ARA-290 (Cibinetide)

Chemical identity and the names it travels under

ARA-290 and cibinetide are two names for one eleven-residue peptide, and the registries treat them that way. PubChem holds a single record: CID 91810664, CAS 1208243-50-8, molecular formula C51H84N16O21, molecular weight 1257.3 g/mol, InChIKey WZTIQQBMSJTRBR-WYKNNRPVSA-N. The FDA substance registry entry under UNII 9W5677JKDA carries the same CAS number and the same PubChem CID, with the ChEMBL identifier CHEMBL3545305 among its primary codes, alongside INN 10264 and the USAN designation BC-123. European orphan designation records spell the active substance out residue by residue as L-Pyr-L-Glu-L-Gln-L-Leu-L-Glu-L-Arg-L-Ala-L-Leu-L-Asn-L-Ser-L-Ser: the same eleven residues in the same order.

The sequence is eleven residues with a pyroglutamate cap: Pyr-Glu-Gln-Leu-Glu-Arg-Ala-Leu-Asn-Ser-Ser. Several names attach to that one structure. ARA-290 and ARA290 are development codes. Cibinetide is the non-proprietary name. pHBSP and PHBSP abbreviate pyroglutamate helix B surface peptide, the descriptive term used through most of the renal and cardiac literature. All resolve to the same PubChem record. That matters when reading a search result set, because a PubMed query on any one of the names returns only part of the file; the combined query used for this page, ARA-290 OR ARA 290 OR ARA290 OR cibinetide OR pHBSP, returns 101 records.

Metabolism in blood has been characterised once, and not for therapeutic reasons. Thomas and colleagues, working on doping-control assays for peptides under 2 kDa, incubated the peptide with EDTA, heparin and citrate plasma and with serum, and reported extensive exopeptidase-driven breakdown with one main metabolite, PyrEQLERALN (PMID 28941172). That is a nine-residue fragment, the parent shortened by two serines from the carboxy end. The same Cologne group's earlier method paper reported limits of detection between 50 and 500 pg/mL by direct urine injection with ion-mobility mass spectrometry (PMID 26578461).

The design brief, and the paper it came from

Brines and fifteen co-authors published the founding paper in the Proceedings of the National Academy of Sciences in 2008, at volume 105, pages 10925 to 10930. The argument was structural. When erythropoietin binds its receptor homodimer, helix B — residues 58 to 82 — faces the aqueous medium rather than the receptor. The paper proposed that the tissue-protective activity attributed to erythropoietin could be separated from the erythropoietic activity by reproducing that exposed face alone, and reported that an eleven-residue peptide made from the adjacent amino acids on that surface reproduced the protective readouts.

Models used in that single paper included ischaemic stroke, diabetes-induced retinal oedema, peripheral nerve trauma, renal ischaemia-reperfusion, wound healing and a rodent cognition task. The paper reported that neither helix B nor the eleven-residue peptide was erythropoietic in vitro or in vivo. That negative result is the reason the molecule exists: the 2015 review by Collino and colleagues states that the use of recombinant erythropoietin for tissue protection has been limited by rises in haematocrit, platelet activation and selectin expression (PMID 25728128), and the design intent was a compound that engaged the protective pathway without those effects. PubMed applies no correction, erratum or integrity notice to the 2008 paper.

One of its co-authors, Thiemermann, is an author on both of the papers in this file that have since been retracted; a second, Patel, is first author on one of them. The retractions concern those later papers and not the 2008 one, and no notice has been attached to the founding publication. The overlap is recorded here because a reader tracing the file will encounter the same names in both places, and because the retracted work sits in the mechanistic layer rather than the identity layer.

Two retractions and four errata

Collino and ten co-authors published a mouse study in the British Journal of Pharmacology in 2014, at volume 171, pages 5802 to 5815, reporting that eleven weeks of the peptide at 30 micrograms per kilogram subcutaneously reduced insulin resistance, hepatic lipid deposition and kidney dysfunction in male C57BL/6J mice fed a high-fat high-sucrose diet for twenty-two weeks. PubMed types that paper as a Retracted Publication. The retraction notice appeared in April 2024 at volume 181, page 1341, PMID 38433016. The publisher returned HTTP 403 to every attempt to retrieve the notice text, so the stated reason is not reproduced here.

