Atrial fibrillation is a common arrhythmia that afflicts humans and other animals.1,2 Understanding the potentially varied mechanisms and circumstances of atrial fibrillation is critical for prevention and treatment in individual patients. For example, atrial fibrillation can develop in dogs with and without obvious structural or functional cardiac disease.1 Also, just as in some humans and horses, the contribution of high vagal tone has been proposed as pivotal for the induction of atrial fibrillation in large- and giant-breed dogs.3–5 Experimentally in dogs and other animals, direct vagal nerve stimulation or acetylcholine infusion alters electrophysiologic properties of the atrial tissue (eg, shortening of the action potential duration through activation of potassium channel IKACh and increase in the heterogeneity of refractoriness) such that, when combined with pacing, atrial fibrillation is induced.5,6 Other means of inducing atrial fibrillation experimentally include aggressive and sustained rapid burst pacing or methods that alter intra-atrial conduction.6 From these models and studies7–9 of spontaneous atrial fibrillation in humans and other animals, hypotheses for the mechanism of atrial fibrillation have been proposed. Two such hypotheses include the multiple wavelet theory and the mother rotor theory.8,10 For the latter, the most recent theory proposes that a stable (more organized) and rapid rotor in the left atrium propagates action potentials irregularly to the right atrium because of spatially varying conduction blocks, resulting in a left to right frequency gradient (left frequency higher than right frequency).9 The frequency gradient is based on the evaluation of intracardiac ECGs by use of frequency domain analysis that identifies a dominant frequency (ie, the frequency of the largest amplitude sinusoidal waveform).11 Previous studies11,12 have revealed a left to right atrial frequency gradient in vagally induced atrial fibrillation; however, evaluation of data based on this technique requires caution when short, unaveraged atrial fibrillation ECGs are used because of potential inaccuracies caused by the changing amplitudes, activation intervals, and complexity of the atrial ECGs characteristic of atrial fibrillation. Instead, averaging the mean dominant frequency of multiple shorter segments provides more robust information and allows detection of differences in dominant frequency between sites. Furthermore, spectral entropy may be a relevant index of disorder that may add to the understanding of the mechanisms driving atrial fibrillation.13 Applied to the power spectrum of left and right ECG recordings, spectral entropy is a measure of how widely power is spread across the available frequencies. When power is unequally distributed in frequency, spectral entropy provides a reasonable measure of the spread in power and therefore the degree of disorder. The lower the spectral entropy, the more organized the electrical activity.
Two considerations brought about the study reported here. First, we had recently noticed that during a routine electrophysiologic, closed-thorax studya to determine the ERP of the atria in German Shepherd Dogs, atrial fibrillation frequently and unintentionally was induced. The study was performed with anesthesia with fentanylb and pentobarbital,c with no other means of increasing vagal tone. The German Shepherd Dogs were from an established colony with inherited arrhythmias that serve as a model of ventricular arrhythmias.14,15 Although we have observed spontaneous atrial arrhythmias, but never atrial fibrillation, during 24-hour ambulatory ECG monitoring (Holter monitoring), the ease with which atrial fibrillation developed in response to a single extra stimulus (premature paced beat) seemed atypical, compared with the more aggressive methods of inducing atrial fibrillation that are published.5,6 In our studies,a the dogs were anesthetized with an infusion of fentanyl and boluses of pentobarbital. Because the most prominent effect was from the fentanyl, based on the doses used, the heart rate was slowed because of the increased vagal tone caused by the narcotic.16,17 Our second contemplation concerned the underlying mechanism of the atrial fibrillation and the use of the dominant frequency and spectral entropy to determine whether a frequency or organizational gradient existed between the left and right atrium in the German Shepherd Dogs under these conditions. Therefore, the purpose of the study reported here was to determine whether atrial fibrillation could be induced in these dogs as an open-thorax model that would permit future studies (eg, epicardial mapping) of arrhythmia initiation and propagation and to determine the dominant frequency and spectral entropy gradients across the atria during atrial fibrillation. We hypothesized that we would be able to induce atrial fibrillation with atrial pacing while the dogs were anesthetized with fentanyl and pentobarbital anesthesia and that, if the vagolytic drug atropine was administered, induction of sustained atrial fibrillation would be hindered. Moreover, we hypothesized that the left atrium would have a higher fibrillating frequency and more organization, giving evidence that the left atrium was driving the right atrium in atrial fibrillation.
Materials and Methods
All procedures were approved by the Cornell University Institutional Animal Care and Use Committee to ensure humane treatment of all dogs and legitimacy of the research. Experiments before this study were not performed on these dogs. Dogs were euthanatized with a lethal dose of pentobarbital and tissues were collected for other investigations to maximize animal use.
Experimental preparation—Seven German Shepherd Dogs from a breeding colony with inherited ventricular arrhythmias and sudden death (weight [mean ± SD], 25.7 ± 4.3 kg; age, 180 ± 20 days) were studied. Each of these dogs was determined to be afflicted on the basis of a minimum of three 24-hour ambulatory ECG recordings (Holter monitoring). Additionally, the presence of atrial arrhythmias was ascertained. Dogs were instrumented for 2 electrophysiologic protocols (extrastimulus and pacedown) for the induction of atrial fibrillation. Anesthesia was induced with fentanyl citrate administered at 0.02 mg/kg and pentobarbital administered at 10 mg/kg, both given IV as 2 sequential boluses. The dogs were intubated and ventilated with oxygen. Dogs were maintained under general anesthesia with a constant fentanyl infusion at a rate of 0.04 mg/kg/h through a cephalic intravenous catheter. Additional boluses of pentobarbital were given as needed through a second cephalic intravenous catheter to maintain adequate anesthesia. An infusion of lactated Ringer's solutiond at 10 mL/kg/h was continued throughout the procedure. A catheter was placed in the pedal artery for recording of peripheral blood pressure. One surface ECG lead in the frontal plane was installed. All the recordings obtained during the experimentation were digitized by use of a data acquisition system.e A left-sided thoracotomy through the fifth intercostal space and a pericardectomy were performed to expose both atria. Five bipolar electrodes were inserted and sutured in place in the atrial wall. The 3 left-sided electrodes were placed in the region of the left auricle, the pulmonary vein–left atrial junction, and the left Bachman bundle, respectively. The 2 right-sided electrodes were placed in the region of the right atrial body and the right auricle. These electrodes were used for pacing. Two MAP cathetersf were held in place by an operator near the left and the right auricle/atrial body juctions for epicardial MAP recordings. The electrophysiologic protocols were performed from each of the 5 atrial electrodes with a stimulatorg by use of square impulses of 2 milliseconds' pulse duration at twice the stimulation threshold.
