Ultra-processed food made up 41% of one study group’s diet, and it tracked with more fat inside their thigh muscles

On MRI scans of 615 people’s thighs, fat was not just sitting under the skin, it was showing up as fine streaks running through the muscle tissue itself, a pattern radiologists call fatty infiltration. Zehra Akkaya, a radiologist who splits her research time between Ankara University and the University of California, San Francisco, led the analysis, which found that people whose diets consisted of about 41% ultra-processed food, 41.4% in the study’s own figure, had consistently more of that intramuscular fat than people who ate less of it. The association held up regardless of how many total calories participants consumed, meaning the effect tracked with what people ate rather than simply how much.

Fat Streaks That Show Up Regardless of Calories

The 615 participants, 275 men and 340 women with an average age of 60 and an average body mass index of 27, were drawn from a group already being monitored for knee health. Akkaya’s team used food frequency questionnaires covering the prior year to estimate how much of each person’s diet came from ultra-processed sources, then compared that figure against thigh muscle composition visible on MRI.

What makes the finding notable is what did not explain it away. In statistical models that adjusted for body mass index, higher ultra-processed intake tracked with more intramuscular fat across the thigh muscles, a relationship the researchers reported in the journal Radiology, and one first flagged in the original coverage of the findings. The association grew stronger, not weaker, once the team adjusted for abdominal circumference instead of body mass index alone, and it persisted after accounting for total calorie intake, physical activity levels, and sociodemographic factors, the kind of adjustment researchers typically run to check whether an association is really about diet quality or just about eating more food overall. Here, it stayed intact, suggesting the type of food mattered independently of the amount.

Defining an Ultra-Processed Diet

Ultra-processed food is not simply a synonym for junk food; it is a category defined by the NOVA classification system, which groups products by how industrially altered they are rather than by their calorie or fat content. The NOVA framework, developed by nutrition researchers and now widely used in cohort studies like this one, places items made with industrial ingredients rarely found in home kitchens, along with added flavorings, colorings and emulsifiers, into the ultra-processed tier regardless of whether the finished product looks indulgent or ordinary. That distinction is why a diet built around packaged breads, flavored yogurts and reconstituted meats can register a high ultra-processed share even without resembling a stereotypical junk-food diet.

A Cohort Built to Study Knees, Repurposed to Study Muscle

The 615 participants came from the Osteoarthritis Initiative, a long-running multicenter study now in its fourteenth year that was originally designed to track knee osteoarthritis risk factors, not diet quality. Funded jointly by the National Institutes of Health and several pharmaceutical companies through the Foundation for the National Institutes of Health, the initiative has followed thousands of participants with repeated imaging and biological sampling, giving Akkaya’s team an existing archive of thigh MRIs to analyze rather than a study built from scratch.

Akkaya emphasized that the imaging method itself was chosen for accessibility rather than novelty, describing in an RSNA-distributed statement how the team paired its questions about modern diet quality with widely available, non-enhanced MRI scanning, a technique she said was practical enough for routine clinical use and future studies rather than requiring specialized research-grade equipment. That choice matters because it means a hospital without an advanced scanner could, in principle, replicate the measurement.

What the Study Cannot Yet Tell a Dieter

The design is cross-sectional, meaning Akkaya’s team measured diet and muscle fat at roughly the same point in time rather than watching either one change first. That makes it impossible to say from this data alone whether cutting ultra-processed food would reduce intramuscular fat that has already accumulated, only that the two were linked in the population studied. Akkaya framed the practical takeaway narrowly rather than broadly: “dietary quality warrants greater attention, and weight loss regimens should take into account diet quality beyond caloric restriction,” she said, stopping short of a specific dietary prescription.

Intramuscular fat matters clinically because it is associated with weaker, lower-quality muscle even in people whose weight looks unremarkable on a scale, which is part of why a knee-focused cohort ended up yielding a finding about diet composition rather than joint cartilage. Researchers studying osteoarthritis have increasingly looked past the joint itself toward the surrounding muscle, on the theory that a thigh muscle riddled with fat provides less support and shock absorption for the knee it moves, even before any cartilage damage becomes visible on a scan.

Whether the pattern Akkaya’s team found in thigh muscles also shows up elsewhere in the body, and whether it predicts osteoarthritis progression specifically over the years the Osteoarthritis Initiative still plans to track its participants, are questions the current paper leaves for the cohort’s next round of imaging to answer.

This article was produced with the assistance of AI and reviewed by Morning Overview editors prior to publication.

More from Morning Overview