Mechanism of repolarization alternans has restitution of action potential duration dependent and independent components

INTRODUCTION: Investigation of relationship between diastolic-interval (DI)-dependent restitution of action potential duration (APD) and alternans of APD has produced conflicting results. We used a novel pacing protocol to determine the role of restitution in alternans by minimizing changes in DI preceding each activation. METHODS: Transmembrane potentials were recorded from right ventricular endocardial tissue isolated from five dogs. We used three pacing sequences: (i) The tissue was paced at a constant DI for 100 beats. (ii) The DIs were changed randomly between two sequences of constant DI. (iii) Each constant DI trial was followed by constant cycle length trial where pacing cycle length was equal to average cycle length during previous constant DI trial. RESULTS: Alternans of APD occurred even when DIs preceding each activation were invariant. Slopes of restitution during constant DI pacing were both negative and positive and were much larger than unity. Alternans amplitude during constant cycle length pacing was larger than during constant DI, 32.2 +/- 12.3 versus 7.5 +/- 2.8 msec, P < 0.01. Random perturbation of DI decreased alternans amplitude during constant DI pacing from 14.7 +/- 4.8 to 10.5 +/- 3.4 msec, P < 0.01. CONCLUSION: Our results indicate that mechanism of repolarization alternans has restitution-dependent and restitution-independent components. However, our results also provide direct evidence that shows that DI-dependent restitution of APD is not a necessary mechanism for the alternans to exist. Ability to pace with explicit control of DI provides a novel approach to dissect mechanisms of alternans into restitution-dependent and restitution-independent effects.     

Effect of action potential duration and conduction velocity restitution and their spatial dispersion on alternans and the stability of arrhythmias

INTRODUCTION: The slope of the action potential duration (APD) restitution curve has been used to explain wavebreaks during arrhythmia initiation and maintenance. This hypothesis remains incomplete to fully describe the experimental data. Other factors contributing to wavebreaks must be studied to further understand arrhythmia dynamics. METHODS AND RESULTS: Control APDs were measured from isolated rabbit hearts using a monophasic action potential probe. APD and conduction velocity (CV) restitution were quantified over the heart surface for two drugs, diacetyl monoxime (DAM) and cytochalasin D (CytoD), using a dual camera video imaging system. For all pacing intervals: (1) control APDs were shorter than for CytoD but longer than for DAM; and (2) CV was greater for CytoD compared with DAM. APD dispersion increased as pacing interval decreased for both drugs. For DAM, increased dispersion was due to a difference in APD restitution between the right and left ventricle. For CytoD, increased dispersion was due to discordant alternans, with no significant spatial variation in restitution. Fibrillation was sustained only in the control hearts; with DAM, stable reentry was sustained with shorter APD and cycle length compared with CytoD for which only nonsustained unstable reentry occurred. CONCLUSION: Alternans and arrhythmia dynamics are affected by the spatial dispersion of APD restitution as well as CV restitution, not simply the slope of APD restitution. Therefore, a direct link of the APD restitution slope to alternans and arrhythmia dynamics in rabbit heart does not exist. Designing antiarrhythmic drugs to alter only the restitution slope may not be appropriate.     

Importance of spatiotemporal heterogeneity of cellular restitution in mechanism of arrhythmogenic discordant alternans

