Dihexa for Cognitive Enhancement

Evidence Review created on 08/10/2026 using AI4L / Opus 5

Also known as: N-hexanoic-Tyr-Ile-(6) aminohexanoic amide, PNB-0408, Dihexa acetate, Hexanoyl-Tyr-Ile-(6)-aminohexanoic amide

Motivation

Dihexa is a laboratory-made compound built from a small fragment of a hormone the body uses to regulate blood pressure. It was designed to pass easily into the brain and to strengthen the signal of a natural growth factor that helps nerve cells form new connections with one another. Interest in it rests on a striking laboratory claim: that it drives the formation of new connections between brain cells far more powerfully than the body’s own growth factors do.

The compound came out of a university laboratory in the United States in the early 2010s and was licensed to a company that later carried a related drug into human testing. That work ran into serious trouble on two fronts. Several of the original laboratory papers were withdrawn by the journal that published them, and the related drug did not improve thinking or daily function in a large study. Dihexa itself has never been given to a person in a registered study, yet it is widely sold as a research chemical.

This review examines what the surviving evidence shows about dihexa, where that evidence comes from, and what remains unknown about its effects and its safety in people.

Benefits - Risks - Protocol - Conclusion

Grokipedia

Dihexa

A dense single-page overview covering chemistry, reported pharmacology, preclinical findings, and the commercial history. It is useful as an orientation map, though several of its claims — the identity of the individual trial registry numbers, and the description of fosgonimeton as a direct prodrug of dihexa — do not survive checking against the trial registry and the trial publications.

Examine

No Examine article exists for dihexa. Examine.com covers dietary supplements, nutrients, and food-derived compounds; dihexa is an unapproved investigational compound sold as a research chemical and falls outside that scope.

ConsumerLab

No ConsumerLab article exists for dihexa. ConsumerLab tests marketed dietary supplements and consumer health products; it does not cover unapproved investigational compounds, and no dihexa product exists in a regulated retail supplement channel for it to test.

Systematic Reviews

One systematic review addresses dihexa, and it does so as part of a broader survey of the peptide family dihexa was derived from.

Cognitive benefits of angiotensin IV and angiotensin-(1-7): A systematic review of experimental studies - Ho & Nation, 2018

A structured review of 32 animal studies of angiotensin IV, angiotensin-(1-7) and their analogues, which explicitly includes dihexa among the analogues assessed. Its central finding for this review is that eight of nine studies in animals with induced cognitive deficits showed benefit, while the picture in normal animals was far weaker — seven of eleven — and that effects were most reliable when the peptides were delivered directly into the brain rather than systemically.

Mechanism of Action

Dihexa’s proposed mechanism has three layers, and the confidence attached to each is very different.

Origin in the brain renin-angiotensin system. Dihexa is a chemically hardened fragment of angiotensin IV, a six-amino-acid peptide produced when the body’s blood-pressure-regulating cascade (the renin-angiotensin system) is broken down further inside the brain. Angiotensin IV had been known since the 1990s to improve memory in rodents, but it is destroyed within minutes in blood and barely crosses the blood-brain barrier — the tight cell layer that controls which molecules can enter brain tissue. Dihexa was built by stripping angiotensin IV down to a three-amino-acid core, capping one end with a fatty (hexanoyl) chain and the other with an amide, which made it resistant to the enzymes that chop peptides apart, more fat-soluble, orally absorbable, and brain-penetrant (McCoy et al., 2013).

The hepatocyte growth factor / MET hypothesis. The originating laboratory at Washington State University, whose senior authors went on to found the company that commercialised the compound and therefore had a direct financial interest in its adoption, proposed that dihexa works by binding hepatocyte growth factor (a natural protein that instructs cells to survive, move, and grow) with very high affinity and helping it activate its receptor, MET (also written c-Met, a receptor on the cell surface that triggers growth and survival signalling when switched on). On this model dihexa is not a growth factor itself but an amplifier: it lets small amounts of the body’s own hepatocyte growth factor switch MET on. Activated MET then drives two intracellular cascades — PI3K/Akt (a pathway that suppresses cell death and promotes growth) and MAPK/ERK (a pathway that relays growth signals to the nucleus) — which in turn increase the number of dendritic spines, the tiny protrusions where nerve cells receive connections from one another.

Why that layer is now unstable. The direct binding measurements and the receptor-activation experiments that underpin this model were published in the Journal of Pharmacology and Experimental Therapeutics between 2011 and 2014. Three of those papers were retracted in April 2025 after an institutional investigation found that images had been altered, and the fourth — the paper that first described dihexa — remains under a 2021 notice of concern. The behavioural rodent data have not been retracted, and the growth-factor amplification model has not been shown to be false; what has happened is that the specific experimental evidence for it has been withdrawn and has not been independently regenerated.

Competing mechanistic explanations. Two serious alternatives exist and both remain live. The first, advanced principally by Albiston, Chai and colleagues in Australia, holds that the angiotensin IV binding site in brain is insulin-regulated aminopeptidase (IRAP, an enzyme that degrades memory-relevant peptides such as vasopressin and oxytocin), and that angiotensin IV analogues improve memory by inhibiting it and letting those peptides persist. On this account dihexa’s growth-factor involvement is incidental. The second comes from the independent replication in Alzheimer’s-model mice, which found that dihexa raised tissue angiotensin IV levels and that its anti-inflammatory and anti-cell-death effects were abolished by a PI3K blocker, implying that the PI3K/Akt pathway is where the action is, whether or not hepatocyte growth factor is the entry point (Sun et al., 2021). A third, more sceptical reading is that a lipophilic peptide amide with a long tissue residence may act through several weak, non-specific routes at once.

Key pharmacological properties. Dihexa’s molecular weight is roughly 500 daltons, at the upper end of what passes a membrane easily but far below a protein. Selectivity: no direct agonist activity at classical angiotensin receptors has been reported, and no broad off-target receptor screen has been published, so selectivity is asserted rather than demonstrated. Half-life: the compound was engineered specifically for metabolic stability, and rodent and patent-derived data describe an in vitro serum half-life of several hours and a terminal elimination half-life of roughly 12 to 13 days after intravenous dosing in rats — a figure that comes from the developing parties and has never been independently confirmed, and which has no human counterpart. Tissue distribution: rodent work describes a very large volume of distribution with brain concentrations at or above plasma concentrations, consistent with a lipophilic molecule that partitions into tissue. Metabolism: clearance is described as slow, with low turnover in rat liver microsomes and a combination of hepatic metabolism and renal excretion; no study has identified which cytochrome P450 enzymes (the liver’s main drug-metabolising enzyme family, of which CYP3A4 handles roughly half of prescription drugs) are involved, and no human metabolism data of any kind exist.

