Abstract

BACKGROUND: Current outcomes from catheter ablation for scar-dependent ventricular tachycardia (VT) are limited by high recurrence rates and long procedure durations. Personalized heart digital twin technology presents a noninvasive method of predicting critical substrate in VT, and its integration into clinical VT ablation offers a promising solution. The accuracy of the predictions of digital twins to detect invasive substrate abnormalities is unknown. We present the first prospective analysis of digital twin technology in predicting critical substrate abnormalities in VT. METHODS: Heart digital twin models were created from 18 patients with scar-dependent VT undergoing catheter ablation. Contrast-enhanced cardiac magnetic resonance images were used to reconstruct finite-element meshes, onto which regional electrophysiological properties were applied. Rapid-pacing protocols were used to induce VTs and to define the VT circuits. Predicted optimum ablation sites to terminate all VTs in the models were identified. Invasive substrate mapping was performed, and the digital twins were merged with the electroanatomical map. Electrogram abnormalities and regions of conduction slowing were compared between digital twin-predicted sites and nonpredicted areas. RESULTS: Electrogram abnormalities were significantly more frequent in digital twin-predicted sites compared with nonpredicted sites (468/1029 [45.5%] versus 519/1611 [32.2%]; P<0.001). Electrogram duration was longer at predicted sites compared with nonpredicted sites (82.0±25.9 milliseconds versus 69.7±22.3 milliseconds; P<0.001). Digital twins correctly identified 21 of 26 (80.8%) deceleration zones seen on isochronal late activation mapping. CONCLUSIONS: Digital twin-predicted sites display a higher prevalence of abnormal and prolonged electrograms compared with nonpredicted sites and accurately identify regions of conduction slowing. Digital twin technology may help improve substrate-based VT ablation. REGISTRATION: URL: https://www.clinicaltrials.gov; Unique identifier: NCT04632394.