Ferrone CR, Ben Porat L, Panageas KS, Berwick M, Halpern AC, Patel A, Coit DG. Clinicopathological Features of and Risk Factors for Multiple Primary
Melanomas. JAMA. 2005;294(13):1647-1654. doi:10.1001/jama.294.13.1647
Author Affiliations: Departments of Surgery
(Drs Ferrone and Coit and Ms Patel), Epidemiology and Biostatistics (Mss Porat
and Berwick and Dr Panageas), and Dermatology (Dr Halpern), Memorial Sloan-Kettering
Cancer Center, New York, NY.
Context The incidence of multiple primary melanomas ranges from 1.3% to 8.0%
in large retrospective reviews; however, the impact of certain risk factors
is not understood.
Objectives To determine the incidence of multiple primary melanomas (MPM) from
a prospective, single-institution, multidisciplinary database, and to describe
the clinical and pathological characteristics and risk factors specific to
Design and Setting Review of a prospectively maintained database at Memorial Sloan-Kettering
Cancer Center in New York, NY.
Patients A total of 4484 patients diagnosed with a first primary melanoma between
January 1, 1996, and December 31, 2002.
Main Outcome Measures Incidence of and risk factors for MPM.
Results Three hundred eighty-five patients (8.6%) had 2 or more primary melanomas,
with an average of 2.3 melanomas per MPM patient. Seventy-eight percent had
2 primary melanomas. For 74% of patients, the initial melanoma was the thickest
tumor. Fifty-nine percent presented with their second primary tumor within
1 year. Twenty-one percent of MPM patients had a positive family history of
melanoma compared with only 12% of patients with a single primary melanoma
(SPM) (P<.001). Thirty-eight percent of MPM patients
had dysplastic nevi compared with 18% of SPM patients (P<.001). The estimated cumulative 5-year risk of a second primary
tumor for the entire cohort was 11.4%, with almost half of that risk occurring
within the first year. For patients with a positive family history or dysplastic
nevi, the estimated 5-year risk of MPM was significantly higher at 19.1% and
23.7%, respectively. The most striking increase in incidence for the MPM population
was seen for development of a third primary melanoma from the time of second
primary melanoma, which was 15.6% at 1 year and 30.9% at 5 years.
Conclusions The incidence of MPM is increased in patients with a positive family
history and/or dysplastic nevi. These patients should undergo intensive dermatologic
screening and should consider genetic testing.
The phenomenon of multiple primary tumors occurring within a single
organ or organ system has been well documented.1 Whether
this is a function of an increased genetic susceptibility of the individual
patient (eg, in familial polyposis and colorectal cancer), consistent exposure
to a common exogenous promoter of malignancy (eg, cigarette smoke and lung
cancer), or a combination of these 2 factors varies from one patient to the
next and from one organ system to the next. Malignant melanoma epitomizes
much of what is known about multiple primary tumors.
In 2005, there will be an estimated 62 000 new cases of invasive
melanoma and an estimated 7600 deaths due to melanoma in the United States.2 Melanoma is the fifth leading cancer in men and the
sixth leading cancer in women in the United States.2 The
incidence of melanoma continues to rise at about 3.0% per year in the United
States, with an estimated lifetime risk for an individual of 1.4%.1,2 This increasing incidence puts a larger
portion of the population at risk not only for 1 primary melanoma but also
for subsequent primary melanomas.3 Among patients
with a history of a primary melanoma, retrospective reviews report the incidence
of a second primary tumor to range from 0.2% to 8.6%3- 13 (Table 1). The Surveillance, Epidemiology, and
End Results (SEER) database from 1996-2001 includes 22 688 patients with
a single melanoma and 465 patients with more than 1 melanoma, for an incidence
Since Pack et al5 first described the
phenomenon of multiple primary melanomas (MPM) in 1952, patients with this
disease have not been well characterized. The clinical manifestations of MPM
are broad. While 63% to 88% of patients with MPM are reported to have 2 primary
tumors,3,5- 10 a
patient with as many as 48 separate primary melanomas has been reported.6
Several risk factors associated with the development of MPM have been
identified. These include a positive family history of MPM and a personal
history of dysplastic nevi (DN), atypical moles that are risk markers but
nonobligate precursors of melanoma.12- 16 Among
patients with MPM, 18% to 38% are reported to have a positive family history
of melanoma4,11 and 38% to 46%
are reported to have a history of dysplastic nevi.3,4 However,
few longitudinal cohort-defined databases have prospectively recorded known
risk factors for all patients with melanoma to assess the impact of these
risk factors on the development of MPM.
