evipedia.ai
← Back

NSI-189 for Health & Longevity

Evidence Review created on 09/04/2026 using AI4L / Opus 5

Also known as: Amdiglurax, ALTO-100, NSI-189 Phosphate

Motivation

NSI-189 (also known as amdiglurax or ALTO-100) is an experimental oral compound designed to do something no marketed antidepressant does directly: prompt the brain to grow new nerve cells in the region most closely tied to memory and mood. It does not act on serotonin or the other chemical messengers that conventional mood medications target.

Interest in it grew out of a long-standing idea that when this brain region loses its capacity to renew itself, low mood and the mental fog that often accompanies it follow. Early human testing suggested improvements in both mood and thinking, and the compound was taken by mouth and well tolerated. On that strength it moved from a laboratory candidate into wide unregulated sale to people experimenting on themselves, years before its testing was finished.

This review examines what the human and animal evidence shows about NSI-189: how it is thought to work, which of its claimed effects have held up under placebo-controlled testing and which have not, what is known about its safety and the doses used, and where the uncertainty still lies.

Benefits - Risks - Protocol - Conclusion

A short list of sources that give a substantive overview of NSI-189, its clinical testing, and its preclinical rationale.

Note on this list: no content on NSI-189 exists on any of the priority expert platforms — FoundMyFitness, Peter Attia, Huberman Lab, Chris Kresser, Life Extension or Lifespan.io — under either web search or the sites’ own search functions; none of them has covered the compound. The five items are drawn instead from the clinical trial record, the mechanistic literature and the specialist trade press; the remaining candidates were either a duplicate record, a re-analysis, or a further paper by an author already listed here, and the rest of the available material is vendor copy or community write-ups that do not meet the quality bar.

Grokipedia

Amdiglurax

The dedicated page for the compound, filed under its international nonproprietary name, covering chemistry, pharmacology, the Neuralstem and Alto Neuroscience development history, and the full clinical trial record.

Examine

No Examine article on NSI-189 exists; a direct search of examine.com returns no results for the compound. Examine covers dietary supplements and food-derived ingredients and does not cover unapproved investigational drugs such as NSI-189.

ConsumerLab

No ConsumerLab article on NSI-189 exists; a direct search of consumerlab.com returns only unrelated supplement-survey news releases. ConsumerLab tests marketed dietary supplements and does not cover unapproved investigational drugs such as NSI-189.

Systematic Reviews

No systematic reviews or meta-analyses for NSI-189 were found on PubMed as of September 4, 2026.

Neither side of the trade-off is represented: there is no systematic review or meta-analysis of the claimed antidepressant and pro-cognitive effect, and none of the principal risk, which is the unknown consequence of sustained stimulation of new nerve cell growth.

Mechanism of Action

NSI-189 is a benzylpiperazine-aminopyridine, chemically (4-benzylpiperazin-1-yl)-[2-(3-methylbutylamino)pyridin-3-yl]methanone. It was picked from a screen for molecules that make human hippocampus-derived neural stem cells divide, and, as an expert-opinion review summarises, it raises neurogenesis (the birth of new neurons) in the mouse hippocampus after oral dosing.

Its action is monoamine-independent: it does not block the serotonin, noradrenaline or dopamine transporters that conventional antidepressants act on. Downstream it activates TrkB (the docking receptor for brain-derived neurotrophic factor, or BDNF, a protein that keeps neurons alive) and the Akt cell-survival pathway, and it enlarges long-term potentiation (LTP, the strengthening of connections between neurons that underlies learning) in hippocampal slices. In diabetic rodents it also restores AMP-activated protein kinase (AMPK, the cell’s energy sensor) activity in sensory nerves.

The direct binding target is unidentified, and NSI-189 does not bind TrkB itself, so the neurotrophic signalling is indirect. The competing mechanistic reading is that there is no drug effect to explain: both trials powered for efficacy missed their primary endpoints, and hippocampal volume did not rise significantly on magnetic resonance imaging (MRI).

Measured in people: peak plasma level 1–2 hours after an oral dose; half-life 17.4–20.5 hours; steady state at 96–120 hours; near dose-proportional exposure from 40 to 120 mg/day; no difference between men and women. It enters the brain and its effects concentrate in the hippocampus. Its metabolic route and the cytochrome P450 liver enzymes (which break down most drugs) involved are unpublished.

Historical Context & Evolution

NSI-189 was not conceived as a longevity compound. It came out of Neuralstem, a company built on human neural stem cell technology, where chief scientific officer Karl Johe ran a phenotypic screen for small molecules that would make human hippocampus-derived neural stem cells proliferate. The intended use was major depressive disorder, on the then-ascendant hypothesis that conventional antidepressants work in part by restoring hippocampal neurogenesis, and that a molecule doing this directly might work faster, more durably, and in people whom serotonin-based agents fail.

The first human evidence, a 24-participant inpatient study published in 2016, reported medium-to-large effect sizes on every depression measure and no serious adverse events, with the improvement still present eight weeks after dosing stopped. That durability claim, unusual for an antidepressant, is what pulled the compound into unregulated sale as a nootropic well before the larger trial read out.

The 220-participant Phase 2 trial then missed its clinician-rated primary endpoint while separating from placebo on three self-rated scales. Neuralstem became Seneca Biopharma and then Palisade Bio; the molecule was licensed to Alto Neuroscience, renamed ALTO-100, and re-tested with a verbal-memory biomarker used to pre-select likely responders. That 301-participant trial also missed its primary endpoint in 2024. The compound has since received the international nonproprietary name amdiglurax and remains in testing for bipolar depression.

Expected Benefits

High 🟩 🟩 🟩

No benefit reaches High: the qualifying class of evidence would be the validated clinician-rated depression scale that both trials chose as their primary measure separating from placebo in more than one trial, and neither trial achieved it.

Medium 🟩 🟩

No benefit reaches Medium either: the qualifying class of evidence would be that same validated clinician-rated endpoint separating from placebo in a single trial without contradiction, and every positive human result for NSI-189 is instead a secondary self-rated or exploratory measure that the primary endpoint in the same trial contradicts.

Low 🟩

Reduction in Self-Reported Depressive Symptoms ⚠️ Conflicted

Two placebo-controlled trials found NSI-189 separated from placebo on self-rated depression scales — the first across pooled doses as an exploratory measure, the second at 40 mg/day alongside a missed clinician-rated primary endpoint. A third, larger trial missed that endpoint. Net reading: the effect is unconfirmed and may be self-report artefact.

Magnitude: In the 220-participant Phase 2 trial, 40 mg/day reduced Symptoms of Depression Questionnaire (SDQ, a 44-item self-report depression scale) scores by 8.2 points more than placebo (95% CI — confidence interval, the range the true effect most likely falls in — −16.2 to −0.2; p = 0.044, p being the chance of a difference this large if the compound did nothing), while the difference on the clinician-rated Montgomery-Åsberg Depression Rating Scale (MADRS, a 10-item scale a clinician scores for depression severity) was 1.8 points and not significant (p = 0.224). The 301-participant Phase 2b trial (NCT05712187) showed no separation on that same scale, according to a news report of its topline results.

Improvement in Cognitive Function ⚠️ Conflicted

Both trials showed better self-rated mental clarity and energy on NSI-189, and the Phase 2 trial found gains on one objective cognitive battery but none on a second run in the same participants. Net reading: a possible pro-cognitive effect that one of two objective tests failed to detect.

Magnitude: Cognitive and Physical Functioning Questionnaire (CPFQ, a seven-item self-report scale of memory, attention and mental energy) scores fell 1.9 points further than placebo at 40 mg/day (95% CI −3.7 to −0.1; p = 0.035), and on the CogScreen delayed-recall subtest the second-stage effect size reached Cohen’s d 1.12 (a standardised effect size where 0.2 is small and 0.8 large; p = 0.002), while every Cogstate measure was null, in the Phase 2 trial.

Speculative 🟨

Stimulation of Hippocampal Neurogenesis and Synaptic Plasticity

Basis is preclinical only: NSI-189 raises new-neuron formation in mouse hippocampus and enlarges long-term potentiation in hippocampal slices. Human imaging in the Phase 1b trial found no significant hippocampal volume change.

Reversal of Diabetic Peripheral Neuropathy

Basis is animal work only: oral NSI-189 prevented and reversed nerve conduction, sensory and memory deficits in type 1 and type 2 diabetic mice. No human trial has tested this.

Functional Recovery After Ischemic Stroke

Basis is a single rat study: dosing started six hours after artery occlusion improved motor and neurological scores for 24 weeks and increased nerve-fibre growth in the hippocampus. No human data exist.

Protection Against Radiation-Induced Cognitive Decline

Basis is one irradiated-rat study: four weeks of oral NSI-189 restored performance on four memory tasks, raised neurogenesis and lowered brain immune-cell activation. No human data exist.

Recovery of Motor and Cognitive Function in Angelman Syndrome

Basis is one mouse study: injected NSI-189 reversed motor and cognitive impairment in a model of Angelman syndrome (a rare genetic disorder causing severe developmental delay). No human data exist.

