Kaplan-Meier melanoma-specific survival probabilities by histopathologically amelanotic and pigmented melanoma are shown for patients with melanoma (n = 2736) with median follow-up of 7.6 years (P < .001, log-rank test).
eMethods. Methods of Follow at Individual Study Centers
eTable 1. Relationship Between Histopathologically Amelanotic and Pigmented Melanoma and Age and Breslow Thickness Stratified by Sex for 3207 Primary Melanomas From 2761 Patients
eTable 2. Hazard Ratios for Overall Death According to Histopathologic Pigmentation Among 2736 Patients With Primary Melanomas
Thomas NE, Kricker A, Waxweiler WT, Dillon PM, Busam KJ, From L, Groben PA, Armstrong BK, Anton-Culver H, Gruber SB, Marrett LD, Gallagher RP, Zanetti R, Rosso S, Dwyer T, Venn A, Kanetsky PA, Orlow I, Paine S, Ollila DW, Reiner AS, Luo L, Hao H, Frank JS, Begg CB, Berwick M, for the Genes, Environment, and Melanoma (GEM) Study Group. Comparison of Clinicopathologic Features and Survival of Histopathologically Amelanotic and Pigmented MelanomasA Population-Based Study. JAMA Dermatol. 2014;150(12):1306-1314. doi:10.1001/jamadermatol.2014.1348
Previous studies have reported that histopathologically amelanotic melanoma is associated with poorer survival than pigmented melanoma; however, small numbers of amelanotic melanomas, selected populations, lack of centralized pathologic review, or no adjustment for stage limit the interpretation or generalization of results from prior studies.
To compare melanoma-specific survival between patients with histopathologically amelanotic and those with pigmented melanoma in a large international population-based study.
Design, Setting, and Participants
Survival analysis with a median follow-up of 7.6 years. The study population comprised 2995 patients with 3486 invasive primary melanomas centrally scored for histologic pigmentation from the Genes, Environment, and Melanoma (GEM) Study, which enrolled incident cases of melanoma diagnosed in 1998 through 2003 from international population-based cancer registries.
Main Outcomes and Measures
Clinicopathologic predictors and melanoma-specific survival of histologically amelanotic and pigmented melanoma were compared using generalized estimating equations and Cox regression models, respectively.
Of 3467 melanomas, 275 (8%) were histopathologically amelanotic. Female sex, nodular and unclassified or other histologic subtypes, increased Breslow thickness, presence of mitoses, severe solar elastosis, and lack of a coexisting nevus were independently associated with amelanotic melanoma (each P < .05). Amelanotic melanoma was generally of a higher American Joint Committee on Cancer (AJCC) tumor stage at diagnosis (odds ratios [ORs] [95% CIs] between 2.9 [1.8-4.6] and 11.1 [5.8-21.2] for tumor stages between T1b and T3b and ORs [95% CIs] of 24.6 [13.6-44.4] for T4a and 29.1 [15.5-54.9] for T4b relative to T1a; P value for trend, <.001) than pigmented melanoma. Hazard of death from melanoma was higher for amelanotic than for pigmented melanoma (hazard ratio [HR], 2.0; 95% CI, 1.4-3.0) (P < .001), adjusted for age, sex, anatomic site, and study design variables, but survival did not differ once AJCC tumor stage was also taken into account (HR, 0.8; 95% CI, 0.5-1.2) (P = .36).
Conclusions and Relevance
At the population level, survival after diagnosis of amelanotic melanoma is poorer than after pigmented melanoma because of its more advanced stage at diagnosis. It is probable that amelanotic melanomas present at more advanced tumor stages because they are difficult to diagnose. The association of amelanotic melanoma with presence of mitoses independently of Breslow thickness and other clinicopathologic characteristics suggests that amelanotic melanomas might also grow faster than pigmented melanomas. New strategies for early diagnosis and investigation of the biological properties of amelanotic melanoma are warranted.
