Dihexa: Evidence Review, Retractions & Risks

Dihexa at a glance

What it is

N-hexanoic-Tyr-Ile-(6)-aminohexanoic amide (PNB-0408, CAS 1401708-83-5) — a small-molecule angiotensin IV analog, C27H44N4O5, molecular weight 504.7 as the free base. The acetate salt is the form named in FDA's compounding listings. Capped at both termini to resist aminopeptidases and cross the blood-brain barrier. Not a true peptide hormone, and not water soluble.

Researched for

Reversal of cognitive deficits in rodent models (scopolamine amnesia, aged rats, APP/PS1 Alzheimer's mice) and dendritic spinogenesis in hippocampal culture. Also used in vitro at 100 nM as a growth-factor-free HGF surrogate in hepatocyte differentiation protocols.

Reported (unvalidated) range

Community and grey-market reports cluster at 5-10 mg daily described as "low" and 10-20 mg daily described as "moderate", for 4-8 weeks. These figures are recorded here because they circulate, not because they are supported: there is no human dose-ranging study, no human pharmacokinetics, and no established safety ceiling behind any of them.

Route reported

Oral capsule or sublingual, or dissolved in DMSO for topical/transdermal use. Preclinical work used oral gavage, intraperitoneal, intravenous and intracerebroventricular routes — all in a DMSO vehicle. Dihexa cannot be reconstituted in bacteriostatic water.

Reported frequency

Once daily is the near-universal community pattern, often with 90-days-on / 30-days-off cycling. That pattern sits against a rodent terminal half-life measured in days, which implies substantial accumulation across a cycle.

Evidence status

Animal-only, and the animal record is compromised. Zero human trials of any phase. Kawas 2011, Kawas 2012 and Benoist 2014 (all JPET) were retracted in April 2025 for image falsification; McCoy 2013 — the source of nearly every dose, half-life and spine-density figure in circulation — carries an unresolved September 2021 Notice of Concern.

Key negative finding

Even inside the tainted literature, dihexa did not improve cognition in cognitively normal rats. The animal data is deficit-rescue, not enhancement in healthy subjects — which contradicts the healthy-brain nootropic use case driving essentially all consumer interest.

Regulatory status

Not approved by the FDA for any indication, and not approved as a drug in any major regulated market. No IND, no registered clinical trial, no recognised US dietary-supplement pathway, and no listing on the DEA controlled substances schedules. Dihexa acetate was removed from FDA's 503A Category 2 list effective on or about April 22, 2026 solely because the nominations were withdrawn — an administrative action, not a safety finding, and not compounding authorization. FDA has indicated the substance is among those scheduled for Pharmacy Compounding Advisory Committee review by February 2027.

WADA status

Prohibited at all times for athletes subject to testing, although the section that captures it is unsettled. Dihexa is not named on the WADA 2026 Prohibited List and WADA has published no determination for it. It falls either under S2.3 (Growth Factors and Growth Factor Modulators), which names hepatocyte growth factor explicitly and extends by non-exhaustive language to growth factor modulators, or — if S2.3 does not reach it — under S0 (Non-Approved Substances), the residual class for pharmacological substances not addressed by any later section of the List and not approved by any governmental regulatory health authority for human therapeutic use. The two are mutually exclusive. Which one applies changes the sanction range rather than the prohibition: WADA classifies S2 substances as non-Specified and S0 substances as Specified. WADA gives BPC-157 as an S0 example despite its regenerative profile, so S0 is at least as plausible for a small-molecule peptidomimetic that is not itself a growth factor.

Reported ranges from research/community — examples, not recommendations.

