Cancer Prevention

Nucleotide Excision Repair Pathway Genes and Oral Premalignant Lesions Yunfei Wang,1Margaret R. Spitz,1J. Jack Lee,2 Maosheng Huang,1Scott M. Lippman,3 and Xifeng Wu1

Abstract Purpose: Oral premalignant lesions (OPL) are associated with tobacco exposure and an increase in risk of oral cancer. Nucleotide excision repair (NER) is one of the major DNA repair pathways involved in the removal of tobacco carcinogen adducts. Polymorphisms in NER genes may cause variations in DNA repair capacity and increase susceptibility to both premalignant lesions and cancer. Experimental Design: In this case-control study of144 OPL patients and 288 controls, we gen- otyped 11polymorphisms in 8 major NER genes, including XPA [A23G at 5¶ untranslated region (UTR)], XPD (Asp312Asn, Lys751Gln), XPC (Ala499Val, Lys939Gln), XPG (His1104Asp), XPF (Pro662Ser), ERCC6 (Met1097Val, Arg1230Pro) Rad23B (Ala249Val), and CCNH (Val270Ala). Results: We found significant or borderline-significant associations between OPL risk and the polymorphisms XPA (A23G), XPD (Lys751Gln ), XPC (Ala499Val), Rad23B (Ala249Val), and XPD (Asp312Asn), with adjusted odds ratios (ORs) of 1.97 [95% confidence interval (95% CI), 1.27-3.06], 1.60 (95% CI, 1.02-2.51), 0.63 (95% CI, 0.40-1.00), 0.67 (95% CI, 0.41-1.07), and 1.42 (95% CI, 0.90-2.23), respectively. When further stratified analyses were done, the decreased risk conferred by the XPC (Ala499Val) variant allele was more evident in older individuals (OR, 0.50; 95% CI, 0.24-1.03), in women (OR, 0.46; 95% CI, 0.21-1.01), in ever smokers (OR, 0.59; 95% CI, 0.33-1.05), and in never drinkers (OR, 0.42; 95% CI, 0.18-1.00). Finally, we found joint effects between these NER gene variants and smoking status. For example, when never smokers with the XPA 23A genotypes were used as the refer- ence group, the ORs for never smokers with the XPA 23G genotype, smokers with the 23A genotype, and smokers with 23G genotypes were 2.19 (1.07-4.46), 2.64 (1.42-4.89), and 5.04 (2.62-9.69), respectively. Gene-gene and gene-smoking interaction for OPLs risk were also confirmed by multifactor dimensionality reduction (MDR) analysis in our study. MDR analysis revealed that a model containing ever smoking, XPA (A23G), XPC (Ala499Val), and XPD (Asp312Asn) was the best model to predict OPL risk with maximum average cross- validation consistency and minimum prediction error (P < 0.001). Conclusion: Our results suggest that polymorphisms in NER genes may contribute to genetic susceptibility to OPLs and may therefore contribute to the development of oral cancer.

Oral white or red patches, known as leukoplakia and to become dysplasia or carcinoma (1, 5, 6). The factors that erythroplakia, are the major forms of oral premalignant lesions most greatly increase the risk of developing OPLs are the con- (OPL; ref. 1). Several clinical and epidemiologic studies have sumption of alcohol and the chewing or smoking of tobacco shown that the malignant transformation rates of oral leu- (7–10); nevertheless, the inherited genetic susceptibility may koplakia range from 8.9% to 17.5% (2–4). The erythroplakia, also play an important role. Individuals with weakened DNA although not as common as leukoplakia, are much more likely repair capacity may have more genetic abnormalities compared with those with normal DNA repair capacity under similar carcinogenetic exposure conditions because DNA damages accumulate faster in individuals with suboptimal DNA repair Authors’Affiliations: Departments of 1Epidemiology, 2Biostatistics, and capacity (11). 3Thoracic/Head and Neck Oncology, The University of Texas M. D. Anderson DNA repair systems play critical roles in protecting against Cancer Center, Houston,Texas mutations and are essential for maintaining the integrity of the Received 8/2/06; revised 1/26/07; accepted 3/22/07. genome. Nucleotide excision repair (NER) is one of the major Grant support: National Cancer Institute grants CA106451and CA097007. The costs of publication of this article were defrayed in part by the payment of page DNA repair pathways and is mainly responsible for the removal charges. This article must therefore be hereby marked advertisement in accordance of bulky DNA adducts induced by chemical carcinogens, such as with 18 U.S.C. Section 1734 solely to indicate this fact. the polycyclic aromatic hydrocarbons (PAH) found in tobacco Requests for reprints: Xifeng Wu, Department of Epidemiology, Box 189, The smoke, car exhaust, coal burning, etc. (12, 13). NER includes University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Boulevard, two subpathways, global-genome repair and transcription- Houston, TX77030. Phone: 713-745-2485; Fax: 713-792-0807; E-mail: [email protected]. coupled repair. XPC-hHR23B and XPA-RPA complexes recog- F 2007 American Association for Cancer Research. nize the initial DNA damage in global-genome repair, whereas doi:10.1158/1078-0432.CCR-06-1911 the ERCC6 (CSB) proteins recognize DNA lesions in the

