Perampanel’s forgiveness factor in a variable medication adherence paradigm in a rat model of chronic epilepsy

doi:10.21203/rs.3.rs-2818916/v1

Background: Poor medication adherence contributes to increased morbidity and mortality in patients with epilepsy and may be under-addressed in clinical practice. Ethical concerns make it impossible to study the impact of medication nonadherence in clinical trials, but our previous work emphasizes the importance of using preclinical approaches to address these questions. With over 30 clinically available antiseizure medicines (ASM’s), it remains an important question to understand the relationship between poor adherence and seizure incidence across mechanistically distinct ASM’s, including the broad-spectrum ASM, perampanel (PER).

Methods: We formulated PER into chow pellets to deliver to rats in a 100% fully adherent or 50% variable nonadherent paradigm via our novel automated medication-in-food delivery system. Chronic oral dosing was initiated in male rats with chronic epilepsy while monitoring 24/7 for videoEEG evidence of seizures during a 4-week placebo baseline and 4-week treatment phase. PER concentrations were monitored in plasma at 1-week intervals and correlated with degree of seizure control. The relationship between missed doses and extended patterns of nonadherence were correlated with breakthrough seizures.

Results: Fully adherent rats demonstrated a median reduction in seizure frequency of 50%, whereas nonadherent rats had a median increase of 54%. Plasma concentrations of PER were stable over the 4-week treatment period in both fully adherent and nonadherent groups, with levels being twice as high in fully adherent animals. There was no correlation between a single missed dose or series of missed doses and the incidence of breakthrough seizures. However, those animals in the nonadherent group that received PER for every meal during a 24-hour period had a reduced likelihood of seizure incidence.

Conclusions. If our preclinical data is supported in the clinic, PER’s favorable pharmacokinetic profile in humans, combined with a lowered risk of breakthrough seizures suggests that it may provide a certain forgiveness factor if a dose is missed within a 24-hour window.

Adherence

epilepsy

antiseizure medicine

perampanel

breakthrough seizures

One third of people living with epilepsy cannot obtain adequate seizure control with their antiseizure medicine (ASM) (1). However, uncontrolled seizures do not necessarily equate to drug-resistant epilepsy. It is possible that poor adherence, or the extent to which a person adheres to their prescribed regimen, may negatively impact seizure control and lead to increased morbidity and mortality (2). Estimates vary but it is thought that up to 50–75% of adults with epilepsy demonstrate some degree of suboptimal adherence (e.g., < 80%) [see (3) for review]. Reasons for not taking medication can be complex and vary from person to person, demonstrating that adherence is a dynamic health behavior that is significantly influenced by socioeconomic, demographic, and therapy-related (e.g., side effects, complexity of prescribing regimens) factors (4). Regardless, the clinical implications of adjusting pharmacotherapy in response to breakthrough seizures (e.g., therapy change or dose escalation) without first addressing the underlying issues with adherence are not fully understood.

Given that clinical studies are time-consuming, costly, and unethical to perform for these types of questions, the availability of animal models has furthered our understanding of the direct relationship between nonadherence and seizure burden. Previous work in our lab utilized a clinically relevant, novel medication-in-food delivery system in a rat model of temporal lobe epilepsy (TLE) to demonstrate that correcting for nonadherence to the ASM, carbamazepine (CBZ), improves seizure control better than changing pharmacotherapy (i.e., dose escalation) (5–7). Moreover, these studies highlight that a single missed dose of CBZ can increase the risk of seizure days after the missed dose occurred (even if subsequent doses have been taken), suggesting that factors beyond a drug’s pharmacokinetics may impact seizure control. Notably, these studies were only completed using a single ASM (e.g., CBZ). With over 30 clinically available ASM’s with distinct pharmacokinetic properties and mechanisms of action (8), it remains an important question to expand upon these results with CBZ to better understand how mechanistically distinct ASM’s impact seizure control in a variable medication adherence paradigm.

Perampanel (PER) is a mechanistically novel ASM that works to prevent seizure initiation and spread at the level of postsynaptic AMPA glutamate receptors (9). Since its approval in 2012 for adjunctive treatment of focal-onset seizures (10), real world and open-label studies have since supported PER’s favorable efficacy and tolerability profile as monotherapy/first-line adjunctive therapy for focal onset and generalized tonic-clonic seizures (11). Moreover, the fact that PER is a once daily oral tablet with a long half-life [t_1/2=105 h, (12)] may allow for prolonged therapeutic concentrations in patients struggling with adherence, and therefore reduce the likelihood of breakthrough seizures after missed doses (13, 14). However, PER’s efficacy in a nonadherence paradigm remains unknown. Ethical concerns over withholding medication from patients make this impossible to study in the clinic, meaning that the precise relationship between poor adherence and seizure incidence remains unknown and underscores the need for the present investigation.

Here we characterize the efficacy of PER when administered in a variably nonadherent medication-in-food paradigm to adult male rats with acquired spontaneous recurring seizures (e.g., epilepsy). We establish novel formulation methods to produce medication-in-food that is compatible with our novel automated delivery system to test the hypothesis that poor adherence results in worsened seizure control and to define the quantitative relationship between a missed dose and seizure incidence. These studies aim to further our understanding of the efficacy of mechanistically distinct ASM’s in a nonadherence paradigm and provide insight into whether certain ASM’s may be more forgiving than others following a missed dose or series of doses.

Animals

Upon arrival, adult male CD IGS Sprague Dawley rats (150–200 g; Charles River Laboratories, Wilmington, MA, USA) were allowed a one-week acclimation period before the start of experimentation. All animals were group housed (5 rats/cage) in standard plastic cages under controlled environmental conditions in a vivarium on a 14:10 light/dark cycle and given ad libitum access to food and filtered water. Following kainic acid (KA) exposures (discussed below), animals were single housed in standard plexiglass cages with two types of enrichment (Nylabones and cardboard tubes) for the duration of the experiment.

