The Antibody That Tells Muscle Not to Leave With the Fat
Everyone who has spent time around GLP-1 medications knows the muscle question. Weight comes off quickly, the scale rewards you weekly, and somewhere in that total is lean tissue you would rather have kept. The standard answer has been resistance training and protein, which is good advice and has never been especially satisfying, because nobody has been able to say how much of the loss it actually prevents. In June 2025 a phase 2 trial called BELIEVE put a different answer on the table. It added bimagrumab, a monoclonal antibody that blocks a receptor controlling muscle growth, on top of semaglutide, and reported that lean mass loss dropped from 7.4 percent to 2.9 percent while fat loss went up rather than down. That is a striking result and it deserves a careful reading, because the same trial also found that roughly two thirds of the people on the combination reported muscle spasms. This is educational background on a compound in active development. Bimagrumab is investigational, is not approved for any indication, and nothing here is medical advice.
The Number That Started the Argument
In BELIEVE, participants on semaglutide 2.4 mg lost 7.4 percent of their lean mass over the course of the study. That figure lands in the range the wider literature has been reporting for several years, and the reason it generates so much discussion is that it is not obviously bad on its own. A man carrying substantial excess weight has more lean tissue than a lean man of the same height, because the body builds skeletal muscle and connective tissue to carry the load. Some of that surplus is not doing anything useful once the load comes off. The honest position has always been that a portion of lean mass loss during weight reduction is appropriate and a portion is not, and no measurement in routine use can tell you which is which. What makes the question urgent rather than academic is who is taking these drugs. A 58-year-old who is already near the bottom of the age-related decline in muscle does not have the same margin as a 34-year-old, and the retrospective cohort work in older adults with type 2 diabetes has been pointing at accelerated sarcopenia in exactly the people with the least to spare. That is the gap BELIEVE was designed to probe.
What Happened to Lean Mass
This is the chart the trial was built to produce. Semaglutide alone took lean mass down 7.4 percent. Bimagrumab alone added 2.5 percent while participants were losing weight, which is an unusual combination. Together, lean mass still fell, but by less than half of what the GLP-1 arm gave up. Bear in mind that lean mass on a DEXA scan is not the same measurement as muscle strength.
What Bimagrumab Actually Blocks
Bimagrumab is a fully human monoclonal antibody that binds activin type II receptors, both ActRIIA and ActRIIB, and physically occupies them so their natural ligands cannot dock. The two ligands that matter most here are myostatin and activin A, and both are brakes. Myostatin exists to limit how much skeletal muscle the body builds, which is why the rare cattle and human cases with loss-of-function mutations in the gene are visibly, unusually muscular. Block the receptor those signals arrive through and the downstream ALK4 pathway goes quiet, and muscle tissue does what it does when the brake is released. That much was established well before anyone thought about obesity. The part that made bimagrumab interesting to a weight-loss program is the second effect. The same receptors are expressed in adipose tissue, and blocking ligand signalling there appears to increase lipid mobilization and lipolysis. So the compound is not simply a muscle drug with a neutral effect on fat. In the trials it reduces fat mass and adds lean mass at the same time, which is a rare pharmacological combination and the reason the body composition results look the way they do.
The 2021 Trial That Made This Worth Testing
The obesity case for bimagrumab did not start with semaglutide. In 2021 a phase 2 trial published in JAMA Network Open randomized 75 adults with type 2 diabetes and obesity to bimagrumab or placebo for 48 weeks, with Steven Heymsfield among the investigators. Fat mass fell about 21 percent, a loss of 7.5 kg, against 0.5 percent in the placebo group. Lean mass rose 3.6 percent, an increase of 1.7 kg, while the placebo group lost a small amount. Total body weight came down 6.5 percent. Read those three numbers together and you can see what happened: participants lost a great deal of fat, gained muscle, and the scale moved only moderately because the two changes partly cancelled. HbA1c improved by about three quarters of a percentage point alongside it. Six and a half percent body weight is not competitive with what GLP-1 drugs deliver, and bimagrumab was never going to win a weight-loss race on its own, because it does not touch appetite. What the trial demonstrated was that the compound changes body composition in a direction nothing else was moving it. That set up the obvious question of what happens when you pair it with a drug that is very good at the part it cannot do.
