The association between XRCC3 rs1799794 polymorphism and cancer risk: a meta-analysis of 34 case–control studies

Studies on the XRCC3 rs1799794 polymorphism show that this polymorphism is involved in a variety of cancers, but its specific relationships or effects are not consistent. The purpose of this meta-analysis was to investigate the association between rs1799794 polymorphism and susceptibility to cancer. PubMed, Embase, the Cochrane Library, Web of Science, and Scopus were searched for eligible studies through June 11, 2019. All analyses were performed with Stata 14.0. Subgroup analyses were performed by cancer type, ethnicity, source of control, and detection method. A total of 37 studies with 23,537 cases and 30,649 controls were included in this meta-analysis. XRCC3 rs1799794 increased cancer risk in the dominant model and heterozygous model (GG + AG vs. AA: odds ratio [OR] = 1.04, 95% confidence interval [CI] = 1.00–1.08, P = 0.051; AG vs. AA: OR = 1.05, 95% CI = 1.00–1.01, P = 0.015). The existence of rs1799794 increased the risk of breast cancer and thyroid cancer, but reduced the risk of ovarian cancer. In addition, rs1799794 increased the risk of cancer in the Caucasian population. This meta-analysis confirms that XRCC3 rs1799794 is related to cancer risk, especially increased risk for breast cancer and thyroid cancer and reduced risk for ovarian cancer. However, well-designed large-scale studies are required to further evaluate the results.


Background
Cancer is the leading cause of death worldwide, and the number of patients with cancer is increasing [1]. The occurrence of cancer is related to many factors, including environmental, lifestyle, genetic and other factors. Among them, gene mutation is a kind of genetic factor, which has a great influence on cancer risk [2]. The mutation in BRCA1 and BRCA2 is related to the increase risk of breast cancer [3]. XPF rs2276466 polymorphism is related to neurogenic cancer [4].
X-ray repair cross-complementing group 3 (XRCC3), functions in the homologous recombination (HR) repair of DNA crosslinks [5] and double-strand breaks [6]. Based on the function of XRCC3, XRCC3 gene mutations are related to the occurrence and development of many diseases. For example, XRCC3 241Thr/Met genotype promotes left ventricular hypertrophy by inhibiting DNA damage repair [7]. Mutations in the XRCC3 gene affect mitochondrial DNA integrity [8]. XRCC3 rs861539 polymorphism is associated with poor prognosis of breast cancer patients [9]. The mutation sites that have been studied more about the relationship between XRCC3 gene and cancer are rs861539, rs1799794 and rs1799796 [10]. However, results remain fairly conflicting rather than conclusive. A number of meta-analyses have investigated the relationship between rs861539 and susceptibility to various cancers . However, there have been few meta-studies on rs1799794 and susceptibility to cancer [28,30,31,33,34]. Therefore, we conducted this meta-analysis to analyze the relationship between rs1799794 and susceptibility to cancer on the basis of more data.

Search strategies
We comprehensively searched five databases (PubMed, Embase, the Cochrane Library, Web of Science, and Scopus) for research published as of June 11, 2019, using relevant MeSH terms and entry terms. The keywords of XRCC3 included X-ray repair cross complementing 3, rs1799794, 4541A/G, XRCC3. The MeSH term and entry terms of polymorphism were genetic polymorphism [MeSH terms]; polymorphisms, genetic; genetic polymorphisms; genetic polymorphism; polymorphism (genetics); polymorphisms (genetics); polymorphism, single nucleotide; nucleotide polymorphism, single; nucleotide polymorphisms, single; polymorphisms, single nucleotide; single nucleotide polymorphisms; polymorphisms; polymorphism; variant; mutation; single nucleotide polymorphism; SNP. The MeSH term and entry terms of cancer were neoplasm [MeSH terms], neoplasms, neoplasia, neoplasias, neoplasm, tumors, tumor, cancer, cancers, carcinoma, carcinogenesis, tumour. Furthermore, we refined the search results of related studies by looking at the list of references included in each article.

Selection criteria
Relevant studies were included in accordance with the inclusion criteria and exclusion criteria, which were similar to those described in the previous study (PMID: 30867406). Original case-control study focused on the relationship between rs1799794 and cancer risk with the frequency of XRCC3 rs1799794 mutant genotypes were included. While conference abstracts or reports, reviews or meta-analyses, republished articles, and studies with insufficient data were excluded.

Data extraction and quality assessment
The following data from each selected article were collected: the surname of the first author, the publication year, country, ethnicity, cancer types, and methods of genotyping XRCC3 rs1799794 polymorphism. The quality of eligible case-control studies was estimated using the Newcastle-Ottawa Scale [35].

