SGLT2 Inhibitors: The Longevity Drug With the Best Human Data

Seventy to ninety grams of glucose per day. That's what your kidneys dump into the toilet when you take an SGLT2 inhibitor. The metabolic equivalent of pouring a 24-ounce Coke down the drain every morning -- except your body is the one pouring it.

The mice that got this drug lived 14% longer. The researchers weren't studying longevity. They were studying diabetes.

Then the human trials came back. All-cause mortality dropped 32%. Heart failure hospitalizations fell by a third. Kidneys held up better. And a 2024 paper in Nature Aging showed that the drug was clearing senescent cells -- not by killing them, but by stripping their immune camouflage and letting T cells finish the job.

The longevity community, meanwhile, is still arguing about NMN dosages.

What SGLT2 inhibitors actually do

SGLT2 stands for sodium-glucose co-transporter 2. It's a protein in your kidneys that reabsorbs glucose from urine back into your blood. Block it, and glucose stays in the urine. You pee out 70-90 grams of sugar per day -- roughly 280-360 calories that never get metabolized.

The FDA approved the first one, canagliflozin (Invokana), in 2013. Empagliflozin (Jardiance) and dapagliflozin (Farxiga) followed in 2014. They were designed to lower blood sugar in type 2 diabetics. Simple mechanism, predictable effect.

What wasn't predicted: the drugs started showing benefits that had nothing to do with glucose control.

The ITP data

The NIA's Interventions Testing Program is the gold standard for aging research. Genetically heterogeneous mice, three independent sites, rigorous protocols. Most compounds tested do nothing. The ITP is where longevity hopes go to die.

Canagliflozin didn't die.

Miller et al., 2020: male mice given canagliflozin from 7 months of age showed a 14% median lifespan extension (p < 0.001) and 9% extension at the 90th percentile. Results replicated across all three sites -- Jackson Lab, University of Michigan, UT Health San Antonio. The effect held across all three sites.

One catch: females got zero benefit. The likely explanation is pharmacokinetic, not biological. Female mice had 20-fold higher plasma drug concentrations than males at the same dose. What looks like a sex difference might actually be a dosing problem -- females may have been getting a toxic exposure rather than a therapeutic one.

A late-start study confirmed the pattern. Canagliflozin begun at 16 months -- roughly equivalent to a human in their late 50s -- still produced 14% lifespan extension in males. Starting late didn't erase the benefit.

Here's where the ITP comparison gets interesting:

Rapamycin: +22% males, +28% females. The king. Acarbose: +22% males, +5% females. 17-alpha-estradiol: +19% males, no female effect. Canagliflozin: +14% males, no female effect. Rapamycin + acarbose combo: +37% males, +28% females. The best ITP result ever recorded.

Canagliflozin sits fourth. Not the biggest number. But the ranking changes completely when you ask a different question.

The human data advantage

Rapamycin has 47 mouse studies and a handful of small human trials. SGLT2 inhibitors have something rapamycin will never have: massive randomized controlled trials in tens of thousands of patients, run for years, measuring hard endpoints like death.

EMPA-REG OUTCOME (empagliflozin, 7,020 patients): 38% cardiovascular mortality reduction. 35% fewer heart failure hospitalizations. 32% reduction in all-cause mortality. The all-cause number is the one that matters for longevity. These people didn't just have fewer heart attacks -- they died less from everything.

CANVAS (canagliflozin, 10,000+ patients): 14% CV event reduction, 40% slower kidney function decline, 27% reduction in albuminuria.

EASEL (25,000+ real-world patients): 43% all-cause mortality reduction compared to matched non-users.

These trials were designed for diabetes and cardiology endpoints. The longevity signal fell out of the data uninvited. Nobody was looking for it. That makes it more credible, not less -- you can't p-hack a finding you weren't hunting for.

James O'Keefe, cardiologist at Saint Luke's Mid America Heart Institute, put it plainly: "This is the first therapy with a large amount of clinical data in humans to suggest that it may indeed slow the pace of aging." He and Nir Barzilai -- the researcher running the TAME metformin trial at Einstein -- co-authored a 2023 review that ranked SGLT2 inhibitors as the #1 most promising geroscience drug. Ahead of metformin. Ahead of rapamycin.

The reasoning: rapamycin has better mouse data. SGLT2 inhibitors have incomparably better human data. And in medicine, human data wins.

The senescence mechanism (this is where it gets wild)

For years, the assumption was that SGLT2 inhibitors worked through caloric restriction mimicry. Dump glucose, trigger AMPK activation, suppress mTOR and insulin-IGF1 signaling -- the same pathways that fasting and rapamycin hit. Reasonable model. Probably partially true.

Then Katsuumi et al. published in Nature Aging in 2024 and blew the model open.

They found that canagliflozin clears senescent cells. But not the way senolytics like dasatinib + quercetin work (direct killing). The mechanism is immunological.

Here's the sequence:

Canagliflozin increases AICAR levels, which activates AMPK. AMPK activation downregulates PD-L1 on senescent cells. PD-L1 is an immune checkpoint protein -- the same molecule that cancer cells use to hide from the immune system. When PD-L1 gets stripped, CD8+ T cells and NK cells can suddenly see the senescent cells. And they kill them.

Seven days of treatment produced measurable senolytic effects. Fourteen days reduced aortic senescence by 40-50%. Atherosclerotic plaque burden dropped 37%. When they knocked out AMPK, the whole effect vanished. When they blocked T cells with antibodies, same thing -- the effect disappeared.

The drug doesn't kill zombie cells. It takes off their disguise and lets your immune system do what it was built to do. Same checkpoint mechanism that oncologists exploit with drugs like pembrolizumab and nivolumab -- except triggered by a $15/month diabetes pill.

