ACE-031 (Ramatercept)

Chemical identity

ACE-031 is a protein, not a peptide. The FDA Global Substance Registration System records it under the name ramatercept, UNII 42HQC6QLEK, CAS 1169766-01-1, ChEMBL CHEMBL2108674, with INN and USAN designations both entered in 2012. The registered structure is a homodimer of two identical 343-residue chains joined at two interchain disulfide bonds, at Cys122 and Cys125 in the residual IgG1 hinge, with seven further disulfide bonds internal to each chain, sixteen in all. Each chain joins the extracellular region of the human activin receptor type IIB to the C-terminal Fc region of human IgG1 through a three-glycine linker. The calculated formula on that record is C3418H5182N928O1062S38 and the calculated mass 77,500 daltons. That figure covers the polypeptide alone; the same record lists three N-glycosylation sites per chain, at residues 23, 46 and 193, so a measured mass on expressed protein will run higher.

PubChem returns no compound identifier for either name. That is unremarkable for a protein of this size, since the small-molecule databases do not index it, but it has a practical consequence. The identity of material carrying this name cannot be settled by the checks that work on a small molecule. A stated molecular weight or a single chromatographic peak constrains a 343-residue disulfide-linked dimer very weakly, and a molecule of this construction is expressed in mammalian cell culture rather than assembled by solid-phase synthesis. The final section, on what analytical chemists have found inside products sold under the name, returns to this point.

The registered trial record

Four interventional studies are registered on ClinicalTrials.gov, all sponsored by Acceleron Pharma, now listed as a wholly-owned subsidiary of Merck. NCT00755638 gave single escalating subcutaneous doses to 48 healthy postmenopausal volunteers between September 2008 and July 2009 and is marked completed. NCT00952887 ran multiple escalating doses in 70 healthy postmenopausal women aged 45 to 75 from August 2009 to February 2011, also completed. NCT01099761 randomised 24 ambulatory boys with Duchenne muscular dystrophy aged four and over, from April 2010, and is marked terminated, with the field reading: based on preliminary safety data. NCT01239758, an open-label extension, enrolled 11 boys and was terminated on the same grounds. Registered enrolment across the programme totals 153.

Two of the four produced a located publication. Attie and colleagues 2013 reports the single-dose study; Campbell and colleagues 2017 reports the Duchenne trial. The 70-participant multiple-dose study, the largest single enrolment in the programme, has no results posted to the registry and no publication that a PubMed or Europe PMC search on ACE-031 and ramatercept returned. The extension study likewise has neither posted results nor a located paper. Roughly half the people who received this molecule in a registered study are therefore represented nowhere in the retrievable literature.

Development ended on 2 May 2013, when the two sponsoring companies announced they had concluded their collaboration and would not restart the programme, stating that additional nonclinical and toxicology studies did not support further development. Trials had been on clinical hold since February 2011, with a public announcement in April 2011. Those dates and that wording come from an advocacy organisation's relay of company statements rather than from a peer-reviewed paper or a regulatory document, and are recorded here on that footing. The registry entries are consistent with them: both Duchenne studies carry termination dates in the first half of 2011.

What the two published human trials measured

Attie 2013 was a double-blind, placebo-controlled single ascending-dose study in 48 healthy postmenopausal women, randomised three to one to active or placebo across doses from 0.02 to 3 mg/kg subcutaneously. Mean AUC and Cmax rose linearly with dose and mean half-life was reported as 10 to 15 days. At day 29 in the 3 mg/kg group, mean total body lean mass was 3.3% greater by DXA (P = 0.03) and thigh muscle volume 5.1% greater by MRI (P = 0.03). The authors described statistically significant changes in serum biomarkers which they read as effects on bone and fat metabolism, without giving those figures in the abstract. Injection-site erythema was among the reported adverse events.

Campbell 2017 reports the Duchenne trial: subcutaneous administration every two to four weeks in a randomised, double-blind, placebo-controlled ascending-dose design, with safety as the primary objective. The abstract records no serious or severe adverse events, a trend toward maintenance of six-minute walk distance against a decline in the placebo group that was not statistically significant, and trends toward greater lean body mass and bone mineral density and reduced fat mass. It gives no effect sizes for any of those trends. The registry posted only adverse-event outcomes for this study, and the full text is subscription-only at the publisher, absent from PubMed Central and from Europe PMC.

Between them the two papers represent 72 participants. One found a measurable difference in muscle mass in healthy women 29 days after a single injection; the other found directional but non-significant differences in a disease population over twelve weeks of dosing. No published human study reports observation beyond the 24-week window of the Duchenne trial, and no human study of any kind was conducted in men, in younger women, or in anyone without either the Duchenne diagnosis or postmenopausal status.

The vascular findings, in the registry's own numbers

NCT01099761 posted results, which makes the safety signal checkable rather than a matter of summary. The adverse-event table covers 24 weeks, twelve of treatment and twelve of follow-up, at a 5% reporting threshold. In the 1.0 mg/kg every-two-weeks arm of nine boys, epistaxis was recorded in five and telangiectasia in five, alongside injection-site erythema in six and headache in three. In the 0.5 mg/kg every-four-weeks arm of nine, epistaxis appeared in one and telangiectasia in none. Among six boys on placebo, neither epistaxis nor telangiectasia was recorded. No serious adverse events were reported in any arm.

A third cohort appears in the same results table, 2.5 mg/kg every four weeks, with zero participants started. That is the published abstract's phrase about the study being stopped after the second dosing regimen, rendered in registry form: the highest planned exposure was never given to anyone. Nine boys is not a safety dataset in any conventional sense, and the counts above are too small to support a rate. What they do show is that the two vascular events clustered in the arm receiving the larger cumulative exposure and were absent from placebo, which is the observation the programme was halted on.

