- Yes, you lose lean mass — and you lose far more fat. In the SURMOUNT-1 body composition substudy, tirzepatide cut body weight 21.3%, fat mass 33.9% and lean mass 10.9% over 72 weeks
- Body composition improves on average. A 2026 meta-analysis of seven randomised trials in 821 patients found lean mass as a share of total weight went up by 1.81 percentage points
- Function improved, it didn't decline. In 106 patients on semaglutide, grip strength rose 4.5 kg at a year and sarcopenic obesity fell from 49% to 33%. Fall and fracture risk data in older adults also looks favourable
- The real danger is the cycle, not the drug. A year after semaglutide was stopped, participants had regained two-thirds of their lost weight — and regained weight is mostly fat. Down as muscle, back as fat, repeat
- Protein is the hard part. Trials point to roughly 1.2–1.6 g/kg/day, but these drugs suppress appetite. Hitting the number requires planning it, not waiting to feel hungry
- Lifting is the lever with the best evidence. In 160 obese older adults losing weight, combined aerobic and resistance training improved physical function 21% versus 14% for either alone
- A muscle-sparing drug is already in phase 2. Bimagrumab plus semaglutide produced 17.8 kg of weight loss versus 14.2 kg for semaglutide alone in a 507-person trial
If you've read anything about GLP-1 drugs in the last year, you've absorbed one of two stories. Either these medications are melting people's muscle away and creating a generation of frail, skinny-fat forty-somethings — or the concern is manufactured panic from people who sell protein powder.
The trial data supports neither. It supports something more useful and more specific: the average outcome is better body composition, not worse; the genuine risk is concentrated in a minority of people who are identifiable in advance; and the biggest hazard is something almost nobody writes about, which is what happens when you stop.
Here is what was actually measured. (For what these drugs may do to aging itself, separate from weight, see our piece on Ozempic and longevity.)
Part 1: What the trials found
The best body composition data comes from DXA substudies inside the big trials — actual scans, not estimates.
In SURMOUNT-1, 160 of the 2,539 participants had DXA scans at baseline and at week 72. They were 73% female, averaging 102.5 kg and a BMI of 38. On tirzepatide, body weight fell 21.3%, fat mass fell 33.9%, and lean mass fell 10.9%. The placebo group lost 5.3% of weight, 8.2% of fat and 2.6% of lean mass.
Bar widths show the approximate share of tissue lost, not absolute kilograms. The striking detail is the bottom row: the placebo group lost tissue in almost exactly the same proportion. This split isn't a drug effect — it's what weight loss looks like.
That last point is the one that gets dropped from headlines. Losing roughly a quarter of your total weight loss as lean tissue is unremarkable. It happens with dieting, with bariatric surgery, with any sustained energy deficit. The study authors noted the fat-to-lean ratio held steady across sex, age and BMI subgroups. The drugs aren't doing something metabolically strange to muscle. They're producing a lot of weight loss, and weight loss has a composition.
Part 2: The case that this is fine
Three separate lines of evidence argue the alarm is overstated.
Body composition improves as a ratio. A 2026 systematic review and meta-analysis pooled seven randomised trials covering 821 patients with obesity treated at obesity doses versus placebo. Lean mass as a proportion of total body weight rose by 1.81 percentage points (95% CI 1.1–2.52). The authors concluded plainly that lean mass loss "should not be considered a limitation for the use of these drugs" — while adding, in the same breath, that nutrition and exercise support is essential. Both halves of that sentence are load-bearing.
Function got better, not worse. This is the finding that should have ended the panic and didn't. The SEMALEAN study followed 115 patients with obesity on semaglutide 2.4 mg, 106 of whom completed 12 months, with a mean starting BMI of 46.3. Lean mass dropped 3 kg in the first seven months and then stabilised. Handgrip strength — a hard functional measure that predicts mortality — improved by 4.5 kg. And the proportion of patients meeting criteria for sarcopenic obesity fell from 49% at baseline to 33% at a year. People got stronger relative to their size, not weaker.
Falls and fractures didn't rise. A 2026 analysis found lower fall and femoral fracture risk with semaglutide and tirzepatide compared with DPP-4 inhibitors in older adults with type 2 diabetes. If these drugs were systematically producing frailty, this is where you would expect it to show up first.
Someone with obesity is often already carrying low muscle relative to their body mass — that's what sarcopenic obesity means, and half the SEMALEAN cohort met the criteria at baseline. Losing 20% of body weight while losing 11% of lean tissue leaves you with more muscle per kilogram of body weight than you started with. Your legs are carrying less. That is the mechanism behind grip strength going up while the scale and the DXA lean number both go down.
Part 3: Who is genuinely at risk
An average is not a person. The population-level answer is reassuring; the individual answer depends on where you're starting and how you're doing it.