Benefit-Modifying Factors

  • Baseline depression severity: a developer-led post-hoc reanalysis found the signal concentrated in participants scoring below 30 on the clinician-rated depression scale and absent above it. Post-hoc, unreplicated, and contradicted by the later trial.
  • Baseline verbal memory: poor word-list recall was the pre-specified biomarker used to enrich the 2024 trial for responders, on the reasoning that it indexes hippocampal plasticity. That enrichment did not produce separation from placebo.
  • Baseline blood biomarkers: in the Phase 1b trial, baseline BDNF, epidermal growth factor, myeloperoxidase (an enzyme immune cells release during inflammation), tumour necrosis factor receptor 2 and alpha-1 antitrypsin predicted response in 17 of 18 participants. A model built on 18 people.
  • Sex: no sex difference in exposure was detectable in the Phase 1b pharmacokinetics, and neither trial reported a sex-by-treatment interaction. Women were 56–67% of Phase 2 participants, so an effect confined to men would have been hard to see.
  • Age: the two published trials capped enrolment at 60, and the later ALTO-100 trials that allowed up to 64–70 published no age breakdown. Hippocampal volume and neurogenic capacity both decline with age, so response above 60 is effectively untested.
  • Pre-existing conditions: the strongest preclinical signals are in diabetic and irradiated animals, where the deficit being corrected is large. Whether a metabolically healthy person has any comparable deficit to correct is unknown.
  • Genetic polymorphisms: no pharmacogenetic data exist. Variants plausibly relevant — BDNF Val66Met (which lowers activity-dependent BDNF release) and COMT Val158Met (which sets prefrontal dopamine clearance) — have never been tested against NSI-189 response.

Potential Risks & Side Effects

High 🟥 🟥 🟥

Headache

Headache was the most frequently recorded treatment-related event in the Phase 1b trial and remained among the most common treatment-related events in the 301-participant Phase 2b trial of the same molecule. Its incidence was identical on drug and on placebo in the first trial and similar to placebo in the second, so it is common in this setting but not clearly caused by the compound. All cases were mild and none led to withdrawal.

Magnitude: 9 of 18 participants on NSI-189 (50%) versus 3 of 6 on placebo (50%) in the Phase 1b trial; reported among the most common treatment-related events at rates similar to placebo in the Phase 2b trial (NCT05712187), per a news report of its topline results.

Nausea

Nausea appeared only in the highest Phase 1b cohort, which took 120 mg/day as three divided doses, and in none of the placebo participants, so it looks dose- or peak-concentration-related rather than incidental. It was also among the most common treatment-related events in the Phase 2b trial. No participant discontinued for it and no serious gastrointestinal event occurred in any trial.

Magnitude: 2 of 18 pooled participants on NSI-189 (11.1%) versus 0 of 6 on placebo in the Phase 1b trial, concentrated entirely in the 120 mg/day cohort at 2 of 6 (33.3%); among the most common treatment-related events in the Phase 2b trial (NCT05712187), per a news report of its topline results.

Sleep Disruption and Vivid Dreams

Insomnia and nightmares or unusually vivid dreams each affected about one in six participants on NSI-189 in the Phase 1b trial and none on placebo, and abnormal dreams were again among the most common treatment-related events in the Phase 2b trial. The pattern fits a long-half-life compound taken in divided doses through the day. All reports were reversible on stopping and none required treatment.

Magnitude: insomnia 3 of 18 (16.7%) and nightmare or vivid dream 3 of 18 (16.7%) on NSI-189 versus 0 of 6 on placebo, with poor sleep quality in a further 1 of 18 (5.6%), in the Phase 1b trial.

Medium 🟥 🟥

Dizziness rose steeply with dose in the Phase 1b trial, affecting two thirds of the highest-dose cohort and none of the once-daily cohort. It is the clearest dose-dependent signal anywhere in the human record and is consistent with a peak-concentration effect, which is the stated reason the developer moved to divided dosing at all. It was recorded in one trial only, every case was mild, and none caused withdrawal.

Magnitude: 5 of 18 pooled participants on NSI-189 (27.8%) versus 1 of 6 on placebo (16.7%), rising from 0 of 6 at 40 mg/day to 4 of 6 (66.7%) at 120 mg/day, in the Phase 1b trial.

Daytime Somnolence

Somnolence (daytime sleepiness) was reported by more than a quarter of participants on NSI-189 in the Phase 1b trial, most often in the once-daily cohort, alongside one report of fatigue. It is the adverse event most likely to matter practically, since it bears on driving and on operating machinery within a few hours of a dose. It was recorded in one trial only and placebo participants also reported it.

Magnitude: 5 of 18 (27.8%) on NSI-189 versus 1 of 6 (16.7%) on placebo, highest in the 40 mg once-daily cohort at 3 of 6 (50%), in the Phase 1b trial.

Paresthesia

Two participants on NSI-189 reported paresthesia (tingling or pins-and-needles sensations) and none on placebo. Two events is too few to interpret, and no later trial has reported it.

Magnitude: 2 of 18 (11.1%) on NSI-189 versus 0 of 6 on placebo in the Phase 1b trial.

