Abstract
CDKN2A is a major susceptibility gene for cutaneous malignant melanoma (CMM) but the variable penetrance and clinical manifestations among mutation carriers suggest the existence of modifier factors. The goal of this study was to identify modifier genes for CMM in CMM-prone families with or without CDKN2A mutations. We genotyped 537 individuals (107 CMM) from 28 families (19 CDKN2A+, 9 CDKN2A−) for 1536 SNPs in 152 genes involved in DNA repair, apoptosis, and immune response pathways. We used conditional logistic regression to account for family ascertainment and differences in disease prevalence among families. Pathway- and gene-based permutation analyses were used to assess the risk of CMM associated with genes in the five pathways (DNA repair, apoptosis, TNF/NFκB, TH1:TH2, and other immune regulation). Our analyses identified some candidate genes such as FAS, BCL7A, CASP14, TRAF6, WRN, IL9, IL10RB, TNFSF8, TNFRSF9, and JAK3 that were associated with CMM risk (P<0.01, gene-based test). After correction for multiple comparisons, IL9 remained significant (Bonferroni P<0.05). The effects of some genes were stronger in CDKN2A-positive families (BCL7A and IL9), while some were stronger in CDKN2A-negative families (BCL2L1). Our findings support the hypothesis that common genetic polymorphisms in DNA repair, apoptosis, and immune response pathways may modify the risk of CMM in CMM-prone families with or without CDKN2A mutations.
Keywords: modifier gene, cutaneous malignant melanoma, high-risk family, CDKN2A, pathway
Introduction
Cutaneous malignant melanoma (CMM) is an etiologically heterogeneous disease with genetic, host, environmental factors, and their interactions contributing to its development. Approximately 10% of CMM cases occur in a familial setting1. To date, two high-penetrance melanoma susceptibility genes, CDKN2A on chromosome 9p21 and CDK4 on 12q14, have been identified. Germline mutations of the CDKN2A gene have been described in approximately 20% of familial melanoma kindreds2,3. Mutations of CDK4 are very rare; only about 10 families worldwide have been found to harbor mutations4, unpublished data. Together, these two genes account for melanoma susceptibility in a small proportion of melanoma-prone families, suggesting the existence of other genetic factors.
Although germline CDKN2A mutations are associated with high risk of CMM, the penetrance of this gene is incomplete and varies by age and geographic location5. Further, phenotypic manifestations, such as age at diagnosis, presence/number of dysplastic nevi (DN), number of melanomas, and co-segregation of pancreatic cancer, vary significantly among mutation carriers even within a single family. These findings suggest that other factors may modify the effect of CDKN2A. Previous studies have demonstrated that DN and melanocortin-1 receptor (MC1R) variants modified penetrance of CDKN2A mutations in melanoma-prone families6-9. MC1R, which encodes the melanocyte-stimulating hormone receptor and plays a crucial role in pigmentation, has been clearly identified as a low-penetrance melanoma susceptibility gene. Other common genetic variants in low-penetrance genes in a number of biologically important pathways, such as cell cycle, DNA repair, apoptosis, and detoxification of metabolites, have been reported to confer modest risk for melanoma10-13. A recent genome-wide association study (GWAS) identified variants in two pigmentation genes (ASIP and TYR) that were significantly associated with CMM risk14. However, the role of these variants in families segregating CDKN2A is undefined.
Sun exposure is believed to be the most important environmental risk factor for the development of CMM. Ultraviolet radiation (UVR) can directly cause DNA damage, influence the expression of apoptosis-related molecules, and induce immunosuppression15. Renal transplantation recipients have increased risk of melanoma compared to the general population, suggesting that altered immune response may contribute to melanoma susceptibility16. Genetic variation of some cytokines and their receptor genes were demonstrated to influence melanoma susceptibility17-19, however, few systematic evaluations of genetic variants in immune response pathways, particularly in CMM high risk families, have been performed. Therefore, in the current study, we evaluated ∼1,500 single nucleotide polymorphisms (SNPs) in 152 genes involved in multiple pathways including DNA repair, apoptosis and immune response in 28 CMM-prone families, mostly with CDKN2A mutations.
Materials and Methods
Study population
The study population of this family study has been previously described8,20,21. In brief, US families with at least two living first-degree relatives with a history of invasive melanoma were ascertained through health-care professionals or self-referrals. All family members willing to participate in the study underwent a full-body skin examination for extensive phenotypes and completed risk factor questionnaires for sun-related exposures. All diagnoses of melanoma were confirmed by histologic review of pathologic material, pathology reports, or death certificates. Clinical and histologic criteria for the diagnosis of DN have been previously described22,23. DN were divided into three categories: absent, indeterminate, or present. Indeterminate was used for prepubertal subjects less than 16 years of age and for subjects greater than 60 years of age without definite clinical or histologic evidence of DN24. The study was approved by the National Cancer Institute Clinical Center Institutional Review Board and conducted according to the Declaration of Helsinki. Informed consent was obtained from all participants.
