There is a question people ask about these medicines that the approved labels essentially refuse to answer. It is not a small question, and the gap is measurable to the word.
The FDA labels for Zepbound and Wegovy were searched in full on 2026-09-10 through openFDA — every section, not just the adverse reactions table. The phrase lean body mass appears zero times. Body composition, zero times. Muscle mass, zero times.
What each label does contain is one sentence, in section 12.2, Pharmacodynamics. Here they are side by side:
Tirzepatide lowers body weight with greater fat mass loss than lean mass loss.
Semaglutide lowers body weight with greater fat mass loss than lean mass loss.
Same sentence, different molecule, two manufacturers, two separate submissions. Eleven words, no figure, no ratio, no cited study, no placebo comparison. And in the Ozempic and Mounjaro labels, covering these same two molecules for type 2 diabetes, the phrase fat mass does not appear at all. The one trial that randomised people to one molecule or the other, SURMOUNT-5, reported total body weight and waist circumference — not what the weight was made of.
This page reports what the public record contains. It does not tell anyone what to do about it; that belongs with a prescriber, as our editorial standards set out.
What "greater" is doing in that sentence
The word carrying the entire claim is greater, and it is a comparative with no second number.
A drug that took nine kilograms of fat for every kilogram of lean tissue satisfies that sentence. So does a drug that took three kilograms of fat for every two of lean tissue. Both are "greater fat mass loss than lean mass loss". The sentence is true in either case and tells a reader nothing about which case they are in. The SURMOUNT-1 body-scan substudy, which put tirzepatide's split at about three to one on drug and placebo alike, and a thigh MRI study of semaglutide 7.2 mg are set out on the Ozempic body page.
That is not an accusation of bad faith. Section 12.2 is a pharmacodynamics section; it describes how a drug behaves, and it is not the section where efficacy numbers live. But the practical consequence stands: the approved labelling for the two most-prescribed weight-management medicines in the United States contains no quantity whatsoever describing what happens to lean tissue, and no page quoting those labels can supply one without going outside them. Aches are a different question from lean mass: joint and muscle pain are absent from every current GLP-1 label, as the joint-pain record sets out.
Most articles on this subject fill the gap with a percentage from a body-composition substudy and present it as though it were settled. This one is going to fill it a different way — by looking at what companies are spending money on, which is a harder thing to fake than a citation.
Nine studies, five sponsors, and no phase 3
On 2026-09-10 the ClinicalTrials.gov registry was queried for studies combining a muscle-sparing agent — bimagrumab, trevogrumab, enobosarm, apitegromab, taldefgrobep or garetosmab — with semaglutide, tirzepatide or a GLP-1 receptor agonist. It returns nine.
| Study | Sponsor | Phase | Enrolment | Status |
|---|---|---|---|---|
| NCT06299098 — trevogrumab, or trevogrumab with garetosmab, plus semaglutide | Regeneron | 2 | 1,005 | Active, not recruiting |
| NCT05616013 — bimagrumab and semaglutide | Eli Lilly | 2 | 507 | Completed |
| NCT06643728 — bimagrumab and tirzepatide, alone or combined | Eli Lilly | 2 | 252 | Active, not recruiting |
| NCT07446998 — enobosarm with a GLP-1 receptor agonist | Veru Inc. | 2 | 239 | Active, not recruiting |
| NCT06282458 — enobosarm dose-finding, body composition | Veru Inc. | 2 | 168 | Completed |
| NCT06890611 — bimagrumab alone and with tirzepatide, healthy participants | Eli Lilly | 1 | 125 | Completed |
| NCT06445075 — apitegromab in adults with overweight or obesity | Scholar Rock | 2 | 102 | Completed |
| NCT05933499 — tirzepatide and bimagrumab, body composition, insulin sensitivity, bone | Mass General | 2 | 63 | Recruiting |
| NCT06901349 — bimagrumab and tirzepatide in obesity with type 2 diabetes | Eli Lilly | 2 | 0 | Withdrawn |
Two facts about that table are worth more than any percentage.
The first: not one of the nine is a phase 3 study. Eight are phase 2 and one is phase 1. Phase 3 is where the evidence required for approval is produced, and no phase 3 programme has begun for any of these agents in this setting. Several of the phase 2 studies have already completed and reported — Lilly's 507-participant bimagrumab study, Veru's 168-participant dose-finding study, Scholar Rock's 102-participant study. Results exist. Approval does not, and on this timeline cannot for years. The nine-arm body-composition results of Lilly's 507-participant study are set out on our bimagrumab trial page.
So anyone reading that a solution is arriving is reading ahead of the registry. There is no approved drug that preserves lean mass during GLP-1 treatment, and the studies that would produce one have not started.
