Extracellular Vesicles—A Source of RNA Biomarkers for the Detection of Breast Cancer in Liquid Biopsies

SIMPLE SUMMARY: Extracellular vesicles (EVs) are nano-sized membrane-bound vesicles released into body fluids by various cell types, including cancer cells, that carry molecular cargo derived from their parental cells. Therefore, they are gaining attention as carriers of cancer biomarkers for liquid biopsies. Here, we sought to identify cancer-derived RNA biomarkers in plasma EVs from breast cancer (BC) patients. We reasoned that the amount of cancer-derived RNAs should decrease after surgical removal of the tumor; therefore, we compared the composition of EV RNA in BC patients at the time of diagnosis and 7 days after surgery and in cancer-free females. The results showed that the fractions of miRNAs, snRNAs, snoRNAs, and tRFs were increased, but the fraction of lncRNAs was decreased in BC EVs compared with healthy controls. We identified biomarker candidates among different RNA biotypes and created biomarker models that can detect BC and inform tumor estrogen receptor and HER2 status with remarkably high accuracy. ABSTRACT: Over the past decade, extracellular vesicles (EVs) have emerged as a promising source of cancer-derived RNAs for liquid biopsies. However, blood contains a pool of heterogeneous EVs released by a variety of cell types, making the identification of cancer RNA biomarkers challenging. Here, we performed deep sequencing of plasma EV RNA cargo in 32 patients with locally advanced breast cancer (BC) at diagnosis and 7 days after breast surgery and in 30 cancer-free healthy controls (HCs). To identify BC-derived RNA biomarkers, we searched for RNAs that had higher levels in BC EVs at the time of diagnosis compared with HCs and decreased after surgery. Data analysis showed that the fractions of miRNAs, snRNAs, snoRNAs, and tRFs were increased, but the fraction of lncRNAs was decreased in BC EVs as compared to HCs. BC-derived biomarker candidates were identified across various RNA biotypes. Considered individually, they had very high specificity but moderate sensitivity for the detection of BC, whereas a biomarker model composed of eight RNAs: SNORD3H, SNORD1C, SNORA74D, miR-224-5p, piR-32949, lnc-IFT-122-2, lnc-C9orf50-4, and lnc-FAM122C-3 was able to distinguish BC from HC EVs with an AUC of 0.902 (95% CI = 0.872–0.931, p = 3.4 × 10(−9)) in leave-one-out cross-validation. Furthermore, a number of RNA biomarkers were correlated with the ER and HER2 expression and additional biomarker models were created to predict hormone receptor and HER2 status. Overall, this study demonstrated that the RNA composition of plasma EVs is altered in BC patients and that they contain cancer-derived RNA biomarkers that can be used for BC detection and monitoring using liquid biopsies.