Methadone Blockade of Inward Rectifier Potassium Current Promotes Both Early and Delayed Repolarization Arrhythmias: Mechanistic Insights From Computational Modeling.
The authors declare that all supporting data are available within the article and the online supplementary file.This is a computational modeling study, no institutional review board approval is required.Early afterdepolarizations (EADs) in isolated cardiac myocytes are often cited as the cause of tissue‐level PVCs, which then propagate through ventricular tissue and become triggers of arrhythmias. However, our computer simulations showed that although cellular‐level EADs can cause tissue‐level PVCs under the right conditions,19, 20 repolarization gradient (RG)‐induced PVCs arising from tissue‐scale dynamical instabilities might be the major mechanism of PVCs, which is supported by experimental observations.21, 22, 23, 24 These PVCs, dubbed “R‐from‐T,”25, 26 can degenerate spontaneously into reentrant arrhythmias under the right tissue conditions. Therefore, to assess the proarrhythmic effects of methadone, we looked for PVC induction in tissue as our marker for arrhythmogenesis.Based on our previous work, we found that a 1‐dimensional (1D) cable is a sufficient test for the genesis of these tissue‐level PVCs and allows fast enough computer simulations. In our simulations, the 1D cable was composed of 200 cells (3 cm), the first half endocardium and the second half epicardium. We used 4 AP models: the 1994 Luo and Rudy (LRd) guinea pig model,27 the 2004 ten Tusscher et al. (TP04) human model,28 the O'Hara et al. (ORd) human model,29 and the modified ORd model by Tomek et al..30 The endocardial versions of the human models were used. For simplicity, we altered the slow component of the delayed rectifier K+ channel (IKs) to model the endo‐epi heterogeneity. We also added transient outward potassium current (Ito) to the epicardium. A stimulus was given at t=0 to all cells to elicit a pacing beat in the cable to assess the genesis of PVCs (see Data S1 for details of the pacing protocol and numerical methods).To take into account interindividual variabilities, we generated 1D cable model populations by randomizing parameters and subjecting the model outputs to preset filters. The LRd, TP04, and ORd models were used for the model populations. (The modified ORd model by Tomek et al. behaves very similarly to the ORd model.) We multiply the maximum conductance of the major ionic currents with a factor drawn randomly from a preset interval. We carried out 1D cable simulations with the randomly selected parameter sets and measured APD at both the endocardium and epicardium. We first dropped the cases that exhibited PVCs and the cases that gave rise to too short an APD in the epicardium (due to a too strong I). After these filters, we obtained a model population with a wide raw APD distribution. We then assigned a Gaussian distribution (the QTc distribution of healthy human is close to a Gaussian distribution31) within this raw APD distribution for the endocardium as the target APD distribution and then dropped parameter sets randomly to achieve the targeted Gaussian APD distribution. The retained parameter sets then formed the normal control model population. For the 3 model populations generated using the 3 AP models, there were ~5000 different 1D cables in each model population (see figure legends for the exact numbers for each case).
to
The blocking effect of methadone on an ionic current was modeled by multiplying the current by a Hill function, that is, for an ionic current Im, the formulation is:(1) Im→11+MIC50nImwhere [M] is the methadone concentration, IC50 is the half‐maximal inhibitory concentration, and n is the Hill coefficient. Im represents IKr, IK1, ICa,L, or INaL. For IKr, we used n=0.8 and IC50=2.9 μM based on Klein et al.13 and Tran et al.12 For IK1, we used n=0.7 and IC50=1.5 μmol/L taken from Klein et al.13 For ICa,L, we used n=1.1 and IC50=5.5 μmol/L taken from Tran et al.12 For INaL, we used n=0.9 and IC50=8.5 μmol/L taken from Tran et al.12 The IC50 estimates for the same ion channel differed in different experiments and conditions (see Klein et al.13 and Tran et al.12 and references cited therein). We took the IC50 data from Klein et al.13 and Tran et al.,12 which were obtained at 37 °C. These IC50s are almost the lowest compared with those from other studies, so that the blocking effects are almost maximum for a specific dose of methadone in our simulations.
RESULTS
To relate the effects of blocking IKr or IK1 on APD to their effects on PVC genesis, we first explored the effects of blocking IKr or IK1 on APD for the 4 AP models simulated in this study (Figure 1). In all 4 models, reducing IK1 by 50% results in roughly a 5% increase in APD but reducing IKr by 50% causes a larger APD increase, as expected. However, their responses to blocking IKr are very different, with the ORd and the modified ORd by Tomek et al. models exhibiting a much larger change in APD than did the other 2 models.
Shown are APD versus percentage of the control value of the maximum IKr conductance (circle) or the maximum IK1 conductance (square) for the 1994 Luo and Rudy model (A), the 2004 ten Tusscher et al. model (B), the O'Hara et al. model (C), and the Tomek et al. model (D). APD indicates action potential duration; IK1, inward rectifier potassium current; IKr, rapid component of the delayed rectifier potassium current; LRd, Luo and Rudy model; ORd, O'Hara et al model; ToRORd, Tomek et al. model; and TP04, ten Tusscher et al. model.
