Intramural Virtual Electrodes in Ventricular Wall
Effects on Epicardial Polarizations
Background— Intramural virtual electrodes (IVEs) are believed to play an important role in defibrillation, but their existence in intact myocardium remains unproven. Here, IVEs were detected by use of optical recordings of shock-induced transmembrane potential (Vm) changes (ΔVm) measured from the intact epicardial heart surface.
Methods and Results— To detect IVEs, isolated porcine left ventricles were sequentially stained with a Vm-sensitive dye by 2 methods: (1) surface staining (SS) and (2) global staining (GS) via coronary perfusion. Shocks (2 to 50 V/cm) were applied across the ventricular wall in an epicardial-to-endocardial direction during the action potential plateau via transparent mesh electrodes, and shock-induced ΔVm were measured optically from the same epicardial locations after SS and GS. Optical recordings revealed significant differences between ΔVm of 2 types that became more prominent with increasing shock strength: (1) for weak shocks, SS-ΔVm were larger and faster than GS-ΔVm; (2) for intermediate shocks, cathodal GS-ΔVm became multiphasic, whereas SS-ΔVm remained monophasic; and (3) for strong shocks, cathodal GS-ΔVm became uniformly negative, whereas SS-ΔVm typically remained positive. The radical differences in the shape and polarity of SS and GS polarizations can be explained by the contribution of subepicardial IVEs to optical signals. Histological examination revealed a dense network of collagen septa in the subepicardium, which could form the IVE substrate.
Conclusions— Intramural virtual electrodes are reflected in optical measurements of shock-induced ΔVm on the intact epicardial surface. These IVEs could be a result of microscopic resistive discontinuities formed by collagen septa.
Received August 12, 2003; de novo received December 5, 2003; revision received February 5, 2004; accepted February 6, 2004.
The current theory of defibrillation is based on the assumption that shocks produce changes of transmembrane potential (ΔVm) in the majority of myocardium, but whether ΔVm occur in the intramural tissue layers is not well known. In a recent study, optical mapping was used to measure shock-induced ΔVm (“virtual electrodes”) on the transmural surface in isolated wedge preparations of the left ventricular (LV) wall.1 It was found that shocks induced widespread ΔVm across the transmural surface, which could possibly account for defibrillation. However, the extrapolation of these findings to the intact myocardium is limited by the differences between electrical properties of the cut transmural surface in wedge preparations and the intact LV wall. Therefore, the goal of the present study was to examine whether intramural virtual electrodes (IVEs) exist in the intact myocardium. It is hypothesized that, because of integration of optical signals from different tissue depths, intramural virtual electrodes can be detected in optical recordings of ΔVm measured from the intact LV epicardium. This hypothesis was examined in isolated coronary-perfused LV preparations that were stained with a Vm-sensitive dye by 2 different methods: (1) by staining only the epicardial tissue layers and (2) by global tissue staining (GS) via coronary perfusion. Radically different shock-induced Vm responses were observed on the epicardium of the same preparations stained by the 2 techniques. These differences provide evidence for the existence of subepicardial intramural virtual electrodes in the intact LV wall.
Porcine LV preparations (length, ≈4.5 cm; width, ≈2 cm; thickness, ≈1.8 cm) were obtained as described previously.1 Preparations were stained with the dye di-4-ANEPPS twice: first by surface staining (SS) and then GS. In the first method, preparations were immersed in 20-μmol/L dye solution for ≈1 hour. Stained preparations were placed in a tissue bath with a glass window for optical mapping (Figure 1A) and perfused for the rest of experiment through a coronary artery with Tyrode’s solution1 containing 15 mmol/L of 2,3-butanedione monoxime. After completion of optical measurements (see below) in SS tissue, the preparations were stained globally by arterial injection of a 5-mL bolus of 20-μmol/L dye solution, and measurements were repeated.
Preparations were paced at a 500-ms interval with a bipolar electrode. To induce action potentials (APs) with a small delay, a 1.5- to 2-V/cm shock (duration=5 ms) was applied across the LV wall in the epicardium-to-endocardium direction via 2 transparent mesh electrodes. A series of test shocks (duration=10 ms) with strengths of ≈2 to 50 V/cm were applied through the same electrodes 40 ms after the pacing shock. A pause of 3 to 5 minutes was allowed for tissue recovery after ≥20-V/cm shocks. Both SS and GS series of measurements lasted ≈1 hour each. To check for measurement reproducibility, each series was normally completed by application of a shock of initial strength. These comparisons showed that polarization and activation maps were highly reproducible in both SS and GS measurements and that they were not affected by strong shocks. Reproducible measurements were also obtained over an ≈2-hour interval (duration of entire experiment) in GS-stained tissue.
