Polymorphisms of the DNA Repair Genes XRCC1, XRCC3, XPD, Interaction with Environmental Exposures, and Bladder Cancer Risk in a Case-Control Study in Northern Italy

Bladder cancer is the most important urinary tract cancer with an estimated 336,000 new cases diagnosed worldwide in 2000 (1). The burden of bladder cancer is high in Italy, with estimated age-standardized incidence rates in 2000 of 28.0 and 5.0/100,000 for men and women, respectively (1).

Tobacco smoking is the leading determinant of bladder cancer to which 66% male cases were attributable in Europe (2). Tobacco smoke contains several potent chemical carcinogens, including PAHs,⁵ aromatic amines, and N-nitroso compounds. Some occupational and industrial activities that involve exposure to aromatic amines and PAHs have also been associated with bladder cancer (3). These carcinogens may lead to direct and indirect DNA damage (4). Different biological mechanisms respond to repair DNA damage and maintain genome integrity. Variation in DNA repair capacity results in different biological response to DNA damage and thus different susceptibility to develop malignant neoplasms (5). Many DNA repair genes have been identified: they are involved in several rare recessive inherited DNA repair syndromes such as ataxia-telangiectasia, Fanconi’s anemia, Bloom’s syndrome, and xeroderma pigmentosum, which are characterized by hypersensitivity to carcinogens and high risk of cancer (6). Polymorphism in DNA repair genes that leads to amino acid substitution may influence the hosts’ capacity to repair DNA damage and thus susceptibility to cancer (7). Although genetic variants of these genes at one or more loci are likely to be associated with only moderate changes in cancer risk, they are prevalent in the population and may contribute to the overall population risk of cancer.

Various types of DNA damage are repaired through multiple repair pathways in which a number of proteins play a role. XRCC1 gene is mapped at human chromosome 19q13.2-13.3 (8) and XRCC1 protein (M_r 70,000) is an important component of the base excision repair pathway, which fixes base damage and DNA single strand breaks caused by ionizing radiation and alkylating agents. The XRCC1 protein has no known catalytic activity but serves to orchestrate base excision repair via its role as a central scaffolding protein physically associated with DNA ligase III at its COOH terminus, DNA polymerase β at its NH₂ terminus, human AP endonuclease, polynucleotide kinase, and poly(ADP-ribose) polymerase, and via its function in recognizing and binding to single strand breaks (9, 10, 11, 12). This gene was found to restore DNA repair activity in Chinese hamster ovary mutant EM9 cells (13). The XRCC1 codon 399 (G→A) transition is located at the COOH-terminal side of the poly(ADP-ribose) polymerase-interacting domain within a relatively nonconserved region between conserved residues of the BRCT domain (10) and leads to amino acid substitution of Arg to Gln (14). Although this polymorphism results in amino acid substitutions, there is no direct evidence on its functional consequences. Two other XRCC1 polymorphisms (Arg¹⁹⁴Trp and Arg²⁸⁰His) were not included in this study because they are infrequent in Caucasians (15).

XRCC3 participates in homologous recombination repair of DNA double strand breaks and cross-links. It is a member of an emerging family of Rad-51-related proteins that may take part in homologous recombination to maintain chromosome stability and to repair DNA damage (16). XRCC3 was shown to interact directly with HsRad51 and, as with Rad55 and Rad57 in yeast, may cooperate with HsRad51 during recombination repair (17). XRCC3-deficient cells were found to be unable to form Rad51 foci after radiation damage and demonstrated genetic instability and increased sensitivity to DNA-damaging agents (18). The Thr²⁴¹Met substitution in XRCC3 is due to a (C→T) transition at exon 7 and is a nonconservative change but does not reside in the ATP-binding domains, which are the only functional domains that have been identified in the protein at this time (14).

The XPD protein takes part in the nucleotide excision repair pathway, which recognizes and repairs a wide range of structurally unrelated lesions such as bulky adducts and thymidine dimers. XPD works as an ATP-dependent (5′→3′) helicase joined to the basal TFIIH complex to separate double helix (19). The XPD protein is necessary for normal transcription initiation and nucleotide excision repair (20). Mutations in the XPD gene can diminish the activity of TFIIH complexes giving rise to repair defects, transcription defects, and abnormal responses to apoptosis (21). The XPD Lys⁷⁵¹Gln substitution is attributed to a (A→C) transversion at exon 23 (14).

