Association between XRCC1 polymorphisms and the risk of cervical cancer: a meta-analysis based on 4895 subjects

The present meta-analysis was intended to explore the relationship between the X-ray repair cross complementing 1 (XRCC1) polymorphisms (Arg194Trp, Arg280His and Arg399Gln) and cervical cancer risk. Several electronic databases were searched systematically and bibliographies of relevant papers were identified carefully. Then, a meta-analysis was performed based on eligible studies in various genetic models. Pooled odds ratios (OR) with 95% confidence intervals (95% CI) were employed to evaluate the strength of associations between the XRCC1 polymorphisms and cervical cancer risk. Additionally, heterogeneity analysis and sensitivity analysis were done if necessary. Totally, 11 articles involving 2092 cases and 2803 controls were included. Taken together, there was no obvious association between the Arg194Trp or Arg280His polymorphism and cervical cancer risk. Considering the great heterogeneity, subgroup analysis was done, but the pooled result remained stable. Nevertheless, the association between the Arg399Gln polymorphism and cervical cancer risk showed distinct statistic significance in the allele model, dominant model, homozygous model and heterozygous model. In view of the exiting heterogeneity, we did subgroup analysis stratified by ethnicity, resulting in the fact that the Arg399Gln polymorphism was related to the decreased risk of cervical cancer. The Begg's test and Egger's test were used to find no publication bias. To conclude, the current meta-analysis indicated that the XRCC1 Arg399Gln polymorphism decreased the risk of cervical cancer, while the Arg194Trp and Arg280His polymorphisms were not associated with cervical caner risk. Certainly, a well-designed large-scale multicenter study is warranted to confirm the finding.


INTRODUCTION
Nowsdays, cervical cancer is one of the most common genital tract carcinomas and has become a challenging health issue confronted by women throughout the world. It seriously threatens women's quality of life arising from reproductive endocrine function's damage caused by this malignancy, and brings about great morbidity and economic burden. Infection with high-risk types of human papillomavirus (HPV) is the main causative factor for developing cervical intraepithelial neoplasia (CIN) which is a precursor lesion for cervical cancer. While, not all women who are infected with HPV will certainly progress into cervical cancer, suggesting that there are still other factors playing a role in the pathogenesis of cervical cancer. For example, ultraviolet, ionizing radiation and environmental chemical agents can lead to DNA damage, initiating certain human cancers [1][2][3][4][5][6][7][8][9].
In human body, DNA repair genes are considerable factors in the prevention of genomic injury and sequential carcinogenesis. So variants of DNA repair genes might be able to impair DNA repair ability and have been suggested to be associated with cancer risk. X-ray repair cross complementing group 1 ( XRCC1) gene is a typical DNA repair gene. It is located at chromosome 19q13. 2-13.3 and encodes the scaffolding protein [10]. The protein functions in the repair of single-strand breaks which is the most common lesions in cellular DNA [11]. Both biological and biochemical evidence indicate XRCC1 interacts with a complex of DNA repair proteins, such as poly(ADPribose) polymerase [11][12][13]. There are three most common polymorphisms in XRCC1, contributing to amino acid
A great many epidemiologic studies have been conducted to evaluate the role of the XRCC1 polymorphisms (Arg194Trp, Arg280His and Arg399Gln) on cervical cancer risk [14][15][16][17][18][19][20][21][22][23][24]. But the results were inconclusive. For example, Zhang et al. found the XRCC1 Arg194Trp polymorphism showed no significant association with CIN and squamous cervical carcinomas (SCC), while the Arg280His polymorphism acted as a protective factor for SCC, and the Arg399Gln polymorphism increased CIN risk among women who first gave birth before 22 years old [14]. Bajpai et al. suggested XRCC1 polymorphisms (Arg194Trp, Arg280His and Arg399Gln) increased cervical cancer risk greatly [23]. Barbisan et al. convinced that XRCC1 polymorphisms (Arg194Trp and Arg399Gln) genotypes and haplotypes contributed to reducing the risk of cervical cancer development in Argentin women [22]. Facing the contradictory, we assumed that a meta-analysis of various studies involving more subjects would offer a more precise conclusion. Thus, we aimed to obtain the summary risk estimating the association between the above mentioned three polymorphisms of XRCC1 and cervical cancer risk through a meta-analysis.
The included studies were all performed in recent years. The objects in eight studies were of Caucasians, two were of Asian and one was Mixed. Eight out of eleven control groups were population-based or healthy-based participants and the ramaining three were hospital-based. The largest number of subjects was 1339, almost 10-fold of the smallest number (n = 133). (Table 1) The quality assessment of included studies showed that all the studies were of high quality except that one study scored 5 points. (Table 2) Hardy-Weinberg equilibrium (HWE) examination results of the included studies and the XRCC1 polymorphisms genotype distribution in cases and controls were displayed in Table 3. All studies were consistent with HWE except for three studies for Arg194Trp [17, 21,22], one study for Arg280His [23], and one study for Arg 399Gln [24].