The American Cancer Society has estimated that 76 100 new invasive melanomas will be diagnosed and 9710 people will die of melanoma in 2014 in the United States.1 Despite newly available targeted and immunomodulatory agents,2- 5 systemic therapies rarely lead to cures. Thus, early detection of primary melanomas followed by surgical excision remains critical for good outcomes. While recognition of all melanomas can be difficult, lack of brown or black color in amelanotic melanoma at the time of clinical diagnosis removes a defining characteristic for melanoma identification. Several studies indicate that amelanotic melanoma is associated with adverse survival because the tumors are more advanced at diagnosis; however, the majority of studies included only small numbers of amelanotic melanoma cases entered at single study sites and used a clinical, not pathologic, definition of amelanotic melanoma or lacked centralized pathologic review.
Amelanotic melanoma has been defined differently between studies, either clinically as melanomas devoid of pigment on visual inspection before biopsy6- 8 or lacking melanin pigment in melanoma cells on routine hematoxylin-eosin (HE)–stained sections.9- 11 Approximately 2% to 20% of melanomas have been classified as amelanotic.6- 11 Amelanotic melanomas can be found among all histologic subtypes, including superficial spreading, nodular, lentigo maligna, and acral lentiginous melanoma.6,12
Of all investigations using either a clinical or histopathologic definition, we are aware of only 3 studies that included more than 100 amelanotic cases.11,13,14 Moreau et al,14 using Surveillance, Epidemiology, and End Results (SEER) registry data, found that amelanotic melanoma was more advanced at diagnosis and more lethal than pigmented melanoma. However, misclassification of histolopathogically amelanotic and pigmented melanomas could occur in SEER data. SEER extracts the term amelanotic from pathology reports and codes it under an International Classification of Diseases for Oncology, Third Edition morphology code, but the SEER registry allows documentation of only 1 morphology code.14,15 Thus, melanomas coded under a different morphology code (such as superficial spreading melanoma, nodular melanoma, lentigo maligna, or acral lentiginous melanoma) would not be coded as amelanotic. In addition, pathologists may not report “amelanotic” on pathology reports, leading to missing data.
The 2 other studies with more than 100 amelanotic cases conducted centralized pathologic review of the melanomas.11,13 In a Norwegian population-based study, Larsen et al13 found that overall survival of patients was significantly worse for amelanotic than for pigmented melanoma, but these results were not adjusted for other melanoma characteristics. In a Danish hospital-based study, Søndergaard et al11 found that patients with amelanotic melanoma did not have poorer melanoma-specific survival once other clinicopathological characteristics including Breslow thickness were taken into account; however, generalization of the findings from a single hospital study would be limited.
We examined histopathologically diagnosed amelanotic and pigmented melanomas in the international population-based Genes, Environment, and Melanoma (GEM) Study of patients with single primary melanomas (SPMs) and multiple primary melanomas (MPMs).16- 19 Expert pathologists reviewed and scored the histologic features, including pigmentation, of 3486 invasive primary melanomas diagnosed in 1998 through 2003 from 2995 GEM Study patients from Australia, the United States, Italy, and Canada. We used a precise definition of histologic pigmentation; melanomas were recorded as histopathologically amelanotic if on light microscopic examination of HE-stained sections no melanin granules were seen in the cytoplasm of the tumor cells. Our goals were to more fully characterize the relationship of histopathologic pigmentation with clinicopathologic features and current American Joint Committee on Cancer (AJCC) tumor stage20 and to compare melanoma-specific and overall survival between patients with histopathologically amelanotic and pigmented melanoma.
The institutional review board at the coordinating center (Memorial Sloan-Kettering Cancer Center) and each participating institution approved the study protocol. Physician approval was sought before contacting eligible participants. All study participants provided written informed consent, including for obtaining diagnostic slides of their melanoma(s) for centralized review.
The GEM Study population included incident primary cutaneous melanomas notified to population-based cancer registries in Australia, Canada, Italy, and the United States.16,17,21- 23 Patients with SPM were diagnosed in 2000 and those with MPM were diagnosed with a second or higher-order invasive or in situ melanoma in 1998 through 2003. We included incident melanomas (SPMs and index MPMs), and, for patients with MPM, also ascertained the previous (usually the first) melanoma (previous MPM) in local cancer registry records. In situ melanomas were eligible as index MPMs when the patient had a previous invasive melanoma.
