ARA-290 Has Real Randomised Trials, and Their Main Endpoint Was a Photograph of an Eye
Its sequence does not appear in erythropoietin at all, which is the interesting part. Across a 1 mg to 8 mg range the response did not rise with the dose, the skin biopsy that diagnoses the disease did not move, and I found no trial registered anywhere after the 2017 announcement of progress.
What it is
Eleven amino acids in a row: pyroglutamate-Glu-Gln-Leu-Glu-Arg-Ala-Leu-Asn-Ser-Ser. Molecular weight 1257.3, PubChem CID 91810664, CAS 1208243-50-8. Pyroglutamate is a glutamine whose side chain has curled round and joined the backbone, capping the front of the chain against the enzymes that chew peptides from that end. At least six names circulate, including cibinetide, the approved generic, and pHBSP, under which both retracted papers below are filed.
The design idea is that erythropoietin does two jobs through two receptors: blood making through a pair of erythropoietin receptors locked together, tissue protection through one of them paired with a protein called CD131, which the group that built this peptide named the innate repair receptor. ARA-290 was built to switch on the second and not the first. (PMID 18676614) The European Medicines Agency describes cibinetide the same way, as activating that receptor, which controls inflammatory processes.
The sequence does not appear in erythropoietin
A 2022 paper calls it "an 11-aa non-hematopoietic peptide sequence within the cardioprotective domain of erythropoietin". (PMID 36741836) Read at speed, "sequence within" means a piece cut out of the protein. It is not.
The originating paper was careful. Helix B, residues 58 to 82 of the mature hormone, faces the water when erythropoietin is docked onto its blood making receptor, and what the authors built was "an 11-aa peptide composed of adjacent amino acids forming the aqueous face of helix B". Adjacent on the surface of the folded protein. Not adjacent in the chain.
Check it in a text editor, as I did. Pull up UniProt entry P01588, strip the 27 residue signal peptide off the front, and search the mature sequence for QEQLERALNSS. It is not there. Mapped back by position, the eleven residues are Q58, E62, Q65, L69, E72, R76, A79, L80, N83, S84 and S85: eight of them scattered across the twenty three positions from 58 to 80, with only A79 and L80 side by side and the last three together at 83 to 85. That is closer to taking a rubbing of eleven raised studs spread across a door and threading them onto a wire than to sawing a panel out of the door. Elegant work, and it deserves the credit. It also means erythropoietin's own enormous literature does not transfer here.
What the trials measured
PubMed returns 60 records for ARA-290 or cibinetide, five tagged clinical trial or randomised controlled trial. Adding the one human study it does not tag that way, my count of published human studies is six.
Heij and colleagues ran the 2012 pilot in 22 patients with sarcoidosis, an inflammatory disease of the organs, who had small fibre neuropathy, damage to the thinnest nerve endings, the ones carrying heat, cold and pain. Twelve got ARA 290 and ten a dummy injection, 2 mg into a vein three times a week for four weeks. It was exploratory and designated no primary endpoint. The symptom questionnaire fell 11.5 points against 2.9. (PMID 23168581) Note the route. The trial that established the effect went into a vein.
Dahan and colleagues followed in 2013: 38 patients, 21 on the drug and 17 on a dummy, 4 mg under the skin daily for 28 days, three primary endpoints. Only the corneal one moved, rising a median 14.5 percent over baseline at a P value of 0.022 while the dummy fell 5.3 percent. Read that carefully, because the story rests on it: both are comparisons of a group against its own starting point, and the paper reports no direct test of the drug against the placebo arm on that endpoint. (PMID 24136731) The skin biopsy, the count of nerve endings in a punch of skin and the test used to diagnose this disease, did not move.
Judging a nerve by photographing it is like judging a house rewire by photographing the cable behind the plaster instead of trying the light switch, and the largest trial says as much itself. Culver and colleagues, 2017, put 64 patients on 1 mg, 4 mg or 8 mg a day under the skin or a dummy for 28 days, sixteen per arm, with corneal nerve fibre area used "as surrogate endpoints for disease modification". A surrogate stands in for what you actually care about. Placebo corrected change at day 28 was 109 square micrometres on 1 mg, 697 on 4 mg at a P value of 0.012, and 431 on 8 mg, with only the 4 mg range clear of zero. (PMID 28475703) Doubling 4 mg to 8 mg made the result smaller and the lowest dose did nothing, so across that eightfold spread the response does not rise with the dose, and I found no published explanation and no follow up study. In the posted registry results every one of the four standard deviations is wider than the average it belongs to. (NCT02039687) Pain improved significantly in all groups, the dummy included.
The 2015 diabetes trial analysed 48 patients with neuropathic symptoms, half on 4 mg under the skin daily for 28 days and half on a dummy. HbA1c, a blood test reflecting average blood sugar over about three months, fell 0.16 percentage points at day 28 and 0.21 at day 56 on the drug, against no change and then a rise of 0.21 on the dummy. The corneal measure across everybody rose 1.1 fibres per square millimetre against 0.1, which the paper calls a non significant increase; the positive result is a subgroup of 18 patients who started more than one standard deviation below normal, and it was significant against their own baseline rather than against the dummy arm. (PMID 25387363)
Two more, both small. A Belfast team gave nine patients 4 mg a day under the skin for 12 weeks with no control arm, and it missed its primary endpoint, vision falling 2.9 letters rather than rising; the registry lists it terminated because the study drug expired with no replacement available. (PMID 32674280) And 36 healthy participants were randomised to a single 2 mg dose or a dummy, so roughly 18 got the drug, with the authors reporting no effect on mood. (PMID 26431906)
That is six studies, 218 people enrolled and roughly 133 of them dosed. Three of the six records on the European register are trials with no published result I could find: rheumatoid arthritis at Leiden, authorised December 2010 as an infusion into a vein, complex regional pain syndrome at Leiden in September 2012, and a Swedish study that register calls completed in December 2014 while the American registry still lists it as unknown. Two are still listed as ongoing, fifteen and thirteen years after authorisation.
The programme's between-group positives belong here too, since leaving them out would be the same selective reading the entry is objecting to. In the 2013 trial the symptom score fell further on the drug than on the dummy at week 5, and pain interference and the six minute walk both separated. In the 2015 diabetes trial the PainDetect score fell 3.3 against 1.1 at day 28 and 4.2 against 0.74 at day 56.
Whose evidence this is
Its own, which on most compounds here it is not. The concentration is in the authorship instead. The 2013 sarcoidosis paper states that A Dunne, A Cerami and M Brines are employees of Araim Pharmaceuticals and hold stock or stock options in it. Brines and Cerami also appear on the 2012 pilot and on the retracted rat paper. That does not make a result wrong. It does mean the positive human record came out of one group, and I found no independent replication of the corneal finding.
The two retracted papers
Filter PubMed to retracted publications, search ARA290 or cibinetide, and exactly two records come back.
The first is a 2012 kidney injury paper in rats given pHBSP at 10 micrograms per kilogram into the belly cavity, retracted on 26 February 2026. The editor found blots in one figure that looked duplicated, and the publisher concluded that "a number of bands selected for publication did not appear to be representative of the observed experimental results". The dosing was a single 10 microgram per kilogram injection six hours into reperfusion, in rats given 30 minutes of restricted blood flow and 48 hours of restored flow. Brines and Cerami are co-authors. Three of the other authors disagreed with the retraction. (PMID 22415011)
The second is a 2014 mouse study of insulin resistance, retracted in April 2024. Eighteen mice received pHBSP at 30 micrograms per kilogram under the skin every two days for 11 weeks, out of 42 animals used across the experiment. The PubMed notice carries no reason on the page I read. (PMID 38433016)
Neither retraction touches a human trial, and I found no clinical result on this compound withdrawn. Both are animal work, and this is a documented fact about a literature rather than an accusation about anybody's trials. It bites in one place: the mouse paper is the preclinical account of why a peptide that does not make blood should improve insulin sensitivity, which is what the human blood sugar number gets sold on.
Where the 4 mg came from
Here the site's usual finding runs backwards. The number has a derivation and it traces to a specific experiment, all of it in the 2013 paper. A crossover study in 10 healthy volunteers, each receiving every exposure, measured how much drug the body actually sees, and 4 mg under the skin came out at 59 nanogram-minutes per millilitre against 65 for the 2 mg into a vein that had produced the first result. So 4 mg went under the skin because it reproduces the exposure of the injection that worked, in a form a patient can give themselves at home. When a vendor guide says 4 mg under the skin once daily for 28 days, it is quoting the trial protocol correctly, and that is worth saying without hedging.
