Comprehensive Pathway-Based Association Study of DNA Repair Gene Variants and the Risk of Nasopharyngeal Carcinoma

Published OnlineFirst March 2, 2011; DOI: 10.1158/0008-5472.CAN-10-0469

Cancer Prevention and Epidemiology Research

Hai-De Qin1, Yin Yao Shugart2, Jin-Xin Bei1, Qing-Hua Pan1, Lina Chen3, Qi-Sheng Feng1, Li-Zhen Chen1, Wei Huang4, Jian Jun Liu5, Timothy J. Jorgensen6, Yi-Xin Zeng1, and Wei-Hua Jia1

Abstract DNA repair plays a central role in protecting against environmental carcinogenesis, and genetic variants of DNA repair genes have been reported to be associated with several human malignancies. To assess whether DNAgenevariantswereassociatedwithnasopharyngeal carcinoma (NPC) risk, a candidate gene association study was conducted among the Cantonese population within the Guangdong Province, China, the ethnic group with the highest risk for NPC. A 2-stage study design was utilized. In the discovery stage, 676 tagging SNPs covering 88 DNA repair genes were genotyped in a matched case-control study (cases/controls ¼ 755/

755). Eleven SNPs with Ptrend < 0.01 were identified. Seven of these SNPs were located within 3 genes, RAD51L1, BRCA2,andTP53BP1. In the validation stage, these 11 SNPs were genotyped in a separate Cantonese population (cases/controls ¼ 1,568/1,297). Two of the SNPs (rs927220 and rs11158728), both

in RAD51L1, remained strongly associated with NPC. The SNP rs927220 had a significant Pcombined of 5.55 10 5,withOR¼ 1.20 (95% CI ¼ 1.10–1.30), Bonferroni corrected P ¼ 0.0381. The other SNP (rs11158728), r2 ¼ P 4 which is in strong linkage disequilibrium with rs927220 ( 0.7), had a significant combined of 2.0 10 , Bonferroni corrected P ¼ 0.1372. Gene–environment interaction analysis suggested that the exposures of salted fish consumption and cigarette smoking had potential interactions with DNA repair gene variations, but need to be further investigated. Our findings support the notion that DNA repair genes, in particular RAD51L1,playaroleinNPCetiologyanddevelopment.Cancer Res; 71(8); 3000–8. 2011 AACR.

Introduction person-years) among the Cantonese-speaking subpopulation within a region of the Guangdong province, China (1). The Nasopharyngeal carcinoma (NPC) is rare in most parts of distinctive geographic and ethnic distribution of NPC suggests the world, but is a leading malignancy in southern China and that NPC is a malignancy with complex etiology involving both Southeast Asia, with the highest incidence rate (40 per 100,000 genetic and environmental factors (2, 3). The risk factors most strongly associated with NPC incidence have been shown to be Epstein-Barr virus (EBV) infection and consumption of Authors' Affiliations: 1State Key Laboratory of Oncology in South China, Department of Experimental Research, Sun Yat-Sen University Cancer salt-preserved fish (4). However, other salt-preserved foods (5) Center, Guangzhou, China; 2Unit of Statistical Genomics, Division of and cigarette smoking (6) have also been consistently reported Intramural Division Program, National Institute of Mental Health, NIH, to be moderate risk factors. Bethesda, Maryland; 3Department of Social Medicine, University of Bristol, White Ladies Road, Bristol, United Kingdom; 4Department of Genetics, Environmental carcinogens such as N-nitrosamines, poly- Shanghai-MOST Key Laboratory of Health and Disease Genomics, Chi- cyclic aromatic hydrocarbons, and aromatic amines, are nese National Human Genome Center, Shanghai, China; 5Human Genet- ics, Genome Institute of Singapore, Singapore, Singapore; and known components of salt-preserved foods and tobacco (7). 6Georgetown University, Lombardi Comprehensive Cancer Center, Jess These chemicals form DNA adducts, which can result in and Mildred Fisher Center for Familial Cancer Research, Washington, carcinogenic mutations if left unrepaired. In particular, nitro- District of Columbia samine metabolism-related DNA damage has been recently Note: Supplementary data for this article are available at Cancer Research linked to NPC (8). Furthermore, the metabolites of these Online (http://cancerres.aacrjournals.org/). carcinogens can generate reactive oxygen species (ROS), H.-D. Qin and Y.Y. Shugart contributed equally to this work. which in turn produce base damage, single-strand breaks The views expressed in this presentation do not necessarily represent the (SSBs) and double-strand breaks (DSBs) in DNA (9). views of the NIMH, NIH, HHS, or the United States Government. There is a growing body of evidence that EBV may promote Corresponding Author: Wei-Hua Jia, State Key Laboratory of Oncology – in South China Department of Experimental Research, Sun Yat-Sen DNA damage (10 14). These multiple reports of DNA damage University Cancer Center, Guangzhou 510060, P.R. China. Phone: 8620 promotion by EBV, taken together with the known DNA 8734 3195; Fax: 8620 8734 3392; E-mail: [email protected] damaging activity of the NPC-associated chemical carcino- doi: 10.1158/0008-5472.CAN-10-0469 gens, support the notion that DNA damage may contribute to 2011 American Association for Cancer Research. NPC induction.