To investigate the effects of blocking IKr or IK1 on PVC genesis, we performed simulations of 1D cables, scanning for what combinations of the maximum conductance of IKr or IK1 and ICa,L produced PVCs. To model repolarization heterogeneity, we reduced maximum conductance of IKs in the endocardial region and added Ito in the epicardial region. We first used cables based on the LRd model to show PVC genesis in more detail (Figure 2) and then cables based on TP04, ORd, and the modified ORd by Tomek et al. models (Figure S1) to verify and compare with the LRd model results.
A, RG‐mediated PVC regions (light blue) on the parameter plane of GK1 and the GCa,L of ICa,L (left) and the parameter plane of GKr and GCa,L (right) when no Ito is present in the epicardial region (maximum Ito conductance =0). The maximum conductance of the slow component of delayed rectifier potassium current is 0.3 folds of control in the endocardium and 2 folds in the epicardium. Red arrows mark the changes in ICa,L threshold for a 50% block of IKr from control. B, Same as (A) but when Ito is present in the epicardial region (maximum conductance of Ito=1.75 mS/cm2). The RG‐mediated PVC regions are colored light blue and the P2R‐mediated PVC regions are colored magenta. The 3 symbols mark the parameter locations for the examples shown in (C through E). Insets are expanded views of the P2R‐mediated PVC regions. C, Time–space‐voltage plot of a case with no PVC (triangle, GK1 is 1 fold and GCa,L 1.1 folds of their control values). D, Time–space‐voltage plot of a RG‐mediated PVC caused by reducing IK1 and increasing ICa,L (diamond, GK1 is 0.7 folds and GCa,L 3.7 folds of control). marks the spontaneous PVC. *E*, Time–space‐voltage plot of P2R‐mediated PVC caused by reducing both IK1 and ICa,L (circle, GK1 is 0.7 folds and GCa,L 0.9 folds of control). marks the spontaneous PVC. Note that the stimulated action potential in the epicardial layer is an abbreviated spike due to lowering GCa,L, which allows the spike‐and‐dome action potential in the middle layer to initiate a PVC in epicardial layer. Endo indicates endothelial; Epi, epicardial; GCa,L, maximum conductance of ICa,L; GK1, maximum conductance of IK1; ICa,L, L‐type calcium current; IK1, inward rectifier potassium current; Ito, transient outward potassium current; P2R, phase 2 reentry; PVC, premature ventricular complex; and RG, repolarization gradient.
For the LRd model, we first excluded Ito in the epicardial layer to avoid Ito‐mediated PVCs (Figure 2A). In this case, PVCs are promoted by enhancing ICa,L and blocking IK1 or IKr (light blue regions). In other words, for a fixed GK1 or GKr, PVCs occur once maximum conductance of ICa,L (GCa,L) is above a certain threshold, or for a high enough GCa,L, PVCs can occur by reducing GK1 or GKr. These PVCs are mediated by EADs in the long APD region or by the RG between endocardium and epicardium (see an example of RG‐mediated PVC in Figure 2D), as shown in previous studies.19, 20 Although blocking IK1 increases APD less than does blocking IKr (Figure 1A), it causes a faster reduction of the threshold ICa,L for PVC genesis than blocking IKr. For example, a 50% reduction of IKr from its control value reduces the ICa,L threshold from 4.5 to 3.9 fold (red arrow in right panel in Figure 2A) of its control value, whereas a 50% reduction in IK1 reduces the threshold from 4.5 to 3 fold (red arrow in left panel in Figure 2A). This indicates that although IK1 exhibits only a small effect on APD and EADs at the cellular scale,32 it plays a critical role at the tissue scale to promote PVCs. This is because both EAD‐ and RG‐mediated PVCs are tissue‐scale phenomena driven by tissue‐scale dynamical instabilities,19, 20 and IK1 plays a critical role in promoting these dynamical instabilities in tissue.
We then added Ito to the epicardial layer and observed qualitative difference in PVC genesis (Figure 2B). Adding Ito created a spike‐and‐dome AP morphology in the epicardial layer (Figure 2C) but had only a small effect on RG‐mediated PVCs, that is, the light blue regions at high GCa,L in Figures 2A and 2B are similar. However, a new PVC region emerged at low GCa,L (magenta regions and insets in Figure 2B). The mechanism of this type of PVCs is via P2R caused by the Ito‐induced spike‐and‐dome AP morphology.33, 34, 35 As shown in Figure 2E, the AP in the epicardial layer became very short (a spike), facilitating the formation of a PVC. The PVC region of this mechanism reduces as GK1 increases until it completely disappears. In other words, blocking IK1 promotes P2R. In general, P2R is promoted by decreasing inward currents or increasing outward currents, so it is counterintuitive that blocking IK1, an outward current, should promote P2R. It promotes P2R because, similar to the RG‐mediated PVCs, the PVCs caused by P2R are also a result of a tissue‐scale instability36 that depends on IK1 strength. Note that the P2R region is insensitive to IKr changes (the width of the PVC region only changes slightly with GKr), agreeing with that shown in our previous studies that P2R is not sensitive to changes of GKr and maximum conductance of IKs. 35, 37
As shown in our previous simulation studies of 2‐dimensional tissue19, 35, 37 or 3‐dimensional heart25 models, the RG‐mediated or P2R‐mediated PVCs can maintain as multibeat focal excitations or degenerate into reentrant arrhythmias, depending on tissue conditions. Therefore, in this study, we used the genesis of spontaneous PVCs as our marker for arrhythmogenesis and a 1D cable model as our test platform.
