XPG rs17655 G>C polymorphism associated with cancer risk: evidence from 60 studies

Xeroderma pigmentosum group G (XPG), a key component in nucleotide excision repair pathway, functions to cut DNA lesions during DNA repair. Genetic variations that alter DNA repair gene expression or function may decrease DNA repair ability and impair genome integrity, thereby predisposing to cancer. The association between XPG rs17655 G>C polymorphism and cancer risk has been investigated extensively, but the results remain contradictory. To get a more accurate conclusion, we performed a comprehensive meta-analysis of 60 case-control studies, involving 27,098 cancer cases and 30,535 healthy controls. Crude odds ratios (ORs) and 95% confidence interval (CIs) were calculated to determine the association of interest. Pooled analysis indicated that the XPG rs17655 G>C polymorphism increased the risk of overall cancer (CC vs. GG: OR=1.10, 95% CI=1.00-1.20; CG vs. GG: OR=1.06, 95% CI=1.02-1.11; CG+CC vs. GG: OR=1.07, 95% CI=1.02-1.12; C vs. G: OR=1.05, 95% CI=1.01-1.09). Stratification analysis by cancer type further showed that this polymorphism was associated with increased risk of gastric cancer and colorectal cancer. This meta-analysis indicated that the XPG gene rs17655 G>C polymorphism was associated with increased overall cancer risk, especially the risk of gastric cancer and colorectal cancer. Further validation experiments are needed to strength our conclusion.


INTRODUCTION
Cancer-related deaths continue to rise in both developed and developing countries. In 2012, there were about 14.1 million new cancer cases and 8.2 million cancerrelated deaths all over the world. Lung and breast cancer are the most common forms of cancer in human beings. Moreover, the incidences of liver, stomach and colorectal cancer are also very high in men and stomach, while cervix uteri and colorectal cancer prevail in women. Cancer is a complex disease. A variety of cancer risk factors have been recognized, such as smoking, drinking, lack of exercise, poor diet, reproductive changes, and genetic lesions [1]. Inherited genetic causations of cancer risk are mainly unidentified. Thus far, great effects have been made to AGING discover genetic variant alleles implicated in the crucial signaling pathways, which may influence individual cancer predisposition.
Genetic DNAs of living organisms are constantly subjected to various types of damages caused by environmental agents and byproducts (e.g., reactive oxygen species) of cellular metabolic processes. To maintain genome integrity, human beings possess a number of systems for the prevention and restoration of DNA damage. Reduced DNA repair ability is a predisposing factor to cancer [2]. Five common DNA repair pathways have been identified, including nucleotide excision repair (NER), base excision repair, double-strand DNA break repair, mismatch repair, and transcription coupled repair [3,4]. Among these pathways, NER is responsible for removing damaged DNA fragments (e.g., bulky adducts) resulting from radiation or chemical agents [5,6]. In the NER pathway, at least eight vital genes [excision repair cross-complementation group 1 (ERCC1), ERCC2/ Xeroderma pigmentosum group D (XPD), ERCC3/XPB, ERCC4/XPF, ERCC5/XPG, XPA, XPC and XPE/damaged DNA-binding protein 1 (DDB1)] have been well studied, which participate in DNA repair, capable of preserving genetic integrity to prevent cells from malignant transformation [7].
ERCC5/XPG is located on chromosome 13q22-33, consisting of 15 exons and 14 introns . Its protein product is a 1,186 amino acid structure-specific endonuclease, and plays an essential role in the two incision steps of NER [4,8]. XPG is highly polymorphic. Among known single nucleotide polymorphisms (SNPs) in this gene, a nonsynonymous Asp1104His (rs17655, G>C) polymorphism is most frequently studied for its association with cancer risk [2,. However the results are inconsistent from study AGING to study. Therefore, we performed this meta-analysis with all eligible publications to investigate the association between the XPG gene rs17655 G>C polymorphism and cancer risk.

Meta-analysis results
As we can see in Table 2 and Figure 2, significant between-study heterogeneity was detected under all the genetic models in the overall analysis. Thus, we used random-effect model. After calculating crude odds ratios (ORs) and 95% confidence interval (CIs), we found that XPG gene rs17655 G>C polymorphism was associated with increased overall cancer susceptibility

