Keeping the muscle on GLP-1

GLP-1 drugs pulled off something genuinely historic: fifteen, twenty, sometimes more than twenty percent of body weight, routinely, from a weekly injection. That fight is essentially won. The exciting part is what comes next — the frontier moving from how much weight comes off to what kind. Because weight isn’t only fat; when the scale drops, some of what leaves is skeletal muscle.

In the body-composition analysis of semaglutide’s pivotal STEP 1 trial, lean mass made up close to 40% of the total weight lost; across the wider literature the figure runs from about a quarter to nearly half (Neeland et al., 2024, review). Read that not as a warning but as an opening: it turns body composition into the next thing we get to engineer.

First, the reassuring part: this isn’t GLP-1 agonists being toxic to muscle. They aren’t. It’s simply the arithmetic of an energy deficit — the same arithmetic behind muscle loss in any diet or fast. Three forces stack up:

• The deficit itself. When intake falls well below expenditure, the body catabolizes protein alongside fat for fuel and to spare amino acids for other needs. Muscle is a reservoir the body will draw down.
• Less load, less signal. Muscle mass is maintained by demand. Eat less and often move less, and the mechanical and nutritional signals that tell muscle to rebuild itself weaken.
• Protein undershoot. The appetite suppression that makes these drugs work also cuts total food — and protein — intake, removing the substrate muscle protein synthesis needs most.

That reframing matters, because it tells you where a solution has to act. You cannot make the deficit disappear — the deficit is the point. So the pharmacological question becomes narrower and sharper: during a large energy deficit, can you bias the body to give up fat and hold on to muscle? To do that, you have to find the switch that sets how much muscle the body is willing to keep.

Skeletal muscle mass is not a free-running maximum — it is actively restrained. The chief brake is myostatin (also called GDF-8), a member of the TGF-β superfamily. Myostatin, and its close relative activin A, signal through the activin type II receptors (ActRIIA/B) on the muscle cell, which activate the intracellular Smad2/3 pathway to hold muscle growth in check.

The proof that this brake is real, and releasable, is dramatic: animals and rare humans born without functional myostatin develop pronounced muscle hypertrophy. The body’s natural antagonist, follistatin, binds and neutralizes both myostatin and activin — the endogenous way of lifting the brake. That is the biology a new class of drugs is trying to borrow: block this pathway during GLP-1 weight loss, and the muscle the deficit would have taken is, in principle, protected.

The programs racing into this space differ in one decisive way: how far up the pathway they cut in. Block a single ligand and you get precision; block the shared receptor and you catch everything signaling through it for a bigger effect. Three antibodies, paired with a GLP-1 drug, map that spectrum from most selective to broadest — each a different bet on the same elegant target.

• Ligand-selective — apitegromab. Scholar Rock's antibody binds only the inactive precursor of myostatin, blocking its activation and nothing else. It is the scalpel of the group.
• Ligand, then two — trevogrumab (± garetosmab). Regeneron's trevogrumab neutralizes myostatin; adding garetosmab layers on activin-A blockade, widening the net one ligand at a time.
• Receptor-level — bimagrumab. Lilly's antibody blocks the activin type II receptor itself, so myostatin and activin A (and their relatives) are shut out together. The broadest lever — and the only one that visibly builds muscle on its own.

The clean way to see it: apitegromab and trevogrumab work at the ligand, bimagrumab at the receptor. Same pathway, different altitude — and, as the trials show, different magnitude.

This is where it gets fun, because the human data is already landing — and it points the same way every time: add the muscle-sparing antibody and a dramatically larger share of the weight lost comes from fat. Three phase-2 obesity programs, one peer-reviewed and two reported as topline/interim readouts, tell a strikingly consistent story.

Read together, these are a genuine proof of concept: the myostatin/ activin brake is druggable in humans, and releasing it during GLP-1 weight loss shifts body composition toward fat loss. Bimagrumab’s receptor-level block is the standout — it does not merely preserve lean mass but adds it, even used alone.

The proof of concept is in hand — the brake is druggable in humans, and releasing it reshapes weight loss toward fat. That’s the hard part done. What’s ahead is the genuinely exciting engineering: turning these body-composition wins into proven strength and function, and building the muscle-sparing arm into the therapy from day one. Where it’s heading:

• From scan to strength. Today's headline numbers are DEXA lean mass; the next wave of trials is wiring in direct strength, mobility, and physical-function endpoints — the readouts that turn 'kept the kilogram' into 'kept the capability.'
• A whole selectivity dial to tune. Ligand-selective, receptor-level, single-target or layered — and beyond antibodies, next-generation candidates like Scholar Rock's SRK-439 aim for more convenient dosing. It's a rich design space where breadth and precision can be tuned to the person.
• Muscle-sparing as the default. The natural endpoint is a GLP-1 that arrives paired with a muscle-preserving agent by design, so high-quality body recomposition is simply how weight loss works — not an add-on you have to chase.

It’s a beautiful example of this site’s whole premise: an elegant, real mechanism — the TGF-β brake — meeting a great problem at exactly the right moment. The biology is understood, the early human data is strong, and the trajectory points somewhere genuinely good: weight loss that spares, and sometimes builds, the muscle underneath.