Interactions between peripheral conduction system and myocardial wave fronts control the ventricular endocardial activation sequence. To assess those interactions during sinus and paced ventricular beats, we recorded unipolar electrograms from 528 electrodes spaced 0.5 mm apart and placed over most of the perfused rabbit right ventricular free wall endocardium. Left ventricular contributions to electrograms were eliminated by cryoablating that tissue. Electrograms were systematically processed to identify fast (P) deflections separated by >2 ms from slow (V) deflections to measure P-V latencies. By using this criterion during sinus mapping (n = 5), we found P deflections in 22% of electrograms. They preceded V deflections at 91% of sites. Peripheral conduction system wave fronts preceded myocardial wave fronts by an overall P-V latency magnitude that measured 6.7 ± 3.9 ms. During endocardial pacing (n = 8) at 500 ms cycle length, P deflections were identified on 15% of electrodes and preceded V deflections at only 38% of sites, and wave fronts were separated by a P-V latency magnitude of 5.6 ± 2.3 ms. The findings were independent of apical, basal, or septal drive site. Modest changes in P-V latency accompanied cycle length accommodation to 125-ms pacing (6.8 ± 2.6 ms), although more pronounced separation between wave fronts followed premature stimulation (11.7 ± 10.4 ms). These results suggested peripheral conduction system and myocardial wave fronts became functionally more dissociated after premature stimulation. Furthermore, our analysis of the first ectopic beats that followed 12 of 24 premature stimuli revealed comparable separation between wave fronts (10.7 ± 5.5 ms), suggesting the dissociation observed during the premature cycles persisted during the initiating cycles of the resulting arrhythmias.
- extracellular mapping
- premature stimulation
- cycle length accommodation
- specialized conduction system
an understanding of electrical interactions between the peripheral conduction system and ventricular myocardium is important because those interactions dictate the endocardial activation sequence. Most investigations to date have focused on interactions across the discrete Purkinje-ventricular junction (PVJ) in superfused canine papillary muscles. This preparation is attractive because there are relatively few PVJs. Furthermore, PVJs are confined to the base of the muscle, which allows antegrade (Purkinje-to-ventricular) conduction to be established by stimulating Purkinje strands attached to the muscle and retrograde (ventricular-to-Purkinje) conduction to be established by stimulating the apex of the muscle. By using this preparation, investigators (22, 26, 27, 32, 35, 40) have demonstrated that PVJ conduction is highly discontinuous because the sparse electrical coupling between Purkinje and ventricular tissue types forms a resistive barrier (35) and that retrograde PVJ conduction is preferential to antegrade PVJ conduction because of the impedance mismatch between tissue types.
These two characteristics ensure the peripheral conduction system dictates the endocardial activation sequence during sinus rhythm and ventricular drives in the canine ventricle, where the distribution of PVJs is relatively coarse compared with other mammalian preparations because of the size of the heart. Long antegrade PVJ conduction delays allow peripheral conduction system wave fronts to remain ahead of myocardial wave fronts during sinus beats. For example, Nagao et al. (31) paced right bundle branches of superfused canine right ventricular (RV) free wall-septum preparations and showed local Purkinje deflections that consistently preceded ventricular deflections throughout the free wall. By using similar preparations, Rawling et al. (35) showed that coarse PVJ distribution was advantageous in this regard because it limited the electrical load imposed by the myocardium on the peripheral conduction system, which allowed high Purkinje conduction velocities to be maintained. During ventricular drives, short retrograde PVJ conduction delays prevent myocardial wave fronts from moving ahead of peripheral conduction system wave fronts and the coarse PVJ distribution allows peripheral conduction system wave fronts to move ahead of myocardial wave fronts once Purkinje cells excite. For example, Ben-Haim et al. (5) found consistent activation sequences with endocardial or atrial stimulation of in situ canine hearts. Pollard et al. (33) frequently saw islands of early activity, suggesting antegrade PVJ conduction in close proximity to endocardial pacing sites in isolated canine hearts. Similarly, Arisi et al. (1) demonstrated independence of epicardial breakthrough sites during epicardial pacing of in situ canine hearts, consistent with the emergence of epicardial wave fronts after transmural expansion from fixed antegrade PVJ conduction sites.
We undertook the present study to assess interactions between peripheral conduction system and myocardial wave fronts in preparations with more densely distributed PVJs. In the rabbit RV free wall, the peripheral conduction system forms a web of Purkinje strands that cover anastomosing trabeculae, which in turn cover the compact myocardium. This anatomic arrangement is consistent with a representation for the peripheral conduction system as a sheet of Purkinje myocytes coupled strongly to the ventricular mass with relatively high PVJ density compared with canine preparations and with papillary muscle preparations from other species. The size of this preparation allowed mapping from the majority of viable tissue during activation sequences resulting from sinus rhythm and endocardial drives. Consistent with the approach used in canine papillary muscle studies, rapid (P) and slow (V) deflections were identified in the electrograms to assess local Purkinje and myocardial activation, respectively. P-V latencies then quantified the extent to which wave fronts separated. We found peripheral conduction system wave fronts that preceded myocardial wave fronts during sinus rhythm but lagged behind those wave fronts during endocardial drives. Separation between wave fronts was comparable during sinus beats and endocardial drives with nominal cycle lengths or with rapid pacing. During premature stimulation and the ectopic cycles that followed many premature responses, myocardial wave fronts also preceded peripheral conduction system wave fronts, suggesting functional uncoupling between the myocardium and the peripheral conduction system despite the dense distribution of PVJs.
