Increased Risk of Breast Cancer Associated with CC Genotype of Has-miR-146a Rs2910164 Polymorphism in Europeans

Background Emerging evidence suggests that microRNAs play a critical role in the pathogenesis of breast cancer. Several molecular epidemiological studies were conducted in recent years to evaluate the association between has-miR-146a rs2910164 polymorphism and breast cancer risk in diverse populations. However, the results remain conflicting rather than conclusive. Methodology/Principal findings We performed a meta-analysis of 6 case-control studies that included 4238 breast-cancer cases and 4469 case-free controls. We assessed the strength of the association, using odds ratios (ORs) with 95% confidence intervals (CIs). Overall, this meta-analysis showed that the rs2910164 polymorphism was not associated with a significantly increased risk of breast cancer in all genetic models (for GC vs GG: OR = 1.00, 95% CI = 0.90−1.09, P heterpgeneity = 0.364; for CC vs GG: OR = 1.16, 95% CI = 0.98−1.36, P heterpgeneity = 0.757; for GC+CC vs GG: OR = 1.02, 95% CI = 0.93−1.12, P heterpgeneity = 0.562; for CC vs GC+GG: OR = 1.10, 95% CI = 0.96−1.26, P heterpgeneity = 0.441). However, in the stratified analysis by ethnicity, we found the rs2910164 polymorphism was associated with increased breast cancer risk among Europeans in homozygote comparison (CC vs. GG: OR = 1.29, 95%CI = 1.02−1.63, P heterpgeneity = 0.950, P = 0.032) and recessive model (CC vs. GC+GG: OR = 1.31, 95%CI = 1.05−1.65, P heterpgeneity = 0.839, P = 0.019). No publication bias was found in the present study. Conclusions/Significance This meta-analysis suggests, for the first time, that the CC homozygote of rs2910164 may contribute to breast cancer susceptibility in Europeans.


Introduction
Breast cancer (BC) is the most frequently diagnosed cancer and the leading cause of cancer death among women globally. About 1.3 million women are diagnosed with BC annually worldwide and about 465,000 die from the disease [1,2,3]. In the United States BC is the second most common cause of cancer death in women [1,3]. Moreover, BC incidence rates have been reported to increase by up to 5% per year in developing countries [2,3,4]. Although environmental factors and lifestyle could contribute to the increased BC risk, genetic factors are also implicated in the pathogenesis of the disease. A current priority in BC research is to identify genetic alterations that are directly involved in breast carcinogenesis. To date, a great number of genetic variants have been identified to be potentially associated with BC risk [5,6,7,8]. However, the molecular mechanisms that contribute to breast carcinogenesis remain poorly understood.
Emerging evidence supports a role for microRNAs (miRNAs) in BC development and progression [9,10]. miRNAs are non-coding, single-stranded RNAs of ,22 nucleotides and constitute a novel class of gene regulators that are found in both plants and animals. miRNAs are predicted to target over 50% of all human proteincoding genes, thus playing regulatory roles in a variety of physiological and developmental processes [11,12]. Recent evidence has shown that miRNA mutations or mis-expression are correlated with various human cancers including BC [13,14]. For example, microRNA-146a has been shown to bind to the 39UTR of the BC susceptibility gene BRCA1 and negatively regulate its expression [15,16]. A G.C polymorphism (rs2910164), which is located in the sequence of miR-146a precursor, results in a change from G:U to C:U in its stem region [16]. Up to now, a few molecular epidemiological studies have investigated the association between the miR-146a rs2910164 polymorphism and BC risk [16,17,18,19,20,21]. However, the results remain controversial and ambiguous. Because a single study might have been underpowered to detect the overall effects, a quantitative synthesis of the accumulated data from different studies is important to provide evidence on the association of rs2910164 polymorphism with BC risk. Thus, in this study we conducted a meta-analysis to combine all studies available and validate whether the miR-146a G.C polymorphisms contribute to BC susceptibility.

Publication search
We searched the PubMed and Embase databases for all articles on the association between has-miR146a rs2910164 polymorphism and BC risk up to September 2011. The following key words were used: ''miR-146a/rs2910164,'' ''breast cancer/carcinoma,'' and ''polymorphism/variant''. The electronic searching was supplemented by checking reference lists from the identified articles and reviews for additional original reports. All the studies must meet the following criteria: (1) case-control study; (2) the outcome had to be breast cancer; and (3) at least two comparison groups (cancer group vs. control group). The major exclusion criteria were: (1) duplicate data, (2) abstract, comment, review and editorial, and (3) no sufficient data were reported.

Data extraction
Two of the authors (H.L. and L.W.) extracted all data independently, complied with the selection criteria, and reached a consensus on all items. In case of disagreement, a third author (JM.Z.) assessed the articles. The following items were collected: first author's name, year of publication, country of origin, ethnicity, definition of study patients (cases), genotyping method, total number of cases and controls, and genotype distributions in cases and controls. PRISMA Checklist for the studies was shown as Checklist S1.

Statistical analysis
The departure of frequencies of hsa-miR-146a polymorphism from expectation under Hardy-Weinberg equilibrium (HWE) was assessed by the chi-square test in controls and a P,0.05 was considered as significant disequilibrium. The strength of the association between the hsa-miR-146a polymorphism and BC risk was measured by odds ratios (ORs) with 95% confidence intervals (CIs). The significance of the pooled OR was determined by the Ztest, and P,0.05 was considered statistically significant. For miR-146a G/C, the meta-analysis examined the association between C allele and BC risk compared with that for G allele in the codominant model (GC versus GG, CC versus GG), dominant model (GC/CC versus GG) and recessive model (CC versus GC/GG). The heterogeneity among the studies was checked by using the chisquare based Q statistic and considered statistically significant at P,0.10 [22]. When P.0.10, the pooled OR of each study was calculated by using the fixed-effects model (the Mantel-Haenszel method, which weights the studies by the inverse of the variance of estimates) [23]; otherwise, the random-effects model (the DerSimonian and Laird method) [24] was used. To adjust for multiple comparisons, we applied step-down Bonferroni method [25], which control for familywise error rate (FWE). The significance threshold was selected to be 0.1 after the Bonferroni correction, as used in the previous studies [26,27]. The Begg's rank correlation method and the Egger's weighted regression method were used to statistically assess publication bias (P,0.05 was considered representative of statistically significant publication bias) [28]. All analyses were done using STATA software, version 11.0 (STATA Corp., College Station, TX, USA), and all tests were two-sided.

Characteristics of the studies
A total of 41 articles were achieved by literature search from the PubMed and EMBASE, using different combinations of key terms.
As shown in Figure 1, 10 eligible studies were retrieved for detailed evaluation. We excluded six studies (four not focused on rs2910164, one was case only study and one did not present usable data). One study identified by manually searching reference lists of retrieved studies was also included [17]. Finally, a total of 6 case-control studies in 5 articles met our inclusion criteria [17,18,19,20,21], including 4238 cases and 4469 controls. Table 1 lists the characteristics of each study. There were four studies of Europeans, one study of Asians and one study of mixed populations. Genotyping methods included PCR-RFLP, TaqMan, MassARRAY and PCR-direct sequencing. The genotype distributions in the controls of all studies were in agreement with Hardy-Weinberg equilibrium except for one study [20].

Quantitative synthesis
The results of the overall meta-analysis did not suggest any associations between rs2910164 polymorphism and BC susceptibility for all genetic models (for GC vs GG: OR = 1.00, 95% CI = 0.9021.09, P heterpgeneity = 0.364; for CC vs GG: OR = 1. 16

Discussion
The identification of single-nucleotide polymorphisms (SNPs) that affect gene function or expression and contribute to BC susceptibility is important to help predict individual and population risk and understand the pathogenesis of BC. A G.C polymorphism (rs2910164) has been identified in the gene for miR-146a, a microRNA that binds to the 39 UTR of BRCA1 [15,16]. This G/C SNP (rs2910164) is located within the seed sequence of pre-miR-146a, which is the miR-146a precursor. [16]. The rs2910164 polymorphism has been associated with the risk of papillary thyroid cancer [29], hepatocellular cancer [30], esophageal squamous cell cancer [31], gastric cancer [32,33], prostate cancer [34,35] and glioma [36]. However, the relationship between rs2910164 polymorphism and BC susceptibility is undetermined. In 2008, shen et al. first reported a correlation between rs2910164 polymorphism in miR-146a and early cancer onset in familial BRCA1/BRCA2-negative breast/ovarian cancer patients [16]. This finding was replicated in a study based on 101 familial BC cases negative for mutations in BRCA1/2 [19], but not in a larger case-control study on 844 German and 760 Italian  familial BC cases negative for mutations in BRCA genes [20]. Moreover, Hu et al. failed to confirm the association of rs2910164 with BC risk in Chinese women [21]. Recently, Garcia et al. showed that the rs2910164 SNP in the miR-146a gene was not associated with BC risk in BRCA1 and BRCA2 mutation carriers [18]. In this meta-analysis, we summarized current data on the association between miR-146a rs2910164 polymorphism and BC risk. We found that hsa-miR-146a rs2910164 G.C polymorphism was associated with a significantly increased risk of BC in the CC homozygote as opposed to the GG homozygote or GC/GG genotype among Europeans. These findings suggest, for the first time, that the CC homozygote of rs2910164 polymorphism may contribute to BC susceptibility. MicroRNAs could function as tumor suppressors or oncogenes and are down-regulated or up-regulated in human tumors, including BC. Based on array data, it was previously reported that miR-146a was significantly upregulated in breast carcinoma tissues compared with normal breast tissues [37]. Overexpression of miR-146a has been reported as a signature in breast, pancreatic and prostate cancers [37,38]. A unique example of a functional miRNA SNP is rs2910164, which is located in the 3p strand of mir-146a. This polymorphism involves a mispairing in the hairpin of the precursor, which leads to altered processing, lower expression of the mature sequence and predisposition to papillary thyroid carcinoma [29]. However, differences in genotype distribution have been found in a few case-control studies. Okubo et al. found that the rs2910164 CC genotype was associated with a significantly higher risk of gastric cancer when compared to non-cancer subjects [32]. In contrast, male individuals with GG genotype were 2-fold more susceptible to hepatocellular carcinoma (HCC) compared with those with CC genotype [30]. In addition, Guo et al. reported that rs2910164 genotype GG was associated with increased risk of esophageal squamous cell carcinoma compared with variant genotype CC in a Chinese Han population [31]. In this meta-analysis, significant association was found between the rs2910164 CC genotype and BC risk in Europeans. These results suggest that the molecular mechanisms underlying the genetic associations of miRNA-SNPs (mir-SNPs) with cancer are complex and the polymorphism might play a different role in different cancers.
One important property of the gene polymorphism is that their incidence can vary substantially between different racial or ethnic populations. Xu et al. found significant differences in the prevalence of the rs11614913 T allele and rs2910164 C allele among controls of Asian (0.544 and 0.498, respectively) and Caucasian (0.375 and 0.246, respectively) [39]. At population level, we found that individuals with rs2910164 C allele could reduce cancer susceptibility in Asians but not in Caucasians, suggesting a possible role of ethnic differences in the genetic background and the environment they lived in (unpublished data). Meta-analysis is a powerful statistical tool that provides a consensus by combining the data from diverse studies that reveal inconsistent results on the same problem. Some results of our study did not show any statistical significance. However, subgroup analysis by ethnicity showed a statistical significant result, suggesting genetic diversity among different ethnicities.
Although many previous studies have been carried out to reveal the potential associations between mir-SNPs and cancer risk, most  of these studies use a candidate gene approach as most mir-SNPs have not been included in current genome-wide association studies (GWAS) designs [14]. The inclusion of mir-SNPs in future GWAS will help to unveil low-penetrance susceptible mutations.
In interpreting the current results, some limitations should be considered. First, lack of the original data of the reviewed studies limited our further evaluation of potential interactions, because the interactions between genes and between genes and environment may modulate breast cancer risk. Second, only published studies were included in this meta-analysis, unpublished data and ongoing studies were not sought, which may have biased our results. Third, all case-control studies were from Asia, Europe, and the United States. Thus our results may be applicable only to these ethnic groups. In spite of these, our meta-analysis has some advantages. First, in the present study, no obvious heterogeneity between studies was observed in overall comparisons by the Q-test. Second, according our selection criteria, the quality of studies included in our meta-analysis was satisfactory. Third, no publication bias was detected, indicating that the whole pooled results may be unbiased.
In conclusion, this meta-analysis suggests that CC genotype of has-miR-146a rs2910164 polymorphism is associated with an increased breast cancer risk in Europeans. Caution must be made about the interpretation of the results because of the limited sample size. Additional large case-control studies are necessary to validate our findings, especially in non-Europeans.

Supporting Information
Checklist S1 (DOC) Author Contributions