Mechanisms of flecainide induced negative inotropy: An in silico study.

It is imperative to develop better approaches to predict how antiarrhythmic drugs with multiple interactions and targets may alter the overall electrical and/or mechanical function of the heart. Safety Pharmacology studies have provided new insights into the multi-target effects of many different classes of drugs and have been aided by the addition of robust new in vitro and in silico technology. The primary focus of Safety Pharmacology studies has been to determine the risk profile of drugs and drug candidates by assessing their effects on repolarization of the cardiac action potential. However, for decades experimental and clinical studies have described substantial and potentially detrimental effects of Na+ channel blockers in addition to their well-known conduction slowing effects. One such side effect, associated with administration of some Na+ channel blocking drugs is negative inotropy. This reduces the pumping function of the heart, thereby resulting in hypotension. Flecainide is a well-known example of a Na+ channel blocking drug, that exhibits strong rate-dependent block of INa and may cause negative cardiac inotropy. While the phenomenon of Na+ channel suppression and resulting negative inotropy is well described, the mechanism(s) underlying this effect are not. Here, we set out to use a modeling and simulation approach to reveal plausible mechanisms that could explain the negative inotropic effect of flecainide. We utilized the Grandi-Bers model [1] of the cardiac ventricular myocyte because of its robust descriptions of ion homeostasis in order to characterize and resolve the relative effects of QRS widening, flecainide off-target effects and changes in intracellular Ca2+ and Na+ homeostasis. The results of our investigations and predictions reconcile multiple data sets and illustrate how multiple mechanisms may play a contributing role in the flecainide induced negative cardiac inotropic effect.

Introduction

While important progress has been made recently via safety pharmacology studies designed to screen drugs for pro-arrhythmia, the primary focus has been on determining the extent to which drugs cause repolarization abnormalities [1,2]. In particular, the emphasis has been on the proclivity for drugs to block the cardiac potassium channel hERG, an infamously promiscuous drug target [3-5]. More recently, studies have begun to take into account the simultaneous effects of drugs on multiple cardiac ion channel. These usually result in complex nonlinear emergent dynamics that can alter the electrophysiological behavior in the heart in unpredictable ways [2]. Computational modeling and simulation approaches have proven to be a useful complement to experimental studies in attempts to rationalize and predict the complex ion channel blocking drug effects on normal cardiac physiology and in pathological states associated with disease.

The focus on repolarization abnormalities in safety pharmacology screening has resulted in insufficient emphasis on closely related depolarization abnormalities and the potential side effects resulting from inhibition of sodium current, although some new technologies are emerging [6]. Such side effects include well-known conduction slowing, but in the ventricle other side effects are also recognized as important risk factors for drug safety [7-10]. One such example is the potential for some sodium channel blocking drugs to reduce the effective mechanical functioning of the heart via a reduction in inotropy, thereby resulting in substantial reduction in the effective pumping function of the ventricles [11,12].

Experimental data obtained from multiple species and from clinical studies have consistently demonstrated a significant negative inotropic effect of flecainide within the clinically relevant dosing range. This strongly contraindicated phenotype is heart rate dependent and exacerbated by rapid heart rate as well in the ischemic and failing heart [8,9,13-20].

Here we have developed a computational framework to predict Na+ channel drug effects on normal cardiac electrophysiology in the Grandi-Bers computational representations of ventricular [21] myocytes and tissue. We have revealed the effects of flecainide on key parameters that regulate cardiac inotropy resulting in new insights regarding Na channel blocker induced negative inotropy. Here we explore the reported plasma concentration ranges reported in humans to assess the physiological relevance and mechanisms of flecainide induced negative inotropy. The therapeutic plasma range of 0.5 to 2.4 μM Flecainide corresponds to administration of ~200 to 1000 ng/mL [22,23].

Methods

We utilized the computational models based on the Grandi-Bers formulations for atrial [24] and ventricular [21] myocyte action potential and [Ca2+] homeostasis as well as the O’Hara-Rudy ventricular human model [25]. The computational model for flecainide interaction with the cardiac Na+ channel is based on publications from the Clancy Group [26,27], with flecainide interaction with the RyR2, IKr, Ito, and ICaL informed by our simulation studies and experimental data [1,27-29].

