VOLUME 26 䡠 NUMBER 18 䡠 JUNE 20 2008

JOURNAL OF CLINICAL ONCOLOGY R E V I E W A R T I C L E

One Hundred Years After “Carcinoid”: Epidemiology of and

Prognostic Factors for Neuroendocrine Tumors in 35,825
Cases in the United States
James C. Yao, Manal Hassan, Alexandria Phan, Cecile Dagohoy, Colleen Leary, Jeannette E. Mares,
Eddie K. Abdalla, Jason B. Fleming, Jean-Nicolas Vauthey, Asif Rashid, and Douglas B. Evans
From the Departments of Gastrointesti-
nal Medical Oncology, Surgical Oncol- A B S T R A C T
ogy, and Pathology, The University of
Texas M.D. Anderson Cancer Center, Purpose
Houston, TX. Neuroendocrine tumors (NETs) are considered rare tumors and can produce a variety of
Submitted November 30, 2007;
hormones. In this study, we examined the epidemiology of and prognostic factors for NETs,
accepted March 13, 2008. because a thorough examination of neither had previously been performed.
Supported in part by Dr. Raymond Methods
Sackler and the Parker family. The Surveillance, Epidemiology, and End Results (SEER) Program registries were searched to
Authors’ disclosures of potential con-
identify NET cases from 1973 to 2004. Associated population data were used for incidence and
flicts of interest and author contribu- prevalence analyses.
tions are found at the end of this
Results
article.
We identified 35,618 patients with NETs. We observed a significant increase in the reported
Corresponding author: James C. Yao, annual age-adjusted incidence of NETs from 1973 (1.09/100,000) to 2004 (5.25/100,000). Using
MD, Department of Gastrointestinal
the SEER 9 registry data, we estimated the 29-year limited-duration prevalence of NETs on
Medical Oncology, Unit 426, The
University of Texas M.D. Anderson
January 1, 2004, to be 9,263. Also, the estimated 29-year limited-duration prevalence in the United
Cancer Center, 1515 Holcombe Blvd, States on that date was 103,312 cases (35/100,000). The most common primary tumor site varied
Houston, TX 77030; e-mail: by race, with the lung being the most common in white patients, and the rectum being the most
[email protected]. common in Asian/Pacific Islander, American Indian/Alaskan Native, and African American patients.
© 2008 by American Society of Clinical Additionally, survival duration varied by histologic grade. In multivariate analysis of patients with
Oncology well-differentiated to moderately differentiated NETs, disease stage, primary tumor site, histologic
0732-183X/08/2618-3063/$20.00
grade, sex, race, age, and year of diagnosis were predictors of outcome (P ⬍ .001).
DOI: 10.1200/JCO.2007.15.4377 Conclusion
We observed increased reported incidence of NETs and increased survival durations over time,
suggesting that NETs are more prevalent than previously reported. Clinicians need to be become
familiar with the natural history and patterns of disease progression, which are characteristic of
these tumors.

J Clin Oncol 26:3063-3072. © 2008 by American Society of Clinical Oncology

and the United Kingdom,2-5 much about them re-

INTRODUCTION
mains unknown. For example, the prevalence of
Neuroendocrine tumors (NETs) consist of a spec- NETs in the general population has not been well
trum of malignancies that can arise from neuroen- described. Furthermore, International Classifica-
docrine cells throughout the body. These tumors are tion of Diseases for Oncology (ICD-O-3) classifica-
characterized by their ability to produce peptides tion of NETs is complex. In particular, a significant
that cause characteristic hormonal syndromes. Most number of NETs are not classified using the ICD-
are more indolent than other epithelial malignan- O-3 codes associated with carcinoid tumors (8240-
cies; however, they can be aggressive and resistant to 8246 and 8249).6 In our present study, we
therapy. Oberndofer1 first described these tumors undertook the most complete analysis of patients
and coined the term carcinoid (or “karzinoide”) with NETs reported to date. We retrospectively an-
in 1907. alyzed the epidemiology of and prognostic factors
Although authors have described the incidence for NETs in patients identified in the Surveillance,
of NETs and the racial, sex, and primary tumor site Epidemiology, and End Results (SEER) database.
distributions and survival durations in patients with Since its inception in 1973, the SEER Program
these tumors in the United States, the Netherlands, has undergone two major expansions to improve its

© 2008 by American Society of Clinical Oncology 3063

Downloaded from ascopubs.org by 218.219.209.169 on March 2, 2017 from 218.219.209.169

Copyright © 2017 American Society of Clinical Oncology. All rights reserved.
 Yao et al