Publication Bias
Symmetry in the funnel plot (

DISCUSSION
In the current meta-analysis, we estimated the association between the XPG gene rs17655 G>C polymorphism and cancer risk based on 60 eligible case-control studies with a total of 27,098 cancer cases and 30,535 healthy controls. Pooled risk estimates revealed that this polymorphism was significantly associated with an increased risk of overall cancer, especially with the risk of gastric cancer and colorectal cancer.
The etiology of cancer is multifactorial [1]. Abnormal accumulation of DNA mutations caused by a variety of www.aging-us.com 1076 AGING  AGING factors might eventually trigger carcinogenic process [68]. Thus, properly repairing DNA damages in time to ensure genome stability and integrity is essential to prevent cancer. NER system includes two pathways: global genome repair and transcription-coupled repair, in both of which XPG plays a crucial role [6][7][8]. XPG gene, one of the eight vital genes in the NER pathway, is responsible for recognizing and excising DNA lesions on the 3' side [3,4]. Loads of SNPs have been identified in the XPG gene over the past decades, among which the rs17655 polymorphism has revoked great attention for its association with cancer risk. The rs17655 polymorphism, leading to the replacement of aspartate with histidine at codon 1104 in ERCC5 protein, may cause an alteration in the protein function, thereby likely affecting DNA repair ability, genome integrity, and cancer predisposition.
Numerous studies were performed to explore the association between the rs17655 polymorphism and the risk of various types of cancer. Feng et al. [22] carried out a study in 2016 to investigate the roles of three SNPs (rs2094258, rs751402 and ra17655) in the XPG gene, consisting of 177 patients and 237 controls. They found that the rs17655 polymorphism was associated with an increased risk of gastric cancer. This association was reconfirmed in different types of cancer, including breast cancer by Hsu et al. [29] with 401 cases and 531controls, colorectal carcinoma by Du et al. [20] with 878 cases and 884 controls, lung cancer by Chang et al. [17] with 255 cases and 280 controls, as well as cancer of other types. However, opposite results were also frequently reported. A population-based case-control study containing 196 gastric cases and 397 controls subjects conducted by Hussain et al. [31] revealed that the XPG rs17655 polymorphism might be associated with reduced gastric cancer risk. Additionally, Ruiz-Cosano et al. [68] reported that this polymorphism did not seem to play a major role in lymphoma susceptibility after studying 213 cases and 214 controls. Ma et al. [62] selected 320 cases and 294 controls and found that the rs17655 polymorphism might not confer susceptibility to breast cancer after adjusting for potential confounding factors. Several meta-analyses were also conducted, and unfortunately the results were still inconsistent [88][89][90][91]. As contradictory results were produced, we performed this meta-analysis to draw a more precise conclusion by including larger sample size and different cancer types from 60 studies. Our result indicated that this polymorphism may increase the risk of overall cancer, especially the risk of gastric cancer and colorectal cancer. The biological function of the rs17655 remains obscure. This polymorphism has been intensively studied for its association with cancer risk as AGING a tagger. It was predicated to be a harmful variant by a sequence homology-based tool [92]. Moreover, its functional potential was further confirmed by SIFT algorithms (scale invariant feature transform) and SNPs3D tools (http://compbio.cs.queensu.ca/F-SNP/) [93]; however, solid in vitro and in vivo data are needed to elucidate biological function of this variant.
There are advantages that strengthened the robustness of our findings. First, we searched five databases to include most of the publications written in English or Chinese. The large sample size provided adequate statistical power. Second, stratified analyses were performed by cancer type, quality score, and source of control. Third, we used the Begg's funnel plot and Egger's linear regression test to assess the possible publication bias.
However, several limitations still existed in this metaanalysis. Firstly, selection bias might occur because only publications written in English or Chinese were included. Researches in other languages were missed. Secondly, the number of individual studies for some cancer types, like HNC and prostate cancer (<5 studies), may be inadequate. Third, more than half of included studies had relative low quality scores (≤ 9). Our results should be interpreted cautiously. Further studies with high quality scores are needed to verify the real association.
Additionally, age, sex, living habits, virus infections or some environmental factors may also influence cancer risk. Our findings based on unadjusted estimates for lack of access to original data might suffer from potential confounding bias. Therefore, the results should be interpreted with caution. Finally, lack of biological evidence of the implication of the rs17655 polymorphism in cancer is also a drawback of the study. Mechanistic studies of the rs17655 polymorphism with cancer should be performed in the future.
In conclusion, this meta-analysis suggests that the XPG rs17655 G>C polymorphism is significantly associated with an increased overall cancer risk, especially with the risk of gastric cancer and colorectal cancer. Moreover, large-scale, well-designed studies in different cancers should be conducted to corroborate our findings.

Publication search
We searched for relevant articles using the following terms: "ERCC5 or XPG", "polymorphism or variant", and "cancer or carcinoma or neoplasm or malignance" in PubMed, EMBASE, CNKI, WANFANG, and Vip databases (the last search was performed on June 17, 2016). We also manually searched the references of the retrieved publications for additional relevant eligible studies.

Inclusion and Exclusion criteria
The publications contained in the meta-analysis had to meet the following criteria: (1) the study was only written in English or Chinese; (2) the study investigated the association between the XPG gene rs17655 polymorphism and the risk of one or more types of cancer; (3) case-control study. If studies had overlapping subjects, the publication including the largest number of individuals were selected.
Exclusion criteria were as follows (1) the study did not report sufficient genotype data to calculate odds ratio (OR) and 95% confidence interval (CI); (2) the study included survival data only. (3) the genotype frequencies of the rs17655 G>C and other polymorphisms were deviated from Hardy-Weinberg equilibrium (HWE) in the controls.

Data Extraction and quality assessment
Two investigators (Chen SS and Zhao J) extracted the following information from each publication independently: first author, publication year, cancer type, country of origin, race, genotyping method, source of controls (hospital-based, population-based and mixed), the genotype counts of cases and controls for the rs17655 G>C polymorphism. We also calculated the score of each publication based on the quality score assessment as described before [94]. All contradictory information was discussed when necessary.

Statistical analysis
We evaluated crude ORs and 95% CIs to assess the association between XPG rs17655 G>C polymorphism and overall cancer risk under the homozygous (CC vs. GG), heterozygous (CG vs. GG), recessive (CC vs. CG+GG), dominant (CG+CC vs. GG), and allele contrast (C vs. G) models. We carried out stratification analyses by cancer type (if one cancer type were investigated in less than three studies, we termed this type as "others"), score (>9 and ≤9), and study design (if a study contained both hospital-based controls and population-based subjects, we termed the study design as "mixed"). We also calculated between-study heterogeneity using the Chi square-based Q-test. When P>0.1 indicating lack of heterogeneity, a fixed-effect model was adopted. Otherwise, a random-effect model would be applied [94]. The potential publication bias AGING was evaluated by Begg's funnel plot [95] and Egger's linear regression test [96]. All of the P values were twotailed. P<0.05 was considered statistically significant. All data analyses were performed by the STATA software (Version 12.0; Stata Corporation, College Station, TX).