We compared the results of the LRd model to those of the other 3 AP models (see Figure S1). The results are similar despite some quantitative differences.
In summary, blocking IKr or IK1 combined with increasing ICa,L conductance promotes RG‐mediated PVCs, and blocking IK1 combined with decreasing ICa,L conductance promotes P2R‐mediated PVCs. Although blocking IK1 only exhibits a small effect on prolonging APD and promoting EADs in single cells, it exhibits a much larger effect on promoting PVCs in tissue, a tissue‐scale phenomenon that cannot be revealed by single‐cell studies.
In general, we can infer, from the results shown in Figures 1 and 2 and Figure S1, that among the proarrhythmic effects of methadone, blocking IKr and IK1 prolongs APD and promotes arrhythmias under high ICa,L via RG‐mediated PVCs and that blocking IK1 promotes arrhythmias at low ICa,L via P2R‐mediated PVCs. However, methadone also blocks other currents and causes hypoxia during sleep; how these effects of methadone integrate together to affect arrhythmogenesis is unclear.
Second, in the general population, there is substantial interindividual variability: individual parameter sets will differ from one individual to another. This interindividual variability results in individuals having different responses to a specific drug, including different responses to different drug doses. The results shown in Figures 1 and 2 are based on a single set of control parameters for each model, which is not representative of the heterogeneity of the general population. To take into account interindividual variability, we generated model populations using randomly selected parameter sets and APD filters (see Methods). This randomization‐plus‐filtering resulted in Gaussian‐like distributions of APD from the endocardium of the normal control populations for the LRd model (Figure 3A), the TP04 model (Figure S2A), and the ORd model (Figure S3A). To differentiate the contributions of methadone's blocking effects on different ionic currents, we also used different combinations of its blocking effects. For example, we might include only its IK1 or IKr blocking effect alone or a combination of the 2, etc.
A, APD histogram of the model population under the control condition. This model population contains 5020 cables. B, Bar graph showing averaged APD () and SD (σ) versus methadone concentration. C, Bar graph showing percentage of individuals exhibiting PVCs versus methadone concentration under 5 different conditions of methadone blocking effects (see labels in the inset). The ICa,L strength is 5 folds of control for the LRd model. D, Dependence of percentage of individuals exhibiting PVCs on baseline under the first 4 conditions of methadone effects. The baseline was obtained at control ICa,L as in (B) and the PVCs are obtained with the same ICa,L as in (C). APD indicates action potential duration; ICa,L, L‐type calcium current; LRd, Luo and Rudy model; PVC, premature ventricular complex.
We first used the model populations to assess the effects of methadone on APD under normal control conditions. Figure 3B shows average APDs and SDs for different methadone concentrations and different combinations of methadone blocking effects, marked by different colors, for the LRd model. In general, the responses of the average APD of the model populations (see also SI Figures S2B and S3B) to methadone agree with the results shown in Figure 1: (1) blocking IK1 alone (yellow) lengthens APD only slightly; (2) blocking IKr alone (red) has a bigger effect on prolonging APD than does blocking IK1; (3) blocking IK1 and IKr together (blue) results in stronger effects on lengthening APD than did blocking IKr alone; and (4) the APD lengthening effects due to blocking IK1 and IKr are attenuated by blocking ICa,L and INaL (green and purple).
We then used the model populations to simulate the arrhythmogenic effects of methadone in the setting of high sympathetic activity, modeled by increased GCa,L. We chose to modify only GCa,L because, although β‐adrenergic stimulation increases the amplitudes of both ICa,L and IKs, the amplitude of ICa,L increases much faster than that of IKs, 38 and thus in the early phase of sympathetic stimulation, the main effect is the increase of ICa,L strength. Figure 3C shows the percentages of individuals in the population exhibiting PVCs for the LRd model. When methadone only blocks IKr (red bars), the PVC risk, although it increases with methadone dose, is very low. If methadone blocks IK1 only (yellow bars), the risk increases with dose to a higher percentage than by blocking IKr alone. Interestingly, when methadone blocks both IKr and IK1 together (blue bars), the risk becomes much higher, indicating that blocking IKr and IK1 synergistically promotes arrhythmias. The TP04 model (Figure S2C) behaves similarly to the LRd model. The ORd model (Figure S3C) exhibits some differences from the other 2 models. In the ORd model, blocking IK1 exhibits almost no effects whereas blocking IKr exhibits a large effect. But again, blocking both IKr and IK1 together exhibits a much larger effect, demonstrating that the synergistic effects of blocking both IKr and IK1 still occur as observed in the other 2 models. For all 3 models, as anticipated, blocking ICa,L reduces PVCs (green bars), which is substantiated at high doses of methadone. Due to its high IC50, the effect of INaL on PVC genesis is small (purple bars) with a larger effect occurring in the ORd model.
To correlate APD to PVC risk under different conditions of methadone blocking effects, we plotted the percentage of the population exhibiting PVCs versus averaged baseline APD for different conditions of methadone blocking effects (Figure 3D). Specifically, we plotted the percentage exhibiting PVCs (under high sympathetic activity) against the averaged APD shown in Figure 3B (under control). Blocking IKr alone lengthens APD but does little in promoting PVCs (red). Blocking IK1 alone increases APD slightly but has a large effect on promoting PVCs (yellow, steep curve). Blocking both IK1 and IKr together, the curve is less steep than for blocking IK1 alone but extends to longer APDs and higher percentages due to the synergy effect. The TP04 model behaves similarly to the LRd model (Figure S2D) whereas the ORd model behaves differently when blocking IKr or IK1 alone (Figure S3D). However, all 3 models demonstrated that blocking IKr and IK1 together greatly enhanced the arrhythmia risk.
Finally, we simulated the effects of methadone under low sympathetic activity (control ICa,L) for the LRd model (Figure 4) and the TP04 model (Figure S4). We simulated 2 scenarios: with or without hypoxia. Because we simulated low sympathetic activity, we used the effects of hypoxia on ionic currents under basal condition used by Gaur et al.,39 in which ICa,L was reduced. In the absence of hypoxia (Figure 4A), when methadone blocks IK1 or IKr but does not block ICa,L and INaL, it exhibits almost no effects on PVC genesis (color bars do not show up in graph). When methadone also blocks ICa,L and INaL (green and purple bars), the PVC risk increases with methadone concentration quickly due to its ICa,L blocking effect. Blocking INaL only enhances the risk slightly. In the presence of hypoxia (Figure 4B), blocking IK1 increases the arrhythmia risk (yellow) but blocking IKr with or without blocking IK1 does not affect the risk of arrhythmia (red and blue bars). If methadone also blocks ICa,L and INaL (green and purple bars), the PVC risk decreases with methadone concentration due to too high an ICa,L blockade at high methadone concentrations.
A, Percentage of individuals in the population exhibiting PVCs versus methadone concentration under 5 different conditions of methadone blocking effects (see labels in the inset) for the control GCa,L of the ICa,L without hypoxia. B, Same as (A) but for the control GCa,L with hypoxia. Effects of hypoxia on ionic current conductance are modeled as follows (from Gaur et al. with zero isoproterenol39): for the INa—INa→0.9INa; for the INaL—INaL→1.5INaL; for ICa,L—ICa,L→0.75ICa,L, and for the IKs—IKs→0.73IKs. GCa,L indicates maximum conductance of ICa,L; ICa,L, L‐type calcium current; IKs, slow component of the delayed rectifier potassium current; INa, sodium current; INaL, late sodium current; and PVC, premature ventricular complex.
Because ICa,L is the current affected by sympathetic stimulation, hypoxia, and methadone, it plays a key role in methadone‐induced arrhythmias. To give a broader view of the effects of ICa,L combined with IK1 and IKr blocking effects, we recorded the percentage of PVCs versus ICa,L strength for a fixed methadone concentration (3 μmol/L) with methadone blocking IKr (red boxes) or IK1 (yellow boxes) alone or both together (blue boxes) for the LRd model (Figure 5). For reference, we also plot the percentage of PVCs in the absence of methadone. In all cases, the PVC risk is low in the intermediate ICa,L strength range. For high ICa,L, PVCs are caused by APD prolongation, and the risk increases with the strength of ICa,L. For low ICa,L, PVCs are caused by spike‐and‐dome AP morphology via P2R, which occurs in a narrow range of ICa,L strength. Blocking IK1 greatly increases the risk of PVCs, but blocking IKr exhibits little effects.
Shown are the percentage of individuals exhibiting PVCs versus the maximum conductance of the L‐type calcium current under no methadone action and 3 different conditions of methadone effects as labeled in the inset. The methadone concentration is 3 μM. The open boxes almost completely overlap with the red boxes. PVC indicates premature ventricular complex.
DISCUSSION
In this study, we used computational simulation to investigate the multiple ion channel blocking effects of methadone on arrhythmogenesis, beyond its role as a hERG blocker. Our major findings are: Blocking IK1 does not prolong APD significantly but can substantially increase the occurrence of RG‐mediated PVCs at high sympathetic activity via tissue‐scale instabilities. Blocking IK1 works synergistically with blocking IKr to greatly increase the susceptibility to arrhythmias. Blocking ICa,L (and INaL) at high methadone concentrations can attenuate the risk of arrhythmias, as anticipated.Blocking IK1 promotes P2R at low or basal sympathetic activity, which is potentiated by the ICa,L blocking action of methadone. Blocking IKr has little effect on this mechanism of arrhythmias.Hypoxia may promote RG‐mediated PVCs at high sympathetic activity by enhancing ICa,L and INaL and also promote P2R at basal or low sympathetic activity by suppressing ICa,L.
