The discovery and development of perampanel for the treatment of epilepsy

Abstract

Introduction: Perampanel is a novel AMPA receptor antagonist, approved in over 35 countries as an adjunctive therapy for the treatment of partial-onset seizures with or without secondarily generalized seizures in patients with epilepsy aged 12 years and older (18 years and older in Canada). These countries include the members of the European Union, the USA, Canada and Switzerland. The AMPA receptor antagonist, perampanel, is the first approved antiepileptic drug to inhibit excitation of postsynaptic membranes through the selective inhibition of glutamate receptors.

Areas covered: This drug discovery case history focuses on the discovery and profiling of perampanel. It analyzes the pharmacological, behavioral and molecular mechanisms of perampanel and how they contribute to the therapeutic benefits of the drug. The article is based on the data reported in published preclinical and clinical studies, product labels and poster presentations.

Expert opinion: Preclinical studies of perampanel have identified its broad-spectrum antiseizure effects in acute seizure models, with a narrow therapeutic index in the rotarod test similar to other AMPA receptor antagonists. This narrow therapeutic index is a potential problem for AMPA receptor antagonists. However, the discovery that perampanel has a very long half-life in humans, with gradual accumulation in plasma, could contribute to the development of tolerance. This, coupled with the identification of an optimal dosing strategy for individual patients, may help to maximize the utility of perampanel in the treatment of epilepsy.

Keywords::

• AMPA receptor antagonist
• anticonvulsant agent
• partial-onset seizures
• perampanel
• refractory epilepsy

1. Introduction

Epilepsy is a common neurological disorder, with estimated median prevalence rates for the active condition ranging from 4.9 patients with epilepsy per 1000 of the overall population in developed countries to 12.7 patients with epilepsy per 1000 of the overall population in rural areas of developing countries [1]. Despite, the introduction of several second-generation antiepileptic drugs (AEDs) over the past two decades, one-third of patients still have poorly controlled epilepsy and continue to experience refractory seizures [2], although a recent report considering prognosis of epilepsy data from longitudinal cohort studies found that 65 – 85% of patients eventually enter long-term remission [3].

While previously available AEDs mainly tended to exert their effects through modulation of sodium channels or enhancement of GABA-related mechanisms, perampanel (Fycompa®, Eisai) is the first clinically available selective AMPA receptor antagonist and the first AED to selectively inhibit excitatory postsynaptic function [4]. Perampanel is approved in the European Union, the USA, Canada and Switzerland as an adjunctive therapy for the treatment of partial-onset seizures with or without secondarily generalized seizures in patients with epilepsy aged 12 years and older (18 years and older in Canada) [5,6].

1.1 The AMPA receptor as a target for epilepsy research

Among the excitatory postsynaptic targets for epilepsy drug development, the AMPA glutamate receptor has received particular attention for more than two decades. This receptor is known to have a substantial role in glutamate-mediated excitatory neurotransmission and synaptic plasticity [7-9], as well as in various types of neurodegenerative conditions [10]. The AMPA glutamate receptor has also been specifically linked to epilepsy in the human hippocampus [11,12], and animal studies have indicated that it has a key role in seizure-induced neuronal injury [13,14], epileptogenesis [15] and seizure activity and expression [16-18]. Furthermore, prototype competitive and noncompetitive AMPA receptor antagonists have demonstrated anticonvulsant activity in a variety of animal models [17,19-21]. Overall, this information suggests the potential of AMPA receptor antagonists as neuroprotective, antiepileptogenic agents alongside already established AEDs in the clinical setting.

1.2 Early evaluation of AMPA receptor antagonists

The AMPA receptor is a type of non-NMDA glutamate receptor. The basic structure of selective competitive non-NMDA receptor antagonists was discovered in 1988 [22], whereas the basic structure of noncompetitive AMPA receptor antagonists was derived from a muscle relaxant [23] that was identified as an AMPA antagonist in 1990 [24]. Since then, pharmaceutical companies have attempted to develop AMPA receptor antagonists as therapeutic agents as described in the following paragraphs.

