Factors affecting epicardial dispersion of repolarization: a mapping study in the isolated porcine heart

Abstract

Objective: Non-uniform drug-induced prolongation of repolarization predominating in the midmyocardial (M) cell layers has been shown to be responsible for perpetuation of reentry, giving rise to torsade de pointes. However, the absence of M cells in immature animals, especially the pig, suggests other possible underlying mechanisms. We sought to examine, in this species, the effects of predisposing factors to torsade de pointes on the dispersion of epicardial repolarization and their contribution to arrhythmogenesis. Methods: Computerized mapping of repolarization and activation was conducted on the epicardial surface in 29 Langendorff-perfused hearts of eight-week-old pigs. Activation–recovery intervals were measured simultaneously from 128 unipolar electrograms. Results: Baseline iso-interval maps were dipolar (41%) or multipolar (59%). Dispersion of repolarization was reverse frequency-dependent but was unaffected by lowering [K⁺]_o. dl-Sotalol (0.1 mmol/l) reinforced local gradients and thus increased epicardial dispersion, whereas intramural recordings did not demonstrate any predominant effect in midmyocardial layers. Phenylephrine (1 μmol/l) notably augmented dl-sotalol effects. After [Mg⁺⁺]_o lowering, although dispersion was not significantly increased, dl-sotalol was associated with the spontaneous occurrence of polymorphic ventricular tachycardia in seven out of nine experiments. When maps of repolarization of escape beats were compared with activation maps of first arrhythmic beats, an arc of functional dissociation was observed in the vicinity of a steep gradient of repolarization in two out of nine tachycardias. Conclusion: Epicardial dispersion of repolarization is increased by slow rates, dl-sotalol and phenylephrine but is not the only requirement for initiation of polymorphic ventricular tachycardia. In combination with other factors, it helps continuation of the arrhythmia in this model.

Arrhythmia mechanisms, Antiarrhythmia agents, Intra/extracellular ions, Mapping, QT-dispersion

Time for primary review 27 days.

1 Introduction

Non-uniform recovery of excitability as a consequence of spatial repolarization heterogeneity was first proposed to set the stage for ventricular arrhythmias by Han and Moe in 1964 [1]. Later, it was shown that dispersion of action potential durations was a possible underlying mechanism for circus movement reentry around refractory barriers [2, 3]. In patients with either the congenital or acquired long QT syndrome (LQTS), the mechanisms that underlie episodes of torsade de pointes are still not well understood. Indirect in vivo and in vitro evidence led to the formulation of two different, sometimes competing, hypotheses as mechanisms for the onset of torsade de pointes. The first one is the early afterdepolarization (EAD) hypothesis, supported by in vitro studies that showed the development of EADs and triggered activity either from Purkinje fibers [4]or from midmyocardial (M) cells [5, 6]in the presence of predisposing factors such as slow driving rates, hypokalemia, hypomagnesemia and drugs that prolong action potential durations. The second one is the dispersion of repolarization hypothesis [7]that has been recently confirmed in a canine model of acquired torsade de pointes using the neurotoxin anthopleurin A [8]. In this model, a marked dispersion of repolarization was demonstrated across the left ventricular wall due to the differential response of M cells to the toxin and to bradycardia. Whereas the initial beat of tachycardia arose as focal activity, reentry was due to infringement of this focal activity on the spatial dispersion of repolarization, resulting in conduction block and recirculating wave fronts.

The M cell is distinguished by the ability of its action potential to prolong disproportionately to other myocardial ventricular cells in response to a slowing of rate or to agents that delay repolarization. Interestingly, Rodrıéguez-Sinovas et al. [9]showed that porcine left ventricular myocardium, unlike canine myocardium, lacks M cells in animals of one to two months of age. This difference has been explained by developmental studies indicating that, in the dog, the M cell is not distinct until approximately three months of age [10], possibly due in part to a gradual increase in the density of the transient outward current, I_to[11]. This maturation is also operative in humans [12]and suggests that M cells are not a critical component in the genesis of torsades de pointes that may be observed in neonates [13]. These considerations, as well as data suggesting that repolarization dispersion computed from body surface mapping is an important risk factor for torsade de pointes in LQTS [14], reinforce interest in studying the factors that control repolarization gradients at the epicardial level.

Thus, our purpose was twofold: (i) systematic investigation of the respective quantitative effects of classical predisposing factors to torsade de pointes (i.e. slow rates, hypokalemia, hypomagnesemia, dl-sotalol and α₁-receptor stimulation [15]) on the occurrence of non-uniform dispersion of repolarization and (ii) identification of the optimal combination of these factors responsible for the occurrence of bradycardia-dependent polymorphic ventricular tachycardia in immature porcine myocardium.