BACKGROUND: Spatially discordant cellular alternans form a substrate for development of unidirectional block and ventricular fibrillation. However, the mechanisms responsible for discordant alternans remain poorly understood. Previous work suggests electrical restitution is critical to the development of alternans in single cells. OBJECTIVES: The purpose of this study was to investigate the hypothesis that spatial and temporal heterogeneities of restitution underlie the mechanism eliciting discordant alternans. METHODS: Steady-state pacing was used to elicit concordant cellular alternans in nine Langendorff-perfused guinea pig hearts. A single extrastimulus (S2) was applied every 51st beat following either the even or the odd beat of alternans. The cellular response to S2 was determined using optical mapping to generate action potential duration (APD) restitution curves from 256 ventricular sites for both the even and the odd beats. RESULTS: Restitution kinetics were temporally heterogeneous during alternans, as restitution curves between the even and the odd beats differed significantly. Temporal heterogeneity was quantified by the average separation of restitution between the two curves, or Delta-restitution. Delta-Restitution was spatially heterogeneous and proportional to the amount of alternans at a given ventricular site. A computer simulation based on the experimental results showed the mechanism of discordant alternans was dependent on both spatial and temporal heterogeneities of restitution. CONCLUSION: Both temporal and spatial heterogeneities of restitution exist during cellular alternans in the intact heart. Temporal heterogeneities of restitution, quantified by Delta-restitution, are proportional to the magnitude of cellular alternans. The combination of spatial and temporal heterogeneities of restitution may underlie the genesis of discordant alternans.     

Restitution dynamics during pacing and arrhythmias in isolated pig hearts

INTRODUCTION: The dependence of action potential duration (APD) on the preceding diastolic interval (DI), i.e., restitution, has been purported to predict the development of alternans and reentrant arrhythmias. However, restitution depends on the history of activation (i.e., memory), and its relevance to arrhythmia induction and maintenance is unknown. METHODS AND RESULTS: Using a dual-camera video imaging system, we recorded action potentials from thousands of sites on the surface of the isolated pig heart. A steady-state pacing (SSP) protocol was performed to generate the SSP APD restitution curve. During SSP, the minimum DI and APD were 57 +/- 6 ms and 107 +/- 6 ms, respectively. The restitution slope was >1 for DIs <85 +/- 5 ms; however, alternans were not observed. Abrupt decreases in cycle length (CL) resulted in a rapid (<5 beats) decrease in APD followed by a slower decrease to "steady state." DI, APD pairs for the initial beats following these rate changes were significantly above the SSP restitution curve. DI, APD pairs measured during sustained ventricular fibrillation clustered significantly below the SSP restitution curve, at significantly shorter APDs (57 +/- 4 ms) and DIs (49 +/- 6 ms) than could be achieved during SSP. In addition, abrupt increases in CL following SSP resulted in APDs significantly shorter than those predicted from the SSP restitution curve. CONCLUSION: Our results indicate that the responses of APD and DI to sudden rate changes and during arrhythmias are not predicted by the SSP restitution relationship. Acute dynamics act to damp out the proarrhythmic oscillations predicted from the SSP restitution curve.     

The restitution portrait: a new method for investigating rate-dependent restitution

INTRODUCTION: Electrical restitution, relating action potential duration (APD) to diastolic interval (DI), was believed to determine the stability of heart rhythm. However, recent studies demonstrate that stability also depends on long-term APD changes caused by memory. This study presents a new method for investigation of rate- and memory-dependent aspects of restitution and for assessment of mapping models of APD. METHODS AND RESULTS: Bullfrog ventricular myocardium was paced with a "perturbed downsweep protocol." Starting from a basic cycle length (BCL) of 1,000 ms, the tissue was paced until steady state was achieved, followed by single beats of longer and shorter cycle lengths. BCL was decreased by 50 to 100 ms and the process repeated. All APDs were plotted as a function of the preceding DI, which allowed simultaneous observation of dynamic, S1-S2, and two constant-BCL restitution curves in a "restitution portrait." Responses were classified as 1:1 (stimulus:response), transient 2:2, or persistent 2:2 (alternans) and were related to the slopes of the restitution curves. None of these slopes approached unity for the persistent 2:2 response, demonstrating that the traditional restitution condition does not predict alternans. The restitution portrait was used to evaluate three mapping models of APD. The models with no memory and with one-beat memory did not produce restitution portraits similar to the experimental one. A model with two-beat memory produced a qualitatively similar portrait. CONCLUSION: The restitution portrait allows a more comprehensive assessment of cardiac dynamics than methods used to date. Further study of models with memory may result in a clinical criterion for electrical instability.     

