Autonomic Modification of the Atrioventricular Node During Atrial Fibrillation
Role in the Slowing of Ventricular Rate
Background—Postganglionic vagal stimulation (PGVS) by short bursts of subthreshold current evokes release of acetylcholine from myocardial nerve terminals. PGVS applied to the atrioventricular node (AVN) slows nodal conduction. However, little is known about the ability of PGVS to control ventricular rate (VR) during atrial fibrillation (AF).
Methods and Results—To quantify the effects and establish the mechanism of PGVS on the AVN, AF was simulated by random high right atrial pacing in 11 atrial-AVN rabbit heart preparations. Microelectrode recordings of cellular action potentials (APs) were obtained from different AVN regions. Five intensities and 5 modes of PGVS delivery were evaluated. PGVS resulted in cellular hyperpolarization, along with depressed and highly heterogeneous intranodal conduction. Compact nodal AP exhibited decremental amplitude and dV/dt and multiple-hump components, and at high PGVS intensities, a high degree of concealed conduction resulted in a dramatic slowing of the VR. Progressive increase of PGVS intensity and/or rate of delivery showed a significant logarithmic correlation with a decrease in VR (P<0.001). Strong PGVS reduced the mean VR from 234 to 92 bpm (P<0.001). The PGVS effects on the cellular responses and VR during AF were fully reproduced in a model of direct acetylcholine injection into the compact AVN via micropipette.
Conclusions—These studies confirmed that PGVS applied during AF could produce substantial VR slowing because of acetylcholine-induced depression of conduction in the AVN.
Atrial fibrillation (AF) has long been recognized as the most frequent chronic arrhythmia.1 Although the abnormal atrial function during AF and its consequences cannot be underestimated, it is important to note that AF is also associated with an elevated and irregular ventricular rate (VR).2 3 4 5 Mortality due to AF is limited, largely because of the filtering role played by the atrioventricular node (AVN).6 Because of the complexity of AV nodal function even at normal rates, our understanding of AVN conduction during AF remains limited.7 8 Thus, the relative roles of the atrial-AVN approaches versus the region of the compact node in the VR reduction during AF are not yet clear, although current clinical procedures are targeting either or both of these structures.
Slow-pathway modification attempts to achieve slower VR during AF,9 10 11 12 while preserving the AVN. AVN modification has shown some encouraging clinical results, but with inconsistent success rates among investigators.13 14 If this is unsuccessful, complete AVN destruction can be performed, rendering the patient pacemaker-dependent and undesirably altering the normal sequence of ventricular activation.13 15
It is conceptually more desirable to slow the VR during AF without requiring permanent ventricular pacing. One attractive idea is to take advantage of the rich supply of vagal nerves to the AVN.16 If these nerves are stimulated with weak, subthreshold currents, most atrial impulses would be blocked within the AVN. The impact of direct autonomic modulation of AVN conduction during AF has not been explored.
The goals of this study were (1) to compare the contribution of the atrial approaches and the region of the compact AVN in the VR-reducing process during AF and (2) to evaluate a new, nondestructive method to slow the VR during AF by applying postganglionic vagal stimulation (PGVS) directly onto the AVN.
Experiments were performed in vitro on 11 isolated heart preparations obtained from New Zealand rabbits (body weight, 2 to 2.2 kg) after anesthesia with sodium pentobarbital injection (50 mg/kg). After a midsternal incision, the heart was excised, the ventricles and left atrium were discarded, and the final preparation contained right atrial tissues (Figure 1⇓).
All solutions were prepared with deionized and subsequently distilled water. The superfusing solution contained (in mmol/L) NaCl 128.5, KCl 4.7, CaCl2 1.3, MgCl2 1.05, NaHCO3 25, NaH2PO4 1.19, and glucose 11.1. This was saturated with 95% O2/5% CO2 and maintained at 35.5°C at a pH of 7.30 to 7.35. Propranolol (3×10−6 mol/L) was added during experiments in which vagal stimulation or acetylcholine (ACh) injections were used.
