Atrial Fibrillation Predictors on Mobile Cardiac Telemetry in Cryptogenic Ischemic Stroke

Background and Purpose: The objective of our study was to evaluate magnetic resonance imaging (MRI) and echocardiographic characteristics that would identify patients with cryptogenic ischemic stroke (IS) and transient ischemic attack (TIA) who subsequently developed paroxysmal atrial fibrillation (PAF) on mobile cardiac outpatient telemetry (MCOT). Methods: All patients with cryptogenic IS or TIA seen at the Emory University Hospital and Emory University Hospital Midtown from January 1, 2009, to June 30, 2013, who underwent MCOT were included in this analysis. Location (cortical, high subcortical, or neither) of current and prior strokes on MRI and left atrial (LA) functional and anatomical echocardiographic parameters were evaluated to determine their association with subsequent detection of PAF. Results: Of 132 patients, 17 (13%) had evidence of newly diagnosed PAF on MCOT (mean duration of monitoring = 25 days). The presence (vs absence) of ≥1 cortical infarct on baseline MRI was a significant predictor of identifying PAF (odds ratio: 5.2, 95% confidence interval: 1.3-19; P = .01). On baseline echocardiography, patients who had PAF (vs non-PAF) had significantly higher mean LA diameters (4.2 vs 3.7 cm, P = .03) and lower tissue Doppler velocity (a’; 5.5 vs 13.5 cm/s, P = .03). In receiver operating characteristic analysis, the ratio of LA volume index to the septal Doppler velocity (LAVI/a’) of >4.6 was associated with a higher likelihood of PAF. Combining MRI with echocardiographic variables did not improve the predictive ability beyond echocardiography alone. Conclusion: Although the presence of cortical-based infarcts on MRI in patients with cryptogenic IS or TIA increases the likelihood of detecting PAF on MCOT, LA functional and anatomic parameters alone best predicted which patients subsequently had PAF.