Risk Factors of Subsequent Primary Melanomas in Austria

Key Points

Question  Which factors influence the risk of developing subsequent melanomas after first melanoma?

Findings  In this case-control study of 1648 patients with melanoma, odds ratios for CDKN2A mutations, nonmelanoma skin cancer, number of nevi, and actinic damage on the back were associated with multiple primary melanomas. Patients with many nevi were younger at the time of the diagnosis of their first melanoma.

Meaning  In patients with increased risk of multiple melanomas, current follow-up intervals should be adapted to prevent delayed diagnosis, and patients should be informed to raise awareness about improving sun protection and to encourage self-examination.

Importance  Information on risk factors of subsequent melanomas would be helpful to identify patients at risk after the diagnosis of their first melanomas.

Objective  To determine risk factors of subsequent melanomas.

Design, Setting, and Participants  In this retrospective case-control study, 1648 participants with histologically verified cutaneous melanoma diagnosed from January 1, 1968, though March 16, 2015, were recruited from a tertiary referral center as part of the Molecular Markers of Melanoma study. CDKN2A was sequenced in 514 and MC1R in 953 participants. Data were analyzed from March 7, 2008, through March 25, 2015.

Main Outcomes and Measures  Phenotypic traits and internal and external risk factors for the development of a second, third, or fourth melanoma.

Results  In total, 1648 patients (53.6% men; mean [SD] age, 54 [15] years) were enrolled, including 1349 with single and 299 with multiple primary melanoma. Mean (SD) age at recruitment was 57 (15) years for the single-melanoma and 62 (14) years for the multiple-melanoma groups. From the internal risk factors, family history (odds ratio [OR], 1.76; 95% CI, 1.22-2.55; P = .006), CDKN2A high-risk mutations (OR, 4.03; 95% CI, 1.28-12.70; P = .02), and high numbers of nevi as a phenotypic risk factor (ORs, 2.23 [95% CI, 1.56-3.28, P < .001] for 20-30 smaller nevi and 2.56 [95% CI, 1.50-4.36; P = .003] for 20-30 larger nevi) were significantly associated with the risk of developing a subsequent primary melanoma using multivariate logistic regression analysis. Nonmelanoma skin cancer (OR, 2.57; 95% CI, 1.84-3.58; P < .001) and signs of actinic skin damage, particularly on the back (ORs, 1.91 [95% CI, 1.12-3.25; P = .04] for freckling and 1.92 [95% CI, 1.29-3.08; P = .007] for solar lentigines), additionally increased risk of a subsequent melanoma. All those factors were also associated with an earlier development of the second melanoma. Patients with 3 melanomas developed their second melanoma earlier than patients with only 2 melanomas (mean [SD] age, 55 [15] years for those with 2 primary melanomas; 52 [15] years for those with 3 primary melanomas). Time spent outdoors, solarium use, outdoor occupation, and hair color had no significant associations in these models.

Conclusions and Relevance  According to the results of this study, internal factors (family history and genetic variants), number of nevi, and actinic damage on the back are more relevant for the development of subsequent melanomas than skin phototype or hair color. Patients with many nevi were younger at the time of the diagnosis of their first melanoma. This finding could help to identify persons at increased risk of developing multiple primary melanomas.

