Polymorphisms in DNA repair genes and epithelial ovarian cancer risk
Article first published online: 27 MAY 2005
Copyright © 2005 Wiley-Liss, Inc.
International Journal of Cancer
Volume 117, Issue 4, pages 611–618, 20 November 2005
How to Cite
Auranen, A., Song, H., Waterfall, C., DiCioccio, R. A., Kuschel, B., Kjaer, S. K., Hogdall, E., Hogdall, C., Stratton, J., Whittemore, A. S., Easton, D. F., Ponder, B. A.J., Novik, K. L., Dunning, A. M., Gayther, S. and Pharoah, P. D.P. (2005), Polymorphisms in DNA repair genes and epithelial ovarian cancer risk. Int. J. Cancer, 117: 611–618. doi: 10.1002/ijc.21047
- Issue published online: 14 SEP 2005
- Article first published online: 27 MAY 2005
- Manuscript Accepted: 20 JAN 2005
- Manuscript Received: 25 NOV 2004
- Cancer Research UK
- Roswell Park Alliance
- National Cancer Institute. Grant Numbers: CA71766, CA16056
- Academy of Finland
- “Deutsche Krebshilfe”
- ovarian cancer risk;
- DNA repair
DNA repair gene polymorphisms and mutations are known to influence cancer risk. We studied whether polymorphisms in DNA double strand break (DSB) repair genes are associated with epithelial ovarian cancer (EOC) risk. Up to 1,600 cases and 4,241 controls from 4 separate genetic association studies from 3 countries were genotyped for 13 single nucleotide polymorphisms (SNP) in 6 genes (BRCA1, NBS1, RAD51, RAD52, XRCC2 and XRCC3) involved in homologous recombination of DNA double strand breaks. Genotype specific risks were estimated as odds ratios (OR) by unconditional logistic regression. No association was detected between EOC risk and BRCA1 Q356R, BRCA1 P871L, RAD51 g135c, RAD51 g172t, RAD52 c2259t, NBS1 L34L, NBS1 E185Q, NBS1 A399A, NBS1 P672P, XRCC2 g4324c, XRCC2 c41657t and XRCC3 T241M. The XRCC2 R188H polymorphism was associated with a modest reduction in EOC risk: OR for heterozygotes was 0.8 (95% confidence interval [CI] = 0.7–1.0) and for rare homozygotes 0.3 (0.1–0.9). The XRCC3 a4541g polymorphism, situated in the 5′UTR, and the intronic XRCC3 a17893g polymorphism were not associated with EOC risk in general, but when the serous EOC subset only was analysed, the OR for heterozygotes for a4541g was 1.0 (0.9–1.2) and for the rare homozygotes 0.5 (0.3–0.9). For the XRCC3 a17893g polymorphism, the OR for the heterozygotes and the rare homozygotes were 0.8 (0.7–0.9) and 0.9 (0.7–1.2), respectively. In our study, some polymorphisms in XRCC2 and XRCC3 genes were associated with EOC risk. Further research on the role of these genes on epithelial ovarian cancer is warranted. © 2005 Wiley-Liss, Inc.
Without properly functioning DNA repair mechanisms, cells may accumulate various types of DNA damage. DNA double-strand breaks (DSB) are a common form of DNA damage induced during normal DNA replication and by environmental agents such as ionising radiation and genotoxic chemicals.1 These lesions are particularly harmful, because if left unrepaired, they can cause chromosomal loss, translocations and deletions that may subsequently lead to the activation of proto-oncogenes, loss of function of tumour suppressor genes or global genomic instability.2
The repair of DSB in human cells is controlled by 2 different pathways: homologous recombination (HR) and non-homologous end-joining (NHEJ). In HR the broken strand is repaired using the homologous chromosome or sister chromatid as a template, whereas in NHEJ the broken strands are crudely joined together at a site of microhomology.3, 4 Homologous recombination is a high fidelity process, whereas NHEJ frequently results in small deletions at the site of fusion and is error-prone. Genes participating in HR include RAD51, RAD52, BRCA1, BRCA2, XRCC2 and XRCC3. The protein products of MRE11, RAD50 and NBS1 form a multi-unit complex that binds to DNA DSB, signalling and recruiting components from the HR and NHEJ pathways.5
Mutations in BRCA1 and BRCA2 cause breast, ovarian and other cancers,6, 7 and an association has been suggested between a common polymorphism in BRCA2 and breast and ovarian cancer risk.8, 9 Some recent studies suggest that polymorphisms in XRCC2 and XRCC3 influence susceptibility to breast cancer, skin cancer or acute myeloid leukemia.10, 11, 12 Profound genetic instability has been noted in XRCC2-knockout mice and in XRCC3-mutant hamster cell lines,13, 14 which gives further evidence for the important role of these DNA repair genes in protecting cells from harmful mutations.
Epithelial ovarian cancer (EOC) accounts for 5% of all cancers among women.15 The etiology of EOC is not fully understood, but both epidemiological and biological observations suggest that ovulation may play a role in ovarian cancer development. During a normal monthly ovulation, the surface epithelium of the ovary is degraded enzymatically to release the ovum and the resulting wound is subsequently repaired. This natural process is speculated to carry the potential for malignant transformation; repetitive ovulation increases the frequency of epithelial cell division, which subsequently increases DNA replication. Replication can induce DNA damage and the resulting lesions must be recognized and repaired to prevent the accumulation of potentially oncogenic mutations.16, 17 In support of the repetitive ovulation hypothesis, nulliparity has been found to be a risk factor for ovarian cancer and lack of ovulation, resulting from oral contraceptive pill use and pregnancy, is associated with a decreased risk of ovarian cancer.18, 19, 20, 21
The known highly penetrant, rare cancer predisposition alleles, such as germline mutations in the BRCA1 or BRCA2 tumor suppressor genes, are estimated to account for <10% of all ovarian cancer cases and <30% of the excess familial risk of ovarian cancer.22 It is likely that the unexplained heritable component of ovarian cancer susceptibility is due to multiple weakly penetrant alleles. Mutations in genes functioning in the HR pathway of DNA DSB repair cause inherited susceptibility to EOC and polymorphisms in these genes are suspected to influence the risk of breast and other cancers. The purpose of our study was to test the hypothesis that polymorphisms in these genes are associated with risk of invasive EOC. We aimed to identify common coding variants that may have functional effects as well as non-coding variants that may be in linkage disequilibrium with variants in unidentified regulatory regions.
Material and methods
Selection of candidate genes and SNP
DSB repair genes considered for our study were: NBS1, RAD50, RAD51, RAD52, XRCC2, XRCC3 and BRCA1. Candidate SNP were initially identified through public SNP databases and literature searches. The URLs of these databases are listed in the section below. SNP were accepted for our study if their population frequency was >0.05. When suitable SNP could not be identified by electronic or literature searches, we carried out a variation search using either single-stranded conformation polymorphism (SSCP) or denaturing high pressure liquid chromatography (DHPLC) as described previously.10 Possible SNP were verified by DNA sequencing using an ABI Prism 377 sequencer.10 The association of a SNP in BRCA2 with ovarian cancer incidence has been described elsewhere.9 The SNP chosen for our study are presented in Table I.
|Gene||DbSNP reference1||SNP||Position||Amino acid change|
|BRCA1||Rs1799950||A/G at nt 1186||Exon 11||Gln356Arg|
|BRCA1||Rs799917||C/T at nt 2731||Exon 11||Pro871 Leu|
|NBS1||Rs1063045||G/A at nt 6582||Exon 2||Leu34Leu|
|NBS1||Rs1805794||C/G at nt 11122||Exon 5||Glu185GIn|
|NBS1||Rs709816||T/C at nt 33890||Exon 10||Asp399Asp|
|NBS1||Rs1061302||A/G at nt 43179||Exon 13||Pro672Pro|
|RAD51||Rs1801320||G/C at nt 135||5'UTR||No change|
|RAD51||Rs1801321||G/T at nt 172||5'UTR||No change|
|RAD52||Rs11226||C/T at nt 2259||3'UTR||No change|
|XRCC2||Rs3218536||G/A at nt 31479||Exon4||Arg188His|
|XRCC3||Rs1799794||A/G at nt 4541||5'UTR||No change|
|XRCC3||Rs1799796||A/G at nt 17893||Intron 5||No change|
|XRCC3||Rs861539||C/T at nt 18067||Exon 7||Thr241 Met|
United Kingdom SEARCH study
The SEARCH ovarian cancer study is an ongoing, population-based ovarian cancer case control study covering the regions served by the East Anglia and West Midlands cancer registries in the United Kingdom. Eligible women are those diagnosed since 1991 with invasive epithelial ovarian cancer under the age of 70 years. Participants are asked to provide written consent, to complete an epidemiological questionnaire and to provide a 20-ml whole blood sample. Currently, of 1,785 eligible patients, 1,019 women have agreed to take part. The first 864 cases (with 864 controls) were available for this analysis. Female controls have been selected randomly from the EPIC-Norfolk component of the European Prospective Investigation of Cancer (EPIC), a prospective study of diet and cancer being carried out in the same population from which the cases have been drawn. The EPIC Norfolk cohort comprises 25,000 individuals resident in Norfolk (East Anglia), aged 45–74 years at first interview in 1993. Blood for DNA extraction was collected during the second health check in 1998–2000. The ethnic background of cases and controls is similar, with over 98% being Caucasian Europeans. DNA was extracted from blood samples by Whatman International Ltd (Ely, UK) using a chloroform/phenol method. The study was approved by the Anglia and Oxford Multi-centre Research Ethics Committee.
