XPG gene rs751402 C>T polymorphism and cancer risk: Evidence from 22 publications

The Xeroderma pigmentosum group G (XPG) gene promotes recognition and excision of damaged DNA during the DNA repair process. We conducted a comprehensive search of the MEDLINE, EMBASE, and Chinese Biomedical databases for publications evaluating the association XPG gene rs751402 C>T polymorphism and overall cancer risk. Pooled odds ratios (ORs) and 95% confidence intervals (CIs) were adopted to assess the strength of the association. A total of 22 publications encompassing 10538 cases and 10511 control subjects were included in the final meta-analysis. We found the polymorphism to be associated with increased cancer risk (TT vs. CC: OR = 1.18, 95% CI = 1.01–1.38, P = 0.040; CT vs. CC: OR = 1.12, 95% CI = 1.01–1.24, P = 0.040; and CT/TT vs. CC: OR = 1.12, 95% CI = 1.002–1.26, P = 0.045). Stratification by cancer type indicated that this polymorphism may increase the risk of gastric cancer and hepatocellular carcinoma, which was further confirmed by a falsepositive report probability analysis. Genotype-based mRNA expression provides further evidence that this polymorphism is associated with altered XPG mRNA expression. This meta-analysis suggests XPG gene rs751402 C>T polymorphism correlates with overall cancer risk, especially for gastric cancer and hepatocellular carcinoma.


INTRODUCTION
According to an estimation by GLOBOCAN, approximately 14.1 million new cancer cases, including 8.2 million deaths, occurred worldwide in 2012 [1]. Approximately 4,292,000 new cancer cases and 2,814,000 cancer deaths occurred in China in 2015, with lung, gastric, esophageal, and liver cancer being the most commonly diagnosed and the leading causes of death [2]. Risk factors for the leading causes of cancer-related deaths are tobacco consumption, overweight/obesity, physical inactivity, and infection [1]. Genetic factors should also be considered [3][4][5][6][7][8].
Human DNA repair genes maintain the integrity and stability of genomic DNA, consequently preventing carcinogenesis and influencing clinical outcomes [9,10].

Research Paper
Many genes promote the diverse DNA repair pathways, including the nucleotide excision repair (NER) pathway [11]. The NER pathway consists of damage recognition, demarcation, dual incision, and gap filling and can repair a variety of damaged DNA [12]. The NER pathway is the main mechanism for the removal of DNA adducts and lesions caused by chemical adducts [13]. Polymorphisms of the genes from the NER pathway might activate cancer risk alteration [14]. As one of the eight core genes in the NER pathway, Xeroderma pigmentosum group G (XPG), which is also known as excision repair cross-complementing group 5 (ERCC5), can recognize and excise DNA lesions on the 3′ side to repair damaged DNA [15].
XPG gene polymorphisms were reported to be associated with the susceptibility of various types of cancers [16][17][18]. Thus, most of the investigations were focused on rs17655 G>C (Asp1104His). The association between XPG gene rs751402 C>T polymorphism (located at the 5′ untranslational region) and cancer risk has been investigated in several studies , but the findings were contradictory and inconclusive. Therefore, we performed this meta-analysis with all eligible publications to comprehensively evaluate the association of XPG gene rs751402 C>T polymorphism with overall cancer risk.

Characteristics of eligible publications
As shown in Figure 1, 227 publications were identified from MEDLINE and EMBASE and 26 additional publications in Chinese were identified from the Chinese Biomedical (CBM) database. After reviewing the abstracts and the full texts, we excluded 222 publications and selected 31 publications with studies of the rs751402 C>T polymorphism for further full-text review. Among these publications, nine were excluded because two studies were repetitive, five studies were clinical outcome studies, and two studies were not on cancers. In the final meta-analysis, 22 publications with studies of 10588 cases and 10511 control subjects were identified, with the duplicated samples counted only once. The characteristics of the included publications are showed in Table 1. In these publications, sample sizes ranged from 96 to 1900 cases and from 101 to 1977 control subjects. Among the studies, 10 focused on gastric cancer [21, 23, 27, 29, 30, 32-34, 38, 39], three focused on breast cancer [25,35,36], two focused on hepatocellular carcinoma [20,37], and one each focused on lung cancer [19], oral squamous cell carcinoma [22], salivary gland tumor [24], nasopharyngeal carcinoma [26], neuroblastoma [28], colorectal cancer [31], and prostate cancer [40]. Of the publications, 12 had quality scores higher than nine, and 10 had quality scores of no more than nine.

Meta-analysis results
As shown in Table 2, significant heterogeneity was presented in all genetic models. As a result, we adopted a random-effect model for all the analyses. We found the XPG gene rs751402 C>T polymorphism associated with increased overall cancer risk (TT vs. CC: odds ratio [OR] = 1.18, 95% confidence interval [CI] =1.01-1.38; CT vs. CC: OR = 1.12, 95% CI = 1.01-1.24; and CT/TT vs. CC: OR = 1.12, 95% CI = 1.002-1.26). As shown in Figure  2, stratification analysis indicated that this polymorphism was associated with increased risk of gastric cancer (TT vs. CC: OR = 1.38, 95% CI = 1. 12 10-2.13). The stratification analysis did not reveal a significant difference between the two strata in any genetic model by quality score.

False-positive report probability analysis for significant findings
We performed false-positive report probability (FPRP) analysis for all significant findings and confirmed that the findings were significant at the priority of 0.1 for gastric cancer and hepatocellular carcinoma ( Table 3).
The genotype-based mRNA expression for XPG gene rs751402 C>T polymorphism As shown in Table 4, the rs751402T allele carriers were associated with decreased XPG mRNA expression among Asians (not significant), Africans (TT vs. CC: P = 0.029), and Caucasians (TT vs. CC: P = 0.013; and TT vs. CC/CT: P = 0.011), as well as all subjects (TT vs. CC: P = 0.010; and TT vs. CC/CT: P = 0.008).

Sensitivity analysis and publication bias
By omitting each publication once in every genetic model in the sensitivity analysis, we did not find any individual publication that could significantly alter the pooled ORs, which indicated that our data were stable and trustworthy. As shown in Figure 3, no obvious publication bias was observed for rs751402 C>T polymorphism (TT vs. CC: P = 0.111; CT vs. CC: P = 0.251; TT vs. CT/CC: P = 0.236; CT/TT vs. CC: P = 0.249; and T vs. C: P = 0.298).

DISCUSSION
In the current meta-analysis, we investigated all available publications that contained studies of the association between XPG gene rs751402 C>T polymorphism and cancer risk. The pooled results suggest that this polymorphism is associated with increased cancer risk, especially for gastric cancer and hepatocellular carcinoma.
The XPG gene, which is located at 13q33 and consists of 15 exons, promotes the removal of damaged DNA in the NER process [41]. When DNA repair capability is decreased, cells might fail to repair the damage. As DNA mutations accumulate, carcinoma might occur [9,21]. The XPG gene is an essential component of the NER pathway, and it activates the cleavage of DNA on the 3′ side of the lesion [42]. Studies reported that the XPG gene promotes cellular processes such as RNA polymerase II transcription and transcription-coupled DNA repair [43]. XPG gene polymorphisms might affect the expression or function of the XPG protein. Studies in several publications investigated the function of XPG gene rs751402 C>T polymorphism in cancer susceptibility. However, inconsistent results have been reported. Duan et al. [21] found that this polymorphism might increase the risk of gastric cancer in a study of 403 gastric cancer cases and 403 healthy control subjects. This association was also confirmed in gastric cancer by Yang et al. [38] in a study of 155 gastric cancer cases and 246 healthy control subjects, in hepatocellular carcinoma by Yoon et al. [20], and in oral squamous cell carcinoma by Zavras et al. [22]. Hua et al. [31] found that this polymorphism might be associated with decreased colorectal cancer susceptibility by studying 1901 colorectal cases and 1976 control subjects, and might have no effect in gastric cancer, as determined by 1142 cases and 1173 control subjects. Others found that this polymorphism might have weak effects on cancer susceptibility. The controversy can possibly be ascribed to the small sample size as well as cancer differences. To overcome the limitations of a single study and to reduce the likelihood of random errors being responsible for false-positive or false-negative associations, we performed the current meta-analysis to assess the association between XPG gene rs751402 C>T polymorphism and overall cancer susceptibility. We included 22 available publications, encompassing 10588 cases and 10511 control subjects, and found that this polymorphism was associated with increased overall cancer risk, especially for gastric cancer and hepatocellular carcinoma. We also performed FPRP analysis to confirm that the significant associations were trustworthy and robust. In addition, the genotype-based mRNA expression analysis as performed also indicated that this polymorphism might be associated with XPG gene mRNA expression alteration.
The current meta-analysis has five advantages. First, we searched the latest publications and we also included the publications written in Chinese. Second, we assessed the quality of each investigation and conducted stratification analysis by the quality score to search for publication bias. Third, we performed genotype-based mRNA expression analysis to provide further evidence that the rs751402 C>T polymorphism can influence the expression of the XPG gene. Fourth, we performed FPRP analysis, which can confirm whether the significant associations are trustworthy and robust. Fifth, we performed TSA to strengthen the robustness and minimize random errors of our conclusions.
Although in the present study we performed the latest and largest meta-analysis for assessing the association between XPG gene rs751402 C>T polymorphism and overall cancer susceptibility, four limitations must be considered. First, because of the heterogeneity in the current meta-analysis, the conclusions on the overall cancer risk should be interpreted cautiously. Second, the results of this study were based on the unadjusted ORs, which might suppress the final results. Third, all the study subjects were Asians. Other ethnicities are needed as subjects in future studies. Fourth, despite the adequacy of the total number of publications, the number of publications that contain studies for some cancers were inadequate. Investigations into other cancers are needed.
Our meta-analysis found that XPG gene rs751402 C>T polymorphism is associated with increased overall cancer risk, especially with respect to gastric cancer and hepatocellular carcinoma. Investigations of different cancers and ethnicities are needed to validate our findings.

Publication search
We systematically searched publications from the MEDLINE, EMBASE, and CBM databases (the last search was updated April 28, 2017) using the following search terms: "cancer or carcinoma or tumor or neoplasm," "excision repair cross-complementing group 5 or ERCC5 or xeroderma pigmentosum group G or XPG or rs751402," and "polymorphism or variant or single nucleotide polymorphism (SNP) or variation." We also manually searched the reference lists of the articles in the included publications.

Inclusion and exclusion criteria
The studies in the included publications met the following criteria: (1) the study evaluated the association between XPG gene rs751402 C>T polymorphism and cancer risk, (2) the study was on human beings, (3) the study was a case-control or cohort design, (4) sufficient data were provided to calculate the ORs and 95% CIs, and (5) the study was published in English or Chinese.
Exclusion criteria were (1) the study was not a casecontrol design, (2) the study was duplicated from previous studies, (3) articles were case reports or review articles, and (4) the studies were without detailed genotype data.

Data extraction and quality assessment
Two authors (Haixia Zhou and Ting-Yan Shi) performed the publication search and data extraction independently. The extracted information includes surname of the first author, publication year, cancer type, country of origin, ethnicity, genotyping methods, and numbers of cases and control subjects with rs751402 CC, CT and TT genotypes. We assessed the quality of each publication based on the quality score assessment [44]. All contradictory information was discussed and resolved through consensus when necessary.

Genotype-based mRNA expression analysis
To determine whether the XPG gene rs751402 C>T polymorphism can influence expression of the XPG gene, we conducted genotype-based mRNA expression analysis as previously described [3,45,46]. Genotype data of XPG gene rs751402 C>T polymorphism for 270 individuals were obtained from HapMap Phase II Release 23. The mRNA expression data for the corresponding individuals were from SNPexp [47].

Statistical analysis
Pooled ORs and 95% CIs were used to investigate the strength of the association between XPG gene rs751402 C>T polymorphism and overall cancer risk under the homozygous (TT vs. CC), heterozygous (CT vs. CC), recessive (TT vs. CT+CC), dominant (CT+TT vs. CC), and allele contrast (T vs. C) models. A goodnessof-fit χ 2 test was adopted to assess the Hardy-Weinberg equilibrium for the control subjects. Stratification analysis was carried out by cancer type (publications with no more than two were merged as the Others Group) and quality score (>9 and ≤9). Heterogeneities were assessed by χ 2based Q test, and a fixed-effect model was adopted when P > 0.1. Otherwise, the random-effect model was applied [48]. Sensitivity analysis was then conducted by omitting each publication in turn to evaluate the stability of the overall results. Potential publication bias was assessed by Begg's funnel plot [49] and Egger's linear regression test [50]. FPRP and TSA were as previously described [8]. All the statistics were two-sided, and P < 0.05 was statistically significant. All statistical analyses were performed by the STATA software (Version 11.0; Stata Corporation, College Station, TX).