The second: one of the five sponsors is Eli Lilly, which manufactures tirzepatide. Four of the nine studies are Lilly's, testing its own bimagrumab against or alongside its own tirzepatide. A company does not run four studies to solve a problem it regards as immaterial to its own product. That is the clearest signal in the public record about how seriously the manufacturer takes lean mass loss — considerably more so, on the evidence of its spending, than its own label's eleven words suggest.
The withdrawn study is worth noting honestly rather than skipping: NCT06901349 is registered as withdrawn with zero participants enrolled, so it represents an intention rather than data. It is left in the table because a count of nine that quietly excluded it would be a count of nine that could not be reproduced from the query.
Combined registered enrolment across all nine, including the withdrawn one at zero: 2,461 participants.
Who pays for a trial is a signal worth reading in both directions. Where manufacturers see a commercial question they fund the work, as four Lilly studies here show. Where they do not, the registry goes quiet: across thirty-one registered studies of these drugs in polycystic ovary syndrome, a comparable body of participants, not one has an industry sponsor.
What the reports database says, and why it is the wrong instrument
FDA's adverse event reporting system was queried on the same day. It is not an incidence measure — no denominator, unverified reports, no finding of causation — and the counts here are small:
| Reported term | Tirzepatide | Semaglutide |
|---|---|---|
| Muscular weakness | 280 | 332 |
| Muscle atrophy | 246 | 220 |
| Sarcopenia | 13 | 14 |
Against 160,663 total reports for tirzepatide and 73,001 for semaglutide, these are fractions of a percent.
It would be easy to present that as reassurance. It is not, and the reason is a limitation of the instrument rather than a fact about the drugs. Spontaneous reporting captures events people notice and attribute — a rash, a bout of vomiting, a hospital admission. Lean mass declining gradually over months is not an event, has no onset date, produces no symptom on any particular day, and is invisible without a scan. It is close to the least reportable thing a drug could do. A low count here is what would be expected whether the underlying change were large or small, which is why this table cannot be used as evidence in either direction, and why the nine trials are a better guide than the 246 reports.
What this page is not going to tell you
Whether resistance training or protein intake changes the outcome is the question most readers arrive with, and it is deliberately not answered here. The sources this page rests on — approved labelling and the trial registry — do not address behaviour, and no approved label for any of these products carries an instruction about either. Filling that gap with general advice would mean leaving the evidence this page is built on, and MuscleLedger covers resistance training and body composition for readers who want that ground properly worked.
The nutrition half of it has since been taken up from the literature rather than the labels, and the answer is an absence with a date on it: the randomised trial built to test protein intake during this therapy is recruiting, targets 1.6 g per kilogram a day, and posts its primary result in 2029.
One thing the record does specify is the calorie deficit itself. Every pivotal trial in the class ran a prescribed diet alongside the drug, and the protocols name the deficit and the minutes — which is where any question about what the body loses has to start.
What can be said from the record is narrower and more useful: the manufacturers have written one unquantified sentence each, the registry contains no approved pharmacological answer and no phase 3 attempt at one, and five organisations including the manufacturer of tirzepatide are spending money as though the problem is real.
Bone is measured about as poorly, and by a similar route: the handful of registered studies touching it enrol postmenopausal participants for reasons of study design rather than to ask a question about that population, which we count on our page about the menopause evidence gap. The fracture counts one label does report, and the bone-density trials, are set out on GLP-1 drugs and bone loss.
The same gap runs through the rest of the labelling: what the labels do and do not count as a side effect is a question of which endpoints were measured, not of what happened to people. And the withdrawal trials that measured what returns after the drug stops counted body weight, which leaves the composition of the regained weight as unmeasured as the composition of the lost weight. The most visible version of the same question — what weight loss takes from the face — is counted separately on Ozempic face, where one small imaging study is the whole of the measured record.
Sources
- FDA label for Zepbound (tirzepatide) injection, effective 2026-08-28, via the openFDA drug label endpoint, full-record text search, read 2026-09-10; and for Wegovy (semaglutide), Novo Nordisk's label effective 2026-06-18, DailyMed, full text searched 2026-09-23.
- FDA labels for Ozempic (effective 2026-01-30) and Mounjaro (effective 2026-08-27, Eli Lilly's copy, re-checked 2026-09-24), same source and date, searched for the same terms.
- ClinicalTrials.gov API v2, combined query for the six named agents with semaglutide, tirzepatide or GLP-1, returning nine studies with sponsor, phase, enrolment and status, queried 2026-09-10. A control query of known scale was run in the same session to confirm the query shape.
- openFDA drug adverse event endpoint, report counts by reaction term, queried 2026-09-10.