Dihexa and Alzheimer's: The Synaptogenesis Research
Dihexa
Failed Phase 2/3 primary endpoint
Fosgonimeton (dihexa's prodrug) did not reach statistical significance on cognition or function in the 26-week LIFT-AD trial for mild-to-moderate Alzheimer's disease, despite promising preclinical synaptogenesis data.
Athira Pharma LIFT-AD Topline Results, 2024
Athira Pharma LIFT-AD Topline Results, 2024
View as imageDihexa (N-hexanoic-Tyr-Ile-(6)-aminohexanoic amide) generated extraordinary claims when it first appeared in the peptide research literature. An angiotensin IV analog that crosses the blood-brain barrier, it was reported to promote synaptogenesis (the formation of new synaptic connections between neurons) through the hepatocyte growth factor (HGF)/c-Met receptor system. In aged rats with memory deficits, dihexa restored cognitive performance. For context on what the original potency claims actually said, the molecule was described as active at picomolar concentrations, leading to widespread but misleading characterizations of it as "millions of times more potent than BDNF."
The trajectory from laboratory promise to clinical reality has been sobering. Athira Pharma developed fosgonimeton, a phosphate prodrug of dihexa designed for subcutaneous injection, and advanced it into human trials for Alzheimer's disease. The phase 2/3 LIFT-AD trial, reported in 2024, failed to meet its primary or key secondary endpoints. The story of dihexa and Alzheimer's is now a case study in the gap between preclinical synaptogenesis data and clinical cognitive outcomes.
Key Takeaways
- McCoy et al. (2013) demonstrated that dihexa promotes dendritic spine formation and synaptogenesis in hippocampal neurons by augmenting HGF/c-Met signaling at subthreshold concentrations
- In aged rats, dihexa restored cognitive performance on spatial learning tasks comparable to young adult levels, with effects persisting after drug washout
- Fosgonimeton (dihexa prodrug) failed the primary endpoint of the phase 2/3 LIFT-AD trial in 302 mild-to-moderate Alzheimer's patients: no significant improvement in cognition or function over placebo at 26 weeks
- Biomarker data from the LIFT-AD trial showed reductions in phosphorylated tau217 (p < 0.01), suggesting biological activity despite the lack of clinical benefit
- A dual GLP-1/CCK receptor agonist improved cognitive performance and synaptogenesis in a 5xFAD Alzheimer's mouse model, showing that synaptogenesis-promoting peptides remain an active research area (Ma et al., 2025)
- No dihexa or fosgonimeton product is approved for any indication
The Original Synaptogenesis Research
The foundational work on dihexa came from the laboratory of Joseph Harding and John Wright at Washington State University, published primarily between 2011 and 2015. The key finding: dihexa activates the HGF/c-Met receptor system in neurons, promoting dendritic arborization (the growth of branching structures from neurons) and synaptogenesis (the formation of new functional connections between neurons).
The mechanism operates through an indirect pathway. Dihexa does not directly activate the c-Met receptor. Instead, it binds to HGF at a site that prevents HGF dimerization inhibition, effectively increasing the local concentration of active HGF available to bind c-Met. In the presence of subthreshold concentrations of HGF (concentrations too low to activate c-Met on their own), dihexa tips the balance toward receptor activation.
This mechanism was demonstrated in several experimental systems:
Hippocampal neuron cultures. Dihexa increased dendritic spine density and promoted the formation of new synapses at concentrations as low as 10 picomolar. The effect was blocked by c-Met inhibitors, confirming it operated through the HGF/c-Met pathway. Notably, dihexa did not directly activate c-Met at these concentrations. It required the presence of at least subthreshold amounts of HGF to produce its effect, functioning as a positive allosteric modulator rather than a direct agonist. This distinction is pharmacologically important: it means dihexa amplifies existing HGF signaling rather than creating signaling where none exists, which should theoretically limit off-target effects.