Historical Context & Evolution

Original intended use: Dihexa was never designed as a cognitive enhancer for healthy adults. It was designed as a candidate treatment for Alzheimer’s disease, specifically to rebuild synaptic connections that neurodegeneration destroys. The entire preclinical programme used models of damage — chemically blocked memory, aged rats, amyloid-overexpressing mice, dopamine-lesioned rats, nerve injury.
The angiotensin IV thread (1990s to 2011): Harding and Wright’s laboratory at Washington State University characterised a distinct brain binding site for angiotensin IV and showed that the peptide improved rodent memory. The obstacles were pharmacological, not conceptual: rapid degradation and poor brain entry. Successive truncations established that the memory-relevant information sat in the first three amino acids of the modified peptide.
Dihexa emerges (2012 to 2013): Chemical capping of that tripeptide core produced dihexa. The reported findings were that oral dosing at 2 mg/kg reversed memory blockade caused by scopolamine (a drug that blocks acetylcholine signalling and reliably impairs memory) and restored maze performance in aged rats, alongside marked increases in hippocampal synapse formation. These are the actual findings, and they have not been withdrawn.
Commercialisation (2011 to 2020): M3 Biotechnology was spun out of the university in 2011, renamed Athira Pharma in 2019, and went public in 2020. Dihexa itself was set aside in favour of fosgonimeton, a separate subcutaneously injected molecule acting on the same hepatocyte growth factor / MET system.
Data integrity (2016 to 2025): Image concerns were raised as early as 2016. In 2021 the company’s chief executive and co-author was placed on leave and then resigned; an internal investigation concluded that images had been altered in her doctoral dissertation and in at least four co-authored papers, and the university subsequently revoked the doctorate. Four papers received notices of concern in September 2021. In January 2025 the company paid a federal settlement of $4,068,698 over allegations it had failed to report the misconduct in connection with grant applications (Department of Justice, 2025). Three of the four flagged papers were retracted in April 2025.
What this does and does not settle: It is worth being precise rather than reaching for a verdict. The retraction notices concern image integrity in specific figures; they are not a demonstration that dihexa lacks activity, and the behavioural experiments — which used automated maze tracking rather than images — were not among the retracted material. Equally, the mechanistic case that made dihexa interesting has lost its evidentiary foundation and no group has rebuilt it. The independent replication in Alzheimer’s-model mice found real effects but attributed them to a different pathway; the independent Huntington’s-model study found nothing at all. The honest current position is that dihexa is an active compound in some rodent damage models, with a mechanism that is now unresolved rather than established.
Migration into the enhancement market (roughly 2015 onward): As the preclinical claims circulated, dihexa appeared on research-chemical vendor sites and in enhancement communities, where it is used by healthy adults for a purpose it was never tested for. Notably, the rodent literature indicates little or no benefit in cognitively normal animals — the effects cluster in damaged systems.

Expected Benefits

High 🟩 🟩 🟩

No benefit of dihexa reaches this evidence level. High-level grading requires human clinical data, and no human study of dihexa has ever been registered or published.

Medium 🟩 🟩

No benefit of dihexa reaches this evidence level either. Medium-level grading requires at least consistent human observational or small-trial evidence; the entire dihexa evidence base is preclinical.

Low 🟩

Restoration of Impaired Spatial Memory in Rodent Models

In rats whose memory had been chemically blocked with scopolamine, and in aged rats with spontaneous decline, oral dihexa restored maze performance toward that of untreated controls. The proposed mechanism is the formation of new synaptic connections rather than a stimulant or cholinergic effect, which is consistent with the delayed and persistent time course reported. The evidence base is several rodent studies, most from the originating laboratory but with partial independent support from a Chinese group working in Alzheimer’s-model mice. The important limitation is that these are all models of deficit: the same literature shows little or no gain in cognitively normal animals, and the systematic review of the wider peptide family found effects far more reliable with direct brain delivery than with systemic dosing.

Magnitude: Restoration to near-control maze performance at 2 mg/kg/day orally in rats; scaled by body surface area this corresponds to roughly 0.3 mg/kg, or about 20 mg/day, for a 70 kg adult.

Increased Dendritic Spine Density and Synapse Formation

Dihexa increases the number of dendritic spines — the receiving points where one nerve cell connects to another — in cultured hippocampal neurons and in the cortex of Alzheimer’s-model mice, measured by the presynaptic marker synaptophysin. The proposed mechanism is amplification of hepatocyte growth factor signalling through the MET receptor and the downstream PI3K/Akt pathway. The cell-culture measurements originate in a paper retracted in 2025 and cannot be relied upon; the tissue-level measurement in mice comes from an independent group and stands. Whether structural synapse counts translate into functional cognitive gain in an intact adult human brain is untested.

Magnitude: Roughly a 50% to 100% increase in cortical synaptophysin expression after three months of oral dosing in Alzheimer’s-model mice; the frequently repeated figure of a 2.5- to 3-fold spine increase at picomolar concentrations comes from retracted work.

Reduction of Neuroinflammatory Signalling

In Alzheimer’s-model mice, dihexa reduced activation of astrocytes and microglia (the brain’s resident immune cells) and shifted the cytokine balance — lowering interleukin-1 beta and tumour necrosis factor alpha while raising interleukin-10, an anti-inflammatory signal. The proposed mechanism is PI3K/Akt activation, since blocking that pathway with wortmannin abolished the effect. The evidence is a single independent study with a clear pharmacological control, which is a stronger design than most of the dihexa literature but is still one experiment in one model.

Magnitude: Not quantified in available studies.

Neuroprotection in Chemically Lesioned Dopamine Systems ⚠️ Conflicted

The originating group reported that dihexa protected dopamine-producing neurons and restored motor function in rats given a chemical lesion that mimics Parkinson’s disease. Against this, an independent group testing the same compound in a chemically induced model of Huntington’s disease found no protection at all against weight loss, motor dysfunction, or memory impairment. Both are chemical-toxin models of movement disorders, and the discrepancy may reflect genuinely different injury mechanisms — mitochondrial poisoning in the negative study versus selective dopaminergic toxicity in the positive one — or it may reflect the difference between originator and independent testing. The conflict is unresolved.

Magnitude: Not quantified in available studies.

Protection of Sensory Hair Cells From Antibiotic Toxicity

In larval zebrafish, dihexa protected the sensory hair cells of the lateral line — cells closely equivalent to those of the mammalian inner ear — against damage from the aminoglycoside antibiotics neomycin and gentamicin. Protection was blocked by a hepatocyte growth factor antagonist and partly reduced by inhibitors of Akt, TOR (target of rapamycin, a master switch that couples nutrient supply to cell growth) and MEK (the relay kinase immediately upstream of ERK in the growth-signal cascade), which supports the growth-factor route in this system. The evidence is a single well-controlled study in a non-mammalian model, and no mammalian hearing study has followed it.

Magnitude: Optimal protection at 1 μM dihexa in the larval zebrafish lateral line; no mammalian dose-response established.

Adjunct to Surgical Peripheral Nerve Repair

In rats whose sciatic nerve had been cut and surgically repaired, dihexa delivered into the target muscle alongside mesenchymal stem cells improved walking-track motor scores and reduced foot contracture at 8 to 16 weeks, compared with repair alone. The proposed mechanism is the same growth-factor amplification acting on peripheral nerve and muscle rather than brain. The evidence is one multi-arm animal study with small group sizes; dihexa was one of several agents tested in combination, so its independent contribution is not cleanly isolated. This benefit sits outside the cognitive goal of this review but is part of the compound’s known profile.

