Three Distinct Directions of Intramural Activation Reveal Nonuniform Side-to-Side Electrical Coupling of Ventricular Myocytes Bryan J. Caldwell, PhD; Mark L. Trew, PhD; Gregory B. Sands, PhD; Darren A. Hooks, MBChB, PhD; Ian J. LeGrice, MBChB, PhD; Bruce H. Smaill, PhD Background—The anisotropy of cardiac tissue is a key determinant of 3D electric propagation and the stability of activation wave fronts in the heart. The electric properties of ventricular myocardium are widely assumed to be axially anisotropic, with activation propagating most rapidly in the myofiber direction and at uniform velocity transverse to this. We present new experimental evidence that contradicts this view. Methods and Results—For the first time, high-density intramural electric mapping (325 electrodes at ⬇4⫻4⫻1-mm spacing) from pig left ventricular tissue was used to reconstruct 3D paced activation surfaces projected directly onto 3D tissue structure imaged throughout the same left ventricular volume. These data from 5 hearts demonstrate that ventricular tissue is electrically orthotropic with 3 distinct propagation directions that coincide with local microstructural axes defined by the laminar arrangement of ventricular myocytes. The maximum conduction velocity of 0.67⫾0.019 ms⫺1 was aligned with the myofiber axis. However, transverse to this, the maximum conduction velocity was 0.30⫾0.010 ms⫺1, parallel to the myocyte layers and 0.17⫾0.004 ms⫺1 normal to them. These orthotropic conduction velocities give rise to preferential activation pathways across the left ventricular free wall that are not captured by structurally detailed computer models, which incorporate axially anisotropic electric properties. Conclusions—Our findings suggest that current views on uniform side-to-side electric coupling in the heart need to be revised. In particular, nonuniform laminar myocardial architecture and associated electric orthotropy should be included in future models of initiation and maintenance of ventricular arrhythmia. (Circ Arrhythmia Electrophysiol. 2009;2:433-440.) Key Words: anisotropy 䡲 mapping 䡲 structure 䡲 computer modeling 䡲 intramural pacing A ccurate information about the electric properties of cardiac tissue is central to understanding the biophysical basis of normal and aberrant heart rhythm. Electric anisotropy influences the spread of activation in the heart, plays a critical role both in the initiation and maintenance of reentrant arrhythmia, and is an important determinant of the effectiveness of cardioversion. Knowledge of the nature and extent of electric anisotropy is required for computer models of heart activation that provide a means of probing intramural electric behavior that cannot be accessed from clinical and experimental measurements made on the surfaces of the heart. Clinical Perspective on p 440 Normal ventricular myocardium is generally thought to function as a syncytium in which side-to-side electric coupling between adjacent myocytes is uniform.1,2 The electric properties of ventricular myocardium are assumed to be axially anisotropic with respect to the local myofiber axis,1,2 with activation propagating most rapidly in the myofiber direction and at uniform velocity in planes transverse to this. This view is not consistent with the laminar model of ventricular myocardium that has resulted from detailed morphometric investigations of 3D cardiac tissue architecture.3–5 Ventricular myocardium is described as having a laminar organization in which myocytes are arranged in layers (myolaminae) approximately 4 cells thick. Adjacent layers branch and interconnect but are separated by cleavage planes across which there can be little direct cell-to-cell coupling. The laminar model of myocardial architecture is supported by the findings of numerous analyses of cardiac mechanical function6 – 8 and by the results of cardiac magnetic resonance diffusion–tensor imaging studies.9,10 A structurally detailed computer model predicted that the laminar architecture of ventricular myocardium will give rise to orthotropic electric Received October 19, 2008; accepted June 8, 2009. From the Auckland Bioengineering Institute (B.J.C., M.L.T., G.B.S., D.A.H., I.J.L., B.H.S.) and the Department of Physiology (I.J.L., B.H.S.), Faculty of Medical and Health Sciences, University of Auckland, Auckland, New Zealand. Correspondence to Bruce H. Smaill, PhD, Auckland Bioengineering Institute, University of Auckland, Private Bag 92019, Auckland 1001, New Zealand. E-mail b.smaill@auckland.ac.nz © 2009 American Heart Association, Inc. Circ Arrhythmia Electrophysiol is available at http://circep.ahajournals.org DOI: 10.1161/CIRCEP.108.830133 433 Downloaded from http://circep.ahajournals.org/ by guest on June 1, 2015 434 Circ Arrhythmia Electrophysiol August 2009 Figure 1. Intramural activation mapping in pig heart. A, Extracellular potentials were recorded from the LV anterior region indicated using a 5⫻5 array of plunge needles, each ⬇4 mm apart at the epicardium and containing 13 electrodes at 1-mm transmural spacing. Fiducial markers outside corners of array are also shown. B, Activation isosurfaces from midwall unipolar stimulus reconstructed at 6-ms