MATERIALS AND METHODS
We studied hearts from 13 New Zealand White rabbits (3–5 kg) anesthetized with intramuscular ketamine (44 mg/kg) and xylazine (20 mg/kg), followed by intraveneous pentobarbital sodium (2.5 ml) and heparin (2.0 ml). The hearts were rapidly excised after a medial sternotomy and retrogradely perfused through the aorta with recirculating, oxygenated (95% O2-5% CO2) normal solution containing (in mmol/l) 126 NaCl, 22 dextrose, 1 MgCl2, 4.4 KCl, 20 taurine, 5 creatine, 5 sodium pyruvate, 1 NaH2PO4, 30.1 NaHCO3, and 1.08 CaCl2 (pH 7.3–7.4). We maintained temperature at 37° ± 1°C and flow rate at 45–50 ml/min. Each heart settled for 10 min before dissection for endocardial electrode placement and protocol initiation.
We mapped responses to endocardial drives in 8 of 13 experiments. In these hearts, the right atria (RA) and left (LA) atria were removed, and the right coronary artery was tied off just proximal to the posterior descending artery to ensure continuous perfusion of the RV free wall. A base-to-apex cut along the posterior descending artery distal to the suture separated the free wall from the septum on one side of the ventricle, forming an RV flap. Chordae tendinae were severed to allow papillary muscles to contract freely. The perfused heart lay horizontally and the RV flap was pinned in place at basal and apical ends to a rubber pad with the endocardial side facing up. Figure1 A shows a schematic of the preparation, including the location for the plaque on the free wall adjacent to the septum and positions for all stimulating electrodes. Figure 1 B shows the plaque, which contained 528 Ag/AgCl tips (102-μm diameter), spaced 0.5-mm apart.
Before placing the plaque on the preparation with a micromanipulator, we cryoablated the left ventricular (LV) free wall, septum, and apex by administering liquid nitrogen for ∼2 min via a metal cryoprobe placed inside the LV. This procedure eliminated LV contributions to the electrograms and controlled the amount of viable tissue outside the mapped region while facilitating collateral perfusion of the RV. A rubber nipple was placed on the end of the probe to prolong the time between epicardial breakthrough of the ablation lesion on the LV free wall and its apical migration to the RV. During cryoablation, the migration of the lesion was carefully monitored to ensure that apical and medial portions of the RV septum remained viable for experimental study, although the basal septum was consistently ablated to destroy the His bundle. Electrograms were recorded with the reference electrode located inside the cryoablated LV. Each signal was amplified (×50), band-pass filtered (0.5–1,000 Hz), and digitized at 2,000 samples/s with 14-bit resolution (42).
Stimulating electrodes were located outside the plaque region on1) the septum, just inferior to the anterior papillary muscle, 2) the apex, away from the septum, and 3) the base, away from the RV outflow tract. Hearts were driven with the use of 1.2–1.5× threshold strength, constant-current square pulses of 2 ms duration, applied by a Bloom Electrophysiology (Fischer Imaging; Denver, CO) stimulator from bipolar electrodes with 1-mm tip separation. Preparations were initially driven at 500-ms cycle length from sites 1–3. The cycle length was then reduced to 125 ms, and the measurements were repeated. To ensure cycle length accommodation to the faster rate, we analyzed signals recorded at least 40 beats after the reduction and after each change in drive site. Cycle length was then increased back to 500 ms to acquire baseline measurements in advance of premature stimulation. Premature stimulation intervals started at 100 ms and were incremented in 5-ms steps until global capture. In this way, we were able to analyze 24 activation sequences to quantify responses to baseline drive, cycle length accommodation, and premature stimulation.
We mapped sinus beats in 5 of 13 experiments. The dissection steps were modified so that the RA was left on the heart when the LA was removed to allow insertion of the cryoprobe into the LV. Cryoablation was performed with a rubber sleeve placed over the probe that included a 5-mm-wide slit along the shaft of the probe. Before being ablated, the slit was directed toward the medial LV free wall to focus the epicardial breakthrough of the lesion distant from the septum. This arrangement prevented migration of the lesion onto the septal crest, where the His bundle courses before its proximal connection to the right bundle branch, which ensured that this tissue was preserved.
All of the electrograms were processed to identify P and V deflections by using the steps shown in Fig. 2. First, we differentiated each raw signal, which allowed straightforward identification of the rapid P deflection. Derivatives were thresholded to −2 V/s for automatic identification and then reviewed manually to ensure deflections with characteristics that clearly indicated peripheral conduction system activation were included in the analysis. P deflections were not identified on all electrograms. Second, we digitally filtered each raw signal by using a ninth-order Chebychev Type II low-pass filter (MATLAB, The Mathworks; Natick, MA) with a stop-band edge frequency of 100 Hz and a stop-band ripple of 25 dB. Because filtering removed the P deflection, it was then straightforward to identify the V deflection automatically as the time of the minimum derivative (37). Each derivative was thresholded to −0.3 V/s for determination. Finally, we measured the latency between the P and V deflections when both were identified. For P-V latency below 2 ms, the electrogram was excluded from analysis because it was impractical to distinguish the V deflection in the filtered electrogram from the P deflection in the raw signal. P-V latencies above 2 ms were recorded. When P deflections preceded V deflections, P-V latencies were positive, and sites were termed “P-led,” whereas “V-led” sites where the V deflections preceded P deflections had negative P-V latencies. Mean P-V latency magnitude was taken as a global measure for the separation between peripheral conduction system and myocardial wave fronts for each activation sequence. Differences in mean P-V latency magnitude between activation sequences were quantified by using paired Student's t-tests.
