Symptomatic stratification based on morphological features of carotid web (SCORE-WEB)

Carotid web (CaW) is an underrecognized variant of fibromuscular disease increasingly implicated in cryptogenic ischemic stroke. Several morphologic features, including lesion length, proximal angulation, and pocket size, have been associated with symptomatic presentation. However, no pragmatic prediction score has been validated to stratify asymptomatic patients.

We performed a cross-sectional analysis of a prospectively maintained database of CaW cases collected between 2014 and July 2025 at two comprehensive stroke centers. Lesions were classified as symptomatic if associated with stroke or TIA involving the lesion vascular territory, and as asymptomatic if incidentally detected. Morphologic parameters extracted from computed tomography angiography (CTA) included pre-web angle, pocket area, lesion length, carotid bulb and distal cervical internal carotid caliber. A multivariable logistic regression model was developed, and a simplified score (SCORE-WEB) was derived based on cutoffs across these variables. Discriminative performance was assessed through ROC curves.

A total of 126 CaW were analyzed, including 80 symptomatic and 46 asymptomatic cases (30 incidental and 16 contralateral to a symptomatic CaW). Symptomatic lesions had more acute pre-web angles, larger pocket areas, and greater distal cervical internal carotid and bulb calibers compared with asymptomatic lesions, while substantial overlap in web length was observed. The multivariable logistic regression model demonstrated good discriminative performance for symptomatic presentation (AUC 0.803). The simplified SCORE-WEB (ranging from 0-11) achieved an AUC of 0.824. The Youden Index (J = 0.51) identified an optimal cutoff of ≥7 points, yielding a sensitivity of 0.54, specificity 0.97, positive predictive value 97.4%, and negative predictive value 44.4%.

The constellation of distinct morphological features appear to collectively and significantly determine the symptomatic status of CaW. The simplified SCORE-WEB, derived from five anatomical parameters, showed good discrimination and potential practical applicability. These findings warrant validation in larger and prospective datasets. The role of clinical variables in enhancing predictive performance merits further investigation.