ACE-083 (Efmitermant Alfa)

Chemical identity

ACE-083 is not a peptide. The FDA Global Substance Registration System records it as a fusion protein under the international nonproprietary name efmitermant alfa, UNII 566QNS333V, CAS 1644543-31-6, with INN number 10811 and USAN code EF-95. ChEMBL carries the same substance as CHEMBL4297867 and types it as a protein at maximum phase 2. The registered structure is a homodimer of two identical 518-residue chains. Each chain places 292 residues of engineered follistatin ahead of a three-glycine linker and a human IgG2 Fc region, and the registry enumerates 42 disulfide links: 20 within each chain and two joining the chains at Cys298 and Cys301. Three N-glycosylation sites per chain are recorded, at residues 95, 259 and 368. The calculated formula is C4928H7718N1386O1548S98 and the calculated mass 114,000 daltons, which covers the polypeptide alone.

Pearsall and colleagues, writing from the originating company in 2019, describe the follistatin portion as FST291, an engineered truncation of the circulating FST315 isoform in which the heparin-binding site remains fully exposed. The registered chain carries 292 residues before the linker rather than 291. That one-residue difference is a numbering convention rather than a dispute about the construct, and it is recorded here because it is the sort of detail that gets copied without checking. The same paper states that the IgG2 isotype was chosen in part to reduce immune effector engagement, a design decision no secondary description of this molecule reproduces.

A query for the exact term ACE-083 in PubChem returns compound identifier 14293713. That record is a small molecule: N-[4-hydroxy-3,5-bis(piperidin-1-ylmethyl)phenyl]acetamide, formula C20H31N3O2, molecular weight 345.5, CAS 81080-10-6, carrying ACE-083 among its supplier-catalogue synonyms. It is not the fusion protein and shares nothing with it. Any page quoting a molecular weight in the hundreds for this compound has retrieved that record. The correct identifiers sit in the substance registries, not in the small-molecule databases, and pubchem_cid is left null here for that reason.

What the molecule was built to do

Follistatin sequesters several transforming growth factor beta superfamily ligands that restrain skeletal muscle growth. The shortest natural isoform, FST288, binds heparan sulfate strongly and is understood to act near where it is released; the longest, FST315, circulates. ACE-083 was assembled to keep the local behaviour and add the persistence of an antibody constant region, and dimerisation through the Fc doubles the heparin-binding sites per molecule.

Surface plasmon resonance in the 2019 characterisation gave apparent equilibrium dissociation constants of 10.3 picomolar for activin B, 15.4 for GDF11, 78.8 for activin A and 348 for myostatin. In a reporter-gene assay in A204 human rhabdomyosarcoma cells, half-maximal inhibitory concentrations ran from 39.2 picomolar for activin A to 676 for myostatin. Elution from an analytical heparin column required 938 millimolar sodium chloride for ACE-083 against 792 for a comparable construct built on full-length FST315. Among bone morphogenetic proteins screened, only BMP6 and BMP7 bound at all, and signalling by BMP9 was not inhibited at concentrations up to 20 micrograms per millilitre.

One property in that paper bears directly on how the molecule behaves outside a muscle. The authors state that ACE-083 reaching the general circulation is rapidly cleaved to inactive fragments, in the manner of the related FST288-Fc construct, and attribute the cleavage to tissue-associated proteinases. Local retention was engineered from three directions at once: the heparin affinity, the paired binding sites, and the proteolysis of whatever escapes. The design intent was containment, not distribution.

The mouse experiments

Wild-type C57BL/6J mice given 100 micrograms of ACE-083 into one gastrocnemius twice weekly for four weeks showed injected-muscle weight 85 percent greater than vehicle, with eight animals per group. Weights of the contralateral gastrocnemius and of the ipsilateral and contralateral rectus femoris were unchanged. In the tibialis anterior the same schedule gave a weight increase of roughly 75 percent, a rightward shift in fibre diameter distribution, and in situ measurements showing twitch force 52 percent greater and peak tetanic force 40 percent greater than vehicle across eight to nine animals. Normalised to the increase in physiological cross-sectional area, the force difference disappeared: absolute force rose, specific force did not.

Seven-month-old Trembler-J mice, a Pmp22 model of Charcot-Marie-Tooth type 1A, received the same regimen in the tibialis anterior. Injected-muscle weight was 73 percent greater than vehicle or the contralateral side, cross-sectional area nearly 70 percent greater, twitch force 44 percent and tetanic force 65 percent greater in five animals, and ankle dorsiflexion torque measured in vivo 137 percent greater than in vehicle-treated animals, also in five. Specific force again did not change.

