DNA repair gene polymorphisms, bulky DNA adducts in white blood cells and bladder cancer in a case-control study
Article first published online: 13 MAR 2001
Copyright © 2001 Wiley-Liss, Inc.
International Journal of Cancer
Volume 92, Issue 4, pages 562–567, 15 May 2001
How to Cite
Matullo, G., Guarrera, S., Carturan, S., Peluso, M., Malaveille, C., Davico, L., Piazza, A. and Vineis, P. (2001), DNA repair gene polymorphisms, bulky DNA adducts in white blood cells and bladder cancer in a case-control study. Int. J. Cancer, 92: 562–567. doi: 10.1002/ijc.1228
- Issue published online: 9 APR 2001
- Article first published online: 13 MAR 2001
- Manuscript Accepted: 5 JAN 2001
- Manuscript Revised: 28 DEC 2000
- Manuscript Received: 1 AUG 2000
- Associazione Italiana per le Ricerche sul Cancro
- University of Turin
- bladder cancer;
- DNA adducts
Individuals differ widely in their ability to repair DNA damage, and DNA-repair deficiency may be involved in modulating cancer risk. In a case-control study of 124 bladder-cancer patients and 85 hospital controls (urological and non-urological), 3 DNA polymorphisms localized in 3 genes of different repair pathways (XRCC1-Arg399Gln, exon 10; XRCC3-Thr241Met, exon 7; XPD-Lys751Gln, exon 23) have been analyzed. Results were correlated with DNA damage measured as 32P-post-labeling bulky DNA adducts in white blood cells from peripheral blood. Genotyping was performed by PCR-RFLP analysis, and allele frequencies in cases/controls were as follows: XRCC1-399Gln = 0.34/0.39, XRCC3-241Met = 0.48/0.35 and XPD-751Gln = 0.42/0.42. Odds ratios (ORs) were significantly greater than 1 only for the XRCC3 (exon 7) variant, and they were consistent across the 2 control groups. XPD and XRCC1 appear to have no impact on the risk of bladder cancer. Indeed, the effect of XRCC3 was more evident in non-smokers [OR = 4.8, 95% confidence interval (CI) 1.1–21.2]. XRCC3 apparently interacted with the N-acetyltransferase type 2 (NAT-2) genotype. The effect of XRCC3 was limited to the NAT-2 slow genotype (OR = 3.4, 95% CI 1.5–7.9), suggesting that XRCC3 might be involved in a common repair pathway of bulky DNA adducts. In addition, the risk of having DNA adduct levels above the median was higher in NAT-2 slow acetylators, homozygotes for the XRCC3-241Met variant allele (OR = 14.6, 95% CI 1.5–138). However, any discussion of interactions should be considered preliminary because of the small numbers involved. Our results suggest that bladder-cancer risk can be genetically modulated by XRCC3, which may repair DNA cross-link lesions produced by aromatic amines and other environmental chemicals. © 2001 Wiley-Liss, Inc.
A potentially important source of interindividual variability in response to carcinogens is DNA-repair capability. Apart from rare recessive inherited syndromes such as ataxia-telangiectasia, Fanconi's anemia and Bloom's syndrome, all of which are characterized by both chromosomal instability and high risk of cancer, and xeroderma pigmentosum, characterized by extreme susceptibility to UV-induced skin cancer,1 individuals differ widely in the ability to repair DNA damage.2 Polymorphisms in DNA-repair genes have been described.3, 4 Their role in modulating the risk of environmentally induced cancer is potentially important but has been studied only occasionally. In particular, XPD, XRCC1 and XRCC3 have been proposed as polymorphic genes that might be involved in environmental carcinogenesis.5–10
The XRCC1 protein is involved in the base excision-repair pathway,11 acting apparently as a scaffold protein, facilitating the repair reaction by binding DNA ligase III at its carboxy and DNA polymerase β to its amino terminus.12 XRCC3 participates in DNA double-strand breaks/recombinational repair and belongs to an emerging family of Rad51-related proteins that participate in homologous recombination (HR) to maintain chromosomal stability.13, 14 It is likely that removal of interstrand cross-links is strongly dependent on efficient HR repair, as suggested by the extraordinary sensitivity of the ERCC1, XPF/ERCC4, XRCC2 and XRCC3 mutants to cross-linking chemicals.15 XRCC3 interacts directly with HsRad51 and, like Rad55 and Rad57 in yeast, may co-operate with HsRad51 during recombinational repair.16 XPD is involved in the nucleotide excision repair (NER) pathway, which recognizes and repairs a wide range of structurally unrelated lesions such as bulky adducts and thymidine dimers.17, 18 XPD functions as an ATP-dependent 5′–3′ helicase joint to the basal transcription factor IIH complex.19 Two of the polymorphisms analyzed, XRCC3-T241M and XRCC1-R399Q, are non-conservative amino acid changes. The XPD-Lys751Gln polymorphism is a conservative substitution which does not reside in known or hypothesized helicase/ATPase domains.
