Influences of Anisotropic Tissue Structure on Reentrant Circuits in the Epicardial Border Zone of Subacute Canine Infarcts

Transcription

1 182 Influences of Anisotropic Tissue Structure on Reentrant Circuits in the Epicardial Border Zone of Subacute Canine Infarcts Stephen M. Dillon, Maurits A. Allessie, Philip C. Ursell, and Andrew L. Wit Excitation in the epicardial border zone of 3-5-day-old canine infarcts was mapped with an array of 192 bipolar electrodes during sustained ventricular tachycardia. Reentrant circuits were found in which activation occurred around long lines of apparent conduction block based on the criterion that excitation on opposite sides of the lines occurred with marked disparity in time. When the lines of apparent block were functional (i.e., occurred only during tachycardia and not during sinus rhythm or ventricular pacing) they were oriented parallel to the long axis of epicardial muscle fiber bundles. Isochrones distal to the lines were oriented parallel to them because widely separate sites within these isochrones were activated nearly simultaneously. This suggested that excitation not only occurred around the lines of block but also slowly across them. This slow activation occurred transverse to the long axis of the myocardial fibers and therefore might result because of the anisotropic tissue properties. To test this hypothesis, the epicardial border zone was stimulated during sinus rhythm through electrodes around its margin and at the center of the recording array. Activation transverse to the myocardial fibers in regions where lines of block occurred during tachycardia was slow, whereas it was rapid parallel to fibers' orientation. During tachycardia electrograms along the lines of apparent block had long durations and were fractionated, a characteristic that can also result from activation transverse to the myocardial fiber long axis. Therefore, we propose that the parallel orientation of the muscle bundles in the epicardial border zone is an important cause of ventricular tachycardia because activation transverse to myocardial fibers is sufficiently slow to permit the occurrence of reentry. (Circulation Research 1988;63: ) Nonsustained and sustained ventricular tachycardia can be induced in the canine heart by programmed premature ventricular stimulation or overdrive pacing during the first week after complete ligation of the left anterior descending coronary artery.'- 3 Tachycardias often From the Departments of Pharmacology and Pathology, College of Physicians and Surgeons of Columbia University, New York, and the Department of Physiology, R(jksuniversiteit Limburg, Maastrict, The Netherlands. Supported by Program Project Grant HL 30557, Grant IR0I- HL 31393, and Postdoctoral Training Award HL from the Heart, Lung, and Blood Institute of the National Institutes of Health and Grant from the Foundation for Medical Research (FUNGO) in The Netherlands. Address for correspondence: Dr. Andrew L. Wit, Department of Pharmacology, College of Physicians and Surgeons, 630 West 168th Street, New York, NY Received July I, 1987; accepted February 10, originate in a thin sheet of parallel-oriented muscle fiber bundles that survive on the epicardial surface of the myocardial infarct the epicardial border zone In 1982 we presented results of experiments in which we mapped excitation of the epicardial border zone during nonsustained ventricular tachycardias induced by stimulation and showed activation patterns consistent with reentrant excitation. 3 Now we report the results of our continuing studies on excitation during sustained tachycardia. In addition to showing that reentry occurs in the epicardial border zone during sustained tachycardia, 6-9 our studies were also designed to determine some of the anatomical and electrophysiological properties of the reentrant circuits and, therefore, some of the underlying mechanisms causing reentry. In particular, the electrophysiological characteristics of the

2 regions around which activation appeared to circulate were closely examined. Our results show that these regions are often not formed by a "gross" anatomical obstacle but may be the result of the functional properties of the myocardial fibers caused by their parallel orientation. Analyses of electrogram characteristics and activation patterns at the "center" of the reentrant circuits along with the results of experiments in which the epicardial border zone was activated by stimulated wavefronts propagating in different directions all suggest that anisotropic conduction properties may have an important influence on the characteristics of reentry. 15 Materials and Methods Surgical Production of Myocardial Infarction Using sterile techniques, myocardial infarction was produced in 40 mongrel dogs, weighing kg, by a two-stage ligation of the left anterior descending coronary artery () approximately 1 cm from its origin. 16 The chest was then closed in layers and an airtight seal established. Thirty-five dogs survived and were studied on the 3rd to the 5th day after the coronary occlusion. We selected this experimental model rather than other models of canine ventricular tachycardia because reentry occurs at 3-5 days in a narrow rim of paralleloriented myocardial fibers on the epicardial surface of the infarct." This model, therefore, was predicted to be useful for determining the effects of tissue anisotropy on reentrant circuits. All care and use of animals conformed to the guidelines of the American Physiological Society and AAALAC. Dillon et al Anisotropic Reentry 183 Electrophysiological Study Electrode arrays and recording instrumentation. On the day of the electrophysiological study the dogs were anesthetized with pentobarbital sodium (15-30 mg/kg) and ventilated by a positive-pressure respirator. The chest was opened by a median sternotomy, the pericardium was opened, and a pericardia! cradle was made to support the heart. An electrode array containing 192 bipolar electrodes was then sutured over the anterior and lateral surface of the left ventricle. The electrodes were made from silver disks with a diameter of 1.0 mm. A distance of 1.0 mm separated the perimeter of each pole of a bipolar pair, and there were 3.5 mm between bipolar pairs. The recording electrodes were embedded in a 0.3-mm-thick square sheet of a rubber-like material (Biomer) that was 5.5 x 5.5 cm in size. When this sheet was sutured to the left ventricle, one side was lined up adjacent to the and extended from the base of the heart toward the apex. We refer to this side as the margin of the array. The opposite side, which we refer to as the lateral left ventricular margin (LL), actually faced the apical aspect of the lateral left ventricle. This location of the electrode array usually corresponded to the location of the infarct, and therefore all bipolar electrodes were concentrated on the infarct epicardial border zone. We did not map impulse propagation in surrounding noninfarcted myocardium. By positioning all electrodes in the infarcted region we were able to increase the spatial resolution of the activation maps 3-6 times greater than that which has been previously reported The increased spatial resolution assisted us in delineating some of the conduction properties of the epicardial border zone during reentry. Placement of this electrode array on the left ventricle did not seem to influence perfusion or electrical activity of the epicardial region. Stable electrograms were recorded for the entire 4-6 hour period of the experiment without any obvious deterioration of the electrical signals or changes in conduction. The electrode leads were led into preamplifiers with automatic gain controlled by a microprocessor. The input stage of the preamplifiers had a band width of 10-1,000 Hz. The signals were multiplexed and sampled at a frequency of 2,000 samples/sec (each signal sampled at 0.5 msec intervals) and digitized by an 8-bit analog-to-digital converter. The digitized signals were converted to Millersquared code by means of an Ampex pulse code modulation system and were stored on wide-band tape. The digitized signals were then transferred to the disk of a PDP 11/34 computer and the electrograms displayed in groups of seven on a Tektronix 4012 graphics terminal. The moments of activation were marked by hand with a cursor (see below). The heart was stimulated through rows of bipolar electrodes located at different sites on the ventricles. Each pole had a 1.0 mm diameter, and the bipoles were 1.5 mm apart. One row of four bipolar electrodes was embedded in a narrow sheet of Biomer 10 mm wide and 3 cm long and sutured adjacent to the on the right ventricle. The other stimulating electrodes were embedded in the recording electrode matrix: a row of four bipolar electrodes toward the basal margin extending over a distance of 3 cm, another row of four bipolar electrodes along the LL margin, also 3 cm long, and a bipolar pair in the center of the electrode matrix. The two rows of bipolar electrodes at the base and LL margins were oriented in an attempt to produce activation along and across the major epicardial fiber axis. The row of electrodes was also used to produce activation along the major axis. After the recording electrode array and stimulating electrodes were sutured in place, the chestcavity was covered with a sheet of plastic to retain heat and moisture. Experimental protocol. The following experimental protocol was used. First, the left ventricle was driven at cycle lengths (in msec) of 350, 300, 250, 280, 180, and 160 from each of the rows of stimulating electrodes around the edges of the electrode array and from the central electrodes. The stimulus

