The Discovery of Cenobamate: A Drug with High Efficacy in Drug-Resistant Epilepsy

Abstract

Cenobamate is a novel alkyl-carbamate antiseizure medication (ASM) that represents a major breakthrough in the treatment of drug-resistant epilepsy, particularly focal seizures. Unlike other ASMs, cenobamate achieves seizure freedom in up to one third of patients with focal drug-resistant epilepsy—a response not observed with other therapies. Herein, we describe the chemocentric and phenotypic screening strategy employed by one of us (Yong Moon Choi) while at SK Biopharmaceuticals and how cenobamate was developed by optimizing earlier alkyl-carbamates such as carisbamate and felbamate. We discuss how the incorporation of a tetrazole moiety likely played a significant role in its successful development. Cenobamate was found to display potent broad-spectrum activity in diverse disease-relevant rodent models, including maximal electroshock, pentylenetetrazole, kindling, and lithium–pilocarpine-induced status epilepticus. When compared to 12 other ASMs, it ranked highest across four models of difficult-to-treat focal seizures. We discuss how the two known molecular activities of cenobamate, i.e., positive allosteric modulation of GABA_A receptors via a non-benzodiazepine site and inhibition of persistent sodium currents (I_NaP), may contribute to its clinical efficacy and how pharmacokinetic/pharmacodynamic modeling confirmed that the effective brain concentrations of cenobamate align with clinically relevant plasma concentrations observed in seizure-free patients. The discovery of cenobamate has significant implications for future ASM development. It validates phenotypic screening as a powerful tool for identifying first-in-class central nervous system therapeutics and highlights the value of chemocentric optimization using known scaffolds. The predictive utility of specific preclinical models—especially kindling and lithium–pilocarpine—suggests that future ASM discovery should prioritize models that better mimic pharmacoresistant epilepsy. The dual mechanisms of cenobamate may serve as a blueprint for designing ASMs with synergistic actions. Approval of cenobamate represents a paradigm shift in epilepsy treatment and offers a promising framework for discovering more effective therapies for drug-resistant epilepsy.

Key Points

Cenobamate appears to achieve a previously unseen seizure freedom in patients with drug-resistant epilepsy and be a breakthrough antiseizure medication.

We describe previously unpublished history of its discovery, which occurred through chemocentric optimization of alkyl-carbamates and phenotypic screening.

Dual mechanism of action as GABAA receptor modulator and inhibitor of persistent sodium current may offer synergistic efficacy.

Introduction

The alkyl-carbamate cenobamate (Fig. 1) is a novel antiseizure medication (ASM) with high efficacy in controlling focal seizures in people with drug-resistant epilepsy (DRE) [1, 2]. In the pivotal, randomized, double-blind, placebo-controlled (RDBPC) phase II add-on study of adults with drug-resistant focal epilepsy, 21% of patients treated with cenobamate 400 mg/day achieved seizure freedom during 12 weeks of maintenance treatment, compared with a median of 9 seizures/28 days at baseline [3]. Seventy-eight percent of those patients were treated with two to three concomitant ASMs. This seizure freedom rate is ~ 3–10 times higher than seizure freedom rates of all other new ASMs evaluated in similarly designed trials since the 1990s [1].

Fig. 1.

Chemical structures of cenobamate and, for comparison, the anticonvulsant alkyl-carbamates felbamate (a dicarbamate) and retigabine (ezogabine)

The differentiated efficacy of cenobamate may be reflected by the fact that the US Food and Drug Administration (FDA) never requested phase III efficacy studies—cenobamate was approved by the FDA, uniquely, based on efficacy results of two phase II studies [3, 4].

In the open-label extension study of the pivotal study, with a median treatment duration of 4.5 years, and a median dose of 300 mg/day, 10% of all patients were seizure free during year 4 of the study, and 24% patients had > 90% seizure frequency reduction [5]. Median duration of seizure freedom was 48 months.

In a pivotal, phase III, open-label safety study of cenobamate with a median treatment duration of 30 months, 13% of patients achieved seizure freedom during the maintenance treatment period, and 40% achieved > 90% seizure frequency reduction [6]. Thirty-six percent of patients had seizure freedom for ≥ 12 months at the last visit. In that study, 35% of patients who had previously failed resective or ablative epilepsy surgery (n = 40) became seizure free for 12 months or longer [7]. These results have been broadly replicated in real-world experience studies [8–11].

Thus, cenobamate may be a paradigm shift and breakthrough in the treatment of focal DRE that has long been searched for [1, 12]. Before cenobamate, none of the about 25 second- and third-generation ASMs developed in the last 40 years have significantly changed the proportion of patients with focal DRE [1, 13–17], although a few retrospective studies have shown ~ 15% seizure freedom at 6–12 months [18, 19].

Understanding why cenobamate differs so markedly in its clinical efficacy from previously developed ASMs is important, as it will help guide the discovery and development of future ASMs [20]. One important aspect is the strategy used to discover cenobamate. However, no information other than that the compound was discovered purely through phenotypic screening is available in the public domain [2, 15, 21]. This prompted us to describe the drug discovery approach used by SK Biopharmaceuticals (Daeduk/Seoul, South Korea) and its US subsidiary, SK Life Science (Fairfield/Paramus, NJ, USA), in more detail. Furthermore, by comparing the antiseizure activity of cenobamate with that of other alkyl-carbamates and various ASMs with other chemical structures in a variety of rodent models of seizures and epilepsy, we try to identify which model(s) may have predicted the unique clinical efficacy of cenobamate.

Phenotypic screening is a drug discovery approach that has been used for decades, aiming to identify compounds in in vitro or in vivo models that display disease-relevant results without prior knowledge of the affected molecular target(s) [22, 23]. Phenotypic screening has been shown to be more effective than “rational” target-based approaches in discovering first-in-class small-molecule drugs [24, 25], leading to a recent resurgence in phenotypic screening in industry as well as academia [23, 26]. As reviewed recently [23, 26] and illustrated by cenobamate here, phenotypic drug discovery has the potential to be much more than random screening in complex systems, in particular when combined with a chemocentric approach, i.e., an approach based on a specific compound or compound class that serves as starting point for further optimization [27].

The Discovery of Cenobamate

Starting in 2001, cenobamate (YKP3089) was designed by one of us (YMC) at SK Biopharmaceuticals. YMC previously conceived of and synthesized several other therapeutically effective alkyl-carbamates, including carisbamate and solriamfetol (see below), and was also involved in the development of felbamate (in the 1980s at Carter-Wallace). The aim of the drug discovery project at SK Biopharmaceuticals described here was to develop alkyl-carbamates that retain the broad spectrum of antiseizure activities but are more effective and better tolerated than the previously developed alkyl-carbamates. Importantly, cenobamate differs from other anticonvulsant alkyl-carbamates by its tetrazole substituent (see Fig. 1). The discovery of more effective alkyl-carbamates included four consecutive steps, finally resulting in cenobamate (Fig. 2).

