Cellular Action Potential and Excitable Gap During Fibrillation

In the past few years, knowledge of the characteristics of the action potentials during fibrillation has increased greatly. This is a direct result of the introduction of techniques for recording action potentials in whole hearts, either in vivo³¹ or in perfused, isolated hearts.^4–6,10,31 During fibrillation, the action potentials are altered: the APD is decreased; the action potential upstroke is slowed (decreased first-order derivative, dV/dt) and of decreased magnitude; the plateau phase is abbreviated; and DIs are abbreviated or absent (Fig. 2-3). During the first few seconds of VF (or atrial fibrillation), the activation rate is quite rapid; the mean cycle length of VF in patients undergoing defibrillator implantation was 213 ± 27 msec.⁶ DIs are rarely seen during early fibrillation, and the upstroke of most action potentials occurs before the transmembrane potential has returned to baseline from the previous action potential. The demonstration of an excitable gap in fibrillating atrial tissue,³⁴ as well as evidence of an excitable gap in fibrillating ventricular tissue,³⁵ suggests that there are periods late in the action potential in the fibrillating myocardium during which an electrical stimulus can capture a portion of the fibrillating myocardium. Knowledge of an excitable gap provides an opportunity to stimulate the tissue just in front of a fibrillating wavefront, to cause wavefront block.

Figure 2-3 Recording during ventricular fibrillation in a human.

Leads I, II, and III are body surface electrocardiograms. Note that there is no period of diastole between action potentials. MAP, Right ventricular monophasic action potentials; ECM, local bipolar electrogram.

(From Swartz JF, Jones JL, Fletcher RD: Characterization of ventricular fibrillation based on monophasic action potential morphology in the human heart. Circulation 87:1907-1914, 1993.)

As described in Chapter 1, the electrical activity of the heart is controlled ultimately by ion channels located in the cell membrane of the myocyte. It has been established that both the voltage-gated fast channels (sodium [Na⁺]) and slow channels (Na⁺ and calcium [Ca²⁺]) are active during the first few seconds of VF.⁸ The fast-channel activity is indicated by the rapidity of the upstroke of the action potential (phase 0) during early fibrillation and its sensitivity to administration of the Na⁺ channel blocker, tetrodotoxin. As fibrillation proceeds, the upstrokes of the action potentials become increasingly slower, with a decreased dV/dt_max, but the action potentials remain sensitive to tetrodotoxin until 1 to 5 minutes after initiation of fibrillation. A transition then occurs in which the action potential upstrokes become insensitive to tetrodotoxin. This suggests that the propagation of the action potential is no longer mediated primarily by the fast voltage-gated Na⁺ channels and may be mediated by slow voltage-gated Ca²⁺ channel activity in the later stages of fibrillation.⁴ The activation complexes recorded from the ventricular myocardium remain active only as long as the coronary arteries are perfused with oxygenated blood, suggesting that ischemia may be responsible for the loss of fast-channel activation during prolonged VF.²⁶

Waveforms, Current Strength, and Distribution During Defibrillation

The two most common waveform shapes used clinically are the monophasic and biphasic waveforms. In monophasic waveforms, the polarity of the shock is unchanged at each electrode for the entire duration of the electrical shock. In biphasic waveforms, the polarity of the shock reverses at each electrode partway through the defibrillation waveform. Many studies, in both animals and humans, have shown that biphasic waveforms can defibrillate with less current and energy than monophasic waveforms, in both internal and transthoracic defibrillation configurations.^37–40 Within each type, waveforms can be described as truncated exponential or damped sinusoidal shapes. ICDs use truncated exponential biphasic waveforms. Most external defibrillators have used damped sinusoidal monophasic waveforms, but because of the inductor necessary to shape the waveform, these defibrillators tend to be large and heavy. More recently, smaller, lighter external defibrillators have been developed that use truncated exponential biphasic waveforms similar to those used in ICDs. Damped sinusoidal biphasic waveforms are used in external defibrillators in Russia; similar to truncated exponential biphasic waveforms, these show improved efficacy over monophasic waveforms.^41,42

However, not all biphasic waveforms are superior to monophasic waveforms, For example, if the second phase of the biphasic waveform becomes much longer than the first phase, the energy required for defibrillation increases and can eventually rise to a level greater than the energy required to defibrillate with a monophasic waveform (with duration equal to the first phase of the biphasic waveform).^40,43,44 The optimum duration of the two phases of the biphasic waveform depends on the electrode impedance and the defibrillator capacitance.^45–48

Several groups have shown that defibrillation efficacy for square waveforms follows a strength-duration relationship similar to that for cardiac stimulation;^49,50 as the waveform becomes longer, the average current at the 50% success point (the current when one half of delivered shocks will succeed) becomes progressively less, approaching an asymptote called the rheobase.⁵¹ On the basis of this observation, some suggest that cardiac defibrillation can be mathematically modeled using a resistor-capacitor (RC) network to represent the heart^45,51–53 (Fig. 2-4). As empirically determined, the time constant for the parallel RC network is 2.5 to 5 msec.^45,47,53 In one version of the model,⁵³ a current waveform is applied to the RC network. The voltage across the network is then calculated for each time point during the defibrillation pulse. The relative efficacies of different waveform shapes and durations can be compared by determining the current strength that is necessary to make the voltage across the RC network reach a particular value, called the defibrillation threshold.

Figure 2-4 Parallel resistor-capacitor (RC) network representation of heart.

Responses of RC network to monophasic and biphasic truncated exponential waveforms with a time constant of 7 msec. The parallel RC network has a time constant of 2.8 msec. A, Input monophasic waveforms. The leading-edge current of the input waveform was 10 A. The waveforms were truncated at 1, 2, 3, 4, 5, 6, 8, and 10 msec. B, Model response, V(t). Initially, as the waveform lengthens, V(t) increases until it reaches a maximum at about 4 msec, after which it begins to decrease. C, Input biphasic waveforms. The leading-edge current was 10 A. Phase 1 was truncated at 6 msec. Phase 2 was truncated after 1, 2, 3, 4, 5, 6, 7, and 8 msec. D, Model response does not change polarity until the phase 2 duration is longer than 2 msec.