intervals after first activation. Electrode locations and PAA are indicated. Activation isochrones (2 ms) are superimposed on tissue structure rendered in a plane (C) parallel to the epicardial surface containing the stimulus (indicated) and (D) transverse to the PAA and also containing the stimulus. Data are from a typical experiment (experiment 1) stimulated 6.5 mm below epicardium. LAD indicates left anterior descending. properties with slow propagation normal to laminae due to the tortuous conduction paths involved.11 Experimental data that would enable this hypothesis to be tested objectively have not previously been available. The few studies in which attempts have been made to reconstruct 3D intramural electric activation12,13 have not had the spatial resolution nor provided sufficient structural data to address this issue. This work, for the first time, provides experimental evidence that the spread of electric activation in ventricular myocardium is not axially anisotropic. We demonstrate preferential transmural activation directions that coincide with the orientation of muscle layers across the left ventricular (LV) wall. Conduction is significantly slower perpendicular to muscle layers, suggesting that structural discontinuities affect side-to-side electric coupling. Methods The authors had full access to and take full responsibility for the integrity of the data. All authors have read and agree to the manuscript as written. All surgical procedures were approved by the Animal Ethics Committee of The University of Auckland and conform to the Guide for the Care and Use of Laboratory Animals (NIH publication No. 85-23). Key features of the study design are represented in Figure 1. Five pigs (weight, 45 to 55 kg) were anesthetized and maintained with halothane (2% to 5%) in oxygen. The heart was exposed, and an array of 25 plunge needles (0.5-mm OD with 13 electrodes at 1-mm spacing; see Rogers14 for details of fabrication) was introduced into the LV free wall. Unipolar extracellular potentials (ECPs), referred to the pulmonary artery root, were acquired (sampling rate, 4 KHz; 12-bit resolution; bandwidth, 0.01 to 2000 Hz) at up to 325 sites using a purpose-developed mapping system (UnEmap, Auckland UniServices Ltd, New Zealand). Recordings were made when ST-segment elevation was fully resolved, 40 to 60 minutes after insertion of all needles. ECPs were recorded during intramural unipolar pacing (typically 80 to 100 bpm for 15 to 30 seconds) from each of the 13 electrodes along the central needle and subsequently in sinus rhythm. This procedure was repeated with bipolar stimulation using adjacent electrodes along the same needle. Constant-current stimulation was used at 1.5⫻ capture threshold. On completion of the experiment, the heart was arrested, excised with needles in place, and immersed in chilled (4°C) physiological saline solution. Hearts were perfusion-fixed (3% formalin in phosphate buffer), and a transmural segment containing the recording array was removed. T2-weighted MR images of the excised LV segment were acquired at 0.313-mm voxel size (1.5-T Siemens Avantos, MAGNETOM Avanto, Siemans AG Medical Solutions) shortly after fixation. Plunge needles were carefully removed from the heart before MR imaging. Previously measured electrode positions were mapped onto best-fit lines superimposed on the segmented needle tracks to determine 3D electrode locations in vivo. Activation times were estimated from dV/dTmin for the intrinsic ECP deflection. Where dV/dT exhibited a prolonged plateau or showed fractionation (17% of all pacing data), the principal deflection was identified by convolving the ECP with adjacent signals from the same plunge needle using the high pass filter weights (⫺0.5 1.0 to 0.5). Activation contours were reconstructed in 3D using Hardy interpolation,15 with R2⫽0.05. Activation data were interpolated onto a 0.5-mm 3D grid for analysis of activation contours in 2D slices and evolution of the wave front in 3D. The direction of most rapid propagation, defined as the principal axis of activation (PAA), was identified as the longest line through the stimulus site enclosed by the 12-ms isosurface. The spread of activation transverse to the PAA was characterized as follows. In the transverse plane containing the origin of activation, the distance from origin to points on a specified isochronal contour was measured and magnitudes and orientations of local maxima and minima were recorded. Local 3D conduction velocities (CVs) were estimated as follows. Isochronal activation surfaces at 1-ms intervals were interpolated at 0.2-mm grid resolution and vectors normal to these surfaces were determined. The orientation and magnitude of local velocity were given by the direction of the normal vector and the distance along it to the next isochronal surface. Maximum and minimum CVs were extracted from volumes immediately adjacent to the PAA and the transverse plane containing the origin of activation. Activation isosurfaces were constructed and analyzed using MATLAB (MathWorks Inc, Cambridge, Mass). Tissue architecture was reconstructed in 3D throughout the volume from which intramural potentials were acquired as follows (Figure 2). The excised LV segment was embedded in wax and the entire volume was imaged using techniques described in detail elsewhere. 