A sugar found only in red meat is linked to a 63% higher diabetes risk

Type 2 diabetes research has spent years circling the same broad target: red meat is linked to higher risk, and nobody is fully sure why. A study presented this week points at a specific molecule as part of the answer, one that is not a nutrient in any conventional sense and that the human body cannot make for itself.

The study

Preliminary results presented at the Annual Meeting of the European Association for the Study of Diabetes (EASD) in Milan, Italy (September 28 to October 2, 2026), led by Dr. Leopold Fezeu at Université Sorbonne Paris Nord, analyzed 89,953 adults from the NutriNet-Santé cohort, a long-running French prospective study. The average participant was 42 years old, 79% were women, and none had type 2 diabetes or cancer at the start of the study. Each person completed three 24-hour dietary records, and the researchers estimated daily intake of N-glycolylneuraminic acid, or Neu5Gc, by matching those records against a database of the compound's content in food. Participants were then split into intake quintiles, calculated separately for men and women, and followed forward for new diagnoses of type 2 diabetes.

This is a conference presentation of preliminary results, not a peer-reviewed published paper. The authors themselves describe it that way, and it should be read with that qualifier attached to every number below.

The finding

Adults in the highest quintile of meat-derived Neu5Gc intake had a 63% higher risk of developing type 2 diabetes than those in the lowest quintile: a hazard ratio of 1.63 (95% CI 1.29 to 2.06). The researchers found no comparable association for Neu5Gc from dairy, and total Neu5Gc intake tracked the outcome mainly through the red-meat-derived portion rather than the dairy-derived portion.

The finding that keeps this honest rather than alarmist: when the analysis additionally adjusted for how much red meat participants ate overall, the association between Neu5Gc and diabetes pointed the same direction but was no longer statistically significant. The researchers frame this as evidence that Neu5Gc may account for part of the established link between red meat and type 2 diabetes risk, not as proof that the sugar itself, independent of red meat, drives the risk.

What this design can and cannot tell you

This is an observational cohort measured through self-reported diet records, not a trial where people were assigned to eat more or less Neu5Gc. Diet is measured with error, quintile cutpoints are somewhat arbitrary, and people who eat the most red meat differ from people who eat the least in a great many ways a dietary record cannot capture. The loss of statistical significance after adjusting for total red meat intake is itself an important limit: it means the current data cannot cleanly separate "Neu5Gc specifically" from "red meat in general" as the operative exposure.

None of that makes the finding unimportant. It is one of the first large studies to test a specific, mechanistically plausible reason red meat might raise diabetes risk, rather than simply reporting that the association exists. But "linked to" is the right verb here, not "causes," and a preliminary conference abstract carries less weight than a published, peer-reviewed paper until it becomes one.

Why a molecule the body cannot even make would matter

Neu5Gc is not a nutrient. It has no known biological function in a human body and no dietary requirement, in either direction. Most mammals synthesize it on their own cell surfaces using an enzyme encoded by the CMAH gene. Humans carry a broken copy of that gene, the result of an ancient mutation, so a human body makes none of its own.

When Neu5Gc from an animal food is absorbed, some of it gets incorporated into the surface of human cells anyway, where the immune system recognizes it as foreign. The body builds antibodies against it, essentially all humans who eat red meat carry some level of these antibodies, and the resulting low-grade, chronic immune activation has been studied under the name xenosialitis. Earlier research, including a widely cited 2015 study in the Proceedings of the National Academy of Sciences, linked this mechanism to inflammation and tumor progression in mouse models, and separate human cohort work has linked circulating anti-Neu5Gc antibodies to cancer risk. Type 2 diabetes is a newer addition to that list, not an established one.

Which foods actually carry it

Neu5Gc content varies a great deal by species, cut and study, in the way a lab-measured compound typically does rather than the way a labeled nutrient does. Ordered by how much a food generally carries:

• Beef: the red meat measured highest, with wide variation by cut and cattle breed in the studies that have measured it
• Pork: measured lower than beef on average but still well above other animal foods, with its own wide range by cut
• Lamb: measured lower than beef and pork in the most-cited food survey, though still clearly present
• Organ meats (heart, liver, kidney): measured higher than the same animal's skeletal muscle in the studies that have compared them directly
• Cow's milk and other dairy: present, but at meaningfully lower levels than red meat
• Yogurt and cheese: present at levels similar to or somewhat below fluid milk
• Caviar and fish roe: one of the few non-mammal foods where it has been detected
• Poultry (chicken, turkey): effectively none in the studies that have measured it
• Fish (excluding roe): effectively none
• Eggs: effectively none
• Plant foods of any kind: none; Neu5Gc is made only by non-human mammals

That list is qualitative rather than a table of milligrams per serving on purpose. Published measurements of Neu5Gc content span wide ranges even within the same meat type, depending on the animal, the cut and how the food was prepared, closer to how heavy-metal contamination varies by batch than to how a vitamin's content holds steady across a food category. A single confident number per serving does not exist in the literature the way it does for iron or vitamin C, so this piece is not going to invent the appearance of one.

Is there a safe amount?

There is no RDA, AI or upper limit for Neu5Gc, because it is not a required nutrient in the first place, unlike vitamins and minerals, less is not a deficiency risk in any direction. The honest summary of where the evidence stands: red meat and, on this preliminary evidence, specifically its Neu5Gc content, tracks with higher long-term risk of type 2 diabetes, on top of previously studied associations with inflammation, cardiovascular disease and cancer. The practical lever is the same one general dietary guidance already points at: eating less red meat, and less organ meat specifically, lowers Neu5Gc exposure alongside everything else red meat brings with it. This study does not establish a specific gram threshold below which the risk disappears.

Too little, and too much

Too little is not a real category here. Because Neu5Gc has no biological function in humans, there is no deficiency to develop and no reason to seek it out. A diet with none of it is not missing anything.

Too much is where the open questions sit. The immune response to ingested Neu5Gc, and the chronic inflammation researchers call xenosialitis, has been studied in connection with cancer and cardiovascular disease for over a decade, and this preliminary result adds type 2 diabetes to that list pending peer review and replication. None of this is a claim that any specific amount of red meat is dangerous; it is a candidate explanation for an association that was already established before anyone had identified Neu5Gc as a possible mechanism.

Tracking it in AIM

Neu5Gc itself cannot be estimated from a photograph, in the same way heavy metal content cannot: it is a lab-measured concentration that varies by cut, animal and preparation in ways no image contains, and there is no established daily target to check it against in the first place. What a photo-based log can do honestly is what it already does for red meat: track the amount eaten. Add "red meat" or "organ meat" as a custom metric, in ounces or grams, with no target set, or a personal ceiling if you want one, and watch how the week actually adds up rather than guessing.

Estimates from a photograph carry real uncertainty for micronutrients, because content varies with soil, cultivar, storage and cooking in ways no image contains. That uncertainty is larger here than usual, since Neu5Gc content is not tracked in standard food composition databases at all. Treat this as a reason to watch total red meat intake, not as a number AIM can put in front of you.