Fig. 2.

Schematic of the drug discovery approach that led to cenobamate. The chemocentric approach was based on two specific compounds (YKP10A and, later, YKP511) that served as starting points for further optimization. Phenotypic screening in mouse models of seizures (and the rotarod test for safety) was used to characterize the various alkyl-carbamate derivatives that were produced by the chemocentric approach. The tetrazole group in pentylenetetrazole (PTZ) and cenobamate is highlighted by the stippled circle. See text for details. MES maximal electroshock seizures

Step 1

The project to develop new, more effective (and less toxic) alkyl-carbamates started with a compound called YKP10A ((2R)-2-amino-3-phenylpropyl carbamate [solriamfetol]; see Fig. 2). In 1995, SK Biopharmaceuticals found that, in rodents, YKP10A exhibits activity-increasing (“antidepressant”) effects in the forced swim test and moderate anticonvulsant activities in rodent seizure models, i.e., the maximal electroshock seizure (MES) test and the pentylenetetrazole (PTZ) seizure test. At that time, YMC was interested in the functionality of the amino group with carbamoyl moiety/carbamate to develop a novel ASM.

YKP10A (solriamfetol) was initially developed and clinically evaluated as a new antidepressant drug [28, 29] but was subsequently found to induce wake-promoting effects [30], and was further developed by Jazz Pharmaceuticals, leading to FDA approval (in 2019) for the treatment of excessive daytime somnolence associated with narcolepsy or obstructive sleep apnea. Solriamfetol inhibits the presynaptic dopamine and norepinephrine transporters, thereby increasing extracellular concentrations of these neurotransmitters [31]. Interestingly, drugs that increase extracellular levels of monoamines in the brain have been reported to exert anticonvulsant activity in rodent models [32–34], which may explain the moderate anticonvulsant activity of solriamfetol in initial screening seizure models described above. Monoamines have also been implicated in the anticonvulsant effect of clinically used ASMs, such as lamotrigine [35], valproate [36, 37], phenytoin [38], carbamazepine and zonisamide [39], and oxcarbazepine [40], and may contribute to the antidepressant and anticonvulsant action of vagal nerve stimulation [41–44]. There is a long-held misconception that antidepressants that potentiate monoaminergic transmission are proconvulsant in epilepsy, which is largely based on case reports involving tricyclic antidepressants at supratherapeutic doses [45, 46]. However, at low, therapeutically relevant doses, most antidepressant drugs exert anticonvulsant effects in seizure models, which is in line with the effect of monoamines on the seizure threshold [45, 46].

Step 2

In a collaboration with the Epilepsy Therapy Screening Program (ETSP) of the US National Institute of Neurological Disorders and Stroke (NINDS) at the University of Utah, which was directed by one of us (HSW) during the period of the development of cenobamate, SK Biopharmaceuticals found one alkyl-carbamate with high anticonvulsant activity, i.e., YKP509 ((2S)-2-chlorophenyl-2-hydroxyethyl carbamate; carisbamate) in 1994 (Fig. 2). Carisbamate was discovered by modifying the structure of the major felbamate metabolite 2-phenyl-2-hydroxy-1,3-propanediol dicarbamate [47], by including a chiral center and R, S, and R&S configuration. The chloride moiety at the phenyl ring was added to increase metabolic stability.

Carisbamate was in-licensed by Johnson & Johnson in the USA in 1998 and evaluated as a novel ASM [21]. However, carisbamate failed to demonstrate consistent efficacy across regulatory trials in patients with focal DRE [48–50], and therefore, the clinical program was discontinued by Johnson & Johnson. SK Life Science is currently performing clinical trials with carisbamate in Lennox–Gastaut syndrome [51–53].

Starting from carisbamate, further analogs were synthesized and tested in vivo using the concept of the structure–activity relationship (SAR) with the aim to increase the potency and efficacy but decrease toxicity. Structure–activity relationship studies involved systematically altering the chemical structure of carisbamate to understand how these changes affect its biological activity, guiding the development of potent, selective, and safer drug candidates. During these SAR studies in 1994–95, the dicarbamate YKP511 ((2R)-2-chlorophenyl-1,2-ethanediol dicarbamate) was derived from carisbamate by replacing the hydroxyl group at C2 with a carbamate moiety (Fig. 2). This structural alteration was prompted by one of the human metabolites of carisbamate, 1-2-chlorophenyl-2-hydroxyethyl carbamate (R307716), which results by a migration of a carbamate group from C1 to C2 position via intramolecular cyclization [54]. YKP511 showed much higher anticonvulsant potency than felbamate and was thus chosen as the main frame for further analog synthesis, replacing one of the two carbamate groups with other groups.

Step 3

After SK Biopharmaceuticals had out-licensed carisbamate to Johnson & Johnson, SK Biopharmaceuticals started a new discovery project that finally led to cenobamate in 2002. A variety of functional amino groups were applied by YMC as a pharmacophore by replacing one of the carbamate groups in YKP511 as described above. The applied groups included acyclic (e.g., guanidine, alkylamine) and cyclic (e.g., triazole, tetrazole) amine radicals. After analog synthesis and screening tests in mouse and rat seizure models, a tetrazole moiety (such as in PTZ; Fig. 2) was identified as an optimal choice.

Tetrazole is a five-membered aromatic ring containing four nitrogen atoms and one carbon atom that is a versatile scaffold in drug discovery, widely used as a metabolically stable, lipophilic bioisosteric equivalent of carboxylic acid, cis-amide, and other functional groups [55–57]. Adding a tetrazole moiety often improves the drug’s properties, such as increased lipophilicity and metabolic stability, stronger ionic and hydrogen bonding that can enhance potency, and reduced adverse effects [57, 58]. Despite the increasing prevalence of tetrazoles in FDA-approved drugs for various conditions (e.g., antihypertensive sartans such as losartan, cephalosporin antibiotics such as cefotetan, anti-chorea drugs such as tetrabenazine), our understanding of their SARs, multifunctional mechanisms, binding modes, and biochemical properties remains limited [57]. Cenobamate is an interesting example of how using a tetrazole scaffold can markedly change the preclinical and clinical efficacy of alkyl-carbamates. Cenobamate is the first FDA-approved ASM in which a tetrazole moiety was used as a pharmacophore. Various tetrazole derivatives have previously been reported to exert anticonvulsant effects [57, 59], including close analogs of PTZ [60]. This encouraged the application of a tetrazole scaffold during the development of cenobamate (Fig. 2).

