Distinct Effects of Orexin Receptor Antagonist and GABAA Agonist on Sleep and Physical/Cognitive Functions After Forced Awakening

Home , Brotizolam, GABA receptor agonist, Rilmazafone, Suvorexant

Distinct effects of orexin receptor antagonist and GABAA agonist on sleep and physical/cognitive functions after forced awakening

Jaehoon Seola,b,1, Yuya Fujiib,1, Insung Parka, Yoko Suzukia, Fusae Kawanaa, Katsuhiko Yajimac, Shoji Fukusumia, Tomohiro Okurad, Makoto Satoha, Kumpei Tokuyamaa, Toshio Kokuboa, and Masashi Yanagisawaa,e,f,g,2

aInternational Institute for Integrative Sleep Medicine, University of Tsukuba, 305-8575 Tsukuba, Ibaraki, Japan; bPhysical Education, Health, and Sport Sciences, University of Tsukuba, 305-8574 Tsukuba, Ibaraki, Japan; cFaculty of Health and Nutrition, Tokyo Seiei College, 124-8530 Tokyo, Japan; dFaculty of Health and Sport Sciences, University of Tsukuba, 305-8574 Tsukuba, Ibaraki, Japan; eLife Science Center for Survival Dynamics (TARA), University of Tsukuba, Ibaraki, 305-8577, Japan; fR&D Center for Frontiers of Mirai in Policy and Technology (F-MIRAI), University of Tsukuba, Ibaraki, 305-8575, Japan; and gDepartment of Molecular Genetics, University of Texas Southwestern Medical Center, Dallas, TX 75390

Contributed by Masashi Yanagisawa, September 14, 2019 (sent for review April 29, 2019; reviewed by Seiji Nishino and Kenneth P. Wright, Jr.) The majority of patients with insomnia are treated with hypnotic Suvorexant is a recently approved orexin receptor antagonist agents. In the present study, we evaluated the side-effect profile of representing an alternative mechanistic approach for the treat- γ an orexin receptor antagonist and -aminobutyric acid A (GABAA)re- ment of insomnia, which specifically inhibits the orexin-mediated ceptor agonist on physical/cognitive functions upon forced awak- wake-promoting system (12). Unlike GABAA agonists, orexin ening. This double-blind, randomized, placebo-controlled, cross-over receptor antagonists promote both NREM and rapid eye move- study was conducted on 30 healthy male subjects. Fifteen mi- ment (REM) sleep, do not disrupt sleep-stage-specific quantitative nutes before bedtime, the subjects took a pill of suvorexant (20 mg), EEG spectral profiles, and allow somnolence indistinct from brotizolam (0.25 mg), or placebo and were forced awake 90 min there- normal sleep (12). In animal studies, sleep induced by an orexin after. Physical- and cognitive-function tests were performed be- receptor antagonist did not impair the ability to selectively re- fore taking the pill, after forced awakening, and the next morning. spond and arouse to emotionally salient auditory stimuli, similar to Polysomnographic recordings revealed that the efficacies of the hyp-

natural sleep (13, 14). Interestingly, after awakening, the animals PHARMACOLOGY notic agents in prolonging total sleep time (∼30 min) and increasing sleep efficiency (∼6%) were comparable. When the subjects were returned to sleep in a dose-proportional manner. allowed to go back to sleep after the forced awakening, the sleep The potential side effects of GABAA agonists include impaired latency was shorter under the influence of hypnotic agents (∼2min) cognitive and physical functions upon awakening from sleep (15, compared to the placebo trial (24 min), and the rapid eye movement 16). Animal studies have suggested distinct effects of GABAA ag- latency was significantly shorter under suvorexant (98.8, 81.7, and 48.8 onists and orexin receptor antagonists on motor and cognitive min for placebo, brotizolam, and suvorexant, respectively). Although functions (17–19). Motor coordination assessed 30 min after the brotizolam significantly impaired the overall physical/cognitive perfor- administration of hypnotic drugs was impaired by the GABAA mance (sum of z score) compared with placebo upon forced awak- agonists zolpidem, eszopiclone, and diazepam in a dose-dependent ening, there was no significant difference in the total z score of performance between suvorexant and placebo. Notably, the score Significance for static balance with the eyes open was higher under suvorexant compared to brotizolam administration. The energy expenditure Insomnia is a common symptom representing an important was lower under suvorexant and brotizolam compared with the health burden. Widely prescribed hypnotic agents enhance the d = placebo. The effect size of brotizolam ( 0.24) to reduce the function of γ-aminobutyric acid (GABA), a major inhibitory d < energy expenditure was larger than that of suvorexant ( 0.01). neurotransmitter. The ability to arouse and respond to unexpected stimuli is a feature of normal sleep, and one of the hypnotics | benzodiazepines | Stroop color-word test | Purdue pegboard concerns of this class of hypnotic agents is that patients may test | body sway become physically and/or cognitively impaired while the drug is in effect. As a new approach for the treatment of insomnia, nsomnia is a common symptom in the general population, and orexin receptor antagonists have been recently approved, Ivarious studies worldwide have shown its prevalence in 10 to which specifically inhibit the orexin-mediated wake-promoting 30% of the population, some even as high as 50 to 60% (1–4). system, supposedly without affecting the whole brain. We Chronic insomnia occurs in 9 to 33% of the adult population (5, found that, compared with the GABA receptor agonist brotizolam, 6), and nearly half (45.6%) of the patients with insomnia take the orexin receptor antagonist suvorexant induced less impair- hypnotic agents (7). Currently, the most commonly prescribed ment in body balance after taking the medicine. hypnotic agents are γ-aminobutyric acid (GABA) agonists, which Author contributions: S.F., T.K., and M.Y. designed research; J.S., Y.F., I.P., Y.S., F.K., T.O., enhance the function of the major inhibitory neurotransmitter M.S., and K.T. performed research; J.S., Y.F., I.P., Y.S., F.K., K.Y., T.O., and K.T. analyzed data; and J.S., Y.F., K.T., and M.Y. wrote the paper. GABA by binding to the allosteric benzodiazepine site on GABAA receptors. Electroencephalographic (EEG) power spectral analyses Reviewers: S.N., Stanford University School of Medicine; and K.P.W., University of have revealed a decrease in low-frequency (0.25 to 7.02 Hz) activity Colorado Boulder. andanincreaseinhigh-frequencyactivity(14.04to21.84Hz)in The authors declare no competing interest. non-rapid eye movement (NREM) sleep with the administration of Published under the PNAS license. 1J.S. and Y.F. contributed equally to this work. these GABAA receptor agonists (8, 9). As a consequence of the 2To whom correspondence may be addressed. Email: yanagisawa.masa.fu@u. widespread expression of GABAA receptors in the central nervous tsukuba.ac.jp. system, GABAA receptor agonists can inhibit neurons throughout This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. the brain and spinal cord, including those not involved in the in- 1073/pnas.1907354116/-/DCSupplemental. duction and maintenance of sleep (10, 11). First published November 11, 2019.