(From Walcott GP, Walker RG, Cates AW, et al: Choosing the optimal monophasic and biphasic waveforms for ventricular defibrillation. J Cardiovasc Electrophysiol 6:737-750, 1995.)

Several observations can be made from this model. First, for square waves, as the waveform duration becomes longer, the voltage across the network increases, approaching an asymptote, or rheobase. For truncated exponential waveforms, however, the model voltage rises, reaches a peak, and then, if the waveform is long enough, begins to decrease (see Fig. 2-4). Therefore, the model predicts that monophasic exponential waveforms should be truncated at a time when the peak voltage across the RC network is reached. Current or energy delivered after that point is wasted. In support of this prediction, strength-duration relationships measured in both animals⁵³ and humans⁵⁴ do not approach an asymptote but rather reach a minimum and remain constant as the waveform lengthens. This minimum occurs over a range of waveform durations and does not extend indefinitely. Schuder et al.⁵⁵ showed that, if the duration of a waveform becomes too long, defibrillation efficacy decreases.

Second, the model predicts that the heart acts as a low-pass filter.⁵² Therefore, waveforms that rise gradually should have better efficacy than waveforms that turn on immediately. This prediction has been shown to hold true for external defibrillation,⁵⁶ internal atrial defibrillation,⁵⁷ and internal ventricular defibrillation.⁵⁸ Ascending ramps defibrillate with a greater efficacy than do descending ramps.^58,59

Third, several groups have suggested that the optimal first phase of a biphasic waveform is the optimal monophasic waveform.^46,53 If so, what does the model predict as the “best” second phase of a biphasic waveform? Empirically, the role of the second phase apparently is to return the model voltage response back to zero as quickly as possible, thereby maximizing the increased efficacy of the biphasic waveform over that of the monophasic waveform with the same duration as phase 1 of the biphasic waveform. If the network voltage does not reach zero, or if it overshoots zero, efficacy is lost. Swerdlow et al.⁴⁷ showed in humans that the “best” phase 2 of a biphasic waveform returns the model response close to zero.

Together, these models allow the clinician to choose optimal capacitor sizes for truncated exponential biphasic waveforms, the waveforms most often used in ICDs. The capacitor must be large enough to be able to raise the network voltage to its threshold value and still hold enough charge to drive the network voltage back to zero. For a 40-Ω interelectrode impedance and a network time constant of 2.8 msec, the minimum capacitor that can accomplish this has a capacitance of 75 microfarads (µF).

These models can also help in choosing shock parameters, including phase duration for biphasic waveforms. Shock duration should be varied depending on defibrillation lead impedance; patients with low impedance should receive short-duration shocks, whereas patients with high impedance should receive longer-duration shocks. Initially, the concept of “tilt” was used to vary shock phase durations with impedance. Tilt is defined as the quotient of the leading-edge voltage minus the trailing voltage at the end of the shock divided by the voltage at the beginning of the shock, expressed as a percentage. For example, a shock with starting voltage of 500 V and ending voltage of 100 V has a tilt of 80%. If tilt is held constant, shock duration varies linearly with shock impedance. Compared with model predictions (Fig. 2-5), fixed-tilt waveforms are too short for patients with low impedance and too long for patients with higher impedance. One defibrillator manufacturer has developed a reference table that gives waveform phase durations as a function of patient impedance using a model similar to the one previously presented.^60,61 Their model uses “model time constants” of 2.5, 3.5, and 4.5 msec. It is recommended that the 3.5-msec constant be used first. If the defibrillation threshold for the 3.5-msec waveform is unsatisfactory, using the 4.5-msec constant is likely to be more effective than 2.5 msec. Waveforms that are too short tend to fail; waveforms that are too long tend to waste shock energy yet still defibrillate.

Figure 2-5 Waveform durations.

Optimal duration of first phase of a biphasic waveform as a function of shock time constant is shown for model responses of 2.0, 3.5, and 4.5 msec. Also shown are the first-phase durations for 50%-tilt and 65%-tilt waveforms.

The location of the defibrillation electrodes affects the magnitude of the shock necessary to defibrillate the heart. Typically, 200 to 360 J of energy are necessary for successful defibrillation, with the defibrillation electrodes located on the body surface, during transthoracic defibrillation with a damped sinusoidal monophasic waveform. Although less energy is required for a truncated exponential biphasic waveform,⁶² only about 4% to 20% of the current that is delivered to transthoracic defibrillation electrodes ever reaches the heart.^63,64 Indeed, when the defibrillation electrodes are placed in the heart itself, usually only 20 to 34 J of energy is required, and the requirement may be as low as only a few joules when large, contoured epicardial electrodes are used.^40,65 The strength of the shock also varies for different locations on or in the heart; epicardial patches defibrillate with a lower shock energy than transvenous electrode configurations.⁶⁶

Although defibrillation efficacy is usually described by some measure of defibrillation shock “strength” (energy, voltage, or current), little insight into the mechanisms of defibrillation can be obtained from these measures. Knowing how the current (or voltage) of a defibrillation shock is distributed over the heart allows greater understanding of how defibrillation occurs. Several studies have measured the potential gradient distribution throughout the heart during a defibrillation shock.^14,67,68 The potential gradient is a measure of the spatial variation of shock voltage across the heart. The potential gradient is measured in volts per centimeter (V/cm) of tissue. In a region with a high potential gradient, the difference in voltage between a given point and an area 1 cm distant from that point is high. Regions of low potential gradient have a measured voltage that is similar to that of nearby points. These studies show an uneven potential distribution for most electrode configurations, with areas of high potential gradient near the defibrillation electrodes and areas of low potential gradient in regions distant from the defibrillating electrodes. It has been hypothesized that a minimum potential gradient must be attained for successful defibrillation to occur, and that this requirement is independent of the current applied or the electrode configuration.^15,16 After a shock that fails to defibrillate VF, the site of earliest activation immediately after the shock can be mapped and related to the electric field that was produced by the shock. For shocks near the defibrillation threshold, the sites of earliest activation after a failed shock occur in the areas of lowest potential gradient.