3000 Cancer Res; 71(8) April 15, 2011

DNA Repair Gene Variation and NPC Risk

Variants and mutations in various DNA repair genes have intake was either the 3 years before diagnosis (cases) or the 3 been associated with several cancers (15–17). Yet, the prior years before interview (controls). For smoking status, indivi- studies with positive findings for NPC associations have duals who had smoked at least 100 cigarettes in their lifetime targeted only a limited number of DNA repair genes were defined as "smokers." Smokers with cumulative cigarette (18, 19). Because environmental risk factors for NPC contri- smoking exposure of <20 pack-years were defined as "light bute to DNA damage, we hypothesized that individuals with smokers," while 20 pack-years were "heavy smokers." inherited deficiencies in DNA repair genes would be more DNA was extracted from about 5 to 10 mL peripheral blood prone to NPC. In this investigation, we sought to determine by QIAamp DNA Blood midi Kit (Qiagen) and stored in –80C whether common variation in DNA repair genes is associated for subsequent processing and analysis. with NPC, either alone or in combination with nongenetic NPC risk factors, among individuals living in the Guangdong EBV antibody testing Province, China, an NPC endemic region. IgA antibody titers against Epstein-Barr virus Capsid Anti- gen (VCA-IgA) and Early Antigen (EA-IgA) were determined Materials and Methods from peripheral blood samples using a commercial kit based on standard immuno-enzymatic methodologies (Zhongshan Study subjects Bio-tech Co. Ltd.; ref. 20), according to the manufacture's This study included NPC patients who were enrolled at the protocol. In brief, the protocol steps were as follows: (1) B95.8 Sun Yat-Sen University Cancer Center (SYSUCC) between cell smears were prepared and aliquoted in the wells of slides October 2005 and October 2007. Clinical information, including as antigen; (2) diluted sera were added and incubated; (3) after age of onset, histopathologic diagnosis and clinical staging washing, peroxidase-conjugated antihuman IgA antibody was were collected from medical records. All cases were born and added and incubated; (4) wells were flooded with aminoethyl- had continuously lived in the Guangdong province at least for 5 carbazole solution and H2O2; and (5) slides were then exam- years. Cases were histopathologically confirmed for NPC, and ined under the microscope. Brown staining was considered were without any prior history of cancer, immunological positive. disease, or mental disease. Once the patients had consented to join the study, interviews were performed and peripheral Candidate gene selection, tagging SNP selection, and blood was drawn. In the end, a total of 1,948 eligible NPC cases genotyping were identified. Among them, 1,845 interviews were conducted; Eighty-eight genes coding for proteins with major functions 103 of them could not complete the questions. The majority of in DNA repair (21) were selected as candidate genes (Table 1). cases (1,767 cases, 95.8%) were interviewed less than 6 months SNP information was obtained from the NCBI dbSNP database after diagnosis of NPC. A subset of 755 newly diagnosed and the International HapMap Project database (22). To (average 28 days since diagnosis, ranging from 1 to 106 days). capture the common haplotypes, tagging SNPs (htSNPs) were Cantonese-speaking cases were randomly selected from the identified among those SNPs that were assayable (i.e., design database to become the case group for the study. score 0.60) with the GoldenGate genotyping platform (Illu- Healthy controls were recruited from 21 health clinics mina Inc.), using the aggressive multimarker selection with an throughout the Guangdong province. In addition, 35 rural r2 > 0.80. A minor allele frequency (MAF) of 0.05 in the Han villages across the Guangdong province were randomly Chinese was used as a cut-off for tagging SNP selection in selected for recruitment to obtain an urban/rural residence Haploview software (23). A total of 676 tagging SNPs across distribution similar to cases. Each healthy subject was selected the 88 genes were selected for genotyping. randomly and contacted by telephone. If they consented to participate, they were interviewed in person and blood was Genotyping quality control drawn. The inclusion/exclusion criteria were the same as for Manufacturer's controls for genotyping quality were incor- cases. The consent rates were 96% for the cases and 66% for porated into the GoldenGate Genotyping (GGGT) assay, the controls. A subset of healthy controls were then randomly including allele-specific extension controls, PCR uniformity selected from the database, matched 1:1 with the selected controls, gender controls, extension gap controls, first hybri- cases by age (5 years), sex, geographic region, and dialect. disation controls, second hybridization controls, and contam- This study was approved by the Human Ethics Committee, ination controls (http://www.lerner.ccf.org/services/gc). In Sun Yat-Sen University. Informed consent was obtained from addition, we also used duplicates of 8 cases and 8 controls each participant at recruitment. to test the genotyping reproducibility. The concordance rate was 100%. Moreover, a call-rate threshold of 95% was the Epidemiology data collection criteria used to identify analyzable SNPs. During 30-min interviews, trained staff interviewers collected data on demographics, dietary habits, cigarette smok- Statistical analysis ing, family history of cancer, and other potential cancer risk Statistical analyses were performed using R scripts (24) and factors. Regarding salted fish intake, subjects were asked to PLink software (25). The quantile–quantile (Q–Q) plot was choose from 5 intake frequency categories (never, sometimes, generated using R to evaluate the overall significance of the monthly, weekly, and daily). The reference point for childhood associations of the SNPs of candidate genes. Genotypic fre- intake was 6 to 12 years old; the reference point for adulthood quencies in control subjects for each SNP were tested for

www.aacrjournals.org Cancer Res; 71(8) April 15, 2011 3001

Qin et al.

Table 1. Tagging SNP selection for the association study

Pathway Gene symbol (genotyped SNPs/failed SNPs)a Subtotal of genotyped Successful (genes) SNPs/failed SNPs genotyping rate (%)

BER(18) XRCC1(15/2), LIG3(7), APEX1(3), PARP1(7), POLE(4), OGG1(7/1), 96/5 95 MBD4(3), NEIL2(15/1), POLD1(5), PNKP(5), SMUG1(3), UNG(5), PCNA(2), MPG(2), POLB(2/1), MUTYH(4), TDG(6), NEIL1(1) HR(16) NBN(10), BRCA2(20/1), RAD51L1(54/3), XRCC2(6), RAD54L(6/1), 156/9 94 RAD52(10), EME1(3), RAD50(3), MRE11A(8), XRCC3(9), RAD51C(3), RAD51L3(8/2), RAD54B(6/1), BRCA1(5), RAD51(4/1), EFEMP2(1) NHEJ(6) XRCC4(29/4), PRKDC(11), LIG4(6/2), XRCC5(19/1), 79/9 89 DCLRE1C(9), XRCC6(5/2) NER(22) RAD23B(5), RPA3(13/1), ERCC5(12), GTF2H3(4), RPA1(18), 143/9 94 ERCC1(10/2), DDB2(5), MNAT1(15/1), GTF2H4(3), ERCC2(7), LIG1(9/1), XPC(6), RPA2(2/1), FEN1(1), ERCC4(2), XPA(6), CCNH(3/2), ERCC3(3), CDK7(4), XAB2(7), GTF2H1(6), GTF2H5(2/1) MMR(5) MSH3(17/1), PMS2(5/1), MLH1(4), MSH2(6), MSH6(4) 36/2 94 DR(1) MGMT(43/5) 43/5 88 DDC(20) RFC1(9/1), TP53(6/2), RFC4(5), HUS1(7/1), CHEK2(13), CHEK1(7/1), 123/15 88 MDC1(5/2), MDM2(6), RAD17(3), TP53BP1(5/1), ATR(5/1), RFC3(18), CDKN1A(6/1), RAD9A(3/2), RAD1(4), ATM(5/1), RFC5(5), DPAGT1(4/1), RFC2(4), ATRIP(3/1) Total (88) 676/54 92