www.aacrjournals.org 3753 Clin Cancer Res 2007;13(12) June 15, 2007 Downloaded from clincancerres.aacrjournals.org on September 28, 2021. © 2007 American Association for Cancer Research. Cancer Prevention transcribed strand of an active gene in transcription-coupled efficient for large-scale molecular epidemiologic studies (31). The repair (14–19). Once the DNA damage is recognized in either controls were matched to the cases by age (F5), sex, and ethnicity subpathway, the DNA helix around the lesion is unwound by (Caucasian, African-American, Hispanics, and other). Questionnaire the XPB and XPD helicases of the multi-subunit transcription data were obtained through personal interviews with the controls. This study was approved by the Institutional Review Boards (IRB) of the factor THIIH, and the two strands are separated to form a ¶ ¶ M.D. Anderson Cancer Center and Kelsey-Seybold Clinic. All study preincision structure (20, 21). Then, 3 and 5 incisions are participants signed an informed consent prior participation. completed by ERCC5/XPG and ERCC1/XPF endonucleases, Genotyping of NER genes. Genomic DNA was isolated from respectively, and finally, the DNA polymerase and ligase peripheral blood samples using proteinase K digestion, followed by complete the repair by filling the gaps (12, 22–24). The isopropanol extraction and ethanol precipitation. PCR-RFLP was used functional mutations in NER genes may cause NER abnormal- for the genotyping of XPA (A23G) and XPD (Asp312Asn) as described ities and reduce the NER capacity and lead to altered genetic previously (25, 28). TaqMan method was used to detect the XPD susceptibility to cancer. (Lys751Gln), XPC (Ala499Val, Lys939Gln), XPG (His1104Asp), XPF Polymorphisms in DNA repair genes have been extensively (Pro662Ser), ERCC6 (Met1097Val, Arg1230Pro), Rad23B (Ala249Val), studied for their associations with diseases. Most NER genes are and CCNH (Val270Ala) genotypes. Briefly, sequences of primers and polymorphic, and many studies have reported the associations probes were either obtained from the National Cancer Institute SNP500 cancer database or designed using Primer Express Software between some polymorphisms in NER genes and the risk of (Applied Biosystems). The probes were labeled fluorescently with either tobacco-related cancer, such as lung cancer, head and neck FAM or VIC on the 5¶ end and a nonfluorescent minor groove binder cancer, and breast cancer (25–30). Spitz et al. (27) reported quencher on the 3¶ end (Applied Biosystems). Typical amplification that polymorphisms in the XPD gene might modulate NER mixes and thermal cycling conditions were recommended by the probe capability in lung cancer patients, and individuals with wide- supplier (Applied Biosystems). Water and internal controls were type genotypes exhibited the most proficient DNA repair included in each plate to ensure the accuracy of the genotyping. capacity. A meta-analysis of 9 individual case-control studies Reactions were run on the dual 384-well GeneAmp PCR System 9700, of 3,725 lung cancer cases and 4,152 controls showed that the and the plates were read in a TaqMan 7900HT sequence detection homozygous variant alleles of these two polymorphisms system (Applied Biosystems). The analyzed fluorescence results were exhibited 20% to 30% increased lung cancer risks (28). then auto-called into genotypes using the built-in software of the system. To our knowledge, there have been no epidemiologic Statistical analyses. All statistical analyses were done using the studies on NER genetic polymorphisms and risk of OPLs. Intercooled Stata 8.0 statistical software package (Stata Corporation). In In this study, we applied a pathway-based approach to comparing cases and controls, the m2 test was used to test for differences systematically examine the associations between a panel of in the distribution of gender, ethnicity, smoking status, and NER polymorphisms in major NER genes and risk of OPLs. Based genotypes, whereas the Student’s t test was used to test for differences on published studies, the polymorphisms that we selected in age and pack-years (among ever smokers). Odds ratios (OR) were in our study included XPA [A23G at 5¶ untranslated region calculated as an estimate of the relative risk. Where appropriate, (UTR)], XPD (Asp312Asn, Lys751Gln), XPC (Ala499Val, unconditional multivariate logistic regression was done to control for Lys939Gln), XPG (His1104Asp), XPF (Pro662Ser), ERCC6 potential confounding by age, gender, ethnicity, and smoking status. (Met1097Val, Arg1230Pro), Rad23B (Ala249Val), and The sample size was too small for minority groups for stratified CCNH (Val270Ala). All these single nucleotide polymor- analyses. However, we checked the distribution of genotype frequencies in different ethnic groups and did not find significant differences. We phisms (SNP) have potential functional significance (non- also restricted the analyses in Caucasians and found that the results synonymous SNPs or located in the promoter and/or were similar to that of all ethnicities combined. Therefore, we only untranslated regions) and have been studied in literature. reported overall results while adjusting for ethnicity. Interaction was We also evaluated the joint effects of these polymorphisms tested on the multiplicative scale by entering product terms in the with smoking in modulating the risk of OPLs. multivariable logistic regression models. Multifactor dimensionality reduction (MDR; refs. 32–34) is done to assess interactions among multiple SNPs and/or risk factors. In MDR, a set of n factors is selected, and the n selected factors and their pos- Materials and Methods sible multifactor classes are represented in n dimensional space. The ratio for the number of cases to the number of controls is calculated Study subjects. A total of 144 OPL patients were identified at the within each multifactor class. Each multifactor class in n-dimensional University of Texas M.D. Anderson Cancer Center from 1997 to 2006. space is then labeled as ‘‘high risk’’ or ‘‘low risk’’, based on whether the The inclusion criteria for cases were the presence of histologically case-to-control ratio has exceeded a threshold. Thus, a new variable confirmed OPL and an age of 18 years or older. Patients with acute incorporating information from several risk factors is defined, and intercurrent illnesses or infections were excluded, as were those who cross-validation and permutation tests are used to assess the capability had retinoid or carotenoid therapy within 3 months before study entry. of this new variable to predict outcome risk. The cross-validation A self-administered questionnaire was used to gather epidemiologic consistency is calculated to estimate the percentage of the same data, including recent and prior tobacco and alcohol uses. Baseline combination of risk factors selected as the best model among cross- blood samples were obtained in heparinized tubes for molecular validation data sets. The average prediction error is calculated as an analyses. A total of 288 controls without any history of cancer were average of the prediction errors across each of the cross-validation recruited from a large pool of potential volunteers from the Kelsey- subsets. Seybold clinic, a large multispecialty managed-care organization in the In this study, we also used the MDR to identify and assess putative Houston metropolitan area, from 1999 to 2006. The potential controls risk factors, as well as gene-gene and gene-smoking interactions, for were identified by reviewing short survey forms distributed to patients NER polymorphisms. We used 100-fold cross-validation and 1,000-fold visiting the Kelsey-Seybold clinics. The majority of control subjects are permutation testing, and all statistical tests were at the significance healthy individuals coming to the clinics for annual health checkups. level of 0.05. Cross-validation and permutation testing were done for The study design to recruit controls has been proven to be valid and the same set of subjects.