Blocking IK1 does not prolong APD significantly but can substantially increase the occurrence of RG‐mediated PVCs at high sympathetic activity via tissue‐scale instabilities. Blocking IK1 works synergistically with blocking IKr to greatly increase the susceptibility to arrhythmias. Blocking ICa,L (and INaL) at high methadone concentrations can attenuate the risk of arrhythmias, as anticipated.
Blocking IK1 promotes P2R at low or basal sympathetic activity, which is potentiated by the ICa,L blocking action of methadone. Blocking IKr has little effect on this mechanism of arrhythmias.
Hypoxia may promote RG‐mediated PVCs at high sympathetic activity by enhancing ICa,L and INaL and also promote P2R at basal or low sympathetic activity by suppressing ICa,L.
The insights from the computer simulations reveal the important roles of IK1 in both RG‐mediated and P2R‐mediated PVCs and the mechanisms of methadone‐induced arrhythmogenesis, as discussed next.
It is well understood that blocking IKr lengthens APD and promotes EADs, which are associated with QT prolongation and arrhythmogenesis in the inherited LQTS type 2 and in most cases of drug‐induced LQTS. Blocking IK1 gives rise to a small increase in APD (Figure 1) and thus its contribution to QT prolongation is small.
But on the other hand, our present and previous studies19, 20 showed that although blocking IK1 plays only a small role in APD prolongation (Figure 1) and in EAD genesis,32 it plays a much greater role in promoting PVCs in tissue (Figures 2 and 3). We showed that RG‐mediated (or phase 2 EAD‐mediated) PVCs originate from a tissue‐scale dynamical instability caused by RG and enhanced ICa,L, 19, 20, 36 and IK1 potentiates this instability without affecting the APD or the RG per se. Therefore, the main effect of IK1on PVC genesis is at the tissue level by promoting dynamical instability in tissue, whereas the effect of IKris mainly at the single‐cell level by prolonging APD and promoting EADs.
Because IK1 plays a key role in stabilizing the resting potential, reducing IK1 can promote automaticity40, 41 or increase the amplitude of delayed afterdepolarizations (DADs) to promote DAD‐mediated triggered activity,42, 43 creating arrhythmogenesis. Note that DADs are caused by spontaneous Ca2+ release, a manifestation of instability in intracellular Ca2+ cycling in the heart. Although methadone's IKr and IK1 blocking effects should not alter Ca2+ handling, partial blockade of ICa,L may have a nontrivial effect on Ca2+ cycling. However, blocking ICa,L will reduce intracellular Ca2+ load and Ca2+ release, suppressing Ca2+ instability for DADs. In other words, methadone may promote DAD‐mediated arrhythmias by destabilizing the membrane potential via its IK1 blocking effect, rather than by destabilizing Ca2+ cycling.
For the same reason, IK1 is thought to be important for the genesis of phase 3 EADs,17 which have been observed in Purkinje fibers44, 45, 46 and in mouse and rat ventricular myocytes.47, 48, 49 However, our previous survey of the literature showed that no phase 3 EADs had been observed in isolated ventricular myocytes of large animals and humans.32 Computer simulations showed that to generate phase 3 EADs in ventricular myocytes of large animal or human AP models, drastic changes in ionic current conductance are needed,50 and these large changes in parameters may well be unphysiological. We did not observe phase 3 EADs by blocking IK1 and increasing ICa,L in the AP models used in this study. Therefore, whether blocking IK1 promotes TdP via promoting phase 3 EADs or not remains unclear.
More intuitively, reducing IK1 makes the resting potential or phase 3 repolarization of the AP less stable, allowing spontaneous occurrence of depolarizations (thus PVCs) in the RG region. Note that phase 3 EADs may be rarely observed in isolated ventricular myocytes but can be observed in tissue22 due to electrotonic coupling and the tissue‐scale genesis of PVCs. For example, in Figure 2D, one can record a phase 3 EAD in the middle of the cable (the RG region) whereas there are no EADs in the long and short APD regions at all.
Another important finding is that blocking IK1 works in synergy with blocking IKr to promote PVCs. In the cases shown in Figures 3; Figures S2 and S3, when only IKr was blocked, increasing the methadone concentration from 0 to 5 μmol/L increased the PVC risk from 0 to 0.18% in the LRd model, to 0% in the TP04 model, and to 4.1% in the ORd model. When only IK1 was blocked, the PVC risk increased from 0% to 9.36%, 3.58%, and 0%, respectively. However, when both were blocked, the PVC risk increased from 0% to 12.56%, 23.54%, and 20.06%, respectively. Therefore, blocking both IKr and IK1 causes an increase in PVC risk that is disproportionate to the sum of blocking them alone, indicating a strong synergistic effect.
The synergy can be understood as follows. Blocking IKr prolongs APD and enhances RG whereas blocking IK1 does not prolong APD or enhance RG but destabilizes the system. When the 2 work together, they work cooperatively or synergistically, amplifying the effect on PVC genesis. This synergistic effect may explain the experimental51 and clinical52 observations that when IKr and IK1 are reduced simultaneously, arrhythmia events increase disproportionately compared with reducing either one alone. Specifically, the study by Hornyik et al.51 showed that the arrhythmia rate was <5% for transgenic LQT2 or LQT2‐5 rabbits under control, but when treated with an IK1 blocker, the rate increased to >50%. However, there were little effects in the wild type or LQT5 rabbits with the same treatment. In the clinical study by Mazzanti et al.,52 they showed that in Andersen‐Tawil Syndrome Type 1 patients (where a Kir2.1 mutation reduces IK1), treatment with the IKr blocker amiodarone (a class III antiarrhythmic drug) caused a >40‐fold increase in arrhythmia risk compared with treatments with β‐blockers or class 1c antiarrhythmic drugs.