Scopolamine-induced amnesia model. Rats treated with scopolamine (which blocks acetylcholine signaling and impairs memory) showed restored performance on spatial learning tasks after dihexa administration. Dihexa was effective when given both intraperitoneally and orally, demonstrating blood-brain barrier penetration via either route. This model is considered a weak proxy for Alzheimer's disease because it disrupts a single neurotransmitter system rather than reproducing the complex pathology of neurodegeneration.
Aged rat model. Dihexa restored cognitive performance in aged rats (22-24 months, equivalent to approximately 70-year-old humans) to levels comparable to young adults on the Morris water maze, a standard test of spatial learning and memory. This finding was particularly striking because the memory deficits of aging are generally considered more clinically relevant than scopolamine-induced amnesia. Cognitive benefits persisted for several days after the last dose, suggesting structural changes (new synapses) rather than transient pharmacological effects. If the drug only modulated acute neurotransmission, the effects would have dissipated within hours of the last dose.
The orally available, blood-brain barrier-permeable properties of dihexa made it unusual among peptide-derived compounds, which typically cannot cross the blood-brain barrier due to their size, polarity, and susceptibility to efflux transporters. Dihexa is a modified hexapeptide fragment (technically a peptidomimetic) with a molecular weight of 451 Da, small enough to passively diffuse across the BBB. This is in contrast to full-length HGF (molecular weight 83,000 Da), which cannot reach the brain after peripheral administration. These properties made dihexa an attractive drug development candidate: a small, orally available molecule that could activate a neurotrophic pathway previously accessible only through direct brain injection.
The HGF/c-Met System in the Brain
Hepatocyte growth factor modulation in the brain is a less-explored therapeutic target compared to the cholinergic, amyloid, and tau pathways that dominate Alzheimer's research. HGF and its receptor c-Met are expressed in neurons, astrocytes, and oligodendrocytes throughout the adult brain. The system promotes neuronal survival, neurite outgrowth, and synaptic plasticity under normal conditions.
In Alzheimer's disease, HGF levels in the brain and cerebrospinal fluid decline with disease progression. Postmortem studies show reduced HGF protein and c-Met receptor expression in the hippocampus and entorhinal cortex, the regions most severely affected by Alzheimer's pathology. The decline in HGF signaling correlates with synaptic loss, suggesting a causal link between HGF pathway failure and cognitive decline.
The hypothesis behind dihexa therapy was that augmenting residual HGF signaling could restore synaptogenesis capacity in degenerating neural circuits, potentially compensating for synaptic loss even without removing amyloid plaques or tau tangles. This hypothesis was particularly appealing because it offered a mechanism of action independent of the amyloid and tau pathways, which have dominated Alzheimer's drug development for three decades with a failure rate exceeding 99%.
The HGF/c-Met system also has roles outside the brain: in liver regeneration, wound healing, and tumor biology. C-Met is a well-known oncogene, and drugs that activate c-Met in the brain could theoretically promote tumor growth elsewhere. This safety concern has followed the dihexa development program from the beginning. The clinical trial data from fosgonimeton did not reveal cancer signals in the 26-week observation period, but 26 weeks is insufficient to rule out long-term oncogenic risk from chronic c-Met activation. This concern is one reason the regulatory bar for an HGF-modulating Alzheimer's therapy would be high: the therapy would need to demonstrate substantial cognitive benefit to justify even a theoretical cancer risk in a population that would need to take the drug for years.
This hypothesis was compelling because synaptic loss, not amyloid plaque burden, is the strongest histological correlate of cognitive decline in Alzheimer's disease. A therapy that rebuilds synapses could theoretically improve cognition regardless of whether the underlying amyloid and tau pathology is addressed.
