ARA-290 (cibinetide): a fragment of erythropoietin that acts on nerves, not on blood

Two jobs, one hormone, and a peptide cut to do only one of them

The reasoning that produced ARA-290 runs in five steps, and each step is a published result rather than a guess.

1. Erythropoietin tells bone marrow to make red blood cells. It does that by clamping two identical erythropoietin docking points together. This is the effect that makes the hormone a doping agent and a stroke risk.
2. Erythropoietin also keeps injured tissue alive and helps it rebuild. That was seen in stroke, in kidney damage, in heart damage and in nerve damage, decades before anyone knew why.
3. The two effects are not the same signal. Brines and Cerami showed that the repair effect runs through a different docking point entirely: one erythropoietin subunit joined to a second protein called CD131. They named the pair the innate repair receptor.

1. Because the two docking points are built differently, the two effects can be pulled apart by design. A molecule shaped to fit the repair pair, but too small to clamp the blood pair, would rebuild tissue without touching the marrow.
2. The part of erythropoietin that faces outward and does the repair binding is called the helix B surface. An eleven-residue peptide that reproduces that surface, with its front end looped shut so enzymes cannot chew it, is ARA-290 — also written pHBSP.

What that produces is not a weaker dose of erythropoietin. It is a different molecule that physically cannot reach the docking point erythropoietin uses to thicken your blood.

What is dying back, and what would have to grow for it to stop hurting

Small-fibre nerve damage is a degeneration story with a specific shape, and the whole case for this compound is that it acts on one particular link in it.

What is breaking down. The nerve endings that carry pain, temperature and sweating signals are the thinnest fibres in the body — unmyelinated C fibres and thinly wrapped A-delta fibres. They end in your skin, in your cornea, in the lining of your gut. They are the furthest thing from the cell body that keeps them alive, so they are the first to die back when anything goes wrong upstream.

What makes it break down faster. Two drivers, and they are different diseases. In sarcoidosis, immune cells clump into granulomas and the inflammation eats the fibres. In diabetes, high blood sugar starves and poisons them from the inside. Both drivers keep working while the fibre is trying to survive, which is why removing the driver matters more than any repair signal.

What the pain actually is. As fibres die back, the ones left behind fire without being touched. That is why the pain is burning, electric and worse at night rather than sharp and located. The pain is not a measure of how much nerve you have left. It can rise while fibres are still dying and fall while fibres are still gone.

What would have to grow. New fibre, sprouting from the surviving stump outward, re-entering the skin. That is a slow, expensive process for a cell, and it is switched off by the same inflammation that caused the damage. Two things have to happen: the surviving cell has to not die, and it has to be given the signal to extend.

What this compound does to that chain, and how strongly the evidence holds at each step. It switches on a repair receptor that only assembles on damaged cells — shown directly in cells and animals, strong. Switching it on stops injured cells killing themselves and quiets the immune cells around them, dropping IL-6, IL-12 and TNF-alpha — shown in cells and animals, strong. Quieting that inflammation lets fibres regrow — shown in people, in two randomised trials, by counting fibres, moderate. Regrown fibres make the pain better — shown once, weakly, and missed twice. That last link is the weak one and it is the one you care about.

The receptor only exists where there is damage

The repair docking point is not sitting on healthy cells waiting to be switched on. The erythropoietin subunit and the CD131 subunit sit apart until injury, inflammation or metabolic stress brings both to the cell surface at the same time. Only then does the pair exist. Only then is there anything for ARA-290 to bind.

Daniel Culver of the Cleveland Clinic, who ran the largest trial, described the assembly to a room of patients in plain terms: a subunit "comes out of the inside of the cell and comes up and joined its partner, joins the beta common receptor here on the surface the cell and it makes this dimer, this two-headed receptor."

Statement

Official disclosure

Webinar transcript: Small Fiber Neuropathy and ARA-290 Results - Dr. Daniel Culver, Cleveland Clinic, 17 May 2017 (Foundation for Sarcoidosis Research)

“By day 56 some of that goes back to the baseline. ... it looks like 28 days is probably not going to be enough to maintain a durable benefit.”

Aug 4, 2026 · stopsarcoidosis.org

2026-08-04 08:29s28 · #6a1f15ea0164

Two things follow from that, and both are practical.

