Second Primary Malignancies in Patients With Melanoma Subtypes: Analysis of 120,299 Patients From the SEER Database (2000-2016)

Purpose Evaluate incidence of second primary malignancies (SPM) after non-acral cutaneous melanoma (NACM), acral lentiginous melanoma (ALM), mucosal melanoma (MM), and uveal melanoma (UM). Patients and Methods First primary NACM, ALM, MM, and UM cases diagnosed 2000-2016 were extracted from SEER. Seer*Stat was used to calculate excess absolute risks (EAR) and standardized incidence ratios (SIR) of SPMs relative to a matched cohort from the general population. P-value of 0.05 determined significance. Results Inclusion criteria was met by 109,385 patients with NACM, 2166 with ALM, 2498 with MM, and 6250 with UM. Increased incidence of malignancies occurred after NACM (SIR 1.51; 95%CI, 1.49-1.54), ALM (SIR 1.59; 95%CI, 1.40-1.81), MM (SIR 2.14; 95%CI, 1.85-2.45), and UM (SIR 1.24; 95%CI, 1.14-1.34) relative to the general population. Cutaneous melanoma occurred more frequently after NACM (SIR 9.54; 95%CI, 9.27-9.83), ALM (SIR 12.19; 95%CI, 9.70-15.14), MM (SIR 10.05; 95%CI, 7.18-13.68), and UM (SIR 2.91; 95%CI, 2.27-3.66). Patients with initial NACM (SIR 2.44; 95%CI, 1.64-3.51) and UM (SIR 44.34; 95%CI, 29.91-63.29) demonstrated increased incidence of eye and orbit melanoma. Renal malignancies occurred more frequently after NACM (SIR 1.24; 95%CI, 1.11-1.38), MM (SIR 3.54; 95%CI, 1.62-6.72) and UM (SIR 1.68; 95%CI, 1.09-2.48). Increased incidence of thyroid malignancies was observed after NACM (SIR 1.83; 95%CI, 1.61-2.06), ALM (SIR 3.74; 95%CI, 1.71-7.11), MM (SIR 4.40; 95%CI, 1.77-9.06), and UM (SIR 3.79; 95%CI, 2.52-5.47). Increased incidence of lymphoma was observed after NACM (SIR 1.20; 95%CI, 1.09-1.31) and ALM (SIR 2.06; 95%CI, 1.13-3.46). Conclusion Patients with NACM, ALM, MM, and UM have increased incidence of SPMs compared to that expected from the general population. Each of these melanoma subtypes had increased occurrence of cutaneous melanoma and thyroid cancer; some, but not all, had increased occurrence of renal malignancies, eye and orbit melanoma, and lymphoma.

NACM generally portends a better prognosis than ALM, MM or UM, with 5-year survival rates of 91.3%, 80.3%, 34.0%, and 78.4% respectively (4,6). If complete remission is attained, patient care becomes increasingly focused on surveillance for recurrences and management of cancer sequalae; second primary malignancies (SPMs) embody one such sequela. Although incidence of SPMs has been investigated for patients with cutaneous melanoma (CM; encompasses NACM and ALM) (9)(10)(11) and UM (12,13), limited literature exists on SPMs specific to MM (14) and ALM (15). Prior study of SPMs associated with mucosal melanoma (14) are limited solely to those arising from the sinonasal cavity and prior study of SPMs associated with ALM (15) focuses on an exclusively Korean population; both lack site-specific SPM risk investigation. With gaps in current literature, consensus cancer guidelines provided by organizations such as National Comprehensive Cancer Network (NCCN) (16)(17)(18), Cancer Care Ontario (CCO) (19), Canadian Medical Association (CMA), and European Society for Medical Oncology (ESMO) (20) provide either no or very limited discussion on SPM risk and follow-up after these malignancies.
In order to address this literature gap, we conducted a retrospective analysis of the SEER database to evaluate if patients with NACM, ALM, MM, and UM demonstrate increased incidence for SPMs compared to the general population in the contemporary era (2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016). We performed additional analysis to identify specific sites and latency periods with elevated risk for secondary malignancies. We conduct, to the best of our knowledge, the first investigation of site-specific SPM risk after MM and ALM. National Cancer Institute's (NCI) Surveillance, Epidemiology, and End Results (SEER) registries (21), a national population-based cancer database, has been used and validated for such analyses in the past (9)(10)(11)(12)22).

Data Source
Cases of melanoma were extracted from the SEER database, which is comprised of up to 21 cancer registries that geographically account for approximately 36.7% of the US population (21). The specific dataset used for this study, "Incidence -SEER 18 Regs excluding AK Research Data, Nov 2018 Sub (2000-2016)", contained data from 18 registries with cases diagnosed between 2000 and 2016. The SEER program tracks incidence of new tumors and documents demographic, treatment, tumor, and survival data; however, it does not include behavioral risk factors (e.g. smoking, physical inactivity) and comorbid diseases. Institutional review board approval was not required for this study, as it utilized only deidentified data with permission from NCI.

Data Collection
Patients diagnosed with NACM, ALM, MM, and UM between 2000-2016 were included in the study; cases that were not first primary malignancies, were diagnosed by death certificate, were diagnosed by autopsy record, or were of unknown age were excluded from analysis.

Statistical Analysis
Demographic and tumor data was tabulated. The multiple primary standardized incidence ratio (MP-SIR) algorithm of the Seer*stat program (version 8.3.6.1) was used to obtain standardized incidence ratios (SIR) and excess absolute risk (EAR) for second primary malignancies in patients with NACM, ALM, MM, and UM compared to a reference group representative of the general population, with similar sex, race (white/unknown, black, other) age-group (5-year interval), and calendar year of diagnosis (5-year interval). The algorithm was then further used to identify specific latency periods in which there was increased incidence of SPMs relative to the reference population. The authors (AL, SP, DSG) examined the sitespecific analysis to identify trends across melanoma subtypes.
Analysis was limited to second malignancies only (early exit at next malignancy) to isolate relationship of subsequent malignancies with the first primary. Only malignant neoplasms diagnosed greater than two months after the melanoma diagnosis were considered to be second primaries, in order to distinguish them from concurrent malignancies discovered during screening. The reference population linked to the SEER database is comprised of Census Bureau data, through partnership with the National Center for Health Statistics (https://seer.cancer.gov/popdata/). An alpha level of significance of 0.05 was used for the study, and EAR was calculated per 10,000 individuals. IBM Statistical Product and Service Solutions (SPSS) version 26 and Microsoft Excel version 16.38 were used to conduct descriptive analysis and generate charts.

Baseline Characteristics
Inclusion criteria was met by 109,385 patients with NACM, 2166 patients with ALM, 2498 patients with MM, and 6250 patients with UM for a total of 120,299 patients. The median (+/-SD) follow-up period for patients with NACM was 5.6 (+/-4.7) years, with ALM was 4.3 (+/-4.5) years, with MM was 1.7 (+/-3.6) years, and with UM was 4.
CM occurred more frequently in patients with initial NACM (SIR 9.54; 95% CI, 9.27
Although historic and smaller retrospective analyses of the SEER database have examined SPMs following CM (9, 10) and UM (12), herein we provide, to the best of our knowledge, the first investigation of site-specific SPM risk after MM and ALM. Moreover, through analysis of UM in a larger and more contemporary cohort, we highlight increased incidence of secondary thyroid malignancies, a finding undetected in prior investigation (12). Bradford, et al. (9) and Spanogle, et al. (10) investigated incidence of SPMs following CM in various subsets of the SEER database and both found increased incidence of secondary CM, eye and orbit melanoma, thyroid cancer, renal cancer, and lymphoma. Similarly, Vakharia, et al. (11) in their investigation of secondary malignancies excluding CM demonstrated increased risk for these sites. Their findings (9-11) were consistent with our findings of SPMs following NACM. Laıńs, et al. (12) investigated risk of second primary malignancies following UM and found increased incidence for secondary CM, eye and orbit melanoma, and renal cancer but not a significant increase in thyroid cancer. However, the study (12) showed a strong trend of increased thyroid cancer (SIR 2.06, 95% CI 0.99 to 3.78) in a cohort of 3976 patients, which with increased power may have captured a significant result similar to our study.
Despite a growing body of literature on SPMs, national consensus guidelines such as NCCN (16-18) (US), Canada CCO (19) (Canada), CMA (Canada) and ESMO (20) (Europe) sparsely address secondary malignancies in their follow-up recommendations. Canadian (19) and European (20) guidelines discuss only an increased risk for secondary cutaneous malignancies after initial CM and the importance of long-term dermatological surveillance; these guidelines lack SPM follow-up recommendations specific to ALM, MM, and UM. American consensus guidelines provide slightly more insight on noncutaneous SPMs following CM by discussing the role of genetic testing in determining SPM risk and by providing guidance on when to consider such testing (16). However, American guidelines do not remark on the increased incidence of lymphoma or thyroid cancer (16). Moreover, these guidelines state that CM is not associated with an increased risk for UM (17).
Indeed, developing follow-up recommendations poses a challenge as cost, clinical benefit, and burden of increased health-care visits must all be balanced. Nonetheless, increased awareness of the associations studied herein are paramount to guiding appropriate clinician judgement when caring for melanoma patients; discussion of up-to-date evidence in national guidelines can improve patient care and long-term health outcomes. With a more appropriate index of suspicion, lesions (e.g. renal cyst, thyroid nodule) and atypical findings discovered during diagnostic or surveillance imaging that may otherwise have been dismissed as benign may instead be deemed to warrant additional follow-up. Furthermore, symptoms concerning for an associated SPM (e.g. visual flashers and floaters) can be interpreted more appropriately and potentially lead to earlier diagnosis. The high-risk latency periods identified by this study may provide additional clinical insight when deciding further management for patients presenting with these signs or symptoms.
The authors propose that cost effective screening such as total body skin exams be recommended for patients with all subtypes (NACM, ALM, MM, UM) of melanoma; prior studies (44,45) support the economic efficiency of targeted screening strategies

Limitations in the Study Design
An important limitation of the SEER database and our study is the possibility of miscoding a recurrence as a second primary malignancy when pathologic evaluation is unavailable and tumor location is identical. This most directly impacts the calculation of secondary eye and orbit melanoma after initial UM (6.67% cases without microscopic confirmation), as UM is largely diagnosed clinically through examination and imaging rather than pathologically (48); indeed, this finding is more likely representative of recurrences rather than true SPMs. Another important limitation is the inability of Seer.Stat's MP-SIR algorithm to analyze site-specific secondary malignancy incidence beyond those preset in the software. As a result, we were unable to provide analysis on incidence of NACM and ALM as secondary malignancies, and instead had them grouped as secondary CM. Other limitations include other possible miscoding and inability to account for variables not included within the database. Additionally, patients that moved to a geographical area not covered by SEER could be lost to followup leading to underreporting. Despite these limitations, however, the national database has been validated for SPM analyses (9)(10)(11)(12)22).

CONCLUSIONS
We found patients with NACM, ALM, MM, and UM to have increased incidence of SPMs compared to that expected from the general population. Each of these melanoma subtypes had increased occurrence of secondary CM and thyroid cancer; some, but not all, had increased occurrence of secondary renal malignancies, eye and orbit melanoma, and lymphoma. These patients may benefit from cost-effective screening methods such as full body skin exams. Patients with NACM should, at a minimum, receive age-appropriate comprehensive eye screening per national guidelines. Increased awareness of these associations is prudent to guiding clinical follow-up and additional studies are necessary to identify best-practice screening guidelines.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding author.