1664

COMMUNICATION
The American College of
Surgeons Commission on
Cancer and the American The National Cancer Data Base
Cancer Society
Report on Cutaneous and
Noncutaneous Melanoma
A Summary of 84,836 Cases from the Past Decade
Alfred E. Chang, M.D.1
Lucy Hynds Karnell, Ph.D.2,3
Herman R. Menck, M.B.A.3

1
Division of Surgical Oncology, University of Michigan Medical Center, Ann Arbor, Michigan.
2
Department of Otolaryngology–Head and Neck Surgery, University of Iowa Hospitals and Clinics,
Iowa City, Iowa.
3
National Cancer Data Base, Commission on Cancer, American College of Surgeons, Chicago, Illinois.

BACKGROUND. This study reviews the case-mix characteristics, management, and

outcomes of melanoma cases occuring in the U. S. within the last decade.
METHODS. Analyses of the National Cancer Data Base (NCDB) were performed on
cases diagnosed between 1985 through 1994. A total of 84,836 cases comprised of
cutaneous and noncutaneous melanomas were evaluated.
RESULTS. The percentages of melanomas that were cutaneous, ocular, mucosal, and
unknown primaries were 91.2%, 5.2%, 1.3%, and 2.2%, respectively. For cutaneous
melanomas, the proportion of patients presenting with American Joint Committee on
Cancer Stages 0, I, II, III, and IV were 14.9%, 47.7%, 23.1%, 8.9%, and 5.3%, respectively.
Factors associated with decreased survival included more advanced stage at diagnosis,
nodular or acral lentiginous histology, increased age, male gender, nonwhite race, and
lower income. Multivariate analysis identified stage, histology, gender, age, and in-
come as independent prognostic factors. For ocular melanomas, 85.0% were uveal,
4.8% were conjunctival, and 10.2% occurred at other sites. During the study period,
there was a large increase in the proportion of ocular melanoma patients treated with
radiation therapy alone. For mucosal melanomas, the distribution of head and neck,
female genital tract, anal/rectal, and urinary tract sites was 55.4%, 18.0%, 23.8%, and
2.8%, respectively. Patients with lymph node involvement had a poor prognosis. For
unknown primary melanomas, the distribution of metastases as localized to a region
or multiple sites at presentation was 43.0% and 57.0%, respectively. Surgical treatment
of patients with unknown primary site of the melanoma resulted in better survival
compared with no treatment.
CONCLUSIONS. Treatment of early stage cutaneous melanoma resulted in excellent
patient outcomes. In addition to conventional prognostic factors, socioeconomic
factors were found to be associated with survival. Cancer 1998;83:1664 –78.
© 1998 American Cancer Society.

Address for reprints: Herman R. Menck, M.B.A., KEYWORDS: melanoma, cutaneous melanoma, outcomes, National Cancer Data
National Cancer Data Base, Commission on Cancer Base (NCDB).
at the American College of Surgeons, 55 East Erie
Street, Chicago, IL 60611.

Received April 17, 1998; revision received August

3, 1998; accepted August 4, 1998.
T he common feature of all melanomas is the cell of origin, the
melanocyte. Melanocytes are pigmented dendritic-like cells lo-
cated in various anatomic sites. These sites include the base of the

© 1998 American Cancer Society

 NCDB Report on Melanoma/Chang et al. 1665

epidermis, the eye, and in epithelia of the nasal cavity, is possible that the counts in this article represent an
oropharynx, anus, vagina, and urinary tract. Cutane- undersampling of these types of patients.
ous melanomas are much more prevalent than non-
cutaneous melanomas, which are comprised of ocular METHODS
and mucosal sites. Hospital-based cancer registries provide their data to
The incidence of cutaneous melanoma is increas- the NCDB through a computerized format.6 To ensure
ing at a greater rate than any other human cancer in standardization, the data are coded at the hospital
the U. S., and the increase in mortality is second only level according to the Data Acquisition Manual,8 the
to lung cancer.1 Since 1973, the incidence rate has third and fourth editions of the American Joint Com-
been rising 4% to 6% each year. In 1998, it is estimated mittee on Cancer (AJCC) Manual for Staging of Can-
that approximately 41,600 new cases will be diagnosed cer,9,10 and the second edition of the World Health
in the U. S.2 It is estimated that there will be 7200 Organization’s International Classification of Disease
deaths due to melanoma in the U. S., or 1 death every for Oncology (ICD-O 2).11
hour and 13 minutes. Lifetime analysis reveals that Cases to be included in this study were extracted
approximately 1 in 75 persons born in the year 2000 from the data base according to histology (ICD-O 2
will develop a cutaneous melanoma during his or her melanoma codes M872–M879), year of diagnosis
lifetime.3 One of the major factors associated with this (1985–1994), the patient’s having been diagnosed
rise in the incidence of cutaneous melanomas is the and/or treated at the reporting hospital, and the pa-
increased exposure of individuals to ultraviolet radia- tients’ having presented with a primary tumor (vs.
tion. Unlike cutaneous melanomas, the incidence of recurrent disease). An algorithm based on selected
both ocular and mucosal melanomas is believed to patient and disease characteristics was employed to
identify and eliminate duplicate cases in which more
have remained stable.4,5
than one hospital submitted a record on the same
The purpose of this article was to review the case-
patient (e.g., a patient who received surgery at one
mix characteristics, management, and outcomes of
facility and radiation therapy at another).
melanoma in the U. S. for the last decade. This report
AJCC staging, used to describe the extent of dis-
includes patients diagnosed with cutaneous, mucous
ease, was applied to melanomas of the skin. This stag-
membrane, and ocular melanomas as well as melano-
ing system employs Breslow’s depth of penetration
mas of unknown primary sites. To depict these pat-
and Clark’s level of invasion, regional lymph node
terns of treatment and survival, analyses were per-
status, and the presence or absence of metastatic dis-
formed on the National Cancer Data Base (NCDB), a
ease. (Breslow’s depth and Clark’s level were not avail-
database of cancer registry information created by the
able for separate analysis.) For selected analyses, stage
American College of Surgeons’ Commission on Cancer
was grouped into early (Stages 0 –II) and advanced
(COC) and the American Cancer Society in 1987.6,7 All (Stages III–IV). Melanomas involving the eyelid were
acute care hospitals are invited to submit their cancer grouped with cutaneous melanomas. Although mela-
registry data on an annual basis. For the most current noma of the eyelid has a separate staging system, it is
NCDB call for data, 1227 hospitals contributed 689,714 equivalent to the staging system for cutaneous mela-
records, representing an estimated 57% of all newly nomas. Histologies were classified as per ICD-O 2
diagnosed cancer patients in 1994. Unlike single-insti- codes. The four main histologic groupings, represent-
tution studies, the NCDB registry data are able to ing cutaneous melanomas, were: nodular (M8721),
provide treatment and survival information on a very lentigo maligna (M8742), superficial spreading
large number of patients from a broad range of com- (M8743), and acral lentiginous (M8744) melanomas.
munity-based and academic hospitals. These cancer The remaining histologies were categorized as “other”
registry data are a direct reflection of the actual infor- (e.g., melanoma, not otherwise specified [NOS],
mation that is being compiled on these patients; for amelanotic, desmoplastic, or spindle cell). Ocular mel-
this reason, the database at times contains a large anomas were lesions involving the conjunctiva and
percentage of cases with unknown variables (such as uvea. Uveal melanomas involved tumors arising in the
stage and grade) with no capacity to audit the infor- iris, ciliary body, and choroid. AJCC staging was re-
mation. corded for uveal melanomas. However, AJCC stage
Although cancer registries are planning to collect groupings for conjunctival melanomas currently are
records directly from physicians’ offices in the year not recommended. Mucosal melanomas are mucous
2000, the data in this article were not collected using membrane lesions arising in the head and neck, fe-
this protocol. Because certain early stage melanomas male genital tract, anal/rectal, and urinary tract. There
can be diagnosed and treated in physicians’ offices, it is no staging system for mucosal melanomas, but
1666 CANCER October 15, 1998 / Volume 83 / Number 8

