Transmural dispersion of repolarization and the T wave

See article by Taggart et al. [50] (pages 454-462) in this issue. Studies conducted over the past dozen years or so have demonstrated that ventricular myocardium is not homogeneous, as previously thought, but is comprised of at least three electrophysiologically distinct cell types: epicardial, M and endocardial cells (see [1,2] for reviews). These three cell types have also been shown to possess different pharmacologic profiles and to respond differently to a variety of pathophysiologic states [3–8]. The three cells types differ principally with respect to repolarization characteristics. Ventricular epicardial and M cells display action potentials with a prominent transient outward current (Ito)-mediated phase 1, giving rise to a notched appearance of the action potential. The absence of a prominent notch in the endocardium is a consequence of a much smaller Ito. Similar regional differences in Ito are found in canine, feline, rabbit, rat and human ventricular myocytes (see [1] for references). Recent studies also indicate that Ito and the action potential notch are much larger in right vs. left ventricular epicardial [9] and M [10] cells. The transmural gradient in the amplitude of the Ito-mediated action potential notch underlies the normal J wave or J point elevation in the ECG [11] and its accentuation, particularly in the right ventricle, contributes to the development of life-threatening arrhythmias in patients with the Brugada syndrome and various forms of idiopathic ventricular fibrillation [12,13]. The presence of a prominent Ito in right ventricular epicardium has also been shown to sensitize this tissue to the effects of ischemia [14]. Accentuation of the action potential notch and eventual loss of the dome in right ventricular epicardium but not endocardium has been shown to contribute to ischemia-induced ST segment elevation [12]. The M cells are distinguished by the ability of their action potentials to prolong more than those of epicardium or endocardium in response to a slowing of rate and/or in response to agents that normally prolong the action potential [3]. These features of the M cell are due at least in part to the presence of a smaller slowly activating delayed rectifier current (IKs) [15], a larger late sodium current (late INa) [1,16] and a larger sodium–calcium exchange current [17]. The rapidly activating delayed rectifier (IKr) and inward rectifier (IK1) currents appear to be homogeneously distributed across the ventricular wall of the canine heart. Cells with repolarization characteristics of M cells have been described in the canine, guinea pig, rabbit, pig and human ventricles [3–8,15,18–33]. Three studies have failed to discern M cells in the ventricles of the pig, guinea pig and rat [7,34,35]. Other studies, while clearly demonstrating the presence of M cells in the ventricles of the canine heart in vitro, failed to delineate the unique cell type in vivo [23,36]. Methodological considerations thought to be responsible for these differences have been discussed at great length [1,2]. When individual myocytes are enzymatically dissociated from the respective layers of the canine left ventricular (LV) wall, transmural dispersion of action potential duration (TD-APD) recorded at slow rates is of the order of approximately 200 ms. When these cells are in the intact wall of the LV, this intrinsic dispersion is greatly reduced due to electrotonic interactions among the different cell types. Studies conducted in perfused canine LV wedge preparations generally display TD-APD values ranging between 40 and 60 ms at slow rates (basic cycle length [BCL]=2000 ms), depending on the size of the heart. At faster rates, TD-APD is further reduced. Transmural dispersion of repolarization (TDR) averages 34±18 ms at a BCL of 1000 ms [31]. In addition to disparities in the time of final repolarization, there are important differences in the voltage of the action potential plateau, which can only be discerned using intracellular microelectrode techniques. These differences in the trajectory of repolarization are responsible for the inscription of the electrocardiographic T wave. Data from the arterially-perfused wedge indicate that currents flowing down voltage gradients on either side of the M region underlie the T wave (Fig. 1) [32]. The interplay between these opposing currents establishes the height and width of the T wave as well as the degree to which the T wave may be interrupted, leading to a bifurcated or notched appearance. The voltage gradients result from a more positive plateau potentials in the M region than in epicardium or endocardium and from differences in the time-course of phase 3 of the action potential of the three predominant ventricular cell types. Under baseline and long QT conditions, the epicardial response is the earliest to repolarize and the M cell action potential is the last. Full repolarization