The main difficulty in estimating the incidence of MPM is that it requires
careful follow-up of large populations of melanoma patients over long periods
of time, with specific effort to capture antecedent, synchronous, or subsequent
This is the first study, to our knowledge, to use a prospective, multidisciplinary,
single-institution database to identify the characteristics of patients at
risk of developing MPM. This study defines the estimated incidence of MPM
among a cohort of consecutive patients treated for melanoma at a tertiary
cancer referral center. Specific patient characteristics, prospectively collected,
included family history and presence of DN. Tumor characteristics were evaluated
with specific attention to the number of melanomas diagnosed, the time course
over which melanomas were diagnosed, the timing of the thickest melanomas
(initial vs subsequent), the location of subsequent melanomas, and the prevalence
of melanoma in situ. In addition, these characteristics among MPM patients
were compared with those of patients with single primary melanoma (SPM) in
the same prospective database.
A prospective, computerized, multidisciplinary melanoma database at
Memorial Sloan-Kettering Cancer Center (MSKCC), New York, NY, was used to
identify a constructed cohort of 4484 patients diagnosed as having an initial
primary melanoma between January 1, 1996, and December 31, 2002, after approval
from the institution’s internal review board. Patients diagnosed as
having in situ and/or invasive melanomas, who had given either written or
oral consent to be entered into the database, were included in the study cohort.
Patients with MPM were defined as those with 2 or more primary melanomas based
on review of pathology reports from MSKCC and outside institutions. All primary
tumors had clear epidermal involvement documented on pathology reports. Recurrences,
in transit, and distant metastases were excluded. Patient characteristics
recorded included age at diagnosis of first melanoma, sex, family history,
and presence of DN. Positive family history was defined as the history of
a first-degree relative diagnosed as having melanoma. The majority of DN were
diagnosed clinically by an MSKCC dermatologist. If a patient had a history
of multiple nevi excisions with a histological diagnosis of DN, the patient
was classified as having DN even if there were no remaining DN on clinical
examination. Dysplastic nevi were defined pathologically as the presence of
a single clinically unequivocal (>5 mm with color variegation and fuzzy border)
or histologically proven dysplastic nevus.14
For each primary melanoma diagnosed, the date of diagnosis, location,
thickness, Clark level, and ulceration status were recorded. Melanomas were
classified as being located in the head/neck, trunk (including chest, back,
abdomen, and inguinal regions), extremity (including shoulders, arms, and
legs), or mucosa (including anorectal and vulvar). Ocular melanomas were recorded
as a separate site. Synchronous melanomas were defined as those diagnosed
within 30 days of each other.
The Fisher exact test was used to compare the single vs multiple primary
groups with regard to sex, family history of melanoma, and presence of DN.
For the continuous variables, age and thickness, the single and multiple primary
groups were compared using a 2-sided t test. The
incidence of second primary melanoma over time was calculated using a competing
risk analysis accounting for death as a competing risk.17 Time
to second melanoma or death was calculated from the date of first melanoma
to the date of second melanoma, death, or last follow-up. The age-adjusted
standardized incidence ratio (SIR) for second primary melanoma was estimated
as the number of patients diagnosed as having MPM (observed) divided by the
number of diagnoses expected based on person-years of follow-up and incidence
rates for each age group from SEER cancer registries. Using SEERStat, version
5.3.1 (National Cancer Institute, Bethesda, Md), we estimated the SIR for
second primary melanoma in the SEER population. P<.05
was considered statistically significant.
Three hundred eighty-five of 4484 patients were identified as having
MPM, for an overall incidence of 8.6%. The 4099 SPM patients were followed
up for a median of 2.2 years (range, 0-8.8 years). An actuarial curve was
constructed for all 4484 patients with the event of interest being the development
of a second primary melanoma. The incidence curve of death without a second
primary tumor was also included (Figure,
A). The incidence of death without a second primary tumor is higher than the
incidence of developing a second primary tumor, which is one of the reasons
more patients do not develop a second primary melanoma. For the entire cohort
of patients, the cumulative estimated 5-year risk of a second primary tumor
was 11.4% (95% confidence interval, 10.2%-12.6%), with more than half (5.5%)
of that risk occurring within the first year. Once a patient had developed
a second primary melanoma, the cumulative estimated 5-year incidence of a
third primary melanoma from the date of the second one was even higher at
30.9%, with half of that risk (15.6%) occurring within the first year.
Sex and Age at Diagnosis. The majority of patients
both with MPM and with SPM were male (Table 2). The mean age at initial diagnosis for all patients was 55 years
(median, 56 years; range, 1-99 years), with a significant difference (P = .003) between those with SPM (mean, 55 years;
median, 56 years; range, 1-99 years) and those with MPM (mean, 58 years; median,
60 years; range, 11-90 years). In both the MPM and SPM cohorts, male patients
were significantly older than female patients at the time of diagnosis of
their first melanoma. Despite developing melanoma at an earlier age, women
had a longer time interval for development of a second primary tumor than
men. This could be because younger patients (<60 years) have a lower incidence
of second primary melanoma than older patients (P<.001).
The estimated 5-year incidence of second primary melanoma was 9.8% for young
patients (<60 years) vs 13.3% for older patients. The mean age at diagnosis
of the first primary tumor for men with MPM was 61 years (range, 11-89 years)
compared with 52 years for women with MPM (range, 20-90 years) (P<.001). These age characteristics are similar to those of patients
in our database with SPM, where the mean age at diagnosis for men was 57 years
(range, 7-97 years) compared with 53 years (range, 1-99 years) for women (P<.001).
Family History and DN. Patients with MPM were
significantly more likely to have a positive family history of melanoma, a
history of DN, or both than patients with SPM (Table 2). A positive family history, the presence or absence of
melanoma in a first-degree relative, was determined in 4459 patients. The
overall prevalence of a positive family history was 578/4459 (13%). Of the
385 MPM patients, 81 (21%) had a positive family history compared with 497
(12%) of 4075 SPM patients. The incidence of second primary melanoma was significantly
higher (P<.001) among those with a positive family
history, with rates at 1 and 5 years of 8.3% and 19.1% compared with 5.1%
and 10.3% for those with a negative family history (Figure, B and Table 3).
Presence or history of DN was determined in 3380 patients, with an overall
prevalence of DN of 788/3380 (23%). Of 375 MPM patients, 145 (39%) had a history
of DN compared with 643 (18%) of 3505 SPM patients. The incidence of MPM was
significantly higher among those with DN, with rates at 1 and 5 years of 11.1%
and 23.6% compared with 4.8% and 9.7% for patients without DN (P<.001) (Figure, C and Table 3).
Patients with both a positive family history and DN did not have a significantly
higher incidence of a second primary tumor compared with patients who had
either a positive family history or DN, most likely because of the strong
effect of DN. The lowest incidence of MPM was seen in patients with neither
a positive family history nor DN, with 1- and 5-year incidences of 4.3% and
9.0%, respectively (Figure, D and Table 3).
Number of Primary Melanomas. A total of 885
primary melanomas were identified in 385 MPM patients (mean, 2.3 melanomas
per patient). Of these patients, 78% had 2 primary tumors, 15% had 3 primary
tumors, 5% had 4, and 2% had 5 or more. The number of primary tumors ranged
from 2 to 7.
