GLP-1 Help

Muscle loss · Weight loss on GLP-1 · Question 15 of 35

Is 40% of GLP-1 weight loss really muscle?

The number is real, but it is not what it sounds like. It comes from one small scan substudy, it measures "lean mass" rather than muscle, and the one trial that used MRI to look at actual muscle found something very different.

By Dr. Chiranjeevi Peetla · Updated November 2026 · 9 min read · 52 sources, each linked

Where the 40% comes from

In the STEP 1 trial, about 1,960 adults with obesity took semaglutide 2.4 mg (sold as Wegovy) or a placebo injection for 68 weeks. Average weight loss was 14.9% on the drug and 2.4% on placebo.[1] A small group of 140 of them also had DXA scans. DXA is the X-ray scan that splits the body into fat, bone and "everything else". That "everything else" is called lean mass.[2]

In those 140 people, fat mass fell 10.4 kg and lean mass fell 6.9 kg on semaglutide (versus 1.2 kg and 1.5 kg on placebo).[1][2] Divide 6.9 by the 17.3 kg total and you get 40%: roughly 60% of the weight lost was fat and roughly 40% was lean mass. A second way of counting in the same appendix gives 38.6%, which is why you see both numbers quoted.[1] That is the "40% muscle" figure. It was never published as a main result; it sits in an abstract and a supplementary table.

STEP 1 DXA substudy (n=140): what the lost weight was made of · semaglutide 2.4 mg, 68 weeks

~60% fat~40% lean mass

Lean mass = water + organs + connective tissue + muscleApproximate; derived from reported % changes

SURMOUNT-1 DXA substudy (n=160): tirzepatide · 72 weeks

74% fat26% lean

SURMOUNT-1 DXA substudy: placebo · same trial, same 72 weeks

~75% fat~25% lean

The fat-to-lean split was the same with and without the drugLook et al. 2025

Three DXA substudies, three different splits. The share of lean mass depends on who was scanned and how much they lost, not on the drug.

The second-largest obesity drug trial, SURMOUNT-1 (tirzepatide, sold as Zepbound), ran its own DXA substudy in 160 people. There, lean mass made up 26% of the weight lost on tirzepatide (5.6 kg of lean against 15.9 kg of fat), and 25% of the much smaller loss on placebo.[4] The split was the same whether people were on the drug or not. The proportion is not drug-specific.[4]

Why the two substudies disagree. Different people, different drugs, different starting weights, and 140 or 160 participants each. Small scan substudies are noisy. The honest range for "share of weight lost that was lean mass" on these drugs is about 25 to 40%, which is where diet and surgery studies also land.[13][38]

"Lean mass" is not muscle

DXA cannot see muscle. Everything that is not fat or bone goes in one bucket: skeletal muscle, but also water, blood, organs such as the liver and kidneys, skin and connective tissue.[16] Skeletal muscle is only about half of lean mass.[16]

Several parts of that bucket are expected to shrink when weight falls, and have nothing to do with muscle wasting:

  • Water. Fat-free tissue is about 74% water, so any shift in hydration shows up as a change in "lean".[16] In a worked example from a body-composition methods paper, a 9.5 kg weight loss came with 1.5 kg of fat-free mass loss; of that, about 0.7 kg was water inside muscle and only about 0.15 kg was actual muscle protein.[16]
  • The scaffolding of fat tissue. Fat tissue is itself about 15 to 20% "fat-free" (water, cells, connective tissue). Losing fat tissue therefore removes some lean mass by definition.[16] In SURMOUNT-1, that alone accounts for roughly 2.8 kg of the 5.6 kg lean-mass loss.[4][16]
  • Organs and other tissue. In the CALERIE calorie-restriction study (no drugs), lean tissue was about 17.5% of what was lost, and organ volumes shrank along with body size.[36]

Putting that together, the NIH-funded nutrition advisory for GLP-1 users estimates that skeletal muscle was about half of the lean-mass loss in STEP 1, so roughly 20% of the weight lost.[18] "40% of the loss was lean mass" is closer to "about 20% of the loss was muscle", and even that figure is an estimate.

