Histopathologic validation of angioscopy and intravascular ultrasound

To establish a histopathologic basis for angioscopic and ultrasound image interpretation we studied 70 postmortem human arterial segments in vitro.

We used 7- to 9-French fiber-optic angioscopes and 20- to 30-MHz intravascular ultrasound imaging catheters. Three observers assigned an angioscopic and ultrasound image classification to each vessel segment. The image and histological classification categories were then compared. The sensitivity, specificity, and accuracy of both methods separately or in combination for normal vessels were each greater than or equal to 95%. The predictive value was better for angioscopy than for ultrasound due to incorrect ultrasound interpretations of normal anatomy in the presence of thrombus. For stable atheroma the sensitivity, specificity, and accuracy of the individual methods were each greater than 90%. However, both angioscopy and ultrasound had classification errors in that disrupted atheroma was identified and classified as stable atheroma. Consequently, the predictive value was 74% for angioscopy and 78% for ultrasound. For disrupted atheroma the sensitivities for angioscopy and ultrasound were only moderate (73% and 81%, respectively), whereas the specificity, accuracy, and predictive value were each high (greater than 90%). For thrombus detection, the specificity, accuracy, and predictive value were high (greater than 93%) for each method. The sensitivity of angioscopy was 100%. However, sensitivity was lower for ultrasound (57%) due to false-negative interpretation of laminar clots in normal vessels and an inability to distinguish disrupted or stable atheroma from intraluminal thrombus.

Contingency analyses showed that each imaging method alone or combined had significant agreement with the results obtained from histology (p less than 0.001). When assessing all cases in which angioscopy and ultrasound were concordant, there was a 92% agreement with the histological classification.