lymph node status was a data element collected by RESULTS

cancer registries and analyzed in this report. Melano- Disease Characteristics
mas presenting as isolated metastatic lesions from The anatomic site of melanomas changed little across
unknown primary sites (i.e., no documentation of aris- these two year groups (Table 1A). The vast majority of
ing from cutaneous, ocular, or mucosal primary mel- melanomas originated in the skin (91.2% for all years),
anomas) were coded as “unknown primary melano- with the next highest percentage being ocular (5.3%
mas” (according to ICD-O 2).11 The presentation of for all years). Other melanoma sites included un-
these unknown primary melanomas as regional (i.e., known primary melanomas and mucous membrane
lymph node) or distant metastasis (i.e., nonlymph melanomas comprising 2.2% and 1.3% of all reported
node) was determined from the General Summary melanomas, respectively. Among the unknown pri-
Stage (according to the Data Acquisition Manual)8 and mary melanomas, 43.0% presented with regional (i.e.,
the presence of lymph node disease. lymph node) metastasis and 57.0% presented with
In addition to tumor site, histology, and staging, distant metastasis. Among mucosal melanomas, the
data regarding patient characteristics including age, incidence of head and neck, female genital, anal/rec-
gender, race/ethnicity, and income also were col- tal, and urinary tract melanomas was 55.4%, 18.0%,
lected. Patient income was based on the average fam- 23.8%, and 2.8%, respectively.
ily income of the patient’s zip code of residence as per The histologic and stage distributions for cutane-
the U. S. Census. Income was grouped into low (, ous melanomas are illustrated in Table 1B. A large
$20,000, which was approximately the lowest 10% of variety of histologic classifications were recorded.
all NCDB cancer cases), high ($ $47,000, which was Among these classifications, four represented the
approximately the highest 10% of all NCDB cancer more traditional “phenotypic” characteristics that
cases), and a middle-income group ranging from commonly are referred to in clinical practice, which
$20,000 – 46,999.6 include lentigo maligna, superficial spreading, nodu-
lar, and acral lentiginous melanomas. There was no
In this report, treatment represents the first
difference in the incidence of these phenotypic cate-
course of cancer-directed therapy for the index dis-
gories between the two year groups. Among these four
ease. More detailed surgical information (e.g., margin
categories for all years (excluding histologies classified
width) and subsequent management of persistent or
as “other”), the percentages of lentigo maligna, super-
recurrent disease are not available.
ficial spreading, nodular, and acral lentiginous mela-
All analyses were performed using SPSS soft-
nomas were 21.4%, 57.6%, 18.9%, and 2.1%, respec-
ware.12 To determine changes across time, patient and
tively. Histologies included in the group labeled
disease characteristics and treatment were broken
“other” comprised 53.4% of all the cutaneous mela-
down by diagnostic year groupings (with 1985–1989
noma cases. Of these “other” histologies, 92.5% were
representing the earlier time period and 1990 –1994
“melanoma, NOS” cases, with the remaining repre-
the later time period). Case-mix and treatment vari- senting unique, descriptive histologic features.
ables were broken down further by patient and dis- A large percentage of cutaneous melanomas were
ease stratifications of interest, with chi-square statis- diagnosed at a localized stage, with 85.7% being Stage
tics performed on selected cross-tabulations. 0 –II for all years (Table 1B). Although this percentage
Outcome information was calculated on pa- changed little between 1985–1989 and 1990 –1994, the
tients diagnosed in the earlier time period only to proportion of early stage melanomas that were re-
allow for sufficient follow-up. Annual survival rates, ported as noninvasive (Stage 0) increased from 11.4%
illustrated through 5 years using the life tables in the earlier time period to 16.4% in the later time
method, represent disease specific rates from the period. For all years, the percentage of patients pre-
date of diagnosis to the endpoint (defined as death senting with localized invasive melanomas (Stage I
with cancer). Pairwise comparisons using the Wil- and II) was 70.8%. Patients who presented with re-
coxon statistic were performed on selected survival gional lymph node disease (Stage III) or metastatic
analyses to determine whether the stratifications disease (Stage IV) at the time of their diagnosis was
had significantly different survival rates. For cuta- 8.9% and 5.3%, respectively, for all years. There was an
neous melanoma, those variables with significantly impressive decrease in the proportion of patients who
different (P , 0.05) survival rates in the univariate were not staged, dropping from 42.0% in 1985–1989 to
analyses were entered into a Cox proportional haz- 17.6% in 1990 –1994.
ards model of regression analysis to determine their The stages at presentation for the different histol-
independent influence on survival. ogies of cutaneous melanoma (all years combined) are
 NCDB Report on Melanoma/Chang et al. 1667

TABLE 1A
Disease Sites by Diagnostic Year for Melanoma

1985–1989 1990–1994 Total: all years

Count Percent Count Percent Count Percent

Site
Cutaneous 29883 91.7 47464 90.9 77347 91.2
Mucous membrane 393 1.2 681 1.3 1074 1.3
Ocular 1647 5.1 2875 5.5 4522 5.3
Unknown primary site 675 2.0 1218 2.3 1893 2.2

Total 32598 100 52238 100 84836 100

Mucous membrane site
Head and neck 212 53.9 383 56.3 595 55.4
Female genital 75 19.1 118 17.3 193 18.0
Anal/rectal 93 23.7 163 23.9 256 23.8
Urinary 13 3.3 17 2.5 30 2.8

Total mucous membrane 393 100 681 100 1074 100

Ocular site
Uveal 1365 82.9 2481 86.3 3846 85.0
Conjunctival 86 5.2 130 4.5 216 4.8
Other site 196 11.9 264 9.2 460 10.2

Total ocular 1647 100 2875 100 4522 100

Unknown primary site at presentation
Regional metastasis 287 42.5 527 43.3 814 43.0
Distant metastasis 388 57.5 691 56.7 1079 57.0

