KarXT's 12.0-Point Open-Label Improvement Still Has to Survive Placebo

By
CTOL Staff Reporter
1 min read

Bristol Myers Squibb has a plausible path into Alzheimer's psychosis treatment. The first Phase III ADEPT numbers are only the beginning of the investment case.

A hallucination and delusion score starts at 14.4 and falls by 12.0 points. Before subtracting placebo, the improvement looks large enough to support a new franchise. Bristol Myers Squibb presented that preliminary result among 266 KarXT patients in the ADEPT-1 open-label lead-in at CTAD on 18 November 2026. We would pass on paying for a broad Alzheimer's psychosis expansion to Cobenfy, KarXT's commercial name, until randomized data assign a substantial share of that improvement to the drug. Improvement during treatment and improvement caused by treatment are different things.

We call that problem Placebo Compression: an impressive treatment-period improvement shrinks when the change in a randomized placebo group is subtracted. In ADEPT-1, 273 of 320 patients completed the 12-week open-label lead-in, or 85.3%. The 12.0-point fall from 14.4 was 83.3% of baseline; general severity fell 1.9 points from 4.5. Bristol Myers called its analyses descriptive, excluded patients still in the open-label period at cutoff and kept all double-blind results blinded. The placebo-adjusted difference is what investors ultimately pay for.

Bristol Myers's strongest commercial argument begins with the available treatments. On its 29 October 2026 earnings call, commercial chief Adam Lenkowsky described roughly 7 million Americans diagnosed with Alzheimer's disease and psychosis in 30% to 50% of them. Seven million times 30% to 50% yields a population scenario of 2.1 million to 3.5 million patients, without predicting sales. Lenkowsky said, "There are no approved treatments today in ADP" and noted the boxed warnings carried by antipsychotics for elderly dementia patients. Half receive treatment in long-term care, he said. The commercial mechanism is straightforward: institutions facing difficult existing choices gain a treatment they can use consistently, even if the incremental symptom benefit is modest.

Jeffrey L. Cummings of UNLV, quoted by Bristol Myers, said, "These preliminary findings further our understanding of the possible impact of KarXT in this setting". The clinical figures give that argument substance. Bristol Myers chief medical officer Cristian Massacesi called the ADEPT-1 improvement and ADEPT-3 safety findings "encouraging". During ADEPT-1's 12 weeks, 4.4% stopped for adverse events, 56.6% experienced treatment-emergent events and 5.0% had serious ones. Completion by 273 people establishes a practical foothold, and those patients matter more than an abstract addressable market. On those same 12 weeks, the 12.0-point decline was measured against baseline; a randomized comparator is what assigns credit for symptom relief. That difference determines the price of an efficacy claim.

The third-quarter 2026 accounts put the temptation in dollars. Cobenfy sold $57 million, up from $43 million in the third quarter of 2025, while Bristol Myers generated $13,069 million of total quarterly revenue. The $57 million over $13,069 million is 0.44% of company revenue. Nine-month Cobenfy sales reached $176 million versus $105 million a year earlier; quarterly operating cash flow was about $4.2 billion. Management said inventory destocking offset increased demand sequentially, making the sales line a somewhat noisy measure of uptake. The expansion prize is larger than the evidence supporting its price today.

Acadia supplied a reason to resist that extrapolation at the same conference. The 17 November 2026 RADIANT disclosure described a six-week placebo-controlled study of remlifanserin in Alzheimer's psychosis. SAPS-H+D fell 12.6 points on drug and 10.4 on placebo. Placebo's 10.4 over 12.6 is 82.5% of the drug-arm fall; the 2.2-point difference missed the primary endpoint at p=0.0603. A separate psychosis measure fell 7.3 points on drug and 5.9 on placebo, with nominal p=0.0385. RADIANT used a different drug and scale, so it warns about attribution rather than prescribing a discount for Cobenfy.

Bristol Myers has seen the same problem. In its ARISE trial announced on 22 April 2025, adjunctive Cobenfy improved schizophrenia symptoms by 14.3 PANSS points against 12.2 on adjunctive placebo; its modeled treatment difference of 2.0 points missed the primary endpoint at p=0.11. The indication and regimen differ, while Placebo Compression is recognizable. A big movement in the treatment arm can leave only a small incremental edge worth paying for.

ADEPT asks two questions with different commercial answers. In ADEPT-1, a 12-week open-label lead-in precedes 26 weeks of randomized withdrawal, where time to relapse is the primary endpoint. A relapse benefit would support keeping initial responders on therapy, a different commercial proposition from starting treatment across the larger population. ADEPT-2 and ADEPT-4 instead measure acute efficacy by the change in NPI-C hallucination and delusion scores. On 29 October 2026, Bristol Myers guided ADEPT readouts throughout 2027 and a potential event-driven ADEPT-1 interim analysis later in 2026. The acute comparison decides how much expansion value to assign.

Acadia won the controlled relapse comparison in HARMONY and still failed to secure a broader psychosis label. In Acadia's HARMONY trial, patients entered randomized withdrawal after 12 weeks of open-label treatment for dementia-related psychosis. Pimavanserin achieved a relapse hazard ratio of 0.353, with one-sided p=0.0023, yet received an FDA complete response letter on 5 April 2021. A second letter on 4 August 2022 cited limits to interpreting the Alzheimer's evidence and a HARMONY effect driven by Parkinson's disease dementia. ADEPT studies Alzheimer's disease alone, so that specific subgroup problem differs. One successful endpoint alone did not secure the new label.

ADEPT-3 offers a longer look at whether patients remain on treatment over 52 weeks. In ADEPT-3, 273 of 401 participants, or 68.1%, finished the 52-week open-label extension; 4.7% discontinued because of adverse events. Among extension participants, the score fell 5.1 points from a separate baseline of 8.9. The extension followed patients from ADEPT-1, ADEPT-2 and ADEPT-4, without splitting them by their earlier trial arms. Comparing its raw decline with ADEPT-1 would muddle the populations. Those figures concern extended treatment in a population with a different baseline. With a positive controlled result, retention and safety can help Bristol Myers persuade long-term care institutions.

Acadia already sells a psychosis treatment, albeit for a different disease. Nuplazid sold $202 million in the third quarter of 2026 under its Parkinson's disease psychosis indication. That exceeds Cobenfy's $57 million quarter in another patient population. With half of patients treated in long-term care by Bristol Myers's estimate, revenue depends on repeat treatment and institutional adoption. A clinical label opens a door; repeated treatment decisions turn the approval into revenue.

We would value Cobenfy on its existing $57 million quarter and exclude a broad Alzheimer's psychosis expansion until a substantial controlled symptom advantage earns its place. We would be wrong if a Bristol Myers public ADEPT-2 or ADEPT-4 release by our 30 June 2027 checkpoint showed at least 6.0 points of placebo-adjusted NPI-C hallucination and delusion improvement, our chosen threshold of half the 12.0-point open-label fall.

You May Also Like

This article is submitted by our user under the News Submission Rules and Guidelines. The cover photo is computer generated art for illustrative purposes only; not indicative of factual content. If you believe this article infringes upon copyright rights, please do not hesitate to report it by sending an email to us. Your vigilance and cooperation are invaluable in helping us maintain a respectful and legally compliant community.

Subscribe to our Newsletter

Get the latest in enterprise business and tech with exclusive peeks at our new offerings

We use cookies on our website to enable certain functions, to provide more relevant information to you and to optimize your experience on our website. Further information can be found in our Privacy Policy and our Terms of Service . Mandatory information can be found in the legal notice