Tracing tumor-specific EV-miRNA signatures via patient-derived explant models to advance liquid biopsy in non-small cell lung cancer

Background: Non-small cell lung cancer (NSCLC) remains a leading cause of cancer mortality, largely due to late diagnosis and limited early-stage biomarkers. Liquid biopsy offers noninvasive monitoring, but most circulating markers lack clear tissue origin, limiting clinical interpretation. Methods: We developed a patient-derived explant workflow to trace the origin of extracellular vesicle-associated microRNAs (EV-miRNAs) and evaluate their clinical utility in plasma. Fifteen resected lung adenocarcinoma specimens and matched normal tissues were cultured ex vivo to collect tissue and secreted small EVs. Small-RNA sequencing of 60 samples (30 EV, 30 tissue) and RNA-seq of 30 tissues identified neuro-metabolic mRNA targets potentially regulated by miRNA shifts. 77 miRNAs were differentially sorted into tumor versus control EVs; 25 candidates were selected for validation. A custom TaqMan OpenArray™ panel demonstrated high concordance with qPCR, enabling high-throughput profiling. Results: We quantified these 25 EV-miRNAs in three cohorts: (1) 38 resectable patients with paired baseline and recurrence plasma; (2) 292 treatment-naïve patients across stages I-IV; and (3) 50 healthy controls. Six miRNAs - three “tumor-high” (miR-21-3p, miR-409-3p, miR-503-5p) and three “tumor-low” (miR-486-5p, miR-486-3p, miR-451a) - tracked recurrence in >50% of relapsing patients but remained stable in non-relapsers. Cross-sectional analysis showed stage IV patients recapitulated EV dysregulation (miR-21-3p high, miR-486-5p low), while localized disease mirrored controls. Prognostic modeling revealed stage-specific patterns: in stage III, low EV-miR-21-3p, miR-503-5p, and miR-486-3p predicted shorter progression-free survival; in stage IV, high EV-miR-21-3p correlated with worse overall survival, whereas high EV-miR-486-5p trended favorably. Exploratory Kaplan-Meier curves combining miR-503-5p/miR-486-5p for PFS and miR-21-3p/miR-486-5p for OS suggested risk stratification potential. Conclusions: Our explant-to-liquid biopsy approach identifies EV-miRNA signatures that reflect tumor biology, monitor recurrence, and stratify prognosis across NSCLC stages. These findings support clinical validation of miR-21-3p, miR-486-5p, and miR-503-5p as prognostic biomarkers and highlight patient-derived explants as a powerful platform for biomarker discovery.