RyR2 – flecainide drug-channel interaction model

The Shannon-Bers Markov model formulation of the RyR2 [30] was modified to include a drug bound state DO with transitions kon = D* [Drug] and koff = D*IC50,Drug to and from the open state O. These represent the drug diffusing to the receptor, and then binding or dissociating from the channel, respectively, as described previously [27]. The diffusion rate of flecainide was estimated at 5500 M−1 ms−1 [31].

Simulation of IKr, Ito, or ICaL Blockade

To simulate the inhibitory effects of flecainide on IKr Ito, or ICaL currents, we decreased the peak conductance, GX of each of these independent channels in a concentration dependent fashion using a concentration response relationship with a Hill coefficient of 1 (n = 1) as follow: where GX,max is the nominal conductance value obtained from each ventricular myocyte model, and the IC50 is the concentration of drug that produces a 50% inhibition of the targeted transmembrane current or intracellular organelle.

Fiber simulations

One-dimensional (1D) tissue was simulated as a fiber of 165 cells (1.65 cm) [32] with reflective boundary conditions. The diffusion coefficient Dx was set to 0.00154 cm2/ms to establish a conduction velocity of 60–73 cm/s in WT conditions [33]. Fibers were paced from 95 BPM to 150 BPM for 1500 beats. The last 100 beats of atrial or ventricular conducted action potentials of the middle cell and the major underlying ion currents were recorded for analysis.

QT intervals

The fiber (5.5 cm) was paced at varying cycle length from 554 to 943 ms for 200 beats in order to match the clinical data [34]. Pseudo ECGs were computed from the transmembrane potential V using the integral expression as in Gima and Rudy [35]. Heart rate corrected QT (QTc) was computed using Bazett’s formula using the cubic root of RR interval [34].

Please see supplemental information for additional details about our previously published models used in this study. All source code to reproduce the simulated data here are freely available on the GitHub: https://github.com/ClancyLabUCD/Mechanisms-of-flecainide-induced-negative-inotropy-An-in-silico-study

Results

Flecainide is a classic “dirty drug” in that it binds to and affects multiple cardiac ionic targets [36]. We developed a computational model of the effects of flecainide to assess and integrate its effects on 5 targets in the heart including: INa (IC50 is dependent upon voltage, state and rate dependent [26]), IRyR (IC50 = 20 μM [27]), IKr (IC50 = 1.5 μM [28]), Ito (IC50 = 15.2 μM [29]) and IcaL (IC50 = 27.1 μM [1]). We then used the model to simulate the rate-dependence and dose-dependence of the effects of flecainide on key cardiac parameters, including the Ca2+ handling as a proxy for the effects of flecainide on inotropy.

In Fig. 1, shows the comparison of experimental data in Panel A from the rat with the indicated doses (Fernandes et al., [8]) as follows: black dot (no drug), purple (0.18±0.03 μM), red (0.67±0.34 μM), gold (1.0±0.12 μM), and green (1.36±0.14 μM) and our simulations. In Panel B experimental data illustrating the effects of flecainide on canine hearts paced at 150±4 BPM is shown (plasma concentration at 5000 ng/mL, from van Middendorp et al., [13]). In Panel C, experimental data from the pig (4.8 μM, from Marum et al., [14]) is shown. Panels D, E and F show simulated data from the Grandi Bers human ventricular computational model indicating the relationship between the flecainide induced reduction in inotropy as a function of increased duration of the QRS complex. Panel D shows the dose dependent effect of flecainide (at 140 BPM with 2.5 (blue), 3.0 (green) and 4.8 (pink) μM) on peak intracellular Ca2+ transient and in Panel E these data are plotted as the integral of the Ca2+ transient (as a proxy for inotropy). The simulated QRS interval change as a function of flecainide reduction in peak INa is shown in Panel F. In Panel G the effects of flecainide from 0, 2.5 (left), 4 and 12 μM (right) on computed ventricular electrograms. Virtual myocytes were paced at 140 BPM. Note that conduction block developed at 4.8 μM flecainide in the simulation resulting in a dynamic timecourse shown during the last 7 beats of the simulation (pink symbols in Panels D, E and F). The simulations also revealed substantial rate-dependent effect of flecainide on QRS widening, as would be anticipated from the well-described use-dependent effects of flecainide in INa (Fig. 1G).

