Predictive value of XPG rs2296147T>C polymorphism on clinical outcomes of cancer patients

The Xeroderma pigmentosum complementation group G (XPG) rs2296147T>C polymorphism is suspected to associate with the clinical outcomes of cancer patients. However, the results are inconsistent. This meta-analysis aimed to evaluate the reliable predictive value of XPG rs2296147T>C polymorphism on clinical outcomes of cancer patients. A total of 11 eligible studies were enrolled in this meta-analysis. Our results indicated that the cancer patients with TT and CT genotypes were significantly associated with better respond rates when compared with the CC genotype (TT versus (vs.) CC: odds ratio (OR) = 2.05, 95% confidence intervals (CIs), 1.32-3.20, P = 0.002; TT+CT vs. CC: OR= 1.57, 95% CI, 1.14-2.17, P = 0.005). The TT genotype and/or T allele might be associated with higher survival time for cancer patients than the CC genotype and/or C allele. The cumulative meta-analyses showed an apparent beneficial objective response of TT genotype on cancer patients. In conclusion, this meta-analysis suggests that the XPG rs2296147T>C polymorphism is associated with the clinical outcomes of cancer patients. The XPG rs2296147T>C polymorphism might be a predictive factor of prognosis in cancers patients and contribute to individual treatment in the future.


INTRODUCTION
Nowadays, cancer has emerged as one of the most serious public health problems worldwide [1][2][3]. Despite intensive efforts have been made to improve the efficacy of cancer diagnosis and therapy, the overall survival (OS) time of cancer patients is still short [1][2][3]. It is very necessary to identify more reliable biomarkers for early diagnosis, accurate prognosis prediction, and efficacy for cancer patients [2]. Emerging evidence has demonstrated that genetic factors are considered to influence the cancer development, treatment effectiveness, survival time of cancer patients, therefore affect the prognosis of patients [3,4]. It has been speculated that DNA damage was significantly associated with the DNA repair capacity [5][6][7]. The genetic variants in DNA repair genes alter the activity of DNA repair, thus influence the effectiveness of therapy, prognosis and survival of patients [1,3,4,6]. The DNA repair genes have been identified in different DNA repair pathways [5][6][7]. The nucleotide excision repair (NER) pathway is the most versatile repair mechanism responsible for repairing bulky DNA damage [5]. The xeroderma pigmentosum complementation group G (XPG), also known as excision repair cross-complementation group 5(ERCC5), is one of the critical DNA repair enzymes of NER pathway. XPG gene is located on chromosome 13q32-q33, and encodes a protein of 1186 amino acids, which containing 15 exons and 14 introns. Previous studies indicate that the XPG rs2296147T>C polymorphism is suspected to have relationship with the clinical outcomes of cancer patients, such as colorectal cancer (CRC) [8][9][10][11], epithelial ovarian cancer (EOC) [12,13], head and neck cancer(HNC) [14], non-small cell lung cancer (NSCLC) [15][16][17][18][19], gastric cancer (GC) [20] and osteosarcoma (OC) [21,22]. Published data from these studies have shown inconsistent results. However, a systematic review and meta-analysis is still lacking. Thus, the aim of this meta-analysis was designed to summarize the currently available published findings and comprehensively assess the reliable predictive value of XPG rs2296147T>C polymorphism on clinical outcomes of cancer patients.

Studies characteristics
Our initial systematic search retrieved 139 publications using different combinations of key terms. Based on our inclusion and exclusion criteria, eleven eligible publications were ultimately enrolled for the data pool in this meta-analysis [8,9,11,[13][14][15][16][17][18][19]22] (Figure 1), altogether 5316 cancer patients. The general characteristics of enrolled studies are summarized in Table 1. These studies included five NSCLC studies, three CRC studies, one EOC study, one HNC study, and one OC study (Table 1). Ten studies were conducted on Asian patients, and one was conducted on Caucasian patients (Table 1). Eight studies were published in English and three studies were published in Chinese. Of these studies, five studies reported the objective response rate (ORR), eight studies reported the OS and hazard ratios (HRs), and seven studies reported the median progressionfree survival (PFS) and HRs ( Table 2). The sample sizes of included studies ranged from 105 to 1901 cancer patients. Several genotyping methods were used in the enrolled studies, including the polymerase chain reaction (PCR)ligase detection reaction (PCR-LDR), PCR-restriction fragment length polymorphism (PCR-RFLP), Taqman realtime-PCR (Taqman RT-PCR), and Illumina GoldenGate assay with Sentrix Array Matrix and 96-well standard microtiter plates (Table 1).