Electrodes and Recordings
Bipolar stimulation and recording electrodes were custom-made from 0.20-mm Teflon-insulated platinum-iridium wire (0.5-mm interelectrode distance). The high right atria were stimulated (2-ms pulse, twice diastolic threshold) via optically isolated stimulator units (WPI, A360) and an 8-channel programmable stimulator (AMPI, Master-8). Surface bipolar electrograms were recorded at the crista terminalis (CrT) and the interatrial septal (IAS) inputs of the AVN and at the bundle of His (Figure 1⇑) via high-resistance, differential-input probes and an 8-channel programmable signal conditioner (Axon Instruments, CyberAmp 380). After amplification and filtering (30 to 3000 Hz), signals were displayed on an oscilloscope and simultaneously recorded on magnetic tape (Vetter Digital, 4000A) for later computer analysis with AxoScope (Axon Instruments) and Origin (Microcal) software programs.
Standard microelectrodes (15 to 30 MΩ) and a 2-channel amplifier system (AxoProbe, Axon Instruments) were used to record action potentials (APs) from distal and proximal AVN cells. The midnodal (compact AVN) region was located in the anterior apex of the triangle of Koch, about 2 mm posterior to the earliest His electrogram recording site. Although the anatomic landmarks were consistent, the approximate location of the compact nodal region was further facilitated by recording of characteristic N-type APs, as well as by the strong effects elicited by PGVS and ACh (see later). Because morphological verification was not attempted, the terms “compact node” and “midnode” will be used to indicate the targeted region for positioning of pacing and/or recording electrodes.
The mean spontaneous sinus cycle length was 355±35 ms. The basic paced cycle length was 300 to 320 ms in all preparations. The mean basic conduction time (CrT-His) was 74±11 ms.
Basic-rate pacing was used during the initial 30 to 40 minutes of adaptation, during determination of vagal stimulation parameters (see below), and between repetitive periods of AF.
The term “AF” will be used to describe the randomized, high-rate atrial pacing and the resulting atrial responses. Custom-written software controlled an analog/digital board (MicroStar 3000A/111) that generated random numbers defining the interval between the subsequent driving stimuli (75- to 150-ms range). The same sequence could be generated multiple times, permitting repetitive initiation of the same AF pattern. In separate experiments (not included in this report), we evaluated the AF reproducibility. Each of the mean time intervals (ie, electrogram intervals measured at the posterior CrT AV nodal input [CrT-CrT], the anterior IAS AV nodal input [IAS-IAS], and the bundle of His recording site [H-H], see below in Data Acquisition) was measured in 2 consecutive trials performed from the same pacing site. Differences between mean interval measurements in trials 1 and 2 did not exceed 3 ms. The intraclass correlation coefficients for the mean intervals ranged from 0.97 to 0.99, confirming the high degree of reproducibility.
In 5 preparations, controlled cooling was applied to the posterior approach to the AVN, mimicking the clinical slow-pathway modification. A thermoelectric cooling probe (3.5-mm2 tip, Novoste Corp) was used to cool discrete areas of perinodal tissue.
Postganglionic Vagal Stimulation
The PGVS technique was based on the previously demonstrated high density of vagal nerve terminals in the compact area of the AVN.16 Importantly, the current threshold needed for nerve excitation (ACh release) was lower than the threshold of myocardial excitation.17 Therefore, subthreshold bursts elicited vagal effects on AVN without impeding atrial or ventricular (His) activation.
Each PGVS burst consisted of 16 rectangular current impulses (2-ms duration, 4-ms interimpulse interval). These 3 parameters were kept constant, resulting in a 92-ms burst.
Two other parameters were varied. The first parameter, timing of the PGVS delivery, was either nonsynchronized or triggered by the His electrogram. In the latter case, PGVS bursts could be initiated either by each trigger or by alternating triggers. For nonsynchronized delivery, PGVS bursts were generated at a constant rate of 1, 2, or 4 Hz.
The second parameter, current intensity, had 5 values expressed as multiples of a threshold: ×0.5, ×0.75, ×1, ×2, and ×3. PGVS threshold was defined as the lowest intensity of a single, synchronized burst delivered in the midnodal region during pacing at a basic cycle length of 300 ms that produced complete AVN block (see Figure 4C⇓ in the Results section). Thus, a total of 25 PGVS combinations (5 timings×5 intensities) were studied in random order in each preparation.
