|| Checking for direct PDF access through Ovid
Introduction: Atrial fibrillation (AF) and heart failure (HF) are associated with cognitive impairment. We used neuroimaging to describe if this association is explained by cardioembolism or other mechanisms.Methods: We included adults with HF (ejection fraction<45%) and AF but no stroke history. Healthy volunteer controls were matched, 1:2 ratio. Participants were assessed with Repeatable Battery for the Assessment of Neurospychological Status (RBANS), Hospital Anxiety and Depression Score (HADS) and 3-T brain MRI. Scans were graded for:infarct, enlarged perivascular spaces, microbleeds, global and regional (hippocampal) atrophy with consensus scoring by four raters using validated, ordinal assessment scales. Brain volumes were semi-automatically acquired using cluster analysis of T1-weighted and FLAIR voxel intensities and diffeomorphic atlas-based segmentation. CSF, hippocampal and white matter volumes were corrected for intracranial volume. We described univariable differences between groups and then created multivariable models where cardiac status was the dependent variable, RBANS and MRI data were the predictors.Results: Of 50 participants, AF-HF (n=34) had poorer RBANS (MD:16.9, SE:3.44; p<0.001). Differences were independent of education and HADS. Infarcts and ordinal markers of atrophy were significantly different between groups, SVD markers were increased in AF-HF but did not reach significance. Quantitative measures of white matter differed between groups but measures of atrophy did not.(Table) On multivariable models, no imaging feature was independently associated with cardiac status.Discussion: The association between cognitive impairment and cardiac disease may not be solely driven by occult cardioembolism; small vessel disease and other, neuroimaging independent, factors also interact. Differences between ordinal scales and quantitative scores suggest that future studies should use robust volumetric analyses.