Partial validation of this processing scheme is presented in Fig.3, which shows raw and digitally filtered electrograms recorded before and after dilute Lugol mixed with normal solution (LNT; in a 1:9 ratio) application in one experiment. Because Lugol binds to the rich glycogen stores in Purkinje cells, it has been used extensively as a stain to visualize the specialized conduction system. It has also been used in electrical mapping studies (11,15, 28, 33) with isolated or in situ canine hearts for selective inhibition of the conduction system. At the end of our initial experiments, the plaque was raised, LNT was applied, and mapping was attempted again. In most experiments, signal degradation suggesting tissue damage to overlying myocardium accompanied inhibition of the peripheral conduction system. In one experiment, however, LNT application eliminated rapid P deflections at many sites where these deflections were evident. That elimination established a closer match between the raw and filtered electrograms, as shown in the post-LNT traces from Fig. 3 A. In addition, the post-LNT electrograms (raw and filtered) matched closely at the sites where no P deflections were evident in the pre-LNT recordings, as shown in Fig. 3 B.
The extent to which P deflections were attenuated by 2,000 samples/s mapping is shown in the electrograms and derivatives from a site with a P deflection in Fig. 4. During this experiment, sampling rate was increased from 2,000 to 4,000, 8,000 to 12,000 samples/s, with accompanying increases in low-pass cutoff frequencies from 1 to 2 kHz and 4 kHz. The 4-kHz cutoff was the mapping system upper limit. As sampling rate increased, the P deflection amplitude increased (2.5–6.9 mV), whereas the minimum derivative decreased (−2.2 to −10.0 V/s). Nevertheless, the timing for P and V deflections was consistent and P-V latency exceeded 2 ms at each sampling rate.
Activation sequence reconstruction.
Activation sequences were assessed in three ways. First, we used the raw electrograms to determine the spatial distribution of active samples and identify the peripheral conduction system and myocardial wave fronts that crossed the mapped region. For every experiment and activation sequence, electrodes were drawn on a schematic as small squares. Those squares were placed into the shape of the recording array. At electrodes where the raw signal derivative fell below −0.5 V/s, those squares were filled to indicate active tissue (24). As described previously, this approach avoided many of the ambiguities associated with accurate identification of unique activation times that complicate traditional isochrone map construction (6). This approach facilitated identification of active sites that collected into thin bands corresponding with peripheral conduction system wave fronts, and active samples that collected into broad regions corresponding with myocardial wave fronts. Second, we assembled isochrone maps with the use of V deflection times from the filtered electrograms to assess myocardial wave front expansion independent of peripheral conduction system wave front expansion. Isochrones were assembled with 1-ms intervals. Third, we quantified activation sequence differences after pacing adjustments with sample Pearson correlation coefficients (CCF) by using V deflection times.
Incidence of P deflections during sinus rhythm and 500 ms pacing.
By using the electrogram processing criteria, we found P deflections in 22% of recordings from the sinus mapping experiments and 15% of recordings from the endocardial pacing experiments (500-ms cycle length). In both the sinus and paced activation sequences, we found P-led and V-led sites. Figure5 A shows means ± SD for the number of P-led and V-led sites from both sets of experiments. For the paced activation sequences, the entries marked overall were obtained by combining data from septal, apical, and basal drives. During sinus mapping, we found 117 ± 51 sites with unique P and V deflections. P deflections preceded V deflections in 91% of the electrograms. During 500-ms pacing, the overall mean ± SD for the number of sites with P deflections was 81 ± 59 (range 13–226). P-led sites accounted for 38% of electrograms with P deflections, and V-led sites accounted for 62%. Septal stimulation produced the most sites (113 ± 61) with P deflections, with 40% P-led and 60% V-led sites. Apical stimulation produced the fewest sites (42 ± 29), with 63% P-led and 37% V-led sites. Figure5 B shows P-V latency magnitudes grouped as in Fig.5 A. Mean PV latency magnitude measured 6.7 ± 3.9 ms during sinus rhythm and 5.7 ± 3.1 ms during 500-ms pacing. Latencies measured 5.7 ± 2.0 ms during septal pacing, 5.1 ± 3.3 ms during apical pacing, and 6.3 ± 3.8 ms during basal pacing. Differences with pacing site were not statistically significant. We therefore combined latencies resulting from all stimulus sites for the analyses in subsequent sections.
Sinus and paced activation sequences.