Kainic Acid (KA) Induced-Status Epilepticus (SE)

Status epilepticus (SE) was induced using a repeated low-dose KA administration paradigm as previously described (15). All animals received an initial bolus of KA (10 mg/kg, i.p., Tocris, Bristol, UK) dissolved in sterile saline followed by a 1-hour observation period. Animals then received subsequent doses of 5 mg/kg (i.p.) every 30 minutes until SE was reached – defined as having two generalized stage 4/5 Racine seizures within a 30 minute period (16). If an animal had a stage 5 seizure before the next 30 min dosing period, the subsequent dose was dropped to 2.5 mg/kg (i.p.) to prevent mortality. All animals were continuously monitored for seizure severity and duration for three hours following SE-onset. At the end of the observation period, all animals received a bolus of Lactated Ringers (3 mL, s.c. Baxter Int. Inc., Deerfield, IL, USA) to replace fluid loss and returned to their home cage. Pedialyte™-moistened chow and Napa Nectar hydration gel (Systems engineering Lab Group Inc., Napa, CA, USA) were provided as supportive care for 7 days post-SE or until animals returned to their pre-SE weight.

Cortical EEG Electrode Implantation

Approximately three to four weeks following KA exposures, each rat was implanted with cortical EEG electrodes following surgical procedures outlined by the University of Washington IACUC rodent survival surgery guidelines. Twenty-four hours prior to surgery, animals received 2 mg Carprofen (p.o., Bio-Serv, New Jersey, USA) as prophylactic analgesia. On the day of surgery, animals were deeply anesthetized with isoflurane (1–3%, i.h.) in medical grade oxygen (flow rate = 1L/min) and the head was stabilized using a stereotaxic apparatus. Ophthalmic ointment (Patterson Veterinary Supply Inc., St. Paul, MN, USA) was applied to the eyes and the surgical site was cleaned with alternating rounds of Betadine® scrub, and 70% isopropyl alcohol before administering a combined Lidocaine/Bupivacaine (1–2 mg/kg, s.c., Hospira, Lake Forest, IL, USA) injection for local anesthesia. A 1” incision was made down the midline of the skull and a three-prong cortical electrode (MS333/1-A, P1 Technologies™, Roanoke, VA, USA) was implanted posterior to bregma with the two recording electrodes placed to the right of the sagittal suture and the ground electrode placed to the left. Four skull mount screws were placed on either side of the sagittal skull suture: two anterior and two posterior to bregma; the entire head mount was stabilized using dental acrylic. All rats were administered Carprofen (2 mg, p.o., Bio-Serv) for post-op analgesia and monitored daily during a 2-week recovery period.

Video EEG (vEEG) Recording and Seizure Detection

Approximately 8–10 weeks post KA-induced SE (Fig. 2), and 4–6 weeks post-surgery, animals were placed in individual custom-made plexiglass recording chambers (11” x 16” x 14”) and tethered to a rotating commutator (P1 Technologies™) for 24/7 vEEG monitoring. EEG data was acquired via a BioPac MP160/ EEC100 (BioPac Systems Inc., Goleta, CA, USA) system coupled to a pair of DVP 7020BE (Advantech, Milpitas, CA, USA) video capture cards. Custom software was used to synchronize video and EEG data while simultaneously controlling feeder system based on previously described methods (6). Data was acquired in 12 h epochs, each recording epoch was analyzed using an unbiased two-step protocol: 1) seizure-like events were detected using an automated rodent seizure detection software, ASSYST (Kaoskey, Thornleigh Australia), where a seizure was defined as spike wave discharges with amplitude at least twice that of background activity in awake animals with at least 100Hz frequency and duration of at least 10s; 2) all events were manually confirmed by two trained investigators and seizures were scored based on a modified Racine scale (16) defined as 1: Mouth and facial clonus; 2: Head bobbing/nodding; 3: unilateral or bilateral forelimb clonus; 4: rearing; 5: rearing with falling, loss of righting reflex, or hindlimb extension. To adequately detect treatment-related effects on seizure control, each animal had to achieve an average > 1 stage 3 or higher seizure per week during their four-week baseline recording period before enrolling into the four-week intervention arm of the study. Any animal that did not reach enrollment criteria or lost their EEG implant over the course of the study was not included in the final analysis.

PCBO and Perampanel Pellet Formulation

Placebo (PCBO) food pellets were formulated in-house using a TDP5 desktop tablet press (LFA Machines, Dallas, TX, USA) equipped with a 12 mm modified ball punch and die tooling set. Formulated chow comprised of 96% (w/w) chocolate rodent chow (Bio-Serv, Flemington, NJ, USA), 2% (w/w) each of magnesium stearate (Sigma- Aldrich, St. Louis, MO, USA) and calcium silicate (Sigma-Aldrich), and < 0.1% (w/w) of filtered water. PER-containing pellets were formulated as described, except that 0.167 mg-PER/g-pellet was thoroughly mixed with the dry Bio-Serv chow before adding the remaining components. PER was generously provided by EISAI (Eisai Co Ltd, Tokyo, Japan) for this study.

Quantitative analysis of PER in pellets and rat plasma

2-(1',6'-dihydro-6'-oxo-1'-phenyl[2,3'-bipyridin]-5'-yl)-benzonitrile-d4 (perampanel-d4), MS-grade acetonitrile (ACN) and methanol (MeOH) were purchased from Fisher Scientific (Pittsburgh, PA). Formic Acid (FA) 88% ACS grade was purchased from Sigma Aldrich (St. Louis, MO). All stock drug solutions, buffers, and HPLC mobile phase were prepared using Milli-Q grade water (Millipore, Bedford, MA). All other consumables were purchased from Fisher Scientific (Pittsburgh, PA)

Chow pellets were prepared by cutting into 4 parts and pulverizing each quarter separately using a mortar and pestle. 15–25 mg of the pulverized pellet portion was weighed out and placed into individual Eppendorf’s. 1 ml of ACN was added to each Eppendorf and vortexed for 10 minutes then centrifuged at 16 rcf for 5 minutes. 10 µl of the supernatant was then added directly into an autosampler vial containing 490 µl of ACN and 500 µl of H₂O. 5 µl of the sample was injected onto the LC-MS platform. Chromatographic separation was achieved using a Water’s Acquity UPLC BEH C8, 2.1x50mm, 1.7 µm column (Waters Corporation, Milford, MA, USA) using a gradient consisting of 0.1% formic acid in H₂O (A) and 0.1% formic acid (FA) in acetonitrile (ACN) (B) at a flow rate 0.3 ml/min (See Table 1).

