ABSTRACT
PURPOSE: Circulating microRNAs (miRNAs) are potential diagnostic biomarkers for breast cancer (BC). The application of miRNA panels could improve the performance of screening tests. Here, we integrated bioinformatic tools and meta-analyses to select circulating miRNAs with high diagnostic accuracy and combined these markers to develop diagnostic panels for BC. METHODS: Analyses across databases were performed to identify potential BC-related circulating miRNAs. Next, a comprehensive meta-analysis was conducted for each miRNA following the PRISMA guidelines. An electronic and manual search for relevant literature was carried out by two reviewers through PubMed, ScienceDirect, Biomed Central, and Google Scholar. The quality of the included studies was assessed using the QUADAS-2, and the statistical analyses were performed using R software 4.1.1. Finally, the accurate biomarkers confirmed through meta-analyses were combined into diagnostic models for BC. RESULTS: Twenty-seven circulating miRNAs were identified as BC-related by bioinformatic tools. After screening, only 10 miRNAs presented in 45 studies were eligible for meta-analyses. By assessing pooled sensitivity, specificity, positive likelihood ratio, negative likelihood ratio, and diagnostic odds ratio, 8 miRNAs (miR-21, miR-30b, miR-125b, miR-145, miR221 miR-222, and miR-335) were revealed as promising BC diagnostic biomarkers. Two panels constructed from these miRNAs showed excellent diagnostic accuracy for BC, with areas under the SROC curve of 0.917 and 0.944. CONCLUSION: We identified 8 potential circulating miRNAs and 2 diagnostic models that are useful for diagnosing BC. However, the established miRNA panels have not been tested in any experimental studies and thus should be validated in large case-control studies for clinical use.