Prototype competitive AMPA receptor antagonists, such as 2,3-dihydroxy-6-nitro-7-sulfamoyl-benzo[f]quinoxaline (NBQX), showed high affinity for the glutamate binding site of AMPA receptors [22,25]. However, poor blood–brain barrier (BBB) penetration was a major problem and such compounds were also poorly soluble at a neutral pH range, resulting in precipitation in the kidneys at therapeutic plasma concentrations [22,25]. Improvement of solubility through the introduction of a polar moiety was explored in an attempt to identify candidate compounds for clinical evaluation, but modification further reduced BBB permeability [25]. Meanwhile, prototype noncompetitive AMPA receptor antagonists, such as the 2,3-benzodiazepine-type compound GYKI52466, demonstrated a modest in vitro inhibitory effect, but good oral bioavailability and in vivo efficacy, indicating effective BBB penetration [25]. However, such compounds were found to have short half-lives [26]. In addition, poor tolerability in terms of CNS-depressant effects was an issue with both competitive and noncompetitive antagonists [21].

Some AMPA receptor antagonists did progress to clinical testing in neurological indications; for example, ZK 200775 was tested in patients suffering from acute ischemic stroke, although the study was terminated prematurely due to strong CNS-depressant effects [27]. Talampanel was evaluated in various disease conditions including epilepsy, brain tumor and amyotrophic lateral sclerosis [28-30]. However, while a positive efficacy signal was demonstrated in epilepsy [28], further clinical development was suspended due to talampanel’s short half-life [26]. Therefore, new AMPA receptor antagonists, with novel chemical structures, were required to overcome the various issues encountered with the early compounds.

2. Discovery and preclinical development of perampanel

2.1 Identification of perampanel

Two high-throughput screening campaigns were run to discover a new chemical template for an AMPA receptor antagonist: the [³H]AMPA binding assay and the neuroprotection assay against AMPA-induced rat cortical neuronal cell death (previously reviewed by Rogawski et al. [31]). The neuroprotection assay identified several promising compounds, including 2,4-diphenyl-4H-[1,3,4]oxadiazin-5-one, which was selected for further development based on its dissimilarity to known AMPA receptor antagonists, as well as its potential for chemical modification [31,32]. Initial attempts at modification included changing of the central ring structure to increase both its inhibitory activity at the AMPA receptors and its metabolic stability. A major milestone during the discovery program was the identification of 1,3,5-triaryl-1H-pyridin-2-one, a compound with effective antagonistic activity at AMPA receptors and good metabolic stability in human liver microsome assays [32]. Furthermore, the ‘3’ position of 1,3,5-triaryl-1H-pyridin-2-one compounds could be easily modified or added to using Suzuki–Miyaura coupling [32,33]. The identification of this core structure accelerated the understanding of the structure–activity relationship of AMPA receptor antagonists and culminated in the discovery of the unique structure of perampanel [32].

2.2 Pharmacology of perampanel based on preclinical studies

Perampanel inhibited AMPA receptor-mediated responses during neurotransmission in rat hippocampal slices and in primary cultures of rat cortical neurons at an IC₅₀ of 230 and 93 nM, respectively [34,35]. In these studies, perampanel did not displace [³H]AMPA binding but instead exerted a noncompetitive effect [34,35]. Perampanel also demonstrated high selectivity for the AMPA receptor, because it did not inhibit kainate- and NMDA-receptor-mediated responses in rat hippocampal neurotransmission or in an NMDA-induced Ca²⁺ influx assay at 10 and 30 M, respectively [34,35]. This suggested a low risk of the psychotomimetic side effects that are known to be elicited by inhibition of NMDA receptors [36]. Ligand binding assays that included 63 physiologically relevant enzymes, ion channels and neurotransmitter transporters were performed to confirm the selectivity of perampanel; at a dose of 1.25 M, perampanel did not show > 50% replacement of ligand binding among any of the 63 targets [32].

Although the binding site of perampanel remains unknown, results from a [³H]perampanel binding assay indicated that it binds to AMPA receptors in rat forebrain membranes with a dissociation constant of 59.8 ± 5.2 nM and maximum binding of 3.2 ± 0.1 pM/mg [35]. [³H]Perampanel binding was not significantly influenced by glutamate (1 mM), AMPA (0.1 mM) or the competitive AMPA receptor antagonist NBQX (0.1 mM). However, [³H]perampanel was displaced by the known selective noncompetitive AMPA receptor antagonists, GYKI52466 (mean inhibition constant ± standard error of the mean, 12.4 ± 1 μM) and CP-465,022 (11.2 ± 0.8 nM), indicating a shared binding site.

