Alteration of Ventricular Fibrillation by Flecainide, Verapamil, and Sotalol
An Experimental Study
Background—The purpose of this study was to determine whether the myocardial electrophysiological properties are useful for predicting changes in the ventricular fibrillatory pattern.
Methods and Results—Thirty-two Langendorff-perfused rabbit hearts were used to record ventricular fibrillatory activity with an epicardial multiple electrode. Under control conditions and after flecainide, verapamil, or d,l-sotalol, the dominant frequency (FrD), type of activation maps, conduction velocity, functional refractory period, and wavelength (WL) of excitation were determined during ventricular fibrillation (VF). Flecainide (1.9±0.3 versus 2.4±0.6 cm, P<0.05) and sotalol (2.1±0.3 versus 2.5±0.5 cm, P<0.05) prolonged WL and diminished FrD during VF, whereas verapamil (2.0±0.2 versus 1.7±0.2 cm, P<0.001) shortened WL and increased FrD. Simple linear regression revealed an inverse relation between FrD and the functional refractory period (r=0.66, P<0.0001), a direct relation with respect to conduction velocity (r=0.33, P<0.01), and an inverse relation with respect to WL estimated during VF (r=0.49, P<0.0001). By stepwise multiple regression, the functional refractory periods were the only predictors of FrD. Flecainide and sotalol increased the circuit size of the reentrant activations, whereas verapamil decreased it. The 3 drugs significantly reduced the percentages of more complex activation maps during VF.
Conclusions—The activation frequency is inversely related to WL during VF, although a closer relation is observed with the functional refractory period. Despite the diverging effects of verapamil versus flecainide and sotalol on the activation frequency, WL, and size of the reentrant circuits, all 3 drugs reduce activation pattern complexity during VF.
Modification of cardiac electrophysiological properties and analysis of its repercussions on the activation patterns have provided useful information for furthering our knowledge of atrial fibrillatory processes.1 2 3 4 5 Interpretation of antiarrhythmic drug action on ventricular fibrillation (VF) entails the added difficulty of myocardial metabolic impairment secondary to the interruption of coronary perfusion.6 7 8 9 10 The use of experimental models in which myocardial perfusion is maintained during VF8 10 11 12 facilitates analysis of the characteristics of the arrhythmia and its relation to the electrophysiological properties of the ventricular myocardium, as in studies of atrial fibrillation.2 3 4 5 13 14 15 16 The present study investigates the pharmacological modifications of the VF activation pattern induced by 3 antiarrhythmic drugs (flecainide, verapamil, and d,l-sotalol) in the isolated rabbit heart using both time- and frequency-domain techniques and analysis of the epicardial activation maps. The purpose of the study is to analyze the variations in activation frequency and the complexity of the fibrillatory pattern produced by these drugs and to determine whether the modifications in myocardial electrophysiological properties are useful for predicting changes in the ventricular fibrillatory processes.
The procedures followed in this study conformed to institutional and NIH guidelines for the care and use of laboratory animals (NIH publication 85-23, revised 1985).
Thirty-two California rabbits (mean weight, 3.9±0.6 kg) were used. After anesthesia with ketamine (25 mg/kg IM) and heparinization, the hearts were removed and immersed in cold (4°C) Tyrode’s solution. After isolation, the aorta was connected to a Langendorff system for perfusing the Tyrode’s solution at a pressure of 50 mm Hg and a temperature of 37±0.5°C. The millimolar composition of the perfusion fluid was NaCl 130, NaHCO3 24.2, KCl 4.7, CaCl2 2.2, NaH2PO4 1.2, MgCl2 0.6, and glucose 12. Oxygenation was carried out with a mixture of 95% O2 and 5% CO2.
A plaque with 121 unipolar stainless steel electrodes (diameter, 0.125 mm; interelectrode distance, 1 mm) was positioned at the epicardial surface of the lateral wall of the left ventricle. The indifferent electrode was a 4×6-mm silver plaque located over the cannulated aorta. Ventricular pacing was performed with bipolar electrodes (diameter, 0.125 mm; interelectrode distance, 1 mm) located in the upper or central zone of the multiple electrode. Pacing (2-ms rectangular pulses, intensity twice the diastolic threshold) was carried out with a GRASS S88 stimulator with a stimulus isolation unit (SIU5). Recordings were obtained with a cardiac electrical activity mapping system (MAPTECH). The electrograms were amplified with a gain of 100 to 300, broadband (1 to 400 Hz) filtered, and multiplexed. The sampling rate in each channel was 1 kHz.