The current study was based on 28 families with 537 genotyped individuals (107 genotyped CMM cases). Nineteen families segregated CDKN2A mutations and 9 families were CDKN2A mutation negative. The distribution of CMM cases by CDKN2A status in each family is shown in Table 1. All study participants were Caucasians.
Table 1.
Distribution of CMM and unaffected individuals by CDKN2A status in each of the 28 CMM families included in this study.
| Family | CMM | Unaffected | ||||
|---|---|---|---|---|---|---|
| CDKN2A+ | CDKN2A− | CDKN2A unknown | CDKN2A+ | CDKN2A− | CDKN2A unknown | |
| CDKN2A+ | ||||||
| AN | 2 | 1 | 0 | 1 | 4 | 0 |
| K | 4 | 0 | 0 | 2 | 12 | 0 |
| D | 4 | 0 | 0 | 2 | 15 | 0 |
| F | 3 | 1 | 0 | 3 | 33 | 0 |
| I | 3 | 0 | 0 | 1 | 5 | 1 |
| AH | 6 | 0 | 0 | 3 | 13 | 0 |
| Q | 2 | 0 | 0 | 1 | 3 | 0 |
| B | 4 | 0 | 0 | 1 | 7 | 0 |
| C | 2 | 0 | 0 | 4 | 12 | 0 |
| L | 5 | 0 | 0 | 7 | 32 | 0 |
| AP | 2 | 1 | 0 | 1 | 8 | 0 |
| E | 2 | 0 | 0 | 6 | 8 | 0 |
| A | 7 | 0 | 0 | 6 | 28 | 0 |
| J | 4 | 1 | 0 | 2 | 14 | 1 |
| O | 3 | 0 | 0 | 5 | 5 | 8 |
| P | 8 | 0 | 0 | 16 | 25 | 0 |
| N | 3 | 1 | 0 | 0 | 11 | 0 |
| H | 2 | 0 | 0 | 2 | 2 | 0 |
| G | 4 | 1 | 0 | 3 | 17 | 0 |
| Total CDKN2A+ families | 70 | 61 | 0 | 66 | 254 | 10 |
| CDKN2A− | ||||||
| M | 0 | 7 | 0 | 0 | 27 | 0 |
| W | 0 | 3 | 0 | 0 | 15 | 0 |
| Z | 0 | 3 | 0 | 0 | 18 | 0 |
| B3 | 0 | 3 | 0 | 0 | 4 | 0 |
| A2 | 0 | 4 | 0 | 0 | 10 | 0 |
| A5 | 0 | 2 | 0 | 0 | 5 | 0 |
| A3 | 0 | 3 | 0 | 0 | 4 | 0 |
| A6 | 0 | 3 | 0 | 0 | 13 | 0 |
| AS | 0 | 3 | 0 | 0 | 4 | 0 |
| Total CDKN2A− families | 0 | 31 | 0 | 0 | 100 | 0 |
| Total (all) | 70 | 37 | 0 | 66 | 354 | 10 |
One of them are from marry-in (non-bloodline) relatives.
Pathway, gene, and SNP selection
152 genes involved in several pathways, including DNA repair, apoptosis, immune response (TH1:TH2, TNF/NFκB, and immune regulation), and oxistress were selected based on their reported roles in cancer development (Supplementary table 1). For each gene, tag SNPs were selected using a minimum minor allele frequency (MAF) criterion of MAF≥5% and r2<0.8, based upon HapMap data for CEU Caucasian samples. SNP validation scores between 0 and 1 were generated using the Illumina Assay Design Tool to estimate the likelihood that SNPs would be successfully genotyped. SNPs with a design score of less than 0.6 were excluded. The final panel of 1536 SNPs was custom designed and genotyped with the Illumina GoldenGate Assay at the National Cancer Institute's Core Genotyping Facility (Gaithersburg, MD). The overall genotype completion rate was 98.5%. The concordance between duplicate pairs was >99%. Forty SNPs that failed QC or significantly violated Hardy-Weinberg equilibrium among controls (P < 0.001) were excluded from the analyses. As an additional check for genotyping quality, family structures were used and 6 SNPs with extensive Mendelian inconsistencies were removed from all analyses.