2 Methods

All procedures for animal care and experimentation followed the NIH guidelines of the Guide for the Care and Use of Laboratory Animals, and were in accordance with the French legislation and approved by an institutional committee.

2.1 Isolated perfused porcine heart preparations

In order to minimize the role of the autonomic nervous system, to achieve very slow pacing rates without hemodynamic compromise and to control extracellular ionic concentrations, non-working hearts of 29 female healthy pigs (eight-week-old, 20–25 kg) were connected to a Langendorff recirculating perfusion system using a modified Tyrode’s solution. The whole procedure has previously been reported in detail [16]. The overall duration of normothermic ischemia comprised between 2.5 and 4.0 min in all cases. After stabilization, a perfusion pressure of 3.9 kPa (30 mmHg) resulted in a perfusion flow of 150±9.5 ml min⁻¹. The temperature of the perfusate was 36.4±0.3°C, and 36.0±0.3°C at the epicardial surface. The solution was equilibrated with a mixture of 95% O₂–5% CO₂, resulting in a pH of 7.43±0.02.

A multiple-terminal electrode was brought into contact with the epicardial surface of the anterolateral wall of the left ventricle. The top border of the plaque electrode was positioned along the nearby segment of the left anterior descending coronary artery. Bipolar pacing electrodes were attached to epicardial tissue via superficial sutures located between the left anterior descending artery and the top border of the plaque electrode. The atrioventricular nodal area was crushed to give rise to slow idioventricular rhythms, thus allowing ventricular pacing at low rates. In some experiments, a fragment of epicardial myocardium was dissected for the purpose of recording action potentials.

2.2 Data acquisition and isochronal mapping

The lead array consisted of 128 evenly spaced unipolar stainless steel electrodes (0.4 mm diameter) mounted on a plaque (interelectrode distance of 2.5 mm). The resolution of the electrode array was chosen in accordance with theoretical and experimental considerations [17, 18], estimating that circus movement due to repolarization nonuniformities requires an area greater than 30–50 mm². The mapping system has been described elsewhere [16, 19]. Briefly, amplified and filtered (0.05–200 Hz) unipolar signals were sampled at 500 Hz and converted to a 10-bit digital format. Data were analyzed in a 1.0-s time window selected from the stored material. The computer automatically detected the local activation time on each electrogram, at the point of most rapid potential decrease (dV/dt_min) and its value was determined according to the three point Lagrange derivative by computing the potential drop between the sampling points preceding and just following the activation point, divided by the time interval (i.e. 4 ms). The threshold for local activation [20]was set at −0.5 V s⁻¹. When this criterion was not fulfilled, it was considered that excitation did not occur at the corresponding recording site, as a result of either local inexcitability or conduction block. The activation times were validated by inspection on a videoscreen and the zero activation time was defined at the site of earliest activation. Isochronal maps depicting the activation sequence of each individual beat in the recording area were computed automatically by linear interpolation and drawn at 10 ms intervals. If a sudden increase in the density of isochrones, indicating local conduction block, as well as electrograms with elevation of the activation–recovery segment, suggesting injury, were observed, the experiment was discarded.

In some experiments, a plunge needle electrode containing five unipolar contacts, each separated by 1.25 mm, was placed in the left ventricular free wall. The most superficial electrode was located approximately 0.5 mm from the epicardial surface. A finger was introduced into the left ventricle through the mitral orifice to make sure of the endocardial location of the electrode tip. Five channels of the mapping system were alternately connected to the plunge electrode or to the plaque electrode in order to record electrograms.

2.3 Data measurements and expression

2.3.1 Activation–recovery intervals (ARIs)

ARI was chosen as an index of local repolarization duration [21, 22]. It was defined as the interval between the time of the minimum first derivative, dV/dt_min, in the activation complex (RS) to the maximum positive slope, dV/dt_max, in the T wave of each of the 128 unipolar electrograms. Computer-generated iso-ARI areas were derived from ARI values and drawn at 5–20 ms intervals. Although previous in vivo studies have shown that ARIs derived from unipolar electrograms approximate the local effective refractory period (ERP) [8, 21], we decided to analyze the relationship of ARI to refractoriness during control and in the presence of dl-sotalol in order to validate the technique of ARI mapping in this ex vivo model.

