Abstract
BACKGROUND The evaluation of aortic valve calcium burden is important when planning for transcatheter aortic valve implantation (TAVI). Although a robust golden standard methodology is available for calcium evaluation on noncontrast-enhanced (NCE) computed tomographic (CT) series, a standard reference for calcium assessment on contrast-enhanced CT series is currently lacking. METHODS Two hundred and forty-four preprocedural CT scans from patients who had received TAVI were analysed. We correlated the aortic calcium volumes obtained on CE series at three thresholds [450, 850, and 'probe + 100' Hounsfield Units (HU)] with the Agatston score obtained on NCE scans. A subgroup analysis was performed taking into account the contrast enhancement of the left ventricular outflow tract (LVOT), with a prespecified cut-off of 300 HU. RESULTS The overall population analysis showed higher correlation with the Agatston score using the 850 HU threshold (r = 0.45, P < 0.0001); no correlation was found with the 450 HU threshold, whilst the 'probe + 100' HU threshold showed a weaker correlation (r = 0.30, P < 0.0001). In patients with LVOT enhancement less than 300 HU, 450 HU showed the highest accuracy in calcium identification (r = 0.70, P < 0.0001), whereas in patients with LVOT enhancement of at least 300 HU, the most accurate threshold was 850 HU (r = 0.46, P < 0.0001). CONCLUSION The thresholds for correct calcium identification using the automatic 3Mensio software depend on the contrast enhancement of aortic and cardiac structures, which can be estimated by measuring the HU in the LVOT. In patients with LVOT HU of less than 300, the correct threshold to be set in the software is 450 HU, whereas in patients with LVOT HU of at least 300 the correct threshold is 850 HU.
