DIHEXA: the '100,000× BDNF' claim, examined

What McCoy et al. (2012) actually reported

The headline paper from Joseph Harding's group at Washington State University reported that DIHEXA produced effects in an in-vitro hippocampal synaptogenesis assay at concentrations approximately seven orders of magnitude lower than BDNF. The finding is precise, methodologically described, and supported by additional rodent behavioral data showing improved spatial-memory performance in aged cognitively-impaired animals.

The "100,000× more potent than BDNF" framing is a fair summary of the in-vitro result. What it does not directly establish is clinical efficacy in humans for cognitive endpoints — which is a much higher bar.

What in-vitro potency translates to (and doesn't)

In-vitro potency comparisons between molecules are highly assay-specific. The same molecules can produce very different rank-order potencies in different cell types, different downstream readouts, and different exposure conditions. A "potency" advantage at one specific assay doesn't automatically translate to a "potency" advantage in tissue-relevant biology, much less in clinical efficacy.

The McCoy 2012 finding is best read as: in this synaptogenesis-promoting cell-culture assay using these specific conditions, DIHEXA showed effects at much lower concentrations than BDNF. Whether DIHEXA at achievable doses in humans produces clinically meaningful cognitive enhancement is a different question that the in-vitro paper does not address.

What happened in the clinical translation

The drug-development effort that grew up around DIHEXA was M3 Biotechnology, later renamed Athira Pharma. The company's clinical lead became fosgonimeton (ATH-1017) — a different molecule targeting the same HGF/MET biology rather than DIHEXA itself. The Phase 2/3 ACT-AD trial of fosgonimeton in Alzheimer's disease did not meet its primary cognitive endpoint in published readouts.

This is informative for the DIHEXA discussion. The successor molecule with substantial commercial backing, designed to target the same biology that the in-vitro DIHEXA work pointed to, did not demonstrate the clinical cognitive benefit that the preclinical work had suggested. That doesn't disprove DIHEXA's potential — it could be a different molecule with different properties — but it does complicate the optimistic reading of the original 2012 finding.

The HGF/MET safety question

HGF/c-Met signaling, the pathway DIHEXA is proposed to activate, is also implicated in cancer biology. Aberrant c-Met activation contributes to tumor invasion, metastasis, and resistance to other targeted therapies in multiple cancer types. Chronic pharmacologic activation of this pathway in humans raises legitimate long-term safety questions that have not been investigated in any human dataset.

This is not a hypothetical concern in the way some safety questions are — c-Met inhibitors are an active oncology drug class precisely because c-Met activation is pro-tumorigenic. Activating the same pathway pharmacologically is a real safety question that the marketing discussions of DIHEXA typically don't engage with.

The honest editorial position

DIHEXA is interesting preclinical pharmacology with real biology behind it. The "100,000× BDNF" framing oversells the certainty of clinical translation. The successor program's Phase 2/3 disappointment complicates the optimistic reading. The HGF/MET pathway safety questions are non-trivial. The combination of these factors places DIHEXA firmly in the "hype-prone" category — the molecule's marketing-discussion profile has substantially outrun what the clinical evidence actually supports.