Total Weight Reduction by Arm at 72 Weeks
The headline result from BELIEVE, taken at the end of the 24-week extension. Bimagrumab on its own moved the scale less than semaglutide did, which is expected given that it is not an appetite drug. Stacked together the two produced more total loss than either alone. The placebo figure is small and not zero, which is normal in a trial where everyone receives lifestyle counselling.
BELIEVE, and Why It Had Nine Arms
BELIEVE enrolled 507 adults with obesity, defined as a BMI of 30 or above, or a BMI of 27 or above with at least one weight-related condition. People with diabetes were excluded, which matters when comparing these results against the 2021 trial. The design was randomized, double-blind and placebo-controlled, and it ran 48 weeks to the primary endpoint with a 24-week extension after it. The nine arms were placebo, bimagrumab at 10 mg/kg and at 30 mg/kg, semaglutide at 1 mg and at 2.4 mg, and every combination of the two drugs at both dose levels. That structure is expensive and it exists for a specific reason. With monotherapy arms at both doses sitting alongside the combinations, the trial can separate what each drug contributes from what the pair does together, rather than reporting a combination result and leaving everyone to guess. Results were presented at the American Diabetes Association Scientific Sessions in June 2025, with Heymsfield covering efficacy and safety, and the full paper subsequently appeared in Nature Medicine. Eli Lilly holds the asset, having acquired Versanis Bio in 2023 in a deal reported at up to 1.9 billion dollars. Versanis had itself been spun out of Novartis, where the compound was developed as BYM338.
What the Body Composition Did
At the 72-week readout the high-dose combination produced 22.1 percent total body weight loss, against 15.7 percent for semaglutide 2.4 mg alone and 10.8 percent for bimagrumab alone. Around 92 percent of the weight lost on the combination was fat. Total fat mass fell 45.7 percent against 27.8 percent for semaglutide alone, and visceral adipose tissue, the deposit packed around the organs that tracks most closely with insulin resistance, came down 58.2 percent against 35.8 percent. Lean mass is where the trial earns its name. Semaglutide alone gave up 7.4 percent. Bimagrumab alone gained 2.5 percent while participants were losing weight. The combination held the loss to 2.9 percent, which the investigators reported as significantly better than semaglutide alone. At the 48-week primary endpoint the corresponding numbers were 20.2 percent weight loss for the combination against 14.8 percent, with lean mass down 2.6 percent against 7.9 percent. Waist circumference fell by roughly 22 cm on the combination. Two markers moved in ways worth noting alongside the composition data: high-sensitivity C-reactive protein dropped 83 percent, and adiponectin more than doubled. Both are consistent with visceral fat being the compartment that emptied.
Where the Fat Came Off
Reduction from baseline in total fat mass and in visceral adipose tissue, comparing semaglutide 2.4 mg against the high-dose combination. Visceral fat is the deposit that sits around the organs and tracks most closely with insulin resistance and cardiometabolic risk, so the gap in the right-hand pair carries more weight than a scale reading would.
- Semaglutide 2.4 mg
- High-dose combination
Why the Same Trial Gets Quoted Two Different Ways
If you go looking at the coverage you will find BELIEVE described as a 20.2 percent result and as a 22.1 percent result, sometimes in articles published within days of each other. Both are correct. The first is the 48-week primary endpoint and the second is the end of the 24-week extension, and not every write-up says which one it is quoting. The same applies to the lean mass figures, where 2.6 percent and 2.9 percent are the same finding at two different time points. This is worth flagging because it is exactly the kind of ambiguity that turns into a bad comparison later, when somebody sets the 22.1 percent number against a competing drug's 48-week result and concludes something the data do not support. Anyone reading trial coverage in this category should be checking the time point before comparing anything to anything, and the habit generalizes well beyond this one study.
