FormalPara Key Points

Lemborexant, an orexin receptor antagonist that induces sleep by inhibiting arousal center function, has demonstrated effectiveness in phase III studies in insomnia patients.

In addition to somnolence, the main adverse drug reaction (ADR) seen in pre-approval clinical trials, nightmares, abnormal dreams, sleep paralysis, hypnagogic hallucinations, suicidal ideation, and suicidal behavior have been suggested to occur as ADRs. Therefore, we conducted a postmarketing observational study to investigate the occurrence of these ADRs in daily clinical practice in Japan, as well as the efficacy of lemborexant and its relationship with background factors.

The incidences of ADRs in this postmarketing observational study were ‘somnolence’ at 7.65%, ‘nightmares’ at 1.76%, ‘abnormal dreams’ at 0.59%, and ‘sleep paralysis’ at 0.20%; no suicidal ideation or behavior were observed. The incidence of all relevant ADRs was the same or slightly lower than in pre-approval clinical trials.

1 Introduction

Insomnia is a typical sleep disorder known to occur frequently in Japan [1], with previous studies showing that it has a prevalence rate of 13.5% [2]. In Japan, the most common treatment for insomnia is drug therapy, and the ‘Clinical guideline for proper usage and tapering of hypnotics’ states that hypnotics should be reduced or withdrawn as soon as possible after remission of insomnia [1].

Benzodiazepine receptor agonists (BZDs) are reported to be the most commonly prescribed hypnotics in drug therapy in Japan [3]. On the other hand, it is considered that long-term use of BZDs should be avoided due to issues of drug dependence [4]. Recently, suvorexant, an orexin receptor antagonist, has been approved in 2014 as a drug with higher safety than BZDs [5]. Lemborexant (Dayvigo® tablets 2.5, 5, and 10 mg; hereinafter referred to as LEM) is an orexin receptor antagonist discovered at Eisai Tsukuba Research Laboratories that was approved in 2020. Orexin receptor antagonists are thought to induce physiological sleep by inhibiting the function of the arousal center [6].

The safety and efficacy of LEM were confirmed in phase III studies (Study 303: E2006-G000-303, NCT02952820; and Study 304: E2006-G000-304, NCT02783729) conducted in insomnia patients [7, 8]. The most common adverse drug reaction (ADR) was somnolence, with an incidence of 10.7% [9], which was also expected to occur post marketing. It has also been suggested that nightmares, abnormal dreams, sleep paralysis, hypnagogic hallucination, and ADRs falling under suicidal ideation and suicidal behavior may also occur in insomnia patients and some due to the mechanism of action of LEM. Therefore, we conducted a postmarketing observational study (hereafter referred to as ‘this study’) to investigate the occurrence of these ADRs associated with LEM administration in daily clinical practice, as well as the efficacy of LEM and its relationship with background factors.

2 Methods

2.1 Study Design

This was a prospective, observational study aimed at verifying the safety and efficacy of LEM administered to insomnia patients in daily clinical practice in Japan. The scientific and ethical appropriateness of the plan was discussed by Eisai Co., Ltd, and the plan was implemented as the ‘Post-Marketing Observational Study of Dayvigo Tablets: Survey on the Occurrence of Adverse Drug Reactions in Insomnia Patients’ in accordance with the ‘Ministerial Ordinance on Good Post-Marketing Study Practice for Drugs (GPSP Ordinance)’. The plan of this study was registered in ClinicalTrials.gov (NCT04573556).

2.2 Subjects and Methods of Administration

This study was conducted in insomnia patients administered LEM for the first time at a medical institution that had signed a contract with Eisai Co. in Japan, and patients were included in the study if they were enrolled by Day 15 from the first day of the treatment. The diagnosis of insomnia was based on investigator assessment, and insomnia both with and without medical or psychiatric comorbidities were acceptable for enrollment. In addition, there were no restrictions on enrollment in terms of types of insomnia symptoms, presence or absence of prior medications, and presence or absence of concomitant insomnia drugs. The enrollment period was from 1 October 2020 to 30 September 2021, and the study period was from 1 October 2020 to 30 April 2022. The observation period per patient was 24 weeks or until discontinuation of the study.

