Predictors and outcomes of Melody valve therapy in structural and congenital heart disease: A retrospective study at the University Hospital Tübingen

Abstract

Die Dissertation ist gesperrt bis zum 02. Juni 2028 !

This retrospective, single-center study (2010–2024) evaluated long-term outcomes and predictors of valve dysfunction in patients with congenital heart disease undergoing transcatheter pulmonary valve replacement (TPVR) with the Melody valve. Data from 91 patients (cumulative follow-up: 450.7 patient-years) were analyzed. The primary endpoint was time to valve dysfunction (reintervention/reoperation). Statistical methods included Kaplan-Meier estimation, Cox proportional hazards regression, and random forest survival analysis to capture non-linear relationships. Thirteen patients experienced valve dysfunction. Overall freedom from dysfunction declined from 98.2% at 38.5 months to 58.0% at 113.0 months. Complications included endocarditis (n=5), graft stenosis (n=1), and stent fractures (n=7). Pre-stenting showed a protective effect. Traditional Cox regression identified higher BMI and residual RVOT pressure gradient as significant risk factors. The random forest model confirmed weight, follow-up RVOT gradient, and BMI as key predictors, revealing heightened risks in younger patients and those at extremes of body size. Despite limitations of a small cohort and low event rate, the findings align with existing literature. Integrating machine learning with traditional survival models successfully uncovered complex, non-linear predictor interactions, highlighting its potential to refine individualized post-TPVR care.