Preprocedural cerebral small vessel disease burden predicts one-year cognitive decline in patients with atrial fibrillation undergoing left atrial appendage occlusion: a prediction model and mediation analysis

Background Cognitive outcomes after left atrial appendage occlusion (LAAO) are variable, and preprocedural markers of cerebral vulnerability are needed. Objectives This study evaluated whether preprocedural total cerebral small vessel disease (CSVD) burden predicts 1-year cognitive decline after LAAO and whether silent cerebral embolism (SCE) mediates this association. Methods This prospective, 2-center observational cohort study enrolled consecutive patients with atrial fibrillation (AF) scheduled for LAAO at the First Affiliated Hospital of Nanjing Medical University and Nantong First People's Hospital between January 2023 and June 2025. Total CSVD burden score was calculated from preprocedural brain magnetic resonance imaging (MRI). Cognitive function was assessed using the Montreal Cognitive Assessment (MoCA) and Mini-Mental State Examination (MMSE) before the procedure and at 1 year. The primary outcome was cognitive decline, defined as a decrease of at least 2 points in MoCA. SCE was assessed on follow-up MRI. Multivariable logistic regression, prediction modeling, bootstrap internal validation, and mediation analysis were performed. Results Among 110 patients, the mean age was 69.10 ± 6.48 years, 64.5% were men, and the median total CSVD burden score was 1.50, interquartile range 1.00 to 2.00. During 1-year follow-up, SCE was detected in 42 patients, 38.2%, and cognitive decline occurred in 32 patients, 29.1%. In a parsimonious multivariable model adjusted for age, sex, hypertension, and baseline MoCA score, total CSVD burden score was independently associated with MoCA-defined cognitive decline, odds ratio: 5.39 per 1-point increase; 95% CI: 2.91 to 11.75; P < 0.001. Adding total CSVD burden score to the clinical model significantly improved predictive performance, increasing the apparent area under the curve (AUC) from 0.670 to 0.890 and the optimism-corrected AUC from 0.608 to 0.863. Exploratory mediation analysis suggested a partial mediating role of SCE, with an estimated proportion mediated of 28.3%. MMSE-based sensitivity analyses were broadly consistent. Conclusions Preprocedural total CSVD burden score was associated with 1-year cognitive decline after LAAO and improved prognostic discrimination beyond clinical variables. SCE may partly mediate this association, supporting total CSVD burden score as a potential marker of vulnerability to postprocedural cognitive decline.