Background: Macroreentrant atrial tachycardia (AT) accounts for 40% to 60% of recurrent atrial arrhythmias after atrial fibrillation (AF) ablation. To describe late gadolinium enhancement magnetic resonance imaging (LGE-MRI)-detected scar-based dechanneling as new ablation strategy to treat ATs after AF ablation.
Methods: Data from 102 patients who underwent initial AF ablation and repeat ablation for recurrent atrial arrhythmia within 1-year follow-up were analyzed. All patients underwent LGE-MRI before initial and repeat ablation. Depending on the recurrent rhythm, patients with AF and AT recurrence were assigned to group 1 or 2, respectively. Group 1 underwent fibrosis homogenization as second procedure. Group 2 underwent LGE-MRI-detected scar-based dechanneling. Both groups underwent reisolation of pulmonary veins if necessary.
Results: Forty-six patients (45%) presented with AF, and 56 patients (55%) presented with AT recurrence during follow-up after initial ablation. In the first 25 patients from group 2, the AT was electroanatomically mapped, and a critical isthmus was defined. It was found that those isthmi were located in the regions with nontransmural scarring detected by LGE-MRI. In the last 31 patients from group 2, an empirical LGE-MRI-based dechanneling was performed solely based on the LGE-MRI results. During 1-year follow-up after second ablation, 67% patients in group 1 and 64% patients in group 2 were free from recurrence (log-rank, P=1.000). In group 2, 64% in the electroanatomically guided and 65% in the LGE-MRI dechanneling group were free from recurrence (log-rank, P=0.900).
Conclusions: Anatomic targeting of LGE-MRI-detected gaps and superficial atrial scar is feasible and effective to treat recurrent arrhythmias post-AF ablation. Homogenization of existing scar is the appropriate treatment for recurrent AF, whereas dechanneling of existing isthmi seems the right approach for patients recurring with AT.
Keywords: arrhythmias, cardiac; fibrosis; heart atria; humans; magnetic resonance imaging.