Proteomic and Transcriptomic Differences in Ischemic Stroke Patients with Atrial Fibrillation Versus Carotid Atherosclerosis

Background: Ischemic stroke occurring in the setting of atrial fibrillation (AF) or carotid atherosclerosis (CA) may reflect distinct underlying biological processes. We integrated proteomic, transcriptomic, and genetic data from different sources to identify circulating proteins and molecular pathways associated with ischemic stroke in patients with AF versus CA. Methods: We conducted a nested proteomic study within the UK Biobank comparing plasma protein levels among ischemic stroke patients with AF (n=539) and CA (n=127). Linear regression models were used to evaluate 2,923 proteins measured using the Olink Explore platform (false discovery rate [FDR] <0.05). In a separate Yale cohort, we evaluated expression of genes encoding identified proteins in thrombectomy clot single-cell RNA sequencing data from ischemic stroke patients with AF (n=7) or CA (n=7), including cell type-specific expression patterns. We then used summary statistics to perform 2-sample Mendelian randomization analyses using cis-protein quantitative trait loci to evaluate associations between genetically predicted levels of proteins identified in prior analyses and ischemic stroke subtypes. Exploratory pathway enrichment analyses were also performed. Results: Twelve circulating proteins differed significantly between ischemic stroke patients with AF versus CA. AF was associated with higher levels of NTproBNP, NPPB, and ACP5, and lower levels of APCS, ANGPT2, PAMR1, PRCP, PROS1, LARP1, F7, F10, and LEO1 (all FDR<0.05). Clot transcriptomic analyses showed corresponding differential expression of ACP5, PRCP, LARP1, ANGPT2, and LEO1 across AF versus CA patients. Pathway analyses suggested enrichment of coagulation-related pathways among proteins associated with CA and natriuretic peptide signaling pathways among proteins associated with AF. Mendelian randomization analyses demonstrated associations between genetically predicted protein levels and ischemic stroke subtypes (AF or CA), including cardioembolic stroke for NTproBNPand ischemic stroke for ANGPT2, ACP5, APCS, and PAMR1. Conclusion: Complementary proteomic, transcriptomic, and genetic analyses identified differing molecular profiles among ischemic stroke patients with AF versus CA. These findings support established biomarkers, including NTproBNP and coagulation-related proteins, while identifying additional candidate pathways that may contribute to biological differences between these stroke-associated conditions. Further validation in clinically adjudicated and longitudinal cohorts is needed.