Older adults who ate closer to the MIND diet carried less of one blood protein, and people with more of it were likelier to get dementia over the next 21 years

The MIND diet has a simple pitch. Take the Mediterranean diet and the blood-pressure diet known as DASH, keep the parts with the best record for the brain, and score people on how closely they follow it: points for green leafy vegetables, other vegetables, nuts, berries, beans, whole grains, fish, poultry, olive oil and a little wine; points for keeping red and processed meat, butter and margarine, cheese, pastries and sweets, and fried or fast food down. For a decade, people with higher scores have kept showing up with slower memory decline and less dementia. What nobody can yet say is how a plate of spinach and salmon would reach the brain.

A team at Indiana University School of Medicine, working with researchers at Johns Hopkins, went looking in the blood. They took 206 people aged 60 and over from the Indiana Alzheimer's Disease Research Center, some with normal thinking, some with mild cognitive impairment and some with Alzheimer's disease, who had filled in a 15-question MIND diet screener and given a blood sample within 30 days of their clinic visit. The sample was run on a platform that measures about 7,000 proteins at once. The paper came out in Alzheimer's & Dementia on 9 October 2026.

The first finding was the familiar one. People with Alzheimer's had the lowest average MIND scores, 8.8 out of 15, against 10.0 for those with normal thinking, and higher scores went with lower odds of an Alzheimer's diagnosis (odds ratio 0.75) after allowing for age, sex, body weight and the APOE4 gene. Higher scores also went with slightly lower neurofilament light, a blood marker that rises when nerve cells are damaged. But diet scores showed no link with the hallmarks of Alzheimer's itself: not with amyloid on brain scans, not with the size of the hippocampus, and not with blood tests for amyloid, tau or a marker of brain inflammation. That is worth noticing. Whatever the diet is doing, in this group it was not showing up in the classic Alzheimer's markers.

To sift 7,000 proteins without drowning in chance findings, the team grouped proteins that rise and fall together into clusters, picked the most connected proteins in the two clusters that tracked diet, and asked which of those also tracked Alzheimer's. Three came through. The two with the strongest case were cytidylate kinase, an enzyme involved in making the building blocks of DNA and RNA, and MCTS1, a protein mostly studied in cancer. In both cases, higher diet scores went with lower levels, and higher levels went with higher odds of Alzheimer's. A statistical model put roughly 38% of the diet's link to Alzheimer's through cytidylate kinase and 21% through MCTS1, but both estimates come with ranges so wide they include zero, and the authors say so.

So they checked elsewhere. In the Atherosclerosis Risk in Communities study, 10,228 middle-aged and older Americans who reported their diet and gave blood in the 1990s and were then followed for a median of 21 years, people with higher MIND scores again had lower MCTS1, and people with more MCTS1 at the start were more likely to be diagnosed with dementia later (about 6% higher risk per standard step up in the protein). In a large international pool of 3,723 people from memory studies, higher MCTS1 again went with Alzheimer's (odds ratio 1.36). Cytidylate kinase was not measured in the long US study and was only borderline in the international pool, so it stays a lead rather than a finding.

Why might MCTS1 matter? The authors point to a recent UK Biobank analysis in which people with more MCTS1 in their blood had more white matter hyperintensities, the bright spots on brain scans that mark wear in the brain's smallest blood vessels and are a common contributor to dementia of all kinds. Diets like MIND and the Mediterranean diet have been linked to fewer of those spots. That gives a plausible story, and one that fits the null results on amyloid and tau: the diet may matter more for the brain's plumbing than for the proteins that define Alzheimer's.

Now the cautions, and there are several. Everything here is observational, so people who eat well may differ in many other ways. Diet was self-reported, which is harder for someone whose memory is failing. Diet and blood were measured at the same visit, so the study cannot show the diet changed the protein rather than the other way round. The Indiana group is small and mostly well educated (about 16 years of schooling on average). The long US study could not score berries or olive oil, and in that group the MIND score's overall link with dementia was too uncertain to call, which is why the authors present MCTS1 as supporting evidence for a pathway, not proof that the diet works through it. Some Indiana participants are also in the international pool and could not be removed. And a protein in the blood may not reflect what is happening in the brain.

What can you do with this? Not much new, and that is fine. Nobody should be testing their MCTS1 or chasing a supplement for it; there is no such test in clinics and no evidence that lowering it helps. The practical message is the one the diet already carries: plenty of leafy greens and other vegetables, beans, nuts, berries, whole grains and fish, and less of the fried, the processed and the sugary. What this study adds is an early hint that part of the benefit, if it is real, may run through the health of small blood vessels in the brain, the same vessels that blood pressure, smoking and diabetes wear down.