Biomech Model Mechanobiol. 2025 Aug 9. doi: 10.1007/s10237-025-01984-1. Online ahead of print.
ABSTRACT
Transcatheter aortic valve replacement (TAVR) has revolutionized the treatment of severe aortic stenosis, yet paravalvular leakage (PVL) remains a significant complication, associated with increased mortality. Clinical studies have identified correlations between PVL and both anatomical features and calcification patterns. Numerical simulations, particularly patient-specific models, offer valuable insights into PVL, but the limited scale of these studies hinders robust statistical analysis. This study introduces a novel in silico clinical trial (ISCT) framework to investigate the correlation between calcification severity, localization and PVL. For this purpose, a synthetic cohort of calcified aortic roots was generated. A conditional convolutional variational autoencoder was used to create calcification patterns for an existing virtual cohort of the aortic root. The workflow includes finite element analyses for pre-dilation and deployment simulations as well as computational fluid dynamic simulations for PVL calculations of 243 virtual TAVR patients. The results show that the absolute amount of calcification in the device landing zone has no significant influence, but its regional distribution does, especially in the combined leaflet regions. In addition, sinotubular junction diameter, annular eccentricity index, oversizing as well as the combination of aortic angle and calcification in the combined non and left coronary leaflet region influence the occurrence of PVL. This framework not only advances our understanding of PVL mechanisms but also demonstrates the potential of ISCT to complement traditional clinical studies, enabling systematic exploration of complex factors influencing TAVR outcomes.
PMID:40782312 | DOI:10.1007/s10237-025-01984-1
