Nucleotide Excision Repair Gene Polymorphisms and Risk of Advanced Colorectal Adenoma: XPC Polymorphisms Modify Smoking-Related Risk

The recent discovery of a novel autosomal recessive form of familial adenomatous polyposis caused by mutations in the base excision repair gene MYH (1) and the well-documented involvement of mismatch repair in hereditary nonpolyposis colorectal cancer suggest the importance of DNA repair mechanisms in colorectal carcinogenesis (2). Based on epidemiologic, molecular, and clinical evidence, colorectal adenomas, particularly those large in size or histologically advanced, are recognized precursor lesions for colorectal cancer (3).

Cigarette smoking is a risk factor for colorectal adenoma and likely also for colon cancer (4). Tobacco products, such as polycyclic aromatic hydrocarbons, heterocyclic amines, nitrosamines, and aromatic amines, reaching the colorectal mucosa, through direct ingestion or via the circulatory system, result in DNA adducts, leading to mutations and potentially to cancer development (5). There is considerable interindividual variation in the response to genotoxic exposures, such as polycyclic aromatic hydrocarbons, and genetic polymorphisms in the DNA repair genes could influence DNA repair capacity and, in turn, cancer susceptibility (6).

The nucleotide excision repair pathway may have particular importance for smoking-related adenoma risk because this system removes complex bulky adducts, including those caused by polycyclic aromatic hydrocarbons in tobacco smoke (7). Excision repair involves global genome repair and transcription coupled repair, involving DNA damage recognition, DNA unwinding, excision of the damage, synthesis of new DNA, gap repair, and ligation (Fig. 1; ref. 8). Damage recognition is carried out either by the XPC-hHR23B (encoded by RAD23B) protein complex in the global genome repair pathway or by RNA polymerase II, which recruits CSA and CSB (ERCC6) in transcription coupled repair. In the second step, DNA is unwound by the action of TFIIH, a nine-subunit protein complex composed of DNA helicases, XPB and XPD, and a kinase subunit cyclin H (CCNH). Replication protein A (RPA) stabilizes the unwound DNA, and XPA facilitates the assembly of DNA repair factors. The damaged DNA is then removed by XPG (ERCC5) and the XPF (ERCC4)-ERCC1 complex, which make 3′ and 5′ incisions, respectively, on either side of the damage. Finally, with the help of several replication accessory factors, the DNA gap is filled with new bases synthesized by DNA polymerases δ and ε and a DNA ligase seals the nick.

Several polymorphisms in nucleotide excision repair genes were found to alter DNA repair, including variants in XPC (9-11) and XPD (ERCC2; refs. 10, 12), although not all studies have been consistent (13-15). Despite the potential role of nucleotide excision repair in colorectal carcinogenesis, few epidemiologic studies of colorectal tumors have evaluated the associations with genetic polymorphisms in this pathway. Studies of colorectal cancer in Taiwan (16) and the United Kingdom (17) found no overall effects for selected polymorphisms in XPD, XPF (ERCC4), XPG (ERCC5), and ERCC1. Genotype effects among smoking subgroups, however, have not been evaluated, and no data were available for colorectal adenoma.

We studied the risk for advanced colorectal adenoma in relation to 15 common single nucleotide polymorphisms (SNPs) in seven genes in the nucleotide excision repair pathway—XPC, RAD23B (hHR23B), CSB (ERCC6), XPD, CCNH, XPF, and XPG—among participants in a large colorectal cancer screening trial in the United States. All SNPs are nonsynonymous, leading to amino acid substitution in the protein product.

The National Cancer Institute Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial, conducted at 10 U.S. screening centers (Birmingham, Alabama; Denver, Colorado; Detroit, Michigan; Honolulu, Hawaii; Marshfield, Wisconsin; Minneapolis, Minnesota; Pittsburgh, Pennsylvania; Salt Lake City, Utah; St. Louis, Missouri; and Washington District of Columbia), is designed to determine whether screening with flexible sigmoidoscopy (60 cm sigmoidoscope), chest X-ray, digital rectal exam plus serum prostate-specific antigen, and CA125 plus transvaginal ultrasound can reduce mortality from cancers of the prostate, lung, colorectum, and ovary, respectively (18, 19). Primary recruitment methods involved mailing informational brochures and letters of invitation to age-eligible persons identified on public, commercial, or screening center-specific mailing lists. To be eligible for the trial, subjects must have been 55 and 74 years of age; not undergoing treatment for cancer (excluding nonmelanoma skin) and with no history of cancer at any of the sites involved in the trial; no history of surgical removal of the prostate, colon, or one lung; no current participation in another cancer screening or primary prevention trial; no use of Finasteride in the previous 6 months; no prostate-specific antigen test in the previous 3 years; no colonoscopy, sigmoidoscopy, or barium enema in the previous 3 years; and the ability to sign the consent form. A total of 77,483 participants (50% men; 87% White) were randomized to the screening arm and a similar number of participants were randomized to the nonscreened arm from 1993 to 2001.