2.3.2 ERP determination

The ERP and corresponding ARI were measured at 15 evenly spaced sites that varied among experiments and were selected among the 128 available leads in the plaque array. The diastolic threshold was determined during cathodal unipolar pacing at a basic driven cycle length of 1500 ms, then the current level was increased 1.5 times. The diastolic threshold was always <0.9 mA, rendering the virtual cathodal effects negligible [23]. Premature stimuli were introduced at increasing intervals (10 ms), beginning with a short coupling interval, after a train of eight basic driven beats. Once there was successful propagation of the premature stimulus on the four local electrograms surrounding the pacing site, its coupling interval was decreased by 20 ms and then again introduced at increasing intervals (5 ms) after 50 basic driven beats until propagation occurred. The ERP of the surrounding sites was defined as the longest coupling interval that failed to evoke a locally propagated response.

2.3.3 Definitions

Mean ARI (mARI) was the arithmetic mean of the 128 ARI values obtained within the same beat, the dispersion index (DI) was expressed as its standard deviation. The maximal gradient on a given ARI distribution was defined as the maximal difference observable between ARIs computed from two adjacent electrodes divided by the interelectrode distance.

2.4 Correlations between in vivo and ex vivo repolarization patterns

Some experiments were carried out to determine if heart insulation did not alter repolarization patterns. After sternotomy, the heart was handled in a pericardial cradle; the epicardial temperature was maintained constant by the use of heat lamps. The epicardial plaque and the pacing electrodes were sutured at the above-mentioned location, then the atrioventricular node was destroyed by direct injection of 37% formaldehyde (0.1–1.0 ml). Data acquisition was performed during epicardial pacing. The heart was then removed and prepared for isolated perfusion as described above.

2.5 Experimental protocol

Following reperfusion and 30 min stabilization, a pacing protocol, consisting of 100 consecutive cycles at different cycle lengths (from 1500 to 500 ms), was applied under control conditions. Then, the preparation was exposed to one of the following conditions: (i) lowering of the extracellular potassium concentration ([K⁺]_o) by slowly adding Tyrode’s solution containing 1.0 mmol/l KCl over 15 min, (ii) consecutive infusion of dl-sotalol (0.1 mmol/l) [24]and phenylephrine (1.0 μmol/l) in a cumulative manner, (iii) infusion of dl-sotalol in the presence of an extracellular magnesium concentration ([Mg⁺⁺]_o) lowered to 0.6 mmol/l. The preparation was allowed to further stabilize for an additional 20 min after each intervention. Under control conditions as well as in the presence of drugs or ionic modification, the stimulation was stopped for 10 min after measurements under pacing were done, to facilitate the possible development of arrhythmias.

2.6 Isolated porcine ventricular muscle preparation and action potential recordings

In order to validate some of the results obtained in the isolated heart preparation, action potentials were recorded from fragments of epicardial muscle using a standard glass microelectrode technique. The protocol has been detailed elsewhere [25]. Briefly, strips of epicardial muscle were dissected, pinned on the bottom of a tissue bath and superfused continuously with oxygenated Tyrode’s solution. The following parameters were measured: resting membrane potential, action potential amplitude, overshoot amplitude, maximal rate of depolarization and action potential duration at 30, 50 and 90% of repolarization (APD₃₀, APD₅₀ and APD₉₀, respectively). Preparations were driven by successive trains of stimuli at varying cycle lengths (2500, 1666, 1250 and 1000 ms) with [K⁺]_o=4.0 mmol/l, then with [K⁺]_o=2.7 mmol/l in the same experiments.

2.7 Statistical analysis

Group values are presented as mean±SEM. Measurements performed before and after interventions were compared by ANOVA followed by Tukey’s post test or by paired Student’s t-test, according to the case. Frequency-dependent effects were assumed by ANOVA. Reproducibility analysis was conducted, comparing 128-paired values of ARIs by linear regression analysis. Comparisons were two-tailed and results were considered to be statistically significant at p<0.05. Analyses were made with INSTAT software (GraphPad, San Diego, CA, USA).

3 Results

3.1 Validation of the experimental protocol

ARIs measured at a pacing cycle length of 1500 ms before and after dl-sotalol infusion from 44 data points pooled from three different experiments and ranging from 320 to 495 ms were compared to corresponding ERPs. The correlation between the two parameters was highly significant [(ERP=0.93·ARI+31.4) r=0.98, p<0.001]. Immediate beat-to-beat reproducibility of the ARI patterns was examined in five control hearts at different pacing cycle lengths and was found to be satisfactory, with correlation coefficients ranging from 0.77 to 0.98 (mean, 0.94) with p<0.001 in all cases.