Total unknown primary site at presentation 675 100 1218 100 1893 100

summarized in Table 1C. Lentigo maligna lesions were cation for mucosal melanomas. However, when infor-
predominantly Stage 0 or I lesions, which comprised mation could be retrieved from the medical records,
88.6%. For superficial spreading melanomas, the most the lymph node status of mucosal melanomas was
prevalent stage was Stage I (68.1%) followed by Stage recorded. Among the 1074 mucosal melanomas for all
II lesions (19.7%). Nodular and acral lentiginous le- years, lymph node status was recorded in 28.4% of
sions were more aggressive according to their stage at cases (Table 2). This represented 305 cases, 40.3% of
presentation. For both nodular and acral lentiginous which had positive lymph nodes and 59.7% of which
histologies, the most common presentation was Stage had negative lymph nodes. A breakdown of lymph
II, representing 51.1% and 37.0% of these histologies, node status by site revealed that the incidence of
respectively. In addition, both nodular and acral len- positive lymph nodes for head and neck, female gen-
tiginous lesions presented with higher percentages of ital tract, anal/rectal, and urinary tract melanomas
Stage III and IV disease compared with the other his- was 26.6%, 23.0%, 61.0%, and 11.1%, respectively. The
tologies. presence of positive lymph nodes in mucosal melano-
Among the 4522 ocular melanomas for all years, mas had a significant adverse effect on survival.
3846 patients (85.0%) had uveal melanomas (Table
1A). Among these uveal melanomas, AJCC staging was Patient Characteristics
recorded in 2286 patients; the percentage of patients Patient characteristics were broken down for all years
with AJCC Stages I, II, III, and IV were 31.3%, 28.9%, with respect to cutaneous, mucosal, ocular, and un-
31.9%, and 7.4%, respectively. There were 216 patients known primary melanoma classifications in Table 3A.
with conjunctival melanomas. The remaining 460 pa- Compared with other histologic groups, there was a
tients with ocular melanomas were classified as “oth- definite trend toward a skewing of older patients di-
er” and comprised cornea, NOS; retina; lacrimal agnosed with mucosal melanomas with 49.0% of pa-
gland; orbit, NOS; overlapping lesion of the eye; and tients being age $70 years. As one would expect, there
eye, NOS. was a higher percentage of females (63.5%) than males
There is no recommended AJCC stage groupings diagnosed with mucosal melanomas due to lesions
for conjunctival melanomas, and no staging classifi- arising in the female genital tract. By contrast, there
1668 CANCER October 15, 1998 / Volume 83 / Number 8

TABLE 1B
Histologic Classifications and AJCC “Combined” Staging by Diagnostic Year for Cutaneous Melanoma

1985–1989 1990–1994 Total: all years

Count Percent Count Percent Count Percent

Histology
Lentigo maligna 2779 9.3 4933 10.4 7712 10.0
Superficial spreading 8430 28.2 12318 26.0 20748 26.8
Nodular 2850 9.5 3969 8.4 6819 8.8
Acral lentiginous 177 0.6 592 1.2 769 1.0
Other 15647 52.4 25652 54.0 41299 53.4

Total 29883 100 47464 100 77347 100

“Combined” stagea
Stage 0 1970 11.4 6420 16.4 8390 14.9
Stage I 8742 50.4 18113 46.3 26855 47.6
Stage II 4026 23.2 9041 23.1 13067 23.2
Stage III 1635 9.4 3461 8.9 5096 9.0
Stage IV 956 5.5 2062 5.3 3018 5.3

Subtotal known stage 17329 100 39097 100 56426 100

Unknown stage 12554 42.0b 8367 17.6b 20921 27.0b
Total 29883 47464 77347

AJCC: American Joint Committee on Cancer.

a
“Combined” stage represents pathologic stage, augmented by clinical stage in cases in which the pathologic stage was not available.
b
Percentage was based on total cases.

TABLE 1C
“Combined” Stage by Histology by Site for Cutaneous Melanoma, All Years

Stage

0 1 2 3 4 Total Cases

Cutaneous 14.9 47.6 23.2 9.0 5.3 100 56426

Lentigo maligna 51.9 36.7 8.2 2.5 0.7 100 5569
Superficial spreading 6.7 68.1 19.7 4.3 1.2 100 15572
Nodular 0.5 20.4 51.1 22.8 5.2 100 5106
Acral lentiginous 9.3 34.0 37.0 14.9 4.8 100 579
Other 14.8 43.8 22.7 10.3 8.4 100 29600

was a higher percentage of males (65.8%) than females with Stage III or IV disease compared with 14.6% of
diagnosed with unknown primary melanoma. With high income patients.
respect to race/ethnicity, there was a larger percent- An analysis of patient characteristics by histology
age of African-American and Hispanic individuals di- for cutaneous melanomas for all years is summarized
agnosed with mucosal melanomas compared with the in Table 3B. There were large differences between
percentage of African-American and Hispanic patients histologies with respect to age and race/ethnicity. A
diagnosed with cutaneous, ocular, or unknown pri- larger percentage of older patients presented with len-
mary melanomas. Approximately 8.8% of patients tigo maligna lesions compared with the other histol-
with mucosal melanomas were of African-American or ogies; 76.7% of the patients diagnosed with lentigo
Hispanic race/ethnicity compared with a substantially maligna were age $60 years compared with 33.5%,
lower percentage of melanomas (,3%) of the other 46.5%, and 61.8% for superficial spreading, nodular,
sites. Levels of income (i.e., low, middle, and high) and acral lentiginous histologies, respectively. With
between the various sites of melanoma were not sub- regard to race/ethnicity, African Americans and His-
stantially different. Levels of income were different for panics comprised 15.9% of all patients diagnosed with
early versus advanced stage within cutaneous mela- acral lentiginous melanomas whereas individuals of
noma, with 21.2% of low income patients presenting these racial/ethnic backgrounds comprised only 1.3%,
 NCDB Report on Melanoma/Chang et al. 1669

TABLE 2
Lymph Node Status by Diagnostic Year for Mucous Membrane Melanoma in which Lymph Nodes Were Examined

1985–1989 1990–1994 Total: all years

Count Percent Count Percent Count Percent

All cases
No positive lymph nodes 85 65.9 97 55.1 182 59.7
Positive lymph nodes 44 34.1 79 44.9 123 40.3

Total 129 100 176 100 305 100

Head and Neck
No positive lymph nodes 36 69.2 44 60.3 80 73.4
Positive lymph nodes 16 30.8 29 39.7 29 26.6

Total 52 100 73 100 109 100

Female genital tract
No positive lymph nodes 24 85.7 31 72.1 55 77.0
Positive lymph nodes 4 14.3 12 27.9 16 23.0

Total 28 100 43 100 71 100

Anal/rectal
No positive lymph nodes 19 44.2 20 35.1 39 39.0
Positive lymph nodes 24 55.8 37 64.9 61 61.0

Total 43 100 57 100 100 100

Urinary tract
No positive lymph nodes 6 100 2 66.7 8 88.9
Positive lymph nodes 0 0 1 33.3 1 11.1