Patel and ten co-authors published a rat acute kidney injury study in Molecular Medicine in 2012, at volume 18, pages 719 to 727, reporting that the peptide given six hours into reperfusion attenuated renal and tubular dysfunction after thirty minutes of ischaemia. That paper was retracted on 26 February 2026, notice at volume 32, article 31, PMID 41749083. The retraction note is specific: the editor-in-chief recorded that in Figure 3a the Total Akt lanes 2, 3 and 4 appeared highly similar, and that in Figure 3d the ph-p38 lane 4 appeared highly similar to Total p38 lane 1. After the authors supplied original autoradiograms, the publisher concluded that a number of bands selected for publication did not appear representative of the observed experimental results. Three named authors do not agree with the retraction; the remainder did not respond to the editor or publisher.

Checking publication types and correction links across all 101 PubMed records returned by the full name set returned no further retractions and no expressions of concern, and four errata. One attaches to a 2018 rat renal allograft study in Transplantation Proceedings, one to a 2014 PLOS ONE study of experimental autoimmune neuritis, and one to a 2020 Bioorganic Chemistry paper describing a technetium-99m labelled dendrimer conjugate of the peptide rather than the parent peptide itself, corrected in May 2020 (PMID 32417522). The fourth attaches to the 2013 sarcoidosis trial, is not cosmetic, and is dealt with next.

The corneal endpoint was corrected three years after publication

Corneal confocal microscopy supplied the objective endpoint for the whole human programme, on the reasoning that small nerve fibres in the cornea can be counted without a biopsy and can be re-counted to detect regrowth. The 2013 sarcoidosis trial reported corneal nerve fibre area, and in October 2016 Molecular Medicine printed an erratum at volume 22, page 674, PMID 28059429. The erratum states that an arithmetical error in the original corneal nerve fibre area algorithm made the reported values too small, and that a reader must multiply the published figure by 2.26, then multiply that result by 6.32 to normalise per square millimetre.

Relative and absolute readouts are affected differently. The trial's headline corneal result was a median increase of 14.5 per cent over baseline, a ratio, which a uniform scaling factor leaves unchanged. Any absolute value in square micrometres taken from the 2013 paper is wrong as printed. Downstream pages quoting absolute corneal areas from that publication without applying the correction are quoting the uncorrected numbers.

Whether the corrected algorithm was applied in the 2017 phase 2b study, which used placebo-corrected change in corneal nerve fibre area as its primary endpoint and reported values around 109, 697 and 431 square micrometres, could not be established. The journal returned HTTP 403 to full-text retrieval and the paper is not deposited in PubMed Central, so only the abstract was available. The abstract does not mention the erratum. This is recorded as an open question rather than a defect.

The receptor the peptide is named for is disputed

The 2008 paper proposed that tissue protection is mediated not by the erythropoietin receptor homodimer but by a heterocomplex of the erythropoietin receptor and CD131, the beta-common receptor, and that proposal was later marketed under the label innate repair receptor. Functional dependence on the beta-common receptor has independent support. Swartjes and colleagues reported in PLOS ONE in 2013 that both ketamine at 50 mg/kg and ARA 290 at 30 micrograms per kilogram relieved allodynia in wild-type mice after spared nerve injury, and that neither did so in mice lacking the beta-common receptor (PMID 23936499). Wang and colleagues reported in 2024 that the protective effect was significantly suppressed in male C57BL/6J mice given small interfering RNA against the beta-common receptor before middle cerebral artery occlusion (PMID 38488446).

Physical association is a separate claim, and the direct test of it was negative. Cheung Tung Shing and ten co-authors published a biophysical study in Scientific Reports in 2018, at volume 8, article 12457, examining the extracellular regions of the two receptors in silico and in vitro, with and without erythropoietin or ARA290 present. Computational and genomic work suggested a possible interaction; the biophysical analysis found that the extracellular regions do not specifically associate. The same paper found no requirement for the beta-common receptor gene under anaemic stress in mice, and concluded that the two receptors do not directly interact. An author correction was printed in May 2019 at volume 9, article 7851.

Functional dependence and physical heterodimerisation are not the same finding, and the literature currently supports the first while the direct biophysical test of the second returned negative. The knockout and knockdown results establish that the beta-common receptor is required somewhere in the chain. They do not establish that the two receptor chains form the complex the name describes.

The human record: four patient trials, the largest with 64 subjects

Heij and colleagues ran the first trial in 2012, in twenty-two people with sarcoidosis and symptoms of small fibre neuropathy, twelve given 2 mg intravenously three times weekly for four weeks and ten given placebo. The screening-list score separated from placebo; the Brief Pain Inventory and the Fatigue Assessment Scale improved in both groups by comparable amounts, and the depressive-symptom inventory did not change. Dahan and colleagues followed in 2013 with thirty-eight subjects, twenty-one active and seventeen placebo, on 4 mg subcutaneously daily for twenty-eight days, and reported separation on the screening list, on the six-minute walk test and on three quantitative sensory thresholds, with no change in intraepidermal nerve fibres at day 28.

