Malignant melanoma, the least common but most deadly form of skin cancer, arises in melanocytes: the cells that produce the dark skin pigment, melanin. Skin exposure to ultra-violet light is the most important risk factor, and fair-skinned people, particularly those living in areas where sunlight is intense, are at greatest risk.

There is much interest in identifying genetic variants associated with melanoma risk, but so far only a few low penetrance variants have been discovered. Most of these are known to be involved in skin pigmentation and the formation of naevi (moles), and have relatively high odds ratios of 1.5 or above.

Two genome-wide association studies published back-to-back in the 9 October issue of Nature have now revealed four further genetic loci associated with melanoma risk; interestingly, none of these has any association with skin colour or naevus formation.

One large study led by David Bishop and his colleagues from Leeds Cancer Research UK Centre, Leeds, UK on behalf of the international GenoMEL research consortium involved analyzing data from 2,804 melanoma patients and 7,618 unaffected controls genotyped at either approximately 300,000 or, later, approximately 600,000 positions.

All melanoma patients had a family history of the disease or disease characteristics, such as early onset, suggestive of genetic susceptibility. All participants were of European ancestry. Initial findings confirmed associations with known genetic risk factors including the gene SLC45A2, which is known to be involved in melanosome maturation and pigmentation.

Bishop and his co-workers also identified seven new genetic regions with at least some evidence of linkage with melanoma susceptibility, three of which, on chromosomes 2, 11 and 21, showed particularly strong evidence of replication with p-values below 10-3. The p-value for the combination of these three genes was 5×10-8.

In each case, the SNP involved was located within a specific gene. These are the CASP8 gene on chromosome 2, which encodes a protease involved in apoptosis; the ATM gene on chromosome 11, which repairs double-stranded breaks in DNA; and the MX2 gene on chromosome 21, which binds GTP and which has not previously been associated with the risk of any cancer.

The researchers point out that the power of their study is low, which indicates that the regions where linkage is only suggested here might be validated in larger studies, and suggest that the identification of these novel genes might indicate mechanisms involved in melanoma formation that do not involve pigmentation.

The second large, international study was led by Stuart Macgregor at Queensland Institute of Medical Research, Brisbane, Australia on behalf of the Lund Melanoma Group. A total of 2,168 melanoma patients and 4,387 non-affected controls were genotyped on Illumina SNP arrays; after imputation from the 1000 Genomes Study the dataset included information on almost 5,500,000 nucleotide positions. All individuals genotyped were Australians of European origin.

This study confirmed the association of three known melanoma susceptibility loci, in the genes MC1R, ASIP and MTAP–CDKN2A, and identified two new potential susceptibility loci, both on chromosome 1 (1q42.12 and 1q21.3). The genotypes at these and another seven loci were analysed using, additionally, data from two large case-control studies from Europe and the USA. One SNP in region 1q21.3, rs7412746, could clearly be associated with melanoma susceptibility; an association with rs3219090, in region 1q42.12, was weaker but still interesting.

These variants were calculated to each explain about 0.1% of total melanoma risk. The SNP rs7412746 was very weakly associated with several pigmentation-related traits, including eye colour and naevus count; the SNP rs3219090 was not associated with pigmentation at all. The DNA region containing rs741246 encodes ten genes, nine of which are expressed in normal melanocytes and several of which have previously been implicated in cancer.

In contrast, the region containing rs3219090 encodes a single gene, PARP1. This gene encodes a DNA repair protein, poly(ADP-ribose) polymerase 1, which is a known cancer drug target.

References

Barrett, J.H., Iles, M.M., Harland, M. and 72 others, on behalf of the GenoMEL Consortium. Genome-wide association study identifies three new melanoma susceptibility loci. Nature, published online ahead of print 9 October 2011. doi:10.1038/ng.959

MacGregor, S., Montgomery, G.W., Liu, J.Z. and 49 others. Genome-wide association study identifies a new melanoma susceptibility locus at 1q21.3. Nature, published online ahead of print 9 October 2011. doi:10.1038/ng.958