aThe number of tagging SNPs and the number of failed SNPs are indicated in the parentheses. Abbreviations: BER, excision repair; DDC, DNA damage checkpoints; DR, direct repair; HR, homologous recombination repair; MMR, mismatch repair; NER, nucleotide excision repair; NHEJ, nonhomologous end joining.

departure from Hardy-Weinberg Equilibrium (HWE) using an The classification and regression tree (CART) method exact test. implemented in the R package (rpart; ref. 29), was used to Association analyses were performed using PLink under identify subgroups with higher risk for NPC classified by the different genetic models. In brief, the individual association of genetic variants with significant marginal effects and 2 well- a SNP with NPC risk was evaluated by the Cochran-Armitage established risk factors. CART is a binary recursive-partition- trend test, logistic regression, and a permutation test in PLink. ing method that produces a decision tree to identify sub- An expectation–maximization (EM) algorithm was used for groups of subjects at higher risk. The trees were pruned and haplotype imputation, reconstruction, and frequency estima- optimized into smaller ones with minimal complexity para- tions. Haplotype association tests were conducted by using meters. This process continued until the terminal nodes had the R package Haplo.stats (26). WGAViewer software (27) was no subsequent significant splits or the terminal nodes reached used to create linkage disequilibrium (LD) plots. The criterion a prespecified minimum size. Five-fold cross-validation (4 of 5 for statistical significance of an association was P < 0.01. of the instances for training, remaining 1 of 5 for test) was used Bonferroni correction and the Benjamini and Hochberg (28) for tree model building. step-up FDR approach were used for multiple comparison adjustments of the P values. Validation study Pair-wise gene–environment interactions were analyzed To validate the significant SNPs identified in the discovery using generalized linear model (GLM) implemented in the stage, we genotyped 1,568 NPC cases recruited in SYSUCC R package. To assess the interactions, the likelihood ratio test during 2 periods, from January 2002 to September 2005, and was used to compare the fit of the full model (with the from November 2007 to March 2009, and 1,297 controls interaction term) and the reduced model (without the inter- recruited in the same period from the medical examination action term). In addition, a stratified classification by cross centers of several hospitals in Guangdong. Only the subjects tabulation was also conducted. Statistic significance (P < 0.05) that were Cantonese speaking and had lived in Guangdong indicated a potential interaction. The pair-wised interactions were included. All participants had signed informed consent between each genotyped SNP and the well-established envir- and were interviewed as done before. The inclusion/exclusion onmental risk factors of smoking status and salted fish con- criteria were the same as that described for the discovery stage. sumption were explored. Individuals homozygous for the Genotyping was performed using the Sequenom DNA Mas- major alleles and without exposures to either smoking or sARRAY platform (Sequenom Inc.), and the procedures and salted fish were used as the referent. analytical methods were the same as for the discovery stage.

3002 Cancer Res; 71(8) April 15, 2011 Cancer Research

DNA Repair Gene Variation and NPC Risk

Table 2. NPC risk estimates of top significant SNPs in Cantonese population in Guangdong during 2005 to 2007

a b Pathway Gene SNP Risk allele Risk allele frequency OR (95% CI) Ptrend Pemp (case/control)

BER OGG1 rs2072668 G 0.65/0.60 1.22 (1.04–1.43) 0.0068 0.0072 NER GTF2H1 rs4150581 G 0.64/0.58 1.33 (1.13–1.57) 0.0006 0.0004 MNAT1 rs4151400 C 0.22/0.18 1.23 (1.01–1.49) 0.0053 0.0043 HR BRCA2 rs206119 C 0.18/0.14 1.37 (1.11–1.68) 0.0038 0.0034 rs4942448 T 0.12/0.09 1.37 (1.07–1.76) 0.0072 0.0145 RAD51L1 rs4902562 G 0.33/0.27 1.27 (1.08–1.5) 0.0007 0.0007 rs11158728 G 0.46/0.40 1.25 (1.07–1.45) 0.0009 0.0008 rs927220 C 0.43/0.37 1.22 (1.04–1.43) 0.0017 0.0022 DDC TP53BP1 rs689647 T 0.42/0.37 1.31 (1.12–1.54) 0.0046 0.0047 rs544122 A 0.26/0.21 1.35 (1.13–1.62) 0.0023 0.0022 CHEK2 rs9620817 A 0.92/0.89 1.53 (1.16–2.00) 0.0037 0.005

aORs were estimated by logistic regression under the additive genetic model, adjusted by sex, age, heavy smokers, and salted fish consumption during childhood. bThe empirical P values based on the statistic in Cochran-Armitage trend test. NOTE: Variants with MAF 5% and P < 0.01 are shown. Abbreviations: BER, excision repair; DDC, DNA damage checkpoints; HR, homologous recombination repair; NER, nucleotide excision repair.