Meta-analysis results
For XRCC1 Arg194Trp polymorphism, there were seven studies, involving 1315 cases and 1633 controls, evaluating the connection between it and cervical cancer susceptibility. All the studies were done among the Asian population apart from one study [22]. Overall, there was no obvious statistic significance between the polymorphism and cervical cancer in all five models (P > 0.05). considering the moderate to great heterogeneity among studies, we performed subgroup analysis stratified by the degree of cervical lesion. However, the finding that the pooled OR still incorporated 1.0 showed that the Arg194Trp polymorphism had no association with the risk of cervical cancer. Then we excluded three studies which were not consistent with HWE [17, 22,23] and reassessed the relationship between this locus and cervical cancer risk. The final results did not change substantially. (Table 4).
With regard to XRCC1 Arg280His polymorphism, four articles including 2015 objects (784 cases and 1231 controls) offered data about the association between it and cervical cancer risk. On the whole, the heterogeneity among studies were quite huge, the random model was employed to weigh the strength of the association. While the remarkable link between this genetic locus and cervical cancer wasn't witnessed in all models (P > 0.05). However, the heterogeneity among studies droped to zero when excluding the study which didn't conform to HWE. Despite of this, the pooled results stayed stable when we eliminated the one [23]. (Table 4) In terms of XRCC1 Arg399Gln polymorphism, ten studies involving 1635 cancer patients and 2361 controls presented available data about this locus and cervical cancer risk. The Arg399Gln polymorphism decreased cervical cancer susceptibility in four genetic models: allele model were in accordance with HWE except one study [24]. The trend of summary ORs remained stable after excluding the one. The subgroup analysis stratified by ethnicity revealed that there still exited obvious association between this polymorphism and decreased cervical cancer among the Asian (Gln vs. Arg: OR = 0.34, 95% CI = 0. 26

Detection for heterogeneity
Considering the great heterogeneity among studies, the random-effect model was applied and subgroup analysis stratified by ethnicity ( Figure 6) and the degree of cervical lesion was performed. Nevertheless, the comprehensive results stayed stable. Furthermore, the meta-regression of ethnicity showed no obvious difference (P > 0.05), implying that the ethnicity exerted no influence on the association between the XRCC1 Arg399Gln polymorphism and the risk of cervical cancer.

Sensitivity analysis
Although some studies wasn't consistent with the balance of HWE in control groups (P < 0.05), yet the final results were not substantially altered after excluding those. Simultaneously, the studies with quite large or small sample sizes were deleted one by one in order to test the stability of pooled results. Moreover, sequential deletion of each study was utilized to perform sensitivity analysis in all models. However, the pooled ORs did not show quantitative changes when excluding any study, suggesting that the results of this meta-analysis were stable and reliable. Sensitivity analysis of the association between the XRCC1 Arg399Gln polymorphism and the risk of cervical cancer in homozygous model was showed Figure 7.

Publication bias
The Begg's test and Egger's test were done in all models showing that there was no statistical evidence for publication bias. Publication bias of the XRCC1 Arg399Gln polymorphism in homozygous model was shown in Figure 8 (P > 0.05).