Magnitude: Statistically significant improvement in walking-track grades at 8 to 16 weeks at 2 to 4 mg/kg; absolute effect size not reported in a form that can be quantified.

Speculative 🟨

Cognitive Enhancement in Healthy, High-Functioning Adults

This is the use that brings most readers of this review to dihexa, and it is the use with the least support of any listed here. No controlled study in healthy humans exists. The rodent literature points the other way: benefits concentrate in damaged or aged nervous systems and are weak or absent in normal animals, which is what a repair mechanism would predict. The basis for the claim is therefore mechanistic reasoning plus uncontrolled self-report from enhancement communities, which is the weakest evidence class available.

Faster Recovery From Concussion and Traumatic Brain Injury

Direct brain delivery of dihexa has been reported to rescue working memory in rats after repeated mild closed-head impact, with the effect blocked by a MET antagonist. The route used was intracerebroventricular — injection directly into the brain’s fluid spaces — which says little about what oral dosing would do, and the finding has not been independently replicated. The basis is a single preclinical report.

Diabetic Wound Healing and Peripheral Tissue Repair

Because hepatocyte growth factor drives new blood vessel formation and skin regrowth, amplifying it has been proposed to accelerate closure of poorly healing wounds. The supporting material comes from patent filings by the developing parties rather than peer-reviewed controlled studies, so the basis is mechanistic and commercial rather than experimental.

Benefit-Modifying Factors

Baseline cognitive status: This is the dominant modifier and it cuts against the enhancement use case. Across the angiotensin IV analogue literature, effects are large where function is impaired and small or absent where it is intact. A high-performing adult with no measurable deficit sits in the population where the animal data predict the least benefit.
Genetic variation in the MET and HGF genes: MET (the gene encoding the receptor dihexa is thought to work through) and HGF (the gene encoding that receptor’s natural ligand, hepatocyte growth factor) both carry common promoter and coding variants that alter receptor expression and signalling strength, and would plausibly set the ceiling on any response. No pharmacogenetic study of dihexa exists, so this is inference from receptor biology rather than measured effect.
APOE4 carrier status: APOE4 (a variant of the apolipoprotein E gene that impairs lipid transport and amyloid clearance in the brain) is the strongest common genetic risk factor for Alzheimer’s disease. Carriers accumulate synaptic loss earlier, which by the repair logic of this compound would put them among the more responsive, but the trial of the related clinical compound was not powered to test this and no dihexa data address it.
COMT Val158Met genotype: COMT (the enzyme that clears dopamine from the prefrontal cortex) comes in two common forms, and the Met/Met variant leaves more dopamine available and is associated with better baseline working memory and a smaller headroom for improvement from any cognitive agent. Relevant here as a general ceiling effect rather than a dihexa-specific interaction.
Baseline neurodegeneration markers: Individuals with an elevated phosphorylated tau 217 (a blood marker of Alzheimer’s-type brain change) or elevated neurofilament light chain (a blood marker of nerve fibre damage) have the kind of ongoing injury the compound was designed for. Individuals with clean markers are, by the same logic, less likely to have anything to repair.
Sex: Unknown, and this is a genuine gap rather than a null finding. The foundational rodent pharmacology used male rats almost exclusively, and the independent nerve-repair study used male rats only. Sex differences in MET expression and in growth-factor signalling are well documented in other tissues, so the absence of female data is a real limitation on generalising any of the reported effects.
Age: Aged rats responded where young rats largely did not. If the mechanism is genuine repair of lost connectivity, adults at the older end of the target range have more substrate for it to act on — while simultaneously carrying the higher background cancer risk discussed under risk modifiers.
Pre-existing conditions that may amplify response: Prior traumatic brain injury, cholinergic-deficit states, chemotherapy-associated cognitive impairment, and early neurodegenerative change are the contexts in which the preclinical signal appears. Conversely, poorly controlled metabolic disease blunts growth-factor signalling generally and would be expected to blunt any response here.

Potential Risks & Side Effects

High 🟥 🟥 🟥

Wholly Uncharacterised Human Safety Profile

Dihexa has never been administered to a human being in any registered clinical study, at any dose, by any route. There is no maximum tolerated dose, no adverse event frequency table, no human pharmacokinetic profile, no organ toxicity screen, and no pharmacovigilance signal — not because the compound has been shown to be safe, but because nothing has been looked for. Even the preclinical toxicology is thin: no published median lethal dose, no chronic dosing study, no genotoxicity assay, no reproductive toxicity study. Every dose taken outside a laboratory is a first-in-human exposure without monitoring.

Magnitude: Zero registered human trials and zero published human pharmacokinetic, tolerability, or adverse event data as of August 2026.

Failure of the Intended Effect Because the Mechanistic Basis Was Retracted

The specific risk here is spending money, time, and unmonitored biological exposure on a compound whose reason for being interesting has been withdrawn from the literature. Three of the four papers flagged by the journal were retracted in April 2025 after an institutional investigation into altered images, and the fourth — the paper that first described dihexa — remains under a notice of concern. No group has independently regenerated the binding and receptor-activation data. The one large human test of the same target, using a different molecule from the same programme, missed every efficacy endpoint.

Magnitude: Three of four flagged papers retracted; the primary human test of the shared target showed a between-group difference on its combined cognitive and daily-function endpoint of -0.08 with a p value (the probability that a difference this large would appear by chance alone) of 0.70, and no secondary endpoint reached significance.

Medium 🟥 🟥

Contaminated, Mislabelled, or Misdosed Gray-Market Product

Dihexa reaches users exclusively through research-chemical vendors operating outside pharmaceutical manufacturing standards. There is no pharmacopeial monograph for it, no compendial identity or purity test, no batch release requirement, and no regulatory inspection of the facilities producing it. The known failure modes for this supply channel are underdosing, overdosing, substitution with a cheaper compound, residual synthesis solvents, and — for anything reconstituted for injection — bacterial endotoxin. The evidence for this risk class is well established across gray-market peptides generally rather than measured for dihexa specifically, which is why it is graded here rather than higher.

Magnitude: Not quantified in available studies.

Low 🟥

Tumour Promotion Through Sustained MET Pathway Activation ⚠️ Conflicted

MET is a proto-oncogene: when it is amplified, mutated, or chronically switched on, it drives uncontrolled cell growth, invasion, and new blood vessel formation. This is not a theoretical worry drawn from a database — it is the reason several MET-directed inhibitor drugs (capmatinib, tepotinib, crizotinib) are approved for lung cancers carrying MET alterations. Dihexa is designed to push that pathway in the opposite direction, systemically, for as long as it is taken. Against this, the developers report that short-duration rodent dosing produced no neoplastic changes, and argue that potentiating existing hepatocyte growth factor is fundamentally different from constitutive receptor activation. The conflict is real and unresolved because the decisive experiment — a lifetime carcinogenicity study — has never been done.

Magnitude: Not quantified in available studies. For context, MET exon 14 skipping or MET amplification is present in roughly 3% to 4% of non-small-cell lung cancers, tumours for which MET inhibition is standard care.

