Susceptibility to basal cell carcinoma: associations with PTCH polymorphisms.

Loss of function of the human patched gene (PTCH) is common and critical in basal cell carcinoma (BCC) development. Indirect evidence suggests polymorphism in PTCH mediates BCC risk. We studied 659 BCC cases and 300 controls to determine if exon 2(318), 3(429), 11(1552), 12(1665), 12(1686), 14(2199) and 23(3944) and intron 9(1336-135) and 15(2560+9)PTCH variants were sufficiently common for use in case-control studies, and if selected markers were associated with risk. Intron 15(2560+9) and exon 23(3944) variants were studied further. Their genotype frequencies were not significantly different in controls and cases, though frequency of the G(2560+9)-C(3944) haplotype was lower in all cases (odds ratio=0.44, p=0.009) and those stratified by BCC site and rate of development of further tumours. This association was not mediated by the extent of UVR exposure. We confirmed the robustness of these findings by showing these associations demonstrated similar odds ratios in two groups of randomly selected cases and controls, and using the false positive report probability (FPRP) approach described by Wacholder et al. (2004). The FPRP value (0.168) was in the noteworthy category. These data, showing for the first time that PTCH polymorphism mediates susceptibility, are compatible with reports showing that PTCH haploinsufficiency influences development of BCC precursor lesions.

Glutathione S-transferase GSTP1 and cyclin D1 genotypes: association with numbers of basal cell carcinomas in a patient subgroup at high-risk of multiple tumours.

We previously described associations between basal cell carcinoma (BCC) numbers and allelic variants at loci that mediate host response to ultraviolet radiation (UV). These associations were largely exerted in cases with the multiple presentation phenotype (MPP). This phenotype describes patients who present at their first or a later presentation with a cluster of BCC (2-10 new BCC). Remaining BCC cases have the single presentation phenotype (SPP) and may develop more than one BCC but only have single new lesions at any presentation. We proposed that the MPP cases comprise a high-risk group as they suffer significantly more lesions than SPP cases. We are attempting to determine, in the total BCC case group and subgroups, how many genes influence BCC numbers and their relative importance. In this study, we assessed the influence of two further candidates, glutathione S-transferase GSTP1 and cyclin D1 (CCND1), on tumour numbers in a total group of 457 patients comprising MPP and SPP cases. The relative importance of these genes in comparison with occupational UV exposure and host response (skin type) was also considered. We found that the frequencies of GSTP1 genotypes based on the Ile105 and Val105-expressing alleles and CCND1 AA, AG, GG genotypes were similar in MPP and SPP cases and that there were no significant associations between GSTP1 or CCND1 genotypes and BCC numbers in the total or SPP groups. However, in the MPP cases, GSTP1 Val105/Val105 was associated with more tumours (P = 0.05, reference GSTP1 Ile105/Ile105). Inclusion of skin type and indoor/outdoor occupation in the negative binomial regression models did not alter the associations of these genotypes with tumour numbers. DNA from 258 cases was analysed to identify GSTP1*A (Ile105-Ala114), GSTP1*B (Val105-Ala114), GSTP1*C (Val105-Val114) and GSTP1*D (Ile105-Val114). In SPP cases, there was no association between BCC numbers and GSTP1 BB, though the association with GSTP1 BC approached significance (P = 0.09). In MPP cases, GSTP1 BC was associated with BCC numbers (P = 0.03). We also found that the interaction term, GSTP1 Val105/Val105 with CCND1 AA, was associated with BCC numbers in the total (P = 0.001) and MPP (P = 0.006) but not SPP (P = 0.68) groups. In a stepwise model including GSTP1 Val105/Val105, CCND1 AA and their interaction terms as well as GSTM1, GSTT1 and CYP2D6 genotypes, skin type 1 and gender, the combination of genotypes was the best predictor of BCC numbers. These data suggest that study of further genes involved in cell-cycle control and protection from oxidative stress will be useful, particularly in high-risk subgroups.

