Intraluminal thrombus, intraplaque hemorrhage, plaque thickness, and current smoking optimally predict carotid stroke.

Abstract

Intraplaque hemorrhage (IPH) is associated with acute and future stroke. IPH is also associated with lumen markers of stroke risk including stenosis, plaque thickness, and ulceration. Whether IPH adds further predictive value to these other variables is unknown. The purpose of this study was to determine whether IPH improves carotid-source stroke prediction. In this retrospective cross-sectional study, patients undergoing stroke workup were imaged with MRI and IPH detection. Seven hundred twenty-six carotid-brain image pairs were analyzed after excluding vessels with noncarotid plaque stroke sources (420) and occlusions (7) or near-occlusions (3). Carotid imaging characteristics were recorded, including percent diameter and mm stenosis, plaque thickness, ulceration, intraluminal thrombus, and IPH. Clinical confounders were recorded, and a multivariable logistic regression model was fitted. Backward elimination was used to determine essential carotid-source stroke predictors with a threshold 2-sided P<0.10. Receiver operating characteristic analysis was performed to determine discriminatory value. Significant predictors of carotid-source stroke included intraluminal thrombus (odds ratio=103.6; P<0.001), IPH (odds ratio=25.2; P<0.001), current smoking (odds ratio=2.78; P=0.004), and thickness (odds ratio=1.24; P=0.020). The final model discriminatory value was excellent (area under the curve=0.862). This was significantly higher than the final model without IPH (area under the curve=0.814), or models using only stenosis as a continuous variable (area under the curve=0.770) or cutoffs of 50% and 70% (area under the curve=0.669), P<0.001. After excluding patients with noncarotid plaque stroke sources, optimal discrimination of carotid-source stroke was obtained with intraluminal thrombus, IPH, plaque thickness, and smoking history but not ulceration and stenosis.