XRCC1 and XPD polymorphisms and esophageal adenocarcinoma risk

Carcinogenesis vol.28 no.6 pp.1254–1258, 2007 doi:10.1093/carcin/bgm020 Advance Access publication January 29, 2007

XRCC1 and XPD polymorphisms and esophageal adenocarcinoma risk Geoffrey Liu1,2,3,
, Wei Zhou2, Beow Y.Yeap1, Li Su2, John C.Wain4, John M.Poneros5, Norman S.Nishioka1, Thomas J.Lynch1 and David C.Christiani1,2 1 Department of Medicine, Massachusetts General Hospital, 55 Fruit Street, Boston, MA 02114, USA, 2Occupational Health Program, Harvard School of Public Health, 665 Huntington Ave, Boston, MA 02115, USA, 3Ontario Cancer Institute/Princess Margaret Hospital, Toronto, Room 7-124, 610 University Avenue, Toronto, Ontario M5G 2M9, Canada, 4Department of Surgery, Massachusetts General Hospital, 55 Fruit Street, Boston, MA 02114, USA and and 5Department of Medicine, Brigham and Women’s Hospital, 75 Francis Street, Boston, MA 02115, USA
To whom correspondence should be addressed. Tel: þ1 416 946 4501 ext. 3428; Fax: þ1 416 946 6546; Email: [email protected]

Introduction In North America, the majority of esophageal cancers are esophageal adenocarcinomas (EAs). EA develops most commonly from the columnar epithelium lining the distal esophagus, and has had the most rapid rate of increase of any solid tumor malignancy (1). The overall North American annual incidence of EA has more than tripled between 1976 and 1995 (2), affecting all ethnic subgroups and both genders (3). Specific subgroups, such as Caucasian males, have had a 4-fold increase in incidence in just the 10 years from 1988 to 1998, a trend that does not appear to be reaching a plateau (3). Abbreviations: AOR, adjusted odds ratio; BER, base excision repair; BMI, body mass index; CI, confidence interval; EA, esophageal adenocarcinoma; GERD, gastroesophageal reflux disease; NER, nucleotide excision repair; OR, odds ratio; XPD, xeroderma pigmentosum group D; XRCC1, X-ray repair cross-complementing gene 1.

Methods Case and control population The study was approved by the Human Subjects Committees of Massachusetts General Hospital, Harvard School of Public Health, and Princess Margaret Hospital. Cases were histologically confirmed, incident (diagnosed ,6 months prior to study entry) EA patients at Massachusetts General Hospital recruited from 1999 to 2003. After 200 EA cases were recruited, we began secondary screening using medical, endoscopic and pathologic reports to ensure that patients met strict criteria for this analysis. Sixteen patients were excluded from the analysis for the following reasons: 11 patients were reclassified as gastric cardia (midpoint of length of tumor was located below the gastroesophageal junction), one patient was found to have a previous cancer diagnosis that was not self-reported, two patients had histologic diagnoses of poorly differentiated carcinomas suspicious for but not diagnostic of adenocarcinoma and two patients had significant missing covariate data due to aborted interviews (both due to acuity of patient illness). The remaining 184 cases formed the basis of the case group. Controls were selected from healthy friends (47%) and non-blood-related family members (53%) of hospital patients, originally recruited as healthy controls for our parallel lung cancer case–control studies. All controls never had any diagnosis of cancer (30); controls were recruited during the years 1999–2003 through Massachusetts General Hospital and were unrelated to EA cases. Selected controls were, as a group, matched with cases to similar race, gender and age distributions. Initially, 400 frequency-matched potential controls were selected. These potential controls underwent secondary screening using their questionnaire data, and were excluded if they had a history of GERD symptoms (n 5 59), missing covariate information (n 5 13) or

Ó The Author 2007. Published by Oxford University Press. All rights reserved. For Permissions, please email: [email protected]

