Davunetide has been studied in patients, but it is not a validated nootropic for healthy adults. Its large trial in progressive supranuclear palsy, or PSP, showed no benefit on its primary endpoints. Subsequent sex-based reanalyses report favorable signals, but are not new confirmatory trials.
The issue, then, is not choosing between commercial enthusiasm and outright rejection. You need to distinguish between the primary result, secondary analysis, exploratory subgroup and new independent evidence. These categories do not become interchangeable just because a publication’s title sounds enthusiastic.
This audit, with a cutoff date of October 17, 2026, compares the trials with later publications, including a PSP reanalysis published in July 2026. It offers no dose, intranasal administration instructions or purchasing advice.
NAP, AL-108 and davunetide: identifying the molecule
Davunetide, davunetide in English-language publications, is an eight-amino-acid peptide derived from ADNP, short for activity-dependent neuroprotective protein. NAP refers to this fragment, also studied under the codes AL-108 and CP201. The code AL-208 was used for its intravenous development, particularly in cardiac surgery.
In preclinical research, it acts on interactions within the neuronal cytoskeleton, particularly those involving microtubules and tau protein. Microtubules help organize cells and transport materials within them. Tau plays a role in their function and may be involved in neurodegenerative diseases.
This biological plausibility explains its clinical development. It does not demonstrate that a healthy adult will remember things better. An interesting target and a measurable benefit are two different steps. The 2026 research describes these mechanisms and the hypothesis of sex-based differences, but its clinical data still come from patients with PSP.
The neurological product Cerebrolysin is a different product · FR, consisting of a mixture of peptides. Its results, risks and routes of administration must not be attributed to davunetide. The word “neuropeptide” does not establish therapeutic equivalence.
Trial table: population and endpoint before the result
The question asked at the outset matters as much as the result highlighted afterward. The table separates the original trials from their reanalyses:
| Publication and population | Original question or endpoint | Result and interpretive limitation |
|---|---|---|
| 2012, 63 patients with schizophrenia, 12 weeks | Cognition measured using the MCCB battery | No significant difference on the MCCB. Signal on UPSA functional capacity, a secondary endpoint in the original publication |
| 2013, 144 people with amnestic mild cognitive impairment, 12 weeks | Composite cognitive memory score | No significant difference on the composite. Signals on two individual memory and attention tasks |
| 2014, 313 patients with PSP, 52 weeks | Two primary endpoints: PSPRS and SEADL | No demonstrated benefit in the primary analysis. Secondary and exploratory measures also show no overall effect |
| 2023, reanalysis of the same PSP trial | Results separated by sex | Favorable signals in women on some outcomes. Retrospective analysis, with no new randomization |
| 2024, reanalysis of the mild cognitive impairment trial | Cognitive tests and anxiety by sex | Signals vary by task and group. The original composite is not used in this reanalysis |
| 2026, new reanalysis of the PSP trial | Ten-item PSPRS scale, longitudinal models and cognition | Favorable interactions and signals reported in women. Same original trial, with endpoints and analyses reworked after the fact |
Sources: Javitt et al., 2012, Morimoto et al., 2013, Boxer et al., 2014, Gozes et al., 2023, Gozes et al., 2024 and Shapira et al., 2026.
The PSP trial’s primary result remains negative
The multicenter PSP trial assessed disease progression and independence. The PSPRS measures clinical severity, while the SEADL assesses daily activities. The lack of benefit therefore concerned more than just a laboratory marker.
The primary analysis included evaluable patients using an intention-to-treat approach, with a method accounting for deaths and loss to follow-up. It showed no difference between groups. An analysis of participants who completed follow-up found none either.
This result weighs against efficacy in the population and conditions tested. It does not prove an absence of effect in every disease, formulation or imaginable subgroup. But a hypothesis that emerged later cannot be presented as the successful primary objective of that trial.
Sex-based analyses: a signal that needs confirmation
Analyzing differences between women and men is a legitimate scientific question. What changes the strength of a conclusion is how the analysis was planned and tested.
PSP: from 2023 to 2026, the same data reexamined
The 2023 publication revisits the PSP trial data and reports favorable results in women, particularly for independence and certain motor domains. Its title claims efficacy, but its design remains a reanalysis conducted after the overall result was known.
The July 2026 publication explores this possibility further. It uses, among other approaches, a ten-item version of the PSPRS and models accounting for time, sex and treatment. It reports favorable interactions as well as cognitive signals in women.
It would therefore be inaccurate to say these authors studied only separate results without interaction testing. However, an interaction test obtained in a retrospective analysis does not turn it into a prespecified confirmatory trial. The authors acknowledge the post hoc nature of their work.
The two publications do not represent two new treated cohorts. They explore the same original trial using different questions and methods. Their value lies in preparing a prospective study capable of confirming or refuting the hypothesis.