Statistical analysis
Conditional logistic regression models were used to estimate the trend P-value for the association between CMM risk and each SNP, using co-dominant coding for genotypes (0, 1, 2) with the homozygote of common alleles as the reference group. Odds ratios (ORs) and 95% confidence intervals (CIs) for each heterozygous and homozygous rare genotype were calculated. Conditioning on families was used to account for family ascertainment and differences in disease prevalence among families. While this approach ignores residual familial correlations among family members, it gives estimates that are attenuated towards the null and is thus conservative25. All analyses were adjusted for age and a three-category nevus factor combining DN (absent, indeterminate, present) and total numbers of nevi. For SNPs with P < 0.01 in single SNP analyses (listed in Supplementary table 1) and SNPs in associated genes (P < 0.05, gene-based permutation analysis, Table 2), we also included adjustment for MC1R variants in the models as a surrogate for pigmentation characteristics. Most pigmentation risk phenotypes, such as red hair color, poor tanning ability, pale/fair skin color, and extensive freckling were previously associated with MC1R variants in our CDKN2A mutation positive families8. MC1R variants were coded as 0=no variant, 1=single variant, 2=multiple variants. SNPs that were associated with CMM risk when all families were combined were also analyzed separately in CDKN2A-positive and negative families to assess effect modification by CDKN2A status. Interactions between CDKN2A status and significant SNPs were also formally tested by including an interaction term (Wald statistics). Since CMM is a late-onset disease, disease phenotype may not be revealed among young unaffected family members, especially among CDKN2A mutation carriers. To reduce phenotypic misclassification, we also conducted separate analyses excluding young unaffected family members (age < 20). The results did not change significantly, so the analyses presented include all genotyped family members.
Table 2.
Pathways and genes showing associations with CMM risk in CMM-prone families.
| Pathways | Genes | #SNPs | P-value1 for most significant SNP | Permutation P-value2 for gene | Permutation P-value2 for pathway |
|---|---|---|---|---|---|
| Apoptosis | FAS | 22 | 0.00026 | 0.00040 | 0.027 |
| BCL7A | 6 | 0.0035 | 0.00065 | ||
| CASP14 | 8 | 0.0026 | 0.0028 | ||
| RIPK1 | 9 | 0.00021 | 0.011 | ||
| BCL2L2 | 2 | 0.030 | 0.013 | ||
| BCL2L1 | 4 | 0.0078 | 0.017 | ||
| DNA repair | TRAF6 | 8 | 0.0018 | 0.0031 | 0.034 |
| WRN | 18 | 0.0046 | 0.0067 | ||
| LIG4 | 28 | 0.0055 | 0.020 | ||
| PRKDC | 11 | 0.0010 | 0.036 | ||
| TH1:TH2 | IL9 | 10 | 0.0058 | <0.00001 | 0.049 |
| IL10RB | 12 | 0.00029 | 0.0036 | ||
| TNF/NFκB | TNFSF8 | 15 | 0.015 | 0.00060 | 0.22 |
| TNFRSF9 | 5 | 0.0068 | 0.0090 | ||
| NFκB1 | 13 | 0.045 | 0.024 | ||
| TNFRSF17 | 9 | 0.018 | 0.027 | ||
| TRAF5 | 6 | 0.064 | 0.030 | ||
| IKBKB | 7 | 0.037 | 0.039 | ||
| Immune regulation | JAK3 | 15 | 0.0068 | 0.0040 | 0.42 |
| FCGR2A | 8 | 0.010 | 0.0048 | ||
| IL1A/IL1B | 13 | 0.013 | 0.0054 | ||
| CD4 | 23 | 0.011 | 0.014 |
Obtained from likelihood ratio test in conditional logistic regression analysis. Genotype using co-dominant coding (0, 1, 2) with the homozygote of the common alleles as the reference group.
Gene-based and pathway-based analyses were performed on the 152 genes and 7 pathways. P-values were computed using rank-truncated test statistics and a permutation-based sampling procedure (20,000 permutations).
Since our primary goal is to identify genes and pathways, rather than single SNPs, that are associated with CMM risk, we applied a gene-based analysis to assess the significance of the joint effect of multiple SNPs genotyped in each gene on CMM risk. We performed gene-based analyses on the 152 genes and computed P-values using rank-truncated test statistics and a permutation-based sampling procedure (20,000 permutations) in the conditional logistic regression models adjusted for the above mentioned variables, taking into account the number of SNPs genotyped in each gene and their LD structure26. We used a Bonferroni correction to account for the number of genes tested, and thus used P < 0.05/152 to define statistical significance. Similar rank-truncated test statistics were used to test associations of each pathway with CMM risk.
All analyses were performed using SAS software, version 9.1 (SAS Institute, Inc., Cary, NC).
Results
Table 1 shows the number of genotyped CMM cases and unaffected family members (including blood-line relatives and spouse controls) and their CDKN2A status in each of the 28 CMM-prone families examined in this study. Since our primary goal was to identify modifier genes for CDKN2A, we over-sampled families segregating CDKN2A (N=19) than families without known mutations (N=9). Among bloodline relatives in CDKN2A mutation positive families, all but 5 CMM cases were CDKN2A carriers. 66 individuals without CMM were also CDKN2A positive. Among the study participants, 537 (107 CMM cases, 136 CDKN2A carriers) had sufficient DNA and were successfully genotyped. After removing SNPs that failed to amplify, were not in Hardy-Weinberg equilibrium among controls, and had substantial Mendelian inconsistencies, a total of 1,491 SNPs were included in subsequent analyses.
