BACKGROUND:
Risk stratification in patients undergoing transcatheter mitral valve replacement (TMVR) remains challenging. Conventional clinical risk scores were not developed for this population, and it is uncertain how well routine baseline clinical and echocardiographic parameters capture mortality risk and generalize across transcatheter mitral interventions.
METHODS:
A baseline clinical risk model for 1-year all-cause mortality was developed in a multicenter TMVR cohort treated with the Tendyne™ system (n = 134). The model included routinely available preprocedural variables—age, sex, body mass index, atrial fibrillation, chronic obstructive pulmonary disease, and renal function—complemented by indexed left ventricular end-diastolic diameter (LVEDDi) as a structural LV parameter not captured by conventional surgical risk scores. Model performance was assessed in complete cases (n = 68). In patients with available Society of Thoracic Surgeons (STS) scores (n = 65), discrimination was compared with an STS-only model using identical samples. To explore transportability, the TMVR-derived baseline model was projected onto an independent MitraClip cohort with harmonized baseline variables and complete 1-year follow-up (n = 48).
RESULTS:
In the TMVR cohort, the baseline model demonstrated good apparent discrimination for 1-year all-cause mortality (AUC 0.77) with good calibration. Among individual baseline variables, atrial fibrillation showed the strongest association with 1-year mortality. In the STS subset, the TMVR-specific baseline model incorporating LV geometry showed numerically higher discrimination than STS alone (AUC 0.76 vs. 0.56) and lower prediction error, although the difference in AUC did not reach statistical significance. When applied to the MitraClip cohort, discrimination was moderate (AUC 0.63), with a broad distribution of predicted risks.
CONCLUSIONS:TMVR-specific modeling of routine baseline parameters, complemented by LV geometry, enables within-cohort risk stratification beyond conventional surgical risk scores, while transportability across transcatheter mitral repair and replacement populations remains limited. These findings highlight the constraints of purely baseline risk models and support the need for more comprehensive phenotyping strategies in this evolving field.

Author Block: Liliane Zillner¹, Miriam Wild², Michaela Hell³, Harald Herkner¹, Elmar Kuhn⁴, Tanja K. Rudolph⁵, Thomas Walther⁶, Lenard Conradi⁷, Andreas Zierer⁸, Francesco Maisano⁹, Marco Russo¹, Fabri Rosati¹⁰, Andrea Colli¹¹, Michael Pinon¹², David Reineke¹³, Gaby Aphram¹⁴, Christophe Dubois¹⁵, Joerg Hausleiter², Stephan Von Bardeleben³, Daniel Zimpfer¹, Juliana Coti¹, Martin Andreas¹.
¹Medical University Vienna, Wien, Austria, ²LMU University Hospital, Munich, Germany, ³University Medical Center Mainz, Mainz, Germany, ⁴University Hospital Cologne, Cologne, Germany, ⁵Heart and Diabetes Center Northrhine-Westfalia, Bad Oeynhausen, Germany, ⁶University Hospital Frankfurt, Frankfurt, Germany, ⁷University Hospital Cologne, Cologne, Austria, ⁸Johannes Kepler University Linz, Linz, Austria, ⁹IRCCS Ospedale San Raaffaele and University Vita Salute, Milano, Italy, ¹⁰Spedali Civili di Brescia, Brecia, Italy, ¹¹Medical and Molecular Pathology and Critical Care Medicine, Pisa, Italy, ¹²Hospital Álvaro Cunqueiro, Vigo, Spain, ¹³Inselspital University Hospital Bern, Bern, Switzerland, ¹⁴Cliniques universitaires Saint Luc, Brussel, Belgium, ¹⁵University Hospital Leuven and Department of Cardiovascular Sciences, Leuven, Belgium.