Topological Data Analysis Identified Prognostically-Distinct Phenotypes in Transcatheter Edge-to-Edge Repair Patients

Area of Science:

• Cardiovascular Medicine
• Medical Informatics
• Data Science in Healthcare

Background:

• Transcatheter edge-to-edge repair (TEER) is a complex procedure with heterogeneous patient outcomes.
• Identifying distinct patient phenotypes is crucial for accurate prognostication and risk stratification in TEER.

Purpose of the Study:

• To apply topological data analysis (TDA) for the first time to identify prognostically distinct clusters in TEER patients.
• To assess the utility of TDA in stratifying TEER patient risk based on morphology and clinical factors.

Main Methods:

• Utilized topological data analysis (TDA) on data from 389 TEER patients (June 2014-September 2020).
• Employed Python Scikit-TDA Kepler-Mapper for network graph creation and Louvain's method for clustering.
• Kaplan-Meier survival analysis was performed to compare all-cause mortality between identified clusters.

Main Results:

• Two major prognostically distinct clusters were identified among 389 TEER patients.
• Cluster 1 exhibited significantly worse survival (HR 2.70, P=.0005) and higher rates of residual mitral regurgitation (41.9% vs 19.4%).
• Cluster 1 was associated with increased tricuspid regurgitation, atrial fibrillation, leaflet prolapse, flail, and calcification.

Conclusions:

• Topological data analysis effectively identifies prognostically distinct clusters in TEER patients.
• TDA provides a valuable tool for risk stratification, improving procedural success prediction and survival assessment in TEER.