dihexa: the angiotensin IV-derived molecule with a retracted foundational paper

dihexa (PNB-0408) is a small angiotensin-IV-derived molecule developed in the Joseph Harding lab at Washington State University as a metabolically stabilized analog of angiotensin IV. it is widely sold as a research-chemical "nootropic" but has never been tested in humans, its foundational HGF/c-Met mechanism paper was retracted in 2025, and the closest in-class clinical compound (fosgonimeton) has failed both a Phase 2 and a Phase 2/3 Alzheimer's trial.

what is dihexa?

dihexa, also called PNB-0408, is N-hexanoic-Tyr-Ile-(6) aminohexanoic amide. it is a Tyr-Ile dipeptide core with two hexanoic appendages, not a six-amino-acid peptide despite what the "hexa" name suggests in marketing copy. its molecular weight is around 462 g/mol.

the parent molecule is angiotensin IV (Ang IV), the hexapeptide Val-Tyr-Ile-His-Pro-Phe, which arises from sequential cleavage of angiotensin II and has procognitive activity in rodents. the engineering problem the Harding lab was trying to solve was that native Ang IV has a very short plasma half-life because terminal exopeptidases destroy it within minutes. dihexa's hexanoyl and 6-aminohexanoyl caps protect the dipeptide core from those exopeptidases and were the rationale for an orally active analog. the rest of the molecule mimics the N-terminal Tyr-Ile-His-Pro-Phe of Ang IV in compressed form, retaining only the residues thought to be essential for cognitive activity [1].

dihexa is fully synthetic; no endogenous parent molecule exists. it is also crucially not the same molecule as fosgonimeton (ATH-1017), which is a structurally distinct subcutaneous prodrug developed by Athira Pharma downstream of the same Harding-lab program. vendor and forum copy frequently conflates the two; they are not the same and their clinical fates do not transfer.

how does it work?

the Harding-lab hypothesis is that dihexa potentiates HGF binding to the c-Met receptor tyrosine kinase, engaging PI3K/AKT and MAPK signaling and driving dendritic spine formation and synaptogenesis in cultured hippocampal neurons. this hypothesis must be taught with the caveat that the canonical paper supporting it (Benoist 2014) was retracted in April 2025 following a Washington State University image-integrity investigation.

the published mechanism story has three nodes. first, dihexa was reported to bind soluble hepatocyte growth factor (HGF) and stabilize HGF dimers, facilitating HGF binding to and activation of the c-Met receptor. second, activated c-Met initiates downstream PI3K/AKT, MAPK, and beta-catenin signaling, all of which are independently implicated in dendritic spine formation, synaptogenesis, and neuronal survival. third, the net cellular outcome reported in primary hippocampal neuron cultures is increased dendritic spine density and increased spine head size, with the behavioral correlate of reversal of scopolamine-induced spatial-memory deficits in rats.

the integrity context is unavoidable. the canonical mechanism-defining paper, Benoist and colleagues in J Pharmacol Exp Ther in 2014, was formally retracted in April 2025 following a Washington State University investigation into image manipulation by then-PhD student Leen Kawas [2]. a second core paper, McCoy and colleagues' 2013 dose-response and oral-activity characterization, carries a 2021 expression of concern from the same investigation. the cleanest non-retracted direct dihexa citation is Benoist 2011, which established hippocampal synaptogenesis and spatial-memory effects in Nle1-angiotensin IV analogs including dihexa [1]. Sun and colleagues in Brain Sciences in 2021 extended the cognitive-rescue picture to APP/PS1 Alzheimer's-model mice with PI3K-dependent signaling, but that is single-group work that has not been independently replicated [3].

what does the evidence show?

no human clinical trial of dihexa itself has been completed or registered. the strongest direct evidence is rodent and cell-culture preclinical work from one or two laboratories. the closest in-class clinical compound, fosgonimeton, has failed both a Phase 2 and a Phase 2/3 Alzheimer's trial, which is the most relevant negative signal for the broader mechanism family.

for dihexa itself, the available preclinical evidence covers spatial memory in scopolamine-impaired rats, dendritic spine density in primary hippocampal neuron cultures, and cognitive-rescue endpoints in APP/PS1 mice. the sample sizes are small, the work is dominated by one Harding-lab line for foundational findings and one Nanjing follow-up group for the APP/PS1 extension, and the most quantitative magnitudes ("near-tripling of spine counts," reported oral bioavailability figures) trace through papers under expression of concern. independent replication of dihexa-specific findings outside those two groups is sparse [3].

fosgonimeton is the closest clinical test of the broader HGF/c-Met-modulator family. Hua and colleagues' 2022 Phase 1 program reported safety and tolerability of fosgonimeton at single and multiple doses, including in Alzheimer's-disease subjects, with the primary adverse events being injection-site reactions [4]. the Phase 2 ACT-AD trial in combination with cholinesterase inhibitors then failed primary and secondary endpoints at 26 weeks, and the Phase 2/3 LIFT-AD trial (n=315) failed the primary endpoint (Global Statistical Test) and key secondary endpoints (ADAS-Cog11 and ADCS-ADL23) in September 2024 [5]. translational implication: the most thoroughly tested compound in dihexa's mechanism family has not delivered clinical benefit in Alzheimer's. fosgonimeton's Phase 1 safety profile does not extend to dihexa, and its Phase 2/3 efficacy failure does not falsify the broader mechanism but does weaken any claim that dihexa "treats Alzheimer's."

regulatory status

dihexa is not FDA approved for any indication. in April 2026 the FDA removed dihexa acetate from 503A Category 2 and scheduled it for Pharmacy Compounding Advisory Committee review before February 2027. removal from Category 2 is not approval; it is procedural. until the PCAC review concludes, dihexa is not a legally compoundable bulk substance under section 503A, and athletes should treat it as banned under WADA categories S0 and S2.

the current regulatory window is best described as limbo. dihexa was previously listed in FDA 503A Category 2, the category for substances raising significant safety concerns. the April 2026 removal from Category 2 reflects that the agency is now formally evaluating the substance, not that it has approved it for any use. a Pharmacy Compounding Advisory Committee review is scheduled before February 2027. if PCAC ultimately approves dihexa for compounding, the supply chain will move toward licensed 503A pharmacies with higher quality standards; if PCAC rejects it, the legal status will harden against compounded supply. either way, the current period is regulatory limbo and not a green light.

on the anti-doping side, dihexa is not listed by name on the WADA Prohibited List, but as a non-approved substance with growth-factor-pathway activity it would fall under category S0 (non-approved substances) at minimum and plausibly category S2 (peptide hormones, growth factors, related substances and mimetics). athletes in any WADA-tested sport should treat it as prohibited.

safety profile and the c-Met oncology concern

there is no controlled human safety dataset for dihexa. the dominant theoretical concern is oncology: c-Met is a proto-oncogene, and aberrant HGF/c-Met activation is a recognized driver of proliferation, invasion, and metastasis in multiple solid tumors. any compound that potentiates HGF/c-Met signaling systemically carries a mechanism-intrinsic oncology risk that has not been characterized for dihexa.

the c-Met oncology concern is not speculative. pharmaceutical development around c-Met has predominantly focused on c-Met inhibitors for oncology indications (capmatinib, tepotinib, savolitinib), precisely because c-Met activation is pro-tumor in epithelial cancers such as hepatocellular carcinoma, gastric, lung, renal, and head-and-neck. a chronic systemic HGF/c-Met-potentiating agent therefore raises a theoretical risk of accelerating occult malignancies or promoting metastasis. this risk has not been characterized in any controlled dihexa exposure study and is not addressed in vendor product information [6].

secondary concerns include a theoretical pro-fibrotic effect from sustained HGF activation, undefined cardiovascular and hepatic risk, and grey-market product-quality variability including identity, purity, residual solvent, and endotoxin contamination. long-term safety data simply do not exist for dihexa; the longest published controlled exposure in the mechanism family is the 26-week LIFT-AD subcutaneous fosgonimeton program, and that is a different molecule. for pregnancy and breastfeeding the default is contraindication on growth-factor-potentiation grounds; for any individual with active or recent cancer history the theoretical oncology risk argues strongly against use.

where it fits in the cognitive-peptide landscape

dihexa is one of several speculative cognitive-enhancement peptides sold through grey-market vendors with thin clinical evidence. its mechanism family (HGF/c-Met potentiation) is genuinely different from cholinesterase inhibitors, NMDA modulators, and anti-amyloid antibodies, but the closest clinical test (fosgonimeton) has failed, and the "10 million times more potent than BDNF" claim that dominates vendor copy is an in-vitro spinogenesis comparison from partially retracted literature, not a clinical efficacy comparison.

relative to standard-of-care cognitive-enhancement medicine, dihexa has zero controlled human efficacy data while the comparators (donepezil, rivastigmine, galantamine, memantine, lecanemab, donanemab) all have controlled efficacy and safety datasets, however imperfect. relative to other speculative cognitive peptides, dihexa is earlier in development than Semax and Selank, both of which have Russian regulatory approval and limited human pilot data, and earlier than cerebrolysin, which has decades of clinical use in Europe and Asia with mixed-quality controlled-trial data.

the most-quoted vendor claim, that dihexa is "10 million times more potent than BDNF," originates from cell-culture spinogenesis potency comparisons in the Harding-lab literature. the supporting papers are partially retracted (Benoist 2014) or under expression of concern (McCoy 2013), and the claim conflates in-vitro potency on one endpoint with clinical therapeutic efficacy that has not been established to within any order of magnitude. for adjacent peptide pathways with cleaner regulatory status, the GH axis story sits in tesamorelin and ipamorelin; the broader pathway tour is in our free peptides and your body module.