Association Between NRAS and BRAF Mutational Status and Melanoma-Specific Survival Among Patients With Higher-Risk Primary Melanoma

Importance  NRAS and BRAF mutations in melanoma inform current treatment paradigms, but their role in survival from primary melanoma has not been established. Identification of patients at high risk of melanoma-related death based on their primary melanoma characteristics before evidence of recurrence could inform recommendations for patient follow-up and eligibility for adjuvant trials.

Objective  To determine tumor characteristics and survival from primary melanoma by somatic NRAS and BRAF status.

Design, Setting, and Participants  A population-based study with a median follow-up of 7.6 years (through 2007), including 912 patients from the United States and Australia in the Genes, Environment, and Melanoma (GEM) Study, with first primary cutaneous melanoma diagnosed in the year 2000 and analyzed for NRAS and BRAF mutations.

Main Outcomes and Measures  Tumor characteristics and melanoma-specific survival of primary melanoma by NRAS and BRAF mutational status.

Results  The melanomas were 13% NRAS+, 30% BRAF+, and 57% with neither NRAS nor BRAF mutation (wildtype [WT]). In a multivariable model including clinicopathologic characteristics, relative to WT melanoma (with results reported as odds ratios [95% CIs]), NRAS+ melanoma was associated with presence of mitoses (1.8 [1.0-3.3]), lower tumor-infiltrating lymphocyte (TIL) grade (nonbrisk, 0.5 [0.3-0.8]; and brisk, 0.3 [0.5-0.7] [vs absent TILs]), and anatomic site other than scalp/neck (0.1 [0.01-0.6] for scalp/neck vs trunk/pelvis), and BRAF+ melanoma was associated with younger age (ages 50-69 years, 0.7 [0.5-1.0]; and ages >70 years, 0.5 [0.3-0.8] [vs <50 years]), superficial spreading subtype (nodular, 0.5 [0.2-1.0]; lentigo maligna, 0.4 [0.2-0.7]; and unclassified/other, 0.2 [0.1-0.5] [vs superficial spreading]), and presence of mitoses (1.7 [1.1-2.6]) (P < .05 for all). There was no significant difference in melanoma-specific survival (reported as hazard ratios [95% CIs]) for melanoma harboring mutations in NRAS (1.7 [0.8-3.4]) or BRAF (1.5 [0.8-2.9]) compared with WT melanoma, as adjusted for age, sex, site, American Joint Committee on Cancer (AJCC) tumor stage, TIL grade, and study center. However, melanoma-specific survival was significantly poorer for higher-risk (T2b or higher stage) tumors with NRAS (2.9 [1.1-7.7]) or BRAF (3.1 [1.2-8.5]) mutations (P = .04) but not for lower-risk (T2a or lower) tumors with NRAS (0.9 [0.3-3.0]) or BRAF (0.6 [0.2-1.7]) (P = .65), as adjusted for age, sex, site, AJCC tumor stage, TIL grade, and study center.

Conclusions and Relevance  Lower TIL grade for NRAS+ melanoma suggests it has a more immunosuppressed microenvironment, which may affect its response to immunotherapies. The approximate 3-fold increased risk of death for higher-risk tumors harboring NRAS or BRAF mutations after adjusting for other prognostic factors compared with WT melanomas indicates that the prognostic implication of these mutations deserves further investigation, particularly in higher–AJCC stage primary melanomas.

Melanomas frequently harbor mutually exclusive BRAF or NRAS mutations that arise early in tumor progression and persist throughout the course of the disease.^1,2 These mutations influence tumor development and maintenance through constitutive activation of the RAS–RAF–mitogen-activated protein (MAP)-kinase kinase (MEK)–extracellular signal-regulated kinases (ERK) pathway.^1,3 Their clinical relevance is underscored by improved survival of patients with stage IV disease with BRAF-mutant melanomas treated with BRAF inhibitors alone or in combination with MEK inhibition.^4-6 These targeted therapies along with new immunotherapies^7,8 are rapidly changing treatment paradigms for metastatic melanoma, and some are under investigation as adjuvant therapies.⁹ Identification of patients at high risk of death from melanoma based on their primary melanoma tumor characteristics before sign of recurrence remains important to inform evidence-based follow-up of patients and adjuvant trials. Equally important is the identification of patients who rarely die of melanoma, as they can be spared the risks of adjuvant therapy. However, it remains unknown whether the primary melanoma NRAS/BRAF mutational status influences survival from melanoma during the natural course of the disease.