A high frequency of chromosomal instability or aberrations has been reported in bladder cancer. Aberrations in chromosomes 8 and 11 and allelic losses on chromosomes 4, 8, 9, 11 and 17 have been described.20–24
The risk of bladder cancer can be modulated by genetic metabolic polymorphisms. GSTM1 is polymorphic25 and plays a role in the metabolism of organic epoxides and peroxides; in particular, it conjugates known carcinogens such as epoxides of polycyclic aromatic hydrocarbons (PAHs). Other GSTs potentially relevant to the risk of bladder cancer are T1 and P1. N-Acetyltransferase-2 (NAT-2) differentially acetylates arylamines, well-known bladder carcinogens, to arylamides. Slow acetylators are at higher risk for developing bladder cancer.25
Previously, we studied the association between the risk of bladder cancer and white blood cell (WBC)–DNA adducts, including modulation by consumption of fruit and vegetables.26, 27 In the present study, we have analyzed 3 DNA-repair polymorphisms and their interaction with WBC–DNA adducts in 124 bladder-cancer cases and 85 controls.
MATERIAL AND METHODS
We performed a hospital-based case-control investigation at the urology departments of 2 Turin hospitals (Gradenigo and S. Giovanni Battista), where about half the cases of newly diagnosed bladder cancer in the Turin metropolitan area are treated. The men making up the case group were aged 45 to 74 years, resident in the Turin metropolitan area and treated from 1994 to 1996 for histologically confirmed bladder cancer. All were incident (newly diagnosed) cases. Cases were identified by daily contact between a trained interviewer and the urology departments. Histological confirmation was obtained from the pathology departments. Controls were recruited daily in random fashion (i) from patients treated at the same urology departments for benign diseases, mainly prostatic hyperplasia and cystitis (all newly diagnosed), and (ii) from patients treated at the medical and surgical departments for hernias, vasculopathies, diabetes, heart failure, asthma or other benign diseases (none was represented in >10% of controls). Patients with cancer, liver or renal diseases and smoking-related conditions were excluded. Like the cases, controls were men aged 40 to 74 years and living in the Turin metropolitan area. Before therapy began, a trained interviewer used a standard questionnaire to interview cases and controls on history of tobacco smoking (including brands and tobacco type) and a 24 hr recall interview to collect information about dietary habits, drug use and occupational history. Before therapy and after informed consent, blood was collected from cases and controls.