16 In brief, specimens were mounted with the circumferential-radial surface uppermost. This surface was trimmed with an ultramiller to expose the tissue, stained with Toluidine blue (0.12% in 1% borax) to a depth of ⬇2 ␮m, and imaged at 8.33-␮m pixel size using a digital camera. The process of milling, surface staining, and imaging was repeated throughout the volume. The full volume was imaged at 50-␮m steps, whereas 16.7-␮m steps were used for a 5-mm-wide central subvolume. MR volume images of the excised LV segment were resliced parallel to the epicardial surface (XY plane). Plunge needle tracks and corner fiducial markers were segmented and the coordinates of epicardial entry and endocardial exit points were determined. Tissue shrinkage and distortion caused by wax embedding were corrected using 3D cubic mappings, constrained so that needle entry and exit points in tissue volumes and corresponding MR images were best matched. RMS distance errors for the corrections varied from 0.62 to 1.75 mm, with a measurement error of 0.62 mm. To determine myofiber orientation, corrected volume images were resliced on XY planes parallel to the epicardial surface, subdivided with a 24⫻24 pixel grid, and dominant angles in each subregion Downloaded from http://circep.ahajournals.org/ by guest on June 1, 2015 Caldwell et al Orthotropic Intramural Spread of LV Activation 435 Figure 2. Reconstruction of 3D morphology in tissue volume containing recording array. The volume was imaged initially with MRI and subsequently using extended volume microscopy. A, 3D reconstruction of extended tissue volume (8.33-␮m voxels). Processing artifact is corrected by transforming the volume to best match key features with corresponding structures in equivalent MR volume. Sections are resliced parallel to the epicardial tangent plane (XY) and transverse to the mean myofiber direction in that plane. B, XY slice with myofiber orientations overlaid. Locations of needle tracks and corner fiducial markers are highlighted. C, Transverse section with muscle layer orientations overlaid. D, Model of 3D geometry and fiber field in the tissue volume used for computation. Epicardial and endocardial surfaces are rendered and local myofiber orientation are represented by ribbons. were extracted (Figure 2B) using a Sobel gradient operator.17 Myolaminar orientations were measured similarly in resliced transmural planes (Figure 2C). A computer activation model was constructed using surface geometry, myofiber orientations, and electrode locations measured in 1 heart (Figure 2D). Electric properties were assumed to be axially anisotropic, with conductivities of 0.4 and 0.1 mS/mm, respectively, in the fiber direction and transverse to the fiber axis. A monodomain activation model with a cubic ionic current was discretized using finite elements at 0.16-mm resolution and solved on the domain containing the recording array (25⫻31⫻18 mm), using a highperformance computer (IBM p595). Data are expressed as mean⫾SEM. Comparison of grouped data were made using multiway, repeated-measures analysis of variance. Results The ECPs acquired in this study were stable and noise-free over the length of the experiment (typically ⬍120 minutes), with no differences distinguishable between consecutive trains of unipolar and bipolar stimulation. ECPs in sinus rhythm were highly reproducible throughout each experiment and exhibited a narrow activation complex. During pacing, activation complexes were broadened and less smooth. STsegment elevation was negligible in all cases. Typical activation isosurfaces generated by intramural point stimulation are shown in Figure 1. The spread of activation was initially ellipsoid with the PAA—the direction of most rapid early propagation—typically lying in a plane adjacent to the stimulus site and parallel to the epicardial surface. Activation isochrones are superimposed on tissue structure rendered in this plane (Figure 1C), and it is evident that the PAA aligns with local myofiber direction. However, the spread of activation transverse to the PAA (Figure 1D) is also markedly nonuniform with most rapid propagation at around ⫺20° to the transmural z-axis. This coincides with the orientation of myolaminae in these transmural planes. Figure 3 shows activation contours generated by stimulation at other intramural sites in the same heart. Panels B and C are XY and transverse planes that contain the stimulus site as outlined above. Again, the spread of activation is elliptical in both planes. The orientation of the PAA varies through ⬇130° in the XY plane as the stimulus is moved progressively across the LV wall (Figure 3B). Transverse to the PAA, activation consistently spreads most rapidly oblique to the transmural z-axis. The correspondence between transverse activation spread in the vicinity of the stimulus and local laminar architecture is shown in Figure 3C. Although the 3D spread of activation was relatively uniform, local inhomogeneity was common. Activation was nonsequential on occasion, and the