We compared PGVS with the direct effects of acetylcholine (ACh) by injecting ACh into the compact region of the AVN with a 10-μm-tip glass micropipette filled with 10 mmol/L ACh connected to a pressure-impulse Pico-Pump (WPI). The pipette was impaled within 100 μm of the PGVS electrode. Using microscopic control and the dial of the micromanipulator, we estimated the tip of the injector to be 100 to 300 μm below the endocardial surface. Each picopulse (duration, 50 or 100 ms) injected a droplet of ≈10−4 to 10−3 μL of ACh. This technique mimicked the discrete release of ACh during PGVS and was evaluated during both paced basic rates and AF. In 3 preparations, ACh picopulses were used along with simultaneous recording of the AVN cellular AP. To ensure that the picoinjection effects were due to ACh release rather than to nonspecific artifacts, separate experiments confirmed that when the pipette was empty or filled with Tyrode’s solution containing low concentrations of ACh (<10−6 mol/L), changes in AVN conduction were negligible, eg, the basic conduction time at a cycle length of 300 ms and the mean H-H interval during AF typically remained within 5% of control.
Data Acquisition and Presentation
Electrograms from the CrT and IAS inputs and the bundle of His were sampled at 1 kHz. Conduction times as well as consecutive CrT-CrT, IAS-IAS, and H-H intervals were determined. Although AF episodes were as long as several minutes, the above measurements were made on a standardized number (100) of H-H intervals.
For simplicity of presentation, all analog records are organized in the following manner. The CrT, IAS, AP, and bundle of His (His) electrograms are displayed from top to bottom. In addition, the consecutive CrT-CrT, IAS-IAS, and H-H intervals were plotted as Lorenz plots. In this format, the abscissa of a given data point represents the value of the nth interval and the ordinate represents the next, nth+1, interval. This facilitates visualization of the range of intervals, as well as the degree of interval dispersion. Thus, a regular rhythm would translate into a single-point Lorenz plot, whereas irregular rates result in dispersed scattergrams.
Coupling intervals (CrT-CrT, IAS-IAS, and H-H) were expressed as mean±SD. H-H distribution data from the cooling experiments were compared with control by the nonparametric paired Wilcoxon test. H-H distributions obtained during AF were compared with control by a 5-level repeated-measurements within-factor analysis, ie, for every mode of PGVS delivery, the effects of 5 different intensities were evaluated. Linear, polynomial, and logarithmic analyses of correlations were performed. A value of P<0.05 was considered statistically significant.
Spatial Heterogeneity of AVN Conduction During AF: Cooling Modification of the Atrial-AVN Approaches and the Region of the Compact Node
Random, high-rate atrial stimulation resulted in chaotic activation of the posterior (CrT) and anterior (IAS) AVN inputs. Concomitant microelectrode recordings revealed that disorganized and decremental APs in the compact AVN region were the major substrate leading to blocks, ie, “filtering” of the atrial impulses. Figure 2⇓ represents traces obtained in the same heart during AF with microelectrodes impaled in different locations. Recorded AF episodes were typically much longer than the 5- to 10-second excerpts shown. Each trace starts with several beats preceding initiation of AF.
The cell in Figure 2A⇑ was impaled in the anterior atrial approach just above the tendon of Todaro, and APAN (2B) was recorded from an AN cell located inferior to the tendon of Todaro. In both regions, numerous APs were inscribed from more positive membrane potentials and had smaller amplitudes and shorter durations than during slow, regular rates. Electrotonic humps were rarely observed. These findings were typical for cells in the atrial-AVN approaches and suggested a reduction of the proximal driving force.
In contrast, rate-induced changes in the compact AVN region were more complex (Figure 2C⇑). APs with multiple humps, reduced and variable amplitude and duration, low dV/dt of the upstroke, and frequent electrotonic depolarizations (arrows) were consistently observed. Although the mechanisms underlying these observations remain unclear, factors such as delayed refractoriness of the N cells that substantially outlasts the repolarization phase of their AP,18 and the irregularity of the atrial input, may be responsible.
The distal AVN was activated in a simpler manner (not shown). The rate was slower, and a steady 1:1 ratio was observed between the distal AP and the His electrogram. Only small AP amplitude variations were noticeable in the NH region, and the AP duration changes paralleled the duration of the diastolic coupling intervals.