Consistent with the relatively small P-V latency magnitudes, activation sequences during sinus rhythm and 500-ms pacing showed peripheral conduction system and myocardial wave fronts that remained close to one another throughout rabbit RV free wall. During sinus mapping, peripheral conduction system wave fronts barely preceded myocardial wave fronts. Figure 6 A shows the spatial distribution of active samples determined from the raw electrograms during an experiment, in which there were 190 sites with unique P deflections, 170 of which were P-led. P-V latency magnitude averaged 5.4 ± 3.2 ms for this experiment. Figure 6 Bshows isochrone maps assembled from the filtered electrograms. Frames were aligned between panels to visualize the combined expansion of the peripheral conduction system and myocardial wave fronts (Fig.6 A) alongside the expansion of the myocardial wave front alone (Fig. 6 B). The earliest activity resulted from one or more peripheral conduction system wave fronts, as active samples in the 0-ms frame from Fig. 6 A were unaccompanied by activity in Fig. 6 B. Peripheral conduction system wave fronts expanded during the 2-, 4-, and 6-ms frames, with the first evidence of a myocardial wave front (W1) in the apicoseptal portion of the 6-ms frame. W1 expanded in the 8-, 10-, and 12-ms frames. Its position was evident in Fig. 6 A as the broad collection of active samples. At 14 ms, a second myocardial wave front (W2) originated in the conus and coalesced with W1 over the 16- and 18-ms frames. A third myocardial wave front (W3) originated in the lateral free wall at 20 ms. It was overtaken by W1 at 22 ms. The last region to excite was in the basolateral corner. All wave fronts moved out of the measurement region by 28 ms.
The extent to which peripheral conduction system wave fronts preceded myocardial wave fronts in different locations is shown in Fig.7 A. The separation between wave fronts was highly variable. For example, in the apicoseptal region where the initial peripheral conduction system and myocardial activity was observed, P-V latencies exceeded 10 ms at most sites on the septal edge of the plaque but were considerably smaller on the apical edge of the plaque. Figure 7 B shows electrograms recorded on a septum-to-free wall line of electrodes from this region. P deflections were separated by more than 2 ms from V deflections at 4 of 6 electrodes on this row. P deflection timing indicated sequential activation at sites i–iv. V deflections were largely aligned, consistent with the entry of W1 into these sites as a broad wave front. Figure 7 C shows electrograms from an intersecting column of electrodes in the path of W1 expansion. V deflection timing in these electrograms indicated sequential activation from the most apical (site i) to most basal (site vi) electrodes.
Because electrograms were recorded in unipolar mode, we were unable to apply the criterion of Veenstra et al. (40) to establish whether P-V latencies decreased because antegrade PVJ conduction occurred in the apicoseptal region. By using a reference electrode retracted 1–2 mm from the tissue surface and directly above the recording electrode, those investigators identified discrete PVJs based on the rapid biphasic P deflection that preceded the uniphasic, completely negative V deflection. By using the filtered electrograms, however, we were able to detect some islands of early myocardial activity that suggested antegrade PVJ conduction, although the close proximity of the peripheral conduction system and myocardial wave fronts throughout the mapped region limited that detection. One exception was in the lateral free wall region, where W3 (Fig.6 B, 18 ms) emerged. Figure 7 D shows electrograms recorded on a septum-to-free wall line of electrodes that spanned the W3 island. The timing for prominent P deflections at sites iand ii indicated peripheral conduction system wave fronts approached this region in advance of W3 initiation. V deflection timing indicated myocardial wave front initiation near sites iiiand iv with components expanding toward the septum (throughsites i and ii) and the free wall (throughsites v and vi). Electrograms from sites on the orthogonal axis shown in Fig. 7 E showed V deflection timing that indicated W3 components expanded toward the apex (throughsites i and ii) and the base (through sites v and vi) as well.
During paced activation sequences, the peripheral conduction system wave fronts generally lagged behind the myocardial wave fronts, although the extent to which wave fronts separated was comparable to that of the sinus activation sequences. Figure8 shows the spatial distribution of active samples and selected electrograms after septal stimulation in an experiment in which there were 159 V-led sites and no P-led sites. From Fig. 8 A, the broad myocardial wave front that emerged from the septum 50 ms after stimulation expanded toward the flap in between 53 and 80 ms during this activation sequence. Small bands of electrodes that formed the peripheral conduction system wave front were scattered throughout the 62- to 80-ms frames. From Fig. 8, B andC, slower V deflections preceded the faster P deflections at all sites where P deflections were evident. P-V latencies at these sites ranged between −3.9 ms and −12.3 ms. The P-V latency magnitude from this experiment measured 7.5 ± 2.8 ms.
After reduction in the steady-state cycle length from 500 to 125 ms, myocardial wave fronts still preceded peripheral conduction system wave fronts, and there was a modest increase in their separation. Across all experiments, P deflections were evident at 57 ± 36 sites with 74% of those sites being V-led. P-V latency magnitude increased from 5.7 ± 3.1 to 6.8 ± 2.6 ms (P = 0.02). Figure 9, A and B, shows activation sequences during 500 and 125 ms (respectively) pacing from the septum in one experiment. Both activation sequences showed peripheral conduction system wave fronts in advance of and in the wake of the myocardial wave front in different parts of the mapped region. The main difference between the two activation sequences was in their timing, as wave front expansion during 125-ms pacing was slower than that during 500 ms pacing. CCF between these two activation sequences measured 0.92. Across all 24 activation sequences, CCF measured 0.79 ± 0.18.
By comparison, premature stimulation caused more pronounced separation between peripheral conduction system and myocardial wave fronts and larger changes to the activation sequence. P-V latency magnitude increased from 5.6 ± 2.3 to 11.7 ± 10.4 ms (P = 0.005), although the number of sites with P deflections only decreased from 85 ± 54 to 72 ± 51. The percentage of V-led sites was 61%. Figure 9 C shows the activation sequence after premature stimulation from the same preparation and septal pacing site as Fig. 9, A andB. During premature stimulation, initial activity within the measurement region was located in the conus. The activation sequence during premature stimulation was poorly correlated to the 500-ms pacing sequence at CCF = 0.04. This example was consistent with correlations between activation sequences during premature stimulation and 500-ms pacing across all experiments (CCF = 0.63 ± 0.37).
