Each daily serving of ultra-processed food was associated with a 13 % higher risk of Alzheimer's disease. That figure comes from a January 2025 paper in the Journal of Prevention of Alzheimer's Disease, which tracked 1,375 participants in the Framingham Heart Study for 12 years. Among people eating 10 or more servings a day, the risk was roughly three times higher. In the United States, according to David Perlmutter, ultra-processed foods make up about 60 % of adult calories.
Perlmutter is a board-certified neurologist, and Brain Defenders is his 16th book. He's careful about what the study can and can't show. It's observational, and he says plainly that it demonstrates correlation, not causation. Then he gives his own view: "If you ask me, and I think you are asking me, yes." He thinks these foods are doing real damage, and his reasons rest on a single biological mechanism he believes sits underneath nearly every neurodegenerative disease.
His argument, made in conversation with Dr Rangan Chatterjee, runs as follows. Alzheimer's is mostly a metabolic disease. The metabolic damage begins decades before anyone forgets a name. The cells that turn that damage into brain destruction answer to choices people make every day.
The deadline is closer than the diagnosis suggests
Perlmutter cites 53 million people living with Alzheimer's worldwide and expects that number to triple by 2050. No drug has meaningfully changed the course of the disease. He sees the growth rate as evidence against a mainly genetic explanation: "Our genetics have not changed. Our metabolism has changed."
Chatterjee mentioned that friends in their 40s keep telling him they're worried about their parents. Perlmutter's answer is the most practical point in the conversation:
"Your friends in your 40s should be worried about themselves because we know that the seeds are sown for Alzheimer's in terms of the metabolic threats to the brain in our 30s and in our 40s."
The metabolic changes, he says, come two to four decades before clinical signs appear. Loss of smell is often one of the first. Memory and executive function problems are "late to arrive at the party." On this view a diagnosis marks the end stage of a long process, which means the period when prevention works best is the period when most people aren't thinking about it.
He also points out that a medical journal devoted entirely to Alzheimer's prevention now exists. In a system built around treating diseases after they appear, he reads that as a real shift.
Microglia switch sides, and metabolism controls the switch
Brain Defenders is built around microglia, the brain's resident immune cells. Perlmutter calls them the good twin and the evil twin. In their supportive state, labeled M2, they protect the blood-brain barrier and help the brain grow new neurons and form new synapses. In their destructive state, M1, they prune synapses, slow the growth of new neurons and damage the blood-brain barrier.
What decides which state they're in? Their metabolism. And their metabolism tracks the metabolism of the whole body. Perlmutter links this to immunometabolism, the growing body of research on how metabolic health shapes immune balance. He notes the same M1/M2 split appears in macrophages elsewhere in the body, and even in chondrocytes, the cells that maintain or break down joint cartilage.
When microglia move from M2 to M1, three things change. They change shape. They change function. And they change how they make energy, moving away from mitochondrial respiration toward glycolysis. That third change is the reason Perlmutter keeps returning to mitochondria. Twenty years ago, speaking at the Institute for Functional Medicine, he called Alzheimer's "an acquired mitochondriopathy": a failure of the cell's energy machinery, mostly not inherited.
He supports the broad claim with brain imaging. TSPO PET scans can show activated M1 microglia in living people. When he lists the conditions where those scans come back positive, the list is long: Alzheimer's, Parkinson's, frontotemporal dementia, multiple system atrophy, progressive supranuclear palsy, long Covid, PTSD and major depression. Mood disorders are on it. Perlmutter says this explains why depression has long been associated with brain inflammation and with synapse loss in areas such as the prefrontal cortex.
That breadth is what makes the framework attractive, and it's also a reason for some caution. A marker of microglial activation showing up across many diseases is consistent with a shared mechanism, but it doesn't prove that microglial metabolism is the primary driver in every one of them. Perlmutter presents it as the unifying explanation. The imaging fits that reading without settling it.
Why the damage keeps spreading after the cause is gone
Inflammation is one of the main things that pushes microglia into the M1 state. Inflammatory signaling molecules called cytokines act on M2 microglia and shift them. Because cytokines circulate, they can start anywhere in the body. Perlmutter's example is a leaky gut, and he says gut-driven inflammation helps explain the consistent link between Parkinson's and bowel disorders.
Then comes the feed-forward loop. M1 microglia produce inflammatory cytokines of their own, and those cytokines convert nearby M2 cells. Perlmutter describes this as spreading "through the brain like a cancer," with the destructive cells eventually becoming the majority.
He uses this to explain why chronic traumatic encephalopathy keeps progressing in football players after they retire, and why Alzheimer's and Parkinson's continue to worsen long after the original head injury or toxic exposure.
It also helps make sense of a result Chatterjee raised from the book. In a 2024 study in Neurology led by a Harvard Medical School researcher, people who switched to Mediterranean, DASH or MIND-style diets after the initial survey still showed the cognitive effects of their earlier ultra-processed food intake. A better diet didn't automatically reverse the damage. If the microglial loop sustains itself, removing one input won't necessarily stop it. That's the basis for Perlmutter's case for acting on several fronts at once.
Single-drug thinking keeps failing the brain
Western medicine, Perlmutter says, is built on monotherapy: one drug for one disease. In Alzheimer's that approach hasn't worked. Drug trials began with the goal of improving cognition, scaled that back to preventing decline, and in his view didn't reach even the lower target.
Parkinson's drugs, he argues, treat "the smoke, not the fire." He values levodopa and anticholinergics because they let patients keep functioning, but the disease continues to progress underneath them.
Two results point toward treating the underlying metabolic problem. He cites a 2024 randomized trial in the New England Journal of Medicine in which Parkinson's patients given a GLP-1 drug held steady or improved slightly on the Unified Parkinson's Disease Rating Scale over a year, while the placebo group declined. He also mentions his own earlier study, published in Movement Disorders, which used intravenous glutathione to support mitochondrial function and reported improvements on the same scale.
He limits what these results mean. The GLP-1 drug in that trial was chosen because it crosses into the brain, and he says it's no longer available in the US because of pancreatic concerns. A recent trial of oral semaglutide in Alzheimer's patients missed its endpoints, possibly because the oral form didn't reach the brain. His interest is in the mechanism: GLP-1 drugs have been shown to improve mitochondrial function in people with type 2 diabetes, and mitochondrial dysfunction is at the center of his model. On whether these drugs belong in Alzheimer's care, he says he's "open to it," provided the risk-benefit ratio holds up.
The approach he endorses more firmly is multi-pronged. He cites Dale Bredesen's protocol, which addresses up to 36 factors, and in the book presents Bredesen case studies in which patients improved cognitively and the hippocampus grew on MRI, as read by radiologists who didn't know about the intervention. Case studies show something is possible, though they can't tell you how often it happens. Perlmutter's narrower claim, that the belief the brain can't improve is out of date, is well supported by them.
Choose a diet by what it does to your blood sugar
After 16 books, Perlmutter has stopped recommending a named diet. Paleo, primal, MIND, Mediterranean and green Mediterranean all have followers, and he thinks requiring an app and a rulebook makes it harder for people to stick with any of them. He now defines a good diet by its results. It should:
- keep blood sugar under tight control
- provide plenty of fiber, which means eating mostly plants
- be colorful, for a wide range of polyphenols
- include adequate protein (he gives targets in grams per kilogram of ideal body weight) and good fats
People can follow their own food traditions as long as those results hold. He also observes, after a recent trip, that people around the Mediterranean no longer eat a traditional Mediterranean diet. The standard American diet has become the global diet.
The change he most wants to make is moving from following a protocol to measuring its effect. If you can't get a continuous glucose monitor (available without a prescription in the US), he suggests a pharmacy finger-stick test once a month. He treats blood sugar as the top priority because most other lifestyle factors show up in it: poor sleep, lack of exercise and stress all affect glucose. He suspects, without citing data, that social isolation does as well.
His main revision since the start of his career concerns fat. He trained when fat was discouraged in every form, and he now considers the right fats important for metabolism, controlling inflammation and maintaining the structure of neurons.
Track the numbers that predict trouble early
Perlmutter strongly supports direct-to-consumer lab testing, because it lets people take responsibility for their health without relying on a doctor's annual fasting glucose check. According to him, about 90 % of Americans with prediabetes don't know they have it.
Chatterjee has a stake in this topic. He co-founded Do Health, a UK service that tests 11 biomarkers chosen because they affect metabolic health and can be changed, and retests every four months. Perlmutter called the approach "an absolute home run" and contrasted it with US companies that offer 400 different tests, which he finds overwhelming. Readers should know about the commercial interest, although the biomarkers themselves are standard ones.
Three markers come up repeatedly in the conversation.
Fasting insulin rises before glucose does
Elevated fasting glucose tells you there's already a problem. Fasting insulin starts rising earlier, which makes it the better early warning. Perlmutter recommends testing both, along with HbA1c, and using a CGM if you want the most detail.
Homocysteine shows whether your B vitamins are working
Perlmutter carries an MTHFR polymorphism, which reduces his ability to methylate, or activate, B6, B12 and folate. As a result his homocysteine was high. He takes methylated folate, sublingual B12, a B complex and occasional B12 injections. His most recent reading was 8.2.
He explains why homocysteine matters. The body converts it to homocysteic acid, which damages blood vessel linings (the cardiovascular risk) and is toxic to mitochondria, which brings it back to the microglia. Normal B vitamin levels in a blood test don't settle the question. What counts is whether your body can use them, and homocysteine is a stand-in measure for that. He says 28 % of people in Western populations carry the variant he has.
He also has a reply to people who doubt supplements because they can't feel any effect. He didn't feel different either. The proof was in the lab result.
Uric acid tells your body winter is coming
Perlmutter traces the uric acid problem to a genetic change in Proconsul, a primate ancestor, about 8 million years ago, when the uricase enzyme that breaks down uric acid was lost. Fructose from fruit ripening in autumn is converted directly to uric acid, which signals the body to store fat, slow mitochondrial function and raise blood pressure as protection against scarce water. Ancestors with higher uric acid were more likely to survive the winter.
Today fructose is in food all year. Perlmutter says more than 80 % of US grocery products contain added sweetener, much of it high-fructose corn syrup. Uric acid inhibits nitric oxide, which makes blood vessels less able to relax and makes insulin less effective. He links higher uric acid to all-cause mortality, dementia and diabetes.
Normal range describes the average, which isn't the goal
Reference ranges are statistical averages, two standard deviations around the population mean. "What is average is not necessarily what is best," Perlmutter says. A fasting glucose of 100 counts as normal. He doesn't consider it good enough.
Muscle acts as a gland that supplies the brain
Perlmutter's simplest piece of advice is general: accept that you're in charge of these choices. On exercise, his explanation is that muscle works like an endocrine gland. When muscles are active they release signaling molecules, including cathepsin B, interleukin-6 (which helps in this setting) and irisin. Irisin reaches the brain and stimulates production of BDNF, brain-derived neurotrophic factor, which supports neurons, pushes microglia back toward M2, and helps protect synapses and the blood-brain barrier.
He cites a one-year trial by Kirk Erickson at the University of Pittsburgh in which aerobic exercise, compared with stretching, raised BDNF, enlarged the hippocampus on MRI and improved memory. He also mentions research linking daily step counts to cognitive decline, with benefits peaking close to the familiar 10,000 steps.
Aerobic and resistance training both matter. Aerobic exercise raises BDNF, and resistance training, he says, "increases the size of the pharmacy." He also recommends flexibility and balance work, mainly to prevent injuries, since an injury can end an exercise habit.
At 71, he lifts weights three times a week, warms up with stretching and a vibration plate, and runs, walks or rides a gamified stationary bike on the other days. He's currently fourth on the bike's leaderboard.
Inflammation weakens the part of the brain that helps you change
This section ties the rest together. In Brainwash, an earlier book written with his son, Perlmutter set out a two-part model of decision-making. The amygdala makes fast, self-focused choices without regard for consequences. The prefrontal cortex, "the adult in the room," weighs the future and considers other people. The anterior cingulate connects them and lets the prefrontal cortex exert top-down control.
That connection is vulnerable to inflammation.
The loop works like this. Ultra-processed food raises inflammation. Inflammation weakens prefrontal control. Weaker control leads to more impulsive choices: more of the same food, later nights, less movement. Those choices add more inflammation. It mirrors the microglial loop at the level of behavior. Perlmutter's conclusion is that any change that restores prefrontal control makes the following changes easier. Chatterjee adds that the first weeks of a new habit may be the hardest because inflammation hasn't come down yet.
Perlmutter applies the same thinking to alcohol. The older U-shaped curve, in which moderate drinkers had lower dementia risk than non-drinkers, isn't supported by more recent research, he says. "Call it what it is. Alcohol is a neurotoxin." He accepts its social and cultural role, but sets the limit at a glass rather than half a bottle, and suggests non-alcoholic red wine as a way to get the polyphenols.
APOE4 raises the stakes on everything else
Perlmutter is clearly in favor of genetic testing, a change from what Chatterjee learned at Edinburgh Medical School in the late 1990s, when the view was that there was no point knowing because nothing could be done.
The APOE gene has three variants. APOE2 is protective, APOE3 is roughly neutral and APOE4 raises risk. One copy of APOE4 raises Alzheimer's risk three to fivefold. Two copies raise it eightfold, and some reports put it as high as twelvefold. Perlmutter's key point is that APOE4 is "a predisposition but not a determinant."
He cites research by Sarah Marzi in the UK, in which microglia engineered to express each APOE variant were transplanted into mouse models of Alzheimer's that lacked microglia. APOE2 microglia reduced inflammation sharply. APOE4 microglia increased inflammation, cell loss and beta-amyloid buildup. If APOE4 acts largely through microglia, then the measures that keep microglia in the M2 state are the same measures that offset the genetic risk.
Chatterjee asked whether the result changes anything if someone plans to follow the whole book regardless. Perlmutter answered that no one follows the whole book. A positive result tells you how hard to work: adopt the more advanced interventions as well as the basics, and monitor cognitive function more closely. As he put it, these are the cards you've been dealt, so how will you play the hand?
Start when nothing seems wrong
The hardest group to reach is Perlmutter's own generation, raised to assume medicine would repair whatever their lifestyle caused. Chatterjee described a friend's mother who has drunk half a bottle of wine every night for years and won't stop. Perlmutter's advice to that friend is honest: do what you can for your parent, and then turn the advice on yourself. He quotes John F. Kennedy: the time to fix the roof is when the sun is shining.
He also explains why this advice gets so little attention compared with new drugs: "It's not monetized. That's why we don't hear about it very much."
His own motivation is personal. His father, a neurosurgeon, died of Alzheimer's. Asked what he would tell someone who has neglected their brain for decades and worries it's too late, he gave the most hopeful answer of the conversation. Anyone thinking clearly enough to worry about it, he said, is "in great shape" to change direction.
