Blinded Computationally Derived Peak Areal Stretch Location to Predict Aortic Root Rupture in TAVR

Background Aortic root rupture is a rare but lethal complication of transcatheter aortic valve replacement (TAVR). It is often the result of excessive radial forces experienced by the aorta during stent deployment. Current pre-procedural planning methods lack the predictive ability to capture interactions between the patient's native anatomical tissue and the transcatheter heart valve. Objectives This study aimed to investigate whether 3-dimensional (3D) computational modeling could be used to derive a biomarker that analyzed peak areal stretch to predict patient-specific risk for aortic root rupture. Methods In a 2-phase study, a derivation cohort (n = 26) was used to establish a peak areal stretch threshold for rupture risk, followed by a blinded validation cohort (n = 40) where modeling was performed using the federal drug administration-cleared PrecisionTAVI software. Researchers were blinded to the clinical outcomes of each case until after computational analysis had been completed. Peak areal stretch's predictive ability was compared to traditional factors such as calcium score or valve oversizing. Results A significant association was found between peak areal stretch and aortic root rupture. Rupture-positive cases had a higher mean stretch (1.956 ± 0.216) than negative cases (1.635 ± 0.284, P = .0285). A threshold of 1.6 showed strong predictive performance with an area under the curve (AUC) of 0.81, outperforming traditional metrics such as calcium score (AUC = 0.643) and valve oversizing (AUC = 0.697). Conclusion Peak areal stretch derived from computational modeling may serve as a patient-specific predictor of aortic root rupture, with potential to improve pre-TAVR planning and patient outcomes.