Metformin, Sulfonylureas, and the Long Walk to 90: What a New Target-Trial Study Actually Shows

The study, published in 2025 by Shadyab and colleagues, asks a narrower and more answerable question than "does metformin extend life?" It asks whether women who started metformin for newly diagnosed type 2 diabetes were more likely to reach age 90 than women who started a sulfonylurea — an older class of glucose-lowering drugs — for the same reason. The researchers used the Women's Health Initiative, a long-running cohort, and applied a framework called target trial emulation, which tries to mimic the discipline of a randomized trial inside observational data.

Most of what we think we know about metformin and longevity comes from messier comparisons: metformin users versus non-users, or versus people without diabetes at all. Those comparisons are riddled with what epidemiologists politely call confounding. People prescribed metformin differ from people who aren't, in ways that have nothing to do with the drug. Target trial emulation tries to clean that up. You specify, in advance, the trial you wish someone had run — who would be eligible, when the clock starts, what the comparison is — and then you build that trial out of real-world records.

Here, the emulated trial enrolled women aged 60 and older with newly diagnosed type 2 diabetes and no prior use of glucose-lowering drugs or insulin. The comparison was metformin monotherapy versus sulfonylurea monotherapy, started at diagnosis. The endpoint was survival to age 90 — what the authors call exceptional longevity. To balance the two groups on the variables that usually muddy these waters — age, lifestyle, body mass index, blood pressure, cardiovascular disease, COPD, cancer, other medications — the investigators used 1:1 propensity score matching, leaving 438 well-matched women.

The headline number: women who initiated metformin had a 30 percent lower adjusted risk of dying before 90 than those who initiated a sulfonylurea, with a hazard ratio of 0.70 and a confidence interval (0.56 to 0.88) that doesn't brush against the null. The incidence of death before 90 was 3.7 per 100 person-years in the metformin group versus 5.0 in the sulfonylurea group. That is a real, consistent gap.

Read carefully, this is a comparison between two drugs, not a verdict on metformin in isolation. Sulfonylureas have their own baggage: they can cause hypoglycemia, they have been linked in some analyses to higher cardiovascular risk, and they don't share metformin's modest weight and metabolic benefits. Part of what looks like a metformin advantage may be a sulfonylurea disadvantage. The authors are upfront that this is a comparative effectiveness question, not a clean test of geroprotection.

The study is also confined to women, all of whom had type 2 diabetes, and the matched sample — 438 — is modest. Target trial emulation reduces confounding; it does not abolish it. Unmeasured differences between the groups, such as adherence patterns or subtle differences in disease severity at the moment of prescribing, can still leak into the result. And the endpoint, survival to 90, is shaped by everything that happens in the decades before it: cardiovascular disease, cancer, falls, dementia. Metformin's fingerprints could plausibly be on several of those, or on none.

Still, the direction of the finding lines up with a broader, if uneven, literature suggesting metformin users tend to do somewhat better on hard outcomes than users of older second-line agents. What this paper adds is methodological seriousness applied to an unusually demanding endpoint. Reaching 90 is not a soft proxy. It is the thing itself.

A predictable question follows: should a man in his sixties, in reasonable metabolic shape, ask his doctor for metformin as an anti-aging measure? Nothing in this study answers that. The trial it emulates enrolled women with incident type 2 diabetes. It tells us about choices made at the point of diabetes diagnosis, not about preventive use in healthy adults. The large randomized trial designed to address the broader question — TAME, Targeting Aging with Metformin — has been discussed for years and has not yet delivered results. Until it does, off-label use for longevity remains a bet placed on indirect evidence.

None of which makes this paper unimportant. For the millions of older adults already weighing first-line diabetes therapies, a 30 percent relative reduction in the risk of not reaching 90 — even with all the caveats — is the kind of number that belongs in the conversation with a clinician. And for the broader geroscience field, it is a useful data point: rigorously framed, modestly sized, pointing in a direction worth pursuing with a real trial.

The long walk to 90 is paved with many small decisions, most of them not pharmacological. Sleep, strength, movement, what's on the plate, who's at the table. A diabetes drug, even a good one, is one tile in a much larger mosaic. But it is worth knowing which tiles are holding up under scrutiny. This one, for now, still is.