In the GEM Study, there were 3578 participants with a total of 4784 primary cutaneous melanomas. This analysis excluded in situ melanomas (n = 302) because the aims were to determine the association of histopathologic pigmentation with clinical and pathologic features of and survival from invasive melanomas. The analyses reported herein included only primary invasive melanomas for which the diagnostic slides were available for review and centrally scored for histopathologic pigmentation—a total of 3486 of 4482 primary invasive melanomas (78%) from 2955 of 3578 GEM Study participants (82%). They comprised 2007 of 2372 index SPMs (85%), 716 of 904 index MPMs (79%), and 763 of 1206 previous MPMs (63%). The 716 index MPMs and 763 previous MPMs occurred in 948 of 1206 patients with MPM (79%), among whom 185 received a pathologic review for only the index MPM, 232 for only the previous MPM, and 531 for both.
Patient age, sex, and melanoma body site were extracted from pathology reports and confirmed during patient interview; histologic subtype and Breslow thickness were also extracted from pathology reports. Centralized review of the melanoma HE-stained slides recorded histologic subtype, Breslow thickness, pigmentation, mitoses, ulceration, tumor infiltrating lymphocytes, adjacent solar elastosis, and coexisting nevus. Melanomas were classified according to previously reported criteria.24,25 Mitoses were defined as present or absent.26
Melanomas were recorded as histopathologically amelanotic if on light microscopic examination of HE-stained sections no melanin granules were seen in the cytoplasm of the tumor cells. In a test set of 19 sections scored for melanin pigmentation by the 3 dermatopathologists who reviewed the GEM melanomas, the kappa (κ) statistic for agreement between the pathologists was 0.48, which indicates moderate agreement.
From 1 study center (North Carolina), we extracted prebiopsy impression of lesional (“clinical”) pigmentation from the pathology reports. Clinical pigmentation was recorded on the pathology reports for only 23% (64 of 274) of the melanomas. Melanomas described as tan, brown, blue, gray, black, or hyperpigmented were grouped as “clinically pigmented,” while melanomas noted as pink, red, white, or amelanotic were grouped as “clinically amelanotic.” Of 60 clinically pigmented melanomas, 57 (95%) were also histopathologically pigmented (as determined by centralized pathologic review), while 4 of 5 clinically amelanotic melanomas (80%) were histopathologically amelanotic (P < .001, Fisher exact test). As clinical pigmentation was often missing on the pathology reports but was associated with histopathologic pigment, we chose histopathologic pigmentation from centralized pathologic review for all analyses. The North Carolina cancer registry ascertained these population-based cases, which were originally diagnosed by multiple providers across North Carolina. We did not have access to medical chart notes to abstract prebiopsy melanoma color descriptions.
All data items were available for the T classification describing the state of the primary tumor in the AJCC TNM (tumor, regional nodes, distant metastasis) melanoma staging system; data on regional nodes and distant metastases were not available.
Melanoma treatment information was not available; however, the follow-up period at all study centers ended before US Food and Drug Administration, Health Canada, European Union, and Australian Therapeutic Goods Administration approvals of CTLA-4, BRAF, and MEK inhibitors for treatment of metastatic melanoma.
Information about deaths from melanoma or other causes was obtained for all participants from National Death Indexes, cancer registries, and municipal records. Patient follow-up for vital status finished at the end of 2007 in most centers and at the end of 2008 in British Columbia and Italy. See eMethods in the Supplement for additional follow-up information.
The association of clinical and pathologic characteristics with amelanotic and pigmented melanoma was examined including all single and multiple (both index and previous) primary melanomas. We used marginal logistic regressions with an independent correlation structure implemented in generalized estimating equations to account for the clustering of melanomas for patients with MPM. All models included study center and lesion status (SPM, index MPM, previous MPM)—the design variables. To identify factors that independently distinguished amelanotic from pigmented melanoma, a multivariable model was developed that included all clinicopathologic features and study design variables (study center and lesion status [SPM, index MPM, or previous MPM]). Statistical significance was evaluated based on Wald tests. We also report results from similar models examining the association of amelanotic vs pigmented melanoma with AJCC tumor stage. Linear trend was tested using the Wald statistic when AJCC tumor stage was treated as a single ordinal variable.
