Clinical Risk Factors and CT Imaging Features of Carotid Atherosclerotic Plaques as Predictors of New Incident Carotid Ischemic Stroke: A Retrospective Cohort Study

Abstract

Parameters other than luminal narrowing are needed to predict the risk of stroke more reliably, particularly in patients with <70% stenosis. The goal of our study was to identify clinical risk factors and CT features of carotid atherosclerotic plaques, in a retrospective cohort of patients free of stroke at baseline, that are independent predictors of incident stroke on follow-up. We identified a retrospective cohort of patients admitted to our emergency department with suspected stroke between 2001-2007 who underwent a stroke work-up including a CTA of the carotid arteries that was subsequently negative for acute stroke. All patients also had to receive a follow-up brain study at least 2 weeks later. From a random sample, we reviewed charts and imaging studies of patients with subsequent new stroke on follow-up as well as those who remained stroke-free. All patients were classified either as "new carotid infarct patients" or "no-new carotid infarct patients" based on the Causative Classification for Stroke. Independently, the baseline CTA studies were processed using a custom, CT-based automated computer classifier algorithm that quantitatively assesses a set of carotid CT features (wall thickness, plaque ulcerations, fibrous cap thickness, lipid-rich necrotic core, and calcifications). Univariate and multivariate statistical analyses were used to identify any significant differences in CT features between the patient groups in the sample. Subsequent ROC analysis allowed comparison to the classic NASCET stenosis rule in identifying patients with incident stroke on follow-up. We identified a total of 315 patients without a new carotid stroke between baseline and follow-up, and 14 with a new carotid stroke between baseline and follow-up, creating the main comparison groups for the study. Statistical analysis showed age and use of antihypertensive drugs to be the most significant clinical variables, and maximal carotid wall thickness was the most relevant imaging variable. The use of age ≥ 75 years, antihypertensive medication use, and a maximal carotid wall thickness of at least 4 mm was able to successfully identify 10 of the 14 patients who developed a new incident infarct on follow-up. ROC analysis showed an area under the ROC curve of 0.706 for prediction of new stroke with this new model. Our new paradigm of using age ≥ 75 years, history of hypertension, and carotid maximal wall thickness of >4 mm identified most of the patients with subsequent new carotid stroke in our study. It is simple and may help clinicians choose the patients at greatest risk of developing a carotid infarct, warranting validation with a prospective observational study.