Fig. 1.

Correlation between flecainide-induced widening of a surrogate measure for the QRS complex and reduction in LV dP/dtmax in a (A) Langendorff rat heart with the indicated doses (Fernandes et al., [8]). (a: 0.18±0.03 μM, b: 0.67±0.34 μM, c: 1.0±0.12 μM, and d: 1.36±0.14 μM) (B) Canine hearts paced at 150±4 BPM and reached a plasma concentration at 5000 ng/mL (van Middendorp et al., [13]). (C) Pig hearts paced at 140 BPM with IT administration of flecainide (1.5 mg/kg, bolus) and reached Cmax at 2000 ng/mL (approximately 4.8 μM - pink). Predicted effects in human ventricular computational tissue of flecainide on the relationship between the widening of QRS complex and (D) the peak intracellular Ca2+ transient, and (E) the integral of the Ca2+ transient, and (F) the peak INa at 140 BPM with 2.5 (blue), 3.0 (green) and 4.8 (pink) μM. (G) Effects of flecainide of 0, 2.5 (left), 4 and 12 μM (right) on computed ventricular electrograms. Myocytes were paced at 140 BPM. Conduction block was observed when flecainide ≥ 4.0 μM.

Fig. 2 shows the comparison of human clinical data from [34] and model predictions from simulated human tissues under drug free conditions and following low dose of flecainide (0.2 μM) and high dose of flecainide (2.0 μM) with varying pacing cycle lengths. The simulated ranges compared to clinically obtained data from humans are in good agreement, thereby providing an indication of the validity and predictive value of the computational results to recapitulate the effect of a flecainide on the human electrophysiology parameters including QRS interval, JT internal and QT interval.

Fig. 2.

Validation of the simulated drugs with human clinical data (A) QT intervals and (B) Heart rate corrected pseudo ECG (ΔQTC interval) was computed from a 1-dimensional strand of Grandi-Bers human cardiac ventricular myocytes for a range of flecainide concentrations (0.2 and 2.0 μM) compared to clinical data (black lines). (C) Comparison of human clinical data showing control and flecainide affected JT intervals (black lines) and simulated mean values under the same conditions ((yellow and red lines). (D) The clinically observed and in silico prediction of QRS intervals changes in a dose-dependent manner. Clinical data: n = 39 for QT, QTc, and JT intervals; n = 47 for QRS.

We next undertook a more detailed investigation into the rate-dependent and dose-dependent effects on the relationship between predicted widening of the QRS interval and drug-induced negative inotropy. In Fig. 3, the predicted dose-dependent effects of flecainide are shown systematically from drug free (black), 0.5 (red), 1.0 (yellow), 1.5 (purple), 2.0 (green) to 2.5 (blue) μM. Rate-dependent (indicated heart rate from 95 to 145 BPM) widening of the QRS interval is shown in Panel A compared to peak INa. Panel B shows the widening of the QRS interval compared to the reduction in the peak intracellular Ca2+ transient and in Panel C as the reduction in the integral of the Ca2+ transient compared to widening of the QRS interval. Interestingly, the approximately linear correlation between the predicted increase in QRS duration and corresponding reduction in inotropy on Ca2+ transient did not significantly change with pacing frequency (Fig. 3D). Inotropy notably continued to decline at fast rates as QRS widening increased, indicated by the similar slopes of the relationships in panels B and C at faster pacing frequencies.

Fig. 3.

Predicted effects of flecainide in human ventricular computational tissue on the relationship between the widening of QRS duration and (A) the peak INa, and (B) the peak intracellular Ca2+ transient, and (C) the integral of the Ca2+ transient, and (C) from 0 (drug free – black), 0.5 (red), 1.0 (yellow), 1.5 (purple), 2.0 (green) to 2.5 (blue) μM flecainide with the indicated heart rate from 95 to 145 BPM. (D) Slopes of correlation between the QRS, the peak INa, peak intracellular Ca2+ transient and integral of the Ca2+ transient as a function of heart rates.

When the underlying ionic currents from a single virtual myocyte were examined in the tissue level simulations described above, we observed a profound slowing of the action potential upstroke (Fig. 4A), which is an expected consequence of INa block (Fig. 4B). In Fig. 4C-D, the mechanism of flecainide induced reduction in inotropy is revealed by the considerably reduced Ca2+ current (Panel C) and Ca2+ transient (panel D). The model predictions in a myocyte within a coupled tissue environment reveal the effect of flecainide to reduce the peak of L-type calcium current. This is due to slow conduction that allows the calcium channels to begin the inactivation process during slow depolarization [37].