WBC DNA was isolated and purified from stored buffy coats by enzymatic digestion of RNA and proteins, followed by phenol-chloroform extraction. WBC-DNA adduct levels were measured using the 32P-DNA post-labeling technique, as described previously.26, 27 DNA-adduct analyses were carried out at the Istituto Nazionale per la Ricerca sul Cancro (Genoa, Italy). Three known slow-acetylator alleles (NAT2*5A/5B, NAT2*6A/6B and NAT2*7A/7B) were identified as described,26, 27 with slight modifications. Rapid-acetylator genotypes are wild-type allele homo-/heterozygotes; slow-acetylator genotypes are those with 2 slow-acetylator alleles.28
PCR followed by enzymatic digestion was also used for genotyping of XRCC1-Arg399Gln, XPD-Lys751Gln and XRCC3-Thr241Met polymorphisms. All PCRs were performed in a total reaction volume of 20 μl containing 10 ng genomic DNA, 0.4 units of Taq polymerase (Perkin-Elmer Applied Biosystems, Foster City, CA) in PCR buffer 1×, 1.5 mM MgCl2, 50 mM dNTPs and 250 nM of each primer. Thermal cycling conditions were as follows: initial denaturation step at 95°C for 3 min; 35 cycles of PCR consisting of 95°C for 20 sec, 20 sec at the appropriate annealing temperature and 72°C for 20 sec; then a final extension step at 72°C for 5 min. The XRCC1-Arg399Gln polymorphism, a transition G→A in exon 10 (position 28152), was determined using the primers (sense) 5′-CAAGTACAGCCAGGTCCTAG-3′ and (anti-sense) 5′-ccttccctca tctggagtac-3′ at 55°C annealing temperature for PCR. The 248 bp PCR product was digested with NciI (Promega, Madison, WI): the Arg allele was cut into 89 and 159 bp fragments (Gln allele not digested). The XPD-Lys751Gln polymorphism, a transversion A→C in exon 23 (position 35931), was determined using the primers (sense) 5′-CTGCTCAGCCTGGAGCAGCTAGA ATCAGAGGAGACGCTG-3′ and (anti-sense) 5′-AAGACCTTCTAGCACCACCG-3′ at 67°C annealing temperature for PCR. The 161 bp PCR product was digested with PstI (Promega): the Gln allele was cut into 41 and 120 bp fragments (Lys allele not digested). The XRCC3-Thr241Met polymorphism, a transition C→T in exon 7 (position 18067), was determined using the primers (sense) 5′-GCCTGGTGGTCATCGACTC-3′ and (anti-sense) 5′-ACAGGGCTCTGGAAGGCACTGCTCAGC TCACGCACC-3′ (underlined base modifies primer sequence introducing a cut site in the presence of the Met allele) at 60°C annealing temperature for PCR. The 136 bp PCR product was digested with NcoI (Promega): the Met allele was cut into 39 and 97 bp fragments (Thr allele not digested).
DNA typing quality control
Methodological validation included a comparison between PCR-RFLP, direct sequencing and denaturing high-performance liquid chromatography (DHPLC).29 Due to the small amount of DNA available, we checked the accuracy of PCR-RFLP genotyping for the 3 polymorphisms using the primer extension technique with DHPLC (Transgenomic, Santa Clara, CA) on a set of 50 individuals from a cardiovascular study, and complete agreement in the typing was obtained. The primer extension technique30 is based on specific incorporation of the complementary dideoxynucleotide (ddNTP) into the base substitution; the rate of right incorporation completely overwhelms ddNTP misincorporations, and heterozygotes are easily detectable. The following primers flanking the base substitution were used: XRCC1-Arg399Gln, 5′-cggcggctgccctccc-3′; XRCC3-Thr241Met, 5′-GGCATCTGCAGTCCCTGGGGGCCA-3′; XPD-Lys751Gln, 5′- ATCTGCTCTATCCTCT-3′. Due to the results described below for XRCC3, we decided to re-analyze the XRCC3-Thr241Met polymorphism by DHPLC in the whole sample (124 cases and 85 controls). All DHPLC typings confirmed the PCR-RFLP results.
Odds ratios (ORs) and the corresponding 95% confidence intervals (CIs) were computed with the SAS (Cary, NC) package for personal computers. Unconditional logistic regression models were fitted. Age was not associated with the distribution of polymorphisms, and age-adjusted ORs were identical to unadjusted estimates. Since we noticed in a previous analysis that the 2 control groups (urological and non-urological) had different levels of DNA adducts,31 we analyzed them separately. In addition, we performed a “case-only analysis” to avoid bias related to the choice of the control group. Case-only studies have proven to be a valid design to test gene–environment and gene–gene interactions, with accurate and precise estimates of association.32 A case-only approach is justified if the variables considered for interaction are independent in the control series.
