Survivors of cutaneous melanoma have been found to have an increased risk of a second primary cutaneous melanoma diagnosis. Although this general epidemiologic risk has been identified in several nationwide studies, most estimates of 5- and 10-year incidences of second melanomas are from older cohorts and/or single-institutions and may not be generalizable.1,2,3,4,5,6,7,8,9,10,11,12,13,14 Furthermore, many population-based studies report a standardized incidence ratio (SIR), which communicates the ratio of observed to expected incidence. Although the SIR is a useful epidemiologic statistical tool, it is less helpful for clinicians trying to discuss second melanoma risk with individual patients.1,2,3,5

Studies that report cumulative incidence have identified a 5-year incidence of second primary cutaneous melanomas in melanoma survivors ranging widely from 2 % in an older Surveillance, Epidemiology, and End Results (SEER) study7 to 12% in a more contemporary single-center study.8,9,10,11,15,16 The broad incidence range reported for second melanomas speaks of the need for updated estimates to better inform patients and clinicians.

Valid estimates of the incidence of a second primary cutaneous melanoma can assist clinicians making recommendations for ongoing screening of melanoma survivors. We aimed to determine the 5- and 10-year incidences of second primary cutaneous melanomas among survivors of cutaneous melanoma in a national, population-based sample.

Methods

Data Source

Data from 18 registries of the SEER Program were used to identify cases of non-metastatic, first primary cutaneous malignant melanoma diagnosed between 1998 and 2012. The last available follow-up date in the SEER data set was 31 December 2017. To identify cases of malignant melanoma, SEER’s combined ICD-O-3 site/histology codes for melanoma of the skin (site codes C442 to C449, histology codes 8720 to 8774) were used.

Study Inclusion and Exclusion Criteria

All cases of first primary cutaneous malignant melanoma diagnosed between 1998 and 2012 were included (n = 210,444). Figure 1 illustrates the inclusion/exclusion flowchart. Cases with a diagnosis before 1998 were excluded based on limited availability of precise and consistent tumor characteristics. We also excluded patients with a diagnosis after 2012 because these patients did not have 5 years of follow-up evaluation within the available data (last follow-up date 31 December 2017). In addition, we excluded patients lost to follow-up evaluation within 3 months after the initial diagnosis (n = 1380). We also excluded those with a first melanoma diagnosed before the age of 18 years (n = 976); those who initially presented with metastatic melanoma, an unknown disease stage, or a diagnosis at autopsy (n = 16,193); those with a missing diagnosis date (n = 75); and those who did not get surgery as part of treatment (“unknown” surgery cases and those with local tumor destruction but no pathologic examination were excluded) (n = 37,266).

Fig. 1
figure 1

Flowchart of study inclusion and exclusion.

Full size image

Data Collection

The primary outcome of interest was a second primary melanoma diagnosis, defined as the first melanoma diagnosed at least 3 months after the initial melanoma diagnosis. We defined melanomas diagnosed within 3 months after the initial melanoma diagnosis as synchronous melanomas. Although the precise period is somewhat arbitrary, this definition is consistent with other literature.17,18 By this definition, 1743 patients in our cohort had synchronous melanomas diagnosed at the initial diagnosis.

We analyzed the incidence of second primary melanomas for this cohort separately because we anticipated an especially elevated risk of additional melanomas in this cohort.

Predictor variables included patient demographics (age at diagnosis, sex, race, ethnicity, insurance status), tumor characteristics (site, histology, disease stage at diagnosis, Breslow depth, ulceration status), and treatment characteristics (general surgical treatment, radiation, chemotherapy).

Statistical Analysis

Descriptive statistics were used to characterize the cohort, with separate analysis of individuals who had synchronous melanomas. Overall 5- and 10-year incidences of second primary cutaneous malignant melanoma were estimated using Kaplan-Meier survival analysis. For subgroups of patients, 5- and 10-year incidences were also estimated based on patient and tumor characteristics at the time of the first primary diagnosis, including age, sex, and extent of disease.

Multivariable Cox regression was used to evaluate risk factors for the development of a second primary melanoma within 5 years after the initial melanoma diagnosis. Histologic subtype was not included in the multivariable analysis due to a high proportion of “not specified” histology (48.6%). Patients were censored at the time of second melanoma diagnosis, loss to follow-up evaluation, death, or 10 years after initial melanoma diagnosis, whichever came first. An identical analytic approach then was followed for the cohort with initial synchronous melanomas.