Overall survival of XPG rs2296147T>C polymorphism
There were eight eligible studies with 4983 cancer patients, qualified for the final analysis of the relationship between the OS of cancer patients and XPG rs2296147T>C polymorphism. We detected a significant association between the OS of cancer patients and XPG rs2296147T>C polymorphism under T vs. C genetic model (HR = 0.50, 95% CI, 0.38-0.66, P < 0.001, Table  3). The stratified analyses by cancer type showed that the XPG rs2296147T>C polymorphism was statistically significantly associated with the OS of cancer patients in NSCLC (TT vs. CC, HR = 0.44, 95% CI, 0.29-0.66, P < 0.001, Figure 3A; TT+CT vs. CC: HR = 0.66, 95% CI, 0.46-0.95, P = 0.024, Table 3). We also observed a significant association between the OS of EOC and XPG rs2296147T>C polymorphism under CC+CT vs. TT genetic model (HR = 0.50, 95% CI, 0.28-0.88, P = 0.017, Number of patients for ORR; in front of oblique line is good responder (complete response (CR) + partial response (PR)) and behind oblique line is poor responder (stable disease (SD) + progressive disease (PD)). b Number of patients for CT and TT genotypes. Table 3). Our results indicated that the TT genotype and/ or T allele might be associated with higher survival time for cancer patients than the CC genotype and/or C allele (Table 3).
No other significant associations between the XPG rs2296147T>C polymorphism and clinical outcomes of cancer patients were observed in this meta-analysis (Table 3).

Publication bias and sensitivity analysis
No obvious publication bias was found through either the inverted funnel plots or Begg's test (data not shown). In the sensitivity analysis, changing the effect models had no significant effects on the pooled ORs/HRs, and did not influence the final strength of association between XPG rs2296147T>C polymorphism and clinical outcomes of cancer patients. The influence of excluding each study on the pooled ORs/HRs was recalculated by repeating the metaanalysis, and the results indicated that excluding each study did not significantly change the overall effects, suggesting that our results are credible.

Cumulative meta-analyses
The cumulative meta-analyses based on year of publication showed an apparent beneficial objective response of TT genotype on cancer patients in the recent studies. As shown in Figure 4, between 2009 and 2014, five studies were enrolled, resulting in an overall effect estimate of OR being 1.26 (95% CI, 0.82-1.95, TT vs. CT+CC, Figure 4).

DISCUSSION
The NER is one of the major pathways of the DNA repair system, and the XPG gene is an indispensable component of the NER system. Many of SNPs in XPG gene have been reported to be associated with the outcomes of various cancers [5,. The association between XPG rs2296147T>C polymorphism and clinical outcomes of cancer patients has been widely studied [8,9,11,[13][14][15][16][17][18][19]22]. Bai et al observed that OC patients with XPG rs2296147T>C TT genotype were associated with better response to chemotherapy and longer OS than CC genotype (OR (95% CI) for ORR: 4.17 (1.64-11.54); HR (95% CI) for OS: 0.47 (0.17-0.94)) [22]. The XPG polymorphisms could be used as predictive markers for prognosis of OC [22]. Zhang et al found a significant decreased risk of death from NSCLC among patients carrying the XPG rs2296147T>C TT genotype when compared with the CC genotype (HR (95% CI) for OS: 0.50 (0.27-0.95)) [18]. The XPG rs2296147T>C polymorphism is associated with response to platinumbased chemotherapy in advanced NSCLC, and could be predictive of prognosis [18]. Yi et al indicated that NSCLC patients carrying XPG rs2296147T>C TT+CT genotype   [16]. Zou et al indicated that patients carrying the XPG rs2296147T>C TT genotype had a significantly reduced risk of developing progressive disease or dying from NSCLC (HR (95% CI) for PFS: 0.31 (0.13-0.73); HR (95% CI) for OS: 0.32 (0.14-0.71)) [15]. These studies suggest that the XPG rs2296147T>C polymorphism could be used as surrogate markers toward individualizing NSCLC treatment strategies [15,16,19]. However, He et al reported that there was no statistically significant association between XPG rs2296147T>C polymorphism and treatment response in NSCLC [17]. In addition, Chen et al suggested that no significant association was found between the XPG rs2296147T>C polymorphism and chemotherapy response in patients with advanced CRC [9,10]. Hu et al observed that the XPG rs2296147T>C C allele was associated with the PFS and OS of EOC compared with the TT genotype (HR (95% CI) for PFS: 0.63 (0.41~0.98); HR (95% CI) for OS: 0.50 (0.28~0.87)) [12]. The XPG rs2296147T>C polymorphism may play as the marker in predicting the clinical outcomes of EOC treated with platinum chemotherapy [12]. Wang reported that CRC patients carrying the XPG rs2296147T>C TT+CT genotype had a significantly shorter median 10 years PFS than those carrying CC genotype (HR = 1.324, 95% CI, 1.05-1.67) [8]. The XPG rs2296147T>C polymorphism could be predictive of unfavorable prognosis of CRC patients [8].
Results from these published studies remain conflicting rather than conclusive.
In the present study, we firstly conducted a systematic meta-analysis to assess the predictive value of XPG rs2296147T>C polymorphism on clinical outcomes of cancer patients. Results from this study could provide more reliable and precise comprehensive evaluation. Fortunately, we are delighted to find that the cancer patients carrying XPG rs2296147T>C T allele are associated with higher survival time and lower risks of death compared with those carrying C allele. Our cumulative meta-analysis indicated a distinct trend toward an apparent better respond rate for TT genotype of XPG rs2296147T>C polymorphism in cancer patients, which The detailed references are given in Table 1 and 2. b The OR for objective response rate. c The HR for overall survival and progression-free survival.  also showed the stable time-dependent trend ( Figure 4).
The sequential year-to-year cumulative meta-analyses could have made the evidence much clearer earlier. It may help planning further clinical trials, preventing many redundant trials, redirecting trial design, and leading to sufficient assessment of clinical outcomes. Thus, findings from this cumulative meta-analysis about the response rates of XPG rs2296147T>C polymorphism in cancer patients should be weighed with caution. Our findings suggested that the XPG rs2296147T>C polymorphism could be a predictive factor for clinical outcomes of human cancers. The XPG rs2296147T>C polymorphism might be used as a biomarker in personalized cancer treatment strategies. Some potential limitations in this meta-analysis should be noted when interpreting our findings. Firstly, although the sample size of our study was relatively large by collecting all enrolled studies, the statistical power was still limited in the stratified analyses because of the relatively small sample sizes. Some of the findings in stratified analyses may have been undervalued, because only one trial was available for analyses. Many important confounding factors were not taken into account in the stratified analyses, such as age, gender, ethnicities, cancer pathology types, cancer stage, cancer history, smoking habit, specific anti-cancer drugs, treatment regimens and test methods. These factors could contribute to the clinical outcomes of cancer patients. However, few of included studies provided sufficient data for stratified analyses, thus making such stratified analyses impossible. Secondly, some significant heterogeneity between studies did observed. However, this heterogeneity was not significantly change the main conclusions, because our findings reflected the most current state of this issue in published studies. Our analyses were based on unadjusted estimates, since not all published studies calculating adjusted estimates. Thirdly, all of included studies were performed in Asians (Chinese), excepting one study in Caucasian populations (Whites and African Americans). Fourthly, only published studies in Chinese and English were included, published studies in other languages, ongoing studies and unpublished data were not obtained. These may have caused some biases in our estimates. Therefore, to make the result more accurate and reliable, further studies should avoid these pitfalls.

CONCLUSIONS
In conclusion, despite the limitations, this meta-analysis indicates that the XPG rs2296147T>C polymorphism is associated with the clinical outcomes of cancer patients. Our findings suggest a predictive role for XPG rs2296147T>C polymorphism in clinical outcomes of cancer patients. The XPG rs2296147T>C polymorphism should be considered as a prognostic factor in human cancers. Further well-designed functional studies with large sample sizes in diverse ethnic populations will be necessary to validate these findings in the future.

Literature search strategies
We systematically retrieved the relevant literatures using the following search terms "XPG or Xeroderma pigmentosum complementation group G'', "ERCC5 or Excision repair cross-complementation group 5'', "rs2296147'', ''SNP or genetic polymorphism or variation'', and "cancer or tumour or tumor or neoplasm or carcinoma'' from the PubMed, ISI Web of Science, Excerpta Medica Database (EMBASE), Google Scholar, Wiley Online Library, ScienceDirect, Springer, VIP Database for Chinese Technical Periodicals, WANFANG Data, and Chinese National Knowledge Infrastructure (CNKI) databases (the last search was updated on May 20, 2016). We searched the articles without the restriction of country, race, languages and publication date. Reference lists were manually searched for further additionally relevant studies.

Inclusion and exclusion criteria
The inclusion criteria were as follows: (1). All patients diagnosed with cancer should be histologically confirmed; (2). XPG rs2296147T>C polymorphism should be genotyped; (3). Treatments and clinical outcomes (i.e., ORR, OS, PFS) should be reported; (4). Sufficient data should be provided to estimate relative risks (i.e., ORs and HRs with corresponding to 95% CIs) for prognostic effects of cancer patients; (5). Only full-text articles were enrolled. The exclusion criteria were utilized to exclude the literatures: (1). Abstracts, letters, comments, reviews and meta-analysis; (2). Duplicated studies; (3). Not reported sufficient data; (4). The corresponding authors were not provided the relevant information upon our request.

Data extraction
Two investigators (Yuhan Wang and Zhengrong Yuan) independently extracted data from the included studies. The discrepancies between investigators were discussed and solved with consensus from our team's decision. The following information was extracted: the first author's name, year of publication, country, ethnicities, cancer type, number of patients, median age (year), clinical stage, evaluation criterion, clinical outcomes (ORR, OS, PFS, time to progression (TTP), overall deaths (OD), disease-specific survival (DSS), HRs with corresponding to 95% CIs), genotyping methods, and the number of responders and non-responders in different genotypes.