The above observations strongly suggest that the filtering of atrial impulses during AF started at the atrial approaches but was most pronounced in the compact AVN region. We performed controlled and reversible cooling (to 18°C) of the posterior AVN approach during AF in 5 preparations. The cooling probe was first placed between the ostium of the coronary sinus and the tricuspid valve ≈3 mm away from the compact AVN, similar to the lead location during radiofrequency ablation procedures, and then moved anteriorly. As demonstrated in Figure 3⇓, during AF, posterior-approach cooling (3A) produced only small slowing of cellular rates and VRs. Moving the cooling probe close to the compact nodal region (3B) produced dramatic effects. Multiple subthreshold depolarizations (stars) not followed by His electrograms indicated only partial penetration of the AVN. This situation resembled the high degree of concealed conduction during AF.19 Similar results were consistently observed in all preparations in which cooling was applied. The degree of ventricular slowing was proportional to the degree of nodal cooling. Thus, in conditions of random, high-rate atrial bombardment, direct depression of conduction in the compact AVN appears to be more effective in slowing VR than modification of atrial engagements of the AVN.
Effects of PGVS During AF
PGVS effects were highly spatially sensitive. As in the above-described observations during cooling, the strongest depression of AVN conduction was observed when vagal stimulation was applied directly to the compact node. ACh released during PGVS induced characteristic hyperpolarization associated with depressed amplitude and duration of the AP (Figure 4⇓). These effects were dependent on both the intensity and the phase (ie, the timing versus the CrT electrogram) of the vagal burst.20 Thus, an early PGVS (phase 0 ms, 4A) had the strongest effect on the beat immediately after the burst, whereas later PGVS (phase 220 ms, 4B) prolonged conduction of the subsequent beat to a greater degree. A burst with proper timing could exert a strong effect at lower intensity (phase 73 ms, 4C).
The above considerations were applicable for slow, regular heart rates. During AF, however, synchronization of a PGVS burst with a random atrial signal would not be warranted. Therefore, in subsequent studies, vagal bursts were synchronized with His electrograms or nonsynchronized at a predetermined frequency. In Figure 5⇓, PGVS was delivered at 2 Hz and 1× threshold (see Methods). Mean VR slowed from 235 bpm (mean coupling interval, 212 ms) to 184 bpm (mean coupling interval, 325 ms). The microelectrode was impaled in the compact nodal region. Note the abrupt change in the cellular response immediately after the start of PGVS. The depressed amplitude and dV/dt of the AP during AF were further reduced, a direct vagal effect on the N cells.20 Multiple electrotonic responses (one marked by arrow) suggested that either the cellular excitability, the driving force, or both were insufficient to permit regenerative depolarization. Double-humped APs (star) were frequently present. Although different mechanisms may be responsible for this phenomenon, likely explanations include the arrival of multiple, fragmented, anterograde wave fronts in the same location or reentrant activation of the impaled fiber.
Quantitative Assessment of Vagally Induced Slowing of VR During AF
A consistent finding in each of the preparations studied was the significant VR slowing during AF by PGVS applied directly to the AVN. The Lorenz plots (Figure 6⇓) clearly illustrate this finding. The scattergrams represent the corresponding intervals during control AF (6A, 6B, and 6C) and during PGVS (6D, 6E, and 6F). Note that PGVS had no effect on the atrial events: the mean value and distribution of both CrT-CrT (6A and 6D) and IAS-IAS (6B and 6E) remained unchanged. Thus, the vagal effect was confined only to the AVN region, although specific involvement of the transitional and compact nodal regions cannot be delineated. A significant prolongation of the H-H intervals can be seen (6C and 6F), with mean H-H interval increasing almost 2-fold versus control (515 versus 272 ms).
The Table⇓ summarized data obtained from all hearts. Mean H-H intervals are shown during control-AF and PGVS. The latter was delivered in 5 modes and with 5 intensities (see Methods). Figure 7⇓ shows that higher-intensity and more frequent bursts were associated with increased prolongation of the H-H intervals (correlation coefficients were in the range 0.94 to 0.99, P<0.001).
To confirm that PGVS effects were induced by ACh release, we compared them with those of minute amounts of ACh injected directly into the compact node (see Methods). Figure 8⇓ represents excerpts from a long AF episode with delivery of ACh every 500 ms (2 Hz). A comparison of Figure 8⇓ with Figure 5⇑ demonstrates the qualitative similarity between the effects of PGVS and ACh. Note the progressively decremental effect on the AP amplitude (8A through 8C) as well as the presence of multiple local electrotonic responses and multihumped APs.
Hundreds of ACh pulses could be delivered in this manner, without any apparent fading of the effect. In contrast, prolonged PGVS was associated with a progressive attenuation and even a reversal of the effect on the VR. In these cases, we first initiated PGVS and then, when fading became apparent, we replaced PGVS with ACh pulse injection. Figure 9⇓ demonstrates that a long-lasting prolongation of the H-H intervals was immediately restored with the delivery of ACh.
