Simultaneous Multisite Mapping Studies During Induced Atrial Fibrillation in the Sterile Pericarditis Model
Insights Into the Mechanism of its Maintenance
Background Chronic atrial fibrillation (AF) is thought to be due to multiple, simultaneously circulating wavelets. In the canine sterile pericarditis model, the mechanisms of maintenance of AF are not yet understood.
Methods and Results During six induced AF episodes in six dogs with sterile pericarditis, 372 unipolar electrograms were recorded simultaneously from an electrode array placed around both atrial free walls, along with 10 to 24 electrodes from the atrial septum, by use of a multiplexing system. Activation maps during 12 consecutive 100-ms windows were analyzed from an episode of sustained AF in each dog (mean duration, 32±24 minutes). In two dogs, two such activation maps during the same episode of AF were analyzed. During AF, multiple unstable reentrant circuits (mean number, 1.4±0.1 per 100-ms analysis window) with very short cycle lengths (mean, 111±8 ms) present primarily in the atrial septum and right atrium were responsible for maintenance of AF. The unstable reentrant circuits frequently disappeared and re-formed. Wave fronts traveling from one atrium to the other and/or from the atrial septum play an important role in re-formation of unstable reentrant wave fronts.
Conclusions In this model of paroxysmal AF, unstable reentrant circuits of very short cycle length principally involving the atrial septum appear to be critical for maintenance of AF. Some reentrant circuits disappear as others re-form, so that at least one reentrant circuit is always present. Because the atria cannot follow their very short cycle lengths in a 1:1 manner, AF is maintained.
The electrophysiological mechanism underlying the maintenance of atrial fibrillation (AF) is not well understood. As recently summarized,1 controversy has existed for a long time regarding whether this arrhythmia was due to a single focus firing rapidly or to some sort of reentrant excitation. It was also suggested that this arrhythmia might result from some combination of the two. It is now generally accepted that AF results from a critical number of simultaneously circulating wavelets, the so-called multiple-wavelet hypothesis proposed by Moe et al.2 3 Multisite mapping studies by Allessie et al4 5 with a Langendorff-perfused canine atrial preparation of AF induced by rapid pacing and sustained by acetylcholine infusion have provided clear evidence to support the multiple-wavelet hypothesis as the basis of AF. Allessie et al4 5 emphasized the importance of a critical number of simultaneously circulating wavelets of the random reentry type and the importance of activation of each atrium by wavelets from the other atrium. Subsequent simultaneous multisite mapping studies by Cox et al6 in a canine mitral regurgitation model of AF and during open-heart surgery in patients with AF also are consistent with this explanation. Wang et al7 performed the activation mapping of vagally induced AF in dogs. However, because of important limitations, particularly in terms of the resolution of recording sites, they did not specify in detail the physiological mechanism of AF. Also, mapping of induced AF in humans was recently performed from a portion of each atrium.8 This latter study characterized the activation patterns seen, but an understanding of the mechanism of AF remains incomplete.
A previous right atrial free wall mapping study of spontaneous and ATP-induced conversion of atrial flutter to transient AF in the canine sterile pericarditis model showed that AF was associated with one or occasionally two unstable reentrant circuits of very short cycle length that frequently disappeared and re-formed.9 These unstable reentrant circuits seemed to be possibly responsible for driving the atria and maintaining the AF.9 In the present study, simultaneous multisite mapping of both atria, including the atrial septum, during induced sustained AF was performed to characterize this arrhythmia in this model and to test the hypotheses that (1) unstable reentrant circuits with very short cycle lengths are responsible for maintenance of AF, (2) the unstable reentrant circuits frequently disappear and re-form, and (3) wave fronts from the atrial septum and/or traveling between the atria play an important role in re-formation of unstable reentrant circuits.
We studied activation patterns of both atria, including the atrial septum, simultaneously during six episodes of induced sustained AF (defined as lasting >5 minutes) 2 days after creation of sterile pericarditis10 in six adult mongrel dogs weighing 20 to 30 kg. All studies were performed in accordance with guidelines specified by our Institutional Animal Care and Use Committee, the American Heart Association Policy on Research Animal Use, and the Public Health Service Policy on Use of Laboratory Animals.
Creation of the Sterile Pericarditis Model
The canine sterile pericarditis model was created as previously described.10 At the time of surgery, pairs of stainless steel wire electrodes coated with polytetrafluoroethylene (Teflon) except at the tip were sutured on the right atrial appendage in three dogs, the interatrial band known as Bachmann's bundle in all dogs, and the posteroinferior left atrium close to the proximal portion of the coronary sinus in all dogs. Another pair was sutured onto the right ventricular apex to be used for pacing, principally after His bundle ablation to create complete AV block as part of the studies in the open-chest state. At the completion of surgery, the dogs were given antibiotics and analgesics and then were allowed to recover. Postoperative care included administration of antibiotics and analgesics.
Studies in the Open-Chest State
For all dogs, on the second postoperative day, each animal was anesthetized, and an open-chest study was performed with standard surgical techniques. The dogs were anesthetized with pentobarbital (30 mg/kg IV) and mechanically ventilated with a Boyle anesthesia machine to deliver 100% oxygen during the experiment. The body temperature of the dogs was kept within the normal physiological range throughout the study with a heating pad and warm drip infusion.
Creation of Complete AV Block
During induced AF in this model, a rapid ventricular response rate usually occurs. Therefore, temporal superimposition of ventricular activation on atrial activation can interfere with interpretation of the recorded unipolar atrial electrograms. To avoid this, as previously described,9 we produced complete AV block by standard radiofrequency ablation techniques before the chest was opened. Then, by use of the previously placed ventricular electrodes, ventricular pacing was initiated at a rate of 80 to 100 bpm with a Medtronic 5375 external-demand pulse generator (Medtronic, Inc). During the periods of simultaneous multisite mapping of AF, ventricular pacing was performed at 40 to 60 bpm to decrease still further the temporal superimposition of atrial and ventricular events.
Induction of AF
AF was induced with a modified Medtronic 5325 programmable, battery-powered stimulator. During these studies, surface ECG lead II and bipolar electrograms obtained from the previously placed atrial epicardial electrodes were monitored on a VR-16 Electronics-for-Medicine oscilloscope and also recorded simultaneously on a Honeywell 101 FM tape recorder for later playback and analysis. The ECG was recorded between a band pass of 0.1 and 500 Hz and the bipolar electrograms between a band pass of 30 to 500 Hz.
AF induction was attempted by use of rapid atrial pacing for ≥20 beats performed from each of the atrial electrode sites beginning at a rate of 500 bpm with the rate incremented by 20 bpm until a rate of 800 bpm was reached or AF was achieved. All pacing was performed with stimuli of at least twice diastolic threshold and up to 20 mA, with a pulse width of 1.8 ms. After AF was induced, its duration was characterized. Only episodes of AF lasting > 5 minutes were used for analysis.
Simultaneous Multisite Mapping
For studies of the sequence of right and left atrial activation during AF, a new electrode array (Fig 1A and 1B⇓⇓) was built to record simultaneously from both atria. This new electrode array contained 372 unipolar electrodes arranged in 186 bipolar pairs (95 pairs for the right atrium and 91 pairs for the left atrium, which includes 14 pairs placed separately on Bachmann's bundle) (Fig 1C⇓). The interelectrode distance of each bipolar electrode in the array was 1.2 mm, and the distance between the center of each bipolar electrode pair and its neighbor was 4.2 mm diagonally and 6 mm perpendicularly. The electrode array for Bachmann's bundle (Fig 1A⇓) was positioned first. It was inserted behind the right atrial appendage, between the superior vena cava and the aortic root, and advanced to the left atrial appendage. Then the electrode array (Fig 1B⇓) was placed around the atria and secured with a hook-and-loop (Velcro) belt. The interatrial septum was also mapped simultaneously with a single 12-polar-electrode catheter with an interelectrode distance of 2 mm in three dogs, two electrode catheters (the previously described 12-polar-electrode catheter and a 12-polar-electrode catheter with an interelectrode distance of 3 mm) in two dogs, and two electrode catheters (20- and 24-polar-electrode catheters with an interelectrode distance of 1 mm) in one dog; all were inserted through the superior vena cava (Fig 2A⇓). The distal tip of one electrode catheter was placed into the orifice of the coronary sinus; it was also used as a marker channel. When two catheters were used, the distal tip of the other electrode catheter was placed into the inferior vena cava so that it lay along the septum. The catheters were sutured in place at the site of insertion in the external jugular vein. The location of the distal pair of electrodes was confirmed by manual palpation. The distance between the two catheters was ≈2 to 3 cm.
An example shown in Fig 2⇑ illustrates the analysis of a representative activation map of the atrial septum. The activation patterns of electrograms recorded from the two catheters were basically similar (Fig 2C and 2D⇑⇑). Atrial electrograms from both atria and the interatrial septum along with ECG lead II were recorded during the period of AF. Data were simultaneously recorded from the right atrial free wall (190 electrodes) and the left atrial free wall (154 electrodes), Bachmann's bundle (28 electrodes), and the atrial septum (10 or 24 electrodes; in the latter instance, electrograms were recorded from 12 electrodes per catheter). Data were recorded and processed with two cardiac mapping systems—that is, one for the right atrium and another for the left atrium, the Bachmann's bundle, and the atrial septum—designed at Case Western Reserve University and described previously.9 During periods of mapping, the start key of the computer keyboard of each system was simultaneously enabled to start recording. For time alignment of the two mapping systems, a common marker channel was used through which a marker was introduced manually at deliberately irregular intervals between 1 and 10 seconds continuously throughout the study. Markers were then numbered consecutively to permit temporal lining up of the data for analysis. Data were archived on either a floppy disk or a hard disk and backed up on tape in their raw format. Because of the differences in size of the right atrium and left atrium from dog to dog, anatomic landmarks (the vena cava, the right atrial appendage, and the AV groove for the right atrium and the pulmonary veins, the left atrial appendage, and the AV groove for the left atrium) were identified and positioned on the grid (electrode array) by visual inspection. For each atrial beat, activation time at each site was placed on an anatomic grid representing activation at each bipolar recording site, and isochronous lines at 10-ms intervals were drawn manually.
Data Acquisition and Analysis
Data were analyzed as previously described.9 11 This permitted selection of activation times and computation of an isochrone map with a maximum resolution of 1 ms. A common marker was recorded on one of the electrode sites of each mapping system and the septal recording and was used to choose an activation reference point. Analysis was based on sequential 100-ms time windows. For each episode, 1.2 seconds of data (12 consecutive time windows) from a midportion of the episode was analyzed, and the activation sequences were depicted by activation maps. In two episodes (episodes 2 and 5), two separate 1.2-second sections of data (eg, at 10 and 20 minutes during a 30-minute episode) during each AF episode were analyzed and compared to validate the assumption that a 1.2-second section of electrogram recordings selected from the middle of a sustained AF episode is representative of the activation sequences throughout the episode (not including the induced onset or spontaneous termination).
The representative example shown in Fig 3⇓ illustrates the analysis of the location and length of the line of functional block and areas of slow conduction. The top panel in Fig 3⇓ shows electrograms recorded from six adjacent sites (a through f) on either side of a line of functional block, depicted by thick dashed lines in the bottom two panels, during one episode of sustained AF. The impulse circulates in a clockwise direction from site a to site f around a central line of block. The bottom panel displays two consecutive activation maps relative to the line of functional block that correspond to the beats in the top panel. In the electrograms recorded from site a, a double potential is present during beat 1, with local activation represented by the first deflection and activation on the other side of the block represented by the second deflection (*).8 11 12 13 14 In the first beat (beat 1), sites a and f are located opposite each other relative to the line of block. When double potentials are recorded at these sites, local activation at each site is reflected by the large electrogram, and activation on the other side of the block by the low-amplitude potential (*). During beat 2, sites a and f and b, d, and e are located opposite each other relative to the line of block. Thus, during beat 2, the line of functional block shifts and lengthens slightly compared with beat 1.
AF was defined as a rapid atrial rhythm characterized by variability of the beat-to-beat cycle length, polarity, morphology, and/or amplitude of the recorded bipolar atrial electrograms.15 Sustained AF was defined as lasting >5 minutes. The mean cycle length of AF was defined as the mean value of all atrial electrogram intervals sampled from a single site in the Bachmann's bundle area during the 1.2 seconds used for analysis. An unstable reentrant circuit was defined as a reentrant circuit in which, from analysis window to analysis window, there were changes in the length and location of the central area of block, the path of the reentrant circuit, and the cycle length of the circulating wave front. For purposes of counting individual wave fronts, the following system was used: when a wave front bifurcated, it was considered that only one additional wave front was present; when two wave fronts fused, it was considered that only one “old” wave front was present, not that a new wave front had formed. A nonactivation zone was defined as the time during a 1.2-second period of analysis in which no depolarization waves were present in either atrial free wall.
Data are expressed as mean±SD. Student's t test was used to compare differences in the number of wave fronts and lines of functional block between atria. A value of P<.05 was considered statistically significant.
Characteristics of Induced AF
Six episodes of sustained AF lasting a mean of 32±24 minutes (range, 7 to 60 minutes) were induced (Table 1⇓). A representative example of induced, sustained AF is shown in Fig 4⇓. Electrograms from the right atrial appendage, Bachmann's bundle, and the posteroinferior left atrium showed a rapid, irregular rhythm with a continuous beat-to-beat variation in electrogram morphology and cycle length. In this case, stable, sustained AF lasting 30 minutes was induced.
Multisite Mapping Studies of Induced AF
Activation Patterns of Both Atria During AF
During AF, the free walls of both atria were activated nonuniformly by a variable number of wave fronts that included various degrees of slow conduction and lines of functional conduction block. During AF, two to five activation wave fronts were observed at any given window of analysis in each atrium. Seven types of activation patterns were seen in the maps: (1) single, broad wave fronts; (2) fusion of wave fronts; (3) bifurcation of wave fronts; (4) collision of wave fronts; (5) block of wave fronts; (6) wave fronts traveling from one atrium to the other and/or from the atrial septum to epicardium via the Bachmann's bundle area, the intercaval region, or beneath the inferior vena cava; and (7) unstable reentrant wave fronts (reentrant circuits). During AF, no mapping data consistent with an ectopic focus or ectopic foci were seen, ie, no activation sequence map suggested an ectopic focus, because no concentric isochrones emanating from a single site were observed. In general, the left atrium was passively activated by wave fronts emerging from the right atrium or septum, although a relatively large reentrant circuit around the pulmonary veins was seen in several windows.
Analysis of Consecutive Activation Patterns During a Representative Episode of Sustained AF
Fig 5A⇓ shows consecutive activation maps from a representative episode (episode 2a) of induced sustained AF in which the mean beat-to-beat cycle length was 104 ms. Twelve separate 100-ms windows of activation recorded at 10 minutes after onset are demonstrated. For each 100-ms window during AF, the right atrium was activated by a variable number of wave fronts: three (window 2), four (windows 1, 3, 5, 6, 7, 8, 10, 11, and 12), or five (windows 4 and 9) wave fronts (mean, 4.1) traveling in different directions. Similarly, the left atrium was activated by a variable number of wave fronts: two (windows 2 and 10), three (windows 1, 6, 7, 8, and 9), or four (windows 3, 4, 5, 11, and 12) wave fronts (mean, 3.3) traveling in different directions. The left atrium was activated by wave fronts coming from the right atrium or septum, as well as by a relatively large reentrant circuit around the pulmonary veins that appeared twice (windows 1 and 2 and windows 9 and 10). As seen in each window, the right atrium was activated by one to three wave fronts (mean, 2.6) coming from the left atrium or septum via the Bachmann's bundle area, the intercaval region, and the region beneath the inferior vena cava, as well as by reentrant circuits on the right atrial free wall (windows 1 to 3 and windows 8 to 11). A reentrant circuit involving the atrial septum was present in each of the twelve 100-ms windows. Examples of fusion (windows 1, 3, 4, 5, 6, 7, 8, 9, 10, and 11) or collision (windows 4, 6, 7, and 11) of wave fronts and local areas of slow conduction (relative crowding of isochrones) (windows 1 to 4 and windows 6 to 12) were found in both atria. Also, bifurcation of a wave front into two separate wave fronts was seen (windows 1 to 12). At least one to three lines of functional block (represented by thick dashed lines) were present in the right atrial free wall in each of the 12 windows (mean, 2.3 per window), whereas in the left atrium, lines of functional block were not seen in every window, with only two being seen most often in one window (mean, 0.9/window). Moreover, the location and length of the lines of functional block changed from analysis window to analysis window.
