Genetic Variation in XPD, Sun Exposure, and Risk of Skin Cancer

The XPD gene is involved in the nucleotide excision repair pathway removing DNA photoproducts induced by UV radiation. Genetic variation in XPD may exert a subtle effect on DNA repair capacity. We assessed the associations between two common nonsynonymous polymorphisms (Asp³¹²Asn and Lys⁷⁵¹Gln) with skin cancer risk in a nested case-control study within the Nurses' Health Study (219 melanoma, 286 squamous cell carcinoma, 300 basal cell carcinoma, and 874 controls) along with exploratory analysis on the haplotype structure of the XPD gene. There were inverse associations between the Lys⁷⁵¹Gln and Asp³¹²Asn polymorphisms and the risks of melanoma and squamous cell carcinoma. No association was observed between these two polymorphisms and basal cell carcinoma risk. We also observed that the association of the ⁷⁵¹Gln allele with melanoma risk was modified by lifetime severe sunburns, cumulative sun exposure with a bathing suit, and constitutional susceptibility score (P for interaction = 0.03, 0.04, and 0.02 respectively). Similar interactions were also observed for the Asp³¹²Asn. Our data suggest these two XPD nonsynonymous polymorphisms may be associated with skin cancer risk, especially for melanoma.

Skin cancer is the most common neoplasm in Caucasians in the United States. The genotoxic effect of sunlight exposure has been clearly shown in the etiology of both melanoma and nonmelanocytic skin cancer (1-3). One important defense mechanism against skin cancer is the ability to repair DNA damage induced by UV light. It has been suggested that reduced DNA repair capacity (DRC) is a susceptibility factor predisposing individuals to skin cancer (4-7). The predominant form of UV-induced DNA damage is DNA photoproducts caused by the direct absorption of UVB by DNA. Cyclobutane pyrimidine dimers and pyrimidine (6-4) pyrimidone photoproducts constitute the two major DNA photoproducts (2). DNA photoproducts are mainly removed by the nucleotide excision repair (NER). The NER is a versatile repair system to remove a variety of bulky, helix-distorting lesions, including UV photoproducts and bulky adducts (8, 9). Individuals with xeroderma pigmentosum, deficient in the NER, have a >1,000-fold increased risk of skin cancer.

Human XPD maps to chromosome 19q13.3 and spans ∼54 kb. It comprises 23 exons and is 761 amino acids in length. The XPD gene encodes an ATP-dependent DNA helicase involved in the NER and in basal transcription as part of the transcription factor TFIIH. Disruption of the mouse Xpd gene results in preimplantation lethality (10). Mutations in the XPD gene lead to NER defects (11) and three clinical syndromes, Cockayne syndrome, xeroderma pigmentosum, and trichothiodystrophy, depending on the location of the mutation (9, 12). In addition, the XPD and p53 proteins can interact with each other to modulate apoptosis and the NER. The p53 binds and modulates the helicase activity of the TFIIH, and the repair of UV-induced dimers was attenuated in Li-Fraumeni syndrome cells (heterozygote p53 mutant; ref. 13). A deficiency in p53-mediated apoptosis was reported in XPD lymphoblastoid cell lines and fibroblasts from xeroderma pigmentosum patients with germ line mutations in the XPD gene (14, 15).

We evaluated two common nonsynonymous XPD polymorphisms (Asp³¹²Asn and Lys⁷⁵¹Gln) in relation to skin cancer risk in a nested case-control study within the Nurses' Health Study along with exploratory analysis on the haplotype structure of the XPD gene. We further investigated the hypothesis that XPD genetic variants modify the associations of sunlight-related risk factors with skin cancer risk.

The Nurses' Health Study was established in 1976, when 121,700 female registered nurses between ages 30 and 55 years completed a self-administered questionnaire on their medical histories and baseline health-related exposures. Updated information has been obtained by questionnaires every 2 years. Between 1989 and 1990, blood samples were collected from 32,826 of the cohort members. Eligible cases in this study consisted of women with incident skin cancer from the subcohort who gave a blood specimen, including squamous cell carcinoma (SCC) and basal cell carcinoma (BCC) cases with a diagnosis anytime after blood collection up to June 1, 1998 and melanoma cases (including in situ cases) up to June 1, 2000 with no previously diagnosed skin cancer. All available pathologically confirmed melanoma and SCC cases and 300 self-reported BCC cases randomly selected from ∼2,600 available self-reported BCC cases were included. The validity of self-report of BCC is high in this medically sophisticated population (90%; ref. 16). All the SCC and BCC cases had no history of melanoma diagnosis. A common control series (case/control, 1:1) was randomly selected from participants who gave a blood sample and were free of diagnosed skin cancer up to and including the questionnaire cycle in which the case was diagnosed. One control was matched to each case by year of birth (±1 year) and self-reported race (Caucasian, Asian, Hispanic, African American, and unknown). More than 95% of cases and controls were Caucasian. At the time we selected cases and controls, 47 cases and 69 controls were deceased. To obtain additional information by supplementary questionnaires, we randomly selected a second matched living control when the first control was deceased and collected supplementary questionnaires from these second living controls. The nested case-control study consisted of 219 melanoma cases (including 77 in situ cases), 286 SCC cases, 300 BCC cases, and 874 matched controls. We mailed to 758 living cases and 804 living controls a supplementary questionnaire on lifetime sun exposure and other skin cancer risk factors; 695 cases responded, 15 cases refused to participate, and 48 cases did not respond after three mailings (participation rate, 92%). Among controls, 713 responded, 9 refused, and 82 did not respond (participation rate, 89%). The study protocol was approved by the Committee on Use of Human Subjects of the Brigham and Women's Hospital (Boston, MA).

Information regarding skin cancer risk factors was obtained from the prospective biennial questionnaires and the retrospective supplementary questionnaire. Information on natural hair color and childhood and adolescent tendency to sunburn or tan was asked in the 1982 prospective questionnaire (ethnic group in the 1992 questionnaire). The retrospective supplementary questionnaire consisted of questions in three major areas: (a) pigmentation, constitutional, and susceptibility factors; (b) history of residence (states and towns), sun exposure habits, and severe sunburns at different ages; and (c) family history of skin cancer (father, mother, and siblings). In addition, the 11 states of residence of cohort members at baseline were grouped into three regions: Northeast (Connecticut, Massachusetts, Maryland, New Jersey, New York, and Pennsylvania), North Central (Michigan and Ohio), and West and South (California, Texas, and Florida).

To estimate sunlight exposure for each subject, a UV database for 50 U.S. states was developed. The database used reports from the Climatic Atlas of the United States, which reported mean daily solar radiation (in Langleys) at the earth's surface for weather stations around the country (17). The records of average annual solar radiation for January and July were extracted to represent winter and summer radiation, respectively. The mean solar radiation for each residence was derived from the UV values measured at the nearest weather station, and both summer and winter radiation indices were developed for each residence. A cumulative lifetime sun exposure was developed by combining the UV database and the information obtained from the supplementary questionnaires. Questions about sun exposure while wearing a bathing suit were used to define a cumulative lifetime intermittent (recreational) sun exposure variable for this behavior.

There are two common nonsynonymous polymorphisms in XPD (Asp³¹²Asn and Lys⁷⁵¹Gln). Because these two single nucleotide polymorphisms (SNP) are not in complete linkage disequilibrium in the database and have potentially functional relevance, we genotyped both of them in our study. The XPD gene was resequenced by the National Institute of Environmental Health Sciences Environmental Genome Project at the University of Washington (http://egp.gs.washington.edu/data/ercc2/) on a subset of 90 samples from the NIH DNA Polymorphism Discovery Resource (18). Among the 136 polymorphisms identified across the genomic region of the XPD gene, we selected 86 polymorphisms with allele frequency over 1%. Based on these 86 polymorphisms, seven common haplotypes (>2% frequency) were inferred by the partition-ligation expectation-maximization algorithm of Qin et al. (19). The threshold of 2% was set to ascertain potential common (>5%) Caucasian-specific haplotypes from this mixed population (20). Five haplotype-tagging SNPs (htSNP) were selected by the BEST algorithm (21) to tag these seven common haplotypes and were genotyped in this study.

Genotyping was done by the 5′ nuclease assay (TaqMan), using the ABI PRISM 7900HT Sequence Detection System (Applied Biosystems, Foster City, CA), in 384-well format. TaqMan primers and probes were designed using the Primer Express Oligo Design software version 2.0 (ABI PRISM). Laboratory personnel were blinded to case-control status and blinded quality-control samples were inserted to validate genotyping procedures; concordance for the blinded samples was 100%. Primers, probes, and conditions for genotyping assays are available on request.