In three other experiments, ARIs were mapped three times at 40 min intervals at different pacing cycle lengths. The pooled data concerning the mean ARI and dispersion values are summarized in Table 1. Comparisons between the first and second series of maps gave correlation coefficients that were between 0.55 and 0.94 (mean, 0.77) with p<0.01 in all comparisons. Comparisons between the third series of maps and the first one gave correlation coefficients ranging from 0.51 to 0.85 (mean, 0.70), p<0.01. These data showed a trend towards a slight decrease in dispersion and mean ARI values over time that did not exceed 5 and 20 ms, respectively, and did not reach the level of statistical significance. In the following, measurements made during the first series will be designated as control.

Table 1

Mean ARI and dispersion values computed in triplicate at 40 min intervals

Time (min)	500 ms		1000 ms		1333 ms		3 s pause
	mARI	Disp	mARI	Disp	mARI	Disp	mARI	Disp
40	226 (13)	15 (5)	313 (15)	17 (5)	327 (11)	18 (6)	280 (6)	15 (6)
80	213 (11)	12 (5)	293 (5)	14 (5)	314 (12)	17 (3)	277 (12)	15 (4)
120	215 (10)	10 (2)	293 (3)	12 (2)	309 (6)	15 (1)	274 (8)	14 (1)

Values are given as means (SEM), expressed in ms, measured at three pacing cycle lengths: 500, 1000 and 1333 ms and after a 3000-ms pause, n=3.

Abbreviations: Disp, dispersion of ARI values, mARI, mean activation–recovery interval.

3.2 Baseline epicardial dispersion of repolarization (n=17)

Repolarization patterns were well organized, as evidenced by ARI mapping. They were grossly dipolar in seven hearts (41%) and multipolar in the remaining ten (59%). Fig. 1 depicts a typical pattern with three extrema (two minima and one maximum) during pacing at different cycle lengths. In panel D, during steady-state pacing at 1500 ms, differences of up to 42 ms were observed between the two margins of the plaque and a steep gradient (9.6 s m⁻¹) was visible in the vicinity of points a and b. In three hearts, acquisitions were also made in vivo at the same pacing cycle length of 1000 ms. The comparison of maps revealed almost identical patterns and similar dispersion values.

Fig. 1

Representative example of reverse frequency dependence of ARI gradients. An epicardial plaque electrode was used to record 128 unipolar electrograms in the anterior left ventricular region during stimulation through pairs of electrodes located between the plaque and the origin of the left anterior descending coronary artery (LAD). The upper side of the plaque was positioned parallel to the nearby segment of the LAD. Panels A to D depict the influence of increasing pacing interval on the ARI pattern, at 500, 750, 1000 and 1500 ms, respectively. Shaded areas corresponding to ARI values are drawn at a constant 15 ms interval. Tracings below each map are electrograms at selected sites a to e. At 1500 ms pacing, a more pronounced lengthening of repolarization in areas where ARIs were long at 500 ms results in the reinforcement of the gradient between sites a and b from 4.8 to 9.6 s m⁻¹.

Fig. 2 shows six unipolar epicardial electrograms that were recorded along the apex–base axis (top to bottom) at three different pacing cycle lengths. During rapid pacing at 500 ms, T waves were either negative, positive or followed a biphasic negative–positive pattern. During intermediate pacing at 750 ms, notched T waves became visible. At a slow pacing cycle length of 1500 ms, split T waves were recorded. In the two upper electrograms, the late component was dominant (ARI=358 ms), whereas in the two lower electrograms, it was the early component (ARI=340 or 342 ms), middle tracings were ambiguous, with superimposition of both components. This was observed in more than three electrograms in 4/17 experiments. These electrograms were identified at slow driving rates and normalized during pacing at 500 ms cycle lengths in all cases.

Fig. 2

Unipolar electrograms and calculated ARIs from six unipolar sites, 2.5 mm apart along the left anterior epicardial apex–base axis (top to bottom). During pacing at 500 ms, T waves were either negative, positive or followed a biphasic negative–positive pattern. During pacing at 750 ms, notched T waves became visible. At a slow pacing cycle length of 1500 ms, split T waves were recorded. Enlarged repolarization signals are represented in the last column: in the two upper electrograms, a late component was dominant, whereas in the two lower electrograms, it was the early one; middle tracings show superimposition of both components.

3.3 Influence of stimulation cycle length (n=17)

A representative example of the change in pacing cycle length on ARI patterns is given in Fig. 1. Increasing the cycle length from 500 to 1500 ms resulted in a non-uniform lengthening of repolarization intervals. Thus, dispersion gradually increased from 5.5 to 9.3 ms as a consequence of a more pronounced lengthening of repolarization in areas where ARIs were long rather than in areas where they were short at a 500-ms cycle length. This led to reinforcement of the maximal ARI gradient from 4.8 to 9.6 s m⁻¹.

