EDITORIAL COMMENTARY The atrial neural network: Ablation minefield or strategic target? Kevin F. Kwaku, MD, PhD, Richard L. Verrier, PhD From the Cardiovascular Division, Beth Israel Deaconess Medical Center, and Harvard Medical School, Boston, MA. Background Neural control of heart rhythm is achieved through complex hierarchal inputs from both the extrinsic autonomic nervous system and the less well understood intrinsic cardiac neural network. 1,2 The latter system consists of an extensive set of neuron types, including afferent sensory fibers as well as sympathetic and parasympathetic efferents. The intrinsic neural network exerts local control of chronotropy, dromot- ropy, and inotropy mediated by ganglionated plexi (GP) in discrete epicardial fat pads initially characterized by the pioneering work of Randall, 3 Zipes, 4 and Armour, 1,2 and their respective coworkers. Several major fat pads have been located in proximity to the atria in mammalian species. The sinoatrial (SA) node fat pad is located at the caudal end of the SA node and contains many of the anterior right (AR) GP. The atrioventricular (AV) node fat pad is located at the junction of the inferior vena cava with both atria and con- tains ganglia from the inferior right (IR) GP. The superior left pulmonary vein (PV) fat pad is proximate to that PV and contains the superior left (SL) GP, as depicted in Fig. 1 of Hou et al. 5 In addition, there is yet another distinct fat pad in canines, which is located between the medial superior vena cava (SVC) and aortic root in proximity to the right pulmonary artery. 4 The human heart also contains epicardial fat pads 6 and an extensive distribution of GPs 1 . The location of these struc- tures generally corresponds to that observed in canines in their proximity to the atria and major vessels. 1,2,6 Cum- mings and colleagues 7 have recently reported on the rele- vance of an anterior fat pad, located between the base of the aorta (Ao) and pulmonary artery in humans, which may correspond to the canine SVC-Ao fat pad. Thus, while fat pad locations may not be identical, there appear to be general interspecies homologies in cardiac neural architec- ture. Functional studies indicate that disruption of the GPs has the potential for enhancing electrical inhomogeneity, lead- ing to arrhythmias. 4 In a recent clinical study, it was dem- onstrated that preservation of the anterior fat pad signifi- cantly decreases the incidence of postoperative atrial fibrillation (AF). 7 Conversely, selective ablative targeting of GPs has been advanced by some centers as a useful adjunct to pulmonary vein isolation in the treatment of AF. 8,9 These incongruous observations underscore the importance of ad- vancing our understanding of the GP network. Main findings and commentary The present study by Hou and coworkers 5 advances an emerging theme that the GPs, which exhibit extensive an- atomical interconnections, function as an interactive net- work in their influence on the SA and AV nodes. The specific hypothesis tested was that as a result of connectivity within the atrial neural network, GPs remote from the SA and AV nodes modulate the effects of GPs more proximate to these structures. The investigators studied the AR, IR, and SL GPs in anesthetized, open-chest canines. A systematic combination of electrical stimulation of the GPs was performed with a handheld bipolar electrical probe in conjunction with blockade of adjoining GPs either with lidocaine or radiofrequency ablation. They measured the responses in sinus and ventricular rates and the atrial-His intervals. From their findings, the investigators deduced that IR GP control of AV nodal dromotropy is modulated by remote inputs from both the AR and SL GP. Similarly, the AR GP control of sinus rate is modulated by remote inputs from the SL GP but not the IR GP. In addition, a weaker, direct influence of the SL GP on the SA node was observed. Overall, the authors concluded that AR GP and IR GP serve integrative roles in modulating SA and AV node function, as depicted in their Fig. 5. 5 The principle emanating from these observations is that the atrial neural network operates through an interactive hierarchy among the GPs with respect to relative influence on the SA and AV nodes. While this basic premise appears sound, a number of caveats need to be stated. The model explores the interactions among three important GPs, but the authors acknowledge that investigation of potential par- ticipation of the SVC-Ao fat pad is not incorporated in their model. This is an important limitation, since this structure appears to receive the majority of efferent vagal innervation Address reprint requests and correspondence: Richard L. Verrier, Ph.D., Associate Professor of Medicine, Harvard Medical School, Beth Israel Deaconess Medical Center, Harvard-Thorndike Electrophysiology Institute, Harvard Institutes of Medicine, 77 Avenue Louis Pasteur, Room 223, Boston MA 02115. E-mail : rverrier@bidmc.harvard.edu. 1547-5271/$ -see front matter © 2007 Heart Rhythm Society. All rights reserved. doi:10.1016/j.hrthm.2006.10.006