The size of the study was sufficient to detect an OR of 2.3 with α = 0.05 and β (type-II error) = 20%, assuming a frequency of the relevant allele of 50% (XRCC3-241 exon 7, Thr/Thr genotype).
We recruited overall 162 bladder-cancer cases and 104 controls, who provided biological samples. All were interviewed except 2 cases and 2 controls who refused and 3 cases who were too ill to answer the questions. We also interviewed 42 cases who did not provide biological samples. Although they did not differ from the others on clinical or pathological traits or smoking habits,27 they were excluded from the present analysis nonetheless. DNA was of a sufficient amount to perform analyses of DNA-repair gene polymorphisms for 124 cases and 85 controls. The following data refer only to these subjects. Table I gives information about the main characteristics of cases and controls. Smoking habits were analyzed in a previous report.26
|Controls(n = 85)||Cases(n = 124)|
|Age groups (years)|
|45–54||20 (23%)||17 (14%)|
|55–64||34 (40%)||45 (36%)|
|65–69||20 (23%)||36 (29%)|
|70–74||11 (13%)||26 (21%)|
|χ2 5.33 (p = 0.15)|
|Never smoked||38 (45%)||13 (10%)|
|Ex-smoker||22 (26%)||37 (30%)|
|Current smoker||25 (29%)||74 (60%)|
|χ2 34 (p = 0.01)|
|Number of cigarettes smoked (highest amount throughout life)|
|0||38 (45%)||13 (10%)|
|1–15||11 (13%)||25 (20%)|
|15–29||24 (28%)||58 (47%)|
|30+||12 (14%)||28 (23%)|
|χ2 32 (p = 0.001)|
No polymorphism showed deviations from the Hardy-Weinberg equilibrium (HWE) in either cases or controls, except XRCC3 in controls (χ = 7.61, p ≤ 0.01), though the XRCC3 allele frequencies in our control group (Met = 0.35 and Thr = 0.65) were comparable with those described by Shen et al.3 (Met = 0.38 and Thr = 0.62). Specifically, the non-urological control group was not in HWE (χ = 7.83, p ≤ 0.01), whereas the urological control group was (χ = 0.83, p > 0.05). There appears to be no biological or functional explanation for this result. We calculated the heterogeneity between samples as G2 (het.) = 7.30, p ≤ 0.01. Even when the deviation from HWE was taken into account (see Appendix), the difference remained statistically significant.
In the whole control group, the XRCC1-399Gln variant frequency (0.39) was similar to that reported by Lunn et al.7 for the Caucasoid population (0.37) but different from that estimated by Shen et al.3 analyzing DNA sequences in 12 individuals (0.25). The XPD-751Gln frequency was higher (0.42) than that estimated by Shen et al.3 in the United States (0.29) and that estimated by Broughton et al.5 in England (0.30) but similar to that estimated by Duell et al.10 in the United States (0.39). Allele frequencies in the case group were as follows: XRCC1-399Arg/Gln = 0.66/0.34, XRCC3-241Thr/Met = 0.52/0.48 and XPD-751Lys/Gln = 0.58/0.42.
The NAT-2 genetic polymorphism was associated with the risk of bladder cancer, with a statistically significant OR of 1.72 (95% CI 1.03–2.87) for slow acetylators compared with rapid acetylators. This was the only metabolic polymorphism among those analyzed that showed a statistically significant association with bladder cancer.26
Table II shows genotype distributions for the 3 DNA-repair gene polymorphisms analyzed (XRCC1-Arg399Gln, exon 10; XRCC3-Thr241Met, exon 7; XPD-Lys751Gln, exon 23). The results are shown separately for the 2 control groups, urological and non-urological. ORs are significantly greater than 1 only for XRCC3 (exon 7), and they are consistent across the 2 control groups. Smoking habits appear to play a modulating role since the effect of the XRCC3 polymorphism was twice as high in non-smokers than in smokers (Table III) (however, the difference among smoking categories was not statistically significant: χ for heterogeneity = 1.046, p = 0.593). No significant difference in bulky DNA-adduct levels was observed for XRCC1-Arg399Gln and XPD-Lys751Gln polymorphisms when stratifying by smoking status (Table III).
