Existing risk scores performed poorly in a contemporary TTVI cohort, with TRIVALVE showing low discrimination (AUC 0.609) and TRI-SCORE significantly overestimating mortality (O:E ratio 0.13).
Cohort (n=457)
Yes
Do existing surgical and TTVI risk scores accurately predict mortality and rehospitalization in patients undergoing contemporary transcatheter tricuspid valve intervention?
Existing surgical and early TTVI risk scores are poorly calibrated or lack discrimination for contemporary transcatheter tricuspid valve interventions, highlighting the need for a dedicated TTVI risk model.
Effect estimate: O:E ratio 0.13 (95% CI 0.07-0.23)
BACKGROUND: Accurate risk stratification is crucial for patients undergoing transcatheter tricuspid valve intervention (TTVI). The performance of existing surgical and TTVI risk scores (TRI-SCORE, STS-TR Society of Thoracic Surgeons Tricuspid Regurgitation, and TRIVALVE International Multisite Transcatheter Tricuspid Valve Therapies Registry) has not been comprehensively evaluated and compared in a contemporary, real-world cohort. OBJECTIVES: The aim of this study was to assess the discrimination and calibration of these scores in a large international multicenter population of patients undergoing TTVI. METHODS: This study population included 457 patients from 6 international centers who underwent TTVI (tricuspid transcatheter edge-to-edge repair, transcatheter tricuspid valve replacement, or transcatheter tricuspid annuloplasty) between 2019 and 2024. The performance of the TRIVALVE score was assessed for the 1-year endpoint of death and rehospitalization. The TRI-SCORE and STS-TR scores were assessed for in-hospital and 30-day mortality, respectively. Performance was evaluated using C statistics for discrimination and smoothed calibration plots for calibration. RESULTS: All 3 scores demonstrated limitations. The TRIVALVE score showed low discrimination (area under the curve: 0.609) and was well calibrated after its endpoint was refined to exclude non-cardiovascular-related hospitalizations. The surgically derived TRI-SCORE and STS-TR scores were miscalibrated and significantly overestimated mortality. The TRI-SCORE showed an observed-to-expected mortality ratio of 0.13 (95% CI: 0.07-0.23), and the STS-TR score had an observed-to-expected mortality ratio of 0.35 (95% CI: 0.19-0.60). CONCLUSIONS: The currently available TTVI risk scores derived from surgical or early TTVI cohorts may not be well suited for accurate risk assessment in contemporary TTVI. The surgical scores when applied to TTVI are miscalibrated, and the TRIVALVE score lacks discrimination. There is a need for the development of a contemporary dedicated TTVI risk model validated specifically for this population.
Potratz et al. (Sun,) conducted a cohort in Transcatheter tricuspid valve intervention (TTVI) (n=457). TTVI risk scores (TRI-SCORE, STS-TR, TRIVALVE) was evaluated on Discrimination and calibration of TRIVALVE, TRI-SCORE, and STS-TR scores (O:E ratio 0.13, 95% CI 0.07-0.23). Existing risk scores performed poorly in a contemporary TTVI cohort, with TRIVALVE showing low discrimination (AUC 0.609) and TRI-SCORE significantly overestimating mortality (O:E ratio 0.13).