Abstract
The association between brain structural connectivity (BSC) and different subtypes of stroke has not been reported. The current study determined whether some BSC patterns may contribute to the risk of stroke. A two-sample, bidirectional, multivariate Mendelian randomization (MR) analysis was performed. Genome-wide association summary statistics for BSC were obtained from the GWAS Catalog at the European Bioinformatics Institute, while stroke outcome data were obtained from the FinnGen study for intracerebral hemorrhage (ICH) and from the MEGASTROKE Consortium for ischemic stroke (IS) and its subtypes. A colocalization analysis was performed to determine whether the association between BSC and stroke was driven by loci within genomic regions. Reverse MR was performed to evaluate potential stroke-induced changes in BSC. Among the significant findings, left hemisphere (LH) somatomotor network-to-LH salience/ventral attention network white matter (WM) structural connectivity (SC) [OR = 1.30; p = 5.96 × 10(-4); p value after Bonferroni's correction [p.bfr] = 0.0125] and right hemisphere (RH) dorsal attention network (DAN)-to-thalamus WM-SC (OR = 1.23; p = 1.60 × 10(-3); p.bfr = 0.0125) were shown to have a positive association with the risk of IS. RH DAN-to-amygdala WM-SC (OR = 0.78; p = 1.26 × 10(-3); p.bfr = 0.0125) showed a negative relationship with the risk of IS. A high LH somatomotor network-to-RH visual network WM-SC (OR = 1.62; p = 9.10 × 10(-3); p.bfr = 0.025) was associated with an increased risk of large atherosclerotic stroke. In conclusion, the results of the current study provided some evidence from the perspective of genetics that different BSCs may have close associations with ICH, IS, and stroke subtypes. These findings may facilitate the screening and the risk stratification for stroke patients.