ACADIA Pharmaceuticals Inc. (ACAD) Earnings Call Transcript & Summary

Okay. Good morning. I assume we're okay to start. Good morning, and welcome. I would first like to remind everyone to please mute your line when you're not speaking. For media and press, the FDA press contact is April Grant. Her e-mail and phone number are currently displayed. My name is Raj Narendran, and I will be chairing this meeting. I would now call the June 17, 2022, Psychopharmacologic Drug Advisory Committee Meeting to order. Dr. Joyce Frempong is the designated Federal Officer for this meeting and will begin with the introductions.

Good morning. My name is Joyce Frempong, and I'm the designated Federal Officer for this meeting. When I call your name, please introduce yourself by stating your name and affiliation Dr. Robert Baker.

Good morning. This is Robert Baker, Deputy Chief Medical Officer at Eli Lilly and Company, so I'm the industry representative.

Dr. Walter Dunn.

This is Walter Dunn, Assistant Clinical Professor at UCLA and the Greater Los Angeles VA.

Dr. Jess Fiedorowicz.

This is Jess Fiedorowicz, Professor at University of Ottawa.

Dr. Satish Iyengar.

My name is Satish Iyengar. I'm from the University of Pittsburgh where I'm Professor and Chair of the Statistics Department.

Dr. Sonia Krishna.

Good morning. This is Dr. Sonia Krishna. I'm affiliate faculty at UT Austin Dell Medical School.

Dr. Rajesh Narendran.

This is Raj Narendran. I'm a psychiatrist at UPMC, Professor of Psychiatry and Radiology at the University of Pittsburgh Medical Center.

Ms. Kim Witczak.

Good morning. Kim Witczak, Woodymatters, Drug Safety Organization out of Minneapolis.

Dr. Liana Apostolova.

This is Liana Apostolova. I'm I am the Barbara and Peer Baekgaard Professor in Alzheimer's Disease Research and Professor of Neurology from Indiana University.

Merit Cudkowicz.

I'm Merit Cudkowicz. I am Chair of Neurology at Massachusetts General Hospital and Professor of Neurology at Harvard Medical School.

Dr. Dean Follmann.

Yes. I'm Dean Follmann, Head of Biostatistics at the National Institute of Allergy and Infectious Diseases.

Colette Johnston.

Colette Johnston. I'm patient advocate and caregiver.

Dr. Madhav Thambisetty.

This is Madhav Thambisetty. I'm a senior investigator at the National Institute of Aging and Chief of the Clinical and Translational Neuroscience Section. I'm also an adjunct Professor of Neurology at the Johns Hopkins School of Medicine.

Dr. Billy Dunn.

This is Dr. Billy Dunn. I'm the Director of the Office of Neuroscience of the FDA.

Dr. Tiffany Farchione.

This is Tiffany Farchione. I'm the Director of the Division of Psychiatry at FDA.

Dr. Bernard Fisher.

This is Bernie Fisher. I'm the Deputy for Psychiatry at the FDA.

Dr. Paul Bossie.

I'm the Clinical Reviewer at the Division of Psychiatry for the FDA.

And Dr. Xiang Ling.

This is Xiang Ling, Statistical Reviewer at the FDA.

Okay. For topics such as those being discussed at this meeting, there are often a variety of opinions, some of which are quite strongly held. Our goal is that this meeting will be a fair and open forum for discussion of these issues and that individuals can express their views without interruption. Thus, as a gentle reminder, individuals will be allowed to speak into the record only if recognized by the Chairperson. We look forward to a productive meeting. In the spirit of the Federal Advisory Committee Act and the Government in the Sunshine Act, we ask that the advisory committee members take care that their conversations about the topic at hand take place in the open forum of the meeting. We are aware that members of the media are anxious to speak with the FDA about these proceedings. However, FDA will refrain from discussing the details of this meeting with the media until its conclusion. Also, the committee is reminded to please refrain from discussing the meeting topic during breaks or lunches. Thank you. Dr. Joyce Frempong will read the conflict of interest statement for the meeting.

The Food and Drug Administration is convening today's meeting of the Psychopharmacologic Drug Advisory Committee under the authority of the Federal Advisory Committee Act of 1972. With the exception of the industry representatives, all members and temporary voting members of the committee are special government employees or regular federal employees from other agencies and are subject to federal conflict of interest laws and regulations. The following information on the status of this committee's compliance with thorough ethics and conflict of interest laws covered by, but not limited, to those found at 18 USC Section 208 is being provided to participants in today's meeting and to the public. FDA has determined that members and temporary voting members of this committee are in compliance with Federal ethics and conflict of interest laws. Under 18 USD Section 208, Congress has authorized FDA to grant waivers to special government employees and regular federal employees who have potential financial conflicts when it is determined that the agency's need for a special government employee services outweighs his or her potential financial conflict of interest or when the interest of a regular federal employee is not so substantial as to be deemed likely to affect the entirety of the integrity of the services, which the government may expect from the employee. Related to today's discussion meeting, members and temporary voting members of this committee have been screened for potential financial conflicts of interest of their own as well as those imputed into them, including those of their spouses or minor children and, for purposes of 18 USC Section 208, their employers. These interests may include investments; consulting; expert witness testimony; contracts, grants, CRADAs; teaching, speaking, writing, patents and royalties and primary employment. Today's agenda involves a discussion of supplemental new drug applications 210793-008 and 207318-011, efficacy supplement resubmission for NUPLAZID pimavanserin tablet, submitted by Acadia Pharmaceuticals Inc. for the proposed treatment of hallucinations and delusions associated with Alzheimer's disease psychosis. This is a particular matters meeting, during which specific matters related to Acadia Pharmaceuticals Inc. supplemental new drug applications will be discussed. Based on the agenda for today's meeting and all financial interest reported by the committee members and temporary voting members, a conflict of interest waiver has been issued in accordance with 18 USC Section 208 (b) (1) to Dr. Walter Dunn. Dr. Dunn's waivers include stock holdings in 4 competing firms. The aggregate market value of his financial interest in the common of the 4 firms is between $17,500 and $37,500. The waiver allows Dr. Dunn to participate fully in today's deliberations. FDA's reasoning for issuing this waiver are described in the waiver document, which is posted on the FDA's website. Copy of the waiver may also be obtained by submitting a written request to the agency's Freedom of Information division, 5630 Fishers Lane, Room 1035, Rockville, Maryland 20857 or request may be sent via fax to (301) 827-9267. To ensure transparency, we encourage all standing committee members and temporary voting members to disclose any public statements that they have made concerning the product at issue. With respect to FDA's invited industry representatives, we would like to disclose that Dr. Robert Baker is participating in this meeting as a nonholding industry representative, acting on behalf of regulated industry. Dr. Baker's role at this meeting is to represent industry in general and not any particular company. Dr. Baker is employed by Eli Lilly & Company. We would like to remind members and temporary voting members that if discussions involve any other products or firms not already on the agenda for which an FDA participant has a personal, imputed financial interest. The participants need to exclude themselves from such involvement, and their exclusion will be noted for the record. FDA encourages all other participants to advise the committee of any financial relationships that they may have with the firm at issue. Thank you.

We will proceed with the FDA's opening remarks from Dr. Tiffany Farchione.

Good morning and welcome to the Psychopharmacologic Drugs Advisory Committee meeting. My name is Tiffany Farchione, and I'm the Director of the Division of Psychiatry here at FDA. Today, we will be discussing Acadia Pharmaceuticals' supplemental New Drug Application for pimavanserin for the treatment of hallucinations and delusions associated with Alzheimer's disease psychosis. The application under review here is a resubmission after a complete response action. In other words, the agency reviewed and did not approve a previous version of this application. I want to emphasize that the committee should not assume that the prior action reflects the agency's position on the current application. The applicant was previously seeking a general indication for the treatment of all dementia-related psychosis, regardless of the underlying disease responsible for dementia but with the current application, has narrowed the proposed indication to the treatment of Alzheimer's-related psychosis and has submitted a number of new analyses in an attempt to address the concerns outlined by the agency with the earlier decision. There are no new studies with this submission, but the agency has agreed to consider the additional analyses in the context of the indication the sponsor now seeks. It's important to acknowledge that the applicant's resubmission for this revised indication focused only on Alzheimer's disease and was discussed with us in multiple pre-submission meetings. And we prospectively agree that their current approach was reasonable and reviewable. Today, our team's presentations will briefly describe the regulatory history, including relevant aspects of the complete response decision and post-action discussions with the applicant, followed by our evaluation of the current application. The applicant is now seeking an indication for the treatment of hallucinations and delusions associated with Alzheimer's disease psychosis. Alzheimer's disease is the most common form of dementia in the United States. The latest estimate puts its U.S. prevalence at 6.5 million individuals. The pathological hallmarks of Alzheimer's disease include: Extracellular deposits of amyloid-beta, known as plaque; an intracellular aggregates of hyperphosphorylated tau, or neurofibrillary tangles. Although cognitive decline is the predominant symptom, neuropsychiatric symptoms, including hallucinations and delusions are common and severe. These neuropsychiatric symptoms cause profound distress for patients and their caregivers, are severely debilitating and are associated with a higher risk of rapid progression to severe dementia, death and out-of-home placement. I think that I advanced one too early. I apologize. Currently, there are no approved pharmacologic treatments for hallucinations and delusions associated with Alzheimer's disease psychosis. Off-label use of antipsychotic medications approved for other conditions occur. However, the American Psychiatric Association practice guideline on the use of antipsychotics to treat agitation or psychosis in patients with dementia notes that the benefits of antipsychotic medications are small at best. There is significant impressing unmet need for the treatment of hallucinations and delusions associated with Alzheimer's disease psychosis. Pimavanserin, the product currently under review, is a serotonin-selective inverse agonist that preferentially targets the 5-HT2A receptor subtype. It is an approved product indicated for the treatment of hallucinations and delusions associated with Parkinson's disease psychosis or PDP. With the current submission, the applicant is seeking a second indication, this time for the treatment of hallucinations and delusions associated with Alzheimer's disease psychosis or ADP. The applicant cites 3 sources of evidence to support this new indication. First, the prior approval of pimavanserin for the treatment of hallucinations and delusions associated with PDP, making the case that ADP and PDP are closely related conditions. Second, Study 019, which was a Phase II 12-week double-blind placebo-controlled study in subjects with Alzheimer's disease psychosis. The primary endpoint in this study was the change from baseline to day 43 on the Neuropsychiatric Inventory–Nursing Home Version psychosis score and the study was positive on the prespecified primary endpoint. Finally, Study 045, which was a Phase III relapse prevention study comprising a 12-week open-label period, followed by a 26-week randomized withdrawal double-blind period. This study included subjects with multiple subtypes of dementia, including a large Alzheimer's disease subgroup. The primary endpoint was time from randomization to relapse in the double-blind period, and the study was positive based on the prespecified primary endpoint. A note on the approved indication. The prior approval in PDP was based on Study 020, a Phase III randomized double-blind, placebo-controlled 6-week study of pimavanserin versus placebo in subjects with Parkinson's disease and psychosis that developed after the diagnosis of Parkinson's. Of the 185 subjects in the intention to treat analysis set, 46 had an MMSE for less than 25 and were considered a Parkinson's disease dementia subset. The primary endpoint was the change from baseline to day 43 on the scale for assessment of positive symptoms, Parkinson's disease or a SAPS-PD total score, which is a 9-item scale derived from the 20-item SAPS-Hallucinations plus Delusions or SAPS-H+D subscales. The study was positive on its prespecified primary endpoint. For the current submission, the applicant presents Study 019 as the primary evidence to support the Alzheimer's disease psychosis indication. Although the agency raised concerns that the design and conduct of this study and the complete response letter to the original submission, the applicant has successfully addressed these concerns with this submission. As previously noted, the study was positive on the primary end point at day 43. The agency seeks the committee's input on the overall persuasiveness of the data from Study 019. The applicant presents Study 045 as additional supportive evidence. This study was positive on the prespecified primary endpoint in a population consisting of subjects with several dementia subtypes. Primary endpoint results by dementia subgroup were strongest in subjects with Parkinson's disease dementia or PDD. The applicant has conducted a series of post-hoc analyses intended to show that pimavanserin's effect in the ADP subgroup is consistent with that in the PDD subgroup. As previously noted, the applicant is citing the prior approval of pimavanserin for the treatment of hallucinations and delusions associated with Parkinson's disease psychosis as evidence to support this application. It is common for companies to seek additional related indications following an initial approval. The agency considers that related initial indication as a source of evidence for subsequent supplemental applications so often requires only a single additional study in the new population. The applicant asserts that ADP and PDP should be considered closely related conditions. The design of Study 045 was based on the a priori assumption that this approach was reasonable, and the agency agreed with that approach. Although there are differences in the pathophysiology of Alzheimer's and Parkinson's disease, psychotic symptoms are present in both. However, the path of physiological underpinnings of psychosis in each condition are unknown. Nonetheless, the efficacy of pimavanserin in Parkinson's disease psychosis contributes to a prior expectation of benefit in a related condition such as Alzheimer's disease psychosis. On phase, the subgroup results of Study 045 may suggest differences in treatment response. However, the successful outcome of Study 019 may also suggest that these observed subgroup differences in Study 045 are not indicative of a lack of efficacy in Alzheimer's disease psychosis. The issues I've outlined, thus far, are all related to the evidence supporting effectiveness. Safety will not be a focus of today's discussion. The findings from the supplemental New Drug Application development program are largely consistent with the known safety profile of pimavanserin, and we do not have any concerns related to safety that would preclude approval. So the charge to the committee today is to discuss the evidence supporting the effectiveness of pimavanserin for the treatment of hallucinations and delusions in Alzheimer's disease psychosis, including the strength, limitations and potential contribution of Study 019, Study 045 and the prior approval of pimavanserin for the treatment of hallucinations and delusions associated with Parkinson's disease psychosis. Following that discussion, we will ask for your vote on the question, does the available evidence support a conclusion that pimavanserin is effective for the treatment of hallucinations and delusions in Alzheimer's disease psychosis? Thank you.

I apologize. I had to -- there's another section I had to read. Both the Food and Drug -- both the FDA and public believe in a transparent process for information gathering and decision-making. To ensure such transparency at the advisory committee meeting, FDA believes that it is important to understand the context of an individual presentation. For this reason, FDA encourages all participants including the applicants, nonemployee presenters, to advise the committee of any financial relationships that they may have with the sponsor, such as consulting fees, travel expenses, honoraria and interest in the sponsor, including equity interest and those based upon the outcome of this meeting. Likewise, FDA encourages you, at the beginning of your presentation, to advise the committee if you do not have any such financial relationships. If you choose not to address this issue of financial relationships at the beginning of your presentation, it will not preclude you from speaking. We will now proceed with presentations from Acadia Pharmaceuticals.

Members of the Psychopharmacologic Drugs Advisory Committee and members of the FDA, my name is Daryl Dekarske, and I'm the Head of Regulatory Affairs and Translational Sciences at Acadia. Thank you for the opportunity to introduce the pimavanserin development program, supporting our resubmission application for pimavanserin for the treatment of hallucinations and delusions associated with Alzheimer's disease psychosis. Unlike currently available multi-receptor acting antipsychotic drugs that primarily act by dopamine-receptor blockade, pimavanserin selectively targets serotonergic 5-HT2A receptors with inverse [ agonist/antagonist.] Acadia studied pimavanserin for the potential to treat psychosis in patients with Parkinson's disease or PDP, without adversely impacting their motor function, a core symptom of Parkinson's disease. This profile was demonstrated in a Phase III study in patients with PDP. And based on these results, the FDA granted pimavanserin a breakthrough therapy designation. In April 2016, FDA approved pimavanserin 34 milligrams once-daily for the treatment of hallucinations and delusions associated with PDP under the trade name NUPLAZID. Today, we'll discuss the data supporting a proposed indication for pimavanserin 34 milligrams once-daily for the treatment of hallucinations and delusions associated with Alzheimer's disease psychosis, or ADP. Evidence of pimavanserin's effectiveness for the newly proposed ADP indication comes from 3 independent placebo-controlled clinical studies: positive Study 019, which demonstrated a clinically meaningful benefit in patients with ADP; confirmatory evidence of effectiveness from positive Study 020 in patients with PDP, a closely related indication. Study 020 was the basis of the FDA approval of pimavanserin for the treatment of PDP, and the pimavanserin treatment effect in ADP patients will be evidenced in the presented clinical study data as being consistent with that observed in PDP patients and supportive evidence from positive Study 045 in patients with dementia-related psychosis, or DRP, in which pimavanserin-treated patients showed a highly statistically significant reduction of risk of psychosis relapse. Patients with ADP were the largest subgroup evaluated, and although not statistically significant, showed a clinically meaningful reduction in the risk of psychosis relapse. We also saw consistent evidence of efficacy across multiple supportive analyses in the ADP subgroup. Relevant FDA guidance states that effectiveness can be established by one adequate and well-controlled clinical study in a new indication for an approved drug supported by confirmatory evidence that comes from existing adequate and well-controlled clinical study data that demonstrated the effectiveness of that same drug for another closely related approved indication. Historical context helps to put into perspective the path taken to the proposed ADP indication. Shortly after the approval of pimavanserin for PDP, positive Study [ 019 ] results in patients with ADP became available. Importantly, we also observed that pimavanserin's treatment effect did not negatively impact a core symptom in these Alzheimer's disease patients, cognition. Acadia aligned with the FDA on a development plan to support a broad indication for the treatment of DRP, specifically, a randomized withdrawal Study 045. The goal of Study 045 was to demonstrate pimavanserin's efficacy for treating psychosis regardless of the underlying dementia diagnosis. Consistent with the clinical understanding of overlapping pathology and psychotic symptoms among dementia subgroups. The study design also had the benefit of mimicking the way patients with DRP are treated in the real world. It would limit the duration of potentially ineffective therapy and would also assess pimavanserin's maintenance of efficacy. The primary analysis evaluated risk of relapse in the overall DRP population. The percentage of patients among the dementia subgroups was targeted to be representative of their epidemiological prevalence. As a result of a prespecified interim efficacy analysis that showed a highly statistically significant reduction of the risk of relapse, the independent data monitoring committee recommended stopping Study 045 early. Acadia shared the study results with the FDA at a [ pre-sNDA meeting ]. In April 2021 FDA Complete Response Letter described concerns regarding a potential differential pimavanserin treatment effect among the dementia subgroups in Study 045. Although the study was not designed to evaluate the risk of relapse in individual dementia subgroups, a robust treatment effect in the Parkinson's disease dementia subgroup was noted, along with a lack of statistical separation in the other dementia subgroups. Concerns regarding the design and conduct of Study 019 were also raised. Following the Complete Response Letter, Acadia presented to FDA sensitivity analyses for Study 019 that confirmed the primary endpoint conclusions as well as the consistency of treatment effect observed in Study 020. We note that FDA have indicated in their briefing document that Study 019 was designed with features that could allow it to be considered adequate and well controlled and that the sensitivity analyses Acadia conducted would allow for FDA to rely on the data for regulatory decision-making. With respect to Study 045, Acadia also presented to FDA new analyses that supported both a consistent and clinically meaningful pimavanserin treatment effect across the dementia subgroups, including in the ADP subgroup. Further, these analyses indicated a basis for the robust findings in the Parkinson's disease dementia subgroup. FDA expressed a readiness to review a resubmission in support of treatment of ADP indication and, in February of this year, Acadia resubmitted. The pimavanserin clinical study efficacy data we provided in our resubmission and which we will present today demonstrate that pimavanserin statistically and meaningfully reduced psychosis and the risk of relapse across multiple clinical studies and measures without adversely impacting cognition or motor function. Importantly, since the approval pimavanserin for the treatment of PDP, an expanded clinical safety data set is now available and corroborates a favorable and differentiated safety profile. Further, 6 years of NUPLAZID post-marketing experience, greater than 44,000 patients with PDP provides continued reassurance of its favorable safety profile. Although safety is not a focus of today's meeting, it is important to consider pimavanserin's differentiated safety profile to inform the overall positive benefit risk in the context of the current treatment landscape and high unmet medical need of patients with ADP. You will hear later in our presentation how much distress psychosis can cause for patients with ADP and the resultant accelerated nursing home placement and increased risk of morbidity and mortality. Unfortunately, there remains no FDA-approved treatment for patients with ADP. While NUPLAZID is the only drug approved in the U.S. to treat PDP, payers require a strict diagnosis of PD for insurance coverage. Consequently, we see virtually no off-label prescriptions for uses outside of PDP. Today, health care providers are left to consider off-label use of available multi-receptor acting antipsychotics, which have not demonstrated efficacy in ADP and are associated with potentially serious safety issues, including adverse impacts on cognition and motor function. With this introduction in mind, here is the agenda for our presentation. Dr. Tariot will discuss the urgent unmet medical need for effective treatment of patients with ADP. Dr. Ballard and Dr. Hendrix will describe the pimavanserin clinical study efficacy results as well as the supportive analyses across these studies. Dr. Tariot will then briefly review key aspects of pimavanserin's safety profile. And Dr. Stankovic will present the benefit-risk [indiscernible] by Dr. Tariot to review the unmet medical need.

Thank you. Good morning. I'm Pierre Tariot. I'm an Internist and Geriatric Psychiatrist and Director of the nonprofit Banner Alzheimer's Institute. I was also closely involved in the 045 trial, the randomized withdrawal study in dementia-related psychosis. It's a privilege to speak with you today to share some of the background on Alzheimer's disease psychosis and the current unmet need that these patients, their families and their loved ones face. I have been caring for and studying patients with Alzheimer's disease and other dementias for more than 30 years. And I can tell you very simply that if left untreated, psychosis has significant and sometimes devastating consequences for our patients. But as you heard, there are no approved treatments for Alzheimer's disease psychosis, and current off-label options are woefully inadequate and often cause more harm than good. Let me begin with a little level setting. Dementia, is a common clinical syndrome that involves crippling cognitive impairment that affects social, occupational and even basic aspects of functioning. Dementia affects at least 7.9 million Americans. And as you've heard, there are various subtypes of dementia, with Alzheimer's being the most common accounting for nearly 70% of cases. The other common forms of dementia are vascular dementia, Lewy body dementia, Parkinson's disease and frontotemporal dementias. Bear in mind that the different dementias can have overlapping pathologies, including Alzheimer's and Parkinson's disease, and that older persons with advanced dementia can have more than one pathology. So these conditions are closely related to one another, and my own experience conforms with the epidemiological data. Next, neuropsychiatric signs and symptoms [indiscernible] dementias at some point in the course of illness, although their frequency time to onset pattern and severity do vary from patient to patient. Among these neuropsychiatric features, psychosis is particularly important. This term refers specifically to hallucinations and/or delusion, which occurs secondary to the underlying disease. Now let's focus just on Alzheimer’s. About 30% of these patients experience psychosis at any given time. This psychosis endures with generally waxing and waning symptoms with gradually increasing severity over time, leading to loss of independence as well as increased distress and burden to the patient, the family and caregivers. The distortion of reality, a core feature of psychosis, can further compound the disorientation that patients experience as their cognition declines, leading to further distress. These and other neuropsychiatric signs and symptoms also lead to decreased quality of life and as you've heard, earlier progression to nursing home care, severe dementia and even death. The takeaway is that there is a close relationship between the clinical manifestations of Alzheimer's disease psychosis and morbidity and mortality. However, despite this close relationship, as you've heard, there are currently no approved drugs in the U.S. for the treatment of patients with Alzheimer's disease psychosis. Now moving on to management. It's considered best practice to begin with nonpharmacological interventions, which I do always attempt. If such methods fail commonly, leaving us clinicians with little choice other than to deploy pharmacological interventions, namely antipsychotic agents. Based on clinical indications, antipsychotic agents are used for patients whose symptoms are frequent, severe, dangerous and cause significant distress. One review of Medicare claims data from 2008 to '16 showed that roughly 2/3 of patients with dementia-related psychosis were prescribed an antipsychotic off-label. But these agents come with significant limitations. Imagine feeling that you might choose a medication with both efficacy at best and an 80% to 90% chance of toxicity because there are no alternatives. The toxicities are due in part to the blockade of dopaminergic, histaminergic and muscarinic receptors contributing to cognitive impairment and increased mortality, among many other adverse effects. Let me briefly discuss the risk-benefit data for these agents. To highlight an example of the limited efficacy and high discontinuation rate we see with atypical antipsychotics. I show here the key findings from the CATIE-AD trial published in the New England Journal. It was the largest NIH-funded randomized placebo-controlled trial of the effectiveness of atypical antipsychotics for psychosis, agitation and aggression in persons clinically diagnosed with Alzheimer's disease. Three antipsychotics and placebo were compared. There were no significant differences among treatments with regard to time to discontinuation of treatment any reason, which was the primary outcome, as shown on the left here. Likewise, a key secondary outcome, the time to discontinuation of treatment due to adverse events, intolerability or death, favored placebo, as shown on the right. Together with Dr. Lon Schneider, I was the Co-Principal Investigator of the study as well as a site Principal Investigator. The first patient that I enrolled was an aged woman being cared for at home by her son. She gradually developed a fixed and increasingly frightening delusion that he was going to harm her. This came to a crisis one day when she pushed him through a plate glass window and attacked him with a fireplace poker. The ambulance brought her to our clinic at the family's request. This case is not an exaggeration. It is an example of what clinicians, patients and families in the real world are struggling with. This is why we sometimes feel that we have no choice but to prescribe atypical antipsychotics, but you can see how limited their effectiveness is. The CATIE-AD study also showed that these drugs were associated with significant decline in cognition shown by worsening on the Mini-Mental State Examination. The participants on drug declined an average of 2.4 point more than those on placebo over the 6-week study. This is equivalent to the decline typically seen over 1 year in dementia. I also want to point out that treating physicians in the trial were likely to switch medications quickly due to lack of efficacy or to adverse effects. So as this trial illustrates, despite the hope that the antipsychotic drugs would be more effective because [indiscernible ] their use. Only a small fraction of patients had both benefit and few or no side effects. The data from this study illustrate what's seen in the rest of the [indiscernible] is high. In addition, antipsychotic drugs increased the risk of death in elderly patients with dementia-related psychosis. The mortality data here comes from the 2005 meta-analysis by Lon Schneider and colleagues show an increased risk [indiscernible]. The FDA had previously conducted its own meta-analysis and came to the same conclusion. These pronounced negative impacts of [indiscernible] and cognitive function are reflected in product labeling as well as our use in clinical practice [indiscernible] the American Psychiatric Association guidelines recommend judicious use of antipsychotics when non-pharmacological therapy alone is ineffective. An individualized treatment plan involving a full discussion of benefits and risks is developed with each patient and family. The medication has started off-label. And if no significant response is seen after 4 weeks, it's to be withdrawn. Even if a patient does respond, we make the decision with the patient and family to attempt to taper and ideally withdraw the medication as soon as possible because of its known toxicities hopefully within 4 months of treatment initiation. So this is the clinical context for today's presentation. The constantly interfaced with the predicament of not treating patients who are desperately in need [indiscernible] marginal, loading the APA marginal efficacy at best available in highly problematic options. Their condition is serious, and the symptomatic consequences can be life altering. Effective and safe treatment of these symptoms means relief of distress, restoration of dignity and allowing our patients to remain at home instead of being institutionalized. This is what is at stake. It would be immensely valuable for patients and their families even to the health care system at large, the therapeutic options is as efficacious if not more so, then available antipsychotics and one that was not associated with their significant toxicities. It would also be immensely valuable if that's safe and effective. It was recognized by health authorities as appropriate for clinical use. Thank you for your attention. I'll now invite Dr. Clive Ballard to speak to Studies 019 and 020.

Thank you. Good morning. I'm Clive Ballard, an academic Old-Age Psychiatrist from the U.K. I've been researching dementia for most of my career and have led many of the pivotal randomized clinical trials focusing on pharmacological and nonpharmacological treatments for psychosis and other neuropsychiatric symptoms in people with dementia. I've also led much of the work demonstrating the limited benefit and significant harms of atypical antipsychotics in these individuals. I know firsthand from my work with patients that the related psychosis across dementias can be devastating and debilitating, and I'm passionate about the development of safer and more effective treatments for these most vulnerable group of patients. Whilst I treat patients in the United Kingdom, much of the practice is the same, and we experience the same treatment gaps that are evident in the United States. I was the principal investigator of Study 019, Chaired the Steering Committee and was closely involved in the design of Study 020. The evidence of efficacy for pimavanserin in ADP is primarily derived from those positive studies. Further supportive evidence comes for efficacy from Study 045 in patients with dementia-related psychosis and additional post-hoc analyses of the Alzheimer's disease psychosis subgroup supporting consistent benefit of pimavanserin treatment. I will discuss biological and clinical evidence regarding the close relationship between Alzheimer's disease psychosis and Parkinson's disease psychosis and data from Studies 019 and 020, providing evidence for the efficacy of pimavanserin in Alzheimer's disease psychosis. Additionally, I would like to address some pertinent discussion points raised by FDA colleagues, namely, the validity and reliability of the neuropsychiatric inventory as the primary outcome measure in Study 019, the meaningfulness relevance, and consistent pattern of the treatment effect, the durability of the effect when controlling for the natural fluctuations in Alzheimer's disease psychosis severity and the lack of separation on the secondary outcomes for symptoms other than hallucinations or delusions. Firstly, let me discuss the relationship between Alzheimer's disease psychosis and Parkinson's disease psychosis. Mechanistically, there are substantial similarities between Parkinson's disease psychosis and Alzheimer's disease psychosis. Common brain regions are involved in both conditions. For example, visual hallucinations are associated with hypometabolism or [ gray matter ] atrophy in the occipital Cortex and visual association areas of the temporal cortex in both disorders based on neuroimaging and postmortem studies. The serotonergic system has been highlighted as a key neurochemical underpinning in both disorders based on neuroimaging, postmortem and genetic polymorphism studies. In addition, there is considerable pathological overlap. More than 90% of people with Parkinson's disease dementia have significant Alzheimer pathology. And even in the absence of dementia, almost all people with Parkinson's disease have at least some amyloid plaque pathology. The clinical picture is also similar in regard to the psychotic symptoms experienced by people with Alzheimer's disease and Parkinson's disease, which are clearly distinct from major psychotic disorders such as schizophrenia. In both conditions, most hallucinations are in the visual modality, usually of people, animals or strangers, the latter often accompanied by the delusional belief that strangers are living in the house. Although the presentation is extremely similar, visual hallucinations do have a higher frequency and are less likely to spontaneously recover in people with Parkinson's disease. In both Alzheimer's disease psychosis and Parkinson's disease psychosis, delusions are simple with common themes such as theft, harm and infidelity. Much of my work is focused on understanding the natural history of psychosis in people with Alzheimer's disease, which shows a fluctuating pattern of recovery followed by relapse or the emergence of new psychotic symptoms. We found that 68% of people recover from their psychotic symptoms by week 12, but majority of these individuals experience a relapse or the emergence of new psychotic symptoms over the subsequent 6 to 12 months. Typically across clinical trials in this area, there is approximately a 50% short-term improvement in the placebo group, probably driven by a Hawthorne effect. These 2 effects combined make design trials for Alzheimer's disease psychosis challenging. And to measure acute treatment response, 6 weeks is the optimal time point for primary assessment. Study 019 in patients with Alzheimer's disease psychosis was conducted in 133 nursing care homes with 20 medical sub-investigators in the U.K. Brief psychosocial therapy was administered during the screening period to mirror clinical guidelines and ensure that only patients requiring a pharmacological treatment enter the randomized period of the study. Patients were randomized 1:1 to pimavanserin, 34 milligrams once-daily or placebo. The primary endpoint was changed from baseline to week 6 in the Neuropsychiatric Inventory-Nursing Home Version Psychosis Score. 6 weeks was chosen as the optimal time point to assess pimavanserin's effect on the speed of symptom recovery. Further assessments from week 6 through week 12 were included principally to address adverse cognitive and global effects with treatment, an important objective given the well-documented adverse impact of dopaminergic atypical antipsychotics on cognition, mobility and motor function. The neuropsychiatric inventory was chosen as the primary measure based on extensive reliability and validation studies and its use in more than 300 clinical studies in Alzheimer's disease. Of particular note, the neuropsychiatric inventory has good concurrent validity with other measures of Alzheimer's disease psychosis. The scale has 12 domains, 2 of which measure psychosis, hallucinations and delusions, respectively. Each domain assesses both symptom severity and frequency to produce a total maximum score of 24 for delusions and hallucinations. Whilst the total score is a multiple of frequency and severity, the frequency and severity scores can be examined separately to give a clear picture of the benefit for specific individuals. Raters were thoroughly trained on the neuropsychiatric inventory. To mitigate any potential for expectancy bias, different raters will utilize the consecutive visits for the same patient. Neuropsychiatric inventory raters were centrally trained by MedAvante. More than 200 NPI assessments were audio recorded to check and ensure high quality. Regular calibration assessments were performed to confirm adherence to standardized procedures with feedback to raters. The raters also received regular refresher training. In addition, the informants were all care staff who knew the individual participants well, and all care staff were trained in the neuropsychiatric inventory to improve the quality of the informant information. These thorough procedures produced high inter-rater reliability, well in excess of 0.9, which is exceptional for a measure of neuropsychiatric symptoms. Now turning to the results. The study enrolled in elderly and frail population with a mean age of 86 years and a modest ethnic diversity representative of U.K. nursing homes. The patients included a representative of a care home population with severe dementia and reflect those patients most in need of treatment. These patients suffered from many comorbidities. And this is, in fact, 1 of only 4 studies that is focused on psychosis treatment in people with severe dementia. The study met its primary endpoint, demonstrating a statistically significant greater mean reduction from baseline to week 6 in the Neuropsychiatric Inventory Psychosis Score for pimavanserin compared to placebo. The standardized effect size versus placebo was a Cohen's D of 0.32, which is clinically meaningful and compares favorably with previous studies of atypical antipsychotics, where effect sizes are less than 0.2. We also evaluated the number of people achieving 30% and 50% improvement on the neuropsychiatric inventory psychosis score. The thresholds usually considered to represent a meaningful benefit in studies of neuropsychiatric symptoms in people with Alzheimer's disease. The study demonstrated statistically significant benefit at week 6 for participants with Alzheimer's disease psychosis treated with pimavanserin compared to those receiving placebo for both thresholds, with the numbers needed to treat or NNTs of 6 and 7 at 30% and 50%, respectively. For context, the 18% greater number of people improving by 30% relative to placebo compares favorably with an 11% advantage from a meta-analysis of studies of atypical antipsychotics. And importantly, benefits compared to placebo are also seen across the full spectrum of improvement. To further put this data into context and help illustrate the tangible benefits of patients and their caregivers, I've split the NPI data to show the frequency and severity scores for improvement. First in delusions. 25% of people treated with pimavanserin improved by 3 or 4 points on frequency, representing a change from multiple times a day to less than once a week. And 30% of people treated with pimavanserin improved by 2 or 3 points on severity, representing a change from severe distress to no distress. For both frequency and severity, these are highly impactful changes that give patients and caregivers significant relief from the terrible burden of psychosis, moving patients towards fewer weekly symptoms and less severe distress. Last participants experienced hallucinations at baseline, and they were generally less severe than the delusions. Nevertheless, we see meaningful improvement in both frequency and severity of hallucinations with tangible benefits for patients and caregivers. Now let's review the subgroup analyses. All subgroups favor pimavanserin and support the primary analysis, including by dementia severity and Alzheimer's disease psychosis that has previously required treatment with atypical antipsychotics. Whilst the benefit observed in the less severe group was modest, there was a particularly favorable response in people with severe psychosis, those patients the most in need. 89% of participants with severe psychosis at baseline achieved clinically meaningful benefit on pimavanserin compared to 43% on placebo. The effect size was substantially higher, greater than 0.7, a large effect size, representing a 4.43-point advantage for the pimavanserin-treated patients compared to placebo with an NNT of 3. Importantly, these data suggest that those individuals with the most frequent and most distressing symptoms are the individuals who benefit the most from pimavanserin. Assessment after 6 weeks were all exploratory in nature and focused on safety. As noted earlier, there's a substantial placebo effect up to week 4. And we know from studies focusing on the natural history of psychosis in people with Alzheimer's disease that many patients spontaneously recover over 12 weeks. It is not, therefore, surprising that whilst the benefits of pimavanserin for the treatment of Alzheimer's psychosis were maintained to week 12, there was also improvement in the placebo-treated group by the week 12 time point, and there was no significant difference between pimavanserin and placebo at that later time. To further address questions around the 12-week outcomes, we assess time to improvement for responders who improved by at least 30% from baseline. The figure on the left shows patients with symptom improvement at a single time point. The figure on the right shows patients with improvement confirmed for 2 consecutive time points indicating sustained improvement. This analysis reduces the influence of fluctuations in symptoms occurring as part of the natural course of Alzheimer's disease psychosis and highlights the individuals with a meaningful and durable treatment response. This confirms the significant acceleration of treatment response and the extended benefit for pimavanserin over placebo for the full 12 weeks of the study. Similar results were also observed for responders who experienced 50% or more symptom relief, with even greater benefit over the full 12 weeks of the study. Now let's review secondary and exploratory end points. The FDA briefing document highlighted the absence of benefit on secondary outcomes, but it's important to emphasize that none of the secondary outcomes measured hallucinations and delusions. Agitation, as a commonly co-occurring neuropsychiatric symptom, was measured as a secondary outcome, and no significant benefit was identified overall. It is interesting to note that in a subsequent publication, we demonstrated a significant benefit in agitation amongst people with a 50% improvement in psychosis. I would interpret this as an additional benefit in treating psychosis rather than a primary effect on agitation. In the context of this study, the Clinical Global Impression Scale was a completely global outcome encompassing cognition and function as well as psychosis and other neuropsychiatric symptoms. This was mainly undertaken to evaluate whether there are any detrimental outcomes on overall function. And importantly, no detrimental impact was observed. Now let's look at cognitive function. As Dr. Tarrio pointed out, a decline in cognitive function is a known side effect of currently used off-label antipsychotics. Pimavanserin treatment had no negative impact on cognitive function. In Study 019, the mini-mental state examination, or MMSE, was used to measure cognitive function at baseline and throughout treatment. Here are the results demonstrating no decline in mean MMSE in pimavanserin-treated patients or difference from placebo-treated patients. We also observed no negative effect on motor function as measured by the Unified Parkinson's Disease Rating Scale or UPDRS Part II. Please note that on this scale, a decrease in score signifies improvement. Scores remain consistent over time, demonstrating no negative impact on motor function, a significant benefit compared to the impact observed with atypical antipsychotics. To conclude, Study 019 demonstrated positive and meaningful efficacy of pimavanserin in Alzheimer's disease psychosis. The study met its primary endpoint with clinically meaningful treatment response; accelerated time to symptom improvement; a consistent pattern of benefit across subgroups and sensitivity analyses; and importantly, severe patients experienced the greatest benefits. Additionally, prespecified safety endpoints demonstrated no impact on cognition or motor function. Now turning to efficacy evidence from Study 020, which led to pimavanserin's FDA approval in patients with Parkinson's disease psychosis. Briefly, study 020 was a randomized, double-blind, placebo-controlled outpatient study in patients with Parkinson's disease psychosis, which, as noted, is a closely related condition to Alzheimer's disease psychosis with similar clinical symptoms, similar treatment response and similar underlying mechanisms for psychosis with significant pathological overlap. Patients had a mean age of about 72 years. Brief psychosocial treatment was again utilized during the screening period. Participants meeting eligibility criteria at the end of screening were randomized 1:1 to either placebo or pimavanserin 34 milligrams once-daily for the duration of the double-blind treatment period. The primary efficacy endpoint was the mean change in the SAPS PD score from baseline to week 6. The SAPS PD is derived from the well-established scale for the assessment of positive symptoms to evaluate hallucinations and delusions. This assessment approach is similar to the neuropsychiatric inventory used in Study 019. The treatment difference based on all randomized patients was 3.06 points with the [ coins D ] of effect size of 0.50. Of note, patients with cognitive impairment at baseline experienced an even greater pimavanserin treatment effect compared to placebo. Now let's review Study 020 results in relation to Study 019. The outcomes from Studies 019 and 020 show a consistent treatment effect. In both studies, patients treated with pimavanserin experienced about 2x greater improvement in symptoms at 6 weeks as compared to placebo groups. In addition, it is important to note the similarity in placebo response in both studies, reinforcing the discussion earlier about the consistent and substantial placebo response over the first 4 weeks in our randomized controlled trials focusing on Parkinson's disease psychosis and Alzheimer's disease psychosis. The similarity in treatment benefit is further illustrated in the responder analysis examining clinically meaningful improvement. Pimavanserin-treated patients experienced more symptom reduction compared to placebo of both 30% and 50% improvement cutoffs in both Alzheimer's psychosis and Parkinson's disease psychosis. To conclude, Studies 019 and 020 provide evidence of efficacy supporting pimavanserin for the treatment of patients with Alzheimer's disease psychosis. Study 019 was an adequate and well-controlled study but greatly informed our understanding of patients with Alzheimer's disease psychosis. It met its endpoint, demonstrating statistically significant, but more importantly, clinically meaningful reductions in psychosis symptoms for elderly patients with substantial comorbidities and was especially effective in those with severe symptoms. Study 020 provides confirmatory evidence in the closely related condition of Parkinson's disease psychosis. The consistent treatment response between these 2 studies supports a common clinical presentation of psychosis and a similar response to effective antipsychotic treatment. Pimavanserin would be a substantial advance in the treatment of our patients with Alzheimer's disease psychosis and would address a critical unmet need. Thank you for your time. I'll now turn the presentation to Dr. Hendrix.

Thank you. and good morning. I'm Susan Hendrix, a statistical consultant who specialized in neurodegeneration for the past 19 years. I will first describe data from Study 045 and then review exploratory analyses that provide additional supportive evidence of pimavanserin's effect in patients with ADP. Study 045 evaluated the durability of effect in pimavanserin in patients with dementia-related psychosis or DRP. The study was a double-blind, placebo-controlled, randomized withdrawal design, treating all patients first in a 12-week open-label period and then randomizing patients to continue treatment or switch to placebo. This mirrors clinical practice and assesses durability of effect. All patients began on pimavanserin 34 mg once-daily with the possibility of an early adjustment to 20 mg. Patients who exhibited a response at both weeks 8 and 12 were then randomized in the double-blind period. The primary efficacy endpoint was time from randomization to relapse of psychosis based on blinded independent adjudication in the double-blind period. Aligned with FDA, a prespecified interim efficacy analysis was performed after 40 relapse events using the [ no Brian forming ] stopping boundary of a one-sided alpha 0.0033. All analyses were prespecified for the ITT analysis set in all DRP patients. This study was conducted in an elderly population, 2/3 of whom had Alzheimer's disease. The SAPS-H+D and MMSE scores reflect moderate dementia and moderate to severe psychosis. The baseline demographic and disease characteristics were similar at randomization and were also balanced between a pimavanserin and placebo groups prior to the double-blind period. During the open-label period, pimavanserin treatment resulted in substantial improvement in psychotic symptoms, as shown in this figure, with a mean reduction of nearly 20 points on the SAPS-H+D score. Additionally, the improvements observed in the ADP and PDP subgroups were very similar to the overall DRP population. Approximately 60% to 70% of patients experienced sustained response and were randomized into the double-blind period. Complete symptom resolution was achieved in approximately 20% of patients overall. The study met the primary endpoint at the interim efficacy analysis with a 2.8-fold risk reduction and a p-value of 0.002. The hazard ratio was 0.35, showing pimavanserin significantly reduced the risk of relapse of psychosis in the overall VRP population, meeting the prespecified stopping criteria. The independent data monitoring committee recommended stopping the study due to this robustly positive efficacy finding. Additionally, the second -- the study met the key secondary endpoint of time to all-cause discontinuation, a measure of both efficacy and tolerability, demonstrating significantly lower discontinuation with pimavanserin versus placebo. The study was positive overall in DRP, and we additionally explored the contribution of dementia subgroups, although the study wasn't powered to show statistical differences. In patients with ADP, we observed a hazard ratio of 0.62, consistent with the clinically meaningful 40% reduction in risk that was not statistically significant. For patients with PDD, which represented about 18% of the overall population, the hazard ratio was 0.05, corresponding to a remarkably high 95% reduction in risk. Based on these results, the FDA questioned the originally planned broad indication for the treatment of DRP. We also wanted to understand why PDD performed better than the other dementia subgroups. Due to the exploratory nature of these analyses, all p-values are nominal and used for descriptive purposes only. The Kaplan-Meier plot on the left shows the placebo group, which has clear heterogeneity between PDD and the other subgroups. But on the right, for pimavanserin, we see homogeneous maintenance of response across dementia subgroups. These results suggest that the dementia subgroups differ in pattern of response only when pimavanserin is withdrawn, resulting in the larger treatment difference observed for PDD. This faster relapse rate after treatment withdrawal for patients with PDD is likely due to the use of dopaminergic medications such as levodopa, which are known to contribute to the psychosis symptoms. Non-PDD patients taking these medications showed consistent results with PDD. Finally, the next several slides will focus on supportive exploratory results just within the ADP subgroup. In addition to the ADP subgroup, I'll present the ADP 34 mg subgroup, as 34 mg is the dose that was used in the 019 and 020 studies and is the FDA-approved dose for the treatment of PDP. In Study 045, all patients who were stabilized on pimavanserin 34 mg prior to randomization were either randomized to stay on pimavanserin 34 mg or switch to the corresponding placebo arm. Nearly all patients were stabilized on pimavanserin 34 mg. In fact, only 7 patients or 6% received pimavanserin 20 mg. The full ADP subgroup demonstrated a nearly 40% reduction in the hazard of relapse, corresponding to the Kaplan-Meier curve shown here on the left. And on the right, the ADP 34 mg group is shown with a 53% reduction in relapse risk. An exposure response analysis was performed to assess the relationship between PK exposure and the risk of relapse based on DRP and ADP population. In the ADP subgroup specifically, this analysis gives a hazard ratio of 0.47 corresponding to a 53% reduction in risk of relapse at the median AUC. We agree with the FDA's conclusion in their briefing document that within ADP, higher PK exposures were associated with a higher relapse-free probability. This analysis indicates that the treatment effect in the ADP subgroup is due to a pharmacologic effect of pimavanserin, consistent with the original effect seen in the prespecified primary analysis for overall DRP, providing another source of evidence that supports the effectiveness of pumevanserin in ADP patients. In order to explore different responder rates, we started by comparing the percentage of patients within each treatment arm who did or did not experience symptom worsening. You can see that 60% of pimavanserin patients did not worsen compared to 48% on placebo. Among those who did worsen, the opposite relationship is observed on the right, with fewer patients on pimavanserin experiencing any worsening. When we further break down those who worsen by additional threshold of worsening, we see that at all levels of worsening on the SAPS-H+D scale, pimavanserin shows less worsening than placebo. And in fact, at a 6-, 9- or 12-point threshold, pimavanserin has half as many patients experiencing those higher levels of worsening, consistent with a clinically meaningful effect. Similarly, we observed this pattern of response with continued pimavanserin treatment versus placebo when assessing the CGI-I, a global clinical assessment of psychosis, with the majority of patients improving or remaining stable, again, avoiding the clinically impactful worsening observed in the placebo arm. Exploratory endpoints also show consistent benefit in Study 045 and reflects several perspectives of patient well-being in addition to the primary relapsed criterion, which was assessed by blinded raters. The SAPS-H+D is based on the clinician's direct assessment of symptoms and the CGII as a clinician's global assessment. The [ Zarit Burden ] inventory reflects the caregivers' burden and the quality-of-life scale on the bottom assesses quality of life of the patient. The exploratory variables of SAPS-H+D and the CGI-I, both achieved a disc significance for the ADP subgroup. The last exploratory outcomes there at burden inventory and quality-of-life scale also showed directionally consistent effect. All of these perspectives show consistent and meaningful effects supporting pimavanserin in ADP. Realizing that the study wasn't powered to assess dementia subgroups and knowing the potential for imbalances in subgroups, we conducted a covaried adjusted model to correct for potential confounding factors known to be clinically important. Here, I show the results of that model as well as several models proposed by my FDA colleagues. It is important to note that regardless of the approach used for covaried adjustment, the resulting point estimates for hazard ratios are consistent and all fall within a clinically meaningful range of 0.48 to 0.64 for the ADP subgroup overall and 0.35 to 0.49 for the 34 mg dose, which is our target. To summarize, the evidence of efficacy is consistent and clinically meaningful. This efficacy has been observed across studies, including Study 019 in the target population of ADP; Study 020 in a closely related condition of PDP; and from positive Study 045 in BRP, with exploratory analyses in the large ADP subgroup that positively inform our understanding of treatment effects for pimavanserin. The totality of efficacy data presented by Dr. Ballard and myself support a true and meaningful benefit of pimavanserin for patients with ADP. Thank you. I'd now like to turn the presentation to Dr. Turner to present the safety data.

Thank you, Dr. Hendrix. My name is Mary Ellen Turner, and I'm Senior Vice President of Pharmacovigilance and Corporate Safety Officer at Acadia. While we agree with the FDA, there are no safety issues to discuss, I'd like to provide information regarding the favorable tolerability and safety profile of pimavanserin to [indiscernible] the benefit risk discussions. Pimavanserin has a well-characterized and favorable safety profile. Across the clinical development program, more than 3,500 patients have been exposed to pimavanserin. This expanded safety data set includes the largest clinical program in patients with neurodegenerative disease or MDD. Pimavanserin post-marketing experience stands more than 6 years and 44,000 PDP patients. The safety profile in the Alzheimer's disease population is favorable and consistent with the known safety profile of pimavanserin. I will present the key safety and tolerability features that differentiate pimavanserin from the current standard of care. These include favorable mortality trends as well as no negative impact on cognitive motor function. Let's review mortality findings, which included over 1,500 elderly patients from our clinical trial program in NDD comparing pimavanserin with placebo patients. In the first line, you will see the incident rate ratio of 1.02 for death within 30 days of last treatment received. For death within the study intended treatment period plus 30 days, the incident rate ratio was 1.28. For clarity, the difference between these 2 analyses is 2 patient deaths that occurred more than 30 days after discontinuing therapy, but still within the intended study period. Both show wide confidence intervals due to small sample size relative to our post-marketing experience. These mortality rates are lower than the mortality rates seen in the original PDP safety data set. And as our clinical safety data set increases in size, the mortality point estimate has become more precise and trend towards placebo. Additionally, since pimavanserin's approval in 2016, Acadia has closely monitored the safety profile in the post-marketing setting. Large observational studies comparing pimavanserin mortality rates with antipsychotics used off-label provide real-world evidence in populations that complement the mortality analysis of pimavanserin from clinical trials. Here, we present two recent large Medicare claims data studies of mortality in patients with Parkinson's disease and in patients with PDP treated with pimavanserin or other antipsychotics. [ Nonfoldering ] colleagues have evaluated all-cause mortality in patients with Parkinson's disease and reported a statistically significant hazard ratio of 0.78 favoring pimavanserin. A subsequent Acadia-sponsored Medicare safety study by Layton and Colleagues demonstrated identical findings with a hazard ratio of 0.78. As previously noted, in our clinical studies, we've observed that pimavanserin has no negative impact on cognitive function. Here, I share data from the open-label period of Study 045. The name change from baseline to week 12 in MMSE score was 1.0. During the double-blind period, there is no decline in meeting MMSE in pimavanserin-treated patients. Here, you can see the full picture of the Study 045 completers, starting from open-label baseline to the 26-week double-blind period, again, showing stability in the MMSE. Additionally, here are the MMSE findings for pimavanserin compared to those for other antipsychotics used in elderly dementia patients taken from the Schneider Med analysis that Dr. Tarrio presented earlier. Again, we see no negative impact on cognitive function with pimavanserin in contrast to other antipsychotics. Additionally, in Study 045, there was no observed worsening in motor function in pimavanserin-treated patients as measured by the change in Extrapyramidal Symptoms Rating Scale A, or ESRSA. During the open-label period, the ESRSA was measured at baseline and at week 12 and showed a trend towards improvement and no worsening of motor function. Again, here are decreasing score signifies improvement. Shown here is the double-blind period during which the main change from baseline was small and similar in the 2 treatment groups at all time points. In conclusion, pimavanserin has a well-established consistent and favorable safety profile across the largest clinical safety data set in patients with MDD, supported by favorable findings from observational studies and our extensive post-marketing experience. The management is well tolerated and differentiated from other antipsychotics currently used off-label. Observed mortality rates are trending favorably. Additionally, pimavanserin has no negative impact on cognitive for motor function. Thank you. Dr. Stankovic will now provide a benefit risk adjustment.

Thank you, Dr. Turner. I'm Serge Stankovic, President of Acadia. I would like to conclude today's review of pimavanserin data in ADP with a discussion of benefit risk. As you heard today, ADP is a serious and debilitating condition with severe consequences for patients and their families. It results in significant mental and physical distress, acceleration of cognitive impairment, nursing home placement and increased mortality and morbidity. Unfortunately, there are no currently approved treatments for patients with ADP. In the absence of a safe and effective treatment option, the antipsychotics used off-label to treat psychotic symptoms associated with dementia expose the frail and elderly patients to great risk as they can worsen the underlying condition with marginal to no benefits. We presented today consistent, robust and clinically meaningful efficacy across multiple studies, end points and over time. Taken together, the aggregate data set provides evidence of effectiveness in ADP. What constitutes evidence of effectiveness of pimavanserin in ADP is a topic of discussion today. In 2019, FDA issued draft guidelines relevant to this topic. It states that one adequate and well-controlled clinical investigation plus confirmatory evidence in a closely related approved indication can be sufficient to establish effectively. Consistent with the above guidance, Acadia presented today data from three positive studies in psychosis: positive Study 019 in the target indication of ADP, confirmatory evidence from positive Study 020 in the closely related approved indication of PDP, and additional supportive evidence from the ADP subgroup from the overall positive study in DRP Study 045. The totality of efficacy data presented reliably meets the standards for evidence of effectiveness in ADP. Particularly important, this efficacy was observed in the context of a favorable safety profile. In conclusion, pimavanserin provides clinical efficacy benefit to patients with ADP and would allow them to manage their psychosis while not exacerbating the underlying condition or introducing new safety risks such as cognitive or model impairment. The totality of data presented today supports a positive benefit risk profile of pimavanserin for the treatment of hallucinations and delusions associated with ADP. Perhaps most important, this is in the context of a disease where there are no approved treatments and the current standard of care has marginal benefit with considerable risks. Thank you. I will now invite Mr. Dekarske to return to moderate the question-and-answer session.

So this is Raj Narendran. It seems like we're ahead of schedule. I do want to give an opportunity for Dr. Paul Stander, who joined us later, to introduce himself. Dr. Paul Stander, if you want to introduce yourself?

Yes. Thank you. This is Dr. Paul Stander. I am the Associate Chief Staff for geriatrics and extended care at the Phoenix VA, and I am a Clinical Professor at the University of Arizona College of Medicine, Phoenix. I'm sorry, I had a few issues early this morning, but I was able to hear the majority of the presentation. Thank you.

Thank you, Dr. Stander.

We will now take clarifying questions for Acadia Pharmaceuticals. Please use the raise hand icon to indicate that you have a question. And remember to clear the icon after you have asked your question. When acknowledged, please remember to state your name for the record before you speak. And direct your question to a specific presenter, if you can. If you wish for a specific slide to be displayed, please let us know the slide number, if possible. Finally, it will be helpful to acknowledge at the end of your question with a thank you, so we know -- or go ahead with a follow-up -- or say you have a follow-up question for you to ask. That way, we can then move to the next panel member. So we -- the first question we have is from Dr. Fiedorowicz.