The investigators administered LEM according to the dosage and administration approved in Japan, as follows: “Usually, for adults, 5 mg of lemborexant should be orally administered once daily just before bedtime. The dose may be adjusted according to symptoms but should not exceed 10 mg once daily”.

Patients were enrolled after obtaining consent to participate in this study, including the use and publication of information obtained orally or through a written document during this study. This study was conducted by the central registration method using the Electronic Data Capture (EDC) system. RaveEDC (Medidata Solutions, Inc.) was used as the EDC system.

Since this was a postmarketing observational study conducted according to the GPSP Ordinance, review by the Institutional Review Board (IRB) was not essential. Also, because this was an observational study conducted in daily clinical practice, no incentive was provided to patients.

2.3 Survey Items

2.3.1 Patient Background

Sex, age, diagnosis (insomnia with or without medical or psychiatric comorbidities), type of insomnia symptoms (difficulty falling asleep, difficulty staying asleep, problems awakening up too early in the morning), disease duration, presence or absence of concomitant therapies (cognitive behavioral therapy, sleep hygiene instructions), etc., were investigated. The diagnosis was made by a physician.

2.3.2 Discontinuation of the Study

For patients who discontinued the study before the end of the 24-week observation period, the reasons for discontinuation (‘lost to follow-up’, ‘symptoms improved’, ‘insufficient effect’, ‘adverse event(s)’, ‘withdrawal of consent for this study’, and ‘other’) were investigated. ‘Insufficient effect’ included cases in which LEM was effective but the effect was insufficient, as well as cases in which LEM was ineffective. ‘Adverse event(s)’ was selected for discontinuation due to an adverse event (AE) to be collected.

2.3.3 Safety

All undesirable or unintended diseases and their symptoms or signs that occurred in patients who received LEM were collected as AEs, regardless of causal relationship with LEM. Among AEs, those for which a causal relationship with LEM could not be ruled out were regarded as ADRs.

AEs were only collected by study physicians for certain target events. These included somnolence, parasomnia (abnormal dreams, nightmare, sleep paralysis, etc.), narcolepsy and associated conditions (hypnagogic hallucination, cataplexy, sleep attacks, etc.), and AEs falling under suicidal ideation and suicidal behavior.

2.3.4 Efficacy

Overall improvement (‘improved’, ‘unchanged’, and ‘worsened’) of insomnia symptoms from baseline, as assessed by the investigator based on the patient’s complaint, was assessed at each evaluation time point (Weeks 8, 16, and 24 [or at discontinuation]).

In addition, average weekday subjective sleep parameters were investigated for 1 week prior to the assessment date. The time from bedtime to falling asleep (sleep latency) and net sleep time (total sleep time) were investigated as sleep parameters.

2.4 Aggregate Analysis Method

The number and proportion of patients who discontinued the study and the reasons for discontinuation were calculated. Summary statistics of the number of days to discontinuation were calculated for patients who discontinued treatment due to ‘symptoms improved’.

In the safety analysis, the absolute numbers and incidences of ADRs were calculated in the safety analysis set, and ADRs leading to discontinuation or withdrawal were also calculated in the same manner. In addition, the incidences of ADRs were calculated by patient background. Furthermore, a factor analysis was performed using a logistic regression model, with the presence or absence of an ADR of somnolence used as an objective variable. For explanatory variables, factors potentially associated with ADRs from a medical point of view were selected. Each factor was examined using a univariate model, and variable selection (stepwise method: significance level 10%) was performed using a multivariate model.

The efficacy rate, calculated in the efficacy analysis set, was defined as the proportion of patients for whom the overall improvement was assessed as ‘improved’, and the evaluation at the final time point in each subject was tabulated and used as the final evaluation. In addition, efficacy rates were calculated by patient background. For the purposes of the tabulation, patients without a history of treatment with insomnia drugs were classified as ‘naïve’, those who discontinued all prior medications at the start of LEM were classified as ‘switched’, and those who continued prior medications even after LEM administration were classified as ‘add-on’. Furthermore, factor analysis was performed using a logistic regression model with the presence or absence of ‘improved’ in the overall improvement rating as an objective variable. For the explanatory variable, ‘history of treatment with insomnia drugs’ was selected instead of ‘presence or absence of prior medications’ or ‘presence of absence of concomitant drugs for insomnia’ when considering the correlation. Each factor was examined using a univariate model, and variable selection (stepwise method: significance level 10%) was performed using a multivariate model. For sleep parameters (subjective sleep onset latency [sSOL], subjective total sleep time [sTST], and subjective sleep efficiency [sSE]), summary statistics of measured values at each evaluation time point and changes from the baseline were calculated, and a one-sample Wilcoxon test was performed. Sleep efficiency was calculated as the percentage of total sleep time relative to the time from bedtime to wake-up time. SAS® version 9.4 was used for data analysis.

3 Results

3.1 Patient Background

In this study, 550 patients were enrolled, and a Case Report Form was obtained from 539 patients (Fig. 1). The safety analysis set included 510 patients and the efficacy analysis set included 498 patients.

Fig. 1
figure 1

Patient disposition. aThe breakdown of the number of excluded patients was counted in duplicate. bPatients who received LEM 5 mg once daily despite the instructions in the ‘Precautions Concerning Dosage and Administration’ that the dosage of LEM is 2.5 mg once daily when used in combination with a drug that moderately inhibits CYP3A, which is considered to correspond to use outside of the dosage and administration. LEM lemborexant, CYP cytochrome P450

Of the 510 patients included in the safety analysis set, 58.82% were female, mean ± standard deviation (SD) age was 53.7 ± 19.6 years, and 67.06% of patients were under 65 years of age (Table 1). The diagnosis was ‘insomnia without medical or psychiatric comorbidities’ (32.35%) and ‘insomnia with medical or psychiatric comorbidities’ (67.65%). The types of insomnia symptoms (multiple count allowed) were ‘difficulty falling asleep’ (84.31%), ‘difficulty staying asleep’ (57.45%), and ‘problems awakening up too early in the morning’ (25.49%).

Table 1 Patient background

3.2 Status of Discontinuation or Continuation

As for the status of treatment discontinuation or continuation by 24 weeks after LEM administration, treatment was discontinued (40.20%) and continued (59.80%) [ongoing at Week 24] (Table 2). The most common reason for discontinuation was ‘symptoms improved’ (16.08%) and ‘insufficient effect’ (9.41%). The proportion of patients who discontinued treatment due to ‘symptoms improved’ by comorbidity was 16.23% for ‘depression’, 13.95% for ‘bipolar disorder’, 16.67% for ‘diabetes mellitus’, and 15.79% for ‘hypertension’. The proportion of patients by history of treatment with insomnia drugs was 24.03% for ‘naïve’, 10.92% for ‘switched’, and 5.26% for ‘add-on’.

Table 2 Presence or absence of study discontinuation and its reasons

The mean ± SD number of days to discontinuation in the 82 patients who discontinued treatment due to ‘symptoms improved’ was 68.2 ± 44.1 days. The number of days to discontinuation by comorbidity was 73.3 ± 39.9 days for ‘depression’, 70.8 ± 39.3 days for ‘hypertension’, and 61.5 ± 50.5 days for ‘anxiety disorder’. The number of days to discontinuation by history of treatment with insomnia drugs was 64.1 ± 41.6 days for ‘naïve’, 87.4 ± 49.0 days for ‘switched’, and 69.0 ± 54.0 days for ‘add-on’.

3.3 Safety

The incidence of ADRs targeted for collection from the 510 patients included in the safety analysis set was 7.65% for ‘somnolence’, 1.76% for ‘nightmares’, 0.59% for ‘abnormal dreams’, and 0.20% for ‘sleep paralysis’; none of these ADRs were serious (Table 3). In the 39 patients who developed somnolence, the action taken for LEM was ‘discontinued’ in 18 patients (46.15%), ‘continued after dose reduction’ in 10 patients (25.64%), ‘no dose change’ in 9 patients (23.08%), and ‘continued after withdrawal’ and ‘none of the above (lost to follow-up)’ in 1 patient each (2.56%). The action taken for LEM by elderly/non-elderly was ‘no dose change’ in 3 of 5 patients (60.00%) aged 65 years or older and 6 of 34 patients (17.65%) aged under 65 years.

Table 3 Occurrence of ADRs

‘Sleep paralysis’ occurred in one patient, and its outcome was recovery after discontinuation of LEM. No narcolepsy-related symptoms including ‘hypnagogic hallucination’ or ADRs of suicidal ideation or behavior occurred.

As for the incidence of ADRs of somnolence by patient background, the incidence by age was 9.94% for ‘under 65 years’ and 2.98% for ’65 years or older’ (Table 4). The incidence of ADRs by presence or absence of concomitant insomnia drugs was 5.48% in patients without concomitant insomnia drugs and 12.27% in patients with concomitant insomnia drugs. As a result of a logistic regression model using presence or absence of somnolence (ADR) as an objective variable, presence of ‘concomitant insomnia drugs’ was significantly associated with the occurrence of ‘somnolence’ (odds ratio [OR] 2.401, 95% confidence interval [CI] 1.131–5.098) (Table 5).

Table 4 Occurrence of somnolence by patient background
Table 5 Background factors affecting somnolence (univariate/multivariate analysis)

3.4 Efficacy

3.4.1 Overall Improvement

Of the 498 patients in the efficacy analysis set, the efficacy rate at the final evaluation was 80.83% for the 459 patients included in the analysis for calculating the efficacy rate, after excluding patients whose evaluation time for overall improvement was outside the range adopted, evaluation date was unknown, or whose overall improvement was ‘unassessable’ (Fig. 2).

Fig. 2
figure 2

Efficacy rate by overall improvement. Note: The final evaluation for each patient was tabulated and the time window for evaluation was up to Week 26.

3.4.2 Efficacy Rates by Patient Background

The efficacy rate by patient background was 81.44% in males and 80.38% in females, 78.90% in patients aged under 65 years, and 84.77% in patients aged 65 years or older (Table 6). The efficacy rate by disease duration was 85.62% for ‘≤ 1’, 86.08% for ‘> 1 to ≤ 5’, 71.79% for ‘> 5 to ≤ 10’, and 66.10% for ‘> 10’. The efficacy rate tended to decrease slightly in patients with disease duration of ≥ 5 years. The efficacy rate by diagnosis was 82.99% for ‘insomnia without medical or psychiatric comorbidities’, and 79.81% for ‘insomnia with medical or psychiatric comorbidities’. The efficacy rate by type of insomnia symptoms (multiple count allowed) was 82.29% for ‘difficulty falling asleep’, 78.03% for ‘difficulty staying asleep’, and 82.35% for ‘problems awakening up too early in the morning’. There was no marked difference in the efficacy rates among the patient groups.

Table 6 Efficacy rates by patient background

The efficacy rate by history of treatment with insomnia drugs was 84.91% for ‘naïve’, 83.81% for ‘switched’, and 70.49% for ‘add-on’. The efficacy rate by comorbidity was 79.14% in ‘presence’ and 81.50% in ‘absence’ of depression, 84.62% in ‘presence’ and 80.43% in ‘absence’ of bipolar disorder, 81.48% in ‘presence’ and 80.74% in ‘absence’ of diabetes mellitus, and 90.14% in ‘presence’ and 79.07% in ‘absence’ of hypertension.

3.4.3 Factors Affecting Overall Improvement

As a result of a logistic regression model using presence or absence of ‘improved’ in overall improvement as an objective variable, presence of ‘concomitant therapies’ was significantly associated with ‘improved’ (OR 1.699, 95% CI 1.004–2.874). Moreover, ‘add-on’ was significantly associated with not being evaluated as ‘improved’ compared with ‘naïve’ (OR 0.365, 95% CI 0.200–0.664). However, ‘switched’ was not related to the evaluation of overall improvement compared with ‘naïve’ (OR 0.654, 95% CI 0.329–1.301) (Table 7). The number of patients by duration of disease was checked by history of treatment with insomnia drugs because many “Add-on” patients had a longer duration of disease, which may have resulted in a lower efficacy rate. As a result, the proportion of patients with the disease duration of 5 years or longer was 8.67% (15/173 patients) for “Naïve”, 36.36% (32/88 patients) for “Switched”, and 64.04% (57/89 patients) for “Add-on”, indicating that the proportion of patients with a longer disease duration was higher in “Add-on” patients.

Table 7 Background factors affecting overall improvement (univariate/multivariate analysis)

3.4.4 Sleep parameters

Changes in sSOL, sTST, and sSE from baseline are shown as sleep parameters (Fig. 3).

Fig. 3
figure 3

Changes in sleep parameters (subjective sleep onset latency, subjective total sleep time, and subjective sleep efficiency) from baseline (efficacy analysis set). Median ± quartile range. *p < 0.001 for comparison with the baseline value. a The time window for evaluation was ±2 weeks of the reference date (Weeks 8, 16, and 24). b The final evaluation for each patient was tabulated and the time window for evaluation was up to Week 26.

Changes in sSOL from baseline (median) were − 30.0 min at Week 8, − 30.0 min at Week 16, − 30.0 min at Week 24, and − 30.0 min at the final evaluation, all of which were significantly decreased (p < 0.001, one-sample Wilcoxon test). The measured values (median) were 60.0 min at baseline, 30.0 min at Week 8, 30.0 min at Week 16, 20.0 min at Week 24, and 30.0 min at the final evaluation.

Changes in sTST from baseline (median) were 60.0 min at Week 8, 80.0 min at Week 16, 90.0 min at Week 24, and 90.0 min at the final evaluation, all of which were significantly increased (p < 0.001, one-sample Wilcoxon test). The measured values (median) were 300.0 min at baseline, 390.0 min at Week 8, 390.0 min at Week 16, 405.0 min at Week 24, and 400.0 min at the final evaluation.

Median changes in sSE from baseline were 13.33% at Week 8, 13.54% at Week 16, 15.38% at Week 24, and 14.68% at the final evaluation, all of which were significantly increased (p < 0.001, one-sample Wilcoxon test). The measured values (median) were 66.67% at baseline, 87.50% at Week 8, 86.67% at Week 16, 87.50% at Week 24, and 87.50% at the final evaluation.

4 Discussion

In this study, 550 patients were enrolled, and 510 patients were included in the safety analysis set. There have been no other prospective observational studies of similar size that investigated the safety and efficacy of LEM in daily clinical practice since marketing approval, and this is the first report of its kind.

The most common reason for discontinuation of the study was ‘symptoms improved’ (16.08%), and the mean number of days of treatment until discontinuation due to improvement of symptoms was 68.2 days (Table 2). The ‘Clinical guideline for proper usage and tapering of hypnotics’ states that hypnotics should be reduced or withdrawn as soon as possible after remission of insomnia [1], but there is a risk of withdrawal symptoms or rebound insomnia when hypnotics are interrupted [10]. In Study 303, withdrawal symptoms and rebound insomnia were shown to be unlikely to occur even if LEM was discontinued without replacing it with placebo 6–12 months after administration [7, 11]. The results of this study also revealed that LEM can be safely discontinued after improvement of insomnia in clinical practice in Japan. In the postmarketing observational study of suvorexant, 17.1% of the patients discontinued the treatment due to improvement, and the mean number of days of treatment in the patients who had discontinued the treatment due to improvement was 61.7 days [12], which was similar to the results of this study.