Clin Cancer Res 2007;13(12) June 15, 2007 3754 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 28, 2021. © 2007 American Association for Cancer Research. Nucleotide Excision Repair Pathway Genes and OPL

Results Table 2. Risk estimates of NER polymorphisms

By study design, the 144 OPL cases and 288 controls were Genotypes Cases, Controls, Adjusted matched by age, gender and ethnicity (Both P = 1.0; Table 1). n (%) n (%) OR* (95% CI) The mean age of controls (59.7 F 11.0 years) was slightly F XPA A23G higher than that of cases (58.3 12.8 years), but the difference AA 11 (8.0) 57 (21.0) Reference was not statistically significant (P = 0.25). As expected, current AG 57 (41.3) 120(44.3) 2.71 (1.29-5.71) smokers were statistically significantly overrepresented among GG 70 (50.7) 94 (34.7) 4.21 (2.00-8.88) the cases (27.1%), as compared with the controls (10.4%; AA + AG 68 (49.3) 177 (65.3) Reference GG 70 (50.7) 94 (34.7) 1.97 (1.27-3.06) P < 0.001). Rad23B Ala249Val In single SNP analysis, the GG genotype for XPA (A23G) was CC 92 (71.9) 172 (61.7) Reference more common in cases (50.7%) than in controls (34.7%). CT 28 (21.9) 93 (33.3) 0.59 (0.35-0.99) Adjusted by age, gender, ethnicity, smoking status, and alcohol TT 8 (6.3) 14 (5.0) 1.22 (0.46-3.25) use, individuals with the GG genotype had a significantly CT + TT 36 (28.1) 107 (38.3) 0.67 (0.41-1.07) XPC Ala 499Val increased OPL risk compared with those with the AA and AG CC 78 (63.4) 153 (55.2) Reference genotypes [OR, 1.97; 95% confidence interval (95% CI), 1.27- CT 40 (32.5) 114 (41.2) 0.62 (0.39-1.00) 3.06]. Individuals with at least one variant allele of XPD TT 5 (4.1) 10 (3.6) 0.69 (0.22-2.18) (Lys751Gln) polymorphism exhibited a significant increase in CT + TT 45 (36.6) 124 (44.8) 0.63 (0.40-0.999) XPC Lys939Gln OPL risks with OR of 1.60 (95% CI, 1.02-2.51). A significant AA 48 (37.5) 103 (36.9) Reference decrease in OPL risks with OR of 0.63 (95% CI, 0.40-0.999) AC 58 (45.3) 128 (45.9) 0.91 (0.56-1.48) was observed in individuals with the variant allele of XPC CC 22 (17.2) 48 (17.2) 1.06 (0.56-2.01) (Ala499Val). The GA + AA genotypes for XPD (Asp312Asn) AC + CC 80 (62.5) 176 (63.1) 0.95 (0.60-1.50) and the CT + TT genotypes for Rad23B (Ala249Val) showed XPD Asp312Asn GG 50 (40.0) 140 (50.4) Reference an increased/reduced OPL risk with ORs of 1.42 (95% CI, GA 59 (47.2) 109 (39.2) 1.47 (0.91-2.36) 0.90-2.23) and 0.67 (95% CI, 0.41-1.07), respectively, but did AA 16 (12.8) 29 (10.4) 1.24 (0.59-2.61) not reach statistical significance (Table 1). No statistically GA + AA 75 (60.0) 138 (49.6) 1.42 (0.90-2.23) significant association was found between other NER SNPs XPD Lys751Gln AA 46 (31.9) 120(42.9) Reference and OPL risk in this study (Table 2). AC 77 (53.5) 132 (47.1) 1.57 (0.99-2.51) Stratified by age, gender, ethnicity, smoking status, and CC 21 (14.6) 28 (10.0) 1.73 (0.86-3.48) drinking status, we found that the reduced risk conferred by the AC + CC 98 (68.1) 160(57.1) 