2.3 Antiseizure profile of perampanel

2.3.1 Acute seizure models

A range of animal models are widely used for drug screening in the field of epilepsy [37]. For example, a protective effect in the MES test can be used to indicate whether a compound has an effect on seizure spread and possible clinical utility in human generalized tonic–clonic seizures. Meanwhile, the subcutaneous pentylenetetrazole (PTZ) seizure test can be used to indicate whether a compound raises the seizure threshold and is useful for identifying drug candidates for the treatment of myoclonic petit mal seizures [38].

AMPA receptor antagonists have been shown to inhibit seizures in various animal models [19-21,35]. Perampanel demonstrated antiseizure activity in the MES and PTZ tests, as well as protecting against audiogenic seizures in DBA2 mice [35]. ED₅₀ values (i.e., the effective dose in 50% of the population) for perampanel in each of the MES-induced, PTZ-induced and audiogenic seizure tests were 1.6, 0.94 and 0.47 mg/kg, respectively, and were consistently lower than those reported with carbamazepine or valproate. Overall, perampanel was deemed to be a potent anticonvulsant in these models and the lower ED₅₀ values had future implications in terms of pill size and dosing because those AEDs with higher ED₅₀ values either required formulation as a large pill or required multiple tablets to be taken.

Perampanel also inhibited seizures in the 6 Hz electroshock-induced test [35], a refractory psychomotor seizure model [39]. Interestingly, many first- and second-generation AEDs lose their protective effect in this model when the stimulus intensity is increased from 32 to 44 mA [39], whereas perampanel was found to inhibit seizures at the higher stimulus intensity with only a slight elevation of the effective dose (ED₅₀ increased from 2.1 to 2.8 mg/kg) [35]. These data indicate that in the tested animal models perampanel showed a similar spectrum of antiseizure activity to valproate [35].

2.3.2 Chronic seizure models

The amygdala kindling model is a chronic seizure model [40]. As the seizures produced are considered equivalent to human complex partial seizures with secondary generalization (SG), this model is often used for the evaluation of anticonvulsants indicated for the treatment of partial-onset seizures. Antiseizure activity in this model is considered to be more predictive of clinical efficacy and toxicity than antiseizure effects observed in acute models.

Oral perampanel 5 or 10 mg/kg increased the after-discharge threshold (ADT) in a rat amygdala kindling model and significantly reduced motor seizure duration, after-discharge duration and seizure severity when an electrical stimulus intensity higher than the ADT was applied [41]. Stimulation at an intensity higher than the ADT often weakens the effect of AEDs, but perampanel showed significant efficacy under such conditions [35,41]. These results indicated that perampanel could inhibit seizure propagation in addition to elevating seizure threshold, suggesting efficacy in secondary generalized seizures. However, while efficacious in a rat amygdala kindling model, oral perampanel was inactive at doses of 1, 3 and 10 mg/kg in an absence seizure model (genetic absence epilepsy rats from Strasbourg [GAERS]) [35].

2.3.3 Status epilepticus

A lithium-pilocarpine rat model of status epilepticus has been used to explore the therapeutic window of perampanel [42]. Monotherapy with intravenous (i.v.) perampanel 8 mg/kg terminated seizures when administered at 10, 30 or 60 min after seizure onset in all tested animals and when administered at 90 min after seizure onset in some of the tested animals. In contrast, diazepam 20 mg/kg i.v. terminated seizures only when administered at 10 min after seizure onset and efficacy was not observed in all animals. Overall, perampanel terminated status epilepticus with a longer therapeutic time window than that of diazepam.

2.4 Pharmacodynamic interactions between perampanel and other AEDs based on preclinical studies

The development of AEDs with novel modes of action has contributed to a renewed interest in the potential of rational AED combination therapies to optimize efficacy and minimize adverse events (AEs), particularly in patients with refractory epilepsy [43]. As the first selective AMPA receptor antagonist, perampanel may be a suitable candidate for rational combination therapies because its discrete mode of action may complement those of other licensed AEDs [35].