representative sampling of the US population. The SEER 9, 13, and 17 registries are linked to different population data sets, we computed the
registries cover approximately 9.5%, 13.8%, and 26.2%, respectively, age-adjusted incidence for three time periods: SEER 9, 1973 to 1991;
of the total US population. In our study, we obtained and analyzed the SEER 13, 1992 to 1999; and SEER 17, 2000 to 2004. We noted a
SEER data based on the November 2006 submission.7 The data set we significant increase in reported annual age-adjusted incidence from
used contained a total of 4,926,760 neoplasms in 4,466,501 patients 1973 (1.09/100,000) to 2004 (5.25/100,000; Fig 1A). Separate time-
diagnosed from 1973 to 2004. trend analyses of the SEER 9, 13, and 17 registries showed significant
increases in the reported incidence of NETs (P ⬍ .001 in all three
analyses). Detailed incidence data for 2000 to 2004 are presented in
METHODS Table 1. We also performed separate time-trend analyses by primary
tumor site (Fig 1B) and disease stage at diagnosis (Fig 1C). These
ICD-O-3 histology codes were used to identify NETs. These codes correspond analyses showed statistically significant increases in the reported
to the following clinical/histologic diagnoses: islet cell carcinoma (8150), insu- incidence rates over time at all primary sites (P ⬍ .001) and disease
linoma (8151), glucagonoma (8152), gastrinoma (8153), mixed islet-cell/ stages (P ⬍ .001).
exocrine adenocarcinoma (8154), vipoma (8155), somatostatinoma (8156),
enteroglucagonoma (8157), carcinoid (8240), enterochromaffin cell carcinoid
In the SEER 9 registry, the estimated 29-year limited-duration
(8241), enterochromaffin-like cell tumors (8242), goblet cell carcinoid (8243), prevalence of NETs on January 1, 2004, was 9,263. We projected this
composite carcinoid (8244), adenocarcinoid (8245), neuroendocrine carci- prevalence into the US standard population and matched by sex, race,
noma (8246), and atypical carcinoid (8249). Small-cell (8002 and 8040-8045) and age. The resulting estimated 29-year limited-duration prevalence
and large-cell neuroendocrine carcinoma (8013) of the lung, pheomochromo- of NETs on January 1, 2004, in the United States was 103,312 cases
cytoma (8700), paraganglioma (8680, 8693), and medullary carcinoma of the or 35/100,000.
thyroid (8510) were excluded.
Because a unified staging system for NETs is lacking, the SEER staging
system was used for analysis. Tumors were classified as localized, regional, Patient Characteristics
or distant. A localized NET was defined as an invasive neoplasm confined Of the 35,618 patients with NETs identified in the SEER database,
entirely to the organ of origin. A regional NET was defined as a neoplasm 18,614 (52%) were women and 17,004 (48%) were men. Eighty-one
that (1) extended beyond the limits of the organ of origin directly into percent of the patients were white, 12% were African American, 5%
surrounding organs or tissue, (2) involved regional lymph nodes, or (3)
fulfilled both of the aforementioned criteria. Finally, a distant NET was
were Asian/Pacific Islander, and 1% were American Indian/Alaskan
defined as a neoplasm that spread to parts of the body remote from the native. The race of the remaining 1% of the patients was unknown.
primary tumor. The median age at diagnosis was 63 years (mean, 62; standard devia-
There is no accepted uniformed grading system for malignant NETs. tion, 15).
Pathologists in the United States typically use the terms “carcinoid tumor” or NETs are commonly classified by embryonic origin as foregut,
“islet-cell tumor” to denote well-differentiated NETs (G1). The term “atypical midgut, or hindgut tumors. Of the 35,825 cases, 14,844 (41%) were
carcinoid” is frequently used to describe a moderately differentiated carcinoid
foregut NETs, 9,266 (26%) were midgut, and 6,963 (19%) were hind-
and is classified as G2 tumor, poorly differentiated tumors are classified as G3
tumors, and anaplastic tumors are classified as G4 tumors. Tumors with mixed gut; in the remaining 4,752 (13%), the primary tumor site was un-
differentiation, such as adenocarcinoid and goblet-cell carcinoid tumors, are known or could not be classified using this system. The disease stage in
classified as having mixed histology. 7,270 cases (20%) went unreported; of the remaining 28,515 cases,
Comparisons of patients, tumor characteristics, and disease extension 14,162 (40%) were localized, 6,718 (19%) were regional, and 7,635
were performed using the ␹2 test. One-way analysis of variance was used for (21%) were distant.
comparison of continuous variables between groups. Survival durations were
measured using the actuarial or Kaplan-Meier method and compared using
the log-rank test. The statistical independence between prognostic variables Primary Tumor Site
was evaluated using the Cox proportional hazards model. The locations of the primary tumors in these patients varied
SEER*Stat software program (version 6.3.5; National Cancer Institute, significantly by sex (P ⬍ .001; Table 1). Female patients were more
Bethesda, MD) was used for incidence and limited-duration prevalence anal- likely to have a primary NET in the lung, stomach, appendix, or
ysis.7 The counting method, which estimates prevalence by counting the
cecum, whereas male patients were more likely to have a primary
number of persons (first NET for patients with multiple primaries) who are
known to be alive at a specific date and adjusting for those lost to follow-up, tumor in the thymus, duodenum, pancreas, jejunum/ileum, or rec-
was used for prevalence analyses.5,8,9 The expected number of cases lost to tum. The primary tumor sites also varied significantly by race (P ⬍
follow-up that were included in the prevalence data was calculated using .001; Table 1). In particular, the lung was the primary NET site more
conditional survival curves for cohorts by age, sex, race, year of diagnosis, and often among white patients (30%) than among patients in the other
primary tumor site. All other statistical calculations were performed using racial groups (P ⬍ .001). Additionally, jejunal/ileal NETs were more
SPSS (version 14.0; SPSS Inc, Chicago, IL). Comparative differences were common in white (17%) and African American (15%) patients than
considered statistically significant when P was less than .05.
in Asian/Pacific Islander and American Indian/Alaskan Native pa-
tients (P ⬍ .001). In contrast, rectal NETs occurred at a markedly
RESULTS higher frequency among Asian/Pacific Islander (41%), American
Indian/Alaskan Native (32%), and African American (26%) patients
Incidence and Prevalence than among white (12%) patients (P ⬍ .001).
We identified a total of 35,825 NETs in 35,618 patients in the
SEER registries. Using population files linked to the SEER database, we Age at Diagnosis
calculated the incidence of NETs per 100,000 per year age-adjusted to We next examined age at diagnosis of NET by race, sex, and
the 2000 US standard population. Because the SEER 9, 13, and 17 primary tumor site. Overall, African American, Asian/Pacific Islander,

3064 © 2008 by American Society of Clinical Oncology JOURNAL OF CLINICAL ONCOLOGY

Downloaded from ascopubs.org by 218.219.209.169 on March 2, 2017 from 218.219.209.169

Copyright © 2017 American Society of Clinical Oncology. All rights reserved.
 Neuroendocrine Tumors: Epidemiology and Prognostic Factors

A Year Incidence (95% CI)