At the baseline, subjects in the screening arm received a flexible sigmoidoscopic visualization of the distal colon (60 cm). If the sigmoidoscopic examination was suspicious for neoplasia (polyp or mass), participants were referred to their primary physicians for subsequent diagnostic follow-up, which generally resulted in a full examination of the colon by colonoscopy and surgery, if needed. Pathology reports regarding the pathologic size and location of each lesion found from colonoscopy or repeat sigmoidoscopy were obtained and coded by trained medical abstractors. Participants provided written informed consent. The study was approved by the institutional review boards of the National Cancer Institute and the 10 screening centers.

As described previously (20, 21), a nested case-control study was conducted among participants randomized to the screening arm of the PLCO Trial, who had a successful sigmoidoscopy (insertion to at least 50 cm with >90% of mucosa visible or a suspect lesion identified), completed a risk factor questionnaire, provided a blood sample, and consented to participate in etiologic studies (n = 42,037) between September 1993 and September 1999. Of these participants, we excluded 4,834 with a self-reported history of ulcerative colitis, Crohn's disease, familial polyposis, colorectal polyps, Gardner's syndrome, or cancer (except basal cell skin cancer). We randomly selected 772 of 1,234 cases with at least one advanced colorectal adenoma (adenoma ≥1 cm or containing high-grade dysplasia or villous characteristics) in the distal colon (descending colon and sigmoid or rectum) and 777 controls without evidence of a polyp or other suspicious lesion in the distal colon (i.e., a negative sigmoidoscopy screening) at baseline, frequency-matched to the cases by gender and ethnicity.

Participants completed a baseline risk factor questionnaire in which they reported their body size and the use of tobacco, alcohol, selected drugs, and hormones, and a 137-item food frequency questionnaire in which they reported their usual dietary intake over the 12 months before enrollment. Detailed information on smoking history was collected, including ages started and stopped, total years of use, amount usually used, and type of tobacco used (cigarettes, pipes, and cigars). Subjects who did not smoke or only smoked cigarettes for <6 months or only smoked pipes or cigars for <1 year were considered to be nonsmokers. For evaluation of risks in relation to time period of cigarette use, cigarette users were classified as long-term quitters (quit ≥10 years before enrollment) and current or recent smokers (quit <10 year before enrollment).

DNA samples were obtained from stored blood samples using Qiagen standard protocols (QIAamp DNA Blood Midi or Maxi kit: (http://www1.qiagen.com/default.aspx). Fifteen SNPs in seven DNA repair genes, XPC (_R492_H, _A499_V, _K939_Q), RAD23B (hHR23B, _A249_V), CSB (ERCC6, _M1097_V, _R1230_P, _Q1413_R), XPD (ERCC2, _D312_N, _K751_Q), CCNH (_V270_A), XPF (ERCC4, _R379_S, R415Q), and XPG (ERCC5, _M254_V, _C529_S, _D1104_H) were genotyped. All assays were validated and optimized at the National Cancer Institute Core Genotyping Facility as described on the website (http://snp500cancer.nci.nih.gov; ref. 22). Internal laboratory quality controls consisted of Coriell DNA samples containing the homozygous major allele, heterozygous, and homozygous minor allele genotypes for each polymorphism under study with four of each control type and four no template controls in every 384 samples. For external blinded quality controls, we interspersed ∼10% repeated quality control samples from 40 individuals. The blinded samples showed >99% interassay concordance for all assays. Approximately 8% of individuals were excluded because of insufficient DNA (7%) or fingerprint profile review showing subject-specific ambiguities (1%). Of those remaining, genotype data were successfully obtained for an average of 94% (90-98%) of study subjects.