Four-week-old mdx mice and age-matched C57BL/10J controls were treated in parallel. Injected tibialis anterior weight was 77 percent greater in mdx animals and 116 percent greater in wild-type controls than the uninjected contralateral muscle, with eight to ten per group. Cross-sectional area rose 42 percent in mdx against 70 percent in wild-type, twitch force 16 against 34 percent, tetanic force 34 against 44 percent. Serum creatine phosphokinase in mdx animals fell 36 percent, which the authors report as not statistically significant. The pattern across all three models is consistent: the effect on the dystrophic muscle was smaller than the effect on the healthy one.

The registered trial record

Four interventional studies appear on ClinicalTrials.gov, all sponsored by Acceleron Pharma, now listed as a wholly-owned subsidiary of Merck. Registered enrolment totals 278: NCT02257489, a phase 1 study in 58 healthy postmenopausal women, completed April 2016; NCT02927080, a two-part phase 2 study in 95 adults with facioscapulohumeral muscular dystrophy, terminated October 2019; NCT03124459, a two-part phase 2 study in 63 adults with Charcot-Marie-Tooth type 1 or X, terminated March 2020; and NCT03943290, an extension enrolling 62 people from the two parent trials, terminated the same day. All four have results posted, which is unusual for a discontinued programme and makes most of this record checkable against registry tables rather than abstracts.

Terminations were attributed to efficacy, not harm. The FSHD entry gives the reason as: Study was discontinued as it did not achieve functional secondary endpoints. The CMT entry opens with an unmatched parenthesis and reads: (Investigation of ACE-083 for use in patients with CMT is being discontinued as it did not achieve functional secondary endpoints in the A083-03 trial. The extension names both parent trials, misspelling one as A083-023. Of the 62 people enrolled in that extension, zero are recorded as completing it.

Phase 1 enrolled women aged 45 to 75 only, 42 to active and 16 to placebo, given 50 to 200 milligrams into one rectus femoris or tibialis anterior as one or two injections three weeks apart. Maximum mean increase in injected muscle volume by MRI was 14.5 percent, standard deviation 4.5, in the rectus femoris and 8.9 percent, standard deviation 4.7, in the tibialis anterior, both three weeks after the last dose. No statistically significant change in strength was observed by hand-held dynamometry or fixed system, including dorsiflexion strength.

Systemic exposure was barely measurable. Fourteen of 252 serum samples from six of twelve treated participants exceeded the assay's lower limit of quantitation of 10 nanograms per millilitre, the highest observed concentration was 37.2 nanograms per millilitre, and the authors state that only limited pharmacokinetic variables could be estimated. No half-life was reported in that paper or in any other publication located.

What the two disease trials measured

In the FSHD study, the randomised portion compared bilateral injection every three weeks against placebo in either biceps brachii or tibialis anterior, with 55 participants evaluable. Total muscle volume by MRI, the primary endpoint, showed a least-squares mean treatment difference of 16.4 percent in the biceps group with a 90 percent confidence interval of 9.8 to 23.0 (P below .0001) and 9.5 percent in the tibialis group, interval 3.2 to 15.9 (P = .01). Contractile muscle volume rose in both, by 23.3 and 18.4 percent. Fat fraction fell significantly in the tibialis group only, by 2.7 percentage points (P = .04).

Function did not follow. Six-minute walk distance differed by minus 5.3 percent in favour of placebo (P = .20), ten-metre walk or run by 4.7 percent (P = .35), four-stair ascent by 0.45 percent (P = .94), and ankle dorsiflexion maximum voluntary isometric contraction by minus 14.54 percent (P = .22). Elbow flexion contraction in the biceps group did reach significance at day 190, at 36.12 percent (P = .02), and the authors record that the improvement was not maintained across twelve months and did not appear when the placebo group crossed over to active treatment. Their own summary calls the mismatch between volume and function apparent and offers six possible explanations without choosing among them.

The CMT study reported a least-squares mean difference in total muscle volume of 13.5 percent, given as P = 0.0096 in the paper and posted as P = 0.01 with a 90 percent interval of 4.9 to 22.1 in the registry. Posted secondary outcomes give a treatment difference of 35.4 percent in quantitative muscle testing of the injected muscle, interval minus 3.2 to 74, P = 0.13; minus 3.1 cubic millimetres of intramuscular fat (P = 0.16); 1.9 percent on walk or run time (P = 0.73); and 3.1 percent on walk distance (P = 0.51).

Two accounts of that trial disagree. The published abstract states that there was a significant difference for contractile muscle volume and for change in ankle dorsiflexion strength. Contractile muscle volume was not among the outcomes posted to the registry, and the strength outcome that was posted is not significant. The full text is subscription-only, absent from PubMed Central and Europe PMC, so the two could not be reconciled from the retrievable record. Paper and registry agree on the conclusion: participants receiving ACE-083 did not show greater functional improvement than those receiving placebo.