6.00 600
Incidence of all malignant neoplasms
1973 1.09 (0.92 to 1.28)
Incidence of neuroendocrine tumors
1974 1.40 (1.22 to 1.61)
1975 1.58 (1.39 to 1.78)
Incidence Per 100,000 for Neuroendocrine Tumors

1976 1.37 (1.20 to 1.56)

5.00 500 1977 1.61 (1.42 to 1.81)
1978 1.43 (1.27 to 1.62)
1979 1.66 (1.48 to 1.86)
1980 1.44 (1.28 to 1.63)
1981 1.30 (1.14 to 1.48)
4.00 400 1982 1.57 (1.40 to 1.76)

Incidence per 100,000

1983 1.63 (1.46 to 1.83)
1984 1.71 (1.53 to 1.91)
1985 1.70 (1.53 to 1.90)
1986 2.65 (2.42 to 2.88)
3.00 300 1987 2.76 (2.53 to 3.00)
1988 2.82 (2.59 to 3.06)
1989 3.00 (2.76 to 3.24)
1990 2.91 (2.68 to 3.15)
1991 3.17 (2.94 to 3.41)
2.00 200 1992 3.21 (3.01 to 3.41)
1993 3.31 (3.11 to 3.51)
1994 3.41 (3.21 to 3.62)
1995 3.95 (3.74 to 4.17)
1996 4.11 (3.89 to 4.33)
1.00 100 1997 4.27 (4.05 to 4.49)
1998 4.71 (4.49 to 4.95)
1999 4.73 (4.50 to 4.96)
SEER9 SEER13 SEER17 2000 4.84 (4.68 to 5.01)
2001 4.63 (4.47 to 4.79)
0 0 2002 5.06 (4.90 to 5.23)
2003 5.21 (5.04 to 5.38)
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2004 5.25 (5.09 to 5.42)

Year

B 1.40 Lung
C 3.00 Localized
Appendix Regional
Stomach Distance
1.20 Colon Unstaged
Small intestine 2.50
Rectum
Cecum
1.00
Incidence per 100,000

Incidence per 100,000

Pancreas
2.00

0.80
1.50
0.60

1.00
0.40

0.50
0.20

0 0
1973
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003

1973
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003

Year Year
Fig 1. These graphs show the incidence of neuroendocrine tumors (NETs) over time, by site and by disease stage. (A) Annual age-adjusted incidence of NETs by year
(1973 to 2004). The incidence is presented as the number of tumors per 100,000 (with 95% CIs) age-adjusted for the 2000 US standard population. Cases were selected
from the Surveillance, Epidemiology, and End Results database (1973 to 2004) using International Classification of Diseases for Oncology histology codes 8150 to 8157,
8240 to 8246, and 8249. (B) Time-trend analyses of the incidence of NETs by primary tumor site (1973 to 2004). Statistically significant increases in incidence at all sites
are shown (P ⬍ .001). (C) The incidence of NETs by disease stage at diagnosis. Statistically significant increases in incidence at all stages are shown (P ⬍ .001).

www.jco.org © 2008 by American Society of Clinical Oncology 3065

Downloaded from ascopubs.org by 218.219.209.169 on March 2, 2017 from 218.219.209.169

Copyright © 2017 American Society of Clinical Oncology. All rights reserved.
 Yao et al

Table 1. Incidence and Distribution of NETs by Sex and Race in the SEER 17 Registry (2000-2004)
Incidenceⴱ Fraction Within Sex and Racial Groups (%)
Sex Race Sex Race
African Asian/P African Asian/P
Distribution All Cases Male Female White American Islander AI/AN Male Female White American Islander AI/AN
All cases 5.00 5.35 4.76 4.92 6.82 3.19 3.07
Disease stage
Localized 2.01 2.00 2.05 1.86 3.24 1.68 1.66 47 52 47 57 65 61
Regional 0.88 0.99 0.79 0.90 1.06 0.38 0.52 24 23 25 21 15 19
Distant 1.03 1.18 0.92 1.08 1.17 0.49 0.48 29 25 28 22 20 20
Unstaged 1.08 1.18 1.01 1.08 1.36 0.53 0.53
Primary tumor site
Lung 1.35 1.30 1.40 1.45 1.17 0.50 0.70 24 30 30 18 15 22
Thymus 0.02 0.02 0.01 0.02 0.01 0.04 0.00 1 0.2 0.4 0.1 1 1
Stomach 0.30 0.29 0.31 0.29 0.39 0.23 0.35 4 6 5 5 6 9
Duodenum 0.19 0.24 0.16 0.15 0.64 0.18 0.03 4 3 2 7 4 2
Jejunum/ileum 0.67 0.80 0.57 0.71 0.88 0.09 0.09 18 14 17 15 4 5
Cecum 0.16 0.16 0.17 0.17 0.21 0.04 0.09 3 4 4 3 1 1
Appendix 0.15 0.14 0.16 0.16 0.14 0.03 0.02 3 4 4 3 2 1
Colon 0.20 0.23 0.17 0.18 0.38 0.12 0.22 4 4 4 5 4 6
Rectum 0.86 0.92 0.81 0.66 1.80 1.25 1.00 16 14 12 26 41 32
Pancreas 0.32 0.38 0.27 0.32 0.36 0.25 0.20 8 6 7 6 8 10
Liver 0.04 0.03 0.04 0.04 0.05 0.01 0.07 1 1 1 1 0.4 1
Other/unknown 0.74 0.84 0.69 0.77 0.79 0.45 0.30 14 14 15 12 12 11
Abbreviations: SEER, Surveillance, Epidemiology, and End Results database; NETs, neuroendocrine tumors; P Islander, Pacific Islander; AI/AN, American
Indian/Alaskan native.
ⴱ
Age-adjusted annual incidence per 100,000 to the 2000 US standard population.