Cytochrome P450 CYP2D6 genotypes: association with hair colour, Breslow thickness and melanocyte stimulating hormone receptor alleles in patients with malignant melanoma.

We previously identified associations between polymorphism in the cytochrome P450 CYP2D6 gene and outcome in several cancers. We have now examined the hypothesis that homozygosity for the mutant alleles, CYP2D6*4 and CYP2D6*3, is associated with susceptibility and outcome in malignant melanoma. Outcome was assessed by Breslow thickness. We first confirmed previous reports that these mutant alleles are associated with increased susceptibility to malignant melanoma. For example, the frequency of homozygosity for CYP2D6*4 was significantly greater (P = 0.006, chi-squared 1 d.f. = 7.4, odds ratio 2.2, 95% confidence interval 1.2, 3.9) in cases (9.1%) than in control individuals (4.3%). The frequency of homozygosity for the mutant alleles was next examined in the malignant melanoma cases grouped on the basis of characteristics associated with malignant melanoma risk. Homozygosity was significantly more common (P = 0.038) in cases with red/blonde hair than in those with brown/black hair. We found no associations between the CYP2D6 genotype and sex, skin type or eye colour. The possible association of CYP2D6 with outcome was assessed by comparing genotype frequencies in patients with tumours of Breslow thickness < 1.5 mm with those whose tumours were > or = 1.5 mm. In patients with red/blonde, but not brown or black hair, homozygosity for CYP2D6*4 was significantly associated with thicker lesions in a multivariate model (P = 0.036). We further examined the association of CYP2D6*4 homozygosity with red/blonde hair by classifying patients on the basis of homo- or heterozygosity for wild-type or val92met, asp294his or asp84glu melanocyte stimulating hormone receptor (MC1R) alleles. None of the nine patients with brown/black hair with the asp294his allele were homozygotes for CYP2D6*4. By contrast, in the patients with red/blonde hair, three of five cases with asp294his were homozygotes for the mutant CYP2D6 allele. The difference in the frequency of CYP2D6*4 homozygotes in the red/blonde cases with wild-type MC1R alleles compared with those with asp294his was significant (exact P = 0.029). No associations between val92his or asp84glu and CYP2D6 alleles were identified.

Susceptibility to melanoma: influence of skin type and polymorphism in the melanocyte stimulating hormone receptor gene.

Allelic variation at the melanocyte stimulating hormone receptor (MC1R) gene has been linked with sun-sensitive skin types, suggesting it is a susceptibility candidate for melanoma. We determined the frequency of the val92met, asp294his, and asp84glu MC1R alleles in 190 Caucasian controls and 306 melanoma cases and studied their association with skin type and hair color. The percentage of controls with at least one val92met, asp294his, or asp84glu allele was 17.3%, 6.8%, and 3.5%, respectively. Individually, frequencies of the val92met, asp294his, or asp84glu alleles in the controls with skin types 3 and 4 were similar to those with skin types 1 and 2. Trend analysis, however, did identify an association (exact p = 0.048, two-sided test) between skin type and MC1R variants in the group comprising all controls with any one or more of these alleles. There was no association between MC1R alleles and hair color. Allele frequencies were not different in melanoma cases and controls. There were no associations between skin types and the proportion of cases with the asp294his or asp84glu alleles, though the association between skin type and the val92met allele approached significance (exact p = 0.09, two-sided test). Unexpectedly, in the group comprising all cases with one or more variant alleles, the proportion of subjects with variant alleles increased with skin types associated with tanning rather than burning, although trend analysis showed that this association did not quite reach statistical significance (exact p = 0.08, two-sided test). Asp84glu (but not val92met or asp294his) variant alleles were more common in subjects with blonde hair, although the relationship between the asp84glu allele and hair color did not achieve statistical significance (chi(2)3 = 6.16, exact p = 0.10). We interpret the data presented as indicating that polymorphism at MC1R does not appear a major determinant of skin type, at least in terms of these allelic variants. Furthermore, considered alone, these alleles are not susceptibility candidates for malignant melanoma.