Introduction
Little data exists on the electrophysiological differences between sustained atrial fibrillation (sAF; >5 minutes) vs self‐terminating nonsustained AF (nsAF; <5 minutes). We sought to investigate the electrophysiological characteristics of coronary sinus (CS) activity during postpulmonary vein isolation (PVI) sAF vs nsAF.
Methods and Results
We studied 142 patients post‐PVI for paroxysmal AF (PAF). In a 50‐patient subset, CS electrograms in the first 30 seconds of induced AF were analyzed manually. A custom‐made algorithm for automated electrogram annotation was derived for validation on the whole patient set. In patients with sAF post‐PVI, CS fractionated potentials were ablated.
Manual analysis showed that patients with sAF exhibited higher activation pattern variability (2.1 vs 0.5 changes/sec; P < .001); fewer proximal‐to‐distal wavefronts (25 vs 61%; P < .001); fewer unidirectional wavefronts (60 vs 86%; P < .001); more pivot locations (4.3 vs 2.1; P < .001); shorter cycle lengths (190 vs 220 ms; P < .001); and shorter cumulative isoelectric segments (35 vs 44%; P = .045) compared to nsAF. These observations were confirmed on the whole study population by automated electrogram annotation and sample entropy computation (SampEn: 0.29 ± 0.15 in sAF vs 0.15 ± 0.05 in nsAF; P < .0001). The derived model predicted sAF with 78% sensitivity, 88% specificity; agreement with manual model: 88% (Cohen's kappa= 0.76). CS defragmentation resulted in AF termination or noninducibility in 49% of sAF.
Conclusion
In PAF patients post‐PVI, induced sAF shows greater activation sequence variability, shorter cycle length, and higher SampEn in the CS compared to nsAF. Automated electrogram annotation confirmed these results and accurately distinguished self‐terminating nsAF episodes from sAF based on 30‐second recordings at AF onset.