This is Jess Fedorowicz from the University of Ottawa. I have a clarifying question. So a lot of the slides and results for Study 019 hinged on this primary outcome where the time frame is stated to be 6 weeks. My clarifying question is that in both the Lancet paper and the clinicaltrials.gov registration, study completion is listed as October 27. All registrations prior to that date show 12 weeks as the time frame after the primary efficacy outcome. And the only registration that shows 6 weeks occurs on July 14, 2017, as you can tell, several months, well after the close of the study. And I was wondering if the agency or if the applicant can clarify that.

Thank you for the question. Happy to clarify. The original study-019 protocol that was submitted to the FDA into the IND had a 6-week primary endpoint, so the day 43 endpoint that was discussed in the core presentation. That endpoint remained throughout the conduct of the study up to unblinding. I'll note your reference to the clinicaltrials.gov website. There was additional clarification subsequent to the trial start. It was initially indicated as a 12-week treatment period, which was indeed true, but there was additional clarification subsequently to be clear that the primary endpoint was at 6 weeks. But I just want to emphasize that, again, the 6-week primary end point was part of the original protocol upon execution of the study.

Our next question is from Ms. Witczak.

Thanks for your presentation. I have [ woody ] matters for the record. I know the original application to the FDA was for dementia-related psychosis, and then it switched over to Alzheimer's disease psychosis. My question is were brain scans or other scans used to objectively diagnose that it was Alzheimer and was able to differentiate from other causes of dementia. And so that's my question.

Thank you for the question. The diagnosis or clinical diagnosis for Study 045, I'd like to ask Dr. Serge Stankovic, to please speak to the inclusion criteria, then I'd like to ask Dr. Pierre Tarrio for follow-up after Dr. Stankovic. Dr. Stankovic?

Serge Stankovic, Acadia. The study 045, our dementia-related psychosis study used clinical diagnosis for different dementia subtype. The study protocol asked the investigator to indicate which dementia subtype is a primary dementia subtype from the clinical perspective and clinical diagnosis. And obviously, they used internationally accepted criteria for diagnosis. But there were no any biomarker used in the study for the diagnosis of subtypes. The reasons for that were that, number one, there is a significant overlap in the underlying neuropathology in the patients of the advanced dementia with psychotic symptoms. And the second nature of the study was that we were approaching the all-comer dementia-related psychosis regardless of the underlying subtype. So from that perspective, the bio margin diagnostics were both [indiscernible]. Important as it would be in the disease modifying treatment or the study of the specific subtypes of dementia. Thank you. I'll now turn it over to Dr. Tarrio.

Thank you. Pierre Terrio here, a consultant, and thank you for the terrific question. Just to build a little bit on what Dr. Stankovic just said, the participants in Study 045 that you asked about were, as you heard, aged, frail and they had advanced dementia and psychosis. So these would not be appropriate candidates for lumbar puncture or amyloid PET scanning, lumbar puncture to collect the CSF to look for elevated amyloid and/or tau. And I agree with what Dr. Stankovic just said, we're very interested in the use of imaging and fluid biomarkers to improve diagnostic accuracy for the presence of Alzheimer's pathology, but in milder forms and milder severity of Alzheimer's disease in younger patients. And this might be a follow-up question of yours. No, there are not plasma samples retained from the study. So we don't have the opportunity to look at those retrospectively. And again, just to repeat at this point that I think is quite important, persons with advanced dementia have a high likelihood of having multiple pathologies. So even if we had, had a way to establish the presence of elevated brain amyloid, it wouldn't rule out other pathologies. And I guess the last point I'd like to make is that the study was designed as a kind of pragmatic clinical trial to inform clinical practice and at least then, and it's changing, but at least then and even now, I would submit that the currently available research from biomarkers are not yet widely used in clinical practice. Thank you.

Our next question is from Dr. Apostolova.

Yes. Can you hear me?

Yes, we can hear you.

So we're -- so I'm glad I'm actually following Pierre and my commentary here because to me, it looks like there is effect observed in both PDD and AD psychosis, however, it's weaker in AD psychosis. And I wonder to what extent presence of [indiscernible] limbic parts actually contributes to that known [indiscernible] oncopathology in Alzheimer's present in 40% possibly? So it might be that the strong responders are those who have Lewy body pathology in at least limbic areas. But how do we know that? There is the already quick assay, but again, at CSF and as your -- as Pierre pointed out that is not attainable in advanced subject. My question though is about the durability of effect, and it's probably directed to Susan Hendrix. Slide 29 and 34 from the presentation do show that over time, regardless of continuous treatment, placebo tends to sort of merge stores the treated group, and the difference is no longer significant. How can that be explained scientifically and statistically? Could it have something to do also with measurement subjectivity as we don't have biomarkers, which are objective measures of treatment response as opposed to subjective measures like MPI? And I'm wondering both what Susan and maybe Pierre think about that.

Thank you for the question. Just want to clarify, you're referring to Study 019. Is that correct?

So Slide 29, Study 019 and then Slide 34, both, Yes.

Yes. Very good. So I'd like to ask Dr. Clive Ballard to comment a little bit further on Study 019, particularly on the durability of that point. But perhaps Dr. Ballard, you could also circle back on the question about the common and overlapping neurobiology and neuropathology between ADP and PDP? Or ask Dr. Ballard to comment. I just do want to remind that Study 019 had a 6-week primary efficacy endpoint. And that the endpoints beyond 6 weeks were exploratory in nature, mostly from the perspective, as you heard Dr. Ballard comment from a safety standpoint with respect to looking for any potential negative impact on cognition. In Study 045, the randomized withdrawal study was -- which is a very conventional design to establish maintenance of efficacy, was intended to demonstrate that following the acute response that was shown in Study 019 and Study 020. Dr. Ballard, can you please comment further?

Thank you, Clive Ballard. Well, firstly, looking at the main outcome slide at week 6, as you said, that was the primary endpoint and the clinically meaningful benefit as well as statistically significant benefit as indicated by benefit at 30% level of improvement and 50% level of improvement. I mean, the study was really designed to focus on acute benefit in 6 weeks was the selected period very deliberately for that outcome. When we're trying to understand sort of longer-term effects, we have to do that in the context of the natural history of the disorder. We conducted a study where we followed people up every month to look at the ongoing course of the symptoms. As I mentioned in the core presentation, over 2/3 of the patients had resolution of those symptoms by 12 weeks. But the underlying pattern is a lot of fluctuation, a lot of bouncing around of these symptoms, a bit of improvement, a bit of worsening, recovery, relapse again. So although there's an overall enduring pattern of these symptoms, it's very much an improve, get worse, recover, relapse type pattern. There's a lot of spontaneous recovery as part of this fluctuating pattern of symptoms of biomarkers as an outcome for psychosis in people with dementia. So I think although I'm sure that will be evolving space. I think where we're at, at the moment is that we have to rely on robust clinical judgment. What we did do, and perhaps I'll ask Dr. Hendrix to comment a little bit further on the analysis, is to try and understand the sort of the pattern of ongoing response beyond 6 weeks by trying to take out that impact of that fluctuating improving relapsing kind of course. And the way that we did that was by looking at -- by requiring people to have 2 consecutive improvements so that they had to be improved to consecutive assessment points in time to response analysis. And when we do that, you can see there in the figure on the right that, that takes a lot of the noise out. It takes a lot of this fluctuation and severity out. And what you see then is that you have an improved early response in pimavanserin, but also that, that response is sustained over sort of a longer period of the study through to the 12 weeks. I mean, I know you asked Dr. Hendrix to comment further, so I'm not sure Dr. Hendrix, whether you'd just like to comment on the statistical approach that you took?

Yes. Thank you, Dr. Ballard. Susan Hendrix, statistical consultant. So if we could just pull up for a minute the slide that shows those plots coming back together at the end, first -- actually, let's first do CO29, that was the original question. So this plot, because it's actually a cumulative distribution plot, the far right of the plot isn't time and coming back together after time, but it's actually the number of people who have achieved 100% improvement on their NPI-NH score on the psychosis component. And so over to the right, there aren't as many people, in fact, there's equivalent numbers in the placebo and pimavanserin group who've achieved a 100% response. But we see really strong separation, the clinically meaningful 30% and 50% effect time points. Now the analysis that Dr. Ballard was referring to, just to point out, again, that when we look for consistent benefit, the active arm has substantial benefit. Pimavanserin has substantial benefit over placebo. And here, when you're looking at the pattern over time, pimavanserin maintains benefit all the way out to 12 weeks. It's the placebo group that comes back down, and that's due to the symptom relapsing and coming back out. And then the slide that he showed that shows the consistent confirmed effect across 2 visits, then we continue to see that separation between the active arms. So all this speaks to two things: number one, a consistent benefit of pimavanserin out to 12 weeks that has been confirmed in the 045 data; and a clinically meaningful effect that, again, is also confirmed in the 045 data.

Thank you. May I have Slide 057 quickly, please? Just circling back around on the maintenance of efficacy point. So as I mentioned, following acute response demonstration in both Study 019 and Study 020, we specifically endeavored to look at maintenance of efficacy in Study 045. As you heard Dr. Hendrix speak to earlier, saw a very statistically significant result in the overall DRP patient population. But in the context of doing further exploration of a positive study, looking at the ADP subgroup, both all doses of ADP as well as ADP 34 milligrams, you saw a clinically meaningful reduction in the risk of relapse in that large ADP subgroup, around 40% to 50%, which for these types of trials is very well recognized as a clinically meaningful reduction in relapse. If I may, briefly, I'd like to ask Dr. Leslie Strome just to speak on the clinical meaningfulness of the hazard ratio in this patient population, Dr. Strome?

Leslie Strome, I'm a psychiatrist and I work with patients, and I do research. And one of the things I do is look at the clinical meaningfulness of clinical trial results and systematic reviews and across the field of psychiatry, whether we're looking at schizophrenia or bipolar disorder, major depressive disorder. And when you look at the maintenance of effect studies, a hazard ratio will describe the likelihood of relapse of recurrence over a given period of time in people who have been stabilized on a medicine of interest and then randomized to either continue that medicine of interest or a placebo. And this hazard ratio, over the course of the period of the study, has ordinarily been around about 0.5, hovering around there. And we would say that people who are maintained on the medicine that got them well, where half is likely to experience a recurrence or relapse, then compared to those who were randomized to go on to placebo. So what we see actually is entirely consistent with clinical trials within psychiatry. So I'm not at all surprised by these results. Thank you.

Our next question is from Dr. Thambisetty.

Thank you, Dr. Narendran. This is Madhav Thambisetty from the NIA. I have a comment and accompanying question. My comment is about Slides 28 or even Slide 43 from the efficacy presentation. I find it somewhat troubling that these graphs are curtailed at the 6-week time point when you have a full data set that extends through to 12 weeks. And I think it's potentially misleading to show graphs curtailed at 6 weeks when you have a full data set at 12 weeks. It's potentially misleading because this graph seems to indicate that there's a divergence in the placebo and treatment groups, which may be continued beyond the 6 weeks, which clearly is not the case. So I would really like to see these graphs show the full data set rather than just break them at 6 weeks, which brings me to the company question. And this is a follow-up question from what Dr. Pillowich asked first off. And I'm not sure I understood the explanation provided in response to that question. So I'll leave this question again because I really think it's very important. The public record on clinicaltrials.gov has a history of changes made to the protocol from the study start date in 2013. And it looks as if all versions of the protocol until July 2017 on clinicaltrials.gov, which consists of information provided by the sponsor, clearly prespecify a primary endpoint outcome at 12 weeks. It isn't until July 14, 2017, that the record indicates that the primary outcome was changed from 12 weeks to 6 weeks, which is 10 months -- nearly 10 months since the last patient was randomized. And then on September 28, 2017, again, clinicaltrials.gov indicates, exactly 1 year after the last patient was randomized, the primary endpoint is again changed from 6 weeks to 43 days. So it's not entirely clear to me why a primary outcome endpoint would be changed 10 months to a year after the last patient was randomized. The rationale is not clear to me. And I really don't know why this was done and what the rationale was.

Thank you for the question. I'll address your further follow-up on the status of the protocol and the primary endpoint. And I'd like Dr. Clive Ballard to speak to your initial question about the 6-week endpoint relative to the 12-week treatment period. Could I have the Slide CO 57, pleas? Sorry, CO34, CO34. As I mentioned earlier in my previous response, although the trial was a 12-week duration, which was indeed the case, the initial posting for clinicaltrials.gov simply referred to that 12-week period. We subsequently clarified on the clinicaltrials.gov website, the primary was actually at 6, although, again, there was a 12-week treatment period. Now importantly, in terms of the actual protocol and its submission to the IND, and at the time of its initiation had a 6-week trial endpoint, that did not change during the conduct of the study nor did it change prior to stopping of the study and unblinding of the data. Dr. Ballard, could you speak a little bit further to the core Slide CO 34, which shows both the 6-week and the 12-week exploratory efficacy?

Clive Ballard, certainly. Firstly, just to confirm that the -- about the 6-week primary outcome. I mean, as the principal investigator for the study, I was responsible for all of the submission of the protocols to the ethics committee for approval. And I can absolutely confirm and we can provide the protocols that the primary outcome was always 6 weeks throughout all of those protocols. So that's definitely not changed at any point during the process. To come back to your question about 6 weeks and 12 weeks, I mean I clearly did show this slide showing the full 12-week outcome during the core presentation. So there was no attempt to conceal anything. This initially presented it up to 6 weeks because that's the primary outcome point. The objective of the study was to focus on acute improvement at 6 weeks, with time out to 12 weeks was largely to look at safety outcomes because of potential concerns with other atypical antipsychotics in terms of impact on cognition and function. So I don't think there was any attempt to conceive anything I presented the data. This placebo pattern that you see over 4 weeks and then starting to improve at about week 6, this is very consistent across trials of atypical antipsychotics in the literature, which is why we very deliberately chose 6 weeks as the optimal time point to look at the acute response, and that was the intention. As you can see from the ongoing period past 6 weeks, as we've already discussed, then clearly, there's a lot of placebo response over between week 6 and 12. I think that, as explained, as I mentioned, by the natural course of the condition, where we know that 2/3 of people in natural follow-up studies are going to have improvement by that 12-week period. So really, the aim of this study was to look at the 6-week acute response. The aim of the 045 study, given that the nature of this condition is a relapsing recurring condition, was to look at the sustained benefit and the prevention of relapse in those individuals. But as I mentioned in the previous response, if we could just have the slide showing the time to response analysis. One of the things that we did to just try and confirm, whether this impact on the loss of response was due to the fluctuation as part of the natural course of the condition, we did do this further analysis, looking at people who had sustained benefits of at least 2 time points. And when we did that and we took the noise out of the situation, we removed the noise from people who are having brief responses and then relapses again. When we did that, there was a much clearer pattern of response to pimavanserin in terms of sustained response across the 12 weeks.

Next question is from Dr. Cudkowicz.

I had a question about Slide 58. If you might explain a little bit more this exposure response. So I was trying to figure out if there -- you're trying to make the point that you needed higher levels in the Alzheimer's, if that might explain some of the difference between the like patients with Alzheimer's versus Parkinson's. And if, in fact, there -- were there different levels in people with all tenders maybe because of different other medications.

Thanks for the question. The principal utility for the exposure response analysis from Study 045 was to reassure on a real pharmacologic treatment effect. So as you heard in the core presentation, what you see in the overall DRP patient population as well as the ADP subgroup that the risk of relapse goes down with increasing concentrations of pimavanserin, as also indicated by FDA in their briefing material. The actual concentration of pimavanserin that you see on the chart in the ADP and the DRP are quite close. And in fact, the steady-state concentration that you would anticipate with a 34-milligram once-daily dose is very well reflected as part of this exposure response analysis. So the profile, the pharmacokinetic profile is what we would expect and if populations are consistent.

I just have one other question on another topic, which goes back to this question of duration of treatment. Do you anticipate this 3 participants or people for 6 weeks or 12 weeks? Or how would this be in clinical practice in the Alzheimer's population?

Thank you for the question. So the 3 clinical study data set that we have, that is supportive of evidence of efficacy in ADP, includes both Study 019 and 020, demonstrating acute response in both ADP and PDP patients. As I mentioned earlier, Study 045, which is typically done in the psychiatry space after demonstrating acute response, was intended to establish maintenance of efficacy to support long-term treatment following acute response, positive acute response. So those are intended to support therapy beyond the acute period. And I'll just add, it's important context with respect to the safety profile of pimavanserin, where we focused on a couple of key safety aspects, that in this class, you're particularly concerned about in terms of negative impact on cognition and motor function. We did not observe that, and we were specifically looking in Study 019 and 045 to make sure with long-term treatment, there weren't any negative impacts. I'd like Dr. Clive Ballard to provide a bit more clinical perspective on duration of treatment and appropriateness.

Thank you. I think obviously, the 6-week effect is really important. These are really distressing symptoms. When they're present, they're very unpleasant for the individuals experiencing and very challenging for everybody. So achieving that benefit over that 6-week period is very, very important. But I think potentially longer-term treatment is also -- can be beneficial. And I don't think that's a blanket approach. It's very much based on the needs of individual patients and the pattern of -- and severity of symptoms in individual patients. But I think one of the challenges we have with currently available, atypical antipsychotics, is that because of the toxicity issues, longer-term treatment for prophylaxis isn't really an option, whereas I think one of the things the relapse prevention data and good tolerability suggests is that, that is a potential option. So whilst that wouldn't be the optimal for every individual for individuals who do have more severe symptoms, you have higher rates of relapse. I think that, that option of been able to provide longer-term treatment to reduce relapse and to give it safely and to maintain that benefit would be a very useful addition to the pharmacological [indiscernible] for those individuals.

Next question is from Dr. Walter Dunn.

This is Walter Dunn from UCLA. So one of the key questions for me for this entire presentation is this notion that or there idea that ADP and PDP are closely related conditions. At least from the 045 results, there's some strong indication that there's a differential response, especially when these patients are randomized to placebo that clearly, there is a much higher relapse rate in the Parkinson's patients. And so in your -- although you didn't emphasize this in your presentation in the briefing documents that Acadia provided, the explanation was that this high relapse rate was due to the presence of dopaminergic agents. Unfortunately, all the PDP or the majority of PDP patients are on those agents, and so it's difficult to disentangle between if is it an illness or the presence of those agents. So I was wondering, do you have data in the open-label period for the patients who did not meet sustained response, and thus did not proceed to the double-blind phase? For those patients who failed to meet sustained response, do you have any data as far as the distribution of those on dopamine agents versus not?

Yes. Thank you for the question. Just following up to your point about our belief that the dopaminergic concomitant therapies were the result of the particularly robust response we saw in the PDD subgroup in the 045 study. I just want to remind that in Study 020, these were PDP patients mostly on dopaminergic medications, where we're looking at improvement in symptoms out to 6 weeks. Study 019, on the other hand, were in ADP patients without dopaminergic therapies. What we saw in those 2 studies, in those 2 different patient populations with respect to use of dopaminergic drugs is a very consistent response. I'm showing Slide CO 43 from the main presentation. As we look at Study 045 and getting to your question in the open-label period, it was reassuring to us that when we looked at the response rate within the first 12 weeks in the open-label period, we saw a very consistent stabilization between the various dementia subgroups. That includes both PDP and ADP, including what we thought was a very robust complete response at the end of the 12-week period of about 20% between the 2 patient populations. We don't have on hand right now the distribution within the open-label period, the use of concomitant dopaminergic therapies, specifically within the subgroups, but we're happy to provide that after the break. But I will say, just following up in the double-blind period, for those patients that continued from the open-label period into the double-blind period, what we saw was that the patients that continued on pimavanserin, the treatment effect was homogeneous across the subgroups. That includes PDD. It was the placebo group where we saw that there was a very early relapse, which is what we would expect based on our belief that the dopaminergic therapies may be exacerbating the underlying psychosis symptoms in the absence of effective antipsychotic treatment. And you can see that clearly here on the core slide CO 54.

Our next question is from Dr. Dean Follmann.

Yes. Thank you. I have two questions. I think they're both addressed to Dr. Hendrix. The first one, in Study 019, you showed a large benefit in people with psychosis with effect estimate of 0.73 and a p-value of 0.01. I was wondering if you had looked at the benefit in 045 in those with severe psychosis to try and replicate those?

Suzanne Hendrix, do to come to the mic and speak to your question. Thank you.

Susan Hendrix, statistical consultant. We did look at that, and we found that those patients who had more room for improvement, which are the ones with more severe symptoms, tended to see more improvement. And in the relapse specifically, we saw that people who were more severe and had treatment removed dropped off more rapidly than those who are less severe. So that same type of pattern have seen a better effect where there's more room for an effect held throughout the 045 data as well.

I have one more question. If you could bring up Slide C0 58 again, this has to do with the effect of drug exposure on the risk of relapse. So drug levels aren't randomized with the drug group, and it could be the drug levels vary where people who don't comply or are sicker or worst outcomes, tend to have lower drug levels. So did you look at the relationship between drug level and sort of baseline characteristics or compliance to try and tease at this to get at this particular issue that we aren't randomizing to drug levels here?

I'll just -- just to point out, the dropout rate in the open-label period was quite low, in fact. So those patients that were going from pimavanserin in the open-label period into the double-blind period, there was a majority of subjects that were doing so. I'd like to ask Dr. Suzanne Hendrix to comment, please.

Yes, it's a very good question because we don't have randomization of those different AUCs. What we did do was we looked at several different covariant analyses, and those covariant analyses tended to make this effect stronger rather than weaker. Very few of those covariant were actually statistically impact none of the covariant were statistically significant, and the overall results after adjusting for them were somewhat stronger statistically.

Our next question is Dr. Apostolova, do you have another question?

Yes. I, do have another question. These are super distressful behaviors, psychosis, hallucinations, delusions, and they have an incredible impact on caregivers. Was their caregiver distress data collected in these studies? And I'm curious what it showed, even though it wasn't presented, if there is any idea.

Yes. Thanks for the question. There was a caregiver burden data that was collected in Study 045 as well as study 020. I'll just ask Dr. Suzanne Hendrix to speak first to the data in the Zarit Burden interview in Study 045 and what we found.

Suzanne Hendrix, statistical consultant. So on this slide, we're showing the secondary endpoints from the 045 study within the Alzheimer's disease psychosis subgroup specifically. And the SAPS-H+C, the CGI were both clinician scales, but the Zarit Burden Inventory, ZBI, shown on the third line here is a caregiver burden assessment. We did not achieve nominal significance on this test, but we did have a 1.25-point benefit in favor of pimavanserin, reflecting a clinically meaningful effect for caregivers having less burden.

Just following up on in Study 020. This is a study in patients with PDP, thus, Zarit Caregiver Burden was also assessed. And there was an exploratory basis, but there was a nominally significant improvement that was seen in the level of burden in those patients.

Dr. Walter Dunn, do you have another question? Please go ahead.

Thank you Dr. Walter Dunn, UCLA. Another question about Study 045 and the open-label data, if you have it. So this speaks to the durability issue again. So do you have any data on the percentage of patients that, again, did not meet sustained response and did not proceed to double blind? And as far as do they never -- what proportion never would have never met criteria at any of the time points for 8 or 12 weeks versus what proportion of patients potentially at that 4-week time point would have met criteria but then symptoms worsened at 8 and 12 and therefore, didn't enter the double-blind phase? My understanding is that there was no 6-week study point. So I'm using the kind of the 4-week as a proxy for like a 6-week time point. Was I clear in my question?

Yes. Thank you. I'm just pulling up the core Slide CO 50 of the open-label period. So to answer the first part of your question, the rate of sustained response or stabilization in the open-label period in the overall group was about 62%. This was consistent across the subgroups. And I think as these randomized withdrawal trials go, that's a pretty high stabilization rate. Usually, you would see around 40% or 50%. To answer the second part of your question, I'm going to ask Dr. Suzanne Hendrix to please come to the microphone.

Suzanne Hendrix, statistical Consultant. Could you please clarify the second part of your question again?

I was just wondering what proportion of patients that did not enter the double-blind phase, would have met or did meet entry criteria or would have met entry criteria 4 weeks, but then relapsed at weeks 8 or 12 and therefore, do not enter the double-blind versus what proportion of patients just never improved enough at any time point to enter double-blind.

So I think -- so the question you're asking is out of that approximately 40% who did not meet the enrollment criteria for the double-blind phase, what percentage of those did meet criteria at some of those earlier time points?

Correct, correct.

Okay. Hold on just a minute. We have a slide for that. It's burried a little bit deep. We'll find it to 1 minute. We'll go ahead and bring that up after the break rather than waiting now.

Next question is from Ms. Witczak.

Yes. Kim Witczak, Woody Matters. I would like to know kind of the philosophy or the rationale behind using a largely primarily white audience in this clinical trial when -- obviously, I know that was in the U.K., But in the U.S., the high prevalence is in the African-American community. I'm Hispanic. And so I'm just curious how that may or may not. And it might also be a question for the FDA as well. But if you can -- and the rationale behind that audience makeup.

Thanks for the question. We recognize this is an issue for our industry, including for us at Acadia. That said, pimavanserin's PK profile data, the clinical study data and our post-marketing experience, we feel, is generalizable to other races and ethnicities. In fact, as we've -- as our development has ensued with pimavanserin, we've improved on a representation of race and ethnicity. We had a post-marketing study commitment upon the original approval to increase the safety data in frail and elderly patients, including Alzheimer's disease patients, which, as you heard Dr. Turner speak to earlier, we've expanded significantly. As you can see, we've got reasonably good representation within the Hispanic and let you know, ethnicity, we're improving in -- with black or African American. But importantly, from a pharmacokinetic perspective, we have done work to identify that pimavanserin's plasma profile is not impacted by race or ethnicity. That all said, in the go forward, we are doing Phase IV studies, and we're working with specialty sites and minority communities to improve diversity. As we know, it's important to inform on that point for our label and for prescribers. But just to land, we do believe it's generalizable with the data we have in hand, both from a PK perspective, from a clinical study perspective as well as post-marketing experience that doesn't indicate a differential impact with race or ethnicity.

We have another question from Dr. Thambisetty.

Madhav Thambisetty, NIA. This is a question directed to the Director of the Division of Sectary from the FDA on her presentation with Slide 8. Maybe this is a question that might be best suited after the agency makes the presentations, but I just thought I'd like to bring this up now, the complete...

I'm sorry to interrupt, but I think it might be better to just ask that during the agency at their own time so that the sponsor doesn't lose their time to answer their questions, if that's okay.

Okay. Okay. Yes. Can I have a quick follow-up question about the decline of Study 045?

Sure. Please go ahead.

Yes. And this is addressed to the sponsor, and I guess anybody can answer it. It's very difficult to find a randomized withdrawal trial in psychiatry that hasn't been effective. I think these designs lend themselves to invariably favoring drug over placebo because there is a topology involved in the design of the study itself. You're selecting out treatment responders in the open-label phase. You're excluding everybody in the placebo group who might have responded. You're excluding people with adverse events. And then in the double-blind phase, we're again measuring the same thing that you measure in the open-label phase. So in fact, treatment response is being measured twice. And so this study design invariably results in outcomes that unequivocally favor drug over placebo. And again, I would maintain that while these designs might be suitable to look at durability of response, they do not provide a lot of useful information in determining efficacy. There is also concern that during the open-label phase, effects of drug withdrawal are almost invariably compounded with the relapse itself. And so for these reasons, I think randomized withdrawal trial designs, especially in psychiatry, where you do not have independent outcomes being measured during [ 2 ] phase of the study, I think they are far from desirable. And again, this is a question that would be suitable both towards the sponsor as well as to the FDA itself.

Thank you for the question. Just for context, generally, the randomized withdraw or maintenance studies are done following demonstration of acute response which, in our case, acute response was demonstrated in studies 019 and 020. So that said, in the open-label period of Study 045 as I mentioned earlier, the response rate was quite high. It was over 60% in these patients. So there was good evidence that there was a high response rate. That's it. I'd like Dr. Serge Stankovic to comment further on the randomized withdrawal design and the level of evidence for the overall package for the effectiveness in ADP patients. Dr. Stankovic?