1254

Downloaded from http://carcin.oxfordjournals.org/ by guest on August 18, 2015

DNA damage is important in the pathogenesis of esophageal adenocarcinoma (EA). Polymorphic variants in DNA repair genes may be modifiers of the risk of EA through their role in altering human host response to gastroesophageal acid reflux, a welldescribed risk factor for EA. We studied the role of genetic polymorphisms of two key DNA repair genes, xeroderma pigmentosum group D (XPD) (Asp312Asn and Lys751Gln) in the nucleotide excision repair (NER) pathway and X-ray repair crosscomplementing gene 1 (XRCC1) (Arg399Gln) in the base excision repair (BER) pathway, in the development of EA in 183 cases and 336 frequency-matched controls for age, gender and race. Genomic DNA was extracted from blood samples. Odds ratios (ORs) and 95% confidence intervals (CIs) were obtained from logistic regression models, adjusted for body mass index at 18 years of age, smoking and alcohol exposure. The variant genotypes of XPD Lys751Gln polymorphism were associated with a higher risk of EA; the adjusted OR comparing Gln/Gln þ Lys/Gln with Lys/Lys was 1.49 (95% CI: 1.02–2.14). Although no significant relationships were found for the XRCC1 Arg399Gln polymorphism alone, this polymorphism did modify the relationship between XPD Lys751Gln and EA risk; when both polymorphisms were evaluated together, adding the number of variant alleles of the two polymorphisms resulted in a significant trend (trend test, P 5 0.008); compared with individuals with no variant alleles (n 5 88), the adjusted ORs of developing EA are 1.49 (95% CI: 0.88–2.59), 1.69 (95% CI: 0.98–2.96) and 2.58 (95% CI: 1.31– 5.06) for one (n 5 195), two (n 5 166) and three or four variant alleles (n 5 70), respectively. No relationships were found for the XPD Asp312Asn polymorphism. We conclude that combined NER and BER pathways are important to the development of EA.

There is little doubt that environmental factors are key to this dramatic rise in incidence. However, many of the known environmental associations with EA risk affect a significant proportion of the population. For instance, chronic gastroesophageal reflux disease (GERD) is a known risk factor for EA, and up to 15–20 million North Americans suffer from GERD. Yet, there are fewer than 15 000 annual cases of EA representing 0.1% or less of the at-risk population (4). Other risk factors include Barrett’s esophagus, smoking, alcohol, obesity and possibly dietary factors (5–9). The small proportion of patients with risk factors who will ultimately develop EA (4) suggests that gene–environment interactions are playing a role. The specific environmental factors involved are difficult to identify, since these factors are likely to involve chronic low-level environmental or dietary exposures rather than acute single-dose exposures. An improved understanding of modifying genetic polymorphisms may help to identify the appropriate environmental factors to test in gene–environment analyses. DNA repair and maintenance is essential in protecting the genome of the cell from environmental hazards. Altered DNA repair capacity can render a higher risk of developing different types of cancer (10– 16). X-ray repair cross-complementing gene 1 (XRCC1), a major participant in the base excision repair (BER) pathway, has multiple roles in repairing base damage and single-stranded DNA breaks in the BER pathway (17). Mutations in the transcription-coupled nucleotide excision repair (NER) gene, xeroderma pigmentosum group D (XPD) (or ERCC2), can give rise to repair and transcription defects, with abnormal responses to apoptosis (18). Polymorphisms of these two genes are associated with lower DNA repair capacities (19–25), and have been associated with a higher risk of esophageal squamous cell carcionoma (26), colorectal and gastric adenocarcinomas (27,28). Recently, both the XRCC1 Arg399Gln and XPD Lys751Gln polymorphisms have been identified as potential modifiers of EA risk (9,29), but the two small published studies evaluating these polymorphisms (n 5 56 and n 5 96 cases, respectively) had opposite conclusions regarding the direction of the associations. In the present study, we investigated the associations between XRCC1 (Arg399Gln) and XPD (Asp312Asn and Lys751Gln) polymorphisms and EA risk in a larger study of 183 cases.

XRCC1, XPD polymorphism and esophageal cancer

self-reported diagnosis of Barrett’s esophagus (n 5 3). Several of the patients met several exclusion criteria. Three hundred and thirty-eight controls remained after completion of the secondary screening. Interview Immediately after enrollment, a trained interviewer administered a questionnaire that collected clinical and demographic information. Cases were interviewed during their hospital/clinic visits. Because controls were recruited when they accompanied other cancer patients to their hospital/clinic visits, interviews for the controls took place in the same area as the cases and by the same interviewers. All interviews were conducted in person using specially trained research assistants, The questions covered demographic variables (current weight, weight 1 year prior to diagnosis/interview, weight at early adult age of 18 years, adult height, age, gender, race, etc.), a detailed smoking and alcohol exposure assessment, past medical history (including exposure to radiation for benign conditions), family history and occupational and environmental history. Lifetime GERD symptoms were assessed up to 1 year prior to diagnosis (cases) or 1 year prior to interview (controls). A Harvard–Willett food frequency questionnaire was also administered, basing the results on food intake 1 year prior to diagnosis for cases and 1 year ago from date of interview for controls. We defined smoking and alcohol status on the basis of whether the cases and controls smoked/drank alcohol 1 year prior to diagnosis/interview.

