Re-evaluating the multiple wavelet hypothesis for atrial fibrillation

There continues to be great interest in better defining mechanisms of atrial fibrillation (AF) to improve therapy and guide ablation beyond pulmonary vein isolation. The multiple wavelet hypothesis (MWH), proposed by Moe et al¹ from canine and computer experiments and further elaborated by Allessie et al,² proposes that disordered AF can self-perpetuate by 3–6 wavelets, which spatially meander within available atrial tissue.^2,3 The MWH is an elegant hypothesis that is widely known, provides a solid foundation to describe disorganized activity in AF, and may explain observations such as why AF is more difficult to treat in patients with larger atria (ie, because there is more atrial space for wavelets to meander, AF is less amenable to treatment). Conversely, it is not clear that disorganization in AF is always self-sustaining rather than downstream from other mechanisms. Moreover, the MWH may not explain consistent gradients in frequency⁴ or electrogram patterns⁵ in AF that have been reported in patients, and it does not explain the success of pulmonary vein isolation or other ablation procedures. The Moe hypothesis arose from studies that, although classic, may have been limited by the mapping techniques available at that time.

Present study

In a study reported in this issue of Heart Rhythm Journal, Lee et al⁶ repeated the original canine studies of Moe et al,¹ using elegant high-resolution simultaneous biatrial mapping. Lee et al initiated AF in 6 adult mongrel dogs by rapid atrial pacing during vagal nerve stimulation. They recorded epicardial electrograms from 510 unipolar electrodes (0.5-mm size, 0.7-mm interelectrode distance) throughout the left atrium (n = 162), including between the pulmonary veins (n = 24), Bachmann bundle (n = 48), and right atrium (n = 276), and analyzed bipolar signals using pairs of unipoles separated by 2–6 mm. This method substantially advances the resolution of Moe et al, who sparsely mapped the right atrium using electrodes in the appendage and body. Lee et al also mapped atrial areas larger than the 10.2-cm² regions mapped by Allessie et al,² which represent a small proportion of atrial areas revealed by magnetic resonance imaging in patients with advanced AF.⁷

The key finding of Lee et al⁶ was not multiwavelet reentry in AF but sites consistent with focal sources that activated after an electrically silent period, lasted several AF cycles, and controlled surrounding atrial tissue. Focal sources repeated for a minimum of 3 cycles but ranged up to 26 cycles (average 8–10 cycles at cycle length 125 ms). There were 2.2 ± 1.2 sources in both atria, although it is not clear whether the sources were identified concurrently or sequentially. The fact that sources controlled surrounding tissue in 1:1 fashion challenges the central concept of multiple meandering and colliding wavelets. Interestingly, clinical studies have recently reported islands of 1:1 atrial activation in AF as potential ablation sites.⁸ It is remarkable that Lee et al could not find meandering wavelets. While these conclusions differ dramatically from those of Moe et al,¹ the AF milieu had a similar number of wavefronts (4.2 ± 1.2), location, and rate (7.5–9.5 Hz).⁶ Lee et al should be commended for repeating the classic studies of Moe et al with state-of-the-art, high-resolution mapping of both atria and meticulous analysis.

Future work to be done

The findings of focal sources in AF reported by Lee et al⁶ are provocative and agree in many ways with recent optical mapping⁹ and clinical studies¹⁰ of human AF. Nevertheless, some aspects of this study should be clarified. A major limitation is that Lee et al mapped only the epicardium in a small number of animals. From recent transmural optical mapping and high-resolution MRI studies of human atria in AF, Hansen et al⁹ reported that electrophysiological features may differ between endocardium and epicardium (rotational sites in the endocardium appeared focal on the epicardium). Transmural or at least bisurface mapping would thus greatly exhance this study. By extension, it could be argued that Lee et al cannot exclude AF wavelets meandering from epicardium to and from endocardium (endocardial–epicardial dissociation) as reported by Allessie et al.⁴ Another limitation of the study, although not its primary goal, is that cause and effect was not established. Future work could use ablation designed to either constrain multiple wavelets¹¹ or target localized sites¹⁰ to address this. Another technical issue is that AF foci were defined by QS and RS electrogram morphology, which in AF could be distorted by summation of multiple waves,¹² and should be validated against optical mapping. Finally, extending the study to patients during cardiac surgery would greatly expand on the impact of their findings.

Lee et al⁶ should be congratulated for an elegant study that re-evaluates the Moe hypothesis using state-of-the-art mapping. Their findings of localized sources rather than multiple wavelet reentry in the Moe studies are provocative. Further clinical translation of these findings is urgently needed and may shed much needed clarity on AF mechanisms and potential new ablation approaches.