Sudden cardiac arrest (SCA) is a global public health problem with an annual incidence of 40–100 per 100,000 individuals (Fishman et al., 2010; Hayashi et al., 2015). SCA accounts for 50% of deaths from cardiovascular disease and 15–20% of all deaths in industrialized societies (Zipes and Wellens 1998; Wong et al., 2019). Most cases of SCA are caused by cardiac arrhythmias (ventricular fibrillation (VF) or ventricular tachycardia (VT)). Such arrhythmias may arise from functional changes in the ion channels that underlie the cardiac action potential (AP) (Antzelevitch and Burashnikov 2011). These functional changes may be evoked by various drugs used for the treatment of cardiac or non-cardiac conditions. This is best known for drugs that affect cardiac repolarization (QT prolonging drugs) (Haverkamp et al., 2000). However, there is increasing recognition that it also applies to drugs that affect cardiac depolarization (Bardai et al., 2013). An example of such drugs are anti-epileptic drugs (AEDs) (Bardai et al., 2015). Some AEDs are primarily developed for blocking neuronal ion channels, e.g., voltage-gated Na⁺, Ca²⁺ or K⁺ channels, while other AEDs act by impacting on neurotransmitters such as γ-aminobutyric acid (Davies 1995; Sills and Rogawski 2020). Importantly, neuronal and cardiac ion channel isoforms are highly homologous (Heinemann et al., 1994; Fozzard and Hanck 1996). Thus, AEDs may not only affect neuronal electrical activity but may also act on cardiac ion channels, thereby causing cardiac arrhythmias (Danielsson et al., 2005). Accordingly, the increased SCA risk of epilepsy patients may be partly explained by AED use (Bardai et al., 2015).

Carbamazepine (CBZ) is a prime example of such drugs, because it has high efficacy in the treatment of epilepsy (Pellock 2000) through various mechanism, including block of neuronal Na⁺ channels (Ragsdale and Avoli 1998; Catterall 1999; Sun et al., 2006; Lasoń et al., 2013). CBZ may also impact on cardiac electrophysiology as suggested by several CBZ-related case reports and retrospective studies, which report bradycardia, sinoatrial and atrioventricular block, QRS interval prolongation, cardiac arrhythmias, and cardiac arrest, as summarized in Table 1 (Beermann et al., 1975; Hamilton 1978; Herzberg 1978; Boesen et al., 1983; Leslie et al., 1983; Benassi et al., 1987; Kasarskis et al., 1992; Hojer et al., 1993; Schmidt and Schmitz-Buhl 1995; Koutsampasopoulos et al., 2014). Still, the underlying electrophysiological mechanism is not completely understood. Our current study has two aims: 1) to establish whether CBZ is associated with increased SCA risk in a large dataset from a cohort that was specifically designed to study SCA in the general population; 2) to establish the effects of CBZ on cardiac APs and individual membrane currents of rabbit and human cardiomyocytes using patch clamp methodology.

The main findings of the present study are: 1) CBZ use is associated with increased SCA risk in the general population; 2) CBZ reduces cardiac AP upstroke velocity and I_Na in human and rabbit cardiomyocytes; 3) CBZ results in a tendency to (in rabbit) and significant (in human) cardiac AP shortening and reduces I_Ca,L, while leaving sarcolemmal potassium currents unaltered. All of the observed effects are consistent with each other: reduction in cardiac AP upstroke velocity is well explained by reduction in I_Na (Berecki et al., 2010), and may, in turn, lead to reduction in cardiac excitability and conduction velocity of the excitation wavefront in the heart, as represented by CBZ-induced QRS interval prolongation (Leslie et al., 1983). It also facilitates reentrant excitation, VF/VT, and SCA, as shown for the use of class IC antiarrhythmic drugs (potent I_Na blockers) (Rogers et al., 1989), and in Brugada syndrome (where 20% of patients have an identifiable loss-of-function mutation in SCN5A, the gene that encodes the Na_V1.5 α-subunit of the cardiac Na⁺ channel) (Meregalli et al., 2005). Previous case reports (Table 1) have reported findings that are consistent with these electrophysiological effects of CBZ. Accordingly, we found that CBZ use is associated with a 90% increase in the risk of SCA in the general population. These epidemiological findings are consistent with a previous study by Bardai et al. (2015) on the association of SCA with epilepsy and with the use of CBZ, which was conducted in a smaller patient set (10 cases used CBZ and 26 controls were included) and with less certain SCA ascertainment (no ECG documentation). In our study, we had no information regarding the epilepsy status. Hence, we could not adjust for epilepsy in the epidemiological analysis. This is an important limitation considering that epilepsy is associated with increased SCA risk (Surges et al., 2009). Therefore, our findings from the epidemiological analysis should be interpreted with caution. However, Bardai et al. (2015) found that the AEDs with putative cardiac I_Na blocking properties such as CBZ are similarly associated with an increased SCA risk. This was not only observed among patients with epilepsy, but also among patients who had no epilepsy (but used AEDs for other indications, e.g., neuralgia). Moreover, the observed association between CBZ and SCA remained unchanged after correction for epilepsy (Bardai et al., 2015). This suggested that the SCA risk associated with CBZ use resulted from the drug effect rather than from suffering epilepsy per se.

