Spatial Distribution of Fibrosis Governs Fibrillation Wave Dynamics in the Posterior Left Atrium During Heart Failure
Heart failure (HF) commonly results in atrial fibrillation (AF) and fibrosis, but how the distribution of fibrosis impacts AF dynamics has not been studied. HF was induced in sheep by ventricular tachypacing (220 bpm, 6 to 7 weeks). Optical mapping (Di-4-ANEPPS, 300 frames/sec) of the posterior left atrial (PLA) endocardium was performed during sustained AF (burst pacing) in Langendorff-perfused HF (n=7, 4 μmol/L acetylcholine; n=3, no acetylcholine) and control (n=6) hearts. PLA breakthroughs were the most frequent activation pattern in both groups (72.0±4.6 and 90.2±2.7%, HF and control, respectively). However, unlike control, HF breakthroughs preferentially occurred at the PLAs periphery near the pulmonary vein ostia, and their beat-to-beat variability was greater than control (1.93±0.14 versus 1.47±0.07 changes/[beats/sec], respectively, P<0.05). On histological analysis (picrosirius red), the area of diffuse fibrosis was larger in HF (23.4±0.4%) than control (14.1±0.6%; P<0.001, n=4). Also the number and size of fibrous patches were significantly larger and their location was more peripheral in HF than control. Computer simulations using 2-dimensional human atrial models with structural and ionic remodeling as in HF demonstrated that changes in AF activation frequency and dynamics were controlled by the interaction of electrical waves with clusters of fibrotic patches of various sizes and individual pulmonary vein ostia. During AF in failing hearts, heterogeneous spatial distribution of fibrosis at the PLA governs AF dynamics and fractionation.
Atrial fibrillation (AF) is the most common sustained arrhythmia in adults and is often associated with congestive heart failure (HF), which induces extracellular matrix remodeling involving atrial fibrosis and dilatation.1,2 Yet, how fibrosis contributes to AF mechanisms has not been thoroughly investigated. Previous experimental and clinical work has emphasized a role for fibrosis in arrhythmia maintenance.3,4 For example, interstitial fibrosis has been implicated in abnormalities during atrial pacing in dogs,3 and patchy fibrosis has been shown to cause activation delays in the human ventricle.5 In addition, chronic AF patients have an increased propensity to develop fibrosis at the posterior left atrium (PLA).4 However, the relation between AF dynamics at the PLA and the percentage and distribution of fibrosis in HF remains unexplored. Here we have characterized how AF frequency and dynamics on the endocardial surface of the PLA and pulmonary vein ostia (PVO) are affected by the amount, type (ie, diffuse versus patchy), and spatial distribution of fibrosis in failing hearts. It is our hypothesis that patchy rather than diffuse fibrosis contributes to wavebreak and intramural rotor formation, and governs the overall AF frequency and dynamics.
Materials and Methods
Heart Failure Model
All procedures were approved by the SUNY Upstate Medical University Committee for Humane Use of Animals. HF was induced in 7 sheep (15 to 25 kg) as previously described6 (see supplemental materials, available online at http://circres.ahajournals.org).
Isolated Heart Preparation
Both HF (n=10) and control sheep (n=6) were anesthetized (Na pentobarbital, 35 mg/Kg iv). The chest was opened through a midsternal incision. The heart was excised, placed in cardioplegic solution, and connected to a Langendorff apparatus for continuous perfusion with Tyrode’s solution at 200 mL/min (36 to 38°C; pH: 7.4; 95% O2, 5% CO2). The PLA endocardial surface was exposed through a minimal surgical opening in the left atrial appendage (LAA) avoiding any visible coronary branches. In Figure 1A, the incision lines are marked in a representative example (left panel) and the mapped area of the PLA that includes the 4 PVO is shown (right panel). In a recent study,7 we demonstrated that this procedure enables optical mapping of the PLA endocardium without any apparent damage to the circulation, and with AF frequencies and dynamics comparable to those observed in intact isolated hearts. In 7 HF and 6 control hearts, AF was initiated by burst pacing in the presence of 4 μmol/L ACh. In 3 additional HF hearts, nonsustained AF and atrial tachycardia (AT) were induced in the absence of ACh (see supplemental materials).
