Abstract TP150: Magnetic Resonance Imaging Predictors of Atrial Fibrillation in Cryptogenic Stroke Cases

Abstract

Introduction: Ischemic stroke is a leading cause of death and disability in the US, with 20-40% of cases, classified as cryptogenic or with an unexplained cause. Detection of one cause, paroxysmal atrial fibrillation (AF), is critical to ensuring optimal treatment with direct oral anticoagulant (DOAC). Currently, the most reliable AF detection strategy is use of an insertable cardiac monitor (ICM). However, earlier detection of occult cardioembolic patterns using MRI may promote earlier decisions for DOAC use. The overall goal of this study is to determine if MRI lesion patterns are predictive of AF detection by ICM. Methodology: Cases of consecutive patients (1/1/2015 - 12/31/2017) with MRI-confirmed stroke performed 48h from time last known well and prior to endovascular treatment were retrospectively analyzed. The primary outcome, presence of occult atrial fibrillation, was detected by ICM placement within 90 days and follow-up within 180 days from stroke. Four imaging patterns were tested as predictors of AF: i) acute stroke lesion involving multiple vascular territories (MVT, i.e. right or left carotid and/or posterior circulation), ii) MVT plus wedge-shaped cortical infarct or chronic stroke on FLAIR, iii) MVT involving 3 territories, and iv) MVT in 3 territories plus chronic FLAIR lesion. Adjustment variables were based on univariate logistic regression predictors of AF at P ≤0.1000. Results: Of the 101 cases in this analysis, the median age was 63 years and 49.5% male. Stroke in multiple vascular territories MVT was present in 22/101 (22%) at baseline. The total AF 6-month detection rate was 36/101 (36%). The imaging pattern most predictive of AF was pattern ii, MVT plus chronic FLAIR with an unadjusted odds ratio (OR) of 3.47, 95% CI of 0.3442-2.1731, P=.0073. The adjusted OR (age ≥ 55, history of stroke, and history of TIA,) was 3.26, 95% CI: 1.0358-11.1860, P =.0480. Conclusion: The presence of acute lesions in MVT and a chronic FLAIR lesion may be a biomarker of occult cardioembolic source that could be used early after the onset of stroke to determine optimal DOAC use for secondary prevention and potential risk reduction of stroke recurrence from suspected but unproven cardioembolic source. Funding: Lone Star Stroke Research Consortium