The story of ACE-083 is a rare case of an experimental molecule that hit its primary endpoint three times running and was cancelled anyway. Understanding why is more instructive than another success story: this is exactly where the difference between "the drug works" and "the drug helps" becomes visible.
The idea: don't flood the whole body
Skeletal muscle growth is restrained by TGF-β superfamily proteins, myostatin above all. The "release the brake" logic has been around for a long time, and the first attempts were systemic: block the ActRII receptor throughout the body. The trouble is that those same ligands do other jobs outside muscle, so a systemic intervention brings side effects that are hard to predict.
Acceleron Pharma took a different route. Rather than acting on the whole organism, it engineered a protein that stays where it is injected. ACE-083 is a modified domain of human follistatin fused to an IgG2 Fc fragment. Follistatin has high affinity for heparin and the extracellular matrix, and that property became the delivery mechanism: after injection the protein is retained in the matrix of one specific muscle, while whatever washes into the circulation is rapidly inactivated.
It is worth stressing what ACE-083 is not. It is not a peptide. Nor is it an antibody like bimagrumab, which blocks the receptor and is given intravenously. ACE-083 is a ligand trap: it does not occupy the receptor but intercepts myostatin, activin A, activin B and GDF11 in the extracellular space, before they ever reach it. The distinction is not cosmetic, one molecule silences the signal body-wide, the other only inside a single muscle.
In animals, everything lined up
In preclinical work published in Scientific Reports, ACE-083 produced dose-dependent hypertrophy of the injected muscle only, in wild-type mice and in models of Charcot-Marie-Tooth disease and Duchenne dystrophy. No systemic muscle effect and no endocrine perturbation were seen. Crucially, it was not just mass: isometric contractile force rose, and in CMT mice so did ankle dorsiflexion torque.
In animals, then, a bigger muscle meant a stronger muscle. That promise carried the programme forward.
In humans, half the promise
The first human study enrolled 58 postmenopausal women. ACE-083 was injected into the rectus femoris or tibialis anterior. Volume increased, by up to 14.5% and 8.9% respectively. Tolerability was good, with no serious adverse events and no dose-limiting toxicity.
And in the same report, a sentence that could at the time be blamed on a small sample: no significant changes in mean muscle strength were observed. The muscle grew. It did not get stronger.
Two phase 2 trials, one result
Next came two conditions with focal weakness, where a local approach looked sensible.
In facioscapulohumeral muscular dystrophy (FSHD), ACE-083 at 240 mg per muscle was injected bilaterally every three weeks. Total muscle volume rose 16.4% versus placebo in the biceps brachii and 9.5% in the tibialis anterior, statistically convincing. Contractile volume also increased and fat fraction fell in the TA group. Yet functional tests and quality-of-life questionnaires showed no consistent improvement over 12 months.
In Charcot-Marie-Tooth disease the story repeated almost verbatim: muscle volume rose 13.5% versus placebo, while fat fraction and every other functional outcome were unchanged. In March 2020 Acceleron closed the programme. All the relevant trials are listed as terminated, and ACE-083 is not registered as a medicine anywhere.
The lesson: a surrogate that misled
It is tempting to explain the failure by saying the added tissue was the wrong kind, oedema, inflammation, fat. The data do not support that. In FSHD it was contractile volume that increased, and fat fraction actually fell. The tissue was muscle. Function still did not move.
The simplest explanation is that in these diseases weakness is not caused by a shortage of muscle mass. In CMT the driver is denervation: the muscle has lost its normal neural input, and making it bigger does not restore control over it. In FSHD a primary dystrophic process is under way. Adding tissue to a muscle that is poorly innervated or structurally damaged is not the same as giving it back its function.
Hence the real methodological value of ACE-083: it is one of the cleanest demonstrations that muscle volume and useful muscle function can be dissociated. MRI (magnetic resonance imaging) hypertrophy looked like a persuasive surrogate endpoint, and proved a poor predictor of clinical benefit. For anyone designing neuromuscular trials, that result is worth more than another positive readout.
What stayed open
The programme stopped at phase 2, so a great deal was simply never tested. There are no data in sarcopenia, or in localised atrophy after injury or immobilisation, conditions where muscle has lost mass but kept its innervation. In theory that is where "adding volume" ought to make sense, but the hypothesis was never tested in humans, and any claim about it would be speculation. There is no phase 3, no regulatory dossier and no long-term follow-up either.
The local-delivery technology itself, anchoring a protein in the matrix of a target tissue, remains a sound engineering idea, regardless of ACE-083 failing as a drug candidate.
See also: a general overview of ACE-083 and the locally acting follistatin-Fc.
ACE-083 sits in our Muscle Growth category; the detailed treatment of mechanism, numbers and limitations is in the monograph. The compound is supplied strictly for laboratory research use (RUO): it is not a medicine, supplement or food, is not intended for human or veterinary consumption, and has no established human dose. All doses cited are parameters of published clinical trials, not recommendations.