Author + information
- Received December 11, 2010
- Revision received January 27, 2011
- Accepted February 22, 2011
- Published online July 5, 2011.
- Chung-Chuan Chou, MD⁎,⁎ (, )
- Po-Cheng Chang, MD⁎,
- Ming-Shien Wen, MD⁎,
- Hui-Ling Lee, MD†,
- Tse-Ching Chen, MD, PhD‡,
- San-Jou Yeh, MD⁎ and
- Delon Wu, MD⁎
- ↵⁎Reprint requests and correspondence:
Dr. Chung-Chuan Chou, Chang Gung Memorial Hospital and Chang Gung University College of Medicine, 2nd Section of Cardiology, 199 North Tung-Hwa Road, Taipei 10591, Taiwan
Objectives The purpose of this study was to provide direct evidences that rotor ablation suppresses atrial fibrillation (AF) inducibility.
Background Micro–re-entrant wavefronts have been suggested to serve as sources of rapid activations during AF. Whether AF inducibility is suppressed by elimination of rotors remains unknown.
Methods We used optical mapping to study Langendorff-perfused left pulmonary vein (PV)–left atrium (LA) preparations from 13 dogs with pacing-induced heart failure. Atrial arrhythmias were induced by pacing and mapped during acetylcholine infusion (1 μmol/l). Rotors were identified from optical recordings. Epicardial ablation was performed targeting the rotor anchoring sites in preparations with sustained (>10 min) or incessant spontaneous AF. Non-rotor ablation was performed in 4 preparations. Repeated pacing was performed to test the AF inducibility after ablation.
Results Sustained AF (n = 12) and incessant spontaneous AF (n = 1) were induced after acetylcholine infusion. Pulmonary vein focal discharge was found in 9 preparations (9.2 ± 4.2 beats/s), and rotor anchoring was found at the left superior PV-LA junction in 13 preparations (9.1 ± 4.6 beats/s) and at the ligament of Marshall-PV-LA junction in 1 preparation. Epicardial rotor ablation successfully inhibited the inducibility of sustained AF in 12 of 13 preparations (p < 0.01), including 4 with the maximal dominant frequency sites located on the PV-LA junctional rotor zones (direct elimination of mother rotors). The longest AF duration was shortened significantly by rotor ablation (Wilcoxon Z = 3.60, p = 0.002, n = 13), but not by non-rotor ablation (Wilcoxon Z = 1.00, p = 0.317, n = 4).
Conclusions Epicardial ablation of the rotor anchoring sites suppresses AF inducibility. The arrhythmogenicity at the maximal dominant frequency sites is directly/indirectly suppressed by the rotor ablation.
Although pulmonary vein (PV) bursting appears to play a role in atrial fibrillation (AF) initiation and maintenance (1,2), this bursting behavior in turn depends on the atrial input and largely disappears when PVs are disconnected from the left atrium (LA) (3). We have previously shown that the left superior pulmonary vein (LSPV)-LA junction is a privileged site for anisotropic re-entry (4). Acetylcholine infusion facilitates both PV focal discharge and micro–re-entry at the LSPV-LA junction to perpetuate AF, and pharmacological suppression of PV focal discharge may inhibit the AF inducibility in a heart failure canine model (5). Whether radiofrequency ablation (RFA) targeting at the rotor anchoring sites is effective in suppressing AF inducibility remains unknown. In this study, we performed high-density optical mapping to test the hypothesis that RFA of the re-entry anchoring sites prevents AF inducibility.
The research protocol was approved by the Institutional Animal Care and Use Committees of Chang Gung Memorial Hospital and conformed to the guidelines of American Heart Association on research animal use. Thirteen beagle dogs (weight 8 to 14 kg) were used in the study.
Surgery and rapid ventricular pacing-induced heart failure
Dogs were pre-medicated with intramuscular injection of ketamine (10 mg/kg) and xylazine (4 mg/kg), intubated, and anesthetized with isoflurane. When the dogs were fully anesthetized and nonresponsive to physical stimuli, the chests were opened through a right thoracotomy. An epicardial lead was placed in the free wall of the right ventricle and connected to a Medtronic Itrel III neurostimulator (Medtronic, Minneapolis, Minnesota) for chronic pacing (250 beats/min for 1 month). Ventricular function was assessed by echocardiography at the baseline and after 1 month of pacing.
Optical mapping studies
Langendorff-Perfused PV-LA Preparation
The hearts were rapidly excised under general anesthesia. All ventricular branches were tied off, and the left coronary artery was immediately cannulated and perfused with cold cardioplegic solution (4°C). Ventricular tissue was quickly removed. The preparation was then placed in a temperature-controlled tissue bath (36.5 ± 0.5°C). It was both perfused and superfused with oxygenated Tyrode's solution (composition in mmol/l: NaCl 125, KCl 4.5, MgCl2 0.25, NaHCO3 24, NaH2PO4 1.8, CaCl2 1.8, glucose 5.5, and albumin 50 mg/l in deionized water), equilibrated with 95% O2 and 5% CO2 to maintain a pH of 7.4. Coronary perfusion pressure was regulated between 80 mm Hg and 85 mm Hg. Two hook electrodes were inserted into the LA appendage (LAA) for recording and pacing.
The preparations were stained with a voltage-sensitive dye (RH 237, Molecular Probes, Eugene, Oregon) and excited with laser light at 532 nm (4). The emitted fluorescence was filtered through a 710-nm long-pass filter and was then acquired with a charge-coupled device camera (CA-D1-0128T, Dalsa, Waterloo, Ontario, Canada) at 269 frames/s. The digital images (128 × 128 pixels) were gathered from the epicardium of the left PVs and adjacent LA over a 30 × 30 mm2 area, resulting in a spatial resolution of 0.24 × 0.24 mm2 per pixel. Motion artifacts were suppressed by 5 μmol/l cytochalasin D (Sigma Aldrich, St. Louis, Missouri). To create color maps, the average fluorescence level (F̄) over the entire data window was first calculated for each pixel. At each pixel, the change in fluorescence (the difference between fluorescence level and the F̄) at each time point was color coded to generate the color maps. Two-dimensional color maps of the membrane potential (Vm) were constructed to illustrate the Vm changes during paced rhythm and arrhythmia. Individual Vm maps showed the depolarized areas in shades of red and repolarized areas in shades of blue and were animated to show the patterns of propagation in the mapped region.
Acetylcholine infusion and pacing protocols
We infused acetylcholine (1 μmol/l) for 15 min. Atrial electrical activities were recorded continuously, and spontaneous atrial arrhythmias were mapped when possible to determine the patterns of initiation and the sources of focal discharge. We paced the LAA with 5-ms pulse width and twice diastolic threshold current. Atrial arrhythmias were induced by standard S1S2 pacing and/or burst atrial pacing protocols (pacing cycle length [CL] 50 ms, 5-ms pulse duration, and 5-mA current for 3 to 5 s, 8 to 12 times). The inducibility of sustained AF was defined as any episode of induced AF >10 min. Three episodes of AF were mapped consecutively within the first 30 s after induction. These episodes, each containing 1,000 frames of optical images, were used for dominant frequency (DF) analyses and identification of micro–re-entrant wavefronts anchoring. After AF sustained for >10 min, defibrillation was performed with 3 to 5 J biphasic shocks delivered by epicardial patch electrodes.
Stepwise catheter ablation to suppress AF
Epicardial RFA was performed in preparations with sustained AF (>10 min) and/or incessant spontaneous AF bursts after acetylcholine infusion. In the first 9 preparations, RFA was performed at the sites that fit the phase singularity clustering (by phase singularity maps) and the re-entry rotor anchoring (by activation movies). We used a thermo control 8-mm ablation catheter (St. Jude Medical, Minneapolis, Minnesota) with a temperature setting of up to 60°C and maximum power of 70 W RF energy (HAT300, Osypka, Berlin, Germany) and delivered multiple burns (2 min/time, range 3 to 6 times) until no further rotor anchoring could be induced at the ablation sites. If sustained AF was still inducible, the post-RF AF episodes were analyzed to determine if other focal mechanisms were present for further ablation. In the remaining 4 preparations, RFA was first performed at non-rotor anchoring sites (LAA), then at the rotor anchoring sites.