The drift is elsewhere, and everything below is reported as convention rather than recommended. Every efficacy trial enrolled patients, and I found no trial in healthy adults beyond that single 2 mg dose study and the crossover in 10. Twenty eight days is the studied course in four of the six studies, and one dosing guide I read gives about four weeks, matching the trial blocks. Another sets out a fifteen week tapered schedule of about 0.03 mg per kilogram that I could trace to no trial. More than one guide states that under the skin is the only route in the trial literature; the intravenous pilot above is why that is wrong. A 250 microgram starting dose also circulates, a sixteenth of the studied amount and below the 1 mg arm that produced nothing, on pages that all refused to load for me, so I report that it circulates without naming a guide I read.
What the regulators have on file
Not approved for any use in any country that I could find. Two European orphan designations are live, EU/3/13/1191 for treatment of sarcoidosis, granted 7 October 2013, and EU/3/16/1721 for prevention of graft loss in pancreatic islet transplantation, granted 29 August 2016. An orphan designation is a development incentive for a rare disease, not permission to sell anything, and this project has seen it read as approval before. For the islet indication I found no human trial, only three PubMed records, all laboratory or mouse work. Araim's press release also claims United States orphan designation for sarcoidosis, United States orphan and Fast Track designations for neuropathic pain in sarcoidosis patients, and a further United States orphan designation for improving pancreatic islet transplantation survival. I could not confirm the American designations against any regulator document, so they rest on the company's own announcement.
Its press release of 8 May 2017 said the phase 2b results "were the basis for a recently-completed successful end-of-phase 2 meeting with the FDA", and that the company looked forward "to taking the next step in clinical development for both neuropathic pain and disease modification indications". I could find no trial of this compound registered anywhere since; the latest start date on ClinicalTrials.gov is April 2016. Araim's own web address now serves a gambling page, which is a fact about what the domain resolves to rather than about the company, still listed as sponsor of the sarcoidosis designation at a Dublin address. I found no statement from any regulator or from the company giving a reason the development stopped.
What could go wrong
Everything here is 28 days or less, in patients, and the whole published exposure is roughly 132 people. The 2013 sarcoidosis trial reported no serious adverse events in its 21 patients on 4 mg under the skin daily for 28 days. The 2015 diabetes trial reported four among its 24 patients on the same schedule, two of which the authors judged unlikely to be related. The phase 2b registry record gives two serious events in the 1 mg arm, one in the 8 mg arm and none in the 4 mg arm or on the dummy. Three on drug and none on the dummy is what chance produces at that size, and the three arms on drug held 48 patients against 16 on the dummy, so the two counts are not measured over the same number of people.
On the central safety claim, that it does not stimulate red blood cell production, the 2013 sarcoidosis trial and the 2015 diabetes trial both report no clinically significant blood count changes, but neither prints haemoglobin, haematocrit or reticulocyte numbers in the sections I read. So that property rests in humans on two statements that nothing changed, over 28 days, in fewer than 50 people between them.
What no one seems to know
Whether the corneal picture means anything to a patient. It is a surrogate, the phase 2b's own abstract says so, and in that trial pain improved in every arm including the dummy. I found no confirmatory trial that would settle it.
Why the response did not rise with the dose across the 1 mg to 8 mg range that trial tested. What the three unreported trials found, and why development stopped after an announcement of progress.
Whether the non blood making property holds beyond 28 days, and what any of this does in a healthy adult, where the only trial I found is a single dose study on emotional processing.
Batch to batch identity in grey market material, which I did not examine at all, so I am flagging it as unexamined rather than absent.
What this adds up to
This one does not take the usual shape. The trials are randomised, placebo controlled and this molecule's own, and the design behind the molecule is among the more elegant things covered on this site.
The endpoint is the problem. Every trial that carried the programme forward was carried by a photograph of the front of an eye, and the skin biopsy a neurologist would use to diagnose the disease sat in the same trial and did not move. The 2013 result the programme was built on was the drug group measured against its own starting point, and the paper never tested it against the placebo arm it had gone to the trouble of recruiting. Then 8 mg did less than 4 mg, and I found no published explanation. The retractions should not be overstated, since both are animal papers and no human result has been pulled that I found, but the mouse one is the mechanism story underneath the blood sugar number and it is gone.
What the vendor pages get right, almost by accident, is the 4 mg. What they get wrong is everything around it, because that amount was worked out for a patient with a diagnosed nerve disease over 28 days, and the person reading the page is healthy and planning to run it for months.
If you have seen ARA-290 sold, open NCT02039687 and read the posted results yourself, then check whether the page that sold it to you named the endpoint. The word to look for is cornea.
Frequently asked
Does ARA-290 have real clinical trials behind it?