To date, studies of NRAS and BRAF mutations in primary melanoma have mostly been retrospective and examined all-cause rather than disease-specific survival.^10-17 Many selected cases based on referral to a particular center,^11-15,17 applied additional criteria such as selection of frozen¹⁶ or metastatic¹⁴ tissues for analysis, or included only nodular¹⁸ or vertical growth phase¹⁹ melanoma. Several studies determined BRAF but not NRAS mutations.^12,17,20 Only 2 studies included more than 1 center and examined NRAS and BRAF mutations in relationship to melanoma-specific survival. Of these, Devitt et al²¹ found that NRAS exon 3 and BRAF V600E mutations translated into worse melanoma-specific survival in a prospective cohort of 249 primary melanoma cases from 2 Australian tertiary melanoma referral centers. Wu et al²² found BRAF V600E mutation to be associated with an unfavorable melanoma-specific survival for 127 primary melanomas diagnosed in women enrolled in the Nurses’ Health Study.

We examined tumor characteristics and melanoma-specific survival by NRAS and BRAF mutation status in 912 incident first primary cutaneous invasive melanomas diagnosed in 2000 in patients from Australia (New South Wales) or the United States (North Carolina, Michigan, and California), who were enrolled in the population-based Genes, Environment, and Melanoma (GEM) Study. The primary melanomas were analyzed for NRAS and BRAF mutations. Our median 7.6-year observation period concluded prior to 2011, when the US Food and Drug Administration and Australian Therapeutic Goods Administration began approving new systemic therapies that improve overall survival in patients with metastatic melanoma.

Box Section Ref ID

The GEM study included patients with single and multiple primary cutaneous melanoma that were diagnosed between 1998 and 2003 from Australia, Canada, Italy, and the United States.^23-27 The institutional review board at the coordinating center, Memorial Sloan Kettering Cancer Center, and each participating institution approved the study protocol. Each study participant provided written informed consent. We sought tumor sections from 1547 participants’ first primary invasive melanoma diagnosed in 2000 from New South Wales (Australia), California, North Carolina, and Michigan.

Histopathology slides were centrally reviewed as previously described.^28,29 Mitoses were defined as present or absent.³⁰ Tumor-infiltrating lymphocyte (TIL) grade was scored as absent, nonbrisk, or brisk using a previously defined grading system.³¹ All data items were available for the T classification describing the state of the primary tumor in the American Joint Committee on Cancer (AJCC) TNM (tumor, regional nodes, distant metastasis) melanoma staging system; data on regional nodal and distant metastases were not available.

Melanoma treatment information was not available; however, the follow-up period at all study centers ended before recent approvals of new systemic agents that alter the natural course of disease.^4-8 Information about deaths from melanoma or other causes was obtained for participants from the National Death Index for the US study centers and the cancer registry for the Australian study center as previously described.²⁸ Patient follow-up for vital status was complete to the end of 2007.

Of 1547 eligible GEM Study participants, 912 (59%) had formalin-fixed, paraffin-embedded melanomas successfully analyzed for NRAS and BRAF mutations. When indicated because of small tumor size or admixture of nonmalignant cells, tumor cells were selectively procured using laser capture microdissection. Tumor DNA was analyzed for BRAF exon 15 (including codon 600) and NRAS exon 2 and 3 (including codons 61, 12, and 13) mutations using single-strand conformational polymorphism (SSCP) analysis and radiolabeled sequencing of SSCP-positive samples as previously described.^32,33 All mutations were confirmed by sequencing an independently amplified DNA fragment to eliminate mutational artifacts. The NRAS/BRAF status of 214 of 218 cases (98%) from North Carolina previously had been reported.³³

BRAF and NRAS mutations were mutually exclusive, and melanomas were grouped as NRAS+ (exon 2 or 3 mutation), BRAF+ (exon 15 mutation), or wildtype (WT) (neither NRAS nor BRAF mutation) for analyses. Pearson χ² tests and Wilcoxon tests were used to compare cases analyzed for NRAS and BRAF mutations with those not analyzed.