Preclinical studies provided insight into the potential combination of perampanel with other AEDs. Enhanced seizure protection was observed in the 6 Hz electroshock-induced test when perampanel was combined with carbamazepine, phenytoin or valproate [35]. Similarly, pharmacodynamic interactions between perampanel and the four most commonly co-administered AEDs in the Phase III registration trials (levetiracetam, lamotrigine, carbamazepine and valproate) were confirmed in an amygdala kindling model using two-way ANOVA analysis. These effects were at least additive with some interactions being classed as synergistic [41].

2.5 CNS-depressant effects of perampanel

Given that AMPA receptors play a key role in the transmission of excitatory signals across neuronal networks; it would be anticipated that AMPA receptor antagonism will cause CNS-depressant effects. Accordingly, muscle-relaxant effects are known consequences of AMPA receptor antagonists (as noted, noncompetitive AMPA antagonists were originally derived from a muscle relaxant).

Poor separation between antiseizure effect and motor incoordination has been demonstrated in animal studies of both competitive and noncompetitive AMPA antagonists [21]. Accordingly, using the rotarod test, perampanel has been associated with dose-dependent motor impairment in mice, with a TD₅₀ value (i.e., the toxic dose in 50% of the population) of 1.8 mg/kg [35]. The protective indices (TD₅₀/ED₅₀) were 1.1, 1.9 and 3.8 in MES-induced, PTZ-induced and audiogenic seizures, respectively, indicating a narrow therapeutic margin [35]. However, this is not necessarily indicative of poor tolerability in humans; valproate has demonstrated a narrow therapeutic margin in the MES test (protective index of 2.8) [39,44] but is generally well tolerated in humans [45], whereas gabapentin and pregabalin had wide therapeutic margins in rodent studies [46,47] but are associated with CNS-depressant effects in humans [48-51]. Indeed, the clinical therapeutic index of AEDs cannot be reliably inferred from preclinical studies as pharmacokinetics and dose regimens may also impact on the potential for CNS-depressant effects.

2.6 Pharmacokinetics of perampanel based on preclinical studies

As discussed, pharmacokinetic issues relating to AMPA receptor antagonists have included poor BBB penetration (competitive AMPA receptor antagonists) and short half-lives (2,3-benzodiazepine AMPA receptor antagonists) [25]. In contrast, animal studies indicated that perampanel offered good BBB penetration, because the ratios of brain to plasma and of cerebrospinal fluid to unbound plasma concentrations were approximately one in male mice or rats [32]. Furthermore, studies in human cell lines indicated that perampanel is not a substrate of P-glycoprotein (P-gp) or the breast cancer resistance protein [5], which are both efflux transporters in the BBB. Therefore, increased levels of such transporters (notably, increased P-gp levels, which are thought to be associated with drug resistance in refractory seizures [52]), are unlikely to limit access of perampanel to the brain.

While perampanel exhibited favorable pharmacokinetic properties in experimental animal models following oral dosing, its terminal half-life was relatively short (rat 1.67 h, dog 5.34 h, monkey 7.55 h) [35]. However, a very low metabolic rate in human liver microsomes (0.009 μl/min/mg protein) suggested that perampanel might have a longer terminal half-life in humans, such that it might be suitable for once-daily dosing [32,35].

Perampanel is mainly metabolized by cytochrome P450 3A4 (CYP3A4) and CYP3A5. AMPA receptor antagonist activity of the metabolites was weaker than that of the parent compound, indicating that the contribution of metabolites to the therapeutic activity of perampanel is negligible [5]. No human-specific metabolites were identified.

3. The clinical development of perampanel

3.1 Phase I clinical trials

Oral perampanel is rapidly and almost completely absorbed in humans, with low systemic clearance and high relative bioavailability [53]. In-line with expectations from human liver microsome assays, the terminal half-life determined in Phase I clinical studies ranged from 53 to 136 h (average of 105 h from 19 studies) [31]. Perampanel plasma concentrations gradually accumulated in healthy male subjects (due to perampanel’s long half-life), reaching steady state after 2 weeks of repeated dosing [53].