Fig. 4.

Illustration of the ionic mechanisms of simulated effects of flecainide on human ventricular computational tissue on 5 targets: INa, IKr, Ito, IRyR, ICaL. Fibers of 165 cells were paced at 150 BPM for 1500 beats. (A) The last conducted action potential of the middle cell (cell #82) is shown. (B) 2.5 μM flecainide (blue) reduced peak INa and depressed cellular excitability (red arrow), as well as (C – D) reduced ICaL activation and the intracellular Ca2+ transient.

As a result, the effect flecainide on a myocyte within coupled tissue (a syncytium) is not the same as what occurs in the single isolated myocyte. The predicted effect in a single simulated myocyte is shown in supplemental Fig. S1. In this case, the main effect of flecainide to reduce the upstroke velocity of the action potential results in a slowing of the depolarizing phase of the membrane potential, but the relative degree of slowing is much larger in coupled tissue. When sodium channels are blocked in the single isolated myocytes, the membrane voltage remains in a range that is favorable for calcium channel activation. In single myocytes shown in supplemental Fig. S1, the slowing of the AP upstroke results in an increase in calcium current at fast heart rates.

In order to further assess the mechanism underlying the negative inotropic effect of flecainide, we next conducted a series of “component dissection” simulations and ion monitoring to assess the primary effects of flecainide that lead to a reduction in the Ca2+ transient. The results are shown in Fig. 5. Our approach was to sequentially remove the effect of flecainide on individual ion current drug targets and then to assess the magnitude of the effect on the changes in the Ca2+ transient. In doing so, we were able to reveal the relative contributions of flecainide effects on specific ion current targets to modification of the Ca2+ transient. In Fig. 5A from left to right, the effects of flecainide on all five simulated targets are shown for comparison on the peak INa, peak Ca2+ transient and the integral of the Ca2+ transient reduction.

Fig. 5.

Correlation between flecainide-induced widening of simulated human ventricular computational tissue derived QRS and reduction in peak Ca2+ transient, and integral of the Ca2+ transient, and peak INa within the therapeutic plasma range of 0.5 to 2.5 μM (0.5 - red, 1.0 - yellow, 1.5 - purple, 2.0 – green and 2.5 – blue). (A) Effects of flecainide on 5 targets, (B) Effects of flecainide on 4 targets and (C) Effects of flecainide on 3 targets. (D) slopes of relationship between flecainide- induced widening of simulated QRS and reduction in peak INa (left), and peak Ca2+ transient (middle) and integral of the Ca2+ transient (right) for each case. Fibers of 165 cells were paced at 150 BPM for 1500 beats. The last conducted action potential of the middle cell (cell #82) is shown.

When the effect of flecainide on the L-type Ca2+ current (IcaL) was eliminated (as in Fig. 5B), but the effect on the other four ion current targets remained, the negative inotropic effect of flecainide was reduced, but still persisted as indicated by peak INa reduction, the reduction in peak [Ca2+] and integral of the Ca2+ transient (from left to right). Interestingly, as shown in Fig. 5C, when the effect of flecainide on the L-type Ca2+ current (IcaL) and the ryanodine receptor (RyR) were both eliminated, but the effect on the other three ion current targets remained, flecainide still resulted in a reduction in the peak Ca2+ transient and the integral of the Ca2+ transient at the rapid heart rate shown (150 BPM).

A summary of the effects of flecainide during the elimination of selected flecainide targets is shown in the bar graphs in panel D, where the magnitude of the changes in the peak INa, the predicted peak Ca2+ transient and the integral of the Ca2+ transient are shown by the change in steepness of the relationships in the left, middle and right columns of panels A, B and C. For peak sodium, we computed the slope by calculating the change from the condition without drug to the point with the highest drug concentration to estimate the rate of change over the therapeutic plasma range. As expected, changes due to the reduction in peak INa is minimal as the effect of flecainide on INa is always included. These results suggest that the effect of flecainide to result in a negative inotropic outcome results in part from direct effects of the drug on proteins key for Ca2+ handling, consistent with earlier reports [38]. Both effects of flecainide on the L-type Ca2+ current (IcaL) and on the IRyR contribute to negative inotropy.