The ARI lengthening effect, which developed with increases in the pacing cycle length, peaked at 1333 ms; at longer cycle lengths, the repolarization duration barely increased. The mARI ranged from 254±6 ms at 500 ms to 346±10 ms at 1333 ms, and the reverse frequency-dependence was statistically significant (ANOVA p=0.0001). The same observations were drawn from the curve depicting the effect on repolarization dispersion; the maximal rise occurred at 1333 ms. Dispersion ranged from 8.7±0.8 ms at 500 ms to 12.6±1.4 ms at 1333 ms, again the reverse frequency-dependence was statistically significant (ANOVA p=0.0004). The 3000 ms pause introduced after a train at 1000 ms induced a striking shortening in the mARI (from 335±9 to 291±8 ms, p<0.0001) as well as a decrease in repolarization dispersion (from 11.2±1.1 to 9.1±1.1 ms; p<0.05). Overall, these results suggested that, at least in this model, repolarization duration reached a ceiling or even shortened at longer cycle lengths. This prompted us to examine the frequency dependence of APD in epicardial preparations. This analysis, summarized in Table 2, part A, actually confirmed a bell-shaped profile of the relationship between APD₃₀, APD₅₀ or APD₉₀ and pacing cycle length.

Table 2

Influence of pacing cycle length on action potential duration recorded from epicardial muscle preparations with [K⁺]_o=4 mmol/l (part A) and [K⁺]_o=2.7 mmol/l (part B)

CL (ms)	1000	1250	1666	2500
A
APD30 (ms)	160 (10)	164 (10)	166 (11)	164 (11)
APD50 (ms)	208 (11)	213 (12)	213 (13)	214 (12)
APD90 (ms)	246 (12)	250 (12)	253 (13)	252 (13.0)
B
APD30 (ms)	144 (10)b	147 (10)b	150 (10)c	154 (10)a
APD50 (ms)	194 (11)b	197 (12)c	201 (11)b	205 (11)a
APD90 (ms)	241 (12)	243 (12)a	246 (12)a	249 (12)

Abbreviations: APD, action potential duration at 30, 50 and 90% of repolarization (APD₃₀, APD₅₀ and APD₉₀, respectively); CL, pacing cycle length.

Values are given as means (SEM) expressed in ms. n=7.

^ap<0.05, ^bp<0.01, ^cp<0.001 vs. [K⁺]_o=4 mmol/l.

3.4 Influence of lowered extracellular potassium concentration (n=7)

In this series of hearts, [K⁺]_o was lowered from 3.62±0.06 to 2.75±0.06 mmol/l and remained stable throughout the experiment, measured at 2.92±0.09 mmol/l 2 h later. As represented in Fig. 3, decreasing [K⁺]_o shortened repolarization duration at every pacing cycle length by shifting the frequency dependence curve towards lower values, however, the effect on repolarization dispersion was not significant. The magnitude of the shortening of the mARI was almost constant at every cycle length, ranging from 7.8 to 11.1%. This effect was confirmed by action potential recordings and was predominant at APD₃₀ and APD₅₀ levels (Fig. 4 and Table 2).

Fig. 4

Epicardial action potential recordings performed at [K⁺]_o=4 mmol/l (dotted lines, designated as control) and then at [K⁺]_o=2.7 mmol/l in the same preparation. Pacing cycle lengths are 2500 ms (A), 1666 ms (B) and 1250 ms (C). Decreasing [K⁺]_o results in a shortening effect, which is predominant at the plateau level.

Fig. 3

Mean ARI (mARI) and dispersion (DI) plotted as a function of pacing cycle length (CL) from data collected on the 128 unipolar epicardial electrograms with [K⁺]_o=3.62 mmol/l (black squares), and then with [K⁺]_o=2.75 mmol/l (white squares) in the same experiments. Data represent mean values and vertical lines indicate SEM from seven porcines. ^ap<0.05 vs. [K⁺]_o=3.62 mmol/l. There was no significant effect on dispersion after [K⁺]_o lowering.