DISCUSSION
Cervical cancer is still the second most common malignant tumor among women and heavily threatens women's health in the world. To improve this embarrassing situation, risk factors concerning cervical cancer should be indentified timely and controlled effectively. There  18,21,24,26]. However, the results remained conflicting rather than conclusive. Because a single study may have been underpowered to detect the effect of XRCC1 polymorphisms on cervical cancer risk, yet a quantitative synthesis of accumulative data from all available studies may provide convincing evidence. So a meta-analysis of ten available studies involving 2092 cervical cancer cases and 2803 controls was performed, expecting to derive a more precise estimation of the association between the XRCC1 polymorphism and cervical cancer susceptibility. Our results showed that there was no obvious association between XRCC1 Arg194Trp or Arg280His and cervical cancer susceptibility. Although we did subgroup analysis and sensitivity analysis, the trend of pooled results still remained identical, suggesting that the comprehensive results were quite stable. As to the Arg399Gln polymorphism, it reduced the risk of cervical cancer sharply. Likewise, we performed subgroup analysis and sensitivity analysis, the summary results still hinted a positive relationship between the Arg399Gln polymorphism and the decreased risk of cervical cancer.
Certainly, there have emerged several other meta-analysises concerning the link between XRCC1 polymorphisms and cervical cancer risk. A latest metaanalysis exploring the association between the Arg399Gln polymorphism and cervical cancer showed that the Arg399Gln polymorphism increased the risk of cervical cancer [27]. The result contradicted ours and the reasons may include the following. On the one hand, the number of databases we searched was bigger, resulting in more available studies in English were included. On the another, the quantity of subjects involving in present meta-analysis was greater, which surely strengthened the persuasive power of this research. Another meta-analysis noted that the Arg399Gln polymorphism elevated the risk of cervical  [24] 2013 Saudi Arabia Asian Hospital-based PCR Bajpai et al. [23] 2016 India Asian Hospital-based PCR-RFLP Barbisan et al. [22] 2011 Argentina Caucasian Population-basd PCR Djansugurova et al. [21] 2013 Kazakhstan Asian Healthy-based PCR Huang et al. [20] 2007 China Asian Population-basd MA-PCR Rozak et al. [18] 2011 Poland Caucasian Hospital-based PCR-RFLP Niwa et al. [19] 2005 Japan Asian Healthy-based PCR Setthetham-Ishida et al. [17] 2011 Thailand Asian Healthy-based PCR-RFLP Wang et al. [16] 2009 Costa Rica Mixed Population-basd Taqman Wu et al. [15] 2004 Taiwan Asian Population-basd PCR-RFLP Zhang et al. [14] 2012 China Asian Healthy-based PCR PCR: polymerase chain reaction; RFLP: restriction fragment length polymorphism. cancer in Chinese population [28]. However, the number of included studies was seven, less than the present one. Moreover, it only included the Chinese population, which undoubtedly weakened the strength of the conclusion. Li et al. held that the Arg194Trp polymorphism increased the risk of cervical cancer, while there was no association between the Arg399Gln or Arg280His polymorphism and cervical cancer risk [29]. But the meta-analysis was done five years ago and the number of databases was less than the present, which may explain the discrepancy in the results. Mei et al. showed that the Arg194Trp polymorphism increased the risk of cervical cancer and the Arg399Gln polymorphism elevated the risk of cervical cancer only in Asian population, while there was no association between the Arg280His polymorphism and cervical cancer risk [30]. While the meta-analysis was performed based on only two databases and it included studies without language limits, which may account for the distinction. As you see, the previous meta-analysises either focused on only one polymorphism or only one race or included fewer studies. Yet the present meta-analysis involved all studies of moderate to high quality according to prescribed inclusion and exclusion criteria, so the strength of this study was stronger than those past studies. Simultaneously, even though we performed subgroup analysis and sensitivity analysis, the pooled results still remained stable, supporting that this study was of great credit and persuasiveness. Likewise, some limitations of this meta-analysis should be mentioned even though considerable effort and resources have been put into testing the possible association between the XRCC1 polymorphism and cervical cancer risk. On the one hand, we retrieved relevant articles only through electronic databases, leading to a potential bias caused by the lack of unpublished articles which would not be available in the electronic databases. On the other, although the great heterogeneity among studies had no effect on the pooled result, yet the heterogeneity could not be neglected completely.
To conclude, the current meta-analysis indicated that the XRCC1 Arg399Gln polymorphism decreased the risk of cervical cancer, while the Arg194Trp and Arg280His polymorphisms were not associated with cervical caner risk. Certainly, to further evaluate the association between

Inclusion and exclusion criteria
The included studies have to meet the following criteria: 1) investigating the association between XRCC1 polymorphisms (Arg194Trp, Arg280His and Arg399Gln) and risk of cervical cancer; 2) studies on human beings; 3) genotype frequencies were available both in case and control groups; 4) subjects in control groups should have no cancer history, previous radiotherapy and chemotherapy history and a family history of tumor; 5) the diagnosis of the cases was based on pathology. The study with the following criteria was excluded: 1) abstracts, case reports, letters, comments, editorials, reviews and mata-analysises; 2) studies lacking relevant data. What's more, the most recent study was included once the studies were duplicated. Any one study was screened by two authors independently and disagreements were resolved by discussing with a third author.

Data extraction and synthesis
Two investigators simutaneously extracted characteristics of the included studies according to the inclusion and exclusion criteria and the results were checked by a third reviewer. The data extracted from each study included first author, year of publication, country of origin, ethnicity, source of the control group, genotyping method and numbers of case and control subjects. Ethnicity was classified as ''Caucasian'', ''Asian'' and ''Mixed''.

Quality assessment
The methodological quality assessment was performed based on the modified scoring system used for studies in genetic epidemiological issues. [31] Points were awarded on the basis of representativeness of cases, source of controls, HWE in controls, genotyping examination and association assessment. Total score ranged from 0 (lowest quality) to 8 (highest quality). A study with a score of 6 or higher was classified as high quality and vice versa.

Statistical analysis
Review Manage version 5.2.0 (The Cochrane Collaboration, 2012) and STATA version 11.0 software (StataCorp LP, College Station, TX) were applied to carry out statistical analysis. The association between XRCC1 polymorphisms and cervical cancer risk was estimated in the allele model, the dominant model, the recessive model, the homozygous genetic model and the heterozygous genetic model. To evaluate the strength of associations, the summary odds ratio (OR) and 95% confidence interval (CI) were calculated through fixed/random effects mode. P < 0.05 was considered statistically significant. To test the heterogeneity among studies, we assumed the I² and Q statistic. We adopted random effect model if there was great heterogeneity (I² greater than 50%). Otherwise, we adopted the fixed effect model. At the same time, we conducted a subgroup analysis according to ethnicities. To assess the stability of the finding, we performed sensitivity analysis. Each study involved in this meta-analysis was deleted respectively to reflect the influence of the individual data exerted on the pooled OR. HWE of the genotype frequencies in the control group of each study was assessed by χ 2 test and P > 0.05 was considered to be consistent with HWE [32]. For the studies which did not live up to HWE, we reassessed the association by eliminating them. The Begg's funnel plot and Egger's test were used to evaluate the possibly exiting publication bias [33,34].