We also investigated the effect of disruption to ion homeostasis under the conditions of rapid pacing in the presence of flecainide. As expected, shown in Fig. 6, simulations from ventricular myocytes indicate that the reduction in peak INa resulting from Na+ channel block has the effect to reduce intracellular [Na+] by ~1 mM.

Fig. 6.

Grandi-Bers Human ventricular model: Ionic mechanisms of flecainide effects for 10 mins simulated time at 150 BPM with 2.0 μM flecainide (green) for (A) effects of flecainide on 5 targets, (B) effects of flecainide on 4 targets and (C) effects of flecainide on 3 targets on ICaL, [Ca2+]i, [Ca2+]SR and [Na+]i.

To determine whether the reduction in intracellular [Na+] contributes to flecainide-induced widening of simulated QRS and reduction in the Ca2+ transient, we paced the cells at 150 BPM within the therapeutic plasma range of 0.5 to 2.5 μM (red dots) with resulting reduced Ca2+ and compared this setting to the condition where intracellular sodium concentration was fixed at 9.7 mM (black asterisks). The results are shown in Fig. 7. There were minimal differences in the predicted effects of flecainide on 5 targets, (B) effects of flecainide on 4 targets and (C) effects of flecainide on 3 targets on the peak INa, peak Ca2+ transient and integral of the Ca2+ transient reduction from left to right. These simulations suggest that the reduction in intracellular Na+ has a minimal effect to promote negative inotropy. Not surprisingly, the ~1 mM reduction in Na+ most affects the amplitude of the Na+ current and consequently the QRS interval, the effect is very small.

Fig. 7.

Flecainide-induced widening of simulated QRS from human ventricular computational tissue and reduction in Ca2+ transient and INa at 150 BPM within the therapeutic plasma range of 0.5 to 2.5 μM (red dots) compared with intercellular sodium concentrations fixed at 9.7 mM (black asterisks) for (A) effects of flecainide on 5 targets, (B) effects of flecainide on 4 targets and (C) effects of flecainide on 3 targets.

Here, we opted to utilize the Grandi-Bers computational model of the cardiac ventricular action potential because it has a well-characterized representation of [Ca2+]i dynamics and homeostasis. However, to ensure that our findings were not dependent on the Grandi-Bers model, we also carried out simulations using the O’Hara-Rudy model, again focusing on assessing the relationship between flecainide-induced widening of simulated QRS and reduction in peak INa, peak Ca2+ transient, and integral of the Ca2+ transient at 150 BPM from drug free to 2.5 μM flecainide (Supplemental Fig. S2). The predictions generated by the O’Hara-Rudy model at drug doses greater than 1 μM were qualitatively similar to the predictions in the Grandi-Bers model.

It is important to note that while the modeling and simulation results in the prediction of the negative inotropy as assessed by a reduction in intracellular calcium in ventricular myocyte simulations, the effects predicted may constitute an underestimate. This is because additional hemodynamic effects are to be expected and likely exacerbated in disease states like atrial fibrillation [14,39-43].

We tested this in model simulations. We applied the flecainide effects on 5 ion current targets in simulated Grandi-Bers atrial cells INa ([26]), IRyR ([27]), IKr (IC50 = 1.5 μM [28]), Ito (IC50 = 15.2 μM [29]) and IcaL(IC50 = 27.1 μM [1]). We tested the rate-dependence and dose-dependence of the effect of flecainide (0.5 and 1.0 μM) on the P wave duration and on the peak intracellular Ca2+ transient (Fig. 8A), and the integral of the Ca2+ transient (Fig. 8B), as a proxy for atrial inotropy. The flecainide effect on all targets results in an increase in the P wave duration a surrogate for the slowing of conduction in atrial tissue. However, when the effect of flecainide on the L-type Ca2+ current (IcaL) were eliminated, but the effect on the other four targets remained, the negative inotropic effect of flecainide persisted, but the AP upstroke was enhanced and the APD was prolonged with restoration of the Ca2+ current (Fig. 8C and D).

Fig. 8.