Cholinergic Depression of Conduction and Multiple Wave Fronts
Direct intranodal ACh injection provided an insight into the nature of the multihumped AVN AP observed during AF and PGVS (Figures 2⇑, 5⇑, and 8⇑). Figure 10⇓ illustrates 1 preparation in which ACh pulses were injected during constant pacing (300 ms) at IAS. Panels A through C represent 4-second excerpts from a continuous recording. The AP was recorded from a fiber in the compact node. During ACh pulses (Figure 10A⇓, 10B⇓, and 10C⇓), progressive hyperpolarization (the dashed horizontal line marks the level of the control diastolic membrane potential) accompanied a splitting of the AP into 2 components (10A and 10B), along with the gradual depression of the AVN conduction until 2:1 block developed (10C). Note that AVN block was correlated with the missing second component of the AP (arrows), although the first one, although very low in amplitude, was always present.
The 2 AP components evolved in a characteristic way. Figure 10D⇑ shows superimposed traces obtained during control and 22 seconds after the start of ACh injection (the thicker AP and His signals). The timing of the first component (star) remained exactly the same as the control AP, although its amplitude decreased rapidly during ACh injections and soon became just a local electrotonus. The second component (dot) appeared later in the subsequent beats in parallel with the prolongation of the AVN delay (the IAS-His interval), from 88 ms in control to 146 ms. In addition, the amplitude of the second component gradually depressed and began to alternate between higher and lower amplitudes (Figure 10B⇑, arrows). Accordingly, the AVN conduction delay was alternating between 140 and 146 ms in consecutive conducted beats (10D, arrows).
These AP changes accompanying ACh injections were reproduced many times, with impalements in different fibers from the compact nodal region. The splitting of the AP into 2 components was a consistent finding. When ACh injection was terminated, the return toward control mirrored the process just described. Initially, the second component began to increase in amplitude (Figure 11A⇓, dashed arrow), and its amplitude alternations gradually disappeared (11B, arrows). The strongly depressed first component recovered more slowly (11C, dashed arrow), during which time the ACh-induced membrane hyperpolarization gradually disappeared. Finally, a fusion of the 2 components into a single AP (11D, arrow) completed the recovery.
Different AP dissociation dynamics were observed outside the compact nodal region. Figure 12⇓ illustrates typical traces obtained from an AN (transitional) fiber located just inferior to the tendon of Todaro. During ACh infusion, the AP again dissociated into 2 components (12B and 12C), as in Figure 10⇑. However, the process developed differently. First, the ACh-induced hyperpolarization was less pronounced than in the N fibers. This can be explained by the facts that the fiber was farther from the ACh-injecting pipette and that the ACh effect was different on atrial-cell–like cells. Second, the first component maintained ≈75% of the amplitude of the original (12A) but progressively shortened. Third, the secondary component was first detected as a late hump (12B, arrow) and eventually was inscribed as a very-low-amplitude depolarization that was present only when there was a successful propagation to the bundle of His (12C, stars).
These observations strongly suggested the presence of 2 wave fronts (Figure 12⇑, inset). According to this model, the earlier wave front should be associated with the anterior septal activation (the putative “fast pathway”) and was dominant (detectable, in both the transitional and compact regions) before the infusion of ACh. The latter led to block of the earlier wave front at the compact nodal region entrance, so that its “signature” on the level of the compact node deteriorated to small electrotonic humps (Figure 10⇑) but could still be detected outside the compact node (Figure 12⇑). Successful conduction to the bundle of His was then maintained by a second, delayed wave front that was apparently more resilient to the ACh effect. The signatures of this now predominant wave front were the second components in the recorded APs and the successful activation of the bundle of His during the ACh infusion. In addition, a retrograde turnaround of the delayed wave front toward the septum cannot be ruled out (Figure 12B⇑ and 12C⇑), although it did not result in atrial echo-beats.
Reversible autonomic modulation of the AVN during AF substantially slowed VR. This effect probably resulted from ACh-induced hyperpolarization of the compact nodal region and was manifest in both depressed conduction and increased inhomogeneity with multiple wave fronts. Rich vagal innervation of the compact node provides the unique opportunity to use subthreshold PGVS during AF as an effective tool for slowing VR while preserving the integrity of all anatomic structures and the anterograde pattern of conduction.