Previous analyses of GEM Study data required the inclusion of an age by sex interaction term because of the higher population incidence of melanoma in younger women than in men but lower incidence in women than in men at older ages.17 In this report, we tested for the presence of such an interaction by adding an age by sex interaction term to the model. We also examined the relation of pigmentation to Breslow thickness and age categories stratified by sex using Pearson χ2 tests.
Survival by amelanotic and pigmented melanoma was examined in all patients, including all patients with SPM and MPM. For patients with MPM with review data for both the index and a previous MPM, we used the pathologic characteristics of the tumor with the greatest Breslow thickness in the analysis. When thickness was the same for both melanomas, we used the characteristics of the index MPM, and, if 1 MPM had thickness missing (n = 30), the pathologic characteristics of the other were used. Because the parent study involved population-based ascertainment of incident SPM and MPM, survival time was accumulated from the diagnosis date of the index lesion, whether SPM or MPM. The end point was date of death due to melanoma or the end of complete follow-up (censored patients). For melanoma-specific survival, patients were censored at the time of death from any cause other than melanoma.
Survival curves by lesion pigmentation were constructed using the Kaplan-Meier method and compared using a log-rank test. Hazard ratios (HRs) and 95% CIs for melanoma-specific survival by pigmentation were estimated in Cox regression models. The timescale used in the Cox regression models was follow-up time, adjusting for baseline age as a covariate. For 96 participants enrolled as patients with SPM who experienced a subsequent melanoma during the period of participant recruitment, a time-dependent covariate for MPM status was included in the model at the date of diagnosis of a second melanoma. An initial Cox model was adjusted for age, sex, site, study center, and whether SPM or MPM and another model also included AJCC tumor stage. A separate fully adjusted Cox model for overall survival by amelanotic compared with pigmented melanoma is also presented herein.
Tests based on Schoenfeld residuals and graphical methods using Kaplan-Meier curves in STATA/IC 12.1 (StataCorp LP) showed no evidence that proportional hazards assumptions were violated for pigmentation. All significance tests were 2 sided. SAS version 9.3 (SAS Institute Inc) statistical software was used for all analyses except for Kaplan-Meier curves, which were implemented in in STATA/IC 12.1.
Overall, 275 of the melanomas (8%) were amelanotic (Table 1). We examined associations of the clinicopathologic characteristics with amelanotic melanoma for 3207 SPMs and MPMs with complete data for all variables of interest (Table 2). Notably, the median thickness of amelanotic melanomas (1.60 mm) was much greater than pigmented melanomas (0.68 mm). We observed some differences between the sexes depending on age and Breslow thickness. More of the melanomas in men older than 70 years were amelanotic (10%) compared with men between the ages of 50 to 69 years (7%) or men younger than 50 years (4%) (P = .007), while in women 8% to 10% of melanomas were amelanotic in each age group (P = .55) (eTable 1 in the Supplement). Amelanotic melanoma was associated (P < .001) with Breslow thickness greater than 2.00 mm in each sex, but a higher percentage of men had thicker amelanotic melanomas (>2 mm in 48%) compared with women (>2 mm in 35%) (P = .05).
Age and each of the clinicopathologic characteristics, but not sex, were associated (P < .05) with amelanotic melanoma when adjusted only for the design variables—study center and lesion status (SPM, index MPM, or previous MPM) (Table 2). When all variables were included in 1 fully adjusted model, the variables independently associated (P < .05) with amelanotic melanoma were female sex, nodular and unclassified or other histologic subtypes, increased Breslow thickness, presence of mitoses, severe solar elastosis, and lack of a coexisting nevus.
The odds ratio (OR) for amelanotic melanoma in women relative to men was significantly increased in the fully adjusted model (OR, 1.4; 95% CI, 1.0-2.0) but not in the model adjusted only for study design variables (OR, 1.1; 95% CI, 0.8-1.4), an effect accounted for by the inclusion of thickness in the fully adjusted model. Ulceration and tumor-infiltrating lymphocytes each became nonsignificant in the fully adjusted model due to addition of thickness to the model. The age by sex interaction term was not significant when added to the fully adjusted model (P value for interaction, .22), and no OR in the model changed by more than 10%; thus, we did not include the interaction term in the models.