The results of repeated-dose studies indicated that perampanel is associated with CNS-depressant effects as measured using Saccade eye velocity and the Bond and Lader sedation subscale. Early CNS-depressant effects occurred from the first dose of 2 mg or more, but after 2 weeks, sedation was only observed following repeated administration of the maximum tested daily dose of 6 mg [53]. It is suggested that gradual accumulation of perampanel plasma concentrations caused self-titration and thus helped develop tolerance to the sedative effect of the drug during repeated dosing.

3.2 Phase II clinical trials

Two randomized, double-blind, placebo-controlled Phase II clinical trials were designed to evaluate the safety and tolerability of perampanel in a population of adult patients with refractory partial-onset seizures despite receiving 1 – 3 AEDs [54]. Study 206 (ClinicalTrials.gov identifier: NCT00144690) evaluated low doses of perampanel and indicated that dosing up to 4 mg/day (using once- or twice-daily dosing) was well tolerated. Subsequently, study 208 (NCT00416195) evaluated higher once-daily doses up to 12 mg and indicated that most patients were able to tolerate once-daily dosing of ≥ 6 mg. While these studies were not powered to confirm efficacy, there was a preliminary indication of efficacy according to data obtained from patient diaries. These studies informed the design of subsequent Phase III clinical trials.

3.3 Phase III clinical trials

The efficacy of adjunctive perampanel at the predicted no-effect daily dose of 2 mg and the predicted effective doses of 4, 8 and 12 mg were further evaluated in three multinational, multicenter, randomized, double-blind, placebo-controlled Phase III registration trials: studies 304 (NCT00 699972) [55], 305 (NCT00699582) [56] and 306 (NCT007 00310) [57]. These studies enrolled patients aged ≥ 12 years who were experiencing simple partial-onset seizures with motor signs or complex partial-onset seizures (CP) (with or without SG) despite receiving 1 – 3 AEDs. Following a 6-week baseline period, patients received once-daily placebo or perampanel 8 or 12 mg (studies 304 and 305) or once-daily placebo or perampanel 2, 4 or 8 mg (study 306), administered at bedtime over a 19-week double-blind treatment phase (comprising a 6-week titration period [weekly dose adjustments in 2-mg increments] and a 13-week maintenance period).

In total, 1478 patients were included in the intent-to-treat analysis sets of studies 304 (n = 387), 305 (n = 386) and 306 (n = 705) [55-58]. In all three studies, randomized doses of perampanel 4, 8 or 12 mg were associated with significantly greater median reductions in the frequency of all partial-onset seizures per 28 days compared with placebo (double-blind treatment phase versus baseline; primary endpoint outside the European Union in all studies; pooled data, perampanel 4 mg −23.3%, 8 mg −28.8%, 12 mg −27.2%, placebo −12.8%, p < 0.01 for each dose versus placebo). There were similar reductions in the frequency of CP+SG seizures (double-blind treatment phase versus baseline; pooled data, perampanel 4 mg −31.2%, 8 mg −35.6%, 12 mg −28.6%, placebo −13.9%, p < 0.001 for each dose versus placebo). Furthermore, in studies 305 and 306, 50% responder rates were also significantly greater with perampanel 4, 8 or 12 mg than with placebo (maintenance period versus baseline; primary endpoint in the European Union) and across the pooled data from all three studies (perampanel 4 mg 28.5%, 8 mg 35.3%, 12 mg 35.0%, placebo 19.3%, p < 0.05 for each dose versus placebo). In study 304, 50% responder rates were not significant for perampanel 8 or 12 mg versus placebo.

Pharmacokinetic/pharmacodynamic analyses of the pooled Phase III data have indicated a significant linear relationship between perampanel plasma concentrations and efficacy [59]. This relationship was not affected by the presence or absence of specific concomitant AEDs, although perampanel plasma concentrations were reduced by AEDs known to induce the CYP3A4 enzyme, which plays a key role in the hepatic metabolism and elimination of perampanel. Such enzyme-inducing AEDs include carbamazepine, which was amongst the most commonly administered concomitant AEDs in the Phase III trials. Further analyses have indicated that the efficacy of perampanel was broadly consistent across subgroups of patients receiving each of these four concomitant AEDs, except that effects were relatively diminished in patients receiving carbamazepine [60]. This is consistent with a reduction in perampanel plasma concentrations in the presence of enzyme-inducing AEDs. Such pharmacokinetic interactions, together with the frequent use of polypharmacy, impede the clear understanding of the extent of pharmacodynamic interactions between perampanel and other AEDs.