We conducted gene-based and pathway-based analyses to examine whether genes in immune response, DNA repair, and apoptosis pathways were associated with CMM risk in our families with and without CDKN2A mutations. Pathway-based analyses suggested that apoptosis, DNA repair, and TH1:TH2 pathways were significantly associated with CMM risk (P < 0.05, Table 2). Gene-based analyses identified 6 genes in the apoptosis pathway (FAS, BCL7A, CASP14, RIPK1, BCL2L2, and BCL2L1), 4 genes in the DNA repair pathway (TRAF6, WRN, LIG4, and PRKDC), 2 genes in the TH1:TH2 pathway (IL9, IL10RB), 6 genes in the TNF/NFκB pathway (TNFSF8, TNFRSF9, NFκB1, TNFRSF17, TRAF5, and IKBKB), and 4 genes in the immune regulation pathway (JAK3, FCGR2A, IL1A/IL1B, and CD4) that were associated with CMM risk (P < 0.05, Table 2). After Bonferroni correction, IL9 in the TH1:TH2 pathway remained statistically significant. There were 7 additional SNPs in genes that were not significant in gene-based analyses that were associated with CMM risk (P < 0.01) in single-SNP analyses, including TNFRSF10C rs10866820 (P=0.0048), TNFRSF1A rs12296430 (P=0.006), and TNFRSF8 rs11569869 (P=0.0064) in the TNF/NFκB pathway, BCL2 rs1381548 (P=0.007) and BCL2L11 rs17484848 (P=0.0005) in the apoptosis pathway, NBN rs1235369 (P=0.0094) in the DNA repair pathway, and CD28 rs11681040 in the TH1:TH2 pathway (Table 3). After Bonferroni correction, none of these SNPs was significant.
Table 3.
SNPs showing suggestive associations (P < 0.01) with CMM risk in CMM-prone families from single-SNP analysis.
| SNP | Gene | Pathway | Ptrend |
|---|---|---|---|
| rs12200314 | RIPK1 | Apoptosis | 0.0002 |
| rs9658727 | FAS | TNF/NFkB | 0.0003 |
| rs2834168 | IL10RB | TH1:TH2 | 0.0003 |
| rs17484848 | BCL2L11 | Apoptosis | 0.0005 |
| rs8178179 | PRKDC | DNA repair | 0.0010 |
| rs331457 | TRAF6/RAG1/RAG2 | DNA repair | 0.0018 |
| rs12827036 | BCL7A | Apoptosis | 0.0035 |
| rs4733225 | WRN | DNA repair | 0.0046 |
| rs10866820 | TNFRSF10C | TNF/NFkB | 0.0048 |
| rs331455 | TRAF6/RAG1/RAG2 | DNA repair | 0.0052 |
| rs1224177 | LIG4/TNFSF13B | TNF/NFkB | 0.0055 |
| rs740002 | IL9 | TH1:TH2 | 0.0058 |
| rs12296430 | TNFRSF1A/LTBR/TNFRSF7 | TNF/NFkB | 0.0060 |
| rs1182927 | BCL7A | Apoptosis | 0.0061 |
| rs13251813 | WRN | DNA repair | 0.0061 |
| rs11569869 | TNFRSF8/TNFRSF1B | TNF/NFkB | 0.0064 |
| rs2493215 | TNFRSF9 | TNF/NFkB | 0.0068 |
| rs11086080 | JAK3 | Other immune | 0.0068 |
| rs1381548 | BCL2 | Apoptosis | 0.0070 |
| rs7270207 | BCL2L1 | Apoptosis | 0.0078 |
| rs2069882 | IL9 | TH1:TH2 | 0.0082 |
| rs12869406 | LIG4/TNFSF13B | DNA repair | 0.0083 |
| rs4934436 | FAS | TNF/NFkB | 0.0089 |
| rs1235369 | NBN | DNA repair | 0.0094 |
| rs11681040 | CD28 | TH1:TH2 | 0.0100 |
We then analyzed individual SNPs showing suggestive associations with CMM risk in either single SNP or gene-based analyses separately in CDKN2A-positive and negative families. Although the stratified analysis was limited by small sample size, it clearly suggested that some SNPs might have stronger risks in CDKN2A-positive families, such as BCL7A, TRAF6, and IL9, whereas some might have stronger effects in CDKN2A-negative families, such as BCL2L1 and TNFSF8 (Table 4). Tests for interaction between CDKN2A and SNPs were significant for rs12827036 in BCL7A (P=0.014), rs7270207 in BCL2L1 (P=0.026), and rs2069882 in IL9 (P =0.035).
Table 4.