Of note, we measured the effects of different concentrations of CBZ (1–100 µM) in vitro, including concentrations corresponding to plasma levels that provide anticonvulsant effects (20–40 µM) (Bertilsson 1978). CBZ displays a high distribution volume, entering the bloodstream from tissue reserves (Charlier et al., 2021), which, together with the fine end-branches of the vasculature of the heart, would make sure that all cardiomyocytes (not only the cells on the surface) are exposed to the compounds in the blood and the extracellular fluid. Thus, the plasma CBZ concentration is a good measure of the concentration of free CBZ “seen” by the cardiomyocytes in the intact heart and in our in vitro experiments. Kennebäck et al. (1995) reported a CBZ plasma concentration of 26.1 ± 5.5 µM (mean ± SD) at a dose of 400 mg/day and 35.6 ± 5.9 μM at 800 mg/day in healthy volunteers. Correspondingly, our study showed that the CBZ-induced reduction of upstroke velocity was present at 10 μM at 2 Hz and faster, and at 30 μM at all pacing frequencies, which is thus within the range of therapeutic concentrations. Our observed reduction of upstroke velocity is consistent with findings in guinea-pig ventricular cardiomyocytes where 75 µM CBZ significantly reduced dV/dt_max at 1 Hz frequency stimulation by ≈ 13% (Delaunois et al., 2015).

We here compared our used CBZ concentrations to plasma concentrations in healthy volunteers. However, as reviewed by Bertilsson (1978), a poor correlation between the prescribed dose and the actual plasma concentration of CBZ is found in epileptic patients. Furthermore, CBZ plasma levels may be affected by several factors, among which age, pregnancy, and pharmacokinetic drug interactions, including interactions with both central nervous system and cardiovascular drugs (Bertilsson 1978; Panday et al., 2017). Consequently, CBZ plasma levels show a considerable inter-individual variability (Bertilsson 1978; Panday et al., 2017). On the one hand, plasma levels can be so low that therapeutic efficacy is lost, while on the other hand the therapeutic range of 4–10 or 4–12 μg/mL (17–42 or 17–51 μM, respectively) is exceeded in a substantial percentage of patients treated with CBZ (Shakya et al., 2008; Al-Balawi et al., 2020; Eroğlu et al., 2021; Grześk et al., 2021), which may have contributed to the observed cardiac arrhythmias of Table 1. Supratherapeutic CBZ plasma levels were found in 4.9% of their patients by Shakya et al. (2008), in 8.6% by Al-Balawi et al. (2020), in 16% by Eroğlu et al. (2021), and in 2.1% by Grześk et al. (2021).

The CBZ-induced changes in upstroke velocity support our epidemiological findings, and suggest that CBZ affects I_Na (Berecki et al., 2010). Indeed, we found that ≥30 µM CBZ reduced cardiac I_Na and that it affected various gating properties (hyperpolarizing shift in voltage dependency of inactivation and slower recovery from inactivation). Our finding is supported by previous studies on CBZ’s effects on cardiac and neuronal I_Na (Kuo et al., 1997; Sun et al., 2007; Sheets et al., 2008; Harmer et al., 2011; Theile and Cummins 2011). For example, Harmer et al. (2011) found an IC₅₀ of 152 µM for Na_V1.5 channels expressed in CHO cells, while IC₅₀ values for “brain-type” Na⁺ channels expressed in HEK293 cells were 2.5 and 1.6 mM for Na_V1.3 and Na_V1.7 channels in resting state, respectively (Sheets et al., 2008). In resting state, tetrodotoxin-resistant (TTX-R) Na_V1.8 channels had an IC₅₀ of 840 µM in dorsal root ganglion cells (Sheets et al., 2008). CBZ-induced shift in voltage dependency of inactivation and slowed recovery of inactivation were also observed for Na_V1.3, Na_V1.7 and Na_V1.8 channels (Sheets et al., 2008; Theile and Cummins 2011). This strengthens the notion that I_Na block is a plausible contributing mechanism of increased SCA risk associated with CBZ and likely other AEDs with similar cardiac electrophysiological effects. This notion may serve as a basis to adapt clinical procedures for prescription of CBZ with the aim of reducing SCA risk (Benassi et al., 1987). This may be achieved by identifying individuals who are vulnerable to this risk when prescription of I_Na blocking CBZ is considered. This may be based on identification of the clinical conditions that increase SCA risk in the context of I_Na block, similar to guidelines regarding the prescription of I_Na blocking (class IC) antiarrhythmic drugs in case of ischemic heart disease and heart failure (Greenberg et al., 1995). Also, procedures to screen for genetic vulnerability (pharmacogenetics) may be developed (Surges et al., 2009). Finally, at set out above, CBZ levels are affected by several factors and supratherapeutic CBZ levels have been found in a substantial percentage of CBZ users. Therefore, CBZ concentrations need to be closely evaluated (Panday et al., 2017; Charlier et al., 2021).