point of first activation was displaced 1.1⫾1.3 mm from the stimulus site. The correspondence between the 3D activation spread and local laminar structure was quantified across all hearts and is represented in Figure 4. The orientation of the PAA is compared with myofiber direction at stimulus sites through the LV wall, for a typical experiment. The correspondence between PAA and myofiber direction is remarkable and this was reproduced across all studies: average difference between pairs 8⫾7°. In Figure 4B, we present distributions of myolaminar orientation adjacent to the stimulus site in the plane transverse to the PAA. Despite local variability, myolamina orientation distributions were surprisingly reproducible both within and across experiments. In all cases, the orientation of the dominant population of myolaminae was tightly distributed around a mean angle of ⫺25° relative to transmural z-axis. A more scattered secondary population with a mean orientation ⬇15° was also observed. To quantify the spread of activation in this plane, maximum and minimum distances from the point of first activation to the 10-ms activation contour are plotted in polar coordinates across all experiments (Figure 4C). The transverse spread of activation is greatest in the upper right and lower left quadrants of the plane (3.8⫾1.3 mm and ⫺42⫾36°) and least in the quadrants at right angles (2.1⫾0.5 mm and 51⫾39°). Features to be noted here are the significant differences between maxima and minima and the fact that they cluster in orthogonal orientations. The directions in which activation spreads most rapidly in the transverse plane coincide with the dominant layer orientation. Local CV was consistently greatest in the fiber direction. In the plane transverse to the PAA, maximum and minimum local velocities clustered as in Figure 4C. For a given stimulus site, maximum velocity was relatively consistent during the interval 5 to 12 ms after first activation. There were too few points for accurate local velocity estimation Downloaded from http://circep.ahajournals.org/ by guest on June 1, 2015 436 Circ Arrhythmia Electrophysiol August 2009 location on mean conduction velocity in the 3 orthogonal directions. To test further the hypotheses that the electric properties of ventricular myocardium are transversely isotropic and that the intramural spread of activation is determined by principally by myofiber rotation, we implemented an experimentspecific activation model. This incorporates detailed measurements of 3D surface geometry and myofiber orientations throughout the region containing the recording array, and electric properties are assumed to be transversely isotropic with respect to the fiber axis. Predicted and experimental 3D spread of activation are compared directly in Figure 5 for midwall and subendocardial unipolar stimulation. It is evident that the model does not capture key features of corresponding experimental data. The predicted spread of activation in the transverse plane containing the stimulus is relatively circular in the model but is elliptical in the experiment. Although the direction of principal activation spread in transverse planes on either side of the stimulus (A and C) is in opposite directions in the model, this is not seen experimentally. The difference between predicted and observed results is marked for stimuli in subendocardial regions where variation of fiber orientation is relatively small (see Figure 5). Discussion Figure 3. Three-dimensional spread of activation from intramural stimulus at a series of sites from subepicardium to subendocardium in experiment 1. A, Broken lines indicate PAA and red arrows the viewing direction of transverse plane (shaded). B, Planes containing the stimulus: transverse to PAA and parallel to the epicardial surface (1-ms isochrones). Red lines indicate orientation of the transverse plane above. C, Isochrones (2 ms) superimposed on 8⫻8 mm region of rendered laminar structure. before this and maximum velocity in this plane increased subsequently, presumably as a result of fiber rotation. Principal local CVs in the fiber direction and transverse to it were averaged across this time interval and are presented across all experiments in Figure 4D; mean values are 0.67⫾0.019, 0.30⫾0.010, and 0.17⫾0.004 ms⫺1. Each of these is significantly different (P⬍0.009, n⫽195), and they separate in the ratio 4:1.8:1. There was no significant effect of intramural In this study, high-density 3D electric mapping has been used for the first time to reconstruct activation sequences and project these directly onto tissue structure imaged at high resolution throughout the same LV volumes. The following important findings flow from this approach. Electric activation spreads in 3 distinct directions from an intramural LV point stimulus, and these directions correspond with local microstructural axes. Active electric properties are therefore anisotropic transverse to the myofiber direction, and this gives rise to preferential transmural activation pathways. That electric activation propagates most rapidly in the fiber direction is wholly expected and has been shown in numerous studies.2,12,18 Our estimates