www.pnas.org/cgi/doi/10.1073/pnas.1907354116 PNAS | November 26, 2019 | vol. 116 | no. 48 | 24353–24358 Downloaded by guest on October 1, 2021 manner, but not following the administration of the orexin receptor N3 N2 N1 R W antagonist DORA-22 (17). However, the effect of orexin receptor 100% antagonists on human cognitive and physical functions upon tnaxerovuS 80% awakening from sleep remains untested. In addition, orexin re- 60% ceptor antagonists may modulate energy metabolism during sleep 40% (20), as orexin infusion into the hypothalamus increases the oxy- 20% Post-test gen consumption of freely behaving and anesthetized rats (21, 22). 0% 100% Moreover, orexin receptor antagonists and GABAA agonists could 80%

modify energy metabolism during sleeping through their effects on malozitorB the sleep architecture, as the sleeping energy expenditure is re- 60% lated to the sleep stages (23). The primary objective of the present 40% study was to compare the pharmacological effects of an orexin 20% Post-test receptor antagonist and a GABA agonist on physical and cog- 0% A 100% nitive functions upon forced awakening when the drug effect was 80%

maximal. We also assessed the effects of hypnotic agents on the obecalP metabolic rate while sleeping. 60% 40% Results 20% Post-test Subject Characteristics. The characteristics of the subjects are 0% presented in SI Appendix, Table S1. All subjects fulfilled the inclusion/exclusion criteria (SI Appendix, SI Materials and Methods). The time in bed, total sleep time, sleep latency, and sleep Fig. 1. Cumulative display of sleep architecture in all 30 subjects. The per- efficiency of the adaptation night were 515 min, 460.4 ± 46.0 centage of subjects in each sleep stage is shown. The subjects were forced min, 6.0 ± 7.5 min, and 89.4 ± 8.9%, respectively. Due to some awake at 00:15 and went back to sleep at 00:40. technical problems, there were missing data for 1 Stroop color-word test, 1 eyes-opened body-sway test, and 2 indirect calorimetry measurements. by brotizolam. A substantial decrease in wake was observed under both hypnotic agents during the 1st quarter of sleep. An Sleep Parameters. The total sleep time during the total time in increase in stage R by suvorexant was also observed during the bed was increased by brotizolam and suvorexant compared with 3rd quarter, and an increase in stage N2 by brotizolam was ob- the placebo (Table 1). Brotizolam and suvorexant decreased the served during the 2nd quarter of sleep time (SI Appendix, Table awake time and increased sleep efficiency. Suvorexant increased S5). When the subjects were allowed to go back to sleep after stage REM (R) compared with brotizolam and placebo. Brotizolam forced awakening, suvorexant significantly decreased the REM increased stage NREM 2 (N2) compared with suvorexant and latency and prolonged stage R compared with the other 2 trials. placebo (Fig. 1 and Table 1). In 40% of the subjects, stage R occurred with a short REM la- When the sleep architecture was separately analyzed before and tency (<15 min) under the influence of suvorexant (SI Appendix, after forced awakening, there were no statistically significant dif- Table S4 and Fig. S1 and Fig. 1). ferences during the 75-min period before forced awakening (SI Appendix, Table S2). The sleep stages immediately before the Physical and Cognitive Functions. forced awakening were comparable among the 3 trials (SI Ap- Total z score. The sum of z scores (24) from all physical- and pendix, Table S3); this was important because the sleep stage from cognitive-function tests performed before bedtime (pretest), which the subject is forced awake affects subsequent performance. after forced awakening (posttest), and on the next morning The sleep latency after forced awakening was significantly (follow-test) is shown in Table 2. The time effect and the in- longer compared with that before forced awakening in placebo teraction of group and time effects were significant. Compared trials (SI Appendix, Tables S2 and S4 and Fig. 1; P = 0.026). with the pretest, the posttest total z score was significantly de- However, in the brotizolam and suvorexant trials, the sleep la- creased in all trials, including placebo. The total posttest z score tencies after forced awakening were significantly shorter than was lower than that of the follow-test (P = 0.033) under the in- that before forced awakening (SI Appendix, Tables S2 and S4 and fluence of brotizolam. The total posttest z score under brotizolam Fig. 1; P = 0.017 and 0.015, respectively). was also lower compared to the placebo (P = 0.001), whereas the The comparison of sleep architecture showed that the effects difference between suvorexant and placebo was not significant of hypnotic compounds were manifested during the 1st quarter (P = 0.264). This was also reflected in the fact that, although there of sleep time after forced awakening (posttest), with increases in were no significant posttest differences in the sum of z scores stages N3 and R by suvorexant and increases in stages N2 and N3 between suvorexant and brotizolam (P = 0.167), the effect size of

