Nevadistinel (NYX-458)

NYX-458 (nevadistinel) is an oral NMDA receptor positive allosteric modulator (PAM) that Aptinyx took into Phase 2 for cognitive impairment in Parkinson's disease and Lewy body dementia (NCT04148391, n=99, 30 mg daily for 12 weeks) [1]. The trial actually read out on February 27, 2023: NYX-458 did not demonstrate clinically meaningful improvements over placebo on any efficacy endpoint, and Aptinyx discontinued development [3]. About three months later the company filed Chapter 7 bankruptcy, following parallel Phase 2 failures of NYX-783 in PTSD (March 2023) and the earlier NYX-2925 failures in painful diabetic peripheral neuropathy (April 2022) and fibromyalgia (Q3 2022) [2,6]. The ClinicalTrials.gov entry still reads 'active, not recruiting' but that status field is stale - the program is dead in this indication. The remaining residual value of the asset is the molecule plus its unpublished Phase 2 dataset; a future acquirer could in principle reposition nevadistinel into a different cognitive indication, but the negative PD/LBD readout substantially weakens the case for the 'gentle modulator' thesis Aptinyx had built the platform around.

Novel small molecule, never approved anywhere. Phase 2 in cognitive impairment associated with Parkinson's disease and Lewy body dementia, target enrollment 99, dosed at 30 mg daily for 12 weeks [1,3]. ClinicalTrials.gov status field still shows 'active, not recruiting' - that's stale. Aptinyx announced negative top-line results on February 27, 2023: no efficacy signal on any endpoint, and the company terminated the program [3]. In May 2023, after a parallel failure of NYX-783 in PTSD and following the earlier NYX-2925 pain failures, Aptinyx filed Chapter 7 bankruptcy and dissolved [2]. No FDA breakthrough, fast track, orphan, or RMAT designation is on record for this compound. Realistic near-term output: peer-reviewed publication of the Phase 2 secondary endpoints is the only credible deliverable left, and even that depends on whether the original academic principal investigators decide to write it up. PIs retain publication rights independent of the corporate sponsor - sponsor dissolution does not extinguish academic authors' ability to publish (subject to whatever data-access provisions the bankruptcy trustee enforces over the locked database). Until and unless a buyer acquires the molecule for a different indication, the development status is best described as terminated, not active.

NMDA receptors are ligand-gated ion channels on neurons that open when glutamate (the brain's main excitatory neurotransmitter) and glycine bind them, letting calcium rush in. That calcium influx triggers synaptic plasticity, the cellular basis of learning and memory [4]. Too much NMDA activity kills neurons (excitotoxicity, the mechanism behind stroke damage); too little starves synapses of the signaling they need to form and reinforce memories. The receptor is a four-subunit complex built from GRIN1 (the obligate glycine-binding subunit present in every NMDA receptor) plus combinations of GRIN2A through GRIN2D (the glutamate-binding subunits, also called NR2A-D). The GRIN2B/NR2B subunit is enriched in cortex and hippocampus - the circuits that fail first in PD and LBD cognitive decline [4]. Nevadistinel is a positive allosteric modulator (PAM): it binds a site distinct from both the glutamate/glycine agonist sites and the channel pore, requires endogenous glutamate and glycine to be present, and amplifies physiological signaling rather than driving the channel on its own [3]. The pharmacology distinction matters: ketamine, esketamine, and memantine are channel-pore blockers, useful but blunt. A PAM in principle nudges the receptor's gain to restore normal signaling without flatlining it. NYX-458 has a novel spiro-β-lactam chemical scaffold and was designed to mimic the activity of rapastinel, an earlier Aptinyx peptide NMDA partial agonist [3]. The exact PAM binding site on NYX-458 has not been publicly disclosed in detail. Aptinyx's whole platform - NYX-2925 in pain, NYX-783 in PTSD, NYX-458 in cognition - was built around this 'restore-don't-block' PAM thesis. The genetic validation for the target is real: GRIN2B loss-of-function variants cause intellectual disability and developmental disorders [5,7]. The preclinical validation was also reasonable: in a nonhuman primate Parkinson's model, NYX-458 produced rapid cognitive improvements that persisted for three months across attention, working memory, and executive function, without affecting motor symptoms or levodopa response [3]. The clinical validation for the PAM approach is now thin-to-negative. NYX-2925 missed in pain, NYX-783 missed in PTSD, and NYX-458 missed in PD/LBD cognition. Target biology is well-established; Aptinyx's specific PAM chemistry failed across three indications.