Pentylenetetrazole (6,7,8,9-tetrahydro-5H-tetrazolo[1,5-a]azepine), also known as pentetrazole, metrazol, or cardiazol, is a widely used convulsant that acts as a GABA_A receptor antagonist via binding to the picrotoxin site of the GABA-gated chloride channel, thereby blocking GABA-activated chloride flux [61]. Importantly, as discussed in Sect. 4, cenobamate acts via the same site of the GABA_A chloride ionophore complex to enhance the effect of GABA. Thus, it is tempting to speculate that the tetrazole moiety in both PTZ and cenobamate is involved in the affinity of these drugs to the picrotoxin site of the GABA_A receptor. This is substantiated by previous findings that various tetrazole derivatives bind to this site [61]. The structural features of the derivatives determine whether they act as agonists or antagonists at the receptor.

Similar to the tetrazole moiety, triazole (a heterocyclic compound featuring a five-membered ring of two carbon atoms and three nitrogen atoms) is widely used in drug development because of its excellent physicochemical properties, ability to form hydrogen bonds, and versatility as a privileged scaffold [62]. The triazole moiety has been used in the development of various compounds with anticonvulsant properties [63, 64]. One anticonvulsant triazole derivative, rufinamide (1-(2,6-difluorobenzyl)-1H-1,2,3-triazole-4-carboxamide), was developed by Novartis in 2004 and FDA approved for the adjunctive treatment of seizures associated with Lennox–Gastaut syndrome in 2008 [65]. Another example is the triazole derivative loreclezole (1-(2-chloro-2-(2,4-dichlorophenyl)ethenyl)-1H-1,2,4-triazole), which is the product of a screening program by Janssen Pharmaceutica in the mid-1980s, which originated with the idea of trying to separate the hypnotic from the anticonvulsant actions of etomidate [66]. Loreclezole displayed broad-spectrum anticonvulsant activity in animal models independently of the benzodiazepine binding site of the GABA_A receptor [67]. Later, it was found to act as a positive allosteric modulator of GABA_A receptors, enhancing peak GABA currents by binding to a novel binding site that is localized at the interface between the principal face of a β2 or β3 subunit and the complementary face of an α subunit at the upper transmembrane domain of the GABA_A receptor [68]. While loreclezole showed promise in early clinical trials for DRE, a double-blind placebo-controlled trial did not indicate any marked efficacy [69], and long-term treatment was affected by unfavorable pharmacokinetics, including a very long half-life of the major metabolite of loreclezole, so further clinical development was discontinued [66]. However, research into the potential of loreclezole for new drug scaffolds continues until today [70]. Thus, both the tetrazole and triazole groups were interesting moieties in the development of more effective alkyl-carbamates.

Step 4

Next, YMC applied chiral chemistry (R, S, R&S) to about 50 tetrazole/triazole carbamate analogs or derivatives to select the optimum candidates based on the potency (MES and PTZ) and safety (rotarod test) shown in preclinical testing. The rationale for the development of enantiomerically pure ASMs with potentially improved efficacy, safety, and commercial viability has been discussed recently [71]. The decision process to finally select cenobamate ((1R)-1-(2-chlorophenyl)-2-(tetrazol-2-yl) ethyl) carbamate)) included the tetrazole moiety as a pharmacophore and the results of testing in several animal models of drug-resistant seizures, including the lithium-pilocarpine model of status epilepticus (SE) in rats. In this model, single-dose treatment with cenobamate was highly potent and effective in suppressing the SE and preventing neuronal damage in the hippocampus, whereas carisbamate was ineffective (see next section).

The Preclinical Profile of Cenobamate

The effects of cenobamate in mouse and rat models of the ETSP were first reported by Bialer et al. [72] and, in more detail, recently by Melnick et al. [2, 73]. Similar to other alkyl-carbamates, such as felbamate, carisbamate, and retigabine (ezogabine), cenobamate has a broad spectrum of antiseizure activity in diverse mouse and rat models of seizures and epilepsy (Tables 1, 2). These models include screening tests of generalized seizures, such as the MES and PTZ tests, the 6-Hz model of focal seizures, the chronic amygdala and hippocampal kindling models of temporal lobe epilepsy, seizures or SE induced by lithium-pilocarpine, and induced or spontaneous generalized seizures in genetic models of epilepsy, i.e., audiogenic seizure-susceptible DBA/2 mice and GAERS (Genetic Absence Epilepsy Rats from Strasbourg) with spontaneous absence-like spike-wave discharges in the electroencephalogram. Apart from cenobamate, only two other clinically used ASMs, phenobarbital and valproate, are effective in all of these models (Tables 1, 2).

Table 1.

Anticonvulsant activity of cenobamate in acute and chronic seizure models in mice^a

Drug	ED50 (mg/kg i.p.) in seizure models in mice							TD50 (mg/kg i.p.) in the rotarod test	Protective index (TD50:ED50 in MES test)	References
	MES (tonic seizures)	s.c. PTZ (clonic seizures)	6 Hz (focal seizures)				DBA/2 (clonic audiogenic seizures)
			22 mA		32 mA	44 mA
Alkyl-carbamates
Cenobamate	9.8	28.5	11.0		17.9	16.5	14	58.0	5.9	[73, 115]
YKP3090	38.2	56.4	NT		NT	NT	NT	143	3.7	[73]
YKP1983	15.8	35.9	NT		NT	NT	NT	92.5	5.9	[73]
Carisbamate	7.9	20.4	20.7		21.4	27.6	NT	46	5.8	[72]
Felbamate	35.5	126	13.1		69.5	241	48.8	220	6.2	[21, 116]
Retigabine (ezogabine)	9.3	13.5	NT		26	33	6.8	20.5	2.2	[21, 116]
SV2A modulators
Levetiracetam	NE	NE	4.6		19.4	1089	30	1,389	NE in MES	[78, 86, 117]
Brivaracetam	113	30	NT		NT	4.4	2.4	195	1.7	[117]
Sodium channel modulators
Phenytoin	6.7	NE	NT		NE	NE	2.5	51	7.6	[87, 116]
Carbamazepine	7.8	NE	10.9		25.8	25.9	4.4	45.4	5.8	[78, 87, 116, 117]
Lamotrigine	5.4	NE	18.6		59.6	43.7	3.5	30	5.5	[78, 87, 116, 117]
Lacosamide	4.5	NE	NT		20	15.2	0.63 (Frings mice)	26.8	5.9	[87, 118]
ASMs with multiple mechanisms
Topiramate	18.3	NE	NT	NE		NT	12.1	234	12.7	[87, 116, 117]
Valproate	263	220	59.6		139	289	43	398	1.5	[78, 87, 116]
Phenobarbital	11.3	27.5	NT		14.8	35.3	3.4	45.5	4.0	[78, 87, 116]

ASM antiseizure medication, ED₅₀ effective dose in 50% of mice, HPD hippocampal paroxysmal discharge, i.p. intraperitoneally, MES maximal electroshock seizure, NE not effective, NT not tested (or not available in the public domain), PTZ pentylenetetrazole, s.c. subcutaneous, SV2A synaptic vesicle glycoprotein 2A, TD minimal neurotoxic dose in 50% of mice

^aOther commonly used antiseizure medications are shown for comparison. Furthermore, the S-enantiomer of cenobamate (YKP3090), the racemate of cenobamate (YKP1983), and the alkyl-carbamates carisbamate and retigabine (ezogabine) are included for comparison with cenobamate. All drugs were administered before (usually 0.25–2 h) seizure induction. Furthermore, minimal neurotoxicity (in the rotarod test; indicated by TD₅₀) and the protective index are shown

Table 2.

