Real-life efficacy of artificial intelligence-guided spatio-temporal dispersion-based strategy in patients undergoing an atrial fibrillation repeat catheter ablation procedure

Background Although repeat catheter ablation procedure is often required after an initial ablation for atrial fibrillation, the optimal ablation strategy for repeat ablation remains uncertain. Purpose To evaluate the real-world effectiveness of an artificial intelligence (AI)-guided spatio-temporal dispersion-based strategy in patients with either isolated or reconnected pulmonary veins undergoing a repeat procedure for recurrent arrhythmia. Methods COMRADE-AI is an observational, prospective, non-randomised, multicentre registry evaluating the use of real-world AI-guided spatio-temporal dispersion ablation with an AI software. This analysis was limited to patients who presented for a repeat procedure. The procedural workflow was as follows: (i) AF induction if patient presents in sinus rhythm, (ii) 3D electro-anatomical mapping and dispersion mapping using the AIsoftware, (iii) ablation including spatio-temporal dispersion ablation, pulmonary vein isolation (PVI) if needed and extra-pulmonary veins (PV) triggers to the operator’s discretion. Procedural characteristics, safety and outcomes up to 12 months were analysed. Results A total of 54 patients with paroxysmal (11%), persistent (78%) and long-standing persistent (11%) atrial fibrillation were included (32 patients had one prior procedure, 21 patients had more than 1 prior procedure). Reconnected PVs was observed in 36/53 patients (68%) patients while 17/53 patients (32%) had isolated PVs. Patients baseline characteristics were similar between patients with isolated or reconnected PVs: 65±7 vs 65±9 years; 12% vs 22% women; 65% vs 67% hypertension; 1.7±0.9 vs 1.4±0.6 previous ablation respectively (p>0.05). However, fewer patients with isolated PVs had a NYHA II score compared to those with reconnected PVs (25% vs 64%, p<0.05), and more patients with isolated PVs had a NYHA III score (19% vs 0%, p<0.05). No differences in cardiac rhythm at the beginning of the procedure between the two groups were observed (p=0.35). Procedure time was similar (118±36 min isolated PVs vs 119±58 min reconnected PVs, p>0.9) as well as the rates of AF termination by ablation and sinus rhythm conversion by ablation (43% vs 30% p=0.5; 53% vs 36% p=0.4, respectively). Kaplan-Meier estimates of freedom from any atrial arrhythmia were 100% (95% CI, 100-100) for the isolated PVs group (median follow-up of 294 days) and 83% (95% CI, 70-98) for the reconnected PVs group (median follow-up of 285 days). Procedure-related complications were observed in 2 patients. Conclusion The AI-guided spatio-temporal dispersion-based ablation strategy appears particularly beneficial for patients with isolated PVs undergoing repeat procedures.