Predictive value for paravalvular regurgitation of 3-dimensional anatomic aortic annulus shape assessed by...

Paravalvular regurgitation (PAR) remains a serious complication after trans-catheter aortic valve replacement (TAVR). Multidetector computed tomography (MDCT)-based measurements of the aortic basal virtual ring (BVR) are considered the gold standard for trans-catheter heart valve (THV) sizing. However, the real anatomic aortic annulus is a 3-dimensional structure. To compare measurement of 3D-anatomic annulus with BVR and secondly to assess independent predictive parameters that may impact on PAR > mild post-TAVR (PAR+).

MDCT was performed in 92 patients before and after balloon or self-expandable TAVR. 3D-AA shape was obtained point by point following the semilunar attachment of aortic cusps (Osirix-MD 2.8.2). 3D-oversizing index (nominal THV area/3D-AA area - 1) × 100 was calculated as well as 2D-oversizing index using BVR area instead of 3D-AA area. PAR was quantified by planimetry of vena-contracta in transthoracic echocardiography short-axis view. Valvular calcium volume and annulus calcium area were measured using Hounsfield-intensity detection. ROC curves and logistic regression for PAR(+) were performed. BVR area overall underestimated 3D-AA area by 19 ± 9% (P< 0.001), significantly more in PAR(+) (26 ± 7%) vs. PAR(-) (17 ± 9%, P< 0.001). 3D-oversizing index had greater predictive value for PAR > mild (area under the curve, AUC = 0.88) with 88% sensibility (Se) and 82% specificity (Sp) than 2D-oversizing index (AUC = 0.68) with 84% Se, but only 41% Sp (P< 0.0001). Also, valvular calcium volume and annulus calcium area were less predictors for PAR > mild (AUC = 0.68, respectively, AUC = 0.75, P = 0.002). In a multivariate analysis, only 3D-oversizing index showed an independent value for PAR > mild (OR = 18.6, P< 0.001).

Basal ring CT measurement significantly underestimated the real 3D-anatomic aortic annulus area. This may impact on THV sizing and PAR incidence. 3D-oversizing index is the most predictive factor for PAR > mild.