After the diagnosis of melanoma, patients are recommended to undergo follow-up examinations for several years. According to current international guidelines,^1-5 the intervals between the physical examinations depend largely on the histopathologic characteristics of the primary tumor, such as invasion thickness, number of mitoses, ulceration, or occurrence of clinically detectable metastases.⁶ This association is mainly due to the focus of such guidelines on avoiding disease progression, rather than detecting subsequent primary tumors at an early stage. However, the diagnosis of melanoma has been repeatedly reported to be a risk factor of subsequent melanomas.^7,8 Therefore, more detailed information on patients who are at risk of acquiring subsequent primary melanomas could be helpful to adjust the follow-up scheme to become more effective in early identification of subsequent primary melanomas. Melanoma risk is widely acknowledged to be based on a complex interaction between genetic and external risk factors.⁹ Genetic testing is currently recommended for patients who fulfil certain criteria. For example, the GenoMEL Consortium, which is composed of scientific groups around the world studying genetic risk factors for melanoma, recommends offering genetic testing if someone fulfills 1 or more of the following criteria: at least 2 cases with melanoma in a pedigree; 2 or more melanomas in a single person; or a diagnosis at younger than 50 years.¹⁰ Mutations of the CDKN2A gene (ENSG00000147889) are currently the best studied high-penetrance genetic risk factors. Only recently, melanoma risk of persons carrying MC1R (ENSG00000258839) variants has been reported to be independent of sun exposure.¹¹ Despite numerous studies describing risk factors for primary melanomas, information on the risk for developing subsequent primary melanomas still remains very limited.^12-17 Until now, studies were limited to genetic variations or sun exposure only, although external as well as internal risk factors are acknowledged risk factors for melanoma. Owing to this interplay of internal and external factors, we aimed to analyze phenotypic traits and internal (ie, inherited) and external risk factors of subsequent melanomas in a population of a region with a moderate risk of melanoma.

In total, 1648 patients with melanoma were included for this study. All participants were recruited for the Molecular Markers of Melanoma study and were diagnosed with melanoma from January 1, 1968, through March 16, 2015.^18,19 The participants were recruited in different hospitals to address a potential selection bias.¹⁸ This study was approved by the ethics committee of the Medical University of Vienna, Vienna, Austria, and all patients signed written informed consent.

Only patients with a histologically verified cutaneous melanoma were included. Participants with missing data were excluded in the specific analysis in which the data were necessary for the calculation. Two hundred and ninety-nine patients developed additional melanomas during their follow-up. They were divided into subgroups with at least 2, at least 3, and at least 4 histologically confirmed primary melanomas. The mean observation period (time from diagnosis of the first primary melanoma till the last check of the records) was 122 (SD, 76) months. For all comparative analysis, patients with only 1 primary melanoma constituted the reference group (controls). We analyzed different risk factors for subsequent primary melanomas and classified them as phenotypic traits (skin phototype, hair color, and the number of nevi), internal or genetic risk factors (family history for melanoma, MC1R and CDKN2A gene status), and external risk factors (signs of actinic damage, occurrence of nonmelanoma skin cancer, outdoor occupation, time spent outdoors at leisure, sunscreen use before melanoma diagnosis, and solarium use). The signs of clinically visible actinic sun damage used in this study included freckling and solar lentigines of the face, neck, and hands as well as wrinkling on the neck and face. Photographs of all participants were taken under standardized conditions. To test for interobserver variability, all signs of actinic damage were independently reevaluated in 50 participants by 2 dermatologists (J.W. and I.O.) using these photographs, as described previously.¹⁹ A person was only indicated as positive for nonmelanoma skin cancer (mainly basal cell carcinoma, squamous cell carcinoma, Bowen disease, actinic keratosis, and keratoacanthoma) if at least 1 cancer was histologically verified.

The number of nevi was assessed in 5 categories (0, <10, 10-19, 20-30, or >30) and additionally separated between nevi smaller than 0.6 mm and at least 0.6 mm. The groups with 0 and less than 10 nevi were consolidated for all analyses.

Genomic DNA was extracted from whole blood using a DNA purification kit (Wizard Genomic; Promega Corporation). Data for CDKN2A exon 1α and 1β and exon 2 were available from 514 patients and were sequenced as described previously.¹⁸ Carriers of CDKN2A mutations associated with a high risk of melanoma were compared with noncarriers. The coding sequence of MC1R was amplified by polymerase chain reaction for 953 patients as described previously.^20,21 Variants p.86_87insA, p.D84E, p.R142H, p.R151C, p.R160W, and p.D294H were classified as R (high risk), whereas the variants p.V60L, p.V92M, p.I155T, and p.R163Q were classified as r (low risk). All other variants and wild-type alleles were classified as 0.