Tissue Accumulation From a Very Long Terminal Half-Life

Rodent and patent-derived data describe a terminal elimination half-life of roughly 12 to 13 days after intravenous dosing, with a very large volume of distribution. If that translates even approximately to humans, daily dosing would accumulate for six to nine weeks before reaching steady state, and any adverse effect that emerged would persist for a comparable period after stopping. The practical consequences are that early tolerability says little about steady-state exposure, and that there is no rapid way to reverse an exposure once it has occurred. These figures come from the developing parties and have never been independently verified.

Magnitude: Terminal half-life of approximately 12 to 13 days in rats implies roughly 6 to 9 weeks to steady state and a comparable washout period.

Proliferative and Angiogenic Effects in Non-Neural Tissue

MET receptors are not confined to the brain; they are abundant on epithelial cells in liver, kidney, lung, gut, and skin, and on vascular endothelium. A systemically distributed MET potentiator would act on all of them. The plausible consequences include stimulation of pre-existing benign proliferative lesions, acceleration of abnormal blood vessel growth in tissues where that is already pathological, and unpredictable effects on wound and scar biology. No systematic tissue survey has been published for dihexa, and the reported volume of distribution indicates broad tissue exposure rather than brain selectivity.

Magnitude: Not quantified in available studies.

Speculative 🟨

Lowered Seizure Threshold From Excessive Synaptic Remodelling

A compound whose stated purpose is to add excitatory connections could in principle push a susceptible cortex toward hyperexcitability. No seizure has been reported in any dihexa study, and no electroencephalographic assessment has been performed in any species, so this rests entirely on mechanistic reasoning about what unregulated synaptogenesis would do.

Maladaptive Rewiring and Loss of Existing Circuit Precision

Synaptic pruning is as important to a functioning adult brain as synaptic formation. Indiscriminate connection-building could in principle degrade the signal-to-noise of established circuits rather than improve them, which would present as reduced mental clarity rather than as a recognisable adverse event. There is no experimental data on this in either direction; it is a structural concern about the mechanism.

Overstimulation, Irritability, and Disrupted Sleep Onset

Uncontrolled self-reports from enhancement communities describe difficulty falling asleep when dosing late in the day, and irritability at higher intakes. No controlled data exist, no mechanism has been proposed that would predict a stimulant-like effect, and self-report from unblinded users taking multiple compounds is the weakest possible basis. It is included because it is the only human-derived observation that exists.

Risk-Modifying Factors

Personal or family history of cancer: The single most important modifier. Any prior malignancy, any monitored precancerous lesion, or a strong family cancer history shifts the MET-activation concern from theoretical to concrete, because the pathway acts on cells that have already taken steps toward transformation.
Germline MET and HGF variants: Activating germline MET mutations cause hereditary papillary renal cell carcinoma, and MET promoter variants that raise receptor expression are documented. An individual who already sits at the high end of MET signalling would be adding an amplifier to an amplified system. This is not routinely tested and would only be known from broad genomic sequencing.
Baseline cancer screening status: Risk is materially different for someone who is current on age-appropriate screening — colonoscopy, dermatological examination, mammography or prostate assessment, low-dose computed tomography for those with significant smoking history — and someone who is not. Occult disease is the scenario in which growth-factor amplification is most consequential.
Sex: Not established for adverse effects, and all foundational rodent toxicology used male animals. Indirectly, sex determines which hormone-responsive proliferative conditions are relevant: uterine fibroids and endometriosis on one side, prostatic hyperplasia on the other, all of which involve tissues carrying MET receptors.
Age: Two opposing pressures. Older adults have more synaptic loss to repair, which is where any benefit would come from, but they also carry substantially higher background incidence of occult malignancy and more accumulated cellular damage, which is exactly where the MET concern bites hardest. Adults at the older end of the target range therefore face both the largest potential upside and the largest unquantified downside.
Liver and kidney function: Clearance is described as combined hepatic and renal. Impairment in either organ would extend an already long half-life, and because no human pharmacokinetic study exists there is no dose adjustment guidance to apply. Reduced function turns an uncharacterised exposure into an uncharacterised and prolonged one.
Pre-existing proliferative and vascular conditions: Proliferative diabetic retinopathy, psoriasis, endometriosis, and uterine fibroids all involve tissue that is already growing or vascularising abnormally, and all involve MET-bearing cells. These conditions convert the angiogenic and proliferative concern from a general one into a specific, locatable one.
Baseline inflammatory and metabolic state: Elevated high-sensitivity C-reactive protein, insulin resistance, and elevated insulin-like growth factor 1 all indicate a growth-permissive internal environment. Layering a growth-factor amplifier onto that state is a different proposition from doing so in a metabolically quiet one.

Key Interactions & Contraindications

No interaction study of dihexa has ever been performed. Everything below is derived from the pharmacology of the target and should be read as predicted rather than observed.

