Abstract T P43: Detection of Advanced Leukoaraiosis Does Not Augment Imaging-Enhanced Prediction of Early Stroke Risk in Transient Ischemic Attack
Background: The risk of stroke following transient ischemic attack (TIA) is high in patients with evidence of transient symptoms with infarction (TSI) on CT or diffusion-weighted imaging (DWI). The extent to which leukoaraiosis augments imaging-based risk prediction is unresolved. We sought to determine whether the presence of advanced leukoaraiosis improves the predictive performance of early CT and MR-based neuroimaging modalities in patients diagnosed with clinical TIA.
Methods: We retrospectively identified 37 cases of 30-day recurrent stroke that had MRI performed within 1 day of presentation, and compared to 43 patients without recurrence from a larger cohort of 1250 patients admitted to the hospital from a single emergency department with suspected TIA between 1/1/2000 and 7/16/2011. In all subjects (n=80), CT and MR (DWI and T2Flair)images were reviewed by a physician reader blinded to outcome and graded for the presence of abnormalities such as acute infarction, prior infarction, and advanced leukoaraiosis as defined by a Van Sweiten scale of 2 in either the anterior or posterior regions. We assessed the predictive performance of DWI alone, CT for infarction, CT for any abnormality and MRI for any abnormality by computing the area under the receiver-operating characteristics curve.
Results: The sensitivity of CT increased from 14% to 71% and MR from 68% to 93% in predicting 30-day stroke recurrence upon inclusion of acute or old infarction or advanced leukoaraiosis. Specificity decreased from 100% to 44% and 67% to 30% respectively. The area under the receiver-operating characteristic curve was 0.57 for CT for infarction alone and 0.58 for CT including leukoaraiosis; 0.68 for DWI alone and 0.61 for DWI or T2Flair evidence of old infarction or advanced leukoaraiosis.
Conclusions: The addition of advanced leukoaraiosis does not appear to add to the predictive ability of CT or MRI-based neuroimaging in the early risk stratification of patients with TIA.
Author Disclosures: M.S. Siket: None. J. Baird: None. H. Ay: None.
- © 2014 by American Heart Association, Inc.