Association between the BRCA2 rs144848 polymorphism and cancer susceptibility: a meta-analysis

The BRCA2 gene plays an important role in cancer carcinogenesis, and polymorphisms in this gene have been associated with cancer risk. The BRCA2 rs144848 polymorphism has been associated with several cancers, but results have been inconsistent. In the present study, a meta-analysis was performed to assess the association between the rs144848 polymorphism and cancer risk. Literature was searched from the databases of PubMed, Embase and Google Scholar before April 2016. The fixed or random effects model was used to calculate pooled odd ratios on the basis of heterogeneity. Meta-regression, sensitivity analysis, subgroup analysis and publication bias assessment were also performed using STATA 11.0 software according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses 2009. A total of 40 relevant studies from 30 publications including 34,911 cases and 48,329 controls were included in the final meta-analysis. Among them, 22 studies focused on breast cancer, seven on ovarian cancer, five on non-Hodgkin lymphoma, and the remaining six studies examined various other cancers. The meta-analysis results showed that there were significant associations between the rs144848 polymorphism and cancer risk in all genetic models. Stratified by cancer type, the rs144848 polymorphism was associated with non-Hodgkin lymphoma. Stratified by study design, the allele model was associated with breast cancer risk in population-based studies. The meta-analysis suggests that the BRCA2 rs144848 polymorphism may play a role in cancer risk. Further well-designed studies are warranted to confirm these results.


INTRODUCTION
Cancer is one of the most common diseases causing considerable morbidity and mortality worldwide. Environmental and genetic factors together contribute to the development of cancers [1][2][3][4]. It has been reported that DNA damage and repair is an important factor in carcinogenesis [5][6][7]. BRCA2 is a well-known cancer susceptibility gene involved in the repair of doublestranded DNA breaks which functions by regulating the intracellular shuttling and activity of RAD51, another critical protein in homologous recombination [8][9][10]. Studies have shown that cancer carcinogenesis is related to abnormalities in DNA repair mechanisms partially caused by a change in gene function which can result from genetic polymorphisms [11,12].
Within the last few years, many studies have focused on the association between BRCA2 gene polymorphisms and cancer risk, including breast cancer, ovarian cancer, non-Hodgkin lymphoma, prostate cancer and others [13-

Review
Oncotarget 39819 www.impactjournals.com/oncotarget 18]. The rs144848 is the only common non-synonymous polymorphism in exon 10 of the BRCA2 gene [19]. The change from A to C in the rs144848 polymorphism results in an asparagine-to-histidine transition (N372H) which may affect BRCA2 structure at residues 290-453, a region which has been determined to interact with the histone acetyltransferase P/CAF prior to transcriptional activation of target genes [20]. Over the past decade, many association studies have been conducted to explore the role of the rs144848 N372H polymorphism in cancer risk [13,15,17,18,[21][22][23][24][25][26][27][28][29][30][31][32][33][34][35][36][37][38][39][40], but it is still inconclusive whether this polymorphism in the BRCA2 gene is associated with susceptibility to cancer. Therefore, we performed a systematic review and meta-analysis of published studies focused on the association between the rs144848 polymorphism and cancer risk. Our in-depth analysis may drive a more precise estimation of risk which could in turn help identify additional genetic targets for future therapeutic interventions.

Study characteristics
A flow diagram for the search strategy is shown in Figure 1. Based on the search strategy, 2,174 articles were identified in the initial search. After reading titles and abstracts, 1,788 articles were excluded and 386 articles were reviewed for full text. According to the study inclusion/exclusion criteria, 40 relevant studies from 30 publications including 34,911 cases and 48,329 controls were used for the final meta-analysis [13-15, 17, 18, 21,

Association between BRCA2 rs144848 polymorphism and cancer risk
As shown in Table 2, there was no heterogeneity in any genetic model. The meta-analysis results showed that there were significant associations between the rs144848 polymorphism and cancer risk in all genetic models (H allele vs.

Meta-regression analysis
The following covariates were considered for metaregression: ethnicity, study design and cancer type. The results showed that cancer type contributed to effect in the meta-analysis (H allele vs. N allele, p = 0.011; HH vs. NN, p = 0.006; dominant model, p = 0.039; recessive model, p = 0.011).