- Already low on muscle. Someone in their seventies with modest muscle mass has less to spare than someone in their forties. The same 10% lean loss lands differently
- Losing very fast. Rapid loss shifts the composition of what you lose. The trials above produced their results over 72 weeks, not 20
- Not eating enough protein. This is the most common and most fixable failure, and the drug itself causes it — appetite suppression is the mechanism of action
- Not doing resistance training. The single biggest controllable variable, and the one most people skip
- Bone, separately. A 2026 review found GLP-1 effects on bone are neutral-to-modestly-positive in type 2 diabetes, but in people without diabetes, greater weight loss was more consistently associated with declines in bone mineral density and increased bone turnover. Follow-up has been short. This is the least settled part of the picture and it matters most for postmenopausal women, who are already losing bone for other reasons
The honest summary is that the risk isn't in the molecule — it's in how the weight loss is executed. Which is good news, because execution is the part you control.
Part 4: The thing almost nobody writes about
Here is the finding that should change how you think about this entire category.
The STEP 1 trial extension followed 327 participants for a year after semaglutide was withdrawn. During treatment they had lost 17.3% of body weight. One year off the drug, they had regained 11.6 of those percentage points — about two-thirds of everything lost — finishing at a net 5.6% below where they started. Cardiometabolic improvements drifted back toward baseline along with the weight.
Now put the two facts together. On the way down, roughly a quarter of what you lose is lean tissue. On the way back up, regained weight is preferentially fat — that's well established across the weight-cycling literature and it is not controversial. Run that loop once and you finish near your old weight with less muscle and more fat than you had before. Run it twice and the arithmetic gets worse.
This reframes the muscle question entirely. The reason to protect lean tissue on a GLP-1 isn't that the drug is dangerous while you're taking it. The evidence says it mostly isn't. The reason is that many people will stop — cost, supply, side effects, insurance, a decision that they're done — and muscle is the thing you cannot easily get back, while fat is the thing that returns on its own.
Treat the months you're on the medication as a window, not a destination. Appetite is suppressed, weight is falling, and training is easier at a lower body weight than it has been in years. That is the best opportunity most people will ever get to build the muscle that makes the next chapter survivable. Spending that window doing nothing but taking the injection is the actual mistake.
Your protein target while losing weight
Dosed the way clinicians actually do it for people with obesity — on an adjusted body weight, not your scale weight, because 1.6 g per kilo of actual weight produces a number nobody can eat. Shows the target, what it looks like per meal, and how hard it will be at the calorie intake these drugs tend to produce.
Part 5: What actually protects muscle
Ranked by how strong the evidence is, and with one honest caveat up front: most of this evidence comes from weight loss generally, not from GLP-1 users specifically. The trials testing these interventions during semaglutide and tirzepatide therapy are running now — one of them, LEAN-PREP, is a randomised trial doing exactly that — but they have not reported yet. What follows is the best available inference, not a finished answer.
Resistance training — and cardio too
In a 2017 NEJM trial, 160 obese older adults losing weight were randomised to aerobic training, resistance training, combined training, or a control group. The combined group improved physical performance by 21%, against 14% for aerobic alone and 14% for resistance alone. An earlier NEJM trial in 107 obese older adults found diet plus exercise beat diet alone (21% vs 12%) and exercise alone (15%).
The lesson people take from this is usually "lift weights." The actual lesson is "do both." Resistance training holds onto the tissue; aerobic work is what a large share of the functional improvement runs through.
In practice: two to three full-body resistance sessions a week, plus walking most days. Start earlier in your treatment than feels necessary.Enough protein, deliberately
A meta-analysis of 47 studies in 3,218 adults with overweight or obesity found higher protein intake significantly prevented muscle mass decline during weight loss. In older adults specifically, a review of 20 randomised trials found they retained more lean mass and lost more fat on higher-protein diets. And a meta-analysis of 74 trials pinned the effective doses: 1.2–1.6 g/kg/day at 65 and over, and 1.6 g/kg/day or more under 65 when combined with resistance training.
The complication unique to this drug class is that the medication removes the signal that normally gets you to eat. Protein has to become a scheduled task rather than a response to hunger — which is exactly the kind of thing that sounds trivial and is the reason most people miss the target.
In practice: protein first at every meal, aim for a similar amount at each, and accept that a shake is often the difference between hitting the number and not.Don't rush the descent
The trial results above came from 68 to 72 weeks of treatment. The composition of weight loss shifts unfavourably when loss is very rapid, which is a general principle of energy restriction rather than anything specific to these drugs. Escalating dose faster than the label schedule to accelerate results is trading tissue you want for a number on a scale.
This is a prescriber conversation, not a self-directed one — but it's worth knowing that patience has a physiological payoff here, not just a tolerability one.
Creatine, if you're training
Creatine monohydrate has the strongest evidence of any training supplement for supporting lean mass and strength, and nothing about GLP-1 therapy changes how it works. It's not a muscle-preservation drug and shouldn't be sold as one, but if you're lifting through a weight loss phase it's among the few supplements with real support behind it — our guide to creatine for women covers the dosing and the specific evidence.