Correlation between flecainide-induced widening of P wave in human atrial computational tissue and reduction in the peak intracellular Ca2+ transient (blue), and the integral of the Ca2+ transient (red) at 80 and 100 BPM with flecainide 0.5 and 1.0 μM. (A) and (B) Simulated effects of flecainide on 5 targets: INa, IKr, Ito, IRyR, ICaL). (C) and (D) Simulated effects of flecainide on 4 targets: INa, IKr, Ito, IRyR). Notice that the * indicates that conduction block (CB) occurs at the indicated concentration.

Discussion

In the study, developed and applied a human cardiac action potential modeling and simulation approach to reveal the mechanisms of negative inotropic affects that have been reported following application of flecainide and other similar sodium channel blockers. A reduction in the capacity of the heart to contract optimally after Na channel block has been described in a variety of species [8,38,44,45]. Understanding the ionic mechanisms of flecainide induced negative inotropy, may allow for the development of new or modified therapeutic approaches for diseases like atrial fibrillation that maintain the beneficial effects of flecainide as an antiarrhythmic drug target to normalize atrial dysrhythmia [46,14].

Indeed, conduction disturbances and mechanical dysfunction are aspects of safety pharmacology that have received relatively little attention. Most drug safety assessments associated with block of repolarization currents and consequent repolarization abnormalities. Even though more recent investigations into repolarization related safety pharmacology and its hallmark, the prolongation of the QT interval, have taken into account multi-channel block by drugs, there has not been a defined assay to consider the effects of depolarization abnormalities and mechanical dysfunction arising from ion channel block.

This important area of translational research is ripe for renewed investigation [6,37,10]. By assessing and revealing the fundamental ionic mechanisms associated with electrophysiological effects of flecainide and the resulting mechanical alterations, insights into promising development of new therapeutic interventions, or combinations of existing drugs used in appropriate doses may allow for minimization of unwanted side effects [47,48]. This may be critical for optimization of new methods for managing prevalent cardiac diseases like atrial fibrillation. Flecainide has been shown to be effective to convert atrial fibrillation to sinus rhythm. The most likely mechanisms for the antiarrhythmic effect is slowing of conduction and concomitant rate-dependent shortening of atrial refractoriness that sufficiently impact the fibrillatory wavelength to extinguish the arrhythmia [49].

A variety of mechanisms have been proposed as primary drivers of the negative inotropic effect of flecainide [38,50,51]. However there have been no experimental studies that systematically considered all plausible mechanisms and their individual contributions to negative inotropy. Moreover, these studies largely utilize drug concentrations that are not therapeutically relevant and so the application of findings to interpret clinical mechanisms is limited. Our findings bridge this gap by taking advantages of the strengths of computational modeling and simulation approaches that are well suited for such investigations. Modeling and simulation can also take into account the therapeutically relevant concentration of flecainide in order to establish the clinical relevance of our findings [52,53]. Moreover, modeling and simulation is uniquely suited to consider pharmacodynamic and kinetics, multi-channel effects and to assess the contributions of drug effects on individual targets to the overall drug efficacy [54-57].

The effects of flecainide on the L-type Ca2+ current [38,50] have been reported in many previous studies and this effect has been suggested as the likely mechanism for the negative inotropy promoted by flecainide. In the study by Kihara et al., the authors examined the differences in negative inotropy induced by flecainide and pilsicainide, both of which are classified as 1C anti-arrhythmic drugs. These two drugs block sodium channels similarly, but flecainide has more potent block of the L-type calcium channel. Indeed, the authors reported that with 10 μM application of flecainide, there was ~25% reduction in peak L-type Ca2+ current when measured in dog ventricular trabeculae [38]. They attributed the larger effect of flecainide to reduce inotropy as resulting from substantial block of the L-type Ca2+ current. The authors postulate that the reduced L-type Ca2+ current then reduced the calcium transient amplitude.

Other studies have reported flecainide to be less potent for inhibition of the L-type calcium current [1]. In the present study, we adopted a conservative approach and selected an IC50 = 27.1 μM [1] for flecainide inhibition of the L-type calcium current. Even the highest doses of clinically relevant flecainide are considerably lower (by an order of magnitude) and so any anticipated effect on the L-type calcium channel is likely to be very small. However, we did observe some rate dependent inhibitory effects presumably due to the build-up of flecainide in the L-type calcium channel and the related rate dependent reduction in the upstroke velocity of the action potential, due to use-dependent Na+ channel block.

