Identification of microRNA biomarkers in type 2 diabetes: a meta-analysis of controlled profiling studies

Aims/hypothesis The aim was to identify potential microRNA (miRNA) biomarkers of type 2 diabetes. Methods Controlled studies were retrieved from PubMed to compare miRNA expression profiles of type 2 diabetes and nondiabetic control samples. Meta-analysis under a random effects model was conducted. Subgroup analyses examined tissue specificity and species specificity. Sensitivity analyses were also performed to explain the heterogeneity among studies. Results were represented as log odds ratios (logOR), 95% confidence intervals (CI) and p values after Bonferroni correction. Results Among 343 differentially expressed miRNAs in 38 miRNA expression profiling studies published between 1993 and March 2014, only 151 miRNAs were tested by multiple studies, out of which 102 miRNAs were reported to be upregulated or downregulated. Meta-analysis identified 51 significantly dysregulated miRNAs. The top upregulated miRNA was miR-142-3p (logOR 6.4721; 95% CI 4.9537, 7.9904; adjusted p = 4.60 x 10(-16)). The top downregulated miRNA was miR-126a (logOR 7.5237; 95% CI 4.7159, 10.3316; adjusted p = 3.01 x 10(-07)). The dysregulation of two miRNAs (miR-199a-3p and miR-223) was highly pancreas-specific and liver-specific. miR-30e was downregulated in patients with type 2 diabetes, while miR-92a was downregulated in animal models of diabetes. In sensitivity analysis, 40 out of 47 miRNAs (85%) were robustly and consistently dysregulated. Conclusion/interpretation This meta-analysis confirms that 40 miRNAs are significantly dysregulated in type 2 diabetes. miR-29a, miR-34a, miR-375, miR-103, miR-107, miR-132, miR-142-3p and miR-144 are potential circulating biomarkers of type 2 diabetes. In addition, miR-199a-3p and miR-223 are potential tissue biomarkers of type 2 diabetes.