Thank you. Serge Stankovic, Acadia. Thank you, Dr. Thambisetty, for that question. It's a very important question. The rationale for using the randomized withdrawal design for the Study 045 was based on the essentially evolution of our development program. We had already demonstrated acute efficacy in 2 models of dementia in PDP and within ADP in the acute setting. Second, the study, as you pointed out, and others, Study 019 left open a question whether there is durability of effect because we did not see separation at week. So the best way to test the maintenance of effect of drug is one of the maintenance of efficacy designs, and that's certainly a randomized withdrawal trial. And that is what we proposed and the FDA agreed at the end of Phase II meeting that it's a reasonable design as a pivotal trial for the dementia-related psychosis program is the next step in our development program. In addition, randomized withdrawal design has obviously some advantages that are important for the patient population that we studied. It minimizes exposure to an effective treatment in this severe and serious medical condition, allows enrollment of patients and their caregivers in the trial because they initially do not have concerns about actually exposing their panel member to ineffective treatment. And there are obviously the escape area following randomization. So it's a design where families are more willing to participate in the trial and really provides a mimic fairly well the standard of practice in treating these patients. So for all of these reasons, this design was chosen and agreed upon for moving into the development of DRP. Thank you.

Thank you. This is Raj Narendran. I mean, I have my own question. I was just curious, I mean, so it seems like looking at Dr. Tarrio's presentation, the objective of having an antipsychotic or a medication to treat psychosis in Alzheimer's dementia is to keep patients out of a nursing home and prevent deterioration, improve their functioning. Why was it that 019 was done in a nursing home where patients are extremely cognitively compromised with a very low mini-mental status as opposed to 045 seems to be more like your target population you want to go forward?

Thanks for the question. I'll just point out quickly before I turn it over to Dr. Ballard to answer your question on 019 in the neuro signal population. Study 020 was also an outpatient study. So there's across the spectrum of patients that are in nursing home or an outpatient basis. We have representation of efficacy and safety data. But in specific to the advanced age and disease for Study 019, Dr. Ballard, can you please explain the design?

Thank you, Clive Ballard. I think both are really important. I mean, obviously, for people living at home with family members or living on their own, trying to help those individuals retaining dependence and reduce the chances of individuals moving into institutional care is really important. But I think we shouldn't also forget the high levels of distress and vulnerability of people already in nursing homes. These are the people with the most severe disease. They are the people who are most likely to have psychosis. Psychosis rates are far, far higher in people with Alzheimer’s living in nursing homes than there are amongst individuals at home nor the people who have the biggest problems tolerating currently available atypical antipsychotics and have the biggest adverse effect. So I think the potential for benefit in terms of improving symptoms, reducing distress, reducing unnecessary harms is extremely great in that nursing home population, but that's not to take away from the important benefit of preventing institutionalization and people living at home.

As a follow-up to that, I get that point, but how do you then divorce agitation and aggression from psychosis in a nursing home population? And you didn't see an improvement in that because it seems like that would be the biggest issue is agitation aggression. And how is that really divorced from psychosis per se in that data set?

Clive Ballard. Well, we exactly applied rigorous criteria for assessing both. As part of the inclusion criteria for the 019 study, we required people to have sufficient verbal ability to be able to describe their symptoms. So people, if they had to leave the [ nation ], they were able to describe those. They weren't inferred from behavior. If they had delusion, they had to be able to verbalize those, and we applied with rigorous criteria the -- in order to both diagnose them and evaluate them with the NPI. So I think it's perfectly possible in people who are able to have that verbal ability to assess psychotic symptoms accurately in people with pretty severe dementia, and that's been done across a number of studies. So in terms of agitation, I mean, concurrent agitation is a big problem. 30% to 50% of people with psychotic symptoms in the context of Alzheimer's disease do have concurrent agitation. The evidence from the literature is that the neurobiological basis of that agitation is different to the basis of psychosis, but the psychosis is one of the factors that might impact on agitation. And although we didn't see any overall benefit in agitation, we did see that when there was a 50% improvement in psychosis, there was also a substantial benefit in agitation associated with that. So improving psychosis did have a knock-on benefit in terms of reducing agitation when psychosis improved. But they are different symptoms. They have a different neurobiological underpinning. And in people with a verbal ability, these symptoms can be assessed accurately and robustly in nursing home patients.

Dr. Narendran, just as a quick follow-up. You mentioned Dr. Tarrio in the context of your question about the utility and nursing homes. I just want to give Dr. Tarrio just a moment to respond directly to you.

Pierre Tarrio, Consultant. Dr. Narendran, I think you actually pointed out a flaw in my presentation. I was actually Director of Psychiatry at a very large long-term care facility in Rochester, New York for 20 years. We did a study evaluating the presence of neuropsychiatric disorders ourselves as opposed to chart diagnosis. And we essentially showed that this is like a state psychiatric hospital, these settings. So half or more of the residents have dementia, and most of those folks have very prominent neuropsychiatric symptoms, including, as you pointed out, agitation and aggression but also including psychosis. I've done many of these trials myself with NIH funding, industry funding. We have yet to find anything that's truly effective for agitation and aggression. And the best we can do is essentially put somebody into a pharmacologic straight jacket, which is just not acceptable. So in fact, for agitation and aggression, our own clinical approach is to focus on nondrug strategies as best we can and really reserve drugs when we're faced with hospitalization. And we really want to try to keep these folks out of the hospital so I did not mean to suggest that there isn't a role in the long-term care setting as well. That was really my main point.

We have time for one last question. I think Dr. Walter Dunn.

Walter Dunn, UCLA. This is a question for Dr. Hendrix. In the briefing document for 045, you conducted a tipping point simulation analysis to mitigate the effects of the Parkinson's group to determine if the study would have been positive just with the ADP cohorts. Can you just describe when they talk about adding 9 events to the PDD group, does that actually mean that you increase the relapse rate for the PDD Group to 11 out of 15? And then as a follow-up, can you just comment briefly on the advantages and disadvantages of this tipping point analysis versus what the FDA did with removing the PDD group altogether from their analysis to determine if ADP would have been significant.

Yes, thank you. Suzanne Hendrix, Statistical Consultant. So the purpose of this analysis was to reproduce in as close a fashion as we can what would have been expected for overall DRP if the PDD group hadn't been so dramatically different than the other group. And of course, because the study wasn't powered within the ADP subgroup specifically, I wanted to perform an analysis that would have similar power the original design of the DRP study. And so if we take out the PDD Group in its entirety and just look at the subgroup of ADP, we lose the power of having those additional patients in the analysis. So the goal of this analysis was to say how much did the PDD results actually drive the analysis of overall DRP. And if we were to have a more comparable hazard ratio in the PDD subgroup, would we still achieve significant overall in DRP? So by adding the additional events, what we find is that is that, once we add as many as 5 or 4 [ look ] 5 events, so near the bottom, the middle of the bottom section here is what I'm looking at in the PDP group. We have 5 events added a p-value 0.028. Right below that, we have 6 events added, and we don't have significance anymore in the ADP subgroup. So we just have a trend. That trend is consistent with the effect that we saw for ADP as a whole. And when we then go up to the top section and look at that same corresponding row with 6 events, we have overall significance. So the point of this was to say, we don't achieve significance with ADP alone, but that's because it's under powered. And we do achieve significance if PDP, ADP and other were all at similar levels of hazard ratio, and then we would have gotten significance overall for DRP. So it's just another way to look at whether the study would have been significant if it had been ADP as a whole with the larger sample size and with similar hazard ratios across all of the groups put together.

A little past to 11:10. So I think it's time for us to take a quick 10-minute break. We could -- panel members, please remember that there should be no chatting or discussion of the meeting topic, other panel members -- or with other panel members during the break. We will resume at 11:20 to start with the agency presentation.

Dr. Narendran?

Yes?

There were 2 questions during the presentation, considering the history of the protocol for 019 as concerns the primary endpoint. We've put together a slide that outlines the history and where we'd be happy to share it just to, I think, perfectly clarify the inception of the protocol, its conduct and up to database lock.

Can we -- maybe we -- if the agency is okay with it, maybe we can do that right before they start. Is that okay? I want to check with them. [ Break ]

Thank you, everyone. Hopefully, everyone's back. I just wanted to give the sponsor a couple of minutes to address their request before we start with the FDA presentation. So if you guys want to go ahead and respond about the Study 019.

This is Pierre Tariot, pinch hitting just for a moment, while the team reassembles. Could we pull up, is it TI-72 because we realize we've kind of not given a crisp response to this. So I'm stalling a little bit. Darryl you're on.

Yes. Serge Stankovic, Acadia speaking. I want to thank [ Dr. Fiedorowicz ] for asking this question. It is very important that we are absolutely clear on this point. I just -- I want to make two points. One is the 6 weeks end point was never a subject on any protocol amendment, and it was never modified in the -- from the beginning of the trial to the end of the trial. In terms of amendments, let me just go through history very quickly. The study protocol was approved in 2010. There were 3 protocol amendments to this protocol -- Study 019. One in 2013, as you can see on the slide, one in 2014 and one in 2015. Last on 16th of November of 2015. Database for this study was locked on December 2, 2016, and the data was unblinded on December 5, 2016, which means the full year plus after the last protocol amendment. And above and beyond all of that, that none of these protocol amendments ever made any changes to the 6 weeks end point. So I hope that this clearly states and clarify that there were no any changes to the end point and no any changes to the protocol per se following database lock and data unblinding. I also want to say that misunderstanding most likely comes from the clinicaltrial.gov, which is evolving, posting often with some mistakes. And in this case, the clinical trial posting was just in error. And -- but it doesn't have anything to do with the protocol amendments, with the implementation of the protocol or with the timing of the database lock and unblinding. So if there are any questions, I'm happy to respond, but I hope that this clarify this on a factual basis. Thank you.

Dr. Narendran, may I make a quick response. This is Dr. Thambisetty.

Sure, very quickly, please. Thank you.

Dr. Stankovic, thanks a lot for that clarification. So my concern is about the clinicaltrials.gov website, which says that there were 2 amendments made in July 2017 and September 2017, which is a year after the blind was broken from your slide. So if that was an error, it might need to be corrected because that is a public record. It's something that U.S. sponsor have submitted to clinicaltrials.gov. And if you, in fact, do side-by-side comparisons of earlier versions of the protocol with the amended protocols in July 2017 and September 2017, you can actually see texts that clearly have been deleted to say that the primary end point at 12 weeks was, in fact, changed to 6 weeks, and this may be a quirk of the clinicaltrials.gov website, I do not know. But I find it a useful resource to track changes to protocol amendments that are not otherwise publicly available, which is why I referenced that source in response -- that source to preface my question. Thank you.

Thank you. I think that kind of addresses the issue. So we will now proceed with the FDA presentation starting with Dr. Paul Bossie.

Thank you. My name is Paul Bossie. I'm the clinical reviewer for the application. I will discuss relevant regulatory history, an overview of the design and results of Study 19 and the resubmission and an overview of the design of Study 45. My statistics colleague, Dr. Xiang Ling will discuss the Study 45 results and resubmission analyses before I return to provide concluding remarks. Pimavanserin was approved in 2016 for the treatment of hallucinations and delusions associated with Parkinson's disease psychosis, or PDP. At a 2008 Pre-Investigational New Drug Meeting, the applicant outlined a plan for Study 19, a Phase II randomized, double-blind, placebo-controlled trial, pimavanserin in subjects with Alzheimer's disease psychosis, or ADP, to serve as part of a multi-study approach to support a dementia-related psychosis indication. At a 2017 End of Phase II meeting, the agency agreed that the treatment of hallucinations and delusions associated with dementia-related psychosis is a potentially approvable indication. The agency expressed concern about basing a regulatory decision on a single randomized withdrawal study, that is Study 45, but ultimately agreed that it would be acceptable as a well-controlled trial for supplement submission for the indication of hallucinations and delusions associated with dementia-related psychosis. The agency agreed with the population as long as subjects were stratified by their current clinical diagnosis that is dementia subtype and noted that labeling would reflect the actual composition in response of subjects enrolled in the study. The applicant submitted the supplement in June 2020, supported by Study 45 with Study 19 and resubmitted data from Study 20 on Phase III study in subjects with Parkinson's disease psychosis, which included a subset of subjects with dementia. The agency issued a complete response in April 2021, concluding the supplemental applicant did not provide substantial evidence of effectiveness for the treatment of hallucinations and delusions associated with dementia-related psychosis. Although it is very important to note that Study 45 was not powered to demonstrate subgroup efficacy, an examination of dementia subgroups revealed the following observations. Results for the Parkinson's disease dementia, or PDD subgroup, were highly nominally statistically significant. Despite being a relatively small subgroup of 35 subjects, the finding in the Parkinson's disease dementia subgroup appeared to drive the overall study results. Again, noting that the study was not powered for subgroup statistical analysis, results for the Alzheimer's disease or AD subgroup were not nominally statistically significant, despite being the largest subgroup with 123 subjects. However, numeric separation of the AD group from placebo was apparent. Too few subjects with dementia with Lewy bodies or frontotemporal dementia were included to adequately represent those subgroup responses. And there was no difference on time to relapse between pimavanserin and placebo in the vascular dementia subgroup. The agency noted that the results of Study 45 essentially demonstrated what was already known that pimavanserin was effective in the treatment of Parkinson's disease psychosis, whether or not patients have dementia. The agency noted that the Study 45 finding suggested a differential response to pimavanserin across dementia subtypes, which called into question whether dementia-related psychosis is a useful construct for a potential indication. Finally, the agency noted concerns related to trial design and conduct with Study 19. The applicant discussed its resubmission plans with the agency at post-complete response meeting, including the intention to change the proposed indication to treatment hallucinations and delusions associated with Alzheimer's disease psychosis. The agency agreed to consider the applicant's point in a resubmission, but advised that an additional adequate and a well-controlled study in subjects with Alzheimer's disease psychosis would likely provide the strongest data in support of a resubmission. Now I'll discuss Study 19. Study 19 was a Phase II, randomized, double-blind, placebo-controlled study of pimavanserin tartrate 40 milligrams once daily versus placebo [indiscernible] 40 milligrams is equivalent to the 34-milligram free base approved dose. The study was conducted within a network of 133 nursing homes in the United Kingdom, overseen by a single principal investigator. The 12-week treatment period was preceded by an approximately 3-week screening period, during which subjects completed an antipsychotic washout, if necessary, and caregivers were trained to provide brief psychosocial therapy to the subject a minimum of 3x per week with a target of 5x per week. For the applicant, the intention of the brief psychosocial therapy was to minimize placebo response prior to randomization and to assure that only subjects requiring pharmacologic therapy were randomized into the study. The study enrolled 181 nursing home residents, at least 50 years old, who met criteria for possible or probable Alzheimer's disease with psychosis, with baseline Mini-Mental Status Examination or MMSE scores between 1 and 22 inclusive. Subjects were to have been nursing home residents for at least 4 weeks and not be confined to bed, psychotic symptoms, including visual or auditor hallucinations or delusions were to have developed [indiscernible] Alzheimer's and subjects were to have verbally [indiscernible] symptoms during the month before screening and [indiscernible] prior to baseline. Subjects were excluded for psychotic symptoms caused by another reason, such as delirium or schizophrenia, anti-dementia drugs, that is acetylcholinesterase inhibitors and memantine, antidepressants and anxiolytics were permitted, if doses were stable before and during the study. Subjects were randomized 1:1 to pimavanserin or placebo and were stratified by baseline MMSE score and Neuropsychiatric Inventory-Nursing Home Version Psychosis Score or NPI-NH PS, which I will describe next. The primary end point was the mean change from baseline to day 43 on the NPI-NH PS. The neuropsychiatric inventory was developed to evaluate 12 neuropsychiatric disturbances or domains, prominent dementia, such as delusions, hallucinations, agitation -- sorry, I don't know what just happened. Okay. Is anyone else in the screen? I don't know why it's same thing. Okay. Sorry about that. The neuropsychiatric inventory was developed to evaluate 12 neuropsychiatric disturbances or domains, common dementia, such as delusions, hallucinations, agitation and disinhibition. I apologize for the slide. I can't figure out why it's moving forward.

Dr. Bossie, which side would you like?

The one I'm on right now. Okay. Perfect. The NPI-NH PS include delusions and hallucinations domains. The score of each item at present represents the product of symptom frequency on a range of 1 to 4 and in severity on a range of 1 to 3, for a maximum score of 12 on each domain, with higher scores denoting worse symptoms. The NPI-NH PS consisted of 2 domains, the maximum possible score is 24. A trained rater was to conduct the assessment with appropriate caregiver at the nursing homes. The NPI has been used in other development programs, but the NPI-NH PS has not been used in other development programs regulated by the agency. Although the NPI-NH PS is considered an adequate end point for exploratory purposes regarding this context of use, the agency's division of clinical outcome assessment has noted that the developed evidence supporting its use has not been optimized. There are residual concerns with the scoring and the interpretation of group and individual differences and limited evidence of content validity for this context of use. The scoring algorithm, which totals the product of severity and frequency item scores for each domain, as seen in the upper left table, yields a metric that may be difficult to interpret. Different permutations of severity and frequency can result in the same score highlighted here in the same color. For example, a severity of moderate and a frequency of often result in this quarter of 6 as does the severity of severe and the frequency of sometimes. As seen in the theoretical examples in the table below, subject A with baseline delusions with moderate severity decreased to mild at end the study, but the delusion frequency of very often did not change. Conversely, their baseline hallucinations frequency of very often decreased often at end of study, but the hallucination severity of severe did not change. Examining the combined delusions plus hallucination scores, the overall change from baseline to end of study is minus 7 points. In terms of implications for subject A, changes in severity or frequency may or may not be meaningful, depending on subject and caretaker input. For example, the reduction in hallucinations frequency from very often to often, but remaining severe considered to be meaningful improvement. The meaningfuless of reduction in frequency when severity levels remain the same in terms of impact on subjects or caregivers may be unclear, for example. Subject B began with mild delusions at baseline, occurring sometimes, they're remitted by end of study. Subject B had severe hallucinations that occurred often at baseline, which decreased to moderate severity at the end of the study with no change in frequency. Examining the combined delusions plus hallucination scores, the overall change from baseline to end of study is minus 5 points. In terms of implications for Subject B, this 5-point change may or may not be considered meaningful because mild delusions remitted, while hallucinations decreased in severity, but not frequency. Additional information, such as more qualitative studies at the end point with feedback from patients and caregivers could help provide understanding of the clinical meaningfulness of the results. Missing evidence or content validity includes research within the development program to provide evidence of content validity or a comprehensive review of the literature with a summary focused on how the items measure the targeted concept of interest that is hallucinations and delusions in the Alzheimer's population. There are overall gaps in psychometric evidence in the literature. The caretaker's rating of hallucinations or delusions may not reflect the entirety of the patient's experience because certain aspects of psychosis are not explicitly presented to [indiscernible] questions. Secondary endpoints included the Alzheimer's Disease Cooperative Study-Clinical Global Impression of Change, or ADCS-CGIC rating at day 43. As of other CGIC scales, the rater is asked to rate the subject functioning relative to the baseline interview using a standardized 7-point scale from 1 marked improvement to 7 marked worsening. Other secondary endpoints included the change from baseline to day 43 on 2 other NPI-NH domains on the Cohen-Mansfield Agitation Inventory-Short Form or CMAI-SF total score and on 3 of its subdomains. The CMAI-SF is a 14-item instrument assessing frequency of manifestations of agitation in the elderly based on directly observable behaviors, including physically and verbally aggressive behaviors within the previous 2 weeks, with each item rated on a scale of 1, never to 5, a few times an hour or continue it for half an hour or more. The score range is 14 to 70 points, with higher scores indicating more frequent agitation symptoms. Relevant exploratory endpoints included analysis of the primary and secondary end points at time points other than day 43, including the NPI-NH PS durability of response from day 43 to day 85, the change from baseline to day 43 on the NPI-NH PS by baseline NPI-NH PS and MMSE score subgroups, and on the Alzheimer's Disease Cooperative Study-Activities of Daily Living or ADCS-ADL instrument total score. The caregiver-rated ADCS-ADL was an exploratory functional end point that includes 23 items related subject ADLs and [indiscernible] ADLs and functioning. The primary endpoint, the analysis was performed using the mixed-effect model for repeated measures or MMRM, method in the full analysis set population. The model included the fixed effects of baseline MMSE category, baseline NPI-NH PS with a continuous [ covariate ] treatment, visit and treatment-by-visit interaction. The statistical analysis plan did not specify multiplicity adjustment for the secondary end points. Among the 181 all randomized subjects, 91 were assigned a placebo and 90 to pimavanserin, with Full Analysis Set or FAS of 178 subjects included randomized subjects with both baseline and at least 1 post-baseline NPI-NH PS, with 91 subjects in the placebo arm and 87 in the pimavanserin arm. For the 181 subjects randomized to the double-blind period, 80% were in the placebo arm and 75% in the pimavanserin arm, completed 12 weeks of double-blind treatment. The most common cause for early termination for the total group was adverse events [indiscernible] withdrawal by subject. I'll provide an overview of major protocol deviations as we will discuss these further regarding the applicant's resubmission. The most common categories of protocol deviation were related to missed study procedures, such as labs or vital signs, informed consent and eligibility criteria. The most frequently reported eligibility criteria deviations included use of exclusionary medication at the time of randomization and enrollment without meeting criteria, most notably inability to confirm psychosis onset after Alzheimer's diagnosis. Regarding medications, the applicant noted that subjects were treated by health care providers and nursing homes, who are not involved in the study, and then it was common for medications to be prescribed during the study without the knowledge or consent of the investigator. Treatment with these medications was often a deviation including if given and discontinued pre-randomization if the medications were not discontinued far enough before randomization on later determination. Regarding eligibility criteria, the applicant noted that for many subjects, neither the nursing home personnel, nor the medical record could provide a date of onset of symptoms. In those subjects, the date was reported as unknown or the same day as Alzheimer's onset because it could not be confirmed if the psychosis onset was after Alzheimer's onset. Though the applicant reports that the investigator had determined that the subjects had Alzheimer's and there was no history of other psychotic disorder. Treatment arms were well balanced by sex, age, race and ethnicity. The mean age of subjects was approximately 86 years across both arms. Approximately 80% of subjects were female across both arms. Race was 98% white subjects in the placebo arm and 93% of white subjects in the pimavanserin arm. No subjects identified their ethnicity as Hispanic or Latino. Although the treatment arms were well balanced, the study population was not representative of the U.S. population in terms of racial or ethnic characteristics, being almost entirely white and entirely non-Hispanic or Latino. It is unclear how these differences between the U.S. population and the study population may affect the generalizability of the study results. Multiple analyses have found a higher risk of dementia in black and Hispanic, Latina populations than in white populations. The treatment arms were also generally well balanced with respect to duration of Alzheimer's and psychosis and baseline NPI-NH total scores, NPI-NH PS, MMSE and CMAI-SF total scores. The median duration of Alzheimer's is approximately 57 months. The median duration of Alzheimer's disease with psychosis was approximately 16 months. At baseline, the median NPI-NH PS was 8 on a possible range of [indiscernible]. And the median CMAI-SF total score was over 27, on a possible range of 14 to 70, where higher scores indicate worse symptoms on both scales. A statistically significant treatment effect for pimavanserin versus placebo was observed on day 43 for the NPI-NH PS. The MMRM least squares mean change from baseline was minus 3.76 for pimavanserin group versus minus 1.93 for the placebo group for a treatment difference of minus 1.84 with a p-value of 0.045. Various sensitivity analyses to support the impact of [indiscernible] yielded similar results to the primary analysis, although pimavanserin achieved statistical significant end point as previously discussed, some of the challenges associated with the interpretation of the NPI-NH PS in terms of clinical meaningfulness. The secondary and relevant exploratory end points, none of the between group comparison met nominal significance and demonstrated no notable numerical separation, including the ADCS-CGIC, the CMAI SF total score or the ADCS-ADL total score. Pimavanserin did not separate from placebo on the NPI-NH PS at day 64 or day 85 as seen in this figure. The figure displays the least squares mean change based -- change from baseline on the NPI-NH PS primary efficacy measure over the 12-week treatment period. The treatment effect appeared largest at day 43, but diminished afterwards. Placebo response at day 43 was notable compared to other time points and appeared to increase the treatment difference. Overall, the lack of support from the secondary efficacy endpoints and the exploratory analyses that do not show discernible differences on the NPI-NH PS at day 64 or day 85 raised the question of whether the treatment difference of day 43 is a chance finding, for example, a sudden onetime worsening in placebo group or questions about the durability of effect. In the resubmission, the applicant responded to the study design and conduct concerns outlined in the agency's complete response letter regarding Study 19. At this time, the agency has concluded that Study 19 is an adequate and well-controlled trial suitable for regulatory decision-making. However, as noted previously, there are limitations of the primary NPI-NH instrument, making interpretation of the primary result somewhat challenging. Regarding study conduct, the Office of Scientific Investigations or OSI, conducted an inspection of the applicant during the initial supplement submission rather than an inspection of the United Kingdom study site because of COVID-19-related limitations. Based on the findings from the applicant inspection, OSI had concerns about data reliability because of the number of protocol deviations, some of which could potentially impact whether the method of selection of subjects provides adequate assurance that they have the disease or condition being studied. As described previously, those eligibility violations principally involved subjects who do not have clear documentation that psychotic symptoms developed after Alzheimer's diagnosis has been established, or subject who received exclusionary medication at the time of randomization. The applicant has noted the proportion of subjects with issues related to documentation and diagnosis or who received exclusionary medications was balanced between the treatment groups. The applicant acknowledge that there were difficulties establishing the date of Alzheimer's diagnosis for some subjects, but pointed out that other eligibility criteria excluded subjects with psychosis caused by other conditions, such as delirium or schizophrenia. The applicant also represented per protocol analysis results to demonstrate the impact of protocol deviations. Here, the left side of the table displays the per protocol set analysis, and the right side of the table displays the statistical reviewers analysis that I'll discuss in a moment. As you can see on the left side of the table, the per protocol set results were in favor of pimavanserin, with a treatment effect estimate of minus 3.31 and a p-value of 0.006 compared to the primary analysis treatment estimate of minus 1.84 with a p value of 0.045. As seen on the right side of the table, the statistical reviewer repeated the primary analysis on the non-per protocol analysis set that is those subjects who are randomized if we're not in the per protocol analysis set. The results showed a treatment effect estimate of minus 0.65 for a nominal P value of 0.648. The decrease in the treatment effect estimate raises questions about the applicants contention that the protocol violations should not have affected the results. Of note, almost 47% of subjects were excluded in the per-protocol analysis. Such a large number of randomized subjects excluded from the analysis could lead to selection bias and exaggeration of treatment effect and the results of this subgroup may not be generalizable to the intended population. The full analysis set should be used to assess treatment effect rather than a per-protocol set. Overall, the agency anticipates that we will be able to rely upon the data from Study 19 for regulatory decision making based on the balanced distribution of the protocol deviations and after examining the nature of the deviation and mitigating factors, such as other eligibility criteria. In summary, the results of Study 19 demonstrated a statistically significant result on the primary end point change from baseline to day 43 on the NPI-NH PS. The end point appears to have face validity for a Phase II exploratory study, but the developmental evidence supporting its use is not optimized. The clinical meaningfulness of the treatment difference may be difficult to interpret and would benefit from support to other outcome assessments. On secondary and exploratory endpoints, there was a lack of notable separation from placebo whether nominal statistical significance or numerical, so we lack evidence from the secondary or exploratory end points to assist our interpretation of the primary end point. The lack of discernible differences on the NPI-NH PS primary outcome measure after day 43 raises questions of whether the difference at day 43 is a chance finding or about the durability of effect. Moving on to Study 45. Study 45 was a Phase III randomized, double-blind, placebo-controlled, multi-center randomized withdrawal study of pimavanserin 34 milligrams once daily versus placebo, with potential dose adjustment to 20 milligrams as described on a later slide. Subjects were screened across 101 international sites, including 27 in the United States. The 3- to 35-day screening period included brief psychosocial therapy and an antipsychotic washout if necessary as in Study 19. The 12-week open-label period was followed by an up to 26-week double-blind period. Eligible subjects were 50 to 90 years old, inclusive with all-cause dementia and in clinical criteria for dementia subtype, with a baseline MMSE score between 6 and 24 inclusive and with psychosis symptoms for at least 2 months. Subtypes included Alzheimer's dementia, Parkinson's disease dementia, dementia with Lewy bodies, frontotemporal dementia spectrum disorders and vascular dementia. Subjects must have had screening and baseline scores on the scale for the Assessment of Positive Symptoms-Hallucinations plus Delusions or SAPS-H+D of at least 10, including scores on hallucinations and delusions global items of at least 4 and a score on the Clinical Global Impression of Severity or CGI-S of at least 4, moderately ill. The full 34-item SAPS was designed to measure hallucinations, delusions, abnormalities in language and behavior and disordered thought processes. This study used the 20 items from the hallucination and delusion subscales, which include global ratings of the severity of each. Each item is rated on a 6-point severity scale from 0, none to 5 severe, for a maximum score of 100 on the 2 subscales, with higher scores denoting more severe symptoms. As in Study 19, subjects were excluded for other causes of psychosis, such as delirium and schizophrenia. As in Study 19, anti-dementia drugs, antidepressants and anxiolytics were permitted, if stable before and during the study. No requirement for stability for anti-Parkinson's dopaminergic agents was included. After the first week of the 12-week open-label period, subjects were permitted to decrease the dose to 20 milligrams for tolerability and reincreased to 34 milligrams for efficacy at any scheduled or unscheduled visit until week 4, at which point the dose remains stable. To enter the double-blind randomized withdrawal period at week 12, subjects had to meet both response criteria at week 8 and 12 and remain otherwise eligible. If not, they were withdrawn and under the safety follow-up period. The response criteria required at weeks 8 and 12 included at least a 30% improvement on the SAPS-H+D and the Clinical Global Impressions-Improvement or CGI-I score of 1, very much improved, or 2, much improved relative to baseline. Responders were randomized 1:1, stratified by dementia subtype and region. During the double-blind, subjects were assessed for relapsed psychosis regularly as well as unscheduled visits in context. Subjects were considered to have relapsed if compared to their double-blind baseline, they demonstrated any of the following: at least a 30% worsening on the SAPS-H+D and the CGI-I score of 6 very much worse or 7 -- or sorry, 6 much worse or 7, very much worse, if they were treated with other antipsychotics for dementia-related psychosis, if they stop study drug or withdrew for lack of efficacy or if they were hospitalized for worsening psychosis. An independent adjudication committee reviewed all termination cases that occurred before the study discontinuation date to determine if protocol-defined relapse criteria were met. The primary end point was time from randomization to relapse in the double-blind period. The secondary end point was time from randomization to discontinuation from the double-blind period for any reason. Exploratory end points relevant to the applicant's resubmission included the SAPS-H+D total score and separate hallucinations and delusions domain scores. The total number of relapse events required to the final analysis was 75. Sample size calculation was based on a placebo relapse rate of 60% over 26 weeks and a pimavanserin relapse rate of 35% over 26 weeks for a hazard ratio of 0.47, a dropout rate of 25% over 26 weeks, an overall 2-sided alpha of 0.05 and a one-sided O'Brien-Fleming stopping boundary of 0.0033 for the interim analysis when half of total planned relapse events occurred. The primary end point was analyzed with the Cox regression model with covariates for treatment group, dementia subtype and region. Of the 392 subjects enrolled in the open-label period, 351 subjects completed or discontinued from the open-label period and 41 were still ongoing in the open-label period at the time of study discontinuation, following interim analysis. Among the 351 subjects, 217 or 62% met sustained response criteria at weeks 8 and 12 and were randomized to the double-blind period. The most common reason for early termination during the open-label period was lack of response for 20% of subjects, followed by discontinuation for adverse events for 8% of subjects. In terms of open-label responses, as you can see here in the left column within each dementia subtype, roughly 60% of subjects with Alzheimer's met the response criteria and were randomized and roughly 71% of the subjects of PDD met the response criteria and were randomized. In the right column, roughly 19% of subject with Alzheimer's were considered to have a complete response, defined as a 100% symptom reduction on the SAPS-H+D and CGI-I of 1 or 2 and roughly 27% of subjects with PDD had a complete response. In the open-label period in both double-blind arms, subjects included roughly 60% females, mean age was roughly 74 years, race was almost entirely white, and ethnicity was roughly 76%, not Hispanic or Latino. In terms of racial characteristics, the study population was not representative of the U.S. population being almost entirely white. Generally, dementia subtype distribution was similar between open-label and double-blind periods in both double-blind arms, with approximately 63% of subjects with Alzheimer's and 19% with PDD in the double-blind period. Double-blind baseline mean MMSE scores were generally similar between the arms. Mean SAPS-H+D scores improved from open-label baseline at 24.4% to similar double-blind baseline in both arms at 5.0 for pimavanserin and 5.2 for placebo. As a reminder, the possible range of SAPS-H+D scores is 0 to 100. I'll turn it over here to my statistics colleague, Dr. Xiang Ling to discuss Study 45s efficacy results and resubmission analyses.