Extrastimulus protocol for induction of atrial fibrillation—On the basis of results of our closed-thorax pilot studies, we knew that simply the attempt to pace or determine the atrial ERP could result in atrial fibrillation.a Therefore, we designed a protocol to determine whether attempts to define the ERP in the open-thorax model, which allowed pacing from the left and right atrium, would result in induction of atrial fibrillation. In the 7 dogs, at each of the 5 atrial sites with the electrodes, the atrial ERP was sought at 2 basic cycle lengths of 500 milliseconds and 350 milliseconds. Each site was paced with a pulse-train of 20 stimuli (S1) followed by an extra stimulus (premature stimulus [S2]) with the coupling interval decreased in 20-millisecond steps between 200 milliseconds and 160 milliseconds and 10-millisecond steps from 150 milliseconds to loss of capture. Pacing was then resumed by adding 10 milliseconds to the S1-S2 interval that resulted in loss of capture, and the S1-S2 interval was again decreased in 1-millisecond steps until loss of capture, defining the ERP; that is, the longest S1-S2 interval that failed to elicit atrial depolarization. A similar protocol was repeated with a train of 20 S1 stimuli followed by 2 premature S2-S3 stimuli. The S1-S2 interval was the shortest S1-S2 interval longer than the ERP resulting in a propagated response. The protocol was applied to the S3 stimulus to determine the S3 ERP. The S2-S3 interval was the shortest S2-S3 interval longer than the ERP resulting in a propagated response. If atrial fibrillation was induced with the extra stimulus, the ERP was not possible to determine. The process of delivering the series of extra stimuli at progressively shorter coupling intervals for each site was defined as an attempt to determine the ERP. Either the ERP or the S1-S2 or S2-S3 interval at which atrial fibrillation occurred was noted for each attempt.
Pacedown protocol for induction of atrial fibrillation—A dynamic pacedown protocol was then performed. Each atrial site was paced with 10 S1 stimuli with no pause between the pacing bursts starting at a pacing rate of 300 milliseconds and decreasing the S1-S1 interval in 20-millisecond steps until 160 milliseconds and then decreasing by 10-millisecond steps. Pacing was stopped when a 2:1 stimulus-to-atrial conduction block ratio developed or at the initiation of atrial fibrillation.
Protocol after atropine for induction of atrial fibrillation—To determine whether the vagomimetic effect of fentanyl contributed to the induction of atrial fibrillation, atropineh at 0.04 mg/kg was administered IV. The pacing protocols were repeated at 1 left and 1 right atrial site for which atrial fibrillation was induced or, if atrial fibrillation had not been induced, from the pulmonary vein–atrial site or the body of the right atrial body. Pacing protocols were repeated after the heart rate had increased by 50% and stabilized for approximately 2 to 3 minutes. Therefore, after IV administration of atropine, the time lapse was > 5 minutes.
Defining atrial fibrillation—Atrial fibrillation was defined as the presence of f waves on the surface ECG with accompanying fractionated or multipotential ECGs with changing amplitudes and morphologies from at least one of the MAP recordings. Each episode of atrial fibrillation was reviewed to determine whether induction was the result of a premature extra stimulus that occurred spontaneously, an extra stimulus from the pacing stimulus, or the pacedown pacing. Results of pilot studies indicated that atrial fibrillation lasting more than 8 minutes did not convert back to sinus rhythm but required cardioversion (monitored for more than 1 hour).a Therefore, to minimize the extensive experimental time, sustained atrial fibrillation was defined as that lasting longer than 8 minutes. When an episode of sustained atrial fibrillation was present, lidocainei (2 mg/kg) was given IV. To avoid excessive prolongation of the experiment, lidocaine was administered after 5 minutes of atrial fibrillation if at least one 8-minute episode of atrial fibrillation requiring lidocaine conversion occurred.
Signal analysis—Left and right MAP recordings of sustained episodes of atrial fibrillation were analyzed to determine spectral entropy and dominant frequency.j Sampling was in 1-minute intervals with 1-minute separations when multiple windows were analyzed within a single episode of atrial fibrillation. Only regions with excellent MAP recordings were included. The 1-minute intervals were long enough to acquire several measures of the spectral entropy (about 15) while potentially short enough so that substantial changes in fundamental properties did not have time to occur. The 4-second moving mean was subtracted from each 1-minute interval prior to analysis to remove any slow drifts in the signal. Fast Fourier transform was used for the spectral analysis. Power spectra were calculated for 4-second time segments with no overlap for each time record. The 4-second time segments used for calculation of the power spectra translated into a frequency resolution of 0.25 Hz. This short segment was sufficient to permit the fastest important changes to be seen and long enough to produce the desired resolution in frequency. A resolution finer than 1 Hz was considered ideal because this would permit typical shifts of 1 to 3 Hz to be well documented. The DC (0 Hz) component was then removed. This procedure was performed to remove any remaining residual baseline drift (found to be necessary; otherwise 0 Hz was sometimes incorrectly marked as the dominant frequency).