Location of MPM. Location of the melanomas
for the SPM and MPM patients is presented in Table 4 and Table 5. The distribution
of initial melanoma sites was similar when comparing MPM and SPM. Additionally,
for MPM patients the distribution of sites of subsequent melanomas was similar
to that of initial melanomas. The initial and second melanomas occurred on
the same body site in 49% of patients (190/385). Of the 61 patients who had
an initial melanoma in the head and neck region, 20 (33%) had a second primary
tumor at this anatomical site. Of the 169 patients who had an initial primary
tumor on the trunk, 81 (48%) also had a second melanoma on the trunk. The
highest body site concordance was seen among the 148 patients who had an initial
primary melanoma on the extremities; 89 (60%) of these patients had a second
primary melanoma on the extremities.
Timing of MPM. Synchronous melanomas occurred
in 139 MPM patients (36%) at some point in their disease course. Most of these
patients initially presented with 2 (n = 105) or 3 (n = 16)
primary melanomas, and the rest (n = 18) developed synchronous primary
melanomas subsequent to the first melanoma diagnosis. Twenty-two of the 121
patients who presented with synchronous melanomas had at least 1 subsequent
Thickness of MPM. Of the 385 MPM patients,
adequate information on the thickness of all melanomas was known for 352 patients.
Of these 352 patients, 60 had melanomas of equal thickness (52 with melanoma
in situ and 8 with invasive melanoma) and 93 patients had synchronous melanomas
of unequal thickness. For the remaining 199 patients with nonsynchronous MPM
of unequal thickness, the thickest melanoma was the initial melanoma in 147
patients (74%) and was the subsequent melanoma in 52 patients (26%). Excluding
the 139 patients with synchronous melanomas, the mean and median thickness
of the initial melanoma were 1.2 mm and 0.5 mm, compared with a mean and median
thickness for subsequent melanomas of 0.4 mm and 0 mm, respectively (range,
0-8 mm; P<.001).
The initial melanoma was also significantly thinner in MPM patients
compared with SPM patients. The mean and median thicknesses of the initial
primary for MPM patients (n = 224) were 1.2 mm and 0.5 mm (range,
0-17 mm) compared with mean and median thicknesses of 2.0 mm and 0.9 mm (range,
0-52 mm) for primary tumors in SPM patients (n = 3312) (P<.001).
Melanoma In Situ. Among MPM patients, 61% (234/385)
had at least 1 melanoma in situ (MIS) at some point in their clinical course.
Of these 234 patients, 52 had exclusively MIS, 175 had MIS and at least 1
invasive melanoma of known thickness, and 7 patients had MIS and a melanoma
of unknown thickness. Melanoma in situ was more prevalent among subsequent
primary melanomas. Among patients with known thickness, 21% of 385 initial
melanomas were MIS, 50% of 385 second melanomas were MIS, 55% of 83 third
melanomas were MIS, and 70% of 23 fourth melanomas were MIS. In contrast,
17% of 3312 SPM patients with known thickness had MIS.
Multiple primary melanomas were identified in 8.6% of a constructed
cohort of patients from our prospective, multidisciplinary, single-institution
database, which is consistent with other large retrospective studies, in which
the incidence ranged from 1.3% to 8.0%5,6 (Table 1). The overall number of melanomas and
number of synchronous melanomas diagnosed are consistent with other large
studies. Similar to the 48% reported in the study by Giles et al,11 49% of our cohort had their second primary tumor
on the same body site as their first primary tumor. However, Giles et al reported
the highest site concordance, 58%, in the head and neck region, whereas the
highest site concordance in our study was 60% in the extremities.
Analysis of the time interval between the first and second melanomas
demonstrated that 59% of MPM patients have their second primary within the
first year, slightly higher than reports by Slingluff et al6 and
Kang et al,8 who documented 51% of second primary
tumors within the first year.