Anticonvulsant activity of cenobamate in acute and chronic seizure models in rats^a

Drug	ED50 (mg/kg i.p.) in seizure models in rats								TD50 (mg/kg i.p.) in the rotarod test	Protective index (TD50:ED50 in MES test)	References
	MES (tonic seizures)	s.c. PTZ (clonic seizures)	Hippocampal kindling (focal onset, sec. gen. seizures)	Amygdala kindling (focal onset, sec. gen. seizures)			Lithium-pilocarpine (or pilocarpine) induced SE	GAERS or WAG/Rij (absence-like SWDs)
				Focal		Gen.
Alkyl-carbamates
Cenobamate	2.9	13.6	16.4	~ 4.8			7.0	~ 20	38.9	13.4	[2, 73, 76]
YKP3090	26.1	16.7	NE (> 50)	NT		NT	NT	NT	50.9	1.95	[73]
YKP1983	8.4	19.3	NE (> 30)	NT		NT	NT	NT	53.3	6.3	[73]
Carisbamate	NT	NE	22.5	> 40			NEb	~ 15	39.5	1.8	[72, 119, 120]
Felbamate	35	238 (p.o.)	296	> 50			80.5	> 100 (p.o.)	> 500	> 14	[21, 114, 121–123]
Retigabine	5.1	NT	NT	3.2			NT	NE	10	1.96	[21, 124]
SV2A modulators
Levetiracetam	NE	NE	~ 350c	33	32		7.1 (mice)	5.4d	1960	NE in MES test	[82, 117, 125–127]
Brivaracetam	NT	NT	0.2d	44			NE (mice)	6.8d	370	NT	[72, 82, 117, 128]
Sodium channel modulators
Phenytoin	7.8	NE	NE	~ 50bc		~ 30c	NE	NE	47.1	6.0	[80, 87, 122, 129]
Carbamazepine	4.1	NE	28.2	15		8	28.9	NE	33.7	8.2	[78, 87, 122, 125]
Lamotrigine	4.4	NE	27	NE			NT	50	37.4	8.5	[87, 125, 130]
Lacosamide	3.1	NE	13.5	7		12	NE (mice)	NE	25.8	8.3	[87, 126, 131, 132]
ASMs with multiple mechanisms
Topiramate	11.5	NE	NE	13.9		10.6	NE (> 50)	~ 10	299	26	[87, 133–135]
Valproate	148	195	214	220		190	286	100	314	2.1	[78, 80, 87, 125]
Phenobarbital	2.6	41	28.5	44		16	23.4	5	41.1	15.8	[78, 80, 87, 125]

ASM antiseizure medication, ED₅₀ effective dose in 50% of rats, GAERS Genetic Absence Epilepsy Rats from Strasbourg, gen. generalized, i.p. intraperitoneally, MES maximal electroshock seizure, NE not effective, NT not tested (or not available in the public domain), p.o. orally, PTZ pentylenetetrazole, s.c. subcutaneous, sec seconds, SE status epilepticus, sec., SV2A synaptic vesicle glycoprotein 2A, SWD spike-wave discharge, TD minimal neurotoxic dose in 50% of rats

^aOther commonly used ASMs are shown for comparison. Furthermore, the S-enantiomer of cenobamate (YKP3090), the racemate of cenobamate (YKP1983), and the alkyl-carbamates carisbamate and retigabine (ezogabine) are included for comparison with cenobamate. All drugs were administered before (usually 0.25–2 hours) seizure induction. Furthermore, minimal neurotoxicity (in the rotarod test; indicated by TD₅₀) and the protective index are shown

^bBut effective after repeated high doses[136]

^c~ 50% protection at this dose, but no increase in protection by increasing the dose

^dMinimally active dose (mg/kg)

In all models, cenobamate was found to be more potent than the alkyl-carbamates felbamate and carisbamate, fulfilling the goal of the development of this novel ASM (see Sect. 2). The only exception was the mouse PTZ test, in which carisbamate was slightly more potent than cenobamate (Table 1). The alkyl-carbamate retigabine (Fig. 1), which was not developed by SK Biopharmaceuticals, exhibited a similar antiseizure potency as cenobamate. However, compared with cenobamate, the protective index of retigabine in mice and rats was markedly lower. Retigabine differs from other alkyl-carbamates in its mechanism of action (MOA) in that it predominantly acts as an activator of voltage-gated neuronal KCNQ (K_v7.2-7.5) potassium channels [21]. These channels play a critical role in modulating susceptibility to seizures, and mutations in genes that encode these channels cause heterogeneous epilepsy phenotypes [74]. However, like cenobamate, retigabine also enhances synaptic and extra-synaptic GABA-evoked responses [21]. Indeed, it has been shown that inhibitory effects of retigabine on seizure-like activity in hippocampal neurons persist in the presence of a blockade of K_v7 channels [75], bolstering the view that positive modulation of GABA_A receptors likely makes a significant contribution to its antiseizure activity.

As expected, cenobamate, an (R)-enantiomer, was more potent than the (S)-enantiomer (YKP3090) and the racemate (YKP1983). More importantly, in contrast to the R-enantiomer, both the S-enantiomer and the racemate were not effective in the hippocampal kindling model (Table 2).

Compared with other clinically used ASMs, cenobamate was not the most potent drug in mouse seizure models (Table 1) but had the lowest median effective dose (ED₅₀) in the rat lithium-pilocarpine model (Table 2). Furthermore, it was among the most potent ASMs in the MES, PTZ, and kindling models in rats.