What MRI found when it looked at muscle directly

MRI can measure muscle itself, and it can tell how much of the "muscle" is actually fat threaded through the muscle. Only one GLP-1-class trial has done this at scale: the SURPASS-3 MRI substudy of tirzepatide in people with type 2 diabetes, published in 2025.[5]

After 52 weeks on tirzepatide, thigh muscle volume fell by about as much as a model built on UK Biobank scans predicts for that amount of weight loss, no more.[5] At the same time, the fat threaded through the muscle (muscle fat infiltration) fell by more than the model predicted. The authors described the change as adaptive: the muscle got smaller in proportion to the body it had to carry, and it got leaner.[5]

A 2026 review of the MRI data puts the scale in perspective: thigh muscle volume fell about 5.5 to 7.3% while body weight fell 9.0 to 11.9%.[52] Muscle shrank less than the body did. That is a very different picture from "40% muscle".

Direct MRI measurements of muscle in people losing weight by diet alone give similar proportions: skeletal muscle made up about 22% of the weight lost in men and about 11% in women.[35] The NIH-funded nutrition advisory for GLP-1 users models muscle at 10 to 15% of weight lost in women and 20 to 25% in men.[18]

Is this normal for weight loss?

Yes. Losing some lean mass is part of losing a lot of weight by any method. A long-standing rule of thumb from diet studies is that about a quarter of weight lost is fat-free mass, and its own authors call it "at best an approximation".[17] An older systematic review found fat-free mass was about 14% of the loss with low-calorie diets, 23% with very-low-calorie diets and 18 to 31% after surgery.[38] In the Look AHEAD lifestyle trial, lean mass was about 29% of the first-year loss.[37] Bariatric surgery sits in the same range.[34]

The GLP-1 numbers fit inside that range. A 2024 network meta-analysis of 22 trials found lean mass was about 25% of the weight lost on GLP-1-class drugs, with the lean share of body weight staying stable.[12] A 2026 meta-analysis of 20 trials using DXA or MRI found lean mass was about 26% of weight lost overall, with no significant difference between incretin drugs and intensive lifestyle programs.[13]

What is different with these drugs is the amount of weight lost. Losing 15 to 20% of body weight means losing more of everything, lean mass included, than losing 5%. The share is ordinary; the total is large.

One more detail: in the STEP 1 substudy, lean mass as a share of body weight actually went up by 3 percentage points, because fat fell faster than lean mass.[2] People finished the trial proportionally more muscular than they started.

Does strength or function drop?

This is the question the big trials did not answer well. The STEP 1 and SURMOUNT-1 DXA substudies measured no strength or function at all.[2][4] In the full trials, people reported better physical function on the drugs than on placebo (SF-36 physical functioning +2.2 vs +0.4 points in STEP 1).[1][3] Where objective tests were done in later trials, sit-to-stand performance did not differ from placebo (STEP UP, REDEFINE 1) and grip strength was mostly unchanged (BELIEVE).[6][7][9] A 2026 meta-analysis found lean mass fell more on the drugs but strength and function data were sparse, with no pooled worsening.[15] The reviews that raise concern point to this gap more than to any measured harm.[22][23] In 2026, the Endocrine Society listed muscle function during medically induced weight loss as a research priority, not an established problem.[24]

The group with the most reason to pay attention is older adults and anyone who already has low muscle for their size. The ESPEN/EASO definition of sarcopenic obesity, meaning low muscle mass and strength together with excess fat, exists because that combination carries higher risk.[29] A 2025 retrospective study of older adults with type 2 diabetes on semaglutide found lean-mass losses in the same range as younger trial participants, with no functional measures.[33] Nobody has a randomized answer yet.

What the labels say. Neither the Wegovy nor the Zepbound prescribing information mentions muscle or lean mass as a warning or an adverse reaction.[26][27] The concern lives in the scientific literature, not on the label.