Optical Mapping Technique
This technique has been detailed elsewhere8 (see also supplemental materials). Briefly, no motion uncoupler was used (the separator exposing the endocardial surface mechanically restrained the PLA). After a bolus injection of 5 to 10 mL Di-4-ANEPPS (10 mg/mL), one CCD camera (DALSA, CA-D1-0128T-STDL) recorded fluorescence changes from an area of ≈5 cm2 at 300 frames/sec to obtain 5-second movies (128×128 pixels).
Spectral Analysis, Dominant Frequency
Dominant frequency (DF) maps for each optical movie were calculated by applying a Fast Fourier Transform (FFT) to the fluorescence signal recorded at each pixel.9 To assess the location of the maximal DF (DFmax) domain at the PLA we superimposed the DF map and the corresponding picture of the field of view according to anatomical landmarks (PVO).
Histological analysis of fibrosis was obtained from transmural PLA slices of HF and normal hearts (see supplemental materials for details).
Activation patterns analyses are described in detail in the online supplement.
2D computer models of characteristic PLA sections were developed based on realistic tissue and fibrosis geometry obtained from the histological study of control and failing hearts. Control and HF membrane ion kinetics of the myocytes and fibroblasts in the models were based on human atrial properties (for details, see supplemental materials).
See online supplement for details on statistical methods. All results were expressed as mean±SEM; P<0.05 was considered statistically significant.
Clinical, Electrocardiographic, and Echocardiographic Findings
All HF sheep manifested variable degrees of anorexia, lethargy, tachypnea, and gain of body weight, as well as episodes of atrial tachyarrhythmia at 182±6.0 bpm. As shown in supplemental Table I, significant clinical and echocardiographic hallmarks of systolic congestive HF were present in all HF sheep.
DF Distribution on the PLA
Figure 1 describes the distribution of DFs in the PLA and other regions of the atria. As shown in panel B, during sustained AF, the frequency of activation at the PLA was significantly larger than at the Bachmann bundle (BB) and the right atrial appendage (RAA) in both HF and control (P<0.05). However, as previously,10–12 regional DFmax values at all locations (PLA, BB, and RAA) were significantly lower in the HF group compared with the control group (11.5±1.3 versus 21.1±1.1 Hz at PLA; 8.1±0.7 versus 17.0±1.3 Hz at BB; 8.0±1.0 versus 13.6±2.0 Hz at RAA; *: P<0.05). Representative examples of DF maps (left panels of Figure 1C and 1D for control and HF, respectively) with single pixel recordings (left traces) and bipolar electrograms (right traces) clearly support such a hierarchical organization. As shown by the maps, in control the DFmax domain (red) spanned the center of the PLA, whereas in HF DFmax was shifted toward the left inferior PVO. In control hearts, the DFmax area was 1.3±0.4 cm2, or 25.4±8.4% of the mapped PLA area. Superposition with an anatomical picture of the PLA similar to that in panel A revealed that in control the DFmax domain extended over the left PVO in 2/6 experiments and over both left PVO and center of the PLA in 4/6 experiments. In contrast, in HF, the DFmax domain spanned an area of 1.1±0.3 cm2 (17.6±5.5% of the PLA, P=0.75) and localized mainly to the margin of the PLA, enclosing one or several PVO (5/7 experiments), or extended over the entire PLA (2/7 experiments). Interestingly, in all HF experiments, the signals recorded in the DFmax area were more fractionated than those recorded in other areas (panel D, trace c compared with trace d). Conversely, the signals from the DFmax area in control hearts were the most regular (panel C, trace a compared with b), in agreement with our recent report.7
Patterns of Propagation
As depicted in Figure 2A, during AF the most common pattern of activation was of a breakthrough type; only some examples of rotors were observed (2 in control and 1 in HF). For all the observed rotors, the period of rotation was equal to 1/DFmax (58.3±8.3 ms in control and 79.9 ms in HF with 4 μmol/L ACh and 217 ms without ACh). Precisely, we observed 2 rotors in 2 different control animals, 1 rotor in one failing heart (in the presence of ACh 4 μmol/L) and one long-lasting rotor in the absence of ACh in one failing heart (see supplemental Movie I). The duration of the rotors varied widely: 2 examples with 1 to 2 rotations, 1 example with 2 to 3 rotations, and 1 example of a long lasting reentrant activity for several minutes. As for the breakthroughs, PLA outward breakthroughs were more frequent than inward breakthroughs during AF in both HF and control (P<0.05), suggesting a driving role of the PLA in both groups. Diagrams of the spatial distribution of the breakthroughs are shown in Figure 2B for control and 2C for HF. Because of variations in heart dimensions, the diagrams metric is normalized to the distances between PVO. AF breakthrough waves in HF hearts tended to cluster peripherally, in closer vicinity to the PVO, compared with breakthrough waves in control (see supplemental Movies II and III for an example in control and HF, respectively). This observation is substantiated by comparing the radii of the areas with 25%, 50%, and 75% breakthroughs in panels B and C. Indeed, the area encircling 50% of HF breakthroughs was found to be on average about 30% larger than control (P<0.01).
Histological Changes in the PLA
We evaluated the extent of remodeling and the distribution of fibrosis in the PLA of 4 HF and 4 control specimens by histological analysis using picrosirius-red staining. As shown in panels B and D of Figure 3, myocytes were larger in the HF group, as expected, and were often surrounded by increased extracellular matrix (ECM) compared with the control group (panels A and C). As previously observed in the ventricles,5 2 distinct patterns of fibrosis were distinguished in both the center (panels A and B) and periphery (near the PVO; panels C and D) of the PLA: diffuse (ie, small and unconnected deposits of fibrous tissue) or patchy (ie, larger clusters of interconnected compact fibrous tissue). Further, quantification of the total amount of fibrosis by measuring the spatial extent of picrosirius-red stains revealed an overall significant ECM increment in the HF group compared with control at all 5 locations examined (23.4±0.4 versus 14.1±0.6%, P<0.001; Figure 3E). In addition, as shown in representative topographic maps of picrosirius-red staining of HF hearts (Figure 4A), the areas of patchy fibrosis were larger in the periphery (right panel) than the center of the PLA (left panel). As shown in Figure 4B and 4C, quantitative analysis of the area of fibrotic obstacles in all PLA specimens revealed a gradient in the amount of fibrosis in the HF specimens with greater ECM deposition in the periphery of the PLA. In 4 HF animals the average area of fibrosis in the periphery was significantly larger than in the center (0.163±0.052 mm2 versus 0.036±0.003 mm2, P<0.01) attributed mainly to the presence of a few patches with areas larger than 1 mm2 (Figure 4B and 4C). The corresponding areas in control animals were 0.052±0.01 and 0.038±0.003 mm2 for PLA periphery and center, respectively (P<0.01), with a remarkable absence of patches larger than 1 mm2.
Propagation of AF Waves
Figure 5A presents 4 consecutive activation maps superimposed on an anatomical picture of the respective PLA in control (upper panels) and HF hearts (lower panels). In control, the direction of activation (black arrows), as well as the breakthrough sites, were highly recurrent from one AF wave to the next (maps 1 to 4). In contrast, the waves changed origin and direction on a beat-to-beat basis in the failing heart (maps 1′ to 4′). Classification of PLA waves into 7 subgroups based on their pattern of propagation (see Methods) allowed calculation of the rate-normalized changes in the AF wave patterns. Figure 5B summarizes such beat-to-beat changes and demonstrates a significantly larger degree of variability in HF than control (1.93±0.14 versus 1.47±0.07 changes/[beats/sec], P<0.05). Further, despite the lower activation frequency in HF hearts, we observed a tendency toward an increased number of singularity points in the DFmax domains or at their borders (data not shown) in HF (4.4±2.2/experiment, range 0 to 13) compared with control (0.5±0.3/experiment, range 0 to 2, P=0.13).
PLA Tissue and Fibrosis
As illustrated in Figure 6, we developed 2-dimensional computer models of transmural PLA sections incorporating human atrial ionic models for control and HF conditions,13 3 different realistic PLA boundary geometries (G-I, G-II, and G-III) and 3 different spatial distributions of fibrosis (patchy, diffuse heterogeneous, and diffuse homogeneous).3 Please refer to the supplemental materials for a detailed description of the methods used.
Intramural Propagation Patterns
Figure 7 illustrates how HF may affect propagation. For this set of simulations a comparison between control and HF dynamics was done by using the same boundary geometry of HF PLA (G-I, see Figure 6), while changing the fibrosis distribution. This strategy was preferred over using the actual anatomical geometry of the control heart (G-III), to focus only on the functional effect of the difference in the fibrotic tissue quantity and architecture between the 2 hearts. In control, S1-S2 stimulation induced intramural reentry (white arrows) at 23.4 Hz (Figure 7A left, supplemental Movie IV). This rotor drifted rapidly toward the right and terminated at the PLA boundary after 4 rotations, without any apparent conduction impairment, as shown by the endocardial (red dashed line) time-space plot (TSP) that was used to compare these data with the experimental results. In contrast, in HF, S1-S2 induced a slower intramural reentry (5.9 Hz) around a central patchy fibrotic area that perpetuated for the entire 2-second episode (Figure 7B left, supplemental Movie V). Moreover, the slower reentry in the HF model generated waves that fragmented when colliding with fibrotic obstacles (white dots). The TSP shows an endocardial breakthrough at location 1 and a delay in the septal-to-lateral propagation at location 2, produced by a large fibrous patch.
To address the possibility of an underlying focal source for the endocardial breakthroughs (see Figures 2 and 5⇑), artificial pacing was applied to the center of the model (pulse symbol) at 6 and 20 Hz, for the HF and control conditions respectively, to represent experimental activation frequencies. In control (Figure 7A, right; supplemental Movie VI), the regularly paced waves propagated uninterrupted, producing a stable endocardial breakthrough site (location 3). On the other hand, as depicted in Figure 7B (right panels, supplemental Movie VII), HF resulted in propagation disturbances with singularity points clustered about fibrotic obstacles. For example, a wave attempting to cross a densely fibrotic region from left to right underwent unidirectional block after the first stimulus (black arrow at 519 ms), resulting in sustained reentry (5.9 Hz) around a large patch of fibrosis and repetitive endocardial breakthroughs at location 4. Then, at 1740 ms, unidirectional block closer to the septum (not shown) gave rise to a much faster rotor (21.5 Hz) that persisted for the remainder of the simulation. In this case, large propagation delays induced by fibrotic obstacles near the lateral wall led to endocardial breakthroughs at location 5. Also, in accordance with the experimentally observed variability in HF breakthrough patterns, the onset of the second intramural reentry shifted the location of the endocardial breakthrough from site 4 to 6. Overall, the simulations presented in Figure 7 predict that regardless of whether the AF mechanism is reentrant or focal, the large size of the fibrotic patches in HF is the major factor responsible for the decreased rate of AF waves.
Peripheral PLA and PV Activity
We further examined the relationship between fibrosis distribution and endocardial breakthrough pattern in a different geometry model (PLA geometry G-II, Figure 6), constructed from a transmural slice that included a PVO of a failing heart (Figure 8A). We decided to focus specifically on the LIPV ostial area because of our experimental observation that PVOs, characterized by large patches of fibrosis, are the preferential region for breakthroughs (see Figures 2 and 4⇑). Figure 8B presents snapshots of the electrical activity at selected time frames. S1-S2 led to intramural figure-of-eight reentry (22.7 Hz) in an area where fibrosis was sparse (frames 94 and 122 ms). Eventually, after 2 full rotations, the left arm of the figure-of-eight shifted toward the LIPV. This resulted in counterclockwise reentry around the PVO at a slower frequency of 5.8 Hz (frames 288, 359, and 423 ms and supplemental Movie VIII). Concomitantly, the right arm of the initial figure-of-eight generated a wave that traveled downward and to the left between a very elongated fibrous tissue septum and the epicardium. This wave vanished when it collided with the lowermost border of the sheet, which rendered the PV reentry dominant. It is noteworthy that the two different reentry frequencies that were measured in this simulation were similar to those obtained for the first geometrical model, in which the HF condition and pacing stimulation were applied. Also, similar to the previous simulation, the rapid change of reentry pattern was accompanied by a rapid change of endocardial breakthrough location, as can be seen in the endocardial TSP of figure 8B (transition from location 1 to 2). Two representative examples of local electrograms (a and b) are shown in Figure 8C, demonstrating fractionation and irregularity of the membrane potential of myocytes near fibrotic tissue. For comparison, we used a model with the same HF ionic properties and tissue geometry, but only diffuse fibrosis (supplemental Movie IX). In this case a reentrant wave formed after S1-S2 stimulation. It did not anchor but drifted toward the PLA center where it terminated after 5 rotations by collision with the epicardial boundary.
