Exploring the Mechanisms of Sex-Specific Proarrhythmia in Long QT Syndrome through Computational Modeling

Abstract

Females exhibit longer QT intervals and a higher risk of Long QT Syndrome (LQTS) associated arrhythmogenesis compared to males. While several studies suggest these sex disparities result from the effect of sex hormones on cardiac ion channels, the underlying mechanisms remain incompletely understood. This research investigates the arrhythmogenic effects, sex-specific risk, and mechanisms associated with LQTS linked to either to loss-of-function of the rapidly activating delayed rectifier K⁺ current (I_Kr), or gain-of-function of the L-type Ca²⁺ current (I_CaL). We primarily utilized the Tomek-Rodriguez (ToR-ORd) model of human ventricular cardiomyocytes and incorporated sex-specific parameterizations based on previous studies. The O’Hara-Rudy and Grandi-Bers models were used to demonstrate model-independence of the findings. We employed a populations-of-models approach to assess early afterdepolarization (EAD) susceptibility in control and LQTS male and female groups. All female models had consistently longer action potentials and were more prone to EADs than male models. In the ToR-ORd model, I_Kr loss-of-function led to EADs in 65.8% of females vs. 22.8% of males. I_CaL gain-of-function led to EADs in 66.2% of females but only 3.6% of males. Using logistic regression analysis, we identified key ionic predictors of EAD susceptibility, with G_CaL and Na⁺/Ca²⁺ exchanger (G_NCX) consistently emerging as positively and G_Kr as negatively associated to EADs across both sexes and LQTS types. Notably, higher G_NCX but lower G_Kr in female vs. male cardiomyocytes could explain heightened female EAD risk. Our studies explore the ionic traits that favor (or confer resilience against) EADs with potential implications for personalized treatments.

NEW & NOTEWORTHY

We explored sex disparities in Long QT Syndrome (LQTS) using sex-specific human ventricular cardiomyocyte models. We showed that females exhibit greater susceptibility to early afterdepolarizations (EADs) than males, and identified key ionic predictors of EAD risk, including increases in the voltage-gated L-type Ca²⁺ current and electrogenic Na⁺/Ca²⁺ exchanger, and downregulation of the rapidly activating delayed rectifier K⁺ current. These findings offer new insights into sex-specific mechanisms underlying arrhythmogenesis in LQTS, with potential implications for personalized treatments.

Graphical Abstract:

INTRODUCTION

Long QT syndrome (LQTS) is a group of disorders (caused by inherited and acquired conditions) that affect cardiac repolarization leading to a prolonged QT interval on the electrocardiogram. Inherited LQTS is caused by mutations in genes encoding ion channels in the heart that are essential for maintaining transmembrane potential stability (1), such as those carrying the voltage-gated K⁺ (KCNH2 and KCNQ1), Na⁺ (SCN5A), and Ca²⁺ (CACNA1C) currents (2). The inherited condition affects approximately 1 in 2,000 people and significantly increases the risk of sudden cardiac death, particularly in young, otherwise healthy individuals (1). Triggers such as physical exertion or emotional stress can further exacerbate the risk of life-threatening arrhythmias like torsades de pointes (1, 3, 4). Acquired LQTS can be caused by certain medications, including some antiarrhythmic drugs, antibiotics, antipsychotics, and antidepressants, that prolong the QT interval (3, 5). Some medical conditions, such as bradycardia or electrolyte abnormalities, can also cause LQTS (6). Prolonged cardiac repolarization can lead to an abnormal reactivation of depolarizing ion currents, especially through Ca²⁺ channels, before the action potential (AP) has fully repolarized. This abnormal activity can cause a secondary depolarization, or early afterdepolarization (EAD), which can favor the formation of premature heart beats and disrupt the normal sequence of generation and propagation of the electrical activation of the heart. Notably, female hearts have a naturally longer QT interval compared to male hearts (7–9). This is mainly attributed to a reduced gene expression for various K⁺ channel subunits (including KCNH2) in female vs. male ventricles (10). The prolonged corrected QT duration increases female susceptibility to LQTS, torsades de pointes, and cardiac mortality (11, 12). While men generally exhibit higher cardiac mortality than women in both the general population and among individuals with cardiovascular disease (13), and sudden cardiac death is less common in women (14), female sex is recognized as a risk factor for the development of malignant arrhythmias associated with both congenital and drug-induced LQTS. Indeed, women are more susceptible than men to developing torsades de pointes during administration of cardiovascular drugs that prolong cardiac repolarization. For example, Lehmann et al. reported that TdP occurred in 4.1% of females compared to 1.9% of males in a cohort over 3,100 patients treated with d,I-stotalol (11) and Makkar et al. found that females comprised of 70% of 332 reported cases of drug-induced TdP (15). As for congenital LQTS, a clinical study of over 1,100 inherited LQT2 patients revealed that women had a 26% chance of experiencing life-threatening cardiac events, compared to 14% in men (16). Locati et al. demonstrated that in congenital LQTS males have a higher risk of cardiac events until puberty, but this risk shifts to females during adulthood, highlighting the influence of both sex and age (17). However, the mechanisms that underlie these sex disparities are not thoroughly understood, partially owing to the underrepresentation of females in fundamental and clinical research studies (18, 19). Biophysically detailed models of cardiac excitation-contraction coupling can complement and accelerate experimental investigation, by providing a mechanistic framework to incorporate the established sex differences derived from experimental data (20), and quantitative predictive tools to unravel the underpinnings of cardiac arrhythmia susceptibility (21, 22).

