Stefanie Aeschbacher (Switzerland)
Cardiovascular Research Institute Basel, University Hospital Basel, University Basel Cardiology Division, Department of MedicineAuthor Of 1 Presentation
PREDICTING NEW BRAIN INFARCTS OR CLINICAL STROKES IN PATIENTS WITH ATRIAL FIBRILLATION
Abstract
Group Name
on behalf of the Swiss-AF Investigators
Background And Aims
Brain infarcts are very common in patients with atrial fibrillation (AF) on systematic magnetic resonance imaging (MRI). We aimed to create a multimodal model to predict new brain infarcts or stroke during two years of follow-up.
Methods
Swiss-AF is a prospective, multicentre cohort study of patients with known AF. We included patients with available brain MRI at enrolment and after 2-years of follow-up. The primary outcome was a composite of any new brain infarct (small non-cortical infarcts and large non-cortical or cortical infarcts) on the follow-up MRI and/or a clinically identified stroke. To predict the primary outcome, we built a multivariable logistic regression model including prespecified clinical, biomarker and baseline MRI variables.
Results
We included 1,232 patients, 89.8% of them taking oral anticoagulants (OAC). The primary outcome occurred in 78 patients (6.3%). The following baseline variables were significantly associated with the primary outcome in the multivariable model: white matter lesion volume (aOR 1.89, 95%CI 1.43-2.55), N-terminal-pro hormone B-type natriuretic peptide (aOR 2.01, 95%CI 1.21-3.44), and heart-type fatty acid binding protein (aOR 0.49, 95%CI 0.31-0.77). Brain infarcts on baseline MRI tended to be associated with the primary endpoint (aOR 1.85, 95%CI 0.97-3.60). The area-under-the-curve (AUC) of the model was 0.82 (95%CI 0.77-0.87), the AUC of the CHA2DS2-VASc score was 0.64 (95%-CI 0.58-0.70) (figure).
Conclusions
A combined clinical-biomarker-MRI model is superior to the CHA2DS2-VASc score in predicting new brain infarcts or stroke over a 2-year of follow-up.
Trial Registration Number
Not applicable