Every molecule I’ve written about in this series had a headline number. 28.3%. 20.7%. 12.4%. Numbers that would have been fantasy a decade ago.
But when I sit across from a patient at MetaboHub, those aren’t the numbers she asks about. She asks two things, and she asks them in the same breath.
“Will I lose muscle?”
“And what happens when I stop?”
I’ve been meaning to write this one for a while, because both questions have a real evidence base now, and because both are being answered badly online. One camp says GLP-1s are “melting your muscle.” The other says regain is proof the drugs “don’t work.” Both takes are in my feed every day. Both are wrong, and they’re wrong for the same reason.
Let me walk you through the data.
Part One: The Muscle Question
What the DXA substudies actually show
Start with the trials that measured body composition properly, with DXA, not with a bathroom scale that guesses.
STEP 1 (semaglutide 2.4 mg), DXA substudy, 68 weeks:
Total weight: −15.0%
Fat mass: −19.3%
Visceral fat: −27.4%
Lean mass: −9.7%
SURMOUNT-1 (tirzepatide), DXA substudy, 72 weeks:
Total weight: −21.3% vs −5.3% placebo
Fat mass: −33.9% vs −8.2%
Lean mass: −10.9% vs −2.6%
In SURMOUNT-1, roughly a quarter of the weight lost was lean mass and three-quarters was fat. Here’s the part almost nobody quotes: the placebo group, losing weight on lifestyle advice alone, showed the same 75/25 split. That ratio is not a tirzepatide effect. It is what weight loss looks like.
And in STEP 1, the proportion of the body that was lean tissue went up by three percentage points. Patients finished the trial with less muscle in absolute terms and a higher muscle-to-fat ratio. Both things are true at once.
Why some lean mass loss is expected, and fine
This is the point I most want to land, so I’ll say it plainly.
A 60 kg body does not need the muscle a 100 kg body needed.
When someone carries an extra 40 kg, not all of it is fat. Forbes described this decades ago: for every kilogram gained in adulthood, roughly a quarter is fat-free mass. The body builds the legs, the back, the heart muscle and the organ mass it needs to carry the load. That’s not a flaw. That’s an adaptive system doing its job.
So when the load comes off, some of that support structure comes off too. Bodies are smart like that. What you are watching in the DXA data is not a drug stripping muscle. It’s a body right-sizing itself to a smaller frame, in exactly the proportion it added it in the first place.
Three technical caveats before anyone quotes me on that.
First, “lean mass” on DXA is not “skeletal muscle.” It’s everything that isn’t fat or bone: water, organs, connective tissue, and yes, muscle. Losing visceral fat pulls water and organ mass with it. Part of that −10% is not muscle at all.
Second, mass is not function. In STEP-HFpEF, patients with obesity-related heart failure lost weight, lost lean mass, and walked about 20 metres further in six minutes than placebo, with a 7.8-point improvement in symptom scores. Less muscle carrying much less weight can mean more capability, not less.
Third, and this is where the concern is legitimate: the average hides the tails. The patient I actually worry about is the 60-year-old woman with sarcopenic obesity, low protein intake, and no resistance training, losing 20% of her body weight in a year. For her, the “proportional” lean loss starts from a base that was already too low. That is a real clinical problem. It’s just not the problem the internet is describing.
The muscle-sparing pipeline, and what Lilly did with it
The industry has heard the question, and it’s answering it with pharmacology.
Bimagrumab (activin type II receptor antibody), BELIEVE trial, 72 weeks:
Semaglutide alone: −15.7% weight, fat −27.8%, lean −7.4%
Bimagrumab alone: −10.8% weight, fat −28.5%, lean +2.5%
Bimagrumab + semaglutide: −22.1% weight, fat −45.7%, lean −2.9%
Read that middle row again. Bimagrumab on its own lost fat and gained lean mass. The combination lost nearly half of total fat mass while giving up under 3% of lean tissue. Those are remarkable numbers, and they were published in Nature Medicine this year.
Then Lilly, which paid roughly $1.9 billion for the molecule, cancelled the Phase 2b trial of bimagrumab plus tirzepatide in September 2025, before enrolment started, “for strategic reasons.” No data problem was cited. I don’t know what the strategy is, and I’m not going to guess in print. But when the company with the best combination data in the field pulls the combination trial, that tells you the commercial question (”will payers fund a second injection to protect muscle?”) is not yet settled.
