For the past few years, obesity pharmacology has measured success with a single number: change in body weight. Semaglutide, tirzepatide, retatrutide, each successive drug announced a bigger minus on the scale. But weight is a sum, not a composition. And once people started looking at that sum more closely, it turned out that part of what is lost is not fat at all.
Bimagrumab (BYM338), a fully human monoclonal antibody against activin type II receptors, is interesting precisely because it is not competing for a bigger minus. It is competing over a different question.
A quarter of the lost mass is not fat
The numbers here are not hypothetical. In the substudy of SURMOUNT-1 participants who underwent DXA (dual-energy X-ray absorptiometry, a body-composition scan) at baseline and at week 72 (n=160), tirzepatide produced −21.3 % body weight, −33.9 % fat mass and −10.9 % lean mass. Expressed as a split: roughly 75 % of the weight lost came from fat and roughly 25 % from lean tissue, and the same proportion held in the placebo group.
A network meta-analysis of 22 randomised trials (2,258 participants) produced the same picture for the class as a whole: lean mass loss accounted for approximately a quarter of total weight loss, and the most potent agents for weight reduction, tirzepatide 15 mg and semaglutide 2.4 mg, were among the least effective at preserving lean tissue. A review in Diabetes Care framed the magnitude differently and more bluntly: a lean mass loss of roughly 6 kg is comparable to a decade or more of ageing.
The limits of that claim matter. Lean mass on DXA is not only muscle: it includes water, glycogen, organs and connective tissue, and part of the decrease is a normal response to a smaller body, not pathology. The point is not that it is bad by definition, but that until recently nobody had a tool to test whether it could be otherwise.
A molecule that was turned around
Bimagrumab entered this story off-plan. Novartis developed it for muscle pathology and tested it wherever muscle is lost: sporadic inclusion body myositis, COPD (chronic obstructive pulmonary disease) with cachexia, sarcopenia, recovery after hip fracture surgery, weight loss in lung and pancreatic cancer. The confirmatory RESILIENT trial in sIBM (n=251) missed its primary endpoint, and the muscle programme was wound down.
But the same trials kept producing a side observation: alongside muscle gain, participants lost fat. At first a secondary result in pharmacokinetic and sarcopenia work, it later became the primary hypothesis. In 2023 Eli Lilly acquired Versanis Bio together with the programme, and the compound was assigned the code LY3985863.
The mechanistic rationale for the combination was tested in diet-induced obese mice (Molecular Metabolism, 2024): bimagrumab alone produced roughly +10 % lean mass while simultaneously reducing fat mass; semaglutide reduced both fat and muscle; the combination produced the greatest fat loss while preserving lean mass despite reduced food intake.
What the human numbers showed
A 48-week phase 2 trial in adults with type 2 diabetes and obesity (Heymsfield et al., 2021; n=75) produced a combination rarely seen together: fat mass −20.5 % (−7.5 kg) versus −0.5 % on placebo, lean mass +3.6 % (+1.70 kg) versus −0.8 %, waist circumference −9.0 cm, and HbA1c −0.76 percentage points.
Then came BELIEVE, a phase 2 trial in 507 adults with obesity (Nature Medicine, 2026): nine groups, 48 weeks, plus an open-label extension to week 72. The absolute change in body weight at week 48 was −9.3 kg for bimagrumab 30 mg/kg, −14.2 kg for semaglutide 2.4 mg and −17.8 kg for the combination, versus −3.3 kg on placebo.
Note the ordering of those numbers. Bimagrumab on its own produced a smaller minus than semaglutide, and by the logic of who took more off the scale it simply loses. The entire point lies in how that minus breaks down: in the highest-dose groups the share of weight loss coming from fat was about 72 % for semaglutide, 100 % for bimagrumab and about 93 % for the combination. Lean mass changed by −7.4 % with semaglutide, +2.5 % with bimagrumab and −2.9 % with the combination.
The third component, mentioned less often
Weight loss costs more than muscle. Bone mass also declines, and in older patients that carries its own consequences, from reduced density to falls and fractures. Since ActRII signalling reaches beyond skeletal muscle, the bone question is not rhetorical. A trial recruiting at Massachusetts General Hospital (NCT05933499, n=63, 52 weeks) compares tirzepatide, bimagrumab and their combination on body composition, insulin sensitivity and bone. No results are available yet, so any claim about a bone effect of bimagrumab in humans would today be speculation.
What still has not been shown
This is exactly where the story gets weaker. The clinical history of bimagrumab in muscle pathology is a history of mass gain failing to translate into functional improvement; the ACE-083 programme showed the same thing independently. In the metabolic setting the question is identical and still unanswered: preserving lean mass during weight loss is an intermediate measure, not a demonstrated clinical benefit. Whether people become stronger, more mobile, or regain less weight after stopping treatment has not been shown in randomised trials.
The programme's status is equally mixed. The largest trials are phase 2 and 2b; there is no phase 3. In 2025 Lilly withdrew a planned tirzepatide combination trial (NCT06901349) before enrolment, citing strategic business reasons; another (NCT06643728) is ongoing. The most frequent adverse events in published work are muscle spasms, diarrhoea and acne. All the data concern intravenous administration in clinical settings, in populations with obesity and type 2 diabetes; there is no basis for extrapolating to healthy individuals.
Bimagrumab sits in the Muscle Growth category of our catalogue; the full account of mechanism, clinical trajectory and sources is in the monograph. It is a monoclonal antibody of roughly 150 kDa, not a peptide. The compound is supplied strictly for laboratory research (RUO): it is not a medicine, not a supplement and not a food product, and it is not intended for consumption by humans or animals. All figures cited are published clinical trial results, not recommendations for use.