Comparison of Survival Between Patients With Single vs Multiple Primary Cutaneous Melanomas

Key Points

Question  Is there a difference in survival between patients with a single primary melanoma vs multiple primary melanomas?

Findings  In this cohort study of 54 645 patients with a single primary melanoma and 2284 patients with multiple primary melanomas in the Netherlands, Breslow thickness was significantly decreased for second primary melanomas, but the likelihood of dying was 31% higher among patients with multiple primary melanomas, a statistically significant finding.

Meaning  Stricter follow-up strategies may be warranted for patients with multiple primary melanomas.

Importance  Melanoma is one of the most rapidly increasing forms of cancer worldwide. Most studies about survival among patients with melanoma consider only the primary tumor and disregard the potential effect of multiple primary tumors. A better understanding of the prognosis of patients with multiple primary melanoma is important for patient counselling and follow-up strategies.

Objective  To describe the epidemiologic features of multiple primary melanoma in patients from the Netherlands.

Design, Setting, and Participants  This retrospective, population-based cohort study included adults with histologically proven, primary, invasive cutaneous melanoma in the Netherlands between January 1, 2000, and December 31, 2014, with a median follow-up of 75.1 months, using data from PALGA, the Dutch Nationwide Network and Registry of Histopathology and Cytopathology. Follow-up data were retrieved from the Netherlands Cancer Registry. Statistical analysis was performed from August 1, 2018, to September 3, 2018.

Main Outcomes and Measures  A multivariable Cox model with a time-varying covariate was performed to assess overall survival between patients with a single primary melanoma vs those with multiple primary melanomas. Secondary outcomes included incidence of multiple primary melanoma, differences in Breslow thickness, and time between first and second multiple primary melanoma.

Results  Of the 56 929 study patients, 31 916 (56.1%) were female, with a mean (SD) age of 56.4 (16.2) years. A total of 54 645 single primary melanomas and 4967 multiple primary melanomas in 2284 patients were included. The median Breslow thickness decreased from 0.90 mm (interquartile range, 0.55-1.70 mm) for the first melanoma to 0.65 mm (interquartile range, 0.45-1.10 mm) for the second melanoma (P < .001). For their second melanoma, 370 patients (16.2%) had a higher T stage, 1112 (48.7%) had the same T stage, and 802 (35.1%) had a lower T stage. In addition, 841 of 2284 second melanomas (36.8%) in patients with multiple primary melanomas were found during the first year of follow-up, whereas 624 of 2284 (27.3%) were found after 5 years of follow-up. These proportions did not vary when stratified for melanoma stage. Worse overall survival was seen among patients with multiple primary melanomas compared with patients with a single primary melanoma (hazard ratio, 1.31; 95% CI, 1.20-1.42; P < .001).

Conclusions and Relevance  A significant decrease in Breslow thickness between the first and second multiple primary melanoma was found, and overall survival among patients with multiple primary melanomas was significantly worse than that among patients with a single primary melanoma. These findings suggest that more strict follow-up strategies may be warranted for patients with multiple primary melanomas.

Melanoma is one of the most rapidly increasing forms of cancer worldwide and is accountable for most skin cancer–related deaths.^1-4 Familial atypical mole and melanoma syndrome leads to multiple primary melanomas, but primary melanoma can manifest multiple times outside the framework of familial syndromes. Follow-up guidelines and survival analyses usually consider only the initial melanoma and disregard the potential effects of multiple primary melanomas, which occur in 0.2% to 2.7% of patients with melanoma.^5-16 Few epidemiologic data about tumor and patient characteristics of patients with multiple primary melanomas are available, necessitating better characterization of and prognostication for this group of patients.

Melanoma staging and prognosis relies on Breslow thickness. A rationale for close follow-up after a first melanoma is diagnosed is to detect metastases and a subsequent melanoma as early as possible, ideally at a decreased Breslow thickness. However, no strong evidence exists to support a specific follow-up interval.^17-20 Current Dutch melanoma guidelines recommend 1 follow-up visit within 1 month after initial diagnosis for stage pathologic (p)T1a melanoma. For pT1b melanoma and higher, a follow-up visit every 3 months is recommended for the first year after diagnosis, twice a year for year 2, and annually for years 3 to 5. Further follow-up is recommended for patients with a family history of melanoma and is debated for patients with a higher risk for developing melanoma, such as those with more than 100 nevi or 5 or more atypical nevi. No recommendation is made regarding follow-up for patients with multiple primary melanomas in Dutch guidelines, similar to most international guidelines.²⁰

Conflicting results have been reported regarding survival among patients with a single primary melanoma vs multiple primary melanomas; increased, equal, and decreased survival has been documented.^16,21-27 Some authors argue that these conflicting results have been found because of methods that disregard survival bias in patients with multiple primary melanomas; despite different methodologic approaches to prevent survival bias, conflicting results remain.^21,24,26

Because knowledge about the epidemiologic characteristics of multiple primary melanomas is important for guidelines, patient guidance, and follow-up, the aim of this study was to gain insight into these epidemiologic characteristics of multiple primary melanomas in patients in the Netherlands. This study focused on incidence, differences in Breslow thickness, time between subsequent melanomas, and overall survival among patients with a single primary melanoma vs multiple primary melanomas.