The drug does something where there is damage and nothing where there is not. That is the mechanical reason its safety record across the trials is as clean as it is — there is no receptor for it to act on in healthy tissue.

And the size of any effect is capped by how much damaged tissue is putting the receptor out. A person with nothing wrong should expect to feel nothing at all. That is not a disclaimer; it is what the trials found. Eleven of the diabetic subjects in the 2015 trial had normal corneal nerve counts at the start. They had nothing to repair, and their numbers did not move.

Six trials in people: three hit, three missed

Every registered or published human study of ARA-290 and cibinetide, with the result stated the way the trial stated it.

A seventh trial was registered and never reported: NCT01933529, a Karolinska study in prediabetes and type 2 diabetes, planned for 24 people, still listed as status unknown against a finish date of December 2015.

Across all six trials, about 132 people have received the real drug rather than placebo. The longest anyone has taken it under observation is 12 weeks, and that was the trial that failed.

Three of the four wins came out of one disease

Three of the four positive results sit inside a single condition, which changes how far you can read them across to anything else.

In sarcoidosis, the small-fibre damage is driven by the immune system. Clumps of inflammatory cells damage the thin fibres that carry pain, temperature and sweating signals. Unlike diabetes, the damage is often not worst at the far end of the limb — it comes in patches rather than the glove-and-stocking pattern. About half of sarcoidosis patients with small-fibre damage carry the inflammatory protein signature of the far-end-worst form.

The people enrolled were not mild cases. In the Phase 2b, more than 80% were already on painkillers, about two-thirds were on nerve-pain drugs, most were taking around two drugs a day for it, and, in Culver's words, "usually not achieving very good benefits." Their skin fibre counts at the start were roughly half those of healthy people the same age and sex.

The diabetes trial is the one win outside sarcoidosis, and its nerve result was a subgroup result. Eleven of the diabetic subjects already had corneal nerve counts within one standard deviation of normal — they had nothing measurable to repair. The +2.6 fibres/mm² gain belongs to the 18 subjects who were genuinely abnormal to start with.

They counted the nerves in a microscope instead of asking how it felt

What makes this data set unusually believable is not the pain scores. It is where they pointed the camera.

Your cornea has more nerve endings per square millimetre than anywhere else on your body, and they sit a few hundred microns under a clear window. A confocal microscope photographs that mesh in a conscious patient in a few minutes — no numbing, no cutting, no biopsy. Software then counts the fibres per square millimetre, the branch points, the total length, and the number the Phase 2b used: corneal nerve fibre area, meaning how much of the picture is nerve.

What that buys you, against the alternatives:

• Against a pain questionnaire. Pain scores moved in every arm of every trial. Culver's summary of the Phase 2b was that "every single group had improvements including the placebo." A photographed count of nerve fibres does not respond to hope.
• Against a skin biopsy. Counting fibres in a punch of skin is the reference method, but it needs holes cut in you at several time points, is read in a handful of specialist labs, and in this drug's own trials was the measure that failed to separate from placebo twice. The corneal picture picks up regrowth better.
• Against a nerve conduction test. Those read the big insulated fibres. Small-fibre disease is invisible to them, which is why so many people with burning feet are told their nerve test was normal.

The Phase 2b then did the thing that turns a stand-in measurement into a real one. It checked whether the eye moved with the rest of the body. Change in corneal nerve fibre area tracked change in newly sprouting GAP-43-tagged fibres in skin, ρ = 0.575, p = 0.025, and change in how far people could walk in six minutes, ρ = 0.645, p = 0.009. The eye, the skin and the legs moved together.

Culver stated the limit of that logic himself, and it is the honest caveat on the entire programme: "heaven forbid that you're measuring something that affects a surrogate endpoint but it doesn't have anything to do with the clinically meaningful endpoint, because then you might end up with a medication that is beneficial to something we measure but not beneficial to how you feel, function or survive."

He reported the durability problem just as plainly. Twenty-eight days of injections produced a measurable gain at day 28, and then: "By day 56 some of that goes back to the baseline... it looks like 28 days is probably not going to be enough to maintain a durable benefit."

Statement

Official disclosure

Webinar transcript: Small Fiber Neuropathy and ARA-290 Results - Dr. Daniel Culver, Cleveland Clinic, 17 May 2017 (Foundation for Sarcoidosis Research)

“By day 56 some of that goes back to the baseline. ... it looks like 28 days is probably not going to be enough to maintain a durable benefit.”

Aug 4, 2026 · stopsarcoidosis.org

2026-08-04 08:29s28 · #6a1f15ea0164

Where the trial record stops

If your problem is sciatica from a disc, you have a different injury from everyone in these trials. Here is exactly what carries across and what does not.

The row that carries across is the inflammation row. Root pain is not purely a squeezing problem: contact between disc material and a nerve root sets off an inflammatory cascade in the spinal cord, and the immune cells there are a large part of what keeps the pain going. That is exactly the target ARA-290 was shown to hit in a mechanical nerve injury. In rats whose nerve was cut in the spared-nerve-injury model, ARA-290 at 3–60 µg/kg on days 1, 3, 6, 8 and 10 reduced pain from light touch and from cold out to 20 weeks, and the animals given 30 µg/kg showed no rise in spinal immune cell activity at all.

Spared nerve injury is a surgical, mechanical cut — closer to a squeezed root than either trial population is. That is the strongest bridge available, and it is a rat.

The row that does not carry across is the squeezing row, and it was put as a question by someone reading the same evidence:

Nothing in the ARA-290 record touches ongoing pressure. The peptide does not widen the gap the nerve is passing through, does not shrink a bulge, and does not change a joint. Every trial population had an injury with no mechanical cause left to remove. If something is still pressing on your root, you have a driver this drug cannot reach.

Searching the trial registry for the spine returns nothing. ClinicalTrials.gov holds four studies under "cibinetide" and four under "ARA-290", and they are the same four: sarcoidosis, type 2 diabetes, depression, swelling at the back of the eye. No sciatica trial. No disc trial. No trial in a pinched nerve root of any kind.

The tested dose is about twelve vials a month, and almost nobody takes it

The number that governs everything practical about this compound is 4 mg a day.

It came out of a crossover study of how the body absorbs and clears it, run inside the Dahan 2013 trial. Blood levels above 1.3 ng/mL were treated as the working range, and the total exposure above that line was 65 ng/mL×min for 2 mg into a vein, 23 for 2 mg under the skin, 59 for 4 mg under the skin and 249 for 6 mg under the skin. Only the 6 mg dose was significantly different from the others. The 4 mg subcutaneous dose was picked because it reproduces the into-a-vein exposure that had already worked, in a form you can inject at home.

A separate run in healthy volunteers gives the shape of the curve. Four milligrams under the skin peaks at about 3 ng/mL in the blood — roughly 2.4 nmol/L — and is half gone in about 20 minutes. Injected into a vein, it is half gone in about 2 minutes.

A drug that clears in twenty minutes, injected once a day, looks like a contradiction. It is not. Binding the repair receptor starts a gene programme inside the cell, and that programme keeps running for days after the peptide itself is gone. Collino and colleagues titled their review of exactly this "flipping the molecular switch", and it is why effects in animals last weeks after five injections.

Then the arithmetic that decides whether any of this is within reach.

A 28-day course at 4 mg a day is 112 mg of peptide. Research-grade ARA-290 is sold in 5 mg and 10 mg vials. So one trial-equivalent month is eleven to twelve 10 mg vials, or twenty-three 5 mg vials.

Nobody buys that. The doses described in public reports run 250 to 1,000 mcg a day — a quarter of a milligram to one milligram, against the four milligrams every successful trial used. That is four to sixteen times less. Say it as bluntly as it deserves: the dose people take has never been tested in a person for anything, and it sits below the lowest arm of the only dose-ranging trial ever run, which itself missed at 1 mg.

Mixing a vial, in numbers a syringe can read

The trials injected 4 mg in 0.5 mL. That is 8 mg per mL, and it is worth copying because it fills exactly half an insulin syringe.

The arithmetic, one step at a time:

1. Take a 10 mg vial. 10 mg is 10,000 mcg.
2. Add 1.25 mL of bacteriostatic water, slowly, running it down the inside wall of the vial. Do not shake. Swirl until it is clear.
3. Concentration = 10,000 mcg ÷ 1.25 mL = 8,000 mcg/mL.
4. A U-100 insulin syringe holds 1 mL across 100 marks, so one mark is 0.01 mL.
5. Mcg per mark = 8,000 mcg/mL × 0.01 mL = 80 mcg per mark.
6. The 4 mg trial dose = 4,000 mcg ÷ 80 = 50 marks = 0.5 mL, the exact volume injected in the diabetes trial.

Other fills, same arithmetic:

For sub-milligram doses, use the weaker fills. At 8,000 mcg/mL a 250 mcg dose is three marks on the barrel, and three marks is not a measurement, it is a guess.

The rest of the regimen, as the trials actually ran it:

• Where it went. Under the skin. Diabetes-trial subjects injected their own front thigh, moving the spot each day. The sarcoidosis trial used upper leg or lower belly and reported no stinging and no irritation at the site.
• How often. Once a day. The only human schedule that was not once a day was the 2 mg into-a-vein pilot at three times weekly, and it was dropped because you cannot run a vein line at home.
• How long. 28 days in five of the six trials, 12 weeks in the sixth. There is no human data on any schedule longer than 12 weeks.
• How long until anything moved. Symptom scores separated from placebo by week 4 in the pilot. Corneal nerve fibre area separated at day 28. Both drifted back toward the starting point by day 56 once the injections stopped.
• What it was mixed in. The trial formulation was 20 mmol/L sodium phosphate buffer at pH 6.5 with 1% sucrose and 4% D-mannitol — a buffered, sugar-stabilised solution, not plain water.

Once water goes in you have four weeks

At 4 mg a day a mixed 10 mg vial lasts two and a half days, so the four-week clock never bites at trial dosing. At 250–500 mcg a day the clock is the thing that decides how the vial gets split, and most of the vial will expire before you use it.

One death, four serious events, and no change in the blood counts

The safety claim that matters here is narrow and specific: ARA-290 does not act on the blood-building docking point, so it should not raise red cell production. The trials tested that, and it held.

• Dahan 2013, n = 38. "No medically significant deviations were noted in the general blood chemistry or hematology assessments." No serious events during dosing or across 12 weeks of follow-up. No pain or irritation at the injection site. One person on ARA-290 had a moderate event: 14 kg of weight loss over several months. The placebo arm had three moderate events — diarrhoea, irritability, light-headedness.
• Brines 2015, n = 48. No meaningful drug-related change in red cells, platelets or white cells. Four serious events happened in the ARA-290 arm. Two were judged unlikely to be related. Two were judged possibly related: one subject on daily furosemide developed worsening borderline kidney failure and stopped at day 15, and one subject was hospitalised for poor blood supply to a leg two weeks after the last dose and then died of a heart attack, which the safety committee judged unrelated to treatment. Non-serious events ran 64 in the ARA-290 arm against 66 on placebo.
• Culver 2017 Phase 2b, n = 64. One person had a serious event that led to stopping the drug, judged "possibly related at all". No deaths. Culver's reading: "there's no clear-cut serious or even not very serious adverse effects that occur very frequently with the medication", qualified in the same breath by "we're analysing small numbers of patients here so we'll need a larger trial to really answer the question."
• Lois 2020, n = 9. "No serious adverse events/reactions or anti-cibinetide antibodies were seen" across 12 weeks, the longest human exposure on record.
• Heij 2012, n = 22. "No safety concerns were raised by clinical or laboratory assessments."

Whether the immune system reacts to it — meaning whether the body starts making antibodies against the injected peptide — was tested in the diabetes trial and in the eye trial. Neither found any. For an injected peptide that is a real question, and it has now been answered twice, at small numbers.

State the limits exactly. About 132 people have had the real drug. Nobody has taken it beyond 12 weeks under observation. The mechanism is switching on a survive-and-repair signal, which is a reason for caution if you have an active cancer, and no trial has looked at that. And the one death on record happened in a diabetic group averaging 63 years old, where a heart attack two weeks after the last injection is what the underlying disease produces anyway — which is why it was judged unrelated, and why 48 people can neither rule it in nor rule it out.

Two regulators gave it orphan status, then the company shut

Statement

Official disclosure

Araim Pharmaceuticals Given FDA Fast Track Designation to ARA 290 for the Treatment of Sarcoidosis-associated Small Fiber Neuropathy (28 October 2014)

“Fast Track Designation ... for new therapies that have the potential to treat serious conditions for which there is no approved treatment option.”

Aug 4, 2026 · prnewswire.com

2026-08-04 08:29s27 · #463c6264b1f5

Statement

Official disclosure

EU/3/16/1721 - EMA orphan designation for cibinetide for prevention of graft loss in pancreatic islet transplantation

“Orphan designation granted 29 August 2016 for the prevention of graft loss in pancreatic islet transplantation.”

Aug 4, 2026 · ema.europa.eu

2026-08-04 08:29s26 · #f517e31ae982

Orphan designation and Fast Track are not approvals. Culver told the patient audience exactly what was still missing: "a phase 2b trial does not equal approval of a medication you must have a phase three trial and sometimes two phase three trials in order for the FDA to approve a medication for commercial distribution." He added, in the same breath, "I don't know if a phase three Cibinetide trial will happen."

It did not. No Phase 3 was ever started, in any condition. Araim Pharmaceuticals, of Tarrytown, New York, stopped operating, and the four registry entries now read completed, terminated or status unknown. Cibinetide is approved nowhere, for anything.

That leaves no pharmaceutical supply at all. What circulates is research-grade material, and an eleven-residue peptide with a looped front end is not the easiest thing to make correctly. A third-party purity run and a mass-spectrometry identity check on the specific batch is the only evidence that a vial holds what the label says.

Statement

Official disclosure

Araim Pharmaceuticals: Cibinetide (ARA 290) Regenerates Small Nerve Fibers and Improves Neuropathic Clinical Symptoms in the Orphan Disease of Sarcoidosis (2017)

“cibinetide demonstrated significant nerve regrowth assessed by two different measures of nerve fiber regeneration, as well as reductions in pain and improvements in functional capacity.”

Aug 4, 2026 · prnewswire.com

2026-08-04 08:29s30 · #a98989a6f085

What the people taking it report, counted

Five public accounts of ARA-290 are catalogued on this page. Three are first-person reports from someone who took it. Two describe improvement. One is a person eight weeks into a course with no result yet. None report nothing happening. None report harm. The remaining two accounts are a sceptic asking a question and a reader quoting the trial numbers back.

Three is not a denominator. Say that plainly rather than dressing it up: for BPC-157 and KPV there are dozens of first-person reports and they can be counted into a rate. Here there are three, and a rate built on three people is noise. Everything below is labelled anecdotal and is here for one reason — it is the only record of what this compound does at doses and durations no trial ran.

Reported improvement, 2 of 3.

That account describes 500 mcg under the skin into the outer hip near the pain, about six hours of tiredness afterward, and then: "I went from 3 weeks of being unable to put on pants or get in the car without stabbing pain, to zero pain." The same writer settled on roughly 400 mcg a day across four months — a tenth of the trial dose, for four times the longest trial.

Under way, no result yet, 1 of 3.

Not a personal report — a question, and a reading of the trial.

The public accounts agree on one thing and are silent on another. They agree that burning, tingling and sensitivity in the feet and legs are what got better. They are silent on nerve root pain: nobody is reporting a resolved pinched root, and nobody is reporting any kind of controlled comparison.

One report belongs here because it came from inside a trial rather than off a forum. Culver, quoting a Phase 2b participant: "Hey I just went to the mall all afternoon and I haven't done that for many, many years. I'm able to do much more than I was ever able to do." That person's 6-minute walk distance was one of the numbers that moved with the corneal nerve count.

Where it sits next to the other compounds here

BPC-157 and TB-500 have animal evidence in tendon, ligament and muscle, and no controlled human trial in any of those tissues. ARA-290 is the mirror image: almost nothing preclinical in muscle or tendon, and the only randomised, placebo-controlled human trials in this whole group, every one of them aimed at nerve.

So the division of labour in a disc protocol is clean. The others are aimed at the tissue around the nerve. This one is aimed at the nerve. The framework is laid out on the disc-stack, herniated-disc and degenerative-disc-disease pages.

What is settled, what missed, and what nobody has measured

Cibinetide is not an approved drug in any country and has no pharmaceutical supply. Nothing here is a dosing or treatment recommendation.

The sibling objects for this page, each one inspectable on its own terms: