Article — "Semaglutide Extends Lifespan in Aged Female Mice—Overlapping but Not Identical to Calorie Restriction": Two longevity researchers looked at the same study and reached opposite worries about the same missing number.
Danica Chen, who led it, is bothered that she can't explain why her drug beat dieting. Nir Barzilai, who didn't, is bothered that she never proved the drug beats dieting at all. "Maybe calorie restriction increases the lifespan more?" he asks. "The data is not there, and it bugs me."
That tension—two experts unsettled by the same absence—is the honest center of a paper that most readers will remember only for one seductive figure.
Two accomplished scientists arrived independently, from opposite directions, at the same missing data. It bugs them both—and that discomfort, not the 12 percent, is the finding.
Here it is: old mice on semaglutide lived about 12 percent longer, a median of 834 days against 742 for controls. Nearly a hundred extra days.
Last week this column parsed the biological-age and biomarker claims swirling around GLP-1 drugs. This week we have something harder to argue with—a real lifespan endpoint, an actual survival curve, from a study published September 2 in Nature.
And the survival curve is the least interesting thing in it.
Treating that 12 percent as evidence that GLP-1 slows human aging misreads what the researchers themselves think is worth arguing about. The lifespan number came from 20-month-old female C57BL/6 mice dosed daily until death. It tells you these particular mice lived longer. It does not tell you why—or whether the mechanism is the drug slowing aging or the drug clearing the metabolic damage of being fat and old.
The paper's real intellectual core is a separate, five-month experiment built to answer that question. Chen has studied calorie restriction for two decades, and she suspected semaglutide might simply be a chemical shortcut to the same thing: eat less, age slower, without the willpower.
So her team pitted the drug against dieting directly. One group got semaglutide. Another was fed 24 percent less—matching the appetite suppression the drug produces—so the calorie deficit was identical.
If semaglutide were just a fancy way to eat less, the two groups should have looked the same. Mostly they did. Muscle strength, coordination, most physiological measures held steady in parallel across both.
The liver told the same story. When the team sequenced gene activity and compared it to existing calorie-restriction data, the overlap was substantial: both interventions quieted inflammation and fat processing, both ramped up genes for handling blood sugar and maintaining proteins.
So far, a clean case for "it's just calorie restriction in a syringe."
Then the story breaks in one place. Metabolic rate.
Calorie-restricted mice slowed their metabolism—the classic hunger-driven adaptation, the body banking energy against scarcity. Semaglutide mice did not. Their metabolic rate stayed largely unchanged.
That is the single measurement that the "it's just dieting" narrative cannot absorb. Both groups cut roughly the same calories. One body responded by throttling down. The other didn't.
And here is what makes it more than a curiosity: the drug that skipped the metabolic slowdown still won on the things that matter. Semaglutide mice surpassed baseline in exploratory drive, spatial memory, and glucose control—not just matching the dieters but beating them, and beating their own starting point.
You would expect the intervention that conserves energy to protect the brain and metabolism better. Instead the one that kept metabolism running full-tilt scored higher on memory, curiosity, and blood sugar.
"In that sense, it's more than just a calorie restriction," Chen says.
This is the study's genuine signal, and it isn't the lifespan figure. It's the divergence—two roads to similar longevity benefits, one of which runs through metabolic suppression and one of which apparently doesn't. That's the first mechanistic hint that GLP-1 may reach aging through a route calorie restriction doesn't use, rather than acting purely as a proxy for eating less or weighing less.
Why does the drug hold this edge? "The short answer is: we don't know yet," Chen says.
That honesty is worth holding onto, because the rest of the paper invites overreach. The authors report reduced cellular senescence, improved hematopoietic and neural stem-cell function, more hippocampal neurogenesis, better mitochondrial function, greater genomic stability—a checklist of aging hallmarks moving the right way.
It reads like a drug slowing aging itself. It might be. But now apply the brakes, because the design cannot yet carry that conclusion.
First: this was one sex and one strain. Female C57BL/6 mice, full stop. Single-sex mouse longevity signals have a poor record of mapping onto humans, and the study offers no reason to assume male mice, other strains, or people would respond the same way.
Second, and this is Barzilai's specific complaint: the lifespan study never raced semaglutide against calorie restriction. The 12 percent extension came from drug versus saline. The head-to-head comparison with dieting was the five-month functional study—which measured behavior, strength, and gene expression, not death.
So the paper cannot tell you whether dieting would have extended lifespan more, less, or the same. The metabolic-rate divergence is real, but nobody has yet shown it translates into a different survival curve. The two things the field most wants to connect—the mechanistic split and the lifespan outcome—were measured in different experiments.
That experiment, Chen says, is currently ongoing. "I don't predict what the results might be. The results always surprise us."
Third: even granting everything, we can't yet separate slowing aging from cleaning up the liabilities of being fat and old. These mice lost weight, predominantly fat mass. A leaner old animal is a healthier old animal for reasons that have nothing to do with tapping a conserved longevity pathway.
The researchers tried to control for this—the behavioral gains held, they report, even after accounting for the mice simply being leaner. That's a point in favor of a genuine aging effect. It is not proof of one.
So where does this leave the field? With a sharp, answerable question, which is more than most aging studies produce.
The question is no longer "does GLP-1 do something in old mice"—it clearly does. It's whether there is a specific, calorie-independent pathway that explains why the drug keeps metabolism up while still delivering the benefits, and whether that pathway is what extends life or just what improves function along the way.
Isolating that route is the work now. Everything else—the human relevance, the dosing, the durability—waits behind it.
For readers watching the GLP-1 longevity story build, the update is narrow but real. We've moved from biomarker arguments to a hard lifespan endpoint in a mammal, and from "these drugs seem good for aging" to a testable claim about mechanism.
What we have not moved to is any statement about people. There is, in the authors' own words, no indication these results could be achieved immediately in humans, and the human evidence—a post-hoc read of the SLIM LIVER trial—remains thin.
The most useful sentence in the whole affair may be the one two accomplished scientists independently arrived at, from opposite directions, about the same missing data.
It bugs them both. It should bug the rest of us too—and that discomfort, not the 12 percent, is the finding.