Thirty minutes after the electrodes were positioned, the extrastimulus test with 2 extrastimuli was applied. The basic cycle (S0S0) was fixed at 250 ms. The first extrastimulus (S1) was fixed at the minimum S0S1 interval that captured the ventricles when this interval was reduced in steps of 10 ms from 200 ms. The second extrastimulus (S2) was delivered at increasing S1S2 intervals in steps of 2 ms until an interval 10 ms longer than the minimum S1S2 that captured the ventricles. VF was induced by pacing at increasing frequencies from 4 to 20 Hz, and coronary perfusion was maintained during the arrhythmia. Recordings were analyzed 30 seconds after the onset of VF. The heart was defibrillated 5 minutes after VF induction by a DC shock (5 J). After 15 minutes, the protocol was repeated when the antiarrhythmic drug was added to the Tyrode’s solution: flecainide (1 μmol, n=10), verapamil (0.2 to 0.8 μmol, n=12), or d,l-sotalol (20 μmol, n=10). The concentrations of flecainide,17 18 19 d,l-sotalol,20 21 and verapamil22 23 were selected from the range in which their characteristic electrophysiological effects are observed. In the case of verapamil, dose adjustment was also made according to the effect elicited on the AV Wenckebach cycle length, choosing the concentration that produced a prolongation of this parameter of ≥50%.
Constant Pacing at the Basic Train (250 ms)
The following parameters were determined: (1) Effective ventricular refractory period (EVRP) for S2, or the maximum S1S2 interval that failed to induce ventricular activation. (2) Functional ventricular refractory period (FVRP), or the minimum V1V2 interval. (3) Ventricular conduction velocity, determined at the minimum V1V2 interval in both the longitudinal (VELG) and transverse (VELTR) direction of propagation. It was estimated by dividing the distance between 2 electrodes positioned 5 interelectrode spaces apart in the direction of maximum velocity as indicated by the isochrones or perpendicular to the former by the time interval between them. (4) Wavelength of the ventricular activation process (WL), or the product of the FVRP and the longitudinal conduction velocity.
Spectral analysis: Welch’s method24 was used to obtain the power spectrum of the signals recorded with 5 electrodes: 4 located in the midzone of each of the sides of the multiple electrode, and the fifth positioned in its central zone. The analysis was performed involving a data block of 2048 points (sampling rate, 1 kHz) (Figure 1⇓). This 2-second window of the VF recordings began at the 30th second after the onset of the arrhythmia. The dominant frequency (FrD) and the energy contained in the segment of the periodogram corresponding to FrD±1 Hz was obtained for each block. Data processing was performed with Matlab software on a Hewlett-Packard 712/80 platform.
Time-domain Analysis: Activation times in each electrode were determined by identifying the moment of maximum negative slope of the electrograms. The minimal threshold for dV/dt to be judged as a local deflection was a percentage (20%) of the maximal negative slope in each channel. The fibrillation interval (VV) histograms and the median (MN) of the consecutive VV intervals were determined during the same 2-second time windows.
Analysis of the Epicardial Activation Maps During VF: The maps were constructed every 100 ms in the 2-second time windows of the VF recordings. Isochrones were drawn semiautomatically at 5-ms intervals, and each map was classified into 3 categories based on its complexity13 : type I, single broad wave fronts propagating uniformly without significant conduction delay; type II, 2 wave fronts or 1 wave front with areas of conduction block or slow conduction; and type III, ≥3 wave fronts associated with areas of slow conduction and conduction block (Figure 2⇓). Likewise, for each map, the presence of activation patterns corresponding to complete reentry was analyzed (Figure 3⇓). In these maps, the electrodes activated on the internal portion of the reentrant wave front were identified by display of successive 10-ms time windows. These electrodes delimited the central core, whose maximum diameter (ie, length of the central core) was measured with the grid electrode matrix as reference. The area encompassed by the central core plus 2 electrodes recording the reentrant activation and the number of consecutive rotations were also determined.
Conduction Velocity During VF (VELVF): In the maps in which the activation front entered the area encompassed by the electrode, without evidence of breakthrough, velocity was calculated by dividing the distance between 2 electrodes positioned 5 interelectrode spaces apart in a direction perpendicular to the isochrones by the difference between their activation times (average of 5 determinations).
Functional Refractory Period During VF (FRPVF): The minimum intervals between 2 successive activations by 2 different wave fronts were determined (Figure 4⇓). FRPVF was regarded as the average of the 5 smallest values.
WL During VF: WLVF, or the product of FRPVF and VELVF, was analyzed.
Data are presented as mean±SD. Comparisons between 2 sets of data were made with Student’s t test for paired and unpaired data. The differences between qualitative variables were analyzed by χ2 test. Differences were considered significant for P<0.05. The linear regressions between pairs of variables were made by the least-squares method. Stepwise regression was used for multiple linear regression analysis.
Effects on Electrophysiological Parameters During Constant Pacing (250 ms)
Both flecainide and d,l-sotalol significantly prolonged FVRP and EVRP, and flecainide reduced the conduction velocity (VELG and VELTR) (Table 1⇓). Verapamil produced a slight shortening of the FVRP, without significant variations in EVRP or conduction velocity. None of the 3 drugs modified WL. Reproducibility of the determinations after a repeated induction of VF without adding any drug was tested, and no significant differences were found (Figure 5⇑).
Drug Effects on the Fibrillatory Pattern
Electrophysiological Parameters During VF
Flecainide and d,l-sotalol significantly prolonged FRPVF and WLVF and reduced VELVF (Table 1⇑). Verapamil reduced FRPVF and WLVF, although without changes in VELVF.
Flecainide and d,l-sotalol produced a decrease in the FrD determined by the 5 selected electrodes, without significant variations in the energy contained in the interval FrD±1 Hz, whereas verapamil increased both parameters (Table 2⇓).
Analysis in the Time Domain
Flecainide and d,l-sotalol produced an increase in the median of the VV intervals (MN) determined in each of the 5 selected electrodes or in the combined electrodes (Table 3⇓). Verapamil produced a decrease in this parameter.
Analysis of the Activation Maps
In all 3 study groups, the most frequent activation maps under control conditions corresponded to type II (type I: flecainide control, 15%; verapamil control, 19%; sotalol control, 11%; type II: flecainide control, 46%; verapamil control, 49%; sotalol control, 52%; type III: flecainide control, 39%; verapamil control, 32%; sotalol control, 37%), and patterns exhibiting complete reentry were recorded in 18 maps (9%) of the flecainide group, in 14 maps (7%) of the verapamil group, and in 12 maps (6%) of the d,l-sotalol group.
There were significant variations in the percentage of activation maps after administration of the 3 drugs (P<0.001), with a decrease in type III maps and an increase in maps of types I and II (type I: flecainide, 36%; verapamil, 24%; sotalol, 28%; type II: flecainide, 51%; verapamil, 58%; sotalol, 58%; type III: flecainide, 13%; verapamil, 18%; sotalol, 14%).
The number of activation maps with complete reentry patterns was 12 (6%) under the influence of flecainide, versus 14 (7%) and 12 (6%) with verapamil and d,l-sotalol, respectively. During flecainide perfusion, the number of consecutive rotations of reentrant activation did not vary (control, 1.4±0.6; flecainide, 1.3±0.5), whereas the length of the central core increased (control, 5±1; flecainide, 7±1 mm; P<0.001), as did the area encompassed by the central zone and the 2 electrodes recording the reentrant activation (control, 45±6; flecainide, 64±6 mm2; P<0.001).
Verapamil produced an increase in the number of consecutive rotations in the maps exhibiting reentry patterns (control, 1.3±0.4; verapamil, 2.1±1.1; P<0.02), with a significant decrease in the length of the central core (control, 5±1; verapamil, 4±1 mm; P<0.02) and the calculated area (control, 49±4; verapamil, 37±5 mm2; P<0.001).
d,l-Sotalol produced no significant variation in the number of consecutive rotations (control, 1.2±0.3; sotalol, 1.0±0.1), although increments were observed both in the length of the central core (control, 5±1; sotalol, 7±1 mm; P<0.001) and in area (control, 43±6; sotalol, 57±7 mm2; P<0.001).
