INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

THE INHERENT ELECTROPHYSIOLOGICAL functional characteristics of atrioventricular (AV) nodal (AVN) tissue have been carefully studied in vitro (1, 2, 4, 13, 29, 32). One of the major factors that governs AVN conduction is AVN recovery time (which is defined in this context as the time between when one impulse exits the AVN and the next impulse enters the AVN) (5, 30, 34). AVN conduction dynamics are a function of the interaction of three intrinsic properties of AVN tissue: recovery, facilitation, and fatigue (2, 4, 25, 34). Recovery is the process of reestablishing excitability after excitation (3, 29). Facilitation is the process that produces a short AV interval after a long AV interval that was preceded by a short recovery time (9, 23, 26, 32). Fatigue is the gradual slowing of AVN conduction during rapid repetitive excitation (21, 32).

As the excitation rate increases, it has been shown that recovery, facilitation, and fatigue can interact in a nonlinear fashion to produce increasingly complex AVN conduction dynamics. The idea that AVN dynamics are nonlinear is not new, having been suggested in the early 1900s by Mobitz (24) and incorporated into most AVN conduction models since that time. The nonlinear dynamic mechanistic theory for AVN conduction, which is consistent with the expanding body of evidence supporting the influence of nonlinear dynamics on cardiac electrophysiological behavior (7, 12), is supported by the observation that AVN conduction time can bifurcate (such bifurcations are hallmarks of nonlinear dynamic systems) and settle into a beat-to-beat AVN conduction time alternation known as alternans. Alternans occurs when conduction fatigues beyond a critical value, after which recovery and facilitation interact in a nonlinear fashion to produce alternating long and short intervals. (For a detailed mathematical analysis of this mechanism, see Ref. 32.)

Although alternans has often been attributed clinically to the presence of dual AVN pathways (11, 31, 33, 35), the aforementioned in vitro studies, along with observations of single AVN pathway alternans in humans during AV orthodromic reciprocating tachycardia (ORT) (8, 19, 27, 28), have provided evidence that the inherent electrophysiological properties of a single AVN pathway may also produce alternans. [ORT is a repetitive reentrant arrhythmia in which normal anterograde ventricular excitation via the AVN is followed by reexcitation of the atria via a pathological retrograde accessory ventriculoatrial (VA) pathway.] Although experimental studies have provided invaluable information about the functional nature of complex AVN dynamics in vitro, knowledge of complex in vivo dynamics in which autonomic influences are also present (17) is lacking. Most in vitro studies have investigated nodal conduction dynamics in the absence of autonomic influences [e.g., the preparations for the experiments that produced the aforementioned nonlinear dynamic model of AVN conduction consisted of only the right atrium, the AVN area, and the upper part of the ventricular septum (1, 32)]. Although such studies are well suited for investigating the dynamics of inherent tissue properties, it is not known how the findings of such studies extrapolate to AVN dynamics in the presence of autonomic modulation. Therefore, the main goal of the present study was to systematically investigate the dynamics and mechanisms of complex AVN conduction (specifically, alternans and related rhythms) during ventricular-triggered atrial pacing in humans.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

An atrial pacing protocol was performed after informed written consent was obtained as a supplementary component of routine clinically indicated electrophysiological studies in 37 patients (18 male, 19 female; age 57 ± 16 yr) with normal AVN conduction. (See Table 1 for patient demographics, relevant cardiac characteristics, and results; 18 patients had dual AVN pathway physiology, detected as described below, whereas 19 patients did not have dual AVN pathway physiology.)

The electrophysiological studies used standardized techniques, which included the introduction of multiple percutaneous catheters from the femoral veins to record intracardiac signals from the right atrium, His bundle region, and right ventricle as well as to pace from the two chambers. In all patients, baseline AVN conduction was assessed using single atrial extra stimuli at two basic cycle lengths (generally 600 and 400 ms). Dual pathway physiology was defined on the basis of a conventional AVN conduction discontinuity criterion. Specifically, dual pathway physiology was confirmed if a 50-ms increment in the atrial-His bundle (A-H) interval occurred for a 10-ms decrement in the atrial-atrial (A-A) interval. A typical AVN conduction curve for a patient with dual AVN pathways is shown in Fig. 1.