To identify factors that independently distinguished NRAS+ or BRAF+ from WT melanoma, a multivariable model was developed that included all clinicopathologic features and study center. We used polytomous logistic regression for this purpose to estimate simultaneously the odds ratios (ORs) and 95% CIs with NRAS+ and BRAF+ compared with WT melanoma, adjusting for study center. Statistical significance was assessed using Wald tests. Linear trend was tested when appropriate using the Wald statistic, with those variables treated as a single ordinal variable. We also report results from a similar model examining the association of NRAS+ and BRAF+ compared with WT melanoma with AJCC tumor stage. Statistical tests were 2 sided, with P < .05 considered statistically significant.

Survival time was accumulated from the diagnosis date until date of death due to melanoma or the end of follow-up (censored patients). Patients were censored at the time of death from any cause other than melanoma. Of the 912 patients who entered the study with first primary melanoma, 40 developed a second primary melanoma during the ascertainment period, and the occurrence of a second primary was included as a time-dependent covariate. The NRAS/BRAF mutational status and pathologic characteristics of their thicker melanoma was used in the survival analysis, as previously published.^28,29

Survival curves by NRAS and BRAF status were visualized using the Kaplan-Meier method and compared using a log-rank test. Hazard ratios (HRs) and 95% CIs by NRAS/BRAF status were estimated in Cox regression models, adjusted for age, sex, study center, and the time-dependent covariate, and then in fully adjusted models that also included anatomic site, TIL grade, and AJCC tumor stage. Scalp/neck and face/ears were included as separate covariates because scalp/neck but not face/ear melanoma predicts worse survival.^34-36 TIL grade was included because higher TIL grade of primary melanoma is associated with better melanoma-specific survival.²⁹ To account for the competing risk of death from other causes, we performed proportional subdistribution hazards regression models according to Fine and Gray³⁷ to assess the effects of covariates on the subdistribution hazard for death as a result of melanoma. The likelihood ratio test was used to test each interaction, comparing a model with the main effects with a model with the main effects and the interaction term, with an a priori α level of .20.³⁸

Tests based on Schoenfeld residuals and graphical methods using Kaplan-Meier curves showed no evidence that proportional hazards assumptions were violated for mutational status. SAS version 9.3 (SAS Institute Inc) was used for all statistical analyses except for Kaplan-Meier curves, which were implemented in STATA/IC 12.1 (StataCorp LP), and the competing risk models, which were analyzed using R (http://www.r-project.org/). The statistical program methods are given in eMethods in the Supplement.

The participants whose tumors were analyzed for NRAS and BRAF mutations (n = 912) were compared with 635 participants whose tumors were unavailable (n = 560), insufficient (n = 43), or failed molecular analysis (n = 32). There were no significant differences (all P > .05) based on median age, sex, site, median Breslow thickness, or death due to melanoma.

Of the 912 participants with NRAS/BRAF mutational status of their first primary invasive melanomas available, 488 (54%) were from Australia and 424 (46%) were from the United States (Table 1). The participants were 54% male, with a median age of 57 years. The median melanoma Breslow thickness was 0.74 mm.

The melanomas were 13% NRAS+, 30% BRAF+, and 57% WT (with neither NRAS nor BRAF mutation (Table 1 and eTable 1 in the Supplement). Of NRAS+ melanomas, 92% harbored mutations in exon 3 and 8% in exon 2; 93% of exon 3 mutations were at codon 61. Of BRAF+ melanomas, 72% carried BRAF V600E, 21% BRAF V600K, and 7% other BRAF exon 15 mutations.

We examined age, sex, and pathologic characteristics comparing NRAS+ and BRAF+ with WT melanoma for the 892 melanomas with complete data for all variables (Table 2). After adjustment for study center, NRAS+ melanoma was significantly associated (P < .05) with each of the pathologic characteristics, but not sex or age; and BRAF+ melanoma was associated (P < .05) with each of the clinicopathologic characteristics, but not sex, ulceration, or TIL grade.

