Association of XPC Gene Polymorphisms with Colorectal Cancer Risk in a Southern Chinese Population: A Case-Control Study and Meta-Analysis

Xeroderma pigmentosum group C (XPC) is a key component of the nucleotide excision repair (NER) pathway. Dysfunctional XPC protein may impair NER-mediated DNA repair capacity and further lead to genomic instability and carcinogenesis. Two common nonsynonymous polymorphisms in the XPC gene, Lys939Gln (rs2228001 A > C) and Ala499Val (rs2228000 C > T), have been investigated in various types of cancer. We genotyped these two polymorphisms in 1141 cases with histologically confirmed colorectal cancer (CRC) and 1173 healthy controls to explore their causative association with CRC susceptibility. Overall, no association was observed between these two variants and the risk of CRC. Our meta-analysis also confirmed a lack of overall association. Stratified analyses were performed by age, gender, smoking status, pack-year, drinking status, tumor sites, and Duke’s stages. We found that XPC Lys939Gln polymorphism was significantly associated with an increased CRC risk in subjects at 57 years of age or younger (adjusted odds ratio (OR) = 1.37, 95% confidence interval (CI) = 1.004–1.86, p = 0.047) and non-drinkers (adjusted OR = 1.53, 95% CI = 1.10–2.12, p = 0.011). Our results indicated that XPC Lys939Gln may be a low-penetrance CRC susceptibility polymorphism. Our findings warrant further validation.


Introduction
DNA damage constantly occurs because of the exposure to endogenous and exogenous mutagens. There are several DNA repair mechanisms that extensively eliminate genetic errors to maintain genome integrity, including base excision repair, mismatch repair, double-strand break repair, and nucleotide excision repair (NER). Among these DNA repair pathways, NER is in charge of repairing bulky DNA lesions, such as pyrimidine dimers, UV-induced DNA damage, chemical adducts, and crosslinks [1,2]. There are at least four critical steps involved in the NER pathway: (a) recognition of damaged DNA via a protein complex containing xeroderma pigmentosum group C (XPC); (b) uncoiling of DNA mediated by the transcription factor II human (TFIIH) complex; (c) excision of the damaged single-stranded fragment; (d) gap-filling DNA synthesis and DNA ligation [2,3]. XPC is an important protein in the NER pathway; it enables DNA repair by complexing with RAD23 homolog B (RAD23B) to facilitate the recognition of DNA damage and the initiation of DNA repair [4][5][6]. The detection of damaged DNA is the rate-limiting step in the NER pathway [5]. Therefore, the function of XPC is critical for proper DNA repair. Polymorphisms in the XPC gene may influence the function of the protein and an individual's DNA repair capacity, and thereby affect genetic instability and modify individual predisposition to cancer.
The XPC gene, located on chromosome 3p25, is highly polymorphic. To date, at least 102 single nucleotide polymorphisms (SNPs) have been reported in the coding region of this gene [7]. Of these coding region SNPs, XPC Lys939Gln (rs2228001 A > C) and Ala499Val (rs2228000 C > T) polymorphisms have been foci of interest for their potential function, as well as association with the risk of various types of cancer, including breast cancer, lung cancer, bladder cancer, and colorectal cancer (CRC) [8]. A number of studies have been conducted to explore the association between these two polymorphisms and CRC susceptibility in different ethnicities which have shown some promising results [9][10][11]. However, apart from one study with 1028 cases [12] and the current study (1141 cases), the number of cases in the included studies was relatively small, ranging from 133 to 665. These studies might be limited by low statistical power; as a result, their conclusions might not be convincing. Moreover, such an association has not been evaluated in the southern Chinese population. With this in mind, we genotyped the two SNPs and assessed their association with CRC susceptibility in 1141 CRC cases and 1173 controls recruited from southern China.

