Dihexa: what the animal studies show, what was retracted and what is unknown

Dihexa is a chemically modified fragment of angiotensin IV, a six-amino-acid hormone made when the blood-pressure hormone angiotensin II is broken down. Angiotensin IV and related peptides had been studied for years as procognitive agents in rats, but they are digested in the gut and cannot cross the blood-brain barrier. A 2013 paper from Joseph Harding’s laboratory at Washington State University described how the group took the three amino acids at the start of an angiotensin IV analogue, Nle-Tyr-Ile, and modified them step by step to make them stable and fat-soluble. The end product, N-hexanoic-Tyr-Ile-(6)-aminohexanoic amide, was named Dihexa, and the authors reported that it was orally active, entered the brain, reversed scopolamine-induced memory deficits in rats, improved performance in aged rats and stimulated synapse formation ^[1].

Strictly, Dihexa is not a peptide in the usual sense. Its core is two amino acids, tyrosine and isoleucine, capped at one end by a six-carbon fatty acid chain (hexanoic acid) and at the other by 6-aminohexanoic amide, hence the name. Its molecular weight is about 505 g/mol, closer to a conventional small-molecule drug than to the peptides it is sold alongside. The development code PNB-0408 belongs to the company set up to commercialise it, M3 Biotechnology of Seattle, which several of the developers were affiliated with when they wrote their 2015 review ^[4].

The Dihexa peptide, as stores label it, is sold as a Dihexa nootropic, usually as a powder or in oil-based solutions because it is poorly soluble in water, and it is frequently compared with Semax and P21. Unlike Semax, it has no registered use in any country, and unlike P21 it has no published data from independent groups supporting its proposed mechanism.

The developers proposed that angiotensin IV analogues, including Dihexa, work by binding hepatocyte growth factor (HGF) and helping it activate its receptor, c-Met, which drives synapse formation. Two papers made that case: a 2012 report that angiotensin IV analogues act as HGF/Met modifiers ^[2], and a 2014 paper concluding that the procognitive and synaptogenic effects of angiotensin IV-derived peptides depend on activation of the HGF/c-Met system ^[3]. The identity of the receptor was contested even then; the group’s own 2015 review acknowledged that Albiston and colleagues had offered convincing evidence that the angiotensin IV receptor is the insulin-regulated aminopeptidase (IRAP), and framed the HGF/c-Met proposal as an alternative ^[4].

In April 2025 the Journal of Pharmacology and Experimental Therapeutics retracted both mechanism papers ^{[5] [6]}. The retraction notice for the 2014 paper states that figures and data in the erratum submitted for the article were found to contain falsified and/or fabricated data, and that an investigation by Washington State University determined that two of the authors, Leen Kawas and Joseph Harding, were solely responsible ^[5]. The 2013 paper introducing Dihexa was not retracted but carries an expression of concern published by the journal in 2021 ^[1]. Anyone reading vendor descriptions of Dihexa as an HGF/c-Met activator should know that the primary evidence for that mechanism has been withdrawn from the literature.

What survives is the older pharmacology of angiotensin IV itself. A 2018 systematic review of experimental studies concluded that angiotensin IV and angiotensin-(1-7) have cognitive benefits in animal models across a range of memory tasks ^[7]. That literature concerns the natural hormone given directly into the brain, not Dihexa given by mouth, and it predates and is independent of the retracted work.

Memory in rats: the 2013 paper reported that Dihexa reversed scopolamine-induced deficits in the Morris water maze, improved performance in aged rats and increased the number of synapses on cultured hippocampal neurons. It described Dihexa as orally active and blood-brain-barrier permeant ^[1]. This paper is the source of nearly every efficacy claim made for the compound, and it comes from the laboratory whose later papers were retracted.

Alzheimer’s-model mice: an independent Chinese group reported in 2021 that oral Dihexa at several doses restored spatial learning in APP/PS1 transgenic mice, increased neuron counts and synaptophysin, reduced activation of astrocytes and microglia, lowered the inflammatory cytokines IL-1β and TNF-α, and activated the PI3K/AKT pathway. The authors proposed the brain angiotensin IV/PI3K/AKT axis as the mechanism rather than HGF ^[8]. This is the only positive efficacy study from outside the developers’ network that we found.

Huntington’s-model rats: a 2024 study gave PNB-0408 to rats exposed to the mitochondrial toxin 3-nitropropionic acid, which produces Huntington’s-like motor and cognitive deficits. Dihexa did not protect the animals from any of the deficits, and the authors concluded it may not be an efficacious strategy in that model ^[9].

Hearing and nerve repair: in zebrafish larvae, Dihexa at 1 µM protected lateral-line hair cells from the toxicity of the antibiotics neomycin and gentamicin, an effect the authors attributed to HGF signalling ^[10]. In rats with a cut and repaired sciatic nerve, Dihexa at 2 to 4 mg/kg combined with mesenchymal stem cells injected into the calf muscle improved motor recovery at 8 to 16 weeks, though Dihexa alone was not singled out as effective ^[11].

Parkinson’s disease: the developers’ 2015 review proposed Dihexa for Parkinson’s as well as Alzheimer’s, on the basis of the HGF/c-Met hypothesis and unpublished motor data ^[4]. We found no peer-reviewed Parkinson’s efficacy study of Dihexa.

There are none. No human study of Dihexa has been published, no clinical trial is registered on ClinicalTrials.gov under Dihexa or PNB-0408, and the US FDA stated in its compounding review that it has not identified any human exposure data on drug products containing Dihexa acetate given by any route, and that it lacks information on whether the compound would cause harm if administered to humans ^[12].

So does Dihexa work? In rats, according to the original developers, and in one Alzheimer’s mouse study from an independent group, yes on memory tests. In a Huntington’s rat model, no. In humans, the question has never been asked in a published study.

In the animal studies, memory testing followed days of dosing: the scopolamine and aged-rat experiments in the 2013 paper measured water-maze performance over training days ^[1], the APP/PS1 mice were dosed for weeks before testing ^[8], and the Huntington’s-model rats were dosed alongside five weeks of toxin exposure ^[9]. Claims that Dihexa is felt within hours come from users of an unstudied compound. No pharmacokinetic study in people exists, so its half-life, time to peak and duration of action in humans are unknown.

These are animal doses. They are not recommendations, they cannot be scaled to a human dose, and the FDA has said there are no human exposure data at all ^[12]. The 2013 paper reported oral and injected dosing in rats, with the compound described as orally active; its abstract does not give a per-kilogram figure ^[1]. The 2021 APP/PS1 mouse study used oral Dihexa at more than one dose level ^[8]. The sciatic nerve study gave 2 to 4 mg/kg body weight in rats, locally, systemically and into muscle ^[11]. In zebrafish, 1 µM in the water was the concentration that protected hair cells ^[10].

Vendor charts quoting daily milligram amounts of Dihexa, with or without a loading phase, are not derived from any of these. The compound is fat-soluble, is sold in oil or as a powder rather than in vials for reconstitution, and is one of the few compounds in the peptide catalogue for which our peptide calculator is of limited use, because the usual vial-and-bacteriostatic-water arithmetic does not apply. The Dihexa calculator page exists for completeness.

Dihexa was designed to be taken by mouth and to reach the brain, and the animal studies used oral dosing as well as injection ^{[1] [8] [11]}. It is sold online as a powder, in capsules, dissolved in oils for oral use or for application to the skin, and occasionally in solutions for injection. None of these forms has been tested in people. Because it is poorly water-soluble, Dihexa is not a typical freeze-dried vial product; where PepFinder lists it, the listing is for the raw material and the purity claim, not for a formulated product.

There are no human safety data. The FDA’s compounding review is explicit that it has not identified any human exposure data for Dihexa acetate by any route and lacks information on whether it would cause harm ^[12]. The animal papers report no formal toxicology, and the Huntington’s-model study, which did not find benefit, did not report harm either ^[9].

The theoretical concern most often raised about Dihexa follows from its proposed mechanism. If Dihexa really does amplify HGF signalling through c-Met, as the developers claimed, that is a growth pathway that is switched on in many cancers, and c-Met inhibitors are established cancer drugs. The 2011 paper from the same laboratory that described HGF-mimicking compounds also described anti-Met, anticancer versions ^[13], which illustrates how closely the pathway is tied to tumour biology. Whether Dihexa affects cancer risk has never been studied in animals over a lifetime or in people, and now that the mechanism papers have been retracted ^{[5] [6]}, it is not even clear that the pathway is the right one to worry about. What can be said is that the compound is potent in rodent brain at low doses, is designed to persist in the body, and has no long-term safety record.

Dihexa is sold as an unregulated research chemical, often from the same catalogues as research peptides. Our guide to spotting a fake peptide supplier applies, with the added difficulty that identity testing of a small lipophilic molecule is not the same as peptide HPLC.

All three are sold as nootropics, and the comparison matters because their evidence differs so much. Semax is a registered nasal drug in Russia with small human studies. P21, a CNTF-derived peptide, has a large body of mouse and rat work from a New York institute but no human data. Dihexa has rat and mouse data, one of which is independent, plus retracted mechanism papers and no human data. On the strength of published evidence, Dihexa sits below both. None of the three has a controlled trial showing cognitive benefit in healthy people.

Dihexa is not an approved medicine anywhere. In the US, Dihexa acetate was nominated for the FDA’s list of bulk substances that may be used in compounding, was placed in the agency’s category 2 of substances that may present significant safety risks, and now appears on the FDA’s list of substances that were nominated but withdrawn by the nominators ^[12]. Withdrawal is not an authorisation: it means the compound has no pathway to legal compounding in the US, and the agency’s stated reason, the absence of any human exposure data, still stands. See our US legal status page.

We found no marketing authorisation or scheduling decision for Dihexa in the UK, Canada, Australia or the EU. Our legal status overview, including the UK and Canadian pages, explains how unapproved research chemicals are treated in each market.

Dihexa was engineered for metabolic stability ^[1] and is a lipophilic solid, so it is more robust than most peptides, but no published stability data exist for the powder or for solutions sold online. Keep it cool, dry and away from light. Our peptide storage guide covers general practice; note that the sections on bacteriostatic water and reconstitution do not apply to an oil-soluble compound.

Dihexa is a synthetic small molecule, so a certificate of analysis should show identity by mass spectrometry with a measured mass near 505 g/mol and purity by HPLC. Our guides to reading a peptide COA and third-party peptide testing explain what a credible certificate looks like and why a vendor-supplied one is not enough.

You can compare Dihexa prices, including Dihexa in the UK and Dihexa in the US, and see our list of third-party tested suppliers. A listing is not a statement that a product is legal to buy or appropriate to use.