The minimum potential gradient required for defibrillation is lower for biphasic than for monophasic waveforms (4 vs. 6 V/cm). A minimum potential gradient of 6 V/cm was required for successful defibrillation using a 10-msec truncated exponential monophasic waveform in the open-chest dog model.¹⁶ Similar findings were observed using a 14-msec truncated exponential monophasic waveform and multiple electrode configurations.¹⁵ In contrast, a minimum potential gradient of 4 V/cm was required for successful defibrillation using a truncated exponential biphasic waveform. Because higher shock strengths are required to induce a higher potential gradient, biphasic shocks successfully defibrillate with lower energy than monophasic shocks (i.e., a lower-voltage gradient is required).

The requirement for a minimum potential gradient may reflect the need for a shock to prevent the generation of new activation fronts that can result in reinitiation of fibrillation.⁶⁹ Examination of activation patterns after failed defibrillation for progressively larger shock strengths indicate that postshock activation occurs at numerous sites throughout the ventricle, and that reentry is common when the shock strength is much lower than that needed for defibrillation⁷⁰ (Fig. 2-6). At shock strengths just lower than those required for defibrillation, postshock activation arises in a limited number of myocardial regions. The activation fronts then propagate to activate other regions of the myocardium for a few cycles before reentry occurs; activation becomes disorganized; and fibrillation is reinitiated. Although postshock activation sites can still arise in regions of lowest potential gradient after a shock slightly greater than that required for defibrillation, the cycles of activation that originate from these sites are slower. These activations terminate after a few cycles without reinitiating fibrillation.^69–71

Figure 2-6 Phase maps of single rabbit heart.

Maps show the last cycle before a shock (left side) and the first cycle after a shock (right side) of 100 V (A), 200 V (B), 600 V (C), and 800 V (D) that failed to defibrillate. The defibrillation threshold was 800 ± 200 V in this series. Colors represent phase, and the symbols + and − indicate phase singularities or centers of reentrant circuits of opposite direction. Phase singularities as a marker of reentry were observed during ventricular fibrillation just before the shock in all cases. A and B, Postshock phase singularities were observed after failed 100-V and 200-V shocks. Visual analysis of animations of the optical recordings indicated that many of the phase singularities represented reentrant activations occurring immediately after the shock, so that the postshock interval was 0 msec. C and D, No phase singularities were observed after the 600-V and 800-V shocks. For the 600-V shock, activation propagated away in all directions from two early sites, at the apex and at the lateral base of the left ventricle, both of which appeared after a postshock interval of 42 msec. For the 800-V shock, a single wavefront of activation appeared at the apex and propagated away in all directions in a focal pattern after a postshock interval of 72 msec.

(From Chattipakorn N, Banville I, Gray RA, Ideker RE: Effects of shock strengths on ventricular defibrillation failure. Cardiovasc Res 61:39-44, 2004.)

At least two inferences can be drawn from the model that the lowest potential gradient region determines whether or not a shock will halt fibrillation. First, different shock electrode configurations with the same defibrillation threshold (DFT) have the same lowest potential gradient value, but in different regions of the heart. Single-coil versus dual-coil defibrillator systems have similar DFTs. Mapping of postshock activations shows that the earliest recorded activity after the shock is the anterior left ventricle for dual-coil shocks but the posterior area for single-coil shocks.⁷² Thus the low gradient region presumably is anterior for dual-coil shocks and posterior for single-coil shocks, although this inference needs to be confirmed experimentally. Crossley et al.⁷³ measured DFTs for shocks from either an electrode in the right ventricular (RV) apex or RV outflow tract to the pulse generator can and showed that DFTs were not significantly different for the two configurations. We can infer that the lowest potential gradient level was the same for each configuration, but likely in a different location.

Second, high DFTs occur in some patients because either (1) the geometry of their heart is such that the low potential gradient region has a lower potential gradient than in most patients, or (2) these patients require a higher gradient level in the low gradient region to defibrillate. In either case, unusual lead configurations may help in these patients. A subcutaneous array can lower high DFTs,⁷⁴ most likely by directing current to the left ventricular free wall. Likewise, an azygos vein electrode can be useful in lowering high thresholds,⁷⁵ likely by directing current to the posterior portion of the heart.

The previous discussion shows how defibrillation can fail because a shock is of insufficient strength. Another issue involves the defibrillation shock that becomes extremely large. At high shock strengths, the probability of defibrillation success again begins to decrease. It is believed that, at large strengths, defibrillation shocks can have detrimental effects on the heart. Increasing the shock strength to very high levels (>1000 V with transvenous electrodes) can result in activation fronts arising from regions of high potential gradient that reinduce VF.⁷⁶ Cates et al.⁷⁷ showed that, for both monophasic and biphasic shocks, increasing shock strength does not always improve the probability of successful defibrillation and may even increase the incidence of postshock arrhythmias. Chapman et al.⁷⁸ showed in dogs that the time required for the heart to recover hemodynamically after a defibrillation episode was shorter for biphasic than for monophasic shocks. Further, they showed that hemodynamic recovery took longer after high-energy shocks than after low-energy shocks. Reddy et al.⁷⁹ showed that transthoracic defibrillation with biphasic shocks resulted in less postshock electrocardiographic evidence of myocardial dysfunction (injury or ischemia) than standard monophasic damped sinusoidal waveforms, and without compromise of defibrillation efficacy.