1.60(1.02-2.51) XPC (Ala499Val) variant alleles was more evident in older XPG His1104Asp individuals (age z 61; OR, 0.50; 95% CI, 0.24-1.03), in women GG 83 (60.1) 165 (58.9) Reference GC 45 (32.6) 109 (38.9) 0.71 (0.45-1.14) (OR, 0.46; 95% CI, 0.21-1.01), in ever smokers (OR, 0.59; 95% CC 10(7.3) 6 (2.1) 2.69 (0.91-7.97) CI, 0.33-1.05), and in never drinkers (OR, 0.42; 95% CI, 0.18- GC + CC 55 (39.9) 115 (41.1) 0.83 (0.53-1.30) 0.998; data not shown). There were no obvious effects of age, ERCC6 Met1097Val gender, smoking, or drinking on the association between other AA 82 (65.6) 175 (63.0) Reference AG 36 (28.8) 94 (33.8) 0.78 (0.48-1.27) polymorphisms and OPL risk (data not shown). However, due GG 7 (5.6) 9 (3.2) 1.53 (0.51-4.55) to the borderline confidence intervals and the multiple AG + GG 43 (34.4) 103 (37.0) 0.85 (0.53-1.35) comparison issue and the small sample size of each stratum, ERCC6 Arg1230Pro the results from these stratified analyses need to be interpreted GG 103 (83.7) 221 (79.5) Reference with caution. GC 18 (14.6) 56 (20.1) 0.72 (0.40-1.31) CC 2 (1.6) 1 (0.36) 2.27 (0.19-27.3) GC + CC 20 (16.3) 57 (20.5) 0.76 (0.43-1.36) XPF Pro662Ser TT 107 (98.2) 273 (98.2) Table 1. Distribution of host characteristics in TC 2 (1.8) 5 (1.8) 1.57 (0.20-12.4) cases and controls CC — — TC + CC 2 (1.8) 5 (1.8) 1.57 (0.20-12.4) Variable Cases, n (%) Controls, n (%) P CCNH Val270Ala TT 79 (62.2) 176 (63.1) Reference Gender TC 43 (33.9) 90 (32.3) 1.09 (0.68-1.76) Male 81 (56.3) 162 (56.3) CC 5 (3.9) 13 (4.7) 0.86 (0.28-2.62) Female 63 (43.7) 126 (43.7) 1.00 TC + CC 48 (37.8) 103 (36.9) 1.06 (0.67-1.69) Ethnicity Caucasian 134 (93.1) 268 (93.1) * Mexican-American 6 (4.2) 12 (4.2) Adjusted by age, gender, ethnicity, smoking status, and alcohol African-American 4 (2.8) 8 (2.8) 1.00 use. Smoking status Never 47 (32.6) 146 (50.7) Former 58 (40.3) 112 (38.9) Current 39 (27.1) 30 (10.4) <0.001 Due to the fact that XPA (A23G), XPD (Lys751Gln), and Drinking status XPC (Ala499Val) polymorphisms showed significant associa- Never 55 (38.7) 95 (34.8) Ever 87 (61.3) 178 (65.2) 0.429 tions with OPL risk in the single SNP analysis above, we Age, mean (FSD) 58.3 (12.8) 59.7 (11.0) 0.246 also evaluated the joint effects of the three SNPs (Table 3). The reference group included individuals with the low-risk

www.aacrjournals.org 3755 Clin Cancer Res 2007;13(12) June 15, 2007 Downloaded from clincancerres.aacrjournals.org on September 28, 2021. © 2007 American Association for Cancer Research. Cancer Prevention

Table 3. Joint effects among XPA A23G, XPD Lys751Gln, and XPC Ala499Val genes

XPA A23G XPD Lys751Gln XPC Ala499Val Multivariate OR* (95% CI) P AA + AG AA CT + TT Reference AA + AG AC + CC CT + TT 2.59 (0.77-8.77) 0.126 AA + AG AA CC 2.27 (0.64-8.12) 0.207 GG AA CT + TT 4.61 (1.16-18.3) 0.030 AA + AG AC + CC CC 3.91 (1.17-13.0) 0.027 GG AA CC 4.27 (1.17-15.6) 0.028 GG AC + CC CT + TT 4.66 (1.25-17.4) 0.022c GG AC + CC CC 8.99 (2.66-30.4) <0.001

