Ozempic-class drugs
GLP-1 drugs work astonishingly well for weight loss in non-diabetics — and the early signal is that they do more than that, including a measurable slowing of biological aging. The catch is that about a third of the weight you lose is lean tissue — the same share you'd lose dieting.
Semaglutide (Ozempic, Wegovy, Rybelsus) and the next-generation dual agonist tirzepatide (Mounjaro, Zepbound) are the first drugs to pair double-digit weight loss with hard clinical outcomes — and, for semaglutide, a measurable deceleration of validated epigenetic aging clocks in non-diabetic adults. That combination is genuinely new in pharmacology, but it is not evenly split between the two drugs. Semaglutide 2.4 mg cut major adverse cardiovascular events from 8.0% to 6.5% over a mean of 39.8 months in non-diabetic adults with obesity and existing heart disease, roughly a 20% relative reduction.[1] Tirzepatide has never been tested against placebo for cardiovascular events; its one outcome trial found it no better than an older GLP-1 drug, dulaglutide.[2] Whether either should be used by healthy adults purely for longevity — and at what dose — is the central question this article tries to answer honestly.
What they are, briefly
GLP-1 (glucagon-like peptide-1) is an incretin hormone gut L-cells release after a meal. It boosts glucose-dependent insulin secretion, suppresses glucagon, slows gastric emptying, and acts centrally to dampen appetite. Native GLP-1 has a half-life of minutes. Semaglutide is a structurally modified analog resistant to degradation by the enzyme DPP-4 (dipeptidyl peptidase-4), allowing once-weekly subcutaneous (Ozempic, Wegovy) or daily oral (Rybelsus, Wegovy pill) dosing. Tirzepatide adds GIP (glucose-dependent insulinotropic polypeptide) receptor agonism on top.
| Agent | Class | Brand | Route | Status (September 2026) |
|---|---|---|---|---|
| Semaglutide | GLP-1 RA | Ozempic / Wegovy injection / Wegovy pill / Rybelsus | weekly SC injection or daily oral | approved for type 2 diabetes and for weight management; the 25 mg tablet was approved for weight in December 2025 |
| Tirzepatide | GIP + GLP-1 dual | Mounjaro / Zepbound | weekly SC | approved; largest weight loss of any agent with head-to-head data, both trials open-label |
| Orforglipron | non-peptide oral GLP-1 RA | Foundayo | daily oral | approved for weight management in April 2026, on the ATTAIN-1 trial[3] |
| Cagrilintide + semaglutide | amylin + GLP-1 | CagriSema | weekly SC | phase 3 complete, filed with the FDA in December 2025, not yet approved |
| Survodutide | glucagon + GLP-1 dual | — | weekly SC | phase 3 complete, not approved |
| Retatrutide | GIP + GLP-1 + glucagon triple | — | weekly SC | phase 3 topline only, not approved |
Efficacy in non-diabetic adults
Weight loss — Strong
In non-diabetic adults with obesity, semaglutide 2.4 mg/week takes off 10–16% of total body weight over 68–104 weeks, with the higher end from the shorter trials and the lower end from four-year follow-up. Placebo-subtracted, the effect is smaller: a meta-analysis of 7 randomised trials in 5,411 non-diabetic adults found 12.24 kg, 12.15 percentage points of body weight and 9.32 cm of waist circumference more lost on semaglutide than on placebo, and people on the drug were about 2.6 times as likely to lose at least 5% of their weight (relative risk 2.63, 95% confidence interval 2.12–3.25 — the confidence interval is the range where the true effect most plausibly lies, and when it stays clear of 1.0 the result is unlikely to be chance).[4] That analysis stopped searching the literature in April 2024. The larger and more current picture comes from a network meta-analysis of 262 trials in 99,791 people: against lifestyle change alone at one year, tirzepatide gives about 14.9% weight loss, CagriSema 14.8%, oral semaglutide 10.9%, orforglipron 9.9% and injected semaglutide 9.8%.[5]
The SELECT trial gives the longest randomised weight data available. Among 17,604 adults with established cardiovascular disease and overweight or obesity but no diabetes, semaglutide 2.4 mg produced 10.2% weight loss and a 7.7 cm waist reduction at 208 weeks, against 1.5% and 1.3 cm on placebo, with the loss reached by about week 65 and held for four years.[6] At the two-year mark, 12.0% of the semaglutide group had reached a body mass index below 25 kg/m², versus 1.2% on placebo — none were in that category at entry. One detail from the same analysis matters for anyone thinking about elective use: people who started at a lower BMI were the most likely to stop the drug.
The same molecule works as a tablet. In OASIS 4, once-daily oral semaglutide 25 mg produced 13.6% weight loss versus 2.2% on placebo over 64 weeks in 307 adults with obesity and no diabetes, a difference of about 11 percentage points.[7] That dose was approved in the United States for weight management in December 2025 and is sold as the Wegovy pill, distinct from the lower-dose Rybelsus tablet used for type 2 diabetes. Manufacturer materials quote 16.6% for the same trial; that is the figure among people who stayed on treatment, and 13.6% is the number counting everyone randomised, which is the one to plan around.
Tirzepatide is meaningfully more effective. In the head-to-head SURMOUNT-5 trial — an open-label randomized comparison in adults with obesity and no diabetes — tirzepatide produced 20.2% weight loss at 72 weeks vs. 13.7% for semaglutide (waist 18.4 vs. 13.0 cm).[8] The honest way to state that gap is 6.5 percentage points of body weight; quoted as a ratio it becomes "47% more weight loss", which is the same fact dressed up. Two design caveats travel with it: each arm was titrated to the maximum tolerated dose, so some of the semaglutide group never reached the licensed 2.4 mg; and in an open-label weight-loss trial, participants who know they were assigned the drug widely reported as stronger have a plausible behavioural head start. The endpoint here is weight, a surrogate rather than a hard clinical outcome, and the trial was funded by the manufacturer, Eli Lilly.
CagriSema, semaglutide combined with the amylin analogue cagrilintide, is the nearest thing to a new arrival: it produced 20.4% weight loss at 68 weeks versus 3.0% on placebo in 3,417 adults without diabetes, and has been under FDA review since December 2025.[9] It has not displaced tirzepatide, though. In the open-label REDEFINE 4 trial, 809 adults over 84 weeks, CagriSema missed its primary endpoint of non-inferiority to tirzepatide — 20.2% versus 23.6% weight loss counting everyone randomised — so the current head-to-head evidence points the same way as SURMOUNT-5 does.
Two further agents are in late-stage development but not yet approved. Survodutide, a glucagon and GLP-1 dual agonist, cut body weight by 13.0% at 76 weeks on the 6 mg dose versus 5.4% on placebo in 725 adults with obesity, a placebo-subtracted difference of about 8 percentage points.[10] Retatrutide, which adds glucagon receptor agonism to the GIP and GLP-1 activity of tirzepatide, reported the largest weight loss yet seen from a drug — 28.3% at 80 weeks and 30.3% at 104 weeks on the top dose, against 2.2% on placebo at 80 weeks — but those are manufacturer topline numbers from the TRIUMPH-1 trial released in May 2026, with no peer-reviewed publication yet, so treat them as provisional. Neither drug is approved, and an unapproved drug bought online is not reliably the same molecule.
