Acadia Plans Phase 3 After Alzheimer’s Drug Endpoint Miss

Catherine Owen Adams was already on Acadia’s investor conference call and webcast at 8:00 a.m. Eastern on Thursday, September 24, 2026. The company had just released the readout on remlifanserin, tested once daily against placebo for hallucinations and delusions in Alzheimer’s disease psychosis.

“This is a highly Phase 3–enabling set of data,” she said.

The 60-milligram arm had improved the primary hallucinations-and-delusions scale more than placebo by week six, yet the difference had come in just short of the pre-set threshold for statistical significance. The p-value was 0.0603. A key secondary measure of overall symptom severity had reached nominal significance on the same dose. The 30-milligram arm had shown little. Safety findings tracked close to placebo across both doses. When the market opened that morning, Acadia’s stock dropped sharply on the miss.

Adams held the call to the path ahead: keep the Phase 3 program moving, strip the weaker dose, leave the primary endpoint alone. The full patient numbers were still sitting there to be read.

The modified full analysis set counted 326 patients who had taken at least one dose, started with a baseline score of at least 10 on the primary scale, and finished at least one later assessment. Of them, 109 received 60 milligrams of remlifanserin once daily, 110 received 30 milligrams, and 107 received placebo. Mean age was about 75. Roughly two-thirds were women.

The primary endpoint measured change from baseline at week six on the Scale for the Assessment of Positive Symptoms-Hallucinations and Delusions, a clinician-scored total of those two psychotic features. The 60-milligram group improved by 12.6 points. Placebo improved by 10.4. The standardized effect size was 0.26. On the key secondary endpoint, the Clinical Global Impression-Severity scale for Alzheimer’s disease psychosis, which rates overall symptom severity from the clinician’s view, the same dose produced a 1.3-point drop from baseline against a 0.9-point drop on placebo. That yielded an effect size of 0.37 and a nominal p-value of 0.0077. The separation from placebo widened across the six weeks on both measures. The 30-milligram arm showed almost none of that movement, with effect sizes of 0.05 on the primary endpoint and 0.11 on the secondary.

Remlifanserin, also called ACP-204, is an investigational highly selective 5-HT2A receptor inverse agonist—a compound that turns down signaling at a serotonin receptor subtype long tied to psychotic symptoms. It works in the same target family as Nuplazid, Acadia’s approved treatment for Parkinson’s disease psychosis, but was built to reduce the risk of QT prolongation, a delay in the heart’s electrical reset after each beat that had limited higher doses of the older drug. The same molecule is already in a separate Phase 2 trial for psychosis associated with Lewy body dementia.

In the RADIANT Phase 2 safety data, adverse events, serious adverse events, and discontinuations due to adverse events occurred at rates similar to placebo across both doses. Electrocardiograms showed no QT prolongation signal versus placebo, and the dataset suggested no negative impact on motor symptoms or cognition. At the 60-milligram dose, only two adverse events rose above 2 percent of patients and above the placebo rate: somnolence in 3.2 percent and nausea in 2.4 percent. No individual adverse event appeared in more than 5 percent of patients on that dose. There were no deaths in the remlifanserin arms. Two deaths occurred on placebo.

About 30 percent of the more than 7 million people in the United States living with Alzheimer’s disease experience psychosis. The symptoms are hallucinations and delusions. They increase caregiver burden and the likelihood of nursing-home placement. There is no FDA-approved therapy specifically indicated for the condition.

“The Phase 2 RADIANT findings provide encouraging evidence that 60 mg once daily remlifanserin may have the potential to reduce hallucinations and delusions in patients with Alzheimer's disease,” Jeffrey Cummings said. “This patient population has a significant need for an efficacious medication that is well tolerated, convenient to administer, and is compatible with the many concomitant medications commonly used by patients with Alzheimer's disease.”

Sahil Pandey’s Reuters report carried the miss to a wider audience: the experimental drug for psychosis associated with Alzheimer’s disease had missed the main goal of the mid-stage study. The Phase 3 design choices came next. Acadia stripped the 30-milligram arm from Phase 3 and locked the program on 60 milligrams once daily. Enrollment continued in the two ongoing Phase 3 studies in Alzheimer’s disease psychosis while the company eyed modest refinements to the enrollment criteria—somewhat higher baseline psychosis—without a broad protocol rewrite.

“We really are trying to get that balance between something that doesn't have a meaningful impact on enrollment but does have an opportunity to increase our overall potential phase 3 efficacy,” Elizabeth H.Z. Thompson said on the call. In one analysis the company discussed, a refined higher-baseline threshold left about 80 percent of the Phase 2 participants still eligible and lifted the primary endpoint effect size to 0.33.

Asked about changing the primary endpoint, Catherine Owen Adams said, “at this point, we're not proposing that.” She put the case for pressing ahead in one stretch: “We believe with the high unmet medical need in this population, the efficacy that we've seen so far ... as well as the safety and tolerability profile, that this molecule offers the potential for patients with Alzheimer's disease psychosis, and it is absolutely worth our investment to move this forward into the phase 3 trial.” BMO’s Evan Seigerman said the near miss on the primary goal and the positive result on a key secondary measure suggested a potential path forward as Acadia modified the design of its ongoing late-stage studies.

The detailed RADIANT package still had a November date in Boston. That date was the Clinical Trials on Alzheimer’s Disease conference, November 16 through 19, 2026. Acadia planned an oral presentation there of the full Phase 2 RADIANT safety and efficacy results. Remlifanserin stayed in its separate Phase 2 trial for psychosis associated with Lewy body dementia as well.

“We are pleased that the Phase 2 RADIANT data support continuation of the Phase 3 ADP program, with new insights helping us to refine the program for future regulatory success,” Catherine Owen Adams said. Elizabeth H.Z. Thompson put the mid-stage purpose simply. “Our goal with this phase two was to get the information that would help us design our phase three program. And I think from that perspective, it has done what we needed it to do.”

The two ongoing Phase 3 studies in Alzheimer’s disease psychosis kept enrolling. Patients came in on the 60-milligram once-daily arm only.