Sleep quality and daytime sleepiness in patients treated with adjunctive perampanel for focal seizures

Epilepsy & Behavior 63 (2016) 57–62

Contents lists available at ScienceDirect

Epilepsy & Behavior journal homepage: www.elsevier.com/locate/yebeh

Sleep quality and daytime sleepiness in patients treated with adjunctive perampanel for focal seizures Manuel Toledo a,⁎, Montserrat Gonzalez-Cuevas a, Julia Miró-Lladó b, Albert Molins-Albanell c, Mercé Falip b, Ana Belén Martinez d, Santiago Fernandez e, Manuel Quintana a, Roser Cambrodi f, Estevo Santamarina a, Javier Salas-Puig a a

Epilepsy Unit, Neurology Department, Vall d'Hebron University Hospital, Barcelona, Spain Epilepsy Unit, Neurology Department, Bellvitge University Hospital, Barcelona, Spain c Epilepsy Unit, Neurology Department, Josep Trueta University Hospital, Girona, Spain d Neurology Department, Son Espases Hospital, Palma de Mallorca, Spain e Neurology Department, Hospital Plató, Barcelona, Spain f Sleep Unit, Vall d'Hebron University Hospital, Barcelona, Spain b

a r t i c l e

i n f o

Article history: Received 8 June 2016 Revised 26 July 2016 Accepted 4 August 2016 Available online xxxx Keywords: Epilepsy Somnolence Sleep quality AED load AMPA receptor Glutamate

a b s t r a c t Purpose: The purpose of this study was to evaluate subjective sleep quality and daytime sleepiness in patients receiving adjunctive perampanel for focal seizures. Methods: We conducted a multicenter, prospective, interventional, open-label study in patients aged N 16 with focal seizures who received adjunctive perampanel (flexible dosing: 2–12 mg). Sleep quality was assessed with the Pittsburgh Sleep Quality Index (PSQI) and daytime sleepiness with the Epworth Sleepiness Scale (ESS) at baseline and 3 and 6 months after initiating perampanel. Patients with modifications in their baseline AEDs or sleep medications were excluded. Results: In 72 patients with drug-resistant focal seizures, mean baseline PSQI score (±standard deviation) was 7.26 (±4.6), and ESS was 6.19 (±4.2). At 3 months (median perampanel dose: 4 mg), there was no significant mean change from baseline in ESS score (n = 61) and a significant improvement in PSQI (−1.51 points; n = 44; p = 0.007), driven mainly by improved sleep efficiency (p = 0.012). In the 31 patients with 6-month data, ESS (but not PSQI) improved significantly at 6 months vs baseline (p = 0.029). The only factor significantly correlated with sleep parameters was number of baseline AEDs (higher number correlated with worse daytime sleepiness). Seizure frequency was reduced significantly from baseline at 3 and 6 months. In bivariate analysis, neither PSQI nor ESS was associated with seizure frequency, suggesting that the changes in daytime sleepiness and sleep quality may be independent of the direct effect on seizures. Conclusion: Adjunctive perampanel did not worsen sleep quality or daytime sleepiness at 3 months and reduced daytime sleepiness in patients continuing perampanel for 6 months. Perampanel may be a suitable AED in patients with sleep disorders, in addition to refractory focal seizures. © 2016 Elsevier Inc. All rights reserved.

1. Introduction There is a well-known reciprocal interaction between sleep and epilepsy. Sleep disorders are common in epilepsy and are one of the factors that may further impact the quality of life of patients with epilepsy [1]. Disorders such as sleep apnea, restless leg syndrome, and periodic limb movement disorder are more likely to occur among patients with epilepsy, and sleep problems are twice as prevalent in patients with epilepsy than in healthy subjects [1]. Complaints of insomnia have been reported in about half of patients with epilepsy and daytime sleepiness

⁎ Corresponding author at: Passeig Vall d'Hebron 119-129, 08035 Barcelona, Spain. E-mail address: [email protected] (M. Toledo).

http://dx.doi.org/10.1016/j.yebeh.2016.08.004 1525-5050/© 2016 Elsevier Inc. All rights reserved.

in 70% [1,2]. Epilepsy can also have a direct impact on sleep. Nocturnal seizures, which are relatively frequent in some types of epilepsy, interrupt sleep and may thus impair sleep architecture by enhancing sleep fragmentation and increasing the percentage of time spent awake or in a light sleep period [1]. Conversely, sleep can influence seizures; for example, sleep deprivation is a common trigger in epilepsies. Another factor in this complex relationship is the impact of antiepileptic drugs (AEDs) on sleep. Several studies have explored the dose– response relationship between AEDs and sleep characteristics in people with epilepsy. The overall consensus is that all AEDs have the potential to influence sleep architecture – some positively, some negatively, and often specific to the type of epilepsy – and that understanding this influence can be important in optimizing seizure control and quality of life for people with epilepsy [3–5].