MET-directed kinase inhibitors — capmatinib, tepotinib, crizotinib, cabozantinib (absolute contraindication): Direct pharmacological opposition. Dihexa is designed to increase MET signalling; these agents are prescribed to abolish it in patients with MET-driven cancer. Concurrent use risks undermining oncological treatment. No mitigation exists other than complete avoidance.
Hepatocyte growth factor and MET-targeting antibodies used in oncology trials (absolute contraindication): Same logic as above, at the ligand rather than the receptor. Anyone enrolled in an oncology study of this class should regard dihexa as disqualifying.
Other tyrosine kinase inhibitors and antiangiogenic agents — sunitinib, sorafenib, bevacizumab (caution): Overlapping downstream signalling and opposing effects on blood vessel growth. Clinical consequence is unpredictable interference with a cancer treatment regimen; the mitigating action is to defer any use until oncological therapy is complete and remission is established.
Angiotensin-converting enzyme inhibitors (lisinopril, enalapril, ramipril) and angiotensin receptor blockers (losartan, valsartan, telmisartan) (monitor): These reshape the entire renin-angiotensin cascade from which dihexa is derived, altering the availability of angiotensin fragments in brain tissue. The predicted consequence is a change in dihexa’s effect size in either direction rather than a safety event. Mitigation is awareness that any perceived response may not be reproducible if antihypertensive therapy changes.
Cholinesterase inhibitors — donepezil, rivastigmine, galantamine (caution, additive): Both act on memory-relevant circuitry, one by preserving acetylcholine and the other by adding connections. Additive cognitive effect is plausible and would not be dangerous, but additive gastrointestinal and sleep-related effects are also plausible. Mitigation is to avoid initiating both within the same four-week window so that any effect can be attributed.
Anticholinergic medications available over the counter — diphenhydramine, dimenhydrinate, doxylamine (monitor): These block the same acetylcholine signalling that scopolamine blocks in the rodent model dihexa was validated against, and would be expected to oppose any cognitive effect. Practical mitigation is to avoid sedating antihistamines during any assessment period.
Over-the-counter non-steroidal anti-inflammatory drugs — ibuprofen, naproxen (monitor): Chronic use suppresses prostaglandin-dependent growth-factor signalling and could blunt the intended effect; conversely no safety interaction is predicted. Mitigation is timing separation only if a response is being formally assessed.
Growth-signalling supplements with additive effects — alpha-glycerylphosphorylcholine, citicoline, huperzine A, Hericium erinaceus, semax, selank, and cerebrolysin (caution): All are taken for neurotrophic or cholinergic effect and are commonly stacked with dihexa. The clinical consequence of stacking is not a specific toxicity but total loss of attribution: no individual compound’s effect or adverse effect can be identified. Mitigation is single-variable use with a minimum four-week separation between additions.
BPC-157 and other angiogenic peptides (caution): These promote new blood vessel formation through vascular endothelial growth factor receptor signalling, which is additive with MET-driven angiogenesis. The predicted consequence is amplification of the proliferative concern described under risks. Mitigation is avoidance of concurrent use, particularly for anyone with any cancer history.
Growth hormone secretagogues and insulin-like growth factor 1 raising agents — ipamorelin, CJC-1295, tesamorelin (caution): Convergent activation of PI3K/Akt and growth signalling. Same reasoning and same mitigation as above.
Rapamycin, metformin, and other growth-pathway suppressors (monitor, opposing): These inhibit mTOR (mechanistic target of rapamycin, the growth-promoting switch downstream of PI3K/Akt) or activate AMPK (AMP-activated protein kinase, the cell’s low-energy sensor, which shifts metabolism away from growth) and would be expected to blunt the downstream limb of dihexa’s proposed mechanism. This is a legitimate strategy for someone who wants to constrain proliferative risk, at the cost of the intended effect. Timing separation of dosing does not resolve the conflict because both act on sustained signalling tone.
Populations who should not use this intervention:
- Anyone with an active malignancy, or any malignancy treated within the past five years, regardless of type or stage.
- Anyone with a monitored precancerous lesion: Barrett’s oesophagus (a change in the lining of the lower gullet after long-term acid reflux), cervical intraepithelial neoplasia (abnormal cervical cells), colonic adenomas on surveillance, monoclonal gammopathy of undetermined significance (a harmless-for-now excess of one antibody-producing cell line that can progress to myeloma), or dysplastic naevi (irregular moles with early abnormal features) under review.
- Anyone with a known hereditary cancer syndrome, particularly hereditary papillary renal cell carcinoma (germline MET mutation), Lynch syndrome, or Li-Fraumeni syndrome.
- Anyone with hepatic impairment of Child-Pugh Class B or C, or renal impairment with an estimated glomerular filtration rate below 60 mL/min/1.73 m².
- Anyone pregnant, attempting to conceive, or breastfeeding: no reproductive or developmental toxicity study exists, and MET signalling is essential to placental and fetal development.
- Anyone under 18 years of age, whose synaptic pruning programme is still running.
- Anyone with an established seizure disorder or a history of unprovoked seizure.
- Anyone with proliferative diabetic retinopathy or another active proliferative vascular condition.

Risk Mitigation Strategies

Baseline cancer screening: Completed before any exposure, covering colonoscopy per standard interval, dermatological examination, mammography or prostate-specific antigen assessment as sex-appropriate, and low-dose computed tomography where there is a 20 pack-year smoking history. This directly addresses the MET-activation concern by establishing that no occult or precancerous lesion is present for the pathway to act on.
Scheduled laboratory panel: Complete blood count, comprehensive metabolic panel, lactate dehydrogenase, high-sensitivity C-reactive protein, and insulin-like growth factor 1 before starting, at 4 weeks, at 12 weeks, and every 3 months thereafter. This creates the reference point without which no laboratory change can be interpreted, and detects the marrow, hepatic, and renal signals most likely to appear first.
Hard time limit on dosing: Blocks of 8 weeks followed by a minimum 8-week interruption. The rationale is direct: the reported terminal half-life of roughly 12 to 13 days means continuous dosing accumulates for two months, and cumulative MET pathway exposure is the variable most plausibly linked to proliferative risk.
Low starting dose: Roughly 5 mg/day rather than the 20 mg/day implied by rodent-to-human body-surface-area scaling, held for at least 2 weeks before any increase. This mitigates the risk that human sensitivity differs from rodent sensitivity in a compound with no human dose-finding data whatsoever.
Batch-specific third-party analysis before use: A certificate of analysis showing high-performance liquid chromatography purity of at least 98% and mass-spectrometric identity confirmation, matched to the batch number on the vial, ideally from a laboratory independent of the vendor. This mitigates the contamination and mislabelling risk, which is the risk most easily reduced by action.
Single-compound use: One compound at a time, with a minimum 4-week separation between additions. Directly mitigates the attribution failure that makes gray-market stacking dangerous — without separation, neither benefit nor an emerging adverse effect can be assigned to a source.
Discontinuation triggers defined in advance: A predefined stopping list covering any new palpable mass, unexplained weight loss exceeding 5% of body weight, unexplained lymphadenopathy (swollen lymph nodes), a changing skin lesion, a first seizure, persistent headache, or a lactate dehydrogenase or liver transaminase rise above the upper reference limit. Predefining these prevents the common failure of rationalising an early signal, which matters more than usual given a washout measured in weeks.
Morning dosing with sleep tracking: Administration before 10:00 with sleep onset latency recorded, addressing the reported overstimulation and delayed sleep onset. Because sleep is when synaptic consolidation occurs, degrading it would work against the intended mechanism as well as being unpleasant.

Therapeutic Protocol

There is no clinical protocol for dihexa. No regulatory authority has approved a dose, no clinical trial has established one, and no medical body has issued guidance. What follows describes the three competing approaches that exist and the parameters each rests on, without presenting any of them as authoritative.