In this study, we leverage established human specific models to investigate the arrhythmogenic effects and sex-specific risks and mechanisms associated with two forms of LQTS, linked to either a deficit in repolarization reserve (loss-of-function the rapidly activating delayed rectifier K⁺ current, I_Kr) or to excessive depolarization reserve (gain-of-function of the L-type Ca²⁺ channel, I_CaL). By examining these specific subtypes, we aim to uncover the sex-specific mechanisms contributing to the development of EADs in cardiac myocytes and the associated arrhythmic risk in females vs. males.

METHODS

Human Ventricular Cardiomyocyte Models

To explore the sex-specific proarrhythmia mechanisms of LQTS in ventricular electrophysiology, we used the Tomek-Rodriguez (ToR-ORd) model (23) as our primary framework due to its recent development and comprehensive nature. The baseline endocardium version of the model, without alterations, was assumed to represent the male physiology. To create the female version, the baseline model was updated to incorporate sex-specific relative changes in ionic currents and transporters, as reported in experimental data and previously implemented in the O’Hara-Rudy (ORd) model (4, 22, 24). All adjustments (summarized in Table 1) were implemented as scaling factors for ionic conductances, except for I_Ks, where a scaling factor was introduced in the female maximal conductance and in the voltage dependence of steady-state activation (x_s1ss, x_s2ss) and activation time constant (τ_xs1, τ_xs2), as done in (22) and described in the Supplementary Materials (equations 1–2). The conductance scaling is based on the original sex-specific parameterization by Yang and Clancy (22), which was informed by a high-throughput quantitative approach by Gaborit et al. (20) that revealed genome-scale sex differences in the expression of key genes encoding cardiac ion channels and transporter subunits in human epicardial and endocardial tissues. Additionally, we incorporated recent updates from Fogli Iseppe et al. (21), including an increase in G_NCX and the removal of sex differences in sarcoplasmic Reticulum Ca²⁺-ATPase (SERCA) and Na⁺/K⁺-ATPase (NKA) formulations, to align with experimental measurements from Park et al. (25) and Papp et al. (26). In this study, we only considered these “chronic” sex-differences in ion channel expression and did not incorporate the acute effects of sex hormones. To mitigate any potential model-specific biases and ensure the universality of our findings, we also employed the ORd model (24) and the Grandi-Bers model (GB) (27). As in the ToR-ORd model, the baseline versions were assumed to represent male physiology. These models were then updated with the same sex-specific parameterizations to maintain consistency across the models. To the same goal, since the GB model did not include a late Na⁺ current (I_NaL) component, we added the I_NaL formulation from the ORd model (24) and confirmed that I_NaL contribution to the AP duration (APD) in the GB model is similar to that in the ORd model (e.g. ~9% APD shortening when I_NaL is fully inhibited compared to baseline). This standardization allowed us to focus on model-independent mechanisms while still benefiting from each model’s unique strengths. For each of the three models, simulations were conducted at a pacing rate of 1 Hz and key electrophysiological parameters such as APD, Ca²⁺ transient (CaT) amplitude, and diastolic Ca²⁺ were measured at steady-state after 200 beats. This baseline was essential to accurately assess and compare the models’ responses when conditions were altered to induce EADs.

Table 1.

Female vs. male parameters reflecting sex differences in ion channel expression and/or function in the endocardium cell type.