One Trial, Two Sets of Numbers
BELIEVE ran 48 weeks and then extended another 24. Both readouts have been reported, and coverage has quoted them interchangeably without always saying which is which. The difference is not large, but it explains why the same study appears in one article as a 20 percent result and in another as a 22 percent result.
- Semaglutide 2.4 mg
- High-dose combination
The Part the Headlines Understate
Muscle spasms were reported by 46.4 percent of participants at the 10 mg dose and 63.6 percent on the high-dose combination. That is not a footnote. Nearly two out of three people on the arm that produced the best body composition result were experiencing muscle cramping, and five participants in the bimagrumab monotherapy arms discontinued because of it. Across the combination groups, about 9 percent stopped treatment for adverse events of any kind over the 72 weeks. Diarrhea and acne were the other common findings, with the gastrointestinal effects difficult to separate from what semaglutide does on its own. Serious adverse events ran 8.8 to 12.5 percent in the bimagrumab arms, 1.8 to 10.7 percent in the semaglutide arms, 3.6 to 9.1 percent in the combination arms and 3.6 percent on placebo. There were no deaths, and the investigators reported no new safety signals beyond what was already known about each drug. Cramping in a trial of a muscle-targeting antibody is not a surprising finding, and the practical suggestions have been the ordinary ones about hydration, electrolytes and checking magnesium. Whether people would tolerate it for years rather than for 72 weeks is a different question, and this trial cannot answer it.
The Side Effect That Defines the Drug
Muscle spasms were reported by 46.4 percent of participants at the lower bimagrumab dose and by 63.6 percent on the high-dose combination. That is not a rare event buried in an appendix, it is the most common thing that happened to people in this trial. Five participants in the bimagrumab monotherapy arms stopped treatment because of it, and roughly 9 percent of those in the combination arms discontinued for adverse events of any kind over 72 weeks.
Lean Mass on a Scan Is Not Strength in a Hand
Here is the caveat that deserves more attention than it has been getting. Lean mass measured by DEXA is everything in the body that is neither fat nor bone mineral, which means skeletal muscle plus organs plus a large amount of water. Glycogen alone binds roughly three grams of water per gram stored, so shifts in carbohydrate intake register as lean mass changes without a single muscle fiber being involved. A drug that increases the number is not automatically a drug that increases the muscle, and even when it genuinely does add tissue, added tissue and added function are separate claims. Bimagrumab has an instructive history on precisely this point. It was tested in sporadic inclusion body myositis, a muscle-wasting disease, and it increased muscle mass without producing the clinical benefit the trial was looking for. The muscle got bigger and the patients did not get better. That result is what makes the physical function data from BELIEVE and its successors the thing to watch, rather than the composition percentages. A 2.9 percent lean mass loss instead of 7.4 percent is only meaningful if it shows up as something a person can do.
What Is Still Unresolved
Bimagrumab is not in phase 3 for obesity. As of the middle of 2026 it remains a phase 2 asset, tested with semaglutide in BELIEVE and under study in combination with tirzepatide, and Lilly withdrew a planned phase 2b trial in type 2 diabetes and obesity in September 2025 while keeping the tirzepatide combination program running. Development decisions of that kind are rarely explained in public and are usually about portfolio strategy rather than a specific finding, but they are also not nothing. The comparison nobody has run is the one most readers actually want. There is no trial setting bimagrumab plus a GLP-1 against a GLP-1 plus supervised resistance training and adequate protein, which means the incremental value of the antibody over the free intervention is unknown. Structured training with protein intake at the upper end of the usual recommendations reliably preserves a meaningful fraction of lean mass during weight loss, and until somebody funds the head-to-head, any claim about how much the drug adds on top of that is an assumption. Cost is unresolved for the same reason: an injected monoclonal antibody stacked onto an incretin is not going to be a cheap protocol, and nobody has had to price the benefit yet.
