Clinical and 2D/3D-Echo Cardiography Determinants of Mitral Valve Reoperation in Children With Congenital Mitral Valve Disease.

Background:Congenital mitral valve disease (CMVD) presents major challenges in its medical and surgical management. Objectives:The purpose of this study was to investigate the value of 3-dimensional echocardiography (3DE) and identify associations with MV reoperation in this setting. Methods:All children <18 years of age who underwent MV reconstruction for CMVD in 2002 to 2018 were included. Preoperative and postoperative 2-dimensional echocardiography (2DE) and 3DE data were collected. Competing risks and Cox regression analysis were used to identify independent associations with MV reoperation. Receiver operating characteristic and decision-tree analysis were implemented for comparison of 3DE vs 2DE. Results:A total of 206 children underwent MV reconstruction for CMVD (mitral stenosis, n = 105, mitral regurgitation [MR], n = 75; mixed disease, n = 26); 64 (31%) required MV reoperation. Variables independently associated with MV reoperation were age <1 year (HR: 2.65; 95% CI: 1.13-6.21), tethered leaflets (HR: 2.00; 95% CI: 1.05-3.82), ≥ moderate 2DE postoperative MR (HR: 4.26; 95% CI: 2.45-7.40), changes in 3D-effective orifice area (3D-EOA) and in 3D-vena contracta regurgitant area (3D-VCRA). Changes in 3D-EOA and 3D-VCRA were more strongly associated with MV reoperation than changes in mean gradients (area under the curve [AUC]: 0.847 vs AUC: 0.676, P = 0.006) and 2D-VCRA (AUC: 0.969 vs AUC: 0.720, P = 0.012), respectively. Decision-tree analysis found that a <30% increase in 3D-EOA had 80% accuracy (HR = 8.50; 95% CI: 2.9-25.1) and a <40% decrease in 3D-VCRA had 93% accuracy (HR: 22.50; 95% CI: 2.9-175) in discriminating MV reoperation for stenotic and regurgitant MV, respectively. Conclusions:Age <1 year, tethered leaflets, 2DE postoperative MR, changes in 3D-EOA and 3D-VCRA were all independently associated with MV reoperation. 3DE parameters showed a stronger association than 2DE. 3DE-based decision-tree algorithms may help prognostication and serve as a support tool for clinical decision-making.