Stata software (version 15) was used for all statistical analysis, with a p value lower than 0.05 considered statistically significant. The analysis of de-identified data was exempted from the University of Wisconsin Institutional Review Board.

Results

Main Analysis

Overall cohort demographics are reported in Table 1. Of the 152,811 individuals in the cohort, 114,470 had at least 5 years of follow-up evaluation, and 53,715 had at least 10 years of follow-up evaluation. Figure 2 displays the cumulative incidence curve for the cohort together with the number of individuals evaluable at each time point. The overall incidence of second melanoma was 3.9% (95% confidence interval [CI], 3.8–4.0%) at 5 years and 6.7% (95% CI, 6.6–6.9%) at 10 years (Table 2).

Table 1 Cohort demographics and multivariable Cox regression analysis for the development of a second primary melanoma within 5 years after initial diagnosis for survivors of cutaneous melanoma
Full size table
Fig. 2
figure 2

Cumulative incidence of second primary cutaneous melanoma among survivors of cutaneous melanoma

Full size image
Table 2 Second primary cutaneous melanoma incidence by patient and first primary cutaneous melanoma characteristics for overall cohort and key subgroups
Full size table

In the multivariable analysis (Table 1), older age, male sex, first melanoma site, first melanoma thickness greater than 1.0 mm, and regional disease were independently associated with an increased risk of second primary melanoma. The cumulative incidences of a second primary melanoma by age, sex, and local/regional nodal status are presented in Table 2.

The majority of second melanomas (65%) were thin (0–1.0 mm), but second melanomas had a larger proportion of unknown Breslow thickness (13%). The distributions of site and histologic type of second melanomas generally were similar to those of the first melanoma diagnoses in the overall cohort, but with more head and neck melanomas (26.1%) and more unspecified histologies (54.2%).

Analysis of Individuals with Synchronous Melanomas (Multiple Diagnoses <3 Months After the Initial Diagnosis)

Survivors with a diagnosis of initial synchronous melanomas (n = 1743) were analyzed separately. This cohort was demographically similar to the overall cohort except for a higher proportion of males (70.9%; Appendix Table 4). The incidence of an additional melanoma 3 months or later after a synchronous melanoma diagnosis was 12.3 % (95% CI, 10.7–4.0%) at 5 years and 18.3% (95 % CI, 16.1–20.8%) at 10 years (Appendix Table 5).

Discussion

This contemporary population-based analysis identified an overall cumulative incidence of second melanomas in survivors of melanoma to be 3.9% at 5 years. This estimate is consistent with prior national and single-institution 5-year incidence estimates, which range between 2% and 12%. For the subset of patients with synchronous melanomas (multiple melanomas diagnosed within 3 months after the index diagnosis), estimates are markedly higher, reaching 12.3% at 5 years for an additional melanoma diagnosis. Our refined estimate can be used in clinical care to counsel on risk of a second melanoma and to encourage skin surveillance for survivors who have completed their initial surgical management. Furthermore, our findings can provide reassurance to patients that the absolute risk of a second melanoma at 5 years is relatively low, especially for certain subgroups such as younger patients.

One reason for the variability in the incidence of a second melanoma reported in prior studies may stem from the variable definitions used to define a second melanoma. The literature clearly shows a subset of patients who have multiple melanomas diagnosed within a short period. Prior studies have varied in the period they use to define a synchronous primary versus a true second melanoma. Table 3 presents prior studies that have reported on the incidence of a second primary melanoma in melanoma survivors. These studies all have variable definitions of synchronous melanoma and variable criteria for excluding individuals with synchronous melanoma from analysis.

Table 3 Summary of previous studies
Full size table

We chose to consider a cutoff of 3 months for a second melanoma while also doing a secondary analysis including those patients who have multiple melanomas diagnosed within 3 months after the initial melanoma diagnosis. The large difference in estimates of a second melanoma event support a separate analysis of these cohorts. We believe this is the most useful estimate for counseling regarding the role of skin surveillance.