Subgroup analysis by cancer type stratification
Based on cancer type, four groups were included in the meta-analysis: breast cancer group, ovarian cancer group, non-Hodgkin lymphoma group and other cancers group. The results showed that the rs144848 polymorphism was not associated with breast cancer or ovarian cancer in any model. However, the rs144848 polymorphism was associated with non-Hodgkin lymphoma in four models (H allele vs. N allele, OR = 1.110, 95% CI = 1.023-1.205, p = 0.012; HH vs. NN, OR = 1.263, 95% CI = 1.035-1.542, p = 0.022; dominant model, OR = 1.118, 95% CI = 1.008-1.240, p = 0.035; recessive model, OR = 1.216, 95% CI = 1.002-1.476, p = 0.048) and with other cancers in all genetic models (Table 3).

Association
between BRCA2 rs144848 polymorphism and breast cancer risk There were 22 breast cancer studies with different ethnicities and study designs. To assess the role of genetic background and the source of the control population in breast cancer risk, we carried out a subgroup analysis. In the analysis of genetic background, the overall population was divided into three subgroups, Caucasian, Asian, and African. The results showed that no statistically significant association was observed in any population (Table 4). In the analysis of study design, the overall population was divided into two subgroups, population-based studies and hospital-based studies. The results showed that the allele model was associated with the risk of breast cancer based on population-based studies (H allele vs. N allele, OR = 1.034, 95% CI = 1.000-1.068, p = 0.047; Table 5).

Sensitivity analysis
To determine the degree to which an individual study affected the overall OR estimates, one-way Debniak et al. [35] 2008   The results showed no influence on pooled ORs and 95% CIs as individual studies were excluded.

Publication bias
Publication bias was observed in only one model (H allele vs. N allele, p = 0.045; Table 2). However, there was no significant publication bias in any genetic model (p > 0.05) after sensitivity analysis. Trim and fill results showed that the adjusted risk estimate remained significant (H allele vs. N allele, OR = 1.028, 95% CI = 1.006-1.050, p = 0.014), which confirmed that the results of this metaanalysis were statistically robust.

DISCUSSION
The mechanisms underlying carcinogenesis are still not fully clear, but it has been suggested that genetic and environmental factors play the most important role in the development of cancer. The BRCA2 protein can regulate homologous recombination by interacting with the RAD51 recombinase, and many studies have suggested that the rs144848 polymorphism in the BRCA2 gene is a susceptibility locus for cancers [8]. However, until now, there has been no consistent result regarding the association between the rs144848 N372H polymorphism and cancer risk. To explain these contradictory results, a meta-analysis including 34,911 cases and 48,329 controls was conducted and five genetic models were utilized to assess the association between the BRCA2 rs144848 polymorphism and the risk of cancer.
In our meta-analysis, the results showed that there was no heterogeneity in any genetic model in overall population, while associations were observed between the rs144848 polymorphism and cancer risk in all genetic models. Meta-regression analysis suggested that ethnicity and study design had no influence on overall effect, but cancer type did contribute to effect (H allele vs. N allele, p = 0.011; HH vs. NN, p = 0.006; dominant model, p = 0.039; recessive model, p = 0.011). Based on cancer type, four groups were included in the metaanalysis: breast cancer group, ovarian cancer group, non-Hodgkin lymphoma group and other cancers group. The results showed that the rs144848 polymorphism was not associated with breast cancer or ovarian cancer in any model. However, the rs144848 polymorphism was associated with non-Hodgkin lymphoma in four models, and associated with other cancers in all genetic models.
The results showed a statistically significant association in all genetic models for overall population. Due to the relatively large number of research studies on  breast cancer, we also did a subgroup analysis in the breast cancer group. To assess the role of genetic background in breast cancer, we stratified the population by ethnicity and found no association in Caucasian, Asian, and African subgroups. Considering that the number of publications in Asian and African populations was small, we believe our Oncotarget 39826 www.impactjournals.com/oncotarget results may not be reliable due to insufficient statistical power, so additional studies should be conducted to confirm our results. However, after subgroup analysis by study design stratification, we found that the BRCA2 rs144848 N372H polymorphism was associated with increasing the risk of breast cancer in population-based studies (H allele vs. N allele, OR = 1.034, 95% CI = 1.000-1.068, p = 0.047). One-way sensitivity analysis suggested no influence of individual studies on pooled ORs and 95% CIs. In 2006, a study from the breast cancer association consortium summarized the common breast cancerassociated polymorphisms but failed to show a significant association between the BRCA2 rs144848 polymorphism and breast cancer [53]. In 2010, Qiu et al. found in a meta-analysis that the BRCA2 rs144848 H allele may be a lowpenetrant risk factor for developing breast cancer [54]. In 2014, Xue et al. conducted a meta-analysis to assess the association between the BRCA2 rs144848 polymorphism and cancer susceptibility [55]. In contrast to Qiu et al., Oncotarget 39828 www.impactjournals.com/oncotarget they did not find an association between the BRCA2 rs144848 polymorphism and breast cancer, but did observe an association with ovarian cancer. Different results from Xue et al. were then obtained in 2015 by Wang et al., who found that the rs144848 polymorphism was not associated with ovarian cancer. Compared with this latter study, we updated and added several new studies which were strictly filtered by a quality assessment. In addition, we used five genetic models to assess the role of the BRCA2 rs144848 polymorphism in our meta-analysis. Another important Oncotarget 39829 www.impactjournals.com/oncotarget difference from Wang et al. was that their results were based on the risk estimates obtained without the original genotype data, whereas all studies included in our metaanalysis provided genotype data, so that our results were more precise by calculating effect directly without potential deviations and biases.
The strength of this meta-analysis is that the most current literature was included. To guarantee the quality of the meta-analysis, the Newcastle-Ottawa scale was conducted to assess the quality of included studies, and a strict procedure for data extraction was performed by two investigators according to inclusion and exclusion criteria. Furthermore, no low-quality literature was included in this meta-analysis which might possibly have influenced our results. One-way sensitivity analysis and meta-regression were also performed to increase the robustness of our conclusions. Subgroup analysis by ethnicity and the source of the control population were used to explain the effect of genetic background and study design.
There are some limitations in this meta-analysis. First, the literature search strategy was limited by language, and only published papers in English were included. Second, because we excluded literature without original data, some studies were excluded. Third, other potential interactions including environment × gene, gene × gene and some potential covariates were not considered due to insufficient information.
In conclusion, our meta-analysis determined that the BRCA2 rs144848 polymorphism was associated with non-Hodgkin lymphoma, and indicated that the rs144848 H allele of the BRCA2 gene may be a low-penetrate risk factor enhancing carcinogenesis in breast cancer. Further well-designed studies are warranted to clarify the mechanism and increase comprehensive understanding of the role of the BRCA2 rs144848 polymorphism in cancer.