What protects muscle, and what is still untested

Outside GLP-1 drugs, two things reliably limit lean-mass loss during weight loss: resistance exercise and adequate protein.

  • Resistance training. In a 2017 trial of about 160 older adults with obesity losing roughly 9% of body weight on a diet, lean-mass loss was about 1 kg with resistance training, 1.7 kg with combined training and 2.7 kg with aerobic exercise alone; combined training was best for physical function.[40] A meta-analysis in older adults found resistance training prevented almost all diet-induced lean loss.[42] A review of 114 randomized trials found resistance training preserves or increases lean mass during weight loss.[41]
  • Protein. A 2024 meta-analysis of 47 trials found that protein intake above about 1.3 g per kg of body weight per day helped maintain muscle during weight loss, with the strength benefit showing mainly when combined with resistance training.[43][25]

Here is the honest part: none of these trials involved a GLP-1 drug. The guidance that GLP-1 users should aim for about 1.2 to 1.6 g of protein per kg of body weight per day (at least 60 g a day) and do muscle-strengthening exercise comes from the 2026 ADA standards and expert advisories that reason from the diet studies above, not from GLP-1 trials that tested those targets.[25][18][19] The first randomized trial designed to test resistance exercise and protein alongside semaglutide or tirzepatide, called LEAN, was registered in 2025 and had not reported at the time of writing.[51]

Two drugs are also being tested to preserve muscle on top of a GLP-1: bimagrumab with semaglutide (BELIEVE) and apitegromab with tirzepatide (EMBRAZE), both published in 2026.[9][10] In both, the add-on reduced lean-mass loss on DXA (EMBRAZE preserved about 1.9 kg more lean mass over 24 weeks in 102 people). Neither is approved for this, neither showed a functional benefit, and both are early.[9][10]

Why your smart scale cannot tell you

Home body-composition scales use bioelectrical impedance (BIA). They send a tiny current through the body, estimate water, and then estimate fat and "muscle" from equations that assume normal hydration.[50] In a methods study of consumer foot-to-foot scales, the muscle or lean estimate for an individual could be off by roughly ±7 kg compared with MRI or DXA (for one device, the range of error was −9.5 to +3.9 kg).[47] Another comparison found fat-free mass errors of −8.2 to +4.3 kg.[49] Even a medical-grade 8-electrode device showed a systematic bias of about 0.5 kg when tracking change against DXA.[48]

The total lean-mass change over a year and a half on these drugs is about 5 to 7 kg. That is the same size as, or smaller than, the error of a consumer scale on a single reading, and fluid loss from nausea or diarrhea can push the reading further.[47] No study has validated consumer smart scales in GLP-1 users.

A smart scale that says you lost 3 kg of muscle this month is reporting a change in water and an equation, not a measurement of muscle. If muscle mass is a real concern for you, a clinician can order a DXA scan, and simple strength tests (grip, chair-stand) tell you more about function than any scale.

Heard vs evidence

"40% of what you lose on Ozempic is muscle."
About 40% of the weight lost was lean mass in one 140-person DXA substudy of semaglutide.[2] Lean mass is mostly water, organ and connective tissue; muscle is roughly half of it.[16] The one MRI study of actual muscle found the thigh shrank no more than weight loss alone predicts, and got leaner.[5][52] The tirzepatide substudy found 25%, same as placebo.[4]
"These drugs melt muscle in a way diets do not."
The lean-mass share of weight loss is within the range seen with diets and surgery.[13][34][38] Meta-analyses find no drug-specific excess once weight loss is accounted for.[12][13]
"Nothing to worry about, then."
Not quite. Losing 15 to 20% of body weight means a large absolute loss of lean mass, strength was not measured in the big trials, and older adults and people with low muscle were under-represented.[22][23][24] The protection strategies (protein, resistance training) are sensible but untested in GLP-1 users.[18][19][51]
"My scale says I lost 3 kg of muscle."
Consumer impedance scales estimate muscle from water, and their individual error (roughly ±7 kg) is as large as the change they are being asked to detect.[47][49][50]

The short version

Short on the way. A 35-second captioned version of this page is in production for the Fewer Bites channel. It will be embedded here when it is live.