The telomere data

A 2025 randomized, double-blind, placebo-controlled trial published in Cell Reports Medicine tested henagliflozin (an SGLT2 inhibitor) in 150 adults with type 2 diabetes for 26 weeks.

Results: 90.5% of the treatment group showed telomere lengthening, versus 65.6% of placebo. Granzyme B -- a marker of cytotoxic T-cell activity -- increased significantly. Beta-hydroxybutyrate rose from 0.07 to 0.10 mmol/L, confirming the caloric restriction mimicry signal. IGFBP-3 (an aging biomarker) held steady in the treatment group while declining in placebo.

Telomere data should always be taken with a grain of salt. Telomere length is a correlate of aging, not necessarily a driver. But the directionality -- combined with the immune activation markers, the senescence clearance, and the mortality data -- paints a consistent picture. (If you're tracking biomarkers like these, our longevity blood panel guide covers optimal ranges for 25+ markers including the metabolic and inflammatory signals SGLT2 inhibitors move.)

Why the longevity community is sleeping on this

Three reasons.

First, the sex difference. The female mice got nothing. In longevity circles, that's treated as a disqualifier. It shouldn't be -- it's a dosing problem, not a mechanism problem. The 20-fold plasma concentration gap between sexes suggests females need roughly half the dose. No one has run that study yet.

Second, the diabetes label. SGLT2 inhibitors are "diabetes drugs" in the same way rapamycin is an "immunosuppressant." The original indication has nothing to do with the longevity mechanism. But the label sticks, and it makes both researchers and biohackers overlook the compound.

Third, there's no narrative. Rapamycin has the Easter Island origin story, the mysterious soil bacteria, decades of aging research. NMN has David Sinclair and a decade of podcasts. SGLT2 inhibitors have cardiologists quietly prescribing them to diabetics. There's no guru, no brand, no movement.

Which is exactly why the evidence is more trustworthy.

The risks (don't skip this section)

SGLT2 inhibitors are generally well tolerated in the diabetic population, but "generally well tolerated" isn't "risk-free."

Euglycemic DKA. Diabetic ketoacidosis with normal blood sugar. About 3x baseline risk. This is the dangerous one. It can present without the classic high-glucose warning, which means it gets missed. If you're combining an SGLT2 inhibitor with a ketogenic diet, the risk compounds -- you're already producing ketones, and the drug pushes that further. Median onset is 43 days after starting.

Genital mycotic infections. Significantly increased in both sexes. You're peeing sugar. Yeast likes sugar. This is the most common side effect and usually manageable, but uncomfortable.

UTIs. Marginally increased. The FDA issued a 2015 warning for severe cases including pyelonephritis and urosepsis.

Fournier's gangrene. FDA class-wide warning. Twelve cases reported in five years versus six cases in 30+ years for other diabetes drugs. Extremely rare, but necrotizing fasciitis of the perineum is as bad as it sounds.

Canagliflozin amputation signal. The CANVAS trial showed increased risk of toe and foot amputations with canagliflozin specifically. Empagliflozin and dapagliflozin have not replicated this signal.

Volume depletion. Osmotic diuresis means more fluid loss. Older adults and anyone on diuretics need monitoring.

For non-diabetics considering off-label use: hypoglycemia monitoring becomes critical. Without the elevated baseline glucose of a diabetic, the margin is thinner.

What to actually do with this information

If you're metabolically unhealthy -- prediabetic, type 2 diabetic, or carrying significant visceral fat -- talk to your doctor about SGLT2 inhibitors. This isn't fringe anymore. The CV mortality data alone makes the conversation worth having, and the longevity signal is a bonus.

If you're metabolically healthy and interested in longevity pharmacology: this is a watch-and-wait situation. The off-label evidence isn't mature enough for a blanket recommendation, and euglycemic DKA is a real risk in people who don't need the glucose-lowering effect. The late-start ITP data is encouraging. The biomarker trials are accumulating. But we're not at "healthy people should take this" yet.

What you should know either way:

The three main options are empagliflozin (Jardiance), canagliflozin (Invokana), and dapagliflozin (Farxiga). Empagliflozin has the cleanest safety profile. Canagliflozin has the ITP lifespan data but also the amputation signal. Dapagliflozin has the broadest indication set (heart failure and CKD, regardless of diabetes status).

The sex difference matters. If human pharmacokinetics mirror the mouse data -- and there's reason to think they might -- women may need lower doses. This is an active area of investigation.

And the mechanism convergence matters. SGLT2 inhibitors, rapamycin, acarbose, and metformin all converge on mTOR suppression and AMPK activation through different upstream pathways. The 70-90 grams of excreted glucose is essentially pharmaceutical caloric restriction -- your body responds as if it's in a mild energy deficit, without the hunger or the willpower.

The bottom line

The ITP comparison table tells one story: rapamycin is the longevity king in mice.

The human evidence tells a different one. SGLT2 inhibitors have more randomized, placebo-controlled mortality data than every other candidate longevity drug combined. A 32% all-cause mortality reduction in EMPA-REG. A 43% reduction in real-world data. Senescent cell clearance through immune checkpoint stripping. Telomere lengthening in a controlled trial. Caloric restriction mimicry without the restriction.

A diabetes drug that clears zombie cells, lengthens telomeres, and cut all-cause mortality by a third. The longevity community is still arguing about NMN.

We rate SGLT2 inhibitors and compare them head-to-head with rapamycin, metformin, and acarbose in the full NeverDie Guide. Evidence tiers, dosing context, and the complete ITP comparison -- all in one place.