Dry Mouth

Two participants on NSI-189 reported dry mouth and none on placebo, both of them in the twice-daily cohort. Two events is too few to interpret, and no later trial has reported it.

Magnitude: 2 of 18 (11.1%) on NSI-189 versus 0 of 6 on placebo in the Phase 1b trial, concentrated in the 40 mg twice-daily cohort at 2 of 6 (33.3%).

Palpitation

One participant on NSI-189 reported palpitation and none on placebo, in the once-daily cohort. One event is too few to interpret, and no later trial has reported it. Cardiac safety is otherwise uncharacterised, which is why every trial ran electrocardiograms.

Magnitude: 1 of 18 (5.6%) on NSI-189 versus 0 of 6 on placebo in the Phase 1b trial, confined to the 40 mg once-daily cohort at 1 of 6 (16.7%).

Rash and Skin Discomfort

One participant on NSI-189 reported a rash and a second reported skin pain, with neither event occurring on placebo. Both were isolated single reports, both were mild, neither led to withdrawal, and no later trial has reported a skin event.

Magnitude: rash 1 of 18 (5.6%) and skin pain 1 of 18 (5.6%) on NSI-189 versus 0 of 6 on placebo in the Phase 1b trial.

Low 🟥

Restlessness and Irritability ⚠️ Conflicted

Isolated Phase 1b reports of restlessness and irritability sit against unstructured accounts from unsupervised users describing early anxiety as the main drawback, and against both trials, where fewer people stopped on drug than on placebo. Net reading: an early activation effect is plausible but unproven.

Magnitude: restlessness 1 of 18 (5.6%) and irritability 1 of 18 (5.6%) on NSI-189 versus 0 of 6 on placebo in the Phase 1b trial; in the Phase 2 trial overall discontinuation across the first six weeks ran 18.9% on placebo versus 9.1% at 40 mg/day and 2.3% at 80 mg/day (p = 0.013), with 7 placebo and 0 NSI-189 withdrawals for intolerance in that stage.

Speculative 🟨

Unknown Consequences of Prolonged Neurogenic Stimulation

Basis is mechanistic only. The longest human exposure on record is 12 weeks, so whether years of driving stem cell proliferation in the hippocampus is harmless has never been tested in people.

Unpredictable Off-Target Effects

Basis is mechanistic only. The molecular target is unidentified and no receptor or enzyme screening panel has been published, so the usual prediction of off-target harms from binding data cannot be made.

Contaminated, Mislabelled or Misdosed Material

Basis is the absence of a regulated supply. NSI-189 is sold only by research-chemical vendors outside pharmaceutical manufacturing rules, so identity, purity and dose accuracy rest on vendor claims. No independent product testing exists.

Risk-Modifying Factors

  • Genetic polymorphisms: no pharmacogenetic data exist, and with the metabolic route unpublished the usual candidates, CYP2D6 and CYP3A4 (the liver enzymes clearing most psychotropic drugs), cannot be ruled in or out as sources of variable exposure.
  • Baseline biomarker levels: no biomarker predicts harm. Baseline liver enzymes and kidney function matter only in the ordinary sense that a compound of unknown clearance route is riskier where elimination organs are already impaired.
  • Sex: the Phase 1b trial found no pharmacokinetic difference between men and women, so no sex-specific dose adjustment is suggested by the data. Reproductive toxicity is untested; trials excluded pregnancy and required contraception in both sexes.
  • Pre-existing health conditions: trials excluded clinically significant cardiovascular, hepatic, renal, neurological and endocrine disease, plus any lifetime mania, hypomania (a milder form of mania) or psychosis. Safety in those groups is unknown rather than established as acceptable.
  • Age-related considerations: the published trials stopped at 60 and the ALTO-100 trials reported no age breakdown. Slower hepatic clearance, greater sensitivity to dizziness and somnolence, and higher fall risk make those events more consequential past 60.

Key Interactions & Contraindications

No formal drug-interaction study of NSI-189 has ever been run, and its metabolic route is unpublished, so every entry below is inferred from trial exclusion rules or from shared pharmacology rather than measured.