均匀心肌组织中心室颤动机制的仿真研究

目的研究心肌中的局部传导阻滞现象,在细胞水平上解释颤动中螺旋波的形成和破裂为多重子螺旋波的机制,强调动作电位时程(APD)恢复特性对心室颤动的诱发和维系具有重要意义。方法从心肌细胞APD恢复特性和传导速度(CV)恢复特性出发,利用非线性动力学方法,基于一维均匀心肌组织模型,观察不同起搏频率下的APD现象。结果证明了单个心肌细胞的APD交替以及均匀心肌组织中的非同步交替现象,并在起搏位置的远端发现了局部传导阻滞。结论均匀心肌组织中,由于APD和CV恢复特性的相互作用,在高频率起搏时,足以产生局部传导阻滞,从而可能进一步诱发螺旋波及破裂成心室颤动。     

Role of calcium cycling versus restitution in the mechanism of repolarization alternans

T-wave alternans, a powerful marker of arrhythmic events, results from alternation in action potential duration (APD). The underlying cellular mechanism of APD alternans is unknown but has been attributed to either intracellular calcium (Ca2+) cycling or membrane ionic currents, manifested by a steep slope of cellular APD restitution. To address these mechanisms, high-resolution optical mapping techniques were used to measure action potentials and Ca2+ transients simultaneously from hundreds of epicardial sites in the guinea pig model of pacing-induced T-wave alternans (n=7). The pacing rates (ie, alternans threshold) at which T-wave (369+/-11 bpm), APD (369+/-21 bpm), and Ca2+ (371+/-29 bpm) alternans first appeared were comparable. Importantly, the site of origin of APD alternans and Ca2+ alternans consistently occurred together near the base of the left ventricle, not where APD restitution was steepest. In addition, APD and Ca2+ alternans were remarkably similar both spatially and temporally during discordant alternans. In conclusion, the mechanism underlying T-wave alternans in the intact heart is more closely associated with intracellular Ca2+ cycling rather than APD restitution.     

Alternans of epicardial repolarization as a localized phenomenon in man

Electrophysiological alternans (beat-to-beat alternation of the configuration of the action potential and/or electrocardiogram) may be important in the causation of ventricular arrhythmias. We recorded monophasic action potentials from the left ventricular epicardium in patients undergoing routine cardiac surgery. We set out to determine: (a) whether a small increase in atrial pacing rate could elicit electrophysiological alternans; (b) whether different phases of the operation influence the incidence and; (c) whether electrical alternans was a localized phenomenon. Thirty-six patients were studied, and alternans of action potential duration was observed in 14 (39%). The difference between alternate long and short action potential durations ranged from 4 to 112 ms. The mean differences during each stage of the protocol were: pacing before bypass 33.3 +/- 22.7 ms (three of 17 patients); pacing after bypass 46.7 +/- 37.8 ms (nine of 36 patients); pacing during transient graft occlusion 28.0 +/- 23.1 ms (five of 17 patients); pacing following release of the grafts 29.6 +/- 30.0 ms (five of 17 patients). None of the patients showed any evidence of alternans in the electrocardiogram or mechanical alternans in radial artery pressure. In seven of the 14 patients showing alternans, recordings were made in closely adjacent (approximately 1 cm) areas showing that alternans could be a localized phenomenon. The findings indicate that electrical alternans was a frequent occurrence in this study (39% of patients) and may be a localized phenomenon.     

Spatially discordant alternans in cardiac tissue: role of calcium cycling

Spatially discordant alternans, where the action potential duration (APD) and intracellular calcium transient (Ca(i)) alternate with opposite phase in different regions of tissue, is known to promote wave break and reentry. However, this phenomenon is not completely understood. It is known that alternans at the cellular level can be caused by dynamical instabilities arising from either membrane voltage (V(m)) attributable to steep APD restitution or to calcium (Ca) cycling. Here, we used a mathematical model of intracellular Ca cycling, coupled with membrane ion currents, to investigate the dynamics of V(m) and Ca(i) transient alternans in an isolated cell, in two electrotonically coupled cells, and in 1D spatially homogeneous tissue. Our main finding is a novel instability mechanism in which the bidirectional coupling of V(m) and Ca(i) can drive the Ca(i) transient of two neighboring cells to be out of phase. This instability is manifested in cardiac tissue by the dynamical formation of spatially discordant alternans. In this case, Ca(i) transient alternans can reverse phase over a length scale of one cell, whereas APD alternans reverses phase over a much longer length scale set by the electrotonic coupling. We analyze this mechanism in detail and show that it is a robust consequence of experimentally established properties of the bidirectional coupling between Ca cycling and V(m) dynamics. Finally, we address the experimental relevance of these findings and suggest physiological conditions under which these patterns can be observed.     

From pulsus to pulseless: the saga of cardiac alternans

Computer simulations and nonlinear dynamics have provided invaluable tools for illuminating the underlying mechanisms of cardiac arrhythmias. Here, we review how this approach has led to major insights into the mechanisms of spatially discordant alternans, a key arrhythmogenic factor predisposing the heart to re-entry and lethal arrhythmias. During spatially discordant alternans, the action potential duration (APD) alternates out of phase in different regions of the heart, markedly enhancing dispersion of refractoriness so that ectopic beats have a high probability of inducing reentry. We show how, at the cellular level, instabilities in membrane voltage (ie, steep APD restitution slope) and intracellular Ca (Cai) cycling dynamics cause APD and the Cai transient to alternate and how the characteristics of alternans are affected by different "modes" of the bidirectional coupling between voltage and Cai. We illustrate how, at the tissue level, additional factors, such as conduction velocity restitution and ectopic beats, promote spatially discordant alternans. These insights have illuminated the mechanistic basis underlying the clinical association of cardiac alternans (eg, T wave alternans) with arrhythmia risk, which may lead to novel therapeutic approaches to avert sudden cardiac death.     

The effects of atrial ganglionated plexi stimulation on ventricular electrophysiology in a normal canine heart

Aims

Atrial ganglionated plexi (GP) have been shown to modulate atrial electrophysiology and play an important role in atrial fibrillation initiation and maintenance. The purpose of this study was to investigate the effects of atrial GP stimulation (GPS) on ventricular refractoriness, restitution properties and electrical alternans.

Methods

In 12 anesthetized dogs, two multiple electrode catheters were sutured at left and right ventricular free walls for recording. Monophasic action potentials were recorded from six epicardial ventricular sites. Ventricular effective refractory period (ERP), action potential duration (APD) restitution properties and APD alternans were measured at baseline and during GPS.

Results

Compared with baseline, GPS significantly prolonged ventricular ERP and APD at all sites and decreased their spatial dispersions (P?<?0.05 for all). GPS also significantly flattened ventricular restitution curves and decreased the maximal slope of restitution curves at each site (P?<?0.05 for all). APD alternans occurred at shorter pacing cycle length at each site during GPS when compared with baseline (P?<?0.05 for all).

Conclusions

GPS prolonged ventricular ERP, decreased the slope of restitution curves and delayed APD alternans, indicating that GPS may exert a protective role for ventricular arrhythmias.     