Data for this retrospective nationwide cohort study were obtained from PALGA, the Dutch Nationwide Network and Registry of Histopathology and Cytopathology.²⁸ Since 1987, PALGA has prospectively collected data from all pathology laboratories in the Netherlands. All data were encoded and used anonymously. Ethical approval was granted by the board of PALGA, Houten, the Netherlands. The ethical board of PALGA approves or disapproves all applications based on internal procedures. Anonymous data were used; therefore, patients could not be asked directly for informed consent. Statistical analysis was performed from August 1, 2018, to September 3, 2018.

The pathology reports of all patients with newly diagnosed melanoma in the Netherlands between January 1, 2000, and December 31, 2014, were analyzed. Melanoma in situ, spitzoid tumors of unknown malignant potential, melanocytic tumors of unknown malignant potential, and superficial atypical melanocytic proliferation of uncertain significance were excluded, as were melanomas lacking or having unclear Breslow thickness. We excluded patients with positive sentinel lymph node biopsy findings, lymph node dissection, fine needle aspiration, or otherwise diagnosed positive lymph nodes within 14 days of the diagnosis of the melanoma to ensure that patients were free of clinically detectable nodal disease in the study. This evaluation led to the exclusion of 188 patients with a single primary melanoma (0.34%) and 15 patients with multiple primary melanomas (0.65%). Furthermore, noncutaneous, desmoplastic melanoma, melanoma of unknown primary, recurrences, intransit melanoma, and melanomas occurring among children (<18 years of age) were excluded. This yielded a data set of adults with histologically proven, invasive, primary cutaneous melanoma diagnosed between January 1, 2000, and December 31, 2014, in the Netherlands.

For each patient, clinical and pathologic variables were extracted from the findings on the pathology files, including date of diagnosis, age, sex, Breslow thickness, T stage, ulceration presence or absence, type of melanoma (superficial spreading, nodular, lentigo maligna, or acral lentiginous), and body site (head and neck, trunk, arms, or legs). The TNM staging was in accordance with the American Joint Committee on Cancer staging at the time of diagnosis. Mitoses were included for melanoma for the period that the seventh edition of the American Joint Committee on Cancer Cancer Staging Manual was valid because mitotic rate (≥1 mitoses/mm²) indicated sentinel lymph node biopsy.

Patients with multiple primary melanomas were defined as those with a new primary melanoma diagnosed on or after the date of first melanoma diagnosis irrespective of topography. Thus, patients diagnosed with 2 simultaneous melanomas were registered with the diagnosis of multiple primary melanomas. Multiple melanomas were counted separately in the analysis, resulting in total number of melanomas instead of patients.

Follow-up data, including vital status (dead or alive), were obtained from the Netherlands Cancer Registry, which gathers information about every patient with cancer in the Netherlands, through January 1, 2018. Some patients classified with a single primary melanoma between January 1, 2000, and December 31, 2014, were reclassified as having multiple primary melanomas if the second melanoma occurred between January 1, 2015, and January 1, 2018.

Categorical variables are presented as numbers and percentages. Continuous variables are presented as medians with interquartile ranges (IQRs) for nonnormally distributed data or means (SDs) for normally distributed data. Univariate variables were analyzed using χ² tests or Mann-Whitney test, as appropriate. For multiple primary melanomas, analyses were performed until the sixth multiple primary melanoma. Patients with multiple primary melanomas diagnosed simultaneously were registered with the diagnosis of multiple primary melanomas, and a random order for difference in Breslow thickness calculations was used. Absolute difference in Breslow thickness was calculated between subsequent primary melanoma and tested for significance on group level using the Kruskal-Wallis test and a post hoc pairwise Mann-Whitney test for significance between groups. The difference in time between subsequent melanoma diagnosis was calculated per day.

To prevent survival (also known as immortal time) bias for multiple primary melanomas, we performed Cox regression analysis with a time-varying covariate to assess differences in survival between patients with a single primary melanoma and those with multiple primary melanomas, yielding a hazard ratio (HR). Immortal time refers to a period of follow-up during which, by design, the study outcome (death) cannot occur. By definition, multiple primary melanomas can only be multiple if patients have survived to develop a second melanoma.^29-31 Death, using overall survival data, was selected as the primary outcome and variables were age, Breslow thickness, ulceration, type of melanoma, localization, and sex. The proportional hazards assumption was checked by plotting a log-minus-log graph for all variables. For multiple primary melanomas, all variables of the first melanoma were considered. We performed an additional worst case analysis, in which we included the pathologic characteristics most likely to be associated with death (defined as highest Breslow thickness and its corresponding ulceration status) for each patient with multiple primary melanomas. For example, the initial melanoma is 0.6 mm thick without ulceration; 1 year later, a 2.5-mm-thick melanoma with ulceration occurs. For the regular analysis, we evaluated the tumor characteristics of the first melanoma (0.6 mm without ulceration), whereas in the additional worst case analysis, the characteristics of the 2.5-mm-thick melanoma were used. Cox regression was performed using SAS, version 9.4 (SAS Institute Inc); all other data were analyzed using SPSS, version 21 (IBM Corporation). Two-sided P < .05 was corrected for multiple hypothesis testing according to Bonferroni before it was considered to be statistically significant.

Of the 56 929 patients included in the analysis, 31 916 (56.1%) were female, with a mean (SD) age of 56.4 (16.2) years. Median follow-up time was 75.1 months (range, 43.5-123.5 months). A total of 56 929 patients with 59 612 primary cutaneous melanomas (54 645 single primary melanomas and 4967 multiple primary melanomas in 2284 patients [4.0%]) met our inclusion criteria. In total, 2008 patients had 2 primary melanomas, 206 patients had 3 melanomas, and 70 patients had 4 or more (up to 10) primary cutaneous melanomas. A total of 339 patients had simultaneous multiple primary melanomas.

When comparing single primary melanomas (n = 54 645) with multiple primary melanomas (n = 2284), more males had multiple primary melanomas (1134 [49.6%] vs 23 879 [43.7%]; P < .001). Melanomas on the trunk (1074 [47.0%] vs 22 818 [41.8%]) were more frequent and melanomas on the leg (529 [23.2%] vs 15 196 [27.8%]) were less frequent for multiple primary melanomas than for single primary melanomas (P < .001). Mean (SD) age (56.4 [16.2] years vs 58.8 [15.2] years; P < .11) and median Breslow thickness (0.90 mm [IQR, 0.52-1.80 mm] vs 0.88 mm [IQR, 0.60-1.45 mm]; P < .64) did not differ significantly between the 2 groups, neither did ulceration (71.3% vs 68.0%) or subtype of melanoma (superficial spreading: 2.4% vs 70.6%; nodular: 14.6% vs 13.4%; lentigo maligna: 4.4% vs 4.5%; and acral lentiginous: 0.4% vs 0.8%) (Table 1). Follow-up of 5 years or more was available for 1542 multiple primary melanomas (67.8%) compared with 31 204 single primary melanomas (60.8%).

In 841 of 2284 (36.8%), the second multiple primary melanoma was found in the first year of follow-up after the first melanoma. However, 624 of 2284 second melanomas (27.3%) were found more than 5 years after diagnosis of the first melanoma (Figure and eTable in the Supplement), with a median difference of 2.2 years (IQR, 0.3 years to 5.4 years) and a maximum time of 16.7 years between the first and second melanoma. When stratifying for stage of the first melanoma, second melanomas later than 5 years after diagnosis of the first melanoma were found in 183 first-stage T1a melanomas (24.7%), 128 T1b melanomas (32.2%), 79 T2a melanomas (21.5%), and 17 T2b melanomas (24.6%) (Table 2).

The median Breslow thickness in multiple primary melanomas was 0.90 mm (IQR, 0.55-1.70 mm) for the first melanoma and 0.65 mm (IQR, 0.45-1.10 mm) for the second melanoma (P < .001). No significant differences were found in Breslow thickness between the second and subsequent melanomas (Figure and eTable in the Supplement). When selecting all first and second multiple primary melanomas, 1379 (60.4%) had a decreased Breslow thickness, 82 (3.6%) had the same Breslow thickness, and 823 (36.0%) had an increased Breslow thickness of the second melanoma. Translated to stages, 370 (16.2%) of second melanomas had a higher T stage, 1112 (48.7%) had the same T stage, and 802 (35.1%) had a lower T stage (P < .001) (Table 3).

