Accuracy of device landing zone calcium volume measurement with contrast-enhanced multidetector computed tomography

Abstract

BackgroundThe extent of aortic valve calcification is an important determinant of procedural success in transcatheter aortic valve implantation (TAVI). We sought to validate device landing zone calcium volume (DLZ-CV) measurements on contrast-enhanced multidetector computed tomography (MDCT) with non-contrast-enhanced scans as reference. MethodsWe determined DLZ-CV in 141 patients undergoing transfemoral TAVI. Non-contrast-enhanced images were analyzed using a threshold of 130 HU as reference (DLZ-CV130). For contrast-enhanced scans, we applied various thresholds including 450 HU (DLZ-CV450), 850 HU (DLZ-CV850), mean aortic attenuation (AttenAo) + 2 SD (DLZ-CV2SD), AttenAo + 4 SD (DLZ-CV4SD), AttenAo + 4 SD + 5 mm3 volume filter (DLZ-CV4SD+), and based on visual estimation (DLZ-CVvis). We compared DLZ-CV values between patients with versus without paravalvular leak (PVL), and between patients with versus without post-dilatation stratified by the type of prosthesis. ResultsAll DLZ-CV measurements on contrast-enhanced scans significantly differed from DLZ-CV130 (p < 0.001 for all comparisons). The best approximation to DLZ-CV130 was achieved with DLZ-CV4SD+ (508 mm3 [332–772]; Pearson correlation: R = 0.87, p < 0.001; Bland-Altman: mean difference 1339 mm3 [limits of agreement 79;2600]). Moreover, DLZ-CV4SD+ allowed for discrimination of PVL ≥1° or the need for post-dilatation in patients receiving self-expanding prostheses. Procedural outcome using balloon-expandable prostheses was independent of DLZ-CV. ConclusionMeasurement of DLZ-CV using contrast-enhanced scans with unadjusted thresholds results in incorrect estimation of the calcium volume. The use of a scan-specific individual HU threshold including a volume filter (DLZ-CV4SD+) provides the best approximation to the reference and allows for discrimination of PVL ≥ 1° in patients receiving the Acurate neo prosthesis.