Utility of Dual-Layer Spectral Detector CTA to Characterize Carotid Atherosclerotic Plaque Components: An Imaging-Histopathology Comparison in Patients Undergoing Endarterectomy

BACKGROUND. The composition of noncalcified portions of carotid atherosclerotic plaque is an important marker of plaque vulnerability and ischemia risk. OBJECTIVE. The purpose of this study was to assess the utility of dual-layer spectral detector CTA (DLCTA) parameters for characterization of carotid plaque components with histologic results from carotid endarterectomy as the reference. METHODS. Seven patients (five men, two women; mean age, 61.6 +/- 8.5 [SD] years) with carotid plaque awaiting carotid endarterectomy were prospectively enrolled and underwent preoperative supraaortic DLCTA. A neuroradiologist and pathologist performed joint slice-by-slice review of histologic slices of resected plaques and CTA images. With the use of anatomic landmarks, ROIs were placed on noncalcified components (lipid-rich necrotic core [LRNC], intraplaque hemorrhage [IPH], fibrous tissue, loose matrix) on CTA images and compared with corresponding histologic slices. For each ROI, attenuation was recorded for conventional polyenergetic images (CTPI) and virtual monoenergetic images with energy ranging from 40 to 140 keV (CT40-140keV), attenuation spectrum curve slope was calculated, and Z-effective value (representing effective atomic number) was recorded. DLCTA parameters were compared among plaque components. RESULTS. Seven plaques with a total of 65 slices and 364 ROIs (159 fibrous tissue, 96 LRNC, 86 loose matrix, 23 IPH) were analyzed. All parameters (CTPI, CT40-140keV, slope from 40 to 140 keV, Z-effective value) had significant differences between LRNC and the other components (all p <.001). For example, mean CTPI was 37.1 +/- 15.1 HU for LRNC, 58.4 +/- 21.6 HU for IPH, 69.7 +/- 20.5 HU for fibrous tissue, and 69.6 +/- 19.6 HU for loose matrix. Mean-CT40keV was 28.1 +/- 36.7 HU for LRNC, 87.5 +/- 48.9 HU for IPH, 106.3 +/- 47.5 HU for fibrous tissue, and 102.6 +/- 48.0 HU for loose matrix. AUC for differentiating LRNC from other components was highest (0.945) for CT40kev and decreased with higher energy; AUC for CTPI was 0.908. CT40kev also had highest accuracy (90.4%); at a cutoff of 55.7 HU, CT40kev had 88.5% sensitivity and 91.0% specificity. For differentiating IPH from fibrous tissue and loose matrix, AUC was highest at 0.652 for CTPI and 0.645 for CT40kev. CONCLUSION. DLCTA showed strong performance in differentiating LRNC from other noncalcified plaque components; CT40kev had highest accuracy, outperforming CTPI. CLINICAL IMPACT. DLCTA parameters may help characterize carotid plaque composition as a marker of vulnerable plaque and ischemia risk.