Data were analyzed from March 7, 2008, through March 25, 2015. Quantitative variables were represented by their mean values including SDs and were compared using a 1-way analysis of variance. Odds ratios (OR) were calculated using logistic regression models. Risk factors for patients with at least 2, 3, and 4 primary melanomas were compared with those of patients with a single primary melanoma. In the multivariate logistic regression model, sex and age at recruitment were included as adjustment variables. This model was also tested for multiple hypotheses using the Bonferroni-Holm correction. In addition, because the development of subsequent melanomas is a sequential event, a multivariate Cox proportional hazards regression analysis was performed from the time of the first to the second melanoma. Sex and recruitment age were used as adjustment variables. Therefore, both results, by logistic regression after testing for multiple testing as well as Cox proportional hazards regression, will be presented. A result was considered to be statistically significant if the 2-sided P < .05 with a 95% CI. For all tests, we used SPSS software (version 24; IBM Corp).

A total of 1648 patients were included in the analysis (765 [46.4%] women and 883 [53.6%] men; mean [SD] age, 54 [15] years). Of the 299 patients with subsequent melanomas, 89 (29.8%) had at least 3 and 33 (11.0%) had 4 or more primary melanomas. Compared with patients with a single primary melanoma, patients with multiple melanomas were more frequently male (185 of 299 [61.9%] vs 698 of 1349 [51.7%]) and were more likely to have a personal history of nonmelanoma skin cancer (89 of 263 [33.8%] vs 173 of 1234 [14.0%]) and a positive family history of melanoma (46 of 298 [15.4%] vs 140 of 1347 [10.4%]) (Table 1). The mean time from diagnosis of the first to the second primary melanoma was 63 (68) months for those with 2 primary melanomas and 42 (57) months for those with 3 primary melanomas (P = .04) (eTable 1 in the Supplement). Patients with 3 melanomas developed their second melanoma earlier than patients with only 2 melanomas (mean [SD] age, 55 [15] years for those with 2 primary melanomas; 52 [15] years for those with 3 primary melanomas). In the univariate model, the variables of recruitment age, sex, nevi count, family history, freckling, wrinkling, solar lentigines, nonmelanoma skin cancer, and solarium use were significantly associated with the risk of subsequent primary melanomas (OR range, 0.66 [95% CI, 0.51-0.85] to 3.14 [95% CI, 2.32-4.24]) (Table 1). All further analyses were performed using a multivariate model, adjusted for sex and recruitment age.

The number of nevi correlated with the number of melanomas detected in a patient. When nevus count was analyzed separately for nevi smaller than and at least 0.6 mm, both nevi groups were significantly associated with increased risk of subsequent melanomas. Patients with 20 to 30 nevi had the highest risk to develop a second primary melanoma (ORs, 2.23 [95% CI, 1.56-3.28; P < .001] for nevi <0.6 mm and 2.56 [95% CI. 1.50-4.36; P = .003] for nevi ≥0.6 mm). A count of more than 30 small nevi was associated with the highest risk to develop 3 melanomas (OR, 3.46; 95% CI, 1.68-7.13; P = .003). Hair color did not have a significant association, and skin phototype had a significant association only for types I and II for the risk of a second primary melanoma (OR, 1.79; 95% CI, 1.14-2.82; P = .04) (Table 2).

In the Cox proportional hazards regression model for larger nevi, higher counts had the highest risk for the development of a second primary melanoma (hazard ratios [HRs], 2.83 [95% CI, 1.81-4.43; P < .001] for 20-30 large nevi and 2.52 [95% CI, 1.46-4.35; P = .001] for >30 large nevi) (eTable 2 in the Supplement). Patients with many small nevi and those with 20 to 30 large nevi were also younger at the time of diagnosis of the first melanoma compared with patients with fewer nevi (mean [SD] age, 49 [14] vs 57 [15] years [P < .001] for those with small nevi and 48 [14] vs 54 [15] years [P = .001] for those with large nevi) (eTable 3 in the Supplement).

For internal risk factors, a positive family history and high-risk CDKN2A mutations significantly increased the risk for developing at least 2 (ORs, 1.76 [95% CI, 1.22-2.55; P = .006] for family history and 4.03 [95% CI, 1.28-12.70; P = .02] for CDKN2A mutations), 3 (ORs, 2.00 [95% CI, 1.11-3.62; P = .04] for family history and 10.18 [95% CI, 2.77-37.46; P < .001] for high-risk CDKN2A mutations), and 4 primary melanomas (ORs, 2.67 [95% CI, 1.12-6.36; P = .03] for family history and 23.83 [95% CI, 4.74-119.65; P < .001] for high-risk CDKN2A mutations). Patients with high-risk CDKN2A mutations were also younger at the time of the diagnosis of the first primary melanoma (mean [SD], 37 [12] vs 51 [16] years; P = .002) (eTable 3 in the Supplement). For MC1R, the combination of 2 or more r variants was associated with the highest risk for developing subsequent melanomas (OR for ≥3 primary melanoma, 3.52; 95% CI, 1.38-8.98; P = .02) in logistic regression (Table 3).

Similar results were obtained by Cox proportional hazards regression (HRs, 2.44 [95% CI, 1.24-4.82; P = .01] for high-risk CDKN2A mutations and 1.63 [95% CI, 1.19-2.23; P = .002] for a positive family history), except for MC1R red hair variants, which failed to remain statistically significant. All results of the internal factors were shown in eTables 3 and 4 in the Supplement.

Sun exposure was assessed by the degree of actinic damage on the skin as described previously.¹⁹ Signs of actinic damage on the back, including freckling and solar lentigines, showed the highest risk of subsequent primary melanomas among all investigated signs of actinic damage (ORs for a second primary melanoma, 1.91 [95% CI, 1.12-3.25; P = .04] for freckling and 1.92 [95% CI, 1.29-3.08; P = .007] for solar lentigines; ORs for a third primary melanoma, 3.46 [95% CI, 1.15-10.36; P = .03] for freckling and 5.61 [95% CI, 1.88-16.77; P = .004] for solar lentigines) (Table 4). The occurrence of nonmelanoma skin cancer increased the risk of 2 (OR, 2.57; 95% CI, 1.84-3.58; P < .001), 3 (OR, 2.95; 95% CI, 1.74-5.00; P < .001), or 4 (OR, 3.52; 95% CI, 1.55-8.01; P = .003) primary melanomas.

Similar to the results of the logistic regression analysis, nonmelanoma skin cancer (HR, 1.90; 95% CI, 1.49-2.66; P < .001) and actinic skin damage on the back (HRs, 1.83 [95% CI, 1.12-2.99; P = .02] for freckling on the back and 1.69 [95% CI, 1.09-2.60; P = .02] for solar lentigines on the back) increased the risk for developing a second primary melanoma in the Cox proportional regression model (eTable 5 in the Supplement). Leisure time spent outdoors, time spent in an outdoor occupation, and solarium use did not increase the risk of subsequent melanoma in any model.

Besides the number of nevi and a positive family history of melanoma, no other risk factor of subsequent melanoma is well known except a previous diagnosis of melanoma itself.^7,8 The frequency of subsequent melanomas was reported to range from 8.2% of previously diagnosed melanoma in European countries to 23% in continents with a more intense ambient UV radiation.^15,22-28 These numbers reflect the complex interplay between internal (ie, inherited) and external risk factors, mainly UV radiation, as the base for melanoma development in general. For example, MC1R red hair variants, the most important contributor of pheomelanin production in melanocytes leading to less effective UV protection, have been reported to increase risk even independently of sun damage.^11,29 On the other hand, sun damage on the dorsal body sites, including wrinkling on the neck and solar lentigo and freckling on the back, increases the risk of melanoma as well.¹⁹

In our present study, we aimed to identify risk factors in patients with melanoma that are associated with subsequent melanomas and found evidence of internal (CDKN2A) as well as external factors (freckling on the back and hands and nonmelanoma skin cancer as a consequence of UV radiation) to increase risk. However, in contrast to studies of risk factors for melanoma in general,^30,31 pigmentation phenotype such as hair color or skin phototype did not seem to have an effect. For MC1R, the combination of 2 or more r variants was associated with subsequent melanomas in the logistic regression analysis that failed to remain significant in the Cox proportional hazards regression analysis. This finding might change when patients are observed for a longer time, and larger studies might be helpful to elucidate the effect of MC1R red hair variants on subsequent melanomas; however, MC1R red hair variants do not seem to be as important as risk factors of subsequent melanomas.