Outside the obesity indication — Weak / no controlled evidence
No randomised trial has ever tested semaglutide for weight loss in adults below a body mass index of 27 kg/m². Every obesity registration programme — STEP, SURMOUNT, SELECT, OASIS 4 — required a BMI of 30, or 27 with a weight-related complication, to enrol. Semaglutide has been randomised without any weight requirement, but only for other diseases: the evoke and evoke+ trials gave oral semaglutide to 3,808 adults with early Alzheimer's disease under eligibility rules that set no BMI limit at all, and found no effect on cognitive decline.[11] Those trials tell you the drug is tolerated across the weight range; they tell you nothing about whether a lean adult gains anything from taking it. What circulates instead are uncontrolled retrospective chart reviews from private clinics dispensing compounded semaglutide, published in low-tier journals; they report that most patients lose weight, which is what an appetite suppressant does and tells you nothing about benefit or harm. This kind of "elective" use sits outside the FDA label and remains regulatorily and ethically contested — the absence of a medical necessity changes the risk-benefit math sharply, especially for muscle and bone (see below).
The geroprotective signal — Weak / preliminary
This is what makes GLP-1 receptor agonists (RAs) interesting on a longevity site rather than just a weight-loss site. The mechanistic case for GLP-1 signalling as an aging intervention has been argued across several hallmarks of aging — inflammation, oxidative stress and stem cell exhaustion among them — almost entirely in rodents and cell culture.[12] Whether any of it is separable from weight loss is contested in the field.[13]
The one strand with randomised human evidence behind it is the inflammation signal. In a prespecified secondary analysis of SELECT's 17,604 participants, semaglutide lowered high-sensitivity C-reactive protein (hsCRP) by 37.8% at 104 weeks, and the fall was already visible at 4 to 8 weeks — before meaningful weight loss — and occurred even in participants who lost no weight.[14]
The first human RCT to test this with validated biomarkers used adults with HIV-associated lipohypertrophy — a population that ages biologically faster than the general population. After 32 weeks of weekly semaglutide:
| Epigenetic clock | Effect vs. placebo, per calendar year | Relevance |
|---|---|---|
| PhenoAge | −4.90 years | morbidity and mortality risk |
| PCGrimAge | −3.08 years | mortality risk, principal-component version |
| GrimAge V2 | −2.26 years | refined mortality risk |
| OMICmAge | −2.20 years | multi-omic integration |
| RetroAge | −2.18 years | transposable element / genomic stability |
| GrimAge V1 | −1.39 years | mortality / lifespan |
| DunedinPACE | −0.09 units (9% slower) | pace of aging |
| Intrinsic Capacity clock | unchanged | functional capacity |
These are rates — epigenetic years accrued per calendar year — not years erased over the 32 weeks of dosing.[15] Organ-specific analyses pointed the same way in all eleven system clocks, but only seven of the eleven reached statistical significance. The largest reductions were in the inflammation, brain and metabolic clocks — each about 5 epigenetic years per calendar year against placebo — while the lung, musculoskeletal, immune and hormone clocks did not move detectably.
Read those numbers with the design in mind. This was a post-hoc, exploratory analysis of a 32-week phase 2b trial — not a trial designed to test aging clocks — with 45 participants on semaglutide and 39 on placebo, in a population selected for accelerated biological aging. Seventeen clocks were tested with no correction for multiple comparisons, and two of the authors work for TruDiagnostic, a commercial epigenetic-testing company, whose scientific adviser is the first author. It has now cleared peer review and been published in Nature Communications, which retires the preprint caveat but none of the design ones; the authors still ask for validation in broader populations.[16]
With that said: few randomised trials have moved several validated aging clocks at once; the larger one, with omega-3 supplements in older adults, found much smaller changes.[17] It does not prove longevity extension — none of these clocks have been validated as causal. For scale, the best-known other randomised trial to move a methylation clock is CALERIE, in which two years of 25% caloric restriction in 220 healthy non-obese adults aged 21–50 slowed DunedinPACE by about 2–3% and moved neither PhenoAge nor GrimAge.[18] That was also a post-hoc analysis, with the clock's patent-holders among the authors, and participants achieved only about half the prescribed restriction. Neither result establishes more than that a clock can be nudged. It places GLP-1 RAs alongside, not above, the rest of the geroprotector class on translational evidence. Compare rapamycin and metformin in Geroprotectors: mechanism rich, hard-outcome evidence still pending.
Cardiovascular and functional benefits — Strong (cardiovascular events); Moderate (functional capacity)
The SELECT trial is the hard-outcome anchor for the entire class: in about 17,600 non-diabetic adults with overweight or obesity and established cardiovascular disease, semaglutide 2.4 mg cut major adverse cardiovascular events — MACE, meaning cardiovascular death, non-fatal myocardial infarction (MI), and non-fatal stroke — by roughly 20% over a mean ~40 months, with events in 6.5% on semaglutide versus 8.0% on placebo (hazard ratio 0.80, 95% confidence interval 0.72–0.90 — the interval stays below 1.0, so the benefit is unlikely to be a chance finding).[19] This is a randomized trial reporting a hard clinical outcome — the strongest tier of evidence — though the trial was funded by the manufacturer, Novo Nordisk. Real-world data point the same way. In a target-trial emulation of 140,169 matched pairs of adults with obesity and no diabetes, starting a GLP-1 drug rather than another anti-obesity medication was associated with about 24% fewer major adverse cardiovascular events (hazard ratio 0.76, 95% CI 0.72–0.81).[20] Treat that as directional only: the same analysis reports 51% lower all-cause mortality, far larger than anything the randomised trials show, which is the classic fingerprint of healthier people being the ones who start and stay on a drug. The cardioprotective effect is only partly explained by the weight loss. In a prespecified SELECT analysis, the benefit held across every baseline weight and waist-circumference category and showed no relationship to how much weight a patient lost, though roughly a third of it was mediated by the fall in waist circumference — leaving most of the effect to direct vascular, myocardial and anti-inflammatory mechanisms rather than to the scale.[21]
Tirzepatide's hard-outcome record is thinner and should not be read across from semaglutide. Its only cardiovascular outcome trial, SURPASS-CVOT, compared it with dulaglutide rather than placebo in 13,165 people with type 2 diabetes and established atherosclerotic disease; events occurred in 12.2% on tirzepatide versus 13.1% on dulaglutide, which cleared the bar for non-inferiority but not for superiority (hazard ratio 0.92, 95.3% confidence interval 0.83–1.01, p = 0.09 for superiority).[22] Where tirzepatide does have randomised hard-outcome data is heart failure: in SUMMIT, 731 patients with obesity and heart failure with preserved ejection fraction had cardiovascular death or a worsening heart-failure event in 9.9% on tirzepatide versus 15.3% on placebo over a median 104 weeks, about 38% fewer (hazard ratio 0.62, 95% confidence interval 0.41–0.95) — a composite driven by heart-failure events, not by deaths, of which there were 8 and 5 respectively.[23]
Oral semaglutide carries the same signal as the injection. In SOUL, once-daily oral semaglutide up to 14 mg was compared with placebo in 9,650 people with type 2 diabetes plus atherosclerotic cardiovascular disease, chronic kidney disease, or both: major adverse cardiovascular events occurred in 12.0% versus 13.8% over a median 49.5 months, about 14% fewer (hazard ratio 0.86, 95% confidence interval 0.77–0.96).[24] The effect is smaller than SELECT's and the population was diabetic, but the route of administration does not appear to cost you the benefit.