*Adjusted by age, gender, ethnicity, smoking status, and alcohol use. cRemains significant after Bonferroni adjustment for multiple comparisons. Family-wise significance is set as 0.05. Thus, significance level for each individual test is 0.0071. genotypes at all three loci. Of these individuals, those with one, The results of the MDR analysis are presented in Table 6. two, or three high-risk genotypes all exhibited increased OPL The model including only smoking was the best single factor risks, with the ORs ranging from 2.27 (95% CI, 0.64-8.12) to model for predicting OPL risk, with average cross-validation 8.99 (95% CI, 2.66-30.4; Table 3). Finally, we found joint consistency of 100% and an average prediction error of effects between NER gene variants and smoking status. 41.0%. Among two-factor models, the combination of Combining heterozygote and homozygote variant genotypes smoking and XPC Ala499Val proved to be the strongest, with and using a reference group that included never smokers with an average cross-validation consistency of 52.7% and an the XPA 23A genotypes, the ORs for never smokers with the average prediction error of 52.9%. When three factors were XPA 23G genotype, smokers with 23A genotype, and smokers considered, the combination of smoking, XPC Ala499Val, and with 23G genotypes were found to be 2.19 (1.07-4.46), 2.64 XPA A23G was the strongest model, with a cross-validation (1.42-4.89), and 5.04 (2.62-9.69), respectively. A similar joint consistency of 100% and an average prediction error of effect was observed between smoking status and the XPD 36.0%. A model containing smoking, XPD Asp312Asn, XPA (Lys751Gln), Rad23B Ala249Val, and XPC Ala499Val adverse A23G, and XPC Ala499Val was the strongest among all four- alleles (Table 4). A gene-drinking interaction was also analyzed, factor models, with a cross validation consistency of 99.9% and no significant results were found (data not shown). We and an average prediction error rate of 34.8%. The average also did haplotype analysis for SNPs of XPD, ERCC6 and XPC cross-validation consistency decreased in five-factor models. (Table 5). Overall, there is no significant association between Among all models, the four-factor model had the highest OPL risk and haplotypes of XPD, ERCC6, and XPC. cross-validation consistency and the lowest prediction error (P < 0.001; Table 6).

Discussion Table 4. Joint effect between polymorphism and smoking status Polymorphisms in NER genes may reduce the efficiency of DNA repair (35). The XPA protein binds to damaged DNA Genotypes Smoking Adjusted P status OR (95% CI)* and maintains an intricate network of contacts with core XPA A23G AA + AG Never Reference Table 5. Haplotype analyses of OPL risk GG Never 2.19 (1.07-4.46) 0.031 AA + AG Ever 2.64 (1.42-4.89) 0.002 Haplotype Controls Cases Adjusted OR* (95% CI) GG Ever 5.04 (2.62-9.69) <0.001 RAD23B Ala249Val XPD Asp312Asn-Lys751Gln CT + TT Never Reference W-W 331 135 Reference CC Never 1.71 (0.78-3.75) 0.180 W-M 46 27 1.53 (0.90-2.60) CT + TT Ever 2.59 (1.12-5.97) 0.025 M-W 25 11 1.07 (0.49-2.34) CC Ever 3.79 (1.77-8.14) 0.001 M-M 136 71 1.40 (0.96-2.02) XPC Ala499Val ERCC6 1097-1230 CT + TT Never Reference W-W 382 176 Reference CC Never 1.46 (0.67-3.17) 0.339 W-M 56 24 0.91 (0.54-1.55) CT + TT Ever 2.22 (1.01-4.86) 0.046 M-W 110 52 0.98 (0.66-1.46) CC Ever 3.63 (1.70-7.77) 0.001 XPC Lys939Gln-Ala499Val XPD Lys751Gln W-W 199 98 Reference AA Never Reference W-M 132 50 0.64 (0.41-1.00) AC + CC Never 2.04 (0.97-4.30) 0.061 M-W 222 100 0.88 (0.62-1.25) AA Ever 3.23 (1.50-6.95) 0.003 M-M 1 0 AC + CC Ever 4.55 (2.21-9.41) <0.001

*Adjusted by age, gender, ethnicity, smoking status, and alcohol *Adjusted by age, gender, ethnicity, and alcohol use. use.

Clin Cancer Res 2007;13(12) June 15, 2007 3756 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 28, 2021. © 2007 American Association for Cancer Research. Nucleotide Excision Repair Pathway Genes and OPL

Table 6. Summary of results for OPL from MDR analysis

Number Best candidate model Cross-validation Average prediction Permutation of variables consistency (%) error (%) test P value 1 Smoking 100 41.0 0.018 2 Smoking, XPC Ala499Val 52.7 52.9 0.738 3 Smoking, XPC Ala499Val, XPA A23G 100 36.0 0.001 4 Smoking, XPC Ala499Val, XPA A23G, 99.9 34.8 <0.001 XPD Asp312Asn 5 Smoking, XPC Ala499Val, XPA A23G, 99.7 38.8 0.006 XPD Asp312Asn, RAD23B Ala249Val

NOTE: Adjusted by age, gender, ethnicity, smoking status, and alcohol use.

repair factors. It plays a key role in accurately positioning the genotypes exhibit a higher OPL risk than other groups. repair machinery around the DNA lesion (36). Previously, we Cigarette smoke contains multiple carcinogenic substrates, have reported that the XPA (A23G) G allele was associated such as polycyclic aromatic hydrocarbons, and may induce with a reduced lung cancer risk for Caucasians (OR, 0.69; DNA damage by covalent binding or oxidation (45, 46). NER is 95% CI, 0.53-0.90), Mexican-Americans (OR, 0.32; 95% CI, the major pathway that repairs tobacco carcinogen-induced 0.12-0.83), and African-Americans (OR, 0.45; 95% CI, bulky DNA adducts. Spitz et al. (27) reported a lower capacity 0.16-1.22; ref. 25). In our current report, however, we found of DNA repair in lung cancer patients when compared with that the homozygous G allele was associated with a signifi- healthy controls, and this risk may be modulated by smoking cantly increased risk for OPL. The reason for this discrepancy is status. OPL is a tobacco-related premalignant lesion, and it is unknown. It may be due to the differences in the etiology of biologically plausible that polymorphisms of NER genes lung cancer and OPL. predispose an individual to OPLs. The XPC gene encodes a 940-amino-acid protein that is The association between NER gene polymorphisms and involved in DNA damage recognition (37). The XPC protein is OPL risk, the gene-gene interaction, and gene-smoking inter- tightly combined with Rad23 and plays an early role in NER action were explored by MDR analysis in our study. This by initially detecting the DNA damage (38, 39). The accurate algorithm may allow rapid identification of potential genetic function of Rad23 in NER is not fully understood. But and environmental interactions when analyzing a large number recently, many studies indicated that the complex of HR23B of variables. (Rad23 homologue) and XPC is involved in the damage This study has several limitations. As a hospital-based case- recognition step of NER, and that the primary function of the control study, this study may be subject to potential selection Rad23 protein is to stabilize the XPC protein (40). In our bias. In addition, recall bias cannot be ruled out. However, as study, the variant T alleles for XPC (Ala499Val) polymor- we tested a genotype-driven hypothesis rather than an phisms were associated with a 0.63-fold decrease in OPL risk, environment-driven hypothesis, selection bias and recall bias which is consistent with the results reported recently. Huang are of less concern. Furthermore, data collected by self- et al. (41) reported that individuals with XPC 499Val exhibited administered questionnaire and in-person interview may a significantly reduced advanced colorectal adenoma risk in produce differential bias in risk assessment. However, in this their case-control study with 772 cases and 777 controls, and study, the influence should be minimal because we only used this finding was supported by a haplotype analysis. Similar demographic information and smoking and drinking status results of the variant T alleles for XPC (Ala499Val) were also data collected by questionnaire. Subjects should be able to reported by Zhu et al. and Zhou et al. (42, 43). The variant T report consistent information in both self-administered ques- allele of the Rad23B (Ala249Val) polymorphism was associ- tionnaire and in-person interview in terms of demographic ated with a borderline significant decrease in OPL risk in our characteristics (age, gender, ethnicity) and smoking and study. However, Shen et al. (44) showed a 2-fold increased drinking status. risk in lung cancer patients with RAD23 Val allele in Chinese In conclusion, the results of this study suggest that after population, and no association was found for this polymor- adjusting for important OPL risk factors, such as smoking and phism in a study of advance colorectal adenoma (41). The alcohol drinking, deficiency in NER pathway predisposes heterogeneity of these results may be attributed to different susceptible individuals to increased OPL risk, supporting the cancer site, race, and sample size. Our results showed a joint hypothesis that the inherited genetic susceptibility also plays an effect among NER polymorphisms and a significant trend for important role in OPL etiology. Therefore, the high progression an increase in OPL risk in subjects with increasing numbers of rate of OPL to oral cancer may also be partly explained by the risk alleles. This result is in agreement with our hypothesis that genetic predisposition factors, such as genetic alterations in these polymorphisms coordinate in the DNA repair pathway, NER pathway. and that the joint effects of their genotypes yield higher risk Although our results suggest that OPL risk is associated with for OPL. variant genotypes of NER genes, these findings are subject to Our results also showed a gene-smoking interaction between scrutiny due to limited power and sample size. Larger studies XPA, XPC, and Rad23 genotypes and smoking status in OPL with a priori hypotheses for these covariates are certainly (Table 4). It suggested that smokers with variant homozygous warranted for validation.

www.aacrjournals.org 3757 Clin Cancer Res 2007;13(12) June 15, 2007 Downloaded from clincancerres.aacrjournals.org on September 28, 2021. © 2007 American Association for Cancer Research. Cancer Prevention