The dominant frequency was defined for each recording as the frequency containing the most power. It was used as an estimation of the atrial activation rate. The spectral entropy in the left and right atria also was determined by use of the MAP recordings. Spectral entropy is an index that indicates how widely power is distributed among various frequencies and therefore is a measure of the randomness of the signal. Thus, low spectral entropy is interpreted as a more organized system, and high spectral entropy corresponds to a less organized system. Spectral entropy simply measures the degree of spread of power across various frequencies, without regard to how far these frequencies are from the mean frequency, and is thus a better measure of disorder.18,19 For example, the presence of alternans spreads energy to only 1 other frequency (ie, the frequency that is half the dominant frequency) and thus increases the spectral entropy only modestly, whereas the difference in frequency that exists between this half frequency and the dominant frequency substantially increases the SD measure. Thus, spectral entropy performs better in the presence of alternans because alternans would not generally be thought of as a phenomenon representing a substantial increase in disorder. For spectral entropy, the fraction p of the total power at each frequency f was obtained by dividing each power spectrum component in a given time segment by the sum of all the power components in that segment. The spectral entropy (SE) for a given 4-second time segment was then calculated from the following equation:
Statistical analysis—Computer software programsk,l were used to perform statistical analysis. The difference in the number of atrial fibrillation inductions by protocol was analyzed by use of a Fisher exact test 2-sided contingency table, and continuous atrial fibrillation duration data were analyzed by use of the Mann-Whitney test because the data were non-Gaussian. Dominant frequency and spectral entropy data were analyzed by use of repeated-measures ANOVA. There were 2 repeated-measures factors: side (left atrium vs right atrium) and time of a given response. The variability among dogs was treated as a between-subject (grouping) factor in the ANOVA. For main effects, a value of P < 0.05 was considered significant, whereas for interactions, a value of 0.01 was considered significant for the 3 factors of interaction. When nonparametric statistics were used, median and range were reported; when parametric statistics were used, mean ± SD was reported.
Results
Induction of atrial fibrillation—In addition to the polymorphic ventricular arrhythmias characteristic of this inherited disorder in the German Shepherd Dogs, 6 of the 7 dogs also had atrial premature complexes, and 2 dogs had short runs of atrial tachycardia detected via Holter monitoring. Holter monitoring did not reveal any episodes of atrial fibrillation.
During the electrophysiologic studies, spontaneous atrial arrhythmias occurred (Figure 1) from both the left and right atria as judged by use of the MAP recording that indicated the atrium with the earliest activation. From these spontaneous atrial depolarizations, atrial fibrillation was triggered in 6 of 7 dogs, with 15 atrial fibrillation episodes. Seven of the 15 episodes were sustained atrial fibrillation (each dog with 1 episode and 1 dog with 2). Also, atrial fibrillation was induced during the electrophysiologic protocols in all 7 dogs (Table 1). During the extrastimulus protocol, all 7 dogs developed atrial fibrillation (Figure 2). Sustained atrial fibrillation was induced in 6 of 7 dogs with the extrastimulus protocol, compared with 1 of 7 dogs with the pacedown protocol (Figure 3).
Distribution of responses to induction of AF in 7 German Shepherd Dogs via spontaneous premature atrial depolarizations, an extrastimulus protocol, and a pacedown protocol.
| Trigger for induction of AF | No. of events | |
|---|---|---|
| Spontaneous premature atrial depolarizations* | ||
| No. of nonsustained AF events | 8 | |
| No. of sustained AF events | 7 | |
| Left atrium | Right atrium | |
| Extrastimulus protocol | ||
| S1-S2 protocol | ||
| Attempts to determine ERP | 44 | 32 |
| S1-S1 500 ms | 22 | 15 |
| S1-S1 350 ms | 22 | 17 |
| No. of nonsustained AF events | ||
| S1-S1 500 ms | 0 | 4 |
| S1-S1 350 ms | 0 | 3 |
| No. of sustained AF events | ||
| S1-S1 500 ms | 3 | 0 |
| S1-S1 350 ms | 2 | 4 |
| S1-S2, S3 protocol | ||
| Attempts to determine ERP | 35 | 13 |
| S1-S1 500 ms | 13 | 6 |
| S1-S1 350 ms | 22 | 7 |
| No. of nonsustained AF events | ||
| S1-S1 500 ms | 1 | 1 |
| S1-S1 350 ms | 2 | 1 |
| No. of sustained AF events | ||
| S1-S1 500 ms | 1 | 0 |
| S1-S1 350 ms | 3 | 0 |
| Extra stimuli during pacedown† | ||
| No. of nonsustained AF events | 2 | 1 |
| No. of sustained AF | 1 | 1 |
| Pacedown protocol | ||
| Attempts | 16 | 10 |
| No. of nonsustained AF events | 7 | 3 |
| No. of sustained AF events | 1 | 1 |
Origin of spontaneous premature atrial depolarization unknown.
Induction of AF was caused by a single extra stimulus and not as the result of the pacedown.
AF = Atrial fibrillation. ms = Millisecond.
Electrocardiographic recording of a German Shepherd Dog during baseline in a study of atrial fibrillation. Increased vagal tone is indicated by the atrioventricular block. A nonsustained run of atrial tachycardia is seen in the middle of the panel. Likely because of the high vagal tone and the rapid rate, atrioventricular conduction does not occur during atrial tachycardia. Lt = Left. Rt = Right. Bar = 1 second.
Citation: American Journal of Veterinary Research 69, 11; 10.2460/ajvr.69.11.1434
Electrocardiographic recording of a German Shepherd Dog during an extrastimulus protocol. Atrial fibrillation was induced with a single premature depolarization. The S1-S1 pulse train was 500 milliseconds with an S1-S2 coupling interval of 150 milliseconds. The stimulation channel indicates the pacing of the left atrium in this dog (2 artifact pulses are seen in the channel also). The right MAP recording appears more fractionated than the left MAP. The left MAP reveals electrical alternans in the latter portion of the frame. SM = Stimulus marker. Bld PR = Blood pressure. See Figure 1 for remainder of key.
Citation: American Journal of Veterinary Research 69, 11; 10.2460/ajvr.69.11.1434
Electrocardiographic recording of a German Shepherd Dog obtained during a rapid pacedown protocol. Atrial fibrillation was induced at an S1-S1 interval of 130 milliseconds (changes to 120 milliseconds at the dotted line of panel B). Numerous ventricular premature complexes are also evident. In panel A, electrical alternans is apparent in the left MAP and right MAP recordings. Panel B overlaps panel A during the bracketed time. Although present in the left MAP and right MAP recordings, electrical alternans is most apparent in the recording from the left auricle. Atrial fibrillation with fractionation of the electrogram is apparent in the MAP recordings on the right of panel B following the electrical alternans. The variation in the amplitude of the stimulus marker is a result of digital recording. See Figures 1 and 2 for key.
