Predictive power of circulating miRNAs in detecting colorectal cancer

Abstract

Many studies indicate that circulating microRNAs (miRNAs) could play important roles in screening human cancers, including colorectal cancer (CRC). However, the conflicting results on the accuracy of miRNA detection lead us to conduct this meta-analysis to access the predictive value of miRNAs for predicting CRC. Eligible studies were identified from the Medline, Embase, CNKI, and Web of Science by the search strategies and screening criterion. We used random effects models to calculate the pooled results from studies. The summary receiver operator characteristic curve (SROC) and the area under the SROC curve (AUC) were used to estimate the predictive accuracy. Subgroup analyses and meta-regression were used to analyze potential sources of heterogeneity. We used Deeks’ funnel plot asymmetry test to test publication bias. This meta-analysis included a total of 24 studies from 19 articles, including 1558 CRC patients and 1085 controls. The overall pooled results from the meta-analysis were as follows: sensitivity was 0.81 (95 % confidence interval (CI) 0.77–0.85), specificity was 0.84 (95 % CI 0.78–0.88), PLR was 5.0 (95 % CI 3.5–6.9), NLR was 0.22 (95 % CI 0.18–0.28), DOR was 23 (95 % CI 14–37), and AUC was 0.89 (95 % CI 0.86–0.91). Subgroup and meta-regression analyses demonstrated that multiple miRNAs (AUC, sensitivity, and specificity of 0.92, 0.84, and 0.87, respectively) had a higher predictive accuracy than single miRNA (AUC, sensitivity, and specificity of 0.84, 0.78, and 0.78, respectively). In addition, we found that serum can be a better matrix for miRNA assays in screening CRC compared with plasma. In summary, our data suggests that circulating miRNAs, particularly multiple miRNAs, which have higher accuracy than single miRNAs, are excellent biomarker for screening CRC with good sensitivity and noninvasive nature.