Acadia Pharmaceuticals Inc. (ACAD): RADIANT is a pivotal near-term test of remlifanserin’s Alzheimer’s psychosis potential

Overview

The report frames ACAD’s upcoming Phase 2 RADIANT readout for remlifanserin in Alzheimer’s disease psychosis. Goldman Sachs identifies potentially improved pharmacology, a more targeted trial design, and a favorable safety proposition, but stresses that durable efficacy, endpoint relevance, and commercial differentiation versus BMY’s Cobenfy remain unresolved.

Core views

ACAD expects Phase 2 RADIANT data for remlifanserin in ADP in September–October 2026. The double-blind, 1:1:1 trial compares 30mg and 60mg doses with placebo in 363 patients over six weeks and is powered at 80% to detect a moderate 0.4 effect size on SAPS-H+D, a clinician-rated scale specifically measuring hallucinations and delusions. Goldman Sachs views the study’s enrichment for more severe, biomarker-confirmed ADP patients as a strength because it seeks to reduce disease heterogeneity and exclude pseudo-dementia. ACAD also screens patients using both NPI-NH and SAPS-H+D. The timing could slip into October if patients do not roll into the open-label extension and instead require a 30-day safety follow-up. ACAD has begun enrollment in two similar Phase 3 studies and expects the Phase 2 and Phase 3 data together to support an FDA NDA submission. The central clinical question is whether 5-HT2A inverse agonism is sufficiently validated in ADP. The supportive case rests on remlifanserin’s potential pharmacologic advantages over pimavanserin, marketed as Nuplazid. ACAD reported 32x–123x 5-HT2A versus 5-HT2C selectivity for remlifanserin, compared with 8x–37x for pimavanserin; a roughly 18–20 hour half-life and about six days to steady state, compared with about 57 hours and 15 days for pimavanserin; and five times higher cerebrospinal-fluid concentrations in a non-human-primate model despite similar plasma concentrations. Goldman Sachs notes that higher Nuplazid exposure was associated with better historical efficacy in ADP, while its 34mg dose is constrained by QT prolongation at higher exposures. ACAD’s data indicate no significant QT-prolongation risk for remlifanserin at twice the marketed-dose equivalent of Nuplazid, which could permit greater dosing flexibility, faster onset, and potentially better efficacy. The counterargument is Nuplazid’s weak regulatory precedent in ADP. In Phase 2 Study 019, pimavanserin improved the NPI-NH psychosis score by 3.76 points at week six versus 1.93 points for placebo, a mean difference of -1.84 points (95% CI -3.64 to -0.04; p=0.045), but showed no significant overall-study advantage at week 12. In the Phase 3 HARMONY dementia-related psychosis study, relapse occurred in 13% of pimavanserin-treated patients versus 28% on placebo, for a hazard ratio of 0.35, but the FDA viewed the result as largely driven by Parkinson’s disease psychosis: the subgroup hazard ratio was 0.054 in PDP versus 0.618 in ADP. The FDA issued a complete response letter for Nuplazid in ADP, citing the small week-six separation and lack of sustained benefit in Study 019 as well as the subgroup issue in HARMONY. Goldman Sachs therefore treats RADIANT as a proof-of-concept test rather than assuming that the prior mechanism translates into durable ADP efficacy. Endpoint choice is another debate. Goldman Sachs explains that SAPS-H+D contains 20 hallucination- and delusion-specific items and may better detect subtle changes in the symptoms targeted by a psychosis medicine than NPI, which covers up to 14 behavioral domains and can dilute a psychosis improvement if unrelated symptoms do not change. Conversely, clinicians view NPI-C and NPI-NH as faster, more practical measures that capture the broader behavioral burden of Alzheimer’s disease, where psychosis commonly coexists with agitation, depression, and apathy. NPI also has established regulatory use in Alzheimer’s neuropsychiatric symptoms, whereas SAPS-H+D originated in schizophrenia research. KOLs consider RADIANT’s six-week duration a risk if placebo separation emerges later, although ACAD argues that remlifanserin’s six-day time to steady state could allow earlier and sustained separation. Competitive positioning versus BMY’s Cobenfy is mixed. KOLs see remlifanserin’s serotonergic mechanism as familiar and potentially better suited to frail older patients, with a cleaner profile than dopamine-blocking antipsychotics and fewer expected QT concerns than pimavanserin or gastrointestinal concerns than xanomeline. Its QD dosing, no observed food effect, and lack of reported drug-drug interactions with commonly used AD medicines are also potential practical advantages. However, remlifanserin is focused on psychosis, which may limit commercial upside if it does not improve broader symptoms. Cobenfy’s distinct muscarinic mechanism could address psychosis alongside agitation, cognition, or other behavioral symptoms, making it the more exciting asset in some experts’ view; it must nevertheless establish efficacy and clean tolerability in ADP. Historical high-dose xanomeline monotherapy had a 52% discontinuation rate, with 12.6% experiencing syncope. In schizophrenia, Cobenfy had higher cholinergic adverse-event rates than placebo, including nausea at 19%, dyspepsia at 18%, constipation at 17%, vomiting at 15%, and diarrhea at 6%, although adverse-event discontinuation was 6% versus 4% for placebo. KOLs flag gastrointestinal tolerability, meal timing, titration, and TID dosing as particular concerns for cognitively impaired ADP patients. The market opportunity is substantial but depends on successful clinical differentiation. ACAD estimates roughly 7 million US Alzheimer’s disease patients, with about 30% experiencing psychosis; using a 50% diagnosis rate, the report estimates 1.1 million diagnosed ADP patients, 65% not well controlled on antipsychotics, and a 0.7 million addressable population. The report calculates an illustrative branded-agent serviceable market of $8.693 billion using a $16,983 net annual price and 75% persistence. Visible Alpha consensus peak sales are about $2.015 billion for remlifanserin, implying 23% market share, and about $2.515 billion for Cobenfy in ADP, implying 29% share. Goldman Sachs does not currently include remlifanserin in its base model. In an illustrative DCF that incorporates consensus estimates, approximately $2 billion peak sales, and a 2038 loss of exclusivity, the current share price of about $27 implies a 42% probability of success, in line with Visible Alpha consensus. If a positive Phase 2 outcome led the firm to apply a 65% probability of success, reflecting Phase 2-to-Phase 3 translation risk in neuropsychiatric disease, its 12-month forward valuation would rise to $29 from $17, around 6% above then-current levels. The report separately states that a positive result could imply 6% upside at a 65% probability of success to 24% upside at 100% probability of success, while options imply an approximately +/-10% to 23% move, or a $20–$33 range, into the data release. Goldman Sachs retains its Sell rating and unchanged $17 12-month target, based on a 100% DCF using a 12% WACC and 2% terminal growth rate.

Analysis framework

Goldman Sachs evaluates RADIANT through trial design, endpoint selection, prior pimavanserin evidence, remlifanserin pharmacology and safety data, KOL feedback, Cobenfy’s competing program, US market sizing, and probability-adjusted DCF scenarios.

Methodology notes

• Valuation methodsDCF (Discounted Cash Flow)

  Probability-adjusted discounted cash flow valuation

  The report uses a DCF to illustrate how a positive Phase 2 result and a higher assumed probability of success for remlifanserin could affect ACAD’s valuation. Its stated base target uses a 12% WACC and 2% terminal growth rate.

• Event-Driven and Behavioral FinanceEvent-driven analysis

  Clinical-readout scenario analysis

  The report frames the RADIANT data release as a catalyst, testing clinical efficacy, safety, endpoint credibility, and the probability of later-stage success, while comparing illustrative valuation outcomes with the options-implied trading range.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• Acadia Pharmaceuticals Inc. (ACAD)

  Primary covered company; value is linked to the upcoming remlifanserin ADP readout.

  Strengths

  Biomarker-enriched trial design, potentially improved QT profile, faster steady state, QD dosing, and experience in long-term-care facilities through Nuplazid.

  Weaknesses

  Remlifanserin’s ADP efficacy remains unproven, and its symptom-specific focus may constrain commercial upside.

  Comparison

  Cobenfy may offer broader neuropsychiatric symptom control but has gastrointestinal and TID-dosing concerns.

  Risks

  RADIANT may not demonstrate durable placebo separation or sufficient efficacy to overcome the prior Nuplazid regulatory precedent.

• Bristol Myers Squibb (BMY)

  Competitor through Cobenfy’s Phase 3 ADEPT program in ADP.

  Strengths

  Distinct muscarinic mechanism may offer broader symptom control beyond psychosis.

  Weaknesses

  Potential gastrointestinal tolerability, titration, meal-timing, and TID-adherence challenges in older ADP patients.

  Comparison

  KOLs generally view Cobenfy as more ambitious, while viewing remlifanserin as potentially safer and more familiar for geriatric psychosis care.

  Risks

  Cobenfy’s approach has not yet been validated in ADP and its ADP trial readouts are expected in early 2027 or later.

Key data

• RADIANT design363 patients; 30mg and 60mg versus placebo; 6-week treatmentDouble-blind, 1:1:1 randomized Phase 2 study in ADP
• RADIANT statistical power80% power for a 0.4 effect sizePrimary endpoint is SAPS-H+D
• Remlifanserin time to steady state~6 daysVersus ~15 days for pimavanserin
• CSF exposure5x higher versus NuplazidReported in a non-human-primate model
• Study 019 week-six result-1.84-point mean difference; p=0.045NPI-NH psychosis-score benefit for pimavanserin versus placebo was not maintained at week 12
• HARMONY relapse result13% versus 28%; hazard ratio 0.35Pimavanserin versus placebo; ADP subgroup hazard ratio was 0.618 versus 0.054 in PDP
• Illustrative remlifanserin valuation$2912-month forward valuation after a positive Phase 2 result under a 65% probability-of-success assumption, versus $17 currently
• US ADP serviceable market$8.693 billionIllustrative branded-agent market estimate

Impact & implications

The report presents RADIANT as the key near-term event for determining whether remlifanserin can convert pharmacologic and trial-design advantages into durable, clinically meaningful ADP efficacy. A positive result could support a higher probability-weighted valuation, while weak or non-durable efficacy would reinforce the historical uncertainty surrounding the mechanism in ADP.

Risks

• RADIANT may fail to show meaningful or durable efficacy versus placebo, particularly given the six-week treatment duration.
• The prior Nuplazid experience creates translation and regulatory risk because ADP benefit was not sustained in Study 019 and HARMONY results were viewed as driven by the PDP subgroup.
• Cobenfy could establish a more compelling efficacy profile through broader symptom control if its tolerability is acceptable.
• The report identifies upside risks to its Sell view from stronger Nuplazid adoption in PDP, ACAD becoming an M&A target, stronger-than-expected Daybue uptake, and positive pipeline updates including remlifanserin in ADP.

What to watch

• The RADIANT Phase 2 readout expected in September–October 2026, including SAPS-H+D efficacy, dose response, onset, safety, and durability of placebo separation.
• Whether patient rollover into the open-label extension permits a September readout or pushes results into October.
• Evidence that remlifanserin’s faster steady state and reduced QT risk translate into a clinically differentiated profile.
• BMY’s ADEPT interim and topline updates, including expected early-2027-or-later data from ADEPT-1, ADEPT-2, and ADEPT-4.
• Cobenfy’s gastrointestinal tolerability and practical dosing profile in older, frail ADP patients.