Stroke Lesion Pattern on MRI and Atrial Fibrillation

Introduction

Atrial fibrillation is the most common cause of cardioembolic stroke and can be detected in about 27% of acute stroke patients treated on stroke units in Berlin. Stroke patients in whom atrial fibrillation is detected benefit from anticoagulation rather than anti-aggregation as a secondary prevention treatment. Lesions on brain imaging that are located in more than one of the main arterial territories of the brain may indicate cardioembolic etiology. Cardioembolic etiology is mainly due to atrial fibrillation. Atrial fibrillation causes cardiac emboli that may embolize in all brain arteries while other embolic sources e.g. carotid plaque cannot. However, current data on the association between lesion patterns and atrial fibrillation are inconclusive.

Purpose

The aim of this retrospective cohort study is to investigate the relationship between lesion pattern on Magnetic Resonance Imaging (MRI) and atrial fibrillation. The investigators hypothesize that a lesion pattern with brain lesions in more than one of the main arterial territories of the brain (left or right internal carotid artery or posterior circulation territory) is associated with atrial fibrillation in patients with acute ischemic stroke. The findings of this study might help to identify patients with atrial fibrillation.

Methods

Blinded for clinical data the investigators will analyze imaging data and radiologic reports of 1000 consecutive stroke patients who received 3-Tesla MRI imaging including Diffusion-weighted Imaging (DWI) and Fluid-attenuated Inversion Recovery (FLAIR) sequence. DWI lesions correspond to acute stroke and FLAIR lesions to subacute and chronic stroke. Brain lesions will be attributed to a vascular territory (left or right internal carotid artery or posterior circulation territory) and the pattern consecutively categorized as "single", "scattered" and "multiple" as described by Braemswig et al..

Fetal origin of the posterior cerebral artery and variability of vascular territories will be taken into account. Vascular territory attribution in ambiguous cases will be conducted by consensus between investigators. Demographic and clinical data will be collected from medical records. Strokes meeting the criteria for Embolic Stroke of Unknown Source (ESUS) will be identified according to Hart et al..

Based on a pilot study including 250 patients the investigators estimate that atrial fibrillation will be present in 38% of patients with a multiple lesion pattern on MRI and in 28% of patients with a non-multiple pattern. Based on these assumptions, 894 patients are required to have a 80% chance of detecting, (as significant at the 5% level), a difference of 10% with 28% atrial fibrillation in patients without multiple lesion pattern and 38% atrial fibrillation in patients with a multiple lesion pattern.

In accordance with legislation in Berlin, Germany, an ethics committee approval is not required for this study.

Outcomes

Primary outcome is multiple lesion pattern on MRI (DWI and FLAIR). Secondary Outcome is multiple lesion pattern on MRI (DWI lesions only) Analyses will be conducted comparing

• Patients with and without atrial fibrillation
• Patients with newly diagnosed atrial fibrillation and patients without atrial fibrillation
• Patients with newly diagnosed atrial fibrillation and patients with known atrial fibrillation
• Patients fulfilling ESUS criteria and patients who do not
• Patients with brain lesions outside of the overlap area of the main vascular territories as defined by Tatu et al.