Results cant association with an increased risk for NPC. Haplotype GTG in block #4, partitioned by rs4902562, rs1957572, and rs6573824, Study population and genotyping success had a P value of 0.0005 (OR ¼ 1.26; 95% CI ¼ 1.06–1.50). In addi- The average age of the cases was 46.6 years old, and the sex tion, the haplotype TGC in block #5, partitioned by rs12432392, ratio was 2.8 males to 1 female, suggesting the case repre- rs11158728, and rs927220, also had a significant association sentation paralleled the demographics of NPC in the general (OR ¼ 1.29; 95% CI ¼ 1.11–1.49; P ¼ 0.0017; Supplementary population (Supplementary Table S1). Of the 676 tagging SNPs Table S2). A fixed length "sliding" technique using 3 consecutive among 88 genes (Table 1), 54 SNPs (8%) were eliminated either SNPs to scan the effects of haplotypes across RAD51L1 indicated due to poor genotyping (call rate <0.95) or significant devia- at least 2 peaks with strong signals of association (Fig. 1). tion from HWE (P < 0.01). The remaining 622 SNPs (92%) were BRCA2 was also shown to be significantly associated with P ¼ P ¼ included in analyses of NPC risk. A quantile–quantile (Q–Q) NPC at 2 SNPs (rs206119: trend 0.0038; rs4942448: trend plot revealed a good match between the distributions of the 0.0072; Table 2). Furthermore, significant associations for TP53BP1 P ¼ observed P values and those expected by chance, except for were detected for 2 SNPs (rs689647: trend P ¼ deviation within the tail of the distribution. 0.0046; rs544122: trend 0.0023; Table 2). Haplotype-based association analyses further supported the association of Variants of DNA repair genes and NPC risk BRCA2 and TP53BP1 with NPC risk (Supplementary Table S2). Eleven SNPs among 7 DNA repair genes (RAD51L1, BRCA2, In addition to RAD51L, BRCA2,andTP53BP1, which had TP53BP1, OGG1, MNAT1, CHEK2, and GTF2H1) showed sta- multiple SNPs associated with NPC, the OGG1, GTF2H1, tistically significant associations with NPC (P < 0.01; Table 2; MNAT1,andCHEK2 genes each had a single SNP variant Supplementary Fig. S1). Three genes—RAD51L1, BRCA2, and associated with NPC. The risk allele "G" of rs2072668 in OGG1 P TP53BP1—had multiple significant loci. had trend of 0.0068; the risk allele "G" of rs4150581 in GTF2H1 P Three neighboring SNPs in RAD51L1 exhibited strong associa- had trend of 0.0006; the risk allele "C" of rs4151400 P ¼ P ¼ MNAT1 P tionwithNPCrisk(rs4902562: trend 0.0007,rs11158728: trend in had trend of 0.0053; and the risk allele "A" of 2 CHEK2 P 0.0009, rs927220: Ptrend ¼ 0.0017). r for the rs4902562-rs11158728 rs9620817 in had trend of 0.0037 (Table 2). Permuta- pair was 0.41 (D0 ¼ 0.85), for rs4902562–rs927220 pair was 0.35 tion-based P values also supported associations with NPC (D0 ¼ 0.74), for rs11158728–rs927220 was 0.81 (D0 ¼ 0.94). An (Table 2). The 11 SNPs all had permutation test P < 0.01, additional 4 loci in this gene showed weak associations with NPC except for rs4942448 which had a P value of 0.0145. risk (rs4899234: Ptrend ¼ 0.0217, rs6573824: Ptrend ¼ 0.0256, rs911263: Ptrend ¼ 0.0242, rs7148412: Ptrend ¼ 0.0236, respectively). Exploring gene–environment interaction in After partitioning RAD51L1 into 11 LD blocks using Gabriel's susceptibility to NPC Confidence Intervals method, haplotype association with NPC In our study, 96% of NPC cases were EBV VCA-IgA positive, was analyzed for each LD block. Two haplotypes showed signifi- compared with only 19% of controls. After adjusting by age, sex,

www.aacrjournals.org Cancer Res; 71(8) April 15, 2011 3003

Qin et al.

Figure 1. Gene structure of RAD51L1, linkage disequilibrium patterns, and results of haplotype and SNP association with NPC risk. A, the ideogram of chromosome 14, indicating the position of RAD51L1 and its gene context (1 of 3 mRNA variants is shown, vertical lines indicate the exons). B, haplotype association scanning was conducted by Haplo.stat. In brief, to evaluate the association of sub-haplotypes (subsets of alleles from the full haplotype) with NPC risk, we evaluated a "window" of alleles by sliding a fixed-width window (3 SNPs) across the entire haplotype in a gene. The y-axis depicts P values on a minus logarithmic scale; the x-axis represents the SNPs ordered along the chromosome according to their positions. The blue arrows indicate the peaks of negative logarithm of P values (–logP) of "fixed-width window" haplotype association. Note that this method used 3 consecutive SNPs to evaluate the subhaplotype score and test the significance; the 3 SNPs (the subhaplotype) will have a single P value. It produces multiple horizontal lines when plotting the –logP versus the SNPs alongside the x-axis. C, P values for single SNP association tests (the 1-df c2 Cochran-Armitage trend test) in RAD51L1 gene.

Red vertical lines indicate significant SNPs with Ptrend < 0.01. Their positions on the linkage disequilibrium diagram are indicated by black squares. D, the estimates of linkage disequilibrium parameter (R-Square). Colors are coded according to the scale and shown at the bottom left.

and education, heavy smokers had a moderate risk with OR of with variants of DNA repair genes TP53BP1, RAD50, RAD54B, 1.81 (95% CI ¼ 1.39–2.35), and the individuals who consumed salt- RAD54L, LIG1, LIG3, LIG4, and POLE, with increased ORs ed fish during childhood had an OR of 3.28 (95% CI ¼ 2.64–4.09). ranging from 3.16 to 4.43 compared with those individuals GLM pair-wise interaction analyses suggested that "con- harboring the homozygous major alleles and without expo- sumption of salted fish during childhood" had interactions sure (the referent). Similarly, smoking status appeared to

3004 Cancer Res; 71(8) April 15, 2011 Cancer Research

DNA Repair Gene Variation and NPC Risk

Figure 2. Application of classification and regression tree analysis to identify the subgroups at higher risk of NPC. The labels on the branches are the environmental variables and/or SNPs used for partitioning. For each terminal, the weight, OR, and the confidence interval (95%CI) are all indicated in the square. Bottom panel depicts the frequencies of the control (open bar) and the case (filled bar) in the corresponding terminal nodes in the upper panel. The model shows the subgroups with higher risk of NPC classified by the salted fish consumption, smoking status, and the variants in RAD51L1, BRCA2, TP53BP1, and MNAT1 genes.