58

M. Toledo et al. / Epilepsy & Behavior 63 (2016) 57–62

Perampanel is an AED approved for the adjunctive treatment of focal seizures and of primary generalized tonic–clonic seizures. Its mode of action is unique among the AEDs, in that it is a selective, noncompetitive antagonist of glutamate AMPA receptors. Evaluation of its efficacy and safety in three placebo-controlled randomized Phase III studies in patients with refractory focal seizures showed that add-on therapy with perampanel decreased seizure frequency significantly at daily doses between 4 and 12 mg/day with good tolerability and safety data compared with that with placebo [6,7]. Clinical case series of perampanel use in routine clinical practice confirm the potential benefits of perampanel regarding seizure reduction and the safety profile [8]. Somnolence is one of the most common side effects reported during perampanel use and is one of the reasons why the drug is recommended to be taken at bedtime [6,8,9]. However, no studies have explored the impact of perampanel on quality of sleep or on daytime sleepiness using sleep-specific measures. We therefore explored the impact of adjunctive perampanel on patient-reported subjective sleep quality using the Pittsburgh Sleep Quality Index (PSQI) and on subjective daytime sleepiness using the Epworth Sleepiness Scale (ESS) and attempted to account for the complex interrelationship between seizures and sleep using bivariate analysis to look for confounding factors. 2. Materials and methods 2.1. Study design We conducted a prospective, multicenter, open-label, interventional study in patients who started adjunctive treatment with perampanel for management of focal epileptic seizures in routine clinical practice. The study was approved by the Local Ethic Committee (MAT-LEV2014-01) and was conducted according to the International Conference on Harmonization notes for Guidance on Good Clinical Practice and the Declaration of Helsinki. All patients provided written informed consent. The primary objective was to assess the impact of adjunctive perampanel on subjective perception of quality of sleep and daytime sleepiness in patients with focal seizures. Secondary objectives included assessing these parameters in the subset of patients who were followed for 6 months and to evaluate efficacy, tolerability, and safety during treatment. 2.2. Setting and participants Patients included in the study were older than 16 years, with a confirmed diagnosis of epilepsy with focal onset seizures (with or without generalization) according to the criteria of the International League Against Epilepsy (ILAE, 2010) and who were prescribed adjunctive perampanel. The decision to prescribe perampanel and the dosage and titration schedule were made on a case-by-case basis by the prescribing neurologist, in accordance with the perampanel prescribing information and taking into account the overall clinical picture. Only after that decision was made were patients considered for eligibility for this study. Perampanel was never used with the intention of improving or modifying sleep characteristics. Patients were enrolled from the epilepsy outpatient clinics of 5 Spanish centers: Vall d'Hebron University Hospital, Bellvitge University Hospital, Josep Trueta University Hospital, Son Espases Hospital, and Hospital Plató; they were followed during the period from June 2014 to June 2015. Patients were required to have been on stable doses of baseline AEDs and any sleep medications for at least three months before the baseline visit. Occasional and short-term use (b 1 week) of sleep inducers such as benzodiazepines was permitted. Patients were not eligible for inclusion if they had nonepileptic seizures, untreated or spontaneous reports of sleep disorders, progressive neurological diseases such as tumors or dementia, systemic progressive or disabling diseases such as heart or hepatic failures, or learning disability or were unable to understand or complete the questionnaires. Potential subjects were prospectively assessed for sleep/wake habits, potential sleep disorders, daytime