To avoid potential population stratification, we excluded one Asian melanoma case and one control, one Hispanic SCC case and two controls, and one African American control. We used a χ² test to assess whether the XPD genotypes were in Hardy-Weinberg equilibrium and to determine Ps for differences in haplotype frequencies between cases and controls. Unconditional logistic regression was employed to calculate odds ratio (OR) and 95% confidence interval (95% CI) to assess the risk of skin cancer for XPD genotypes among all women. A test for trend was calculated across the three genotypes for each polymorphism. We used a likelihood ratio test to evaluate heterogeneity in the effects of the XPD genotypes on different types of skin cancer in polytomous logistic regression models (22). To summarize multiple variables, we constructed a multivariate confounder score to create a constitutional susceptibility score for skin cancer (23). Briefly, we applied the logistic regression coefficients from a multivariate model, including age, race, natural skin color, natural hair color, child or adolescent tendency to burn, and the number of palpably raised moles on arms, to each individual's values for the latter four of these variables and summed the values to compute a susceptibility risk score in the logit scale. We used median value of this score among controls to define women with low and high constitutional susceptibility. The number of severe lifetime sunburns and cumulative sun exposure with a bathing suit were categorized into tertiles based on the distribution of controls.

To evaluate interactions between the environmental exposures and the XPD genotypes, we modeled them as ordinal variables to test significance of a single multiplicative interaction term. We modeled the genotype into three levels (homozygous wild-type, heterozygotes, and homozygous variants). For environmental risk factors, we modeled the number of lifetime severe sunburns into three levels, cumulative sun exposure with a bathing suit into three levels, and constitutional susceptibility score as a dichotomous variable. All statistical tests were two sided.

Detailed description of characteristics of cases and controls will be reported elsewhere.⁷

In brief, the mean age at diagnosis of melanoma cases was 63.4 years and that of SCC and BCC cases was 64.7 and 64.0 years, respectively. Women in the West and South regions were more likely to be diagnosed with SCC or BCC compared with those in Northeast. Cases of each type of skin cancer were more likely to have used sunlamps or attended tanning salons. A family history of skin cancer was a risk factor for the three types of skin cancer. Cases of each type of skin cancer were more likely to have had higher cumulative sun exposure with a bathing suit, more lifetime severe sunburns that blistered, and higher constitutional susceptibility risk score (Table 1).

The genotype distributions of the two polymorphisms were in Hardy-Weinberg equilibrium among controls. The two polymorphisms were in linkage disequilibrium in controls (D′ = 0.77, P = 8.7 × 10⁻⁸⁴; r² = 0.50, P = 8.4 × 10⁻⁴³). Genotype concordance between the two polymorphisms was 76.0% in controls, which is consistent with a previous report (24). The linkage disequilibrium statistics and genotype concordance in each case series was similar to those in controls. Inverse associations of the ⁷⁵¹Gln and ³¹²Asn alleles with the risks of melanoma and SCC were observed (Table 2). No association was observed between these two polymorphisms and BCC risk. There was no significant heterogeneity in the main effect of each polymorphism on the three types of skin cancer (Table 2). We also evaluated the combined effect of both polymorphisms. We grouped our population into three genotype categories, double wild-type (carriage of both Asp/Asp and Lys/Lys genotypes), double homozygous variants (carriage of both Asn/Asn and Gln/Gln genotypes), and others. The results of this combined analysis were similar to the results of the individual polymorphism analysis (Table 2).

We evaluated gene-environment interactions between the two nonsynonymous polymorphisms and the number of lifetime severe sunburns, cumulative sun exposure while wearing a bathing suit, and constitutional susceptibility score on skin cancer risk (Table 3). An interaction was observed between the number of lifetime severe sunburns and the Lys⁷⁵¹Gln polymorphism on melanoma risk (P for interaction = 0.03; Table 3). The number of lifetime severe sunburns was significantly associated with an increased risk of melanoma among women with the ⁷⁵¹Lys/Lys genotype (≥5 versus never, OR, 3.26; 95% CI, 1.39-7.63), and this excess risk was attenuated among those who carried the Gln allele. The significantly inverse association of the ⁷⁵¹Gln allele with melanoma risk was limited to women with five or more lifetime sunburns (P for trend = 0.02), whereas no association between the ⁷⁵¹Gln allele and melanoma risk was observed among women who had four or fewer lifetime sunburns. A similar interaction pattern was also seen between the number of lifetime severe sunburns and the polymorphism Asp³¹²Asn on melanoma risk (P for interaction = 0.03).

