Hsa-mir-499 rs3746444 polymorphism and cancer risk: a meta-analysis

MicroRNAs (miRNAs) are gene regulators involved in numerous diseases including cancer, heart disease, neurological disorders, vascular abnormalities and autoimmune conditions. Although hsa-mir-499 rs3746444 polymorphism was shown to contribute to the susceptibility of multiple genes to cancer, the data have yielded conflicting results. Therefore, this meta-analysis was performed to provide a comprehensive assessment of potential association between hsa-mir-499 rs3746444 polymorphism and cancer risk. In this meta-analysis, a total of 9 articles regarding 10 eligible case-control studies in English (including 6134 cases and 7141 controls) were analyzed. No significant association between hsa-mir-499 rs3746444 polymorphism and overall cancer risk was demonstrated. However, an increased risk was observed in the subgroup of breast cancer patients (G allele vs A allele: OR = 1.10, 95% CI = 1.00-1.20; Pheterogeneity = 0.114; I2 = 53.9%) and population-based studies (G allele vs A allele: OR = 1.12, 95% CI = 1.00-1.25; Pheterogeneity = 0.062; I2 = 64.0%). The findings suggested an association between hsa-mir-499 rs3746444 polymorphism and increased risk to breast cancer.


INTRODUCTION
MicroRNAs (miRNAs) are short (around 22-nt) non-coding single-stranded RNA molecules involved in both physiological and pathological processes, exerting their regulatory effects by suppressing translation or by inducing the cleavage of target RNA transcripts [1] . MiRNAs are regulators of gene expression that provide a regulation for a broad range of biological processes including cancer development, cellular differentiation, proliferation, apoptosis and metabolism [2][3][4][5][6] . MiR-499-5p, a gene with frequently increased expression in colorectal cancer, may function to promote migration and invasion of malignant cells [7] . One study has demonstrated that both the αand β-isoform of the calcineurin catalytic subunit are direct targets of miR-499, which inhibits apoptosis of cardiomyocytes through a suppression of calcineurin-mediated dephosphorylation of dynamin-related protein-1 (Drp1) [8] .
Single nucleotide polymorphisms (SNPs), the most common type of sequence variations in the human genome, contribute to human phenotypic differences [9] , and sequence variations could potentially affect the processing and/or target selection of miRNAs in miRNA genes, pri-miRNAs, pre-miRNAs and mature miRNAs [10] . Landi et al. [11] reported seven SNPs located in seven pre-miRNA hairpin regions, which include the polymorphism within hsa-mir499. Although a sizeable number of studies have been performed to investigate the role of hsa-mir-499 rs3746444 polymorphism in several cancer types such as breast cancer [12,13] , lung cancer [14] , gallbladder cancer [15] , squamous cell carcinoma of the head and neck (SCCHN) [16] , prostatic cancer [17] , gastric cancer [18] , cervical squamous cell carcinoma (CSCC) [19] and bladder cancer [20] , these existing eligible studies have yielded contradictory results, needing to be investigated further. Here, we performed a meta-analysis including subgroup analysis from all eligible studies to obtain a more precise assessment of the association between hsamir-499 rs3746444 polymorphism and cancer risk.

Publication search
We conducted a computerized literature search of PubMed, EmBase databases and Chinese National Knowledge Infrastructure (CNKI) databases (the last search was updated on August 10, 2011). The following search terms were used in isolation and combination with one another: "mir-499 or rs3746444" and "polymorphism or variant or mutation" and "cancer or carcinoma". The search was limited to papers published in the English language. The related reference articles were reviewed to identify other potentially eligible publications. All studies matching the eligible criteria were included in our meta-analysis. Fig. 1 shows the study capture procedure.

Inclusion criteria
Studies included in the meta-analysis met the following criteria: 1) articles on hsa-mir-499 rs3746444 polymorphism and cancer risk; 2) studies using a case-control design; 3) studies containing sufficient published data for the estimation of odds ratios (ORs) with their 95% confidence interval (CI).

Data extraction
Two investigators extracted necessary data from all eligible publications independently according to the inclusion criteria listed above. From each of the included articles the following information was abstracted: the name of first author, year of publication, country origin, ethnicity (Caucasian, Asian or others), cancer type, source of controls (population-or hospital-based controls), genotyping methods, total number of cases and controls, the number of cases and controls with hsa-mir-499 rs3746444 polymorphism genotypes, G allele frequency in controls and P value for Hardy-Weinberg equilibrium (HWE), respectively.

Statistical analysis
We first assessed HWE for the controls in each study. The strength of the association between polymorphism and cancer risk was assessed by ORs with 95% CIs. The statistical significance of the summary OR was determined by Z test (P < 0.05 was considered statistically significant). The pooled ORs were Fig. 1 Study flow chart calculated and used for comparisons between two homozygotes (GG vs AA), two heterozygotes (AG vs. AA), dominant models (GG+AG vs AA), recessive models (GG vs AG+AA) and the allele contrast models (G vs A), respectively. Subgroup analyses were also performed by cancer types (a cancer type with less than three individual studies was combined into other cancer groups), ethnicity and source of controls. Inter-study heterogeneity was estimated using a chisquare-based Q-test. The pooled ORs were analyzed using a random effects model (the DerSimonian and Laird method) [21] in case of a significant result (P < 0.05) provided by the Q-test, and using a fixed-effects model (the Mantel-Haenszel method) [22] if insignificance (P > 0.05) was found. The I 2 (I 2 = 100% × (Q-df)/Q) statistic was then used to quantitatively estimate heterogeneity, where I 2 < 25%, 25-75% and >75% represent low, moderate and high inconsistency, respectively [23,24] . Additionally, sensitivity analyses were performed by omitting each study to reflect the influence of the individual data on the summary ORs. Finally, publication bias of literatures was estimated using the Begg's funnel plot and Egger's test (P < 0.05 was considered a significant publication bias) [25] . All statistical analyses were performed with the software Stata (Version 11; Stata Corporation, College Station, Texas, USA), and all tests were two-sided.