Modification of the Atrial-AVN Approaches Versus Depression of the Compact Node
Localized cooling of the posterior approaches, a procedure similar to posteroseptal radiofrequency ablation, had limited effect on the VR during AF (Figure 3A⇑). In contrast, when the cooling was applied close to the compact node (Figure 3B⇑), the H-H intervals prolonged dramatically. The lower efficiency of the VR control obtained with modification of one atrial approach can be explained by the persistent high-rate bombardment of the compact node via the remaining atrial approaches. It may be speculated that only an arch of ablation along the boundaries of the triangle of Koch, rather than only localized posterior modification, may result in a successful VR slowing during AF. Alternatively, direct depression of the compact AVN by cooling invariably resulted in a dramatic slowing of the VR without alteration of the pattern of atrial bombardment.
PGVS as a Tool for VR Slowing During AF
Subthreshold PGVS produces transient hyperpolarization and subsequent depression of AVN conduction.17 20 21 As this study has demonstrated, PGVS applied directly over the AVN during AF resulted in a substantial slowing of the VR (Figures 5⇑ and 6⇑). This effect was not associated with a modification of the rate of atrial bombardment (Figure 6⇑) but rather was confined to the AVN region. We speculate that PGVS worked by increasing the inhomogeneity of conduction (producing fragmented and desynchronized wave fronts) and by decreasing AP amplitude. This would result in a decremental proximal driving force associated with multiple local electrotonic events and an overall high rate of concealed conduction. However, this speculation involves very complex mechanisms that are not inclusive and/or directly supported by the experimental results.
The effects of PGVS on AVN conduction have been shown to be dependent on the phase of the bursts in the cardiac cycle.21 Although phasic effects were probably present during AF as well, the random arrival of the atrial impulses made it difficult to evaluate their role. We investigated several different synchronized and nonsynchronized modes of application of PGVS (Table 1⇑) and found that the VR slowing was proportional to both the intensity of the current and the frequency of the PGVS bursts.
Fading of the PGVS Effects
The hyperpolarizing and depressive effects of ACh on the AVN transmembrane potentials during regular slow atrial rhythms are well known from classic works.22 23 Later studies24 demonstrated that prolonged, high-frequency stimulation of the intramural vagal nerves, but not exposure to ACh, was associated with fading effects on the sinus node. In this study, persistent application of strong PGVS bursts during AF in the isolated rabbit heart preparation saturated vagal effects on the AVN. Because direct ACh pulses reversed the fading effect over prolonged periods of time (Figures 8⇑ and 9⇑), desensitization of the muscarinic receptors25 was an unlikely underlying mechanism. An alternative possibility is the release of neuropeptide Y from sympathetic terminals. Neuropeptide Y attenuates the global vagal effects26 but is evoked by strong and long neural stimulation, in contrast to the brief and subthreshold PGVS used in our studies. Thus, the most likely mechanism for the fading phenomenon is the limited pool of choline in our in vitro experiments. Because choline uptake in the synaptic cleft is a slow process and because this is the main source of choline in vitro,27 repetitive PGVS may have resulted in more released than synthesized ACh per second. In vivo, where normally an unlimited pool of choline is provided by the plasma,28 the synthesis of ACh during repetitive PGVS would be less hampered.
Are Dual Pathways Involved in Conduction Through the Depressed AVN?
The dynamic development of 2 distinct components of the AVN AP during direct ACh injection (Figures 8⇑ and 10⇑ through 12) and during PGVS (Figure 5⇑) is compatible with the presence of fast and slow wave fronts with different ACh sensitivities. Although unproven, a reasonable hypothesis can be considered. The fast wave front may have produced the earlier component of the cellular responses (Figures 10 through 12⇑⇑⇑) and was readily depressed by ACh, whereas the later slow wave front was more resilient. This later wave front was not detectable in the microelectrode recordings during control, because it arrived when the area of impalement was already depolarized by the fast wave front. However, the presence of the slow wave front was revealed during the ACh injection, and it was this delayed wave front that maintained the conduction to the bundle of His.
Limitations of the Study
Random, high-rate atrial pacing was used in these studies to mimic AF. The major advantage of this model was the possibility to repeatedly reproduce nearly identical patterns of random atrial activation. However, further in vivo studies are needed to evaluate the autonomic modulation of the AVN during spontaneous AF. The reported observations must be extrapolated cautiously to humans, even though direct parasympathetic effects on the AVN in humans can be evoked by PGVS29 30 or stimulation of 1 of the fat pads.31 In addition, recent reports indicated that the autonomic control of the AVN is preserved after radiofrequency ablation of AVN reentrant tachycardia32 and that electrical stimulation of parasympathetic fibers near the posteroseptal and anteroseptal areas could induce a negative dromotropic effect.33 These observations, along with those from our study, suggest the need for further systematic evaluation of the effects of long-lasting PGVS on the human AVN during AF.