As described in Sect. 2, the high efficacy of cenobamate in the lithium-pilocarpine model was crucial for the decision to choose cenobamate as the lead compound for further development. Cenobamate blocked the SE in this model with an ED₅₀ of 7 mg/kg intraperitoneally [76]. Furthermore, a single injection of cenobamate prevented neurodegeneration in the hippocampus, while carisbamate was ineffective. The pilocarpine or lithium-pilocarpine models, albeit widely used as models of pharmaco-resistant SE [77] and, after development of spontaneous recurrent seizures, as models of temporal lobe epilepsy (TLE) [78], are not routinely used (or validated) as models of DRE [78]. Pilocarpine (with or without lithium that potentiates the effect of pilocarpine) is a muscarinic acetylcholine receptor agonist that induces a severe SE characterized by continuous limbic (focal nonconvulsive) and generalized convulsive seizures that are initially induced by the convulsant but become self-sustained, lasting several hours [77, 79]. To our knowledge, the sensitivity of pilocarpine-induced seizures to pretreatment with ASMs was first described by Turski et al. [80], who found that the limbic (focal) seizures produced by pilocarpine are resistant to several major ASMs. As shown in Table 2, levetiracetam (LEV) was the only ASM that exhibited similar antiseizure potency to cenobamate in this model (but the experiments with LEV were performed in mice and not in rats).

The kindling model of TLE, which results from chronic electrical stimulation of the amygdala or hippocampus, was crucial in the discovery of LEV [81, 82] and predicted the clinical efficacy of most novel ASMs in patients with focal DRE [78]. As shown in Fig. 3, cenobamate is quite effective in hippocampal and amygdala kindled rats and blocks both the focal nonconvulsive (Racine stages 1 and 2) and the secondarily generalized convulsive seizures (stages 4 and 5) induced by electrical stimulation [2, 73]. In addition, cenobamate significantly increases the focal seizure threshold in this model [73]. When compared with LEV and felbamate, cenobamate is markedly more potent and effective in kindled rats (Fig. 3).

Fig. 3.

Anticonvulsant effects of cenobamate (CNB) and, for comparison, levetiracetam (LEV) and felbamate (FBM), in amygdala and hippocampal kindled rats. Rats were fully kindled until they exhibited reproducible stage 5 seizures upon electrical stimulation with a fixed suprathreshold stimulus, which was applied via a depth electrode to the amygdala or hippocampus. At a fixed interval after single-dose drug (or vehicle) intraperitoneal treatment (typically 0.25–2 h), the stimulus was applied and the seizure severity was scored by the Racine scale in groups of 6–14 fully kindled rats. Data are shown as means + standard error of the mean. Significant difference to vehicle control (“0”) is indicated by an asterisk (P < 0.05). Data on CNB in amygdala kindled rats are from Melnick et al. [2]; data on CNB in hippocampal kindled rats are from Melnick et al. [73]; data on LEV in amygdala kindled rats are from Löscher and Hönack [81]; data on LEV in hippocampal kindled rats are from De Smedt et al. [113]; data on FBM in amygdala kindled rats are from Wlaz and Löscher [114]; data on FBM in hippocampal kindled rats are from the NINDS PANAChE database (https://panache.ninds.nih.gov). A limitation of this comparison of three drugs in two kindling models is the heterogeneity across preclinical studies, including, but not restricted to, varying rat strains, sexes, drug formulations, pretreatment times, and the equipment (e.g., stimulators) used. i.p. intraperitoneally

In the 6-Hz mouse model of pharmaco-resistant focal seizures, cenobamate exhibited similar potency at stimulation currents that ranged from the electroconvulsive threshold (CC₉₇; 22 mA) to twice the CC₉₇ (44 mA), while most other ASMs lost efficacy at the suprathreshold currents (Table 1). Exceptions were retigabine, brivaracetam, lacosamide, and carbamazepine. Thus, this model failed to differentiate cenobamate from other ASMs.

As shown in Tables 1 and 2, cenobamate was not only effective in models of difficult-to-treat focal seizures but also exhibited a broad spectrum of antiseizure activity in several other models, including models of absence-like seizures such as the PTZ test in mice and rats and the GAERS model, and models of primarily generalized convulsive seizures such as the MES test in mice and rats and audiogenic seizure-susceptible DBA/2 mice. Retrospective clinical studies and case reports suggest the effectiveness of cenobamate in a wide array of generalized seizure types, including absence seizures [2]. A 22-week RDBPC study of cenobamate in adolescents and adults with idiopathic generalized epilepsy and primary generalized tonic-clonic seizures is ongoing (NCT03678753).

Since the preclinical evaluation of cenobamate by the ETSP, which was first described by Bialer et al. [72], the ETSP added several rodent models to its testing scheme for DRE [78, 83]. Importantly, the program now uses three chronic models of TLE with spontaneous recurrent seizures: (1) the systemic kainate rat model, in which spontaneous recurrent seizures develop after intraperitoneal injection of the convulsant; (2) the intrahippocampal kainate mouse model; and (3) the intra-amygdala kainate  mouse model. Spontaneous recurrent seizures in these models are refractory to many clinically used ASMs [78]. The ETSP is currently testing cenobamate in these models (Brian Klein, personal communication). Respective data will show whether any of these models predict the remarkable clinical efficacy of cenobamate in focal DRE.

Ranking of Cenobamate in Models of Difficult-to-Treat Focal Seizures

The ED₅₀s of cenobamate, carisbamate, and 12 FDA-approved first-, second-, and third-generation ASMs shown in Tables 1 and 2 were used to rank all drugs based on their potency in the four rodent models of difficult-to-treat focal seizures, i.e., the 6-Hz (44-mA) mouse model, the hippocampal and amygdala kindling rat models, and the lithium-pilocarpine model in rats. As shown in Fig. 4, when the ranking in the four models was combined, cenobamate was markedly more potent and effective than any of the other 12 ASMs. Thus, this ranking seems to predict the higher efficacy of cenobamate as an add-on treatment in patients with focal DRE [1, 84]. This is an unexpected finding because anticonvulsant ED₅₀s are thought to primarily relate to drug potency and not efficacy [85]. It is important to note that for the present ranking, only those ED₅₀s were used for ranking that were below the “neurotoxic” TD₅₀s in the rotarod test (Tables 1, 2).

Fig. 4.