Based on absolute thickness, the thickest melanoma was the first melanoma
for the majority of MPM patients. In the MPM cohort, the percentage of melanoma
in situ also increased with each subsequent primary melanoma, further confirming
the trend toward thinner subsequent melanomas. Brobeil et al7 made
a similar observation. This trend toward thinner melanomas is likely due to
improved surveillance and early detection, but less aggressive disease biology
in the MPM patients may also play a role, since the initial melanomas in MPM
patients were significantly thinner than the melanomas in the SPM cohort.
This may also explain the difference in disease-specific death between the
MPM and SPM cohorts. Of the MPM patients, 5.6% (29/385) died from melanoma
compared with 15.8% of SPM patients (649/4099).
Considerable debate surrounds the clinical significance of the diagnosis
of MIS. Blackwood et al,4 Slingluff et al,6 Kang et al,8 and Johnson
et al10 all included MIS in their analyses.
However, the 37 700 new annual cases of MIS are not included in the yearly
incidence of melanoma cases reported.2 To address
the concern surrounding the true biology of MIS and its classification as
a true melanoma or a precursor lesion, we performed another analysis excluding
patients with MIS. Interestingly, when patients who had only MIS or only 1
invasive melanoma and MIS were excluded from the analysis, the incidence of
invasive MPM dropped to 5.7% (165/2905); however, the demographics of the
subset analyses remains constant. Of the 165 patients with at least 2 invasive
melanomas, 68% were male, 20% had a positive family history, and 38% had DN.
Although the subset of patients with 2 invasive melanomas was smaller, the
trends remained consistent with the entire cohort of MPM patients.
Because the entity of DN is relatively recent, many of the older retrospective
analyses were not able to determine its prevalence. In our study, MPM patients
with DN presented at a significantly younger age than MPM patients without
DN. Despite presenting at a younger age, they did not appear to have a greater
number of primary tumors per patient, with 2.4 melanomas per patient. The
thickest melanomas in patients with SPM and MPM and with DN were significantly
thinner than in patients without DN. This may be the result of better surveillance
or less aggressive disease biology, despite the predisposition to multiple
The estimated cumulative 5-year risk of a second primary melanoma in
our 4484 patients is 11.4%, which is somewhat higher than reported in the
other studies (Table 1). This difference
may be due to the prospective collection of data or to the referral pattern
of a tertiary cancer center such as MSKCC. To assess the potential impact
of lead time bias, we analyzed the subset of patients whose diagnoses were
made at MSKCC. Since the resulting prevalence and incidence results were similar
to the entire cohort, we believed that the effect of selection bias was minimal.
Of the 11.4%, roughly half of the risk was accrued within the first
year, in large part because of the frequency of MPM patients presenting with
more than 1 primary melanoma. After the first year, the remaining 6.0% risk
accrued over years 2 through 5 at a rate somewhat higher than the yearly risk
calculated by Slingluff et al6 of 0.38% per
year. One melanoma put patients at risk of a second primary melanoma, and
a second primary melanoma put patients at even higher risk of a third primary
melanoma. After a second primary melanoma, the incidence of a third primary
melanoma almost tripled. The 1- and 5-year incidences of a third primary melanoma
from the date of the second primary melanoma were 15.6% and 30.9%, respectively.
When comparing our cohort with the SEER database, our SIR for a second
primary tumor was significantly higher. The SEER database for 1996-2001 included
22 688 patients with a primary melanoma, of whom 465 had second primary
tumors over 55 932 person-years at risk, for an SIR of 18.58. For our
cohort of 4484 patients with 385 MPM over 10 957 person-years at risk,
the SIR was 122. If we exclude patients with synchronous melanomas from the
SEER cohort as well as from our own cohort, the number of patients with MPM
drops to 310 with an SIR of 12.8 in the SEER database and 246 with an SIR
of 78.7 in the MSKCC data set. The significant difference in the incidence
and SIR between the SEER data and our cohort can be partially attributed to
the difference between a population-based cohort compared with an institutional
cohort. The SEER database comprises 10 statewide registries, whereas the MSKCC
cohort includes patients from a single tertiary cancer center. Also contributing
to the difference in SIR between the 2 data sets is the prospective nature
of the MSKCC database and the arduous task of reporting MPM.
