Muscle During Rapid Weight Loss: What Has Been Measured

The claim, and the number underneath it

You will have seen the figure: about 25% of the weight lost on an incretin drug is lean mass rather than fat.

That number is real, it comes from published substudies, and it is usually quoted without the two things that make it interpretable — how many people it was measured in, and what the placebo arm did.

The tirzepatide substudy, with its denominator

SURMOUNT-1 randomised 2,539 adults. Its body-composition substudy had 160 participants with usable baseline and week-72 DXA scans: 124 across pooled tirzepatide doses and 36 on placebo¹.

In that substudy, change from baseline to week 72 was −21.3% in body weight, −33.9% in fat mass and −10.9% in lean mass on tirzepatide, against −5.3%, −8.2% and −2.6% on placebo, all p < 0.001¹.

Now the sentence that matters most. Of the body weight lost, approximately 75% was fat mass and 25% was lean mass — *for both tirzepatide and placebo*¹.

That is the finding. It is not that the drug preferentially strips muscle. It is that losing weight, by whatever route, comes with a lean-mass component in roughly that ratio.

Our tirzepatide page carries the trial's main results.

The retatrutide substudy, and how quickly the numbers thin out

The retatrutide phase 2 body-composition substudy is a useful worked example of how fragile this literature is.

189 participants were enrolled into the substudy. 155 had a baseline DXA scan. 103 completed treatment and had both baseline and week-36 scans².

Across arms, percent reduction from baseline in total fat mass was 4.9% on the lowest retatrutide dose, 15.2% on the pooled 4 mg arms, 26.1% on the pooled 8 mg arms, 23.2% on 12 mg, 2.6% on dulaglutide and 4.5% on placebo².

The authors' interpretation was that the proportion of lean mass loss to weight loss was similar to other obesity treatments². Note that this is a comparison of proportions across a substudy where individual arms are in the low thirties, and where the 8 mg arm's fat reduction came out numerically above the 12 mg arm's. Retatrutide is investigational; the retatrutide page tracks the programme.

The systematic review that put a range on it

A systematic review published in 2026 searched six databases plus ClinicalTrials.gov from January 2003 to February 2026 for randomised trials reporting body-composition outcomes on liraglutide, semaglutide, tirzepatide or dulaglutide³.

From 8,102 titles and abstracts it kept 35 primary studies. Ten of those 35 — 28.6% — prespecified body composition as a primary outcome. The other 25 measured it as a secondary or exploratory endpoint³.

Within the incretin groups, the median proportion of total weight loss attributable to reductions in muscle-based indices was 28.3%, with an interquartile range of 15.9% to 39.9%. In the studies using BIA or DXA the median was about 29%; in those using CT or MRI it was about 25.3%³.

And the comparator: across the 13 studies reporting weight loss in lifestyle or placebo groups, 38% exceeded their respective benchmark for muscle-based loss³. Again, the pattern is not unique to the drugs.

The gap the review names explicitly

Here is the sentence in that review worth reading twice: across the 35 included studies, no study reported objective physical function outcomes³.

DXA measures tissue. It does not measure whether someone can climb stairs, carry shopping, or get out of a chair without using their hands. Those are the outcomes the concern is actually about.

So as of the review's February 2026 search window, the field has a body of tissue-compartment data and, within that set of 35 randomised trials, no measured functional consequence. Uncertainty here is the finding, not a hole in this page.

The two interventions that have been randomised against it

Exercise. A one-year trial randomised 195 adults, after an eight-week low-calorie diet that produced a mean 13.1 kg loss, to exercise plus placebo, liraglutide 3.0 mg plus usual activity, both, or neither. The combination reduced body-fat percentage by 3.9 percentage points — roughly twice the reduction in the exercise-only group (−1.7 points) and the liraglutide-only group (−1.9 points)⁴. The combination was also the only strategy associated with improvements in HbA1c, insulin sensitivity and cardiorespiratory fitness⁴.

A drug. Bimagrumab is an investigational antibody targeting type II activin receptors, intended to reduce fat mass while promoting muscle growth. A 507-participant phase 2 trial randomised adults with obesity to nine groups for 48 weeks. Least-squares mean absolute change in body weight at week 48 was −9.3 kg with bimagrumab 30 mg/kg, −14.2 kg with semaglutide 2.4 mg, and −17.8 kg with the high-dose combination, against −3.3 kg on placebo⁵. Common adverse events with bimagrumab included muscle spasms, diarrhoea and acne⁵.

An earlier phase 2 trial of bimagrumab in adults with type 2 diabetes and obesity had already tested the body-composition question directly⁶.

How to read a lean-mass claim

Ask for the substudy size, not the trial size. A 2,539-person trial can carry a 160-person DXA substudy.

Ask what the placebo arm did. If roughly a quarter of placebo-arm weight loss was also lean mass, the drug is not the variable.

Ask whether anything functional was measured. Within the 35 trials in that 2026 review, nothing was.

For how these compounds compare on weight itself, see weight loss, and for the mechanism behind the class, incretins explained.