P-21 Is Not a Cerebrolysin Derivative, and Every Study I Found Is Rodent
Every number I could find attached to this compound came out of a mouse or a rat, fed in chow, put down a tube into the stomach, leaked from a pellet under the skin, or injected into the abdomen. I found no study, in any species, of the injection under the skin people buy.
What it is
P-21 is built on four amino acids, Asp-Gly-Gly-Leu, residues 148 to 151 of human ciliary neurotrophic factor, CNTF for short, a protein the body makes that helps nerve cells survive and grow. Onto the tail the lab hung an adamantane, a rigid ball of carbon from the Parkinson's drug amantadine, to make it greasier, help it into the brain and slow the enzymes that break peptides apart. I pulled the 200 residue sequence out of UniProt and counted: the four letters sit where the papers say. The shorthand every source prints, Ac-DGGLAG-NH2, reads as six amino acids and is not six: PubChem's systematic name caps the leucine with 3-carbamoyl-1-adamantylamine, so the trailing "AG" is the cage, not two residues.
What the trials found
None that I could find. In the United States registry the term P021 returned five records, every one a false match on an identifier string inside an unrelated study, and the sponsor name Phanes returned three oncology trials from a different company. The Alzheimer's Drug Discovery Foundation logged zero on both the registry and DrugBank in its May 2025 report.
So the record is rodent. Li and colleagues published the first in 2010: normal C57Bl6 mice aged 8 to 10 months, 25 nmol a day for 35 days from a pellet implanted under the skin that released continuously, after which object recognition and spatial memory improved against vehicle and new neuron counts rose (PMID 20600002). The dose and route come from ADDF's review, not the abstract.
The longest exposure is Baazaoui and Iqbal 2017: female triple transgenic Alzheimer's mice fed P-21 in chow at 60 nmol per gram of feed from 3 to 21 months of age, in three groups of 41 wild type, 35 vehicle fed transgenic and 32 P-21 fed transgenic, 108 animals in all. Against the vehicle fed group, plaque area at 18 months was down about 20 percent in the CA1 and about 40 percent in the subiculum, two parts of the hippocampus, where memories are laid down. Survival ran the same way: 87 percent of the P-21 fed mice were alive at week 71 against 41 percent of the vehicle fed ones, in ADDF's reading. (PMID 28655344, PMID 28387677)
Aged Fisher rats, 22 to 24 months old, got it by gavage, a tube into the stomach, at 500 nmol per kilogram a day for 88 days and declined less in learning and memory than vehicle treated aged rats (PMID 24702821), though ADDF reads it more carefully than the market: the error bars overlapped the vehicle treated animals and the treated rats never approached young ones.
The rest is mice fed it in the diet. Kazim and colleagues used more than three times the Alzheimer's dose, 200 nanomoles per gram, and fed it to pregnant Down syndrome model mice from day 8 of pregnancy to weaning, so the pups measured were dosed in the womb and through the milk (PMID 28368015). Wei and colleagues ran that same prenatal window at 60 nanomoles per gram in Alzheimer's model mice (PMID 32854771). Falangola 2026 imaged the brains of 79 female mice, 30 of them Alzheimer's model animals on the peptide fed 60 nanomoles per gram from 2.5 to 8 months, reported white matter measures moving toward normal, and states that no behavioural testing was done (PMID 41740658).
Whose evidence is it
The derivation chain is documented: the lab cut its eleven amino acid parent, Peptide 6, CNTF residues 146 to 156, to the four active residues and added the adamantylated glycine. The 2007 patent, US 8,796,214, states that derivation and names Cerebrolysin only as background.
Cerebrolysin is in the history, but as the reason the lab looked at CNTF at all. Chen and colleagues reported in 2007 finding CNTF and three other growth factor activities inside Cerebrolysin, CNTF the most active at growing new neurons in cultured rat hippocampal progenitor cells (PMID 16859812). That is a lineage of attention, not a chemical derivation: Cerebrolysin's record is about a mixture given into a vein.
The only study I found that put the two side by side was retracted in June 2026: Rockenstein and colleagues had injected 5 nmol a day into the abdominal cavity of amyloid precursor protein transgenic mice for one month, n of 8 per group, and reported both raising new neuron counts (PMID 21860085). Its senior author is Eliezer Masliah, whose papers appear on Alzforum's misconduct dossier list, and the notice carries no abstract, so I could not retrieve the reason (PMID 42015030). It tested Peptide 6 and Peptide 6A rather than P-21, so P-21's evidence base loses nothing.
I could not find a study of P-21 in living animals, as opposed to cells in a dish, without Khalid Iqbal on the author list. He invented it, and co-founded the company licensed to develop it. The only test I found by a group outside his laboratory largely failed: Mottolese and colleagues in Bologna gave P-21 to mice missing the CDKL5 gene, by mouth at 60 nmol per gram of diet for 70 days, n of 6 against 7 vehicle and 9 wild type, and into the abdominal cavity at 750 nmol per mouse a day for 30 days, n of 4 against 5 and 8. Learning, memory, social behaviour and most motor measures were unchanged, and BDNF, the growth signal the mechanism runs through, did not rise (PMID 39592934). Iqbal is a coauthor there too.
The declarations of interest do not agree. Baazaoui and Iqbal 2017 and Wei 2020 state that the authors have no competing interests; three other papers, the 2026 imaging one among them, declare his patents, and two of those name his position at Phanes Biotech, which holds the licence. The patent priority date is 2007, before all of them.
How it compares
Four names get used near each other. Only one is in the vial.
| What it actually is | Where its evidence comes from | |
|---|---|---|
| P-21 | Four residues of human CNTF, 148 to 151, plus an adamantane cage, 578.7 daltons | Rodents only: chow, a stomach tube, an implanted pellet, one course of injections into the abdomen |
| Peptide 6 | The eleven amino acid parent, CNTF 146 to 156 | Its own rodent studies. A traumatic brain injury result credited to P-21 is this molecule's: 50 nmol a day for 30 days in mice (PMID 25255260) |
| Peptide 6A | A different four residue fragment, CNTF 145 to 148 | The Cerebrolysin comparison, retracted in June 2026 |
| Cerebrolysin | A mixture made by enzyme digestion of animal brain protein | Its own separate literature, none of it about this molecule |
Where the dose came from
Not from anywhere I could trace. What circulates is 500 micrograms to 1 milligram once daily by injection under the skin, with a wider cited range of roughly 100 micrograms to 2 milligrams. That is vendor and community convention, reported and not recommended, and no trial I found produced it. One seller shows its working: the rat gavage figure of 289 micrograms per kilogram scales to 2,981 micrograms a day for a 65 kilogram person, and the mouse chow intake to about 25 milligrams, about six and about fifty times above the 500 micrograms its own market runs.
The route is the harder problem. A pellet leaking the compound for 35 days against one shot a day is a dripping tap against a jug emptied over the plant once a day: the same water goes in across a week, and I could find nobody who has published which of the two the plant lives on. I found no bioavailability figure published for any route in any species, so no arithmetic gets you from one to the other.
What could go wrong
ADDF's reviewers state that safety in humans has not been assessed, and I found no human safety data and no dedicated toxicology study in my searches. What exists is tolerability watched alongside efficacy experiments in rodents: 18 months of chow at 60 nmol per gram in those triple transgenic mice with no weight loss, no tumours and no signs of pain.
The theoretical risk the literature raises is antibody formation against the compound, noted by Lozupone and colleagues in a 2023 review of peptide and oligonucleotide tau drugs, which adds that no immune reaction to P-21 has been reported to date (PMID 37042028). I found no record of it being given to a person, so "to date" is doing all the work there.
The interesting fact runs the other way: the parent protein failed in people. In 1996 the ALS CNTF Treatment Study Group randomised 730 patients with motor neurone disease to that protein at 15 or 30 micrograms per kilogram under the skin three times a week for 9 months, or to a dummy injection. Muscle strength, the primary measure, declined no more slowly than on the dummy and fell faster early on in the treated patients, deaths were the same in all three arms, and side effects limited dosing in many (PMID 8628460). In mice fed 60 nmol per gram in chow, P-21 does the opposite on weight, raising it over six months without changing food intake.
What nobody knows
Whether it does anything in a human being. Nobody has looked, in any published record I could find.
How much reaches the brain. The lab's papers call it able to cross the blood brain barrier, the filter that keeps most of what is in the blood out of brain tissue: Khatoon 2015 says so directly (PMID 26401692) and Liu 2019 credits a commercial assay service (PMID 31803044). ADDF's reviewers state that no study has assayed the level reaching the brain in mice, and their summary line reads "not documented".
What is in the vial. A certificate has to confirm an adamantane cap that is about a third of the mass, not just four amino acids, and I found no independent testing of a sold product. How to read a COA.
My take
The derivation is the whole entry. P-21 is sold as a piece of Cerebrolysin, and that framing borrows a licensed medicine's forty years of clinical argument for a compound with no human record at all. The two are not the same kind of thing: one is an extract with a label, the other a defined molecule with an adamantane cap bolted on to get it into the brain, and the only published comparison between the two families was retracted in June 2026.
If you have bought P-21, I want to know what mass the certificate reported, and whether the word adamantane appeared on it anywhere.
Frequently asked
Is P-21 derived from Cerebrolysin?
It is not. P-21 is a four amino acid fragment of human ciliary neurotrophic factor, residues 148 to 151, with an adamantane cage attached, and Cerebrolysin is a mixture made by enzyme digestion of animal brain protein. The real connection is historical: Khalid Iqbal's laboratory reported finding ciliary neurotrophic factor activity inside Cerebrolysin in 2007, picked that factor as the most active of the four it found for growing new neurons, and went on to build P-21 from it. That is a lineage of attention, not a chemical derivation, and Cerebrolysin's clinical record does not transfer to this molecule.
Has P-21 ever been tested in humans?
Not in any published record I could find. My searches of the United States trials registry returned five records under the term P021 and all five were false matches inside unrelated studies, and the Alzheimer's Drug Discovery Foundation's reviewers logged zero trials and zero DrugBank entries in their May 2025 report, recording no human research available in either direction. Every efficacy and safety figure attached to P-21 comes from a mouse or a rat.
What dose did the animal studies use, and how does it relate to what people take?
The published animal figures are 25 nanomoles a day from a pellet implanted under the skin of mice for 35 days, 60 nanomoles per gram of feed in mouse chow, 200 nanomoles per gram of diet fed to pregnant mice from day 8 of pregnancy to weaning in a Down syndrome model, so the offspring were exposed in the womb and through the milk rather than eating it themselves, 500 nanomoles per kilogram a day by stomach tube in rats for 88 days, and 750 nanomoles per mouse a day injected into the abdominal cavity for 30 days. What circulates as a human figure, 500 micrograms to 1 milligram a day by injection under the skin, is vendor and community convention rather than anything a study I found produced, and it is reported here and not recommended. One seller's own scaling of the animal doses lands about six times and about fifty times above it.
Do people inject P-21 the way the studies gave it?
They do not. Every published result came from feeding it in chow, delivering it by a tube into the stomach, an implanted pellet that releases it continuously, or one course of daily injections into the abdominal cavity of mice. I could not find a single study in any species that gave P-21 by daily injection under the skin, which is the route it is sold for, and I found no bioavailability figure published for any route in any species that would let anyone convert between them.
Is P-21 safe?
I could find no answer to that. The Alzheimer's Drug Discovery Foundation's reviewers state that safety in humans has not been assessed, and I found no human safety data of any kind, no case series and no tolerability study, in my own searches. Rodents fed it for up to 18 months showed no weight loss, tumours or signs of pain alongside the efficacy experiments, but that is tolerability watched in passing rather than a toxicology study, and I could not find a dedicated one. The full length parent protein, given under the skin to 730 patients with motor neurone disease in 1996, caused side effects severe enough to limit dosing in many of them.
Is P-21 approved anywhere?
It is not approved for anything in any market I checked, and the Alzheimer's Drug Discovery Foundation records its availability as none, in clinical development. It is not on FDA's 503A bulks list, the set of raw substances a compounding pharmacy may lawfully work with, and it was not among the seven peptides the agency's advisory committee considered on 23 and 24 July 2026. I could not retrieve FDA's full list of nominated substances, so I can say P-21 does not appear on those agendas and that I found no nomination for it, not that none exists.
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 Cognitive and neuro
- Cerebrolysin Is a Licensed Medicine Made of Pig Brain, and Its Own Label Says Nobody Can Measure ItAustria has prescribed it since 1996 with a dosing table by condition. Ten cerebrolysin papers are flagged retracted on PubMed, and the amount people inject at home stops exactly where the label's syringe limits stop, 5 mL into a muscle and 10 mL into a vein.
- Dihexa: The Ten Million Times Figure Is From a Press Release, Not a PaperNeither founding paper mentions BDNF, three of the four were retracted in April 2025, and the same molecule went into a 554 person trial under another name and did not beat a dummy injection.
- PE-22-28: The Antidepressant Claim Is Ten Mice in a Water CylinderOne laboratory in every paper I found, male mice only, injected into the abdominal cavity. The 23 hour duration figure was measured on a different molecule, and the same group later reported the effect reverses direction with dose.