/83
cycles, P , 0.01). For treatment-resistant patients, adjuvant chemotherapy (lower worse), Swallowing 8/9 (higher worse), Dry Mouth 14/45 (higher worse),
following IC+CCRT reduced risk of distant metastasis (HRDMFS = 0.42 and Sticky Saliva 9/28 (higher worse). Mean pre- and 2 year post-treatment
[0.24–0.74], P , 0.01). Conclusions: We propose investigate risk-adapted PRO-CTCAE (0 to 4 scale, higher worse) scores were: Swallowing 0.5/0.7 and
chemotherapy de-intensification and intensification strategies based on the Dry mouth 0.4/1.4. There were no $ Grade 3 late adverse events.
four novel phenotypes, which could shape the individualized treatment of LA- Conclusions: Clinical outcomes with a highly de-intensified CRT regimen of
NPC. Our study highlights the feasibility of liquid biopsy for real-time thera- 60 Gy IMRT with concurrent low-dose cisplatin are excellent in patients with
peutic adaptation. HPV-associated OPSCC. Clinical trial information: NCT02281955.

6023 Poster Discussion Session; Displayed in Poster Session (Board #12), 6024 Poster Session (Board #13), Sat, 1:15 PM-4:15 PM
Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Afatinib versus methotrexate as second-line treatment for patients with re-
Sat, 4:30 PM-6:00 PM current and/or metastatic (R/M) head and neck squamous cell carcinoma
NRG-HN003: Phase I and expansion cohort study of adjuvant cisplatin, (HNSCC) progressing on or after platinum-based therapy: LUX-Head & Neck 3
intensity-modulated radiation therapy (IMRT), and MK-3475 (Pembrolizumab) phase III trial. First Author: Ye Guo, Shanghai East Hospital, Tongji University,
in high-risk head and neck squamous cell carcinoma (HNSCC). First Author: Shanghai, China
Julie E. Bauman, University of Arizona Cancer Center, Tucson, AZ
Background: In a previous global phase III trial (LUX-Head & Neck 1),
Background: Pembrolizumab, an anti-PD1 monoclonal antibody, improves survival in second-line (2L) afatinib significantly improved PFS vs methotrexate (MTX)
advanced HNSCC. Patients with pathologic high risk, HPV-negative HNSCC have a high in pts with R/M HNSCC. Here, we compared efficacy/safety of 2L afatinib vs
recurrence rate despite adjuvant cisplatin-IMRT (CRT), the current standard. Immu- MTX in Asian pts with R/M HNSCC. Methods: Pts progressing on/after
nosuppression is induced by HNSCC and CRT, and may be reversible by targeting PD1. platinum therapy were randomized (2:1) to 40 mg/day afatinib (feeding
Methods: We conducted a phase I trial with expansion cohort to determine the rec-
tube or oral) or 40 mg/m2/week iv MTX. Primary endpoint was PFS by in-
ommended phase II schedule (RP2S) for adding fixed-dose pembrolizumab to adjuvant
CRT (NCT02775812). Eligibility: oral cavity, pharynx, or larynx primary; HPV-negative;
dependent review. Secondary endpoints were OS, ORR, and patient-reported
pathologic high risk (positive margin or extranodal extension [ENE]); Zubrod 0-1. outcomes. Results: 340 pts were randomized (afatinib 228, MTX 112).
During phase I, patients enrolled in descending cohorts of 12 (Table). RP2S was Median (range) duration of treatment (Tx) was 3.0 ( , 0.1–35.9) and 1.4 ( ,
declared if # 3 dose-limiting toxicities (DLT) occurred. DLT was defined as $ Grade 3 0.1–8.8) mos, respectively. Afatinib significantly decreased the risk of
non-hematologic adverse event (AE) related to pembrolizumab, immune-related (ir)AE progression or death by 37% compared with MTX (HR 0.63; 95% CI: 0.48,
requiring . 2 weeks of systemic steroids, or unacceptable delay in IMRT. The ex- 0.82 p = 0.0005, median PFS, 2.9 vs 2.6 mos; landmark analysis at 12 and
pansion cohort enrolled 20. Results: From Nov 2016-Oct 2018, 34 eligible patients 24 wks, 58 vs 41%, 21 vs 9%). There was no significant difference in OS
enrolled at 22 NRG institutions. During the first cohort, 1 DLT was observed (Grade 3 (HR 0.88; 95% CI: 0.68, 1.13; median 6.9 vs 6.4 mos). ORR was 28% with
fever). RP2S was declared as Schedule 3 and the expansion cohort triggered. Among all afatinib and 13% with MTX (OR 2.8; 95% CI: 1.5, 5.2, p = 0.016). More pts
34 patients, median age was 60 years (26-83); 68% were male; 74% had Zubrod 1; had clinically relevant improvements in global health status/quality of life
85% had oral cavity; 88% had ENE; 21% had positive margin. During expansion, 3
(GHS/QoL; 40 vs 23%, p , 0.01), swallowing (34 vs 18%, p = 0.01) and
additional patients with DLT were observed: wound infection; diverticulitis; nausea. No
DLT unacceptably delayed IMRT. Twenty-eight of 34 (82%) received $ 5 doses of
pain (34 vs 25%, p = 0.22) with afatinib vs MTX. Post-baseline change in
pembrolizumab; 17 (50%) got all 8 doses. Thirty-one of 32 (97%) DLT-evaluable GHS/QoL score was more favorable with afatinib (p , 0.001). Treatment-
patients received all adjuvant RT; 1 withdrew consent after starting protocol. related adverse events (TRAEs; all/grade $3) were reported in 89/16% and
Conclusions: The RP2S is pembrolizumab 200 mg IV q 3 weeks for 8 doses, starting the 67/23% pts with afatinib and MTX. The most common grade $3 TRAEs were
week before adjuvant CRT. This regimen was safe and feasible in a cooperative group rash/acne (4%), diarrhea (4%), and stomatitis (3%) with afatinib, and
setting. irAE were rare in this population. Clinical trial information: NCT02775812. anemia, leukopenia, and fatigue (all 5%) with MTX. Fatal AEs were reported
in 23 and 11% pts with afatinib and MTX. Two ( hypoglycemia, pneumonitis/
Week of Adjuvant IMRT
lung infiltration) and 4 pts had fatal AEs considered related to Tx with
CRT Maintenance afatinib and MTX. 11% and 17% pts discontinued Tx due to TRAEs.
Loading
Modality -1 1 2 3 4 5 6 9 12 15 18 21 24 27 Conclusions: LUX-Head & Neck 3 achieved its primary endpoint. Two
IMRT (60 Gy, 2 Gy/Fx/day) – all schedules X X X X X X randomized phase III trials have now demonstrated clinical benefit with 2L
Cisplatin 40 mg/m2/week IV – all schedules X X X X X X afatinib vs MTX. Safety data were consistent with the known tolerability
Pembrolizumab 200 mg IV profiles of afatinib and MTX. Clinical trial information: NCT01856478.
Schedule 3 (Starting) X X X X X X X X
Schedule 2 (1st De-escalation) X X X X X X X X
Schedule 1 (2nd De-escalation) X X X X X X X X

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
348s Head and Neck Cancer

6025 Poster Session (Board #14), Sat, 1:15 PM-4:15 PM 6026 Poster Session (Board #15), Sat, 1:15 PM-4:15 PM
CDX3379-04: Phase II evaluation of CDX-3379 in combination with cetuximab Pembrolizumab (pembro) for recurrent head and neck squamous cell carci-
in patients with advanced head and neck squamous cell carcinoma (HNSCC). noma (HNSCC): Post hoc analyses of phase 3 KEYNOTE-040 prior radiation
First Author: Julie E. Bauman, University of Arizona Cancer Center, Tucson, AZ treatment (RT) and disease state. First Author: Kevin J. Harrington, The In-
stitute of Cancer Research/The Royal Marsden NHS Foundation Trust National
Background: CDX-3379, an anti-ErbB3 monoclonal antibody with a half-life-
Institute of Health Research Biomedical Research Centre, London, United
extending YTE modification in its Fc region, binds a unique epitope that locks
Kingdom
ErbB3 in an inactive form and inhibits ErbB3 signaling, the latter implicated in
tumor growth/resistance to anticancer therapies. CDX-3379 enhances anti- Background: The open-label, randomized, phase 3 KEYNOTE-040 study (NCT02252042)
tumor activity of targeted therapies in preclinical models. In a Phase 1 clinical showed that pembro vs standard of care (SOC) chemotherapy prolonged survival in patients
study, CDX-3379 was well-tolerated alone and in combination with cetuximab. (pts) with recurrent and/or metastatic HNSCC whose disease progressed during/after
platinum-based therapy. Post hoc analyses were conducted to evaluate pembro vs SOC
A durable complete response (CR) to CDX-3379 + cetuximab was observed by prior RT and disease state (metastatic, locoregionally recurrent [referred to as recurrent
(8.3 months) in a patient (pt) with cetuximab-refractory HNSCC (Falchook herein], or recurrent and metastatic [R/M]). Methods: Pts (N = 495) were randomly
ASCO 2016). Methods: This open-label phase 2 study (NCT03254927) was assigned (1:1) to receive pembro (200 mg Q3W) or investigator choice of methotrexate
designed to enroll up to 30 pts with advanced, HPV-, HNSCC, previously (40 mg/m2 QW), docetaxel (75 mg/m2 Q3W), or cetuximab (400 mg/m2 loading dose, then
treated with cisplatin, anti-PD-1 antibodies, and cetuximab-resistant (pro- 250 mg/m2 QW). Primary end point was OS; PFS and ORR were secondary end points.
gression within 6 months), according to a Simon’s 2-stage design (13 Results: 175, 97, and 195 pts had metastatic, recurrent, and R/M disease, respectively (28
evaluable pts in Stage 1 with $1 objective response allows enrollment of 14 pts had unknown disease state); 64 pts had no prior RT; 431 pts had prior RT. As in the ITT
population, prolonged survival benefit and trend toward improved PFS and ORR was
more pts in 2nd stage). Pts receive CDX-3379 (initial dose 12 mg/kg IV every
observed with pembro vs SOC in pts with prior RT (Table), and prolonged survival benefit
21 days) + cetuximab (loading dose 400 mg/m2; 250 mg/m2 IV weekly) until was observed with pembro vs SOC in pts with metastatic and R/M, but not recurrent,
disease progression/toxicity. Endpoints include objective response rate (pri- disease. Conclusions: In this post hoc analysis, patients with prior RT benefited from
mary), progression-free and overall survival, safety, pharmacokinetics, im- treatment with pembro vs SOC. For patients without RT, sample sizes are too small to draw
munogenicity, and exploratory biomarkers. Results: Stage 1 accrual is any definitive conclusions. Survival benefit of pembro vs SOC was observed in pts with
complete with 14 evaluable pts treated. All pts were heavily pretreated; prior metastatic and R/M disease. Clinical trial information: NCT02252042.
therapies included surgery (10/14) and chemotherapy (13/14). All pts had Metastatic Pembro Metastatic SOC Recurrent Pembro Recurrent SOC R/M Pembro R/M SOC
prior radiation, cetuximab and PD-1 targeted therapy. One confirmed ongoing n = 90 n = 85 n = 43 n = 54 n = 104 n = 91

CR (8.1+ months) was observed. 7/14 pts experienced stable disease (SD), OS: Median, mo
OS: pembro vs SOC
8.9
0.72
7.9
2
7.2
1.07
9.1
2
6.7
0.79
5.6
2
including 4 with tumor shrinkage (8-27.5% reduction). Three pts continue HR
(95% CI)
(0.52-1.03)
0.038
(0.67-1.69)
0.607
(0.58-1.07)
0.065
treatment. Treatment-related adverse events were generally grade 1-2 and P

included diarrhea (53%), hypokalemia (20%), prolonged QT interval (13%) Prior RT Prior RT No Prior RT No Prior RT
Pembro SOC Pembro SOC
and rash (13%). Conclusions: CDX-3379 in combination with cetuximab is n = 217 n = 214 n = 30 n = 34
OS: Median, mo 8.4 6.6 7.2 9.4
well tolerated with the primary toxicity of diarrhea. Signs of antitumor activity OS: pembro vs SOC 0.75 2 1.01 2
were observed in these cetuximab-resistant HNSCC pts, including an ongoing, HR
(95% CI)
(0.60-0.93)
0.004
(0.59-1.74)
0.520
durable CR. Complete stage 1 results will be presented. Clinical trial in- P
PFS: Median, mo
2.1 2.2 2.1 3.9

formation: NCT03254927. PFS: pembro vs SOC

HR
0.87
(0.71-1.07)
2 1.32
(0.80-2.18)
2
2
(95% CI) 0.086 0.854
P 15.7 11.2 6.7 2.9
ORR: %
ORR: pembro vs SOC 4.5 2 3.7 2
Difference (22.0-11.0) (29.3-19.0)
(95% CI) 0.088 0.243
P

6027 Poster Session (Board #16), Sat, 1:15 PM-4:15 PM 6028 Poster Session (Board #17), Sat, 1:15 PM-4:15 PM
Induction chemotherapy with docetaxel, cisplatin and cetuximab versus Safety and antitumor activity of accelerator-based boron neutron capture
docetaxel, cisplatin and 5-fluorouracil followed by radiotherapy with cetux- therapy in patients with inoperable recurrent and locally advanced head and
imab for locally advanced or inoperable squamous cell carcinoma of the head neck cancer: A phase II study. First Author: Katsumi Hirose, Southern
and neck: Promising results of a randomized phase II AGMT-trial. First Author: Tohoku BNCT Research Center, Koriyama, Japan
Felix Keil, Medical Dept. Int. Med. 3, Hematology and Oncology, Hanusch-
Background: To assess safety and efficacy of accelerator-based boron neutron
Krankenhaus, Vienna, Austria
capture therapy (AB-BNCT) using cyclotron-based neutron generator,
Background: Induction chemotherapy (ICT) with Cisplatin (P), 5-FU (F) and Taxanes BNCT30, and 10B-boronophenylalanine (borofalan(10B)) agent, SPM-011, in
(T) is a therapeutical option in patients suffering from locally advanced or unre- patients with recurrent squamous cell carcinoma (R-HNSCC) or recurrent/
sectable stage III or IV squamous cell carcinoma of the head and neck (SCCHN). The locally advanced non-squamous cell carcinoma (R/LA-HNNSCC) of the head
role of ICT is controversial and toxicity and/or delay of radiotherapy may reduce the and neck. Methods: The multi-institutional open-label, a world-first phase II
potential benefit of this treatment regimen. Here we report promising results of a trial of AB-BNCT in patients with inoperable R-HNSCC which present re-
randomized phase II trial comparing TPF with TP and Cetuximab (C) replacing F. sistance to platinum-based chemotherapy, or with inoperable R/LA-HNNSCC,
Methods: In our trial, N= 100 patients with locally advanced or unresectable stage III was conducted to assess safety and antitumor activity of AB-BNCT with
or IV SCCHN were randomly assigned to either Arm A (N= 49), receiving TPF, or Arm
BNCT30 and SPM-011. SPM-011 was administered at 200 mg/kg/h in-
B (N= 51), receiving TPC, both followed by radiotherapy (RT) + C. The primary end-
point of the study was overall response rate (ORR) three months after RT + C was
travenously for 2 hours, followed by neutron irradiation with continuous in-
finished. Results: We observed a remarkable response rate (CR + PR) of 86.4% in fusion of SPM-011 at 100 mg/kg/h. The irradiated dose for tumor was
the TPC-arm that compared favorably with 77.5% responding patients in the TPF- determined passively as a mucosal maximum dose was given 12 Gy-Eq in
arm three months after RT + C was completed. OS and PFS were similar in both arms. calculation with a blood boron concentration measured just before the start of
After 400 days we observed an OS rate of 79% in the TPF and 86% in the TPC arm, neutron irradiation. Primary endpoint was objective response rate (ORR) by
and a PFS rate of 67% in the TPF and 70% in the TPC arm. TPC containing ICT led to central review. Tumor response was assessed using Response Evaluation
less serious adverse events (SAEs), including blood and lymphatic disorders (40.8% Criteria in Solid Tumors (RECIST) ver 1.1 every 4 weeks for the first 3 months
in TPF arm, 27.5% in TPC arm) and metabolism and nutrition disorders (22.4% in and every 12 weeks thereafter. Results: Eight R-HNSCC and thirteen R/LA-
TPF arm, 9.8% in TPC arm) during ICT. Interestingly, in HPVp16 positive patients, HNNSCC patients were enrolled and received AB-BNCT. All R-HNSCC pa-
88.24% in the TPF-arm and 93.33% in the TPC-arm showed CR or PR three months tients had prior radiotherapy with a median dose of 65.5 Gy (range
after RT + C, whereas only 69.57% in the TPF-arm and 82.76% in the TPC-arm 59.4–76.0). The median irradiation time was 43 min (range 26–65). The
showed CR or PR. We only lost one patient because of treatment-related mortality median tumor minimum dose was 31.0 Gy-Eq (range 16.1–42.6). For adverse
(TRM) and no delay from the end of ICT to local radiotherapy was observed in any event, nausea (81%), dysgeusia (71%), parotitis (67%) were observed more
patient. All patients received RT + C within three weeks after ICT was completed. frequently. The ORR for all patients were 71.4%, and CR/PR were 50.0%/
Conclusions: In conclusion, TPC is a feasible and tolerable therapy regimen and can 25.0% in R-HNSCC and 7.7%/61.5% in R/LA-HNNSCC. At a median follow
be applied within one day with less hematological toxicities. In contrast, more local
up of 18.8 months (range 9.2–29.0), 1-year PFS and OS by investigator
reactions were observed after TPC. TPC containing ICT leads to improved response
rates, while OS and PFS were similar in both arms. TRM was extremely low with 1%. assessment were 70.6% and 100%, respectively. The data for antitumor
Therefore, we conclude, that TPC containing ICT could be a considerable ther- activity is still immature and will be further updated. Conclusions: AB-BNCT
apeutical alternative for patients with locally advanced or unresectable stage III or IV for R-HNSCC and R/LA-HNNSCC demonstrated an acceptable safety profile
SCCHN, who are eligible for ICT. Clinical trial information: 2011-005540-99. and a promising antitumor activity.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Head and Neck Cancer 349s

6029 Poster Session (Board #18), Sat, 1:15 PM-4:15 PM 6030 Poster Session (Board #19), Sat, 1:15 PM-4:15 PM
Hyperprogressive disease (HPD) in head and neck squamous cell carcinoma Apatinib for locoregionally recurrent or metastatic nasopharyngeal carcinoma
(HNSCC) patients treated with immune checkpoint inhibitors (ICI). First after failure of first-line chemotherapy: A multicenter, phase II trial. First
Author: Salvatore Alfieri, Head and Neck Cancer Medical Oncology 3 Unit, Author: Wei Jiang, Affiliated Hospital of Guilin Medical University, Guilin,
Fondazione IRCCS Istituto Nazionale dei Tumori, Milan, Italy China
Background: HPD was described in 9% of cancer patients (pts) treated in phase I Background: Concordant programs for patients with nasopharyngeal carci-
trials, in 13.8% of advanced non-small cell lung cancer and 29% of 34 HNSCC pts noma (NPC) who failed to first-line chemotherapy after locoregional recurrence
upon ICI. A better definition of the hallmarks and survival outcomes of HPD pts in a or metastasis are not yet available. Here, we investigated the efficacy and
larger cohort of HNSCC is still lacking. Methods: We retrospectively analyzed all safety of apatinib as an second-line treatment in these patients. Methods: In
advanced HNSCC pts treated with ICI at our Institution between October 2014 and this multicenter, phase II trial, patients of NPC with disease progression after
December 2018. Three scans, performed before ICI, at baseline and at first evaluation failure of first-line chemotherapy were treated with apatinib (500mg/d).The
during ICI, were assessed according to RECIST 1.1. Tumor Growth Kinetics (TGK) primary endpoint of this study was objective response rate (ORR), secondary
pre- (TGKpre) and post-baseline (TGKpost) were measured as previously reported
endpoints included progression free survival (PFS), overall survival (OS) and
(Saâda-Bouzid E, Ann Oncol 2017). Pts were defined HPD if they had progression at
first radiological evaluation and TGKpost/TGKpre $2. Correlation between HPD and
toxicity. Results: Between January and December 2017, 33 patients were
clinical characteristics was performed by Fisher or t-student test. Median overall finally enrolled onto the analysis from three centers in China. The baseline
survival (mOS) and progression free survival (mPFS) were estimated using the Kaplan- characteristics were summarized in Table. Of the 12 patients achieved a partial
Meier method and compared between HPD and non-HPD using the log-rank test. response and no complete responses were observed, yielding an ORR of
Results: Ninety pts were eligible: 18% were female, 4% had ECOG PS $ 2, 73% 36.3%. Additionally, 6 patients (18.2%) experienced stable disease of at least
smoking history, 37% oropharyngeal cancer (61% HPV+), 65% locoregional disease 5 months in duration, and the disease control rate was 54.5%. At a median
(89% previously irradiated), 54% received combined immunotherapy, 75% in $ 2nd follow-up time of 14 months (range 1-22), median PFS was 5.0 months (95%
line. Two out of 90 pts had TGKpre = 0 and were not evaluable for TGK ratio. HPD was CI, 2.3 to 7.7). The median OS had not reached, and the 1-year OS rate was
observed in 7.9% (7/88) of pts. HPD pts were significantly younger compared to non- 83.1%. The most common adverse events (grade 1 to 2) related to apatinib
HPD pts (median age 53 6 3.7 vs 63.3 6 0.9 years, p = 0.002) and had a sig- were hypetension (42.4%), hand-foot syndrome (54.5%), proteinuria (12.1%)
nificantly higher median neutrophil-lymphocyte ratio (NLR) (11.5 6 3.5 vs 6.4 6 0.4, and oral ulcer (24.2%). Conclusions: Apatinib showed a well therapeutic
p = 0.004). Overall, mOS and mPFS were 7.5 (95% CI: 4.2-10.8) and 2.2 months effect and a manageable safety profile for patients of advanced NPC after
(95% CI: 0.9-3.4), respectively. At a median follow-up of 20.9 months (95% CI: 19- previous chemotherapy. Further study in combination with chemotherapy and
22.8), HPD pts had a significantly worse mPFS compared to non-HPD pts [1.8 (95% other targeted agents in patients with NPC is warranted. Baseline demographic
CI: 1.5-2.2) vs 3.5 (95% CI: 2.2-4.8) months; p = 0.001]. HPD correlated with a not and disease characteristics. Clinical trial information: NCT03130270.
significant trend in lower mOS compared to non-HPD group [3.7 (95% CI: 2.4-5.1) vs
8.3 (95% CI: 4.1-12.5) months; p = 0.348]. Three (43%) out of 7 HPD pts early Apatinib
switched to chemotherapy after PD to ICI having a mOS of 8.1 months (range 3.7- contents (n=33)
25.3). Excluding these 3 pts, HPD correlated with a significantly worse mOS Age in years, median (range) 48(23-70)
compared to non-HPD [2.6 (95% CI: 1.9-3.3) vs 8.3 (95% CI: 4.1-12.5) months; p = Male/female, n (%) 28(84.8%)/5(15.2%)
NPC histology, n (%)
0.006]. Conclusions: HPD was identified in 7.9% of HNSCC and correlated with Non-keratinization Differentiated/Undifferentiatied 3(9.1%)/30(90.9%)
younger age and higher NLR. HPD pts who did not receive a subsequent treatment Recurrent or metastatic sites, n (%)
had poorer mPFS and mOS. The assessment of HPD in a control cohort of advanced Nasopharynx/Regional lymph nodes 5(15.2%)/6(21.2%)
Lung/Bone/Liver 13(39.4%)/6(18.2%)/10(30.3%)
HNSCC upon standard chemotherapy is ongoing. Previous chemotherapy, n (%)
Platinum/ Gemcitabine/Docetaxel/Fluorouracil 33(100%)14(42.4%)/23(69.7%)/17(51.5%)

6031 Poster Session (Board #20), Sat, 1:15 PM-4:15 PM 6032 Poster Session (Board #21), Sat, 1:15 PM-4:15 PM
Preclinical efficacy of copanlisib in cetuximab sensitive and resistant tumors A safety study of nivolumab in patients with recurrent and/or metastatic
of HNSCC. First Author: Konrad Friedrich Klinghammer, Charite Compre- platinum-refractory squamous cell carcinoma of the head and neck (R/M
hensive Cancer Center, Berlin, Germany SCCHN): Interim analysis on 199 patients—The TOPNIVO study on behalf of
the GORTEC and the Unicancer Head & Neck Group. First Author: Caroline
Background: Copanlisib is a highly selective, pan-class I PI3K inhibitor with
Even, Gustave Roussy, Villejuif, France
preferential activity against the p110a and p110d isoforms that lead to
downregulation of PI3K signaling. Copanlisib has been approved for the Background: In the randomized phase III Study CA209141, Nivolumab (N)
treatment of follicular lymphoma in the US. Here, we explored the anti-tumor demonstrated significant overall survival (OS) benefit with favorable safety
activity of copanlisib in head and neck squamous cell carcinoma (HNSCC), profile for platinum refractory R/M SCCHN and is now approved for these
where Pi3K signaling has been defined as alternate signaling in cetuximab patients (pts). The objectives of the study are to provide additional insight into
resistant tumors. Further, TCGA data show up to 56% of HNSCC display either the frequency of high-grade AEs related to N and the efficacy of N in real life.
amplification or mutational changes in the Pi3K pathway making Pi3K an Methods: Between August and December 2017, 203 pts were included in the
attractive target. Methods: Using a mouse-clinical trial set-up we profiled 20 multicenter, non-controlled phase II TOPNIVO. The main inclusion criteria
patient derived HNSCC xenograft models for their sensitivity to cetuximab or were patients with platinum refractory R/M SCCHN with progressive disease,
copanlisib as single agent as well as in combination. Models were selected ECOG 0-2. Pts received N 3mg/kg every 2 weeks intravenously over 30
from our HNSCC PDX platform based on Pi3K mutational status, with 6 models minutes. Four pts did not receive N. We report here the safety during the first
harboring hot spot mutations, HPV positivity (n=3) and/or cetuximab re- 6 months (mo) after inclusion and OS results on the first 199 treated pts.
sistance based on previous drug screenings (n=12). Treatment response was Results: Median age was 62 yr, 83% were male, 84% were ECOG 0-1, 16% 2.
defined as tumor regression, stabilization or progression expressed as relative The primary site of cancer was oral cavity 26%, oropharynx 38%, larynx 16%,
tumor volume (RTV) after 3 weeks of treatment: RTV,0,7 responder, hypopharynx 21%. 33% had loco regional relapse, 32% metastatic disease
RTV.1,2 progressor. Results: Copanlisib single agent treatment resulted in and 35% both. 49% had received one prior line of chemotherapy and 30% two
moderate activity with 5 responders (25%). In cetuximab resistant tumors prior lines. 157 (79%) pts ended their treatment within the first six mo: 5 for AE
(n=12) combined treatment led to an improved tumor response in 75% (n=9) related to N (pneumonitis 3 pts, hepatitis 1 pt, diarrhea 1 pt), 107 for pro-
whereas 41% (n=5) resulted in tumor control. Pi3KCA mutation was not gression, 33 for death (24 related to progression, 9 to intercurrent disease), 12
predictive for treatment response to either cetuximab or copanlisib. No PTEN other. 132 pts (66%) experienced at least 1 AE grade $3. On the 226 AEs
mutation was detected in the selected cohort. Increased Pi3K signaling activity grade 3-4, 21 (mainly pneumopathy, lipase increase and asthenia) were re-
evaluated through gene expression profiling and computed with GSEA pathway lated to N and occurred in 18 pts. On the 51 AEs grade 5, 3 were considered
analyses was positively correlated with response. Conclusions: The anti-tumor related to N (2 pneumonitis, 1 cardiac arrest). The median OS was 7.7 mo (CI
responses observed in monotherapy or in combination treatment support 95% [6.0; 9.5]) in the whole population; 9.2 mo [6.8; 12.1] in the 167 pts
further investigation for the potential of PI3K inhibition in HNSCC with high with ECOG 0-1, 3.0 mo [1.1; 6.0] in the 32 pts with ECOG 2; 12.1 mo [7.6;
expression of Pi3K pathway signature as a potential predictive biomarker. NR] in the 64 pts with metastatic disease, 7.7 mo [5.0; 9.6] in the 66 pts with
locoregional disease and 4.6 mo [3.1; 7.9] in the 69 pts with both. OS was
similar in pts older or younger 70 yr. Conclusions: The interim analysis of the
TOPNIVO study shows no additional toxicities of N compared to what has been
described previously, confirms the previous results of OS and provides new
survival data in subgroups of pts. Clinical trial information: NCT03226756.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
350s Head and Neck Cancer

6033 Poster Session (Board #22), Sat, 1:15 PM-4:15 PM 6034 Poster Session (Board #23), Sat, 1:15 PM-4:15 PM
An open label, nonrandomized, multi-arm, phase II trial evaluating pem- Clinical implications of hyperprogression with immune checkpoint inhibitors
brolizumab combined with cetuximab in patients with recurrent/metastatic in patients with head and neck squamous cell carcinoma (HNSCC). First
(R/M) head and neck squamous cell carcinoma (HNSCC): Results of cohort 1 Author: Panagiota Economopoulou, Attikon University Hospital, Athens,
interim analysis. First Author: Assuntina Gesualda Sacco, University of Greece
California San Diego Moores Cancer Center, San Diego, CA
Background: Hyperprogressive disease (HPD) refers to paradoxical accelera-
Background: Pembrolizumab (a humanized monoclonal antibody blocking tion of tumor growth kinetics (TGK) after initiation of treatment with anti- PD-1/
programmed death receptor-1[PD-1]), and cetuximab (a chimeric monoclonal PD- L1 agents and has been reported across tumor types in 4-29% of patients
antibody inhibiting epidermal growth factor receptor) are both FDA-approved, using different definitions. Preliminary data suggest that HPD might affect
second-line monotherapies for R/M HNSCC. This is the first trial to evaluate response to subsequent therapies. Methods: We compared TGK prior and TGK
anti-tumor efficacy of dual therapy with pembrolizumab and cetuximab. upon immunotherapy (IO) in 62 patients (pts) with recurrent/metastatic (R/M)
Previously reported safety data demonstrated favorable toxicity. An interim HNSCC treated with PD-1/PD-L1 inhibitors. The TGK ratio (TGKR, ratio of
futility analysis of cohort 1 (anti-PD-1/PD-L1 and cetuximab naı̈ve) was tumor growth velocity before and upon treatment) was calculated. The first
completed per protocol. Methods: Patients (pts) with platinum-refractory/ imaging assessment was performed 3 months (mo) after IO initiation. HPD was
ineligible, R/M HNSCC were treated with pembrolizumab 200mg IV on day defined as 1. Radiological HPD (TGKR$2) or 2. Clinical HPD (Disease-related
1 and cetuximab 400mg/m2 loading dose followed by 250mg/m2 weekly (21- rapid clinical deterioration post IO). Results: After median follow-up of 12.3
day cycle). Primary endpoint: overall response rate (complete and partial re- mo (range, 0.4-28.1), 43 pts progressed and 38 died. Median PFS was 2.8 mo
sponses) by 6 months (mo). Secondary endpoints: 12-mo progression-free (95%CI, 2.2-3.4) and median OS 8.6 mo (95%CI, 4.2-12.9). HPD was
survival (PFS) probability, overall survival, response duration, safety, correl- observed in 16 pts (25.8%), while 15 pts had early PD (Time to Treatment
ative analyses. Results: 14 evaluable pts were enrolled March 2017-October failure, TTF , 3 mo) and 31 late PD (TTF . 3mo). Among 16 pts with HPD, 11
2018. Median age 60y (range 47-86y), M:F 6:8, ECOG (0:1) 2:12, 14 had radiological HPD and 10 had clinical HPD. 4 pts had both clinical and
mucosal primaries (9 oral cavity, 2 HPV-mediated oropharynx, 2 non-EBV- radiological HPD. Pts with late PD had median OS 11.3 mo (95%CI, 9.3-
associated nasopharynx, 1 larynx). 11 pts (79%) had no prior lines of systemic 13.3), those with early PD 5.2 mo (95%CI, 3.1-7.3 months) and those with
therapy for R/M HNSCC (range 0-1). 6 pts (42.8%) had a partial response by HPD 5.1 mo (95%CI, 4.4-5.9) (p , 0.005). Regarding post-progression OS,
6 months, meeting pre-planned criteria for trial continuation. 4 pts (28.6%) pts with late PD had median 11.3 mo (95%CI 0-22.8), those with early PD 2.5
had stable disease and 4 (28.6%) had progressive disease. Median PFS was mo (95%CI 0.6-4.4) and those with HPD 4.2 mo (95%CI 1.7-6.7) (p =
128 days (4.3 mo). Median duration of response was 160.5 days (5.4 mo) for 0.001). Pts with HPD had a trend for longer median post-progression OS
partial responders and 133 days (4.4 mo) for pts with stable disease. Disease compared to pts with early PD (p = 0.121). Median PFS with chemotherapy
control rate (partial + stable) was 71.4%. There were 7 grade 3 treatment- after immunotherapy failure was 3.0 mo (95%CI 2.4-3.6) for pts with late PD,
related toxicities. 2 pts discontinued cetuximab due to toxicity, however, both 2.1 mo (95%CI 0.9-3.4) for pts with early PD and 6.1 mo (95%CI 3.0-9.3) for
continued pembrolizumab. Conclusions: Interim analysis indicates that those with HPD (p = 0.040). HPD was associated with longer median PFS with
pembrolizumab plus cetuximab is potentially active for platinum-refractory/ chemotherapy compared to pts with early PD (p = 0.016), while the difference
ineligible pts with R/M HNSCC. These results meet protocol specifications in median PFS with chemotherapy between pts with HPD and late progressors
for trial continuation. Final results will include PD-L1 expression data. Clinical was non-statistically significant (p = 0.260). Conclusions: Radiological or
trial information: NCT03082534. clinical HPD was observed in 25.8% of patients with R/M HNSCC treated
with IO. Early progression to immunotherapy is an important predictor of
short survival, while HPD was associated with improved PFS to subsequent
chemotherapy.

6035 Poster Session (Board #24), Sat, 1:15 PM-4:15 PM 6036 Poster Session (Board #25), Sat, 1:15 PM-4:15 PM
Efficacy and safety of immune checkpoint inhibitors in elderly patients ($70 A multicenter prospective observational study of nutritional status on survival
years) with squamous cell carcinoma of the head and neck. First Author: in locally advanced nasopharynx cancer treated by induction chemotherapy
Caroline Even, Gustave Roussy, Villejuif, France and chemoradiotherapy. First Author: Jingjing Miao, Department of Naso-
pharyngeal Carcinoma, Sun Yat-sen University Cancer Center, State Key
Background: Recent meta-analysis showed that immune checkpoint in-
Laboratory of Oncology in South China, Collaborative Innovation Center of
hibitors (ICI) have comparable activity in younger vs older patients (pts) ($65
Cancer Medicine, Guangzhou, China
years (y)). However little is known about efficacy and safety of ICI in elderly pts
with relapsed/metastatic squamous cell carcinoma of the head and neck (R/M Background: We conducted a multicenter prospective study (NCT02575547) to
SCCHN). The aim of this study is to compare efficacy and grade $3 immune- investigate the association between longitudinal nutritional status and survival in
related adverse events (irAEs) of ICI in pts $70 y with R/M SCCHN to younger locally advanced nasopharynx cancer (LA-NPC) treated with induction chemotherapy
pts. Methods: A retrospective study was conducted at 4 French hospitals. and concurrent chemoradiotherapy (IC-CCRT). Methods: All patients with biopsy-
Eligibility criteria were pts treated with ICI for R/M SCCHN between September proven LA-NPC and planned for IC-CCRT were recruited from ten institutions. IC
entailed 2 cycles of docetaxel 75mg/m2/3w and cisplatin 75mg/m2/3w; CCRT
2014 and December 2018. Clinical and radiological data and outcome were
entailed 2-3 cycles cisplatin 100mg/m2/3w and IMRT (70-72Gy/30-32fr). Study
collected from review of medical records. Results: Two hundred twenty six pts parameters included weight loss (WL), % of ideal body weight (%IBW), body mass
were enrolled including 67 pts $ 70 y. Median age of elderly pts was 75y index (BMI), nutrition risk screening 2002 (NRS2002), patient-generated subjective
(range 70-87). Elderly pts received ICI as first-line treatment in 21% of pts vs global assessment (PG-SGA), and EORTC QLQ-C30 that were collected at the fol-
17% in younger pts. In elderly pts, 9% had ECOG of 0, 72% had ECOG of 1 and lowing time-points: baseline (T1), 1 w pre-2nd IC (T2), 1 w pre-CCRT (T3), 4 w mid-
15% had ECOG of 2 at ICI initiation vs 34%, 62% and 4% respectively in CCRT (T4), end-CCRT (T5), 3 mo post-CCRT (T6), 1 y post-CCRT (T7), 2 y post-CCRT
younger pts (p = 0.0006). In elderly pts, 22% had only loco-regional relapse at (T8). Results: 186 patients were recruited; 171 were eligible for analysis. Median
ICI initiation, 30% only distant recurrence and 49% had both vs 42%, 32% follow-up was 35.8 mo (range 12.3-46.1 mo). Compliance rates were 97.7% (167/
and 26% respectively (p = 0.0014). Elderly pts received ICI as monotherapy in 171) and 87.7% (150/171) for IC and CCRT, respectively; all except one completed
73% of pts vs 52% (p = 0.0027). The ORR in elderly pts was 23% vs 13% in RT. Longitudinal assessment indicated the worst nutritional status at T5, followed by
younger pts (p = 0.071). After a median follow-up of 16.8 months (m) (range recovery at T8: 27.1% with %IBW ,90%; 69.2% with WL $5%, which was also
10.7-23.7), median OS was 9.7m in elderly pts vs 8.7m in younger pts (p = associated with a worsened QOL (OR = 6.23 for QOL change $25.0, P = 0.012).
0.87). Median PFS was 2.7 m in elderly pts vs 1.9 m (p = 0.2). After ad- Interestingly, T1 nutritional status was not associated with prognosis (P .0.05 for all).
However, nutritional parameters at T5 were significantly associated with survival;
justment for ECOG, type of evolution, number of ICI drugs, time between initial
%IBW ,90% was the strongest predictor for inferior DMFS (HR = 2.669) and OS (HR
diagnosis and ICI start and number of previous lines, age $70 years was
= 4.661) among all parameters (multivariable-adjusted P ,0.05). Subgroup analyses
significantly associated with a better PFS (HR = 0.66 (95%CI = 0.47;0.93), revealed that %IBW ,90%, WL $10% at T5 represented the most adverse subset of
p = 0.02) but was not significantly associated with OS (HR = 0.91 (95%CI = patients (Table). Conclusions: Here, we show that poor nutrition despite systemic
0.61;1.34), p = 0.62). Grade $3 irAEs occurred in 15% of elderly pts vs 8% of intensification leads to inferior QOL and disease control in LA-NPC patients. This is
younger pts (p = 0.13). Patients with grade $3 irAEs had a significantly higher counter-intuitive and highlights the crucial importance of paracrine factors in opti-
ORR than pts without Grade $3 irAEs (36% vs 14%, p = 0.007). mising treatment efficacy. Clinical trial information: NCT02575547.
Conclusions: Elderly pts treated with ICI had significantly higher PFS but not Subgroup analysis.
OS after adjustment. Grade $3 irAEs were associated with significantly higher
3-y LRRFS 3-y DMFS 3-y DFS 3-y OS
ORR to ICI in the whole population.
%IBW‡90% & WL<10% 95.1% 88.3% 87.1% 96.4%
%IBW‡90% & WL‡10% 91.6% 97.4% 88.9% 100.0%
%IBW<90% & WL<10% 92.9% 89.3% 85.7% 92.9%
%IBW<90% & WL‡10% 100.0% 58.8% 58.8% 81.9%

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Head and Neck Cancer 351s

6037 Poster Session (Board #26), Sat, 1:15 PM-4:15 PM 6038 Poster Session (Board #27), Sat, 1:15 PM-4:15 PM
Metformin treatment of locally advanced head and neck squamous cell carci- High-dose, short-duration, intra-arterial cisplatin therapy for oral cancer. First
noma (LAHNSCC) patients induces an anti-tumorigenic immune response. First Author: Masatoshi Ohmae, Rinku General Medical Center, Izumisano, Japan
Author: Vidhya Karivedu, University of Cincinnati, Cincinnati, OH
Background: We developed a superselective intra-arterial chemotherapy (iaCT)
Background: Metformin is a biguanide, widely used oral hypoglycemic agent. approach for oral cancer wherein an intra-arterial catheter is retrogradely
Metformin has also shown to inhibit tumor growth and progression in a wide inserted via the superficial temporal artery (STA) and/or occipital artery (OA). In
variety of cancers including Head and Neck Squamous Cell Carcinoma most iaCT cases, high-dose anticancer agents are administrated via the intra-
(HNSCC). Metformin activates AMP protein kinase (AMPK) related pathways arterial catheter weekly or biweekly with daily irradiation. This approach re-
leading to inactivation of mammalian target of rapamycin (mTOR) and sup- markably improves curative efficiency, but some adverse events, e.g., severe
pression of its downstream effectors. In addition, metformin is postulated to mucositis, dysphagia, dysgeusia, dry mouth, and radiation osteonecrosis,
alter immune regulation in the tumor microenvironment leading to increased mainly because of irradiation, decrease the quality of life. Methods: Thirty-two
tumor cell killing. Here, we report our findings on the impact of metformin on patients with stage II, III, or IV oral squamous cell carcinoma were treated using
T cells, NK cells and cytokines from patient peripheral blood mononuclear this new iaCT approach. The catheter was superselectively placed in the tumor-
cells (PBMCs) from a phase I open-label single site dose escalation study feeding arteries by cut-down of STA or OA. The catheter was completely placed
combining metformin and chemoradiation (CRT) in HNSCC (NCT02325401). under the skin and was connected to an infusion reservoir that was sub-
Methods: In this study, we evaluated the immune cell phenotypes and cytokine cutaneously implanted around the mastoid process via the subcutaneous
profiles of peripheral blood in patients before and after metformin treatment on tunnel, ensuring little possibility of catheter-related issues such as infection and
trial by using flow cytometry and cytokine magnetic bead assays (Luminex). displacement of catheter. Anticancer agents (30 mg/m2of cisplatin with/without
Cytokine profiles were further studied in co-culture experiments combining 10 mg/m2of docetaxel) were intra-arterially administered via the reservoir
PBMCs, HNSCC cell lines, and metformin. Results: Patients who received twice a week for 3 weeks, 180 mg/m2/6 times in total, without irradiation. The
metformin developed expanded NK cell populations, increased NKG2D ex- treatment effect was assessed using computed tomography, positron emission
pression, and a shift in their CD8+ T-cell memory phenotypes. Patient serum tomography, and biopsy. Results: The response rate of this approach was
ELISA examination revealed increased anti-tumorigenic cytokine profiles. 100%, with 31 and 1 case having complete response (CR) and partial response
Metformin treatment of HNSCC cell lines in vitro as well as HNSCC PBMCs (PR), respectively. Five patients with delayed regional lymph node metastasis or
ex vivo resulted in downregulation of STAT3 compared to healthy controls. PR underwent salvage surgery; 28 patients (87.5%) had disease-free survival,
Downregulation of STAT3 may be a potential mechanism in which metformin while 2 (6.2%) died due to local recurrence and 2 due to distant metastasis. All
stimulates NK cells. Conclusions: Here we show evidence that metformin patients developed CTCAE v4.0 Grade 2 oral mucositis in the flow area of the
treatment has a direct effect on the innate immune system in patients with intended artery, most of which disappeared in half a year. No dry mouth,
HNSCC, inducing an anti-tumorigenic immune response suggesting that dysgeusia, and eating disorder were observed because the patients did not
metformin continues to be a good candidate to yield improved clinical out- receive radiotherapy. No systemic adverse events such as hematologic toxicity
comes in patients with advanced stage HNSCC. Clinical trial information: and renal and/or hepatic injuries occurred. Conclusions: This method improved
NCT02325401. the adverse event of iaCT with radiotherapy, and the main advantage of
superselective iaCT was not lost.

6039 Poster Session (Board #28), Sat, 1:15 PM-4:15 PM 6040 Poster Session (Board #29), Sat, 1:15 PM-4:15 PM
Phase 1b/2, open label, multicenter study of intratumoral SD-101 in combi- Personalized TPF induction chemotherapy on the basis of stathmin expres-
nation with pembrolizumab in anti-PD-1 treatment naı̈ve patients with re- sion in patients with locally advanced oral squamous cell carcinoma. First
current or metastatic head and neck squamous cell carcinoma (HNSCC). First Author: Lai-Ping Zhong, Department of Oral & Maxillofacial-Head & Neck
Author: Ezra E.W. Cohen, University of California, San Diego, La Jolla, CA Oncology, Ninth People’s Hospital, Shanghai Jiao Tong University School of
Medicine, Shanghai, China
Background: SD-101, a synthetic CpG-ODN agonist of TLR9, stimulates
dendritic cells to release IFN-alpha and mature into antigen presenting cells - Background: A randomized phase 3 trial failed to demonstrate a survival
activating T cell anti-tumor responses. Pembrolizumab has demonstrated advantage for TPF (docetaxel, cisplatin, and 5-fluorouracil) induction che-
activity in HNSCC. Study DV3-MEL-01 (NCT02521870) assesses safety and motherapy in the overall study population of patients with locally advanced and
efficacy of SD-101 in combination with pembrolizumab in patients with resectable oral squamous cell carcinoma (OSCC). It is possible that person-
recurrent/metastatic HNSCC. We have previously reported a 27.3% ORR in 22 alized TPF-based induction chemotherapy might improve outcomes in
patients receiving 8 mg SD-101/injection in the modified ITT after at least 2 CT biomarker-defined subsets of patients. Methods: Immunohistochemical
scans due to late responses (Abstract 3560, ESMO 2018). Higher efficacy at a staining against stathmin was performed in pre-treatment biopsy specimens in
lower SD-101 dose, 2 mg/injection, has been reported in advanced melanoma 170 OSCC patients from our randomized trial. Chemoresistance to TPF
patients (LBA 45, ESMO 2018). Consequently, this dose is now being chemotherapy drugs regulated by stathmin expression was investigated. The
assessed in HNSCC. We report preliminary data with the 2 mg/injection dose in anti-cancer activity of TPF chemotherapy drugs alone, a combination of TPF
23 patients in mITT at the first CT scan. Methods: Anti-PD-1/PD-L1 naı̈ve drugs and PI3K-AKT-mTOR pathway inhibitors, and vincristine were in-
patients received 2 mg SD-101 intratumorally in 1 - 4 lesions (weekly x 4 doses vestigated using in vitro and in vivo OSCC models. Results: OSCC patients with
then Q3W x 7 doses). Pembrolizumab is was administered IV at 200 mg Q3W. low stathmin expression benefited from TPF induction chemotherapy in terms
Responses were assessed per RECIST v1.1. Results: 28 patients enrolled: of overall survival (HR = 0.102, 95% CI:0.013-0.781, P = 0.028), disease-
median age 63 y/o, male 68%; ECOG PS 0-1 (18%/82%); mean prior lines of free survival (HR = 0.070, 95% CI:0.009-0.525, P = 0.010), locoregional
systemic therapy 1 (0-3); mean treatment duration 70 days (1-253). Primary recurrence-free survival (HR = 0.070, 95% CI:0.009-0.524, P = 0.010), and
tumors: 19 (68%) oropharyngeal; 3 (10%) laryngeal; 2 (7%) hypopharyngeal; distant metastasis-free survival (HR = 0.101, 95% CI:0.013-0.767, P =
4 (14%) unknown. Mean number of target lesions: 1.82 (1 to 5). HPV status: 7 0.027). Stathmin overexpression promoted cellular proliferation and che-
(25%) +, 9 (32%) -, 12 (43%) unknown. 18 (64 %) discontinued treatment: moresistance to TPF drugs in OSCC (P , 0.05). PI3K pathway inhibitors
12 (42%) due to PD, 4 (16%) deaths, 1 (3%) consent withdrawn, 1 (3%) went decreased stathmin expression and phosphorylation, and improved the che-
to hospice. Mean follow up 2.70 months. Safety: 16 non-treatment-related mosensitivity to TPF drugs in vitro and in vivo (P , 0.05). Vincristine decreased
SAEs in 9 patients. 2 treatment-related Grade $3 AEs: sepsis (4%) and stathmin expression and phosphorylation, and OSCC lines were significantly
lymphopenia (4% ). No treatment-related deaths. Efficacy: 23 patients in the sensitive to vincristine in vitro and in vivo (P , 0.01). Conclusions: Our results
mITT population with first CT scan at day 64: ORR: CR: 2, PR: 3 (22%); SD: 6 suggest a potential personalized TPF induction chemotherapy on the basis of
(26%), PD: 7 (30%), non-evaluable: 5 (22%). Disease control rate (48%). 5 stathmin expression in OSCC patients: patients with low stathmin expression in
patients on study have not had a CT scan. Conclusions: SD-101 with Pem- the biopsy samples are suggested to receive TPF induction chemotherapy
brolizumab shows early promising data and is well tolerated. Additional follow- followed by surgical resection and post-operative radiotherapy. For patients
up scans from both dose cohorts are being evaluated and will be presented. with stathmin overexpression PI3K inhibitor is a good choice to improve the
Clinical trial information: NCT02521870. inductive effect of TPF chemotherapy drugs; vincristine is a potential alter-
native for a chemotherapy drug when a patient is chemoresistant to or un-
suitable to receive TPF chemotherapy drugs; however, more clinical trials are
warranted to verify this optimization protocol.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
352s Head and Neck Cancer

6041 Poster Session (Board #30), Sat, 1:15 PM-4:15 PM 6042 Poster Session (Board #31), Sat, 1:15 PM-4:15 PM
Predictors of early immunotherapy response in head and neck cancer: Per COX-2 expression and mesenchymal-transition status on circulating tumor
lesion analysis of a prospective randomized trial with nivolumab. First Author: cells to predict survival in patients with nasopharyngeal carcinoma: A pro-
Yao Yu, Memorial Sloan Kettering Cancer Center, New York, NY spective analysis. First Author: Yongjiang Li, Department of Oncology, West
China Hospital, Sichuan University, Chengdu, China
Background: The capacity for radiation and checkpoint inhibitors to elicit
clinical responses is impacted by tumor immunogenicity and the immune Background: The prognostic significance of circulating tumor cells (CTCs) in
microenvironment. We sought to determine whether head and neck primary head and neck cancer is still under active investigation. It remains unclear
site and metastatic tumor location was associated with initial response in non- whether just CTC count alone is sufficient to predict outcomes or whether the
irradiated lesions. Methods: We evaluated response in 144 non-irradiated functional status of the CTCs is also needed. We investigated the clinical
lesions from 59 patients with metastatic head and neck cancer enrolled significance of determining epithelial-to-mesenchymal transition status and
on a phase II randomized controlled trial of nivolumab with stereotactic body geno-/pheno-typic biomarkers of aggressiveness of CTCs in predicting out-
radiotherapy (n=30) vs. nivolumab alone (n=29). Nivolumab was administered comes in nasopharyngeal carcinoma (NPC). Methods: The prospective study
3 mg/kg intravenously every 2 weeks. Radiated lesions were treated with 27 Gy enrolled 131 patients with NPC. CTCs were isolated at baseline and at the end
/ 3 fractions to a single lesion within 14 days of the first dose of nivolumab. of concurrent chemoradiotherapy using the CanPatrol system. Subsequently,
Non-target lesion progression was defined $30% increase in the greatest axial the epithelial-mesenchymal transition (EMT) biomarkers and cyclooxygenase-
diameter 8 weeks after enrollment. Fisher’s exact test with nested bootstrap 2 (COX-2) expression status of the CTCs were identified by RNA-in situ hy-
resampling was used for univariate analysis. Logistic regression with a mixed bridization (ISH) method. Results: COX-2 expression was found in 87/131
random effects term was used for multivariate analysis. Results: Primary tumor (66.4%) patients at baseline and 53/115 (46.1%) patients post-treatment.
site, metastatic tumor organ sites, and the unadjusted likelihood of progressive Independent of initial COX-2 expression status, the patients with post-
disease by site are listed in the table. On multivariate logistic regression treatment COX-2 expression on CTCs had significantly poorer treatment re-
controlling for PD-L1 status (p=0.66) and viral status (p=0.29), lymph node sponse (P = 0.011), and higher risk of tumor relapse (P = 0.026) and me-
metastases (OR 0.79, p=0.0064) were associated with decreased risk of tastasis (P = 0.007). Similarly, post-treatment mesenchymal transition was
progression, while liver metastases (OR 1.39, p=0.014) and oral cavity pri- also associated with higher risk of tumor relapse and metastasis. In multi-
maries (OR 1.56, p=0.018) were associated with increased risk of progression variate analysis, post-treatment COX-2 expression on CTCs remained an in-
at 8 weeks, using lung metastases and larynx/hypopharynx primaries as ref- dependent prognostic indicator of poorer overall survival (HR 2.41, 95% CI
erence. Conclusions: Primary tumor subsite and metastasis location were 1.12-5.19; P = 0.024). Post-treatment COX-2 expression and mesenchymal
predictors of response or stable disease following treatment with nivolumab. transition in CTCs was the strongest prognostic indicator of overall survival
Metastases from oral cavity primaries and metastases to the liver were at on multivariate analysis (HR 2.73, 95% CI 1.28-5.83; P = 0.009).
increased risk of early progression. Clinical trial information: NCT02684253. Conclusions: Post-treatment COX-2 expression on CTCs, especially on the
Primary Tumor Subsite Total (n=59) Unadjusted Likelihood of Early Progression (p=0.050)
mesenchymal subtype, is a novel and promising prognostic indicator for NPC
patients treated with chemoradiation therapy. Future studies are needed to
Oropharynx 23 30%
Nasopharynx 10 32% validate our findings and further clarify the value of integrating the indicators
Larynx/Hypopharynx 13 31% with current clinical strategies in improving survival of NPC patients.
Oral cavity 7 82%
Other 6 33%
Metastatic Tumor Sites Total (n=144) Unadjusted Likelihood of Early Progression (p=0.001)
Lung 57 40%
LN 46 15%
Liver 23 74%
Soft Tissue 16 37%
Bone 2 0%

6043 Poster Session (Board #32), Sat, 1:15 PM-4:15 PM 6044 Poster Session (Board #33), Sat, 1:15 PM-4:15 PM
Development of a clinicomolecular risk stratification system for nonmeta- Association between human papillomavirus (HPV) status and duration of
static nasopharyngeal carcinoma using Epstein–Barr virus DNA and TNM response of anti-programmed cell death protein-1 (PD-1) inhibitors in patients
stage: A “Big data” analysis of 9,160 endemic cases. First Author: Fo-Ping with recurrent/metastatic (R/M) head and neck squamous cell carcinoma
Chen, Sun Yat-sen University Cancer Center, Guangzhou, China (HNSCC). First Author: Kedar Kirtane, Moffitt Cancer Center, Tampa, FL
Background: To construct a clinicomolecular index integrating circulating Background: Human papillomavirus status is known to be prognostic for
Epstein-barr virus (cfEBV) DNA with T- and N- categories for better prognos- patients with HNSCC. Current data suggests that HPV-positive HNSCC
tication in nasopharyngeal carcinoma (NPC). Methods: Clinical and treatment tumors exhibit increased infiltration of immune cells and higher levels of T-
records of 9,160 biopsy-proven, non-metastatic NPC cases were identified from cell exhaustion markers compared with HPV-negative tumors, possibly
an institutional “Big-data” platform. Decision tree modeling (DTM), recursive suggesting a difference in response patterns to immunotherapy. We eval-
partitioning analysis (RPA) and adjusted hazard ratio (AHR) methods were used uated whether HPV status is associated with duration of response in patients
to generate clinicomolecular risk models. Outcome prediction of the models receiving anti-PD-1 inhibitors. Methods: We performed a retrospective chart
were compared against 8th edition TNM stage and two RPA-original models. review of 54 patients at Moffitt Cancer Center who received either pem-
Results: We observed linearity between cfEBV DNA and DFS; cfEBV DNA brolizumab (N = 32) or nivolumab (N = 22) from February 2016 to July 2018
of . 2,000 copies was consistent for risk discretisation (HR . 1.0) for for R/M HNSCC. We collected the following data for our patient population:
DFS, OS and DMFS in our cohort of 9,160 patients. DTM, RPA-new and primary site of disease, stage, smoking status, duration of treatment, and
AHR modelling using a two-tiered stratification by cfEBV DNA (#2,000 overall survival (OS). Overall survival time was defined as the date of starting
and . 2,000 copies) and T- and N- categories yielded five risk groups with anti-PD-1 inhibitors to death. Primary disease site was oropharynx (N = 25),
significantly disparate DFS (P , 0.001 for all subgroup comparisons). AHR oral cavity (N = 13), larynx (N = 11), nasopharynx (N = 3) and unknown
model outperformed all other models and the TNM stage classification with primary (N = 2). HPV status was available for 37 patients. Analysis of survival
better hazard consistency, hazard discrimination, explained variation, sample and time on treatment was done using log-rank test. Results: Overall survival
size balance and likelihood difference. Importantly, our clinicomolecular AHR was not different with respect to primary site of disease, smoking, ECOG
groupings were significantly associated with the efficacy of different thera- status, or type of anti-PD-1 inhibitor, but was significantly longer for patients
peutic regimes. Outcomes were comparable between concurrent chemo- with HPV-positive vs HPV-negative HNSCCs (17 months vs 4.5 months; log
radiotherapy and intensity-modulated radiotherapy (IMRT) and IMRT alone for rank p , 0.001). Time on anti-PD-1 inhibitor was also significantly longer for
AHR1-2 (3-y DFS [AHR1] = 96.2% [IMRT] vs 95.3% [chemo-IMRT]; 3-y DFS patients with HPV-positive HNSCCs (7 months vs 3 months; log rank p ,
[AHR2] = 91.1% vs 90.4%). Neoadjuvant chemotherapy and chemo-IMRT 0.001). Conclusions: Our data suggests patients with HPV-positive R/M
was superior to chemo-IMRT alone for AHR4-5 (AHRDFS 0.77[0.65-0.93], HNSCCs have longer duration of response and OS on anti-PD-1 inhibitors
P = 0.005; 0.77[0.61-0.97], P = 0.027), but not for AHR3 (AHRDFS 1.07 compared to HPV-negative patients.
[0.86-1.34]). Conclusions: Here, we present a robust clinicomolecular
risk stratification system that outperforms the TNM stage classification in
non-metastatic NPC. Our clinicomolecular model is associated with the
efficacy of different therapeutic regimes.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Head and Neck Cancer 353s

6045 Poster Session (Board #34), Sat, 1:15 PM-4:15 PM 6046 Poster Session (Board #35), Sat, 1:15 PM-4:15 PM
Role of the oral and gut microbiota as a biomarker in locoregionally advanced Impact of adjuvant chemotherapy and cumulative cisplatin dose in locally
oropharyngeal squamous cell carcinoma (ROMA LA-OPSCC). First Author: advanced nasopharyngeal carcinoma (LA-NPC) treated with definitive
Marc Oliva Bernal, Princess Margaret Cancer Centre, University of Toronto, chemoradiotherapy. First Author: Marc Oliva Bernal, Princess Margaret
Toronto, ON, Canada Cancer Centre, University of Toronto, Toronto, ON, Canada
Background: The ROMA LA-OPSCC (NCT03759730) study prospectively Background: Total cumulative cisplatin dose (CDDP-D) (concurrent/induction/
evaluated the oral and gut microbiota in a single-centre cohort of LA-OPSCC adjuvant) in multimodality therapy for LA-NPC has been associated with
patients (pts) receiving chemoradiotherapy (CRT). Methods: LA-OPSCC pts survival at centers in Asia. We evaluated the survival impact of adjuvant
treated with definitive CRT (IMRT plus single-agent cisplatin) were eligible. chemotherapy (adj chemo) and total CDDP-D in a large, single institution
Oral swabs over the tumor site and stool samples were collected at baseline and Canadian cohort of LA-NPC. Methods: Patients (Pts) withWHO type II and III
end of CRT (EOT). Taxonomic profiles were generated by 16S rRNA se- LA-NPC treated with concurrent IMRT with high-dose CDDP and adj chemo
quencing. ANOSIM/Kruskal-Wallis tests were used to identify differences with CDDP/Carboplatin and 5-FU (maximum total/adjuvant CDDP-D= 540/
between baseline and EOT samples. Results: A total of 96 samples were 240 mg/m2) between 2003-2016 were analyzed. EBER status was tested by
collected from 24 evaluable pts (100% compliance). Baseline characteristics: ISH. Staging was classified by UICC/AJCC7thedition TNM. Kaplan-Meier 5-
median age = 61 (range, 50-71); smoking status current/former/never = 5/11/ year (5y) for overall survival (OS) and recurrence-free survival (RFS) were
8; HPV+/- = 23/1; stage I/II/III/IVA = 7/7/9/1; use of antibiotics = 12 pts. In oral calculated and compared by log-rank test betweenstage, adj chemo (yes vs no)
swabs, decreased Shannon diversity (p, 0.01) and changes in abundance and total CDDP-D (.300 vs #300mg/m2). Multivariable analysis (MVA)
(adjusted p value: q, 0.05) of multiple taxa including Prevotella, Veillonella, identified survival predictors. Results: A total of 312 pts were evaluated:
andStreptococcuswere observed at EOT vs baseline. Stool diversity did not median age = 49.8 (range 17.4-75.9); EBER+/-/unknown=67%/1%/32%;
differ between baseline and EOT (p= 0.42), but abundance of Ruminoccocus stage II/III/IV=2%/51%/47%; T4=36%; N3=17%; adj chemo=83% (21%
and Roseburia decreased (q, 0.05). CRT-associated changes remained switched to carboplatin); median total/adjuvant CDDP-D=380/160 mg/m2;
significant when controlled for stage, smoking, antibiotics, cisplatin dose and median follow-up 7.6 years (range 0.6-14.9). 5y OS differed by stage II-III vs IV
mucositis grade (p, 0.01). In HPV+ pts, stage I-II baseline oral swabs had (95% vs 80%, p,0.001) and total CDDP-D .300 vs #300mg/m2 (89% vs
higher relative abundance of Clostridium IV (p= 0.02) and Escherichia (p= 83%, p=0.02). Adj chemo and total CDDP-D impacted 5y OS in stage IV
0.04) than stage III, which had higher Fusobacterium (p =0.03) and Gemella (table). 5y RFS was higher in stage IV with total CDDP-D .300 vs #300mg/m2
(p, 0.01). Relative abundance of Actinobacteria (p , 0.01), Proteobacteria (74% vs 59%, p=0.03), with a trend in locoregional control (LRC) (91% vs
(p , 0.01) andFirmicutes (p = 0.03) was higher in stool from stage III pts. 80%, p=0.05) but not significant on distant control (DC) (78% vs 72%,
Akkermansia muciniphila was present in 57% of the stage I-II stool samples, p=0.36). Conclusions: Total CDDP-D .300 mg/m2 impacts OS in the overall
and 11% in stage III (p= 0.04). Conclusions: CRT in LA-OPSCC is associated cohort. The benefit of adj chemo and total CDDP-D on OS and RFS is sig-
with increases in potentially pathogenic genera in the oropharynx. HPV+ stage nificant in stage IV but not stage II-III LA-NPC, mainly due to higher LRC rather
III disease was associated with higher Fusobacterium in the oropharynx, which than DC.
has been implicated in tumor metastases, and with decreased prevalence of 5y OS and MVA by stage.
the immunotherapy-response-associated species Akkermansia in stool. These Variables
preliminary observations suggest an opportunity for the evaluation of IO based Adj chemo Total CDDP-D (mg/m2)
therapies or manipulation of the gut microbiota in this patient population. Stage Yes vs No >300 vs £300
Clinical trial information: NCT03759730. 5yOS % II-III 95% (90-99) vs 94% (86-100), p=0.52 94% (89-99) vs 96% (90-100), p=0.87
(95%CI) IV 82% (76-89) vs 61% (40-91), p,0.01 85% (56-84) vs 69% (78-93), p,0.01
MVA HR II-III 0.71 (0.25-2.02, p=0.52) 0.92 (0.34-2.47, p=0.87)
(95%CI) IV 0.32 (0.14-0.74, p,0.01) 0.43 (0.23-0.81, p,0.01)

6047 Poster Session (Board #36), Sat, 1:15 PM-4:15 PM 6048 Poster Session (Board #37), Sat, 1:15 PM-4:15 PM
Impact of tobacco smoking on radiotherapy outcomes in 1875 HPV-positive Correlation of angiogenic and immunomodulatory proteins with clinical
oropharynx cancer patients. First Author: Pernille Lassen, Department of outcomes of durvalumab (anti-PDL1) in recurrent/metastatic head and neck
Experimental Clinical Oncology, Aarhus, Denmark squamous cell carcinoma. First Author: Xiang Guo, MedImmune, Gai-
thersburg, MD
Background: This study investigates the impact of smoking on radiotherapy
(RT) outcome and survival in a population based cohort of HPV+ oropharynx Background: The potential for durvalumab, a PD-L1 blocking monoclonal
cancer (OPC). Methods: We identified all OPC with positive p16 staining from antibody, to treat head and neck squamous cell carcinoma (HNSCC) is being
2007 –2015 who received curative IMRT according to approved guidelines in evaluated in multiple clinical trials. We assessed circulating protein biomarkers
two oncology groups. Associations between smoking and locoregional control in HNSCC patients prior to treatment to better understand pathways related to
(LRC) and distant control (DC) were estimated by competing risk regression. clinical outcomes and potentially relevant for targeting in combination with
Disease free survival (DFS) and overall survival (OS) were estimated by durvalumab. Methods: Sixty-six selected serum proteins were measured by
proportional-hazards regression model. Multivariable analyses (MVA) adjusted multiplex immunoassay at baseline in HNSCC patients receiving durvalumab
for age, gender, performance status (PS), T- and N-category, and treatment treatment: 106 patients with high PD-L1 ($25% tumor cells; SP263 assay) in
regimen. Results: A total of 1875 patients were included. Median age was 59.2 phase II HAWK trial and 52 patients with low/no PD-L1 in phase II CONDOR
[31.3-86.8]; 79% (1481) were males; 96% (1651) had PS ,2; 71% (1337) trial. Results: Multivariate Cox modeling demonstrated that higher baseline
received concurrent chemo-radiotherapy (CRT) +/- hypoxic modification concentrations of angiogenic, pro-inflammatory, and myeloid-associated pro-
(Nimorazole); and 538 (29%) received RT alone +/- Nimorazole. 23% (425) teins (ANGPT2, CRP, IL6, S100A12) were associated with shorter overall
were current smokers (at time of diagnosis) and 46% (853) were ex-smokers. survival (OS), while higher concentration of a bone formation marker and
Median smoking pack-years (PY) was 20.1 in the total cohort, and higher in immunostimulatory hormone (BGLAP) correlated with longer OS in 158
current smokers vs ex-smokers (38 vs 20 PY, p,0.001). 63% of current durvalumab-treated HNSCC patients (P, 0.05). These 4 proteins also showed
smokers had .30 PYs. Median follow-up was 4.8 years. Actuarial 5-year higher baseline levels in patients with progressive disease (PD) compared to
univariate analysis showed that current smokers had a reduced probability of stable disease (SD) and partial or complete responses (PR/CR), while BGLAP
LRC (85% vs 92%, p=0.002), DC (88% vs 92%, p=0.046), DFS (69% vs 84%, had lower levels in PD compared to SD or PR/CR (Mann-Whitney P, 0.05). The
p,0.001), and OS (73% vs 88%, p=,0.001) compared to never-smokers 5 proteins remained significantly associated with OS in a multivariate model
(n=567). Outcomes for ex-smokers and never-smokers were similar. In MVA including PD-L1, ECOG, tumor size, and neutrophil count. Bone metastasis
current smoking retained strong independent significance for LRC (HR 1.73 status had no impact on the association of BGLAP with OS, which has not been
[1.18-2.53]), DFS (1.79 [1.35-2.36]) and OS (2.06 [1.49-2.84]). However, reported before in HNSCC. Interestingly, ANGPT2 level above median showed
DC was not significantly influenced by current smoking status (1.27 [0.83- the highest hazard ratio (HR = 2.2, P ,0.001) among all evaluated variables.
1.95]). Similar observations were found for .30PY. Conclusions: Heavy life- Furthermore, higher levels of VWF, an angiogenesis-related protein, correlated
time and current smoking negatively impacts LRC and survival in HPV+ OPC. with shorter OS by univariate survival analysis (P , 0.001). Conclusions: Our
While smoking mediated hypoxia could interfere with RT efficacy, a putative results suggested an important role of angiogenesis in the resistance of HNSCC
impact on tumor biology remains uncertain in the absence of a detriment to patients to durvalumab treatment, and ANGPT2 may have predictive utility for
distant metastasis risk. The findings support encouraging smoking cessation to durvalumab combination with an anti-angiogenic agent. The predictive value of
improve therapeutic efficacy of RT and to avoid excess smoking related BGLAP remains to be evaluated in a randomized clinical study. Clinical trial
mortality. information: NCT02207530; NCT02319044.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
354s Head and Neck Cancer

6049 Poster Session (Board #38), Sat, 1:15 PM-4:15 PM 6050 Poster Session (Board #39), Sat, 1:15 PM-4:15 PM
Phase 1b study of chemoprevention with green tea polyphenon E (PPE) and Connective tissue growth factor (CTGF) methylation status is associated with
epidermal growth factor receptor (EGFR)-tyrosine kinase inhibitor (Erlotinib) prognosis of patients with head and neck squamous cell carcinoma (HNSCC)
in patients (pts) with advanced premalignant (AP) lesions of the head and neck. treated with radiochemotherapy (RCHT): A multicenter study of the German
First Author: Dong Moon Shin, Emory Winship Cancer Institute, Atlanta, GA Cancer Consortium Radiation Oncology Group (DKTK-ROG). First Author:
Bouchra Tawk, German Cancer Research Center (DKFZ), Heidelberg and
Background: Based on the strong synergistic effects between green tea pol-
German Cancer Consortium (DKTK), Core Center Heidelberg, Heidelberg,
yphenon E (PPE) and EGFR-TKI in our preclinical studies (Int J Cancer, 2008;
Germany
Cancer Prev Res, 2009; JCO, 2009), we conducted a phase 1b study with PPE
and erlotinib combination for APL (mild-, moderate-, severe-dysplasia or Background: CTGF plays a central role in tissue remodeling and has emerged
carcinoma in situ [CIS]) of the oral cavity and larynx from 2/2011 to 11/2017 as an attractive novel therapeutic target. We sought to investigate the impact of
at Emory Winship Cancer Institute. Methods: All pts were enrolled after signing CTGF methylation status (CTGF-M) in predicting outcome of HNSCC patients
the IRB approved Informed Consent Form. Tissue biopsy before and at 6- undergoing RCHT. Methods: CTGF-M was discovered by applying Illumina
months (6-M) treatment was mandatory, and cytobrushed samples of the APL 450K/850K arrays to DNA extracted from FFPE material of patients homo-
and normal buccal mucosa at 3-, 6-, and 12-M were obtained for biomarker geneously treated with surgery followed by adjuvant cisplatin-based RCHT in
studies. Treatment included fixed dose of PPE (200 mg, P.O.,TID) and dose frame of the DKTK-ROG multicenter retrospective trial (n = 194, training
escalation of erlotinib P.O., (50mg [level 1], 75mg [level 2] and 100mg [level cohort). Methylation probes correlating with overall survival (OS) and pro-
3]) for 6-M. The primary endpoint was safety and toxicity, and secondary gression were identified using random forest. Validation was done in 4 cohorts
endpoints were evaluation of pathologic responses, cancer free survival (CFS) including DKTK-ROG definitive RCHT (n = 110) and 3 RCHT cohorts from
and biomarker modulation. Results: Out of 27 enrolled pts, 6 control subjects Heidelberg, Dresden and Munich (n = 222). CTGF-M and RNAseq data from
for biomarker studies, 2 ineligible, and 19 were treated with PPE and erlotinib The Cancer Genome Atlas (TCGA) (n = 206) were analyzed to identify dif-
for 6-M. Clinical characteristics of treated patients included median age, 63 ferentially expressed genes (DEG) as a function of CTGF-M. Results: Increased
yrs. (range,33-78); 9 M/10 F; 10 former or current smokers/9 never smokers; methylation of 2 probes (mapping to CTGF 3’UTR and gene body) was as-
15 severe dysplasia or CIS, 2 moderate dysplasia, 2 mild dysplasia; 13 had sociated with significantly improved OS in the training cohort (HR = 0.51, p =
surgical resection; 17 oral cavity primary; and 2 at larynx. 3 pts were treated at 0.044) and the validation cohort (HR = 0.67, p = 0.016), in multivariate cox
dose level 1, 4 at level 2, and 12 at level 3. Toxicity (G0 or G1 excluded) were: regressions adjusting for HPV status, age, T and N stages, location and
skin rash (1 G3, 1 G2), pruritus/dry skin (1 G2), fatigue (1 G2), diarrhea (1 G2), treatment (adjuvant vs definitive RCHT). In the TCGA, probes’ methylation was
epistaxis (1 G2), and hypertension (2 G3, 1 G2). Skin rash (associated with significantly inversely correlated with CTGF gene expression (r = -0.18 and
erlotinib) may be DLT and MTD has not been reached. The recommend doses -0.51, p , 0.05). 1843 DEGs were found at FDR , 0.05 as a function of
for phase 2 or 3 studies will be PPE 200mg TID plus erlotinib 100mg QD. 17 CTGF-M. Pathways mapping to tissue remodeling were significantly enriched
pts were assessed for pathologic responses at 6-M: pCR 7/17 (41%), pPR 2/17 for among downregulated genes in CTGF hypermethylated tumors. Increased
(12%), pSD 5/17 (29%) and pPD (3/17 (18%). The median follow up was CTGF methylation was inversely correlated with the mesenchymal subtype
32 months. Median CFS has not been reached. 16 pts are alive at the time of mRNA gene signature (r = -0.21, p = 0.0026) and correlated with the atypical
data analyses and 1 pt died (by noncancerous reason). Biomarker studies are subtype (r = 0.32, p = 0.000002). Conclusions: Implementation of CTGF-M in
ongoing for tissues and/or cytobrushed samples. Conclusions: The treatment of routine diagnostic is feasible and correlates well with CTGF gene-expression
the combination of green tea PPE plus erlotinib for 6-M was well tolerated in levels and activation of tissue remodeling pathways. Therefore, CTGF-M might
pts with APL of the head and neck, and showed significant pathologic response be instructive for stratifying HNSCC patients for CTGF targeting therapies.
rates (pCR and pPR, 53%). This combination therefore deserves further in- CTGF emerges as a promising prognostic marker independently of HPV status.
vestigation for efficacy testing. Clinical trial information: NCT01116336.

6051 Poster Session (Board #40), Sat, 1:15 PM-4:15 PM 6052 Poster Session (Board #41), Sat, 1:15 PM-4:15 PM
Racial disparities in outcome among head and neck cancer patients in the Pretreatment obesity prolongs survival in elderly patients ($65 years) with
United States: An analysis using SEER-Medicare linked database. First head and neck cancer (HNC). First Author: Parth Anil Desai, UT Health San
Author: Ikumi Suzuki, Univ of Maryland Marlene and Stewart Greenebaum Antonio, San Antonio, TX
Cancer Ctr, Baltimore, MD
Background: Pre-treatment Body mass Index (BMI) is an important prognostic factor in
Background: Despite overall decline in cancer mortality, African Americans HNC with variable survival benefit reported till date. Despite increasing incidence of
suffer from higher mortality in most cancer types including cancers of the head HNC in geriatric patients there is limited information on prognostic variables in this
and neck. These differences likely result from a complex interplay of clinical group. Methods: This is a single center, retrospective cohort study of patients with
HNC$65 years of age at their diagnosis from 2012 to 2016. Patients were stratified by
and non-clinical factors. We aim to estimate disparities in overall survival
BMI (Class 1- BMI,25 kg/m2 (underweight & normal), Class 2- BMI$25-29.9 kg/m2
across racial groups in HNSCC in the United States. Methods: This study used (overweight), Class 3- BMI$30 kg/m2 (obese) ). Various variables were collected &
SEER-Medicare linked database. We identified all patients aged 66 years or appropriate statistical analysis was done. Results: 188 elderly HNC patients with non-
older diagnosed with HNSCC as their first cancer from 1992 to 2011. We metastatic disease were stratified into three BMI groups (Table) & found to have evenly
excluded those in HMO, diagnosed by death certificate or autopsy, non-SCC, distributed co-variates. The Median OS was significantly higher in class 3 patients
unknown race, and missing month and/or year of diagnosis. Further exclusions (53 months) compared to class 1 (21 months) (p =0.02). In multivariate analysis, class
included metastatic disease, salivary gland cancers, receiving no treatment in 3 was an independent good prognostic indicator (Hazard ratio=0.44;Range-0.21-0.90,
the first 180 days, and unknown stage. Analytic data set included oropharynx, p=0.03). Stage, CCI score & gastrostomy tube were adverse prognostic factors in the
oral cavity, nasopharynx, hypopharynx, and larynx. Primary treatment was study. Conclusions: Obese elderly HNC patients have survival benefit over normal/
defined as any treatment modality received within 180 days after diagnosis. underweight patients. Predictive prognostic models incorporating BMI with other
Overall survival (OS) parameters were estimated across ethnic groups by the prognostic factors are needed to determine appropriate management in these patients.
Cox regression model stratified by site and stage of cancer at diagnosis, ad- Class 1 Class 2 Class 3
BMI <25 BMI 25-29.9 BMI ‡ 30
justed for clinical and demographic characteristics, and propensity score CHARACTERISTICS (78 patients) (67 patients) (43 patients) p-value
weighted. Results: Our study population included 15, 547 patients. Median MEAN AGE AT DIAGNOSIS (SD) Years 75.12 (7.46) 73.38 (6.13) 73.96 (6.14) 0.290
OS was 3.5 years (95% CI: 3.4-3.7) across all ethnic groups. African SEX (%)
WOMEN 27 (34.6) 16 (23.9) 15 (34.9) 0.305
Americans (AA) had inferior outcome with median OS of 2.0 years (95% CI: MEN 51 (65.4) 51 (76.1) 28 (65.1)
PATHOLOGICAL TYPE (%)
1.9-2.3) compared to 3.7 years (95% CI: 3.6-3.8) for Caucasian Americans SCC 71 (90.9) 58 (86.6) 36 (83.7) 0.409
OTHERS 8 (9.1) 9 (13.4) 7 (16.2)
(CA) (p , 0.0001). This difference was seen despite AA patients receiving TNM STAGE (%)
comparable treatments and presenting at similar stage of disease, except for STAGE 1
STAGE 2
5 (6.4)
13 (16.7)
11 (16.4)
9 (13.4)
8 (18.6)
9 (20.9)
0.745

cancers of the oral cavity where AA were more likely to present with advanced STAGE 3 9 (11.5) 7 (10.4) 4 (9.3)
STAGE 4A 34 (43.6) 24 (35.8) 13 (31.0)
disease (67% versus 47%; P , 0.001). The difference was most pronounced STAGE 4B 12 (15.3) 11 (16.0) 5 (11.6)
in the oropharynx where median OS was 1.9 years (95% CI: 1.7-2.1) for AA and UNKNOWN
CHARLSON COMORBIDITY INDEX (CCI) MEAN (SD)
5 (6.4) 5 (7.5) 4 (9.3)

3.8 years (95% CI: 3.5-4.1) in CA (P , 0.0001). AA also had consistently INITIAL TREATMENT (%)
CHEMOTHERAPY ONLY
5.03 (2.34)
5 (6.4)
5.54 (2.92)
1 (1.5)
5.28 (1.78)
2 (4.7)
0.460
0.628
worse OS over time from 1992 to 2011. This study clearly demonstrated AA CHEMOTHERAPY + SURGERY 0 (0.0) 0 (0.0) 1 (2.3)
CHEMORADIATION 16 (20.5) 13 (19.4) 8 (18.6)
have inferior outcomes despite similar treatments, comorbidities, age at di- CHEMORADIATION + SURGERY 6 (7.7) 10 (14.9) 5 (11.6)
agnosis, stage at presentation, tumor location, year of diagnosis and sex. RADIATION ONLY
RADIATION +SURGERY
7 (9.0)
8 (10.3)
9 (13.4)
10 (14.9)
7 (16.3)
7 (16.3)
Conclusions: The current study demonstrates inferior overall survival for Af- SURGERY ONLY
NONE/SUPPORTIVE ONLY/HOSPICE
19 (24.4)
17 (21.8)
14 (20.9)
10 (14.9)
7 (16.3)
6 (14.0)
rican American head and neck cancer patients independent of primary site and MEAN SURVIVAL TIME (SD) Days 368.35 (353.9) 527.08 (319.10) 590.27 (495.62) 0.080
DECEASED AT LAST FOLLOW-UP (%)
treatment modalities. NO 35 (44.9) 42 (62.7) 28 (65.1) 0.037
YES 43 (55.1) 25 (37.3) 15 (34.9)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Head and Neck Cancer 355s

6053 Poster Session (Board #42), Sat, 1:15 PM-4:15 PM 6054 Poster Session (Board #43), Sat, 1:15 PM-4:15 PM
Association of single nucleotide polymorphisms within genes in NF-kB, TGF-b, A phase II randomized control trial of erlotinib in combination with celecoxib
and JNK signaling pathways with the risks of nasopharyngeal carcinoma in in patients with operable oral squamous cell carcinoma (OSCC): Erlo-Xib
Chinese Han. First Author: Hanyi Zhang, Sichuan Cancer Hospital, Chengdu, Study. First Author: Sudhir Vasudevan Nair, Tata Memorial Centre, Mumbai,
China India
Background: Nasopharyngeal Carcinoma(NPC) is an Epstein-Barr virus(EBV) Background: When combined with COX-2 inhibitors, the EGFR tyrosine
associated malignancy with remarkable ethnic and geographical distribution. kinase inhibitor Erlotinib has shown a better antitumor response in pre-
The EBV oncoprotein latent membrane protein 1 (LMP1) is the primary on- clinical studies. Since high volume hospitals in many countries usually
cogene of EBV infection through the its signaling cascade and its connections have a longer waiting period for surgery, neoadjuvant targeted therapy may
to other pathways including NF-kB, TGF-b and JNK signaling, which plays an be helpful in reducing disease progression and downstaging oral squamous
important role in the pathogenesis of NPC. In GWASs(Genome-wide associ- cell cancers. Methods: Sixty-four treatment-naı̈ve operable oral cancer pa-
ation studies) associations these pathways were also identified. Single nu- tients were randomized into a four-arm window of opportunity study con-
cleotide polymorphisms (SNPs) in the regulatory regions may regulate the sisting of treatment with erlotinib 150mg daily, celecoxib 200mg twice
expression of genes in these pathways, or affect the function of the coded daily, the combination of both or observation alone (NCT02748707). Since
protein. Methods: Altogether 149 SNPs were covered by the 15 SNPs in the the regular wait period for surgery at our hospital was four to five weeks, we
TRAF2, TRAF3, NFKBIA, MAP2K4, and CHUK genes were genotyped in a planned a 21-day drug treatment versus observation followed by definitive
hospital-based case-control study of 350 NPC cases and 587 healthy controls surgery in the fourth week. MRI scans and biopsies were done before and
from the Chinese Han. The observed genotype frequencies in the controls were after drug treatment. Post-operative adjuvant treatments were given as per
tested for Hardy–Weinberg equilibrium (HWE) using the chi-square test. Odds the standard guidelines used for regular patients. Results: There were 10
ratios (ORs) and 95% confidence intervals (95% CIs) for associations between females and 54 males with a mean and median age of 44 and 45 years
genotypes and NPC risk and tumor characteristics were calculated by logistic respectively. Taking a 20% reduction in the maximum tumor dimension after
regression, and they were adjusted for multiple testing using the SNP spectral drug treatment (assessed clinically and radiologically) as partial response,
deposition (SNPSpD) approach for multilocus analyses. Results: We found the combination arm had a 60% partial response and a 25% stable disease.
one NFKBIA SNP was associated with NPC risk after adjustment for multi- Whereas, 60% in the control arm had disease progression. The ratio of the
ple comparisons. Minor allele carriers of the NFKBIA had an increased risk longest tumor dimension at day 21 versus day 0 (Clinical & MRI assessment)
of NPC (P＜0.05). The analyses were adjusted for age and gender. For a also showed a significant difference between the observation vs erlotinib
polymorphism with a variant allele frequency between 10 %and 50%, the arms (p , 0.001) using Mann-Whitney Test. Grade II/III rashes was the
study had greater than 90% power to detect an OR of 1.50 at a signifi- commonly observed toxicity predominantly in the combination arm. Though
cance level of 0.05 (PS—software for power and sample size calculation, not powered for survival analysis, a significant difference (p = 0.048) was
http://biostat.mc.vanderbilt.edu/twiki/bin/view/Main/PowerSampleSize). The observed for two-year overall survival for celecoxib + control (60%) versus
other genotyped SNPs were not associated with NPC risk. Conclusions: Our erlotinib + combination groups (86%) using Kaplan Meier estimator. Bio-
data suggests that genetic variation especially in the NFKBIA maybe a useful marker analysis (transcriptome sequencing and IHC) is being done on pre
biomarker for NPC screening and further studies are warranted. and post-treatment tumor specimens and the final results will be presented.
Conclusion: Preoperative targeted therapy with erlotinib and celecoxib
combination can arrest disease progression and downstage tumors with
possible impact on survival. The identified biomarkers can further refine a
future cohort for effective neoadjuvant targeted therapy in oral cancers.
Clinical trial information: CTRI/2012/07/002828.