of the epicardial action potential is coincident with peak of the T wave and repolarization of the M cells coincides with the end of the T wave so that the Tpeak–Tend interval provides an index of transmural dispersion of repolarization [32,37,38]. Voltage gradients on either side of the M region underlie the inscription of the electrocardiographic T wave. Top: action potentials simultaneously recorded from endocardial (Endo), epicardial (Epi) and M region sites of an arterially-perfused canine left ventricular wedge preparation. Middle: ECG recorded across the wedge. Bottom: computed voltage differences between the epicardium and M region action potentials (ΔVM−Epi) and between the M region and endocardium responses (ΔVEndo−M). If these traces are representative of the opposing voltage gradients on either side of the M region, responsible for inscription of the T wave, then the weighted sum of the two traces should yield a trace (middle trace in bottom grouping) resembling the ECG, which it does. The voltage gradients are weighted to account for differences in tissue resistivity between M and Epi and Endo and M regions, thus yielding the opposing currents flowing on either side of the M region. (A) Under control conditions the T wave begins when the plateau of epicardial action potential separates from that of the M cell. As epicardium repolarizes, the voltage gradient between epicardium and the M region continues to grow giving rise to the ascending limb of the T wave. The voltage gradient between the M region and epicardium (ΔVM−Epi) reaches a peak when the epicardium is fully repolarized — this marks the peak of the T wave. On the other end of the ventricular wall, the endocardial plateau deviates from that of the M cell, generating an opposing voltage gradient (ΔVEndo−M) and corresponding current that limits the amplitude of the T wave and contributes to the initial part of the descending limb of the T wave. The voltage gradient between the endocardium and the M region reaches a peak when the endocardium is fully repolarized. The gradient continues to decline as the M cells repolarize. All gradients are extinguished when the longest M cells are fully repolarized. (B) d-Sotalol (100 μM) prolongs the action potential of the M cell more than those of the epicardial and endocardial cells, giving rise to a widening of the T wave and a prolongation of the QT interval. The greater separation of epicardial and endocardial repolarization times also gives rise to a notch in the descending limb of the T wave. Once again, the T wave begins when the plateau of epicardial action potential diverges from that of the M cell. The same relationships as described for (A) are observed during the remainder of the T wave. The d-sotalol-induced increase in dispersion of repolarization across the wall is accompanied by a corresponding increase in the Tpeak–Tend interval in the pseudo-ECG. Modified from [32] with permission. Studies designed to quantitate TDR in vivo have yielded variable results, depending on the anesthesia and recording methods employed [39]. Most if not all anesthetics are thought to importantly reduce transmural dispersion of repolarization owing to their action in blocking late INa. Block of this current preferentially abbreviates the action potential of the M cell and dramatically flattens its APD–rate relationship [28,40]. Agents like pentobarbital are particularly effective in reducing TDR because of the anesthetic's potent effect of blocking late INa as well at IKs. The effect of this dual ion channel inhibition is to prolong the epicardial and endocardium action potential, but abbreviate that of the M cell in the canine heart. A relatively small transmural dispersion of repolarization has been reported (at slow rates) in in vivo studies that have used pentobarbital or α-chloralose for anesthesia [22,23,36] vs. studies that have used other agents including isoflurane [25,41,42] or halothane [22]. Consistent with these findings, the development of Torsade de Pointes (TdP) in vivo has often met with failure when sodium pentobarbital was used for anesthesia [22,43], whereas TdP could be readily induced when halothane or isoflurane was employed [22,25,41,44] or when no anesthesia was used [45,46]. In the case of α-chloralose anesthesia, TdP is generally observed when measures in addition to IKr block are included in the protocol (α-adrenergic agonists and/or hypokalemia) [47]. A recent study by Yamamoto et al. [48] demonstrates the capability of both pentobarbital and isoflurane to suppress quinidine and astemizole-induced TdP, suggesting that both reduce TDR [49]. Studies conducted using myocytes and tissue slices isolated from the human heart have demonstrated cell characteristics and ionic distinctions very similar to those described in the dog. The extent to which transmural dispersion of repolarization exists within the human heart has been inferred on the basis of measurement of Tpeak−Tend values. Lubinski et al. reported that the normal human heart exhibits a Tpeak−Tend interval of 62.4±7.5 ms (in unanaesthetized subjects free of medication) and that this index of transmural dispersion of repolarization is augmented to 79.6±9.6 ms in patients with the long QT syndrome. Direct measurement of transmural dispersion of repolarization in the human heart was previously unavailable. In this issue of Cardiovascular Research, Taggart et al. [50] report the result of a unique study designed to measure transmural dispersion in hearts of patients undergoing coronary artery surgery. The authors sought to quantify TD-APD under baseline and ischemic conditions. The results, while interesting, must be interpreted cautiously because the conditions under which the recordings were obtained and the methodologies employed, for reasons beyond the control of the investigators, are less than ideal. First, the patients were medicated with a number of drugs (morphine, Hyoscine, Midazolam, fentanyl, pancuronium, nitrous oxide and isoflurane), several of which are known to block INa in excitable cells and the heart in particular [51–54]. As discussed above, block of late INa alone is sufficient to dramatically reduce transmural dispersion [28,40]. The extent to which these agents and anesthetics block other ion channels in the heart and thus further reduce transmural dispersion (e.g. IKs) is not known. Before accepting the data as representative of transmural dispersion of repolarization in the ‘normal’ human heart, it would be prudent to ascertain the effects of these medications on the electrical activity of the heart, particularly on TDR, the parameter being evaluated. An approximation of this can be obtained by comparing the ECG (especially in left precordial leads) before and after administration of the sundry medications. For example, pentobarbital dramatically reduces the amplitude of the T wave, owing to the effect of the anesthetic to reduce voltage differences among the three cell types [28]. Regrettably, such ECG data are not presented in the study by Taggart et al. and we are told that none are available. The authors argue the point that isoflurane, one of the anesthetics used in their study does not diminish TDR, but in fact may augment this index secondary to cellular uncoupling, which should facilitate the induction of TdP under long QT conditions. A recent study designed to contrast the effects of isoflurane, halothane and pentobarbital fails to support this contention, demonstrating instead an effect of isoflurane in reducing the incidence of TdP [48]. Second, as the authors point out, these studies were done on diseased hearts. In order to avoid ischemic zones, recordings were obtained from different regions of the LV in each patient. This would not be problematic if the data were not averaged. As work from our group has pointed out repeatedly, the position of the M cells within the wall is not fixed. The M cell layer displaying the longest action potentials appears to shift from the deep subendocardium to the deep subepicardium as one moves from the anterior wall to the lateral and posterior wall of the left ventricle of the canine heart. If the same is true in the human heart, it is critically important that measurements be made in the same region in each patient to minimize this problem. If the longest APDs are encountered in a different region in each patient, the average data would be expected to show no significant transmural gradient, despite the fact that a pattern may be present in each heart. Third, the authors indicate that “in only two out of twenty-one patients was the longest activation recovery interval (ARI) observed in one of the midmyocardial electrodes”. In the dog anterior LV wall, M cells with the longest action potential are not found in the midmyocardium, but in the deep subendocardium (Fig. 2) [32]. If the same is true in the human heart, it is unlikely that Taggart and coworkers would have been able to record the cells with the longest action potential since the recording electrodes used (spanning 6 mm from the first recording node to the last) did not extend to the deep subendocardium, let alone the endocardium of the human heart whose anterior wall depth usually ranges between 9 and 11 mm. Moreover the placement of the epicardial electrode 0.85 mm from the surface is unlikely to have permitted recording of the briefest surface epicardial response in many cases. Thus, the study may have further underestimated TDR due to the fact that the recording methodology by design precludes measurement of the shortest and longest action potentials. While unipolar recordings usually provide a relatively accurate approximation of local repolarization, they do not approach the accuracy of a map obtained using an intracellular microelectrode probe, as illustrated in Fig. 2. The wide spacing of electrodes and the shorting of the extracellular space by the transmural needle electrode used are additional factors that may contribute to underestimation of intrinsic TDR in this study. Fourth, the study challenges both classical and modern concepts for the genesis of the T wave, based on its failure to observe either a linear or biphasic transmural ventricular gradient of repolarization, previously invoked to explain the T wave. For lack of another explanation, the authors suggest apico–basal differences in repolarization (not measured in this study) as the basis for the T wave. While the unipolar recordings used in their study can reasonably approximate repolarization time, they are unable to discern differences in plateau voltage. Recent studies involving canine LV wedge preparations have shown that much of the T wave is inscribed as a consequence of opposing currents flowing down transmural voltage gradients on either side of the M region, caused by differences in plateau voltage of the three predominant cell types and that apico–basal gradients contribute little to the T wave (Fig. 1) [32]. Thus, transmural voltage gradients can contribute importantly to inscription of the T wave even if transmural dispersion of final repolarization, as measured using an ARI, is small. Finally, Taggart et al. also recorded transmural signals during a 3-min period of global ischemia produced by clamping the aorta between the input from the pump oxygenator and the coronary arteries. They report a homogeneous abbreviation of ARI throughout the anterior wall of the human LV. This result appears to be at variance with in vitro data demonstrating a preferential abbreviation of the epicardial action potential under ischemic conditions [14]. As the authors point out, this disparity may be due to electrotonic interaction of epicardium with the subtending tissues or to the relatively brief period of ischemia used in their protocols. Other factors not discussed include the fact that preferential abbreviation of the epicardial action potential under ischemic conditions has previously been demonstrated only in RV epicardium and that this phenomenon depends critically on the presence of a prominent Ito[14]. Ischemia-induced APD abbreviation occurs most readily in RV epicardium because Ito is much more prominent in this tissue than it is in either LV epicardium or any other tissue of either right or left ventricular origin [9,10,18]. Thus, the homogeneous response to ischemia may also be attributable to the fact that the left ventricle was studied and to the possibility that the medications or anesthesia used may have inhibited the intrinsic Ito. While it is challenging and exciting to obtain data from the human heart in vivo, we must be mindful of the fact that experimental conditions are often far from ideal. Accordingly, the data must be interpreted with great caution. I applaud the authors for pointing out many of the methodological pitfalls and have attempted to amplify these further in the discussion above. I think it safe to conclude that the extent to which transmural electrical heterogeneity exists within the human heart remains to be definitively determined. Transmural distribution of action potential duration and tissue resistivity. (A) Schematic diagram of the arterially perfused canine LV wedge preparation. The wedge is perfused with Tyrode's solution via a small native branch of the left descending coronary artery and stimulated from the endocardial surface. Transmembrane action potentials are recorded simultaneously from epicardial (Epi), M region (M) and endocardial (Endo) sites using three floating microelectrodes. A transmural ECG is recorded along the same transmural axis across the bath, registering the entire field of the wedge. (B) Histology of a transmural slice of the left ventricular wall near the epicardial border. The region of sharp transition of cell orientation coincides with the region of high tissue resistivity depicted in (D) and the region of sharp transition of action potential duration illustrated in (C). (C) Distribution of conduction time (CT), APD90 and repolarization time (RT=APD90+CT) in a canine left ventricular wall wedge preparation paced at BCL of 2000 ms. A sharp transition of APD90 is present between epicardium and subepicardium. Epi, epicardium; M, M cell; Endo, endocardium; RT, repolarization time; CT, conduction time. Width of LV wedge ranged between 9 and 14 mm. Error bars represent S.E.M. (n=5). (D) Distribution of total tissue resistivity (Rt) across the canine LV wall. Transmural distances at 0% and 100% represent epicardium and endocardium, respectively. *, P<0.01 compared with Rt at mid-wall. Tissue resistivity increases most dramatically between deep subepicardium and epicardium. Error bars represent S.E.M. (n=5). Modified from [31] with permission. Supported by grants from the National Institutes of Health (HL 47678), the American Heart Association, New York State Affiliate, and the Masons of New York State and Florida.