Subgroup analyses
Subgroup analyses were performed of hsa-mir-499 rs3746444 polymorphism by cancer type, showing that the rs3746444 polymorphism was associated with elevated risk in breast cancer (G vs A, OR = 1.10, 95% CI = 1.00-1.20; P heterogeneity = 0.114; I 2 = 53.9%) rather than in other cancer types ( Table 2). When stratified by ethnicity, the results showed no significant association between hsa-mir-499 rs3746444 polymorphism and cancer risk among Asians and Caucasians in all comparison models tested ( Table 2). In the subgroup analysis by the source of controls, significantly increased risk was found in the population-based studies (G vs A, OR = 1.12, 95% CI = 1.00-1.25; P heterogeneity = 0.062; I 2 = 64.0%), not in the hospital-based studies ( Table 2).

Sensitivity analysis
In most of the studies, the frequency distributions of genotypes in the controls are generally consistent with the principles of HWE except one research conducted by Zhou et al. [19] . Removal of this study did not significantly change the corresponding pooled ORs. None of the pooled ORs were significantly affected by any single study, suggesting robustness of our results (Fig. 4).

Assessment of bias
We used Begg's funnel plot and Egger's test to assess the publication bias. The graphical funnel plots for the comparison of the G allele and the A allele appeared to be symmetrical (Fig. 5). The results of Egger's test did not show any evidence of publication bias (t = 1.90, P = 0.095) for the G vs A allele contrast model.

DISCUSSION
MiRNAs are short nucleotide RNAs that may influence mRNA stability and translation [26] . Recent studies have demonstrated that miRNAs can act as either tumor suppressors or oncogenes for cancers [26][27][28][29] and are associated with many other diseases including heart disease, neurological disorders, and autoimmune conditions. SNPs are the most common source of genetic polymorphism in the human genome. Various genetic association studies have explored the association between pre-miRNA polymorphisms and cancer risk. The SNP rs3746444 located in its corresponding 3p mature miRNAs regions may influence both the binding of 3p mature miRNAs to target mRNAs and pre-miRNA maturation of 5p and 3p miRNAs [12] . Srivastava et al. [15] reported that genetic variations in the miR-499 gene could alter PDCD4 expression levels, subsequently progressing to gallbladder epithelium transformation. In this study, to clarify controversial results from previous studies, we carried out a metaanalysis to investigate the association of hsa-mir-499 rs3746444 polymorphism with cancer. However, we failed to find any significant association between this polymorphism and cancer risk, which may be due to the limited number of studies included in this metaanalysis, the different matching criteria of the control group across studies, selection bias, and ethnic variation.
As tumor origin can influence the results from metaanalysis, we performed subgroup analyses by cancer type for hsa-mir-499 rs3746444 polymorphism. Our meta-analysis on the available studies showed that this Hu (2008) Tian ( Meta-analysis estimates, given named study is omitted | Lower CI Limit | Upper CI Limit ○ Lower CI Limit Fig. 4 The influence of individual studies on the summary odds ratio (OR). The middle vertical axis indicates the overall OR and the two vertical axes indicate its 95% confidence interval (CI). Every hollow round indicates the pooled OR when the left study was omitted in this meta-analysis. The two ends of every broken line represent the 95% CI. polymorphism was associated with the risk of breast cancer, suggesting that this polymorphism might be biologically functional in the pathogenesis of this malignancy. Some factors might account for these. First, hsa-mir-499 rs3746444 polymorphism may exert varying effect on the mechanisms of carcinogenesis, which may differ in different cancers. Second, hsamir-499 rs3746444 polymorphism detected in an invasive breast cancer cell line [30] suggested its relevance for promoting breast cancer progression. In the stratified analysis by ethnicity, no significantly increased cancer risks were found in Asians or Caucasians. The null result may be due to the limited number of studies, which had insufficient statistical power to detect a significant effect. Further studies are required to examine this association.
The results of meta-analysis often depend on control selection procedures. According to the source of controls, significantly increased risks were observed in the population-based studies. This may be due to the fact that the population-based controls might be typically representative of the general population and be better to reduce biases in such genetic association studies. Since there are only 3 population-based studies available, the data presented in this meta-analysis have limited power to reveal a reliable association. Further meta-analysis in the population-based studies subgroup with a larger sample size is necessary to be performed to examine this association.
Our meta-analysis should be interpreted within the context of its limitations. First, the total number of studies was too small to perform subgroup analyses. Second, our results were obtained based on the unadjusted estimates due to a lack of the genotype information stratified for the main confounding variables in the original papers. More accurate OR should be corrected by age, gender, smoking status, alcohol consumption and other exposure factors. Third, in this meta-analysis, the effect of gene-gene and geneenvironment interactions was not addressed.