The preclinical-derived approach: The only dose with experimental support is the 2 mg/kg/day oral dose used in the rodent memory studies from the Washington State University laboratory of Joseph Harding and John Wright, who originated the compound. Converted by standard body-surface-area scaling — multiplying by the ratio of rat to human correction factors — this gives approximately 0.3 mg/kg, or about 20 mg/day for a 70 kg adult. This conversion is a regulatory convention for estimating a starting dose, not a demonstration that the dose is effective or safe in humans.
The clinical-development approach: The commercial route abandoned dihexa itself. M3 Biotechnology, later Athira Pharma, developed fosgonimeton — a distinct molecule acting on the same hepatocyte growth factor / MET system — for daily subcutaneous injection at 40 mg and 70 mg. This approach was chosen specifically because dihexa’s own drug-like properties were considered inadequate for development, which is a substantive argument against oral dihexa that the enhancement community rarely engages with. It also failed its efficacy endpoints.
The community approach: Within enhancement communities the conventional intake is 5 to 20 mg/day, most often as a powder dissolved in dimethyl sulfoxide (a solvent that carries compounds through skin) and applied topically, or taken orally in capsules, in cycles of 4 to 8 weeks. No expert or clinic is credibly associated with popularising this; it emerged from vendor marketing and forum convention rather than from any practitioner. Its dose range broadly brackets the scaled rodent dose, which is likely where it came from.
Best time of day: Morning administration is the consistent convention, based on reports of delayed sleep onset with later dosing. If the mechanism is genuinely structural rather than stimulant, timing should matter little for effect and only for tolerability.
Expected half-life: Reported as an in vitro serum half-life of several hours with a terminal elimination half-life of roughly 12 to 13 days in rats. No human figure exists. A terminal half-life of this length is the pharmacological fact that most shapes any sensible protocol, because it means the compound is still accumulating for weeks after dosing begins.
Single versus split dosing: A terminal half-life measured in days makes split dosing pharmacologically pointless — plasma concentration is governed by accumulation, not by peaks. Single daily administration is the only rational schedule, and less-than-daily administration is defensible on the same reasoning.
Genetic polymorphisms relevant to protocol choice: MET and HGF variants would be expected to set responsiveness, APOE4 status to set how much synaptic loss there is to address, and COMT genotype to set baseline cognitive headroom. None has been studied with this compound, so none can currently guide dose selection; they are listed because they are where pharmacogenetic guidance would come from if it existed.
Sex-based differences in dosing: Unknown. The rodent pharmacology that generated the only defensible dose used male animals, so the scaled dose is, strictly, a male-derived figure. There is no basis for a sex-adjusted protocol beyond body weight.
Age-related considerations: Older adults have both the greatest theoretical responsiveness and the greatest exposure to the proliferative concern. Renal and hepatic clearance also decline with age, which for a compound with a two-week terminal half-life and no human data means longer effective exposure at the same nominal dose. Nothing in the literature supports a specific age-adjusted dose.
Baseline biomarkers influencing response: Phosphorylated tau 217 and neurofilament light chain indicate whether there is active neurodegeneration to act on; high-sensitivity C-reactive protein and fasting insulin indicate whether the internal environment is growth-permissive. These do not change the dose but do change the expected probability of any effect at all.
Pre-existing conditions influencing response: Prior head injury, cholinergic deficit, and early neurodegenerative change define the states in which the preclinical signal appears. Uncontrolled metabolic disease, chronic sleep restriction, and untreated depression all suppress neurotrophic signalling and would be expected to override any contribution from this compound.

Discontinuation & Cycling

Lifelong versus short-term use: Nothing in the evidence supports indefinite use. The compound was developed as a disease-modifying treatment intended to rebuild lost connectivity, which implies a finite repair phase rather than permanent administration, and the unquantified proliferative concern scales with cumulative exposure. Time-limited use is the only position the evidence supports.
Known withdrawal effects: None documented in any species. No dependence, tolerance, or rebound phenomenon has been reported, and none is predicted by the mechanism — growth-factor amplification does not produce the receptor adaptations that underlie withdrawal syndromes.
Tapering: Pharmacologically unnecessary and practically irrelevant. With a terminal half-life of roughly 12 to 13 days, stopping abruptly still produces a gradual decline over six to nine weeks. A deliberate taper adds exposure without changing the concentration curve meaningfully.
Cycling for maintaining efficacy: No tolerance has been demonstrated, so there is no efficacy argument for cycling. There is, however, a safety argument: interrupting exposure limits cumulative MET pathway activation and allows the long terminal phase to clear before the next block. Blocks of 8 weeks on and at least 8 weeks off follow directly from the reported half-life rather than from any efficacy data.
Washout before medical procedures: Any planned surgery, oncological investigation, or biopsy warrants a washout of at least 8 weeks, on the reasoning that growth-factor amplification could confound histological interpretation and act on tissue that is being assessed precisely because it may be proliferating.

Sourcing and Quality

Regulatory position determines where it can be obtained: Dihexa is not an approved drug, not a dietary supplement, and not currently available through legitimate compounding. The United States Food and Drug Administration removed dihexa acetate from Category 2 of the interim list of bulk drug substances for compounding under section 503A in April 2026, but removal from that category is not authorisation — the substance still has to clear Pharmacy Compounding Advisory Committee review and be added to the 503A bulks list before any compounding pharmacy may legally prepare it. Until then the only supply channel is research-chemical vendors (FDA, Bulk Drug Substances Used in Compounding Under Section 503A).
What a certificate of analysis must actually show: A batch-specific document, with the batch number matching the vial, showing high-performance liquid chromatography purity of at least 98%, mass-spectrometric confirmation of the expected mass for the chemical identity, and a residual solvent screen. A generic certificate not tied to a batch number is marketing material rather than analysis.
Independent verification is the meaningful step: Vendor-supplied certificates are self-reported. Independent analytical laboratories that accept consumer submissions for peptide and small-molecule identity and purity testing exist and cost a fraction of a typical order. For a compound with no regulatory oversight at any point in the chain, third-party testing is the only quality control that exists.
Identity checks a buyer can perform: The compound should be listed by its full chemical name, N-hexanoic-Tyr-Ile-(6) aminohexanoic amide, and by Chemical Abstracts Service registry number 1401708-83-5. Vendors that cannot state both, or that describe dihexa as a peptide hormone or as approved for any indication, are describing a product they do not understand.
Formulation and solubility considerations: Dihexa is poorly water-soluble, which is why topical preparations use dimethyl sulfoxide as a carrier. This has two consequences worth weighing: dimethyl sulfoxide carries whatever else is present in the preparation through the skin alongside the intended compound, which makes impurity a more serious matter than it would be for an oral capsule, and it commonly causes local irritation and a characteristic garlic-like taste and body odour.
Sterility applies only to injectable preparation: Anything reconstituted for injection requires bacterial endotoxin testing and sterile handling. Research-chemical material is not manufactured to sterile-injectable standard, and no vendor certificate should be read as implying that it is.

Practical Considerations

Time to effect: Unknown in humans. In rodent studies behavioural improvement emerged over days to weeks of daily dosing rather than acutely, which is consistent with a structural mechanism rather than a stimulant one. Any perceived effect within hours of a first dose is more likely to be expectation than pharmacology, and the long accumulation phase means a fair assessment period is at least 6 to 8 weeks.
Common pitfalls: Treating the “far more potent than the body’s own growth factors” claim as an effect size when it is a concentration comparison in cell culture — potency describes the dose needed, not the magnitude of benefit. Stacking several neurotrophic compounds simultaneously, which destroys attribution. Assuming that because a related compound reached late-stage clinical trials, dihexa itself has been through any part of that process. Buying without a batch-matched certificate of analysis. Skipping baseline cancer screening. Dosing in the evening and then attributing the resulting poor sleep to something else.
Regulatory status: Not approved by any national regulator for any indication, anywhere. Not a dietary supplement under United States law, so it cannot legally be marketed for human consumption; it is sold under research-use-only labelling. For competitive athletes it falls under the World Anti-Doping Agency’s S0 category covering pharmacological substances not approved by any governmental health authority for human therapeutic use, which prohibits it at all times regardless of whether it is named individually.
Cost, access, and payer incentives: The compound itself is inexpensive to synthesise and is priced accordingly; cost is not the barrier. Access is legally and analytically constrained rather than financially constrained. It is worth noting where the money in this field does sit: no insurer or national health system covers any synaptogenic agent, and institutional payers have a clear financial preference for the cheapest adequate option in cognitive decline — a preference that has historically favoured generic cholinesterase inhibitors over novel high-cost biologics and that shapes which mechanisms attract development funding. Growth-factor approaches have been chronically underfunded relative to amyloid-directed programmes for reasons that are as much about reimbursement expectations as about biology.
Practical monitoring burden: The screening and laboratory schedule that makes use of this compound defensible costs more, in money and appointments, than the compound does. Anyone unwilling to complete baseline cancer screening is carrying the compound’s principal risk without the one measure that reduces it.