What it is / mechanism

The proposed mechanism of dihexa has nothing to do with angiotensin receptors, despite its parentage. Dihexa is built on the angiotensin IV (Ang IV) backbone, but the published claim — developed at Washington State University by Joseph Harding and John Wright — is that it acts on the hepatocyte growth factor (HGF) / c-Met (MET) receptor tyrosine kinase system. The reported model is that dihexa binds HGF itself with extremely high affinity (a figure of Kd approximately 65 pM is quoted everywhere online), modulates HGF dimerization, and thereby potentiates c-Met autophosphorylation at HGF concentrations that would otherwise be subthreshold. In that framing dihexa is not a receptor agonist at all: it is a positive allosteric-style modulator of a growth factor, and it does nothing where HGF is absent. Downstream of MET activation the signaling is well characterized and is not in dispute as general biology: MET phosphorylation recruits GAB1 and GRB2 and drives PI3K/AKT and MAPK/ERK cascades. In CNS tissue those cascades are linked to dendritic spinogenesis, synaptogenesis, neurite outgrowth, and neuroprotection against excitotoxic and ischemic insult. HGF and MET are both upregulated in brain after injury, ischemia and in neurodegenerative states, which is the biological rationale for describing dihexa as a repair-state amplifier rather than a general-purpose enhancer — it is proposed to raise the gain on a signal the tissue is already producing, not to create one. That account is the single most-repeated block of text on the peptide web, and its primary experimental basis no longer stands. The 65 pM binding figure comes from Kawas et al. 2012 (J Pharmacol Exp Ther), and the demonstration that dihexa's procognitive effect is HGF/c-Met-dependent — the intracerebroventricular HGF-antagonist blocking experiment — comes from Benoist et al. 2014 (JPET). Both papers were formally retracted in April 2025. A third paper from the same lab, Kawas et al. 2011, was retracted at the same time; that one reported that closely related Ang IV analogs mimicking the HGF dimerization domain were anti-Met, anticancer agents. So the same chemical family was published by the same group as both a MET potentiator and a MET inhibitor, and both directions have now been withdrawn from the record. The honest statement is that the published literature does not reliably establish whether dihexa potentiates MET, antagonizes it, or engages it at all, nor by what molecular route. What survives is thinner and more indirect, but it is real and it is independent. Siller et al. 2015 (Stem Cell Reports) and Mathapati et al. 2016 use 100 nM dihexa as a growth-factor-free substitute for HGF to mature human pluripotent stem cell-derived hepatoblasts into hepatocyte-like cells — an independent laboratory result that is difficult to explain unless dihexa does something HGF-like at the cellular level. Uribe et al. 2015 (Frontiers in Cellular Neuroscience) showed an HGF mimetic of this family protecting zebrafish lateral line hair cells from aminoglycoside damage. And Sun et al. 2021 (Brain Sciences, China Pharmaceutical University) — the strongest independent in vivo result — attributed oral dihexa's effects in APP/PS1 mice to PI3K/AKT engagement and raised endogenous Ang IV levels, and did not demonstrate a discrete HGF/MET step. Taken together: something pharmacologically active is happening, the HGF/MET story is plausible and partly corroborated in vitro, and the specific quantitative claims that made dihexa famous are not currently supported by any standing publication.

Researched effects

The effects attributed to dihexa in the animal literature are specific and, on their face, striking. In the foundational rodent work, dihexa reversed scopolamine-induced deficits in the Morris water maze at the top dose of every route tested, improved performance in 24-month-old aged rats, and produced roughly a threefold increase in dendritic spine density in hippocampal culture (41 versus 15 spines per 50 micrometres), with wider spine heads (0.80 versus 0.67 micrometres) and confirmation via AMPA-receptor miniature EPSCs that the new spines were functionally wired rather than morphological artifacts. Independent work in APP/PS1 transgenic mice — a standard Alzheimer's model — reported restored water-maze performance, higher synaptophysin and neuronal counts, reduced astrocyte and microglial activation, lower IL-1-beta and TNF-alpha, and higher IL-10. The finding that almost never makes it into vendor copy is the one that matters most for the people actually buying dihexa: it did not improve cognition in cognitively normal rats. In the same experiments, the HGF antagonist alone had no effect on normal learning either. The pathway behaves as injury-responsive, not enhancement-responsive. Every rodent result above is deficit rescue in an animal that had been chemically amnesic, aged, or transgenically loaded with amyloid. There is no animal dataset showing dihexa makes a healthy brain work better — and that is the use case generating essentially one hundred percent of search demand and grey-market sales. The related viral claim, that dihexa is "seven orders of magnitude more potent than BDNF," originated in a 2012 WSU press release and not in any publication; the two compounds were never directly compared in the underlying paper, as the Alzheimer's Drug Discovery Foundation's independent Cognitive Vitality review has pointed out. In humans, the reported effect profile is anecdotal only and thoroughly confounded. The most consistently described subjective effect across community reports is unusually vivid dreams. Beyond that, reports are scattered and non-specific: some describe improved verbal fluency or word retrieval, some describe nothing, and a meaningful fraction describe headache, overstimulation, anxiety or irritability at higher doses. These accounts are unblinded, self-selected, dose-uncertain, and frequently confounded by the DMSO vehicle — DMSO has its own systemic and subjective effects and carries whatever else is on the skin into circulation. There is no controlled human efficacy data of any kind, and the only clinical test of the underlying mechanism failed: Athira Pharma's fosgonimeton (ATH-1017), a structurally distinct HGF/MET positive modulator from the same scientific lineage, missed both its primary and key secondary endpoints in the LIFT-AD Phase 2/3 trial reported in September 2024, in a primary analysis population of 312 mild-to-moderate Alzheimer's patients not taking acetylcholinesterase inhibitors (GST change -0.08, p = 0.70, over a 26-week treatment period).

Dosage — reported ranges (overview)

There is no established dihexa dose in humans, and no honest way to derive one. This section exists to describe what is reported and to show why none of it constitutes a protocol — it is not a dosing guide and nothing in it should be read as a recommendation. The preclinical numbers are the only ones with any experimental basis: in rats, oral gavage 1.25-2.0 mg/kg/day (2.0 mg/kg was the effective dose), intraperitoneal 0.05-0.50 mg/kg/day (0.5 mg/kg effective), and intracerebroventricular 0.1-1.0 nmol, with intravenous dosing used principally in the pharmacokinetic arm from which the terminal half-life figures come. In that work dihexa was dissolved in DMSO at 1 mg/mL and serially diluted, with rats receiving roughly 200 microlitres of a vehicle that was up to 75 percent DMSO — approximately 150 microlitres of neat DMSO per dose, which against a total blood volume on the order of 25 to 30 millilitres for a rat of that size works out to roughly half a percent of blood volume (about 0.5 to 0.6 percent). All of those figures come from the paper currently under a Notice of Concern, so they are provisional rather than settled. The figures circulating in community and vendor settings have no relationship to that data. What is commonly reported is 5-10 mg once daily as a "low dose" and 10-20 mg once daily as a "moderate dose", typically run for four to eight weeks, with widely repeated claims of diminishing returns above 20 mg and cycling schemes such as 90 days on and 30 days off. A flat 10-20 mg adult dose is not an allometric scaling of the 2 mg/kg rat oral dose, is not derived from any pharmacokinetic model, and is not anchored to any measured human exposure. It appears to be a convenience number that propagated through forums and vendor pages and then hardened into consensus. There is no human dose-ranging study, no human pharmacokinetics, no NOAEL, and no established ceiling — the absorbed fraction of an oral dose in a human is unknown, because oral bioavailability has never been measured in one. One structural point deserves particular attention and is almost universally ignored in these discussions. The rodent terminal half-life is measured in days, not hours. Once-daily dosing of a compound with a multi-day terminal half-life and a very large volume of distribution means exposure accumulates continuously across a multi-week cycle, and the steady-state concentration reached in week six of a "90 days on" schedule is unknown and considerably higher than the concentration after the first dose. The same property means that if an adverse effect emerges, stopping does not quickly reverse it. Reported routes are oral capsule, sublingual, or dissolution in DMSO for topical/transdermal application; anecdotal reports are further confounded because DMSO itself is pharmacologically active and transports co-present contaminants through skin into systemic circulation. None of the above constitutes a protocol, a recommendation, or an endorsement — it is a description of what is reported, published so that the weakness of the foundation is visible.

The full step-by-step protocol examples, titration schedule, and a printable protocol sheet with a dosing and injection log are included in the paid Protocol Playbook for this guide.