Questions to bring to your prescriber

Print these or screenshot them. They are conversation starters, not a checklist of things to do.

  • Given my age and how much weight I have lost, is there a reason to measure body composition (DXA) rather than rely on the scale?
  • Would a referral for a strength or function check (grip strength, sit-to-stand) make sense for me?
  • How much protein do you want me to aim for, and how do I get there when I am barely hungry?
  • Is there a resistance-training program you would suggest, or someone you refer patients to?
  • Do I have any condition (sarcopenia, kidney disease, older age) that changes how we think about lean mass on this medicine?

Education, not advice. This page explains what trials, labels and guidelines report. It does not tell anyone to start, change or stop a medicine. Your prescriber knows your history; this page does not.

Sources

Every number on this page traces to one of these. "Primary read" means the full paper or label was read; "abstract only" means only the abstract was available; "secondary" means the figure was taken from a named review.

  1. Wilding JPH et al. Once-weekly semaglutide in adults with overweight or obesity (STEP 1). NEJM 2021; supplementary appendix. primary read · industry-funded
  2. STEP 1 body-composition (DXA) substudy abstract. J Endocr Soc 2021;5(Suppl 1):A16. abstract only · industry-funded
  3. Jastreboff AM et al. Tirzepatide once weekly for the treatment of obesity (SURMOUNT-1). NEJM 2022. primary read · industry-funded
  4. Look M et al. Body composition changes during weight reduction with tirzepatide in SURMOUNT-1 (DXA substudy). Diabetes Obes Metab 2025. primary read · industry-funded
  5. SURPASS-3 MRI substudy: tirzepatide and muscle volume and composition. Lancet Diabetes Endocrinol 2025. primary read · industry-funded
  6. Hjelmesæth J et al. STEP UP body composition (EASD 2025 oral presentation). conference · industry
  7. Ravussin E et al. REDEFINE 1 body composition (ECO 2026 poster). conference · industry
  8. SUSTAIN 8 DXA substudy: semaglutide vs canagliflozin body composition. Diabetologia 2020. primary read · industry-funded
  9. BELIEVE trial: bimagrumab and semaglutide. Nature Medicine 2026. primary read · industry-funded
  10. Pratley RE et al. EMBRAZE: apitegromab with tirzepatide. Nature Medicine 2026. primary read · industry-funded
  11. Coskun T et al. Retatrutide body composition in type 2 diabetes. Lancet Diabetes Endocrinol 2025. primary read · industry-funded
  12. Karakasis P et al. Network meta-analysis of lean mass with GLP-1 receptor agonists. Metabolism 2024. abstract only
  13. Eisa N, Barood O. Lean mass changes with incretin therapy versus lifestyle intervention: meta-analysis. Diabetes Obes Metab 2026. abstract only
  14. Laverde et al. Meta-analysis of GLP-1 RAs and lean mass. Int J Obes 2026. abstract only
  15. Beaudart C et al. Incretin-based drugs, muscle mass and function. Drugs 2026. primary read
  16. Tinsley GM, Heymsfield SB. Fundamental body composition principles provide context for fat-free and skeletal muscle loss with GLP-1 RA treatments. J Endocr Soc 2024. primary read
  17. Heymsfield SB et al. Weight loss composition is one-fourth fat-free mass: a critical review. 2014. primary read
  18. Mozaffarian D et al. Nutritional priorities to support GLP-1 therapy for obesity: joint advisory (ACLM/ASN/OMA/TOS). 2025. primary read · expert consensus
  19. Strategies for minimizing muscle loss during use of incretin-mimetic drugs. Obesity Reviews 2025. primary read · review
  20. DXA appendicular lean soft tissue versus skeletal muscle. Obesity Reviews 2025. abstract only
  21. Neeland IJ et al. Changes in lean body mass with GLP-1-based therapies and mitigation strategies. Diabetes Obes Metab 2024. primary read · review