Cumulative ranking of 13 antiseizure medications (ASMs) in four rodent models of difficult-to-treat focal seizures, i.e., the 6-Hz (44-mA) mouse model, the hippocampal and amygdala kindling rat models, and the lithium-pilocarpine model in rats. In each model, the potency (effective dose in 50% of rats [ED₅₀]) was ranked from 1 (highest potency) to 13 (lowest potency). Only drugs for which data were available for at least three of the four models (see Tables 1, 2) were used for the ranking. If a drug was not effective (i.e., an ED₅₀ could not be determined), it received a rank of 13. Similarly, if the ED₅₀ was ≥ TD₅₀ in the rotarod test, the drug received a rank of 13. If a drug was not tested in one of the four models, it received an intermediate rank of 6.5. For each drug, the ranks obtained in the four models were summed up for overall ranking in rodent models of difficult-to-treat focal seizures. These cumulative ranks are shown in the figure, indicating that cenobamate was by far the most potent and effective ASM. This was substantiated by statistical analyses, using Barnard’s test, which indicated that the rank obtained for cenobamate was significantly lower than the rank of any other ASM (*P < 0.05; ***P < 0.0001). A limitation of this comparison of ASMs in a panel of animal models is the heterogeneity across preclinical studies, including, but not restricted to, varying rodent strains, sexes, drug formulations, pretreatment times, and the equipment (e.g., stimulators) used. BRV brivaracetam, CBZ carbamazepine, CNB cenobamate, CRS carisbamate, FBM felbamate, LCM lacosamide, LEV levetiracetam, LTG lamotrigine, PB phenobarbital, PHT phenytoin, RTG retigabine (ezogabine), TPM topiramate, VPA valproate

In animal models of drug-resistant focal seizures, efficacy is more difficult to determine than potency, but one approach is determining ED₅₀s in the 6-Hz model at increasing current intensities (CC₉₇, 1.5xCC₉₇, 2xCC₉₇) as proposed by Barton et al. [86]. As shown in Table 1, at threshold current (CC₉₇), most ASMs block seizures in this model. However, when the threshold current is increased, most ASMs become less effective or ineffective in this test. Only a few second- and third-generation ASMs, in particular cenobamate, remain effective [87], thus allowing the differentiation of ASMs in terms of anti-seizure efficacy in this mouse model.

Indeed, an alternative view of anticonvulsant ED₅₀s being a pure measure of potency is the fact that often an increase in seizure threshold is involved in a drug’s ED₅₀ in a seizure model. For instance, LEV significantly increases the electroconvulsive threshold for tonic seizures in mice and rats, but the increase is not sufficiently large to allow the drug to be effective in the MES test with suprathreshold stimulation [81]. Thus, the resistance of the MES test to LEV (Table 1) results from the low efficacy of LEV to increase the seizure threshold in this model. In contrast, LEV induces large increases in the seizure threshold in amygdala kindled rats [81, 82], so that seizures induced by suprathreshold stimulation in this model are blocked by LEV (Fig. 3A). Thus, the low ED₅₀ of LEV in amygdala kindled rats (Table 2) results from a high efficacy to increase the seizure threshold. Many ASMs, including cenobamate, act by increasing the seizure threshold [73, 88, 89]. As a consequence, the ED₅₀s in the diverse models shown in Tables 1 and 2 may reflect both potency and efficacy, and resistance of a model to an ASM primarily relates to insufficient efficacy in this model. The possibility that the ED₅₀ may be a surrogate of efficacy seems to be substantiated by the results of ranking reported here (Fig. 4).

The MOA of Cenobamate

The presumed MOAs of cenobamate were not identified until later in its development. This is typical in phenotypic drug discovery, which relies on chemical interrogation of a disease-relevant biological system in a molecular target-agnostic approach [23]. Only two mechanistic studies of the MOA of cenobamate have been published [90, 91]. Two MOAs emerge from these studies: (1) an inhibitory effect on the persistent sodium current (I_NaP) and (2) potentiation of synaptic and extra-synaptic GABAergic inhibition by a non-benzodiazepine mode of action [21, 92].

The persistent sodium current (I_NaP) is an important activity-dependent regulator of neuronal excitability [93]. As shown in Fig. 5A, B, I_NaP is upregulated in monogenetic epileptic syndromes and TLE and thought to be critically involved in neuronal hyperexcitability associated with epilepsy [94]. Cenobamate inhibits the non-inactivating I_NaP more potently than the transient Na⁺ current (I_NaT) [90], which likely contributes to the ability of this drug to suppress sustained depolarizations while sparing single action potentials and low-frequency firing. In addition, cenobamate was shown to enhance the inactivated state of voltage-gated Na⁺ channels [90].

Fig. 5.

Perceived mechanisms of action of cenobamate. Cenobamate has as a dual mechanism of action: preferential inhibition of the persistent sodium current (I_NaP) while sparing the transient sodium current (I_NaT), combined with positive allosteric modulation of synaptic and extra-synaptic GABA_A receptors, leading to an increase in phasic and, more potently, tonic inhibition. Inhibiting I_NaP and augmenting the hyperpolarizing GABA_A-mediated tonic current are effective in preventing the membrane potential of the principal cell from developing a paroxysmal depolarization shift and reaching the threshold for firing a burst of action potentials and thus contributing to epileptic activity [92]. These combined mechanistic effects of cenobamate are unique among the presently known antiseizure medications and could explain the broad-spectrum effect of cenobamate in animal seizure models and its efficacy for focal seizures in humans. A Normal function of persistent sodium current (I_NaT). B Increase in persistent sodium current by channelopathies or temporal lobe epilepsy (TLE). C A GABAergic nerve terminal with a synaptic and extra-synaptic GABA_A receptor. Synaptic GABA_A receptors function to mediate rapid (phasic) inhibition (i.e., the inhibitory postsynaptic potential), triggered by relatively short-lived peak levels of GABA acting directly at the synapse. In contrast, extra-synaptic GABA_A receptors mediate a persistent tonic inhibition via the activity of low levels of ambient GABA that have escaped presynaptic GABA transporter reuptake. These unique properties of synaptic and extra-synaptic GABA_A receptors are conferred by their specific subunit composition (as illustrated in C), which dictates the GABA binding affinity, kinetics, and subcellular localization [95]. See text for details

As shown in Fig. 5C, synaptic and extra-synaptic GABA_A receptors mediate two distinct modes of inhibition: fast phasic inhibition at synapses and slower tonic inhibition via receptors outside synapses [95]. Cenobamate acts as a positive allosteric modulator of GABA_A receptors in hippocampal neurons, with effects on both synaptic and, more potently, extra-synaptic GABA_A receptor isoforms that mediate phasic and tonic inhibitory currents, effects that were not antagonized by the benzodiazepine antagonist flumazenil [91]. In a barbiturate-like effect, cenobamate has been reported to displace the binding of TBPS (t-butylbicyclophosphorothionate) to the picrotoxin site of the GABA-gated Cl⁻ channel, whereas the binding of GABA, muscimol, flunitrazepam, and Ro-15-1788 (flumazenil) to GABA_A receptors was not inhibited [96]. Thus, the effect of cenobamate on GABA_A receptors resembles that of other alkyl-carbamates and phenobarbital [21].

