Nearly half of people with Alzheimer’s can’t sleep at night, waking up more often and staying awake longer than others their age. This causes nighttime wandering, and worsens cognition and depression. Despite the burden for both patient and caregiver, many physicians do not prescribe available hypnotics, sedatives, or antipsychotics for fear of causing falls, stroke, or hastening cognitive decline. Now, results of a Phase 3 trial testing Merck’s insomnia drug suvorexant, sold as Belsomra, in patients with mild to moderate AD, suggest that this drug prolongs slumber, with more overall sleep time and shorter bouts of nighttime wakefulness than placebo. The drug is FDA-approved for insomnia and has been reported safe in the elderly population. Merck scientists presented this data at the American Academy of Neurology’s 71st Annual Meeting, held May 4–10 in Philadelphia.

  • Many Alzheimer’s patients have serious problems sleeping.
  • Four weeks of suvorexant nudged up nighttime slumber by a half hour.
  • Results may encourage physicians to prescribe this orexin inhibitor.

“Clinicians often say they are hesitant to try suvorexant in AD patients because of insufficient data to support the efficacy and safety in this key subpopulation,” W. Joseph Herring, Merck, Kenilworth, New Jersey, who presented the results, told Alzforum. “We’ve shown that suvorexant was effective in improving sleep and generally well tolerated in patients,” he said.

“They did a really nice study,” said David Holtzman, Washington University in St. Louis. “It clearly shows the drug caused more sleep, and there was a pretty strong effect. I think it’s worth trying in some of our dementia patients with sleep problems.”

Suvorexant tablets are to be taken 30 minutes before bedtime. The drug targets orexin receptors in the brain. A neurotransmitter produced in the hypothalamus, orexin binds to receptors in several brain regions, and the ensuing signaling works to keep a person awake. By binding and blocking orexin receptors, suvorexant temporarily halts orexin signaling.

Some people with Alzheimer’s have high orexin levels in their cerebral spinal fluid, which is thought to contribute to the disturbance in their sleep/wake cycle (Gabelle et al., 2017). This is the first study to evaluate suvorexant in AD.

The 277 participants who completed the double-blinded trial averaged 70 years of age and scored between 12 and 26 on the Mini Mental State Examination (MMSE) when they enrolled. Eight participants dropped out. They all suffered from insomnia as defined by the DSM-IV, meaning they had enough trouble falling asleep or staying asleep to impair their daily functioning. All participants first got placebo for two weeks and spent two nights in a sleep lab to record a baseline sleep pattern. Then 136 were randomized to 10 mg suvorexant and 141 to placebo. After two weeks, 77 percent of treated patients tolerated the drug well enough to double their doses. Five mg, 10mg, and 20 mg tablets are FDA-approved. After four weeks, a final overnight polysomnography test assessed the impact of drug or placebo.

Suvorexant met both its primary and secondary efficacy endpoints, improving total sleep time and shortening wake time in the middle of the night. At baseline, people on drug slept an average of 278 minutes, similar to the 274 minutes for those on placebo. After four weeks of treatment, those on suvorexant snoozed 73 minutes longer at night while the placebo controls improved by 45 minutes, a 28-minute advantage for the drug. The 45 minutes more sleep in the non-drug group—a large placebo effect—was due to sleep hygiene coaching that was also part of the trial, Herring said. A 20-minute improvement is considered clinically meaningful, Herring said. Treated individuals were twice as likely as those on placebo to gain an hour of extra sleep.

After falling asleep, patients on the drug woke up and then lay awake in the middle of the night for 45 minutes less than they had at baseline. People on placebo reduced their wake time by 30 minutes, a 15-minute improvement for the drug. In addition, treated patients scored lower on the Clinician Global Impression of Severity (CGI-S) of insomnia versus placebo, while caregivers rated the patients’ sleep quality higher.

Herring said one concern about using sleep medications in dementia is that the drug might worsen cognition. The researchers assessed MMSE and the Digit Symbol Substitution test at baseline and after four weeks and saw no difference between scores. This was but a short-term trial. Merck did not present longer-term data on sleep or cognition.

Adverse events occurred in 22 percent of drug-treated patients, compared with16 percent of the placebo group. They included mild-to-moderate drowsiness during the day in six patients treated with suvorexant versus two people on placebo, headache in five treated and six controls, dry mouth in three treated patients and one on placebo, and falls in three from the treatment group.