and American Indian/Alaskan Native patients were younger at diag- Tumor Stage
nosis than white patients were (P ⬍ .001). We observed no difference Next, we examined factors associated with extent of disease and
in age at diagnosis by sex (P ⫽ .44). The ages at diagnosis did varied observed a strong correlation between primary tumor site and disease
significantly by primary tumor site (P ⬍ .001). Details regarding age at stage, among the 28,515 cases where stage information was available
diagnosis are presented in Table 2. (Table 2; P ⬍ .001). We also found that histologic grade was strongly

Table 2. Age and Disease Stage at Diagnosis of NETs by Race, Sex, and Primary Tumor Site
Age at Diagnosis (years) Disease Stage (%)

Characteristic Median Mean Standard Deviation Localized Regional Distant

Race
White 64 62 15 47 25 28
Black 59 59 14 57 21 22
Asian/P Islander 59 59 14 65 15 20
AI/AN 58 57 16 61 19 20
Sex
Male 63 62 14 47 24 29
Female 63 62 15 52 23 25
Primary tumor site
Lung 64 62 15 49 23 28
Thymus 59 56 16 28 41 31
Stomach 65 64 15 76 9 15
Duodenum 67 65 14 81 10 9
Jejunum/ileum 66 65 13 29 41 30
Cecum 68 66 14 14 42 44
Appendix 47 48 18 60 28 12
Colon 65 64 14 45 23 32
Rectum 56 57 13 92 4 5
Pancreas 60 59 15 14 22 64
Liver 67 64 15 45 27 28

NOTE. Cases selected from the SEER Program database (1973-2004) using ICD-O-3 histology codes 8150-8157, 8240-8246, and 8249.
Abbreviations: NETs, neuroendocrine tumors; P Islander, Pacific Islander; AI/AN, American Indian/Alaskan native.

3066 © 2008 by American Society of Clinical Oncology JOURNAL OF CLINICAL ONCOLOGY

Downloaded from ascopubs.org by 218.219.209.169 on March 2, 2017 from 218.219.209.169

Copyright © 2017 American Society of Clinical Oncology. All rights reserved.
 Neuroendocrine Tumors: Epidemiology and Prognostic Factors

A 1.0

0.8
Survival Probability

0.6
Median Survival
Months 95%CI
0.4 Carcinoid/islet cell: well-differentiated 124 101 to 147
Carcinoid/islet cell: unspecified grade 129 124 to 134
Carcinoid/islet cell: moderately differentiated 64 56 to 72
Neuroendocrine: poorly differentiated 10 9 to 11
0.2
Neuroendocrine: anaplastic 10 9 to 11
Neuroendocrine: unspecified grade 10 9 to 11

0 12 24 36 48 60 72 84 96 108 120
Time (months)
B 1.0 Median Survival
Months 95%CI
Localized 223 208 to 238
0.8 Regional 111 104 to 118
Distant 33 31 to 35
Survival Probability

Survival probability
0.6
Year Localized Regional Distant
1 .94 .89 .70
2 .90 .83 .57
0.4 3 .87 .78 .48
4 .85 .73 .40
5 .82 .68 .35
0.2 6 .79 .63 .29
7 .76 .59 .25
8 .74 .55 .22
9 .71 .51 .19
10 .69 .48 .17
0 12 24 36 48 60 72 84 96 108 120
Time (months)
C 1.0
Median Survival
Months 95%CI
Localized 34 27 to 41
0.8
Regional 14 13 to 15
Survival Probability

Distant 5 4.5 to 5.5

0.6
Survival probability
Year Localized Regional Distant
1 .76 .59 .25
0.4
2 .57 .36 .11
3 .48 .28 .07
4 .42 .24 .05
0.2 5 .38 .21 .04
6 .35 .20 .03
7 .32 .18 .03
8 .30 .17 .03
9 .28 .15 .02
0 12 24 36 48 60 72 84 96 108 120
10 .27 .13 .02
Time (months)
Fig 2. Survival duration by (A) histology (B) well- and moderately differentiated histology, and (C) poorly differentiated histology. Neuroendocrine tumor cases identified
at autopsy or solely on the basis of death certificates were excluded. Median survival durations are presented in months (with 95% CIs).

www.jco.org © 2008 by American Society of Clinical Oncology 3067

Downloaded from ascopubs.org by 218.219.209.169 on March 2, 2017 from 218.219.209.169

Copyright © 2017 American Society of Clinical Oncology. All rights reserved.
 Yao et al