Therefore, it is likely that the combined (and perhaps synergistic) effect of cenobamate on GABA-mediated inhibition and on I_NaP, which occurs at similar, clinically relevant concentrations (see Sect. 5), contributes to its broad antiseizure profile and impressive clinical efficacy [87, 92]. However, this mechanistic profile is not unique. It is also shared, at least in part, by topiramate (TPM), which both blocks I_NaP and enhances GABA-mediated inhibition [93, 97, 98]. However, TPM has a different efficacy profile in preclinical models than cenobamate (Tables 1, 2). These differences may be explained by the additional cellular effects of TPM (i.e., inhibitory effects on voltage-gated Ca²⁺ channels and ionotropic glutamate receptors) that are not known to be shared by cenobamate [87]. Furthermore, differences in the degree of blocking I_NaP may be involved. Thus, cenobamate has been found to block both the non-inactivating and the slowly inactivating current component of the I_NaP with an half-maximal inhibitory concentration (IC₅₀) of 50–70 μM, while TPM is a mild blocker at clinically relevant concentrations in both step and ramp protocols, with an effect size that is limited to a maximum of −30%; thus, TPM only exerts a partial block of I_NaP [93].

Given the current dearth of mechanistic information, we cannot rule out the possibility of other, yet unknown mechanisms that may critically contribute to the unique efficacy of cenobamate. In view of the monoaminergic effects of YKP10A (solriamfetol; see Sect. 2), it would be interesting to examine whether cenobamate also affects dopamine and norepinephrine transporters. Furthermore, given the activating effect of the alkyl-carbamate retigabine on neuronal K_v7.2/7.3 potassium channels [99], this potential target should be examined when further exploring the pharmacology of cenobamate. So far, it has only been shown that cenobamate has little effect on K_V7.1 potassium channels (IC₅₀ 1.3 mM) that are not expressed in the brain and are also not affected by retigabine [93]. At high supratherapeutic concentrations, cenobamate inhibits L-type calcium currents with an IC₅₀ of 350 μM [93].

Previously designed tetrazole derivatives with anticonvulsant activity have been reported to act as antagonists at glutamate receptors [57, 59]. Examples are tezampanel (LY293558), a 5-substituted tetrazole derivative that acts as an antagonist at the AMPA and kainate subtypes of glutamate receptors [100], and LY2333053, which is a selective NMDA receptor antagonist [101]. Furthermore, two anticonvulsant alkyl-carbamates, felbamate and meprobamate, have been reported to block NMDA receptors [21]. Thus, cenobamate should be examined for effects on glutamate receptors.

Pharmacokinetic/Pharmacodynamic Modeling of Cenobamate

Using the MES model in mice and rats, Bettio et al. [102] performed pharmacokinetic/pharmacodynamic (PK/PD) modeling with various ASMs, including cenobamate. The oral ED₅₀ of cenobamate was 7.05 in mice and 6.0 in rats, determined 2 h after administration. EC₅₀ values for plasma and brain (determined shortly after the MES test) were 66.0 µM (plasma) and 24.9 µM (brain) in mice and 15.9 µM (plasma) and 10.7 µM (brain) in rats, which are clinically relevant concentrations of cenobamate (see below). The brain:plasma ratio at ~ 2 h after oral administration was 0.38 in mice and 0.67 in rats. In the study of Palamanda et al. [103], the brain:plasma ratio of 1.14 was determined 1 h after a single dose of cenobamate, i.e., at time of maximal drug concentration. The brain:plasma ratio of cenobamate decreased thereafter (the elimination half-life of cenobamate is only 2.9 h in rats), explaining the lower brain:plasma ratios determined by Bettio et al. [102] at ~ 2 h after oral drug administration.

Green and Kamin [104] determined a reference plasma concentration range for cenobamate by analyzing steady-state trough plasma concentrations during the maintenance phase of two placebo-controlled clinical studies [3, 4] in patients who achieved 50% seizure reduction or seizure freedom. The range of concentrations during the maintenance phase for 90% of patients with seizure freedom ranged from 5 to 35 µg/mL (20–130 µM). A similar range was reported by Steinhoff et al. [105]. The brain:plasma ratio of cenobamate is not known for humans, but in rats, it is 1.14 at the time of maximal drug concentration [103].

Assuming that the brain:plasma ratio of cenobamate is similar in humans and rodents (as shown for numerous ASMs [106]), trough brain concentrations of this drug during maintenance treatment will be in the range of 20–130 µM and display considerably higher at the time of maximal drug concentration. Following multiple dosing, maximum concentration values at doses of 50–400 mg/day ranged between 19 and 172 µM [107]. Thus, only effects on molecular targets of cenobamate that occur within (or below) this concentration range are therapeutically relevant. Cenobamate blocks both the non-inactivating and the slowly inactivating current component of the I_NaP with an IC₅₀ of 50–70 μM [90], which is in this relevant concentration range. Similarly, cenobamate potentiated GABA-induced currents in acutely isolated CA3 pyramidal cells with an EC₅₀ of 164 μM and enhanced extra-synaptic tonic GABA_A currents in hippocampal slices with an EC₅₀ of 36 μM [91], which are clinically relevant drug concentrations.

Clinical Development of Cenobamate

The clinical development of cenobamate has been highly unusual in that there have been no phase III efficacy studies, reflecting its differentiating efficacy. Following a positive phase IIa proof-of-concept study of the safety, tolerability, and efficacy of cenobamate in the photoparoxysmal response, intermittent, photic stimulation study paradigm [108], there were two phase II RDBPC studies of cenobamate add-on treatment in adults with focal DRE [109]. Both enrolled patients with resistant focal DRE who had not responded to two or more antiseizure medications and had high baseline seizure frequency while treated with one to three concomitant ASMs. The first study evaluated cenobamate 200 mg compared to placebo, over a short treatment that included 6 weeks of titration and 6 weeks of maintenance treatment periods. Seizure freedom rate during the maintenance period was 28%, compared with 9% placebo [4]. The second study had a traditional phase IIb RDBPC add-on design with 6 weeks of titration and 12 weeks of maintenance treatment, evaluating 100-, 200-, and 400-mg/day doses [3]. As noted, seizure freedom in the 400-mg/day arm was 21%, compared with 1% in the placebo arm. The FDA did not request a phase III efficacy study, and no phase III RDBPC efficacy study has ever been done. This is unprecedented in the history of ASM development. Safety issues included the occurrence of DRESS (Drug Reaction with Eosinophilia and Systemic Symptoms) among the first 954 individuals exposed to cenobamate, in whom starting doses ranged from 50 to 200 mg/day and titration increases occurred in 5- to 7-day steps. This led to a third pivotal study, a large open-label safety study (n = 1339) with long treatment (median treatment duration 30 months), using a titration schedule with low-dose initiation and a slow-dose increment [6]. No cases of DRESS occurred with this titration schedule, which became the recommended titration schedule.