Others are still running:
Apitegromab (Scholar Rock, EMBRAZE): added to tirzepatide, weight lost was 85% fat / 15% lean versus 70/30 on tirzepatide alone. Total weight loss was slightly lower with apitegromab (−12.3% vs −13.4%), which is exactly what you’d expect if you’re keeping the heavier tissue.
Trevogrumab ± garetosmab (Regeneron, COURAGE): lean mass fell from 33% of weight lost on semaglutide alone to 7% on the triple combination. Two deaths occurred in the triple-combination arm. Small trial, causality unclear, but this is the class where I want to see safety before I get excited about DXA.
Enobosarm (Veru, QUALITY): positive Phase 2b topline, heading for Phase 3.
The intervention that already works, costs nothing, and nobody prescribes
The S-LiTE trial randomised people after an 8-week diet (−13 kg) to liraglutide, supervised exercise, both, or placebo for a year. The combination arm kept the most weight off, cut body-fat percentage roughly twice as much as either intervention alone, and preserved lean mass. In the one-year follow-up after all treatment stopped, the groups that had exercised held on to their body-composition gains; the drug-only group regained.
That study is five years old. It is still the best answer to the muscle question that exists. Resistance training two or three times a week plus adequate protein is not a “lifestyle add-on.” It is part of the treatment, and if we don’t prescribe it with the same seriousness as the dose titration, we are under-treating.
Part Two: What Happens When You Stop
The trial data is consistent, and it’s stark
STEP 1 extension: one year after stopping semaglutide, participants regained about two-thirds of what they’d lost. −17.3% at week 68 became −5.6% at week 120. Blood pressure, lipids and HbA1c drifted back toward baseline with the weight.
STEP 4: patients switched from semaglutide to placebo at week 20 gained back 6.9% by week 68; those who continued lost a further 7.9%. A 14.8-point difference from a single decision.
SURMOUNT-4: patients switched from tirzepatide to placebo at week 36 regained 14.0% by week 88; continuers lost a further 5.5%. A 19.5-point gap.
The 2026 meta-regression in eClinicalMedicine pooled 48 studies and modelled the curve: roughly 60% of lost weight regained at one year, plateauing around 75%, with about 5% of original body weight kept in the long run. Notably, the authors could not assess the body composition of the regain, because almost nobody measured it. So the honest answer to “do you regain fat or muscle?” is: we don’t know from the trial data, and anyone telling you confidently either way is guessing.
The real world is messier, and more hopeful
Two 2026 datasets complicate the trial picture in a useful way.
Cleveland Clinic followed 7,938 patients who stopped a GLP-1. One year later, the obesity group had regained on average just 0.5%. Why the gap with the trials? Because in real life, stopping is not stopping. Within 12 months, 27% had switched to a different obesity medication and 20% had restarted the original. The drug was withdrawn; the treatment wasn’t.
A separate analysis of about 135,000 patients found that six months after discontinuing, a third had maintained, a third had continued losing, and a third had regained. Median change: zero.
And here’s the persistence data, from Prime Therapeutics: 63% of people who started a modern obesity GLP-1 in early 2024 were still on it at one year, up from 33% for 2021 starters. At three years, 8%.
This is the part I actually want you to take away
Put the trial data and the real-world data together and one picture emerges: people go on, come off, go back on, switch molecules, take breaks for cost or side effects or pregnancy or supply, and come back. On and off, for a very long time. Like it or not.
And I want to be clear that this is not a failure of the drugs. It is the natural history of the disease being treated.
The World Obesity Federation called obesity a “chronic relapsing progressive disease process” back in 2017. The Lancet Commission on Clinical Obesity said it again in 2025. We already know this. We don’t behave as if we know it.
Nobody says a statin “doesn’t work” because your LDL comes back when you stop. Nobody calls antihypertensives a “temporary fix.” Regain after discontinuation is not evidence against GLP-1s. It’s evidence that obesity is what the guidelines say it is, and that we should stop designing treatment, and pricing, and patient expectations, around a 12-month cure.
Which brings the two questions together. If a patient is going to be on therapy, in some form, on and off, for decades, then the muscle question isn’t “how much lean mass did she lose in trial week 72.” It’s “what does her body composition look like at 60, after three cycles of loss and partial regain, and did anybody ever tell her to lift something.”
That’s the question I want the muscle-sparing trials to answer. Not 72 weeks. Ten years.