Because of missing data, a total of 38 816 cases were analyzed: 37 049 single primary melanomas and 1767 multiple primary melanomas. Corrected for all variables, an HR of 1.31 (95% CI, 1.20-1.42) (P < .001) for multiple primary melanomas vs a single primary melanoma was found (Table 4). In multivariable analysis, age per year (HR, 1.06; 95% CI, 1.05-1.06), Breslow thickness per mm (HR, 1.11; 95% CI, 1.10-1.12), and presence of ulceration (HR, 2.20; 95% CI, 2.14-2.25) increased. Females were less likely than males to die at any given time (overall HR, 0.71; 95% CI, 0.67-0.76). Compared with a single primary melanoma, all other types of melanoma had a lower HR, although the HR for lentigo maligna melanoma was not significant. Analysis including pathologic characteristics most likely to be associated with death yielded an HR of 1.12 (95% CI, 1.01-1.24; P = .04).

We found a 4.0% prevalence of multiple primary melanomas among all patients with melanoma in the Netherlands from 2000 through 2014. A total of 36.8% of second multiple primary melanomas were found during the first year of follow-up, and 27.3% of these melanomas were found after 5 years. We observed a decrease in Breslow thickness between subsequent melanomas and that overall survival among patients with multiple primary melanomas was significantly worse compared with that among patients with a single primary melanomas.

The 4.0% observed multiple primary melanoma prevalence that we found is in line with the literature, in which a range of 0.2% to 12.7% has been described.^5-16 Some studies incorporated melanoma in situ in this percentage.^6-8,13 We observed a significant decrease in Breslow thickness between the first and second multiple primary melanomas. Whereas many other studies confirmed this finding,^{5,8,9,12,14,16,32} few assessed the question of why subsequent melanomas tend to have decreased Breslow thickness. It has been argued that patients themselves were likely to detect their first primary melanoma.^17,33 However, little is known about the detection of subsequent primary melanomas. Several studies^18,33,34 reported that subsequent melanomas were most often detected by physicians (up to 94%-95%). Francken et al³⁵ showed that a history of melanoma was not associated with an increase in the ability of patients to detect new primary melanoma themselves. De Giorgi et al³⁶ showed that patients with melanoma who did not attend follow-up visits had significantly increased Breslow thickness for the second melanoma, suggesting that decreased Breslow thickness is associated with dermatologic surveillance.

A total of 36.8% of second multiple primary melanomas were found in the first year of follow-up, but no less than 27.3% of second melanomas were found after 5 years of follow-up, comparable with another study.⁶ Ferrone et al⁸ observed a 59% rate within the first year and Murali et al⁹ found a 58% rate in the first 3 years, but neither reported on the incidence of second melanoma after 5 years. Menzies et al¹³ and Moore et al³² found a median time of 2.8 years and 3.8 years between the first and second multiple primary melanomas, comparable with the 2.2 years that we observed. We stratified the 36.8% of multiple primary melanomas by stage; when selecting only patients with initial T1a melanoma, this percentage stayed almost stable because 24.8% of the second melanomas in this selected group of patients were found after 5 years. Current Dutch and international guidelines recommend discontinuation of follow-up in all patients with melanoma 5 years after the initial diagnosis.³⁷ International follow-up guidelines differ considerably, ranging from no follow-up for pT1a melanoma to lifelong follow-up for all patients.²⁰ Few guidelines comment about patients with multiple primary melanomas. The American Association of Dermatology states that multiple primary melanoma is one of the factors that “may influence follow-up interval.”^38(p22) Other guidelines do not comment on follow-up but instead focus on the hereditary component in patients with multiple primary melanomas.^39,40 The review by Francken et al¹⁷ suggests that no strong evidence exists to support a specific follow-up interval for regular melanoma. Even though the article is from 2005, it reported that most investigators advocated long-term or even lifelong follow-up for patients with multiple melanoma. Follow-up surveillance in patients with melanoma can serve several goals: detect metastases, find a subsequent melanoma, or reassure and educate patients. One could argue that finding a subsequent melanoma is not a main goal, as in the Netherlands, because pT1a melanomas are not incorporated in the follow-up surveillance. Because the risk of metastasis in this group of patients is low, it is not useful to screen these patients for a subsequent melanoma. However, because the prevalence of melanoma has increased and the risk of metastasis and thus death is low, paradoxically, this group of patients may have a high risk of developing a subsequent melanoma.

