Canagliflozin (Invokana) is an FDA-approved drug for type 2 diabetes that works by making the kidneys dump glucose into urine instead of reabsorbing it. That mechanism has earned it real, well-established cardiovascular and kidney benefits in large human trials — and, separately, a replicated lifespan extension in male mice in the National Institute on Aging's flagship testing program. Those are two different bodies of evidence, and the gap between them is the whole story: canagliflozin is a genuinely effective, genuinely risky prescription drug being taken off-label by some longevity enthusiasts on the strength of a mouse result alone. Here is what is actually known, and what is not.
What canagliflozin actually does
Canagliflozin belongs to the SGLT2 inhibitor class (sodium-glucose cotransporter-2 inhibitors), alongside empagliflozin and dapagliflozin. The kidney normally filters glucose out of the blood and then reclaims almost all of it back into circulation through transporters in the proximal tubule, chiefly SGLT2. Canagliflozin blocks that transporter, so a meaningful share of filtered glucose is never reabsorbed and instead leaves the body in urine — a controlled, pharmacologic glucosuria. The result is lower blood glucose without depending on insulin at all, plus a modest calorie loss (the excreted glucose is calories the body does not keep) and mild fluid loss from the accompanying osmotic diuresis.
On that basis canagliflozin is approved as an adjunct to diet and exercise to improve glycemic control in type 2 diabetes, to reduce the risk of major cardiovascular events in adults with type 2 diabetes and established cardiovascular disease, and to reduce the risk of end-stage kidney disease and related outcomes in adults with type 2 diabetes and diabetic nephropathy.[5] Those three indications were not assumptions — each is backed by a dedicated outcomes trial, which is more clinical-trial support than most longevity compounds will ever have.
The human trial record: CANVAS and CREDENCE
The CANVAS Program pooled two trials totaling just over 10,000 people with type 2 diabetes and high cardiovascular risk, followed for a mean of roughly 3.6 years. Canagliflozin reduced the composite of cardiovascular death, non-fatal heart attack, and non-fatal stroke compared with placebo, and also cut hospitalization for heart failure and slowed progression of kidney disease.[3] The same program, however, is where canagliflozin's most consequential safety signal showed up: participants on canagliflozin had roughly double the rate of lower-limb amputation — mostly of the toe or forefoot — compared with placebo.[3]
CREDENCE, a dedicated kidney-outcomes trial in over 4,400 people with type 2 diabetes and established diabetic kidney disease, followed patients for a median of about 2.6 years and found canagliflozin lowered the risk of the composite endpoint of kidney failure, doubling of serum creatinine, and cardiovascular or renal death.[4] Together, CANVAS and CREDENCE are the reason canagliflozin's cardiovascular and renal benefits in diabetes are considered settled, not speculative. Neither trial measured, or was designed to measure, all-cause lifespan in people without diabetes — which is the population the longevity claim would actually need to be tested in.
The mouse lifespan signal, read honestly
The finding that put canagliflozin on the geroscience radar comes from the NIA Interventions Testing Program (ITP), the same multi-site, genetically heterogeneous mouse program that produced the acarbose result — see our explainer on how the ITP is designed to resist false positives. Started on a cohort of mice at seven months of age, canagliflozin extended median lifespan by about 14% and maximum lifespan by roughly 9% — but only in male mice; females on the same dose showed no significant lifespan benefit at all.[1] A follow-up study from an overlapping group looked at aged mice and found the same lopsided pattern in a different domain: canagliflozin improved insulin sensitivity in the brain and measures of exploratory and locomotor activity in aged male mice, again with no equivalent benefit in females.[2]
Two things about that result deserve equal weight. First, it is a real, non-trivial effect size in a study design built specifically to catch flukes — that is worth taking seriously as a hypothesis. Second, it is mouse data, and sex-skewed mouse data at that. A drug whose lifespan benefit shows up in males and disappears in females is a mechanistic clue, not a detail to round away, and it is one more reason a mouse hit does not translate automatically into a human recommendation — the same caution that applies to acarbose's ITP result and to metformin's still-unproven TAME hypothesis.
Why an SGLT2 inhibitor might touch aging
The leading mechanistic idea borrows from caloric restriction. By forcing out several hundred calories a day as excreted glucose and by mildly shifting metabolism toward fat oxidation and ketone production, canagliflozin produces a metabolic state that looks, in some respects, like a pharmacologic echo of eating less — without actually eating less.[1] That kind of caloric-restriction-mimetic profile is a recurring theme among the ITP's few repeatable hits. It is a plausible story for why the drug might matter for aging biology, not a proven pathway, and it does not explain the sex difference, which remains unresolved.
The real safety profile
This is not a supplement with a mild side-effect list. Canagliflozin's FDA label carries several warnings that matter for anyone considering it outside a diabetes indication: diabetic ketoacidosis, including the “euglycemic” form where ketoacidosis develops with near-normal blood sugar and can be missed; genital mycotic infections, which are common; necrotizing fasciitis of the perineum (Fournier's gangrene), a rare but life-threatening infection flagged across the whole SGLT2 inhibitor class; and volume depletion, which can cause symptomatic hypotension, especially in older adults, people on diuretics, or anyone with impaired kidney function.[5] The lower-limb amputation signal from CANVAS led the FDA to add a boxed warning in 2017; newer trial data lowered the estimated magnitude of that risk and the boxed warning was removed in 2020, but the label still lists amputation as a warning to monitor for, particularly in people with prior amputation, neuropathy, or peripheral vascular disease.[3][5]
None of that makes canagliflozin an unreasonable drug — millions of people take it safely for diabetes under medical supervision. It does mean the risk side of the ledger is real, monitorable, and not optional reading for anyone weighing it against a 14% mouse result.
Should you take canagliflozin for longevity?
Be clear-eyed about what would actually be happening: taking a prescription drug, off-label, for an indication with zero human trial evidence, on the strength of one line of mouse data that did not even replicate in females. That is a fundamentally different risk calculus than trying an over-the-counter supplement with a similarly thin evidence base — canagliflozin can cause ketoacidosis, serious infection, and volume depletion in people who do not have diabetes, and those risks do not pause because the reason for taking it changed. Anyone seriously considering it needs a prescriber willing to order it for that purpose, baseline and follow-up labs (kidney function, electrolytes, ketones if symptomatic), and an honest conversation about the fact that the human longevity evidence does not yet exist. This is a decision for a physician relationship, not a purchase.
The honest bottom line
Canagliflozin is a well-studied, genuinely effective diabetes drug with real cardiovascular and kidney outcome data behind it, and a striking, replicated, male-specific lifespan signal in NIA mice that makes it a legitimate object of geroscience interest. It is not a proven human longevity drug: there is no human lifespan or healthspan trial, the mouse benefit is sex-restricted for reasons that are not understood, and the drug's real safety profile — ketoacidosis, serious infection risk, and a historical amputation signal — is not something to take lightly for an unproven indication. The defensible 2026 reading is the same one that applies to acarbose: a credible hypothesis worth watching closely in aging research, not a verdict to self-prescribe around.
This article is general information, not medical advice, and it is not an endorsement of off-label use. Canagliflozin is a prescription-only medication with serious potential side effects, including diabetic ketoacidosis, genital and other serious infections, volume depletion, and amputation risk. Any decision to use it — for diabetes or for any other reason — should be made with a licensed prescriber who can order appropriate monitoring and weigh your individual medical history. Do not start, stop, or repurpose this medication without medical supervision.