A circulating microRNA-based diagnostic model for breast cancer, in which the miR-139-3p/RPA2 axis regulates the sensitivity to DNA-damaging agents

Background: To date, no robust, non-invasive biomarker for breast cancer (BC) diagnosis has been identified. The study aimed to develop a circulating microRNA (miRNA or miR)-based diagnostic model and explore potential mechanisms of miRNA regulating DNA damage sensitivity. Methods: A differential analysis was performed on The Cancer Gene Atlas (TCGA)-Breast Invasive Carcinoma (BRCA) and Gene Expression Omnibus (GEO) datasets. Three machine-learning algorithms were employed to screen key circulating miRNAs. The serum miRNA levels were detected using real-time quantitative polymerase chain reaction (RT-qPCR). The effect of miR-139-3p on sensitivity to DNA damage and the underlying mechanisms were investigated in vivo and in vitro. Results: Our diagnostic model was based on the identification of four miRNAs (i.e., miR-139-3p, miR-134-3p, miR-629-3p, and miR-191-3p). The model achieved high diagnostic performance in two external datasets and a clinical cohort, which had area under the curve (AUC) values of 0.952, 0.847, and 0.869, respectively. The survival analysis showed that high levels of miR-139-3p were only linked with a better prognosis in the subgroups of patients that underwent chemotherapy [hazard ratio (HR) =0.6, P