Total 6 100 3 100 9 100

1.2%, and 1.9% of patients diagnosed with lentigo 1990 –1994). The treatment of ocular melanomas also
maligna, superficial spreading, or nodular melano- demonstrated a large rise across these years in the
mas, respectively. Levels of income were not substan- percentage treated with radiation therapy alone (from
tially different between these histologic classifications. 17.4% to 30.4%). There was a concomitant drop in the
percentage of ocular melanomas treated with surgery
Treatment alone (from 60.6% to 55.2%). Information regarding
The vast majority (91.5%) of cutaneous melanomas the use of biologic response modifier therapy was
were treated with surgery alone (Table 4). Surgery recorded for 16,987 of these patients with melanoma.
alone also was the primary modality, but to a lesser Within this group, 498 patients (2.9%) received a bio-
extent, for managing mucous membrane melanomas logic response modifier as part of their treatment.
(56.4%) and ocular melanomas (57.0%). A higher per-
centage of these noncutaneous melanoma patients Survival
also received radiation therapy, either alone (25.7% of Disease specific survival analyses, performed on pa-
ocular melanomas) or in combination with surgery tients diagnosed between 1985–1989, indicated that
(19.3% of mucous membrane melanomas). Approxi- 78.8% were alive 5 years after diagnosis (Table 5A).
mately 58% of patients with unknown primary mela- When stratified by anatomic site, cutaneous melano-
nomas were treated surgically with or without radia- mas had the highest 5-year survival rate (80.8%), fol-
tion therapy and/or chemotherapy for management of lowed by ocular melanomas (74.6%). The proportion
the metastasis. However, a substantial percentage of patients with mucous membrane melanoma who
(17.7%) received no treatment in the form of surgery, survived 5 years was much lower (25.0%). Patients
radiation therapy, or chemotherapy. Although not with unknown primary site of the melanoma also had
shown in Table 4, there were few changes across the a low survival rate, with only 29.1% alive at 5 years.
years in treatment for the various sites. However, sur- Mucosal melanomas were analyzed with respect
gery with radiation therapy has been used with in- to site and lymph node status. Patients with head and
creasing frequency for treating mucous membrane neck, female genital tract, and anal/rectal mucosal
melanomas (from 11.3% in 1985–1989 to 15.3% in melanomas had 5-year survival rates of 31.7%, 11.4%,
1670 CANCER October 15, 1998 / Volume 83 / Number 8

TABLE 3A
Patient Characteristics by Anatomic Site for Melanoma, All Years

Cutaneous Mucosal Ocular Unknown primary site

Count Percent Count Percent Count Percent Count Percent

Age (yrs)
, 30 5398 7.0 10 0.9 149 3.3 104 5.5
30–39 11036 14.3 35 3.3 322 7.1 253 13.3
40–49 13500 17.4 108 10.1 619 13.7 316 16.7
50–59 13202 17.1 149 13.9 839 18.5 368 19.4
60–69 15778 20.4 244 22.7 1210 26.8 416 22.0
70–79 12574 16.2 306 28.5 980 21.7 299 15.8
801 5795 7.5 220 20.5 400 8.8 136 7.2
Unknown 64 0.1 2 0.2 3 0.1 1 0.1

Total 77347 100 1074 100 4522 100 1893 100

Mean age (yrs) 55.3 67.0 60.4 56.0
Gender
Male 42013 54.3 391 36.4 2352 52.0 1245 65.8
Female 35132 45.4 682 63.5 2158 47.7 646 34.1
Other/unknown 202 0.3 1 0.1 12 0.3 2 0.1

Total 77347 100 1074 100 4522 100 1893 100

Race/ethnicity
White non-Hispanic 72500 93.7 918 85.5 4168 92.2 1794 94.8
African-American, non-Hispanic 590 0.8 72 6.7 37 0.8 34 1.8
Hispanic (any race) 911 1.2 42 3.9 86 1.9 19 1.0
Other/unknown 3346 4.3 42 3.9 231 5.1 46 2.4

Total 77347 100 1074 100 4522 100 1893 100

Incomea
Low (, $20,000) 5634 7.3 137 12.8 526 11.6 192 10.2
Middle ($20,000–46,999) 56922 73.6 768 71.5 3350 74.1 1365 72.1
High ($$47,000) 10852 14.0 118 11.0 359 7.9 194 10.2
Unknown 3939 5.1 51 4.7 287 6.4 142 7.5

Total 77347 100 1074 100 4522 100 1893 100

a
Income groupings based on the lowest (, $20,000) and highest ($$47,000) 10% of all cancer patients on the National Cancer Data Base data set.

and 19.8%, respectively. There were too few cases of mas, patients with AJCC Stages I, II, III, and IV had
urinary tract melanomas for survival analyses. The survival rates of 83.9%, 84.9%, 64.4%, and 59.3%, re-
survival rate for head and neck mucosal melanoma spectively (Fig. 1). All stage subgroups were signifi-
was significantly better (P , 0.05) than female genital cantly different from each other.
tract or anal/rectal mucosal melanomas. Combining Cutaneous melanoma survival rates were disag-
all sites, 96 patients were identified with known lymph gregated by “combined” stage when staging informa-
node status in cases diagnosed between 1985–1989. tion was available in 13,547 patients (Fig. 2). For Stage
Patients with positive lymph nodes had a poorer out- 0 melanomas (also known as melanomas in situ), the
come, demonstrated by a 5-year survival rate of 16.4% 5-year survival rate was 96.0%. For localized, invasive
compared with 38.7% for all patients with negative melanomas (representing Stages I and II), the survival
lymph nodes. There were too few cases within the rates were 92.5% and 74.8%, respectively. Lymph node
positive lymph node group to perform statistical anal- positive Stage III patients had a survival rate of 49.0%,
yses. whereas Stage IV patients with metastatic disease had
Patients with ocular melanomas had survival bro- a survival rate of only 17.9% (Fig. 2). All stage sub-
ken down between uveal and conjunctival sites. The groups were significantly different from each other.
overall, 5-year survival for uveal and conjunctival mel- A breakdown of cutaneous melanoma survival ac-
anomas was 83.5% and 75.7%, respectively, which was cording to histologic subtype is summarized in Table
not significantly different. The majority of ocular mel- 5B. The 5-year survival rates for lentigo maligna, su-
anomas were of uveal origin. Among uveal melano- perficial spreading, nodular, and acral lentiginous
 NCDB Report on Melanoma/Chang et al. 1671

TABLE 3B
Patient Characteristics by Histology for Cutaneous Melanoma, All Years

Lentigo maligna Superficial spreading Nodular Acral lentiginous

Count Percent Count Percent Count Percent Count Percent

Age (yrs)
, 30 44 0.6 1790 8.6 443 6.5 31 4.0
30–39 195 2.5 3885 18.7 908 13.3 56 7.3
40–49 550 7.1 4419 21.3 1114 16.3 91 11.8
50–59 1001 13.0 3687 17.8 1176 17.3 116 15.1
60–69 2163 28.0 3759 18.1 1398 20.5 196 25.5
70–79 2468 32.0 2340 11.3 1113 16.3 193 25.1
801 1286 16.7 860 4.1 659 9.7 86 11.2
Unknown 5 0.1 8 0.1 8 0.1 0 —

Total 7712 100 20748 100 6819 100 769 100

Mean age (yrs) 67.5 51.3 56.4 61.6
Gender
Male 4613 59.8 10451 50.4 4005 58.7 346 45.0
Female 3091 40.1 10231 49.3 2792 41.0 422 54.9
Other/unknown 8 0.1 66 0.3 22 0.3 1 0.1

Total 7712 100 20748 100 6819 100 769 100

Race/ethnicity
White non-Hispanic 7185 93.1 19586 94.4 6477 95.0 617 80.2
African-American, non-Hispanic 35 0.5 49 0.2 55 0.8 89 11.6
Hispanic (any race) 63 0.8 206 1.0 78 1.1 33 4.3
Other/unknown 429 5.6 907 4.4 209 3.1 30 3.9

Total 7712 100 20748 100 6819 100 769 100

Incomea
Low (, $20,000) 573 7.4 1361 6.5 577 8.5 101 13.2
Middle ($20,000–46,999) 5642 73.2 15378 74.1 5151 75.5 543 70.6
High ($$47,000) 1063 13.8 2876 13.9 776 11.4 84 10.9
Unknown 434 5.6 1133 5.5 315 4.6 41 5.3

Total 7712 100 20748 100 6819 100 769 100

a
Income groupings based on the lowest (, $20,000) and highest ($$47,000) 10% of all cancer patients on the National Cancer Data Base data set.