Alternans of action potential duration and amplitude in rabbits with left ventricular dysfunction following myocardial infarction

T-wave alternans may predict the occurrence of ventricular arrhythmias in patients with left ventricular dysfunction and experimental work has linked discordant repolarization alternans to the induction of re-entry. The aim of this study was to examine the occurrence of transmural repolarization alternans and to investigate the link between alternans and ventricular arrhythmia in rabbits with left ventricular dysfunction following myocardial infarction. Optical mapping was used to record action potentials from the transmural surface of left ventricular wedge preparations from normal and post-infarction hearts during a progressive reduction in pacing cycle length at 30 and 37°C. Data were analyzed using custom software, including spectral analysis. There were no significant differences in baseline transmural electrophysiology between the groups. Post-infarction hearts had a lower threshold for both repolarization alternans (286 vs. 333 bpm, p<0.05) and ventricular arrhythmias (79 vs. 19%, p<0.01) during rapid pacing, which was not accounted for by increased transmural discordant alternans. In VF-prone hearts, alternans in optical action potential amplitude was observed and increased until 2:1 block occurred. The degree of optical action potential amplitude alternans (12.0 ± 7.0 vs. 1.8 ± 0.3, p<0.05), but not APD(90) alternans (1.4 ± 0.6 vs. 1.1 ± 0.1, p>0.05) was associated with VF inducibility during rapid pacing. Post-infarction hearts are more vulnerable to transmural alternans and ventricular arrhythmias at rapid rates. Alternans in optical action potential amplitude was associated with conduction block and VF. The data suggest that changes in optical action potential amplitude may underlie a mechanism for alternans-associated ventricular arrhythmia in left ventricular dysfunction.     

Dynamics and cardiac arrhythmias

Cardiac arrhythmias, characterized by single or multiple reentrant circuits, represent a dynamic phenomenon in an excitable medium. In this review, we provide a brief overview of how cardiac action potential duration restitution, conduction velocity restitution, and intracellular calcium cycling regulate the dynamics of action potential excitation and wave propagation in relation to the genesis and maintenance of cardiac arrhythmias.     

Effect of 60 Minutes of Rapid Atrial Pacing on Atrial Action Potential Duration in the In-Situ Canine Heart

Right atrial monophasic action potentials were recorded before and after 60 minutes of rapid atrial pacing (pacing cycle length (CL); 127 ± 10 ms) in 12 closed-chest dogs. The right atrial (RA) monophasic action potential (MAP) duration at 90% repolarization (RAMAPD) was measured at CLs of 400 ms and 250 ms. CL-dependent changes in RAMAPD (CL 400 ms – 250 ms) before and after rapid atrial pacing were 24 ± 1 ms and 16 ± 5 ms, respectively (p < 0.02). RAMAP was recorded at each atrial pacing CL starting at 240 ms decreasing by 10-ms increments. RAMAPD alternans was observed in 10 of 12 dogs at a CL of 163 ± 17 ms before and in 10 of 12 dogs at s CL of 198 ± 29 ms (p < 0.01) after rapid atrial pacing. Sustained atrial fibrillation (AF) (>5 minutes) was induced in 1 of 12 dogs at a pacing CL of 130 ms before rapid atrial pacing and in 4 of 12 dogs at a pacing CL of 135 ± 17 ms after rapid atrial pacing. Onset of AF was always preceded by the RAMAPD alternans. Sixty minutes of rapid atrial pacing leads to diminution of rate adaptation of atrial action potential duration (APD) and appearance of APD alternans of greater magnitude at longer CL, both of which may contribute to the initiation and perpetuation of AF during its early phase.     

Action potential duration restitution and alternans in rabbit ventricular myocytes: the key role of intracellular calcium cycling