References 1. Neville BW, Day TA. Oral cancer and precancerous 19. Kamiuchi S, Saijo M, Citterio E, de Jager M, 33. Hahn LW, Ritchie MD, Moore JH. Multifactor lesions. CA Cancer J Clin 2002;52:195 ^ 215. Hoeijmakers JH, Tanaka K. Translocation of Cockayne dimensionality reduction software for detecting 2. Bouquot JE,Whitaker SB. Oral leukoplakia-rationale syndrome group A protein to the nuclear matrix: pos- gene-gene and gene-environment interactions. Bioin- for diagnosis and prognosis of its clinical subtypes or sible relevance to transcription-coupled DNA repair. formatics 2003;19:376 ^ 82. ‘‘phases’’. Quintessence Int 1994;25:133 ^ 40. Proc Natl Acad Sci U S A 2002;99:201 ^6. 34. Moore JH. Computational analysis of gene-gene 3. Silverman S, Jr., Gorsky M, Lozada F. Oral leukopla- 20. Schaeffer L, Roy R, Humbert S, et al. DNA repair interactions using multifactor dimensionality reduc- kia and malignant transformation. A follow-up study of helicase: a component of BTF2 (TFIIH) basic tran- tion. Expert Rev Mol Diagn 2004;4:795 ^ 803. 257 patients. Cancer 1984;53:563 ^ 8. scription factor. Science 1993;260:58 ^ 63. 35. XiT,JonesIM,MohrenweiserHW.Manyaminoacid 4. Lind PO. Malignant transformation in oral leukoplakia. 21. Schaeffer L, Moncollin V, Roy R, et al. The ERCC2/ substitution variants identified in DNA repair genes Scand J Dent Res 1987;95:449 ^ 55. DNA repair protein is associated with the class II during human population screenings are predicted to 5. Shafer WG, Waldron CA. Erythroplakia of the oral BTF2/TFIIH transcription factor. EMBO J 1994;13: impact proteinfunction. Genomics 2004;83:970^9. cavity. Cancer 1975;36:1021 ^ 8. 2388 ^ 92. 36. Park CH, Sancar A. Formation of a ternary complex 6. Mashberg A, Samit A. Early diagnosis of asymptom- 22. Habraken Y, Sung P, Prakash L, Prakash S. A con- byhumanXPA,ERCC1,andERCC4(XPF)excisionre- atic oral and oropharyngeal squamous cancers. CA served 5¶ to 3¶ exonuclease activity in the yeast and pair proteins. Proc Natl Acad Sci U S A 1994;91: Cancer J Clin 1995;45:328^ 51. human nucleotide excision repair proteins RAD2 and 5017^21. 7. Hashibe M, Sankaranarayanan R, Thomas G, et al. XPG. J Biol Chem 1994;269:31342 ^ 5. 37. Sugasawa K, Ng JM, Masutani C, et al. Xeroderma Body mass index, tobacco chewing, alcohol drinking 23. O’Donovan A, Davies AA, Moggs JG, West SC, pigmentosum group C protein complex is the initiator and the risk of oral submucous fibrosis in Kerala, India. Wood RD. XPG endonuclease makes the 3¶ incision of global genome nucleotide excision repair. Mol Cell Cancer Causes Control 2002;13:55 ^ 64. in human DNA nucleotide excision repair. Nature 1998;2:223 ^ 32. 8. Campisi G, MargiottaV. Oral mucosal lesions and risk 1994;371:432 ^ 5. 38. Rybicki BA, Conti DV, Moreira A, Cicek M, Casey habits among men in an Italian study population. 24. Sijbers AM, de Laat WL, Ariza RR, et al. Xeroderma G, Witte JS. DNA repair gene XRCC1 and XPD poly- J Oral Pathol Med 2001;30:22 ^8. pigmentosum group F caused by a defect in a struc- morphisms and risk of prostate cancer. Cancer Epide- 9. Silverman S, Jr. Prevention, early detection and diag- ture-specific DNA repair endonuclease. Cell 1996;86: miol Biomarkers Prev 2004;13:23 ^ 9. nosisoforalcancer.DermatolClin1987;5:675^80. 811^22. 39. SugasawaK,NgJM,MasutaniC,etal.Twohuman 10. Silverman S, Jr. Precancerous lesions and oral can- 25. Wu X, Zhao H,Wei Q, et al. XPA polymorphism as- homologs of Rad23 are functionally interchangeable cer in the elderly. Clin Geriatr Med 1992;8:529^ 41. sociated with reduced lung cancer risk and a modulat- in complex formation and stimulation of XPC repair 11. Wu X, Lippman SM, Lee JJ, et al. Chromosome in- ing effect on nucleotide excision repair capacity. activity. Mol Cell Biol 1997;17:6924 ^ 31. stability in lymphocytes: a potential indicator of pre- Carcinogenesis 2003;24:505 ^ 9. 40. Xie Z, Liu S, ZhangY,Wang Z. Roles of Rad23 pro- disposition to oral premalignant lesions. Cancer Res 26. Park JY, Park SH, Choi JE, et al. Polymorphisms of tein in yeast nucleotide excision repair. Nucleic Acids 2002;62:2813^ 8. the DNA repair gene xeroderma pigmentosum group Res 2004;32:5981 ^ 90. 