As described above, high nevus count is an established risk factor of primary melanomas in general.^32,33 In our study, a high nevi count was associated with subsequent melanomas in the logistic regression as well as Cox proportional hazards regression analysis. In fact, patients with many nevi were also younger at the time of their first melanoma, supporting the idea that nevi and multiple melanomas share the same driving force. Despite previous reports describing a higher risk for larger nevi,^34,35 risk of multiple melanomas appeared to be independent of size in our study.

Most previous studies have used logistic regression to analyze potential risk factors. In our current study, we have used Cox proportional hazards regression analysis additionally as the event for which we are studying risk factors that occur over time. Using Cox proportional hazards regression analyses, almost all factors associated with the risk of subsequent melanomas in the logistic regression analysis remained significant. Factors that were not confirmed by Cox proportional hazards regression analysis were solar lentigo located on the hands and MC1R red hair variants. Despite our common assumption that patients with a more sun-sensitive skin phototype or lighter skin complexion would require more attention, our results put sun-sensitive skin in another perspective with regard to subsequent melanomas. This finding is in concordance with the fact that MC1R red hair variants were not associated with subsequent melanomas.

One potential limitation of this study was the retrospective design. Moreover, as a single-center study, we cannot exclude a potential selection bias of our patients. In particular, subsequent in situ melanomas could have been missed when they were removed in other clinics or physicians’ offices or after the observation period. However, loss due to missed reports by the patients seems unlikely, because patients tend to report even removal of benign lesions by an outside physician. Besides, our rate of subsequent melanomas is higher than reported in other studies from Europe.^22,36

We demonstrated that high-risk CDKN2A germline mutations were associated with the highest increase of subsequent melanoma risk, followed by occurrence of nonmelanoma skin cancer, a high count of large or small nevi, and actinic skin damage using 2 statistical models. Despite our finding that nonmelanoma skin cancer and actinic damage are associated with subsequent melanomas in both models, light skin complexion or skin phototype do not seem to affect risk. Taken together, a combination of consulting patients for appropriate sun-avoiding behavior and continued regular follow-up examinations, particularly of patients with the above-mentioned features, at a specialized clinic appears meaningful to avoid delayed diagnosis of subsequent melanomas. The intervals between the follow-up examinations should be further adapted for those carrying these risk factors.

Accepted for Publication: October 18, 2018.

Corresponding Author: Ichiro Okamoto, MD, Department of Dermatology, Medical University of Vienna, Waehringer Guertel 18-20, 1090 Vienna, Austria (ichiro.okamoto@meduniwien.ac.at).

Published Online: December 19, 2018. doi:10.1001/jamadermatol.2018.4645

Author Contributions: Drs Müller and Okamoto had full access to all data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Müller, Okamoto.

Acquisition, analysis, or interpretation of data: Müller, Wendt, Rauscher, Sunder-Plassmann, Richtig, Fae, Fischer.

Drafting of the manuscript: Müller, Okamoto.

Critical revision of the manuscript for important intellectual content: Wendt, Rauscher, Sunder-Plassmann, Richtig, Fae, Fischer.

Statistical analysis: Müller, Wendt.

Obtained funding: Okamoto.

Administrative, technical, or material support: Rauscher, Sunder-Plassmann, Fae, Okamoto.

Supervision: Okamoto.

Conflict of Interest Disclosures: None reported.

Funding/Support: This work was supported by grant 15079 from the Anniversary Fund of the Austrian National Bank and grant 10077 from the Medical and Scientific Fund of the Mayor of the City of Vienna.

Role of the Funder/Sponsor: The funders were not involved in design and conduction of the study; collection; management, analysis and interpretation of the data; preparation, review or approval of the manuscript; and decision to submit the manuscript for publication.

Additional Contributions: We thank all participants of the Molecular Markers of Melanoma study for their contribution.