Pooled across the class, the mortality question now has a high-certainty answer — in people who were already ill. A 2025 meta-analysis of 21 randomised trials and 99,599 patients found GLP-1 receptor agonists reduced death from any cause by about 12% (incidence rate ratio 0.88, 95% confidence interval 0.84–0.92; 121 people treated to prevent one death), cardiovascular death by about 13%, and major adverse cardiovascular events by about 13% (66 treated to prevent one), alongside a 63% relative increase in gastrointestinal disorders and a 26% increase in gallbladder disorders.[25] Every one of those trials enrolled people who were already ill; none tells you what the drug does to a healthy adult's mortality.
Functional capacity also improves. Across the STEP-HFpEF programme — the pooled STEP-HFpEF and STEP-HFpEF DM trials, 1,145 patients with obesity-related heart failure with preserved ejection fraction — semaglutide added 17.1 m to the 6-minute walk distance over placebo at 52 weeks (95% confidence interval 9.2–25.0), and most of that gain was already present at 20 weeks, before maximal weight loss.[26] The phase 3b STRIDE trial randomised 792 adults with type 2 diabetes and symptomatic peripheral artery disease with intermittent claudication to semaglutide 1 mg weekly or placebo for 52 weeks; maximum walking distance on a constant-load treadmill improved by about 13% over placebo (estimated treatment ratio 1.13, 95% confidence interval 1.06–1.21; median ratio to baseline 1.21 on semaglutide versus 1.08 on placebo, p = 0.0004).[27] The trial reported ratios only; against the 186 m median baseline recorded in the STRIDE design paper, 13% is on the order of 24 additional metres — a figure derived here, not one the trial states. Joint pain moves too. In STEP 9, 407 adults with obesity and moderate knee osteoarthritis with at least moderate pain, semaglutide 2.4 mg reduced the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) pain score by 41.7 points on a 0–100 scale over 68 weeks versus 27.5 points on placebo, with a parallel gain in 36-Item Short Form Health Survey (SF-36) physical function of 12.0 versus 6.5 points.[28] These are patient-reported outcomes, not hard endpoints — but staying mobile is close to the whole point of a healthspan argument.
These are real but modest gains. Anchored to how patients themselves rate their change, the minimal clinically important improvement in the 6-minute walk test has been estimated at 14–15 m for a "little improvement" and 20–24 m for a moderate one, in heart failure with reduced ejection fraction.[29] Semaglutide's 17.1 m clears the first bar and not the second. Still, these are populations that are very hard to move with anything else — though note STRIDE enrolled only diabetic patients, so it does not generalise to non-diabetic claudication.
The kidney — Strong (in kidney disease and cardio-kidney-metabolic populations)
Kidney outcomes are the second hard endpoint this class moves, and they rarely make consumer coverage. In FLOW, semaglutide 1 mg weekly versus placebo in 3,533 people with type 2 diabetes and chronic kidney disease cut a composite of kidney failure, a halving of filtration rate, or kidney or cardiovascular death by about 24% (hazard ratio 0.76, 95% confidence interval 0.66–0.88); the trial was stopped early for benefit, and death from any cause was about 20% lower (0.80, 0.67–0.95).[30] Pooling participant-level data from SELECT, FLOW and SOUL — 30,787 people, with and without diabetes — the kidney composite fell by about 16% (0.84, 0.77–0.91), and the authors note the benefit is not fully explained by glucose or weight effects.[31]
Sleep apnoea — Strong
In SURMOUNT-OSA, two phase 3 trials in 469 adults with obesity and moderate-to-severe obstructive sleep apnoea, tirzepatide cut the apnoea-hypopnoea index — apnoeas and hypopnoeas per hour of sleep — by 20.0 events per hour more than placebo in people not using positive airway pressure (95% confidence interval 14.2–25.8) and by 23.8 more in those who were (17.9–29.6), against a baseline of roughly 50 events per hour.[32] On that evidence tirzepatide (Zepbound) was approved for moderate-to-severe obstructive sleep apnoea in adults with obesity in December 2024 — the first drug ever licensed for the condition. Unlike the cardiovascular benefit, this one tracked weight loss closely, so it looks adiposity-mediated. See Sleep-disordered breathing.
Liver — Moderate (histology, not outcomes)
In ESSENCE, a phase 3 trial in adults with biopsy-confirmed metabolic dysfunction-associated steatohepatitis (MASH) and stage 2–3 fibrosis, steatohepatitis resolved without fibrosis worsening in 62.9% on semaglutide 2.4 mg versus 34.3% on placebo at 72 weeks, and fibrosis improved without steatohepatitis worsening in 36.8% versus 22.4%.[33] Semaglutide received accelerated approval for non-cirrhotic MASH with moderate-to-advanced fibrosis in August 2025.[34] Read "accelerated" literally: the approval rests on liver biopsies, and the trial runs to week 240 before anyone knows whether the histology translates into fewer cirrhoses, transplants or deaths.
The brain: food noise, mood, and addiction
The most psychologically striking effect patients report is the silencing of "food noise" — the relentless, intrusive cognitive preoccupation with food. The usual mechanistic account is that GLP-1 receptor agonists reach reward circuitry, including the ventral tegmental area (VTA), and blunt the dopamine response to hyperpalatable food, weakening the cue-driven reinforcement loop. That chain has been worked out in rodents rather than in people, so treat it as a plausible story for a real and well-attested experience, not as an established human mechanism.
The blunting generalises beyond food (Moderate — large, but observational). A Swedish national register study followed 95,490 people who already carried a diagnosis of depression or anxiety and were taking antidiabetic medication, comparing each person's own periods on and off a drug. Against their own non-use periods, semaglutide was associated with a 44% lower risk of worsening depression (aHR 0.56, 95% CI 0.44–0.71), a 38% lower risk of worsening anxiety (0.62, 0.52–0.73), and a 47% lower risk of worsening substance use disorder (0.53, 0.35–0.80); GLP-1 RAs as a class were associated with less self-harm (0.56, 0.34–0.92).[35] Two limits on how far that stretches: the outcome is deterioration (psychiatric hospitalisation, long psychiatric sick leave, self-harm, suicide) in people who were already ill, not prevention of new depression in healthy adults; and the effect was specific to semaglutide and liraglutide — exenatide and dulaglutide showed nothing. Patients also spontaneously report reduced cravings for alcohol, opioids, nicotine, and even compulsive shopping — but where that has been tested, craving and behaviour have come apart. The only randomised smoking trial gave nine weeks of low-dose semaglutide to 24 adults smoking at least five cigarettes a day: nicotine craving and body weight fell, while both co-primary smoking outcomes — laboratory smoking resistance and cigarettes per day — did not move.[36] Craving is not the same thing as quitting.
Three randomised trials have now tested semaglutide in alcohol use disorder, and they are small and mixed. The largest and longest randomised 108 treatment-seeking adults who had both moderate-to-severe alcohol use disorder and obesity to weekly semaglutide 2.4 mg — the full weight-management dose — or placebo for 26 weeks, on top of standard cognitive behavioural therapy; heavy drinking days fell by 41 percentage points from baseline on semaglutide against 26 points on placebo, a drug-attributable difference of about 14 percentage points (estimated treatment difference −13.7 percentage points, 95% CI −22.0 to −5.4).[37] An earlier NIH-funded trial gave low-dose semaglutide (0.25–1.0 mg weekly) to 48 non-treatment-seeking adults for nine weeks with no therapy attached: it cut laboratory alcohol self-administration and craving, but left drinks per calendar day and number of drinking days unchanged.[38] A third trial, of oral semaglutide in 50 treatment-seekers, missed its primary endpoint of cue-elicited craving while still reducing heavy drinking days.[39] See also Alcohol for context on alcohol harms.
Caveat: profound dopamine modulation can theoretically blunt non-food pleasure (anhedonia) in a susceptible minority. Watch for it on initiation and if mood drops, treat it as a real signal.
Alzheimer's: clinical decline didn't move — Strong (null for clinical benefit)
The phase 3 evoke and evoke+ trials randomised 3,808 adults aged 55–85 with amyloid-confirmed mild cognitive impairment or mild dementia due to Alzheimer's disease to once-daily oral semaglutide up to 14 mg or placebo for up to 156 weeks, and measured global clinical and functional decline on the Clinical Dementia Rating–Sum of Boxes at week 104. Both missed: the estimated difference was −0.08 points (95% CI −0.35 to 0.20) in evoke and 0.10 points (−0.17 to 0.38) in evoke+, and the programme was discontinued for negative clinical outcome.[40] A cerebrospinal fluid (CSF) substudy did move in the right direction — tau and synaptic markers fell from baseline, p-tau181 by 8.3%, total tau by 6.6% and neurogranin by 8.6% — but that comes from a manufacturer's conference presentation rather than the peer-reviewed report, and should be read as provisional. The honest interpretation: GLP-1 RAs may modify pathology biology without rescuing already-established clinical disease, offering nothing for current symptomatic patients while leaving the prophylactic-decades-earlier hypothesis open. That hypothesis is where the observational data sit. Among 214,442 propensity-matched adults with type 2 diabetes and no baseline neurodegeneration, starting a GLP-1 RA rather than a dipeptidyl peptidase-4 (DPP-4) inhibitor was followed over a mean four years by new neurodegenerative disease in 2.2% versus 2.9% — an absolute difference of 0.6 percentage points, about 19% lower relative risk (hazard ratio 0.81, 95% CI 0.77–0.86) — with dementia and Alzheimer's disease moving together and Parkinson's disease not moving at all.[41] A separate matched cohort of 60,860 adults with diabetes and obesity found roughly 37% lower dementia incidence over seven years (hazard ratio 0.63, 95% CI 0.50–0.81).[42] Both are observational and open to confounding by indication. The one randomised echo is a post-hoc analysis of SELECT in which semaglutide slowed the rise of a validated 25-protein plasma dementia-risk score over 104 weeks in 2,970 adults aged 65 and over — a predicted-risk surrogate, not diagnosed dementia, and analysed by the manufacturer.[43]
The catch nobody talks about enough: muscle and bone
This is where elective use among healthy adults diverges sharply from cost-benefit analysis in obesity treatment.
About a third of the weight is lean tissue — Moderate
Across 20 randomised trials in 15,782 adults, fat-free mass accounted for about 35% of the weight lost on semaglutide (95% CI 32–39%), 25% on tirzepatide and 27% on liraglutide — but also 26% (95% CI 24–28%) on intensive lifestyle intervention alone, a difference that was not statistically significant.[44] That is the honest frame: this is what losing 15–20% of body weight costs by any route, not a drug-specific toxicity.
The proportion itself does not appear to differ by sex or age. In the SURMOUNT-1 body-composition substudy, roughly 75% of weight lost was fat and 25% lean in both the tirzepatide and the placebo arm, and that split held across sex and across age bands under 50, 50 to 64, and 65 and over.[45] What differs is the consequence. Older adults and postmenopausal women start with less muscle and bone in reserve, and the joint advisory names older age, menopausal status, low testosterone and sedentary behaviour as the factors that turn a proportionally ordinary loss into sarcopenia — and identifies older individuals and women as losing more bone for a given weight reduction.[46]
Two caveats cut in opposite directions. Fat-free mass on a dual-energy X-ray absorptiometry (DXA) scan is not contractile muscle — it bundles water, connective tissue and organ mass, some of which is shed alongside adipose tissue.[47] But the absolute losses are real: pooled across seven placebo-controlled obesity trials in 821 adults, lean mass fell 1.7 kg overall and 5.4 kg on semaglutide specifically — even though lean mass as a share of body weight rose, which is why those authors do not treat it as a reason against the drug.[48] Muscle is the largest sink for postprandial glucose; losing it lowers resting energy expenditure and, in someone who starts with little reserve, accelerates frailty risk.
The BELIEVE trial tested combining semaglutide with bimagrumab, a monoclonal antibody that blocks type II activin receptors — the receptors through which myostatin and related ligands restrain muscle growth. Across 507 adults with obesity but without diabetes, the double-blind phase ran 48 weeks with an open-label extension to 72. At week 72 the high-dose combination produced 22.1% total weight loss, of which 92.2% was fat mass; semaglutide 2.4 mg alone lost 15.7% with 75.6% from fat. Bimagrumab alone lost 10.8%, all of it fat, while adding 2.5% lean mass.[49] Activin-pathway drugs aren't approved, so for now the answer is mechanical (resistance training) and nutritional (protein) — see below.
A nuance: while muscle quantity drops, muscle quality in the remaining tissue appears to improve. In the SURPASS-3 magnetic resonance imaging substudy, 52 weeks of tirzepatide cut thigh muscle volume by 0.64 L — statistically indistinguishable from what UK Biobank data predict for that much weight loss — while reducing intramuscular fat by significantly more than expected.[50] On evidence like this a Circulation primer concluded the muscle response looks adaptive rather than pathological, with improved insulin sensitivity and less fat infiltration lowering the probability of losing strength and function.[51] The mitochondrial part of the story is still rodent-only, and the mouse finding — improved oxidative phosphorylation efficiency — is framed by its own authors as a possible driver of weight regain rather than a benefit.[52] None of this rescues absolute strength in someone who starts sarcopenic, but it explains why functional outcomes often improve despite lean mass loss.
Bone mineral density — Moderate
A 52-week phase 2 randomised trial in 64 adults selected for increased fracture risk — 86% postmenopausal women, mean body mass index 27.5 kg/m² — found semaglutide 1.0 mg reduced total hip bone mineral density by 2.6% (95% CI 1.1–4.1%) and lumbar spine by 2.1% (95% CI 0.5–3.6%) versus placebo alongside about 9% weight loss, with tibial cortical thickness down 1.8%.[53] Read the whole trial, though. Its primary endpoint — the bone formation marker P1NP — was null; femoral neck density did not move; and estimated bone strength by micro-finite-element analysis was unchanged at both the tibia and the radius. What did change was bone resorption, up about 55%, which is the signature of mechanical unloading and rapid energy deficit rather than direct osteoblast toxicity. A matched-cohort scan study places the signal more in non-diabetics losing weight on GLP-1s than in people with type 2 diabetes: total-hip loss exceeded matched controls only in the non-diabetic group (−1.0% vs −0.6%), while in type 2 diabetes it was no different from controls — the metabolic improvements appear to offset the mechanical unloading.[54]
Bone mineral density is a surrogate, and the fracture data do not track it cleanly. In a target-trial emulation of 133,606 propensity-matched adults aged 50 to 90 with type 2 diabetes, starting a GLP-1 receptor agonist rather than a dipeptidyl peptidase-4 inhibitor was associated with about 21% lower three-year fragility fracture risk (hazard ratio 0.79, 95% CI 0.76–0.83) — an absolute reduction of 0.79 percentage points, roughly one fracture avoided per 126 people treated — with the biggest reductions at the hip and spine.[55] A Danish nested case-control study of 337,648 fracture cases pointed the same way, with about 12% lower odds of major osteoporotic fracture among users (odds ratio 0.88, 95% CI 0.83–0.94).[56]
The signal that survives is narrower — and it is the one that matters for elective use. In that same emulation the protective association held only in people with diabetes and reversed in those without (hazard ratio 1.13, 95% CI 1.04–1.23). A matched study of 33,210 pairs of overweight and obese adults without diabetes found one-year fracture risk of 3.05% on a GLP-1 receptor agonist versus 2.61% without, concentrated almost entirely in those with a body mass index of 40 or above and those aged 68 or older.[57] An Israeli cohort of 46,177 adults aged 65 and over with type 2 diabetes found an 11% higher risk (hazard ratio 1.11, 95% CI 1.01–1.21).[58]
These are observational studies pointing in different directions, and none of them is a trial. The reasonable reading: no broad fracture excess, with a plausible one in the oldest and heaviest non-diabetic users. Baseline DXA, calcium and vitamin D repletion, and weight-bearing and resistance loading remain worth doing — more as insurance against a real but unresolved risk than as a settled requirement.
How to take it without wrecking yourself — Weak
If you are going to use a GLP-1 receptor agonist — especially as a healthy adult electing it for cardiometabolic optimization rather than treating obesity — the lifestyle scaffolding is the difference between a longevity tool and an accelerated path to sarcopenic obesity. One caveat about everything that follows: a 2026 international expert consensus on precisely this question found "a significant lack of direct evidence to guide clinical practice" and built its 52 recommendations mostly from bariatric nutrition practice and clinical experience rather than from trials in GLP-1 users.[59] The protocol below is cheap, low-risk and mechanistically sound. It is not, yet, proven.
Protein
Intake has to go above the ordinary adult reference of 0.8 g/kg/day, but the joint guidance from the American College of Lifestyle Medicine, American Society for Nutrition, Obesity Medicine Association, and The Obesity Society is more conservative than the figures circulating online: it notes that targets of 1.2–1.6 g/kg/day have been proposed during active weight reduction, and states that sustained intake at or above 2 g/kg/day should be avoided.[60] It also warns that in obesity, scaling off actual body weight can significantly overestimate what someone needs — where body composition is known, 1.5 g per kilogram of fat-free mass is the better anchor, and where it is not, the practical fallback is an absolute 80–120 g/day. For a 70 kg adult in the proposed band that's 84–112 g/day. Spreading it across 3–4 meals of roughly 25–30 g is a sensible way to hit the total rather than a finding of this advisory, which says nothing about meal timing — see Protein.
The hard part: GLP-1s blunt appetite and slow gastric emptying enough that hitting these targets from whole food alone is genuinely difficult for most users. Whey protein (high leucine) or a complete plant blend (pea + rice) is essentially mandatory. The strongest completed evidence for pairing the drug with exercise is not a resistance-training trial but comes close. In S-LiTE, 195 adults with obesity were randomised after an 8-week low-calorie diet to a year of supervised exercise, liraglutide 3.0 mg, both, or placebo. The combination cut body-fat percentage by 3.9 percentage points — roughly twice what exercise alone (1.7 points) or liraglutide alone (1.9 points) achieved — and only the combination improved HbA1c (a measure of average blood sugar), insulin sensitivity and cardiorespiratory fitness.[61] A year after all treatment stopped, the combination group was still 5.1 kg lighter and 2.3 percentage points leaner than those who had taken liraglutide alone.[62] The LEAN-PREP trial is the first randomised test of the specific protein-plus-resistance-training combination, in 232 adults at a single centre, using magnetic resonance imaging of quadriceps cross-sectional area, with a protein target of 1.6 g/kg/day and three home-based sessions per week — it began recruiting in August 2025 and is not due to report before 2029.[63] See also Protein for the underlying physiology.
Resistance training
Mandatory, not optional — and the dose is specific. The joint advisory calls for strength training at least three times per week, plus at least 150 minutes of moderate-intensity aerobic activity weekly, prescribed alongside the drug rather than suggested after it.[64] It is the intervention most consistently linked to a better ratio: pooled across randomised trials, lifestyle weight-loss arms that added resistance training lost 17.5% (95% CI 14–21%) of their weight as fat-free mass, against roughly a quarter for intensive lifestyle intervention alone — a comparison between trial arms, not a randomised test of training on top of a GLP-1 drug.[65] It matters for bone too: the advisory cites a randomised trial in which a year of GLP-1 therapy plus exercise preserved bone mineral density while GLP-1 therapy alone reduced it. Without mechanical loading, a high-protein diet is unlikely to be enough: the advisory states that protein intake alone is likely inadequate without structured resistance or strength training. See Resistance training.
Micronutrients and fiber
The energy deficit these drugs create is large. In a randomised trial in 114 adults without diabetes, three weeks of tirzepatide cut intake at an ad libitum lunch by about 525 kcal versus placebo (95% CI 401–648 kcal), alongside falls in appetite, cravings and perceived hunger.[66] A cut of that size means proportional drops in micronutrient intake unless the food that remains gets denser. A narrative review of six studies covering 480,825 adults on GLP-1 receptor agonists found vitamin D deficiency to be the commonest abnormality — 7.5% at six months, rising to 13.6% at twelve — with users showing 26–30% lower ferritin than a comparator group on SGLT2 inhibitors, and more than 60% consuming below estimated requirements for calcium and iron.[67] The authors synthesised these studies descriptively rather than pooling them, and are explicit that the data are mainly observational and cannot establish causality. Case reports of severe thiamine deficiency, including Wernicke encephalopathy, sit at the tail. Highest-priority adds:
- Vitamin D and calcium — vitamin D deficiency in 13.6% of users at one year; critical for the bone density side.
- Vitamin B12 and iron — altered gastric emptying and reduced intake; users show 26–30% lower ferritin than an SGLT2-inhibitor comparator group.
- Vitamins A, E, and magnesium — reduced fat-soluble vitamin intake on low-fat anti-nausea diets.
- Fiber 22–34 g/day — counteracts severe constipation from delayed gastric emptying. Titrate up slowly with water.
A Mediterranean-pattern baseline (Dietary patterns) handles much of this without explicit supplementation.
Hydration and electrolytes — Weak
No professional body sets a fluid target for people taking these drugs, and the claim that they blunt thirst centrally does not survive contact with the trials. In a randomised crossover trial in 20 healthy volunteers, three weeks of dulaglutide left thirst perception and blood sodium unchanged and cut fluid intake by a non-significant 100 mL/day.[68] The one trial that did reduce thirst studied patients with primary polydipsia, a disorder of pathological over-drinking, and its functional MRI substudy — the actual test of a central mechanism — found no difference in brain activity.[69] In both trials urine output fell in step with intake.
The dehydration risk that is real is gastrointestinal, and it is on the label: nausea, vomiting and diarrhoea cause volume depletion, and the label carries a warning about acute kidney injury following exactly that sequence (see Cautions below). So: drink to thirst normally, and drink deliberately more — with sodium, potassium and magnesium — during dose escalation and any stretch of GI side effects, which is when the risk is concentrated. The commonly quoted ~2.5–3.5 L/day is a rule of thumb at most, not a prescription. See Water for the filtration and remineralization side.
Bone surveillance
Baseline DXA before initiating, repeat at 12 months, then per risk profile. The joint advisory sets no numeric vitamin D or calcium target; it says supplements for at-risk nutrients such as vitamin D and calcium can be proactively considered, at appropriate doses tailored to the person[70]. The general adult recommended intake for calcium is 1,000–1,200 mg/day depending on age and sex, from food plus a supplement if needed[71]. For vitamin D, the recommended intake of 15–20 µg (600–800 IU) a day corresponds to a blood 25(OH)D of at least 50 nmol/L[72]. The often-quoted ≥75 nmol/L is not a current guideline target: the Endocrine Society's 2024 guideline found no clear evidence defining an optimal 25(OH)D level and no trial evidence supporting routine screening, including in people with obesity, and suggests against routine testing[73]. Bone turnover markers (CTX, P1NP) for high-risk patients.
| Pillar | Target | Why |
|---|---|---|
| Protein | 1.2–1.6 g/kg/day, or 80–120 g/day; avoid sustained ≥2 g/kg | Defend lean mass against the deficit |
| Resistance training | ≥3 sessions/week | Mechanical stimulus for muscle retention |
| Aerobic activity | ≥150 min/week | Prescribed alongside the drug, not after it |
| Fiber | 22–34 g/day, titrated | Counter delayed gastric emptying |
| Hydration | Drink to thirst; more during escalation and GI upset | Volume depletion from GI losses, not blunted thirst |
| Vitamin D + calcium | Intake at least the recommended amount (calcium 1,000–1,200 mg/day); no proven blood-level target | Bone density support |
| Electrolytes | Na, K, Mg | GI losses, exercise demand |
| DXA | Baseline + annual | Detect bone loss early |
Dosing, microdosing, and the maintenance question
Standard titration
Wegovy titrates monthly: 0.25 → 0.5 → 1.0 → 1.7 mg/week, then a maintenance dose of either 2.4 mg (recommended) or 1.7 mg once weekly. The label now also permits escalation to a 7.2 mg weekly maximum for adults who tolerate 2.4 mg and need further weight reduction, and an oral form (25 mg once daily) has been approved alongside the injection.[74] Slower titration reduces nausea and improves adherence, and the label explicitly allows any escalation step to be delayed by four weeks. The 7.2 mg dose buys more weight loss at a specific cost: dysesthesia — abnormal skin sensation — affected 22% at 7.2 mg versus 6% at 2.4 mg and 0.3% on placebo, and recurred in 45% of those re-escalated after it had resolved.
"Microdosing" for longevity
A growing off-label practice among biohackers and longevity-focused clinicians is to hold at 0.25 mg/week or less indefinitely, never escalating. The argument: most of the systemic geroprotective effects (lower hsCRP, improved lipids, glycemic stabilization, reduced food noise, vascular insulin sensitization) plausibly occur at sub-weight-loss doses, while side effects, muscle loss, and bone loss scale with dose and weight loss magnitude.[75] Every one of those effects was measured at 1.0–2.4 mg weekly in people with obesity or type 2 diabetes.
Honest assessment of microdosing as of 2026:
- No RCT has tested low-dose chronic semaglutide for longevity endpoints in healthy adults.
- The epigenetic clock data above came from full doses in a non-healthy population — extrapolating to healthy adults at 10–25% of that dose is speculation.
- Compounded semaglutide is where microdosing demand goes, and the legal ground under it has moved. Compounders could copy semaglutide and tirzepatide only while those drugs sat on the FDA shortage list; the FDA declared the tirzepatide shortage resolved on 19 December 2024 and the semaglutide injection shortage resolved on 21 February 2025, and the grace periods ran out during 2025. Anything still marketed as "compounded" or "personalised-dose" semaglutide therefore sits outside that exemption, and much of the online supply is unapproved material sold under "research purposes only" labelling. The FDA notes that semaglutide sodium and semaglutide acetate "are different active ingredients than are used in the approved drugs", that multi-dose vials carry contamination risk beyond 28 days, and that as of 31 May 2026 it had received 990 adverse-event reports for compounded semaglutide and more than 730 for compounded tirzepatide; it has issued successive rounds of warning letters to online sellers and telehealth firms through 2025 and 2026.[76] None of it is reviewed for dose accuracy, sterility or salt form.
- Real risk: at sub-therapeutic plasma concentrations, you may capture none of the supposed benefits and still face the GI, gallbladder, and pancreatitis risk profile.
If you go this route, do it with a real prescriber, not a compounding mill, and keep DXA + protein + training in place anyway.
You probably can't ever stop
The withdrawal evidence is unambiguous about direction, if not about mechanism. In the STEP 1 off-treatment extension, an exploratory subset of 327 participants stopped both semaglutide and the lifestyle programme at week 68 and were followed for a further year. They regained 11.6 percentage points of body weight — about two-thirds of the 17.3% they had lost — leaving them a net 5.6% below baseline at week 120, and most cardiometabolic improvements reverted toward baseline.[77] What that extension did not measure is what the regained weight is made of — no body-composition data were collected off treatment — so the widely repeated claim that it returns as pure fat is an extrapolation rather than a finding. Nor is there human evidence of a permanently suppressed metabolism: in a phase 1 trial in adults with obesity, tirzepatide had no measurable effect on metabolic adaptation while increasing fat oxidation.[78]
Like statin therapy and antihypertensives, semaglutide treats a chronic, relapsing condition; benefits last only while the drug is on board. Realistic expectation: indefinite maintenance. The label now recognises 1.7 mg once weekly as an alternative to the recommended 2.4 mg maintenance dose, so a step down that far is on-label; 1.0 mg is a titration step, not a labelled maintenance dose, and no randomised trial has tested reducing the dose after reaching goal and then measuring whether the weight holds. What is documented is only what prescribers actually do: in a Danish primary-care cohort of 206 people, among those still on treatment at 12 months, 40% were on 1.0 mg weekly or less, 30% on 1.7 mg and 30% on 2.4 mg, with mean weight loss among continuers of 13.6% (95% CI 12.2–15.0).[79] Treat de-escalation as an experiment you monitor, not a plan with evidence behind it.
Most real-world users do not manage indefinite maintenance. Among 15,811 commercially insured US adults who started semaglutide for weight management, 46% had stopped by the fifth month, and the rate climbed from 41% in the lowest monthly-copayment fifth to 51% in the highest.[80] In the same Danish cohort, 38% had permanently discontinued and another 15% temporarily by 12 months.[81] Cost is a dominant driver, which makes affordability of indefinite therapy part of the decision to start, not an afterthought.
Cautions and interactions
The tolerability profile is not incidental to the decision — it is most of the reason people stop, and a few of the signals are serious. Incidence figures below come from the randomised trials and from post-marketing pharmacoepidemiology.
- Gastrointestinal adverse events — the near-universal cost. In the pooled STEP 1–3 analysis of semaglutide 2.4 mg vs placebo, nausea affected 43.9% vs 16.1%, diarrhoea 29.7% vs 15.9%, vomiting 24.5% vs 6.3%, and constipation 24.2% vs 11.1%. The great majority were transient, clustered around dose escalation, and mild — 98.1% mild or moderate and 99.5% non-serious — and 4.3% of the semaglutide arm discontinued permanently because of them.[82] Slower titration is the main mitigation.
- Acute kidney injury from volume depletion. The label warns of postmarketing acute kidney injury, some cases requiring haemodialysis, occurring mostly in patients whose nausea, vomiting or diarrhoea led to dehydration, and instructs prescribers to check renal function in anyone reporting volume-depleting adverse reactions, especially during initiation and dose escalation.[83] This is why the hydration and electrolyte advice above matters, and why a bad week of vomiting is a reason to call a prescriber rather than push through. The same label cycle recorded severe gastrointestinal reactions in 4.1% on the injection versus 0.9% on placebo, and states the drug is not recommended in severe gastroparesis.
- Gallbladder and biliary disease. A meta-analysis of 76 randomised trials in 103,371 patients found GLP-1 RAs raised the risk of gallbladder or biliary disease by about 37% (RR 1.37, 95% CI 1.23–1.52), driven by cholelithiasis and cholecystitis. The risk was more than doubled in trials using GLP-1 RAs for weight loss specifically (RR 2.29, 1.64–3.18) and scaled with dose and duration. Rapid weight loss is itself lithogenic, so this is partly the weight loss and partly the drug.[84]
- Pancreatitis, gastroparesis, bowel obstruction. In a claims analysis of adults using GLP-1 RAs for weight loss, the semaglutide arm was only 613 people followed for a median of seven months, so the headline incidences — 4.6 per 1,000 person-years for pancreatitis and 9.1 for gastroparesis — rest on two and four events respectively and should be read as order-of-magnitude figures only. Against the comparator weight-loss drug bupropion-naltrexone, the pooled GLP-1 group, mostly liraglutide, carried roughly nine times the risk of pancreatitis (hazard ratio 9.09, 95% CI 1.25–66.00), about four times the risk of bowel obstruction (4.22, 1.02–17.40) and about 3.7 times the risk of gastroparesis (3.67, 1.15–11.90); biliary disease was not raised in this analysis (1.50, 0.89–2.53).[85] The absolute risks are low, the relative increases are not, and the confidence intervals are wide enough to be nearly uninformative about magnitude. A prior episode of pancreatitis is a reason not to start.
- Sudden vision loss (NAION). Non-arteritic anterior ischemic optic neuropathy is a painless, largely irreversible loss of vision in one eye caused by optic-nerve infarction, and it is now a labelled side effect in Europe. A 2024 matched cohort at a single neuro-ophthalmology referral centre raised the alarm, reporting a 36-month cumulative incidence of 8.9% on semaglutide versus 1.8% on non-GLP-1 comparators in people with type 2 diabetes — figures that describe a referral population and cannot be read as population risk.[86] The population-scale numbers are far smaller. Across all 424,152 Danes with type 2 diabetes, semaglutide roughly doubled the risk (hazard ratio 2.19, 95% CI 1.54–3.12), with an incidence of 0.23 versus 0.09 cases per 1,000 person-years — about one extra case per 10,000 person-years of treatment — and a median 22 months from first prescription to the event.[87] A cohort of 174,584 matched pairs found the same direction, with roughly 2.4 times the risk at two and three years and no detectable excess in the first year.[88] The size of the association is now contested. A US network study across 14 databases (OHDSI), with about 810,000 new semaglutide users with type 2 diabetes, found cohort hazard ratios against other diabetes drugs of 1.23–1.44 that did not reach significance under its sensitive case definition; only the comparison with empagliflozin under the specific definition did (HR 2.27), and the within-person self-controlled estimate was a modest 1.32.[89] Two 2026 meta-analyses of observational studies split: one pooled six studies to an adjusted HR of 1.63 (95% CI 0.88–2.39), not significant and of low certainty, with a significant signal in type 2 diabetes (1.70) but none in obesity (0.47, 0.19–1.13);[90] the other, a Bayesian analysis, put the relative risk at 2.52 (credible interval 1.56–4.72).[91] A pooled analysis of placebo-controlled liraglutide and semaglutide trials in 96,829 participants found no excess, about 3 against 6 confirmed cases per 100,000 person-years, though three of its authors are Novo Nordisk employees and its result sits awkwardly with the regulator's statement that trials show a slightly higher risk than placebo.[92] European regulators concluded in June 2025 that NAION is a very rare side effect of semaglutide — up to 1 in 10,000 people — and added it to the product information; the UK followed in February 2026,[93] while the US label still carries no NAION language. Any new painless visual loss on a GLP-1 warrants same-day ophthalmology, and regulators are explicit that treatment should be stopped if NAION is confirmed.
- Muscle and bone loss. Covered above. About 35% of weight lost is fat-free mass — the same proportion as weight loss by diet alone; hip bone mineral density fell about 2.6% over 52 weeks. This is the caution most relevant to a healthy adult using the drug electively.
- Hair loss. Now a labelled common adverse reaction, and strongly sex-skewed. In the trials supporting the 7.2 mg dose, hair loss was reported by 5.8% of patients on 7.2 mg (8.4% of women, 0.2% of men), 3.3% on 2.4 mg (5.4% of women, no men) and 1.0% on placebo (1.5% of women).[94] A target-trial emulation in adults with type 2 diabetes found GLP-1 receptor agonists associated with roughly 40–70% higher risk of non-scarring alopecia against two active comparators (hazard ratio 1.37, 95% CI 1.08–1.73 versus SGLT-2 inhibitors; 1.68, 1.28–2.20 versus DPP-4 inhibitors), on a low absolute base.[95] Whether this is the drug or the rapid weight loss is unresolved, but for a cosmetically motivated elective user it belongs in the decision, particularly for women.
- Two scares that did not survive scrutiny. Suicidality: European regulators concluded in April 2024 that the evidence does not support a causal association between GLP-1 receptor agonists and suicidal or self-injurious thoughts and actions, and in January 2026 the FDA — after reviewing 91 placebo-controlled trials in 107,910 patients and a claims study of more than 2.2 million people — found no increased risk and asked manufacturers to remove the warning; it came off the Wegovy label in February 2026.[96] Thyroid cancer: the rodent C-cell tumour signal has not shown up in people. In a Scandinavian cohort of 145,410 adults starting a GLP-1 receptor agonist versus 291,667 starting a DPP-4 inhibitor, thyroid cancer occurred at 1.33 versus 1.46 cases per 10,000 person-years over a mean 3.9 years — about 7% lower risk, well within the range expected from chance (hazard ratio 0.93, 95% CI 0.66–1.31) — with no excess of medullary thyroid cancer specifically.[97] Follow-up is under four years, so this does not settle multi-decade exposure, and the labelled contraindication for medullary thyroid carcinoma and multiple endocrine neoplasia type 2 (MEN2) stands.
- Contraindications and cautions. The formal labelled contraindications are narrow: personal or family history of medullary thyroid carcinoma (MTC) or MEN2 and prior serious hypersensitivity to semaglutide. Prior severe pancreatitis and active eating disorders are clinical reasons not to start rather than label contraindications. Pregnancy is handled separately and the number matters: semaglutide has a half-life of about a week and stays in the circulation for five to seven weeks after the last dose, so the label directs anyone taking it for weight or cardiovascular indications to stop as soon as pregnancy is recognised and to discontinue at least two months before a planned pregnancy.[98] Frail older adults at high sarcopenia risk should be screened before, not after.