TABLE 4
Treatment by Anatomic Site for Melanoma (All Years)

Surg only RT only CH only Surg1 RT Surg1 CH RT1 CH All three None Unknown Total Cases

Site
Cutaneous 91.5 0.7 0.5 1.4 1.5 0.3 0.4 3.0 0.7 100 77347
Mucous membrane 56.4 8.5 1.6 19.3 2.0 2.0 2.8 6.1 1.3 100 1074
Ocular 57.0 25.7 0.3 8.8 1.5 0.2 0.2 4.6 1.7 100 4522
Unknown primary 39.4 8.2 8.4 11.0 5.5 4.5 2.6 17.7 2.7 100 1893

Surg: surgery; RT: radiation therapy; CH: chemotherapy.

melanoma were 93.2%, 91.6%, 64.6%, and 66.4%, re- subtypes were comprised of a higher percentage of
spectively. All histologic subgroups were significantly Stage II, III, and IV cases compared with the lentigo
different from each other except the nodular and acral maligna and superficial spreading histologic subtypes.
lentiginous subgroups. One potential factor contrib- When accounting for stage, patients with nodular mel-
uting to differences in survival between certain histo- anomas still fared significantly worse than individuals
logic subtypes most likely is due to the distribution of with superficial spreading melanomas for all stages of
stages within each group. As indicated in Table 1C, invasive disease (Table 5B). There were no significant
both the nodular and acral lentiginous melanoma differences in survival for patients with superficial
1672 CANCER October 15, 1998 / Volume 83 / Number 8

TABLE 5A
Five-Year, Disease Specific Survival for 1985–1989 Melanoma

Years after diagnosis

1 2 3 4 5 Cases P value

All cases 93.1% 87.8% 84.2% 81.1% 78.8% 25795

Site
Cutaneous 94.1 89.3 85.9 83.0 80.8 23696
Mucous membrane 74.5 49.3 38.4 30.2 25.0 306
Ocular 96.4 89.2 84.6 79.2 74.6 1275
Unknown primary 51.7 41.1 34.9 31.6 29.1 518
Mucous membrane sitea
Head and neck 82.0 54.4 46.5 37.4 31.7 163 0.0034b
Female genital tract 68.1 40.1 26.3 21.1 11.4 59
Anal/rectum 61.6 44.0 27.4 19.8 19.8 74
Known lymph node status for mucous membrane
Positive 61.9 33.6 25.7 16.4 16.4 33
Negative 84.9 55.6 44.0 38.7 38.7 63
Ocular site
Conjunctival 100 94.6 86.4 86.4 83.5 64 ,0.0001c
Uveal 97.3 90.5 86.6 80.6 75.7 1062
Other ocular 88.8 78.8 70.3 66.5 63.2 149

A pairwise comparison of “combined” Stages I–IV showed that each stage is significantly different from all other three stages, using the Wilcoxon (Gehan) statistic.
a
There were too few cases for the urinary tract for survival analyses.
b
At the P , 0.05 level, all site subgroups were significantly different from each other except female genital tract and anal/rectum.
c
At the P , 0.05 level, all site subgroups were significantly different from each other except conjunctival and uveal.

respect to age, gender, race/ethnicity, and income for

cutaneous melanoma patients (Table 5C). Patients
were classified as presenting with “early” (Stages I and
II) or “advanced” (Stages III and IV) disease. Patients
age $60 years did significantly worse than patients age
,60 years, with 5-year survival rates of 81.4% and
90.1%, respectively, for patients with early stage dis-
ease. For patients with advanced stage disease, the
5-year survival rate of patients age $60 years (32.0%)
was significantly less than the rate of patients age ,60
years (40.5%). Males fared significantly worse than
females, with the two genders having 5-year survival
rates of 83.5% versus 90.5% for early stage and 32.7%
versus 43.7% for advanced stage disease, respectively.
In regard to race/ethnicity, white individuals had sig-
nificantly better survival compared with nonwhite in-
dividuals for early stage melanoma (86.8% and 74.0%,
FIGURE 1. Five-year, disease specific survival by “combined” stage for 1985– respectively). For advanced melanoma, nonwhite pa-
1989 uveal melanoma. (Note:“combined” stage represents pathologic stage, aug- tients fared worse than white patients, with a 5-year
mented by clinical stage in cases in which the pathologic stage was not available) survival rate of 33.9% compared with 36.6%, respec-
tively, although the difference was not statistically sig-
nificant. In regard to income, there was a significant
spreading or lentigo maligna melanomas within trend toward improved survival in patients with early
Stages I or II. There were too few patients with acral melanomas as income increased. In early melanomas,
lentiginous melanoma to analyze for survival differ- the 5-year survival rates for patients with low income
ences. versus high income were 83.2% and 90.9%, respec-
Survival analyses by stage were performed with tively. For patients with advanced stage melanomas,
 NCDB Report on Melanoma/Chang et al. 1673

sentation. Patients with regional disease (i.e., lymph

node) had a 5-year survival rate of 46.3% compared
with a rate of only 15.8% for patients who presented
with distant (i.e., nonlymph node) sites of disease, a
statistically significant difference (P , 0.0001) (Table
5E). The survival of individuals with unknown primary
melanomas was analyzed further with respect to treat-
ment. Treatment was grouped into surgical resection
(with or without adjuvant treatment) or no treatment.
For all patients with unknown primary melanomas,
the 5-year survival rate for those patients undergoing
surgical resection (38.8%) was significantly better than
that for patients who received no treatment (19.6%).
This was broken down further into individuals with
regional (i.e., lymph node) or distant disease at pre-
sentation. For regional presentation, the 5-year sur-
vival rates for patients with surgical treatment versus
those with no treatment were 54.5% and 26.5%, re-
spectively (with the number of patients receiving no
FIGURE 2. Five-year, disease specific survival by “combined” stage for
treatment being too small for statistical comparisons).
1985–1989 cutaneous melanoma. (Note: “combined” stage represents patho-
For distant presentation, 5-year survival rates associ-
logic stage, augmented by clinical stage in cases in which the pathologic stage
ated with surgical treatment versus no treatment were
was not available)
20.3% and 15.9%, respectively, which represented a
statistically significant difference.