|Cases||Urological controls||Non-urological controls|
|XRCC1, exon 10|
|Arg/Arg||53 (43%)||12 (33%)||19 (40%)|
|Arg/Gln||58 (47%)||19 (51%)||22 (46%)|
|Gln/Gln||13 (10%)||6 (16%)||6 (14%)|
|Arg/Gln + Gln/Gln||72 (57%)||25 (67%)||28 (60%)|
|OR (95% CI)||0.60 (0.21–1.71)||0.80 (0.28–2.25)|
|p = 0.69 (1 missing value)|
|XRCC3, exon 7|
|Thr/Thr||33 (26%)||19 (50%)||23 (49%)|
|Thr/Met||64 (52%)||14 (37%)||13 (28%)|
|Met/Met||27 (22%)||5 (13%)||11 (23%)|
|Thr/Met + Met/Met||91 (74%)||19 (50%)||24 (51%)|
|OR (95% CI)||2.84 (1.34–6.04)||2.72 (1.37–5.43)|
|OR (95% CI), controls combined||2.77 (1.55–4.93)|
|p = 0.007|
|Lys/Lys||39 (32%)||12 (32%)||12 (26%)|
|Lys/Gln||66 (53%)||23 (60%)||27 (57%)|
|Gln/Gln||19 (15%)||3 (8%)||8 (17%)|
|Lys/Gln + Gln/Gln||85 (68%)||26 (68%)||35 (74%)|
|OR (95% CI)||0.94 (0.42–2.10)||0.71 (0.32–1.58)|
|p = 0.69|
|XRCC1, exon 10|
|Arg/Arg||31 (0.55 ± 0.10)||7 (0.31 ± 0.18)||16 (0.33 ± 0.06)||5 (0.15 ± 0.08)||6 (0.29 ± 0.10)||19 (0.13 ± 0.05)|
|Arg/Gln + Gln/Gln||43 (0.55 ± 0.08)||17 (0.26 ± 0.07)||21 (0.30 ± 0.06)||17 (0.26 ± 0.13)||7 (0.55 ± 0.15)||19 (0.17 ± 0.04)|
|OR (95% CI)||0.43 (0.15–1.25)||0.27 (0.07–1.02)||1.05 (0.29–3.89)|
|XRCC3, exon 7|
|Thr/Thr||18 (0.54 ± 0.11)||10 (0.22 ± 0.10)||13 (0.31 ± 0.16)||13 (0.37 ± 0.08)||2 (0.28 ± 0.06)||19 (0.11 ± 0.04)|
|Thr/Met + Met/Met||56 (0.55 ± 0.08)||15 (0.29 ± 0.09)||24 (0.28 ± 0.05)||9 (0.12 ± 0.05)||11 (0.45 ± 0.11)||19 (0.18 ± 0.05)|
|OR (95% CI)||1.8 (0.7–4.8)||5.2 (1.2–21.5)||4.8 (1.1–21.2)|
|XPD, exon 23|
|Lys/Lys||24 (0.55 ± 0.11)||13 (0.16 ± 0.04)||10 (0.28 ± 0.08)||4 (0.63 ± 0.53)||5 (0.43 ± 0.15)||7 (0.13 ± 0.08)|
|Lys/Gln + Gln/Gln||50 (0.55 ± 0.08)||12 (0.37 ± 0.13)||27 (0.32 ± 0.05)||18 (0.15 ± 0.04)||8 (0.43 ± 0.14)||31 (0.15 ± 0.04)|
|OR (95% CI)||2.53 (0.92–6.96)||0.60 (0.14–2.68)||0.36 (0.09–1.55)|
Table IV shows that XRCC3 apparently interacts with the NAT genotype. Although based on small numbers, Table IV shows that the effect of XRCC3 is limited to the NAT-2 slow genotype, suggesting that XRCC3 might be involved in a common repair pathway of bulky DNA adducts. In a logistic regression model including age, XRCC3 (exon 7, dichotomized as Thr/Thr vs. Thr/Met+Met/Met), NAT-2 and an interactive term for XRCC3 and NAT-2, ORs were 1.38 (95% CI 0.60–3.17) for XRCC3, 1.09 (95% CI 0.52–2.29) for NAT-2 and 2.28 (95% CI 0.78–6.66) for the interactive term. The effect of XRCC3 was present also in the case-only analysis, which was not influenced by the choice of controls, and when DNA adducts