Acute ischemic stroke (AIS) has a varying impact on men and women. For instance, women feature a higher AIS severity than men that cannot be fully explained by key clinical variables. Further, we observed distinct sex divergences in lesion topographies in a recent study of 555 AIS patients: when explaining AIS severity, women were characterized by more widespread lesion pattern. Only in women, AIS severity was strongly affected by lesions in the left posterior circulation. In the current work, we aimed to determine sex-specific lesion pattern effects on long-term functional outcomes.
We relied on data of 822 AIS patients of the MRI-GENIE study (age:64.7(15.0), 39%female, 27%poor outcome, mRS>2). AIS lesions were automatically segmented from DWI-images, spatially normalized and parcellated into atlas-defined (sub)cortical regions and white matter tracts. Subsequently, we employed data-driven matrix factorization to obtain quintessential lesion patterns. Unfavorable functional outcome was modeled via Bayesian hierarchical logistic regression, taking the sex-specific lesion patterns and covariates age, age2, sex, hypertension, atrial fibrillation, diabetes, ischemic heart disease, prior stroke, smoking and total lesion volume as inputs.
We derived ten anatomically plausible lesion patterns that predominantly represented right- and left-hemispheric strokes with varying emphases on cortical-subcortical and anterior-medial-posterior regions. Three out of these ten patterns substantially contributed to the explanation of poor outcome for both men and women (AUC=0.81). These lesion patterns primarily comprised bilateral subcortical grey matter regions and left-lateralized regions in proximity to the insula. Additionally, a lesion pattern of left posterior circulation brain regions had a substantially higher relevance in women compared to men.
We here present evidence that female-specific lesion pattern effects relating to left-hemispheric posterior circulation regions are not only implicated in the explanation of AIS severity, yet have a lasting effect and also impact long-term outcomes. Future work is needed to examine links to more granular behavior.