Optical signals were recorded from epicardium through a shock electrode (Figure 1A) that was ≈4 mm from the epicardium. Because the electrode was out of focus, it did not obscure the preparation, which was verified by imaging a test pattern (Figure 1B). Shock-induced ΔVm were measured with an optical system described previously1 with a 200-W mercury/xenon lamp and a 16×16 photodiode array at a sampling rate of 5 kHz/channel and spatial resolution of 1.18 mm/diode. A shock-induced ΔVm was measured as the difference between a linear regression fit of AP plateau and the Vm level at a given time (Figure 1C) and normalized by AP amplitude (APA). For weak shocks (E<≈3 V/cm), ΔVm during the 0.2- to 9-ms interval was fitted with an exponential function by use of nonlinear regression (Origin 7, OriginLab Corp), and the decay time constant was calculated. Signals measured from SS preparations had relatively low signal-to-noise ratio. To improve signal quality, excitation light was focused on the central epicardial region with dimensions of ≈6×10 mm2, which was separated by ≈6 mm from the edges of the preparation. Statistical data were expressed as mean±SD. Differences were compared by the 2-tailed t test. Results were considered statistically significant at a value of P<0.05.
To evaluate the depth of dye penetration into the tissue, stained preparations were cut across the wall, and the transmural tissue surface was imaged under a fluorescence microscope. The viability of myocardium was accessed in 3 preparations after the end of the experiments by staining transmural tissue slices with triphenyl tetrazolium chloride. No ischemic damage was detected. Subepicardial tissue structure was examined in 5-μm-thick transmural slices of formalin-fixed myocardium that were stained for collagen by the picrosirius red technique.2
Shock-induced ΔVm were measured in 7 LV preparations. In each preparation, there were significant differences in ΔVm waveforms measured after SS and GS, and these differences became more prominent with increasing shock strength. Activation patterns were not substantially different (not shown).
ΔVm Produced by Weak Shocks
Figure 2 shows ΔVm produced by 3-V/cm shocks in SS and then GS tissue. In measurements of both types, cathodal shocks produced only positive ΔVm (B), and anodal shocks produced only negative ΔVm (C). Whereas the SS and GS polarizations were qualitatively similar, there were significant differences in their magnitudes and time courses, with SS-ΔVm being substantially larger and faster than corresponding GS-ΔVm (D and E).
Similar ΔVm were observed in all preparations (n=7). All responses to weak shocks had the same simple monophasic shape, but SS-ΔVm were significantly larger and faster than GS-ΔVm. For cathodal shocks with E=4.7±1.6 V/cm, SS-ΔVm and GS-ΔVm averaged over the mapping area were 54±20% and 16.4±10% APA (P<0.001), respectively; a large difference was also observed during anodal shocks. The time constant of exponential fit (R2=0.96±0.05) during weakest cathodal shocks (E=2.9±0.3 V/cm) was ≈23% larger for GS-ΔVm than for SS-ΔVm (2.7±0.4 versus 2.2±0.7 ms, n=5, P<0.05). An even larger difference (≈52%) between SS and GS time constants was measured during anodal shocks.
ΔVm Induced by Intermediate-Strength Shocks
Increasing shock strength caused significant changes in the ΔVm shape. Figure 3 demonstrates the effects of ≈14-V/cm shocks. During cathodal shocks, SS-ΔVm remained positive and, in most cases, monophasic (C). In contrast, the majority of GS-ΔVm became biphasic, whereby an initial Vm rise was followed by a decline (traces 1, 2, 4, and 5). Moreover, the polarity of GS-ΔVm could change from positive to negative during the shock (trace 5). ΔVm maps (A) showed that SS-ΔVm were always larger than GS-ΔVm.
During anodal shocks (B and D), both SS- and GS-ΔVm remained negative. The SS-ΔVm became biphasic, whereby initial Vm decline was followed by Vm increase (D). The GS-ΔVm either remained monophasic (not shown) or became slightly biphasic. ΔVm maps (B) show that the SS-ΔVm were larger than GS-ΔVm at the shock beginning (t=1 ms), but these differences attenuated toward the shock end.
Similar findings were obtained for shocks with E=16±5 V/cm in 7 preparations. In SS tissue, cathodal ΔVm were always positive, and anodal ΔVm were always negative. In GS tissue, anodal ΔVm were always negative, whereas cathodal ΔVm could change their polarity from positive to negative during the shocks. In most cases, SS-ΔVm were larger than GS-ΔVm. Thus, average cathodal SS- and GS-ΔVm were 21±16% and 4±4% APA (P<0.001) at shock beginning and 44±25% and −9±14% APA (P<0.001) at shock end. For anodal shocks, large differences between average SS- and GS-ΔVm were observed at the shock beginning (P<0.001), but they became closer at the end (P=NS).
The threshold for occurrence of biphasic negative ΔVm was lower in SS than in GS measurements. The threshold was determined as the field strength when ΔVm at the shock end became more positive than ΔVm in the middle (t=5 ms). During anodal shocks, this threshold was 9±3 and 16±5 V/cm (n=7, P<0.01) for SS- and GS-ΔVm, respectively.
ΔVm Induced by Strong Shocks
Further increase of shock strength resulted in more radical differences between SS and GS polarizations. Figure 4 illustrates the effects of ≈39-V/cm shocks. The GS-ΔVm became negative and biphasic during both cathodal and anodal shocks (C and D). In contrast, the shape of SS-ΔVm was different for different shock polarities and different locations. During the anodal shock (D), SS-ΔVm were strongly biphasic, with a rapid rise after the initial decline. The SS-ΔVm map (B) was uniformly negative at the shock onset but, because of rapid Vm rise, became predominantly positive at the shock end.
During the cathodal shock (C), SS-ΔVm were either positive monophasic (traces 3 to 5) or biphasic with changing polarity (traces 1 and 2). The SS-ΔVm maps (A) showed both positive and negative polarizations at the shock beginning but uniformly positive polarization at the shock end.
Qualitatively similar results were obtained in all 7 preparations in response to shocks with E=38±6 V/cm. Shocks of both polarities produced mostly negative biphasic GS-ΔVm, whereas SS-ΔVm were of several waveform types, described above.
Figure 5 presents data on maximal and minimal ΔVm measured over the entire mapping area in all preparations after SS and GS. Maximal and minimal ΔVm were achieved during cathodal and anodal shocks, respectively. In general, SS-ΔVm were larger than GS-ΔVm. At the shock beginning, the largest maximal and minimal SS-ΔVm were ≈50% and −100% APA (A), respectively, whereas corresponding GS-ΔVm did not exceed 25% and −70% APA (B). Large differences were also measured at the shock end, when largest maximal and minimal SS-ΔVm were ≈100% and −150% APA (C), respectively, whereas corresponding GS-ΔVm were ≈50% and −75% APA (D). The maximal SS-ΔVm remained positive for all shocks (A and C). In contrast, maximal GS-ΔVm became negative for shocks stronger than ≈30 V/cm (B and D).
Depth of Dye Staining
Figure 6 illustrates distributions of di-4-ANEPPS in subepicardial regions of SS and GS preparations. In SS preparations (A, left), dye fluorescence was confined to a narrow subepicardial layer. The average profile of fluorescence intensity (right) reveals a nearly exponential decay, with intensity decreasing by 80% within 0.23 mm from the epicardium. From all SS preparations (n=5), the average distance of 80% fluorescence decay was 0.25±0.06 mm. In the GS preparations (n=4), dye distributions were nearly uniform (B).
Subepicardial Collagen Distribution
Figure 7 presents fluorescence images of collagen distribution in subepicardial regions of LV at different magnifications. A dense layer of collagen with a thickness of 50 to 100 μm can be seen at the epicardium (C). From this layer, collagen protrudes into the subepicardial muscle layers. Below this region, there are multiple collagen septa, some of them reaching a length of ≈1 mm, oriented primarily parallel to the epicardium (A and B). At some locations (A, upper right), however, a change in collagen septum orientation could occur. In addition to collagen septa, there are discontinuities related to blood vessels of varying diameters. Qualitatively similar collagen distribution was observed in 2 LV preparations.
This work examines the existence of intramural virtual electrodes by comparing optical ΔVm signals measured from intact LV wall after SS and GS with a Vm-sensitive dye. It was found that ΔVm of 2 types measured from the same locations exhibited the following differences: (1) for weak shock, SS-ΔVm were larger and faster than GS-ΔVm; (2) for intermediate shocks, cathodal GS-ΔVm became multiphasic, whereas SS-ΔVm remained monophasic; and (3) for strong shocks, cathodal GS-ΔVm became uniformly negative, whereas SS-ΔVm typically remained positive. The radical differences between ΔVm of 2 kinds reflect the existence of intramural virtual electrodes.
Source of Optical Signals
It is known that optical signals measured from cardiac tissue represent a weighted average of contributions from tissue layers at different depths. The depth of signal collection is determined by a combination of multiple factors, including light absorption and scattering, wavelengths of excitation and emission light, spatial resolution, etc.3,4 Direct experimental determination of the collection depth is difficult, and most reliable estimates were obtained with mathematical models. Simulations in a Monte Carlo model of light propagation3 estimated that 80% of fluorescent photons measured over an area of 1 mm in diameter originate from a depth of ≈1.2 mm; a similar estimate was obtained with a diffusion approximation of light propagation.4
Measurements of transmural dye distribution performed here indicate that optical signals measured from SS tissue originate from a much thinner tissue layer. In such tissue, an 80% decrease in dye fluorescence was observed at a depth of ≈0.25 mm. Assuming no light absorption or scattering, this value would represent the collection depth of optical signals in SS preparations. Because of significant light absorption and scattering by the tissue, however, the collection depth is expected to be smaller than that and, therefore, much smaller than the depth of signal collection in GS preparations.
Intramural Virtual Electrodes
Electrical shocks are believed to produce intramural virtual electrodes that play an important role in defibrillation, but their existence is still debatable. A recent study1 found that shocks produced widespread ΔVm on the transmural surface of isolated LV wall, which could potentially account for defibrillation. Extrapolation of this finding to the intact LV wall, however, is limited by the differences in boundary conditions existing at the cut tissue surface from those inside the intact LV. Therefore, the present study examined the existence of intramural ΔVm in the intact myocardium.
To evaluate ΔVm generated in the deep tissue layers, optical signals were measured twice from the same locations, first after dye staining on the tissue surface and then after GS via coronary perfusion. Because of different collection depths, the difference between signals of 2 types reflects the contribution of deep cell layers to shock-induced ΔVm. Analysis of these signals revealed both quantitative and qualitative differences between SS- and GS-ΔVm observed at different shock strengths.
During weak shocks, there were differences in magnitudes and time constants of SS- and GS-ΔVm (Figure 2). Such differences can be explained by the classic cable theory, according to which the ΔVm magnitude rapidly decays and the time constant increases with the distance from the tissue-bath interface.5 Therefore, depth signal averaging should produce smaller and slower ΔVm in GS tissue than SS tissue, which corresponds to the present observations.
During stronger shocks, profound differences in the shape of SS and GS responses were found. One such difference was observed during shocks of moderate strength when cathodal GS-ΔVm became biphasic, whereas SS-ΔVm at the same locations remained monophasic (Figure 3C). Moreover, at some locations, GS-ΔVm even changed their polarity from positive to negative toward the shock end. Such changes in the GS responses and their differences from SS responses cannot be explained by the continuous cable theory. The only plausible explanation is that they are a result of intramural virtual anodes formed under the epicardial surface.
The most radical differences between GS- and SS-ΔVm were observed during the strongest shocks when cathodal SS-ΔVm remained mostly positive but GS-ΔVm became uniformly negative (Figures 4C and 5⇑C). Such negative cathodal GS-ΔVm indicate 2 things: (1) that there are intramural virtual anodes and (2) that these virtual anodes became much stronger than the primary positive polarizations produced at the epicardium-bath interface. The latter is not possible in a system with linear membrane response, in which primary polarizations are always larger than secondary polarizations.6 The reason for the dominance of virtual anodes is the negative asymmetry in the membrane response during the AP plateau, when negative polarizations are much larger than positive polarizations.7,8 It should be added that even cathodal SS-ΔVm exhibited negative polarizations at some locations, indicating that there were intramural virtual anodes immediately under the epicardial surface (within <250 μm).
Epicardial GS-ΔVm showed important similarities with ΔVm measured previously from the transmural cut surface.1 Thus, in both cases, cathodal polarizations became uniformly negative at E≈30 V/cm; maximal positive and negative ΔVm were comparable as well. These findings are in contrast to epicardial measurements in rabbit hearts, in which only positive ΔVm were measured on the cathodal side of the hearts.9 The difference between the results of this and the present study could be because of differences in animal species and differences in shock strengths.
Mechanism of Intramural Virtual Electrodes
In general, intramural virtual electrodes can arise via 2 different mechanisms, one dependent on spatial gradients of the shock field10,11 and the other dependent on spatial gradients in resistive tissue properties.6,12 It is highly unlikely that there were significant gradients of electrical field over a relatively small epicardial mapping area faced with a much larger plate electrode oriented parallel to the epicardium. Therefore, the most likely reason for intramural ΔVm was nonuniform resistive tissue properties.
