REVIEW OF: Tariot PN et al, NEJM 2021;385(4):309–319

STUDY TYPE: Randomized double-blind placebo-controlled trial

In the never-ending search for a safe and effective antipsychotic in dementia, the newest kid on the block is pimavanserin (Nuplazid), an antipsychotic with a unique mechanism of action based on 5-HT2A inverse agonism/antagonism and no effect on dopamine receptors. As a reminder, atypical antipsychotics (which are dopamine/serotonin antagonists) are modestly effective for treating agitation and psychosis in dementia, but they are saddled with a black box warning regarding increased incidence of mortality vs placebo. Though the absolute risk is small—about 4% mortality on antipsychotics vs 2% on placebo—the warning has caused regulatory authorities to create policies that have limited the use of antipsychotics in this population. 

Enter pimavanserin, which was approved for psychosis in Parkinson’s disease in 2016. Although it carries the same black box warning, there wasn’t evidence that it increased mortality in dementia-related psychosis early on. Its manufacturer, Acadia, has been funding clinical trials for psychosis in dementia. Thus far, one study was published showing no benefit over placebo at 12 weeks for patients with psychosis and Alzheimer’s disease residing in long-term care settings (Ballard C et al, Lancet Neurol 2018;17(3):213–222). 

The latest study (the HARMONY trial) enrolled patients with any type of dementia, not only Alzheimer’s disease. In this study, 392 patients with dementia-related psychosis were initially enrolled in an open-label trial of pimavanserin (20–34 mg/day). The primary end point was relapse of psychosis. Of these patients, 217 responded to the medication with symptom reduction and were then enrolled in a double-blind phase—105 were randomized to continuing pimavanserin, and 112 were switched to placebo. This phase was supposed to last 26 weeks, but it was terminated early because an interim analysis showed that pimavanserin was clearly more effective than placebo. Specifically, a relapse of psychosis occurred in 12 of the 95 patients assigned to pimavanserin (13%) and 28 of the 99 patients assigned to placebo (28%). Of note, only 44 patients in the pimavanserin group and 35 patients in the placebo group completed the 26-week trial. Pimavanserin and placebo were comparable in terms of side effects, though asymptomatic QTc prolongation occurred more frequently in patients who received pimavanserin.

Carlat Take 

This study tells us that if patients with dementia and psychosis respond well to pimavanserin over 12 weeks, then they should probably be continued on the medication to prevent relapse. However, there are a couple of flaws that shake our confidence in the study’s results. First, it’s possible that the “relapses” in the placebo group were in part just withdrawal symptoms from having discontinued pimavanserin. Second, the fact that different subtypes of dementia patients were enrolled makes it hard for us to decide which patients will respond to the drug. We will have to wait for more data for clarity, especially given pimavanserin’s high cost, which insurance will likely not cover for an off-label use. For now, the FDA has declined to approve the drug for this indication.