Publication research
Studies were retrieved by searching PubMed, Embase and Google Scholar following the guidelines in Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2009 [41]. The last search was updated on April 2016 with the terms "cancer", "tumor", "BRCA", "polymorphism", "genetic", "variant", "rs144848" and "N372H". References in potential articles were also included in order to find more relevant studies.

Inclusion criteria
All articles were reviewed by two investigators independently. Studies were included in the meta-analysis if they met the following criteria: (1) Studies were case-control or cohort studies; (2) articles were original studies of human participants; (3) genotype distributions were available; (4) studies were published in English; and (5) articles were association studies between rs144848 polymorphism and cancer risk. If studies were drawn from the same population, only the study with the largest sample size or with a sufficient quantity of useful data was included. If an article reported the results from different studies, each study was treated as a separate comparison in our meta-analysis.

Quality score assessment
The Newcastle-Ottawa scale was used to assess the quality of studies [42]. Three items including selection, comparability and exposure were used to calculate the score of studies with a maximum score of nine. Any disagreements were adjusted by a third reviewer. A total score of three or lower, four to six and seven or greater was considered to indicate low, medium and high quality studies, respectively.

Data extraction
Data were extracted from included studies using a standardized form. For each study, the following information was extracted: (1) name of first author, (2) year of publication, (3) ethnicity of population, (4) source of control population and (5) sample size and genotype distribution. Ethnicity was categorized as Caucasian, Asian or African, and the study design was categorized as population-based study, hospital-based study or nested study.

Statistical analysis
The odds ratios (ORs) with corresponding 95% confidence intervals (95% CIs) were calculated to assess the association between the rs144848 polymorphism and cancer risk. Five models were used in this meta-analysis: (1) H allele vs. N allele, (2) NH vs. NN, (3) HH vs. NN, (4) dominant model, (NH+HH vs. NN), and (5) recessive model, (HH vs. NH+NN). Statistical analysis was performed using STATA 11.0 (Stata Corporation, College Station, TX, USA). The chi-square test was conducted to evaluate if the studies deviated from Hardy-Weinberg equilibrium, and the threshold for disequilibrium was p < 0.05. Cochran's Q test and I 2 statistic test were performed to assess heterogeneity across individual studies (p < 0.10 and I 2 > 50% suggested heterogeneity). The fixed-effects model (the Mantel-Haenszel method) was used to estimate the pooled OR if I 2 < 50%; otherwise, the random-effects model (the DerSimonian and Laird method) was used [43]. A value of p < 0.05 was accepted as the significance threshold for each genetic model.