The Two Papers That Explained How Dihexa Works Were Retracted for Fabricated Data

What happened

In April 2025 the Journal of Pharmacology and Experimental Therapeutics retracted two papers that together built the case for how dihexa is supposed to work. Both retraction notices say the same thing. Following an investigation at Washington State University, specified figures were found to contain falsified and/or fabricated data, and two named authors, Leen H. Kawas and Joseph W. Harding, were found solely responsible.

The papers are Kawas and colleagues, 2012, on angiotensin IV analogues as hepatocyte growth factor modifiers, and Benoist and colleagues, 2014, on whether the procognitive and synaptogenic effects of those analogues depend on the hepatocyte growth factor system. A third paper from the same group, McCoy and colleagues, 2013, has carried a Notice of Concern since 2021 and has not been retracted. A Notice of Concern is a flag, not a withdrawal, and it should not be described as one.

The background is not obscure. Harding invented dihexa in his lab at Washington State University. Kawas was his graduate student, founded the company that became Athira Pharma to commercialise it, and went on to run it. She resigned in October 2021 after the university's investigation found she had altered images in her doctoral dissertation and in at least four co-authored papers published between 2011 and 2014.

What a retraction does and doesn't settle

So the accurate statement is narrower than either side of this argument would like. The mechanism everyone cites was proposed in papers that have since been withdrawn for data fabrication. It was not disproved, it was left unsupported, and nobody has gone back and redone that work in a way that would settle it.

What those papers were carrying

If you have read a product listing for dihexa, you have read that it potentiates hepatocyte growth factor and its receptor, that it drives new dendritic spines, and very often that it is seven orders of magnitude more potent than brain-derived neurotrophic factor. The first two claims trace back to the 2012 and 2014 papers, which are the retracted ones. The third is worth pausing on. The Alzheimer's Drug Discovery Foundation went looking for the source and reports that the two compounds were never directly compared in the publication, and that the comparison came from a press release.

The one place any of this reached humans

No trial of dihexa itself has ever been registered or published. What exists instead is fosgonimeton, a successor compound out of the same Washington State programme, developed by Athira. Be careful with that relationship, because product pages describe it as dihexa in trials and that is more than anyone can support. It is a different molecule from the same lineage.

Fosgonimeton had a clean Phase 1 in 88 people, safe and well tolerated across the dose range, though that trial measured brain wave and evoked potential markers rather than cognition. Then it got the real test. LIFT-AD randomised 315 people aged 55 to 85 with mild to moderate Alzheimer's disease to 40 milligrams a day by injection or to placebo for 26 weeks. In September 2024 Athira reported that it missed its primary endpoint and its key secondary endpoints, and development was discontinued.

That is the most useful thing anyone can tell you about this pathway, and it is a fact no product listing mentions. The hepatocyte growth factor mechanism did get its shot in humans, at proper scale, in the right population, and it did not produce a benefit.

What is known about safety

Very little, and the gaps have a shape worth knowing. The Alzheimer's Drug Discovery Foundation's review states that no studies in animals or humans have examined dihexa's long-term safety, and that no dose has been established for human use.

Two specifics follow from that. In rats the half-life ran to well over a week depending on route, and the human figure is unknown, so a compound taken daily accumulates toward a plateau nobody has measured. And the receptor at the centre of the proposed mechanism, c-Met, is an oncogene. Most of clinical oncology's interest in that pathway is in blocking it, while dihexa is an agonist of it that crosses into the brain, never tested for tumour promotion in any species over any duration. The same foundation review names this as a theoretical concern rather than a demonstrated one, which is the correct weight to give it. It is also worth noting that Athira's own trials of the successor compound excluded people with a history of malignant tumours.

One more finding sits oddly against how this compound is sold. In the animal work the effect was a reversal of a deficit, either chemically induced or age-related, and the same foundation review notes that dihexa did not improve cognitive function in rats whose cognition was normal to begin with. Normal cognition describes most of the people buying it.

What this means for you right now

If you are already buying dihexa, nothing changed this week, and that is rather the point. The retractions happened in April 2025 and they are still not reflected in most of what you will read about the compound. What has changed is what you can reasonably believe about the mechanism, which is now less than the product copy claims and less than it claimed two years ago.

If you are deciding whether to start, the retractions are not really the deciding factor. There is no human dosing data, no human safety data, no established dose, a very long animal half-life, and a structural question about an oncogenic pathway that nobody has answered. All of that was true before April 2025 and would still be true if the papers had never been withdrawn.

And if you are comparing vendors on it anyway, the variables you control are the ones you always control. What you are paying per milligram, and whether the seller will show you third party identity and purity testing on the lot you are buying. Our price comparison tool tracks per-milligram pricing across vendors, and the vendor directory tracks who publishes testing and who does not.

What to watch

The development that would genuinely change this is an independent group, with no connection to the original lab, replicating the hepatocyte growth factor mechanism and publishing it. That would be news and we would cover it. Short of that, dihexa stays where it is, an interesting idea whose foundational evidence was withdrawn and whose closest clinical relative failed a well-powered trial.