It is important to note that the effect of flecainide on the L-type calcium channel measured “in isolation” does not reflect what we observed in a more physiological setting. The effect of flecainide to reduce the upstroke velocity of the action potential results in a slowing of the depolarizing phase of the membrane potential (Fig. 4A). What this means practically is that when sodium channels are blocked the membrane voltage remains in a range that is favorable for calcium channel activation. Indeed, this effect in single simulated myocytes (shown in Supplemental Fig. S1) actually results in an increase in calcium current at fast heart rates. However, in simulated tissue under the same conditions assessed in the single simulated myocytes, the model simulations predicted the opposite. As shown in Fig. 4, the predictions in a model myocyte within a coupled tissue environment suggest a substantial effect of flecainide to reduce the peak of L-type calcium current. Part of the effect results from slow conduction that allows the calcium channels to begin the inactivation process during slow depolarization.

Other studies have suggested the importance of the Na/Ca exchanger in promoting the negative inotropic effect of flecainide [51]. Ito and coauthors reported that inhibition of Na+ channels can result in changes to the Na+/ Ca2+ exchange current and ultimately leads to a decrease in the Ca2+ content in both the sarcoplasmic reticulum (SR) and Ca2+ entry through the exchanger. Fig. 6 shows the results of our simulations of these effects. Indeed, as suggested by Ito et al., we identified a substantial impact of Na+ channel inhibition by flecainide to reduce intracellular [Na+] levels. We therefore tested the effect of preventing the reduction in intracellular [Na+] by clamping it to maintain drug free levels as shown in Fig. 7. The graphs in Fig. 7 indicate a small effect of the change in intracellular [Na+] that affects the magnitude of the Na+/ Ca2+ exchange current. Meme et al., [58], suggest under conditions of reduced intracellular [Na+], Ca2+ entry into the myocyte is likely to occur during diastole Na+/ Ca2+ exchange current rather than in systole via the L-type Ca2+ current. In our study, we primarily examined the effects of flecainide to reduce intracellular [Na+] during rapid pacing. In this setting the diastolic interval is short and does not allow time for loading via during diastole Na+/ Ca2+ exchange.

In summary, our human action potential and intracellular Ca2+ model simulations predict an interaction of multiple mechanisms that combine to result in the flecainide mediated reduction in cardiac contractile performance. In single isolated simulated myocytes, the reduction in INa acts to slow the upstroke velocity of the action potential and this can favor an increase in L-type Ca2+ current during the depolarizing phase. However, the modest increase in L-type Ca2+ current is offset by a reduction in outward NCX current that brings in notably less Ca2+ during the early phases of the action potential. Direct block of the RyR and L-type Ca2+ current by flecainide results in a reduction in both of these currents relative to the drug free condition. The net effect is a reduction in the peak and the integral of the Ca2+ transient which in turn results a reduction in inotropy. In the tissue level simulations, conduction of the depolarizing excitatory wave is hampered by flecainide block of INa that is exacerbated by electronic effects. In the tissue setting, simulated cells in the tissue exhibited a reduction in peak L-type Ca2+ current because the action potential upstroke was slow enough to allow channel inactivation. This effect is added to the reduction in outward NCX current and small amount of block of RyR by flecainide. Together, these effects lead to a substantial reduction in the predicted peak and the integral of the Ca2+ transient (Fig. 4).

We further applied a modeling and simulation approach and examined the relationship between flecainide application and the Ca2+ transient in computational atrial myocytes. Because of the different morphology of the atrial action potential, the dose and rate dependent reduction in the atrial Ca2+ transient following application of flecainide was more profound than in the ventricular simulations. The effect of the reduced capacity to carry out mechanical “squeeze” by the atria in the presence of flecainide would be expected to reduce ventricular filling and preload and consequently reduce dP/dtmax [59-61].