3 184 Circulation Research Vol 63, No 1, July 1988 pulse was 2 msec in duration and 2-4 times greater than diastolic threshold. By stimulating through a row of electrodes, a broad wavefront was initiated. As a result of stimulating through the different electrodes, impulse propagation occurred in different directions relative to the long axis of the myocardial fiber bundles over the epicardial border zone of the infarct. Analysis and comparison of activation maps of excitation waves moving in different directions enabled us to evaluate the influence of the anisotropic structure of the border zone on impulse propagation." Ventricular tachycardia was sometimes induced at the shorter stimulus cycle lengths (see below). After these data were obtained, programmed premature stimulation was used to initiate tachycardias. The ventricles were driven at the longest basic cycle length that ensured capture (usually msec but occasionally 250 msec), and single premature stimuli were interpolated at progressively decreasing coupling intervals until they could not excite a premature response. Both the basic drive stimulus and the premature stimulus had a duration of 2 msec and were 2-4 times greater than diastolic threshold. The basic drive stimuli were omitted after the premature stimuli to allow repetitive activity or tachycardia to manifest itself. Programmed premature stimulation was done from all stimulation sites in case tachycardia might be induced preferentially from one site. If sustained tachycardia was not initiated by this protocol, rapid "burst" pacing was used. The ventricles were stimulated in bursts of 4-10 impulses at progressively decreasing cycle lengths from 180 msec to 140 msec. Electrogram recordings were made both during the procedures to induce tachycardia and during the tachycardia. Data analysis: Determination of activation times and construction of activation maps. Standard criteria (the largest negative slope or highest amplitude of a deflection 20 ) were used to mark activation times. However, during the course of the data analysis it became obvious that these criteria were sometimes inadequate, in particular at sites where extracellular recordings were characterized by longduration, polyphasic waveforms containing a number of deflections. We relied on previously published studies by Spach and Dolber 21 on the origin of these complex waveforms, which provides a theoretical and experimental basis for their interpretation. According to their data from unipolar recordings with fine-tipped electrodes in human atria, such polyphasic waveforms represent "zigzag" conduction in the direction transverse to the long axis of fibers in which side-to-side connections are few. 21 Similar polyphasic shapes can also be recorded with bipolar electrodes 22 although it is not possible to tell which of the two poles record all, none, or some of the activity that generates the waveforms. In any case, these waveforms indicate that electrical activity persists in the immediate vicinity of the bipoles for the duration of the electrogram and there is no unique activation time at these sites Therefore, whereas activation times at sites where single deflections were recorded could clearly be determined in relation to a reference time, activation times at sites where polyphasic waveforms were recorded were arbitrarily assigned to the "major" deflection (highest amplitude or fastest downstroke) for the purpose of plotting the activation maps. However, our interpretation of these maps were dependent on the morphology and location of these waveforms and the consideration that they represented transverse propagation during a time interval of activation (see "Results"). After the activation times were plotted, isochrones were drawn at 10-msec intervals. Regions in which a high density of isochrones occurred (>3 per interelectrode distance of 3.5 mm) were indicated by thick black lines ("bunched" isochrones). Although such regions are often interpreted as indicating conduction block, our interpretation of the meaning of these lines (slow activation or block) was assisted by analysis of the waveforms in these regions and the isochronal patterns around them; for example, bunched isochrones associated with polyphasic waveforms indicated slow activation transverse to the long axis of the muscle fibers while uniphasic deflections indicated real block. 23 Histology. The exact location of the recording electrode array on the anterolateral left ventricle was marked by placing sutures around its perimeter after the electrophysiological study was finished. The heart was then removed and placed in 10% neutral buffered formalin. After complete fixation, the heart was rinsed in water and the surface blotted dry. The electrode was then replaced in its original position and the location where reentrant circuits were found from analysis of the activation maps, also marked with sutures. The region of the anterolateral left ventricular wall where the electrode array was located was then dissected free from the heart (full thickness of the wall). This block was processed for histological studies as we have previously described. 311 From the histological sections, we determined the thickness of the epicardial sheet of ventricular muscle that survived over the infarcts at specific sites by counting the number of surviving cell layers along the length of each section (900-1,000 sites per experiment)." The orientation of the long axis of the muscle fibers relative to the direction of impulse propagation was also determined. Results Characteristics of Tachycardia We report here only data from the experiments in which sustained ventricular tachycardia was induced by either burst pacing or by single premature ventricular stimuli (eight different dogs) (Table 1). Some tachycardias in each experiment lasted at least 3-5 minutes, and in each of the experiments persisted

4 Dillon et al Anisotropic Reentry 185 TABLE 1. Experiment Characteristics of Sustained Tachycardia No. of different tachycardias Mode of induction P,B B P B B B P P Tach. Cl. (msec) Times induced Maximum duration (min) Activation pattern D,S N S,N D D,N N N D P, premature stimulus; B, "burst" stimuli; D, double reentrant loop; S, single reentrant loop; and N, no reentrant loop. until they were terminated by either premature or overdrive ventricular stimulation. These tachycardias are designated as being sustained. Sustained tachycardias according to our definition did not degenerate into ventricular fibrillation when no attempt was made to terminate them (with electrical stimulation) although fibrillation occurred in two dogs after stimuli were applied during tachycardia. Each tachycardia had a uniform cycle length and QRS morphology but sometimes during the initial two to five impulses, cycle length and QRS morphology were changing (Figure 1). Stable tachycardia cycle lengths ranged from msec in the different experiments (Table 1). In six experiments several different sustained tachycardias with different cycle lengths and QRS morphologies were induced. Each tachycardia in each dog was initiated at least five times during the course of an experi- FlGURE 1. Lead II electrocardiograms recorded from four different experiments in which sustained tachycardia was initiated. In Panels A, C, and D the last sinus beat prior to a period of rapid stimulation, indicated by the horizontal black bar, is shown at the left. A break in the top trace represents 10 stimulated impulses. Stimuli were applied through the electrodes at the left lateral margin of the recording array. In Panel B, tachycardia was initiated by a single stimulated premature impulse at the arrow, delivered during basic drive through the electrodes. The sustained tachycardia is shown at the right in each trace and the cycle length of tachycardia is indicated in milliseconds.

5 186 Circulation Research Vol 63, No 1, July Photomicrographs of histological sections taken from different regions of an epicardial border zone. In both panels the epicardial surface is at the top. In A the surviving epicardial muscle fibers are widely separated by edema and the extracellular matrix while in B the muscle fibers are more closely packed together. In both cases, however, the muscle fibers are parallel-oriented. FIGURE D» '- * *- - «i lm ment. QRS morphology and tachycardia cycle length were reproducible each time a tachycardia was initiated. In each of these experiments, nonsustained tachycardias were also initiated (tachycardias that terminated spontaneously within 1 minute). Activation of Epicardial Border Zone During Sinus Rhythm The epicardial border zone in experiments in which reentrant circuits were found consisted of a narrow rim of surviving muscle over the infarct in contact with normal myocardium around its margins (Figure 2)." The lack of intramural connections, from below, to the surviving sheet, except at its margins, was evident by the activation patterns during sinus rhythm (Figure 3). Whereas in noninfarcted left ventricle, transmural breakthrough may result in initial activation of the epicardial surface at the center of the anterior left ventricle with some radial spread, in transmural infarcts, activation initially occurred around the margins of the epicardial border zone (within the 0-20-msec isochrones in Figures 3A and 3B) and then proceeded toward the center (as indicated by the arrows) where the wavefronts from all the margins eventually collided (the regions within the msec isochrones in Figures 3A and 3B were activated last). In the experiment shown in Panel C, activation also occurred from the margins of the epicardial border zone (from the left and above); part of the apical region was not activated during sinus rhythm because here the infarct extended all the way to the epicardial surface and there were few surviving muscle fibers (this region was also not activated during tachycardia). The locations of the lines of block and slow activation during tachycardia (see next section) are also shown on these maps by the dashed lines. Much of these regions were activated homogeneously during sinus rhythm, and the marked dispar-

6 Dillon et al Anisotropic Reentry 187 LL LL LL FIGURE 3. Each panel shows an activation map of the epicardial border zone during one sinus beat from three different experiments. The left margin of each map is the margin of the electrode array adjacent to the left anterior descending coronary artery (). The right margin of each map is the margin of the electrode array on the lateral left ventricle (LL). The top margin is along the base and the bottom at the apex. On each map is plotted the activation times at each of the recording sites indicated by small numbers. Isochrones are drawn at 10 msec intervals and labeled with larger numbers. Arrows indicate the sequence of isochrones and thus, the direction of activation. In addition, the dashed lines on each map point out the location of the lines of apparent block that occurred during the reentrant excitation in Figure 4.

7 188 Circulation Research Vol 63, No 1, July 1988 ity of activation times between adjacent electrodes that occurred during tachycardia was not evident. It is also apparent from these activation maps that there were no gross anatomical obstacles in the sheet of surviving muscle, the kind that might result from replacement of myocardial fibers by connective tissue forming in the healing infarct. From the spatial arrangement of our electrodes, "holes" in the sheet greater than about 10 mm 2 could probably be detected but not smaller anatomical defects. We point out the absence of gross obstacles because later it is proposed that reentry occurs because of functional properties of muscle fibers in the sheet and not because of gross anatomical obstacles around which activation circulates. Activation of Epicardial Border Zone During Sustained Ventricular Tachycardia The patterns of activation of the epicardial border zone during 17 morphologically distinct sustained tachycardias in eight dogs were determined. We have designated a pattern of activation as being reentrant if the sequence of isochrones delineates continuous movement of a wavefront through a pathway that eventually leads back to the original location of the wavefront. Demonstration of reentry according to this definition does not prove that the reentry is the cause of the tachycardia. Such proof requires additional postulates (see "Discussion"). Several different general patterns of activation of the epicardial border zone occurred during the sustained tachycardias. We have subdivided them as follows: 1) circular activation in the pattern of double (reentrant) "loops," described as "figure of eight" by El-Sherif and his coworkers 24 (four tachycardias in four experiments), 2) circular activation in the pattern of single reentrant loops (four tachycardias in two experiments), and 3) activation patterns in which no reentrant excitation was apparent on the epicardial surface (nine tachycardias in five experiments). Different patterns occurred during tachycardias with different QRS morphologies in the same heart (Table 1). For example, in experiment 1 both single- and double-loop reentry occurred and in experiment 3 single-loop reentry and no epicardial reentry occurred during tachycardia (Table 1). Only one pattern of activation was associated with a particular tachycardia QRS morphology. This pattern of activation and the associated QRS morphology were not dependent on the site of stimulation that initiated tachycardia. For our initial description of activation, the sequence of isochrones in the epicardial border zone during four of the sustained tachycardias is shown in Figure 4. It became apparent during the detailed analysis of this data that the isochronal sequence, while accurately describing the relative times of activation of different regions, is not always an accurate indicator of the exact pathways of activation within the reentrant circuit. This will be pointed out in subsequent figures. Figure 4A shows the isochronal sequence during one cycle of a tachycardia in an experiment in which the doubleloop reentrant pattern occurred. The tachycardia was initiated by burst pacing from the lateral margin electrodes (activation maps during initiation are described later). The electrocardiogram recorded during this tachycardia is shown in Figure 1A. The tachycardia persisted for 5 minutes, and this activation pattern was repeated in a nearly identical manner during each beat of the 30 beats of tachycardia that were analyzed. The left-hand margin of the activation map (electrode array) in this and all subsequent figures was adjacent to the (see "Materials and Methods"). In Figure 4A, activation times progress from the 10-msec isochrone at the lateral (LL) margin in directions both upward toward the base and downward toward the apex; the isochrones indicate that two wavefronts moved in opposite directions. The upper activation wave moved toward the margin (isochrones 10-80), which it reached after msec. The lower activation wave also moved toward the margin (isochrones 10-90) and arrived there after 90 msec. The two activation waves, one from the base and one from the apex, coalesced, and activation proceeded back toward the LL margin (isochrones ), completing the reentrant cycle. The time for activation of the epicardial border zone during one reentrant circuit is shown by the arrows on the ECG in Figure 1A. The same pattern of excitation was then repeated. Although we did not map activation in normal ventricle around the epicardial border zone, the Q wave of the ECG occurred several milliseconds after the activation wave reached the LL margin, suggesting that the exit from the reentrant circuits to the rest of the ventricles might have been in this region. The epicardial border zone in this experiment, in the region where the reentrant circuits were located, was composed of 4-60 layers of ventricular muscle cells overlying the transmural infarct. The narrowest region (fewest cells thick) was located where the common central pathway of the two circuits was located (4-20 cell layers). The thickest regions were located toward the margins of the infarct with normal myocardium, which coincided with the margins of the electrode array. The infarct reached the epicardial surface at only 14 of the 950 sites where cells were counted and each of these sites was less than 1.0 mm in diameter. Intramural surviving cells or transmural connections were not seen. Also, the long axis of the myocardial fibers was in the direction from the to LL margins of the electrode array and all fibers were parallel to each other (see Figure 2). This orientation was consistently found for each experiment and was also described in detail in a previous publication." The map shown in Figure 4B is an example of another activation pattern during tachycardia in which there were two reentrant circuits. Tachycardia was initiated by a single stimulated premature