  • Prescription antidepressants (sertraline, escitalopram, venlafaxine, bupropion): caution; plausible consequences are additive sedation and unattributable mood shifts. The Phase 2 trial excluded them, so this is unstudied. Mitigation: monotherapy only.
  • Antipsychotics (quetiapine, olanzapine, aripiprazole), lithium and buspirone: caution for the same reason; all three were protocol-excluded. Consequence is unquantifiable rather than known to be severe. Mitigation: none established beyond avoiding the combination.
  • Sedatives allowed alongside in trials (benzodiazepines, zolpidem, zaleplon, eszopiclone, low-dose trazodone): monitor for additive daytime somnolence. Mitigation: these were permitted only at doses stable for four weeks before screening.
  • Over-the-counter sedating antihistamines (diphenhydramine, doxylamine) and alcohol: monitor; additive sedation and impaired driving are the plausible consequence given a somnolence rate near 28%. Mitigation: separate from the dose or avoid on dosing days.
  • Over-the-counter stimulants (caffeine, pseudoephedrine): monitor; additive insomnia and restlessness on a compound that already disrupts sleep in about one in six users. Mitigation: no caffeine after midday.
  • Enzyme-perturbing supplements (St John’s wort, a strong CYP3A4 inducer; grapefruit juice, a CYP3A4 inhibitor): caution. Because NSI-189 clearance is uncharacterised, exposure could shift in either direction unpredictably. Mitigation: avoid concurrent use.
  • Supplements with additive effects (5-HTP, or 5-hydroxytryptophan, a serotonin precursor; SAM-e, or S-adenosylmethionine; ashwagandha; melatonin): monitor. Serotonergic and sedating supplements compound the mood-activation and somnolence effects. Mitigation: introduce one new agent at a time.
  • Other neuroplasticity interventions (ketamine, psilocybin, repetitive transcranial magnetic stimulation): caution; the plausible consequence is compounded mood activation and an uninterpretable response, since all act on overlapping plasticity pathways and none has been combined with NSI-189. Mitigation: sequential rather than concurrent use.

Populations who should avoid NSI-189:

  • Anyone with a lifetime history of mania, hypomania or psychosis — excluded from every trial, and plasticity-targeting agents carry a theoretical switch risk.
  • Anyone with clinically significant suicidal ideation, or non-response to three or more adequate antidepressant trials in the current episode — both were protocol exclusions.
  • Anyone with a seizure disorder or an abnormal electroencephalogram; the Phase 1b protocol built electroencephalogram monitoring around peak concentration into every dosing day, and one participant showed transient sharp activity.
  • Anyone with an alcohol or substance use disorder active within the past 12 months.
  • Pregnant or lactating women, and anyone of either sex not using contraception; the Phase 1b protocol required contraception for three months after the last dose.
  • Anyone under 18 or over 60, or with a body mass index below 19.5 or above 35 kg/m², all outside the range of the two published trials.
  • Anyone with clinically significant cardiovascular, hepatic, renal, respiratory, endocrine, neurological or immunological disease, including Child-Pugh Class B or C liver impairment (a severity grade for chronic liver disease) and estimated glomerular filtration rate below 60 mL/min/1.73 m².

Risk Mitigation Strategies

  • A 40 mg daily ceiling: the trials found no added benefit at 80 or 120 mg/day and the 40 mg arm outperformed the 80 mg arm on every self-rated measure. The ceiling avoids dose-related dizziness and nausea.
  • Single morning dose: a 17–20 hour half-life means splitting doses raises peak concentration late in the day. Morning dosing mitigates insomnia and vivid dreams, which appeared in about one in six trial participants.
  • Exclusion of three-times-daily dosing: dizziness reached 66.7% and nausea 33.3% in the 120 mg/day divided-dose cohort, against 0% and 0% at 40 mg once daily. Dropping that regimen removes the steepest adverse-event gradient in the record.
  • Baseline liver enzymes, kidney function, electrocardiogram and electroencephalogram: the clearance route is unpublished and one trial withdrawal followed an abnormal electroencephalogram. Baselines give a comparator for later abnormality and screen for the excluded cardiac, hepatic and seizure groups.
  • A 12-week exposure limit: 12 weeks is the longest exposure any human has been studied at, so that limit bounds the wholly untested risk of prolonged neurogenic stimulation.
  • Batch certificate of analysis with mass-spectrometric identity confirmation: the only supply is unregulated. A batch-specific certificate mitigates the contamination, mislabelling and misdosing risk that no trial data can speak to.
  • A four-hour driving pause after early doses: somnolence affected 27.8% of trial participants and peaks near the 1–2 hour concentration maximum, so the first several doses carry the highest impairment risk.
  • No concurrent prescription psychotropics or second new agent: every such combination is unstudied, and simultaneous changes make it impossible to attribute either benefit or harm.

Therapeutic Protocol

No clinical practice exists for NSI-189, because no regulator has approved it and no clinic prescribes it; what follows is the dosing actually used in the trials rather than a practitioner consensus.