Trend P-values and odds ratios of SNPs associated with CMM risk from single-SNP and gene-based analyses in CDKN2A-positive and negative families.
| Gene | SNP | CDKN2A+ | CDKN2A− | ||
|---|---|---|---|---|---|
| P-trend | OR (95% CI) | P-trend | OR (95% CI) | ||
| BCL7A | rs1182927 | 0.0014 | 0.31(0.15,0.63) | 0.77 | 1.22(0.32,4.68) |
| rs12827036* | 0.0002 | 0.36(0.21,0.61) | 0.62 | 1.19(0.61,2.33) | |
| CASP14 | rs4808901 | 0.05 | 1.72(1.00,2.97) | 0.22 | 1.80(0.70,4.60) |
| RIPK1 | rs12200314 | 0.0025 | 2.22(1.32,3.71) | 0.083 | 1.95(0.92,4.13) |
| BCL2L2 | rs1884056 | 0.045 | 1.69(1.01,2.81) | 0.23 | 1.60(0.74,3.48) |
| BCL2L1 | rs7270207* | 0.76 | 0.84(0.29,2.47) | 0.0017 | 0.16(0.049,0.50) |
| BCL2L11 | rs14784848 | 0.013 | 0.14(0.029,0.66) | 0.14 | 0.42(0.13,1.33) |
| BCL2 | rs1381548 | 0.0074 | 0.50(0.30,0.83) | 0.31 | 0.67(0.31,1.45) |
| rs4987827 | 0.023 | 0.49(0.27,0.91) | 0.48 | 0.71(0.27,1.86) | |
| TRAF6 | rs331457 | 0.0007 | 3.19(1.63,6.24) | 0.51 | 1.34(0.56,3.21) |
| rs331455 | 0.008 | 0.44(0.24,0.81) | 0.31 | 0.65(0.29,1.48) | |
| WRN | rs4733225 | 0.0024 | 2.40(1.36,4.21) | 0.82 | 1.15(0.35,3.75) |
| rs13251813 | 0.0057 | 4.59(1.56,13.52) | 0.74 | 1.52(0.13,18.23) | |
| LIG4 | rs1224177 | 0.028 | 1.92(1.07,3.42) | 0.094 | 1.91(0.90,4.09) |
| rs12869406 | 0.032 | 2.11(1.07,4.16) | 0.078 | 2.04(0.92,4.52) | |
| rs1473792 | 0.088 | 0.54(0.26,1.10) | 0.050 | 0.42(0.17,1.00) | |
| PRKDC | rs8178179 | 0.0033 | 0.065(0.01,0.40) | 0.42 | 0.55(0.13,2.36) |
| NBN | rs1235369 | 0.021 | 5.37(1.29,22.34) | 0.21 | 2.45(0.60,10.01) |
| IL9 | rs740002 | 0.054 | 1.67(1.00,2.80) | 0.050 | 2.01(1.00,4.03) |
| rs2069882* | 0.0043 | 0.37(0.19,0.73) | 0.89 | 0.94(0.37,2.40) | |
| rs1859428 | 0.60 | 1.16(0.67,2.02) | 0.0030 | 3.57(1.54,8.27) | |
| IL10RB | rs2834168 | 0.0065 | 2.15(1.24,3.72) | 0.036 | 2.25(1.05,4.82) |
| FAS | rs9658727 | 0.0061 | 0.40(0.21,0.77) | 0.045 | 0.22(0.051,0.97) |
| 0.021 | 1.76(1.09,2.84) | 0.28 | 1.62(0.67,3.90) | ||
| TNFSF8 | rs7872878 | 0.28 | 1.34(0.79,2.27) | 0.0096 | 2.95(1.30,6.69) |
| rs927375 | 0.33 | 0.78(0.47,1.29) | 0.0048 | 0.27(0.11,0.67) | |
| TNFRSF9 | rs2493215 | 0.10 | 1.53(0.92,2.56) | 0.038 | 2.40(1.05,5.47) |
| rs679563 | 0.38 | 1.25(0.76,2.08) | 0.011 | 3.11(1.29,7.52) | |
| NFκB1 | rs230510 | 0.098 | 1.47(0.93,2.33) | 0.19 | 1.62(0.79,3.32) |
| TNFRSF17 | rs9922891 | 0.017 | 3.88(1.28,11.79) | 0.59 | 1.64(0.27,9.88) |
| TNFRSF10 | rs10866820 | 0.062 | 1.74(0.97,3.11) | 0.012 | 4.73(1.40,16.00) |
| TNFRSF1A | rs12296430 | 0.010 | 0.46(0.25,0.83) | 0.24 | 0.58(0.23,1.46) |
| JAK3 | rs11086080 | 0.12 | 1.60(0.89,2.89) | 0.031 | 2.52(1.09,5.85) |
| IL1A/IL1B | rs17042407 | 0.043 | 1.81(1.02,3.23) | 0.36 | 1.49(0.64,3.46) |
| CD4 | rs2071081 | 0.011 | 0.51(0.31,0.86) | 0.44 | 0.74(0.35,1.58) |
: Test for interaction between CDKN2A and SNP was significant (P < 0.05).
To evaluate whether associations of SNPs with CMM were influenced by pigmentation characteristics, we adjusted for MC1R variants as a surrogate for pigmentation variables in the models. Associations of SNPs with CMM risk did not change significantly (data not shown). To assess the magnitude of possible bias due to residual familial correlations among family members, we also used a two-level random effect model25. Single-SNP analyses based on the random effect model showed stronger significance and therefore confirmed that our findings obtained from conditional logistic regression were conservative (data not shown).