The same findings in an approved relative

Sotatercept, marketed as Winrevair, is a fusion protein of the same general design: the extracellular domain of the activin receptor type IIA joined to an IgG1 Fc region. It was approved in the United States in 2024 and its label carries an adverse-reaction table. In the STELLAR trial, 163 patients on sotatercept against 160 on placebo, epistaxis was reported in 22.1% against 1.9% and telangiectasia in 16.6% against 4.4%. In ZENITH, 86 against 86, epistaxis appeared in 45.3% against 9.3%, telangiectasia in 25.6% against 3.5%, and gingival bleeding in 10.5% against 2.3%. Median time to onset of telangiectasia was 36.1 weeks.

None of those figures describes ACE-031. Sotatercept traps a different receptor's ligands, was given to adults with pulmonary arterial hypertension, and none of its numbers transfer. What the label does establish is that bleeding and telangiectasia were reported in a licensed molecule built on the same principle, at frequencies large enough to appear in its prescribing information: those reactions sit in section 6.1, and serious bleeding carries its own Warnings and Precautions entry at 5.3. Whether these events reflect a property shared across activin-receptor traps is not established for ACE-031; the sotatercept label describes a different molecule at a different receptor in a different population.

Rodent and primate work, and a naming problem

Cadena and colleagues 2010 treated eight-week-old C57BL/6 mice with ACE-031 or vehicle for 28 days. Mean body weight in the treated group was 16% greater than control, and wet weights of soleus, plantaris, gastrocnemius and extensor digitorum longus were greater by 33, 44, 46 and 26% respectively, all at P < 0.05. Mean fibre cross-sectional area in the soleus rose 22% in type I and 28% in type II fibres, and 57% in the plantaris. Soleus fibre-type distribution did not change. Neither the route of administration nor the group sizes appear in the retrieved abstract.

Fifteen years later, Cadena and colleagues published a non-human primate study: twelve healthy adult common marmosets, six male and six female, eight given 3.0 mg/kg by interscapular subcutaneous injection once weekly for fourteen weeks and four given vehicle. Lean body mass by EchoMRI was greater than baseline in treated animals at weeks 2 to 14; the time-by-treatment interaction was significant (P = 0.0012) but the main effect of treatment was not (P = 0.7506). In biceps brachii, type II fibre cross-sectional area was 20% greater and type I 34% greater than in vehicle controls. Specific twitch and specific tetanic force in the extensor digitorum longus were greater ex vivo, while absolute twitch and absolute tetanic force were not significantly different; the paper's own abstract and discussion instead report an increase in both absolute and specific force.

Dystrophic-model results attributed to ACE-031 usually belong to a different molecule. Carlson and colleagues 2011, Sanchez and colleagues 2015 and Nielsen and colleagues 2017 all worked in mdx mice using RAP-031 or RAP-435, activin receptor traps built with a murine Fc region for administration to mice. The receptor domain is shared; the Fc is not, and neither is the pharmacokinetic behaviour that the Fc governs. Wagner's 2020 review of myostatin inhibition in muscular dystrophy addresses the wider version of this problem directly, asking why a class that improved the phenotype of many mouse models has not produced functional improvement in randomised trials.

What is sold under the name

Reichel and colleagues published a 2025 doping-control study of fourteen products sold as ACE-031 on the grey market. Two contained no protein reactive to an activin receptor IIB antibody at all. The remaining twelve did, but SDS-PAGE showed each carried many additional proteins alongside a main band of about 58.4 kDa, and mass spectrometry with immunoblotting identified that band as the full-length human activin receptor IIB rather than the Fc fusion. Digestion with IdeS protease, which cleaves IgG, failed to cut the products, confirming the absence of an Fc region. None of the fourteen contained ACE-031.

The same paper records ACE-031 as prohibited in sport under chapter S4.3 of the 2024 World Anti-Doping Agency Prohibited List, the category covering agents preventing activin receptor IIB activation, and states that it has never been pharmaceutically approved. The current list could not be retrieved to check whether the compound is still named on it, and the Thevis 2026 banned-substance review corroborates the class heading rather than the compound: its tabulation of the 2025 list gives domagrozumab, stamulumab and bimagrumab as that subgroup's named examples. A search of the FDA Drugs@FDA database for ramatercept returns no match. No medicines regulator has licensed it in any jurisdiction, development stopped thirteen years ago, and the last registered administration to a human took place in 2011.

What is not known

Human exposure to this molecule stopped in 2011 and the published record covers 72 people. Half the registered participants are invisible: NCT00952887, the 70-woman multiple-dose escalation study, posted no results and no publication for it was located in PubMed or Europe PMC, and the 11-boy extension study is in the same position. Nothing was published on multiple-dose pharmacokinetics, on accumulation across repeat administration, or on how quickly the lean mass difference reverses after discontinuation. No trial enrolled anyone outside the two groups that were recruited: postmenopausal women aged 45 to 75, and boys with Duchenne muscular dystrophy aged four and over. No adult male, no premenopausal woman and no child without the Duchenne diagnosis received the molecule in a registered study, and no study addressed long-term administration, immunogenicity, carcinogenicity, or reproductive and developmental effects. The toxicology work the sponsors said informed the 2013 decision has never been published, so the specific finding that ended the programme is not in the public record at all. The mechanism behind the epistaxis and telangiectasia is not established for this molecule; the parallel in the sotatercept label is suggestive of a class effect and is not evidence about ACE-031. On the material side, the position is worse than an evidence gap. The only published analysis of products sold under this name found that none of fourteen contained the fusion protein, which means the published human and animal data describe a molecule that grey-market buyers were not receiving.

Questions