A Pilot Double‐Blind Randomized Trial of Perampanel for Essential Tremor

ABSTRACT

Background

Perampanel is a noncompetitive antagonist of alpha‐amino‐3‐hydroxy‐5‐methylisoxazole propionic acid glutamate receptors suggested to modulate tremor.

Objectives

To assess the efficacy and tolerability of perampanel for essential tremor.

Methods

This was a double‐blind, placebo‐controlled, randomized, cross‐over trial involving 26 patients titrated to 8 mg/day or a lower maximally tolerated dose as monotherapy or adjunct to antitremor medication. Tremor was assessed at the beginning and end of each 14‐week treatment arm. The primary endpoint was change in the videotaped performance subscale of The Essential Tremor Rating Assessment Scale, scored by a blinded rater. Secondary endpoints included change in The Essential Tremor Rating Assessment Scale Activity of Daily Living and Quality of Life in Essential Tremor and Subject Global Impression of Change subscales.

Results

Data are available for 15 and 11 participants who completed placebo and perampanel arms, respectively. Perampanel was superior to placebo on the primary endpoint (P = 0.028), Activity of Daily Living (P = 0.009), and Subject Global Impression of Change (P = 0.016), but not Quality of Life (p = 0.48). Video scores were rated >50% improved in 3/11 on perampanel and 0/15 on placebo. Adverse events were more likely on perampanel (especially at >4 mg/day) than on placebo, leading to withdrawal (36% vs. 10%) and dose reduction (41% vs. 15%). Adverse events more common with perampanel included imbalance/falls (50% vs. 10%), dizziness (36% vs. 10%), and irritability (27% vs. 5%).

Conclusions

These findings suggest that perampanel exerts efficacy for some persons with essential tremor, but this population appears prone to adverse events.

Propranolol, primidone, and topiramate (>200 mg/day) are the only pharmacological interventions for which evidence‐based medicine strongly supports efficacy for essential tremor (ET).1 However, many patients abandon treatment because of side effects or poor efficacy.2

Indeed, half or more of ET patients reportedly stop taking each of the most commonly used antitremor medications, and one third, even the most disabled, give up taking any medication for tremor.3 More treatment options for ET are clearly needed.

Alpha‐amino‐3‐hydroxy‐5‐methylisoxazole propionic acid (AMPA) receptors are located in several regions within the oscillatory network revealed by magnetoencephalography4 in ET, including the frontal cerebral and cerebellar cortex.5 Moreover, AMPA receptors increase the degree of inferior olivary coupling, which in turn increases Purkinje cell complex spike synchrony6, 7 that may be associated with increased movement amplitude and tremor.6, 8 Systemically administered AMPA receptor antagonists suppress tremor in the harmaline animal model,9, 10 whereas PF‐4778574, an allosteric AMPA receptor activator, causes tremor in primates and other species.11 These considerations suggest that an AMPA receptor antagonist might exert efficacy for ET.

Perampanel is a noncompetitive AMPA‐type glutamate receptor antagonist approved in the United States as an adjunct treatment of partial‐onset seizures in patients with epilepsy. Perampanel has been studied for Parkinson's disease12 and restless legs syndrome.13 A case report has described efficacy for orthostatic tremor.14 Here we report the results of an exploratory efficacy and safety study of perampanel for ET.

Methods

This was a 2‐site, double‐blind, randomized, placebo‐controlled, cross‐over clinical trial of perampanel for ET (NCT02668146). The primary objective was to compare the efficacy of perampanel (8 mg or a lesser maximally tolerated dose once daily) with that of placebo on tremor.

Men and nonpregnant women ages ≥18 years were recruited. Women could not have child‐bearing potential or were using adequate contraception. Participants had ET as defined by the Tremor Investigation Group criteria15 and had at least 1 upper extremity with a tremor score ≥ 1.5 in forward posture, wing‐beating posture, or finger‐to‐nose movement using the Performance subscale of The Essential Tremor Rating Assessment Scale (TETRAS‐P). All participants had no significant hepatic or renal disease, were in general good health without a significant psychiatric history or substance abuse, had no history of suicide attempt or aggression in the past 2 years, were taking stable doses of up to 3 antitremor medications in the 4 weeks prior to screening, had no prior surgery for tremor, had no botulinum injections for at least 6 months, did not have significant imbalance or fall risk, were not taking inducers or inhibitors of cytochrome 3A4, and had not previously taken perampanel.

The protocol was approved by Institutional Review Board B of Veterans Affairs Greater Los Angeles and by the Institutional Review Board of the Mount Sinai School of Medicine, New York, in accordance with all applicable national and international guidelines/laws, and written informed consent was obtained from all participants.