What I’m Watching
Body composition of regain. The single biggest gap in the literature. Every withdrawal trial from here on should carry a DXA substudy. If regain is preferentially fat, as the physiology suggests it might be, then cycling on and off is worse for body composition than either staying on or never starting, and that changes how we counsel patients.
Lilly’s bimagrumab strategy. The BELIEVE data are too good to shelve. Watch what they do with it and in which population.
Safety in the myostatin/activin class. Two deaths in COURAGE’s triple arm need an explanation before the class scales.
Maintenance dosing. There is no good RCT of lower-dose or intermittent maintenance. Clinicians are doing it anyway. Somebody needs to study what everybody is already doing.
Long-term persistence. 63% at one year is a huge improvement. 8% at three years is the real number for a chronic disease, and it’s the one payers and clinicians should be planning around.
The Bigger Picture
The clinical researcher in me looks at the DXA data and sees a body doing exactly what bodies do, in exactly the proportion the physiology predicts. The pharmacist in me looks at the withdrawal trials and sees a chronic disease behaving like a chronic disease. And the evidence nerd in me is frustrated that after five years and tens of thousands of patients, nobody has measured what the regained weight is made of.
Some muscle loss is expected. Some regain is expected. Neither is the scandal it’s being sold as. The scandal would be treating a lifelong, relapsing condition as a one-time fix, and then blaming the patient, or the molecule, when it relapses.
This is metabolic medicine’s moment. And it’s just getting started. ✌️
Key References
Wilding JPH, et al. Once-Weekly Semaglutide in Adults with Overweight or Obesity (STEP 1). N Engl J Med. 2021;384:989-1002 (DXA substudy in supplementary appendix).
Look M, et al. Body composition changes during weight reduction with tirzepatide in the SURMOUNT-1 study of adults with obesity or overweight. Diabetes Obes Metab. 2025.
Forbes GB. Lean body mass–body fat interrelationships in humans. Nutr Rev. 1987;45:225-231.
Kosiborod MN, et al. Semaglutide in Patients with Heart Failure with Preserved Ejection Fraction and Obesity (STEP-HFpEF). N Engl J Med. 2023;389:1069-1084.
Heymsfield SB, et al. Bimagrumab with or without semaglutide in adults with overweight or obesity (BELIEVE). Nat Med. 2026.
Scholar Rock. Phase 2 EMBRAZE trial results, apitegromab with tirzepatide. Press release, June 2025.
Regeneron. Phase 2 COURAGE trial results, trevogrumab and garetosmab with semaglutide. Press release, 2025.
Lundgren JR, et al. Healthy Weight Loss Maintenance with Exercise, Liraglutide, or Both Combined (S-LiTE). N Engl J Med. 2021;384:1719-1730.
Jensen SBK, et al. Healthy weight loss maintenance with exercise, GLP-1 receptor agonist, or both combined followed by one year without treatment. eClinicalMedicine. 2024.
Wilding JPH, et al. Weight regain and cardiometabolic effects after withdrawal of semaglutide: The STEP 1 trial extension. Diabetes Obes Metab. 2022;24:1553-1564.
Rubino D, et al. Effect of Continued Weekly Subcutaneous Semaglutide vs Placebo on Weight Loss Maintenance (STEP 4). JAMA. 2021;325:1414-1425.
Aronne LJ, et al. Continued Treatment With Tirzepatide for Maintenance of Weight Reduction (SURMOUNT-4). JAMA. 2024;331:38-48.
Trajectory of weight regain after cessation of GLP-1 receptor agonists: systematic review and meta-regression. eClinicalMedicine. 2026.
Cleveland Clinic. Real-world weight trajectories after GLP-1 discontinuation (n=7,938). Diabetes Obes Metab. 2026.
Prime Therapeutics. GLP-1 persistence for obesity at one and three years. June 2025.
Bray GA, Kim KK, Wilding JPH. Obesity: a chronic relapsing progressive disease process. Obes Rev. 2017;18:715-723.
Lancet Diabetes & Endocrinology Commission on the definition and diagnosis of clinical obesity. 2025.
Disclaimer: Bimagrumab, apitegromab, trevogrumab, garetosmab and enobosarm are investigational compounds not approved by the FDA, EMA or any other regulatory agency. Several figures above come from company press releases or conference presentations and have not been peer reviewed. This article is for educational and informational purposes only and does not constitute medical advice. Always consult a qualified healthcare professional before starting, stopping, or changing any medication.