Abstract
Growing evidence from recent studies has revealed that microRNA-203 (miR-203) might be an attractive prognostic biomarker for cancer. But controversy still remains. The aim of this meta-analysis was to summarize available evidences and clarify the preliminary predictive value of miR-203 for prognosis in cancer patients. Eligible studies were identified through multiple research strategies in PubMed, EMBASE and Web of Science up to October 2015. Key statistics such as pooled hazard ratios (HR) with 95 % confidence intervals (CIs) were utilized to calculate patient survival. 13 eligible studies with 1600 patients were ultimately enrolled in this meta-analysis. Our results failed to show a significant relation between upregulated miR-203 expression and a favorable overall survival (OS) (HR 1.00, 95 % CI 0.65-1.36) in a random effect model. However, in subgroup analysis, we found that high expression of miR-203 was significantly associated with poor OS in Caucasian patients (HR 1.31, 95 % CI 1.06-1.55). In contrast, for Asian patients, over-expression of miR-203 was an independent prognostic factor for better and OS (HR 0.59, 95 % CI 0.22-0.96). It also suggested that cancer types and miRNA assay method were significant associated with prognosis. The over-expression of miR-203 was effectively predictive of worse prognosis in breast cancer (HR 6.35, 95 % CI 1.34-11.36), pancreatic cancer (HR 1.19, 95 % CI 1.08-1.30), ependymoma (HR 1.35, 95 % CI 1.10-1.61), but for glioma patients, elevated miR-203 is a potential biomarker for predicting better progression of cancer (HR 0.26, 95 % CI -0.02 to 0.54). Besides, for direct miRNA profiling studies, over-expression of miR-203 was an independent prognostic factor for worse OS (HR 6.35, 95 % CI 1.34-11.36). This meta-analysis indicated that ethnicity, tumor type and miRNA assay method mainly contributed to heterogeneity. Considering the insufficient evidence, further relevant studies are warranted.