GLP-1 and muscle

Body composition by sex on GLP-1s: what the substudies did not report

This page contains no dose, schedule or route, and makes no claim about muscle, strength or appearance. It reports what the body-composition substudies contained, and specifically what they did not report by sex.

How small the body-composition evidence actually is

STEP 1 randomised 1,961 adults and reported percentage change in body weight as its primary endpoint. Its body-composition endpoints were measured only in a subgroup that received a dual-energy X-ray absorptiometry scan, and the posted results give the analysis set for those endpoints as 83 participants on semaglutide and 39 on placebo with available data.¹²

That is 122 people out of 1,961 — roughly one in sixteen. Every published figure for fat mass, lean body mass and visceral fat mass change in that trial rests on that group.¹

The registry describes the set precisely: participants in the full analysis set who had a scan performed at baseline. It was not a randomised subgroup, and sites without a scanner could not contribute to it.¹

The subgroup was also filtered

The subgroup was not just small; it was selected on a characteristic. A published review of the STEP programme states that body composition was evaluated by dual-energy X-ray absorptiometry in a subgroup of participants with a body-mass index at or below 40 kg/m².³

That filter removes everyone at the top of the body-size distribution the trial recruited. A subgroup defined by a body-mass index ceiling is a selected slice rather than a neutral sample, and the registry does not publish the characteristics of the people who fell inside it.¹³

The split that is not there

STEP 1's posted baseline characteristics give sex for the whole randomised population — 1,453 female and 508 male. They do not give sex for the DXA analysis set, and the body-composition outcome measures are posted only by treatment arm.¹

So the public record does not permit anyone to say how many men were in the body-composition subgroup, and therefore does not permit a sex split of the fat-mass or lean-mass results to be computed or checked.¹

SURMOUNT-1 is thinner still on this point. Its posted outcome measures include no body-composition endpoint at all, so there is nothing to split.⁴

Why it would be hard even if the split existed

Suppose the sex breakdown of a 122-person analysis set were published, and suppose it mirrored the parent trial at roughly three-quarters female. That would leave about 30 men across both arms combined.¹

A comparison between roughly 30 men and roughly 90 women, in a subgroup that was not randomised and was filtered on body-mass index, would be a subgroup analysis of a subgroup. The standard methodological objections to subgroup claims apply with full force, and the sample would be far too small to support an interaction test.¹⁵

That is the real reason no credible sex split exists here. It is not an oversight in reporting; the substudy was never large enough to produce one.¹⁵

What the label says, and does not

The approved US labeling for semaglutide in weight management describes weight-related efficacy. It carries no claim about fat mass, lean mass or muscle for any group, which is consistent with a body-composition dataset of this size.⁶

A regulator declining to make a claim is informative. It means the evidence submitted did not support one, in either direction and for either sex.⁶

Semaglutide has the only posted body-composition dataset in this group of compounds, and it covers 122 participants with no published sex breakdown. That is the whole of the registry-checkable record.

Common questions