3.5 Effects of dl-sotalol and phenylephrine (n=7)

dl-Sotalol increased both repolarization duration and dispersion, shifting the frequency dependence curves towards higher values (Fig. 5 and Fig. 6). The effect on the mean ARI was statistically significant at every pacing cycle length but was more pronounced at long pacing cycle lengths. The effect on repolarization dispersion reached the level of statistical significance at 500 and 1333 ms only. When phenylephrine was added, the lengthening effect of dl-sotalol on the duration of repolarization was reinforced at 1333–1500 ms, unchanged at 750–1000 ms and even reversed at a pacing cycle length of 500 ms. On the other hand, dispersion notably increased at every pacing cycle length, approximately 1.5 to 2-times compared to baseline values (Figs. 5 and 6). Again, this effect was stronger at long pacing cycle lengths.

Fig. 6

Mean ARI (mARI) and dispersion (DI) plotted as a function of pacing cycle length (CL) at baseline (squares) in the presence of dl-sotalol (0.1 mmol/l; circles) and then in the presence of dl-sotalol (0.1 mmol/l) in combination with phenylephrine (1 μmol/l; triangles). Data represent mean values and vertical lines indicate SEM from seven porcines. dl-Sotalol increases both repolarization duration and dispersion, shifting the baseline curve towards higher values. When phenylephrine was added, the lengthening effect was reinforced at 1333–1500 ms, unchanged at 750–1000 ms and even reversed at a pacing cycle length of 500 ms. Dispersion increases at every pacing cycle length by approximately 1.5 to 2-times compared to baseline. ^ap<0.05 dl-sotalol vs. baseline, ^bp<0.05 dl-sotalol+phenylephrine vs. baseline.

Fig. 5

Representative example of dl-sotalol and phenylephrine effects on ARI gradients. Iso-ARI patterns are recorded at baseline (A) after dl-sotalol (0.1 mmol/l; B) then after dl-sotalol (0.1 mmol/l) in combination with phenylephrine (1 μmol/l) at a constant pacing cycle length of 1333 ms (C). Shaded areas corresponding to ARI values are drawn at a constant 25 ms interval. Tracings beside each map are electrograms at selected sites a to e. dl-Sotalol exerts a non-uniform lengthening over ARIs leading to a reinforcement of the gradient between sites a and b from 8.8 to 12.0 s m⁻¹. Overall, dispersion increases from 10.2 to 12.9 ms. After the addition of phenylephrine, the maximal gradient reaches 15.3 s m⁻¹ and dispersion further augments to 18.3 ms. Note that, compared to panel B, phenylephrine prolongs repolarization in areas where ARIs were long and slightly shortens it in areas where they were short (notably the d–e zone).

The effects of dl-sotalol on intramural repolarization were also examined in six hearts, especially after the 3000 ms pause, in order to ensure the absence of any predominant effect of the drug in deep subepicardial to midmyocardial layers. Indeed, these measurements, detailed in Table 3, were unable to demonstrate such regional differences.

Table 3

Distribution of ARIs across the porcine left ventricular free wall in the presence of dl-sotalol (0.1 mmol/l)

Number	EPI	1.25 mm	2.5 mm	3.75 mm	5 mm
1	364	358	356	358	348
2	300	300	310	310	290
3	414	394	400	384	387
4	394	394	392	394	398
5	424	436	432	440	442
6	282	286	284	290	286
mean (SEM)	363 (24)	361 (24)	362 (23)	362 (23)	358 (25)

The values of ARI are expressed in ms as a function of the distance of the recording site from the epicardial surface (EPI), and were measured after an abrupt transition from a cycle length of 1000 ms to a cycle length of 3000 ms, n=6, ANOVA: p=ns.

The effects of dl-sotalol were reassessed in nine hearts exposed to a lowered [Mg⁺⁺]_o and compared to the above-mentioned data, which was obtained with a normal [Mg⁺⁺]_o of 1.2 mmol/l. This comparison did not disclose any significant modification of the drug’s effects on repolarization duration or dispersion. At a pacing cycle length of 1333 ms, where the effects of dl-sotalol were predominant, mARI and dispersion were 356±13 and 18.7±5.3 ms with [Mg⁺⁺]_o=0.6 mmol/l compared to 379±11 and 17.2±3.5 with [Mg⁺⁺]_o=1.2 mmol/l. Moreover, lowering [Mg⁺⁺]_o did not result in any significant increase in repolarization dispersion by itself (i.e. in the absence of dl-sotalol).