The hearts were fixed in 4% formalin for 1 h and stored in 70% alcohol. The proximal LSPV and ablated tissues overlying the LSPV-LA junction were selectively excised and paraffin embedded. Sections (5 μm) were stained with hematoxylin and eosin for light microscopic examinations.
Focal discharge was defined as an activation propagating centrifugally from a central site. Alternatively, an activation originating from the distal end of the PV muscle sleeve and propagating only in the direction toward the LA was also considered a focal activation. The frequency of focal discharge and micro–re-entry were calculated by the average of total number of focal discharges in 4 s, which was the duration of each episode of optical recording (6). The phase maps (7) were constructed using a time-delay embedding method by convolving the image sequence with 2-dimesional kernels, as described previously (8). Phase singularity is an area with ambiguous phase where neighboring elements exhibit a continuous progression of phase from −π to +π. Cumulative phase singularity maps were constructed by plotting phase singularities of 100 consecutive frames in the same map to determine spatial clustering of the phase singularities. To calculate the DF, the signals were subtracted by their direct current component, windowed through a flat-top window, and went through a standard Fourier transformation to obtain their spectral contents and to determine the frequency at each pixel during 4-s AF. The DF maps were constructed by plotting the DF of activation at different mapped regions, and were color coded to reveal the DF distribution during AF.
Continuous variables with normal distribution were expressed as the mean ± SD and median (first quartile to third quartile). The Kolmogorov-Smirnov test was used to test normal distribution of all continuous variables. Student t tests were used to compare the means of variables with normal distribution. Inducibility of sustained AF before and after RF ablation was tested with the McNemar test (for proportions) and the Wilcoxon signed-rank test (for means). The proportions that sustained AF could not be induced post-ablation in the 2 groups receiving RF at the LSPV-LA junction only (n = 9), and RF at the LSPV-LA junction and LAA (n = 4) were compared with Fisher's exact test. The deltas in AF duration between the 2 groups were compared with the Wilcoxon rank-sum test. A p value <0.05 was considered statistically significant.
The left ventricular ejection fraction was decreased from 55 ± 11% to 31 ± 5% (p < 0.01, n = 13) after pacing. Sustained AF could be induced in 12 of 13 preparations during acetylcholine infusion. In 1 preparation, incessant spontaneous AF bursts rather than sustained AF were induced. PV focal discharges were found in 9 preparations (frequency of 9.2 ± 4.2 beats/s), and intermittent re-entrant wavefronts anchored at the LSPV-LA junction in 13 preparations (frequency of 9.1 ± 4.6 beats/s) and at the ligament of Marshall (LOM)-LSPV-LA junction in 1 preparation. We performed epicardial RFA in 13 preparations. The inducibility of sustained AF was suppressed, that is, sustained AF was inducible in 92% (12 of 13) preparations, but only in 8% (1 of 13) of preparations after RF at the rotor anchoring sites (p = 0.001 for McNemar test), and the longest AF duration was shortened (680 ± 145 s vs. 63 ± 198 s; median [first to third quartile]: 720 ms [720 to 720 ms] vs. 5 ms [2 to 15 ms]; Z of Wilcoxon signed-rank test = 3.60, p = 0.002) by RFA at the rotor anchoring sites. On the contrary, the longest AF duration was not significantly changed (720 ± 0 s vs. 619 ± 203 s; median [first to third quartile]: 720 ms [720 to 720 ms] vs. 720 ms [416 to 720 ms]; Z of Wilcoxon signed-rank test = 1.00, p = 0.317) by RFA at the LAA. There was no difference in proportions that sustained AF could not be induced (p = 0.692 for Fisher exact test) and the deltas in AF duration (576 ± 275 s vs. 710 ± 12 s; median [first to third quartile]: 708 ms [450 to 717 ms] vs. 713 ms [698 to 720 ms]; Z of Wilcoxon rank-sum test = −0.85, p = 0.395) between the 2 groups receiving RFA at the LSPV-LA junction only and RFA at the LSPV-LA junction and LAA, respectively.
Effects of epicardial ablation on acetylcholine-induced AF
Maximal Dominant Frequency Within Area Harboring the LA Rotor at the LSPV-LA Junction
In 5 preparations (1 with and 4 without inducible PV focal discharge), including 4 with sustained AF and 1 with incessant spontaneous AF bursts, the maximal DF (DFmax) zone was co-located with the rotor area adjacent to the LSPV-LA junction. Figure 1 shows an example. The Vm recordings at LA rotor zone and PV (Fig. 1, sites “a” and “b,” respectively) during the initiation of burst pacing-induced AF are shown in Figure 1B. The DF maps sampled during the period of AF initiation (Fig. 1C) and the sequentially sampled 2 AF episodes (Fig. 1D) show that the DFmax area was constantly located adjacent to the LSPV-LA junction where the rotor anchored. Figure 1E shows the isochronal map of the rotor (left) and its corresponding phase singularity on the phase map (white arrow, right). The re-entry core meandered along the LSPV-LA junction during AF (Online Video 1), resulting in dense phase singularity accumulation in this area (Fig. 1F, yellow arrow) where we performed RFA. Before ablation, burst pacing could induce sustained AF, and cardioversion (red arrows) was needed to convert it (Fig. 1Ga). After ablation, burst pacing could only induce nonsustained AF (6.0 ± 4.7 s, n = 9; the longest, 15.4 s) (Fig. 1Gb). The RFA suppressed the inducibility of sustained and/or incessant spontaneous AF in 4 preparations without PV focal discharge in this subgroup. In the RFA-refractory preparation, the mean DFmax of AF was higher than that of the other 4 preparations (28.4 ± 1.4 Hz vs. 16.1 ± 4.9 Hz, p < 0.05). The RFA at the rotor anchoring sites failed to suppress the AF inducibility but significantly decreased the mean DFmax of AF (28.4 ± 1.4 Hz vs. 22.1 ± 1.4 Hz, p = 0.01). Also, multiple wavefronts that came from the border of the mapped field were noted in the residual AF episodes.
Maximal Dominant Frequency Within the PV
In 7 preparations with inducible PV focal discharge, the site of DFmax was on the discharging PV (example, Fig. 2). Pseudo-electrocardiograms in Figure 2 subpanels Aa and Ab (the expanded portion in the red rectangle, Aa) show S1S2 pacing induced sustained AF (S1S2 = 70 ms). Figure 2C shows the Vm recordings at the LA (site a) and LSPV (site b) labeled in Figure 2B during the initiation of AF. The red asterisks indicate LSPV focal discharges. The DF map sampled during the period of AF initiation (Fig. 2D) shows the DFmax site was at the LSPV (17.5 Hz), corresponding to the focal discharging site shown in Figure 2E (black arrow, left). The micro–re-entrant wavefront meandered along the LSPV-LA junction (right, Fig. 2E), resulting in dense phase singularity accumulation in this area (Fig. 2F, yellow arrow) where we performed RFA. As shown in Figure 2G (pseudo-electrocardiograms), only short runs of atrial arrhythmias were induced by S1S2 pacing (subpanel Ga) and burst pacing (the longest duration 3.17 s) (subpanels Gb and Gc) after ablation.
The RFA targeting the rotor also influences the PV focal discharge generation and thus AF perpetuation (Fig. 3).Fig. 3A shows the pre-RFA frame shots corresponding to the period labeled in Figure 2C (red bar) (Online Video 2). The LSPV focal discharge (frame 1,524 ms) induced micro–re-entry, circling around the anterior LSPV-LA junction, and lasted for 13 beats (mean CL 61 ± 9 ms). The wavefronts from LA rotor were feedback activations to LSPV (frames 1,612 ms, 1,660 ms). Figure 3B represents the Vm tracings of post-RFA nonsustained AF corresponding to Figure 2Gc (red bar). The DF map shows the DFmax site was at the LSPV (7.7 Hz) (Fig. 3C, site b). Figure 3D shows the post-RFA frame shots corresponding to the period labeled in Figure 3B (red bar) (Online Video 3). A wavefront from the left border of the mapped field propagated to LSPV, followed by a PV focal discharge that detoured around the ablation line (white line, frame 2,384 ms) and failed to form micro–re-entry at the LSPV-LA junction. Then the other wavefront from the left border of the mapped field activated LSPV, which was repolarized earlier, met the ablation line (white line, frame 2,452 ms) and extinguished. Without the electrical feedback from the fast-activating LSPV-LA junctional rotor, the electrical feedback to LSPV was reduced, and LSPV focal discharge became quiescent, thus AF was stopped. The inducibility of sustained AF was suppressed by RF ablation in all 7 preparations in this subgroup.
Maximal Dominant Frequency Between 2 Rotors
In 1 preparation, 2 rotors were induced during pacing-induced atrial arrhythmias. Figure 4A shows the pseudo-electrocardiograms of burst pacing-induced sustained atrial arrhythmia, which spontaneously interchanged atrial tachycardia with AF. Atrial tachycardia was maintained by sustained re-entry circling around the LOM-LSPV-LA route (Fig. 4C). During AF, LSPV-LA junctional rotor occurred intermittently, interacting with wavefronts from the LOM-dependent rotor and LAA (Fig. 4D, right subpanel; Online Video 4). Figure 4E shows frame shots demonstrating the interplay between the 2 rotors. The LOM-dependent rotor activated faster (mean CL 75 ± 5 ms, n = 53) than the LSPV-LA junctional rotor (mean CL 84 ± 5 ms, n = 26). The wavefronts from the 2 rotors (frame 1,852 ms) collided with each other, leading to the formation of phase singularities at the broken endpoints of the daughter wavelets. As shown in Figure 4F, phase singularities accumulated not only at the rotors anchoring sites (yellow arrows), but also at the sites where wavefronts collided (white arrow). This area was 1 of the DFmax sites (Fig. 4G, site b) with more fragmented Vm potentials (Fig. 4H, red arrows) than those at sites a and c (labeled in Fig. 4G). The RFA targeting the LSPV-LA junctional rotor did not inhibit AF inducibility, but significantly decreased the mean DFmax of AF (13.8 ± 1.2 Hz vs. 8.9 ± 2.0 Hz, p = 0.01). The LOM-dependent rotor survived to maintain the residual AF (Fig. 4I). The DF map of the residual AF (Fig. 4J) shows the DFmax site was at the LOM-dependent rotor zone. Further ablation targeting the LOM-dependent rotor suppressed the AF inducibility.
Epicardial ablation on nonrotor anchoring sites
To exclude the possibility that the AF-suppressing effects were from the RFA lesion itself rather than the RFA location, we performed 2-stage RFA (RFA at the LAA first, then at the rotor anchoring sites) in 4 preparations (Fig. 5).Figure 5B shows the Vm recordings during the initiation of AF at the LA (site a) and LSPV (site b) labeled in Figure 5A. Figure-of-8 re-entry at the LSPV-LA junction (Fig 5C, left; Online Video 5) and LSPV focal discharge (Fig. 5C, black arrow, right) were found during AF. The phase singularity map showed clustering of phase singularities at the LSPV-LA junction (Fig. 5D, yellow arrow). The DFmax zone was located at the discharging LSPV (Fig. 5E, left). Sustained AF was inducible at baseline (Fig. 5Fa). The RFA was first performed at the LAA (Fig. 5G, RF1), but LSPV-LA junctional rotor (Online Video 6) and sustained AF were still inducible (Fig. 5Fb), and the DFmax location was unaltered (Fig. 5E, right). After performing RFA targeting the rotor (Fig. 5G, RF2), AF inducibility was suppressed (Fig. 5Fc). It suggests that rotor elimination plays a critical role in AF suppression. The histological studies showed transmural myocardial coagulative necrosis evidenced by pyknotic nuclei and acidophilic cytoplasm at the RF site (Fig. 5H).
The following major findings were obtained in this study: First, epicardial ablation at the rotor anchoring sites inhibits the inducibility of acetylcholine-induced sustained AF in all 4 preparations without discharging PVs and in 8 of 9 preparations with PV focal discharges, that is, elimination of rotors can suppress the DFmax sites directly or indirectly. Second, the DFmax may be high for reasons other than active activation, which may account for ineffective AF ablation at the DFmax sites.
Mechanisms of drivers perpetuating AF
Drivers mean high-frequency sources that maintain fibrillatory activity. Experimental (8) and clinical (9) AF mapping studies have recognized the presence of rapid activations emanating from the PV regions, which play a role in maintaining AF by demonstrating the DFmax at these sites. Ablation of high-frequency sites results in AF termination in a significant proportion of patients with paroxysmal AF, even if the underlying mechanisms remain unresolved (9). We previously showed that PV focal discharge sites were co-located with the DFmax sites during acetylcholine-induced AF (5). Infusion of ryanodine and thapsigargin suppressed PV focal discharge and resulted in noninducibility of sustained AF in 4 of 6 preparations, suggesting that inhibition of intracellular calcium accumulation may suppress the triggers of AF within the PVs. However, sustained AF was still inducible in 2 preparations, and intermittent LSPV-LA re-entrant wavefronts, but not PV focal discharge, were found in the residual AF episodes, implying that PV-LA re-entry also plays an important role in AF perpetuation. Saunders et al. (9) reported that ablation at the PV regions harboring the DFmax sites resulted in AF termination, and Atienza et al. (10) indirectly demonstrated that re-entrant mechanism underlies these DFmax sites by intravenous administration of adenosine. Consistent with their findings, we directly demonstrated that mother rotors underlay the mechanisms at the DFmax sites, and applying RFA to eliminate mother rotors successfully suppressed the inducibility of sustained AF in 4 preparations without inducible PV focal discharge. In the meantime, the DFmax sites were found in the discharging PVs in 7 preparations with PV focal discharge. Elimination of the LSPV-LA junctional rotors also suppressed AF inducibility in these preparations by decreasing electrical feedback to the discharging PVs. This mechanism mimics ibutilide's anti-AF actions by suppressing re-entrant excitations and the interplay between LA and PV without directly suppressing PV focal discharge (11).
Dominant frequency mapping is predicated on the hypothesis that finding the highest activation sites can identify driver regions of AF; thus, the adoption of DF measurements with proper interpretation can potentially provide additional information to improve AF treatment. However, the existence of double and/or complex fractionated potentials was shown to have the potential to affect the DF measurement. These high DF sites can be considered to reflect a pivot point of a re-entrant wavefront or collision of multiple wavefronts (12). As shown in Figure 4, the DFmax sites might reflect driven rather than driver behavior. A better understanding of the mechanisms leading to localized highest activation rate at the DFmax sites is warranted before DF mapping can be used efficiently as a clinical electrophysiological tool to terminate AF.
We did not do 3-dimensional simulation studies, which help discern mechanisms involving the complex anatomy. We only mapped the epicardial surface of Langendorff-perfused PV-LA preparations in a specific model. We did not assess the potential contribution of autonomic ganglia near the PV. Therefore, caution is advised when extrapolating our model to clinical AF.
The authors are indebted to Dr. Peng-Sheng Chen and Dr. Shien-Fong Lin for their invaluable comments and critique of the study.
For supplementary figures and Videos 1 to 6, please see the online version of this article.
This work was supported by CMRPG33016-3 to Dr. Chou. The authors have reported that they have no relationships to disclose.
- Abbreviations and Acronyms
- atrial fibrillation
- cycle length
- dominant frequency
- maximal dominant frequency
- left atrium
- left atrial appendage
- ligament of Marshall
- left superior pulmonary vein
- pulmonary vein
- radiofrequency ablation
- membrane potential
- Received December 11, 2010.
- Revision received January 27, 2011.
- Accepted February 22, 2011.
- American College of Cardiology Foundation
- Chou C.C.,
- Nihei M.,
- Zhou S.,
- et al.
- Wu T.-J.,
- Ong J.J.C.,
- Chang C.-M.,
- et al.
- Sanders P.,
- Berenfeld O.,
- Hocini M.,
- et al.
- Atienza F.,
- Almendral J.,
- Moreno J.,
- et al.
- Chou C.C.,
- Zhou S.,
- Tan A.Y.,
- Hayashi H.,
- Nihei M.,
- Chen P.S.