Correlation of Clinical and Ultrasound Variables to Vulnerability of Carotid Plaques in Patients Submitted to Carotid Endarterectomy

Abstract

Background: The aim of this study is to investigate the correlation of clinical and ultrasound parameters with characters of vulnerable atherosclerotic carotid plaque, as evaluated at preoperative magnetic resonance angiography (MRA), in patients submitted to carotid endarterectomy (CEA), in order to develop a clinical risk score for plaque vulnerability.

Methods: Preoperative data of patients submitted to CEA for significant carotid stenosis from January 1, 2012 to December 31, 2016 were retrospectively collected. The available case series was randomly divided into 2 groups, including a training (60%) and a validation series (40%). Data of plaque vulnerability were assessed at preoperative MRA scans. Univariate analysis was used on the training series to correlate the preoperative covariates available to the features of plaque vulnerability. Therefore, a backward selection procedure was performed again on the training series and on the validation series to assess if the same variables were associated to data of plaque vulnerability, in order to obtain a prediction model for the risk of plaque vulnerability. Odds ratios (ORs) with 95% confidence intervals were reported. P values <0.05 were considered statistically significant.

Results: The training case series consisted of 352 patients, while the validation case series of 248 patients. After univariate analysis and logistic regression, on the training and the validation series respectively, 6 variables were significantly associated to features of vulnerable plaque at preoperative MRA. These included male sex (OR 2.05), diabetes mellitus (OR 3.06), coronary artery disease (OR 1.95), neutrophil/lymphocyte ratio (OR 17.99), platelet counts (OR 1.03), and gray-scale median value (OR 0.84). A nomogram was then obtained from the final logistic model, in order to predict the probability of the presence of vulnerable carotid plaque, using a weighted points system. This risk score was then applied to the validation series. The validation data were found to have a C-index of 0.934.

Conclusions: Sex, diabetes mellitus, coronary artery disease, neutrophil/lymphocyte ratio, platelet counts, and gray-scale median value were significantly associated to the features of vulnerable plaque at preoperative MRA in patients undergoing CEA. In particular, when combined together in a "risk score," these variables provided an accurate probability of the presence of a vulnerable plaque at MRA scans.