  • Trial-derived standard regimen: 40 mg once daily by mouth, the dose that outperformed both placebo and the 80 mg arm on every self-rated measure in the Phase 2 trial run by the Massachusetts General Hospital Clinical Trials Network and Institute.
  • Competing regimen — divided dosing: the Phase 1b trial, designed by Neuralstem with Fava’s group, escalated to 40 mg twice and three times daily to raise exposure without raising peak concentration. No efficacy gain, markedly more dizziness.
  • Competing regimen — biomarker-stratified use: Alto Neuroscience, founded by Amit Etkin, pre-selected participants by poor verbal memory before dosing 40 mg daily, reasoning the compound should help only where hippocampal plasticity is impaired. That approach failed in the 2024 trial.
  • Best time of day: morning. The trials dosed in the daytime, and a compound producing insomnia and vivid dreams in about one in six users is worst placed near bedtime.
  • Half-life and accumulation: 17.4–20.5 hours, with steady state reached at 96–120 hours regardless of how the daily dose is split, which is why once-daily dosing is sufficient and effects should not be judged before day five.
  • Single versus split dosing: once daily is supported. Splitting was introduced only to blunt peak-related adverse events at total doses above 40 mg, and total doses above 40 mg showed no benefit.
  • Genetic polymorphisms: none can be used to guide dosing. With the metabolic route unpublished, no CYP2D6, CYP3A4, COMT or MTHFR (the folate-processing enzyme gene) genotype has any established bearing on exposure or response.
  • Sex-based differences: none in dosing. Phase 1b pharmacokinetics showed no obvious male-female difference in exposure, and no sex-by-treatment efficacy interaction was reported.
  • Age-related considerations: no dosing guidance exists above 60; the published trials excluded that range and the ALTO-100 trials published no age breakdown. Reduced clearance and greater sensitivity to dizziness argue against extrapolating the 40 mg dose upward in age.
  • Baseline biomarker levels: baseline depression severity below 30 on the clinician-rated scale, and poor baseline verbal memory, are the two response predictors that have been proposed. Both come from post-hoc or prospectively failed analyses.
  • Pre-existing health conditions: trials enrolled only medically healthy adults with recurrent depression and no more than two failed antidepressant trials in the current episode, so response in medically complex people is unstudied.

Discontinuation & Cycling

  • Not intended as a lifelong agent: every trial dosed for a fixed 4 or 12 weeks, and the compound was developed as a course of treatment for a depressive episode rather than as continuous maintenance therapy.
  • No withdrawal syndrome reported: neither trial recorded discontinuation symptoms after stopping, in explicit contrast to the discontinuation effects seen with serotonergic antidepressants.
  • No taper required by the available data: the Phase 1b trial stopped dosing abruptly at day 28 and followed participants to day 84 without a taper and without rebound.
  • Effects appear to outlast dosing: improvement on the self-rated depression scale was still present eight weeks after the last dose in that trial, which is the single most distinctive claim made for the compound and the least replicated.
  • Cycling is untested: no study has compared continuous with intermittent dosing, and no tolerance has been documented over 12 weeks, so there is no evidence that cycling maintains or restores any effect.
  • A practical stopping rule: since the durability claim implies benefit should persist, continuing past 12 weeks adds untested exposure without a documented reason.

Sourcing and Quality

  • No regulated supply exists: NSI-189 is not approved anywhere, is not a dietary supplement, and is not stocked by compounding pharmacies, so no reputable pharmaceutical brand or compounder can be named. Every available source is a research-chemical vendor.
  • Formulation used in trials: NSI-189 phosphate, the phosphate salt, in oral capsules. Vendor material is usually sold as the same salt in bulk powder, where the salt-versus-free-base distinction shifts the actual delivered dose.
  • What to look for: a batch-specific certificate of analysis, dated and matching the lot number, with identity confirmed by mass spectrometry and purity by high-performance liquid chromatography — not a generic certificate reused across batches.
  • Independent verification matters more here than for supplements: third-party testing commissioned by the buyer, rather than testing supplied by the seller, is the only check on identity for a compound with no regulatory oversight of any kind.
  • Heavy metal and solvent residue testing: synthesis of a piperazine-containing molecule uses catalysts and solvents that a certificate of analysis should quantify. Absence of residual solvent data is a meaningful gap, not a formality.
  • Powder dosing is a quality problem in itself: accurate delivery of 40 mg from bulk powder requires a milligram-resolution balance; volumetric scoops routinely miss by a wide margin.
  • Purity claims cannot be assumed: no published survey has tested marketed NSI-189 products for identity or purity, so the failure rate of the vendor market is simply unknown.

Practical Considerations

  • Time to effect: steady state arrives at 96–120 hours, and the trials measured outcomes at 4 and 12 weeks. Self-rated improvement in the Phase 2 trial emerged mainly in the second six-week stage, so a month is the minimum meaningful trial.
  • Common pitfall — dosing too high: the 80 mg arm underperformed the 40 mg arm on every self-rated measure, and 120 mg/day produced the worst adverse-event profile. More is measurably worse.
  • Common pitfall — judging it in days: with a 17–20 hour half-life and a five-day approach to steady state, any impression formed in the first week is uninterpretable.
  • Common pitfall — trusting the volume claim: vendor copy widely cites a fixed percentage increase in hippocampal volume. The Phase 1b imaging found no significant change, and a similar non-significant drift appeared in the amygdala control region.
  • Regulatory status: not approved by the Food and Drug Administration or any other regulator, not a dietary supplement under United States law, and not lawfully marketed for human consumption. Purchase is legal only as a research chemical.
  • Cost and accessibility: inexpensive as research-chemical powder, but that price reflects the absence of pharmaceutical manufacturing, quality control and liability rather than genuine affordability.
  • A structural bias: approved, NSI-189 would be a branded agent competing against generic antidepressants costing pennies a day. Insurers and national health systems have a systematic financial incentive to favour those generics, which shapes which trials get funded and which guidelines get written.

Interaction with Foundational Habits

  • Sleep: direct and mostly adverse. Insomnia and vivid dreams each affected about one in six trial participants while daytime somnolence affected over a quarter, consistent with a long-half-life agent shifting sleep architecture. Practical consequence: morning dosing, with worsening sleep read as a reason for discontinuation rather than for adding a sedative.
  • Nutrition: largely indirect. No food-effect study has been published, so whether a meal alters absorption is unknown; the trials dosed in a controlled inpatient or outpatient setting without dietary restriction. Practical consequence: constant timing relative to food, so that variable absorption is not mistaken for variable response.
  • Exercise: potentiating in principle. Aerobic exercise is the best-validated stimulus for hippocampal neurogenesis and BDNF signalling in humans — the same pathway NSI-189 targets indirectly — and it has clinical outcome data the compound lacks. Practical consequence: exercise is the higher-evidence route to the same endpoint, not an alternative to be displaced.
  • Stress management: indirect and important. Chronic stress and sustained cortisol elevation suppress hippocampal neurogenesis, which is the substrate the compound is meant to expand, so unmanaged stress plausibly works against it. Practical consequence: no interaction has been measured, and any stress-reduction effect on response remains a mechanistic inference.

Monitoring Protocol & Defining Success

Because NSI-189 has no approved label and no established monitoring standard, the sensible baseline panel is the one the trials themselves used to screen and to follow participants: liver enzymes, kidney function, a complete blood count, an electrocardiogram, vital signs, and a thyroid screen to rule out a treatable cause of the same symptoms. A baseline is not optional here — with the metabolic route unpublished, any later abnormality is uninterpretable without a pre-treatment comparator. Baseline cognitive and mood measures belong on the same day, since the outcomes at issue are subjective and memory drift makes retrospective comparison unreliable.

For ongoing monitoring, a workable cadence is week 1 for tolerability and vital signs, week 4 for the first full repeat panel and the first meaningful efficacy read, week 12 at the end of the studied exposure window, and every 6 months thereafter if use continues beyond the tested period.

Biomarker Optimal Functional Range Why Measure It? Context/Notes
Alanine aminotransferase 10–26 U/L Liver injury from a compound of unknown clearance route ALT (alanine aminotransferase) is a liver enzyme; conventional labs flag only above 40–55 U/L, well above the functional ceiling. Fasting not required; pair with aspartate aminotransferase.
Aspartate aminotransferase 10–26 U/L Confirms and contextualises any alanine aminotransferase rise AST (aspartate aminotransferase) also rises with muscle damage; conventional upper limit is about 40 U/L. Avoid testing within 48 hours of hard training.
Estimated glomerular filtration rate >90 mL/min/1.73 m² Renal clearance capacity, unmeasured for this compound eGFR (estimated glomerular filtration rate) is calculated from creatinine; conventional labs call 60–89 normal. Cystatin C is the better paired test in lean or muscular individuals.
Resting heart rate 55–70 bpm Palpitation was a recorded treatment-related event Conventional labs call anything from 60 to 100 bpm normal, well above the functional ceiling. Measure seated after five minutes’ rest, same time of day. Best paired with blood pressure at the same sitting.
Blood pressure <120/80 mmHg Baseline for any autonomic effect and a screen for the excluded cardiovascular group Conventional thresholds treat below 130/80 as acceptable. Take two readings a minute apart and record the second.
Corrected QT interval <430 ms (men), <450 ms (women) Trials ran electrocardiograms; cardiac safety is otherwise uncharacterised QTc (corrected QT interval) measures ventricular repolarisation on an electrocardiogram; conventional cut-offs are wider, at 450 ms for men and 470 ms for women. Repeat only if a QT-prolonging drug is added.
Thyroid stimulating hormone 1.0–2.0 mIU/L Thyroid dysfunction mimics the depressed mood and cognitive dysfunction being targeted TSH (thyroid stimulating hormone) has a conventional range of 0.4–4.5 mIU/L, far wider than the functional target. Draw in the morning; pair with free T3 and free T4 (the circulating thyroid hormones).
Haemoglobin A1c 4.8–5.4% Metabolic status, given the diabetic-model preclinical data HbA1c (haemoglobin A1c) reflects average glucose over about three months; conventional labs call below 5.7% normal. Fasting not required.
High-sensitivity C-reactive protein <1.0 mg/L Inflammation suppresses neurogenesis and blunts antidepressant response hs-CRP (high-sensitivity C-reactive protein) is a general inflammation marker; conventional cardiac cut-offs run to 3.0 mg/L. Invalid within two weeks of infection or injury.
White blood cell count 4.0–7.0 ×10⁹/L Trials monitored haematology; no marketed comparator exists to predict effects Part of a complete blood count. Conventional range extends to 11.0 ×10⁹/L. Draw with the same morning fasting sample as the metabolic panel.
Serum brain-derived neurotrophic factor No established target range exists; track the change from the individual’s own baseline The proposed mechanism runs through this protein, and baseline levels predicted response in the Phase 1b trial BDNF (brain-derived neurotrophic factor) supports neuronal survival and growth. Assay variability between labs is large, so only within-person, same-lab comparisons are meaningful. Platelet-poor plasma preferred over serum.

Qualitative markers are at least as informative as the panel here, because the only human signals that ever separated from placebo were self-rated:

  • Mood and emotional reactivity, tracked with the same self-report scale each time rather than by impression.
  • Verbal memory, tested as recall of an unrelated word list — the specific measure used to pre-select likely responders in the 2024 trial.
  • Mental clarity, concentration and the ability to make decisions, the domains where the self-rated cognitive scale separated from placebo.
  • Sleep quality, sleep onset latency and dream intensity, the earliest adverse signals to appear.
  • Daytime alertness within four hours of dosing, given the somnolence rate.
  • Restlessness, irritability or anxiety in the first two weeks, the window unsupervised users most often describe.

Emerging Research

  • Bipolar depression trial: NCT06656416, a 200-participant Phase 2 study of ALTO-100 in bipolar I or II depression, recruiting with completion expected in October 2026. Its primary endpoint is the clinician-rated depression score at week 6 in a pre-defined biomarker subgroup. A positive result would be the first.
  • The unpublished negative dataset: NCT05712187 enrolled 301 adults across 34 United States sites and reported no separation from placebo. No results are posted and no journal publication has appeared, leaving the strongest evidence against the compound outside independent scrutiny.
  • Whether any biomarker identifies responders: NCT05117632, a 245-participant study in depression and post-traumatic stress disorder, generated the verbal-memory biomarker that the larger trial then failed to replicate. Whether the enrichment strategy or the compound failed is the central open question.
  • Peripheral nerve and metabolic indications: the animal work of Jolivalt et al., 2019 and Jolivalt et al., 2022 on diabetic neuropathy and mitochondrial function is the strongest case for a non-psychiatric use. No human trial in this indication has been registered.
  • Stalled injury and neurodevelopmental indications: the stroke work of Tajiri et al., 2017, the Angelman syndrome work of Liu et al., 2019 and the radiation work of Allen et al., 2018 remain animal-only. That none has advanced to human testing in seven years weakens the case.
  • Durability, still unmeasured: NCT02724735, a 220-participant observational cohort designed to measure time to starting a new antidepressant, would have tested the persistence claim directly. Its registry status is listed as unknown and no results have appeared.
  • Identifying the molecular target: no published work has resolved what NSI-189 actually binds. Until it does, off-target risk cannot be predicted and the neurogenic explanation for any clinical effect stays an inference rather than a demonstrated chain.

Conclusion

NSI-189 is an experimental oral compound built on a genuinely different idea: rather than adjusting the brain’s chemical messengers, it was designed to make the brain region that governs memory and mood grow new cells. In animals that idea works, repeatedly and across very different kinds of damage. In people the picture is far weaker: the two trials designed to prove the effect both missed the outcome they set out to change. What separated from placebo, twice, were participants’ own ratings of their mood and mental clarity — a real signal, but one the doctors’ assessments in the same trials did not confirm, and one a larger, more carefully selected group failed to reproduce.

The safety record is reassuring as far as it goes. Nothing serious happened in any trial, fewer people stopped for side effects on the compound than on placebo, and the events that did occur — headache, nausea, dizziness, sleepiness and vivid dreams — were mild and rose with dose. But the longest anyone has been studied is twelve weeks, the published trials stopped at sixty, and no one knows what the compound binds to or how the body clears it.

Two things shape how the evidence should be read. Almost every study was authored by staff of the company developing it, so the published record carries a direct commercial interest. The entire supply is unregulated, adding a quality risk on top of the uncertainty about the drug. The mechanism remains interesting; the human case for it remains unmade.

Top - Benefits - Risks - Protocol