Results regarding safety in daily clinical practice in this study showed that the incidences of all ADRs targeted for collection were similar to or slightly lower than those in the clinical study before approval (884 patients included in the safety analysis set) (Table 3) [9]. The incidence of ‘somnolence’ was 7.65% and 10.75% in this study and the clinical study before approval (Study 303), respectively. The incidences of ‘nightmare’ were 1.76% and 1.36%, ‘abnormal dreams’ were 0.59% and 1.81%, and ‘sleep paralysis’ were 0.20% and 1.58%, respectively. No ADRs falling under suicidal ideation or behavior occurred in this study or in the clinical study before approval. In addition, the mechanism of action of LEM suggests that symptoms such as sleep attacks, cataplexy, and hypnagogic hallucination seen in narcoleptic patients may occur. Although ‘hypnagogic hallucination’ was observed in 0.68% (6/884) of patients in the clinical study before approval (Study 303), these ADRs did not occur. In Study 304, reduced REM latency was observed [13], but the only ADR of REM sleep-related symptoms, such as sleep paralysis and hypnagogic hallucination, was sleep paralysis in one patient (0.20%). In the examination of factors affecting the occurrence of somnolence, the presence of concomitant insomnia drugs was significantly related (Table 5). These results indicate that the safety of LEM in actual clinical practice is not significantly different from the safety profile in the clinical study.

Regarding efficacy, the overall improvement based on the investigator's evaluation was ‘improved’ in 80.83%, ‘unchanged’ in 16.1%, and ‘worsened’ in 3.1%. On the other hand, the final overall improvement of suvorexant as evaluated by investigators was ‘improved’ in 74.0%, ‘unchanged’ in 22.2%, and ‘worsened’ in 3.7% [12]. A meta-analysis including LEM and suvorexant also reported that both drugs significantly improved sleep latency and total sleep time, etc., at Week 4 compared with the placebo group [14]. These results suggest that LEM also shows at least the same efficacy as suvorexant in actual clinical practice in Japan.

When the efficacy rates were examined by patient background, the efficacy rates in patients with ‘difficulty falling asleep’, ‘difficulty staying asleep’, or ‘problems awakening up too early in the morning’ were all approximately 80%, showing certain effects (Table 6). The efficacy for ‘difficulty falling asleep’ was also confirmed by improvement in sSOL. Improvements in sSE and sTST have also been observed, suggesting that LEM also contributes to improvements in difficulty maintaining sleep. The efficacy rate was around 80% for insomnia both with and without medical or psychiatric comorbidities, and there was no marked decrease in efficacy due to the presence/absence of underlying disease (Table 6). Study 303 suggested the efficacy of LEM for insomnia with medical or psychiatric comorbidities [7], and the results of this study also supported this finding. Regarding efficacy rate by comorbidity, the efficacy rate was almost the same in the presence/absence of each disease, and there were no comorbidities that markedly decreased the efficacy rate (Table 6). However, for ‘disease duration’, the efficacy rate decreased in patients with a longer period of disease, likely because insomnia became chronic and refractory.

In the examination of factors affecting overall improvement, the presence of concomitant therapies and ‘add-on’ rather than ‘naïve’ were significantly associated (Table 7). Among concomitant therapies, cognitive behavioral therapy in particular has been shown to be effective as a concomitant therapy with medications [1], and the results were considered to be reflected. In addition, more ‘add-on’ patients had longer disease duration than ‘naïve’ and ‘switched’ patients and were also taking LEM in combination with other insomnia drugs. Therefore, it was considered possible that they were refractory patients.

Regarding the effects of LEM on sleep parameters, sSOL, sTST, and sSE were all significantly improved at Week 8 compared with those at baseline (Fig. 3).

As a limitation of this study, it cannot be denied that bias may have occurred in the enrolled patients due to the investigators’ selection of patients who had no safety problems and were expected to respond to the drug. Since this was an observational study without a control group, no statistical analysis was performed. In addition, biases such as patient background and the effects of other drugs used in treatment may have occurred. Since no data have been collected after the end of the observation period for LEM, additional verification is necessary for the prognosis, including recurrence of symptoms after discontinuation of LEM.

5 Conclusion

Based on the results of this study, the safety of LEM was not different from the safety profile described in the current package insert in actual clinical practice in Japan. A certain level of clinical usefulness was also observed for efficacy.