- Drug interactions. Delayed gastric emptying alters the absorption of orally administered drugs — clinically relevant for narrow-therapeutic-index agents and for oral contraceptives around dose escalation. In insulin-treated or sulfonylurea-treated patients, doses need reducing to avoid hypoglycaemia. Anaesthetists now routinely ask about GLP-1 use because of retained gastric contents and aspiration risk, but blanket holding is no longer the advice. The 2024 multisociety guidance — the American Society of Anesthesiologists with the gastroenterology, bariatric and endoscopic surgery societies — says the drug may be continued before a procedure in patients without elevated risk of delayed gastric emptying, and reserves holding for those in the dose-escalation phase, on a high dose, or currently reporting nausea, vomiting or abdominal pain; the mitigations are a liquid diet for at least 24 hours, point-of-care gastric ultrasound if there is same-day concern, and rapid-sequence induction rather than cancellation.[99] In a claims cohort of 366,476 surgical patients, 5,931 of them on a GLP-1 receptor agonist, 30-day postoperative aspiration pneumonia was no more common among users (odds ratio 0.78, 95% CI 0.57–1.06 — a 22% lower rate, but well within the range expected from chance).[100] Tell the anaesthetist well in advance; expect a risk assessment, not an automatic instruction to stop.
- Compounded semaglutide. Beyond the regulatory problem covered under microdosing above, the documented harm is arithmetic: patients drawing doses from multi-dose vials with insulin syringes have given themselves five to twenty times the intended dose, and prescribers converting between milligrams and units have made five- to ten-fold errors, with some patients hospitalised and with pancreatitis and gallstones among the reported consequences.[101] If you use the drug, use the licensed product.
What we don't know yet
- Hard longevity outcomes in healthy adults. Every long-term trial in this class — SELECT, SOUL, FLOW, SURPASS-CVOT, SUMMIT, the STEP-HFpEF programme, STRIDE, SURMOUNT-OSA, ESSENCE, STEP 9 — required an existing disease at entry: cardiovascular disease, diabetes, kidney disease, heart failure, peripheral artery disease, sleep apnoea, steatohepatitis or symptomatic osteoarthritis. No randomised trial has tested whether GLP-1 drugs extend healthspan in metabolically healthy adults.
- Decades-long safety. Pancreatitis, gallbladder pathology, the rodent medullary thyroid signal, gastroparesis, and the NAION vision signal are all documented (see Cautions). Multi-decade exposure data don't exist yet.
- Microdosing dose-response. Plausible but unproven.
- Who shouldn't take it. A personal or family history of MEN2 or medullary thyroid carcinoma is the labelled contraindication; prior severe pancreatitis and active eating disorders are clear clinical reasons not to start; and anyone planning a pregnancy should stop at least two months beforehand. Frail older adults at high sarcopenia risk should be screened carefully and only treated with aggressive scaffolding.
Bottom line for a healthy midlife adult
For a healthy adult at a normal BMI seeking longevity benefit alone, the case is interesting but not yet made. The epigenetic clock signal is intriguing, but it's one trial, in one population, with one dose. Microdosing is plausible, regulatorily grey, and largely untested.
For an adult with overweight/obesity, established cardiovascular risk, or metabolic dysfunction, the case is much stronger: hard MACE reduction, durable weight loss, plus the longevity-relevant biomarker shifts on top.
In either case, the single most important sentence in this article: a GLP-1 without protein and resistance training is a slow path to sarcopenic obesity, not longevity. If you are not willing to lift weights and hold protein at 1.2–1.6 g/kg/day (or 80–120 g/day), do not start. If you are, then it becomes one of the more interesting tools in current geroscience.
Further reading
- SELECT — semaglutide 2.4 mg cut major adverse cardiovascular events from 8.0% to 6.5% over a mean of 39.8 months in about 17,600 non-diabetic adults with overweight or obesity and established cardiovascular disease, roughly a 20% relative reduction (hazard ratio 0.80, 95% confidence interval 0.72–0.90); hard-outcome randomised trial, Novo Nordisk-funded.[102]
- Ryan DH et al. Long-term weight loss effects of semaglutide in obesity without diabetes in the SELECT trial. Nat Med 2024.[103]
- Non-diabetic semaglutide weight-loss meta-analysis (7 RCTs, n=5,411).[104]
- OASIS 4 — oral semaglutide 25 mg for non-diabetic weight loss.[105]
- Wharton S et al. Gastrointestinal tolerability of once-weekly semaglutide 2.4 mg in adults with overweight or obesity. Diabetes Obes Metab 2022.[106]
- He L et al. Association of GLP-1 Receptor Agonist Use With Risk of Gallbladder and Biliary Diseases: systematic review and meta-analysis of 76 RCTs. JAMA Intern Med 2022.[107]
- Sodhi M et al. Risk of Gastrointestinal Adverse Events Associated With GLP-1 Receptor Agonists for Weight Loss. JAMA 2023.[108]
- Hathaway JT et al. Risk of Nonarteritic Anterior Ischemic Optic Neuropathy in Patients Prescribed Semaglutide. JAMA Ophthalmol 2024.[109]
- Hsu AY et al. Semaglutide and Nonarteritic Anterior Ischemic Optic Neuropathy Risk Among Patients With Diabetes. JAMA Ophthalmol 2025.[110]
- Urbina S et al. Micronutrient and Nutritional Deficiencies Associated With GLP-1 Receptor Agonist Therapy: A Narrative Review. Clin Obes 2026.[111]
- SURMOUNT-5 — tirzepatide vs. semaglutide head-to-head, 20.2% vs. 13.7% weight loss (a 6.5-percentage-point gap); open-label RCT on a surrogate endpoint, Eli Lilly-funded.[112]
- Corley MJ et al. Semaglutide slows epigenetic aging in a randomized trial of HIV-associated lipohypertrophy. Nat Commun 2026.[113]
- Ortiz GU et al. Semaglutide as a possible therapy for healthy aging: Targeting the hallmarks of aging. Ageing Res Rev 2024.[114]
- STEP-HFpEF — exercise capacity in obesity-related HFpEF, pooled programme analysis.[115]
- STRIDE — semaglutide and walking capacity in symptomatic peripheral artery disease with type 2 diabetes; phase 3b RCT.[116]
- evoke / evoke+ — primary endpoint missed in early Alzheimer's disease.[117]
- Taipale H et al. Association between GLP-1 receptor agonist use and worsening mental illness in people with depression and anxiety in Sweden: a national cohort study. Lancet Psychiatry 2026.[118]
- BELIEVE — semaglutide + bimagrumab, 92.2% of weight lost was fat at week 72; phase 2 RCT.[119]
- Hansen MS et al. Once-weekly semaglutide versus placebo in adults with increased fracture risk: a phase 2 trial (hip BMD −2.6%, lumbar −2.1%, tibial cortical −1.8%). EClinicalMedicine 2024.[120]
- Liu Y et al. Skeletal effect of semaglutide and tirzepatide — bone loss greater in non-diabetics than in type 2 diabetes. J Clin Endocrinol Metab 2026.[121]
- Nutritional priorities advisory (ACLM/ASN/OMA/TOS).[122]
- LEAN-PREP muscle-preservation protocol trial.[123]
- Wilding JPH et al. Weight regain after withdrawal of semaglutide: STEP 1 extension. Diabetes Obes Metab 2022.[124]
- FDA, FDA's Concerns with Unapproved GLP-1 Drugs Used for Weight Loss (content current 1 September 2026).[125]