There were 1480 patients included in the safety analysis sets of studies 304 (n = 388), 305 (n = 386) and 306 (n = 706) [55-58]. In general, perampanel 2 – 12 mg had an acceptable safety profile, and most AEs were mild or moderate in intensity. The most commonly reported AEs with perampanel in all three studies occurred in the CNS and were dose dependent: dizziness (10.0 – 47.9%) and somnolence (9.3 – 18.2%) [55-58]. Other AEs reported in ≥ 5% of patients treated with perampanel in any of the studies were headache, fatigue, irritability, nausea and fall. Across the three studies, AEs necessitated discontinuation of perampanel in 99 patients (9.5%) and of placebo in 21 patients (4.8%). The AEs most frequently resulting in the discontinuation, reduction or interruption of perampanel dosing included aggression, ataxia, blurred vision, convulsion, dizziness, dysarthria, fatigue, headache, hypersomnia, somnolence and vertigo. Prescribing guidelines recommend that patients taking perampanel should be monitored for psychiatric events since in the Phase III trials, more patients receiving perampanel experienced psychiatric and behavioral AEs (hostility or aggression) compared with the placebo group. For further information on the safety and tolerability profile of perampanel, the reader is referred to the recent review by Serratosa et al. [61], which provides an overview of the AE profile based on the two Phase II dose-finding studies, the three Phase III registration trials and the two ongoing extension studies.

Among frequent AEs and AEs of special interest, the predicted probability of dizziness, somnolence, fatigue, irritability, gait disturbance, weight increase, dysarthia and euphoric mood were increased significantly in patients with higher plasma concentrations of perampanel compared with those patients who had lower plasma concentrations of perampanel (p < 0.001) [59]. In general, concomitant AEDs were not found to have any effect on the frequency of AEs. However, the likelihood of developing fatigue was increased in patients receiving levetiracetam, irritability was increased in those receiving phenobarbitol and decreased appetite was found in those receiving oxcarbazepine or primidone compared with patients receiving perampanel alone.

4. The long-term evaluation of perampanel

Extension studies have provided insight into long-term outcomes with perampanel. The extension studies 207 (NCT00368472) [62] and 307 (NCT00735397) [63] were designed to explore the long-term safety and tolerability of adjunctive perampanel in the treatment of refractory partial-onset seizures, with the secondary objective of evaluating the maintenance of efficacy.

In total, 138/180 (76.7%) patients from the earlier Phase II trials were enrolled in the extension study 207 [62]. At the time of interim analysis (performed at approximately 4 years after the start of the study), more than one-third of patients (38.4%) remained on perampanel; 69.6% of patients had > 1 years’ exposure, 50.7% had > 2 years’ exposure, 41.3% of patients had > 3 years’ exposure and 13.0% had ≥ 4 years’ exposure. The overall median change in the frequency of all partial-onset seizures per 28 days was -31.5% (entire duration of open-label treatment versus pre-perampanel baseline), and 50% responder rates at the end of years 1, 2, 3 and 4 were 43.8, 51.5, 49.0 and 50.0%, respectively. No new safety signals emerged with long-term treatment.

There were 1218/1264 (96.4%) patients from the three Phase III registration trials that were enrolled in the extension study 307 [63]. An interim analysis was performed approximately 2 years after the study start, when the median duration of perampanel exposure was 51.4 weeks. At this time, 91.4% of patients had been titrated to perampanel 10 or 12 mg (mean dose 10.1 mg) and 70.8% of patients remained on perampanel. Median changes from pre-perampanel baseline in the frequency of all partial-onset seizures per 28 days were −39.2% over weeks 14 – 26, −46.5% over weeks 40 – 52 and −58.1% over weeks 92 – 104. For the same time periods, the overall responder rates were 41.4% for weeks 14 – 26 (n = 1114), 46.9% for weeks 40 – 52 (n = 731) and 62.7% for weeks 92 – 104 (n = 59). AEs were reported in 87.4% of patients, the most frequent of which were dizziness (43.9%), somnolence (20.2%), headache (16.7%) and fatigue (12.1%).