The Fosgonimeton Clinical Trials
Athira Pharma (now rebranded as Athira Pharma) developed fosgonimeton as a subcutaneous injectable prodrug of dihexa. The phosphate group is cleaved after injection, releasing active dihexa. Three clinical trials tested the compound:
ACT-AD Phase 2 (2022)
The exploratory ACT-AD study tested fosgonimeton in 77 patients with mild-to-moderate Alzheimer's disease over 26 weeks. While the study was not powered to detect clinical cognitive improvements (its primary purpose was safety and biomarker assessment), it provided the first human pharmacodynamic data for the HGF/c-Met modulation approach. Blood biomarkers showed changes consistent with HGF pathway engagement, and the compound was reported as generally well-tolerated. These results were presented at the Alzheimer's Association International Conference in 2022 and were considered sufficient to advance to a larger pivotal trial.
LIFT-AD Phase 2/3 (2024)
The pivotal LIFT-AD trial enrolled 302 patients with mild-to-moderate Alzheimer's disease in a 26-week, randomized, placebo-controlled design. The primary endpoint was the Global Statistical Test (GST), combining cognition (measured by ADAS-Cog) and function (measured by ADCS-ADL) into a single composite score.
Results: Fosgonimeton improved the GST by 0.08 points over placebo. This effect was not statistically significant (p = 0.7), meaning the difference was indistinguishable from random variation. Key secondary endpoints, including ADAS-Cog (cognition) and ADCS-ADL (daily function) individually, also did not reach significance. The magnitude of the primary effect (0.08 points on the composite) was so small that even doubling or tripling the sample size would likely not have changed the conclusion. This was not a near-miss; it was a clear failure on cognitive and functional endpoints.
The biomarker data told a different story. Patients receiving fosgonimeton showed statistically significant reductions in phosphorylated tau217 compared to placebo (p < 0.01). Phosphorylated tau217 is considered one of the most accurate blood-based biomarkers of Alzheimer's pathology. The drug appeared to be doing something biologically, but whatever it was doing did not translate into measurable cognitive or functional improvement over 26 weeks.
Subgroup analyses suggested fosgonimeton might have greater effect in patients with more advanced disease and in APOE e4 carriers (a genetic risk factor for Alzheimer's). These exploratory findings have not been confirmed in a dedicated trial.
SHAPE Phase 2 (Parkinson's Disease Dementia)
A separate Phase 2 trial tested fosgonimeton in Parkinson's disease dementia (PDD) and dementia with Lewy bodies (DLB). This trial showed more encouraging results, with improvements in attention and processing speed, cognitive domains that are more prominently affected in Lewy body dementias than in Alzheimer's. The different outcome may reflect the distinct pathophysiology: Lewy body dementias involve more prominent synaptic dysfunction in cortical cholinergic circuits that may be more responsive to HGF modulation than the hippocampal circuits primarily affected in Alzheimer's. The PDD/DLB population also tends to have less severe neuronal loss at the time of diagnosis, potentially providing a more favorable substrate for synaptogenesis.
Whether Athira will pursue the PDD/DLB indication further remains unclear. The company's stock price dropped substantially after the LIFT-AD failure, and the financial viability of pursuing a narrower indication in a smaller patient population is uncertain. The SHAPE results, while encouraging, came from a small study and require confirmation in a larger trial.
Why Synaptogenesis in Rodents Did Not Predict Human Outcomes
The disconnect between robust preclinical cognitive improvement and failed clinical efficacy is not unique to dihexa. It is the central problem of Alzheimer's drug development, where over 99% of compounds that work in animal models fail in human trials.
Several factors likely contributed to the translation failure:
The rat models do not reproduce Alzheimer's pathology. Aged rats have age-related memory decline but do not develop amyloid plaques or tau tangles. Scopolamine-induced amnesia models disrupt a single neurotransmitter system. Neither model captures the progressive neurodegeneration, neuroinflammation, and vascular dysfunction of human Alzheimer's disease. A drug that restores synapses in a brain with intact neurons but reduced synaptic activity (the rodent situation) may not work in a brain where neurons are dying and the inflammatory environment actively degrades new connections (the human Alzheimer's situation).
26 weeks may be insufficient. Synaptogenesis is a structural process. Even if dihexa promotes new synapse formation in human brains, the functional integration of those synapses into cognitive circuits takes time. The LIFT-AD trial's 26-week duration may have been too short to detect a structural effect that manifests as cognitive improvement only over 12-18 months.
Dose and exposure may have been wrong. Converting picomolar in vitro activity to an appropriate human dose is notoriously difficult. The relationship between plasma levels, CSF penetration, and brain tissue concentrations is poorly characterized for most peptide-derived compounds. Fosgonimeton was administered subcutaneously, and while dihexa is known to cross the blood-brain barrier in rodents, the degree of brain exposure achieved in human patients at the doses used in LIFT-AD was not directly measured. It is possible that the subcutaneous dosing regimen simply did not deliver enough active compound to the hippocampus and cortex to replicate the synaptic effects seen at picomolar concentrations in cell culture.
The disease may be too advanced. By the time Alzheimer's is clinically diagnosed as mild-to-moderate, substantial neuronal loss has already occurred. The average Alzheimer's patient has lost roughly 30-40% of hippocampal volume by the time of clinical diagnosis. Promoting synaptogenesis between surviving neurons may not compensate for the loss of entire neural populations. The subgroup finding of greater effect in more advanced disease is paradoxical and may be a statistical artifact of small subgroup sizes.
Synaptogenesis alone may not be sufficient. Even if new synapses form, they must be integrated into functional circuits, supported by adequate blood supply, and protected from the ongoing inflammatory and oxidative processes that characterize the Alzheimer's brain. The amyloid and tau pathology that drives neurodegeneration continues regardless of whether new synapses are being formed. Building new connections in a brain that is actively destroying them may be analogous to constructing a building during an earthquake: the structural work is real, but the environment undermines it.
The Broader Synaptogenesis Peptide Landscape
Despite the fosgonimeton setback, peptide-based synaptogenesis remains an active research area. Several approaches are being pursued:
Ma et al. (2025) demonstrated that a dual GLP-1/CCK receptor agonist improved cognitive performance and promoted synaptogenesis in the 5xFAD Alzheimer's mouse model, one of the most widely used transgenic models of Alzheimer's pathology. The 5xFAD model develops amyloid plaques, synaptic loss, and neuroinflammation, making it a more stringent test than the aged rat models used in dihexa research. The dual agonist reduced amyloid burden and improved synaptic markers, suggesting that incretin-based peptides may promote synaptic repair while simultaneously addressing amyloid pathology. This multi-target approach contrasts with dihexa's single-target HGF modulation and may be more appropriate for a disease with as many contributing factors as Alzheimer's.[1]
Verma et al. (2025) reviewed the therapeutic potential of small peptides in Alzheimer's disease, cataloguing advances in memory restoration and targeted delivery systems. The review highlights that multiple peptide scaffolds are under investigation for cognitive enhancement, from BDNF-modulating peptides to mitochondria-targeted sequences.[2]
Pahal et al. (2026) reviewed brain peptides in Alzheimer's pathophysiology and therapeutic development, noting that the field has shifted from single-target approaches (like anti-amyloid antibodies) toward multi-target peptide strategies that address synaptic loss, neuroinflammation, and metabolic dysfunction simultaneously.[3]
Das et al. (2026) reviewed protein and peptide-based nanotherapeutics for Alzheimer's management, emphasizing that delivery across the blood-brain barrier remains the primary challenge. Dihexa's ability to cross the BBB orally was one of its most unusual properties, and achieving this with other neuroprotective peptides requires nanoparticle encapsulation or receptor-mediated transcytosis.[4]
Rehra et al. (2026) demonstrated brain delivery of a neurotrophic peptide derived from secreted amyloid precursor protein (APPsalpha) as a therapeutic strategy for Alzheimer's. This approach targets the same synaptic repair goal as dihexa but through a different molecular pathway, leveraging a naturally occurring neuroprotective fragment of the amyloid precursor protein.[5]