linked with disease stage (P ⬍ .001). Among patients with NETs with patients with metastatic disease, those with mixed-histology tumors
explicitly stated tumor histologic grades, 21% of those with well- had worse outcomes than did those with G2 NETs.
differentiated (G1) tumors and 30% of those with moderately differ- Age at diagnosis (P ⬍ .001; Fig 3C), sex (P ⬍ .001; Fig 3D), and race
entiated (G2) tumors had synchronous distant metastasis at diagnosis, (P ⬍ .001; Fig 3E) were also prognostic of survival. Women had better
whereas 50% of those with poorly differentiated (G3) tumors or survival durations than men did in all stage categories. Also, Asian/Pacific
undifferentiated (G4) tumors had synchronous distant metastasis Islander and American Indian/Alaskan Native patients had the best sur-
at diagnosis. vival durations among patients with localized disease (median survival
Other factors associated with disease stage included race and sex duration not reached), whereas white patients had the best survival dura-
(Table 2). White patients were the most likely to present with advanced tions among patients with metastatic disease. We also examined the effect
disease (P ⬍ .001), with 28% having synchronous distant metastasis at of age at diagnosis on survival by separating the patients into three groups
diagnosis. Also, male patients were more likely to have metastasis at pre- (ⱕ 30, 31 to 60, and ⬎ 60 years). We found age to be a strong predictor of
sentation than female patients were (29% v 25%; P ⬍ .001). survival duration (P ⬍ .001; Fig 3C).
Next, we sought to determine whether the survival durations
improved in patients with NETs over time. Because the somatostatin
Survival
analog octreotide was the only new drug introduced for use against
For survival analyses, we excluded 521 cases that were identified
NETs during this period (in 1987), we compared the survival dura-
at autopsy or solely on the basis of death certificates. The median
tions in patients who received diagnoses from 1973 to 1987 with those
overall survival duration in the remaining 35,097 cases was 75 months.
who received diagnoses from 1988 to 2004 (Fig 3G). Although the
When we examined survival by histologic grade (Fig 2A), we found
survival durations did not improve significantly among patients with
that the median survival duration in patients with G1 and G2 NETs
localized NETs (hazard ratio [HR] ⫽ 0.96; 95% CI, 0.87 to 1.06; P ⫽
was 124 and 64 months, respectively. Patients with G3 and G4 tumors
.43) or regional NETs (HR ⫽ 0.91; 95% CI, 0.82 to 1.01; P ⫽ .08), they
had identical survival curves; the median survival duration in these
improved dramatically among patients with metastatic disease (HR ⫽
patients was 10 months. Among cases where histologic grade was not
0.67; 95% CI, 0.62 to 0.73; P ⬍ .001).
explicitly stated, those with ICD-O-3– designated neuroendocrine Finally, we performed multivariate survival analysis of G1/G2
histology and those with G3 or G4 tumors had identical survival NETs using the Cox proportional hazards model. We included poten-
curves; the median survival duration in these patients was 10 months. tially prognostic parameters such as disease stage, primary tumor site,
The survival curves for those with ICD-O-3– designated carcinoid or histology, age, sex, race, and period of diagnosis (1973 to 1987 and
islet cell histology but an unspecified tumor grade were similar to 1988 to 2004) in this model. We found that all of the parameters that
those for patients with G1 tumors; the median survival duration in were significant in the univariate analysis were also significant in the
these patients was 129 months. The difference in survival duration multivariate analysis (Table 3).
between the patients with G1, G2, and G3/G4 NETs was statistically Survival for G3/G4 tumors. Poorly differentiated NETs, which are
significant (P ⬍ .001). also known as high-grade NETs, are aggressive and associated with poor
Survival for G1/G2 tumors. We found several factors, including survival. We analyzed the survival of 4,054 patients with G3/G4 NETs in
disease stage (P ⬍ .001), to be predictors of outcome. The median the SEER registries (1973 to 2004). The median survival durations in
survival durations in patients with G1/G2 NETs who had localized, patientswithlocalized,regional,anddistantdiseasewere34months(95%
regional, and distant disease was 223 months, 111 months, and 33 CI, 27 to 41 months), 14 months (95% CI, 13 to 15 months), and 5
months, respectively (Fig 2B). We then examined potential prognostic months (95% CI, 4.5 to 5.5 months), respectively (Fig 2C).
factors for survival duration stratified by disease stage and found the
primary tumor site to be a powerful predictor of survival duration
(P ⬍ .001). The median survival durations among patients with local- DISCUSSION
ized NETs varied from greater than 360 months (appendiceal tumors)
to 111 months (jejunal/ileal tumors) to 50 months (liver tumors). In this study, we took advantage of the vast amount of data collected by
Among patients with regional NETs, the median survival durations the SEER Program to examine the largest series of NET cases reported
varied from 360 months (appendiceal tumors) to 36 months (colon to date with a focus on incidence, prevalence, and prognostic factors.
tumors [excluding cecal and rectal tumors]) to 14 months (liver tu- Similar to those of previous reports,3 our results indicated a significant
mors). In addition, among patients with metastasis, the median sur- increase in the reported incidence of NETs over time. This increase
vival durations varied from 56 months (jejunal/ileal tumors) to 5 was likely caused in part by improvements in classification of these
months (colon tumors [excluding cecal and rectal tumors]). Details tumors. Also, widespread use of endoscopy for cancer screening likely
regarding the results of these analyses by primary tumor site are contributed to the increase in reported incidence of rectal carcinoid
presented in Figure 3A. NETs. Whether changes in dietary habits, environmental factors, and
Another significant predictor of outcome was histopathology. In use of certain medications such as proton pump inhibitors resulted in
addition to tumor grade, the presence of adenocarcinoma features in increased reported incidence of NETs of various types is unknown.
mixed-histology NETs has been thought to portend a poor prognosis. Prevalence of a disease is defined as the number of people alive on
We compared the survival durations in patients with G1, G2, and a certain date in a population who have never had a diagnosis of that
mixed-histology NETs stratified by disease stage. Those with G1 tu- disease. In our study, we used the counting method8-10 to estimate
mors had the best outcomes in all stage groups (P ⬍ .001; Fig 3B). prevalence from incidence and follow-up data. Complete prevalence
Interestingly, patients with local/regional mixed-histology tumors had can be determined using this method with registries containing data
better outcomes than did those with G2 NETs. However, among obtained over long periods of time. Given the long survival durations

3068 © 2008 by American Society of Clinical Oncology JOURNAL OF CLINICAL ONCOLOGY

Downloaded from ascopubs.org by 218.219.209.169 on March 2, 2017 from 218.219.209.169

Copyright © 2017 American Society of Clinical Oncology. All rights reserved.
 Neuroendocrine Tumors: Epidemiology and Prognostic Factors

A Localized Regional
1.0 1.0
Survival Probability

Survival Probability
0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0 12 24 36 48 60 72 84 96 108 120 0 12 24 36 48 60 72 84 96 108 120

Time (months) Time (months)

1.0
Distant Median Survival (months)
Color Site Localized Regional Distant
Survival Probability

0.8
Appendix >360 >360 27
Cecum 135 107 41
0.6 Colon 261 36 5
Duodenum 107 101 57
Gastric 154 71 13
0.4
Liver 50 14 12
Lung 227 154 16
0.2 Pancreas 136 77 24
Rectum 290 90 22
Small bowel 111 105 56
Thymus 110 68 40
0 12 24 36 48 60 72 84 96 108 120

Time (months)
B 1.0 Localized 1.0 Regional 1.0 Distant
Median Survival
Survival Probability

