The Alzheimer’s Cleaning Cycle Most Adults Don’t Get Enough Of
A practical guide to improving the hidden nightly reset that supports focus, memory, and long-term mental health.
Every night you sleep, your brain runs a cleaning cycle. Cerebrospinal fluid moves through channels between your neurons, flushing out the proteins that accumulate from a day of thinking. Chief among them: beta-amyloid and tau, the same proteins whose buildup defines Alzheimer’s disease.
This cleaning cycle doesn’t happen uniformly across your sleep. It happens almost exclusively during one specific stage. And most people are spending far too little time there.
The public health message about sleep has focused on hours for thirty years. Get seven to nine. Go to bed earlier. Put the phone down. That advice isn’t wrong, but it’s looking at the wrong number. A 2025 meta-analysis of 79 cohort studies confirmed the U-shaped relationship: both too little and too much sleep increase mortality risk, with seven hours per night as the lowest-risk duration.¹ What it can’t tell you is whether those seven hours are doing the job they’re supposed to do.
Seven hours of fragmented, architecturally disrupted sleep and seven hours of well-structured sleep carry completely different biological consequences. You can sleep eight hours and emerge having almost completely skipped the stage where Alzheimer’s proteins are cleared. Your tracker will tell you you slept fine. The number of hours recorded tells you almost nothing about whether your brain was adequately maintained overnight.
What Sleep Architecture Actually Is
A normal night of sleep is not a uniform state. It’s a sequence of repeating 90-minute cycles, each containing multiple distinct stages with different physiological functions, different electroencephalographic signatures, and different consequences when they’re disrupted or shortened.
Each cycle moves through the same general progression:
N1 (light sleep): the transition from wakefulness. It’s brief, easily disrupted, and minimal restorative function. 5-10% of a normal night.
N2 (intermediate sleep): the largest component of sleep, approximately 45-55% of a normal night. Sleep spindles (bursts of neural activity) and K-complexes (slow negative waves) are the characteristic features. Memory consolidation of procedural and semantic information occurs here.
N3 (slow-wave sleep, or SWS): the deepest and most physiologically critical sleep stage. Characterized by slow, high-amplitude delta waves. Growth hormone is secreted almost exclusively during SWS. Tissue repair, immune consolidation, and, most critically, the brain’s overnight waste clearance system operate most efficiently here. Approximately 15-20% of a normal night in young adults.
REM (rapid eye movement sleep): the stage most people associate with dreaming, though dreaming occurs in other stages too. Brain activity during REM resembles wakefulness. This is when emotional memory is processed, fear extinction occurs, creative problem-solving is consolidated, and the brain integrates new information with existing knowledge. Approximately 20-25% of a normal night.
The cycles don’t all look the same. The first half of the night is dominated by SWS. The second half is dominated by REM. This distribution reflects the biological hierarchy of sleep functions. Brain waste clearance and growth hormone secretion happen first. Emotional processing and memory integration happen later.
This means that cutting sleep from the front — going to bed later than usual — disproportionately reduces SWS. Cutting it from the back — waking earlier than usual, or having an alarm interrupt the final cycles — disproportionately reduces REM. And many common behaviors, substances, and age-related changes disrupt specific stages without touching the others.
What Slow-Wave Sleep Actually Does to Your Brain
In 2013, Maiken Nedergaard and her colleagues at the University of Rochester published a paper in Science describing a discovery that changed how neuroscience understands the function of sleep.
They identified a previously unknown waste-clearance system in the brain — which they named the glymphatic system (from glial + lymphatic) — that uses cerebrospinal fluid (CSF) flowing through channels around blood vessels to flush the brain’s interstitial fluid and remove metabolic waste products.²
The glymphatic system is, in the simplest possible terms, the brain’s overnight cleaning crew.
During SWS, the interstitial space between neurons expands by up to 60%, dramatically increasing the flow of CSF through the brain parenchyma.²
The cleaning crew has access to areas that are much more tightly packed during wakefulness. What they’re clearing, specifically, is the accumulation of metabolic byproducts that neurons produce during their daytime activity — including beta-amyloid and phosphorylated tau: the proteins whose accumulation is the defining pathological hallmark of Alzheimer’s disease.
A 2025 paper in Cell by Hauglund and colleagues identified the precise driving mechanism: during non-REM sleep, synchronized oscillations in norepinephrine levels, cerebral blood volume, and CSF flow create a slow pumping motion — slow vasomotion — that is the strongest predictor of glymphatic clearance efficiency.³ When norepinephrine falls during sleep, blood vessels dilate and contract rhythmically, driving CSF through the glymphatic channels.
The clinical implication is direct: insufficient or architecturally disrupted slow-wave sleep means incomplete overnight amyloid and tau clearance.