We observed a similar interaction pattern between cumulative sun exposure with a bathing suit and the Lys⁷⁵¹Gln polymorphism on melanoma risk (Table 3). Cumulative sun exposure with a bathing suit was significantly associated with an increased risk of melanoma among women with the ⁷⁵¹Lys/Lys genotype (highest tertile versus lowest tertile, OR, 4.84; 95% CI, 2.16-10.82), which was attenuated among those who carried the Gln allele. We observed an inverse association of the ⁷⁵¹Gln allele with melanoma risk among the women in the highest tertile of sun exposure with a bathing suit (P for trend = 0.06) but not among those in the lowest or intermediate tertile (P for interaction = 0.04). No significant interaction was observed between the Asp³¹²Asn polymorphism and cumulative sun exposure with a bathing suit on melanoma risk.

We also observed interactions between the constitutional susceptibility score and both polymorphisms Lys⁷⁵¹Gln and Asp³¹²Asn on melanoma risk (Table 3). A significantly inverse trend between the ⁷⁵¹Gln allele and melanoma risk (P for trend = 0.008) was only seen among women with low susceptibility score but not among those with high score (P for interaction = 0.02). A similar interaction pattern was observed for the Asp³¹²Asn polymorphism (P for interaction = 0.08).

We also evaluated the combined effect of both polymorphisms in the gene-environment interactions. Compared with the individual polymorphism, similar patterns of interactions were observed when we assessed the interactions of three genotype groups (double wild-type, double homozygous variants, and others) with history of sunburns (P for interaction = 0.007), cumulative sun exposure while wearing a bathing suit (P for interaction = 0.11), and constitutional susceptibility score (P for interaction = 0.14).

There was no significant difference in the Lys⁷⁵¹Gln and Asp³¹²Asn genotype distributions in different categories of exposures among controls (i.e., the interactions we observed were not driven by controls). In addition, we did not observe any interaction between the genotypes and geographic region on melanoma risk. No significant interactions were observed between the two polymorphisms and the above exposure risk factors on SCC or BCC risk.

We did exploratory analysis of the XPD haplotype structure. There appears two haplotype blocks in the XPD gene based on the resequencing data on 90 samples from the Environmental Genome Project. We evaluated the long-range haplotype across the two blocks. Five htSNPs were selected, two htSNPs in the first block (G1305C and G7988A), one in the second block (T19938C), and two in the linker region between the two blocks (T12733G and C12799T). Nine common inferred haplotypes with allele frequency >5% in controls accounted for 83.3% of the alleles of controls in the present study (Table 4). We observed that three different common haplotypes were less common in the cases of melanoma, SCC, and BCC than controls, respectively.

In this nested case-control study, we observed that the XPD ⁷⁵¹Gln allele was associated with decreased risks of melanoma and SCC along with finding that the effect of the ⁷⁵¹Gln allele on melanoma risk was modified by the number of lifetime severe sunburns, cumulative sun exposure while wearing a bathing suit, and constitutional susceptibility score. Similar main effect and interactions were also observed for the Asp³¹²Asn in melanoma risk. The nested case-control design, high follow-up rate, and high response rate for the retrospective supplementary questionnaire strengthen the validity of this study.

Several studies have examined the functional significance of the two XPD nonsynonymous polymorphisms Asp³¹²Asn and Lys⁷⁵¹Gln using DNA adducts and UV-induced photoproducts, which are mainly repaired by the NER. The ³¹²Asn and ⁷⁵¹Gln alleles were associated with decreased DRC for removal of UV photoproducts (25) and benzopyrene diol epoxide-DNA adducts (26) by host cell reactivation assays. Hemminki et al. (27) reported that the combined ³¹²Asn/Asn and ⁷⁵¹Gln/Gln genotype was associated with reduced DRC of UV-induced cyclobutane dimers in human skin in situ and the ⁷⁵¹Gln allele was associated with reduced repair among individuals ages ≥50 years. Consistent with these observations, a positive association was observed of the ³¹²Asn or ⁷⁵¹Gln allele with increased aromatic DNA adduct level in peripheral lymphocytes (28). Seker et al. (29) reported that the ³¹²Asn allele was associated with enhanced apoptotic response to UV. These authors did not detect any effect of the two polymorphisms in in vitro p53 binding or DRC. Nevertheless, the increased UV-induced apoptosis may reflect slightly impaired DRC (i.e., the subtle reduction in DRC may cause the accumulation of excess DNA damage and in turn trigger apoptosis). It was reported that unrepaired UV-induced damage on the transcribed strand blocks transcription, triggers the accumulation of p53, and induces apoptosis (30).

We observed that the ⁷⁵¹Gln allele was associated with a decreased risk of melanoma, and this decreased risk was more apparent among women with five or more severe lifetime sunburns or those in the highest tertile of cumulative sun exposure with a bathing suit. Similar interactions were also observed for the ³¹²Asn allele. A suggestive positive association of the variant allele with melanoma risk was observed among women who never had severe blistering sunburns or were in low category of sun exposure. Given previous data suggesting the reduced DRC of the ⁷⁵¹Gln and ³¹²Asn alleles, our data showed that the positive association of the variant allele and melanoma risk occurred in the context of low levels of DNA damage. However, when challenged by an overwhelmingly high dose of exposure measured as five or more sunburns or high level of cumulative sun exposure with a bathing suit, melanocytes with impaired DRC may accumulate excess DNA damage, inducing apoptosis and in turn decreasing the risk of melanoma.

We also observed a significant interaction between the constitutional susceptibility score and the ³¹²Asn and ⁷⁵¹Gln variants. A significant association of the ⁷⁵¹Gln or ³¹²Asn allele with melanoma risk was seen among women with low score but not among those with high score. In addition, we also observed a decreased risk of SCC associated with the ⁷⁵¹Gln allele.

Although melanocytes are less sensitive to apoptosis than keratinocytes because of the lower levels of cell cycle and proliferation and the higher content of antiapoptotic proteins (31, 32), we observed that the ³¹²Asn and ⁷⁵¹Gln allele were associated with a decreased risk of melanoma potentially due to apoptosis. As suggested by the previous functional data (29), the two ³¹²Asn and ⁷⁵¹Gln alleles, which are in linkage disequilibrium, may enhance the cellular apoptotic response to UV exposure. This is consistent with our data that the decreased risk was stronger among women with high sun exposure as measured by lifetime sunburns and cumulative sun exposure while wearing a bathing suit.

We did exploratory analysis of XPD long-range haplotype. Across two haplotype blocks, we inferred nine common haplotypes (>5% allele frequency) from five htSNPs. We used the data on 90 multiethnic samples from the Environmental Genome Project to infer haplotype structures. Two recent studies of haplotype variation in different ethnic groups (33, 34) reported substantial conservation of haplotypes among ethnicities with the fewest population-specific common haplotypes in Caucasians. In addition, using the BEST algorithm to tag the haplotypes of 105 genes, Sebastiani et al. showed that an average of 95% (and in most cases 100%) of the htSNPs in the European American sample is a subset of the htSNPs of the African American sample (21). These suggest that use of the data from the Environmental Genome Project to infer haplotype in our study of mainly Caucasian individuals is appropriate to define common Caucasian haplotypes. We observed that three different common haplotypes were less frequent in the cases of melanoma, SCC, and BCC risk than controls, respectively. Block-specific haplotype analysis is under investigation to better understand the associations of the polymorphisms in the regulatory regions of the XPD gene.

In summary, this is the first report of interactions between two XPD nonsynonymous polymorphisms (Asp³¹²Asn and Lys⁷⁵¹Gln) and sun exposure on skin cancer risk. We observed an inverse association of the XPD ⁷⁵¹Gln allele on melanoma and SCC risk and effect modification of the ⁷⁵¹Gln allele on sun exposure related factors for melanoma risk along with the similar findings for the Asp³¹²Asn. Additional functional data on XPD polymorphisms are warranted to elucidate the role of these variants in the development of skin cancer.

We thank Dr. Hardeep Ranu and Craig Labadie for their laboratory assistance, Rong Chen for her programming support, Dr. David Cox for his program to calculate linkage disequilibrium statistics, and the participants in the Nurses' Health Study for their dedication and commitment.