FDA staff reveal flaws with Acadia's 2nd Nuplazid try in Alzheimer's psychosis

The FDA is poking holes in Acadia Pharmaceuticals’ second try for expanding Nuplazid into a new neurological disorder—and in doing so, the agency has revealed a chasm in the data that the company likely won’t be able to remedy without an additional clinical trial.

The FDA pointed out that Nuplazid’s benefit in a phase 3 trial for a broader psychosis population was actually driven by the Parkinson’s disease subgroup, according to a review document (PDF) prepared ahead of a Friday advisory committee meeting. The expert panel will discuss Nuplazid for treating hallucinations and delusions associated with Alzheimer’s disease psychosis.

Nuplazid is already approved in the U.S. for treating Parkinson’s disease psychosis. Acadia’s application in the broader psychosis indication was rejected last April as the FDA demanded more clinical data to examine the drug’s effect in each patient subgroup. In its resubmission, the California company is aiming for the Alzheimer’s subpopulation.

Previously, Acadia said Nuplazid’s showing in the Alzheimer’s subgroup in the phase 3 Harmony study wasn’t statistically significant because of the trial design. For its part, the FDA is now saying there’s more problem than just statistical powering.

Despite Acadia’s argument that Nuplazid showed consistent benefit, the FDA says the Alzheimer’s and Parkinson’s subgroups “appeared to respond differently” to Nuplazid in the Harmony trial. About two-thirds of patients in Harmony belong to the Alzheimer’s group, which mirrors the estimated demographic in the real world.

The Harmony trial hit its primary endpoint and stopped early after accumulating 40 relapse events, showing a significant risk reduction of 65% in favor of Nuplazid. However, only the treatment effect in the group of Parkinson’s disease dementia or dementia with Lewy body “appears to separate from placebo,” the FDA reviewers said.

After removing that small group of patients, the risk reduction dropped to 28%, with insufficient statistical evidence to suggest that the benefit didn’t happen just by chance, an FDA analysis showed.

Acadia believes the big response observed for the Parkinson’s group stemmed from the concomitant use of dopaminergic medications, which led to a more rapid relapse in this subgroup. The company has run some additional analyses, including a simulation of a final calculation when the effect of dopamine therapy winded down.

Because Harmony was designed for the all-comers psychosis group, Acadia is only using it as supporting evidence for the refiling. Instead, it’s leaning on a phase 2 Alzheimer’s-specific trial coded Study-019 as the main data package. There, rather than a direct rejection, the FDA said the trial features “could allow it to be considered an adequate and well-controlled trial suitable for regulatory decision making,” but “questions remain related to whether the methods of assessment of subjects' response are well defined and reliable.”

Study-019, conducted in the U.K., linked Nuplazid to a 1.8 advantage over placebo on a psychosis score called NPI-NH-PS at day 43. For its part, the FDA has also raised concerns that this magnitude of improvement on a 24-point scale may not be clinically meaningful. Besides, Nuplazid didn’t show any strong benefits on any of the secondary endpoints partly thanks to insufficient statistical power.

Acadia argues that Alzheimer’s psychosis and Parkinson’s psychosis are similar conditions. But based on the review document, the company will likely have a hard time convincing the FDA.

Industry watchers have largely written off the Alzheimer’s psychosis opportunity. As SVB Securities analysts said in a note last Wednesday, “We doubt that the FDA will change its mind regarding these issues, given that it should have known Acadia’s arguments already.” That said, the SVB team still gave Nuplazid a 35% chance of success in the disease, considering the agency has convened the advisory committee meeting to seek external input.