When all clinicopathologic characteristics were included in 1 model adjusted for study center, NRAS+ tumors were significantly associated (P < .05) with anatomic site other than scalp/neck (OR, 0.1 [95% CI, 0.01-0.6] for scalp/neck vs trunk/pelvis), presence of mitoses (OR, 1.8 [95% CI, 1.0-3.3]), and lower TIL grade (ORs for nonbrisk, 0.5 [95% CI, 0.3-0.8]; and brisk, 0.3 [95% CI, 0.5-0.7] [vs absent TILs]). In this model, BRAF+ melanoma was associated with younger age (ORs for ages 50-69 years, 0.7 [95% CI, 0.5-1.0]; and ages >70 years, 0.5 [95% CI, 0.3-0.8] [vs <50 years]), superficial spreading subtype (ORs for nodular, 0.5 [95% CI, 0.2-1.0]; lentigo maligna, 0.4 [95% CI, 0.2-0.7]; and unclassified/other, 0.2 [95% CI, 0.1-0.5] [vs superficial spreading]), and presence of mitoses (OR, 1.7 [95% CI, 1.1-2.6]) (Table 2).

The relationships between NRAS+ and BRAF+ tumors with AJCC tumor stage relative to WT tumors were examined, adjusted for other prognostic factors (age, sex, anatomic site, and TIL grade) and study center (Table 3). NRAS+ and BRAF+ melanomas were each more frequent among higher tumor stages (P values for trend, <.001 and .04, respectively).

There were 62 melanoma deaths in 892 patients with complete AJCC tumor stage and TIL grade information during a median follow-up time of 7.6 years. Five-year survival was 91% (95% CI, 86%-96%) with NRAS+; 95% (95% CI, 93%-98%) with BRAF+; and 95% (95% CI, 94%-97%) with WT melanoma (log-rank test, P = .09) (Figure, A).

In a Cox model adjusted for age, sex, and study center, NRAS+ (HR, 1.8; 95% CI, 0.9-3.4) and BRAF+ (HR, 1.3; 95% CI, 0.7-2.4) relative to WT melanoma were not significantly associated with melanoma-specific survival (P = .19). After further adjustment for anatomic site, tumor stage, and TIL grade, the HR for NRAS+ melanoma was 1.7 (95% CI, 0.8-3.4) and the HR of BRAF+ melanoma increased to 1.5 (95% CI, 0.8-2.9); the results remained nonsignificant (P = .27) (Table 4). In the fully adjusted model, younger age, upper extremities relative to trunk, and lower tumor stage were significantly (P<.05) associated with improved melanoma-specific survival, while scalp/neck site was associated with worse melanoma-specific survival (HR, 2.1; 95% CI, 0.9-5.1) (eTable 2 in the Supplement). We found a significant interaction of NRAS/BRAF mutational status with tumor stage (P value for interaction, .04) but not with age, sex, site, TIL grade, or study center in the full model.

Given the significant interaction with stage, we categorized tumors as in higher-risk (T2b/T3a/T3b/T4a/T4b) and lower-risk (T1a/T1b/T2a) AJCC stages (Table 4).³⁹ In our study, 36 of 144 patients (25%) with higher-risk tumors died of melanoma compared with 26 of 748 patients (3.5%) with lower-risk tumors. For higher-risk tumors, 5-year survival was 73% for NRAS+; 71% for BRAF+; and 82% for WT melanoma (log-rank test, P = .28) (Figure, B). For lower-risk tumors, 5-year survival was 98% for NRAS+; 99% for BRAF+; and 98% for WT melanoma (log-rank test, P = .61) (Figure, C).

For higher-risk tumors adjusted for age, sex, and study center, the HRs were 1.7 (95% CI, 0.8-3.9) for NRAS+ and 2.3 (95% CI, 1.0-5.1) for BRAF+ compared with WT melanoma (P = .13) (Table 4). After further adjustment for anatomic site, tumor stage, and TIL grade, the HRs for NRAS+ and BRAF+ melanoma strengthened to 2.9 (95% CI, 1.1-7.7) and 3.1 (95% CI, 1.2-8.5), respectively, compared with WT melanoma (P = .04). The addition of anatomic site in the model explained the strengthening of the estimates for NRAS and BRAF mutations in the full model. For lower-risk tumors, NRAS/BRAF mutational subtype was not positively associated with hazard of death in either the partially or fully adjusted models. Similar patterns of higher ORs for higher- compared with lower-risk tumors were seen in reanalyses stratified by continent.

In a reanalysis including only NRAS codon 61 and BRAF V600E and WT melanomas, melanoma-specific survival differences based on mutational status remained limited to higher-risk tumors (Table 4).