6055 Poster Session (Board #44), Sat, 1:15 PM-4:15 PM 6056 Poster Session (Board #45), Sat, 1:15 PM-4:15 PM
Immune signatures associated with response to neoadjuvant PD-1 blockade Risk of mortality varies by type of fat consumed in a longitudinal cohort of
in oral cavity cancer. First Author: Hannah Knochelmann, Medical University head and neck cancer patients. First Author: Hania M. Taha, Department of
of South Carolina, Charleston, SC Food Science and Human Nutrition, University of Illinois at Urbana-
Champaign, Urbana, IL
Background: PD-1 inhibition therapy has revolutionized clinical medicine as it
can mediate durable responses in a small cohort of patients. Yet, it remains Background: Dietary interventions have promise for improving cancer out-
incompletely understood why these patients respond. To address this question, comes, but remain an understudied area of cancer care. The relationship
we studied patients with oral cavity squamous cell carcinoma (OCSCC) between head and neck squamous cell carcinoma (HNSCC) mortality and
to elucidate immune phenotypes associated with response to nivolumab. dietary fat intake has not yet been examined. The objective of this study was
Methods: We defined the immune profile from the blood and tumor of patients to determine how pre- and post-treatment intake of various types of fat are
on neoadjuvant nivolumab. We tested if tumor-infiltrating lymphocytes (TIL) associated with disease-specific and all-cause mortality in adults diagnosed
could be preferentially expanded ex vivo from nivolumab-responsive patients with HNSCC. Methods: Our sample included 472 newly diagnosed HNSCC
versus those who were either non-responsive or had never received nivolumab. patients recruited into the University of Michigan Head and Neck Spe-
During the course of therapy, we comprehensively profiled a number of surface cialized Program of Research Excellence (HN-SPORE) between 2008 and
markers on patients’ T cells to define their activation status, cytotoxic capacity 2012. Participants completed pre-treatment and post-treatment Food
and memory phenotype. Moreover, the immune profile of the peripheral blood Frequency Questionnaires (FFQs) and health surveys. Multivariable Cox
was assessed pre- and post-nivolumab using high dimensional mass cytometry. Proportional Hazards models were used to test the associations between both
Results: Regardless of PD-1 therapy, TIL were successfully expanded from 11 the type and quantity of fat intake (categorized into tertiles: low, medium and
of the 12 patients. TIL were comprised of both CD4+ and CD8+ T cells. Ad- high intake) and time to mortality, after adjusting for relevant covariates. Fat
ditional investigation revealed that the frequency of CD4+ T cells and effector types included animal, vegetable, medium-chain-fatty-acids (MCFA), long-
memory T cells in TIL correlated with disease progression (CD4: p = 0.04, r = chain-fatty-acids (LCFA), unsaturated, saturated, and trans. Results: During
0.74, effector memory: p = 0.046, r = 0.74). TILs from responders expressed the study period, there were 144 total deaths and 97 cancer-specific deaths.
higher CD26 (p = 0.007, r = -0.88) and Tim3 (p = 0.045, r = -0.74) while PD- In considering pre-treatment dietary intake, compared to low intake levels of
1, Lag3, and Ox40 were not differentially expressed based on response. LCFA, high intake was associated with a reduced risk of all-cause mortality
Spearman correlation and Mann Whitney U test were used to assess pheno- (HR: 0.57; 95% CI: 0.34–0.94). High intakes of unsaturated-fats were
typic differences. Conclusions: We demonstrate, for the first time, that TIL can associated with a reduced risk of HNSCC-specific mortality compared to low
be reliably expanded from OCSCC patients on neoadjuvant nivolumab. intake (HR 0.52; 95% CI: 0.29–0.93). Considering post-treatment dietary
Moreover, individuals who were responsive to PD-1 blockade had TIL variables, medium (HR: 0.21; 95% CI: 0.08–0.49) and high (HR: 0.41;
expressing high levels of CD26 and Tim3. Future studies will explore if these 95% CI: 0.21–0.78) total fat intakes were associated with reduced risk of
markers are predictive of responses and if they contribute to treatment all-cause mortality compared to low intake. Medium (HR: 0.25; 95% CI:
outcome. 0.08–0.67) and high (HR: 0.26; 95% CI: 0.09–0.67) total fat intakes were
associated with reduced risk of HNSCC-specific mortality compared to low
intake. Conclusions: Our data suggest that HNSCC prognosis may vary
depending on both the type and quantity of fats consumed, specifically total
fat and long chain fatty acids. Clinical intervention trials are needed to
further examine this hypothesis.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
356s Head and Neck Cancer

6057 Poster Session (Board #46), Sat, 1:15 PM-4:15 PM 6058 Poster Session (Board #47), Sat, 1:15 PM-4:15 PM
Identifying adverse molecular features of HPV+ head and neck cancers using Depth of invasion in early oral cancers: Is it an independent prognostic
patient-derived models. First Author: Devraj Basu, The University of factor? First Author: Harsh Dhar, Tata Memorial Hospital, Mumbai, India
Pennsylvania, Philadelphia, PA
Background: Depth of invasion (DOI) has been incorporated in the new AJCC
Background: Advancing therapy for human papilloma virus-related (HPV+) TNM staging (8th edition) for oral cancers. We hypothesized that the negative
head and neck squamous cell carcinoma (HNSCC) is hindered by inadequate effect of increasing DOI on outcomes was a result of an increased propensity to
preclinical models and risk stratification tools. This study addresses both node metastasis and appropriate neck treatment would negate its detrimental
barriers by characterizing a panel of patient-derived xenografts (PDXs) that effect on outcomes. Methods: Patients with T1/ T2 oral squamous cell car-
includes nine new models of HPV+ disease. Methods: Exome-sequenced cinoma, clinically node negative, from a previously reported Randomized
genetic features of the PDXs were compared to their growth properties and Controlled Trial (NCT 00193765) formed the cohort for this study. Patients
the outcomes of their patients of origin. Genetic traits with potential were restaged according to the new staging system . Overall survival(OS) was
prognostic utility were validated in multiple retrospective patient cohorts. estimated by the revised T stage for the entire cohort and separately for those
Results: The HPV+ PDXs avoided known artifacts in HPV+ cell lines, in- who underwent END and those who did not (TND arm) using Kaplan Meier and
cluding 3q amplifications and loss of PIK3CA mutants, while enriching for log rank test . Multivariate analysis was performed using Cox proportional
alterations in H3K4 methyltransferases and Notch pathway genes. A positive hazard model making adjustment for neck treatment, T stage, site, prognostic
association emerged between PDX tumor mutational burden (TMB) and their factors and the interaction between revised T stage and neck treatment.
growth efficiency both in vivo and as organoids. This finding led to identi- Results: Of the 596 patients 577 were evaluable, with a median follow up of
fication in The Cancer Genome Atlas (TCGA) of an association between high 77.57 months. Initial pT staging was pT1, 389(67.4%); pT2, 181(31.4%);
TMB and worse survival of early-stage HPV-negative cancers but not HPV+ pT3, 7(1.2%) and was modified to pT1, 195(33.8%); pT2, 280(48.5%); pT3,
ones. Insight into aggressive HPV+ disease came from a PDX established 102(17.7%) on restaging . 288 patients underwent END and 289 did not
from a patient before lethal relapse. The reduced levels of viral E7 and (TND arm). For the entire cohort 5-year OS rates were 79.0% [95 %CI, 73.12-
p16INK4A present in this model were also detected in early lethal HPV+ cases 84.88] for pT1, 69.4% [95% CI, 63.91-74.89] for pT2 and 53.0% [95% CI,
in TCGA as well as the recurrences in a second HPV+ HNSCC cohort (JHU). 43.2 -62.8] for pT3 with significant difference between the 3 groups (p ,
This observation suggested a diminished contribution of viral oncogenes to 0.001). In those without upfront neck treatment( TND ), OS difference was
cell cycle dysregulation in aggressive cases. To evaluate this possibility, maintained between the pT1 and pT2 groups [81.1% (95%CI, 73.26-88.94)
hierarchical clustering of both cohorts was performed based on expression of vs 65.0% (95%CI, 56.77-73.23)], p = 0.004. This difference was not ap-
E2F target genes. This analysis discovered a distinct cell cycle-related parent in the END arm ,pT1 -76.9% (95 %CI, 68.47-85.33) vs pT2 -73.7%
transcriptional pattern in the clusters of cases containing the recurrences (95%CI, 66.25-81.15), p = 0.73. T3 tumours had uniformly poor survival
and early lethal events. Furthermore, a subset of these transcripts proved to irrespective of neck treatment. On multivariate analysis of the revised pT1/T2
be stage-independent predictors of survival for the HPV+ HNSCCs in both cohort (n = 475), pT stage, neck treatment and grade were independent
TCGA and JHU cohorts. Conclusions: Characterizing the most HPV+ patient- prognostic factors impacting OS. There was a significant interaction between
derived models described to date revealed novel prognostic utility for E2F the T stage and neck treatment (p = 0.03). Conclusions: When DOI , 10 mm,
target expression in HPV+ HNSCCs and TMB in HPV-negative HNSCCs. END supplants the prognostic implication of depth with similar outcomes for
These features have potential for application to risk stratification, biomarker T1 and T2 tumours (new AJCC staging). The exact role of DOI on outcomes
development, and trial design. warrants further research. Clinical trial information: NCT00193765.

6059 Poster Session (Board #48), Sat, 1:15 PM-4:15 PM 6060 Poster Session (Board #49), Sat, 1:15 PM-4:15 PM
Prophylactic gabapentin decreases fatigue and swallowing difficulty in Predictive biomarkers for response to nivolumab in head and neck squamous
patients undergoing concurrent chemo-radiation (CCR) for head and neck cell carcinoma (HNSCC) (NCT#03652142). First Author: Amanda Psyrri,
cancer (HNC): Interim results from a randomized controlled trial. First Author: Attikon University Hospital, Athens, Greece
Barbara A. Murphy, Vanderbilt University Medical Center, Nashville, TN
Background: Tumor immune cell compositions determine response to im-
Background: Preliminary data suggest that gabapentin (G) administered munotherapy. For a better understanding of the mechanisms of resistance to
during HNC radiation may decrease treatment associated pain. To confirm nivolumab in HNSCC, we sought to investigate a prospective cohort of longi-
this, we undertook a prospective, randomized trial in HNC patients un- tudinal HNSCC samples from recurrent/metastatic HNSCC pts treated with
dergoing CCR. Primary outcomes: pain severity (scale of 0-10) and opioid nivolumab and identify biomarkers of response and resistance. We will spe-
use. Exploratory outcomes: local and systemic symptoms measured by the cifically focus on modulation of immune markers following two cycles of
Vanderbilt Head and Neck Symptom Survey version 2 plus the general nivolumab. Methods: Patients with platinum-refractory HNSCC with no con-
symptom survey (VHNSSv2/GSS). We report results of the exploratory traindication to nivolumab therapy are included in this study. Tumor biopsies
endpoints from the interim analysis. Methods: Measures: VHNSSv2 - 50 are performed at baseline, 24-72 hours after the second cycle and at pro-
items, demonstrated reliability/validated, captures acute/chronic local HNC gression with appropriate written informed consent. Samples were assessed for
specific symptoms; GSS – 11 item checklist, demonstrated content validity, the presence of Tertiary Lymphoid Structures (TLS), PD-L1 expression (TPS and
captures acute/chronic systemic symptoms. Population: HNC patients ($ CPS) and CD8 T cell infiltrates combined with Ki67 (CD8/Ki67 double IHC
stage 2) undergoing primary or adjuvant CCR. Procedures: Randomized to stain). The primary outcome measure of the study is change in the percentage of
standard pain management (SPM) or SPM + G dose escalated to 900mg tid. immune cells in post treatment compared to baseline biopsies. Secondary
VHNSSv2/GSS completed weekly during treatment beginning week 1 and endpoints include safety of performing a second biopsy, best overall response
ending the last week of radiation therapy. Results were analyzed using a rate, biomarker expression in association with response and survival. Evaluation
mixed-effects regression analysis adjusting for baseline levels of each of other biomarkers including tumor mutational burden, HLA class I and II
symptom. Results: 71 patients completed a mean of 5.5 surveys. Patients on expression and adaptive immunity cell populations using multiplex IF is on-
G experienced a reduction in overall systemic symptoms as measured by the going. Results: Of 20 patients evaluable for response, 14 had PD (8 of whom
GSS (11-items, p = 0.0073), fatigue (two-items, p = 0.013 ) sleep dis- showed hyper-progression) and 6 attained disease control (1 with PR). PD-L1
turbance (five items, p , 0.0001) , neurosensory eating (3 items, p = status (CPS or TPS) was not altered by treatment (p = 0.905) and CPS . 20 pre-
0.026), phlegm-related symptoms (4 items, p = 0.004), and trend to better treatment showed a favorable trend towards response (p = 0.117). Absence of
smell (2 items, p = 0.055). No impact on swallow, xerostomia, voice, dental tertiary lymphoid structures was associated with disease progression (p =
or musculoskeletal symptoms was noted. Conclusions: This exploratory 0.0374). Infiltrating plasma cell count remained unchanged pre- and post-
analysis suggests that G may moderate neurological and neuropsychiatric treatment and was unrelated to response (p = 0.458). The percentage of
toxicities in HNC patients undergoing CCR. Further studies are warranted. proliferating CD8+ T cells (CD8+/Ki67+) increased in post-treatment biopsies
in the entire population (p = 0.022) and especially in progressors (p = 0.039).
Pre-treatment CD8+ T cell density was higher in patients with hyper-progression
compared to progressors (p = 0.029). Conclusions: Increased percentage of
proliferating CD8+ T cells in progressors might represent dysfunctional T cells
as has been recently shown in melanoma pts (Li H et al Cell 2019) and clinical
efforts to reactivate intratumoral T cells may augment the efficacy of PD1
checkpoint inhibitors. Clinical trial information: NCT03652142.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Head and Neck Cancer 357s

6061 Poster Session (Board #50), Sat, 1:15 PM-4:15 PM 6062 Poster Session (Board #51), Sat, 1:15 PM-4:15 PM
Prognostic impact of baseline circulating tumor cells (CTCs) detected by the Tumor volume, circulating tumor cells, and cfDNA changes during radio-
isolation by size of epithelial tumor cells (ISET) in locally advanced head and therapy in patients with head and neck cancer. First Author: Sweet Ping Ng,
neck squamous cell carcinoma (LAHNSCC): Results of a prospective study. Peter MacCallum Cancer Centre, Melbourne, Australia
First Author: Thiago Bueno Oliveira, A.C. Camargo Cancer Center, S~ ao Paulo,
Background: Head and neck (HN) cancer treatment response relies heavily
Brazil
on macroscopic clinical findings. Monitoring of circulating markers during
Background: The prognostic impact of CTCs in LAHNSCC is yet to be de- treatment may improve detection of responders versus non-responders
termined, with conflicting results in previous trials, the majority utilizing during radiotherapy (RT). Our work prospectively describes the changes
cytokeratin dependent techniques for identification and counting of CTCs. The in gross tumor volume (GTV), circulating tumor cell (CTC), and cell-free DNA
primary objective of this study is to determine the detection rates using the (cfDNA) enumeration during RT. Methods: Patients with intact HN squa-
ISET method, and the prognostic role of CTCs in LAHNSCC patients (pts) mous cell carcinoma were enrolled in a prospective IRB-approved study. Pre-,
treated with a curative intent. Methods: In this prospective study, peripheral after first RT, weekly in-, and post-RT blood samples were collected. Serial
blood samples of pts with non-metastatic LAHNSCC, stages III/IV, were an- pre-, weekly in-, and post-RT magnetic resonance imaging (MRI) was ob-
alyzed for CTCs using the ISET method, in two scenarios: curative surgical tained. Serial GTV measurements were recorded. CTC was enumerated using
resection and adjuvant radiotherapy (CTCs before RT) and candidates for a the FDA-approved CellSearch (Menarini Silicon Biosystems) system. Plasma
non-surgical strategy (unresectable or organ preservation) either with upfront were collected and cfNA from pre-, mid- and post-RT timepoints were
RT concurrent with chemotherapy (CT) or cetuximab (CTCs before RT), or isolated using the MagMAX Nucleic Acid Isolation Kit (Thermo Fisher Sci-
preceded by induction CT (CTCs before ICT). Results: Eighty-three pts were entific), and cfDNA were quantified with Qubit high sensitivity dsDNA assay
included, the majority males (83%), with oropharynx primary (50%) and (Invitrogen). Results: 40 patients were eligible for analysis. Median age
submitted to ICT (40%). The detection rate of baseline CTCs was 94% (78/ was 60 years and 36 were males. The median pre-RT GTV was 14.1cc
83), and CTCs counts were significantly correlated with survival. For each (range 1.3 – 44.9cc). There was a median reduction of 81% in GTV by week
increase of 1 CTC at baseline there was a relative increase of 18% in the risk 4 (p , 0.0001). Of the 341 samples analyzed for CTC, 146 (43%) had
of death (HR = 1.18; 95%CI: 1.06-1.31; p , 0.001), 16% in the risk of detectable CTC. 7 patients had detectable pre-RT CTCs (1-3/7.5ml blood).
progression (HR = 1.16; 95%CI: 1.04-1.28; p = 0.004) and a reduction of There was no correlation between cancer stage, nodal status, and GTV with
26% in the odds of complete response to treatment (non-surgical group only - detection of CTC. After 1 fraction of RT, 14 patients had CTCs detected,
OR = 0.74; 95%CI: 0.58-0.95; p = 0.022). Using the maximum of the including 11 who had no CTC detected prior. All patients had CTC detected
standardized log-rank statistic proposed by Lausen and Schumacher we es- at some point during RT except for 2 patients who had none. In week 4,
tablish cut off points for overall survival (OS) and progression-free survival with significant reduction in GTV, 25 (63%) had detectable CTC. 16 and
(PFS). Pts with CTCs , 6.5/ml had an estimated 2y OS of 85.6% versus 11 patients had detectable CTC in final week and post-RT timepoints. The
22.9% for CTCs $ 6.5/ml (HR = 0.18; 95%CI: 0.06-0.49; P , 0.0001) and cfDNA levels increased during RT, with its highest level in the final week of
pts with CTCs # 3.8/ml had an estimated 2y PFS of 71.8% versus 37% for RT and lowest at post-RT time-point, inversely correlated with GTV.
CTCs . 3.8/ml (HR = 0.32; 95%CI:0.15-0.67; p = 0.001). Conclusions: The Conclusions: Our study showed that CTCs can be detected during RT,
detection rate of baseline CTCs using the ISET method was very high in suggesting mobilization into peripheral circulation during RT with un-
LAHNSCC and the counts of CTCs were strongly correlated with survival and known viability. cfDNA kinetics during RT correlated with CTC release, and
response to treatment. may indicate apoptotic change during RT. Combined cfDNA-CTC as an
early marker of treatment response should be investigated further.

6063 Poster Session (Board #52), Sat, 1:15 PM-4:15 PM 6064 Poster Session (Board #53), Sat, 1:15 PM-4:15 PM
Biomarker predictors of outcome from a randomized trial of nivolumab +/- Safety and efficacy of docetaxel combined with cisplatin as induction
stereotactic body radiotherapy (SBRT) in metastatic (M1) head and neck chemotherapy followed by cisplatin concurrent chemoradiotherapy plus
squamous cell carcinoma (HNSCC). First Author: Sean Matthew McBride, gemcitabine as adjuvant chemotherapy in locally advanced nasopharyngeal
Memorial Sloan Kettering Cancer Center, New York, NY carcinoma: A prospective and multicenter phase II trial. First Author: Zhi Hui
Wang, The Fifth Hospital Of Sun Yat-sen University, Zhuhai, China
Background: A minority of patients with M1 HNSCC respond to the anti-
programmed death (PD-1) monoclonal antibody, Nivolumab (Nivo). We Background: Radiation therapy is the only curative treatment modality for
sought to determine biomarker predictors of outcome to Nivo in M1 HNSCC. nonmetastatic nasopharyngeal cancer (NPC). Concurrent chemoradiation
Methods: Patients with M1 HNSCC were randomized (n = 60) with strati- (CCRT) is the standard treatment strategy for NPC in locally advanced stages.
fication by virus status (EBV/HPV+ vs not) to either Nivo alone or Nivo+SBRT However, the results after such treatment are suboptimal. Clearly, novel
to a single lesion. The primary end-point was objective response rate (ORR) treatment strategies are needed to further improve patients’ survival rates. This
in non-irradiated lesions with overall survival (OS) as a secondary end-point. trial aimed to determine the safety and efficacy of a new treatment strategy.
PD-L1 staining in $ 1% of tumor cells was regarded as positive. Tumor Methods: Patients with stage III – IVa-b NPC received TP (docetaxel 75 mg/
mutation burden (TMB) was determined using a next generation sequencing m2, cisplatin 75 mg/m2 every 3 weeks for 2-3 cycles) followed by cisplatin
assay (MSK IMPACT). Predictive model selection was done to minimize the chemotherapy concurrently with either 3-dimentional conformal radiation
Akaike Information Criterion (AIC). Results: There was no difference in ORR therapy or intensity-modulated radiation therapy plus gemcitabine (1000mg/
comparing the two treatment arms (p = 0.86). On univariate analysis, virus m2 every 2 weeks for 2 cycles) as adjuvant chemotherapy. Objective response
status trended towards predicting both ORR (Positive 41.9% vs Negative rates and acute toxicity were assessed based on RECIST (1.1) and CTCAE
20.7%, (p = 0.14)) and OS (1-yr OS for Positive, 65.9% vs 1-yr OS for v.4.0, respectively. Kaplan-Meier analysis was used to calculate survival rates.
Negative 40.6%, p = 0.08). In the sub-group of patients for whom PDL1 This trial is registered with the Chinese Clinical Trials Registry, number
staining was available (n = 56), there was a trend towards association with ChiCTR-OIC-17011464. Results: From July 2010 to July 2017, 20 eligible
ORR: PDL1+ 50% vs PDL1- 23.5% (p = 0.08). PDL1+ patients demon- patients with nonmetastatic stage III-IVb NPC were enrolled. The objective
strated significantly longer OS (1-yr OS 63% vs 47%, p = 0.02). There was response rates were 90% (3 complete responses [CRs] and 15 partial re-
no association between virus status and PDL1 staining. IMPACT data was sponses [PRs]) after two or three cycles of induction chemotherapy (ICT) and
available in 46 patients. TMB was significantly higher in virus negative 100% (17 CRs and three PRs) after CCRT plus gemcitabine adjuvant che-
(mean 8.1 mut/mB) vs. virus positive (mean 4.6 mut/mB) (p = 0.01); TMB motherapy, respectively. With a median follow-up time of 41 months, the 3-
was similar comparing PDL1+ to PDL1- (p = 0.47). TMB was higher in year overall survival rates were 90% (18/20,95% confidence interval [CI],
responders than non-responders (p = 0.02); TMB was significantly higher 76.9%-100%).The 3-year progression-free survival, distant metastasis-free
in virus positive responders (p = 0.01) and trended towards being higher in survival, and local progression-free survival rates were 80% (16/20,95% CI,
virus negative responders (p = 0.10). A model to predict ORR that included 62.5%-97.5%), 85% (17/20,95% CI, 69.4%-100%),95% (19/20,95% CI,
PDL1, virus status, and TMB minimized AIC with a C-index of 0.72. A model 85,4%-100%), respectively. The most frequent grade 3–4 toxicities were
to predict OS that included PDL1 and virus status minimized AIC with a neutropenia (3/20,15%) and nausea (2/20,10%) after ICT and thrombocy-
C-index of 0.62. Conclusions: A multi-variate model including viral status, topenia (6/20,30%) and leukopenia (6/20,30%) after CCRT plus gemcitabine
PDL1 status, and TMB predicts well response to Nivolumab in M1 HNSCC. adjuvant chemotherapy. Conclusions: Neoadjuvant TP followed by concurrent
A model containing PDL1 and virus had moderate predictive value for OS. chemoradiation plus gemcitabine as adjuvant chemotherapy was well tolerated
Clinical trial information: NCT02684253. and produced promising outcomes in patients with LA-NPC in this hypothesis-
generating study. The authors concluded that randomized controlled trials are
warranted to definitively confirm this aggressive and potentially efficacious
strategy. Clinical trial information: ChiCTR-OIC-17011464.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
358s Head and Neck Cancer

6065 Poster Session (Board #54), Sat, 1:15 PM-4:15 PM 6066 Poster Session (Board #55), Sat, 1:15 PM-4:15 PM
Safety of radiotherapy with concurrent and adjuvant MEDI4736 (durvalumab) A randomized trial of laryngeal organ preservation evaluating two cycles
in patients with locoregionally advanced head and neck cancer with a con- of induction chemotherapy with platinum, docetaxel, and a novel Bcl-xL
traindication to cisplatin: NRG-HN004. First Author: Loren K. Mell, University inhibitor. First Author: Paul Swiecicki, University of Michigan Rogel Cancer
of California San Diego Moores Cancer Center, La Jolla, CA Center, Ann Arbor, MI
Background: MEDI4736 (durvalumab), a PD-L1 inhibitor, has shown prom- Background: Laryngeal bio-selection with induction chemotherapy (platinum +
ising antitumor activity and safety in head and neck squamous cell carcinoma 5-FU infusion: PF) has been demonstrated to result in impressive rates of survival
(HNSCC). A phase II/III trial with lead-in component was designed to evaluate and organ preservation with moderate toxicities. We sought to reduce the toxicity
the safety and efficacy of concurrent and adjuvant MEDI4736 with radiation of PF by substituting docetaxel (T) for 5-FU and improve the tumor response rate
therapy (RT) for HNSCC patients with a contraindication to cisplatin. Safety with the addition of the Bcl-xL inhibitor AT-101. Methods: Pts with advanced
data for 10 patients on the lead-in study are reported. Methods: Eligible stage laryngeal cancer were treated with 1 cycle of T 75 mg/m2 + platinum (P)
patients had previously untreated locoregionally advanced unresected SCC of (cisplatin 100 mg/m2 or carboplatin AUC 6) and were randomized 2:1 to the
the larynx, hypopharynx, oropharynx (OPX), oral cavity, or unknown head/neck addition of AT-101. Patients with a CR proceeded to chemoradiation (CRT) with
primary (AJCC 7th stage III-IVB). Contraindications to cisplatin included renal concurrent weekly P. All pts with PR or NR underwent a second cycle of induction
or hearing impairment, age $ 70 with moderate or severe comorbidity/ with TP + AT-101. Pts with a . 50% response (CR or PR) after the second cycle
vulnerability to cisplatin, or age, 70 with severe comorbidity/vulnerability, of induction chemotherapy underwent CRT. Pts with a # 50% response (NR)
based on 6 validated indexes. Intravenous MEDI4736 1500 mg was delivered underwent laryngectomy. PET-CT was performed 12 weeks after CRT. Pts with
at weeks -2, 2, 6, 10, 14, 18, and 22 with RT (70 Gy in 35 daily fractions residual disease underwent salvage laryngectomy (SL). Results: 54 eligible pts
weeks 1-7). The primary endpoint was dose-limiting toxicity (DLT), defined as a were enrolled; 46 M, 8 F; median age 59; 26 T4; stage III 16, stage IV 38; site: 2
high-grade adverse event (AE; NCI CTCAE version 4.0) definitely/probably hypopharyngeal, 39 supraglottic, 11 glottic, 2 subglottic. After cycle 1 of in-
related to MEDI4736 up to 4 weeks following completion of RT; 0-2 DLTs in 8 duction, 2/54 (4%) died prior to assessment by DL and 2 were removed from
evaluable patients was considered acceptable. Results: Characteristics of the protocol due to adverse events (AEs). 29/50 pts (58%) had $ 50% response, 3 of
10 enrolled patients were: 30% age $ 70, 90% male, 100% Caucasian, 40% which had CR; 2 proceeded to CRT & 1 received a 2nd cycle of IC. 21/50 pts
ECOG performance status 0, 60% modified Charlson Comorbidity Index $ 1, (42%) had , 50% response. A total of 48 pts received cycle 2 of IC. After the 2nd
60% .10 pack-years, 20% larynx, 60% p16+ OPX, 50% T3-4 and 80% N2-3 cycle, a total of 39/50 pts (78%) had $ 50% response & received CRT. No
disease. All 10 patients had $ 2 contraindications to cisplatin. All 10 patients difference in response was seen with addition of AT-101. 11/50 pts (22%) had
completed RT and were evaluable. 8 of 10 patients received all 7 doses of , 50% response- 8 had laryngectomy & 3 refused. Following CRT, 10 pts
MEDI4736 and 1 patient is still on MEDI4736 after 6 doses. 1 patient re- underwent SL, 2 for residual disease & 8 late SL. The organ preservation rate was
ceived 2 doses then discontinued due to AE (diarrhea possibly related to 31/50 (62%). The most common grade 3/4 toxicities associated with IC were
MEDI4736). No DLTs were observed. No grade 4-5 AEs were observed. nausea 9%, neutropenia 9%, & infection 7%. 1 pt died of CNS hemorrhage
Grade $ 3 AEs possibly related to MEDI4736 were: diarrhea (n=1), nausea (1), unrelated to therapy & 1 from sepsis. The median follow-up time is 28 months.
and vomiting (1). No grade $ 3 AEs were rated as definitely or probably related The 2 yr laryngectomy-free survival is 54% (95% CI:38-68) & 2 yr overall survival
to MEDI4736. Conclusions: MEDI4736 is safe and feasible to administer is 81% (95% CI: 64-90). Conclusions: AT-101 did not improve responses to P.
concurrently with RT for patients with HNSCC with a contraindication to Treatment with 2 cycles of IC with PT produced similar response rates to our
cisplatin. Clinical trial information: NCT03258554. institutional controls, but organ preservation was somewhat less following
treatment with weekly P + RT. Toxicities were overall improved with this
treatment strategy. Clinical trial information: NCT01633541.

6067 Poster Session (Board #56), Sat, 1:15 PM-4:15 PM 6068 Poster Session (Board #57), Sat, 1:15 PM-4:15 PM
Induction chemotherapy with and without erlotinib in patients with oral cavity Impact of smoking cessation in locally advanced head and neck cancers
squamous cell carcinomas (OCSCCs) amenable for surgical resection. First undergoing radiation. First Author: Ruth Lauren Sacks, MD Anderson,
Author: Xiuning Le, The University of Texas MD Anderson Cancer Center, Houston, TX
Houston, TX
Background: Multiple studies have highlighted the negative outcomes asso-
Background: We have previously demonstrated activity of erlotinib in head and ciated with smoking during radiation (XRT) for locally advanced head and neck
neck SCCs as monotherapy prior to surgical resection, or in combination with cancer. However, there has been little research investigating the potential
chemotherapy for recurrent/metastatic disease (William et al. ASCO 2011, benefit of smoking cessation prior to XRT and the effect on response rates,
2017). The aim of this study was to evaluate the efficacy of induction che- relapse, distant metastases, secondary malignancies, and overall survival.
motherapy with a platinum-taxane regimen and explore the potential benefit of Methods: We reviewed 680 patients at the University of Texas MD Anderson
erlotinib as part of induction therapy in patients with resectable OCSCCs. Cancer Center from 2005-2012 with locally advanced head and neck cancer
Methods: This was a randomized, placebo-controlled, phase II trial of in- undergoing XRT. 127 were referred to the Tobacco Treatment Program (TTP)
duction chemotherapy (cisplatin 75 mg/m2 or carboplatin AUC 6 with based on provider referrals, self-referrals, or screening. Of those referred and
docetaxel 75 mg/m2 every 3 weeks for 3 cycles) with erlotinib (150mg oral retrospectively reviewed, 89 were identified as current smokers and 41 of them
daily) or placebo in patients with OCSCCs stage III-IVB amenable for surgical participated in the TTP for smoking cessation. Among these 89 patients, 50
resection. The primary endpoint was major pathological response (MPR, de- patients (18 participated in the TTP) quit smoking prior to XRT and 29 patients
fined as , 10% viable tumor cells in the surgical specimen). Secondary (19 participated in the TTP) continued to smoke, which are referred to as
endpoints included safety and long-term efficacy outcomes. Results: From Quitters and Smokers, respectively. 10 patients (2 participated in the TTP) had
April 1, 2014, to June 7, 2017, 52 patients were enrolled, of whom 47 incomplete data and were excluded from further analysis. Results: Quitters
underwent planned surgery. MPR was achieved in 7/23 (30%) in the erlotinib had 100% complete response (CR) on initial assessment following XRT. 7/50
group and 10/24 (41%) in the placebo group. With a median follow up of (14%) developed relapsed disease with 4 local recurrences (LR) and 3 distant
26.5 months, the 2-year long-term progression-fee survival (PFS) were esti- metastases (DM). 6/50 (12%) developed secondary malignancies. By contrast,
mated at 75% (95% CI: 59.5-94.5) in the erlotinib arm, and 58.6% (95% CI: Smokers had 96.5% CR on initial assessment following XRT. 8/29 (27.5%)
40.9-84.1) in the placebo arm, and 2-year overall survival at 73.5% (95% CI: developed relapsed disease with 6 LR and 2 DM. 6/29 (20.6%) developed
57.2-94.5) for the erlotinib group and 73.1% (95% CI: 55.9-95.6) for the secondary malignancies. The median follow ups for Quitters and Smokers were
placebo group. In patients who achieved MPR (n = 17), the 2-year PFS was 57.5 and 54 months with overall survival rates of 82% and 79%, respectively.
77.4% (95% CI: 57.3-100), compared to 64.5% (95% CI: 49.0-84.8) in Conclusions: Current smokers that achieved smoking cessation prior to XRT
patients who did not achieve MPR (n = 29, p = .16). All 7 patients in the demonstrated lower rates of relapse, DM, and secondary malignancies com-
erlotinib group who achieved MPR remained disease-free. The majority of pared to those that continued to smoke. Thus, smoking cessation is an integral
patients (87%) completed all 3 cycles of induction chemotherapy. The part of head and neck cancer treatment and needs to be further incorporated
common side effects were expected and distributed similarly between erlotinib in cancer care to improve cancer treatment outcomes. As a future direction,
and placebo groups. As expected, rash, diarrhea and dehydration were more a comparable group of patients who did not smoke from the same time range
common in the erlotinib group. Conclusions: Platinum and docetaxel-based will be compared for response rates, LR, DM, secondary malignancies, and
induction chemotherapy induced major pathological response in 17/47 (36%) survival.
of resectable OCSSC patients. Two-year overall survival was 73%. Responders
had improved long-term outcome. Addition of erlotinib did not improve the rate
of MPR, but might have contributed to improved PFS. Clinical trial information:
NCT01927744.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Head and Neck Cancer 359s

6069 Poster Session (Board #58), Sat, 1:15 PM-4:15 PM 6070 Poster Session (Board #59), Sat, 1:15 PM-4:15 PM
Hafnium oxide nanoparticles NBTXR3 activated by radiotherapy as a new Safety of nivolumab and ipilimumab in combination with radiotherapy in
therapeutic option for elderly/frail HNSCC patients. First Author: Christophe patients with locally advanced squamous cell carcinoma of the head and
Le Tourneau, Institut Curie, Paris, France neck (LA SCCHN). First Author: Jennifer Maria Johnson, Thomas Jefferson
University Hospital, Philadelphia, PA
Background: New therapeutic approaches are needed for elderly or frail head
and neck squamous cell carcinoma (HNSCC) patients (pts) ineligible for Background: Standard cisplatin-based chemoradiotherapy for LA SCCHN is
standard of care treatment. NBTXR3, a crystalline solution of hafnium oxide associated with toxicity and less than desirable survival in high-risk patients
nanoparticles may represent such an option. Injected intratumorally, NBTXR3 (pts). The combination of the anti-PD1 antibody nivolumab (nivo) and the
enters tumor cells and yields an increased cell-localized energy deposit upon anti-CTLA-4 antibody ipilimumab (ipi) has been effective in melanoma and
exposure to radiotherapy (RT), leading to increased tumor cell death compared renal cell carcinoma and is under investigation in other solid tumors in-
to the same dose of RT alone. Methods: Phase I study of NBTXR3 activated by cluding SCCHN. Combining RT and immunotherapy (IO) has a strong
RT in pts $70 years old or $65 years old and unable to receive cisplatin, preclinical rationale and is being evaluated in the clinic. This pilot trial
eligible for exclusive RT with stage III or IV HNSCC of the oral cavity or combines IO and RT to build upon these observations. Methods: We enrolled
oropharynx [NCT01946867]. A 3+3 dose escalation design was implemented previously untreated high-risk pts with AJCC 7th edition stage IVA-IVB
with dose levels corresponding to 5%, 10%, 15% and 22% of baseline tumor SCCHN of the oral cavity, oropharynx (OP), hypopharynx, larynx. HPV+
volume, followed by an expansion phase. Pts received an intratumoral (IT) OP tumors were T4, or N2c or N3 by AJCC 7th edition criteria. Nivo 3 mg/kg
injection of NBTXR3 and intensity modulated RT (IMRT; 70 Gy/35 fractions/7 was administered every 2 weeks IV x 17 doses and ipi 1 mg/kg every 6 weeks x
weeks). Determination of Recommended Phase 2 Dose (RP2D) and Dose 6 doses beginning 2 weeks before IMRT 2 Gy/fraction/day given to total dose
Limiting Toxicities (DLT) were primary endpoints of phase I. Absence of 70 Gy. The primary safety endpoint was acute in-field toxicity. Exploratory
NBTXR3 leakage and preliminary efficacy using RECIST 1.1 principles were correlative studies include tumor PD-L1 expression, tumor immune bias, and
also evaluated. Results: The dose-escalation is complete. Nineteen pts were exosome quantity and composition. The total sample size is 24 pts with 12
enrolled: 3 at 5%, 3 at 10%; 5 at 15% and 8 at 22% with no observed DLT or enrolled in the first stage and 12 in the expansion cohort. Results: A planned
SAE related to NBTXR3 or IT injection. One grade 1 NBTXR3-related AE safety analysis was performed in the first 12 pts (8 OP; 1 hypopharynx; 3
(asthenia at 22%) and four IT injection-related AE (grade 2 oral pain; grade 1 larynx). No acute grade (G) 4 or dose-limiting toxicities were seen. Acute G 3
tumor hemorrhage; grade 1 asthenia, and grade 1 injection site hemorrhage) in-field toxicity occurred during RT in 50% of pts: dysphagia (4 pts),
were reported. RT-related toxicity was as expected with IMRT. RP2D has been mucositis (3), odynophagia (2), hoarseness (1), dermatitis (1). 50% of pts
determined to be 22%. CT-scan assessment between 24h and 7 weeks post-IT had immune-related toxicity at any time. 3 pts discontinued treatment at .
injection demonstrated absence of NBTXR3 leakage in the surrounding tis- 3 months post RT: 1 for an immune-related cause (G 3 colitis that resolved
sues. Among 13 evaluable pts treated at doses $10%, 9 achieved a complete with steroids), and 2 for non-immune-related causes (1 G 5 bleeding due to
response of the injected lesion. Conclusions: These results show that NBTXR3 carotid rupture secondary to an in-field ulcer at 4 months post RT in a pt with
activated by RT is safe and well tolerated at all doses with preliminary en- complete response and 1 G 3 OP ulcer resolving with hyperbaric oxygen). 1 pt
couraging efficacy results. It thus represents a promising future treatment for with a laryngeal primary developed a solitary lung metastasis vs new primary
frail and elderly pts with locally advanced HNSCC with limited therapeutic 6 months post RT. 10 of 12 pts are alive with no evidence of disease (follow
options. Expansion phase has started at the RP2D. Clinical trial information: up 7.2-18.4 months). Conclusions: Preliminary results with this non-
NCT01946867. platinum-containing RT plus IO regimen are encouraging. Longer follow-
up is needed for assessment of late effects and efficacy. Enrollment is
ongoing in the expansion cohort. This study is supported by BMS. Clinical
trial information: NCT03162731.

6071 Poster Session (Board #60), Sat, 1:15 PM-4:15 PM 6072 Poster Session (Board #61), Sat, 1:15 PM-4:15 PM
Long-term survival of adjuvant high-dose (HDC) vs weekly cisplatin (WC) for Phase II study: Induction chemotherapy and transoral surgery as definitive
human papilloma-virus (HPV) and non-HPV head and neck squamous cell treatment (Tx) for locally advanced oropharyngeal squamous cell carcinoma
carcinoma (HNSCC). First Author: Katharine Andress Rowe Price, Mayo (OPSCC)—An update and retrospective review of non-study patients. First
Clinic, Rochester, MN Author: Robert S. Siegel, George Washington University School of Medicine,
Washington, DC
Background: Phase III data suggests a benefit of HDC in the adjuvant setting, but
the effect of HDC and WC on long term survival and for HPV+ HNSCC is unknown. Background: The standard of care for OPSCC includes chemoradiation (CRT)
Methods: Data from a published retrospective study (Geiger Oral Onc 2013) of or surgery with adjuvant radiation (RT). However, RT is associated with
HDC vs WC in resected HNSCC was updated. Overall survival (OS) and recurrence- significant life long morbidity. We assessed the efficacy of a two-drug in-
free survival (RFS) was assessed by Kaplan-Meier method for all pts and by HPV duction regimen, followed by transoral robotic assisted surgery (TORS) &
status. Multivariate analyses were performed to assess impact of HPV status, neck dissection for locally advanced OPSCC. Methods: This is an IRB ap-
smoking, age, HDC vs WC, and cumulative cisplatin dose ( , 200mg/m2 proved single-arm phase II study for untreated stage III or IVA (AJCC 7th
vs $200 mg/m2). Results: 51 patients (pts) received HDC and 53 WC. Median edition) OPSCC patients (pts) with an ECOG , 2 and GFR . 50 cc. Induction
follow-up was 8.7 yrs (0.8-13.7). For the whole cohort, HDC had significantly cisplatin 75 mg/m2 and docetaxel 75 mg/m2 was administered every
improved OS over WC (p = 0.0095; 5- and 10-yr OS 84% and 80% vs 72% and 21 days for 3 cycles. Patients then underwent TORS and neck dissection(s).
60%). No OS benefit for HDC was seen in pts with HPV+HNSCC (5- and 10-yr OS At post-op visits, flexible laryngoscopy, blood tests, and imaging with PET/CT
90% and 87% for HDC and 81% and 81% for WC; p = 0.51). For HPV-negative and/or MRI were done. Short and long term toxicity, progression-free sur-
HNSCC, OS had borderline significance with HDC vs WC (5- and 10-yr OS 73% vival, overall survival, and quality of life (QOL) were evaluated in all pts.
and 68% vs 65% and 44%; p = 0.06). For the whole cohort, there was no Results: Twenty oropharyngeal pts were treated, 19 were male, 17 were
difference in 5- and 10-yr RFS (78% and 74% for HDC vs 72% and 62% for WC; Caucasian, and 19 were HPV+. Median age at dx was 57. Three pts were
p = 0.32). When analyzed by HPV status, there was no difference in RFS with HDC stage III, and 17 were stage IVA. Pathologic CR at the primary site occurred
or WC for either HPV+ (p = 0.43) or HPV-negative HNSCC (p = 0.97). On in 15 pts and CR among LN neck dissections occurred in 13 pts. Four pts
multivariate analyses of OS for all pts, only HPV status was significant were given dose-reduced chemo and 1 pt was changed to carboplatin per
(p = 0.0011; HR 0.27, CI 0.12-0.62). For HPV+ HNSCC, there was no
protocol because of renal dysfunction. Pre vs post tx QOL scores did not
significant predictor of OS. For non-HPV HNSCC, the benefit of HDC
change. At a mean follow-up of 33 months (range 19.6 to 44.1), 18 pts are
approached significance with a decreased risk of death (HR 0.38; p = 0.07).
alive and NED. Three pts recurred a mean of 2.2 mos after surgery, and were
For all pts, those who received $200mg/m2 had significantly improved OS
treated with salvage CRT. Two pts died of metastatic disease, the third is
(5-yr 90% vs 72% and 10-yr 86% vs 61%; p = 0.004). By HPV status,
cumulative dose had no significant effect on OS. Conclusions: OS is better
alive and well. All 3 pts had positive LN (9 LN, 3 LN and 1 LN) at surgery. A
with HDC and with cumulative dose . 200 mg/m2 in unselected patients. The fourth pt had 12 pos LN and received radiation. He has not recurred. A
benefit of cisplatin is likely higher for non-HPV HNSCC. A difference in OS with retrospective review of an additional 20 twenty pts treated in the same way,
no difference in RFS suggests non cancer-related causes of death in the WC were also reviewed for efficacy. Mean age was 61.5. Two pts died of met-
cohort. Ability to receive HDC could be a surrogate marker of comorbidity. astatic disease. Fourteen pts have been followed for . 7 mos, and their mean
overall survival is 44 mos. Conclusions: Cisplatin + docetaxel followed by
Original study cohort characteristics.
TORS & neck dissection(s) appears to be an effective model for the definitive
HDC WC p-value
treatment for OPSCC, while avoiding the adverse effects of RT.
Median age 53 61 , 0.01
Male 90% 83% 0.28
Pack-yr smoking 6 15 0.04
HPV+ 63% 42% 0.03
T, N-stage, primary site NS

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
360s Head and Neck Cancer

6073 Poster Session (Board #62), Sat, 1:15 PM-4:15 PM 6074 Poster Session (Board #63), Sat, 1:15 PM-4:15 PM
Safety and disease control achieved with the addition of nivolumab (Nivo) to Is there a benefit of adding surveillance imaging to frequent history and
chemoradiotherapy (CRT) for intermediate (IR) and high-risk (HR) local- physical exams in patients treated definitively for head and neck squamous
regionally advanced head and neck squamous cell carcinoma (HNSCC): cell carcinoma? First Author: Jeffrey Chi, North Well Health, Lake Success,
RTOG Foundation 3504. First Author: Maura L. Gillison, The University of NY
Texas MD Anderson Cancer Center, Houston, TX
Background: In head and neck squamous cell carcinoma (HNSCC) patients
Background: Nivo, which inhibits the programmed death-1 (PD-1) receptor, (pts) who completed curative-intent definitive treatment (tx), close sur-
improved survival for pts with platinum-refractory recurrent/metastatic HNSCC. veillance is important. Across all centers, pts are closely monitored for
A clinical trial evaluated the safety of adding nivo to 4 standard intensity symptoms and undergo frequent dedicated head and neck evaluation. Role
modulated (chemo) radiotherapy (RT) regimens (see table) for pts with newly of surveillance imaging after the initial 12 week post-treatment PET/CT
diagnosed IR/HR HNSCC. Primary endpoint was safety and feasibility. however is less clear. Our institutional practice is to follow pts with regular
Methods: Eligibility included IR (p16+ oropharynx [op], T1-2N2b-N3/T3-4N0- interval imaging for two years after treatment. However this carries a fi-
3, .10 pack-years [pys], or T4N0-N3/T1-3N3, #10 pys) & HR HNSCC (oral nancial cost, and risk for false positives and unnecessary biopsies.
cavity, larynx, hypopharynx, p16- op, T1-2N2a-N3/T3-4N0-3). 10 pts/arm (8 Methods: This is a retrospective chart review of pts treated definitively for
evaluable; 0-2/8 DLTs acceptable). Nivo (dose & schedule varied per arm) HNSCC at our institution from 2012 to 2016. Pts who had a biopsy (bx) post-
started 2 wks pre-RT & continued 3 months post-RT. Feasibility of adjuvant nivo tx due to suspicion for recurrence were included. Pts belonged to 3 groups: In
months 3-12 post-RT defined as $4 of 8 pts/arm received 7 doses. Arm 4 the first group (A), biopsy was prompted by findings on surveillance imaging
limited to age $70, Zubrod performance status (PS) 2, CrCl ,50 ml/min, (SI); in the second (B), biopsy was prompted by symptom triggered imaging
grade $2 hearing loss or $ grade 3 neuropathy. Results: Characteristics of 39/ (STI) and in the third (C), biopsy was based on physical exam (PE). We
40 treated pts: median age 62, 79% male, 49% PS0, 38% HR, 67% .10 pys, recorded the aggregate results of bx in each group and calculated the positive
62% p16+ op, 72% T3-4, 85% N2-3. Grade $3 nivo-related AEs: adrenal predictive value (PPV) for each. Results: Of 353 HNSCC pts, 66 underwent
insufficiency, diarrhea-3, anemia, fatigue-2, mucositis-3, nausea, vomiting, post-tx bx for suspected recurrence of which 46 were positive. Of the 30 pts
lipase increase-6, amylase increase-2, lymphocyte/neutrophil/WBC decrease-4, in group A, 21 had positive bx (PPV = 70%). Within this group, PPV was
hyponatremia-3, anorexia, maculo-papular rash. SAE in 4/10, 4/9, 5/10 & 4/ highest with PET/CT (81.82%) followed by magnetic resonance imaging
10. DLTs, adjuvant chemo feasibility, median follow-up (mo), progression or (66.67%) and CT (62.5%). 20 out of 20 pts in group B had bx-proven
death events per arm shown in table. Conclusions: Nivo concomitant with all recurrence (PPV = 100%). 27 out of 36 pts in group C had positive bx (PPV =
(chemo)RT regimens was safe for patients with newly diagnosed IR/HR HNSCC 75%). While there was no overlap between groups A and B, there was some
but adjuvant nivo was infeasible after high-dose cisplatin or in cisplatin- overlap between groups A and C; and B and C. 45.45% of all recurrences
ineligible patients (NCT02764593). Preliminary data on progression/death were captured because of SI. When both imaging and PE conducted were
is provided. Acknowledgements: Support for this study was provided by positive simultaneously, 54.35% of recurrences were detected first by PE
Bristol-Myers Squibb Company. Clinical trial information: NCT02764593. and 45.65% by imaging. Conclusions: Bx triggered by STI has the highest
Arm Regimen DLT (type) Feasibility Follow-up Events
PPV. SI has the lowest PPV, but 45.65% of recurrences were diagnosed
because of SI alone. Our study suggests that routine SI for at least two years
1 cisplatin 40mg/m2/wk x 7 0/8 6/8 17 0/10
2 cisplatin 100mg/m2 0/8 3/8 16 1/9 post treatment for HNSCC patients may add to the surveillance value of
q21 days x 3 frequent PE but larger studies are needed to determine the optimal fre-
3 cetuximab 400 mg/m2 load, 250 mg/m2/wk x 7 1/8 4/8 10 1/10 quency and type of SI modality.
(mucositis)
4 RT alone 2/8 2/7 6 3/10
(lipase/mucositis;
fatigue)

6075 Poster Session (Board #64), Sat, 1:15 PM-4:15 PM 6076 Poster Session (Board #65), Sat, 1:15 PM-4:15 PM
Cost-effectiveness analysis of chemoradiation compared to radiation alone in Analysis of hydration and antiemetics policies in preventing cisplatin-related
the treatment of nonmetastatic oropharyngeal cancer. First Author: Husam gastrointestinal and renal toxicities in low-risk human papillomavirus
Albarmawi, University of Maryland School of Pharmacy, Baltimore, MD positive-oropharyngeal cancer (HPV+OPC) patients undergoing chemoradia-
tion in De-ESCALaTE trial. First Author: Anthony Hee Kong, University of
Background: Using concurrent chemoradiation (CRT) to treat oropharyngeal
cancer (OPC) has increased since 2000. However, there is limited information
Birmingham, Birmingham, United Kingdom
regarding the cost-effectiveness of CRT compared to radiation alone (RT) espe- Background: The De-ESCALaTE trial confirmed the superiority of cisplatin
cially given the approval of cetuximab (cetux) in 2006. We conducted a cost- over cetuximab in combination with radiotherapy for the treatment of low risk
effectiveness analysis of 1) platinum-based CRT compared to RT and 2) cetux- human papillomavirus positive oropharyngeal cancer (HPV+OPC). The most
imab plus RT (cetux+RT) compared to RT to determine the value of CRT over time. common serious adverse events (SAEs) for cisplatin were due to vomiting and
Methods: In this retrospective cohort study, we identified non-metastatic OPC nausea, in contrast with oral mucositis and vomiting for concurrent cetuximab.
patients aged 66 years or older diagnosed between 2000-2011 using the linked
In this study, we examined the efficacy of different hydration and anti-emetic
Surveillance, Epidemiology and End Results -Medicare dataset. We defined two
cohorts based on the diagnosis period: 2000-2005 (Cohort I) and 2006-2011 policies in preventing cisplatin-related gastrointestinal and renal toxicities as
(Cohort II). Cetux+RT was identified in Cohort II only. We matched the platinum- well as related SAEs in the cisplatin arm of the De-ESCALaTE trial.
based CRT and cetux+RT groups to the RT groups using propensity score models Methods: This was a post-hoc pre-specified analysis of data collected within
that included age, race, marital status, income, Charlson Comorbidity Index and the De-ESCALaTE trial including pre-hydration, diuretics, the amount of in-
stage at diagnosis. The outcomes were incremental cost, incremental life-year travenous (IV) fluids before, during and after chemotherapy, whether oral fluid
gained (LYG) and incremental cost-effectiveness ratio (ICER) during the 3 years hydration was advised and type of antiemetic regimen prescribed, if any, after
after diagnosis. Costs were estimated from the Medicare perspective and using chemotherapy administration, including if a triple antiemetic regimen with a
2017 USD. Results: 2,646 OPC patients were eligible for the study. The esti- NK1 receptor antagonist, steroids and a serotonin 5-HT3 antagonist was given
mated parameters with the corresponding 95% confidence intervals (CI) are before and after chemotherapy. The primary outcome was number of SAEs per
shown in the table. Conclusions: From 2000-2005, platinum-based CRT was a patient; secondary outcome was number per patient of cisplatin-induced
cost-effective option compared to RT. From 2006-2011 and compared to RT, severe toxicity events of interest: nausea, vomiting, dehydration or renal
platinum-based CRT provided a survival benefit at higher costs while cetux+RT toxicity. Results: 166 (mean age 57 yrs; 132 m, 34 f) patients received
patients incurred higher costs with no survival benefit. cisplatin. Hydration and anti-emetics policies for cisplatin treatment are
Incremental Incremental significantly correlated with the rate of SAEs and acute severe nausea,
Cohort / cost life-year gained Incremental cost-effectiveness ratio vomiting, dehydration or renal toxicities. Using stepwise backwards multi-
Comparison groups (95% CI) (95% CI) (95% CI) variable ordinal logistic regression in the presence of baseline characteristics,
I: Diagnosis year use of a triple anti-emetics regimen (OR 0.41,p = 0.032) and 2.5 to 3L IV
2000-2005 fluids given before and during cisplatin chemotherapy (OR 0.161, p = 0.009)
Platinum-based $18,097 0.26 $70,318 / LYG
CRT vs. RT (6,012 to (0.09 to 0.43) (21,559 to 213,994)
as well as oral fluids advised post chemotherapy (OR 0.365, p = 0.03) were
31,051) associated with a significantly lower incidence of SAEs and severe toxicities of
II: Diagnosis year interest. We will also present data on relative cost-effectiveness of the different
2006-2011 regimens. Conclusions: Based on our results, we recommend the use of a triple
Platinum-based $23,521 0.16 $148,224 / LYG
CRT vs. RT (9,734 to (0.01 to 0.31) (34,470 to 1,284,469) anti-emetic regimen, adequate hydration of 2.5-3L before and during che-
36,332) motherapy as well as advising patients to take oral fluids advised to reduce
Cetux+RT vs. RT $31,869 -0.13 Cetux+RT is more expensive with no cisplatin toxicities related to nausea, vomiting, dehydration and/or renal injury.
(17,694 to (-0.30 to 0.05) survival benefit (dominated) Clinical trial information: ISRCTN33522080.
44,959)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Head and Neck Cancer 361s

6077 Poster Session (Board #66), Sat, 1:15 PM-4:15 PM 6078 Poster Session (Board #67), Sat, 1:15 PM-4:15 PM
Randomized phase II study with or without induction chemotherapy com- Impact of antiviral prophylaxis in HSV positive patients treated with con-
bined with accelerated high dose radiotherapy and cetuximab in locally current chemoradiotherapy for head and neck cancer. First Author: Nathalie
advanced unresectable HPV positive squamous cell carcinoma of the head Letarte, Centre de recherche du Centre hospitalier de l’Université de
and neck. First Author: Signe Friesland, Dep. of Oncology, Karolinska In- Montréal (CRCHUM), Montreal, QC, Canada
stitute, Stockholm, Sweden
Background: Chemoradiotherapy used for the treatment of locally advanced
Background: Concomitant chemoradiotherapy (CT/RT) is the standard treat- head and neck cancer (HNC) causes a high incidence of mucositis that may
ment for locally advanced non- resectable squamous cell carcinoma of the head be accentuated by a reactivation of herpes simplex virus (HSV). To date, no
and neck. Acute side effects can be serious and late effects important. Patients study has evaluated the impact of antivirals used as prophylaxis to prevent
with HPV+ tumors carry a better prognosis than other patients. De-escalation mucositis or their severity. Methods: This is a retrospective observational
studies are explored with respect to RT, dose, fractionation, target volumes, study including patients who received at least one cycle of concurrent
adjuvant pharmacotherapy, with encouraging results. However, several patients chemoradiotherapy for the treatment of head and neck cancer between
still recur locoregionally, also in high dose RT volumes. Some patients have January 2014 and June 2017 at the Centre hospitalier de l’Université de
distant metastases, often with massive tumor burden and late during follow up Montréal (CHUM). HSV negative patients were excluded. After approval by
(FU). Purpose of the study: 1.To evaluate the effect of induction chemotherapy the IRB, we compared the incidence and severity of mucositis in HSV
(IC) on (a) progression free survival (b) distant metastases as first site of failure (c) positive patients who started an antiviral prophylaxis before cycle 1 or 2
locoregional failure (d) pattern of tumor response, spatial and timely, exploring (prophylaxis group) to HSV+/unknown HSV patients who did not receive
the possibility to reduce RT dose and/or target volume(s) in future protocols. 2. to antiviral prophylaxis (control group). Emergency visits and hospitalizations
address the impact of high dose RT for bulky disease, T3/T4 given concurrent related to mucositis were collected. Mucositis were assessed regularly by
with cetuximab (E). Methods: Patients had previously untreated stage III/IV radiation oncologists during the treatment. Results: Of 482 patients who
(.80 % st IV; TNM 7th), M0 disease, WHO 0-1, unresectable squamous cell received concurrent chemoradiotherapy for HNC, 75 were HSV negative and
carcinoma of the head and neck, HPV positive as evaluated with p16 and/or 407 were included in this study. In the group with (n = 94) and without
PCR. Pts were randomised between 2 arms. Pts in arm A were scheduled for 2 prophylaxis (n = 313), patients received carboplatin and 5-FU (77% vs 62%)
cycles of TPF : T (docetaxel) 75 mg/m2 day 1, P (cisplatin) 75 mg/m2 day 1 and and cisplatin (23% vs 38%) with concurrent radiation respectively. The rate
F (fluorouracil) 1000 mg/m2 over 24 hours day 1-4. RT was delivered with IMRT of all grade mucositis in patients with and without prophylaxis (99% vs 96%;
to all pts : 68 Gy in 6 weeks for T1/T2 tumors, 76 Gy in 6.5 weeks for T3/T4 p = 0.19) was not statistically significant. The rate of grade 3 and 4 mucositis
tumors with E given one week before and weekly during RT. Tumor response was (42% vs 49%; p = 0.29), the rate of emergency visit (29% vs 28%; p = 0.91)
evaluated according to RECIST with CT, MRI or PET/CT after IC, at 6-8 weeks, 1 and hospitalization (9% vs 8%; p = 0.80) were not statistically significant
and 2 years FU. Results: From january 2011 to february 2016, 152 consecutive between each group. However, in a subgroup of patient receiving carboplatin
pts were enrolled, 77 in arm A. All pts had oropharyngeal cancer. In arm A, 7 pts and 5-FU, antiviral prophylaxis seems to decrease significantly the rate of
had CR after TPF, 19 had PR out of 36 evaluable pts. At 2 years FU 70/77 pts grade 3 (49% vs 63%; p = 0.04). Conclusions: The addition of antiviral
(91%) were alive in arm A, 69/75 (92%) in arm B. Distant metastases as first site prophylaxis in HSV positive in patients undergoing concurrent chemo-
of failure was 3 (3.9%) in arm A and 7 (9.3%) in arm B. Adverse events grade 3- radiotherapy for locally advanced HNC didn’t decrease the rate of all grade
4, ever registered, were seen in 71 pts in arm A and 63 in arm B, were transient, mucositis. In the subgroup of patients receiving carboplatin and 5-FU mainly
most often related to RT. Conclusions: Survival and locoregional control at 2 of oropharynx origin, HSV prophylaxis decreased the severity of mucositis.
years was high and similar in both arms. Distant metastases as first site of failure
was more than doubled in arm B, not having induction chemotherapy (IC).
Clinical trial information: EudraCT number: 2009-013438-26.

6079 Poster Session (Board #68), Sat, 1:15 PM-4:15 PM 6080 Poster Session (Board #69), Sat, 1:15 PM-4:15 PM
A retrospective analysis of cisplatin dosing strategies when used with Outcomes of postoperative treatment with concurrent chemoradiotherapy
radiation on outcomes in head and neck squamous cell carcinoma of the (CRT) in high risk resected oral cavity squamous cell carcinoma (OCSCC): A
oropharynx. First Author: Vatche Tchekmedyian, Memorial Sloan Kettering multi-institutional collaboration. First Author: Jessica Lyn Geiger, Cleveland
Cancer Center, NY, NY Clinic, Cleveland, OH
Background: Cis is the gold standard radiosensitizer for CRT to treat head Background: Adjuvant CRT with high-dose cisplatin remains standard treat-
and neck cancer. Prospective trials have required that cis be administered ment for OCSCC with high risk pathologic features of positive surgical margins
early in the week (Mon/Tues) to optimize radiosensitization without evidence (SM+) and/or extranodal extension (ENE). High-dose cisplatin is associated
to support this practice. This retrospective analysis considers the impact of with significant toxicities, and alternative dosing schedules or treatments are
cis day of week (DOW) administration and other variables on OPSCC pt used. We evaluated outcomes associated with different systemic therapies
outcomes. Methods: We reviewed OPSCC cases treated with primary CRT at concurrent with RT and the effect of cumulative dosing of cisplatin.
our center. Pts treated with non-cis or induction chemotherapy were ex- Methods: An IRB-approved collaborative database of patients (pts) with pri-
cluded. Data collected includes age, DOW (Mon/Tues vs Wed/Thurs/Fri), mary OCSCC (Stage I-IVB AJCC 7th edition) treated with primary surgical
smoking status, total dose (TD) of cis (#200mg/m2 vs . 200mg/m2), single resection between 1/1/2005 and 1/1/2015 with or without adjuvant therapy
dose (SiD) [100mg/m2 x1 day] vs split dose (SpD) [50mg/m2 x2 days] was established from 6 academic institutions. Pts were categorized by sys-
administration, T stage (0-2b vs 2c-3), N stage (0-2b vs 2c-3), overall temic therapy received, and resultant groups compared for demographic data,
survival (OS) (from start of RT), local/regional/distant failure, KPS and HPV/ pathologic features, and outcomes by t-test and Chi-squared tests. Kaplan-
p16 status. Univariate Cox proportional hazards regression was used to Meier curves, log-rank p-values, and multivariate analysis (MVA) for disease
analyze OS and multivariate Cox proportional hazard model investigated the free survival (DFS) and freedom from metastatic disease (DM). Results: From a
relationships between OS and cis dosing variables while controlling for other total sample size of 1282 pts, 196 pts were identified with high risk features
factors. Results: 745 pts with OPSCC were treated with CRT from 7/31/ (SM+, ENE) who were treated with adjuvant CRT. Median age was 56 years,
2001-2/7/2014. 459 used cis based regimens and were included. Median 63.3% of pts were men, 81.1% were Caucasian, 70.9% had significant to-
age at start of RT was 55. 311 pts (67.8%) received SpD, 124 (27%) re- bacco history. 35.7% of pts had SM+, 82.7% ENE, 65.3% with perineural
ceived SiD, 8 (1.7%) received weekly cis and 16 (3.5) received mixed cycles. invasion (PNI), 49% had lymphovascular space invasion (LVSI). There was a
269 (58.6%) received #200mg/m2. 232 (50.5%) were HPV/p16 positive, trend associating higher cisplatin dose delivered with improved locoregional
40 (8.7%) were negative and 187 (40.7%) were unknown. Median f/u was control, DM, and overall survival (OS) (p-values 0.131, 0.084, and 0.187,
7.9yrs (1.8m -18.9yrs). There were 92 (20%) deaths, and 75 (16.4%) respectively). DFS was significantly better with higher cisplatin dose (HR =
recurrences (local/regional and distant) with 44 (9.6%) competing events. In 0.95 per 100 mg/m2 increase in cisplatin). Administration schedule of cis-
univariate analysis, age, N stage, T stage, KPS and HPV/p16 status were platin (weekly versus high-dose) was not significantly associated with DFS. On
significantly associated with OS, while DOW, TD, and SiD/SpD were not. In MVA, PNI and higher cisplatin dose remained statistically significant for DFS
multivariate analysis (MVA), none of the associations between cis dosing and (p , 0.001 and 0.007). Median OS by cisplatin dose was 10.5 ( , 200 mg/
OS were significant (although MVA was limited by low number of events and m2) vs. 20.8 months ( . / = 200 mg/m2). Conclusions: This multi-institutional
total variables included). Conclusions: This retrospective analysis suggests analysis demonstrated cumulative cisplatin dose . / = 200 mg/m2 was as-
that the DOW cis is given has no impact upon CRT outcomes for OPSCC pts. sociated with improved DFS in high risk resected OCSCC pts. It remains
SpD cis represents an alternative administration approach. unclear by this analysis if cisplatin administration schedule has any prognostic
implication. Further study is warranted to elucidate the optimal cisplatin
schedule for this population.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
362s Head and Neck Cancer

6081 Poster Session (Board #70), Sat, 1:15 PM-4:15 PM 6082 Poster Session (Board #71), Sat, 1:15 PM-4:15 PM
Influence of tumor size and Eastern Cooperative Oncology Group performance A randomized phase II study of pembrolizumab with or without radiation in
status (ECOG PS) at baseline on patient (pt) outcomes in lenvatinib-treated patients with recurrent or metastatic adenoid cystic carcinoma. First Author:
radioiodine-refractory differentiated thyroid cancer (RR-DTC). First Author: Jonathan Daniel Schoenfeld, Dana-Farber Cancer Institute, Boston, MA
Lori J. Wirth, Massachusetts General Hospital Cancer Center, Harvard Uni-
Background: Adenoid cystic carcinoma (ACC) is a salivary gland malignancy
versity, Boston, MA
characterized by a high rate of distant recurrence. Systemic therapy has
Background: In SELECT, lenvatinib significantly improved progression-free survival generally failed to produce durable benefit. Radiation (RT) is used for localized
(PFS) of pts with RR-DTC versus placebo (18.3 v 3.6 months; hazard ratio [HR]: disease and as directed treatment for metastases. Here, we report the safety and
0.21 [99% CI: 0.14, 0.31]; P,0.001). Here we examine the treatment of RR-DTC efficacy of pembrolizumab (pembro) administered with or without hypo-
with lenvatinib in relation to tumor size (sum of all targeted lesions) and ECOG PS. fractionated RT in a phase II randomized study. Methods: Eligible patients (pts)
Methods: In this post hoc analysis of SELECT with pts randomized to receive had recurrent or metastatic ACC with evidence of progressive disease (PD)
lenvatinib, Kaplan-Meier estimates of time to ECOG PS $2 were calculated for within the last 12 mos and .=1 measurable non-CNS lesion, along with 1-5
subgroups of pts according to baseline ECOG PS or tumor size. Objective response
additional lesions deemed appropriate for RT to 30 Gy in 5 fractions. Pts were
rate (ORR) and Kaplan-Meier estimates of overall survival (OS) and PFS according to
randomized to pembro alone (200 mg IV q3 weeks) or in combination with RT
ECOG PS (0 or 1) at baseline were calculated. Correlations between ECOG PS at
baseline (0 or 1) and maximum tumor shrinkage were calculated using one-way given within 7 days of cycle 1, day 1. The primary endpoint was objective
analysis of variance. Results: Pts with ECOG PS 0 or 1 at baseline had similar response rate (ORR) outside the RT field by RECIST 1.1. Using a parallel two-
demographic and disease characteristics. ORR was 78.5% and 51.0% for pts with stage design, if .=1 response out of 10 was observed in either arm, 10 more pts
ECOG PS 0 and 1 at baseline, respectively (odds ratio [95% CI]: 3.508 [2.018, would be enrolled to that arm. If .=3 responded, the null hypothesis
6.097]). Mean maximum percent decrease in tumor size was significantly greater in (ORR=5%) would be rejected in favor of a 25% ORR. Predefined secondary
pts with baseline ECOG PS 0 (-46.13%) versus pts with ECOG PS 1 (-37.16%; endpoints included progression free survival (PFS) and toxicity. Analyses of
P=0.0017). For pts with ECOG PS 1 at baseline, time to ECOG PS $2 was nu- tumor growth rate (TGR) excluding RT lesions and immune biomarkers were
merically shorter with tumor size .60 mm versus tumor size #60 mm (HR [95% CI]: exploratory. Results: Ten pts per arm were randomized into the trial’s first stage
1.450 [0.708, 2.967]). Additional results are summarized in the table. with median age 65 (45-79). No objective responses were seen. Stable disease
Conclusions: Among pts with RR-DTC, PFS, OS, ORR, and time to ECOG $2 were (SD) was observed in 13 pts; 6 had PD as best response, 1 was unevaluable.
generally better for patients with lower ECOG PS or smaller tumor size at baseline. Median PFS was 7 mos 95% CI (3 - 13 mos), with 9 pts without progression at 6
These results may indicate that it is beneficial to start lenvatinib in pts with RR-DTC mos. 3 pts remain on study treatment (range 8-11 mos). In pts with SD, TGR
early, before ECOG PS worsens and tumor size increases. Clinical trial information: decreased by .25% in 7 of 12 pts and by .75% in 4 pts. There was
NCT01321554. no difference in likelihood of SD or PFS between arms. Treatment related AEs
Baseline population Baseline ECOG PS n HR (95% CI) (TRAEs) occurred in 18 pts but there were no G3-5 TRAEs. Among 8 biop-
Time to Tumor size* # 35 mm 0 46 0.328 (0.114, 0.941) sies analyzed, PD-L1+ tumor/immune cells ranged from 12-52%.
ECOG PS ‡ 2 1 19
. 35 – # 60 mm 0 46 0.344 (0.092, 1.287) Conclusions: Pembro alone or with hypofractionated RT was well tolerated. We
. 60 – # 92 mm
1
0
25
26 0.269 (0.088, 0.820)
observed no objective responses, but 65% of pts with PD prior to study entry
1 33 achieved SD, the majority with decreased TGR, and 15% had prolonged SD.
. 92 mm 0 26 0.253 (0.051, 1.257)
1 27 Additional strategies are needed to further delay progression and effect re-
Baseline population Baseline
ECOG PS
n Median (months) HR (95% CI)
sponse. Clinical trial information: NCT03087019.
OS ECOG PS 0 or 1 0 144 NE 0.420 (0.256, 0.690)
1 104 22.1
PFS ECOG PS 0 or 1 0 144 NE 0.523 (0.353, 0.775)
1 104 12.6

*Sum of all targeted lesions at baseline.

6083 Poster Session (Board #72), Sat, 1:15 PM-4:15 PM 6084 Poster Session (Board #73), Sat, 1:15 PM-4:15 PM
NISCAHN: A phase II, multicenter nonrandomized trial aiming at evaluating A phase II trial cohort of nivolumab plus ipilimumab in patients (Pts) with
nivolumab (N) in two cohorts of patients (pts) with recurrent/metastatic recurrent/metastatic adenoid cystic carcinoma (R/M ACC). First Author:
(R/M) salivary gland carcinoma of the head and neck (SGCHN), on behalf of Vatche Tchekmedyian, Memorial Sloan Kettering Cancer Center, NY, NY
the Unicancer Head & Neck Group. First Author: Jerome Fayette, Centre
Background: R/M ACC is a malignant neoplasm most commonly of salivary
Léon Bérard, Medical Oncology, Lyon, France
gland origin with no standard treatment. The impact of combined PD-1/
Background: SGCHN are rare tumors including adenoid cystic carcinoma (ACC) CTLA-4 checkpoint blockade in R/M ACC is unknown. Methods: In a two-
and non-ACC, with no standard systemic treatment for R/M pts. We evaluated N stage minimax phase II trial, pts with progressive R/M ACC (non-salivary
monotherapy in R/M SGCHN pts. Methods: R/M SGCHN pts (ACC or non-ACC) primaries allowed) were enrolled and treated with nivolumab 3 mg/kg every
not eligible to local treatment and with centrally confirmed RECIST 1.1 disease 2 weeks plus ipilimumab 1 mg/kg every 6 weeks (1 cycle = 6 weeks). Im-
progression over the last 6 months were enrolled and received N 3 mg/kg IV, aging, using RECIST v1.1 response assessment, was scheduled to be
every 2 weeks for a maximum of 12 months (mo). Possibility was given to re-start performed approximately every 12 weeks. The primary endpoint was overall
N in case of progression during the 2-year follow-up phase. Primary endpoint response rate (ORR = complete response [CR]+partial response [PR]) per
was 6-mo non-progression Rate (NPR6m) as per RECIST 1.1. Secondary RECIST v1.1. To detect a difference between an unacceptable ORR of 5%
endpoints included ORR, PFS, OS, and safety. Considering that N would be and a desirable ORR of 20% (one-sided type I error of 10%, power of 90%),
uninteresting if NPR6m # 20% and promising if $ 40% and using a Fleming’s at least 1 in the first 18 pts required an observed response. At least 4 re-
single-stage design (a: 5% unilateral, power: 90%), at least 14 successes/42 sponses of 32 total pts were needed to meet the primary endpoint. Treatment
evaluable pts were required for each cohort to be positive. Results: 46 ACC and beyond progression of disease (PD) was allowed at the discretion of the
52 Non-ACC pts (median age 61 yrs (range 29-81), 43.9% female, 55.1% PS1 investigator. A second cohort of pts with non-ACC salivary cancer is still
and 2.0% PS2) were enrolled and received at least one dose of N. Median accruing for separate analysis. Results: From 6/12/2017-6/20/2018, 32 pts
treatment duration was 5.5 mo (ACC) and 3.3 (Non-ACC). Median FU was 10.8 were enrolled and evaluable for the primary endpoint. There was 1 confirmed
mo (ACC) and 8.3 mo (Non-ACC). 95 patients were evaluable for the primary PR in the first 18 pts, therefore enrollment of the second stage continued.
endpoint. NPR6m was 15/45 pts (33.3%, 90%CI :21.8;46.6) and 7/50 pts ORR was 6% (2/32). One additional pt had an unconfirmed PR (-31%
(14.0%, 90%CI:6.8;24.7) for ACC and non-ACC pts respectively. 4 (8.7%) regression before CNS PD). For best overall response, there were 2 PRs, 15
partial responses (PR) and 26 (56.5%) stable diseases (SD) were observed in SD, and 11 PD. Four pts never reached a first disease assessment: 3 due to
ACC cohort while 2 (3.8%) PR and 22 (42.3%) SD were observed in non-ACC. death from clinical PD and 1 was removed for toxicity. Six pts discontinued
Median PFS was 4.9 mo (95%CI = 3.4;5.6) in ACC pts and 1.8 mo (95%CI = the trial for toxicities (Grade 4 (G4) neutropenia/sepsis and G3 adrenal
1.7;3.5) in non-ACC pts. The most common related adverse events (AE) ( . 10% insufficiency (1), G2 hypophysitis (2), G3 arthritis . 7 days (1), G3 colitis
by cohort) were asthenia, hyperthyroidism, diarrhea, rash, pruritus and (1), and G3 hepatitis/G4 creatinine kinase (CK) elevation (1)). The 2 con-
hypothyroidism. 7/98 pts (7.1%) presented at least one related AE Grade 3-4 firmed PRs consisted of -73.1% and -58.4% regressions, with a duration of
(mainly hepatic) and 9 pts (9.2%) prematurely discontinued Nivolumab due therapy of 18.4 and 7.8 months, respectively (treatment ongoing for both).
to toxicity. Conclusions: Limited efficacy was observed with N in R/M SGCHN Conclusions: The study did not meet its primary endpoint, though the re-
pts. N in combination might be of interest and deserves exploration in ACC pts. sponses observed were dramatic. Paired biopsy and peripheral blood
Clinical trial information: NCT03132038. samples will be analyzed to elucidate insights into mechanisms of response
and resistance to dual checkpoint blockade. Clinical trial information:
NCT03172624.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Head and Neck Cancer 363s

6085 Poster Session (Board #74), Sat, 1:15 PM-4:15 PM 6086 Poster Session (Board #75), Sat, 1:15 PM-4:15 PM
Development and validation of a prediction-score model for distant metastases PROSPERO: A study to determine the utility of focused genomic profiling to
in major salivary gland carcinoma. First Author: Jelena Lukovic, Department of guide selection of drug therapy in salivary gland cancer. First Author: Samuel
Radiation Oncology, Princess Margaret Cancer Centre, Toronto, ON, Canada Rack, University of Manchester, Manchester, United Kingdom
Background: We developed and validated a prediction-score for distant metastases Background: For most patients with recurrent or metastatic salivary gland
(DM) in major salivary gland carcinoma (SGC). Methods: Patients with SGC treated cancer (RM-SGC), there are no standard therapies. Many patients undergo
with curative-intent surgery +/- postoperative radiation therapy (PORT) at 4 tertiary genomic profiling to guide selection of targeted therapy. The MSK-IMPACT
cancer centers were divided into discovery (institution A&B) and validation (in- study applied a 468 gene next generation sequencing (NGS) panel, identifying
stitution C&D) cohorts. Multivariable analysis using competing risk regression was actionable mutations in 34/114 patients (30%) with RM-SGC. Minimising
used to identify predictors of DM in the discovery cohort and create a prediction cost will facilitate application within publically funded healthcare systems. We
score. The optimal score cut-off for high vs low-DM risk was determined using a therefore sought to determine the utility of genomic profiling using a focused
minimal p-value approach. The results were subsequently evaluated in the validation
24 gene targeted NGS panel to identify actionable mutations in RM-SGC with a
cohort. The cumulative incidence and Kaplan-Meier methods were used to analyze
DM and overall survival (OS), respectively. Results: Overall, 1035 patients were sub-group analysis in adenoid cystic carcinoma (ACC) and non-ACC sub-types.
included (Table). In the discovery cohort, DM predictors (risk score coefficient) were: Methods: From January 2017 to 2018, 125 patients with RM-SGC provided
positive margin (0.6), pT3-4 (0.7), pN+ (0.7), lymphovascular invasion (LVI; 0.8), informed consent to an ethically approved study. Clinical and demographic
and high risk histology* (1.2). High DM-risk SGC was defined by sum of coefficients characteristics were collected. DNA was extracted from FFPE samples and
greater than 2. In the discovery cohort, the 5-year cumulative incidence of DM for analysed using Qiagen GeneRead DNAseq Targeted Panel V2 in the Man-
high vs low risk SGC was 50% vs 8%; p , 0.01; these results were similar in the chester Centre for Genomic Medicine Diagnostic Laboratory, an NHS clinically
validation cohort (44% vs 4% at 5 years; p , 0.01). In the combined cohorts, this accredited lab. A custom bioinformatic pipeline was validated to detect single
model predicted distant-only failure (40% vs 6%, p , 0.01) and late ( . 2yr post nucleotide variants and indels ( , 40bp) to 5% mutant allele frequency.
surgery) DM (22% vs 4%; p , 0.01). Patients with high DM-risk SGC had an in- Alterations were categorised following American College of Medical Genetics
creased incidence of DM in the subgroup receiving PORT (46% vs 8%; p , 0.01) or guidelines and Association for Molecular Pathology tiering. Results: DNA from
concurrent chemotherapy (71% vs 34%; p , 0.01). The 5-yr OS for high vs low risk 101 tumours (69 major, 32 minor salivary gland) was sequenced with 95%
SGC was 48% vs 92% (p , 0.01). Conclusions: This validated prediction score coverage at . 350x read depth over the target enrichment. 65 patients had
model may be used to identify SGC patients at increased risk for DM and select those adenoid cystic carcinoma (ACC) and 36 had non-ACC SGC. Median age was 55
who may benefit from prospective evaluation of treatment intensification and/or years (range 18-80). 43 actionable alterations were identified in 33 patients
surveillance strategies. Baseline characteristics.
within the following genes: TP53 (21%), PIK3CA (8%), ERBB2 (6%), PTEN
Discovery (n=619) Validation (n=416) (3%), BRAF (2%), EGFR (T790M) (1%), and AKT1 (1%). Targeted therapy
N (%) N (%) p
was selected based on genomic findings in 12% of these patients. In ACC
Median follow up, yrs 5.3 5 patients, actionable alterations were seen in 25% compared with 55% of non-
Median age, yrs 56 57 0.19
Parotid subsite 518(84) 329(79) 0.07 ACC patients (9 adenocarcinoma, 5 salivary duct carcinoma, 3 carcinoma ex
LVI 92(15) 47(11) 0.11
Positive margin 215(35) 158(38) 0.6
pleomorphic adenoma, 2 mucoepidermoid carcinoma and 1 myoepithelial
High risk histology* 372(60) 224(54) 0.04 carcinoma). Conclusions: This study identified actionable alterations in 33%
pT3-4 224(36) 129(31) 0.09
pN+ 145(23) 89(21) 0.45 of SGC patients using focused genomic profiling, demonstrating comparable
PORT 472(76) 285(69) ,0.01 utility to larger research panels. This focussed panel is being expanded to
5-yr DM (95% CI) 20% (17-24%) 14% (11-18%) 0.03
include emerging biomarkers such as NOTCH gene mutations, with NOTCH
*High risk histology: adenoid cystic carcinoma, salivary duct carcinoma, carcinosarcoma, undifferentiated car-
cinoma, grade (G) 2-3 adenocarcinoma, G2-3 mucoepidermoid carcinoma, G2-3 carcinoma ex-pleomorphic
inhibitors currently in trials in ACC. Greater access to basket studies in-
adenoma, other G3 histology. corporating therapies matched to genomic alterations will maximise the
clinical utility of this approach.

6087 Poster Session (Board #76), Sat, 1:15 PM-4:15 PM 6088 Poster Session (Board #77), Sat, 1:15 PM-4:15 PM
Genomic landscape of FNAs positive for medullary thyroid cancer (MTC) and Pilot study combining PD-L1 antibody durvalumab (D) with CTLA-4 antibody
potential impact on systemic therapy. First Author: Lori J. Wirth, tremelimumab (T) and stereotactic body radiotherapy (SBRT) to treat metastatic
Massachusetts General Hospital Cancer Center, Harvard University, Boston, anaplastic thyroid cancer (ATC). First Author: Eric Jeffrey Sherman, Memorial
MA Sloan Kettering Cancer Center, New York, NY
Background: Systemic therapies targeting specific genomic alterations in Background: ATC is a rare and aggressive cancer with very limited treatment
advanced MTC are available or under investigation. The Afirma Genomic options. The thyroid is one of the most immunogenic organs in the body and PD-
Sequencing Classifier (GSC) uses RNA sequencing to assess FNA specimens L1 is commonly expressed on ATC tumor cells and PD-1 in the inflammatory
from cytologically indeterminate thyroid nodules, which are also tested for cells in the ATC microenvironment. However, antibodies to PD-1 as single
specific molecular aberrations associated with thyroid cancer via a suite of agents have a poor record in this disease. Methods: This study evaluated the
highly accurate malignancy classifiers. This suite can be applied in- addition of T (75 mg every 4 weeks up to 4 doses) to D (1500 mg every 4 weeks).
dependently to Bethesda V/VI nodules. The Afirma Xpression Atlas (XA) is an SBRT 9Gy x 3 fractions was given within the first 2 weeks of treatment to
additional test that can be combined with Afirma GSC to report nucleotide produce an “abscopal” effect. Major inclusion criteria: Metastatic ATC; ECOG
variants and fusions across 511 cancer-associated genes. Here we report the PS 0-2; No prior immunotherapy; Last anti-cancer treatment . 7 days prior to
prevalence and genomic landscape of MTC classifier positive (MTC+) FNA starting study. Primary objective 1-year overall survival with target of $ 2 out of
samples. Methods: All Afirma GSC and malignancy classifier tests run in the 12 patients. Results: 12 patients were accrued. Male – 50%; Median PS 1;
Veracyte Clinical Laboratory between July 2017 and January 2019 were Median Age – 71 (49-82); Prior radiation to neck (75%); Prior chemotherapy
deidentified and examined for MTC+ cases. Afirma XA was run on all such (75%). MSI-High was noted in 2/11 subjects. BRAFV600E mutation in 3/12
cases, and all variants and fusions were tabulated. Results: Examination of subjects. There were 0 confirmed responses and only 1 subject with SD for 4
29,895 FNAs revealed 90 MTC cases. Of 22,793 Bethesda III cases, 32 cycles or longer. Median time on treatment was 11 weeks (1-28+ weeks). MSI
(0.14%) were MTC+. Of 5,491 Bethesda IV cases, 33 (0.60%) were MTC+. status did not affect treatment response. MSI-High patients were on treatment
Provider-ordered testing was done on an additional 16 and 9 MTC cases from before progression for 8-14 weeks. Median overall survival was 14.5 weeks with
Bethesda V and VI nodules, respectively. 58% of all MTC+ samples harbored a only one person alive past 1 year. Neither the presence of a BRAF or p53
RET variant (+/- others), 9% contained a KRAS variant (+/- others), 6% in- mutation appeared to affect either outcome. Conclusions: T/D with SBRT was
cluded an HRAS variant, 1% had a BRAF fusion, 1% demonstrated other not active in metastatic ATC. Future studies looking at other novel immuno-
fusions, and 26% held no variant/fusion. Conclusions: In our cohort, Afirma XA therapy combinations in ATC should be evaluated. Biopsies done on study are
identified a variant or fusion in 74% of MTC+ FNAs. Currently approved or being analyzed. Clinical trial information: NCT03122496.
investigational therapies exist for cancers with RET, BRAF and HRAS alter-
ations, suggesting that 64% of our series might be eligible for treatment based
on genomic information from FNA. In advanced MTC, noninvasive FNA sample
collection at the time of diagnosis may ultimately impact on targeted therapy
selection, with the option to repeat FNA testing should the disease progress.
Future studies may investigate how finding a genomic alteration by FNA can
inform the management of MTC and, in the case of progressive disease,
improve our understanding of the mechanisms of disease progression and drug
resistance.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
364s Head and Neck Cancer

TPS6089 Poster Session (Board #78a), Sat, 1:15 PM-4:15 PM TPS6090 Poster Session (Board #78b), Sat, 1:15 PM-4:15 PM
A multicenter, randomized, double-blind, placebo-controlled phase III study KEYNOTE-689: Phase 3 study of adjuvant and neoadjuvant pembrolizumab
of anlotinib or placebo in combination with gemcitabine and cisplatin (GP) combined with standard of care (SOC) in patients with resectable, locally
as first-line treatment for recurrent or metastatic nasopharyngeal carcinoma advanced head and neck squamous cell carcinoma. First Author: Ravindra
(R/M NPC). First Author: Yunpeng Yang, State Key Laboratory of Oncology in Uppaluri, Dana-Farber Cancer Institute and Brigham and Women’s Hospital,
South China; Collaborative Innovation Center for Cancer Medicine; Sun Yat- Boston, MA
sen University Cancer Center, Guangzhou, China
Background: Evidence of efficacy and pathological response at the time
Background: GP is the standard first-line chemotherapy for R/M NPC. How- of surgery was reported in two phase 2 studies (NCT02296684 and
ever, the outcome of patients who are refractory to first-line chemotherapy is NCT02641093) of preoperative pembrolizumab in patients with high-risk,
poor. There remains an unmet need for more effective first-line treatment. resectable, locally advanced (LA) head and neck squamous cell carcinoma
Overexpression of vascular endothelial growth factor (VEGF) is common in NPC (HNSCC). The randomized, open-label, phase 3 KEYNOTE-689 trial
and the higher expression is related to lower OS. This feature makes NPC (NCT03765918) will evaluate efficacy and safety of pembrolizumab as
potentially suitable for antiangiogenic treatment. Anlotinib is a novel multi- neoadjuvant and adjuvant therapy in combination with SOC (radiotherapy 6
target tyrosine kinase inhibitor that targets VEGFR 1 to 3, fibroblast growth cisplatin) in patients with previously untreated, resectable LA HNSCC.
factor receptor 1 to 4, and platelet-derived growth factor receptor a and b. Our Methods: Patients with newly diagnosed LA HNSCC will be randomly assigned
phase I study of anlotinib in R/M NPC patients who failed from standard 1:1 to two treatment arms. Patients in arm A will receive neoadjuvant pem-
treatment had shown a manageable safety profile and promising antitumor brolizumab (200 mg Q3W for two cycles) followed by surgical resection then
activity with an ORR of 25%. This phase 3 trial aims to compare the efficacy SOC plus adjuvant pembrolizumab (15 cycles). Patients in arm B will undergo
and safety of anlotinib versus placebo in combination with GP in patients with only surgical resection followed by adjuvant SOC. Eligibility criteria will include
R/M NPC in the first-line setting. Methods: Key eligibility criteria of this study age $18 years; newly diagnosed, resectable, stage III/IVA HNSCC (AJCC
are that the patient has metastatic disease after curative radiotherapy, or is Cancer Staging Manual, 8th edition); and ECOG performance status 0-1.
primarily metastatic; has an ECOG PS of 0 or 1; has adequate organ function; Randomization will be stratified by primary tumor site (oropharynx/oral cavity vs
and has at least 1 measurable lesion according to RECIST 1.1. Eligible pa- larynx vs hypopharynx), tumor stage (III vs IVA), and HPV p16 status (oro-
tients will be randomized in a 1:1 ratio to receive intravenous gemcitabine at 1 pharynx p16 positive vs oropharynx p16 negative or larynx/hypopharynx/oral
g/m² on days 1 and 8, cisplatin at 75 mg/m² on day 1, plus anlotinib or placebo cavity). Treatment will continue until disease progression, unacceptable
12 mg daily orally on days 1–14 every 3 weeks for a maximum of 6 cycles toxicity, or decision to withdraw. Patients in arm A will undergo the first ra-
followed by anlotinib or placebo 12 mg daily on days 1–14 every 3 weeks as diologic imaging assessment after two cycles of neoadjuvant pembrolizumab
maintenance therapy. The primary endpoint is PFS. Secondary endpoints and before surgery. In both arms, postoperative imaging will be performed
include OS, ORR, quality of life and safety profile. Independent Data Moni- 12 weeks after SOC, then every 3 months until the end of year 3, and then every
toring Committee and Independent Review Committee will be used in this 6 months until the end of year 5. Dual primary end points are major patho-
study. We assume that the median PFS will be 10 mos in anlotinib group and 7 logical response, defined as #10% invasive squamous cell carcinoma within
mos in placebo group. To detect a 3-month improvement of PFS in anlotinib resected primary tumor and sampled regional lymph nodes per blinded central
group at a two-sided significant level of 0.05 and power of 0.8, allowing for a pathology, and event-free survival. Secondary end points include overall
dropout rate of 10%, a total of 336 patients will be enrolled. From August survival, pathological complete response, and safety and tolerability. Re-
2018, 58 patients have been enrolled. Clinical trial information: cruitment is ongoing and will continue until ~600 patients are enrolled.
NCT03601975. Clinical trial information: NCT03765918.