P2R is a term used to describe a PVC or reentry caused by the spike‐and‐dome morphology induced by Ito 33, 34 or by altered Na+ channel kinetics.37 It has been proposed as one mechanism of arrhythmogenesis in Brugada syndrome (BrS) and early repolarization syndrome (ERS).53 In general, P2R is promoted by reducing inward currents or increasing outward currents, in contrast to PVC genesis in delayed repolarization or LQTS. For example, our simulations showed that P2R is promoted by increasing Ito (Figures 2 and Figure S1) or by reducing ICa,L (Figure 5). A non‐intuitive observation from our simulation is that blocking IK1, an outward current, can greatly increase the risk of P2R (Figures 4 and 5; Figure S4). Blocking IK1 also promotes P2R caused by altered Na+ channel kinetics.37 Here, the role of IK1 is the same as its role in promoting RG‐mediated PVCs, ie, by promoting tissue‐scale dynamical instabilities36 or by making the resting potential and phase 3 repolarization less stable.
Based on the mathematical formulation by Gaur et al.,39 hypoxia alters INa, INaL, ICa,L, and IKs and their responses to isoproterenol. For example, in the mathematical formulation by Gaur et al.,39 ICa,L→Isofactor×ICa,L with Isofactor=1+31+5.3Iso for the normal condition and Isofactor=0.75+31+1.6Iso for hypoxic condition. For low isoproterenol (eg, for Iso=0, Isofactor=1 for the normal condition and 0.75 for hypoxia), ICa,L is lower under hypoxia than under the normal condition. For intermediate or high isoproterenol (eg, for Iso=1.6, Isofactor=1.69 for the normal condition and 2.25 for hypoxia), ICa,L is higher under hypoxia than under the normal condition. Therefore, at low sympathetic activity or basal condition, hypoxia reduces ICa,L, but at normal or high sympathetic activity, it increases ICa,L. The consequence is that hypoxia promotes arrhythmias at both low and high sympathetic activity via promoting P2R‐mediated and RG‐mediated PVCs, respectively.
QT prolongation has been shown for methadone use13, 14, 54, 55 with a high risk of TdP.7 Like many other LQTS or TdP‐genic drugs, methadone also blocks IKr and thus causes QT prolongation and TdP as widely anticipated. However, biophysical studies showed a wide range of IC50, from ~3 to 10 μmol/L8, 10, 11, 12, 13 or even higher.9 For example, for a methadone concentration of 1 μmol/L, if the IC50 is 3 μmol/L, IKr is blocked by only ~25%. If the IC50 becomes 10 μmol/L, only ~10% IKr is blocked. Therefore, the IKr blocking effect may not be sufficient to explain methadone's proarrhythmic effect.
On the other hand, it has been shown that methadone blocks IK1 with an IC50 of 1.5 μmol/L.13 Our simulations showed that blocking IK1 alone, although it has only a small effect on prolonging APD or the QT interval, can increase the risk of RG‐mediated PVCs. More interestingly, blocking IK1 and IKr simultaneously results in a synergistic effect that increases the risk of arrhythmias many‐fold more than blocking either one alone. In other words, moderately blocking IKr combined with blocking IK1 increases QT moderately, but the synergistic effect of blocking the 2 greatly increases the TdP risk, making methadone a drug causing a high incidence of TdP with moderate QT prolongation.
For the same reason, due to the IK1 blocking effect of methadone, there are many potential drug interactions which would promote arrhythmia toxicity. For example, a hERG blocking drug combined with methadone may have a disproportionately higher arrhythmia risk than each alone. Clinicians should therefore avoid combining methadone with known hERG blockers.
Although the major focus of methadone effect is on QT prolongation and TdP, BrS‐like ECGs have also been observed in methadone users.5, 56, 57, 58, 59 Our simulations in this study showed that methadone's IK1 blocking effect combined with its ICa,L blocking effect can promote P2R at low or basal sympathetic activity, agreeing with arrhythmias occurring at low sympathetic or high vagal activity in BrS and ERS.60, 61, 62, 63 Note that unlike QT prolongation, BrS ECGs are not persistent even in the same patient and thus cannot be easily or widely detected. Clinically, Na+ channel blockers are often used to unmask BrS. Furthermore, our simulation showed that blocking IKr prolongs QT but has little effect on either promoting or suppressing P2R, and thus QT prolongation can still be a dominant ECG marker, but the mechanism of arrhythmias can be P2R. Therefore, the prevalence of BrS or ERS‐related arrhythmias in methadone users might be highly underestimated.