the 5-year survival rates for patients with low income DISCUSSION
versus high income were 30.0% and 45.4%, respec- Based on hospital registry data during a 10-year pe-
tively, but the difference was not statistically signifi- riod, this report documents the relative percentages of
cant. cutaneous and noncutaneous melanomas diagnosed
For cutaneous melanoma, age, gender, race, in- in .86,000 patients. To our knowledge, this represents
come, stage, and histology indicated significantly dif- one of the largest databases comprising patient char-
ferent survival rates within their stratifications. These acteristics, histology, staging, therapy, and survival for
six variables were entered into a Cox regression anal- melanomas. The last NCDB study of melanoma, re-
ysis as dichotomous variables (Table 5D). Age was ,60 ported in 1994, reviewed 20,165 cases of cutaneous
or $60 years, gender was male or female, race was melanoma.13 Prior to that, the American College of
white or minority individuals, income was individuals Surgeons performed a Patient Care Evaluation study
in the lower 50% or the upper 50%, stage was early or of cutaneous and noncutaneous melanoma compris-
advanced, and histology was nodular and acral len- ing 11,904 patients that was reported in 1992.14 In the
tiginous (with low 5-year survival rates of 64.6% and current study, cutaneous melanomas comprised
66.4%, respectively) or lentigo maligna and superficial 91.2% of the cases, followed by ocular (5.3%), un-
spreading (with high 5-year survival rates of 93.2% and known primary (2.2%), and mucosal (1.3%) melano-
91.6%, respectively). Race was the only variable that mas. The peak age at diagnosis for cutaneous, ocular,
was not significantly associated with survival. The and unknown primary site of melanomas was between
strongest predictors were stage (with advanced dis- ages 60 – 69 years. By contrast, for mucosal melano-
ease having a 2.08 relative risk value) and histology mas, the peak age of incidence was between ages
(with the classifications demonstrating lower survival 70 –79 years. There was a predominance of white in-
rates [nodular and acral lentiginous] having a 1.78 dividuals diagnosed with cutaneous, ocular, and un-
relative risk value). The relative risk values for the known primary site of melanomas, with African-
remaining variables were 1.33 for males, 1.20 for pa- American or Hispanic individuals comprising only
tients age $60 years, and 1.12 for patients in the lower 5.4% and 3.4% of cases, respectively.
50% income bracket. In discussing the results by anatomic site of ori-
The survival of patients determined to have me- gin, cutaneous melanoma patients represented the
tastases from unknown primary sites were evaluated majority of individuals in the data base, totaling 77,347
with respect to regional versus distant disease at pre- cases. Among the patients diagnosed between 1985–
1674 CANCER October 15, 1998 / Volume 83 / Number 8

TABLE 5B
Five-Year, Disease Specific Survival of Histology by “Combined” Stage for 1985–1989 Cutaneous Melanoma

Years after diagnosis

1 2 3 4 5 Cases P value

Histology for cutaneous cases

Lentigo maligna 99.1 97.4 96.2 94.8 93.2 2157 , 0.0001a
Superficial spreading 99.0 97.1 95.2 93.0 91.6 6761
Nodular 91.5 81.7 74.4 68.7 64.6 2267
Acral lentiginous 96.1 88.4 80.0 73.8 66.4 138
“Combined” stage by histology for cutaneous casesb
Stage 0c
Lentigo maligna 99.4 99.1 98.9 97.7 97.7 511 0.9355
Superficial spreading 100 100 100 98.8 96.0 229
Stage Id
Lentigo maligna 99.4 98.7 97.4 95.2 93.6 513 , 0.0001e
Superficial spreading 99.7 99.0 98.2 96.6 95.7 2772
Nodular 97.0 91.4 87.8 82.4 79.8 351
Stage IId
Lentigo maligna 100 98.1 95.0 90.4 85.1 122 , 0.0001e
Superficial spreading 99.2 95.0 90.8 85.1 81.3 753
Nodular 96.9 86.7 78.9 73.6 68.0 663
Stage IIIf
Superficial spreading 97.9 80.9 73.4 63.5 58.4 204 0.0040
Nodular 86.5 71.4 61.8 54.0 48.5 299
Stage IVf
Superficial spreading 63.3 58.5 47.6 44.7 44.7 60 0.0263
Nodular 52.1 33.6 22.2 17.2 13.6 87

a
At the P , 0.05 level, all histologic subgroups were significantly different from each other except nodular and acral lentiginous.
b
“Combined” stage represents pathologic stage, augmented by clinical stage in cases in which the pathologic stage not available.
c
There were too few cases of Stage 0 nodular and acral lentiginous to be reported.
d
There were too few cases of Stage I or Stage II acral lentiginous to be reported.
e
At the P , 0.05 level, all histologic subgroups were significantly different from each other except lentigo maligna and superficial spreading.
f
There were too few cases of Stage III or Stage IV lentigo maligna and acral lentiginous to be reported.

1989, the overall 5-year survival was 80.8%. As ex- nodular, and acral lentiginous) and survival. Both the
pected, survival was associated with stage. Stage 0 (in lentigo maligna and superficial spreading subgroups
situ melanoma) and Stage I disease were highly cur- presented with a higher percentage of earlier stage
able, with 5-year survival rates of 96.0% and 92.5%, disease (Stages 0 through II). By contrast, the nodular
respectively. Fortunately, 62.6% of patients diagnosed and acral lentiginous subgroups presented with a
with melanoma fell into these stage groups. Surgical higher proportion of advanced disease (Stages III and
excision was the major therapeutic modality, em- IV), underscoring the aggressive biology of these his-
ployed in 94.8% of patients alone or in concert with tologic subtypes. Even when accounting for stage, the
other modalities. Of the patients in this database who nodular histologic subtype behaved more aggressively
had information regarding the administration of bio- than the lentigo maligna and superficial spreading
logic response modifier therapy, only 2.9% were subgroups. These observations validate the continued
treated with this modality. Because of the recent find- use of these histologic descriptions in the reporting of
ings that interferon-a is an effective adjuvant therapy pathologic diagnoses of melanomas. There have been
after surgery for patients with Stage III disease, it is other histologic features that may account for the ob-
anticipated that there will be an increase in the use of served differences in this report, including ulceration,
this biologic response modifier in the future.15 More- vertical growth phase, angiolymphatic invasion, mi-
over, vaccine therapies have seen a resurgence of in- totic activity, presence of infiltrating lymphoid cells,
terest in the management of patients with melano- and evidence of regression.18 –20 However, these fea-
ma.16,17 tures were not recorded within the hospital cancer
Of interest was the association between histologic registries from which the NCDB is drawn, and there-
subtype (i.e., lentigo maligna, superficial spreading, fore could not be analyzed.
 NCDB Report on Melanoma/Chang et al. 1675

TABLE 5C
Five-Year, Disease Specific Survival by Patient Characteristics by Early versus Advanced “Combined” Stage for 1985–1989 Cutaneous Melanoma