Discussion
CDKN2A confers the strongest risk for CMM in our CMM-prone families segregating CDKN2A mutations27. Among 86 CMM cases from CDKN2A-positive families analyzed in this study, only 5 bloodline cases did not have CDKN2A mutations. This finding is consistent with Melanoma Genetics Consortium (GenoMel) data that about 10% of melanomas in CDKN2A families occurred in non-mutation carriers5. However, the penetrance of CDKN2A is known to be incomplete. Among 136 CDKN2A mutation carriers genotyped, only half were affected with CMM. Additional genetic factors are likely to modulate cancer susceptibility. In this exploratory study, we evaluated the association between genetic variants of genes involved in DNA repair, apoptosis, and immune regulation and response pathways and the risk of developing CMM in CMM high-risk families with or without CDKN2A mutations. Our analyses identified some candidate genes such as FAS, BCL7A, CASP14, TRAF6, WRN, IL9, IL10RB, TNFSF8, TNFRSF9, and JAK3 that were associated with CMM risk (P<0.01, gene-based test). After adjusting for multiple testing, IL9 in the TH1:TH2 pathway remained significant.
The key environmental risk factor for CMM is UVR exposure. Apoptosis and DNA repair are the two protective mechanisms against DNA damage caused by sun exposure. Genetic polymorphisms in genes regulating apoptosis and DNA repair were associated with CMM risk28,29. We also observed that genetic variants of some genes, such as FAS, CASP14, WRN, and LIG4, in these two pathways were associated with CMM risk in our melanoma-prone families. We showed that Werner Syndrome protein (WRN), a member of the RecQ family of helicases with a role in maintaining genomic stability, was associated with CMM risk. This finding is consistent with a previous report that CMM occurs in excess in Werner Syndrome30. In addition to directly causing DNA damage, UVR depresses the immune response in the skin, which can permit the growth of emerging tumors produced by the effects of UV-induced DNA damage31. Studies investigating the role of genetic variants in genes involved in immune regulation in CMM etiology are limited. However, several studies have suggested that genetic polymorphisms of some cytokines and their receptors might influence the risk and development of CMM and other skin cancers17,18. Consistent with these results, our data support the hypothesis that common genetic polymorphisms in immune response/regulation pathways may play important roles in CMM etiology.
Our analysis was limited by the small number of CMM cases particularly in analyses stratified by CDKN2A status. The stronger effects of most genes in CDKN2A-positive families may be due to the smaller number of CDKN2A-negative families examined in our study. Our study should be viewed as an exploratory study and replication in larger samples is warranted. Also, we could not adequately assess the interaction of genetic factors and host factors and sun exposure-related variables. Because of small numbers, we used MC1R variants as a surrogate for skin type, eye/hair color, and sun burn/tanning abilities. We included nevi as a covariate in all models with CMM as the outcome variable. Adjustment for MC1R and nevi did not change results significantly, suggesting that these SNPs might be risk factors of CMM independently from host factors and sun exposure. Our study was also limited by the selection of genes and pathways included as the panel did not include all genes or SNPs that were found to be important in CMM by previous studies. Despite these limitations, we used a gene/pathway-based permutation analysis that allowed us to identify biologically relevant genes/pathways that are associated with disease risk in a moderately sized study population. In addition, our high-risk families provide a unique resource for identifying genetic variants since familial cases are enriched for genetic alterations. Although based on CMM-prone families, these results may also have the potential of being generalized to sporadic CMM. In summary, our findings support the hypothesis that common genetic polymorphisms in different genes in DNA repair and immune response pathways may modify the risk of CMM in CMM-prone families with and without CDKN2A mutations.
Supplementary Material
Supplemental Table 1. Candidate genes and pathways included in this study.
Acknowledgements
We are indebted to the participating families, whose generosity and cooperation have made this study possible. We also acknowledge the contributions to this work that were made by Virginia Pichler, Deborah Zametkin, and Mary Fraser. This research was supported by the Intramural Research Program of the NIH, NCI, DCEG.
Footnotes
Statement of novelty and impact: Using a pathway and gene-based statistical approach, we demonstrated that common genetic variants in several biologically important pathways may act as modifier genes in melanoma high-risk families. These results may improve our understanding of phenotypic variation in families segregating a major susceptibility gene and may provide new clues in identifying additional genes for melanoma in general.