3.6 Arrhythmogenesis

All hearts developed a slow idioventricular rhythm in the absence of pacing, with an escape cycle ranging from 2000 to 4000 ms. At times, these rhythms were interrupted by pauses that could reach a duration of several tens of seconds. Arrhythmias did not occur under baseline conditions, neither in the presence of a lowered [K⁺]_o, nor in the presence of dl-sotalol (alone or in combination with phenylephrine), despite the above-mentioned increase in repolarization dispersion. On the other hand, nine polymorphic ventricular tachycardia episodes initiated by a ventricular pause occurred spontaneously in seven out of the nine hearts exposed to dl-sotalol and a lowered [Mg⁺]_o. A temporal relationship appeared between the mARI value of the last escape beat and the coupling interval of the first arrhythmic beat: the longer the mARI, the longer the coupling interval, r=0.98, p<0.0001. Four episodes terminated spontaneously (duration three–eight beats) the five others degenerated into ventricular fibrillation. Runs of self-terminating tachycardia displayed longer cycle lengths (303–431 ms) than those that degenerated into fibrillation (188–286 ms) (p<0.05).

Interestingly, during two tachycardias, an arc of functional dissociation was observed in the propagation of the first ventricular beat in the vicinity of a steep gradient of repolarization intervals (>10 s m⁻¹). An example of such a relationship between the iso-ARI map of the penultimate escape beat and the isochronal activation of the first ventricular beat is given in Fig. 7. From the site of earliest epicardial activation, the impulse was conducted in the large region where repolarization intervals were shorter, circumventing a line of dissociation that corresponded to the gradient between low and high ARI values.

Fig. 7

Panel A: Occurrence of a sustained episode of polymorphic ventricular tachycardia. Panel B: Iso-ARI map of the penultimate idioventricular beat indicated by an asterisk in panel A. Panel C: Isochronal activation of the first ventricular beat of the arrhythmia (number 1 in panel A). In panel B, the shaded areas depicting the ARI pattern are drawn at 15 ms intervals, the electrograms at the bottom of the map illustrate the calculated ARIs at sites a to f. In panel C, isochrones are drawn at 5 ms intervals, the dotted area indicates functional conduction block, the electrograms at the bottom of the map illustrate activation from site g to site k and from site n to site k, where an electrotonic potential is visible. Vertical lines and numbers beside these electrograms refer to activation times (AT). Note that the functional conduction block in activation map C corresponds to the steep gradient of ARIs of the repolarization map (B).

4 Discussion

Repolarization dispersion is an intrinsic feature of left ventricular epicardium that can be quantitatively assessed. This dispersion is present in the centimetric range and is organized, leading to the observation of local gradients that increase in some of the conditions predisposing to the occurrence of torsade de pointes, i.e. slow driving rates, exposure to dl-sotalol and α-receptor stimulation by phenylephrine. On the other hand, dispersion is not altered by decreasing extracellular potassium or magnesium concentrations. An exaggerated dispersion is not sufficient by itself to provoke the occurrence of this impulse (i.e. the first arrhythmic beat). Accordingly, a lowered extracellular magnesium concentration appears to be such an additional and independent factor that promotes the arrhythmia. We found no differences in ARI durations registered from different myocardial layers in the presence of dl-sotalol, in agreement with the hypothesis that the occurrence of bradycardia-dependent polymorphic ventricular tachycardia does not rely on intramural dispersion of repolarization in immature porcines.

4.1 Baseline repolarization dispersion in the isolated porcine heart

The cellular mechanisms underlying repolarization dispersion are unknown. The preservation of repolarization patterns during rate and pharmacological interventions strongly suggests the existence of regional electrophysiological differences within the epicardial tissue. Such differences might affect some of the currents involved in the time course of repolarization. For example, a varying localization of ERG, the channel protein responsible for the rapid component of the delayed rectifier current (I_Kr), has recently been described in ferret epicardium [26]. Additionally, postnatal development of a variety of ionic currents has been reported in different species. It could be suspected that such developmental changes involving inward [27]or outward currents [11, 12]are present in the pig and affect the epicardial tissue in a non-uniform fashion.

The observation that a 3000-ms pause intercalated during stable pacing at 1000 ms intervals strikingly reduces both repolarization duration and dispersion is probably not transposable to all species. The possibility remains notably in humans that the duration of the epicardial action potential continues to lengthen at very long cycle lengths or after a pause.

4.2 Effects of lowered extracellular potassium

Our demonstration that rapid lowering of [K⁺]_o does not increase dispersion and even slightly shortens both epicardial ARI and APD conflicts with studies conducted in M cells or Purkinje fibers [4, 28]. On the other hand, Gettes and Surawicz [29]have shown that low [K⁺]_o (0.6 to 0.8 mmol/l) prolongs the plateau in Purkinje fibers but shortens it in porcine ventricular fibers. In both types of cells, the duration of phase three of the action potential is prolonged. In ventricular fibers, the net change in action potential duration is determined by the sum of these opposing effects. Thus, one should consider that [K⁺]_o=2.7 mmol/l is not low enough to reveal action potential prolongation and results in a predominant shortening effect at the plateau level. In addition, I_Kr, unlike most other potassium currents, decreases when [K⁺]_o is decreased, whereas I_Ks increases in parallel with the increase in the driving force generated by low [K⁺]_o[30]. A lesser contribution of I_Kr to repolarization, either species-dependent or age-dependent, might be another explanation for this shortening effect.

