My NCBI Sign In
Jump to: Authorized Access | Attribution | Authorized Requests

Study Description

The results from a small number of melanoma GWAS have been published. Initial studies identified several pigmentation- and nevus-associated loci that mediate an effect on melanoma risk, however, these studies were somewhat limited in that sample number (hence power) was low, they used first-generation low density SNP arrays, or used pools of DNA samples. Subsequently, additional melanoma GWAS have been completed, including a "Phase 2" study by the International Melanoma Genetics Consortium (GenoMEL), a full GWAS of the Australian sample resource previously used in a pooled-DNA GWAS, and study conducted through the MD Anderson Cancer Center and the Brigham and Women's Hospital. The groups who performed these studies have recently shared data and replicated a number of new melanoma susceptibility loci.

Aside from the identification of novel loci, there are two additional notable outcomes of these studies. Firstly, an examination of the observed versus expected test statistics (Q-Q plot) from the Australian study, even after removing data from SNPs within known replicated susceptibility loci, reveals that there remains an excess of positive results. Similarly, the latest GenoMEL study finds almost three times as many SNPs reaching p-values between 10-4 and 10-5 as would be expected by chance. In the light of other studies of complex traits [1], these data suggest that studies based on additional melanoma samples from the same population are likely to identify an even larger "polygenic tail" and should prove invaluable in identifying and characterizing the underlying loci.

A second observation, is that the effect size for replicated associations differs very little between GWAS. Given the strong effect of ultraviolet radiation (UVR) on risk of melanoma this might be expected to affect penetrance, particularly in the Australian and English samples, which are ethnically similar, but have very different sun exposures. We hypothesize that those SNPs whose effects are strongly mediated by UVR exposure may have smaller marginal effects on risk and are therefore harder to detect initially, suggesting a whole category of genetic effects that are unexplored.Indeed, most melanoma GWAS performed to date used samples from both European and Australian populations in discovery and or replication, and thus, existing studies have been poorly powered to detect associations with a wide difference in effect between these low- and high-UVR populations.

Here, we wish to pursue this hypothesis further and plan to genotype additional cases drawn from the English and Australian populations. This will serve two complementary objectives: firstly, it will greatly increase our collective power to detect additional new melanoma risk alleles; secondly, it will highlight associations that are stronger specifically within the context of either low-UVR or high-UVR exposure environments. We posit that these data will be invaluable, not just for identifying novel melanoma-associated loci, but also for addressing gene-environment interactions.

We have available a collection of valuable resources to help fulfill our aims. These include a considerable number of well-characterized and ungenotyped Australian and UK melanoma cases, as well as existing genotype data from sets of Australian and UK cases and population-based controls.

We propose the following Specific Aims:

  1. To extend our current collective melanoma GWAS by genotyping a large number of additional UK and Australian cases.
  2. To jointly analyze these data and identify new melanoma risk loci.
  3. To determine whether pre-existing and new melanoma risk alleles show different degrees of association between the English and Australian populations, and assess whether these differences might be due to differences in sunlight exposure.

  • Study Type: Case Set
  • dbGaP estimated ancestry components using GRAF-pop
  • Number of study subjects that have individual level data available through Authorized Access: 3912

Authorized Access
Publicly Available Data (Public ftp)

Connect to the public download site. The site contains release notes and manifests. If available, the site also contains data dictionaries, variable summaries, documents, and truncated analyses.

Study Inclusion/Exclusion Criteria

The new case populations to be genotyped by CIDR (Illumina OmniExpress) are described briefly below:

Western Australian Melanoma Health Study (WAMHS): led by Lyle Palmer and Michael Millwall

1274 melanoma cases.

Participants were recruited through the Western Australian Cancer Registry, which is notified of all incident cancers in the state of Western Australia by law. All eligible adult cutaneous melanoma cases diagnosed between January 2006 and September 2009, were contacted by mail and invited to participate. Clinical and phenotypic data and blood samples were collected from consenting cases. Clinical data consisted of all pathological data recorded by the cancer registry and the questionnaire, administered by telephone interview, covered major risk factors for melanoma, such as pigmentation characteristics, naevi and skin type.

University of Leeds, UK: led by Tim Bishop and Julia Newton Bishop

1243 melanoma cases.

Population ascertained incident melanoma cases identified via clinicians, pathology registers and the cancer registry, were recruited from 2000 to December 2006 in a geographically defined area of Yorkshire and the Northern region of the UK.

SEARCH, Cambridge University, UK: led by Doug Easton and Alison Dunning

1395 melanoma cases.

Melanoma cases were recruited through cancer registries in the East Anglia, West Midlands and Trent regions of the UK as part of the SEARCH Programme of the University of Cambridge Cancer Research UK activities. Additional cases (half the sample) have been identified and recruited from the pigmented lesions clinic at Addenbrookes/Hinchingbrooke Hospitals in East Anglia. Cases have provided information on the skin color, hair color, eye color, extent of freckling and naevus count.

Molecular Data
TypeSourcePlatformNumber of Oligos/SNPsSNP Batch IdComment
Whole Genome Genotyping Illumina HumanOmniExpressExome BeadChip N/A N/A
Study History

GWAS of Melanoma and Melanoma-Associated Phenotypes. GWAS by multiple research groups, including those by the investigators of this study, have been successful in identifying genes and chromosomal regions associated with melanoma susceptibility. These data suggest three broad classes of susceptibility loci: those associated with pigmentation, those associated with naevus count, and those mediating melanoma risk via pathways independent of these known risk phenotypes.

Pigmentation genes associated with melanoma: Variants of the melanocortin-1 receptor gene (MC1R) are associated with the combination of red hair, freckling, and sun sensitivity, and have long been implicated in melanoma risk [1-4]. A GWAS by GenoMEL strongly confirmed an association between melanoma risk and MC1R [5], while a second melanoma GWAS using Australian samples established the genomic region harboring ASIP as a novel susceptibility locus (P < 1 x10-15; per-allele OR of 1.75 [1.53, 2.01]) [6]. Evaluation of specific variants in other genes previously identified to be associated with pigmentation via GWAS [5,7-10] has also identified TYR, SLC45A2 and TYRP1 (as well as ASIP) being associated with melanoma [6,9,11,12].

Naevus genes associated with melanoma: The GenoMEL melanoma GWAS [5] also strongly established a locus on chromosome band 9p21 adjacent to MTAP and flanking CDKN2A and CDKN2B (P = 2.60 x 10-11 for rs7023329) as associated with CMM. Independently, a GWAS for naevus count in the UK population identified two loci that replicated in an Australian naevus study: rs4636294 (combined P = 3.4 x 10-15) in the 5'- UTR of MTAP, and rs2284063 (combined P = 3.4 x 10-8) in an intron of PLA2G6 (chromosome band 22q13). Evaluation of these two loci in the Leeds (UK) and Australian melanoma case-control sets demonstrated that both of these naevus loci are also significantly associated with melanoma risk. Based on the Leeds sample, the combined risk for melanoma at both loci was estimated. Subjects homozygous at the two loci have double the risk (OR = 1.94; 95% CI = 1.19 - 3.16) of subjects carrying no risk alleles. More recently, a second naevus GWAS performed on an Australian sample identified a locus on chromosome 6 (close to IRF4), for which evaluation in the Leeds and Australian case-control studies also confirmed an association with melanoma [13].

