Abstract
AIMS: This study aimed to investigate the correlation between clot composition and stroke mechanisms in patients undergoing endovascular therapy (EVT), using proteomic analysis. METHODS: This study included 35 patients with ischemic stroke (cardioembolism [CE], n = 17; large artery atherosclerosis [LAA], n = 6; cancer-related [CR], n = 4; and undetermined (UD) cause, n = 8) who underwent EVT. Retrieved clots were proteomically analyzed to identify differentially expressed proteins associated with the three stroke mechanisms and to develop the machine learning model. RESULTS: In the discover stage, 3838 proteins were identified using clot samples from 27 patients with CE, LAA, and CR mechanisms. Through functional enrichment and network analysis, 149 proteins were identified as potential candidates for verification studies. After verification experiments, 34 proteins were selected as the final candidates to predict stroke mechanisms. Furthermore, the machine learning-based model identified three proteins associated with each mechanism (Pleckstrin in CE; CD59 glycoprotein in LAA; and Immunoglobulin Heavy Constant Gamma 1 in CR) in the UD group. CONCLUSIONS: This study identified specific protein markers of clots that could differentiate stroke mechanisms in patients undergoing EVT. Therefore, our results could offer valuable insights into elucidating the mechanisms of ischemic stroke, which could provide information on more effective secondary prevention strategies.