Because blocking ICa,L or INa can cause BrS and ERS,53 blocking IK1 may work synergistically with blocking ICa,L or INa to greatly increase the risk of P2R. Therefore, a Na+ channel blocking drug or a nondihydropyridine Ca2+ channel blocking drug combined with methadone might substantially increase the risk of P2R‐induced arrhythmias and likely should be avoided, particularly for patients with evidence of BrS or ERS.
Elevated U‐waves are another ECG feature in methadone users,5, 6 which agrees with methadone's IK1 blocking effect.13, 64 Although the exact mechanism of the U‐wave is unclear, it is related to phase 3 repolarization and phase 4 depolarization, where IK1 becomes important. One of the links of U‐waves to arrhythmias is via DAD‐mediated arrhythmogenesis65, 66 in which reducing IK1 lowers the threshold for a DAD to trigger an AP for arrhythmogenesis.42, 67 DADs are caused by spontaneous intracellular Ca2+ release events, which can be triggered by high sympathetic activity.68
Besides its ion channel blocking effects, another arrhythmia risk factor for methadone use is hypokalemia.6, 69, 70, 71 One of the effects of hypokalemia is reduction of IK1 conductance, and lowering IK1 is a key parameter for arrhythmias in all 3 mechanisms, as discussed previously. Therefore, in the case of methadone use, IK1 is already lowered by the drug action and hypokalemia makes it even worse.
Our simulation results may provide insights into the observation that most methadone deaths occur during sleep.1, 16, 18 It is known that cardiac arrhythmias for LQTS occur during physical or emotional stress except for LQT3 where the cardiac events occur mainly during sleep.72, 73 For catecholaminergic polymorphic ventricular tachycardia, arrhythmias occur in the afternoon to evening due to higher physical activity,74 and for BrS and ERS, most arrhythmias occur in the nighttime due to low sympathetic activity.60, 61, 62, 63 Our simulations show that blocking IK1 combined with lowering ICa,L caused by low sympathetic activity and methadone‐related ion channel blockade and hypoxia during sleep promotes P2R for arrhythmogenesis. This agrees with the observations that BrS‐like ECGs occur in methadone users and that methadone‐related deaths mainly occur during sleep. Furthermore, serum K+ concentration is lower in the nighttime (bottoming around 9 p.m.),75, 76, 77 which can further increase the likelihood of arrhythmias during sleep.
Studies have shown that men account for 60% to 70% of methadone‐related deaths.1, 78 The causes of this sex difference are unknown. Our current study may provide mechanistic insights into this observation. It is known that women have larger ICa,L and smaller K+ currents than men,79 causing women to be more susceptible to delayed repolarization arrhythmias and men to be more susceptible to early repolarization arrhythmias. If methadone related deaths occur primarily via delayed repolarization arrhythmias, it is difficult to explain why men have a higher methadone‐related death rate since women are more susceptible to LQTS arrhythmias. However, as shown in our study, methadone's effects on blocking ICa,L directly or indirectly via hypoxia, combined with its IK1 blockade, can substantially increase the risk of P2R. As shown in our previous computer simulations,35, 37 due to a higher ICa,L, women are much less susceptible to P2R than men, which provides a mechanism for the fact that the majority of patients with BrS and ERS are men.53 Similarly, we can apply the same mechanism to the observation that men have a high methadone‐related death rate due to a higher susceptibility to methadone‐induced P2R.
Therefore, methadone is a multichannel blocking drug with different IC50s. Its blocking effects on IKr and IK1 cause QT prolongation to promote delayed repolarization or LQTS‐related arrhythmias, but its blocking effects on ICa,L and INaL are protective from these arrhythmias. On the other hand, its blocking effects on ICa,L and INaL combined with its blocking effect on IK1 promote early repolarization arrhythmias. Furthermore, blocking IK1 may also potentiate automaticity or DAD‐related arrhythmias. These arrhythmogenic effects are dose‐dependent due to the different IC50s. Because the IC50 for blocking IK1 is the lowest among those for other currents, blocking IK1 promotes all mechanisms of arrhythmias, and blocking IK1 works in synergy with other ion channel blocking effects, the IK1 blocking effect could be the biggest risk factor for arrhythmias for methadone. Moreover, due to the synergistic effects, caution must be taken when prescribing it with other drugs.
Finally, we would like to note that, similar to methadone, many drugs exhibit multichannel blocking effects,80, 81 which make their proarrhythmic and antiarrhythmic effects complex. For example, amiodarone blocks IKr, ICa,L, and INa as well as β‐receptors. Although amiodarone is one of the most commonly prescribed antiarrhythmic drugs in the United States, its proarrhythmic effects are not negligible.82 It can be intuitively understood that the antiarrhythmic effect of amiodarone is largely due to its ICa,L and β‐receptor blocking actions, whereas its proarrhythmic effect mainly stems from its IKr blockade. These effects are also highly dependent on the IC50s. However, our current study introduces 2 novel aspects regarding a drug's pro‐ and antiarrhythmic effects. The first is the synergistic effect of multichannel blockade on arrhythmogenesis. When the blocking effects act synergistically, the arrhythmia risk can increase dramatically compared with when each acts alone. For instance, amiodarone is highly proarrhythmic in patients with Andersen‐Tawil Syndrome Type 1,52 most likely because its IKr blocking effect interacts synergistically with the loss of function IK1 mutation, whereas it remains antiarrhythmic in many other clinical settings.82 The second novel aspect is the proarrhythmic effect on early repolarization. Current cardiotoxicity studies on drug‐induced cardiac arrhythmias predominantly focus on QT‐prolongation‐related arrhythmias. To the best of our knowledge, no cardiotoxicity studies have specifically addressed the proarrhythmic effects of drugs on early repolarization‐related arrhythmias. It is not difficult to conceive that a multichannel blocking drug can suppress delayed repolarization arrhythmias but promote early repolarization arrhythmias, or vice versa, which need to be addressed in the drug's cardiotoxicity assessment. Furthermore, some of the proarrhythmic effects may only be observable in tissue‐level studies, not in single cells, as shown in this study.
Several limitations are worth mentioning. First, we simulated only 1D cable models for spontaneous genesis of PVCs, to allow for a large number of simulations, but arrhythmias are tissue‐level phenomena in the 3‐dimensional heart. To fully model arrhythmias, 2‐ or 3‐dimensional tissue models are needed. Furthermore, the shape of the geometry of the heterogeneous region in higher dimensions than 1D may also be nontrivial for arrhythmogenesis. However, as shown previously,26, 35, 37 once a spontaneous PVC occurs, it may exhibit as a benign ectopic beat or degenerate into a reentrant arrhythmia, depending on properties of the heterogeneous region. Therefore, the propensity of spontaneous PVCs in the 1D cable is a good measure of propensity for arrhythmias in higher dimensional tissue. Second, we used 1 fixed IC50 for each current and combinations of inclusion and exclusion of currents for methadone action to assess arrhythmogenesis related to methadone use. In reality, the IC50 for methadone action may differ in different experimental conditions, in different species, as well as in different individuals. These differences may give rise to different conclusions for the contributions of different ionic currents caused by methadone. Third, we modeled sympathetic activity by altering ICa,L only. It is known that β‐adrenergic stimulation causes strong increases of both ICa,L and IKs. Because the maximum ICa,L strength elevates much faster than that of IKs in response to β‐adrenergic stimulation,38 there is a time window in which ICa,L enhancement is the dominant effect. Furthermore, β‐adrenergic stimulation may affect other ionic currents via the protein kinase signaling pathways,83 which may need to be taken into account. Particularly, experimental studies41, 49, 84, 85, 86, 87 have shown that β‐adrenergic stimulation can either increase or decrease IK1, which may also differ across species and in acute and chronic effects.84 If β‐adrenergic stimulation increases IK1, it is protective of arrhythmias triggered by sympathetic surge. The protective or proarrhythmic effects may also depend on the time course of IK1 relative to that of ICa,L in response to β‐adrenergic stimulation. Nonetheless, in a more modern view of evolution,88, 89, 90 the complex effects of β‐adrenergic stimulation on IK1 and other currents may just be a “good enough solution” for normal life. Methadone blocking IK1 breaks the natural balance of the solution, greatly increasing the susceptibility to arrhythmias of different mechanisms as discussed previously. Fourth, although we simulated 4 AP models and obtained generally consistent results, there are still differences between models, such as the roles of IKr and IK1. It is known that different AP models may exhibit very different responses to the same ionic current blocker,20, 81 which is one of the reasons that we employed 4 AP models in this study. Nevertheless, the computer simulation results in this study provide mechanistic insights into methadone‐induced arrhythmogenesis and sudden cardiac death.
CONCLUSIONS
The computer simulations in this study reveal that blocking IK1, which exhibits only a small effect on APD prolongation, can promote both early repolarization and delayed repolarization arrhythmias by promoting tissue‐scale instabilities for PVC genesis. The arrhythmia risk may increase disproportionately when an IK1 blocker is combined with IKr blocker or ICa,L blocker or enhancer due to synergistic effects. Methadone blocks multiple ionic currents including IK1, IKr, ICa,L, and INaL with different IC50s, and these blocking effects alone or in combination can promote both early and delayed repolarization arrhythmias, namely. The synergistic effect due to the multichannel blocking effects of methadone may greatly increase the methadone's propensity for arrhythmias and sudden cardiac death. Particularly, our results suggesting that methadone promotes early repolarization arrhythmias provides a mechanism for the clinical observation that methadone related deaths mainly occur during sleep and are higher in men than in women. Furthermore, the results of our simulation also imply that due to the synergistic effects of multi‐channel blockade, methadone (mainly through its IK1 blocking effect) may interact with other drugs to disproportionately increase its proarrhythmic effects, which needs to be avoided in drug prescription.
Sources of Funding
This study is partially supported by National Institutes of Health grants R01 HL175074, R01 HL157116, R01 HL139829, and P01 HL164311. Mark C. Haigney is supported by Defense Health Agency grant HU00012420018. Zhaoyang Zhang is supported by National Natural Science Foundation of China Grants 12375033 and 12235007; Zhejiang Provincial Natural Science Foundation of China Grant LY23A050002; Ningbo Youth Leading Talent Project 2024QL032; and Ningbo University K. C. Wong Magna Fund.