This research was supported in part by grant 9807701 from the American Heart Association (Northeast Ohio Affiliate).
- Received September 18, 1998.
- Revision received February 10, 1999.
- Accepted February 23, 1999.
- Copyright © 1999 by American Heart Association
Lewis T. The Mechanism of the Heart Beat. London, UK: Shaw & Sons; 1911:194–234.
Bootsma BK, Hoelsen AJ, Strackee J, Meijler FL. Analysis of R-R intervals in patients with atrial fibrillation at rest and during exercise. Circulation. 1970;41:783–794.
Rawles JM, Rowland E. Is the pulse in atrial fibrillation irregularly irregular? Br Heart J. 1986;56:4–11.
Meijler FL, van der Tweel I, Herbschleb JN, Hauer RN, Robles de Medina EO. Role of atrial fibrillation and atrioventricular conduction (including Wolff-Parkinson-White syndrome) in sudden death. J Am Coll Cardiol. 1985;5:17B–22B.
Markowitz SM, Stein KM, Lerman BB. Mechanism of ventricular rate control after radiofrequency modification of atrioventricular conduction in patients with atrial fibrillation. Circulation. 1996;94:2856–2864.
Murgatroyd FD, Camm AJ. Atrial fibrillation: the last challenge in interventional electrophysiology. Br Heart J. 1995;74:209–211.
Imaizumi S, Mazgalev T, Dreifus LS, Michelson EL, Miyagawa A, Bharati S, Lev M. Morphological and electrophysiological correlates of atrioventricular nodal response to increased vagal activity. Circulation. 1990;82:951–964.
Mazgalev T, Dreifus LS, Michelson EL, Pelleg A. Effect of postganglionic vagal stimulation on the organization of atrioventricular nodal conduction in isolated rabbit heart tissue. Circulation. 1986;74:869–880.
Merideth J, Mendez C, Mueller WJ, Moe GK. Electrical excitability of atrioventricular nodal cells. Circ Res. 1968;23:69–85.
Mazgalev T, Dreifus LS, Michelson EL, Pelleg A. Vagally induced hyperpolarization in atrioventricular node. Am J Physiol. 1986;251:H631–H643.
Mazgalev T, Dreifus LS, Michelson EL, Pelleg A, Price R. Phasic effects of postganglionic vagal stimulation on atrioventricular nodal conduction. Am J Physiol. 1986;251:H619–H630.
Cranefield PF, Hoffman BF, Paes de Carvalho A. Effects of acetylcholine on single fibers of the atrioventricular node. Circ Res. 1959;28:199–219.
Paes de Carvalho A, Hoffman BF, Paula de Carvalho M. Two components of the cardiac action potential, I: voltage-time course and the effect of acetylcholine on atrial and nodal cells of the rabbit heart. J Gen Physiol. 1969;54:607–635.
Salata JJ, Jalife J. “Fade” of hyperpolarizing responses to vagal stimulation at the sinoatrial and atrioventricular nodes of the rabbit heart. Circ Res. 1985;56:718–727.
Warner MR, Levy MN. Sinus and atrioventricular nodal distribution of sympathetic fibers that contain neuropeptide Y. Circ Res. 1990;67:713–721.
Blusztajn JK, Wurtman RJ. Choline and cholinergic neurons. Science. 1983;221:614–620.
Keim S, Mazgalev T, Kinn R, Tchou P. Non captured burst stimulation near the AV node: evidence for vagal stimulation as the mechanism for modulation of AV conduction in humans. Pacing Clin Electrophysiol. 1992;15:519. Abstract.
Keim S, Mazgalev T, Tchou P. Physiologic effects of subthreshold burst stimulation on the human AV node. Pacing Clin Electrophysiol. 1993;16:158. Abstract.
Quan KJ, Lee JH, Mackall JA, van Hare GF, Biblo LA, Carlson MD. Identification and characterization of selective vagal innervation to the atrioventricular node and low right atrium in humans. Circulation. 1997;96(suppl I):I-359. Abstract.