of both axial and maximum transverse CV are very similar to results reported for ventricular epicardium.2,12 However, anisotropic spread of activation transverse to the fiber axis has not previously been demonstrated. This is because the few studies in which the spread of electric activation has been reconstructed in 3D12,13,19 had insufficient spatial resolution to detect such directional variation. We inserted 325 electrodes at 1-mm transmural spacing into a segment approximately 16⫻16⫻12 mm of the anterior free wall in the pig heart. This enabled us to map 3D electric activation with sufficient spatial resolution to reconstruct accurate isochronal contours transverse to the PAA in planes containing the site of first excitation. Transverse activation follows a consistent transmural pattern across all experiments, with maximum and minimum transverse local velocities separating in a ratio ⬇2:1. This is not consistent with the widely held view that the electric properties of ventricular myocardium are isotropic transverse to the fiber axis. The strong association between the direction of preferential transverse activation and the orientation of laminae in the same transverse plane does not prove that there is a mecha- Downloaded from http://circep.ahajournals.org/ by guest on June 1, 2015 Caldwell et al Orthotropic Intramural Spread of LV Activation 437 Figure 4. Comparison of myocardial structure with spread of activation. A, Myofiber and PAA orientations with respect to circumferential y-axis as a function of transmural depth in experiment 1. B, Distributions of transverse myolaminar orientations (binned at 5°) for each experiment. C, Maximum (circles) and minimum (squares) transverse spread 10 ms after first activation across all experiments. Polar plots for C and D oriented with respect to the transmural z-axis. D, Principal local conduction velocities 5 to 12 ms after first activation. Mean CV in fiber direction is 0.67⫾0.019 ms⫺1, whereas maximum and minimum transverse CVs are 0.30⫾0.010 ms⫺1 and 0.17⫾0.004 ms⫺1, respectively. nistic link between the two. However, it provides additional support for the view that the laminar architecture of LV myocardium gives rise to orthotropic electric properties. In our initial article reporting laminar arrangement of myocytes in ventricular myocardium,3 we argued that, in the absence of electric coupling across the clefts that separate adjacent layers, the spread of activation perpendicular to laminae would be forced to a follow a relatively tortuous path and therefore occur more slowly than the transverse spread of activation within layers. Hooks et al11 tested this hypothesis in a computer model that incorporated detailed information about laminar structure through the rat LV free wall.5 The model predicted a transverse velocity ratio of 1.68 in the directions parallel with and normal to laminae, respectively. Most recently, Hooks et al20 used a combination of highresolution intramural electric mapping, detailed structural Figure 5. Predictions of structurally detailed, experiment-specific activation model, incorporating axially anisotropic electric properties, compared with observed data (experiment 1). The stimulus depth (indicated) is 6.5 mm in the upper panel and 10.5 mm (in anterior papillary muscle) in the lower panel. The predicted 3D spread of electric activation is represented by isosurfaces at 6-ms intervals after first activation. Planes transverse to the PAA (A, behind, B, containing, and C, in front of the stimulus site) are superimposed on the 3D maps (viewing direction indicated by red arrows). Predicted and observed activation contours in these planes, and the differences between them are presented. Downloaded from http://circep.ahajournals.org/ by guest on June 1, 2015 438 Circ Arrhythmia Electrophysiol August 2009 measurement, and experiment-specific modeling to analyze the passive electric response of LV myocardium to injection of a current pulse. It was found that bulk conductivity transverse to the fiber axis (affected by extracellular and intracellular conductivities) divided in a ratio of approximately 2:1, with maximum bulk conductivity in the direction of myolaminae and minimum conductivity perpendicular to this. Although most laminae are oriented between ⫺20 to ⫺30° to the transmural z-axis, there are regions with interspersed myolaminae approximately at right angles to this (see Figure 4B), as has been reported elsewhere.3,9,10,20 Although the direction of the principal laminar orientation is surprisingly consistent across all studies, the relative frequency of the second orthogonal direction varies between and within hearts and appears to be greatest in the midwall. This would be expected to introduce a stochastic component into the spread of activation. An important further question is whether it is necessary to incorporate orthotropic electric properties in computer activation models to capture key features of intramural propagation. The results of computational studies that have sought to address this issue21,22 have been equivocal. Colli-Franzone et al22 have argued that the effects of intramural fiber rotation would mask the orthotropic nature of the medium, particularly for epicardial activation maps. They state that, from a qualitative point of view, both axially anisotropic and orthotropic models are compatible with the experimental findings “. . . and only a quantitative comparison with the experimental data could provide a means for validating one of the two assumptions.” We have used experiment-specific computer modeling to test the hypothesis that it is not necessary to use locally orthotropic electric properties in activation models, because the intramural spread of activation is determined principally by myofiber rotation. The 3D spread of electric activation in a typical heart has been interpreted using a computer model that assumes transverse isotropy and incorporates detailed information on myofiber orientation and surface geometry in that heart (see Figure 5). The transverse isotropic model predicts a relatively uniform spread of activation in the plane transverse to the fiber direction at the point of activation. In parallel planes in front of and behind the stimulus, however, the predicted spread is elliptical with the major axis of the ellipse in opposite directions due to transmural fiber rotation. These predictions are consistent with the results of Colli-Franzone et al22 for an axially anisotropic medium but fail to reproduce even the qualitative features of experimental data obtained under exactly comparable conditions. The spread of activation in transverse planes in front of and behind the stimulus is in the same direction, whereas differences between predicted and observed results are most marked in subendocardial regions where variation of fiber orientation is relatively small. Whole-heart computer models of cardiac electric activation23–25 are playing an increasingly important role in aiding understanding of the mechanisms that give rise to reentrant arrhythmia and ventricular fibrillation (VF). Electric anisotropy and structural heterogeneity are clearly implicated in the initiation of ventricular tachycardia, wave break, and the induction of VF.26,27 Moreover, the spatio-temporal characteristics of reentrant wave front dynamics during VF26 –28 are all expected to depend on the electric orthotropy of ventricular tissue. Whereas the laminar arrangement of myocytes probably increases conduction safety29 transmurally by minimizing current-load mismatch, intrinsically slow electric propagation perpendicular to myolaminae may contribute to the initiation and maintenance of reentrant arrhythmia, particularly in ischemia. The present study strongly reinforces the view that the nonuniform laminar architecture of myocardium and associated electric orthotropy should be included in future computer models of cardiac arrhythmia. Possible artifacts introduced by the techniques used in this study warrant careful consideration. First, there is no evidence that introduction of the electrode array has affected the results. During data collection, ST-segment elevation was negligible and results were highly repeatable both within and between experiments. Second, reliable interpretation of the results presented here depends on accurate spatial registration of electric and structural data. Although the 3D locations of individual electrodes were identified with considerable precision using MR imaging, structural data were acquired from wax-embedded specimens that undergo substantial shrinkage and distortion during processing. This deformation was corrected using a mapping procedure that matches features in the wax-embedded specimens with corresponding structures in MR volumes. The effectiveness of this correction is independently corroborated by the near-perfect alignment of the principal axis of propagation with reconstructed myofiber orientation at the stimulus site in all experiments. Third, interpolation of 3D isochrones from activation times at electrodes that are nonuniformly distributed (1-mm spacing along each needle, but around 4 mm spacing between needles) can introduce bias, particularly in the initial phases of activation. We have elected to use a conservative interpolation scheme15 that tends initially to default to isotropic interpolation and have taken pains to identify a value of the R2 fitting parameter that is robust across the experimental data sets used (see Methods). Finally, it could be argued that virtual electrodes generated by intramural stimulation30 may play some role in the 3D activation patterns reported here. We have used near-threshold stimulation in all cases to minimize this possibility and can demonstrate no difference between unipolar and bipolar stimulation. In addition, we are confident that our results are not biased by intramural activation of the specialized conduction system, which has much greater penetration across the LV wall in pigs than in dogs or human beings.31 The transmural spread of activation was relatively uniform, with little evidence of the rapid local spread expected with activation of the conduction system. The spread of excitation was also consistent in magnitude and direction (relative to tissue microstructure) for each of the 13 sites across the LV wall at which intramural stimulation was applied. Moreover, the mean value of the principal velocities estimated transverse to the fiber direction in this study lies within the range of transverse velocities measured on the epicardial surface in the dog.2,12 Downloaded from http://circep.ahajournals.org/ by guest on June 1, 2015 Caldwell et al Limitations For more detailed analysis of the electric properties of ventricular myocardium, it would be desirable to characterize intramural activation at even higher spatial resolution. Ideally, the data obtained from such a study would be interpreted with a fully orthotropic computer model that incorporates detailed information not only on myofiber orientation and boundary geometry but also the laminar arrangement of myocytes and the distribution of other structures such as coronary blood vessels. This is beyond the scope of the current study, but work is proceeding in each of these areas in our laboratory. Conclusions This study provides experimental evidence that the spread of electric activation in LV myocardium is not uniform transverse to the local myofiber axis. We have demonstrated distinct directions of slow and faster propagation, transmurally. Transverse to the fiber direction, maximum CV coincides with the orientation of muscle layers, and minimum CV is approximately normal to the myolaminae. The effects of myocardial tissue architecture on the spread of electric activation have also been probed by comparing the predictions of an experiment-specific computer model with measured electric activation. Despite incorporation of detailed information on surface geometry and fiber orientation, models with axially anisotropic electric properties do not capture key features of the observed 3D spread of electric activation in the heart. This work therefore provides a consistent body of evidence that conflicts with the widespread assumption that electric coupling in ventricular myocardium is uniform transverse to the local myocyte axis and suggests that current views on side-to-side electric coupling in the heart should be revised. Acknowledgments We thank Dr Denis Loiselle and Professor Andrew Pullan for their contributions to this research. Sources of Funding This work was funded by a programme grant from the Health Research Council of New Zealand. Disclosures None. References 1. Saffitz JE, Kanter HL, Green KG, Tolley TK, Beyer EC. Tissue-specific determinants of anisotropic conduction velocity in canine atrial and ventricular myocardium. Circ Res. 1994;74:1065–1070. 2. Kleber AG, Rudy Y. Basic mechanisms of cardiac impulse propagation and associated arrhythmias. Physiologic Rev. 2004;84:431– 488. 3. LeGrice IJ, Smaill BH, Chai LZ, Edgar SG, Gavin JB, Hunter PJ. Laminar structure of the heart: ventricular myocyte arrangement and connective tissue architecture in the dog. Am J Physiology. 1995;269: H571–H582. Orthotropic Intramural Spread of LV Activation 439 4. Young AA, Legrice IJ, Young MA, Smaill BH. Extended confocal microscopy of myocardial laminae and collagen network. J Microsc. 1998;192:139 –150. 5. Sands GB, Gerneke DA, Hooks DA, Green CR, Smaill BH, Legrice IJ. Automated imaging of extended tissue volumes using confocal microscopy. Micros Res Technique. 2005;67:227–239. 6. LeGrice IJ, Takayama Y, Covell JW. Transverse shear along myocardial cleavage planes provides a mechanism for normal systolic wall thickening. Circ Res. 1995;77:182–193. 7. Costa KD, Takayama Y, McCulloch AD, Covell JW. Laminar fiber architecture and three-dimensional systolic mechanics in canine ventricular myocardium. Am J Physiol. 1999;276:H595–H607. 8. Arts T, Costa KD, Covell JW, McCulloch AD. Relating myocardial laminar architecture to shear strain and muscle fiber orientation. Am J Physiology Heart Circ Physiol. 2001;280:H2222–H2229. 9. Chen J, Liu W, Zhang H, Lacy L, Yang X, Song SK, Wickline SA, Yu X. Regional ventricular wall thickening reflects changes in cardiac fiber and sheet structure during contraction: quantification with diffusion tensor MRI. Am J Physiology Heart Circ Physiol. 2005;289: H1898 –H1907. 10. Helm PA, Tseng HJ, Younes L, McVeigh ER, Winslow RL. Ex vivo 3D diffusion tensor imaging and quantification of cardiac laminar structure. Magn Reson Med. 2005;54:850 – 859. 11. Hooks DA, Tomlinson KA, Marsden SG, LeGrice IJ, Smaill BH, Pullan AJ, Hunter PJ. Cardiac microstructure: implications for electrical propagation and defibrillation in the heart. Circ Res. 2002;91:331–338. 12. Frazier DW, Krassowska W, Chen PS, Wolf PD, Danieley ND, Smith WM, Ideker RE. Transmural activations and stimulus potentials in threedimensional anisotropic canine myocardium. Circ Res. 1988;63:135–146. 13. Taccardi B, Punske BB, Macchi E, MacLeod RS, Ershler PR. Epicardial and intramural excitation during ventricular pacing: effect of myocardial structure. Am J Physiol Heart Circ Physiol. 2008;294:H1753–H1766. 14. Rogers JM, Melnick SB, Huang J. Fiberglass needle electrodes for transmural cardiac mapping. IEEE Trans Biomed Eng. 2002;49: 1639 –1641. 15. Ni Q, MacLeod RS, Lux RL. Three-dimensional activation mapping in ventricular muscle: interpolation and approximation of activation times. Ann Biomed Eng. 1999;27:617– 626. 16. Gerneke DA, Sands GB, Ganesalingam R, Joshi P, Caldwell BJ, Smaill BH, Legrice IJ. Surface imaging microscopy using an ultramiller for large volume 3D reconstruction of wax- and resin-embedded tissues. Microsc Res Technique. 2007;70:886 – 894. 17. Karlon WJ, Covell JW, McCulloch AD, Hunter JJ, Omens JH. Automated measurement of myofiber disarray in transgenic mice with ventricular expression of ras. Anat Rec. 1998;252:612– 625. 18. 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;44:701–712. 19. Taccardi B, Punske BB, Sachse F, Tricoche X, Colli-Franzone P, Pavarino LF, Zabawa C. Intramural activation and repolarization sequences in canine ventricles: experimental and simulation studies. J Electrocardiol. 2005;38:131–137. 20. Hooks DA, Trew ML, Caldwell BJ, Sands GB, LeGrice IJ, Smaill BH. Laminar arrangement of ventricular myocytes influences electrical behavior of the heart. Circ Res. 2007;101:e103– e112. 21. Colli-Franzone P, Guerri L, Taccardi B. Modeling ventricular excitation: axial and orthotropic anisotropy effects on wavefronts and potentials. Math Biosci. 2004;188:191–205. 22. Colli-Franzone P, Pavarino LF, Taccardi B. Simulating patterns of excitation, repolarization and action potential duration with cardiac Bidomain and Monodomain models. Math Biosci. 2005;197:35– 66. 23. Nielsen PM, Le Grice IJ, Smaill BH, Hunter PJ. Mathematical model of geometry and fibrous structure of the heart. Am J Physiol Heart Circ Physiol. 1991;260:H1365–H1378. 24. Berenfeld O, Jalife J. Purkinje-muscle reentry as a mechanism of polymorphic ventricular arrhythmias in a 3-dimensional model of the ventricles. Circ Res. 1998;82:1063–1077. 25. ten Tusscher KH, Hren R, Panfilov AV. Organization of ventricular fibrillation in the human heart. Circ Res. 2007;100:e87– e101. 26. Qu Z, Kil J, Xie F, Garfinkel A, Weiss JN. Scroll wave dynamics in a three-dimensional cardiac tissue model: roles of restitution, thickness, and fiber rotation. Biophys J. 2000;78:2761–2775. 27. Berenfeld O, Wellner M, Jalife J, Pertsov AM. Shaping of a scroll wave filament by cardiac fibers. Phys Rev E Stat Nonlinear Soft Matter Physics. 2001;63:619 – 624. Downloaded from http://circep.ahajournals.org/ by guest on June 1, 2015 440 Circ Arrhythmia Electrophysiol August 2009 28. Xie F, Qu Z, Yang J, Baher A, Weiss JN, Garfinkel A. A simulation study of the effects of cardiac anatomy in ventricular fibrillation. J Clin Invest. 2003;113:686 – 693. 29. Rohr S, Kucera JP, Fast VG, Kleber AG. Paradoxical improvement of impulse conduction in cardiac tissue by partial cellular uncoupling. Science. 1997;275:841– 844. 30. Wikswo JP, Lin S-F, Abbas RA. Virtual electrodes in cardiac tissue: a common mechanism for anodal and cathodal stimulation. Biophys J. 1995;69:2195–2210. 31. Tranum-Jensen J, Wilde AA, Vermeulen JT, Janse MJ. Morphology of electrophysiologically identified junctions between Purkinje fibers and ventricular muscle in rabbit and pig hearts. Circ Res. 1991;69:429 – 437. CLINICAL PERSPECTIVE The anisotropy of cardiac tissue is a key determinant of 3D electric propagation, affecting the spread of activation from an intramural focus, the spatio-temporal dynamics of ventricular arrhythmia, and the response of the heart to defibrillation. The traditional view of anisotropy is that electric propagation is fastest in the direction of the myofiber and uniformly slower in all directions transverse to the myofiber. In this study, high-resolution activation maps combined with tissue microstructural analysis reveal that in fact, 3 distinct velocities of propagation can be defined at any point within ventricular myocardium. Most rapid propagation is observed in the direction of myofibers, and intermediate and slowest propagation is seen within and across microscopic laminar groupings of myocytes. This new view of electric anisotropy, when incorporated into models of the electric behavior of the heart, will enhance understanding of all anisotropy-dependent phenomena. Downloaded from http://circep.ahajournals.org/ by guest on June 1, 2015 Three Distinct Directions of Intramural Activation Reveal Nonuniform Side-to-Side Electrical Coupling of Ventricular Myocytes Bryan J. Caldwell, Mark L. Trew, Gregory B. Sands, Darren A. Hooks, Ian J. LeGrice and Bruce H. Smaill Circ Arrhythm Electrophysiol. 2009;2:433-440; originally published online June 18, 2009; doi: 10.1161/CIRCEP.108.830133 Circulation: Arrhythmia and Electrophysiology is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231 Copyright © 2009 American Heart Association, Inc. All rights reserved. Print ISSN: 1941-3149. Online ISSN: 1941-3084 The online version of this article, along with updated information and services, is located on the World Wide Web at: http://circep.ahajournals.org/content/2/4/433 Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published in Circulation: Arrhythmia and Electrophysiology 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: http://www.lww.com/reprints Subscriptions: Information about subscribing to Circulation: Arrhythmia and Electrophysiology is online at: http://circep.ahajournals.org//subscriptions/ Downloaded from http://circep.ahajournals.org/ by guest on June 1, 2015