8 Dillon et al Anisotropic Reentry 189 iso ITO no no too LL 100 no TO y SO 130 MO' IM LL LL FIGURE 4. Each panel shows an activation map of the epicardial border zone during one beat from each of the sustained tachycardias shown in Figure 1. The left margin of each map is the margin of the electrode array adjacent to the left anterior descending coronary artery (). The right margin of each map is the margin of the electrode array on the lateral left ventricle (LL). The top margin is along the base and the bottom at the apex. On each map b plotted the activation times at each of the recording sites, indicated byjmall numbers. Isochrones are drawn at 10 msec intervals and labelled with larger numbers. Arrows indicate the sequence of isochrones and, thus, the direction of movement of activation. Descriptions of the activation patterns are in the text. M impulse at the margin. The electrocardiogram recorded during this tachycardia is shown in Figure IB. The map illustrates how the pattern of activation in different experiments with double reentrant loops was quite different (compare with Figure 4A). The sequence of isochrones shows that activation progressed as a single wavefront from the LL margin toward the margin (isochrones ) although a large component of the broad wavefront stopped at 50 msec (described in detail later). The activation wave then split, and one wavefront progressed up toward the base (black arrows) and then from left to right along the base to the LL margin, which it reached after 210 msec (completing the upper circuit). The second wavefront moved down along the apical margin toward the LL margin (from left to right, indicated by the black arrows) and also reached the LL margin after 210 msec (completing the lower circuit). These two activation waves coalesced. The region where these reentrant circuits was located also consisted of a narrow (1-35 cell layers) rim of parallel-oriented ( toward LL margin) surviving muscle cells with no transmural connections except around the peripheral margins. The infarct reached the epicardial surface at 23 sites within the shaded area on the activation map. Electrograms were not recorded in this region during tachycardia, although they were recorded during sinus rhythm. The map shown in Figure 4C is from an experiment in which only one reentrant circuit was apparent within the field of the recording electrodes. The electrocardiogram is shown in Figure 1C. The tachycardia was initiated by stimuli applied at the LL

9 190 Circulation Research Vol 63, No 1, July 1988 margin. Activation in the time frame that is shown began within the 10-msec isochrone at the LL margin of the electrode array near the base of the ventricle. The sequence of isochrones progresses from right to left along the base to the margin, which was activated after 100 msec. After reaching the margin, activation occurred back toward the LL margin (from left to right; isochrones ). In this experiment, electrograms were not recorded from a large part of the apical region (see Figure 3C). At the few sites where electrograms were recorded toward the apex, they had very low amplitudes and a broad but uniform morphology. Histology of the region where no electrograms were recorded showed a completely transmural infarct extending to the epicardial surface with few surviving muscle fibers, whereas in the region where the reentrant excitation occurred there was a layer of parallel-oriented surviving muscle cells, 5-30 cells thick, over the transmural infarct. An activation map of the epicardial border zone during one of the tachycardias in which a reentrant circuit was not seen is shown in Figure 4D. This tachycardia was initiated by a burst of stimuli applied through the left lateral electrodes. Activation began at the margin (0-msec isochrone) and spread across toward the opposite side. The wavefront reached the LL margin after msec, which is 70 msec less than the tachycardia cycle FIGURE 5. Characteristics of electrograms recorded along the line of apparent block. At the top, the isochrones from the activation map described in Figure 4C are shown. The letters indicate recording sites of electrograms shown in Panels A, B, and C. Below each letter is a number indicating the designated activation time at that site. Panel A shows electrograms recorded at sites A-I, along the line of apparent block. The arrows indicate the point on the electrogram taken as the moment of activation plotted on the map (see text). Panels B and C show electrograms recorded on either side of the line of apparent block. length of 160 msec (the electrocardiogram recorded during this tachycardia is shown in Figure ID). A surviving epicardial sheet of muscle cell layers thick corresponded to the region where activation was mapped. The isochronal activation patterns of the reentrant circuits shown in Figures 4A-4C are roughly in the shape of ovals. The long axis of each oval extends in the general direction from the margin to the LL margin of the electrode array. In the maps shown in Panels A and C, and in the lower reentrant circuit in B, isochrones are rotating around long lines (indicated by the thick black line(s) in each panel) extending in the direction of the long axis of the oval. In the upper circuit in B, the line is bent in the form of an S, with the two end components of the line oriented in the -to-ll direction, connected by a vertical line. Horizontal lines in the activation maps are oriented parallel to the long axis of the myocardial fibers that composed the epicardial border zone. This characteristic was determined in each experiment from the histological analysis of these regions. Vertical components of the lines (Figure 4B; midsection of upper line) are perpendicular to the long axis of the muscle fibers and were found in regions where there were few viable fibers. The extent and location of these lines are defined by the marked differences in activation times at the recording sites on either side of them.

10 Dillon et al Anisotroplc Reentry 191 FIGURE 6. Effects of activation patterns on electrograms recorded along the line of apparent block that was described in Figure 5. The calibration at the top is 70 msec. Panel A shows these electrograms during tachycardia, Panel B shows them during sinus rhythm, and Panel C shows them during stimulation at a cycle length of 190 msec at the margin. Arrow in C is stimulus artifact. The biggest differences in activation times across each line is at its center, while the differences decrease toward the ends of the lines. According to convention, interpolation of a number of isochrones between the markedly disparate activation times results in the thick black lines, which are actually formed by closely bunched isochrones (see "Materials and Methods")- Such marked differences in activation time between adjacent electrodes have sometimes been interpreted as indicating conduction block between areas on either side of the line. 5, However, a large number of isochrones bunched closely together to form the thick black lines in Figures 4A-4C might also result from very slow propagation (across the lines). Therefore, the lines are referred to as lines of "apparent" block. Analysis of the waveforms recorded in regions where a number of isochrones were interpolated enabled slow activation across the lines to be distinguished from real conduction block. Characteristics of electrograms near lines of apparent block during sustained tachycardia: Block or slow activation? At many recording sites that were not in the vicinity of the lines of apparent block, electrograms recorded during tachycardia were characterized by relatively discrete deflections. Electrograms at the lines were more complex. Figure 5A shows an example of the characteristics of these electrograms recorded at sites A-I on the accompanying activation map of the "single loop" reentry (previously described in Figure 4C). Moving from right to left (A-I on the map), the electrogram duration and number of components increased as the line was reached (compare electrograms A, B, and C with D and E in Panel A). Toward the opposite end of the line, electrogram (H and I) duration and fractionation decreased. The marked difference in activation times between electrodes located on opposite sides of the line of apparent block is shown in Panels B and C. Electrograms J and L were recorded from locations above the line and K and M from locations below the line. These electrograms had relatively discrete deflections. Between these sites long-duration, fractionated electrograms were recorded (D and E). Identical electrogram characteristics were also found in other experiments; long-duration, fractionated electrograms were located along the lines of apparent block that were oriented in the horizontal direction, while discrete electrogram deflections occurred on either side of the lines. In contrast, at vertical lines of block (Figure 4B for example), longduration, fractionated electrograms were not found. We interpret the fractionated waveforms that we recorded to mean that there was slow activation transverse to the long axis of the muscle bundles in these regions, which were nonuniformly anisotropic 21 (see "Materials and Methods" section on data analysis). Thus, in Figure 5, activation occurred slowly across the line of apparent block (from site J to K, for example) rather than stopping at the line (real block). The orientation of the interpolated isochrones (lines of apparent block) assists in this interpretation; the lines are parallel to the fibers,

