Were We Wrong About Fish Oil and the Brain?

For thirty years, the supplement aisle has sold fish oil as one of the simplest decisions you can make for your brain. In 2026, two research teams on opposite sides of the world published papers that complicate that story, and the literature behind them has been building for almost twenty years.

IN THIS INVESTIGATION

• What a 2026 ADNI cohort study of 800+ older adults actually found
• Why faster cognitive decline in fish oil users showed up on every measure
• What the brain imaging revealed when classic Alzheimer's pathology came back negative
• The 2026 Cell Reports mechanism work in repetitive brain injury
• Why EPA and DHA do different things inside the brain, and why most trials ignored the difference
• Twenty years of cognitive trials nobody connected, including a 2006 trial whose biomarkers, re-analyzed in 2021, told a different story
• What the 2026 authors themselves said about how to read their own findings
• Dose, baseline status, oil quality, and physiological context; the conditions the supplement aisle has never asked about
• What twenty years of brain literature actually supports if you take fish oil today

Subscribe for more investigations.

Transcript

Two Fish Oil Papers Shock Readers

SPEAKER_00 0:00

In the spring of 2026, two research teams on opposite sides of the world published papers about fish oil. Both findings landed most major science news outlets within days. The first came out of cell reports on March 25th. The research group included the Boston University CTE Center, the same team that has spent the last two decades examining the brains of DC's football players and military veterans for signs of chronic brain injury. Their finding omega-3 fats accumulated in the brain and after head injury were responsible for damage, leaving behind injured blood vessels and the kind of protein deposits seen after repeated head trauma. Three weeks later, the Journal of Prevention of Alzheimer's disease published the second paper, a team built around the Alzheimer's disease neuroimaging initiative, also called ADNI, that's a U.S. research program that's been tracking the brains of aging adults for over 20 years. They found that older adults who took fish oil declined faster on every cognitive test they were given over five years compared to a carefully matched group who took nothing. Among older Americans with cardiovascular disease, roughly one in five takes fish oil, most of them on the advice that it protects the aging brain. For years, we've been told that fish oil is uniformly good for us. Two papers, two continents three weeks apart, both made the science news, both of them alarming in their own right, and the literature behind them has been building for almost 20 years. The human data, the ADNI paper, had access to something that earlier trials did not. Adne tracked the same older adults over years with brain scans, spinal fluid genetics, and detailed medication and supplement logs. They could look at what happened inside the brains of fish oil users, not just what happened to their test scores. The fish oil users were also matched, the non-users, on a gene called ApoE4. That's the single biggest genetic risk factor for Alzheimer's. Carriers had to be evenly distributed between the groups, or the comparison wouldn't mean anything. Then they ran the numbers. On three separate measures of cognitive function, covering memory, daily function, and overall mental status, the fish oil group declined faster than the match controls. Every measure. Highly unlikely to be due to chance, the size of the effect was roughly 8 to 15% of what an Alzheimer's patient typically loses in a year. Small but consistent and present across every measure. Anyone reading a result like this should ask the obvious question. What if the people who started taking fish oil were already declining faster? And the supplement was a response to early symptoms, not a cause of later ones? The research team asked the same question. They compared the pre-supplement trajectory of the future fish oil users to the matched non-users. The fish oil users were not declining faster before they started. If anything, they were trending slightly better. The decline emerged after supplementation began. Next question was where in the brain the harm was showing up. The team looked at three classic signatures of Alzheimer's pathology: amyloid plaques, towel tangles, and the loss of brain tissue itself. None of them were different between the groups. So they looked at one more thing: a different kind of brain scan that measures how much sugar the brain is using. When synapses start failing, glucose uptake drops. In the fish oil group, glucose uptake in the brain regions most vulnerable to Alzheimer's, it was declining faster than in the controls. The decline was associated with reduced brain energy use, not with plaques and tangles of Alzheimer's disease. This was just published, but it's not the first omega-3 study in the cognitive literature to come back with results that no one was hoping for. The first major trial came back in 2006, a Swedish study called Omega-AD. 174 Alzheimer's patients given either a daily dose of omega-3 or a placebo for six months. The two groups declined at the same rate. The trial was reported as negative. Fifteen years later, a different team reanalyzed the spinal fluid samples that had been collected during Omega-AD. They were looking at two specific markers, neurofilament light, that's a structural protein that leaks into spinal fluid when neurons are damaged, and YKL40, a marker of brain inflammation. Both proteins had risen in the omega-3 group, not in the placebo group. The increases were small, but the results were statistically significant on both markers, and the authors used very precise language. The supplement they wrote had produced a possible increase of inflammatory response and axonal damage. In 2022, a randomized trial in Taiwan went even further. The researchers enrolled 163 patients with mild cognitive impairment or early Alzheimer's and split them into four groups. Placebo, DHA alone, EPA alone, DHA, and EPA combined. 24 months, the headline result was familiar. No benefit on the primary cognitive measures, but on the specific cognitive subdomains, two of the supplement groups had gotten worse than placebo. In 2010, the largest trial of its kind had just been published in JAMA, 400 Alzheimer's patients, 2 grams of DHA per day, 18 months across 51 sites. No benefit on daily function, no effect on brain shrinkage measured by MRI. The 2026 ADE paper was the first to find faster cognitive decline in a large group of fish oil users, followed over years, but the damaged signals had been there all along, buried in trials that nobody had connected, and the explanation had been sitting in the biology for years. Somewhere the cognitive trials had never even thought to look. The cognitive trials had been asking the same question for almost 20 years. Does taking omega-3 slow cognitive decline? Almost none of them asked a different one. What does omega-3 actually do in the brain? DHA is the dominant omega-3 in brain tissue by a wide margin. DHA comprises approximately 40% of the polyunsaturated fatty acids in the brain and up to 60% in the retina. Brain DHA levels are roughly 250 to 300 times higher than EPA. EPA is something else entirely. The brain barely takes EPA in, the brain incorporates EPA into phospholipids approximately 70 times more slowly than the liver does. And of what does enter the brain, between a third and a half is broken down within the first few minutes, converted to other compounds rather than incorporated into the tissue itself. This is not a new discovery. Some of those cognitive trials still treated EPA and DHA as the exact same intervention. In 2020, a paper named in Nature Communications that went further. The researchers gave mice an acute dose of EPA and tested their learning in memory. The mice were worse than controls. The cellular processes underlying memory formation in the hippocampus were impaired. The researchers then genetically engineered mice to produce elevated levels of EPA inside their own brains, the same result. Then they identified the mechanism. EPA was quieting the brain's neurons, enhancing inhibitory signaling through a specific serotonin receptor. When the team added DHA back at the ratio found in marine fish oil, the impairment disappeared through a different serotonin receptor. The paper's conclusion was specific. EPA exhibited an unexpected detrimental impact on cognitive function. Caution was required in omega-3 supplementation. The ratio of EPA to DHA mattered there. The one trial that did separate EPA and DHA, a four-arm Taiwan trial in 2020, found exactly the pattern the mouse data would have predicted. The EPA only arm worsened on specific cognitive measures. That finding was buried in a trial reported as null. In the spring of 2026, the cell reports paper landed in this literature. Mice on a fish oil-rich diet with or without repetitive subconcussive head impacts and the uninjured mice EPA had accumulated in the brain. DHA remained stable. After the head impacts, EPA drained out. The injured brain was actively pulling it from cell membranes and using it as a part of the injury response. The injured brains had switched on the genes for fatty acid metabolism and switched off the genes for building new blood vessels. They were burning lipids and starving the repair process at the same time. The damage that followed was vascular, thickened cerebrovascular walls, impaired blood flow regulation, and tau accumulation around damaged vessels. At six months, the spatial learning of the injured mice was worse than controls. EPA selectively impaired the cells' ability to repair themselves. DHA did not. The author stated it directly. The cerebral accumulation of EPA constitutes a latent metabolic liability that becomes unmasked in the context of repetitive mild traumatic brain injury, not a liability in a healthy brain, a liability that emerges when the brain is undergoing injury-driven metabolic remodeling, when it shifts to burning fatty acids, and EPA gets pulled into that burn. The honest read of the first half of this episode starts with what the 2026 authors said about their own findings. The Adney paper did not conclude omega-3 as harmful. The authors wrote that the harmful association should be interpreted with caution, that it could reflect a genuine adverse effect, or it could reflect the specific conditions under which omega-3 was consumed in this particular cohort. Then they named those conditions. The first was dose. The failed cognitive trials had used doses between one and two grams of omega-3 per day. That dose was the standard. The adne authors pointed at it. In 2020, a randomized trial at the University of Southern California gave 33 older adults two grams of DHA per day for six months. They put needles into the participants' spines and pulled cerebrospinal fluid at the start and the end of the trial. The DHA in the participant's cerebrospinal fluid rose by 28%. EPA rose by 43%. The authors concluded that doses at or below 1 gram per day, the doses used in most of the failed cognitive trials, may simply not have been delivering enough omega-3 to the brain to act on it. The second condition was baseline status. The ADNI trial population was a sample of older Americans who chose to take fish oil. Nobody measured whether they actually needed it. Most of the failed trials didn't either. In the Framingham Heart study, 899 dementia-free older adults were followed for nine years. The ones in the top quartile of baseline plasma DHA had a 47% lower risk of developing dementia over the next decade. In a 2014 imaging study of more than a thousand women, those with higher baseline omega-3 had larger total brain volume and larger hippocampi when imaged eight years later. The strongest signal in the entire brain literature for omega-3 is not from supplementation in the already replete individuals. It's from carrying adequate omega-3 across decades into older age. The third condition was the supplement itself. Most participants in the ADNI cohort had used commercially available fish oil, which oxidizes more readily than the purified preparations used in earlier trials. A 2015 New Zealand study found 83% of fish oil products tested exceeded industry recommended oxidation limits. Most products on the shelf is, by industry definition, partially oxidized. So was oxidized fish oil causing harm? It sounds intuitive, but the claim is heavily contested. In 2012, a randomized trial actually fed healthy adults oxidized fish oil for seven weeks and found no difference from fresh fish oil on standard markers of oxidative stress, lipid peroxidation, and inflammation. What's more is that the Global Organization for EPA and DHA and the European Food Safety Authority both note that oxidation values in fish oil are arbitrary with respect to human health. Those values used to measure fish oil freshness, those are a regulatory threshold and not really a biological one. The mice in the cell report study that ate the fish oil-rich diet for months and didn't have head impacts, the healthy mice accumulated EPA in their brains and developed no thickened vessels, no impaired blood flow, no perivascular towel, no learning impairment, six months on the same diet, no damage. Both 2026 papers, when read carefully, used the same phrase to describe their findings. Context dependent. The adne authors used it about the aging brain. The cell reports team used it about the injured brain. Neither paper concluded that omega-3 is bad. Both concluded that what omega-3 does depends on a set of conditions. The supplement aisle has never been built to consider. Baseline omega-3 status, age, inflammatory burden, the presence or absence of underlying neuropathology, the dose, the form, and the specific molecule in the capsule. The first half of this episode is real. The trials really failed. The cognitive decline signal really does exist. The brain injury mechanism really is biologically plausible. What changes is the question those findings answer. The question the headlines posed, is fish oil bad for your brain, was never really the right question. The right question is harder because it has more than one answer. If you take fish oil, here's what 20 years of brain literature actually supports. The clear signal in the entire field is for people carrying adequate omega-3 across decades, not for people supplementing once they get worried about actual decline. So the first question is whether you need it. An omega-3 index test from a finger prick will tell you if you're already replete, the evidence for adding more is weak. And the 2026 work is a reason to be careful about adding more if you don't actually need more. If you have less than adequate omega-3 stores, then dose matters. The trials that used one gram a day or less may not have delivered enough to the brain to matter. The trial that worked used 900 milligrams of DHA. The brain delivery study used more than two grams. If you're going to supplement, the dose has to be high enough to do what you bought it for. And if you have a history of repeated head injury, contact sports, military service, anything along those lines, the 2026 mechanism work in CELT reports is the strongest reason to have a real conversation about whether high-dose fish oil is the right call. The healthy mice didn't get hurt, the injured ones did. But again, those were rodents consuming large amounts daily for months, which is years in relation to how humans age. The supplement aisle has sold omega 3 as a single decision for almost 30 years. Whether it helps you, does nothing, or harms you depends on conditions the bottle doesn't ask about, and conditions you've probably never been told to find out. Thank you for joining me on today's episode. Please subscribe on whatever channel you're listening on. Like and comment as that helps the channel. Until next time, stay healthy.