Statistical analysis After secondary screening, the remaining cases and controls were compared again for age, gender and race distributions to ensure that frequency matching was maintained. We analyzed all subjects with complete information on race, age, gender, smoking status (never, ex- and current smokers), alcohol status (ever and never alcohol drinker) and body mass index (BMI) at 18 years old. Demographic and genotype information was compared across genotype using Pearson chi-square tests (for categorical variables) and non-parametric Wilcoxon rank sum test (for continuous variables) where appropriate. Analyses of all genotype associations with EA risk were based on logistic regression models (34). Because the matching was for the entire sample, rather than for each individual, unconditional logistic regression was used for all analyses. We adjust for predetermined potential confounding factors such as race, age, gender, smoking status, alcohol status and BMI at the age of 18 years (BMI-18, a surrogate measure of adult BMI). We chose to evaluate BMI at 18 years of age instead of current BMI because the majority of patients with EA lost weight within the year prior to diagnosis, rendering current weight an inappropriate substitute for usual adult weight. In addition, we also investigated the joint effects of the combined XRCC1 and XPD polymorphisms. A lack of fit test was performed to summarize the goodness of fit for each logistic regression model (34). Where appropriate, odds ratios (ORs) and 95% confidence interval (CI) for the risk of EA were calculated from these models. Statistical analyses were all undertaken using SAS statistical packages (SAS 8.1, SAS Institute, Cary, NC).

Results There were no significant differences in recruitment rates (race, age or gender) between enrolled and unenrolled eligible cases (87%

Table I. Demographic characteristics of EA cases and controls Case (n 5 183) Racea Caucasian 180 (98%) Non-Caucasian 3 (2%) 65 (33–91) Ageb Gendera Female 19 (10%) Male 164 (90%) Smoking statusa Non-smokers 30 (16%) Ex-smokers 120 (66%) Current smokers 33 (18%) Alcohol usea Never 15 (8%) Ever 168 (92%) a,b 23 (15–43) BMI at 18 years 25 139 (76%) 25–30 33 (18%) .30 11 (6%) XRCC1 Arg399Gln polymorphisma Arg/Arg 77 (42%) Arg/Gln 77 (42%) Gln/Gln 29 (16%) XPD Asp312Asn polymorphisma,c Asp/Asp 75 (41%) Asp/Asn 92 (50%) Asn/Asn 16 (9%) XPD Lys751Gln polymorphisma Lys/Lys 61 (33%) Lys/Gln 98 (54%) Gln/Gln 23 (13%)

Control (n 5 336)

P

332 (99%) 4 (1%) 65 (38–84)

0.67 0.37

42 (12%) 294 (88%)

0.47

100 (30%) 180 (53%) 56 (17%)

,0.01

17 (5%) 319 (95%) 22 (14–33) 269 (80%) 58 (17%) 9 (3%)

0.16 0.15 0.16

153 (46%) 142 (42%) 41 (12%)

0.48

144 (43%) 160 (48%) 32 (9%)

0.86

143 (43%) 161 (48%) 32 (10%)

0.12

a

Pearson’s chi-square test. Median (range), Wilcoxon rank test. c Percentages may not add up to a hundred due to rounding. b

1255

Downloaded from http://carcin.oxfordjournals.org/ by guest on August 18, 2015

Genotyping DNA was extracted from peripheral blood samples using the Puregene DNA Isolation Kit (Gentra Systems, Minneapolis, MN). The XRCC1 Arg399Gln polymorphism was detected using modified polymerase chain reaction–restriction fragment length polymorphism methods and published primer sequences (21,31). In brief, a 242 bp polymerase chain reaction product that included the Arg/Gln (A/C) allele in exon 10 was amplified, followed by MspI enzyme digestion (New England BioLabs, Beverly, MA). The genotyping methods for the XPD Asp312Asp (exon 10) and Lys751Gln (exon 23) polymorphisms have been described in detail (21,32). Briefly, two separate polymerase chain reaction assays were used to detect the polymorphisms in exon 10 and exon 23 of XPD using published primer sequences. DpnII and MspI (for exon 10) and MboII enzyme (for exon 23) digestions (New England BioLabs) were used for restriction fragment analyses. For quality control purposes, a random 5% of the samples were repeated to assess the reproducibility of results. Two authors independently reviewed 100% of the agarose gels, while a third arbitrated, when necessary. In addition, all cases and controls were also genotyped using a Taqman approach (33). Concordance in results between the two techniques was found for 99.4% of the samples. Discrepant results were genotyped several times using both platforms. One case and two controls had inconclusive or conflicting results from the two techniques for at least one of the three polymorphisms and were excluded from further analysis. Thus, 183 cases and 336 controls were analyzed.

participation rate) and controls (90% participation rate) for the base populations. After secondary screening and genotyping, a total of 183 EA cases and 336 controls were analyzed. Table I details the demographic characteristics of cases and controls. Age, race and gender distributions were similar between cases and controls. There were greater proportions of never smokers in the controls than the cases. There was a non-significant trend for cases to have higher BMI at the age of 18 years than controls at that age. Among the cases, 48 (26%) were Stage I–IIA (node-negative localized disease), 73 (40%) were Stage IIB–III (node-positive regional disease) and 62 (34%) were Stage IV (metastatic disease). There were statistically significant associations between smoking status and BMI at the age of 18 years and the risk of EA. When adjusted for race, age and gender, the AOR for the risk of EA of ever smokers versus never smokers was 2.21 (95% CI: 1.39–3.52), and the AOR of each increase in 1 U of BMI at 18 years of age was 1.07 (95% CI: 1.02–1.13; continuous variable). The relationship with alcohol (ever drinkers versus never drinkers) was borderline, with an adjusted odds ratio (AOR) of 1.43 (95% CI: 0.89–2.92). These variables were adjusted in the subsequent analyses of gene–EA associations. Genotype frequencies in both cases and controls are shown in Table II. Each polymorphism in this control population was consistent with Hardy–Weinberg equilibrium (P . 0.05, chi-square goodness of fit) and genotype frequencies were similar to Caucasians in the SNP500 database (http://snp500cancer.nci.nih.gov). There was no statistically significant association between the XRCC1 Arg399Gln and the XPD Asp312Asn polymorphisms and the risk of EA before or after adjustment for race, age, gender, smoking status and BMI-18 (Table II). However, for the XPD Lys751Gln polymorphism, the presence of at least one Gln allele was associated with a greater risk of EA; the AOR of Gln/- versus Lys/Lys was 1.51 (95% CI: 1.03–2.17; P 5 0.048). We

G.Liu et al.

Table II. Crude and adjusted ORs of XRCC1 and XPD polymorphisms in the risk of EAs

XRCC1 Arg399Glna Arg/Arg Arg/Gln Gln/Gln Gln/XPD Asp312Asna Asp/Asp Asp/Asn Asn/Asn Asn/XPD Lys751Glna Lys/Lys Lys/Gln Gln/Gln Gln/-

Crude

Adjusteda

1.00 1.08 (0.73–1.59) 1.41 (0.81–2.43) 1.15 (0.80–1.66)

1.00 1.12 (0.75–1.67) 1.48 (0.85–2.59) 1.19 (0.83–1.76)

1.00 1.10 (0.76–1.61) 0.96 (0.50–1.86) 1.08 (0.75–1.56)

1.00 1.14 (0.79–1.69) 0.94 (0.49–1.95) 1.10 (0.78–1.63)

1.00 1.43 (0.97–2.11) 1.68 (0.91–3.11) 1.47 (1.01–2.14)

1.00 1.47 (0.99–2.20) 1.73 (0.94–3.15) 1.51 (1.03–2.17)

a

Adjusted for race, age, gender, smoking status, alcohol status and BMI at 18 years. The results were expressed as ORs and 95% CIs.

Table III. Combined XRCC1 Arg399Gln and XPD Lys751Gln polymorphisms in the risk of EAs Case (%)

Zero-variant alleles One-variant alleles Two-variant alleles Three- or fourvariant alleles

23 (13%) 65 (19%) 1.00 1.00 66 (36%) 129 (38%) 1.45 (0.82–2.53) 1.49 (0.88–2.59) 61 (33%) 105 (31%) 1.64 (0.93–2.91) 1.69 (0.98–2.96)

a

Control (%)

Crude

Adjusteda

No. of variant alleles

33 (18%) 37 (11%) 2.52 (1.29–4.91) 2.58 (1.31–5.06)

Adjusted for race, age, gender, smoking status, alcohol status and BMI at 18 years. The results were expressed as ORs and 95% CIs.

1256

phism (D# 5 0.65), whereas the D# for XPD Lys751Gln and XRCC1 Arg399Gln is only 0.13. The simple addition of variant alleles of two linked polymorphisms with a third but unlinked polymorphism potentially poses a problem with data interpretation. Further, we always intended to evaluate the joint effects of the XPD Lys751Gln and XRCC1 Arg399Gln polymorphisms from the start of the study, given recently published data (9,29), but neither study found an important association with the XPD Asp312Asn polymorphism. Nonetheless, as part of an exploratory analysis, when all three polymorphisms were placed in a similar model, the results were not statistically significant, probably the result of the XPD Asp312Asn polymorphism diluting the risk effects of the other two polymorphisms (data not shown). We also created a weighted model where the variant allele of the XPD Asp312Asn polymorphism contributes between 0 and 0.9 of the weight of the variant allele of the other two polymorphisms, and found that the greatest association was detected when the XPD Asp312Asn polymorphism had a weight of zero (data not shown). We also re-analyzed the data using a recessive model (where we grouped XPD 312 Asp/ Asp þ Asp/Asn) and found no change in results. Finally, we also analyzed the XPD 312–751 haplotype (using the HAPPY SAS macro; www.hsph.harvard.edu/faculty/kraft/softetc/HAPPY.pdf) and found that the association between XPD Lys751Gln and EA risk was substantially greater than the association by XPD haplotype. Discussion Despite a significantly increasing incidence rate, EA is still a rare cancer. To our knowledge, this is the largest case–control study evaluating the role of XPD and XRCC1 polymorphisms and EA risk. In our North American mixed Caucasian population, we found that the variant allele of the XPD Lys751Gln polymorphism conferred an increased risk of EA. Our results for all three polymorphisms (XPD Asp312Asn, XPD Lys751Gln and XRCC1 Arg399Gln) are in concordance with results from the population-based case–control of Ye et al. (29) performed in a Swedish population. Furthermore, we extend this knowledge to suggest a combined NER and BER pathway role in EA development through our joint effects analysis. Strengths of this study include a high recruitment rate, high rate with complete information, detailed uniform methods used to minimize misclassification of tumors, careful ascertainment of potential confounding factors and relatively large sample size. In addition, our cases had similar age, gender, race and stage distributions as the underlying population of esophageal cancer patients in Massachusetts (data from Massachusetts Cancer Registry, 2001). Our controls had primary smoking exposures similar to the general Massachusetts adult population over the age of 45 years (data from Massachusetts Tobacco Survey, 1999) and frequency matching for age, gender and race distributions. All three polymorphisms were in Hardy–Weinberg equilibrium. We also utilized published data to determine which polymorphisms to evaluate, rather than attempting to genotype a myriad of DNA polymorphisms. This helped reduce the number of multiple comparisons. Finally, similar to previous studies, we found significant relationships or trends for smoking, BMI and alcohol as risk factors of EA. A wide diversity of DNA damage could be induced by normal metabolic processes in endogenous origin or by environmental carcinogens in exogenous sources. If not repaired, such damage can be converted into gene mutations and genomic instability, which in turn result in cellular malignant transformation (35). Therefore, the normal function of DNA repair enzymes is of great importance, and lower DNA repair capacities (or their associated polymorphic variants) have been associated with higher risk of lung (10–12), breast (13,14), skin (15), prostate (16) and esophageal cancers (36,37,38). In EA, DNA repair mechanisms are critical in areas exposed to constant physical insults. Reflux of stomach contents is associated with esophageal inflammation (esophagitis). Chronic exposure to this reflux leads to a host response, and the development of Barrett’s esophagus or intestinal metaplasia (4). Continued exposure to such physical insults can lead to dysplasia and eventually develop into frank EA (7). As the current theories of EA development involve

