GLP-1 and muscle

What controlled trials actually measured about lean mass on GLP-1s

Almost every headline about muscle loss on a GLP-1 receptor agonist traces back to a small number of body-composition substudies rather than to the large trials themselves. This page separates what was measured from what has been inferred.

What the large trials actually measured

STEP 1 randomly assigned adults with overweight or obesity to once-weekly semaglutide or placebo for 68 weeks. Its primary endpoints were the percentage change in body weight and the proportion of participants reaching a weight-loss threshold. Body composition was not among them.¹

SURMOUNT-1 was built the same way. Investigators reported percentage change in body weight and categorical weight-loss responses over 72 weeks for tirzepatide. Tissue composition was a secondary question handled in a subset, not a trial-level endpoint.²

The approved U.S. labeling reflects that design. The prescribing information for semaglutide in weight management describes weight-related efficacy. It carries no claim about fat mass, lean mass, or muscle, because no such claim was studied at registration scale.⁶

• Primary outcomes in the large obesity trials were expressed as change in total body weight.
• Body-composition data came from imaging substudies run at a minority of trial sites.
• Labeling describes weight outcomes, not tissue outcomes.¹²⁶

The STEP 1 DXA substudy, and what it can carry

Within STEP 1, a subgroup of participants underwent dual-energy X-ray absorptiometry at baseline and again at week 68. The substudy was small relative to the parent trial and was limited to sites where a scanner was available.¹

Investigators reported that in the semaglutide group, total fat mass fell proportionally about twice as much as lean body mass, and that the share of body weight made up of lean tissue rose. Lean mass still fell in absolute terms. Both statements are true at the same time, and quoting only one of them is how a single substudy has been used to argue opposite conclusions.¹

That is also why published fat-versus-lean percentages for this trial disagree. A percentage reduction within a compartment and a share of total weight lost are different quantities, and both have been reported from the same scans. Neither figure should be treated as a settled constant.¹³

A substudy is not a trial. Participants were not randomised to receiving a scan, sites with imaging capability may differ from those without, and the sample was too small to resolve who loses more lean tissue — older participants, for instance, or those starting with less muscle.¹

STEP 1 is the semaglutide trial most often cited on this question. Its DXA substudy is the source of nearly every published figure for fat versus lean mass change during semaglutide treatment.

The one-fourth rule and its critique

Outside pharmacotherapy, a familiar approximation holds that about one quarter of the weight lost during energy restriction is fat-free mass. It is convenient, easy to remember, and frequently transplanted onto GLP-1 results as if it were a benchmark.³

Heymsfield and colleagues traced where that number came from and argued it is far less stable than its use suggests. The fat-free share of weight lost varies with baseline adiposity, the size and duration of the energy deficit, protein intake, whether resistance training is performed, and the measurement method itself.³

• People carrying more fat mass at baseline tend to lose a smaller fraction of weight as fat-free mass.
• Larger and faster energy deficits shift the composition of the loss.
• Protein intake and resistance training change the ratio.
• Different measurement methods return different fat-free mass estimates from the identical weight change.³⁷

The practical consequence is that a fixed one-quarter figure cannot be used to declare a GLP-1 result normal or alarming. It was never a constant, and the studies that produced it did not involve these drugs.³

Fat-free mass is not the same as muscle

Fat-free mass is a compartment defined by what it is not. It contains skeletal muscle, but also total body water, glycogen and the water bound to it, connective tissue, bone mineral, and the mass of internal organs.⁴⁷

That distinction matters early in weight loss, when glycogen depletion and the accompanying water shift can register as a fat-free mass reduction within days. It matters again over months, because organ mass and connective tissue also decline as overall body size falls.⁷

So a shrinking fat-free compartment does not by itself demonstrate lost contractile muscle, and it does not demonstrate lost strength. Strength and physical function are measured with dynamometry, timed walking tests, and performance batteries — instruments that the large obesity trials did not use.⁷

How DXA and BIA can mislead

DXA is treated as a practical reference in obesity research, but it is a model, not a direct measurement. It infers tissue from differential X-ray attenuation and assumes a stable hydration fraction of fat-free mass. When hydration shifts — as it does during rapid weight loss, with sodium and carbohydrate changes, and around illness — the lean estimate moves with it.⁴

DXA results are also not interchangeable between manufacturers, and often not between software versions on the same machine. Buckinx and colleagues catalogued this across methods and argued that the field still lacks an agreed reference standard for muscle mass, which is why estimates from different studies cannot be pooled naively.⁴

What impedance adds, and what it does not

Bioelectrical impedance analysis does not measure muscle at all. It measures the body’s opposition to a small alternating current and converts that into a fat-free mass estimate using a regression equation derived from a reference population. Kyle and colleagues set out how sensitive that conversion is to hydration status, recent food and fluid, exercise, skin temperature, electrode placement, and whether the equation fits the person being measured.⁵

• Compare scans from the same device, the same software version, and similar conditions — or do not compare them.
• A single scan is a point estimate with error attached, not a measurement of muscle.
• Consumer impedance scales inherit every assumption above and rarely disclose the equation they use.⁴⁵

What is still unknown

The honest summary is that this question is under-measured rather than settled. No large randomised trial has been powered on muscle strength or physical function as a primary endpoint, so there is no direct evidence on whether the lean-mass changes seen on DXA translate into measurable weakness.⁷

Follow-up after discontinuation is thinner still. Weight regain after stopping treatment has been described, but whether the fat-free compartment recovers, how quickly, and in what proportion relative to fat has not been characterised in a dedicated trial with repeated imaging.¹²

Older adults — the group with the most to lose from a decline in muscle mass and function — are underrepresented in these obesity programmes, and the imaging substudies inside them are smaller still. The same applies to people who already meet criteria for low muscle mass before treatment begins.⁴⁷

Common questions