Melanoma-associated genes independent of known risk phenotypes: More recently, two other melanoma GWAS have been completed and have identified multiple novel replicated susceptibility loci that appear to mediate risk independently of pigmentation and naevus count [14,15]. Firstly, a full GWAS of the Australian-based Queensland study of Melanoma: Environment and Genetic Associations (QMEGA) and Australian Melanoma Family Study (AMFS) sample sets was recently completed (on Illumina Hap660 and Omni1M arrays), identifying two novel susceptibility loci [14]. In total, data from 2,168 melanoma cases from these sources were compared to 4,387 controls, re-identifying several known loci (MC1R, ASIP, MTAP/CDKN2A). The study also identified two novel loci at 1q21.3 and 1q42.12 (rs7412746, P = 2.5 x 10-7; rs3219090, P = 9.5 x 10-7; respectively) that replicated in GWAS conducted in the United States (1,804 cases, 1,026 controls) and Europe (2,804 cases, 7,618 controls, GenoMEL study), as well as via genotyping in an additional un-arrayed set of 585 cases and 6,500 controls from the United States (rs7412746 meta-analysis P = 9.0 x 10-11; rs3219090 meta-analysis P = 9.3 x 10-8). Neither locus appears to be strongly associated with pigmentation or naevus count. The 1q21.3 locus encompasses ten genes including ARNT and MCL1. In contrast the locus at 1q42.12 encompasses a single gene in its entirety, PARP1. Simultaneously, GenoMEL completed a second GWAS and jointly analyzed data from 2,804 GenoMEL melanoma cases and 7,618 controls [15]. The study replicated several known loci, including SLC45A2 and TERT/CLMPT1L (TERT/CLMPT1L was originally identified as a melanoma risk locus in a meta-analysis across multiple cancer types [16,17]), but also identified three additional loci that replicated at P < 10-3 in sample sets from Australia and the United States, including a missense variant in ATM (rs1801516, GenoMEL P = 4.8 x 10-7), a SNP within MX2 (rs45430, GenoMEL P = 5.6 x 10-7), and a SNP adjacent to CASP8 (rs13016963, GenoMEL P = 5.7 x 10-7). Similarly to the two chromosome 1 loci from the Australian GWAS, none of these three loci is associated with pigmentation or naevus phenotypes.

Cumulatively, the multiple low-penetrance risk variants described above fail to explain the majority of heritable melanoma risk in the general population. An unpublished estimate from the GWAS data is that they explain only 10-20% of the excess familial risk. In the Australian study cohort, there remains an excess of positive results in the Q-Q plot after the removal of SNPs located within previously identified melanoma susceptibility regions. Only a very small proportion of this excess is explained by the two novel chromosome 1 regions described here. Likewise, the latest GenoMEL study finds almost three times as many SNPs reaching P-values between 10-4 and 10-5 as would be expected by chance, and the three loci most recently published (ATM, CASP8 and MX2) have powers from 10-20%. This suggests many more susceptibility loci exist on the margins of detectability. In the light of other studies of complex traits [18], these data suggest that studies based on additional melanoma samples are likely to identify an even larger "polygenic tail" and should prove invaluable in identifying and characterizing the underlying loci.

Effect sizes for melanoma-associated loci differ little between the Australian and UK samples. For example, for the novel GenoMEL loci the UK and Australian effect sizes respectively are:

  1. CASP8 rs1093196, UK OR=1.16 (1.03, 1.30), Australian OR=1.09 (0.96, 1.13);
  2. ATM rs1801516, UK OR=0.84 (0.71, 0.98), Australian OR=0.87 (0.85, 1.05); and
  3. MX2 rs45430, UK OR=0.88 (0.79, 0.98), Australian OR=0.91 (0.89, 1.04).

  4. Of loci from the UK naevus GWAS:
  5. PLA2G6 rs6001027, UK OR=1.19 (1.07, 1.34), Australian OR=1.06 (0.95,1.19); and
  6. CDKN2A/MTAP rs7023329, UK OR=1.22 (1.10, 1.27), Australian OR=1.22 (1.11, 1.37). We see here that the confidence intervals of the ORs for Australian and UK samples always overlap.

Given the strong effect of sun exposure on melanoma risk it is surprising that no genetic factor has been found yet that influences risk only in those with either continuously high sun exposure or those with very intermittent sun exposure. Such genetic effects are likely harder to find, because they have a smaller marginal effect. To date, most GWAS have relied on a mix of samples from both European and Australian populations [5] in discovery and/or replication stages, and have thus been poorly powered to detect such loci.


  1. Holly, E.A., et al., Cutaneous melanoma in women. II. Phenotypic characteristics and other host-related factors. Am J Epidemiol, 1995. 141(10): p. 934-42. (PMID: 7741123)
  2. Raimondi, S., et al., MC1R variants, melanoma and red hair color phenotype: a meta-analysis. Int J Cancer, 2008. 122(12): p. 2753-60. (PMID: 18366057)
  3. Kanetsky, P.A., et al., Population-based study of natural variation in the melanocortin-1 receptor gene and melanoma. Cancer Res, 2006. 66(18): p. 9330-7. (PMID: 16982779)
  4. Cress, R.D., et al., Cutaneous melanoma in women: anatomic distribution in relation to sun exposure and phenotype. Cancer Epidemiol Biomarkers Prev, 1995. 4(8): p. 831-6.(PMID: 8634653)
  5. Bishop, D.T., et al., Genome-wide association study identifies three loci associated with melanoma risk. Nat Genet, 2009. 41(8): p. 920-5. (PMID: 19578364)
  6. Brown, K.M., et al., Common sequence variants on 20q11.22 confer melanoma susceptibility. Nat Genet, 2008. 40(7): p. 838-40. (PMID: 18488026)
  7. Sulem, P., et al., Genetic determinants of hair, eye and skin pigmentation in Europeans. Nat Genet, 2007. 39(12): p. 1443-52. (PMID: 17952075)
  8. Han, J., et al., A genome-wide association study identifies novel alleles associated with hair color and skin pigmentation. PLoS Genet, 2008. 4(5): p. e1000074. (PMID: 18483556)
  9. Nan, H., et al., Genetic variants in pigmentation genes, pigmentary phenotypes, and risk of skin cancer in Caucasians. Int J Cancer, 2009. (PMID: 19384953)
  10. Nan, H., et al., Genome-Wide Association Study of Tanning Phenotype in a Population of European Ancestry. J Invest Dermatol, 2009. (PMID: 19340012)
  11. Gudbjartsson, D.F., et al., ASIP and TYR pigmentation variants associate with cutaneous melanoma and basal cell carcinoma. Nat Genet, 2008. 40(7): p. 886-91. (PMID: 18488027)
  12. Fernandez, L.P., et al., SLC45A2: a novel malignant melanoma-associated gene. Hum Mutat, 2008. 29(9): p. 1161-7. (PMID: 18563784)
  13. Duffy, D.L., et al., IRF4 variants have age-specific effects on nevus count and predispose to melanoma. Am J Hum Genet, 2010. 87(1): p. 6-16. (PMID: 20602913)
  14. Macgregor, S., et al. Genome-wide association study identifies a new melanoma susceptibility locus at 1q21.3. Nat Genet., 2011. 43(11):1114-8. (PMID: 21983785)
  15. Barrett, J.H., et al. Genome-wide association study identifies three new melanoma susceptibility loci. Nat Genet., 2011. 43(11): p. 1108-13. (PMID: 21983787)
  16. Stacey, S.N., et al., Common variants on 1p36 and 1q42 are associated with cutaneous basal cell carcinoma but not with melanoma or pigmentation traits. Nat Genet, 2008. 40(11): p. 1313-8.(PMID: 18849993 )
  17. Stacey, S.N., et al., New common variants affecting susceptibility to basal cell carcinoma. Nat Genet, 2009. 41(8): p. 909-14. (PMID: 19578363)
  18. Teslovich, T.M., et al., Biological, clinical and population relevance of 95 loci for blood lipids. Nature, 2010. 466(7307): p. 707-13. (PMID: 20686565)

Selected publications
Diseases/Traits Related to Study (MESH terms)
Links to Related Resources
Authorized Data Access Requests
Study Attribution