sleepiness, and quality of sleep and by means of a standardized clinical interview, which also included the Epworth Sleepiness Scale (ESS) and the Pittsburgh Sleep Quality Index (PSQI). Potential subjects were excluded if the presence of sleep disorders such as restless leg syndrome, sleep apnea, period limb movement, parasomnia, REM sleep disorder, or narcolepsy was suspected based on the interview. 2.3. Interventions and assessments Baseline characteristics before starting perampanel were recorded from patients' charts and were confirmed at the baseline visit. Seizure types and seizure frequency (per 30 days) at baseline were calculated from prebaseline patient seizure diaries. At the baseline visit, ESS and PSQI questionnaires were completed. The ESS is an 8-item self-rated scale, with a minimum score of 0 and possible maximum score of 24 (excessive daytime sleepiness); the ‘normal’ range is 0–10 [10,11]. The PSQI is a 9-question self-rated scale (with optional items that can be scored by bed partners or roommates, which we did not include). For questions 1–4, patients provide numerical data (bedtime, get-up time, sleep latency in minutes, and sleep duration in hours), and questions 5–9 are rated on a 4-point Likert scale from 0 (very good) to 3 (very bad). To obtain the total score, the response to each question is converted into a component score (0–3) based on established thresholds and formulae, and the 7 component scores are combined to give a total score that can range from 0 (no difficulties) to 21 (severe difficulties in all sleep areas). The ‘normal’ range is usually considered to be b 5 points [12]. Incomplete or inadequate PSQI evaluations were excluded from analysis. We evaluated the total PSQI score, the 7 individual component scores, the proportion of patients who improved/worsened/did not change for each individual component score (i.e., from 0 to 1 would be worsening; from 3 to 2 would be improvement), and the numerical data for key questions (sleep latency [minutes]; sleep efficiency [time asleep as % of total time in bed]). Perampanel was added to baseline AEDs and dosed in accordance with the prescribing information: once daily, at bedtime, with a starting dose of 2 mg, increasing to 4 mg after the first 1–2 weeks, based on seizure control and tolerability. Individual investigators used their discretion to select the perampanel titration schedule and maintenance dose; thus, the titration schedule was different for all patients, and the starting dose of 2 mg was often maintained for up to one month before up-titration. Only patients whose concomitant AED dosage/regimen and sleep medications or dosages remained unchanged for the duration of perampanel treatment were included in the analysis. Follow-up visits to reassess ESS and PSQI were scheduled 3 months after the baseline visit, and patients completed these questionnaires based on sleepiness and sleep quality during the previous month. When available, patients were also reassessed at 6 months. Patient seizure diaries were used to calculate the median monthly (30-day) seizure frequency based on the total seizure count over the entire 3-month period (baseline to 3-month visit or 3-month to 6-month visit). Median percentage change in seizure frequency relative to baseline was calculated at 3 and 6 months, and responder rate was reported as the proportion of patients who achieved ≥50% reduction in seizure frequency relative to baseline. Side effects reported by patients during the previous 3 months were collected, and the presence of specific side effects of interest was specifically elicited (dizziness, nausea/vomiting, headache, gait, irritability, blurred vision, psychiatric side effects, suicide ideation). In patients taking carbamazepine, phenytoin, phenobarbital, lamotrigine, or valproic acid, blood samples were taken at baseline and at follow-up visits, and blood levels of these AEDs were recorded. 2.4. Statistical methods All statistical analyses were performed using SPSS 21.0 software. p b 0.05 was considered statistically significant. Categorical variables

M. Toledo et al. / Epilepsy & Behavior 63 (2016) 57–62

59

were represented as frequency (% of patients) and numeric variables as the mean (± standard deviation) for normal distributions and the median (± interquartile range) for non-normal distributions as assessed by QQ Plot and Kolmogorov–Smirnov test. Paired-sample t-test was used to evaluate mean change in ESS and PSQI values between baseline and 3 months for the primary objective and between baseline and 6 months in patients who were evaluated for the secondary analysis. Missing data (i.e., values without a matching pair) were excluded. Because PSQI component scores are ordinal variables with ≥2 categories, the McNemar–Bowker test was used to evaluate changes in the PSQI component scores. To study the presence of potential confounding factors related to the variations in ESS or PSQI during the follow-up, we performed bivariate analysis, assessing continuous variables using Pearson's correlation or Spearman's correlation coefficients depending on the normal distribution, and categorical with Student's t test or ANOVA. Seizure change from baseline was assessed with the nonparametric Wilcoxon signedrank test.

and valproate (15.3% each); phenobarbital (11.1%); zonisamide (9.7%); clonazepam (6.9%); and phenytoin, topiramate, diazepam, and retigabine (b5% each). The majority of patients (80%) were taking 2 or 3 concomitant AEDs (Table 1). The perampanel dose ranged from 2–8 mg during the first 3 months, and the median dose at the time of the 3-month visit was 4 mg (range: 2–6 mg). Of the 72 patients enrolled and who received perampanel, 61 were ongoing at the 3-month visit, and 31 were ongoing at the 6-month visit (Fig. 1). All 11 patients who dropped out of the study during the first 3 months did so because they discontinued perampanel, because of either lack of efficacy (n = 5) or adverse events (n = 6; mostly dizziness). A total of 30 patients discontinued between 3 and 6 months, mostly because of lack of efficacy (n = 18, which included 10 patients who were excluded because of changes in baseline AEDs), with fewer patients discontinuing during this period because of adverse events (n = 5) and 7 patients who were lost to follow-up (including 4 patients who were excluded because of changes in baseline AEDs).