View larger version (17K): [in this window] [in a new window]   Fig. 1.   Baseline atrioventricular (AV) nodal (AVN) conduction curve for a patient with dual AVN pathways (patient 22). Atrial-His bundle conduction times (A2H2) were measured for single atrial extra stimuli that were delivered at atrial-atrial intervals (A1A2) after the last of 8 consecutive stimuli spaced at basic cycle lengths (A1A1) of 400 and 600 ms. The existence of dual pathways was verified by the presence of a 50-ms increment in A2H2 for an ~10-ms decrement in A1A2 (dotted lines).

During each trial, ORT was simulated via a protocol, called fixed-delay stimulation (Fig. 2), in which the right atrium was stimulated (at time A) at a fixed time VA interval after detection of ventricular activation (at time V). The fixed delay is consistent with the nearly constant retrograde conduction time that accompanies ORT. Because AVN recovery time is equal to VA plus a constant (HV), changes to VA are used to manipulate the recovery time and, therefore, affect AVN conduction dynamics. In studies in which isoproterenol was administered during the clinical evaluation stage, the atrial pacing protocol was not initiated until at least 15 min had passed since the termination of isoproterenol delivery and the heart rate had returned to its baseline value.

View larger version (29K): [in this window] [in a new window]   Fig. 2.   A schematic showing normal conduction from the sinoatrial (SA) node, through the right (RA) and left atria (LA), the AVN, and the right (RV) and left ventricles (LV). In the absence of an abnormal retrograde pathway, orthodromic reciprocating tachycardia can be simulated (loop containing the computer) by fixed-delay stimulation of the RA (at time A) at a ventriculoatrial (VA) interval after detection of ventricular activation (at time V).

A surface electrogram was sampled at 1 kHz by a National Instruments AT-MIO-16E-10 (Austin, TX) data acquisition board in a 266-MHz Intel Pentium II-powered computer running real-time Linux and a custom C++ experiment interface system (6). This system automatically detected R waves in the surface electrogram (denoted as time V) via a threshold-crossing algorithm and then, at an operator-controlled VA interval (with 1-ms resolution) after the detected R wave, outputted a voltage pulse via the AT-MIO-16E-10 to trigger a Bloom DTU215 stimulator (Fischer Imaging; Denver, CO) to stimulate the right atrium (at time A) (Fig. 2).

In in vitro animal AVN preparations, when the VA interval is reduced (simulating faster reentry), the period 1 rhythm fatigues and then destabilizes such that the conduction time through the AVN bifurcates into alternans. Thus, to determine if the human AVN undergoes similar nonlinear dynamic behavior, the nominal VA interval was decreased gradually until an alternation in the AV interval was observed, AVN conduction was blocked, or the minimum VA interval (1 ms) was reached.

Twelve patients underwent a second fixed-delay atrial pacing trial (during the same electrophysiological study, immediately after the first atrial pacing trial) after pharmacological autonomic blockade, which was achieved via intravenous delivery of 0.2 mg/kg propranolol and 0.04 mg/kg atropine (15). The pharmacological blockade trials were used to investigate the relationship between autonomic function and the pacing-induced AVN conduction dynamics.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

AV interval alternans occurred in 18 patients. (Unless specifically noted, the results given are for the trials without the propranolol-atropine autonomic blockade.) There were no significant correlations between the occurrence of alternans and the existence of dual AVN pathways: alternans occurred in 9 of 18 (50%) patients who had dual AVN pathways and in 9 of 19 (47%) patients who did not have dual AVN pathways [P = not significant (NS); significance computed using Fisher's exact test]. Similarly, there were no significant correlations between the occurrence of alternans and the use of antiarrhythmic medications: alternans occurred in 9 of 21 (43%) patients who were taking antiarrhythmic medications and in 9 of 16 (56%) patients who were not taking antiarrhythmic medications (P = NS).

Figure 3 shows the surface and intracardiac electrograms from one trial during simulated ORT pacing. Throughout this segment of the trial, the VA pacing interval was held at 60 ms. The pacing caused a bifurcation (at approximately the time of the fourth beat) in the atrial-His bundle (AH) conduction time (HBE tracing) from period 1 conduction (at AH  120 ms) to a period 2 alternation (between AH  148 and 106 ms by the end of the tracing). The AH interval, which is the best measurement of the AVN conduction time, is quantifiable during posttrial analysis of electrograms, as shown in Fig. 3. However, because automated His bundle activation detection is unreliable, the AV interval was used in this study as a surrogate for AVN conduction. This substitution is based on the assumption that the His bundle-ventricle interval (i.e., ventricular conduction) is constant, an assumption that was verified for each patient during atrial pacing at multiple cycle lengths.

View larger version (35K): [in this window] [in a new window]   Fig. 3.   Surface (leads I, aVF, V1, and V6) and intracardiac [high RA (HRA), His bundle (HBE), and RV apex (RVA)] electrograms from a segment of a fixed-delay atrial pacing trial (patient 32). The two rows of numeric annotations mark the VA (top) and AH (bottom) intervals in milliseconds (H, His-bundle excitation). Note that the onset of atrial activity (A) in HBE occurs later than in HRA, reflecting intra-atrial conduction time. The VA intervals are measured as the time between the electrogram threshold (dotted horizontal line superimposed over V6) crossing and the stimulus delivery. VA was held fixed at 60 ms, which initially produced a relatively constant AVN conduction of AH  120 ms (left). A bifurcation to period 2 alternans occurred at approximately the time of the fourth beat.

Figure 4 shows the time course of alternans over a period of 500 beats in a fixed-delay pacing trial on another patient. This figure demonstrates a gradual fatigue of the AV intervals, followed by a period 2 bifurcation, followed by an extended stage of alternans. The inset in Fig. 4 shows 25 consecutive AV intervals, magnified and connected to demonstrate that the intervals are indeed alternating on a beat-to-beat basis. The gradual onset via a period-doubling bifurcation [as opposed to the rapid (>50 ms) AV conduction time jump typical of dual pathway conduction], as well as the fact that this patient had no evidence of dual AVN pathways, demonstrates that alternans of this type is a function of the nonlinear dynamic properties of a single nodal pathway. In fact, in all 19 alternans-positive patients, alternans developed via a bifurcation (typical of nonlinear dynamic function) instead of a discrete jump (dual pathway function).

View larger version (27K): [in this window] [in a new window]   Fig. 4.   The AV intervals for a fixed-delay pacing trial with no pharmacological autonomic blockade (patient 9). During the shown time segment, VA was held at 2 ms. The AV intervals gradually fatigued and then bifurcated into alternans. The inset shows 25 consecutive AV intervals, magnified and connected to demonstrate that the intervals did, in fact, alternate on a beat-to-beat basis.

The alternans shown in Fig. 4 were stable over a long time period (nearly 500 beats), terminating only upon discontinuation of fixed-delay pacing. Two other patients demonstrated similar stable alternans. However, most alternans-positive patients (15 patients) had alternans that terminated either via spontaneous reversion along a reverse bifurcation to period 1 conduction (6 patients), via AVN conduction block (8 patients), or via a change in pacing rate (1 patient). Each patient typically had multiple runs of alternans. For the 15 patients with spontaneously terminating alternans, the maximum duration of alternans during the trial ranged from 14 to 150 beats (49 ± 37 beats).

An example of alternans that terminated spontaneously via blocked AVN conduction is shown in Fig. 5. In this segment, the AV intervals repeated a characteristic dynamic pattern consisting of a fatigue stage accompanied by a gradual increase in the magnitude of the alternans (i.e., the difference between consecutive AV intervals increased). Eventually, a particularly short AV interval was followed by AVN conduction block, which can be thought of as an infinitely long AV interval. AVN conduction remained blocked for one or more beats, after which the AVN tissue recovered, thereby allowing AVN conduction to resume at approximately the same conduction time as at the beginning of the previous cycle. In this example, alternans reinitiated immediately after conduction block. This might suggest that the alternans rhythm was merely masked, rather than terminated, by the conduction block. Such masking would be characterized by a continuity of the preblock long-short-long-short alternans pattern during the blocked beats and upon alternans reinitiation (i.e., the phase of the preblock alternans would be propagated through the blocked beats to the postblock alternans). However, such phase-locked behavior did not dominate, suggesting that block did, indeed, terminate the alternans dynamic. Similar dynamics were also observed in the in vitro rabbit experiments of Sun, Amellal, Glass, and Billette. (Ref. 32, Fig. 5, p. 82).

View larger version (18K): [in this window] [in a new window]   Fig. 5.   The AV intervals for a fixed-delay pacing trial with no pharmacological autonomic blockade (patient 5). During the shown time segment, VA was held at 100 ms. A repetitive pattern of fatigue, widening alternans, conduction block (in which the AV intervals extend beyond the top of the graph), and resetting of the pattern can be seen.

In some trials, more complex AVN conduction rhythms were seen. One such example is shown in Fig. 6. In this segment, the AV intervals initially alternated, then spontaneously resumed period 1 conduction, then rebifurcated to period 2 alternans, then bifurcated again to period 4 conduction for 16 beats (3,363  n  3,378), and then reverted back to period 2 alternans. This cascade of period-doubling bifurcations is particularly interesting from a nonlinear dynamic perspective, given that chaotic processes often initiate via period-doubling cascades. This was the only trial in which period 4 conduction was observed.

View larger version (22K): [in this window] [in a new window]   Fig. 6.   The AV intervals for a fixed-delay pacing trial with no pharmacological autonomic blockade (patient 22). During the shown time segment, VA was held at 30 ms. Multiple bifurcations (including period 2 to period 4) and reverse bifurcations occurred. The AVN conduction periodicities are annotated [period 1 (P-1), period 2 (P-2), and period 4 (P-4), respectively].

The complex rhythms observed in these experiments were not limited to variations of AVN alternans. In fact, in 16 patients (13 of whom had no alternans), a very different AVN conduction rhythm occurred: visually apparent quasisinusoidal AV interval oscillations. [Three patients had both alternans and oscillations (two were concurrent); six patients had neither.] These approximately periodic oscillations were characterized by a gradual increase in AVN conduction time, followed by a nearly symmetric decrease of AVN conduction time (i.e., the AV intervals gradually shortened at approximately the same rate at which they lengthened during the increasing stage), followed by another increase, followed by another decrease, etc. The AVN conduction oscillations had a mean period (for the 16 patients in whom oscillations occurred) of 55 ± 18 s (0.02 Hz). There were no significant correlations between the occurrence of AV oscillations and the existence of dual AVN pathways: oscillations occurred in 9 of 18 (50%) patients who had dual AVN pathways and in 7 of 19 (37%) patients who did not have dual AVN pathways (P = NS). Similarly, there were no significant correlations between the occurrence of AV oscillations and the use of antiarrhythmic medications: oscillations occurred in 7 of 21 (33%) patients who were taking antiarrhythmic medications and in 9 of 16 (56%) patients who were not taking antiarrhythmic medications (P = NS).

One example of such oscillations is shown in Fig. 7A. In Fig. 7A, the AV intervals show a very clear increasing-decreasing oscillation. (Because the AV oscillations occurred during the initial VA rampdown stage of the trial, the VA intervals are indicated for this trial by annotated arrows. Because the oscillations are not phase locked to the pacing decrements, it is clear that the VA decrements were not the cause of the oscillations; i.e., the oscillations were independent of, and coincidental to, the VA decrements.) Figure 7B shows a segment of another trial in which oscillations occurred during constant VA pacing.

View larger version (33K): [in this window] [in a new window]   Fig. 7.   A: AV intervals for a pacing trial with no pharmacological autonomic blockade (patient 2). The VA intervals are indicated (in ms) by annotated arrows. A quasisinusoidal oscillation (with a period of ~105 beats or 52 s) of increasing and decreasing AVN conduction time occurred, independent of the VA interval changes. B: quasisinusoidal oscillations (with a period of ~145 beats or 66 s) in AV intervals during a fixed-delay pacing trial (VA = 50 ms) with no pharmacological autonomic blockade (patient 27).

Twelve patients underwent a second trial after autonomic blockade. The results of these trials are summarized in Table 2, which shows the occurrences of alternans and oscillations in the trials before and after autonomic blockade. Five patients had alternans both in their pre- and postblockade trials. Three patients had alternans after blockade but not before. Three patients had oscillations before blockade but not after. One patient had neither alternans nor oscillations before or after blockade. Note that alternans was never eliminated by blockade and was actually induced in three cases in which it did not occur before blockade. In contrast, oscillations never occurred after blockade, and preblockade oscillations actually disappeared in three postblockade trials. Thus it appears that autonomic blockade facilitated alternans but inhibited oscillations.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

It is becoming increasingly recognized that many aspects of cardiac electrophysiology are governed by the principles of nonlinear dynamics (7, 12). Experimental evidence (1, 14, 32) suggests that the conduction dynamics of the AVN are no exception. These experiments have shown that the inherent functional characteristics of the nodal tissue (i.e., recovery, facilitation, and fatigue) interact in a nonlinear manner as AVN excitation rate increases and can produce a period-doubling bifurcation (to alternans).

The in vitro experiments (1, 32) that have led to this nonlinear dynamic formulation, along with isolated clinical examples (8, 19, 27, 28), have led to an understanding of the functional behavior of the AVN during rapid excitation. However, it is not known how well such findings extrapolate to the AVN in the presence of autonomic influences, because, until now, such dynamics have not been investigated in a systematic study in vivo.

In this study, we have demonstrated that alternans resulting from rapid atrial pacing (which occurred in 18 of 37 patients) do, in fact, occur in a single AVN pathway. Furthermore, the mechanism of alternans appears to be the same as that observed in previous in vitro rabbit AVN experiments (1, 32): a gradual fatigue stage precedes a period-doubling bifurcation. Thus this study supports the concept that alternans in humans does not require dual pathways and, in fact, can be a nonlinear dynamic single pathway phenomenon. However, it should be noted that, although these results demonstrate that typical dual pathways are not required for alternans, this study does not definitively rule out more subtle dual pathway influences on complex AVN conduction dynamics.

We also observed AVN conduction behavior that has not been reported previously (in vitro or in vivo): quasisinusoidal AV interval oscillations (which occurred in 16 of 37 patients). The occurrence of oscillations and alternans were typically mutually exclusive (with the exception of 3 patients who had alternans and oscillations). The observation of these low-frequency (0.02 Hz) oscillations is consistent with previous studies that have identified a strong low-frequency component in electrocardiogram P-R interval variability (10, 18, 20).

Because autonomic blockade prevented oscillations from occurring in all 12 patients (including 3 patients in which oscillations had occurred before blockade), it appears that the oscillations are autonomically mediated. This is consistent with the absence of such oscillations in in vitro denervated AVN preparations (1, 32). It should be noted that in one previous in vitro trial (14) there was an oscillation superimposed on an alternans rhythm. The oscillations were at a higher frequency (1 cycle per 20 beats, which corresponded to 0.35 Hz) than those observed here; however, it is not clear whether this quantitative discrepancy is significant in such a cross-species comparison. In contrast to all but one trial in the present study, the in vitro oscillations were superimposed on an alternans rhythm. The mechanism for the in vitro oscillatory behavior was not determined, but the oscillations could not have been autonomically mediated because the preparation was denervated. Given the evidence of this study supporting autonomic mediation of oscillations, we suspect, but cannot be certain, that the underlying mechanism for the oscillations in the two studies was different.

The fact that oscillations always started with a fatigue stage (i.e., lengthening of the AV interval) suggests a potential baroreceptor-mediated mechanism. Furthermore, increases in AV interval produce corresponding decreases in cardiac output (assuming that there is an optimal AV interval beyond which cardiac output decreases) and blood pressure. This can disengage arterial baroreceptors, which causes an increase in sympathetic tone as well as parasympathetic withdrawal, which, in turn, causes more rapid AVN conduction. As the AV intervals decrease toward their optimal value, the mechanism reverses, the baroreceptors engage, and conduction again slows. This hypothesis is supported by the fact that the frequency of oscillations is near the low-frequency component of heart rate variability spectral analysis (0.04-0.15 Hz), which has been associated with baroreceptor-mediated blood pressure control (16) [it should be noted, however, that this heart rate variability mechanistic association is not uniformly accepted (22)].

Although autonomic blockade inhibited oscillations, it had the opposite effect on alternans: postblockade alternans occurred in every patient that had preblockade alternans and in three patients who did not have preblockade alternans. Given that the nonlinear dynamic mechanism previously proposed for alternans (32) is only a function of inherent AVN properties (i.e., the proposed nonlinear dynamic mechanism has no autonomic component), it is not surprising that autonomic blockade would not terminate alternans.

The proalternans effect of the autonomic blockade (in 3 patients), the antioscillation effect of the autonomic blockade (in 3 patients) and the fact that alternans and oscillations rarely occurred in the same patient suggest that oscillations in autonomic tone might serve to prevent alternans while at the same time inducing AV oscillations.

In summary, this study has shown that the complex functional properties of single AVN pathways, in conjunction with autonomic tone, are responsible for a range of AVN conduction dynamics during rapid AVN excitation. Alternans, which develops via the type of period-doubling bifurcation typical of nonlinear dynamic systems, is most likely mediated solely by the nonlinear dynamic properties of the AVN tissue. Quasisinusoidal oscillations appear to be mediated by oscillations in autonomic tone (potentially baroreceptor mediated). These findings offer important insight into the integrated functionality of the AVN in vivo, thereby complementing previous in vitro studies while concurrently extending the understanding of AVN function with information only attainable through clinical experiments.