The supplement aisle's answer
HMB, BCAAs, collagen and various "muscle preservation" blends are being aggressively marketed to GLP-1 users right now. None has evidence in this population, and BCAAs in particular add little when total protein is adequate. Meeting your protein target with food does everything they claim to do, for less money.
Part 6: What's coming
The pharmaceutical industry has clearly decided this is a real problem worth solving, which is itself informative.
Bimagrumab is an antibody that blocks activin type II receptors — the signalling pathway that limits muscle growth. In a phase 2 trial published in 2026, 507 adults with obesity were randomised across nine groups for 48 weeks. Bimagrumab alone produced 9.3 kg of weight loss, semaglutide 2.4 mg produced 14.2 kg, and the combination produced 17.8 kg, against 3.3 kg on placebo. Side effects included muscle spasms, diarrhoea and acne.
Several other muscle-sparing agents are in development on the same logic. It's early — phase 2 is a long way from your pharmacy — but the direction is clear: the next generation of these drugs will likely be paired with something that protects lean tissue. In the meantime, the tools that do the same job are protein and a barbell, and they're available now.
- This is not a reason to stop, delay or avoid a medication your doctor has prescribed. Obesity is a disease with its own mortality, and these drugs treat it more effectively than anything that came before
- Nothing here is dosing guidance, and no article should change your titration schedule — that is a conversation with your prescriber
- The point is narrow and practical: if you are going to do this, do the two things that make the result durable, and start them early
The drug isn't the risk. Stopping without having built anything is.
On the evidence as it stands, GLP-1 drugs improve body composition on average: much more fat leaves than lean tissue, the ratio of muscle to body weight goes up, grip strength improves, and fracture risk doesn't rise. The frailty story that spread through social media is not what the trials found.
But averages hide the people at the edges, and one specific hazard is real and badly under-discussed. Two-thirds of the weight comes back within a year of stopping, and it comes back as fat. Lose a quarter of your weight loss as muscle, regain it as fat, and one round leaves you a little worse off than you started. That is the scenario worth defending against.
The defence is unglamorous and well established: enough protein, scheduled rather than left to an appetite the drug has switched off, and resistance training started early rather than promised later. The months on the medication are the easiest training window most people will ever have. Use them for something.
References
Every figure above traces to one of these. Links go to the published record.
| Tirzepatide body composition | SURMOUNT-1 DXA substudy, Diabetes Obes Metab 2025 — 160 participants scanned at baseline and week 72. PMID 39996356 |
| Lean mass meta-analysis | Int J Obes 2026 — 7 RCTs, 821 patients; lean mass as proportion of weight +1.81%. PMID 42321502 |
| Muscle function on semaglutide | SEMALEAN, Diabetes Obes Metab 2026 — 106 patients completed; grip +4.5 kg, sarcopenic obesity 49%→33%. PMID 41068996 |
| Weight regain after stopping | STEP 1 trial extension, Diabetes Obes Metab 2022 — 327 participants followed one year off treatment. PMID 35441470 |
| Exercise type in dieting older adults | Villareal et al. N Engl J Med 2017 — 160 obese older adults randomised to aerobic, resistance, combined or control. PMID 28514618 |
| Weight loss plus exercise | Villareal et al. N Engl J Med 2011 — 107 obese adults 65 and older. PMID 21449785 |
| Protein doses for lean mass | Nunes et al. J Cachexia Sarcopenia Muscle 2022 — meta-regression of 74 RCTs. PMID 35187864 |
| Protein during weight loss | Clin Nutr ESPEN 2024 — 47 studies, 3,218 adults with overweight or obesity. PMID 39002131 · Older adults, Nutr Rev 2016 PMID 26883880 |
| Muscle-sparing drug | Bimagrumab plus semaglutide phase 2, Nature Medicine 2026 — 507 adults, 48 weeks. PMID 41772149 |
| Trial in progress | LEAN-PREP protocol, BMJ Open 2026 — resistance exercise and protein during semaglutide and tirzepatide therapy. PMID 42020128 |
| Bone | GLP-1 receptor agonists and bone metabolism, J Clin Endocrinol Metab 2026. PMID 42563411 |
| Falls and fractures | Osteoporos Int 2026 — semaglutide and tirzepatide vs DPP-4 inhibitors in older adults with type 2 diabetes. PMID 42435064 |
| Sarcopenic obesity criteria | ESPEN and EASO consensus statement, Obes Facts 2022. PMID 35196654 |
| Older adults and sarcopenia risk | Scoping review, Curr Nutr Rep 2026. PMID 42303931 |
- This article is educational and is not medical advice. It was written by a layperson who reads the primary literature carefully, not by a physician or dietitian, and it has not been clinically reviewed
- Nothing here is a reason to start, stop, delay or change the dose of any medication. GLP-1 receptor agonists are prescription drugs and those decisions belong with your prescriber
- Protein targets are not universal. Kidney disease, liver disease and pregnancy all change them. Check yours with a clinician who knows your history