“The absolute and placebo-relative improvements are quite dramatic, and hopefully reflect physiologically healthy sleep,” wrote Logan Schneider, Stanford University School of Medicine, California, to Alzforum. “The undesirable consequences of non-physiologic sleep that standard hypnotics or sedatives provide often make me quite cautious about using traditional medications in this population,” Schneider wrote. He was less concerned about the daytime drowsiness experienced by some patients, though he wondered about the duration of symptoms and impact on quality of life.

Merck plans to file the data with the FDA for potential inclusion in the BELSOMRA prescribing information.

Sleeplessness reportedly drives the accumulation of amyloid and release of tau (Jan 2018 news; Apr 2018 news; Jan 2019 news). Could suvorexant potentially slow disease progression by lengthening sleep? No longer-term studies are currently planned to explore this question, though it is on the radar for the future, Herring told Alzforum.

The possibility is intriguing, Holtzman agreed, saying, “Researchers might want to explore this mechanism as a prevention strategy, to see if we could actually delay the amyloid and tau-induced changes promoted by too much wakefulness.”

Suvorexant is also being studied for insomnia in multiple sclerosis, PTSD, and diabetes, as well as bipolar, major depressive, and alcohol and cocaine use disorders.—Gwyneth Dickey Zakaib


  1. The absolute improvement and placebo-relative improvements [in this study] are quite dramatic, and hopefully reflect physiologically healthy sleep. It's nice to see that there is robust efficacy in patients with Alzheimer’s disease, as the undesirable consequences of non-physiologic sleep that standard hypnotics/sedatives provide often make me quite cautious using traditional medications in this population.

    For context, 75 minutes of extra sleep comprises more than 17 percent of a normal sleep duration of seven hours, and these improvements are greater than the ~30 min. improvements in total sleep time seen with existing medications. The drowsiness is not particularly concerning (happening in a paucity of patients), though more clarity on duration of symptoms and impact on quality of life might change that, or would be something worth adapting to in clinical practice.

    Of note, common concerns in these patient populations include risks of falls at night; if that is improved by taking this medication rather than a standard sedative/hypnotic, then we really have moved in the right direction.

    Also, a few things make me pause. The first is the natural adaptation to polysomnography (PSG). Improvement over baseline might be partially due to habituation to the regularity of PSG that is really demonstrating resolution of the first-night effect. Nonetheless, these findings look legitimate.

    Another aspect is the reporting on wake after sleep onset (WASO) bouts >10 min. This may not accurately capture the true nature of sleep fragmentation by not accounting for meaningful sleep disruptions that are less than 10 minutes in duration. Given that suvorexant is essentially recapitulating narcolepsy, sleep fragmentation is expected, such that WASO bouts >10 minutes may be providing false assurance of sleep continuity.

    Looking at the strengthening sleep-Alzheimer’s connection related to sleep macro and microarchitecture, we'll have to see the long-range impact of this effort to improve sleep, as we may just be reassuring ourselves that "more" sleep was gained, but in fact that sleep may not be healthy. I don't know of an increased prevalence of AD in narcoleptic patients (Scammell's 12-patient evaluation in 2012 suggested normal prevalence), so this may be a moot point.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Skimping on Sleep Makes For More Aβ in the Brain
  2. Does Amyloid Accumulate After a Single Sleepless Night?
  3. Another Reason to Catch Some Zzzs: Sleep Regulates Tau Release

Paper Citations

  1. . Cerebrospinal fluid levels of orexin-A and histamine, and sleep profile within the Alzheimer process. Neurobiol Aging. 2017 May;53:59-66. Epub 2017 Jan 18 PubMed.

External Citations

  1. Phase 3 trial

Further Reading


  1. . Sleep-wake regulation and the hallmarks of the pathogenesis of Alzheimer's disease. Sleep. 2019 Apr 1;42(4) PubMed.
  2. . Orexin receptor antagonists for the treatment of insomnia and potential treatment of other neuropsychiatric indications. J Sleep Res. 2019 Apr;28(2):e12782. Epub 2018 Oct 18 PubMed.
  3. . Brain and cognitive correlates of sleep fragmentation in elderly subjects with and without cognitive deficits. Alzheimers Dement (Amst). 2019 Dec;11:142-150. Epub 2019 Feb 8 PubMed.
  4. . Associations between sleep duration and cognitive impairment in mild cognitive impairment. J Sleep Res. 2019 Apr 21;:e12864. PubMed.