The associations for melanoma-specific death according to tumor BRAF and NRAS mutational status and clinicopathologic characteristics remained similar in competing risk models (Table 4 and eTable 2 in the Supplement).

We present data from the largest population-based study to date, to our knowledge, analyzing tumor characteristics and melanoma-specific survival by NRAS and BRAF mutational subtypes. Independently of other clinicopathologic characteristics, NRAS+ melanoma was associated with anatomic site other than the scalp/neck, presence of mitoses, and lower TIL grade and BRAF+ melanoma was associated with younger age, superficial spreading subtype, and presence of mitoses. We found no significant difference for the risk of melanoma-related death from NRAS+ or BRAF+ compared with WT melanoma adjusted for other prognostic factors. However, after adjusting for other prognostic factors, there was an approximate 3-fold increase in melanoma-related death for higher-risk (T2b or higher stage) NRAS+ and BRAF+ tumors compared with WT melanoma but not for lower-risk (T2a or lower stage) tumors.

The NRAS and BRAF mutational frequencies, 13% and 30%, respectively, in our study are within previously reported ranges for primary melanoma.^13,21,40 Other studies similarly reported associations of NRAS+ melanoma with older age, trunk and extremity locations, nodular subtype, increased Breslow thickness, and presence of mitoses.^{13,14,21,40,41} We also confirm BRAF+ melanoma associations with younger age, trunk location, superficial spreading melanoma, presence of mitoses, and vertical growth phase.^{11,13,14,21,40-44} Ellerhorst et al¹³ in a hospital-based study similarly found that NRAS+ and BRAF+ melanomas tended to present at a more advanced AJCC tumor stage, while Devitt et al²¹ found that NRAS+ tended to be higher stage.

To our knowledge, no prior study has reported an association of mitoses with NRAS+ and BRAF+ compared with WT melanoma independently of Breslow thickness and other clinicopathologic characteristics. This association may reflect NRAS and BRAF oncogenic activation of the mitogenic RAS-RAF-MEK-ERK pathway.¹ Mitoses are considered as a marker for tumor growth.⁴⁵ Melanoma growth rate, as self-reported by patients, correlates positively with mitotic rate⁴⁶; thus, the association of NRAS+ and BRAF+ melanomas with mitoses suggests that they may grow faster than WT melanomas. It is in agreement with a significant association between either BRAF or NRAS mutation and fast growing melanomas, calculated by using self-reported time on the skin and Breslow thickness.⁴⁷

Similar to our results, NRAS+ melanoma has been identified frequently arising on the trunk⁴⁰ or on the upper^13,14 or lower extremities.²² We further refine this knowledge with our report of an inverse association between NRAS+ melanoma and scalp or neck location; the majority of scalp/neck melanomas in the GEM Study were WT. This finding and the 2-fold worse survival in the GEM Study for scalp/neck melanoma adjusted for mutational subtype indicate that the poor prognosis of scalp/neck melanoma^34-36 is unlikely to be related to NRAS/BRAF mutational status.

To our knowledge, our study is the first to report lower TIL grade for NRAS+ compared with WT melanoma. Notably, TIL grade remained associated with NRAS+ melanoma independently of other factors (age, anatomic site, histologic subtype, and Breslow thickness) that we previously found to be associated with TIL grade in the GEM Study.²⁹ Our observation is plausible because oncogenic RAS pathway activation can disrupt antitumor immunity by decreasing expression of antigen-presenting major histocompatibility complexes on the surface of tumor cells and recruiting immunosuppressive regulatory T cells and myeloid-derived suppressor cells to the tumor site.⁴⁸ Unlike Edlundh-Rose et al,¹⁴ we did not find BRAF+ melanoma, relative to WT melanoma, to be associated with higher lymphocyte infiltration; however, their study design and lymphocyte scoring method differed from those in the GEM Study.

We compare our results with other multisite studies examining melanoma-specific survival by NRAS/BRAF primary melanoma status. Although not reaching statistical significance, our findings of poorer melanoma-specific survival for NRAS+ and BRAF+ (adjusted HRs of 1.7 and 1.5, respectively) compared with WT melanoma are in the same direction found by Devitt et al²¹ for NRAS+ and BRAF+ (adjusted HRs of 2.96 and 1.7, respectively) melanoma despite different study designs and adjustments.²¹ Wu et al²² similarly found that NRAS+ and BRAF+ tumors had shorter melanoma-specific survival than WT melanoma, with BRAF+ compared with WT melanoma reaching statistical significance. Thus, these studies and our results combined indicate a modestly worse prognosis for NRAS+ and BRAF+ tumors overall for melanoma-specific survival.

Our study suggests that melanoma-specific survival differences based on NRAS and BRAF mutational status are limited to higher-risk tumors. Few deaths occurred in lower-risk tumors, and we found no effect of mutational status on survival among lower-risk tumors. Thus, our results provide evidence that NRAS/BRAF mutational status may add prognostic information for higher-risk tumors. A possible explanation for the increased proportion of deaths for NRAS+ and BRAF+ melanoma being limited to higher-risk tumors is that higher-risk tumors may have acquired another contributing genetic alteration during their progression. Our finding, however, requires confirmation. We are not aware of another study that has analyzed survival by NRAS and BRAF status stratified by tumor stage.

Advantages of our study are its large size, use of current AJCC tumor staging, centralized pathology review by expert dermatopathologists, and comparatively long observational period ending before recent approvals of new systemic agents that alter the natural course of disease.^4-8 Any future study examining NRAS and BRAF mutations in primary melanomas in relationship to survival will be confounded by these new treatments.

Our tumor collection and mutational analysis rates of all eligible primary melanomas are similar to or higher than the rates in comparable melanoma studies.^21,22,49-51 Further, our results are representative of the entire population of participants with melanoma enrolled in the GEM study, since we found no significant differences comparing clinicopathologic characteristics of cases with and without mutation analysis. Population-based prevalence estimates of mutations may be useful for budgetary and economic evaluations in present and future pharmacoeconomics studies. Some mutations may have been misclassified, but we minimized this possibility by using laser capture microdissection for all small samples and independently confirming mutations on a separately amplified DNA fragment.

A limitation is that we did not obtain sentinel lymph node status and thus could not determine whether NRAS/BRAF status provides information beyond sentinel lymph node status for outcome prediction. We also did not obtain information regarding potentially used therapies, such as regional radiation, systemic interferon, or clinical trial participation, which could confound our results. Information on relapse was also not available.

Our finding that NRAS+ and BRAF+ melanomas are associated with higher tumor stage at diagnosis indicates that NRAS+ and BRAF+ melanomas are less likely than WT melanoma to be diagnosed when lower risk and surgically curable. The association of NRAS+ melanoma with lower TIL grade may influence its response to immunotherapies. In the GEM Study, the approximate 3-fold increased risk of death for higher-risk NRAS+ and BRAF+ melanomas after adjusting for other prognostic factors compared with WT melanoma indicates that mutational status may be prognostic for this group. This finding could be useful in the identification of patients at high risk of death from melanoma based on their primary melanoma tumor characteristics to inform evidence-based follow-up of patients and determination of eligibility for novel systemic therapy adjuvant trials.

Accepted for Publication: February 24, 2015.

Corresponding Author: Nancy E. Thomas, MD, PhD, Department of Dermatology, University of North Carolina, 413 Mary Ellen Jones Bldg, CB#7287, Chapel Hill, NC 27599 (nthomas@med.unc.edu).

Published Online: April 9, 2015. doi:10.1001/jamaoncol.2015.0493.

Author Contributions: Dr Thomas had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Thomas, Edmiston, Armstrong, Marrett, Gallagher, Rosso, Ollila, Begg,

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

Drafting of the manuscript: Thomas, Anton-Culver, Luo, Ollila, Begg.

Critical revision of the manuscript for important intellectual content: Thomas, Edmiston, Alexander, Groben, Parrish, Kricker, Armstrong, Gruber, From, Busam, Hao, Orlow, Kanetsky, Luo, Reiner, Paine, Frank, Bramson, Marrett, Gallagher, Zanetti, Rosso, Dwyer, Cust, Ollila, Begg, Berwick, Conway.

Statistical analysis: Thomas, Anton-Culver, Kanetsky, Luo, Bramson, Begg.

Obtained funding: Thomas, Kricker, Armstrong, Gallagher, Zanetti, Berwick.