12. Aboussekhra A, Biggerstaff M, Shivji MK, et al. A and risk of primary lung cancer. Cancer Epidemiol 41. Huang WY, Berndt SI, Kang D, et al. Nucleotide ex- Mammalian DNA nucleotide excision repair reconsti- Biomarkers Prev 2002;11:993 ^ 7. cision repair gene polymorphisms and risk of ad- tuted with purified protein components. Cell 1995; 27. Spitz MR,Wu X,WangY, et al. Modulation of nucle- vanced colorectal adenoma: XPC polymorphisms 80:859^68. otide excision repair capacity by XPD polymorphisms modify smoking-related risk. Cancer Epidemiol Bio- 13 . Wood RD. DNA repair. Variants on a theme. Nature in lung cancer patients. Cancer Res 2001;61:1354 ^ 7. markers Prev 2006;15:306 ^ 11. 1999;399:639^40. 28. HuZ,WeiQ,WangX,ShenH.DNArepairgeneXPD 42. ZhuY, Lai M,Yang H, et al. Genotypes, haplotypes, 14. Lindahl T,Wood RD. Quality control by DNA repair. polymorphism and lung cancer risk: a meta-analysis. and diplotypes of XPC and risk of bladder cancer. Car- Science 1999;286:1897^905. Lung Cancer 2004;46:1 ^10. cinogenesis 2006. Epub ahead of print. 15. He Z, Henricksen LA, Wold MS, Ingles CJ. RPA in- 29. Smith TR, Levine EA, Perrier ND, et al. DNA-repair 43. Zhou RM, Li Y,Wang N, Zhang XJ, Dong XJ, Guo volvement in the damage-recognition and incision genetic polymorphisms and breast cancer risk. Cancer W. Correlation of XPC Ala499Val and Lys939Gln steps of nucleotide excision repair. Nature 1995;374: Epidemiol Biomarkers Prev 2003;12:1200 ^ 4. polymorphisms to risks of esophageal squamous cell 566^9. 30. Forsti A, Angelini S, Festa F, et al. Single nucleotide carcinoma and gastric cardiac adenocarcinoma. 16. Li L, Lu X, Peterson CA, Legerski RJ. An interaction polymorphisms in breast cancer. Oncol Rep 2004;11: Ai Zheng 2006;25:1113^ 9. between the DNA repair factor XPA and replication 917^ 22. 44. Shen M, Berndt SI, Rothman N, et al. Polymor- protein A appears essential for nucleotide excision re- 31. Hudmon KS, Honn SE, Jiang H, et al. Identifying phisms in the DNA nucleotide excision repair genes pair. Mol Cell Biol 1995;15:5396 ^ 402. and recruiting healthy control subjects from a man- and lung cancer risk in Xuan Wei, China. Int J Cancer 17. MatsudaT, Saijo M, Kuraoka I, et al. DNA repair pro- aged care organization: a methodology for molecular 2005;116:768 ^ 73. tein XPA binds replication protein A (RPA). J Biol epidemiological case-control studies of cancer. 45. Hecht SS. Tobacco smoke carcinogens and lung Chem 1995;270:4152^7. Cancer Epidemiol Biomarkers Prev 1997;6:565 ^ 71. cancer. J Natl Cancer Inst1999;91:1194 ^ 210. 18. van Gool AJ, Citterio E, Rademakers S, et al. The 32. Ritchie MD, Hahn LW, Roodi N, et al. Multifactor 46. Duell EJ, Wiencke JK, Cheng TJ, et al. Polymor- Cockayne syndrome B protein, involved in transcrip- dimensionality reduction reveals high-order interac- phisms in the DNA repair genes XRCC1and ERCC2 tion-coupled DNA repair, resides in an RNApolymerase tions among estrogen-metabolism genes in sporadic and biomarkers of DNA damage in human blood II ^ containing complex. EMBO J1997;16:5955^65. breast cancer. Am J Hum Genet 2001;69:138 ^ 47. mononuclear cells. Carcinogenesis 2000;21:965 ^ 71.

Clin Cancer Res 2007;13(12) June 15, 2007 3758 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 28, 2021. © 2007 American Association for Cancer Research. Nucleotide Excision Repair Pathway Genes and Oral Premalignant Lesions

Yunfei Wang, Margaret R. Spitz, J. Jack Lee, et al.

Clin Cancer Res 2007;13:3753-3758.

Updated version Access the most recent version of this article at: http://clincancerres.aacrjournals.org/content/13/12/3753

Cited articles This article cites 45 articles, 16 of which you can access for free at: http://clincancerres.aacrjournals.org/content/13/12/3753.full#ref-list-1

Citing articles This article has been cited by 2 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/13/12/3753.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/13/12/3753. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.

Downloaded from clincancerres.aacrjournals.org on September 28, 2021. © 2007 American Association for Cancer Research.