Relation Between FrD and Electrophysiological Parameters
When the data corresponding to the control condition and after drug administration were considered jointly, the regression lines obtained on relating FrD (in Hz) to the MN (in ms) or its inverse (IMN, in Hz) were as follows: FrD=0.19 MN+28.6; r=0.88; n=64; P<0.0001; SEE=1.9 Hz; SE of the constant=1; SE of the coefficient=0.01. FrD=1.0 IMN−0.02; r=0.92; n=64; P<0.0001; SEE=1.6 Hz; SE of the constant=0.82; SE of the coefficient=0.06.
Simple regression analysis yielded significant results with the following parameters: EVRP (r=0.51, P<0.0001), FVRP (r=0.59, P<0.0001), FRPVF (r=0.66, P<0.0001), VELG (r=0.31, P<0.01), VELVF (r=0.33, P<0.01), and WLVF (r=0.49, P<0.0001). By stepwise multiple regression, the independent variables entered in the function were in first place FRPVF, and in second place FVRP (r=0.71, P<0.0001).
Effects of Flecainide, d,l-Sotalol, and Verapamil on the Frequency of Activation During VF
Spectral analysis and VV histograms demonstrate the slowing effect of flecainide and d,l-sotalol and the accelerating action of verapamil. The slowing of VF has also been reported with both sotalol and tedisamil.12 25 Flecainide has been found to slow arrhythmia in experimental models of atrial fibrillation,2 although the information available in the context of VF is limited. In an earlier study,9 we observed a decrease in the FrD of VF. This slowing effect has also been reported for other type I drugs, such as lidocaine,6 7 propafenone,11 and procainamide.26 A series of studies have found nifedipine, diltiazem, and verapamil to retard the slowing process of the arrhythmia in VF models in which coronary perfusion was not maintained.6 7 8 9 In addition, FrD has been seen to increase in the first moments of the arrhythmia under the action of nifedipine,8 and a shortening of the VF VV intervals after local and intravenous verapamil perfusion has been reported.27 The results of our study support the existence of a direct effect of verapamil on the VF pattern that is not related to protection against ischemia or to sympathetic reflexes triggered by the drug.
Modification of the WL During Arrhythmia
WL exhibits an inverse relation to inducibility of arrhythmia.1 28 29 Opposite actions on refractoriness and conduction velocity lead to variable results in WL.1 2 4 26 Flecainide reduces conduction velocity and increases refractoriness. These effects have been explained in terms of prolonged binding kinetics to the sodium channels and postrepolarization refractoriness.2 30 31 In the present study, the determinations made with 250-ms cycles show a balanced and counterposed action on both parameters; during VF, however, the predominance of the action on refractoriness led to a significant increase in WL. This fact has been described in atrial fibrillation,2 although it has also been reported that WL is not prolonged by flecainide or hydroquinidine.4 The effects of d,l-sotalol centered predominantly on refractoriness. This action is attributable primarily to cardiac potassium channel block.3 20 21 30 32 The effects may be expected to decrease during VF because of a reverse use-dependence. However, the increase in VF cycle length produced by class III antiarrhythmic drugs12 25 is explained in terms of concomitant increments in refractoriness. In the present study, we observed a slowing of the arrhythmia associated with an increase in WL attributable to the effect on refractoriness. With verapamil, a slight decrease in the functional refractory period and a shortening of the WL during the arrhythmia were observed, associated with VF acceleration under the influence of the drug.
During atrial fibrillation, the increase in WL implies a decrease in both the number of wave fronts and the activation frequency.2 3 In the present study, significant and inverse correlations were obtained between FrD and WL; however, more important correlations were obtained when the refractory periods were used. Other factors, such as the presence of an excitable gap, could determine the activation frequency during the arrhythmia.4 The widening of the excitable gap would imply a time increment between 2 successive activations.
Variations in Reentrant Circuit Size and Arrhythmia Complexity
Faster atrial reentrant rhythms based on smaller intramyocardial circuits and the opposite effects have been described.28 Procainamide produces an increase in the core size of the reentrant wave fronts and a slowing of VF.26 Cromakalim (an ATP-sensitive potassium channel opener) reduces the central core of functional reentries and the reentry cycle length.33 In the present study, we observed concordant variations in reentrant circuit size, activation cycle length, and WLVF. Conversely, flecainide and d,l-sotalol produced a reduction in the complexity of the activation maps. This observation would be related to the decrease in ventricular activation frequency during VF and to the increase in WLVF and reentrant circuit size induced by these drugs. However, verapamil was found to produce the opposite effects, and these phenomena were not associated with a greater complexity of the activation maps. There was also an increase in the energy contained in the segment of the periodogram corresponding to FrD±1 Hz that indicates more regular activation patterns. The increases in intracellular calcium levels after the development of VF34 and in the minimal level of calcium after decrease of pacing intervals35 suggest that drugs that modify intracellular calcium interfere with the fibrillatory processes.
Various factors could be related to the reduction in VF complexity produced by verapamil. On one hand, we see the reduction in the slope of the electrical restitution and in the action potential duration alternans produced by the drug (Riccio et al36 ). The decrease in the oscillation of action potential duration would reduce the wave breaks during VF, thus increasing the spatiotemporal organization of the arrhythmia.36 On the other hand, during VF, it has been shown that both fast and slow channels participate in myocardial activation.7 27 The L-type calcium channel blocking effect of verapamil could modify the oscillations of the conduction velocity, the reduction of which has been related to increased organization in cardiac activation.37 Further research is needed to clarify this subject.
Limitations of the Study
VF occurs in a 3-dimensional setting. In this context, the influence of breakthrough phenomena has been diminished in the determination of VELVF by use of only those maps in which both the input and output of wave fronts are identified without difficulty in the lateral zones of the activation maps. Although this measure does not avoid the influence of breakthrough patterns or wave fronts that proceed in an oblique way with respect to the epicardial surface, it does minimize this source of error, which would imply overestimation of the true conduction velocity. Although we have not established the magnitude of this error, it appears to be constant, because no significant differences were observed in analysis of the reproducibility of the determinations. Conversely, the estimation of FRPVF based on the wave-front dynamics implies an overestimation of this parameter in the presence of an excitable gap. In this sense, the use of the minimum values found contributes to reduction of the influence of this factor.
(1) Flecainide and d,l-sotalol prolong WLVF, whereas verapamil shortens it. (2) An inverse relation exists between FrD and WLVF, although this relation is more accentuated when ventricular refractoriness is considered. (3) The drug-induced modifications in activation frequency and WL are accompanied by significant changes in reentrant wave-front size, with increments induced by flecainide and d,l-sotalol and decrements under the action of verapamil. (4) Despite the divergent effects on activation frequency, WL, and reentrant circuit size of verapamil with respect to flecainide and d,l-sotalol, all 3 antiarrhythmic drugs diminish the complexity of the activation patterns during VF.
This work was supported in part by a grant (GV97-TI-05-14) from the Consellería de Cultura, Educació, i Ciència de la Generalitat Valenciana.
- Received September 17, 1999.
- Revision received November 4, 1999.
- Accepted November 8, 1999.
- Copyright © 2000 by American Heart Association
Rensma PL, Allessie MA, Lammers WJEP, Bonke FIM, Schalij MJ. Length of excitation wave and susceptibility to reentrant atrial arrhythmias in normal conscious dogs. Circ Res. 1988;62:395–410.
Wang Z, Pagé P, Nattel S. Mechanism of flecainide’s antiarrhythmic action in experimental atrial fibrillation. Circ Res. 1992;71:271–287.
Wang J, Feng J, Nattel S. Class III antiarrhythmic drug action in experimental atrial fibrillation: differences in reverse use dependence and effectiveness between d-sotalol and the new antiarrhythmic drug ambasilide. Circulation. 1994;90:2032–2040.
Allessie MA, Wijffels MC, Dorland R. Mechanisms of pharmacologic cardioversion of atrial fibrillation by class I drugs. J Cardiovasc Electrophysiol. 1998;9(suppl):S69–S77.
Elvan A, Pride HP, Eble JN, Zipes DP. Radiofrequency catheter ablation of the atria reduces inducibility and duration of atrial fibrillation in dogs. Circulation. 1995;91:2235–2244.
Stewart AJ, Allen JD, Devine AB, Adgey AAJ. Effects of blockade of fast and slow inward current channels on ventricular fibrillation in the pig heart. Heart. 1996;76:513–519.
Martin G, Cosin J, Such M, Hernandez A, Llamas P. Relation between power spectrum time course during ventricular fibrillation and electromechanical dissociation: effects of coronary perfusion and nifedipine. Eur Heart J. 1986;7:560–569.
Chorro FJ, Sánchez-Muñoz JJ, Sanchis J, Cortina J, Bataller M, Guerrero J, Espí J, Ruiperez JA, López-Merino V. Modifications in the evolution of the dominant frequency in ventricular fibrillation induced by amiodarone, diltiazem, and flecainide. J Electrocardiol. 1996;29:319–326.
Chorro FJ, Guerrero J, Cánoves J, Martínez-Sober M, Mainar L, Sanchis J, Calpe J, Llavador E, Espí J, López-Merino V. Quantification of the modifications in the dominant frequency of ventricular fibrillation under conditions of ischemia and reperfusion: an experimental study. Pacing Clin Electrophysiol. 1998;21:1716–1723.
Dorian P, Newman D. Tedisamil increases coherence during ventricular fibrillation and decreases defibrillation energy requirements. Cardiovasc Res. 1997;33:485–494.
Konings KTS, Kirchhof CJHJ, Smeets JRLM, Wellens HJJ, Penn OC, Allessie MA. High-density mapping of electrically induced atrial fibrillation in humans. Circulation. 1994;89:1665–1680.
Kirchhof CJHJ, Chorro FJ, Scheffer GJ, Brugada J, Konings KTS, Zetelaki Z, Allessie M. Regional entrainment of atrial fibrillation studied by high-resolution mapping in open-chest dogs. Circulation. 1993;88:736–749.
Gray RA, Pertsov AM, Jalife J. Incomplete reentry and epicardial breakthrough patterns during atrial fibrillation in the sheep heart. Circulation. 1996;94:2649–2661.
Carmeliet E. Electrophysiologic and voltage clamp analysis of the effects of sotalol on isolated cardiac muscle and Purkinje fibers. J Pharmacol Exp Ther. 1985;232:817–825.
Vaughan Williams EM, Somberg JC. Control of Cardiac Rhythm. Philadelphia, Pa: Lippincott-Raven Publishers; 1998.
Oppenhein A, Schafer R. Digital Signal Processing. Hertfordshire, UK: Prentice Hall Inc; 1975.
Kwan YY, Fan W, Hough D, Lee JJ, Fishbein MC, Karagueuzian HS, Chen PS. Effects of procainamide on wave-front dynamics during ventricular fibrillation in open-chest dogs. Circulation. 1998;97:1828–1836.
Zhou X, Guse P, Wolf PD, Rollins DL, Smith WM, Ideker RE. Existence of both fast and slow channel activity during the early stage of ventricular fibrillation. Circ Res. 1992;70:773–786.
Smeets JLRM, Allessie MA, Lammers WJEP, Bonke FIM, Hollen J. The wavelength of the cardiac impulse and reentrant arrhythmias in isolated rabbit atrium: the role of heart rate, autonomic transmitters, temperature, and potassium. Circ Res. 1986;58:96–108.
Wang Z, Fermini B, Nattel S. Mechanism of flecainide’s rate-dependent actions on action potential duration in canine atrial tissue. J Pharmacol Exp Ther. 1993;267:575–581.
Uchida T, Yashima M, Gotoh M, Qu Z, Garfinkel A, Weiss JN, Fishbein MC, Mandel WJ, Chen PS, Karagueuzian HS. Mechanism of acceleration of functional reentry in the ventricle: effects of ATP-sensitive potassium channel opener. Circulation. 1999;99:704–712.
Koretsune Y, Marban E. Cell calcium in the pathophysiology of ventricular fibrillation and in the pathogenesis of postarrhythmic contractile dysfunction. Circulation. 1989;80:369–379.
Thandroyen FT, Morris AC, Ziman B, Pai L, Willerson JT, Buja LM. Intracellular calcium transients and arrhythmia in isolated heart cells. Circ Res. 1991;69:810–819.
Riccio ML, Koller ML, Gilmour RF. Electrical restitution and spatiotemporal organization during ventricular fibrillation. Circ Res. 1999;84:955–963.
Cao JM, Qu Z, Kim YH, Wu TJ, Garfinkel A, Weiss JN, Karagueuzian HS, Chen PS. Spatiotemporal heterogeneity in the induction of ventricular fibrillation by rapid pacing: importance of cardiac restitution properties. Circ Res. 1999;84:1318–1331.