TPS6091 Poster Session (Board #79a), Sat, 1:15 PM-4:15 PM TPS6092 Poster Session (Board #79b), Sat, 1:15 PM-4:15 PM
EACH: A randomised phase II study evaluating the safety and anti-tumour Tabelecleucel in combination with pembrolizumab (Pembro) in platinum-
activity of the combination of avelumab and cetuximab relative to avelumab pretreated, recurrent/metastatic Epstein-Barr virus (EBV)-positive nasopha-
monotherapy in recurrent/metastatic head and neck squamous cell cancer. ryngeal carcinoma (EBV+NPC). First Author: Lillian L. Siu, Princess Margaret
First Author: Martin David Forster, University College London Hospitals, Cancer Centre, University of Toronto, Toronto, ON, Canada
London, United Kingdom
Background: Approximately 25% of patients (pts) with NPC develop RM
Background: Patients with recurrent/metastatic head and neck squamous cell disease, which has a poor prognosis (median overall survival [mOS]: 12–16
carcinoma (R/M HNSCC) have low response rates to licensed second line mo), despite standard treatments with radiation and/or chemotherapy. NPC is
therapies, including PD-1 inhibitors nivolumab and pembrolizumab, and an EBV-associated cancer in which programmed cell death ligand 1 (PD-L1)
represent an area of unmet clinical need. The chimeric IgG1 epithelial growth expression is upregulated upon EBV activation. Pembro showed antitumor
factor receptor (EGFR) monoclonal antibody cetuximab potentiates the activity activity in a phase 1b study of pts with RM-NPC (objective response rate [ORR]:
of radiotherapy in locally advanced HNSCC and chemotherapy in R/M HNSCC 26%; mOS: 16.5 mo) (Hsu, J Clin Oncol 2017;35:4050-56). Targeting RM
and is also licensed with modest activity as a single agent. Cetuximab initiates EBV+ NPC with tab-cel immunotherapy (off-the-shelf, allogeneic EBV-specific
Natural Killer (NK) cell antibody-dependent cell-mediated cytotoxicity (ADCC), T cells) in pts has also shown promise, with 2-yr OS rates of 84% (Prockop,
resulting in an anti-tumour immune response and the potential to augment the ASCO 2016;34:3012). The favorable safety profile of tab-cel offers the op-
activity of PD-1/PD-L1 inhibition. EACH aims to examine the safety and ef- portunity for combination immunotherapy with pembro for increased efficacy.
ficacy of the potentially synergistic interaction between cetuximab and Methods: This multicenter, open-label, single-arm phase 1b/2 study evaluates
avelumab, a fully human IgG1 anti-PD-L1 monoclonal antibody in R/M safety and efficacy of tab-cel in combination with pembro. Study participants
HNSCC. Methods: EACH is a randomised phase II trial preceded by a are $12 yrs of age with incurable, locally recurrent or metastatic EBV+ NPC
safety run-in phase. Eligible patients have histologically or cytologically previously treated with platinum-containing therapy. Pts are checkpoint-
confirmed measurable recurrent or metastatic squamous cell carcinoma of any inhibitor naı̈ve (phase 1b/2) or refractory to anti-PD-1 or anti-PD-L1 therapy
site in the safety run-in phase, and HNSCC in phase II, that is considered (phase 1b). Tab-cel is selected from a bank based on matching $2 HLA
incurable by local therapies. The safety run-in has a single arm de-escalating alleles, including $1 restricting HLA allele, between pts and donors. Tab-cel
design, aiming to establish the safety of cetuximab with avelumab and de- will be administered intravenously (IV) on days 1, 8, and 15 of a 21-day cycle.
termine the optimal dose of cetuximab within this combination. The safety run- Initial tab-cel dose is 2x106 cells/kg and the de-escalated tab-cel dose (if
in has a dosing schedule of avelumab (10 mg/kg) + cetuximab (500 mg/m2) needed) is 1x106 cells/kg. Pembro is administered at 200 mg IV Q3W in adults
intravenously every 2 weeks, with de-escalation of cetuximab to 400 mg/m2 and 2 mg/kg IV Q3W in pts aged 12 to 17 yrs. Primary outcomes of phase 1b
and 300 mg/m2 if necessary. The safety run-in phase commenced recruitment are to characterize dose-limiting toxicities, identify the maximum tolerated
in July 2018 and is ongoing. The phase II component will randomize 114 dose (MTD) or in the absence of MTD, the recommended phase 2 dose, and
HNSCC patients between either avelumab + cetuximab at the dose determined assess safety. Primary outcomes for phase 2 are ORR and safety. Secondary
by the safety run-in phase or avelumab (10 mg/kg) alone. Treatment will be in endpoints include progression-free survival, OS, and duration of response.
4-week cycles for up to one year. The primary endpoint in the safety run-in Enrollment is ongoing for 12-24 participants in the phase 1b portion of the
phase is the occurrence of dose limiting toxicities, and in phase II is Disease study with a 6+6 design. Phase 2 is expected to enroll 36 pts. Clinical trial
Control Rate at 24 weeks, using iRECIST. Blood and fresh tissue will be information: NCT03769467.
collected for exploratory translational studies, which will focus on the iden-
tification of potential novel predictive biomarkers for response. Clinical trial
information: NCT03494322.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Head and Neck Cancer 365s

TPS6093 Poster Session (Board #80a), Sat, 1:15 PM-4:15 PM TPS6094 Poster Session (Board #80b), Sat, 1:15 PM-4:15 PM
Comparative effectiveness trial of transoral head and neck surgery followed A global phase III multicenter, randomized, double-arm, open label trial of
by adjuvant radio(chemo)therapy versus primary radiochemotherapy for oro- ASP-1929 photoimmunotherapy versus physician’s choice standard of care
pharyngeal cancer (TopROC). First Author: Chia-Jung Busch, Department of for the treatment of patients with locoregional, recurrent head and neck
Otorhinolaryngology and Head and Neck Surgery, University Medical Center squamous cell carcinoma (rHNSCC). First Author: Merrill A. Biel, Rakuten
Hamburg Eppendorf, Hamburg, Germany Aspyrian Inc., San Diego, CA
Background: For locally advanced, transorally resectable oropharyngeal Background: rHNSCC commonly affects local or regional sites and is associated
cancer (OPSCC), both, surgical resection and risk-adapted adjuvant (chemo) with considerable morbidity and mortality. Outcomes of these patients remain
radiotherapy or definite chemoradiotherapy with or without salvage surgery are poor with limited curative treatment options and low response rates. New
considered the current standard of care. To date, these two different thera- modalities that are targeted, minimally invasive, and provide improved tumor
peutic approaches for transorally resectable OPSCC have not been compared response and control while having limited systemic side effects are needed.
head to head in a randomized trial yet. The goal of this study is to compare Photoimmunotherapy (PIT) is a new cancer-targeted platform technology. It is a
primary transoral surgery followed by adjuvant treatment with definitive combination drug and device treatment that utilizes monoclonal antibodies
chemoradiation for resectable OPSCC, especially with regards to loco-regional conjugated to a dye (IRDye 700DX) that is photoactivated using nonthermal red
control as well as organ function. Methods: TopROC is a prospective, two-arm, light to induce rapid and selective tumor cell death. The objective of this phase 3
open label, multicenter, randomized controlled comparative effectiveness study is to evaluate the efficacy and safety of ASP-1929 (EGFR-directed an-
study. Eligible pts. are $18 years old with treatment-naı̈ve, histologically tibody cetuximab-IR700 conjugate) PIT treatment as a monotherapy in patients
proven OPSCC (T1, N2a-c, M0; T2, N1-2c, M0; T3, N0-2c, M0 TNM 7th ed.) with locoregional rHNSCC. Methods: A global, multicenter phase 3, random-
which are amenable to transoral resection, ECOG PS #2 and no distant ized, double-arm, open-label, controlled trial of ASP-1929 PIT vs physician’s
metastasis. p16 immunohistochemistry by local pathology or FFPE tissue must choice standard of care (SOC) for the treatment of locoregional, rHNSCC in
be available for central diagnostic. 280 pts. will be randomly assigned (1:1) to patients who have failed or progressed on or after at least two lines of therapy, of
surgical treatment (arm A) or chemoradiation (arm B). Standard of care which at least one line must be systemic therapy, is currently underway. Pri-
treatment will be done according to daily clinical practice. Arm A consists of mary endpoints of the study are PFS and OS and the key secondary endpoint is
transoral surgical resection with neck dissection followed by risk-adapted ORR. Key inclusion criteria include: disease not amenable to curative therapy;
adjuvant (chemo)radiation. Pts. treated in arm B receive standard chemo- tumor(s) accessible for PIT light treatment and measurable by CT or MRI; male
radiation, residual tumor may be subject to salvage surgery. Follow-up visits are or female $ 18 yrs old with life expectancy . 6 months; ECOG score of 0 to 1.
planned until three years after treatment. Primary endpoint is time to local or Key exclusion criteria include: history of $ Grade 3 cetuximab infusion reaction;
locoregional failure or death from any cause (LRF). Secondary endpoints in- distant metastatic disease; tumors invading a major blood vessel unless
clude overall and disease-free survival, toxicity, patient reported outcomes and embolized. The study will include ~275 subjects in a 2:1 randomization (ASP-
cost-effectiveness analysis. Approximately 20 centers will be involved in 1929 PIT: Physician’s choice SOC). The physician’s choice SOC arm includes
Germany. This trial is supported by the German Cancer Aid and accompanied cetuximab, methotrexate, or docetaxel. Tumor(s) are illuminated with 690 nm
by a large scientific support program. Recruitment started in January 2018. PIT light treatment 24 hrs following completion of ASP-1929 infusion (640 mg/
Clinical trial information: NCT03691441. m²). Clinical trial sites will be in the USA, EU and Asia. Clinical trial information:
NCT03769506.

TPS6095 Poster Session (Board #81a), Sat, 1:15 PM-4:15 PM TPS6096 Poster Session (Board #81b), Sat, 1:15 PM-4:15 PM
Multicenter randomized controlled phase III study of nivolumab alone or in Roman: Reduction in oral mucositis with avasopasem manganese (GC4419)—
combination with ipilimumab as immunotherapy vs standard follow-up in Phase 3 trial in patients receiving chemoradiotherapy for locally-advanced,
surgical resectable HNSCC after adjuvant therapy. First Author: Chia-Jung non-metastatic head and neck cancer. First Author: Jon Holmlund, Ascenta
Busch, Department of Otorhinolaryngology and Head and Neck Surgery, Therapeutics, Carlsbad, CA
University Medical Center Hamburg Eppendorf, Hamburg, Germany
Background: Approximately 70% of patients receiving intensity-modulated
Background: Surgically treated locally advanced head and neck squamous cell radiotherapy (IMRT) plus cisplatin for locally advanced head and neck
carcinoma (LA HNSCC) often requires postoperative chemoradiation with high cancer (HNC) develop SOM, defined as WHO Grade 3 or 4, which limits
risk of acute and late toxicity. Disease-free survival (DFS) after 2 years is patients’ ability to eat solids (Gr 3) or liquids (Gr 4, requiring enteral nu-
approximately 70%. Combining Nivolumab (N), a PD-1inhibitor, and trition). An RT-induced burst of superoxide initiates oral mucositis (OM)
Ipilimumab, a CTLA4 inhibitor, as maintenance therapy may improve DFS due development. GC4419, a superoxide dismutase mimetic, interrupts this
to anti-tumor effects of immunotherapy by enhancing cross-presentation of process by converting superoxide to H2O2. It showed promising reduction of
tumor antigens. The IMSTAR HN study compares neoadjuvant N and N6I SOM in a published open-label Phase 1b/2a trial (IJROBP 1 Feb 2018). In a
6 months after adjuvant therapy vs the standard therapy as first-line treatment subsequent randomized, double-blind placebo-controlled trial in 223 pa-
for LA HNSCC. Methods: Eligible pts are $18 years old with treatment-naive tients receiving IMRT/cisplatin for HNC (ASCO 2018), 90 mg of GC4419
LA HNSCC (oral cavity, oropharynx p16-, hypopharynx, and larynx), ECOG administered M-F prior to IMRT demonstrated statistically significant re-
PS #1, and no distant metastasis. 276 pts will be randomized (2:1) into 2 duction in SOM duration (p=0.024, median 1.5 days @ 90 mg vs 19 days
arms and approximately 10 centers in Germany will be involved. Standard of placebo) and meaningful reductions @ 90 mg in SOM incidence (43% vs
care (arm II) consists of surgical resection followed by risk-adapted adjuvant 65%) and severity (incidence of Grade 4, 16% vs 30%). The safety profile
(chemo)radiation. The experimental arm I receives neoadjuvant N 3mg/kg. was acceptable and consistent with the known toxicities of IMRT/cisplatin.
After treatment according to standard arm a second randomization will be Methods: 335 patients at multiple centers in the U.S. and Canada with
performed: In arm Ia N 3mg/kg will be given every 2 weeks until progression or locally-advanced, nonmetastatic head and neck cancer (oral cavity/
up to 6 months. In arm Ib I 1mg/kg will be applied additionally every 6 weeks oropharyngeal) receiving 70 Gy IMRT (.50 Gy to . 2 oral sites) plus cis-
also until progression or up to 6 months. Primary endpoints is DFS in arms I and platin (40 mg/m2 qwk x 6-7, or 100 mg/m2 q3wk x 3) are being randomized
II. Secondary endpoints: Local regional control (LRC), distant metastasis free (double-blinded) 3:2 to 90 mg of GC4419 or placebo, M-F before each RT
survival (DMFS), overall survival (OS), quality of life (QoL), survival depending fraction. Enrollment is stratified by cisplatin schedule and treatment setting
on PD-L1 status, comparison of arm Ia vs arm II and Ib vs. II. AEs, graded per (definitive vs post-op). OM by the WHO scale will be assessed twice weekly
CTCAE v4.03, are evaluated for at least 12 months after randomization. The during RT & weekly for 2 weeks post RT. The primary efficacy endpoint is
translational program includes investigations concerning immunmodulation, incidence of SOM through the end of IMRT. Secondary efficacy endpoints
mutational load in general, but also specific mutations in targets involved in include severity (incidence of Grade 4 OM through the end of IMRT), & days
immune function and antigen presentation. Recruitment started in August of SOM (days from first to last SOM for all patients, with patients never
2018. Clinical trial information: NCT03700905. developing SOM having 0 days of SOM by definition). Days of SOM for the
subset developing SOM will be analyzed descriptively. Patients will be
followed for one year post IMRT for tumor progression/recurrence and for two
years for survival. Supported by Galera Therapeutics, Inc. Clinical trial in-
formation: NCT03689712.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
366s Head and Neck Cancer

TPS6097 Poster Session (Board #82a), Sat, 1:15 PM-4:15 PM TPS6098 Poster Session (Board #82b), Sat, 1:15 PM-4:15 PM
A phase 3 (COSMIC-311), randomized, double-blind, placebo-controlled ACCURACY: phase (P) 2 trial of AL101, a pan-Notch inhibitor, in patients
study of cabozantinib in patients with radioiodine (RAI)-refractory differen- (pts) with recurrent/metastatic (R/M) adenoid cystic carcinoma (ACC) with
tiated thyroid cancer (DTC) who have progressed after prior VEGFR-targeted Notch activating mutations (Notchact mut). First Author: Renata Ferrarotto,
therapy. First Author: Marcia S. Brose, Head and Neck Surgery, Radiology, The University of Texas MD Anderson Cancer Center, Houston, TX
University of Pennsylvania, Philadelphia, PA
Background: Notch signaling plays a key role in tumorigenesis. Notch cleavage
Background: Treatment options are limited for patients with RAI-refractory by g-secretase frees the Notch intracellular domain, which promotes the
DTC that is resistant to VEGFR-targeted therapy. Cabozantinib inhibits re- expression of target genes involved in cancer. AL101, a small molecule, is a
ceptor tyrosine kinases including VEGFR2, MET, AXL, and RET, which are g-secretase inhibitor that potently inhibits Notch1-4, resulting in robust an-
implicated in the development of DTC, and has shown clinical activity in early- titumor activity in vivo (PMID 26005526), including ACC xenograft models
phase studies of patients with RAI-refractory DTC. This study evaluates the with Notchact mut (Ferrarotto, AACR 2019, Abstr 4885). Three P1 trials tested
efficacy and safety of cabozantinib in patients with RAI-refractory DTC who AL101 as monotherapy or in combination regimens in . 200 solid/
have progressed during or after prior VEGFR-targeted therapy. Methods: This hematologic cancer pts. In the P1 trial of AL101 monotherapy, conducted
is a phase 3, multicenter, randomized, double-blind, placebo-controlled trial in 94 pts with advanced/metastatic solid tumors refractory to standard ther-
(NCT03690388). The co-primary endpoints are progression-free survival and apies (Tx), AL101 was generally well tolerated, with manageable AEs, and the
objective response rate evaluated by blinded independent radiology committee recommended P2 dose was 4 mg IV once weekly (QW; El-Khoueiry, ASCO
(BIRC) per RECIST v 1.1. Additional endpoints include safety, overall survival, 2018, Abstr 2515). 4 pts had objective responses, 2 of those had Notchmut (1
quality of life, and changes in relevant biomarker levels (eg, thyroglobulin). of which had ACC). ACC, a rare cancer that most commonly develops in the
Approximately 300 patients will be randomized in a 2:1 ratio to receive either major salivary glands, but can also arise in minor salivary glands in the trachea,
cabozantinib (60 mg QD orally) or placebo. Randomization is stratified by prior lacrimal gland, and other sites, is refractory to chemotherapy, with a high
treatment with lenvatinib and age (# 65 yrs vs . 65 yrs). Eligible patients must recurrence rate. Notchact mut are found in a subset of ACC pts (11%–22%),
have a pathologic diagnosis of DTC and must have been previously treated with with particularly aggressive disease and poor prognosis. There is no proven
or deemed ineligible for treatment with iodine-131 for DTC. Patients must have active treatment for R/M ACC pts (PMID 27870570). Methods: ACCURACY
received lenvatinib or sorafenib for DTC and progressed during or following (NCT03691207) is an open-label, single-arm, multicenter study of AL101
treatment with a VEGFR inhibitor. Up to 2 prior VEGFR-targeting TKI agents (4 mg IV QW) in pts with R/M ACC (bone-exclusive disease included) with
are allowed. Patients randomized to placebo may be eligible for real time on- known Notch1-4act mut. Pts with disease progression within 6 months of en-
study crossover to cabozantinib based on BIRC confirmation of disease pro- rollment or newly diagnosed metastatic pts are allowed; pts who received . 3
gression. Unblinded patients randomized to cabozantinib may continue on prior systemic Tx are excluded. Primary endpoint: ORR by RECIST v1.1 (or
study treatment if there is clinical benefit per investigator. Key words: modified MDA bone criteria), by independent review committee (IRC). Sec-
Radioiodine-refractory differentiated thyroid cancer, cabozantinib, VEGFR- ondary endpoints: ORR by investigator review (IR), duration of response by IRC
targeted therapy, trial-in-progress. Clinical trial information: NCT03690388. and IR, PFS by IRC, OS, and safety. Per the Simon optimal design, 12 pts are
enrolled in stage 1; if $2 pts respond, 24 additional pts are enrolled in stage 2.
If $4 pts in stage 2 respond, the trial is deemed positive. This design yields 5%
type I error rate and 80% power, if ORR is 25%. 4 of planned 36 pts have been
enrolled as of 2/12/19. Clinical trial information: NCT03691207.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Health Services Research, Clinical Informatics, and Quality of Care 367s

6500 Oral Abstract Session, Fri, 2:45 PM-5:45 PM 6501 Oral Abstract Session, Fri, 2:45 PM-5:45 PM
Effect of montelukast and rupatadine on rituximab infusion time, rate, Comparison of normal saline versus heparin flush solutions for maintaining
severity of reactions, and cost of administration. First Author: Rouslan patency of central venous catheter in cancer patients. First Author: Amy Pai,
Kotchetkov, Simcoe Muskoka Regional Cancer Program, Barrie, ON, Canada University of Texas MD Anderson Cancer Center, Houston, TX
Background: Rituximab is associated with frequent infusion reactions which carry significant Background: Published guidelines recommend normal saline (NS) flushes
burden to patients and health care practitioners. Standard pre-medications (SP) do not prevent for maintaining patency of central venous catheters (CVCs) in cancer patients.
reactions sufficiently. Rupatidine (R) and Montelukast (M) are used for symptomatic treatment
of urticaria and allergic rhinitis. We assessed addition of R, M and their combination on
The practice at a large oncology ambulatory facility was modified to use NS
Rituximab infusion, rate, severity of reactions and cost of administration. Methods: Adult flushes instead of heparin for CVC maintenance (centrally inserted catheters
patients with lymphoproliferative disorders treated at our cancer center between Jan 2018 to and peripherally inserted central catheters, excluding implanted ports). The
Jan 2019 were evaluated with Rituximab-containing regiments. Since the majority of reactions number of catheter occlusion events, pre- and post-implementation, was
occur during the first infusion, our study was limited to the initial Rituximab treatment. utilized to determine the effectiveness and safety of this practice change.
Patients received either SP with diphenhydramine/acetaminophen and additional R, M or R+M Methods: Retrospective data of patients aged 18 years and older who pre-
combination. Comparative analysis of infusion time/rate, severity of infusion reactions, number
sented to the ambulatory centers were collected from the electronic health
of rescue medications and cost of Rituximab infusions among groups was performed using
one-way ANOVA with Tukey post-hoc or chi-square. The study was approved by our institutional record (EHR). The number of alteplase instillations was used as a surrogate
IRB. Results: Patients received either: 1) SP; 2) SP + Rupatadine (R) 10 mg; 3) SP + measure for occlusion events. Ambulatory line-days, defined as patients ar-
Montelukast (M) 10 mg; or 4) both (SP+R+M). Patient demographics are shown in the table. riving for an appointment and with either an active CVC or a CVC placement,
Compared to SP, the R, M and R+M groups had greater improvement in Rituximab delivery. was the denominator. The numerator was the number of ambulatory line-days
Mean infusion time was 306 [range 235-441] min. in SP, 254 [105-390] in M, 265 [193-350] with a catheter line and alteplase dose(s) given. The pre-intervention CVC
in R and 248 [196-342] in R+M groups, (p=0.0001). Infusion reactions occurred in 92%
maintenance period using heparin flushes was March 2016 - April 2017. The
in SP vs. 38, 45, 33% in M, R and R+M groups (p=0.0001). Median reaction grade was
2 in SP, 1 (M), 0 (R and R+M). Median number of rescue medications was 3 [0-10] in SP post-intervention CVC maintenance period using NS flushes was May 2017 -
vs 0 [0-7] in M, R and R+M groups. Cost of rescue medications (US$) was 38 [0-63] (SP), September 2018. Results: 95,089 line-days and 4,452 unique patients were
11 [0-50] (M), 17 [0-63] (R), 12 [0-58] (R+M) groups (p,0.0001). Mean nursing cost analyzed pre-intervention and compared to 115,194 line-days and 5,575
(US$) per patient infusion was calculated as 269 [207-388] in SP vs 222 [92-343] (M), unique patients post-intervention. The baseline incidence rate of occlusion
233 [170-308] (R), 218 [174-301] (R+M) group. Conclusions: Addition of R, M and par- was 0.91%, compared to the post-intervention incidence rate was 2.67% (p ,
ticularly R+M significantly improved Rituximab delivery, lowered the rate and severity of
0.01). The increase started immediately within the first month of the practice
infusion reactions, and lowered the cost of Rituximab administration.
change (2.26%) and continued throughout the post-implementation obser-
Overall Survival: Age and Toxicity.
vation time. Conclusions: In an ambulatory oncology-practice setting with a
SP M R M+R
high volume of CVC utilization, an increase in line occlusions rate was observed
N (pts) 25 21 20 21 after implementing the practice change of flushing CVCs from heparin to NS.
Mean age 65 73 65 69
Male (%) 43 71 75 80 This raises concerns regarding safety and additional burdens to patients and
Hx of allergy (%)
Disease (%): DLBCL
40
24
33
14.3
50
20
25
28.6
caregivers, and additional analyses are ongoing to evaluate the further impact
FL 36 9.5 20 19 of this evaluation.
CLL/SLL 16 19 20 9.5
Other iNHL 24 57.1 20 52.4
Advance stage/bulky (%) 64 76 85 67
Tx naı̈ve (%) 96 86 80 90
Chemotherapy (%): R-CHOP 32 23.8 30 38.1
B+R 56 42.8 30 38.1
FCR 8 9.6 25 9.5
Other 4 23.8 15 14.3
Mean Rituximab dose (mg): 727 750 740 753

6502 Oral Abstract Session, Fri, 2:45 PM-5:45 PM 6503 Oral Abstract Session, Fri, 2:45 PM-5:45 PM
Preventing excess narcotic prescriptions in MIS urologic oncology discharges Significance of examined lymph node number in accurate staging and long-
(PENN): A prospective cohort quality improvement initiative. First Author: term survival in resected stage I-II pancreatic cancer: More is better? A large
Ruchika Talwar, University of Pennsylvania, Philadelphia, PA international population-based cohort study. First Author: Lei Huang, German
Cancer Research Center (DKFZ), Heidelberg, Germany
Background: In the setting of the national opioid crisis, there is increasing
interest in non-narcotic pain strategies, particularly for oncology patients. Background: Examined lymph node (ELN) number is an important quality
Robotic urologic surgeries for cancer have been shown to result in less pain metric in cancer care. This large international cohort study aimed to investigate
than open approaches. We hypothesized that the majority of these patients the associations of ELN number with accurate staging and long-term survival in
could be safely discharged with adequate analgesia without opioids. pancreatic cancer (PaC) and to robustly determine the minimal and optimal
Methods: This prospective cohort study aimed to reduce narcotics prescribed ELN thresholds. Methods: Population-based data on patients with stage I-II
at discharge after robotic radical prostatectomy (RARP), robotic radical ne- PaC resected in 2003-2015 from the US Surveillance, Epidemiology, and End
phrectomy (RARN) and robotic partial nephrectomy (RAPN). Prior to 9/2018, Results (SEER)-18 Program and Netherlands National Cancer Registry (NCR)
100% of patients were discharged on varying amounts of oxycodone (range: were analyzed. Associations of ELN number with stage migration and sur-
75-337.5 oral morphine milligram equivalents [MME]). We implemented a vival were evaluated using multivariable-adjusted logistic and Cox regression
standard non-opioid analgesia pathway with escalation options across the models, respectively. The series of odds ratios (ORs) for stage migration and
continuum of care. Patients received gabapentin 300 mg and acetaminophen hazard ratios (HRs) for survival with more ELNs were fitted using a LOWESS
975 mg once PO pre-operatively, as well as gabapentin 300 mg every 8 hours, smoother, and structural breakpoints were determined by Chow test.
acetaminophen 975 mg every 8 hours PO, and ketorolac 15 mg every 6 hours Results: Overall 18,303 patients were analyzed. With increasing ELN number,
IV post-operatively. If complaining of persistent pain despite the standing both cohorts exhibited significant proportional increases from node-negative to
regimen, patients were given 50 mg or 100 mg of tramadol every 6 hours as node-positive disease (ORSEER-18= 1.05, 95% CI = 1.04-1.05; ORNCR= 1.10,
needed for pain level 5-7 or 8-10 on the visual analog scale, respectively. If 95% CI = 1.08-1.12) and serial improvements in survival (HRSEER-18= 0.98,
requiring further escalation, patients were given 5 or 10 mg of oxycodone every 95% CI = 0.98-0.99; HRNCR= 0.98, 95% CI = 0.97-0.99) per additional ELN
6 hours as needed on the aforementioned scale. Regardless of escalation after controlling for confounders. Associations for stage migration and survival
status, all patients were discharged on the standing non-narcotic protocol. If remained significant in most stratifications by patient, tumor, and treatment
escalated, ten pills of tramadol 50 mg or oxycodone 5 mg were prescribed factors. Cut-point analyses suggested a minimal threshold ELN number of 12
accordingly. Results: Our cohort (n = 170) consisted of patients undergoing and an optimal number of 19, which were validated both internally in the
RARP (n = 87), RARN (n = 25), RAPN (n = 58) between 9/1/2018-1/9/2019. derivative US cohort and externally in the Dutch cohort with the ability to well
Overall, 67.7% were discharged without opioids, 24.4% with ten pills of discriminate different probabilities of both survival and stage migration.
tramadol 50 mg (50 MME) and 8.2% with ten pills of oxycodone 5 mg (75 Conclusions: In stage I-II PaC, more ELNs are associated with more precise
MME). On multivariate analysis, older age (OR: 0.961, 95% CI: 0.923-0.995, nodal staging, which might largely explain the survival association. Our results
p = 0.026) was associated with lower odds of needing opioids at discharge. robustly conclude 12 ELNs as the minimal and suggest 19 ELNs as the optimal
There was no difference in postoperative telephone encounters between those cut-points, for evaluating quality of lymph node examination and possibly for
discharged with or without opioids. Conclusions: The majority of robotic stratifying postoperative prognosis. Our findings provide important references
surgery patients do not require opioids upon discharge. An escalation protocol for defining population-based quality metrics in PaC care.
allows for a patient centered approach to reduce narcotic prescribing while still
addressing cancer and surgical pain.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
368s Health Services Research, Clinical Informatics, and Quality of Care

6504 Oral Abstract Session, Fri, 2:45 PM-5:45 PM 6505 Oral Abstract Session, Fri, 2:45 PM-5:45 PM
Economic analysis of alternative pembrolizumab and nivolumab dosing A process to reduce hospital admissions in the OCM population through
strategies at an academic cancer center. First Author: Evan Thomas Hall, focused intervention of super-utilizers. First Author: Manuel Perry, Crystal
Stanford University School of Medicine, Stanford, CA Run Health Care, LLP, Middletown, NY
Background: Pembrolizumab (P) and nivolumab (N) were initially investi- Background: A small number of patients (super-utilizers) are responsible for
gated and FDA-approved with weight-based dosing strategies, but later the much of the cost in healthcare. In several studies, 5% of patients are responsible
approval label was amended to a fixed dose administration. Given increasing for 50% of the cost. Healthcare hot-spotting is the strategic use of data to deliver
concerns about financial toxicity of cancer therapies, we hypothesize that enhanced resources to selected super-utilizers in an effort to improve the quality
weight-based dosing of P and N and allowing vial sharing among patients will of care and to reduce costs. Methods: Patients enrolled in the Oncology Care
result in substantial cost savings. Methods: We obtained IRB approval to Model (OCM) who were scheduled to receive cytotoxic, targeted therapy or
retrospectively examine all outpatient doses of P and N given at three immunotherapy beginning in the fall of 2017 were screened by nurse practi-
Stanford Medicine infusion centers between July 1, 2018 and Oct 31, 2018 tioners prior to initiating treatment in an effort to identify super-utilizers. Risk
using the Stanford Medicine Research Data Repository (STARR) database. factors included high risk disease (stem cell transplant, myelodysplastic syn-
We performed cost-minimization analysis modeling the impact of dosing drome, acute leukemia, stage IV disease), significant co-morbidities (CHF,
strategies based upon patient weight versus fixed dosing (2 mg/kg vs 200 mg COPD, renal failure, Insulin Dependent Diabetes , presence of a gastrostomy
q3wks for P; 3 mg/kg vs 240 mg q2wks or 6 mg/kg vs 480 q4wks for N). tube), or an admission to the hospital in the past 6 months. Such patients were
“Dose-minimization” (DM) was defined as whichever dose was lower provided with enhanced services which included twice weekly outreach by a
(weight-based or fixed dose). The impact of allowing vial sharing (considering nurse practitioner between 8am and 10am for a telephonic evaluation of their
commercially available vial sizes) between patients treated at the same site health status. If the patient was found to have either a new problem that required
and on the same date was assessed. Average sales price (ASP) from Center intervention or a worsening chronic problem, then a same-day appointment was
for Medicare and Medicaid Services for Part B drugs was used for cost made with the nurse practitioner, oncologist, primary care provider or appropriate
estimates. Results: A total of 1,029 doses of P or N were administered across a specialist. Results: Inpatient admissions decreased from 21.3 per 100 OCM
variety of cancer types. For most doses (N = 789, 77%), the calculated weight- beneficiaries to 18.7 per 100 OCM beneficiaries representing a 12.2% de-
based dose was less than the fixed dose. DM resulted in decreased usage and crease. Inpatient costs were reduced from $979 per beneficiary per month
expenditures of both P and N, and a further decrease was observed with vial (PBPM) to $885 PBPM representing a decrease of $94 PBPM (9.6% decrease).
sharing. Total savings estimated with DM and vial sharing strategy were . $1.4 Conclusions: Targeting oncology super-utilizers based on the nature of their
million (Table). This amounted to savings of . 22,000 mg of P (112 fixed diagnosis, co-morbidities, or history of a recent hospital admission and offering
doses) and . 11000 mg of N (47 fixed doses). Savings were greatest at the telephonic evaluation reduced hospitalizations and decreased cost resulting in a
highest volume infusion center. Conclusions: Alternative dosing strategies of P savings of nearly $600,000 per year for our OCM cohort of 600 beneficiaries.
and N would result in significantly less drug utilization and pharmaceutical
expenditure without anticipated impact on efficacy. Potential barriers to this Quarter Admissions (per 100/beneficiaries) Cost (PBPM)
approach include existing policies regarding vial sharing and drug vial sizes. April-June 2017 21.3 $979
July-Sep 2017 20.7 $944
Actual cost Estimated cost ($) of Ab Estimated cost ($) of Ab Est savings ($) Oct-Dec 2017 20 $882
($) of Ab used with DM w/o vial used with DM w/ vial with DM + vial Jan-March 2018 19.7 $903
Drug Name used sharing sharing sharing
April-June 2018 18.7 $885
Pembrolizumab $6,419,497 $6,371,085 $5,310,060 $1,109,437
Nivolumab $3,300,090 $3,000,207 $2,987,539 $312,551

Total $1,421,988

6506 Oral Abstract Session, Fri, 2:45 PM-5:45 PM 6507 Oral Abstract Session, Fri, 2:45 PM-5:45 PM
The financial impact of physician practice patterns: An analysis of as-needed Quality of end-of-life cancer care at minority-serving US cancer centers: A
filgrastim versus prophylactic pegfilgrastim in patients with prostate cancer retrospective study of Medicare claims data. First Author: Garrett Wasp,
receiving docetaxel. First Author: Christine Barrett, Allegheny General Dartmouth-Hitchcock Medical Center, Lebanon, NH
Hospital, Pittsburgh, PA
Background: Higher EOL treatment intensity is burdensome and has been
Background: Docetaxel improves survival in metastatic prostate cancer. Pa- defined as low-quality care. We explored cancer centers’ EOL quality outcomes
tients receiving docetaxel are at an intermediate risk of febrile neutropenia among minority and white patients and evaluated whether minority-serving
(FN). Granulocyte-colony stimulating factor (G-CSF) reduces the incidence of cancer centers had systematically lower EOL quality. Methods: We conducted a
neutropenia and infection. There is no consensus among physicians on G-CSF retrospective cohort study of Medicare beneficiaries with poor-prognosis
use with docetaxel in prostate cancer, and physician preference may signif- cancers who died between April 1, 2016 and December 31 2016. We at-
icantly impact healthcare costs. The purpose of this study is to assess phy- tributed patients’ EOL treatment to the cancer center where they received the
sician practice patterns by comparing efficacy and cost of as-needed filgrastim preponderance of inpatient services during the last 6 months of life. We then
(ANF) to prophylactic pegfilgrastim (PPF) for neutropenia in patients with compared age-sex-comorbidity adjusted center-level measures of EOL care
prostate cancer receiving docetaxel. Methods: This retrospective cohort study (EOL chemotherapy, emergency department (ED) use, intensive care unit (ICU)
included patients with prostate cancer that received standard docetaxel and G- admission, hospice use, life-sustaining treatment use, palliative care, and
CSF, either ANF or PPF, from February 2016 to October 2018 at our in- advance care planning) between minority (black, Hispanic, Asian, other) and
stitution. In the ANF arm, patients with an ANC less than 1000 cells/mL on day non-Hispanic white (white) patients within the same center and across centers,
8 of the cycle received filgrastim. In the PPF arm, patients received pegfil- grouped by concentration of minorities served (low: , 15%, medium 15-30%,
grastim on day 2 for primary prophylaxis. Patient characteristics, chemo- and high . 30%). Results: Among 126,434 patients, 10,006 (21.4% mi-
therapy dose intensity, delays, and reductions, and incidence of FN were nority) received treatment at one of 53 National Cancer Institute-designated
collected. Results: Of the 65 patients included in this study, 32 received ANF and/or National Comprehensive Cancer Network-affiliated cancer centers. Only
and 33 received PPF. Patients in the ANF arm received an average of 6.0 4/8 quality measures had sufficient sample size to calculate a minority-specific
cycles versus 4.8 cycles in the PPF arm (p = 0.013). The completion rate of rate for $10 centers. Those measures showed high within-center correlation for
planned cycles differed, with 93.8% completion in the ANF arm versus 66.7% minority and white patients (ICU admission: r = 0.79,p , 0.001; no hospice
in the PPF arm (p = 0.001). Average total drug exposure per patient was greater referral: r = 0.70,p , 0.0001; life sustaining treatment: r = 0.73,p = 0.004;
in the ANF arm than the PPF arm (437.5 mg/m2 versus 343.4 mg/m2; p = and palliative care: r = 0.78,p , 0.0001), but the mean adjusted rate for
0.007). In both groups, there was no significant difference in dose delays or minority versus white patients was significantly worse for two measures: no
reductions, or incidence of FN. However, there was a significant difference in hospice referral (40.2% v. 37.2%; p , 0.02) and life-sustaining treatments
cost, with an average of $740.26 per cycle per patient for filgrastim, versus (21.8% v. 19.4%; p , 0.02). When grouped by concentration of minorities
$7327.72 per cycle per patient for pegfilgrastim, based on current average served (low/medium/high), 5/8 measures showed systematically lower quality
wholesale price. Conclusions: In our study, use of ANF was safe, and did not as the concentration of minorities increased: more than 1 ED visit (6.0/8.5/
impact planned dose, intensity, or duration when compared to PPF. In ad- 7.7%; p = 0.002), ICU admission (29.1/29.7/35.1%; p = 0.0004), no hospice
dition, the administration of ANF was associated with significant cost savings. referral (34.3/38.7/36.8% (p = 0.005), and life sustaining treatments (14.8/
Analysis of G-CSF use in patients with prostate cancer receiving docetaxel at 16.7/17.9%; p = 0.005). Conclusions: There were systematic differences in
our institution has identified an opportunity for substantial healthcare cost end-of-life quality measures across US cancer centers. For many measures,
savings. quality was lower at centers that served a greater concentration of minorities.
However, EOL care quality for minority and white patients was similar for most
but not all measures within any given center.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Health Services Research, Clinical Informatics, and Quality of Care 369s

6508 Oral Abstract Session, Fri, 2:45 PM-5:45 PM 6509 Oral Abstract Session, Sun, 8:00 AM-11:00 AM
Impact of a same-day breast biopsy program on disparities in time to biopsy. The impact of routine ESAS use on overall survival: Results of a population-
First Author: Matthew Seidler, Massachusetts General Hospital, Boston, MA based retrospective matched cohort analysis. First Author: Lisa Catherine
Barbera, Tom Baker Cancer Centre, Calgary, AB, Canada
Background: While screening rates have improved among minorities, racial/
ethnic disparities in diagnosis and treatment persist. Many steps in the Background: The study objective was to examine the impact of routine
diagnostic pathway can delay tissue diagnosis, and in usual practice breast Edmonton Symptom Assessment System (ESAS) use on overall survival among
biopsies are performed days to weeks after biopsy recommendation. The adult cancer patients. We hypothesized that patients exposed to ESAS would
purpose of this study was to identify if racial/ethnic disparities exist in time have better overall survival rates than those who didn’t have ESAS.
from biopsy recommendation to biopsy, and if a same-day biopsy program Methods: The effect of ESAS screening on survival was evaluated in a retro-
(biopsy on the same day as the recommendation) eliminates these dispar- spective matched cohort study. The cohort included all Ontario patients aged
ities. Methods: After IRB approval, we identified all diagnostic mammogram 18 or older who were diagnosed with cancer between 2007 and 2015. Patients
and ultrasound exams leading to biopsy pre- (September 2016-March 2017) completing at least one ESAS assessment during the study were considered
and post- (September 2017-March 2018) implementation of our same-day exposed. The index date was the day of their first ESAS assessment. Follow up
biopsy program. We compared the distribution of age, race, language, in- time for each patient was segmented into one of three phases: initial, con-
surance type, days to biopsy and proportion of same-day biopsies in pre- vs. tinuing, or palliative care. Exposed and unexposed patients were matched 1:1
post-implementation groups using the Wilcoxon test (for continuous vari- using hard (birth year 6 2 years, cancer diagnosis date 6 1 year, cancer type
ables) and the Pearson’s chi-squared test (for categorical variables). Mul- and sex) and propensity-score matching (14 measures including cancer stage,
tivariable linear and logistic models were estimated in pre and post periods to treatments received, and comorbidity). Matched patients were followed until
assess if days from biopsy recommendation to biopsy (linear) and having a death or the end of study at Dec 31, 2015. Kaplan-Meier curves and mul-
same-day biopsy (logistic) were associated with age, race, language, and tivariable Cox regression were used to evaluate the impact of ESAS on survival.
insurance type. Results: 663 and 482 patients underwent biopsy during pre- Results: There were 128,893 pairs well matched on all baseline character-
and post-implementation, respectively. Age, race, language, and insurance istics (standardized difference , 0.1). The probability of survival within the
type were similar between time periods. For all patients, the same-day biopsy first 5 years was higher among those exposed to ESAS compared to those who
program decreased mean time from diagnostic examination to biopsy from 8 were not (73.8% vs. 72.0%, P-value , 0.0001). In the multivariable Cox
(IQR: 4-13) to 0 (IQR: 0-4) days (p , 0.001). During the pre time period, regression model, ESAS assessment was significantly associated with a de-
non-white patients and having government insurance were significantly creased mortality risk (HR: 0.49, 95% CI: 0.48-0.49) and this protective
associated with longer days to biopsy (non-white aCoef: 2.30 (95% CI: 0.58- effect was seen across all phases. Conclusions: ESAS exposure is associated
4.03); insurance aCoef: 1.67 (95% CI: 0.02-3.32); p , 0.05), and in- with improved survival in cancer patients, in all phases of care. To the extent
creasing age and having government insurance were significantly associated possible, extensive matching methods have mitigated biases inherent to ob-
with decreased odds of having a same-day biopsy (age aOR: 0.97 (95% CI servational data. This provides real world evidence of the impact of routine
0.95-0.99); insurance aOR: 0.35 (95% CI 0.14-0.88); p , 0.05), after symptom assessment in cancer care.
adjustment. During the post time period there was no evidence of these
disparities. Conclusions: A same-day biopsy program eliminated racial/
ethnic disparities in time from breast biopsy recommendation to biopsy.

6510 Oral Abstract Session, Sun, 8:00 AM-11:00 AM 6511 Oral Abstract Session, Sun, 8:00 AM-11:00 AM
Phase III non-randomized controlled trial of PROMPT-Care, an eHealth inter- Longitudinal toxicity analysis with novel summary metrics of lenalidomide
vention utilizing patient reported outcomes in routine clinical care: Impact on maintenance in follicular lymphoma in ECOG-ACRIN 2408. First Author:
emergency department presentations. First Author: Afaf Girgis, Centre for Gita Thanarajasingam, Mayo Clinic, Rochester, MN
Oncology Education and Research Translation (CONCERT), Ingham Institute
Background: Conventional adverse event (AE) analysis (ToxC) focuses on
for Applied Medical Research and University of New South Wales, Sydney,
incidence of grade (gr) 3+ toxicities, and fails to capture AE time profile.
Australia
Novel metrics that reflect chronic low gr and overall AE burden are needed.
Background: The significant impact of routine assessment and clinical utili- We applied the Toxicity over Time (ToxT) approach to ECOG-ACRIN 2408 to
zation of patient-reported outcomes (PRO) on patient and survival outcomes and depict time-dependent toxicity of lenalidomide (L) with rituximab mainte-
reduced emergency department (ED) presentations has been demonstrated in nance (MR) in follicular lymphoma (FL), and we developed a novel summary
specific patient populations (e.g. advanced cancer). This controlled trial metric of symptomatic AE burden, the maximum gr over time (MGOT).
evaluated the impact of an eHealth system, PROMPT-Care, on ED presentations Methods: In E2408, high risk FL patients (pts) were randomized (1:2:2) to:
in a diverse population of cancer patients from four oncology treatments centers. A) bendamustine-rituxumab (BR) x 6 then MR x 2 years (yrs) vs B) BR-
Methods: All adult patients receiving cancer care (including adjuvant therapy bortezomib x 6 then MR x 2 yrs vs C) BR x 6 then MR x 2 yrs + L x 1 yr (MRL).
and follow-up) were eligible, excepting those with a diagnosis of a hematological Analysis included 3 laboratory and 5 symptomatic AEs of highest incidence
malignancy, insufficient English literacy or no internet access outside of the during maintenance on arms A and C. Treatment-related AEs of any gr were
clinic. Intervention (PROMPT-Care) patients completed monthly online as- analyzed by ToxC and ToxT. Repeated measures, time-to-event (TTE) and
sessments comprising 61 items of distress, common symptoms and unmet area under the curve (AUC) analyses capture trends over time in ToxT. MGOT
needs, with PRO results electronically transferred into the electronic medical combines the 5 symptomatic AEs. Results: 104 randomized pts (30 MR, 74
record (EMR). In “real-time”, the care team accessed patients’ PRO summary MRL) were included. For the laboratory AEs, by ToxC, neutropenia incidence
reports to guide their care, an email alert notified nurses of ongoing unresolved was significantly higher in MRL (84%) than MR (47%, p , .001). ToxT
issues between visits, and patients received links to support self-management. additionally shows neutropenia does not worsen over time (10/14/20% gr 1/
Control group patients (n = 2,288) comprised the general cancer patient 2/3+ at c1, 6/21/12% gr 1/2/3+ at c12). For the symptomatic AEs, ToxC
population receiving usual care at the participating cancer therapy centers. indicates 2% gr 3+ GI AEs. However, gr 1-2 GI AEs are more common on
Multivariable negative binomial regression was used to compare between-group MRL (59%) than MR (26%, p , .001). ToxT AUC captures a higher burden
differences. Results: From April 2016 to March 2018, 345 eligible patients of GI AEs over time on MRL(2.8) vs MR(1.4, p = .002). TTE depicts sooner GI
(mean age 62, 58% female, 27% stage IV) participated and were sent at least AE onset in MRL (10% vs 0% gr 2+ GI by day 50, p = 0.03). Bar charts of
four assessments within the first six months on trial. On average, control patients incidence and grade by cycle illustrate this improves over time (34/7/4% gr
had 30 ED presentations and PROMPT-Care patients had 21 ED presentations 1/2/3+ at c1, 13/0/0% gr 1/2/3+ at c12). ToxT MGOT analyses demonstrate
per 10,000 patient days. After adjusting for age, sex and stage of disease, earlier time to gr 2+ symptomatic AE on MRL vs MR (63% vs 31% by day 50,
PROMPT-Care patients had 26% (95% CI 0.2%, 57%) fewer ED presentations p , .001) and suggest that overall AE burden over time is higher for patients
compared to the control group (p = 0.0483). Conclusions: The PROMPT-Care on MRL(AUC 18.2) than MR(11.8, p , .001). Conclusions: ToxT depicts AE
intervention resulted in significantly fewer ED presentations in a broad pop- time profile and can guide AE interventions. Summary metrics suggest that
ulation of cancer patients, including those currently undergoing cancer treat- symptomatic AEs occur earlier and their burden over time is higher on MRL.
ment or are in follow-up, and patients with a wide range of tumor types. The We are implementing ToxT in patient-reported AE data to better characterize
results support its utility as an improved model for ongoing supportive care for a pt tolerability.
diverse population, with potential healthcare cost savings. Clinical trial in-
formation: ACTRN12616000615482.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
370s Health Services Research, Clinical Informatics, and Quality of Care

6512 Oral Abstract Session, Sun, 8:00 AM-11:00 AM 6513 Oral Abstract Session, Sun, 8:00 AM-11:00 AM
Population-based cohort of prostate cancer patients on active surveillance (AS): Clinical benefit of breakthrough cancer drugs approved by the United States
guideline adherence, conversion to treatment and decisional regret. First Au- Food and Drug Administration. First Author: Consolacion Molto, Hospital de
thor: Sabrina Peterson, University of North Carolina School of Medicine, Chapel Sant Pau, Barcelona, Spain
Hill, NC
Background: The Breakthrough Therapy program was established in July 2012
Background: AS is recommended for early-stage prostate cancer, for which to expedite drug development and approval by the FDA. We compared the
over-treatment has been widely described. In published studies from large characteristics of clinical trials leading to FDA approval as well as the magnitude
academic institutions and/or controlled clinical trials, where patients are of clinical benefit and value framework scores of breakthrough-designated and
monitored rigorously, AS is safe and results in low rates of cancer-specific non-breakthrough-designated cancer drugs. Methods: We searched the Drugs@
mortality. However, active surveillance in the community setting has not been FDA website for cancer drug approvals from July 2012 and December 2017. For
previously examined. Methods: In collaboration with the North Carolina state each indication, we applied the value frameworks and used thresholds of high
cancer registry, 346 men with newly-diagnosed low- or intermediate-risk clinical benefit developed by American Society of Clinical Oncology Value
prostate cancer throughout the state from 2011–13 who pursued active Framework version 2 (ASCO VF v2; scores $45), the ASCO Cancer Research
surveillance were enrolled in an observational cohort; medical records and Committee (OS gains $2.5 months PFS gains $3 months), the European
patient-reported outcomes (validated measures of prostate cancer anxiety Society for Medical Oncology-Magnitude of Clinical Benefit Scale version 1.1
[MAX-PC] and Clark’s prostate cancer decision regret) were collected pro- (ESMO-MCBS v1.1; grade of A or B for trials of curative intent and 4 or 5 for
spectively. Guideline-adherent monitoring during active surveillance was those of non-curative intent), and the National Comprehensive Cancer Network
assessed using contemporary NCCN guidelines: PSA testing every 3–6 months (NCCN) Evidence Blocks (scores of 4 and 5). Trial characteristics and value
and prostate biopsy within 18 months of initial diagnosis. Results: 58% of framework scores were compared using Chi squared or Mann Whitney U tests.
patients received adequate PSA testing and 45% prostate biopsy; overall, 32% Results: We identified 106 pivotal trials supporting the approval of 52 indi-
of patients received guideline-adherent monitoring. Urology follow-up in Year 1 vidual drugs for 96 indications. Of these indications, 38 (40%) received
was 97% but dropped to 67% in Year 2. Within the first 2 years, 16% of breakthrough designation. Compared with trials for non-breakthrough drugs (n =
patients converted to treatment. Multivariable analysis showed MAX-PC scores 62), trials for breakthrough drugs (n = 44) had smaller sample size (median 373
(OR 1.8, p = 0.008) and younger age were significantly associated with vs 612, P= .03), were less often randomized (57% vs 86%; P= .001) and more
conversion; no other sociodemographic (race, education, marital status, rural/ likely to be open-label (84% vs 53%, P= .001). Trials for breakthrough drugs
urban) or diagnostic variable (risk group) was associated. At 2 years, 94% were more likely to demonstrate high clinical benefit using ASCO VF (68% vs
expressed no regret. Conclusions: In a non-controlled setting of patients 31%, P= .002) and NCCN Evidence Blocks (86% vs 56%, P= .002). A similar
pursuing AS in the community, adherence to guideline-recommended mon- proportion of trials supporting breakthrough and non-breakthrough drugs
itoring was only 32%. Few patients expressed decisional regret. Conversion to demonstrated high clinical benefit using the ASCO Cancer Research Committee
treatment was likely driven by patient anxiety but not disease-related factors. (82% vs 68%, P= .25) and ESMO-MCBS (35% vs 33%; P= .87) frameworks.
While there are continued efforts to increase AS uptake, these results highlight Conclusions: In patients with advanced solid tumors, cancer drugs approved
the importance of behavioral interventions during active surveillance to reduce under breakthrough therapy designation were more likely to demonstrate high
anxiety and improve monitoring adherence. Whether AS in non-controlled clinical benefit as defined by the ASCO VF and NCCN value frameworks. A
settings is safe and effective requires further study. similar proportion of approved breakthrough and non-breakthrough therapy
drugs met the high benefit thresholds using the ASCO Cancer Research
Committee and ESMO-MCBS frameworks.

6514 Oral Abstract Session, Sun, 8:00 AM-11:00 AM 6515 Oral Abstract Session, Sun, 8:00 AM-11:00 AM
The use of optimal evidence-based chemotherapy (chemo) regimens in phy- Integrating psychosocial care into routine cancer care: A stepped-wedge
sician offices versus hospital outpatient facilities. First Author: Michael Jordan design cluster randomized controlled trial (SWD-RCT) to evaluate effective-
Fisch, AIM Specialty Health, Chicago, IL ness of the HuCare Quality Improvement Strategy (HQIS) on health-related
quality of life (HRQoL). First Author: Rodolfo Passalacqua, Oncology Unit,
Background: The proportion of infused chemo administered in hospital out-
Oncology Department, ASST of Cremona, Cremona, Italy
patient facilities (HOF) increased from 6% in 2004 to 43% in 2014. The
average annual cost for patients receiving chemo was significantly higher in Background: Cancer patients (pts) often do not receive evidence-based psy-
HOFs than in physician offices (POs). One option to explore differences in the chosocial care. We evaluate the effects of an implementation strategy we
quality of care between these two settings is to examine the use of chemo previously demonstrated feasible, which includes communication skill training
regimens, which, based on their efficacy, toxicity, and costs, have been for all physicians and nurses; four support visits at the centers by an im-
designated as “on-pathway.” This study compared on-pathway rates among provement team to assist staff in identifying obstacles, finding solutions, and
patients receiving infused chemo administered in POs vs. those in HOFs. strengthening motivation; screening for distress and social needs; individu-
Methods: Using administrative claims data, we identified 61,496 breast, lung, alized pts’ education with a referring nurse; use of a question prompt list.
or colorectal cancer patients receiving chemo from 2013 to 2018. Chemo Methods: Multicenter incomplete SWD-RCT with 3 clusters of 5 centers each.
regimens were considered “on-pathway” when they were on payer’s program Consecutive outpatients requiring medical treatment and diagnosed in the
list of optimal regimens when administered. Generalized linear models ex- previous 2 months were eligible. Primary endpoint: difference of at least one of
amined the association between site of service and on-pathway prescribing the 2 domains of HRQoL emotional or social functions, at 3 months from
rates, and costs of care. Models adjusted for age, sex, year, rural status, baseline, in pts of the centers that implemented the HQIS vs standard of care
cancer type and setting, and comorbidities, with fixed effects for providers. (SoC). Secondary endpoints include: patient mood, long-term effect, overall
Results: Percentage of infused chemo administered in HOFs increased from HRQoL. Analyses were performed using a beta-binomial regression model.
44.2% in 2013 to 54.7% in 2018. After adjustment, on-pathway prescribing Results: 762 pts were enrolled. At baseline, 41% showed high anxiety (HADS-
rate did not differ significantly between HOFs and POs (50.1%, 95% CI: A.7), and 88% had at least one psychosocial need. 299 health professionals
48.6%-51.5% vs. 49.8%, 95%CI: 48.3%-51.3%, p = 0.65). 6-month chemo attended 3-day courses (84% of all clinical staff). 647 pts (85%) were
cost ($56,885, 95% CI: $54,364-$59,524 vs $32,240, 95% CI: $30,929- available for analysis. The 315 pts who received HQIS exhibited better quality
$33,605, p , 0.001) and 6-month medical cost ($114,280, 95% CI: of life for the emotional domain than those assigned to SOC (OR=1.115,
$110,716-$117,960 vs $79,455, 95% CI: $77,089-$81,893, p , 0.001) p=0.016). Pts who showed the greatest improvement were the older
were significantly higher in HOFs vs. POs. Conclusions: Quality of care as (OR=1.003, p=0.035), had lower anxiety basal levels (OR=0.853, p,0.001),
measured by use of optimal chemo regimens was similar in hospital and office and social needs were met (OR=1.182, p,0.001). The difference was not
setting. Cost continues to be significantly higher in hospital setting. These significant for the social domain (OR=0.955, p=0.353). The HQIS’s long-term
findings provide a strong basis for site-neutral reimbursement policies. effect was confirmed for the emotional domain at 12 months. No effect on
mood (HADS-D) and overall HRQoL was observed. Conclusions: To our
knowledge this is the first RCT demonstrating the effectiveness of a psy-
chosocial care implementation strategy on cancer patients’ emotional well-
being. Clinical trial information: NCT03008993.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Health Services Research, Clinical Informatics, and Quality of Care 371s

6516 Oral Abstract Session, Sun, 8:00 AM-11:00 AM 6517 Oral Abstract Session, Sun, 8:00 AM-11:00 AM
Randomized trial of text messaging (TM) to reduce early discontinuation of Impact of Electronically Accessible Pathways (EAPathways) on clinical trial
aromatase inhibitor (AI) therapy in women with breast cancer: SWOG S1105. enrollment at a large multisite cancer center. First Author: Jeryl Jean Villadolid,
First Author: Dawn L. Hershman, Columbia University Medical Center, New Levine Cancer Institute, Charlotte, NC
York, NY
Background: Clinical pathways streamline evidence-based treatment deci-
Background: Non-adherence to AI’s for breast cancer is common and in- sions and provide consistent, high-quality, value-based care. A high-quality
creases risk of recurrence. Text messaging (TM) has been shown to increase clinical pathway should enhance screening and access to clinical trials. Our
adherence to medications for chronic conditions. We conducted a multicenter healthcare system utilizes EAPathways to allow providers to select treatment
randomized trial to evaluate if TM reminders improve AI adherence. regimens vetted by section experts, inquire about clinical trials, and refer to
Methods: Patients taking an AI for $30 days and having $36 mos of planned relevant programs (e.g. palliative medicine) or testing (e.g. genomics) at a main
therapy were eligible. Patients were randomly assigned 1:1 to receive either TM cancer center and 22 regional sites. With over 400 clinical trials, our goal is to
or NO-TM twice a week for 36 mos. Randomization was stratified by length of provide access regardless of where a patient lives or receives treatment. We aim
time on prior AI therapy ( , 12 (64%) vs. 12-24 mos (36%)) and AI class to explore the impact of EAPathways on clinical trial enrollment at our
(anastrozole, letrozole, exemestane). Content themes of the 36 TMs focused on healthcare system. Methods: This study is a retrospective review to compare
overcoming barriers to adherence. Patients were assessed for discontinuation clinical trial inquiries through EAPathways and clinical trial enrollment using
of AIs every 3 mos for 36 mos. The primary outcome was time to non- Oncore between 1/1/2017 and 7/31/2018. The primary outcome is the
adherence, where non-adherence was defined as urine AI metabolite assay success rate reported as the total number of inquiries that resulted in clinical
results satisfying the following: , 10 [units], undetectable, or no submitted trial enrollment. Other outcomes include a comparison of inquiries and en-
specimen. A stratified Log-rank test was conducted. Multiple sensitivity an- rollments for hematology and solid tumor oncology, cancer treatment and non-
alyses were performed using Cox regression. Results: In total, 724 patients treatment (e.g. specimen collection), and our main cancer center and regional
were registered between May, 2012 and September, 2013, among whom 696 sites. The number of and reason for opting out of treatments or trials was also
(338/360 (93.9%) on TM and 338/364 (92.9%) on NO-TM) were eligible and analyzed. Results: A total of 29.1% (740/2539) of clinical trial inquiries
adherent at baseline. Observed (time-independent) adherence at 36 mos was through EAPathways resulted in clinical trial enrollment. Success rates for the
55.4% for TM and 55.4% for NO-TM. The primary analysis showed no dif- following settings were reported: 39.5% (223/564) in hematology, 26.2%
ference in time-to-adherence failure between patients on the TM and NO-TM (517/1975) in solid tumor oncology, 27.0% (594/2203) in treatment trials,
arms (HR = 0.89, 95% CI:0.76,1.05 p = .18). An analysis of negative urine and 43.5% (146/336) in non-treatment clinical trials. Sixty-three percent of
tests alone resulted in similar non-significant results. With missed appoint- enrollments were at our main cancer center compared to regional sites. A total
ments not counted as failures, time to self-reported discontinuation (89.6% vs. of 39.7% (3356/8453) of patients were enrolled into an opt-out pathway due
89.7%; HR = 1.17, 95% CI:0.69-1.98, p = .57) and site reported discon- to reasons such as performance status, organ dysfunction, or hospice.
tinuation (78.1% vs. 81.1%; HR = 1.31, 95% CI:0.86-2.01, p = .21) were Conclusions: Clinical pathways can provide access to clinical trial enrollment
also similar between the 2 groups. Conclusions: As the first large long-term in multiple settings. These baseline metrics will help assess process im-
randomized trial of an intervention directed at improving AI adherence, we provement needs to increase clinical trial enrollment success rates and address
found high rates of AI discontinuation. Bi-weekly text reminders did not im- reasons for opt-out.
prove adherence to AIs compared to usual care. Improving long—term ad-
herence will likely require sustained behavioral interventions and support.
Clinical trial information: NCT01515800.

6518 Poster Discussion Session; Displayed in Poster Session (Board #209), 6519 Poster Discussion Session; Displayed in Poster Session (Board #210),
Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Sat, 4:30 PM-6:00 PM Sat, 4:30 PM-6:00 PM
The financial impact of fractionation scheme and treatment planning method Quantification of the financial burden of antineoplastic agent price increases.
for rectal cancer in the United States. First Author: Assaf Moore, Tel Aviv First Author: Michail Alevizakos, University of Pittsburgh Medical Center,
University, Tel Aviv, Israel Pittsburgh, PA
Background: Preoperative long-course chemoradiotherapy (CRT) and short- Background: Antineoplastic medication prices are overall increasing yet this
course radiotherapy (SCR) for locally advanced rectal cancer (LARC) were phenomenon is not limited to new medications but can also be observed in
found to have equivalent outcomes in three randomized trials. SCR may have already established medications. Methods: We accessed the yearly payment
lower acute toxicity and the down-staging following CRT is more well- files from Medicare Part B for injectable antineoplastic agents (codes J8501-
established. At present, SCR is frequently used in Europe but has not been J9999) for the years 2010-2017 and all costs were adjusted to 2017 USD to
widely adopted in the United States (US). It is standard to deliver radiotherapy adjust for inflation. We then calculated the price-per-dose for every medication
by 3D planning, while the use of Intensity-modulated radiotherapy (IMRT) is and compared that price with the price-per-dose that the medication would
controversial. In recent years there has been an increasing focus on under- have if its initial price was only affected by inflation. We subsequently mul-
standing the cost and value of cancer care. In this study we aimed to assess the tiplied the difference with the total doses of the medication administered in
economic impact of fractionation scheme and treatment planning method for order to calculate the additional cost accrued by Medicare from medications
payers in the US. Methods: We performed a population-based analysis of the whose price had increased more than the inflation rates. Only medications with
total cost of radiotherapy for LARC in the US annually. The national annual total annual payments .10 million USD/yr were included in the analysis.
target population of patients was calculated using the Surveillance, Epide- Notably, Medicare provides reimbursement based on average U.S. market
miology, and End Results (SEER) database. Treatment costs for various prices. Results: Price increases were noted on average in 64.5% of already
fractionation schemes were based on billing codes and 2018 pricing by established medications (median 69.6%, range 45.4-74.1%), leading to an
Medicare’s Hospital Outpatient Prospective Payment System (OPPS). The cost average additional extra cost of 243 million USD per year (range 140-330
of chemotherapy was based on the Payment Allowance Limits for Medicare million USD), for a total of 1.7 billion USD over the 7 years of observation.
Part B Drugs by Centers for Medicare and Medicaid Services (CMS). Rituximab (539 million USD), trastuzumab (221 million USD), and bev-
Results: We estimate that 12,945 patients with LARC are treated with ra- acizumab (178 million USD) accrued the highest extra costs. This extra cost
diotherapy annually in the US. The cost of CRT with 3-D or IMRT is US$ represented 4.6-8.9% of the annual total Medicare Part B spending for an-
15,881.76 and US$ 23,744.82 per patient, respectively. With 3-D SCR the tineoplastic medications (Table). Conclusions: The majority of already
cost is US$ 5,457 per patient. The use of SCR would lead to 64-77% annual established injectable chemotherapeutics demonstrate price increases that
savings of US$ 125,701,387 - US$ 236,727,934 in the US compared with 3- lead to substantial additional financial cost to Medicare and likely other U.S.
D and IMRT based CRT, respectively. IMRT based planning increases the total markets as well. Price increases of Medicare Part B antineoplastic medications
cost of CRT by 49% and if adopted widely would lead to an excess cost of US$ with cost .10 million USD/yr.
101,787,312 annually. Conclusions: SCR may have the potential to save in
Medications with Already available % medications with Extra cost due to Total % extra cost/
the region of US$ 0.12-0.23 billion annually in the US, likely without Year price increase medications price increase price increases1 spending1 total spending
impacting outcomes. Lack of evidence showing benefit with costly IMRT 2011 16 23 69.6% 225 3684 6.1%
should limit its use to clinical trials. SCR may also lead to lower personal 2012 14 24 58.3% 210 3412 6.2%
2013 10 22 45.4% 140 3064 4.6%
financial toxicity. It would be reasonable for public and private payers to 2014 13 23 56.5% 184 3052 6%
consider which type of radiation is most suited to reimbursement. 2015 20 28 71.4% 289 3240 8.9%
2016 18 25 72% 330 3729 8.8%
2017 20 27 74.1% 321 3965 8.1%
1
In million USD

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
372s Health Services Research, Clinical Informatics, and Quality of Care

6520 Poster Discussion Session; Displayed in Poster Session (Board #211), 6521 Poster Discussion Session; Displayed in Poster Session (Board #212),
Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Sat, 4:30 PM-6:00 PM Sat, 4:30 PM-6:00 PM
Trends in financial relationships between industry and individual medical Comparative effectiveness of proton therapy versus photon therapy as part of
oncologists in the United States from 2014 to 2017: A cohort study. First concurrent chemoradiotherapy for locally advanced cancer. First Author:
Author: Deborah Catherine Marshall, Icahn School of Medicine at Mount Brian Christopher Baumann, Washington University in St. Louis, Department
Sinai, New York, NY of Radiation Oncology, St. Louis, MO
Background: Industry-physician financial relationships in medical oncology Background: Concurrent chemo-radiotherapy is standard-of-care curative
are common and introduce conflicts of interest. The Open Payments (OP) treatment for many cancers, but is associated with substantial morbidity.
program collects and discloses data on industry payments to physicians, in Proton therapy may increase the tolerability or effectiveness of concurrent
part to discourage inappropriate relationships. However, the effect of OP on chemo-radiotherapy by reducing radiation to normal tissues. Methods: We
how oncologists engage with industry is unknown. Our aim was to evaluate conducted a comparative effectiveness study of adult non-metastatic cancer
trends in physician-level payments to test whether the implementation of OP patients treated with curative intent with proton chemo-radiotherapy vs.
has resulted in fewer physicians engaging with industry and has shifted the photon chemo-radiotherapy from 2011-2016 at the University of Penn-
nature of interactions towards those considered more appropriate. sylvania. Re-irradiation and disease sites treated with photon-only therapy
Methods: We performed a retrospective cohort study of US medical oncol- were excluded. Data on adverse events and survival was gathered pro-
ogists in 2014 from the National Plan and Provider Enumeration System. OP spectively. Primary endpoint was 90-day adverse events associated with
data for general (non-research) payments between 2014-2017 were unplanned hospitalizations (CTCAEv4 grade $3 adverse events). Secondary
matched to physician to evaluate receipt of payments over time. We cal- endpoints included decline in ECOG performance status during treatment,
culated the percentage of physicians receiving payments, annual value and 90-day grade $2 adverse events, disease-free survival (DFS) and overall
number of payments, and average annual trends over time, including by survival (OS). Modified Poisson regression models with inverse propensity
nature of payment. Results: From 2014-2017, medical oncologists received score weighting were fit for both outcomes. Propensity scores were estimated
1.4 million industry payments totaling $330.6 million. The absolute number using an ensemble machine-learning approach. Results: 1,483 patients
of medical oncologists receiving payments decreased 4% on average an- were included (391 proton/1,092 photon). Proton patients were significantly
nually (P= .006), and proportionally from 67.2% to 59.6% overall. The value older (median 66 vs. 61), had less favorable Charlson-Deyo comorbidity
and number of payments have not significantly changed. The value and scores (median 3.0 vs. 2.0), and had lower integral radiation dose to tissues
number of payments increased for accredited/certified CME (+821% and outside the target (p , 0.05 for all). Baseline toxicity and performance status
+209% annually) and decreased for non-accredited/certified CME (-18% were similar (p . 0.05). In propensity score weighted-analyses, proton
and -25% annually). The value and number of food/beverage payments chemo-radiotherapy was associated with significantly lower relative risk (RR)
remained the same. The value and number of royalty/licensing payments of 90-day grade $3 adverse events [11.5%(95%CI 8.3-14.7%) for protons;
increased. Conclusions: Fewer oncologists are receiving payments, but 27.6%(95%CI 24.9-30.2%) for photons; RR 0.31, 95%CI 0.15-0.66,
spending has not decreased suggesting that physicians are less likely to p , 0.01]; 90-day grade $2 adverse events (RR 0.78, 95%CI 0.65-0.93,
engage and industry is more selective. Increased payments for accredited p , 0.01); and decline in performance status during treatment (RR 0.51,
CME suggest that less appropriate speaker’s fees are being avoided. Food/ 95%CI 0.37-0.71, p , 0.01). There was no difference in DFS or OS.
beverage payments are not decreasing, thus these interactions may not be Conclusions: In adults with locally advanced cancer, proton chemo-
recognized as problematic. Increasing royalty/licensing payments require radiotherapy was associated with significantly reduced acute adverse
ongoing scrutiny. Changes in physician payments since the inception of OP events causing unplanned hospitalizations with similar disease-free and
highlight the importance of transparency in policymaking. overall survival.

6522 Poster Discussion Session; Displayed in Poster Session (Board #213), 6523 Poster Discussion Session; Displayed in Poster Session (Board #214),
Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Sat, 4:30 PM-6:00 PM Sat, 4:30 PM-6:00 PM
Enhancing community capacity to improve cancer care delivery and the effect Early case ascertainment and prospective multidisciplinary review for man-
on patient-reported outcomes, healthcare utilization and total costs of care. agement of new melanoma diagnoses within an integrated healthcare system:
First Author: Manali I. Patel, Division of Oncology, Department of Medicine, The Kaiser Permanente Northern California experience. First Author:
Stanford University School of Medicine, Stanford, CA Thach-Giao Truong, Kaiser Permanente Northern California, Vallejo, CA
Background: To curb rising expenditures and improve patient-reported outcomes Background: Appropriate surgical treatment of early-stage melanoma yields a
(PROs), we designed an intervention with patient, caregiver, provider, and payer high cure rate, but this management can be nuanced. In particular, surgical
input. The intervention is based on prior work using a lay health worker (LHW) to management, including sentinel lymph node biopsy (SLNB), of thin melanoma
assess advanced cancer patients’ symptoms. In this study, we trained the LHW (#1.0mm) is not well-defined. Methods: Biopsies with new melanoma di-
to refer patients to palliative care and/or behavioral health services in response to agnoses were identified electronically and manually reviewed. In a community
positive assessments and expanded the intervention to all cancer stages. We oncology setting, we organized a review panel of physicians specialized in
implemented the intervention with a health plan and a community oncology group melanoma from dermatology, medical oncology, nuclear medicine, radiation
serving elderly racially/ethnically diverse patients to test the effect on symptoms, oncology, and surgical subspecialties (oncology, plastics, head and neck).
healthcare use, and total costs. Methods: We enrolled all newly diagnosed health Patients were assigned to care pathways based on NCCN and ASCO guidelines,
plan beneficiaries with solid and hematologic malignancies from 10/2016 to including guidance on SLNB for thin melanomas with high-risk features like
11/2017 and compared outcomes to all cancer patients diagnosed in the lymphovascular invasion, high mitotic rate, positive deep margin, and ulcer-
year prior to the intervention (control arm). Our primary outcome was change in ation. These recommendations were documented in the chart and commu-
patient-reported symptoms using the Edmonton Symptom Assessment Scale and
nicated directly to the patient s care team. Results: From 11/2016 through 10/
Personal Health Questionnaire-9 at baseline, 6 and 12 months post-enrollment.
2018, our multidisciplinary committee reviewed 3626 patients with new
Secondarily, we compared 12 month healthcare use and costs. All generalized
linear regression models were adjusted for age, stage, comorbidities, diagnosis,
melanoma from 22 sites in our integrated, regional hospital system. Median
and follow-up. Results: 425 patients were in the intervention; 407 in the control. age was 66 (range 19-99); 60% were male. cT2N0 tumors comprised 7%, cT3
In both groups, mean age was 79 years; 48% were non-Hispanic White, 43% 3%, and cT4 2%. Thin melanomas #1.0mm represented 71% of cases, of
Hispanic, 3% Black, 6% Asian/Pacific Islander; 60% had advanced stages; 28% which 34% were #0.5mm. SLNB was performed in 9.8% of thin melanomas,
had breast, 28% had gastrointestinal, and 10% had thoracic cancers. In- and 18% were positive, much higher than historical positive rates of 3-4%.
tervention patients had significantly decreased symptom burden over time as Conclusions: Early case ascertainment and prospective multidisciplinary re-
compared with the control (Mean Difference: intervention (-0.77 +/- 0.28 p = view in a community oncology setting resulted in increased identification of
0.01) vs. control: (0.45 +/- 0.25 p = 0.06)); difference in difference: (-0.68 +/- high-risk thin melanoma, and consequently increased identification of nodal
0.25 p = 0.007)). Depression scores also significantly decreased over time among disease through SLNB. Positive SLNB triggers important clinical decision-
intervention patients as compared with the control (Mean Difference: intervention making regarding need for node dissection versus clinical surveillance, and
(-1.10 +/- 0.38 p = 0.04)) vs control: (1.21 +/- 0.34 p = 0.01); (difference in need for adjuvant therapy, which have been shown to improve survival. This
difference: -2.03 +/- 0.3 p , 0.001)). As compared to the control arm, in- clinical practice structure improved risk-stratification and adherence to na-
tervention patients had lower inpatient admissions (0.7 vs. 0.5 p = 0.01) and tional guidelines. We plan to further study the impact of these improvements to
emergency department visits per thousand patients per year (0.6 vs. 0.42 p = melanoma care on disease-free survival and overall survival.
0.02), and lower median total healthcare costs ($32,270 versus $25,512 p =
0.01). Conclusions: An LHW intervention significantly improved patient-reported
outcomes and the value of cancer care delivery and may be a solution to improve
burdensome and costly care for patients.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Health Services Research, Clinical Informatics, and Quality of Care 373s

6524 Poster Discussion Session; Displayed in Poster Session (Board #215), 6525 Poster Discussion Session; Displayed in Poster Session (Board #216),
Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Sat, 4:30 PM-6:00 PM Sat, 4:30 PM-6:00 PM
Effect of exercise during adjuvant chemotherapy for breast cancer. First Clinical outcomes and cost-effectiveness of breast cancer screening for child-
Author: Barbara K Haas, The University of Texas at Tyler, Tyler, TX hood cancer survivors treated with chest radiation: A comparative modeling
study. First Author: Jennifer Yeh, Boston Children’s Hospital and Harvard
Background: In 2018, an estimated 266,120 women faced the challenge of
Medical School, Boston, MA
living with breast cancer and approximately 40,920 died from their disease.
Nearly 100% of these women experienced significant treatment related side Background: Survivors of childhood cancer previously treated with chest ra-
effects that negatively impact quality of life (QOL). Exercise has repeatedly diation face elevated breast cancer risk similar to BRCA1 carriers. Children’s
demonstrated to alleviate many of the side effects, improve QOL, and de- Oncology Group (COG) guidelines recommend annual mammography with
crease cancer recurrence and mortality. In spite of this evidence, women do breast MRI, yet the benefits and costs of various screening strategies are
not maintain exercise during treatment for breast cancer. The purpose of this uncertain. Methods: We used two breast cancer simulation models (Model 1
randomized controlled trial was to determine the effectiveness of exercising and 2) from the Cancer Intervention and Surveillance Modeling Network
the day chemotherapy is administered on 1) the persistence with an exercise (CISNET) and data from the Childhood Cancer Survivor Study to reflect high
program, 2) side effects, and 3) QOL. Methods: Eligible women were ran- breast cancer and competing mortality risks among survivors. We simulated
domly assigned to a control or experimental group. As part of their treatment 3 screening strategies: annual mammography with MRI starting at age 25
plan, all participants were referred to one of 14 community-based exercise (COG25), annual MRI starting at 25 (MRI25), and biennial mammography
centers to exercise. Experimental group also exercised at one of two cancer starting at 50 (Mammo50). Performance of mammography+/-MRI was based
centers each day chemotherapy was administered. Outcome measures in- on published studies in BRCA1/2 carriers who have similar cancer risk. Costs
clude exercise retention and chemotherapy, completion; cancer-related and quality of life weights were based on US averages and published studies.
fatigue, nausea/vomiting, peripheral neuropathy, weight gain, and QOL. Results: Among a simulated cohort of 25-year-old survivors treated with chest
Outcome measures were assessed prior to every second course of chemo- radiation, the lifetime breast cancer mortality risk in the absence of screening
therapy and 3- and 6-months post-chemotherapy. Results: 273 women with was 10-11% across models. Compared to no screening, Mammo50, MRI25,
Stage I-III breast cancer receiving chemotherapy were enrolled in the study. and COG25 screening avert approximately 23-25%, 56-62% and 56-71% of
The number of participants who withdrew from exercise was higher among deaths, respectively; averted deaths for COG25 compared to MRI25 were
those in the control group (n= 16; 12.4%) compared with those in the higher in Model 1 than Model 2 (9% vs. ,1%). In Model 1, both MRI25 and
experimental group (n = 10; 6.9%). At cycle 5, those in the experimental COG25 were cost-effective; in Model 2, MRI25 was preferable (more effective,
group reported less motor peripheral neuropathy than those in the control less costly than COG25). Conclusions: Compared to no screening, initiating
group (p = .018) and higher physical well-being scores than those in the annual screening at younger ages for at-risk survivors averts .50% of breast
control group (p = .047). Conclusions: The highest impact of the intervention cancer deaths and is cost-effective. Additional data on test performance are
was on attrition from exercise. Since participants in both the control and needed to inform recommendations on screening modality.
experimental groups exercised throughout chemotherapy, it is not surprising
Model 1 Model 2
that the groups performed comparably with regard to side effects and QOL.
False- False-
Given the positive effects exercise has demonstrated on persons receiving positive Incremental QALYs positive Incremental QALYs
cancer treatment in numerous studies, having patients exercise in the cancer Strategy tests* Costs*† gained*† ICER tests* Costs*† gained*† ICER
center on the day of chemotherapy is a significant step toward engaging Mammo50 259 – – – 257 – – –
No screening 0 1,033,840 -74.4 ‡ 0 839,750 -65.4 ‡
persons receiving chemotherapy in an exercise program. MRI25 3283 5,629,340 285.3 $19,730 3764 6,651,870 175.4 $37,920
COG25 4188 8,439,630 350.5 $43,100 4879 9,463,560 171.6 ‡

ICER, incremental cost-effectiveness ratio ($/QALY) *Per 1000 †Discounted ‡Dominated

6526 Poster Discussion Session; Displayed in Poster Session (Board #217), 6527 Poster Discussion Session; Displayed in Poster Session (Board #218),
Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Sat, 4:30 PM-6:00 PM Sat, 4:30 PM-6:00 PM
Previsit breast cancer educational microlearning videos: Impact on patient Growth factor use and rate of neutropenic complications in breast cancer
satisfaction and engagement. First Author: Rachel M Hurley, Mayo Clinic patients treated with dose-dense paclitaxel: A 5-year experience from a safety
Alix School of Medicine, Rochester, MN net hospital. First Author: Racha Halawi, University of Texas Southwestern
Medical School, Dallas, TX
Background: Most patients diagnosed with breast cancer turn to the Internet
to learn about their diagnosis; however, information online is often generic, Background: The NCCN guidelines recommend growth factor (G-CSF) support
challenging to navigate, and not expert-curated. To facilitate patient education to reduce the risk of febrile neutropenia and maintain dose density in pa-
and engagement in our breast clinic, we piloted the implementation of pre-visit tients receiving dose dense chemotherapy. We retrospectively reviewed growth
education via brief microlearning videos organized within an online platform. factor utilization with dose dense paclitaxel (ddT) in breast cancer patients
Methods: Seventeen videos of 2-4 minute duration were developed by mul- treated at our institution. Methods: Electronic medical records of patients
tidisciplinary content experts. Videos covered a variety of educational topics treated at Parkland Health and Hospital System between 2012-2017 for
relevant to breast cancer, including treatment options. Patients received a link breast cancer with dose dense adriamycin and cyclophosphamide (ddAC)
via email to create an account, which provided access to the platform. Ag- followed by ddT were reviewed. Data on patient characteristics as well as G-
gregate viewing data and optional patient surveys (online after viewing 3 videos CSF use and neutropenic complications were collected. Results: Two-hundred
and at the clinical appointment) were used to assess opt-in rates, engagement, sixty eight patients received a total of 1019 cycles of ddT. Only one physician in
and satisfaction. Results: Between September 2018 and January 2019, the practice routinely prescribed G-CSF after ddT. The majority of ddT cycles
57.4% (240/418) of women with biopsy proven breast cancer who were sent were administered without G-CSF support (781 vs 238 cycles). There were no
an email invitation registered on the platform. On average, patients watched 11 episodes of neutropenic fever in either group. The rate of grade 3/4 neutropenia
of the 17 videos, with 93.7% of users (225/240) viewing at least one. Overall, was 2.1 % with G-CSF support (all grade 3), and 2.7% without G-CSF support
85% (168/198) of women recommend the microlearning format for patient (85% grade 3), p = 0.61. Treatment delays were longer in patients who did not
education. The most-viewed video topics included types of breast cancer, receive G-CSF support, but this difference was not statistically significant
breast abnormality and biopsy, understanding biopsy results, tumor markers, (mean of 4 vs 2.2 days, p = 0.07). The number of cycles needed to treat to
and staging. Seventy-eight percent (154/198) of women reported that they prevent 1 episode of grade 3/4 neutropenia was 167. Based on Medicare
planned to share the videos with family or caregivers, and 67% (133/198) felt average sales price (ASP) for pegfilgrastim, routine use of G-CSF in our patient
that the educational content increased their satisfaction with their overall population would have added over $3.6M to the cost of care over the study
experience. Barriers to video access were emails marked as junk, not having an period. Conclusions: Our results show a similarly low rate of neutropenic
email address, and difficulties with videos loading. Conclusions: The majority complications in patients receiving dose dense paclitaxel with or without G-
of patients participating in this pilot registered on the platform and watched CSF support. Therefore routine use of G-CSF with this regimen is not war-
pre-visit microlearning educational videos. These expert-curated videos con- ranted. Judicious use of expensive medications such as G-CSF would reduce
tributed to high levels of satisfaction and engagement. Strategies to overcome the cost of care and financial toxicity to patients, and promote high value care.
barriers to access will be developed to expand the reach of this new valuable
component of breast cancer care.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
374s Health Services Research, Clinical Informatics, and Quality of Care

6528 Poster Discussion Session; Displayed in Poster Session (Board #219), 6529 Poster Session (Board #220), Sat, 1:15 PM-4:15 PM
Sat, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Evaluation of a risk-based approach to follow-up care for breast cancer survivors:
Sat, 4:30 PM-6:00 PM From oncology specialist to primary care provider (PCP). First Author: Kathleen
Improving time to initiation of bone modifying agents in patients with newly Keenan, Memorial Sloan-Kettering Cancer Center, New York, NY
diagnosed multiple myeloma. First Author: Nathaniel Rosko, Cleveland
Background: The number of cancer survivors in the Unites States is expected
Clinic, Cleveland, OH
to exceed 18 million by the year 2020. Predicted care demands coupled with
Background: Current ASCO and IMWG guidelines recommend that all pa- an anticipated shortage of oncologists have led to alternate follow up care
tients (pts) on active anti-myeloma therapy receive concurrent supportive models and providers. This nurse practitioner- (NP) led survivorship team
care treatment with a bone modifying agent (BMA) to decrease the risk of was charged with developing a risk-based discharge plan for the . 8,000
skeletal related events (SRE). Unfortunately, recent data shows only 51% of breast cancer survivors in our survivorship clinic at this NCI-designated cancer
Medicare pts with myeloma received a BMA within 90 days of first che- center. Methods: In collaboration with breast oncologists and surgeons, a risk-
motherapy. We implemented a quality improvement project to identify the based strategy to identify and transition eligible survivors to their PCPs was
average time to BMA initiation at all Cleveland Clinic affiliated sites, with the developed. Low risk was determined by: stage, absent disease and significant
primary goal to improve time to initiation of BMA in newly diagnosed multiple late-effects, and time since diagnosis. Interventions: Evidence supporting the
myeloma (NDMM) pts by 4 weeks at Cleveland Clinic Main Campus (MC). willingness, availability and expectations of PCPs to provide follow-up care to
Methods: Barriers with BMA initiation were identified using quality im- cancer survivors was reviewed. Criteria for eligibility were developed; patient
provement tools developed at the ASCO Quality Training Program. The first and PCP educational documents were created. Continued breast cancer
PDSA cycle was implemented between September 2018 – January 2019 screening at our center was offered with results sent directly to PCP. Outreach
with an emphasis on education. This included review of updated guidelines, educational programs were provided to local PCP groups to build relationships
literature review of risks and toxicities associated with BMAs, and strategies and provide breast specific follow up recommendations. To promote com-
for choosing BMA based on pt factors. Baseline data on time to start BMA munication and collaboration between NP and PCP prior to and at the time
and data to evaluate impact of PDSA intervention was completed via chart of transition, clinic notes were shared after each survivorship visit. A com-
review. Results: 161 NDMM pts were evaluated between 2015 to 2018 at all prehensive transition note including detailed follow-up recommendations
sites. The average time difference between the start of anti-myeloma therapy was provided at the time of transition. Patients were assured of rapid return
and the start date of a BMA in NDMM pts was 10.5 weeks. Subset analysis to the oncology provider in the event of recurrence or cancer related issue.
based on whether pts were treated at MC vs affiliate sites was 10.6 weeks vs Results: 5080 patients were offered transition, with 3642 accepting. Miti-
9.1 weeks, respectively. During the first PDSA cycle, 14 NDMM pts were gating factors and facilitators to transition for both patients and PCP will be
treated at MC. 86% (12/14) pts were treated with a BMA. The average time discussed and educational strategies to overcome barriers will be described.
between cycle 1 day 1 of first line treatment and first dose BMA was Conclusions: Transition of breast cancer survivors’ care to PCPs is successful
4.3 weeks (range 4-12 weeks). Conclusions: With increased physician ed- approximately 56% of the time. Key factors include seamless communication
ucation and awareness of internal baseline data, we achieved our initial goal among providers and early discussions with patients to set expectations and
and observed a significant improvement in time to initiation of BMA from normalize the transition early in the care pathway.
10.5 weeks to 4.3 weeks. Obstacles regarding effective communication with
patients on the benefit of BMAs as well as need for dental clearance were
barriers identified early on. We plan to incorporate BMA guidelines in our
institutional care path with the goal to decrease time to initiation at all
affiliated practices. Further mechanisms to ensure reinforcement of BMA
initiation in NDMM patients is warranted to maintain therapeutic benefit.

6530 Poster Session (Board #221), Sat, 1:15 PM-4:15 PM 6531 Poster Session (Board #222), Sat, 1:15 PM-4:15 PM
Association of providers’ prescribing patterns with postsurgical opioid use Use, attitudes, and perceptions of tumor genomic testing: Survey of TAPUR
among cancer patients undergoing curative-intent surgery. First Author: physicians. First Author: Suanna S. Bruinooge, American Society of Clinical
Yuan Xu, University of Calgary, Calgary, AB, Canada Oncology, Alexandria, VA
Background: Patients with cancer are vulnerable to chronic opioid use. Al- Background: This survey of Targeted Agent and Profiling Utilization Registry
though opioid use may be appropriate, preliminary data suggest that a sig- (TAPUR) Study physicians examined use, attitudes, and perception of tu-
nificant proportion may be using opioids inappropriately. This study aims to mor genomic testing (TGT), defined as any DNA test performed on tumor
evaluate the association between the history of the providers’ opioid-prescribing specimen/plasma. TAPUR is a multibasket study of marketed agents tar-
patterns and post-surgical opioid use in cancer patients undergoing curative- geting tumor genomics. Methods: 333 physicians at 54 TAPUR sites were
intent surgery. Methods: This population-based study included all patients surveyed (2016-2017). Survey domains included use of TGT, barriers to
diagnosed with common solid tumors who received curative-intent surgery and ordering TGT, and genomic confidence. Surveys included 3 scenarios for
were non-opioid users prior to surgery between 2009 and 2015 in Alberta, TGT ordering 1) pretreated advanced cancer patients (pts) without options,
Canada. Based on previously published methods, a new persistent opioid user 2) newly diagnosed, untreated, metastatic pts and 3) early stage/potentially
was defined as opioid-naı̈ve prior to surgery and who subsequently filled at least curable pts with standard options. Data were analyzed with descriptive sta-
one opioid prescription between 60 and 180 days after surgery. The opioid- tistics. Results: 112 physicians responded (33%). The table displays demo-
prescribing patterns of a patient’s most responsible provider (MRP) were graphics and genomic confidence. Respondents reported a median of 25% of
measured as the mean daily dosage (oral morphine equivalent, OME) that was their pts had TGT in past 12 months for trials/routine care (range 0-85%).
prescribed to all other patients by that provider prior to the surgical date. Barriers to testing included access to tumor specimen (86%), insurance
Multivariable logistic regression was performed to identify associations between coverage (67%), concerns that results will not be actionable (55%), and test
the MRP’s prescribing patterns and the patient’s opioid use after surgery. issues (wait time, unsure which test/lab to use, test accuracy) (54%). TGT was
Results: 14,780 patients met the inclusion criteria and were associated with ordered most often for scenario 1 (96%) and 2 pts (70%). Few respondents
2,880 MRPs, among which 2,364 (16%) patients became new persistent (32%) would order testing in scenario 3. Of those who reported testing for
opioid users after surgery. Multivariate analysis demonstrated that patients with scenarios 1 & 2, most told pts that results could inform treatment/prognosis/
MRPs who routinely prescribed higher doses of opioids ($60 vs. 0-59 mg OME: trials (97%) or may be uninformative (84%). In all scenarios, pt expectations of
OR = 2.33, P , 0.0001) for their patients were associated with a greater risk of TGT results were discussed prior to testing. A minority reported frequently
new persistent opioid use after surgery. In addition, those with a higher Charlson telling pts in advance that results could inform heritable cancer susceptibility
comorbidity index (P = 0.006), visited more prescribers (P , 0.0001), had a (37%). Conclusions: Confidence in using TGT was high. TGT was performed
specific tumor type (breast, colorectal, lung, prostate, melanoma or kidney vs. most for pts with advanced cancer and few options. Availability of specimens
others, P , 0.0001), received adjuvant chemotherapy (OR = 1.37, P , was largest barrier reported, indicating the importance of blood-based tests.
0.0001), and received adjuvant radiation (OR = 1.3, P = 0.0004) were also Few respondents discussed implications of germline findings in advance,
associated with greater risk of new persistent opioid use after surgery. despite growing evidence of germline findings in somatic testing.
Conclusions: Our results suggest that prescribers with a history of prescribing
higher opioid doses are an important predictor of chronic opioid use among Male 70%
cancer patients undergoing curative-intent surgery. Awareness of physician Hem/Oncs 89%
prescribing practices and their unintended consequences may inform strategies >15 yrs after med school 66%
to minimize persistent post-operative opioid use in cancer patients. Somewhat/very confident TGT knowledge 98%
Somewhat/very confident ability to recommend treatment from 95%
TGT results
Somewhat/very confident ability to explain TGT results to pts 99%

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Health Services Research, Clinical Informatics, and Quality of Care 375s

6532 Poster Session (Board #223), Sat, 1:15 PM-4:15 PM 6533 Poster Session (Board #224), Sat, 1:15 PM-4:15 PM
The effects of immunotherapy and novel therapies on medical oncology work A prospective blinded study of 1000 cases analyzing the role of artificial
load in a Canadian province. First Author: Ravi Ramjeesingh, Division of intelligence: Watson for oncology and change in decision making of a Mul-
Medical Oncology, Nova Scotia Cancer Center, Dalhousie University, Halifax, tidisciplinary Tumor Board (MDT) from a tertiary care cancer center. First
NS, Canada Author: S.P. Somashekhar, Manipal Comprehensive Cancer Center, Banga-
lore, India
Background: Both novel targeted therapies and immunotherapies have dramatically
changed the landscape in a number of disease sites with previously limited treatment Background: Artificial intelligence is being used to provide support for
options. This has resulted in an impact on clinical workload for oncologists with information-intensive decision making. In this report, we present our ex-
subspecialty practices in the areas of non-small cell lung, (NSCLC), melanoma (M), and perience in explaining how artificial intelligence adds value to MDT’s de-
genitourinary (GU) cancer. Our aim was to investigate the shift in workload amongst
cision making ability & paves way for personalized therapy. Methods: 1000
these practices as compared to other disease sites within a single academic cancer
center in Nova Scotia (NS), Canada. Methods: The NS Cancer Center is the academic
cases involving breast, lung, and colorectal cancer were evaluated by a
cancer center for the province of NS providing consultative and ongoing care for multidisciplinary tumor board at a major cancer center in India between
approximately 72% of provincial patients. We manually quantified appointment visits 2016 and 2018. After the tumor board decision was made, MDT was
(new consultation, treatment and follow up visits) as well as telephone toxicity and chart presented with the Watsons recommendations contemporaneously. MDT
checks booked from February 1 to April 30 across a 3-year interval (2016, 2017, and reviewed their decision after going through Watson’s recommendations and
2018) and then extrapolated this data to derive full year estimates. Disease sites most also the evidences that it put forth supporting its decision. Cases in which
impacted by therapies that have changed treatment landscape (NSCLC, M and GU) decision was changed, objective assessment was done by asking MDT to
were compared with the Breast and Gastrointestinal disease sites. Results: Clinical quote the reasons for reviewing and changing their decision. Results: Of
workload increased across all domains over the 3 year period but the majority of the 1000 cases, breast, lung, colon & rectal cancers were 620, 130,126 & 124
increase is attributed to the 3 disease sites (Table). Conclusions: Medical oncology respectively. There were 712 non-metastatic & 288 metastatic cases. Mean
workloads are increasing over time and novel treatments (including immunotherapy) in
age of the patients was 54.3 6 12.2. Treatment concordance was observed
disease sites with previously limited options likely account for a significant portion of
that increase. New patient consultation metrics, taken in isolation, do not reflect
in 92% for all cancers combined, 93% for rectal cancer, 92% for breast
current trends in medical oncology workload. Hiring practices, space allocation and use cancer, 89% for lung cancer, and 81% for colon cancer.MDT changed their
of physician extenders must take into account these shifting workload dynamics to decision in 136 cases (13.6%). The reasons for tumour board to change their
mitigate physician burnout and potential impacts on quality and timeliness of care. decision was, Watson provided recent evidences for newer treatment in
55%, better personalized alternative in 30% & new insights from genotypic
3 months 1 year Estimates
% increase and phenotypic data and evolving clinical experiences in 15% of time.
2016 2017 2018 2016 2017 2018 2016 - 2018 Conclusions: The study suggest that cognitive computing decision support
Combined cohort: system holds substantial promise to reduce the cognitive burden on on-
Teletoxicity 150 145 228 600 580 912 52%
Chart Review 1409 1495 1737 5636 5980 6948 23.3% cologists by providing expert, updated, recent evidence-based insights for
Consultations
Follow up visits
514
2175
560
2216
554
2505
2056
8700
2240
8864
2216
10020
7.8%
15.1%
treatment-related decision-making. The 13.6 % incremental advantage over
NSCLC/M/GU: and above in a tertiary cancer centre with functioning MDT speaks in itself
Teletoxicity 93 85 181 372 340 724 94.6%
Chart Review 924 948 1225 3696 3792 4900 32.6%
the value of having a learned colleague like Watson for oncology at our
Consultations 219 251 287 876 1004 1148 31.1% disposal. It will certainly add more value in settings lacking ready access to
Follow up visits 1104 1149 1524 4416 4596 6096 38.0%
GI/Breast: high quality cancer expertise and information. These systems can be
Teletoxicity 57 60 47 228 240 188 -17.5% valuable adjuncts to strong patient-clinician relationships in the delivery of
Chart Review 485 547 512 1940 2188 2048 5.6%
Consultations 295 309 267 1180 1236 1068 -9.5% high quality cancer care.
Follow up visits 1071 1067 981 4284 4268 3924 -8.4%

6534 Poster Session (Board #225), Sat, 1:15 PM-4:15 PM 6535 Poster Session (Board #226), Sat, 1:15 PM-4:15 PM
Leveraging a conversational agent to support adherence to oral anticancer Risk stratification and daily symptom monitoring for oncology patients. First
agents: A usability study. First Author: Bethany Mooney Berges, Hospital of Author: Robert Michael Daly, Memorial Sloan Kettering Cancer Center, New
the University of Pennsylvania, Philadelphia, PA York, NY
Background: Identifying effective, scalable strategies to ensure patient ad- Background: Monitoring and managing patient reported outcomes (PROs) has
herence to oral anticancer agents (OACAs) is a major challenge. Previous been recommended for oncology patients on active treatment but can be time
studies have shown widely variable rates of adherence, and suboptimal ad- and resource intensive. Identifying patients likely to benefit and the optimal
herence is associated with decreased effectiveness and higher costs. A small frequency of PRO capture is still under investigation. We tested the feasibility of
but growing literature supports digital health behavioral interventions across a monitoring patients who are high-risk risk for acute care with daily PROs.
variety of chronic illnesses, including in cancer. In particular, conversational Methods: Using data from our institution, we developed a model that employs
agents—or technologies that mimic human conversation using text or spoken over 400 clinical variables to calculate a patient’s risk of an emergency room visit
language—have shown early promise in supporting behavior change, but have within 6 months following the onset of treatment. From October 15, 2018 to
January 23, 2019, we enrolled patients identified as high risk through a
yet to be rigorously tested in the context of OACAs. Methods: A rapid cycle
technology-enabled program to monitor and manage those patients’ symptoms.
prototyping approach led to the development of ‘Penny’ – a bidirectional,
Enrolled patients entered PRO assessments daily via an online portal. Symptoms
conversational agent that engages patients via text messaging, and leverages were monitored and managed by a centralized clinical team. Tiered notifications
natural language processing and machine learning to learn from clinical in- informed the team of concerning or escalating symptoms. We assessed how
teractions. Core functionalities include: (1) real-time dosing instructions, (2) frequently patients completed symptom assessments and the frequency of
motivational reminders, and (3) symptom monitoring with self-management symptom notifications. Results: During the pilot, 28 patients were identified as
support. We conducted a four-month usability study between December 24, high risk and enrolled in the program (median age 65; 64% percent female).
2017 and May 1, 2018 in a large academic cancer center. At monthly intervals Disease types were: 15 (54%) thoracic, 7 (25%) gynecologic, 6 (21%) gas-
for the first 12 weeks of follow-up, research staff conducted qualitative in- trointestinal. Median time in the program was 50 (6-98) days. Patients com-
terviews with participants to evaluate usability and acceptance. Results: 11 pleted 840 of 1,350 assessments (62%). There were 328 assessments that
patients with gastrointestinal neuroendocrine cancer on capecitabine and triggered moderate alerts (39%) and 220 that triggered severe alerts (26%). The
temozolomide were approached regarding the study. Of these, 10 agreed to table describes the prevalence of symptoms at the patient-level. Conclusions: A
participate (ages 45 to 71). Overall, participant satisfaction was high with a model can be employed to identify high-risk patients in collaboration with cli-
Net Promoter Score of 100. Reliability of Penny’s algorithmic branching to nicians. Our adherence rate with a daily symptom assessment was similar to those
provide accurate dosing information and symptom triage was also high: found in studies of less frequent PRO capture. Future work will expand to a larger
symptoms were accurately graded 100% of the time, and there was appro- patient population with other cancer types, evaluate impact on outcomes, and
priate self-management advice or provider triage 100% of the time. Average assess optimal frequency for PRO collection and alert thresholds.
daily adherence (based on self-report) was 98%. Participants reported that 3 Prevalence of symptoms reported at moderate or severe levels one or more days % (n=28).
emergency room visits were avoided during the study period. Conclusions: In Symptom Moderate Severe
preliminary testing, a mobile phone-based conversational agent was a usable Anxiety 86% 11%
and acceptable means of supporting OACA adherence. Expanded study testing Pain 75% 54%
patient safety and efficacy is underway. Fatigue 71% 54%
Nausea 50% 14%
Constipation 46% 0%
Diarrhea 46% 0%
Dyspnea 43% 7%
Decreased oral intake 39% 11%
Emesis 21% 4%

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
376s Health Services Research, Clinical Informatics, and Quality of Care

6536 Poster Session (Board #227), Sat, 1:15 PM-4:15 PM 6537 Poster Session (Board #228), Sat, 1:15 PM-4:15 PM
Evaluation of quality of life, satisfaction and cost of care in metastatic colorectal Proactive, multidisciplinary approach to supportive medicine: Improving
cancer patients receiving ambulatory chemotherapy. First Author: Phichai health outcomes and care utilization. First Author: Brooke Worster, Thomas
Chansriwong, Ramathibodi Hospital, Mahidol University, Bangkok, Thailand Jefferson University, Philadelphia, PA
Background: Colorectal cancer is an important health problems in Thailand; Background: Evidence suggests that cancer patients who receive palliative
chemotherapy remains the most suitable treatment for metastatic patients. care early in their disease have improved quality of life, decreased emer-
Chemotherapy treatment that was once delivered only in hospital environ- gency department (ED) visits, and less aggressive end-of-life care. In 2017, the
ments is now administered at patient’s home that helping patients to live Sidney Kimmel Cancer Center at Jefferson established the Neu Center for
normal lives during receiving chemotherapy. The chemotherapy regimens Supportive Medicine and Cancer Survivorship (NCSMCS) as a model for in-
are based on a 48 hours 5-fluorouracil infusion combined that need pa- tegrated care in the outpatient setting for all cancer patients. A multidisci-
tients to be hospitalized, consequence to decrease QOL and increase cost of plinary team consisting of palliative care physicians, social work, psychology,
treatment. Aims: Compares the QOL score, patients’ satisfaction and cost and navigation conducts biopsychosocial screening and initiates a personal-
difference in treating ambulatory chemotherapy (AC) patients compared with ized care plan for each patient to clarify treatment goals and offer assistance.
inpatient treatment. Methods: An observational cohort which enrolled 156 Objectives: To use biopsychosocial screening at specified time points to
patients at the Ramathibodi hospital from Dec 2015 to Nov 2016. AC identify needs and evaluate the impact of supportive care as part of stan-
administered by the central venous access device (CVAD). The regimen as dardized oncology care regardless of stage. Methods: This assessment utilized
FOLFOX or FOLFIRI, 5-FU were in the elastomeric infusion pump and Oncology Care Model (OCM) data for Jefferson Medicare patients between 7/1/
administered at the patients’ home. Nurse coordinators followed up with the 16 to 7/31/18. Incidence of ED admits ED/Observation and admissions were
patients by phone. The FACT-G and FACT-C scale, patients’ satisfaction and evaluated as well as ICU utilization and advanced care planning. Poisson
cost of treatment questionnaire were collected at time of enrolment, regression was used to generate incidence rate ratios (IRR) and 95% confi-
2 months and end of treatment. Results: 156 patients are enrolled that 111 dence intervals (CI) to facilitate the comparison of post- vs. pre- incidence rates
patients treated with AC and 45 patients treated with inpatient. 134 of hospitalization. Results: The post-intervention hospital admissions de-
returned the questionnaire (response rate 86%). Intention to treat analysis creased by 31% in NCSMCS (IRR 0.69; 95% CI 0.48-0.98) and by 10% in
revealed significantly improved in social wellbeing and FACT-G (p ,0.001) Non-NCSMCS (IRR 0.90; 0.84-0.96) and advanced care plans were more
in AC group. Significant higher administration, service and overall satis- likely to be on file for NCSMCS (9.0% vs. 4.9%). The intensive care unit (ICU)
faction score in AC group. The AC reduced cost about 483 US dollars per admissions were decreased by 17% among Non-NCSMCS (IRR 0.83; 95%
cycle of chemotherapy. Conclusions: Ambulatory chemotherapy helps co- CI 0.74-0.93). The utilization rates for ED admissions were not statistically
lorectal cancer patients to live normal lives by administer treatment at different among both the groups. Conclusions: The preliminary data is
patients’ home and results to significantly improve in quality of life especially promising and impact will be monitored as the intervention is expanded.
in social wellbeing and more satisfaction. Moreover, ambulatory chemo- Reducing admissions has benefits from both a cost savings as well as quality of
therapy reduced cost of chemotherapy treatment. life perspective. Future analyses will consider the impact of the intervention
on a patient’s quality of life.

6538 Poster Session (Board #229), Sat, 1:15 PM-4:15 PM 6539 Poster Session (Board #230), Sat, 1:15 PM-4:15 PM
Deployment and integration of a cognitive technology in China: Experiences Capacity to provide specialized care for older adults in community oncology
and lessons learned. First Author: Tianle Li, Qingdao Baheal Intelligent practices: Results of the NCI Community Oncology Research Program (NCORP)
Technology Co., LTD, Qingdao, China Landscape survey. First Author: Grant Richard Williams, University of Alabama
at Birmingham, Birmingham, AL
Background: Cognitive technologies are rapidly being introduced in oncology
for decision-support, prescribing therapy, predicting risk, reducing medical Background: American Society of Clinical Oncology guidelines recommend
errors and for care management. Few studies have reported on successful that patients $65 years of age starting chemotherapy undergo a geriatric
approaches for clinical adoption. We report the early adopter experience of assessment (GA) to inform and guide management; however, little is known
BahealIntelligenceTechnologyCo., Ltd. (Baheal), across China within a 2-year about resources available in community oncology practices to facilitate ge-
period (April 2017-January 2019). We also describe lessons and experiences riatric specialty care and implement these guidelines. Methods: Community
of oncology users. Methods: Baheal developed collaborative agreements for oncology practices were electronically surveyed in 2017 regarding the avail-
use of IBM Watson for Oncology (WFO) in 96 hospitals across 8 provinces. Key ability of various providers, supportive services, and practice characteristics, as
opinion leaders who saw the potential for AI were recruited as champion part of a larger survey of cancer care delivery research (CCDR) capacity at
advocates. A 29-item survey conducted included usability and integration NCORP sites. Designated CCDR leads provided information about their site.
within clinical workflow. 85 questionnaires were distributed to oncologists who Descriptive statistics were used to report prevalence of resources available at
were major WfO users; 51 were completed. All questionnaires were completed each community practice. Results: Of the 925 NCORP practice locations, 504
anonymously and de-identified prior to analysis. Results: As of January 31, (54%) responded to the survey, representing 227 practice groups. Of re-
2019, 866 physicians have entered a total of 52,537 cancer cases into WfO. spondents, 58% included a free-standing clinic or private/group practice and
Most users approved of both the quality (44/51, 86.3%) and comprehensibility 82% included inpatient services. The median number of new cancer cases per
(45/51, 88.2%) of treatment options, rationales, and literature references. year $65 years of age was 443 (Interquartile range [IQR] 220-903). The
WfO was most frequently applied in the context of inpatient cases reviewed median number of medical oncology providers was 5 (IQR 3-11). Only 1.8% of
by a multidisciplinary tumor board (MDT) (44/51, 86.3%). A lack of locally practices had a dual fellowship trained geriatric oncologist on staff. Geria-
available treatments in WfO was cited as an area for improvement by two-thirds tricians were available for consultation or co-management for 34% of sites,
of users (34/51, 66.7%). CDS was rated from 0 (lowest) to 10 (highest) for but only 13% of those had availability within the oncology clinic. Among those
each of the following uses: EBM medical education (8.1); assistance with with access to geriatricians, consultations were primarily outpatient (90%)
literature (7.7); medical care quality control (7.3); second opinion consul- versus inpatient (54%). Ancillary services that could support GA were variably
tations (7.0); case review with tumor board (6.9); and decision support (6.4). available onsite: social work (83%), nurse navigators (78%), pharmacist
Overall, users were willing to recommend CDS to patients and other clinicians (77%), dietician (69%), supportive caregiver services (62%), rehabilitative
(7.3). Conclusions: WfO CDS, employed in a variety of settings, was viewed medicine (57%), psychologist (41%), and psychiatrist (39%). Most sites
positively by more than 86% of users, with perceived benefits differing by utilized electronic health record systems (84%) and patient portals (89%).
context. Future incorporation of locally available treatments and under- Conclusions: Availability of geriatric-trained providers is limited in community
standing reasons for their omission in CDS may increase perceived value, oncology practices. Use of primarily self-administered GA tools that direct
improve standardization, quality of cancer care and equity of care in China. referrals to available ancillary services may be an effective implementation
strategy for guideline-based care.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Health Services Research, Clinical Informatics, and Quality of Care 377s

6540 Poster Session (Board #231), Sat, 1:15 PM-4:15 PM 6541 Poster Session (Board #232), Sat, 1:15 PM-4:15 PM
Using real-world cohorts to assess the generalizability and relevance of Delay in receipt of newly prescribed oral anticancer drugs. First Author:
randomized clinical trials (RCTs). First Author: Caroline Savage Bennette, Daniel O’Neil, Columbia University Medical Center, New York, NY
Flatiron Health, New York, NY
Background: Oral anticancer drug (OACD) prescriptions require coordination
Background: RCTs are the gold standard for understanding the efficacy of new between clinicians, payers, specialty pharmacies, and financial assistance
treatments, however, patients (pts) in RCTs often differ from those treated in the (FA) groups, which may delay patient receipt of the drug. Factors associated
real-world. Further, selecting a standard of care (SOC) arm is challenging as with delay in receipt of OACDs are unknown. Methods: We prospectively
treatment options may evolve during the course of a RCT. Our objective was to collected data on all new OACD prescriptions (RXs) from the medical oncology
assess the generalizability and relevance of RCTs supporting recent FDA ap- practice at the Herbert Irving Comprehensive Cancer Center from 1/1/2018 to
provals of anticancer therapies. Methods: RCTs were identified that supported 12/1/2018. We collected patient demographic, insurance and clinical in-
FDA approvals of anticancer therapies (1/1/2016 - 4/30/2018). Relevant pts formation; date of prescription; date of drug delivery; and staff interactions with
were selected from the Flatiron Health longitudinal, EHR-derived database, payers and FA groups. Federal Drug Association (FDA) labels and Micromedex
where available. Two metrics were calculated: 1) a trial’s pt generalizability were reviewed for initial drug approval dates, approved indications and average
score (% of real-world pts receiving treatment consistent with the control arm wholesale price. We used multivariable linear and logistic regression to de-
therapy for the relevant indication who actually met the trial’s eligibility criteria) termine factors associated with number of days from prescription to receipt of
and 2) a trial’s SOC relevance score (% of real-world pts with the relevant OACD. Results: During the study period 510 OACD RXs were evaluated. Of
indication and meeting the trial’s eligibility criteria who actually received these, 84 (16%) were never filled. The most common OACDs were capeci-
treatment consistent with the control arm therapy). All analyses excluded real- tabine (90, 18%), abiraterone (45, 9%), palbociclib (35, 7%) and osimertinib
world pts treated after the relevant trial’s enrollment ended. Results: 14 RCTs (28, 6%). Of 426 filled RXs, the median time from prescription to receipt was
across 5 cancer types (metastatic breast, advanced non-small cell lung cancer, 8 days (IQR 5-13), with 193 RXs (46%) received in #7 days, 145 (34%) in 8-
metastatic renal cell carcinoma, multiple myeloma, and advanced urothelial) 14 days and 65 (15%) in 14-28 days, and 23 (5%) at . 28 days. Linear
were included. There was wide variation in the SOC relevance and pt gener- regression showed time to receipt of OACD (log transformed) was associated
alizability scores. The median pt generalizability score was 63% (range 35% - with having commercial primary insurance (p = 0.02), pursing FA (p = ,
88%), indicating that most real-world pts would have met the RCT eligibility 0.001), RX of a drug approved by the FDA , 2 years earlier (p = 0.008), drugs
criteria. The median SOC relevance score was 37% (range 15% - 74%), in- without an approved indication for the primary tumor (p = 0.03) and estimated
dicating that most RCT control arms did not reflect the way trial-eligible real- drug cost (p = 0.002). The other included covariates, patient age and prior
world pts in the US were actually treated. Conclusions: There is great variability authorization, were not associated with time to receipt. Logistic regression
across recent RCTs in terms of pt generalizability and relevance of SOC arms. comparing receipt at #14 versus . 14 days found association with FA (OR
Real-world data can be used to inform selection of control arms, predict impact 3.17; 95%CI 1.78-5.65), FDA approval within 2 years (OR 3.52; 95%CI
of inclusion/exclusion criteria, and also assess the generalizability of the results 1.31-9.45) and off-label use (OR 2.30; 95%CI 1.18-4.50). Conclusions: Over
of completed trials. Incorporating real-world data in planning and interpretation 20% of new OACDs were received 14 days or longer from the date of RX.
of prospective clinical trials could improve accrual and enhance relevance of Financial and insurance related factors; and more expensive and recently
RCT outcomes. approved drugs were associated with longer delays in receipt of therapy. Policy
changes to improve the timeliness of OACD access are needed.

6542 Poster Session (Board #233), Sat, 1:15 PM-4:15 PM 6543 Poster Session (Board #234), Sat, 1:15 PM-4:15 PM
Safety and outcomes of a cancer patient urgent care clinic. First Author: Validation of an oncology-specific instrument to measure cancer patients’
Jack S Bevins, University of Texas Southwestern Department of Internal perception of care coordination. First Author: Izumi Okado, University of
Medicine, Dallas, TX Hawaii Cancer Center, Honolulu, HI
Background: Several cancer centers describe cancer-patient dedicated ur- Background: According to the IOM, effective coordination of care (CC) is a
gent care clinic (UCC) that address commonly anticipated complaints of critical component of high-quality cancer care; however, lack of a reliable and
adults with cancer. UCC may be capable of preventing some ED visits, but validated measure limits our current understanding of cancer care coordi-
little is known of the safety and outcomes for patients after a UCC visit. nation. We examined psychometric properties and utility of a Care Coordination
Methods: We identified UCC visits made by adults at our comprehensive Instrument (CCI), a survey developed to assess cancer patients’ perceptions of
cancer center between 2013-2016 and compared the cohort to patients who care coordination. Methods: The 29-item CCI was administered to 200 pa-
did not visit the UCC. We linked patients to tumor registry data and their tients receiving active treatment for cancer at private oncology practices
electronic health record from the UCC visit, then tracked ED visits, inpatient and hospital-based facilities from Oct. 2018 to Jan. 2019. The CCI includes
and intensive care unit (ICU) admissions occurring within 24 hours of the subscales that evaluate CC in 3 domains (Communication, Navigation, Op-
UCC visit. Results: Between 2013-2016, 551 patients generated 772 UCC erational) across 4 areas of CC (patient-physician; between health providers;
visits, compared to 17,496 who did not visit. UCC users had signifi- during inpatient-to-ambulatory care transitions; during transitions across
cantly (p,0.001) more advanced-stage cancer than non-UCC users (37.3% different phases of care). All items were rated on a 4-point Likert scale.
vs 18.9%), but there were no significant differences in mean age, race/ Results: Psychometric analyses of the CCI demonstrated that it has good
ethnicity, or death within 180 days of diagnosis. The most common chief internal consistency reliability (a = .917) and the three-factor solution was an
complaints accounted for nearly half of all UCC visits: (17.4%), URI acceptable fit (CFI = .853, SRMR = .065). Overall, cancer types (leukemia,
symptoms/fever (12.6%), nausea/vomiting/diarrhea (7.8%), and fatigue/ myeloma) and having an identified patient navigator significantly predicted
weakness (7.6%). After 10.0% of UCC visits, patients had an ED visit, higher patients’ ratings of CC (p , .05). Similar trends were found for
while 12.3% were admitted to the hospital; only 5 UCC visits (0.7%) had an Communication and Operation subscale scores (p , .05). Having an identified
associated ICU stay. Most patients (75.7%) only had a single UCC visit, but navigator predicted higher Navigation scores (p , .05). Marginally significant
patients who visited the UCC more often tended to have higher rates of ED differences were found for practice setting, with patients receiving care in
visits and hospitalizations within 24 hours (Table). The mean time from UCC hospital-based facilities reporting better CC (p = .085). Item-level analyses
arrival to ED arrival was 3.0 hours, and 6.5 hours from UCC arrival to in- revealed significant differences in specific aspects of CC (e.g., physician-
patient arrival. Conclusions: The majority of patients seen in UCC did not patient communication) across cancer type, presence/absence of a patient
require ED or inpatient hospitalization. Patients with subsequent ED or navigator, and practice setting. Conclusions: The results demonstrate that the
inpatient visits had minimal delays in care. Findings suggest that triaging CCI is a reliable and valid instrument for measuring cancer patients’ per-
cancer patients for commonly anticipated complaints to a UCC does not ceptions of care coordination. Perception of CC correlated with the presence
result in high rates of mis-triaging or major delays in care. Patients with ED, of a navigator, underlying cancer type and (trending) practice setting. Use of
Inpatient, or ICU visit after UCC, stratified by UCC visits per patient (2013- this instrument may reveal important information about cancer care coordi-
2016). nation and may identify areas of targets for improvement in patient-centered
cancer care delivery.
UCC Visits/ # of Patients ED within 24 hrs. Inpatient within 24 hrs. ICU within 24 hrs.
Patient (%) (%) (%) (%)
1 417 (75.7) 43 (10.3) 50 (12.0) 4 (1.0)
2 81 (14.7) 16 (19.8) 19 (23.4) 0
3 30 (5.4) 10 (33.3) 11 (36.7) 1 (3.3)
4+ 23 (4.2) 7 (30.4) 5 (21.7) 0

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
378s Health Services Research, Clinical Informatics, and Quality of Care

6544 Poster Session (Board #235), Sat, 1:15 PM-4:15 PM 6545 Poster Session (Board #236), Sat, 1:15 PM-4:15 PM
Nurse Navigators telephone and web application follow-up intervention for Watson for Oncology applied to teaching and remote consulting model. First
patients treated with oral anticancer medication: an optimization of oncol- Author: Juemin Fang, Shanghai Tenth People‘s Hospital, Tongji University,
ogists’ medical time. First Author: Marie Ferrua, Gustave Roussy, Villejuif, Shanghai, China
France
Background: IBM Watson for Oncology (WFO) is developed with artificial
Background: Many interventions to improve safety and quality of oral anti- intelligence to assist the oncologists in treatment decision-making by pro-
cancer medication have occurred the past few years. Frequently, these systems viding evidence-based treatment recommendations with priority and per-
involve Nurse Navigators (NN) but rather limited data is actually available sonality. We have been exploring the value of applying WFO in teaching and
describing their activities and quantifying which intervention could have been remote consulting among members in the hospital union to improve medical
performed without an oncologist’s expertise. This is a major issue in optimizing quality and conformity. Methods: Four modes were followed to apply WFO
the workload of oncologists statewide. Methods: The study was conducted at to teaching and remote consulting. Firstly, a teaching hospital conducted
Gustave Roussy (Villejuif, France) as part of an assessment program of an instructional teaching for a primary hospital through WFO, or a primary hospital
intervention combining NN and mobile app dedicated to patients receiving oral consulted with a teaching hospital on cancer treatment combined with WFO.
cancer medication (cytotoxic, targeted therapies). A coding grid was developed Secondly, a teaching hospital conducted conference-based teaching and
to classify the NNs’ activities from the intervention reports. NN interventions remote consulting for two or three primary hospitals through demonstrating
consist in regular follow-up calls and requests from patients and their relatives WFO. Thirdly, we cooperated with other teaching hospitals to carry out in-
(via phone calls or mobile app). NN interventions recorded over a 24-month teractive discussions on cases through WFO with several primary hospitals.
period were all encoded by two researchers. Results: 2395 interventions were Fourthly, we cooperated with other teaching hospitals to conduct WFO
analyzed concerning 236 patients, 1880 of which were regular follow-ups, and demonstration and discussion for several primary hospitals. Remote com-
515 patient requests. The majority of contacts were carried by phone con- munication was performed by using third-party video conference software. A
versation (94%). 52% of the interventions were followed-up upon by at least survey was conducted to collect feedback from 56 primary hospitals in our
one additional action by the NN (n = 1250). In 25% (n = 318) of the in- hospital union. Results: More than 80% of primary hospitals were willing to
terventions, the oncologist’s expertise was required mainly because of the learn the standardized treatment and recent treatment progress of tumors by
presence or aggravation of symptoms and/or toxicities or concerning onco- participating in WFO remote consulting in our hospital. The value of four modes
logical treatment (therapeutic protocol, biological assessment). 75% (n = 932) to promote the standardized cancer treatment and improve the medical quality
of the interventions have been processed successfully by NN by themselves. and conformity in primary hospitals were all recognized. WFO also contributed
The principal actions carried out by the NN were: A. Contact or provide referral to academic exchanges and learning among high-level teaching hospitals, and
to a health professional or institution 24%, B. Advice to the patient 43%, C. facilitated the use of artificial intelligence in oncology therapy more rational
Management of medical documents and appointments 6%; D. Administrative and appropriate. Conclusions: Non-standardization in cancer care remains to
management 2%. Conclusions: Patients’ needs for oral cancer drugs are be a big problem in primary hospitals in China. Although limitation of WFO
mainly related to advice, information and coordination of the care pathway. In exists in the complex cases of certain tumor types, the rational use of WFO in
the majority of interventions, the NNs were able to manage these situations by the teaching and remote consulting could help and promote the standardized
themselves, which could optimize the workload of oncologists. cancer treatment in China.

6546 Poster Session (Board #237), Sat, 1:15 PM-4:15 PM 6547 Poster Session (Board #238), Sat, 1:15 PM-4:15 PM
Creation of a virtual cancer care network for remote oncology treatment. First National lung cancer screening utilization trends in the Veterans Health
Author: Vida Almario Passero, VA Pittsburgh Healthcare System, Pittsburgh, Administration. First Author: Jennifer A. Lewis, Veterans Health Adminis-
PA tration, Tennessee Valley Healthcare System Geriatric Research Education
Clinical Center, Nashville, TN
Background: The VA Pittsburgh Healthcare System (VAPHS) Virtual Cancer
Care Network was launched in January 2018 after we had established an Background: Low-dose CT (LDCT) is an effective means for early lung cancer
electronic consult service where 555 hematology electronic consults were detection, but is often underutilized. An estimated 900,000 Veterans are
completed at VAPHS in FY17. The clinical video telehealth (CVT) clinic allows eligible for lung cancer screening. We are the first to describe national lung
veterans from central Pennsylvania to receive their anticancer therapy at the VA cancer screening utilization trends in the Veterans Health Administration
in Altoona, Pennsylvania where the oncology pharmacy, nursing, telehealth, (VHA). Methods: We assembled a retrospective cohort of patients within the
and supportive oncology staff are on site. Patients follow regularly and remotely VHA’s Observational Medical Outcomes Partnership (OMOP) Common Data
during treatment via CVT visits with their oncologist located 93 miles away at Model who underwent lung cancer screening. LDCT scans with Common
the VA in Pittsburgh. Methods: A chemotherapy pharmacy and nursing in- Procedure Terminology (CPT) codes G0297 or 71250 from January 1, 2011 to
fusion clinic were created at the VA in Altoona. CVT visits started in January May 31, 2018 were eligible for inclusion. We further selected exams described
2018. Data including treatment, adverse events defined through CTCAE v5.0, as “lung cancer screening,” “screening,” or “LCS.” We used descriptive
gender, age, zip code, and other details were examined retrospectively. statistics with frequencies and medians to calculate the total exams per
Results: 279 CVT visits for 89 patients were completed January 2018 through Veteran and evaluate utilization trends over time and by region. Results: At
Sept 2018. 87 were male, 2 were female. Average age was 70 (range 45-90). initial screening, Veterans had a median age of 66 (IQR 61, 70), 95% were
Most common primary disease sites were prostate (19.1%), colorectal male, 76% Caucasian. From January 1, 2011 to May 31, 2018, 75 VHA
(13.4%), and lung (9%). 61.8% of patients were on treatment. Non-treatment facilities performed 129,363 LDCT exams for lung cancer screening; 87,950
visits were for surveillance and survivorship. Treatment administered included (68%) initial and 41,413 (32%) subsequent exams. Screening has increased
platinum doublets, fluorouracil doublets, immunotherapy, and oral anticancer over time (226 in 2011-2012; 7848 in 2013-2014; 41,225 in 2015-2016;
therapy. 5.4% of patients had Grade 3-4 events due to febrile neutropenia, 80,064 in 2017 until May 31, 2018) in all regions. Providers in primary
increased liver enzymes, and hemolytic anemia. 41.7% had grade 1-2 events care/internal medicine (56%), family medicine (16%), pulmonology (6%),
due to peripheral neuropathy, neutropenia, anemia, thrombocytopenia, and oncology (0.3%), other specialties (21%) ordered screening exams.
infusion-related reactions. Using an average commuting speed of 60 mph Conclusions: Lung cancer screening with low-dose CT within the VHA in-
and a travel cost of $ 0.56 per mile, the total commute distance averted creased over time within all geographic regions. Future strategies aimed at the
was 49,579 miles. Mean distance averted per patient was 557 miles. Total Veteran, provider, and healthcare system levels are needed to increase lung
commute time saved for veterans was 826 hours. Total mileage costs saved for cancer screening utilization among eligible Veterans.
veterans was $27,764. Conclusions: The Virtual Cancer Care Network reduced
the travel time and costs for veterans who previously would have travelled from Northeast South Mid-West West
Year n (%) n (%) n (%) n (%) Total
central Pennsylvania to VAPHS for their oncology treatment. Adverse events
were tolerable and managed by the VA in Altoona. Integration of CVT secures 2011-12 1(0) 225(0.4) 0 0 226(0.2)
2013-14 2,325(7.5) 4,007(6.7) 1,200(4.5) 316(2.7) 7848(6.1)
safe access to cancer care and maintains patients’ primary relationships with 2015-16 10,095(32.6) 19,314(32.1) 8,283(31.2) 3,533(30.2) 41,225(31.9)
their oncologists. 2017-18* 18,500(59.8) 36,665(60.9) 17,031(64.2) 7,868(67.2) 80,064(61.9)
Total 30,921(100) 60,211(100) 26,514(100) 11,717(100) 129,363(100)
*Data collected through May 31, 2018

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Health Services Research, Clinical Informatics, and Quality of Care 379s

6548 Poster Session (Board #239), Sat, 1:15 PM-4:15 PM 6549 Poster Session (Board #240), Sat, 1:15 PM-4:15 PM
Active search for toxicity reduces delay on chemotherapy. First Author: Prospective comparative effectiveness trial of multidisciplinary lung cancer
Juliana Souza, ONCOCLINICA, Rio De Janeiro, Brazil (LC) care. First Author: Raymond U. Osarogiagbon, Baptist Cancer Center,
Memphis, TN
Background: Interdisciplinary teamwork involves health care professionals
working as a team with the purpose of discussing individual cases and Background: LC is common and lethal; care-delivery is complex, varies in
recommending care plans. In october 2015, the breast cancer interdisci- quality and outcomes, stimulating calls for multidisciplinary treatment planning
plinary team (BCIT) was implemented in our outpatient oncology clinic. This (MTP) involving key specialists. This much-advocated model lacks rigorous
team includes oncology physician, pharmaceutist, psychologist, nutritionist evaluation. We conducted a prospective cohort study of MTP v Serial Care (SC)
and oncology nurse. Since 2018, the oncology nurse does active search for in a community healthcare system. Methods: Newly-diagnosed LC patients with
toxicity (AST) in order to solve early symptoms. Methods: We evaluated performance status (PS) 0-2, and their caregivers, were enrolled from a LC
outcomes of breast cancer patients who finished neoadjuvant/adjuvant multispecialty group clinic (MGC) or single-specialty general oncology clinics. A
chemotherapy and received BCIT care at our oncology clinic from october subset of general oncology clinic patients were discussed in a Multidisciplinary
2015 to November 2018. Results: Of 200 pts who had finish treatment, 139 Thoracic Oncology Conference (MTOC), others were not (Serial Care [SC]). In
pts (69.5%) received adjuvant chemotherapy and 61 pts (30.5%) neo- this analysis, we compare MGC and MTOC patients (MTP recipients) to SC
adjuvant schedule. Median age was 52,64 years (range:30.6-82.3y). At patients. Primary endpoint was overall survival (OS); secondary endpoints were
diagnosis, 50 pts (25%) were stage I, 101 pts (50.5%) were stage II, and 49 measures of quality: staging practices, guideline-concordant treatment, time-
pts (24.5%) were stage III. When we compared before and after AST, there liness of care, patient and caregiver satisfaction. We adjusted proportional
were no significant difference between mean age (54.13y vs 54.06y, p: hazards and logistic models for age, sex, histology, stage, PS, insurance, and
0.97), colony stimulating factor use (p:0.10), cold-cap use (p:0.81), and race. Results: 254 patients received MTP v 272 SC. After a median 30 months’
timing of chemotherapy (p:0.13). There were 7 hospitalizations during follow up, there was no difference in OS (adjusted hazard ratio 1.10 [CI 0.87-
chemotherapy, with no significant difference with AST (5pt vs 2pt, p: 0.82). 1.40], p = .43). Stage-confirmatory biopsy was done in 61% MTP v 45% SC
However, patients in AST had significantly lower mean delay to com- patients (adjusted odds ratio [aOR] 2.59, CI 1.74-3.86, p , .0001); 81% MTP
plete chemotherapy (9.4 days vs 4.2 days p:0.003). After median follow up v 68% SC patients received guideline-concordant treatment (aOR 2.04, CI
17 months, there were 9 progression disease and 4 deaths. 1-year and 3-year 1.31-3.19, p , .002). Although the time from lesion detection to diagnostic
overall survival rate was 99,2% and 95,1%, respectively. Conclusions: AST biopsy (25 v 15 days, p = .004) or staging biopsy (29 v 20 days, p = .007) was
increased adherence to chemotherapy, with lower delay on chemotherapy higher with MTP, there was no difference in time to definitive treatment (60 v
treatment. There were no significant difference with colony stimulating 57 days, p = .06). MTP patients and their caregivers reported greater satis-
factor use, cold-cap use, timing of chemotherapy, and hospitalization. faction with the combined quality of care received from all team members (p ,
.0001) at baseline, 3 and 6 months. Conclusions: MTP for LC significantly
improved the quality of care including the thoroughness of staging, use of
guideline-concordant care, and patient satisfaction. Contrary to reports from
retrospective analyses, timeliness of care was worse with MTP. Patient and
caregiver satisfaction was superior with MTP. Despite improved quality, MTP
was not associated with improved LC survival. Clinical trial information:
NCT02123797.

6550 Poster Session (Board #241), Sat, 1:15 PM-4:15 PM 6551 Poster Session (Board #242), Sat, 1:15 PM-4:15 PM
Minimizing drug wastage (DW) and cost of cabazitaxel used to treat meta- Trends in the use of proton beam therapy among newly diagnosed cancer
static castrate-resistant prostate cancer (mCRPC). First Author: Di Maria patients in the United States. First Author: Leticia Maciel Nogueira,
Jiang, Princess Margaret Cancer Centre, University Health Network, Toronto, American Cancer Society, Atlanta, GA
ON, Canada
Background: Proton Beam Therapy (PBT) is a potentially superior radio-
Background: Cabazitaxel is indicated for mCRPC, but is associated with substantial therapy technology for tumors with complex anatomy surrounded by sensitive
DW and financial strain on hospital budgets. It is only available in single-dose 60mg tissues and for childhood cancers where sparing surrounding normal tissue is
vials and has short reconstituted drug stability of , 24 hours. We aimed to determine better achieved than with photon radiotherapy. The first conditions for
feasibility and cost savings of an aggressive batching strategy to facilitate vial sharing payment of PBT claims went into effect in 2009. In 2014, the American
of Cabazitaxel. Methods: Our mitigation strategy was to administer Cabazitaxel Society of Radiation Oncology categorized PBT clinical indications into
20mg/m2 q3-weekly (without prophylactic G-CSF) on a single weekday whenever Group 1, for which health insurance coverage is recommended, and Group 2,
possible. Drug was prepared after patient (pt) arrival. Remaining amount from each
for which coverage is recommended only if additional requirements are met.
vial was saved for subsequent pts on the same day. Amount administered, discarded
Methods: We evaluated 21,920 newly diagnosed patients in the National
and number of (#) vials used were obtained from pharmacy records. We estimated
drug cost without batching by assigning 1 vial/treatment, and drug cost with batching
Cancer Database (NCDB) who received PBT between 2004 and 2016.
from the actual # vials used. Cost of DW was determined from the amount discarded. Joinpoint analyses were used to evaluate the Annual Percent Change (APC)
All cost calculations were based on market price ($96.7CAD/mg) accounting for in the number and characteristics of patients treated with PBT. Results: The
Sanofi’s discount incentive (5 vials for the price of 4), allowing a real-world cost number of patients treated with PBT in NCDB facilities increased from
assessment. Results: Between 09/2015 and 09/2018, 74 pts received 404 Cab- 1,114 in 2004 to 3,173 in 2016 (APC = 8.78, p , .001), due mainly to
azitaxel treatments on 164 days using 319 vials. Multiple pts were batched on 68% increases in Group 1 cancers after 2010 (from 271 patients in 2010 to
treatment days. Every 3 pts batched saved 1 vial. Average dose/treatment was 37mg 1,124 in 2016, APC = 26.4, p , .001). The number of Group 2 patients
(20-45mg). Among 10 treatment cancellations, prepared drug was administered for treated with PBT increased slower (from 937 in 2004 to 2,049 in 2016, APC
subsequent pts in 9 cases. Drug and DW costs over the 3-year period with and = 6.1, p , .05). Breast and prostate cancers were most common, although
without batching are shown in Table. Conclusions: Batching $3 pts on a single trends varied substantially by cancer site. Between 2010 and 2016, receipt
weekday was feasible and significantly lowered drug cost of Cabazitaxel by reducing of PBT increased for breast cancer patients from 40 in 2010 to 405 in 2016
wastage. This strategy could help mitigate costs associated with wastage for other (APC = 48.5, p , .001), but decreased for prostate cancer patients from
oncology drugs. 1,205 in 2011 to 680 in 2016 (APC = -14.06, p , .001). While most of
No Batching Batching Group 1 patients had private insurance coverage (59.3% of patients treated
# Pts # Drug DW DW Cost / Drug Drug DW DW Cost / Drug Drug Cost in 2016), Medicare was the most common primary insurance type among
batched days Cost ($) Cost ($) Cost (%) Cost ($) Cost ($) Cost (%) Saved $ (%) Group 2 patients (50% of patients treated in 2016). Conclusions: The
1 53 235,561 119,229 51 235,561 119,229 51 0 number of newly diagnosed cancer patients treated with PBT has increased
2 36 327,233 158,685 48 327,233 158,685 48 0
3 39 536,105 257,609 48 365,526 31,331 9 170,579 (32) between 2004 to 2016 in the US, with a sharp increase for cancers with
4 24 446,758 211,193 47 330,718 71,945 22 116,040 (26) clinical indications for health insurance coverage since 2010. While most of
5 6 139,248 70,881 51 92,832 12,861 14 46,416 (33)
6 5 139,248 67,883 49 92,832 9,863 11 46,416 (33) these patients have private insurance coverage, the steady increase in the
7 1 34,820 15,665 45 23,208 4,061 16 11,612 (33) number of patients being treated with PBT for cancers with additional re-
Total 1,879,848 721,285 - 1,467,910 326,846 - 411,938 (22)
quirements for health insurance coverage is primarily in those with Medicare
coverage.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
380s Health Services Research, Clinical Informatics, and Quality of Care

6552 Poster Session (Board #243), Sat, 1:15 PM-4:15 PM 6553 Poster Session (Board #244), Sat, 1:15 PM-4:15 PM
The diagnosis and outcomes when the outpatients receiving chemotherapy A blinded evaluation of a clinical decision-support system at a regional cancer
visited the emergency room: A tertiary referral center retrospective study of care center. First Author: Suthida Suwanvecho, Horizon Cancer Center,
734 cases. First Author: Takatsugu Ogata, Department of Clinical Oncology, Bumrungrad International Hospital, Bangkok, Thailand
Kobe City Medical Center General Hospital, Kobe, Japan
Background: Clinical decision-support systems (CDSS) such as Watson for
Background: Today, the chemotherapy is performed in outpatient, but there Oncology (WFO) may reduce treatment variation in oncology, provided op-
is no research on the safety. This study aimed to determine the safety of tions offered by the system are at least as acceptable as expert, evidence-
outpatients receiving chemotherapy and the points to note when they visit based options. Deviation from expert consensus in practice is not well
the emergency room. Methods: We retrospectively collected data from July documented. In this blinded study, WFO therapeutic options and treatment
2011 to October 2018 in tertiary referral center of outpatients receiving decisions made by individual oncologists at Bumrungrad International
chemotherapy when they visited the ER within 3 months from the admin- Hospital (BIH) were evaluated by expert panel. Methods: Treatments se-
istration of chemotherapy. Results: Seven hundred and thirty-four cases lected by BIH that were labeled as either “for consideration” or “not rec-
(345 patients) were enrolled (median age, 71 years; male 410, female 324). ommended” by WFO were evaluated by a panel of 3 oncologists in 2018. The
The median days of the patients visiting the ER was 16 days after the ad- panel evaluated WFO options and previous BIH treatments for prospective
ministration of chemotherapy. The tumor types were gastrointestinal (226 cases from 2016-2018, blinded to the source of treatment option. Con-
cases), urological (199 cases), respiratory (112 cases), and the other (197 sensus of panel rated treatment pairs as: identical; both acceptable and
cases). The cytotoxic agents, antibody, or hormonal agents were 530 cases, roughly equivalent; both acceptable, but one preferred; one is acceptable
150 cases, or 173 cases, respectively. The median body temperature and and the other, unacceptable; neither is acceptable. The results of 321
systolic blood pressure were 36.8 °C and 130 mmHg, respectively. The treatment choices for breast, lung, colon and rectal cancers were analyzed,
median estimated glomerular filtration rate was 68. The cases of emergent and McNemar’s test, a modified pairwise chi-square, was applied to identify
admission were 256 cases. The tumor-associated disease was 184 cases differences between BIH and WFO. Results: 71% of both BIH and WFO
and the chemotherapy-associated disease was 105 cases. The most frequent treatments across all 4 cancer types were considered acceptable or identical
chemotherapy-associated disease was febrile neutropenia (FN) (35 cases). by the panel. In 18 cases (5.6%), WFO treatments were preferred; in 14
The admission of tumor-associated disease or chemotherapy-associated cases (4.4%), BIH cases were preferred. Unacceptable treatments by either
were 94 cases or 41 cases, respectively (p = 0.273). Conclusions: The out- BIH or WFO were identified in 15% and 23% of treatments, respectively.
patients receiving chemotherapy visited the emergency room because of the Statistical analysis of treatment pairs revealed no significant difference
tumor-associated symptoms rather than chemotherapy-associated symptoms. between BIH and WFO treatments for breast, colon and rectal cancer.
It was safe that the chemotherapy is performed in outpatient. FN was the most Treatment for lung cancer differed significantly (p = 0.004); in 6% of cases,
frequent chemotherapy-associated disease. It is important to be careful that WFO was unacceptable and BIH acceptable; in 1% of cases, BIH was
not only chemotherapy-associated symptoms but also tumor-associated unacceptable and WfO was acceptable. Conclusions: This study is one of the
symptoms when the outpatients receiving chemotherapy visited the ER. first to compare therapeutic options from CDSS to treatment decisions made
in practice, evaluated in a blinded fashion by an expert panel. 71% of
treatments suggested by WFO CDSS were as acceptable as those selected by
clinicians at the point of care, and some were considered superior. Deci-
sions made in practice were unacceptable to the panel in 15% of cases,
suggesting a role for CDSS.

6554 Poster Session (Board #245), Sat, 1:15 PM-4:15 PM 6555 Poster Session (Board #246), Sat, 1:15 PM-4:15 PM
A framework for building a clinically relevant risk model. First Author: Reasons for discordance in treatment approaches between oncology practice
Robert Michael Daly, Memorial Sloan Kettering Cancer Center, New York, NY and clinical decision support in China. First Author: Jun Liang, Peking Uni-
versity International Hospital, Beijing, China
Background: Acute care accounts for half of cancer expenditures and is a
measure of poor quality care. Identifying patients at high risk for emergency Background: Therapeutic clinical decision-support systems (CDSS) are often
department (ED) visits enables institutions to target resources to those most evaluated by comparisons between CDSS options and actual practice de-
likely to benefit. Risk stratification models developed to date have not been cisions or expert opinions. Few such studies have carefully examined reasons
meaningfully employed in oncology, and there is a need for clinically relevant for discordance. Methods: We reviewed 11 concordance studies from different
models to improve patient care. Methods: We established and applied a hospitals across 8 provinces in China, published between 2017 and 2018. The
predictive framework for clinical use with attention to modeling technique, studies compared IBM Watson for Oncology (WfO) therapeutic options to
clinician feedback, and application metrics. The model employs electronic treatments selected by oncologists or a tumor board involved in review of cases
health record data from initial visit to first antineoplastic administration for for lung, colon, rectal, breast, gastric, and gynecological cancers. We identified
patients at our institution from January 2014 to June 2017. The binary given reasons for discordance and summarized themes across studies.
dependent variable is occurrence of an ED visit within the first 6 months of Results: Of the 11 studies, 9 provided 1 or more reasons for discordance which
treatment. The final regularized multivariable logistic regression model was could be analyzed. We found three major themes related to discordance:
chosen based on clinical and statistical significance. In order to accom- formulary restrictions, treatment-protocol differences, and physician or patient
modate for the needs to the program, parameter selection and model cal- preferences (Table). Formulary differences between WfO and regional prac-
ibration were optimized to suit the positive predictive value of the top 25% of tices included off-label drug uses or unavailable therapies. Treatment-protocol
observations as ranked by model-determined risk. Results: There are 5,752 differences included variations in regimens, such as simultaneous versus
antineoplastic administration starts in our training set, and 1,457 in our test sequential treatments. Physician or patient preferences included factors such
set. The positive predictive value of this model for the top 25% riskiest new as the cost of treatment and logistics associated with various treatments.
start antineoplastic patients is 0.53. From over 1,400 data features, the Conclusions: This study identified multiple reasons for discordance between
model was refined to include 400 clinically relevant ones spanning de- an oncology CDSS option and oncologists’ treatment choices in China.
mographics, pathology, clinician notes, labs, medications, and psychosocial Treatment differences arose from local formulary or protocol differences as well
information. At the patient level, specific features determining risk are as provider and patient preferences. Future studies of CDSS should include
surfaced in a web application, RiskExplorer, to enable clinician review of reasons for discordance when assessing system performance in this manner.
individual patient risk. This physician facing application provides the in- Reasons for discordance.
dividual risk score for the patient as well as their quartile of risk when
compared to the population of new start antineoplastic patients. For the top % of Studies
Source of Discordance Reporting Reason
quartile of patients, the risk for an ED visit within the first 6 months of
treatment is greater than or equal to 49%. Conclusions: We have Formulary restrictions 77 % Off-label uses or availability of a therapy
constructed a framework to build a clinically relevant risk model. We are now Treatment protocol 33 % Simultaneous versus sequential admin-
piloting it to identify those likely to benefit from a home-based, digital differences istration regimens
Physician or patient 22 % Cost of treatment or logistics associated
symptom management intervention. preference with treatments

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Health Services Research, Clinical Informatics, and Quality of Care 381s

6556 Poster Session (Board #247), Sat, 1:15 PM-4:15 PM 6557 Poster Session (Board #248), Sat, 1:15 PM-4:15 PM
Development of an artificial intelligence model to predict survival at specific Accelerating advanced precision medicine through a harmonized data
time intervals for lung cancer patients. First Author: Smita Agrawal, Concerto exchange platform and research consortium (PMEC). First Author:
HealthAI, Bengaluru, India Davendra Sohal, Cleveland Clinic, Cleveland, OH
Background: Survival prediction models for lung cancer patients could help Background: Clinico-genomic data sharing is consistently identified by the
guide their care and therapy decisions. The objectives of this study were to global oncology community as a critical requirement to accelerate the dis-
predict probability of survival beyond 90, 180 and 360 days from any point in a covery and development of new targeted therapies. However, lack of effective
lung cancer patient’s journey. Methods: We developed a Gradient Boosting collaborative models, fragmented and lengthy legal contracting processes,
model (XGBoost) using data from 55k lung cancer patients in the ASCO paucity of funding, and inadequate technological platforms have historically
CancerLinQ database that used 3958 unique variables including Dx and Rx been obstacles for effective data sharing. Methods: In 2015, 10 US academic
codes, biomarkers, surgeries and lab tests from #1 year prior to the prediction medical centers (AMC) and Foundation Medicine Inc. (FMI) formed PMEC.
point, which was chosen at random for each patient. We used 40% data for Feasibility assessments included creation of a master agreement across
training, 25% for hyper-parameter tuning, 20% for testing and 15% for sites and willingness to use a central IRB. Oversight and research steering
holdout validation. Death date available in the Electronic Health Record was committees were created within the consortium. Through a centralized,
cross checked by linkage to death registries. Results: The model was validated secure web-based platform, FoundationInsight, we combined and shared
on the holdout set of 8,468 patients. The Area Under the Curve (AUC) for the de-identified, harmonized comprehensive FoundationOne genomic pro-
model was 0.79. The precision and recall for predicting survival beyond the filing data. Research proposals mining this data warehouse are invited
three time points were between 0.7-0.8 and 0.8-0.9 respectively (see table). quarterly from participant AMCs and peer-reviewed; approved studies are
This compares favourably to other lung cancer survival models created using executed at all sites. Results: All 10 AMCs collaborated to execute a
different machine learning techniques (Jochems 2017, Dekker 2009). A Cox- master registry participation agreement, followed by a master IRB protocol
PH model created using the top 20 variables also had a significantly lower (New England IRB # 120180008), subsequently approved by individual
performance (see table). Analysis of input variables yielded distinctive patterns site IRBs. Since its launch, the PMEC database has grown, on average,
for patient subgroups and time points. Tumor status, medications, lab values 60% per year, to now house over 14,000 cases. The shared dataset covers
and functional status were found to be significant in patient sub cohorts. all tumor types (most commonly lung [17.2%], gastrointestinal [13.8%]
Conclusions: An AI model to predict survival of lung cancer patients built and breast [9.2%]), encompasses genomic alterations in .300 genes, and
using a large real world dataset yielded high accuracy. This general model can reports relevant supplementary data such as tumor mutation burden and
further be used to predict survival of sub cohorts stratified by variables such as microsatellite instability status. To date, 15 studies have been proposed
stage or various treatment effects. Such a model could be useful for assessing and evaluated using this platform, with 2 projects currently approved and
patient risk and treatment options, evaluating cost and quality of care or in progress. Conclusions: We demonstrated the feasibility of creating a
determining clinical trial eligibility. collaborative academic consortium that facilitates data sharing and po-
tential discovery efforts in oncology. Technology solutions can accelerate
Precision Recall F1 score Support AUC Cox PH AUC the ability of AMCs, in partnership with central labs, to share and har-
Death in 90D 0.61 0.44 0.51 2235 0.79 0.69 monize data to advance precision medicine. This approach lays the
Alive after 90D 0.82 0.9 0.86 6233 0.79 0.69 groundwork for conducting prospective, biomarker-enriched clinical trials
Death in 180D 0.69 0.53 0.6 3151 0.79 0.7 among participating AMCs.
Alive after 180D 0.76 0.86 0.8 5317 0.79 0.7
Death in 360D 0.74 0.67 0.7 4073 0.79 0.71
Alive after 360D 0.72 0.78 0.75 4395 0.79 0.71

6558 Poster Session (Board #249), Sat, 1:15 PM-4:15 PM 6559 Poster Session (Board #250), Sat, 1:15 PM-4:15 PM
Use of machine learning to identify relevant research publications in clinical Patient reported outcomes and predictors of distress. First Author:
oncology. First Author: Fernando Jose Suarez Saiz, IBM Watson Health, New Patrick Leland Meadors, Levine Cancer Institute, Atrium Health, Charlotte, NC
York, NY
Background: ASCO/NCCN guidelines recommend screening for multifacto-
Background: Finding high-quality science to support decisions for individual rial distress in cancer patients. Understanding predictors of cancer related
patients is challenging. Common approaches to assess clinical literature distress can lead to early intervention and improve clinical outcomes,
quality and relevance rely on bibliometrics or expert knowledge. We describe a symptom management, and operational efficiency. Through electronic distress
method to automatically identify clinically relevant, high-quality scientific screening (EDS), patient reported outcomes (PRO) were collected across 42
citations using abstract content. Methods: We used machine learning practice locations. Methods: EDS has 39 questions related to cancer related
trained on text from PubMed papers cited in 3 expert resources: NCCN, NCI- distress including: distress, cancer symptoms/side effects, malnutrition, de-
PDQ, and Hemonc.org. Balanced training data included text cited in at least pression, anxiety, social/family support, financial, and spiritual concerns.
two sources to form an “on topic” set (i.e., relevant and high quality), and an 27,106 patients completed screens between 2017-2018. Multivariate
“off-topic” set, not cited in any of the above 3 sources. The off-topic set was analysis and logistic regressions determined predictors of distress for com-
published in lower ranked journals, using a citation-based score. Articles pleted screens overall, registry matched, and within 60 days of diagnosis.
were part of an Oncology Clinical Trial corpus generated using a standard Results: Median age was 59 (IQR 18-101) and 65% were female. Five
PubMed query. We used a gradient boosted-tree approach with a binary symptoms consistently predicted clinically significant distress $ 4: anxiety,
logistic supervised learning classification. Briefly, 988 texts were processed fatigue, pain, poor emotional coping, and sleep. Diagnosis (dx), staging at time
to produce a term frequency-inverse document frequency (tf-idf) n-gram of dx, and timing of screen did not independently predict distress. Factors
representation of both the training and the test set (70/30 split). Ideal predicting clinically significant distress varied across geographic regions.
parameters were determined using 1000-fold cross validation. Results: Our Conclusions: In large patient population, five key PROs are predictive of
model classified papers in the test set with 0.93 accuracy (95% CI (0.09: clinically significant distress and could potentially impact clinical outcomes.
0.96) p # 0.0001), with sensitivity 0.95 and specificity 0.91. Some false Early PROs predictive of distress were consistent along the continuum, thus
positives contained language considered clinically relevant that may the importance of early symptom identification. EDS can help custom tailor
have been missed or not yet included in expert resources. False negatives supportive oncology programs to mitigate symptoms related to cancer distress.
revealed a potential bias towards chemotherapy-focused research over ra- Multivariate Analysis of Distress $ 4 (Odds Ratio/95% CI).**
diation therapy or surgical approaches. Conclusions: Machine learning can
Total Screened Registry Matched Registry Matched and within 60 days
be used to automatically identify relevant clinical publications from bio- (N=27,106)* (N=9,438)* of DX (N=5,894)*
graphic databases, without relying on expert curation or bibliometric
Pain 1.21 [1.94, 1.23] 1.18 [1.15, 1.21] 1.20 [1.16, 1.24]
methods. The use of machine learning to identify relevant publications may Fatigue 1.13 [1.11, 1.14] 1.10 [1.07, 1.12] 1.07 [1.04, 1.10]
reduce the time clinicians spend finding pertinent evidence for a patient. Sleep Concern 1.08 [1.07, 1.09] 1.11 [1.09, 1.13] 1.11 [1.08, 1.15]
This approach is generalizable to cases where a corpus of high-quality Anxiety (5/6) 6.38 [4.54, 8.95] 7.75 [4.30, 10.33 [4.68, 22.82]
13.98]
publications that can serve as a training set exists or cases where docu- Anxiety (6/6) 5.76 [4.32, 7.69] 6.87 [4.31, 7.07 [3.96, 12.63]
ment metadata is unreliable, as is the case of “grey” literature within on- 10.94]
cology and beyond to other diseases. Future work will extend this approach Concern for Family 1.35 [1.23, 1.48] 1.31 [1.14, 1.51] N/A
and may integrate it into oncology clinical decision-support tools. Quite Poor Emo- 6.30[3.85, 8.59 [3.06, 6.93 [2.08, 23.11]
tional Coping 10.33] 24.09]
*p,.0002 **all test variables not included in chart but will be included in presentation

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
382s Health Services Research, Clinical Informatics, and Quality of Care

6560 Poster Session (Board #251), Sat, 1:15 PM-4:15 PM 6561 Poster Session (Board #252), Sat, 1:15 PM-4:15 PM
Impact of travel time on healthcare costs and resource utilization by phase of Esophageal cancer in Hispanic patients: A demographic analysis of the
care for older cancer patients. First Author: Gabrielle Betty Rocque, National Cancer Database. First Author: Juan Ricardo, Florida State Uni-
University of Alabama at Birmingham, Birmingham, AL versity College of Medicine, Sarasota, FL
Background: Closures of hospitals and clinics may have unintended conse- Background: Hispanics are the fastest-growing minority accounting for 18%
quences, including increasing patient travel time. Increased patient travel time of the US population. The National Cancer Institute estimated 17,290 new
to healthcare facilities has the potential to adversely impact patient outcomes. cases of esophageal cancer (EC) in the US in 2018. Hispanics are reported to
Limited data exist on the impact of travel time on healthcare costs and resource have lower EC prevalence. We sought to interrogate the demographic pat-
utilization. Methods: This retrospective cohort study from 2012-2015 eval- terns of EC in Hispanics. Secondary objective was to examine evidence of
uated drive time to cancer care site for Medicare beneficiaries age $ 65 in the socioeconomic disparities and differential therapy. Methods: We queried the
Southeastern US. The primary outcome was Medicare spending by phase of National Cancer Database to identify patients with EC between 2005–2015.
care (initial, survivorship, end of life [EOL]). Secondary outcomes included Patients were divided into two groups, Hispanic vs Non-Hispanic (NH). De-
resource utilization measured by hospitalization rates, hospitalization sites, mographics compared were age, sex, tumor data, surgical intervention, type of
intensive care unit (ICU) admissions, and chemotherapy-related hospitaliza- treatment, insurance status, income, residence area, and Charlson/Deyo score.
tion rates. Hierarchical linear models with patients clustered within cancer Pearson’s Chi-square test was used to compare categorical variables. Groups
care site and adjusted for pertinent covariates were used to determine the were matched by propensity score-matched analysis (PSM). Survival analysis
effects of drive time on average monthly phase-specific Medicare spending. was estimated by the Kaplan-Meier method and associated log-rank test.
Results: Median drive time was 32 minutes (IQR 18-59) for the 23,382 in- P-value #0.05 was considered significant. Results: We identified 85,004
cluded Medicare beneficiaries, with 24% of patients driving . 1 hour to their patients with EC; 3,205 were Hispanic (3.8%). In this US population we
cancer care site. During the initial phase of care, Medicare spending was 14% identified significant disparities between the Hispanic and NH groups. Sta-
higher for patients traveling . 1 hour than those traveling # 30 minutes. tistically significant differences among Hispanics included higher prevalence
Hospitalizationrates were 4-13% higher for patients traveling . 1 hour vs. of squamous EC (24.7% vs 19.6%), higher likelihood of stage IV cancer
# 30 minutes in the initial (61 vs. 54), survivorship (27 vs. 26), and EOL (310 vs. diagnosis (40.7% vs. 34.8%), younger age, higher uninsured status (10.4% vs
86) phases of care (all p , .05). The majority of patients traveling . 1 hour 3%) with income , $38,000 (26.4% vs 15.9%), and Charlson/Deyo score 0
were hospitalized at a local hospital rather than at their cancer care (72.3% vs 70.7%) when compared to NH. However, Hispanics were less likely
site, whereas the converse was true for patients traveling # 30 minutes. to have surgical intervention (29% vs 36.3) and overall less likely to receive any
Conclusions: As healthcare locations close, patients living farther from treat- type of treatment (30.1% vs 26.1%). PSM showed that any treatment, in-
ment sites may experience more limited access to care, and healthcare spending surance status and lower income were predictors of survival. Treated Hispanics
could increase for Medicare. survived longer than NH (median survival 17 vs 15 months). Overall survival at
5 years was 22% vs 17%, respectively, p , 0.05. Conclusions: Despite lower
prevalence of EC in Hispanics compared to NH, there is a disproportionately
higher number of metastatic and untreated cases among Hispanics. This
disparity may be explained by Hispanics’ limited access to medical care ex-
acerbated by their socioeconomic and insurance status. Further clinical and
epidemiologic research is warranted to reveal other factors impacting these
health disparities.

6562 Poster Session (Board #253), Sat, 1:15 PM-4:15 PM 6563 Poster Session (Board #254), Sat, 1:15 PM-4:15 PM
Access to care and financial burden for patients with breast cancer in Ghana, Racial comparisons in receipt of timely guideline-based colon cancer treat-
Kenya, and Nigeria. First Author: Majid Twahir, Aga Khan University Hos- ment in an equal-access health system. First Author: Yvonne L Eaglehouse,
pital, Nairobi, Kenya Henry M. Jackson Foundation for the Advancement of Military Medicine,
Bethesda, MD
Background: Breast cancer is the most frequently diagnosed malignancy
and the most common cause of cancer-related death in women in Ghana, Background: Non-Hispanic Black (NHB) adults with colon cancer may have
Kenya, and Nigeria. We evaluated healthcare resource use and financial longer time-to-treatment and be less likely to receive guideline-based therapy
burden for patients treated at tertiary cancer centers in these countries. than Whites (NHW) in the U.S. This may be largely related to racial differences
Methods: Records of breast cancer patients treated at the following in access to care and insurance coverage. This study aimed to determine
government/private tertiary centers were included – Ghana: Korle-Bu whether there were racial differences in receipt of timely guideline-based
Teaching Hospital and Sweden Ghana Medical Centre; Kenya: Kenyatta colon cancer treatment in the equal-access Military Health System (MHS).
National Hospital and Aga Khan University Hospital; Nigeria: National Methods: Patients age 18-79 years diagnosed with colon adenocarcinoma
Hospital Abuja and Lakeshore Cancer Center. Patients presenting within a between January 1, 1998 and December 31, 2007 were identified in linked
prespecified 2-year period were followed until death or loss to follow-up. databases from the Department of Defense Central Cancer Registry and MHS
Results: The study included 299 patient records from Ghana, 314 from Data Repository. Odds ratios (ORs) and 95% confidence intervals (CIs) of
Kenya, and 249 from Nigeria. The use of common screening modalities (eg, receiving stage-specific treatment within recommended timeframes [surgery
mammogram, breast ultrasound) was , 45% in all 3 countries. Use of core within 6 weeks of diagnosis (stages I-III); adjuvant chemotherapy within
needle biopsy was 76% in Kenya and Nigeria, but only 50% in Ghana. Across 8 weeks of surgery (stages II-III); treatment within 4 weeks of diagnosis (stage
the 3 countries, 91-98% of patients completed blood count/chemistry, IV)] for NHB relative to NHW patients were estimated using multivariable
whereas only 78-88% completed tests for hormone receptor and human logistic regression. Results: Patients (n = 2,170) had a mean age at diagnosis
epidermal growth factor receptor 2 (HER2). Most patients underwent sur- of 59.6 (SD 11.8) years and the racial distribution was 78.6% NHW and
gery: mastectomy (64-67%) or breast-conserving Most patients in Ghana 21.4% NHB. The likelihood of receiving timely surgery between races was
and Nigeria (87-93%) paid for their diagnostic tests entirely out of pocket similar across the stage groups (I-III). NHB patients were equally likely to
(OOP) compared with 30-32% in Kenya. Similar to diagnostic testing, the receive adjuvant chemotherapy as NHW patients (OR 0.90, 95% CI 0.57,
proportion of patients paying OOP only for treatments was high: 72-89% in 1.41) and to receive it within 8 weeks of surgery (OR 1.19, 95% CI 0.75,
Nigeria, 45-79% in Ghana, and 8-20% in Kenya. Among those receiving 1.87). The likelihood of receiving timely treatment for patients with stage IV
HER2-targeted therapy, average number of cycles was 5 for patients paying disease was similar between races (OR 0.82, 95% CI 0.39, 1.69). The overall
OOP only vs 14 for patients with some level of insurance coverage. likelihood of receiving treatment adherent to stage-specific guidelines in the
Conclusions: Patients treated in tertiary facilities in sub-Saharan African study sample was similar between NHB and NHW patients (OR 1.00, 95% CI
countries lack access to common imaging modalities and systemic thera- 0.77 to 1.31). Conclusions: In the MHS population, the likelihood of receiving
pies. Most patients in Ghana and Nigeria bore the full cost of their breast timely treatment adherent to recommended guidelines was similar between
cancer care, suggestive of privileged financial status. Access to screening/ races. Our results support the role of equal access to medical care and in-
diagnosis and appropriate care is likely to be substantively lower for the surance coverage in reducing racial disparities in colon cancer treatment.
general population.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Health Services Research, Clinical Informatics, and Quality of Care 383s

6564 Poster Session (Board #255), Sat, 1:15 PM-4:15 PM 6565 Poster Session (Board #256), Sat, 1:15 PM-4:15 PM
Assessment of enrollment characteristics for Children’s Oncology Group (COG) Development of a financial navigation program to ease the burden of financial
upfront therapeutic clinical trials 2004-2015. First Author: Kelly Faulk, toxicity. First Author: Melissa Monak, Cleveland Clinic, Cleveland, OH
University of Colorado School of Medicine, Children’s Hospital Colorado,
Background: Due to the rising cost of cancer care, even well-insured patients
Center for Cancer and Blood Disorders, Department of Pediatrics, Aurora, CO
are at risk for financial toxicity. Cancer treatment costs can be challenging for
Background: Improvements in pediatric cancer survival are attributed to co- patients and families because of increasing deductibles, increasing premiums
operative clinical trials. Under representation of specific demographic groups and frequent co-payments. At Cleveland Clinic Cancer Centers, our goal is to
has been described in adult and pediatric cancer and poses a threat to the identify patients at risk for financial distress and help ease the financial stress
generalizability of trial results. A comprehensive evaluation of data provided by and hardship associated with cancer care. Methods: In 2015 we initiated
the Children’s Oncology Group (COG) of upfront trial enrollment for US patients huddles for all new patients to ensure seven day access for our patients as well
0 to 29 years old between 2004 and 2015 was performed to assess for as out of network authorizations. During this work we identified the need for a
disparities in participation. Methods: Estimates of cancer cases were calcu- position to help manage the costs of care for our patients. The Financial
lated using the Surveillance, Epidemiology, and End Results registry and the Navigator (FN) role was created in 2016 to complete benefits investigations to
US Census and compared to observed COG cases. Percent enrollment and determine covered services and minimize the patient’s out of pocket expenses.
Standardized Ratios of enrollment were calculated across various de- We are participants in the Oncology Care Model and are expected to provide
mographic, disease, and socioeconomic groups. The COG website was utilized and document the estimated out-of-pocket costs in the patient’s medical
to quantify available upfront trials during the study period and assess age record. Navigators conduct telephone outreach calls with treatment patients to
eligibility criteria. Results: 21.3% of estimated US cancer patients age 0 to 19 educate them on their benefits. The FN also assists with applications on copay
years enrolled on COG trials. Younger patients were consistently more rep- assistance and free drug options. Finally, the FN will connect patients with
resented across disease types and race/ethnicities. Patients with hematologic resources for premium assistance and household expenses when copay as-
malignancies were more represented compared to solid and central nervous sistance isn’t an option. The FN team has grown to 7 FTE’s since 2016.
system (CNS) tumors. Conclusions: COG clinical trial enrollment rates are Results: Since 2016 the dollar amount approved through copay assistance
declining, which may be due to challenges in pediatric drug development, applications has increased by 70%, the free drug value received increased by
difficulty designing feasible trials for highly curable diseases, and issues in 55%, and end of year 2018 data shows that we more than doubled what was
ensuring trial availability for the heterogeneous group of solid and CNS tumors. collected towards patient out of pocket costs from copay assistance programs
Though racial/ethnic groups and county-level socioeconomic factors were compared to 2017. We have also been able to comply with the out of pocket
proportionally represented, under representation of the adolescent/young adult expense metric for OCM, not only for our Medicare patients, but for all of our
(AYA) population and younger patients with solid and CNS tumors remain treatment patients and compliance has risen from 44% in December 2017 to
significant concerns. Targeted enrollment efforts should focus on the iden- 98% in April 2018 and we continued to obtain a goal of 96% or higher
tified subgroups and further research should evaluate AYA enrollment across throughout 2018. Conclusions: In conclusion our Navigators have built re-
all available trials to provide continued treatment advances for all patients. lationships with Social Workers, Care Coordinators, and community services
that are available to patients. This added benefit helps patients with non-
medical financial stressors. The team can connect patients to organizations
that can assist with household utilities, childcare concerns, food concerns, and
mortgages. Utilization of the FN role has helped the healthcare team and
patients to identify ways to mitigate the costs of care and distress related to
financial concerns.

6566 Poster Session (Board #257), Sat, 1:15 PM-4:15 PM 6567 Poster Session (Board #258), Sat, 1:15 PM-4:15 PM
Unique perspectives from the transgender community: A retrospective chart Effectiveness of mobile computerized tomographic (CT) lung scanning unit
review of cancer care needs for transgender patients. First Author: Carolyn for early diagnosis of lung cancer in under-served populations. First Author:
Moloney, Cork University Hospital, Cork, Ireland, Ireland Derek Raghavan, Levine Cancer Institute, Atrium Health, Charlotte, NC
Background: It is estimated that 1% of a population experience some degree Background: The National Lung Screening Trial (NLST) demonstrated that
of gender non-conformity. There is scant information worldwide on cancer screening high-risk patients with low-dose CT (LDCT) of the chest reduces
incidence and mortality for this population however due to a lack of investi- lung cancer mortality compared to screening with chest x-ray. Uninsured and
gating large-scale prospective studies. National cancer registries do not hold Medicaid patients lack access to this hospital-based screening test due to
demographics on this population. Current literature indicates transgender geographic isolation/socio-economic factors. We hypothesized that a mobile
people may face an increased cancer risk. Transgender patients may avoid screening unit would improve access and confer benefits demonstrated by
screening programmes for cancers which are themselves gendered. Trans- the NLST to this under-served group, which is most at risk of lung cancer
gender patients can feel excluded from gender specific cancer support groups. deaths. Methods: In collaboration with Samsung Inc, we inserted a BodyTom
We set out to identify how cancer services in Ireland can better meet trans- portable 32 slide low-dose CT scanner into a 35-foot coach, reinforced to
gender people’s unique needs. Methods: Medical oncology consultants in the avoid equipment damage, to function as a mobile lung scanning unit. The
South/South-West of Ireland were contacted to identify patients who identified unit includes a waiting area, high speed wireless internet connection for
as transgender or gender non-conforming. We carried out a retrospective chart rapid image transfer, and electronic tablets to deliver smoking cessation and
review of the four transgender patients identified. We analysed staging at health education programs and shared decision-making video aids. It has
diagnosis, family supports, smoking history, alcohol use and whether cancer been certified as a lung cancer screening Center of Excellence by Lung
treatment affected gender transitioning treatment and if this had documented Cancer Alliance. We employed the LUNG RADS approach to lesion classi-
effects on mental well-being. We also noted if medical records reflected a new fication, yielding high sensitivity and specificity in assessment. All films were
name or change of gender and if not, whether original name and gender used reviewed by a central panel of oncologists, pulmonologists and radiologists.
for chemotherapy and blood product administration. Results: All four patients The protocol was approved by Chesapeake IRB, which oversees all LCI
were diagnosed with relatively advanced disease at diagnosis- Stage IIIc high cancer trials. Interim analysis at this time was approved by the Oversight
grade ovarian cancer, stage IV gastrointestinal tumour, stage IVb diffuse large Committee. Results: We screened 470 under-served smokers between 4/
B Cell and locally advanced extra-abdominal desmoid tumour. Of the four 2017-1/2019; M:F 1.1:1, mean age 61 years (range 55-64), with average
patients, three had a smoking and alcohol history on diagnosis. All four pa- pack year history of 45.7 (30-150) (25% African-American; 3% Hispanic;
tient’s recent medical correspondence reflected a name and gender change 65% rural; 100% uninsured, under-insured or Medicaid - NC Medicaid does
but the medical records did not reflect this. Three patients had documented not cover lung cancer screening). Patients over the age of 64 years were
depression for which they were attending psychiatry services. It was noted that excluded as they are covered by Medicare for lung cancer screening. We
two patients had gender transitioning treatment postponed due to cancer care. found at initial screen 35 subjects with LUNG RADS 4 lesions, 49 subjects
Minimal family support was noted for two patients. Conclusions: The trans- with LUNG RADS 3 lesions, 10 lung cancers (2.1%), including 4 at stage
gender community is a growing population that will continue to integrate into I-II. 4 non-lung cancers were identified and treated. Other incidental non-
mainstream society. Our retrospective chart review adds to a growing body of oncologic findings are the subject of another presentation. Conclusions: In
evidence which suggests gender minorities may suffer from cancer-related this small sample using the first mobile low dose CT lung screening unit in
disparities and have an increased need for psychosocial support. As in other the United States, the initial cancer detection rate is comparable to that
studies, it is difficult to identify these individuals. We should identify gender reported in the NLST but with marked improvement of screening rates in
minority individuals and report this data in medical records in order to build underserved groups and with better anticipated outcomes at lower cost than
much needed epidemiological information. if they had first presented with metastatic disease.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
384s Health Services Research, Clinical Informatics, and Quality of Care

6568 Poster Session (Board #259), Sat, 1:15 PM-4:15 PM 6569 Poster Session (Board #260), Sat, 1:15 PM-4:15 PM
Australian population-based study of single- versus multi-fraction palliative Clinical impact of the Mexican healthcare system "Seguro Popular" on breast
radiotherapy for bone metastases. First Author: Wee Loon Ong, Department cancer survival. First Author: Luis Antonio Cancel, Centro Universitario
of Radiation Oncology, Olivia Newton John Cancer Wellness and Research Contra el Cáncer, Hospital Universitario, UANL, San Pedro Garza Garcia, NL,
Centre, Austin Health, Heidelberg, Australia Mexico
Background: Single fraction palliative radiotherapy (SFRT) has been shown to Background: Breast cancer (BC) is one of the leading issues in public health in
be equivalent to multi-fraction radiotherapy (MFRT) for bone metastases low and middle-income countries. In Mexico, access to healthcare is frag-
symptom management. Remuneration for radiotherapy (RT) in Australia are mented according to the patient´s employment and not by its needs; IMSS and
largely determined by fractions delivered. We aim to determine the use of SFRT ISSTE (Social Security) provide access to prepaid medicine to those under the
for bone metastases in Australia. Methods: We did a population-based linkage formal sector of the economy, leaving up to 50 million Mexicans without access
study of multiple administrative healthcare databases in Victoria, Australia: the to a prepaid scheme. In 2003, the Seguro Popular (SP) was created in order to
Victorian Radiotherapy Minimum Data Set (VRMDS), the Victorian Cancer bring universal access to prepaid medicine in Mexico, and in 2007 expanded
Registry (VCR), and the Birth, Death and Marriage registry (BDM). All patients its coverage for BC. Methods: Retrospective and comparative study. The
with solid tumour (excluding primary bone cancer) who received palliative primary endpoint was to determine the impact on survival of SP on BC. Records
radiotherapy for bone metastases between 2012 and 2017 were included. The were obtained from the electronic database of the Hospital Universitario “Dr.
primary outcome was use of SFRT. The Cochrane-Armitage test for trend was José Eleuterio González”. We included patients with invasive BC stage I-IV.
used to evaluate SFRT use over time. Multivariable logistic regression was used Patients with any other kind of healthcare schemes other than SP, patients who
to identify factors associated with SFRT use. Results: A total of 15,668 underwent treatment outside our institution, and those with a follow up no
courses of RT for bone metastases were delivered to 10,351 patients. The greater than 3 months were excluded. 104 patients from the period prior the
overall proportion of SFRT was 18% (2,746/ 15,668). There was no significant implementation of the SP (2000-2007) met the criteria for evaluation;
change in SFRT use over time, from 18% in 2012 to 20% in 2017 (P = 0.06). thereafter we randomly selected a second cohort with the same size from the
Older patients were more likely to have SFRT (mean age 69.4 vs. 68.2, P , period after the implementation of the SP (2008-2013). Results: Median age
0.001). Patients who had lung cancer (21%) and prostate cancer (19%) were at diagnosis was 48 and 51 years, respectively, for the periods before and after
more likely to have SFRT compared to other tumour types (P , 0.001). Spine the implementation of SP. Distribution by clinical stage (Non-SP vs SP): CS I,
RT was associated with lower use of SFRT compared to other treatment sites 4.8 vs 10%, CS II, 31 vs 44%, CS III, 52 vs 38%, and CS IV, 10 vs 6.7%.
(14% vs. 22%, P , 0.001). Patients from remote area were more likely to have Molecular subtypes distribution (Non-SP vs SP): Luminal, 61 vs 62%, HER2
SFRT compared to patients from major cities (22% vs. 17%, P , 0.001). Positive (IHC+++/FISH+) 17 vs 22%, TNBC, 21 vs 18%, unknown 6.7 vs
Patients treated in private institutions were less likely to have SFRT compared 5.7%. Regarding survival, we observed a statistically significant difference on
to those treated in public institutions (10% vs. 22%, P , 0.001). In multi- progression-free survival and overall survival favoring the SP cohort; PFS at 5
variate analyses, patients’ age, tumour type, area of residence, and treatment years, 54 vs 81% (p = , 0.0001) and OS at 5-year, 72 vs 86% (p = 0.01).
institutions were independently associated with SFRT use. Conclusions: This Conclusions: We present evidence that the Mexican healthcare scheme SP,
is the largest Australian population-based cohort treated with RT for bone created to bring medical access to those patients without prepaid health
metastases, with low utilisation of SFRT over time. There is large variation in protection, provides a significant clinical benefit on survival (PFS and OS) in
SFRT use depending on patient-, tumour-, geographical and institutional women with breast cancer.
factors. Further work is needed to increase uptake, and reduce unwarranted
variation, in SFRT use.