Unstable reentrant circuits (denoted by colored lines with arrows in Fig 5A⇑)
In Fig 5A⇑, window 1 shows an activation wave front of the reentrant circuit at time zero. This area of zero activation time was reactivated at 100 ms. This reentrant circuit (orange) had one rotation around a central line of functional block at the right atrial appendage in a clockwise direction with a cycle length of 100 ms, and subsequently one rotation that included the superior vena cava as part of the central line of block with a cycle length of 130 ms (windows 2 and 3). Window 3 shows persistence of the reentrant circuit but with a change, ie, the line of functional block shortened. The shorter line of functional block allowed the free wall close to the right atrial appendage now to be activated from this reentrant circuit. Because of the short cycle length (80 ms) of the reentrant circuit during window 3, that area (at 170 ms) was probably still refractory, and block of the reentrant wave front at 250 ms (represented by T markers) occurred. The electrograms recorded from six sites (a through f) along this reentrant circuit (orange) are shown in Fig 5B⇑. The first beat corresponds to window 1 in Fig 5A⇑. Note that the impulse circulates from site a to site f. The second beat corresponds to window 2. Note that at site c, there is now a double potential, indicating that activation at this site now reflects a region of block, with each potential of the double potential reflecting activation on either side of the line of block. Thus, activation that previously proceeded directly from site b to site c is now blocked. Furthermore, site c is now activated by a wave front that travels around the superior vena cava (Fig 5A⇑, window 2; Fig 5B⇑). The impulse then continues around the area of functional block, pivoting again at site d (Fig 5A⇑, window 2; Fig 5B⇑). With the third beat in this sequence (Fig 5A⇑, window 3; Fig 5B⇑), the wave front pivots again on the right atrial side of the superior vena cava to activate sites a, b, and c, after which it blocks, presumably because of refractoriness (it occurs ≈70 ms after the previous activation wave front).
In the left atrium (Fig 5A⇑), an activation wave front came from the intercaval region at 20 ms (window 1). This wave front went toward the left atrial appendage, then entered the Bachmann's bundle area and reactivated a starting point of the reentrant circuit at 120 ms. This reentrant circuit (green) had 1.5 rotations around the pulmonary veins and line of functional block in a clockwise direction with a cycle length of 100 ms and then disappeared at 170 ms, probably because of refractoriness from the previous activation (Fig 5A⇑, windows 1 and 2). However, in window 2, a wave front coming from the left atrium around the inferior vena cava penetrated the septum from the intercaval region on the right side, forming a new reentrant circuit. This reentrant circuit (blue) involved the septum and the atrial epicardium and lasted for 2 rotations with a cycle length of 100 ms. In window 4, the wave front that broke through from the septum at Bachmann's bundle went down the intercaval region and formed another reentrant circuit. This reentrant circuit (blue) also involved the septum and the atrial epicardium. After 370 ms, no electrical activity was seen in the entire right atrial free wall until a new wave front from the septum entered Bachmann's bundle and activated the right atrial free wall at 430 ms (Fig 5A⇑, windows 4 and 5).
In the activation maps from 400 to 1200 ms, the reentrant circuit (blue) that involved the septum and the atrial epicardium lasted for 7.5 rotations, with a cycle length of 100 to 120 ms (Fig 5A⇑, windows 5 through 12). Electrograms recorded from six sites (g through l) along this septal reentrant circuit (blue) are shown in Fig 5C⇑. The four beats in Fig 5C⇑ correspond to windows 5 through 8 in Fig 5A⇑. The impulse from Bachmann's bundle (*) travels down along the intercaval region from site g to site i, goes up along the septum from site j to site l, and then breaks through at Bachmann's bundle. The electrograms from Bachmann's bundle in Fig 5C⇑ are, for purposes of this figure, unipolar potentials. Note the biphasic potential (+,−), indicating that the activation wave front first approached this site and then traveled away from it.
The activation wave front coming from the septum (Fig 5A⇑, window 8) entered the right atrium via Bachmann's bundle superior and inferior to the superior vena cava. The broad inferior wave front blocked unidirectionally in the mid right atrial free wall at 760 ms and bifurcated (Fig 5A⇑, window 8). The upper branch of the bifurcated inferior wave front joined the superior wave front at the right atrial appendage (Fig 5A⇑, window 8) and formed a reentrant circuit (orange) in the right atrial free wall (Fig 5A⇑, window 9). This latter reentrant circuit turned around a line of functional block (Fig 5A⇑, windows 9 through 12), shifting from the right atrial appendage to the center of the free wall in a clockwise direction for 3 rotations through to window 11 with a mean cycle length of 110 ms. Then, it blocked in tissue that was probably still refractory from the previous activation at 1110 ms. In window 10 in Fig 5A, 4⇑⇑ reentrant circuits were observed. One (orange) circulated around a line of functional block in the right atrial free wall, one (purple) around the inferior vena cava, one (green) around the pulmonary veins, and one (blue) involved the septum and the atrial epicardium.
In summary, in this 1.2 seconds of a representative episode of sustained AF, 18 reentrant circuits in total (mean, 1.5 reentrant circuits per 100-ms window) were observed (10 involving the septum and the atrial epicardium, 4 in the right atrial free wall, 2 around the pulmonary veins, 1 around the superior vena cava, and 1 around the inferior vena cava). Moreover, these reentrant circuits disappeared and subsequently re-formed, usually with variations in the course and location of the circuits. Furthermore, these reentrant circuits were characterized by very short cycle lengths (mean, 104 ms) and 1 to 7.5 (mean, 2.6) consecutive rotations of the reentrant circuits.
Nonactivation zones in the atrial free wall
During 1.2 seconds of this episode of AF, no activation was recorded in the right atrial free wall from 370 to 430 ms (zone=60 ms) and from 680 to 730 ms (zone=50 ms). After electrical activity disappeared from the right atrial free wall, reactivation occurred via a wave front coming from the septum via the Bachmann's bundle area at 430 and 730 ms, respectively. Similarly, in the left atrial free wall, no activation was observed from 180 to 210 ms, from 280 to 310 ms, from 380 to 410 ms, from 490 to 530 ms, from 600 to 620 ms, from 700 to 720 ms, from 800 to 840 ms, from 1010 to 1040 ms, and from 1110 to 1140 ms. Reactivation of the left atrial free wall always occurred via wave fronts coming from the septum via Bachmann's bundle. Moreover, no activation was seen from 380 to 410 ms and from 700 to 720 ms in both atria, and their reactivations occurred by the wave front from the septum via Bachmann's bundle.
In two episodes of AF, we analyzed the activation sequence during an additional 1.2-second period to compare with the activation sequence of the first 1.2-second period analyzed. As seen in the analysis of these two 1.2-second periods during the same episode of sustained AF and as tabulated in Table 2⇓ (dogs 2a, 2b and 5a, 5b), although the exact locations and activation patterns of the unstable reentrant circuits varied, their numbers and other quantitative characteristics (cycle length, number of rotations) remained quite constant, as did reactivation of the atrial free walls after a nonactivation period.
Analysis of Consecutive Activation Patterns During Another Representative Episode
In Fig 6⇓, consecutive activation maps are shown from another episode (episode 5a) of induced sustained AF in another dog. A tabulation of the characteristics of the activation sequence during the 1.2 seconds of analysis is shown in Tables 2 and 3⇑⇓. In window 1, a reentrant circuit was seen in the right atrial free wall (orange). It had one full rotation and part of another rotation and disappeared at 140 ms (window 2). A similar right atrial reentrant circuit re-formed in window 7. It lasted for about 2.5 rotations and disappeared at 1010 ms (window 11). Also, in the left atrium, an unstable reentrant circuit around the pulmonary veins (green) that disappeared and re-formed was observed in windows 1 through 4, 6 through 8, and 9 through 12. Likewise, a septal reentrant circuit (blue) was seen that lasted for several windows (windows 1 through 6), then disappeared and re-formed in window 8. It lasted through window 10. It then re-formed once again in window 11 and continued through window 12.
In this 1.2-second period of an episode of AF, 16 reentrant circuits in total (mean, 1.3 reentrant circuits per 100-ms window) were observed: 7 involving the septum and the atrial epicardium (blue), 6 around the pulmonary veins (green), and 3 in the right atrial free wall (orange). Moreover, they lasted for 1.5 to 4.5 (mean, 2.7) consecutive rotations, with a mean cycle length of 118 ms. Thus, as tabulated in Tables 2 and 3⇑⇑, although there were some differences between this representative example (episode 5a) and that shown in Fig 5A⇑ (episode 2a), the fundamental similarities are clear.
Characteristics and Location of Unstable Reentrant Circuits During Sustained AF
The relevant data are summarized in Table 2⇑. The mean cycle length of induced AF in all episodes was 110±7 ms (range, 101 to 119 ms). The mean number of wave fronts per window (3.8±0.6 in the right atrium versus 2.6±0.3 in the left atrium, P<.01) and lines of functional block per window (1.6±0.5 in the right atrium versus 0.9±0.5 in the left atrium, P<.02) were significantly greater in the right atrium than in the left atrium. The mean number of unstable reentrant circuits during the 1.2 seconds of AF analyzed was 17±2 (1.4±0.1 per window), with a range of 15 to 19. They had very short cycle lengths (mean, 111±8 ms; range, 101 to 119 ms) and 1 to 7.5 (mean, 2.4±0.5) consecutive rotations per reentrant circuit. These reentrant circuits disappeared and re-formed and, we suggest, drove the atria at cycle lengths that could not be followed 1:1, thus maintaining the AF. The re-formation of reentrant circuits seems to be caused largely by wave fronts coming from the atrial septum, the opposite atrium, or both.
The locations of the unstable reentrant circuits are summarized in Table 3⇑. The unstable reentrant circuits were observed primarily (1) involving the septum and the atrial epicardium (68/134; 51%); (2) in the right atrial free wall (31/134; 23%); and (3) sometimes around the pulmonary veins (23/134; 17%), the superior vena cava (8/134; 6%), and the inferior vena cava (4/134; 3%) (Fig 7⇓).
Nonactivation Zones in the Atria and their Reactivation via Sites From the Septum or the Opposite Atrium
During 1.2 seconds of analysis, in all episodes of sustained AF, the mean duration of nonactivation zones was 62±19 ms (range, 30 to 120 ms) in the right atrial free wall and 47±21 ms (range, 20 to 90 ms) in the left atrial free wall. In the right atrium, this zone was reactivated by wave fronts coming primarily from the septum (29/32; 91%) or the left atrial free wall (only 3/32; 9%). The reactivation from the septum occurred via four sites: (1) Bachmann's bundle (14/29; 48%), (2) the posteroinferior aspect of the superior vena caval region (6/29; 21%), (3) the intercaval region (6/29; 21%), and (4) the peri-inferior vena caval region (3/29; 10%). In the left atrium, a nonactivation zone was reactivated by wave fronts coming primarily from the septum (47/60; 78%) or the right atrial free wall (13/60; 22%). The reactivation from the septum occurred via two sites: (1) Bachmann's bundle (36/47; 77%) and (2) the peri-inferior vena caval region (11/47; 23%). Eight instances of nonactivation zone in both atria were observed in 3 dogs; their reactivation occurred from the septum via Bachmann's bundle in 6 (75%) and the peri-inferior vena caval region in 2 (25%).
Thus, wave fronts coming from the septum via the Bachmann's bundle area make the greatest contribution to the reactivation of the right atrial free wall during AF. The presence of these nonactivation zones demonstrates that neither the right atrium alone nor the left atrium alone is responsible for maintenance of AF. In addition, the fact that the reactivation almost always results from an exiting wave front from a septal reentrant circuit demonstrated that a reentrant circuit in the septum is a major factor in the maintenance of AF.
The data presented in this study show that during sustained AF in the canine sterile pericarditis model, multiple (up to 4 per 100-ms window) unstable reentrant circuits of relatively short cycle length are present. These unstable reentrant circuits involve the septum and atrial epicardium and are short-lived, so that they disappear and re-form. Also, nonactivation zones are periodically observed in each atrium. Wave fronts primarily from the septum play an important role in reactivation of the nonactivation zones and re-formation of the unstable reentrant circuits.
Role of Unstable Reentrant Circuits in Maintenance of AF in This Model
On the basis of a right atrial mapping study from our laboratory9 of the brief period of AF during the transition from atrial flutter to AF and back again to atrial flutter in the canine sterile pericarditis model, we had suggested that AF might be generated by one or possibly two unstable reentrant circuits of short cycle length. However, this previous study was limited to mapping from the right atrial free wall only. On the basis of the current, much more extensive mapping studies of sustained AF, we suggest that the multiple, unstable reentrant circuits present in our model may indeed be responsible for the maintenance of sustained AF in the following way. First, unstable reentrant circuits (fixed reentrant circuits were never seen) that disappeared while others were re-formed are always present. Most important, at least one unstable reentrant circuit is always present and circulates for several rotations (1 to 7.5). These unstable reentrant circuits have variable but rather short cycle lengths—mean, 111±8 ms. These short cycle lengths serve to drive the atria. Because the cycle lengths are so short and there is usually more than one circulating reentrant circuit (up to 4 per 100-ms window), the atrial tissue cannot follow 1:1. The result is that areas of functional block, slow conduction, and multiple wave fronts are generated, which in turn serve to re-form unstable reentrant circuits. In sum, there is an interaction in which unstable and short-lived reentrant circuits (ie, mother waves) of short cycle length generate wave fronts (ie, daughter waves), which serve to re-form the unstable reentrant circuits. It is because the unstable reentrant circuits are always generating multiple wave fronts and the wave fronts are continuously re-forming reentrant circuits of short cycle length that AF is maintained in this model. This implies that if the reentrant circuits do not re-form, for instance, after a nonactivation period, the AF would spontaneously terminate.
The following illustrates this explanation. As shown in the maps during sustained AF (windows 4 and 5 in Fig 5A⇑), a wave front in the left atrium disappeared at 380 ms, and a nonactivation zone was seen in both atrial free walls until another wave front coming from the septum entered the atrial free wall via the Bachmann's bundle area at 420 ms. This wave front not only reactivated the atrial free wall but also, by reentering the atrial septum at 460 ms, continued the septal reentrant circuit. If a reentrant circuit had not been present in the atrial septum and a new wave front had not entered the atrial free wall from the septal reentrant circuit, we suggest that the AF would have terminated with the block of the wave front at 380 ms. This indicates that AF could not sustain itself unless, in this case, a reentrant circuit involving the atrial septum were present. In addition, the presence of nonactivation zones demonstrates that neither the right atrium alone nor the left atrium alone is responsible for maintenance of AF, and the presence of a wave front propagating from the atrial septum is an important factor in perpetuating activation in each atrium and therefore in maintaining AF. Thus, continuous atrial activation during AF is caused by the interdependence of multiple unstable (short-lived) reentrant circuits of short cycle length and multiple wave fronts in both atria and the atrial septum.
Relation to Previous Studies
Mapping of AF has been performed in a number of studies.4 5 6 7 8 16 17 In these studies, AF was the result of cholinergic stimulation4 5 7 or atrial enlargement resulting from mitral valve incompetence.6 These studies indicate that AF is based on multiple simultaneously circulating wave fronts. The several studies showed a wide spectrum of activation patterns, principally showing multiple reentrant activation wavelets but also showing some short-lived reentrant circuits.5 6 7 8 In the model of cholinergic AF in Langendorff-perfused canine atria by Allessie et al,4 5 total mapping of both atria exhibited complex activation patterns resulting from the interaction of an average of about six randomly reentrant wavelets. This number of reentrant wavelets coincides with our observations, in which the mean number of wave fronts was 3.8 in the right atrium and 2.6 in the left atrium (6.4 in both atria) during the period of sustained AF. Although Allessie et al5 principally emphasized that “…intra-atrial reentry as a basis for fibrillation means reexcitation of a given area that has already been activated before by another propagating wavelet,” they did describe an occasional unstable reentrant circuit (up to two revolutions) around the right or left atrial appendage. They explained the perpetuation of AF on the basis of a critical number of circulating random reentrant wavelets, with the number of wavelets determined by a “…balance between disappearance of ‘old’ wavelets and genesis of ‘new’ ones….” The latter were thought to result from division (bifurcation) of an existing wavelet or from activation coming from the other atrium. They also speculated on the possible contribution of new impulse formation from an automatic (pacemaker) site.
The simultaneous multisite mapping study of both atria during AF reported by Cox et al6 described findings in 25 dogs with mitral regurgitation as well as in 13 patients with Wolff-Parkinson-White syndrome who were operated on for division of an accessory AV connection. Six patients had a single reentrant circuit in the right atrium in which conduction block occurred along the sulcus terminalis. The left atrium was activated by wave fronts emerging mostly from the right atrial reentrant circuit, and reentrant circuits in the left atrium could not be detected. Their study demonstrated the presence of a single reentrant circuit.6 They did not record from the septum, but on the basis of the demonstration and location of incomplete reentrant circuits on the epicardium, they speculated that the septum might be involved. Furthermore, they suggested that AF could “…be caused by a single reentrant circuit.”
Another simultaneous multisite mapping study of both atria (but not including the septum) was performed during vagally induced AF in dogs by Wang et al.7 They indicated that their study “…suggests several coexistent reentry circuits of relatively small diameter.” However, the authors were careful to refer to them as “apparent reentry circuits,” because they never demonstrated complete reentrant circuits, and in consecutive beats, it was not clear that one could follow a reentrant circuit. That interpretation is consistent with the data from Allessie et al5 and was suggested by Wang et al as well.7
Wells et al15 classified AF into four types based on the morphology of bipolar atrial electrogram recorded from one to three sites in patients after open-heart surgery. Recently, on the basis of the pattern of activation obtained by simultaneous multisite mapping of only a part of the right and left atrium during induced AF in humans, Konings et al8 distinguished three types of AF based on activation patterns. In type I, the right atrial free wall was activated by single broad wave fronts propagating without conduction delay. Type II was more complex, showing a higher degree of delayed conduction and intra-atrial conduction block. Type III activation was highly complex, with three or more wavelets associated with areas of slow conduction and conduction block. However, neither the Wells nor the Konings study addressed mechanism, nor was the type of AF correlated with the duration of induced AF. However, Konings et al did demonstrate leading circle–type reentry during their type III AF. In the present study, during sustained AF, unstable reentrant circuits were observed, principally involving the atrial septum and the right atrial free wall. More recently, Wijffels et al18 described a study of chronically instrumented goats in which 7.1±4.8 days of artificial maintenance of AF by a fibrillation pacemaker produced sustained AF (>24 hours) in 10 of 11 goats. However, no mapping studies were reported.
Finally, previous sequential site mapping studies after placement of aconitine on the canine atria suggested that a single site firing rapidly could be responsible for the maintenance of AF.19 Subsequent studies in which aconitine was placed on rabbit atria produced a rapid atrial rate most likely due to abnormal automaticity.20 21 Thus, it appears that a single focus firing rapidly can produce AF. In the case of aconitine, the focus presumably was automatic, but in theory, a focus firing rapidly could even be due to a triggered or reentrant mechanism. Schuessler et al22 recently demonstrated a stable figure-eight reentrant circuit of very short cycle length in the inferior wall of an in vitro canine atrial preparation superfused with acetylcholine. In this example, the cycle length of the reentrant circuit was so short that the rest of the atrial tissue could not follow with 1:1 activation, resulting in what could reasonably be called AF. Furthermore, continuous rapid atrial pacing from a single site has long been known to precipitate and sustain AF in patients.23 24 And most recently, we have obtained evidence in patients after open-heart surgery that suggests that a focus of type II atrial flutter with a very short cycle length can drive the atria at a rate so rapid that the remainder of the atria cannot follow 1:1 activation, thereby producing AF.25 The common thread of these models of AF is the necessity of an arrhythmia mechanism to produce a very rapid rate at which the atria cannot respond with 1:1 activation. This principle is also part of the mechanism of AF in our model.
An understanding of the mechanism of maintenance of AF may provide insights into devising therapeutic strategies for the treatment of AF. For example, for paroxysmal AF, medications that prolong the refractory period such that the unstable reentrant circuits either could not form or would be of such long cycle length that they could no longer drive the atria and sustain AF might be predicted to be effective. Recently, Cox et al26 developed the maze operation, a surgical approach to treat AF by producing numerous selected through-and-through incisions in both atria. This surgical therapy is based on preventing continued random reentry by presumed multiple wavelets as a result of the incisions (a maze) and thereby preventing AF. This therapeutic approach has also been mimicked by radiofrequency catheter ablation techniques.27 28 29 30 However, our mapping data may provide new insights into the decision of where anatomically to place the critical incisions or even ablative lesions in patients with paroxysmal AF. As seen in the present study, the reentrant circuit in the atrial septum is particularly important in generating activation wave fronts that re-form the unstable reentrant circuits in both atrial free walls. This activation occurs primarily via Bachmann's bundle. Because of this, the region of Bachmann's bundle seems a critical crossroad and thus a potential vulnerable site (Achilles' heel) for surgical or catheter ablation. Thus, block of wave fronts at this area might prevent re-formation of unstable reentrant circuits in the atria, thereby preventing AF from becoming sustained. Preliminary studies in the canine sterile pericarditis have supported this hypothesis.31 32
A limitation of our study is the fact that we had relatively poor recording resolution from the atrial septum. We used only 10 or 24 electrodes to record from the septum. Clearly, this is a less-than-optimal recording resolution. Nevertheless, analysis of the data from the septum was most helpful and was consistent with the data obtained from the atrial epicardial activation maps.
In addition, because our model is not a model of chronic AF, we cannot extrapolate from our data to the mechanism of chronic AF. However, it is possible that AF starts with a mechanism much as we have demonstrated in our model, but when the AF becomes persistent, pathophysiological adaptations resulting from the very rapid atrial rate (remodeling), as described by Wijffels et al,18 may result in an evolution of the mechanism of AF to that described by Moe et al2 3 and Allessie et al.5
This study was supported in part by grant RO1-HL-38408 from the US Public Health Service, National Institutes of Health, National Heart, Lung, and Blood Institute, Bethesda, Md.
- Received April 18, 1996.
- Revision received August 5, 1996.
- Accepted August 28, 1996.
- Copyright © 1997 by American Heart Association
Waldo AL. Mechanisms of atrial fibrillation, atrial flutter, and ectopic atrial tachycardia: a brief review. Circulation. 1987;75(suppl III):III-23-III-40.
Moe GK. On the multiple wavelet hypothesis of atrial fibrillation. Arch Int Pharmacodyn Ther. 1962;140:183-188.
Allessie MA, Lammers WJEP, Smeets JRLM, Bonke FIM, Hollen J. Total mapping of atrial excitation during acetylcholine-induced atrial flutter and fibrillation in the isolated canine heart. In: Kulbertus HE, Olsson SB, Schlepper M, eds. Atrial Fibrillation. Molndal, Sweden: Lindgren and Soner; 1982:44-62.
Allessie MA, Lammers WJEP, Bonke FIM, Hollen J. Experimental evaluation of Moe's multiple wavelet hypothesis of atrial fibrillation. In: Zipes DP, Jalife J, eds. Cardiac Arrhythmias. New York, NY: Grune & Stratton; 1985:265-276.
Cox JL, Canavan TE, Schuessler RB, Cain ME, Lindsay BD, Stone C, Smith PK, Boineau JB. The surgical treatment of atrial fibrillation, II: intraoperative electrophysiologic mapping and description of the electrophysiologic basis of atrial flutter and atrial fibrillation. J Thorac Cardiovasc Surg. 1991;101:406-426.
Wang Z, Page´ P, Nattel S. Mechanism of flecainide's antiarrhythmic action in experimental atrial fibrillation. Circ Res. 1992;71:271-287.
Konings KTS, Kirchhof CJHJ, Smeets JRLM, Wellens HJJ, Penn OC, Allessie MA. High-density mapping of electrically induced atrial fibrillation in humans. Circulation. 1994;89:1665-1680.
Ortiz J, Niwano S, Abe H, Gonzalez HX, Rudy Y, Johnson NJ, Waldo AL. Mapping the conversion of atrial flutter to atrial fibrillation and atrial fibrillation to atrial flutter: insights into mechanism. Circ Res. 1994;74:882-894.
Feld GK, Shahandeh-Rad F. Mechanism of double potentials recorded during sustained atrial flutter in the canine right atrial crush-injury model. Circulation. 1992;86:628-641.
Olshansky B, Moreira D, Waldo AL. Characterization of double potentials during ventricular tachycardia: studies during transient entrainment. Circulation. 1993;87:373-381.
Schuessler RB, Kawamoto T, Hand DE, Mitsuno M, Bromberg BI, Cox JL, Boineau JP. Simultaneous epicardial and endocardial activation sequence mapping in the isolated canine atrium. Circulation. 1993;88:250-263.
Kirchhof CJHJ, Chorro F, Scheffer GJ, Brugada J, Konings KTS, Zetelaki Z, Allessie MA. Regional entrainment of atrial fibrillation studied by high-resolution mapping in open-chest dogs. Circulation. 1993;88:736-749.
Wijffels MCEF, Kirchhof CJHJ, Dorland R, Allessie MA. Atrial fibrillation begets atrial fibrillation: a study in awake chronically instrumented goats. Circulation. 1995;92:1954-1968.
Scherf D. Studies on auricular tachycardia caused by aconitine administration. Proc Exp Biol Med. 1947;64:233-239.
Goto M, Sakamoto Y, Imanaga I. Aconitine-induced fibrillation of the different muscle tissues of the heart and the action of acetylcholine. In: Sano T, Matsuda K, Mizuhira B, eds. Electrophysiology and Ultrastructure of the Heart. New York, NY: Grune & Stratton; 1967:199-209.
Schuessler RB, Grayson TM, Bromberg BI, Cox JL, Boineau JP. Cholinergically mediated tachyarrhythmias induced by a single extrastimulus in the isolated canine right atrium. Circ Res. 1992;71:1254-1276.
Waldo AL, MacLean WAH, Karp RB, Kouchoukos NT, James TN. Continuous rapid atrial pacing to control recurrent or sustained supraventricular tachycardia following open heart surgery. Circulation. 1976;54:245-250.
Waldo AL, Cooper TB. Spontaneous onset of type I atrial flutter in patients. J Am Coll Cardiol. In press.
Avitall B, Hare J, Mughal K, Silverstein E, Krum D, Natale A, Deshpande S, Dhala A, Akhtar M. Ablation of atrial fibrillation in a dog model. J Am Coll Cardiol. 1994;23:276A. Abstract.
Kempler P, Littmann L, Chuang CH, Dezern KR, Tuntelder JR, Wu G, Tatsis GP, Svenson RH. Radiofrequency ablation of the right atrium: acute and chronic effects. Pacing Clin Electrophysiol. 1994;17(pt 2):797. Abstract.
Elvan A, Pride HP, Zipes DP. Replication of the ‘maze’ procedure by radiofrequency catheter ablation reduces the ability to induce atrial fibrillation. Pacing Clin Electrophysiol. 1994;17(pt 2):774. Abstract.
Swartz JF, Pellersels G, Silvers J, Patten L, Cervantez D. A catheter-based curative approach to atrial fibrillation in humans. Circulation. 1994;90(pt 2):I-335. Abstract.
Kumagai K, Uno K, Khrestian C, Waldo AL. Single site radiofrequency catheter ablation of atrial fibrillation. J Am Coll Cardiol. 1996;27(suppl A):4A. Abstract.
Nakagawa H, Kumagai K, Imai S, Yamanashi W, Pitha J, Uno K, McClelland J, Beckman K, Arruda M, Lazzara R, Waldo AL, Jackman W. Catheter ablation of the Bachmann's bundle from the right atrium eliminates atrial fibrillation in a canine sterile pericarditis model. Pacing Clin Electrophysiol. 1996;19(pt 2):581. Abstract.