Hanak et al. (2025) found that inhibiting the angiotensin-converting enzyme N-terminal catalytic domain prevents endogenous opioid degradation in the brain, connecting the angiotensin system (dihexa's origin as an angiotensin IV analog) to broader neuropeptide signaling networks.[6]
Ugalde-Trivino et al. (2025) developed a brain-accessible peptide targeting BDNF-receptor TrkB-T1, modulating stroke inflammatory response and neurotoxicity. This work illustrates how neurotrophic peptide research extends beyond Alzheimer's to stroke, traumatic brain injury, and other conditions where synaptic repair is needed.[7]
Qian et al. (2024) created a hybrid peptide fusing cell-penetrating sequences with neuroprotective domains (SS31 and Humanin), achieving amplified multimodal neuroprotection. The fusion design approach may represent the next generation of brain-targeted peptide therapeutics.[8]
Ding et al. (2021) reviewed the role of mitochondria-targeted peptide SS-31 in both diabetes mellitus and Alzheimer's disease, highlighting the convergence of metabolic and neurodegenerative pathology that peptide therapeutics are uniquely positioned to address. Mitochondrial dysfunction is a feature of both conditions, and SS-31's ability to stabilize mitochondrial membranes and reduce oxidative stress may protect synapses from the energy failure that contributes to their loss in Alzheimer's. This metabolic approach to synaptic protection represents a complementary strategy to dihexa's growth-factor-based synaptogenesis promotion.[9]
The diversity of these approaches reflects a maturing understanding of synaptic loss in Alzheimer's disease. It is not a single-pathway problem. Synapses are lost due to amyloid toxicity, tau pathology, neuroinflammation, vascular insufficiency, mitochondrial dysfunction, and growth factor deprivation. A therapy that addresses only one of these drivers, as dihexa attempted to do with HGF augmentation, may be insufficient to produce measurable cognitive improvement in a disease where all six processes are operating simultaneously. The next generation of synaptogenesis-promoting peptides will likely need to address multiple drivers or be combined with other therapies that cover the remaining pathological mechanisms.
The Gray Market Problem
Dihexa is widely available as a "research chemical" from peptide suppliers. It is not FDA-approved for any indication and has no established human safety profile beyond the clinical trial data from fosgonimeton. The failed LIFT-AD trial means there is no evidence that dihexa improves cognition in humans. Despite this, online forums and peptide vendors market dihexa for cognitive enhancement, often citing the preclinical rodent data without mentioning the clinical trial failure.
The potency claims that circulate about dihexa are based on comparing in vitro active concentrations (picomolar for dihexa vs micromolar for BDNF), which is not how potency is meaningfully assessed. Two molecules that work through completely different mechanisms, different receptors, and different downstream pathways cannot be compared by concentration alone. The fact that dihexa is active at 10 picomolar while BDNF requires nanomolar concentrations reflects the different binding kinetics and receptor systems, not a meaningful comparison of therapeutic potency.
The failed LIFT-AD trial should temper expectations from the gray market. The compound that was described as "millions of times more potent than BDNF" in online marketing failed to improve cognition in the most rigorous human test conducted to date. This does not mean the molecule is biologically inert. The phospho-tau biomarker data suggests it was doing something in the brain. But the clinical outcome that matters, whether patients think and function better, was not achieved.
The Bottom Line
Dihexa's preclinical data on synaptogenesis through HGF/c-Met modulation was among the most compelling in peptide neuroscience. In aged rats, it restored cognitive performance to young-adult levels. But fosgonimeton, its clinical prodrug, failed the phase 2/3 LIFT-AD trial in Alzheimer's patients: no significant cognitive or functional improvement over placebo at 26 weeks, despite biomarker evidence of biological activity (reduced phospho-tau217). The translation failure reflects the broader challenge of Alzheimer's drug development, where rodent cognitive models consistently fail to predict human outcomes. Synaptogenesis-promoting peptides remain under active investigation through multiple pathways, but the HGF/c-Met approach has yet to demonstrate clinical efficacy.