NCT04148391 enrolled 99 patients with mild cognitive impairment or mild dementia due to Parkinson's disease or Lewy body dementia [1]. Sponsor: Aptinyx. Phase 2, randomized, placebo-controlled, 30 mg nevadistinel daily for 12 weeks [3]. The ClinicalTrials.gov record lists the primary endpoint as 'change from baseline in physical examination' - a safety endpoint sitting in the primary slot. That signals the study was designed as a Phase 2a safety, tolerability, and PK (pharmacokinetic - how the drug is absorbed, distributed, metabolized, and cleared) exercise, with cognitive endpoints as secondary or exploratory measures. That is a reasonable choice for a first study in a fragile dementia population. The trade-off is that even a clean readout would not independently support efficacy claims; it would only justify a larger Phase 2b. Enrollment of 99 is small but consistent with proof-of-mechanism work in PD cognition, where recruiting cognitively impaired patients who can still consent is genuinely hard. The trial population mixes PD and LBD, a design choice that broadens recruitment but combines two related-but-distinct neurodegenerative processes, complicating signal interpretation. In the end those trade-offs were academic: Aptinyx announced on February 27, 2023 that NYX-458 produced no clinically meaningful improvement over placebo on any efficacy endpoint [3]. The structural Phase 2a design was reasonable; the molecule did not deliver.

Most drugs that reach Phase 2 never make it to approval - historically, about 24% do in this area. Our model takes that starting point and adjusts it using ten specific facts about this trial and its sponsor, arriving at a 4% chance of eventual approval. The biggest drags are the sponsor's weak approval record, limited earlier-phase results, heavier-than-usual blinding, and a randomized trial design. The remaining factors are close to average for this stage and don't move the number much either way.

The risks here are mostly historical because most have already materialized. (1) Efficacy risk - realized. The Phase 2 trial failed on all efficacy endpoints in February 2023 [3]. (2) Sponsor risk - realized. Aptinyx dissolved in Chapter 7 in May 2023 [2]. No corporate entity exists to run a Phase 3, submit revised IND (Investigational New Drug) paperwork to FDA, or commit to publishing the full Phase 2 dataset. Asset disposition by the bankruptcy estate has not been publicly disclosed. (3) Class-efficacy risk - also realized. All three Aptinyx NMDA-PAM clinical programs failed at Phase 2: NYX-2925 in painful diabetic peripheral neuropathy (April 2022) and fibromyalgia (Q3 2022) [6], NYX-783 in PTSD (March 2023) [2], and NYX-458 in PD/LBD cognition (February 2023) [3]. Memantine - an approved NMDA channel-pore blocker - offers modest benefit in moderate-to-severe Alzheimer's (around 3 points on ADAS-Cog, the Alzheimer's Disease Assessment Scale-Cognitive Subscale, an 11-task cognition test), but the PAM sub-mechanism specifically has now failed across three indications. (4) Indication-difficulty risk - confirmed. PD and LBD cognitive impairment has chewed through dozens of programs. Rivastigmine is FDA-approved for PD dementia and donepezil is used off-label, both with marginal benefit; brexpiprazole (Otsuka) has been studied in LBD-related agitation; almost everything novel has produced flat trials or safety signals. Mixing PD and LBD in one Phase 2 made any directional signal harder to read - and in the end there was no signal to interpret. The remaining open issues are around the unpublished dataset (will it ever surface?) and asset disposition by the bankruptcy estate, not clinical or scientific risk for the molecule.

Noise. This is a failed Phase 2 from a dissolved sponsor in a notoriously hard indication. Data points that would partially flip this back toward signal: (a) academic publication of NCT04148391 secondary endpoints showing any directional cognitive benefit - PIs can publish independently of the defunct sponsor and that is the most likely near-term output for the program; (b) asset acquisition by a neurology-focused buyer with a credible plan to test nevadistinel in a different indication (e.g., mild cognitive impairment without Lewy body pathology, where the NHP preclinical data was strongest); (c) post-hoc subgroup analysis showing a responder population (PD vs LBD split, ApoE4 stratification, baseline cognitive severity). None of these are commercially actionable on their own. Worth checking back: PubMed for any publication from the original PIs; ClinicalTrials.gov for a results posting; bankruptcy estate filings for asset-disposition records. Smaller neurology-focused buyers of distressed assets (Acadia, Anavex, and historically Cerevel before the AbbVie acquisition) are the realistic acquirer profile, but the negative Phase 2 readout makes the asset substantially less attractive than the 'zombie with unread data' framing would suggest. Market context for any acquirer: dementia with Lewy bodies affects roughly 1.4 million people in the US (with overlap into PD dementia, which develops in 25-30% of long-duration PD patients) - a real unmet need, but Phase 3 in neurodegenerative cognition typically runs $100-300M+ given long study durations, large enrollment, and high screening attrition. That cost stack, against a failed Phase 2 and a three-for-three platform miss, explains why no acquirer has stepped forward. The mechanism is interesting and the molecule is real, but Aptinyx's PAM platform now has three Phase 2 failures and zero successes. Keep the node in the graph for target-class connectivity and historical completeness; do not expect commercial activity. Re-audit if the Phase 2 results are published with any directional signal, if an acquirer is named, or if the bankruptcy estate publicly disposes of the asset.