Citation: American Journal of Veterinary Research 69, 11; 10.2460/ajvr.69.11.1434
The induction of either sustained or nonsustained atrial fibrillation (relative to the number of attempts) was more frequent with the pacedown protocol (P = 0.01). Atrial fibrillation was sustained in 2 of 26 (7.7%) pacedown attempts, and atrial fibrillation was nonsustained in 10 of 26 (38.5%) pacedown attempts, whereas atrial fibrillation was sustained in 13 of 124 (10.5%) extrastimulus attempts, and atrial fibrillation was nonsustained in 12 of 124 (9.7%) extrastimulus attempts. Atrial fibrillation developed 16 times during 76 (21.1%) attempts with a single extra stimulus (S2) and 9 times during 48 (18.8%) attempts with 2 extra stimuli (S2-S3). Sustained atrial fibrillation occurred in 13 of 25 (52%) episodes of atrial fibrillation induced during the extrastimulus protocol and in 2 of 12 episodes of atrial fibrillation induced during pacedown (P = 0.07). Additionally, during the pacedown protocol, 5 episodes of atrial fibrillation were attributed to a premature complex rather than the rapid pacing train (eg, initiation at the beginning of train that acted as an extra stimulus). These episodes were not attributed to the pacedown protocol in the statistical analysis. The duration of the nonsustained AF between the extrastimulus protocol (median, 96.5 seconds; range, 2 to 474 seconds) and pacedown protocol (median, 39 seconds; range, 1 to 240 seconds) did not differ (P = 0.31). Although brief (< 1-second) electrical alternans would intermittently develop during atrial fibrillation (Figure 2), electrical alternans (Figure 4) was noted to occur prior to atrial fibrillation induction during pacedown, indicating a probable different mechanism for induction of atrial fibrillation, compared with the extrastimulus protocol. The total number of episodes of atrial fibrillation induced by means of the extrastimulus protocol were not different for the left versus right atrium (P = 0.10), although sustained atrial fibrillation was more likely to occur from a left versus right atrial stimulus (P = 0.047) The S1-S2 coupling intervals during the extrastimulus protocol at 350 (median, 112.5 milliseconds; range, 106 to 131 milliseconds) and 500 milliseconds (median, 120 milliseconds; range, 110 to 200 milliseconds) from the right atrium and at 350 (median, 125 milliseconds; range, 95 to 155 milliseconds) and 500 milliseconds (median, 110 milliseconds; range, 78 to 200 milliseconds) from the left atrium that induced atrial fibrillation were not significantly different. The S2-S3 coupling intervals were too few for comparisons.
Dominant frequency recorded in a German Shepherd Dog during a sustained run of atrial fibrillation. The frequency pattern in the right atrium is broader than in the left atrium, with the frequency of the left atrial activity clustering closer to 9 Hz. However, during some periods of atrial fibrillation, the frequency decreases to half in the left atrium. Inspection of the left monophasic action potential recordings revealed electrical alternans during these periods. Darker-colored circles represent overlapping values. LA = Left atrium. RA = Right atrium.
Citation: American Journal of Veterinary Research 69, 11; 10.2460/ajvr.69.11.1434
Induction of atrial fibrillation after atropine—The induction of longer runs of atrial fibrillation was more difficult after atropine administration. Before atropine administration, 37 episodes of atrial fibrillation were induced by either protocol with 150 attempts; after atropine administration, 6 episodes of atrial fibrillation were induced with 41 attempts (P = 0.20; Table 2). However, after atropine administration, no episodes of sustained atrial fibrillation occurred with either protocol (P = 0.026), and there were no episodes of spontaneous atrial fibrillation. Also, the duration of the nonsustained atrial fibrillation after atropine (median, 5 seconds; range, 1 to 26 seconds) was significantly (P = 0.008) shorter than before atropine (median, 65 seconds; range, 1 to 474 seconds) was administered. Atrial tachycardia was induced during the extrastimulus protocol once in a dog and during the pacedown protocol 5 times in 3 dogs. Five of the 6 episodes of atrial tachycardia were sustained. Before atropine administration, sustained atrial tachycardia was not induced.
Distribution of responses to induction of AF and atrial tachycardia in 7 German Shepherd Dogs via spontaneous premature atrial depolarizations, an extrastimulus protocol, and a pacedown protocol after administration of atropine.
| Triggers for induction of AF and AT | No. of events | |
|---|---|---|
| Spontaneous premature atrial depolarizations* | ||
| No. of nonsustained AF events | 0 | |
| No. of sustained AF events | 0 | |
| Left atrium | Right atrium | |
| Extrastimulus protocol | ||
| S1-S2 protocol | ||
| Attempts to determine ERP | 13 | 8 |
| S1-S1 500 ms | 3 | 2 |
| S1-S1 350 ms | 10 | 6 |
| No. of nonsustained AF events | ||
| S1-S1 500 ms | 0 | 0 |
| S1-S1 350 ms | 0 | 0 |
| No. of sustained AT events | ||
| S1-S1 500 ms | 1 | 0 |
| S1-S1 350 ms | 0 | 0 |
| S1-S2, S3 protocol | ||
| Attempts to determine ERP | ||
| S1-S1 500 ms | 0 | 0 |
| S1-S1 350 ms | 7 | 4 |
| No. of nonsustained AF events | ||
| S1-S1 500 ms | 0 | 0 |
| S1-S1 350 ms | 0 | 1 |
| No. of sustained AT events | ||
| S1-S1 500 ms | 0 | 0 |
| S1-S1 350 ms | 1 | 0 |
| Extra stimuli during pacedown† | ||
| No. of nonsustained AF events | 1 | 0 |
| No. of sustained AT | 0 | 0 |
| Pacedown protocol | ||
| Attempts | 6 | 3 |
| No. of nonsustained AF events | 3 | 2 |
| No. of sustained AT events | 3 | 2 |
AT = Atrial tachycardia.
See Table 1 for remainder of key.
Frequency and entropy signal analysis—Five episodes of sustained atrial fibrillation from 5 dogs had MAP signals that were suitable for spectral analysis (Figure 5). Less than ideal quality or temporary loss of the left or right atrial MAP signal precluded the use of the other recordings. We identified sixteen 40-second segments for frequency domain analysis, with a maximum of 6 segments/episode of atrial fibrillation. Three runs of sustained atrial fibrillation were triggered by spontaneous atrial premature complexes and 2 by a left-sided extrastimulus protocol.
Illustration of a single selected 1-minute window from cardiologic evaluation of a German Shepherd Dog, including ECG, left MAP, and right MAP recordings during atrial fibrillation (A). Spectral entropy (B) and dominant frequency (C) were analyzed. Mean spectral entropy was significantly (P < 0.001) higher on the right (mean ± SD, 3.85 ± 0.036) than the left (3.72 ± 0.07). Mean dominant frequency was significantly (P < 0.001) higher on the left (10.5 ± 0.46 Hz) than the right (7.21 ± 1.69 Hz). See Figure 1 for key.
Citation: American Journal of Veterinary Research 69, 11; 10.2460/ajvr.69.11.1434
There was no significant dog effect on spectral entropy (nondimensional, unitless; P = 0.059). There was a significant (P < 0.001) side effect (left [3.67 ± 0.08] versus right [3.81 ± 0.03] atrium) on spectral entropy; mean spectral entropy for the right atrium was higher than that for the left atrium. Moreover, there were no individual dogs in which the mean spectral entropy recorded from the left side was higher than that recorded from the right side (Table 3). There were no other significant interactions among the factors side, time, and dog. The time effect, when adjusted for the lack of sphericity relating to the spectral entropy responses recorded over time, was significant (P = 0.008), with an increasing spectral entropy over time (Table 4). Polynomial contrast analysis indicated that only a linear trend in time of the spectral entropy was significant (P = 0.011), with a positive slope.
Spectral entropy and dominant frequency of the left atrium and right atrium during atrial fibrillation in 5 German Shepherd Dogs.
| Dog No. | Atrium | Spectral entropy | Dominant frequency | ||
|---|---|---|---|---|---|
| Mean ± SD | 95% confidence interval | Mean ± SD | 95% confidence interval | ||
| 1 | Left | 3.61 ± 0.24 | 3.47–3.74 | 9.71 ± 1.76 | 8.74 – 10.68 |
| Right | 3.80 ± 0.10 | 3.75 – 3.86 | 7.63 ± 2.52 | 6.24 – 9.01 | |
| 2 | Left | 3.55 ± 0.24 | 3.41 – 3.68 | 9.35 ± 1.76 | 8.39 – 10.32 |
| Right | 3.80 ± 0.10 | 3.75 – 3.86 | 10.7 ± 2.52 | 9.37 – 2.13 | |
| 3 | Left | 3.76 ± 0.14 | 3.68 – 3.83 | 8.93 ± 1.02 | 8.37 – 9.49 |
| Right | 3.83 ± 0.06 | 3.80 – 3.86 | 6.58 ± 1.45 | 5.78 – 7.38 | |
| 4 | Left | 3.75 ± 0.11 | 3.69 – 3.81 | 9.83 ± 0.79 | 9.40 – 10.26 |
| Right | 3.79 ± 0.04 | 3.76 – 3.81 | 10.12 ± 1.12 | 9.50 – 10.74 | |
| 5 | Left | 3.71 ± 0.10 | 3.65 – 3.76 | 10.35 ± 0.72 | 9.96 – 10.75 |
| Right | 3.84 ± 0.04 | 3.82 – 3.86 | 7.92 ± 1.03 | 7.36 – 8.49 | |
Spectral entropy and dominant frequency of atrial fibrillation over time in 5 German Shepherd Dogs.
| Time division | Spectral entropy | Dominant frequency | ||
|---|---|---|---|---|
| Mean ± SD | 95% confidence interval | Mean ± SD | 95% confidence interval | |
| 1 | 3.72 ± 0.06 | 3.69 – 3.75 | 9.42 ± 1.25 | 8.73 – 10.10 |
| 2 | 3.72 ± 0.06 | 3.69 – 3.75 | 8.90 ± 1.50 | 8.07 – 9.72 |
| 3 | 3.73 ± 0.04 | 3.71 – 3.76 | 8.94 ± 1.21 | 8.27 – 9.61 |
| 4 | 3.74 ± 0.04 | 3.71 – 3.76 | 8.42 ± 0.94 | 7.90 – 8.94 |
| 5 | 3.74 ± 0.04 | 3.72 – 3.77 | 9.31 ± 1.12 | 8.69 – 9.93 |
| 6 | 3.75 ± 0.06 | 3.72 – 3.79 | 8.88 ± 1.66 | 7.97 – 9.78 |
| 7 | 3.75 ± 0.07 | 3.71 – 3.79 | 9.38 ± 1.32 | 8.66 – 10.11 |
| 8 | 3.76 ± 0.06 | 3.72 – 3.79 | 9.23 ± 1.10 | 8.63 – 9.84 |
| 9 | 3.76 ± 0.05 | 3.73 – 3.78 | 8.89 ± 1.69 | 7.96 – 9.82 |
| 10 | 3.76 ± 0.05 | 3.74 – 3.79 | 9.79 ± 1.26 | 9.10 – 10.49 |
In contrast to the spectral entropy, the mean dominant frequency was significantly (P < 0.001) different between dogs. In the ANOVA, although the effect of side was significant (P < 0.001), with the left atrium having a higher overall mean dominant frequency, it was associated with a significant (P < 0.001) interaction between the factors side and dog. Subsequent analysis revealed that mean dominant frequency for the left atrium was higher than that for the right atrium in 3 dogs, with a mean difference of 2.29 Hz, whereas dominant frequency was higher in the right atrium than the left atrium in 2 dogs, with a mean difference of 0.84 Hz. These results revealed an inconsistency when dominant frequency is evaluated. The other interactions (all P values > 0.18) and the time effect (P = 0.14) were not significant (Table 4).
Discussion
The major findings of this study in German Shepherd Dogs with inherited ventricular arrhythmias anesthetized with fentanyl and pentobarbital supported our original hypotheses that atrial fibrillation was readily induced with 2 pacing protocols (extrastimulus and pacedown) and that an organizational gradient exists between the left and right atria.5 It should be emphasized that before any pacing protocols were performed, 6 of the 7 dogs developed spontaneous atrial fibrillation. Throughout all recordings, careful examination of all induced episodes of atrial fibrillation was performed to identify specifically the initiation or trigger of the atrial fibrillation. The pacedown protocol is one that is used experimentally to induce atrial fibrillation, whereas the single extra stimulus is not the generally preferred method for inducing atrial fibrillation in experimental models. Nevertheless, the latter, less aggressive protocol induced atrial fibrillation in these German Shepherd Dogs under the described conditions.5,6 Results of spectral entropy analysis indicated that the electrical activity in the atria was more organized in the left atrium during the induced atrial fibrillation. Also, we identified electrical alternans as a precedent event to atrial fibrillation for the pacedown protocol.
From this study, it could not be determined whether these particular dogs had a greater propensity for the development of atrial fibrillation than other dogs. The ventricular arrhythmias that exist in German Shepherd Dogs with this inherited disorder are well established; however, the atrial arrhythmias have not been investigated thoroughly, although it should be emphasized that the number of premature atrial arrhythmias detected by use of 24-hour Holter monitoring is low.20 Moreover, atrial fibrillation in these dogs has never been detected except under the conditions of anesthesia with fentanyl and pentobarbital. Conversely, it is important to note that these dogs developed spontaneous atrial fibrillation without initiation from the pacing protocols, induced instead by presumed inherent atrial premature beats. A pilot study with 3 age- and weight-matched mongrel dogs anesthetized with the same protocol revealed that 2 mongrel dogs also developed atrial fibrillation. However, we still do not know whether there may be some greater tendency for the German Shepherd Dogs to develop atrial fibrillation during conditions that increase vagal tone. Increased vagal tone that is caused by clinically used anesthetics can be a strong modifier or trigger for atrial fibrillation. This is supported in part by the finding that the induction and maintenance of atrial fibrillation after administration of the vagolytic drug atropine resulted in shorter runs of nonsustained atrial fibrillation and no sustained atrial fibrillation. Recently, we have determined that in large-breed dogs administered drugs that are vagomimetic, spontaneous atrial fibrillation has developed.3 The results of that study therefore support the theory that vagal tone is an important contributor not only to the initiation of atrial fibrillation but also to its maintenance or perpetuation. It is possible that these particular German Shepherd Dogs with inherited ventricular arrhythmias may have more of a propensity to develop atrial fibrillation under conditions of high vagal tone with less of a trigger than other dogs; however, this cannot be determined from the present study. The number of German Shepherd Dogs and control dogs required to answer this question would likely be prohibitive.
Increased vagal tone through direct vagal nerve stimulation or the use of cholinergic agonists has been extensively applied to experimental models of atrial fibrillation.5 We used fentanyl, a synthetic opiate analgesic responsible for vagally mediated bradycardia. Its action is via pre- and postsynaptic μ-opioid receptors and results in a reduction of γ-aminobutyric acid neurotransmission to cardiac vagal neurons.17 The resulting cholinergic stimulation reduces the ERP of the atrial myocytes in a spatially heterogeneous fashion. These 2 electrophysiologic changes promote some forms of atrial fibrillation because they sustain reentry. Waves of depolarization propagate through the atria as the myocytes recover from their refractory period in a heterogeneous manner, creating pathways of excitable tissue around areas of block.5,6,21 This nonuniform effect of vagal stimulation on the atria is attributed to heterogeneous vagal innervation, unequal distribution of muscarinic receptors, and higher current density of IKACh in the left atrium than in the right atrium. Also, IKACh is responsible for action potential duration abbreviation when stimulated by acetylcholine.22,23
Potential mechanisms to explain why rapid pacing can cause atrial fibrillation, even in a normal heart, have been hypothesized. One explanation is that a stimulus train triggers the release of acetylcholine at the atrial level and decreases atrial fibrillation threshold.2 In the present study, in which the pacedown protocol was used, atrial fibrillation was preceded by electrical alternans, which is a mechanism that has been proposed for the induction of fibrillation.24 Although most studies concern electrical alternans and degeneration to ventricular fibrillation, recent attention has been given to its role in atrial fibrillation.25 However, identification of electrical alternans as a precedent to fibrillation of the atria has rarely been reported.25 Development of fibrillation can be related to the relationship of the diastolic interval and the action potential duration. As heart rate increases, both of these time periods shorten with a specific relationship known as the restitution curve.24 If this relationship, or slope of the line representing it, is too steep, with a slope > 1, electrical alternans will develop with an alternating action potential. The faster the heart rate, the steeper the slope of the restitution curve, and this leads to instability and fibrillation. The electrical alternans is hypothesized to be the result of alternating intracellular calcium concentration at the fast rates that affect repolarization. Calcium current is decreased with parasympathetic stimulation and increased with sympathetic stimulation. Moreover, each of these effects is likely not homogeneous throughout the atrial tissue, which promotes arrhythmias. What is notable from the present study is the ease with which atrial fibrillation was induced by simple, spontaneous, premature atrial beats or a single extra stimulus during pulse trains that were not fast. Therefore, in the dogs reported here, the role of the parasympathetic system was important for modulating the susceptibility to atrial fibrillation, which either may be specific to this affected breed or may occur simply because of tendencies inherent in a large-breed dog.