Danish MALOVA study
The MALOVA study is a population-based, Danish case-control study of ovarian cancer. Eligible cases were women aged 35–79 years, who were diagnosed with an ovarian tumour from December 1994 to May 1999. The study included 18 different hospitals from the municipalities of Copenhagen and Frederiksberg as well as the counties of Copenhagen, Frederiksborg, Roskilde, Vestsjßlland, Storstrøm, Funen, Southern Jutland and Northern Jutland. In total 698 invasive epithelial ovarian cancers, 219 ovarian borderline tumors and 450 benign ovarian tumors were enrolled. Controls were drawn from the general female population within the study area (aged 35–79 years) selected at random using the computerized Central Population Register. After providing written consent, cases and controls had a personal interview and gave blood samples (pre-operatively for cases). Samples from 438 cases and 1,112 controls were available for our study. DNA was extracted from blood using the QIAamp DNA Blood Mini Kit (Qiagen, Valencia, CA).
United States FROC study
The United States subjects were ascertained from the Family Registry for Ovarian Cancer (FROC) of 6 counties in Northern California. Included were ovarian cancer cases (n = 327) from Caucasians (20–64 years old) diagnosed between March 1, 1997 and July 31, 2001 and age/ethnic matched self-reported healthy female controls without cancer (n = 427), obtained from the same 6 counties by random-digit dialing. Buccal rinses were obtained from 83 cancer cases and 55 controls of FROC and bloods from all other subjects. DNA was purified from peripheral blood leucocytes using the Puregene Kit (Gentra Systems, Minneapolis, MN). DNA was isolated from exfoliated cells in buccal mouthwash rinses as described previously.23 The study protocol was approved by the Institutional Review Boards at all participating study sites.
United Kingdom Royal Marsden Hospital and young ovarian cancer study (UK RMH/YOV)
The fourth case control study comprised cases drawn from 2 different sources: 244 were women with invasive EOC who were <70 years of age at the time of diagnosis and were seen at the Royal Marsden Hospital, London between July 1993 and September 199524 and 118 were women who had taken part in a United Kingdom national study of early onset ovarian cancer.25 Women diagnosed in the United Kingdom between 1984–93 with EOC diagnosed under 30 years of age who were still alive in 1998 were eligible for our study. Control samples (n up to 1,878) for these cases were a set of EPIC samples (separate from the samples selected as controls for SEARCH) that had been previously genotyped for a breast cancer case-control association study10 (see above for description of EPIC). Genomic DNA was extracted by standard methods using an Extrragen automated extractor. Both these studies were approved by the relevant local research ethics committees.
Genotyping was carried out centrally in Strangeways Research Laboratory using the 5′-nuclease assay (TaqMan, Applied Biosystems, Warrington, UK) and the ABI PRISM 7700/7900 sequence detection system (Applied Biosystems). Primers and probes were designed with the Primer Express Oligo Design Software v1.0 (Applied Biosystems). The primer and probe sequences and amplification conditions are listed in Table II.
|SNP||Forward primer (5′-3′)||Reverse primer (5′-3′)||Vic-probe1||Fam-probe1|
For the United Kingdom RMH/YOV study, 15-ul assays (25-ul for BRCA1 P871L) were carried out on 20 ng of genomic DNA according to the manufacturer's instructions. Amplification was carried out using MJ Tetrad thermal cyclers (GRI), and the plates were read “post-PCR” on an ABI PRISM 7700 Sequence detector using the Allelic Discrimination sequence detection software (Applied Biosystems). We included 8 non-template controls (H2O) and 8 positive controls (allele-specific) in each 96-well plate. Positive controls were artificial DNA templates made by annealing together long oligonucleotides that span each SNP.
For the other 3 studies, each assay was carried out using 10 ng DNA in a 5 μl reaction with primers at 900 nM and probes at 200 nM concentrations. All reactions were carried out using 384-well arrays with 12 duplicate samples in each plate for quality control. There were no discordant genotypes in duplicates. Genotypes were called using the Allelic Discrimination sequence detection software (Applied Biosystems,). DNA samples that did not give a clear genotype result at the first attempt were not repeated because this is a high-throughput process. There are variations in the number of samples successfully genotyped for each polymorphism. We were unable to obtain satisfactory genotype calling for the BRCA1 Q356R assay in the Danish samples.
We present age-unadjusted odds ratios (OR) as there was no association of genotype with age in controls, and so age is not a confounding variable. Odds ratios were not adjusted for other known risk factors for ovarian cancer such as reproductive factors because it is unlikely that any such factors will act as true confounders.
Genotype frequencies in the controls were tested for deviation from Hardy-Weinberg Equilibrium using a standard χ2 test. (1 df). The primary test of association used was a comparison of genotype frequencies in cases and controls. This was done for each study separately using χ2 tests (2 df). The data were then pooled and genotype frequencies were compared in cases and controls using unconditional logistic regression models with terms for genotype and study and an appropriate likelihood ratio test. Genotype specific risks, with the common homozygote as the baseline comparator, were estimated as OR by unconditional logistic regression.
For genes with more than one polymorphism we investigated possible haplotype effects, using the logistic regression procedure suggested by Cordell and Clayton.26 This method includes a main effect term for each polymorphism in the logistic regression model, rather than modelling the full haplotype effect. A likelihood ratio test was used to test this model for significance. All statistical tests and p-values are 2-tailed.
The observed genotype and allele frequencies for each of the 13 polymorphisms are presented in Table III. The genotype frequencies in the controls did not differ significantly from those expected under Hardy-Weinberg equilibrium (HWE) apart from RAD52 c2259t that was marginally significant in the United Kingdom SEARCH controls only (p = 0.02). This is likely to be a chance finding, as the discrimination of genotypes for this assay was good. There were no significant differences in genotype frequencies in cases and control for any of the polymorphisms in BRCA1, NBS1, RAD51 and RAD52 when each study was analysed separately or when the data were combined (Table III). Genotype specific risks for the SNP in these genes were not significantly different from unity (Table IV).
|Study||Total||Rare allele frequency||HWE p-value||Case-control comparison of genotype frequencies|
|χ2 (2 df)||p-value|
|Study||Heterozygote risk||Rare homozygote risk|
|OR||95% CI||OR||95% CI|
There was a moderately strong association of the XRCC2 R188H polymorphism with EOC (combined data genotype frequency heterogeneity test p = 0.003). The rare allele was associated with a reduced risk of disease in a dose-dependent manner: women having one rare (histidine) allele had a 20% reduction in risk (combined data OR = 0.8 [0.66–0.96]), whereas women who were homozygote for the histidine allele had their EOC risk reduced to less than half of the common homozygote risk (combined data OR = 0.31 [0.11–0.88]). There was no evidence for heterogeneity of risk between the studies (p = 0.62). The results were similar when the analysis was restricted to the 737 cases with serous disease (data not shown).
There was some evidence for an association of the XRCC3 SNP a4541g and a17893g with invasive EOC (combined data p = 0.087 and 0.049 respectively). The genotype specific risks are shown in Table IV. These suggest that the rare allele is associated with a weak protective effect for each SNP. These associations were somewhat stronger when analysis was restricted to the 737 cases with serous type histopathology (p = 0.024 and 0.027). In the serous cases, the g-allele of a4541g in the 5′UTR was found to be associated with a recessive protective effect (OR ag vs. aa = 1.0 [0.87–1.2]; OR gg vs. aa = 0.50 [0.28–0.89]). The g-allele of a17893g in the 5′UTR was also found to be associated with a protective effect (OR ag vs. aa = 0.79 [0.0.66–0.94]; OR gg vs. aa = 0.95 [0.72–1.2]). There was no association with the missense T241M variant in XRCC3. Multiple SNP logistic regression models provided no evidence for haplotype specific effects for BRCA1, NBS1, RAD51, XRCC2 and XRCC3 (p = 0.86, 0.84, 0.84, 0.23 and 0.80 respectively).
