Mice on the longevity diet ate as many calories as any other group and lost fat anyway—without losing lean mass. That single result, from Valter Longo's team at USC, is the crack in a definition of protein quality that has stood for a century.
Here is the paradox stated flatly. The official scorecard for protein quality—the Digestible Indispensable Amino Acid Score, and leucine density behind it—says animal protein wins for aging muscle. A fresh analysis of more than 200,000 people says the biggest animal-protein eaters got more obesity and twice the rate of diabetes.
Both can be true. They measure different things.
Start with the case for animal protein, because it is real and well-argued. Two reviews from the University of Arkansas geriatrics group—one in Frontiers in Nutrition in 2024, one in the Journal of Nutrition in 2025—make it carefully.
Their claim is precise: protein quality is a food's capacity to deliver essential amino acids and stimulate protein synthesis. DIAAS captures the EAA composition and digestibility of a source. Leucine, one of nine essential amino acids, switches on the muscle-building machinery.
Older adults face "anabolic resistance"—their muscles use dietary protein less efficiently, which is why experts recommend 1.0 to 1.5 g/kg/d for people over 65, well above the 0.83 g/kg/d RDA. On that logic, high-EAA-density protein isn't a preference. It's a requirement to overcome the resistance.
And the modeling backs the muscle argument on its own terms. Diet modeling studies show EAA density and protein quality are higher in omnivorous and lacto-ovo-vegetarian diets, with whole-plant-protein diets requiring greater total protein and energy to compensate for lower quality.
So if your only endpoint is muscle protein synthesis, the Arkansas group is right. Maximize high-quality grams.
Now the turn. Muscle protein synthesis is not healthspan, and the two optima appear to diverge.
Muscle protein synthesis is not healthspan, and the two optima appear to diverge—one maximizing essential amino acids, the other keeping methionine low but sufficient.
Longo's Cell Metabolism study fed 20-month-old mice one of four diets: standard, Western, ketogenic, or a low-protein, methionine-supplemented "longevity diet" modeled on the traditional Italian and Okinawan patterns. The longevity-diet mice showed longer healthspan, less frailty, and reduced fat mass.
The lever was a single amino acid. "Modulating just a single amino acid, methionine, in the longevity diet could produce such dramatic metabolic changes," said first author Maura Fanti. Too little methionine caused frailty; too much abolished the benefit. Low but sufficient was the target.
Then the human layer, run with University of Toronto and Harvard collaborators. Across those 200,000-plus people, the highest animal-protein eaters—therefore the highest methionine and EAA consumers—had more obesity and twice the diabetes rate of those eating little to no animal protein.
The detail that should unsettle anyone reading is this: the association persisted even though the high-animal-protein group ate fewer calories and had otherwise healthier diets. You cannot wave it away as junk food or overeating riding along with the meat.
Longo's reading: "overall protein intake may be less important than specific amino acid intake." The anabolic optimum—maximize EAAs to feed synthesis—and the healthspan optimum—just enough methionine, no more—are different targets pointing in opposite directions.
The mouse data even carried a metabolic fingerprint: longevity-diet animals showed higher levels of signaling molecules including GLP-1. Read that only as a biomarker of a favorable metabolic state, not as a story about drugs. It's the diet moving the same hormone axis, downstream of amino acid composition.
Now the cautionary case study, because the muscle-building logic has already failed once in exactly the way this argument predicts. Omega-3 supplementation had a plausible mechanism for preserving lean mass. It was supposed to sensitize aging muscle to protein.
An overview of reviews in Clinical Nutrition ESPEN found no evidence that omega-3 supplementation enhances lean mass, muscle strength, or physical function—in healthy adults or clinical populations. Not a weak effect. No effect.
That is the pattern worth internalizing. A mechanistically attractive input, isolated and supplemented, was expected to build or preserve muscle. At the whole-organism level, the benefit evaporated.
Betting on isolated "muscle-building" inputs—whether an omega-3 capsule or a maximized leucine load—is the recurring mistake. The molecule works in the assay; the person doesn't get the outcome.
Which brings the whole thing to policy, and to a warning that predates Longo's paper. In 2019, David Katz, David Jenkins, and colleagues published what they called a public health case for modernizing the definition of protein quality.
Their charge: the prevailing definition has more to do with biochemistry than with the overall effects of food on human health. It rewards animal protein for its EAA profile while chronic disease—not kwashiorkor—is the actual threat in high-income countries.
The mechanism candidates aren't mysterious. Plant protein is generally low in branched-chain amino acids, and lower BCAA consumption improves metabolic health. Animal protein raises IGF-1, a growth signal linked to diabetes and heart disease even in people born with higher levels.
The scale of the observational signal is hard to dismiss. In the NIH-AARP cohort—400,000 people, the largest diet study in history—swapping just 3% of calories from animal to plant protein was associated with 10% lower mortality, and roughly double that if eggs went too.
So guidelines that keep encoding a DIAAS-and-leucine definition of quality will keep nudging people toward more meat, eggs, and dairy. The metric optimizes muscle synthesis in isolation and stays silent on the metabolic bill.
The honest caveat, stated plainly: the human evidence here is observational, and observational data cannot prove that methionine or animal protein causes the diabetes. Confounding is always the alternative story, even after adjustment.
But the mouse arm is not observational—it is a controlled feeding study where a single amino acid was the manipulated variable, and it points the same direction. That convergence is what raises this above association.
There is also a real risk on the other side. Too little methionine caused frailty in the mice. This is not an argument for protein minimalism; it's an argument for a specific composition—plant-plus-fish, with methionine low but sufficient.
The thing that would actually settle it is already named. Longo and Fanti say the next step is a controlled clinical trial of the methionine-supplemented longevity diet in humans.
Until that runs, the defensible position is not "eat more high-quality protein" and not "eat as little protein as possible." It is that source and amino acid composition—a mostly plant-based pattern with fish and a measured amount of methionine—may matter more for healthspan than maximizing high-quality grams.
The scorecard was built to answer a wartime question about survival and growth. The question most people in rich countries are actually asking is about the decades after fifty—and for that, the metric may be measuring the wrong thing.