We evaluated the consequences of HF remodeling on PLA AF dynamics. Specifically, we examined how the heterogeneous HF architecture of PLA fibrosis alters AF wave propagation. Our work represents a substantial extension of previous studies demonstrating that fibrosis provides a substrate for AF14 because of atrial conduction disturbances,3,15 and showing that AF dynamics in HF differs from that in normal hearts.10 A major result of these combined experimental-simulation results was that whether the mechanism of atrial fibrillation is reentrant or focal, the large size of the fibrous patches in heart failure provides the major factor responsible for different dynamics of AF waves in HF versus control hearts. Here, we demonstrate for the first time that (1) during AF, endocardial breakthroughs waves, that are by far the most common activation pattern at the PLA in both normal and HF groups, are more variable in HF hearts and distribute more peripherally, closer to the PVO. (2) The amount, distribution, and type of fibrosis are significantly different in HF and control hearts. Importantly, in HF the architecture of fibrosis in the PLA consists primarily of patches with the larger fibrotic obstacles found preferentially in the vicinity of the PVO. (3) Both large PLA fibrous patches and the PVO act to anchor reentrant circuits and impair wave propagation to generate delays, wavebreaks and signal fractionation.
Frequency of Activation and Location of AF Sources
Our experiments demonstrate that PLA sources maintain AF in both normal and HF hearts. In both groups, AF waves were visualized as breakthroughs traveling from inside to outside the field of view (Figure 2) with a significantly higher DF in the PLA compared with other atrial areas (Figure 1). These data corroborate previous animal and human findings10,12,16–18 confirming that the PLA plays a key role in maintaining AF activity. Specifically, Wu et al16 have shown in patients in permanent AF and organic heart disease that epicardial AF dynamics was characterized by rapid repetitive activations originating near the PVOs.16 Also, our numerical simulations show that the activity in HF hearts revolved around obstacles, formed by either non-excitable patches of fibrosis or one of the PVO, with a cycle length that depended on the obstacle size (Figures 7B and 8⇑), and was found significantly larger in HF.
Mechanism of AF Maintenance
Both reentry and focal activities at or near the PV area have been shown to sustain AF in the atria of failing dog hearts,10,11,18,19 and in the presence of ACh.20–22 Our experiments demonstrate that the periostial area (PLA periphery) harbors the largest fibrotic obstacles (Figures 3 and 4⇑), the most breakthrough sites (Figure 2), and the highest frequency domains. Furthermore, our computer simulations illustrate that the largest fibrotic patches and the PVO are potential anchoring sites for “micro-anatomical” reentry. Thus, our results suggest that the fibrillatory activity in our experiments was maintained by intramural reentry centered on fibrotic patches and that it appeared at the PLA as breakthroughs. Additional measurement of PLA myocardial thickness, transillumination experiments, as well as a 3D simulation of an intra-atrial scroll presented in the supplemental materials further support this assertion. Nonetheless, we cannot fully exclude a spontaneous pacemaker or triggered activity mechanism possibly interacting with an evolving reentry, as a source of AF. This was supported by the simulated triggered activity in our numerical simulations that yielded reentry impulse propagation disturbance and anchoring shift (see Figure 7). Altogether, in HF, the remodeled substrate of the PLA seems to be more important for maintaining AF than the nature of the source, whether reentry or triggered activity.
AF Dynamics in HF and Control: Role of Fibrosis Amount and Architecture
Both human and animal studies have described the changes of AF dynamics induced by HF. For instance, local conduction abnormalities attributable to interstitial fibrosis have been reported in a HF model in dogs,3 and patients in chronic AF have demonstrated increased interstitial fibrosis in the PLA in comparison with the LAA and also with the PLA of sinus rhythm patients.4 However, to the best of our knowledge, no previous study has addressed the role of the architecture of interstitial fibrosis on PLA impulse propagation during AF. We demonstrate here for the first time that in the PLA major changes in AF dynamics are caused by an increased amount and a different architecture of fibrosis caused by HF. Interestingly, the fibrotic patches govern AF dynamics at the PLA by playing various roles that are sometimes conflicting in nature. First, because of their ability to act as micro-anatomical obstacles they have the propensity to anchor reentrant sources. However, the patches need not be confluent to be the central pivoting attachment of reentry. In additional simulations presented in the supplemental materials, we demonstrate that a simple heterogeneous distribution of diffuse fibrosis can attach a reentry (see supplemental Figure VIA and VIB and supplemental Movies X to XIII) at a specific PLA location. Second, the existence of large fibrotic patches increases the likelihood of low frequency reentrant activities. As shown in Figures 1B and 4⇑B, we observed experimentally larger fibrotic obstacles in failing hearts and correspondingly a lower DFmax. Such observations are supported by the numerical experiments shown in Figure 7A and 7B. Third, the patches of fibrosis increase dramatically the variability and complexity of the patterns. They are, for instance, responsible for propagation delays and unidirectional blocks (see Figure 5A and 5B and Figure 7B). However, it should be noted that the fact that the DFmax is lower in HF than in control does not exclude some short transitional episodes of fast reentrant activity. In fact, their ability to anchor to obstacles of different size and shape make transient reentrant sources of various frequencies likely to appear sequentially (see Figures 7B, right panel and 8B and supplemental Movies VII and VIII) or even to coexist (see supplemental Figure VIA and supplemental Movie X). Fourth, in good accordance with the experimental results (Figure 2B and 2C), the simulations predict that the presence of PV ectopic activity related to stretch-related lengthening of the myocytes might reproducibly initiate a reentrant circuit that would anchor to fibrotic patches adjacent to the PVO.
Whereas AF ablation for HF patients has been proven feasible,23 the difference of AF wave dynamics at the PLA in those patients had not been previously explored. In this work, we present experimental and computational evidence suggesting that the sites of AF sources and their related endocardial breakthrough sites appear toward the periostial area of the PLA, close to large fibrous patches. These data open a new perspective for improvement of radiofrequency catheter ablation procedure for HF patients, suggesting that in HF the outer PLA and areas with large fibrotic obstacles are critical for AF maintenance.
Sustained AF was induced in the presence of 4 μmol/L ACh. In the absence of ACh in failing hearts we usually obtained AT and nonsustained AF episodes. However, detailed analysis of the data obtained in 3 hearts demonstrated that AF-AT dynamics in the absence of ACh were very similar to those in the presence of 4 μmol/L ACh. As shown in supplemental Figure II, the majority of breakthrough sites were located at the periphery of the PLA. In the same Figure (see also supplemental Movie I), we also demonstrate that during sustained AF in the absence of ACh a stable rotor anchored in the vicinity of the LSPV ostium.
Although we show that fibrosis plays a strong role in determining AF dynamics we cannot exclude additional mechanisms of frequency slowing and of other AF dynamics changes in HF hearts, such as anisotropy, heterogeneous coupling changes as suggested by Qu et al,24 CHF-related ionic changes,25 or reduced IKACh density in the HF sheep atria (similarly to what was reported in humans26). However, as demonstrated by our simulations in which the same ACh level and IKACh density were used for both HF and control and by experiments without ACh, fibrous patches were sufficient to decrease dramatically the frequency of activation.
We thank Jianling Deng, Jiang Jiang, and Sharon E Chase for their technical assistance.
Sources of Funding
This work was supported by grants from the NHLBI (P01-HL39707, R01-HL70074, R01-HL60843, to J.J.; R01-HL087055, to J.K.), American Heart Association postdoctoral fellowships (to J.K.), and SDG (0230311N, to O.B.).
↵*These authors contributed equally to the study.
Original received April 10, 2007; revision received July 27, 2007; accepted August 8, 2007.
Li D, Fareh S, Leung TK, Nattel S. Promotion of atrial fibrillation by heart failure in dogs: atrial remodeling of a different sort. Circulation. 1999; 100: 87–95.
Corradi D, Callegari S, Benussi S, Maestri R, Pastori P, Nascimbene S, Bosio S, Dorigo E, Grassani C, Rusconi R, Vettori MV, Alinovi R, Astorri E, Pappone C, Alfieri O. Myocyte changes and their left atrial distribution in patients with chronic atrial fibrillation related to mitral valve disease. Hum Pathol. 2005; 36: 1080–1089.
Kawara T, Derksen R, de Groot JR, Coronel R, Tasseron S, Linnenbank AC, Hauer RN, Kirkels H, Janse MJ, de Bakker JM. Activation delay after premature stimulation in chronically diseased human myocardium relates to the architecture of interstitial fibrosis. Circulation. 2001; 104: 3069–3075.
Moreno J, Zaitsev AV, Warren M, Berenfeld O, Kalifa J, Lucca E, Mironov S, Guha P, Jalife J. Effect of remodelling, stretch and ischaemia on ventricular fibrillation frequency and dynamics in a heart failure model. Cardiovasc Res. 2005; 65: 158–166.
Kalifa J, Tanaka K, Zaitsev AV, Warren M, Vaidyanathan R, Auerbach D, Pandit S, Vikstrom KL, Ploutz-Snyder R, Talkachou A, Atienza F, Guiraudon G, Jalife J, Berenfeld O. Mechanisms of wave fractionation at boundaries of high-frequency excitation in the posterior left atrium of the isolated sheep heart during atrial fibrillation. Circulation. 2006; 113: 626–633.
Skanes AC, Mandapati R, Berenfeld O, Davidenko JM, Jalife J. Spatiotemporal periodicity during atrial fibrillation in the isolated sheep heart. Circulation. 1998; 98: 1236–1248.
Okuyama Y, Miyauchi Y, Park AM, Hamabe A, Zhou S, Hayashi H, Miyauchi M, Omichi C, Pak HN, Brodsky LA, Mandel WJ, Fishbein MC, Karagueuzian HS, Chen PS. High resolution mapping of the pulmonary vein and the vein of Marshall during induced atrial fibrillation and atrial tachycardia in a canine model of pacing-induced congestive heart failure. J Am Coll Cardiol. 2003; 42: 348–360.
Courtemanche M, Ramirez RJ, Nattel S. Ionic targets for drug therapy and atrial fibrillation-induced electrical remodeling: insights from a mathematical model. Cardiovasc Res. 1999; 42: 477–489.
Verheule S, Sato T, Everett T 4th, Engle SK, Otten D, Rubart-von der Lohe M, Nakajima HO, Nakajima H, Field LJ, Olgin JE. Increased vulnerability to atrial fibrillation in transgenic mice with selective atrial fibrosis caused by overexpression of TGF-beta1. Circ Res. 2004; 94: 1458–1465.
Spach MS, Miller WT 3rd, Dolber PC, Kootsey JM, Sommer JR, Mosher CE Jr. The functional role of structural complexities in the propagation of depolarization in the atrium of the dog. Cardiac conduction disturbances due to discontinuities of effective axial resistivity. Circ Res. 1982; 50: 175–191.
Sanders P, Berenfeld O, Hocini M, Jais P, Vaidyanathan R, Hsu LF, Garrigue S, Takahashi Y, Rotter M, Sacher F, Scavee C, Ploutz-Snyder R, Jalife J, Haissaguerre M. Spectral analysis identifies sites of high-frequency activity maintaining atrial fibrillation in humans. Circulation. 2005; 112: 789–797.
Ryu K, Shroff SC, Sahadevan J, Martovitz NL, Khrestian CM, Stambler BS. Mapping of atrial activation during sustained atrial fibrillation in dogs with rapid ventricular pacing induced heart failure: evidence for a role of driver regions. J Cardiovasc Electrophysiol. 2005; 16: 1348–1358.
Derakhchan K, Li D, Courtemanche M, Smith B, Brouillette J, Page PL, Nattel S. Method for simultaneous epicardial and endocardial mapping of in vivo canine heart: application to atrial conduction properties and arrhythmia mechanisms. J Cardiovasc Electrophysiol. 2001; 12: 548–555.
Sarmast F, Kolli A, Zaitsev A, Parisian K, Dhamoon AS, Guha PK, Warren M, Anumonwo JM, Taffet SM, Berenfeld O, Jalife J. Cholinergic atrial fibrillation: I(K, ACh) gradients determine unequal left/right atrial frequencies and rotor dynamics. Cardiovasc Res. 2003; 59: 863–873.
Burashnikov A, Antzelevitch C. Reinduction of atrial fibrillation immediately after termination of the arrhythmia is mediated by late phase 3 early afterdepolarization-induced triggered activity. Circulation. 2003; 107: 2355–2360.
Qu Z, Karagueuzian HS, Garfinkel A, Weiss JN. Effects of Na(+) channel and cell coupling abnormalities on vulnerability to reentry: a simulation study. Am J Physiol Heart Circ Physiol. 2004; 286: H1310–H1321.
Li D, Melnyk P, Feng J, Wang Z, Petrecca K, Shrier A, Nattel S. Effects of experimental heart failure on atrial cellular and ionic electrophysiology. Circulation. 2000; 101: 2631–2638.
Koumi S, Arentzen CE, Backer CL, Wasserstrom JA. Alterations in muscarinic K+ channel response to acetylcholine and to G protein-mediated activation in atrial myocytes isolated from failing human hearts. Circulation. 1994; 90: 2213–2224.