The concern is straightforward. Semaglutide and tirzepatide produce weight loss at a rate previously only achievable with surgery, and some of what comes off is not fat. The question worth answering is how much, whether that proportion is unusual, and what it means over a longer horizon.

The numbers

The most cited figure comes from the STEP 1 body composition substudy. In that arm, 140 participants had DEXA scans at baseline and week 68. Total body fat mass fell from 46.4 percent to 43.3 percent of body mass in the semaglutide 2.4 mg group, and the proportion of lean body mass rose correspondingly, which is the direction you want. In absolute terms, participants lost roughly 15 kg total, of which about 6 kg was lean mass. That works out to a bit under 40 percent of the total.

SURMOUNT-1 ran a similar substudy for tirzepatide with 160 participants. Over 72 weeks, total fat mass fell 33.9 percent while lean mass fell 10.9 percent. Lean mass accounted for around 25 percent of total weight lost, and the ratio of fat to lean loss was better than in the lifestyle comparison group in that analysis.

So the headline range across the major trials is roughly a quarter to two fifths of lost mass being lean tissue.

What that number is not

Three things need saying before anyone panics.

First, lean body mass on DEXA is not muscle. It is everything that is not fat and not bone mineral: skeletal muscle, organs, connective tissue, glycogen and the water bound to it, and blood volume. Glycogen binds roughly three grams of water per gram, so a meaningful early drop in lean mass reading is glycogen and fluid. Organ mass genuinely does shrink with substantial weight loss, and so does the extra muscle that was carrying the extra body mass around. Some of the lean loss is appropriate.

Second, this proportion is not specific to GLP-1 drugs. Longstanding data on caloric restriction, reviewed by Heymsfield and colleagues, puts lean mass at about 20 to 30 percent of total weight lost across diet interventions, and bariatric surgery cohorts report figures in the 20 to 30 percent range as well. The GLP-1 numbers sit inside that band, at the higher end in the semaglutide substudy. The drugs are not producing an unusual pattern, they are producing a normal pattern faster and to a greater extent.

Third, both substudies were secondary analyses on a fraction of the trial population, not powered to detect differences in lean mass, and neither controlled protein intake or resistance training. Participants in these trials were not being coached to lift.

Why it still matters

Two populations should care more than the average.

  • Older adults. Skeletal muscle mass declines about 1 percent per year after 50 and strength faster than that. Someone at 68 with borderline sarcopenia who loses 6 kg of lean mass over 15 months is in different territory to someone at 35.
  • Anyone stopping the drug. The STEP 1 extension found that a year after withdrawal, participants had regained about two thirds of the weight lost. Regain is disproportionately fat. Lose lean, regain fat, repeat, and body composition worsens across cycles even if the number on the scale returns to where it started.

What the mitigation evidence supports

The interventions with real backing from the wider weight loss literature, applied here by reasonable extrapolation rather than by direct GLP-1 trial evidence:

  1. Protein intake. Higher protein during energy restriction preserves lean mass, shown repeatedly in controlled feeding studies. Around 1.2 to 1.6 g per kg of body weight per day is the range that comes out of that work, and it is hard to reach when a drug has suppressed your appetite to the point where you eat 1200 calories without noticing. Front-loading protein at each meal is the practical answer.
  2. Resistance training. The single strongest lever. Trials combining energy restriction with progressive resistance training consistently show most or all of the lean mass loss abolished compared with restriction alone.
  3. Rate of loss. Faster loss shifts the composition unfavourably. Titrating to a dose that produces steady rather than maximal loss is a defensible choice.

The bimagrumab question

There is active pharmaceutical interest in combining GLP-1 agonists with myostatin or activin pathway blockers. A phase 2 trial of bimagrumab, an activin type II receptor antibody, reported a 20.5 percent reduction in fat mass with a 3.6 percent increase in lean mass over 48 weeks in adults with type 2 diabetes and obesity. Combination trials with semaglutide are running. This is promising and it is also, at time of writing, phase 2 data in a specific population, so treat conclusions as provisional.

Where the evidence stops

No published trial has randomised GLP-1 users to structured resistance training versus none with body composition as the primary outcome, which is exactly the study that would answer the practical question. Nobody has measured muscle function (grip strength, gait speed, one rep max) alongside DEXA in these cohorts at scale, and mass is a poor proxy for function. And there is no long-term data on what repeated cycles of GLP-1 use and discontinuation do to body composition over a decade, because the drugs have not been in wide use that long.