References
- 1.Goldstein AM, Tucker MA. Genetic epidemiology of cutaneous melanoma: a global perspective. Arch.Dermatol. 2001;137:1493–6. doi: 10.1001/archderm.137.11.1493. [DOI] [PubMed] [Google Scholar]
- 2.Eliason MJ, Larson AA, Florell SR, Zone JJ, Cannon-Albright LA, Samlowski WE, Leachman SA. Population-based prevalence of CDKN2A mutations in Utah melanoma families. J Invest Dermatol. 2006;126:660–6. doi: 10.1038/sj.jid.5700094. [DOI] [PubMed] [Google Scholar]
- 3.Goldstein AM. Familial melanoma, pancreatic cancer and germline CDKN2A mutations. Hum Mutat. 2004;23:630. doi: 10.1002/humu.9247. [DOI] [PubMed] [Google Scholar]
- 4.Helsing P, Nymoen DA, Ariansen S, Steine SJ, Maehle L, Aamdal S, Langmark F, Loeb M, Akslen LA, Molven A, Andresen PA. Population-based prevalence of CDKN2A and CDK4 mutations in patients with multiple primary melanomas. Genes Chromosomes Cancer. 2008;47:175–84. doi: 10.1002/gcc.20518. [DOI] [PubMed] [Google Scholar]
- 5.Bishop DT, Demenais F, Goldstein AM, Bergman W, Bishop JN, Bressac-de PB, Chompret A, Ghiorzo P, Gruis N, Hansson J, Harland M, Hayward N, et al. Geographical variation in the penetrance of CDKN2A mutations for melanoma. J Natl Cancer Inst. 2002;94:894–903. doi: 10.1093/jnci/94.12.894. [DOI] [PubMed] [Google Scholar]
- 6.Box NF, Duffy DL, Chen W, Stark M, Martin NG, Sturm RA, Hayward NK. MC1R genotype modifies risk of melanoma in families segregating CDKN2A mutations. Am J Hum Genet. 2001;69:765–73. doi: 10.1086/323412. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Chaudru V, Laud K, Avril MF, Miniere A, Chompret A, Bressac-de PB, Demenais F. Melanocortin-1 receptor (MC1R) gene variants and dysplastic nevi modify penetrance of CDKN2A mutations in French melanoma-prone pedigrees. Cancer Epidemiol Biomarkers Prev. 2005;14:2384–90. doi: 10.1158/1055-9965.EPI-04-0777. [DOI] [PubMed] [Google Scholar]
- 8.Goldstein AM, Landi MT, Tsang S, Fraser MC, Munroe DJ, Tucker MA. Association of MC1R variants and risk of melanoma in melanoma-prone families with CDKN2A mutations. Cancer Epidemiol Biomarkers Prev. 2005;14:2208–12. doi: 10.1158/1055-9965.EPI-05-0321A. [DOI] [PubMed] [Google Scholar]
- 9.van d V, Sandkuijl LA, Bergman W, Pavel S, van ML, Frants RR, Gruis NA. Melanocortin-1 receptor variant R151C modifies melanoma risk in Dutch families with melanoma. Am J Hum Genet. 2001;69:774–9. doi: 10.1086/323411. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.James MR, Roth RB, Shi MM, Kammerer S, Nelson MR, Stark MS, Dumenil T, Montgomery GW, Hayward NK, Martin NG, Braun A, Duffy DL. BRAF polymorphisms and risk of melanocytic neoplasia. J Invest Dermatol. 2005;125:1252–8. doi: 10.1111/j.0022-202X.2005.23937.x. [DOI] [PubMed] [Google Scholar]
- 11.Jannot AS, Meziani R, Bertrand G, Gerard B, Descamps V, Archimbaud A, Picard C, Ollivaud L, Basset-Seguin N, Kerob D, Lanternier G, Lebbe C, et al. Allele variations in the OCA2 gene (pink-eyed-dilution locus) are associated with genetic susceptibility to melanoma. Eur J Hum Genet. 2005;13:913–20. doi: 10.1038/sj.ejhg.5201415. [DOI] [PubMed] [Google Scholar]
- 12.Kanetsky PA, Holmes R, Walker A, Najarian D, Swoyer J, Guerry D, Halpern A, Rebbeck TR. Interaction of glutathione S-transferase M1 and T1 genotypes and malignant melanoma. Cancer Epidemiol Biomarkers Prev. 2001;10:509–13. [PubMed] [Google Scholar]
- 13.Winsey SL, Haldar NA, Marsh HP, Bunce M, Marshall SE, Harris AL, Wojnarowska F, Welsh KI. A variant within the DNA repair gene XRCC3 is associated with the development of melanoma skin cancer. Cancer Res. 2000;60:5612–6. [PubMed] [Google Scholar]
- 14.Gudbjartsson DF, Sulem P, Stacey SN, Goldstein AM, Rafnar T, Sigurgeirsson B, Benediktsdottir KR, Thorisdottir K, Ragnarsson R, Sveinsdottir SG, Magnusson V, Lindblom A, et al. ASIP and TYR pigmentation variants associate with cutaneous melanoma and basal cell carcinoma. Nat Genet. 2008;40:886–91. doi: 10.1038/ng.161. [DOI] [PubMed] [Google Scholar]
- 15.Muller HK, Malley RC, McGee HM, Scott DK, Wozniak T, Woods GM. Effect of UV radiation on the neonatal skin immune system- implications for melanoma. Photochem Photobiol. 2008;84:47–54. doi: 10.1111/j.1751-1097.2007.00246.x. [DOI] [PubMed] [Google Scholar]