Atropine administration prevented induction of prolonged episodes of atrial fibrillation, but pacing protocols resulted in sustained atrial tachycardia. It has been revealed in an experimental preparation with canine cardiac tissue that flutter-like activity can be triggered with low concentrations of acetylcholine, and fibrillation occurs at higher concentrations. This effect is associated with detachment of the reentrant wave fronts from anatomic structures.26 It is possible that the dose of atropine used in the present study did not totally inhibit cholinergic influences on atrial myocyte. We did wait for > 5 minutes after atropine administration before the pacing protocols were repeated, and this would have eliminated any vagomimetic effects of atropine that can occur, usually during the first 1 to 2 minutes after IV administration.27
In the present study, we consistently identified a higher degree of organization of the electrical activity in the left atrium, as reflected by a lower mean spectral entropy number, compared with the right atrium. At the same time, the left auricular region was activated at a higher frequency in 3 dogs and a slightly lower frequency in 2 dogs, compared with the right atrium. These results agree with the current hypothesis that under conditions of increased vagal tone, atrial fibrillation has some degree of spatio-temporal organization.2,5,9 Studying 4-second-long segments of sustained atrial fibrillation that were initiated by burst pacing in a Langendorff-perfused sheep heart preparation after addition of acetylcholine, Skanes et al28 detected spatio-temporal periodicity in the left atrium in 100% of the segments studied. Dominant frequency was higher in the left atrium in 74% of the recordings, whereas in the remaining 24%, the dominant frequency was similar in the left and the right atria.28 In the experiments of Skanes et al,28 optical mapping of the left and right atria identified a high-frequency periodic wave of reentry in the left atrium. Activation of the right atrium resulted from complex patterns of propagation from this left atrial source. Thus, it is likely that, in our experiment, the left atrial MAP recording was close to a similarly stable, high-frequency area, resulting in more dynamic organization and lower spectral entropy of the MAP signal. Moreover, the temporal changes in the left and right spectral entropy may reflect periodic changes of atrial fibrillation spatio-temporal organization that have not been identified by other groups that were studying much shorter segments of sustained atrial fibrillation (3 to 4 seconds).28 In a recent study29 of vagally induced atrial fibrillation in dogs, researchers concluded that atrial fibrillation resulted from multiple rotors spinning off several daughter wavelets that were colliding with each other. It is possible that, in our experiment, multiple rotors were present. This could explain the variability of mean dominant frequency between dogs.
The organization of atrial activity has been studied by applying fast Fourier transform analysis to the power spectrum of the atrial ECG. Instead of being distributed homogeneously across a wide range of frequencies, power is usually concentrated around 1 dominant frequency.12 The presence of a dominant peak suggests that the atria were potentially activated in a spatially ordered manner.25 In our experiment, we used the same method to evaluate atrial fibrillation organization, but with the mean value of multiple samples (15 windows/ min).11 The spectral entropy had more consistency than the dominant frequency, with regards to suggesting that the left atrium was the source of the dominant wavefront. Problems with only using the dominant frequency have been recently addressed.11,30 Additionally, electrical alternans can cause an inaccurate assessment of the dominant frequency. Specifically, only the frequency with the largest amplitude is detected by the dominant frequency analysis; thus, it is possible for half of the main frequency to be recorded as the dominant frequency when alternans, which can contribute substantial power at this frequency, is strong. We saw this type of activity not only before the induction of atrial fibrillation during the pacedown protocol but also during atrial fibrillation regardless of the initiating event.
As with many studies, the number of animals studied was small. Although all dogs developed episodes of atrial fibrillation, the analysis by episode may have been skewed because each episode may not have been independent. We did not randomize between the attempts to induce by premature stimulus and pacedown protocols. We wanted to use the less aggressive protocol first to assess the frequency of atrial fibrillation induction by this technique, compared with the pacedown method used in other published studies of atrial fibrillation. Electrical remodeling can occur during rapid pacing, and we wanted to limit this effect. Having said this, regardless of the method selected once atrial fibrillation was induced, this had the potential of altering the electrical environment for rhythms, paced and spontaneous, that followed. Moreover, we wanted to ascertain the difference in left versus right dominant frequency and spectral entropy and did not know the effects of order on these variables. Finally, during the sustained atrial fibrillation, although we waited for what we believed was an adequate amount of time between the administration of lidocaine and the next pacing sequence, we did not know the tissue concentrations of this drug at each time point. Consequently, it was difficult to escape the downstream effects of previous pacing, rhythm, or treatment with certainty.
During this experiment, blood concentrations of anesthetic agents and particularly fentanyl were not monitored. Variation of blood concentrations among dogs and during an experiment may have influenced the inducibility of atrial fibrillation. We evaluated atrial fibrillation organization from 2 MAP recordings located near the atrial appendages. Optical mapping of left and right auricles during atrial fibrillation has revealed that the organization of the atrial signal can substantially vary over short distances. The variability of the signal increases with the distance from the source of periodic activity.28 Therefore, the variability of atrial fibrillation organization among recordings may have been influenced by the location of the MAP catheters that were manually held on both auricles. It is likely that more information could be obtained from multiple atrial recordings, which would in particular allow determination of the origin of atrial fibrillation. We did not study a control group of dogs to determine whether our results were restricted to these German Shepherd Dogs or solely the result of the anesthetic-vagal impact. That was not the purpose of this investigation. Results of the pilot study indicated that a large number of dogs would be required to determine a breed predisposition, and this could not be justified.
The results reported here brought our attention to dogs with accompanying disorders that could also have increased vagal tone that developed atrial fibrillation after the administration of narcotics.3 Since that report,3 we have documented the same scenario in an additional 8 large-breed dogs.
ABBREVIATIONS
| ERP | Effective refractory period |
| MAP | Monophasic action potential |
Pariaut R, Koetje BD, Renaud-Farrell S, et al. Spatiotemporal organization of induced atrial fibrillation in German shepherd dogs with inherited ventricular arrhythmias (abstr). J Vet Intern Med 2005;19:416.
Fentanyl citrate, Hospira Inc, Lake Forest, Ill.
Nembutal, Abbott, Chicago, Ill.
Baxter Healthcare Corp, Deerfield, Ill.
BioPac Systems, Model MP 150, AcqKnowledge 3.7.3, Santa Barbara, Calif.
EP Technologies, Sunnyvale, Calif.
Bloom Stimulator, Fischer Imaging Corp, Denver, Colo.
Atropine, Vedco, St Joseph, Mo.