The Unglamorous Version That Already Works
None of this changes what a person on a GLP-1 today should be doing, which is the thing that has been true since before bimagrumab entered the conversation. Resistance training two or three times a week, protein intake somewhere around 1.6 grams per kilogram of body weight, and a rate of weight loss slow enough that the body is not forced to catabolize tissue to keep up. The rate point is the one most often skipped. Aggressive titration to the highest tolerated dose produces the fastest scale movement and the largest lean mass cost, and a slower approach at a lower dose trades some speed for a better composition of what comes off. That trade is available now, requires no additional prescription, and is the reason lower-dose protocols keep coming up in these discussions. What BELIEVE contributes is a demonstration that the lean mass loss is not fixed. It is a variable that responds to intervention, and the size of the response, 7.4 percent down to 2.9 percent, sets a rough ceiling on how much of the problem is addressable at all. Whether the eventual solution is an antibody, a barbell, or a slower dose curve is a question the next several years of trials will answer, and the barbell has a considerable head start on evidence for the thing that actually matters, which is what the muscle can do rather than what it weighs.
Frequently Asked Questions
Have more questions? Contact us or explore our full peptide intelligence platform.
Related Articles
The Entire Case for Magnesium and Testosterone Rests on Thirty Men in Turkey, Four Weeks, and a Dose Twice the Supplemental Upper Limit. The Ten Who Took No Magnesium Went Up Too.
Magnesium appears in almost every testosterone stack on the market, and the number behind that placement comes from one 2011 study by Cinar and colleagues: thirty men aged 18 to 22, split into three groups of ten, four weeks, 10 mg per kilogram of body weight a day, which is about 800 mg for an 80 kg man against a 350 mg upper limit for supplements. Total and free testosterone rose in all three groups, including the ten athletes who took no magnesium at all. This article walks through every human study that has measured magnesium against testosterone, explains why a normal serum magnesium result tells you almost nothing, compares the elemental content and absorption of every form on the shelf, and separates what magnesium plausibly does from what it is sold for.
One Week of Five-Hour Nights Cut Daytime Testosterone by 10 to 15 Percent in Ten Healthy Young Men, a Drop That Normal Aging Takes 10 to 15 Years to Produce. Here Is What Every Human Sleep and Testosterone Study Actually Found.
Sleep is the only testosterone intervention with a tightly controlled human experiment behind it. In the 2011 JAMA study by Leproult and Van Cauter, ten healthy men aged about 24 slept five hours a night for eight nights in a laboratory and their daytime testosterone fell 10 to 15 percent, with the deepest trough between 2 pm and 10 pm. Older men who slept around four hours had morning testosterone roughly half that of men sleeping eight, and in 2,295 US men each hour of sleep lost tracked with about 5.85 ng/dL less testosterone. This article walks through every controlled restriction study, explains why a single 8 am blood draw can miss the effect, and separates the sleep signal from the obesity and sleep apnea it travels with.
A 20-Second Squeeze Test Predicted Death Better Than Blood Pressure in 139,691 People Across 17 Countries. Every 5 kg of Grip Strength a Man Loses Adds 16 Percent to His Mortality Risk, and It Is Measured in Almost No Physical.
In the PURE study of 139,691 adults in 17 countries, each 5 kg drop in grip strength was associated with a 16 percent higher risk of death from any cause and a 17 percent higher risk of cardiovascular death, and grip predicted mortality more strongly than systolic blood pressure. UK Biobank reproduced the finding in 502,293 people. Men peak at roughly 51 kg in their thirties, cross the 27 kg sarcopenia threshold in their eighties if nothing changes, and lose strength faster on GLP-1 medications unless they train. An educational review of what grip strength measures, why it predicts survival, what testosterone and resistance training do to it, and where the evidence stops.