ABSTRACT NUMBER: ESOC2026A714 CARDIAC IMAGING MARKERS AND INCIDENT ISCHEMIC STROKE: RESULTS FROM A POPULATION-BASED STUDY AND MENDELIAN RANDOMIZATION

Abstract Background and aims Imaging markers of cardiac structure and function could inform precision-medicine approaches to prevent cardio-embolism, but the optimal choice of biomarkers for this purpose is uncertain. Methods We investigated the association between markers of atrial and ventricular structure/function quantified on cardiac-MRI using automated machine-learning and incident ischemic stroke (IS) in a population-based study (UK-Biobank), excluding participants with baseline atrial fibrillation (AF), heart failure (HF), or prior stroke. Cox regression analysis was performed to estimate the hazard ratio (HR) for incident IS for 23 imaging markers after adjustment for cardiovascular risk factors/incident AF. Two-sample Mendelian randomization was then performed to investigate the causal role of selected cardiac traits in stroke pathogenesis. Results Data for 37,708 participants (264 outcomes) with cardiac-MRI were available. After correction for multiple comparisons, left-atrial (LA) ejection fraction (EF) and volume were independently associated with IS. Indices of impaired left-ventricular (LV) function/structural remodeling were also associated with IS, including EF, global longitudinal strain, both global and apical circumferential strain, and LV mass. Only a small proportion (0-8%) of the effect of LA/LV parameters on incident IS was mediated through AF/HF. Mendelian randomization analyses indicated that genetically-determined LA markers, but not LV markers, were associated with IS. Conclusions Cardiac-MRI markers of atrial and ventricular cardiopathy are associated with IS, mostly independently of AF/HF, but only LA markers are likely present on the causal pathway. These imaging markers could help advance the definition of cardiopathy and enrich patient selection in future trials of therapies for cardio-embolic stroke prevention. Conflict of interest John McCabe, funding from the Irish Institute of Clinical Neuroscience, University College Dublin, and the Mater Foundation, and in-kind contributions from Roche Diagnostics Ltd. Figure 1 - belongs to Results Figure 2 - belongs to Conclusions