Carotid Plaque-RADS: Impact of its ancillary features on improving the detection of symptomatic plaques.

Abstract

This study aimed to evaluate the predictive value of the carotid plaque reporting and data system (Plaque-RADS) combined with ancillary features for identifying symptomatic carotid plaques, and to explore its utility in cerebrovascular risk stratification. We retrospectively included patients with carotid atherosclerosis who underwent both computed tomography angiography and high-resolution vessel wall MR imaging between January 2021 and December 2024, and classified them according to the presence of symptoms and the degree of stenosis. Carotid plaques were evaluated using the Plaque-RADS and multiple ancillary features, including the rim sign, plaque burden, remodeling index, enhancement ratio, and perivascular fat density (PFD). Logistic regression was used to identify independent risk factors for symptomatic plaques, and predictive models were subsequently developed and evaluated. Multivariate logistic regression identified the degree of stenosis, Plaque-RADS, plaque burden, remodeling index, enhancement ratio, and mean value of PFD as independent risk factors for symptomatic plaques. In the overall cohort, the integrated model combining Plaque-RADS, degree of stenosis, and multiple ancillary features achieved the best predictive performance (AUC = 0.876, 95% CI: 0.828-0.924). In the moderate stenosis subgroup, the combination of Plaque-RADS and ancillary features also further improved diagnostic accuracy (AUC = 0.919, 95% CI: 0.867-0.971). Models incorporating ancillary features demonstrated good calibration and provided the greatest net clinical benefit in decision curve analysis. Ancillary features provide additional diagnostic value to Plaque-RADS for predicting cerebrovascular events. The integrated model incorporating Plaque-RADS, ancillary features, and the degree of stenosis effectively identified symptomatic plaques and further improved risk stratification. Question The incremental value of ancillary features of Plaque-RADS for identifying symptomatic carotid plaques remains unclear. Findings Ancillary features (plaque burden, remodeling index, enhancement ratio, and mean value of PFD) improved the predictive performance of Plaque-RADS, with the integrated model achieving the highest AUCs in the overall cohort and moderate stenosis subgroup (0.876 and 0.919, respectively). Clinical relevance Incorporating ancillary features into Plaque-RADS facilitates the identification of high-risk plaques and enhances cerebrovascular risk stratification.