At the screening visit, laboratory tests and examinations were performed to verify general good health, and tremor measures, including videotaping, were performed to accustom the participants to test procedures. In the first‐arm baseline visit, up to 1 month later, examinations and tremor measures were repeated.

The research pharmacist at each site used computer‐generated random numbers prepared in advance to assign participants sequentially to receive either perampanel or matching placebo 2‐mg tablets in a 1:1 ratio for balanced groups of 4. Apart from dispensing the blinded study medication, provided by Eisai Inc., (Woodcliff Lake, New Jersey, USA) the pharmacist was not involved with study conduct. None of the investigator team members or participants were aware of treatment assignment. On completion of study conduct, the pharmacist revealed which participants had received treatments “A” or “B” in the first treatment arm, then received the opposite treatment in the second arm. Not until the external statistician had finished analyses comparing treatments A and B and the descriptive statistics were completed was the identity of these treatments revealed to the investigators.

The participants were instructed to take study medication once daily before bedtime, starting at 1 tablet, and to increase the dose by 1 tablet each 21 days to no more than 4 tablets (8 mg) a day. The choice of 8 mg/day as a target dose was based on epilepsy trials that indicated that this dose was generally tolerated.16

The participants were followed in telephone visits at 2‐week intervals. During titration, the treating physician (A.H., W.T.) could lower the dose if treatment‐emergent adverse events (TEAEs) occurred. The investigator could also refrain from raising the dose further if the participant reported good tremor control. Inclusion in efficacy analysis required a constant dose to be taken during the 21 days prior to the end‐of‐arm treatment visit at 14 weeks. Following examinations and tremor measures at the treatment visit, the study drug was stopped for a 4‐week washout, and then the participants returned for the baseline visit of the second treatment arm, during which the procedures were as in the first arm but with the opposite blinded study drug. At the final visit, 4 weeks after the second end‐of‐arm treatment visit, safety evaluations included examinations, laboratory tests, and review of adverse events (AEs).

The TETRAS‐P exam was videotaped at each treatment arm's baseline and treatment visits by a study team member who was separate from the treating physician and did not have knowledge of participants’ symptoms. The 16‐item TETRAS‐P Scale evaluates postural and kinetic tremor of body parts affected by ET, with emphasis on upper extremity tremor. Items are scored from 0 to 4, with 4 representing the highest severity. The maximum score is 64.17 A blinded external rater (R.J.E.) received coded videotapes without knowledge of visit or treatment. The participants also completed a 12‐item (0–4 ratings) Activity of Daily Living questionnaire (TETRAS‐ADL, maximum score = 48) and a 30‐item Quality of Life in Essential Tremor (QUEST) questionnaire, also with 0 to 4 ratings.18, 19 On treatment visits only, the participants rated overall change with the Subject Global Impression of Change Scale (SGIC) in which change was rated between minus 3 and plus 3, with 1 connoting mild change, 2 moderate change, and 3 marked change, with minus indicating improvement and plus worsening. Safety testing included laboratory tests, examinations, and collection of TEAEs at clinic and telephone visits.

The primary outcome endpoint was tremor change from baseline to treatment in the perampanel treatment arm compared with the placebo treatment, according to the TETRAS‐P blinded video ratings. Changes in questionnaire scores constituted secondary endpoints. The intent‐to‐treat population included randomized participants who had efficacy data from at least 1 treatment arm. The completer population included participants from the intent‐to‐treat population who completed both treatment arms with efficacy data. The safety population included all participants who received at least 1 dose of study drug.

Statistical Methods

Estimates were computed using the completer population (n = 10 of 26) and again using data from the intent‐to‐treat population (n = 16). In the latter analysis, missing values were imputed using the method of multivariate imputation by chained equations, as described by van Buuren and Groothuis‐Oudshoorn.20 As none of the endpoints had a normal (Gaussian) distribution, standard errors and P values were computed using the nonparametric resampling (bootstrap) method with 10,000 resamplings. Computations were carried out using the R software system (R Foundation for Statistical Computing, Vienna, Austria). Means and standard errors are reported for scores. A P value less than 0.05 was considered significant. TEAEs are summarized using descriptive statistics.

Results

Between May 2016 and September 2017, 28 patients were screened, of whom 26 were randomized. Demographic and descriptive data for the randomized, intent‐to‐treat, and completer populations are shown in Table 1. Participant flow is summarized in Figure 1. The most common reason for early withdrawal was TEAEs, occurring in 8/22 participants on perampanel but in only 2/20 on placebo. One patient left due to a serious AE of arm fracture during a fall while on perampanel, and another on placebo had a serious AE of facial tumor. Of the 2 sites, Mount Sinai had more dropouts (P = 0.046, Fisher exact test) due to serious AEs and lost to follow‐ups and had too few participants to permit efficacy comparisons of the sites.

Table 1.

Participant demographic data

	Randomized	Intent to Treat	Completer
Attribute	n = 26	n = 16	n = 10
*Age, y	70.4 (14.1)	73.2 (9.9)	72.3 (6.8)
Tremor duration, y	29.7 (18.4)	30.8 (18.2)	30.3 (16.3)
No. prior tremor drugs
Tried and stopped	1.6 (1.7)	1.4 (1.1)	1.3 (1.2)
Male	15 (58%)	10 (63%)	6 (60%)
Race
White	20 (77%)	15 (94%)	10 (100%)
African American	3 (12%)	0	0
Hispanic	3 (12%)	1 (6%)	0
Alcohol effect on tremor
Reduces	15 (58%)	10 (62%)	5 (50%)
No effect	3 (12%)	1 (6%)	1 (10%)
Unknown	8 (31%)	5 (31%)	4 (40%)
Concurrent antitremor meds
None	10 (38%)	7 (44%)	5 (50%)
1	13 (50%)	6 (38%)	3 (30%)
2	2 (8%)	2 (12%)	2 (20%)
3	1 (4%)	1 (6%)	0

^* Means (standard deviations) and n (%) are shown.

Figure 1.

Subject flow. SAE, serious adverse event; TEAE, treatment‐emergent adverse event.

Efficacy Outcomes

Data for the intent‐to‐treat group are available from 15 who completed the placebo arm and from 11 who completed the perampanel arm. At the treatment visit, the participants were taking 3.73 (0.80) placebo tablets (mean [standard deviation]) daily or perampanel 3.09 (0.92) 2‐mg tablets daily. The change in video‐rated tremor in the TETRAS‐P Scale met the primary endpoint (P = 0.028; Table 2), with a mean − 3.13‐point score reduction. This point reduction is estimated to correspond to an average 20% reduction in tremor amplitude according to the Weber‐Fechner relationship between tremor ratings and tremor amplitude:21 Percent change = [10^(αR/N) − 1]100%, where α = 0.5, R = −3.1, and N = 16, the number of items in TETRAS‐P. In the perampanel arm, 3/11 had TETRAS‐P score reductions >50% compared with 0/15 in the placebo arm.

Table 2.

Efficacy endpoint outcomes

	Placebo		Perampanel
Measure	Baseline Scores	Change in Score	Baseline Scores	Change in Score	P
Intent to treat	n = 15		n = 11
TETRAS‐P	22.9 (2.1)	0.72 (0.64)	21.9 (1.9)	−3.13 (1.45)	0.028
TETRAS‐ADL	27.7 (1.8)	0.13 (1.56)	27.9 (1.9)	−6.67 (2.36)	0.009
QUEST	42.7 (4.1)	−3.71 (2.89)	43.7 (4.3)	−6.53 (3.57)	0.475
SGIC		−0.34 (0.24)		−1.46 (0.38)	0.016
Completers	n = 10		n = 10
TETRAS‐P	24.1 (2.4)	0.17 (0.37)	25.0 (2.2)	−3.27 (1.79)	0.093
TETRAS‐ADL	30.5 (2.3)	−1.17 (1.53)	29.9 (2.3)	−7.38 (2.54)	0.029
QUEST	48.8 (7.1)	−2.83 (4.14)	48.6 (5.4)	−7.42 (4.28)	0.460
SGIC		−0.17 (0.19)		−1.67 (0.47)	0.003

Means (standard errors) are shown for scores at baseline and change at the end‐of‐arm visit. Negative values indicate improvement. P values are for comparisons of mean changes, as computed by the nonparametric resampling (bootstrap) method.

TETRAS‐P, Performance subscale of The Essential Tremor Rating Assessment Scale; TETRAS‐ADL, Activity of Daily Living subscale of TETRAS; QUEST, Quality of life in Essential Tremor Scale; SGIC, Subject Global Impression of Change.

The participants rated their TETRAS‐ADL score as barely changed on placebo (0.13 points increased on average) but as improved in the perampanel arm (6.67 points lower on average). This endpoint was met (P = 0.009), with 3/11 on perampanel rating their score as >50% lower versus 0/15 on placebo. Perampanel treatment also met the SGIC endpoint (P = 0.016), with 4/11 on perampanel rating their impression as marked improvement versus 0/15 on placebo. In contrast, perampanel did not affect QUEST scores compared with placebo (P = 0.48; Table 2). Of the 6/11 participants who rated their global impression of change as markedly improved and/or had >50% score reductions on the TETRAS‐ADL and/or TETRAS‐P Scales, 1, 1, 1, and 3 were taking perampanel 4, 5, 6, and 8 mg/day, respectively, at the end‐of‐arm treatment visit.

For the subset of 10 participants who completed both treatment arms of this cross‐over study, the primary endpoint of video‐rated change in TETRAS‐P in the perampanel arm compared with the placebo arm did not achieve significance (P = 0.093; Table 2), nor did the change in QUEST scores (P = 0.46). The perampanel‐associated change in TETRAS‐ADL scores was greater than that for the placebo arm (P = 0.029), as it was for the SGIC (P = 0.003).

A significant period effect occurred for QUEST scores in the intent‐to‐treat population (P = 0.034) due to a greater reduction of scores during treatment arm 1 than for arm 2, regardless of treatment. This was not significant in the completer population, nor was any other period effect significant for the other 3 outcome measures in either population. There was no significant treatment by period interaction for any outcome in either population.

Adverse Symptoms

Compared with the placebo treatment arm, those in the perampanel arm were more likely to report TEAEs (91% vs. 35%) and to experience symptoms leading to dose reduction (41% vs. 15%) or early withdrawal (36% vs. 10%). Of those on perampanel who withdrew, 1 did not tolerate 4 mg/day, whereas the other 7 did not tolerate a dose higher than 4 mg/day (6.2 mg/day on average) and chose to withdraw at that dose or despite adjustment to a lower dose. The most common TEAEs while taking perampanel were imbalance, dizziness, irritability, fatigue, impaired cognition, and somnolence (Table 3). TEAEs occurring more often at 7 to 8 mg/day than at 3 to 4 mg/day perampanel included imbalance (6/10 [60%] vs. 7/22 [32%]), irritability (3/10 [30%] vs. 3/22 [14%]), fatigue (3/10 [30%] vs. 4/22 [18%]), and slurred speech (2/10 [20%] vs. 0/22 [0%]).

Table 3.

Treatment‐emergent adverse events (TEAEs)

	Placebo	Perampanel
	n = 20	n = 22
TEAEs, n (%)	7 (35%)	20 (91%)
SAEs, n (%)	1 (5%)	1 (5%)
TEAEs leading to study withdrawal, n (%)	2 (10%)	8 (36%)
TEAEs leading to dose reduction, n (%)	3 (15%)	9 (41%)
Mean maximum daily intakea	3.5 tablets	3.3 tablets
Specific TEAEs, n (%)
Imbalance, including falls	2 (10%)	11 (50%)
Dizziness	2 (10%)	8 (36%)
Fatigue	4 (20%)	6 (27%)
Irritability	1 (5%)	6 (27%)
Impaired cognition	1 (5%)	4 (18%)
Somnolence	2 (10%)	4 (18%)
Worse tremor	1 (5%)	3 (14%)
Confusion	0 (0%)	2 (9%)
Slurred speech	0 (0%)	2 (9%)

Only TEAEs occurring in >5% are shown.

^aTablets with active drug contained 2 mg perampanel.

SAE, serious adverse events.

Discussion

Although the sample size was small, this placebo‐controlled, randomized, cross‐over trial of the AMPA antagonist perampanel met the primary endpoint of blinded video ratings of tremor on the TETRAS‐P. In addition, the secondary endpoints of the TETRAS‐ADL and Subject Global Impression of Change were met, but not the responses to the QUEST Scale, which considers quality of life as affected by tremor. The lack of change in the QUEST scores may be due to a lack of sensitivity, as this scale has questions about items not likely to change rapidly during a trial, such as jobs, finances, and hobbies. Improved quality of life has not been demonstrated for any oral medication for ET. In studies of deep brain stimulation, quality‐of‐life measures improved only modestly despite marked tremor reduction.22, 23

The target perampanel dose was 8 mg/day, but due to dose adjustments triggered by TEAEs, the average dose at the efficacy visit was 6 mg/day. To minimize the effects of possible unblinding due to TEAEs, the participants were instructed not to discuss their symptoms with the on‐site tremor assessor, who performed videotaping and oversaw the completion of questionnaires. The rater for the primary endpoint received videotapes that were coded without identification of the arm or baseline versus treatment visit, thus avoiding the confounding effects that may occur with on‐site tremor rating.

Gironell and colleagues24 reported that of 12 participants titrated to 4 mg/day perampanel in an open trial, 4 withdrew due to AEs, whereas the remaining 8 exhibited significant tremor reduction on several scales. By comparison, in our blinded trial, 4/11 participants in the perampanel arm reported marked improvement on the SGIC Scale, 3/11 had >50% total score reduction in the TETRAS‐ADL Scale, and 3/11 were rated >50% improved by the blinded rater in the TETRAS‐P Scale, whereas none in the placebo arm (n = 15) had such responses.

Comparisons of our efficacy to that of propranolol or primidone are not possible as studies of these drugs mostly relied on accelerometric measures of postural hand tremor rather than validated rating scales.2 The average tremor reduction in the topiramate trial was 39%25 compared with an average reduction in our study of 20%. Given the heterogeneity of ET, the small sample size in our study, and the very large sample size in the topiramate trial (n = 208), head‐to‐head comparison of perampanel with topiramate, propranolol, and primidone are needed to determine their relative efficacy.

Limitations of the study include small sample size and high dropout rate. Outcome effects were strong enough to yield positive P values, but a larger sample would have provided more definitive estimations of average tremor reduction. The multivariate imputation by chained equations statistical method imputed values in the intent‐to‐treat population with the assumption of random dropouts, which may be questioned in view of frequent perampanel‐related AE‐triggered dropouts. However, the outcomes data for the completer population, in which the data were complete, are very similar to the outcomes for the intent‐to‐treat population, suggesting that little bias occurred.

Our study was designed to reduce TEAEs by excluding participants considered to be a fall risk, once‐a‐day bedtime dosing, and a slow titration of 2‐mg increments every 3 weeks,26 with close monitoring. Compared with the AEs seen at 8 mg/day in 3 pooled phase 3 epilepsy trials,16 our rates of dizziness (36% vs. 32% in epilepsy) and somnolence (18% vs. 11%) were comparable, but the rates of imbalance and falls (50% vs. 10%), fatigue (27% vs. 8%), and irritability (27% vs. 7%) were greater. In addition, our participants were more likely to report impaired cognition, confusion, and worse tremor. As a result, more of our participants discontinued perampanel than did epilepsy participants taking 8 mg/d (36% vs. 9%).

Age does not substantially affect perampanel clearance, but a role for age in perampanel tolerability is suggested by a meta‐analysis finding that epilepsy participants aged 65 and older in phase 3 trials were more likely than younger participants to experience dizziness, fatigue, and falls.27 The appreciable rate of poorly tolerated TEAEs in the ET population may thus be in part age related, as many are aged >65 years. In addition, imbalance and irritability were more common in our participants than in older epilepsy patients,27 which may reflect vulnerability due to inherent balance and psychiatric/personality anomalies in the ET population.28, 29, 30

Our ET patients experienced more imbalance, fatigue, irritability, and slurred speech at 7 to 8 mg/day compared with when they took 3 to 4 mg/day. Only 1 did not tolerate 4 mg/day, whereas 7 withdrew at doses >4 mg/day. Two of our high‐efficacy responders were taking 4 to 5 mg/day. These considerations, combined with the open‐label findings of Gironell and colleagues,19 suggest that further evaluation of perampanel for ET may be appropriate at doses lower than the target dose employed in this trial.

In conclusion, a fraction of our participants displayed a clinically significant tremor reduction in response to perampanel. The incidence of AEs was appreciable, leading to frequent dose reduction or withdrawal, especially at doses >4 mg/day. Further evaluation of perampanel at 4 mg/day for tremor may be justified.

Author Roles

(1) Research Project: A. Conception, B. Organization, C. Execution; (2) Statistical Analysis: A. Design, B. Execution, C. Review and Critique; (3) Manuscript Preparation: A. Writing of the First Draft, B. Review and Critique.

A.H.: 1A, 1B, 1C, 2A, 3A

W.T.: 1B, 1C, 3B

R.E.: 1C, 2B, 2C, 3B

Disclosures

Ethical Compliance Statement

We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this work is consistent with those guidelines.

Funding Sources and Conflicts of Interest

The study was funded by Eisai, Inc. and supported by the U.S. Department of Veterans Affairs. The authors declare that there are no conflicts of interest relevant to this work.

Financial Disclosures for the Previous 12 Months

Adrian Handforth has received research support from Medtronic, Inc. Rodger Elble has received consulting fees from Cavion, Cadent, and Praxis Medicines and receives research funds from the Kiwanis Neuroscience Research Foundation, Illinois–Eastern Iowa District. The authors declare that there are no additional disclosures to report.