It does, and unusually they are its own rather than borrowed from a parent molecule. Six published human studies enrolled 218 people, of whom roughly 133 received the drug, and five of the six were randomised against a dummy injection. The largest randomised 64 patients with sarcoidosis nerve fibre loss and neuropathic pain, sixteen to each of three daily amounts under the skin and sixteen to a dummy injection, for 28 days.
Is ARA-290 a fragment of erythropoietin?
It is not a fragment. I searched UniProt entry P01588 for its sequence, QEQLERALNSS with the first residue in a ring form, and it is not there. Someone identified eleven residues that sit next to each other on one face of the folded hormone and joined them into a chain, so it is a designed copy of a surface. Open UniProt entry P01588 and search the sequence yourself. The practical consequence is that erythropoietin's own literature does not transfer to this peptide.
What did the ARA-290 trials actually measure?
A photograph. The endpoint that carried the programme was corneal nerve fibre area, imaged with a microscope at the front of the eye, and the largest trial's abstract calls that a surrogate endpoint, meaning a measurement standing in for the thing you care about. In the 2013 sarcoidosis trial the skin biopsy that diagnoses the disease did not move, and in the 2017 trial pain improved in every arm including the dummy injection.
Where does the 4 mg figure come from?
From the trials, which is unusual for anything on this site. A crossover study in 10 healthy volunteers found that 4 mg under the skin reproduced the drug exposure of the 2 mg into a vein that had produced the first positive result, and 4 mg daily for 28 days was then what patients received in the main trials. That is the trial protocol, reported here as what was studied in diagnosed patients rather than as a schedule for anyone to run.
Are there retracted papers about ARA-290?
Two, both animal work. Filtering PubMed to retracted publications and searching ARA290 or cibinetide returns exactly two records: a 2012 rat kidney injury paper retracted in February 2026 over duplicated blot images, on which the programme's two principals are co-authors, and a 2014 mouse insulin resistance paper retracted in April 2024 whose PubMed notice gives no reason on the page I read. Neither concerns a human trial, and I found no clinical result on this compound withdrawn.
Is ARA-290 approved anywhere?
Not in any country that I could find. It holds two European orphan designations, granted in October 2013 for sarcoidosis and in August 2016 for prevention of graft loss in pancreatic islet transplantation, and an orphan designation is a development incentive for a rare disease rather than permission to sell anything. The company announced in May 2017 that it had cleared an end of phase 2 meeting with the American regulator, and I could find no trial of this compound registered anywhere since.
Is ARA-290 safe?
The published record is too small to answer that, and its size is the useful part. Roughly 133 people have received the drug across six published studies, nearly all for 28 days or less, and the longest exposure I found was 12 weeks in nine patients. Brines and colleagues recorded four serious adverse events among 24 patients on the drug over 28 days, two of which the authors judged unlikely to be related, and the registry results of the 64 patient trial record three serious events across the 48 patients on drug against none among the 16 on the dummy injection. On the claim that it does not stimulate red blood cell production, two trials report no clinically significant blood count changes over 28 days in fewer than 50 people between them, without printing the red cell numbers in the sections I read.
This content is for educational and informational purposes only and is not medical advice. Peptides discussed are research compounds and may not be approved for human use. Nothing here should be used to diagnose, treat, cure, or prevent any disease. Always consult a qualified healthcare professional before starting any peptide, supplement, or protocol. Individual responses vary. Do not self-administer compounds without proper medical supervision.
Get the next one in your inbox
Free. One email a week at most. Unsubscribe in one click.
Double opt-in · No spam · Unsubscribe anytime
More in Immune and thymic
- LL-37 Has a Human Dosing Record. I Found None of It Under the SkinThree randomised trials I found put it on a wound and one injected it into tumours. The largest missed its primary endpoint, its sponsor went into liquidation in October 2023, and the dose response ran backwards twice, the second time in a different molecule.
- Thymalin Is Calf Thymus Extract, and Its Clinical Record Is Being Lent Forward to Two Other PeptidesThe registered active substance is the words thymus extract, 10 mg from the glands of calves and cattle under one year. Khavinson's group now says the active substances are the dipeptides KE and EW, sold separately as Vilon and Thymogen. In the geroprotection paper everyone cites, 24 people received Thymalin on its own.
- Thymosin Alpha-1: The Approval and the Evidence Are for Different ThingsThe only European registration I could verify is Italian, and its indication section is one sentence about influenza vaccination in immunocompromised patients. The two largest placebo controlled trials I found, in sepsis and in hepatitis C, both came back null.