3. Results

3.2. Daytime sleepiness

3.1. Participants

The mean baseline ESS score (±SD) was 6.19 (±4.24, n = 72), indicating that average daytime sleepiness was within the normal range for healthy subjects (score of b 10). In the 61 patients with values at baseline and 3 months, there was no significant change in mean ESS score from baseline to 3 months (mean change: +0.39; 95% CI: −1.36–0.58; p = 0.427); absolute mean scores at each time point are shown in Fig. 2. Bivariate analysis at 3 months showed that ESS improvement was negatively correlated with number of baseline AEDs (Spearman's correlation coefficient R = 0.302; p = 0.020); i.e., improvement from baseline in the ESS scores was less likely in patients with a higher number of baseline AEDs. Other variables, including perampanel dose and seizure frequency, were not associated with ESS score.

During the study period, we recruited 72 consecutive patients. Baseline demographic and clinical data are shown in Table 1. In general, our patients had similar characteristics that indicated severe, drug-resistant epilepsy. Patients had used a median of 4 previous AEDs, there was a wide variety of seizure types and frequency, and most patients had a history of epilepsy of over 20 years. Previous history of psychiatric disorders was present in 27 patients (37.5%; Table 1), although all were considered stable or nonactive when perampanel was initiated. The most commonly used concomitant AED was eslicarbazepine (31.9%) followed by levetiracetam and lacosamide (25% each); clobazam and carbamazepine (22.2% each); lamotrigine (20.8%); oxcarbazepine

3.3. Sleep quality Table 1 Patient characteristics and epilepsy history at baseline. Characteristic (±SD/IQR) [range]

Patients (N = 72)

Mean age, years (SD) Female gender, % (n) Median baseline seizure frequency/month (IQR) Median duration of epilepsy, years (IQR) Median number of previous AEDs Median number of concomitant AEDs at baseline 1 AED, % (n) 2 AEDs, % (n) 3 AEDs, % (n) 4 or 5 AEDs, % (n) Psychiatric comorbidity, % (n) Depression Anxiety/irritability Psychotic disorder Personality disorder Hyperactivity disorder Seizure etiology, % (n) Unknown Vascular Mesial temporal sclerosis Tumor Malformation of cortical development Brain injury Cerebral palsy Others Seizure origin, % (n) Frontal Temporal Parietal Occipital Multifocal

42.41 (±14.3) [17–75] 56.9% (41) 24.0 (11–66) 21.0 (13–33.5) 4 [1–9] 2 [1–5] 9.7% (7) 44.4% (32) 36.1% (26) 9.7% (7) 37.5% (27) 19.4% (14) 12.5% (9) 2.7% (2) 1.4% (1) 1.4% (1)

The mean baseline total PSQI score (±SD) was 7.26 (±4.55, n = 72). In the 44 patients with PSQI scores available at baseline and 3 months, mean total PSQI score improved significantly from baseline to 3 months (mean change: − 1.51 points; 95% CI: − 2.59 to − 0.43; p = 0.007). Absolute mean total scores at each time point are shown in Fig. 2. When the PSQI components were analyzed individually, no significant change from baseline was detected: sleep quality (p = 0.623), sleep latency (p = 0.572), sleep duration (p = 0.442), sleep efficiency (p = 0.101), sleep disturbances (p = 0.105), use of hypnotics (p = 0.504), and daytime dysfunction (p = 0.384). However, for each component score, a higher proportion of patients had improvement in component score than worsening (Table 2). When PSQI answers 1–4 were analyzed as numerical variables, a significant change from baseline was detected for sleep efficiency. This

27.8% (20) 9.7% (7) 18.1% (13) 6.9% (5) 12.5% (8) 15.3% (9) 5.6% (4) 8.3% (6) 31.6% (21) 53.7% (36) 5.6% (4) 6.9% (5) 8.4% (6)

AED: antiepileptic drug; IQR: interquartile range; SD: standard deviation.

Fig. 1. Patient disposition. Flow diagram showing number of patients evaluated at each stage of the study. (*n) indicates number of patients excluded because of changes in baseline AEDs, and (**n) indicates number of patients excluded because of changes in concomitant sleep medications; these numbers are not mutually exclusive, and the same patient may be counted in both groups but only once in the total discontinuations.

60

M. Toledo et al. / Epilepsy & Behavior 63 (2016) 57–62

Fig. 2. Self-assessed sleep quality and daytime sleepiness at baseline and 3 months. Mean Pittsburgh Sleep Quality Index (PSQI) and Epworth Sleep Scale (ESS) scores at baseline and 3 months after initiating adjunctive perampanel. Significant reduction (improvement) in sleep quality, as assessed by global PSQI score was seen after 3 months of perampanel treatment, relative to baseline, in the 44 patients with evaluable data. There was no change in self-assessed daytime sleepiness, as assessed by ESS score, from baseline to 3 months, in the 61 patients with ESS data.

parameter (time spent asleep as a percentage of time spent in bed) improved from 80.4% (±SD 17.7) at baseline to 87.4% (±15.5) at 3 months (p = 0.012). Sleep duration also improved significantly, from 6.8 h (±1.7) at baseline to 7.5 h (±1.7) at follow-up (p = 0.010). In the bivariate analysis, no factor was found to be associated with PSQI improvement (age, gender, seizure frequency, perampanel dosage, duration of epilepsy, number of baseline AEDs, number of previous AEDs, or presence of adverse events). 3.4. Sleep parameters in patients with 6-month data We analyzed the evolution of sleep parameters in the 31 patients who remained on perampanel and had both ESS and PSQI data at baseline and 6 months. The mean baseline ESS score in this subgroup (8.1 ± 4.4) was slightly higher than in the overall group; and mean ESS score at 6 months (6.4 ± 3.7) was significantly lower than at baseline (p = 0.029; mean change: −1.73; 95% CI: −3.27 to − 0.19). Absolute mean scores are shown in Fig. 3. In this subgroup of 31 patients, mean PSQI score was improved from baseline (7.1 ± 4.6) to 6 months (5.9 ± 3.6; p = 0.185; mean change: − 1.16; 95% CI: − 2.91–0.59), but the difference was not significant (Fig. 3). The median perampanel dose between the 3- and 6-month visits was 4 mg in these patients, ranging from 2–8 mg. Again, the only risk factor associated with worse ESS performances in this group of patients was number of baseline AEDs (p = 0.003). 3.5. Seizure responses Seizure frequency improved significantly during the first 3 months of follow-up relative to baseline. The median monthly seizure frequency Table 2 Proportion of patients with worsening, no change, and improvement in individual-item PSQI scores from baseline to three months. Pittsburgh items

Worsening

No change

Improvement

Quality Latency Duration Efficiency Fraction Sleep-medications Daytime dysfunction

13.3% 15.6% 11.1% 11.6% 8.9% 6.7% 15.6%

57.8% 64.4% 60% 53.5% 71.1% 80% 51.1%

28.9% 20% 28.9% 34.9% 20% 13.3% 33.3%

Fig. 3. Self-assessed sleep quality and daytime sleepiness in patients with 6 months of data. Mean Epworth Sleepiness Scale (ESS) score and Pittsburgh Sleep Quality Index (PSQI) global score in patients with data available at baseline and at both 3 and 6 months after initiating perampanel (n = 31).

decreased from 24 (interquartile range [IQR]: 11–66; range: 2–314) at baseline (n = 67) to 21 (8–52; 0–259) at 3 months (n = 54; p = 0.002); a median 16.7% reduction in seizure frequency. In the patients who had data available at 6 months, the response was sustained; median monthly seizure frequency was 20 (3–45, 0–150; n = 40, p b 0.001 vs baseline); a median 33.3% reduction in seizure frequency. Two of the 72 patients who started perampanel remained seizure-free from first dose of perampanel to the 3-month visit (2.8% pragmatic seizure-free rate or 3.7 of the 54 patients with 3-month seizure data). Two of the 40 patients with 6-month data remained seizure-free from the first dose of perampanel to the 6-month visit (5% seizure-free). All seizure types, including focal seizures with and without impairment of consciousness and those with evolution to a bilateral convulsion, improved during the follow-up. At 3 months, 16 of 54 patients were responders (29.6%); at 6 months, 19 of 40 patients were responders (47.5%). 3.6. Adverse events At the baseline visit, adverse events (AEs) associated with existing AEDs were reported by 47.2% of patients (Table 2). The most frequent background AE before starting perampanel was fatigue (n = 18, 25.0%), followed by dizziness (n = 17, 23.6%), mood disorders/irritability (n = 15, 20.8%), and somnolence (mild to moderate; n = 14, 19.4%). Treatment-emergent AEs (TEAEs) were reported by 62.5% of patients during the first 3 months and by 65.6% of patients followed up at 6 months. However, the majority of TEAEs at each of these time points were either transient or tolerated by patients and treatment continued (39/45 of patients with AEs at 3 months continued treatment, 86.7%, and 35/40 patients at 6 months, 87.5%). The most frequent TEAEs were somnolence and dizziness (Table 3). Fourteen patients reported somnolence as a baseline AE before starting perampanel: six of these patients had increasing somnolence after adding perampanel, 6 had no worsening of somnolence, and 2 had reduced daytime sleepiness (ESS score N 10 at baseline improved by 4 points). Of the 16 patients reporting somnolence as an AE during perampanel treatment, 8 had also reported somnolence at baseline; somnolence was mild and transient in 14, and in the other two patients, perampanel had to be discontinued because of somnolence. Eleven of the 72 patients (15.3%) discontinued perampanel treatment because of AEs — 6 during the first 3 months and 5 between the 3- and 6-month visits. The TEAEs leading to discontinuation were dizziness (n = 3), ataxia (n = 3), increased seizure frequency (n = 3), and

M. Toledo et al. / Epilepsy & Behavior 63 (2016) 57–62 Table 3 Adverse events before and during perampanel treatment.

Any AE, n (%) AEs leading to discontinuation Most common AEs Dizziness Somnolence Ataxia Anxiety/irritability Headache Increased seizure frequency Fatigue

Baseline (N = 72)

AEs between baseline and 3 months (N = 72)

AEs between 3 and 6 months (N = 61)

34 (47.2) –

45 (62.5) 6 (8.3)

40 (65.6) 5 (8.2)

17 (23.6) 14 (19.4) 0 15 (20.8) 0 0 18 (25.0)

12 (16.6) 16 (22.2) 7 (9.7) 6 (8.3) 4 (5.5) 3 (4.2) 1 (1.3)

5 (8.2) 8 (13.1) 3 (4.9) 1 (1.6) 0 0 4 (6.5)

somnolence (n = 2). No major psychiatric disorders or suicidal ideation was reported. A total of 7 patients (9.7%) reported anxiety or irritability; all of whom had previous history of psychiatric disorders. 3.7. Concomitant AEDs We explored the sleep and seizure-response data for patterns relating to concomitant AEDs. We found no differences between different individual concomitant AEDs or when AEDs were categorized by a mechanism of action or by cytochrome p450 status — inducer (carbamazepine, phenytoin, phenobarbital) vs inhibitor (valproic acid) vs no metabolic impact. There were no significant changes in the blood levels of carbamazepine, phenytoin, phenobarbital, lamotrigine, or valproate between the baseline and the follow-up visit(s). At baseline, 14 patients (19.4%) were taking hypnotics (lorazepam, n = 10; lormetazepam, n = 4). During follow-up, 9 of these patients (13.3% of total) stopped taking the hypnotic and were therefore excluded from the study. Hypnotics were discontinued in all 9 patients because of patient-perceived improvement in the quality of sleep and not because of excessive daytime somnolence in any patients. No patients who discontinued hypnotics needed to restart them during follow-up. 4. Discussion To our knowledge, this is the first prospective study of subjective sleep quality and daytime sleepiness in patients treated with adjunctive perampanel for focal seizures. In line with the existing literature, perampanel improved seizure control in our patients, and the adverse events were in line with those previously reported [7,8]. Somnolence is the most commonly reported adverse event in perampanel trials [6], so we expected to see an increase in daytime sleepiness after introducing perampanel. However, we found no increase in daytime sleepiness with adjunctive perampanel, a reduction in daytime sleepiness in the patients who remained on perampanel for 6 months, and overall improvements in sleep quality from 3 months, particularly sleep efficiency (time asleep as a proportion of time in bed). We suggest that perampanel may help to improve sleep in some patients, thus reducing daytime sleepiness over the longer term. Whether the effect on sleep is direct, or indirect through effects on seizures, is not clear. Most AEDs have been shown to impact sleep architecture and/or daytime sleepiness — some positive and some negative. For example, phenobarbital, valproic acid, and levetiracetam are associated with daytime sleepiness, and carbamazepine, tiagabine, and pregabalin improve total sleep time and sleep efficiency [1]. It is not clear whether these effects are related to seizure control or a direct effect of the drug on sleep architecture. The mean baseline PSQI score in our sample was as expected for patients with epilepsy, which is usually in the upper limit of normality [2]. We found improvements in the global sleep quality and specifically in sleep efficacy, which were independent of

61

any confounding factors such as demographic characteristics, seizures, perampanel dosing, or baseline treatments. The fact that none of the variables we analyzed were significantly associated with changes in PSQI score suggests that the improvement in PSQI at 3 months could be attributable to the addition of perampanel. However, our data are preliminary data in a relatively small population, so this should be tested further in a larger cohort. Based on our results, however, we believe perampanel could be added to the group of AEDs to consider in patients who have seizures with comorbid insomnia. In such patients, carbamazepine, tiagabine, and pregabalin are often considered as they have demonstrated favorable effects on total sleep time and efficiency [1]. Clinical experience and the perampanel label recommend bedtime dosing because of somnolence [8,9,13]. However, we did not observe a significant impact of perampanel on sleep latency or the total sleep time in our patients. Despite this, we saw improvements in sleep efficiency (subjective total sleep time as a percentage of total time in bed) and in overall sleep quality. Therefore, these preliminary data suggest that perampanel may be one AED option for patients with epilepsy with sleep problems, perhaps specifically sleep maintenance. But this needs to be validated in further, larger studies. Studies using the ESS have shown that excessive daytime sleepiness is more common in patients with epilepsy than in healthy subjects [2,11]. In addition, somnolence is a common AE associated with some AEDs [1], including perampanel. The pooled analysis of three perampanel clinical trials in focal seizures reported somnolence as an AE in 12%, 9%, 16%, and 18% of patients taking 2-, 4-, 8-, and 12-mg perampanel, respectively (vs 7.2% with placebo) [6]. In our analysis, somnolence was reported as an AE in 22% of patients in the first 3 months and 13% in 3–6 months, which is in line with the clinical trial data. Somnolence in our patients was usually mild and transient, and only 2 patients discontinued perampanel because of somnolence. Rather than be impaired by the addition of perampanel to baseline AEDs, daytime sleepiness (measured by the ESS) did not change from baseline to 3 months and improved at 6 months in the 31 patients who remained on adjunctive perampanel. Overall, perampanel does not appear to increase daytime sleepiness. The sleepiness and somnolence typically reported as side effects with perampanel do not necessarily equate with true daytime sleepiness as measured by a validated scale. Side effects of sleepiness and somnolence are thought to be related to peak-dose effects at the AMPA receptor, and if bedtime dosing is used, then problem daytime sleepiness may not necessarily occur. Our patients' clinical profiles were quite similar to those included in the pivotal clinical trials [7]; however, the slow titration and the low maintenance dose of perampanel (median: 4 mg) in our sample may account for the better tolerability we saw compared with those in clinical trials. For example, up to 42% of patients reported dizziness in the clinical trials in the 12-mg perampanel group, but the rates were very similar if we compare our results with the 4-mg group in the clinical trials (16.6% vs 16.3%, respectively) [6]. As has been suggested previously [14,15], baseline AED overload (i.e., greater number of baseline concomitant AEDs) was associated with worse daytime sleepiness but, interestingly, had no impact on sleep quality in our study. The potential reciprocal benefits between sleep and seizure control – i.e., improved seizure control with greater sleep quality and improved sleep with better seizure control [1,2] – were not observed in our patients, but the study was not designed to assess this complex relationship. It was a small population sample of patients with medically refractory seizures with poor prognosis for seizure control. However, we saw a favorable clinical response with significant seizure reduction, no worsening of daytime sleepiness, and improvements in sleep quality, despite the relatively low doses of perampanel used in our patients, and with good tolerability. There are many limitations inherent to an open-label cohort study of this type, and this study should be considered as justification for further study in this area, rather than as robust evidence in itself. The open-label nature and the flexible dosing of perampanel according to the decision

62

M. Toledo et al. / Epilepsy & Behavior 63 (2016) 57–62

of the treating neurologist have the advantage of resembling routine clinical practice but also introduce a potential for bias and increase the heterogeneity of the populations. In addition, the sleep measures we used are subjective, rather than objective, and subjective reports of sleep parameters (sleep latency and total time awake, in particular) are known to mismatch with objective measures, especially in patients with insomnia [16]. Objective polysomnography to accurately measure sleep parameters and objective tests such as actigraphy or multiple sleep latency test to confirm suspected sleep disorders, time of sleep, or the presence of daytime somnolence would have been useful and would have excluded patients with sleep apnea and periodic limb movements more definitively than we were able. However, many objective techniques are not possible in this type of study. Objective polysomnography requires subjects to spend the night in a sleep laboratory, which disturbs normal routines and is not practical for many patients with epilepsy. Although nonlaboratory objective sleep measures are being developed, most are still prototypes and not in common usage [17]. The cohort design and lack of control group also limit the conclusions, and it would be interesting to see if our results can be confirmed in a randomized placebo-controlled study with multiple treatment groups. Another limitation, in this naturalistic setting, was our decision to exclude from analysis patients who had changes in their concomitant AEDs or hypnotics. Some patients who responded to perampanel, and whose concomitant AEDs were adjusted as a result, or reduced to optimize tolerability, therefore had to be excluded from the analysis. This contributed to the fact that less than half of patients had evaluable sleep data at the 6-month visit. Also, some patients were able to drop their sleep aid medications after perampanel addition but were excluded from the analysis. The reduction of drug load in patients with refractory epilepsy has been associated with improved quality of life, because of a reduction in side-effect burden [14,15]. This is an important potential benefit of perampanel that deserves further study. In conclusion, our data give preliminary evidence that adjunctive perampanel may improve sustainably the global quality of sleep, primarily because of improved sleep efficiency, in patients with focal seizures. Daytime sleepiness was not increased by the addition of perampanel to the existing AED regimen, and in fact, daytime sleepiness was reduced in patients who continued with perampanel treatment for 6 months. The single risk factor we found to be associated with sleep parameters was number of baseline AEDs — increasing number of AEDs reduced the chances of seeing improvements in daytime sleepiness. Acknowledgments Editorial support in the preparation of this manuscript was provided by Katherine Carpenter and funded by Eisai Farmacéutica, S.A.

Disclosure of conflicts of interest EISAI Europe Ltd. gave financial support for this investigatorinitiated study (reference number: MAT-LEV-2014-01). Manuel Toledo, Xavier Salas-Puig, Estevo Santamarina, Mercé Felip, Albert Molins-Albanell, and Julia Miró-Lladó have received honoraria for speaking engagements and advisory board contributions from EISAI Europe Ltd., BIAL, UCB-pharma, GSK, and Laboratorios Esteve. References [1] Jain SV, Glauser TA. Effects of epilepsy treatments on sleep architecture and daytime sleepiness: an evidence-based review of objective sleep metrics. Epilepsia 2014;55: 26–37. http://dx.doi.org/10.1111/epi.12478. [2] Ismayilova V, Demir AU, Tezer FI. Subjective sleep disturbance in epilepsy patients at an outpatient clinic: a questionnaire-based study on prevalence. Epilepsy Res 2015; 115:119–25. http://dx.doi.org/10.1016/j.eplepsyres.2015.06.009. [3] Malow BA, Bowes RJ, Lin X. Predictors of sleepiness in epilepsy patients. Sleep 1997; 20:1105–10. [4] Piperidou C, Karlovasitou A, Triantafyllou N, Terzoudi A, Constantinidis T, Vadikolias K, et al. Influence of sleep disturbance on quality of life of patients with epilepsy. Seizure 2008;17:588–94. http://dx.doi.org/10.1016/j.seizure.2008.02.005. [5] Manni R, Terzaghi M. Comorbidity between epilepsy and sleep disorders. Epilepsy Res 2010;90:171–7. http://dx.doi.org/10.1016/j.eplepsyres.2010.05.006. [6] Steinhoff BJ, Ben-Menachem E, Ryvlin P, Shorvon S, Kramer L, Satlin A, et al. Efficacy and safety of adjunctive perampanel for the treatment of refractory partial seizures: a pooled analysis of three phase III studies. Epilepsia 2013;54:1481–9. http://dx.doi. org/10.1111/epi.12212. [7] Kramer LD, Satlin A, Krauss GL, French J, Perucca E, Ben-Menachem E, et al. Perampanel for adjunctive treatment of partial-onset seizures: a pooled dose– response analysis of phase III studies. Epilepsia 2014;55:423–31. http://dx.doi.org/ 10.1111/epi.12527. [8] Steinhoff BJ, Hamer H, Trinka E, Schulze-Bonhage A, Bien C, Mayer T, et al. A multicenter survey of clinical experiences with perampanel in real life in Germany and Austria. Epilepsy Res 2014;108:986–8. http://dx.doi.org/10.1016/j.eplepsyres. 2014.03.015. [9] Fycompa SPC. Fycompa summary of product characteristics; 2015. [10] Johns MW. A new method for measuring daytime sleepiness: the Epworth sleepiness scale. Sleep 1991;14:540–5. [11] Khatami R, Zutter D, Siegel A, Mathis J, Donati F, Bassetti CL. Sleep–wake habits and disorders in a series of 100 adult epilepsy patients—a prospective study. Seizure Eur J Epilepsy 2006;15:299–306. http://dx.doi.org/10.1016/j.seizure.2006.02.018. [12] Buysse DJ, Reynolds CF, Monk TH, Berman SR, Kupfer DJ. The Pittsburgh sleep quality index: a new instrument for psychiatric practice and research. Psychiatry Res 1989; 28:193–213. [13] Krauss GL, Perucca E, Ben-Menachem E, Kwan P, Shih JJ, Clément J-F, et al. Long-term safety of perampanel and seizure outcomes in refractory partial-onset seizures and secondarily generalized seizures: results from phase III extension study 307. Epilepsia 2014;55:1058–68. http://dx.doi.org/10.1111/epi.12643. [14] Dash D, Aggarwal V, Joshi R, Padma MV, Tripathi M. Effect of reduction of antiepileptic drugs in patients with drug-refractory epilepsy. Seizure 2015;27:25–9. http://dx. doi.org/10.1016/j.seizure.2015.02.025. [15] Ben-Menachem E. Medical management of refractory epilepsy—practical treatment with novel antiepileptic drugs. Epilepsia 2014;55(Suppl. 1):3–8. http://dx.doi.org/ 10.1111/epi.12494. [16] Bianchi MT, Williams KL, McKinney S, Ellenbogen JM. The subjective–objective mismatch in sleep perception among those with insomnia and sleep apnea. J Sleep Res 2013;22:557–68. http://dx.doi.org/10.1111/jsr.12046. [17] Van de Water ATM, Holmes A, Hurley DA. Objective measurements of sleep for nonlaboratory settings as alternatives to polysomnography—a systematic review. J Sleep Res 2011;20:183–200. http://dx.doi.org/10.1111/j.1365-2869.2009.00814.x.