Systematic Review of Associations Between the Presence of Acute Ischemic Lesions on Diffusion-Weighted Imaging and Clinical Predictors of Early Stroke Risk After Transient Ischemic Attack

Abstract

Early risk of stroke after a transient ischemic attack can be reliably predicted with risk scores based on clinical features of the patient and of the ischemic event, but it is unclear how these features correlate with findings on brain imaging. We performed a systematic review of the literature and identified all previous studies which reported patient characteristics and the nature of transient ischemic attack symptoms in relation to appearances on diffusion-weighted imaging (DWI). We then performed a meta-analysis of the associations between the components of the risk scores and positive DWI. Authors were contacted for additional unpublished data. Nineteen studies were identified by the systematic review, and additional unpublished data were obtained from 11 of these studies. On meta-analysis, several components of the risk scores were associated with positive DWI, including symptom duration > or =60 minutes (13 studies, odds ratio [OR], 1.50; 95% CI, 1.16 to 1.96; P=0.004), dysphasia (9 studies, OR, 2.25; 95% CI, 1.57 to 3.22; P<0.001), dysarthria (8 studies, OR, 1.73; 95% CI, 1.11 to 2.68; P=0.03) and motor weakness (9 studies, OR, 2.20; 95% CI, 1.56 to 3.10; P<0.001). However patient age, sex, hypertension and diabetes were not associated with the presence of DWI lesions. From an etiologic perspective, atrial fibrillation (9 studies, OR, 2.75; 95% CI, 1.78 to 4.25; P<0.001) and ipsilateral > or =50% carotid stenosis (10 studies, OR, 1.93; 95% CI, 1.34 to 2.76; P=0.001) were associated with positive DWI. Presence of acute ischemic lesions on DWI correlates with several clinical features known to predict stroke risk after transient ischemic attack. Large studies (sample size >1000) will therefore be required to determine the independent prognostic value of DWI and its interactions with these clinical characteristics.