We examined in more detail the association of amelanotic melanoma with individual AJCC tumor stage adjusted for factors known to be associated with survival (sex, age, and anatomic site) along with the design variables (Table 3). Melanomas with higher AJCC tumor stage were more likely to be amelanotic, with ORs (95% CIs) between 2.9 (1.8-4.6) and 11.1 (5.8-21.2) for tumor stages between T1b and T3b and ORs (95% CIs) of 24.6 (13.6-44.4) for T4a and 29.1 (15.5-54.9) for T4b relative to T1a (P value for trend <.001).
There were 208 melanoma deaths in 2736 GEM study patients with complete AJCC tumor stage information; the median follow-up time was 7.6 years. Kaplan-Meier survival curves show 5-year melanoma-specific survival of 88% (95% CI, 84%-92%) in amelanotic and 95% (95% CI, 94%-96%) in pigmented melanoma (P < .001, log-rank test) (Figure). The HR for melanoma death was 2.0 (95% CI, 1.4-3.0; P < .001) for amelanotic relative to pigmented melanoma in a Cox regression model adjusted for age, sex, anatomic site, study center, and whether SPM or MPM (Table 4). However, the addition of AJCC tumor stage to this model removed the association of amelanotic relative to pigmented melanoma with melanoma-specific survival (HR, 0.8; 95% CI, 0.5-1.2) (P = .36).
In a reanalysis of SPMs alone, death from melanoma remained higher for amelanotic than pigmented melanoma (HR, 2.4; 95% CI, 1.5-3.8) (P < .001) adjusted for age, sex anatomic site, and study design variables, but survival did not differ once AJCC tumor stage was also taken into account (HR 0.8; 95% CI, 0.5-1.2) (P = .26) (not shown in Tables).
Hazard of death from all causes was higher for histopathologically amelanotic than pigmented melanoma (HR, 1.6; 95% CI, 1.2-2.1), adjusted for age, sex anatomic site, and study design variables (P = .001; not shown in Tables); however, survival did not differ once AJCC tumor stage was also taken into account (HR, 0.9; 95% CI, 0.7-1.2) (P = .44) (eTable 2 in the Supplement).
Female sex, nodular and unclassified or other subtypes, increased Breslow thickness, presence of mitoses, severe solar elastosis, and lack of a coexisting nevus were independently associated with histopathologically amelanotic melanoma in the international population-based GEM Study. Amelanotic melanomas were more frequent among melanomas with higher AJCC stage at diagnosis. Melanoma-specific fatality was higher for amelanotic compared with pigmented melanoma but not once AJCC tumor stage was taken into account.
The 8% frequency of histopathologic amelanotic melanoma in our study is within the range of the 2% to 20% of melanomas previously reported as histopathologically amelanotic.7,9- 11,13,14 Other studies have found amelanotic melanoma to be associated with older age,6,14 head/neck site,14 and sun-damaged skin,27 as we did. However, solar elastosis, which we have shown to be associated with cumulative site-specific ambient erythemal UV exposure,28 was a stronger predictor of amelanotic melanoma in the GEM Study than age or site. Like us, others have reported that nodular melanomas6,13,29 were more likely to be amelanotic. We contribute to the literature that unclassified and histological types other than nodular, superficial spreading, or lentigo maligna melanoma were independently associated with amelanotic melanoma. Further, we found lack of a coexisting nevus to be independently associated with amelanotic melanoma also, apparently, for the first time.
Other studies have variably reported predilection of amelanotic melanoma for women,6,7 men,14,30 or neither sex.31 In unadjusted analyses, we observed that younger (<50 years) women were more likely than men of a similar age to have amelanotic than pigmented melanoma, but there was little difference at older (≥50 years) ages. Men generally had thicker melanomas. Being female increased the likelihood that a melanoma was an amelanotic melanoma but only after taking into account Breslow thickness and age. We speculate that women on the whole might self-refer for suspicious lesions that prove to be amelanotic melanomas more often than men do.