Several studies have analyzed survival among patients with a single primary melanoma vs that among patients with multiple primary melanomas; 3 found better survival,^16,22,23 1 found similar survival,²¹ and 4 found worse survival among patients with multiple primary melanomas vs patients with a single primary melanoma,^24-27 as we did in the present study. Only 3 studies corrected for immortal time bias of multiple primary melanomas.^21,24,26 Pardo et al²⁴ used Cox time-varying analysis and found an HR of 1.32 (95% CI, 1.17-1.50) among 1210 patients with multiple primary melanomas. Youlden et al²⁶ applied delayed-entry methods and found worse survival among 2330 patients with multiple primary melanomas (HR, 2.01; 95% CI, 1.57-2.59). On the contrary, Grossmann et al²¹ recently analyzed survival data obtained from the Surveillance, Epidemiology, and End Results registry using a 1:1 matching technique to prevent bias and found no survival difference in 887 cases (HR of 1.07 [95% CI, 0.87-1.31] for all melanomas and HR of 0.99 [95% CI, 0.76-1.29] for only invasive melanomas). Our results are in line with Pardo et al²⁴ and Youlden et al.²⁶ We found an HR of 1.31 (95% CI, 1.20-1.42; P < .001) for multiple primary melanomas vs a single primary melanoma, indicating that the hazard of dying of multiple primary melanomas at any given time was 1.31 times higher than the hazard of dying of a single primary melanoma, corrected for all potential confounders included in the analysis.

Because patients with multiple primary melanomas had worse survival than patients with a single primary melanoma in our multivariable analysis, we believe that a patient, once proven to have developed a second melanoma, may benefit from more thorough surveillance. Adjuvant therapies are being developed and studied, and in the near future, their use may extend beyond patients with stage IV disease. We argue that patients with multiple primary melanomas may benefit from being monitored more closely, not only for subsequent melanoma, but especially for metastases.

One of the strengths of this study is that, to our knowledge, we used the largest epidemiologic cohort thus far for answering this research question. We also used the appropriate statistical techniques to assess survival between patients with a single primary melanoma and those with multiple primary melanomas (time-dependent exposures), thus best preventing immortal time bias.^29-31

A limitation is the lack of information about family history because patients with familial atypical mole and melanoma syndrome are known to have multiple primary melanomas and to be younger at the time of diagnosis of their first melanoma. Another possible limitation was that we chose not to include melanoma in situ. Previous studies were not always clear if they included melanoma in situ. We argue that we would have found even more multiple primary melanomas if we had included melanoma in situ, since Leiter et al⁶ showed 33.6% of secondary melanoma to be melanoma in situ. However, since we aimed to analyze differences in Breslow thickness and association with overall survival, 2 features not related to melanoma in situ, we chose not to include noninvasive melanoma. All multiple primary melanoma studies are inherently hampered by the lack of an adequate definition of multiple primary melanoma. Some studies excluded^8,15,32 and 1 study⁹ separately analyzed simultaneous multiple primary melanomas, defined as a second primary melanoma diagnosed within 1 month after the first melanoma. Others defined multiple primary melanoma as 2 or more primary melanomas diagnosed 1 year or further from each other,²¹ but most have not reported their definition. We defined patients with multiple primary melanomas as having another melanoma on or after the date of the first diagnosis of melanoma irrespective of topography. A final limitation was that we did not have data available about immunosuppressive therapy, which is a known risk factor for survival in general and for developing melanoma.⁴¹

In this study, Breslow thickness decreased with subsequent melanomas. A high percentage of second melanomas occurred later than 5 years after the first melanoma, and overall survival among patients with multiple primary melanomas was significantly worse compared with that among patients with a single primary melanoma. The findings suggest that current melanoma follow-up strategies need to be reconsidered for patients with multiple primary melanomas and guidelines should comment on this.

Accepted for Publication: April 6, 2019.

Corresponding Author: Mary-Ann El Sharouni, MD, University Medical Center Utrecht, Heidelberglaan 100, Utrecht 3584 CX, the Netherlands (m.a.elsharouni-2@umcutrecht.nl).

Published Online: June 26, 2019. doi:10.1001/jamadermatol.2019.1134

Author Contributions: Drs El Sharouni and van Diest 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.

Concept and design: El Sharouni.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: El Sharouni, Sigurdsson, van Diest.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: El Sharouni.

Supervision: Witkamp, Sigurdsson, van Diest.

Conflict of Interest Disclosures: None reported.

Additional Contributions: Rinus Voorham, PhD, (The Dutch Nationwide Network and Registry of Histopathology and Cytopathology [PALGA]), provided data, and Paco Welsing, PhD (Julius Center, University Medical Center Utrecht), provided statistical insight. These individuals were not compensated for their contributions.