The primary endpoint of this analysis was patient survival and the main secondary endpoints were significant residual MS (defined as immediate postprocedure mean gradient ?10 mm Hg), significant residual MR (defined as regurgitation ? moderate), and rate of repeat MV replacement ([MVR] either transcatheter or surgical). The mechanism of bioprosthetic valve failure was defined according to European Association of Echocardiography and American Society of Echocardiography criteria. 15 The presence of at least moderate MR and MS was defined as mixed failure. Surgical risk was estimated by the Society of Thoracic Surgeons (STS) MVR score. Chronic kidney disease was defined as estimated glomerular filtration rate ?60 mL/min·1.73 m 2 (ie, stage III and above). Clinical endpoints are reported according to the Mitral Valve Academic Research Consortium (MVARC) definitions (Expanded Methods). 16 Body surface area was calculated with the Mosteller formula. 17 Severe prosthesis-patient mismatch was defined as indexed effective orifice area ?0.9 cm 2 /m 2 for patients with body mass index <30 kg/m 2 and indexed effective orifice area ?0.75 cm 2 /m 2 for those with body mass index ?30 kg/m 2 . 18 Left ventricular outflow tract (LVOT) obstruction was defined as outflow mean gradient ?10 mm Hg 16 or cardiogenic shock that was clinically related to that complication as reported by the center. The true internal diameter for each model and size of surgical valve/ring was derived from previously published tables, when available. 19 Malposition was reported by the principal operator and defined as inadequate final position of the transcatheter heart valve for any cause, according to MVARC definitions.
Results are presented as mean±SD for continuous variables with normal distribution, median (interquartile range [IQR]; 25th–75th percentiles) for nonnormally distributed continuous variables and number (percentage) for categorical data. Student’s t test was used to compare means of normally distributed continuous variables between 2 groups. The Mann–Whitney U-test was used to compare distributions of nonnormally distributed continuous variables between 2 groups and the Kruskal–Wallis test was used to compare nonnormally distributed continuous variables between 3 or more groups. ? 2 and Fisher’s exact tests were used to compare proportions of categorical variables, as appropriate. Time-to-event curves were truncated at the last point with ?10% of patients at risk for the primary endpoint (4-year follow-up). Logistic regression was utilized to establish independent correlates of significant residual MS and significant residual MR. Cox regression was utilized to establish independent correlates of survival. Given the competing risk of mortality in the evaluation of repeat MVR, a Fine and Gray cause specific subdistribution hazards model was used. 20 The following variables were included in univariable models for significant residual MR and MS: body mass index, age, label size, true internal diameter, baseline MV area, baseline mean mitral gradient, baseline left ventricular ejection fraction, transcatheter heart valve diameter, male sex, mitral ViR (versus ViV), and MR versus MS as the mechanism of failure. In addition to the aforementioned variables, the following were also included in the survival and repeat MVR models: diabetes, peripheral vascular disease, chronic kidney disease, cerebrovascular disease, chronic lung disease, baseline New York Heart Association (NYHA) class IV symptoms (versus others), immediate postprocedural residual mean gradient ?10 mm Hg, residual MR ? moderate, baseline pulmonary artery systolic pressure, transseptal access, and LVOT obstruction, as well as if the case was performed before or after the tenth mitral ViR/ViV of a center (ie, median number of cases performed per center). The proportional hazards assumption was tested for each covariate of the Cox regression and for the final model. A center effect was included in the Cox proportional hazards model in the form of a shared frailty variable. Variables with a P value <0.1 in the univariable model were considered for inclusion in the multivariable model, with consideration also given to collinearity and overfitting. Odds ratio (OR), hazard ratio (HR), and subhazard ratio (SHR) are reported for binary logistic, Cox and Fine and Gray models, respectively, with the associated 95% CI. The first author and the corresponding author had full access to the data and vouch for its integrity. A 2-tailed P value <0.05 was considered statistically significant. Statistical analyses were performed with SPSS 23 (IBM Corporation, Armonk, NY) and Stata 14.1 (StataCorp, College Station, TX).