Genetic variation near the receptor that the GIP half of tirzepatide acts on lined up with a slower-aging profile more consistently than variation near the GLP-1 receptor behind semaglutide. That is the headline result of a drug-target Mendelian randomization study ↗ published July 22 in Cardiovascular Diabetology by Li and colleagues.
The method matters here, because it is easy to misread. This was not a trial. Nobody was given a drug. Mendelian randomization uses common gene variants people are born with as stand-ins for a drug's effect. Variants near the GIPR gene that nudge blood sugar or body weight down a little, for life, are treated as a lifelong, low-dose imitation of a GIP-targeting drug. The same trick was run for the GLP-1 receptor gene, GLP1R, and for both together, the combination that tirzepatide hits.
The researchers then asked whether those genetic nudges tracked with four different readouts of aging: a frailty index, a blood-marker estimate of biological age called PhenoAge, telomere length (the protective caps on chromosomes that shorten with age), and plain longevity, meaning the odds of living to an old age. Genetically modeled blood-sugar lowering through GIPR lined up with all four. People with those variants were less frail, showed a younger biological age, carried longer telomeres, and had close to three times the odds of reaching old age (odds ratio 2.90, rising to 6.54 in a second dataset).
The GLP-1 side was narrower. Modeled blood-sugar lowering through GLP1R was tied to longevity alone, with a large but unstable estimate (odds ratio 3.72 in the first dataset, 16.53 in the replication, with a confidence range wide enough to distrust the exact figure). And modeling GLP1R through weight loss rather than sugar lowering was tied to none of the four aging measures. The dual GIPR plus GLP1R model looked much like GIPR on its own.
Two things in that pattern cut against the usual story. The first is that GIP, often treated as the junior partner riding along with GLP-1 in tirzepatide, carried the broader signal. The second is subtler and sharper: the aging associations came mostly from the genes' modeled effect on blood sugar, not their modeled effect on weight. The dominant public explanation for why these drugs might help people age better is that they take weight off. In this genetic model, the weight arm of GLP1R did nothing measurable, and the glucose arm did the work.
That is a provocative claim, and it deserves the caveats that come with the method. Mendelian randomization models a lifetime of small inborn tendencies, not a drug started in middle age at a pharmacologic dose, so the size of any real drug effect could be very different. The aging readouts are proxies, not outcomes anyone lived: PhenoAge is a formula over blood tests, telomere length is a noisy measurement, and a frailty index is a composite. The wildest confidence intervals here, longevity odds swinging past 16 with a ceiling above 60, are a tell that the longevity data are thin and the estimates unstable. And separating a "sugar" pathway from a "weight" pathway when the two travel together genetically is exactly the kind of thing this design does uneasily.
What it is good for is pointing. On peptidemodel, semaglutide sits on the GLP-1 receptor ↗ and tirzepatide bridges GLP-1 ↗ and the GIP receptor ↗. Most of the attention, and most of the aging speculation, has ridden on the GLP-1 half. This analysis says the GIP half is the one worth a hard, dedicated look for aging biology, and that glucose control may matter more than the scale does. It is a hypothesis with a genetic address on it, not an answer. The trials that could settle it have not been run.