As we found, amelanotic melanoma has been reported as positively associated with increased Breslow thickness10,14,29 and presence of mitoses in other studies.13 Notably, Liu et al32 reported rapid rate of melanoma growth, which they found correlated moderately to strongly with mitotic rate, to be associated with histopathologically amelanotic melanoma. We report that the association of mitoses with amelanotic melanoma is independent of Breslow thickness and other clinicopathologic characteristics. Because mitoses are generally considered as a marker for tumor growth,33 our finding suggests that amelanotic melanomas grow faster than pigmented melanomas.
Previous studies variably found ulceration to be more frequent in14,34 or not associated with6 amelanotic melanoma. In the GEM Study, ulceration and higher tumor-infiltrating lymphocyte (TIL) grade were each positively associated with amelanotic melanoma in GEM Study patients but not after accounting for thickness. No study that we are aware of has examined TIL grade in relationship with histopathologic pigmentation of melanoma; however, amelanotic melanomas have been reported to frequently retain pigment cell differentiation antigens.30,35,36 We speculate that this may render histopathologically amelanotic and pigmented melanomas similarly antigenic and lead to comparable lymphocytic infiltrates.
Like the GEM Study, one other large study that included more than 100 amelanotic melanomas and conducted centralized pathologic review found no difference in melanoma-specific survival for histopathologically amelanotic vs pigmented melanoma after accounting for their generally more advanced tumor stage.11 It is probable that amelanotic melanomas tend to present at a more advanced tumor stage because they are more difficult to diagnose. A high rate of clinical misdiagnosis for amelanotic melanoma has been reported.6,27,37 Lack of melanin granules in histopathologically amelanotic melanoma could affect use of the color criterion in the ABCDE (asymmetry, border irregularity, color variegation, diameter ≥6 mm, and evolution) algorithm38 and affect early recognition. One group attempted to measure diagnostic delay using a clinical, not pathologic, definition of amelanotic melanoma. Betti et al39 reported a greater delay in amelanotic melanoma diagnosis due principally to physician diagnostic delay, but the same group29 subsequently reported no significant diagnostic delay. Additional studies could clarify the issues surrounding delays in diagnosis of amelanotic melanoma.
We are not aware of another international study comparing survival from amelanotic and pigmented melanoma, nor has another population-based study examined survival by pigmentation of centrally reviewed melanomas while taking into account tumor stage. Other advantages of our study include its large size, collection of information about cause of death, objective histopathologic definition of tumor pigmentation, use of current AJCC tumor staging, and long observational period ending before recent US Food and Drug Administration approvals of new agents,2- 5 which could differentially alter the natural course of disease. As BRAF-mutant melanoma has been reported to be associated with histologic pigmentation,40 future survival studies of amelanotic melanoma in relationship with survival could be confounded by treatments targeting the BRAF-MEK pathway.2,4,5
A study weakness is that histopathologically amelanotic and pigmented melanoma may be misclassified for some cases; the interobserver agreement for scoring of histopathologic pigmentation was moderate (κ = 0.48). Others have also noted difficulty grading pigmentation,13 although we are not aware of any prior interobserver studies for histopathologic pigmentation. Another limitation is that we were unable to access prebiopsy color descriptions of the melanomas from provider charts.
The poorer melanoma-specific and overall survival for histopathologically amelanotic compared with pigmented melanoma in the GEM Study was entirely due to higher tumor stage at diagnosis of amelanotic melanomas. It is very likely that amelanotic tumors presented at a higher AJCC tumor stage because of delayed diagnosis. The positive association of mitoses with amelanotic melanoma independent of tumor thickness and other clinicopathologic characteristics suggests that amelanotic melanomas may grow more rapidly than pigmented melanomas. If this is the case, amelanotic melanomas may not only be more difficult to diagnose, but the window of opportunity for early diagnosis might be smaller. Although amelanotic melanoma was associated with severe solar elastosis, it occurred on all body sites, supporting complete skin examinations for its detection. Studying a subset of cases, we found that “clinical” pigmentation was associated with “histopathologic” pigmentation. Future research is needed to identify the best methods for diagnosis of amelanotic melanoma. Studying the biological properties of amelanotic melanoma could shed light on the possibility that it has a faster growth rate than pigmented melanoma.