Table 1. Total sleep time during total time in bed (23:00 to 00:15 and 00:40 to 08:00) Sleep parameters Placebo, mean (SD) Brotizolam, mean (SD) Suvorexant, mean (SD) ANOVA, P Post hoc

Total sleep time, min 459.7 (46.6) 492.7 (15.9) 488.7 (19.4) <0.001 S, B > P Sleep latency, min 5.8 (6.6) 5.2 (7.1) 6.8 (8.8) 0.369 — WASO, min 49.2 (8.5) 16.7 (2.5) 19.2 (2.9) <0.001 S, B < P Sleep efficiency, % 89.3 (9.1) 95.7 (3.1) 94.9 (3.8) <0.001 S, B > P Stage R, min 89.7 (21.0) 92.3 (20.0) 105.3 (16.6) <0.001 S > B, P Stage N1, min 44.2 (14.1) 38.2 (17.9) 41.7 (19.2) 0.146 — Stage N2, min 244.9 (44.4) 277.7 (39.7) 257.2 (32.7) <0.001 B > S, P Stage N3, min 81.0 (27.8) 84.6 (28.9) 84.6 (30.1) 0.473 —

B, brotizolam; P, placebo; S, suvorexant. n = 29. Sleep latency is for the initial sleep onset

24354 | www.pnas.org/cgi/doi/10.1073/pnas.1907354116 Seol et al. Downloaded by guest on October 1, 2021 Table 2. Comparison of physical- and cognitive-functions test by trial Physical and cognitive function Placebo Mean (SD) Brotizolam Mean (SD) Suvorexant Mean (SD) Interaction P Group P Time P

Total z-score Pre 0.28 (0.17)* 0.36 (0.15)* 0.33 (0.13)* † † ‡ Post −0.03 (0.18)*, −0.61 (0.20)*, , −0.25 (0.19)* Follow 0.15 (0.20) 0.13 (0.19)‡ 0.17 (0.16) Comparison <0.001 0.136 <0.001 Cognitive function Stroop color-word test, ms Neutral tasks Pre 646.8 (122.2) 638.1 (128.6) 627.2 (109.7) Post 711.8 (131.9) 752.7 (152.5) 740.5 (175.1) Follow 650.4 (137.7) 651.7 (120.6) 651.6 (125.6) Comparison 0.201 0.693 <0.001 Incongruent tasks Pre 728.6 (142.5) 727.8 (155.6)* 712.7 (128.1)* Post 774.2 (162.1)‡ 845.1 (181.9)*,‡ 820.4 (212.4)*,‡ Follow 713.0 (155.9)‡ 718.6 (120.1)‡ 732.3 (170.1)‡ Comparison 0.031 0.323 <0.001 Stroop interference Pre 81.8 (46.3) 89.7 (59.0) 85.6 (38.8) Post 62.4 (72.0) 92.4 (73.4) 79.9 (78.5) Follow 62.6 (41.6) 66.9 (61.0) 80.7 (66.9) Comparison 0.396 0.139 0.072 Physical functions Purdue pegboard test, count Pre 40.7 (3.9) 40.5 (3.7) 40.3 (3.5)

Post 38.6 (3.9) 36.4 (4.1) 36.7 (3.2) PHARMACOLOGY Follow 39.6 (4.0) 39.0 (3.2) 38.9 (3.1) Comparison 0.083 <0.001 0.028 Body sway: trajectory length of center of foot pressure, cm Eyes open Pre 268.5 (63.0) 253.6 (56.8) 259.8 (64.3) Post 250.1 (55.5) 271.3 (78.6) 253.5 (70.0) Follow 268.9 (69.9) 262.6 (66.8) 265.4 (73.6) Comparison 0.124 0.776 0.516 Eyes closed Pre 341.1 (110.8) 329.3 (77.0) 338.3 (77.8) Post 325.1 (88.9) 329.3 (92.4) 328.5 (104.9) Follow 341.9 (102.2) 341.8 (116.1) 337.7 (92.5) Comparison 0.947 0.964 0.376 Body sway: rectangular area, cm2 Eyes open Pre 2.5 (1.7) 2.7 (1.7)* 2.3 (1.3) † † ‡ Post 2.6 (1.5) 6.0 (5.8)*, , ,§ 2.7 (1.5)§ ‡ Follow 3.0 (3.7) 3.1 (2.1) 2.5 (1.6) Comparison 0.001 <0.001 0.002 Eyes closed Pre 3.5 (3.3) 2.8 (1.4) 3.4 (2.2) Post 4.2 (2.7) 6.4 (4.7) 4.7 (4.1) Follow 3.9 (3.3) 4.4 (3.8) 3.4 (1.8) Comparison 0.074 0.082 <0.001 Agility and dynamic balance test, s Pre 15.2 (1.7) 15.1 (1.7)* 15.2 (1.5)* Post 15.5 (2.0)‡ 15.8 (1.9)*,‡ 15.7 (1.8)*,‡ ‡ ‡ ‡ Follow 15.1 (1.8) 14.9 (1.6) 15.1 (1.6) Comparison 0.042 0.815 <0.001 Choice stepping reaction time, ms Pre 803.0 (101.4) 790.6 (85.6) 799.0 (85.3) Post 838.0 (114.0) 856.5 (97.6) 835.3 (96.9) Follow 815.1 (105.4) 815.1 (98.5) 811.4 (97.7) Comparison 0.054 0.838 <0.001

Pre, pretest at before bedtime; Post, posttest at forced-awakening; Follow, follow-test at after waking up in morning. *Significant difference between pre and post. † Significant differences between brotizolam and placebo. ‡ Significant difference between post and follow. §Significant differences between brotizolam and suvorexant (P < 0.05); there are no post hoc differences between suvorexant and placebo.

Seol et al. PNAS | November 26, 2019 | vol. 116 | no. 48 | 24355 Downloaded by guest on October 1, 2021 suvorexant was smaller than that of brotizolam (SI Appendix, first 75 min, the sleep architecture was comparable among the 3 Table S6). There were significant order effects observed in the trials. After the forced awakening, all subjects maintained a total z score (P = 0.001) (i.e., the total z scores were improved wakeful state to complete the tests. When the subjects were as the subjects repeated the trials). allowed to go back to sleep, the sleep latency of the placebo Stroop color-word test. A time effect was observed on the neutral trial was prolonged (P = 0.026), presumably because of the task score, which deteriorated in the posttest condition. For the strong psychological arousal induced by the physical- and incongruent task, there were significant time effects and interac- cognitive-assessment tasks (27, 28). In contrast, the sleep latency tions. The score of the incongruent task deteriorated in the under suvorexant (P = 0.015) and brotizolam (P = 0.017) was posttest condition with brotizolam and suvorexant administration. shortened compared with the first sleep latency of the night. The In Stroop interference, there was no significant effect of group, effects of the hypnotic drugs were manifested until late at night time, or interaction. (SI Appendix, Table S5), and both hypnotics were similarly ef- Purdue pegboard test. There were significant group and time ef- fective in increasing sleep efficiency after forced awakening. fects. The scores deteriorated in the posttest condition. In the The observed order effect on the total z score underscores the follow-test condition, the scores returned to levels similar to earnest attempt of the subjects at each trial. In order to offset the those of the pretest condition in the 3 trials. The Purdue pegboard order effect, we adopted a randomized cross-over study design. test score was lower with brotizolam and suvorexant administra- The poor performance after forced awakening may have been tion compared with the placebo. partly due to the influence of sleep inertia, the transitional state Body sway. The rectangular area of the center of foot pressure of lowered arousal occurring immediately after awakening from during 30 s of test with the eyes opened showed significant ef- sleep. Indeed, the total z score of the posttest condition was fects of time, group, and interaction. With the eyes closed, there significantly decreased, even with placebo (Table 2). As signifi- were significant effects of time, but there were no statistically cant differences in the sleep architecture were observed imme- significant effects of group or interaction. When the center of diately after the subjects were allowed to go back to sleep, it was foot-pressure trajectory length was evaluated during the test, assumed that the physical- and cognitive-function tasks were there were no significant effects of time, group, or interaction performed under the influence of the hypnotic agents. Com- z (Table 2 and SI Appendix, Fig. S2). pared with the placebo, the sum of scores was further decreased z Agility and dynamic balance test. Significant effects of time and in- at the posttest condition by brotizolam. The sum of scores teraction were observed. After forced awakening, agility and under suvorexant was not significantly different from that of dynamic balance significantly deteriorated with brotizolam and the placebo or brotizolam (Table 2). When the effect size was evaluated, it was found to be smaller under suvorexant than suvorexant administration, but not placebo. SI Appendix Choice stepping reaction time. There was a significant time effect. under brotizolam ( , Table S6). The reaction deteriorated in the posttest condition, but returned The distinct effect of suvorexant and brotizolam on body sway to a level similar to the pretest results. was observed when it was evaluated as the circumscribed rectangular area of foot pressure trajectory with the eyes open (SI Appendix Energy Metabolism. The energy expenditure showed significant , Fig. S2). In theory, the outer circumference of the area effects of time (P = 0.001), group (P = 0.005), and interaction of the center of foot pressure during the sway test is determined (P = 0.001). The post hoc test revealed significant differences by big sways. Compared with trajectory length, the sway area of among average energy expenditure during the 20 min from 00:40 the center of foot pressure is more sensitive to alcohol con- to 01:00 and subsequent hourly averages (Fig. 2). The cumulative sumption (29). Because visual information is essential for energy expenditure over the entire sleeping period was smaller maintaining body balance (30), the level of difficulty increased with brotizolam (532 ± 53 kcal during sleep) than with suvorexant when the same balance task was performed with the eyes closed, ± P = ± and this might mask the differences in body sway among the (544 52 kcal during sleep, 0.027) and placebo (544 49 kcal SI Appendix during sleep, P = 0.013) administration. experimental trials ( , Fig. S2). Suvorexant had little effect on body sway, evaluated as the rectangular area of the Discussion center of pressure with the eyes open, suggesting a lower risk of – In the present study, to enhance the assay sensitivity, forced falls (15, 16, 31 34). awakening was targeted when the plasma concentration of hyp- Our present findings were consistent with previous animal notic drugs presumably peaked 90 min after suvorexant ingestion studies comparing the effects of orexin receptor antagonists and GABA agonists on the time spent on an accelerating rotating rod (25). Among the GABA agonists currently prescribed, brotizolam A A (rotarod). The rotarod performance was impaired by the GABA was selected as a positive control because its plasma concen- A agonists zolpidem, eszopiclone, and diazepam, but not by the trationpeaked60to120minafteringestion(26).Duringthe orexin receptor antagonist DORA-22 (17, 18). The differential effects of suvorexant and brotizolam are likely related to the fundamental differences in their mechanisms of action. Balance is (kcal/min) Placebo Brotizolam Suvorexant coordinated by the cerebellum (35), in which benzodiazepine re- 1.30 #¶ ceptors are abundantly expressed (36). In contrast, the cerebellum erutidnepxe is free from the direct effects of orexin; orexin-producing neurons 1.20 ¶ y ¶ grenE ˒ essentially send no projections to the cerebellum (37), where very 1.10 ˒ little expression of orexin receptors has been documented (38). 1.00 Post-test The effect size on the Stroop incongruent task and the agility and 0.90 dynamic balance test under suvorexant were smaller compared to brotizolam (SI Appendix, Table S6). However, the performance of the Stroop incongruent task and agility and dynamic balance test was impaired by both suvorexant and brotizolam at the posttest Fig. 2. Time course of energy expenditure. The hourly average, average of 15 min between 00:00 and 00:15, and average of 20 min between 00:40 and compared with the pretest condition. Thus, further clinical studies 01:00 are shown with the SEM. Physical- and cognitive-function tests were on incidents of falling under suvorexant are warranted. performed outside the metabolic chamber. §Significant differences between The suppressive effect of hypnotic agents on energy expenditure suvorexant and brotizolam; #significant differences between suvorexant and was detected in the present study, although indirect calorimetry placebo; {significant differences between brotizolam and placebo. was interrupted for 25 min when the plasma concentration of

24356 | www.pnas.org/cgi/doi/10.1073/pnas.1907354116 Seol et al. Downloaded by guest on October 1, 2021 hypnotic compounds presumably peaked. The reduction of energy their safety. Thus, further study with patients with insomnia, expenditure might be due to the drug’s direct effects on energy particularly elderly subjects, is necessary. metabolism (39). Alternatively, hypnotic agents might suppress the energy expenditure by inducing sleep. Applying a semiparametric Materials and Methods regression analysis on the time course of energy expenditure after Subjects. We used advertisements to recruit 30 healthy male subjects. The forced awakening, the effect of hypnotic agents was decomposed study concepts were explained to all of the subjects, and they all provided into the effect through changes in sleep stages and the effect in- signed informed consent. This study was approved by the Ethics Committee of dependent of sleep stages (23). However, the relative importance the University of Tsukuba (reference no. Tai 27-143). The study protocol of these 2 factors remained inconclusive (SI Appendix, Figs. S3 and (UMIN000022752) was registered with the University Hospital Medical In- formation Network Center (https://www.umin.ac.jp/english/). S4). Additionally, the energy metabolism measured after the subjects reentered the metabolic chamber might be affected by the Study Design and Procedure. This double-blind, randomized, placebo- residual effect of physical activity during the test, known as excess controlled, 3-way cross-over study was conducted from July 2016 to Janu- postexercise oxygen consumption (40). Further study is warranted ary 2018. The washout period was more than 2 d, and the 3 trials were to assess the effects of hypnotic agents on energy metabolism completed within 8 wk. The experiment was preceded by an adaptation during the entire sleeping period. night in a metabolic chamber, during which the sensors and electrodes of a Consistent with previous studies (41–43), the effects of suvorexant polysomnographic recording system were attached to the subjects. Prior to and brotizolam on the sleep architecture were different. When the physical- and cognitive-function tests, the subjects ate dinner 5 h before the subjects were allowed to go back to sleep under the influence bedtime, and the electrodes were attached. A series of physical- and of suvorexant, stage R with a short REM latency was observed cognitive-function tests before sleep (pretest) were performed 3.5 h before in 12 of 30 subjects (with a mean latency of 4.7 min in the 12 bedtime. Subsequently, the subjects entered the metabolic chamber and assumed a sitting position. Then, they took a pill of either suvorexant (20 mg), subjects). This was substantially more frequent than the sleep- brotizolam (0.25 mg), or placebo 15 min before bedtime. Brotizolam was onset REM periods (SOREMPs) observed in clinical trials of selected as a positive control to demonstrate the assay sensitivity versus suvorexant; in 1 such study (41), SOREMPs were observed in placebo, as in other studies (51, 52). At 23:00, polysomnography recording 4.1% of subjects taking 20 mg (15 mg for elderly subjects) of commenced with the lights out. At 00:15, 90 min after taking the pill, the suvorexant at bedtime compared with 1.0% of subjects taking a subjects were forced awake to repeat the same physical- and cognitive- placebo. This was likely due to the fact that, in the present study, function tests (posttest). Thereafter, they returned to the metabolic cham- the subjects were allowed to go back to sleep when suvorexant ber and were allowed to sleep from 00:40 to 08:00. At 08:30, after main- was already exerting maximal effects. The subsequent peak in the taining a sitting position for 30 min, the subjects repeated the physical- and percentage of subjects in REM sleep was observed at ∼190 min cognitive-function tests (follow-test) (Fig. 3). PHARMACOLOGY Measurements. after the initial bedtime in the trial with suvorexant (Fig. 1). In Indirect calorimetry using a whole-room metabolic chamber. The subjects slept the placebo and brotizolam trials, the peak increase in the per- ∼ in a whole-room indirect calorimeter (Fuji Medical Science) (23), and the centage of subjects in REM sleep was also observed at 190 min energy equivalence of the measured O2 uptake and CO2 production was after the first bedtime, without the initial R stage shortly after calculated according to Weir’s equation (53). The effects of sleep stages and going back to bed. We speculated that the NREM–REM cycle time after going back to bed on the energy expenditure were analyzed by might be discontinued and reset by sleep interruption as pre- semiparametric regression analysis, as described (23). viously observed (44, 45) in the placebo and brotizolam trials, Polysomnography. Sleep was recorded polysomnographically by using a PSG- whereas suvorexant might have partially suppressed this resetting 1100 system (Nihon Kohden) (23). The records were scored every 30 s to of the NREM–REM cycle. stages wakefulness (W), N1, N2, N3, and R. Measurements during sleep- The ability to arouse and respond to unexpected environmental onset latency were classified as stage W. Wake after sleep onset (WASO) was defined according to standard criteria (54). stimuli is a feature of normal sleep, which is crucial when people Physical- and cognitive-assessment tasks. Physical- and cognitive-assessment are faced with urgent situations. There is a general concern that tasks were performed in the following order: Stroop color-word, Purdue hypnotic agents may impair physical and cognitive functions, pegboard test, body sway, agility and dynamic balance, and choice stepping eliciting muscle atonia, ataxia, loss of balance, retrospective am- reaction time test. nesia, attention deficits, and slower response time, and patients might become temporarily incapacitated and unable to appropri- Statistical Analyses. We used the sum of z scores for all of the physical- and ately respond under the effect of hypnotic agents. When patients cognitive-assessment tasks. Each score of the physical- and cognitive- need to be awake under the influence of a hypnotic agent, the function tasks was analyzed by using 2-way ANOVA with repeated mea- impairment of physical and cognitive functions might manifest as a sures and post hoc tests with Bonferroni’s correction for changes at different fall or serious misjudgment. The use of GABA receptor agonists time points between treatments. In order to assess differences in sleep pa- A rameters and energy expenditure between treatments, 2-way ANOVA and has been associated with an increased risk of falling and traffic Bonferroni’s correction as post hoc tests were used. The comparison of the accidents (10, 46–49). In the present study, to maximize the assay sensitivity, physical and cognitive functions were assessed when the plasma concentration of the hypnotic agent peaked. However, this Forced- protocol may not be optimal with respect to real-world clinical Wake Sleep Sleep Wake significance, as falls and other incidents related to leaving the bed, awakening such as urination, occur at later hours (15). pre-test take a pill post-test follow-test Next-day sedation and cognitive dysfunction have been reported after the use of GABAA receptor agonists (50), but the z score in the follow-test condition returned to a level similar to 19:30 22:45 00:40 that of the pretest in all experimental trials in our present study. This lack of residual effects of hypnotic agents 9.75 h after dosing 18:00 23:00 00:15 08:00 suggested that the dose of the hypnotic agents was within a reasonable range. Consistently, it has been reported that, after single and repeated doses of suvorexant, its residual effects on PSG, energy expenditure recording recording driving performance in the morning were negligible (35, 36). As this was an initial study with forced awakening under the maxi- Fig. 3. Experimental protocol. Physiological- and cognitive-function tests mal influence of suvorexant, a new class of hypnotics, we en- were performed in the evening (19:30), after forced awakening (00:15), and rolled relatively young, healthy individuals in consideration of on the next morning (08:30). PSG, polysomnography.

Seol et al. PNAS | November 26, 2019 | vol. 116 | no. 48 | 24357 Downloaded by guest on October 1, 2021 sleep stage before forced awakening was performed by using χ2 statistics. on Innovative Areas Grant JP15H05942 (to M.Y.); the MEXT World Premier Statistical significance was established at P < 0.05 (2-tailed). All analyses International Research Center Initiative “Living in Space” (to M.Y.); Japan were carried out by using IBM SPSS statistics software (Version 25.0; IBM Society for the Promotion of Science (JSPS) KAKENHI Grant 17H06095 (to Corporation) for Windows. M.Y.); MEXT CREST Grant A3A28043 (to M.Y.); the JSPS Funding Program for World-Leading Innovative R&D on Science and Technology (M.Y.); a Uehara ACKNOWLEDGMENTS. We thank Naruki Kitano for hard work on the pilot Memorial Foundation research grant (to M.Y.); a Takeda Science Foundation study and Momoko Kayaba for helping keep polysomnography electrodes research grant (to M.Y.); and the Japan Sports Agency Sports Research In- attached. This work was supported by Ministry of Education, Culture, Sports, novation Project (K.T.). Also, J.S. is a recipient of a scholarship from the Science and Technology–Japan (MEXT) Grant-in-Aid for Scientific Research Otsuka Toshimi Scholarship Foundation.

1. S. Ancoli-Israel, T. Roth, Characteristics of insomnia in the United States: Results of the 29. M. Nieschalk et al., Effects of alcohol on body-sway patterns in human subjects. Int. J. 1991 National Sleep Foundation Survey. I. Sleep 22 (suppl. 2), S347–S353 (1999). Legal Med. 112, 253–260 (1999). 2. D. J. Taylor et al., Comorbidity of chronic insomnia with medical problems. Sleep 30, 30. O. Aoki, Y. Otani, S. Morishita, Effect of eye-object distance on body sway during 213–218 (2007). galvanic vestibular stimulation. Brain Sci. 8, E191 (2018). 3. D. J. Buysse et al., Prevalence, course, and comorbidity of insomnia and depression in 31. A. Vermeeren et al., On-the-road driving performance the morning after bedtime use young adults. Sleep 31, 473–480 (2008). of suvorexant 20 and 40 mg: A study in non-elderly healthy volunteers. Sleep 38, 4. S. Schutte-Rodin, L. Broch, D. Buysse, C. Dorsey, M. Sateia, Clinical guideline for the evalua- 1803–1813 (2015). tion and management of chronic insomnia in adults. J. Clin. Sleep Med. 4,487–504 (2008). 32. A. Vermeeren et al., On-the-road driving performance the morning after bedtime use 5. J. A. Costa e Silva, M. Chase, N. Sartorius, T. Roth, Special report from a symposium of suvorexant 15 and 30 mg in healthy elderly. Psychopharmacology (Berl.) 233, 3341– held by the World Health Organization and the World Federation of Sleep Research 3351 (2016). Societies: An overview of insomnias and related disorders–Recognition, epidemiol- 33. M. Nakamura, M. Ishii, Y. Niwa, M. Yamazaki, H. Ito, Temporal changes in postural ogy, and rational management. Sleep 19, 412–416 (1996). sway caused by ultrashort-acting hypnotics: Triazolam and zolpidem. ORL J. Oto- 6.S.Bhaskar,D.Hemavathy,S.Prasad,Prevalence of chronic insomnia in adult patients and rhinolaryngol. Relat. Spec. 67, 106–112 (2005). its correlation with medical comorbidities. J. Family Med. Prim. Care 5, 780–784 (2016). 34. C. L. Drake et al., Arousability and fall risk during forced awakenings from nocturnal 7. J. Scalo, P. Desai, K. Rascati, Insomnia, hypnotic use, and health-related quality of life sleep among healthy males following administration of zolpidem 10 mg and doxepin – in a nationally representative sample. Qual. Life Res. 24, 1223 1233 (2015). 6 mg: A randomized, placebo-controlled, four-way crossover trial. Sleep 40, zsz086 8. C. H. Bastien, M. LeBlanc, J. Carrier, C. M. Morin, Sleep EEG power spectra, insomnia, (2017). – and chronic use of benzodiazepines. Sleep 26, 313 317 (2003). 35. M. A. Watson, F. O. Black, M. Crowson, The human balance system—A complex co- 9. S. V. Fox et al., Quantitative electroencephalography within sleep/wake states dif- ordination of central and peripheral systems. Vestibular disorders association (2014). ferentiates GABAA modulators eszopiclone and zolpidem from dual orexin receptor https://vestibular.org/. Accessed 15 March 2019. – antagonists in rats. Neuropsychopharmacology 38, 2401 2408 (2013). 36. M. Hirouchi, H. Mizutani, Y. Kohno, K. Kuriyama, Characteristics of the association of – 10. D. J. Buysse, Insomnia. JAMA 309, 706 716 (2013). brotizolam, a thieno-triazolo diazepine derivative, with the benzodiazepine receptor: 11. U. Rudolph, F. Knoflach, Beyond classical benzodiazepines: Novel therapeutic po- A selective and high affinity ligand of the central type I benzodiazepine receptor. Jpn. tential of GABA receptor subtypes. Nat. Rev. Drug Discov. 10, 685–697 (2011). A J. Pharmacol. 59, 387–391 (1992). 12. P. J. Coleman, A. L. Gotter, W. J. Herring, C. J. Winrow, J. J. Renger, The discovery of 37. T. Sakurai, The neural circuit of orexin (hypocretin): Maintaining sleep and wake- suvorexant, the first orexin receptor drug for insomnia. Annu. Rev. Pharmacol. Toxicol. fulness. Nat. Rev. Neurosci. 8, 171–181 (2007). 57,509–533 (2017). 38. J. N. Marcus et al., Differential expression of orexin receptors 1 and 2 in the rat brain. 13. P. L. Tannenbaum et al., Orexin receptor antagonist-induced sleep does not impair J. Comp. Neurol. 435,6–25 (2001). the ability to wake in response to emotionally salient acoustic stimuli in dogs. Front. 39. O. Kirvelä, A. Katila, R. Aantaa, J. Kanto, J. J. Himberg, Metabolic and subjective re- Behav. Neurosci. 8, 182 (2014). sponses to oral diazepam and midazolam. Eur. J. Anaesthesiol. 11, 365–369 (1994). 14. P. L. Tannenbaum et al., Inhibition of orexin signaling promotes sleep yet preserves 40. E. Børsheim, R. Bahr, Effect of exercise intensity, duration and mode on post-exercise salient arousability in monkeys. Sleep 39, 603–612 (2016). oxygen consumption. Sports Med. 33, 1037–1060 (2003). 15. S. I. Uemura et al., Residual effects of zolpidem, triazolam, rilmazafone and placebo 41. E. Snyder et al., Effects of suvorexant on sleep architecture and power spectral profile in healthy elderly subjects: A randomized double-blind study. Sleep Med. 16, 1395– in patients with insomnia: Analysis of pooled phase 3 data. Sleep Med. 19,93–100 1402 (2015). (2016). 16. D. J. Frey, J. D. Ortega, C. Wiseman, C. T. Farley, K. P. Wright, Jr, Influence of zolpidem 42. C. D. Cox et al., Discovery of the dual orexin receptor antagonist [(7R)-4-(5-chloro-1,3- and sleep inertia on balance and cognition during nighttime awakening: A ran- benzoxazol-2-yl)-7-methyl-1,4-diazepan-1-yl][5-methyl-2-(2H-1,2,3-triazol-2-yl)phenyl]methanone domized placebo-controlled trial. J. Am. Geriatr. Soc. 59,73–81 (2011). (MK-4305) for the treatment of insomnia. J. Med. Chem. 53, 5320–5332 (2010). 17. A. D. Ramirez et al., Dual orexin receptor antagonists show distinct effects on loco- 43. C. J. Winrow et al., Promotion of sleep by suvorexant-a novel dual orexin receptor motor performance, ethanol interaction and sleep architecture relative to gamma- antagonist. J. Neurogenet. 25 ,52–61 (2011). aminobutyric acid-A receptor modulators. Front. Neurosci. 7, 254 (2013). 44. M. Grözinger, D. G. Beersma, J. Fell, J. Röschke, Is the nonREM-REM sleep cycle reset 18. M. A. Steiner, H. Lecourt, D. S. Strasser, C. Brisbare-Roch, F. Jenck, Differential effects by forced awakenings from REM sleep? Physiol. Behav. 77, 341–347 (2002). of the dual orexin receptor antagonist almorexant and the GABA(A)-α1 receptor modulator zolpidem, alone or combined with ethanol, on motor performance in the 45. A. Miyasita, K. Fukuda, M. Inugami, Effects of sleep interruption on REM-NREM cycle in – rat. Neuropsychopharmacology 36, 848–856 (2011). nocturnal human sleep. Electroencephalogr. Clin. Neurophysiol. 73,107116 (1989). 19. J. M. Uslaner et al., Orexin receptor antagonists differ from standard sleep drugs by 46. F. Barbone et al., Association of road-traffic accidents with benzodiazepine use. – promoting sleep at doses that do not disrupt cognition. Sci. Transl. Med. 5, 179ra44 (2013). Lancet 352, 1331 1336 (1998). 20. S. Nicolaidis, Metabolic mechanism of wakefulness (and hunger) and sleep (and sa- 47. P. Woratanarat et al., Alcohol, illicit and non-illicit psychoactive drug use and road – tiety): Role of adenosine triphosphate and hypocretin and other peptides. Metabo- traffic injury in Thailand: A case-control study. Accid. Anal. Prev. 41,651 657 lism 55 (suppl. 2), S24–S29 (2006). (2009). 21. K. Kiwaki, C. M. Kotz, C. Wang, L. Lanningham-Foster, J. A. Levine, Orexin A (hypocretin 1) 48. C. M. Chang et al., Psychotropic drugs and risk of motor vehicle accidents: A population- – injected into hypothalamic paraventricular nucleus and spontaneous physical activity based case-control study. Br. J. Clin. Pharmacol. 75,1125 1133 (2013). in rats. Am. J. Physiol. Endocrinol. Metab. 286, E551–E559 (2004). 49. P. S. Wang, R. L. Bohn, R. J. Glynn, H. Mogun, J. Avorn, Zolpidem use and hip fractures 22. J. Wang, T. Osaka, S. Inoue, Orexin-A-sensitive site for energy expenditure localized in in older people. J. Am. Geriatr. Soc. 49, 1685–1690 (2001). the arcuate nucleus of the hypothalamus. Brain Res. 971, 128–134 (2003). 50. M. Lader, Benzodiazepine harm: How can it be reduced? Br. J. Clin. Pharmacol. 77, 23. M. Kayaba et al., Energy metabolism differs between sleep stages and begins to in- 295–301 (2014). crease prior to awakening. Metabolism 69,14–23 (2017). 51. A. N. Nicholson, B. M. Stone, Midazolam: Sleep and performance studies in middle 24. E. A. Haapala et al., Associations of physical performance and adiposity with cogni- age. Br. J. Clin. Pharmacol. 16 (suppl. 1), 115S–119S (1983). tion in children. Med. Sci. Sports Exerc. 47, 2166–2174 (2015). 52. M. Mamelak, A. Csima, L. Buck, V. Price, A comparative study on the effects of brotizolam 25. Merck & Co, Inc., Belsomra [Package Insert] (Merck & Co., Inc., Whitehouse, NJ, 2014). and flurazepam on sleep and performance in the elderly. J. Clin. Psychopharmacol. 26. M. S. Langley, S. P. Clissold, Brotizolam. A review of its pharmacodynamic and 9, 260–267 (1989). pharmacokinetic properties, and therapeutic efficacy as an hypnotic. Drugs 35, 104– 53. J. B. Weir, New methods for calculating metabolic rate with special reference to 122 (1988). protein metabolism. J. Physiol. 109,1–9 (1949). 27. N. Goel, T. Abe, M. E. Braun, D. F. Dinges, Cognitive workload and sleep restriction 54. R. B. Berry et al., The AASM Manual for the Scoring of Sleep and Associated Events: interact to influence sleep homeostatic responses. Sleep 37, 1745–1756 (2014). Rules, Terminology and Technical Specifications (Version 2.4, American Academy of 28. A. Suzuki, C. M. Sinton, R. W. Greene, M. Yanagisawa, Behavioral and biochemical Sleep Medicine, Darien, IL, 2017). dissociation of arousal and homeostatic sleep need influenced by prior wakeful ex- 55. J. Cohen, Statistical Power Analysis for the Behavioral Sciences (Routledge Academic, perience in mice. Proc. Natl. Acad. Sci. U.S.A. 110, 10288–10293 (2013). New York, NY, 1988).

24358 | www.pnas.org/cgi/doi/10.1073/pnas.1907354116 Seol et al. Downloaded by guest on October 1, 2021