As expected, the risk of subsequent primary melanoma is even higher
for patients with a positive family history, DN, or both, at 19.0%, 23.6%,
and 29.9% at 5 years from initial diagnosis of melanoma, respectively. Interestingly,
the risk for patients with no family history or DN is still roughly 1.2% per
year after the first year. However, unlike in the other patient subsets, the
risk begins to level off for these patients, with a risk of 8.2% at 4 years
and 9% at 5 years. This is in contrast with patients with a positive family
history and DN, whose risk increases 1.5% and 1.1% in years 4 and 5, respectively.
The substantially higher rate of positive family history and DN in MPM
patients offers a unique opportunity to further characterize melanoma risk
at a genetic level. Mutations in the tumor suppressor gene CDKN2A and the proto-oncogene CDK4 have been
identified in patients with a positive family history and MPM; however, the
clinical relevance and impact of these mutations is not clear.15,16 The
frequency of a CDKN2A mutation is less than 2% in
all melanoma patients.17 This rises to 8% to
12% in patients with sporadic MPM and to an astonishing 47% in patients with
MPM and a positive family history.18,19 By
further studying this high-risk cohort with an increased frequency of mutations,
better understanding of the genotype, phenotype, penetrance, and clinical
variability can be gained.
Dysplastic nevi are estimated to be present in 5% to 10% of the general
population.20,21 In our cohort
of SPM patients, the rate was 18%, which may be attributed to MSKCC’s
being a tertiary care center that attracts high-risk patients. Similar to
other reported series, the rate of DN in our cohort of MPM patients was 39%.
Patients with DN are predisposed not only to SPM but also to MPM. Further
genetic characterization of these patients could help to identify some of
the underlying molecular mechanisms specific to melanoma susceptibility.
All melanoma patients are counseled to undergo not only lifelong annual
dermatologic follow-up for the detection of additional primary melanomas but
also to perform self-examinations and behavior modifications. In the higher-risk
cohorts with MPM, positive family history, or DN it may be prudent for patients
to be seen more often than annually. For patients with multiple atypical nevi,
a photographically assisted follow-up may also be indicated.
Patients diagnosed as having melanoma are at increased risk of developing
additional primary melanomas. In our patient cohort, the incidence of MPM
was 8.6%, with a 5-year estimated cumulative risk of a second primary melanoma
of 11.4%. Patients with a positive family history or a history of DN are at
significantly greater risk of developing MPM and should be enrolled in more
intensive dermatologic surveillance programs. This high-risk subset of patients
should also be further characterized genetically to further elucidate the
biology and etiology of melanoma.
Corresponding Author: Daniel G. Coit, MD,
Memorial Sloan-Kettering Cancer Center, 1275 York Ave, New York, NY 10021
Author Contributions: Drs Ferrone and Coit
had full access to all of the data in the study and take responsibility for
the integrity of the data and the accuracy of the data analysis.
Study concept and design: Ferrone, Coit.
Acquisition of data: Ferrone, Panageas, Berwick,
Analysis and interpretation of data: Ferrone,
Ben-Porat, Panageas, Berwick, Halpern, Patel, Coit.
Drafting of the manuscript: Ferrone, Ben-Porat,
Critical revision of the manuscript for important
intellectual content: Ben-Porat, Panageas, Berwick, Halpern, Patel,
Statistical analysis: Ben-Porat, Panageas,
Administrative, technical, or material support:
Study supervision: Coit.
Financial Disclosures: None reported.
Acknowledgment: We thank Javier Betancourt,
New York College of Osteopathic Medicine, and Sue Clinco, Memorial Sloan-Kettering
Cancer Center, for their contributions in maintaining the melanoma database.