Michelle L. Mellion, MD, Chief Medical Officer at AstronauTx discusses why small molecules that can restore deep, restorative sleep, could be key to cognitive diseases such as Alzheimer’s.  

Sleep is the bedrock of good brain health. For a long time, the focus has been on the quantity of sleep. A large body of evidence has built up showing 6-8 hours per night is optimal1. 

But we now know that quantity alone is not enough. To deliver its real benefits, studies show, we must have good quality sleep, as described by what sleep experts and neurologists call the architecture of sleep. 

‘Sleep architecture’ is how our sleep state changes during the night. It comprises ‘macro-architectural’ features – the main phases of sleep, which last from minutes to tens of minutes at a time – and ‘micro-architectural’ features, fleeting bursts of brain activity which endure for less than half a minute, and might only last a fraction of a second. 

Our knowledge of sleep architecture, how it changes as we age, and how it differs in people with neurological problems, is now growing at a rapid rate, making it an exciting time to be involved in the field.  

Our increasing understanding of the biology of sleep and how it supports cognitive function is opening up incredible opportunities, as we begin to apply this knowledge to help people with neurological conditions such as Alzheimer’s disease by enabling them to achieve better quality sleep. 

Furthermore, we have identified molecules that do not simply help people sleep more, but could help them sleep better – and reap all the neurological rewards that entails. 

Why deep sleep/NREM3 sleep matters for brain health 

Sleep consists of two broad states: rapid eye movement (REM) sleep and non-REM (NREM) sleep. NREM sleep includes a deep stage known as NREM3, where most slow-wave sleep occurs. This stage is critical for memory consolidation, brain maintenance and the removal of waste products linked to neurodegenerative disease. 

During NREM3 sleep, slow-oscillation brainwaves help move the day’s experiences from the hippocampus, a temporary and space-limited memory store, to the larger cortex. Here memories are stored ‘permanently’, while the hippocampus is freed up to receive fresh information the next day.  

These slow waves are assisted by a micro-architectural feature called ‘sleep spindles’ –strong bursts of electrical activity lasting 0.5 to 3.0 seconds which originate in the thalamus and help with this consolidation of information. 

The slow waves also facilitate an essential maintenance process called glymphatic flow, in which the brain is flushed of toxins strongly linked to neurodegenerative diseases, including β-amyloid and tau. It’s a little like the garbage men coming out to clean the streets every night, or a regular oil change for a car’s engine. 

Unfortunately, as we move through middle age our ability to enter deep (NREM3) sleep, and stay there long enough, diminishes. We spend more of the night in lighter sleep and wake more frequently.  

This ‘sleep fragmentation’, as it is known, reduces our ability to consolidate memories and maintain a healthy brain. Analysis of people over 60 years old in the Framingham Heart Study, a US longitudinal health survey, found for every 1% annual reduction in slow-wave sleep duration there was a 27% increase in dementia risk and a 32% increase in Alzheimer’s risk2. 

As we age, our ability to enter into and stay in the deep sleep stage where powerful sleep spindles coordinate properly with slow-wave brainwaves can be compromised. 

For best information transfer, each spindle should fire just as the slow wave reaches its peak. But with age, the ‘pacemaker’ driving this coordination (known as spindle / slow oscillation coupling) can malfunction. Strong sleep spindles are also known to help keep us in deep sleep, so if their power and frequency diminish so does our ability to stay in deep sleep. 

Multiple studies have confirmed that both fewer, weaker spindles, and poorer spindle / slow oscillation coupling, are associated with an increased risk of Alzheimer’s disease3,4,5. 

While this article focuses mainly on NREM sleep, it should also be noted that reductions in REM sleep – which too plays important, albeit different, roles in memory and brain hygiene – have also been linked with an increased risk of dementia.  This relationship is not as well understood and is an active area of research6. 

The causality problem – and why we think poor quality sleep really does matter 

While the relationship between sleep and neurodegeneration is likely bi-directional, a growing body of evidence suggests that chronic disruption of slow-wave sleep contributes directly to brain ageing and dementia risk. Given the established role of deep sleep in memory consolidation and glymphatic clearance, preserving sleep quality by stabilising sleep architecture – by reducing sleep fragmentation, lengthening NREM bouts and increasing deep sleep –  is increasingly viewed as an important component of brain health. 

Why the ageing brain can’t sleep as well – and what we can do about it 

One of the first brain regions to atrophy as we age is the medial prefrontal cortex – exactly the region where slow-oscillation brainwaves are generated. In 2013, researchers definitively linked this atrophy with reduced NREM sleep7. 

Besides these structural changes, we now know that levels of various neurotransmitters, which must be right to trigger and maintain slow-wave sleep, also change as we get older, as do the interactions between them. 

AstronauTx has been working to identify small molecules that are able to modulate certain neurotransmitters involved in slow-wave sleep. 

This therapeutic approach opens the possibility of restoring sleep quality, specifically by improving the ability to access NREM3 and maintain it for longer. If you can improve sleep architecture, we believe that you can improve cognitive function and potentially alter the course of disease. 