Survival Probability

Survival Probability

0.8 0.8 0.8 Grade 1 35 months

Grade 2 24 months
Mixed histology 19 months
0.6 0.6 0.6
Grade 1
Grade 1 Grade 1 Grade 2
Grade 2 Grade 2 Mixed histology
0.4 Mixed histology 0.4 Mixed histology 0.4

Median Survival Median Survival

0.2 Grade 1 226 months 0.2 Grade 1 116 months 0.2
Grade 2 168 months Grade 2 64 months
Mixed histology 199 months Mixed histology 93 months

0 12 24 36 48 60 72 84 96 108 120 0 12 24 36 48 60 72 84 96 108 120 0 12 24 36 48 60 72 84 96 108 120

Time (months) Time (months) Time (months)

C 1.0
Localized
1.0
Regional
1.0
Distant
< 30
31-60
Survival Probability

Survival Probability

Survival Probability

0.8 0.8 0.8 > 60

0.6 0.6 0.6

< 30 < 30
0.4 31-60 0.4 31-60 0.4
> 60 > 60
Median Survival Median Survival Median Survival
0.2 < 30 > 360 months 0.2 < 30 > 360 months 0.2 < 30 76 months
31-60 362 months 31-60 218 months 31-60 53 months
> 60 116 months > 60 69 months > 60 22 months

0 12 24 36 48 60 72 84 96 108 120 0 12 24 36 48 60 72 84 96 108 120 0 12 24 36 48 60 72 84 96 108 120

Time (months) Time (months) Time (months)

Fig 3. (A) Survival duration by primary tumor site. Neuroendocrine tumor cases identified at autopsy or solely on the basis of death certificates were excluded, as were those with
missing site and/or stage data. Median survival durations are presented in months. (B) Survival duration by histology. G1 tumors had the best outcome in all staging groups (P ⬍ .001).
(Continued on next page)

www.jco.org © 2008 by American Society of Clinical Oncology 3069

Downloaded from ascopubs.org by 218.219.209.169 on March 2, 2017 from 218.219.209.169

Copyright © 2017 American Society of Clinical Oncology. All rights reserved.
 Yao et al

D 1.0 Localized 1.0 Regional 1.0 Distant

Survival Probability

Survival Probability

Survival Probability
0.8 0.8 0.8

0.6 0.6 0.6

0.4 0.4 0.4

0.2 Median survival 0.2 Median survival 0.2 Median survival

Male 201 months Male 106 months Male 31 months
Female 251 months Female 122 months Female 35 months

0 12 24 36 48 60 72 84 96 108 120 0 12 24 36 48 60 72 84 96 108 120 0 12 24 36 48 60 72 84 96 108 120

Time (months) Time (months) Time (months)

E 1.0
Localized
1.0
Regional
1.0
Distant
Median survival
AI/AN 12 months
Asian/PI 25 months
Survival Probability

Survival Probability

Survival Probability
0.8 0.8 0.8 Black 33 months
White 34 months

0.6 0.6 0.6

0.4 0.4 0.4

Median survival Median survival
AI/AN > 360months AI/AN 109 months
0.2 Asian/PI > 360 months 0.2 Asian/PI 96 months 0.2
Black 199 months Black 100 months
White 224 months White 113 months

0 12 24 36 48 60 72 84 96 108 120 0 12 24 36 48 60 72 84 96 108 120 0 12 24 36 48 60 72 84 96 108 120

Time (months) Time (months) Time (months)

F 1.0 Localized 1.0 Regional 1.0 Distant

Median survival
Survival Probability

Survival Probability

Survival Probability

0.8 0.8 0.8 1973-1987 18 months

1988-2004 39 months

0.6 0.6 0.6

0.4 0.4 0.4

0.2 Median survival 0.2 Median survival 0.2

1973-1987 223 months 1973-1987 104 months
1988-2004 203 months 1988-2004 114 months

0 12 24 36 48 60 72 84 96 108 120 0 12 24 36 48 60 72 84 96 108 120 0 12 24 36 48 60 72 84 96 108 120

Time (months) Time (months) Time (months)

Fig 3 (Continued). (C) Survival duration by age at diagnosis. Patients were separated into three groups according to their age at diagnosis (ⱕ 30, 31 to 60, and ⬎ 60
years). Age was found to be a strong predictor of outcome (P ⬍ .001). (D) Survival duration by sex. Women had statistically significantly longer survival durations over
all three categories histologies (P ⬍ .001). (E) Survival duration by race. Patients were separated into four categories on the basis of race (American Indian/Alaskan
Native [AI/AN], Asian/Pacific Islander [Asian/PI], African American, and white). American Indian/Alaskan Native and Asian/Pacific Islander patients had the longest
survival durations for localized tumors, whereas white patients had the longest survival durations for metastatic disease. (F) Survival duration by period of diagnosis.
Patients were separated into two groups by year of diagnosis (1973 to 1987 and 1988 to 2004). Patients with metastatic disease had an improvement in median survival
duration (P ⬍ .001; from 8 to 39 months). There were no significant improvements in survival duration among patients with localized or regional disease. Each set of
three graphs shows localized, regional, and distant survival from left to right.

often experienced by patients with NETs, we report here only 29-year than esophageal cancer (28,664), gastric cancer (65,836), pancre-
limited-duration prevalence, which estimates the number of people atic cancer (32,353), and hepatobiliary cancer (21,427) in the
alive on January 1, 2004, who were diagnosed with NET during the United States.11
preceding 29 years. Clearly, however, NETs are more common than Using multivariate survival analysis, we found that disease stage,
generally believed. For example, when compared with other GI neo- primary tumor site, histology, age, sex, race, and period of diagnosis
plasms, the estimated 29-year limited-duration prevalence of NETs of (1973 to 1987 and 1988 to 2004) were important predictors of out-
103,312 in 2004 makes these tumors significantly more common come. We found the primary tumor site to be perhaps the most useful