Years after diagnosis

1 2 3 4 5 Cases P value

Age (yrs)
Early stage
,60 99.2 96.8 94.6 92.1 90.1 6230 , 0.0001
$60 97.9 93.0 89.0 84.4 81.4 3760
Advanced stage
,60 75.3 59.2 49.8 44.7 40.5 1131 , 0.0001
$60 66.6 49.1 41.5 36.0 32.0 941
Gender
Early stage
Male 98.4 94.2 90.4 86.5 83.5 5194 , 0.0001
Female 99.0 96.7 94.6 92.2 90.5 4799
Advanced stage
Male 69.9 52.8 42.8 36.6 32.7 1306 0.0010
Female 73.8 57.6 51.7 47.9 43.7 767
Race/ethnicity
Early stage
White 98.7 95.4 92.5 89.2 86.8 9420 0.0001
Nonwhite 97.3 89.4 81.4 75.2 74.0 158
Advanced stage
White 71.7 54.7 46.0 40.8 36.6 1945 0.3202
Nonwhite 63.0 48.8 45.4 35.9 33.9 77
Income
Early stage
Low 98.5 93.7 90.1 85.5 83.2 631 , 0.0001a
Middle 98.6 95.3 92.2 89.0 86.5 7335
High 98.8 96.7 9.6 92.5 90.9 1578
Advanced stage
Low 66.7 45.7 37.5 33.6 30.0 202 0.0914
Middle 71.7 54.8 45.9 40.1 35.9 1491
High 71.1 58.2 50.8 47.8 45.4 283

“Early” stage represents “combined” Stages I and II, and “advanced” stage represents “combined” Stages III and IV.
a
At the P , 0.05 level, all income subgroups were significantly different from each other.

Because of the NCDB’s large database of cuta- women have a higher percentage of extremity mel-
neous melanoma patients, the effects of different anomas, whereas males have a higher percentage of
patient characteristics on survival could be evalu- truncal melanomas.21 The axial location of mela-
ated. Age $60 years was found to be an adverse noma has been reported to be associated with a
prognostic factor for both early stage and advanced worse prognosis.22 In the NCDB the percentage of
stage disease. Analysis of age by increasing decades males and females with axial melanomas (i.e., head
also demonstrated the same trend of decreased sur- and neck or truncal sites) was 64.8% and 39.1%,
vival (data not shown). Deaths due to causes unre- respectively. The proportion of males with cutane-
lated to melanoma were censored and only disease ous melanomas of the extremities (i.e., arm/shoul-
specific survival data were calculated. Hence, co- der and leg/hip) was 39.2% compared with 55.4% for
morbid conditions associated with increasing age females. The gender differences identified in this
were unlikely to be reasons for decreased survival database also may be related to hormonal influ-
from cutaneous melanoma. ences on the prognosis of melanoma. With respect
Gender was found to be a significant prognostic to race/ethnicity, whites fared better than non-
factor, with males having worse survival than fe- whites (i.e., African Americans and Hispanics) for
males. This was true for both early stage and ad- early stages of melanoma. There was a trend toward
vanced stage patient groups. It is known that the poorer survival in nonwhite patients who presented
distribution of the anatomic origin of cutaneous with advanced melanoma compared with whites;
melanomas differs between males and females; however, the numbers were too small to reach sta-
1676 CANCER October 15, 1998 / Volume 83 / Number 8

TABLE 5D which were conjunctival lesions. Surgery has been the

Patient and Disease Characteristics Influencing Disease Specific mainstay of treatment over all years; however, there
Survival for 1985–1989 Cutaneous Melanoma
was a significant increase in the use of primary radi-
Confidence ation therapy as the sole treatment between 1985–
intervals 1989 and 1990 –1994 (from 17.0% to 30.4%).
Variables entered A total of 1074 mucosal melanomas was recorded
into model P value Relative risk Lower Upper in the database. The overall, 5-year survival rate of
those patients diagnosed between 1985–1989 was only
Stage (0–II vs. III–IV) , 0.0001 2.08 1.93 2.24
Histology (low vs. 25.0%. The most prevalent site of mucosal melanomas
high survival was the head and neck region (55.4%), followed by
rate)a , 0.0001 1.78 1.65 1.92 anal/rectal melanomas (23.8%). Surgery was the main
Age (yrs) (, 60 vs. component of treatment in 80.5% of cases. The com-
601) , 0.0001 1.20 1.12 1.28
bined approach of utilizing surgery plus radiation
Income (lower vs.
upper 50%) 0.0011 1.12 1.05 1.20 therapy for mucosal melanomas involved 19.3% of
Gender (male vs. patients compared with 1.4% for cutaneous melano-
female) , 0.0001 1.33 1.24 1.43 mas. Approximately one-third of patients with muco-
Race (white vs. sal melanomas had positive lymph nodes when lymph
African American) 0.9971 —b —b —b
node status was known. The presence of positive
These results were obtained using Cox regression analysis. lymph nodes was associated with a 5-year survival of
a
The cutaneous histologies with higher 5-year survival (lentigo maligna and superficial spreading, with 16.4% compared with 38.7% for patients with negative
93.2% and 91.6% rates, respectively) were grouped together and compared with the histologies with lymph nodes. These findings document the important
lower 5-year survival (nodular and acral lentiginous, with 64.6% and 66.4% rates, respectively). prognostic effect of lymph node status in mucosal
b
Not applicable because race was not significant.
melanomas, which should be incorporated in any for-
mal staging system established for this disease entity.
Patients with unknown primary melanomas fared
tistical significance. Also interesting was the differ- poorly. The database contained information that per-
ences in survival related to income levels. There was mitted identification of whether the patients pre-
a direct relationship between increasing income sented with regional or distant metastases. If one as-
level and improved survival in patients with early sumes that the predominant source of these unknown
stages of cutaneous melanoma. This trend also was primary melanomas arose from a cutaneous site that
observed for advanced stage melanoma; however, spontaneously regressed, it is interesting to note that
the numbers again were too small to reach statisti- the 5-year survival rate of patients with regional me-
cal significance. tastasis (46.3%) was similar to the rate of survival for
A multifactorial analysis was performed to deter- patients with Stage III cutaneous melanoma (49.0%).
mine which case-mix variables (AJCC stage, histologic In addition, patients with distant metastases from un-
subtype, age, gender, race, and income) were predic- known primary sites had a 5-year survival rate (15.8%)
tive of survival for cutaneous melanoma. The factors that was similar to the survival of Stage IV cutaneous
that proved to be independent predictive factors for melanoma (17.7%). Despite the poor prognosis of this
survival, in decreasing order, were: stage, histology, category of patients, surgical treatment did result in a
gender, age, and income. We are not aware of any significant 5-year benefit in this subgroup of patients
report documenting income as a prognostic factor for compared with the subgroup of patients given no
cutaneous melanoma. In cross-tabulations, higher in- treatment. Although there most likely is a selection
come levels (i.e., low, middle, and high) were associ- bias in favor of patients who received surgical therapy
ated with lower stages of disease at presentation (data versus those who received no treatment, several re-
not shown). Nevertheless, the multivariate analysis ports have observed that the surgical management of
demonstrated income to be an independent prognos- patients with lymph node disease from unknown pri-
tic factor in addition to stage, which indicates that the mary melanomas fared as well as patients with known
effects of income on survival are due to influences cutaneous primary sites.23 The data in this current
besides stage. These effects of income on melanoma report appear to substantiate the potential benefit of
survival are unknown and should be investigated fur- surgical resection of patients with isolated lymph
ther because it is a factor that potentially can be al- node or visceral disease from unknown primary mela-
tered. nomas.
For all years, there was a total of 4522 ocular In summary, patients with cutaneous melanomas
melanomas, 85.0% of which were uveal and 4.8% of fared much better than patients with noncutaneous
 NCDB Report on Melanoma/Chang et al. 1677