Polymorphisms of DNA repair genes have been suggested to be risk factors for various neoplasms such as cancer of the lung, stomach, and head and neck (15). Several studies have been conducted to assess the relationship between polymorphisms of several DNA repair genes and the risk of bladder cancers (15), but the results are fairly inconsistent, and no conclusions can be drawn at present. In this study, we hypothesized that there exists an association between bladder cancer and three genetic polymorphisms in DNA repair genes, XRCC1 (Arg³⁹⁹Gln), XRCC3 (Thr²⁴¹Met), and XPD (Lys⁷⁵¹Gln). To investigate the role of polymorphisms of these three genes and their joint effect with smoking, occupational PAHs, and aromatic amines exposure, we performed a hospital-based case-control study on men in Brescia, Italy.

This study was carried out in the Brescia province, northern Italy, a highly industrialized area with extensive metal and mechanic industries, construction, and manufacture of textiles. Although there are no obvious industrial activities traditionally associated with an increased risk of bladder cancer such as rubber and dyestuffs industry, this area has one of highest mortality rates from bladder cancer among men in Italy and in Western Europe (22).

Eligible subjects were male residents in the province of Brescia, ages 20–80 years. The case group comprised 216 male incident bladder cancer patients who were admitted to the Urology Department of two main hospitals in the province of Brescia, where almost all incident cases in the town of Brescia and the majority of those occurring in the province are admitted. All bladder cancer diagnoses were histologically confirmed. Controls were 220 men admitted to the urology departments of the same hospitals, who were diagnosed with nonneoplastic diseases, including hydronephrosis, urolithiasis, malformative urological diseases, prostatic adenomas and hypertrophia, urological traumas, orchiepididymitis, hydrocele, or unspecified urinary symptoms. Controls were frequency matched to cases by age (±5 years), period of recruitment, and hospital of admission. A written informed consent describing aims, methods, and personal responsibility for the study was obtained from each subject, except 15 cases and 6 controls, who refused either the interview or the blood test. The final study population consisted of 201 cases and 214 controls. The study period was from July 1997 through December 2000. Women were excluded from this study because of their low incidence of this disease and low prevalence of exposure to tobacco smoking and occupational bladder cancer carcinogens. All study subjects were Caucasians of Italian nationality.

Cases and controls were interviewed face-to-face during their hospitalization by three interviewers, who were experienced in occupational medicine and aware of the case and control’s status. A semistructured questionnaire was developed, which included sections dedicated to sociodemographic status, clinical and histological data, occupational history, dietary habits, tobacco smoking, beverages and alcohol consumption, frequency of diuresis, total fluid consumption, environmental PAH exposure, and leisure-time activities entailing chemical exposures. Lifetime occupational history was collected for each job that lasted for at least 1 year. Information included job title, plant activity, type of production, exposure to chemicals, with detailed description of workplaces and job tasks, as well as the use of personal protective devices in the workplace. Job titles and plant activities were coded according to the “International Standard Classification of Occupation” and “International Standard Classification of All Industrial Activities.” For each specific job title, exposure to PAHs and aromatic amines was assessed by an expert in occupational medicine and industrial hygiene who was blind to case/control status. Exposure was classified as absent, possible, probable or definite. In the case of exposure, level and frequency of exposure were classified on a three-category scale, and mode of exposure was also classified as respiratory or dermal according to the methodology described in previous studies (23, 24). The average duration of an interview was ∼1.5 h.

During hospitalization, an aliquot (20–25 ml) of venous blood was drawn from each subject and sent to a local laboratory for centrifugation and cell extraction. WBC samples were shipped to IARC for genotype identification.

PCR followed by enzymatic digestion analysis was used for genotyping of XRCC1 Arg³⁹⁹Gln, XRCC3 Thr²⁴¹Met, and XPD Lys⁷⁵¹Gln polymorphisms. The methods used were developed and validated by Matullo et al. (25). Reproducible RFLP patterns were obtained from the literature and tested on a series of subjects with known genotype before their routine application to the study population. All uncertain results were reanalyzed with the same technique, and usually one more assay was sufficient to clarify any doubts. All analyses were conducted in one laboratory by technicians who were blind to case/control status.