  22. Conte C, Hall KD, Klein S. Is weight loss-induced muscle mass loss clinically relevant? JAMA 2024 (Viewpoint). primary read · opinion
  23. Prado CM et al. Muscle matters: the effects of medically induced weight loss on skeletal muscle. Lancet Diabetes Endocrinol 2024. primary read · opinion
  24. Jastreboff AM et al. Obesity science, research gaps and opportunities in the new era of obesity medicines: Endocrine Society scientific statement. Endocrine Reviews 2026. primary read · expert statement
  25. American Diabetes Association. Standards of Care in Overweight and Obesity 2026. primary read · guideline
  26. FDA. Wegovy (semaglutide) prescribing information, rev. 02/2026. label
  27. FDA. Zepbound (tirzepatide) prescribing information, 2026. label
  28. EMA. Wegovy summary of product characteristics and EPAR. label
  29. ESPEN/EASO consensus definition and diagnostic criteria for sarcopenic obesity. 2022. abstract only · consensus
  30. Haines MS et al. Semaglutide and lean mass (ENDO 2025 abstract). conference
  31. SEMALEAN trial. Diabetes Obes Metab 2026. primary read
  32. SLIM LIVER: semaglutide and body composition in people with HIV. Clin Infect Dis 2024. primary read
  33. Semaglutide and muscle in older adults with type 2 diabetes (retrospective). Drug Des Devel Ther 2025. primary read · observational
  34. Bariatric surgery and lean/skeletal muscle loss: meta-analysis. primary read
  35. Skeletal muscle share of weight loss by sex (MRI). primary read
  36. CALERIE: organ and tissue changes with calorie restriction. Sci Rep 2025. primary read
  37. Look AHEAD long-term body composition. Obesity 2015; 8-year follow-up. primary read
  38. Chaston TB et al. Changes in fat-free mass during significant weight loss: a systematic review. Int J Obes 2007. abstract only
  39. Lundgren JR et al. Healthy weight loss maintenance with exercise, liraglutide, or both combined. NEJM 2021. primary read
  40. Villareal DT et al. Aerobic or resistance exercise, or both, in dieting obese older adults. NEJM 2017. primary read
  41. Resistance training and body composition in overweight/obesity: review of 114 RCTs. Obesity Reviews 2022. abstract only
  42. Resistance training prevents muscle loss induced by caloric restriction in obese elderly: meta-analysis. primary read
  43. Enhanced protein intake and muscle in adults with overweight/obesity during weight loss: meta-analysis. Clin Nutr ESPEN 2024. abstract only
  44. Incretin-based weight loss pharmacotherapy: can resistance exercise optimize changes in body composition? Diabetes Care 2024. primary read · opinion
  45. Gradual vs rapid weight loss on body composition: systematic review and meta-analysis. Br J Nutr 2020. abstract only
  46. Vink RG et al. Rate of weight loss and long-term weight regain. 2016. abstract only
  47. Bosy-Westphal A et al. Accuracy of bioelectrical impedance consumer devices. Obes Facts 2008. primary read
  48. BIA versus DXA for tracking change during weight loss. primary read
  49. BIA versus DXA agreement study. primary read
  50. ESPEN. Bioelectrical impedance analysis, part I: review of principles and methods. primary read · guideline
  51. LEAN trial: resistance exercise and protein with semaglutide or tirzepatide (protocol, NCT06885736). protocol · no results yet
  52. Muscle health with incretin therapy: review (thigh muscle volume −5.5 to −7.3% with weight −9.0 to −11.9%). European Heart Journal 2026. primary read · review

Eating on a GLP-1

Full after three bites: what to eat first

Read

Stopping

What happens to the weight when you stop

Read

Hub

GLP-1 side effects, by the numbers

Read

Nausea

Why GLP-1 nausea happens and how long it lasts

Read

All 35 questions