Departures from Hardy-Weinberg equilibrium among controls for each self-described ethnic group were assessed by comparing the observed genotype frequencies to the expected frequencies using a χ² test or exact test if cell counts were small. Odds ratios (OR) and 95% confidence intervals (95% CI) for the genotype associations were obtained using unconditional logistic regression, adjusting for gender, race (White, Black, and others), and age (55-59, 60-64, 65-69, and 70-74 years). Trend tests were done using a score of 0, 1, or 2 for each genotype in a logistic regression model. Statistical significance of the multiplicative interaction between DNA repair genotype and smoking was assessed by comparing nested models with and without cross-product terms (assuming linearity for smoking-related risk) using a likelihood ratio statistic. Adenoma risks associated with haplotypes defined by the SNPs within the same gene were assessed using HaploStats (http://www.mayo.edu/hsr/people/schaid.html), which uses the expectation-maximization algorithm to estimate haplotypes. A global score test was used to evaluate overall differences in haplotype frequencies between cases and controls adjusted for covariates (23), and the effect of individual haplotypes on adenoma risk was assessed using the generalized linear model implemented in HaploStats (23, 24). Multiplicative interaction between XPC haplotypes and smoking was assessed by comparing nested models, with and without cross-product terms (assuming linearity for smoking-related risk), using a likelihood ratio statistic.

Cases and controls were similar with regard to the matching factors, gender and race (Table 1). Cases tended to have a higher body mass index (P = 0.3), were older (P < 0.0001), more likely to have reported a first-degree family history of colorectal cancer (P = 0.02), and had less education (P = 0.001). Consistent with results from the entire cohort (25), cigarette smoking was associated with increased risk for advanced adenoma (ever versus never, OR, 1.4; 95% CI, 1.2-1.8) with greater risks among current or recent cigarette smokers (OR, 2.4; 95% CI, 1.8-3.2) than long-term quitters (OR, 1.1; 95% CI, 0.9-1.4 for quit ≥10 years). Risk relationships for level (OR, 1.7; 95% CI, 1.3-2.2 for >20 cigarettes/d) and pack-years of use (OR, 1.8; 95% CI, 1.3-2.3 for ≥30 pack-years) were weaker, showing the importance of recent exposure in relation to adenoma risk.

Genotype distributions among controls were consistent with Hardy-Weinberg equilibrium (P > 0.05 for all SNPs). None of the studied polymorphisms were independently associated with risk of advanced adenoma (Table 2). After considering smoking status (in terms of nonsmoker, long-term quitter, and current or recent smoker), all three XPC SNPs were shown to modify smoking-related risks (P_interaction ranging from 0.003 to 0.04; Table 3).

Trends for increased risks associated with current or recent smoking were statistically significant only for carriers of XPC 492_RR, XPC 499_AA, and XPC 939_KQ+QQ (all P_trend < 0.0001) with risks increasing to 2.6 (95% CI, 1.9-3.6), 3.0 (95% CI, 2.0-4.5), and 2.0 (95% CI, 1.3-3.0), respectively. In stratified analyses among current and recent smokers, we observed decreased risks of 0.4 (95% CI, 0.2-0.9) comparing carriers of the XPC 492_H allele to those homozygous for the XPC 492_R allele, and 0.6 (95% CI, 0.4-1.0) comparing carriers of the XPC 499_V allele to those homozygous for the XPC 499_A allele).

The three XPC SNPs were in strong linkage disequilibrium (D′ = 1.0 for all pairwise comparisons; specific r² in Fig. 2), yet, including all three SNPs in a logistic regression analysis in current and recent smokers, tended to show statistically independent risk relationships only for the XPC 499 polymorphism (P_trend = 0.06). Haplotype analysis inferred by the expectation-maximization algorithm showed that the three SNPs distributed in four common haplotypes (A, 40%; B, 36%; C, 18%; D, 5%; specifics by smoking subgroups in Table 4). As with the genotypes, haplotype analysis revealed no overall association with adenoma risk (P_{global permutation} = 0.6); however, again, risk differentials were noted among current and recent smokers (P_{global permutation} = 0.03) due largely to lower risks associated with haplotypes C and D, possibly driven by the above described genotype relationship with XPC 499_V.

No statistically significant differences were observed for the XPC genotype effects by age, gender, race, adenoma size, or histologic type (i.e., presence of high-grade dysplasia or villous characteristics). When analyses were restricted to Whites only (94%), all results were essentially the same.

We found no evidence that polymorphisms in seven-nucleotide excision repair genes had an overall effect on colorectal adenoma risk; however, smoking-related risk differentials for this disease varied in relation to carriage of three linked nonsynonymous polymorphisms in XPC (_R492_H, _A499_V, and _K939_Q), with protective effects most clearly associated with XPC 499_V.