11 192 Circulation Research Vol 63, No I, July 1988 A B ( _ FIGURE 7. The effect of assigning different activation times to fractionated electrograms on the location of the line of apparent block. Electrograms are shown from sites L, E, and M in Figure 5. In Panel A activation times are assigned only to electrogram L (37) and M (125). To the right of the electrogram traces, the electrodes with the activation times are indicated by the circles and the isochrones are drawn. The isochrones between 40 and 130 msec are interpolated and form a line of apparent block. Panel B shows isochrones when an activation time is assigned to the first major component of the fractionated electrogram E, in addition to electrograms L and M. Because the activation time of electrogram E is 44 msec, isochrones must be interpolated between 50 and 130 msec, still forming a line of apparent block. In Panel C the moment of activation was assigned to the second major component of the fractionated electrogram E (125 msec). When this time is used, isochrones are still interpolated between the 40- and the 130-msec isochrones but the line of apparent block shifts upward. In panel D activation is considered to occur throughout the duration of electrogram E and activation is assigned every 10 msec. This leads to the construction of isochrones every 10 msec between the 37-msec activation time of the first major deflection and the 125-msec activation time of the second, and a line of apparent block formed by the bunched isochrones. indicating that slow activation across the lines is transverse to the fibers. The histology of the section taken from the region of fractionated activity (sites C-E) is shown in Figure 2A and suggests that interconnections among the parallel-oriented muscle fibers may be decreased (compare with Figure 2B, which shows a section taken near the lateral margin of the circuit, site I). The long duration and fractionated characteristic of electrograms during tachycardia represented a dramatic change from the characteristics of these electrograms during sinus rhythm or ventricular stimulation. Figure 6A shows again these electrograms along the line of apparent block during the tachycardia described in Figure 5. Figure 6B shows recordings from the same sites during sinus rhythm. Although the marked fractionation that occurred at most recording sites during tachycardia was decreased, some fractionation remained, especially at sites C and D. It is important to note that during sinus rhythm, activation occurred transverse to the long axis of the myocardial fibers at these sites (see Figure 3C). However, in Figure 6C, during stimulation at the margin and activation in the direction parallel to fiber orientation most of the fractionation disappeared and electrogram duration decreased. As described in "Materials and Methods," a unique activation time was assigned to the polyphasic waveforms for the purpose of plotting the isochronal maps, although these electrograms probably represent a time interval of slow activation extending from the first to the last deflection. However, Figure 7 shows that, no matter what the criteria are, assigning an arbitrary activation time does not alter the interpretation of the isochronal map as long as it is not the intention to define the activation pattern within the small (1 mm) region where the polyphasic waveform was recorded. Figure 7 shows activation between recording sites L, E, and M across the line of apparent block during the single-loop reentry discussed in Figure 5. Activation times were assigned unambiguously to electrodes L and M at the arrows because discrete deflections were recorded. When we initially ignored recording E, for which no discrete moment of activation could be assigned, and interpolated isochrones between sites L and M at 10-msec intervals, the thickly bunched isochrones appeared (Panel A); the difference in activation times was large compared to the isochrone intervals, so a relatively large number of isochrones were interpolated (giving rise to the thick line). In Panels B and C, different activation times were assigned to electrode E, one corresponding to the initial deflection (Panel B) (and used in the activation map in Figures 4C and 5), and the other to the second deflection (Panel C). No matter which was chosen, it was necessary to interpolate a number of isochrones between this moment of activation and the moment of activation of the adjacent electrode. The line of apparent block shifted slightly but the pattern of activation was unaltered. When activation was considered to occur throughout the whole duration of electrogram E, as

12 Dillon et al Anisotropk Reentry 193 BO A 60 B FIGURE 8. Comparison of activation patterns during stimulation and during tachycardia. The format for each map is the same as described in the legend to Figure 4. In A, activation during tachycardia is shown at the left (cycle length 190 msec). The circled activation times are discussed in the text. Electrograms recorded at the four sites within isochrone 130 are shown at the right. In B, the activation pattern during stimulation at the margin (S), at a cycle length of 190 msec, is shown. The same five recording sites as in A, are circled and electrograms recorded at four of the sites are shown at the right. The dashed line on the map indicates the location of the line of apparent block that occurred during the tachycardia (apparent block does not occur in this region during stimulation). In C, activation during stimulation at the margin of the electrode array at the base (cycle length 190 msec) (S) is shown. Again, the same five recording sites as in A and B are circled and electrograms are shown at the right. The dashed line on the map indicates the location of the line of apparent block that occurred during tachycardia. we interpret these results, an activation time was assigned every 10 msec as shown in Panel D. This still resulted in the closely bunched isochrones between electrograms L and M. Therefore, longduration, fractionated electrograms result in closely bunched isochrones (lines of apparent block) that indicate slow activation irrespective of how activation times are chosen. Characteristics of activation adjacent to lines of apparent block. The patterns of the isochrones immediately adjacent to the lines of apparent block (interpolated isochrones) provided further support for our conclusions that the polyphasic waveforms indicated slow activation across the lines, transverse to fiber orientation, and that these lines did not indicate conduction block. For example, in the map shown in Figure 8A (previously described in Figure 4C), activation above the line (toward the base) in general appears to have progressed from lateral (LL) to margin (from right to left), as shown by the sequence of isochrones from 10 to 90, while activation below the line is in the general direction from the to lateral margin according to the sequence of isochrones 100 to 160 (from left to right). However, the isochrones immediately adjacent to the line on this side (isochrones 130 and 140) are parallel to the line, and isochrone 130 extends for a distance of about 20 mm along its length. Activation of the four recording sites within the 130-msec isochrone (circled activation times in Panel A) occurred within 8 msec of each other. In fact, within the 130-msec isochrone, one recording site (121 msec) was activated earlier than the adjacent sites on either side of it (see electrograms at the

13 194 Circulation Research Vol 63, No 1, July 1988 LL 70 FIGURE 9. Comparison of activation patterns during stimulation and during tachycardia. The format for each map is the same as described in the legend to Figure 4. In A, activation during tachycardia is shown (cycle length 150 msec). The circled activation times are discussed in the text. In B, the activation pattern during stimulation at the lateral left ventricular margin (LL) of the electrode array (S) at a cycle length of 150 msec is shown. The dashed lines indicate the location of the lines of apparent block that occurred during tachycardia. (Lines of apparent block do not occur during stimulation.) In C, activation during stimulation at the margin of the electrode array at the (S) is shown, and in D, activation during stimulation from the center of the electrode array (S) (cycle length 150 msec) is shown. The dashed lines in both C and D indicate the location of the lines of apparent block that occurred during tachycardia. LL right in Panel A). Activation at recording sites in isochrone 140 occurred within 7 msec of each other. If there was conduction block at the line and activation was occurring around it (from left to right), conduction velocity would need to be more than 2 m/sec to account for the small difference in activation times over the 20 mm distance within isochrones 130 and 140 and could not account for the earlier activation at the 121-msec site within isochrone 130. Therefore, activation must have occurred across the line of interpolated isochrones as we concluded from the electrogram analysis. Other examples of the characteristics of isochrones next to lines of apparent block (interpolated isochrones) are shown in the activation maps in Figure 9A (previously described in Figure 4A) and Figure 10A (previously described in Figure 4B). In the upper reentrant circuit toward the base in Figure 9A, the 10- and 20-msec isochrones are parallel to the line and extend about 16 mm along its length. Widely disparate sites (circled activation times within the 10-msec isochrone) were activated nearly simultaneously so conduction only in a rightto-left direction, which would be necessary if there was block at the line, was unlikely. In the pathway common to both circuits, the electrodes adjacent to the lower line in isochrone 120 were also activated with a time difference of only 3-7 msec (circled activation times), which we explain by activation occurring across this line (from below); electrograms recorded here were also polyphasic (not shown). In Figure 10A, isochrones 130 and 140 adjacent to the horizontal component of the upper line (at the left) are also parallel to it, and the small differences in activation times along the line indicate conduction across it (see circled activation times), although the isochrones distal to the lower line are not parallel to it. Relation between regions of slow activation in reentrant circuits (lines of apparent block) and anisotropic conduction properties of epicardial border zone. The observations that the lines of apparent block were often parallel to the orientation of the long axis of the myocardial fibers, that extracellular waveforms recorded in these regions were polyphasic and had a long duration, and that

14 Dillon et al Anisotropk Reentry ZOO 1Z0 DID 1M \3XJ \1*J lit 1W UJ III 111 FIGURE 10. Comparison of activation patterns during stimulation and during tachycardia. The format for each map is the same as described in the legend to Figure 4. In A, activation during tachycardia is shown (cycle length 210 msec). The circled activation times are discussed in the text. In B, activation during stimulation at the margin (S) is shown (cycle length 210 msec). The dashed lines in this and subsequent panels indicate the location of the lines of apparent block that occurred during tachycardia. In C, activation during stimulation at the lateral left ventricular margin (LL), and in D, activation during stimulation at the center of the electrode array are shown (cycle length 210 msec). isochrones adjacent to these lines were often parallel to them led us to conclude that there was slow activation across the lines resulting in the large time difference between activation on either side of them. This resulted in merging of wavefronts that crossed the line with other wavefronts that conducted around the line as shown by the diagrams in Figure 11. The center of the reentrant circuit, therefore, was smaller than indicated by the isochronal maps alone. Activation that occurred across the line was transverse to the long axis of the fibers and was slow because of the effects of anisotropy This conclusion was further supported by the results of experiments in which the epicardial border zone was stimulated during sinus rhythm from the margins and from the center so that activation occurred in different directions. We compared activation times in the regions where lines of very slow activation were located during tachycardia with activation during stimulation from the different sites. The comparison of the activation maps during tachycardia with maps during stimulation showed that activation within the parallel isochrones adjacent to the lines of apparent block (during tachycardia) occurred too rapidly to be explained by conduction only around the lines and was a result of transverse activation across them (discussed above). In the experiment shown in Figure 8, the stimuli were applied through electrodes on the margin of the epicardial border zone (at S in Panel B) (during sinus rhythm) at the same cycle length as tachycardia (190 msec). Stimulation from this location caused activation to occur from the toward the LL margin as shown in Panel B, parallel to the long axis of the myocardial fibers. Activation time was compared between the same electrodes during stimulation (Figure 8B) as during tachycardia (Figure 8A). Five recording sites were designated as reference sites for determining activation times under the two different conditions. These recording sites are circled on the activation maps in Figures 8A and 8B. In Figure 8A, during tachycardia, total activation time between the two most separated recording sites was 24 msec from a time of 105 msec for the electrode at the left to 129 msec for the electrode at the right. There was also a 4-msec difference between activation of the left and

15 1% Circulation Research Vol 63, No 1, July 1988 B 30 vation time difference between the far left and far right reference sites was 50 msec (from 4 to 54 msec), while there was a 22-msec difference between the reference electrodes previously in the 130-msec tachycardia isochrone (from 32 to 54 msec). The same four electrograms that are shown in Panel A are also shown during stimulation at the right in Panel B, emphasizing the increased activation time between them during stimulation. Therefore, acti- J7Q- FIGURE 11. Diagrams illustrating the possible effects of anisotropy on reentrant circuits. The isochrones from the reentrant circuits discussed in Figures 4C and 4 A are shown in Panels A and D. In Panel A the large black arrows point out a sequence of excitation that appears to be progressing around the long line of apparent block, indicated by the horizontal thick black line, as discussed in the description of Figure 4C. From an analysis of electrogram characteristics and isochrones adjacent to this line, we have proposed that there is slow activation across it (see text), and therefore, the center of the circuit is smaller as indicated by the smaller black arrows within the shaded circle. This region is enlarged in Panel B where the thick black line of apparent block has been resolved into closely bunched isochrones, indicating slow activation in the part of the circuit transverse tofiberorientation. In Panel D the large black arrows point out a sequence of excitation that appears to be progressing around two long lines of apparent block, indicated by the horizontal thick black lines as discussed in the description of Figure 4A. From a closer analysis of electrogram characteristics and isochrones adjacent to the lines of apparent block we have proposed that there is very slow activation across these lines (see text). The smaller arrows in Panel D within the shaded circles indicate these proposed activation patterns. As a result, the center of the reentrant circuits may be small. These proposed circuits are shown enlarged in Panel C (upper circuit) and Panel E (lower circuit). According to this interpretation, there is rapid activation in the parts of the circuit where activation occurs parallel to the long axis of the fibers and slow activation in the parts of the circuit where activation occurs transverse. The slow activation is indicated by resolving the thick black lines of apparent block into the closely bunched isochrones. right designated electrodes in the 130-msec isochrone (from 125 to 129 msec). The four electrograms recorded within isochrone 130 are also shown to the right of the map. During stimulation (Figure 8B) the isochrones were in the vertical direction indicating conduction from left to right. The line of interpolated isochrones that occurred during tachycardia was not evident (the dashed line in Figure 8B indicates its location during tachycardia). The acti-

16 vation along this row of electrodes during tachycardia was much more rapid than conduction in the left-to-right direction during stimulation, consistent with the interpretation that activation during tachycardia was occurring across the line and not just around it. The same region was then activated in a direction transverse to the myocardial fiber long axis by stimulating from the electrodes along the basal margin. This activation map is shown in Figure 8C. Activation perpendicular to the long axis occurred much more slowly over the first msec than activation parallel to the long axis, as can be seen from the bunching together of the isochrones (compare Panel C with Panel B). Activation was very slow through the same region where the line of apparent block was located during tachycardia (indicated by dashed line) although it was not as slow as during tachycardia. Further away from the basal stimulating electrodes, activation of the epicardial border zone also occurred from the margin toward the center (from left to right) and seemed to merge with the wavefront conducting from the base. Since this region was not activated only perpendicular to fiber orientation, activation (after 80 msec) occurred more rapidly. Figure 9 shows the results of activating the epicardial border zone by stimulating at different sites during sinus rhythm (150-msec cycle length) in the experiment in which double-loop (figure of eight) reentry occurred. We previously pointed out the characteristics of the isochrones adjacent to the lines of apparent block in Panel A. When stimulating through the lateral electrodes, activation occurred obliquely from the LL toward the margin (Figure 9B). The recording sites that were in the 10-msec isochrone during tachycardia and that were activated nearly simultaneously (circled in Figure 9A), were activated with a 23-msec difference during stimulation (circled in Figure 9B). (The location of the lines during tachycardia are indicated by the dashed lines in Panels B-D). Recording sites that were activated nearly simultaneously along the lower line of apparent block, within the 120- msec isochrone during tachycardia (circled in Figure 9A) were activated with a 27-msec difference during stimulation at the margin (circled in Figure 9C), which caused activation to occur in the left-to-right direction (Figure 9C). During stimulation from the margin (Figure 9C) activation occurred parallel to the long axis of myocardial fiber orientation resulting in rapid and uniform activation in regions where lines of apparent block were located during tachycardia (dashed lines). In this experiment, stimulation from the base did not cause activation transverse to the long axis in these regions because of invasion of the epicardial border zone by stimulated wavefronts from the margin that apparently conducted outside the infarcted area. To activate the muscle fibers in the Dillon et al Anisotropic Reentry 197 transverse direction, the epicardial border zone was stimulated through the central electrodes during sinus rhythm in the region where the two lines of apparent block that served as the common pathway for the two reentrant circuits was located (the location of these lines during tachycardia is indicated by the dashed lines in Figure 9D). During central stimulation at the site S, there was rapid activation toward the and LL margins parallel to the myocardial fiber orientation; these margins were activated after 30 and 40 msec, respectively. There was much slower activation toward the base above and toward the apex below transverse to fiber orientation; both of these margins were activated after about 80 msec. The regions toward the base and apex that were activated slowly during central stimulation were activated more rapidly during stimulation at the margin (see Panel C), when activation occurred from left to right, parallel to myocardial fiber orientation. The results also show that there was slow activation perpendicular to the long axis of the myocardial fibers in the regions where the lines of apparent block occurred during tachycardia; the circled electrodes adjacent to the dashed lines were activated nearly simultaneously, as during tachycardia. However, the regions of slowest anisotropic conduction during stimulation in Panel D do not correlate exactly with the location of the lines during tachycardia. It is possible that some of this lack of correlation results from the influence of the stimulus on propagation. Since the location of the lines was relatively close to the stimulating electrodes, some fibers in these regions may have been excited directly by the stimulus and as a result, some of the slow conduction was obscured. These results (Figures 8 and 9) show that slow activation (approximately 0.05 to 0.2 m/sec) occurred in the regions of horizontal lines of apparent block when wavefronts moved perpendicular to the long axis of the myocardial fibers. Since activation was more rapid (approximately 0.3 to 0.9 m/sec) in the same regions in the direction parallel to fiber orientation, the anisotropic tissue structure is implicated as a cause of slow activation rather than depressed transmembrane potentials (slow responses). However, this was not the case for vertical components of lines of apparent block such as shown by the small arrows in the activation map in Figure 10A (previously described in Figure 4B); lines occurred irrespective of the site of stimulation as shown by the thick black lines in Figures 10B and 10C during stimulation from the and lateral margins (activation parallel to fiber orientation) and in Figure 10D during stimulation from the central electrodes. Thus vertical regions of lines of block cannot be caused by slow activation transverse to fiber orientation. In this experiment, regions where horizontal components of lines of apparent block were located had similar responses to stimulation as we described for the experiments in Figures 8 and 9; activation

17 198 Circulation Research Vol 63, No I, July LL ZOO FIGURE 12. Activation maps of the epicardial border zone showing initiation of tachycardia by a single premature stimulus applied near the margin. The electrocardiogram is shown in Figure IB. The format of each of the activation maps is the same as the one described in Figure 4. Panel A shows activation during the last driven impulse initiated at S prior to the premature stimulus. The dashed lines in this and subsequent panels indicate the location of the lines of apparent block during the sustained tachycardia (activation map shown in D). Panel B shows activation map of the stimulated premature impulse, from its initiation to 190 msec. Panel C shows the continuation of propagation of this premature impulse to the margin and then, again around the circuit. Panel D shows the activation map of one beat of the sustained tachycardia. LL, lateral left ventricular margin. was homogeneous and relatively rapid when it occurred parallel to fiber orientation (Panel B) and slow when it occurred transverse to fiber orientation (Panel D, upper dashed line). Anisotropy and Conduction Block During Initiation of Sustained Tachycardia During initiation of tachycardia by either programmed premature stimuli or burst pacing, relatively large differences in activation times between adjacent electrodes also resulted in lines formed by interpolated isochrones. We have concluded that these lines indicate conduction block rather than slow activation for reasons that are described below. We examined the location and characteristics of these lines of block and compared them with the location and characteristics of the lines of apparent block formed by slow activation during the sustained tachycardia. During initiation of tachycardia in all experiments orientation of a large part of the line of block was in a vertical direction (transverse to fiber orientation) indicating block of conduction of wavefronts moving 120 parallel to fiber orientation. This orientation occurred irrespective of the site at which the initiating stimuli were applied. Following the initiating impulse, this vertical line disappeared and the (horizontal) lines of apparent block that were characteristic of the sustained tachycardia appeared in another location. Figures 12 and 13 compare activation maps during initiation of tachycardia by single premature stimuli applied through electrodes at the margin (Figure 12) and at the basal margin (Figure 13). The activation pattern during the sustained tachycardia was previously described in Figure 4B and was not dependent on the site of stimulation. The ECG during initiation from the margin is shown in Figure IB. Figure 12A shows the activation pattern during basic drive at the margin at a cycle length of 300 msec. A broad wavefront swept across the left side of the epicardial border zone (from the 10- to the 50-msec isochrones). The pattern then became more inhomogeneous because of a small region of conduction block and wavefront collision shown by the thick black line that appeared

18 Dillon et al Anisotropic Reentry FIGURE 13. Activation maps of the epicardial border zone showing initiation of tachycardia by a single premature stimulus applied near the basal margin. The format of each of the activation maps is the same as the one described in Figure 4. Panel A shows activation during the last driven impulse initiated near S, prior to the premature stimulus. The dashed lines in this and subsequent panels indicate the location of the lines of apparent block during the sustained tachycardia (activation map shown in Figure I2D). Panel B shows activation map of the stimulated premature impulse, from its initiation to the 180 msec isochrone. Panel C shows continuation of propagation of this premature impulse. Some of the electrograms recorded in the vicinity of the line of block of the initiating premature impulse are shown in Panel D. The label at the left of each trace indicates the site of each recording as designated on the map in Panel B. Electrogram traces A, C, and E were recorded to the left of the line of block (see Panel B). The premature excitation wave reached these sites before dying out. The electrograms of the premature impulse (black arrows) are mostly upright. Electrogram traces B, D, and F were recorded to the right of the line of block. These sites were not excited until the reentering impulse returned to these regions (unfilled arrows). during drive. (The dashed lines in Panel A and subsequent panels show the location of the lines formed by interpolated isochrones during the sustained tachycardia.) Activation occurred around this line of block from the base and apical margins. Panel B shows the activation pattern of the stimulated premature impulse (coupling interval of 200 msec) that initiated tachycardia. This wavefront blocked along the margin at the msec isochrones (denoted by the thick black line and unfilled arrows). Conduction block of the premature impulse is indicated by the markedly different activation times at sites to the left and right of the region designated as block (a cause of the interpolated isochrones and the thick black line) and by the observation that activation waves on either side of the line are moving toward each other (note that during sustained tachycardia, although activation times across lines of apparent block related to anisotropy may also be quite different, wavefronts do not move toward each other). In addition, electrograms recorded at lines of block during initiation were quite different from the electrograms recorded at the lines of interpolated isochrones caused by anisotropy during sustained tachycardia. The characteristics are described below. Much of the line of block caused by the premature impulse was in the base-to-apex direction and perpendicular to the fiber orientation, indicating block of wavefronts moving parallel to the fiber long axis. The line of block also hooked to the right above, oblique to fiber orientation. Activation during the premature impulse circulated around this block along the basal margin above (isochrones 10-60) to the LL margin at the right (isochrones 60-80) and then proceeded back to the margin (isochrones ). This sequence of activation is pointed out by the black arrows. Activation during the premature impulse

19 200 Circulation Research Vol 63, No 1, July 1988 also occurred around the line of block towards the apex (below, out of the recording field) and proceeded back toward the margin while merging with the activation wave from the base and LL margin. Some additional regions of very slow activation or block occurred in this return pathway as indicated by dark black lines. The location of the block caused by the premature impulse was not the same as the site of (anisotropic) slow activation (apparent block) during sustained tachycardia, which is also shown in Panel B by the dashed lines. The circulating premature impulse returned to the distal side of the line of block between 180 and 190 msec (Panel B). This period was sufficient for recovery of excitability on the proximal side of the line of block. Panel C, which continues from where the map in panel B ends, shows that activation then continued through the region where the line of block was originally located (isochrones 10-30). After activation passed through the region of the initial line of block it spread again toward the base and apex and then toward the LL margin, returning to the margin as a second reentrant impulse (Panel C). Much of the line of block that occurred during initiation was no longer evident and the lines of interpolated isochrones that occurred during sustained tachycardia (Panel D) formed during the next 10 tachycardia beats. What is also apparent during these initial reentrant excitations is that activation proceeded through the regions where the lines of apparent conduction block eventually formed during the sustained tachycardia (dashed lines in Panels B and C). Activation through these regions slowed progressively until the lines of apparent block were formed. Figure 13 shows initiation of tachycardia in the same experiment by a premature impulse stimulated at the basal margin. In Panel A, during basic drive (also from the base S) activation proceeded slowly in broad wavefronts transverse to fiber orientation from the 10- to the 70-msec isochrones before wavefronts also invaded the epicardial border zone from the and LL margins as well. The premature impulse in Panel B (coupling interval 190 msec), blocked along an extensive horseshoe-shaped line (thick black line), with long components of the line of block in the vertical direction toward both the and LL margins. Block was again distinguished by the markedly different activation times between adjacent electrodes leading to interpolation of isochrones, wavefronts moving toward each other on either side of the line formed by the bunched isochrones, and electrogram characteristics. The morphology of electrograms typically found at the lines of block during initiation are exemplified by the recordings shown in Panel D. The electrograms of the premature impulse for the most part had uniphasic waveforms, either devoid of fractionated components or with only a few small notches (black arrows). The waveforms did not extend for the entire duration until activity occurred distal to the line of block, as was described for electrograms at lines of apparent block during sustained tachycardia; there was a long period of quiescence (isoelectric segment) before excitation distal to the line of block occurred as a result of the returning wavefront (unfilled arrows). Wavefronts moved around these long lines of block at the apex (below) and then entered the epicardial border zone from this direction (Panel B, isochrones 170 and 180). Activation then proceeded from the apex toward the and basal margins (isochrones , Panel C), passing through regions where the line of block was located at the margin, resulting in the first reentrant circuit. (Activation also occurred through the regions where lines of apparent block were eventually located during sustained tachycardia, which are indicated by the dashed lines.) Figure 14 shows initiation of tachycardia by burst pacing from the LL margin in the experiment previously described in Figure 4A (electrocardiogram during initiation is shown in Figure 1A), and illustrates the relation of the line of block that occurred during initiation (thick black lines) with the lines of apparent block caused by anisotropic slow activation that occurred during sustained tachycardia (dashed lines). Activation of the epicardial border zone during the first two stimulated impulses occurred uniformly. Panel A shows the map of one of these stimulated beats; broad wavefronts moved from the LL to margin. The location where the lines of apparent block eventually formed during sustained tachycardia are indicated by the dashed lines in this and subsequent panels. Activation during subsequent stimulated beats slowed and a line of block then occurred with a vertical orientation by the 5th stimulated impulse. The activation map of this stimulated impulse is shown in Panel B; the vertical line of block is shown by the thick black line and unfilled arrows. Activation in this map occurred around the line of block (black arrows). The line of block then extended during each subsequent stimulated impulse until it was in the shape of a horseshoe with vertical and horizontal components by the 8th impulse (Panel C). Activation occurred around the line at both of its extremities toward the margin and moved back towards the lateral margin (isochrones ). The activation wave reached the distal side of the vertical component of the block line at msec. This wavefront, returning toward the LL margin collided with the subsequent stimulated impulse, and this pattern of activation continued for several more stimulated impulses. The activation map showing the collision is in Panel D. The shaded area is the returning impulse resulting from the previous stimulus. This wavefront collided with the antegradely conducting impulse evoked at site S. After the last stimulated impulse the returning

20 Dillon et al Anisotropic Reentry 201 B 30 LAO LL 10 D.Oll 104 / TL70 50«FIGURE 14. Activation maps of the epicardial border zone showing initiation of tachycardia by rapid pacing (electrocardiogram shown in Figure IA). The format of each of the activation maps is the same as the one described in Figure 4. Panel A shows activation of the second stimulated impulse (S), initiated at the lateral left ventricular margin (LL). The dashed lines in this and subsequent panels indicate the location of the lines of apparent block during the sustained tachycardia (activation map shown in Figure 4A). Panel B shows the activation map of the fifth stimulated impulse, Panel C of the eighth stimulated impulse, Panel D of the 10th stimulated impulse and Panel E of the reentrant impulse after the last stimulated beat.

21 202 Circulation Research Vol 63, No 1, July 1988 wavefront continued toward the LL margin (Panel E, isochrones 10-50) and the reentrant circuit was established. Discussion Ventricular Tachycardia During the Healing Phase of Myocardial Infarction El-Sherif et al 2-4 first showed evidence obtained from composite electrograms that tachycardias might be caused by reentry on the epicardial surface of infarcts caused by occlusion, where a thin "sheet" of muscle often survives (the epicardial border zone) Subsequent studies utilizing simultaneous multiple electrode recordings to construct activation maps have shown circuitous patterns of activation (Figures 4A-4C), consistent with the reentry hypothesis, 5-9 although in some instances the epicardial border zone may comprise only a part of the reentrant circuit. 9 Evidence supporting the relation between the reentrant patterns of activation in the epicardial border zone and the genesis of tachycardia is provided by the experiments of El- Sherif et al 25 and Gessman et al, 26 who abolished the arrhythmias by cooling the epicardial region and blocking conduction. Therefore, because the relation between reentry in the epicardial border zone and the origin of tachycardia had been established, we did not attempt to prove in our study that the reentrant patterns we mapped caused the tachycardias. In other canine infarction models with inducible tachycardias, such as after coronary occlusion and reperfusion 3-27 or occlusion of the epicardial branches of the circumflex along with the, reentrant circuits have also been described but their location may not always be in the epicardial border zone because of the differences in the anatomical features of the infarcts. 327 Importance of Resolution In the studies in which isochronal activation maps have been constructed during tachycardia in the canine infarction model that we used, electrogram recordings have often been obtained both from the epicardial region of the infarct as well as from other regions of the ventricles This approach provided the important data necessary to show that impulses originating in the epicardial region caused the tachycardia. However, the total number of recordings has been limited to between 48 and 232 in the different experiments, and distributing electrodes throughout the ventricles is done at a cost to the spatial resolution of the maps of the reentrant circuits (spatial resolution is a function of the number of electrodes per unit area). As a result, between 20 and 80 recording sites have been located in regions of about 4 cm diameter where reentrant circuits occur With this number of recordings, if the revolution time in a circuit is 200 msec and isochrones are drawn at 10-msec intervals, one to four activation times on average might determine the location of each isochrone and, thus, the pattern of activation. In actuality, because unequal conduction times in different regions causes uneven isochrone distribution, some isochrones usually have to be drawn without real activation times. Although the resultant maps still accurately describe general patterns of activation and are valid for showing that reentry is present, some of the details of the activation patterns may not be apparent. In our study, we did not map activation of the ventricle outside the epicardial region over the infarct since it was not our purpose to relate activation in the reentrant circuits to that of the rest of the ventricle. Instead, 192 recordings were obtained from the 5.5 x 5.5 cm area of the epicardial border zone (the size of the electrode array) to improve the spatial resolution of the maps, providing greater accuracy for following the pathway of activation. Although the data we obtained from this approach has led to new conclusions and hypotheses, still higher spatial resolution and different recording techniques would be useful for solving some of the unanswered questions that arise from our experiments. We used bipolar electrodes with 1 mm between poles and a 3.5-mm spacing between bipolar pairs. Bipolar electrodes enabled us to record well-defined local signals devoid of a large extrinsic deflection from a small density of tissue. The 3.5-mm spacing enabled us to cover most of the epicardial border zone with the array. However, it is obvious after the complete data analysis that very low conduction velocities occur in regions of nonuniform anisotropy that are smaller than the spacing of the bipolar pairs and sometimes even smaller than the distance between poles. Activation patterns on a microscale in these regions (lines of apparent conduction block) are undoubtedly very complex, as indicated by the polyphasic extracellular waveforms. 21 These complex patterns cannot be resolved by the recording techniques we employed but might require the use of very fine-tipped (50 ^m) unipolar electrodes. 21 Our data, therefore, show the limits of isochronal mapping since isochrones cannot depict the complex slow activation patterns that probably occur in regions of nonuniform anisotropy. Block and Apparent Block: Causes of Lines of Block During Sustained Reentry In general, repetitive circus movement occurs around a region of conduction block; this is crucial for reentry since it prevents the reentrant impulse from "short circuiting" the reentrant pathway and blocking in refractory tissue. The region of block around which excitation revolves may result either from an anatomical obstacle (for example, the tricuspid orifice in the experiments of Frame et al 31 ) or may be caused by functional properties of viable cardiac fibers as in the leading circle mechanism. 32 In previous studies on reentry in the epicardial border zone, it has been emphasized that reentry occurs around long lines of block, although the

22 Dillon et al Anisotropic Reentry 203 FIGURE 15. Nonuniform conduction velocities in different regions of the anisotropic reentrant circuits described previously in Figure 4A and 4C and in Figure 11. The direction of wavefront movement was approximated from the orientation of the isochrones and the conduction velocities calculated between contiguous bipolar pairs. The velocities are only an estimate since the exact direction of wavefront movement cannot be determined. The orientation of the myocardialfiberlong axis is indicated by the long arrow below each panel. Thefigureshows that conduction velocity in the circuit where activation occurs transverse to fiber orientation may be as low as 0.04 to 0.07 mlsec while in the parts of the circuit where activation occurs parallel to fiber orientation it may be as high as 0.5 to 1.0 mlsec. cause for these lines of block has not been characterized. 24 Conduction block has sometimes been interpreted to occur when there was greater than a 40-msec difference between activation at adjacent electrodes. However, only several electrodes were often located in regions where it was concluded that there was block so some of the fine details of activation were not apparent We also found long lines during sustained reentry, across which activation was delayed significantly. However, because of the increased number of electrodes in these regions enabling a detailed analysis of extracellular waveforms and a more accurate construction of isochrones (see previous section) it was suggested to us that block might not always be occurring; therefore, they might be called lines of apparent block. Many of the electrograms at the lines of apparent block had a long duration and fractionated activity during tachycardia, whereas electrograms recorded at a distance from the line were usually characterized by more discrete deflections. Fractionated electrograms in reentrant circuits in canine infarcts have also been noticed by other investigators Although there is still some uncertainty concerning the cause of such electrograms in infarcts, the study of Spach and Dolber 21 in human atrium clearly show that they can be a result of slow activation in a complex "zigzag" pattern transverse to fiber orientation in nonuniformly anisotropic myocardium. Our own studies correlating microelectrode recordings with bipolar electrogram recordings in isolated tissues from infarcts also suggest that longduration, fractionated electrograms result from slow activation in the vicinity of the recording electrodes and that activation persists for the duration of the electrogram. 22 Each component ("spike") of a fractionated electrogram may arise from action potentials in different muscle bundles that may be poorly coupled to each other because of the loss of side-toside connections. The structure of the epicardial border zone provides an anatomical basis for this interpretation. This region is composed of parallel myocardial fibers that course from the toward the lateral left ventricle, while curving in the direction of the apex This arrangement of muscle bundles influences conduction properties. In the present study, activation of the epicardial border zone in vivo was shown to be more rapid in the direction of the long axis of the myocardial fiber bundles than transverse to the long axis (see Figures 8 and 9).'' It was also apparent from analysis of our activation maps that anisotropy in the epicardial border zone was nonuniform. Although transverse propagation in general was much slower than longitudinal propagation, it was also much slower in some regions than in others. This nonuniformity has been shown in the atrium to result from loss of side-to-side couplings between muscle fiber bundles. 21 Such uncoupling may also have occurred in the epicardial border zone as a result of the uneven distribution of edema, which separates the

23 204 Circulation Research Vol 63, No I, July 1988 myocardial fiber bundles," or some other aspect of the healing process. It is therefore predicted that the anisotropy should slow conduction in the reentrant circuits since wavefronts in these circuits move both parallel and transverse to the long axis of the muscle fibers. The results of the stimulation experiments also show that slow activation in the regions where lines of apparent block were located must be a result of anisotropy and not abnormalities in the transmembrane potentials such as low membrane potentials or a prolonged refractory period 36 since such abnormalities would be expected to influence conduction irrespective of the direction of activation. When activation occurred parallel to fiber orientation in these regions, it was not slow. The isochrones in the vicinity of the regions of lines of apparent block and polyphasic waveforms also supported the interpretation that there was slow activation caused by anisotropy rather than block. When the lines of apparent block were oriented parallel to the long axis of the myocardial fibers (horizontally in our figures) isochrones distal to the lines were often parallel to them. This means that activation distal to the lines occurred nearly simultaneously along much of their length, something that would not be possible if activation occurred only around lines of real conduction block. The isochrones parallel to the lines of apparent block might not have been obvious if only a few electrodes had been located in these regions. Since there was no evidence of gross anatomical obstacles in most of the regions of horizontal lines of apparent block, as shown by the surviving thin layers of muscle and from activation maps during sinus rhythm and ventricular stimulation, activation across the lines during tachycardia should be possible. On the other hand, vertically oriented lines of block during tachycardia (transverse to fiber orientation) (Figure 4B) were associated at least partly with a gross anatomical defect. In these regions, the myocardial infarct extended to the epicardial surface and there were few surviving muscle fibers. The proposal that isochrones parallel to lines of apparent block result from activation across the lines is also based on an assumption that epicardial breakthrough is not responsible for the nearly simultaneous activation. Such breakthrough could not result from transmural impulse propagation since the reentrant circuits occurred in narrow sheets of muscle overlying necrotic myocardium. Comparison With Lines of Block During Initiation of Tachycardia The lines of apparent block during sustained tachycardia can be distinguished from the lines of real block that occur during initiation of tachycardia by premature impulses (Figures 12-14). These lines of block during initiation were also indicated by a marked disparity in activation times of adjacent electrodes. However, activation waves on each side of the lines of block were moving toward each other, unlike activation on opposite sides of the lines of apparent block during sustained tachycardia, and recordings at these lines were not characterized by long-duration polyphasic waveforms. The uniphasic deflections found in these regions during initiation are consistent with the expected characteristics of electrograms in regions of block. 23 The lines of block during initiation were, for the most part, oriented in the base-to-apex direction irrespective of the site of stimulation and were located toward the margin of the epicardial border zone with normal tissue. This orientation indicates block of activation of wavefronts moving parallel to the long axis of the myocardial fibers. Gough et al 37 have shown that block of premature impulses in the epicardial border zone occur where there is a marked disparity in refractory periods. However, Spach et al 13 have proposed that the safety factor for conduction of premature impulses is lower for wavefronts propagating parallel to fiber orientation than transverse to it. 13 If this applies to the epicardial border zone, anisotropy may also contribute to the block of premature impulses. After tachycardia was initiated, these lines of block disappeared as the circuit stabilized. A Model of Reentry in an Anisotropic Matrix: Anisotropic Reentry Spach et al 131 * 21 originally showed that nonuniform anisotropy can cause single reentrant impulses in atrial myocardium. Based on our interpretation of the data we have presented, we propose that nonuniform anisotropy caused by side-to-side uncoupling is also an important factor that causes the slow conduction necessary for sustained reentry in the epicardial border zone of infarcts. If the lines of apparent block, in fact, are formed by slow activation transverse to the myocardial fiber orientation, then reentry is most likely occurring around a small central fulcrum rather than around long lines of block as we have diagrammed in Figure 11. Reentry on this small size scale has also been shown in reperfused infarcts by Richards et al. 30 The size of the central fulcrum may vary in different circuits; we cannot precisely define its extent without higher resolution maps. We also have no data relating to the events occurring at this fulcrum. It may be a region of collision and block of centripetal impulses from the circulating wavefront as in the leading circle model of reentry 32 or it may be a small area damaged by the infarction. Sometimes part of the central region may be formed by a larger anatomical obstacle as shown by the reentry in the upper loop in Figure 4B. Because of anisotropy, activation around the circuit is inhomogeneous. This characteristic of anisotropic reentry is shown in Figure 15, where we have calculated apparent conduction velocities in the reentrant circuits previously described in Figure 4A and 4C. Conduction velocities were nearly normal 38 and range from about 0.3 to 1.0 m/sec in the parts of the circuits where activation

24 was parallel to fiber orientation. Conduction velocities were very slow and ranged from about 0.2 to 0.04 m/sec in parts of the circuit where activation was transverse to fiber orientation. The slow transverse activation in anisotropic reentry provides the necessary time for recovery of excitability that enables reentry to continue. We do not mean to imply from our interpretation of the data that abnormal transmembrane potentials of muscle fibers in the epicardial border zone do not play any role in causing reentry. The possible role of inhomogeneous refractory periods in the initiation of reentry has already been discussed. In addition, effects of a reduced membrane potential and action potential upstroke velocity may sometimes be superimposed on the effects of the anisotropic tissue structure. 36 The relative role of both abnormal transmembrane potentials and anisotropic tissue structure requires further investigation. References 1. Kaplinsky E, Yahini JH, Ncufeld HN: On the mechanism of sustained ventricular arrhythmias associated with acute myocardial infarction. Cardiovasc Res 1972;6: El-Sherif N, Hope RR, Scherlag BJ, Lazarra R: Reentrant ventricular arrhythmias in the late myocardial infarction period. I. Conduction characteristics in the infarction zone. Circulation 1977;55: Karagueuzian HS, Fenoglio JJ Jr, Weiss MB, Wit AL: Protracted ventricular tachycardia induced by premature stimulation of the canine heart after coronary occlusion and reperfusion. Circ Res 1979;44: El-Sherif N, Hope RR, Scherlag BJ, Lazarra R: Reentrant ventricular arrhythmias in the late myocardial infarction period. 2. Patterns of initiation and termination of reentry. Circulation 1977^5: Wit AL, Allessie MA, Bonke FIM, Lammers W, Smeets J, Fenoglio JJ Jr: Electrophysiological mapping to determine the mechanism of experimental ventricular tachycardia initiated by premature impulses. Experimental approach and initial results demonstrating reentrant excitation. Am J Cardiol 1982;49: El-Sherif N, Smith A, Evans K: Canine ventricular arrhythmias in the late myocardial infarction period. 8. Epicardial mapping of reentrant circuits. Circ Res 198l;49: Mehra R, Zeiler RH, Gough WB, El-Sherif N: Reentrant ventricular arrhythmias in the late myocardial infarction period. 9. Electrophysiological-anatomic correlation of reentrant circuits. Circulation I983;67: Cardinal R, Savard P, Carson DL, Pag6 P: Mapping of ventricular tachycardia induced by programmed stimulation in canine preparations of myocardial infarction. Circulation 1984;70: Kramer JB, Saffitz JE, Witkowski FX, Corr PB: Intramural reentry as a mechanism of ventricular tachycardia during evolving canine myocardial infarction. Circ Res 1985;- 56: Wit AL, Allessie MA, Fenoglio JJ Jr, Bonke FIM, Lammers WJEP, Smeets J: Significance of the cndocardial and epicardial border zones in the genesis of myocardial infarction arrhythmias, in Harrison D (ed): Cardiac Arrhythmias: A Decade of Progress. Boston, GK Hall, 1982, pp Ursell PC, Gardner PI, Albala A, Fenoglio JJ Jr, Wit AL: Structural and electrophysiological changes in the epicardial border zone of canine myocardial infarcts during infarct healing. Circ Res 1985;56: Dillon et al Anisotropic Reentry Clerc L: Directional differences of impulse spread in trabecular muscle from mammalian heart. J Physiol (Lond) 1976; 255: Spach MS, Miller WT, Geselowitz D, Barr RC, Kootsey JM, Johnson EA: The discontinuous nature of propagation in normal canine cardiac muscle. Evidence for recurrent discontinuities of intracellular resistance that affect the membrane currents. Circ Res 1981;48: Spach MS, Miller WT, Dolber PC, Kootsey JM, Sommer JR, Mosher CE: The functional role of structural complexities in the propagation of depolarization in the atrium of the dog: Cardiac conduction disturbances due to discontinuities of effective axial resistivity. Circ Res 1982^0: Dillon S, Ursell PC, Wit AL: Pseudo-block caused by anisotropic conduction: A new mechanism for sustained reentry (abstract). Circulation 1985;72(suppl III):III Harris AS: Delayed development of ventricular ectopic rhythm following experimental coronary occlusion. Circulation 1950;l: Garan H, Fallon JT, Rosenthal S, Ruskin JN: Endocardial, intramural and epicardial activation patterns during sustained monomorphic ventricular tachycardia in late canine myocardial infarction. Circ Res 1987;60: El-Sherif N, Gough WB, Zeiler RH, Hariman R: Reentrant ventricular arrhythmias in the late myocardial infarction period. 12. Spontaneous versus induced reentry and intramural versus epicardial circuits. J Am Coll Cardiol 1985; 6: El-Sherif N, Mehra R, Gough WB, Zeiler RH: Reentrant ventricular arrhythmias in the late myocardial infarction period. II. Burst pacing versus multiple premature stimulation. J Am Coll Cardiol 1984 ;4: Itatsu H: Theoretical interpretation of contiguous bipolar ECG and its relationship to the time of arrival of activation. Jpn Circ J 1954;18:1-IO 21. Spach MS, Dolber PC: Relating extracellular potentials and their derivatives to anisotropic propagation at a microscopic level in human cardiac muscle. Evidence for uncoupling of side-to-side fiber connections with increasing age. Circ Res 1986^8: Gardner PI, Ursell PC, Fenoglio JJ Jr, Wit AL: Electrophysiologic and anatomic basis for fractionated electrograms recorded from healed myocardial infarcts. Circulation 1985; 72: Spach MS, Barr RC, Serwer GS, Johnson EA, Kootsey JM: Collision of excitation waves in the dog Purkinje system. Extracellular identification. Circ Res 1971 ;24: El-Sherif N: The figure 8 model of reentrant excitation in the canine post infarction heart, in Zipes DP, Jalife J (eds): Cardiac Electrophysiology and Arrhythmias. 1985, pp El-Sherif N, Mehra R, Gough WB, Zeiler RH: Reentrant ventricular arrhythmias in the late myocardial infarction period. Interruption of reentrant circuits by cryothermal techniques. Circulation 1983;68: Gessman LJ, Agarwal JB, ErdoT, Helfant RH: Localization and mechanism of ventricular tachycardia by ice mapping one week after onset of myocardial infarction. Circulation 1983,68: Michelson EL, Spear JR, Moore EN: Electrophysiologic and anatomic correlates of sustained ventricular tachyarrhythmias in a model of chronic myocardial infarction. Am J Cardiol 1980;45: Garan H, Fallon JT, Ruskin JN: Sustained ventricular tachycardia in recent canine myocardial infarction. Circulation 1980;62: Garan H, Ruskin JN: Localized reentry: Mechanism of induced sustained ventricular tachycardia in canine model of recent myocardial infarction. J Clin Invest 1984,74: Richards DA, Blake GJ, Spear JF, Moore EN: Electrophysiologic substrate for ventricular tachycardia: Correlation of properties in vivo and in vitro. Circulation 1984;69:

25 206 Circulation Research Vol 63, No 1, July Frame LH, Page RL, Hoffman BF: Atrial reentry around an anatomic barrier with a partially refractory excitable gap. A canine model of flutter. Circ Res 1986^8: Allessie MA, Bonke FIM, Schopman FJG: Circus movement in rabbit atrial muscle as a mechanism of tachycardia. III. The "leading circle" concept: A new model of circus movement in cardiac tissue without the involvement of an anatomical obstacle. Circ Res I977;41: Boineau JP, Cox JL: Slow ventricular activation in acute myocardial infarction: A source of reentrant premature ventricular contraction. Circulation 1973;48:7O2-7I3 34. Waldo AL, Kaiser GA: A study of ventricular arrhythmias associated with acute myocardial infarction in the canine heart. Circulation 1973,47: Streeter DD: Gross morphology and fiber geometry of the heart, in Berne RM, Sperelakis N, Greger SR (eds): Handbook of Physiology. The Cardiovascular System. Vol. I. The Heart. Baltimore, Maryland, Williams & Wilkins Co, 1979, pp Lazarra R, Scherlag BJ: Role of the slow current in the generation of arrhythmias in ischemic myocardium, in Zipes DP, Bailey JC, Elharrar V (eds): The Slow Inward Current and Cardiac Arrhythmias. The Hague, The Netherlands, Martinus Nijhoff Publishing, 1980, pp Gough WB, Mehra R, Restivo M, Zeiler RH, El-Sherif N: Reentrant ventricular arrhythmias in the late myocardial infarction period in the dog. 13. Correlation of activation and refractory maps. Circ Res 1985^7: Roberts DE, Hersh LT, Scher AM: Influence of cardiac fiber orientation on wavefront voltage, conduction velocity and tissue resistivity in the dog. Circ Res 1979;44: KEY WORDS myocardial infarction ventricular tachycardia anisotropy reentry uncoupling

26 Influences of anisotropic tissue structure on reentrant circuits in the epicardial border zone of subacute canine infarcts. S M Dillon, M A Allessie, P C Ursell and A L Wit Circ Res. 1988;63: doi: /01.RES Circulation Research is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX Copyright 1988 American Heart Association, Inc. All rights reserved. Print ISSN: Online ISSN: The online version of this article, along with updated information and services, is located on the World Wide Web at: Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published in Circulation Research can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office. Once the online version of the published article for which permission is being requested is located, click Request Permissions in the middle column of the Web page under Services. Further information about this process is available in the Permissions and Rights Question and Answer document. Reprints: Information about reprints can be found online at: Subscriptions: Information about subscribing to Circulation Research is online at: