Four-dimensional Cardiac Modelling of Multiphase Computed Tomography for Predicting Outcomes After Transcatheter Aortic Valve Replacement.

BACKGROUND

Multiphase computed tomography angiography (mpCTA) is routinely performed prior to transcatheter aortic valve replacement (TAVR) to determine eligibility and enable preprocedural planning. Incremental prognostic value may be realized from full-cycle, multiphase reconstructions to assess the contractile health of the cardiac chambers. In this study we assessed the feasibility of 4-dimensional chamber modelling of the left ventricle (LV) to support 3-dimensional minimum principal strain (3DminPS)-based predictions of clinical outcomes after TAVR.

METHODS

Two hundred five patients undergoing pre-TAVR mpCTA were studied. UNet-based 3D chamber segmentation was followed by mesh modelling and 3D feature tracking-based deformation to determine global 3DminPS for endocardial, epicardial, and transmural layers. Independent associations of 3DminPS with the primary outcome of heart failure hospitalization or death are described.

RESULTS

Of the 205 patients studied, 196 (96%) had analyzable mpCTAs (median age, 85 years; 55% male; Society of Thoracic Surgeons Predicted Risk of Mortality score = 3.10; 60.0% echocardiographic LV ejection fraction). At a median of 25 months after TAVR, 55 patients (28%) experienced the primary outcome. After adjustment for baseline variables, patients with an endocardial 3DminPS amplitude worse than -23.7% experienced a 2.7-fold higher risk of the outcome (adjusted hazard ratio, 2.7; 95% confidence interval, 1.4-5.1; P = 0.001), with this high-risk cohort having 1- and 3-year event rates of 32% and 49%, respectively.

CONCLUSIONS

Four-dimensional chamber modelling of mpCTA using UNet-based segmentation and standardized mesh deformation is feasible and enables delivery of 3D deformation markers with strong prognostic value for the prediction of outcomes after TAVR. Prospective validation in a multicentre setting is currently being undertaken.