A consistent feature that highlights the functional nature of coupling between the peripheral conduction system and myocardium that we observed with cycle length accommodation and premature stimulation was a dynamic shifting of P-V latencies at individual recording sites. Figure 10 shows electrograms recorded in the conus during 500-ms pacing, 125-ms pacing, and premature stimulation from the experiment presented in Fig. 9. In the septum to flap direction, P-V latencies that measured 7.4 ms (site ii), 6.9 ms (site iii), and 5.5 ms (site iv) were observed during 500-ms pacing. Those delays shortened during cycle length accommodation, with a P-V latency exceeding 2 ms limited tosite iii (2.3 ms). During premature stimulation, we only observed V deflections that preceded P deflections along this line of electrodes. In the base-to-apex direction, no P-V latencies were measured during 500-ms pacing or premature stimulation. During cycle length accommodation, however, P-V latencies of 2.5 ms (site ii), 2.1 ms (site iii), and 2.3 ms (site v) were observed. Across all experiments, rapid pacing shifted 81% of the P-led sites to V-led sites and 61% of V-led sites to P-led sites. With premature stimulation, 70% of P-led sites became V-led and 60% of V-led sites became P-led sites.
Initiating ectopic cycles.
Ectopic beats were observed after 12 of the 24 premature stimuli. Cycles persisted for <1 s in 2 of 12 cases, between 1 and 2 s in 4 of the 12 cases, and between 5 and 24 s in 6 of the 12 cases. The initiating ectopic cycles showed comparable separation between peripheral conduction system and myocardial wave fronts to that measured during premature stimulation, as mean P-V latency magnitude measured 10.7 ± 5.5 ms. There were 60 ± 37 sites with P deflections, and the percentage of sites that were V-led (69%) exceeded the percentage of sites that were P-led (31%). Overall, initiating ectopic activation sequences were poorly correlated to the paced (CCF = 0.32 ± 0.45) and premature (CCF = 0.49 ± 0.29) sequences.
The distinguishing characteristic of these cycles compared with the sinus and paced sequences was the presence of several slow moving wave fronts that appeared and disappeared in different portions of the measurement region. In many instances, it appeared as if the peripheral conduction system and myocardium were functionally uncoupled. For example, Fig. 11 shows the initiating ectopic activation sequence from an experiment in which first activity was observed in the conus. Electrograms recorded from this region are denoted as a, b, and c, and the responses during 500-ms pacing, premature stimulation, and the ectopic cycle are included. Note that all three sites were V-led (6.0–7.0 ms) during 500-ms pacing and switched to P-led (2.5 ms) during premature stimulation. During the ectopic cycle, separation between peripheral conduction system and myocardial wave fronts was pronounced at these sites, as the P-V latencies measured 19.0–22.0 ms. Comparable separation between wave fronts was also observed at the basal sites marked d, e, and f. However, these sites were P-led during 500-ms pacing (2.0–2.5 ms) but became V-led during premature stimulation (6.5–7.0 ms) and the ectopic cycle (18.0–24.0 ms).
We used 528-channel mapping of perfused rabbit RV endocardium to measure the relative timing for P and V deflections during sinus rhythm and during endocardial pacing at basic cycle length, cycle length accommodation, and premature stimulation. Because premature responses were followed by ectopic beats in 50% of the activation sequences analyzed, we were further able to compare the timing of P and V deflections during arrhythmia formation to that during pacing. New findings that arose from the study include demonstration of1) comparable P-V latency magnitudes during sinus rhythm and 500-ms pacing, 2) a consistent preference for V deflections that preceded P deflections during endocardial pacing, and3) pronounced separation between peripheral conduction system and myocardial wave fronts during premature stimulation and ectopic beats. These findings are important because they highlight the dynamic nature of electrical interactions between the peripheral conduction system and myocardium in preparations with relatively strong PVJ coupling between layers.
528-Channel mapping of perfused RV endocardium.
We established these relationships by using an integrated methodology that overcame limitations commonly associated with endocardial mapping. We used high spatial and temporal resolution during simultaneous recording with fixed electrode geometry in continually perfused preparations. This approach differed from endocardial mapping studies that used monophasic action potential (19) or transmembrane potential-dependent fluorescence (3, 18, 23) recordings in that it provided information on both P and V deflections. P deflections have been reported in studies with intramural needles for three-dimensional mapping of in situ (2,5) and Langendorff-perfused (17) hearts. However, such studies typically focused on intramural contributions to activation sequence and presented less detail regarding endocardial contributions than we present.
Details regarding the separate expansion of peripheral conduction system and myocardial wave fronts have been presented in previous studies (30, 31, 40) using sequential recordings from superfused canine septum-free wall preparations. We recognize an advantage of such studies is their unambiguous demonstration of Purkinje cell upstrokes from impalements. However, maintenance of midwall and epicardial myocardium distinguished our study from their work because midwall uncoupling resulting from healing over for spared subendocardium likely reduces the electrical load imposed by the ventricular layer on the peripheral conduction system layer. Load is an important determinant of antegrade PVJ conduction delay (21,32). Therefore, maintenance of midwall myocardium is an important factor in the use of P-V latencies to assess peripheral conduction system and myocardial wave front expansion.
