High dominant frequency (DF) sources have been described as drivers of atrial fibrillation (AF). However, the spatial stability of these high frequency drivers across long durations of AF and distinct episodes of AF is unknown.
A paced canine model of chronic persistent AF (n=6, AF>12 months) was used for this study. A 64-electrode constellation basket catheter (Boston Scientific) was used to acquire local endocardial electrograms in the left atrium. A 30-minute-long segment during AF was recorded. The animals were then cardioverted to sinus rhythm, and rapid pacing re-induced AF was recorded for 15 minutes. We computed the power spectral density (PSD) function for 30 minutes of AF. We defined “high DF” sites as those with normalized DF > 0.8 and computed the variance of DF for each electrogram. We then correlated the location of the high DF sites between the native and the re-induced AF.
DF ranged from 5.3±0.2 to 7.7±0.9 Hz for native and 5.4±0.2 to ± 7.3±0.7 Hz for re-induced AF. We also found that the high DF sites had significantly lower variance. For every 1 unit increase in DF variance, the median DF decreased by 0.195 (95% CI: (-0.246, -0.145), P-value <0.0001) unit. High DF sites in native AF correlated 87.2210.39 % with high DF sites during the re-induced AF.
High DF sites have low variability in a 30 min continuous period of AF, which makes them likely sites of AF drivers. The high DF sites are also stable after cardioversion and re-induction of AF. Low DF sites, on the contrary, have higher variability suggesting that they may arise from collisions of waves emerging from other driver sites like the high DF sites. This makes the high DF sites a potential target for AF ablation.