Initially, we are focusing on Mild Cognitive Impairment (MCI) and Alzheimer’s. This is because there is a strong body of evidence linking poor quality sleep to these indications, clear mechanisms at play (such as the accumulation of β-amyloid and tau), and tools such as EEG/polysomnography that can assess if an intervention is making the intended impact. 

However, such therapeutics have the potential to treat people with a wide range of conditions where poor sleep quality is implicated – from other neurodegenerative diseases including Parkinson’s and Huntington’s, to neurological conditions which typically affect younger people, like ADHD. 

There are serious questions to ask along the way. While a causal link between chronic poor sleep quality and numerous neurological conditions is strong, proving that a sleep intervention actually improves the symptoms of a disease, or delays its progression, is another matter. Each intervention – pharmaceutical or otherwise – will have to be rigorously tested in each indication. 

There are good reasons to think success is possible though. For example, Takeda has demonstrated that its drug Orzeyful (oveporexton), recently approved by US Food and Drugs Administration in August, which mimics a neuropeptide called orexin that people with narcolepsy type 1 (NT1) lack, can help them stay awake during the daytime and prevent sleepiness showed improvement in cognitive symptoms8. 

In March, Eli Lilly announced it was buying Centessa – which has a rival orexin-based asset for narcolepsy in Phase 3 trials, and is developing assets for a range of sleep disorders – for up to $7.8 billion9. As The Financial Times noted in June, investors are waking up to the merits of biotechs focused on sleep10.  

The same month, Eli Lilly’s Chief Scientific Officer, Dr Dan Skovronsky, told Bloomberg he thought sleep conditions like narcolepsy could be “just the tip of the iceberg”11. He said: “Maybe someday we can make medicines that can help a broad range of people sleep better and be better awake during the daytime.”   

In the longer term, therapeutics which improve sleep architecture could be used more widely to help maintain our sleep quality throughout life, so we continue to sleep well into middle and older age. Maintenance of sleep quality will help us to preserve our cognitive function later into life, and perhaps delay the onset of neurodegenerative diseases. Given that populations around the world are ageing rapidly – one in six will be 65 or older by 2050, up from 10% today12 – such therapeutics could be of major societal benefit. 

About the author  

Michelle L. Mellion, MD, Chief Medical Officer at AstronauTx, is a neurologist and drug development leader with 20+ years of experience in academic medicine and biotechnology. Formerly on Brown University’s faculty, she has held leadership roles at Biogen, Vertex, Fulcrum, PepGen, and EveryONE Medicines, advancing therapies for neurological, neuromuscular, rare, and ultra-rare diseases. 

References 

  1. Sleep chart of biological ageing clocks in middle and late life The MULTI Consortium., O’Toole, C.K., Song, Z. et al. Sleep chart of biological ageing clocks in middle and late life. Nature (2026). https://doi.org/10.1038/s41586-026-10524-5  
  2. Association Between Slow-Wave Sleep Loss and Incident Dementia. Himali JJ, Baril AA, Cavuoto MG, et al. JAMA Neurology. 2023;80(12):1326-1333. doi:10.1001/jamaneurol.2023.3889. 
  3. Association of Sleep Spindle Activity With Cognitive Decline in Early Clinical Stages of Alzheimer Disease. Bender AC, Berezuk C, Pellerin KR, et al. Neurology. 2026;106(1):e214459. doi:10.1212/WNL.0000000000214459. 
  4. Sleep as a Potential Biomarker of Tau and Β-Amyloid Burden in the Human Brain. Winer JR, Mander BA, Helfrich RF, et al. The Journal of Neuroscience : The Official Journal of the Society for Neuroscience. 2019;39(32):6315-6324. doi:10.1523/JNEUROSCI.0503-19.2019. 
  5. Deep Sleep Slow Wave-Spindle Coupling Is Selectively Linked to Plasma Amyloid-Β Levels in Older Adults in Clinical Trials. Wunderlin M, Wicki K, Teunissen CE, Züst MA. Scientific Reports. 2026;16(1):11716. doi:10.1038/s41598-026-47886-9. 
  6. Sleep Architecture and the Risk of Incident Dementia in the Community. Pase MP, Himali JJ, Grima NA, et al. Neurology. 2017;89(12):1244-1250. doi:10.1212/WNL.0000000000004373. 
  7. Prefrontal atrophy, disrupted NREM slow waves and impaired hippocampal-dependent memory in aging. Mander, B., Rao, V., Lu, B. et al.  Nature Neuroscience 16, 357–364 (2013). https://doi.org/10.1038/nn.3324 
  8. https://www.fda.gov/news-events/press-announcements/fda-approves-first-drug-treat-full-range-narcolepsy-type-1-symptoms 
  9. JAMA Neurol. 2026;83(2):145-152. doi:10.1001/jamaneurol.2025.4825  
  10. https://www.nytimes.com/2026/03/31/business/eli-lilly-narcolepsy-centessa.html 
  11. https://www.ft.com/content/0c71ad46-6d76-4ff0-a952-76329881eee5?syn-25a6b1a6=1 
  12. https://www.bloomberg.com/features/2026-eli-lilly-weight-loss-drug-boom/ 
  13. https://www.un.org/en/global-issues/ageing