While I_Na block is a plausible mechanism underlying the higher SCA risk observed during CBZ use, there is less compelling evidence to support the notion that increased SCA risk results from changes in AP repolarization. We found mild effects of CBZ on AP repolarization as indicated by the tendency to (in rabbit cardiomyocytes) and significant (in human cardiomyocytes) APD₉₀ shortening at 100 µM CBZ, which is above the reported plasma concentrations (Kennebäck et al., 1995). An AP shortening was also observed at 75 µM CBZ in guinea-pig ventricular myocytes at 1 Hz stimulation frequency (Delaunois et al., 2015), but QT intervals, ECG measures of the ventricular AP durations, were not affected by therapeutic doses of CBZ (Arhan et al., 2009; Amin et al., 2010; Dogan et al., 2010; Sathyaprabha et al., 2018). The mild extent of CBZ effects on AP repolarization fits with our voltage clamp experiments. We observed a lack of CBZ effects on the main cardiac repolarizing currents, I_K1, I_K and I_to1, consistent with previous findings in other tissues and expression systems (Wooltorton and Mathie 1993; Rundfeldt 1997; Kobayashi et al., 2009). CBZ (10–50 µM) had no effect on I_K in rat isolated sympathetic neurons (Wooltorton and Mathie 1993) and NG108-15 neuronal cells (Rundfeldt 1997), while it did not affect Kir2.1 currents (Kobayashi et al., 2009), with Kir2.1 as the major Kir isoform of I_K1 channels in cardiac myocytes. Although one study reported that CBZ inhibited the I_Kr tail current, the CBZ dosages used in that study (250–500 µM) were much higher than recommended therapeutic concentrations (Danielsson et al., 2003). We found a mild reduction of the depolarizing current I_Ca,L at 100 µM. Although it agrees with findings in cultured rat hippocampus neurons (Ambrósio et al., 1999) and rat sensory spinal ganglion cells (Schirrmacher et al., 1995), it is unlikely that such a decrease contributes to the SCA increase and relates to CBZ-induced changes in whole heart parameters, because the reduction is rather small and only observed at 100 μM, which is above the therapeutic plasma concentrations (Kennebäck et al., 1995). It has been demonstrated that CBZ reduced connexin43 expression in cultured cardiomyocytes (Schirrmacher et al., 1995), but more studies are required to determine the exact role of cardiac connexins in the altered ECG parameters and arrhythmias by CBZ use, and our observation of increased SCA.

The effects of CBZ on APs and I_Na density of freshly isolated human atrial cardiomyocytes were only tested at 100 µM due to the limited availability of Ca²⁺-tolerant, non-depolarized cells (Verkerk et al., 2021). We used human atrial cardiomyocytes isolated from explanted hearts of patients (with various medications) with end-stage heart failure caused by ischemic cardiomyopathy. Although such cells may be in a diseased state, the main effects of CBZ on those human atrial cardiomyocytes were largely similar to those on ventricular cardiomyocytes of control rabbits, indicating that the effects of CBZ are also present in human conditions. The K⁺ currents and I_Ca,L were measured with very general voltage clamp protocols without specific solutions and/or blockers. Although such measurements might also involve small contributions of other membrane currents, the CBZ effects were assessed in paired experiments. In addition, our findings match with CBZ findings on membrane currents in non-cardiomyocytes (Wooltorton and Mathie 1993; Rundfeldt 1997; Ambrósio et al., 1999; Danielsson et al., 2003; Sheets et al., 2008; Kobayashi et al., 2009; Theile and Cummins 2011), indicating that the CBZ effects on these (net) currents were reliably characterized.

CBZ reduces cardiac depolarization by reducing I_Na, and inducing an associated reduction of the AP upstroke velocity, in cardiomyocytes at therapeutic plasma concentrations. CBZ also affects cardiac repolarization, by reducing I_Ca,L, and an associated reduction of AP duration, but only at relatively high concentrations. These electrophysiological effects may contribute to the found increased SCA risk upon CBZ use in the general population.