3070 © 2008 by American Society of Clinical Oncology JOURNAL OF CLINICAL ONCOLOGY

Downloaded from ascopubs.org by 218.219.209.169 on March 2, 2017 from 218.219.209.169

Copyright © 2017 American Society of Clinical Oncology. All rights reserved.
 Neuroendocrine Tumors: Epidemiology and Prognostic Factors

Table 3. Survival Analysis of Patients with Well-Differentiated to Moderately Differentiated NETs: Univariate and Multivariate Cox Proportional Hazards
Analysis of G1/G2 NETs Diagnosed From 1973 to 2004
Univariate Multivariate

Parameter Median Survival (months) Hazard Ratio 95% CI Hazard Ratio 95% CI Multivariate P
Disease stage
Localized 223 1ⴱ — 1ⴱ — ⬍ .001
Regional 111 1.89 1.79 to 2.01 1.60 1.50 to 1.71
Distant 33 4.93 4.68 to 5.21 3.85 3.60 to 4.11
Primary tumor site
Jejunum/ileum 88 1ⴱ — 1ⴱ — ⬍ .001
Lung 193 0.53 0.50 to 0.57 1.01 0.93 to 1.08
Thymus 77 1.12 0.82 to 1.53 1.47 1.06 to 2.03
Stomach 124 0.83 0.75 to 0.91 1.54 1.38 to 1.73
Duodenum 99 0.89 0.78 to 0.99 1.42 1.24 to 1.62
Cecum 83 1.16 1.05 to 1.29 1.06 0.96 to 1.18
Appendix NR 0.33 0.29 to 0.37 0.66 0.57 to 0.76
Colon 121 0.93 0.84 to 1.03 1.54 1.38 to 1.71
Rectum 240 0.32 0.29 to 0.34 0.74 0.67 to 0.82
Pancreas 42 1.65 1.53 to 1.78 1.65 1.53 to 1.79
Liver 23 2.20 1.76 to 2.75 2.92 2.25 to 3.79
Histology
Well-differentiated 134 1ⴱ — 1ⴱ — ⬍ .001
Moderately differentiated 64 1.67 1.53 to 1.82 1.26 1.15 to 1.40
Mixed 135 1.02 0.92 to 1.14 1.65 1.45 to 1.88
Sex
Female 145 1ⴱ — 1ⴱ — ⬍ .001
Male 114 1.21 1.16 to 1.26 1.20 1.14 to 1.25
Race
White 126 1ⴱ — 1ⴱ — ⬍ .001
AI/AN NR 0.56 0.36 to 0.87 0.79 0.50 to 1.26
Asian/P Islander 204 0.65 0.58 to 0.72 0.94 (0.83 to 1.07)
African American 117 1.04 0.98 to 1.10 1.28 1.19 to 1.37
Age, years
ⱕ 30 NR 1ⴱ — 1ⴱ — ⬍ .001
31-60 247 3.31 2.74 to 4.00 3.03 2.41 to 3.81
ⱖ 61 71 10.08 8.36 to 12.15 9.23 7.34 to 11.61
Year of diagnosis
1973-1987 95 1ⴱ — 1ⴱ — ⬍ .001
1988-2004 138 0.75 0.72 to 0.79 0.73 0.69 to 0.77

Abbreviations: NET, neuroendocrine tumor; NR, not reached; AI/AN, American Indian/Alaskan Native; P Islander, Pacific Islander.
ⴱ
Referent.

predictor of outcome in patients with NETs. Using the primary tumor effect in patients with NETs,12-14 conclusive data from randomized
site as a prognostic marker, we were better able to separate outcomes human studies are lacking.
into categories. We therefore included a table of survival duration by We acknowledge that our analysis of data obtained from the
primary tumor site and disease stage for patients who received diag- SEER registries likely underestimated the total number of patients
noses from 1988 to 2004 as a practical guide for clinicians in Table 4. with NETs. Only patients with malignant NETs are included in the
In our analyses, we did not observe a statistically significant SEER registries. Thus, data on many small, benign-appearing tumors
difference in survival duration among patients with local and regional (ie, appendiceal tumors) likely are excluded from the registries.
NETs over time. However, we observed a dramatic improvement in Whereas histologic evidence of invasion of a basement membrane
survival duration among patients with metastatic NETs diagnosed in defines malignant behavior for most epithelial malignancies, the def-
the later period (1988 to 2004). One possible explanation is that the inition of malignant behavior for NETs is more complex. In the
introduction of octreotide in 1987 improved the control of carcinoid absence of obvious malignant behavior, such as direct invasion
syndrome and changed the natural history of NETs. For example, of adjacent organs and metastasis to regional lymph nodes or
carcinoid crisis with severe flushing, diarrhea, and hemodynamic in- distant sites, classifying a NET as benign or malignant may be
stability, which was a major cause of morbidity and mortality in the difficult. Thus, whereas SEER registry data provide important
past, now occurs rarely. Organ failure, which tends to occur later in the information about malignant NETs, the extent to which these
course of illness, is now the major cause of mortality. Whereas many data underestimate the frequency of small, benign-appearing
researchers have speculated that octreotide has a disease-stabilizing NETs is unknown.

www.jco.org © 2008 by American Society of Clinical Oncology 3071

Downloaded from ascopubs.org by 218.219.209.169 on March 2, 2017 from 218.219.209.169

Copyright © 2017 American Society of Clinical Oncology. All rights reserved.
 Yao et al

Table 4. Survival Analysis of Patients with Well-Differentiated to Moderately Differentiated NETs: Actuarial Survival by Disease Stage and Primary Tumor
Site in Patients With G1/G2 NETs Diagnosed From 1988 to 2004
Localized Regional Distant

Median Survival Rate (%) Median Survival Rate (%) Median Survival Rate (%)
Primary Tumor Survival Survival Survival
Site (months) 3-Year 5-Year 10-Year (months) 3-Year 5-Year 10-Year (months) 3-Year 5-Year 10-Year
Thymus 92 93 93 52 68 78 65 49 40 62 32 0
Lung NR 89 84 70 151 77 72 56 17 34 27 15
Pancreas NR 83 79 58 111 73 62 46 27 42 27 11
Liver 47 64 43 — 14 32 27 — 12 34 26 0
Gastric 163 80 73 56 76 75 65 43 13 33 25 9
Duodenum 112 80 68 48 69 75 55 44 57 60 46 27
Jejunum/ileum 115 73 65 49 107 83 71 46 65 70 54 30
Cecum 135 74 68 55 107 78 71 44 55 61 48 23
Colon NR 90 85 74 52 60 46 33 7 20 14 6
Rectum NR 94 90 80 90 74 62 47 26 37 24 3
Appendix NR 93 88 72 NR 86 78 67 31 42 25 11

Abbreviations: NET, neuroendocrine tumor; NR, not reached.

At present, surgery is the only curative treatment for NETs, and is Declaration and the Disclosures of Potential Conflicts of Interest section in
recommended for most patients for whom cross-sectional imaging sug- Information for Contributors.
gests that complete resection is possible.15,16 Although NETs generally Employment or Leadership Position: None Consultant or Advisory
Role: Jean-Nicolas Vauthey, Sanofi-Aventis (C), Genentech (C) Stock
have a better prognosis than adenocarcinomas at the same site, NETs are Ownership: None Honoraria: Jean-Nicolas Vauthey, Sanofi-Aventis,
incurable once they advance to unresectable metastatic disease. New ther- Genentech Research Funding: Jean-Nicolas Vauthey, Sanofi-Aventis
apeutic approaches for NETs, such as peptide receptor radiotherapy and Expert Testimony: None Other Remuneration: None
systemicagentstargetingvascularendothelialgrowthfactorandmamma-
lian target of rapamycin, are under development.17 AUTHOR CONTRIBUTIONS

Conception and design: James C. Yao

AUTHORS’ DISCLOSURES OF POTENTIAL CONFLICTS Financial support: James C. Yao
OF INTEREST Administrative support: James C. Yao
Collection and assembly of data: James C. Yao, Cecile Dagohoy,
Although all authors completed the disclosure declaration, the following Jeannette E. Mares
author(s) indicated a financial or other interest that is relevant to the subject Data analysis and interpretation: James C. Yao, Manal Hassan,
matter under consideration in this article. Certain relationships marked Alexandria Phan, Colleen Leary, Jean-Nicolas Vauthey, Asif Rashid,
with a “U” are those for which no compensation was received; those Douglas B. Evans
relationships marked with a “C” were compensated. For a detailed Manuscript writing: James C. Yao, Asif Rashid, Douglas B. Evans
description of the disclosure categories, or for more information about Final approval of manuscript: James C. Yao, Jeannette Mares, Eddie K.
ASCO’s conflict of interest policy, please refer to the Author Disclosure Abdalla, Jason B. Fleming, Douglas B. Evans

6. Fritz AG: International Classification of Dis- 11. SEER Cancer Statistics Review, 1975-2004.
REFERENCES eases for Oncology: ICD-O (ed 3). Geneva, Switzer- http://seer.cancer.gov/csr/1975_2004/
land, World Health Organization, 2000 12. Schally A: Oncological applications of soma-
1. Obendorfer S: Karzinoide tumoren des 7. Surveillance, Epidemiology, and End Results tostatin analogue. Cancer Res 48:6977-6985, 1988
dunndarms. Frankf Z Pathol 1:425-429, 1907 (SEER) Program: SEER*Stat Database: SEER 17 Regs 13. Hillman N, Herranz L, Alvarez C, et al: Efficacy
2. Godwin JD: Carcinoid tumors. An analysis of Nov 2006 sub (1973-2004). Bethesda, MD, National of octreotide in the regression of a metastatic car-
2,837 cases. Cancer 36:560-569, 1975 Cancer Institute, 2007 cinoid tumour despite negative imaging with In-111-
8. Gail MH, Kessler L, Midthune D, et al: Two pentetreotide (Octreoscan). Exp Clin Endocrinol
3. Modlin IM, Lye KD, Kidd M: A 5-decade
approaches for estimating disease prevalence from Diabetes 106:226-230, 1998
analysis of 13,715 carcinoid tumors. Cancer 97:934-
population-based registries of incidence and total 14. Oberg K: Treatment of neuroendocrine tu-
959, 2003
mortality. Biometrics 55:1137-1144, 1999 mors. Cancer Treat Rev 20:331-355, 1994
4. Lepage C, Rachet B, Coleman MP: Survival
9. Gigli A, Mariotto A, Clegg LX, et al: Estimating 15. Kulke MH, Mayer RJ: Carcinoid tumors.
from malignant digestive endocrine tumors in En- the variance of cancer prevalence from population- N Engl J Med 340:858-868, 1999
gland and wales: A population-based study. Gastro- based registries. Stat Methods Med Res 15:235- 16. Schnirer II, Yao JC, Ajani JA: Carcinoid: A compre-
enterology 132:899-904, 2007 253, 2006 hensive review. Acta Oncol 42:672-692, 2003
5. Yao JC, Eisner MP, Leary C, et al: Population- 10. Byrne J, Kessler LG, Devesa SS: The preva- 17. Yao JC: Molecular targeted therapy for carci-
based study of islet cell carcinoma. Ann Surg Oncol lence of cancer among adults in the United States: noid and islet-cell carcinoma. Best Pract Res Clin
14:3492-3500, 2007 1987. Cancer 69:2154-2159, 1992 Endocrinol Metab 21:163-172, 2007

■ ■ ■

3072 © 2008 by American Society of Clinical Oncology JOURNAL OF CLINICAL ONCOLOGY

Downloaded from ascopubs.org by 218.219.209.169 on March 2, 2017 from 218.219.209.169

Copyright © 2017 American Society of Clinical Oncology. All rights reserved.