5. Summary

Perampanel has been approved in over 35 countries including the European Union, the USA, Canada and Switzerland for the treatment of partial-onset seizures with or without secondarily generalized seizures in patients with epilepsy aged 12 years and older (18 years and older in Canada) [5,6]. Not only is it the first marketed selective AMPA receptor antagonist but it is also the first marketed selective inhibitor of excitatory postsynaptic activation [4,6].

Historically, many barriers have hindered the development of AMPA receptor antagonists as valuable drug candidates, including structure-related safety issues, short half-lives and, importantly, side effects associated with pharmacological mechanisms [21,22,25-30]. In attempts to develop an AMPA receptor antagonist to overcome these challenges, the basic structure of perampanel was discovered through high-throughput screening using AMPA-induced cell death assays in primary cultured neurons [32]. This approach identified 2,4-diphenyl-4H-[1,3,4]oxadiazin-5-one, which was selected as a starting compound, and later 1,3,5-triaryl-1H-pyridin-2-one, which was a key turning point of the project and culminated in the development of perampanel.

Preclinical studies indicated that perampanel exerts a highly selective and strong inhibitory effect on AMPA receptor activity [35]. In animal models of acute and chronic seizures, perampanel was shown to have broad-spectrum antiseizure effects that elevated the ADT and inhibited seizure propagation [35,41]. Furthermore, perampanel terminated seizures in an animal model of benzodiazepine-refractory status epilepticus [42].

Metabolic turnover of perampanel in human liver microsome assays was suggestive of a long half-life in humans [32]. This was confirmed in Phase I clinical trials, which also indicated the gradual accumulation of perampanel plasma concentrations over time, accompanied by improvements in tolerability [53]. Dose titration and administration at bedtime were used as approaches to improve tolerability in subsequent studies. Phase III registration trials provided evidence for the consistent efficacy of perampanel, combined with an acceptable safety profile [55-57]. Moreover, interim analyses of long-term extension studies indicated that efficacy may be sustained, and doses of 10 – 12 mg may be tolerated by most patients, over time [62,63].

6. Expert opinion

Over the years, many pharmaceutical companies have tried to discover an AMPA receptor antagonist for the treatment of neurological disorders. Perampanel is the first such compound to reach the market, despite its development starting later than that of many other AMPA receptor antagonists. In fact, when perampanel was identified, the AMPA receptor antagonist talampanel had already shown positive signals in a Phase II clinical trial in patients with refractory partial-onset seizures [28], although its subsequent development was later suspended. Notably, talampanel had a short half-life (3 h), resulting in the requirement for multiple daily dosing, and AEs occurred around the time that maximum plasma concentrations were reached [26].

By searching for novel compounds with different chemical structures, perampanel was developed to overcome the challenges of tolerability and multiple daily dosing. There were several notable milestones that led to the discovery of perampanel, including the identification and chemical manipulation of its basic structure and the subsequent discovery of 1,3,5- triaryl-1H-pyridin-2-one, which exhibited high metabolic stability and efficient medicinal chemistry [32]. In contrast to talampanel, perampanel was shown to have a long half-life in humans, such that once-daily dosing was sufficient [53]. By dosing before bedtime, plasma concentrations were expected to peak during sleep, which was deemed an effective approach to minimize the emergence of somnolence and dizziness associated with AMPA receptor antagonism. In addition, the long half-life was found to result in the gradual accumulation of perampanel in the plasma, which appeared to improve tolerability [53].

Also of significance was the selection of an indication for perampanel, as it had failed to demonstrate efficacy in both Parkinson’s disease [64] and neuropathic pain [65,66]. The ultimately successful selection of epilepsy was based on preclinical data from animal models of the disease, which are generally predictive of efficacy in the clinic. Perampanel was found to reduce the severity of seizures in a rat amygdala kindling model stimulated with higher stimulus intensity than the ADT [35,41]. In contrast, other AEDs (carbamazepine, lamotrigine, levetiracetam, valproate) had no inhibitory effect at the same stimulus intensity. This suggests that perampanel may strongly inhibit seizure propagation in addition to elevating seizure thresholds compared with other AEDs, supporting the potential of perampanel for the treatment of secondary generalized seizures, which are caused by seizure propagation to remote regions. Furthermore, the fact that excitatory synaptic transmission plays a significant role in seizure propagation also indicates that AMPA receptor antagonism could impact on the initiation of generalized seizures. Indeed, Phase III data have indicated that perampanel can have a significant effect on this severe seizure type [58,67].

Repeated administration in humans appears to improve the tolerability of perampanel [53]. Also, it is suggested that tolerability may be improved by slowing the rate of up-titration as observed in the conversion phase of the long-term extension Phase III (bi-weekly titration) [63] compared with the weekly forced titration in the Phase III double-blind studies [55-57]. Accordingly, the US prescribing information for perampanel suggests that the dose is generally increased no more frequently than once weekly, but a maximum of once fortnightly is recommended for elderly patients or those with mild to moderate hepatic impairment [6]. In the extension study 307, where 2-week titration intervals were used, most (91.4%) patients were able to receive treatment with perampanel 10 or 12 mg, with a mean dose of 10.1 mg [63]. Overall, the 2-week titration intervals may be appropriate to minimize tolerability issues in some patients.

Perampanel is metabolized by CYP3A4 and, therefore, its plasma concentration is reduced by the concomitant use of enzyme-inducing AEDs, such as carbamazepine. Accordingly, pooled analyses of Phase III data indicated that the therapeutic effect of adjunctive perampanel was diminished in the presence of the enzyme-inducing AED carbamazepine, compared with nonenzyme-inducing AEDs [60]. However, while enzyme-inducing AEDs may reduce plasma concentrations of perampanel, they are not thought to affect the linear exposure/efficacy relationship demonstrated across a perampanel actual dose range of 2 – 12 mg [59], and no specific pharmacodynamic interactions with different types of concomitant AEDs have been identified in the Phase III analyses. This is consistent with a preclinical rat amygdala kindling model, in which the antiseizure effects of perampanel were additively or synergistically augmented when combined with carbamazepine, lamotrigine, levetiracetam or valproate, but again no specific pharmacodynamic interactions were identified with one type of concomitant AED over any other [41].

Perampanel is the first clinically available AMPA receptor antagonist approved for the adjunctive treatment of partial-onset seizures in over 35 countries including the European Union, the USA, Canada and Switzerland [5,6] and is now under evaluation in patients with primary generalized tonic–clonic seizures (NCT01393743). Preclinical study results indicated that perampanel exerts broad-spectrum antiseizure effects; as such, a multiple-seizure population might be an appropriate target, although further clinical evaluation in this setting is warranted.

Article highlights.

• Perampanel is the first clinically available AMPA receptor antagonist and is approved as an adjunctive therapy for the treatment of partial-onset seizures, with or without secondarily generalized seizures, in patients with epilepsy aged 12 years and older (European Union, USA and Switzerland) or 18 years and older (Canada).

• The development of earlier AMPA receptor antagonists had been hindered by poor BBB permeability, structure-related toxicity and short half-lives; perampanel was discovered as a result of high-throughput screening campaigns that were designed to identify new compounds to overcome these obstacles.

• In preclinical studies, perampanel exhibited potent activity in both in vitro assays and in vivo seizure models.

• Oral perampanel was found to be rapidly and almost completely absorbed in humans, with a low systemic clearance, a high relative bioavailability and a long half-life.

• In three pivotal, randomized, double-blind, placebo-controlled Phase III registration trials in patients experiencing partial-onset seizures despite treatment with 1 – 3 antiepileptic drugs, adjunctive perampanel provided significant reductions in seizure frequency and was associated with an acceptable adverse-event profile.

Acknowledgments

The manuscript was language edited by H FitzGibbon of Complete Medical Communications, with funding from Eisai, Inc.

Declaration of interest

T Hanada is an employee of Eisai, Inc.

Notes

This box summarizes key points contained in the article.