6570 Poster Session (Board #261), Sat, 1:15 PM-4:15 PM 6571 Poster Session (Board #262), Sat, 1:15 PM-4:15 PM
Impact of race/ethnicity in the clinical presentation and outcomes of patients Forecasting the global need for high-dose methotrexate to prevent and
with multiple myeloma in an underserved urban population. First Author: mitigate shortages. First Author: Scott C. Howard, University of Tennessee
Christian Torres, John H. Stroger Jr. Hospital of Cook County, Chicago, IL Health Sciences Center, Memphis, TN
Background: Patients with multiple myeloma (MM) who are part of racial/ Background: Causes of drug shortages in low-income and middle-income
ethnic minority groups have been typically underrepresented in large de- countries are numerous and often more than one cause contributes. Short-
scriptive and randomized-controlled studies. Despite the identification of ages are especially problematic when they involve drugs like high-dose
biological and genetic risk factors, the impact of race/ethnicity in the out- methotrexate (HDMTX), which is required for several highly curable cancers
comes of patients with MM remains largely unknown. We aimed to describe and has no good substitute. Morocco is currently experiencing a national
the racial/ethnic differences in clinical presentation and outcomes of pa- shortage of high-dose methotrexate that has so far lasted several months and
tients with MM in an ethnically-diverse underserved urban population. has impacted the care of all patients whose protocol includes HDMTX. In this
Methods: We conducted a single-center retrospective study of patients with case, a single supplier left the market due to government-imposed pricing that
MM from Jan 1st 2008 – Dec 31st 2016 using ICD coding from our tumor made continued sales increasingly unprofitable. Practicing hematologist/
registry. We abstracted demographic, clinical and treatment variables. We oncologists in Myanmar, the Philippines, Vietnam, and Cambodia also re-
used Chi-square to compare categorical variables and Kaplan-Meier method ported shortages that reached patients during 2018. Methods: We created a
for survival analysis. Statistical analysis was performed using IBM SPSS forecasting model to identify the total annual national need for HDMTX and
version 25. Results: We identified 73 patients with MM with a median follow number of patients based on incidence by age, age distribution, and numbers
up time of 42 months (Range 1 to 81 months). Patients had a median age of of doses in typical protocols for osteosarcoma, primary CNS lymphoma, pe-
59 years (Interquartile Range [IQR] =17) and were predominantly male diatric B-cell NHL, and ALL, curable cancers for which HDMTX is essential.
(54.8%). The most frequent racial/ethnic group was African American (AA) Results: More than 200,000 patients per year need HDMTX (approximately
(59%) followed by Hispanic (27%) and Caucasian (8%). When compared to 950,000 doses, see Table). 86% of these live in low- and middle-income
other ethnicities, patients who were AA had higher ISS-3 scores (41% vs countries, where access to inexpensive generic drugs is especially important.
23% p=0.101) worse cytogenetic risk (65% vs 30% p=0.009) and worse Conclusions: This forecasting tool will facilitate national purchasing and ne-
response after induction (Complete response [CR] 47% vs 77% p=0.047). gotiating to assure continuous supply of HDMTX. It can be adapted for other
They were also more likely to have medical insurance coverage than other drugs, and is especially relevant for low-cost generic drugs, which are dis-
ethnicities (67% vs 27% p=0.003) but had similar access to autologous proportionately affected by shortages.
bone marrow transplant (23% vs 23% p= 0.99). Overall, AA patients had
worse overall survival (OS) compared to all other ethnic groups (mean OS: World bank income Number of newly-diagnosed Number of HDMTX doses
group, cancer type, and patients who need HDMTX needed annually for newly-
58.3 months vs 79 months p=0.014). Conclusions: AA patients with MM age group each year diagnosed patients
had more aggressive disease and worse OS compared to other ethnicities
which may suggest an underlying genetic predisposition towards high-risk Low-income 21032 95854
genetic features. Improvement of access to autologous bone marrow Lower-middle-income 85720 400848
Upper-middle-income 65931 316440
transplantation may improve survival in high-risk racial/ethnic groups. High-income 28667 139394
Osteosarcoma 20907 167256
Primary CNS 31783 190698
lymphoma
Pediatric B-cell NHL 16111 64444
Acute lymphoblastic 132626 530504
leukemia

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Health Services Research, Clinical Informatics, and Quality of Care 385s

6572 Poster Session (Board #263), Sat, 1:15 PM-4:15 PM 6573 Poster Session (Board #264), Sat, 1:15 PM-4:15 PM
Cervical cancer screening in incarcerated women: An experience from the first A novel patient-centric approach between sites, CRO, and sponsor to accel-
cervical cancer screening campaign in a southern Thailand correctional erate FPI and drive more patients into oncology protocols. First Author:
facility. First Author: Ingporn Jiamset, Division of Gynecologic Oncology, Krystyna Kowalczyk, Optimal research, Raleigh, NC
Department of Obstetrics and Gynecology, Prince of Songkla University Faculty
Background: According to the Journal of Clinical Oncology, 50% of sites
of Medicine, Hatyai, Thailand
performing clinical trials never enroll a patient. And on top of that, it can take
Background: Cervical cancer is one of the most preventable cancers, not only several months to activate a site for an oncology clinical trial; precious time
presence of effective HPV vaccination but also simple and robust screening that patients cannot afford when they’ve had a cancer diagnosis and precious
methods such as Pap test. Nevertheless, there were some women at risk whom time sponsors need when developing new compounds for market. While
were unable to access screening cause of incarceration. Hence, in 2018, many patients are interested in participating in trials, they are limited in their
together with Songkhla Woman Correctional Institute, we launched a cancer opportunities because they do not live near a research site or work with a
screening campaign including clinical breast exam, mobile mammography and physician performing clinical trials. So with this crisis in the oncology field,
Pap test. This is the first report of cervical cancer screening result demon- research needs to be more efficient and inclusive. Methods: A combined
strated the essential of cervical cancer screening in these disadvantaged partnership of sites, physicians, CRO and sponsor leveraging Just-in-Time
women. Methods: Due to the regulation of the jail, we had to limited bring-in enrollment methodology helped expedite clinical trial enrollment and di-
tools, allowed staffs and operating-time, therefore we used a pre-screening versify trial access driving faster first patient enrolled and expanding the
questionnaire, included 5 items: HIV infection, number of partner, parity, age potential patient denominator. Results: This on-demand methodology
at first sexual intercourse and number of term baby and each of them scored as augmented existing sites that had access to oncology patients by providing
2 for “high-risk” and 1 for “low-risk”, total score ranged from 5 to 10. We broader access, faster, and with no quality loss. Strong partnerships between
ranked and chose the volunteer participants, who have HIV infection and/or CRO and sponsor then facilitated two-week site activation allowing every
with highest risk score, to undergo Pap test. Results: Of the 1328 ques- identified patient to be converted to a study subject. This methodology was
tionnaire responders, Their mean risk score was 7.3 (SD= 1.3). HIV infected repeated across seven protocols driving patents on trials within six weeks of
participants number were 34 (2.5%). Of the 200 screened-participants, None trial available. Conclusions: The benefits of this Just-in-Time methodology
of them had ever received HPV vaccination before, and all participant did not touch all areas of clinical trials: Patients have greater clinical trial access: A
have Pap test since imprisonment. (mean 53.8 m, range 13-236 m, SD 36.7). larger denominator of patients across broader geographies have local access
Their score ranged between 8 to 10, 42.5% of them had score level 8, 54.5% to portfolios of clinical trials; Trials start to enroll faster: Patients can be
had score level 9 and 3% had score level 10. Mean age was 37.7 years. randomized into oncology clinical trials within two weeks of study start up
10 (5%) of them had abnormal Pap test; 1 of them showed ASC-US, 1 was driving trial time to completion; Sites have more trial options to consider:
LSIL, 1 was ASC-H, 5 of them showed HSIL and 2 of them showed squamous Sites have a broader portfolio of trials to access on demand without added
cell carcinoma and small round cell carcinoma. Final histopathological test administrative burden; Trials complete faster: Sponsors accrue patients
resulted in 6 of cervical intraepithelial neoplasia (CIN) I, metaplasia and faster driving expedited timelines and accelerating drug development.
cervicitis, 3 were diagnose CIN III and 1 diagnosed microinvasive carcinoma.
Incidence of cervical cancer was higher than normal population in this region.
(0.5% vs 0.02%). Conclusions: Incarcerated women were at high risk of
cervical cancer compared to normal population. Unfortunately, in many
places, they were unconditionally inaccessible to the cervical cancer pre-
ventive healthcare system for years. Social should increase awareness to
decrease this health disparity.

6574 Poster Session (Board #265), Sat, 1:15 PM-4:15 PM 6575 Poster Session (Board #266), Sat, 1:15 PM-4:15 PM
Investigation of HBOC germline mutations in women diagnosed with breast Diagnostic and treatment delays in young women with breast cancer. First
cancer in Trinidad and Tobago. First Author: Gerneiva Parkinson, Yale New Author: Philip D Poorvu, Dana-Farber Cancer Institute, Boston, MA
Haven Hospital, New Haven, CT
Background: Delays in diagnosis (dx) and treatment (tx) affect breast cancer
Background: Trinidad and Tobago (T&T) is the southern-most Caribbean (BC) outcomes. We sought to identify factors associated with delays among
island, and according to the WHO/PAHO, it has the 2nd highest breast young women, who do not undergo routine screening and often have pregnancy
cancer mortality rate in the region. Notably, a large proportion of breast or breastfeeding-related breast changes that may mask a BC. Methods: The
cancer cases in T&T occur at a young age; with nearly 36% of them being Young Women’s Breast Cancer Study is a multicenter, prospective cohort that
diagnosed under the age of 50. There is a known association between a enrolled 1302 women with newly dx BC age #40 between 2006-2016.
younger age at diagnosis and Hereditary Breast and Ovarian Cancer syn- Women reported the method and timing of cancer detection on the baseline
drome (HBOC). Yet, the prevalence of HBOC mutations remains unknown in survey. 231 were ineligible or excluded due to missing information. Among
T&T, as genetic counseling and testing services are extremely limited in the those reporting self-detected cancers, using multivariable regression we
region. Therefore, we sought to determine the prevalence and spectrum of evaluated factors associated with delays $90 days (d) from symptom to
HBOC mutations in T&T. Methods: At the National Radiotherapy Center, presentation (self delay) and presentation to dx (care delay); in stage 0-III BC
T&T’s main oncology unit, female breast cancer patients, who met NCCN we evaluated delays $60d from dx to tx (tx delay). Results: 1071 eligible
criteria for further genetic counseling and testing were recruited through women had median age at dx of 37 yrs (17-40) and 74% reported self-detected
chart reviews. After pre-test counseling, enrolled subjects had a detailed cancers. Self delay or care delay $90d was reported in 17% and 13%, re-
interview about their personal breast cancer diagnosis and family history. A spectively. Factors inversely associated with self delay included pregnancy at
saliva sample was collected using an Oragene kit, and analyzed by Color dx (vs nulliparous, OR 0.10, CI 0.01-0.78) and perceived financial comfort (vs
Genomics Inc. for 30 genes associated with hereditary cancers. Finalized not, OR 0.62, CI 0.41-0.93). Women dx #1 year post-partum who breastfed
results were returned to patients by genetic counselors from Color Genomics. (vs nulliparous, OR 2.60, CI 1.14-5.93) and those with a family history of
Results: A total of 118 female patients who met NCCN guidelines for HBOC breast/ovarian cancer (vs none, OR 1.79, CI 1.00-3.19) were more likely to
testing received genetic testing. A majority were 50 years of age or younger have a care delay. Age was inversely associated with care delays (OR 0.94, CI
(69/118, 59%). The cohort was ethnically diverse: 34% African, 15% Asian, 0.89-0.99). Tx delay was reported by 10% (105/1015), and associated with
48% multiple ethnicity, and 3% other/unknown. A pathogenic or likely being single (vs partnered, OR 1.61, CI 1.02-2.56 ), non-white (vs white, OR
pathogenic variant (positive result) was identified in 21.2% of the cohort 1.85, CI 1.09-3.13) and having Stage 0 BC (vs stage 1, OR: 3.08, CI 1.65-
(25/118) - most commonly identified in the BRCA1 gene (13/25, 52%), 5.77); women with stage 3 BC (vs stage 1, OR 0.13, CI 0.03-0.56) were less
followed by BRCA2 (5/25, 20%), PTEN (2/25, 8%), BRIP1 (1/25, 4%), likely to have a tx delay. Conclusions: In this cohort, most young women with
CHEK2 (1/25, 4%), MSH6 (1/25, 4%), PALB2 (1/25, 4%), and RAD51C (1/ BC underwent timely dx and tx initiation. Women dx #1 year post-partum who
25, 4%). Conclusions: We found a strikingly high HBOC germline mutation breastfed were more likely to experience a care delay, likely because lactational
prevalence rate of 21.2% among a cohort of female breast cancer patients changes may mask BC signs and symptoms. The associations of perceived
meeting NCCN criteria for HBOC testing in T&T. Given the growing impli- financial status with self delay and non-white race with tx delay underscore the
cations of germline HBOC mutations for breast cancer treatment and pre- need for additional support to ensure timely care for underserved populations
vention, our results demonstrate an urgent need for funding, as well as the with the goal of eliminating disparities in outcomes.
development of robust genetic counseling and testing services in T&T.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
386s Health Services Research, Clinical Informatics, and Quality of Care

6576 Poster Session (Board #267), Sat, 1:15 PM-4:15 PM 6577 Poster Session (Board #268), Sat, 1:15 PM-4:15 PM
Association of diagnosing physician and hospital characteristics with the use Disparities in lung cancer outcomes for veterans with comorbid mental disorders.
of radical cystectomy among patients with muscle-invasive bladder cancer. First Author: Jacob E Berchuck, Dana–Farber Cancer Institute, Boston, MA
First Author: Hemalkumar B Mehta, The University of Texas Medical Branch
Background: Emerging evidence suggests that cancer patients with comorbid
at Galveston, Galveston, TX
mental disorders have worse outcomes. Lung cancer is the leading cause of
Background: Only one out of five muscle-invasive bladder cancer patients receive cancer-related death in the United States, yet the impact of specific mental
radical cystectomy, a guideline-recommended treatment. Prior studies evaluated disorders on outcomes for patients diagnosed with lung cancer is not
patient characteristics associated with radical cystectomy use. We aimed to de- well known. Methods: Stage at diagnosis and receipt of stage-appropriate
termine bladder cancer diagnosing physician and hospital characteristics associated treatment were assessed for 55,315 Veterans with non-small cell lung
with the use of radical cystectomy. Methods: This cohort study used linked SEER- cancer (NSCLC) in the Veterans Affairs (VA) Central Cancer Registry from
Medicare Data from 2002 to 2011. We included older adults (age.65 years) di- 2000 to 2011. Patients were stratified by the presence of specific comorbid
agnosed with muscle-invasive bladder cancer. For each patient, a urologist who mental disorders. Multivariate analysis evaluated the association between
performed transurethral resection of bladder tumor was assigned as a diagnosing
mental disorders and survival, as well as the impact of VA treatment pro-
physician. The diagnosing physician was assigned to one hospital based on where he/
she performed more than half of all urologic surgeries. Two-level hierarchical model
grams on survival. Results: Patients with schizophrenia had lower rates of
(patients nested within hospitals) were constructed to determine the association of localized disease at diagnosis compared to those without mental disorders.
patient, physician and hospital characteristics with radical cystectomy use. Schizophrenia and dementia were associated with significantly lower rates
Results: A total of 7,097 patients were diagnosed by 4601 physicians who were of stage-appropriate treatment for localized, locoregional, and metastatic
affiliated with 822 hospitals. Overall, the radical cystectomy utilization rate was disease. Conversely, patients with depression or PTSD were more frequently
26.5%. Only 4.8% of the variation in radical cystectomy was attributed to the diagnosed with early-stage disease and significantly more likely to receive
hospital level. In the two-level hierarchical model, patients diagnosed by female stage-appropriate treatment across all stages. After adjusting for baseline
physicians were more likely to undergo radical cystectomy (32.8% vs. 25.8%; differences, stage, and rates of stage-appropriate treatment, hazard of death
OR=1.63, 95% CI=1.31-2.02). Higher radical cystectomy volume by diagnosing was higher for patients with schizophrenia (hazard ratio [HR] 1.10; 95% CI,
physicians and hospitals increased the radical cystectomy use (Table). Diagnosing 1.03-1.16; P , .005) and dementia (HR 1.11; 95% CI, 1.06-1.18; P ,
physician characteristics (age, years in practice, employment status), and hospital .0005). Participation in VA-based programs to address mental illness,
characteristics (teaching states, location, type of hospital) were not associated with substance use, and homelessness was associated with a significant re-
radical cystectomy use. Patient characteristics such as age, male, higher comor- duction in all-cause mortality (HR 0.71; 95% CI, 0.68-0.75; P , .0001)
bidity burden were associated with lower likelihood of radical cystectomy use. and lung-cancer specific mortality (HR 0.73; 95% CI, 0.69-0.77; P ,
Conclusions: Physician and hospital factors do not largely contribute to the receipt of .0001). Conclusions: Schizophrenia and dementia are strong negative
radical cystectomy. Radical cystectomy volume by diagnosing physicians and
predictors of survival among Veterans diagnosed with NSCLC. VA-based
hospital drives radical cystectomy use.
mental health treatment programs are associated with reductions in all-
Association of physician and hospital volume with radical cystectomy use.
cause and lung cancer-related mortality, highlighting the importance of
Characteristics Radical Cystectomy Use (%) Odds ratio (95% CI)
funding and promoting mental health and supportive programs.
Diagnosing physician radical cystectomy volume (ref=0) 18.3
1-5 36 2.75 (2.39-3.17)
6-10 34.2 2.50 (1.85-3.40)
10+ 41.2 2.41 (1.84-3.16)
Hospital volume of radical cystectomy (ref = 0) 20.6
1-20 24.5 1.14 (0.90-1.45)
21-40 26 1.16 (0.86-1.56)
40+ 36 1.62 (1.13-2.32)

6578 Poster Session (Board #269), Sat, 1:15 PM-4:15 PM 6579 Poster Session (Board #270), Sat, 1:15 PM-4:15 PM
Impact of smoke-free ordinance strength on smoking prevalence and lung Emergency department visits for prescription and synthetic opioid overdoses
cancer incidence. First Author: Ryan Nguyen, Indiana University School of among patients with cancer. First Author: Vikram Jairam, Department of
Medicine, Indianapolis, IN Therapeutic Radiology, Yale University School of Medicine, New Haven, CT
Background: Smoke-free ordinances (SFO) have been shown to decrease the Background: Patients with cancer may be at high risk of opioid dependence
prevalence of cardiovascular and pulmonary disease, but there is limited data due to physical and psychosocial factors, although little data exists to inform
on the impact of such policies on lung cancer incidence. We investigated the providers and policymakers. Our aim is to examine overdoses from pre-
relationship between strength of county-level SFO with smoking prevalence scription and synthetic opiates leading to emergency department (ED) visits
and lung cancer incidence in Indiana. Methods: Following IRB approval, we among patients with cancer in the United States. Methods: The Healthcare
queried the Indiana State Cancer Registry and Indiana Tobacco Prevention and Cost and Utilization Project Nationwide Emergency Department Sample
Cessation Commission’s policy database between 1995 and 2016. County- (HCUP-NEDS) was queried for all patient visits with a primary diagnosis of
level characteristics included population, income, poverty, education, race/ prescription or synthetic opioid overdose between 2006 and 2015. Baseline
ethnicity, sex, and rurality. Lung cancer diagnosis and stage were also col- differences between patients with and without cancer were assessed using
lected. Using generalized estimating equations (GEE) with robust standard chi-square and ANOVA testing. Overdose rates by primary cancer site were
errors and accounting for the clustering effect at county level, we performed normalized using prevalence data from the Surveillance, Epidemiology,
multivariable analyses of smoking prevalence and age-adjusted lung cancer and End Results (SEER) Program. Weighted frequencies were used to create
rates with respect to the strength of smoke-free ordinances at the county level national estimates for all data analyses. Results: There were 682,820
over time. Results: Indiana consists of 92 counties, 24 of which had SFO by weighted ED visits for synthetic opioid overdose, among which 34,547 (5.1%)
2011. In 2012, Indiana enacted a law mandating at least a moderate state- visits were also associated with a diagnosis of cancer. During this timeframe,
wide SFO. From 1995 to 2016, 110,935 Indiana residents were diagnosed ED visits for opioid overdose among patients with cancer increased 2.5-fold,
with lung cancer. Indiana also had an average age-adjusted yearly lung cancer compared to 1.7-fold among those without cancer. 16.5% of patients with
incidence of 76.8 per 100,000 population and average yearly smoking cancer had metastatic disease. Patients with cancer presenting for opioid
prevalence of 25% during this time. Smoking prevalence was 1.2% (95% CI overdose had higher risk of hospital admission (74.8% vs 49.6%), respiratory
[-1.88, -0.52]) lower in counties with comprehensive or moderate SFO intubation (13.2% vs 12.2%), mortality (2.1% vs 1.1%), and cost-of-hospital-
compared with those with weak or no SFO. Counties that had comprehensive or stay ($32,665 vs $31,824) compared to their non-cancer counterparts (all
moderate SFO had an 8.36 (95% CI [-11.45, -5.27]) decrease in new lung P , 0.05). Primary cancers with the highest normalized overdose rates (ED
cancer diagnosis per 100,000 population per year compared with counties visits per 10,000 patients) were esophagus (134), liver & intrahepatic bile
that had weak or no SFO. Conclusions: Stronger municipal smoke-free air duct (124), and cervical cancer (124). Other common cancers had the fol-
ordinances are associated with decreased smoking prevalence and fewer new lowing normalized overdose rates: lung (105), head and neck (70), and breast
lung cancer cases. Strengthening smoke-free ordinances is paramount to (26). Conclusions: Approximately 5% of all ED visits due to prescription and
decreasing lung cancer incidence. synthetic opioid overdose are among patients with cancer. The rate of increase
in ED visits due to opioid overdose from cancer patients was nearly 50% higher
than that from non-cancer patients over the 10-year study period. Patients with
esophageal, liver, and cervical cancer may be at highest risk.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Health Services Research, Clinical Informatics, and Quality of Care 387s

6580 Poster Session (Board #271), Sat, 1:15 PM-4:15 PM 6581 Poster Session (Board #272), Sat, 1:15 PM-4:15 PM
Is immunotherapy toxicity associated with improved overall survival among older Low nationwide utilization of adjuvant chemotherapy (AC) in elderly patients
adults with advanced cancer? First Author: Andrew Johns, Dept. of Internal with localized non-small cell lung cancer (NSCLC). First Author: Kaushal
Medicine, The Ohio State University Wexner Medical Center, Columbus, OH Parikh, Mayo Clinic, Rochester, MN
Background: There is growing evidence that checkpoint inhibitor immuno- Background: While AC is recommended for all patients with stage IIA, IIB, and IIIA NSCLC, its
use and benefit among the elderly population is unclear. Methods: We identified patients with
therapy (IO) toxicity is associated with improved treatment response. There
stage IIA, IIB, or IIIA NSCLC using the National Cancer Database from 2006-2014. Patients
is a paucity of evidence examining the relationship between toxicity and were divided into age groups #65, 66-70, 71-75, 76-80, . 80 years. Trends in AC use, factors
overall survival (OS) in older adults. Methods: We performed a single in- influencing AC administration, and outcomes associated with AC were studied. Results: Out of
stitution retrospective cohort study of adults who received IO for advanced 27368 patients, 13464 received AC and 13904 did not. 11% had stage IIA disease, 50% had
cancer from 2011-2017. Baseline clinical characteristics were abstracted stage IIB disease and 39% had stage IIIA disease. AC use was lower with increasing age (49%
age #65 vs. 3% age . 80 (P , 0.0001)). Temporal use of chemotherapy for each age group was
from the electronic health record. Immune-related toxicities were graded by unchanged from 2006-2014. In multivariate logistic regression analysis, factors predictive of
physicians based on Common Terminology for Adverse Events criteria, v4.0. lower AC use were older age, Medicare or Medicaid insurance, academic center, and higher
Bivariate analysis with chi-squared statistics was used to describe baseline Charlson Comorbidity Index (CCI). Stage IIIA (OR 1.4, 95% CI 1.3-1.6) and Stage IIB (OR 1.3,
characteristics of patients $70 years (y) vs. ,70y. Survival outcomes were 95% CI 1.2-1.4) had a higher chance of receiving AC than stage IIA. In a Cox proportional hazard
estimated by the Kaplan-Meier method (time zero = start of first-line IO) and model, younger age, female sex, academic center, private insurance, higher income, lower CCI,
West coast center and AC were associated with better outcomes. When stratified by age and
compared using the log-rank test. The association of age and $ grade stage, AC was still associated with better survival (Table). Conclusions: AC utilization in stages
3 toxicity with OS was tested with a Cox proportional hazards model. IIA, IIB and IIIA NSCLC remains low, despite its association with improved survival in the elderly.
Results: Among 676 patients treated with IO, 238 (35.4%) were $70y. Future clinical trials may be needed focused on elderly patients to establish the best regimen to
There was no difference in baseline characteristics of each age group except optimize outcomes.
cancer type (P,0.01). The incidence of $ grade 3 toxicity did not differ by Stage IIA Stage IIB Stage IIIA
Overall
age (,70y: 14.5% vs. $70y: 13.5%, P=0.71). Median OS was significantly Chemo No Chemo Chemo No Chemo Chemo No Chemo
longer for adults ,70y (16.4 vs. 13.2 months, P,0.01) or those with Age
mOS
(months) mOS
HR(95%
CI) mOS mOS
HR(95%
CI) mOS mOS
HR(95%
CI)
HR(95%
CI)
$ grade 3 toxicity (18.3 vs. 14.7 months, P,0.01). When stratified by age £65 NR NR 83.0 70.5 57.3 51.9
and toxicity, patients ,70y with $ grade 3 toxicity had longer OS vs. those 66-70
(NR-NR)
NR
(71.1-NR)
NR 1.1
(73.4-87.4)
71.1
(62.8-79.8)
43.6 1.2
(54.4-63.0)
49.1
(47.0-60.0)
37.2 1.1 1.1
without $ grade 3 toxicity (P,0.01). However, there was no OS difference 71-75
(53.6-NR)
63.3
(NR-NR)
NR
(0.9-1.3)
1.2
(59.9-79.7)
55.3
(38.9-50.6)
44.8
(1.1-1.2)
1.2
(41.7-53.2)
40.9
(32.6-45.3)
39.0
(1.0-1.2)
1.1
(1.1-1.2)
1.2
among adults $70y with vs. without $ grade 3 toxicity (P=0.78). Adjusted 76-80
(52.7-NR)
51.7
(44.7-NR)
43.9
(1.0-1.5)
1.6
(47.5-60.1)
49.4
(40.0-49.3)
31.7
(1.1-1.4)
1.5
(37.3-48.7)
38.9
(34.8-44.4)
30.0
(1.0-1.3)
1.4
(1.1-1.3)
1.5
hazard ratios with an interaction term are below. Conclusions: Though the > 80
(45.3-NR)
45.1
(37.4-59.2)
34.6
(1.3-1.9)
1.9
(43.7-55.7)
37.8
(28.4-34.7)
28.8
(1.4-1.7)
1.8
(34.1-47.2)
32.2
(26.9-34.8)
26.3
(1.2-1.5)
1.7
(1.4-1.6)
1.7
incidence of $ grade 3 toxicity did not significantly differ by age, there was (30.5-NR) (28.9-41.3) (1.6-2.4) (31.7-47.2) (25.6-31.6) (1.6-1.9) (26.6-39.0) (23.3-30.0) (1.5-1.9) (1.6-1.9)

no significant OS advantage for older adults with $ grade 3 toxicity as *Adjusted for age, insurance, median income quartile, CCI, region, and race (not shown)

compared to younger adults. Caution should be used in considering a

toxicity-survival relationship in older adults.
Overall Survival: Age and Toxicity.
Covariates HR 95% CI P-value
<70 y; no ‡ Grade 3 toxicity Ref
<70 y; ‡ Grade 3 toxicity 0.53 0.35 0.81 ,0.01
‡70 y; no ‡ Grade 3 toxicity 1.19 0.96 1.48 0.12
‡70 y; ‡ Grade 3 toxicity 1.13 0.73 1.75 0.59
*adjusted for sex, race, ECOG PS, BMI, cancer type

6582 Poster Session (Board #273), Sat, 1:15 PM-4:15 PM 6583 Poster Session (Board #274), Sat, 1:15 PM-4:15 PM
Self-reported health and survival in older patients diagnosed with multiple Real-world prevalence of autoimmune disease (AD) among patients (pts)
myeloma. First Author: Nadia Azmi Nabulsi, University of Illinois at Chicago, receiving immune checkpoint inhibitors (ICI) in ASCO’s CancerLinQ database.
Department of Pharmacy Systems, Outcomes and Policy, Chicago, IL First Author: Li Chen, Concerto HealthAI, Boston, MA
Background: The strength of associations between pre-diagnosis self- Background: Although pts with AD are routinely excluded from ICI clinical
reported health (SRH) and mortality differ by medical condition, with a trials, evidence suggests they may be receiving ICI therapy once approved.
moderately strong association reported among cancer patients. Less is We sought to understand the prevalence of AD among all pts receiving ICIs in
known about the impact of SRH on survival among patients diagnosed with real world clinical care, as well as in advanced non-small cell lung cancer
multiple myeloma (MM). We aimed to evaluate pre-diagnosis SRH in relation (aNSCLC) alone, and to describe the characteristics of ICI pts with and
to survival in a cohort of older MM patients. Methods: We analyzed a pro- without evidence of AD. Methods: We conducted a retrospective, observa-
spective cohort from the Surveillance, Epidemiology, and End Results tional cohort study using statistically de-identified data from January 2011
(SEER)-Medicare Health Outcomes Survey (MHOS) database of patients 65 to November 2018 in CancerLinQ, ASCO’s real-world oncology database.
years and older diagnosed with first primary MM. Survey responses to a single Adult pts who received $ 1 dose of an ICI and had $ 2 clinical visits were
general health question (asking patients to self-report their health as ex- eligible for inclusion. A sub-analysis examining only aNSCLC pts was also
cellent, very good, good, fair, or poor) were used to determine pre-diagnosis carried out. To reduce the likelihood of capturing pts who may have been on a
SRH, grouped as high (excellent/very good/good) or low (fair/poor). We used clinical trial, pts were excluded if they received the ICI prior to its first FDA
multivariable Cox proportional hazards models to estimate adjusted hazard approval date. AD status was determined by the presence of select ICD-9/
ratios (HR) and 95% confidence intervals (CI) for associations between SRH ICD-10 codes or a medication used to treat autoimmune disease (including
and risks of all-cause and cancer-specific mortality. Results: Of 521 MM steroids) prior to ICI treatment start date. Symphony claims data were linked
patients with pre-diagnosis SRH data, the mean (SD) age at diagnosis was to CLQ via tokenization to build out cohorts. Characteristics of pts with and
76.8 (6.1) years with 60% of patients identifying as white, 18% as black, without autoimmune disease were compared using Chi-square or Fisher’s
and 32% reporting low SRH. Compared to patients reporting high SRH, exact tests. Results: Prevalence of AD was 23% (538/2425 pts) in the
patients reporting low SRH were older, had lower education levels, more aNSCLC population and 27% (3407/12712 pts) in the all ICI patient
comorbidities, and lower Veterans-RAND 12 physical health and mental population. Median age did not differ between AD pts and those with no
health component summary scores. In multivariable analyses, MM patients evidence of AD (All ICI: 67.6 v 67.3 years; aNSCLC: 68.5 v 67.9). AD pts
with low SRH had a 28% increased risk of all-cause mortality (HR = 1.28, were more likely to be female (All ICI: 46% v 40%, p , 0.001; aNSCLC: 55%
95% CI = 1.00, 1.64) and a non-statistically significant 19% increased risk v 44%, p , 0.001). Among all ICI pts, AD pts were less likely to be Stage IV
of cancer-specific mortality (HR = 1.19, 95% CI = 0.87, 1.61) compared to (62% v 65%) or to have melanoma (4.6% versus 8.7%) compared to pts with
MM patients reporting high SRH. Conclusions: Our findings suggest that no evidence of AD. The most common ADs among all ICI and aNSCLC
lower SRH is highly prevalent among MM patients prior to diagnosis and is patients were glucocorticoid deficiency (6.3% and 3.9%), rheumatoid ar-
associated with modestly increased all-cause mortality. At a minimum, low thritis (4.2% and 5.8%), and sacroiliitis (2.7% and 3.9%), respectively.
SRH deserves clinical attention to determine how older MM patients’ quality Conclusions: This analysis of real-world data finds that a large proportion of
of life may be compromised. The mechanism by which SRH affects mortality pts receiving ICI may have pre-existing AD. Further examination is warranted
in MM should be further assessed and efforts should be made to identify to examine how AD status may impact outcomes.
whether any of the underlying mechanisms linking SRH and mortality in MM
are mutable.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
388s Health Services Research, Clinical Informatics, and Quality of Care

6584 Poster Session (Board #275), Sat, 1:15 PM-4:15 PM 6585 Poster Session (Board #276), Sat, 1:15 PM-4:15 PM
Caveat medicus: Harrowing experiences of clinicians who identify and publish Can high-volume surgeons achieve optimal outcomes at low-volume hospitals?
near-fatal oncology associated adverse drug reactions. First Author: Ashley Caitlin Implications for the Leapfrog Initiative and regionalization of high-risk surgical
Godwin, University of South Carolina College of Pharmacy, Columbia, SC oncology procedures. First Author: Christopher Thomas Aquina, University of
Rochester Medical Center, Rochester, NY
Background: Less than 1-10% of adverse drug reactions are reported. It is
even more rare to find clinicians who choose to publish these reports in the Background: The Leapfrog group recently released surgeon and hospital procedure
literature. We identified clinicians who had treated a patient for an oncology- volume standards for several surgical oncology procedures. This study investigated trends
associated serious adverse drug reaction and published these findings. This in volume and whether high-volume surgeons at low-volume hospitals achieve equivalent
is the first study that has investigated personal experiences of clinicians outcomes to high-volume surgeons at high-volume hospitals. Methods: New York’s
Statewide Planning and Research Cooperative System was queried for esophagectomy,
choosing to publish information about serious oncology-associated drug lung resection, pancreatectomy, and proctectomy for cancer from 2004-2015. Mixed-
reactions they see in their patients. Methods: Clinicians treating individual effects analyses assessed the association among Leapfrog surgeon/hospital volume
patients who developed oncology-associated serious adverse drug events standards and 90-day mortality. Results: Among 55,528 cases, high-volume surgeons
were asked to participate. Inclusion criteria included having index patient performed 64.7% of cases (esophagectomy = 52%; lung resection = 75.6%; pancre-
information, reporting case series, and being collaborative with investigators atectomy = 56.7%; proctectomy = 53%), and high-volume hospitals performed 59.5% of
from two National Institutes of Health funded pharmacovigilance networks. cases (esophagectomy = 55.5%; lung resection = 58.3%; pancreatectomy = 63.4%;
Thirty-minute interviews addressed feedback from pharmaceutical manu- proctectomy = 61%). After risk-adjustment, high-volume surgeons at high-volume
facturers, FDA personnel, and academic leadership, and recommendations hospitals had lower odds of 90-day mortality compared to high-volume surgeons at
for improving pharmacovigilance. Responses were analyzed using constant low-volume hospitals for each organ system except for pancreas. Despite trends toward
regionalization, between 2012-2015, there were large differences in the number of
comparative methods of qualitative analysis. Results: 18 clinicians met in-
hospitals and median annual case number between high-volume and low-volume centers
clusion criteria and 14 interviewees are included. Toxicities included central for esophagectomy (8 vs. 56 hospitals; 31.5 vs. 3 cases), lung resection (22 vs. 89
nervous system infections, arterial/venous thromboembolism, gastrointesti- hospitals; 69.5 vs. 7 cases), pancreatectomy (15 vs. 56 hospitals; 36 vs. 3 cases), and
nal toxicity, cardiac arrhythmias, and cancer development/progression. These proctectomy (38 vs. 117 hospitals; 28 vs. 3 cases). Conclusions: This study supports the
investigations were frequently followed by label warnings and/or convening of Leapfrog initiative for performance of high-risk surgical oncology procedures by high-
Food and Drug Administration (FDA) Advisory Committee reviews of safety volume surgeons at high-volume hospitals. However, it remains unclear whether full
findings. Five studies were disseminated in four high-impact factor medical regionalization to high-volume centers is feasible.
journals (JAMA, Lancet, Annals of Internal Medicine, and New England Journal Adjusted 90-Day Mortality
of Medicine). Six clinicians received feedback characterized as supportive from Unadjusted 90-Day Mortality Rate [OR (95% CI)]

academic leaders, while four clinicians received feedback characterized as Esophageal Cancer Resection* (N = 4,177)
HVS/HVH 4.8% 0.37 (0.22-0.63)
negative. Responses from pharmaceutical manufacturers were characterized HVS/LVH 9.6% Reference
as negative by 12 clinicians. Responses from FDA employees were characterized LVS/HVH
LVS/LVH
5.4%
11%
0.38 (0.21-0.76)
0.68 (0.40-1.14)
as negative by six clinicians. Three clinicians recommended that pharmacovi- Lung Cancer Resection† (N = 27,809)
HVS/HVH 2.9% 0.74 (0.58-0.92)
gilance should include simplified clinician reporting systems. Conclusions: Our HVS/LVH 4.7% Reference
LVS/HVH 4.1% 0.82 (0.57-1.18)
study finds that clinicians who published reports of serious oncology-associated LVS/LVH 5.3% 1.11 (0.87-1.37)
drug reactions experienced negative feedback from pharmaceutical manufac- Pancreatic Cancer Resection‡ (N = 6,531)
HVS/HVH 3.9% 1.03 (0.54-2.05)
turers. Feedback from FDA employees and academic clinicians were viewed as HVS/LVH 3.9% Reference
LVS/HVH 7.1% 1.62 (0.75-3.30)
supportive by some and negative by others. Most clinicians recommended that LVS/LVH 9.1% 1.56 (0.77-2.92)
future pharmacovigilance involve big data analyses. Rectal Cancer Resection£ (N = 17,011)
HVS/HVH 1.9% 0.55 (0.39-0.84)
HVS/LVH 4.0% Reference
LVS/HVH 3.9% 0.71 (0.49-1.07)
LVS/LVH 4.7% 0.83 (0.61-1.19)

6586 Poster Session (Board #277), Sat, 1:15 PM-4:15 PM 6587 Poster Session (Board #278), Sat, 1:15 PM-4:15 PM
Impact of cardiovascular comorbidities on mortality in patients admitted for Prospective study of pain outcomes associated with contralateral prophy-
neutropenic fever in 2016. First Author: Suheil Albert Atallah-Yunes, lactic mastectomy in women with nonhereditary breast cancer. First Author:
University of Massachusetts -Baystate, Springfield, MA Demetria Joy Smith-Graziani, The University of Texas MD Anderson Cancer
Center, Houston, TX
Background: Neutropenic fever (NF) remains one of the most common causes for
hospitalization and mortality in oncology patients. Concomitant cardiovascular Background: Women with nonhereditary breast cancer are increasingly un-
disease in patients with cancer is not uncommon. There is limited data on the impact dergoing contralateral prophylactic mastectomy (CPM). We examined pain
of cardiovascular (CVS) comorbidities on mortality in cancer patients with NF. severity and the impact of pain on the lives of women who underwent CPM
Methods: This is a retrospective cohort study using the 2016 National Inpatient compared to those who did not. We also examined the associations between
Sample database (NIS) of adults ( . 18 years) admitted for NF based on the ICD-10 age, race/ethnicity, reconstruction and pain outcomes. Methods: Between
code. Mortality was the primary outcome. Multivariate linear regression adjusted for 2012 and 2015, we recruited women with newly diagnosed nonhereditary
potential confounder of age, sex, race, Charlson comorbidity index and all the CVS
breast cancer who were planned for surgery. We assessed pain with the Brief
comorbidities of the study including atrial fibrillation (AF), heart failure with pre-
served ejection fraction (HFpEF), heart failure with reduced ejection fraction Pain Inventory at initial surgical consultation and at 1, 6, 12, and 18 months
(HFrEF), coronary artery disease (CAD), peripheral vascular disease (PVD), hyper- after surgery. The repeated measures model was used to assess the association
tension (HTN), history of smoking, history of cerebrovascular accident (CVA) or TIA between pain severity or interference and CPM status over different time points
and dyslipidemia. STATA 15 was used for analysis. Results: We identified 31,310 adjusting for other covariates. Results: Of 288 women enrolled (mean age 56
patients (mean age 44.6) (49.6% females) admitted with NF, among which 250 years, 58% non-Hispanic White, 17% non-Hispanic Black), 50 underwent
died during same admission. On multivariate linear regression there was a significant CPM, 163 had unilateral mastectomy, and 75 had breast conserving surgery.
increase in adjusted all-cause mortality in patients with AF (OR: 2.39; 95%-CI 1.06- Mean pain severity was higher at 1 month (2.78 vs 1.9, p = .01) and 6 months
5.40, P = 0.035) and HFpEF (OR: 4.30; 95%-CI 1.08- 17.17, P = 0.039). There (2.79 vs 1.96, p = .03) after surgery in women with CPM versus those without.
was no significant increase in mortality in patients with HFrEF, dyslipidemia, HTN, In the multivariable repeated measures model adjusted for time, age, race/
PVD, CAD, history of CVA/TIA and smoking. Conclusions: Patients with NF and ethnicity and reconstruction status, there was a significant interaction between
concomitant history of AF or HFpEF have an increased risk of mortality during time and CPM for pain severity (p , .01) but not interference (p = .13). This
hospitalization. Inflammation is emerging as a key player in AF pathogenesis. This suggests that CPM patients had higher pain severity in the first 6 months after
may explain why AF appears to correlate with mortality, as those with more severe surgery, but their pain scores decreased by 12 months becoming similar to
presentations are more likely to have a heightened state of inflammation. Patients women without CPM. Black women had higher pain severity (mean difference
with NF are more likely to receive fluids in the setting of infectious complications 1.35, standard error [SE] 0.35; p , .01) and interference (mean difference
which could explain the increased mortality in CHF patients with NF. Identifying risk
0.91, SE 0.32; p , .01) compared to White women with or without CPM. There
factors for increased mortality in patients with NF is important for risk stratification
and in guiding clinicians in the management of this delicate population. was no association between age or reconstruction status and pain severity or
interference. Conclusions: Pain severity in patients undergoing CPM is highest
CVS Comorbidity Odds ratio (95% CI) for mortality P value during the first 6 months after surgery. Women considering CPM should be
Afib 2.39 (1.06 to 5.40) 0.035 counseled about this potential outcome. Race/ethnic disparities exist in pain
HFpEF 4.30 (1.08 to 17.17) 0.039 management, pain perceptions and communication of pain. Black women
HFrEF 2.63 (0.79 to 8.72) 0.112
Dyslipidemia 0.59 (0.28 to 1.25) 0.174 undergoing breast surgery report worse pain outcomes than White women
Smoking 0.75 (0.35 to 1.59) 0.448 regardless of CPM status.
HTN 1.67 (0.90 to 3.11) 0.103
CAD 1.22 (0.51 to 2.96) 0.648
Hx of CVA/TIA 0.88 (0.31 to 4.93) 0.756
PVD 0.54 (0.06 to 5.37) 0.602

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Health Services Research, Clinical Informatics, and Quality of Care 389s

6588 Poster Session (Board #279), Sat, 1:15 PM-4:15 PM 6589 Poster Session (Board #280), Sat, 1:15 PM-4:15 PM
Analysis of racial distribution amongst patients in phase III cancer clinical Real-world effectiveness of approved anticancer agents among Medicare
trials. First Author: Swathi Gopishetty, Navicent Health, Macon, GA beneficiaries. First Author: Angela Green, Memorial Sloan Kettering Cancer
Center, New York, NY
Background: Minority races are often under-represented in cancer clinical
trials as enrollment often occurs in large centers. Racial diversity may vary by Background: The overall survival (OS) among elderly patients (pts) treated with
geographical location and socio-economically backward areas may have a very novel anticancer agents in the real world may be inferior to the OS reported in
different racial mix. This study explores the representation of different races in pivotal trials for drug approval. Pts $ 65 yrs old constitute 60% of cancer pts,
phase 3 clinical trials conducted in the past 10 years. Methods: Details about yet only 40% of cancer clinical trial participants. Elderly pts have greater
adult patients involved in phase 3 trials was extracted from the clinical comorbidities and experience a higher risk of toxicity from cancer drugs. Using
trials.gov for 3 common solid organs and 3 hematological malignancies the Surveillance Epidemiology and End Results-Medicare database (SEER-
[breast, colon, lung, diffuse large B-cell lymphoma (DLBCL), acute myeloid Medicare), we compared the OS among elderly pts treated with FDA-approved
leukemia (AML) and acute lymphoblastic leukemia (ALL)]. The racial dis- cancer drugs for advanced solid tumors to the OS reported in the clinical trial.
tribution of the patient population in these trials was analyzed. Methods: We identified cancer drugs FDA-approved for metastatic solid tu-
Results: African American and Asian patients are under-represented in all mors between 1/1/08-12/31/12. In a retrospective analysis, for each indi-
phase 3 cancer clinical trials. The table shows the average racial distribution cation we identified pts in SEER-Medicare meeting disease eligibility criteria
in clinical trials for each organ specific malignancy. Conclusions: Most (stage, histology, prior therapies) in the trial associated with approval. Pts were
phase 3 clinical trials except for lung cancer, predominantly consisted of diagnosed with cancer from 2010-2013 with follow-up through 2014.
Caucasian patients. Applying the results of these trails to patients of other Treatment (tx) was determined from national drug codes in Medicare Part B for
races should be done with caution. This study highlights the disparity of race in intravenous (IV) drugs and Part D for oral drugs. Indications were included if
patients enrolled in clinical trials when compared to diverse and different $ 30 pts receiving tx met eligibility. Kaplan-Meier methods were used to
populations that are encountered in practice. calculate median OS and cancer-specific survival (CSS) in Medicare pts.
Median duration of therapy (DOT) was estimated from date of the first through
% of Afri- completion of the last prescription claim for oral drugs and number of cycles for
Number Number of Phase % of can % of % of
S. of Phase 3 trials with Race Asian American White Other Total
IV drugs. Results: OS data were available and sample size parameters met for
No Malignancy 3 trials specified Pts Pts Pts pts No pts 14 drug indications. The median OS among SEER-Medicare pts was shorter
than the reported trial OS of tx arms for 13 of 14 drug indications (median
1 Colon 20 9 21.15 2.6 74.25 2 5537 difference -7.6 mos, range +3.4 to -28.7 mos). CSS was similar to OS among
cancer
2 Lung 73 51 26.06 30.63 39.66 3.65 23594
Medicare pts. Median DOT among SEER-Medicare pts was shorter than the
cancer reported trial DOT of tx arms for 13 of 14 indications (median difference -2.9
3 Breast 74 39 17.41 3.55 73.76 5.27 24170 mos, range 0 to -8.9 mos). Conclusions: The OS and DOT among SEER-
cancer Medicare pts treated with FDA-approved cancer drugs was inferior to the
4 DLBCL 10 3 16.59 1.46 77.73 4.22 687 reported OS in pivotal clinical trials for nearly all indications analyzed. The
5 AML 16 7 2.28 2.34 92.86 2.52 4479 shorter DOT among Medicare pts may explain survival differences. Trials
6 ALL 8 6 5.88 6.7 77.53 9.9 1940 leading to regulatory approval may not be generalizable to cancer pts $ 65 yrs.

6590 Poster Session (Board #281), Sat, 1:15 PM-4:15 PM 6591 Poster Session (Board #282), Sat, 1:15 PM-4:15 PM
Reporting of patient reported outcome (PRO) in clinical trials: A systematic Hormone replacement therapy (HRT) and risk of distant recurrence in newly
review of clinical trials. First Author: Liat Vidal-Fisher, Syneos Health, Tel diagnosed ER+, node-negative (N-) breast-cancer (BC) patients: A retrospec-
Aviv, Israel tive population-based matched-cohort study. First Author: Ariel Hammerman,
Clalit Health Services Headquarters, Tel-Aviv, Israel
Background: Clinically relevant outcomes as improvements in overall survival
(OS) or quality of life (QOL) should guide decision-making. The FDA encourages Background: Observational studies have shown an increased risk of BC with
the implementation of patient-centric measures in clinical trials. Over the past use of HRT. However, data on the prognosis of BC that develop in HRT users are
decade a growing number of protocols have included PROs in their outcome inconsistent. The association between HRT use and results of the 21-gene
measures. We aimed to evaluate the frequency at which PRO measures, in- Recurrence Score (RS) assay (Oncotype DX, Genomic Health Inc.) has not
corporated into clinical trials, are made publicly available, when trial results are been investigated. We aimed to analyze this association, and examine the
published. Methods: We searched Citeline Trialtrove database, a registry of actual rate of distant recurrence or death in this population. Methods: Clalit
clinical trials, for randomized phase 3 trials of patients with non-small cell lung Health Services (CHS) is the largest health maintenance organization (HMO) in
cancer (NSCLC), recruiting during 2006-2016, with PRO listed. We excluded Israel. We identified all CHS newly diagnosed ER+, N- breast-cancer patients,
trials evaluating supportive treatment, and those unpublished. We identified aged 45-60 that performed a RS assay between 01/2006-12/2012 and that
publications associated with the trial, and extracted various parameters, in- were treated for at least three months with HRT during the eight years before
cluding whether or not PRO outcome was published. Results: Of 158 NSCLC BC diagnosis. A 1:4 matched-cohort analysis was performed, with matching
trials identified, 99 listed at least 1 PRO. 24 trials were excluded (supportive made according to age and year of BC diagnosis. Clinical and demographic
care, unpublished). The commonly used scales were EORTC-QLQ C30, QLQ- data were extracted from the CHS centralized registry for all patients. RS assay
LC13, LCSS, FACT-L, and EQ-5D. Study sponsor was industry in 45 trials, and scores was grouped according to the TAILORX categorization and distribution
industry-academic in 12 trials. Of 75 trials analyzed, 41 (55%) published the was compared using Chi-square test. Kaplan-Meier analysis with log-rank test
results of the PRO endpoint. Only 37% of the 75 included a comprehensive was performed in order to compare time to a combined outcome of distant-
report of PRO, and many publications referenced the PRO briefly. Of 41 trials recurrence and mortality. Results: A cohort of 259 HRT-treated patients was
that reported PRO, only 21 (51%) provided information about missing PRO identified and matched with 1001 controls, not treated with HRT. The pro-
data. 59 trials were published as full-text, of which 40 (68%) reported the PRO, portions of low-risk patients (RS 0-25) and high-risk patients (RS 26-100)
and 53% reported the PRO more than 6 months after the initial full publication. were 76.8% and 23.2%, respectively, within HRT-treated patients, and
Of the trials that showed PFS benefit, with no OS benefit, 22/34 (65%) 80.4% and 19.6% within controls. Chi square test was not found significant
published the results of the PRO endpoints. Conclusions: Despite a growing (x2= 1.634, p = 0.201). The mean follow-up time was 148.4 months for the
emphasis on QOL and the inclusion of PROs in oncology clinical trials, and cases and 146.9 months for controls, with log-rank test not showing a sig-
despite patient and healthcare provider efforts to record PRO data, a significant nificant difference between groups. Conclusions: These data did not show
number of NSCLC randomized trials do not report the PRO. The collection of significant association between HRT use and higher RS assay scores, and also
PRO data should result in routine, timely and appropriate reporting as part of the did not find an association between HRT use and actual distant recurrence or
trial outcome publication, to allow for a thorough assessment of investigational death. Although the proportion of patients with high risk RS appeared to be
clinically relevant treatment effect. slightly higher within HRT treated patients, this difference had not reached
significance and further studies are required.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
390s Health Services Research, Clinical Informatics, and Quality of Care

6592 Poster Session (Board #283), Sat, 1:15 PM-4:15 PM 6593 Poster Session (Board #284), Sat, 1:15 PM-4:15 PM
Association of sex, age and ECOG performance status with cancer immuno- Clinical utility and reimbursement for expanded genomic panel testing in adult
therapy efficacy in randomized controlled trials. First Author: Fang Yang, The oncology. First Author: Susan Jean Hsiao, New York Presbyterian Hospital/
Comprehensive Cancer Centre of Drum Tower Hospital, Medical School of Columbia University Medical Center, New York, NY
Nanjing University & Clinical Cancer Institute of Nanjing University, Nanjing,
Background: The routine use of large next generation sequencing (NGS) cancer
China
panels is required to identify the increasing number of, but often uncommon
Background: Sex, age and ECOG performance status (PS) may affect immune actionable alterations present across multiple tumor histologies to guide
response and the efficacy of cancer immunotherapy with immune check- therapy. Inconsistent coverage and variable payment is hindering adoption of
point inhibitors (ICI). We did a meta-analysis to assess the potential sex, age these tests into clinical practice. A review of clinical utility, coverage and re-
and ECOG PS differences of immunotherapy efficacy in advanced cancer. imbursement was conducted in a cohort of adult oncology patients who received
Methods: PubMed was searched up to January 15, 2019 for randomized expanded genomic panel testing as part of their clinical care. Methods: The
controlled trials (RCT) comparing overall survival (OS) in patients with ad- Columbia Combined Cancer Panel (CCCP), a 467 gene panel designed to detect
vanced cancer treated with ICI immunotherapy vs control therapy (without ICI). single nucleotide variants, indels, and copy number variations in solid and liquid
For sex difference analysis, pooled hazard ratio (HR) of death for men and tumors was performed in a CLIA-certified laboratory at Columbia University
women was calculated separately, and the heterogeneity between the two Irving Medical Center. Clinical utility categories included: immediate change
estimates was assessed using an interaction test by pooling study-specific in management; informed future treatment options; provided diagnostic/
interaction HR. Age ( , 65 vs $65) and ECOG PS (0 vs $1) difference was prognostic information; and other impact. Claims were submitted between
analyzed similarly. Subgroup analysis by cancer type and line of therapy 1/1/17 and 4/30/18. Carriers were categorized into commercial, managed-
(frontline vs subsequent) was explored. All analyses were done in Compre- government, and government plans. Results: 300 tumors underwent NGS.
hensive Meta Analysis (v2) with random effects models. Results: Thirty phase Reimbursement data were available for 258 cases. 57% of testing was per-
2/3 RCTs involving 17,728 patients were included. An OS benefit of im- formed for a treatment-resistant, recurrent, or high stage cancer, or for a cancer
munotherapy was found for both men (HR 0.75, 95% confidence interval [CI] of rare/mixed histology (21%). Findings were clinically actionable in 183 cases
0.69-0.81, P , 0.01) and women (HR 0.79, 95% CI 0.69-0.90, P , 0.01); (61%). Results led to an immediate change in management (n = 6, 2%),
for both younger ( , 65; HR 0.75, 95% CI 0.68-0.83, P , 0.01) and older informed future treatment options (n = 140, 47%), and provided diagnostic/
($65; HR 0.76, 95% CI 0.69-0.83, P , 0.01) patients; and for both ECOG PS prognostic information (n = 29, 10%). Only 57 tests (22%) received coverage.
0 (HR 0.78, 95% CI 0.69-0.89, P , 0.01) and PS $1 (HR 0.77, 95% CI In 59% of denials (118/201), a clinically-actionable result was found. Com-
0.71-0.84, P , 0.01) patients. No significant difference of relative benefit mercial plans reimbursed 29/119 tests (24%) and managed-government
from immunotherapy over control therapy was found in patients with different plans reimbursed 28/54 tests (52%). Government plans provided no cover-
sex (P = 0.283), age (P = 0.906) or ECOG PS (P = 0.783). In melanoma RCTs, age for 85 tests. On average, insurers reimbursed 10% of the total CCCP service
compared with women (HR 0.77, 95% CI 0.64-0.94, P , 0.01), men (HR charges: 12.5% for commercial and 22% for managed-government plans.
0.56, 95% CI 0.42-0.76, P , 0.01) had more OS benefit from immuno- Conclusions: Expanded genomic testing identified clinically-impactful alter-
therapy (P = 0.037). No significant difference was found in other subgroup ations in 61% of cases. Limited coverage and low reimbursement remain a
analyses by cancer types or line of therapy. Conclusions: Overall we found no barrier and broader reimbursement policies are needed to adopt expanded
evidence of association of sex, age, or ECOG PS with cancer immunother- genomic testing that benefits patients into clinical practice.
apy efficacy. However, in melanoma, men might benefit more from immu-
notherapy than women.

6594 Poster Session (Board #285), Sat, 1:15 PM-4:15 PM 6595 Poster Session (Board #286), Sat, 1:15 PM-4:15 PM
Quality of life (QOL) in patients undergoing CAR-T therapy versus stem cell Incidence and outcome of interval breast cancer among women participating
transplant (SCT). First Author: Surbhi Sidana, Mayo Clinic, Rochester, MN in the provincial population based screening program in Manitoba, Canada.
First Author: Saroj Niraula, CancerCare Manitoba and Univ of Manitoba,
Background: Given the significant short-term adverse effects of CAR-T cell therapy,
it is important to evaluate its impact on QOL of patients in addition to efficacy, Winnipeg, MB, Canada
compared with established forms of cellular therapy like SCT. Methods: QOL was Background: The province of Manitoba, Canada has an organized population
evaluated prospectively in patients undergoing CAR-T therapy, autoSCT & alloSCT based biennial mammographic screening program. Here we report outcomes
for hematologic malignancies. QOL was assessed with FACT-G at baseline, 2 weeks of women diagnosed with Interval Cancers (IC), defined as cancer diagnosed
and monthly for 6 months thereafter. Functional well-being (FWB), physical within 24 months of a normal screening mammogram and before the next
WB (PWB) emotional WB (EWB) & social WB (SWB) and change over time were
screening mammogram, compared to Screen Detected (SD) cancers.
compared across groups. Results: 45 patients were recruited (CAR-T: 10; Auto SCT:
Methods: The Manitoba Cancer Registry was used to obtain data about tumor
22; Allo SCT: 13) with follow up for 2 weeks & 1 month available for 23 &15 patients,
respectively (Table). There was no statistically significant difference in baseline total and host characteristics and cause-specific mortality for women 52 to 64
QOL scores (p=0.13), though scores were lower in the alloSCT group (85,84,68). years of age diagnosed with invasive breast cancer from January 2004 to
EWB &FWB were numerically higher in the CAR-T group, followed by autoSCT group. December 2010. Lead time bias in SD cancers was adjusted based on
At 2 weeks, overall QOL decreased by only 2 points in CAR-T group vs. 22 & 18 points Duffy’s correction factor. Competing risk analysis was used to examine the
in auto & alloSCT groups (p=0.09). Change in PWB vs. baseline was less pronounced risk of death by cancer detection method. To examine the relationship between
in the CAR-T group (-1, -9, -13, p=0.03). At 1 month, overall QOL was 6 points lower breast cancer detection type and personal and tumour characteristics, we
than baseline in CAR-T group vs. 3 and 14 points lower in auto & alloSCT groups, performed multinomial logistic regression analysis with age, area-level income
respectively (p=0.34). Importantly, PWB had at least returned to baseline in the quintile, tumour grade, hormone receptor, and HER2 receptor as independent
CAR-T group. Conclusions: Preliminary data show that patients undergoing CAR- variables. Results: There were 5,884 women diagnosed with invasive breast
T cell therapy do not experience a more significant decline in QOL compared with cancer during the study period of which 1,338 were SD, 362 were IC, and the
auto & allo SCT, and may experience fewer physical side effects in the short-term. remainder were diagnosed outside the screening program or were non-
Accrual & follow-up are ongoing. compliant to screening. IC were more likely than SD cancers to be high
CAR-T Auto SCT Allo SCT grade [Odds Ratio (OR) 3.8, 95% Confidence Interval (CI): 2.1-6.8], and ER
N=10 N=22 N=13 P-value negative [OR 1.7, 95% CI: 1.02-3.12]. At data cut-off date of June 30, 2012,
Median age, years 62.5 62.2 61.4 0.9 risk of death from breast cancer was significantly higher for IC compared to SD
FACT-G Baseline Median
Physical WB (0-28) 19 22 20 0.70 cancers [Hazard Ratio (HR) 4.18, 95% CI: 1.97-8.87)], for sojorn time (period
Social WB (0-28) 24 24.5 22 0.12 when tumour is asymptomatic but screen detectable) of 2 years adjusting for
Emotional WB (0-24) 20 19 16 0.12
Functional WB (0-28) 20.5 18 14 0.09 area-level average income quintile and age. Sensitivity analyses with sojourn
Total FACTG (0-108) 85 84 68 0.13 times of 1, 3, and 4 years showed similar results. Risk of non-breast cancer
QOL Change Week 2 vs. Baseline N=7 N=10 N=6 death was not increased with IC compared to SD cancers (HR 1.33, 95% CI:
Physical WB
Social WB
-1
0
-9
0.5
-13
0
0.03
0.72
0.43-4.15). Conclusions: Among women who participated in a systematic
Emotional WB 1 1 1 0.88 population-based screening program, IC occurred frequently; breast cancer-
Functional WB -2 -8 -6 0.40 related death for IC was 4-times that of SD cancers. These results highlight the
Total FACTG -2 -22 -18 0.09
discordance between the principles underpinning population based breast
QOL Change Month-1 vs. Baseline N=5 N=5 N=5
Physical WB 1 -2 -4 0.73
cancer screening and natural history of these more lethal breast cancers.
Social WB -1 0 -2 0.41
Emotional WB 2 2 1 0.93
Functional WB -9 -1 -8 0.21
Total FACTG -6 -3 -14 0.34

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Health Services Research, Clinical Informatics, and Quality of Care 391s

6596 Poster Session (Board #287), Sat, 1:15 PM-4:15 PM 6597 Poster Session (Board #288), Sat, 1:15 PM-4:15 PM
A living systematic review of immune checkpoint inhibitors in cancer patients: Association between age, risk of severe (Grade 3-4) immune-related adverse
A novel platform for evidence synthesis in oncology. First Author: Irbaz Bin events (sirAE), and mortality in patients receiving immune checkpoint in-
Riaz, Mayo Clinic, Rochester, MN hibitors (ICI). First Author: Asrar Alahmadi, University Hospitals Seidman
Cancer Center, Cleveland, OH
Background: Several previous systematic reviews and meta-analyses have attempted to
summarize toxicity of Immune checkpoint inhibitors (ICIs). However, very soon after each Background: ICI are used in the treatment of advanced malignancies with on-
one of these reviews has been published, it became outdated. ICIs are currently used in target adverse events of non-tumor inflammation. Many studies have shown
14 different cancers and data is rapidly evolving from new clinical trials. A living Sys- conflicting safety results with regard to older vs younger adults where 65 - 70
tematic review, which is defined as a systematic review that is continually updated to
years of age has been used as the discrete cut-off variable. We sought to
incorporate relevant new evidence as it becomes available, is necessary in this situations.
Therefore, we performed an updated systematic review and a meta-analysis which will
investigate the incidence and the association between age and sirAE using age
serve as a foundation of a living Systematic review. Methods: MEDLINE, EMBASE and as a continuous variable at a large tertiary cancer center. Methods: Under IRB
Cochrane were searched to identify phase 2 and 3 RCTs of PD-1/PD-L1 ICIs. Included approval, our ICI outcomes database was queried for those who were hospi-
studies compared either immunotherapy alone or combination with existing standard of talized and received an immunosuppressant. Charts were individually reviewed
care treatment and reported data for AE’s of interest. DerSimonian-Laird random effects to identify hospital admissions due to a sirAE (a grade 3 or 4 AE per CTCAE
Meta-Analysis was performed to derive pooled odds Ratio (OR) estimates for AE’s of v4.0) and all-cause mortality at 12 months post ICI start. Non-linear analyses
interest. An infrastructure of a living systematic review is being developed and it includes using Cox regression models with penalized smoothed splines were performed
monthly literature searches, cumulative meta-analysis and an online reporting platform. to explore association between age and sirAE. Results: There were 6.3% of
Results: We screened 6746 studies and 31 phase 3 and 2 phase 2 RCTs (n = 21,421) 1043 patients who had a sirAE, and a total of 83 sirAE events. Mean age was
were included in the analysis. 22 RCTs used PD-1/PD-L1 ICIs as a single agent and 11 646 13 years.ICI included anti-PD-1 (77.8%), anti-CTLA-4 (18.1%), and anti
as a combination therapy. Selected toxicity estimates are summarized in a table. PD-L1 (4.1%). Pts with sirAE had a thirty day, and one year mortality of 12%
Conclusions: The meta-analysis updates previously published toxicity estimates and
provides additional information about the risk of toxicities in single versus combination
and 49%, respectively. These events included: colitis (30.1%), pneumonitis
regimens. We have initiated the first living systematic review in oncology that will be (16.9%), hepatitis (10.8%), hypophysitis (6%), and thyroiditis (6%). There
continuously updated, incorporating relevant new evidence as it becomes available, and were 5 neurologic and 4 myocarditis/myositis sirAEs. The 1-year cumulative
will provide accurate and up to date toxicity estimates to support clinical decision making. incidences of sirAE and all-cause mortality were 8.4%, and 48%, respectively.
Spline analysis showed a U-shaped association between age and hazard of
Disease Trials(n) Odds Ratio
sirAE (P = 0.03). Every 10 years above age 60 was associated with increased
SINGLE AGENT
Hypothyroidism 16 4.92 (3.71-6.43) sirAE (HR 1.65 [1.14-2.40], P = 0.008), while every 10 years below 60 was
Thyroiditis
Diabetes mellitus
7
5
4.54 (1.79-11.53)
3.79 (1.61-12.38)
associated with increased sirAE (HR 1.50 [0.99-2.27], P = 0.054). However,
Hepatitis 8 2.72 (1.13-4.55) age and mortality showed a linear association (P = 0.003). Conclusions: We
Nephritis 5 1.7 (0.52-5.61)
Pneumonitis 14 1.39 (1.09-1.78) observed a curvilinear U-shaped association between age and the risk of sirAE,
Colitis
Adrenal Insufficiency
14
4
1.26 (0.89-1.79)
1.22 (0.5-2.98)
with minimum risk at age of 60, compared with a linear relationship between
Pancreatitis 4 1.03 (0.33-3.17) age and mortality. Further studies are needed to understand this relation-
Fatigue 24 0.75 (0.69-0.82)
Diarrhea 24 0.67 (0.61-0.73) ship and its impact on outcomes, clinical care, and underlying host immune
COMBINATION TREATMENT
Disease Odds Ratio context.
Hepatitis 3 6.10 (1.08-34.4)
Hyperthyroidism 4 4.27 (2.27-8.06)
Pneumonitis 4 3.12 (1.79-5.42)
Hypothyroidism 9 1.46 (1.19-1.79)
Colitis 3 1.29 (0.83-2)
Fatigue 10 0.91 (0.82-1.01)
Diarrhea 10 0.88 (0.79-0.99)

6598 Poster Session (Board #289), Sat, 1:15 PM-4:15 PM 6599 Poster Session (Board #290), Sat, 1:15 PM-4:15 PM
Impact of timing of lung resection on survival for clinical stage I and II lung Online advertising and marketing claims by providers of proton beam therapy:
cancer. First Author: Isabel M Emmerick, University of Massachusetts, Are they guideline based? First Author: Mark Thomas Corkum, Department of
Worcester, MA Radiation Oncology, London Health Sciences Centre, London, ON, Canada
Background: Lung cancer has the highest mortality among the leading cancers Background: Proton beam therapy (PBT) is a radiotherapy platform that
in the U.S. Surgical resection is considered as the most effective treatment for purports an improved therapeutic ratio by way of a rapid radiation dose fall-
lung cancer in early stages, providing greater long-term survival. Clinical off. Despite this technology being hindered by significant capital and patient
guidelines on delays in resection of early-stage lung cancer do not exist. This work costs, the number of centres offering PBT is increasing exponentially.
aims to assess whether increasing time between diagnosis/first doctor visit and Consensus guidelines support PBT use in a limited number of disease sites or
surgery for early stage non-small cell lung cancer (NSCLC) is associated with on clinical trials. As patients frequently obtain information about PBT from
poorer survival. Methods: We identified a retrospective cohort of incident lung hospital or cancer centre websites, the purpose of this study was to evaluate
cancer cases who had surgical treatment for lung cancer at our institution be- direct to consumer advertising (DTCA) content and claims made by proton
tween January 2009 and December 2017, and no prior radiation or chemo- therapy centre (PTC) websites. Methods: English PTC websites worldwide
therapy. We assessed overall survival (OS) and predictors included a) time from were identified using the Particle Therapy Co-Operative Group website. Data
first contact to surgery; and b) time from diagnosis to surgery. The association abstraction of website content was performed independently by two in-
between time from diagnosis and time from first contact to surgery, and survival vestigators. Eight international guidelines were consulted to determine in-
for patients with early stage NSCLC was assessed using multivariable Cox dications for PBT. Univariate and multivariate logistic regression models
proportional hazard. We investigated four cut-off points: surgery within 15 days, were used to identify website characteristics that were associated with a
30 days, 60 days and 90 days. We controlled for sociodemographic charac- higher likelihood to make non-evidence-based claims of PBT such as im-
teristics as well as clinical outcomes. Results: Our cohort comprised 491 pa- proved disease control or cure. Results: From the 48 PTCs with 46 English
tients. The age average was 66.9 years, 61% female, 94.7% white, and 9% websites, most (58%) did not provide any references for claims made re-
never smoked. Clinical Stage 1A and 1B corresponded to 86.5% of patients garding PBT. These included: improved disease control or cure (61%), fewer
while in pathological stage it was 76%. The 5-year overall survival was 56.0%.
side effects (85%), or was the standard of care (13%). Prostate (87%), head
Surgery occurred a median 40 days after the diagnosis and 43 days after the first
and neck (87%) and pediatric (83%) cancers were the most frequently listed
visit and within 35 days if the first appointment was with a thoracic surgeon. The
PBT-indicated disease sites, consistent with international guidelines. However,
threshold time associated with statistically significant worse survival was 60 days
pancreatobiliary (52%), breast (50%) and esophageal (44%) cancers were
after diagnosis. Surgery was performed more than 60 days of diagnosis in 115
(25.7%) patients, their OS was significantly worse than patients who had surgery
frequently advertised despite not being endorsed in any consensus guidelines.
earlier (HR=1.7 [95% CI: 1.1-2.6]). Conclusions: Greater intervals between On multivariate analysis, an increasing number of listed disease sites and claims
diagnosis of early-stage NSCLC and surgery are associated with worse survival. of being a regional PTC leader were associated with indicating that PBT offers
Efforts to minimize delays, particularly factors that prolong the period from greater disease control or cure. The availability of PBT through a clinical trial was
diagnosis to first contact with a lung cancer provider may improve survival. mentioned on 57% of websites. Conclusions: PTC websites often contain in-
formation and DTCA claims inconsistent with international consensus guide-
Cox proportional hazard model.
lines. As online marketing information may have significant influence on patient
Variables in the final model HR 95% CI p-value
decision-making, alignment of such information with accepted guidelines and
Male vs female 1.8 1.3 2.6 0.00 consensus opinion should be adopted by PBT providers.
Late stage vs early stage 3.8 2.2 6.5 0.00
Non-white vs white 5.6 1.4 23.1 0.02
Sub-lobar resection vs Lobectomy 1.9 1.3 2.8 0.00
Medicare/Medicaid vs non-Medicare 1.6 1.0 2.5 0.05
Time from surgery greater than 60 days vs less than 60 days 1.7 1.1 2.6 0.01

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
392s Health Services Research, Clinical Informatics, and Quality of Care

6600 Poster Session (Board #291), Sat, 1:15 PM-4:15 PM 6601 Poster Session (Board #292), Sat, 1:15 PM-4:15 PM
Mortality within 30 days of immunotherapy (checkpoint inhibitors) in meta- Does the innovative 4R Cancer Care Delivery Model improve patient self-
static cancer patients treated at Australian tertiary cancer center. First Author: management in safety net and non-safety net centers? First Author:
Hiren A. Mandaliya, Calvary Mater Newcastle Hospital, Waratah, Australia Julia Rachel Trosman, Center for Business Models in Healthcare, Chicago, IL
Background: Cancer treatment has evolved rapidly since the advent of im- Background: Under the NCI ASCO Teams Project, we proposed a 4R Model of
munotherapy (checkpoint inhibitors). As compared to chemotherapy, immu- teamwork and patient self-management (pSM) (Trosman JOP ’16). 4R (Right
notherapy is associated with a more favourable but distinct side effect profile. Info / Care / Patient / Time) enables patient (pt) and care team to manage
Mortality within 30 days of chemotherapy in cancer patients has been ac- complex care continuum with an innovative multimodality 4R Care Project
cepted as a clinical indicator of preventable harm and used as an auditing tool Plan. 4R includes a novel “project” feature – a graphical description of care
for clinical practice and improving quality of life. This should be investigated in interdependencies. 4R was previously piloted at 3 Chicago centers (Trosman
the current era of immunotherapy, as it has been the standard treatment for ASCO ‘18). Methods: In this new study, we improved and tested 4R for impact
advanced melanoma, lung cancer, renal cell cancer and others. Methods: We on pSM at 4 safety net and 3 non safety net centers across the US. 4R Plans
conducted a retrospective study on patients with advanced cancer treated with were provided to stage 0-III breast cancer pts Jan - Nov’18 (4R cohort). We
immunotherapy and died within 30 days of treatment. Clinical data on patients surveyed the 4R cohort and a historical control cohort of pts who received care
treated with immunotherapy at Calvary Mater Newcastle between 2006 and at same centers pre-4R, Jan - Dec ’17. Results: Survey response rates: 65%,
2018 was collected. Data were compared with 30-day mortality statistics of 105/162 (4R cohort); 44%, 190/432 (control). 4R markedly improved 4 of 5
chemotherapy. Results: A total of 601 metastatic cancer patients received pSM metrics vs control (Table). Additional analyses showed that safety net pts
immunotherapy agents (Pembrolizumab, Nivolumab, Ipilimumab, Atezolizu- had a significant increase in 4R vs control cohort in “seldom overwhelmed”
mab, Tislelizumab and MSB0011359C) on 5022 occasions. Seventy-six (84%, 49/58 vs 64%, 67/104 respectively, p = .007), while non safety net pts
(12.6%) patients died within 30 days of receiving immunotherapy. Median had nonsignificant increase. Other metrics improved to a similar extent for
age was 68 years (35-90). Melanoma was the most prevalent cancer type safety net vs non safety net pts. Within the 4R cohort, 85% found 4R useful in
(63%) followed by lung (20%). Forty-seven (47%) of patients received im- organizing their care and 73% found 4R’s novel “project” feature useful in
munotherapy as first-line treatment and 39% as second-line. Patients died understanding care interdependencies. Safety net pts reported similar use-
within 30 days received an average 2 (1-16) immunotherapy doses. A quarter fulness of 4R in organizing their care as non safety net pts (88%, 51/58 vs
of patients had ECOG 3 and ECOG 4 before last dose. Majority of deaths were 81%, 38/47, NS) and similar usefulness of the “project’ feature in un-
related to disease (86%). Nearly 80% of patients died in hospital. One patient derstanding care interdependencies as non safety net pts (74% vs. 72%, NS).
died due to treatment-related pneumonitis. In univariate analysis, there was no Conclusions: 4R significantly improved patient self-management, but further
association between mortality and patients’ demographic variables such as efforts are needed to expand the benefit to as close to a 100% of pts as
age, sex, BMI, cancer type, ECOG performance status, immunotherapy agent feasible. Safety net pts benefited from 4R at similar or higher rates than non
and prior treatment. Conclusions: To our knowledge, this is the first ever real- safety net pts, indicating that 4R may reduce care disparities. An expansion of
world data on 30-day mortality after immunotherapy in advanced cancer. 4R across the US continues this work.
Thirty-day mortality rates were comparable to published data on patients
% 4R cohort % Historical control P
treated with chemotherapy. Results emphasise significance of careful se- pSM metric N = 105 cohort N = 190 value
lection of advanced cancer patient for immunotherapy. Due to small sample
size, the power to detect a significant association between patients de- Know your cancer stage 94 78 .0002
Care plan clear 85 71 .01
mographics and survival is reduced. Timing & sequence of care clear 65 41 .0001
Seldom overwhelmed, not in control of 82 69 .03
your cancer care
Able to explain your care to others 80 76 NS

6602 Poster Session (Board #293), Sat, 1:15 PM-4:15 PM 6603 Poster Session (Board #294), Sat, 1:15 PM-4:15 PM
Patient-reported outcomes from two global multicenter clinical trials of chil- Where’s my doctor? the impact of the primary oncologist’s visit with their
dren and adults with tropomyosin receptor kinase (TRK) fusion cancer re- hospitalized patients. First Author: Brianna R. Bakow, Brown Univeristy
ceiving larotrectinib. First Author: Shivaani Kummar, Stanford Cancer Center, Alpert School of Medicine, Providence, RI
Stanford University, Palo Alto, CA
Background: Continuity of care is a cornerstone of the patient-practitioner
Background: TRK fusions involving NTRK1, NTRK2, and NTRK3 occur in di- relationship and patient satisfaction. The inpatient continuity visit (ICV), a
verse tumor types in children and adults. Larotrectinib, a selective TRK inhibitor, face-to-face patient-provider interaction, involves a discussion regarding
was recently approved by the FDA based on high objective response rates, durable hospital course and care goals and decisions. We theorize that the ICV in-
disease control, and a favorable toxicity profile in in patients with tumors harboring fluences patient satisfaction. Previously, patient satisfaction has been re-
TRK fusions (Drilon et al, NEJM 2018; Laetsch et al, Lancet Oncol 2018). To lated to patient perception of physician conduct, including communication
explore the impact of larotrectinib on quality of life (QoL), patient-reported out- skills. Currently, there are no studies investigating the impact of an ICV on
comes (PROs) are reported here for pediatric patients (SCOUT; NCT02637687) inpatient oncology patients on a hospitalist service. Objectives: To assess the
and adolescents and adults (NAVIGATE; NCT02576431). Methods: Patients relationship between the ICV and patient satisfaction. We hypothesized that
completed questionnaires assessing QoL (i.e., PedsQL, FACES, EORTC QLQ-C30, one or more visits by the outpatient oncologist would enhance satisfaction of
EQ-5D-5L) at baseline and planned cycle visits. Data were analyzed using de- oncology inpatients. Methods: Subjects (N=82) were comprised of adult
scriptive statistics, longitudinal data models, and graphical displays. Results: As inpatients on the oncology unit at Miriam Hospital, a teaching hospital of the
of July 2017, 37 patients (13 pediatric and 24 adolescent or adult) completed Alpert Medical School of Brown University. All participants had an oncologist
questionnaires at baseline and had at least one post-baseline measurement. at the hospital based cancer center. A survey, given at discharge, included a
Improvements compared to baseline in PROs were seen in both groups of patients 5-point Likert scale ranging from greatly worsened to greatly improved to
in all functional domains of the PedsQL (physical, emotional, social, and school) assess the impact of the ICV on patient satisfaction. Results: Of 82 par-
and of the EORTC QLQ-C30 (physical, role, cognitive, emotional, and social). ticipants, 46 reported a visit by their outpatient oncologist. Forty-two
Number of patients with MID-improvement for PedsQL (Total Score), EORTC (91.3%) reported that this visit either greatly or somewhat improved sat-
QLQ-C30 (Global Health Score), and EQ-5D-5L (VAS) are presented in the table isfaction with their hospital stay, while 8.7% reported no impact. Of patients
below. Improvements were rapid (by cycle 3 or 5), seen across most tumor types whose oncologist visited once, 94.4% reported either greatly or somewhat
and sustained a minimum of 2 cycles. Conclusions: Although necessarily viewed improved satisfaction compared to 89.3% who had more than one visit. Out
with caution due to relatively small sample sizes, these results nonetheless show a of 36 subjects who did not receive a visit, 16.7% reported that the lack of
meaningful early improvement in QoL in pediatric and adult patients with TRK visit either greatly or somewhat worsened their hospital stay, while 83.3%
fusion cancer treated with larotrectinib. Clinical trial information: NCT02637687 reported no impact. Conclusions: Our study suggests that an ICV improves
and NCT02576431. satisfaction of care in cancer patients on a hospitalist service. Furthermore,
Number of patients with MID-improvement. one of every six subjects who did not receive an ICV reported a negative
impact on satisfaction. Results highlight a possible intervention to the
Baseline and at least one Best post-baseline score MID-
post-baseline measurement greater than baseline score improvement* discontinuity of care that may be perceived by patients. While the practicality
N N (%) N (%) of this intervention requires evaluation, the efficacy of a single continuity
PedsQL (Total Score) 13 (pediatric) 11 (85) 10 (77) visit to improve satisfaction is reassuring.
EORTC QLQ-C30 24 17 (71) 15 (62)
(Global Health
Score)
EQ-5D-5L (VAS) 24 18 (75) 14 (58)
*MID: $10 points for EORTC QLQ-C30 and EQ-5D-5L; $4.5 points for PedsQL.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Health Services Research, Clinical Informatics, and Quality of Care 393s

6604 Poster Session (Board #295), Sat, 1:15 PM-4:15 PM 6605 Poster Session (Board #296), Sat, 1:15 PM-4:15 PM
The impact of NCI-sponsored network group clinical trials on guideline care Survivorship care: Improving delivery of care plans for hematology patients at
and new drug indications. First Author: Joseph M. Unger, Fred Hutchinson the Washington Cancer Institute (WCI). First Author: Hira Latif, Medstar
Cancer Research Center, Seattle, WA Washington Hospital Center, Washington, DC
Background: National Cancer Institute Clinical Trial Network (NCTN) groups Background: Cancer and long-term sequalae of its treatment impact the future
serve a vital role in identifying new antineoplastic regimens. However, the health and psychosocial wellness of these survivors. ASCO guidelines rec-
clinical impact of their trials has not been systematically examined. We ommend providing survivorship care to cancer patients who have completed
analyzed the influence of network group cancer clinical trials on clinical treatment with curative intent. The Commission on Cancer (COC) recommends
guidelines and new drug approvals. Methods: We evaluated Phase III cancer that survivorship care including treatment summaries be delivered to 50% of
clinical trials which the SWOG Cancer Research Network coordinated or eligible patients. In our ASCO Quality Training Program project, we aimed to
participated in (1980-2017). Included trials were completed and its results achieve this COC goal of 50% for the year 2018. Methods: Baseline data
published. A documented practice influential (DPI) trial was one with ver- collected from Jan 1, 2016 to June 30, 2018 indicated that 33% of he-
ified influence on National Comprehensive Cancer Network (NCCN) clinical matologic malignancy survivors at WCI received treatment summaries and
guidelines (available starting in 1996) or on U.S. Food and Drug Admin- survivorship care. For the year 2018, there were 11 survivors of hemato-
istration (FDA)-approved package inserts. We estimated the rate of DPI trials logic malignancy, and 4 of them (36%) had received survivorship care prior to
overall and over time. The total federal investment supporting the set of initiation of the project. We surveyed 12 providers to obtain data for perceived
trials was also determined based on public data. Results: In total, 182 trials challenges to deliver survivorship care. Large volume of patients, lack of re-
comprising 148,028 patients were studied. We identified 79 DPI studies sources, no standardized process and high no-show rates were identified as the
(43.4%); 73 influenced NCCN guidelines, 12 influenced new drug ap- most important barriers. A bi-monthly survivorship clinic run by hematology/
provals, and 6 influenced both. The rate of DPI trials was 72.3% (47/65) oncology fellows was initiated in September 2018 to address some of these
among formally positive trials (i.e., achieved their protocol specified end- barriers. Patient referral forms were mounted in the clinic work rooms to assist
point) and 27.4% (32/117) among negative trials. Thus 40.5% (32/79) of providers with identifying patients that qualify for survivorship; a pathway for
DPI trials were based on negative studies, half of which (16/32 = 50.0%) referrals to the survivorship clinic was created and providers were informed
reaffirmed standard of care over experimental therapy. There were no dif- about the clinic. Information flyers regarding survivorship care and the clinic
ferences between DPI and non-DPI trials in key study design characteristics. were placed in the waiting room to increase awareness amongst patients.
Total federal investment for the programs conducting the trials was $1.36 Results: By November 30, 2018, 63% of hematological cancer survivors
billion (USD2017), a rate of $7.5 million per trial, or $17.2 million per DPI received survivorship care and treatment summaries. Compared to the average
trial. Conclusions: Nearly half of all phase III trials by one of the NCTN’s from the preceding two years, survivorship care delivery increased by 30%.
largest groups had documented practice influence on clinical care guide- Conclusions: Our institution was able to meet the COC requirement by de-
lines or new drug approvals. Even many negative trials impacted guideline livering survivorship care to 63% of survivors of hematological malignancies
recommendations. Compared to the costs of a new drug approval in phar- through the intervention from this quality improvement study. We intend to
maceutical companies – typically estimated at . $1 billion – the amount extend this process to other tumor types to increase the delivery of consolidated
invested by federal funders to provide this valuable evidence was modest. survivorship care at the WCI.
These findings highlight the major role of the NCTN’s clinical trial program in
advancing oncology practice.

6606 Poster Session (Board #297), Sat, 1:15 PM-4:15 PM 6607 Poster Session (Board #298), Sat, 1:15 PM-4:15 PM
Multi-institution quality improvement in supportive oncology: Results of the Utilization of value stream mapping to improve chimeric antigen receptor
Coleman Supportive Oncology Collaborative (CSOC). First Author: Paramjeet (CAR) T-cell patient experience at an academic medical center (AMC). First
Khosla, Mount Sinai Hospital, Chicago, IL Author: Julie Porter, Stanford Health Care, Stanford, CA
Background: The Institute of Medicine and Commission on Cancer recommend Background: Value Stream Mapping (VSM) is commonly used to identify waste,
systematic delivery of supportive oncology care for cancer patients. The CSOC is reduce process cycle times and implement process improvements in the
focused on quality improvement (QI) of supportive care across Chicago cancer product and manufacturing space. VSM was applied to analyze and improve
centers (Weldon ASCO ’17). Supportive oncology includes distress, practical, the patient experience for individuals receiving CAR T cell therapy at this AMC,
family, physical, nutrition, pain, fatigue and care concerns. To support QI, cross- with emphasis on safety and quality of care, efficient patient flow and improved
institution teams developed unique, relevant tools, methods, care delivery communication across the care team. Methods: Fifty-three multidisciplinary
processes, patient handouts and online training. Methods: Ten centers (5 ac- representatives from 21 teams across the AMC participated in at least one of
ademic, 1 VA, 1 public, 2 safety net, 1 community) implemented supportive the five 4-hour VSM sessions. Comprehensive VSM workflows were created for
oncology screening and care delivery quality improvements. Centers collected each step in the current CAR T cell patient journey, from initial referral through
data for relevant Quality Oncology Practice Initiative (QOPI) metrics. Analyses post-infusion care. Future state analysis was performed, with emphasis on
used simple frequencies and Fishers exact test. Results: Five of six QOPI identification of information technology improvements, communication and
measures were improved at statistically significant levels from 2014 to 2017, pain points. Baseline outcome metrics were established to assess effective-
p , .00001. Improvements are more modest in 2016 & 2017 as 4 of the centers ness of future state VSM improvements. Results: Visual mapping of the CAR
started this QI in 2017. Conclusions: The CSOC achieved significant im- T cell patient journey identified several opportunities for improvement. A total
provements in supportive oncology screening and identifying and addressing of 72 improvement projects were outlined, which could be grouped into four
patients’ needs and concerns. Additional work is needed to improve these main categories: (1) establish channels for improved care team communi-
measures to achieve the best quality of cancer care possible for every patient cation, (2) build information technology infrastructure to support efficient
based on their needs and concerns. patient flow and quality of care, (3) develop teaching tools for patient and care
team education, and (4) delineate standard of care and clinical trials oper-
QOPI P-value 2014
Measure # 2014 2015 2016 2017 to 2017 ational workflows. Baseline outcome metrics were measured across the patient
Patients screened for supportive oncology needs, 24 17% 38% 53% 69% , 0.0001
population (n = 33), including time from consent to cell collection, length of
including distress (148/ (415/ (678/ (537/ inpatient stay, intensive care unit (ICU) utilization, and percent of read-
843) 1075) 1283) 779) missions within 7 days. Outcome metrics will be tracked and reported after
Stage IV patients screened for supportive on- 24 6% 15% 22% 35% , 0.0001
cology needs, within 30 days of diagnosis (27/ (58/ (78/ (91/
future state changes to determine effectiveness of VSM-based process im-
452) 379) 359) 261) provements. Conclusions: CAR T cell therapy is a novel therapy, which poses
Patients with one or more supportive oncology 25 36% 38% 46% 56% , 0.0001 new challenges for coordination across a multidisciplinary care team at an
need identified and addressed (303/ (413/ (587/ (439/
843) 1075) 1283) 779) AMC. The VSM approach is a useful tool to identify process and infrastructure
Documented discussion on understanding of 2 54% 55 66 55 NS improvements that can improve the patient journey. This approach can be
illness (453/ (592/ (627/ (432/ generally applied to drive process improvements across clinical care.
843) 1075) 945) 779)
Patients given prognosis time frame (days to 2 24% 34% 43% 44% , 0.0001
weeks, weeks to months, months to years, (205/ (364/ (403/ (340/
years+) 843) 1075) 945) 779)
Stage IV patients with documented palliative 43 15% 43% 36% 36% , 0.0001
referral (68/ (163/ (128/ (94/
452) 379) 359) 261)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
394s Health Services Research, Clinical Informatics, and Quality of Care

6608 Poster Session (Board #299), Sat, 1:15 PM-4:15 PM 6609 Poster Session (Board #300), Sat, 1:15 PM-4:15 PM
Venous thromboembolism (VTE) in patients with ALK-rearranged non-small Characterization of oncologic patient visits at the emergency department
cell lung cancer (ALK-LC): A population-based cohort. First Author: Alona Zer, (ED) at an Austrian tertiary care center. First Author: Christoph Mini-
Sarcoma Unit, Davidoff Cancer Institute, Rabin Medical Center, affiliated to chsdorfer, Department of Medicine 1 and Comprehensive Cancer Centre,
the Sackler Faculty of Medicine, Petach Tiqwa, Israel Medical University of Vienna, Vienna, Austria
Background: Advanced NSCLC is associated with an increased risk for VTE, Background: Oncologic patients frequently seek out the ED as first contact
with a reported rate of 8-15%. Clinical observations suggest a higher rate of point. Besides the rare classic oncologic emergencies other causes are
VTE in patients with ALK-rearranged NSCLC. Clalit Health Services (CHS) is supposed to be more frequent. However, little is known about reasons of
both a healthcare payer and provider, covering over 50% of the population in patients with active malignancies to consult the ED. Methods: Retrospective
Israel, with individual patient data recorded in a centralized electronic da- data regarding emergency visits of patients with a medical history of active
tabase. We aimed to determine the incidence of VTE in patients with ALK-LC cancer or adjuvant therapy, who were taken by ambulance to the ED in the
using a population-based cohort. Methods: We identified all patients di- time period from 01.01.2017 – 31.07.2017, was collected. For OS Kaplan-
agnosed with NSCLC between 01/2012-12/2017 within CHS. Clinical and Meier curves were created and univariate COX proportional hazards model
demographic data (including VTE risk factors, i.e. components of the Khorana analysis was used and log rank tests were performed. For statistical analysis
score) were extracted from the CHS registry for all patients. Patients with ALK- the chi2-test was used to investigate significant relations between the three
LC were identified according to crizotinib prescriptions (dispensed after an months mortality (3MM) and patient characteristics. 3MM was defined as
approved CHS’ pre-authorization for an ALK-LC diagnosis). VTE was identified death within 3 month of the first ED consultation. Results: From a total of
by ICD diagnosis codes 415.xx, 444.xx, 451.xx and 453.xx. VTE risk factors 1029 patient contacts 743 met the inclusion criteria. In total 552 patients
(Khorana score) were also extracted. Association between ALK-LC and VTE were included. The median age was 67 years (Minimum: 18y; Maximum:
were analyzed using logistic regression, estimating univariate and multivariate 95y). The vast majority of patients (480, 87%) were in a palliative setting.
Odds Ratios (OR). Results: Overall, 4327 patients with a diagnosis of NSCLC 243 patients (44%) were under active treatment. The biggest group of
were identified. 149 (3%) had at least one prescription for crizotinib for patients suffered from lung cancer (18%) followed by hematologic malig-
advanced ALK-LC. The rate of VTE in these patients was 25% (38 of 149 nancies (14%), pancreaticobiliary tumours (11%), breast cancer (8%) and
patients), while in the non-ALK-LC the rate was 14% (596/4178), OR = 2.05, colorectal cancer (7%). In total 111 patient contacts (15%) were treatment
p = 0.0004. The association was significant also in a multivariate analysis associated, 384 patient contacts (52%) were tumor related, the remaining
adjusting for, age, smoking status, BMI, platelet count, hemoglobin and (248, 33%) were unrelated to either. In 533 cases (72%) inpatient treat-
Charlson co-morbidity score (OR 1.66, p = 0.018). Conclusions: This pooled ment was necessary. The average length of stay was 9, 36 days. Furthermore
analysis of individual patient data confirms prior data from smaller retro- the 3MM was significantly higher in elderly patients . 65y (p = 0,001),
spective studies, suggesting ALK-LC is associated with a high risk of VTE. palliative patients not undergoing active treatment (p = 0,003) and patients
Randomized trials with prophylactic anti-coagulation are unlikely to be per- with elevated CRP levels ( . 0,5mg/dl) (p = 0,001) at time of ED contact. In
formed in this rare subtype, thus increased awareness and consideration of addition in our univariate analysis factors that significantly decreased the OS
VTE prophylaxis in high risk patients is warranted. were age . 65, visceral metastasis and palliative patients without active
treatment. Conclusions: The majority of cancer patients in this study were
in a palliative setting. Moreover, cancer associated complications posed the
most frequent cause for ED consultation. We could identify a higher risk for
mortality for elderly patients, patients with visceral metastasis and patients
undergoing best supportive care. The results may help to inform both cancer
patients and primary care units about frequent complications.

6610 Poster Session (Board #301), Sat, 1:15 PM-4:15 PM 6611 Poster Session (Board #302), Sat, 1:15 PM-4:15 PM
A randomized study comparing physician-directed or fixed-dose dexameth- QOL assessment integrated into the clinical care of cancer survivors to
asone replacement following incomplete steroid premedication for docetaxel identify needs and direct care. First Author: Christian S. Adonizio, Geisinger
chemotherapy. First Author: Tina Hsu, Ottawa Hospital, Ottawa, ON, Canada Cancer Institute, Danville, PA
Background: Prior to receiving docetaxel-based chemotherapy patients often Background: Validated survey tools have been used to measure the quality of
incorrectly take all or part of their steroid-premedication. The lack of stand- life (QOL) of patients treated for cancer, however, there are newer studies that
ardised steroid-replacement strategies can lead to variability in care and delays have shown both an improvement in QOL, and improvement in overall survival
in starting chemotherapy while nursing/pharmacy/physicians establish an in- using these tools. We integrated the Functional Assessment of Cancer Therapy –
dividualized patient plan, which can use up valuable chemotherapy chair time. General Population (FACT-GP v.4) to direct the deployment of resources and
A randomised controlled trial comparing a fixed-oral dose of dexamethasone interventions to improve the care of patients who have completed potentially
and physician-directed replacement was performed. Methods: Patients who curative therapy for cancer. Methods: This is an observational study of patients
missed at least one dose of steroid-premedication were randomised to either who received cancer therapy with curative intent in the last 18 months. The
standard replacement with dexamethasone 8mg orally or physician-directed FACT-GP was administered by an RN via telephone. Patients contacted re-
replacement (any steroid, dose or route). The primary outcome was time from ceived and reviewed a Survivorship Care Plan (SCP) as defined by the American
randomisation to starting docetaxel. Secondary outcomes included rates of College of Surgeons Committee on Cancer. Patients who had a total score less
acute and delayed hypersensitivity reactions, fluid retention and skin rashes. than 60 on FACT-GP and/or had a score less than 12 on the Emotional Well-
Results: Sixty patients were randomized. Most patients were enrolled during Being subscale (EWB) were considered high-risk and were referred to the
cycle 1 (47.5%) and cycle 2 (22%) of docetaxel. The most frequent total doses Survivorship MDC for in-person evaluation. Results: From 10/1/2018 to 12/
of dexamethasone omitted were 24 mg (27%), 12 mg (20%), and 8 mg (19%). 31/2018, 114 patients were referred to the cancer survivorship program. Of
There were 7 different replacement strategies used by physicians. The most these, 64 (56%) patients had FACT-GP administered and were evaluated. 45
frequently used strategies were: dexamethasone 8mg IV (34.5%), 12mg IV of these (70%) only completed the FACT-GP and received an SCP. 21 patients
(17.2%) and 20mg IV (13.8%). Patients in the fixed-dose arm received had a total score less than 60 and/or an EWB sub-score less than 12 and were
docetaxel earlier than patients in the physician-choice arm, at a median of 47.5 identified as high-risk. 15 (72%) patients were seen in MDC, 4 (19%) patients
and 61 minutes after randomization (mean = 62.2 vs 83.4 minutes) (p = 0.033). were seen in conjunction with a scheduled appointment by the MDC team, 2
No significant difference in rates of acute (0 vs 2)/delayed allergic reactions (1 vs (9%) patients refused further evaluation. 66.7% of patients in the survivorship
0), fluid retention (2 vs 1), or skin rashes (1 vs 0) was observed between the fixed- program were referred to Oncology Behavioral Health compared to 18.2% of all
dose and physician-choice arms respectively. Conclusions: This is the first oncology patients. Survivorship patients in the cohort had a baseline utilization
randomised trial to compare steroid-replacement strategies in this patient of the emergency department (ED) of 4.1% (10 of 241) from 1/1/2018 to 9/
population. Fixed-dose replacement with dexamethasone 8 mg PO should be the 30/2018 and 0 (0 of 64) after the initiation of the intervention from 10/1/2018
preferred standard of care, as it reduces both the time to starting docetaxel and to 12/31/2018. Conclusions: Integrating a validated QOL tool (FACT-GP) as a
treatment variability, with no apparent increase in toxicity. Clinical trial in- therapeutic intervention is feasible and may both identify needs and direct
formation: NCT02815319. services for cancer survivors while possibly decreasing ED utilization. Clinical
trial information: NCT03835052.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Health Services Research, Clinical Informatics, and Quality of Care 395s

6612 Poster Session (Board #303), Sat, 1:15 PM-4:15 PM 6613 Poster Session (Board #304), Sat, 1:15 PM-4:15 PM
Practice variation on hospital level in the systemic treatment of metastatic Impact of cancer in pulmonary embolism presentation and outcomes: A large
colorectal cancer in the Netherlands: A population-based study. First Author: academic center study. First Author: Cherry Au, Sidney Kimmel Medical
Lotte Keikes, Department of Medical Oncology, Academic Medical Center, College at Thomas Jefferson University, Philadelphia, PA
University of Amsterdam, Amsterdam, Netherlands
Background: The risk of venous thromboembolism is increased 4- to 7-fold in
Background: Population-based data on the implementation of guidelines for patients with malignancy, emphasizing the need to identify and treat these
cancer patients in daily practice are scarce. Therefore, we evaluated practice patients early to improve outcomes. We aimed to study the clinical presentation
variation patterns and associated variables in the systemic treatment of and outcomes of pulmonary embolism (PE) in patients with and without cancer.
metastatic colorectal cancer (mCRC) between 2008 and 2015 in the Neth- Methods: We performed a retrospective analysis of consecutive patients di-
erlands. Methods: We selected a random sample of adult mCRC patients agnosed with PE via CT scan from 2014-2016 at Jefferson Hospital. We
diagnosed from 2008 to 2015 from the National Cancer Registry in 20 Dutch compared patient characteristics, presentation, PE characteristics and mor-
hospitals. We examined the influence of patient, demographic and tumor tality of patients with and without cancer. Cox proportional regression hazards
characteristics on the odds of being treated with systemic therapy according to model was used for survival-time analysis. Results: Our study included 581
the current guideline and assessed its association with survival. Results: Our patients, of which 187 (32.1%) had active cancer. Cancer patients were less
study population consisted of 2222 mCRC patients of whom 1307 patients likely to have chest pain (18.2% vs 37.4% p , 0.01), syncope (2.7% vs 6.6%
received systemic therapy for mCRC. Practice variation was most obvious in the p = 0.05), bilateral PEs (50% vs 60% p = 0.025), and right heart strain (RHS)
use of bevacizumab and anti-EGFR therapy in patients with (K)RAS wild-type (48% vs 58% p = 0.024). Indwelling catheters (IC) were present in 41.2% (n =
tumors. Administration rates did not differ between hospital types but fluc- 77) of cancer patients. However, presence of IC was not associated finding of
tuated between individual hospitals for bevacizumab (8-92%; p,0$0001) incidental PEs (26% vs 18.2% p = 0.201). There was no difference in hospital
and anti-EGFR therapy (10-75%; p=0$05). Bevacizumab administration was length of stay (8.9 vs 9.4 days p = 0.61) or intensive care unit admission (31.9%
inversely correlated to higher age (OR:0$2; 95% CI: 0$1-0$3) comorbidity vs 33.3% p = 0.75). There were fewer massive PE (3.2% vs 7.1% p = 0.06) in
(OR:0$6; 95% CI: 0$5-0$8) and the presence of metachronous metastases patients with cancer, but this difference was not statistically significant. Cancer
(OR:0$5; 95% CI: 0$3-0$7), but patient characteristics did not differ between patients elected comfort care at higher rates (15.2% vs 5.4% p = 0.01). Cancer
hospitals with low or high bevacizumab administration rates. Exposure to patients had higher 1-year mortality as compared to non-cancer (adj HR 6.9,
bevacizumab (HR:0$8; 95% CI: 0$7-0$9) and anti-EGFR therapy (HR:0$6; 95% CI 3.3- 14.7, p , 0.01). Among cancer patients, 52.7% had metastasis
95% CI: 0$5-0$8) was associated with prolonged survival. Conclusions: We with a higher 1-year mortality (adj HR 2.5, 95% CI 1.8- 4.9, p , 0.1) and
identified significant inter-hospital variation in targeted therapy administration 35.8% were on active chemotherapy with no difference in 1-year survival (adj
for mCRC patients, which may affect outcome. Age and comorbidity were HR 1.1, 95% CI 0.6-1.8, p = 0.79). The most represented cancers were
inversely correlated with non-administration of bevacizumab, but did not genitourinary, lung and head and neck (35.3%, 23.0%, 13.4%, respectively).
explain inter-hospital practice variation. Our data strongly indicate that Conclusions: Cancer patients presented with less severe pulmonary emboli
practice variation is based on individual strategy of hospitals rather than which may be due to increased health care contact and pre-clinical suspicion.
guideline recommendations or patient-driven decisions. Individual hospital The presence of IC did not affect the size, location of PE or incidental PEs
strategies are an additional factor that may explain the observed differences among cancer patients. Although cancer patients have higher 1-year mortality,
between real-life data and results obtained from clinical trials. PE may not be as large as a contributor as previously perceived.

6614 Poster Session (Board #305), Sat, 1:15 PM-4:15 PM 6615 Poster Session (Board #306), Sat, 1:15 PM-4:15 PM
Impact of specialist palliative care delivered over three months prior to death Impact of an oncology urgent care clinic on emergency department rates.
on a colorectal cancer patient’s risk of experiencing aggressive end-of-life First Author: Tannaz Sedghi, Cancer Outcomes, Public Policy, and Effec-
care. First Author: Aynharan Sinnarajah, University of Calgary, Calgary, AB, tiveness Research (COPPER) Center, Yale Cancer Center and Yale University
Canada School of Medicine, New Haven, CT
Background: More patients are experiencing aggressive end-of-life (EOL) care. Background: Oncology-specific urgent care clinics (UCC) may play a key role
This is concerning as aggressive EOL care, on a population level, is associated in reducing unscheduled emergency department (ED) visits among patients
with poor quality care. Specialist palliative care (PC) has been shown to help with cancer. We sought to determine if establishment of an Oncology UCC was
relieve EOL symptoms, improve patient quality of life, and reduce aggressive associated with lower ED utilization among patients receiving cancer care at
EOL care. This study aimed to estimate the impact of the timing of specialist Yale’s Smilow Cancer Hospital (SCH) and two nearby, integrated community
PC, specifically PC delivered at least 3 months prior to death, on a colorectal practices. Methods: SCH opened its UCC in April 2017 to provide supportive
cancer (CRC) patient’s risk experiencing aggressive care in the last 30 days of care and symptom management for patients with cancer who need acute
life. Methods: A population-based retrospective cohort study of adult patients medical attention outside of regular clinic visits. We identified patients who
who died from CRC in Alberta, Canada from 2011-2015. The Alberta Cancer had at least one visit with an oncology provider during the Pre-UCC period (9/1/
Registry was used to identify the cohort, which was linked to healthcare re- 16 – 12/31/16) or Post-UCC period (9/1/17 – 12/31/17) and received che-
source use data in local, provincial, and national databases. Individuals who motherapy within a year preceding their provider visit. For each patient, we
died , 30 days from CRC diagnosis were excluded. Patients who accessed any captured all ED visits in a four-month window starting from their last provider
of the provinces specialist PC services were deemed exposed to specialist PC visit in each study period. The ED visit rate for both periods was defined as the
(includes PC consult team, intensive PC unit, palliative home care, hospice). total number of ED visits divided by the total number of unique patients in the
Aggressive EOL care was defined as having experienced at least one of: hospital period. To determine the impact of the UCC on ED utilization, we evaluated the
death, . 1 emergency department visit, . 1 hospital admission, . 14 days of absolute difference in the ED visit rate between the Pre- and Post-UCC period
hospitalization, $1 intensive care unit admission, $1 new chemotherapy using a two-sample t test. Results: There were 3,754 patients in the Pre-UCC
program (or any treatment in the last 14 days of life). Logistic regression was period and 4,734 patients in the Post-UCC period. In the full study sample, the
used to model factors (specialist PC timing and clinical characteristics) as- mean age was 62.9 and most common cancer types were Hematologic,
sociated with aggressive EOL care. Results: The cohort comprised 2979 Gastrointestinal, and Breast. Prior to opening the UCC, the ED visit rate was
patients. Most patients received specialist PC before death (58%); 60% 0.27 per unique patient. After opening the UCC, we found a 13.9% relative
had $1 indicator of aggressive EOL care. Relative to patients who received decrease in the overall ED visit rate from 0.27 to 0.23 (p = 0.02). The SCH
specialist PC . 3 months before death, patients who received specialist PC , patient ED visit rate declined by 12.5% (p = 0.03) and the community practice
3 months before death were 1.5 times more likely to experience aggressive EOL rate declined by 37.1%; however, the latter decline was not statistically
care (CI: 1.2-1.9). Patients who received no specialist PC were 2.1 times more significant, potentially due to a small sample size (p = 0.19). Conclusions: Our
likely to experience aggressive EOL care (CI: 1.7-2.8). Short disease duration study found a decrease in the ED visit rate after the opening of an Oncology
( , 1 year from diagnosis to death), younger age at death, living in a rural area, UCC. An urgent care strategy for cancer centers may serve as an efficient way to
and male sex, were also associated with higher odds of experiencing aggressive manage patients while minimizing ED use.
EOL care. Conclusions: Specialist PC delivered . 3 months before death
Pre-UCC Post-UCC
reduces a CRC patient’s risk of experiencing aggressive EOL care over PC
Unique Patients ED Visit Rate Unique Patients ED Visit Rate Relative Change P
delivered , 3 months before death.
Overall 3754 0.27 4734 0.23 -13.9% 0.02
SCH 3590 0.26 4551 0.23 -12.5% 0.03
Community 164 0.35 183 0.22 -37.1% 0.19

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
396s Health Services Research, Clinical Informatics, and Quality of Care

6616 Poster Session (Board #307), Sat, 1:15 PM-4:15 PM 6617 Poster Session (Board #308), Sat, 1:15 PM-4:15 PM
Quality control system of Watson for oncology: Artificial intelligence for support- Combination of statin/vitamin D and metastatic castration-resistant prostate
ing clinical decisions in oncology. First Author: Juemin Fang, Shanghai Tenth cancer (mCRPC): A post-hoc analysis of two randomized clinical trials. First
People‘s Hospital, Tongji University, Shanghai, China Author: Guillermo de Velasco, Department of Medical Oncology, University
Hospital 12 de Octubre, i + 12, Madrid, Spain, Madrid, Spain
Background: Watson for Oncology (WFO) is an artificial intelligent clinical
decision-support system (AI-CDSS) developed by IBM and trained by Me- Background: Retrospective database studies have suggested that statins and vi-
morial Sloan Kattering Cancer Center to assist in cancer care by providing tamin D have a positive impact on prostate cancer survival and specifically in mCRPC
evidence-based treatment options with priority. However, there are dis- patients (pt). Methods: We conducted a post-hoc analysis of individual pt data of
agreements argue that WFO recommends “unsafe and incorrect” cancer mCRPC pts treated with abiraterone (AA) and/or Prednisone (P) on two randomized
treatments. Also, guidelines and drug availability in China are different from phase III clinical trials COU -AA-301 and COU-AA-302 to analyze the impact of
USA. Therefore, a quality control system of WFO is urgently needed to help statins and vitamin D in overall survival (OS). Statistical analyses were performed
oncologists better use WFO in China. Methods: Experts from medical on- using the Kaplan Meier method and Independent predictors were investigated using
Cox regression analysis. This study, carried out under YODA Project #2016-1136,
cology, surgical oncology, radiology, intervention, radiology and pathology
used data obtained from the Yale University Open Data Access Project.
etc. forming a Multiple Disciplinary Team (MDT) to score Watson recom- Results: These two studies included 2280 patients (1340 treated with AA/P and
mendations in 6 aspects (shown in the table). Results: With this quality 640 with P). Use of Statin + vitamin D was associated with a 38% reduction in
control system, the value of WFO was carefully evaluated by MDT. Rec- mortality in the postdocetaxel setting and 32% in the predocetaxel setting in patients
ommendations with higher score(especially more than 80) were more treated with abiraterone (Table 1 and 2). No significant reduction in the rate of
standardized, reasonable and evidence-based thus more likely to be chosen. skeletal-related events was seen in patients treated with vitamin D or statins.
Localization and drug availability problem was solved by taking Chinese Conclusions: To our knowledge this is the first report suggesting the impact of vi-
guidelines and drug approval into evaluation within this scoring system. tamin D+statin in mCPRC treated with abiraterone. The potential benefits of vitamin
Treatment options unsuitable or unavailable for patients by the system will D do not seem to be secondary to concomitant statin use in this population. Further
be removed and replaced by the advices of MDT. Conclusions: Reliability studies are needed to confirm these results.
and security are the top concerns of applying new technology in healthcare. Impact of vitamin D and statin in overall survival.
With the MDT quality control system, AI-CDSS can be used safely and ef- Medication use n Median OS HR for OS (95% CI) P value
ficiently before it is fully mature. Also, the accuracy and advancement are COU302 Prednisone (control, no statin, no vit D) 228 29.2 (26.0; 31.6) 1
assessed in this system to help oncologists better use WFO in China in the Abiraterone (no statin, no vit D) 215 33.7 (28.8; 37.2) 0.76 (0.61; 0.96) 0.0191
future. Indicators evaluating the WFO recommendations. Abiraterone+statin+vitD 95 35.0 (30.6; 42.8) 0.68 (0.51; 0.92) 0.0108
COU301 Prednisone (control) 254 10.8 (9.6; 13.3) 1
Abiraterone (no statin, no vitD) 473 14.7 (13.9; 16.0) 0.91 (0.77; 1.07) 0.2598
Actual Abiraterone+statin+vitD 45 21.7 (20.0; 25.5) 0.62 ( 0.44; 0.77) 0.0075
Evaluation index Score score
1 WFO recommendations and guidelines (NCCN, ESMO, CSCO) 20
compliance rate Multivariate analysis of overall survival.
2 WFO recommendations and MDT suggestions compliance rate 20 Multivariate analysis Multivariate analysis
3 Evidence-based level of WFO recommendations 20 COU302 COU 301
4 Evidence-based level of references supporting options by WFO 20 HR for OS (95% CI) p-value HR for OS (95% CI) p-value
5 Patient-specific level 10
6 Drug availability 10 ECOG 1.32 (1.04; 1.68) 0.1577 1.55 (1.36; 1.76) ,0.0001
BMI 0.98 (0.96; 1.01) 0.0226 0.98 (0.97; 1.01) 0.1766
Total score 100 Prednisone (control, no statin, no vit D) 1 - 1 -
Abiraterone (no statin, no vit D) 0.76 (0.61; 0.95) 0.0172 0.89 (0.75; 1.05) 0.1812
Note：NCCN, the National Comprehensive Cancer Network；ESMO, European Society Abiraterone+statin+vitD 0.68 (0.50; 0.91) 0.0095 0.54 (0.38; 0.76) 0.0005
for Medical Oncology; CSCO, Chinese Society of Clinical Oncology.

6618 Poster Session (Board #309), Sat, 1:15 PM-4:15 PM 6619 Poster Session (Board #310), Sat, 1:15 PM-4:15 PM
Adherence to the guidelines: A comparison of biomarker testing implementa- Qualifying sites for oncology clinical trials. First Author: Dax Kurbegov, Sarah
tion in metastatic non-squamous, non-small cell lung cancer in university Cannon Research Institute, Nashville, TN
versus community institutions. First Author: Sufana Shikdar, Mercy Catholic
Background: Current methods to assess trial sites for clinical trial partici-
Medical Center, Darby, PA
pation are onerous, with unnecessary redundancies and “no-value” steps
Background: Molecular biomarkers have become essential in determining that impact clinical trial participation. This project assessed the impact of
optimal treatment for patients with advanced non-small-cell-lung cancer current sponsor and contract research organizations (CRO) methods to
(NSCLC). Few studies have evaluated the implementation of biomarker evaluate sites for trials. Methods: A survey was conducted with community-
assessment (e.g. EGFR, ALK and ROS1) in routine clinical practice. We and academic-based trial sites. Samples of feasibility questionnaires (FQs)
endeavored to assess adherence to biomarker testing guidelines in different used by sponsors and CROs were also compiled. An ASCO sponsored multi-
clinical settings, specifically in a university hospital versus a community stakeholder meeting was held to identify strategies to more effectively assess
hospital in the same region. Methods: A retrospective analysis was con- trial sites. Results: 113 oncology practices (63 community, 50 academic)
ducted in newly diagnosed metastatic non-squamous-NSCLC patients reported completing 11 FQs and 4 pre-study site visits (PSV) on average per
comparing the compliance of biomarker testing based on nationally month. On average, each FQ took 4 hours (528 hours/site and 59,664 hours
established guidelines available at the time of diagnosis. De-identified for all respondents, annually) and each PSV took 10 hours (480 hours/site
electronic health record (EHR) data were collected from Mercy Catholic and 54,240 hours for all respondents, annually) to complete. Thus, the total
Medical Center (MCMC) and Hahnemann University Hospital (HUH) be- staff hours required to complete site feasibility assessments was 113,904
tween 1/1/15- 1/30/19. Results were compared in each setting to determine annually. Respondents reported that content in both FQs (82%) and PSVs
utilization of biomarker testing. Results: 27 patients were identified at (91%) was redundant to information previously provided and FQs were
MCMC and 41 at HUH. 22 (81%) patients at MCMC and 36 (88%) patients redundant between different sponsors (86%). The 42 sample FQs had a
at HUH underwent appropriate molecular testing based on guidelines median 45 questions (range 13 to 96). Respondents noted that sponsors/
available at the time of diagnosis. Conclusions: Our data suggests that both CROs provided insufficient study documentation to accurately complete
university and community institutions have appropriately adapted the FQs. It took 7 months on average from first contact to first patient enrolled.
evolving guidelines for molecular testing for patients with non-squamous Respondents also provided feedback about standardizing and streamlining
NSCLC. In the rare instances that molecular testing was not performed, the site qualification processes. Conclusions: The current methods of assessing
most common reason was inadequate amounts of tissue available. Newer site feasibility for clinical trials poses tremendous burden on site resources
technologies, such as next-gen sequencing and serum based testing, will and is not sustainable. New methods are needed that standardize, har-
make compliance with guidelines easier in the future, particularly as in- monize, and streamline criteria and site assessments. Such changes will
creasing numbers of molecular markers will need to be assessed. reduce burden and costs for all stakeholders, and will expedite and increase
patient enrollment onto clinical trials.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Health Services Research, Clinical Informatics, and Quality of Care 397s

6620 Poster Session (Board #311), Sat, 1:15 PM-4:15 PM 6621 Poster Session (Board #312), Sat, 1:15 PM-4:15 PM
Impact of Oncology Care Model (OCM) reporting requirements on quality of Decreasing postoperative opioid prescriptions in ambulatory extended re-
care. First Author: Emily Castellanos, Flatiron Health, New York, NY covery patients. First Author: Nkechi Fearon, Memorial Sloan Kettering
Cancer Center, New York, NY
Background: The OCM is a voluntary Center for Medicare and Medicaid In-
novation alternative payment model pilot program. As of Oct 2017, OCM Background: Over-prescription of opioids after surgery contributes to the
practices are required to report the biomarker status for NSCLC pts. Our opioid abuse epidemic. Optimum post-operative opioid dosing is not de-
objective was to assess the effect of OCM reporting on quality of care in fined. We evaluated prescribing patterns among different surgical services
aNSCLC. Methods: We developed a decision-analytic model to compare the and created a standardized practice to reduce dispensation of unnecessary
likelihood of receiving biomarker testing and corresponding appropriate opioids. Methods: Opioid-naı̈ve patients over 18 who underwent urologic,
therapy. We populated the model using real-world data from pts (n=7,075) at gynecologic, or breast surgery between March 2018 and January 2019 were
OCM sites (n=45) and non-OCM sites (n=105) in the Flatiron Health electronic eligible. A 4-month pre-intervention evaluation of number of opioid pills
health record (EHR)-derived database. The pre-period control included pts prescribed, number of pills taken, additional refills, and pain-control was
diagnosed with aNSCLC from Jan 1, 2011 - Dec 31, 2015; the post-period obtained by contacting patients 7-10 days post-operatively. Findings were
included pts diagnosed Oct 2017 - Nov 2018. For OCM vs non-OCM sites, we used to standardize prescriptions. Following implementation, patients un-
estimated probabilities and unadjusted odds ratio (OR) of biomarker testing dergoing surgery for the following 4-months were contacted to assess the
(EGFR, ROS1, or ALK) and subsequent delivery of appropriate therapy, de- impact of standardized opioid prescriptions. Data was compared with the
fined as use of a biomarker-guided tyrosine kinase inhibitor (TKI) for positive institution’s electronic prescription system. Results: Pre-intervention, 368
pts, or non-TKI therapy for negative pts. Results: No differences in rates of eligible urology and gynecology patients (75.6%) responded and were
biomarker testing or delivery of appropriate therapy were detected between prescribed between 6 and 40 opioid pills. Urology patients received median
OCM and non-OCM practices prior to the reporting requirement. In the post- 28 (20, 30) tablets and 33% reported taking none. Gynecology patients
period, OCM was associated with higher odds of biomarker testing and ap- received a median 20 (19, 28) tablets and 41% took none. Of 238 mas-
propriate therapy (Table). Conclusions: To our knowledge, this is the first study tectomy patients, 176 (74%) reported taking median 3 and 4.9 of 20
of the association of OCM reporting requirements with downstream quality of prescribed opioid pills and 39% or 61% took no opioid pills (without vs with
care. Our results suggest that OCM documentation and reporting requirements reconstruction). Prescriptions were standardized to 8, 7, and 10 tablets for
are associated with modestly higher quality of care for pts with aNSCLC. urology, gynecology, and breast services. Post-intervention surveys revealed
Ongoing sensitivity analyses will determine the relative contribution of provider opioid tablets taken to be unchanged with minimal increase in refill requests.
and practice characteristics to these findings. Careful measurement of the Conclusions: Prior to standardization, a large variation in opioids prescribed
impact of reporting requirements is essential to measure the impact of pay- was observed. Standardizing opioid prescriptions resulted in fewer opioids
ment reform interventions and inform policy. dispensed without impacting pain control or refill requests.
Non-OCM OCM p-
Outcomes (%) (%) OR value
Biomarker testing rate 81.3 85.2 1.32 0.036
Positive result 18.1 20.6
Negative result 78.5 76.8
Overall Total: testing and appropriate therapy delivered 72.1 76.4 1.25 0.042
Appropriate first-line therapy (TKI) among biomarker 63.6 68.5
positive
Appropriate first-line therapy (non-TKI) among bio- 98.3 98.3
marker negative

6622 Poster Session (Board #313), Sat, 1:15 PM-4:15 PM 6623 Poster Session (Board #314), Sat, 1:15 PM-4:15 PM
Virtual visits for children, adolescents, and young adults with cancer. First Evaluating the impact of telephonic nurse care management during high
Author: Peter Meade Anderson, Cleveland Clinic, Cleveland, OH toxicity chemotherapy treatment. First Author: Craig A Hunter, CVS Health,
Northbrook, IL
Background: Children, adolescents, and young adults have rare cancers and
standard-of-care treatment is commonly very aggressive. Virtual visits provide Background: Care management is aimed at improving quality and cost of
include many of the nuances of face-to face communication. These are much cancer care. To measure this we evaluated inpatient (IP) and emergency room
friendlier than phone calls or email and can be scheduled and structured to (ER) admissions and costs of chemotherapy patients receiving versus not
provide a large amount of information efficiently. Methods: Cleveland Clinic receiving telephonic nurse care management. Methods: 1,609 patients were
uses HIPPA-compliant software from American Well (Boston, MA) that allows studied over a 35 month period, of which 802 enrolled in a voluntary holistic
the health care provider and patient to use a phone, tablet, or desktop telephonic nurse care management program. Patients were identified based on
computer for video visits. Our intake process involves obtaining a Medical having been prescribed or initiating treatment with a high toxicity chemo-
Record Number (MRN), sending a brief summary, uploading or sending a CD therapy agent within the previous 60 days, following which notification was
with images in DICOM, and having an administrative assistant schedule the received from an oncology pathways intermediary. Nurse care managers en-
virtual visit. Telemedicine sessions typically last ,60 minutes. During the gaged patients one or more times per month based on patient need, educating
visit a summary is updated, images are reviewed, and this and other in- on side effects, pain management, where to seek care, and general well-being.
formation shared via email after the visit. Results: In 2017+ 2018 we con- IP and ER were measured during the 14 days following chemotherapy ad-
ducted 223 virtual visits; 85% were ,30 years old (table). The summary has ministration (Risk-Period) and compared to a baseline of all remaining patient
been a key to efficient and effective organization and includes not only contact claims history, excluding 28 days following chemotherapy administration
information and past medical history, but also an “Opportunities to Improve (14 day Risk Period and 14 day washout period). Results: IP rates per 1,000
Health" section (problem list /action plan). Topics discussed in solid tumor Risk Period Days for managed patients decreased during the 14 days following
patients include: 1) local control, 2) medical therapy (chemotherapy), 3) chemotherapy by 44% as compared to their baseline (3.02 v 5.39, p#0.001),
imaging and tumor markers, 4) control of side effects and nutrition, 5) social while non-managed patients increased by 29% (2.33 v 1.81, p=0.003). ER
issues and goals of care (which can include palliative care and hospice), and 6) rates per 1,000 Risk Period Days were not significantly different for managed
follow-up. A power point with key images and the updated summary and patients (2.72 v 2.35, p=0.188) and significantly increased by 117% for non-
articles are emailed at the end of the visit to the patient & caregivers and often managed patients (2.34 v 1.08, p#0.001). Baseline utilization for managed
referring physician, NP, or PA . Visit diagnoses have included osteosarcoma patients was more than twice the rate of non-managed patients, suggesting
and Ewing sarcoma (73%), but also other rare cancers such as rhabdomyo- that a greater need for support may influence voluntary participation in care
sarcoma, DSRCT, paraganglioma, and adrenal cortical carcinoma. Survivor- management. Conclusions: Telephonic nurse care management meaningfully
ship and cGVHD have also been discussed. Conclusions: A structured virtual reduced IP and ER admissions and their associated costs.
visit to help young people and their caregivers understand complex multi-
disciplinary cancer care is now possible for all regions of North America. A Managed Patients Non-Managed Patients
major source of satisfaction has been two-way sharing of information to im- N 802 807
prove not only cancer control, but also improved nutrition, communication, and Average Days at Risk per Member 71.4 104.2
proactive toxicity reduction. Number of IP Admits (B; RP) 363; 173 582; 196
Number of ER Admits (B; RP) 158; 156 348; 197
Age Range % of Visits Average Cost per IP Admit (B; RP) $44,006; $38,606 $52,863; $49,557
Average Cost per ER Admit (B; RP) $1,292; $1,173 $1,146; $1,211
0-12 20.4
13-18 30.9 IP Rate Change (44%); p=,0.001 29%; p=0.003
19-30 33.2 ER Rate Change 16%; p=0.188 117%; p=,0.001
31+ 15.5
B=Baseline; RP=Risk Period

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
398s Health Services Research, Clinical Informatics, and Quality of Care

6624 Poster Session (Board #315), Sat, 1:15 PM-4:15 PM 6625 Poster Session (Board #316), Sat, 1:15 PM-4:15 PM
Evaluating the American Society of Clinical Oncology (ASCO) and European Is neratinib following trastuzumab in early-stage HER2-positive breast cancer
Society of Medical Oncology (ESMO) value frameworks for Food and Drug cost-effective? First Author: Naomi RM Schwartz, University of Washington,
Administration (FDA) approved checkpoint inhibitors (CIs). First Author: Seattle, WA
Sophie Feng, Princess Margaret Cancer Centre, Toronto, ON, Canada
Background: Neratinib after adjuvant trastuzumab significantly improves
Background: Although the development of CIs has led to a dramatic change in disease-free survival (DFS) in human epidermal growth factor receptor 2-
the oncology landscape, these agents are associated with significant costs and postiive (HER2+) breast cancer, but the median absolute DFS gain is only
toxicity. ASCO and ESMO have developed separate frameworks to define the 1.3 months. There has been much controversy in the clinical and lay media as to
value of emerging cancer treatments in order to encourage cost-effective whether the substantial cost of neratinib is justified by its effects, including a
therapies. We apply these frameworks to trials supporting FDA approvals prominent ASCO Post article from Dr. Vogl about a year ago. We performed a
of CIs and explore the correlation between these two scoring systems. cost-utility analysis to formally assess the value of adding neratinib based on
Methods: We searched the FDA database for CIs and indications approved Phase III ExteNET trial results. Methods: We developed a Markov state-
between January 1, 2011 and January 1, 2019. Only randomized phase II/III transition model to assess the value of neratinib in Stage I-III HER2+ breast
trials for solid tumors were included. Data on survival, toxicity and quality of life cancer. Five-year recurrence rates were derived from ExteNet. Mortality and
were extracted from the most recent publications by two reviewers in- recurrence rates after 5 years were derived from the HERceptin Adjuvant
dependently. A trial showing a substantial benefit was defined as an ASCO (HERA) trial. Costs were derived from wholesale acquisition costs and peer-
score of $ 45, or ESMO Grade 4-5 (palliative setting) or Grade A/B (curative reviewed literature. Health state utilities were derived from ExteNET and prior
setting). Concordance for substantial benefit was assessed using Cohen’s publications. Outcomes included life years (LY), quality-adjusted life years
Kappa while Spearman coefficients were used to determine the degree of (QALYs), direct medical expenditures, and cost per QALY gained. The analysis
correlation in individual scores. Results: We identified 40 FDA indications for took a payer perspective over a lifetime horizon and results were discounted at
7 CIs. Of these, 18 indications based on Phase I/II single-arm trials were 3% per year. One-way and probabilistic analyses were conducted to evaluate
excluded. The remaining 22 indications were based on 21 randomized phase uncertainty. As neratinib conferred more clinical benefit in hormone receptor-
II/III trials (3 adjuvant, 18 metastatic). In the palliative setting, 73% and 86% positive (HR+) disease, we also assessed value in that specific subgroup.
trials showed substantial benefit based on ASCO and ESMO frameworks re- Results: Base case results are presented in Table. At typical U.S. willingness to
spectively [median ASCO score: 54.8, interquartile range (IQR) 46.2-64.0; pay thresholds of $100,000 and $150,000 per QALY gained, neratinib had
median ESMO score: 5, IQR: 4-5]. 27% of trials were scored intermediate or 16.7% and 27.2%, probabilities of cost-effectiveness, respectively. In the HR+
low benefit by ASCO, while 9% were ineligible for ESMO scoring. Weighted subgroup, neratinib had a cost of $275,311 per QALY gained (19.9% & 31.2%
kappa was 0.719 between the two frameworks. Spearman rho was 0.84. All 3 probability of cost-effectiveness at $100,000 & $150,000 per QALY).
adjuvant trials were assigned ESMO grade A but low benefit with ASCO Conclusions: In the first independent assessment of the value of neratinib after
(median 37.7, IQR 20.5-40.9). Conclusions: In the palliative setting, the adjuvant trastuzumab, neratinib is not projected to be cost-effective, even
majority of trials supporting FDA approved CI indications demonstrated sub- among patients who derived the most clinical benefit. Future analyses should
stantial benefit using both ASCO and ESMO frameworks. There was a strong reassess the cost-effectiveness of neratinib treatment as trial data mature. Base
correlation between the two frameworks. However, in the curative setting scores case results.
were discordant. The ASCO framework may require further refinement for
Results Neratinib Observation Difference
adjuvant trials.
Cost $317,619 $152,812 $164,806
LYs 18.31 18.17 0.14
QALYs 15.67 15.28 0.40
Cost per LY -- -- $1,183,223
Cost per QALY -- -- $416,106

6626 Poster Session (Board #317), Sat, 1:15 PM-4:15 PM 6627 Poster Session (Board #318), Sat, 1:15 PM-4:15 PM
Relationship of emergency department use pre- and post-cancer diagnosis in A weighted criterion-based approach to value assessment of oncology drugs.
safety-net adults. First Author: Arthur Hong, University of Texas South- First Author: Doreen Anuli Ezeife, Tom Baker Cancer Centre, Calgary, AB,
western Medical Center, Dallas, TX Canada
Background: Safety-net adults generate a high rate of emergency department Background: The rising cost of anti-cancer therapy has motivated recent ef-
(ED) visits within the 180 days after a new cancer diagnosis, many of which forts to quantify the overall value of new cancer treatments. Multi-criteria
could be alternatively triaged to an urgent care clinic. It is unclear how much of decision analysis offers a novel approach to establish an explicit framework to
this ED use is attributable to the cancer and treatment vs. ED-seeking behavior. evaluate new cancer treatments. Methods: We recruited a diverse multi-
To identify patients at risk of frequent ED use, we explored whether a patient’s stakeholder group who identified and weighted key criteria to establish the
pre-cancer ED visit use predicted ED use after diagnosis. Methods: We Drug Assessment Framework (DAF). Strength of evidence (SOE) modifiers
identifiably linked adults from the tumor registry in the Dallas County safety-net deducted points for lower quality evidence. Through one-on-one meetings with
health system to a regional hospital database with claims-like data for all stakeholders, face and content validity of the DAF were established in an
patients from 98% of non-federal hospitals in North Texas. We applied a iterative process. Construct validity assessed the degree to which DAF scores
mixed-effects multivariate logit model, using frequent ED use ($4 visits) in the were associated with the pan-Canadian oncology drug review (pCODR) funding
6-12 months or 12-18 months before diagnosis to predict frequent ED use decisions and European Society for Medical Oncology Magnitude of Clinical
after diagnosis, adjusting for demographics, comorbidities; cancer type, stage, Benefit score (ESMO-MCBS, version 1.1). Sensitivity analyses were performed
initial treatment modalities; and grouping visits at the patient level. Results: Of on the final results. Results: The final validated DAF includes ten criteria:
8,610 adults diagnosed from 2012-2016, 76.2% had Medicaid or were overall survival, progression-free survival, response rate, quality-of-life, toxicity,
uninsured, 30.9% had lung, breast, or colorectal cancer, and 25.9% had unmet need, equity, feasibility, disease severity and caregiver well-being. The
advanced-stage cancer at diagnosis. In the 180 days after diagnosis, 42.5% of first five clinical benefit criteria represent 64% of the total weight. DAF scores
patients had zero ED visits, 45.7% had 1-3 visits, and 11.8% were frequent ED range from 0 to 300, reflecting both the expected impact of the drug and the
users ($4). In multivariate analysis, patients with frequent ED use in the 6- quality of the supporting evidence. When the DAF was retrospectively applied
12 months before a cancer diagnosis had 6.7 higher odds (95% CI: 4.8, 9.3) of to the last 60 drugs (in blinded fashion) reviewed by pCODR (2015-2018), the
having frequent ED use after diagnosis, compared to patients who had zero ED mean total DAF score was 94 (range, 18-179). Drugs with positive pCODR
visits prior to diagnosis. This compared to 1.3 higher odds (95% CI: 1.1, 1.5) of funding recommendation had higher DAF scores than drugs not recommended
frequent ED use if the patient had advanced-stage cancer, and 2.1 higher for reimbursement (103 vs. 63, t-test p = 0.0007). Funded drugs had fewer
odds (95% CI: 1.8, 2.4) if chemotherapy was part of initial treatment. Al- SOE points deducted than those that were not funded (median 0 vs. 24 points
though most post-diagnosis frequent ED users generated zero visits (62.2%) deducted, Wilcoxon p = 0.03). The correlation coefficient for DAF and ESMO-
or 1-3 visits (30.7%) in the 6-12 months prior to diagnosis, 38% of patients MCBS was 0.37 (95% CI, 0.10 to 0.59). Sensitivity analyses that varied the
with frequent ED use pre-diagnosis continued frequent ED visits after di- subjective criteria either positively or negatively did not change the results.
agnosis. Results were similar for ED use 12-18 months prior to diagnosis. Conclusions: Using a structured and explicit approach, a criterion-based
Conclusions: Among safety-net adults, prior ED-seeking behavior strongly valuation framework was designed and validated. The DAF can provide a
predicted ED use after a new cancer diagnosis. This may represent a high-risk transparent and consistent method to value and prioritize cancer drugs, in
group that might benefit from care delivery innovation. order to facilitate the delivery of affordable cancer care.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Health Services Research, Clinical Informatics, and Quality of Care 399s

6628 Poster Session (Board #319), Sat, 1:15 PM-4:15 PM 6629 Poster Session (Board #320), Sat, 1:15 PM-4:15 PM
Followup mammography after breast conservation therapy: Is 3D Tomosyn- Cost disparities with age in the treatment of advanced non-small cell lung
thesis(3DT) worth it? First Author: Shaakir Hasan, Department of Radiation cancer (NSCLC) in Ontario, Canada. First Author: Ying Wang, Department of
Oncology, Allegheny Health Network Cancer Institute, Pittsburgh, PA Medical Oncology, BC Cancer Agency, Vancouver, BC, Canada
Background: Screening three-dimensional tomosynthesis mammography (3DT) Background: Previous studies noted an association between age and cost of
is more cost-effective than two-dimensional mammography (2DM) for care in NSCLC. The drivers of these cost disparities have not yet been fully
detecting breast cancer, however cost-effectiveness as a follow-up for examined. We conducted a cost analysis study examining the differences in,
treated breast cancer is unknown. We retrospectively analyzed the down- and drivers of, costs of NSCLC care across age groups in Ontario, Canada.
stream workup and costs associated with 3DT compared to 2DM when Methods: We conducted a retrospective cohort study of patients diagnosed
employed as initial follow-up imaging in breast conservation therapy (BCT). in Ontario from Apr 1, 2007 to Mar 30, 2014, who received palliative
Methods: Between the years 2015-2017, 450 consecutive BCT patients chemotherapy for stage IV NSCLC. Variables of interest were extracted from
ages 32 – 89 with a follow-up 3DT (n = 162) or 2DM (n = 288) were reviewed registry data linked by the Institute for Clinical Evaluative Sciences (ICES).
in this IRB-approved study. The primary endpoint was further workup after The mean total cost of care including systemic therapy and supportive care,
follow-up mammogram and associated healthcare costs at 1 year. Down- was calculated in 2015 CAD dollars by fiscal year of diagnosis. Results: Of all
stream workup was secondarily tested for correlation with clinical and NSCLC cases diagnosed in Ontario (n = 37,786), 17,203 (45.5%) were de
treatment-related variables. A single 3DT cost an estimated $149 compared novo stage IV, of which 29.7% of patients received any chemotherapy for
to $111 for a 2DM, based on Centers for Medicare claims data Oncology Care their disease (n = 5,113), and 281 patients are presumed alive. In this
Model. Results: Patient clinical characteristics were : 6% DCIS, 10% T1a, population, median age was 65 to 69 years, 51.9% were male, 43.5% were
29% T1b, 35% T1c, 19% T2, 88% N0, 9% N1, 3% N2, 76% ER+/PR+/ adenocarcinomas, and 25.1% received second line chemotherapy. After
HerNeu2-, 12% TNBC, and 14% Her2Neu+. Whole breast radiation was adjusting for comorbidities, income, gender, year of diagnosis, and rural
given with conventional (59%) and hypo (39%) fractionation (81% with a versus urban living, the average lifetime costs per patient remains signifi-
boost), and 10% received accelerated partial breast irradiation. First post- cantly inversely related to age (p , 0.001). Belonging to the highest income
treatment mammogram was received within 3 months (20%), 3-6 months quintile (p = 0.006) and being diagnosed in more recent years (p , 0.001)
(32%), and after 6 months (48%) following RT. There were no differences in contributes significantly to increasing overall healthcare costs. Elderly pa-
breast density, patient age, T/N stage, receptor status, type of RT, or tients (80+) cost less (71%) and have shorter survival time (HR of death
mammogram timing between those in the 2DM and 3DT groups. The fol- 1.28, 95% C.I. 1.10 to 1.50) compared to younger patients (#45 years old).
lowing downstream workup ensued for 3DT compared to 2D imaging: 18% Accounting for longer survival in younger patients, the youngest group still
vs 29% short-interval (6-month) mammogram (OR = 1.83, P = 0.01), 6% vs incur a higher cost per day alive than other age groups ($471/day in #45
11% breast MRI (OR = 1.90, P = 0.08), 4% ultrasound for each, and 3% group, $301/day in . 85 group). Hospitalization accounts for ~30% of total
biopsy for each (1 positive in the 2D group). Including downstream workup, cost in both age groups. Chemotherapy accounts for 1% of total health care
the estimated cost per patient in the 3DT group = $249.00 compared to costs amongst the elderly (80+) age group and 10% of costs in #45 group.
$253.64 in the 2D group. With multivariable analysis the independent Conclusions: Our study shows that, despite everyone receiving systemic
predictors for reduced downstream workup was the use of 3DT and follow-up therapy in this patient population, younger patients incur significantly higher
mammogram at least 6 months after radiation (P , 0.05). Conclusions: Excess costs than elderly patients with advanced NSCLC, both before and after
workup was reduced with 3DT compared to 2DM in the post-treatment setting. adjusting for survival. While hospitalization accounts for the biggest com-
A single 3DT costs approximately 34% more than 2DM, however in this study ponent of total costs, patients with high income and more recent years of
the associated reduction in downstream workup with 3DT actually made it diagnosis drive the higher costs of care, and chemotherapy remains a driver
more cost-effective. of higher costs amongst younger patients.

6630 Poster Session (Board #321), Sat, 1:15 PM-4:15 PM 6631 Poster Session (Board #322), Sat, 1:15 PM-4:15 PM
Economic impact of the 31-gene expression profile test in the Medicare- Cost of oral cancer drugs: Trends in the United States Medicare population in
eligible population with cutaneous melanoma. First Author: Federico A. 2013 and 2015. First Author: Kira Rose-Madison Seiger, Harvard Medical
Monzon, Castle Biosciences, Inc., Friendswood, TX School, Boston, MA
Background: Sentinel lymph node biopsy (SLNB) is recommended as a Background: Rising pharmaceutical costs threaten affordability and access to
staging procedure for patients with cutaneous melanoma (CM), but SLNB is cancer care. Methods: The 2013 and 2015 Medicare Part D claims databases
associated with additional surgical risks and costs. The SLN positivity rate is were queried for CareFirst’s 2017 list of oral cancer drugs. Drugs with on-label
approximately 12-16% overall and varies by age. Older patients have lower indications for conditions other than cancer were excluded. Cost in 2013 were
SLN positivity rates despite a higher metastatic rate. A 31-gene expression adjusted for inflation to 2015 dollars according to the Bureau of Labor Sta-
profile (31-GEP) test classifies 5-year metastatic risk (Classes 1A [lowest tistics consumer price index. Results: The total cost of oral cancer drugs paid
risk], 1B, 2A, and 2B [highest risk]), including SLN metastasis. As previously by Medicare nearly doubled from 2013 to 2015, rising from over $4.8 billion to
demonstrated using a large multicenter database, patients with Class 1A, over $9.1 billion. The number of prescriptions increased by 21.6% from
T1-T2 (#2.0 mm thickness) CM have an overall SLN positivity likelihood , 640,193 to 778,357, suggesting that increased spending is driven more by
5%, below the NCCN guideline threshold for SLNB. The aim of this study was rising drug prices than by increased utilization. Cost per patient increased for
to determine the relative cost impact of patient management via traditional 81% (46) of the 57 drugs prescribed in both years and decreased for the
care versus using the 31-GEP test to guide SLNB and surveillance plans for remaining 19% (11). While only seven drugs cost more than $50,000 per
16,572 patients $65 years (Medicare-eligible) with T1-T2 CM who are patient in 2013, more than fifteen drugs cost more than $50,000 per patient
SLNB-eligible. Methods: Decision tree models were used to compare in 2015. The overall average cost per beneficiary increased by 56% from
management strategies (surveillance, SLNB) according to traditional care $7,521.64 to $11,734.68, and seven drugs more than doubled in per patient
based on SLNB staging and with 31-GEP test utilization. Models used T cost. Cost per patient increased by 4.5 times for imbruvica (ibrutinib), 3.5
category from AJCC TNM staging guidelines and SEER data for incidence times for targretin (bexarotene), 3.2 times for soltamox (tamoxifen citrate),
estimations. Model outcomes included total Medicare costs of melanoma 3.1 times for nilandron (nilutamide), 2.6 times for gilotrif (afatinib dimaleate),
treatment and patient management. Cost impact of the 31-GEP test was 2.6 times for cyclophosphamide, and 2.5 times for iclusig (ponatinib hy-
calculated based on the difference in Medicare costs between the traditional drochloride). Conclusions: Medicare costs for oral cancer drug nearly doubled
care and 31-GEP test paradigms. Results: Under the paradigm in which the from 2013 to 2015, largely due to increased per patient costs.
31-GEP guided SLNB and surveillance in patients $65 years with T1-T2 CM
2013 2015
who are eligible for SLNB, 11,157 fewer SLNBs would be performed per
year, a 67% reduction in SLNB procedures in this population. This would Total cost $4,815,303,459.27 $9,133,772,699.60
result in a net annual cost savings of $68M, a 31% reduction in total costs. Number of drugs* 61 74
Number of 640,193 778,357
Use of the 31-GEP testing to guide surveillance in patients $65 with T3-T4 prescriptions
melanoma, in addition to T1-T2 melanoma, were also analyzed and also Mean cost per $7,521.64 $11,734.68
demonstrated a reduction in costs. Conclusions: Results from this study beneficiary
suggest that using the 31-GEP test to guide SLNB and surveillance offers Mean total cost per $78,939,400.97 $123,429,360.81
substantial cost savings compared to traditional care for Medicare-eligible drug
Drugs with highest 1. Revlimid (lenalidomide), 2. 1. Revlimid (lenalidomide), 2.
patients. Medicare Gleevec (imatinib mesylate), 3. Gleevec (imatinib mesylate), 3.
spending Zytiga (abiraterone acetate), 4. Xtandi (enzalutamide), 4. Zytiga
Tarceva (erlotinib HCl), 5. Xtandi (abiraterone acetate), 5. Imbruv-
(enzalutamide) ica (ibrutinib)
*Brand names and generics, where applicable, were considered separately

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
400s Health Services Research, Clinical Informatics, and Quality of Care

6632 Poster Session (Board #323), Sat, 1:15 PM-4:15 PM 6633 Poster Session (Board #324), Sat, 1:15 PM-4:15 PM
Sustaining the gains in cancer care from the oncology care model. First Identifying and predicting the most costly patients in oncology. First Author:
Author: Valerie P Csik, Sidney Kimmel Cancer Center, Philadelphia, PA Mark Raymond Waddle, Mayo Clinic, Jacksonville, FL
Background: The Oncology Care Model (OCM) is a 5-year demonstration project Background: Quality based payment programs in medicine are currently being
led by the Centers for Medicare and Medicaid Services (CMS) to create a introduced nationally, aimed to improve care and reduce cost. This study
framework for the future of oncology care in the United States. More than half aimed to evaluate the top spenders (TS) after cancer diagnosis and predict TS
way through the project, our large, urban NCI-designated cancer center chose to at two separate time points using predictive analytics. Methods: Patient
focus on and invest in resources and personnel for patient navigation and the characteristics, cancer details, treatments, adverse events, and outcomes were
development of clinical pathways. Although navigation has shown to reduce collected for patients treated for cancer at Mayo Clinic from 2007 - 2017.
emergency department (ED) visits by as much as 6% per quarter compared to Standardized costs over a 2 year period after first treatment were obtained from
non-navigated patients, sustaining it is a challenge because it is a nonbillable the Mayo Clinic Cost Data Warehouse with Medicare reimbursements assigned
service. Clinical pathways are a tool to reduce care variation by addressing drug to all services and adjusted to the 2017 GDP Implicit Price Deflator for in-
expenditures, and represent an opportunity to reduce outpatient costs by as flation. TS were identified as those with greater than 93rd percentile costs
much as 35% when patients are treated on pathway.3 Many OCM practices ($$113,158) due to a sharp rise in cost at that level. Descriptive statistics and
made similar investments and all are facing the question: How will the in- univariate analysis were used for comparison. A prediction model with a
frastructure and efforts developed during OCM be sustainable after the dem- training (80%) and validation set (20%) using multivariate selection was used
onstration project ends? Methods: An analysis of average ED cost and utilization to predict TS and was repeated using information available at 1) the time of
as well as drug expenditures was conducted using OCM feedback data (Q1-Q8). consultation and 2) at last follow-up. Results: A total of 80,385 patients were
Total utilization of ED visits and ED admits were used to determine a projected included and 5,626 TS were identified. Mean cost (25th, 75th percentile)
annualized cost which was compared to a budgeted navigation team. Similarly, overall was $44,953 ($16,776, $51,889). Prediction models at time 1 and 2
projected annualized drug expenditures were compared to the annual cost of the had ROC AUC statistics of 0.82 and 0.89 in training and 0.82 and 0.88 in the
pathways tool. Results: We found that ED visits and ED admits would need to be validation indicating good prediction of high costs. Factors most predictive of
reduced by 11% to cover navigation costs. Similarly, a 0.7% reduction in total TS included need for blood transfusions within 90 days of treatment (OR 5.3),
drug expenditures would cover the cost of clinical pathways. The OCM data bone marrow transplant (OR 4.0), mild liver disease (OR 3.5), hemiplegia (OR
represents a timeframe prior to implementation of these programs and an 3.4), weight loss . 10% within 90 days of treatment (OR 3.3), upper GI cancer
average increase of 1.6% per quarter for ED admits, a 0.6% decrease in ED (OR 3.0), ‘other’ cancer type (OR 2.8), immunotherapy use (OR 2.7), and
visits and 2.7% increase in drug expenditures. This will serve as a baseline to hospitalizations within 90 days (OR 2.4), all p , 0.001, among others. The
measure progress towards our sustainability targets. Conclusions: Long term largest costs were due to hospital services in the TS and non-TS groups, mean
sustainability of the infrastructure developed during OCM to support cancer care costs $114,258 and $13,185 respectively. Conclusions: This is the first study
transformation will be dependent on reducing high cost and highly utilized to predict with high accuracy the top spenders in Oncology. Patient comor-
services. Aligning impact areas with resources/tools to ensure sustainability is an bidities and toxicities were among the strongest predictors of high costs, along
approach that can help define targets for OCM practices. with certain treatments (bone marrow transplant and immunotherapy). Our
findings suggest that quality payment programs should adjust for comorbid-
ities, and that reducing toxicity may be an effective method at reducing costs.

6634 Poster Session (Board #325), Sat, 1:15 PM-4:15 PM 6635 Poster Session (Board #326), Sat, 1:15 PM-4:15 PM
Clinical outcomes of patients with stage IV cancer receiving immune check- The effect of guideline-concordant novel therapy use on meeting cost targets
point inhibitors in the inpatient setting. First Author: Sienna Durbin, Harvard in OCM: Results from a large community oncology network. First Author:
Medical School, Boston, MA Stephen Matthew Schleicher, Tennessee Oncology, Nashville, TN
Background: Immune checkpoint inhibitors (ICI) represent a major leap in Background: The Oncology Care Model (OCM) is intended to incentivize phy-
the treatment of many cancers. Use has rapidly expanded in recent years, yet sicians to improve the quality and reduce the cost of cancer care. In OCM,
it is unknown whether hospitalized patients, who are often sicker than those providers are accountable for all costs during six month episodes of care relative
who were studied in clinical trials, derive benefit from ICI. The primary to target costs (TC) derived from a baseline spending period (BSP; 2013-2015).
objectives of this study were to characterize the clinical features and out- This accountability is intended to foster care coordination to reduce preventable
comes of inpatients receiving ICI at a single institution, and to identify emergency department visits and hospitalizations (EDH). Benefits of reducing
predictors of survival. Methods: After IRB approval, we conducted a retro- EDH may be diluted when new treatment indications for costly immunotherapies
spective chart review of inpatients with Stage IV solid tumors receiving ICI (IO) are introduced into clinical practice after BSP. Methods: We identified all
between 2015 – 2018 at a tertiary care referral hospital. Patients receiving non-small cell lung cancer (NSCLC) and bladder cancer (BC) OCM episodes
ICI on clinical trial were excluded. We examined the clinical characteristics, attributed to Tennessee Oncology (TO), a large community oncology network of
readmission rate, and post-discharge survival. We then conducted a Cox over 90 oncologists, during performance period 2 (PP2; the most recent PP with
multivariable regression analysis to identify predictors of post-discharge available data). We selected NSCLC and BC because both diseases have IO
survival. Results: A total of 103 patients with Stage IV solid tumors were indications that became standard of care after BSP. Using claims data analytics
treated with ICI as inpatients between 2015 – 2018. Average age was 57 software, we identified all NSCLC and BC episodes with spending above TC, and
years (range = 26 to 85); 57% were male; 27% had ECOG performance found a subset of these above target episodes (ATEs) without any EDH that
status (PS) 3-4; average Charlson Comorbidity Index score was 8.3. Most remained above TC due to IO use. Two medical oncologists reviewed these cases
common tumor types were melanoma (35%) and lung (22%). Seventy-six in duplicate to assess guideline concordance of IO. Results: During PP2 there
percent began ICI as an inpatient and 24% received ICI as continuation of were 2,623 OCM episodes attributed to TO, including 240 NSCLC and 31 BC
episodes. Spending was above TC in 118 (49%) and 13 (42%) of NSCLC and BC
outpatient therapy. Seventeen percent experienced an immunotherapy re-
episodes, respectively. For these NSCLC and BC ATEs, EDH was prevented in 62
lated adverse event, most commonly colitis and pneumonitis. The 30 day
(53%) and 5 (38%) of cases, respectively. In NSCLC and BC ATEs without EDH,
readmission rate was 41%. The median post-discharge survival was 31 days;
43 (69%) and 5 (100%) of episodes included IO, respectively. Clinician review
47% of patients died during admission or within 30 days of discharge; 14%
in duplicate (S.M.S.; C.A.W.) found that the use of IO was NCCN guideline
survived more than 6 months. Factors predictive of shorter post-discharge concordant in 33 (77%) and 4 (80%) of these NSCLC and BC cases, respectively
survival were PS of 3-4 relative to PS 0-2 (HR 2.0, p , 0.004), and lung (K = 0.87). Conclusions: Guideline-concordant use of expensive IO as its
cancer (HR 2.0, p , 0.024) and other tumor types (HR 2.1, p , 0.004) treatment indications expand poses substantial challenges to meeting cost
relative to melanoma. Conclusions: While the majority of inpatients receiving targets in OCM, even when practices prevent EDH.
ICI died during admission or within 30 days of discharge, a subset of patients
with stage IV disease were alive at 6 months. Tumor type and ECOG PS # (%) ATEs w/o
predict post-discharge survival and may be used to identify inpatients more # # (%)ATE w/o EDH* but include % guideline
likely to benefit from ICI. These novel findings, which are unique to a single Disease episodes # (%)ATE EDH* IO concordant
institution, require additional validation. NSCLC 240 118 62 (53%) 43 (69%) 77%
(49%)
BC 31 13 5 (38%) 5 (100%) 80%
(42%)
*also without inpatient post-acute care

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Health Services Research, Clinical Informatics, and Quality of Care 401s

6636 Poster Session (Board #327), Sat, 1:15 PM-4:15 PM 6637 Poster Session (Board #328), Sat, 1:15 PM-4:15 PM
Clinical and economic outcomes of pegfilgrastim in metastatic colorectal Burden of cytokine release syndrome (CRS) and neurologic events (NE) in
cancer. First Author: Lalan S. Wilfong, Texas Oncology, Dallas, TX patients (Pts) with relapsed/refractory non-Hodgkin lymphoma (NHL) receiving
lisocabtagene maraleucel (Liso-cel; JCAR017) in TRANSCEND NHL 001.
Background: Febrile neutropenia (FN) resulting from myelosuppressive che-
First Author: Jeremy S. Abramson, Massachusetts General Hospital Cancer
motherapy can lead to increased hospitalizations and mortality. Pegfilgrastim
Center, Boston, MA
can be used to reduce the risk of FN; however, few studies address pegfil-
grastim’s value in patients with metastatic solid tumors. This observational Background: We assessed incidence, healthcare resource utilization (HCRU), and costs of
study compared outcomes for pegfilgrastim-treated (peg-tx) and peg-untreated CRS and NE in pts in TRANSCEND receiving liso-cel, an investigational, anti-CD19 CAR
(no peg) patients with metastatic colorectal (CRC) at US Oncology practices T cell product administered as a defined composition of CD4+/CD8+ CAR T cells.
Methods: HCRU within the dates of onset and resolution of CRS and NE was identified from
(USO) participating in the Oncology Care Model. Methods: Patients with case report forms and trial management guidelines. Costs associated with grade (G) of CRS
metastatic CRC treated at USO from July 1, 2013 – December 31, 2014 and a or NE were applied using public databases and literature on national average costs from a
qualifying baseline OCM episode were included. Propensity scoring was used provider perspective. Results: Of 102 pts, 21 (21%) had CRS only (no G$3), 6 (6%) had
to match (1:2) peg-tx and no peg cohorts based on line of therapy, number of NE only (no G$4), 6 (6%) had nonconcurrent CRS and NE, and 11 (11%) had concurrent
comorbidities, age, gender, ECOG performance status, chemotherapy neu- CRS and NE (including 1 G4 CRS). Most pts had G#2 events (31/44, 70%). Of pts with
tropenic fever risk, and dose reduction. Outcomes assessed included all-cause both CRS and NE, 16/17 (94%) had CRS first. Of pts with nonconcurrent events, NE
and infection-related hospitalization rates; total cost of care per 6-month OCM developed a median of 2 days after CRS resolved. Of 10/11 pts with concurrent events and
CRS before NE, NE developed a median of 3.5 days after CRS onset. 11/38 pts with CRS
episode; and overall survival (OS). Results: Matched peg-tx and no peg
(29%) received tocilizumab and 21/44 with CRS and/or NE (48%) received dexameth-
samples were 149:298. The all-cause hospitalization rate was higher in the asone. Overall median costs for G#2 vs G$3 CRS and/or NE were $16,479 vs $70,549.
peg-tx vs. no peg population, 45% vs. 32% (OR 1.7, (1.1, 2.5), p = 0.011). Management costs for CRS and/or NE were largely driven by hospitalizations (Table), and
Infection-related hospitalization rates were no different in peg-tx vs. no peg ranged $177–24,947 for G#2 CRS or NE only, $30,749 for G#2 concurrent CRS and NE,
cohorts, p = 0.367. Cost of care was significantly higher for peg-tx patients vs. and $43,974–264,149 for G$3 CRS and/or NE. Conclusions: In TRANSCEND, in-
no peg ($58,787 6 $20,490 vs. $37,811 6 $19,593 respectively, p , cidences of G$3 CRS and NE were low. G$3 events resulted in a 328% increase in overall
.001). OS was 19.5 months (peg-tx) vs. 19.7 months (no peg), p = 0.882. median costs vs G#2 events. Clinical trial information: NCT02631044.
Conclusions: While peg use in curative treatment settings for high risk patients HCRU and median costs per category for each pt subset.
is standard of care, in our Medicare population use in metastatic CRC did not Median
Length
result in a lower all-cause or infection-related hospitalization rate or impact in Severity of Stay, Diagnostics, Drugs,
OS. There was a higher 6-month total cost of care associated with those Gradea n ICU, n (%) days US$ US$ Facility, US$ Total, US$
patients who received peg during chemotherapy. CRS only 1|2 13 | 8 0 | 2 (25) 5|6 70 | 249 297 | 167 13,341 | 13,707 |
16,009 16,426
NE only 1|2 1 | 3 0 | 1 (33) 0|9 63 | 527 0 | 408 114 | 24,013 177 | 24,947
3 2 2 (100) 17 1390 689 76,987 79,066
Non- CRS & NE 2 0 6 771 2174 16,180 19,125
concurrent #2
CRS & NE
CRS or NE 4 1 (25) 16 861 423 42,690 43,974
$3
Concurrent CRS & NE 4 0 10.5 606 2127 28,015 30,749
CRS & NE #2
CRS or NE 6 3 (50) 16 1450 8711 75,653 85,814
$3
CRS & NE 1 1 (100) 56 3407 10,498 250,244 264,149
$3
a
Lee Criteria (CRS); CTCAE 4.03 (NE).

6638 Poster Session (Board #329), Sat, 1:15 PM-4:15 PM 6639 Poster Session (Board #330), Sat, 1:15 PM-4:15 PM
Clinical benefit and prices of cancer drugs in the United States and Europe. A cost-utility analysis of atezolizumab in the second-line treatment of
First Author: Kerstin Noëlle Vokinger, Harvard Medical School Program on metastatic urothelial carcinoma. First Author: Ambika Parmar,
Regulation, Therapeutics, and Law, Boston, MA Sunnybrook Health Sciences Centre, Odette Cancer Centre, University of
Toronto, Toronto, ON, Canada
Background: Given rising cancer drug costs, Medicare recently proposed to tie
some US drug prices to average prices paid by comparable countries. To Background: Despite early promising results, IMvigor211 failed to demon-
understand the potential scope of this policy, we assessed differences in strate an overall survival benefit for atezolizumab, compared to chemo-
cancer drug prices in the US and selected European countries. We also therapy, in the second-line treatment of metastatic urothelial carcinoma.
evaluated the correlation between drug prices and their clinical benefit, as However, given improvements in adverse events (AE) and quality of life with
measured by two value frameworks: the American Society of Clinical Oncology atezolizumab, there may still be investment value. We conducted a cost-
Value Framework v2 (ASCO VF) and the European Society for Medical On- utility analysis (CUA) of atezolizumab compared to chemotherapy from a
cology Magnitude of Clinical Benefit Scale v1.1 (ESMO-MCBS). Methods: We public-payer healthcare perspective. Methods: We developed a partitioned
identified all new drugs for adult solid and hematologic cancers, approved by survival model to evaluate atezolizumab versus chemotherapy (i.e. doce-
the FDA from 2009-2017 and that have also been approved by the EMA by taxel, paclitaxel or vinflunine). IMvigor211 informed rates of treatment
December 31, 2018. US average sales prices (and if not available, wholesale receipt (initial and subsequent), AE, effectiveness and utility estimates. Cost
acquisition costs) were extracted as of February 1, 2019, and compared to for treatment, AE and death were based on published literature (adjusted to
comparable currency-adjusted ex-factory drug costs in England, France, 2018 Canadian dollars). Per health state, cost of treatment (initial and
Germany, and Switzerland. ASCO VF and ESMO-MCBS scores were assessed subsequent) and AE were incorporated. Outcomes included quality-adjusted
for pivotal trials supporting solid tumor drugs; in case of multiple trials, we life-years (QALY), cost per treatment, and incremental cost-effectiveness
focused on the highest score. Consistent with the developers of the rating ratio (ICER). QALY and cost were discounted at 1.5% (Canadian guidelines).
scales, “high benefit” was defined as scores of A-B (neo/adjuvant setting) and Parameter uncertainty was assessed through one-way and scenario analyses.
4-5 (palliative setting) on the ESMO-MCBS scale and scores$45 on the ASCO Time horizons of 2 (within trial) and 5 years (extrapolated lifetime) were
VF. Linear regression models and non-parametric Kruskal-Wallis test and were evaluated. Results: QALY of atezolizumab and chemotherapy over 2 years
used to assess the association between drug prices and benefit scores. (lifetime) were 0.65 (0.93) and 0.58 (0.64), respectively. Cost of atezoli-
Results: The study cohort included 63 drugs approved by the FDA and the EMA zumab and chemotherapy over 2 years (lifetime) was $77,614.64,
during the study period. 46 (73%) were approved for solid tumors, and 17 ($92,484.34), and $62,212.35 ($67,606.65), respectively. ICER over 2
(27%) were approved for hematologic malignancies. Overall, median cancer years and lifetime was $220,032.71/QALY and $85,785.14/QALY. Sce-
drug prices in included European countries were 52% (interquartile range: 37- nario analysis from a North American perspective with only taxane che-
72%) lower than US prices. There was no statistically significant association motherapy with and without third line immunotherapy (IO) revealed an ICER
between monthly treatment cost and ASCO-VF or ESMO-MCBS scores in any of $80,144.90/QALY, $125,332.76/QALY over a lifetime horizon, re-
country. There was also no association between price differential between US spectively. Conclusions: The difference in ICER dependent on time horizon,
and median European drug prices and either ASCO-VF (P = 0.599) or ESMO- driven by extrapolated survival benefits, highlights the importance of long-
MBCS (P = 0.321) scores. Conclusions: Cancer drug prices in the US were term follow-up to examine whether early evidence for durable response
significantly higher than in the compared European countries. Drug prices of translates into long-term survival and improvements in cost-effectiveness.
cancer drugs were not associated with clinical benefit in the US or in European Thus, CUA of IO require careful interpretation and warrant dynamic as-
countries. sessment as new data becomes available.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
402s Health Services Research, Clinical Informatics, and Quality of Care

6640 Poster Session (Board #331), Sat, 1:15 PM-4:15 PM 6641 Poster Session (Board #332), Sat, 1:15 PM-4:15 PM
Desire to address costs at time of treatment decisions among patients with Drug funding price negotiations: Towards achieving a balance between
metastatic breast cancer. First Author: Yvonne Y. Lei, Massachusetts General individual and population gains in health benefits. First Author: Ambika
Hospital, Boston, MA Parmar, Sunnybrook Health Sciences Centre, Odette Cancer Centre, University
of Toronto, Toronto, ON, Canada
Background: Costs of cancer care may impact access to therapy, adherence,
and distress among patients. However, the degree to which patients with Background: Drug price negotiation to lower cost to a cost-effectiveness
metastatic breast cancer (MBC) wish to discuss financial issues when making threshold (l) is a recognized approach to improve health care opportunities
treatment decisions is unknown. Methods: In a single arm feasibility trial, 40 for the greater benefit of the population. Critics have raised concerns for this
women with newly diagnosed or progressive MBC completed a 1-page survey approach given the additional time required and speculated loss of quality-
regarding goals and priorities for discussion with the oncology team. The survey adjusted life-years (QALY) for patients. The current study aimed to quantify the
included 17 potential priorities for discussion in the domains: treatment incremental net health benefit (INHB) of publicly funded oncology drugs, if
options, symptom management, emotional concerns, planning for the future, funding occurred at list prices without (w/o) negotiations. Methods: Pan-
and lifestyle. We evaluated participants’ interest in prioritizing discussion of Canadian Oncology Drug Review submissions were reviewed to identify
financial issues and sociodemographic and clinical correlates of this prefer- funded drugs with unique indications. For included drug indications eco-
ence. We examined the relationship between desire to discuss financial issues nomic guidance panel (EGP) reports were reviewed for incremental costs
and both distress on the Distress Thermometer (DT) and satisfaction with (DC) and DQALY from manufacturer’s base case cost-effectiveness analyses,
cancer care using Fisher’s exact test. Results: Among 40 participants, 11 EGP lower (LL) and upper limit (UL) re-analyzed estimates (based on list
(28%) reported interest in discussing financial issues when making treatment prices). Number of new cases in Ontario (most populous province in Canada)
decisions, 29 (72%) were not interested. Average age was 57 (range 31-73), per indication (2017-18) was obtained from provincial databases. Annual
and the majority were white (85%) and college graduates (66%). Only 18% of QALY gain for each indication was determined by: (DQALY3cases). Pro-
white patients were interested in addressing cost, while 83% of non-white vincial QALY gain/loss w/o price negotiations to reference l was estimated
patients were interested (p , 0.01). Those with a college education were less by: (INHB= [DQALY2 (DC/l)] 3cases). Incremental net monetary benefit
likely to prioritize financial discussion compared to no college (16% vs. 47%, demonstrated annual monetary gain/loss w/o price negotiations to reference
p = 0.04). Patients interested in discussing cost were more likely to have a l: (INMB= [(DQALY3l) 2DC] 3cases). Results: 34 drug indications in-
household income , $50,000 (50% vs. 22% . $50,000, n.s.) and to have cluding 4,629 new cases were identified. Annual QALY gain for funded
Medicaid (50% vs. 25% other insurance, n.s.). Additionally, patients with indications using manufacturer, LL and UL estimates was 1,851, 1,617 and
higher levels of distress (35% vs. 21% DT , 4, n.s.) and those on novel 1,301, respectively. At reference l CAD$100,000/QALY, funding w/o ne-
targeted or biologic therapy (42% vs. 21% other therapy, n.s.) were more likely gotiations resulted in loss of 2,176, 2,368, 2,451 QALY, representing bud-
to prioritize discussion of costs. Desire to discuss cost was not related to getary diversions away from other health care interventions. This would result
satisfaction with care. Conclusions: A substantial minority of patients with in a provincial net annual loss of 325, 751 and 1,150 QALY, respectively.
MBC, particularly those from less advantaged backgrounds, wish to discuss INMB demonstrated provincial net monetary loss of CAD$32,472,389,
financial issues at time of treatment decisions. Financial toxicity research $75,113,684 and $115,022,331, respectively. Conclusions: Despite an
should recognize that not all patients desire this discussion and evaluate annual gain in QALY for funded drug indications, a net provincial loss in QALY
methods to screen for financial concerns and barriers to care. w/o price negotiations was demonstrated. Thus, supportive evidence exists for
drug price negotiations towards the promotion of health benefits for the
population.

6642 Poster Session (Board #333), Sat, 1:15 PM-4:15 PM 6643 Poster Session (Board #334), Sat, 1:15 PM-4:15 PM
Cost-minimization analysis of using tumor cell-free DNA as monitoring tool in Progression free survival (PFS) benefits of immuno-oncology agents (IOs)
cancer immunotherapy. First Author: Alexander Kuhlmann, Leibniz Universität and implications for market cost-sharing inefficiencies. First Author:
Hannover, Center for Health Economics Research Hannover (CHERH), Syed Muhammad Mushtaq Ashraf, University of Toledo Medical Center/
Hannover, Germany Dana Cancer Center, Toledo, OH
Background: Albeit showing great benefit in individual cancer patients, only Background: Cancer care expense has high financial toxicity with 6 in 10
the minority of patients benefit from checkpoint inhibitor immunotherapies patients reporting distress about finances. System wide cost has also increased
(IMTs). Mutation load and PD-L1 staining are used for response prediction, with new treatment approvals; from 2011 to 2016 FDA approved 52 cancer
but are imperfect predictors. A universal method to quantify tumor cell-free treatments. During this time, US healthcare spending increased from $26.8
DNA (TcfDNA) enables the early and effective evaluation of individual IMT billion to $42.1 billion. Lung cancer has high mortality- in 2018 American
efficacy[1], by comparing TcfDNA to pre-therapeutic values. Here we present a Cancer Society estimated 234,030 new cases with 50% likely to die from the
health economic evaluation of test usage. Methods: This cost-minimization disease. Hence new treatments are important societal need but costs are
study determines the economic efficiency of TcfDNA-test from the perspective concerning. Methods: We selected Phase 3 trials of IO based regimens for non-
of the statutory health insurance in Germany. The assumption is that the small cell lung cancer (NSCLC) in first line setting, approved by FDA in 2018.
effectivity of the intervention (TcfDNA monitoring) and of the comparator (no Median PFS was compared between intervention and no-intervention group
test), is comparable with regards to the patient-relevant-endpoints (morbidity, (NIG). Total costs were evaluated using 2018 Medicare reimbursements rates.
mortality, quality of life). The model simulates the course of treatment for each Cost and PFS were converted to represent expenditures accrued for 12 months
patient with and without TcfDNA testing, calculating the respective cost. (mths) PFS gain. Results: Study 1 (NCT02125461) for stage 3 NSCLC; looked
Treatment details, outcome (RECIST) and TcfDNA results are derived from an at maintenance Durvalumab after platinum doublet chemoradiation induction.
earlier clinical trial.[1] Costs are obtained from publicly accessible data bases. Median PFS 17.2 vs 5.6 mths in NIG. 1 year PFS cost $155,391 vs $27,477.
Two testing strategies are explored. Strategy 1: Testing all patients only before Study 2 (NCT02578680) for stage 4 NSCLC; looked at maintenance Pem-
the second cycle. Strategy 2: Same as strategy 1 plus a confirmation test brolizumab, Pemetrexed or Pemetrexed alone after platinum doublet in-
before the third cycle, in patients with initial result in a defined grey zone. duction. Median PFS 8.8 vs 4.9 mths in NIG. 1 year PFS cost $283,949 vs
Results: Both testing strategies correctly classified 68% of progressive disease $116,159. Study 3 (NCT02775435) for stage 4 NSCLC (squamous); looked
patients. Testing strategy-1 misclassifies about 8% of disease control patients, at maintenance Pembrolizumab after platinum doublet induction. Median PFS
which would lead to discontinuation of successful therapies in those. In 6.4 vs 4.8 mths in NIG. 1 year PFS cost $195,820 vs $119,146. Study 4
contrast strategy-2 correctly classifies all patients with disease control, whilst (NCT02366143) for stage 4 NSCLC; looked at maintenance Atezolizumab,
requiring only 24% more tests, since ~50% of patients are undoubtedly Bevacizumab or Bevacizumab alone after platinum doublet induction. Median
classified after one cycle. Assuming six cycles at average costs of 4,781 EUR in PFS 8.3 vs 6.8 mths in NIG. 1 year PFS cost $316,499 vs $199,278.
the no-test-setting, average costs savings per patient with strategy-1 are 5,735 Conclusions: Median PFS doubled from IOs in NSCLC but costs accrued, more
EUR (20% of treatment costs), with strategy-2 are 2,010 EUR (7%) compared than doubled. The expense has to be quantified in relation to per capita GNP in
to no testing. Conclusions: TcfDNA monitoring is a cost-saving strategy. US- $53,128 in 2017. Accepting even a higher threshold of $100,000 for 1
However, a confirmatory strategy is desirable to avoid early discontinuation of quality adjusted life year gained, costs appear economically unsustainable.
successful IMT. [1] Weiss GJ et al.: Clin Cancer Res: 2017;5074-81. But PFS gains are significant indicating need to enable IOs utilization, by
making innovative cost sharing platforms.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Health Services Research, Clinical Informatics, and Quality of Care 403s

6644 Poster Session (Board #447), Sat, 1:15 PM-4:15 PM 6645 Poster Session (Board #446), Sat, 1:15 PM-4:15 PM
Cost-effectiveness of CHOP in treating DLBCL in Malawi. First Author: Cost-minimization analysis for biosimilar pegfilgrastim in the prophylaxis of
Matthew Painschab, Lineberger Comprehensive Cancer Center; University of chemotherapy induced (febrile) neutropenia and expanded access based on
North Carolina, Chapel Hill, NC budget neutral basis. First Author: Weijia Wang, Sandoz Inc., Princeton, NJ
Background: DLBCL is common in Africa, and often curable, but treatment Background: Pegfilgrastim (pegfil) injection is indicated to decrease the in-
costs and cost-effectiveness are key considerations. WHO defines extremely cidence of infection, as manifested by febrile neutropenia, in patients with
cost-effective interventions as having an incremental cost-effectiveness ra- non-myeloid malignancies receiving myelosuppressive anti-cancer drugs as-
tio (ICER) , GDP per capita. Methods: We used a decision tree model to sociated with a clinically significant incidence of febrile neutropenia. As-
conduct a cost-effectiveness and budget impact analysis from a health systems suming biosimilarity of biosim-pegfil and pegfil in the prevention of febrile
perspective in Malawi (2017 GDP per capita $340). Comparisons were made neutropenia, we estimated cost minimization of conversion from pegfil to
between CHOP vs. palliative care with diagnosis (PC+D), and palliative care biosim-pegfil and subsequent potential expanded access due to cost savings.
without diagnosis (PC-D). Microcosting and clinical outcomes were derived Methods: A cost minimization model was conducted based on a hypothetical
from published prospective data. Costs reflect treatment and 2 years of follow- panel of 20,000 patients using average selling price (ASP) for one chemo-
up. Outcomes reflect a lifetime time horizon. Life expectancies were derived therapy cycle. ASP was obtained from 1Q 2019 payment allowance limits. The
from UNdata, and disability-adjusted life year (DALY) weights from the Global simulation included two steps: 1) cost minimization was calculated per cycle
Burden of Disease Study. Costs were analyzed in 2017 US $, and costs and (biosim-pegfil and pegfil is administered 1 dose per cycle) when patients were
outcomes were discounted at 3% annually. Annual estimates for new DLBCL converted to biosim-pegfil from pegfil on ratio of 10% to 100% at 10% in-
cases (n=161) were used as input incidence. Probablistic sensitivity analysis tervals and at a discount from 15% to 35% at 5% intervals, and 2) expanded
was conducted using Crystal Ball software over 1000 simulations. Results: For access to biosim-pegfil was calculated based on budget neutrality. Since pegfil
the base case, the ICER of CHOP versus PC+D is $150/DALY averted, and has two forms of availability (Neulasta and Neulasta-Onpro) with the same
versus PC-D is $200/DALY averted (Table). The ICER was stable across a wide price, results were to either conversion scenario. Results: Per-cycle per-patient
range of sensitivity analyses. The ICER varied most across the range of cost minimization from converting pegfil to biosim-pegfil ranged from
progression-free survival estimates ($117-209), and range of costs for CHOP $702.27 (15% discount) to $1,638.63 (35% discount). For 20,000 patients,
plus follow-up ($71-308). CHOP was extremely cost-effective by the WHO this yields savings of over $14 million (15% discount) to $32 million (35%
definition in 99% of simulations versus PC+D, and 94% of simulations versus discount) at 100% conversion rate. When half the patients were converted to
PC-D. In the base case, total annual cost of DLBCL treatment with CHOP in biosim-pegfil, savings could range from . $7 million (15% discount) to . $16
Malawi was $306,221. Conclusions: This is one of the first rigorous cost- million (35% discount). With 100% conversion rate and 15% discount, 3,529
effectiveness and budget impact analyses for cancer treatment in a low-income additional patients could be treated with the savings generated. At 50%
country. CHOP is extremely cost-effective compared to palliative care, with conversion rate, cost savings could be applied to another 1,765 patients with
~$300,000 needed annually to treat all DLBCL cases in Malawi. These 15% discount, 3,333 patients with 25% discount, and 5,385 patients with
findings merit external validation, and support continued regional investments 35% discount, respectively. Conclusions: Conversion from pegfil to biosim-
in cancer care. pegfil can lead to potential cost savings and these savings can be applied to
offer increased access to supportive care with biosim-pegfil for patients re-
Incremental Incremental Incremental ceiving chemotherapy on a budget-neutral basis. For payers with larger
Treatment Life Total $/death $/Life year $/DALY
Alternative Deaths Years DALYs Costs averted gained averted
populations, savings can be substantial. More studies are warranted to eval-
uate such potential cost savings due to use of biosim-pegfil over reference
CHOP 97 1369 1279 306,221 -- -- -- pegfil.
PC+D 161 40 19 117,098 2,934 142 150
PC-D 161 40 19 53,958 3,913 190 200

6646 Poster Session (Board #445), Sat, 1:15 PM-4:15 PM 6647 Poster Session (Board #444), Sat, 1:15 PM-4:15 PM
Scope of practice of advanced practice providers (APP) in US community National cancer expenditure analysis in the United States Medicare pop-
oncology. First Author: Andrew Klink, Cardinal Health, Dublin, OH ulation, 2013. First Author: Emily S Ruiz, Department of Dermatology,
Brigham and Women’s Hospital, Harvard Medical School, Boston, MA
Background: Oncology practices are increasingly employing nurse practi-
tioners (NPs) and physician assistants (PAs) known collectively as advanced Background: Cancer is the second leading killer in the United States, but
practice providers (APPs) to improve practice workflow, increase efficiency, there is no comprehensive analysis evaluating total cancer costs by cancer
and enable physicians to focus on complex patient care. Understanding and treatment modality. Methods: Data from the 2013 inpatient (100%),
variations in scope of practice for APPs may help establish a benchmark outpatient (100%), and carrier (5%) Medicare Limited Data Set Standard
against which future changes are measured. Methods: US community phy- Analytic Files were queried for claims filed for International Classification of
sicians responded to a web-based survey from Sep to Nov 2018. Physicians Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes for all
were asked how frequently their APPs performed certain tasks on a 5-point cancer diagnoses. Claims associated with the ICD-9 codes for professional
scale (i.e., never, occasionally, sometimes, frequently, and always). Responses fees were analyzed by Current Procedure Terminology (CPT) code to de-
have been summarized using descriptive statistics. Results: In this study, 163 termine resource allocation by disease. Outpatient oral chemotherapy costs
physicians were surveyed, most (81.0%, n = 132) used APPs in their practice. were obtained from the National Average Drug Acquisition Cost database for
Among physicians using APPs, 91.7% (n = 121) used NPs and 49.2% (n = 65) 50 drugs used for cancer treatment in 2013. Results: $27.9 billion was
used PAs. Most physicians stated that APPs were frequently/always involved in spent on cancer treatment of which $12 billion (43%), $6.5 billion (23%),
providing patient education (84.1%), ordering imaging and laboratory studies and $1.1 billion (4%) was allocated to professional, inpatient facility, and
(68.9%), and/or making supportive care decisions (62.1%). Over 85% outpatient facility fees, respectively. Oral and hospital-based chemotherapy
(57.6%-59.8% occasionally/sometimes; 28.0%-28.8% frequently/always) of accounted for 17% ($4.7 billion) and 15% ($4.3 billion) of all cancer costs,
physicians agreed that APPs discussed imaging reports and end of life (EOL) respectively. Lung/thoracic cancer ($2.9 billion) had the highest total an-
care (57.6% occasionally/sometimes, 28.8% frequently/always) with patients. nual cost and multiple myeloma ($9,019, SD $19,962) is the most ex-
Regarding procedures: 51.9% (28.0% occasionally/sometimes; 24.1% pensive cancers to treat annually per patient. Average out-of-pocket
frequently/always) responded that APPs performed bone marrow biopsies and expenses are $470 (SD 287), which is 12% of the annualized per patient
intrathecal chemotherapy. Regarding systemic therapy: 68.2% (58.3% oc- costs. Conclusions: In 2013, cancer accounted for only 5% of Medicare
casionally/ sometimes; 9.8% frequently) allowed APPs to modify existing spending. Chemotherapy costs accounted for approximately a third of total
regimen e.g., dose/schedule change; 39.4% responded that APPs made spending and are likely to rise as innovation continues. Out-of-pocket
decisions about new therapy selection. Conclusions: While substantial vari- spending can place a financial burden on cancer patients requiring more
ation in the role of APPs in community oncology practices was observed, similar expensive treatments.
themes emerged. APPs appear to be integral in patient education, ordering
laboratory and imaging studies, and discussing EOL care. Fewer are involved in
managing and selecting supportive care and systemic therapy. Longitudinal
and longer follow up are warranted to ascertain whether the scope of these
practices change over time.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
404s Health Services Research, Clinical Informatics, and Quality of Care

6648 Poster Session (Board #443), Sat, 1:15 PM-4:15 PM TPS6649 Poster Session (Board #442a), Sat, 1:15 PM-4:15 PM
Total cost of care for patients (pts) with stage II colon cancer (CC-II): A SEER- Reproductive health in cancer. First Author: Erika K Radeke, Cook County
Medicare analysis. First Author: Afsaneh Barzi, University of Southern Health, Chicago, IL
California Norris Comprehensive Cancer Center, Los Angeles, CA
Background: Reproductive health needs of females ages 15 to 55 with cancer
Background: Adjuvant chemotherapy (AC) for stage II colon cancer remains a are poorly understood and overlooked, despite their importance to patients.
controversial topic. We used Surveillance Epidemiology End Results (SEER) Uncertainties regarding fertility and pregnancy are intricate and challenging
linkage with Medicare claims to explore the benefits of AC and cost of care in for a patient and a cancer care team. The goals of reproductive health care may
pts with CC-II diagnosed between 2004-2010. Methods: Colon cancer was come in conflict with the primary objective of cancer care. However, with
identified using ICD-O-3 codes. TNM staging was used to classify pts as stage appropriate counseling and preventive measures, this conflict can be absolved
II. Cohort was restricted to pts who had surgery within 4months (mos) of the to unify objectives. The primary objective of this trial is to evaluate the success
diagnosis and excluded pts who died within 3mos after the surgery as well as of the implementation of reproductive health programming among re-
those who were enrolledon a health maintenance organization. We searched productive aged females with cancer. To expand knowledge on this subject, the
claims in the 4mos after surgery to identify pts who received AC using ICD-9 NCI approved the ECOG-ACRIN EROS trial: Engendering Reproductive Health
diagnosis and procedure codes, HCPCS, and revenue center codes. Kaplan- within Cancer Survivorship, with two ancillary studies: Endocrine Disruption
Meier method was used for survival analysis. Cost of care from payer’s per- in Cancer Care and Sexuality in the first 5 years after Cancer Diagnosis.
spective (Medicare) starting from surgery service to death or 3 years after Methods: The primary study is a multicenter, cluster randomized control trial,
surgery was captured from claims data. Results: A total of 16948 pts with CC-II with NCI Community Oncology Research Programs randomized to either the
diagnosed 2004-2010 were included in this analysis. Among those 14% non-intervention arm (usual standard practice related to reproductive health)
received AC and 33% of them oxaliplatin. After adjusting for pts and tumor or the intervention arm (using study-specific training and tools). The accrual
characteristics, probabilityof survival at 3years was 72.9% for pts who received goal is 668 patients based on the expectation that the intervention can in-
AC and 74.2% for those who did not, HR=1.06 (95% CI, 0.96-1.17), with P- crease the adoption of appropriate reproductive health management within
value: 0.229. The cost of care for the AC was significantly higher than the no- 3 months from the baseline visit from 50% to 80%, with the first 200 to
AC group (median $254,116 vs. $91,086). The biggest difference in between consent to the endocrine disruption substudy. All patients participate in the
two groups was cost of outpatient and physician services. Costs for AC group sexuality part. Pre-menopausal female patients ages 15 to 55 with an initial
was higher in years 1, 2, and 3 as seen in table. Conclusions: AC for CC-II is diagnosis of any type of cancer who have not initiated treatment of any type are
has a low value in elderly pts and should be avoided. eligible to participate. Patients are asked to complete 2 questionnaires at 8
timepoints regarding their reproductive health interests. Providers are also
Year 1 Year 2 Year 3
asked to complete questionnaires regarding their healthcare practice in
No AC AC No AC AC No AC AC
Mean $ Mean $ Mean $ Mean $ Mean $ Mean $ general as well as specific to patients enrolled in this study. The reproductive
(SD) (SD) (SD) (SD) (SD) (SD) health management rate at each time point will be summarized, by arm, with
Total cost 83,589 323,191 ,.0001 60,696 96,576 ,.0001 53,195 75,517 ,.0001 frequency and percentage along with its 95% confidence interval. The
(151,945) (403,142) (178,371) (245,029) (182,332) (253,964)
Hospital 21,428 23,999 0.0002 6,171 7,731 0.0042 4,814 5,483 0.1100 comparison between the two arms, using a GEE model, can evaluate the
(28,695) (31,151) (18,194) (24,725) (15,151) (18,901) intervention effect. EROS was activated in September 2015. To date, 264
Physician 24,218 168,041 ,.0001 17,772 39,811 ,.0001 14,369 27,636 ,.0001
49,397) (205,493) (70,991) (134,554) (68,632) (101,471) patients have been enrolled to the main study and half of these patients are
outpatient 30,306 127,109 ,.0001 30,614 45,749 ,.0001 27,950 39,460 0.0186 participating in the endocrine disruption correlative. Outcomes pertaining to
(118,451) (372,056) (150,264) (189,755) (159,147) (222,892)
management and treatment implementation and modification are the cor-
nerstones of this study. It should inform organizations in cancer care to improve
guidelines and to include a reproductive health assessment for all young fe-
males with cancer. Clinical trial information: NCT01806129.

TPS6650 Poster Session (Board #442b), Sat, 1:15 PM-4:15 PM TPS6651 Poster Session (Board #441a), Sat, 1:15 PM-4:15 PM
Patient related outcomes in cancer patients in Croatia. First Author: Dragan Digitally captured step counts for evaluating performance status in advanced
Trivanovic, General Hospital Pula, Pula, Croatia cancer patients: A single cohort, prospective trial (Digi-STEPS). First Author:
Gillian Gresham, Cedars-Sinai Medical Center, Los Angeles, CA
Background: Accurate evaluation of symptom intensities is essential for optimal
cancer care and improving the quality of life of patients. An inappropriate in- Background: Advanced cancer patients undergo dynamic changes in their
terpretation of symptoms may lead to treatment outcomes failure, overdose of functionality and physical activity over the course of their treatment. Moni-
medication, or may leave the patients undertreated. However, the perception of toring patient function is important because it can inform treatment decisions
symptoms can vary between the treating physician and patient. Physicians appear and allow for timely and appropriate intervention. Current scales that assess
to underestimate the patient symptoms. And this variation in the perception of side patient function, such as the ECOG Performance Status (PS), are limited in
effects can lead to wrong assumptions and subsequent treatment changes, af- their ability to capture the wide range in activity that cancer patients can
fecting treatment effectiveness and quality of life. There is growing interest to experience on a daily basis outside of the clinic setting. Given recent tech-
enhance symptom monitoring during routine cancer care using patient-reported nological advances in wearable activity monitors, we can collect real-time,
outcomes, leaving open the question of whether the benefits of systems to reveal objective information about a patient’s daily activity including steps, stairs,
self-reports outweigh their added cost. There are several tools for assessment of heart rate, sleep, and activity intensity. Thus, the primary objective of this study
symptoms in oncology. In cancer treatment clinical trials, the standard source of
is to determine whether longitudinal changes in objectively-assessed activity
adverse symptom data is clinician reporting by use of items from PRO-CTCAE,
are associated with change in physician-rated ECOG PS. Methods: This is a
developed by NCI. To address these questions, we conducted a single-center
prospective, single cohort trial being conducted at Cedars-Sinai Medical
prospective trial to test whether systematic tablet computer-based collection of
patient-reported symptoms during chemotherapy treatment, with automated
Center. Stage 3/4 cancer patients who are English or Spanish-speaking,
alerts to clinicians for severe adverse events (grade 3-4) will change in ques- ambulatory (assistive walking devices are allowed) and expected to be seen
tionnaire score at 6 months compared with baseline. Secondary endpoints will for treatment or follow-up with their oncologist at least every 8 weeks are
include difference in unscheduled clinic visits frequency, and survival. eligible for study. Consenting patients will be asked to wear a Fitbit Charge HR
Methods: Patients initiating chemotherapy at General Hospital Pula Oncology continuously for 8 weeks during the study period and for one week prior to the
Clinic for advanced or metastatic gastrointestinal, lung, breast, genitourinary, or 6 month and 1 year follow-up visits. Primary outcomes are change in average
gynecologic cancers will be enrolled in a nonblinded, prospective trial of self- daily step counts and ECOG PS at 8 weeks from baseline. Secondary outcomes
reporting of symptoms, compare with usual care. Patients receiving chemotherapy include: 1) Change in NIH PROMIS patient-reported outcomes (physical
and their clinicians will be independently asked on the same day to complete 10 function, pain, sleep, emotional distress, and fatigue), 2) Change in frailty
symptoms (including fatigue, pain, nausea, vomiting, diarrhea, dysgeusia, ap- status at 8 weeks, 3) Occurrence of adverse events, and 4) 6-month and 1-year
petite, sleep disturbance, fever and hair loss). Participants will remain on study survival outcomes. Baseline assessments include a physical exam, medical
until discontinuation of cancer treatment, withdrawal, or death. All participants history, and frailty assessment. The attending oncologist will rate the patient’s
will provide written informed consent and followed for up to 28 months or until ECOG PS at baseline and at the end-of-study visit. Weekly NIH PROMIS
death. To compare how patient’s vs clinician’s reports relate to clinical events, a questionnaires will be administered online over the 8-week study and again at
time-dependent Cox regression model adjusted for covariates including cancer 6 months and 1 year follow-up. The occurrence of serious cancer-related
type, age, sex, and education level will be used to measure associations between adverse events, chemotherapy-associated toxicities, and hospitalizations will
reaching particular grade severity thresholds with the risk of death and un- be documented up to 12 weeks from baseline. Survival will be assessed at
scheduled clinic visits. Clinical trial information: 2019-000855-15. 6 months and 1 year. Accrual is ongoing with 20 patients currently enrolled of a
target sample size of 60 patients. Clinical trial information: NCT03757182.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Health Services Research, Clinical Informatics, and Quality of Care 405s

TPS6652 Poster Session (Board #441b), Sat, 1:15 PM-4:15 PM

Design and accrual of S1417CD: Development of a prospective financial
impact assessment tool in patients with metastatic colorectal cancer
(mCRC). First Author: Veena Shankaran, Fred Hutchinson Cancer Re-
search Center, Seattle, WA
Background: Few studies have assessed the financial impact of cancer di-
agnosis (dx) in diverse patients (pts) and caregivers (cgs) using objective and
standard financial measures. S1417CD, led by the SWOG Cancer Research
Network, is the first prospective cohort study assessing financial outcomes to
be conducted in the NCI Community Oncology Research Program (NCORP).
We present our experience with design and accrual. Methods: Pts age $ 18
within 120 days of mCRC dx were considered eligible and asked to identify a
caregiver (cg) who could participate concurrently. The primary endpoint is
incidence of treatment-related financial hardship, defined as $ 1 of the
following: debt accrual, selling/refinancing home, $ 20% income decline, or
borrowing money. Measures include 1) pt and cg surveys (baseline (BL), 3, 6, 9
and 12 months (mo)) assessing out-of-pocket spending, financial impacts, cg
burden, and quality of life and 2) pt credit reports (BL, 6, and 12 mo). Linkage
to records from TransUnion, a national credit agency, required pt social se-
curity number (SSN) and processes for batched credit report transfer via secure
web portal. The accrual goal was n = 374 pts in 3 years. The study activated on
Apr 1, 2016 and closed on Feb 1, 2019 after reaching its accrual goal. A total
of 380 pts (median age 59.7 years) and 155 cgs enrolled (41% cg partici-
pation). Enrollment steadily increased during the study period; 56% enrolled
in the last 12 mo. Credit data were not obtainable for 76 (20%) pts due to early
death, lack of credit, or inability to match records. S1417CD, the first co-
operative group led study assessing financial outcomes in the community
setting, completed enrollment faster than anticipated. Required SSN col-
lection was not a barrier to enrollment, which improved as sites became fa-
miliar with data security measures. Robust accrual to S1417CD demonstrates
pts’ and cgs’ desire to improve understanding of financial toxicity and its
solutions. Follow-up will conclude in 12 mo with results to follow. SWOG plans
to launch a randomized study (S1912) assessing the impact of financial
navigation on household finances, using credit data for primary endpoint
assessment. Clinical Trials Registry Identifier NCI-2015-01885. Clinical trial
information: NCT02728804.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
406s Hematologic Malignancies—Leukemia, Myelodysplastic Syndromes, and Allotransplant

7000 Oral Abstract Session, Sat, 3:00 PM-6:00 PM 7001 Oral Abstract Session, Sat, 3:00 PM-6:00 PM
Effect of gilteritinib on survival in patients with FLT3-mutated (FLT3mut+) A randomized phase II trial of CX-01 with standard therapy in elderly patients
relapsed/refractory (R/R) AML who have common AML co-mutations or a with acute myeloid leukemia (AML). First Author: Tibor Kovacsovics,
high FLT3-ITD allelic ratio. First Author: Mark J. Levis, The Sidney Kimmel Huntsman Cancer Institute, Salt Lake City, UT
Comprehensive Cancer Center at Johns Hopkins University, Baltimore, MD
Background: Elderly AML patients have poor outcomes irrespective of therapy.
Background: The FLT3 inhibitor, gilteritinib, showed superior response and CX-01 is a low anticoagulant heparin derivative that retains heparin’s ability to
overall survival (OS) compared with salvage chemotherapy (SC) in patients alter the activity of the CXCL12/CXCR4 axis, P-selectin, extracellular histones,
(pts) with FLT3mut+ R/R AML in the phase 3 ADMIRAL study. We analyzed the and Platelet Factor 4. A pilot study of CX-01 combined with standard therapy
impact of baseline co-mutations and FLT3-ITD allelic ratio (AR) on response for AML led to a complete remission (CR) rate of 92% (Blood Adv 2:381,
and OS. Methods: A total of 37 recurrently mutated genes in AML (Archer Core 2018). We conducted a randomized, dose-finding study of the same com-
Myeloid Panel) were analyzed by next-generation sequencing; the cutoff for co- bination in newly diagnosed elderly AML patients. Methods: 76 fit patients
mutation positivity (co-mut+) was $0.027. Baseline FLT3-ITD AR (FLT3-ITD older than 59 were randomized to induction with idarubicin and cytarabine
to FLT3 wild-type DNA) was measured by the LeukoStrat CDx FLT3 Mutation on a 7+3 schedule only (Group 1); 7+3 with a lower dose of CX-01 (0.125 mg/
Assay. The median FLT3-ITD AR value of 0.77 was used to define high kg/hour) (Group 2); or 7+3 with a higher dose of CX-01 (0.25 mg/kg/hour)
($0.77) vs low (,0.77) FLT3-ITD AR. Results: Analysis of 361 FLT3mut+ pts (Group 3). Patients in CR received consolidation therapy consisting of up to 3
identified four major co-mutation cohorts, each with $10% of pts: NPM1 cycles of intermediate dose cytarabine (1000 mg/m2 every 12 hours on Days 1,
(n=173; 47.9%), DNMT3A (n=115; 31.9%), DNMT3A/NPM1 (n=86; 3, 5) without or with the same dose of CX-01 for Groups 1, 2, and 3, re-
23.8%), and WT1 (n=65; 18.0%). In addition, seven pts (1.9%) had all three spectively. CX-01 was given as a continuous infusion after a 4 mg/kg bolus until
co-mutations (ie, NPM1, DNMT3A, and WT1). The gilteritinib arm had su- completion of chemotherapy. Results: 66 of 75 treated patients were evalu-
perior response rates and OS across all four major co-mutation cohorts, with the able for response. Ten patients were not evaluable due to withdrawal of consent
greatest survival benefit in pts with DNMT3A/NPM1 co-mut+ (Table). In FLT3- (6 patients), introduction of midostaurin after its approval (3 patients), or death
ITD AR analyses (n=335), gilteritinib conferred longer OS than SC in pts with a due to hepatic sinusoidal obstructive disease at Day 21 (1 patient in Group 2).
high or low FLT3-ITD AR (gilteritinib: high FLT3-ITD AR, 7.1 mos vs low FLT3- We present results for evaluable patients and not on an intent to treat basis.
ITD AR, 10.6 mos; SC: high FLT3-ITD AR, 4.3 mos vs low FLT3-ITD AR, 6.9 Baseline characteristics were similar across groups. The composite CR rate (CR +
mos). In both arms, OS was longer in the low FLT3-ITD AR cohort than the high CRi) was highest for patients in Group 3 with 89% patients achieving a
FLT3-ITD AR cohort but the difference in the gilteritinib arm was not sta- composite CR as compared to 58% and 50% in Groups 1 and 2, respectively.
tistically significant (gilteritinib: HR=1.341, P=0.0712; SC: HR=2.01, Kaplan-Meier curves indicated a statistically significant improvement in event
P=0.0021). Conclusions: The ADMIRAL trial shows that the clinical benefit of free survival (EFS) (P = 0.019) and a non-significant trend (P = 0.10) to
gilteritinib in FLT3mut+ R/R AML is maintained regardless of NPM1, DNMT3A, improvement in OS in Group 3 as compared to Group 1. Groups 1 and 2 were
DNMT3A/NPM1, or WT1 co-mut+ or high FLT3-ITD AR. Clinical trial in- comparable for EFS and OS. CX-01 was well tolerated without increased
formation: NCT02421939. incidence of bleeding in Groups 2 and 3. Conclusions: The encouraging CR
rate and EFS in elderly fit patients with newly diagnosed AML suggests that CX-
CR/CRh (%) Median OS (mos)
01 may potentiate the efficacy of standard AML induction therapy. A ran-
Patients Gilteritinib SC Gilteritinib SC HR P-Value domized study to confirm these findings with the higher dose of CX-01 is
ITT population (n=371) 34.0 15.3 9.3 5.6 0.637 0.0007 warranted. Clinical trial information: NCT02873338.
Co-mut+ cohorts
NPM1 (n=173) 32.2 12.1 8.3 5.1 0.419 ,0.0001
DNMT3A (n=115) 37.3 12.5 9.1 5.5 0.504 0.0031
DNMT3A/NPM1 (n=86) 40.0 9.7 10.8 5.0 0.252 ,0.0001
WT1 (n=65) 35.6 5.0 9.1 3.4 0.309 0.0001

7002 Oral Abstract Session, Sat, 3:00 PM-6:00 PM 7003 Oral Abstract Session, Sat, 3:00 PM-6:00 PM
Association of smoking with poor risk ELN 2017, cytogenetics/molecular Randomized phase II study of cladribine with simultaneous or delayed
profile, and survival outcomes in acute myeloid leukemia. First Author: rituximab in patients with untreated hairy cell leukemia. First Author: Dai
Mansour Alfayez, The University of Texas MD Anderson Cancer Center, Chihara, Medical Oncology Service, NCI, NIH, Bethesda, MD
Houston, TX
Background: Single-agent purine analog therapy, usually cladribine, has been
Background: Smoking increases the relative risk of AML by 40% and 25% in the standard 1st-line therapy of hairy cell leukemia (HCL) for ~30 years.
active and former smokers, respectively, compared with non-smokers (Fir- Relapse is attributed to minimal residual disease (MRD) persisting despite
canis et al., 2014). While the relationship of smoking with AML development complete remission (CR). Rituximab can clear MRD, but there are no long-term
is recognized, whether smoking impacts underling AML biology and clinical data for how frequently and for how long this happens, or even how frequently
outcome remains ill-defined. Methods: Newly diagnosed, treatment naı̈ve MRD remains after cladribine. We therefore conducted a randomized phase II
AML pts seen at MDACC between 2012 and 2017 with available smoking trial of 1st-line cladribine with concurrent vs delayed rituximab. Methods: The
history were evaluated, along with baseline parameters, co-occurring mu- primary endpoint was to determine if 8 weekly doses of rituximab (375 mg/m2
tations, cytogenetics and clinical outcome. Results: We identified 858 pts iv) begun the same day as 5 daily doses of cladribine (0.15 mg/kg iv) reduce
[486 (57%) male; median age 67 yrs (14-97)], representing 535 (62%) HCL MRD 6 months later compared to cladribine alone. Secondary endpoints
treatment naı̈ve and 323 (38%) salvage pts. Smoking status was recorded as included testing delayed rituximab, 8 weekly doses beginning if/when MRD is
smokers (active = 39 pt, former = 380 pt), versus never smoker (439 pt). In tx detected in blood (or if HCL-related cytopenias persist preventing consider-
naı̈ve group, smoking is associated with lower remission rates (OR 0.63, ation of CR) $ 6 months after cladribine. Both groups could receive a 2nd
95% CI 0.43-0.94, p = 0.02) and inferior OS (HR = 1.6, 95% CI 1.27-2.02, course of rituximab if the same situation recurred $ 6 months after starting the
p , 0.001). Smoking status was not significant in multivariate analysis 1st course of rituximab. MRD was assessed in blood (PB) and bone marrow
including AML biologic characteristics and ELN 2017 risk stratification. (BM) using flow cytometry (FC), consensus PCR, and immunohistochemistry.
Therefore we postulated that worse OS may be driven by smoking associated Results: Sixty-eight patients were randomized 1:1 to concurrent vs delayed
AML biology rather than smoking associated comorbidities. Indeed, in arms. At 6 months after cladribine, 97% concurrent vs 24% delayed (p ,
univariate analysis smoking was associated with poor ELN risk (p = 0.015), 0.0001) patients were MRD-free by all tests. 100% concurrent vs 41%
complex karyotype (p = 0.0002), and TP53 mutation (p = 0.0235) while delayed patients were MRD-free by all tests except BMA FC (p , 0.0001). At a
negatively associated with NPM1 (p = 0.018), FLT3-ITD (p = 0.032) and median follow-up of 87.3 months, delayed rituximab was required by 1
GATA2 (p = 0.0497). Age was a significant cofounder between smokers vs concurrent patient vs 27 courses in 21 delayed patients. MRD-free survival
non-smoker ( , 0.0001). After controlling for age, significance was retained after 1st rituximab was not reached for 34 patients after concurrent rituximab,
for ELN risk, complex karyotype and GATA2 at p = 0.0454, p = 0.0006, p = with 32 still MRD-free after a median of 72 months (94% MRD-free survival),
0.048 respectively, while significance was lost for NPM1 (p = 0.079), FLT3- compared to median MRD-free survival 60.1 months after delayed rituximab,
ITD (p = 0.1) and TP53 (p = 0.084). In analysis of young pts ( , 60 yr), with 10 (48%) of 21 remaining MRD-free (p , 0.0001, hazard ratio for
smoking is positively associated with complex karyotype (p = 0.0042) and concurrent rituximab 0.09). Conclusions: Achievement of MRD-free CR after
TP53 (p = 0.0289), and negatively associated with RUNX1 (p = 0.0143) and 1st line cladribine for HCL is greatly enhanced by concurrent rituximab. Use of
IDH2 (p = 0.0357). Conclusions: We report the largest analysis of smoking delayed rather than concurrent rituximab could achieve MRD-free CR but
status and impact on molecular, cytogenetics, and AML clinical outcomes. durability was clearly inferior to concurrent rituximab. Longer follow-up will
Smoking history is associated with poorer risk molecular and cytogenetics, determine if MRD-free survival leads to cures of HCL, increased CR dura-
lower response rate and shorter survival in treatment naı̈ve patients. tion, and/or less need for additional therapy. Clinical trial information:
NCT00923013.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Hematologic Malignancies—Leukemia, Myelodysplastic Syndromes, and Allotransplant 407s

7004 Oral Abstract Session, Sat, 3:00 PM-6:00 PM 7005 Oral Abstract Session, Sat, 3:00 PM-6:00 PM
Frontline flumatinib versus imatinib in patients with chronic myeloid leukemia ENESTop 192-week results: Treatment-free remission (TFR) in patients (pts)
in chronic phase: Results from the China randomized phase III study. First with chronic myeloid leukemia in chronic phase (CML-CP) after stopping
Author: Zhang Li, State Key Laboratory of Experimental Hematology, Institute second-line (2L) nilotinib (NIL). First Author: Timothy P. Hughes, SA Pathology
of Hematology, Chinese Academy of Medical Sciences, Tianjin, China and South Australian Health and Medical Research Institute, University of
Adelaide, Adelaide, Australia
Background: Flumatinib (FM), a derivative of imatinib (IM), is a novel BCR-
ABL1 tyrosine kinase inhibitor (TKI). The aim of this open-label phase III Background: In the ENESTop study (NCT01698905) of TFR in pts with
study was to validate the efficacy and safety of FM in comparison with IM as CML-CP who achieved a sustained deep molecular response (MR) with 2L
frontline treatment in Chinese patients with newly diagnosed Philadelphia NIL, 57.9% remained in TFR 48 wks after stopping NIL (primary endpoint).
chromosome–positive chronic myeloid leukemia (CML) in chronic phase Analyses at 144 wks showed durability of TFR. Data from longer follow-up
(CML-CP). Methods: Randomization was stratified by Sokal score with a 1:1 (192 wks) evaluating the maintenance of TFR are reported. Methods: Pts
allocation to each arm. Primary endpoints were major molecular response treated with $2 y NIL after . 4 wks imatinib ($3 y total) and achieving
(MMR = BCR-ABL IS#0.1%) rates at 6 and 12 months. Molecular responses MR4.5 (BCR-ABL1IS #0.0032%) on NIL were eligible. After a 1 y consol-
were assessed at a central laboratory blinded to treatment allocations during idation, pts with no confirmed loss of MR4.5 could attempt TFR. NIL was
the study. Efficacy endpoints were analyzed for the intention-to-treat pop- resumed upon loss of major MR (BCR-ABL1IS #0.1%) or confirmed loss of
ulations. This study is registered with ClinicalTrials.gov, number NCT02204644. MR4 (BCR-ABL1IS #0.01%). At the data cut-off (Sep 24 2018), all pts had
Results: 400 eligible patients were randomized and patient characteristics at completed $192 wks of TFR, resumed NIL, or discontinued the study.
baseline were similar in each arm. The full analysis set (FAS) consisted of 393 Results: By the data cut-off, of 126 pts entering TFR, 56 were ongoing, 59
patients who received FM 600 mg (n = 196) or IM 400 mg (n = 197) tablets had resumed NIL, and 11 had discontinued. TFR rate at 192 wks was 46.0%
once daily. Compared with IM, FM resulted significantly higher induction of (58/126; 95% CI, 37.1–55.1%); all but 1 of the 58 pts were in MR4.5. Only
MMR rate (%; 95%CI) at 6 month (33.7; 27.06-40.29 vs 18.3; 12.88-23.67; 1/61 pts in TFR at 144 wks lost response by 192 wks (confirmed loss of
P = 0.0005) and 12 month (48.5; 41.47-55.47 vs 33.0; 26.43-39.56; P = MR4); another 2 pts discontinued due to serious AE (polycythemia vera) and
0.0021) and also at 3 month (8.2; 4.33- 12.00 vs 2.0; 0.06-4.00; P = 0.0058). pt/guardian decision, respectively. Of 59 pts who resumed NIL, 56 (94.9%)
Significantly more patients in the FM than in the IM arm achieved a complete and 55 (93.2%) regained MR4 and MR4.5, respectively. 40/56 pts (71.4%)
molecular response (BCR-ABL IS#0.0032%) at 12 months. Early molecular who regained MR4 had stable MR4 at 96 wks (12 discontinued , 96 wks,
response (BCR-ABL IS # 10%) at 3 months and early CCyR at 6 months were and 4 remained on study with , 96 wks, after regaining MR4); 37/55 pts
also significantly higher with FM than IM (82.1; 76.78-87.50 vs 53.3; 46.33- (67.3%) who regained MR4.5 had stable MR4.5 at 96 wks (12 discontinued
60.27; P , 0.0001 and 60.71; 53.88-67.55 vs 49.75, 42.76, 56.73; P = , 96 wks, and 6 remained on study with , 96 wks, after regaining MR4.5). There
0.0332). FM has a safety profile similar to IM. The rates of grade 3/4 TEAEs of were no disease progressions, deaths due to CML, or new deaths since the 144-
FM were similar to IM, 56.57% (112 of 198) vs 41.38% (87 of 196). However, wk analysis. The 192 wk treatment-free survival rate was 50.3% (95% CI,
the frequencies of some nonhematological and hematological adverse events 41.2–58.7%). Of 62 pts who remained in TFR for . 144 wks, 11.3%, 53.2%,
were significantly lower in the FM than in the IM arm, such as rash (4.59% vs 21.0%, 14.5% and 3.2% had musculoskeletal pain AEs during consolidation
12.63%, P = 0.0064) and eyelid edema (0.51 vs 14.65, P , 0.0001); leu- and each subsequent 48 wk period of TFR. Among 59 pts who resumed NIL,
kopenia (30.61 vs 62.63, P , 0.0001) and neutropenia (30.10 vs 59.60, P , most common AEs were hypertension (20.3%) and arthralgia (13.6%); the
0.0001). No specific TEAE was identified in each arm. Conclusions: This phase majority of AEs were grade 1/2. Conclusions: Results demonstrate long-term
III study met its primary endpoints. Our study results suggest that FM is com- durability and safety of TFR following 2L NIL, with no disease progressions or
parable to IM in its safety and superior in its efficacy profile at 3, 6 and 12 month CML-related deaths, and musculoskeletal pain AEs were transient. Clinical trial
time points. These results support FM as a frontline treatment option for patients information: NCT01698905.
with newly diagnosed CML-CP. Clinical trial information: NCT02204644.

7006 Oral Abstract Session, Sat, 3:00 PM-6:00 PM 7007 Oral Abstract Session, Sat, 3:00 PM-6:00 PM
End of phase I results of ZUMA-3, a phase 1/2 study of KTE-X19, anti-CD19 Post-transplant cyclophosphamide in matched and haploidentical trans-
chimeric antigen receptor (CAR) T cell therapy, in adult patients (pts) with plant recipients receiving myeloablative timed sequential busulfan condi-
relapsed/refractory (R/R) acute lymphoblastic leukemia (ALL). First Author: tioning regimen: Results of a phase II study. First Author: Uday R. Popat, The
Bijal D. Shah, H. Lee Moffitt Cancer Center & Research Institute, Tampa, FL University of Texas MD Anderson Cancer Center, Houston, TX
Background: KTE-X19 is an autologous anti-CD19 CAR T cell therapy under Background: Myeloablative stem cell transplants have a lower rate of relapse than re-
investigation for adult R/R ALL. In an interim analysis of Phase 1 of ZUMA-3, duced intensity regimens. Timed sequential busulfan (TSB) with fludarabine (Flu) is a
we reported manageable safety and encouraging efficacy of KTE-X19; 72% promising myeloablative regimen for patients undergoing matched sibling (MSD) or
unrelated (MUD) donor transplantation (HCT) with low non-relapse mortality (NRM)
of pts achieved a complete remission (CR) or CR with incomplete bone marrow
(Popat et al Lancet Haematology 2018), but was not tested in haploidentical (haplo). Also,
(BM) recovery (CRi; Wierda et al, ASH 2018. #897). Here, we present end of whether this approach can be used with post-transplant cyclophosphamide (PTCy) in
Phase 1 results. Methods: Adults with R/R B cell ALL, . 5% BM blasts, and MSD, MUD and haplo HCT is unknown. To address these issues, we conducted a pro-
ECOG 0-1 received 2, 1, or 0.5 3 106 KTE-X19 cells/kg after conditioning spective phase II study. Methods: Patients with hematological malignancies with MSD,
chemotherapy. Revised adverse event management (rAE mgmt) was imple- MUD or haplo donor were eligible. They received fixed doses of Busulfan(BU) 80mg/m2
mented for additional pts in a 1 3 106 dose cohort: corticosteroids were given either on day -13 and -12 (n=45) or on -20 and -13 (n=10). Then, Flu 40mg/m2 was given
earlier at onset of Grade $ 2 neurologic events (NEs) and tocilizumab was used on day -6 to -2 followed by Bu dosed to achieve target area under the curve (AUC) of
only for active toxicity. The primary endpoint was the dose-limiting toxicity 20,000 umol/min for the whole course based on pharmacokinetic studies. Thiotepa 5mg/
(DLT) rate. Key additional endpoints were KTE-X19 levels, incidence of AEs, kg was given on day -7 to haplo group. GVHD prophylaxis was PTCy 50mg/kg on day 3 and
4 and tacrolimus. Haplo and later MUD recipients also received mycophenolate mofetil.
minimal residual disease (MRD), and CR/CRi rate. Results: As of 9/27/18, 45 Results: 55 patients with a median age of 47 (15-65) years were enrolled. 30 patients had
pts had received KTE-X19 (median follow-up [f/u], 16 mo). The median age AML or MDS, 9 CML or MPD, 5 lymphoma, 5 myeloma and 6 ALL. About half had haplo
was 46 y (range, 18–77); 30 pts (66%) had $ 3 prior therapies and the median 26 (47%); others had MUD 18 (33%) or MSD 11(20%). Disease risk index was high in 18
pre-conditioning BM blasts was 70% (range, 0–97). Six, 23, and 16 pts (32%), intermediate in 32 (58%), and low in 5 (9%) patients. Comorbidity score was $3
received 2, 1, and 0.5 3 106 cells/kg, respectively. There were no DLTs in the in 22 (40%) patients. With a median follow up of 17 months (5-28), 1-year OS, PFS, NRM
DLT-evaluable pts. The most common Grade $ 3 AEs were hypotension and relapse rates were 71% (60-84%), 63% (51-77%), 20% (9-31%), and 17% (7-
(38%), pyrexia (38%) and thrombocytopenia (31%). There were 2 previously 27%), respectively [Table]. There were no graft failures. Day 100 grade II-IV and III-IV
reported KTE-X19–related Grade 5 AEs of cerebral infarction and multiorgan acute GVHD rates were 38% (25-51%) and 9% (95% CI 1-17%), respectively; 1-year
chronic GVHD and extensive chronic GVHD rates were 10% (2-28%) and 8% (0-15%),
failure, both in the context of CRS. Grade $ 3 CRS and NEs occurred in 13
respectively. 1-year OS in MSD, MUD and haplo groups were 91% (75-100%), 72% (54-
(29%) and 17 (38%) pts, respectively. Of 41 pts with $ 2 mo of f/u, 68% had 96%), and 62% (45-83%) respectively (P=0.11). Conclusions: Myeloablative TSB with
CR/CRi, and 73% had undetectable MRD. Of 19 pts with $ 2 mo of f/u treated PTCy is feasible in MSD, MUD and haplo HCT. It lowers the incidence of severe acute and
with 1 3 106 cells/kg, 16 (84%) had a CR/CRi and the median event-free chronic GVHD without apparent increase in relapse. Clinical trial information:
survival was 15 mo. In 9 pts treated with 1 3 106 cells/kg and rAE mgmt, 2 NCT02861417.
(22%) had Grade 3 CRS and 1 (11%) had Grade 3 NE with no Grade 4/5 Results (N=55) 95% CI
events. Conclusions: KTE-X19 dosing and safety mgmt have been successfully
OS, 1 yr. 71% (60-84%)
refined by testing 3 cell doses and evaluating a new AE mgmt guideline with MSD (N=11) 91% (75-100%)
altered corticosteroids/tocilizumab use for NE/CRS. Pivotal Phase 2 is ongoing MUD (N=18)
Haplo (N=26)
72%
62%
(54-96%)
(45-83%)
at the 1 3 106 dose with rAE mgmt. Clinical trial information: NCT02614066. PFS, 1 yr. 63% (51-77%)
Relapse, 1 yr. 17% (7-27%)
NRM, 1 year 20% (9-31%)
Acute GVHD II-IV 38% (25-51%)
Acute GVHD III-IV 9% (1-17%)
Chronic GVHD, 1 yr. 10% (2-18%)
Extensive Chronic GVHD 8% (0-15%)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
408s Hematologic Malignancies—Leukemia, Myelodysplastic Syndromes, and Allotransplant

7008 Oral Abstract Session, Sat, 3:00 PM-6:00 PM 7009 Poster Discussion Session; Displayed in Poster Session (Board #384),
Off-the-shelf virus specific T-cells for therapy of adenovirus disease in Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
immunosuppressed patients. First Author: Hind Rafei, Division of Cancer Mon, 4:30 PM-6:00 PM
Medicine, The University of Texas MD Anderson Cancer Center, Houston, TX The first-in-class anti-CD47 antibody Hu5F9-G4 is active and well tolerated
alone or with azacitidine in AML and MDS patients: Initial phase 1b results.
Background: Adenovirus infection can cause significant morbidity and mortality in
immunosuppressed patients. Cidofovir is commonly used, but its nephrotoxicity is
First Author: David Andrew Sallman, Moffitt Cancer Center, Tampa, FL
concerning and efficacy limited. Another approach is to restore the anti-adenovirus Background: Hu5F9-G4 (5F9) is an antibody targeting CD47, a macrophage
immunity. Indeed, virus specific T-cells have been shown to be safe and effective in immune checkpoint and “don’t eat me” signal on cancers. CD47 blockade
stem cell transplant recipients. Methods: Immunosuppressed pts with either ade- induces tumor phagocytosis and eliminates leukemia stem cells (LSC) in AML
novirus viremia or adenovirus-related end organ damage were enrolled. Most closely models. Azacitidine (AZA) synergizes with 5F9 by inducing “eat me” signals on
HLA-matched adenovirus cytotoxic T lymphocytes (CTLs) were generated by AML, enhancing phagocytosis. This trial explored the safety/efficacy of 5F9
expanding donor derived T-cells with a peptide library derived from the hexon protein
alone or with AZA in AML/MDS patients (pts). Methods: This Phase 1b treated:
of adenovirus serotype 3 in the presence of IL-2 20 IU/ml, IL7 10 ng/ml, IL4 10 ng/ml.
After receiving 2x105 /kg T cells, pts were monitored for response and adverse r/r AML/MDS pts with 5F9; and untreated AML (induction chemo ineligible) and
events. Results: Eight pts received adenovirus CTLs with one infusion. The Table higher risk MDS pts with 5F9+AZA. A 5F9 priming/intrapatient dose escalation
summarizes their characteristics and responses. Seven pts had complete resolution regimen (1-30 mg/kg weekly) was used to mitigate on target anemia.
of their symptoms (CR) and adenovirus becoming undetectable (ND). Those pts are Results: 10 (6 AML, 4 MDS) r/r pts received 5F9 (median 2 prior therapies
alive to date. The remaining patient initially responded but then lost the response (range 1-6). 24 untreated pts (15 AML, 9 MDS) received 5F9+AZA. In total,
when started on high dose prednisolone for treatment of GVHD to which she median age was 73, 62% of AML pts were intermediate or poor cytogenetic risk
eventually succumbed. Best response was achieved after a median time of 13 days (38% unknown), all MDS pts were intermediate or high risk by IPSS-R. 5F9
[10-36]. No cytokine release syndrome occurred and we did not observe any side alone or with AZA was well-tolerated with no MTD reached. 5F9 did not po-
effect attributable to the CTLs. Conclusions: The use of off-the-shelf adenovirus tentiate AZA toxicities. Treatment-related AEs ( . 10% of pts) for 5F9+AZA
CTLs is a feasible, safe, and effective approach to treat severe adenovirus infections were anemia (25%), thrombocytopenia (20%), and infusion reactions (15%). In
in immunosuppressed pts. Clinical trial information: NCT03425526. 25 efficacy evaluable pts, 8/15 (53%) untreated AML/MDS pts had a CR/CRi to
Patient demographic and clinical characteristics. 5F9+AZA (5/10 (50%) in AML, 3/5 (60%) in MDS). 1/10 (10%) r/r AML/MDS
Viremia (day Adenovirus- Viremia (day pts had a response (MLFS) to 5F9 alone. LSC frequency was reduced/
0), tran- related 28), tran- Response
Patient Disease Age scripts/ml disease scripts/ml (day 28) Comments eliminated in most 5F9+aza responders; 50% of responders were MRD neg-
1 AML post- 67 19,300 Pneumonia ND CR ative by flow cytometry. 4/10 (40%) AML pts became RBC transfusion in-
2
allotransplant
DLBCL post- 60 45,200 Pneumonia ND CR
dependent and 4/5 (80%) MDS pts had hematologic improvement. Time to
allotransplant response was more rapid (median 1.9 mos) than expected for AZA alone. As of
3 CML post- 37 53,900 Exudative ND CR
allotransplant tonsillitis Jan 2019, no responder has relapsed (median follow-up of 3.4 mos (range 1.1 –
4 CML post- 48 No viremia Pneumonia - CR 6.8 mos). 2 pts had successful allogeneic transplant. Conclusions: 5F9+AZA
allotransplant
5 Myelofibrosis 58 8,700 Cystitis ND CR is a novel immunotherapy blocking a key macrophage checkpoint. It has been
post-
allotransplant
well tolerated with robust activity in AML/MDS pts with rapid CRs and MRD
6 Follicular lym- 65 11,000 Bone mar- ND CR negativity. Adding 5F9 to cytotoxic agents may be a promising treatment
phoma post- row aplasia
allotransplant strategy. An expansion cohort is ongoing. Funded by Forty Seven and the
7 MDS post-
allotransplant
64 2,900 Viremia 2,700 Viral load at day 7 increased to 843,000
then gradually decreased by day 28.
California Institute for Regenerative Medicine. Clinical trial information:
Response was lost after starting sys- NCT03248479.
temic steroids at day 32
8 CLL 62 3,800 Pneumonia CR

7010 Poster Discussion Session; Displayed in Poster Session (Board #385), 7011 Poster Discussion Session; Displayed in Poster Session (Board #386),
Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Mon, 4:30 PM-6:00 PM Mon, 4:30 PM-6:00 PM
Low-dose cytarabine with or without glasdegib in newly diagnosed patients Mutant IDH1 inhibitor ivosidenib (IVO; AG-120) in combination with
with acute myeloid leukemia: Long-term analysis of a phase 2 randomized azacitidine (AZA) for newly diagnosed acute myeloid leukemia (ND AML).
trial. First Author: B. Douglas Smith, The Sidney Kimmel Comprehensive First Author: Courtney Denton Dinardo, The University of Texas MD Anderson
Cancer Center, Johns Hopkins, Baltimore, MD Cancer Center, Houston, TX
Background: Glasdegib is a potent, selective, oral inhibitor of the Hedgehog signaling Background: IVO, a mutant IDH1 (mIDH1) inhibitor, is approved for the
pathway, approved in the US – in combination with low-dose cytarabine (LDAC) – for treatment of relapsed/refractory mIDH1 AML. We report results from an ongoing
treatment of newly diagnosed acute myeloid leukemia (AML) in patients (pts) unable phase 1b study of patients (pts) with mIDH1 ND AML ineligible for intensive
to receive intensive chemotherapy due to comorbidities or age ($75 y). Methods: In treatment who received combination IVO+AZA (NCT02677922). Methods: Pts
this follow-up analysis from the phase 2 BRIGHT AML trial (NCT01546038), newly received oral IVO 500 mg daily continuously and subcutaneous AZA 75 mg/m2
diagnosed pts with AML ineligible for intensive chemotherapy were randomized 2:1 to on D1–7 in 28-d cycles. ORR comprised CR + CRi/CRp + PR+ MLFS. CR with
glasdegib + LDAC or LDAC alone (study design: Cortes et al., Leukemia 2018). This partial hematologic recovery (CRh) was defined as CR with ANC . 0.53109/L
long-term analysis evaluated efficacy and safety after ~20 mo of additional follow-up.
and platelets . 503109/L. Exploratory analysis included digital PCR as-
Results: As of Oct 11, 2018, 116 pts were assigned to treatment with glasdegib +
LDAC (n = 78) or LDAC alone (n = 38) (median follow-up: 43.4 and 42.0 mo, re-
sessment of mIDH1 allele frequency in bone marrow mononuclear cells
spectively). Median overall survival (OS) was higher with glasdegib + LDAC vs LDAC (#0.04% sensitivity). Results: As of 9Oct2018, 23 pts received IVO+AZA (11
alone (Table). Improvement in OS was consistent across the prespecified subgroups. male; median age 76 yrs [range 61–88]). Median duration of exposure was 11
The main cause of death in both arms was disease progression (both during study and mo (0.3–25.3); 12 pts remained on treatment at data cutoff. All-grade adverse
follow-up). The incidence of adverse events (AEs) and serious AEs on glasdegib was events (AEs) regardless of cause in $30% pts were thrombocytopenia (65%),
generally lower long term (after 90 days) than short term (during the first 90 days) nausea (61%), diarrhea (57%), anemia (52%), constipation (52%), febrile
(83.7% and 51.2% vs 98.7% and 65.3%, respectively). Clinical trial information: neutropenia (39%), pyrexia (39%), vomiting (35%), fatigue (35%), hypoka-
NCT01546038. Conclusions: Addition of glasdegib to LDAC vs LDAC alone con- lemia (35%), dizziness (35%), insomnia (35%), and neutropenia (30%). AEs of
tinued to demonstrate improved OS in pts with AML in this analysis; improvement was special interest included ECG QT prolonged (26%), IDH differentiation syn-
consistent across groups stratified by cytogenic risk. Long-term follow-up confirmed drome (17%), and leukocytosis (13%). Grade 3/4 AEs in $10% pts were
treatment with glasdegib was associated with an acceptable safety profile. thrombocytopenia (61%), anemia (44%), febrile neutropenia (39%), neu-
HR (95% CI) tropenia (26%), sepsis (22%), and ECG QT prolonged (13%). ORR was 78%
Glasdegib + LDAC LDAC Pvalue (n = 18): CR 57%, CRi/CRp 13%, and MLFS 9%. CR+CRh rate was 70% (n = 16).
OS n = 78 n = 38 Median time to response was 1.8 mo (0.7–3.8) and to CR 3.5 mo (0.8–6.0);
Median (95% CI), mo 8.3 (4.7–12.2) 4.3 (1.9–5.7) 0.495 (0.325–0.752) median response duration not yet reached. mIDH1 clearance was seen in 10/16
0.0004 pts (63%) with CR/CRh, including 9/13 (69%) with CR. Conclusions: IVO+AZA
Deaths, n (%) 69 (88.5) 35 (92.1)
OS, good / intermediate cytogenic risk* n = 53 (67.9%) n = 22 (57.9%) was well tolerated with a safety profile consistent with IVO or AZA monotherapy.
Median (95% CI), mo 12.2 (6.9–16.5) 5.3 (3.5–8.7) 0.510 (0.294–0.886) All-grade cytopenia-related AEs were infrequent relative to other non-intensive
0.0074 therapies. CR and ORR rates exceeded those of AZA alone (Dombret et al.,
Deaths, n (%) 45 (84.9) 19 (86.4)
OS, poor cytogenic risk† n = 25 (32.1%) n = 16 (42.1%) Blood 2015) and most responders achieved mIDH1 mutation clearance. Based
Median (95% CI), mo 4.4 (2.6–7.4) 2.1 (1.0–4.9) 0.514 (0.264–1.000) on these findings, a phase 3 double-blind placebo-controlled study of IVO +AZA
0.0229
Deaths, n (%) 24 (96.0) 16 (100)
(AGILE, NCT03173248) is actively enrolling pts. Clinical trial information:
NCT02677922.
*Favorable, intermediate-I and intermediate-II risk groups † Adverse risk group CI = confidence
interval; HR = hazard ratio

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Hematologic Malignancies—Leukemia, Myelodysplastic Syndromes, and Allotransplant 409s

7012 Poster Discussion Session; Displayed in Poster Session (Board #387), 7013 Poster Discussion Session; Displayed in Poster Session (Board #388),
Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Mon, 4:30 PM-6:00 PM Mon, 4:30 PM-6:00 PM
Primary results of the phase 4 BYOND study of bosutinib (BOS) for pretreated Treatment-free remission (TFR) following frontline (1L) nilotinib (NIL) in
chronic phase (CP) chronic myeloid leukemia (CML). First Author: Carlo patients (pts) with chronic myeloid leukemia in chronic phase (CML-CP):
Gambacorti-Passerini, University of Milano-Bicocca, Monza, Italy 192-week data from the ENESTfreedom study. First Author: Francis J. Giles,
Developmental Therapeutics Consortium, Chicago, IL
Background: The tyrosine kinase inhibitor (TKI) BOS is approved for patients (pts)
with Philadelphia chromosome (Ph)+ CML resistant/intolerant to prior therapy and Background: In ENESTfreedom (NCT01784068), which evaluated TFR fol-
newly diagnosed pts in CP. Methods: The ongoing phase 4 BYOND study is further lowing 1L NIL in CML-CP pts, 51.6% remained in TFR 48 wks after stopping
evaluating efficacy and safety of BOS (starting dose 500 mg/d) for CML resistant/ treatment (primary endpoint) and durability of TFR was demonstrated at 144
intolerant to prior TKIs. Primary endpoint (not powered) in Ph+ CP CML cohorts is wks. Data from longer follow-up (192 wks) evaluating maintenance of TFR are
cumulative confirmed major cytogenetic response (MCyR) by 1 y. Results: Of 163 reported. Methods: Pts with MR4.5 (BCR-ABL1IS #0.0032%) and $2y of 1L
pts who received BOS, 156 had Ph+ CP CML (46, 61 and 49 after 1, 2 and 3 prior
NIL entered a 1y consolidation; pts with sustained deep molecular response
TKIs, respectively). Across Ph+ CP CML cohorts, 51.9% of pts were male; median
(MR) were eligible for TFR. NIL was resumed after loss of major MR (MMR;
age was 61 y. As of 1 y after last enrolled pt (median follow-up 30.4 mo), 56.4%
remained on BOS. Median BOS duration was 23.7 mo and median dose intensity BCR-ABL1IS #0.1%). At the latest data cut-off (Sep 17 2018), all pts had
after adjustment due to adverse events (AEs) 313 mg/d. Of 144 evaluable pts with a completed $192 wks of TFR, resumed NIL, or discontinued the study.
valid baseline assessment, cumulative confirmed MCyR by 1 y was 71.5% (95% Results: By the data cut-off, of 190 pts entering TFR, 87 were ongoing, 91 had
confidence interval [CI] 63.4–78.7). Cumulative complete cytogenetic response resumed NIL, and 12 had permanently discontinued. The TFR rate at 192 wks
rate anytime on treatment was 81.3% (95% CI 73.9–87.3). Cumulative molecular was 44.2% (84/190, 95% CI: 37.0–51.6%). Of 89 pts with successful TFR at
response (MR) rates were high across lines of therapy (Table). 10 deaths occurred (5 144 wks, 5 were not assessable for TFR at 192 wks as 2 had discontinued by
on treatment); 1-y overall survival rate was 98.0%. No pt progressed to accelerated/ 192 wks due to pt/physician decision, and 3 with MR4.5 previously did not have
blast phase on treatment. 25.0% discontinued BOS due to AEs and 5.1% due to 192 wk PCR data. Of 91 pts who resumed NIL, 90 (98.9%) regained MMR and
insufficient response. Most common treatment-emergent AEs (TEAEs) were diarrhea 84 (92.3%) regained MR4.5. 75/90 and 73/84 pts, respectively, had stable
(87.8%) and nausea (41.0%). Grade 3/4 TEAEs in . 10% of pts were diarrhea MMR and MR4.5 at 48 wks later. There were no cases of disease progression or
(16.7%) and increased alanine aminotransferase (ALT; 14.7%). The only TEAE new deaths. 10 deaths were reported in the 144-wk analysis, none due to CML.
leading to discontinuation in . 5% of pts was increased ALT (5.1%). The 192-wk treatment-free survival rate was 48.7% (95% CI 41.4–55.6%). Of
Conclusions: Most pretreated pts with Ph+ CP CML had MCyR by 1 y with BOS; a 89 pts remaining in TFR for . 144 wks (including 87 pts for . 192 wks), all-
substantial proportion achieved or preserved major MR (MMR) and deep MR in all grade AE rates during consolidation and each subsequent 48 wk period of TFR
therapy lines. Results further support BOS use for Ph+ CP CML resistant/intolerant to
were 84.3%, 77.5%, 70.8%, 48.3%, and 52.8%, respectively. All-grade
prior TKIs. Clinical trial information: NCT02228382.
musculoskeletal pain AE rates were 15.7%, 40.4%, 9.0%, 3.4% and 3.4%,
Ph+ CP CML, No. Prior TKIs respectively; cardiovascular event rates were low across these periods. Among
Cumulative rate any time on treatment, % (95% CI) 1 2 3 pts who resumed NIL, most common AEs were nasopharyngitis (18.7%) and
Evaluable, n 46 55 48 pruritus, fatigue, and increased lipase (14.3% each); the majority of AEs were
MMR
MR4
82.6 (68.6–92.2)
69.6 (54.2–82.3)
76.4 (63.0–86.8)
61.8 (47.7–74.6)
56.3 (41.2–70.5)
39.6 (25.8–54.7)
grade 1/2. Conclusions: Results continue to support the long-term durability
No baseline MMR, n 25 28 26 and safety of TFR at 192 wks after stopping 1L NIL; overall AE rates declined
MMR 76.0 (54.9–90.6) 64.3 (44.1–81.4) 38.5 (20.2–59.4)
MR4 52.0 (31.3–72.2) 42.9 (24.5–62.8) 19.2 (6.6–39.4)
during the TFR phase and musculoskeletal pain AEs were transient. Pts should
continue to be regularly monitored during TFR. Clinical trial information:
NCT01784068.

7014 Poster Discussion Session; Displayed in Poster Session (Board #389), 7015 Poster Discussion Session; Displayed in Poster Session (Board #390),
Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Mon, 4:30 PM-6:00 PM Mon, 4:30 PM-6:00 PM
Pooled safety summary for patients treated with the CD22-directed cytotoxin Inferior survival after microbiota injury: A multicenter allo-HCT study. First
moxetumomab pasudotox-tdfk. First Author: Robert J. Kreitman, Laboratory Author: Jonathan U. Peled, Memorial Sloan Kettering Cancer Center, New
of Molecular Biology, NCI, NIH, Bethesda, MD York, NY
Background: Moxetumomab pasudotox-tdfk was FDA-approved in 2018 for Background: Relationships between microbiota composition and clinical
the treatment of adults with relapsed/refractory hairy cell leukemia (HCL) outcomes following allogeneic hematopoietic cell transplantation (allo-HCT)
who have received $2 prior therapies, including a purine nucleoside analog. have been described in single-center studies. Geographic variations in human
Methods: The pooled integrated safety summary included all adult patients microbial communities and differences in clinical practices across institutions
(N=165) treated with $1 dose of moxetumomab across 5 open-label, single- raise the question of whether these associations are generalizable. We report
arm clinical trials. Two trials included patients with HCL (N=129, with N=80 the first multi-center study of the intestinal microbiota in allo-HCT.
in the pivotal phase 3 trial) and 3 trials included patients with non-Hodgkin Methods: Intestinal communities in 8,768 fecal samples from 1,362 allo-
lymphoma (N=15) or chronic lymphocytic leukemia/small lymphocytic HCT patients at 4 centers on 3 continents were profiled by 16S sequencing.
lymphoma (N=21). Overall, 81% were male, median age was 60 years (range Associations between microbiota composition and clinical outcomes were
34–84) and 94% had an ECOG #1. Results: The most common adverse analyzed with proportional-hazards analysis in an observational study.
events (AEs) with moxetumomab were peripheral edema (41%), hypo- Results: We observed reproducible patterns of microbiota injury characterized
albuminemia (35%), and nausea (35%). Serious AEs occurred in 47 (28.5%) by loss of diversity and domination by single taxa. Low diversity in the neu-
patients and were considered treatment-related in 21 (12.7%) patients. trophil engraftment period was reproducibly associated with increased risk of
Treatment-related AEs were mainly grade 1/2; the most common were pe- death (multivariate HR 0.48 [0.30-0.77] p = 0.002 in the largest cohort).
ripheral edema (32%), hypoalbuminemia (32%), increased ALT (29%), in- These reductions in OS were in part due to an increased risk of transplant-
creased AST (28%), nausea (26%), headache (24%), pyrexia (23%), fatigue related mortality and graft-vs-host disease. Baseline pre-HCT samples already
(22%), and myalgia (20%). In total, 19 deaths (11.5% of patients) occurred bore evidence of microbiome disruption; low diversity prior to transplantation
(including 3 in the pivotal trial), with most (13) due to underlying disease and 6 was associated with poor survival. A bacterial-composition risk score that was
due to AEs (none were assessed as treatment-related by the investigators, trained in one cohort predicted mortality in the other three cohorts (multivariate
however, 1 fatal AE in a non-HCL patient was re-assessed by the Sponsor as HR 1.42 [1.04-1.93] p = 0.03), indicating that not only a diversity metric but
possibly related to moxetumomab). In 16 patients (9.7%), a treatment-related also a signature of specific bacterial abundances is informative about post-HCT
AE led to treatment discontinuation, with hemolytic uremic syndrome HUS; mortality risk across independent institutions. Conclusions: We demonstrate a
3.6%) and capillary leak syndrome (CLS; 2.4%) the most common. Treatment- relationship between microbiota and survival after allo-HCT that is in-
related AEs resulted in dose delay, omission or treatment interruption in 11 dependent of transplant center and geographic location. The diversity of
patients (6.7%). Conclusions: Moxetumomab had an acceptable safety profile clinical practices across institutions imposed significant heterogeneity in the
based on pooled data from 165 adult patients with hematologic malignancies, study, yet we observed reproducible microbiota injury patterns and associa-
with few treatment-related discontinuations. Proactive monitoring of patients tions with outcomes. This concordance suggests that approaches to manip-
is important to manage AEs. These results are consistent with results from the ulate the intestinal microbiota in allo-HCT may be generalizable.
pivotal phase 3 trial in patients with HCL. Clinical trial information: NCT01829711,
NCT00586924, NCT00587457, NCT00587015, NCT01030536.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
410s Hematologic Malignancies—Leukemia, Myelodysplastic Syndromes, and Allotransplant

7016 Poster Discussion Session; Displayed in Poster Session (Board #391), 7017 Poster Discussion Session; Displayed in Poster Session (Board #392),
Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Mon, 4:30 PM-6:00 PM Mon, 4:30 PM-6:00 PM
Development and validation of a novel disease risk model for patients with Impact of geriatric vulnerability on outcomes of older patients in allogeneic
AML receiving allogeneic hematopoietic cell transplantation. First Author: hematopoietic cell transplantation. First Author: Richard Jirui Lin, Memorial
Piyanuch Kongtim, Faculty of Medicine Thammasat University, Pathumthani, Sloan Kettering Cancer Center, New York, NY
Thailand
Background: Older patients are at increased risk for complications and death
Background: Molecular data and minimal residual disease (MRD) have been following allogeneic hematopoietic cell transplantation (allo-HCT). Traditional
shown to influence outcomes in AML patients undergoing allogeneic hema- transplant-specific prognostic indices such as hematopoietic cell transplant
topoietic cell transplantation (AHCT). Here we developed and validated a novel comorbidity index (HCT-CI) may not capture all underlying geriatric vulnera-
AML-specific Disease Risk Group (AML-DRG) and revise our previously de- bilities, and in-depth evaluation by a geriatrician prior to transplant may not
veloped Hematopoietic Cell Transplant - Composite Risk (HCT-CR) model by always be available. We hypothesize that routine pre-transplant assessments
incorporating molecular data and MRD status before transplant to predict post- by interdisciplinary clinical providers may help uncover additional geriatric
transplant outcomes of patients with AML. Methods: The study included 1414 deficits. Methods: Using an institutional database of 457 adults age 60 years
consecutively treated adult AML patients, who received first AHCT between and older (range 60-78.7) who underwent first allo-HCT for hematological
1/2005-8/2018 at our institution. Patients were randomly assigned into a malignancies from 2010 to 2017, we retrospectively examined the prevalence
training (N = 944) and validation set (N = 470). To develop the AML-DRG and the prognostic impact of pre-transplant geriatric deficits identified by
model, the coefficient of all significant AML-related variables in multivariable interdisciplinary clinical providers including geriatric domains of functional
Cox’s regression analysis (MVA) in a training dataset was converted into activity, cognition, medication, nutrition, mobility, and routine laboratory tests.
scores.The AML-DRG was the sum of scores from all significant covariates, Results: With a median follow-up of 37 months for survivors, the 3-year
while the AML-HCT-CR was the sum of disease-related factors assessed by the probability of overall survival (OS) was 50% (95% CI 45-55). The 2-year
AML-DRG model with the addition of weighted scores from patient-related cumulative incidence of non-relapse mortality (NRM) was 25% (95% CI 22-
factors. Results: Based on results of MVA, the following scores were assigned 28). Among pre-transplant geriatric variables, we found that impairment in
to each AML-related variable; 1 point to secondary AML; 1 point to the 2017 instrumental activities of daily living (IADL) was associated with increased
ELN adverse risk; 2 points to CR with MRD positive/unknown; and 4 points to NRM and inferior PFS and OS. In multivariate analyses, mismatched donor,
active disease at transplant. These were used to generate the AML-DRG (low- age-adjusted HCT-CI . 4 (aaHCTCI), and IADL impairment were associated
risk score 0-2; intermediate-risk score 3-4; and high-risk score . 4) with with NRM, while high/very high disease risk index (DRI), IADL impairment, and
significantly different OS with HR of 2.02 (P , 0.001), and 3.85 (P , 0.001) positive CMV status were associated with OS. The combination of IADL im-
for intermediate and high risk group compared with low risk group. By adding 1 pairment with either aaHCTCI or DRI readily stratifies NRM and OS, re-
point for patients . / = 60 years or HCT-CI . 3 to the AML-DRG, we created 4 spectively. Conclusions: Our findings establish a simple assessment tool to risk
risk groups of AML-HCT-CR (low-risk: score 0-2; intermediate-risk: score 3; stratify older patients prior to allo-HCT using IADL and aaHCTCI and DRI.
high-risk: score 4 and very high-risk: score $5) with distinct survival outcomes. These results may provide an entry point for prospective, interventional trials to
The AML-DRG and AML-HCT-CR model had C-index of 0.672 and 0.715, reduce NRM and toxicities for older allo-HCT patients.
respectively which were better compared with DRI, ELN2017 and cytogenetic
risk model. Conclusions: Prognostic models incorporating molecular data and
MRD status before transplant allow better stratification and improved survival
estimates of AML patients post-transplant.

7018 Poster Discussion Session; Displayed in Poster Session (Board #393), 7019 Poster Discussion Session; Displayed in Poster Session (Board #394),
Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session,
Mon, 4:30 PM-6:00 PM Mon, 4:30 PM-6:00 PM
Adult Langerhans cell histiocytosis: A contemporary single-institution series Correlation of BRAF V600E mutation with cardiac involvement assessed by
of 186 patients. First Author: Gaurav Goyal, Mayo Clinic, Rochester, MN heart imaging in a monocentric series of 205 patients with Erdheim-Chester
disease. First Author: Marine Bravetti, Departement of Radiology, Hôpital
Background: Langerhans cell histiocytosis (LCH) is a rare histiocytic neoplasm
driven by MAPK-ERK mutations in majority of patients. Contemporary data on Pitié-Salpêtrière, APHP, Sorbonne Université, Paris, France
treatments and outcomes in adult LCH are lacking. Hence, we undertook this study Background: Erdheim–Chester disease (ECD), an inflammatory myeloid neo-
to analyze a large cohort of adult LCH patients. Methods: This was a retrospective plasm, is an histiocytosis associated with multisystem infiltration. Cardiovas-
study of adult ($18 years) LCH patients seen at our institution between 1998 and cular involvement in ECD is under-diagnosed and associated with poor outcome.
2018. Results: We included 186 patients with adult LCH (median age 43; 19-88), The targetable BRAFV600Emutation is present in up to 70% of all ECD.
and 54% were females. 70% of patients were diagnosed after 2007. Common
Methods: Retrospective study of 205 patients (pts) with ECD who had cardiac
presenting symptoms were cough/dyspnea (30%), rash (17%), pain/swelling in head
imaging (195 MRI, 10 CT when MRI was contraindicated). We identified the
(17%), and diabetes insipidus (10%). 70 (38%) patients had multisystem LCH, 62
(33%) had isolated pulmonary LCH, and 35 (19%) had unifocal LCH. Common sites
types of lesions (infiltration, tumor, and effusion), localization (pericardial,
of involvement included lung (59%), bone (37%), skin (21%), and nervous system myocardial, valvular) and consequences on cardiac function (coronary stenosis,
(16%). 121 (65%) were smokers; 48% of these had lung disease, while 52% had atrial wall dyskinesia, diastolic and systolic functions). Results: 141 (68.8%) pts
multisystem disease. 18 of 31 tested (58%) patients had BRAF-V600E mutation. were male. 30 (14.6%) had mixed histiocytosis (mainly ECD + langerhans cell
Most common first-line treatment was smoking cessation in 24 patients, and led to histiocytosis). BRAF mutation (BRAFm) was found in 112 (54.6%) cases, while
an overall response rate (ORR) of 83% in pulmonary lesions. Radiation therapy was 59 pts (28.8%) were Wild Type (WT) and 34 pts (7.6%) had unknown BRAF
used in 11 patients, and led to an ORR 82%. Surgical resection of lesion was done in status. Among the 205 cardiac imaging, 101 (49.3%) were abnormal. Cardiac
23 patients, with relapses in 24%. Systemic therapies were used in 78 (42%) involvement was found in 93 pts (49%). Among these, 72 had an impairment of
patients (Table). Most common first-line systemic therapy was cladribine with ORR the right ventricular atrioventricular sulcus (74%), 65 of the right atrium (RA)
of 78%. Vemurafenib was used in 3 patients with BRAF-V600E, leading to an ORR enclosure (69%). Alteration of Tricuspid Annular Plane Systolic Excursion was
of 67% . After a median follow-up of 23 months (0-261), 21 patients had died. Of found in 15% and correlated with the size of the tumor. Pericardial involvement
these, 10 died of progressive LCH. Median OS was not reached, and mean OS was (effusion, thickening or contrast enhancement) was found in 59 pts (29%).
196 months. Conclusions: This is the largest contemporary series of adult LCH. It Among BRAFm pts, 75 (67%) had a heart abnormality while 37 (33%) had
shows that diverse clinical spectrum, ranging from benign course to a progressive normal imaging; Among WT pts 14 (23.7%) showed heart abnormality, whereas
multisystem disease. Although smoking cessation was an effective treatment for 45 (76.3%) had normal imaging (RR 2.8 (CI: 1.8-4.5); p = 1.8*10-7). A RA
pulmonary LCH, a large subset required systemic chemotherapy.
tumor was present in 51 (45.5%) BRAFm but only 6 (10.2%) WT pts re-
1st line ORR Median remis- Relapse Subsequent ORR Median remis- Relapse spectively (RR 4.5 (CI : 2.0-9.8); p = 7*10-6). BRAFm was also associated with
Systemic Treatment (n) % sion (mo.) % line (n) % sion (mo.) %
aortic infiltration (RR 1.76 (CI: 1.2–2.5)) and pericardial involvement (RR 2.12
Cladribine 18 78 28 (5 -124) 21 16 69 67 (16 - 164) 9 (CI: 1.1–3.9), p = 0.0017). Conclusions: Cardiac infiltration is frequent in ECD
Vinblastine + 5 80 18 (12- 28) 75 6 17 9 (9-9) 100
prednisone (49.3%), especially RA tumor. BRAFm is associated with RA, aortic and
Vemurafenib 1 100 13 (13-13) 0 2 50 19 (19-19) 0 pericardial involvements.
Prednisone 9 0 - - - - - -
Prednisone + smoking 10 27 63 (14 - 73) 0 - - - -
cessation
Cytarabine - - - - 3 33 5 (5-5) 100
Etoposide 3 0 - -
Methotrexate - - - - 2 100 16 (15- 16) 0
Bisphosphonates - - - - 3 100 26 (5 - 29) 33

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Hematologic Malignancies—Leukemia, Myelodysplastic Syndromes, and Allotransplant 411s

7020 Poster Discussion Session; Displayed in Poster Session (Board #395), 7021 Poster Session (Board #396), Mon, 8:00 AM-11:00 AM
Mon, 1:15 PM-4:15 PM, Discussed in Poster Discussion Session, Evaluation of etiologic genetic polymorphisms by whole exome sequencing
Mon, 4:30 PM-6:00 PM (WES) in the Mayo Clinic AML epidemiology cohort (MCAEC). First Author:
Clinical features, molecular aberrations, treatments, and outcomes in histio- Zaid Abdel Rahman, Mayo Clinic, Jacksonville, FL
cytic sarcoma. First Author: Gordon Ruan, Mayo Clinic Rochester, Rochester,
Background: Case-control studies (CCS) have identified germline variants
MN
(GVs) & genes associated with AML risk and/or outcome. However, whether
Background: Histiocytic sarcoma (HS) is a rare and aggressive malignancy of the these genes also harbor somatic mutations (SMs) in AML and whether these
monocyte/macrophage lineage. There is a paucity of data on treatments and SMs are associated with clinical phenotypes and outcome is unknown. We
outcomes in HS. Methods: This is a retrospective study of patients with histo- systemically interrogated PubMed (terms "AML genetic polymorphisms"),
logically confirmed diagnosis of HS from 2000-2018. Kaplan-Meier and log-rank yielding 612 reports (1981-2018), from which we identified unique variants/
tests were used to perform overall survival (OS) analyses. Hazard ratios (HR) with single nucleotide polymorphisms (SNPs) in 100 genes associated with AML
confidence intervals (CI) were calculated using Cox-proportional hazards. risk and/or outcome in CCS. To determine the prevalence and relevance of SMs
Results: There were 27 patients in the study (median age 59; range 3-83) and
in these genes in an unselected AML population, we performed an analysis
63% were males. Nine had unifocal involvement and 18 had multifocal disease.
using WES of cases in the highly-annotated MCAEC (Finn, Ca Epidemiol
Common sites involved were lymph node (50%), soft tissue (40%), bone (36%),
and bone marrow (22%). Two of 8 patients had the BRAF-V600E mutation. Next- 2015). Methods: We performed WES using remnant cytogenetic pellets from
generation sequencing was performed on 8 patients and showed $1 oncogenic AML diagnosis in 98 MCAEC patients with available samples, sequenced at a
alterations (DNMT3A, FLT4, KRAS, NRAS, NUP98, PTCH1, PTPN11, TP53, depth of ~100 million paired-end 100bp reads using Agilent SureSelectXT
TSC1, PDGFR, or BRAF genes) as well as a CLIP2-BRAF fusion. The median OS Human All Exon V5 + UTRs target enrichment kit. Sequencing reads were
for the cohort was 12 months. Factors associated with worse OS included mul- aligned to human reference genome, and SMs including non-synonymous (NS)
tifocal disease (OS 10 vs. 125 months, p = 0.03; HR 5.0, CI 1.1-21.8) and and truncating single nucleotide variants (SNVs) and small INDELs were
marrow involvement (OS 3 vs. 43 months, p = 0.003; HR 8.9, CI 1.7-45.0). identified and filtered using Exome Sequencing Project, 1000 genome,
Twelve (44%) had surgery with median OS of 42 months (range 1-125). Fifteen HapMap, & Mayo Clinic (MC) internal Biobank GV database. Copy number
had chemotherapy with median OS of 12 months (range 2-67; Table). aberrations (CNAs) were identified & filtered using public CN polymorphism
Conclusions: HS is an aggressive neoplasm, with multifocal disease and bone databases. For this analysis we evaluated the prevalence of somatic SNVs,
marrow involvement portending worse outcomes. Most patients have somatic INDELs, & CNAs in the 100 nominated genetic risk genes in collaboration with
oncogenic alterations involving the MAPK-ERK pathway and other genes. Che- MC Cancer Center to determine their association with clinical phenotype and
motherapeutic regimens have variable response rates but are not durable. More outcome. Results: Of the 100 unique genes, there were 40 with SM in our
studies on targeted kinase inhibitors in HS are needed to improve outcomes. dataset, with 126 total mutations. Most were mutated in a very small number of
Median # of Time to Progression patients, allowing insufficient statistical power to perform association ana-
Chemotherapy N cycles (months) lyses. However, an NS mutation in CYP2B6 (c.785A . G, p.Lys262Arg) had a
Alemtuzumab 1 11 3 minor allele frequency of 12.0%; this variant has been reported in association
Allogenic stem cell transplantation after carmustine, etoposide, 3 NA 14 (6-41)
cytarabine, and melphalan
with breast cancer risk, but not with AML. Conclusions: We studied SM’s in
Cladribine 6 4 8.5 (1-12) AML genetic risk genes using WES of tumor samples. While representing a
Cyclophosphamide, etoposide, procarbazine, and prednisone 1 6 9 smaller sample size, this suggests their contribution to AML development may
Cytarabine and idarubicin 1 1 0.5
Doxorubicin and ifosfamide 1 4 3 be limited in a broader clinical population. These results & prevalence of the
Gemcitabine and docetaxel 3 3 7 (2-23) novel CYP2B6 rs2279343 variant in AML must be confirmed in larger cohorts.
Ifosfamide, carboplatin, and etoposide 1 2 0.5
Rituximab, cyclophosphamide, hydroxydaunorubicin, vincris- 1 6 3
tine, and prednisone

7022 Poster Session (Board #397), Mon, 8:00 AM-11:00 AM 7023 Poster Session (Board #398), Mon, 8:00 AM-11:00 AM
Granulocyte transfusions in patients with skin and soft tissues infections and Association of the GATA3 rs3824662A allele with clinical outcomes in adult
leukemia: Are they useful? First Author: Yamil Michelen, University of Texas patients with adult B-ALL. First Author: Paul B. Koller, The University of
MD Anderson Cancer Center, Houston, TX Texas MD Anderson Cancer Center, Houston, TX
Background: Granulocyte transfusions (GTX) have been proposed to improve Background: Inherited GATA3 variants (rs3824662) have been described in
clinical outcome of neutropenic patients (pts) with serious infections. We higher frequency in both adults and children with Ph-like acute lymphoblastic
evaluated the impact of unirradiated GTX in pts with skin and soft tissue leukemia (Ph-like ALL) (Perez-Andreu, Nat Genetics 2013; Jain, ASH 2017).
infections (SSTI) and leukemia. Methods: We did a retrospective analysis of The clinical outcomes of Ph-like ALL are well known, and as we have previously
pts with leukemia and SSTI that received GTX from 2014 to 2018. We an- described, it as a high-risk subtype of ALL in both children and adults (Roberts,
alyzed infection outcome and changes in ANC after GTX. Donors were mo- NEJM 2014; Roberts, JCO 2017; Jain, Blood 2017). However, the clinical
bilized using G-CSF plus oral dexamethasone 12 hours prior to apheresis. outcomes of ALL patients with different germline variants of GATA3 that are
Results: Twenty-seven pts received 141 GTX for 33 SSTI. Transfusions were commonly seen in Ph-like ALL is unknown. Methods: Of the newly diagnosed
unirradiated, except for 10 (7%) radiated units administered due to avail- patients treated at MD Anderson Cancer Center (MDACC) with B-ALL, we
ability. Twenty pts were male (74%); median age was 59 yrs (20–83 yrs). performed analyses for the GATA3 rs3824662 variant in 85 patients (42 Ph-
Hematologic diagnoses included AML (23, 70%), MDS (3, 9%), ALL (3, 9%), like ALL, 43 non-Ph-like ALL). Testing for Ph-like ALL, CRLF2 rearrangement,
CLL (2, 6%), CML (1, 3%), and MF (1, 3%). In 24 (73%) SSTI, pts had a and GATA3 genotypes was performed as previously described (Jain, ASH
baseline ANC,0.5 x109/L, 3 (9%) had ANC between 0.51x109L to 0.99 x 2017). Results: Of the 85 patients, the rs3824662 AA genotype was iden-
109L, and 6 (18%) with ANC .1 x109/L. After GTX, ANC increased in 99 tified in 22 patients, the AC genotype in 33 patients, and the CC genotype in 29
(70%) cases by a median of 0.7 x 109/L (0.02 to 10.03) with a median peak patients. There was a trend towards increased frequency of Hispanic heritage
time of 9 hrs (4 to114 hrs), with a median time from GTX to first drop of 34 hrs in the AA genotype (86%) vs AC genotype (61%), vs the CC genotype (31%).
(10 to 136 hrs). ANC decreased in 27 (19%) (by a median -0.5 x 109/L,-0.02 Additionally, amongst patients with the AA genotype, 91% had the Ph-like ALL
to -2.41). There was no change of ANC in 15 (11%). Improvement was de- vs 45% of the AC genotype vs 10% of the CC genotype. The median white blood
termined by reduction in fever and inflammation, or resolution of SSTI 7 days count at diagnosis was 52.1 x 109/L in the AA genotype group, 20 x 109/L in
after first GTX. Twenty-seven (82%) episodes improved. In those improved, the AC genotype group, and 5.9 x 109/L in the CC genotype group. The 5 year
ANC remained ,1.5x109/L in 15 (56%) episodes after last GTX and after event-free-survival (EFS) of the CC group was 55% vs 36% of the AC group vs
improvement. Main adverse reactions were fever in 21% (29), and respiratory 14% of the AA group. The EFS between the 3 genotypic groups was statistically
complications in 6% (9) (pulmonary effusion, respiratory distress and acute significant with a p value of 0.018. The 5 year overall survival (OS) of the
hypoxemia). One pt (3%) required intubation. There was no transfusion-related CC group was 64% vs 42% of the AC group vs 14% of the AA group.
GVHD. Cumulative survival from first GTX on each SSTI by Kaplan-Meier Conclusions: We observed significant differences in the EFS between patients
Survival was 82% (8) at 30 days, 49% (17) at 90 days, 41% (17) at 180 days, with 0, 1, 2 copies of this genetic polymorphism in the GATA3 gene. The role of
and 24% (12) at 360 days. Eight of this pts died of infections, 4 of which were genomic AA GATA3 variant on the biology of ALL and understanding of the
SSTI described on this paper. Conclusions: GTX have a beneficial effect on downstream mechanisms needs to be determined.
clinical improvement in patients with SSTI and underlying hematological
malignancies and severe neutropenia. It can be safely administered without OS EFS
GVHD. Further prospective studies are warranted. Genotype 2 yr % 5 yr % Median (Months) 2 yr % 5 yr % Median (Months)
CC 73 64 NR 73 55 NR
AC 64 42 32 42 36 21
AA 67 14 29 33 14 19

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
412s Hematologic Malignancies—Leukemia, Myelodysplastic Syndromes, and Allotransplant

7024 Poster Session (Board #399), Mon, 8:00 AM-11:00 AM 7025 Poster Session (Board #400), Mon, 8:00 AM-11:00 AM
Allogeneic hematopoietic stem cell transplantation contributes to eradica- Time from randomization to first subsequent induction/salvage therapy (ST)
tion of minimal residual disease in adult Philadelphia chromosome-positive in patients with relapsed/refractory (R/R) acute lymphoblastic leukemia (ALL)
acute lymphoblastic leukemia patients treated with dasatinib. First Author: treated with inotuzumab ozogamicin (InO) in the phase 3 INO-VATE trial. First
Jeremy Chang, University of Southern California + LA County Medical Center Author: Elias Jabbour, The University of Texas MD Anderson Cancer Center,
Internal Medicine Residency, Los Angeles, CA Department of Leukemia, Houston, TX
Background: Philadelphia Chromosome-Positive (Ph+) disease is associated Background: Additional salvage regimens are burdensome for R/R ALL patients (pts)
with a poor prognosis in acute lymphoblastic leukemia (ALL). Recent studies due to therapy-associated toxicities, which may affect quality of life with poor ef-
have shown that the eradication of minimal residual disease (MRD) in this ficacy. In INO-VATE (Kantarjian et al, NEJM 2016), pts receiving InO vs standard
population leads to improved survival outcomes. While hematopoietic stem chemotherapy (SC) had significantly greater remission rates and longer overall
cell transplantation (HSCT) has demonstrated clinical benefits in Ph+ ALL survival; after study treatment discontinuation, overall fewer STs (mainly burden-
patients treated with the tyrosine kinase inhibitor (TKI) imatinib, its role is some chemotherapies) were used in the InO group vs SC. Methods: Study design was
published. Adults with CD22+ ALL who were due to receive 1st or 2nd salvage
less clear with the use of more potent, newer-generation TKIs such as
treatment were randomized 1:1 to InO (n = 164) or SC (n = 162). Time to the first ST
dasatinib. Methods: This was a retrospective study analyzing the impact of
(TST) was time from randomization to the start of the first ST. Pts who did not receive
allogeneic HSCT on MRD status in Ph+ ALL patients treated with dasatinib. any ST were censored. Data cutoff: Jan 4, 2017. Results: Fewer InO pts had ST vs SC
Patients were divided into 2 groups: those treated with chemotherapy plus pts (Table). Among the censored pts, 83/108 (76.9%) in the InO vs 54/69 (78.3%)
dasatinib followed by allogeneic HSCT and those who received chemo- in the SC group died, 23/108 (21.3%) vs 5/69 (7.2%) were alive at the end of the
therapy plus dasatinib alone. All patients underwent bone marrow biopsy study, and 2/108 (1.9%) vs 10/69 (14.5%) were no longer being followed for
with MRD analysis following induction therapy and subsequent re-evaluation survival. TST was longer in pts receiving InO vs SC. Overall median (95% CI) TST was
of MRD status at day 100 post-transplant in the HSCT group and after further 18.8 (14.7–NA) vs 3.9 (2.4–5.1) months; hazard ratio (HR) = 0.34, 97.5% CI:
cycles of chemotherapy plus dasatinib in the non-transplant group. MRD- 0.23–0.49, 1-sided P, 0.0001. For pts (InO: 85, SC: 126) who never received
negative disease was defined as the absence of a BCR-ABL1 transcript by hematopoietic stem cell transplantation (HSCT) on study, fewer InO pts had ST vs SC
real-time quantitative polymerase chain reaction (qRT-PCR) with a sensi- pts; TST was significantly longer in InO vs SC. For pts (InO: 70, SC: 18) who had
tivity of 0.01%. Results: A total of 51 adult Ph+ ALL patients with MRD- HSCT directly after salvage, TST was extended in InO vs SC. Conclusions: In this
positive disease following induction therapy were included. Twenty-seven study, treatment with InO provided the benefit of extended TST, effectively allowing
patients (53%) were male and the median age at time of diagnosis was 42 patients a longer time period until an ST was needed in both patients who proceeded
years (range 23-68). There were 29 patients in the transplant group and 22 to as well as those who did not proceed to HSCT. Clinical trial information:
patients in the non-transplant group. When analyzing rates of MRD eradi- NCT01564784.
cation, 18 (62%) patients in the transplant group were found to have MRD- Total, Had % (n/ 1-sided
negative disease at day 100 post-transplant compared to 7 (32%) patients in N ST, n N) TST, median (95% CI), months HR (97.5% CI) P

the non-transplant group who only received further cycles of chemotherapy Overall 0.336 ,
(0.228–0.494) 0.0001
plus dasatinib (risk ratio 0.56, 95% confidence interval 0.32-0.96, p = InO 164 56 34.1 18.8 (14.7–NA)
0.048). Conclusions: In the era of newer-generation TKIs, allogeneic HSCT SC 162 93 57.4 3.9 (2.4–5.1)
Never had HSCT 0.461 0.0001
continues to have notable benefits in Ph+ ALL such as a significantly higher (0.283–0.752)
rate of MRD eradication as demonstrated in this study. InO 85 32 37.6 11.5 (6.8–23.0)
SC 126 66 52.4 3.0 (2.1–5.0)
Had HSCT directly after 0.477 0.0370
study treatment (0.184–1.233)
InO 70 15 21.4 NA (KM curve . 50%, median
not reached)
SC 18 9 50.0 16.0 (12.3–NA)

7028 Poster Session (Board #403), Mon, 8:00 AM-11:00 AM 7029 Poster Session (Board #404), Mon, 8:00 AM-11:00 AM
Ivosidenib (IVO; AG-120) in IDH1-mutant newly-diagnosed acute myeloid Inotuzumab ozogamicin (InO) treatment in patients with relapsed/refractory
leukemia (ND AML): Updated results from a phase 1 study. First Author: acute lymphoblastic leukemia (R/R ALL): Outcomes of patients treated in
Gail J. Roboz, Weill Cornell Medical College, New York, NY salvage one with a long duration of first remission. First Author: Hagop M.
Kantarjian, The University of Texas MD Anderson Cancer Center, Houston, TX
Background: IVO is an oral, potent, targeted inhibitor of mutant IDH1 (mIDH1)
approved for the treatment of adults with mIDH1 relapsed or refractory AML. Background: In R/R ALL, data are limited supporting targeted therapies like
We report the safety and efficacy of IVO in pts with ND AML not eligible for InO as first salvage in patients (pts) who had a long first complete remission
standard therapy enrolled in the first-in-human, phase 1, dose escalation and (CR1). InO, a calicheamicin-conjugated antibody, targets CD22 on ALL blasts.
expansion study of pts with mIDH1 advanced hematologic malignancies Here we report outcomes in pts with ALL in first salvage (S1) who had a long
(NCT02074839). Methods: Enrollment completed on 08May2017. Overall CR1 before receiving InO vs standard of care chemotherapy (SC).
response rate (ORR) comprised complete remission (CR) + CR with incomplete Methods: Adults with CD22+ R/R ALL, stratified by salvage and length of
hematologic or platelet recovery + partial response + morphologic leukemia- remission (CR1 $ or , 12 mos), were randomized to receive InO (n = 164) or
free state. CR with partial hematologic recovery (CRh) was CR except absolute SC (n = 162). Methods were previously described (Kantarjian et al, NEJM
neutrophil count . 0.5 3 109/L and platelet count . 50 3 109/L. 2016), with data shown up to the last patient last visit (January 4, 2017).
Results: Baseline characteristics for 34 pts with ND AML who received IVO Outcomes including complete remission (CR)/CR with incomplete hematologic
500 mg once daily were: 19 men/15 women; median age 76.5 yrs (range 64- recovery (CRi) and overall survival (OS) were determined for S1 pts who
87); 56% $75 yrs of age; 76% had secondary AML and 53% prior MDS; 47% achieved CR1 $ 12 mos and CR1 $ 18 mos. Results: For S1 pts with CR1
had $1 hypomethylating agent for an antecedent hematologic disorder. As of $ 12 mos or CR1 $ 18 mos, InO and SC arms had generally comparable baseline
02Nov2018, 7/34 (21%) pts remained on treatment; 3 (9%) had dis- characteristics. For S1 pts with CR1 $ 12 mos and CR1 $ 18 mos, fewer pts
continued treatment for allogeneic stem cell transplant. Median duration of had received a prior stem cell transplant in the InO vs SC arm (10.4% vs 25.0%
IVO exposure was 4.3 mo (range 0.3–40.9). Adverse events of any grade and and 9.7% vs 21.4%). For S1 pts with CR1 $ 12 mos and CR1 $ 18 mos
causality in $25% of pts were diarrhea (53%), fatigue (47%), nausea (38%), respectively, InO vs SC treatment led to a higher CR/CRi rate (85.4% vs 27.5%
decreased appetite (35%), leukocytosis, anemia, thrombocytopenia, periph- [P, 0.0001] and 83.9% vs 32.1% [P, 0.0001]) and improved OS (HR =
eral edema (all 26%). Most were grade 1–2 and unrelated to treatment. IDH 0.547 [P= 0.0086] and 0.504 [P= 0.0163]), with veno-occlusive liver disease
differentiation syndrome (DS) was seen in 6/34 (18%) pts, grade $3 in 3 (9%); reported in 12.5% and 12.9% of InO pts Clinical trial information:
IVO was held due to DS in 3 pts. QT prolongation was seen in 6 pts. In 33 pts NCT01564784. Conclusions: Improved outcomes were seen with InO vs SC
confirmed mIDH1-positive by the companion diagnostic test, CR rate was 30% among S1 pts who had a long first complete remission (CR1 $ 12 mos or
(95% CI 16%, 49%), CR+CRh 42% (95% CI 26%, 61%), and ORR 55% CR1 $ 18 mos), supporting the benefit of InO vs SC in this population.
(95% CI 36%, 72%), with median durations not estimable (95% CI lower HR
bound 4.2, 4.6, and 4.6 mo, respectively); 12-mo response durations were CR/CRi MRD-negativity* Median OS (97.5% CI) 24 mo survival
n (%) n (%) (95% CI) months 1-sided P probability % (95% CI)
78%, 62%, and 63%, respectively. Of 21 pts who were transfusion dependent
CR1 ‡ 12 months 41 (85.4) 37 (90.2) 10.3 0.547 37.4
at baseline, 43% became transfusion independent for $56 consecutive days InO (n = 48) (7.7-29.6) (0.31-0.97) (23.7-51.1)
on treatment. Updated co-occurring mutation and mutation clearance data will P= 0.0086
SC (n = 40) 11 (27.5) 5 (45.5) 7.7 11.7
be reported. Conclusions: IVO was well tolerated and induced durable re- (4.5-14.6) (3.7-24.7)
missions in elderly pts with poor-prognosis mIDH1 ND AML. Clinical trial CR1 ‡ 18 months
InO (n = 31)
26 (83.9) 25 (96.2) 10.3
(7.7-NE)
0.504
(0.24-1.05)
45.6
(27.2-62.2)
information: NCT02074839. P= 0.0163
SC (n = 28) 9 (32.1) 4 (44.4) 9.5 12.3
(4.7-16.5) (3.1-28.2)

*In pts achieving CR/CRi MRD = minimal residual disease, NE = Not evaluable, CI = confidence interval, HR = un-
stratified hazard ratio

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Hematologic Malignancies—Leukemia, Myelodysplastic Syndromes, and Allotransplant 413s

7030 Poster Session (Board #405), Mon, 8:00 AM-11:00 AM 7031 Poster Session (Board #406), Mon, 8:00 AM-11:00 AM
Results from the first-in-human study of mivebresib (ABBV-075), a pan-inhibitor Does intensity of induction chemotherapy affect the impact of measurable
of bromodomain and extra terminal proteins, in patients with relapsed/refractory residual disease (MRD) on prognosis in acute myeloid leukemia (AML)? First
acute myeloid leukemia. First Author: Olatoyosi Odenike, University of Chicago Author: Michael Jamie Hochman, University of Washington, Seattle, WA
Medical Center, Chicago, IL
Background: Absence of MRD in first complete remission (CR1) of AML
Background: Bromodomain and extra-terminal (BET) proteins bind to ace- decreases the probability of subsequent relapse both in patients given more
tyllysines and upregulate oncogenic target genes. Mivebresib (ABBV-075) is a (e.g. 7+3 or FLAG) or less (e.g. azacitidine) intense induction. Although high-
pan-BET inhibitor with antitumor activity in vitro and xenograft models of AML. intensity induction seems to increase the chance of CR without MRD, it is
This 2-part phase 1 study evaluates the safety and pharmacokinetics unknown whether induction intensity affects outcomes beyond eradication of
of mivebresib at monotherapy or combination dosing schedules in patients MRD. Methods: We retrospectively studied adults with newly diagnosed AML
with solid tumors (part 1) and acute myeloid leukemia (AML; part 2) or MDS with 10-20% blasts who received induction at the University of
(NCT02391480). Here, we report preliminary data from part 2 in patients with Washington from 2008 through 2015. Induction was classified as high or low
relapsed/refractory (RR) AML. Methods: Mivebresib monotherapy (MIV-mono), intensity, CR defined by standard criteria, and marrow MRD by presence of
or combined with venetoclax (MIV-VEN), were administered daily to adult blasts on multiparameter flow cytometry within 1-2 weeks of CR. Post-
patients with AML. The dose-limiting toxicity (DLT) period was 28 d. remission therapy in 10 patients included allogeneic hematopoietic cell
Results: As of Dec 2018, 41 patients (median age: 69 y [range, 29–84]; 19 transplant (HCT). Multivariate regression analysis examined the independent
patients had . 2 prior therapies) were enrolled: 19 in MIV-mono (5 of whom effects of age, initial cytogenetics (favorable, intermediate, adverse), perfor-
switched to MIV-combo) and 22 who began treatment in MIV-VEN cohorts. 23 mance status, de novo vs secondary AML, induction intensity, and MRD on
patients had high cytogenetic risk. Median time on treatment was 28 d (range, post-CR outcomes. Results: 217 patients received high- and 38 low-intensity
8–562). There were no DLTs. All patients experienced a treatment-emergent induction. 75% of the former vs. 42% of the latter achieved CR without MRD.
adverse event (AE), most commonly ($40% patient incidence), fatigue (56%), Although high-intensity therapy was associated with longer overall survival
dysgeusia (46%), decreased appetite (44%), diarrhoea (42%), nausea (42%), (OS), this difference disappeared after accounting for the above covariates
vomiting (42%). 40 patients had grade $3 AEs (febrile neutropenia (37%), (HR 0.94, 95% CI 0.58 1.52, P = 0.8) and OS was poor regardless of
anemia (34%) and thrombocytopenia (32%). 33 patients had serious AEs, chemotherapy intensity if MRD was present. Multivariate analysis found no
most commonly febrile neutropenia (19%). 25 deaths were reported; 15 interaction between the presence or absence of MRD and induction intensity
patients died of causes unrelated to mivebresib and 10 patients due to AML (HR 0.95, 95% CI 0.4 2.29, P = 0.9). Conclusions: Receiving high-intensity
progression. The median best % bone marrow blast change for 26 evaluable induction chemotherapy does not appear to confer a separate advantage in
patients was -20% (range, -98% to +300%). Gene expression analysis in pre- long-term outcomes once CR without MRD is achieved, at least in patients
and post-treatment peripheral blood samples showed that HEXIM1, DCXR and mostly not receiving HCT. We are currently including more such patients who
CD93 genes were reliable PD biomarkers of ABBV-075 which were consis- received HCT. Our results suggest that any effect of high-intensity induction
tently modulated in a dose-dependent manner. At the cutoff date, median largely reflects its association with a higher probability of achieving an MRD-
overall survival for all patients was 2.6 m. Conclusions: Mivebresib was well negative CR.
tolerated and showed antileukemic effects in patients with RR AML. Clinical
trial information: NCT02391480.

7032 Poster Session (Board #407), Mon, 8:00 AM-11:00 AM 7033 Poster Session (Board #408), Mon, 8:00 AM-11:00 AM
Nonclinical safety assessment of AMG 553, an investigational anti-FLT3 Clinical validation of long-range PCR assay for MLL partial tandem dupli-
CAR-T therapy. First Author: Christine M Karbowski, Amgen, Thousand cations for acute myeloid leukemia. First Author: Sun Hee Rosenthal, Quest
Oaks, CA Diagnostics Nichols Institute, San Juan Capistrano, CA
Background: FMS-like tyrosine kinase 3 (FLT3) is a tyrosine-protein kinase Background: Partial tandem duplication of the mixed lineage leukemia gene
involved in hematopoiesis. In acute myeloid leukemia (AML), higher FLT3 (MLL-PTD) occurs in about 3% to 5% of adult acute myeloid leukemia
transcript levels, independent of the presence of FLT3 mutations, correlate (AML), mostly in patients with normal karyotypes, and is associated with poor
with higher leukocyte counts and higher degrees of bone marrow infiltration prognosis. Real-time PCR on cDNA or RT-PCR followed by gel electro-
by leukemic cells. FLT3 protein is detectable on the cell surface of more than phoresis are the most commonly used methods for the detection of the
80% of leukemia isolates from adult AML patients whereas FLT3 mutations mutation. However, those methods require RNA extraction, which is not
are present only in approximately one third of patients. AMG 553 is a novel, often performed for molecular cytogenetic testing or next generation se-
investigational, adoptive cellular immunotherapy for the treatment of re- quencing (NGS) for AML. In this study, we developed an MLL-PTD detection
lapsed refractory AML, consisting of autologous T cells genetically modified assay that uses long-range PCR on extracted DNA, a commonly processed
ex vivo to express a transmembrane chimeric antigen receptor (CAR) to target sample type for AML molecular testing, to reduce sample processing re-
FLT3 protein on the surface of AML cells irrespective of FLT3 mutational quirements. Methods: DNA was extracted from de-identified 72 whole blood
status. Methods: The nonclinical safety evaluation of AMG 553 includes and 30 bone marrow aspirate specimens. Two PCR reactions were performed
assessment of FLT3 expression in normal tissues; in vitro assessment of for each DNA sample: one to detect MLL-PTD and another to detect CHEK2
cytotoxicity against human cells with or without FLT3 expression; toxicology to monitor DNA integrity. The 3 most frequent forms of MLL-PTD (exons 2~8,
studies in cynomolgus monkeys administered autologous T-cells transduced 2~9, and 2~10) were detected using a forward DNA primer on exon 8 and a
with an anti-FLT3 CAR or administered anti-FLT3 bi-specific T-cell engager reverse primer on exon 2. MLL-PTD positivity was determined by the in-
(BiTE) antibodies. Results: Assessment of FLT3 expression in normal tissues tensity of amplified DNA fragment at or above the lower detection limit
indicates that FLT3 protein is detected on the cell surface of a subpopulation control of this assay. Intra-assay precision, inter-assay precision, and limit of
of human bone marrow hematopoietic stem and progenitor cells; expression detection (LOD) were performed according to standard laboratory protocols.
in other tissues is cytoplasmic and not anticipated to be accessible to AMG Results of this method were compared to those of NGS. Results: Intra-assay
553. AMG 553-induced cytotoxicity was only observed in a percentage of precision studies on 5 specimens yielded 100% genotype concordance.
primary bone marrow CD34+ cells. There was no evidence of CAR-T cell- Inter-assay precision studies on 20 specimens yielded 100% concordance
mediated toxicity, expansion, or persistence in cynomolgus monkeys likely between replicates. LOD studies using a human eosinophilic leukemia cell
due to restricted cell surface FLT3 protein expression in healthy animals line (EoL-1) with 12.5-kb PTD in size showed assay sensitivity of 5%; for
which demonstrates the limited value of such studies for this target. All smaller MLL-PTD positives (PTD ~3 kb and ~5 kb), LOD was 1%. For method
findings in toxicology studies with anti-FLT3 BiTEs were consistent with comparison studies, 71 specimens (52 whole blood and 19 bone marrow)
the targeted killing of cells expressing FLT3 on the plasma membrane. were also analyzed by NGS; concordance between the methods was 97.2%
Conclusions: The nonclinical safety data support AMG 553 as a novel (69/71). Conclusions: We developed and validated long-range MLL-PTD
therapeutic to target FLT3 protein on AML cells irrespective of FLT3 mu- detection assay for AML. The developed method uses DNA, which is
tational status, while only affecting a percentage of normal hematopoietic commonly used for molecular testing for AML, and thus may reduce sample
stem and progenitor cells. processing time by consolidating sample type across tests.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
414s Hematologic Malignancies—Leukemia, Myelodysplastic Syndromes, and Allotransplant

7034 Poster Session (Board #409), Mon, 8:00 AM-11:00 AM 7035 Poster Session (Board #410), Mon, 8:00 AM-11:00 AM
Activity of venetoclax-based therapy in TP53-mutated acute myeloid leukemia. Mechanism of biologic therapy for pre B-cell acute lymphoblastic leukemia
First Author: Mahran Shoukier, University of Texas MD Anderson Cancer Center, with CRLF2 overexpression. First Author: WayAnne Watson, Loma Linda
Houston, TX University, Loma Linda, CA
Background: Mutations in TP53 are associated with low response rates to standard Background: Acute lymphoblastic leukemia (ALL) is the most common childhood
therapy and poor outcomes in patients (pts) with acute myeloid leukemia (AML). cancer, of which pre B-cell ALL with CRLF2 overexpression is a high risk subtype.
Combination therapy with the BCL2 inhibitor venetoclax (VEN) has emerged as an CRLF2 B-ALL is associated with poor outcome, increased rate of relapse, and
effective treatment option for pts with AML. Methods: We reviewed pts with TP53- health disparities in Hispanic children. Together with IL-7 receptor alpha (IL-7Ra),
mutated AML treated with VEN-based therapy between 2014-2018. Mutation CRLF2 makes up a receptor complex that is activated by the cytokine, thymic
testing was performed using a whole-exome next-generation sequencing panel. We stromal lymphopoetin (TSLP). Activation of the receptor complex leads to JAK2/
analyzed the characteristics of these pts, responses to therapy, and outcomes. STAT5 and PI3/AKT/mTOR signals that promote leukemia cell survival and pro-
Results: Sixty nine pts with TP53-mutated AML treated with VEN were identified, 36 liferation. To study the role of TSLP in CRLF2 B-ALL, we developed a patient-
(52%) in frontline & 33 (48%) in the salvage (R/R) setting (Table). The median follow
derived xenograft (PDX) model of CRLF2 B-ALL that allows us to alter circulating
up was 4.5 [0.5 - 48.5] and 8 [1 - 46.5] months for frontline & R/R pts, respectively.
levels of human TSLP (hTSLP). We generated PDX from CRLF2 B-ALL cells
Karyotype was complex in 32 (88%) and 29 (88%) pts in the frontline & R/R cohorts,
respectively. In the R/R cohort, the number of median prior treatments was 2 [0 – 8]. harvested from patient samples and compared leukemia burden in mice with
VEN was given in combination with: 1) Hypomethylating agents (HMA) (87%), 2) varying levels of hTSLP. In PDX models with hTSLP levels at or below physiological
FLAG-Ida (3%), 3) Low dose Ara-C (4%), or 4) CPX-351 (6%). The overall response levels in pediatric cancer patients, CRLF2 B-ALL cells grew robustly. However, in
rate (ORR) was 47% & 24% in frontline and R/R pts, respectively. All 6 pts with PDX with elevated levels of hTSLP, leukemia cells were essentially eliminated. Our
negative minimal residual disease (MRD) achieved complete cytogenetic response objective is to elaborate on the mechanism of high dose hTSLP’s antileukemic
after taking VEN % remain in complete remission (CR) with a median of 3.4 [1.7-4.7] effect. Methods: We performed TSLP dose response studies and used flow
months. Two pts (both R/R) underwent allogeneic stem cell transplantation. cytometry to evaluate the effect of TSLP on SOCS protein expression, CRLF2
Conclusions: VEN based therapy was associated with similar ORR, but higher CR signaling shutdown, and loss of TSLP receptor components. Results: CRLF2 B-
rates in TP53 mutated AML compared with HMAs alone. Larger studies with longer ALL cells cultured with hTSLP showed a dose-dependent loss in the ability to
follow up are needed to determine the role of VEN-based therapy in this difficult induce STAT5 and S6 phosphorylation following hTSLP stimulation. This loss was
subset. Patient characteristics and outcome. correlated with the loss of IL-7Ra, and maintained for 24-48 hours following a
Frontline (N=36) R/R (N=33)
pulse of high-dose, but not low-dose, hTSLP. The loss of signaling and surface IL-
Characteristics N (%) / Median [range] N (%) / Median [range] 7Ra could be prolonged if high-dose hTSLP levels were maintained. Flow
Age 74 [30 - 86] 67 [22 - 85] cytometry analysis showed that high-dose hTSLP upregulated SOCS1 and SOCS3
BM Blasts 25 [6-88] 32 [6-86]
Secondary AML 11 (31%) 4 (12%) proteins in CRLF2 B-ALL cells. Whole genome microarray showed that SOCS1,
Therapy related AML 10 (28%) 6 (18%)
Karyotype SOCS2, SOCS3 and CISH mRNA were upregulated in primary CRLF2 B-ALL cells
Complex (‡ 3 abnormalities)
Diploid
32 (88%)
1 (3%)
29 (88%)
1 (3%)
cultured with high dose hTSLP. Conclusions: These data support the hypothesis
Other 3 (9%) 3 (9%) that TSLP exerts its anti-leukemia effects through shutdown of CRLF2-mediated
VEN Combo
VEN + HMAs 33 (91%) 27 (82%) signals and that these effects are at least partially mediated by the loss of the IL-
VEN + FLAG -ida 0 2 (6%)
VEN + LDAC 3 (6%) 0 (3%) 7Ra component, and potentially through SOCS family proteins. These studies
VEN + CPX-351
Response
1 (3%) 3 (9%) elucidate the mechanism of the human TSLP cytokine as a potential biologic
CR 13 (36%) 5 (15%) therapy to treat CRLF2 B-ALL. Supported in part by 1R01CA209829 (KJP),
CR without count recovery (CRi) 4 (11%) 3 (9%)
ORR (CR/CRi) 17 (47%) 8 (24%) 1R43CA224723 (KJP), ASH HONORS Award 2018-2019 (WBW), and Alpha
MRD Negative 5 (14%) 1 (3%)
Median overall survival (months) 3.6 2.5 Omega Alpha Carolyn L. Kuckein Student Research Fellowship 2018-2019 (WBW).
6-month OS 36% 27%
Median CR duration (CRD) (months) 6.4 3.6
6-month CRD 51% 22%

7036 Poster Session (Board #411), Mon, 8:00 AM-11:00 AM 7037 Poster Session (Board #412), Mon, 8:00 AM-11:00 AM
Phase II results of mitoxantrone in combination with clofarabine in children Interim results from a phase Ib/II clinical study of the glutaminase inhibitor
with relapsed/refractory acute leukemia. First Author: Jessica Hochberg, telaglenastat (CB-839) in combination with azacitidine in patients with
New York Medical College, Valhalla, NY advanced myelodysplastic syndrome (MDS). First Author: Veronica Guerra,
MD Anderson Hematology/Oncology Fellowship, Houston, TX
Background: Despite improved outcomes in pediatric acute leukemia, those
who relapse or are refractory do poorly. Mitoxantrone and clofarabine as single Background: Glutaminase (GLS) is an enzyme that catalyzes conversion of
agents have proven efficacy in pediatric leukemia. Our previously reported glutamine to glutamate, providing key metabolic fuel for tumor cells. GLS is
phase I results demonstrated the MTD of this combination to be clofarabine highly expressed in AML and high-risk MDS, particularly in the setting of
35 mg/m2 x 5 days and mitoxantrone 12 mg/m2 x 4 days. Here, we present our complex cytogenetics. Preclinical studies of primary AML cells demonstrate
phase II data utilizing the RP2D. Methods: Prospective, open label safety and dependence on glutamine; glutaminase inhibition led to reduced cell growth
efficacy study (NCT01842672). Patients 0-30.99 yr old with ALL or AML with and induced apoptosis. This study was designed to evaluate the safety and
relapse OR induction failure were given 1 to 3 cycles of clofarabine (RP2D 35 efficacy of the oral glutaminase inhibitor CB-839 in combination with
mg/m2/day) Day 1-5, in combination with mitoxantrone 12 mg/m2/day on Day azacitidine for MDS. Methods: This is a single arm Phase Ib/II trial of CB-
3-6. Dexrazoxane was given prior to mitoxantrone. CNS prophylaxis was 839, evaluating the dose of 600 mg BID orally daily in combination with
achieved with intrathecal liposomal cytarabine. MRD was defined by flow azacitidine (AZA) for intermediate and high-risk MDS in 28-day cycles. The
cytometry ($ 0.01%). Results: Total 21 patients have enrolled at the RP2D primary outcome for the Phase 1 portion was to confirm the safety and
clofarabine. Median age 8 yrs (0.5-22 yrs). Patient characteristics include 13
recommended Phase 2 dose of CB-839 in combination with AZA. Secondary
ALL (8=Induction Failure, 4=Relapse 1, 1=Relapse 2) and 8 AML (4=Induction
endpoints evaluate efficacy and clinical activity using IWG response criteria
Failure, 4=Relapse 1). Overall regimen was well tolerated. One patient with
grade IV prolonged myelosuppression. Median time to neutrophil recovery was
for MDS including hematological improvement (HI), complete response
24 days. Eighteen of 21 (85.7%) patients achieved an MRD negative CR. Three (CR), marrow CR (mCR), stable disease (SD) and no response (NR).
patients died of progressive disease. All 18 patients achieving CR went on to Results: A total of 10 pts with MDS were enrolled; the Phase I portion is now
receive an allogeneic HSCT with no documented transplant associated toxicities complete, confirming CB-839 600 mg BID with standard AZA. Median age
attributed to prior MITCL therapy. One patient developed MRD (2.6%) at Day + was 71 [47-76], 90% were men. 7 pts were treatment naı̈ve, 3 pts had prior
270 post HSCT and underwent subsequent therapy. The remaining 17 pa- HMA exposure. 6 pts were intermediate-2, and 4 pts intermediate-1 by IPSS
tients continue to demonstrate MRD negativity. Two patients died of transplant with high-risk mutations. 3 pts had complex cytogenetics. The most frequent
associated complications. The overall and event free survival for patients who mutations were ASXL1 (n = 5), TET2 (n = 4) and TP53 (n = 4). Median
responded to therapy are 89% (CI95 0.64-0.98) and 83% (CI95 0.58-0.96), number of cycles was 3 [1-9]. The most common non-hematological AEs
respectively at a median follow up time of 24 months (range 2-58). were grade 1-3 nausea (60%) and constipation (50%). No DLTs were
Conclusions: The combination of clofarabine and mitoxantrone reinduction identified; one pt experienced grade 3 transaminitis requiring dose reduction
therapy for relapsed or refractory acute leukemia has been demonstrated to be of CB-839 (to 400 mg BID) with resolution. Seven pts (70%) achieved mCR/
safe and well tolerated at a RP2D of 35 mg/m2 clofarabine in children with poor CR with HI, 2 pts had SD, and 1 pt had NR. The median time to mCR was
risk acute leukemias. Phase II data is encouraging with 85.7% MRD negative 3 months [1-3]. 4 pts proceed to hematologic stem cell transplant (HSCT)
CR rate in leukemic patients allowing patients to safely proceed to AlloHSCT after 2-4 cycles. 1 pt progressed to AML after 3 cycles of therapy, and 3 pts
with a 2 year survival rate of 89%. Longer follow up and a larger cohort is remain on study. Conclusions: 600 mg BID CB-839 is safe and well tolerated
planned and ongoing. Clinical trial information: NCT01842672. in combination with azacitidine in pts with MDS. The Phase II portion is
ongoing. Clinical trial information: NCT03047993.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Hematologic Malignancies—Leukemia, Myelodysplastic Syndromes, and Allotransplant 415s

7038 Poster Session (Board #413), Mon, 8:00 AM-11:00 AM 7039 Poster Session (Board #414), Mon, 8:00 AM-11:00 AM
Identifying Mcl-1 protein dependencies using dimerization-specific antibody Impact of luteinizing hormone suppression on hematopoietic recovery after
biomarker for predicting response to targeted apoptosis inducing therapies. intensive chemotherapy in patients with leukemia. First Author: Iman
First Author: Michael H. Cardone, Eutropics, Pharmaceuticals Inc., Cam- Aboudalle, The University of Texas MD Anderson Cancer Center, Houston, TX
bridge, MA
Background: Treatment of acute leukemia with intensive chemotherapy (IC)
Background: The anti-apoptotic Bcl-2 family proteins facilitate pro-survival and leads to increased risk of infection and bleeding because of myelosup-
resistance to anti-cancer therapies. Measuring the function of these proteins has pression. Luteinizing hormone (LH) blockade was found to improve he-
shown utility in predicting response to therapy. A method that expands the matopoietic recovery in mice after radiation or chemotherapy through
principle of BH3 profiling to solid tumors is presented. Measuring the occurrence protection of the hematopoietic stem cells (HSCs) which express the LH
of heterodimers of Myeloid Leukemia Cell Differentiation Protein (Mcl-1), and receptor (Velardi et al, Nat Med 2018). We hypothesized that LH blockade
pro-apoptotic binding protein Bim is an indicator of cancer cell apoptotic priming improves hematopoietic recovery following IC in patients (pts) with leuke-
state. The readout of Mcl-1 containing complex-specific biomarkers can identify mia. Methods: We assessed gene expression of the LH receptor (LHR) in
survival dependencies in cancer cells potentially providing clinical utility in lineage-specific normal and AML hematopoietic cells from a reference
guiding cancer treatments. Methods: Engineered immunogens that recapitulate dataset (Corces et al, Nat Gen 2016). We conducted a retrospective analysis
conformation-specific epitopes induced during binding of the Mcl-1/Bim protein on pre-menopausal women with acute myeloid leukemia (AML) or acute
complex were used to generate. monoclonal antibodies. One Heterodimer lymphoblastic leukemia (ALL) who received IC and Lupron, given for pre-
Specific Mcl-1 Bim (HsMcB) was chosen. The selective binding was confirmed vention or treatment of abnormal uterine bleeding. Results: LHR was
using ELISA, fluorescence polarization, immunofluorescence microscopy and greatest in HSCs, with little or no expression in mature subtypes within the
flow cytometry in Bim or Mcl-1 knockdown cells, and by immunohistochemistry hematopoietic hierarchy. Surprisingly, LHR was expressed on blasts. Since
in formalin-fixed paraffin-embedded patient tissue. Correlation of HsMcB Lupron was more commonly given in younger pts, we performed propensity
measurements to BH3 profiling readouts from the Mcl-1 restricted Noxa peptide matching between the Lupron groups (AML N = 64; ALL N = 49) and control
was explored. Results: HsMcB signal depends on both Mcl-1 and Bim protein (Ctrl) groups (AML N = 128; ALL N = 98 pts). Baseline characteristics
levels. The ratio of the HsMcB to unbound protein (Mcl-1) signal ([HsMcB]/[Mcl- including blood counts were well balanced. Pts with AML who had received
1]) was measured. Disruption of the complexes by BH3 mimetics targeted to Lupron had a significantly higher increase in their platelet count following IC
Mcl-1 and depletion of Mcl-1 level using CDK9 inhibitors diminished the (13.8x109/L/year vs Ctrl; p = 0.02). Pts with ALL who had received Lupron
[HsMcB]/[Mcl-1] readout. We see correlations between these readouts and BH3 had significantly higher increase in their absolute neutrophil count
mimetic peptides readouts from the Mcl-1 specific Noxa and MS1 BH3 mimetic (0.37x109/L/year vs Ctrl; p = 0.02). AML pts in the Lupron group received
peptides in AML patient samples that have been treated with Mcl-1 inhibitors. significantly less blood transfusions vs Ctrl (mean: 23.9 vs 34.7 units; P =
Conclusions: Mcl-1 dependence is a predictive biomarker for venetoclax re- 0.002) and less platelet transfusions (mean: 24.4 vs 32.8 units; P = 0.06).
sistance and for response to Mcl-1 targeted therapies. Flow cytometric and IHC There was no difference in event-free and overall survival between the groups
based measurements of a heterodimer complex offer a direct and simpler ap- in each leukemia cohort. Conclusions: Lupron use in leukemia pts receiving
proach that harbors potential for use in clinical settings. Additional antibodies IC was associated with improved long-term blood count recovery. It was also
targeting Mcl-1/Bak and Mcl-1 Noxa complexes are being tested. associated with decreased transfusion requirements in AML. Despite ex-
pression of the LHR in blasts in addition to normal HSCs, there was no effect
of LH blockade on rates of leukemia relapse or death.

7040 Poster Session (Board #415), Mon, 8:00 AM-11:00 AM 7041 Poster Session (Board #416), Mon, 8:00 AM-11:00 AM
A video decision aid to improve acute myeloid leukemia patients’ illness IL2-STAT5 immune signatures to predict responses to PD-1 inhibition and
understanding: Results of a pilot trial. First Author: Thomas William LeB- azacitidine treatment in acute myeloid leukemia (AML): A subset analysis
lanc, Duke University Medical Center, Durham, NC of a phase 2 study. First Author: Hussein Abbas, University of Texas MD
Anderson Cancer Center, Houston, TX
Background: Many acute myeloid leukemia (AML) patients harbor mis-
understandings about their illness, overestimating both their likelihood of cure Background: Combining PD-1/PD-L1 blockade with hypomethylating agents
and risks of intensive therapies. Decision aids (DA) can improve illness un- (HMA) shows encouraging preliminary efficacy in AML, but immune features
derstanding and reduce decisional conflict, but are not routinely used in AML. predictive of response are lacking. Methods: We treated 11 relapsed/refractory
Methods: We developed an AML DA with input from patients, caregivers, (R/R) AML patients with azacitidine (AZA) and nivolumab (Nivo) in a phase 2
clinicians, and laypersons, via the International Patient Decision Aids Stan- clinical trial (Daver N et al Cancer Discovery 2018). Patient characteristics:
dards (IPDAS) process. It includes 10 short animated videos with voiceovers, median (med) age 65 years (47-73), 63% adverse cytogenetics, 27% TP53
covering AML basics, etiology, outcomes, treatment paradigms, and risks/ mutated. Pretreatment bone marrow aspirates had immune-phenotypic 17-
benefits of various treatment approaches. We enrolled 20 patients in a pilot color flow analysis and NanoString RNA quantification of 1469 immune-
feasibility and efficacy trial, with pre/post survey assessments of AML relevant genes. Results were correlated with clinical, pathological and mo-
knowledge via an 18-item questionnaire, decisional conflict (Decisional lecular data. Results: The median courses of AZA+Nivo administered was 3
Conflict Scale; DCS), anxiety (State Trait Anxiety Inventory, Short Form; STAI- (range 1-17). The CR/CRi rate was 45% (including 2 CR, 1 CRi, 1 CRN and 1
6), and measures of DA usability and satisfaction. Results: Participants were a CRP), with a median time to response of 1.8 months (range 0.8-4.9 months).
mean of 62.4 years old, 12 (60%) were male, 17 (85%) white, and 15 (75%) The median overall survival was 13 months with 27% patients alive at 1 year.
had newly-diagnosed disease. Mean time since AML diagnosis was 145 days We found significant positive correlations between proportions of T-effector
(median 31; range 2-1092). 16 (80%) exhibited high-school-level un- cells at baseline, and CD3+, CD8+, and T-regulatory cells at end of cycle (EOC)
derstanding of medical terms per the REALM-SF, and participants on average 1 (r . 0.75, p , 0.01 for all). At EOC2, these correlations were no longer
exhibited moderate numeracy (mean score of 4.1 on the Subjective Numeracy significant. However, there was a significant positive correlation between T-
Scale). All participants completed the study, exceeding our pre-determined effector cells at baseline and T-regulatory cells (r = 1, p , 0.001) at EOC4.
feasibility threshold. AML knowledge scores generally improved, from a mean Using NanoString analysis, we found 105 differentially expressed genes (fold
of 11.8 correct items on pre-test, to 15.2 on post-test assessment (p , change = 1.5, p , 0.05) between responders (5/11) and non-responders (6/
0.0001), with 80% of participants achieving improved scores. Struggles 11) at pretreatment. IL2-STAT5, TP53 and TNF Hallmark pathways and
remained regarding patients’ understanding of the role that genetic tests play immune response from GO gene sets were highly enriched (q , 5x10-4) in
in AML care. There was no increase in anxiety after watching the videos, but responders. We then utilized z-score distribution analysis to quantify the
decisional conflict was significantly reduced, from a mean of 28.5 at baseline degree of activation of known immunologic pathways. We found that signatures
to 22 in the post-test (p = .019). Participants reported high satisfaction and highly specific to neutrophils, NK cells, T-cells and eosinophils were signif-
usability scores for the DA. Conclusions: Our AML decision aid exhibits fa- icantly (p , 0.05) upregulated in patients with CR compared to non-
vorable performance characteristics, with high satisfaction and usability, a responders at pretreatment. Conclusions: Our data demonstrates that a sig-
marked increase in patient knowledge, and reduced decisional conflict. nature suggestive of lymphocyte activation in the pretreatment BM may be
Further testing is warranted in a randomized trial. Clinical trial information: associated with augmented clinical response to PD-1 based therapies. Similar
NCT03442452. underlying pathways that have consistently predicted for responses to PD-1
inhibition in solid cancers, primarily IL2-STAT5 genes, may have predictive
relevance in AML. Such pretreatment flow and NanoString signatures may help
select AML patients most likely to benefit from PD-1 blockade plus HMA,
further enhancing the benefit-risk ratio with such therapies.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
416s Hematologic Malignancies—Leukemia, Myelodysplastic Syndromes, and Allotransplant

7042 Poster Session (Board #417), Mon, 8:00 AM-11:00 AM 7043 Poster Session (Board #418), Mon, 8:00 AM-11:00 AM
Outcomes with CPX-351 versus 7+3 by baseline bone marrow (BM) blast First-in class selective AXL inhibitor bemcentinib (BGB324) in combination
percentage in older adults with newly diagnosed high-risk/secondary acute with LDAC or decitabine exerts anti-leukaemic activity in AML patients unfit
myeloid leukemia (sAML). First Author: Ellen K. Ritchie, Weill Cornell for intensive chemotherapy: Phase II open-label study. First Author: Sonja
Medical College of Cornell University, New York, NY Loges, Department of Oncology, Hematology, BMT with Section Pneumology
and Institute of Tumor Biology, University Medical Center Hamburg-
Background: CPX-351, a liposomal encapsulation of cytarabine (C) and daunoru-
bicin (D) at a synergistic ratio, is approved as Vyxeos in the US and EU for adults with
Eppendorf, Hamburg, Germany
newly diagnosed therapy-related AML or AML with myelodysplasia-related changes. Background: The RTK AXL represents a therapeutic target promoting AML cell
In a phase 3 study, CPX-351 significantly improved OS and remission rates vs 7+3 in proliferation and survival by pleiotropic mechanisms and is a negative regulator
patients (pts) aged 60-75 y with newly diagnosed high-risk/sAML. Some studies of anti-tumour immunity. Bemcentinib is a first-in-class, highly selective, oral
suggest a high baseline blast percentage may portend a worse prognosis in AML. This AXL inhibitor that has previously shown encouraging anti-leukaemic activity
post hoc analysis of phase 3 data assessed outcomes by baseline BM blast per- as a monotherapy in r/r AML and hr-MDS. Methods: A monotherapy dose-
centage. Methods: Pts diagnosed with AML per 2008 WHO criteria ($20% blasts in
escalation and expansion part of this trial is complete. In this second, phase II
peripheral blood or BM) were randomized 1:1 to receive #2 inductions of CPX-351
(100 units/m2 [C 100 mg/m2 + D 44 mg/m2] on Days 1, 3, 5 [2nd induction: Days 1,
part of the study, 11 and 15 AML pts unfit for intensive chemotherapy received
3]) or 7+3 (C 100 mg/m2/d continuously for 7 d [2nd induction: 5 d] + D 60 mg/m2 bemcentinib at RP2D (200 mg po/d) in combination with low-dose cytarabine
on Days 1-3 [2nd induction: Days 1-2]). Pts achieving complete remission (CR) or CR (LDAC) and decitabine, respectively. Median age was 77 yr (range: 50-83),
with incomplete platelet or neutrophil recovery (CRi) could receive #2 consolida- median screen myeloblast count 39% (3-95%) and 2/19 (11%) of pts
tions. Results: CPX-351 had longer median OS and higher remission rates vs 7+3 evaluable for FLT3 were FLT3+. Plasma protein biomarker levels were mea-
irrespective of baseline BM blast percentage; median OS was worse in higher blast sured using the DiscoveryMap v3.3 panel (Myriad RBM) at screen and fol-
groups for both treatments (Table). The incidence of grade $3 TEAEs was .80% for lowing treatment. Results: The most common TRAEs ($ 15% of pts) were ECG
both arms; febrile neutropenia was the most common. Conclusions: Improved QT prolongation (35%) and diarrhoea (15%). Among these, 3 were Grade 3,
outcomes were observed with CPX-351 vs 7+3 irrespective of baseline BM blast and none 4 or 5. All TRAEs were manageable and/or reversible. As of Feb ‘19, 9
percentage in older adults with newly diagnosed high-risk/sAML. Clinical trial in- pts (2 de novo, 1 secondary, 6 r/r) in the bemcentinib + LDAC group were
formation: NCT01696084. evaluable for response and 4 (44%; 2 de novo + 2 relapsed) achieved rapid CRi
Baseline BM blasts at C2D1. Responses were durable (range: 7 – 11 cycles) in 3 of the 4 re-
<20%a 20%-40% >40%-60% >60%
sponders. A further 2 pts (22%, 1 secondary + 1 relapsed) achieved durable
SD (5 and 6 cycles). mPFS among the 5 pts with durable CRi or SD was
CPX-351 7+3 CPX-351 7+3 CPX-351 7+3 CPX-351 7+3
n = 21 n = 22 n = 62 n = 65 n = 34 n = 34 n = 32 n = 29 5 months (range: 3.5-7.7). Further, at the time of writing, 11 pts (8 de novo, 3
Median OS, mo 12.62 7.31 11.56 5.95 8.77 4.86 4.65 2.92
r/r) in the bemcentinib + decitabine group were evaluable for response of which
HR 0.45 0.58 0.71 0.59 4 (36%, all de novo) achieved CRi after $ 4 cycles. One additional de novo pt
95% CI 0.22-0.93 0.39-0.89 0.41-1.24 0.34-1.03 achieved durable SD lasting for 5 cycles. Conclusions: Bemcentinib in
Median EFS, mo 2.96 1.13 2.50 2.23 4.99 1.03 1.84 1.18
HR 0.64 0.85 0.46 0.60 combination with LDAC exerted early durable responses in patients with both
95% CI 0.34-1.21 0.58-1.26 0.27-0.78 0.35-1.04 de novo and relapsed AML whilst the combination of bemcentinib + decitabine
CR, n(%) 7 (33) 5 (23) 25 (40) 21 (32) 16 (47) 8 (24) 9 (28) 4 (14)
OR 3.78 1.22 2.89 1.72
exerted comparably fewer and later responses in de novo AML. Soluble bio-
95% CI 0.68-21.05 0.56-2.66 0.99-8.44 0.41-7.21 marker correlations will be presented at the meeting. Both combinations were
CR+CRi, n(%) 11 (52) 8 (36) 30 (48) 29 (45) 20 (59) 9 (26) 12 (38) 4 (14) generally well-tolerated and further exploration is warranted. Clinical trial
OR 2.67 0.94 3.91 2.20
95% CI 0.62-11.50 0.46-1.93 1.37-11.15 0.55-8.82 information: NCT02488408.
a
AML diagnosis confirmed by an independent pathologist.

7044 Poster Session (Board #419), Mon, 8:00 AM-11:00 AM 7045 Poster Session (Board #420), Mon, 8:00 AM-11:00 AM
Phase II trial of CPX-351 (cytarabine:daunorubicin) liposome injection in The Transplant Optimization Program (TOP): Implementing a geriatric assess-
patients with acute myeloid leukemia .60 years of age who have not been ment (GA)-guided interdisciplinary clinic (IDC) prior to allogeneic hematopoi-
treated with intensive chemotherapy. First Author: Ellen K. Ritchie, Weill etic cell transplantation (HCT) in older adults. First Author: Benjamin Avi
Cornell Medical College of Cornell University, New York, NY Derman, University of Chicago Medical Center, Chicago, IL
Background: This study enrolled older patients without age limitation to Background: Limitations found on GA correlate with worse outcomes after HCT,
intensive chemotherapy with CPX 351 allowing for prior treatment with low but no data exists on prospectively utilizing GA prior to HCT. We established the
intensity regimens for MDS or AML. Methods: 30 patients aged $65yrs with TOP IDC in March 2013 to implement a cancer-specific GA and an IDC to risk-
diagnosed AML received up to two induction cycles and consolidation cycles stratify HCT candidacy and create an individualized care plan for allograft
of CPX-351 as a first intensive therapy. The primary efficacy endpoint was candidates 60+ yrs. Methods: The IDC consisted of a HCT physician, advanced
overall survival and the primary safety endpoint was 30-day mortality. practice provider, dietician, PT/OT, social worker, ID physician, and a geriatric
Secondary efficacy endpoints included response rate and duration, and oncologist to devise a pt specific optimization strategy. We compared con-
assessment of the relationship between cognitive function and treatment secutive HCT pts $ 60 years undergoing GA prior to TOP implementation (pre-
outcome. Although the primary efficacy endpoint data is pending, these TOP) from 2005-2012 (n=75) to TOP pts from 2013-2018 (n=86). Results: 3/
interim results describe promising primary safety and efficacy endpoint 89 HCT pts 60+ yrs who did not attend TOP were excluded; all 3 died before 1
results. Of the 30 patients enrolled, 10 patients (33.3%) were $75yrs old, year post-HCT. Compared with controls, the TOP group was older (median age
13 patients (43.3%) had had previous hematologic malignancies, 17 pa- 67 vs. 64 yrs, p,0.001) but was similar in HCT-CI $ 3 (37% vs. 48%, p=0.2),
tients (56.6%) had adverse ELN risk stratification, and 16 patients (53.3%) use of myeloablative regimens (20% vs. 19%, p=0.8), and advanced ASBMT
had failed to respond to previous non-intensive treatments. In addition, all risk disease (46% for both). Relative to the pre-TOP group, TOP pts at baseline
patients had an average of 2.1 co-morbid medical conditions. Results: 14 had fewer impairments in independent activities of daily living (30% vs. 48%,
patients (47%) achieved best response criteria, with 12 patients (40.0%) p=0.02) and fewer frail 4-meter walk tests (7% vs. 31%, p,0.001). Pts
achieving complete remission (CR) and 2 patients (6.6%) achieving com- undergoing optimization in TOP fared better versus pre-TOP (Table). 1-yr non-
plete remission with incomplete platelet or neutrophil recovery (CRi). 6 relapse mortality (NRM) and 1-yr overall survival (OS) continued to improve
patients (20%) had a morphologic leukemia free state with pancytopenia including 11% NRM and 89% OS in 2017. Conclusions: A GA-guided in-
and 10 patients had persistent disease. 8 patients (26.6%) went to stem cell terdisciplinary optimization clinic for allograft recipients age 60+ reduced
transplant. 2 patients (6.6%) died within 30-days. Although final analysis on transplant associated morbidity and mortality, with marked improvements in
adverse events data is pending, 6 patients (20%) experienced a Grade 2 NRM and OS over time. A GA-based IDC can facilitate selection and optimi-
decrease in left ventricular ejection fraction (LVEF) at the end of first in- zation of older pts considering HCT.
duction. Conclusions: These results suggest that in elderly AML patients Outcomes of HCT recipients age $ 60: pre-TOP vs TOP.
regardless of age and prior treatment with non-intensive chemotherapy can Pre-TOP Early TOP Mature TOP
be treated safely with CPX 351 with low 30 day mortality. CR in these 2005-2012 3/2013-2014 2015-2018
n=75 n=28 n=58
patients is 40% with 26% going forward to stem cell transplant. Complete
Inpatient Deaths During Initial Hospitalization 13 (17%) 0 (0%)* 0 (0%)**
data will be assessed for OS and rate of MRD positivity. The relationship of Length of inpatient stay after HCT, median days (IQR) 19 (15-28) 14.5 (13-19)** 14 (13-17)**
cognitive and social factors to outcome will also be addressed. Events by Day 100
Discharge to Nursing Facility 11/62 (18%) 1 (4%) 1 (1.7%)**
Readmission 37/62 (60%) 19 (68%) 34 (59%)
Death 16 (21%) 3 (11%) 3 (5%)*
1-year NRM¶ 43% 36% 18%**
1-year OS¶ 44% 46% 70%**

*p,0.05/**p,0.01 relative to pre-TOP by Fisher’s exact or Mann Whitney U tests. ¶Excludes 2018 pts.

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Hematologic Malignancies—Leukemia, Myelodysplastic Syndromes, and Allotransplant 417s

7046 Poster Session (Board #421), Mon, 8:00 AM-11:00 AM 7047 Poster Session (Board #422), Mon, 8:00 AM-11:00 AM
Late infectious complications in hematopoietic cell transplantation Allogeneic stem cell transplantation (AlloSCT) for patients (pts) with acute
survivors. First Author: Eric Jessen Chow, Fred Hutchinson Cancer Re- leukemia following venetoclax-based therapy. First Author: Akash Mukherjee,
search Center, Seattle, WA The University of Texas MD Anderson Cancer Center, Department of Stem Cell
Transplantation and Cellular Therapy, Houston, TX
Background: Infections are a major complication of hematopoietic cell
transplantation (HCT). Few studies have compared the incidence of late Background: BCL-2 inhibitor Ven has shown a promising benefit in pts with acute
infections occurring $2y post-HCT to other cancer patients and the general myeloid leukemia (AML) and acute lymphoblastic leukemia (ALL). There is paucity of
population. Methods: Single center records of $2y HCT survivors who were information about the safety and efficacy of alloSCT post Ven. Methods: We
Washington residents treated from 1992-2009 (n = 1,792; median age 46y; conducted a retrospective analysis of 35 AML/ALL pts who received alloSCT fol-
53% allogeneic; 90% hematologic malignancies) were linked to the state’s lowing Ven-based therapies between 2013-2018 at MDA. Results: Median age at
hospital discharge and death registries. Individuals randomly selected from alloSCT was 60 years and 15 (43%) pts had an age-adjusted HCT-CI score $ 4.
the state cancer registry (n = 5,455, non-HCT) and driver’s license files (n = Disease diagnosis – AML (n = 31; 89%), ALL (n = 4; 11%). Disease status at
transplant was CR1 (n = 17; 49%), CR2/CR3 (n = 9; 26%) or refractory (n = 9; 26%).
16,340, DOL) who survived $2y formed two comparison groups, matched
20/26 (77%) CR pts were MRD-negative. Median # of prior therapies was 2 (range 1-
on sex, age, year, and cancer diagnosis (non-HCT group only). Based on 7) and 4 (11%) pts had failed a prior alloSCT. AML pts were classified by ELN 2017
hospital and death registry codes, incidence rate ratios (IRR) with confi- criteria to have favorable, intermediate and adverse risks in 16%, 23% and 61%
dence intervals (CI) of infections by organism type and organ system were respectively. Ven was provided in combination of hypomethylating agents (HMA) or
estimated using Poisson regression. Results: With 6y (range 2-20) median other chemotherapies in 26 (74%) and 9 (26%) pts, respectively. Among pts treated
follow up, the incidence rate (per 1000 person-y) of all infections was 65 in with Ven + HMA, some also received IDH1/2 inhibitors (n = 7, 20%), FLT3 inhibitors
HCT survivors vs. 40 in the non-HCT group (IRR 1.6, 95% CI 1.3-1.9). In (n = 4; 11%) or anti-PD1 (n = 3, 9%). Median duration of Ven-based treatment was
contrast, the DOL group’s infection rate was 7 (HCT vs. DOL IRR 10.0, 95% 2 months (range 0.5- 4.6). Ven was discontinued in 6 (17%) pts due to adverse
CI 8.3-12.1). Specifically, bacterial and fungal infections were each 70% events (n = 4) or progression (n = 2); the remaining pts (83%) continued their Ven-
more common in HCT vs non-HCT cancer survivors (IRRs 1.7; p , 0.01). therapy as a bridge to alloSCT. The median time from last Ven dose and transplant
Differences in viral infection rates were more modest (IRR 1.4, p = 0.07). was 26 days. Conditioning regimens were melphalan-based reduced intensity (n =
Infections attributed to staphylococcus, streptococcus, and non-Candida 26, 74%), or busulfan-based myeloablative regimens (n = 9; 26%). Donor source
fungi including Aspergillus were twice as common in the HCT vs. non-HCT was matched -unrelated (n = 14, 40%), -related (n = 9; 26%) or haplo- (n = 12;
cancer survivors (IRRs 2.1-2.3; p , 0.05). IRRs for nervous system, re- 34%). GVHD prophylaxis consisted of tacrolimus with either PT-Cy in 25 (71%) pts
spiratory, and musculoskeletal infections between these 2 groups were 1.9- or methotrexate in 10 (29%) pts. All pts engrafted (median day 30 donor cells =
2.8 (p , 0.05). Among potentially vaccine-preventable organisms, the IRR 100%). Median days to ANC . 500 and platelets . 20K was 15.5 and 22.5,
respectively. With a median follow up of 5.7 months (range 0.7-15.4), the 1-year
was 3.2 (95% CI 2.2-4.6). While the absolute incidences decreased with
rates of OS, PFS, and NRM were 71%, 63% and 3% respectively. CI of acute grade
time, the relative risks in almost all categories were even greater when re- 2-4 and 3-4 GVHD were 26% and 3% respectively. Four pts died: 3 because of
stricted to $5y HCT vs. non-HCT cancer survivors. Conclusions: $2y HCT disease relapse and 1 of infection. Conclusions: AlloSCT is a safe and feasible
survivors had a significantly increased incidence of infections vs. matched consolidation treatment option in acute leukemia pts who were pre- treated with Ven,
non-HCT cancer survivors. Providers caring for long-term HCT survivors without excessive risk of NRM or acute GVHD. Larger prospective studies are re-
should maintain high vigilance for infections in this population and ensure quired to validate our observations.
adherence to HCT antimicrobial prophylaxis and vaccination guidelines.

7048 Poster Session (Board #423), Mon, 8:00 AM-11:00 AM 7049 Poster Session (Board #424), Mon, 8:00 AM-11:00 AM
Rapid reduction of peripheral blasts in older patients with refractory acute Role of killer cell immunoglobulin-like receptor (KIR)-ligand interactions to
myeloid leukemia (AML) using reinduction with single agent anti-CD45 prevent relapse in patients (pts) receiving matched unrelated stem cell
targeted iodine (131I) apamistamab [Iomab-B] radioimmunotherapy in the transplant (SCT) for acute myeloid leukemia (AML). First Author: Hind
phase III SIERRA trial. First Author: Ben Kent Tomlinson, Adult Hematologic Rafei, Division of Cancer Medicine, The University of Texas MD Anderson
Malignancies and Stem Cell Transplant Section, Seidman Cancer Center, Cancer Center, Houston, TX
University Hospitals Case Medical Center, Cleveland, OH Background: NK cells play a pivotal role in cancer immune vigilance. NK cell function is regulated
Background: The SIERRA trial is a prospective, randomized, phase 3, open-label, by a balance between activating and inhibitory signals derived from various receptor-ligand in-
ongoing multicenter trial for patients aged $55 years with active, relapsed/refractory teractions. Methods: 390 AML pts received 10/10 unrelated donor SCT. KIR genotype was
performed by PCR with sequence specific primers. Hazard ratio (HR) for 2-year relapse was
(R/R) AML evaluating allogeneic hematopoietic cell transplantation (HCT) versus
calculated using Fine-Gray regression and adjusted for disease risk index, remission status, pre-SCT
conventional care (CC). Recent preliminary data demonstrated robust donor en- MRD, conditioning regimen and presence of HLA-DP mismatch. KIR-ligand (K-L) match was
graftment in all patients treated with Iomab-B (Blood 2018 132:1017) despite active defined as the presence a given KIR in the donor and the presence of its reported ligand in the
disease. We hypothesize that successful engraftment was due to rapid disease re- patient (ex. 2DL1 and HLA-C2). KIRs that have no known ligands were not included in this analysis.
duction with Iomab-B. Methods: Patients are randomized to receive Iomab-B and HCT The Table shows pt characteristics and K-L matches. Results: There was no correlation between the
or to a CC therapy including approved targeted agents followed by HCT if in remission. number of inhibitory or activating KIRs or KIR haplotype (A or B) and the probability of relapse after
Majority of patients (79%) in the CC arm did not achieve CR and the study allowed SCT. However, donor KIRs had a dramatic effect on relapse when they were considered together
crossover to receive Iomab-B. Results: Data were evaluated for the first 25% of with the presence of the corresponding ligand in the pt. The 210 pts who had $3 inhibitory K-L
patients (N=38). Twenty-nine patients received Iomab-B, either directly (N=19) or via matches had a significantly higher probability of relapse (HR=1.748, CI=1.147-2.667, p=0.009)
crossover (N=10). Median baseline marrow blasts were 30% (4-74) for Iomab-B and than the remaining 180 pts. Similarly, the 96 pts who had at least one known activating K-L match
24% (6-70%) for CC, which increased to 45% (10-70%) at crossover. Peripheral had a lower probability of relapse (HR=0.581, CI=0.345-0.978, p=0.04). When we considered
blast data was available in 16 patients (Iomab 7, Crossover 9). By day 3 post-Iomab, inhibitory and activating K-L matches together, we found that the 168 pts who had $3 inhibitory
and no activating K-L matches had a significantly higher probability of relapse (HR 2.001,
blasts were reduced by 98% with 100% reduction by day 8 (assuming 0% blasts due
CI=1.376-2.908, p,0.001) than the 222 remaining pts. Conclusions: Donor-pt KIR-ligand
to lack of differential at WBC 0.1; Table). All patients engrafted with ANC at a median matching should be taken into account when choosing unrelated donors for AML pts.
of 13 days (9-22 days). Conclusions: Targeted re-induction radioimmunotherapy with
single agent Iomab-B rapidly decreases peripheral blasts in chemotherapy refractory Characteristic No. of pts (N= 390)

AML. This significant reduction of leukemia burden followed by successful en- Median age, years (range) 54.3 (18.1-79.8)
Male sex 184
graftment after HCT is encouraging in this underserved population. Clinical trial Conditioning Regimen
Fludarabine-Melphalan 63
information: NCT02665065. Fludarabine-Busulfan (RIC) 59
Fludarabine-Busulfan (MAC) 268
WBC x 103 Abs. Blasts x 103 Remission status
(N=28) % Reduction (N=16) % Reduction CR 222
No CR 138
Disease specific risk index
Pre-Therapeutic In- 0.85 - 0.37 - Low 22
fusion (TI) Intermediate 125
(0.1 - 36.8) (0.02 – 18.77) High 177
Day 1 Post TI 0.4 60% 0.09 (N=11) 80% Very high 66
(0.1 – 20.4) (p , 0.0001) (0 – 16.32) (p , 0.0001) Number of inhibitory K-L matches, Median (range) 3 (1-5)
Number of activating K-L matches, Median (range) 0 (0-4)
Day 3 Post TI 0.3 (N=26) 80% 0.03 (N=8) 98% K-L matches
(0.1 – 2.7) (p , 0.0001) (0 – 0.42) (p , 0.0001) 2DL1-HLA C2 226
Day 8 Post TI 0.1 88% 0.0* 100% blast reduction 2DL2-HLA C1, B46:01, B73:01 150
assuming 0% blasts* 2DL3-HLA C1, B46:01, B73:01 300
3DL1-HLA Bw4 212
(Pre-Fludarabine (0.1 – 0.3) (p , 0.0001) 3DL2-HLA A3, A11 147
Conditioning) 2DS1-HLA C02:02, C04:01, C 05:01, C06:02, C17:0, C18:02 39
2DS2-HLA A11:01 21
*Differential not done due to low WBC 2DS4-HLA C01:02, C02:02, C05:01, C14:02, C16:01, A11:01, A11:02 44
3DS1-HLA B27:05, Bw4T80 60

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
418s Hematologic Malignancies—Leukemia, Myelodysplastic Syndromes, and Allotransplant

7050 Poster Session (Board #425), Mon, 8:00 AM-11:00 AM 7051 Poster Session (Board #426), Mon, 8:00 AM-11:00 AM
Mast cells as mediators of fibrosis and effector cell recruitment in dermal Cardiac, vascular, and hypertension safety of bosutinib versus imatinib for
chronic graft-versus-host disease. First Author: Ethan Strattan, University of newly diagnosed chronic myeloid leukemia in the BFORE trial. First Author:
Kentucky, Lexington, KY Jorge E. Cortes, The University of Texas MD Anderson Cancer Center, Houston,
TX
Background: Allogeneic hematopoietic stem cell transplantation (allo-HCT)
is a potentially curative treatment for patients with malignant neoplasms or Background: Tyrosine kinase inhibitor therapy has been linked to cardiac and
inborn defects of hematopoiesis. Benefits of allo-HSCT are hampered by vascular events. Cardiac, vascular and hypertension treatment-emergent ad-
graft-versus-host-disease (GVHD), which can be debilitating and potentially verse events (TEAEs) with bosutinib or imatinib for newly diagnosed chronic
lethal. In chronic GVHD (cGVHD), inflammation and aberrant wound healing phase chronic myeloid leukemia were analyzed. Methods: Patients (pts) who
lead to pathological fibrosis across multiple organs, most frequently in the received $1 dose of bosutinib (n = 268) or imatinib (n = 265) 400 mg/d in the
skin, yet the exact pathophysiology is not well-understood. Mast cells (MCs) phase 3 BFORE trial were included. Prespecified MedDRA terms comprised
are primarily known for their role in atopic disease. However, recent studies the clusters of investigator assessed TEAEs. Exposure-adjusted TEAE rate was
have demonstrated new roles for MCs, showing that they can be involved in defined as the number of pts with TEAEs / total pt-yr (pt-yr = sum of total time to
wound healing and in the pathogenesis of fibrotic disease. Given these new first TEAE for pts with TEAEs and treatment duration for pts without TEAEs).
paradigms and the MC tropism to skin, alongside their reported role in other Results: After $36 mo follow-up, 65% vs 62% of pts in the bosutinib vs
fibrotic diseases, we investigated whether MCs may play a role in the path- imatinib arm were still on treatment. Rates of TEAEs, treatment withdrawals
ogenesis of dermal cGVHD. Methods: Cells: MCs were grown ex vivo from and drug-related TEAEs in the clusters of interest were low in both arms
murine bone marrow. Transplant: 8 Gy radiation, followed by injection of LP/J (Table). The most common cardiac, vascular and hypertension TEAEs, re-
marrow and splenocytes into C57BL/6J (WT) or B6.Cg-KitW-sh MC-deficient spectively, were sinus bradycardia (2%), angina pectoris (3%) and hyper-
recipients. Results: Ex vivo, we show that MCs survive and are functional after tension (7%) vs prolonged QT (3%), peripheral coldness (1%) and
lethal radiation, such as that used in conditioning prior to HCT. In a murine hypertension (9%) with bosutinib vs imatinib; corresponding grade 3/4/5 TEAE
model of cGVHD WT mice had significantly more cGVHD symptoms than MC- rates in the respective clusters were 3%, 3% and 4% vs 1%, 0.4% and 4%.
deficient mice as measured by clinical scoring. This scoring correlated with a Hypertension was the only grade 3/4 TEAE occurring in $1% of pts in either
significant increase in skin pathology, collagen deposition, and expression of arm (4% each); 1 grade 5 TEAE each was noted for bosutinib (cardiac failure)
pro-fibrotic genes in WT as compared to B6.Cg-KitW-sh mice. Dermal MC and imatinib (cerebrovascular accident). Exposure-adjusted rates of cardiac,
numbers were increased in WT mice, but were nearly undetectable in B6.Cg- vascular and hypertension TEAEs, respectively, were 0.04, 0.03 and 0.04 vs
KitW-sh mice, implying that the MCs that are present were recipient-derived and 0.03, 0.01 and 0.04 (grade 3/4/5 only: 0.01, 0.01 and 0.02 vs 0.01, 0.002
had survived conditioning. Skin from WT but not B6.Cg-KitW-sh mice was and 0.02) for bosutinib vs imatinib. Conclusions: Cardiac, vascular and hy-
enriched in cGVHD effector cells and in inflammatory cytokines and che- pertension TEAE rates were low with bosutinib and imatinib. A majority of
mokines. Murine MCs, upon stimulation were sources of many of these factors, TEAEs were low grade and few led to treatment withdrawal. Clinical trial in-
production of which was blocked when treated with ibrutinib and ruxolitinib, formation: NCT02130557.
drugs used in cGVHD treatment. Conclusions: In summary, we show here a n (%) Bosutinib (n = 268) Imatinib (n = 265)
previously unknown role for MCs in the pathogenesis of dermal cGVHD,
Any cardiac TEAE 23 (9) 17 (6)
suggesting that MCs may be targetable to prevent and treat this devastating Led to withdrawal 1 ( , 1) 0
complication of allo HCT. Drug-related 7 (3) 6 (2)
Any vascular TEAE 18 (7) 8 (3)
Led to withdrawal 3 (1) 1 ( , 1)
Drug-related 5 (2) 0
Any hypertension TEAE 22 (8) 25 (9)
Led to withdrawal 0 0
Drug-related 2 (1) 5 (2)

7052 Poster Session (Board #427), Mon, 8:00 AM-11:00 AM 7053 Poster Session (Board #428), Mon, 8:00 AM-11:00 AM
Update on lower-dose dasatinib 50 mg daily as frontline therapy in newly Discrepancy of blast percentage between the bone marrow aspirate and
diagnosed chronic phase chronic myeloid leukemia (CML-CP). First Author: biopsy and its impact on survival outcomes in patients with myelodysplastic
Kiran Naqvi, The University of Texas MD Anderson Cancer Center, Department syndromes excess blast (MDS-EB). First Author: Meera Yogarajah, Mayo
of Leukemia, Houston, TX Clinic, Rochester, MN
Background: Dasatinib, a potent BCR-ABL tyrosine kinase inhibitor (TKI), is Background: The revised International Prognostic Scoring System (IPSS-R) aids in
approved for the treatment of chronic phase CML (CML-CP) in the frontline and prognosticating MDS. The percentage (%) of blasts in the bone marrow is one of
salvage settings. Notable side effects include pleural effusions and myelo- the major determinants of the scoring system. The aspirate blast % is utilized as the
suppression. We previously reported dasatinib 50 mg daily to be active and standard of care, but there could be discrepancies in the blast % reported by the
better tolerated than the approved 100 mg daily dose (CANCER. 2018 Jul 1; aspirate and the biopsy. We aim to study the possible use of bone marrow biopsy
124(13):2740-2747). We present an update on the efficacy and toxicity blasts in MDS-EB in calculating IPSS-R. Methods: The MDS database was
profile of lower dose dasatinib 50 mg orally daily in patients with early CML-CP. reviewed for cases of MDS-EB after due IRB approval. We calculated IPSS-R scores
Methods: All patients presenting to our institution in early CML-CP were eli- based on the aspirate blast % (IPSS-RAsp) and biopsy blast % (IPSS-RBx). The
biopsy blast % was reported morphologically or by the CD34 stain. Whenever a
gible to participate. Prior TKI therapy for up to 1 month was allowed. Re-
range was reported the highest value was utilized as the blast %. Suboptimal
sponses were assessed according to the European LeukemiaNet guidelines
aspirates were excluded from the study. The overall survival (OS) was determined by
(Baccarani et al. Blood 2013 122.872:884). Results: From March 2016 to IPSS-RAsp, IPSS-RBx and IPSS-R highest blast (IPSS-RHi). OS estimates were
March 2018, 81 patients have been enrolled. Median age is 47 years (20-84). calculated by Kaplan-Meier curves and log-rank testing using JMP v.13. Uno’s
Patients categorized by Sokal risk are: low 53; intermediate 22 and high 6. concordance statistic was used to compare all 3 risk scoring systems. Results: Of
Median follow-up is 18 months (9-31). Cumulative response rates over time 1322 patients, 431 (33%) cases were identified with MDS-EB; out of which 173
are shown below: At 3 months, 96% patients achieved early molecular re- cases had both blasts reported in the biopsy and the aspirate. Out of 173 cases, 35
sponse (BCR-ABL PCR #10%). Median time to CCyR was 4.6 months, MMR (20%) had MDS-EB1, and 61 (35%) had MDS EB-2 based on both biopsy and
6.0 months, MR4.0 11.4 months and MR4.5 12.2 months. Eighteen patients aspirate (concordant cases). Seventy seven (45%) patients changed from EB-1 to
had treatment interruption: pleural effusion 4 (possibly related 3, unrelated 1 EB2 or vice versa based on the biopsy blast (44/77 (57%) cases were upstaged).
due to pneumonia); gastrointestinal bleed 2; thrombocytopenia 3; trans- The OS outcomes based on the IPSS-RBx biopsy showed a clear and meaning-
aminitis 2; renal dysfunction 1; asthma exacerbation 1; pneumonitis 1; lower ful separation with median OS decreasing with increased risk but IPSS-RAsp and
extremity edema 1; myalgias 1, and pregnancy 2. Four patients had dose IPSS-RHi did not (Table). We compared the 3 models for observed OS differences
reductions: pleural effusion 3; myalgias 1. Four patients had dasatinib dose using the Uno model and there was no statistically significant difference.
increased to 100mg: lack of CCyR at 6 months, 3; lack of MMR at 12 months, Conclusions: IPSS-RBx (but not IPSS-RAsp and IPSS-RHi) identified prognostic
1. Four patients are off study: no response 2, pneumonitis 1, and insurance 1. groups for OS with median OS decreasing with increased risk. The small sample size
None of the patients have transformed or died. Conclusions: These updated may have led to an insignificant effect on model power by Uno model. This finding
results continue to support dasatinib 50 mg daily as an effective and safe dose needs to be validated by other centers.
for early CML-CP. Clinical trial information: NCT02689440. IPSS-R
Risk groups Low Intermediate High Very high P value Uno concordance
3 months 6 months 12 months 18 months
IPSS-R Asp 23.8 41.5 21.9 9.2 0.0011 Asp-Bx
CCyR 28% 78% 96% 96%
Median OS, months (P=0.6)
MMR 19% 52% 79% 85%
IPSS-R Bx 47.5 39.3 21.9 8.3 0.0001 Bx-Hi
MR4.0 4% 21% 55% 69%
Median OS, months (P=0.8)
MR4.5 2% 15% 49% 58% Hi
R-IPSS 31.6 41.5 25.5 9.2 0.0001 Hi- Asp
Median OS, months (P= 0.4)

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
 Hematologic Malignancies—Leukemia, Myelodysplastic Syndromes, and Allotransplant 419s

7054 Poster Session (Board #429), Mon, 8:00 AM-11:00 AM 7055 Poster Session (Board #430), Mon, 8:00 AM-11:00 AM
Patterns of leukemic transformation in patients with TP53-mutant myelodys- Proposal of an endpoint for a phase III clinical study of essential thrombo-
plastic syndromes. First Author: Kelly Sharon Chien, The University of Texas cythemia: Balancing between short term effects and long term benefits. First
MD Anderson Cancer Center, Division of Cancer Medicine, Houston, TX Author: Ruben A. Mesa, UT Health San Antonio Cancer Center, San Antonio,
TX
Background: TP53 is the most frequently mutated gene in human cancers,
including 7-13% of acute myeloid leukemia (AML) and myelodysplastic Background: Essential thrombocythemia (ET) is a chronic myeloproliferative
syndrome (MDS) patients. Although it is associated with transformation to AML neoplasm (MPN), covering broad spectrum of clinical scenarios, from
in patients with MDS, the additional genomic events leading to transformation asymptomatic patients with only isolated high platelets to highly morbid pa-
are poorly understood. Methods: We retrospectively evaluated 312 patients tients in late stage disease. Due to heterogenous patient population, designing
with TP53-mutated AML or MDS diagnosed between 2013-2016. Patient clinical studies in ET is difficult, and after anagrelide there was no new drug
characteristics and bone marrow data, including cytogenetic and next gen- approval to treat ET during the past 15 years, thus, high unmet medical to treat
eration sequencing information, were assessed at the time of diagnosis and patients with ET remains. Interferons alpha (IFNa) are known to have bene-
progression to AML. Results: There were 151 TP53-mutated MDS patients ficial effects in MPN (Kiladjian et al, 2016). P1101 is a next generation
and 161 TP53-mutated de novo AML patients with a median follow-up time of monopegylated IFNa, developed specifically to treat MPNs, including ET.
34.1 months. Forty-one patients with TP53-mutated MDS transformed to Methods: External published clinical data in ET were analyzed to design
AML. Sequencing data at transformation was available in 17 patients (41%). optimal clinical study. Proposed endpoints are meant to cover all relevant
At diagnosis, median age was 67 with 2 patients with intermediate-risk, 7 clinical aspects of ET, and suffice for a regulatory relevant pivotal clinical
patients with high-risk, and 32 patients with very high-risk MDS by IPSS-R. study. Results: Composite primary endpoint scale is based on modified
Complex karyotype was seen in 40 patients, and 12 patients had 1, 25 patients ELN criteria. Short term study endpoints should have clinical meaningfulness
had 2, and 4 patients had 3 TP53 abnormalities. Predictors of transformation at time of measurement but also predict the later outcomes. The scale consists
to AML include TP53 loss of heterozygosity (p = 0.008), 3 TP53 abnormalities of: normalization of platelets (,400 G/L) and leukocytes (,9.5 G/L); nor-
(p = 0.049), complex cytogenetics (p = 0.023), and female gender (p = malization in size or non-progression of palpable spleen; lack of major car-
0.002). Median time to transformation was 10.4 months. At transformation, diovascular event during the observation period; and improvement of
an increase in TP53 variant allelic frequency was observed in 7 patients (41%), Myeloproliferative Neoplasm Symptom Assessment Form total symptom score
and new mutations, particularly NRAS, IDH1, TP53, MLL, KDM6A, TET2, and (MPN-SAF TSS). MPN-SAF TSS is a 10 items questionnaire, allowing concise,
NOTCH1, were acquired by 10 patients (59%). Cytogenetic evaluation valid, and accurate assessment of MPN symptom burden over time (Emanuel
revealed identical clones in 5 patients, linear acquisition of cytogenetic ab- et al, 2012). For TSS-10 score, to qualify for response, following rules were
normalities in 12 patients, and new clones in 8 patients. Patients with TP53- elaborated: 10 points or higher reduction for patients with baseline score
mutated AML from MDS also had worse median overall survival than patients of $20, for patients with baseline score 15-19 – 5 points reduction, baseline
with TP53-mutated de novo AML at 2.5 months vs. 5.7 months respectively scores 10-14 – reduction below 10 points, and for TSS baseline score ,10 –
(p , 0.001). Conclusions: TP53 mutations confer a worse prognosis, espe- stay ,10. Bone marrow analysis is not to be a mandatory test for assessment of
cially in the setting of AML transformed from MDS. This may be due to the overall benefit of therapy. Extensive genetic workup is planned to document
acquisition of new mutations and cytogenetic abnormalities. Further explo- any change in observed genetic and chromosomal abnormalities, including
ration of the biological mechanisms leading to transformation in TP53-mutant level of circulating mutant CALR. This would allow for objective evaluation of
MDS is warranted. IFNa’s ability to modify the disease. Long term observation, going beyond the
12 month of initial observation, is planned. Conclusions: Authors conclude
that proposed endpoint scale covers well all clinically relevant aspects of ET, in
order to make clinically relevant conclusions on durable benefits and risks of
P1101 therapy.

7056 Poster Session (Board #431), Mon, 8:00 AM-11:00 AM 7057 Poster Session (Board #432), Mon, 8:00 AM-11:00 AM
A phase II study of cpi-0610, a bromodomain and extraterminal protein Fedratinib (FEDR) in myelofibrosis (MF) patients previously treated with
inhibitor (BETi) alone or with ruxolitinib (RUX), in patients with myelofibrosis ruxolitinib (RUX): A reanalysis of the JAKARTA-2 study. First Author: Claire N.
(MF). First Author: Marina Kremyanskaya, Mount Sinai School of Medicine, Harrison, Guy’s and St Thomas’ NHS Foundation Trust, London, United
New York, NY Kingdom
Background: BETi have been shown to regulate NF-kB, MYC, BCL2, and TGF- Background: MF is a life-threatening MPN for which RUX is the only approved
b signaling, important drivers of marrow fibrosis. Preclinical studies have treatment (Tx) option. Patients (pts) who are relapsed/refractory (R/R) or in-
suggested that combined BETi and JAK2 inhibition synergistically reduce MF- tolerant to RUX have a particularly high unmet medical need. FEDR is an oral
related splenomegaly, bone marrow fibrosis and the malignant allele burden selective JAK2 inhibitor active against wt and mut JAK2. The JAKARTA-2
(Kleppe, 2018). CPI-0610 is a selective and potent oral BETi, being evaluated study demonstrated $35% spleen volume responses (SVR) in pts resistant or
in the first study of a BETi in MF. Methods: Phase 2 trial with 3 arms: CPI-0610 intolerant to RUX per investigator assessment. This JAKARTA-2 reanalysis
monotherapy (Arm 1) or RUX + CPI-0610 “add-on” (Arm 2) in pts who have employs a more stringent definition of RUX failure than used in the previous
progressed/ had an inadequate response to RUX, or CPI-0610 + RUX in JAK analysis. Methods: Adult pts previously treated with RUX with intermediate or
inhibitor-naı̈ve pts with anemia (Arm 3). Arms 1 and 2 are stratified: trans- high-risk primary, post-PV, or post-ET MF, palpable splenomegaly, ECOG
fusion dependence (TD) yes: A/no: B. The primary objectives are to evaluate the PS #2, and platelet counts $50 3 109/L received FEDR 400 mg QD in
effect of CPI-0610 on transfusion dependence (TD, 1A and 2A) and spleen continuous 28-day cycles. The primary endpoint was spleen volume response
volume (1B, 2B and 3). A Simon two-stage design: if 2 responses are seen will rate (SVRR): $35% SVR from baseline (BL) at cycle 6 end per blinded central
advance to the 2nd stage. Results: 4 pts enrolled in Arm 1, 14 pts in Arm 2, no review of MRI/CT scans. A key secondary endpoint was symptom RR ($50%
pts accrued to Arm 3 yet. Median age: 69 years (46-83), gender: 9 male pts; decrease in total symptom score from BL). Results: 79/97 enrolled pts (81%)
11 pts received $1 prior therapy besides RUX. JAK2/MPL/CALR mutations: met the more stringent criteria for RUX R/R (n=65, 82%) or intolerance (n=14,
17/18 pts, $3 mutations: 10 pts, ASXL1 mutations: 11 (61%) pts. 18%). Median BL spleen volume was 2946 mL (~143 normal). Median prior
Hemoglobin ,10 g/dL at baseline: 11 (61%) pts. 6 pts received $ 24 weeks of RUX Tx duration was 11.5 mo (range 1.0–62.4). Median number of FEDR Tx
CPI-0610 treatment at this analysis. 2 TD pts in Arm 2 became transfusion cycles was 7 (1–20). SVRR with FEDR was 30% (95% CI 21, 42). KM es-
independent, both remain on treatment free of transfusions. Hgb increase timated median spleen response duration was not estimable (95% CI 7.2 mo,
of $1.5 g/dL from baseline was observed with successive cycles of therapy in NE). Symptom RR was 27%. Safety was consistent with prior reports.
anemic pts: 2/2 pts in Arm 1 (100%) and 3/9 pts in Arm 2, (33%). Spleen Conclusions: FEDR provided clinically meaningful reductions in splenomegaly
volume reduction, by MRI (6-44%) was observed in all 10 evaluable pts and symptom burden in pts with MF who met more stringent criteria for R/R or
irrespective of their driver mutation. Symptom improvement and reductions of intolerance to RUX. Clinical trial information: NCT01523171.
cytokine levels were observed. In Arm 1: 2/2 evaluable pts had marrow fibrosis RUX failure.
improvement with Hgb increase; additionally, thrombocytosis resolved in 2/2 Prior Analysis Current Analysis
pts (baseline $791 103/uL). Most common adverse events were mild diarrhea, Resistance RUX Tx $ 14 d with no response or stable Relapsed: RUX Tx $ 3 mo with regrowth,
nausea/vomiting; and reversible and non-cumulative thrombocytopenia. disease per investigator, disease progression, defined as , 10% SVR or , 30% decrease in
Conclusions: CPI-0610 is a well-tolerated, and an effective therapeutic agent or loss of response spleen size from BL, following an initial re-
sponse
for the treatment of MF. Collectively, these data indicate that CPI-0610 +/- Refractory: RUX Tx $ 3 mo with , 10% SVR
RUX might have disease modifying effects. Updated data will be provided. or , 30% decrease in spleen size from BL
Intolerance RUX Tx $ 14 d before discontinuing Tx due to RUX Tx $ 28 d complicated by development of
Clinical trial information: NCT02158858. unacceptable toxicity RBC transfusion requirement ($ 2 units/mo
for 2 mo); or grade $ 3 thrombocytopenia,
anemia, hematoma/hemorrhage while on RUX

Visit abstracts.asco.org and search by abstract for the full list of abstract authors and their disclosure information.
420s Hematologic Malignancies—Leukemia, Myelodysplastic Syndromes, and Allotransplant

7058 Poster Session (Board #433), Mon, 8:00 AM-11:00 AM 7059 Poster Session (Board #434), Mon, 8:00 AM-11:00 AM
Results from ongoing phase 1/2 clinical trial of tagraxofusp (SL-401) in Results from ongoing phase 1/2 clinical trial of tagraxofusp (SL-401) in
patients with intermediate or high risk relapsed/refractory myelofibrosis. patients with relapsed/refractory chronic myelomonocytic leukemia (CMML).
First Author: Naveen Pemmaraju, Department of Leukemia, The University First Author: Mrinal Mahesh Patnaik, Mayo Clinic, Rochester, MN
of Texas MD Anders