The Quantitative Genetics of Nevus Count and Other Pigmentary Characteristics of the Skin

Australia has the highest incidence of melanoma in the world. Melanocytic nevi and mutations in the CDKN2A gene are the main risk factors for the development of cutaneous melanocytic melanoma, and particularly in those of European descent. My study uses genetic epidemiological methods to investigate causes of variation in the number of melanocytic nevi and pigmentary traits such as freckles, eye colour, hair colour and skin colour collected on a sample of adolescent twins and siblings from the Brisbane Twin Nevus Study (1992-2006). Information was available for 2524 individuals from 973 families (from the first visit when the twins were aged 12 years), and from a repeat visit (two years later) for 1598 individuals from 791 families. Using the twin study design extended to siblings and parents, variance components analyses showed that the proportion of phenotypic variance explained by genetic factors ranged from 43 to 99 percent for the traits studied. The atypical nevus count and freckles showed sex differences in the magnitude of genetic and environmental effects. Genetic correlations among counts of three types of nevus (flat, raised and clinically atypical) ranged from 0.46 to 0.63. Nevus count was genetically correlated with skin colour (r=0.23). I analysed genome-wide linkage data using a total of 1190 microsatellite (STR) markers from three scans for 644 families with 1646 twins and siblings, plus genotypes for 1033 parents. These were combined with an additional 169 families with genome-wide association 100K SNP (single nucleotide polymorphism) data, where I selected a linkage analysis panel of 13,000 SNPs making a total of 3365 individuals from 811 families (each individual had more than 200 markers typed). Suggestive linkages for flat nevus count (FNC) were identified on chromosomes 2p25 and 9p21 with lod scores of 3.19 and 2.62 respectively. For raised nevus count (RNC), a suggestive QTL with a lod score of 2.20 was found on chromosome 2q37.2, and for atypical nevus count (ANC) a lod score of 2.71 was found on chromosome 7p14.1. There was suggestive evidence of linkage for freckling on chromosomes 2 and 9. Eye colour was strongly linked (lod=17.86) to chromosome 15, at the OCA2 locus. I have also carried out genetic association analyses using the 100K SNP data in 169 families, and additional fine mapping using SNPs (as well as STR markers) in the complete data set. A sample size of 169 families (461 twins) for genome-wide association data means that statistical power is low. From the 100K SNP data, the best association for total nevus count (TNC) was with SNP rs2420070, p=6.0×10-6 on chromosome 10, and included another two nearby SNPs; rs7086663, p=2.0×10-4 and rs7090904, p=1.5×10-4. These 3 SNPs also showed possible association with FNC; rs2420070, p=3.9×10-6, rs2420070, p=1.3×10-4 and rs2420070, p=8.9×10-5. For raised nevus count the top three associated SNPs were rs1885238, p=1.8×10-5 on chromosome 9; rs10503048, p=3.7×10-5 on chromosome 18 and rs4769189, p=4.0×10-5 on chromosome 13. SNP rs1412341 which is located near CDKN2A on chromosome 9, was also associated with p=2.8×10-4. There were a total 18 SNPs which showed evidence of association with atypical nevus count, the strongest signal being with rs951099 (p=3.7×10-5) on chromosome 9. In a fine-mapping dataset, I studied the association of CDKN2A SNP rs2218220 with TNC, FNC, RNC and ANC. The best SNP, rs2218220, gave p values of 2.8×10-10, 7.7×10-8, 2.7×10-12 and 9.1×10-8, respectively. A SNP, rs1800407 (R419Q) located in the OCA2 gene (chromosome 15q11.2-15q12) showed evidence of association with eye colour and particularly with blue and green eye colours, (p=1.7×10-12 and p=6.0×10-7). SNP rs12913832 from the Hect Domain and RCC1-like Domain 2 gene (HERC2) on chromosome 15q13.1, was also strongly associated with eye colour p=3.6×10-155. This SNP was associated with blue (p= 7.9×10-150) and brown (p=5.3×10-158), but not green eye colour. In addition I confirmed the association of the MC1R SNP rs1805007 and freckling (p=4.8×10-12). This SNP was also associated with FNC (p=3.5×10-8), a finding not previously described in the literature. I also carried out multi-allelic association analysis using STR markers with these traits and uncovered suggestive findings for several regions. Finally, I conducted a multivariate association analysis searching for SNPs with pleiotropic effects. The most interesting results for all types of nevi were with rs801840, p=3.5×10-5, and rs10487075, p=4.9×10-5, both on chromosome 7q21.13. Another four SNPs on chromosome 8p23.1 also showed associations, rs7009724, p=1.4×10-4, rs10503389, p=1.9×10-4, rs7832398, p=6.2×10-4 and rs7005133, p=6.9×10-4 (close to a candidate gene, MFHAS1, implicated in sarcoma risk). In conclusion I have characterised a number of definite and possible genetic factors influencing important risk factors for melanoma