Characteristics of Study Subjects
We analyzed in total 1141 CRC cases and 1173 healthy controls in the present study (Table 1). There was no significant difference between cases and controls (p = 0.518) when compared by the matching factor, gender. However, cases were significantly older than controls (55.7 ± 13.7 vs. 45.2 ± 11.6). The percentage of ever smokers in cases (27.3%) was significantly lower than that in controls (43.6%). The pack-years also differed significantly between cases and controls. Moreover, there were significantly fewer drinkers in cases than in controls. In this study, 44.3% (505) of cases had colon cancer, while 55.7% (636) had rectal cancer. In terms of the clinical stage, 130 (11.4%), 363 (31.8%), 359 (31.5%), and 289 (25.3%) patients were diagnosed with Duke's stage A, B, C, and D colorectal carcinoma, respectively.
Genotype frequency distributions of the two SNPs in controls were in agreement with the Hardy-Weinberg equilibrium (HWE) (p = 0.300 for rs2228001 A > C, and p = 0.095 for rs2228000 C > T). Overall, neither XPC Lys939Gln (rs2228001 A > C) nor Ala499Val (rs2228000 C > T) polymorphism was associated with the risk of CRC (Table 2). We next combined these two polymorphisms, which might increase the effect of individual polymorphisms and improve cancer risk prediction [13]. However, we found that carriers of one or two risk XPC genotypes had similar cancer risk to those carrying two wild-type genotypes (adjusted odds ratio (OR) = 1.13, 95% confidence interval (CI) = 0.77-1.66, p = 0.524 for carriers of one risk genotype; adjusted OR = 1.03, 95% CI = 0.69-1.54, p = 0.884 for carriers of two risk genotypes) ( Table 2).

Stratification Analysis
We performed a stratified analysis by age, gender, smoking status, pack-year, drinking status, tumor sites, and Duke's stages (Table 3). A stratified analysis by age indicated that XPC rs2228001 A > C polymorphism was significantly associated with an increased CRC risk in subjects at 57 years of age and younger (adjusted OR = 1.37, 95% CI = 1.004-1.86, p = 0.047). A stratified analysis also revealed increased CRC risk associated with rs2228001 for never smokers (OR = 1.42, 95% CI = 1.05-1.92, p = 0.024); however, the significance of the association disappeared after adjustment for age, gender, smoking and drinking status. Moreover, non-drinkers carrying rs2228001 AA/AC genotypes showed a significantly increased risk of CRC (adjusted OR = 1.53, 95% CI = 1.10-2.12, p = 0.011), which was likely to be due to chance. In term of tumor sites, XPC rs2228001 A > C did not seem to confer predisposition to either colon cancer or rectal cancer. XPC rs2228001 A > C polymorphism was not associated with Duke's stage, either. Moreover, no significant association with CRC risk was found for the rs2228000 C > T polymorphism in the stratified analysis.

Meta-Analysis Results
The two XPC polymorphisms have been frequently studied for their ability to modify CRC susceptibility. However, results remain open to more than one interpretation. We next carried out meta-analysis to further explore the association between these two polymorphisms and CRC risk by including eligible publications and our data (Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram). Overall, there were 12 publications consisting of 13 studies (6107 cases and 8207 controls) for the rs2228001 A > C polymorphism and five publications consisting of six eligible studies (3533 cases and 4860 controls) for the rs2228000 C > T polymorphism (Table 4). After pooling all studies together, risk estimates indicated a lack of correlation between the rs2228001 A>C polymorphism and CRC risk (Table 5 and Figure 1). Similarly, null association with CRC risk was found for the rs2228000C > T polymorphism in overall and stratified analyses (Table 5 and Figure 2).   (6107 cases and 8207 controls) for the rs2228001 A > C polymorphism and five publications consisting of six eligible studies (3533 cases and 4860 controls) for the rs2228000 C > T polymorphism (Table 4). After pooling all studies together, risk estimates indicated a lack of correlation between the rs2228001 A>C polymorphism and CRC risk ( Table 5 and Figure 1). Similarly, null association with CRC risk was found for the rs2228000C > T polymorphism in overall and stratified analyses (Table 5 and Figure 2).