XPC, located on chromosome 3p25, is a 33-kb-long gene containing 16 exons, encoding a 940-amino-acid protein involved in DNA damage recognition (15). The XPC SNPs evaluated in this study are located in exons 8 and 15 (9, 26, 27). Although the coding variants lead to amino acid substitution, their functional significance is uncertain; the 939_K allele was associated with reduced repair of single-strand breaks measured by alkaline Comet assay (11), although XPC _A499_V and XPC _K939_Q did not show functional correlations in the post-UV host cell reactivation-based complementation assay.

Two other common polymorphisms in XPC are of interest (Fig. 2). A common biallelic poly(AT) insertion (+) polymorphism in intron 9 has been associated with an increased risk of head and neck cancer (28) and is related to differential removal of UV photoproducts in the host-cell reactivation assay (10). A common C > A SNP 3′ of the XPC intron 11 splice acceptor site was related to increased formation of the XPC mRNA non-exon 12 isoform, resulting in diminished DNA repair (9).

These polymorphisms are in strong linkage disequilibrium with the _K939_Q SNP in exon 15 [939_Q variant was correlated with the poly(AT) insertion and −5 A splice acceptor site with D′ = 1.0 and r² = 0.9 for both; Fig. 2] in Caucasians (9), and, by the linkage disequilibrium relationships shown here, must also be partially linked to the XPC _R492_H and XPC _A499_V variants (939_Q variant was correlated in our study with the 492_R and 499_A alleles, r² = 0.04 and 0.2, respectively; refs. 13, 15).

The functional data on the XPC poly(AT) insertion (+) and −5 A acceptor site variants (9-11) are of further interest in relation to differential smoking-related risks of colorectal adenoma found in our study, as similar predictive effects were noted for smoking-related head and neck cancer (28) and lung cancer (29), and the tightly linked XPC 939_Q variant was associated with increased bladder cancer risk in one study (30), although not in another study (31).

Subjects in this study came from 10 different screening centers representing a broad population distribution in the United States. The cases and controls were comparable, randomly selected from the same source population, and screened by a standardized procedure (i.e., cases were not screened based on symptoms). The large study population allowed us to confine the analysis to cases with advanced adenoma, a particularly meaningful intermediate outcome with higher potential for malignant transformation. Although this fairly large sample size allowed us to explore genetic interrelationships with smoking-related adenoma risk, we had limited statistical power to make any confident conclusion. Reliable genotype data were ensured by high success and reproducibility rates. Our subjects, however, were largely Caucasians and better educated. As another consideration, in studying prevalent cases identified from the initial baseline screening, we may be oversampling slow-growing adenoma (the so-called length-sampling bias). As the PLCO Trial continues to identify additional prevalent adenomas as well as incident adenomas and colorectal cancers, we will have opportunities to address some of the limitations.

In summary, we found that three linked XPC alleles, 492_R, 499_A, and 939_Q, were associated with increased risk of colorectal adenoma in smokers, potentially related to functional changes due to amino acid substitution or to linkage with other functional polymorphisms in the gene. Our finding on adenoma and supporting data from some functional (9-11) and epidemiologic studies of smoking-related cancers (28-30) suggest a critical role of XPC in repairing DNA damage (e.g., bulky adducts) caused by tobacco carcinogens (e.g., polycyclic aromatic hydrocarbons, heterocyclic amines, nitrosamines, and aromatic amines). These results for XPC variants, tobacco use, and adenoma, along with the smoking-related adenoma risk modifications by variants of polycyclic aromatic hydrocarbon metabolism-related genes (e.g., EPHX1, NQO1, CYP1A1, and GSTT1) that we have previously observed in this population (20, 21, 32), support the importance of genetic susceptibility to tobacco-related colorectal carcinogenesis. Although we found no relation between other nucleotide excision repair genes and smoking-related adenoma risk, potential effects by unidentified genetic variants cannot be ruled out. Confirmation in other studies of colorectal tumors and investigations of other genes and variants in this pathway and other major DNA repair pathways are needed to clarify the underlying mechanisms of DNA repair in the development of colorectal cancer.

We thank Drs. Christine Berg and Philip Prorok (Division of Cancer Prevention, National Cancer Institute); the Screening Center investigators and staff or the PLCO Cancer Screening Trial; Tom Riley and staff (Information Management Services, Inc.); Barbara O'Brien and staff (Westat, Inc.); and Drs. Bill Kopp, Wen Shao, and staff (Science Applications International Corporation-Frederick) for their contributions to making this study possible.