Spatial gradients in tissue resistivity can result from different anatomic structures of cardiac muscle, including fiber rotation, microscopic tissue discontinuities such as collagen septa, and blood vessels. Most likely, IVEs measured here were not a result of fiber rotation, because it occurs predominantly in the plane parallel to the epicardial surface.12 Therefore, these IVEs were most likely related to microscopic tissue discontinuities. Histological examination of subepicardial tissue layers revealed a dense network of collagen septa and blood vessels (Figure 7). These collagen septa could have lengths of up to ≈1 mm, and they were oriented primarily parallel to the epicardial surface, which is similar to the distribution of collagen septa in the LV of human hearts.13 Such collagen septa create an efficient substrate for IVE formation, with positive and negative ΔVm produced at opposite septa sides. Because of the negative asymmetry of Vm response, spatial averaging of optical signals from microscopic IVE results in negative ΔVm observed during cathodal shocks.
Shock-induced ΔVm had complex shapes that varied with shock strength. On the qualitative level, all waveforms can be explained by spatial and temporal summation of Vm responses of 3 basic types. The ΔVm of the first type are passive ΔVm that have a monophasic exponential shape, similar to ΔVm observed during the weakest shocks (Figure 2). The ΔVm of the second type are asymmetrical, with negative ΔVm exceeding positive ΔVm. Responses of this type have been described in various cardiac preparations.7,8,14–16 Summation of linear surface ΔVm with asymmetrical intramural ΔVm can explain the biphasic shape of positive GS-ΔVm observed during cathodal shocks of moderate strength (Figure 3C). The negative ΔVm asymmetry is also an important factor in the explanation of why GS-ΔVm were uniformly negative during both anodal and cathodal strongest shocks (Figure 4, C and D).
In addition to linear and asymmetrical ΔVm, a prominent role was played by Vm responses of the third type, characterized by a positive shift of Vm after an initial negative polarization. The ΔVm of this type were previously described in cell cultures.7,16,17 Recently, it was found that the occurrence of such ΔVm was paralleled by cell uptake of the membrane-impermeable dye propidium iodide in the areas of negative but not positive ΔVm,18 indicating that such biphasic ΔVm are the result of electroporation at sites of negative polarization. Membrane electroporation was implicated in the detrimental effects of shocks, such as loss of excitability and mechanical function,19 generation of postshock arrhythmias, and failure of defibrillation at very strong shocks.20 In the present work, signs of electroporation were observed in both SS and GS measurements, but the threshold shock strength for their occurrence was significantly lower in SS than GS recordings, indicating that SS measurements provide a more sensitive indicator of detrimental shock effects.
Implications and Limitations
The demonstration of intramural virtual electrodes in the intact LV wall validates the current theory of defibrillation, which postulates that the electrical field causes Vm changes in a critical mass of myocardium. Whereas the existence of IVEs was proven here for a relatively thin subepicardial tissue layer, the fact that the same intramural tissue structure is found in much deeper tissue layers indicates that similar IVEs should be produced by shocks there as well.
This study was supported by National Institutes of Health grant HL-67748.
Fast VG, Sharifov OF, Cheek ER, et al. Intramural virtual electrodes during defibrillation shocks in left ventricular wall assessed by optical mapping of membrane potential. Circulation. 2002; 106: 1007–1014.
Fast VG, Rohr S, Ideker RE. Non-linear changes of transmembrane potential caused by defibrillation shocks in strands of cultured myocytes. Am J Physiol. 2000; 278: H688–H697.
Cheek ER, Ideker RE, Fast VG. Non-linear changes of transmembrane potential during defibrillation shocks: role of Ca2+ current. Circ Res. 2000; 87: 453–459.
Efimov IR, Aguel F, Cheng Y, et al. Virtual electrode polarization in the far field: implications for external defibrillation. Am J Physiol. 2000; 279: H1055–H1070.
Trayanova N. Concepts of ventricular defibrillation. Phil Trans Roy Soc London A. 2001; 359: 1327–1337.
Spach MS, Dolber PC. Relating extracellular potentials and their derivatives to anisotropic propagation at a microscopic level in human cardiac muscle: evidence for electrical uncoupling of side-to-side fiber connections with increasing age. Circ Res. 1986; 58: 356–371.
Tung L, Kleber AG. Virtual sources associated with linear and curved strands of cardiac cells. Am J Physiol. 2000; 279: H1579–H1590.
Fast VG, Cheek ER. Optical mapping of arrhythmias induced by strong electrical shocks in myocyte cultures. Circ Res. 2002; 90: 664–670.
Cheek ER, Fast VG, Nonlinear changes of transmembrane potential during electrical shocks: role of membrane electroporation. Circ Res. 2004; 94: 208–214.
Gold JH, Schuder JC, Stoeckle H, et al. Transthoracic ventricular defibrillation in the 100 kg calf with unidirectional rectangular pulses. Circulation. 1977; 56: 745–750.