Functional substrate versus anatomical ablation approaches in structural ventricular tachycardia, in-silico comparison with a novel digital twin framework

Abstract Background Catheter ablation is a cornerstone therapy for scar-related ventricular arrhythmias in patients with structural heart disease. While scar homogenization remains an established and effective approach, it is often associated with extensive ablation volumes. Alternative strategies, such as cardiac magnetic resonance-guided corridor ablation and targeting of deceleration zones (DZ) identified by isochronal late activation mapping, have emerged as potentially more efficient and substrate-specific options. Purpose To compare in terms of in-silico sustained arrhythmia inducibility different ablation approaches for structural VT. Methods Fifteen consecutive patients prospectively enrolled in the VOYAGE clinical trial for CMR aided/guided VT ablation at our referral center were retrospectively analyzed. Multidetector computed tomography (MDCT), and LGE-CMR were acquired before procedure. LGE-CMR was segmented with ADAS3D for scar characterization. LGE-CMR derived normalized-pixel-intensity maps were processed with a novel in-house software (CardioMat) for VT induction using a standardized protocol (S1 600 ms x 6, S2 350 ms, S3 300 ms, S4 280 ms) pacing from all 17 segments of the left ventricle (LV) to obtain 3D local activation time (LAT) maps, including endocardial, mid-myocardial and epicardial activation. Different VTs with identical circuits were counted as one. In-silico ablation with 5 mm radius lesions (endo-epicardial when necessary) was performed along CMR corridors centerlines, at primary DZ, at secondary DZ elucidated after virtual remapping, and compared in terms of residual VT inducibility with scar homogenization in the same patients. Results A total of 63 distinct sustained ventricular tachycardias (VTs) were induced in 12 patients (mean 5.25 VTs per patient); 3 patients were non-inducible. Mean VT cycle length was 385 ± 49 ms (range 310–657 ms). All three ablation strategies—scar homogenization, CMR-guided corridor ablation, and DZ ablation—significantly reduced the number of inducible VTs compared with baseline (residual VTs per patient: 1.2 −82%, P=0.003; 2.41 −62%, P=0.003; 2.9 −52%, P=0.006, respectively), with no significant difference among the strategies (P = ns). Scar homogenization required a significantly greater ablation volume (17.41±7.22 mL) than both corridor (9.17±4.4 mL, P=0.001) and DZ ablation (10.62±5.7 mL, P=0.001), while ablation volume did not differ between corridor and DZ ablation (P=ns). Conclusion Our novel framework for generating LGE-CMR-derived in-silico VT activation maps enabled simulation of different ablation strategies, demonstrating that both corridor and DZ ablation effectively reduce VT inducibility with greater efficiency than conventional scar homogenization. Further validation is warranted to assess the clinical applicability of these imaging-guided, substrate-specific approaches.Figure 1