Longevity Diet in Mice: Gains Without a Longer Life

Old mice cut body fat and improved metabolic and frailty markers, but did not live longer. Applicability to humans is unknown.

In old mice, a low-protein “longevity” diet with a little methionine added cut body fat and improved metabolic and frailty markers without cutting calories, with FGF21, GLP-1 and growth hormone implicated in the change — but it did not extend lifespan. A separate, unadjusted comparison of survey data from more than 200,000 US adults linked the highest animal-protein intake to roughly double the prevalence of type 2 diabetes, an association that cannot show cause.

What this means

In old mice, a low-protein “longevity” diet supplemented with a moderate amount of methionine cut body fat and improved metabolic and frailty markers without cutting calories, working through shifts in FGF21, GLP-1 and growth hormone — a knock-out experiment showed FGF21 was required for the fat and insulin benefits — but it did not extend the animals’ lives. A separate human figure, from an unadjusted cross-sectional look at data from more than 200,000 US adults, found about double the prevalence of type 2 diabetes in the group reporting the most animal protein; a snapshot like that cannot show cause, and the study’s own authors say it does not prove methionine is responsible. No human has been put on the mouse diet, and the study’s senior author holds equity in a fasting-diet company and related patents, interests the paper discloses.

A mouse result and a human number, told as one story

A single Cell Metabolism paper, published online on June 23, 2026, carries two findings that coverage has tended to braid into one: an intervention in mice and a survey result in people. They are not the same kind of evidence. The mouse arm is a controlled experiment; the human arm is a cross-sectional snapshot that cannot show cause.

What the mouse experiment did (animal evidence)

The experiment used UM-HET3 mice, a genetically diverse strain, starting at 20 months of age — late in a mouse’s life, roughly equivalent to a 60- to 65-year-old person. The animals ate freely; nothing restricted their calories. The researchers compared a standard diet with a Western diet, a ketogenic diet, a low-amino-acid “Longevity Diet” (LD), that diet supplemented with a moderate dose of the amino acid methionine (LDMM), and a Fasting-Mimicking Diet given as two 4-day cycles a month. A healthspan cohort of 30 female and 16–20 male mice per group was assessed within six months of starting the diets; a separate lifespan cohort used 40–44 mice of each sex per group.

The methionine-supplemented diet produced the strongest reduction in fat mass of any group (p<0.0001), while lean mass held steady. It improved glucose tolerance (p=0.002) and insulin sensitivity (p<0.0001) relative to every other group, including the standard diet. FGF21 rose sharply (p<0.0001, with an early rise by day 14), GLP-1 rose at both 14 days and 5 months, growth hormone was higher and IGF-1 lower. When the researchers bred mice that could not make FGF21, the strong fat loss and the insulin-sensitivity gains disappeared — evidence the hormone was doing real work, not merely marking it.

The benefit depended on getting the methionine dose right. A 0.3% supplement stabilized weight and raised the hormones; a 0.7% dose, matching the standard diet, reversed the diet’s effects and raised IGF-1 instead.

The diet did not extend lifespan in mice

The treatments that improved healthspan in old mice did not extend their lives. Because the diets began late in life, the authors write, the lifespan test may understate any benefit from starting younger — an idea this study did not test. The paper reports a healthspan result, not a lifespan one.

The “ate more” detail is a trend, not a result

Some coverage said the mice ate more. The paper reports only a non-significant trend toward higher food intake during the inactive phase in the methionine-supplemented group (p=0.06). A trend at that threshold is a hint to follow up, not a finding.

The human numbers are a different kind of evidence (human evidence)

The second tier is not an experiment. The authors re-analyzed cross-sectional data from three cohorts of US health professionals — 72,992 women in the Nurses’ Health Study, 92,088 in NHS II and 40,722 men in the Health Professionals Follow-up Study, together more than 200,000 people — at a single mid-follow-up time point (1998–1999). Comparing the highest with the lowest fifth of animal-protein intake, the top group had 34% more overweight or obesity, 2.61% more hypertension and about double the prevalence of type 2 diabetes (+105.82%; all p<0.01); methionine intake was 81.47% higher. The comparison was not adjusted for health and lifestyle factors — the paper says so — and cannot show that protein, or methionine, caused any of it.

the top group had 34% more overweight or obesity, 2.61% more hypertension and about double the prevalence of type 2 diabetes

Two details sharpen the caveat. The high-animal-protein group also reported a healthier profile on some measures (about 9.5% fewer calories, 35.17% less alcohol, 73.75% fewer sugar-sweetened beverages), which makes an unadjusted disease comparison harder to read, not easier. And the top plant-protein fifth showed 18.04% less overweight or obesity and 2.88% less hypertension, yet still had 65% higher diabetes prevalence than the lowest fifth — a pattern that complicates any simple “animal bad, plant good” reading.

What the human figure cannot show

Cross-sectional data describe people at one moment; they cannot establish what came first or whether diet caused the difference. The study’s own authors state that their results “do not demonstrate that methionine is responsible” and that they “cannot conclude that methionine is causing weight gain or promoting diabetes in these human studies.” The new figures also sit above the properly adjusted analysis of the same three cohorts: when Malik and colleagues analyzed this data prospectively with adjustment, the highest total- and animal-protein intakes carried 39% and 49% higher type 2 diabetes risk, and that excess shrank to 7% and 13% after adjusting for body-mass index. Reading the same data without adjustment would be expected to produce larger numbers.

Where this sits

Protein restriction and longevity is a large, active and unsettled field, dominated by animal work. Sulfur-amino-acid restriction — the broader category that includes methionine — continues to draw mechanistic mouse studies, such as a 2026 paper linking it to glucose control through liver serine synthesis. An adjacent review of protein and amino-acid restriction in aging surveys the same territory, again almost entirely in model organisms. The pattern holds: the mechanistic case for tuning amino acids is real and growing in animals, while human evidence on protein and type 2 diabetes stays observational and sensitive to adjustment. No human trial has put anyone on the specific methionine-supplemented low-protein diet this paper headlines. Some framing around the study came from a press release whose headline merged the mouse and human tiers; that framing is not evidence.

Funding and conflicts

The work was funded by National Institutes of Health and National Institute on Aging grants (GR1045540 and AG084485) and the USC Edna Jones Chair Fund, with acknowledged assistance from the Integrative Liver Cell Core at the Southern California Research Center for ALPD and Cirrhosis (5 R24AA012885). The senior author, Valter D. Longo, holds equity in L-Nutra, a medical-food company; he, T.E.M. and S.B. have filed patents related to the Fasting-Mimicking Diet at USC; USC has licensed intellectual property to L-Nutra with potential royalties; and Longo and M.F. are inventors on a USC provisional patent application covering aspects of the methods and findings. That is a substantial commercial interest in the fasting and longevity-diet space, and the paper discloses it.

Terms explained

Methionine: an essential amino acid, one of the building blocks of protein, and the nutrient the mouse diet was tuned around; it comes from dietary protein. The experiment added a moderate amount to a low-amino-acid diet, and the benefit depended on that amount.
FGF21: fibroblast growth factor 21, a hormone released mainly by the liver that influences energy use and fat metabolism. It rose on the supplemented diet, and mice engineered to lack it lost the fat and insulin benefits — a sign it was part of the mechanism, not just a bystander.
GLP-1: glucagon-like peptide 1, a gut hormone that raises insulin release after eating; it is the pathway drugs such as semaglutide amplify. Here it rose modestly on the mouse diet.
Growth hormone and IGF-1: two linked growth and metabolism signals. Growth hormone rose and IGF-1 fell on the supplemented mouse diet — an unusual split, since growth hormone usually raises IGF-1. The higher methionine dose reversed the pattern.
Healthspan vs lifespan: healthspan is how long an animal stays healthy and free of frailty; lifespan is how long it lives. The mouse diet improved the first and did not change the second.
Cross-sectional association vs prospective cohort: a cross-sectional study measures diet and disease at one moment and cannot show which came first. A prospective cohort follows people forward and can adjust for other factors, so it can support stronger — though still not causal — inferences than a snapshot.

Sources

Primary source: Fanti M, Brandhorst S, Navarrete G, Diaz A, Giuliani G, Chowdhury D, Antunes GC, Morgan TE, Dubeau L, Villani V, Perin L, Malik VS, Hu FB, Longo VD. “Methionine-supplemented Longevity Diet increases growth hormone, GLP-1, and FGF21, reduces frailty and promotes healthspan.” Cell Metabolism 2026;38(8):1579–1598.e6; published online 2026-06-23. DOI: 10.1016/j.cmet.2026.05.015 · PMID: 42335894 · PMCID: PMC13480414. Peer-reviewed study with two evidence tiers — a preclinical mouse intervention and a cross-sectional human reanalysis. Full text read via the NCBI PMC author manuscript (a pre-issue, unedited accepted manuscript, not the OA version of record); supplementary tables not read.
Landscape — the adjusted human counterpart (full text verified via PMC): Malik VS, Li Y, Tobias DK, Pan A, Hu FB. “Dietary Protein Intake and Risk of Type 2 Diabetes in US Men and Women.” Am J Epidemiol 2016;183(8):715–728. DOI: 10.1093/aje/kwv268 · PMID: 27022032 · PMCID: PMC4832052. The properly adjusted prospective analysis of the same three cohorts; its risk increments shrink markedly after BMI adjustment.
Landscape — adjacent field review (full text verified via PMC; context, not the same story): Knopf & Lamming. “The hallmarks of protein and amino acid restriction in aging and longevity.” Cell Press Blue 2026;1(5):100079. DOI: 10.1016/j.cpblue.2026.100079 · PMID: 42639474 · PMCID: PMC13501984. A review of the protein/amino-acid-restriction field, almost entirely model-organism evidence; adjacent context, not a duplicate of this study.
Landscape — recent mechanistic mouse work (abstract-level access): “Dietary sulfur amino acid restriction improves glucose homeostasis through hepatic de novo serine synthesis.” Mol Metab 2026. DOI: 10.1016/j.molmet.2026.102325 · PMID: 41644011 · PMCID: PMC12924129.
Press signal (discovery and framing only, not evidence): “Scientists found a longevity diet that helped mice eat more and lose fat.” ScienceDaily, 2026-07-08. https://www.sciencedaily.com/releases/2026/07/260708022206.htm