Abstract WMP69: Profiling Extracellular Vesicle Surface Markers as Diagnostic Tool in Patients With Transient Ischemic Attack (TIA)

Abstract

Introduction: -TIA diagnosis is challenge. A circulating blood-based biomarker, supporting clinical and imaging evaluation, would be fundamental for optimizing the diagnosis of true brain ischemic events. Surface proteins carried by cellular secreted extracellular vesicles (EV), reflect abnormal inflammation and coagulation. We aimed to test the performance of EV surface proteins in discriminating true ischemic TIA. Methods: - We analyzed n=40 patients with suspicion of TIA, and n=20 asymptomatic matched controls. According to the PREDISC score, which includes clinical and imaging (multimodal MRI) criteria, patients were stratified as “unlikely” (score 0-1, n=10), “possibly” (score 2-3, n=15), or “very likely” (score 4-8, n=15) of having true ischemic TIA. Serum-derived EV were quantified by nanoparticle tracking analysis and EV-surface profile characterized by flow cytometry assay. Linear discriminant analysis (LDA) and random forest (RF) were built and validated through supervised machine learning algorithms. Results: - The concentration of circulating serum-derived EV was significantly increased in patients with proven ischemic events ranging in PREDISC score of 4-8 (p<0.001) and for those classified as “possibly” (score 2-3, P<0.05) as compared to controls. No significant differences were observed in patients stratified as “unlikely” versus controls. Overall EV concentration was directly correlated with the PREDISC score (R = 0.712; P < 0.001). Among EV-surface markers, CD8, CD2, CD62P, MCSP, CD42a, CD44, CD326, CD142, CD31, and CD14 were identified as discriminants between groups (p<0.05). Their expression was correlated with the PREDISC score (R range 0.320-0.461; P < 0.05), and associated with the occurrence of true ischemic TIA, as confirmed by univariate analysis (p<0.05). We combined 37 differentially expressed EV antigens in two models to predict PREDISC and achieved an accuracy of 95.0% and 90.0%, for the LDA and the RF model respectively. Conclusion: - Patients with recent true ischemic TIA appear to have a different cellular secreted EV profile in blood compared to controls. EV profile measurement is a promising novel biomarker to be validated in future studies addressing the optimization of TIA diagnosis.