Brines and colleagues then studied forty-eight analysable subjects with type 2 diabetes on the same regimen, reporting a divergence in glycated haemoglobin over fifty-six days among the forty-two subjects with values at all three time points, and a difference on the PainDetect questionnaire, with corneal nerve fibre density increasing only in the subgroup whose baseline density was already more than one standard deviation below normal. The largest study, a four-arm dose-ranging phase 2b in sixty-four subjects with sarcoidosis-associated nerve fibre loss, found its primary corneal endpoint separated from placebo at 4 mg per day but not at 1 mg or 8 mg. Pain in that trial improved in every arm including placebo, and the moderate-to-severe subgroup comparison the authors described as clinically meaningful carried a p value of 0.157.

Registry coverage is thin and partly retrospective. A ClinicalTrials.gov search across cibinetide, ARA290 and ARA 290 returns four interventional records, none of which has posted results, and none registered as phase 3. The Karolinska prediabetes and type 2 diabetes study, NCT01933529, carries a status of UNKNOWN with its last update in September 2015 and no publication traceable to it. The Leiden sarcoidosis trial sits in the Netherlands Trial Register as NTR3575, with EudraCT 2012-001492-37, and the Leiden diabetes trial as NTR3858, rather than on ClinicalTrials.gov. The published report of the 2012 pilot names no registry identifier at all, and no registry record for that trial was located.

The diabetic macular oedema trial, run in Belfast in nine patients with no control arm, was first posted to ClinicalTrials.gov in October 2024, seven years after it finished. Its two sources disagree on when it ran: the registry gives a start in April 2016 and completion in August 2017, while the publication gives recruitment from May 2016 to April 2017. The registry entry gives the reason for termination as expiry of study drug with no replacement available. No sponsor has registered a study of the compound since.

Orphan designations, no approval, and a place on the prohibited list

Four United States orphan drug designations are recorded against the substance in the FDA Office of Orphan Products Development database, all to a single commercial sponsor and all listed as Not FDA Approved for Orphan Indication: prevention of delayed graft function following renal transplant, designated 18 September 2009; treatment of neuropathic pain in patients with sarcoidosis, 13 September 2011; treatment of sarcoidosis, 13 June 2016; and treatment to increase survival and improve functioning of pancreatic islets following transplantation, also 13 June 2016. Two European designations exist: EU/3/13/1191 for sarcoidosis, granted 7 October 2013, and EU/3/16/1721 for prevention of graft loss in pancreatic islet transplantation, granted 29 August 2016.

Orphan designation is a development incentive awarded on the basis of disease prevalence and a plausible rationale. It is not a finding of efficacy, not a finding of safety, and not an authorisation. Cibinetide is not an approved medicine in any jurisdiction, no marketing application has reached a public regulatory docket, and material circulating outside a registered trial has been subject to none of the identity, purity or sterility controls that govern investigational supply. Two published analytical-chemistry papers list ARA-290 among the peptides under 2 kDa covered by the World Anti-Doping Agency prohibited list and describe assays developed to detect it in urine.

What is not known

Total human exposure across the published record is roughly 220 subjects: 173 enrolled across the four randomised patient trials, 36 in a randomised study in healthy volunteers and 9 in an uncontrolled study. The largest single trial enrolled 64, and no trial dosed for longer than twelve weeks. Nothing is known about exposure beyond that horizon: no durability data after treatment stops, no repeat-course data, no long-term safety signal detection, and no cardiovascular or renal outcome measurement of any kind. Pharmacokinetics are effectively uncharacterised in public; no dedicated study in any species was located, and the frequently quoted plasma half-life traces to an unreferenced line in a review whose two lead authors are co-authors on both retracted papers. The proposed mechanism is contested at its foundation: functional dependence on the beta-common receptor is supported by knockout and knockdown experiments, but the direct biophysical test of the receptor heterocomplex the compound is named after returned negative. Endpoint validity is a second open question, because the corneal measurement used across the programme required an arithmetical correction three years after first publication, and whether the corrected algorithm was applied in the later phase 2b could not be established from the abstract alone. Dose-response is not monotonic in the one dose-ranging trial: the middle dose separated from placebo on the primary endpoint and the highest dose did not, and no explanation for that pattern has been published. Pain, the symptom the compound is most often associated with, improved in every arm including placebo in that trial. No paediatric, pregnancy, lactation or elderly data exist. No trial has been run against an active comparator. Development appears to have stopped rather than concluded: the last trial to finish was terminated because the study drug expired and could not be replaced, and no sponsor has registered a study since.

Questions