Reconstitution — bac-water math

Dihexa is the exception in this library: it cannot be reconstituted in bacteriostatic water. It is not water soluble, and it does not go into solution in bacteriostatic water regardless of vial size, swirling or warming — what results is a cloudy suspension or undissolved powder rather than a usable solution, and there is no valid concentration arithmetic to perform. Medibact's bacteriostatic water has no application to dihexa, and this guide does not present one. Vendors advertising a "dihexa injection" are selling a DMSO-solubilised preparation; injecting DMSO solutions is tissue-irritant, requires heavy dilution, and represents a meaningful additional hazard rather than a normal subcutaneous peptide route. The table below therefore describes laboratory stock-solution arithmetic in DMSO — the solvent the published in vitro and preclinical work actually used. It is not a dose-preparation aid, and no column in it corresponds to a human dose. Supplier-reported DMSO solubility for dihexa spans roughly 20 mg/mL to 100 mg/mL depending on vendor and on whether sonication and warming are applied, so 50 mg/mL is best treated as a central estimate rather than a fixed ceiling. Volumes are expressed per 100 mg of powder. The two right-hand columns give each stock in molar terms (molecular weight 504.7, so 1 mg/mL is approximately 1.98 mM) and the fold dilution required to reach 100 nM — the working concentration used in the independent, non-retracted hepatocyte-differentiation work of Siller et al. 2015 and Mathapati et al. 2016. At those dilutions the residual DMSO in culture medium is on the order of 0.0001 percent for the 50 mg/mL stock and 0.001 percent for the 5 mg/mL stock, in both cases far below the roughly 0.1 percent commonly treated as a cytotoxicity threshold, which is why such stocks are stepped down by serial dilution rather than in a single transfer.

This is concentration math, not a dose recommendation.

Injection / administration basics

Dihexa is not a subcutaneous reconstituted peptide and is not handled as one. There is no insulin-syringe workflow here, no bacteriostatic water, and no reconstitution-calculator entry — the standard peptide administration model simply does not apply. In the preclinical literature the routes were oral gavage, intraperitoneal, intravenous and intracerebroventricular, all in a DMSO vehicle at DMSO concentrations (up to 75 percent in the dosing solution) that would be unacceptable for repeated human parenteral use. In community settings the reported routes are oral capsule, sublingual, or a DMSO stock applied topically to skin for transdermal absorption. This section describes those routes; it does not provide instructions for any of them. The DMSO dimension carries its own distinct hazards that are separate from anything dihexa does. DMSO is a powerful penetration enhancer: it crosses skin readily and carries dissolved solutes and surface contaminants with it, which means anything present at an application site — residual soap, lotion, environmental contamination, or impurities in a grey-market powder — is transported systemically alongside the intended compound. At high concentration DMSO is a tissue irritant, produces a characteristic garlic-like taste and body odour within minutes of any exposure route, and is not appropriate for direct injection without substantial dilution. Because grey-market dihexa is sold as a research-use-only powder with no pharmacopoeial identity, purity or endotoxin testing, the composition of what is actually being carried across the skin barrier is unverified. Finally, the accumulation problem intersects directly with administration. Because rodent terminal half-life is measured in days rather than hours, any route that delivers a daily dose produces rising exposure over weeks. There is no human pharmacokinetic data to model this against, no way to estimate steady state, and no rapid washout available if something goes wrong. Medibact does not sell dihexa, does not supply solvents for it, and does not provide administration guidance for it beyond this factual description.

Half-life & frequency rationale

In rats, dihexa's terminal half-life was reported as approximately 12.68 days after intravenous administration and approximately 8.83 days after intraperitoneal administration — extraordinarily long for a molecule of 505 Da, and attributed to a very large volume of distribution with slow release from tissue rather than to slow metabolism per se. Supporting in vitro data are consistent with slow clearance: rat serum metabolic half-life of roughly 335 minutes (about 5.6 hours), and very low Phase I metabolism in rat liver microsomes (mean intrinsic clearance around 2.72 microlitres/min/mg, microsomal half-life about 509 minutes). Brain-to-plasma ratio was reported as greater than 1 in rodents, consistent with the design intent of the N- and C-terminal caps. All of these figures come from McCoy et al. 2013, the paper currently under an unresolved Notice of Concern, and are therefore provisional. The human half-life of dihexa is completely unknown. No human pharmacokinetic study has ever been published, so there is no measured Cmax, Tmax, AUC, clearance, volume of distribution or elimination half-life in any person. Oral bioavailability in humans has likewise never been measured; the "high oral bioavailability" claim that circulates online is a physicochemical prediction — a predicted human jejunal permeability (Peff) of about 1.78 × 10⁻⁴ cm/s, bracketed by the reported values for enalapril and piroxicam — not an observed value. The multi-day rodent half-life is not a trivia item; it is a safety-relevant property. If it translates even partially to humans, three consequences follow. First, once-daily dosing accumulates: exposure in week four of a cycle is materially higher than after a single dose, and the resulting steady state has never been characterized. Second, there is no meaningful way to titrate by subjective effect on a daily timescale, because the effect of the previous day's dose has not cleared. Third, and most important, an adverse effect cannot be quickly reversed by stopping. For a compound whose proposed mechanism runs through a proto-oncogenic growth pathway, slow washout is the opposite of a desirable property.

Stacking — overview

Medibact's position on dihexa stacking is that there is nothing here to build on. Stacking discussions assume a known dose-response, a known duration of action, and a characterized safety profile for the base compound; dihexa has none of the three. Adding a second compound to an agent with zero human pharmacokinetics, a multi-day rodent half-life, and an uncharacterized adverse event profile does not produce a synergy — it produces an experiment with two uncontrolled variables and no way to attribute either an effect or a problem to its source. The combinations below are documented because they are what appears in community discussion, and because readers searching for them deserve to encounter the caveats rather than a vendor stack chart. They are not recommendations. Two interaction categories are worth flagging specifically. First, dihexa is the only HGF/MET compound anywhere in this library, so there is no shared-pathway precedent to reason from — the cognitive-cluster compounds it gets paired with (Semax and Selank and their amidated variants, DSIP, PE-22-28, Epithalon, Cartalax) are associated with BDNF/NGF expression modulation, GABAergic and serotonergic pathways, sleep architecture, or telomerase and pineal signaling, and none of them touches MET. Second, and more seriously, any compound with its own proliferative or angiogenic signature — growth hormone secretagogues, IGF-1 LR3, MGF, or tissue-repair agents — layers an additional growth signal on top of a pathway whose oncogenic potential is already the main open question. That is the combination category with the least justification and the most theoretical downside. On formulation: dihexa cannot be co-reconstituted with anything, because it does not dissolve in bacteriostatic water at all. Any "stack" involving dihexa is necessarily two separate preparations by two separate routes — a DMSO stock or oral capsule alongside whatever the second compound requires.

For how combinations are grouped by research context, the named blends, and why a pre-mixed blend vial cannot be calculated from its total milligrams, see the peptide stacks guide.

Storage & handling

• Dihexa is supplied as a lyophilized or crystalline research-use-only powder. Cold, dark, desiccated storage is what supplier documentation describes: refrigeration at 2-8 °C for short-term holding and -20 °C or below for long-term, protected from light and moisture. The dry powder is markedly more stable than the compound in any solution.
• Bacteriostatic water is not a storage medium for dihexa. Because the compound is not water soluble, adding bacteriostatic water yields an undissolved suspension rather than a stable stock, and the resulting preparation has no defined concentration. For context, Medibact's USP-grade bacteriostatic water carries a 28-day in-use period after first puncture per USP and manufacturer labeling — but that period has no bearing here, because dihexa never enters solution in it.
• DMSO stocks are handled differently. DMSO is strongly hygroscopic and draws atmospheric water into the vial on every opening, progressively changing the effective concentration; standard laboratory practice is tightly sealed vials, minimal open time, and single-use aliquots. Pure DMSO also freezes at approximately 18.5 °C, so a refrigerated or cool-room stock will be solid, and it is warmed to room temperature and mixed to homogeneity before any volume is measured.
• Glass or DMSO-compatible polypropylene is the conventional container choice for DMSO stocks. DMSO dissolves or leaches many common plastics and elastomers, which can introduce plasticizers and extractables into the stock — a real concern given DMSO's capacity to carry whatever it contains through skin.
• Research labelling convention is that every vial is marked research use only, with compound, solvent, concentration and date, and kept segregated from anything intended for administration. Because dihexa is sold exclusively through grey-market research-use-only channels, there is no pharmacopoeial identity, purity, potency or endotoxin verification behind the label on the vial as received.

Guide FAQ

Quick answers about guide scope, access, and educational use context.