Action potential duration (APD) restitution properties and repolarization alternans are thought to be important arrhythmogenic factors. We investigated the role of intracellular calcium (Ca2+i) cycling in regulating APD restitution slope and repolarization (APD) alternans in patch-clamped rabbit ventricular myocytes at 34 to 36 degrees C, using the perforated or ruptured patch clamp techniques with Fura-2-AM to record Ca2+i. When APD restitution was measured by either the standard extrastimulus (S1S2) method or the dynamic rapid pacing method, the maximum APD restitution slope exceeded 1 by both methods, but was more shallow with the dynamic method. These differences were associated with greater Ca2+i accumulation during dynamic pacing. The onset of APD alternans occurred at diastolic intervals at which the APD restitution slope was significantly <1 and was abolished by suppressing sarcoplasmic reticulum (SR) Ca2+i cycling with thapsigargin and ryanodine, or buffering the global Ca2+i transient with BAPTA-AM or BAPTA. Thapsigargin and ryanodine flattened APD restitution slope to <1 when measured by the dynamic method, but not by the S1S2 method. BAPTA-AM or BAPTA failed to flatten APD restitution slope to <1 by either method. In conclusion, APD alternans requires intact Ca2+i cycling and is not reliably predicted by APD restitution slope when Ca2+i cycling is suppressed. Ca2+i cycling may contribute to differences between APD restitution curves measured by S1S2 versus dynamic pacing protocols by inducing short-term memory effects related to pacing-dependent Ca2+i accumulation.     

自主神经干预对心房恢复性质的影响

目的 研究心脏自主神经干预对心房恢复性质的影响.方法 正常成年杂种犬10只,开胸后将多极电生理导管缝置于肺静脉、左右心耳和左右心房处,应用Ag-AgCl电极记录标测部位单相动作电位,在基础状态和颈部迷走神经刺激条件下构建标测部位恢复曲线,分别对标测部位进行快速电刺激,记录心房颤动（房颤）诱发时的起搏周长和持续时间.心脏自主神经节（GP）消融后重复上述步骤.结果 GP消融前迷走神经刺激同基础状态相比显著缩短动作电位时限（APD）,降低恢复曲线最大斜率（Smax）,抑制APD电交替,但房颤容易发生（P＜0.05）.GP消融后,APD较消融前显著延长,恢复曲线Smax增大,APD电交替提前,但房颤不易诱发（P＜0.05）;GP消融后迷走神经刺激效应明显减弱.GP消融前迷走神经刺激能显著增加APD恢复曲线Smax离散度（0.5±0.2对0.3±0.1,P＜0.05）,而GP消融能显著降低APD恢复曲线Smax离散度（0.2±0.1对0.3±0.1,P＜0.05）.结论 恢复曲线的斜率并不能完全解释房颤的诱发和维持,心房APD电交替可能对房颤的诱发并无预测作用,恢复性质的离散可能是诱发房颤的重要因素.     

Adaptive diastolic interval control of cardiac action potential duration alternans

INTRODUCTION: Recent experimental and computational studies have shown that beat-to-beat alternation in action potential duration can trigger cardiac reentry, suggesting that such "alternans" is a mechanistic precursor to arrhythmias. Given such a link, termination of alternans may help prevent the onset of arrhythmias. To this end, recent efforts have shown that chaos control methods can modulate the timing of electrical stimulation to eliminate alternans. METHODS AND RESULTS: We have developed an alternative control method founded entirely in cardiac electrophysiology (rather than borrowing techniques from the control of physical systems as with existing control techniques). Using computer simulations, we show that this method, which exploits the rate-dependent behavior of cardiac tissue, can be used to control alternans (and higher-order) rhythms, and is robust to drift and noise. When applied to individual model cells exhibiting alternans, the algorithm converges to the period-1 rhythm over as wide, and in some cases a wider, range of feedback proportionality constant values relative to existing methods. Control success comparable to existing methods is achieved when the algorithm is applied to a simulated one-dimensional Purkinje fiber exhibiting alternans. CONCLUSION: We have developed a method that adaptively controls the timing of electrical stimulation to rapidly eliminate action potential duration alternans in cardiac tissue. This control method may prove valuable in future arrhythmia prevention therapies.     