Downloaded from http://carcin.oxfordjournals.org/ by guest on August 18, 2015

found no significant interactions between genotype and BMI at the age of 18 years or smoking status. For the joint effects of the XRCC1 Arg399Gln and XPD Lys751Gln polymorphisms (Table III), we dichotomized the population into four genotype groups based on the number of variant alleles of the two polymorphisms: the reference group (no variant alleles of either gene, i.e. individuals with wild genotype for both the polymorphisms); with one variant allele; with two variant alleles and with three or four variant alleles. In the crude analysis, there was a significant trend for a greater number of variant alleles and cases compared with controls (P 5 0.008, Mantel–Haenszel trend test). There was a statistically significant higher risk for three or four variant alleles when compared with the reference: AOR, 2.58 (95% CI: 1.31–5.06; Table III). Similar associations were found between never smokers and ever smokers, and between subjects with different BMI at 18 years. The AORs were 2.60 (95% CI: 0.50–13.66) for never smokers and 2.56 (95% CI: 1.14–5.12) for ever smokers. The AORs were 3.22 (95% CI: 1.02–10.04) for BMI at 18 years ,22 (the median BMI for all study participants) and 2.38 (95% CI: 0.95–5.82) for subjects with BMI at 18 years .22, respectively. Stratified analyses were not performed for race and alcohol use variables since Caucasians and ever drinkers each represented over 90% of all study participants. We also considered a differential role between individuals who were heterozygous for both polymorphisms and individuals who were wild-type for one polymorphism while being homozygous variant for the other, since both of these scenarios would lead to the same classification of having Ôtwo-variant allelesÕ; results in these different subgroups were similar, allowing us to group all these individuals into a single twovariant allele category. We had concerns about incorporating the XPD Asp312Asn polymorphism into the joint effects analysis. The XPD Asp312Asn is in moderate linkage disequilibrium to the XPD Lys751Gln polymor-

XRCC1, XPD polymorphism and esophageal cancer

Acknowledgements We thank Peggy Suen, Barbara Bean, Andrea Solomon, Andrea Shafer, Richard Rivera-Massa, Ian James and William Puricelli and the generous sup-

port of the staff at the Massachusetts General Hospital Cancer Center. This research is supported by the National Institutes of Health Grants CA109193, CA110822, CA74386, ES/CA06409, Doris Duke Charitable Foundation and the Kenneth Jackson Memorial Fund. Conflict of Interest Statement: None declared.

References 1. Ilson,D.H. (2003) Oesophageal cancer: new developments in systemic therapy. Cancer Treat. Rev., 29, 525–532. 2. Devesa,S.S., Blot,W.J. et al. (1988) Changing patterns in the incidence of esophageal and gastric carcinoma in the United States. Cancer, 83, 2049– 2053. 3. Younes,M. et al. (2002) Incidence and survival trends of esophageal carcinoma in the United States: racial and gender differences by histological type. Scand. J. Gastroenterol., 37, 1359–1365. 4. Katzka,D.A. et al. (2000) Gastroesophageal reflux disease and Barrett’s esophagus. Med. Clin. North Am., 84, 1137–1161. 5. Cameron,A.J. (1997) Epidemiology of columnar-lined esophagus and adenocarcinoma. Gastroenterol. Clin. North Am., 26, 487–494. 6. Sharma,P. et al. (1997) Dysplasia in short-segment Barrett’s esophagus: a prospective 3-year follow-up. Am. J. Gastroenterol., 92, 2012–2016. 7. Drewitz,D.J., Sampliner,R.E. et al. (1997) The incidence of adenocarcinoma in Barrett’s esophagus: a prospective study of 170 patients followed 4.8 years. Am. J. Gastroenterol., 92, 212–215. 8. Naidoo,R. et al. (1999) DNA repair gene status in oesophageal cancer. Mol. Pathol., 52, 125–130. 9. Casson,A.G. et al. (2005) Polymorphisms in DNA repair genes in the molecular pathogenesis of esophageal (Barrett) adenocarcinoma. Carcinogenesis, 26, 1536–1541. 10. Wei,Q. et al. (2000) Repair of tobacco carcinogen-induced DNA adducts and lung cancer risk: a molecular epidemiologic study. J. Natl Cancer Inst., 92, 1764–1772. 11. Spitz,M.R. et al. (2001) Modulation of nucleotide excision repair capacity by XPD polymorphisms in lung cancer patients. Cancer Res., 61, 1354–1357. 12. Shen,H. et al. (2003) DNA repair capacity and risk of nonsmall cell lung cancer. Int. J. Cancer, 107, 84–88. 13. Shi,Q. et al. (2004) Reduced DNA repair of benzo(a)pyrene diol epoxideinduced adducts and common XPD polymorphisms in breast cancer patients. Carcinogenesis, 13, 1081–1083. 14. Ramos,J.M. et al. (2004) DNA repair and breast carcinoma susceptibility in women. Cancer, 100, 1352–1357. 15. 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. 16. Hu,J.J. et al. (2004) Deficient nucleotide excision repair capacity enhances human prostate cancer risk. Cancer Res., 64, 1197–1201. 17. Thompson,L.H. et al. (2000) XRCC1 keeps DNA from getting stranded. Mutat. Res., 459, 1–18. 18. Benhamou,S. et al. (2002) XPD/XPD gene polymorphisms and cancer risk. Mutagenesis, 17, 463–469. 19. Lunn,R.M. et al. (1999) XRCC1 polymorphisms: effects on aflatoxin B1DNA adducts and glycophorin A variant frequency. Cancer Res., 59, 2557–2561. 20. Abdel-Rahman,S.Z. et al. (2000) The 399Gln polymorphism in the DNA repair gene XRCC1 modulates the genotoxic response induced in human lymphocytes by the tobacco-specific nitrosamine NNK. Cancer Lett., 159, 63–71. 21. Duell,E.J. et al. (2000) Polymorphisms in the DNA repair genes XRCC1 and XPD and biomarkers of DNA damage in human blood mononuclear cells. Carcinogenesis, 21, 965–971. 22. Hu,J.J. et al. (2001) Amino acid substitution variants of APE1 and XRCC1 genes associated with ionizing radiation sensitivity. Carcinogenesis, 22, 917–922. 23. Palli,D. et al. (2001) DNA adduct levels and DNA repair polymorphisms in traffic-exposed workers and a general population sample. Int. J. Cancer, 94, 121–127. 24. Qiao,Y. et al. (2002) Modulation of repair of ultraviolet damage in the hostcell reactivation assay by polymorphic XPC and XPD/ERCC2 genotypes. Carcinogenesis, 23, 295–299. 25. Wang,Y. et al. (2003) From genotype to phenotype: correlating XRCC1 polymorphisms with mutagen sensitivity. DNA Repair, 2, 901–908. 26. Lee,J.M. et al. (2001) Genetic polymorphisms of XRCC1 and risk of the esophageal cancer. Int. J. Cancer, 95, 240–246.

