Skip Navigation


Carcinogenesis Advance Access originally published online on January 8, 2007
Carcinogenesis 2007 28(5):1087-1093; doi:10.1093/carcin/bgl257
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow Supplementary Data
Right arrow All Versions of this Article:
28/5/1087    most recent
bgl257v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (1)
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by Povey, J. E.
Right arrow Articles by Melton, D. W.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Povey, J. E.
Right arrow Articles by Melton, D. W.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

© The Author 2007. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oxfordjournals.org

DNA repair gene polymorphisms and genetic predisposition to cutaneous melanoma

Joanne E. Povey, Fatemeh Darakhshan, Karen Robertson1, Yvonne Bisset1, Magda Mekky2, Jonathan Rees1, Val Doherty1,3, Gina Kavanagh1,3, Niall Anderson2, Harry Campbell2, Rona M. MacKie3,4 and David W. Melton*

Sir Alastair Currie Cancer Research UK Laboratories, Molecular Medicine Centre, University of Edinburgh, Western General Hospital, Crewe Road, Edinburgh, EH4 2XU, UK
1 Department of Dermatology, University of Edinburgh, Lauriston Building, Lauriston Place, Edinburgh EH3 9HA, UK
2 Public Health Sciences Section, Division of Community Health Sciences, University of Edinburgh, Medical School, Teviot Place, Edinburgh EH8 9AG, UK
3 On behalf of the Scottish Melanoma Group
4 Department of Public Health and Health Policy, University of Glasgow, Glasgow G12 8RZ, UK

* To whom correspondence should be addressed. Tel: +44 0 131 651 1079; Fax: +44 0 131 651 1072; Email: david.melton{at}ed.ac.uk


   Abstract
Top
 Abstract
Introduction
 Materials and methods
 Results
 Discussion
 Supplementary material
 References
 
The incidence of cutaneous melanoma is rising rapidly in a number of countries. The key environmental risk factor is exposure to the ultraviolet (UV) component in sunlight. The nucleotide excision repair (NER) pathway deals with the main forms of UV-induced DNA damage. We have investigated the hypothesis that polymorphisms in NER genes constitute genetic susceptibility factors for melanoma. However, not all melanomas arise on sun-exposed sites and so we investigated the hypothesis that genes involved in other pathways for the repair of oxidative DNA damage may also be involved in susceptibility to melanoma. Scotland, with its high incidence of melanoma and stable homogeneous population, was ideal for this case–control study, involving 596 Scottish melanoma patients and 441 population-based controls. Significant associations were found for the NER genes ERCC1 and XPF, with the strongest associations for melanoma cases aged 50 and under [ERCC1 odds ratio (OR) 1.59, P = 0.008; XPF OR 1.69, P = 0.003]. Although an XPD haplotype was associated with melanoma, it did not contain the variant 751 Gln allele, which has been associated with melanoma in some previous studies. No associations were found for the base excision repair and DNA damage response genes investigated. An association was also found for a polymorphism in the promoter of the vitamin D receptor gene, VDR (OR 1.88, P = 0.005). The products of the two NER genes, ERCC1 and XPF, where associations with melanoma were found, act together in a rate-limiting step in the repair pathway.

Abbreviations: CI, confidence interval; NER, nucleotide excision repair; NMSC, non-melanoma skin cancer; OR, odds ratio; PCR, polymerase chain reaction; RFLP, restriction fragment length polymorphism; SNP, single-nucleotide polymorphism; UV, ultraviolet


   Introduction
Top
 Abstract
 Introduction
Materials and methods
 Results
 Discussion
 Supplementary material
 References
 
Melanoma is the most lethal form of skin cancer and an increasingly common disease worldwide (1). Cutaneous melanoma is the eighth most common cancer in the UK and the second commonest in young people (aged 20–39). In Scotland the incidence of melanoma is rising rapidly: from 3.5 in 1979 to 10.6 per 105 population in 1998 for men (300% increase) and from 7.0 to 13.1 for women (180% increase) (2).

The key environmental risk factor is exposure to the ultraviolet (UV) component in sunlight, with incidences of sunburn during early childhood being particularly important (3). The nucleotide excision repair (NER) pathway deals with the main types of UV-induced DNA damage (cyclobutane pyrimidine dimers and 6–4 photoproducts). Studies on the human-inherited disease xeroderma pigmentosum identified complete NER deficiency as a dramatic genetic risk factor for skin cancer (4). Reduced NER capacity has also been associated with an increased risk of melanoma (5). We wished to investigate the hypothesis that NER gene polymorphisms constitute genetic susceptibility factors for melanoma.

In a small pilot case–control study, we found an association between three XPD polymorphisms and melanoma (6). Subsequently, other case–control studies have investigated possible associations for XPD and other NER gene polymorphisms with melanoma. Significant associations have been reported for XPD (7–10), XPC (11) and XPF (12). Other studies found no association for XPD (12), XPC (9,13), XPF (9), XPG (9,13), ERCC1 (12), CSB and HR23B (9). Thus, the situation is confused and additional large studies are required to resolve the role of NER genes in melanoma.

UV causes many types of mutation, but C-T and CC-TT are regarded as UV signature mutations. UV signature mutations will only be seen in melanoma if C residues occur in functionally critical areas of the genome. Not all melanomas arise on sun-exposed sites and not all mutations identified in melanoma have a UV signature. This raises the issue of the importance of other forms of DNA damage and repair pathways in melanoma. The commonest mutation in the BRAF gene, found in both sporadic melanoma and benign naevi, does not have a UV signature (14), indicating that other lesions, possibly UVA-induced oxidative lesions, may also be involved. An association has been reported for the XRCC1 gene in a subset of non-melanoma skin cancer (NMSC) patients with multiple incidences of sunburn (15), although no association was found for melanoma (12,16,17). This gene is involved in the base excision repair pathway for the removal of oxidative DNA damage. 8-Oxoguanine glycosylase (OGG1) is also involved in base excision repair of the commonest oxidized base 8-oxoguanine. A codon 326 variant has been associated with elevated rates of lung and prostate cancer (reviewed in ref. 18) and, in one study, with increased sensitivity to cytotoxic agents (19). Its possible involvement in melanoma merits investigation. One study (12), but not others (20–22), has also reported an association between the recombinational repair gene XRCC3 and melanoma.

In addition to DNA repair genes themselves, many other genes are involved in determining the consequences of UV-induced and spontaneous oxidative DNA damage and may therefore also be involved in susceptibility to melanoma. This is most clearly demonstrated by the frequent occurrence of mutations in the CDKN2A and CDK4 genes in familial melanoma (reviewed in ref. 23). The normal form of the p53 protein induces cell-cycle arrest and activates target DNA repair genes more effectively than the codon 72 arginine variant (24,25); this polymorphism has been associated with increased susceptibility to melanoma (26,27). UV irradiation generates reactive oxygen species that can damage DNA. Glutathione S-transferases contribute to the defence against oxidative stress. A null allele of the GSTT1 gene has been associated with increased risk of NMSC (reviewed in ref. 28) and with increased sunburn sensitivity (29). Its possible involvement in melanoma merits investigation. Although not related to DNA repair, we have also investigated the vitamin D receptor (VDR) gene, which has an important role in skin biology and where an association between a polymorphism in the promoter region and melanoma has been reported (30).

Scotland is a particularly suitable location to investigate genetic susceptibility to melanoma. In addition to the high incidence, the population is relatively homogeneous and stable compared with some other parts of the UK (31). Finally, the Scottish Melanoma Group has collected information on all melanoma cases in Scotland since 1979. Here, we report the results of a large case–control study involving 596 Scottish melanoma patients and 441 population-based controls. In addition to analysing all cases, we also analysed cases aged 50 and under because genetic effects are often more easily detected in younger cases (32).


   Materials and methods
Top
 Abstract
 Introduction
 Materials and methods
Results
 Discussion
 Supplementary material
 References
 
Participants and study design
This was a case–control study involving 596 melanoma patients recruited from dermatology services in Lothian region (referred to as Edinburgh cases) and the west of Scotland (referred to as Glasgow cases) and 441 controls matched for age (±7 years) and sex and residing not only predominantly in the same regions but also from other main population centres in mainland Scotland. The parents of 78% of cases and 76% of controls were both born in Scotland. Patients were interviewed by a research nurse and recruited between November 2001 and November 2004. The inclusion criteria for cases were patients attending dermatology clinics, either for the first time with a primary cutaneous melanoma or for subsequent follow-ups following surgery. Thus, they were considered as incident cases. No patients were excluded from study participation. The controls were recruited from an existing study by the Medical Research Council national DNA case–control collection for colorectal cancer. This study used the Community Health Index, a Scottish index of individuals registered with a general practitioner and estimated to be >95% complete, as a source of 2500 healthy Scottish inhabitants. Individuals were selected at random from the index by the Information and Statistics Division of the National Health Service in Scotland and invitations to participate were sent via their general practitioner. A standard questionnaire was administered to participants to assess control status. In addition, all controls were linked to the cancer registry so that their cancer status could be checked. Twenty millilitres of blood was collected in two aliquots from both cases and controls and stored for subsequent DNA analysis.

DNA isolation
Case and control samples were handled similarly and analysed contemporaneously. DNA was isolated from each 10 ml blood sample using a Nucleon BACC3 kit and the manufacturer's protocol (GE Healthcare Bio-Sciences, Amersham, UK).

Genotyping
Genotyping was carried out by polymerase chain reaction (PCR) and restriction fragment length polymorphism (RFLP) assays. Details of the genes, single-nucleotide polymorphisms (SNPs) and RFLP assays are given in Table I. SNPs were chosen that had a variant allele frequency >0.1, could be detected by RFLP assay and, ideally, resulted in a non-synonymous (coding) change, although, of course, silent (synonymous) changes are equally useful for association studies. The exception was the GSTT1 gene, where the null allele results from a large deletion and separate PCR assays were used for the active and null alleles. PCRs (50 µl) were carried out on genomic DNA (~100 ng) using Taq DNA polymerase (Promega, Southampton, UK), according to the manufacturer's instructions. Details of the primers and cycle conditions for each assay are given in Supplementary Material (supplementary Table I). Twenty microlitres of each PCR was then made up to 50 µl, digested with the appropriate restriction enzyme according to the manufacturer's instructions (New England Biolabs, Hitchin, UK) and subjected to agarose gel electrophoresis. Genotypes were assigned by a member of the research team blinded to the status of the samples. Genotypes were verified by a second member of the team and any discrepancies resolved by repeating the genotyping process.


View this table:
[in this window]
[in a new window]

  		Table I. RFLPs studied

 
Statistical analysis
Cases and controls were examined by calculating summary measures to assess comparability on the basis of age and gender. A {chi}2 approximation test for Hardy–Weinberg equilibrium (33) was carried out on all the polymorphisms to determine whether the chromosomes could be assumed to be independent, both to simplify and to increase the power of the study. The genes were then investigated individually for association between genotype and disease status using {chi}2 techniques. The {chi}2 exact test was used in some cases throughout the study when predicted values were small (i.e. <5) and might otherwise have altered the accuracy of the {chi}2 approximation. Binary logistic regression was used to correct for age and sex. XPD haplotype analysis was conducted using the Web-based program, ‘Haplotype Resolution using Imperfect Phylogeny’ (34). This program infers haplotypes from genotype input and was selected for its ease of use, reputation and popularity.

The possibility of gene–gene interaction was investigated by utilizing the Classification and Regression Tree method. This technique was used as an exploratory tool in order to identify rationally the combinations of genes that would probably be of interest for further investigation in a model generated for the data. The size of the tree was limited by cost-complexity pruning for clarity and in order to show the most important splits. A plot of the complexity parameters against deviance was used in order to generate the tree with a moderate number of nodes. The cost-complexity parameter was chosen at the point where little additional deviance was explained by including subsequent nodes and a value of 2.6 was ultimately chosen for the final product. Genes that might have been interacting were hypothesized on the basis of early splits in the tree. Combinations of two and up to a maximum of three genes were selected (using more than this could have created issues with insufficient sample size) from the tree for further testing using binary logistic regression.

All P values reported are two-sided. All analyses were conducted in SPSS version 11.5, with the exception of the Classification and Regression Tree analysis, which was generated in S-Plus version 6.0.


   Results
Top
 Abstract
 Introduction
 Materials and methods
 Results
Discussion
 Supplementary material
 References
 
The case and control groups were first compared for age and gender distribution since any differences would need to be considered as possible confounders later in the study (see Tables II and III). The proportion of males was 43% for all cases and controls; for subjects aged 50 and under, 40% of cases and 38% of controls were male. Male participants were on average 2.4 years older and the mean age of the control group was 2.9 years older than cases. Both these differences were significant. Although the gender division between Edinburgh cases, Glasgow cases and controls was significant for all subjects and subjects aged 50 and under, there was no significant difference in the gender distribution between case and control groups. This was the division used throughout the rest of the study.


View this table:
[in this window]
[in a new window]

  		Table II. Comparison of study subjects' age by category

 


View this table:
[in this window]
[in a new window]

  		Table III. Comparison of study subjects' gender by category

 
Each polymorphism was checked in cases and controls for Hardy–Weinberg equilibrium. All were found to be in equilibrium with the exception of ERCC1 exon 4 controls (P = 0.008). However, use of the Bonferroni correction for the 24 tests would indicate that P < 0.002 was needed to reject Hardy–Weinberg equilibrium. Re-examination of the original gels for this polymorphism revealed no genotyping errors, so it seems likely that this result was due to chance.

Single-gene {chi}2 association tests were then carried out on all subjects and on subjects aged 50 and under using the case and control genotype frequencies for each of the polymorphisms (Table IV). For all subjects, only the XPF exon 11 polymorphism was associated with disease status (P = 0.011). After logistic regression correction for age and sex imbalance between cases and controls, this association strengthened (P = 0.004). For subjects aged 50 and under, the XPF exon 11 polymorphism was again associated with melanoma (P = 0.003 and P = 0.005 after logistic regression correction). For subjects aged 50 and under, the ERCC1 exon 4 polymorphism was also associated with melanoma (P = 0.003) and the association strengthened after logistic regression correction (P = 0.001). In addition, for subjects aged 50 and under, the VDR promoter polymorphism was weakly associated with melanoma (P = 0.017), but the association disappeared after logistic regression correction (P = 0.140). Use of the Bonferroni correction for the 10 genes tested would indicate that P < 0.005 is needed to achieve significance. No other significant single-gene associations were found.


