Deep Learning-Based Single-View Echocardiographic Analysis for Prediction of Left Ventricular Outflow Tract Gradient Elevation After Transcatheter Aortic Valve Replacement

Background and Objectives: Dynamic left ventricular outflow tract (LVOT) gradient elevation can occur following transcatheter aortic valve replacement (TAVR) and remains difficult to predict with conventional transthoracic echocardiography. We aimed to determine whether a deep learning (DL) model, originally developed in hypertrophic cardiomyopathy (HCM) to predict severe LVOT obstruction (LVOTO) from resting 2D parasternal long-axis videos without Doppler inputs, could identify susceptibility to post-TAVR LVOT gradient elevation from pre-TAVR echocardiograms. Methods: This retrospective study included 302 consecutive patients undergoing TAVR. The model was applied to pre-TAVR resting parasternal long-axis transthoracic echocardiography (TTE) videos to generate a patient-level DL index of LVOTO (DLi-LVOTO; range 0-100). Post-TAVR Doppler-defined LVOTO was defined as a peak pressure gradient [&ge;]30 mmHg on follow-up TTE. Results: Pre-TAVR LVOTO was present in 32 patients (10.6%) and post-TAVR LVOTO in 35 (11.6%). DLi-LVOTO was significantly elevated in patients who developed post-TAVR LVOTO (P<0.001). In multivariable analysis adjusting for conventional TTE parameters and pre-TAVR LVOTO status, DLi-LVOTO remained independently associated with post-TAVR LVOTO (adjusted OR 1.24, 95% CI 1.02-1.51 per 10-score increase, P=0.026), with an area under the receiver operating characteristic curve (AUROC) of 0.78 (95% CI 0.72-0.85). Among patients without pre-TAVR LVOTO, DLi-LVOTO retained independent predictive value (adjusted OR 1.61, 95% CI 1.27-2.14, P<0.001; AUROC 0.84, 95% CI 0.77-0.91). Conclusions: A HCM-trained DL model predicts Doppler-defined post-TAVR LVOTO from pre-TAVR TTE, suggesting that DLi-LVOTO may reflect a shared obstructive phenotype beyond conventional structural assessments.