There are several limitations present in our study. Here, we tracked the peak of the Ca2+ transient (CaT) that has been linked to maximum contractile force [62], and the integral of the CaT as a more comprehensive indicator of the time course of the force of contraction over the entire cardiac cycle [63,64]. Clearly, the myofilament calcium sensitivity is a critical parameter that can affect the specific correlation of the relationship between CaT and contractile force. A limitation of our current model is that we do not explicitly account for the myofilament calcium sensitivity although it could be extended in a future state to include this and other features relevant to disease states [65]. Moreover, we do not explicitly represent stochastic phenomena such as calcium sparks that have shown to be inhibited by flecainide [66,67], although we and others have developed detailed models based on these data previously [27,68,69].

There are notable differenced in the results from the Grandi-Bers (GB) and O’Hara-Rudy (ORd) model predictions of the dose-dependence of flecainide effects on inotropy. Unlike the results predicted in Grandi-Bers, where we observed a dose-dependent reduction in inotropy, over the full range of simulated dose response in ORd, we observed a biphasic response. Whereas lower doses of flecainide caused an increase in the peak CaT in the ORd model, higher drug doses (1.5 μM and above) resulted in a decrease in contractility related parameters as shown in Supplemental Fig. S2. Indeed, the model described by Tomek et al. in the ToR-ORd model was developed in part to address the observed increase in Ca2+ in response to Na+ channel block [70].

In addition to the differences in the representation of the Na+ current in earlier models, the GB and ORd models have different approaches to of Ca2+ handling and incorporate different local subspaces. The GB model predicts separate Na+ and Ca2+ concentrations in the junctional cleft, sub-sarcolemma (SL) and cytosol as well as accounting for the concentration of Ca2+ in the sarcoplasmic reticulum (SR). However, the ORd model included two compartments representing separate cytosol and junctional cleft concentrations for Na+ and Ca2+, along with junctional and network SR Ca2+ concentrations. ICaL is also distributed differently in the GB model, with 10% of channels in the SL and 90% in the junctional cleft (in close proximity to the T-tubules and RyR Ca2+ release channels). In the ORd model, the ICaL was located only in the junctional cleft, and INaCa was distributed as 20% in the junctional cleft and 80% in cytosol. In addition, in the GB model, the ICaL peak was broader than in the ORd model, and the decay rate of CaT was slower in the Ord model compared to GB model.

The ToR-ORd model [70] includes additional modifications and represents an updated version that includes ICaL, INa, Ikr (replaced by a Markov model), and Ik1. ICaL was particularly revised in terms of its driving force, activation curve, and location – 20% of ICaL was placed in cytosolic space. The calcium-sensitive chloride current and background chloride current from GB model were also added into the ToR-ORd model. Other changes in ionic conductance parameters included Ito, IkKs, INaCa, INaK, IKb, INab, ICab, IpCa, Jrel, and Jup.

We recognize that all models have strengths and weaknesses and that models continue to undergo continuous development. Indeed, a potential perceived weakness of the model used to represent the interaction of flecainide with the Na+ channel in this study is complex. However, despite having recently constructed a reduced simplified Hodgkin-Huxley model of lidocaine interaction with the Na current where we show that we can reproduce the basic properties of lidocaine block (arXiv:2102.02342), we were not able to similarly develop a reduced model of the Na+ channel interaction with flecainide. We were unable to reduce the model and still reproduce the slow complex time course of recovery from block and the voltage dependence of use-dependent block, which we have shown in earlier studies is the signature of flecainide interaction with the Na+ channel [26,27]. In other words, we know that a simple reduction of the conductance of the Na+ channel will not yield realistic block of the Na+ current. In the case of flecainide, the rapid gating kinetics of the Na+ channel and the slow drug kinetics of flecainide create the complex dynamics of flecainide block of Na+ current.

Future studies could consider the impact of additional disease states, which may further complicate the interpretation of safety of Na+ channel blockers as effective antiarrhythmic agents [9,14,39-43]. Indeed, the study by Gao et al. revealed compromised contractility in heart failure, but also identified an altered response to pharmacological manipulation by a variety of agents, including flecainide. In de Antonio et al. the importance of rate-dependence of disease can allow for selective atrial versus ventricular drug efficacy, a promising approach in atrial fibrillation [41]. A limitation of our results presented in this study is that they do not include predictions in the diseased heart. Future studies will be carried out to allow for study of disease states.

IC50 values of drug to inhibit currents
Currents	IC50	Reference
IRyR	20 μM	[27]
IKr	1.5 μM	[28]
Ito	15.2 μM	[29]
ICaL	27.1 μM	[1]