Tests for Hardy-Weinberg equilibrium among cases and controls were conducted on observed and expected genotype frequencies using Pearson’s χ² test with one degree of freedom. Multivariate ORs and 95% CIs of suspected environmental risk factors were calculated using unconditional logistic regression with SAS (version 8, SAS Institute) and adjusted for age, education, residence, and accumulative tobacco consumption (pack-year). Effect of genotypes were estimated by unconditional logistic regression and adjusted for age as a continuous variable.

Gene-environment interactions were estimated and tested on a multiplicative scale by combining genotypes and environmental exposure information and adding a multiplicative term in the logistic model. Joint effects were encoded using as reference individuals unexposed to the environmental factor with the common genotype. Cumulative lifetime smoking was classified into two subgroups, light and heavy, according to the value of pack-years, with a cutoff point of 26, the median pack-year level among controls. Occupational exposure to PAHs and aromatic amines was classified as never or ever exposed.

The distribution of age, education, and residence was similar between case and control groups (Table 1). The mean of age was 63 years, and ∼85% subjects were −50 years. Approximately 55% of subjects had low educational level and ∼20% of them received higher education. Less than 40% subjects lived in the town of Brescia and the remaining subjects lived in other districts of Brescia province. Cases were more likely to be current or ex-smokers than controls. The adjusted OR for current smoking versus never smoking was 5.22 (95% CI, 2.74–9.93), and OR for ex-smoking versus never smoking was 2.38 (95% CI, 1.27–4.48). The adjusted OR of ever exposure to PAHs and aromatic amines compared with never exposed was 1.22 (95% CI, 0.79–1.90) and 1.57 (95% CI, 0.70–3.52), respectively.

The XRCC1 codon 399-A allele frequency among cases and controls was 0.32 and 0.34, respectively, that of XRCC3 codon 241-T allele was 0.34 among cases and 0.40 among controls, and that of XPD 751-C allele was 0.39 among cases and 0.40 among controls. The distribution of common and variant genotypes of XRCC1 codon 399 and XPD codon 751 among cases and controls was consistent with Hardy-Weinberg equilibrium, but the frequency of XRCC3 241-T allele among controls was not in Hardy-Weinberg equilibrium (P = 0.05).

The genotype frequencies of the XRCC1 399 and XPD 751 polymorphisms were rather similar between cases and controls and showed no significant association with bladder cancer. In the case of the XRCC3 codon 241 polymorphism, genotype frequencies of both heterozygote (Thr/Met) and homozygote (Met/Met) were different between cases and controls, and variant carriers (241-T) exhibited a significantly reduced risk of bladder cancer (OR, 0.63; 95% CI, 0.42–0.93; Table 2).

In the analysis by smoking subgroups, the XRCC1 ³⁹⁹Gln variant was not a risk factor among never and light smokers but showed a nonsignificant protective effect among heavy smokers, especially for the homozygous variant (OR, 0.38; 95% CI, 0.14–1.02). XRCC3 ²⁴¹Met variant was associated with a significant reduction of risk of bladder cancer only among heavy smokers (OR, 0.49; 95% CI, 0.28–0.88). The XPD ⁷⁵¹Gln polymorphism was not a risk factor among any smoking subgroups (Table 3). In the joint effect analysis, there was little evidence of a multiplicative interaction between these polymorphisms and tobacco smoking (Table 4).

There was no suggestion of a modification of the effect of PAHs exposure according to any of the three polymorphisms (Table 5). A nonsignificant increased risk of bladder cancer associated with occupational aromatic amines exposure was observed among common XRCC1 genotype carriers (OR, 2.74; 95% CI, 0.70–10.70) and common XRCC3 (OR, 2.92; 95% CI, 0.77–11.07) genotype carriers but not among mutant carriers. The tests for interaction were not statistically significant (Table 5, P of test for interaction 0.39 for XRCC1, 0.28 for XRCC3, 0.89 for XPD).

The repair of carcinogen-induced DNA damage is an important aspect in carcinogenesis. It is under genetic control, and there is increasing evidence that genetic variation leads to different DNA repair capacities in the human population (5). Rare mutations of DNA repair genes result in lethal genetic diseases and common polymorphisms may act as genetic susceptibility factors. Therefore, integrating information on common allelic variants of these genes may be informative in clarifying the causes and mechanisms of cancers and determining groups of exposed individuals at higher risk. We investigated the role of the polymorphisms of three DNA repair genes as bladder cancer risk factors in a case-control study conducted in northern Italy. Allele frequencies in our study were close to those of Caucasians observed in other studies (15). Our results indicated that the polymorphism at XRCC1 codon 399 might decrease the risk of developing bladder cancer, in particular among heavy smokers. The XRCC3 Thr²⁴¹Met polymorphism had a protective effect against bladder cancer risk, largely confined to heavy smokers. The impact of aromatic amines on bladder cancer might be restricted to common genotype carriers of XRCC1 and XRCC3 in the local population.

A possible relationship between XRCC1 Arg ³⁹⁹Gln polymorphism and cancer has been investigated in >30 studies, although results remain inconsistent. A few experimental studies observed that ³⁹⁹Gln allele carriers had significantly increased DNA adducts level, sister chromosome exchange frequency, or other indicators of damaged DNA repair capacity (26, 27). However, these results were not reproduced, and contradictory results were obtained in an additional study (28). Similarly, most epidemiological case-control studies did not find elevated risk of tobacco-related cancers, including lung cancer and head and neck cancer, associated with ³⁹⁹Gln variant (29, 30). Similar to our study, a case-control study suggested that the homozygous variant provides moderate protection against bladder cancer in a recessive genetic model (OR, 0.7; 95% CI, 0.4–1.3) in American whites with limited evidence of interaction with smoking (31). In another case-control study conducted in Italy, an apparent protective impact was restricted to ever-smokers (crude OR, 0.48; 95% CI, 0.22–1.02; Ref. 25).

We found supportive evidence for the hypothesis that XRCC1 codon 399 polymorphism affected bladder cancer among heavy smokers but obtained little evidence of its interaction with smoking, PAHs, or aromatic amines. The XRCC1 Arg³⁹⁹Gln polymorphism might have subtle and multiple influences on cancer formation, which depend on other genetic factors or environmental exposures. Carcinogenesis is a complex process, and there exist interrelated pathways of biological response to DNA damage in which multiple genetic mechanisms are involved (32). The (Arg→Gln) substitution might give protection against bladder cancer to heavily exposed subgroups. The biological mechanism of this dose-dependent gene exposure relationship needs additional clarification.

Our study is the first to suggest a protective role of XRCC3 codon 241 polymorphism against bladder cancer risk, which was stronger among heavy smokers. Matullo et al. observed higher DNA adducts level among Italian subjects who were Met/Met homozygote carriers (33). This difference was present among never- and ex-smokers, although not current smokers, and was explained by up-regulation of DNA repair enzymes due to heavy exposure to carcinogens, reversing the difference among different genotypes. In a separate case-control study, they found that a significantly increased risk of bladder cancer was associated with the variant genotype (OR, 2.77; 95% CI, 1.55–4.93), and the impact was restricted to nonsmokers and ex-smokers (25). In their study, the distribution of genotypes among controls departed significantly from Hardy-Weinberg equilibrium (P = 0.006), and the observed number of putative high-risk homozygote carriers in control group was larger than predicted by Hardy-Weinberg equilibrium, which may increase the chance of a false-positive association (34). A case-control study from the United States provided little evidence of a positive association between ²⁴¹Met variant carriers and bladder cancer (OR, 1.3; 95% CI, 0.9–1.9), although ²⁴¹Met carriers appeared to have higher risk than did common genotype carriers from heavy smoking (OR, 8.3 versus 3.9; Ref. 35).

XRCC3 is one of five identified paralogs of the strand-exchange protein RAD51 in human beings and functions through complex interactions with other relevant proteins to repair double strand breaks and maintain genome integrity in multiple phases of homologous recombination (36). XRCC3-deficient hamster cells showed high frequency of multiple centrosomes and abnormal spindle formation (18). Polymorphisms of this gene may result in reduced DNA repair capacity, but direct functional research evidence is absent, and epidemiological research results are inconclusive at present. A recent experimental study did not show a functional difference between the 241 variant and the common genotype of XRCC3 in homologous recombination repair of DNA in XRCC3-deficient irs1SF cells and in sensitivity to the interstrand cross-linking agent, mitomycin C (37). A protective effect of XRCC3 ²⁴¹Met allele on human cancer is plausible, although little is known about the biochemical properties and biological functions of XRCC3 protein and the functional changes associated with this polymorphism. It may also be in linkage disequilibrium with another gene responsible for the observed protective association. In addition, the frequency of the XRCC3 genotype distribution in our control group departed from Hardy-Weinberg equilibrium and thus we cannot rule out the possibility that such an association occurred as a result of bias. However, when we repeated the analysis on the basis of expected numbers under Hardy-Weinberg equilibrium of controls (38), the apparent protective effect persisted, although the OR was no longer significant. Corrected ORs of heterozygotes, homozygous variant, and variant carriers compared with common homozygote carriers were 0.74 (95% CI, 0.51–1.07), 0.66 (95% CI, 0.30–1.41), and 0.72 (95% CI, 0.50–1.04), respectively.

The (A→C) variation of the XPD 751 gene leads to a change of configuration of the coded protein and may alter the XPD protein’s interaction with helicase activator p44 protein inside the TFIIH complex (21). Similar to the studies regarding the XRCC1 and XRCC3 polymorphisms, the role of the XPD polymorphism in human cancer is unclear. In a study conducted in northern Italy, significantly elevated DNA adducts among nonsmokers were associated with XPD codon 751 variants (33). However, another study observed reduced DNA repair proficiency in Caucasian women associated with the common genotype of XPD (OR, 7.2; 95% CI, 1.01–87.7; Ref. 39). Several case-control studies reported that XPD 751 polymorphism carriers were at significantly or borderline increased risk of smoking-related cancers (40, 41, 42). Hou et al. (40) found in a Swedish population that the XPD 751 variant allele was associated with an increased risk of lung cancer among nonsmokers, particular younger than 70 years (OR, 3.2; 95% CI, 1.3–8.0), and had no effect among ever-smokers. In a study among non-Hispanic whites in the United States, XPD 751 homozygous variants had a borderline increased risk of squamous cell carcinoma of the head and neck (OR in recessive genetic model, 1.51; 95% CI, 0.94–2.43), and this risk was higher among current alcohol drinkers and current smokers (41). In a Chinese population, this polymorphism (Gln) increased the risk of squamous cell carcinoma of the lung (OR, 1.52; 95% CI, 0.94–2.46; Ref. 42). On the contrary, another study in a Chinese population reported a large increase in risk of lung cancer associated with the common genotype (OR of Lys/Lys + Lys/Gln versus Gln/Gln, 3.19; 95% CI, 1.01–10.07; Ref. 29). There are also several nonsignificant results reported. Interestingly, Matullo et al. (25) reported a nonsignificant decreased risk of bladder cancer among nonsmokers (OR, 0.36; 95% CI, 0.09–1.55) and ex-smokers (OR, 0.60; 95% CI, 0.14–2.68) and an increased risk among current smokers (OR, 2.53; 95% CI, 0.92–6.96). In contrast, Stern et al. (43) observed little overall association between the XPD codon 751 variant and bladder cancer risk but a negative statistical interaction with smoking (P = 0.03). In our study, neither the overall analysis nor the subgroup analysis suggested a significant association between XPD ⁷⁵¹Gln allele and bladder cancer.

Our study had several strengths: our cases represent an incident series from a high-risk area; information of history of occupational exposure to chemicals was obtained using a detailed questionnaire coded with assessment by occupational health and industrial hygiene experts; and the study had power of 94% to detect a factor with 2-fold increased risk and 40% frequency in the population. There were also several limitations in our study, which may bias the results and restrict us in making conclusive statements. Our control group came from hospital urology patients, which would not be an unbiased sample of local residents. Selection bias might originate from an association between the studied polymorphisms and diseases of controls. However, the variability of diseases among controls and the lack of evidence of an association with DNA repair (6) minimize the possibility of bias. Similarly, different referral patterns of patients with bladder cancer versus other urological conditions may lead to selection bias, although it is unclear how this might lead to genetic differences between cases and controls.

In conclusion, among the three genetic polymorphisms included in our study, XRCC3 codon 241 polymorphism had a favorable impact against bladder cancer, the effect and magnitude of which depend on environmental chemical exposures. The XRCC1 codon 399 polymorphism also had a protective influence on bladder cancer among heavy smokers. The XPD codon 751 polymorphism was not, however, associated with bladder cancer risk.

We thank Dr. Mia Hashibe for comments of this manuscript.