Multimodal Approach for the Prediction of Atrial Fibrillation Detected After Stroke: SAFAS Study

Background Intensive screening for atrial fibrillation (AF) has led to a better recognition of this cause in stroke patients. However, it is currently debated whether AF Detected After Stroke (AFDAS) has the same pathophysiology and embolic risk as prior-to-stroke AF. We thus aimed to systematically approach AFDAS using a multimodal approach combining clinical, imaging, biological and electrocardiographic markers. Methods Patients without previously known AF admitted to the Dijon University Hospital (France) stroke unit for acute ischemic stroke were prospectively enrolled. The primary endpoint was the presence of AFDAS at 6 months, diagnosed through admission ECG, continuous electrocardiographic monitoring, long-term external Holter during the hospital stay, or implantable cardiac monitor if clinically indicated after discharge. Results Of the 240 included patients, 77 (32%) developed AFDAS. Compared with sinus rhythm patients, those developing AFDAS were older, more often women and less often active smokers. AFDAS patients had higher blood levels of NT-proBNP, osteoprotegerin, galectin-3, GDF-15 and ST2, as well as increased left atrial indexed volume and lower left ventricular ejection fraction. After multivariable analysis, galectin-3 ≧ 9 ng/ml OR 3.10; 95% CI (1.03–9.254), p = 0.042, NT-proBNP ≧ 290 pg/ml OR 3.950; 95% CI (1.754–8.892, p = 0.001, OPG ≥ 887 pg/ml OR 2.338; 95% CI (1.015–5.620), p = 0.046) and LAVI ≥ 33.5 ml/m 2 [OR 2.982; 95% CI (1.342–6.625), p = 0.007 were independently associated with AFDAS. Conclusion A multimodal approach combining imaging, electrocardiography and original biological markers resulted in good predictive models for AFDAS. These results also suggest that AFDAS is probably related to an underlying atrial cardiopathy. Clinical Trial Registration www.ClinicalTrials.gov , identifier NCT03570060.