interact with genetic variants of MGMT, RFC3, RAD51L1, and 4.13 (95% CI ¼ 2.41–7.17), 5.59 (95% CI ¼ 2.59–12.60), and 6.42 RAD52, with ORs ranging from 1.52 to 1.82 (Supplementary (95% CI ¼ 3.83–10.96), respectively. Moreover, a subgroup with Table S5). Due to the fact that almost all cases were EBV higher risk was also identified in the salted fish nonconsump- positive (i.e., 96% positive by VCA-IgA titer), we did not have tion populations. In node #16, the heavy smokers showed adequate power to identify any genotypic interaction with higher risk to NPC when carrying the variant allele of EBV exposure. rs11158728, OR ¼ 3.39 (95% CI ¼ 2.04–5.70). Nevertheless, in The CART method identified a few subgroups with higher node #12, although nonheavy smokers, the populations carry- risk of NPC among the risk factors and the genetic variants of ing multiple variant alleles still showed a higher risk, OR ¼ 2.72 the TP53BP1, RAD51L1, BRCA2, and MNAT1 genes (Fig. 2; (95% CI ¼ 1.53–4.84; Fig. 2; Supplementary Table S6). Supplementary Table S6). There was an initial split on salted However, when we tested the potential interactions between fish consumption, suggesting that salted fish consumption was rs544122 and salted fish consumption during childhood by one of the most important risk factors for NPC. More inter- comparing the full model (z ¼ b0 þ b1x1 þ b2x2 þ b3x1 x2) estingly, distinct patterns for consumption and nonconsump- to the reduced model (z ¼ b0 þ b1x1 þ b2x2), we did tion were observed in the CART tree. That is, in the right branch not observe any interaction between these 2 variables of the tree, the nodes with consumption of salted fish showed (P ¼ 0.3076). higher ORs than the nonconsumption nodes in the left branch. As compared with the referent (Node #5), those individuals Validation of the significant SNPs in a separate who consumed salted fish and carried allelic variants showed population dramatically increased risk of NPC (e.g., those carriers of Table 3 shows the results of the analyses of the 11 most variant rs544122 in TP53BP1 were at the highest risk (OR ¼ significantly associated SNPs in the validation set. rs927220 in RAD51L1 P ¼ 7.6; 95% CI ¼ 4.75–12.34, node #25). Similar results were was significantly associated with NPC risk ( trend 5 observed in rs927220, rs492448, and rs4151400, with OR ¼ 0.0085 in the validation data set alone and Ptrend ¼ 5.55 10

www.aacrjournals.org Cancer Res; 71(8) April 15, 2011 3005

Qin et al.

Table 3. Validation in a separate Cantonese population recruited in Guangdong during 2002 to 2005 and 2007 to 2009 for the 11 top significant SNPs in the discovery stage

Validation (n ¼ 1,568/1,297)a Combined (n ¼ 2,323/2,052)*

c c Gene SNP Risk Phwe/Call Risk allele OR (95% CI) Ptrend Risk allele OR (95% CI) Ptrend allele rate %b frequency frequency (case/control) (case/control)

OGG1 rs2072668 G 0.0599/99.4 0.60/0.61 0.99 (0.88–1.10) 0.7531 0.62/0.61 1.06 (0.97–1.16) 0.2051 GTF2H1 rs4150581 G 0.4899/99.5 0.60/0.60 0.98 (0.88–1.10) 0.8204 0.61/0.59 1.09 (1.00–1.19) 0.0701 MNAT1 rs4151400 C 0.5415/99.5 0.20/0.19 1.06 (0.92–1.21) 0.3527 0.21/0.19 1.13 (1.02–1.26) 0.0173 BRCA2 rs206119 C 0.4885/99.6 0.17/0.17 0.99 (0.86–1.14) 0.7296 0.17/0.16 1.08 (0.96–1.21) 0.1375 BRCA2 rs4942448 T 0.0858/99.9 0.10/0.11 0.97 (0.82–1.16) 0.0369 0.11/0.10 1.08 (0.94–1.24) 0.2400 RAD51L1 rs4902562 G 0.0092/99.9 0.31/0.32 0.97 (0.86–1.08) 0.6289 0.32/0.30 1.09 (0.99–1.19) 0.0923 RAD51L1 rs11158728 G 0.0005/99.5 0.45/0.42 1.12 (1.01–1.24) 0.0335 0.45/0.41 1.17 (1.08–1.27) 0.0002 RAD51L1 rs927220 C 0.0755/98.5 0.43/0.40 1.14 (1.03–1.27) 0.0085 0.43/0.39 1.20 (1.10–1.30) 5.55105 TP53BP1 rs689647 T NA –– –––– TP53BP1 rs544122 A 0.3444/99.3 0.25/0.24 1.05 (0.93–1.19) 0.3234 0.26/0.23 1.14 (1.03–1.26) 0.0080 CHEK2 rs9620817 A Failed –– ––––

an, the number of cases vs. controls. bNA, the SNP could not be subjected to the Sequenom genotyping platform. cOdds ratio was adjusted by sex and age.

