Bristol Myers Squibb has released the first clinical data on Cobenfy (xanomeline/trospium, also called KarXT) in psychosis associated with Alzheimer's disease, and the numbers look encouraging but prove nothing yet. Every result presented comes from open-label treatment, meaning all participants knew they were taking the drug and no one received a placebo for comparison.
Cobenfy is an oral drug approved in September 2024 for schizophrenia in adults. It works differently from older antipsychotics, targeting muscarinic receptors in the brain rather than blocking dopamine. The question for families dealing with hallucinations and delusions in dementia is whether that different mechanism also helps them — and the trial that answers it has not reported.
Medical information disclaimer: This article is for general educational purposes only and does not provide medical advice, diagnosis, or treatment. Always consult a physician or other qualified health professional about symptoms, medications, tests, or treatment decisions.
Table of Contents
- What the CTAD data actually showed
- Why "open-label" changes how much this means
- Safety and dropout signals worth watching
- The trial that decides it, and when
- What this means for families right now
- Frequently Asked Questions
What the CTAD data actually showed
At the Clinical Trials on Alzheimer's Disease conference on November 18, 2026, BMS presented results from two Phase 3 studies in its ADEPT program, according to the company's announcement. Both reported changes on the NPI-C hallucinations-and-delusions score, a clinician-rated measure of how frequent and severe those symptoms are. In the 12-week open-label lead-in of ADEPT-1, that score dropped a mean of 12.0 points from a baseline of 14.4 among 266 participants. Of 320 people enrolled, 273 — 85.3% — finished the period.
That is a large drop on a scale where the starting score was 14.4. ADEPT-3, a 52-week open-label extension, showed a mean 5.1-point fall from an open-label baseline of 8.9 in 270 participants. Participants entering an extension have usually already improved, which is why the baseline is lower. Here 273 of 401 people, or 68.1%, completed treatment.
Why "open-label" changes how much this means
Open-label means no placebo group and no blinding. In psychosis associated with dementia, that matters more than in most conditions: symptoms fluctuate on their own, caregivers and clinicians who expect improvement tend to report it, and placebo response in trials of behavioral symptoms is historically substantial. BMS described these as descriptive analyses, and said the double-blind data across the ADEPT program remain blinded.
In plain terms, nobody outside the trial — including the company — has seen the comparison that would separate drug effect from everything else. So the honest reading is narrow. The data show that people taking Cobenfy for Alzheimer's psychosis got better over 12 and 52 weeks, and that most of them stayed on it. They do not show that Cobenfy caused the improvement.
Safety and dropout signals worth watching
Treatment-emergent adverse events — any new medical problem during treatment, whether or not caused by the drug — occurred in 56.6% of ADEPT-1 participants, with 5.0% serious and 4.4% stopping the drug because of them. In ADEPT-3 the rate was about 65%, with 7.7% serious and 4.7% discontinuing. BMS reported no newly identified safety signals, meaning nothing appeared beyond what is already known from the schizophrenia program.
Reporting in Psychiatric Times nonetheless flagged the discontinuation figures and safety events as open questions. The 68.1% completion rate in the 52-week extension is the number to hold onto. Nearly a third of participants did not finish a year of treatment. Some of that is expected in a frail, older population over 12 months, but it sets a realistic ceiling on how many patients would stay on this drug long term.
The trial that decides it, and when
ADEPT-2 is the randomized, placebo-controlled study. It compares Cobenfy against placebo in roughly 400 patients with mild-to-severe Alzheimer's and moderate-to-severe psychosis, and it is registered as NCT06126224. Its topline readout has slipped to early 2027. The delays have a documented history.
In December 2025, BMS said it had excluded data from several ADEPT-2 sites before database lock after finding irregularities, and chose to enroll additional patients and continue the study rather than halt it. More recently the company cited slow enrollment in ADEPT-2 and ADEPT-4, plus slow relapse accrual in ADEPT-1. Until that readout, the Alzheimer's indication is a hypothesis with supportive open-label data behind it. Cobenfy earned $176 million in the first nine months of 2026 on its schizophrenia approval alone, and BioSpace reported that the dementia expansion rests on the 2027 results.
What this means for families right now
About half of the roughly 6 million Americans with Alzheimer's experience hallucinations or delusions, by BMS's estimate; other estimates put the share between a quarter and a half, or 1.75 to 3.5 million people. No therapy is FDA-approved for the condition. What gets used instead is off-label antipsychotics, which carry boxed warnings for increased mortality in elderly patients with dementia.
Cobenfy cannot be prescribed on-label for Alzheimer's psychosis today. Its approval covers schizophrenia in adults, and insurers generally follow the label. Practical steps if psychosis is affecting someone you care for:.
- Ask the prescriber to rule out treatable triggers first — urinary tract infection, pain, dehydration, constipation, new medications, poor sleep, or vision and hearing loss.
- Ask specifically about the boxed mortality warning on any antipsychotic being proposed, and what the plan is for reviewing whether it is still needed.
- Ask whether behavioral and environmental approaches have been tried: consistent routines, reduced evening stimulation, better lighting, not arguing with the content of a delusion.
- Ask the care team to search ClinicalTrials.gov for open ADEPT sites if trial participation interests you; enrollment in the program has been slow, which means some sites are still recruiting.
- Write down what the hallucinations or delusions are, how often they occur, and whether they distress the person. Symptoms that cause no distress and no danger may not need drug treatment at all.
Frequently Asked Questions
Can my doctor prescribe Cobenfy off-label for dementia psychosis today?
Off-label prescribing is legal, but Cobenfy is approved only for schizophrenia in adults, so insurance coverage for an Alzheimer's indication is unlikely and no efficacy data from a placebo-controlled trial exists to guide dosing or expectations in this population.
What is the NPI-C hallucinations-and-delusions score?
It is a clinician-rated measure of how often hallucinations and delusions occur and how severe they are. In ADEPT-1 the average starting score was 14.4; in the ADEPT-3 extension it was 8.9.
Why would a 12-point drop not count as proof?
Because every participant received the drug and none received placebo. Without a comparison group, there is no way to separate the drug's effect from natural symptom fluctuation, regression to the mean, and expectation on the part of raters and caregivers.
What should I watch for in early 2027?
The ADEPT-2 topline. It is the randomized comparison against placebo in about 400 patients, and it is the result that determines whether BMS can seek approval for this use.