in WBCs were considered in relation to the NAT-2 and XRCC3 genotypes. In a logistic regression model including a 3-way interactive term (based on NAT-2, the median of DNA adducts and dichotomized XRCC3), ORs were 1.42 (95% CI 0.64–3.13) for XRCC3, 0.95 (95% CI 0.47–1.92) for NAT-2, 3.17 (95% CI 1.77–5.69) for the median of DNA adducts and 1.81 (95% CI 0.92–3.56) for the interactive term. Since intake of fruit and vegetables modified the association between DNA adducts and bladder cancer, we also included this variable in the analyses, but there was no change in the estimates.
|Case-control analysis: Cases/controls (OR and 95% CI in parentheses)|
|XRCC3||NAT-2 rapid||OR (95% CI)||NAT-2 slow||OR (95% CI)||NAT-2 rapid2||NAT-2 slow2|
|A||B||OR (95% CI)||A||B||OR (95% CI)|
|Thr/Thr||15/17||1.0||18/25||1.0||8/5||7/12||2.7 (0.6–11.7)||11/8||7/17||3.3 (0.9–11.8)|
|Thr/Met||18/15||1.4 (0.5–3.6)||45/12||5.2 (2.2–12.2)||8/4||10/11||2.2 (0.5–9.6)||32/5||13/7||3.4 (0.9–12.8)|
|Met/Met||6/6||1.1 (0.3–4.3)||21/10||2.9 (1.1–7.6)||3/3||3/3||1.0 (0.1–9.6)||13/1||8/9||14.6 (1.5–138)|
|Thr/Met + Met/Met||24/21||1.3 (0.5–3.2)||66/22||3.4 (1.5–7.9)||11/7||13/14||2.2 (0.6–8.5)||45/6||21/16||5.2 (1.8–15.1)|
|XRCC3||NAT-2 rapid||NAT-2 slow||OR (95% CI)||NAT-2 rapid3||NAT-2 slow3||OR (95% CI)|
|Thr/Met||18||45||2.1 (0.9–5.0)||7||11||24||21||1.8 (0.6–5.5)|
|Met/Met||6||21||2.9 (0.95–9.1)||2||4||11||10||2.2 (0.3–14.9)|
|Thr/Met + Met/Met||24||66||2.25 (1.0–4.9)||9||15||35||31||1.9 (0.7–4.9)|
In a case-control study, we found a statistically significant association between the Thr241Met polymorphism of the XRCC3 gene, involved in DNA repair, and the risk of bladder cancer. The association was particularly apparent among those with the slow-acetylator genotype and among non-smokers. In addition, increased levels of bulky DNA adducts in WBCs were more frequent in bladder-cancer patients with the XRCC3-241Met variant who were slow acetylators.
XRCC3 participates in DNA double-strand break and cross-link repair through homologous recombination and contributes, as other RAD51-related proteins, to the maintenance of chromosomal stability.13, 14, 33 One study reported an association between a rare microsatellite polymorphism in the XRCC3 gene and cancer in patients with varying radiosensitivity.6 The Thr241Met substitution in XRCC3 is a non-conservative change with possible biological implications for the functionality of the enzyme and/or for the interaction with other proteins involved in DNA repair.