Accepted for Publication: May 14, 2014.
Corresponding Author: Nancy E. Thomas, MD, PhD, Department of Dermatology, University of North Carolina, 405 Mary Ellen Jones Bldg, CB#7287, Chapel Hill, NC 27599 (email@example.com).
Published Online: August 27, 2014. doi:10.1001/jamadermatol.2014.1348.
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, Kricker, Waxweiler, Dillon, Busam, Armstrong, Marrett, Gallagher, Begg.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Thomas, Kricker, Waxweiler, Dillon, Busam, Armstrong, Ollila, Reiner, Hao, Begg.
Critical revision of the manuscript for important intellectual content: Thomas, Kricker, Waxweiler, Dillon, From, Groben, Armstrong, Anton-Culver, Gruber, Marrett, Gallagher, Zanetti, Rosso, Dwyer, Venn, Kanetsky, Orlow, Paine, Ollila, Reiner, Luo, Frank, Begg, Berwick.
Statistical analysis: Thomas, Kricker, Waxweiler, Dillon, Kanetsky, Luo, Begg.
Obtained funding: Thomas, Kricker, Anton-Culver, Gallagher, Venn, Berwick.
Administrative, technical, or material support: Thomas, Waxweiler, Dillon, Busam, From, Groben, Gruber, Marrett, Gallagher, Zanetti, Rosso, Venn, Orlow, Paine, Hao, Frank, Berwick.
Study supervision: Thomas, Gallagher.
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 (coinvestigator), Klaus J. Busam, PhD (dermatopathologist), Anne S. Reiner, MPH (biostatistician), Pampa Roy, PhD (laboratory technician), Ajay Sharma, MS (laboratory technician), Emily La Pilla (laboratory technician). University of New Mexico, Albuquerque: Marianne Berwick, MPH, PhD (PI), Li Luo, PhD (biostatistician), Kirsten White, MSc (laboratory manager), Susan Paine, MPH (data manager).
Study centers included the following: 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 (coinvestigator); Menzies Research Institute Tasmania, University of Tasmania, Hobart, Australia: Alison Venn, PhD (current PI), Terence Dwyer, PhD (PI, currently at International Agency for Research on Cancer, Lyon, France), Paul Tucker, PhD (dermatopathologist); British Columbia Cancer Research Centre, Vancouver, Canada: Richard P. Gallagher, MA (PI), Donna Kan (coordinator); Cancer Care Ontario, Toronto, Canada: Loraine D. Marrett, PhD (PI), Elizabeth Theis, MSc (coinvestigator), Lynn From, PhD (dermatopathologist); CPO, Center for Cancer Prevention, Torino, Italy: Roberto Zanetti, MD (PI), Stefano Rosso, PhD (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, PhD (Director of Melanoma Program), Shu-Chen Huang, MS, MBA (coinvestigator, joint at USC-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, PhD (coinvestigator), Kathleen Conway, PhD (coinvestigator), Pamela A. Groben, PhD (dermatopathologist), Sharon N. Edmiston, BA (research analyst), Honglin Hao (laboratory specialist), Eloise Parrish, MSPH (laboratory specialist), Jill S. Frank, MS (research assistant); University of Pennsylvania, Philadelphia: Timothy R. Rebbeck, PhD (PI), Peter A. Kanetsky, MPH, PhD (coinvestigator); UV data consultants: Julia Lee Taylor, PhD and Sasha Madronich, PhD, National Centre for Atmospheric Research, Boulder, Colorado.
Conflict of Interest Disclosures: None reported.
Funding/Support:National Cancer Institute (NCI) grants R01CA112243, R01CA112524, R01CA112243-05S1, R01CA112524-05S2, CA098438, U01CA83180, R33CA160138, and P30 CA014089; National Institute of Environmental Health Sciences (P30ES010126); University of Sydney Medical Foundation Program grant (Bruce Armstrong); Michael Smith Foundation for Health Research Infrastructure Award (Richard Gallagher).
Role of Funder/Sponsor: The sponsors had no role in the design and conduct 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.