Psychosis associated with Alzheimer’s diseaseNovember 18, 2026

KarXT’s Alzheimer’s Psychosis Signal: What Open-Label Progress Adds to BMY’s Investment Case

BMY

Executive Summary

KarXT’s preliminary ADEPT findings support further investigation, while the investment question remains whether controlled evidence and tolerability can support a durable new treatment opportunity. BioScore unverified; no editorial investment rating assigned.

KarXT’s Alzheimer’s Psychosis Signal: What Open-Label Progress Adds to BMY’s Investment Case

Treating Alzheimer’s disease is not only a question of preserving memory. A therapy that reliably reduces hallucinations and delusions could address a different, immediately important problem for patients and caregivers. For investors, that distinction matters: symptom management can have substantial value without changing the underlying course of dementia. It also requires its own evidence, prescribing logic and commercial assumptions.

What was actually disclosed

Bristol Myers Squibb’s November 18 company disclosure at CTAD concerns descriptive, open-label KarXT results; double-blind ADEPT data remain blinded. In ADEPT-1, 273 of 320 participants completed 12 weeks. The hallucination-and-delusion scale fell 12.0 points from 14.4 among 266 assessed participants. Treatment-emergent adverse events occurred in 56.6%, serious events in 5.0%, and event-related discontinuations in 4.4%.

In the 52-week ADEPT-3 extension, 273 of 401 completed treatment. Corresponding adverse-event percentages were 65.1%, 7.7% and 4.7%. Hallucination-and-delusion scores declined 5.1 points from 8.9 among 270 assessed participants. These analyses excluded participants still ongoing at cutoff and included those completing or prematurely discontinuing the relevant period. KarXT combines xanomeline and trospium. These are company-reported preliminary observations, not a disclosed randomized efficacy result. BMS, November 18

The commercial opportunity is a chain, not a patient count

Our investment interpretation starts with three separate propositions. First, symptoms must improve because of treatment rather than observation, expectations or natural variation. Second, that improvement must justify treatment burden and risk. Third, clinicians, families and payers must find it practical enough to sustain use. A large disease population does not establish all three.

The headline change from baseline is relevant to the first proposition, but cannot settle it. Without the concurrent randomized comparison, one cannot isolate the treatment contribution. An investor should therefore avoid converting the observed score change directly into either a success probability approaching certainty or a penetration forecast.

The same caution applies in the other direction. Descriptive data are not worthless simply because they are not definitive. They can help frame whether the next test is worth pursuing, which outcomes deserve attention and what a realistic development thesis must explain. The disciplined response is to improve the questions, not dismiss the program or declare it de-risked.

Persistence is part of the product

For a chronic treatment, initial symptom change and sustained usability are distinct assets. A commercial model should therefore distinguish eligible patients, patients starting therapy, patients benefiting and patients remaining on treatment. Applying one penetration percentage to the entire addressable population hides where the uncertainty actually sits.

Completion rates are informative but do not identify every reason someone stopped. Nor is a serious adverse event automatically treatment-caused. Investors need event types, timing, adjudication, exposure and an appropriate comparator before translating aggregate safety figures into a definitive causal judgment. Detailed mortality information and comparative safety remain important diligence items; this note does not infer either absence of risk or a drug-related mortality effect from the available aggregate disclosure.

Extension studies also deserve a separate lens. People reaching and entering an extension need not represent all patients who might eventually receive treatment. Persistence observed in that setting should not be mechanically inserted into a real-world revenue forecast. A sensitivity analysis is more honest than a single confident duration assumption.

What would change the valuation argument?

The next material step is evidence that separates a treatment effect from a within-patient trajectory. After that, the relevant questions become durability, safety, label scope and adoption economics. These are sequential uncertainties, not interchangeable milestones.

For BMY, the appropriate framework is an indication-specific scenario analysis. A conservative case leaves this use as unproven optionality; a constructive case requires persuasive controlled results and a usable risk-benefit profile; a stronger case also needs evidence supporting sustained prescribing. None should borrow certainty from a different indication or assume that a favorable study automatically produces reimbursement and routine uptake.

This is not a price target or a forecast of incremental sales. No verified event-specific share-price reaction is assigned here. The useful conclusion is narrower: ADEPT has supplied observations worth examining, while the investment thesis still depends on the controlled experiment and on whether the resulting treatment proposition works outside a trial.

Source and evidence boundary

BioScore

3.2/10
High Risk