One mechanism implicated in the means by which shocks cause damage to the myocardium is electroporation, the formation of holes or pores in the cell membrane. Electroporation may occur in regions where the shock potential gradient is high (>50 to 70 V/cm) and may even occur in regions where the potential gradient is much less than 50 V/cm.⁸⁰ The very high voltage can result in disruption of the phospholipid membrane bilayer and in the formation of pores that permit the free influx and efflux of ions and micromolecules. Electroporation can cause the transmembrane potential to change temporarily to a value almost equal that of the plateau of the action potential. At this transmembrane potential, the cell is paralyzed electrically, being both unresponsive and unable to conduct an action potential. Exposure of the myocardium to yet higher potential gradients, probably greater than 150 V/cm, results in arrhythmic beating, and at extremely high potential gradients, necrosis may occur.⁸¹

The shape of the waveform alters the strength of the shock at which these detrimental effects occur. Use of a 10-msec, truncated exponential monophasic waveform for VF in dogs resulted in conduction block in regions where the potential gradient was greater than 64 ± 4 V/cm.⁸² Shocks that created even higher potential gradients in the myocardium (71 ± 6 V/cm) were required for conduction block when a 5-msec/5-msec, truncated exponential biphasic shock was used. Adding a second phase to a monophasic waveform, thereby making it a biphasic waveform, reduced the damage sustained by cultured chick myocytes compared with that induced by the monophasic waveform alone.⁸³ Therefore, biphasic waveforms are less likely to cause damage or dysfunction in high-gradient regions than monophasic waveforms.

Models Proposed to Explain Induction of Changes in Transmembrane Potential throughout Heart during Defibrillation Shock

A shock in the form of a square wave given across the defibrillation electrodes appears almost immediately as a square wave in the extracellular space of the heart. There is no significant distortion, because the extracellular space throughout the body is primarily resistive, with little reactive component. Phase delays and alterations of the appearance of the shock wave occur in the transmembrane potential, however, because of the capacitance and ion channels of the myocyte membrane.⁸⁴ Consequently, a square-wave shock can elicit an exponential change with time in the transmembrane potential (Fig. 2-7).

Figure 2-7 Effect of square-wave shock on extracellular potential and transmembrane potential.

The square-wave shock appears immediately as a relatively undistorted square wave in the extracellular space. It appears as an exponentially increasing change in the transmembrane potential. When a shock of a given polarity is delivered during the action potential plateau (A), the depolarization obtained is different in magnitude and time course from the hyperpolarization obtained with a shock of the opposite polarity (B).

(From Walcott GP, Knisley SB, Zhou X, et al: On the mechanism of ventricular defibrillation. Pacing Clin Electrophysiol 20:422-431, 1997.)

The nonlinear behavior of the membrane caused by the ion channels also affects the outcome of reversing the polarity of the defibrillation shock. Reversing the polarity may reverse the sign of the change in the transmembrane potential in some regions of the myocardium, and the nonlinear behavior of the membrane can alter the magnitude and the time course of this change. As discussed previously, reversing the polarity of the shock may not reverse the sign of the change in the transmembrane potential in all regions of the heart; some areas may be hyperpolarized with both shock polarities.⁸⁵ This may reflect the nonlinear behavior of the membrane ion channels.

Several models have been formulated in an attempt to explain the mechanisms by which the defibrillation shock is distributed throughout the myocardium to restore coordinated, effective action potentials. As yet, none of the models adequately describes all the experimental findings on the action potential changes during defibrillation. It is well established experimentally that changes occur many centimeters from the defibrillating shock electrodes. These changes in transmembrane potential can result in new action potentials or prolongation of the action potential as described previously.^23,86 Direct excitation can be observed,²³ even far from the electrode (>30 mm),⁸⁷ as well as across the entire heart.⁸⁸

Although the one-dimensional cable model described in Chapter 1 adequately describes the generation of self-propagating action potentials close to an electrode as required for pacing, it fails to account for the far-field changes observed during defibrillation. During stimulation or defibrillation, this model predicts that the tissue near the anode should be hyperpolarized, whereas the tissue near the cathode should be depolarized.⁸⁹ The magnitude of the hyperpolarization or depolarization decreases exponentially with the distance from the electrodes according to the membrane space constant, the distance at which the hyperpolarization or depolarization has decreased by 63%. For cardiac tissue, the space constant is only 0.5 to 1 mm.^89,90 Therefore, the one-dimensional cable equations predict that tissue more than 10 space constants (~1 cm) distant from the defibrillation electrodes should not directly undergo changes in transmembrane potential because of the shock field. That is, new action potentials should not arise by direct excitation at distances greater than 1 cm from the electrodes. This model fails to describe the experimentally observed global distribution of action potentials during defibrillation.

Therefore, several additional mathematical formulations have been proposed, including the sawtooth model,^21,91–93 the formation of secondary sources at barriers in the myocardium,^20,94 and the bidomain model,⁹⁵ to explain how a defibrillation shock affects the transmembrane potential a long distance from the shocking electrodes. In the simplest formulation of these models, the extracellular and intracellular spaces are considered to be low-resistance media and the membrane to be a high-resistance medium in parallel with capacitance. The simple case models incorporate only passive myocardial properties. The models have been rendered more realistic by the addition of active components to represent the ion channels in the membrane, gap junctions, and membrane discontinuities.^96,97 By convention, the current is defined as the flow of positive ions from the anode to the cathode.

Sawtooth Model

The one-dimensional cable model positions two low-resistance continuous spaces that conduct current from the shock, the intracellular space and the extracellular space, separated by a high-resistance cell membrane. In the sawtooth model the intracellular space is divided by a series of high-resistance barriers, the gap junctions. Because of these high-resistance barriers, current moving in the intracellular space is forced to exit into the extracellular space and reenter the cell on the other side of the barrier. Exit and reentry of the current from the intracellular domain results in hyperpolarization near the end of the cell closest to the anode and depolarization near the end of the cell closest to the cathode. A tracing of the changes in transmembrane potential along a fiber during the shock should therefore resemble the teeth of a saw, with each tooth corresponding to an individual cell^21,91–93 (Fig. 2-8). Increases in the junctional resistances are predicted to increase the magnitude of the potential changes at the ends of the cell.²¹ Although gap junctions are of low resistivity, they can present significant junctional resistance under certain conditions, such as hypoxia^98,99 and calcium depletion.¹⁰⁰ As the resistance of the gap junctions increases, it is predicted that a greater fraction of current passes preferentially across the cell membrane rather than along the cell.

Figure 2-8 Sawtooth model of transmembrane potential during a shock.

A, Transmembrane potential shown is a summation of the membrane potential profile of the cable model and the periodicity arising from the periodic changes in intracellular resistance. The anode is to the left and the cathode is to the right in this one-dimensional model. The fiber is divided into 31 cells of equal length separated by junctions of high resistance. In the figure, the junctional resistance is shown much greater than is believed to occur in cardiac fibers, to allow the sawtooth pattern to be seen. B, Two parts of the summation in panel A are shown; V_m⁰, transmembrane potential profile of cable model; V_m¹, periodicity arising from periodic changes in intracellular resistance.

(Modified from Krassowska W, Pilkington TC, Ideker RE: Periodic conductivity as a mechanism for cardiac stimulation and defibrillation. IEEE Trans Biomed Eng 34:555-560, 1987.)

The sawtooth model adequately describes the requirement for a minimum potential gradient, because the magnitudes of the hyperpolarization and polarization at the ends of the cells are directly proportional to the strength of the stimulus. It also adequately describes the generation of action potentials at a distance from the electrodes and the differences in threshold stimuli between cathodal and anodal stimulation. Sawtooth changes in transmembrane potential have been observed in preparations of isolated cardiomyocytes;^101,102 however, such a pattern in isolated cells would be consistent with the cable model. This pattern has not been observed in a syncytium of cardiac cells.^20,103,104

Secondary Source Model

Although the resistivity of the gap junctions at the boundaries between the cells may not adequately explain the physiologic effects of defibrillation, the resistivity of other intracellular discontinuities and interruptions may well play a role. Most theories concerning the generation of action potentials and their propagation across the ventricle, such as the bidomain theory described later, consider the myocardium to be a uniform electrical continuum. This assumption does not take into account the discontinuities of the intracellular domain, where the myocardium is interrupted by barriers such as connective tissue septa, blood vessels, and scar tissue. As described previously for the sawtooth model, the intracellular current, on encountering such a barrier, must leave the intracellular space, cross the barrier, and reenter the intracellular domain on the other side. Depolarization should occur on one side of the barrier and hyperpolarization on the other side. Therefore, the barrier acts as a set of electrodes during the shock, becoming a secondary source of action potentials (Fig. 2-9). These secondary sources are important causes of depolarization and hyperpolarization throughout the myocardial tissues during a shock.²⁰ The resistive barriers act in a manner similar to that described for the sawtooth model. In this case, however, the resistive barriers represent larger discontinuities, which tend to increase with age and cardiac hypertrophy.¹⁰⁵

Figure 2-9 Computer model of secondary sources adjacent to scar elicited by single, large, pacing pulse.

Area to the right represents myocardium that contains a rectangular scar (stippled region). On the left is the blood pool with a pacing catheter in it. Note that a pacing pulse depolarizes not only tissue near the cathode but also tissue near the scar.

(From Street AM: Effects of connective tissue embedded in viable cardiac tissue on propagation and pacing: implications for arrhythmias. Durham, NC, 1996, Department of Biomedical Engineering, Duke University, p 134.)

Computer simulations have shown that the cathodal stimulation delivered to the myocardium near an oval scar results in three distinct activation fronts: the primary activation front and secondary fronts at the distal and proximal edges, which are generated by the exit and reentry of current from the intracellular and extracellular spaces¹⁰⁶ (see Fig. 2-8). Optical techniques have directly recorded changes in transmembrane potentials throughout a monolayer of ventricular myocytes.²⁰ Localized regions of depolarization and hyperpolarization coincided with discontinuities in the monolayer, resulting in slow conductance. Microscopic regions of depolarization and hyperpolarization have also been observed in isolated slabs of left ventricle, although their correlation with anatomic structures was not possible using optical mapping techniques because of the light-scattering properties of myocardium.¹⁰⁷ The significance of secondary sources was demonstrated in whole hearts by mapping of action potentials and determination of shock thresholds before and after the generation of a transmural lesion in the myocardial walls of dogs.⁹⁴ Generation of the lesion resulted in the development of a region of direct activation in the area of the lesion, in addition to the region of direct activation resulting from the stimulating electrode observed before the lesion. Furthermore, the strength of the shock required to cause direct activation in the area of the lesion was less than one half of that required before generation of the lesion (Fig. 2-10).

Figure 2-10 Isochronal activation maps.

Maps obtained after cathodal stimulation before and after creation of a transmural incision that caused secondary sources adjacent to the lesion. Isochrones are drawn at 5-msec intervals, timed from the onset of the S1 or S2 stimulus. Arrows represent direction of activation. Darkened regions represent areas directly activated by the stimulus. Black vertical bars represent the approximate location of the transmural incision. A, S1 stimulus delivered before incision. B, 75-mA S2 stimulus delivered before incision. C, 250-mA S2 stimulus delivered before incision. D, S1 stimulus delivered after incision. E, 75-mA S2 stimulus delivered after incision. F, Orientation of the long axis of myocardial fibers.

(From White JB, Walcott GP, Pollard AE, et al: Myocardial discontinuities: a substrate for producing virtual electrodes to increase directly excited areas of the myocardium by shocks. Circulation 97:1738-1745, 1998.)

The effects of secondary sources have major implications for the probability of successful defibrillation at different shock strengths in individual patients, particularly elderly patients, as well as the potential for reentry. Furthermore, the size and placement of operative lesions may play significant roles in the success of subsequent defibrillation.

Bidomain Model

The bidomain model is an extension of the one-dimensional cable model into two or three dimensions. That is, the extracellular and intracellular spaces are represented as single, continuous domains that extend in two or three dimensions and are separated by the highly resistive cell membrane⁹⁵ (Fig. 2-11). If the conductivities of the intracellular and extracellular spaces are constant in all directions, the model collapses to the one-dimensional cable model. Anisotropy refers to the manner in which conductivities change with the direction of myocardial fiber orientation. Clerc¹⁰⁸ showed that conductivity is higher in the direction parallel to the long axis of myocardial fibers (longitudinal) than in the direction perpendicular to the fibers (transverse) for both the intracellular and extracellular spaces. If conductivities change with direction but change the same for the intracellular and extracellular spaces, the bidomain model collapses to the one-dimensional cable model.

Figure 2-11 Circuit diagram of two-dimensional bidomain model.

Top of the resistor network represents the extracellular space; bottom of network represents the intracellular space. The symbol reΔx represents the extracellular resistivity in the x direction; reΔy represents the extracellular resistivity in the y direction; riΔx and riΔy represent the intracellular resistivities in the x and y directions. The rectangles represent the cell membrane. For a passive model, the rectangle would be replaced by a parallel resistor-capacitor network. For an active model, the rectangle would be replaced by a membrane ion model.

Studies have shown that the anisotropy ratio in ventricle is about 3 : 1 in the extracellular space and 10 : 1 in the intracellular space. When anisotropy ratios are used, the bidomain model begins to give new insights into how shocks change the transmembrane potential. Similar to the one-dimensional cable model, the bidomain model predicts that hyperpolarization occurs in tissue that lies under the extracellular anodal electrode. Likewise, depolarization occurs in tissue under the extracellular cathodal electrode. Unlike the one-dimensional cable model, the bidomain model also predicts that depolarization occurs along the long axis of the myocardial fibers at distances just a few millimeters from the anode. A similar effect is predicted to occur at the cathode, with hyperpolarization at distances of a few millimeters.¹⁰⁹ Therefore, the effect on the transmembrane potential near the shocking electrode is predicted to be much more complicated by the bidomain model than by the one-dimensional cable model.

The power of the bidomain model, however, is that it hypothesizes that there should be changes in the transmembrane potential, either hyperpolarization or depolarization, across the entire heart. In this model, the change in transmembrane potential elicited by the shock depends on the distribution of intracellular and extracellular current, which is affected by the change in potential gradient with distance, the distance from the electrode, and the orientation of the myocardial fibers. Experimental studies show a complex pattern of transmembrane potential changes during the delivery of a defibrillation shock, similar to those predicted by the bidomain model.^85,110,111

The transmembrane potential changes that occur during the delivery of a defibrillation shock can lead to the initiation of reentry and subsequent reinitiation of fibrillation after the shock. Reentrant circuits can be described by the mathematical concept of a phase singularity.¹¹² Phase can be used to describe the cardiac action potential, with 0 phase assigned to the upstroke of the action potential and 2Π phase assigned to the end of the action potential. A reentrant circuit can be thought of as a circle of phase starting at 0 (excitation) and continuing to 2Π (recovery). The reentrant circuit moves around a central point, called a phase singularity. Efimov et al.¹¹⁰ showed that a defibrillation shock can impose changes on the transmembrane potential extending from 0 phase through 2Π phase (Fig. 2-12). Thus, a reentrant circuit is generated and fibrillation induced. The authors suggest that induced reentrant circuits may be one way that defibrillation shocks can fail.

Figure 2-12 Creation of shock-induced phase singularity.

Left upper panel shows the change in transmembrane potential at the end of a +100/−200 V biphasic shock (i.e., at 15th msec of 16-msec shock), which resulted in a single extra beat. Scale is shown in millivolts, calibrated to a control 100-mV action potential. Point of phase singularity is indicated by the black circle. Middle upper panel shows a 5-msec isochronal map that depicts the initiation of the postshock spread of activation. The map starts at the onset of the 8-msec phase 2 of the shock (polarity reversal). Lower panels show optical recordings from several recording sites used to reconstruct the activation maps. Right upper panel shows a continuation of the reentrant activation shown in the middle panel. Reentrant activity self-terminates after encountering refractory tissue in the lower right corner of the field of view (see traces in right lower panel).

(From Efimov IR, Cheng Y, Van Wagoner DR, et al: Virtual electrode-induced phase singularity: a basic mechanism of defibrillation failure. Circ Res 82:918-925, 1998.)

The secondary source and bidomain models may not be mutually exclusive; rather, both may contribute to the changes in transmembrane potential. The exact mechanism by which an electrical pulse results in defibrillation remains incompletely understood at the level of the cell membrane and the ion channels. As the transmembrane potential attains values closer to the typical resting transmembrane potential than the usual minimum of −65 mV observed in fibrillating myocytes, this may allow the voltage-gated Na⁺ channels to recover sufficiently and the myocytes to regain full excitability.

Effect of Defibrillating Shock Field on Cellular Action Potential

The final common pathway of changes in the transmembrane potential caused by a defibrillation shock involves effects on the shape and duration of the cellular action potential. The shock can have one of three effects on the myocardium, depending on the local strength of the shock and its timing with respect to the local action potential. If the shock is delivered during the early plateau, there will be little or no prolongation of the action potential. If the shock is strong enough and is delivered relatively late during the action potential, it will initiate a new action potential. A shock that is strong enough but is delivered during early phase 3 of the action potential will modify and prolong an ongoing action potential without initiating an entirely new action potential (Fig. 2-13).

Figure 2-13 Transmembrane recordings from guinea pig papillary muscle.