The association between bulky DNA-adduct formation and XRCC3 Thr/Met and Met/Met genotypes (particularly in the slow NAT-2 group) may be related to environmental exposure to genotoxic aromatic amines, such as trans-4-dimethylaminostilbene and 4-trans-acetylaminostilbene,34 which are capable of forming DNA adducts to guanine and adenine and of inducing other secondary lesions of equal or greater importance, e.g., cross-links between bases. 4-Aminostilbene has been reported to induce high levels of chromosomal aberrations.35 The association between DNA adducts and the XRCC3 polymorphism may also be due to oxidation reactions, which might cause formation of intrastrand cross-links between adjacent nucleotides, leading to bulky oxidative DNA modification, i.e., dimer formation, detectable by 32P-DNA post-labeling.36
In studies on tobacco use, smoking has been clearly associated with the risk of bladder cancer, yet no relationship between smoking and WBC-DNA adduct levels (p > 0.05) was observed in the present study.26 The level of DNA adducts was strongly associated with case/control status.27 The age-adjusted OR associated with an adduct level above the limit of detection was 3.7 (95% CI 2.2–6.3), and a dose-response relationship with adduct levels was apparent.27
Surprisingly, the association between bladder cancer and the XRCC3 polymorphism was higher in non-smokers and ex-smokers (Table III), which is difficult to explain but consistent with observations made on the association between DNA adducts and DNA-repair polymorphisms in a group of healthy subjects (data not shown). A potential explanation is that smoking induces DNA-repair enzymes so that the difference in adduct levels observed in peripheral leukocytes among genotypes is overcome in current smokers, whereas the difference becomes detectable in non-smokers.
Our results concerning bulky DNA adducts are rather surprising because such adducts are known to undergo NER rather than recombinational repair. This might means that an important fraction of cross-links could be detectable by the 32P-post-labeling36 and/or that XRCC3 can lower DNA-adduct levels by acting on different repair pathways.
If our findings are correct, a high frequency of chromosomal instability or aberrations induced by carcinogens in bladder cancer is to be expected. Indeed, a number of studies have shown aberrations and allelic losses in different chromosomes,20–24 so bladder cancer can be considered one of the cancers prone to chromosomal instability.
Our study has some limitations. Specifically, statistical power was relatively limited and the HWE was not met in 1 of the control groups (surgical controls). The latter problem might be explained by selection bias. However, the association with XRCC3 was statistically significant when using either control group; also, the response rate was very high, both control groups were heterogeneous (with few subjects having the same diagnosis) and there was no evidence of selection bias. In addition, the association with the XRCC3 polymorphism was confirmed in a case-only analysis, i.e., excluding controls. Despite limited power, 95% CIs were reasonably narrow. Multiple comparisons were not a major problem in this study since we looked at 3 polymorphisms, based on a priori choice.
In conclusion, the XRCC3-241Met variant can be associated with human bladder cancer, showing an interaction with the NAT-2 gene. However, any discussion of interactions should be considered preliminary because of the small numbers involved. Our results suggest the need to further investigate possible synergistic effects between DNA repair and metabolic genetic polymorphisms in modulating cancer risk.
We tested for deviations from the HWE using the G2 statistic:37
This statistic is used to test whether an observed number is significantly different from the number expected on the basis of a specific hypothesis (e.g., HWE), and it is distributed as a χ2 statistic. When phenotypes (p) are compared in a number of samples (n), the heterogeneity between their gene frequencies (g), which takes into account the effect of the deviation from HWE, can be written as follows:
where G2 (T) with (p – 1)(n – 1) degrees of freedoms (df) tests for the difference of phenotypes as the usual (p × n) contingency table; G2 (H.W.) with (p – g) df tests for the deviation from HWE in the total sample formed by pooling n samples; G2 (Shw) with n(p – g) df is the sum of n tests of deviation from HWE calculated in each sample; and G2 (het.) with (g – 1)(n – 1) tests for the heterogeneity of the gene frequencies (g) between n samples.
- 1DNA repair and mutagenesis. Washington DC: ASM Press, 1995., , .
- 25Metabolic polymorphsims and susceptibility to cancer. IARC Scientific Publication 148, Lyon: IARC, 1999., , , , , .
- 37Biometry. San Francisco: Freeman, 1969., .