The short-term efficacy in the phase II trials largely held up in long-term open-label studies. In the open-label extension study of the pivotal study, with a median treatment duration of 4.5 years, 10% of all patients were seizure free during year 4 of the study, and 24% patients had a > 90% seizure frequency reduction [5]. In a pivotal, phase III, open-label safety study of cenobamate with a median treatment duration of 30 months, 13% of patients achieved seizure freedom and 40% achieved a > 90% seizure frequency reduction during the maintenance treatment period [6]. Subsequent real-world experience studies have broadly confirmed these results but have also highlighted some challenges, including potential for drug–drug interactions and slow titration [8–11, 107, 110].

Discussion

Epilepsy, unlike many other central nervous system disorders, has benefited from the highly predictive animal models that have been developed over recent decades [111]. However, while rodent models of seizures and epilepsy carry a high predictive value for antiseizure effects of drugs in humans, they have been less predictive in differentiating highly effective from less effective drugs in patients with DRE. This is despite the efforts of the NINDS-funded ETSP to include various rodent models that are classified into screening models used for identification (e.g., MES, PTZ, 6-Hz) and more complex chronic models for differentiation of potential therapies for DRE [83]. Until cenobamate, none of the new ASMs tested in recent years with these models significantly changed the proportion of patients with focal DRE [1, 15]. It was the combined efforts of the ETSP and in-house testing at SK Biopharmaceuticals that helped to discover cenobamate, a novel ASM with unsurpassed efficacy in patients with focal DRE. The lithium-pilocarpine rat model of induced seizures or SE is not part of the ETSP but was crucial in the decision to choose cenobamate as the lead compound for further development. Cenobamate is the most potent ASM in this model, while other carbamates were either ineffective (carisbamate) or much less effective (felbamate, > 10 times less effective). This model may therefore be important in future testing of ASMs for DRE treatment.

Cenobamate was also highly potent in the amygdala and hippocampal kindling models of TLE, only surpassed by retigabine. The kindling model is the only model that adequately predicted the clinical utility of most ASMs against focal and secondarily generalized convulsive seizures as well as the lack of clinical efficacy of NMDA receptor antagonists [78, 111]. The lamotrigine-resistant amygdala-kindled rat is included in the differentiation phase of the ETSP [83] and replaced the hippocampal-kindled rat that was used by the ETSP at the time of cenobamate development [72]. The high potency and efficacy of cenobamate in the kindling models correctly predicted its clinical utility.

Interestingly, when combined potency rankings in the kindling, pilocarpine, and 6-Hz models of drug-resistant focal seizures were used to compare cenobamate with numerous other clinically established ASMs, cenobamate displayed by far the highest potency of all drugs across the four models. This indicates that such a combined ranking has a high predictive validity and provides added value when differentiating ASMs.

Conclusions

Cenobamate is an important example of how chemocentric and phenotypic drug discovery can be combined in the search for novel, highly effective, first-in-class therapies. For cenobamate discovery, known alkyl-carbamates with anticonvulsant activity served as a starting point for optimization, and the numerous resulting analogs were characterized by rodent seizure models.

It remains to be shown whether the high clinical efficacy of cenobamate in patients with focal DRE is a consequence of its dual MOA or whether other, yet unknown MOAs contribute, which we consider likely. Because of cenobamate, seizure freedom has become attainable in a significant proportion of patients with focal DRE [1]. In view of other recent advances, such as fenfluramine for patients with Dravet syndrome, and a rich pipeline of new compounds under clinical development [112], the long-sought-after breakthrough in the treatment of epilepsy may finally be in sight.

Funding

Open Access funding enabled and organized by Projekt DEAL. Open Access funding enabled and organized by Project DEAL. No funding was received for the preparation of this article.

Declarations

Conflicts of interest/competing interests

Yong Moon Choi founded SK Biopharmaceuticals (Fairfield, NJ, USA and Daeduk/Seoul, South Korea) in 1993, was Executive Vice President of the company during the development of cenobamate (till 2006), and, since 2008, is the CEO of Bio-Pharm Solutions (Seoul, South Korea). H. Steve White directed the NINDS ETSP at the University of Utah during the development of cenobamate. He has received grant funding from Neurelis and consultant fees from BioPharm Solutions and JAZZ Pharmaceuticals, and speaker honoraria from SK Life Sciences, Takeda, and UCB Pharma. He is also a co-founder of NeuroAdjuvants, Inc., Salt Lake City, UT, USA. Pavel Klein took part in all three pivotal clinical trials of cenobamate, co-authored two of the three foundational manuscripts (Krauss et al., 2020; Sperling et al., 2020), with access to complete study reports for both trials, and lead-authored the open-label extension manuscript of the most important pivotal trial (Klein et al., 2022), in addition to co-authoring a host of secondary cenobamate efficacy and safety manuscripts derived from the pivotal trial data. He has served as a consultant, advisory board member, or speaker (2020–25) for Abbott, Angelini Pharma, Aquestive Therapeutics, Arvelle Therapeutics, Aucta Pharmaceuticals, Dr. Reddy’s, Eisai, GRIN Therapeutics, Jazz Pharmaceuticals, Longboard Pharmaceuticals, Neurelis, Inc., Neurona Therapeutics, Paladin Pharma, SK Life Science, Sunovion, UCB Pharma, UNEEG, UniQure, and Xenon Pharma, is a member of the Data and Safety Monitoring Board for Neurona Therapeutics, and is a member of the Medical Advisory Board of Stratus and the Scientific Advisory Boards of OB Pharma and of NEUmirna, is the CEO of PrevEp, Inc. (Bethesda, MD, USA), and has received research support from CURE/Department of Defense and from the NIH/SBIR. Wolfgang Löscher is a member of the External Consultant Board of the NINDS ETSP program. He is a co-founder and CSO of PrevEp, Inc. (Bethesda, MD, USA). He has received in the past 5 years consultancy fees from Lundbeck, Angelini, Clexio, Selene, Axonis, SynapCell, Sintetica, ND Capital, Atlas Venture, Cogent Biosolutions, Ovid, Idorsia, and Addex.

Ethics approval

Not applicable.

Consent to participate

Not applicable.

Consent for publication

Not applicable.

Availability of data and material

No primary data were collected for this study. The original data extraction is available from the corresponding author upon reasonable request.

Code availability

Not applicable.

Authors’ Contributions

WL: conceptualization (lead); investigation (equal); visualization (lead); writing (original draft preparation [lead]); writing (review and editing [equal]); YMC: writing (review and editing [equal]); investigation (equal). HSW: writing (review and editing [equal]); investigation (equal). PK: writing (review and editing [equal]); investigation (equal). All authors reviewed and approved the final paper.