1257

Downloaded from http://carcin.oxfordjournals.org/ by guest on August 18, 2015

chronic physical injury by acid reflux to the esophageal mucosa, there is a greater potential for DNA repair genetic polymorphisms to alter the risk of progression to EA. Thus, even subtle alterations in DNA capacity can lead to substantial differences in disease risk, when the injury occurs continuously over years (36). BER and NER are two distinct excision repair pathways. BER operates on small lesions such as oxidized or reduced bases, fragmented or non-bulky adducts or those produced by methylating agents (39). The NER pathway repairs bulky lesions such as pyrimidine dimers, larger chemical adducts and cross-links (40). Of the genes involved in these two pathways, polymorphisms of XRCC1 and XPD are the most frequently evaluated. The XRCC1 Arg399Gln polymorphism is located within the BRCA1 C-terminus functional domain, with the Gln allele associated with higher levels of DNA adducts (19,23) and glycophorin A variants (19) increased sister chromatid exchange frequencies (20,21), higher sensitivity to ionizing radiation (22) and higher values of chromosomal breaks per cell (25). For the XPD polymorphisms, the variant Gln allele of the Lys751Gln polymorphism or the Asn allele of the Asp312Asn polymorphism is associated with lower DNA repair capacities measured by the host cell reactivation assay, with one study of 360 healthy controls using BPDE-treated plasmids (11) and another study of 102 healthy individuals using UV-irradiated plasmids (24). However, in a small study of 31 women at risk for breast cancer, the Lys/Lys codon 751 XPD genotype was found to be associated with a reduced repair of X-ray-induced DNA damage, and the Asp312Asn polymorphism did not appear to affect DNA repair proficiency (41), which is consistent with our results of the XPD polymorphisms. The joint effects of XPD and XRCC1 polymorphisms do not imply that a direct biologic interaction exists between these two genes, but more likely that these parallel pathways promote carcinogenesis independently and that these effects are at least additive. This study has several limitations. First, the design was a hospitalbased case–control study. Second, though we incorporated the major confounding variables, we did not adjust for diet or occupational exposure data in the logistic regression models due to missing data. Nonetheless, even with a relatively large sample size for main effects, we did not have adequate sample sizes for proper evaluation of gene– environment interactions or entire pathway analyses; thus, we were not able to confirm the preliminary results from the Swedish study, which found preliminary interactions between the XPD Lys751Gln polymorphism and either BMI or GERD symptoms. Data must be interpreted cautiously when there are inconsistent results across studies. Our results are quite different from the study of Casson et al. (9) from Halifax, Canada, which found a protective effect of the XRCC1 Arg399Gln and XPD Lys751Gln variant alleles. Similar to their study, we used a strict definition of EA, and we also adjusted for age, gender, smoking and alcohol in logistic regression models. The ever-smoking and ever-drinking proportions were similar in both studies, and both used frequency-matched healthy GERD-free controls from a Northeastern North American area. The only major difference was the fact that the present study had three–four times the number of study participants as the Halifax study. In contrast, the present study results strongly support the Swedish population-based case–control study (29), and combined with the independent strengths of both studies would suggest a true association with XPD Lys751Gln. Nonetheless, when compared with studies in more common cancer sites (e.g. breast, lung or colorectal), our study consisted of modest sample sizes and the possibility of spurious results still exists. We also found that the combined effects of XPD Lys751Gln and XRCC1 Arg399Gln polymorphisms are even greater than those of the XPD Lys751Gln alone. Future larger studies should focus on BER and NER pathway analyses, including analyses of other NER–BER polymorphisms and possibly tagging single nucleotide polymorphisms, as well as gene–environment interactions.