We were able to establish relationships between peripheral conduction system and myocardial wave fronts by applying criteria from experiments with superfused preparations. Specifically, we analyzed timing for P and V deflections that represented peripheral conduction system and ventricular layer activation, respectively. This approach is specifically supported by Nagao et al. (31), who analyzed their sequentially recorded electrograms for rapid P deflections (1- to 3-ms duration) that preceded slower V deflections (6- to 10-ms duration) because accompanying microelectrode impalements showed those deflections corresponded to Purkinje and ventricular cell upstrokes, and by other investigators who have used similar approaches (25,31, 32, 35). Consistent with these reports, we found that the rapid deflections most likely resulted from peripheral conduction system activation because they were eliminated during LNT application (Fig. 3).
Rabbit RV free wall.
Rawling et al. (35) suggested that limited PVJ coupling is advantageous for endocardial excitation. Those investigators simulated action potential propagation with a fast conducting Purkinje cable connected to a slow conducting myocardial cable at 7-mm intervals. Such spacing was advantageous because the distribution of early sites for action potential initiation in the myocardium established multiple wave fronts that coalesced to achieve rapid ventricular excitation. Furthermore, the limited coupling lowered the electrical load imposed by the myocardial cable on the Purkinje cable, which allowed the Purkinje cable to maintain high conduction velocity. Such sparse PVJ coupling is typical of superfused canine papillary muscle and free wall preparations and is also consistent with our observations during endocardial mapping from 14 × 14-mm regions of Langendorff-perfused canine hearts (33). In those experiments, we identified 2–3 early sites from which secondary myocardial wave fronts were initiated in advance of the primary wave front established by pacing on the periphery of the region. Our limited ability to routinely identify early sites during sinus mapping or 500-ms pacing in the present experiments supports a much higher degree of PVJ coupling in the rabbit RV free wall than in canine hearts. In this regard, it is also important to distinguish our preparation from previous studies that used rabbits. Overholt et al. (32) used rabbit LV papillary muscles and showed the basal halves of the rabbit preparations were devoid of peripheral conduction system, consistent with canine preparations. Stimulation of free running Purkinje strands attached to the muscle established distinct peripheral conduction system and myocardial wave fronts that allowed early site identification and assembly of separate activation maps for the peripheral conduction system and myocardial components. Tranum-Jensen et al. (38) studied superfused rabbit LV preparations in which free running strands were driven to facilitate PVJ identification with electrical recordings in advance of fixation to allow histologic reconstruction of the junctional region.
In contrast to the rabbit LV preparations that share similarities with canine preparations, the rabbit RV free wall includes a web of Purkinje strands that overlie a network of anastomosing trabeculae. Although relatively long (6–10 mm) free running strands course from the septum to the apical free wall in most preparations, numerous short strands course between trabeculae. Furthermore, LNT staining shows significant Purkinje cell distribution along trabeculae. Assuming that PVJs are more densely distributed over these trabeculae than in other preparations provides a plausible explanation for our observation that comparable P-V latency magnitudes were measured during sinus rhythm and 500-ms pacing. A high degree of coupling would prevent either the peripheral conduction system or the myocardial wave front from becoming unsynchronized. In this regard, use of either the proximal conduction system during a sinus beat or a myocardial wave front during endocardial pacing would be expected to establish synchronous wave front expansion between peripheral conduction system and ventricular layers in approach to the measurement region, consistent with our findings.
Preference for V-led sites during endocardial pacing.
By analyzing separate P and V deflections, we did not expect to find the preference for V-led sites that we observed during 500-ms pacing. Our 500-ms pacing protocol was designed to promote Purkinje-mediated activation in the measurement region. We used bipolar stimulation of septal, apical, and basal tissue because we anticipated such placement would provide sufficient time for action potential propagation from the myocardium to the peripheral conduction system in advance of both wave fronts reaching the plaque. We believe the V-led preference relates both to the developmental similarities between the trabecular component of the ventricular layer and the specialized conduction system, and to electrical loading of the peripheral conduction system by the myocardium. Studies (29) of specialized conduction system development suggest that trabecular tissue may be more closely related to the conduction system than to the free wall myocardium and may in fact be the precursor to the entire conduction system. The genetic and molecular similarities between cells in the trabecular component and the conduction system correspond to similar functions in the activation sequences of developing hearts. For example, De Jong et al. (16) found that endocardial electrograms recorded from 4-day-old embryonic chicken hearts before conduction system development revealed earlier activation near the apex than near the atrioventricular canal. This result was similar to the pattern of activation spread after conduction system development. In subsequent simultaneous endocardial and epicardial electrogram recordings, those investigators found that activation occurred earlier on trabeculae than at corresponding epicardial sites, suggesting that activation spread preferentially through the ventricular trabeculae. In addition to producing a similar pattern of ventricular activation to that produced by the fully developed conduction system, trabeculae display a conduction velocity intermediate to Purkinje and myocardial fibers due to their higher connexin43 expression than myocardium.
These developmental links suggest refinement to the interpretation of Rawling et al. (35) for Purkinje-mediated activation sequence. In that interpretation, PVJ conduction delays minimally influenced the activation sequence because Purkinje conduction velocity was much higher than myocardial conduction velocity and PVJs were separated from one another. During myocardial stimulation, overall activation was similarly enhanced because retrograde PVJ conduction near the stimulus site initiated a Purkinje wave front that established antegrade conduction at the distant PVJ. In canine preparations, both sinus and endocardially paced activation sequences show antegrade conduction across most PVJs. This occurs during ventricular drives because the high Purkinje conduction velocity and the coarse PVJ distribution allow peripheral conduction system wave fronts to accelerate past myocardial wave fronts and establish antegrade conduction at distant PVJs. In contrast, we believe coordinated development with enhanced gap junction expression along trabeculae (29) led to conditions in which the discrepancy between conduction velocities in the rabbit peripheral conduction system and trabecular layers was much smaller than in canine papillary muscles. More continuous PVJ coupling, with accompanying retrograde PVJ conduction delays during propagation initiated in the ventricular layer, would necessarily result in V deflections that preceded P deflections.
Cycle length accommodation.
Further evidence for this interpretation is provided by the relatively modest changes to endocardial activation sequence and mean P-V latency magnitudes that we observed during cycle length reduction. The spatial characteristics of activation sequences were maintained because the sequences during 125-ms pacing were highly correlated to sequences during 500-ms pacing. The sequences slowed, however, which was consistent with reports demonstrating steady-state accommodation to fast rates included conduction velocity slowing in isolated Purkinje strands (8, 9) and papillary muscle preparations (9,12). Also, V-led sites remained preferential to P-led sites. This likely occurred because retrograde PVJ conduction delays were minimally influenced by the rapid rate. In isolated guinea pig papillary muscle preparations, Chen and Gettes (12) showed that cycle length reduction from 333 to 160 ms reduced upstroke velocity by only 10%. Such a small change would have negligible effect on source charge available for Purkinje cell activation via retrograde PVJ conduction, especially because downstream sink charge requirements for maintenance of such conduction are relatively modest due to the high membrane resistance that establishes high excitability of Purkinje cells (10, 13, 14, 21, 34). By comparison, antegrade PVJ conduction time likely increased with rate acceleration. In isolated canine Purkinje strands, Bigger et al. (8) reported that cycle length reduction from 500 to 200 ms reduced upstroke velocity by 20%. Such reduction places increased importance on the maintenance of a depolarized action potential plateau for antegrade PVJ conduction. That importance is highlighted by prolongation of delays in canine Purkinje-papillary muscle preparations exposed to cadmium by Wiedmann et al. (41). Cadmium partially blocks L-type calcium current, which maintains the action potential plateau. A plausible explanation for our observation that P-led sites were still observed relates to the prominent transient outward current of rabbit Purkinje cells (14). At slow rates, that current dramatically repolarizes the early plateau of the Purkinje cell action potential (21). Rate acceleration markedly inhibits transient outward current with a concomitant depolarization of the plateau, which would in turn facilitate antegrade PVJ conduction.
Our finding that endocardial activation sequences were altered by premature stimulation is seemingly contradictory to findings by Van Dam and Janse (39). By completing multiple needle electrode mapping studies with isolated canine hearts, those investigators showed similar transmural activation sequences occurred during premature stimulation and during epicardial stimulation at basic drive cycle lengths. However, the premature sequence was slower, and the lack of clearly defined Purkinje potentials (i.e., rapid deflections) on the endocardial electrodes led those investigators to conclude that the peripheral conduction system contributed minimally to activation sequence. Whereas our finding that endocardial activation sequence slowed is consistent with this earlier report, we note important differences that suggest the peripheral conduction system contributed to the response to premature stimulation. For example, we found that the activation sequence changed considerably, as evidenced by the low correlation coefficient between the paced and premature sequences (0.63 ± 0.37). This occurred despite the existence of more sites where P deflections were detected than during cycle length accommodation. We believe the explanation for these collective changes is related to the marked increase in P-V latency. That increase suggests that delayed or failed PVJ conduction allowed peripheral conduction system and myocardium to functionally uncouple. Failed PVJ conduction is consistent with measurements by Gilmour and Watanabe (20), who used canine Purkinje-papillary muscle preparations to show that premature papillary muscle excitation prolonged retrograde conduction time, and with measurements by Sasyniuk and Mendez (36), who used canine papillary muscle-false tendon preparations to show that false tendon premature stimulation produced conduction block at PVJs proximal to the stimulus site. This interpretation is also consistent with the report of Ben-Haim et al. (5), who used multiple needles arranged in an apex-to-base line in the LV of in situ canine hearts to study relationships among Purkinje, endocardial, and epicardial activation sequences. Those investigators found conduction delays in the peripheral conduction system with premature stimulation changed activation sequence by allowing remote epicardium to be activated by alternate routes.
Our finding regarding the pronounced separation between peripheral conduction system and myocardial wave fronts during the initiating ectopic cycles is consistent with the suggested contribution of the conduction system to arrhythmia formation by Berenfeld and Jalife (7). Those investigators assembled a three-dimensional model of the mammalian ventricles and attached a representative conduction system to 1-mm3 ventricular elements at 214 PVJ sites. Reentry was initiated by initial conditions in the FitzHugh-Nagumo equations that were intended to represent triggered Purkinje activity in elemental excitation and recovery processes. Initiation gave rise to polymorphic ventricular tachycardia with drifting epicardial breakthrough sites and endocardial foci that originated at PVJs. Evolution of separate reentrant activation sequences in specialized conduction system and ventricular elements was essential to the formation of intramural reentrant circuits. Once these latter circuits formed, complete uncoupling at PVJs had no effect on activation sequence. Although our methodology prevented reconstruction of separate activation sequences during ectopic beats, we consistently observed P-V latencies exceeding 10 ms. Latencies of such magnitude are well beyond reported measures for PVJ conduction times. This asynchrony of P and V deflections in the present study supports the concept that separate activation sequences in the peripheral conduction system and overlying ventricular layer precede arrhythmia formation.
In assessing our findings, it is important to recognize certain limitations. To facilitate endocardial access and plaque positioning with electrodes covering most of the RV free wall, preparations were pinned in a chamber during aortic perfusion. The horizontal orientation may have deformed the aortic valve annulus and could potentially have caused valvular insufficiency. Although signal characteristics were maintained through the protocols and we did not observe marked the S-T segment elevation that suggests ischemia, the possibility that perfusion was incomplete cannot be excluded. In this regard, horizontal positioning, the process of cryoablation, and the free wall dissection required for endocardial access may all have contributed to the relatively high incidence of ectopy that we observed in response to single premature stimuli.
It would have been advantageous to complete all protocols after LNT application to eliminate contributions of the specialized conduction system to the endocardial activation sequences. This information would have been especially beneficial in analyzing the transition from premature to ectopic responses, because a demonstration that endocardial-programmed pacing following LNT application failed to establish nonsustained arrhythmias of the type observed pre-LNT would have been compelling evidence for a contribution of the peripheral conduction system to endocardial arrhythmia formation. Unfortunately, we had limited success with maintaining preparations after LNT application because there was marked signal degradation despite the relatively small amounts (0.2–0.4 ml) used in our applications. This finding was in contrast to experiments with isolated and in situ canine hearts (11, 15, 28, 33), where LNT has been used to effectively ablate specialized conduction system. Our observation suggests some destruction of endocardial myocytes accompanies Purkinje cell ablation during LNT application. Because the rabbit RV free wall is so thin compared with the canine heart, accompanying myocyte damage would necessarily be expected to have a more pronounced influence in rabbit RV free wall than in preparations from larger hearts. The use of an animal model with larger heart size may therefore have been advantageous to facilitate the completion of pacing protocols pre- and post-LNT.
We used a 2-ms threshold for the separation between P and V deflections to measure P-V latencies. Although antegrade PVJ conduction delays on the order of 2 ms were reported by Tranum-Jensen et al. (38) in experiments using superfused rabbit LV preparations, it was not our intent to relate conduction delay to this threshold. The threshold was selected on practical grounds, because we were unable to identify unique deflections without ambiguity when they were spaced closer than 2 ms. One consequence of this selection is that we potentially underreported the number of sites with P deflections. Sites at which peripheral conduction system and myocardium excited within 2 ms of one another were treated in the same way as sites at which no peripheral conduction system was present. Likewise, sampling at a higher rate may also have allowed the identification of more P deflections. By using isolated canine Purkinje strands, Barr and Spach (4) found a rate of 15,000 samples/s was necessary for complete reconstruction of unipolar electrograms.
Although our use of a plaque recording array was advantageous in terms of completing the experiments because simultaneous data acquisition with fixed-electrode geometry was achieved, the array itself imposed certain limitations that potentially influenced our analyses. Because the Ag/AgCl electrodes were embedded in epoxy, it was not practical to integrate a 1–2 mm retracted reference for each electrode. As described by Veenstra et al. (40), the primary advantage to using such a reference is that discrete PVJs can be identified from characteristic signal components, without the need to analyze those components alongside the myocardial activation sequence. As confirmed by the detailed electrophysiological and histological measurements by Tranum-Jensen et al. (38), a rapid biphasic P deflection that precedes a uniphasic, completely negative V deflection corresponds with underlying Purkinje and ventricular cell depolarization in such electrograms. If we had been able to integrate such electrode pairs into our recording array, it may have been possible to identify more sites with P deflections and to label more regions as antegrade PVJ conduction sites. Additionally, the use of a plaque prevented direct electrode contact at all sites on the RV free wall because of the trabeculation that is characteristic of this preparation. Local signal attenuation associated with differences in contact likely contributed to the relatively low percentage of P deflections we measured in all experiments.
It would also have been advantageous to complete sinus mapping in advance of endocardial pacing protocols using the same preparations. This was a practical limitation that resulted from our goal of limiting viable tissue to the measurement region in the pacing experiments, which was especially important for mapping of the initiating ectopic cycles to ensure the arrhythmias did not simply result from atrioventricular conduction. Mapping of sinus beats in advance of endocardial pacing would have allowed measurements of the change in P-V latencies at individual sites and might additionally have allowed identification of more early sites suggesting discrete PVJs as described for larger hearts.
The authors acknowledge the editorial comments of Dr. Delilah Huelsing.
This work was supported by National Science Foundation Awards BES-9457212 and BES-9903466, American Heart Association Southeast Affiliate Award 0051196B, and National Heart, Lung, and Blood Institute Grants HL-54024, HL-33637, and HL-28429.
Address for reprint requests and other correspondence: A. E. Pollard, Cardiac Rhythm Management Laboratory, Univ. of Alabama at Birmingham, Volker Hall B140, 1670 University Blvd., Birmingham, AL 35294 (E-mail:).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2001 the American Physiological Society