The association of six non‐synonymous variants in three DNA repair genes with hepatocellular carcinoma risk: a meta‐analysis

Abstract Hepatocellular carcinoma is a complex polygenic disease. Despite the huge advances in genetic epidemiology, it still remains a challenge to unveil the genetic architecture of hepatocellular carcinoma. We, therefore, decided to meta‐analytically assess the association of six non‐synonymous coding variants from XRCC1,XRCC3 and XPD genes with hepatocellular carcinoma risk by pooling the results of 20 English articles. This meta‐analysis was conducted according to the PRISMA statement, and data collection was independently completed in duplicate. In overall analyses, the minor alleles of four variants, Arg280His (odds ratio, 95% confidence interval, P: 1.37, 1.13–1.66, 0.001), Thr241Met (1.93, 1.17–3.20, 0.011), Asp312Asn (1.22, 1.08–1.38, 0.001) and Lys751Gln (1.42, 1.02–1.97, 0.038), were associated with the significant risk for hepatocellular carcinoma. There were low probabilities of publication bias for all variants. Subgroup analyses revealed significant association of XRCC1 gene Arg399Gln with hepatocellular carcinoma in Chinese especially from south China (odds ratio, 95% confidence interval, P: 1.57, 1.16–2.14, 0.004), in larger studies (1.48, 1.11–1.98, 0.007) and in studies with population‐based controls (1.33, 1.06–1.68, 0.016). Taken together, our findings demonstrated that XPD gene Asp312Asn and XRCC1 gene Arg399Gln might be candidate susceptibility loci for hepatocellular carcinoma. Considering the ubiquity of genetic heterogeneity, further validation in a broad range of ethnic populations is warranted.


Introduction
Liver cancer is rapidly escalating to epidemic proportions worldwide, and it constitutes a major burden on individual and public health. Liver cancer is much more common in men than in women, and it is a common malignancy in Asia and Africa [1]. The most common type of liver cancer is hepatocellular carcinoma, which accounts for more than 70% of cases [2]. Despite the huge advances in genetic epidemiology, to unveil the genetic architecture of hepatocellular carcinoma still remains a challenging task, as apparent inconsistency from many individual underpowered studies has clouded the true effect of culprit genotype [3,4]. A practical and popular way is to pool the results of these individual studies to generate a more reliable estimate, just as a meta-analysis does.
DNA damage is widely recognized as a major primary cause of cancer, and without correct repair, such damage can give rise to genetic instability and eventually carcinogenesis [5,6]. Individuals differ greatly in their ability to repair DNA damage, and it is of clinical importance to interrogate DNA damage repair genes to account for this interindividual difference. The task of moving from the selection of candidate DNA repair genes to the enumeration of causal variants poses a challenge. It represents a useful shortcut to interrogate a non-synonymous coding variant with strong biological credentials in a compelling gene. To generate convincing information, we in this study decided to employ a meta-analytical method to assess the association of six widely evaluated non-synonymous coding variants from three DNA repair genes, that is, XRCC1 (gene ID: 7515), XRCC3 (gene ID: 7517) and XPD (gene ID: 2068), in susceptibility to hepatocellular carcinoma, by pooling the summarized data from 20 English articles from medical literature.

Search strategies
Identification of relevant articles investigating the association of XRCC1, XRCC3 or XPD genetic variant(s) with the risk for hepatocellular carcinoma was carried out through searching PubMed and Embase using the predefined key words: ('hepatocellular' or 'liver cancer' in Title) and ('XRCC*' or 'XPD' or 'DNA damage repair' or 'DNA repair' in Title/Abstract) and ('genotype' or 'allele' or 'polymorphism' or 'variant' or 'mutation' in Title/Abstract). The latest search checkpoint was on October 19, 2015. Each source of relevant articles was integrated into the EndNote X5 software (Thomson Table 1 The baseline characteristics of 20 qualified studies in this meta-analysis Reuters EndNote, Times Square, NY, USA) for reference management. In addition, the bibliographies of some major narrative reviews, original and meta-analytical articles were also abstracted to make sure there was no missing hit. The conduct of this meta-analysis was in agreement with the guidelines formulated in the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) statement (Table S1) [7].

Inclusion/exclusion criteria
Articles were included if one or more of the six variants (XRCC1 gene: Arg194Trp, Arg280His, Arg399Gln; XRCC3 gene: Thr241Met; XPD gene: Asp312Asn, Lys751Gln) were examined in association with hepatocellular carcinoma risk, and the count of each genotype was provided in both hepatocellular carcinoma patients and controls. In case of unavailable genotype counts, the article was includable if odds ratio (OR) and its 95% confidence interval (95% CI) were provided. Articles were excluded if they were conference abstracts or posters due to insufficient information or if they were editorials, narrative reviews, meta-analyses or lacking of control groups.

Data collection
Using a uniform Excel form, two authors (Yan-Hui Shi and Bin Wang) independently collected and cross-checked all necessary data from each qualified article, with any disagreement solved by a discussion. All necessary data included surname of the first author, year of publication, country where study samples collected and their acquisition time, sources of patients and controls, genotyping method, matched status, sample sizes, the genotype counts of six examined variants between the two groups and some baseline characteristics, including age, gender, smoking, drinking, hepatitis B and C virus infection, and family history of cancer.

Statistical analysis
Given the small number of mutant homozygous genotypes, the association of six examined variants with hepatocellular carcinoma risk was assessed under allelic and dominant models only. The random-effects model using the DerSimonian and Laird method was employed to compute weighted OR and its 95% CI. The inconsistency index (I 2 ) was used to quantify the magnitude of heterogeneity across studies, and the I 2 , which takes values from 0% to 100%, represents the percent of the observed variability that results from heterogeneity rather than chance. Practically, the I 2 of greater than 50% was indicative of significant heterogeneity.
To account for the potential sources of clinical heterogeneity, some stratified analyses were undertaken according to country, source of controls, sample size, genotyping method and matched status, respectively. Only subgroups with three or more independent studies were presented. In addition, a meta-regression analysis was used to detect other sources from some continuous confounders, including age, gender, smoking, drinking, hepatitis B and C infection, and family history of cancer. In addition, sensitivity analysis was performed by removing an individual study each time to check whether it can bias the overall estimate.
The probability of publication bias was assessed by the trim-and-fill method, which was used to assess the potential effect of missing studies with negative findings that may have had on the observed estimates.
The above analyses were completed by Stata software version 12.0 for Windows (StataCorp, College Station, TX, USA).
Using the trim-and-fill method, one study was reported to be missing for Arg280His under allelic model and for Asp312Asn under both models (Fig. 1). After taking missing studies into account, the magnitude of association was weakened, and still there was statistical In view of the limited number of qualified studies, stratified analyses were only undertaken for Arg399Gln (13 studies, Fig. 2) to seek possible sources of heterogeneity (Table 3). First, by country, the 399Gln allele was associated with the significant risk of hepatocellular carcinoma in Chinese subjects, especially from south China (OR, 95% CI, P: 1.57, 1.16-2.14, 0.004 under allelic model and 1.69, 1.18-2.42, 0.004 under dominant model), and no significance was found in subjects from the Indo-Pakistani region and France. Second, after grouping studies by total sample size at a cut-off of 500, there was no significance in smaller studies under both genetic models, but in larger studies, the 399Gln allele was associated with the significant risk of hepatocellular carcinoma under both allelic (OR, 95% CI, P: Further in meta-regression analysis, the percentages of hepatitis B virus infection in patients (P = 0.005) and controls (P = 0.008) were found to be significant confounders for the association between Arg399Gln variant and hepatocellular carcinoma risk.

Discussion
The aim of this comprehensive meta-analysis was to systematically assess the susceptible roles of six non-synonymous coding variants in three DNA repair genes from the English literature. The key findings of this meta-analysis are that XPD gene Asp312Asn was a candidate locus predisposing individuals to hepatocellular carcinoma, and for the widely evaluated variant Arg399Gln in XRCC1 gene, country, sample size, hepatitis B virus infection and source of controls might confound its association with hepatocellular carcinoma.
It is widely accepted that hepatocellular carcinoma is a multifactorial disease, and generally, carcinogenesis can be promoted by a single dominant mutation leading to expression of an oncogene. It is of importance to capture targeted genetic culprits responsible for the functional changes of these oncogenes. The interrogation of non-synonymous variants in coding region is a promising choice as they might change protein's function and consequently tailor susceptibility to disease [28]. To increase the chance of identifying susceptibility loci, we focused on six non-synonymous variants in three DNA repair genes, and 20 qualified studies systematically assessed their potential roles in susceptibility to hepatocellular carcinoma. Our overall analyses revealed that the mutation of Asp312Asn in XPD gene increased the risk of having hepatocellular carcinoma, irrespective of genetic models, and there was no indication of heterogeneity and publication bias. However, a note of caution should be sounded when interpreting this finding as only four studies were available for summarization, and moreover as we only analysed data in articles published in English language, the possible existence of publication bias cannot be excluded convincingly [29].
Although the overall association between XRCC1 gene Arg399Gln and hepatocellular carcinoma was negative, further stratified analyses revealed that country, sample size, hepatitis B virus infection and source of controls constituted the potential sources of heterogeneity. It is of particular interest to notice that Arg399Gln variant was associated with the significant risk of hepatocellular carcinoma in Chinese subjects, especially from south China, suggesting the existence of genetic heterogeneity. This association was deemed robust as significance remained after restricting analysis to larger studies only. Moreover, it is widely recognized that the coinfection of hepatitis B virus is an established risk factor for hepatocellular carcinoma, and this infection was unlikely able to cloud the susceptibility of Arg399Gln variant to hepatocellular carcinoma. On the basis of these observations, our meta-analytical findings suggested that XRCC1 gene Arg399Gln was another susceptibility locus for hepatocellular carcinoma in Chinese.
However, considering the limited number of qualified studies, this result cannot be directly extrapolated to the other ethnic groups, and further investigations in groups of other continents are timely required.
Several possible limitations should be acknowledged in this metaanalysis. First, only English articles were retrieved, and selection bias cannot be totally ruled out, although our trim-and-fill method revealed a low probability of publication bias. Second, only six non-synonymous variants from three DNA repair genes were meta-analysed, and other functional variants such as in the promoter regions of other relevant genes also deserved attention pending sufficient published data. Third, besides Arg399Gln, there were limited numbers of qualified studies for the other examined variants, which precluded further exploration on heterogeneity by using stratified and meta-regression analyses. Fourth, all involved studies were retrospective in nature, and it is intriguing to see whether the two significant variants were associated with the relapse, metastasis and survival in patients with hepatocellular carcinoma following hepatectomy.
In conclusion, we through a comprehensive meta-analysis demonstrated that XPD gene Asp312Asn and XRCC1 gene Arg399Gln might be candidate susceptibility loci for hepatocellular carcinoma. Considering the ubiquity of genetic heterogeneity and in view of small