Dihexa

What it is

In the 1990s, researchers noticed that a breakdown product of the blood pressure hormone angiotensin improved memory in rats. Nobody knew why. The group around Joseph Harding and John Wright at Washington State University followed the trail and ended up at hepatocyte growth factor HGF and its receptor c-Met — a system from liver and cancer research that apparently also makes synapses grow in the brain.

Dihexa was the designed product of this work: small, stable, orally swallowable. It was meant to pull exactly this lever. Today it is sold as a research chemical, usually with the promise that it makes synapses sprout.

How it is supposed to work

The chain of action was clearly formulated. Dihexa binds to HGF, amplifies its signal at the receptor c-Met and thereby triggers the formation of new contact points between nerve cells.

Exactly this chain was set out in the 2014 paper by Benoist and colleagues. It showed 4 things: binding to HGF with high affinity, activation of c-Met even at concentrations that on their own would not suffice, amplification of the cell migration typical of HGF — and, as the keystone, that the memory effect disappears when an HGF blocker is given directly into the brain.

Why the foundation is missing

In April 2025, the Journal of Pharmacology and Experimental Therapeutics retracted 3 papers by this group, including the one by Benoist. The reason was manipulated figures. An investigation by Washington State University concluded that co-author Leen Kawas had altered figures, in her 2011 doctoral thesis and in at least 4 publications.

Kawas was at the time chief executive of Athira Pharma, the company that emerged from this research and went public in 2020. In June 2021 she was placed on leave, and in October she resigned. In January 2025, Athira reached a settlement of just over 4 million dollars with the US Department of Justice — not because of the manipulation itself, but because the company concealed the allegations from the health authorities while applying for grant funding. There was no admission of guilt.

The figure quoted everywhere

Practically every product page states that Dihexa is 7 orders of magnitude more potent than BDNF. This figure does not come from a scientific paper but from a university press release from October 2012. In the associated publication, Dihexa and BDNF were not compared directly at all; 2 separately measured concentration ranges were put side by side.

Even if calculated correctly, the figure would say the wrong thing. Potency only means that something works at a lower concentration. It says nothing about whether something works better, works longer or is of any use at all. Nicotine is also more potent than caffeine. The best-known thing about Dihexa is thus also the least well supported.

What the studies show

McCoy 2013 — the introduction, with reservations

The paper that introduces Dihexa in the first place is by McCoy and colleagues from 2013, with rat experiments in the water maze. It has not been retracted. Since September 2021, however, it has been under an Expression of Concern, a formal reservation by the journal, for 5 years without a decision. That is not an acquittal but a state of limbo.

4 independent papers that contradict each other

In 2021, a Chinese group found improved cognition and less inflammation in an Alzheimer’s mouse model. In 2024, a group at Pacific University tested the same substance in a Huntington’s model in 40 rats and writes verbatim that it did not protect the animals. In addition, there are 2 stem cell papers in which Dihexa replaces HGF. To this day, nobody has independently measured whether Dihexa actually binds to HGF.

LIFT-AD — the principle tested in humans

With fosgonimeton, Athira developed a different molecule following the same school of thought: ramping up the HGF pathway in the brain. LIFT-AD was a placebo-controlled phase 2/3 trial with 554 enrolled Alzheimer’s patients, 287 of them in the primary analysis, over 26 weeks. Result in September 2024: missed, the primary endpoint had a p-value of 0.7. The threshold is 0.05; a value of 0.7 means practically no difference from placebo. 2 smaller studies before it were also negative.

What is missing in humans

For Dihexa itself there is nothing in humans: no study, no case report, not a single registered clinical trial, neither ongoing nor terminated. Since the retraction, the advertised mechanism has been left hanging, because no independent group has ever measured the binding to HGF.

Strictly speaking, a retraction for misconduct only says that these data are worthless. It does not say that the idea is wrong. The idea was, however, tested, with a different molecule and in a large trial, and it failed. The program was discontinued; in January 2026, Athira Pharma renamed itself LeonaBio and pivoted to oncology. An oral successor for the HGF pathway is still in development, for the nerve disease ALS. Two things therefore have to be separated: the signaling pathway is real. What is left hanging is the claim that this peptide of all things engages it.

Status, approval and legal

Dihexa is not approved as a medicine anywhere. Anyone who offers it for an effect in the body is placing an unapproved medicine on the market; research chemical is not a legal status but a label. For athletes there is an additional point: Dihexa is not named on the Prohibited List, but it falls under the category of non-approved substances, and these are prohibited at all times. The argument that it is not on the list does not hold.

Safety

Safety has not been studied. The honest wording is not dangerous but untested — for a substance that is supposed to switch on a receptor that oncology deliberately blocks. c-Met is an established cancer gene, and there are approved drugs that switch off exactly this receptor. For Dihexa there is not a single long-term or cancer study. In addition, there is a half-life of almost 13 days in the rat experiment: with repeated administration, the substance accumulates. Memory problems and neurological complaints belong in the hands of a physician.

Sources

• Benoist et al., Journal of Pharmacology and Experimental Therapeutics 2014 (retracted in April 2025)
• McCoy et al., Journal of Pharmacology and Experimental Therapeutics 2013 (Expression of Concern since September 2021)
• Press release of Washington State University, October 11, 2012
• ClinicalTrials.gov NCT04488419 (LIFT-AD, phase 2/3 with fosgonimeton, result September 2024)
• Porsteinsson et al., Journal of Alzheimer’s Disease Reports 2025 (LIFT-AD, full publication: 554 randomized, 287 in the primary analysis, p = 0.70)
• Settlement between Athira Pharma and the US Department of Justice, January 2025
• Investigation report of Washington State University on image manipulation