TABLE 5E
Five-Year, Disease Specific Survival by Presentation and by Treatment (for All Cases, Regional Presentation, and Distant Presentation) for 1985–
1989 Unknown Primary Melanoma

Years after diagnosis

1 2 3 4 5 Cases P value

Presentation status
Regional metastasis 72.8 62.4 54.9 50.1 46.3 225 , 0.0001
Distant metastasis 35.4 24.5 19.4 17.2 15.8 293
Treatment
All cases
Surgerya 64.9 54.4 46.2 41.8 38.6 306 , 0.0001
No treatment 36.9 28.0 24.2 20.9 19.6 125
Regional presentation
Surgerya 80.4 71.3 63.1 58.0 54.5 160 —b
No treatment 57.9 49.0 39.2 31.4 26.5 40
Distant presentation
Surgerya 47.5 35.0 26.8 23.2 20.3 146 0.0001
No treatment 27.4 18.7 17.4 15.9 15.9 85

a
Surgery with or without adjuvant radiation therapy.
b
The number of patients in the group receiving no treatment was too small for statistical comparisons.

melanomas after treatment. Fortunately, the majority Cutaneous melanoma. 3rd edition. Philadelphia: J.B. Lip-
of individuals with cutaneous melanomas present at pincott, 1998:195–206.
6. Steele GD, Jessup JM, Winchester DP, Menck HR, Murphy
an early stage when therapy is highly successful in
GP, editors. National Cancer Data Base: annual review of
achieving 5-year cure rates. The mainstay of therapy
patient care, 1995. Atlanta: American Cancer Society, 1995.
for patients with melanomas of all sites has been sur- 7. Jessup JM, Menck HR, Winchester DP, Hundahl SA, Murphy
gical excision, except for patients with ocular melano- GP. The National Cancer Data Base report on patterns of
mas, in whom a significant trend toward primary ra- hospital reporting. Cancer 1996;78:1829 –37.
diation therapy has been observed. Surgical resection 8. Commission on Cancer. Data acquisition manual. Chicago:
of isolated disease due to unknown primary site of the American College of Surgeons Commission on Cancer, 1990.
9. Beahrs OH, Henson DE, Hutter RVP, Myers MH, editors.
melanomas should be instituted when medically fea-
Manual for staging of cancer. 3rd edition. Philadelphia: J.B.
sible because a subgroup of those patients fare well.
Lippincott, 1988.
Hopefully, with a better understanding of the biologic 10. Beahrs OH, Henson DE, Hutter RVP, Kennedy BJ, editors.
and socioeconomic factors influencing melanoma Manual for staging of cancer. 4th edition. Philadelphia: J.B.
treatment, the survival of patients with melanoma will Lippincott, 1992.
improve. In this regard, adjuvant strategies utilizing 11. ICD-O: International classification of disease for oncology.
biologic response modifiers after surgical resection 2nd edition. Geneva: World Health Organization, 1990.
12. SPSS, Inc. SPSS user’s guide. Release 6.1. Chicago: SPSS,
appear promising.
Inc., 1992.
13. Urist MM, Karnell LH. The National Cancer Data Base:
report on melanoma. Cancer 1994;74:782– 8.
REFERENCES 14. Sutherland CM, Chmiel JS, Henson DE, Cunningham MP,
1. Wingo PA, Ries LAG, Rosenberg HM, Miller DS, Edwards BK.
Peace BB, Winchester DP. Patient characteristics, methods
Cancer incidence and mortality, 1973–1995. Cancer 1998;82:
of diagnosis, and treatment of melanoma in the United
1197–207.
States. Surg Gynecol Obstet 1992;175:129 –34.
2. Landis SH, Murray T, Bolden S, Wingo PA. Cancer statistics,
15. Kirkwood JM, Strawerman MH, Ernstoff MS, Smith TJ, Bor-
1998. CA Cancer J Clin 1998;48:6 –29.
3. Rigel DS. Malignant melanoma: incidence issues and their den EC, Blum RH. Interferon alpha-2b adjuvant therapy of
effect on diagnosis and treatment in the 1990s. Mayo Clin high-risk resected cutaneous melanoma: the Eastern Coop-
Proc 1997;72(4):367–71. erative Oncology Group trial EST 1684. J Clin Oncol 1996;
4. Sahel JA, Skeves RA, Albert DM. Intraocular melanoma. In: 14:7–17.
DeVita VT, Hellman S, Rosenberg SA, editors. Cancer: prin- 16. Livingston PO, Wong GYC, Adluri S, Tao Y, Padavan M,
ciples and practice of oncology. 5th edition. Philadelphia: Parente R, et al. Improved survival in stage III melanoma
J.B. Lippincott, 1997:1995–2011. patients with GM2 antibodies: a randomized trial of adju-
5. Ross MI, Stern SJ. Mucosal melanomas. In: Balch CM, vant vaccination with GM2 ganglioside. J Clin Oncol 1994;
Houghton AN, Milton GW, Sober AJ, Soong S-J, editors. 12:1306 – 44.
1678 CANCER October 15, 1998 / Volume 83 / Number 8

17. Rosenberg SA, Yang JC, Schwartzentruber DJ, Hwu P, noma based on tumor progression. J Natl Cancer Inst 1989;
Marincola FM, Topalian SL, et al. Immunologic and thera- 81:1893–904.
peutic evaluation of a synthetic peptide vaccine for the 21. Balch CM, Karakousis C, Mettlin C, Natarajan N, Donegan
treatment of patients with metastatic melanoma. Nat Med WL, Smart CR, et al. Management of cutaneous mela-
1998;4:321–7. noma in the United States. Surg Gynecol Obstet 1984;158:
18. NIH Consensus Development Panel on Early Melanoma. 311– 8.
Diagnosis and treatment of early melanoma. JAMA 1992; 22. Stadelmann WK, Rapaport DP, Soong S-J, Reintgen DS,
268:1314 –9. Buzaid AC, Balch CM. Prognostic clinical and pathological
19. Thorn M, Ponten F, Bergstrom R, Sparen P, Adami H-O. features. In: Balch CM, Houghton AN, Milton GW, Sober AJ,
Clinical and histopathologic predictors of survival in pa- Soong S-J, editors. Cutaneous melanoma. Philadelphia: J.B.
tients with malignant melanoma: a population-based study Lippincott, 1998:11–35.
in Sweden. J Natl Cancer Inst 1994;86:761–9. 23. Wong JH, Cagle LA, Morton DL. Surgical treatment of lymph
20. Clark WH, Elder DE, Guerry D, Braitman LE, Trock BJ, nodes with metastatic melanoma from unknown primary
Schultz D, et al. Model predicting survival in stage I mela- site. Arch Surg 1987;122:1380 –3.