Comparison of Upper Limit of Vulnerability and Defibrillation Probability of Success Curves Using a Nonthoracotomy Lead System
Background An upper limit to the strength of shocks that induce fibrillation during the vulnerable period, the upper limit of vulnerability (ULV), has been shown to exist in both humans and animals. The purpose of this study was to compare ULV and defibrillation (DF) probability of success curves for a clinically useful nonthoracotomy lead system.
Methods and Results Sixteen pentobarbital-anesthetized pigs were studied. Single-capacitor biphasic waveforms with both phases 5.5 ms in duration were used for ULV and DF testing. A right ventricular catheter electrode served as first-phase cathode and a superior vena cava catheter electrode coupled with a cutaneous R2 patch electrode served as common first-phase anodes. A pacing catheter was placed in the right ventricle to deliver a train of 15 S1 stimuli at a pacing interval of 250 to 300 ms. A ULV shock was delivered on the peak of the T wave as measured from the surface ECG; if ventricular fibrillation was induced, a DF shock was delivered after 10 seconds of fibrillation. Shock voltages were determined by an up-down protocol. Ventricular fibrillation was induced an average of 53 times in each animal. The composite data indicate that below V97, that is, the voltage that leaves the animal in normal sinus rhythm 97% of the time when delivered on the peak of the T wave or the voltage that defibrillates 97% of the time, ULV is lower than DF. ULV and DF became significantly correlated at V80 and maximally correlated at V97. Even at V97, however, ULV and DF differed by more than 100 V in 2 of the 16 animals.
Conclusions ULV approximately equaled DF at V97. This is fortunate because it is clinically important to set the device voltage at the uppermost portion of the probability of success curve. Estimating DF V97 from ULV V97 would reduce the number of fibrillation inductions needed to establish defibrillation shock strength requirements. However, the large difference between ULV V97 and DF in a few animals indicates that further improvement and testing of algorithms for determining ULV V97 must be developed before the technique is used clinically.
In addition to a defibrillation threshold (DFT), an upper limit of vulnerability (ULV), which can be defined as the shock voltage level above which ventricular fibrillation cannot be initiated during the vulnerable period, has been shown to exist in both animals and humans.1 2 3 4 5 6 7 8 9 10 Information regarding the relation between ULV and DFT may prove to be clinically useful when determining shock voltage requirements for automatic implantable cardioverter-defibrillator implantation. Several investigators have noted that a single shock strength delivered on the peak of the T wave may induce ventricular fibrillation on one attempt and fail to induce ventricular fibrillation on another attempt.1 2 7 This suggests that ULV data may be described with a probability of success curve,2 7 similar to defibrillation.11 12 The purpose of this study is to construct ULV and defibrillation (DF) probability of success curves and examine the relations between them at various probability of success levels.
This study was performed according to the “Position of the American Heart Association on Research Animal Use” adopted by the American Heart Association on November 11, 1984. Sixteen pigs of either sex (heart weight, 141.4±17.7 g) were studied. The animals were anesthetized with intravenous pentobarbital (30 to 35 mg/kg) and maintained with a pentobarbital infusion at a rate of 3 mg/kg per hour.13 Succinylcholine (1 mg/kg) was given at the time of anesthesia induction followed by 0.25 to 0.50 mg/kg per hour as needed for skeletal muscle paralysis.14 The pigs were intubated with a cuffed endotracheal tube and mechanically ventilated with a volume respirator (Harvard Apparatus). The left femoral artery was exposed and cannulated for hemodynamic monitoring and arterial blood sampling. Blood was drawn every 60 minutes, more frequently if needed, to monitor arterial pH, Pco2, Po2, and HCO3 as well as Na+, K+, and Ca2+. The ventilator was adjusted to keep oxygenation and acid/base status within normal limits; electrolytes were supplemented as needed. ECG leads were applied, and the ECG was continuously monitored (Lifepak, Physio-Control).
The pigs were positioned on their backs. The right and left internal and external jugular veins were exposed. A triple-lumen catheter (7F) was inserted for optimal intravenous access. Two catheter electrodes (2.95 cm2) were used (CPI). One catheter was placed in the superior vena cava and the other in the right ventricular apex. A 113-cm2 cutaneous R2 patch electrode (Darox Corp) was placed on the chest wall such that the lowermost anterior quadrant of the patch covered the point of maximal impulse. The right ventricular electrode served as cathode and the superior vena cava and cutaneous patch electrodes served as anode for the first phase of the biphasic waveform. Additionally, a pacing catheter was placed in the right ventricle (Fig 1⇓). Cutaneous defibrillation patch electrodes were placed on the right and left thorax to facilitate the delivery of an external rescue shock.
Upon completion of the study, the animals were killed with a potassium chloride injection. A median sternotomy was performed, the lead positions were verified, and the heart was removed and weighed.
ULV and DF Protocol
The pacing threshold current was determined at the beginning of the study using a step-up protocol. For the remainder of the study, the pacing stimulus current was set to twice the pacing threshold. The pacing interval was initially set at 250 ms but was decreased if necessary to eliminate competition between the pacing stimulus and the intrinsic rate of the heart. Chen et al2 found no significant difference in ULV50, that is, the shock strength inducing fibrillation 50% of the time, with varying pacing intervals.
The time from the pacing stimulus to the peak of the T wave was measured from an oscilloscope using limb leads II or III. In the first animal, the time to the peak of the T wave was measured before each shock during the first 8 hours of the experiment. Analysis of this information demonstrated that the time to the peak of the T wave remained relatively constant throughout the experiment. For the remainder of the study, the time from the pacing stimulus to the peak of the T wave was measured at the beginning of the experiment and at least every 60 minutes thereafter during pacing with a 10-ms monophasic pulse. If the pacing rate changed during the study, the time to the peak of the T wave was measured again.
The first step in a testing sequence was overdrive pacing. After 15 paced beats (S1) without competition, a ULV test shock (S2) was delivered to coincide with the peak of the T wave (Fig 2⇓). If the ULV test shock left the animal in normal sinus rhythm, at least 10 intrinsic beats elapsed before overdrive pacing was restarted and a subsequent ULV test shock was given. However, if ULV testing resulted in ventricular fibrillation, a defibrillation test shock was delivered 10 seconds after the onset of fibrillation. A rescue shock of known efficacy was given if the defibrillation test shock failed. A minimum of 4 minutes elapsed between fibrillation-defibrillation episodes.
ULV and DF test shock strengths were selected using a modification of the up-down algorithm.15 16 ULV and DF shocks began at 500 V. The ULV and DF voltages were independently adjusted according to each postshock rhythm. For ULV measurements, if the ULV shock did not induce ventricular fibrillation (that is, the animal remained in normal sinus rhythm), the ULV voltage was decreased by 20 V. Alternatively, if a ULV shock induced ventricular fibrillation, the voltage was increased by 40 V. Likewise, if a DF shock was successful (that is, the animal returned to normal sinus rhythm), the DF voltage was decreased by 20 V. If a DF shock was unsuccessful, the DF voltage was increased by 40 V. The incremental and decremental step sizes were unequal so that the test shocks would be concentrated in the upper portion of the probability of success curve. This modified up-down algorithm was repeated until fibrillation had been induced at least 40 times by ULV shocks.
Waveform Used for ULV and DF Testing
All test shocks were generated by a microprocessor-based external defibrillator (Ventritex HVS-02). ULV and DF test shocks were constructed using simulated single-capacitor biphasic waveforms as previously described.17 The shocks were 11 ms in total duration, with each phase lasting 5.5 ms.
The probit procedure from SAS18 was used to perform regression analysis on ULV and DF shocks separately to determine a probability of success curve for each. probit was also used to estimate the voltages corresponding to the .50, .75, .80, .95, and .97 probability of success (V50, V75, V80, V95, and V97, respectively). Pearson correlation coefficients for ULV and DF voltages at each probability of success level were determined using sas, and the significance of the correlation coefficient was determined for V50 and V95. For all analyses, a P value <.05 was considered significant. The statistical significance of these tests depends on the number of animals in the study,19 20 the variance of the voltage estimates, and the population variance of the animals,21 based on previous studies. The variance of the estimates depends on the number of fibrillation episodes per animal.22 From these two dependences, a minimum number of fibrillation episodes (40) and a minimum number of animals (15) were determined before the study began.
For the entire study, a total of 855 DF shocks and 2570 ULV shocks were delivered, averaging 53±12 DF shocks and 161±44 ULV shocks per animal. The mean lead impedance was 39 Ω.
Composite Animal Data
The ULV and DF measurements for all 16 animals were pooled; the composite data are shown in Table 1⇓ and Fig 3⇓. The slope of the DF curve is steeper than the ULV curve (Fig 3⇓). Below V97, the voltage that leaves the animal in normal sinus rhythm 97% of the time when delivered on the peak of the T wave (ULV) or the voltage that defibrillates 97% of the time (DF), DF was greater than ULV. The difference between ULV and DF became smaller at each successively higher percent success level up to V97 (Fig 4⇓). At V50, the voltage that leaves the animal in normal sinus rhythm 50% of the time when delivered on the peak of the T wave or defibrillates 50% of the time, ULV and DF were not significantly correlated (Table 1⇓). ULV and DF became more highly correlated as the probability of success increased to V97 (Fig 5⇓). Graphically, the two probability of success curves move closer together until they intersect just above V97. Beyond V98, however, the ULV and DF curves again diverge; ULV becomes greater than DF (Fig 3⇓), and they become less positively correlated (Fig 5⇓).
Individual Animal Data
Data for each animal are listed in Table 2⇓. As previously noted, ULV was less than DF for the composite data. There were six animals with ULV greater than DF for at least one probability of success level. All six animals had ULV greater than DF for the uppermost probability of success levels. Specifically, DF was greater than ULV in three animals for V95 through V97, two animals for V75 through V97, and one animal for V50 through V97.
Time to Peak of the T Wave
The average time to the peak of the T wave for all animals tested was 209±7.7 ms. The average variance of the time to the peak of the T wave for an individual animal was 5.8 ms. The time to the peak of the T wave remained relatively constant throughout an experiment, as shown in Fig 6⇓.
The ULV hypothesis is one of several theories regarding the mechanism of defibrillation.1 2 23 It is based on the ability of premature stimuli to induce ventricular fibrillation during the vulnerable portion of the T wave during regular rhythm.24 When the stimuli are made sufficiently large, they no longer induce fibrillation during the vulnerable portion of the cardiac cycle. This “upper limit of vulnerability” has been shown to exist in animals and humans.1 2 3 4 5 6 7 8 9 10 During ventricular fibrillation, activation fronts are present continuously so that likewise, areas of repolarization also should be present continuously. The ULV hypothesis states that unsuccessful shocks slightly weaker than necessary to defibrillate halt all activation fronts during fibrillation but stimulate regions of myocardium during their vulnerable period, giving rise to new activation fronts that reinitiate fibrillation.25 Since there is an upper limit to the shock strength that can induce ventricular fibrillation during the vulnerable period, shocks above this upper limit during fibrillation are hypothesized to defibrillate by extinguishing all fibrillation activation fronts without causing new activation fronts that reinitiate fibrillation. Therefore, the ULV should be related to the DFT. If so, information regarding the ULV and its correlation with the DFT may be useful during implantation and testing of automatic implantable cardioverter-defibrillators.
The composite data from this study show that below V97, the DF voltage is greater than the ULV voltage for the same probability of success. The ULV and DF voltages become more positively correlated as the probability of success level increased to V97; beyond V97, ULV and DF voltages become less positively correlated. As the probability of success level is increased, the estimation variance for both the ULV and DF voltages also increases.22 The increased estimation variance will tend to decrease the correlation.26 The fact that the correlation was observed to increase up to V97 despite this natural tendency toward decreasing correlation suggests that the ULV and DF voltages are most correlated at the upper portion of the probability of success curves. The observed drop in correlations beyond V97 may reflect the fact that the estimation variance increases more rapidly above V97 than below.22
Our observations are consistent with the results of Chen and colleagues1 that DF is greater than ULV. In their study, although a statistically significant difference between ULV and DFT energies did not exist for a lead configuration consisting of right atrium as anode and ventricular apex as cathode, a lead system consisting of the left atrium as anode and ventricular apex as cathode produced a statistically significant difference between ULV and DFT energies, with DFT greater than ULV. Another study by Chen et al2 compared V50 for ULV and DFT. Their results again showed that V50 for DFT is greater than ULV for monophasic and biphasic shocks given at the peak of the T wave. The V50 values of ULV and DFT were not significantly different if the ULV test shock was delivered at the mid upslope of the T wave; significant differences between ULV and DFT energy requirements were noted if the ULV test shocks were delivered on the peak of the T wave or during the mid downslope of the T wave. More recently, Chen et al8 compared V50 for ULV and DFT in 13 patients using 6-ms truncated exponential shocks delivered on the mid upslope of the T wave. The results are consistent with their previous studies in animals and our study in that DFT is greater than ULV and that ULV is significantly correlated with DFT. The significant positive correlation between the ULV and DF as shown in our study and previous studies1 2 7 8 9 further supports the ULV hypothesis for defibrillation. It is more difficult to identify the mid upslope and mid downslope of the T wave than to identify just the peak of the T wave because the beginning and the end of the T wave, in addition to the peak of the T wave, must be measured. Therefore, we chose to deliver all ULV test shocks on the peak of the T wave.
Contrary to our composite data, Wharton et al3 have reported ULV to be greater than DFT in all animals for both monophasic and biphasic shock waveforms. Bacon and associates6 examined ULV and DFT in humans before automatic implantable cardioverter-defibrillator placement and found mixed results: ULV was essentially the same as DFT in 8 patients, less than DFT in 4 patients, and greater than ULV in 3 patients. Chen et al8 reported 4 of 13 patients having ULV greater than DFT. Our individual animal data (Table 2⇑) also show some variability with regard to ULV and DF voltages; in 6 of the 16 animals, ULV was greater than DF for several probability of success levels.
Differences in experimental design, including different defibrillation electrode configurations, different S1 pacing sites, and different techniques for determining the ULV, may be responsible for the variation in results obtained in these studies. In some studies, the ULV was determined by scanning the T wave, and therefore the ability of each shock voltage to induce fibrillation was tested at numerous time points throughout the T wave. Other studies, including our own, gave each shock voltage at only one single time point within the T wave to greatly reduce the number of shocks given to determine the ULV. Since the scanning method tests a shock voltage at many time points, the ULV determined by scanning is likely to be higher than that determined at a single time point within the T wave.
If scanning is not used, the best single time point within the T wave to find the ULV is probably different for different electrode configurations and S1 pacing sites. Reentry leading to fibrillation is thought to be induced by a premature electrical stimulus given during the vulnerable period when a critical value of potential gradient field created by the stimulus intersects a critical degree of relative refractoriness of the myocardium.27 28 A functional reentrant activation front revolves about this critical point and soon degenerates into fibrillation. The ULV voltage that achieves a very high probability of success may be the shock voltage that creates a potential gradient distribution in which the minimum potential gradient just exceeds the critical value; thus, no critical points are formed. According to the critical point hypothesis, a shock slightly weaker than the ULV will generate the critical value of potential gradient in that part of the heart in which the shock potential gradient field is weakest. A shock delivered during the portion of the T wave in which the myocardium in this low potential gradient region is in its critical degree of relative refractoriness will lead to reentry and fibrillation. This time point within the T wave will be different for different electrode configurations because they have different potential gradient distributions and for different S1 pacing sites because they cause different sequences of activation and, hence, repolarization and refractoriness. If these ideas are correct, there is no one time point in the T wave that is best for determining the ULV of all electrode configurations and S1 pacing sites.
If the critical point hypothesis is correct, the ULV and DF probability of success curves should not be superimposable over their entire range. For the ULV determination, the dispersion of refractoriness should be nearly the same for all shocks given during the same point on the T wave, whereas shocks of the same strength given during different DF episodes will encounter different dispersions of refractoriness due to the relatively disorganized activation patterns of fibrillation. Therefore, the creation and location of critical points for shocks of identical strengths should be more reproducible during the ULV determination than during the DF determination, suggesting the ULV probability of success curve should be steeper than the DF curve. Alternatively, if a shock is given during that portion of the T wave when the shock field happens to be passing through the critical degree of refractoriness in that part of the ventricle where the shock field is weakest, the entire ULV probability of success curve should correspond to the uppermost portion of the DF probability of success curve. This would occur because a shock slightly lower than the ULV should always create a critical point leading to fibrillation, whereas the same shock strength given during fibrillation may or may not create a critical point, depending on the refractory state of the heart in the region where the shock field is weakest. Our results support the second conclusion, that the ULV probability of success curve will intersect at a high point on the DF probability of success curve. Our data do not support the hypothesis that the ULV curve will have a steeper slope than the DF curve [mean ULV, (V80−V50)/V50=0.32; mean DF, (V80−V50)/V50=0.14]. This suggests either (1) the critical point hypothesis is incorrect or (2) small changes in the T wave caused by changes in the autonomic, metabolic, and hemodynamic state of the animal or small changes in the shock electric field caused by slight changes in heart geometry, blood filling, and impedance can shift the critical point sufficiently to alter the ULV at a single point on the T wave.
To be clinically useful, the method of determining the ULV must be improved to eliminate these discrepancies. A recent study suggests a possible method by which this might be accomplished. Walker et al29 demonstrated that rapid pacing during ULV testing mimics the altered geometry, blood volume, and electrophysiological state seen in ventricular fibrillation and moves the ULV higher on the DF probability of success curve.
Limitations of the Study
First, this study examined only one waveform and a single pulse duration. Second, electrode polarity was held constant. It is not clear if polarity influences ULV. Third, we did not scan the entire T wave during ULV measurements. Scanning the T wave may have found a higher ULV value than that found by shocking only at the peak of the T wave. Fourth, probit analysis provides only an estimate of the relation between the probability of success and shock voltage. The fit of the probability of success curve to the measured data in some cases is poor. Therefore, the estimates of the probability of success for ULV and DF may be in error, particularly at very high and very low probabilities of success. Last, all studies were performed in healthy hearts; results may differ in diseased hearts.
ULV and DF are best correlated at the uppermost portion of the probability of success curves. This is fortunate, since our clinical interest is in setting the defibrillator voltage at this level. The ability to use ULV to estimate DF would allow for fewer ventricular fibrillation inductions during intraoperative testing of implantable devices. The smaller number of ventricular fibrillation inductions and subsequent rescue shocks should reduce the amount of myocardial damage and hemodynamic consequences that arise during DF testing.
Although the composite data appear promising, there were large discrepancies between ULV and DF voltages for a few animals. V97 was 329 V lower for ULV than for DF in one animal and was 244 V higher for ULV than for DF in another animal. The possibility of such an underestimate or overestimate of DF requirements by the ULV indicates that the method used in this study to determine ULV V97, and from it to estimate DF V97, should not be applied to clinical situations without additional development and study.
This study was supported in part by National Institutes of Health research grants HL-28429, HL-42760, HL-44066, and National Science Foundation Engineering Research Center grant CDR-8622201. The authors wish to thank Dr Patrick Wolf, Dr Gregory Walcott, Dr William Smith, Dennis Rollins, Ellen Dixon-Tulloch, Sharon Melnick, Jenny Hagler, Robert Walker, and Katie Rogers for their technical assistance.
- Received April 27, 1994.
- Revision received September 19, 1994.
- Accepted October 3, 1994.
- Copyright © 1995 by American Heart Association
Chen P-S, 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.
Wharton JM, Richard VJ, Murry CE Jr, Dixon EG, Reimer KA, Meador J, Smith WM, Ideker RE. Electrophysiological effects of monophasic and biphasic stimuli in normal and infarcted dogs. PACE. 1990;13:1158-1172.
Stanton MS, Mehra R, Morris J, Fetter J, DeGroot P. Relationship between defibrillation threshold and upper limit of vulnerability in humans. PACE. 1992;15:563. Abstract.
Bacon ME, Vitullo RN, Kasell JH, Blitchington T, Pressley JC, Smith PK, Lowe JE, Ideker RE, Wharton JM. Upper limit of vulnerability and its relationship to defibrillation threshold in humans. PACE. 1992;15:530. Abstract.
Kavanagh KM, Harrison JH, Dixon EG, Guse P, Smith WM, Wharton JM, Ideker RE. Correlation of the probability of success curves for defibrillation and for the upper limit of vulnerability. PACE. 1990;13:536. Abstract.
Chen P-S, Feld GK, Kriett JM, Mower MM, Tarazi RY, Fleck RP, Swerdlow CD, Gang ES, Kass RM. Relation between upper limit of vulnerability and defibrillation threshold in humans. Circulation. 1993;88:186-192.
Topham SL, Cha Y-M, Peters BB, Chen P-S. Effects of lidocaine on relation between defibrillation threshold and upper limit of vulnerability in open-chest dogs. Circulation. 1992;85:1146-1151.
Swartz JF, Karasik PE, Bacon ME, Fletcher RD. Relationship of upper limit of vulnerability to defibrillation threshold for monophasic and biphasic waveform human nonthoracotomy defibrillation. J Am Coll Cardiol.. 1994;23:293A. Abstract.
Dixon WJ, Mood AM. A method for obtaining and analyzing sensitivity data. J Acoust Soc Am. 1948;43:109-126.
McDaniel WC, Schuder JC. An up-down algorithm for estimation of the cardiac ventricular defibrillation threshold. Med Instrum. 1988;22:286-292.
Statistics: probit procedure. In: sas User’s Guide. Cary, NC: SAS Institute, Inc; 1982:287.
Finney DJ. Probit Analysis: a Statistical Treatment of the Sigmoid Response Curve. New York: Cambridge University Press; 1971:1-333.
Rao CR. Advanced Statistical Methods in Biometric Research. New York: John Wiley & Sons; 1952:230.
Wetherhill GB. Sequential estimation of quantal response curves. J R Stat Soc. 1962;B25:1-48.
Ideker RE, Chen P-S, Shibata N, Colavita PG, Wharton JM. Current concepts of the mechanism of ventricular defibrillation. In: Breithardt G, Borggrefe M, Zipes D, eds. Non-Pharmacological Therapy of Tachyarrhythmias. Mount Kisco, NY: Futura; 1987:449-464.
King BG. The Effect of Electric Shock on Heart Action With Special Reference to Varying Susceptibility in Different Parts of the Cardiac Cycle. New York: Columbia University; 1934. Thesis.
Chen P-S, Wolf PD, Ideker RE. Mechanism of cardiac defibrillation: a different point of view. Circulation. 1991;84:913-919.
Mendenhall W, Scheaffer RL, Wackerly DD. In: Payne M, ed. Mathematical Statistics With Applications. Boston: Duxbury Press; 1986:391.
Frazier DW, Wolf PD, Wharton JM, Tang ASL, Smith WM, Ideker RE. Stimulus-induced critical point: mechanism for electrical initiation of reentry in normal canine myocardium. J Clin Invest. 1989;83:1039-1052.
Winfree AT. When Time Breaks Down: the Three-dimensional Dynamics of Electrochemical Waves and Cardiac Arrhythmias. Princeton, NJ: Princeton University Press; 1987.
Walker RG, Idriss SF, Malkin RA, Ideker RE. Rapid pacing moves the upper limit of vulnerability higher on the defibrillation probability of success curve. J Am Coll Cardiol. 1994;23:293A. Abstract.