Purpose Carotid web is an intimal variant of fibromuscular dysplasia that has been associated with recurrent ischemic strokes in the young. The accuracy of different diagnostic tests is unknown. The objective of the present study is to evaluate the diagnostic performance of different imaging modalities for the detection of carotid webs.

Methods Using an institutional neurovascular database, we retrospectively identified consecutive patients diagnosed with carotid web and internal carotid artery atherosclerosis. Multimodality imaging including at minimum neck CT angiography (CTA) at baseline plus either Digital Subtraction Angiography (DSA) and/or carotid Ultrasound (US) was mandatory. CTA images were evaluated by two readers (neurointerventionist#1 and diagnostic neuroradiologist) in a consensus read for confirmation of diagnosis, and degree of stenosis based on NASCET criteria. DSA studies were evaluated by two blinded readers (neurointerventionists #2 and #3) to categorize the presence or absence of carotid web or atherosclerotic plaque in individual carotid arteries. US studies were evaluated by two blinded readers (two neurosonology-credentialed stroke neurologists) to characterize diagnosis (carotid web vs atherosclerosis vs normal), degree of certainty, and the degree of stenosis. In the setting of discrepancies, consensus read was obtained. The performance of DSA and of US was compared to CTA.

Results Median age of the cohort was 57 years (IQR 46–61 years) with 36.4% (12/33) male gender. Overall comorbidities included 60.6% (20/33) hypertension, 33.3% (11/33) hyperlipidemia, 21.2% (7/33) diabetes mellitus, and 27.3% (9/33) smoking. Multimodal imaging was available for 33 carotid arteries. Twenty-eight carotids had CTA and DSA available for review. DSA readers identified correctly all normal vessels, but diagnosed carotid webs less frequently (43%, 12/28; DSA group vs 50%, 14/28; CTA group) and atherosclerosis more frequently (43%, 12/28; DSA group vs 36%, 10/28; CTA group) when compared to CTA. Inter-rater correlation for the two DSA readers (κ=0.86, p<0.0001) and correlation between CTA and DSA (κ=0.92, p<0.0001) demonstrated very good strength of agreement. Ultrasound readers diagnosed carotid webs less frequently (12%, 4/33; US group vs 42%, 14/33; CTA group), and over-identified number of normal (42%, 14/33; US group vs 27%, 9/33; CTA group) and atherosclerotic vessels (45%, 15/33; US group vs 30%, 10/33; CTA group) as compared to CTA. Among the ultrasound readers discrepant interpretation was more common (39%; 13/33) leading to mere moderate strength of agreement among readers (κ=0.552, p=0.01). The correlation between CTA and US was moderate (κ=0.553, p=0.001). In assessing degree of carotid artery stenosis per etiology, ultrasound and CTA were quite comparable in normal vessels (89% concordance; 8/9 vessels), carotid web (86% concordance; 12/14 vessels), and atherosclerosis (90% concordance; 9/10 vessels). In 16 vessels where all three imaging modalities were present, the correlation between DSA and CTA had perfect agreement (κ=1.00, p<0.0001), while correlation with US and CTA was less conspicuous (κ=0.62, p=0.01).

Conclusion CTA and DSA should be preferred imaging modalities to diagnose and characterize carotid webs.

Disclosures T. Madaelil: None. R. Nogueira: None. J. Grossberg: None. A. Anderson: None. C. Barreira: None. D. Haussen: None.