Common variants in genes involved in double strand break DNA repair are good candidates for low penetrance susceptibility to epithelial ovarian cancer. We have found some evidence that the missense variant R188H in XRCC2 is associated with risk of invasive epithelial ovarian cancer, the rare allele having a protective effect under a co-dominant genetic model. This result needs to be interpreted with some caution. Even though the null hypothesis would be rejected at the 0.005 level for the primary test of association, the possibility of a Type I statistical error (false positive) must be considered. It has been suggested that stringent criteria should be applied to statistical tests for genetic association, e.g., p < 0.0001, because of the large number of candidate polymorphisms across the human genome. A total of 7,600 cases with controls would be needed to detect a co-dominant allele with risk of 0.8 with 80% power at this level of significance.
Hidden population stratification is an alternative explanation for a spurious association. This occurs when allele frequencies differ between population sub-groups and cases and controls are drawn differentially from those sub-groups. It seems unlikely that population stratification is relevant in this investigation because the cases and controls in the four studies reported here were drawn from the same ethnic groups and minor allele frequencies were similar for all SNP in the four studies. Furthermore, if stratification were present, it is unlikely that the same degree of stratification would be seen in all four studies. It is also worth noting that the existence of significant population stratification that has resulted in a false genetic association has never been demonstrated empirically.27
Other direct evidence to support or refute our findings is lacking as there have been no studies published previously of XRCC2 R188H in ovarian cancer. Two studies, however, have found marginal evidence for an increased risk of breast cancer associated with the rare allele, either by itself28 or only when modulated by plasma folate levels.29 Another study of this polymorphism in breast cancer found no association.10 Furthermore, there is evidence for a functional effect of R188H that might predict the rare allele to be associated with a decreased risk of cancer.28 Cell lines, in which the positively charged arginine at position 188 has been changed to the neutral alanine or deleted altogether, show significantly decreased survival compared to normal cell lines when treated with mitomycin C. Cell lines with the polymorphic histidine also showed decreased survival, but this effect was subtle and not significant statistically.
We also found weaker evidence that 2 SNP in XRCC3 are associated with EOC. The association for a4541g was not significant at the 5% level (p = 0.087) and that for a17893g was only marginally significant (p = 0.049). As with R188H, these observations may have occurred by chance. Some support for the findings comes from a report that both these SNP were associated with breast cancer risk.10
We found no evidence for an association of 9 polymorphisms in BRCA1, NBS1, RAD51 and RAD52 with EOC. Our study included a minimum of 1,561 cases and 2,602 controls providing us with at least 93% power at the 5% level of significance to detect a co-dominant allele with frequency 0.3 that confers a relative risk of 1.2 or 86% power to detect a dominant allele with frequency 0.1 that confers a relative risk of 1.3. We cannot, however, exclude the possibility that the alleles investigated are associated with smaller risks, or that there are other susceptibility variants in these genes that are not correlated strongly with the polymorphisms examined.
Several other studies investigating putative associations between ovarian cancer and various SNP have been published,30, 31, 32, 33, 34, 35, 36, 37, 38, 39 but the number of patients in the studies has rarely exceeded 500. Apart from the BRCA1 polymorphisms, none of the polymorphisms in our study have been examined previously. The small but significant differences in risk associated with XRCC2 R188H variant and the 2 non-coding XRCC3 variants, with what is known of the importance of these genes in homologous recombination pathway of DNA DSB, might indicate a true association, but confirmation of our results in other data sets is needed before drawing definitive conclusions.
The authors thank J. MacIntosh, H. Munday, B. Perkins, C. Jordan, K. Driver and the East Anglian Cancer Registry for recruitment of the UK cases, the EPIC-Norfolk investigators for recruitment of the UK controls, C. Pye for managing the RMH ovarian cancer patient data, K. Head, A. Nachaev and J. Mikula for expert technical assistance and all the study participants who contributed to our research. A.A. was partly funded by the Academy of Finland, B.K. was funded by the “Deutsche Krebshilfe”, H.S. was funded by a grant from WellBeing, B.A.J.P. is a Gibb Fellow, P.D.P.P. is a Senior Clinical Research Fellow and D.F.E. is a Principal Research Fellow of Cancer Research UK.
- 2DNA double-strand breaks, chromosomal rearrangements, and genomic instability. Mutat Res 1998; 40: 125–8., , , , , .
- 15Clinical gynecologic oncology. 6th ed. St. Louis : Mosby Inc., 2002. p 289., .
- 23A simple mouthwash method for obtaining genomic DNA in molecular epidemiological studies. Cancer Epidemiol Biomark Prev 1998; 8: 719–24., .
- 38Polymorphisms in BRCA1 and BRCA2 and risk of epithelial ovarian cancer. Clin Cancer Res 2003; 12: 4396–403., , , , , , .