Lidocaine, Vedco, St Joseph, Mo.
MatLab, version 6.5, The Math Works Inc, Natik, Mass.
GraphPad Prism, version 4, GraphPad Software, San Diego, Calif.
SPSS, version 15.0, SPSS Inc, Chicago, Ill.
References
- 1.↑
Menaut P, Belanger M, Beauchamp G, et al. Atrial fibrillation in dogs with and without structural or functional cardiac disease: a retrospective study of 109 cases. J Vet Cardiol 2005;7:75–83.
- 3.↑
Moise NS, Pariaut R, Gelzer ARM, et al. Cardioversion with lidocaine of vagally associated atrial fibrillation in two dogs. J Vet Cardiol 2005;7:143–148.
- 4.
McGuirk SM, Muir WW. Diagnosis and treatment of cardiac arrhythmias. Vet Clin North Am Equine Pract 1985;1:353–356.
- 6.↑
Wilber DJ, Morton JB. Vagal stimulation and atrial fibrillation: experimental models and clinical uncertainties. J Cardiovasc Electrophysiol 2002;13:1280–1282.
- 7.
Gelzer ARM, Moise NS, Vaidya D, et al. Temporal organization of atrial activity and irregular ventricular rhythm during spontaneous atrial fibrillation: an in vivo study in the horse. J Cardiovasc Electrophysiol 2000;11:773–784.
- 8.
Moe GK. A conceptual model of atrial fibrillation. J Electrocardiol 1968;1:145–146.
- 9.↑
Jalife J. Experimental and clinical AF mechanisms: bridging the divide. J Interv Card Electrophysiol 2003;9:85–92.
- 10.
Jalife J, Berenfeld O, Mansour M. Mother rotors and fibrillatory conduction: a mechanism of atrial fibrillation. Cardiovasc Res 2002;54:204–216.
- 11.↑
Ng J, Kadish AH, Goldberger JJ. Effect of electrogram characteristics on the relationship of dominant frequency to atrial activation rate in atrial fibrillation. Heart Rhythm 2006;3:1295–1305.
- 12.↑
Mansour M, Mandapati R, Berenfeld O, et al. Left-to-right gradient of atrial frequencies during acute atrial fibrillation in the isolated sheep heart. Circulation 2001;103:2631–2636.
- 13.↑
Barbaro V, Bartolini P, Calcagnini G, et al. Automated classification of human atrial fibrillation fromintraatrial electrograms (Erratum published in Pacing Clin Electrophysiol 2000;23:viii). Pacing Clin Electrophysiol 2000;23:192–202.
- 14.
Moise NS, Meyers-Wallen V, Flahive WJ, et al. Inherited ventricular arrhythmias and sudden death in German shepherd dogs. J Am Coll Cardiol 1994;24:233–243.
- 15.
Moise NS. Inherited arrhythmias in the dog: potential experimental models of cardiac disease. Cardiovasc Res 1999;44:37–46.
- 16.
Moise NS, Moon PF, Flahive WJ, et al. Phenylephrine-induced ventricular arrhythmias in dogs with inherited sudden death. J Cardiovasc Electrophysiol 1996;7:217–230.
- 17.↑
Griffioen KJ, Venkatesan P, Huang ZG, et al. Fentanyl inhibits GABAergic neurotransmission to cardiac vagal neurons in the nucleus ambiguus. Brain Res 2004;1007:109–115.
- 18.
Rezek IA, Robert SJ. Stochastic complexity measures for physiological signal analysis. IEEE Trans Biomed Eng 1998;45:1186–1190.
- 19.
Powell GE, Percival IC. A spectral entropy method for distinguishing regular and irregular motion of Hamiltonian systems. J Phys A Math Gen 1979;12:2053–2071.
- 20.↑
Moise NS, Gilmour RF Jr, Riccio ML, et al. Diagnosis of inherited ventricular tachycardia in German shepherd dogs. J Am Vet Med Assoc 1997;210:403–410.
- 21.
Liu L, Nattel S. Differing sympathetic and vagal effects on atrial fibrillation in dogs: role of refractoriness heterogeneity. Am J Physiol 1997;273:H805–H816.
- 22.
Sarmast F, Kolli A, Zaitsev A, et al. Cholinergic atrial fibrillation: I(K,ACh) gradients determine unequal left/right atrial frequencies and rotor dynamics. Cardiovasc Res 2003;59:863–873.
- 23.
Euler DE, Scanlon PJ. Acetylcholine release by a stimulus train lowers atrial fibrillation threshold. Am J Physiol 1987;253:H863–H868.
- 24.↑
Koller ML, Maier SKG, Gelzer ARM. Altered dynamics of action potential restitution and alternans in humans with structural heart disease. Circulation 2005;112:1542–1548.
- 25.↑
Narayan SM, Bode F, Karasik PL, et al. Alternans of action potential during atrial flutter as a precursor to atrial fibrillation. Circulation 2002;106:1968–1973.
- 26.↑
Wu TJ, Kim YH, Yashima M, et al. Progressive action potential duration shortening and the conversion from atrial flutter to atrial fibrillation in the isolated canine right atrium. J Am Coll Cardiol 2001;38:1757–1765.
- 27.↑
Rishniw M, Tobias AH, Slinker BK. Characterization of chronotropic and dysrhythmogenic effects of atropine in dogs with bradycardia. Am J Vet Res 1996;57:337–341.
- 28.↑
Skanes AC, Mandapati R, Berenfeld O, et al. Spatiotemporal periodicity during atrial fibrillation in the isolated sheep heart. Circulation 1998;98:1236–1248.
- 29.↑
Everett TH IV, Wilson EE, Verheule S, et al. Structural atrial remodeling alters the substrate and spatiotemporal organization of atrial fibrillation: a comparison in canine models of structural and electrical atrial remodeling. Am J Physiol Heart Circ Physiol 2006;291:H2911–H2923.
- 30.
Ng J, Kadish AH, Goldberger JJ. Effect of electrogram characteristics on the relationship of dominant frequency to atrial activation rate in atrial fibrillation. Heart Rhythm 2006;3:1295–1305.
