Coronary plaque composition assessed by cardiac computed tomography using adaptive Hounsfield unit thresholds

Quantitative computed tomography (QCT) may be useful in detecting high-risk patients with coronary atherosclerosis. Assessment of plaque composition using fixed Hounsfield unit (HU) thresholds is influenced by luminal contrast density. A method using adaptive HU thresholds has therefore been developed. This study investigates agreement between plaque volumes derived using fixed and adaptive HU thresholds and the influence of luminal contrast density on the determination of plaque composition.

We performed QCT in 260 patients with recent acute-onset chest pain without acute coronary syndrome. Plaque volumes of necrotic core (NC), fibrous fatty (FF), fibrous (FI) and dense calcium (DC) tissue were measured in 1161 coronary segments. Agreement between plaque volumes using fixed and adaptive HU thresholds was tested using the Bland-Altman method. Additionally, patients were stratified into tertiles of ascending aortic luminal contrast density and plaque volumes were compared.

Bland-Altman plots revealed that fixed HU thresholds underestimated FI and FF plaque volumes and overestimated NC and DC plaque volumes compared to adaptive HU thresholds. Volumes of dense calcium plaque differed with increasing tertiles of luminal contrast density when using fixed HU thresholds but not when using adaptive HU thresholds: DC for fixed HU thresholds (mm³, median (95%CI)): 7.73 (5.17;12.31), 9.83 (6.55;13.57), 12.02 (8.26;16.24); DC for adaptive HU thresholds (mm³, median (95%CI)): 7.34 (5.12;12.03), 7.78 (5.40;12.61), 8.56 (5.22;12.69).

Plaque volumes by fixed and adaptive HU thresholds differed. Plaque volumes by adaptive HU thresholds were more independent of luminal contrast density for higher attenuation tissues compared to fixed HU thresholds.