A landmark study by Lucey and colleagues found that a single night of sleep deprivation resulted in a 30% increase in soluble amyloid-beta in cerebrospinal fluid compared to normal sleep.⁴
A 2025 randomized crossover study published in Fluids and Barriers of the CNS confirmed in healthy adults that sleep — compared to sleep deprivation — significantly reduced CSF concentrations of both beta-amyloid and tau.⁵
This is happening every night you compromise your deep sleep.
What Happens to Your Deep Sleep With Age
Here is the finding that makes sleep architecture a longevity conversation rather than just a performance one.
Slow-wave sleep declines by approximately 75% between the ages of 20 and 60.⁶ A 25-year-old typically spends 80-100 minutes in SWS per night. A 60-year-old averages 20-30 minutes. The cleaning crew doesn’t disappear, but it has far less time in the building. Over decades, incomplete nightly clearance leads to progressive accumulation of amyloid and tau, beginning potentially 20 years before any cognitive symptom appears.
This decline is partially biological and partially behavioral. Age-related changes in slow wave oscillatory power are partially intrinsic. But a substantial proportion of SWS loss in aging is driven by modifiable factors — chronic stress, alcohol use, poor sleep timing, sedentary lifestyle, temperature disruption, and the use of medications that suppress slow-wave activity — that could be addressed if they were recognized as sleep architecture threats rather than general lifestyle concerns.
The bidirectional trap is the most clinically important feature of this picture. Amyloid accumulation disrupts slow-wave sleep, and disrupted slow-wave sleep causes more amyloid accumulation.
A 2025 review of sleep abnormalities and Alzheimer’s risk in Current Neurology and Neuroscience Reports confirmed that reductions in slow-wave and REM sleep are linked to early alterations in amyloid-β and tau biomarkers even in cognitively unimpaired individuals, meaning the loop begins long before any diagnosis.⁷
This is a self-reinforcing downward spiral that starts in midlife and accelerates with each passing decade of inadequate sleep architecture.
What REM Sleep Does That Nothing Else Can
REM sleep is less well understood than SWS in the context of neurodegeneration, but the evidence connecting its disruption to Alzheimer’s risk has been accumulating rapidly.
The biological functions of REM are distinct from those of SWS. REM sleep is when the brain:
Consolidates emotional memories. Specifically, it extracts the emotional significance of experiences while dampening the raw emotional charge of distressing ones. REM sleep is the neurological mechanism of emotional regulation. People deprived of REM show hyperreactive amygdala responses to negative stimuli: the emotional buffer is offline.
Performs synaptic homeostasis. During REM, the brain prunes and reorganizes synaptic connections established during the previous day, keeping the strongest and discarding the redundant. This process is essential for the integration of new learning with existing knowledge and for the kind of insight and creative problem-solving that feels like “sleeping on it.”
Drives neuroplasticity. Acetylcholine surges during REM, activating neural circuits that support the formation of new connections and the consolidation of complex, contextual memories.
What the Alzheimer’s data specifically shows is striking.
A 2017 study in Neurology by Pase and colleagues — following 321 participants over 12 years — found that lower REM sleep percentage and longer time to enter REM were both independently associated with dementia and Alzheimer’s disease.⁸
A January 2026 preprint tracking 88 adults with home polysomnography and neuroimaging over up to 16 years found that REM characteristics were strongly associated with both the current state of AD biomarkers and the rate of change in those biomarkers over time, suggesting a bidirectional relationship between REM disruption and accumulating Alzheimer’s pathology.⁹
The mechanistic question of why REM specifically predicts Alzheimer’s risk is not fully resolved.
Candidate mechanisms include: REM’s role in synaptic pruning that may affect amyloid clearance pathways; the high acetylcholine activity during REM, which is the neurotransmitter system most devastated early in Alzheimer’s disease; and the emotional stress processing function of REM, which when impaired may sustain the hypothalamic-pituitary-adrenal (HPA) axis — the body's central stress response system — activation that drives neuroinflammation and tau phosphorylation.
What the evidence collectively says is that both major sleep stages — SWS and REM — are independently important for Alzheimer’s prevention, through different mechanisms, at different points in the night. Protecting one while losing the other is not sufficient.
An Honest Note on What’s Still Being Debated
The glymphatic hypothesis — that brain waste clearance is predominantly sleep-dependent and specifically SWS-dependent — is compelling and well-supported by substantial mechanistic evidence. It’s also being actively contested.
A debate at the 2025 SLEEP Annual Meeting raised the question of whether glymphatic clearance may in some respects operate more efficiently during wakefulness than during sleep, based on more recent data suggesting that some fluid dynamics in the brain are more active in the awake state. This does not invalidate the SWS-clearance relationship, but it complicates the clean story of “sleep = cleaning crew, wakefulness = building in use.”
The honest position: the mechanistic evidence that poor sleep quality leads to greater amyloid and tau accumulation, and that amyloid and tau accumulation further disrupts sleep quality, producing a vicious cycle — is robust, replicated, and clinically meaningful. The precise cellular mechanics of how clearance happens are still being worked out. The intervention target — protecting sleep quality and architecture — is clear regardless of exactly which mechanism is doing the work.
This is a field in active development. Some of what we know will be refined. The direction of the evidence is not in doubt.
What Is Destroying Your Sleep Architecture
The most important thing about sleep architecture disruption is that the most potent causes are neither obvious nor discussed. When people are told their sleep is “fine” based on total hours, the architecture-level damage happening beneath that number is invisible.
Alcohol is perhaps the most important. A 2025 systematic review and meta-analysis in Sleep Medicine Reviews confirmed what the laboratory evidence has shown for decades: even moderate alcohol consumption suppresses REM sleep in the first half of the night and produces compensatory REM rebound in the second half, leading to fragmented, early-waking sleep.10 The first cycle or two of REM — when amygdala processing and fear extinction are most active — is disproportionately disrupted. The common experience of “sleeping well” after a glass of wine reflects the sedating effect of alcohol on sleep onset, not the quality of the sleep that follows.
Benzodiazepines and Z-drugs (zolpidem, zopiclone, temazepam) suppress SWS and significantly alter REM sleep. These are among the most commonly prescribed medications for sleep. They produce sleep by a different mechanism than natural sleep, and the result is architecturally different from natural sleep — sedation rather than restorative oscillatory activity. Long-term benzodiazepine use has been independently associated with dementia risk in multiple large-scale studies, and the architectural suppression of SWS may be part of the mechanistic pathway.
Stress and elevated cortisol suppress SWS by keeping the brain’s arousal circuits active. Slow-wave sleep requires the brain to reduce norepinephrine and cortisol to the levels that allow the vasomotion driving glymphatic clearance to occur. Chronic HPA activation from psychological stress, poor sleep hygiene, or cortisol-elevating medications fragments the slow-wave architecture regardless of total sleep duration.
Temperature regulation is a mechanistic requirement for sleep onset and SWS maintenance. Core body temperature must fall by approximately 1-2°C for sleep to initiate and SWS to consolidate. Sleeping in a room that is too warm, or having core temperature elevated by alcohol, evening exercise, or late meals, impairs this thermoregulatory drop and reduces SWS duration and depth.
Sleep apnea fragments sleep architecture by producing hundreds of brief arousals per night — too brief to register as conscious wakefulness, but sufficient to prevent sustained SWS. Obstructive sleep apnea specifically disrupts slow-wave sleep through sympathetic activation, hypoxia, and blood pressure surges during apnea events. Untreated sleep apnea is one of the strongest known drivers of amyloid accumulation and dementia risk, and its effects operate precisely through the SWS and glymphatic mechanisms described above.
The Architecture Assessment You Haven’t Had
The clinical management of sleep focuses almost exclusively on total sleep hours, sleep latency (how long it takes to fall asleep), and the presence or absence of insomnia complaints. Sleep architecture — the proportion of time in SWS and REM, the quality of the oscillatory activity in each stage — is essentially never assessed in standard primary care.
An overnight polysomnography (PSG) provides complete sleep architecture data, but it requires a sleep laboratory, is expensive, and is reserved primarily for sleep apnea diagnosis. Home sleep trackers with accelerometers and heart rate sensors (consumer devices including Oura Ring, Whoop, and advanced Apple Watch models) provide estimates of sleep stages that, while imperfect, are clinically informative as tracking tools. Their absolute accuracy is limited but their sensitivity to directional changes over time is meaningful.
The AHA’s 2024 Scientific Statement on sleep disorders and brain health — published in Stroke — called for greater clinical attention to sleep quality and architecture as modifiable risk factors for cerebrovascular and neurodegenerative disease.¹¹ The statement specifically identified sleep apnea, insomnia, and circadian disruption as priority targets for clinical intervention in the context of brain health preservation.
Most people presenting to their GP with cognitive complaints — fatigue, brain fog, word-finding difficulty, reduced concentration — are not having their sleep architecture assessed. They are told they are getting enough hours, that their blood tests are normal, and that they should manage their stress.
The question that needs to be asked is not “how many hours are you sleeping?” It’s “what is your architecture like within those hours?”
Below, I’ll talk about how to assess your sleep architecture without a lab, the specific behaviors and substances that most powerfully disrupt SWS and REM (with the evidence ranked), what the science supports for protecting and enhancing each stage, the women’s specific protocol for perimenopausal sleep architecture disruption, how sleep apnea screening fits into this, the 90-minute cycle and what it means for sleep timing, and the conversation to have with your doctor.