View this table:
[in this window]
[in a new window]

  		Table IV. Genotype and allele frequencies for RFLPs

 
Odds ratios (ORs) were calculated for the four associations found after pooling genotypes (supplementary Table II). For the ERCC1 exon 4 polymorphism, genotypes containing the G allele were significantly over-represented in melanoma cases aged 50 and under (OR 1.59, P = 0.008). For the XPF exon 11 polymorphism, genotypes containing the C allele were significantly over-represented in all melanoma cases (OR 1.31, P = 0.041) and in cases aged 50 and under (OR 1.69, P = 0.003). For the VDR promoter polymorphism, genotypes containing the A allele were significantly over-represented in melanoma cases aged 50 and under (OR 1.88, P = 0.005).

No association was found between the XPD exon 23 Lys 751 Gln polymorphism and melanoma. Since we had typed two other SNPs in this gene, it was also possible to test for association between melanoma and the most common XPD haplotypes. Studying haplotypes can give an indication of SNPs that are interacting or provide an increase in power if the disease-causing variant is part of a haplotype block in which all SNPs are in strong linkage disequilibrium with each other. Haplotypes were compared on the basis of having the haplotype of interest as a homozygote or heterozygote or not having it at all against disease status. They were also compared on the basis of having the haplotype of interest (in homozygote or heterozygote form), or not having it at all, against disease status in order to attempt to identify deleterious or protective haplotypes (Table V). XPD haplotypes are presented as XPD exon 6, XPD exon 22 and XPD exon 23 (e.g. ACA = XPD exon 6 A, XPD exon 22 C and XPD exon 23 A).


View this table:
[in this window]
[in a new window]

  		Table V. Association analysis between XPD haplotypes and disease status

 
Results indicated a strong association between the ACA (P < 0.001), CCA (P = 0.040) and CTC (P = 0.007) haplotypes and disease status when comparing cases versus controls for the homozygotes, heterozygotes and lack of haplotype of interest. Use of the Bonferroni correction for multiple testing would indicate that P < 0.006 is needed to achieve significance. Similar results were obtained when comparing the cases and controls for the presence or lack of the haplotype of interest. Having the ACA haplotype increased the odds of being a case (OR 1.46, P = 0.007). In contrast, having the CCA haplotype (OR 0.72, P = 0.014) or the CTC haplotype (OR 0.68, P = 0.002) had a protective effect. ATC and CCC haplotypes were not tested comparing both classes of homozygotes and heterozygotes against disease status because there were too few homozygotes for such analysis. Presence or absence of the ATC and CCC haplotypes was not found to have a statistically significant effect on the OR.

Finally, the possibility of gene–gene interactions was investigated. The Classification and Regression Tree program was used to generate a classification tree for the genotype information. The pruning of the tree was based on the need for balance between clarity and the desire not to lose important detail. The tree indicated that it would be prudent to investigate interactions between XPG exon 15 and VDR promoter, as well as XPG exon 15 and XRCC3 exon 7. The interaction of XPG exon 15, XRCC3 exon 7 and XPD exon 6 would also be a valid option for investigation. Upon model building using binary logistic regression and correcting for age and sex, the interaction of XPG exon 15 and VDR promoter was not found to be significant (P = 0.152), neither was the interaction between XPG exon 15 and XRCC3 exon 7 (P = 0.157). The XRCC3 exon 7 and XPD exon 6 interaction was also not significant (P = 0.768). Thus, no evidence for significant gene–gene interaction with melanoma was found.


   Discussion
Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
Supplementary material
 References
 
The relatively small size of many of the previous studies of NER gene polymorphisms and melanoma and the lack of carefully matched population-based controls could explain the contradictory results. Our small pilot study involved only 28 patients with primary melanoma who were <50 years of age and 33 controls from blood donations collected at the same hospital (6). The Oxford study involved 125 patients with late-stage melanoma and 211 controls taken from cadaveric kidney transplant donors (12). The Italian study involved 176 melanoma patients with 177 controls consisting mostly of spouses and friends of patients, but also out-patients and staff at the same hospital (7). The German study involved 294 melanoma patients with 375 controls consisting of local blood donors and health care staff (13). The American Nurses’ Health study involved 219 female nurses who developed melanoma and 873 unaffected nurse controls (16). The Polish study involved 426 melanoma patients and 421 controls (10). One notable exception is the recently reported Genes Environment and Melanoma Study (9), which involved 2485 controls with a single primary melanoma and 1238 cases with second, or higher order, primary melanomas collected from centres in Australia, Canada, Italy and the USA. However, this was not a conventional case–control study in that controls were patients with a single primary melanoma and cases were patients with additional primary melanomas.

Our study is larger than other conventional case–control studies for melanoma and is also notable for the controls being selected from a large population-based registry covering the same geographical area. The homogeneous nature of our sample set is indicated by the equivalent high percentage of both cases and controls with both parents born in Scotland, rendering population stratification an unlikely explanation for the associations found. The study was powered to detect an OR of 1.5. Variants showing lack of association could have small effects that this study was not powered to detect. Although the study was large, we carried out many association tests on numerous loci, so there is a chance of false positives due to multiple testing. However, we consider that applying the Bonferroni corrections given in the Results section would be over-conservative since we have conducted a candidate gene study focussed on the specific hypothesis that DNA repair and cell-cycle control genes are genetic risk factors for melanoma rather than an exploratory search. Information on known melanoma risk factors was not analysed in this study. Controlling for such potential confounders could have strengthened the results obtained.

Only two of the NER gene polymorphisms we studied resulted in non-synonymous protein changes: XPD Lys 751 Gln and XPG Asp 1104 His. Functional assays have not revealed a consistent difference between the XPD alleles (reviewed in ref. 35), whereas no functional studies have been reported for the XPG alleles. Of course, both silent and non-synonymous changes are equally useful for association studies, which provide no information about the functional role of a particular allele.

Our small pilot study identified an association between three different polymorphisms in the XPD gene and melanoma (6). The exon 23 Lys 751 Gln polymorphism (rs13181) has been the subject of a number of subsequent studies. Our original pilot study found an association between the Lys 751 allele and melanoma [OR 2.8, 95% confidence interval (CI) 1.2–7.0, P = 0.02]. In subsequent studies, where an association has been reported, it has mostly been for the variant Gln 751 allele. Baccarelli et al. (7) reported an association between the Gln 751 allele and melanoma in older patients (OR 2.3, 95% CI 1.1–4.9, P = 0.03). However, Han et al. (8) reported an inverse association between the Gln 751 allele and melanoma (OR 0.63, 95% CI 0.38–1.05, P = 0.09). The Genes Environment and Melanoma study (9), found an association between the XPD 751 Gln/Gln genotype and melanoma (OR 1.4, 95% CI 1.1–1.7, P = 0.004). The same study also reported an association with another XPD polymorphism Asp 312 Asn, where the variant 312 Asn/Asn genotype was associated with melanoma. In addition the Asn 312 Gln 751 haplotype was significantly more frequent in cases than controls. Other studies have not reported an association between single XPD polymorphisms and melanoma (10,12), although Debniak et al. (10) did report that XPD haplotypes containing Gln 751 were associated with melanoma.

This study failed to find an association between three separate XPD polymorphisms [synonymous changes in exon 6 (rs238406) and 22 (rs1052555) and the non-synonymous Lys 751 Gln change] and melanoma. However, an XPD haplotype at exons 6, 22 and 23 was significantly over-represented in melanoma cases, but this haplotype incorporated 751 Lys rather than 751 Gln. Thus, overall the data suggest that, if XPD polymorphisms are involved in susceptibility to melanoma, any effects are at best modest and that larger and better controlled studies than most of those reported to date are needed to detect them.

In our study, modest, but significant, associations were found for two synonymous changes, in ERCC1 exon 4 (rs11615) and XPF exon 11 (rs1799801). We are encouraged by our finding that the associations were stronger in melanoma patients aged 50 and under, despite this group only constituting 56% of total patients (ERCC1 exon 4: OR 1.59, 95% CI 1.11–2.27, P = 0.008; XPF exon 11: OR 1.69, 95% CI 1.18–2.43, P = 0.003). One previous study has reported an association between a different XPF polymorphism (rs1799797) and melanoma (OR 1.65, 95% CI 1.03–2.66, P = 0.038) (12). Another study failed to find an association for rs1799797 and a third XPF polymorphism, rs1800067 (9). The ERCC1 exon 4 polymorphism has been investigated before, in our small pilot study, where we failed to find an association with melanoma (6), and in a separate study (12), where no association was found. Clearly, interpretation of this result is complicated by the difference in goodness of fit to Hardy–Weinberg equilibrium between cases and controls for ERCC1 exon 4; however, given that ERCC1 is essential for the repair process that protects against the key environmental risk factor for melanoma, further investigation of this polymorphism would still seem to be warranted.

The two NER genes ERCC1 and XPF, where we have found an association with melanoma, act together in a complex to make an incision to the 5' side of a DNA lesion, after another endonuclease, XPG, has made an initial incision to the 3' side of the lesion. This could constitute a particularly critical stage of the repair process, after a single-strand break has been introduced into the DNA, where minor changes in the efficiency of making the second incision by ERCC1/XPF or in immediate processing at the incision sites to reseal the DNA could result in an increased frequency of mutagenesis or genome instability and thus an increased risk of melanoma. Indeed, in this regard, it has been reported that levels of ERCC1/XPF are rate limiting for NER (36). Given that UV is also the key environmental risk factor for NMSC, similar associations for these ERCC1 and XPF polymorphisms might be anticipated. However, the situation may be complicated by the different UV exposure risk profiles for the diseases (cumulative for some forms of NMSC, intermittent for melanoma). There is one report of an association between the ERCC1 exon 4 polymorphism and basal cell carcinoma, but with the A allele (OR 12, 95% CI 1.17–124, P = 0.019) rather than the G allele as reported here (37).

No significant interactions were found between combinations of the DNA repair and DNA damage response genes in our study. Despite the moderately large sample size, the study had limited power to assess this. The major weakness of this interaction analysis is that models with interaction terms involving more than three genes could not be fitted. The larger the number of genes involved in the interaction term, the greater the size of the data set required to ensure that each combination of genotypes is represented. Thus, it was impossible in this data set, and is extremely difficult in general, to model more complex interaction patterns.

Association studies cannot identify causative genetic changes. Both the XPF and ERCC1 associations identified involve synonymous changes, making them unlikely to be causative unless they affect RNA stability or processing. More likely, they are in linkage disequilibrium with adjacent causative changes. This could be investigated by selecting additional SNPs in the vicinity for additional single-gene association and haplotype studies.

We found no association between the XPG exon 15 Asp 1104 His polymorphism (rs17655) and melanoma, confirming results reported for the same polymorphism (9,13). Concerning other NER genes not included in our study, Blankenburg et al. (13) reported no association for three XPC polymorphisms, G1580A, T1601C and G2166A, but association was found for three other XPC markers, intron 9 PAT, intron 11-6A and Lys 939 Gln (11). No association for XPC Lys 939 Gln was found by Millikan et al. (9), who also found no association for two SNPs (Arg 1213 Gly and Arg 1230 Pro) in CSB and one SNP (Ala 249 Val) in HR23B.

We found no evidence for a role for genes involved in other DNA repair pathways in melanoma. We were particularly interested in the possibility that genes in the base excision repair pathway, responsible for the repair of UVA-induced oxidative DNA damage, might be genetic susceptibility factors for melanoma because the commonest BRAF mutation in sporadic melanoma does not have a classic UVB signature mutation (14), although an indirect mechanism by which BRAF mutations could be generated as a result of UVB-induced cyclobutane pyrimidine dimer formation and error-prone translesion synthesis has been proposed (38). No association was found for XRCC1 exon 10 Arg 339 Gln (rs25487), confirming previous studies (12,16,17), nor for OGG1 exon 7 Ser 326 Lys (rs1052133), which has not been investigated before for a role in melanoma. We also found no association for a gene involved in the homologous recombination repair pathway, XRCC3 exon 7 Thr 241 Met (rs861539). One previous study had reported an association with melanoma (OR 2.36, 95% CI 1.44–3.86, P = 0.0004) (12), but other studies found no association (20–22).

Turning to other pathways involved in protective responses to UV-induced DNA damage, the p53 codon 72 Arg variant (rs1042522) has been associated with increased susceptibility to melanoma (OR 1.43, 95% CI 1.03–1.99, P = 0.031) (26). In another study, the 72 Pro form showed association (OR 1.58, 95% CI 0.85–2.95, P = 0.07) (27). We found no evidence for any p53 codon 72 association in our study. The variant form of p53 has been shown previously to induce cell-cycle arrest and activate p53-dependent DNA repair target genes less effectively than the normal form (24,25). Similarly, although the glutathione S-transferases protect against oxidative stress and the GSST1 null allele is linked with increased risk of NMSC (reviewed in ref. 28), we found no evidence for an association with melanoma. A small Slovenian study also found no association (39).

Finally, in melanoma cases aged 50 and under, an association with a polymorphism in the VDR promoter (rs4516035) was seen, but the association disappeared after logistic regression correction. Genotypes containing the A allele were significantly over-represented in melanoma (OR 1.88, P = 0.005). A previous smaller study (174 melanoma patients, not stratified by age and 80 controls) found that the AA (OR 3.3, P = 0.007) and AG genotypes (OR 2.5, P = 0.03) were both over-represented in melanoma, and it was suggested that altered transcription of the VDR gene may influence the immune response to cancer (30). Other polymorphisms in the VDR gene have recently been investigated. In one report, association with melanoma was found (40), whereas in the other no association was seen (41).

Although significant, the effects we have reported for NER genes are modest by comparison with those for well-established, low-penetrance melanoma susceptibility genes. For instance, having two red hair colour variants of the MC1R gene conferred an OR of 4.8 (42).


   Supplementary material
Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Supplementary material
References
 
Supplementary tables I and II can be found at http://carcin.oxfordjournals.org/


   Acknowledgments
 
This work was supported by a project grant (CZB/4/248) from the Chief Scientist Office to D.W.M.

Conflict of Interest Statement: None declared.


   References
Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Supplementary material
 References
 

  1. Kamangar F, et al. (2006) Patterns of cancer incidence, mortality, and prevalence across five continents: defining priorities to reduce cancer disparities in different geographic regions of the world. J. Clin. Oncol. 24:2137–2150.[Abstract/Free Full Text]
  2. MacKie RM, et al. (2002) Incidence of and survival from malignant melanoma in Scotland: an epidemiological study. Lancet 360:587–591.[CrossRef][Web of Science][Medline]
  3. Gandini S, et al. (2005) Meta-analysis of risk factors for cutaneous melanoma: II. Sun exposure. Eur. J. Cancer 41:45–60.[CrossRef][Web of Science][Medline]
  4. Kraemer KH, et al. (1994) The role of sunlight and DNA repair in melanoma and nonmelanoma skin cancer. The xeroderma pigmentosum paradigm. Arch. Dermatol. 130:1018–1021.[Abstract/Free Full Text]
  5. Wei Q, et al. (2003) Repair of UV light-induced DNA damage and risk of cutaneous malignant melanoma. J. Natl Cancer Inst. 95:308–315.[Abstract/Free Full Text]
  6. Tomescu D, et al. (2001) Nucleotide excision repair gene XPD polymorphisms and genetic predisposition to melanoma. Carcinogenesis 22:403–408.[Abstract/Free Full Text]
  7. Baccarelli A, et al. (2004) XPD gene polymorphism and host characteristics in the association with cutaneous malignant melanoma risk. Br. J. Cancer 90:497–502.[CrossRef][Web of Science][Medline]
  8. Han J, et al. (2005) Genetic variation in XPD, sun exposure, and risk of skin cancer. Cancer Epidemiol. Biomarkers Prev. 14:1539–1544.[Abstract/Free Full Text]
  9. Millikan RC, et al. (2006) Polymorphisms in nucleotide excision repair genes and risk of multiple primary melanoma: the Genes Environment and Melanoma Study. Carcinogenesis 27:610–618.[Abstract/Free Full Text]
  10. Debniak T, et al. (2006) XPD common variants and their association with melanoma and breast cancer risk. Breast Cancer Res. Treat. 98:209–215.[CrossRef][Web of Science][Medline]
  11. Blankenburg S, et al. (2005) Assessment of 3 xeroderma pigmentosum group C gene polymorphisms and risk of cutaneous melanoma: a case-control study. Carcinogenesis 26:1085–1090.[Abstract/Free Full Text]
  12. Winsey SL, et al. (2000) A variant within the DNA repair gene XRCC3 is associated with the development of melanoma skin cancer. Cancer Res. 60:5612–5616.[Abstract/Free Full Text]
  13. Blankenburg S, et al. (2005) No association between three xeroderma pigmentosum group C and one group G gene polymorphisms and risk of cutaneous melanoma. Eur. J. Hum. Genet. 13:253–255.[CrossRef][Web of Science][Medline]
  14. Davies H, et al. (2002) Mutations of the BRAF gene in human cancer. Nature 417:949–954.[CrossRef][Medline]
  15. Nelson HH, et al. (2002) The XRCC1 Arg399Gln polymorphism, sunburn, and non-melanoma skin cancer: evidence of gene-environment interaction. Cancer Res. 62:152–155.[Abstract/Free Full Text]
  16. Han J, et al. (2004) Genetic variation in XRCC1, sun exposure, and risk of skin cancer. Br. J. Cancer 91:1604–1609.[Web of Science][Medline]
  17. Li C, et al. (2006) Genetic variants of the ADPRT, XRCC1 and APE1 genes and risk of cutaneous melanoma. Carcinogenesis 27:1894–1901.[Abstract/Free Full Text]
  18. Goode EL, et al. (2002) Polymorphisms in DNA repair genes and associations with cancer risk. Cancer Epidemiol. Biomarkers Prev. 11:1513–1530.[Abstract/Free Full Text]
  19. Yarosh DB, et al. (2005) After sun reversal of DNA damage: enhancing skin repair. Mutat. Res. 571:57–64.[Web of Science][Medline]
  20. Duan Z, et al. (2002) DNA repair gene XRCC3 241Met variant is not associated with risk of cutaneous malignant melanoma. Cancer Epidemiol. Biomarkers Prev. 11:1142–1143.[Free Full Text]
  21. Bertram CG, et al. (2004) An assessment of a variant of the DNA repair gene XRCC3 as a possible nevus or melanoma susceptibility genotype. J. Invest. Dermatol. 122:429–432.[CrossRef][Web of Science][Medline]
  22. Han J, et al. (2004) Polymorphisms in DNA double-strand break repair genes and skin cancer risk. Cancer Res. 64:3009–3013.[Abstract/Free Full Text]
  23. Hayward NK. (2003) Genetics of melanoma predisposition. Oncogene 22:3053–3062.[CrossRef][Web of Science][Medline]
  24. Siddique M, et al. (2006) Trp53-dependent DNA-repair is affected by the codon 72 polymorphism. Oncogene 25:3489–3500.[CrossRef][Web of Science][Medline]
  25. Pim D, et al. (2004) p53 polymorphic variants at codon 72 exert different effects on cell cycle progression. Int. J. Cancer 108:196–199.[CrossRef][Web of Science][Medline]
  26. Shen H, et al. (2003) p53 codon 72 Arg homozygotes are associated with an increased risk of cutaneous melanoma. J. Invest. Dermatol. 121:1510–1514.[CrossRef][Web of Science][Medline]
  27. Han J, et al. (2006) The p53 codon 72 polymorphism, sunburns, and risk of skin cancer in US caucasian women. Mol. Carcinog. 45:694–700.[CrossRef][Web of Science][Medline]
  28. Lear JT, et al. (2000) Detoxifying enzyme genotypes and susceptibility to cutaneous malignancy. Br. J. Dermatol. 142:8–15.[CrossRef][Web of Science][Medline]
  29. Kerb R, et al. (2002) Influence of GSTT1 and GSTM1 genotypes on sunburn sensitivity. Am. J. Pharmacogenomics 2:147–154.[CrossRef][Medline]
  30. Halsall JA, et al. (2004) A novel polymorphism in the 1A promoter region of the vitamin D receptor is associated with altered susceptibility and prognosis in malignant melanoma. Br. J. Cancer 91:765–770.[Web of Science][Medline]
  31. Vitart V, et al. (2005) Increased level of linkage disequilibrium in rural compared with urban communities: a factor to consider in association-study design. Am. J. Hum. Genet. 76:763–772.[CrossRef][Web of Science][Medline]
  32. Farrington SM, et al. (2005) Germline susceptibility to colorectal cancer due to base-excision repair gene defects. Am. J. Hum. Genet. 77:112–119.[CrossRef][Web of Science][Medline]
  33. Wigginton JE, et al. (2005) A note on exact tests of Hardy-Weinberg equilibrium. Am. J. Hum. Genet. 76:887–893.[CrossRef][Web of Science][Medline]
  34. Halperin E, et al. (2004) Haplotype reconstruction from genotype data using imperfect phylogeny. Bioinformatics 20:1842–1849.[Abstract/Free Full Text]
  35. Benhamou S, et al. (2005) ERCC2/XPD gene polymorphisms and lung cancer: a HuGE review. Am. J. Epidemiol. 161:1–14.[Abstract/Free Full Text]
  36. Ferry KV, et al. (2000) Increased nucleotide excision repair in cisplatin-resistant ovarian cancer cells: role of ERCC1-XPF. Biochem. Pharmacol. 60:1305–1313.[CrossRef][Web of Science][Medline]
  37. Yin J, et al. (2003) Twelve single nucleotide polymorphisms on chromosome 19q13.2-13.3: linkage disequilibria and associations with basal cell carcinoma in Danish psoriatic patients. Biochem. Genet. 41:27–37.[CrossRef][Web of Science][Medline]
  38. Thomas NE, et al. (2006) Could BRAF mutations in melanocytic lesions arise from DNA damage induced by ultraviolet radiation? J. Invest. Dermatol. 126:1693–1696.[CrossRef][Web of Science][Medline]
  39. Dolzan V, et al. (2006) Genetic susceptibility to environmental carcinogenesis in Slovenian melanoma patients. Acta Dermatovenerol. Alp. Panonica Adriat. 15:69–78.[Medline]
  40. Li C, et al. (2007) Genetic variants of the Vitamin D receptor gene alter risk of cutaneous melanoma. J. Invest. Dermatol 127:276–280.[CrossRef][Web of Science][Medline]
  41. Han J, et al. (2006) Polymorphisms in the MTHFR and VDR genes and skin cancer risk. Carcinogenesis doi: 10.1093.
  42. Kennedy C, et al. (2001) Melanocortin 1 receptor (MC1R) gene variants are associated with an increased risk for cutaneous melanoma which is largely independent of skin type and hair color. J. Invest. Dermatol. 117:294–300.[CrossRef][Web of Science][Medline]
Received November 13, 2006; revised December 21, 2006; accepted December 25, 2006.


Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?


This article has been cited by other articles:


Home page
 CarcinogenesisHome page
S. Mocellin, D. Verdi, and D. Nitti
DNA repair gene polymorphisms and risk of cutaneous melanoma: a systematic review and meta-analysis
Carcinogenesis, October 1, 2009; 30(10): 1735 - 1743.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow Supplementary Data
Right arrow All Versions of this Article:
28/5/1087    most recent
bgl257v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (1)
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by Povey, J. E.
Right arrow Articles by Melton, D. W.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Povey, J. E.
Right arrow Articles by Melton, D. W.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?