Statistical analysis
The relationship between XRCC3 rs1799794 polymorphisms and cancer risk were evaluated using odds ratios (ORs) and 95% confidence intervals (CI) under five genetic models (G vs. A, GG vs. AA, GG + GA vs. AA, GG vs. GA + AA, GA vs. AA).as previous study. If P < 0.05 or the 95% CI did not include 1, the result was considered statistically significant. Cochran's Q with chi-square (with P Q ) and the Higgins I 2 test were used to determine heterogeneity in between-study variability. When P Q < 0.1 or I 2 > 25% indicated significant heterogeneity [36][37][38], we analyzed the data using a random effects model [39]. If the opposite held, a fixed effects model was chosen. We also performed subgroup analyses and a sensitivity analysis to explore sources of heterogeneity. Subgroup analyses stratified studies by cancer type (ovarian cancer, acute lymphoblastic leukemia, breast cancer, thyroid cancer, bladder cancer, lung cancer, other), ethnicity (Arabian, Asian, Caucasian, mixed), sample size (< 100, > 100), the publication year (≤ 2010, > 2010), detection method (PCR-RFLP, sequencing, TaqMan, PCR, ND, other), and source of control (HB, PB, mixed, nested). We assessed publication bias using funnel plots and Egger's test (P < 0.05). Statistical calculations were performed with Stata 14.0.

Literature search and study characteristics
Finally, 3,467 potentially relevant published works were identified (997 in PubMed, 27 in the Cochrane library, 855 in Embase, 696 in Scopus, and 889 in Web of Science). Of these, duplicates (1959) and works not related to cancer and rs1799794 polymorphism (1451) were excluded. Then 23 of these studies were excluded after reviewing full texts. The remaining 37 works (43 studies) were included in this meta-analysis [10,. Because two studies in Auranen et al. [10] were duplicated in Quaye et al. [62], we only extracted data from these studies from Auranen et al. [10] to avoid duplication; thus, one article included four studies [66], and three articles included two studies each [10,68,70]. The flow chart of the literature selection process is shown in Fig. 1.

Meta-analysis and subgroup analyses
The value of I 2 in the five genetic models was greater than 25%, and P Q < 0.10, so pooled ORs for the five genetic models were calculated with a random effects model. There was no obvious correlation between rs1799794 and cancer risk (P Z > 0.05; Table 2).
Subgroup analyses were then performed based on cancer type, ethnicity, detection method, the publication year, source of control, and sample size to investigate sources of heterogeneity (Table 3). In the subgroup analysis based on cancer type, a significantly increased risk for thyroid cancer was observed in the five models (G vs. In the subgroup analysis based on ethnicity, rs1799794 was associated with increased cancer risk in the Caucasian population according to the heterozygous model (AG vs. AA: OR = 1.05, 95% CI = 1.01-1.10, I 2 = 0.0%).

Sensitivity analysis
Eight studies [41, 42, 48-50, 53, 54, 56] had P HWE < 0.05, but for two studies [51, 63] P HWE was not available. We compared the combined results before and after excluding these 10 studies and there were slight changes in the results. When the subgroup analysis was performed according to cancer type, there were no significant associations between rs1799794 polymorphism and increased risk for thyroid cancer in the recessive model,

Discussion
Our study shows that XRCC3 rs1799794 is irrelevant to cancer risk. In addition, the risk for thyroid cancer and breast cancer increase significantly in patients with rs1799794, and Caucasian populations are more likely to develop these cancers while having a decreased risk for ovarian cancer. We excluded articles that did not conform to HWE and reanalyzed the data. Compared to the previous results, rs3116496 was related to a decreased risk for lung cancer in the five models, although the other results were not much changed (data not shown).
Moderate heterogeneity was found in this meta-analysis. First, we used random models when significant heterogeneity. Second, we performed subgroup analyses to explore sources of heterogeneity. As shown in Table 3, in the subgroup analysis based on ethnicity, heterogeneity increased in Arabian/Asian populations but was 0% in Caucasian populations, which suggests that ethnicity may be a factor in heterogeneity. Furthermore, we analyzed studies stratified by cancer type, detection method, source of control, and sample size. Ethnicity, cancer type, source of control, and sample size may be the source of inter-research heterogeneity. In addition, a sensitivity analysis suggested that the current findings were reliable.
To date, five meta-analyses of the impact of rs1799794 on cancer risk have been performed [28,30,31,33,34] on rs1799794 and susceptibility to pan-cancer [28], breast cancer [30,34], bladder cancer [33], and ovarian cancer [31]. To the best of our knowledge, ours is currently the most comprehensive meta-analysis of correlations between rs1799794 polymorphisms and cancer. There are many differences between the results of this study and previous studies. According to Qiu et al. 's research on rs1799794 and susceptibility to breast cancer, which included four studies in three papers, rs1799794 was associated with a statistically significant increase in cancer risk in the dominant model (GG + AG vs. AA: OR = 1.09, 95% CI = 1.01-1.17, P H = 0.15), whereas our results showed an increased risk for breast cancer in AG carriers, different from the protective effect found previously [48]. In addition, our study found that the G allele might be a dominant gene and found an increased risk for thyroid cancer.
Our study included a large number of samples and conducted a stratified analysis, which played an important role in the reliability of the research results. At the same time, there are problems that cannot be ignored: the presence of heterogeneity that may due to ethnicity, source of control, status, or cancer type; the lack of relevant data published in other languages and evaluation of the interaction between cancer-related factors.

Conclusion
In conclusion, this meta-analysis found no association between XRCC3 rs1799794 and cancer risk, but XRCC3 rs1799794 was associated with breast cancer and thyroid cancer as well as with Caucasian    populations. In addition, detection method, source of control, and sample size played a role in heterogeneity and in the results. Well-designed large-scale studies are required to further evaluate the results.