in the combined data set). In addition, rs11158728 in RAD51L1 and remained significant after multiple comparison correction provided a weak association (Ptrend ¼ 0.0335 for the validation (P ¼ 0.0381 after Bonferroni correction). In addition, another data set and Ptrend ¼ 0.0002 for the combined data set; see the SNP in RAD51L1 (rs11158728) also had a significant combined Table 3). Moreover, rs544122 in TP53BP1 and rs4942448 in P value (2.0 10 4), but fell below significance after multiple BRCA2 had Ptrend < 0.05 either in the validation set or in the comparison correction. These 2 SNPs are in strong LD with combined data set. each other (D0 ¼ 1.0, r2 ¼ 0.7). We then used the Bonferroni correction and the Benjamini Further, we explored the LD patterns between SNPs with and Hochberg step-up FDR approach for multiple comparison possible function in RAD51L1 and the significantly associated adjustments. In particular, SNP rs927220 suggested evidence SNPs (rs11158728, rs4902562, and rs927220), but no putative of association after being corrected for multiple testing (Bon- functional SNPs have yet been identified as being in strong LD ferroni adjusted P ¼ 0.085 for the validation data set, and P ¼ with these SNPs. A total of 53 putative functional SNPs were 0.0381 for the combined data set; Supplementary Table S4). identified using the latest dbSNP (Build 131) based on their gene locations, including 17 SNPs within the coding region, 30 Discussion SNPs located within 2 kb upstream of the gene, 1 SNP located within 0.5 kb downstream of the gene and 5 SNPs in the 30- A haplotyping-tagging approach was used to systematically UTR. They are considered "potential functional SNPs". Using investigate the associations between hundreds of common the Chinese Han population (CHB) genotype data within variants of DNA-repair genes and NPC risk. The advantage of HapMap, we found that only 2 out of 53 SNPs had qualified this comprehensive approach is that it covered all of the major genotypes, and LD between the 3 NPC-associated SNPs and 2 human DNA repair pathways and virtually all known human potential functional SNPs was low (Supplementary Table S3). DNA repair genes. This is, thus far, the most comprehensive In contrast, the SNPs in strong LD (r2 0.5) with rs11158728, multigenic study evaluating associations between a large rs4902562, and rs927220 in RAD51L1 are all located within number of DNA repair gene variants and NPC. introns (data not shown). Further investigations need to be In the discovery stage, the most notable finding was that 7 conducted to make conclusive comments about the relation- of the 11 most significant variants were located within just 3 ship of these SNPs and the putative functional SNPs. genes, RAD51L1, BRCA2, and TP53BP1. These associations RAD51L1 is a member of the RAD51 protein family, which were maintained at both the individual SNP and at the regulates the central activity of homologous recombination haplotype levels. (HR) DNA repair (30). RAD51 and its homologues have been In the validation stage, within a separate Cantonese popu- reported to be very important to cancer risk, particularly for lation, 2 of the SNPs, both located in RAD51L1, retained their breast and ovarian cancers (31). Recently, a 3-stage GWAS on association with NPC. Specifically, rs927220 in RAD51L1 was breast cancer in 9,770 cases and 10,799 controls mapped the the most strongly associated, with a combined P ¼ 5.55 10 5, susceptibility locus to RAD51L1 (32). RAD51L1 has been

3006 Cancer Res; 71(8) April 15, 2011 Cancer Research

DNA Repair Gene Variation and NPC Risk

reported to form a stable heterodimer with RAD51 family Taiwanese (46). The third relatively small-scale GWAS found member RAD51L2, which then further interacts with the other that the integrin-alpha 9 gene (ITGA9) was associated with family members, such as RAD51, as well as XRCC2 and XRCC3 NPC risk (47). (33). RAD51L1(–/–) cells are hypersensitive to DNA cross- Our candidate gene association study provided a focused linking agents and have much reduced formation of RAD51 view of specific DNA repair genomic regions. Only 1 SNP in nuclear foci (34), suggesting that DNA repair is compromised RAD51L1 (rs927220) survived the Bonferroni correction. This when the gene is dysfunctional. same SNP had also been genotyped in a previous GWAS study In addition to RAD51L1, we found that BRCA2 was also (45) and had a moderate P value of 0.02 in Cantonese associated with NPC in the discovery stage. BRCA2 is another population. The P value below the threshold of our GWAS major gene involved in the HR repair pathway and the BRCA2 phase I study for further validation study might due to its lack protein physically interacts with RAD51 proteins (35), and of power. More importantly, the directions of the association thus has functional relevance to RAD51L1. Nevertheless, the effect revealed by the previous GWAS and current candidate BRCA2 SNPs failed to validate. gene study are consistent [OR in GWAS is 1.17 (95% CI ¼ 1.02– Previous studies have reported that OGG1 was associated 1.35); OR in candidate gene study is 1.22 (95% CI ¼ 1.04–1.43)]. with NPC (36, 37) and other cancers (38, 39). In our study, We believed that this candidate gene study could be a com- rs2072668 in OGG1 showed statistically significant evidence of plement to the findings from previous GWAS. association in the discovery stage (P ¼ 0.0068). This SNP is in In summary, rs927220 in RAD51L1 was discovered to be strong LD with nonsynonymous SNP rs1052133 (r2 ¼ 1.0; associated with NPC, and subsequently validated at a sig- Supplementary Table S3) that may be functional. Yet, nificant level in another study group. This SNP is located in rs2072668 was not significantly associated with NPC in the an intron, and therefore is unlikely to be functional itself, but validation stage. it may be in LD with a yet to be identified functional locus. Because NPC is a malignancy that fits a multifactorial model, Other SNP associations identified in the discovery stage it is important to account for environmental risk factors when could not be validated and further studies are needed. This assessing genetic associations. Pair-wise interaction results study supports the notion that DNA repair genes, in parti- suggested possible interactions between environmental risks cular the RAD51L1 gene, might play a role in NPC etiology and genetic components. As indicated in Table S5, many of the and development. variants involved in the putative interactions were from the RAD gene family (i.e., RAD50, RAD54B, RAD54L, and RAD51L1). Disclosure of Potential Conflicts of Interest Although the pattern of genes with potential interactions shows a degree of commonality, some displayed negative The authors declared no conflict of interest. interactions, which are hard to interpret mechanistically. This raises some question as to whether the statistical interactions Acknowledgments found here have biological relevance. We gratefully acknowledge the participation of all individuals, without whom CART models yielded large ORs (ranging from 4.1 to 7.6) for this work would not have been possible. We also thank Professor Shao-Qi Rao individuals consuming salted fish and carrying more than 1 for his help in manuscript revision. risk genotype in RAD51L1, TP53BP1, and BRCA2. Statistically significant departures from multiplicative models were not Grant Support observed; however, much larger sample sizes will be needed to This work was supported by the National Natural Science Foundation of assess the joint effects of DNA repair variation and nongenetic China (30671798, 30471487, u0732005), the National Science and Technology risk factors of NPC. Support Program of China (2006BAI02A11), the National Major Basic Research Program of China (863: 2006AA02A404) and the National Institute of Health Previous studies have identified other loci associated with (RO3CA113240–01), USA. NPC. Among them, HLA has consistently been reported as a The costs of publication of this article were defrayed in part by the payment susceptibility region (40–42). [Other susceptibility loci include of page charges. This article must therefore be hereby marked advertisement in 4p15.1-q12 (43) and 3p.21 (44).] Recently, 2 independent accordance with 18 U.S.C. Section 1734 solely to indicate this fact. genome-wide association studies (GWAS) supported the Received February 5, 2010; revised January 5, 2011; accepted February 17, HLA region as an NPC risk locus in the Cantonese (45) and 2011; published OnlineFirst March 2, 2011.

References 1. Jia WH, Huang QH, Liao J, Ye W, Shugart YY, Liu Q, et al. 4. Yu MC, Ho JH, Lai SH, Henderson BE. Cantonese-style salted fish as Trends in incidence and mortality of nasopharyngeal carcinoma a cause of nasopharyngeal carcinoma: report of a case-control study over a 20–25 year period (1978/1983–2002) in Sihui and in Hong Kong. Cancer Res 1986;46:956–61. Cangwu counties in southern China. BMC Cancer 2006;6: 5. Yu MC. Nasopharyngeal carcinoma: epidemiology and dietary fac- 178. tors. IARC Sci Publ 1991;105:39–47. 2. Young LS, Rickinson AB. Epstein-Barr virus: 40 years on. Nat Rev 6. Nam JM, McLaughlin JK, Blot WJ. Cigarette smoking, alcohol, and Cancer 2004;4:757–68. nasopharyngeal carcinoma: a case-control study among U.S. whites. 3. Jia WH, Feng BJ, Xu ZL, Zhang XS, Huang P, Huang LX, et al. Familial J Natl Cancer Inst 1992;84:619–22. risk and clustering of nasopharyngeal carcinoma in Guangdong, 7. Hecht SS. Tobacco smoke carcinogens and lung cancer. J Natl China. Cancer 2004;101:363–9. Cancer Inst 1999;91:1194–210.

www.aacrjournals.org Cancer Res; 71(8) April 15, 2011 3007

Qin et al.

8. Dodd LE, Sengupta S, Chen IH, den Boon JA, Cheng YJ, Westra W, 29. Therneau TM, Atkinson B, Ripley B. rpart: Recursive Partitioning. R et al. Genes involved in DNA repair and nitrosamine metabolism and package (V. 3.1–42). [http://cran.r-project.org/web/packages/rpart/]. those located on chromosome 14q32 are dysregulated in nasophar- 30. Thacker J. The RAD51 gene family, genetic instability and cancer. yngeal carcinoma. Cancer Epidemiol Biomarkers Prev 2006;15:2216– Cancer Lett 2005;219:125–35. 25. 31. Antoniou AC, Sinilnikova OM, Simard J, Leon e M, Dumont M, Neu- 9. Frenkel K. Carcinogen-mediated oxidant formation and oxidative DNA hausen SL, et al. RAD51 135G–>C modifies breast cancer risk among damage. Pharmacol Ther 1992;53:127–66. BRCA2 mutation carriers: results from a combined analysis of 19 10. O’Nions J, Allday MJ. Deregulation of the cell cycle by the Epstein- studies. Am J Hum Genet 2007;81:1186–200. Barr virus. Adv Cancer Res 2004;92:119–86. 32. Thomas G, Jacobs KB, Kraft P, Yeager M, Wacholder S, Cox DG, et al. 11. Liu MT, Chen YR, Chen SC, Hu CY, Lin CS, Chang YT, et al. Epstein- A multistage genome-wide association study in breast cancer iden- Barr virus latent membrane protein 1 induces micronucleus formation, tifies two new risk alleles at 1p11.2 and 14q24.1 (RAD51L1).Nat Genet represses DNA repair and enhances sensitivity to DNA-damaging 2009;41:579–84. agents in human epithelial cells. Oncogene 2004;23:2531–9. 33. Miller KA, Yoshikawa DM, McConnell IR, Clark R, Schild D, Albala 12. O’Nions J, Turner A, Craig R, Allday MJ. Epstein-Barr virus selectively JS. RAD51C interacts with RAD51B and is central to a larger protein deregulates DNA damage responses in normal B cells but has no complex in vivo exclusive of RAD51. J Biol Chem 2002;277:8406– detectable effect on regulation of the tumor suppressor p53. J Virol 11. 2006;80:12408–13. 34. Takata M, Sasaki MS, Sonoda E, Fukushima T, Morrison C, Albala JS, 13. Gruhne B, Sompallae R, Marescotti D, Kamranvar SA, Gastaldello S, et al. The Rad51 paralog Rad51B promotes homologous recombina- Masucci MG. The Epstein-Barr virus nuclear antigen-1 promotes tional repair. Mol Cell Biol 2000;20:6476–82. genomic instability via induction of reactive oxygen species. Proc 35. Pellegrini L, Yu DS, Lo T, Anand S, Lee M, Blundell TL, et al. Insights Natl Acad Sci USA 2009;106:2313–8. into DNA recombination from the structure of a RAD51-BRCA2 14. Gruhne B, Sompallae R, Masucci MG. Three Epstein-Barr virus complex. Nature 2002;420:287–93. latency proteins independently promote genomic instability by indu- 36. Liu N, Zhou XX, Lu LX. Expression and implication of base excision cing DNA damage, inhibiting DNA repair and inactivating cell cycle repair genes in human nasopharyngeal carcinoma and non-tumor checkpoints. Oncogene 2009;28:3997–4008. nasopharyngeal tissues. Chin J Cancer 2008;27:126–32. 15. Goode EL, Ulrich CM, Potter JD. Polymorphisms in DNA repair genes 37. Cho EY, Hildesheim A, Chen CJ, Hsu MM, Chen IH, Mittl BF, et al. and associations with cancer risk. Cancer Epidemiol Biomarkers Prev Nasopharyngeal carcinoma and genetic polymorphisms of DNA repair 2002;11:1513–30. enzymes XRCC1 and hOGG1. Cancer Epidemiol Biomarkers Prev 16. Jiang J, Zhang X, Yang H, Wang W. Polymorphisms of DNA repair 2003;12:1100–4. genes: ADPRT,XRCC1, and XPD and cancer risk in genetic epide- 38. Yuan W, Xu L, Feng Y, Yang Y, Chen W, Wang J, et al. The hOGG1 miology. Methods Mol Biol 2009;471:305–33. Ser326Cys polymorphism and breast cancer risk: a meta-analysis. 17. Benhamou S, Sarasin A. ERCC2/XPD gene polymorphisms and Breast Cancer Res Treat 2010;122:835–42. cancer risk. Mutagenesis 2002;17:463–9. 39. Sliwinski T, Przybylowska K, Markiewicz L, Rusin P, Pietruszewska W, 18. Yang ZH, Dai Q, Kong XL, Yang WL, Zhang L. Association of ERCC1 Zelinska-Blizniewska H, et al. MUTYH Tyr165Cys, OGG1 Ser326Cys polymorphisms and susceptibility to nasopharyngeal carcinoma. Mol and XPD Lys751Gln polymorphisms and head neck cancer suscept- Carcinog 2009;48:196–201. ibility: a case control study. Mol Biol Rep 2011;38:1251–61. 19. Cho EY, Hildesheim A, Chen CJ, Hsu MM, Chen IH, Mittl BF, et al. 40. Simons MJ, Wee GB, Chan SH, Shanmugaratnam K. Probable iden- Nasopharyngeal carcinoma and genetic polymorphisms of DNA repair tification of an HL-A second-locus antigen associated with a high risk enzymes XRCC1 and hOGG1. Cancer Epidemiol Biomarkers Prev of nasopharyngeal carcinoma. Lancet 1975;7899:142–143. 2003;12:1100–4. 41. Hildesheim A, Apple RJ, Chen CJ, Wang SS, Cheng YJ, Klitz W, et al. 20. Ye YZ, Sun Y, Li LJ, Xiao XB, Zhang CQ, Feng KT. Preparation and Association of HLA class I and II alleles and extended haplotypes application of a quality control sera in VCA IgA immuno-enzymatic with nasopharyngeal carcinoma in Taiwan. J Natl Cancer Inst 2002; assay (Chinese). Chin J Cancer 1997;16:79–80. 94:1780–9. 21. Wood RD, Mitchell M, Sgouros J, Lindahl T. Human DNA repair genes. 42. Lu SJ, Day NE, Degos L, Lepage V, Wang PC, Chan SH, et al. Linkage Science 2001;5507:1284–9. of a nasopharyngeal carcinoma susceptibility locus to the HLA region. 22. The International HapMap Consortium. The International HapMap Nature 1990;346:470–1. Project. Nature 2003;426:789–96. 43. Feng BJ, Huang W, Shugart YY, Lee MK, Zhang F, Xia JC, et al. 23. Barrett JC, Fry B, Maller J, Daly MJ. Haploview:analysis and visualiza- Genome-wide scan for familial nasopharyngeal carcinoma tion of LD and haplotype maps. Bioinformatics 2005;21:263–5. reveals evidence of linkage to chromosome 4. Nat Genet 2002;31: 24. R Development Core Team. R: A language and environment for 395–9. statistical computing. (V. 2.8.1). [http://www.R-project.org]. 44. Xiong W, Zeng ZY, Xia JH, Xia K, Shen SR, Li XL, et al. A susceptibility 25. Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, Bender D, locus at chromosome 3p21 linked to familial nasopharyngeal carci- et al. PLINK: a tool set for whole-genome association and population- noma. Cancer Res 2004;64:1972–4. based linkage analyses. Am J Hum Genet 2007;81:559–75. 45. Bei JX, Li Y, Jia WH, Feng BJ, Zhou G, Chen LZ, et al. A genome-wide 26. Schaid DJ, Rowland CM, Tines DE, Jacobson RM, Poland GA. Score association study of nasopharyngeal carcinoma identifies three new tests for association between traits and haplotypes when linkage susceptibility loci. Nat Genet 2010;42:599–603. phase is ambiguous. Am J Hum Genet 2002;70:425–34. 46. Tse KP, Su WH, Chang KP, Tsang NM, Yu CJ, Tang P, et al. Genome- 27. Ge D, Zhang K, Need AC, Martin O, Fellay J, Urban TJ, et al. wide association study reveals multiple nasopharyngeal carcinoma- WGAViewer: software for genomic annotation of whole genome associated loci within the HLA region at chromosome 6p21.3. Am J association studies. Genome Res 2008;18:640–3. Hum Genet 2009;85:194–203. 28. Benjamini Y, Hochberg Y. Controlling the false discovery rate: A 47. Ng CC, Yew PY, Puah SM, Krishnan G, Yap LF, Teo SH, et al. A practical and powerful approach to multiple testing. J Roy Stat Soc genome-wide association study identifies ITGA9 conferring risk of 1995;57:289–300. nasopharyngeal carcinoma. J Hum Genet 2009;54:392–7.

3008 Cancer Res; 71(8) April 15, 2011 Cancer Research

Comprehensive Pathway-Based Association Study of DNA Repair Gene Variants and the Risk of Nasopharyngeal Carcinoma

Hai-De Qin, Yin Yao Shugart, Jin-Xin Bei, et al.

Cancer Res 2011;71:3000-3008. Published OnlineFirst March 2, 2011.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-10-0469

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2011/03/02/0008-5472.CAN-10-0469.DC1

Cited articles This article cites 45 articles, 11 of which you can access for free at: http://cancerres.aacrjournals.org/content/71/8/3000.full#ref-list-1

Citing articles This article has been cited by 1 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/71/8/3000.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://cancerres.aacrjournals.org/content/71/8/3000. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.