A Water’s Xevo-XS coupled to a Water’s I-Class Ultra high-pressure liquid chromatography system was used (Waters Corporation, Milford, MA, USA). Analytes were monitored in electrospray positive ionization mode (ESI+), using MRM mode. Instrument settings were as follows: Capillary (kV) – 3.1, Source Temperature 150°C, Desolvation Temperature 350°C, Cone Gas Flow 150 L/Hr, Desolvation Gas Flow 1000 L/Hr, and Collision Gas Flow 0.16 mL/min. Two fragments for each analyte were used; one used for quantifying the other as a qualifier. A parent m/z of 350.2 and 354.2 fragmented to 247.2 and 251.1 was used for perampanel and parempanel-d4, respectively (see Table 2). An example of the lowest calibrator used as well as a subject sample can be found in Fig. 1B & C. Data was processed using MassLynxs (Milford, MA) and linear equations were formulated by using peak area ratios (PAR’s) allowing for 15% variability across calibrators and quality controls for acceptability.

Internal Standard (I.S.) was purchased as a 100 µg/ml in MeOH of perampanel-d4 solution which was further diluted to 100 ng/ml in ACN. Calibration Curve and QC samples were prepared by making dilutions of stock solutions containing perampanel from 1 mg/ml stocks in MeOH and stored at -20°C. Calibration curves were created by analyzing drug-free plasma samples fortified with perampanel at 0, 5, 10, 50, 100, 500, 1000 ng/ml. Quality control (QC) samples in plasma were prepared at 50 and 500 ng/ml from separate dilutions of stocks than those used for the calibration curves. Neat solutions of perampanel were made in ACN for pellet concentration determination.

Table 1

LC Gradient for LC Separation of Perampanel
Time (min)	Flow (ml/min)	% A	% B
Initial	0.300	80	20
1.00	0.300	80	20
4.00	0.300	0	100
5.00	0.300	0	100
5.10	0.300	80	20
A: 0.1% Formic Acid in Water; B: 0.1% Formic Acid in Acetonitrile

Table 2

MS Transitions for perampanel quantitation
Compound	Parent (m/z)	Daughter (m/z)	Dwell time (s)	Cone (v)	collision (eV)
Perampanel	350.2	219.1	0.072	52	32
Perampanel	350.2	247.2	0.072	52	24
Perampanel-d4	354.3	223.1	0.072	52	36
Perampanel-d4	354.3	251.1	0.072	52	32

Plasma was deproteinated prior to LC–MS analysis. 10 µl of subject plasma, calibration, or quality control samples were pipetted into a 1.2 ml Eppendorf. 10 µl of internal standard consisting of permapanel-d4 (100 ng/ml in ACN) was added to each Eppendorf and vortexed for 5 seconds. 100 µl of ACN was then added to each Eppendorf and samples were vortexed for 15 seconds and then centrifuged at room temperature for 5 minutes at 16 rcf. 80 µl of supernatant was removed and placed directly into a 450 µl nunc 96 well plate and directly analyzed. 5 µl of the sample was injected onto the LC-MS platform. All LC-MS conditions for plasma evaluation were performed as described above.

Oral Dosing Paradigm and Plasma Collection

Animals were maintained on a feeding regimen of 60 g/kg/day, delivered as 1 g pellets via a custom-built automated feeding system (7). Upon the start of vEEG recording, animals were enrolled into a 4-week baseline period (weeks 1–4), receiving 60 g/kg/day PCBO chow at four equally spaced intervals throughout the day (15 g/kg, q.i.d). Upon reaching an average seizure frequency of 1 or more seizures/week (Racine Stage 3 or higher) for 4 weeks, animals were randomly enrolled to receive PER either 100% or randomly 50% of the time for 4 weeks. The 100% adherent group (n = 10) received an average daily dose of 10 mg/kg/day (2.5 mg/kg, q.i.d), while 50% adherent group (n = 12) received medicated pellets for only 50% of their meals at random over the course of 1-week intervals (~ 5 mg/kg/day). 10 mg/kg/day was determined to be the maximally tolerated oral dose when sub chronically delivered to age- and sex-matched naïve rats (7 days) (Supplemental File 1), the same dose with demonstrated efficacy in rodent acute seizure models (17). The number of medicated pellets was adjusted on a weekly basis to maintain a 10 mg/kg/day dosing regimen. Food consumption was monitored throughout the study and leftover pellets were discarded daily.

At one-week intervals following initiation of PER treatment, blood samples were collected from the lateral tail vein approximately 1 hour following a scheduled meal from all rats (100% and 50% groups). Blood samples were processed using microtainer tubes pre-coated with K2EDTA anticoagulant and separated using centrifugation (3,000 x g for 10 min). Plasma was transferred to individual microcentrifuge tubes and stored at -80°C until LC-MS analysis of PER content as described above.

Study Outcomes and Statistical Analyses

All EEG recording data was stored electronically and contained time-stamped events of feeding history (medicated vs. PCBO), seizure events, and detailed annotation of reviewer-identified seizure behavior. Data were analyzed and figures were generated using R version 4.0.3. (R Core Team, 2020) with the package ggplot2 (18), and GraphPad Prism (Version 9.0.0, La Jolla, CA, USA).

Primary outcomes for this study included seizure burden, a metric that summarizes the frequency and severity of seizures, as well as seizure frequency. Based on our previous data with CBZ in a nonadherence paradigm (5), sample size was powered to detect a 2-fold difference in average seizure burden between the two treatment groups with 80% power. Treatment-related effects on seizure burden or frequency was determined via 2-way ANOVA. Intergroup differences between study groups were compared using normalized values, e.g., each animal was normalized to its own baseline period and the groups compared to each other, via two-tailed Mann-Whitney test. Secondary outcomes evaluated changes in pellet consumption and systemic PER concentrations in plasma over time via mixed effects model with Geisser-Greenhouse correction. Results were reported as mean ± standard error of the mean (SEM) and significance was reported at level p < 0.05. The relationship between PER delivery, e.g., individual meal vs. patterns of acute adherence, and seizure occurrence via two-tailed binomial and Chi-square tests respectively. Results were reported as calculated odds ratio ± 95% confidence interval.

Study Approval

Experiments involving animals complied with the ARRIVE guidelines and were performed in accordance with the National Institutes of Health guide for the care and use of laboratory animals (NIH publication No. 8023, revised 1978) following protocols approved by the University of Washington, Institutional Animal Care and Use Committee (protocol 4387-01, approval date: 05/05/2019).

Development of Medication-In-Food Pellets and Analytical Quantification Methods

Formulation of both medicated and PCBO pellets produced uniform, 1 g pellets that were consistent in size and shape (Fig. 1A). Daily administration of PCBO pellets to naïve Sprague Dawley male rats (60 g food/kg body weight/day) resulted in normal growth rates of animals as previously established (5, 6, 19). Importantly, tandem liquid chromatography mass spectrometry bioanalytical methods were adapted from previous work (20, 21) for quantitation of PER content in both pellets and plasma. Chromatographic separation yielded high specificity and sensitivity with sharp, well-defined peaks in calibrator (Fig. 1B) and experimental plasma samples (Fig. 1C). Lower limit of quantification (LLOQ) was determined to be 5 ng/ml, which is significantly lower than expected plasma concentrations for our dosing range of 5–10 mg/kg/day p.o. (17). Analytical quantification of PER content in formulated pellets resulted in consistent intra- and inter-pellet levels (Fig. 1D) that aligns with a total daily dose of 10 mg/kg/day, or 0.167 mg PER/g pellet when adjusted for 60 g/kg/day feeding regimen.

Experimental Schema and Animal Characteristics

The in-life portion of this study was completed over three independent cohorts with an initial enrollment of 50 animals. A total of 28 animals were lost to attrition from expected KA mortality (n = 22, 44% of total), loss of EEG head cap implant (n = 2, 4% of total), or failure to meet the enrollment criteria of ≥ 1 Racine Stage 3 or higher seizure/week to be included in the study (n = 4, 8% of total) (16). By the end of the study, 22 remaining animals (cohort 1: n = 7; cohort 2: n = 7; cohort 3: n = 8) were randomly assigned to one of two experimental adherence groups: 100% PER (n = 10) or 50% PER (n = 12).

General characteristics of animals were the same across both groups, including baseline and study duration body weights, as well as general KA-SE characteristics (Supplemental File 2). Animals in both groups met enrollment criteria (≥ 1 convulsive Stage 3 seizure/week) during a 4-week baseline recording period (starting at 8–10 weeks post-KA) and were enrolled onto PER at an average of 14.9 ± 1.9 weeks (100% PER) or 13.6 ± 1.1 weeks (50% PER) post-KA (Fig. 2A). Both groups received the same PER pellets (0.167 mg PER/g pellet) delivered on the same automated feeding schedule (15 g food/kg, q.i.d.) except that the nonadherent animals received medicated pellets at random only 50% of the time over the course of the week as depicted in Fig. 2B. No differences were found in weight gain (Supplemental File 2) or pellet consumption (100% PER: 86 ± 4% or 50% PER: 85 ± 10%) between groups. Feeder errors were relatively uncommon, occurring only 12 times over the total 4,928 meals delivered in the study (< 0.2%).

Overall degree of seizure control with PER is likely dose-dependent

The impact of PER nonadherence on seizure control was evaluated relative to fully adherent animals over the duration of an 8-week study. Baseline seizure frequencies (Table 3) and seizure burden (Fig. 3B & 3C) were similar between experimental groups, albeit there was significant inter-animal variability within each group as represented in Fig. 3A. Addition of PER did not significantly change the average daily seizure frequency (Table 3) or average daily seizure burden (Fig. 3C) compared to baseline values in the either the 100% (mean ± SD: base: 3.63 ± 4.56 vs treat: 1.07 ± 1.34) or 50% (mean ± SD: base: 3.38 ± 4.24 vs. treat: 4.72 ± 4.59) groups. However, the cumulative seizure burden was significantly higher in nonadherent animals upon initiation of PER treatment (Fig. 3B; time x treatment interaction: F (55, 770) = 1.710, p < 0.001). Given the significant inter-animal variability in seizure frequency in this model (15), seizure burden and frequency were normalized to each animal’s baseline period. When comparing between groups, fully adherent animals had a median reduction of 55% in their seizure burden compared to their untreated baseline phase [95% CI: -100%, 10%], which was significantly different from the 65% median increase [95%CI: -38%, 165%] in nonadherent animals (U = 21, p = 0.0089). Similar trends were observed in the changes in seizure frequency between treatment and baseline for both groups (Table 3). The 50% responder rate, or the number of animals that had greater than 50% reduction in seizure frequency, was 5/10 animals for fully adherent animals, while 0/10 animals in the variably nonadherent group reached this level of seizure control (Table 3). Similarly, 3/10 animals were completely seizure free in the 100% PER group, while 0/12 achieved seizure freedom in the 50% PER group.

Table 3

Summary of Epilepsy Seizure Burden and Responder Rates for 100% and 50% PER Groups
	100% Adherent (~ 10 mg/kg/day) N = 10	50% Adherent (~ 5 mg/kg/day) N = 12
Avg Seizure Frequency (# per day)
Baseline	0.871 ± 1.115	0.747 ± 0.981
Treatment	0.246 ± 0.303	1.262 ± 1.352
Median Δ [95% CI]	-50% [-100, 51]	+ 54%* [-43, 187]
> 25% Responder Rate # / n	5/10	5/12
> 50% Responder Rate # / n	5/10**	0/12
Seizure Freedom # / n	3/10	0/12
Data presented as mean ± SD
*Significantly different at p < 0.05 as determined by Mann-Whitney test
**Significantly different at p < 0.01 as determined by Fishers Exact test

The degree of seizure control between the two groups were unlikely due to differences in weekly pellet consumption as there was no significant main effect of treatment (Fig. 4A; F(1, 20) = 0.1402, p = 0.7120) during baseline or treatment phases. Moreover, a significant effect of time (F(2.3, 46.6) = 12.66, p < 0.0001) but not time x treatment interaction (F(7,140) = 0.3790, p = 0.9134) suggests that changes in pellet consumption are likely due to factors that affect both groups equally, such as reduced metabolic rate because of age (22). PER concentrations measured in plasma at 1-week sampling intervals following the start of treatment reveal that plasma concentrations were dependent on delivered daily dose since PER concentrations in the 100% group were about twice as high as the 50% group (Fig. 4B; F (1, 20) = 6.126, p = 0.02). Importantly, there was no significant time effect (F (1.9, 23.9) = 0.2834, p = 0.7457), suggesting that rats in both groups had consistent plasma levels over the duration of the study, which were not significantly correlated with weekly pellet consumption (Fig. 4C). Moreover, the degree of seizure control, or change in seizure frequency compared to baseline, was negatively correlated with PER plasma levels in fully adherent animals, highlighting that seizure control was associated with greater PER levels in plasma (Fig. 4D, r = -0.6933, p = 0.03). While nonadherent animals trended toward the same relationship, this did not reach statistical significance (r = -0.4965, p = 0.1).

Acute patterns of nonadherence are not associated with increased risk of breakthrough seizures

To quantify the relationship between a missed dose and subsequent seizure, we analyzed the distribution of meals occurring before every seizure in nonadherent animals (Table 4). Of note, one or more seizures occurring between mealtimes, e.g., seizure cluster, was collapsed into a single event for analysis. Being that the addition of PER did not change seizure cluster phenotype or frequency in either fully adherent or nonadherent animals, nor was there a significant relationship between any meal pattern and the risk of a seizure cluster, clusters were collapsed into single events as to not unequally weight certain meal patterns over others (Supplemental File 3). The medication delivery schedule was analyzed for a total of 206 seizure events across 12 animals as previously described (6). For instance, the distribution of PCBO vs. PER delivery was tabulated for every meal preceding a seizure for each animal (e.g., one meal before, two meals before, three meals before etc.) and the distribution was represented as the percentage of the total of delivered meals (Table 4). For example, if there was no relationship between a single missed dose and the incidence of a seizure, the expected distribution would be 50:50 (PCBO vs. PER) for every meal preceding a seizure. A two-tailed binomial test did not identify any individual meal that deviated from the expected distribution, although the third meal before a seizure trended towards being unmedicated (i.e., 44% of meals were delivered as PCBO for the third meal, p = 0.0828). Together, these data suggest there is no relationship between a single missed dose of PER and incidence of a seizure.

Table 4

Expected vs. Observed Distribution of PER Meals (%) Before a Seizure
Meal Before a Seizure	% Medicated			Statistical Significance
Meal Before a Seizure	Expected	Observed	95% CI	P value
1	50	52	45–58	0.6259
2	50	50	43–57	> 0.999
3	50	44	37–51	0.0828
4	50	54	47–61	0.2960
5	50	54	47–61	0.2960
6	50	47	41–54	0.4435
7	50	46	39–52	0.1854
8	50	51	44–57	0.8895
9	50	52	45–58	0.7277
10	50	52	45–58	0.7277
Data presented as % of meals that were medicated with PER before a seizure. The expected distribution is 50:50. A total of 206 seizure events were analyzed. No statistically significant differences as determined by two-tailed binomial test.

From a clinical perspective, medication nonadherence can take various forms across different individuals. For instance, some may forget a single dose at sporadic times, while others may have short, sustained periods of nonadherence. Therefore, in addition to quantifying the relationship between a single missed dose of PER and seizure occurrence, we also evaluated whether acute periods of nonadherence within a 24-h window were associated with seizure incidence. Based on q.i.d. dosing, there are 16 possible patterns of medication delivery that range from completely nonadherent to fully adherent (Fig. 5A). We did not identify any meal patterns associated with increased incidence of a seizure, including 24 h periods where PER was missed for every dose (OR [95% CI]: 1.335 [0.6448, 2.716], z = 0.754, p = 0.4482). However, we found that the odds of seizure occurrence were lower following periods of full adherence within 24 h (OR [95% CI]: 0.2194 [0.0658, 0.7658], z = 2.55, p = 0.011). Moreover, when grouping meal patterns into acute periods of adherence (e.g., 0, 25, 50, 100%), we similarly found no association between acute nonadherence and seizure incidence (Fig. 5B). Likewise, when evaluating different scenarios of nonadherence within a 24 h window, e.g., 2 consecutive doses are missed, or the final 3 doses before a seizure were missed (Fig. 5C), there was no situation that resulted in an increased risk of seizure incidence. Taken together, these data highlight an unexpected relationship between medication delivery and breakthrough seizures suggesting that a missed dose, or series of missed doses of PER within a 24-h period may not significantly increase the risk of unexpected seizures.

Herein, we describe a study where we have successfully formulated drug-in-chow pellets that are uniform in size, shape, and ASM content, which are compatible with our automated feeder delivery system utilized for long-term drug studies in preclinical research (5–7). While useful across all disease types, we utilized our novel formulation methods to expand on previous work with CBZ (5, 6) to determine how nonadherence to a mechanistically distinct first-line therapy for focal seizures, perampanel (PER), impacts seizure control in an etiologically relevant animal model of acquired epilepsy.

Medication-in-food delivery approaches are not novel for chronic drug studies in preclinical models of epilepsy (19, 23, 24). However, few studies outside of our own have taken advantage of an automated medication-in-food delivery system to deliver ASM therapy on a schedule that supports pharmacokinetic-based dosing (5, 6). Limitations of such systems require pellets to be formulated in a way that is compatible in both size and shape with the system and investigators may be limited to the pre-determined selections that are offered from commercial vendors. Moreover, commercial vendors may not have the regulatory license in place to manufacture food pellets with controlled substances such as PER, which is listed as a Schedule III agent in the U.S. Drug Enforcement Agency (DEA) Controlled Substances Act (25). We overcame these hurdles by using commercially available tooling resources that can be customized to create pellets of any shape and size. Moreover, our formulation methods generate pellets with uniform medication content that can be replicated with any compound of interest. Together, these methods may reduce the barriers to chronic drug studies and provide an essential tool for improving preclinical drug discovery across all models of chronic diseases.

These results support the hypothesis that poor ASM adherence results in worsened seizure control. One interpretation may be that seizure control is dose-dependent and that the 50% group may not have reached a large enough dose of PER to render any antiseizure effects. It is worth noting that PER levels in both the 100% and 50% animals remained stable and were well within the range of effective concentrations in adult patients that respond to PER therapy (425 ± 270 ng/ml, (26)). Moreover, our data also highlight that any 24 h period where PER was administered for all four doses was associated with a lower risk of breakthrough seizures, suggesting that our findings in nonadherent animals cannot fully be explained by dose alone. These data do not rule out the possibility that increasing the dose of PER beyond 10 mg/kg/day may have resulted in better seizure control in the refractory animals. However, the goal of ASM therapy is to attain seizure freedom without untoward side effects (27), and early pilot data in age- and sex-matched naïve rats suggest that motor impairment and sedation occur at oral doses as low as 20 mg/kg/day. One might imagine the clinical implications of pushing the dose into an intolerable range; e.g. increased adverse events could perpetuate nonadherent behavior, rendering any dose-dependent improvement in efficacy negligible. It is not surprising that few animals attained seizure freedom given that the post-KA model of chronic epilepsy more closely resembles human TLE (28), which is the most refractory type of epilepsy (~ 30% of patients achieve seizure freedom with adequate therapy (29)). However, this may also emphasize a strength of this model for further investigation into whether medication nonadherence may contribute to future pharmacoresistance with different ASM therapy, a limitation that was not addressed by the present investigation.

One of the most surprising findings from this study is the apparent lack of pharmacokinetic dependence between a missed dose of PER and breakthrough seizures. Given the short half-life of PER in rodents (~ 1.67 hours), one would expect that a missed dose (i.e., 6 hours, ~ 4 half-lives) would result in PER concentrations that fall below therapeutic levels, leading to breakthrough seizures (17), which was not the case in our animals. Interestingly, PER concentrations remained consistent in the 50% group at levels that have been reported to be effective in patients. One limitation is that our therapeutic drug monitoring was completed at 1-week intervals, and while PER concentrations did not fluctuate over time, we do not have brain or plasma measurements within any acute 24-hour period to confirm how PER concentrations change with different patterns of adherence and whether this correlates with breakthrough seizures. This is an ongoing line of investigation. Nonetheless, these findings raise an interesting possibility that missing a dose of PER may not negatively impact seizure threshold despite changes in drug concentrations within the local microenvironment.

On the surface, variable adherence to PER differs from published work with CBZ (5, 6). While a missed dose of PER does not increase the risk of breakthrough seizures, a missed dose of CBZ may lead to unexpected seizures much later in time (e.g., days later), even if subsequent doses have been taken. Granted, the CBZ studies initiated nonadherence at the time of “epilepsy diagnosis” in animals, which is relevant for newly diagnosed patients who may be more likely to adopt nonadherent behavior (30), but makes it difficult to interpret the impact of nonadherence on disease progression. For instance, it is unclear whether CBZ nonadherence resulted in an improvement or worsening of seizure clusters, and whether those seizure clusters were accounted for in the analyses of the relationship between missed dose and seizure events. Conversely, we found that PER nonadherence did not alter disease phenotype, allowing us to limit our analyses to the relationship between missed dose(s) and seizure events (e.g., 1 + seizure occurring between meal periods). Despite the differences in study design and analysis, both CBZ and PER highlight an interesting non-pharmacokinetic relationship between ASM nonadherence and breakthrough seizures that should be considered in clinical practice.

If substantiated in patients, the clinical implication of our work supports consideration of a patient’s adherence profile when undergoing ASM selection, rather than only selecting an ASM that meets the appropriate tolerability and efficacy profile for a given patient’s seizure type (31). A long-half life and favorable pharmacokinetic profile reduce barriers for attaining optimal adherence (13), however, certain ASM’s that are less likely to result in breakthrough seizures because of missed dose(s) may be an important consideration for patients at higher risk of nonadherence. If our preclinical data is substantiated in the clinic, PER’s favorable pharmacokinetic profile in humans (t_1/2 ~105 h) (32), combined with a lowered risk of breakthrough seizures suggests that it may provide a certain forgiveness factor if a dose is missed within a 24-hour window. It is worth noting that, to date, preclinical studies have only investigated the relationship between nonadherence and seizure control using male subjects. While clinical data suggest that men may be more prone to adopting nonadherent behavior (3), further investigation is warranted into whether sex as a biological variable changes the relationship between adherence and breakthrough seizures.

Patterns of medication nonadherence may not always be random and may take different forms that are influenced by affordability, accessibility, or other patient-specific factors (i.e., perceived lack of benefit, intolerable adverse events etc.) (3, 30, 33). For example, brief drug holidays, or prolonged periods where patients abstain from ASM therapy represent other patterns of nonadherence; however, there is very little clinical or preclinical data to suggest whether these periods may increase the risk of conferring treatment resistance later on when therapy is restored (34, 35). Preclinical work with CBZ suggests that a brief drug holiday does not necessarily predispose an animal to developing CBZ resistance when adherence is restored (6), but whether this generalizes to ASM’s across different mechanistic classes is unknown and warrants further investigation. Moreover, clinical practice may dictate a switch in monotherapy or escalation to polytherapy after a loss of seizure control, however, further investigation is needed to determine whether poor adherence to initial ASM therapy will extend to poor outcomes in subsequent ASMs that are tried.

The unique experimental approach presented herein highlights the important relationship between medication adherence and seizure control, which may provide clinically meaningful insights into ASM management in patients with epilepsy. These results suggest that a missed dose of the mechanistically novel ASM, PER, does not immediately increase the risk of breakthrough seizures in rats with acquired epilepsy. If upheld in the clinic, our findings highlight a non-pharmacokinetic relationship between PER nonadherence and breakthrough seizures that have important implications for clinical practice.

ASM, antiseizure medicine

CBZ, carbamazepine

CNS, central nervous system

KA, Kainic Acid

PCBO, Placebo

PER, Perampanel

SD, standard deviation

SE, status epilepticus

SEM, standard error of the mean

TLE, temporal lobe epilepsy

vEEG, video electroencephalography

Ethics approval and consent to participate.

Experiments involving animals complied with the ARRIVE guidelines and were performed in accordance with the National Institutes of Health guide for the care and use of laboratory animals (NIH publication No. 8023, revised 1978) following protocols approved by the University of Washington Institutional Animal Care and Use Committee (protocol 4387-01, approved 05/05/2019).

Consent for publication

Not applicable.

Availability of data and materials

The data presented in this study are available on request from the corresponding author and presented as individual datapoints in all accompanying tables and figures in this manuscript.

Competing Interests

The authors declare no competing interests.

Funding

This research was supported by an unrestricted grant from Eisai, Inc. M.G. was supported by a postdoctoral fellowship from the American Epilepsy Society.

Author Contributions

MG contributed to the conception and design of the study, contributed to data acquisition and statistical analyses of the results, generated figures, drafted and revised the manuscript; AC contributed to data acquisition and statistical analyses of the results, drafted the manuscript; JV contributed to data acquisition and revision of the manuscript; DW designed analytical method for PER-detection in food and biological samples, drafted text for methods and results; HSW was involved in the conception and design of the study, oversaw the completion of studies, obtained funding for the project, and substantively revised the manuscript.

Acknowledgements

The authors are grateful for the technical assistance of Nicholas Uribe.

