Abstract
Plasma small extracellular vesicles (sEVs) are a promising liquid biopsy tool. This study aims to delineate and validate a multimodal plasma sEV biomarker signature for glioma. We use size exclusion chromatography to separate sEVs from plasma (1 mL) and a combination of multi-spectral (Fourier transform infrared/Raman) and orthogonal multi-omics (proteomic/microRNA) approaches on 206 plasma samples (159 individuals) across three independent cohorts. We identify distinct glioma sEV biomolecular profiles, including differences in sEV protein/nucleic acid composition, and consistent alterations in 45 proteins and 20 microRNAs. Machine learning models derived from training cohort data achieve high diagnostic performance (areas under the curve [AUCs] 0.931-0.971), while external validation across independent cohorts confirms the signature's diagnostic potential, with 100% accuracy for the proteomic and multimodal signatures in the longitudinal cohort. Our findings, generated through a rigorous multi-cohort and multi-algorithmic framework, establish the potential of plasma sEV signatures as a clinically relevant diagnostic liquid biopsy approach for glioma.