What GLP-1 Trials Actually Show About Muscle Loss

GLP-1 body-composition trials often show losses of both fat and lean mass. They do not show that every kilogram of lean mass lost is skeletal muscle. 1

The often-quoted 25 to 40% range describes selected findings, not a rule for everyone. Its meaning depends on what was measured and which loss was used as the denominator. 2 3

The concern deserves attention. The evidence needs more careful wording than either “these medicines destroy muscle” or “there is nothing to worry about”.

Start with the tissue that was measured

DXA separates fat, lean soft tissue and bone mineral. Fat-free mass includes bone mineral; lean soft tissue does not. Neither compartment is identical to skeletal muscle. 1

Fat-free mass includes muscle, organs, skin, connective tissue, water and glycogen with its associated water. Adipose tissue also contains non-fat material, so losing adipose tissue can reduce fat-free mass. 1

Tinsley and Heymsfield’s explanation is useful: a change in this broad compartment cannot be translated directly into the same amount of muscle lost. 1

Use the trial’s own noun. Report lean mass as lean mass. Report strength as strength. A functional improvement is valuable, but it does not itself measure muscle size. 1

Where the 25 to 40% figures come from

Karakasis and colleagues’ network meta-analysis estimated that approximately one-quarter of weight loss was lean mass. It found no significant reduction in lean mass as a percentage of body weight. 4

That is a pooled finding, not an individual forecast. Combining medicines, populations and study methods does not make their results interchangeable. 4

SUSTAIN 8 illustrates a higher proportion. Over 52 weeks, semaglutide was associated with reductions of 3.4 kg in fat mass and 2.3 kg in lean mass. Participants had type 2 diabetes. 3

Dividing 2.3 by the combined 5.7 kg reduction gives approximately 40%. This is a calculation from the reported component changes, not a percentage stated by the authors. 3

Its denominator is combined fat and lean mass lost, not necessarily total scale-weight loss. The comparison was with canagliflozin, not placebo, and semaglutide was given at 1 mg weekly. 3

These distinctions explain why “40% of everyone’s weight loss is muscle” misrepresents the result. They also prevent a subtler mistake: treating unlike percentages as a precise class-wide range.

The placebo arm changes the story

The SURMOUNT-1 DXA analysis included 160 participants: 124 receiving tirzepatide, a dual GIP/GLP-1 agonist, and 36 receiving placebo. 2

At 72 weeks, weight had fallen by 21.3% and 5.3%, respectively. Approximately 75% of the lost weight was fat mass and 25% lean mass in both groups. 2

Same approximate split. Different total loss.

That supports a narrower interpretation than “the medicine uniquely strips away lean tissue”. It does not prove there is no direct drug effect, because the groups were not matched for weight loss. 2

It also does not make absolute loss irrelevant. As an arithmetic example, one-quarter of a 20 kg loss is 5 kg; one-quarter of a 4 kg loss is 1 kg. Identical percentages can hide very different quantities.

The useful questions are therefore separate: how much tissue changed, which compartment was measured, and what happened to strength or daily function? A favourable percentage cannot answer all three.

These were not trials without nutrition support

STEP 1’s trial record describes diet and physical-activity counselling for 68 weeks in both the semaglutide and placebo groups. The comparison was medicine plus lifestyle support versus placebo plus support. 5

SURMOUNT-1 likewise included a reduced-calorie diet, activity advice and counselling across its treatment groups. 2

However, receiving advice is not the same as achieving a particular protein intake or completing progressive resistance training. Those are separate interventions that need their own assessment.

For these trials, “no nutrition intervention” is the wrong description. But their existing counselling also cannot establish how much benefit a more intensive nutrition programme would add.

A coaching service should not present its clients as outperforming an unsupported trial group when that was not the comparator. A valid comparison needs comparable participants, measurements, follow-up and treatment conditions.

Protein targets are guidance, not guarantees

The 2025 joint advisory discusses proposed protein intakes of 1.2 to 1.6 g/kg/day during weight reduction. It explicitly notes uncertainty about which weight basis to use in people with obesity. 6

Actual body weight, adjusted weight and fat-free mass are different denominators. Using actual weight can overestimate requirements; the same number cannot simply be transferred between them. 6

For arithmetic only, 1.2 g/kg means 120 g at 100 kg body weight, but 72 g at 60 kg fat-free mass. These are not interchangeable prescriptions.

The distinction in wording matters:

• Accurate: Guidance supports individualised protein intake alongside resistance exercise. 6
• Too strong: GLP-1 trials prove that one protein number preserves everyone’s muscle.

Ask whoever sets your target to explain the denominator, the reason for choosing it and how it fits your appetite and health needs. A precise calculation is only useful when its starting assumptions are appropriate.

Exercise has been tested, but important gaps remain

It is incorrect to say that exercise has never been tested alongside GLP-1 treatment. Lundgren and colleagues published a relevant randomised trial in 2021. 7

After an eight-week low-calorie diet, 195 adults with obesity were assigned to exercise, liraglutide, both treatments or placebo for one year. They did not have diabetes. 7

The combination reduced fat mass while preserving lean mass. Exercise-containing groups also improved cardiorespiratory fitness. These findings provide direct evidence for an exercise-plus-medicine strategy. 7

But the programme was predominantly vigorous aerobic exercise, mostly cycling and running. It studied maintenance after initial dietary weight loss, not a resistance-only programme during semaglutide or tirzepatide initiation. 7

LEAN-PREP addresses a more specific question. Its four groups compare control, resistance exercise, increased protein and their combination during semaglutide or tirzepatide treatment. 8

Its planned primary outcome is MRI-measured quadriceps cross-sectional area, with additional physical-function assessments. The protocol describes what will be tested, not what works. 8

The search checked on 6 September 2026 located that protocol, not a results paper. The remaining uncertainty is about specific prescriptions and populations, not whether exercise has any supporting evidence.

The intake problem is measured, but cautiously

Korus and colleagues studied 387 adults recruited through Polish online support groups. Participants reported their treatment and completed food records covering one weekday and one weekend day. 9

Reported average intake was 753 kcal and 33.4 g of protein daily. Fewer than 10% met the study’s protein recommendations. These are findings from this sample, not expected intakes for all GLP-1 users. 9

Self-selection, possible under-reporting and only two recorded days limit interpretation. There was no untreated comparison establishing that medication caused these intakes. 9

Food records also do not diagnose biochemical deficiencies or show that low protein caused skeletal-muscle loss. The study supports checking dietary adequacy, not assuming that every user is malnourished. 9

Persistent nausea, vomiting or very low intake warrants contact with your prescriber. Seek urgent assessment if you cannot keep fluids down or have signs of dehydration. 6

What to track in practice

Keep three evidence layers separate: what a trial measured, what guidance recommends, and what is happening to you. None should be presented as a substitute for the others.

A useful review can cover:

• Food and fluids: record a few representative days, including difficult days, rather than judging intake from appetite alone. 6
• Strength and function: discuss changes in lifting, stairs or getting out of a chair, not just changes on the scales. 6
• Composition: DXA and BIA estimates do not directly count kilograms of skeletal muscle. Hydration and measurement conditions matter when interpreting trends. 1

Bring these observations to a review rather than treating one scan or one percentage as a verdict. Nutrition coaching can support food structure; prescribing decisions remain with your clinician.

For practical detail, read GLP-1 nutrition, strength and monitoring and how much protein you need.

For measurement context, see how body-composition scans work. Our coaching programmes provide nutrition support alongside your own prescriber.