4.3 Drug effects (dl-sotalol and phenylephrine)

In the presence of dl-sotalol, the relationship between dispersion of repolarization and pacing cycle length is shifted towards higher values, which confirms the hypothesis that this drug-related increase in dispersion is a marker of the arrhythmogenic potential of dl-sotalol that does not depend on its well-known reverse frequency-dependent effects on repolarization duration [7]. This finding apparently conflicts with two studies, conducted with I_Kr blockers, which were aimed at quantifying their effect on dispersion [31, 32], however, in these experiments, the effects of long pacing cycle lengths were not examined. On the other hand, Zabel et al. [33]also reported rate-dependent and concentration-dependent increases in dispersion computed from six–eight monophasic action potential recordings by d-sotalol in the isolated rabbit heart [33]. Because of the presence of the l-sotalol isomer, which exerts a β-blocking effect, the assumption can be made that phenylephrine mainly acts through α stimulation. We showed potentiation of dl-sotalol effects by phenylephrine, which is consistent with the demonstration in other species that α₁ receptor stimulation results in the inhibition of potassium currents, delaying repolarization, especially I_K1 and the 4-aminopyridine-sensitive component of I_to[34, 35].

We were unable to disclose intramural gradients in the presence of dl-sotalol after a 3000-ms pause, a condition where expected differences in action potential duration between M cells and cells from other sites are magnified [5, 6]. Since we did not record action potentials from superfused transmural strips, and given the very limited number of intramural recordings, the presence of intramural gradients cannot be definitely ruled out. However, this suggests that M cells are not essential for the genesis of torsade de pointes and, therefore, reinforces the hypothesis of EADs arising from Purkinje fibers.

4.4 Arrhythmogenesis

Interestingly, lowering [Mg⁺⁺]_o in the presence of dl-sotalol is of importance for the observation of some polymorphic ventricular tachycardias but does not increase repolarization dispersion. This further supports the theory that initiating and perpetuating mechanisms of arrhythmias are different in the setting of prolonged repolarization [7, 8, 36]. The role of magnesium has been suspected in humans for a long time [37]. Lowering its extracellular concentration promotes the observation of EADs in vitro [29]. Conversely, it has been reported that MgSO₄ suppressed subthreshold EADs from monophasic action potential recordings, shortened the QT interval and prevented induction of torsade de pointes by d-sotalol in the dog [38].

Mapping was restricted to a limited area, therefore conclusions about the site of origin of the reported polymorphic ventricular tachycardias cannot be drawn from the present study. Although the majority of these tachycardias were initiated by a pause, were closely related to the T wave of the preceding escape beat and were able to terminate spontaneously, their occurrence was observed in extreme conditions, rendering their relevance questionable. However, in both experiments where conduction block was observed in the propagation of the first arrhythmic beat, the gradient value was greater than 10 s m⁻¹, a value similar to the difference associated with the induction of arrhythmia in canine experiments in which the gradient was generated by production of a difference in the temperature of adjacent areas [2, 3].

Besides the possible generation of reentry, the occurrence of conduction block in areas with a high repolarization gradient may result in nonuniform propagation. The hypothesis that nonuniform conduction during triggered activity may facilitate the shift of the origin from one site to another has been recently formulated [36]. Indeed, a focus capable of undergoing EADs may dominate during the first activation, then a second focus can gain control of the succeeding activity if it remains protected by a conduction block from the initial activation.

5 Conclusion

The data presented may contribute to an understanding of torsade de pointes, especially in immature hearts. Our model combines features of both acquired LQTS (especially the form that develops in patients exposed to dl-sotalol) and congenital LQTS. With regard to the latter, I_Kr blockade with dl-sotalol can be considered as a surrogate to the second phenotype of QT prolongation (LQT2). An exaggerated dispersion of repolarization is not sufficient to provoke the occurrence of a polymorphic tachycardia in this model but helps its perpetuation. Accordingly, additional factors are of importance in arrhythmogenesis.

Acknowledgements

This work was supported by grant 96/33/9739 from CH et U Lille.