Abstract WP290: Identification of Plasma Protein Biomarkers for Early-Onset Stroke through Mendelian Randomization

Introduction: Circulating plasma proteins have emerged as promising biomarkers for vascular disease, but their genetic contribution to early-onset stroke (EOS) risk remains underexplored. Leveraging large-scale proteome-wide association data and EOS case-control genotypes, we performed Mendelian randomization analyses to evaluate the causal effects of inflammation and cardiometabolic-related proteins on susceptibility to EOS and its subtypes. Methods: We used published GWAS summary statistics obtained from 54,219 participants in the UK Biobank for 1,281 plasma proteins to calculate genetic risk scores (GRS) for association with EOS in 6,728 genotyped cases and 33,764 controls of European ancestry from the Early-Onset Stroke Consortium (stroke onset < 60 years). Protein levels were measured on the Olink platform. All variants with frequency > 0.01 and within ± 200Kb of each protein-coding gene that were associated with protein levels (p ≤ 1×10 -5 ) were identified; those present in the EOS genotype data were then pruned to remove those in LD. GRS were calculated in PRSice (clumping: 250 kb, r 2 =0.1, p=1e -5 ). We tested for association of each GRS with EOS using logistic regression, adjusting for sex and 10 principal components for ancestry. Stroke subtypes were determined by TOAST classification. Results: GRS for 14 of the 1,281 proteins analyzed showed significant associations (P < 1×10 -3 ) with EOS or its subtypes. For all-stroke analyses, the strongest association was observed with the ABO GRS (P = 5.16×10 -9 ), with the association driven primarily by associations with the large artery atherosclerosis and undetermined subtypes. Other GRS associated with all-stroke included F11 , driven primarily by association with undetermined stroke subtype; BGLAP , driven primarily by association with the cardioembolic subtype; and BSG ; CXCL8 , FN1 , and IGFBP1. Subtype-specific associations included GRS for NCAM1 (with cardioembolic EOS), EGF and IGFBPL1 (with large artery atherosclerosis); LGALS3 (with small artery occlusion); PARP1 (with other determined etiology); and PDGFB and KHK (with undetermined stroke subtype). Conclusions: By integrating proteome-wide genetic data with EOS case-control genotypes, we identified novel and established proteins causally associated with stroke risk. These results fill a critical gap in understanding the molecular pathways underlying EOS. Some of these proteins may serve as useful biomarkers or therapeutic targets for stroke prevention.