Spatiotemporal transition to conduction block in canine ventricle

Interruption of periodic wave propagation by the nucleation and subsequent disintegration of spiral waves is thought to mediate the transition from normal sinus rhythm to ventricular fibrillation. This sequence of events may be precipitated by a period doubling bifurcation, manifest as a beat-to-beat alternation, or alternans, of cardiac action potential duration and conduction velocity. How alternans causes the local conduction block required for initiation of spiral wave reentry remains unclear, however. In the present study, a mechanism for conduction block was derived from experimental studies in linear strands of cardiac tissue and from computer simulations in ionic and coupled maps models of homogeneous one-dimensional fibers. In both the experiments and the computer models, rapid periodic pacing induced marked spatiotemporal heterogeneity of cellular electrical properties, culminating in paroxysmal conduction block. These behaviors resulted from a nonuniform distribution of action potential duration alternans, secondary to alternans of conduction velocity. This link between period doubling bifurcations of cellular electrical properties and conduction block may provide a generic mechanism for the onset of tachycardia and fibrillation.     

Electrical and mechanical restitution of the human heart at different rates of stimulation

Action potential duration and contractility are reduced following premature excitations, and gradually increase as the stimulus interval is lengthened. To examine these phenomena of electrical and mechanical restitution in the human heart, we simultaneously measured action potential duration and the maximum rate of left ventricular pressure in five patients undergoing electrophysiological study. Test beats were introduced at varying intervals after the last of a series of steady state intervals. By plotting action potential duration and maximum rate of left ventricular pressure as a function of the test interval, we formed electrical and mechanical restitution curves. When the rate of steady state pacing was increased, there was a decrease in action potential duration and an increase in the maximum rate of left ventricular pressure for all test intervals; i.e., a change in pacing rate affected action potential duration and maximum rate of left ventricular pressure of test responses in a reciprocal fashion. In addition, a higher steady state pacing rate allowed action potentials and contractile responses to be elicited at shorter test intervals, thereby displacing the electrical and mechanical restitution curves to the left. The magnitude of the leftward shift of both curves corresponded closely to the shortening of the steady state action potential duration induced by the increase in pacing rate. These findings confirm for the human heart that both electrical and mechanical restitution occur after membrane repolarization, i.e., as a function of the electrical diastolic interval preceding a beat, and not the stimulus interval.     

Mechanism linking T-wave alternans to the genesis of cardiac fibrillation

BACKGROUND: Although T-wave alternans has been closely associated with vulnerability to ventricular arrhythmias, the cellular processes underlying T-wave alternans and their role, if any, in the mechanism of reentry remain unclear. METHODS AND RESULTS:-T-wave alternans on the surface ECG was elicited in 8 Langendorff-perfused guinea pig hearts during fixed-rate pacing while action potentials were recorded simultaneously from 128 epicardial sites with voltage-sensitive dyes. Alternans of the repolarization phase of the action potential was observed above a critical threshold heart rate (HR) (209+/-46 bpm) that was significantly lower (by 57+/-36 bpm) than the HR threshold for alternation of action potential depolarization. The magnitude (range, 2.7 to 47.0 mV) and HR threshold (range, 171 to 272 bpm) of repolarization alternans varied substantially between cells across the epicardial surface. T-wave alternans on the surface ECG was explained primarily by beat-to-beat alternation in the time course of cellular repolarization. Above a critical HR, membrane repolarization alternated with the opposite phase between neighboring cells (ie, discordant alternans), creating large spatial gradients of repolarization. In the presence of discordant alternans, a small acceleration of pacing cycle length produced a characteristic sequence of events: (1) unidirectional block of an impulse propagating against steep gradients of repolarization, (2) reentrant propagation, and (3) the initiation of ventricular fibrillation. CONCLUSIONS: Repolarization alternans at the level of the single cell accounts for T-wave alternans on the surface ECG. Discordant alternans produces spatial gradients of repolarization of sufficient magnitude to cause unidirectional block and reentrant ventricular fibrillation. These data establish a mechanism linking T-wave alternans of the ECG to the pathogenesis of sudden cardiac death.