Recordings show an all-or-none response to a weak field stimulus and action potential prolongation in response to a larger stimulus. A, Recordings that illustrate the response to an S2 stimulus of 1.6 V/cm oriented along the fibers. The S1-S2 stimulus interval for each response (milliseconds) is indicated to the right of the recordings. The responses are much different, even though the change in S2 timing was only 3 msec. An S1-S2 interval of 222 msec caused almost no response, whereas an interval of 225 msec produced a new action potential. B, Range of action potential extensions produced by an S2 stimulus generating a potential gradient of 8.4 V/cm oriented along long axis of myofibers; recordings from same cell as in A. The action potential recordings, obtained from one cellular impairment, are aligned with the S2 time. An S1 stimulus was applied 3 msec before phase 0 of each recording. The longest (230 msec) and shortest (90 msec) S1-S2 intervals tested are indicated beneath their respective phase 0 depolarizations. The S1-S2 interval for each response is indicated to the right.

(From Knisley SB, Smith WM, Ideker RE: Effect of field stimulation on cellular repolarization in rabbit myocardium: Implications for reentry induction. Circ Res 70:707-715, 1992.)

To defibrillate the heart successfully, a defibrillating shock (1) must stop most or all activation wavefronts on the heart and (2) must not reinitiate fibrillation. The extension of refractoriness hypothesis helps to explain how a shock can stop fibrillation. A shock can prolong the refractory period of an action potential without triggering a new potential if it is of sufficient strength and is delivered at an appropriate interval with respect to the upstroke.^23,113 If the first activation front that forms after a defibrillation shock encounters tissue with an extended refractory period, the front will be stopped because it cannot propagate into the region of refractory tissue.

Also, a defibrillation shock must not restart fibrillation. If only part of the front encounters tissue with an extended refractory period, only part of the front will be halted. The rest of the activation front will propagate forward and will eventually move into the area that could not be stimulated or that would not allow propagation (unidirectional block). This process of stimulating some tissue and creating unidirectional block in adjacent regions creates a reentrant circuit that eventually breaks down into fibrillation. The critical point is that point at which a critical shock strength intersects a critical level of refractoriness, leading to the formation of a reentrant circuit.^114–116

Optical mapping has revealed another type of critical point about which reentry can occur in response to the shock electric field.¹¹⁰ This type of critical point is formed when the shock electric field causes a large region of hyperpolarization immediately adjacent to a large region of depolarization. According to Efimov et al.,¹¹⁷ a biphasic shock waveform is more efficacious because the second phase of the shock obliterates the regions of hyperpolarization and depolarization so that the critical-point reentrant wavefront does not arise. Both types of critical points share two attributes; first, they produce a wavefront that appears immediately after the shock, and second, the wavefront forms a reentrant circuit.

Refibrillation Caused by Defibrillation Shock

Although activation wavefronts consistent with the two types of critical points are observed following shocks given during the vulnerable period of paced rhythm and after shocks much weaker than the DFT during VF, these wavefronts are usually not observed after shocks near DFT strength during VF.^118–123 Instead, earliest activation is not recorded until 50 to 60 msec after the shock, an interval called the isoelectric window, and wavefronts spread centrifugally away from this early site in a focal activation pattern. This pattern is present not only in electrical epicardial recordings (Fig. 2-14) but also in optical epicardial recordings (Fig. 2-15), which are not subject to saturation of the amplifiers by the shock as are electrical recordings. This activation pattern may represent epicardial breakthrough of an intramural reentrant circuit caused by a critical point. However, the same focal activation pattern has been observed in three-dimensional transmural electrical recordings with plunge needles, during which an isoelectric window was also observed intramurally.¹²² This window has also been observed in optical recordings of a thin layer of spared epicardial tissue in which the intramural myocardium was frozen.¹²³

Figure 2-14 Example of postshock cycles after failed shock in pig.

Defibrillation shock was of a strength near the defibrillation threshold (DFT), delivered from electrodes in the right ventricular apex and the superior vena cava. Recordings were made simultaneously from 504 electrodes distributed globally over the porcine ventricular epicardium. Electrode sites are indicated in gray on a polar projection, with the atrioventricular groove at the periphery and the left ventricular apex in the center. Anterior is at the bottom of the projection, and the left ventricle is to the right. Each panel shows in black the electrode sites at which an activation occurred at any time during each 10-msec interval. Numbers above the panels indicate the start of each 10-msec interval relative to the shock onset. Red arrows indicate the site of earliest recorded activation for each cycle. The first cycle appeared on the epicardium 64 msec after the shock at the anteroapical left ventricle. The second cycle (154 msec) arose on the epicardium in the same region as the first cycle and also propagated away in a focal pattern. The third (235 msec) and fourth (315 msec) cycles arose before the activation front from the previous cycle disappeared.

(From Chattipakorn N, Fotuhi PC, Ideker RE: Prediction of defibrillation outcome by epicardial activation patterns following shocks near the defibrillation threshold. J Cardiovasc Electrophysiol 11:1014-1021, 2000.)

Figure 2-15 Similar activation patterns of postshock cycle 1 after failed (A) and successful (B) shocks.

Maps from a pig heart are oriented as shown at top of figure; RV, right ventricle; LV, left ventricle; LAD, left anterior descending (coronary artery). Numbers above maps are times in milliseconds relative to defibrillation shock onset. Recording sites where action potential upstroke reached 50% of the action potential amplitude at any time during each 8-msec interval are black. Activation arose focally at the left ventricular apex where the shock electric field is weak and appeared 84 msec after the shock in A and 72 msec after the shock in B. Activation wavefronts propagated focally away from the apex toward the base in an organized pattern for both the failed and the successful shock episodes.

(From Chattipakorn N, Banville I, Gray RA, Ideker RE: Mechanism of ventricular defibrillation for near-defibrillation threshold shocks: a whole-heart optical mapping study in swine. Circulation 104:1313-1319, 2001.)

