What Is the Glymphatic System?
Your brain runs hot. It makes up about 2% of your body weight and burns roughly 20% of your energy. All that metabolic activity produces waste: amyloid beta, tau fragments, inflammatory byproducts. The rest of your body has a lymphatic system to haul that trash away. The brain lacks traditional lymphatic drainage, so for decades the question of where brain waste went stayed murky.
The answer came into focus about a dozen years ago. Your brain has its own drainage network, and it does most of the work while you sleep. We call it the glymphatic system. The "g" stands for glia, the support cells around your neurons that run this plumbing.
Here is the mechanism. Cerebrospinal fluid (CSF), the non-blood fluid in your brain, flows along channels called perivascular spaces. These are water channels that run alongside your blood vessels. CSF gets produced largely by the choroid plexus in the lateral ventricles, mixes with the interstitial fluid between your cells, and that mixing dissolves and pulls proteins and toxins out of the tissue. The waste-laden fluid then drains out along those perivascular spaces, through the meninges wrapping your brain, and eventually into the spinal CSF and the body's lymphatic system. It is a water-channel exchange system running parallel to your blood supply.
How Does Aquaporin-4 Control Brain Drainage?
The whole system is gated by a water-channel protein called aquaporin-4 (AQP4). These channels sit at the end-feet of astrocytes, the star-shaped glial cells that wrap around blood vessels. When AQP4 is properly polarized, meaning the channels are concentrated at those astrocyte end-feet, CSF moves efficiently through the tissue. When that polarization is lost, flow drops.
This detail matters because AQP4 polarization degrades with aging, and that degradation appears to track with Alzheimer's risk. The protein is becoming a target for medication. In Alzheimer's mouse models, manipulating aquaporin changed the lymphatic flow, which tells us we may be able to take pressure off the system pharmacologically down the line.
Why Does Deep Sleep Drive Brain Clearance?
The system is dramatically more active when you sleep. During deep non-REM sleep, the interstitial space around your brain expands. Some estimates put the increase near 60%, which physically opens the door for drainage.
At the same time, delta activity (1-4 Hz slow-wave sleep) drives oscillatory fluid waves through the tissue. Picture a washing machine on its agitation cycle, spinning and shaking to flush the dirt out. That is roughly what these delta-triggered surges are doing. Increased volume plus increased wave action means increased waste clearance, and it happens when you are deep in dreamless sleep. Amyloid beta and tau clearance depends on this process as much as memory consolidation does.
A practical note for the self-trackers in the audience. Ignore the REM numbers on your Oura Ring or Whoop strap. Those are largely made-up. Deep sleep readings are usually valid, with one caveat: some people get poor-quality deep sleep that does not descend all the way down, and the trackers cannot tell that apart from healthy deep. If your tracker says your deep sleep is good but you do not feel rested, or if it says you are getting too little, the rough rule is to double the number you see and you will land closer to reality. You have real leverage over deep sleep. You can push it with fasting before bed and with afternoon exercise that burns off cortisol when it is naturally low. For a full breakdown, see my guide on biohacking sleep, and the role of delta and SMR in SMR neurofeedback.
How Does the Glymphatic System Connect to Alzheimer's?
This is where it gets clinically important. The proteins the glymphatic system is supposed to clear, amyloid beta and tau, are the same ones that build up and turn sticky in Alzheimer's. Those tissues also glycate and oxidize faster in the presence of excess sugar and oxidative stress, producing advanced glycation end products that drive tangles and plaques and damage tissue further.
The core hypothesis is straightforward: impaired clearance allows protein accumulation, and accumulation starts the cascade. If you have ever lived through a trash strike in a big city, you know how fast a smelly mountain piles up on every street corner once collection stops. Your brain is a hot, smoky engine pushing out byproduct all day. It needs to cool down at night and run the drainage.
Two human studies have moved this from theory toward something measurable. Benedetti and colleagues, published in Brain in 2025, ran a randomized crossover in healthy young people. A single night of normal sleep reduced CSF levels of amyloid beta and tau compared with staying awake. No change in neurofilament, no change in GFAP. The effect was specific to the Alzheimer's-defining proteins.
Then Iliff and colleagues, published in Nature Communications in early 2026, ran another randomized crossover with 39 adults. Sleep-active physiology, the pumping and drainage cycles, increased brain-to-plasma clearance of these biomarkers. Time spent in non-REM stage 2 and stage 3 sleep correlated directly with clearance. Clearance rates explained roughly half the variance in the signal. Sleep is physically moving these proteins out of the brain and into the bloodstream, and we can now quantify it.
What Does Glymphatic Failure Look Like in the Brain?
When researchers look at Alzheimer's brains, the vascular, lymphatic, glymphatic, and sleep signals converge into one sobering picture. A 2024 study by Lee, published in Alzheimer's & Dementia, mined a classic imaging dataset using a proxy for glymphatic function called the DTI-ALPS index. ALPS is a diffusion-tensor-imaging measure of water diffusion along the perivascular channels. Worse ALPS scores tracked with faster brain atrophy, more memory complaints, and reduced volume.
In groups of elders with mild memory complaints, the DTI-ALPS measure shows a negative correlation with progression to Alzheimer's, around -0.7, which is strong for this kind of work. In cognitively unimpaired older adults, the pattern holds: better vascular flow goes with intact function. The DTI-ALPS variant is tuned to pick up the diffuse water around the vessels rather than the blood within them, so it functions as a readout of how open the drainage highways are.
What Is the Difference Between Alzheimer's and Dementia?
This question comes up every time the topic surfaces, so let me draw the line clearly.
Alzheimer's is a specific disease process, a syndrome with inputs and mechanisms we still do not fully understand. Dementia is a symptom that lives inside it. Dementia means significant memory impairment, usually long-term retrieval. Many different diseases can produce dementia.
The retrieval problem that shows up early in Alzheimer's dementia is a different phenomenon from the everyday brain fart of hunting for a word or a name. That word-hunting is usually a speed-of-processing issue driven by inadequate deep sleep, which loops us right back to tonight's topic. Alzheimer's hits episodic, experiential memory first: forgetting what you wore at your wedding, forgetting the big party you planned last year. That pattern comes from how the disease affects the hippocampus and parahippocampus and which tissues handle that input and output.
Slow processing and word-finding trouble in healthy aging is often sleep decline rather than cognitive decline. Fix your sleep architecture and your habits, get deeper sleep, and your brain speeds up, your verbal fluency improves, your thinking smooths out. For the slow-processing version of this, see biohacking brain fog and the critical aging window.
Why You Cannot Catch Most Dementias Early by Symptoms Alone
A point about timelines that should change how you think about prevention. Most dementias are diseases you carry for a decade, sometimes two, before symptoms appear roughly halfway through the process. Parkinson's tremor does not show up until you have lost 75 to 80% of your dopaminergic neurons in the pars compacta of the substantia nigra. That is a decade or more of degradation before the visible sign.
Does that mean the first 75% of those neurons are irrelevant? No. The brain has a remarkable ability to tune the dynamic range around any absolute level of neurotransmitter. This is also why measuring neurotransmitter levels in metabolic testing is generally meaningless. The brain tunes around the range, and you are sampling far downstream from where the action happens, after layers of degradation and other metabolic noise. If a biohacker is selling you on neurotransmitter level testing, treat those numbers as noise rather than signal for now.
Does Metabolic Health Affect Glymphatic Flow?
Yes, and the data keep converging. In insulin-resistant diabetic animal models, you see reduced CSF influx alongside reduced outflow. The "type 3 diabetes" framing of Alzheimer's, the glycation-end-product framing, the oxidative-stress framing, and the waste-accumulation framing are all imperfect models with holes poked in each one. They point in the same direction. Inflammation, reduced flow, and oxidative stress all read out as the same metabolic tissue-loss phenomenon across datasets.
That convergence is good news for prevention, because metabolic health is something you can act on. Time-restricted eating is one lever; I cover it in strategic fasting.
What Can You Actually Do About Glymphatic Function?
Two interventions sit clearly under your control with mechanistic support behind them.
Sleep first. Go to bed early and fasted. Protect your deep sleep, because that is when delta-driven clearance peaks. Everything in my sleep optimization and morning practice work feeds this.
Exercise second. Large movements of the thigh muscles drive lymphatic pumping, which is part of why walking is so effective. Exercise also appears to improve AQP4 polarity, meaning better drainage. In Alzheimer's mouse models, exercise shifted glial cells to drain better, reduced amyloid, and improved memory. The exact mechanism is still being worked out, and a twelve-week protocol does not reliably improve memory in older adults with established decline. If you are over 40 or 50, and especially if you are a woman, weight this more heavily. I recommend resistance training two to four times a week plus something gentle each morning after you wake and before coffee: yoga, tai chi, or a walk. Get a puppy if you need an external reason to go outside every morning.
Beyond those two, here is what is on the horizon and what I would consider personally.
Vasomotion training. I use a technique called hemoencephalography (HEG), an infrared sensor you wear while you practice toning your cerebral vasculature, leaning into the natural surges of vasomotion and making them stronger. The pacemaker neurons that set vasomotion sit right around the vasculature, and those same regulatory cells almost certainly influence the perivascular water channels running alongside. My clinical extrapolation is that HEG is likely training glymphatic and lymphatic outflow coupled to blood flow, because the regulatory tissue is sitting right there and it would be strange for it to leave the adjacent fluid channels untouched. Reduced vasomotion tracks with migraines, brain fog, dysautonomia, and aging with impairment.
Gamma stimulation. 40 Hz light and sound stimulation rejuvenates tissue in mouse models. I have not seen solid human research yet. If I or a parent were facing this, I would likely try 40 Hz flicker stimulation, often paired with 1070 nm near-infrared light, on the strength of the animal literature. It probably cannot hurt beyond the cost of a good headset. For the red-light side of this, see photobiomodulation.
Micronized purified flavonoid fraction (MPFF). For the body's lymphatic system, MPFF acts as a lymphagogue, promoting pumping and drainage. It seems to help with mast cell issues, some dysautonomia, and conditions like lipedema and lymphedema. I have not found studies showing a brain effect, so this is a theory: if you have a backed-up body lymphatic system and you improve its drainage, the brain likely benefits, either by preventing a backup or by participating in a drainage signal. You can feel it act, with a mild histaminic quality, which at least tells you the system is engaged.
Neuromodulation. Neurofeedback is a reasonable tool in this space. If you want to understand what it is and whether it holds up, start with is neurofeedback legitimate and the QEEG brain mapping guide.
A Note From the Q&A: Could Infra-Low Neurofeedback Train Drainage?
Someone asked whether infra-low and infraslow neurofeedback might affect the lymphatic or brainstem systems. It is plausible. My hunch is that what people are training there is the pacemaker neurons around the vasculature, the same cells that set vasomotion. I have not seen good evoked-potential brainstem studies or solid imaging to support the specific claim, only hand-waving theories.
The experiment is doable. Take someone running infra-low training who claims a flow effect, and measure DTI-ALPS before and after on a DTI-capable MRI. ALPS reads out water diffusion in the perivascular spaces, so it gives you a flow metric: how open are the highways. The hard part is the usual problem in this field. Experiments are expensive, hard to replicate, and neurofeedback requires iteration that does not sit comfortably inside rigid experimental designs. Someone still has to run it.
What to Do Starting Tonight
The mechanistic story is converging from several directions. Impaired glymphatic clearance, fading aquaporin-4 polarization, reduced vasomotion, and metabolic dysfunction all track with the same tissue-loss pattern underlying Alzheimer's risk, and the clearance window opens during delta-rich deep sleep. We do not yet have a clean, validated human intervention that prevents Alzheimer's by fixing drainage. Two levers have real mechanistic support and decades of safety behind them: protect your deep sleep, and move your body, with resistance training a few times a week and a fasted walk most mornings.
If you want to see what your own brain looks like, listeners of the show can get a QEEG brain map at any Peak Brain office for $250 once a year, with unlimited repeats for that year, and I will teach you to read the data yourself. For the cost and coverage questions people always ask, see how much neurofeedback costs and whether neurofeedback is covered by insurance.