Another reason for the wide-ranging incidences reported in prior studies could be variable inclusion of melanoma in situ as a first melanoma diagnosis. Studies that examined the incidence of cutaneous melanoma after melanoma in situ and invasive melanoma generally reported higher 5-year incidences.9,15 Melanoma in situ is an entity that has pathologic diagnosis challenged by low reproducibility and accuracy.19 The increasing incidence of melanoma in situ in the United States over time has been observed across varied sub-populations, and overdiagnosis is likely a contributing factor to this trend.20 Given the uncertainty concerning the clinical significance of melanoma in situ, we focused on individuals with invasive melanomas as a distinct group with a more advanced disease process in this study.

Single-center studies also have tended to report higher incidences of a second melanoma.8,9,15 It is possible that closer and more comprehensively documented follow-up evaluation at specialized cancer centers further contributes to increased detection and reporting of second melanomas. We aimed to identify population-level incidence of second melanoma for a more generalizable estimate for melanoma survivors and their providers.

Our 5-year cumulative incidence of a second primary melanoma is nearly double that reported in a similarly focused historical analysis of the SEER database by Goggins and Tsao,7 who evaluated invasive melanomas diagnosed between 1973 and 1997. In that cohort of melanoma survivors, the authors identified the 5-year the cumulative incidence of a second primary melanoma to be 2.06%.

The increased incidence of a second primary melanoma identified in our study follows an overall trend of an increasing melanoma incidence in the general population in recent years. Although the population-wide increase in first melanomas is likely multi-factorial, there are concerns that overdiagnosis contributes to this trend. Welch et al.21 describes a “cycle of melanoma overdiagnosis” characterized by increased referrals to dermatology, increased biopsies, and lower thresholds used by pathologists to diagnose melanoma, without concomitant improvement in outcomes such as survival. This has led Welch et al.21 to argue against population-wide screening for skin cancer in asymptomatic individuals. The causes and approaches to the overall increase in melanoma incidence are admittedly controversial, with serious consequences for “missed” melanoma diagnoses, but it is generally agreed that accurate diagnosis of clinically meaningful melanomas is a challenging area warranting further improvement.22,23,24,25

Overdiagnosis also likely has some contribution to the increased incidence of second melanomas reported in our study compared with historical studies. We anticipate that melanoma survivors may be even more susceptible to receiving a biopsy recommendation by dermatology and a subsequent melanoma diagnosis by pathology for a borderline atypical skin lesion than the general population. Efforts to limit unnecessary biopsies and pathologic overdiagnosis will unquestionably benefit survivors of melanoma. However, melanoma survivors experience second melanomas at an incidence that likely represents an underlying set of risk factors for these patients.3,7 Current practices of skin surveillance for melanoma survivors leads to the diagnosis of most second primary melanomas when they are thin and prognostically favorable.8,9,11 For melanoma survivors, skin surveillance has the additional benefit of monitoring for recurrence. We believe that despite the possibility of overdiagnosis, melanoma survivors benefit from regular skin surveillance as a component of their follow-up evaluation.

Study Limitations

Our study had a few limitations. First, complete oncologic data such as tumor histology, disease stage, and thickness were not available for a number of cases.

Second, melanomas had a larger volume of such missing data, which limited our ability to make assumptions about the clinical significance of a second melanoma. Furthermore, our study was not able to account for all the risk factors for a second melanoma because the SEER database does not capture certain known melanoma risk factors such as family history,15,26 nevi characteristics,15,16,26,27 and ultraviolet damage.28,29 We attempted to assess the risk of a second melanoma for certain subgroups of interest, specifically non-white patients and those with acral melanomas. However, the sample sizes for these subgroups were too small to allow an accurate estimation of risk.

Finally, there is not one accepted definition of what counts as a synchronous melanoma versus a second melanoma. We chose to define synchronous melanomas as those diagnosed fewer than 3 months after the initial diagnosis, consistent with other studies in the literature.17,18 Although this definition may be imperfect, by providing a separate analysis for these patients who have melanomas diagnosed within 3 months, we believe we provide the most relevant information to inform counseling. Despite these limitations, this study provides population-level data that offer the most comprehensive estimate of second melanoma incidence.

Conclusion

At 5 years, melanoma survivors have a 3.9% risk of a second melanoma. With current practice, most second melanomas are identified at an early stage. Our findings support the recommendation of routine skin surveillance as a component of survivorship care for melanoma survivors. Our study contributes accurate information on the risk of additional melanoma diagnoses, which is highly valuable for survivors and their health care providers planning follow-up care.