Interaction with Foundational Habits

Sleep: Bidirectional and potentially self-defeating if mishandled. Direction is potentiating in one direction and blunting in the other: synaptic consolidation and pruning both occur predominantly during slow-wave sleep, so a compound that adds synapses depends on sleep to integrate and refine them. Uncontrolled reports of delayed sleep onset with evening dosing point the other way. Practical consideration is morning administration and protection of sleep duration, since chronic restriction suppresses growth-factor signalling and would blunt the mechanism directly.
Nutrition: Direct and modest. Dihexa is lipophilic and poorly water-soluble, so oral absorption is likely improved by taking it with a meal containing fat, by the same mechanism that applies to fat-soluble vitamins. Downstream, the PI3K/Akt and mTOR signalling that the mechanism converges on is nutrient-sensitive: adequate protein and leucine support it, while prolonged fasting and ketogenic states suppress it. Anyone using extended fasting protocols is running the intervention and a physiological brake on the same pathway simultaneously.
Exercise: Potentiating, and the interaction is the best-evidenced item in this section. Aerobic exercise raises circulating hepatocyte growth factor as well as brain-derived neurotrophic factor. Since dihexa’s proposed mechanism is amplification of hepatocyte growth factor rather than replacement of it, more available ligand should mean more effect from the same dose — the amplifier has more signal to work with. Practical consideration is placing aerobic sessions on dosing days, and recognising that regular aerobic exercise produces measurable cognitive benefit in humans on evidence that is incomparably stronger than anything supporting dihexa.
Stress management: Blunting. Sustained cortisol elevation suppresses neurotrophic signalling and causes measurable dendritic retraction in the hippocampus and prefrontal cortex — the precise structures and the precise structural change this compound is meant to reverse. Chronic psychological stress is therefore not a background variable here but a direct pharmacological antagonist. Practical consideration is that addressing chronic stress will do more for the outcome being sought than the compound plausibly will, and that failure to address it may make any effect undetectable.

Monitoring Protocol & Defining Success

Baseline testing exists to answer two questions before any exposure occurs: is there occult proliferative disease that a growth-factor amplifier could act on, and what are this individual’s normal values so that a later change can be recognised as a change. Because dihexa has no human safety record, the panel below is constructed from the organ systems its mechanism touches rather than from any observed adverse event profile. It should be drawn fasting and in the morning, alongside the age-appropriate cancer screening described under risk mitigation.

Ongoing monitoring follows a fixed cadence: repeat the full panel at 4 weeks, at 12 weeks, and every 3 months for as long as use continues, with age-appropriate cancer screening maintained annually rather than at the standard longer intervals. Because the terminal half-life is measured in weeks, a final panel 8 weeks after stopping is worth drawing to confirm that any change has resolved.

Biomarker	Optimal Functional Range	Why Measure It?	Context/Notes
Complete blood count with differential	White cells 4.5–7.0 K/μL; platelets 175–250 K/μL; haemoglobin mid-reference for sex	Detects marrow, immune, or occult proliferative change	Complete blood count (CBC) is the standard cell-line panel. Conventional laboratories flag only outside 4.0–11.0 K/μL for white cells; a persistent upward drift within range is the signal of interest here
Comprehensive metabolic panel: alanine and aspartate aminotransferase	Alanine aminotransferase 10–26 U/L (men), 10–19 U/L (women); aspartate aminotransferase 10–26 U/L	Liver is a MET-rich organ and a clearance route for the compound	Comprehensive metabolic panel (CMP) covers liver, kidney, and electrolytes in one draw. Conventional upper limits reach 40–55 U/L, well above the functional range; fast 10–12 hours and avoid intense exercise for 48 hours beforehand
Estimated glomerular filtration rate and creatinine	Estimated glomerular filtration rate above 90 mL/min/1.73 m²	Renal clearance contributes to elimination; impairment prolongs an already long half-life	Estimated glomerular filtration rate (eGFR) is a calculated measure of kidney filtration. Conventional practice treats anything above 60 as normal; creatinine is inflated by recent creatine supplementation and heavy protein intake
Lactate dehydrogenase	140–180 U/L	Non-specific marker of cell turnover and tissue proliferation; a rising trend warrants investigation	Lactate dehydrogenase (LDH) is an enzyme released by damaged or rapidly dividing cells. Conventional range extends to roughly 250 U/L. Haemolysis during the draw falsely elevates it, so a repeat before acting on a single high value is essential
High-sensitivity C-reactive protein	Below 0.9 mg/L	Establishes whether the internal environment is inflammatory and growth-permissive before starting	High-sensitivity C-reactive protein (hs-CRP) is a sensitive general inflammation marker. Conventional cardiovascular cut-off is 3.0 mg/L. Invalid within 2 weeks of any infection, injury, or vaccination
Insulin-like growth factor 1	Mid-reference for age, roughly 120–160 ng/mL at ages 40–60	Indicates systemic growth signalling tone that dihexa’s pathway would add to	Insulin-like growth factor 1 (IGF-1) is the main mediator of growth hormone. Conventional ranges are very wide (80–250 ng/mL); the upper half of the range in combination with this compound is the configuration to avoid
Fasting insulin and haemoglobin A1c	Insulin 2–5 μIU/mL; haemoglobin A1c 4.9–5.3%	Insulin resistance both blunts neurotrophic signalling and raises proliferative risk	Haemoglobin A1c (HbA1c) reflects average glucose over roughly 3 months. Conventional thresholds are insulin below 25 μIU/mL and HbA1c below 5.7%. Requires a 12-hour fast; A1c is falsely low in anaemia and shortened red cell survival
Plasma phosphorylated tau 217	Below the assay-specific cut-off for Alzheimer’s-type change	Indicates whether there is active neurodegeneration for the mechanism to address	Phosphorylated tau 217 (p-tau217) is a blood marker of Alzheimer’s-type brain change. No universal reference range exists; results are only comparable within the same assay and laboratory
Neurofilament light chain	Below 10 pg/mL under age 50; below 20 pg/mL over age 60	Marker of ongoing nerve fibre damage; a rising value on treatment would be a stop signal	Neurofilament light chain (NfL) is released when nerve axons are injured. Rises with age and with reduced kidney function, so interpret alongside estimated glomerular filtration rate
Prostate-specific antigen (men over 45)	Below 1.0 ng/mL at ages 40–50; below 2.5 ng/mL thereafter	The most accessible marker of a common hormone-responsive proliferative process	Prostate-specific antigen (PSA) is produced by prostate tissue. Conventional threshold for investigation is 4.0 ng/mL. Falsely elevated by cycling, ejaculation within 48 hours, and recent examination

Qualitative markers matter as much as the panel here, because the intended effect is cognitive and no blood test measures it. The following should be recorded at baseline and reviewed at each laboratory timepoint:

Word-finding fluency and the frequency of tip-of-the-tongue episodes in ordinary conversation
Working memory in practice: holding a multi-step instruction, keeping track of several threads in a discussion
Sustained attention during a demanding task, and how long it holds before it degrades
Sleep onset latency and subjective sleep quality, given the reported evening-dosing effect
Mood stability and irritability, which are the two effects most commonly self-reported
A repeatable objective cognitive battery administered at the same time of day at baseline, 4 weeks, and 12 weeks — subjective impressions of one’s own cognition are unreliable, and an unblinded self-assessment of an expensive intervention is the least trustworthy measurement available
New physical findings: palpable masses, changing skin lesions, unexplained weight loss, persistent headache

Success, defined honestly, means a measurable and reproducible change on the objective battery that is not explained by practice effects, accompanied by an unchanged safety panel. Absence of that change after 12 weeks is a result, not a reason to increase the dose.