Current/Transporter	Definition	F vs M	Female Scaling Factor	Reference
GpCa	Maximal conductance of the plasmalemmal Ca2+ pump current	↑	1.6	Gaborit et al., 2010
CMDNmax	Maximal buffering capacity of calmodulin	↑	1.21	Gaborit et al., 2010
Gtos	Maximal conductance of the slowly inactivating transient outward K+ current	↓	0.64	Gaborit et al., 2010
GKr	Maximal conductance of the rapidly activating delayed rectifier K+ current	↓	0.8	Gaborit et al., 2010
GKs	Maximal conductance of the slowly activating delayed rectifier K+ current	↓	0.83	Gaborit et al., 2010
GK1	Maximal conductance of the inward rectifier K+ current	↓	0.86	Gaborit et al., 2010
GNCX	Maximal transport rate of the Na+/Ca2+ exchanger	↑	1.15	Parks et al., 2013; Papp et al., 2017

Long QT/EAD Simulations

In this study, we simulated both a loss-of-function in I_Kr and a gain-of-function in I_CaL to determine whether the mechanisms leading to sex differences in arrhythmia susceptibility are consistent across both types of alterations. To simulate the I_Kr loss-of-function, we reduced the maximal conductance (G_Kr). For each ventricular cell model, we chose the scaling factor for G_Kr through incremental testing across a range of basic cycle lengths (BCLs) from 1,000 to 3,000 ms. This process involved systematically evaluating a series of scaling factors in both male and female models to identify the reduction that would result in the emergence of EADs in either model at the fastest pacing rate (shortest BCL). Through this process, we identified that 90% reduction in G_Kr at a BCL of 3,000 ms successfully induced EADs in the female ToR-ORd model, with no EADs seen in the male model under the same conditions. Similarly, an 80% reduction in G_Kr induced EADs in the female (but not male) ORd model at the same BCL. In the GB model, G_Kr reduction did not produce EADs in either the male or female models.

I_CaL gain-of-function was simulated by increasing the maximal conductance (G_CaL) and shifting the steady-state activation toward negative potentials. Through incremental testing across a range of BCLs from 500 to 2,000 ms, we found that in the ToR-ORd model a 45% increase in G_CaL combined was a −6 mV shift in the voltage-dependence of steady-state activation resulted in the development of EADs in the female but not the male model at a BCL of 2,000 ms. In the GB model, a 5% increase in G_CaL with a −3 mV shift was sufficient in creating an EAD in the female parameterization at the same pacing rate, while no EADs were observed in the male for the same perturbation. In the ORd, I_CaL changes led to AP prolongation and failure of repolarization with drastic parameter modifications, but no EADs were detected.

Using these scaling factors, we simulated the mutations in the baseline male and female versions of each respective model. We then calculated key electrophysiological biomarkers, including APD measure at 90% (APD₉₀) and 50% (APD₅₀) repolarization, diastolic Ca²⁺ concentration, CaT amplitude, CaT duration (CaD₅₀), and sarcoplasmic reticulum (SR) Ca²⁺ content. These measurements provided a comprehensive dataset that enabled a detailed comparison of the mutation effects and sex-specific differences, allowing us to identify distinct patterns of arrhythmogenic risk across sexes and models.

Populations of Human Ventricular Cardiomyocyte Models

We employed a “population of models” approach combined with multivariable linear and logistic regression analyses (28–31) to study the parameter sensitivity of AP and CaT characteristics and correlate the EAD risk to underlying ionic traits. To simulate intrinsic physiological variability, we generated populations of 500 cell model variants for each sex and model lineage by randomly varying key parameters in the models, such as ion channel conductances and maximal ion transport rates, as described by Dr. Sobie (32). Baseline parameter values were adjusted by multiplying them with log-normally distributed scale factors, calculated as $scaling=exp({\sigma }_G\left(i\right)·randn\left(\mathrm{1,500}\right))$, where ${\sigma }_G=0.1$ is the standard deviation and randn is a MATLAB function that generates normally distributed pseudorandom numbers. This results in 95% of the samples expressing scale factors ranging between 82% and 122%. Population size was determined as previously described by Dr. Sobie and our group (32–34). The definitions of the parameters from the ToR-ORd model used to build the model populations are presented in Table 2. We applied multivariable linear regression analysis using the nonlinear iterative partial least squares (PLS) algorithm (29, 32), to correlate changes in model parameters in the baseline models to the resulting effect on APD₉₀, CaT amplitude, and CaT duration. We then simulated the I_Kr loss-of-function and I_CaL gain-of-function effects in each variant of the populations. Each model variant was tested for the presence or absence of EADs, with outcomes recorded in binary form—1 for the presence of EADs (identified as a secondary peak during the repolarization phase of the AP) and 0 for absence of EADs. This categorization allowed us to perform logistic regression analysis, carried out using MATLAB’s mnrfit function, to correlate the changes in model parameters with the probability of EAD (28).

Table 2.

Definition of the ToR-ORd model parameters that randomly perturbed to introduce variability in the population of models.

Parameter	Definition
GNa	Maximal conductance of the fast Na+ current
GCaL	Maximal conductance of the L-type Ca2+ current
Gto	Maximal conductance of the transient outward K+ current
GNaL	Maximal conductance of the late Na+ current
GKr	Maximal conductance of the rapidly activating delayed rectifier K+ current
GKs	Maximal conductance of the slowly activating delayed rectifier K+ current
GK1	Maximal conductance of the inward rectifier K+ current
GKb	Maximal conductance of the background K+ current
GNCX	Maximal transport rate of the Na+/Ca2+ exchanger
GNaK	Maximal transport rate of the Na+/K+ pump
GNaB	Maximal conductance of the background Na+ current
GCaB	Maximal conductance of the background Ca2+ current
GpCa	Maximal transport rate of the plasmalemmal Ca2+ pump
GClCa	Maximal conductance of the Ca2+ dependent Cl− current
GClB	Maximal conductance of the background Cl− current
Jrel	Maximal transport rate of SR Ca2+ release via ryanodine receptors (RyRs)
Jup	Maximal transport rate of the sarcoplasmic reticulum (SR) Ca2+ ATPase
Jleak	Maximal transport rate of SR Ca2+ leak via RyRs

Numerical Methods and Code

Simulations of the updated baseline models (ToR-ORd, GB, and ORd models) were performed with a standard desktop using MATLAB (MathWorks, Natick, MA), v. R2023b. Population simulations were performed using MATLAB R2023b on a system running Ubuntu 22.04.4 LTS (64-bit) OS, equipped with an Intel^® Core^™ i7–8700 CPU @ 3.20 GHz with 12 CPUs (24 threads). Data analysis was performed with MATLAB using a standard desktop. Model codes are freely available on the authors’ Github page: https://github.com/drgrandilab.

RESULTS

In this study, we employed the ToR-ORd model to investigate sex-specific differences in APD and Ca²⁺ handling in cardiomyocytes in the context of loss of I_Kr function and gain of I_CaL function mimicking LQT2 and LQT8 respectively. When parameterizing the ToR-ORd to include genomic and functional differences across sex in ion channels and transporters, we found that APD₉₀ and APD₅₀ were significantly elongated in female cardiomyocytes compared to their male counterparts across all pacing rates (Fig. 1A,B, top), which are consistent with QT interval prolongation in females (7–9). These results align with the Yang and Clancy model (22), which demonstrated similar sex-dependent APD differences (21). The female model exhibited smaller CaT amplitude, as shown in rodents (35). The diastolic Ca²⁺ concentration (Fig. 1B, middle) and the SR Ca²⁺ content (Fig. 1B, bottom) were similar in female and male cardiomyocytes, whereas the time to 50% CaT decay was ~12% longer in the female model (vs. male). These Ca²⁺ handling sex differences were observed across the full range of pacing cycle lengths examined (500 to 4,000 ms). Overall, these modest sex-dependent differences were comparable to experimental data from patients, which reported no significant male vs. female differences in any of the Ca²⁺ measurements (21, 36–38). The elongation of the APD and smaller CaTs in female cardiomyocytes were consistently replicated across both the ORd and GB models (Fig. S1–S2).

Figure 1. Sex-specific parameterization of the ToR-ORd biophysical model of the human cardiac action potential and Ca²⁺ handling.

A: Simulated transmembrane potential and cytosolic CaT traces upon pacing at 1,000ms BCL in the male (red) and female (blue) ventricular endocardial cardiomyocyte models. B: Pacing rate dependence of simulated biomarkers: APD₉₀, APD₅₀, diastolic Ca²⁺ concentration, CaT amplitude, CaD₅₀, and SR Ca²⁺ content.

We next sought to understand how loss-of-function in I_Kr could affect the sex differences in ventricular APs and CaTs. The loss-of-function in I_Kr in the ToR-ORd model exhibited a significant prolongation of APD (Fig. 2), which was exacerbated at slow pacing rates. Simulations of I_Kr loss-of-function revealed sex-specific susceptibilities to EAD formation. In the ToR-ORd model, a 90% reduction in G_Kr at a BCL of 3,000 ms successfully induced EADs in the female model, while the same reduction in the male model did not produce EADs under the same conditions (Fig. 2A,B). The ORd model showed a similar trend, with an 80% reduction in G_Kr leading to EADs in the female model at 3,000 ms BCL, whereas the male model did not exhibit EADs under the same reduction (Fig. S3). The GB model did not develop EADs under I_Kr loss-of-function conditions (not shown). In addition, in both ToR-ORd and ORd models the loss-of-function of I_Kr elongated the CaT duration and modestly reduced diastolic Ca²⁺, CaT amplitude and SR Ca²⁺ content (Fig. 1B, Fig. S3B).

Figure 2. Simulation of I_Kr loss-of-function in male and female ToR-Ord models.

A: Simulated transmembrane potential and cytosolic Ca²⁺ concentration traces upon pacing at 3,000 ms BCL of female (blue) and male (red) in the presence and absence of I_Kr loss-of-function (LoF) (corresponding to a 90% decrease in G_Kr, dashed lines). EADs are seen in the female LQTS model at 3,000 and 4,000ms BCLs. B: Pacing rate dependence of simulated biomarkers: APD₉₀, APD₅₀, diastolic Ca²⁺ concentration, CaT amplitude, CaD₅₀, and SR Ca²⁺ content.

We used a “population of models” approach (28, 29, 32), which involved generating populations of ventricular myocytes for both male and female models by randomly varying the maximal conductances and transporter rates. This approach allowed us to capture the observed cell-to-cell variability in electrophysiological properties. The simulations revealed a pronounced sex-specific difference in the susceptibility to EAD formation under I_Kr loss-of-function conditions. In the ToR-ORd model, a 90% reduction in G_Kr combined with a long cycle length led to the development of EADs in 65.8% of the female population, whereas only 22.8% of the male population exhibited EADs under the same conditions. In the ORd model we saw an even greater susceptibility in the females with a 91.4% EAD occurrence and only a 16.0% occurrence in the male. To quantify the impact of variations in ion channel conductances and transport rates on the development of EADs in human ventricular myocytes, we performed a logistic regression analysis. The regression coefficients obtained from this analysis quantify the contribution of each parameter to the likelihood of EAD development in both sexes. A positive (negative) coefficient indicates that an increase in the associated model parameter correlates with a higher (lower) probability of EAD occurrence. In the ToR-ORd model, logistic regression analysis yielded consistent results across sexes: G_CaL, G_NCX, G_NaL, and G_CaB had the highest positive coefficients, indicating that an increase in these parameters is most strongly associated with an increase in the likelihood of EADs. Conversely, G_Kb, J_Up, G_Kr, and G_Ks were identified as the parameters exerting the strongest negative influence on EAD occurrence (Fig. 3C), suggesting that their increase may offer the greatest protection against EAD formation. Similar results were observed in the ORd model populations (Fig. S4C).

Figure 3. Male and female populations of ToR-Ord models with I_Kr loss-of-function.

A: Representative traces of transmembrane potential and cytosolic Ca²⁺ concentration for 50 female (EAD+ dark blue, EAD− light blue) and 50 male (EAD+ dark red, EAD− light red) model variants extracted from sex-specific populations of 500 models upon pacing at a BCL of 3,000ms. B: Distribution of EAD+ APD₉₀ durations (dark blue female, dark red male) and EAD− APD₉₀ durations (light blue female, light red male) across the total population. EADs occurred in 65.8% of female and 22.8% of male models in the population. C: Results of logistic regression analysis for female (R²=0.976) and male (R²=0.999) populations quantifying the sensitivity of development of EADs to changes in the listed model parameters.

The I_CaL gain-of-function in the ToR-ORd model resulted in a markedly prolonged APD in females, particularly at slow pacing rates (Fig. 4), while males exhibited only slight prolongations. In the ToR-ORd model, a 45% increase in G_CaL combined with a −6 mV shift in the steady-state activation resulted in EAD formation only in the female model at a BCL of 2,000 ms (Fig. 4A). In the GB model, EADs were induced in the female model with a 5% increase in G_CaL and a −3 mV shift. The male model did not show EADs under the same conditions, although the APD was markedly prolonged (Fig. S5). I_CaL gain-of-function in the ORd model led to AP repolarization failure but did not produce EADs (not shown). The gain-of-function in I_CaL also resulted in increased diastolic Ca²⁺ concentration, CaT amplitude, and SR Ca²⁺ content compared to baseline values (Fig. 4C).

Figure 4. Simulation of I_CaL gain-of-function in male and female ToR-ORd models.

A: Simulated transmembrane potential and cytosolic Ca²⁺ concentration traces upon pacing at 2,000ms BCL of female (blue) and male (red) in the presence and absence of I_CaL gain-of-function (GoF) (corresponding to a 45% increase in G_CaL and −6mV shift in the steady-state activation, dashed lines). EADs are seen in female LQTS model at BCL of 2,000, 3,000, and 4,000 ms. B: Pacing rate dependence of simulated biomarkers: APD₉₀, APD₅₀, diastolic Ca²⁺ concentration, CaT amplitude, CaD₅₀, and SR Ca²⁺ content.

Using the same “population of models” approach as detailed above, we were able to test sex-specific differences in susceptibility to EAD formation under I_CaL gain-of-function and observed that this susceptibility was more pronounced in females, similar to what was seen with I_Kr loss-of-function. In the ToR-ORd model, a 45% increase in G_CaL and a −6 mV shift in the steady-state activation led to the development of EADs in 66.2% of the female population. However, the male population exhibited a much lower EAD occurrence, with only 3.6% developing EADs under the same conditions. In the GB model, the susceptibility to EADs was more evenly distributed between the sexes, with 52.2% of the female population and 38.4% of the male population exhibiting EADs under the gain-of-function conditions. Due to the low sample size of male model variants displaying EADs, we did not run logistic regression analysis for the male population in the ToR-ORd model. For the female population, the logistic regression analysis identified G_CaL, G_NCX, G_NaL and G_K1, as the strongest positive coefficients, indicating that increases in these parameters significantly raise the probability of EAD formation. G_Kr, G_Kb, and J_Up were the most negative parameters emerging as the most protective factors against EAD development in females (Fig. 5C). The logistic regression analysis for the GB model revealed some similar parameters. In both sexes, the positive coefficients were G_CaL, G_NaK, and G_CaB, highlighting these parameters as key contributors to EAD formation. In female, the negative coefficients included G_Clb, G_Kr, G_pCa, and J_Leak, indicating their protective roles. In male, the negative coefficients included G_Clb, G_Kr, and a similar value contribution from G_ClCa, G_tof, J_Leak, and G_Na (Fig. S6).

Figure 5. Male and female populations of ToR-Ord models with I_CaL gain-of-function.

A: Representative traces of transmembrane potential and cytosolic Ca²⁺ concentration for 50 female (EAD+ dark blue, EAD− light blue) and 50 male (EAD+ dark red, EAD− light red) model variants extracted from sex-specific populations of 500 models upon pacing at a BCL of 2,000 ms. B: Distribution of EAD+ APD₉₀ durations (dark blue female, dark red male) and EAD− APD₉₀ durations (light blue female, light red male) across the total population. EADs occurred in 66.2% of female and 3.6% of male models in the population. C: Results of logistic regression analysis for the female (R²=0.987) population quantifying the sensitivity of development of EADs to changes in the listed model parameters. We did not run an analogous analysis with the male populations given the low sample size of the model variants displaying EADs.

DISCUSSION

This study highlights likely sex differences in the susceptibility to EADs across both loss and gain-of-function variations of LQTS. Using computational modeling, we demonstrated that females are more prone to developing EADs compared to males, regardless of whether the underlying mechanism is due to a loss or gain-of-function in the repolarizing or depolarizing ionic currents respectively. We induced broad changes in these ionic currents rather than targeting specific genetic mutations, and thus our findings are generalizable to both inherited and acquired LQTS. Using a population of models, we simulated a range of parameter combinations to examine the influence of key ionic currents on APD prolongation and the development of EADs through regression analysis. Our analysis revealed that many of the same ionic factors influence EAD susceptibility in both sexes and across LQTS types. However, the sex-specific differences in the levels of these ionic currents, particularly the reduced G_Kr and increased G_NCX in females, may amplify arrhythmia vulnerability in the female sex.

Sex-Specific Arrythmia Risk

Our findings provided evidence, based on computational modeling, of sex-specific differences in arrhythmia susceptibility in LQTS. In the baseline model populations, females exhibited prolonged APD relative to males, which is consistent with clinical data and observations reporting naturally longer QT interval in females (7–9). This prolongation of APD in females likely underpins their increased vulnerability to arrhythmias in the LQTS models and is again consistent with the previously shown sex-dependent susceptibility to LQTS (11). The simulations of LQTS also revealed a significantly higher incidence of EADs in females compared to males. In both cases, a larger percentage of the female population displayed EADs. This is consistent with in vitro observations in adult rabbit ventricular myocytes, whereby in all myocytes, the I_Kr blocker E4031 produced a marked APD prolongation, but EADs could only be observed in females (39). In the same study, perfusion of adult male rabbit hearts with E4031 led to marked APD lengthening but failed to develop torsade, whereas adult female hearts consistently developed torsade under the same conditions. Similarly, an in vivo study showed that female rabbits are more susceptible to drug-induced long QT and cardiac arrhythmias than are male rabbits (40). These findings mirror clinical observations that adult females are at greater risk for developing arrhythmias, such as torsades de pointes, particularly in the context of loss of I_Kr function (11, 17), as well as clinical studies highlighting sex-specific risk in both LQT2 and LQT8 patients (16).

Ionic Underpinnings of Sex Differences in Arrhythmogenesis

Results from logistic regression-based parameter sensitivity analysis revealed that, across both sexes and LQTS types, EAD development demonstrated similar sensitivities to many of the same ionic conductances and transporter rates. Many models exhibiting EADs in the I_Kr loss-of-function scenario also displayed EADs in the I_CaL gain-of-function case. For example, in the ToR-ORd model, 260 female models exhibited EADs with both I_Kr loss-of-function and I_CaL gain-of-function. These results highlight that shared ionic factors drive EAD development (or confer resilience against EADs) across both LQTS conditions. The top positive and negative predictors of EAD susceptibility remained consistent regardless of sex, suggesting that while overall sensitivities are shared, sex-specific differences in certain ionic currents may amplify female vulnerability to arrhythmias. Linear regression analysis also highlighted similar ionic factors influencing APD₉₀ (Figs. S7–S8), reinforcing the idea that these currents play a central role in modulating arrhythmogenic risk. For example, G_NCX was identified as a positive predictor of EAD development and APD lengthening. G_NCX plays a key role in prolonging the AP by contributing to CaT decay through Ca²⁺ extrusion and Na⁺ influx (with a 1:3 stoichiometry), which generate an inward depolarizing current. G_NCX was larger in the female vs. male model parametrization (Table 1) (41), which could exacerbate female arrhythmia risk. Similarly, G_Kr, which consistently emerged as a negative predictor of EADs, was reduced in the female vs. male parametrization (Table 1). As a major repolarizing current, I_Kr is critical in driving the phase 3 repolarization of the cardiac AP (42). Therefore, the diminished G_Kr in the female model can at least in part explain the longer APD and increased EAD propensity in females. These findings suggest that while the same ionic factors are critical for arrhythmogenesis across both sexes, the sex differences in the levels of G_NCX and G_Kr may underlie the observed sex disparities in EAD susceptibility.

The identification of key ionic predictors highlights that certain conditions may pose greater risks to individuals with specific ionic profiles. For instance, individuals with elevated I_CaL, which emerged as the primary positive predictor of EAD susceptibility, may be more vulnerable to arrhythmogenesis, especially when faced with additional stressors such as genetic factors or external influences such as medications. Notably, the increased propensity to EAD and torsade in adult female rabbit myocytes and hearts was associated with enhanced I_CaL at the base of the heart (39), which is higher in post-pubertal females than males. In fact, this difference is reversed in pre-pubertal hearts and associated with increased proarrhythmia in males in an experimental model of LQT2. Similarly, electrolyte imbalances, bradycardia, or hypokalemia—all known causes of acquired LQTS (3, 5)—could compound the likelihood of EADs and arrhythmias for individuals with altered versions of the ionic predictors.

Implication for Potential for Therapeutics Strategies

Current treatment options for LQTS are limited, with the main methods being pharmacologic nonselective β-blockers and invasive interventional approaches such as device implantation (1). Although these treatments can be effective for some patients, they have significant limitations and fail to work for many others (43). Thus, there is a pressing need for identifying novel strategies. One promising therapeutic approach for LQTS is the development of hERG activators (43). These activators could enhance the function of the repolarizing current I_Kr to attenuate APD prolongation and increased susceptibility to arrhythmias. By directly targeting this channel, hERG activators could restore the critical repolarizing current, addressing the root cause of arrhythmias in LQTS type 2, but could also reduce the risk seen in other LQTS types. Similarly, targeting G_NCX with a blocker, such as SEA0400, could be a potentially worthwhile approach for treating LQTS, particularly given its upregulation in female patients and its contribution to APD prolongation. In the future, known drugs could be systematically screened for sex-dependent efficacy and/or safety (21) in LQTS models by incorporating established effects—such as changes in conductance or gating properties.

Limitations and Future Extensions

AP prolongation and EADs can cause Ca²⁺ overload and indirectly favor Ca²⁺-dependent arrhythmogenesis. Excess Ca²⁺ can enhance the likelihood of EADs or delayed afterdepolarizations by activating Ca²⁺ extrusion via depolarizing Na⁺/Ca²⁺ exchanger current, and triggered APs though subsequent Na⁺ current activation (44), further promoting reentrant and focal arrhythmia. This highlights the importance of incorporating detailed Ca²⁺ dynamics and Ca²⁺-V_m coupling into future models to fully capture the arrhythmogenic potential in both sexes. While our analysis focused on simulating and analyzing mechanisms of single cells, arrhythmias are inherently a multiscale phenomenon with their propagation often depending on the complex interactions between many individual cells. It is important to note that arrhythmias can behave differently in a more complex environment and future studies will expand the cellular simulations to the tissue level to investigate potential sex differences in the functional and structural substrate for sustaining these arrhythmias.

While our study focused on sex-dependent differences in EAD frequency and mechanisms, it is conceivable that other dynamical instabilities, like APD and CaT alternans, are affected by sex and LQTS. In fact, Yang and Clancy reported that alternans developed at slower pacing rates in the female compared to male model, and showed that a modest degree of I_Kr block shifted the alternans pacing threshold toward slower rates in both sexes (22). Our preliminary simulations showed that ORd female model exhibits alternans at a longer BCL than the male model, but the opposite trend was observed in the ToR-ORd model (i.e., alternans develop at longer BCL in male vs female, not shown). In both models, modest I_Kr block increased the BCL threshold for alternans, as shown by Yang and Clancy (22). Thus, further work is warranted to investigate these sex-specific predictions and address model dependency.

Another important factor to be considered when exploring these arrhythmia sex differences is the role of sex hormones on cardiac electrophysiology. While estrogen has been shown to prolong the QT interval, testosterone has a shortening effect (45). We utilized known sex-specific scaling factors for ionic and transporter currents but did not explicitly consider sex hormonal influences such as estrogen, progesterone, and testosterone and the acute fluctuations that come with those. Incorporating hormonal regulation into future models could better reflect additional sex-differences as altering concentrations have been shown to affect QT duration as well as arrythmia risk (46). Additionally, the models used in this study do not account for the heart’s autonomic responses, which regulate cardiac function by influencing pacemaker cells, thereby altering the heart’s beating rate. These responses have direct effects on ventricular dynamics and the expression of various proteins that are critical for normal heart function. Integrating these autonomic mechanisms into future simulations could capture their impact on arrhythmogenesis, improving the physiological relevance of the models.

Conclusions

Our study suggests that females are more prone to arrhythmias in LQTS, with greater abnormalities and EADs seen at slower heart rates. Key ionic factors identified through logistic regression, such as G_Kr and G_NCX, point to potential mechanisms contributing to this increased risk. Targeting these ionic imbalances may offer more effective therapies in treating LQTS, particularly for female patients.

Abbreviations

AP

Action potential

APD

Action potential duration

APD₅₀

APD measured at 50% repolarization

APD₉₀

APD measured at 90% repolarization

BCL

Basic cycle length

CaD₅₀

Ca²⁺ transient duration measured at 50% decay

CaT

Ca²⁺ transient

CMDN_max

Maximal buffering capacity of calmodulin

EAD

Early afterdepolarization

GB

Grandi-Bers model

G_CaB

Maximal conductance of the background Ca²⁺ current

G_CaL

Maximal conductance of the L-type Ca²⁺ current

G_ClB

Maximal conductance of the background Cl⁻ current

G_ClCa

Maximal conductance of the Ca²⁺ dependent Cl⁻ current

G_K1

Maximal conductance of the inward rectifier K⁺ current

G_Kb

Maximal conductance of the background K⁺ current

G_Kr

Maximal conductance of the rapidly activating delayed rectifier K⁺ current

G_Ks

Maximal conductance of the slowly activating delayed rectifier K⁺ current

G_Na

Maximal conductance of the fast Na⁺ current

G_NaB

Maximal conductance of the background Na⁺ current

G_NaK

Maximal transport rate of the Na⁺/K⁺ pump

G_NaL

Maximal conductance of the late Na⁺ current

G_NCX

Maximal transport rate of the Na⁺/Ca²⁺ exchanger

G_pCa

Maximal transport rate of the plasmalemmal Ca²⁺ pump

G_to

Maximal conductance of the transient outward K⁺ current

I_CaL

L-type Ca²⁺ current

I_Kr

Rapidly activating delayed rectifier K⁺ current

I_NaL

Late Na⁺ current

J_leak

Maximal transport rate of SR Ca²⁺ leak via RyRs

J_rel

Maximal transport rate of SR Ca²⁺ release via ryanodine receptors (RyRs)

J_up

Maximal transport rate of the sarcoplasmic reticulum (SR) Ca²⁺ ATPase

LQTS

Long QT syndrome

ORd

O’Hara-Rudy model

PLS

Partial least squares

SERCA

Sarcoplasmic Reticulum Ca²⁺-ATPase

SR

Sarcoplasmic reticulum

ToR-ORd

Tomek-Rodriguez model

τxs1, τxs2

I_Ks Activation time constant

xs1ss, xs2ss

I_Ks Steady-state activation

DATA AVAILABILITY

All our source codes (and related documentation) used in this study and all simulated data presented here are freely available for download at github.com/drgrandilab.