Discussion
The association between two common nonsynonymous XPC polymorphisms (rs2228001 A > C and rs2228000 C > T) and the risk of CRC was evaluated in 1141 CRC cases and 1173 healthy controls enrolled from southern China. We found no statistical evidence of a significant association between these two polymorphisms and the risk of CRC. A combined analysis with these two polymorphisms suggested that subjects carrying one or two risk genotypes did not show increased CRC risk, when compared with non-carriers. Stratified analyses by age, gender, smoking status, pack-year, drinking status, tumor sites, and Duke's stages demonstrated a significant association between the XPC Figure 2. Forest plot of effect estimates for XPC rs2228000 C > T polymorphism with overall colorectal cancer risk by a recessive model. For each study, the estimates of OR and its 95% CI are plotted with a box and a horizontal line. ♦, pooled ORs and its 95% CIs.

Discussion
The association between two common nonsynonymous XPC polymorphisms (rs2228001 A > C and rs2228000 C > T) and the risk of CRC was evaluated in 1141 CRC cases and 1173 healthy controls Genes 2016, 7, 73 9 of 14 enrolled from southern China. We found no statistical evidence of a significant association between these two polymorphisms and the risk of CRC. A combined analysis with these two polymorphisms suggested that subjects carrying one or two risk genotypes did not show increased CRC risk, when compared with non-carriers. Stratified analyses by age, gender, smoking status, pack-year, drinking status, tumor sites, and Duke's stages demonstrated a significant association between the XPC rs2228001 variant allele and increased CRC risk in participants at 57 years of age or under, non-smokers, and non-drinkers. However, in the stratified analysis, age, smoking status, and drinking status had no effect on the association between the rs2228000 C > T polymorphism and the risk of CRC. The role of the XPC gene in carcinogenesis started to attract attention because of the observation that mutations in the DNA repair genes can cause the inherited disease xeroderma pigmentosum (XP). This disease is characterized by hypersensitivity to UV light due to impaired DNA repair capacity. XP patients have an approximately 1000-fold higher risk of developing multiple skin cancers [14]. XPC participates in NER by recognizing specific changes in the DNA structures, instead of the lesions themselves [15]. Upon recognition of damaged DNA, the XPC complex can interact with xeroderma pigmentosum group A (XPA) or TFIIH to enable the downstream steps of NER, including excision and restoration of DNA [16]. There are three widely investigated common polymorphisms in the XPC gene, XPC rs2228001 A > C, rs2228000 C > T [8], and a poly (AT) deletion/insertion on intron 9 [17]. The XPC rs2228001 A > C polymorphism, an A-to-C transition in exon 15, leads to a substitution of glutamine for lysine in codon 939 (Lys939Gln), while rs2228000 C > T polymorphism is a C-to-T transition in exon 8, resulting in the replacement of alanine with valine in codon 499 (Ala499Val) [18]. The XPC rs2228001 A > C polymorphism is located in the domain interacting with TFIIH. Both biological and biochemical evidence showed that XPC rs2228001 A > C polymorphism could alter the DNA repair capacity [19,20]. Moreover, lymphocytes from individuals with this XPC polymorphism were shown to have a reduced capacity of repairing benzo(a)pyrene DNA adducts [21].
Our findings were in line with some previous studies. A number of case-control studies have been carried out to investigate the association between XPC rs2228001 A > C polymorphism and the risk of CRC among Caucasians [22][23][24][25][26][27][28]. Despite biological plausibility, none of those investigations provided evidence of the association of this polymorphism with CRC susceptibility [22][23][24][25][26][27][28]. Several studies have also evaluated the role of XPC rs2228001 A > C polymorphism in CRC among Asians [11,12,29,30]. Wu et al. [29] reported that carriers of the CC genotype had a significantly increased risk of CRC in an eastern Chinese population with 421 CRC patients and 845 controls. Liu et al. [12] assessed the association in a northeastern Chinese population including 1028 CRC cases and 1085 controls. They found that the XPC rs2228001 A > C polymorphism was significantly associated with increased CRC risk under the heterozygous (OR = 1.40, 95% CI = 1.16-1.69) and dominant (OR = 1.31, 95% CI = 1.10-1.56, p = 0.001) genetic models. In contrast, Yue et al. [30] failed to repeat the reported association in Chinese patients in a study population of 428 CRC cases and 450 controls. The conflicting results regarding the association from different studies might be, in part, attributed to the variations in samplings, sample sizes, genotyping methods, genetic backgrounds, and ethnicities. It was also worth noting that we identified XPC rs2228001 A > C polymorphism as a CRC susceptibility variant in some subpopulations, including participants at 57 years of age and under, non-smokers, and non-drinkers, but these findings could lack causality. Taken together, these findings suggest that XPC rs2228001 A > C polymorphism may be a low-penetrance common polymorphism with a moderate effect on CRC susceptibility. Large, well-designed multicenter studies, with consideration of epidemiological characteristics and lifestyles, should be performed to clarify the role of XPC rs2228001 A > C polymorphism in CRC carcinogenesis.
When compared with the XPC rs2228001 A > C polymorphism, relatively fewer studies have examined the association of the rs2228000 C > T polymorphism with CRC susceptibility [25,[27][28][29]. The results from three studies were similar to ours [25,28,29]. Interestingly, Steck et al. [28] reported a significantly positive association between this polymorphism and CRC risk in African Americans, while another study demonstrated an association in the reverse direction in Caucasians [27].
Several meta-analyses on the association between these two polymorphisms and CRC risk have been published [8,10,31]. Since then, many relevant studies have been reported. Therefore, we also performed a meta-analysis to reevaluate the association of interest, with the addition of more studies. To date, the current meta-analysis was the largest pooled study on this topic and showed null association for these two polymorphisms.
Replication study is the golden method for the validation of the genetic associations. We tested the association of interest in the study with a relatively large sample size. As a result, the statistical power was largely increased in comparison with previous studies and the results were convincing. In the present study, the statistical power to detect a significant finding in younger subjects, never smokers, and never drinkers was 77.2%, 65.4%, and 47.9%, respectively. While using our sample size, the statistical power to detect the OR of 1.20 was approximately 70% in a recessive genetic model. Therefore, this study was important in defining genetic susceptibility to CRC. Given that the genetic backgrounds of populations may vary depending on regions and ethnicities, findings from other regions and races may not be applicable to the southern Chinese population. This study was the first one to evaluate the association in the southern Chinese population. Moreover, single case-control studies often yield conflicting results, partially due to the relatively small sample sizes of each study and variations in samplings, ethnicities, genotyping methods, and study designs (e.g., population-based vs. hospital-based) among studies. To overcome these drawbacks, we also conducted a meta-analysis on the associations of interest by pooling all eligible studies and ours together. The meta-analysis results verified our findings from the current case-control study.
However, several limitations of the studies were deserving of attention. First, the present study only had a moderate sample size. Especially, the sample size in the stratified analysis was even smaller and for this reason the statistical power might be limited. Second, we only analyzed two potentially functional polymorphisms in the XPC gene. More potentially functional polymorphisms in the XPC gene should be studied. Third, details on other risk factors, including lack of exercise, being overweight or obese, and high consumption of red meat, were not available for cases and controls, which prevented us from further investigating the effects of these parameters on the association. Fourth, the current study was a gender matched case-control study; age and other factors, such as smoking and drinking status, were not factored into the study design, which might lead to a selection bias. Fifth, an information collection bias might exist. For example, the cases had a lower percentage of individuals with a history of smoking or alcohol consumption than controls did in our current study. Sixth, 32.2% of the cases with colorectal cancer were 50 years of age or younger, which was a little higher than that (21.9%) in previous investigations performed in central China [32]. This discrepancy may be ascribed to geographical diversities. Moreover, some of our cases might have a family history of cancer. Finally, future in vitro and in vivo studies are needed to further provide biological evidence of the risk effects of XPC rs2228001 A > C polymorphism on the development of CRC and clarify the underlying mechanisms.
In conclusion, we did not find any predominant effects of the XPC rs2228001 A > C and rs2228000 C > T polymorphisms with CRC risk in the southern Chinese population. Our findings should be interpreted cautiously and warrant further validation in prospective investigations.

Study Population
We recruited 1141 cases with histologically confirmed CRC between January 2000 and May 2010 in Sun Yat-sen University Cancer Center. Patients with primary cancer in sites other than the colon and rectum were removed from the final study. Moreover, 1173 healthy residents in the same region were randomly recruited as controls in the same period of time, and they were frequency-matched to cases on gender. All the participants were of the ethnic Han Chinese population from southern China and genetically unrelated. Potential participants were contacted to screen for eligibility and the written informed consent was signed by each of the study subjects. A self-administered questionnaire was employed to collect information from individuals during a personal interview by their attending doctors, including demographic characteristics (age and gender), lifestyle (e.g., smoking habits and alcohol consumption), and family history of cancer. A blood sample of approximately 5 mL was obtained from each subject at the end of the interview. Overall, more than 80% of cases and controls responded. This investigation was approved by the institutional review board of Sun Yat-sen University Cancer Center (Guangzhou, China). Written informed consent was obtained from all patients.

Genotyping
Genotyping procedures were described elsewhere [33][34][35]. Briefly, genomic DNA was isolated from peripheral blood samples using the Qiagen Blood DNA Mini Kit (Qiagen Inc., Valencia, CA, USA). The Taqman method was chosen for genotyping these two polymorphisms with a 7900 HT sequence detector system (Applied Biosystems, Foster City, CA, USA). Quality control was strictly conducted, with four duplicate positive controls and four negative controls (omitting the DNA template) loaded in each of 384-well plates. Moreover, genotyping assays were repeated with 10% of the samples that were randomly selected and genotyping results were 100% concordant. Laboratory staff were blind to the case/control status of the samples.

Statistical Analysis
We performed Pearson's χ 2 test and the Student's t test to compare categorical variables and continuous variables between cases and controls, respectively. A goodness-of-fit χ 2 test was used to check the significant deviation of SNP genotype frequencies from HWE in the control subjects. An unconditional logistic regression analysis was conducted to calculate ORs and 95% CIs. An additional adjustment for age, gender, smoking and drinking status was made in the multivariate logistic regression analysis. We also carried out a stratified analysis by age, gender, pack-years, smoking and drinking status to study whether these factors could modify the association of interest. All analyses were performed using SAS software (Version 9.1; SAS Institute, Cary, NC, USA). All tests were two-sided. Results were considered statistically significant at p < 0.05.

Meta-Analysis
The meta-analyses were conducted on the association between the two SNPs (rs2228001 A > C and rs2228000 C > T) and CRC susceptibility following the PRISMA checklist [36]. Systematic literature searches of MEDLINE, EMBASE, and PubMed databases were conducted. The search terms were as follows: "Xeroderma pigmentosum group C or XPC", "Lys939Gln or rs2228001", "Ala499Val or rs2228000", "polymorphism or variant or variation", and "colorectal cancer or tumor or carcinoma or neoplasm or CRC". Literature searches were performed between October and November 2015. Detailed procedures referred to previous publications [8,[37][38][39][40]. Cross-study heterogeneity was determined by a chi-square-based Q-Test. The fixed-effects model (the Mantel-Haenszel method) [12] or the random-effects model (the DerSimonian and Laird method) [13] would be performed in the absence or presence of heterogeneity, respectively. Publication bias was measured by the funnel plot and the Egger's linear regression test [14]. Furthermore, a sensitivity analysis was also performed. We performed all analyses using STATA version 11.0 (StataCorporation, College Station, TX, USA).