Administrative, technical, or material support: Thomas, Edmiston, Alexander, Groben, Parrish, Kricker, Gruber, From, Busam, Hao, Frank, Marrett, Berwick.

Study supervision: Thomas, Edmiston, Gallagher, Rosso, Berwick.

Conflict of Interest Disclosures: None reported.

Funding/Support: This study was supported by grants R01CA112243, R01CA112524, R01CA112243-05S1, R01CA112524-05S2, U01CA83180, CA098438, R33CA10704339, P30CA016086, P30CA014089, and P30CA008748 from the National Cancer Institute; grant P30ES010126 from the National Institute of Environmental Health Sciences; and a University of Sydney Medical Foundation Program grant (Dr Armstrong).

Role of the Funders/Sponsors: The sponsors had no role in the design and conduct of the study; in the collection, analysis, and interpretation of data; in the preparation of the manuscript; or in the review or approval of the manuscript.

GEM Study Group:Coordinating Center, Memorial Sloan-Kettering Cancer Center, New York, New York: Marianne Berwick, MPH, PhD (principal investigator [PI], currently at the University of New Mexico), Colin B. Begg, PhD (co-PI), Irene Orlow, PhD (co-investigator), Klaus J. Busam, MD (dermatopathologist), Anne S. Reiner, MPH (biostatistician), Pampa Roy, PhD (laboratory technician), Ajay Sharma, MS (laboratory technician), and Emily La Pilla (laboratory technician). University of New Mexico, Albuquerque: Marianne Berwick, MPH, PhD (PI), Li Luo, PhD (biostatistician), Kirsten White, MSc (laboratory manager), and Susan Paine, MPH (data manager). Study centers:The University of Sydney and The Cancer Council New South Wales, Sydney, Australia: Bruce K. Armstrong, MBBS, DPhil (PI), Anne Kricker, PhD (co-PI), Anne E. Cust, PhD (co-investigator); Menzies Research Institute Tasmania, University of Tasmania, Hobart, Australia: Alison Venn, PhD (current PI), Terence Dwyer, MD (PI, currently at University of Oxford, United Kingdom), Paul Tucker, MD (dermatopathologist); British Columbia Cancer Research Centre, Vancouver, British Columbia, Canada: Richard P. Gallagher, MA (PI), Donna Kan (coordinator); Cancer Care Ontario, Toronto, Ontario, Canada: Loraine D. Marrett, PhD (PI), Elizabeth Theis, MSc (co-investigator), Lynn From, MD (dermatopathologist); CPO, Center for Cancer Prevention, Torino, Italy: Roberto Zanetti, MD (PI), Stefano Rosso, MD, MSc (co-PI); University of California, Irvine: Hoda Anton-Culver, PhD (PI), Argyrios Ziogas, PhD (statistician); University of Michigan, Ann Arbor: Stephen B. Gruber, MD, MPH, PhD (PI, currently at University of Southern California, Los Angeles), Timothy Johnson, MD (Director of the Melanoma Program), Duveen Sturgeon, MSN (co-investigator, joint at USC and University of Michigan); University of North Carolina, Chapel Hill: Nancy E. Thomas, MD, PhD (PI), Robert C. Millikan, PhD (previous PI, deceased), David W. Ollila, MD (co-investigator), Kathleen Conway, PhD (co-investigator), Pamela A. Groben, MD (dermatopathologist), Sharon N. Edmiston, BA (research analyst), Honglin Hao (laboratory specialist), Eloise Parrish, MSPH (laboratory specialist), Jill S. Frank, MS (research assistant), David C. Gibbs, BS (research assistant), Jennifer I. Bramson (research assistant); University of Pennsylvania, Philadelphia: Timothy R. Rebbeck, PhD (PI), Peter A. Kanetsky, MPH, PhD (co-investigator, currently at H. Lee Moffitt Cancer Center & Research Institute, Tampa, Florida). Consulting Center:National Centre for Atmospheric Research, Boulder, Colorado: Julia Lee Taylor, PhD, and Sasha Madronich, PhD (UV data consultants).

Correction: This article was corrected on May 14, 2015, to fix an error in wording in the Statistical Methods subsection of the Methods section. In the fourth paragraph, fourth sentence, the word “lower” should haved read “better.” The corrected sentence reads “TIL grade was included because higher TIL grade of primary melanoma is associated with better melanoma-specific survival.”