Mild cognitive impairment: the composite remains part of the evidence
The 2024 reanalysis concerns the 144 participants in the amnestic mild cognitive impairment trial. It describes different signals by sex, test and treatment group. The authors state that they did not reuse the original composite score, citing patient variability and the difficulty of certain tests, among other reasons.
This decision must remain visible: a selection of positive tasks does not replace a nonsignificant original composite. Statistical corrections are described for some comparisons, but that does not make the full set of retrospective analytical choices equivalent to a confirmatory plan established in advance.
The title refers to prodromal Alzheimer’s disease. The inclusion criteria described are based on amnestic mild cognitive impairment, with no requirement for amyloid or tau confirmation. This population therefore cannot be recast as a cohort with biologically confirmed Alzheimer’s disease, nor can we conclude that the treatment prevents dementia.
Evidence timeline
New papers do not mean new trials
- 2013–2014: original trialsMild cognitive impairment and PSP: no significant benefit on primary endpoints.
- 2023: PSP reanalysisSex-based hypotheses and results from the existing trial.
- 2024: cognitive reanalysisAlternative readings of tests from the mild cognitive impairment trial.
- 2026: PSP reanalysisNew models and a reworked scale, with no new confirmatory cohort.
Development ties matter too
The original PSP trial was funded by Allon Therapeutics. The reanalyses are led in part by davunetide’s inventor, who is involved in its development. The 2026 publication discloses patent protection and development or advisory roles at ExoNavis for some authors.
These ties do not invalidate their results. They make independent prospective confirmation particularly valuable, with endpoints, subgroups and analytical methods announced before access to the results.
What the word “subgroup” should prompt you to check
A significant effect in one group and a nonsignificant effect in the other do not, on their own, demonstrate a difference between them. That difference must be tested directly. Its robustness and clinical relevance must then be examined.
When reading a new announcement, keep these questions in mind:
- Was the subgroup planned before the analysis began?
- Are sample sizes and missing data reported?
- Is an interaction tested, with uncertainty reported?
- How many tests, time points and models were explored?
- Does the result concern independence, a cognitive test or only a biomarker?
- Has a new cohort confirmed the same hypothesis?
These questions are not meant to dismiss a signal automatically. They prevent a lead worth testing from being turned into an established benefit. A p value does not tell you how much someone’s everyday memory will improve.
Cardiac surgery and healthy adults: different contexts
A 2025 publication on coronary artery bypass surgery reanalyzes a trial involving AL-208, the intravenous formulation. It reports sex-based signals in markers and certain memory measures. The original study showed no overall cognitive difference, and the small number of women limits the reanalysis.
This hospital setting addresses a surgery-related risk. It represents neither cognitive enhancement in healthy adults nor validation of a spray. The trial registry helps keep that indication front and center when interpreting its results.
No demonstration of a lasting cognitive gain in healthy adults was identified in the trials reviewed. The existence of human studies is a step forward from a promise based only on animal research. It does not justify changing the population studied when drawing the conclusion.
Tolerability and unknowns: what “well tolerated” means
In the PSP trial, nasal events were more frequent with davunetide. Adverse events led to discontinuation of the product in 21 participants on davunetide versus 12 on placebo, mainly in connection with nosebleeds or congestion. There were 11 and 10 deaths respectively, in a population with a serious disease: these numbers alone do not establish drug-related causality.
These data document the tolerability of a trial formulation under monitoring. They do not provide the frequency of risks associated with a product bought online. Identity, impurities, storage and microbiological quality raise additional questions.
Interactions, pregnancy, breastfeeding and prolonged use in healthy people lack a validated framework in this review. There is therefore no list of precautions that can make self-experimentation “safe.” If memory problems persist, worsen or interfere with your independence, seek a medical assessment rather than trying a peptide.
Status in France: orphan designation and marketing authorization are distinct
The situation must be assessed by indication and date:
- PSP: European designation EU/3/10/728, granted in 2010, was withdrawn from the register in April 2013 at the sponsor’s request. The revised EMA document specifies this.
- ADNP syndrome: another designation, EU/3/24/3007, was granted on December 13, 2024. The EMA entry explicitly states that an orphan designation is not an authorization for use.
In the French medicinal product and composition files dated September 29, 2026, and the EMA medicines export downloaded on October 17, 2026, no entry matching davunetide was identified under the names checked. Sources: French Public Medicines Database and EMA export.
This check does not constitute an exhaustive opinion on all exceptional access pathways or import rules. It does not identify a French marketing authorization for nootropic use. A rare disease designation, a registered trial and an online sale are not three ways of saying “authorized.”
The Verdict
Main sources
The full texts of the PSP trial and the 2023, 2024, 2025 and 2026 reanalyses were reviewed. For the original 2012 and 2013 publications, the reported conclusions are based on their verified abstracts, supplemented by the methods described in the reanalyses:
