Influence of Postshock Epicardial Activation Patterns on Initiation of Ventricular Fibrillation by Upper Limit of Vulnerability Shocks
Background—Shocks of identical strength and timing sometimes induce ventricular fibrillation (VFI) and other times do not (NoVFI). To investigate this probabilistic behavior, a shock strength near the upper limit of vulnerability, ULV50, was delivered to yield equal numbers of VFI and NoVFI episodes.
Methods and Results—In 6 pigs, a 504-electrode sock was pulled over the ventricles. ULV50 was determined by scanning the T wave. S1 pacing was from the right ventricular apex. Ten S2 shocks of approximate ULV50 strength were delivered at the same S1-S2 coupling interval. Intercycle interval (ICI) and wave front conduction time (WCT) were determined for the first 5 postshock cycles. ICI and the WCT of cycle 1 were not different for VFI versus NoVFI episodes (P=0.3). Beginning at cycle 2, ICI was shorter and WCT was longer for VFI than NoVFI episodes (P<0.05).
Conclusions—The first cycle after shocks of the same strength (ULV50) delivered at the same time has the same activation pattern regardless of shock outcome. During successive cycles, however, a progressive decrease in ICI and increase in WCT occur during VFI but not NoVFI episodes. These findings suggest shock outcome is (1) deterministic but exquisitely sensitive to differences in electrophysiological state at the time of the shock that are too small to detect or (2) probabilistic and not determined until after the first postshock cycle.
Astrong stimulus during the vulnerable period can induce repetitive responses that either halt without inducing ventricular fibrillation (VF) or degenerate into VF.1 Most proposed mechanisms of VF induction based on this finding, such as the nonuniform dispersion of refractoriness hypothesis,2 imply that activation immediately after the shock in a successful VF induction (VFI) differs from that in a failed VF induction (NoVFI).2 3 4
Previous studies that used a range of shock strengths and timings showed that the interval between the shock and the first global postshock activation is shorter for VFI than for NoVFI shocks.3 5 However, comparison of VFI and NoVFI episodes after shocks of the same strength has not been reported.
In this study, we determined activation patterns after shocks of identical strength and timing. A shock strength near the upper limit of vulnerability that induced VF in ≈50% of the trials (ULV50) was used. We tested the hypothesis that the activation pattern immediately after VFI shocks differed from that after NoVFI shocks.
Thirty of the 60 shocks in the 6 pigs were VFI episodes. One VFI episode was excluded because the last S1 stimulus did not capture. A shock at ULV50 produced a shock voltage of 498±112 V. The mean coupling interval (CI) at which VF was induced with ULV50 shocks was used as the S1-ULV50 CI (sCI); the sCI was 214±14 ms. To determine ULV50, 15±11 shocks were required. The diastolic pacing threshold (DPT) was 0.2±0.1 mA. Heart weight was 154±45 g. For each animal, delivered shock voltage for the 10 ULV50 shocks was nearly constant (%SD of 0.1 to 0.3). Repeatability of the shock potential distribution at the 504 electrodes was measured in 1 pig. The mean correlation coefficient of the potentials was 0.995±0.004 for all 10 shocks.
The preshock interval and wave front conduction time (WCT) of the last paced cycle before the shock was not different between VFI and NoVFI episodes (Table 1⇓). The similarity of the last paced cycle before the shock (Table 2⇓) was not different, which suggests that the activation sequence of the last paced cycle was constant for all 10 shock episodes.
Cycle 1 Site of Earliest Activation
Cycle 1 sites of earliest activation (SEAs) were always at the anteroapical left ventricle (LV). While the cycle 1 SEA varied slightly between animals, it was highly repeatable for each animal (Table 3⇓), showing that the first postshock cycle appeared in the same epicardial region regardless of shock outcome.
Maximum first derivative (dV/dt) of activation at the cycle 1 SEA was not different for VFI and NoVFI episodes (Table 3⇑). Mean dV/dt at the SEA of cycles 2 to 5 was less negative than that of cycle 1 for VFI episodes (P<0.01). However, no dV/dt differences were found among the first 5 postshock cycles in NoVFI episodes. For VFI episodes, SEA repeatability for cycles 2 to 5 (Table 3⇑) was slightly lower than for cycle 1 because the SEA of cycles 2 to 5 moved a short distance (2±3 electrodes) away from the cycle 1 region of earliest activation (REA). However, the SEA of cycles 2 to 5 in NoVFI episodes moved a much greater distance away from the cycle 1 REA, 9±4 electrodes (P<0.002 vs VFI episodes).
Propagation Pattern in VFI Episodes
A typical VFI episode is shown in Figure 1A⇓ (first cycle) and Figure 2A⇓ (subsequent cycles). All 5 cycles began in the anteroapical LV 36, 140, 229, 320.5, and 430 ms after the shock, respectively. Cycle 1 initially propagated toward the LV base, blocking at the right ventricular (RV) apex. It continued bilaterally around the apex, rejoining on the posterior RV to complete activation. Subsequent cycles did not block at the RV apex and activated both ventricles in a more radial pattern from apex to base. WCTs increased progressively up to cycle 3, then slightly decreased (89, 161, 215, 170, and 160 ms, respectively). Cycle 2 did not overlap temporally with cycle 1; however, subsequent cycles all overlapped with their immediate predecessor.
Propagation Pattern in NoVFI Episodes
Cycle 1 after a NoVFI shock in the same animal (Figure 1B⇑) was nearly identical to cycle 1 of the VFI episode. This first-cycle similarity is also apparent in the electrograms (Figure 3⇓). Because cycle 1 for both episodes blocked at the RV apex, we paced from the anterobasal LV in the absence of shocks to establish there were no anatomic or functional barriers at the apex to cause the block (Figure 1C⇑).
Cycle 2 from this NoVFI episode (Figure 2⇑, B1) followed a pathway similar to the VFI cycle 2 but began later and conducted faster. Overlapping cycles were absent. Cycles 3 (B2) and 4 (B3) arose after long delays from different SEAs and had fast WCTs (73 and 38 ms, respectively). Cycle 4 was sinus and not analyzed.
Postshock Intercycle Intervals
The mean intercycle interval (ICI) (Figure 4A⇓) of cycle 1 did not differ between VFI (51±23 ms) and NoVFI episodes (n=24, 68±78 ms). The SD of NoVFI episodes was high because of 1 episode with an extremely long postshock interval (461.5 ms). This episode had its SEA near the posterobasal LV, whereas the SEAs from other episodes in this animal were at the anteroapical LV. If this NoVFI episode is excluded, the mean postshock interval for NoVFI episodes becomes 54±23 ms (P=0.6 vs VFI). ICIs for VFI episodes were significantly shorter than for NoVFI episodes for cycle 2 (138±38 vs 387±312 ms) and cycle 3 (132±31 vs 389±193 ms) (n=12 and 10, respectively). For cycle 4, the mean ICI for VFI (126±40 ms) was shorter than for NoVFI (484±182 ms) episodes; however, the difference was not significant, probably because there were only 5 NoVFI episodes with a fourth ectopic cycle. ICIs of cycle 5 were not compared because only 1 NoVFI episode had a cycle 5 (537 ms). ICIs of cycle 5 for VFI episodes were 128±41 ms.
The WCT (Figure 4B⇑) of cycle 1 did not differ between VFI (116±19 ms) and NoVFI episodes (113±16 ms). The mean WCTs of cycles 2 (172±51 vs 99±44 ms), 3 (203±51 vs 83±18 ms), and 4 (197±55 vs 59±25 ms) were significantly longer for VFI than for NoVFI episodes (P<0.01). WCT of the single cycle 5 for NoVFI episodes was 58 ms. WCT of cycle 5 for VFI episodes was 210±53 ms.
In all VFI episodes, overlap occurred during cycles 2 to 3, 3 to 4, and 4 to 5 (overlapping index >1) but not cycles 1 to 2 (overlapping index <1) (Figure 4C⇑). In contrast, there was no overlap among the first 5 ectopic cycles in any NoVFI episode.
Correlation of 504 Electrograms of Cycle 1
The similarity function, sij, and temporal lag, τm, were compared for all possible combinations of first cycles in each animal and divided into 3 groups: VFI versus VFI, VFI versus NoVFI, and NoVFI versus NoVFI episodes (Table 4⇓). There were no differences in any group for either variable. The very high similarity and short temporal lag among different episodes indicate that first postshock cycles were nearly identical whether or not VF was initiated.
Cycle 2 in NoVFI Episodes
In NoVFI episodes, cycle 2 could be divided into 2 distinct subgroups. Subgroup 1 (n=7) had a short ICI, long WCT, and same REA as cycle 1. Subgroup 2 (n=6) had a long ICI, short WCT, and different REA than cycle 1.
ICIs in subgroup 1 were all <200 ms (142±27 ms) but were all >400 ms (672±228 ms) in subgroup 2 (P<0.002). Although the cycle 2 ICI of subgroup 1 was not different from that in the VFI episodes, in subgroup 2 it was significantly longer than in the VFI episodes (P<0.002). Subgroup 1 all had WCTs >100 ms (135±21 ms), whereas subgroup 2 all had WCTs <80 ms (56±12 ms, P<0.01). WCTs of both subgroups were significantly shorter than WCTs of cycle 2 in VFI episodes (P<0.004).
Cycle 2 SEA repeatability in subgroup 1 (63±48%) was different (P<0.04) from subgroup 2 (0±0%) in NoVFI episodes. Cycle 2 SEA repeatability in VFI episodes (60±33%) was different from that of subgroup 2 (P<0.007) but not subgroup 1. Thus, cycle 2 SEAs in the No-VFI subgroup with longer ICIs (subgroup 2) moved to a different site, whereas in the NoVFI subgroup with shorter ICIs (subgroup 1), as well as in VFI episodes, they mostly remained in the REA of cycle 1. The mean dV/dt for cycle 2 in VFI episodes also differed from that of subgroup 2 (−3.2±1.7 V/s) but not of subgroup 1 (−2.2±0.6 V/s) in the NoVFI episodes.
Thus, episodes could be distinctly divided into 3 groups: (1) VFI, (2) NoVFI subgroup 1, and (3) NoVFI subgroup 2. The propagation patterns (Figure 5⇓) as well as electrograms (Figure 6⇓) of cycle 1 from these 3 groups were nearly identical. Cycle 2 for groups 1 and 2 were also similar but not quite identical (Figure 5⇓). Group 3 had a different REA and shorter WCT than did groups 1 and 2.
Similarity of First Postshock Cycle Regardless of Shock Outcome
Our major finding is that the first cycle after a ULV50 shock that induces VF cannot be distinguished from that after a shock of identical strength and timing that does not induce VF. This finding was apparent in the activation sequence animations and in the similarity of the following first-cycle variables: (1) SEA repeatability, (2) postshock interval, (3) WCT, (4) dV/dt at the SEA, (5) similarity function, and (6) temporal lag.
Most studies reporting differences of activation sequences and dispersion of refractoriness between VFI and NoVFI episodes used multiple shock strengths and coupling intervals.6 7 Those studies found that NoVFI episodes correlated with a lower dispersion of refractoriness immediately after the shock, whereas VFI episodes correlated with a greater dispersion of refractoriness. However, in those studies the shock strengths of NoVFI episodes were higher than those of VFI episodes. Thus, absence of VF induction may not have been secondary to a lower dispersion of refractoriness; rather, both lack of VF induction and a lower dispersion of refractoriness could have been secondary to higher shock strength. To evaluate this possibility, we kept shock strength and timing constant. Our results suggest that when shock strength and timing are constant, differences in the dispersion of refractoriness are not large enough to cause measurable differences in postshock activation sequences and potentials. However, differences may exist immediately after the shock that are too small to be detected. These very small differences, according to the Chaos theory,8 could cause prominent differences after several cycles.
Association of Overlapping Cycles With VFI
Although differences between VFI and NoVFI episodes were first seen at cycle 2, a prominent distinction was not universally seen until cycle 3. This is because cycle 2 in 1 NoVFI subgroup behaved similarly to cycle 2 in the VFI episodes. However, no overlapping cycles were present in either No-VFI subgroup, whereas they were always present in the VFI group. Overlapping cycles may not be the direct cause of VF induction but may be a marker for short ICIs and long WCTs that are responsible for unstable reentry and VF induction. However, these results imply that at least 3 ectopic cycles with overlap by the third cycle may be required to initiate VF. If so, a method to halt the initiation of ectopic cycles in the REA could prevent VF even if it is applied as late as the third postshock cycle. Recent defibrillation9 and VF induction studies10 support this hypothesis.
Implications for Mechanism of VF Induction
It has been proposed that VF occurs by 2 mechanisms, an initiating mechanism that may involve ectopic unifocal impulses11 12 and a maintaining mechanism that involves reentry.13 Our results are consistent with reentry as a consequence of the initial accelerating, overlapping cycles observed in the VFI episodes. The rapid activation rate that led to the overlap of wave fronts also probably caused action potential duration to shorten and dV/dt to slow in the second to fifth cycles,13 possibly leading to unidirectional block, wave front fractionation, and reentry. In addition, the degeneration from the first postshock activation to VF could also be due to the nonuniform recovery of excitability of cardiac muscle after the premature stimulation of the first few postshock cycles.14
Although we did not observe epicardial reentry, intramural reentry cannot be ruled out because we did not record transmurally. The similarity of the first postshock cycle was based on epicardial recordings; therefore differences may have existed intramurally. The first postshock cycles could have differed in ways too small to be detected in our maps. Epicardial reentry may have been missed because it was too small to be detected by electrodes 4 mm apart or because part of the pathway consisted of activations that were so slow and small that the recordings did not meet our activation criterion.
The first postshock cycle has the same epicardial activation sequence regardless of shock outcome (VFI vs NoVFI), which suggests that large global differences in conduction or tissue refractoriness caused by the shock are not always the primary factors determining VF induction. Unidirectionally propagating epicardial activation followed by several ectopic, radially spreading impulses precedes VF. A progressive decrease in ICI and increase in WCT, resulting in overlapping cycles, heralds VF initiation.
Six pigs (weight 30 to 35 kg) of either sex were anesthetized, monitored, and maintained under physiological conditions as described previously.15 The chest was opened through a median sternotomy, and the heart was suspended in a pericardial cradle.
Two catheter-mounted, platinum-coated titanium coil electrodes delivered S2 shocks (CPI-Guidant Corp). A 34-mm electrode was inserted into the RV apex (Figure 7A⇓) and served as the cathode for the first phase of the S2 shock. A 68-mm electrode was positioned at the junction of the superior vena cava and right atrium. An electrode at the catheter tip was used for S1 pacing. Unipolar pacing electrodes were sutured to the RV outflow tract and anterobasal LV free wall.
The elastic sock had 14 rows of electrodes15 and was pulled over the ventricles to record unipolar epicardial electrograms (Figure 7B⇑). Each silver–silver chloride electrode was 1 mm in diameter and ≈4 mm from its neighbors. Two 3-mm-diameter, silver–silver chloride disk electrodes were sutured to the aortic root 5 mm apart to serve as reference for the unipolar recordings and ground for the mapping system.
Electrograms were bandpass filtered between 0.5 and 500 Hz and recorded at 2 kHz with 14-bit precision.16 Five milliseconds before each shock, amplifier coupling was changed from AC to DC, and amplifier gains were decreased. Approximately 9 ms after the shock, initial conditions were restored.
Initially, normal sinus rhythm and pacing from each pacing electrode were recorded to orient the mapping array. The intrinsic R-R interval and diastolic pacing threshold (DPT) were measured. S1 stimuli were constant-current, unipolar, 5-ms, monophasic pulses. S2 shocks were biphasic truncated exponential single-capacitor waveforms (Figure 7C⇑) (Ventritex Corp, HVS-02). Delivered voltage and current were displayed and total delivered energy was calculated by a waveform analyzer (DATA 6100, Analogic Inc).
Ten S1 stimuli were delivered 3 times, and the average coupling interval (CI) between the last S1 and the beginning, peak, and end of the T wave in limb leads I or II were determined.17 These intervals were recalculated every 5 VF episodes.
Ten S1 stimuli at 5 to 10 times cathodal DPT were delivered at an interval of 300 ms or 80% of the intrinsic R-R interval, whichever was shorter. S2 leading edge voltage was initially 500 V. The first S2 was delivered at the peak of the T wave. Subsequent S1- S2 CIs scanned the T wave in 10-ms steps. For example, if the initial CI was 200 ms, then subsequent CIs were 210, 190, 220, 180 ms, and so forth.
ULV shock strength was determined by the use of a modified up-down protocol with 40-V and 20-V steps, as described previously.17 Successive shocks were separated by 15 seconds. The lowest shock strength that did not induce VF at any CI was defined as the ULV.17 Next, the T wave was scanned with a shock 10 V below the ULV (sULV) in 10-ms steps, starting at the last S1- S2 CI that induced VF during the ULV determination. The T wave was scanned, alternately increasing and decreasing the S1- S2 interval until all CIs inducing VF were determined. The mean CI at which VF was induced with sULV shocks was used as the S1-sULV CI (sCI).
Mapping of VF Induction
Activations before and after 10 shocks of sULV strength delivered at the sCI were mapped. Ten seconds after a shock induced VF, a 20- to 30-J rescue shock was delivered. If the percentage of VFI was not 40% to 60%, sULV strength was increased or decreased in 10-V steps and a new sCI was determined. Then, 10 shocks at the new sULV strength were delivered at the new sCI. This protocol was repeated until the incidence of VFI episodes was 40% to 60%, which indicated that S2 shock strength was approximately ULV50. Only this last group of 10 shocks were analyzed. VF episodes were at least 4 minutes apart. At the end of the study, the animal was euthanized by VF.
Activations were analyzed by animating the dV/dt of the unipolar electrograms on a polar projection of the ventricular epicardium (Figure 7D⇑) on a computer.15 The dV/dt was computed with the use of a 5-point digital filter, and activation was identified when dV/dt was less than or equal to −0.5 V/s.15
The preshock interval was defined as the time from the beginning of the last paced cycle to the beginning of the shock. The first 5 activation cycles after the shocks were analyzed. In NoVFI episodes, only ectopic cycles were analyzed, so <5 activations were sometimes examined. The site of earliest activation (SEA) of each cycle was defined as the location of the first electrode that detected activation. The intercycle interval (ICI) for cycle [n+1] was the interval between earliest activation for cycle [n] and earliest activation for cycle [n+1]. The ICI for cycle 1 was the postshock interval. Wave front conduction time (WCT) was the time between activation at the SEA and at the site of latest activation of each cycle. Overlapping cycles, that is, activation from 2 cycles on the epicardium simultaneously, were detected by dividing the WCT of cycle [n] by the ICI of cycle [n+1] (the overlapping index). Overlapping cycles were absent when the index was ≤1.
To compare the first postshock cycles (or the last paced cycles) of 2 shock episodes, we computed a similarity cross-correlation function,18 sij, where i and j identify the 2 episodes. Pi and Pj are matrixes containing the unipolar potentials during the 2 cycles.Each row (n) represents a recording electrode and each column (t) a time sample. Comparison began 20 ms after the shock (or 5 ms after the last S1 pacing for the last paced cycles) and ended when activation from the cycle was no longer recorded. Respective durations of the 2 cycles are Ti and Tj, n is the electrode number, and τ is a temporal lag. We found the maximum of this function, cij=max[cij(τ)] and the lag, τm, at which it occurred. sij was evaluated by normalizing cij: Although this method does not provide specific information about the activation sequence, it is a sensitive indicator of the similarity between 2 different cycles (sij=1, indicating an exact match) and of any temporal shift between the 2 cycles (indicated by a nonzero lag, τm). The similarity function and lag were computed for all 45 possible pairs of first postshock cycles in each animal.
The SEA of the first postshock cycle was determined from the animations of activation. SEAs of different shock episodes from an animal were considered to be in the same REA if they were at the same site or at the nearest neighbors of the site. For each animal, the percentage of first-cycle SEAs in the REA with the most SEAs and the percentage of subsequent-cycle SEAs in this first-cycle REA were determined. This parameter is called SEA repeatability for each cycle.
VFI and NoVFI episodes was compared by use of the Student’s t test for paired and unpaired data. Similarity of the first postshock cycle and of the last paced cycle from different shock episodes and of the dV/dt at the SEA of the postshock cycles were compared by use of 1-way ANOVA. When statistical significance was found, individual comparisons were performed with Fisher’s post hoc test. Values are shown as mean±SD. Differences were considered significant at a level of P≤0.05.
This study was supported in part by National Institutes of Health research grants HL-28429 and HL-42760.
The Methods section of this article can be found at http://www.circulationaha.org
- Received February 16, 1999.
- Revision received September 14, 1999.
- Accepted October 7, 1999.
- Copyright © 2000 by American Heart Association
Wiggers CJ, Wégria R. Ventricular fibrillation due to single, localized induction and condenser shocks applied during the vulnerable phase of ventricular systole. Am J Physiol. 1940;128:500–505.
Han J, Moe GK. Nonuniform recovery of excitability in ventricular muscle. Circ Res. 1964;14:44–60.
Shibata N, Chen P-S, Dixon EG, Wolf PD, Danieley ND, Smith WM, Ideker RE. Influence of shock strength and timing on induction of ventricular arrhythmias in dogs. Am J Physiol. 1988;255:H891–H901.
Chen P-S, Wolf PD, Dixon EG, Danieley ND, Frazier DW, Smith WM, Ideker RE. Mechanism of ventricular vulnerability to single premature stimuli in open-chest dogs. Circ Res. 1988;62:1191–1209.
Chen P-S, Shibata N, Dixon EG, Wolf PD, Danieley ND, Sweeney MB, Smith WM, Ideker RE. Activation during ventricular defibrillation in open-chest dogs: evidence of complete cessation and regeneration of ventricular fibrillation after unsuccessful shocks. J Clin Invest. 1986;77:810–823.
Kuo CS, Reddy CP, Munakata K, Surawicz B. Arrhythmias dependent predominantly on dispersion of repolarization. In: Zipes DP, Jalife J, eds. Cardiac Electrophysiology and Arrhythmias. Orlando, Fla: Grune & Stratton; 1985:277–285.
Kaplan DT, Goldberger AL. Chaos in cardiology. J Cardiovasc Electrophysiol. 1991;2:342–354.
Kenknight BH, Walker RG, Ideker RE. Dual shock defibrillation with a new lead configuration involving an electrode in the left posterior coronary vein. Pacing Clin Electrophysiol. 1998;21:806. Abstract.
Scherf D, Schott A. Extrasystoles and Allied Arrhythmias. London, UK: William Heinemann Medical Books Limited; 1973.
Sano T, Sawanobori T. Mechanism initiating ventricular fibrillation demonstrated in cultured ventricular muscle tissue. Circ Res. 1970;26:201–210.
Pastore JM, Girouard SD, Laurita KL, Rosenbaum DS. Mechanism linking T wave alternans to the genesis of cardiac fibrillation. Circulation. 1999;99:1385–1394.
Chattipakorn N, Kenknight BH, Rogers JM, Walker RG, Walcott GP, Rollins DL, Smith WM, Ideker RE. Locally propagated activation immediately after internal defibrillation. Circulation. 1998;97:1401–1410.
Smith WM, Wolf PD, Simpson E, Danieley ND, Ideker RE. Mapping ventricular fibrillation and defibrillation. In: Shenasa M, Borggrefe M, Breithardt G, eds. Cardiac Mapping. Mount Kisco, NY: Futura Publishing Co, Inc; 1993:251–260.
Huang J, Kenknight BH, Walcott GP, Rollins DL, Smith WM, Ideker RE. Effects of transvenous electrode polarity and waveform duration on the relationship between defibrillation threshold and upper limit of vulnerability. Circulation. 1997;96:1351–1359.
Stearns SD, David RA. Signal Processing Algorithms in Fortran and C. Englewood Cliffs, NJ: Prentice Hall Press; 1993.