Following a failed shock slightly weaker than the DFT, the focal activation pattern is seen for several cycles before the less organized pattern of VF reappears (see Fig. 2-14). However, this same focal activation pattern can be observed for one or more cycles after a successful defibrillation shock (see Fig. 2-15, B). These findings suggest that the failed shock reinitiates VF by causing a trigger (the focus) that interacts with the substrate (the ventricular tissue). If the trigger focus lasts long enough for reentry to develop in the ventricular myocardium substrate, the shock fails. If the trigger focus halts before reentry is induced in the myocardial substrate, the shock succeeds. This concept raises the possibility that cardiac disease or drugs can alter DFT by altering the trigger, the substrate, or both. For example, conditions that increase the number of cycles of the focal trigger or that decrease the cycle length of the trigger might increase DFT. Similarly, conditions that allow more easily induced reentry in the substrate, such as an increased dispersion of refractoriness, could also increase DFT.

Although several mechanisms have been hypothesized, the cause of the focal activations after the shock is not definitively known, nor is it known why increasing the shock electric field decreases the number of focal postshock activations. One hypothesis is that the focus is not a true focus but arises from “tunnel propagation.”¹²⁴ According to this hypothesis, a small part of a VF activation wavefront is not extinguished but propagates intramurally during the postshock isoelectric window, where it is undetected by three-dimensional mapping because it is so small. When sufficient tissue recovers, it is able to break out as a large wavefront. This wavefront appears to arise focally because the small, slowly propagating tunnel of activation is not detected by mapping. According to this hypothesis, a biphasic waveform has a lower DFT than a monophasic waveform because it decreases the number and size of the tunnels for the same-strength shock. However, the finding at multiple focal cycles is difficult to explain with the tunnel propagation hypothesis; multiple focal cycles, not just one, are present after the shock (see Fig. 2-14).

Another hypothesis is that the focal activation arises where an intracellular calcium (Ca_i) “sinkhole” is present after the shock.¹²³ This hypothesis is based on optical recordings demonstrating that Ca_i decreases during the isoelectric period in the region where the focus will later appear (Fig. 2-16). After the formation of the sinkhole, the action potential in this region is immediately preceded by an increase in Ca_i, suggesting the potential arises from triggered activity. According to this hypothesis, a biphasic waveform has a lower DFT than a monophasic waveform because it causes less postshock heterogeneity in Ca_i.¹²⁵

Figure 2-16 Intracellular calcium (Ca_i) “sinkhole” and earliest postshock activation during unsuccessful shock in isolated rabbit heart.

A, Optical signals from the site marked by an asterisk in B. The isoelectric window is the time between the time of the defibrillation shock (red line) and the time when the postshock Vm tracing crosses F (fluorescence strength, black line). B, Optical maps after failed shock (200 V), with time of the shock as time 0. C, Consecutive isochronal activation maps (left), isochronal repolarization map (middle), and Ca_i maps (right). Ca_i sinkholes (white arrows) are seen at the same site before the onset of the repetitive activations; Act, activation; Rep, repolarization. D, Optical tracings from the center (top two tracings) and edge (bottom two tracings) of the Ca_i sinkhole. Horizontal line segments on the tracings represent the levels of F.

(From Hwang GS, Hayashi H, Tang L, et al: Intracellular calcium and vulnerability to fibrillation and defibrillation in Langendorff-perfused rabbit ventricles. Circulation 114:2595-2603, 2006.)

In one study the DFT after VF of 10 seconds was significantly decreased by flunarizine, a drug that prevents activity triggered by delayed afterdepolarization (DAD).¹²⁶ However, another study did not find that DFT after 10 seconds of VF was altered either by flunarizine or by pinacidil, a drug that prevents activity triggered by early afterdepolarization (EAD).¹²⁷ A third study found that DFT after 20 seconds of VF was not altered by flunarizine, but that DFT was reduced by flunarizine after 7 minutes of VF by 82%.¹²⁸ This suggests that after long-duration VF, but not short-duration VF, the focal postshock activation may arise from a DAD.

Another finding indicates that the mechanism of defibrillation may differ depending on the duration of VF. The earliest postshock activation after the isoelectric period arises primarily in the working myocardium following VF of about 30 seconds, but arises primarily in Purkinje fibers after VF of 180 seconds.¹²⁹ These differences in defibrillation after short- versus long-duration VF are probably related to the different characteristics of the activation wavefronts maintaining these short or long VFs.^130–132 It has even been reported recently that endocardial ventricular activation is not reentrant but arises focally from Purkinje fibers in up to 30% of cases during long-duration VF¹³³ (Fig. 2-17). This finding raises the startling possibility that to defibrillate long-duration VF, a shock must not only halt reentrant activation wavefronts, but also halt focal, possibly triggered, activity.

Figure 2-17 Recordings and activation times for 15 cycles after ventricular fibrillation in dog.

A, Soon after onset of VF; B, after 5 minutes of VF. Left ventricular endocardial basket recordings from 64 unipolar electrodes are shown at the top of each panel, with a right ventricular (RV) unipolar recording and ECG lead (aVR) below them. Below the ECG are the ventricular activation times recorded at the 64 basket electrodes during the same period, with each short vertical line representing an activation. At the bottom of each panel, the activations for all 64 electrodes are shown on a single line. The 15 cycles are numbered at the top of B. In A, endocardial activations are present throughout the cardiac cycles, consistent with reentry. In B, however, activation is highly synchronized with activation present during only a small part of the cardiac cycle, so that 14 large temporal gaps are present between the 15 cycles of activation in which no activations are present. This finding is inconsistent with reentry near the left ventricular endocardium. In both A and B, RV activation and ECG appear disorganized, as expected for VF.

(From Robichaux RP et al: Periods of highly synchronous, non-reentrant endocardial activation cycles occur during long-duration ventricular fibrillation. J Cardiovasc Electrophysiol 21:1266-1273, 2010.)

Thus, although significant advances have been made in our understanding of the mechanisms of defibrillation, it is clear that there is still much more to be learned.