- 16.Hollenbeak CS, Todd MM, Billingsley EM, Harper G, Dyer AM, Lengerich EJ. Increased incidence of melanoma in renal transplantation recipients. Cancer. 2005;104:1962–7. doi: 10.1002/cncr.21404. [DOI] [PubMed] [Google Scholar]
- 17.Gu F, Qureshi AA, Niu T, Kraft P, Guo Q, Hunter DJ, Han J. Interleukin and interleukin receptor gene polymorphisms and susceptibility to melanoma. Melanoma Res. 2008;18:330–5. doi: 10.1097/CMR.0b013e32830658b2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Howell WM, Turner SJ, Bateman AC, Theaker JM. IL-10 promoter polymorphisms influence tumour development in cutaneous malignant melanoma. Genes Immun. 2001;2:25–31. doi: 10.1038/sj.gene.6363726. [DOI] [PubMed] [Google Scholar]
- 19.Nikolova PN, Pawelec GP, Mihailova SM, Ivanova MI, Myhailova AP, Baltadjieva DN, Marinova DI, Ivanova SS, Naumova EJ. Association of cytokine gene polymorphisms with malignant melanoma in Caucasian population. Cancer Immunol Immunother. 2007;56:371–9. doi: 10.1007/s00262-006-0193-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Hussussian CJ, Struewing JP, Goldstein AM, Higgins PA, Ally DS, Sheahan MD, Clark WH, Jr, Tucker MA, Dracopoli NC. Germline p16 mutations in familial melanoma. Nat Genet. 1994;8:15–21. doi: 10.1038/ng0994-15. [DOI] [PubMed] [Google Scholar]
- 21.Goldstein AM, Struewing JP, Chidambaram A, Fraser MC, Tucker MA. Genotype-phenotype relationships in U.S. melanoma-prone families with CDKN2A and CDK4 mutations. J Natl Cancer Inst. 2000;92:1006–10. doi: 10.1093/jnci/92.12.1006. [DOI] [PubMed] [Google Scholar]
- 22.Reimer RR, Clark WH, Jr, Greene MH, Ainsworth AM, Fraumeni JF., Jr Precursor lesions in familial melanoma. A new genetic preneoplastic syndrome. JAMA. 1978;239:744–746. [PubMed] [Google Scholar]
- 23.Clark WH, Jr, Reimer RR, Greene M, Ainsworth AM, Mastrangelo MJ. Origin of familial malignant melanomas from heritable melanocytic lesions. ‘The B-K mole syndrome’. Arch Dermatol. 1978;114:732–8. [PubMed] [Google Scholar]
- 24.Tucker MA, Greene MH, Clark WH, Jr, Kraemer KH, Fraser MC, Elder DE. Dysplastic nevi on the scalp of prepubertal children from melanoma-prone families. J Pediatr. 1983;103:65–9. doi: 10.1016/s0022-3476(83)80777-x. [DOI] [PubMed] [Google Scholar]
- 25.Pfeiffer RM, Gail MH, Pee D. Inference for covariates that accounts for ascertainment and random genetic effects in family studies. Biometrika. 2001;88:933–48. [Google Scholar]
- 26.Dudbridge F, Koeleman BP. Rank truncated product of P-values, with application to genomewide association scans. Genet Epidemiol. 2003;25:360–6. doi: 10.1002/gepi.10264. [DOI] [PubMed] [Google Scholar]
- 27.Goldstein AM, Martinez M, Tucker MA, Demenais F. Gene-covariate interaction between dysplastic nevi and the CDKN2A gene in American melanoma-prone families. Cancer Epidemiol Biomarkers Prev. 2000;9:889–94. [PubMed] [Google Scholar]
- 28.Figl A, Scherer D, Nagore E, Bermejo JL, Dickes E, Thirumaran RK, Gast A, Hemminki K, Kumar R, Schadendorf D. Single nucleotide polymorphisms in DNA repair genes XRCC1 and APEX1 in progression and survival of primary cutaneous melanoma patients. Mutat Res. 2009;661:78–84. doi: 10.1016/j.mrfmmm.2008.11.011. [DOI] [PubMed] [Google Scholar]
- 29.Povey JE, Darakhshan F, Robertson K, Bisset Y, Mekky M, Rees J, Doherty V, Kavanagh G, Anderson N, Campbell H, MacKie RM, Melton DW. DNA repair gene polymorphisms and genetic predisposition to cutaneous melanoma. Carcinogenesis. 2007;28:1087–93. doi: 10.1093/carcin/bgl257. [DOI] [PubMed] [Google Scholar]
- 30.Goto M, Miller RW, Ishikawa Y, Sugano H. Excess of rare cancers in Werner syndrome (adult progeria). Cancer Epidemiol Biomarkers Prev. 1996;5:239–46. [PubMed] [Google Scholar]
- 31.Hanneman KK, Cooper KD, Baron ED. Ultraviolet immunosuppression: mechanisms and consequences. Dermatol Clin. 2006;24:19–25. doi: 10.1016/j.det.2005.08.003. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplemental Table 1. Candidate genes and pathways included in this study.