Emerging Research

No registered dihexa trial: A search of the trial registry returns zero studies of dihexa, PNB-0408, or the chemical name, at any phase, in any country. This is the single most important fact about the compound’s research status and it has not changed in more than a decade since it was described.
LIFT-AD, the shared-target read-out: NCT04488419 enrolled 554 participants with mild-to-moderate Alzheimer’s disease across Phase 2/3, testing daily subcutaneous fosgonimeton against placebo, with a combined cognitive and daily-function endpoint at 26 weeks. It completed in July 2024 and missed its primary and all secondary endpoints, with only a marker of Alzheimer’s-type brain change showing a nominal difference (Porsteinsson et al., 2025). This is the closest existing test of whether potentiating the hepatocyte growth factor / MET system improves human cognition, and it did not.
Supporting trials in the same programme: NCT04491006 was a 77-participant Phase 2 study in mild-to-moderate Alzheimer’s disease; NCT04831281 tested the same agent in Parkinson’s disease dementia and dementia with Lewy bodies in 28 participants and was terminated; NCT04886063 was a 423-participant open-label extension, also terminated; NCT05511558 was an 8-participant radiolabelled absorption, metabolism and excretion study. The Phase 1 first-in-human study NCT03298672 enrolled 88 participants and reported the agent safe and well tolerated with a signal of faster brain electrical responses (Hua et al., 2022).
Next-generation oral agent: NCT06432647 is a completed 80-participant Phase 1 single- and multiple-ascending-dose study of ATH-1105, an orally administered small-molecule positive modulator of the same system, in healthy volunteers. If it advances, it would provide the human tolerability data for oral MET potentiation that dihexa has never generated — and would be the most direct route to strengthening the case for this class.
Independent binding replication: This is the decisive open question. The direct evidence that dihexa binds hepatocyte growth factor and promotes MET receptor activation was withdrawn from the literature in April 2025. No group has published a replication attempt. A clean, independently performed binding and receptor-phosphorylation study would either restore the mechanistic foundation or remove it definitively, and it is a modest experiment relative to its importance.
Long-term carcinogenicity data: These would resolve the principal safety question. No lifetime rodent carcinogenicity study, genotoxicity assay, or reproductive toxicity study of dihexa has been published. Until one exists, the proliferative concern cannot be quantified in either direction. A negative carcinogenicity study would materially strengthen the case for extended use; a positive one would end it.
Alternative mechanistic accounts: These remain testable and could weaken the growth-factor case further. The competing proposal that angiotensin IV analogues act by inhibiting insulin-regulated aminopeptidase, and the independent finding that dihexa’s effects in Alzheimer’s-model mice depend on PI3K/Akt and are accompanied by raised tissue angiotensin IV (Sun et al., 2021), both offer routes by which the compound could be active without the mechanism it is marketed on being correct.
Accumulating negative evidence: The negative independent result in a chemically induced movement-disorder model (Wells et al., 2024) is the first published failure to replicate a neuroprotective effect, and the systematic review of the wider peptide family found the effects most reliable with direct brain delivery rather than the systemic route used outside the laboratory (Ho & Nation, 2018). Further independent negative results in oral dosing models would make the enhancement case difficult to sustain.

Conclusion

Dihexa is a synthetic compound designed to strengthen a natural growth signal that helps nerve cells build new connections. In rodents it restored memory damaged by drugs, ageing, or injury, and it increased the density of connections between nerve cells. Those are the strongest results available, and they stop at animals. No person has ever taken dihexa in a registered study, so nothing is known about its dose, its side effects, or its long-term safety.

The evidence base carries an unusual burden. Nearly all of the original work came from one university laboratory and the company its founders spun out, both of which stood to gain financially from the compound being taken up. Three of the founding papers were withdrawn after an investigation found that images had been altered, and a fourth remains flagged. A later drug from the same programme, acting on the same target, did not improve thinking or daily function in people.

The most serious open question is biological rather than administrative: the growth signal dihexa strengthens is the same one that several approved cancer drugs are built to block, and whether strengthening it can promote tumours has never been examined in a long-term study.

For people who accept experimental risk in pursuit of sharper thinking, dihexa pairs a vivid laboratory story with a very thin human record, sold outside any system that checks what is in the vial.

Top - Benefits - Risks - Protocol

Dihexa for Cognitive Enhancement

Motivation

Recommended Reading

Grokipedia

Examine

ConsumerLab

Systematic Reviews

Mechanism of Action

Historical Context & Evolution

Expected Benefits

High 🟩 🟩 🟩

Medium 🟩 🟩

Low 🟩

Restoration of Impaired Spatial Memory in Rodent Models

Increased Dendritic Spine Density and Synapse Formation

Reduction of Neuroinflammatory Signalling

Neuroprotection in Chemically Lesioned Dopamine Systems ⚠️ Conflicted

Protection of Sensory Hair Cells From Antibiotic Toxicity

Adjunct to Surgical Peripheral Nerve Repair

Speculative 🟨

Cognitive Enhancement in Healthy, High-Functioning Adults

Faster Recovery From Concussion and Traumatic Brain Injury

Diabetic Wound Healing and Peripheral Tissue Repair

Benefit-Modifying Factors

Potential Risks & Side Effects

High 🟥 🟥 🟥

Wholly Uncharacterised Human Safety Profile

Failure of the Intended Effect Because the Mechanistic Basis Was Retracted

Medium 🟥 🟥

Contaminated, Mislabelled, or Misdosed Gray-Market Product

Low 🟥

Tumour Promotion Through Sustained MET Pathway Activation ⚠️ Conflicted

Tissue Accumulation From a Very Long Terminal Half-Life

Proliferative and Angiogenic Effects in Non-Neural Tissue

Speculative 🟨

Lowered Seizure Threshold From Excessive Synaptic Remodelling

Maladaptive Rewiring and Loss of Existing Circuit Precision

Overstimulation, Irritability, and Disrupted Sleep Onset

Risk-Modifying Factors

Key Interactions & Contraindications

Risk Mitigation Strategies

Therapeutic Protocol

Discontinuation & Cycling

Sourcing and Quality

Practical Considerations

Interaction with Foundational Habits

Monitoring Protocol & Defining Success

Emerging Research

Conclusion