Authors’ Information

Not applicable

Kwan P, Brodie MJ. Early identification of refractory epilepsy. N Engl J Med. 2000;342(5):314–9.
Terman SW, Kerr WT, Aubert CE, Hill CE, Marcum ZA, Burke JF. Adherence to Antiseizure vs Other Medications Among US Medicare Beneficiaries With and Without Epilepsy. Neurology. 2022;98(4):e427–e36.
Malek N, Heath CA, Greene J. A review of medication adherence in people with epilepsy. Acta Neurol Scand. 2017;135(5):507–15.
Tan M, Allemann SS, Qin XS, D'Souza WJ. Adherence patterns in antiseizure medications influencing risk of sudden unexplained death in epilepsy: A data linkage study using dispensed prescriptions. Epilepsia. 2023;64(3):641–53.
Hill AC, Thomson KE, Newell TG, White HS. Correction of medication nonadherence results in better seizure outcomes than dose escalation in a novel preclinical epilepsy model of adherence. Epilepsia. 2019;60(3):475–84.
Thomson KE, Modi AC, Glauser TA, Rausch JR, Steve White H. The impact of nonadherence to antiseizure drugs on seizure outcomes in an animal model of epilepsy. Epilepsia. 2017;58(6):1054–62.
Thomson KE, White HS. A novel open-source drug-delivery system that allows for first-of-kind simulation of nonadherence to pharmacological interventions in animal disease models. J Neurosci Methods. 2014;238:105–11.
Löscher W, Klein P. The Pharmacology and Clinical Efficacy of Antiseizure Medications: From Bromide Salts to Cenobamate and Beyond. CNS Drugs; 2021.
Greenwood J, Valdes J. Perampanel (Fycompa): A Review of Clinical Efficacy and Safety in Epilepsy. P T. 2016;41(11):683–98.
French JA, Krauss GL, Steinhoff BJ, Squillacote D, Yang H, Kumar D, et al. Evaluation of adjunctive perampanel in patients with refractory partial-onset seizures: results of randomized global phase III study 305. Epilepsia. 2013;54(1):117–25.
Wheless J, Chourasia N. Safety evaluation of perampanel as monotherapy or first adjunctive therapy in patients with epilepsy. Exp Opin Drug Saf. 2022;21(10):1239–47.
Franco V, Crema F, Iudice A, Zaccara G, Grillo E. Novel treatment options for epilepsy: focus on perampanel. Pharmacol Res. 2013;70(1):35–40.
Gidal BE, Ferry J, Reyderman L, Pina-Garza JE. Use of extended-release and immediate-release anti-seizure medications with a long half-life to improve adherence in epilepsy: A guide for clinicians. Epilepsy Behav. 2021;120:107993.
Gidal BE, Majid O, Ferry J, Hussein Z, Yang H, Zhu J, et al. The practical impact of altered dosing on perampanel plasma concentrations: pharmacokinetic modeling from clinical studies. Epilepsy Behav. 2014;35:6–12.
Hellier JL, Patrylo PR, Buckmaster PS, Dudek FE. Recurrent spontaneous motor seizures after repeated low-dose systemic treatment with kainate: assessment of a rat model of temporal lobe epilepsy. Epilepsy Res. 1998;31(1):73–84.
Racine RJ. Modification of seizure activity by electrical stimulation. II. Motor seizure. Electroencephalogr Clin Neurophysiol. 1972;32(3):281–94.
Hanada T, Hashizume Y, Tokuhara N, Takenaka O, Kohmura N, Ogasawara A, et al. Perampanel: a novel, orally active, noncompetitive AMPA-receptor antagonist that reduces seizure activity in rodent models of epilepsy. Epilepsia. 2011;52(7):1331–40.
Wickham H. ggplot2: Elegant Graphics for Data Analysis. Cham: Springer International Publishing : Imprint: Springer; 2016.
Grabenstatter HL, Clark S, Dudek FE. Anticonvulsant effects of carbamazepine on spontaneous seizures in rats with kainate-induced epilepsy: comparison of intraperitoneal injections with drug-in-food protocols. Epilepsia. 2007;48(12):2287–95.
Mano Y, Takenaka O, Kusano K. High-performance liquid chromatography-tandem mass spectrometry method for the determination of perampanel, a novel alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor antagonist in human plasma. J Pharm Biomed Anal. 2015;107:56–62.
de Grazia U, D'Urso A, Ranzato F, De Riva V, Contarato G, Billo G, et al. A Liquid Chromatography-Mass Spectrometry Assay for Determination of Perampanel and Concomitant Antiepileptic Drugs in the Plasma of Patients With Epilepsy Compared With a Fluorescent HPLC Assay. Ther Drug Monit. 2018;40(4):477–85.
McCarter RJ, Palmer J. Energy metabolism and aging: a lifelong study of Fischer 344 rats. Am J Physiol. 1992;263(3 Pt 1):E448–52.
Ali A, Dua Y, Constance JE, Franklin MR, Dudek FE. A once-per-day, drug-in-food protocol for prolonged administration of antiepileptic drugs in animal models. Epilepsia. 2012;53(1):199–206.
Nobili P, Cattalini A, de Grazia U, Cagnoli C, de Curtis M, Battaglia GS, et al. Early Chronic Carbamazepine-in-Food Administration to MAM/Pilocarpine Rats Does Not Affect Convulsive Motor Seizures. Front Pharmacol. 2020;11:181.
Drug Enforcement Administration DoJ. Schedules of controlled substances: placement of perampanel into Schedule III. Final rule. Fed Regist. 2013;78(231):72013–6.
Steinhoff BJ, Hubers E, Kurth C, Jurges Kehl-Kork U. Plasma concentration and clinical effects of perampanel-The Kork experience. Seizure. 2019;67:18–22.
Talati RSJ, Phung OJ. Effectiveness and Safety of Antiepileptic Medications in Patients with Epilepsy. 2011. In: Comparative Effectiveness Reviews, No 40 [Internet]. Rockville, MD: Agency for Healthcare Research and Quality. Available from: https://www.ncbi.nlm.nih.gov/books/NBK83926/.
Lévesque M, Avoli M, Bernard C. Animal models of temporal lobe epilepsy following systemic chemoconvulsant administration. J Neurosci Methods. 2016;260:45–52.
Kim WJ, Park SC, Lee SJ, Lee JH, Kim JY, Lee BI, et al. The prognosis for control of seizures with medications in patients with MRI evidence for mesial temporal sclerosis. Epilepsia. 1999;40(3):290–3.
Modi AC, Rausch JR, Glauser TA. Patterns of nonadherence to antiepileptic drug therapy in children with newly diagnosed epilepsy. JAMA. 2011;305(16):1669–76.
Kim H, Kim DW, Lee ST, Byun JI, Seo JG, No YJ, et al. Antiepileptic Drug Selection According to Seizure Type in Adult Patients with Epilepsy. J Clin Neurol. 2020;16(4):547–55.
Patsalos PN. The clinical pharmacology profile of the new antiepileptic drug perampanel: A novel noncompetitive AMPA receptor antagonist. Epilepsia. 2015;56(1):12–27.
von Gaudecker JR, Buelow JM, Miller WR, Tanner AL, Austin JK. Social determinants of health associated with epilepsy treatment adherence in the United States: A scoping review. Epilepsy Behav. 2021;124:108328.
Azar NJ, Lagrange AH, Wang L, Song Y, Abou-Khalil BW. Transient improvement after brief antiepileptic drug withdrawal in the epilepsy monitoring unit–possible relationship to AED tolerance. Epilepsia. 2010;51(5):811–7.
Howland RH. Medication holidays. J Psychosoc Nurs Ment Health Serv. 2009;47(9):15–8.

Perampanel’s forgiveness factor in a variable medication adherence paradigm in a rat model of chronic epilepsy

Status:

Journal Publication

Version 1

Abstract

Figures

Introduction

Methods

Animals

Kainic Acid (KA) Induced-Status Epilepticus (SE)

Cortical EEG Electrode Implantation

Video EEG (vEEG) Recording and Seizure Detection

PCBO and Perampanel Pellet Formulation

Quantitative analysis of PER in pellets and rat plasma

Oral Dosing Paradigm and Plasma Collection

Study Outcomes and Statistical Analyses

Study Approval

Results

Development of Medication-In-Food Pellets and Analytical Quantification Methods

Experimental Schema and Animal Characteristics

Overall degree of seizure control with PER is likely dose-dependent

Acute patterns of nonadherence are not associated with increased risk of breakthrough seizures

Discussion

Conclusions

Abbreviations

Declarations

References

Supplementary Files

Status:

Journal Publication

Version 1