G.Liu et al.

27. Abdel-Rahman,S.Z. et al. (2000) Inheritance of the 194Trp and the 399Gln variant alleles of the DNA repair gene XRCC1 are associated with increased risk of early-onset colorectal carcinoma in Egypt. Cancer Lett., 159, 79–86. 28. Shen,H. et al. (2000) Polymorphisms of the DNA repair gene XRCC1 and risk of gastric cancer in a Chinese population. Int. J. Cancer, 88, 601–606. 29. Ye,W. et al. (2006) The XPD 751Gln allele is associated with an increased risk for esophageal adenocarcinoma: a population-based case-control study in Sweden. Carcinogenesis, 27, 1835–1841. 30. Miller,D.P. et al. (2002) Combinations of the variant genotypes of GSTP1, GSTM1, and p53 are associated with an increased lung cancer risk. Cancer Res., 62, 2819–2823. 31. Zhou,W. et al. (2002) Gene-environment interaction for the XPD polymorphisms and cumulative cigarette smoking exposure in lung cancer. Cancer Res., 62, 1377–1381. 32. Zhou,W. et al. (2003) Polymorphisms in the DNA repair genes XRCC1 and XPD, smoking, and lung cancer risk. Cancer Epidemiol. Biomarkers Prev., 12, 359–365. 33. de las Penas,R. et al. (2006) Polymorphisms in DNA repair genes modulate survival in cisplatin/gemcitabine-treated non-small-cell lung cancer patients. Ann. Oncol., 17, 668–675. 34. Hosmer,D. et al. (1989) Applied Logistic Regression. John Wiley and Sons, New York.

35. Xing,D. et al. (2002) Polymorphisms of DNA repair genes XRCC1 and XPD and their associations with risk of esophageal squamous cell carcinoma in a Chinese population. Int. J. Cancer, 100, 600–605. 36. Shao,L. et al. (2006) Deficiency of cell cycle checkpoints and DNA repair system predispose individuals to esophageal cancer. Mutat. Res., 602, 143–150. 37. Cai,L. et al. (2006) Dietary selenium intake, aldehyde dehydrogenase-2 and X-ray repair cross-complementing 1 genetic polymorphisms, and the risk of esophageal squamous cell carcinoma. Cancer, 106, 2345–2354. 38. Hao,B. et al. (2004) Identification of genetic variants in base excision repair pathway and their associations with risk of esophageal squamous cell carcinoma. Cancer Res., 64, 4378–4384. 39. Goode,E.L. et al. (2002) Polymorphisms in DNA repair genes and associations with cancer risk. Cancer Epidemiol. Biomarkers Prev., 11, 1513–1530. 40. Squire,J.A. et al. (1998) Genetic basis of cancer. In Hill,R.P. and Tannock,I.F. (eds.) The Basic Science of Oncology. McGraw-Hill Professional, NY, pp. 48–78. 41. Lunn,R.M. et al. (2000) XPD polymorphisms: effects on DNA repair proficiency. Carcinogenesis, 21, 551–555. Received December 1, 2006; revised December 31, 2006; accepted January 16, 2007

Downloaded from http://carcin.oxfordjournals.org/ by guest on August 18, 2015

1258