Safety of Transvenous Atrial Defibrillation
Studies in the Canine Sterile Pericarditis Model
Background It is recognized that a ventricular vulnerability period exists during which atrial shock delivery may induce a ventricular tachyarrhythmia. This study was designed to define the zone in which the ventricles are vulnerable to induction of ventricular tachyarrhythmia during delivery of atrial shocks in the sterile pericarditis canine model of atrial fibrillation.
Methods and Results Two days after creation of sterile pericarditis, 24 dogs underwent either a four-part or five-part ventricular vulnerability protocol during which atrial shocks were delivered between transvenous catheters, one in the distal coronary sinus and one in the right atrial appendage. The protocol included part 1, shocks during induced atrial fibrillation; parts 2 through 4, shocks delivered synchronously with the last ventricular beat of one of the following three ventricular pacing protocols: constant ventricular rates (S1S1), short-long-short cycles (S1S2S3-V), and ventricular premature beats (S1); and part 5, shocks delivered synchronously with the last R wave resulting from an atrially paced short-long-short cycle (S1S2S3-A). Ventricular tachyarrhythmia was induced 122 times: 2 of 665 shocks in two dogs in part 1, 29 of 786 shocks in nine dogs in part 2, 67 of 734 shocks in 15 dogs in part 3, 24 of 919 shocks in five dogs in part 4, and none in part 5. All ventricular proarrhythmia resulted from shocks delivered during the T wave of a preceding ventricular beat. No episodes of ventricular tachyarrhythmia were induced by atrial shocks synchronized to R waves with the previous RR at intervals above the QT+60 ms interval (absolute interval >320 ms), with one exception, at the QT+100 ms interval (absolute interval 360 ms).
Conclusions With transvenous electrode catheters used to deliver atrial shocks, life-threatening ventricular rhythms were induced but were limited to a specific zone defined by the QT interval.
There is a growing interest in transvenous atrial defibrillation as a treatment for recurrent or chronic AF.1 2 3 4 5 6 7 8 For this technique to become acceptable as a clinical tool, it must be shown to be effective, reliable, easy to use, tolerable, and safe. Clearly, a major potential acute risk of this procedure is the inadvertent induction of VT or VF. The existence of a portion of the cardiac electrical cycle during which the ventricles are vulnerable to the initiation of VT by a shock has been well described.9 10 Others have begun to describe this zone of ventricular vulnerability as it relates to atrial defibrillation.1 2 6 11 In this study, we further defined this zone of vulnerability in an attempt to better understand its characteristics and to define zones of safety for delivery of shocks to accomplish transvenous atrial defibrillation.
This study of adult mongrel dogs was performed in accordance with guidelines specified by our Institutional Animal Care and Use Committee, the American Heart Association Position on Research Animal Use, the current Public Health Service Policy on Humane Care and Use of Laboratory Animals, and the American Association for Accreditation of Laboratory Animal Care.
Preparation of the Animals
Sterile pericarditis was created in 24 adult male mongrel dogs (weight, 20 to 30 kg) as previously described.12 For this surgery, the animals underwent sodium pentothal induction and isoflurane anesthesia. Pairs of epicardial stainless steel wire electrodes, insulated except for the distal 5 mm, were placed on Bachmann’s bundle, the right atrial appendage, and the right ventricular apex (Davis and Geck). Through a cutdown on the right external jugular vein, two customized defibrillation electrode catheters, each with a 60-mm silver-plated stainless steel electrode coil at the tip, were placed by manual palpation, one in the distal coronary sinus under the left atrial appendage and one in the right atrial appendage (Fig 1⇓). The leads and lead configuration used were those that should yield the lowest energy requirements on the basis of our prior work with this model13 and the work of others.7 14 A commercially available bipolar endocardial lead with passive fixation was placed via the right jugular vein into the right ventricle (Biotronik). The catheters were sutured in place at their site of insertion into the jugular vein, and the external tips were placed in a subcutaneous pocket in the neck. Chest radiographs were performed to confirm the cardiac location of the catheters. The animals were given buprenorphine (0.3 mg SC q 12 hours) for postoperative pain management and oral ampicillin with clavulanic acid for infection prophylaxis.
On postoperative day 2, the dogs were given sodium pentobarbital (20 mg/kg) as sedation. After radiographs were taken to confirm the position of the defibrillation catheters, the animals were intubated and ventilated with 100% oxygen and 1 to 2 L/min of isoflurane as needed to maintain general anesthesia. The femoral artery was cannulated for blood pressure monitoring and blood sampling. Blood gases, sodium, potassium, calcium, glucose, and magnesium were monitored at the beginning and end of the experimental protocol and as indicated. Adjustments in ventilation and intravenous replacement of electrolytes and bicarbonate were performed to maintain normal pH, Paco2, Pao2, and serum electrolyte concentrations.
With a surgical plane of general anesthesia having been successfully obtained, the subcutaneous pocket was opened, lead ends were exposed, and the transvenous catheters were connected to both a customized, isolated interface/amplifier (InControl) and an Electronics-for-Medicine switched-beam oscilloscopic recorder (model VR-16, Honeywell). During each study, ECG lead II, femoral arterial blood pressure, and bipolar electrograms from the catheter electrodes and the stainless steel epicardial wire electrodes, which were placed at the three selected sites, were monitored on the Electronics-for-Medicine VR-16, recorded on magnetic tape (Honeywell model 101 Tape Recorder/Reproducer), and digitized and stored on a Macintosh IIfx computer (Apple Computer) with custom LABVIEW software (National Instruments).
Determination of ADT
Stable AF was induced via rapid atrial pacing through either Bachmann’s bundle or right atrial electrodes at a CL <100 ms with a stimulus strength of 20 mA and a stimulus pulse width of 1.8 ms. ADTs were determined as described by Ortiz et al13 and were defined as the lowest-energy shock that would yield a successful conversion of AF to sinus rhythm at least 10% of the time and not more than 90% of the time (Fig 2⇓). The initial peak leading-edge voltage was 50 V. This initial shock intensity was selected because it was suspected that it would be uniformly ineffective. If the shock failed to defibrillate the atria, the shock intensity was increased in 10-V steps until successful defibrillation was achieved. The percent success of the lowest successful intensity was then determined by attempts to defibrillate the atria 20 times at this voltage (after each successful shock, AF was reinduced). If the percent success was <10%, the shock intensity was increased by 10 V; if the percent success was >90%, the shock intensity was decreased by 10 V, and an additional 20 shocks were delivered to determine the percent success at the new intensity. If the percent success was between 10% and 90%, this voltage was considered to be the ADT.
All shocks were delivered from a custom defibrillator (InControl) and were biphasic, with both phases 3 ms in duration. Shock delivery was controlled by customized LABVIEW software on the Macintosh IIfx computer. For the initial phase of each shock, the distal coronary sinus catheter served as the cathode and the right atrial appendage catheter served as the anode. For this portion of the protocol, all shocks were delivered 20 ms after an R wave sensed through the transvenous right ventricular catheter.
The QT interval was measured in a standard fashion from surface ECG lead II at the beginning of ventricular vulnerability testing while the atria were paced at a CL of 300 or 400 ms. Ventricular vulnerability was tested in a four-part protocol in the first 10 animals and in a five-part protocol in the remaining 14. The order of the parts of the protocol was determined by randomization at the beginning of the study. At selected intervals after a previous R wave (see below), an atrial shock of 300 V was delivered 2 ms after a sensed R wave. In the first 10 dogs, atrial shocks were delivered during induced AF or during ventricular pacing in the presence of the underlying spontaneous supraventricular rhythm. In the remaining 14 dogs, the study protocol included the addition of an atrial shock delivered during atrial pacing. Ventricular pacing was performed via the apical epicardial wire electrodes and atrial pacing either via the right atrial appendage or Bachmann’s bundle wire electrodes, both at twice diastolic threshold. Both the pacing and delivery of shocks were controlled by a customized LABVIEW software program running on the MacIntosh IIfx computer.
Three hundred volts was used as the defibrillation voltage in an attempt to maximize the ventricular vulnerability. The following is the reasoning behind this: Chen et al15 found that the range of energies applied directly to the ventricles that resulted in VF was 0.025 to 6.6 J. If shocks are applied between the right atrial appendage and the distal coronary sinus, maximum energy will be applied between these two locations, with lesser energies transmitted to other portions of the heart. Therefore, to maximize the risk to the ventricles, we chose to use the maximum voltage output available on the atrial defibrillator used, namely 300 V (≈3.5 J), which would be expected to yield energies within the vulnerable range. This concept is also consistent with the work of Dunbar et al,2 who found that 2.4% of shocks delivered to the atria that were not synchronized to an R wave resulted in VF. The energies used in Dunbar’s study were between 0.01 and 5 J, the shocks that resulted in VF were >0.25 J, and the rate of induction increased with increasing energies.
Vulnerability Part 1: Studies During AF (RR Protocol)
In this part of the protocol, atrial shocks were delivered during induced, sustained AF or during continuous rapid atrial pacing to ensure sustained AF (with the atria paced from Bachmann’s bundle electrodes at CLs of ≈100 ms). The ventricular RR intervals were continuously sensed. Shocks were delivered only when synchronized to R waves that were within a specified range of intervals from the preceding R wave (Fig 3⇓). These intervals were specified in windows of 50 ms in the first 10 dogs. For example, if the defibrillator was set to deliver shocks synchronously with an R wave that was within a window 200 to 250 ms from the previous R wave, the defibrillator would discharge on the first R wave that followed a preceding R wave if the interval fell within this 50-ms window. In an effort to increase the incidence of ventricular tachyarrhythmia and to better define the zone of vulnerability, windows of 20 ms were specified in the final 14 dogs. For each specified window, a shock was delivered after each of four separate RR intervals that fell within the selected window unless VT or VF was induced, in which case no further shocks were delivered during that window. If no shock was delivered within a 2-minute period of scanning for RR intervals within a specified window, this window was no longer tested. To ensure that all possible windows that might produce a defibrillator shock were tested for a given animal, at least two windows longer than that containing the longest RR interval (two intervals that did not result in a shock) were scanned, and one window without a shock at the shortest RR interval was scanned.
Vulnerability Parts 2 Through 5
Intervals Tested: Relation to QT Interval
Recall that the QT interval was determined from ECG lead II during a paced atrial rhythm at a CL of 300 or 400 ms. Because the QT interval during ectopic ventricular activation is longer than that during normal ventricular activation from a supraventricular rhythm, we began vulnerability testing in the subsequent parts of the protocol at intervals 50 to 100 ms above the previously determined QT interval. For the first three animals, the longest S1S1 interval tested in vulnerability part 2 (ventricular pacing at constant rates), the longest S2S3 tested in vulnerability part 3 (short-long-short ventricular cycles), and the longest S1 interval tested in vulnerability part 4 (ventricular premature beats) were the QT+50 ms. However, during part 3 testing in the second animal, there was induction of ventricular tachyarrhythmia at all S2S3 intervals except the QT+50 ms interval. Therefore, during this testing there was only one interval defining the “safe zone.” To increase the number of intervals defining this zone, for the remaining 21 animals, the longest pertinent intervals tested in each of the parts was increased to the QT+100 ms.
Vulnerability Part 2: Studies During Ventricular Pacing at Constant Rates (S1S1 Protocol)
In this portion of the protocol, a regular rapid ventricular rhythm was simulated by pacing the ventricles with a train of eight stimuli (S1), with delivery of the atrial shock synchronized to the ventricular depolarization resulting from the eighth stimulus of the train (Fig 3⇑). For the first three animals, the longest S1S1 interval tested was the QT+50 ms, and the S1S1 interval was decreased in 10-ms decrements until there was failure of consistent ventricular capture. In the subsequent 21 animals, the longest S1S1 interval tested was the QT+100 ms. The S1S1 interval was decreased by 20 ms until the QT+60 ms was reached; then the S1S1 interval was decreased by 10 ms until loss of consistent ventricular capture. Each sequence of eight S1s with a shock delivered on the ventricular depolarization resulting from the eighth S1 was repeated four times for each S1S1 interval unless VT or VF was induced. If a ventricular tachyarrhythmia occurred, no further pacing and shock delivery was performed at that interval, and the next shortest interval was tested. Each S1S1 interval was also tested four times without delivery of a shock to ensure that the pacing itself did not induce ventricular tachyarrhythmia.
Vulnerability Part 3: Short-Long-Short Ventricular Cycles During Ventricular Pacing (S1S2S3-V Protocol)
This part of the protocol simulated a sequence thought to be particularly associated with the initiation of ventricular tachyarrhythmia, namely a short-long-short sequence.16 All pacing was ventricular for this part of the protocol. After a train of eight stimuli (S1) at a CL of 300 ms, an S2 equal to the sinus node recovery time minus 20 ms was delivered followed by an S3 delivered at varying CLs (Fig 3⇑). (Sinus node recovery time is the longest interval between the last beat of the train of eight stimuli at a CL of 300 ms and the next spontaneous A wave.) In the first two cases, the sinus node recovery times were 610 and 760 ms, respectively. It was decided that using S1S2 intervals >600 ms would lessen the clinical applicability of this portion of the study. In subsequent animals, if the sinus node recovery time was >600 ms, the S1S2 interval was set to 600 ms. In the first three dogs, the longest CL for the S3 was the QT+50 ms, and the CL was decreased in 10-ms decrements until failure of ventricular capture. In the remaining 21 dogs, the longest CL for the S3 was the QT+100 ms. The CL was decreased in 20-ms decrements until the QT+60 ms was reached, after which it was decreased in 10-ms decrements until failure of ventricular capture. This sequence was repeated four times, with the atrial shock delivered synchronized with the V3, ie, the response to S3. If ventricular tachyarrhythmia was induced, no further shocks were given at that S2S3 interval. Each pacing sequence was also repeated four times without the delivery of a shock to ensure that pacing itself did not induce ventricular tachyarrhythmia.
Vulnerability Part 4: Premature Ventricular Beats (S1 Protocol)
This part of the vulnerability protocol mimicked the delivery of an atrial defibrillation shock during a ventricular premature beat (Fig 3⇑). After R waves were sensed, single paced beats (S1) were delivered at intervals starting at the QT+50 ms in the first three dogs and at the QT+100 ms in the remainder. Between the QT+100 ms and the QT+60 ms, the S1 interval was decreased in 20-ms decrements, and when less than the QT+60 ms, it was decreased in 10-ms decrements until loss of ventricular capture. An atrial shock was delivered synchronized with the R wave resulting from each S1, and the S1 with an atrial shock was delivered four times at each interval unless ventricular tachyarrhythmia was induced, in which case no further testing was done at that interval.
Vulnerability Part 5: Short-Long-Short Ventricular Cycles During Atrial Pacing (S1S2S3-A Protocol)
After analysis of the data from the first 10 dogs, it was evident that delivery of an atrial shock during the S1S2S3-V (part 3) protocol was associated with the greatest number of induced ventricular tachyarrhythmias. Because this ventricular pacing protocol most likely enhanced ventricular dispersion of refractoriness, it was decided to add a short-long-short (S1S2S3) atrially paced sequence to the protocol to achieve short-long-short ventricular cycles with a supraventricular activation sequence. Thus, the vulnerability part 3 protocol (short-long-short ventricular cycles) was also performed with atrial pacing in the last 14 animals, with the atrial shock synchronized to the R wave resulting from the S3 (Fig 3⇑).
To supplement observational methods used in this study, three statistical methods were used so that specific conclusions could be drawn about the relative incidence of ventricular proarrhythmia observed in each part of the study and so that each part (1 through 5) could be compared with each other part with respect to arrhythmogenic potential. Each of the methods required certain assumptions and data transformations to permit the data to be analyzed. Therefore, to be conservative, we required that all the statistical methods used to compare the several parts of the protocol agree as to the acceptance or rejection of the null hypothesis at the specified value of P=.05 before conclusions would be drawn regarding the result. The statistical methods used in this study may also be useful in future studies comparing different physiological perturbations (eg, drug effects or ventricular ischemia) with respect to the relative risks of ventricular proarrhythmia induced by atrial defibrillation shocks.
The methods used to analyze our data were the Kaplan-Meier nonparametric survival analysis, proportional-hazards analysis, and parametric methods. This survival analysis reflects the fact that the risk of ventricular tachyarrhythmia induction is proportional to the QT+X. The Weibull extreme-value parametric survival functions were fit by the SAS Lifereg procedure, and Kaplan-Meier nonparametric survival estimates were fit by the SAS Lifetest procedure.17 In addition, the three protocols were tested to see whether they were significantly different regarding the incidence of induction of ventricular tachyarrhythmia with nonparametric and parametric χ2 tests. Finally, to assess safety, 90%, 95%, and 99% predicted survival was estimated under each protocol, and 95% CIs for QT+X were given.
Atrial Defibrillation Threshold
Correct electrode catheter positions were confirmed before postoperative studies in 20 dogs. In 4 dogs, however, the coronary sinus catheter electrode had migrated as follows: in 3 animals this catheter electrode was found in the right atrium, and in 1 animal it was found in the right ventricular outflow tract. These four catheter electrodes were successfully repositioned, and the study protocol was completed.
The mean ADT was 80±35 V (±SD), similar to those reported by others.2 4 5 7 11 At the ADT, the mean energy output was 0.26±0.035 J, and the mean impedance was 40±7 Ω. The mean percent success at this energy level was 42±22%. No induced ventricular arrhythmia occurred during this portion of the study, which included 652 shocks.
No ventricular tachyarrhythmia was induced by a pacing protocol performed alone, ie, in the absence of an atrial shock. In total, 3580 atrial shocks were delivered to 24 animals during vulnerability testing with shocks of 300 V. The average QT interval to which shock timing was referenced was 236±21 ms (±SD) (range, 190 to 280 ms). VF or polymorphic VT, defined as VT with multiform QRS complexes in which the tachycardia spontaneously terminated, was induced a total of 122 times in 15 animals (Figs 4⇓ and 5⇓). In 9 animals, no ventricular arrhythmia was induced. All episodes of ventricular tachyarrhythmia resulted from shocks that were delivered on the T wave of the preceding beat. (Actually, this could not be confirmed in 1 animal; see below. In this animal, however, the interval between the beat to which the atrial shock was synchronized and the previous R wave was consistent with a shock on the T wave.) No monomorphic VT was induced.
Vulnerability Part 1: Studies During AF (RR Protocol)
In the 24 animals, 665 atrial shocks were delivered during part 1 (shocks delivered during AF), with shocks being delivered after previous RR intervals ranging from the mean shortest RR interval of 203±34 ms (±SD) (range, 127 to 255) to the mean longest RR interval of 535±132 ms (range, 317 to 793). There was induction of ventricular tachyarrhythmia in 2 animals, after RR intervals of 134 and 180 ms, respectively, or 2 of 665 (incidence, 0.3%). The interval of 134 ms was shorter than expected for a ventricular response to AF. Unfortunately, this episode inadvertently was not recorded on FM magnetic tape, so confirmation of the mechanism of initiation of VF was not possible. We nevertheless felt obligated to report it despite the uncertainty about the mechanism. Also in these two animals, windows shorter than those containing 134 and 180 ms, respectively, were not tested.
Vulnerability Part 2: Studies During Ventricular Pacing at Constant Rates (S1S1 Protocol)
During part 2, atrial shocks were delivered during a paced ventricular rhythm at constant CLs. In total, 786 atrial shocks were delivered to 24 animals, resulting in 29 episodes of induced ventricular tachyarrhythmia in 9 of the animals. All episodes occurred with atrial shocks delivered between the QT−60 ms and the QT+60 ms intervals (between absolute S1S1 intervals of 170 to 310 ms) (Fig 6⇓). Of the 9 animals who had tachyarrhythmia, 7 had more than one episode of induction. These episodes occurred at consecutive intervals tested in 6 animals, denoting a “zone of vulnerability.” In one animal (dog 16), this zone of vulnerability was interrupted by one interval at which no tachyarrhythmia was induced (Fig 6⇓). Of interest, although an episode of tachyarrhythmia was induced at almost every interval tested inside each animal’s zone of vulnerability, it did not always occur with either the first shock or every shock delivered at each interval (recall that up to a total of four shocks were delivered at each interval).
Vulnerability Part 3: Short-Long-Short Ventricular Cycles During Ventricular Pacing (S1S2S3-V Protocol)
In part 3, atrial shocks were delivered after a short-long-short sequence of ventricular pacing. In total, 734 shocks were delivered to 24 animals. This resulted in 67 episodes of induced ventricular tachyarrhythmia in 15 of the animals (Fig 7⇓). The mean S1S2 (long) interval was 567±75 ms (±SD). The mean shortest S2S3 tested was 230±41 ms (range, 150 to 280 ms). The induction of ventricular tachyarrhythmia occurred when the atrial shock was delivered at S2S3 intervals between the QT−60 ms and the QT+60 ms intervals (between absolute intervals of 170 to 310 ms) except for one outlier animal in which ventricular tachyarrhythmia was induced at intervals including the QT+100 ms (absolute interval, 360 ms) (dog 20). As in the S1S1 protocol, the episodes occurred at consecutive intervals tested in 11 animals, defining a zone of vulnerability. In 2 animals, this zone was interrupted by three intervals free of arrhythmia induction. In another 2 animals, this zone was interrupted by one interval (Fig 7⇓). Once again, although an episode of tachyarrhythmia was induced at almost every interval tested in each zone of vulnerability, it did not always occur with either the first shock or every shock delivered at each interval (up to a total of four shocks were delivered at each interval). In animal 2, of the four intervals tested, only the longest interval did not result in ventricular tachyarrhythmia. In the other animals with induced ventricular tachyarrhythmia, there were at least two intervals free of ventricular tachyarrhythmia induction beyond the longest interval resulting in this induction except animal 20 (see above).
Vulnerability Part 4: Premature Ventricular Beats (S1 Protocol)
In part 4, atrial shocks were delivered during simulated ventricular premature beats. A total of 919 shocks were delivered to 24 animals. Twenty-four episodes of ventricular tachyarrhythmia were induced in 5 animals. This arrhythmia occurred when the atrial shock was delivered at intervals between the QT−70 ms and the QT+40 ms (absolute intervals, 210 to 320 ms) as in the S1S1 and S1S2S3-V protocols; the episodes occurred at consecutive intervals tested in 3 animals, defining a zone of vulnerability. In 1 animal, this zone was interrupted by one interval free of arrhythmia induction. In another animal, this zone was interrupted by six intervals free of arrhythmia induction (Fig 8⇓). Once again, each episode of tachyarrhythmia was not always induced with the first or every shock delivered at each interval (Fig 8⇓).
Vulnerability Part 5: Short-Long-Short Ventricular Cycles (Atrial Pacing) (S1S2S3-A)
In part 5, a total of 467 shocks were delivered, and no episodes of ventricular arrhythmia were induced. The mean shortest RR interval in response to the atrially paced S2S3 was 281±43 ms (mean±SD) (range, 250 to 320 ms). (For these animals, the mean shortest S2S3 interval tested in vulnerability part 3, ventricularly paced short-long-short cycles, was 230±41 ms; range, 150 to 280 ms. However, there were intervals that were tested in both vulnerability parts 3 and 5 at which there was tachyarrhythmia induction during vulnerability part 3 but not during vulnerability part 5.) Although premature atrial stimuli were introduced until the atrial effective refractory period was reached, an atrial shock was never delivered on the initial portion or the peak of the T wave resulting from the S2. Occasionally, an atrial shock was delivered on the terminal portion of the T wave resulting from the S2.
Comparison of Vulnerability Protocols
The predicted survival proportions from vulnerability parts 2 through 4 are shown in Fig 9⇓. Part 1, studies during AF, and part 5, atrially paced short-long-short sequences, were not included in the statistical comparisons because of a lack of variability in the response. (Only two episodes of ventricular tachyarrhythmia were induced in vulnerability part 1, and no episodes were induced during vulnerability part 5.) In Fig 9⇓, Kaplan-Meier nonparametric estimates are shown with a dotted line and the parametric survival estimates (Weibull model) are shown with a solid line. There is good agreement between the Kaplan-Meier estimates of survival and the parametric estimates of survival. At a fixed QT+X interval, vulnerability part 4 had the highest survival followed by vulnerability part 2. Vulnerability part 3 had the lowest survival. The survival analysis compared the three protocols and found them significantly different (all df=2; log-rank χ2=12.1, P<.0024; Wilcoxon χ2=12.2, P<.0023; Weibull χ2=10.5, P<.0052). This multiple comparison analysis showed that the paced ventricular short-long-short cycles had the highest risk for ventricular tachyarrhythmia (P<.05 for all these methods).
Under the Weibull model, it was possible to predict what value of QT+X would yield a fixed proportion of survival, the results of which are shown in the Table⇓. For example, with vulnerability part 4, ventricular premature beats, if 99% safety is desired, intervals of QT+63.9 ms or greater should be used. Under vulnerability part 2, ventricular pacing at constant rates, intervals of QT+107.7 ms should be used. Under vulnerability part 3, ventricularly paced short-long-short sequences, intervals of QT+159.2 ms should be used. These safety estimates were not done for parts 1 and 5 because of the lack of variability of response in these parts (only two episodes of ventricular tachyarrhythmia were induced in vulnerability part 1, and no episodes were induced in vulnerability part 5).
These data from structurally normal canine hearts with sterile pericarditis show that the atria can be readily defibrillated with a low-energy shock (mean, 80 V, 0.26 J). The data also show that under some circumstances during atrial shock delivery, unintended induction of VF is possible, but if atrial shock delivery is limited to certain intervals after a preceding R wave and thereby shocks delivered synchronously with the T wave of a ventricular beat are avoided, no ventricular arrhythmia should occur.
Atrial Defibrillation Thresholds
The ADTs we obtained in this study were similar to those obtained by us13 and others2 4 5 7 11 in animal models of atrial defibrillation using transvenous systems. (Recall from the “Methods” section that we did not require 100% shock efficacy.) The philosophy of ADT determination in this study was the same as that described by Ortiz et al.13 Briefly, this philosophy considers that AF is virtually never a clinical emergency and that AF is usually clinically well tolerated for relatively long periods of time. Therefore, it is not necessary that atrial defibrillation be done as soon as AF is recognized nor that it be achieved with each delivered shock. In short, one should be able to minimize and perhaps even eliminate patient discomfort by using a defibrillation shock intensity with a <100% success rate and perhaps even a <50% success rate (recall that the threshold was determined for a success rate between 10% and 90%). The average success rate of 42±22% (±SD) obtained in this study is a rate in keeping with this philosophy. It was also noteworthy that during this testing, when the ventricles were activated normally, ie, as expected during a supraventricular rhythm, despite the very large number of threshold shocks delivered, not a single ventricular arrhythmia was induced. These data are particularly relevant because this part of the study most closely simulates clinical conversion of AF.
Previous studies have reported a low occurrence of VF as a result of transcatheter atrial defibrillation, and in most of these studies, the episodes resulted from shocks that were not synchronized with the R wave.1 2 11 In this study in the sterile pericarditis model, a model in which the ventricles are essentially normal (only the atria have florid pericarditis12 ), the incidence of induction of ventricular tachyarrhythmia was also low. Most important, VF or polymorphic VT was induced during zones of vulnerability that could be defined with reference to the QT interval measured during a paced atrial rhythm with normal ventricular activation. This is consistent with the work of Dunbar et al,2 who found that all episodes of VF induced by atrial defibrillation shocks delivered to the atria in dogs with sterile pericarditis occurred within 116 to 180 ms of the onset of the QRS complex. The work published by Ayers et al11 began to explore the proarrhythmic effects of atrial defibrillation shocks synchronized to an R wave. In this study, we have further defined this vulnerability by defining zones of vulnerability and by exploring the difference in the incidence of induction of tachyarrhythmias found between pacing protocols. Our work further differed from that of Ayers et al in that we used the canine sterile pericarditis AF model as opposed to a sheep model that used β-methacholine as needed.
Also, there was a conspicuous difference between the incidence of induction of ventricular tachyarrhythmia with an atrial shock when the ventricles were activated normally (ie, during the ADT testing, the RR protocol, and the S1S2S3-A protocol) and when they were activated ectopically (ie, the S1S1, S1S2S3-V, and S1 protocols). Within the ectopic ventricular activation group, the short-long-short sequence (ventricular pacing, S1S2S3-V) was significantly more strongly associated with the induction of ventricular arrhythmia. This coincides with the observations by others that the short-long-short CL sequence is associated with an important incidence of the initiation of ventricular tachyarrhythmia.16 It has been postulated that short-long-short ventricular CL sequences are associated with the development of ventricular tachyarrhythmias because of an increase in the dispersion of ventricular refractoriness that results from this sequence,16 and ventricular pacing increases the inhomogeneity of ventricular repolarization beyond that produced by conduction via the His-Purkinje system. In this regard, it is of interest that no ventricular arrhythmia was induced with a short-long-short ventricular CL sequence produced with atrial pacing. When the short-long-short CL sequence vulnerability studies during ventricular pacing were compared with those during atrial pacing, the mean shortest preceding RR interval was shorter during the former (230±41 ms) than during the latter (281±43 ms), although the upper range of the former overlapped with the lower range of the latter (150 to 280 ms versus 250 to 320 ms, respectively). There were intervals that were tested in the short-long-short CL sequence vulnerability studies with both ventricular and atrial pacing at which there was tachyarrhythmia induction with ventricularly paced sequences and not atrially paced sequences. Despite the S2S3 interval being decreased when atrially paced short-long-short CL sequences were delivered until the atrial effective refractory period was reached, the atrial shock delivered synchronously with the R wave resulting from the S3 was never delivered on the initial portion or the peak of the T wave resulting from the S2. This may have been due to conduction delay in the AV node.
No episodes of VT or VF were induced by an atrial shock if the preceding RR interval was ≥QT+60 ms (absolute interval, ≥320 ms, which is a relatively short interval) except for one animal (absolute interval, 340 ms), no matter which protocol was used, despite use of protocols designed to maximize the risk of induction of ventricular arrhythmia. Recall that the QT interval was determined during baseline atrial pacing at a CL of 400 or 300 ms. An additional time period was added to this QT interval determined in this way to account for the fact that during ectopic ventricular activation (ie, ventricular pacing in our studies and, conceivably, spontaneous ventricular ectopy during clinical states), the QT interval would be longer and therefore associated with increased dispersion of refractoriness compared with that during a supraventricular rhythm with normal ventricular activation. Recall also that all episodes of ventricular tachyarrhythmia occurred with atrial shocks that were delivered during a T wave of a ventricular beat. This implies that transvenous atrial defibrillation can be done safely by timing the delivery of the shock properly, ie, taking into account the preceding RR interval, so that an atrial defibrillation shock of low energy can be delivered after the T wave of the preceding beat.
Selected Abbreviations and Acronyms
|ADT||=||atrial defibrillation threshold|
This study was supported in part by a grant from InControl, Inc, Redmond, Wash. We would like to thank José Ortiz, MD, for his technical support and Al M. Best, PhD, for his work with the statistics.
- Received May 16, 1996.
- Revision received February 24, 1997.
- Accepted February 28, 1997.
- Copyright © 1997 by American Heart Association
Nathan AW, Bexton RS, Spurrell RA, Camm AJ. Internal transvenous low energy cardioversion for the treatment of cardiac arrhythmias. Br Heart J. 1984;52:377-384.
Cooper RAS, Alferness CA, Smith WM, Ideker RE. Internal cardioversion of atrial fibrillation in sheep. Circulation. 1993;87:1673-1686.
Ferris LP, King BG, Spence PW, Williams HB. Effect of electric shock on the heart. Elec Eng. 1936;55:498-515.
Wiggers CJ, Wegria 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.
Ayers GM, Alferness CA, Ilian M, Wagner DO, Sirokman WA, Adams JM, Griffin JC. Ventricular proarrhythmic effects of ventricular cycle length and shock strength in a sheep model of transvenous atrial defibrillation. Circulation. 1994;89:413-422.
Chen PS, Shibata N, Dixon EG, Martin RO, Ideker RE. Comparison of the defibrillation threshold and the upper limit of ventricular vulnerability. Circulation. 1986;73:1022-1028.
SAS Institute Inc. SAS/Stat User’s Guide, Version 6, 4th ed, Volume 2. Cary, NC: SASG Institute, Inc; 1989.