The Gale readResearch

Ultraprocessed food linked to 40% higher odds of brain injury on MRI

A Canadian study of 8,783 middle-aged adults links the highest quartile of ultraprocessed food intake to more body fat and to MRI-defined vascular brain injury, but not to worse memory or thinking test scores, and the brain finding shifted with how the food was defined.

By Gale Staff · October 19, 2026 · JAMA Network Open

Not reviewed by a clinician.

Illustration: Brain health & aging

The short answer

In 8,783 Canadian adults aged 35 to 69 without cardiovascular disease, those in the highest quarter of ultraprocessed food intake had 1.7 percentage points more body fat (95% CI 1.2 to 2.2), 5.7 mL more visceral fat (3.2 to 8.3) and 40% higher odds of MRI-defined vascular brain injury (adjusted OR 1.40, 1.10 to 1.77) than those in the lowest quarter, with no difference in liver fat and no association with two cognitive test scores. This was a cross-sectional study, so it shows what travels together at one point in time and cannot say that the food caused any of it, and the authors report that the brain-injury link was sensitive to how ultraprocessed food was defined.

Who to see

Primary care clinician

Adults eating the most ultraprocessed food had more body fat and more signs of vascular brain injury on MRI; a primary care clinician can talk through your diet and health.

Find care →

The headline reads like a warning about the brain. What does it actually show?

A person reads that ultraprocessed food is linked to brain injury and then asks the practical questions in order. Is this about dementia? Does it mean the brain is already damaged? Is it the food, or the weight that tends to come with it? The paper, published online on 15 October 2026 in JAMA Network Open by Russell de Souza, Sonia Anand and colleagues, answers some of those questions and leaves others open, and the gap between the two is the more useful part of the story.

What it measured was not dementia and not memory loss. It looked at the body, at the brain on an MRI scan, and at two short cognitive tests, in people who were middle-aged and had no known cardiovascular disease. The brain finding concerns what the authors call vascular brain injury, which they define as covert brain infarcts or high white matter hyperintensities on MRI. Covert here means the scan shows it although the person has not had a recognised stroke. The associations are modest, they are statistical, and one of them depends on a definitional choice that the authors flag themselves.

What the researchers did

The study is cross-sectional: every measurement was taken in the same snapshot, so none of it can show which came first. It pooled 8,783 adults aged 35 to 69 from two Canadian research programmes, the Canadian Alliance for Healthy Hearts and Healthy Minds (CAHHM) and Prospective Urban and Rural Epidemiology-MIND (PURE-MIND). Participants in CAHHM were recruited between 2014 and 2018, and those in PURE-MIND between 2010 and 2018. The mean age was 57.7 years (standard deviation 8.7), and 4,917 participants, or 56.6%, were female.

Ultraprocessed food intake came from semiquantitative food frequency questionnaires that the authors describe as previously validated, expressed as servings per 2,000 kilocalories a day. That scaling matters: it compares what share of a person's diet is ultraprocessed, not how much they eat in total. Participants were then split into four equal groups, or quartiles, by intake.

The outcomes were measured directly. Body fat percentage came from bioelectrical impedance, which estimates fat from how an electrical current passes through the body. Visceral adipose tissue volume, the fat packed around the organs, and the fat fraction of the liver were also measured, and the brain findings were defined from MRI. Cognition was assessed with the Montreal Cognitive Assessment (MoCA) and the Digital Symbol Substitution Test (DSST), two brief screens of thinking and processing speed. The data were analysed between July 2024 and August 2026.

What they found

Comparing the highest quartile of intake with the lowest, body fat percentage was 1.7 percentage points higher (95% confidence interval 1.2 to 2.2). Visceral fat volume was 5.7 mL greater (3.2 to 8.3). Both intervals sit clear of zero. Liver fat fraction showed no significant difference.

For the brain, the highest quartile had higher odds of vascular brain injury, with an adjusted odds ratio of 1.40 (95% CI 1.10 to 1.77, P for trend 0.009). An odds ratio of 1.40 means the odds were 40% higher in the top group than in the bottom group. The abstract does not give how many people in each group had the injury, so the absolute difference cannot be calculated from what was published there, and a 40% rise in odds can be a small or a large change in the number of people affected depending on how common the condition is to begin with. The P for trend suggests the pattern ran across the four groups rather than appearing only at the extreme.

On thinking, the result was null. No significant association appeared between ultraprocessed food intake and either the MoCA or the DSST. The authors' own summary is that higher intake was associated with greater adiposity and higher odds of MRI-defined vascular brain injury, but not with cognitive performance.

The caveat that carries the most weight

The authors state that the association with vascular brain injury was sensitive to the definition of ultraprocessed food. That is a plain admission that the brain result moved when the way of classifying foods moved. There is no single agreed line between a processed food and an ultraprocessed one, and food frequency questionnaires lump many products together, so a finding that depends on where the line is drawn is a weaker finding than one that survives every reasonable drawing of it. The abstract does not say which definitions weakened the result or by how much, so the size of that sensitivity is not something this account can supply.

Two further limits come with the design. Because everything was measured once, the study cannot rule out that people who already carry more body fat, or who are on a different health path, eat differently, and it cannot tell whether the food, the extra body fat or something shared by both sits behind the brain scans. The authors close by calling for longitudinal studies to establish temporality, which is the same point made by the people who ran the study. Self-reported diet adds its own error as well, since a questionnaire records what people remember and are willing to report.

Why it matters anyway

The study is useful for what it separates. Body fat and visceral fat tracked with ultraprocessed food intake, which fits the authors' framing that such foods are consistently associated with adiposity and cardiometabolic disease. The brain signal is the newer and less certain piece. The authors describe it as subclinical, meaning the scan shows it before any symptoms, which is why it is of interest in midlife.

The cognitive null is worth holding next to the scan finding rather than behind it. In a group averaging 57.7 years, brain changes visible on MRI did not yet show up as a difference on the two tests used. That is compatible with the injury being too early or too small to affect these screens, and equally compatible with the scan association being weaker than it looks. The study cannot choose between those readings, and a longer follow-up in the same participants would be the way to find out.

What this study can't tell you

  • Whether ultraprocessed food causes more body fat or vascular brain injury: this is a cross-sectional study, so exposure and outcomes were captured at the same time and the order of events is unknown.
  • How many people in each intake group had vascular brain injury: the abstract reports an adjusted odds ratio but not the counts, so absolute risks and the size of the real-world difference cannot be given.
  • Which definitions of ultraprocessed food weakened the brain association, and by how much: the authors say the finding was sensitive to definition, and the abstract does not detail the alternatives.
  • Whether the brain finding would hold in younger or older adults, people with cardiovascular disease, or populations outside Canada: participants were aged 35 to 69, free of cardiovascular disease and drawn from two Canadian programmes.
  • Whether the diet reports were accurate: intake came from food frequency questionnaires, which depend on memory and on how foods are grouped.
  • Whether the lack of a cognitive association means no effect on thinking: only two short tests were used, in middle-aged adults, at one point in time.
  • What the study adjusted for and who paid for it: the abstract says the odds ratio was adjusted but does not list the adjustment factors or name a funder, and this account did not read the full text.

The Gale read

The paper is a measured piece of work with a result that is easy to over-read. Body fat and visceral fat are the firmest findings, with intervals that sit well away from zero, and they fit the earlier associations of ultraprocessed food with adiposity that the authors cite. The brain-injury association is the headline, yet it comes with the authors' own warning that it shifted with how ultraprocessed food was defined, and with an odds ratio whose lower bound of 1.10 sits close to no effect. The absence of any link with MoCA or DSST scores is not a footnote; it means the brain signal has not been shown here to matter for how these middle-aged adults think. Because the study measured everything once, the honest reading is that heavier ultraprocessed food intake, more fat around the organs and more silent vessel injury on MRI appeared together in this Canadian sample, and nothing in the design says one produced the others. The authors reach the same place when they ask for longitudinal studies, and that request is the right next step rather than a formality.

Common questions

does ultraprocessed food cause brain damage

This study cannot say so. It was cross-sectional, so it measured diet and brain MRI findings at one point in time. People in the highest quarter of ultraprocessed food intake had 40% higher odds of MRI-defined vascular brain injury (adjusted OR 1.40, 95% CI 1.10 to 1.77), but the authors report that the association was sensitive to how ultraprocessed food was defined and call for longitudinal studies to establish temporality.

is ultraprocessed food linked to belly fat

In this study of 8,783 Canadian adults aged 35 to 69, the highest quartile of intake had 5.7 mL more visceral fat (95% CI 3.2 to 8.3) and 1.7 percentage points more total body fat (1.2 to 2.2) than the lowest quartile. Liver fat did not differ significantly. It is an association in a single snapshot, not proof that the food produced the fat.

does ultraprocessed food affect memory and thinking in middle age

Not in this analysis. Neither of the two cognitive tests, the Montreal Cognitive Assessment and the Digital Symbol Substitution Test, showed a significant association with ultraprocessed food intake in adults averaging 57.7 years, even though the MRI brain-injury finding was present. The study cannot say whether cognitive differences would appear later.

Sources

  1. 1.de Souza RJ, Schulze KM, Desai D, Despres JP, Dummer TJB, Friedrich MG, Hicks J, Ho V, Lear SA, Lee DS, Mente A, Miller V, Poirier P, Smith EE, Tardif JC, Teo KK, Vena JE, Yusuf S, Anand SS. Ultraprocessed Food Intake and Adiposity, Vascular Brain Injury, and Cognition in Middle-Aged Adults. JAMA Network Open. 2026;9(10):e2639118. doi:10.1001/jamanetworkopen.2026.39118 link

Read the study →

1 sources, numbered by first appearance. General health information, not medical advice. AI-assisted editorial content — every citation independently verified. Editorial policy

More from Gale Staff

The Latest Health Research →

Published October 19, 2026 · Gale Staff

All Gale Staff analyses →