Adverse events and immunogenicity

Injection-site events dominate every posted table. In the FSHD randomised portion, injection-site erythema was recorded in 6 of 14 in the tibialis active arm against 0 of 15 on placebo, and pruritus in 6 of 14 against 0 of 15; in the biceps arms bruising appeared in 9 of 14 active against 10 of 15 placebo, which is the pattern of a needle rather than a molecule. Serious events across that trial were one lung abscess, one sepsis, one pneumonia, one diverticulitis on placebo, and one invasive ductal breast carcinoma during open-label treatment. The CMT trial recorded one hypokalaemia, one suicide attempt on placebo and one cellulitis. No deaths were reported in any arm of any of the four studies.

One neurological event is worth its own line. A participant receiving ACE-083 in the tibialis anterior during the FSHD double-blind period had left foot paraesthesia, left partial peroneal nerve axonal injury and hypoesthesia of the dorsum of the left foot, all graded 2 and assessed as probably drug-related, leading to dose reduction. The paper records that these resolved with sequelae, describing progressively improving weakness. Another participant withdrew after grade 2 paraesthesia beginning in the feet.

Antibodies against the drug were common. In part 1 of the FSHD study, 8 of 37 participants tested positive, 21.6 percent. During the randomised portion, 11 of 28 receiving ACE-083 were positive, 39.3 percent, against one participant on placebo. Seventeen more tested positive during open-label treatment. In the phase 1 study three participants had confirmed positive results, one with titres ranging from 40 to 2,560 that had fallen to 80 twelve weeks after the last dose; that participant was lost to follow-up. Whether the antibodies affected exposure or response is not established, since the assays that would answer it were near their limits at the concentrations this route produced.

Status, and what circulates

No regulator has approved this molecule. A query of the FDA drug approval database for efmitermant returns no match, and no product label mentions ACE-083. Three orphan drug designations are recorded on the substance registry, numbered 312310 for Duchenne muscular dystrophy, 573617 for facioscapulohumeral muscular dystrophy and 674918 for Charcot-Marie-Tooth disease, and all three carry the status Designated/Withdrawn. Administration in a registered study last occurred in the trials that ended in March 2020.

On the 2026 World Anti-Doping Agency Prohibited List, effective 1 January 2026, section S4.3 covers agents preventing activin receptor IIB activation and includes, but is not limited to, myostatin-binding proteins, giving follistatin and myostatin propeptide as its examples. ACE-083 is not named anywhere in that document; ACE-031 is, as an example of a decoy activin receptor. A follistatin-based fusion protein falls within the class heading by construction rather than by listing. Analytical methods exist: Walpurgis and colleagues published assays in 2020 using monomeric follistatin and dimeric follistatin-Fc constructs as model compounds, with an estimated detection limit of 10 nanograms per millilitre, and a 2025 equine method covers follistatin-Fc among nine Fc-containing proteins.

Material sold under this name has never been analysed in the published literature. A PubMed search returns five records for the exact term ACE-083: two phase 2 trial reports, the phase 1 report, the preclinical characterisation, and a Chinese-language review of Charcot-Marie-Tooth type 1A that mentions the compound in a list. Nothing analogous to the 2025 doping-control analysis of grey-market ACE-031 products exists here, so no published work establishes what a product carrying this name contains. Vendor pages describe a 114-kilodalton glycosylated protein expressed in mammalian cell culture using the vocabulary of a synthesised peptide, and several offer it in vial sizes an order of magnitude or more below the amount given per muscle in every registered study.

What is not known

The programme answered one question and left the important one open. Injecting a muscle with this protein enlarged it, reproducibly, in mice, in healthy women and in two disease populations, and that result is as well documented as an abandoned compound's results get: three published trials, four posted registry result sets, consistent effect sizes. What the enlarged muscle could do afterwards was measured repeatedly and never improved in a way that held. The mouse work had already shown the shape of it, in that absolute force rose while specific force did not. No published study establishes why, and the FSHD investigators list six candidate explanations without adjudicating between them. Nothing is known about administration beyond twelve months, about what happens to the added volume after treatment stops, or about repeated dosing into more than one or two muscles at a time. No man received it in the phase 1 study, which enrolled only women aged 45 to 75, and no one under 18 received it in any study. Anti-drug antibodies appeared in roughly a third to two-fifths of treated participants, and whether they affected exposure or response was not determined, partly because the concentrations this route produces sit near the floor of the assay. The unpublished extension study holds 62 people's worth of data on administration for up to two years, with results posted only as adverse-event tables. On the material side the position is simply blank: unlike its relative ACE-031, no published analysis has ever examined what is inside a product sold under this name, so nothing can be said about whether such material contains a 114-kilodalton glycosylated fusion protein at all.

Questions