Thank you, Dr. Bossie. My name is Xiang Ling. I'm the strategical reviewer for Study 45. I'll cover the next few slides on the strategical analysis. Study 45 met its primary end point of current randomization to relapse in the double blind period. In accordance with the statistical analysis plan, an interim analysis was conducted after 40 relapse events had occurred. The prespecified stopping criteria was made of interim analysis because of one sided p-value of 0.0023 was less than the O'Brien-Fleming stopping boundary of 0.0033. The study was stopped early for efficacy. However, there are large differences in the estimates of the treatment effect in terms of [ hazard ] ratio across the dementia subtypes. Only the treatment effect in the subgroup that include PDD subjects appear to differ from placebo with [ confidence ] intervals, excluding low effect, hazard ratio of 1. For the AD subgroup, the confidence interval includes hazard ratio of 1 and is white, indicating large statistical uncertainties about the estimated treatment effect. Additionally, the confidence interval for the AD and the PDD subgroups did not overlap, suggesting differential treatment may effect across dementia subtypes. However, it is important to note that the study was not powered to provide reliable estimates of the subgroup effects and differences. Additional exploratory analysis of the primary endpoint, excluding the PDD subset did not meet the O'Brien-Fleming stopping boundary of 0.0033, not the nominal one sided significant level of 0.025. Of note, this exploratory analysis have reduced power. In the next few slides, we'll discuss the resubmission with a focus on the analysis of the AD subgroup in accordance with the revised indication of the treatment of hallucinations and delusions associated with Alzheimer's disease psychosis. In the resubmission, the applicant ascertain that there was consistency of response across dementia subtypes. The applicant hypothesized that the PDD subgroup's smaller hazard ratio was caused by the use of dopaminergic therapy to manage motor symptoms of Parkinson's disease, which could cause a worsened psychotic symptoms. Additionally, the applicant conducted reanalysis of the primary and the exploratory efficacy end point for the AD subgroup as well as on exposure-response analysis that examines the relationship between plasma pimavanserin concentration and the primary efficacy endpoint. We'll discuss each of them in the following slides. The applicant conducted a test for qualitative or crossover interaction and concluded that the treatment effects are directionally consistent. However, there's apparent variation in the magnitude, though not the direction, of the treatment effect across subgroups. The treatment effect estimates are very different between the AD subgroup and the PDD subgroup, and the confidence intervals for the AD and PDD subgroup do not overlap. We conducted an analysis that includes the interaction of the treatment by the dementia subgroup stratification factor in the primary analysis model, but that appears to show evidence of our quantitative or [ noncore ] interaction for differential treatment effects across subgroups. As we have seen, subgroup analysis by dementia subtype suggest differential results. In particular, there is a big difference in placebo response across subgroups, which may be due to the dopaminergic medication use according to the applicant's hypothesis. However, dopaminergic medication use was almost completely confounded with the dementia subtype. Almost all subjects with PDD were on dopaminergic therapy, while a few subjects in the non-PDD subgroup were on this therapy. Therefore, it is not possible to statistically adjust for the dopaminergic medication effect for the PDD subjects receiving placebo. Furthermore, it's unclear whether the effect of dopaminergic medication and the risk of relapse is the only explanation for the possible difference in the treatment effect between the AD and PDD subgroup. Still, this does not affect the assessment of the treatment effect for the AD subgroup, which is a focus on the resubmission. The prespecified primary analysis for time to relapse was based on the Cox regression model with treatment, designated dementia subtype and the region as factors for the analysis of the overall population. Both the dementia subtype and the region were stratification factors for the randomization. The applicant conducted a modified Cox regression analysis. That included 4 factors selected post-hoc for the AD subgroup and excluded the prespecified region factor. The results showed a smaller hazard ratio of 0.475 and the smaller P value of 0.1 compared to the prespecified primary Cox model. There are some caveats to the post-hoc analysis. The choice of covariates should be prespecified and post-hoc data-driven analysis are difficult to interpret and may be prone to bias. In addition, for the covariate of the baseline severity of psychosis, the applicant used the open-label baseline SAPS-H+D score instead of the double-blind baseline score without providing justification. Arguably, the double-blind baseline score maybe more appropriate when testing the treatment effect in the double-blind period. And there's no reason to exclude region, which was a stratification factor and the prespecified covariate for the preliminary analysis. We conducted a similar post-hoc analysis adjusting for the same covariate the applicant selected, except that the open-label SAPS-H+D score was replaced with double-blind baseline score. In addition, we added back the prespecified region covariate. The resulting hazard ratio is similar to that of the prespecified primary model. In summary, none of the p values reached nominal statistical significance. The modified model used by the applicant is not justified. And post-hoc and potentially data-driven analysis are very challenging to interpret. Influence on the treatment impact should be based on the prespecified primary analysis, unless in the real situation where the primary analysis is clearly invalid, which is not a case here. The most relevant exploratory end point for this study was changed from double blind baseline in SAPS-H+D core. The applicant conducted post-hoc analysis on this end point for the AD subgroup using a nonparametric test on rank scores. Specifically, the applicant assigned the same best or second-best rank to over half of the subjects, whose scores never worsened during the double blind period. This analysis yielded a normal p value of 0.0375. However, for these subjects whose scores never worsened, there were still differences in terms of how much SAPS-H+D score changed from baseline. Additionally, relapses may be considered the worst outcome regardless of the change in SAPS-H+D score. We conducted analysis using the same nonparametric test, but with ranks assigned differently. We assigned worse ranks to subjects whoever relapsed based on the time to relapse and assigned better ranks to those who never relapsed on the maximum change of score. This analysis yielded a nominal p-value of 0.1355. In summary, results of the exploratory end point of SAPS-H+D score did not provide much additional support for efficacy. Dr. Bossie will now present the exposure-response analysis and the concluding remarks.

The applicant also conducted an exposure-response analysis to evaluate the relationship between pimavanserin plasma concentrations and time to relapse in Study 45 to provide supportive evidence for efficacy. The exposure-response analysis assessed for the efficacy difference between Alzheimer's and PDD subgroups associated with plasma concentration and its variability. However, it does not appear that differences in subgroup efficacy are related to pharmacokinetic exposure differences, as exposures were similar between the Alzheimer's and PDD subgroups. Higher pharmacokinetic exposures were associated with a higher relapse-free probability for both subgroups, but the drug effect was lower for the Alzheimer's subgroup than PDD subgroup. In summary, Study 045 demonstrated a statistically significant result on its primary endpoint of time to relapse in the double-blind period. However, overall results appear driven primarily by the PDD subgroup, suggesting a possible differential response to pimavanserin across dementia subtypes. It is unclear whether the effect of dopaminergic medication on the risk of relapse is the only explanation for possible differences in treatment effect between the PDD and Alzheimer's subgroups and use of the medication is confounded by dementia subtype. Finally, post-hoc analyses demonstrated mixed results and are subject to inherent limitations. I'll summarize the overall evidence and uncertainties to conclude. In terms of evidence, both Study 019 and Study 045 demonstrated statistically significant results on their primary endpoints. In Study 019, on the NPI-NH PS change from baseline to day 43; and in Study 45 on the time from randomization to relapse in the double-blind period in Study 045. In terms of uncertainties, for Study 019, the primary endpoint, NPI-NH PS appears to have face validity for a Phase II exploratory study, but the developmental evidence supporting its use is not optimized. The clinical meaningfulness of the treatment difference may be difficult to interpret and would benefit from support by other outcome assessments. There was a lack of notable separation from placebo on secondary and exploratory endpoints. So we lack evidence to assist our interpretation of the primary endpoint. In the lack of discernible differences in the primary outcome, NPI-NH PS measure after day 43 raises questions of whether the difference of day 43 is a [ chance ] finding or about the durability of effect. For Study 045, the primary endpoint results appear driven by the PDD subgroup for whom pimavanserin is already indicated as a population with Parkinson's disease psychosis, with and without dementia. It is unclear dopaminergic medication use is the only explanation for the subgroup efficacy difference between PDD and Alzheimer's. Post-hoc analyses offer mixed results and are subject to inherent limitations. That concludes our presentation. Thank you for your attention.

We will now take clarifying questions for the agency. Please use the raise hand icon to indicate that you have a question and remember to lower your hand by clicking the raise hand icon once again after you've asked your question. When acknowledged, please remember to state your name for the record before you speak and direct your question to a specific presenter, if you can. If you wish for a specific slide to be displayed, please let us know the slide number, if possible. Finally, it would be helpful to acknowledge the end of your question as a thank you and end of your follow-up question with that is all for my questions. So we can move on to the next panel member.

The first question from Dr. Follmann.

Yes, thanks. I had a question about the effect of dopaminergic medication. So I look at the [indiscernible] that you have a very small p-value saying there's a difference in the treatment effect between the PDD and AD groups. So that's an important result. And whether or not that is driven -- I mean, it's driven by the PDD group, but whether or not that is further caused or driven by dopaminergic medication? Why does that matter? And if you concluded, it was entirely due to dopaminergic medication, that changed your conclusions. It seems to me you'd still have essentially an underpowered study in the AD group.

This is Tiffany Farchione, the Director of the division. So I think the issue here is that it seems like a reasonable explanation to say that if you have a dopaminergic medication on board that potentially when the pimavanserin is withdrawn that, that could drive a faster relapse of psychotic symptoms. Unfortunately, that's a hypothesis. We aren't able to say one way or the other based on the data that we have available. It does seem like a reasonable hypothesis, but we just can't -- we can't answer that question with any kind of certainty at this point.

Many of [indiscernible] this hypothesis was true, how would I interpret the effect in the AD group differently than what it is, which is sort of marginal or not really significant and underpowered?

Right. Well, I mean that's one of the questions that we're really asking the committee to opine on in terms of the strength of the data that we have available to us. I mean I think that ultimately, the best way to respond to that would be to have a study in the Alzheimer's-only group, but we don't have that at the moment. But we do have a package available that has some -- that has some evidence for us to review. These are things that we agree would be review issues at the time of resubmission. But it does add a layer of uncertainty. So we're certainly interested in the committee's opinions about the overall strength of that data.

Our next question is from Dr. Thambisetty.

Dr. Narendran. Madhav Thambisetty from the NIA. The FDA sent in its complete response letter in April 2021, it did not consider Study 019 to be adequate and well controlled. In the Type A Review meeting in June 2021, they advised the sponsor to perform a new study of specific dementia populations, example, Alzheimer's disease. This advice was again reiterated in December 2021 in the Type B guidance meeting, where, again, the agency continued to advise the sponsor that an additional adequate and well-controlled study in AD Psychosis would likely provide the strongest data in support of a resubmission. And with all of the analysis presented today, Study 019, in my opinion, still remains not adequate and not well controlled. The most substantial analysis presented to support Study 019 in the resubmission, to me, it looks as it consists of throwing out 47% of data from protocol violators and then showing that there is a large treatment effect, which to me is not valid in any way because you cannot throw out nearly half of data of randomized participants to support the analysis. And the fact that, that seems to be the only substantial analysis in the resubmission in support of 019 to me seems quite inadequate. Now does the FDA believe that the resubmission analysis, excluding 47% of the data now renders Study 019 adequate and well controlled because that is not entirely clear to me.

So this is Tiffany Farchione again. So it's not the per protocol analysis that renders it adequate and well controlled. It was the deeper examination of the nature of the violations and the balance of the protocol deviations across the 2 groups. So we're still looking at the overall results from the full analysis set. But in following up on some of those individual deviations, we're reassured about the quality of the data.

Our next question is Dr. Walter Dunn.

Walter Dunn, UCLA. This is a question for Dr. Bossie about Study 019. So 2 questions about it. Number one, I think there's been discussion about the lack of racial and ethnic representation in terms of generalizability to the U.S. population. Has there been a precedence in the FDA about accepting the result of a trial that were conducted exclusively outside the U.S. for an acute phase treatment study supporting either a supplemental or initial novel drug approval? And then the second question also related to 019. How often do you see protocol deviations in the 50% to 55% range? Obviously, you noted that it's quite high, but generally, what's the baseline that you see across your other studies?

So quickly, I know you direct -- I know -- this is Dr. Farchione again. I know you directed that towards Dr. Bossie, but perhaps that would be a better question for Dr. Dunn considering his broader perspective of agency precedent.

Sure. I'll start [indiscernible], just unmuting. So to the first question, I want to make sure I got those in order since they're directed to Dr. Bossie. Your first question was about basing approvals on foreign data essentially?

Correct, -- any foreign population, non-U.S. population?

Yes. At a high level, it's easy to answer that is yes. We are able to base approvals and considerations data on foreign data. We do need to have work with the sponsor to understand the applicability of the foreign data to the domestic population. And as you heard in Dr. Bossie's presentation, some of the characteristics of that population are obviously different from our overall demographic makeups. Quite honestly, and sadly, that's not different than many of our domestic trials as well. But as I think we all know, this is an area of tremendous focus, but we do routinely encounter data from nondomestic or ex-U.S. sources. And as long as there's no scientific reason to believe that those data are [ inapplicable ] to our population, we can rely on them for regulatory action.

Okay. And then regarding the protocol deviation rate of 50% to 55% in 019?

Right. Yes. So the second part of your question, it's less about the rate. I don't know that anybody on the team is going to have those data at their fingertips in terms of a comprehensive analysis of what is typically seen. I think it's about the character and understand the potential impact. And I think you heard some presentations from the team pretty clearly discussing this and Dr. Bossie and Dr. Xiang Ling can refer you back to the slides where this was discussed, but it seems that the team has looked at this and felt that the character, notwithstanding their quantitative count, but the character of the deviations has been considered at some detail internally. And I think you heard the team's assessment that the study is suitable for consideration. It's the primary study offered for support of the Alzheimer's population, and it won, right? It's a positive study on its endpoint. And so the question for the committee is really the same question that we're facing here, which is what is the persuasiveness of the data that are provided by the sponsor. They do have a study in Alzheimer's disease as Study 019. And you've heard some discussion about the character of that study. And they have some support from other studies, and we also explore studies to sort out how much supportive evidence comes from within a study or from other sources of data. So you're presented with the same things that we're thinking about. We're trying to sort out what that primary source of evidence is and what else might support it. You kind of heard from the team about some of the issues related to the secondary endpoints in that study and how those might play a role. But Study 019 in Alzheimer's patients did win, and those deviations are not felt to detract on that by the team at this point.

Our next question is from Dr. Iyengar.

I guess my question is basically this, one of the features of randomized withdrawal design is that because only responders are included at the second stage, the treatment effect is generally believed to be overestimated. So there's a bias inherent to the design. Is there any -- did anyone either at the FDA or with ACADIA do any sort of assessment of what the magnitude of that bias might be? That's my question.

This is Dr. Farchione again. So what kind of analysis would you have in mind in terms of evaluating that?

I guess some sort of simulation study -- well, I -- perhaps using some of the Study 019 data to get an estimate of an effect and then use that in a simulation study to assess what the bias -- magnitude of the bias might be in the Study 045. I know that this is fuzzy, but one of the things that I've heard repeatedly about randomized withdrawal design is, we expect it to be biased in favor of the treatment. I've just never heard about, okay, how biased is it? And that's all I'm asking at this point.

Right. I mean I don't believe that anyone on our team has done an analysis of that kind. I don't know if the applicant has anything to add to that.

Next question is from Ms. Witczak... [Technical Difficulty]

Sorry, this is Daryl Dekarske with ACADIA. We had a technical difficulty. Thanks for passing the question, Dr. Farchione. I'd ask Serge Stankovic just to comment briefly on the randomized withdrawal design and the question around the bias.

Yes. Serge Stankovic, ACADIA. Direct answer to your question, we do not have or did not perform any estimate in that regard. Frankly, we're not quite sure how would that analysis look at all. So we really don't have the response to your question either. I would just say is that in respect of the purpose of the randomized withdrawal trial, which is to evaluate a maintenance of effect, we do not consider that, that design is inherently biased in demonstration of that maintenance of effect. And the second point that I would like to make is that Study 045 data on ADP in the context of our overall submission is a supportive data to the evidence of efficacy we presented with other studies. And the fact that it's overall positive in the -- another closely related indication of dementia-related psychosis, it's also supportive of the evidence of efficacy.

Next question, Ms. Witczak.

Kim Witczak [indiscernible] Consumer Rep. In preparation for this meeting, I was -- it looks like NUPLAZID has been considered an atypical antipsychotic, but it looks that the mechanism or something with the inverse agonist. And I'm curious if you could explain what that is and that mechanism. And then who -- like who determines that? Is that something that the company determined or did the FDA determined it to be an inverse agonist?

So this is Dr. Farchione again. So the designation of atypical antipsychotic is -- it's a limitation of our terminology in terms of medication class. I mean, pretty much anything that's not like an old [ school held out ] type antipsychotic. Any of the newer generations from Risperidone onward would be considered atypical antipsychotics, even though they all have sort of different profiles of receptor activity. Most of the atypical antipsychotics, the ones that are used for treatment of schizophrenia are dopaminergic in terms of their action. So in response to your second question, yes, the company provides data from animal studies and receptor occupancy studies and things like that. The agency does evaluate that, and we determine what goes into the label in terms of the description of the mechanism of action. And oftentimes, we are narrowing that. I can't recall on this application in particular, like if there were any differences between what the company said and what we said. But there have been occasions where a company will try to make broader claims and will be like, no, no, no, this is what we are going to say about the mechanism of action. So yes, it is something that we would review very carefully from our nonclinical team.

Okay. Because I was wondering if that would be -- thanks for that clarification of like [ what is ] atypical. Also then I was thinking is that what the point of differentiation is from a marketing standpoint. So -- but it sounds like it still came from the company, and then you have to analyze it. Is that correct?

Yes. I mean the data always originates with the company, but we do our own independent evaluation of what they submit, yes.

Our next question is Dr. Cudkowicz.

I have 2 questions. First, about 019, I mean I understand that this is the trial really being considered as whether it's persuasive [indiscernible] in Alzheimer's. And one thing has come up from the FDA is the concern about the primary outcome measure. I wanted to learn a little bit more about that because we did hear from the experts that ACADIA'S brought in who are treating patients and leaders in the seal that this is a good outcome measure. And not having that much familiarity about it. I'd like to understand more about that because this trial was positive and, in my opinion, was persuasive on that outcome measure. So can you explain a little bit more besides some of the examples you gave or why you don't think it's a good outcome measure. What's better in this field for psychosis in Alzheimer's?

So this is Dr. Farchione again. I will pass that over to David Reasner from our Clinical Outcome Assessment.

Yes, David Reasner, Division of Clinical Outcome Assessment. Well, that's a very broad question. So I will identify a couple of areas where we have an interest in additional evidence that supports the endpoint. One area is what we would describe as content validity. Often that comes from qualitative research with patients, caretakers and treating health care professionals. Another area is on the quantitative side, that would be the psychometric properties. Interesting psychometric properties might include reliability between raters, for instance. With regard to this particular assessment, while there was qualitative research conducted, it did include all the areas we would typically expect when a sponsor provides a supportive evidence [indiscernible] for a particular endpoint. With respect to the psychometric properties, we've already pointed out, to a certain extent, the difficulty of working with the total score. But in general, the psychometric properties neither correct nor undermine the core content validity of the instrument, which we believe reflects relevant and important concepts. However, the dividends package as a whole is not as broad as we would have expected. I think in part, there was a focus on other targeted concepts of interest. So we have fewer assessments, secondary endpoints, secondary assessments with which to rely on -- thank you for your question.

Okay. I have now just a question about 045, which I view kind of is a supportive study. The sponsors provide some other analyses around the number of -- percent of patients with worsening symptoms by degree and I didn't see that in the FDA's [ released ] presentation or I missed it in the briefing book. I was just wondering your thoughts on that and the relevance. This is like the percent of people who worsened by a certain amount comparing in the Alzheimer's group treated versus placebo.

This is Xiang Ling, Statistical Reviewer.

Thank you, Xiang Ling. I was just about to throw it over to you. Thanks.

So we did talk about the analysis of SAPS-H+D score. And that [indiscernible] is just another presentation. It's a discrete presentation of the SAPS-H+D score. It's descriptive in nature. But the analysis applicant conducted related to that [indiscernible] test that we mentioned in our presentation. So in their conclusion, they said that [ the analysis ] showed nominally statistically significant result with a p-value of about 0.04 and our analysis take into consideration of the relapses, but will add to actual maximum [ change ] score for all the patients and our analysis resulted in a p-value of 0.1355.

We have another question from Dr. Walter Dunn.

Walter Dunn, UCLA. This question is regarding the statistical approach used by the FDA and there were subsequent analyses for Study 045. So this is a similar question to what I posed to Dr. Hendrix. It looks like there were 2 different approaches, the FDA looked at the effect of ADP or an ADP subgroup by removing PDD. And as you mentioned in your presentation, Dr. Ling, that you lose power and the conclusion, at least was that it was not significant based off the nominal p-value. ACADIA went out in a different way where they did a tipping point simulation when they preserved that population that decreased the contribution from the Parkinson's group. So can you -- and just kind of qualitatively comment on the advantages and disadvantages of both approaches and perhaps why the FDA chose that approach versus a similar tipping point simulation that the sponsor carried out.

Sure. The applicant conducted simulation to show that the overall study still had a large probability of success after final analysis, even if the true main effects in the PDD subgroup were attenuated. However, trial conclusion should be based on actual data instead of simulated data. Additionally, as the applicants' proposed indication was the resubmission and it is changed to ADP, we are interested in the probability of success at the final analysis for the AD subgroup instead of the overall population, if in truth the treatment effect in the AD and the PDD defer. The probability of success at the final analysis for the AD subgroup is about 19%, assuming that the treatment effect would not change over time. This suggests that even if the trial was not stopped early, the study would have [ low ] chance of success for the AD subgroup at the completion of the study. This is due to the smaller sample files for the AD subgroup as well as the smaller treatment effect size for the AD estimated at interim analysis compared to the assumed effect size at the trial designing stage.

So if I understand the kind of the main differences -- yes, if I understand the main differences in the 2 approaches. So the applicants' approach would maintain a similar effect size but have a larger end, while the FDA's approach would maintain a similar effect size have a smaller end, hence, the kind of the different conclusions?

That's correct. The applicants approach by adding more relapses will actually increase the number of events and increase the power. And our analysis with subgroup AD will decrease the power.

So when the applicant added events, did they actually change the overall end? Or did they just switch from non-relapsed to relapse?

The power is related to the number of events, not the number of subjects. So by increasing the number of events, it increases the study power.

Next question is Dr. Follmann.

This is just a comment on the point about the bias of the randomized withdrawal study if that's a pay to talk about. Yes. So yes. So an analogy is that like in blood pressure trials, you'll try and identify people who are hypertensives so you'll get people who are -- have true high blood pressures, but they might be particularly high on that day, sort of a [ randomize ] in addition to having a true high blood pressure. So if you [ mentioned ] the next day, the blood pressure goes down. I mean it's known statistically is [indiscernible] problem and you can correct for that. And I think what's going on here is just the themes that are -- maybe you don't call it relapsing and remitting but there are periods when you [indiscernible]. And if you sort of grab people when they're not having psychosis, which is sort of what you're doing during the open-label phase and maybe they're doing for a bad episode later. So I think that's the fundamental thing. It doesn't really lead to a biased estimate of the between group difference because you're selecting both groups during open label, everything is fine. But if you want to know what is the risk of relapse, then within the drug arm, then you do [ have this bias ] problem.

Thank you for that comment. Our next question is Dr. Baker.

Yes, this is Robert Baker, the Industry Representative. I also was going to ask about the randomized withdrawal design for Dr. Farchione, or whomever she'd like to designate on this. I think we've heard a few concerns that maybe accepting the last one were tied to the exclusion of patients who don't respond in the open period or even though it might be particularly biased in psychiatry. So from a perspective of the industry, I was thinking about randomized withdrawal, which is an enrichment design and has to be interpreted fairly as to how generalizable it is for the population outside the enriched cohort, but nevertheless, is commonly used across therapeutic areas. And I wouldn't see a particular reason why psychiatry would be not a place for it to be used and as it looked like the division had -- after some discussion with the sponsor agreed to the approach. So I just would be interested in your thoughts on this or confirming that in the context of other sources of evidence it is a way to establish a drug effect.

Right. So this is Dr. Farchione again. And I think your last comment in terms of it being in the context of other evidence is the key point here. In terms of standing on its own, I'm not sure that, that would be appropriate. In this case, we have 2 other potential sources of evidence, right? You have the assertion from the applicant that we should be considering these as closely related conditions. So we do have the prior of the approval in Parkinson's disease psychosis, but then you also have the data from Study 019. So again, that's really why we're asking about the overall strength of the evidence and we're ultimately going to ask the committee to discuss the contribution of Study 020 to the overall evidence base for this program because that's really the crux of the question here is how much can we glean from those other studies, given that this Study 045, the randomized withdrawal is really intended to be supportive data in this context, not as a primary source of evidence.

Our next question is Dr. Thambisetty.

Sorry about that. Madhav Thambisetty, NIA. This is a question again for the FDA. It's not entirely clear to me that looking to Study 020, the initial study that's on the basis for the approval is valid here in this situation. Because the data that the FDA has presented and analyzed clearly show that the treatment response is different in AD Psychosis from PD psychosis. And so given that there is convincing evidence of a treatment by subgroup interaction, to me, it's far more compelling that these subgroups behave differently in response to treatment. Then looking to Study 020 as a prior indicating that, that is -- that lends some support. I think the analysis presented today is very convincing, at least in my mind, that there is a very strong interaction for treatment by subgroup. And the results presented clearly show that the AD Psychosis subgroup behaves entirely differently from the PD dementia subgroup. My question was with regards to the uncertainties presented on Slide 56. And I think this is important, at least in my mind, because it draws a sharp contrast between the interpretation of the results by the FDA's reviewers and those of the sponsor presented earlier this morning. So to my mind, I think Slide 56 clearly summarizes the reasons why Study 019 is not persuasive because it seems to be driven by the placebo worsening at week 6. It's not durable because the curve do not separate out at 9 and 12 weeks. And the magnitude of the effect calls into question the clinical meaningfulness. And that is well summarized in Slide 56. And I just wanted to clarify with the FDA that, that is, in fact, the position that there is a clear difference between how they interpret the results of 019 from what the sponsor presented earlier in the morning with respect to the placebo worsening driving the results and the lack of durability beyond week 6.

This is Dr. Farchione. So these are -- as you know, these are things that we're presenting as uncertainties, stuff that we have questions about. But again, this is one of the primary reasons for seeking advice from the committee at this point is because these remain unresolved issues in the review process. So we're very interested in your opinion. And I think that you've stated your opinion fairly clearly at this point. But again, you'll have an opportunity to summarize in the discussion portion and with the vote.

Our next question is Dr. Iyengar.

Sorry. I had just forgotten to put my hand down. Sorry.

Then the next question is from Dr. Krishna.

This is Sonia Krishna at University of Texas, Austin. I'm also very concerned about the change at day 43, like Dr. Thambisetty's last comment was there. In the morning presentation by ACADIA, it was clear that, that's the endpoint showing that the medicine is efficacious. And then the FDA presentation, it looks like this could be a random error. Is there any recommendation by the FDA of how to determine whether or not this is random? And then my second question is related is just do we know how long it actually takes the medicine to start working, if it appears that the main point is by 6 weeks and maybe not be sustained after.

This is Dr. Tiffany Farchione again. I can start with the second question first in terms of the other source of information that we have for this would be the original pimavanserin development program in Parkinson's, where that was also a 6-week endpoint. So -- and again, that was a very strongly positive study at the time. So I think that 6 weeks is a reasonable expectation, and the study was designed based on the assumption that we'd be able to see in effect at 6 weeks. As for trying to determine whether it was a random blip or a real effect, this is again one of the reasons why we list the uncertainties that we have, we would typically look to things like related secondary endpoints or things of that nature. And in this case, when we look at the secondary endpoints, we don't have additional support. Now not all [indiscernible] I mean, really the secondaries that are really measuring the same thing are the things looking at time point. And you can see on the various graphs, what that looks like. The other secondary endpoints are -- would not be considered supportive either because they're not nominally statistically positive or they're measuring different things. So it makes it more difficult to really understand what that effect at day 43 is.

Our next question is Dr. Walter Dunn.

Walter Dunn again, here from UCLA. So kind of a broader question for the division and also a clarifying question. So in the briefing documents, there was not extensive mention about Study 020, although the applicant certainly emphasize the deposit results in that study. But in none of the voting questions or discussion questions talk about opining on the results of that in terms of [ influencing ] our decisions. But is that something that you would want us to formally consider when talking about overall effectiveness or about the evidence?

This is Tiffany Farchione again. No, Study 020 was the study that supported the original approval. So we're not here to relitigate those findings. That was a positive study. It led to the original approval. We believe that pimavanserin works and works well in the population for whom it's indicated right now. The question of the relevance of Study 020 for this application has to do with the relatedness question and it's being positioned as the idea that you have a closely related condition. Like we said, there's psychosis present in both Parkinson's disease and Parkinson's with dementia as well as in Alzheimer's. Now normally speaking, if you were just looking at the symptom across different disorders. So you have 2 different types of dementia, you have 2 neurodegenerative disorders and both of them have psychosis. Again, a priority, that's why we thought that it was reasonable to include the 2 populations together in a single study because they seem related. We have no reason to believe otherwise. Now what we're asking the committee to discuss related to that is in that context of using it as support for this application and looking at the data from Study 045, which again was not powered to detect subgroup differences. How would you interpret that? And how would you weigh Study 020 in your overall evaluation of the evidence for this program.

I probably should have clarified why I asked that question, you addressed it specifically about, yes, I think another key or probably the critical question for me is how related are they, ADP and PDP? And what does the current evidence tell us. So okay. So that sounds like that's something that you would certainly want to kind of hear about our opinions as far as why we either believe or do not believe that the 2 conditions are either closely related or completely unrelated.

Absolutely, yes. That's what we're hoping for in the discussion. Thank you.

We have another question from Dr. Thambisetty.

Thank you, Dr. Narendran. I'd like to call attention to Slide 48 from the FDA's presentation if possible. These are -- this slide refers to one of several post-hoc analysis performed for Study 045. While they are pulling up the slide, I can also reference Page 33 of the applicants' submission and Figure 19 under the heading, Substantial Evidence for Effectiveness for AD Psychosis. And this is again relevant to the one of many post-hoc analyses that were performed by the applicant. And it looks as this -- the results that the applicant chose to present on Page 33 are what are being referred to here as the applicant modified [indiscernible] at Slide 48, the previous slide. And so to me, I respect the fact that the applicant did say that all of these analyses were post hoc and therefore, should be considered exploratory, I think that's commendable. But to me, it looks as if this is really an exercise in data dredging because you're using a set of post-hoc covariates that were not prespecified. You're dropping a covariate that was, in fact, prespecified. So the region factor was a prespecified covariate that has been dropped for no reason, at least no reason in the materials that we were presented with. So I'd like to ask if the FDA reviewers or other people who analyze the data, the FDA, had any rationale presented to them by the sponsor provide this particular set of covariates were chosen? Why a prespecified covariate was dropped from these analyses? And was there a list of other models that were run with the covariates that did not show comparable reserves. So I'm just trying to understand the choice for the covariates used in this analysis by the applicant and whether the FDA had any data or information as to why these were chosen?

Dr. Ling?

The applicant didn't provide a rationale for dropping the region factor, but they did provide rationale for selecting [ the 4 ] covariates. It's basically based on [ literature ] and the results of prior studies. And maybe the applicant could add more details.

So there's no reason to only choose baseline severity of psychosis during the open-label phase and not in the double-blind phase. So you've clearly shown in your own analysis, Dr. Ling, that when you use the psychosis severity in the double-blind phase, you get a different set of results. So to me, it looks as if this -- the impression that I'm getting is that a variety of models were run with various permutations of covariates, and what has been shown here is the model that used the most ideal combination of covariates to show the result that we're seeing here?

So this is Dr. Farchione. I would like to point out that we don't have any evidence that the applicant would have done a bunch of analyses and then only presented the most favorable to us. Again, perhaps ACADIA can comment on the specifics of why they chose this model and talk a little bit about their model development process.

Thanks Dr. Farchione. Now I'll ask Dr. Susan Hendrix to speak a little bit further about various covariate adjusted models.

Thank you. Susan Hendrix, Statistical Consultant. When we were developing the model for the covariate adjustment, we were looking at a couple of things. The first is whether there were baseline imbalances in some of these factors and then correcting for those imbalances because of the post hoc subgroup nature of the ADP population specifically. And we had achieved significant overall in the DRP, but because we weren't powered to see significance in the smaller subgroup, we knew that those baseline imbalances could make a bit of difference. We excluded region primarily because there were smaller sample sizes in some of the regions in 4 separate regions. And so with those smaller sample size, there were some potential convergence issues with that model. When we received the response from the FDA, we went back and took their model, which they had determined with an AIC criteria, and we actually did another model where we included our baseline, which was the double-blind baseline with their model. So we put region in, did double-blind baseline. We got actually even a better AIC again, using the FDA's criteria for the model selection. And the main reason we use double-blind baseline -- sorry, that we use open label baseline rather than double-blind baseline was that at the open-label baseline, there was a lot more difference in the patients because it was prior to treatment. So they came in with all their different severities of disease and at the double-blind baseline, everyone was on treatment, so they looked much more homogeneous. When we put both regions or sort of both baseline models -- baseline terms in the model together, the double-blind baseline does not add significantly to the open-label baseline. But across all of these different models within the ADP Group and the ADP 34 mg dose, we get consistency of the hazard ratios with hazard ratios on the top of the figure here from 0.48 to 0.64 within the all doses group, and with 34 mg 0.35 to 0.49. And the primary model in my mind based on the AIC is actually the second from the bottom, where we have a 0.42 hazard ratio, a p-value 0.064 that had the best AIC and included both terms that the FDA had suggested and the terms that we had prespecified or that we had designated from the literature and from past experience.

Another question from Dr. Apostolova.

It's not a question. I'm just again will postulate a bit and extend some observations from the pathology literature, which might actually explain the smaller effect size in Alzheimer's. And that is that we know that cytosis, first of all, that is the defining -- one of the defining criteria for dementia with Lewy body and also is extremely frequent in Parkinson's disease dementia. And that is because it's strongly associated with the presence of Lewy bodies. And in Parkinson's disease dementia, everybody has Lewy body. In Alzheimer's disease and all other disorders, about 50% of patients have concomitant Lewy body pathology in the limbic at a minimum part of the brain. So that could explain why there is a little bit of differential effect. We know that Lewy body pathology is associated with the dopaminergic dysfunction. So I'm just offering the explanation. We don't have to anticipate similar effect size in these disorders based on what we know pathologically. Thank you.

If there are any other questions.

I have a quick question for Dr. Farchione. I mean one of the things is NPI psychosis scale, it seems suboptimal. There's questions about content validity and there was a very small effect. I mean, obviously, was this discussed early on? I mean all the other trials were done with the SAP, which seems like a lot more robust and reliable. I don't know to what extent did this come up ahead of the trial? Or was it just kind of let go because it's Phase II and was exploratory at that point?

So this is Dr. Farchione. So the earliest discussions of the study design and the endpoint and everything happened back in like 2008. So, to give you some impression, that was before I even started at the agency. So it was -- at that time, we didn't actually even have -- you know the current iteration is the clinical outcome assessment division. But prior to that, it was something called SEALD, which was the study endpoints and labeling development team. We didn't even have SEALD yet at that time. So really the assessment of endpoints back then was primarily one of phased validity more so than anything else. But again, with it being initially conceptualized as an exploratory study and to be part of a larger development program, the idea of going back to look at that endpoint with greater scrutiny even as time went on, didn't really come up. So that's sort of the history there.

Thank you. Any other questions? [indiscernible] screen. Dr. Apostolova I see your hand is still raised, do you have another comment or question?

No, sorry.

Okay. Is that it. I guess, we could break for lunch 10 minutes earlier than anticipated. So we will now break for lunch. We'll reconvene at 2:00 p.m. Eastern Time. Panel members, please remember that there should be no chatting or discussion of the meeting topic with other panel members during the lunch break. Additionally, you should plan to rejoin at around 1:45 to ensure that you're connected before we reconvene at 2:00. Thank you. [Break]

We will now begin the open public hearing. Both the FDA and the public believe in a transparent process for information gathering and decision-making. To ensure such transparency at the open public hearing session of the advisory committee meeting, FDA believes that it is important to understand the context of an individual's presentation. For this reason, FDA encourages you the open public hearing speaker at the beginning of your written or oral statement to advise the committee of any financial relationship that you may have with the sponsor, this product, and if known, its direct competitors. For example, this financial information may include the sponsors, payment of your travel, lodging or other expenses in connection with your participation in this meeting. Likewise, FDA encourages you at the beginning of your statement to advise the committee if you do not have any such financial relationships. If you choose not to address this issue of financial relationships at the beginning of your statement, it will not preclude you from speaking. The FDA and the committee place great importance in the open public hearing process. The insights and comments provided can help the agency and the committee in their consideration of the issues before them. That said, in many instances and for many topics, there will be a variety of opinions. One of our goals for today which for the open public hearing to be conducted in a fair and open way where every participant is listened to carefully and treated with dignity, courtesy and respect. Therefore, please speak only when recognized by the chairperson. Thank you for your cooperation. Speaker number one, your audio is connected now. Will speaker number one begin and introduce yourself. Please state your name and any organization you are representing for the record.

This is Gus Alba. Well, thank you very much for the opportunity. This is Dr. Gus Alba speaking. I am currently the Medical Director of ATP Clinical Research in Costa Mesa, California, and the Medical Director for the senior brain health program at Hoag Hospital in Newport Beach, as well as being an associate professor at the University of California in Riverside for the Department of Neuroscience. I'm speaking on my own behalf. And obviously, the testimony that I am giving is practician clinician, but I also need to let you know that I was one of the investigators in the HARMONY trial that you are reviewing right now. I sit on the scientific advisory group for ACADIA Pharmaceuticals, and I lecture extensively nationally and internationally, and obviously have had support from all of the major companies that are out there, including Acadia. But the reason for wanting to share some thoughts with you right now is that there's a serious unmet need of patients that suffer with Alzheimer's disease and then subsequently the psychosis that comes about with them. Unfortunately, this neurodegenerative condition, just like Parkinson's thesis, which you've had an opportunity of reviewing where we oftentimes see behavioral dysregulation and psychotic symptoms flare in individuals between 1/3, up to 40% of individuals thus experiencing it. And we clearly note that there's a serious ripple that affects not just the patient, but also their loved ones or family members. So this is something that we see on a daily basis. On neuropsychiatrists and my patient population is such that I see quite a few patients with both Alzheimer's as well as Parkinson's disease, and oftentimes, these conditions lead to dementia. I obviously also treat other dementias, including frontotemporal dementia, Lewy body and so on, and thus, my interest in having served as an investigator in HARMONY trial. At the present time, we have sense of urgency in that there are no approved agents to address the psychosis associated with dementia. And unfortunately, patients oftentimes get prescribed off-label antipsychotics without a proven positive benefit risk. We note that cognition, motor function and increased morbidity and mortality are clearly documented based on multiple studies that have been done in this particular arena. And as a consequence of that, we need something that's been proven and safe for our patients. The important thing right now is that there's a serious unmet need the current medication that you are reviewing is obviously indicated for Parkinson's see psychosis, but as has been noted by individuals working for the FDA, the overall mortality in patients that suffer with dementia, including Parkinson's, when treated with an agent like pimavanserin versus an atypical antipsychotic. And in most cases, the most common atypical antipsychotic that people reach for is quetiapine. We certainly note higher overall morbidity and mortality in individuals being prescribed medicines that do not have an FDA approval right now. So obviously, the reason that I wanted to chime in is that, again, I see this on a daily basis. There's a serious important unmet medical need. We certainly know that the overall risk benefit ratio for patients is something that is important to consider. We have the fortune of having committees like yours that can review data and then take a look at potentially helping us. We obviously need guidance, and we obviously need individuals to peruse all of the information and as a consequence of that, garner potential aid for many of the individuals that suffer with this illness. When someone can't trust their family members when they're thinking that their spouse is.

Sorry to interrupt. We have to move on to the next speaker.

I apologize. Well, I thank you kindly for your consideration of my thoughts. Thank you.

Speaker #2, your audio is connected now. Will speaker #2 begin and introduce yourself. Please state your name and any organization you're representing for the record.

My name is Gary Small, and I've spent most of my career studying and caring for patients suffering from Alzheimer's disease and their families. Both my mother and mother-in-law are among the approximately 6 million Americans living with the disease. My geriatric psychiatry practice at UCLA for 3 decades and Hackensack Meridian Health the past 2 years has focused on patients with cognitive impairment. I've served as an adviser and speaker for ACADIA in the past, but today, I speak on my own behalf. Most people think of Alzheimer's disease as a cognitive problem, but some of the scariest symptoms for patients and caregivers of their psychotic symptoms that afflict 30% of patients with the disease. These symptoms may worsen insomnia, confusion and agitation and signal greater risk for nursing home placement and mortality. Most of the family caregivers in my practice work outside the home all day and return home to care for their loved one. These 2 full-time jobs often lead to burnout and depression. Imagine a daughter's frustration when her mother accuses her of stealing her wallet, when really her mother's wallet is simply out of sight. Despite my best efforts to explain that such paranoid thoughts shouldn't be taken personally, caregivers still feel hurt when the person they love and care for lashes out at them. The burden of caregiving is intense, and we need to do a better job supporting caregivers. We also need safe and effective therapies to manage the symptoms in caregiver burden. Currently, there is no approved treatment for Alzheimer's related psychosis and clinicians often prescribe off-label antipsychotics with limited efficacy. Such off-label use increases risk for further cognitive decline, infection and even death. There is an urgent need for approved therapies to treat the psychosis related to Alzheimer's disease. Patients, families and caregivers need help in recognizing the onset of symptoms that are psychosis, so they can better address them sooner rather than later. We also need more institutional support for caregivers, including affordable community resources, personalized medical care, education and advocacy. I remember how painful it was for my sisters and me to observe my mother once a brilliant and vital force in our lives as her mental abilities and engaging personality gradually slipped away from us. That emotional anguish was almost unbearable when she then started accusing us of stealing her clothing and jewelry as we tried our best to help. This disease impacts the entire family, and we've got to do a better job in providing support. Thank you.

Thank you. Speaker #3, your audio is connected now. Will speaker #3 begin and introduce yourself. Please state your name and any organization you're representing for the record.

Good afternoon, and thank you for allowing me to speak today. My name is Chad Worz, and I'm Chief Executive of ASCP, the American Society of Consultant Pharmacists. ASCP represents thousands of pharmacist members managing drug therapies and improving the quality of life of geriatric patients and others in various settings, long-term care facilities and home and community-based care. Every day, pharmacists like me and members of ASCP are in communities helping people live better lives by effectively managing medications. This experience has led me to speak today about pimavanserin. At present, this medication is approved for Parkinson's disease psychosis. Since its approval for this indication, it has proven to be an effective and reliable tool for many clinicians and family caregivers. I have witnessed pimavanserin improving the quality of life of patients with Parkinson's disease psychosis. It quells harmful hallucinations and delusions that can manifest in advancing Parkinson's. I can recount stories of improvement that lessen the intensity, frequency and sometimes eliminated those hallucinations and delusions. One such patient I helped to manage was seeing children outside her window who seemed to be in danger. The anxiety and agitation associated with the hallucination was significant, impacting everything from that person's eating and social habits to their behavioral management. Pimavanserin was able to eliminate those hallucinations and delusions from daily occurrences to monthly occurrences in a short 2-month time span. The patient's use of supportive medications for anxiety and agitation were able to be eliminated. Her eating habits improved and her participation in social activities returned. Those kinds of real-world outcomes are common in patients treated with pimavanserin with PDP, and represent an opportunity in people with hallucinations and delusions and other conditions, specifically dementias. Based on the evidence available, pimavanserin shows effectiveness and reliability for hallucinations and in Alzheimer's disease. Adding this new indication would add another tool to providers working to support patients living with Alzheimer's disease and its associated neuropsychiatric symptoms like psychosis. At present, there are no tools in this toolbox, and providers are left to select between inaction and using other medications off-label and against an existing black box warning. The safety of pimavanserin and the evidence of its utility in patients with dementia make it a safe and potentially effective option in a devastating condition, which has no safe options. An approval would bring hope to millions of patients, family members and health care professionals struggling with this terrible disease. We know that nearly half of families who turn to nursing homes do so because of their loved ones behaviors. In many cases, a direct result of their psychosis. The ability of providers and families to try this medication could allow thousands of patients to stay home longer and age in place. As America ages, the ability of patients to remain in their homes and communities is critical. Geriatrics like pediatrics is a sensitive and vulnerable population. It is common and crucial that we ensure access to safe and potentially beneficial treatments where often no other safe or effective options exist. I ask the committee to allow clinicians to practice good medicine and recommend approval, and put potentially powerful and already proven tool in the hands of providers for patients, families and caregivers. Thank you again for your time and attention.

Thank you. Speaker #4, your audio is connected now. Will speaker #4 begin and introduce yourself. Please state your name and any organization you're representing for the record.

Hi, everyone. I'm Sue Peschin and I serve as President and CEO of the Alliance for aging Research. The Alliance receives funding from the sponsor for non-branded health education and advocacy on neuropsychiatric symptoms of dementia. But today, I'm here as a great granddaughter who loved her bubbe. I was lucky to know my great grandmother until I was 13. We spent countless hours together at the Riverview Senior Apartments in Pittsburgh. We played cards, took walks and visited people at the nursing home up the path from her building at the Jewish home for the aged. When I was 11, my mom and I started noticing how bubbe would forget to turn the stove off or leave the water running. She slowly lost the cadence in her step and her quick wit. When my mom made the decision that bubbe needed nursing home care, it was really hard, and I think the weight of those decisions are often not recognized. It helped that we knew many of the residents and staff there. We were allowed to sometimes help with bubbe's bathing and making sure her hair was properly done. For a few months, bubbe would occasionally mention that she saw Hitler sneaking around the building. When constant coverage of Princess Diana's wedding was on TV, bubbe started to believe I was Princess Diana. She would kiss my hand and ask me to promise to keep kosher in the castle. The staff taught my mom and me to go with bubbe wherever she went in their mind. They knew validation before it was a recognized thing to do. We used distraction or told Hitler left, and that seemed to calm her. But after many months, her hallucinations and delusions came more intensely and more often and they were harder to redirect. She became very scared to the point of not wanting to leave a room. When I listen to somebody speak about psychosis as something that only needs to be managed with behavioral techniques, I wonder to myself, has that person ever seen someone they care about thrash around, screaming in abject fear to the point of soiling themselves and crying uncontrollably. Have they ever seen it happen multiple times or even more than once in a given day? If not, I would ask them to think about what that might be like for the person experiencing it and for the people around that person trying their best to help. I recently saw a slide presentation against antipsychotic use that included a picture of a crying toddler. The presenter framed Alzheimer's psychosis as if it were a developmental issue that just needed proper prompting to fix. In truth, my bubbe would have been badly injured had she not been given Haldol back then. Today, there are better therapies being developed for neuropsychiatric symptoms, but we still don't talk about symptoms like psychosis and agitation as openly as we do about memory loss or about the importance of diagnosing and treating them. The impact of this on care for people with Alzheimer's is significant. Moderate to severe neuropsychiatric symptoms diminish quality of life and they hasten death in people with Alzheimer's. Please consider their perspectives of patients and families as you make your important decisions today, and thank you.

Thank you. Speaker #5, your audio is connected now. Will speaker #5 begin and introduce yourself. Please state your name and organization you are representing for the record.

I am John Schall, Chief Executive Officer of Caregiver Action Network. CAN is the nation's leading nonprofit family caregiver organization for the more than 90 million Americans who care for loved ones with chronic conditions and the frailties of old age. ACADIA is one of more than 40 companies that support CAN's nonprofit mission. On behalf of family caregivers, millions of them, I'm speaking in support of NUPLAZID for the proposed treatment of hallucinations and delusions associated with Alzheimer's disease psychosis. If approved, the drug would be the first therapy indicated for this purpose. Alzheimer's takes a huge toll not only on our loved ones, but on us as family caregivers as well. There are 17 million family caregivers of over 6 million loved ones with Alzheimer's in the United States. Family caregivers provided 15 billion hours of unpaid care in 2020, [indiscernible] $257 billion to people living with Alzheimer's. Family caregivers suffer higher levels of depression, face disruptions in their jobs and careers and sacrifice financially and emotionally for their loved ones. A recent survey of family caregivers of loved ones with dementia identified paranoid delusions visual hallucinations and lack of trust as common symptoms. For example, someone's mother might have a false belief that her son or daughter is stealing her personal items and then be verbally and physically aggressive towards them. In fact, more than 3/4 of family caregivers reported paranoid delusions as occurring at least weekly. So hallucinations and delusions are much more common than many people realize. We desperately need an FDA-approved treatment for these symptoms. Right now, if nothing else available, the off-label use of antipsychotics is sometimes prescribed. But antipsychotics often pose safety risks associated with increased mortality and hospital admissions and they can actually worsen cognitive decline. This puts us as family caregivers in a no-win situation, having to make hard choices between doing nothing or treating our loved ones with antipsychotics and maybe creating even greater cognitive loss. Hallucinations and delusions don't just go away, and the problems these symptoms present are very real. Hallucinations and delusions lead to increased risk of hospitalization. They can make our loved ones to take actions that could be harmful to themselves or their families, and they make it difficult for us as family caregivers to care for our loved ones at home. In fact, these challenges are the leading reason why many family caregivers decide that they need to place their love words in a nursing home. To finally have a therapy available, we as family caregivers will be better able to care for our loved ones at home longer, and at last give us hope that these very serious symptoms can be treated. For these reasons, we strongly support the approval of NUPLAZID for hallucinations and delusions associated with Alzheimer's-related psychosis. Thank you.

Thank you. Speaker #6, your audio is connected now. Please introduce yourself and state your name and organization for the record. I guess we will move to speaker #7. Speaker #7, your audio is connected now.

Hi. I'm Dr. Karl Steinberg. I'm a long-term care geriatrician, and I've been a nursing home and hospice medical director in the San Diego area for over 25 years. Most of my patients are nursing home residents and probably just over half of those suffer from dementia, mostly of the Alzheimer's type. I don't have any financial disclosures. I am the immediate past President of AMDA, a national medical specialty society for nursing facility medical directors and other professionals who practice in that setting, and I take my dogs to work with me in the nursing home whenever I can. As a frontline physician attending to many people with Alzheimer's, I want to emphasize just how devastating the psychotic symptoms of this disease can be. Most importantly, to the patients themselves who may be suffering extreme and distressing hallucinations or paranoid delusions, but also to their caregivers, both family and professional and to those around them like other nursing home residents, including their roommates. Alzheimer's psychosis and agitated behaviors related to psychosis are also very common, affecting well over 25% of the population at some point in their disease trajectory. There are well over 1 million nursing home residents in the U.S. and millions more in other congregate care settings. Symptoms can range from crying to screaming to actual physical violence against caregivers. In geriatrics, we try to avoid using medications of all types whenever we can, and especially in Alzheimer's psychosis, since there are no medications approved for its treatment. For Alzheimer's psychosis, we always try to use nonpharmacological interventions first. Unfortunately, though, they are often ineffective. So when these patients continue to experience severe distress, or present a danger to themselves or others because of psychosis. We're left with the off-label use of generally atypical antipsychotics or other medications like anticonvulsants or antidepressants. Antipsychotic use is very highly scrutinized in nursing homes, as it should be, considering the known risks of their use, including cardiovascular, metabolic, cognitive and motor issues. And while they've been historically overutilized because of the scrutiny, many prescribers and facilities today are reluctant to use antipsychotics even when the patient is having severe distress. Of course, even when we do use antipsychotics, they don't always work either. The lack of an FDA-approved medication for Alzheimer's related psychosis is a major gap for us and for our patients. There's an urgent need for us to have something in our armamentarium that we can use to alleviate the extreme, severe and sometimes enduring distress that these unfortunate patients and those around them suffer without any understanding of what's going on, terrified and acting out in ways that would no doubt mortify them if their previous intact selves could see them in their current state. I very much appreciate your attention and the time today, and I ask that you please consider the severe unmet need these patients have, and don't let the good be the enemy of the perfect. I urge you to help us on the front lines to help this vulnerable population we serve in nursing homes, dementia units and private homes across the country by making a medication approved and available for them. And to continue the research to find more pharmaceuticals that can make a difference in this large and growing unfortunate group of patients. Thank you so much.

Thank you. Speaker #8, your audio is connected now. Please introduce yourself.

Good afternoon. Thank you for allowing me to speak today. My name is Aaron Ritter. I'm currently a cognitive disorder specialist at the Cleveland Clinic Lou Ruvo Center for Brain Health in Las Vegas, Nevada. My practice is entirely focused on the care of patients with neurodegenerative disorders such as Alzheimer's disease, Parkinson's disease, Lewy body disease or frontotemporal dementia. A particular interest is treating the neuropsychiatric symptoms that emerge in dementia, and I have over 30 publications in my 6 years of practice and have received over more than $2 million in NIH funding for various research projects. Today, I will be speaking on behalf of the patients I treat. I have participated in clinical trials sponsored by ACADIA but have never received any direct salary support or financial compensation from Acadia or any of its competitors. Simply put, the behavioral manifestations that accompany Alzheimer's disease and other related dementias are devastating and often have a greater impact than the cognitive symptoms. In fact, many patients may not remember that they don't remember or aren't bothered that they cannot remember the data what they ate for breakfast. But on the other hand, patients and the families are acutely aware and frequently tormented by the belief that a spouse is cheating, a son or daughter is emptying money from a bank account, a phantom border is hiding in the shadows or the appearance of the variable cast of characters emerges from the shadows each night to watch over them as they sleep. Research evidence is very clear that the behavioral manifestations of Alzheimer's and related dementias, including psychosis, are the primary determinants of institutionalization and the #1 driver of caregiver burden. As I'm sure you're well aware of Black Box Warning accompanies all of the known medications that may provide relief from psychosis and AD, and most expert commentary rightly so argued against the use of antipsychotics in the elderly. These recommendations, however, fail to acknowledge the situation in the clinic when you're presented with a family and patients in desperate need of relief from the torment of AD-related psychosis. Clinicians such as myself are left with facing the decision of one, treating those dreadful and terrible symptoms using dopamine blocking antipsychotic agents, which by all accounts slowly kill patients over time; or two, offer nothing, which I believe most practitioners do. Offering no medication for ADP, however, have unintended consequences and leaves patients sick and untreated. This is precisely why many of our inpatient psychiatric wards and emergency rooms are filled with patients with dementia. I am in a unique position of having extensive experience of using pimavanserin in both clinically and in the Phase IV study for patients with Parkinson's-related psychosis. I'm also in a unique position of not having it available for my patients with Alzheimer's and Lewy body disease. The bottom line is that after 4 years of experience with pimavanserin, I believe that to be an effective -- an important and effective treatment in most cases. I would urge the committee to consider providing some weapons in our armamentarium that is currently empty. Thank you very much.

Thank you. Speaker #9, your audio is connected now.

Good afternoon. I am Dr. Nina Zeldes, a Senior Fellow at National Center for Health Research. We analyze scientific data to provide objective health information to patients, health professionals and policymakers. We do not expect funding from drug companies so I have no conflicts of interest. As we all know, in 2018, FDA was concerned about "the number of reports of death and other serious adverse events" regarding this drug, which already carries a black box warning that there is "increased mortality in elderly patients with dementia-related psychosis." And in 2021, a study published in Neurology found a statistically significant increase in hospitalization along Parkinson's patients taking this drug compared to nonusers. To outweigh such serious safety risks, the benefits of this drug would need to be substantial. There is no such evidence. Moreover, the resubmission of a previously rejected application for a broader indication relies on the same 2 studies, which FDA previously criticized. They described Study 019 as "not an adequate[indiscernible] controlled study" and noted that there are several study design concerns and protocol deviations. And although NUPLAZID showed a statistically significant improvement compared to placebo, it only translated to a treatment difference of less than 2 points on a 24-point scale. Is that a clinically meaningful improvement for patients, especially since there is no evidence that this timing improvement lasts more than a few days or weeks? The validity is questionable since statistical significance was not reached for the secondary end point. FDA Study 045 as not "powered to determine an effect in the included dementia subgroups." The results for AD patients are not statistically significant. We agree with the FDA that the proposed post-hoc analysis for this subgroup are "very challenging to interpret." For example, there is no scientific reason for the sponsor to use the open-label baseline score instead of the double-blind baseline score when testing the treatment effect on relax in a double-blind period. Lack of diversity is a serious problem. For example, Study 019, only 3 patients were black and only 17 were men in the treatment group. This is not enough to draw any conclusions about either group. And together, these 2 groups comprise close to half of Alzheimer's patients. If the sponsor had made a serious effort to recruit men and more nonwhite patients, they could have done so. My final point is that AD drugs are taken for years. The 12-week Study 019 will not provide adequate evidence of long-term benefit or safety. To determine the benefits outweigh the risks, we need longer placebo-controlled studies. In conclusion, I respectfully urge you to consider whether the evidence of a possible small benefit is clinically meaningful? And if so, does it outweigh the known serious safety risk of Nuplazid. Thank you for your time.

Thank you. Speaker #10, your audio is connected now.

Hello. I am Dr. Leigh Callahan, a Professor of Medicine at the University of North Carolina Chapel Hill. I have no financial relationships with the sponsor, and I'm here today representing myself. I am here because I recently lost my husband, Dr. John Winfield, to Alzheimer's disease. John was a nationally recognized physician scientist. I watch my brilliant husband decline from this devastating disease over 10 years. There are many terrible aspects to Alzheimer's disease with the worst symptom that John experienced with psychosis, including hallucinations, delusions and paranoia. These symptoms were not only scary and heart wrenching for me, but they were absolutely terrifying for John. Let me give you a few examples. John would often think we were living in a different city, but in our same home. As the disease progressed, his suspicions and paranoia grew. Just this past December, when one of our long-time caregivers was taking down the Christmas tree and putting the ornaments away, John entered the room and became enraged, convinced Glenn was stealing the dead tree and taking our family heirlooms. He remains highly agitated and the task had to stop. A far more disturbing event happened a few years ago when John had his first real psychotic break. I heard crashing sound on our screen porch. I found John surveying a room of wreckage, tables were broken and glass shattered. I am a slight woman, I could not intervene physically, but had to convince him there were no aliens and to join me inside. It was powerful, harmfully disturbing and crushing to see John come back to reality, survey the same and ask me in disbelief. "Are you telling me that I did this?" Following this event, John's treating physicians suggested Seroquel, and I felt strongly that this was not an acceptable choice. This would have blunted John's ability to function in general, whereas this event was transitory. We need something very different for this disease in its dimension of psychosis. The drug you are considering has been approved by the FDA for use in Parkinson's, so patients and caregivers like me can rely on evidence of its safety profile in an elderly population. If you find that [indiscernible] is effective in treating Alzheimer's related psychosis, this will have the potential of addressing a very high unmet need. My primary goal was always for John feel safe. It broke my heart that these hallucinations and delusions costs in such distress and even care. As you consider your task and why the benefits and risks of this potential therapy, please keep the dimensions of psychosis in the context of this unique disease and the experience of the patient and caregiver at the forefront of your consideration. Thank you.

Thank you. Speaker #11, your audio connected now.

Thank you. Hello. My name is Agustin Artiles. I have been a research manager with Premier Clinical Research since 2012. We are located in Miami, Florida with the majority of our patients and families being of Hispanic origin. I'm here to share my experience working with patients and their caregivers in the HARMONY trial with pimavanserin. In my role, I have heard many stories from caregivers where at the beginning of the study expressed feeling scared and worried, not knowing how they will continue to care for their loved one as they became disoriented from the illness and progressed and hallucinations. Some expressed already feeling burned out and desperate either due to the behaviors or the constant supervision needed for the patient. And there are many stories I've heard about just how much this treatment has changed patients and families lives. One story I will share today is that the patient and his caregiver, his wife, because I think we'll help you understand why this treatment is needed, especially in the Latin community we serve, and just how much of a difference it can make in the patient's life and in the lives of their families because ADP impacts everyone. We saw a gentleman in his late 60s in a HARMONY trial with wife as the only caregiver. Her husband had lost interest in family hobbies, social interactions, a complete departure from the man he once was. He needed constant supervision. His wife experienced her own health issues due to the burden of having to constantly provide care for her husband. She worried what would happen next if he were to fall ill. One of the recurring worries from caregivers is that in the Latin culture, it is not well accepted to place a family member in a long-term care facility. This forces caregivers and families to make many changes and sacrifices in their lives with lasting impact to support the care of the beloved family members at home. When this gentleman entered the trial, he was unable to write a sentence or draw a simple figure. It was a challenge to have them communicate or engage in a brief conversation. His wife worried about episodes that might occur where she wouldn't be able to manage things on her own. Since taking pimavanserin, this man is now engaged, listening to music, watching TV, socializing, caring for her dog, talking appropriately and coherently with his children and friends. His wife has expressed feeling extremely grateful for "giving me my life back and giving me my husband back." These kinds of findings can be seen in many other patients in the HARMONY trial with symptoms declining enough to allow patients to, in most cases, regain interest in their surroundings, family, hobbies and social life and function independently or semi independently. I ask you to remember what I've shared as a treatment like pimavanserin will help families, and the Latin community care for their loved ones at home and honor their culture while sustaining their own well-being. Thank you for your time and consideration.

Thank you. Speaking #12, your audio is connected now.

Thank you very much. This is Dr. George Grossberg. I'm an academic geriatric psychiatrist, and I've spent my whole career at St. Louis University as Director of the Division of Geriatric Psychiatry. I have over 25 years of clinical experience in dealing with Alzheimer's patients as well as their family care partners. I'm actually speaking to you this afternoon from one of our teaching nursing homes. And one of the new patients that we've been asked to see is a lovely 83-year-old woman with Alzheimer's disease in the kind of middle to later stages, who was also accompanied by her daughter at the bedside. Her daughter is increasingly distressed and anxious because her mom is now starting to become accusatory toward her as well as towards the staff to the point where her mom believes that the staff is trying to harm her, maybe kill her. She's refusing to take her medication. She's also maybe moving toward refusing to eat, and her daughter is obviously very, very concerned. These delusions or paranoid-type symptoms, psychotic symptoms are not rare, as you have heard, in patients with Alzheimer's disease. They significantly impact the quality of life of the patient, the family, the professional caregivers in this kind of scenario. Unfortunately, the current antipsychotic medications sometimes do or do not work, but they come with a lot of baggage with a lot of side effects, particularly for patients in their 80s and 90s, as we often see with Alzheimer's disease, whether Parkinsonian side effects, sedation, orthostasis, so on and so forth, even further impairing cognition. So there's a great need for a safe and effective treatment that can really improve the quality of lives of patients and their care partners, whether it's family or professional caregivers. I'm hoping that with the development of pimavanserin, we're going to be able to fill this significantly needed void. Thank you all for listening, and thank you for the work that you're doing.

Thank you. Speaker #13, your audio is connected now.

My name is Meryl Comer. I'm a co-founder and Board member of UsAgainst Alzheimer's. My written statement is abbreviated here to respect the time limit. I have no conflicts of interest. For more than 2 decades, I cared for my husband and my mother with Alzheimer's, both of whom exhibited a range of psychosis that put them in harm's way and complicated their care. My 57-year-old husband, a respected physician and researcher at NIH, was misdiagnosed for 4 years with everything from depression to pernicious anemia and even mad cow's disease. While we were privately held captive to his paranoia, hallucinations and delusions. The private advice to me from his attending, you may want to get out while you can. His other advice, call 911 if he gets to danger. Several months later, my husband was admitted to John Hopkins for evaluation. For the next 2.5 months, he was confined on the locked board where every available antipsychotic was tried, slowly titrated and then discarded. My husband's final diagnosis read Alzheimer's disease with a behavior disorder, discharged to me with descriptions that included 16 Depakote, an antiseizure medication and 4 Ativan day. And there was nothing left to try. The damage had been done. No facility would take him. I brought him home and slowly weaned him off all the medication that in turn put me in harm's way. He passed 2 years ago, 24 years later. My other experience is the garden variety psychosis suffered by my 80-year-old mother. During her early Alzheimer's paranoia, she was insistent, she was being spied on by neighbors and that her personal items were being stolen, though she let no one in her house. She would scream out the window of the car to strangers to rescue her, and even called 911 to report she was being held against her will. The doctor's prescription for Seroquel never filled because we feared the long list of potential side effects more. The reality is that whatever the FDA approves and doctors prescribe, we are left to manage the consequences. The real numbers in societal impact of psychosis and dementia are masked. As a family caregiver, we keep the secret about these behaviors even from our adult children to support and protect the loved ones day to day. An FDA-approved drug if deemed effective by this panel in treating Alzheimer's related psychosis, will help us support them at the intersection where the scaffolding in their identity begins to fall apart due to the ravages of the [indiscernible] neurodegenerative disease with no cure. Thank you for your consideration.

Thank you. Speaker #14, your audio is connected now.

Good afternoon. I am Jed Levine, President Emeritus at CaringKind, The Heart of Alzheimer's Caregiving, formerly known as the Alzheimer's Association's New York City Chapter. CaringKind is the premier resource for all things related to dementia care in New York City, and I should say that ACADIA is a financial supporter of CaringKind. We provide guidance and support for individuals diagnosed with Alzheimer's and related disorders, and most importantly, those who care for them. I have over 40 years of experience with this population. Caring for a relative who is now experiencing progressive cognitive decline is unlike any other caregiving. Unless you've lived it, done it day in and day out, you don't really know what it's like, how exhausting and demanding it is. The challenges evolve as the disease progresses from the early stage where the individual is still interacting in many ways as they did to the middle stage where the confusion, memory loss, anxiety, frustration, psychotic symptoms and functional disabilities become more pronounced, to the end stage where the individual lost language and the ability to walk, sit up and are dependent on someone else for all personal care. Caregivers report that the neurobehavioral symptoms, agitatedly asking the same question, aggression during personal care, resistance to bathing or washing hair, anxiety, pacing, sleep disruption, apathy are particularly distressing. Significantly adding to the stress are the psychotic features such as hallucinations and paranoid delusions. The hallucinations, almost always visual, might not be upsetting, but often they are. Delusions too can be extremely troubling for the individual. I recall one member in our Early Stage Center who had an extremely fearful reaction to his reflection in a mirror, believing there was a threatening stranger in the home, or the former church organist who had the persistent delusion that she had to play for a service, no amount of distraction or reassurance would calm her. Common delusions include that a family member is an imposter, that the home is not their home, that a spouse is cheating or has stolen money or property. These symptoms are a frequent feature of dementia with some studies showing that they exist in 15% to 75% of patients with delusions happening in up to 30% of patients. And I have heard from family caregivers who we're fearful for their own safety when their person with Alzheimer's was experiencing delusions, threatening to harm them or at times striking out at them. We teach non-pharmacological approaches that are useful but some individuals have persistent and resistant psychotic features that are distressing not only for that family caregiver or staff member, but for the individual themselves. The experience of psychotic symptoms can be frightening and extremely upsetting for the person having them. Current antipsychotic medications, as you've heard, are often ineffective, contraindicated for use with people with dementia and can result in overly sedating the patient and have concerning adverse effects. Having a new drug to address the psychotic features will be an extremely helpful adjunct to the repertoire of non-pharmacological approaches we use now and can greatly improve quality of life for the diagnosed individual as well as those caring for him. The lack of an FDA-approved antipsychotic for Alzheimer's psychosis is a substantial unmet clinical need. On behalf of the millions of individuals with Alzheimer's and their caregivers, I thank you for your consideration of this supplemental new drug application for the treatment of Alzheimer's psychosis. Thank you.

Thank you. Speaker #15, you line is connected now.

Yes. Hello, everyone. Thank you for allowing me to speak. My name is Howard Kirshner, and I am a Professor of Neurology at Vanderbilt University Medical Center in Nashville. I've been the Vice Chair of the Department and the Head of the Behavioral and Cognitive Neurology division. I also am speaking on behalf of the Clinical Neurology Society of America, which is working on a white paper on the issue of psychosis in Alzheimer's disease. I would just say from my own experience of 44 years of practice as an attending neurologist that Alzheimer's is one of the most distressing diseases we deal with, and the psychotic features, the delusions and hallucinations are the most troubling. They are the single leading cause of patients being institutionalized, which is distressing for patients and families. And then there's also a major problem in the extended care facilities because patients are often oversedated. There is a tremendous need for a good treatment for Alzheimer's-related psychosis. There is no FDA-approved treatment as of now, as you all know, and frequently off-label use of either benzodiazepines or antipsychotic drugs, it's definitely either one or is definitely harmful to the patient. So when you consider that there are no alternatives, I think a new drug is particularly appealing. It may not be perfect, it may not have been tested for long enough courses, but it appears relatively safe, at least in short-term use, much more so than the existing medications that are tried. So I would urge at least an interim approval of the drug pimavanserin and hope for other future treatments. I want to thank everyone on the committee for your attention to this very important issue and for the service you do every day. Thank you.

Thank you. Speaker #16, your audio is connected now.

Good afternoon, everyone. I am [ Nadina Arce ] in [indiscernible] Medical Research Center, specifically [ Dr. Navarro ]. We are in Miami, Florida. I am the caregiver of my father, Raul Arce, who has been suffering from Alzheimer's psychosis since September 2018 when he was officially diagnosed after countless tests and medications. He was only 69-year old. I begin by telling you that I'm a single mother of 2 beautiful children, a 13-year old and 3-year old. To this day, we share our life with my father. [indiscernible] where he worked and still is today the most loving, affectionate and hardworking man. The example to follow of our family and the most respectful of being that exists. The person was always present for anything we needed in good and bad. In May 2015, my father began to show signs of mental decline, signs that I didn't recognize, I think, because I didn't want to recognize reality. I realized that something was very wrong the day he called me because he didn't remember how to get home. The next day, I took him to his doctor. After the appointment, everything began to change in our life and not for the good. Even though a best doctor put him on medications that helped him later, he suffered from the side effects of these medications, irritable, depressed, anxious, having trouble sleeping and a million other things more. He also could not be left at home alone because of his illness. So I decided to start to work from home to take care of him. My income decreased, and my life situation also changed. I got divorced not because of my father's situation, but I think that influenced our decision. I remain strong for my family, but a broken soul from seeing such an amazing man dying day by day to such a [indiscernible] disease. Thank God in early June 2021 we were told about the pimavanserin trial. Once my father was accepted into the trial, we saw changes. This medication helped us manage my father's care and kept him at home with us where he belongs, at the center of our family, no matter how ill he might be. I ask you to think about my story and my father's story as you make your decision today. There are so many other families like mine out there who want to keep the people they love at home, and give them the care they deserve and the life they deserve. Pimavanserin will have a great impact on the person who suffered Alzheimer's-disease psychosis and their families. Thank you so much. Please make it available to us.

Thank you. Speaker #17. Your audio is connected now.

My name is Anita Louise Royal. I have no financial relationships to disclose. For 21 years, I served at the Pima County Public Fiduciary as a lawyer, providing guardian services to several hundreds of vulnerable adults, many of whom suffered from neurocognitive disorders. Asked that our goal was to preserve their self-independence, their self-determination while ensuring their safety. Many stuffed from Alzheimer's dementia with psychotic features, including delusions and hallucinations, requiring often them to leave their own homes and be placed in residential extended-care facilities. However, I'm not here to talk about my prior experience as a lawyer, I'm here to talk about the fact that I am one of 19 million in this country, who serve as a full-time caretaker for my beloved mother who was diagnosed with dementia almost 20 years ago. I have served as her caretaker for 12 years. During that period of time, she had begun experiencing -- actually, in the last 5 or 6 years, she's begun experiencing auditory and less often visual hallucinations. We've tried medications, we started with Depakote, went to Lexapro and Seroquel to deal with some of her very disruptive behaviors. Nothing has worked. Thankfully, we are in the United -- I'm sorry, University of Colorado's Senior Center, where we are now getting sufficiently appropriate care for her dementia and her behavior. However, she continues to suffer from auditory hallucinations in which she hears children crying and often loud music. When this happens, she becomes so distressed and so upset that she often tries to get out of bed despite her limited mobility, which has resulted in her falling and having injuries. I need as a caretaker, some medication to help her from being so upset and distraught when she hears the babies crying. She thinks she has a duty to in fact help them, but she does not. I need a medication as so many other Alzheimer's and dementia caretakers in this country do to help our loved ones, to deal with these symptoms and so that they can have more palatable loving lives. And I just want to also thank all the rest of the speakers before me, I learned so much from them. And I just want to thank you all for the work that you're doing on behalf of this population.

Thank you. Speaker #18, your audio is connected now.

Good afternoon. My name is Jani Moreira. I am a caregiver of my mother in law, [indiscernible] Medical Center. I am blessed to have a beautiful Cuban family here in Miami that really enjoys spending time together and taking a good Cuban coffee in the morning, especially my mother-in-law, [indiscernible]. This is her story. In 2014, Amada had starting to put salt instead of sugar in the coffee or forget the water in the coffee maker due to depression and other psychotic symptoms. No more coffee for us. This is a small and simple thing, I know. But here, I told you about the story of the Alzheimer's-disease psychosis and how to [indiscernible] people. Due to that. I had to stop working to take care of my mother-in-law full time at home, something that causes problems in our family. The medication she was prescribed helped a little, but she continued depressed and sicker. In August 2021, after no improvement, her doctor told us about the pimavanserin trial. Once my mother-in-law was accepted into the trial, we saw big change in Amara. The loving mother and dedicated mother came back to us. She pays more attention to the life today, and she gained interest in her hobbies and activities. In fact, with some assistance of course, she has retained to make even coffee in the morning. Our family tradition is back, thanks to that medication. So please same story into consideration when you are making your decision today. You can help many families like me who need this important treatment for family members. Thank you very much for your attention.

Thank you. Speaker #19, your audio is connected now.

This is Dr. Daniel Claassen, and I am the Chief of Behavioral Neurology at Vanderbilt University Medical Center and Professor of Neurology here at Vanderbilt. And I've been able to listen to the other 18 calls, and it's been quite a powerful story to hear some of the caregiver and burden and some of the patient stories. I just wanted to give you some perspective as a physician, what things we deal with. I don't speak for ACADIA. We do have some research grants from them in people in my division, but this is from my own accord. As a clinician, especially one that takes care of those with neurodegenerative disorders, I just want to convey to you that what we do is really a practice of medicine. And I know that you've probably been spending a lot of time looking at numbers and data. But I really want to give you the perspective that what we do is an art form. And just like an artist would need to have different colors to paint a picture, I think we need to have different colors to treat psychosis. And the colors that we have right now, especially when you think about Alzheimer's-disease psychosis and other related psychoses, they're dark colors. They have a lot of side effects, Parkinsonian symptoms, weight gain, sedation, [indiscernible], just to name a couple of them. And I think our current practice is limited based on the labeling of pimavanserin. And I think that -- I just want to encourage you as from a clinician point of view, we really need new colors in our color palette. I think pimavanserin could be an important color as we practice our art with -- as we partner with patients and families and try to find remedies for these terrible symptoms. So I know you have a difficult decision, and you have to make a decision based on numbers and statistics and data. But perhaps if you could consider how this decision really has profound implications for how me as a doctor, as my colleagues as physicians, lot of colleagues as nurse practitioners, how we take care of our patients, pimavanserin really does have a lot of opportunities for us to practice better medicine. Especially how we -- the lack of a titration schedule, the side effect profile, the clinical benefits. So I ask you as a clinician to give us a chance to practice with these new colors. And I thank you for considering this, and I thank you for your hard work and your service to make these important decisions.

Thank you. Speaker #20, your audio is connected now.

Hello. Good afternoon. My name is Stephen Chambers. I'm a physical therapist. I lived in Oakhurst, New Jersey. I've been practicing in both New York and New Jersey for the past 20 years. But I'm not speaking today in any sort of professional capacity, I have no financial disclosures or conflicts to disclose. I'm speaking today because I know the emotional pain that the loved ones of those stricken with Alzheimer's disease endure. I support and advocate for the approval of safe medications that minimize delusions, hallucinations, and other symptoms of Alzheimer's-dementia psychosis. My father, Chester Chambers, was diagnosed with Alzheimer's in 2014, but started showing the signs as early as 2012. We lost my father on September 8, 2021, 2 months prior to his 79th birthday. My father, Chester, was drafted into the U.S. Army within 2 months of graduating from college and marrying my mother, as college sweetheart. After 2 years of service during the Vietnam War, he was honorably discharged and began his career as a recruitment manager at a Social Security Administration where he spent the next 36 years in various human-resources roles. He became a lifelong mentor and friend to many of whom who hired him at the agency. He was also a loving and devoted son, husband, father and friend. He was the patriarch of our family and a caregiver to us all right up until he was unable to provide that care for us anymore, and we had to take over caring for him. He was a friend to everyone and beloved in our community. He was truly one of the kindest people you could ever meet. My family has a history of Alzheimer's disease. My father's mother and older brother both preceded him in death due to Alzheimer's-related dementia. Despite this history, it took me and my sister some time before we started comparing notes and realized that he needed increasing amounts and systems to manage his day-to-day tasks. As the disease progressed, he began to experience delusions and hallucinations. For example, he often believed there were parties going on in his basement or actual people visiting, even when it was simply he and his home health aide alone in the house. On another occasion, he was convinced that a food delivery person had stolen his wallet when he provided a tip, although he'd actually hidden the wallet in the cabinet. This incident led to me running out of the house barefoot and chasing down the delivery guy to assure me that there was no wallet or need for money to change hands as I'd placed the order online and paid and tipped in advance. This is just one of the many instances where delusions caused extreme distress to both my father and me and those who cared for him. My sister and I struggled with the fact that our father's delusions were just close enough to being plausible that they had to be checked out regularly, adding to the extreme levels of stress and effort required to ensure his safety and care. And these extreme levels of effort became to be too much, and in 2016, we had to move him to a place where he could receive around-the-clock care, which was a very difficult decision for us as my father was a very independent, strong willed and highly functioning individual until that time. Dad was prescribed Namenda and donepezil with the intent of slowing the progression of this dementia and sertraline as well, as a mood stabilizer. We were told that there was no cure of magic bullet and that these drugs will only potentially slow down the inevitable. We were not given the option of an off-label treatment, and honestly, I think that as the disease progressed, whatever effect these medications had was minimal. Families like mine, who [indiscernible] in our task of caring for a loved one with Alzheimer's or some other form of dementia are desperate for an approved treatment for Alzheimer's dementia and psychosis. I hope and pray that we are not far [indiscernible], a successful way to minimize or eliminate amyloid plaque. However, until that day comes, if there are medications like pimavanserin, that can help minimize the trauma of Alzheimer's-dementia psychosis, then I implore you to approve it for this purpose. Please think of the emotional and physical distress and frankly, trauma that this disease can inflict on those experiencing this kind of psychosis and their families who love and care for them. If treatments like pimavanserin can help families to preserve and salvage the quality time that they have left with their loved ones, then please make it available to patients who are experiencing Alzheimer's-dementia psychosis. I thank you for your time and for allowing me to speak here today.

Thank you Speaker #21, your audio is connected now.

Good afternoon. I'm Adriane Fugh-Berman, a Physician and Professor at Georgetown University Medical Center and Director of PharmedOut, a rational prescribing project at my university. My conflict of interest disclosures that I'm a paid expert witness on behalf of plaintiffs in litigation regarding pharmaceutical marketing practices. PharmedOut opposes ACADIA's application to expand pimavanserin's indications to include Alzheimer's-disease psychosis. Last year, the FDA rejected a broader indication for dementia-related psychosis. That rejection was correct and pimavanserin should be rejected for any subset of dementia-related psychosis. ACADIA has already begun its prelaunch marketing through its MoreThanMemoryLoss and MoreThanCognition websites. It bears noting that these websites focus on dementia-related psychosis despite the black box warning on NUPLAZID that states the drug should not be used for patients with dementia-related psychosis. Tellingly, even the Alzheimer's association, UsAgainstAlzheimer's and the Alliance for Aging Research, all industry-friendly organizations that receive funding from ACADIA couldn't bring themselves to wholeheartedly back their sponsor's drug, perhaps because that drug has the potential to kill their constituents. You've heard the term unmet need numerous times in this session, but unmet need doesn't trump data. There's always an unmet need for a symptom turned into a disease by a drug maker. The symptom is certainly real. Psychotic episodes may accompany many diseases, including depression, bipolar disorder, Huntington's, HIV and malaria. Use of cannabis, alcohol, other recreational drugs and prescription drugs can cause psychotic symptoms, so can hypoglycemia. A tedious reframing of Parkinson's-disease psychosis and now Alzheimer's-disease psychosis as unique diseases benefits ACADIA, but changing symptoms into diseases won't benefit patients. At $4,173 a month, the price of the drug itself is high but with 2.4 million vulnerable patients who could be prescribed pimavanserin outside of the controlled conditions of a trial, only a markedly effective relatively safe drug can address an unmet need. NUPLAZID is not that drug. A 2021 study by [ Hwang ] found both increased risk of hospitalization and mortality with pimavanserin. And although it's been claimed that pimavanserin causes fewer deaths than other antipsychotics, a very recent study by [indiscernible] or Medicare beneficiaries found that it's true only in the first 6 months. After that, the risk is the same. Antipsychotic drugs are already excessively used among vulnerable elders, especially in nursing homes. 1 in 5 nursing home residents are on antipsychotics. As the head of ACADIA said himself, the symptoms of dementia can overlap with symptoms of psychosis, a committee member noted that psychosis can be difficult to separate from agitation and aggression. If any drug is approved for Alzheimer's-related psychosis, the diagnosis as well as the drug will be legitimized. Diagnoses for this questionable condition will skyrocket, hundreds of thousands of elders will be sedated into oblivion and many will die prematurely as the direct effect of the drug. It's bad enough the FDA approved pimavanserin for Parkinson's-disease psychosis. Please don't compound this error by recommending additional approval for an unclear diagnosis in a vulnerable population in whom pimavanserin can only cause harm. Thank you.

We're going to give Speaker #6, whose audio was blocked at that time, there were some technical difficulties, to have an option. Speaker #6, your audio is connected now.

I'm Sidney Wolfe, I'm the founder of the Health Research Group at Public Citizen. I have no financial conflict of interest. Next slide. A study published in 2018, which now is known as Study 019, this is from the original public study with Dr. Ballard, [indiscernible] meeting today said pimavanserin showed efficacy in patients with Alzheimer's disease at 6 weeks. But follow-up at 12 weeks did not show significant advantage over placebo. This is in a published article 4 years ago. The FDA added a few other concerns when they finally were able to get a hold of the documents from this study. The FDA inspectors have concerns about the reliability of Study 019 because of many protocol deviations. You've seen a chart with all of these. They principally involved subjects who did not have clear documentation that psychotic symptoms developed after AD diagnosis that have been established or will receive exclusionary medications at the time of [indiscernible]. As we stated earlier, in addition to that, roughly half of the patients did not get [ added informed consent ]. Next slide. So we now go to slide 45 -- Study 045. And again, in Dr. Terry's and his colleague's paper published in the New England Journal, he stated, "longer and larger trials are required to determine the effects of pimavanserin in dementia-related psychosis. Approximately 15% of the patients in the trial had Parkinson's disease, which may have skewed the results in favor of pimavanserin." The FDA has looked into this more carefully and has data pretty much showing that, whereas that subgroup who had Parkinson's disease had a statistically significant improvement and the data are out there, it's a narrow confidence interval, whereas the people in the AD group had a nonsignificant improvement. So the apparent differential effects of pimavanserin in the PD subgroup relative to the other dementia subgroup was the main reason that FDA filed a complete response action in the first review and a reason that the broad and this is quote from the FDA, the broad dementia-related psychosis indication is no longer being considered. Next slide. In addition, the FDA concluded they would need a much larger sample size to be able to really find robust findings if they exist in the AD group. Because the trial was terminated early at the initial analysis, the conclusion can be based only on the IA results. And that's the conclusion, again, in the briefing documents, the study failed to demonstrate a treatment effect in the AD population. Last slide. So the voting is really on do these two studies support a conclusion that it works for AD. Given the serious flaws in both studies, we would agree with FDA's conclusion that, "the study failed to demonstrate a treatment effect in the AD population." It's not much more than a year ago where the FDA mistakenly approved aducanumab for treating Alzheimer's disease despite the fact that the evidence was as weak or possibly weaker than here. So the idea of the FDA approving a drug that's been studied with a mixture of not only Alzheimer's patients, but patients who had Parkinson's disease and the conclusion of the FDA is that that's why the study overall looks good. I've never been in the 50 years I've been going through FDA advisory committees, I've never seen a situation where someone is asking for a supplementary approval where the [indiscernible] study includes not only the one you are trying to approve it for, Alzheimer's disease, but also Parkinson's disease. This drug should not be approved. Thank you very much.

Thank you. The open public hearing portion of this meeting is now concluded, and we will no longer take comments from the audience. I just wanted to -- the sponsor wanted to respond to Dr. Walter Dunn's question with a slide. So if they can do that very quickly in 2 to 3 minutes, we would really appreciate that. So I'm going to give the sponsors a second to respond to Dr. Dunn's comments.

Thank you, Dr. Narendran for giving us a few minutes to speak to Dr. Dunn's question. Dr. Hendrix, can you please come to the mic?

Thank you, Suzanne Hendrix, Statistical Consultant. In Study 045, among those who did not achieve stable response at both 8 and 12 weeks, and therefore, were not randomized, approximately 20% to 30% of people had early response at 2, 4 or 8 weeks, as shown in this figure, we saw a similar pattern also in the ADP population. The second question, we have confirmed that 100% of the PDD patients that did not qualify for randomization were on dopaminergic therapies, and the non-PDD patients had a low rate of dopaminergic use which is consistent with the randomized patient population. Thanks for the opportunity.

Thank you. Dr. Dunn, do you have anything to add to this?

No, thank you very much.

The committee will now turn its attention to address the task at hand, the careful consideration of data before the committee as well as the public comments. We will proceed with questions to the committee in final discussion. I would like to remind public observers that while this meeting is open for public observation, public attendees may not participate, except at the specific request of the panel. The discussion question #1. Discuss whether the evidence supports the effectiveness of pimavanserin for the treatment of hallucinations and delusions in the Alzheimer's disease psychosis population. In your discussion, comment on the strength, limitations and the extent to which each of the following potential sources of evidence contribute to your overall assessment of effectiveness, study 019, Study 045 and then the prior approval of pimavanserin for the treatment of hallucinations and delusions associated with Parkinson's disease psychosis. Are there any questions about the question from the committee to the agency before we open this up for discussion? Questions about the question. No questions. So I assume it's clear. So I think I would like to call on every committee member to kind of weigh in on their thinking on this question. We would like to get everybody's opinion on this discussion question. Is there anybody who wants to go first? Dr. Thambisetty, we'll start with you.

Thank you, Dr. Narendran. This is Madhav Thambisetty, NIH. I'd like to thank you for the opportunity to go first in this open discussion. In my opinion, Study 019 remains not adequate and not well controlled as assessed by the FDA in the complete response letter with a substantial number of major protocol deviations, 55% in the placebo group and 56% in the pimavanserin group. Most importantly, the separation of drug and placebo groups at [ weekly ], a coincidence with the marked placebo group version. The small treatment effect at this point is not maintained at any other subsequent time points. There is also no support of efficacy from analysis of any of the secondary or exploratory end points. The use of the NPH-NH to measure the primary outcome have limited convergent validity as pointed out by the FDA's analysis. And the FDA's concerns with the scoring and interpretation of group and individual differences within this instruments appears to be well founded. With regards to Study 045, this used a randomized withdrawal design that is associated with several well-known limitations because it collects out frequent responders in the open-label phase and measures the same treatment response in the double-blind phase, which likely overestimates drug versus placebo differences in favor of the drug. The study design also requires an abrupt withdrawal of the drug and this likely results in confounding the effect of drug withdrawal with relapse of psychosis that are further undermining validity in the results from this study results. The primary endpoint results in Study 045 are clearly driven by the PD subgroup. The FDA's analysis clearly shows a strong treatment by subgroup effect with the AD psychosis subgroup showing a lack of treatment signal on the primary outcome as well as virtually all of the secondary and exploratory post hoc analysis done. So in my opinion, Study 045 does not provide any supportive efficacy for pimavanserin in AD psychosis. With regards to the third question about whether the prior approval is relevant here, I would go by the data that we have before us. So rather than look to the prior approval to Study 020, what I would focus on and what I would emphasize on is the actual data analysis that clearly shows a strong treatment by subgroup effect in 045 showing that the two subgroups, PD dementia and AD dementia, in fact, behave very differently in response to this drug. Therefore, I do not think that the prior approval of pimavanserin here is relevant because the data that we have and the analysis of the data that we have clearly indicates otherwise.

Thank you. I have Dr. Cudkowicz next.

Second Study 019, I do think it's a positive -- it's definitely a positive study. It's primary and it's supportive of an effect. It's not a perfect study. I am reassured by the FDA, their audit and via conclusion that deviations were balanced. The complexity of doing a study in nursing home in people with advanced Alzheimer's, I would have been surprised not to see deviations. Not that you never want them, but again, I'm going to go with the FDA's conclusion that it's still a study that could be considered for registration, and that those deviations were balanced. So I'm less concerned about the deviations. That's a secondary since [ a line ], of course, it was a better phase they did, but they're measuring different things that it's not that clear whether this drug will work on. So again, I think it has a short-term effect. It brings faster relief to people with an awful symptom and awful disease, and I think we heard from the community, how important that is. Of course, it would be better if it was sustained over the 12-week period, and that's a concern whether another study or other ways to get us that is still going to be important long term. 045 is a more complicated study. It wasn't really designed to answer the question if this works in AD psychosis because it was broader and it was empowered for that, if there are some trends for it, but it's not conclusive. So I'll put in a little less weight on that than 019 in my thoughts about this. If we believe that the mechanism of hallucinations and delusions is similar in Alzheimer's and Parkinson's and the prior approval of pimavanserin is actually entirely relevant, I think there's no clarity on that. We heard from some of the experts in the field that there are overlapping biologies. And so there's certainly some, whether it's all of it, I don't think the [indiscernible] actually knows. I do think that safety part is important from the prior experience from pimavanserin in Parkinson's disease. And my understanding from the FDA briefing, [indiscernible] that was not something that's up for a discussion or concern, it's really about the efficacy. So I think if something is safe for Parkinson's, it's probably safe for Alzheimer's. So it's really down to whether we think it's the case or not. And again, I do think that Study 019 is persuasive.

Thank you, Dr. Cudkowicz. Dr. Dunn, you're next.

Walter Dunn, UCLA. So in reviewing the data and listening to the discussions today, for myself there's this reoccurring theme of viewing today's issues, either from a broad, all-encompassing approach versus a narrow precision perspective. I mean, each has its merits and drawbacks. But unfortunately, I think many of the issues before the committee today, they've been applied in questionable context. And I'll refer back to this theme as I outlined my opinions on the questions posed by the agency. So for Study 019, from a narrow consideration of the data, I agree that Study 019 is technically a win for the applicant and additional strength in the study was that despite a considerable number of protocol deviations that actually appear to work against the study drug there was still an overall statistical separation for placebo, which suggests that there's a potential for a large effect from active treatment. However, I think the totality of evidence for 019 questions a conclusion of drug efficacy. The pattern of response in the placebo arm is quite concerning as it does suggest a chance effect at week 6 driven by worsening in the placebo arm. While I can appreciate the waxing and waning nature of psychosis in Alzheimer's dementia, I think the fact that we do not see a similar pattern in the active treatment arm suggests that the worsening of symptoms at week 6 in the placebo arm may be an artifact completely unrelated to the disease or treatment. However, as a side note, you actually do see a bump in symptom severity at week 9 for the drug arm. However, I don't have a good explanation as to why the waxing and waning nature of the psychosis would be "delayed" by drug treatment. The lack of signal for all the other secondary outcomes is another concerning observation that places into question drug efficacy. And also to the question of clinical utility, the use of a primary outcome that only captured a narrow slice of symptom presentation, I think, goes against what I think we should be aiming for in drug development, treatments that have an impact on functional outcomes. I agree it's challenging to win on all of your outcomes. However, I think at least a signal on the agitation and aggression domain would have been -- would have made the case more compelling. And as also has already been discussed, the differences in ethnic and racial composition of the U.K. study population compared to the U.S. population, I think, is a limitation of the study design. Obviously, this is something that occurs even within U.S. studies and something that we should endeavor to resolve. Regarding Study 045. So I think the results from Study 045 is a prime example of why we need to be working towards more precise diagnosis and treatments for our neuropsychiatric illnesses. The field clearly appreciates that the future of medicine is about developing precision treatments, which can only occur with precision diagnoses. So while not powered to do so, I strongly believe the results from Study 045 suggests that Parkinson's-disease and Alzheimer's-disease psychosis are different illnesses with different responses to the study drug. Even within a unitary diagnosis such as schizophrenia, those of us in the field clearly believe that multiple underlying pathophysiologies across different patients drive similar clinical presentation, but with different responses to medications. To the question of the dopaminergic drugs driving the high relapse rate, so while that proposal is plausible, I think there's unfortunately no other evidence to support that conclusion. In fact, the differential response in the open-label phase between Parkinson's and Alzheimer's patients suggests otherwise. While there was not a formal comparison, there was a higher numerical response in the Parkinson's patients compared to the Alzheimer's patients. And these were the same patients who are on the dopaminergic drugs. If that explanation were true or partially true, I would have at least expected a numerically lower response rate in the Parkinson's-disease psychosis as their symptoms would have been complicated by the presence of dopamine agents. But in fact, you actually see a better response in those patients. So finally, to the final question about essentially Study 020, so this leads me to the question. So I'll preface my comments by saying that I'm only addressing the question of whether I believe the 2 conditions are closely related because that's obviously essential and necessary if I'm going to give Study 020 consideration in this new indication. So it's ultimately up to the agency to decide what level of evidence is required for approval. But I do not believe the prior approval of pimavanserin in Parkinson's supports efficacy in Alzheimer's because there's limited evidence suggesting psychosis between the 2 conditions is being driven by similar mechanisms or that they respond similarly to pimavanserin. In fact, as I outlined earlier, from my interpretation of Study 045, they were actually probably quite different.

Thank you. Dr. Iyengar, you're next.

Thank you. Actually, my concerns were very well articulated by the earlier speakers. So I'll be a little bit brief. I appreciate the unmet need for the ADP patients and the attempts by ACADIA to marshal evidence from the studies 019 and 045. However, I see just too many problematic issues with the evidence. First, there's the unexplained blip for placebo at week 6 in Study 019. There's a demographic mismatch between the population in England and here. And there's a lack of support from secondary endpoints. And there's also this issue of the treatment by subgroup interaction, which makes taking evidence from Study 020 and trying to support the current application. In short, I think what's really needed is a well-powered study on an appropriate Alzheimer's-disease psychosis sample. Thank you.

Thank you. Dr. Follmann.

Yes. Thanks. So sometimes, I struggle with the decision on an advisory committee but not today. And I think it's really just a simple and unfortunate story. Study 045 stopped early at an interim analysis, which was almost entirely driven by the result in the PDD. And then after the study had stopped, you find this enormous statistical interaction between PDD and AD, suggesting and sort of telling us that it's not appropriate to combine these 2 groups and that we need to look at the evidence individually. And so now you're left with the AD subgroup, which is simply underpowered. And while it might have had a positive numerical effect in this study, the evidence really is not there to support 019. I thought Study 019 by itself was just significant in the ITT. It just met the P0.05 bar, but like the comments of the FDA and sort of a lack of consistency across the results for the secondary endpoints and not maintaining the effect for the second half of the study, didn't really sort of give further support to the story of the [indiscernible] 045 in Study 019. So I didn't find that very helpful. In Study 045, the sponsor did many analyses, some of which showed the similarity of the response of the 3 different subgroups on pimavanserin, but these aren't randomized comparisons. What we really care about is the randomized comparison of the treatment effect for the different subgroups is which is where we see this enormous effect, which allows us or sort of ensures that we should look at the subgroup separately. A lot of the points that I want to make or I agree a lot with what [ Dr. Sam Bozzette ] was saying, in particular, there are a lot of analysis that the sponsor made, but I didn't -- these are sort of, I think, the best analysis that would support the sponsor's case. And I just own them, in aggregate, weak. And then finally, Study 020, I don't think it's really relevant here given I think the Study 045 shows the 2 diseases are not comparable. So I don't think that's a supportive evidence. So I think it's just not a very compelling story. Thank you.

Dr. Fiedorowicz?

Jess Fiedorowicz, University of Ottawa. So my comments are going to be much in line what you've heard already. Hold on. I'm getting some feedback. Right. So Study 019 was initially designed as a Phase II study. It included a questionably validated clinical outcome measure. The measure was significant at the 6-week time point from the protocol, although that was not appropriately publicly registered, and I do think that is an important issue. Differences at that time point were also not consistent with surrounding time points, and they're equally supported by secondary and exploratory endpoints. And there were some concerns about protocol deviations, some of which I think are understandable but there are concerns nonetheless. The consent ones concern me most. The ones that have the diagnosis were in the time line of that were a little less concerning because that is very difficult to tease out in clinical history. Overall, I felt Study 019 supported the conclusions that were published in that Ballard Lancet paper, where it was published, where they said, "the findings from the study suggest potential efficacy and acceptable tolerability of pimavanserin for psychosis in Alzheimer's disease, encouraging the development of a Phase III clinical trial program." And so while Study 19 indeed supports the design of such a study and program, it falls short in my mind of the FDA definition of an adequate and well-controlled clinical study. Moving on to Study 045, while the overall study was positive, it was indeed strongly influenced by the Parkinson's subgroup, where there already is an indication given. There is evidence of differential results by diagnostic group, and it's not clear to me whether this is induced by concurrent use of dopaminergic agents is a relevant question. So when [ Dr. Follmann ] asked the question of why does it matter, it's not clear to me that it matters. We know that there's differential response to those groups and presuming those differences also apply to other study designs where those with Parkinson's are going to be more likely on this medicine. The subgroup analysis to Alzheimer's was not statistically significant. And ultimately, when you look out at the raw data, it boils down to 14 relapsing with placebo versus 9 with pimavanserin for the Alzheimer's disease and FTD spectrum disorders. On the prior approval, I think the overlapping dementia pathophysiology provides biological plausibility for consideration of that, but the results of Study 045 showing differential response ultimately question [indiscernible] have noted.

Dr. Stander?

I appreciate the opportunity to participate in this panel. I need to make sure people understand that my expertise or experiences or to entirely as a clinician over the years. And I really appreciate a lot of the insight that has been given by the other members of the panel who have greater expertise in study, design, analysis and statistical interpretation. I am currently at Phoenix, VA and faculty at the University of Arizona. I won't repeat a lot of the details from the other speakers, but I would agree that my concerns of Study 019 or some of the conclusions based relative short duration, the makeup of the population and a relatively small statistical benefit. And the -- it was [indiscernible] was on Study 045, who was acknowledged that it was empowered to distinguish subgroup [indiscernible]. And I don't think that the efficacy prior approval [indiscernible]. Listening to many of the comments, I do empathize with those individuals because speaking from not just experience of treating patients with Alzheimer's but I did spend considerable amount as a caregiver and have a mother in law with dementia -- in [ my mother ]. So I surely recognize the desperate need for effective treatment here. But I do think caution it's necessary because a desperate need is not necessarily different approval of medication that has limited and short-term benefits. Particularly, and I also need to weigh in on what is likely to be relatively very high [indiscernible] might be.

I think you're breaking up, Dr. Stander.

I said I was finished. I'm sorry.

Okay. I heard the agency wanted to comment on the registration -- clinical trials registration issue before we continue. I'll be happy to hand it to the agency to...

Yes, Dr. [ Narendran ], this is Tiffany Farchione. I guess this issue about the clinicaltrials.gov and 12 weeks versus 6 weeks, I think, keeps coming up. I just want to make sure that, that -- we have plenty of data to consider. I don't want that to color the committee's opinion. We -- obviously, the company submitted their protocols to us prior to initiating any studies from the very first submission, week 6, was listed as the endpoint for that study. So it's always been week 6 since the beginning. So regardless of what was posted on clinicaltrials.gov, it sounds like there may have been a snafu there, but it has always been week 6. That's it.

Thank you for that clarification. Dr. Krishna? I want to give you a chance to weigh in.

This is Sonia Krishna. That week 6 question is what I had followed up before, and I appreciate the clarification. If you look at that, it looks like Study 019 is positive, but I wanted more data points to confirm that, and I would like to benefit to be a bit more sustained since we do have at least the 12 weeks of data. And so it does make me more concerned about the side of the placebo variation. Also, it would have been nice if any of the secondary endpoints were also positive. I'm also curious because we have spent a long time talking about this drug labeling for PPP, and we've talked a lot actually about the off-label use of atypical antipsychotics. But in the 6 years this drug has been out, there's no discussion about the off-label use of people who've been using it for [ EDP ], obviously understanding that we're trying to consider belaboring now. Thank you.

Ms. Witczak.

Kim Witzek, Consumer Rep. First of all, I'd like to just start out with this idea of unmet need. It seems like a lot of the drugs that are coming before committee do come with this unmet need and this idea of is it symptoms, is it -- it just feels like there's a -- what happens is it does lower -- because of using the fast-tracking mechanism, it does lower clinical trial requirements. So with that being said, that's just an overall comment. But the first study, 019, I go back to the fact that it was a broad-sweeping dementia-related psychosis and that the FDA at that time said we rejected it and it wasn't an adequate well-controlled study. And it sounds like from what -- from our papers that the FDA really would have liked to have seen a new study. And that's really what I would have liked to see because especially when Alzheimer's, when I asked that question earlier, how do we know that they were actually Alzheimer’s versus just dementia and were they given brain scans, so I feel like it's become very subjective. And especially because now we're trying to go, okay, it didn't work here, let's try to go here and then narrow applications. So that I would have liked to have seen and agree with the agency that a new study, and yet I know it's extensive, et cetera, to do all of that. Then the other issues of that it was over in the U.K. with predominantly a white population when we know that here -- and not a whole lot of men as well, but the one we know that according to CDC's numbers that the predominant Alzheimer's is in the black population, followed by Latin and white. So that's just something that I know that where overall a lot of trials have this issue, but I hope that and encourage sponsors and the industry to do a better job on that. In terms of the 045,I think they're just -- well, I've been going back to the primary, it didn't reach it. Secondary, in my mind, the fact that there wasn't any substantial and really it's 2 points meaningful enough to the patient population. And I heard and I sympathize when people are talking about what it is like to live in real-world situations with this. But I think it's not set for that. And then 045,we've got the different subgroups. And I think whether it's Parkinson's or the [ Lewy bodies ] or Alzheimer's, it just feels like it was to all like put together and it targets parcel out. That, again, from what I have understood in my lay person is that a lot of times, Parkinson's is that -- could that be the psychosis caused by the drugs that are used to treat it as opposed to Alzheimer’s, which is a different type of psychosis or mechanism that does create that. And then in terms of the last study using the original Parkinson's psychosis, I just don't think it's far reaching. It doesn't feel relevant. It feels like we're just reaching for strides so that we can get this unmet need and we can get out there in market. And I think -- and I don't know who is right before me, but she mentioned something of answer -- I mean, I would assume since I've heard many times at this committee that the FDA is not in the business of regulating off-label. But I would think there are probably physicians out there right now that have been using this drug off-label, and so I'll just say that. And then, of course, I am not even touching safety because I know that's not what we're doing. But safety is always a concern. And given what the -- even just the last Parkinson's psychosis, and we know what's happening in the latest rounds of data on the safety that's coming out of that. And Alzheimer's is a longer-term disease, so those are some of my concerns with the 2 studies.

This is Raj Narendran. And I'm just going to add in my -- I do feel like I agree that it kind of met the primary endpoint of Study 019, but I feel like there are too many issues in terms of the scales used are not well quantified or it was a very severely ill -- severely cognitively compromised population where symptoms fluctuate a lot. So maybe that explains some of the issues there. And the single time point was a concern. I thought the study sample was relatively small because it was designed as a Phase II study. And the lack of functional outcome issue was -- and lack of response on agitation and aggression and things like that, that go along with psychosis kind of give me pause to think how effective this drug is based on the Study 019. Study 045, I felt the randomized controlled trial design works very well if we know that there is an established data set. Like antipsychotics in schizophrenia, you remove them and they worsen, it seems fair. Or if your efficacy in a short-term trial prospectively is clear-cut defined based on some mechanism, I think the randomized controlled trial design is a good way to look at durability and maintenance. But I don't know that was the right design to go for. And I think as we all know, it didn't work out. And unfortunately, it was terminated early. And it was -- the study was underpowered to engage its efficacy in Alzheimer's disease psychosis. The last thing I think -- I do not think that the 020 study is relevant here. I think Parkinson's disease psychosis, as we know, is mostly LBD, Lewy-body dementia. There's a lot of inclusions. It has a predominance of visual hallucinations. And there's a lot more stability for hallucinations in Parkinson's disease as opposed to an Alzheimer's dementia psychosis, it's mostly dilution. So I'm not sure we can really use that data to support this particular indication. So that was kind of my thoughts. Is there anybody else who wants to weigh in before I summarize it? Everybody have a chance? Do we see any raised hands? People forgot to put their hands...

I guess the one with the more positive outlook from the data that was presented, I do -- first of all, I'm a neurologist, dementia doctor, 100% care for patients with -- largely with Alzheimer's, very little with Parkinson's disease, dementia, to be honest, last with Lewy body dementia. And in my experience in Alzheimer's per se, unlike in schizophrenia, psychosis doesn't always associate that strongly with agitation and aggression, cognitive decline and inability of patients to actually understand what's going on. It's more likely to cause aggressive behavior and agitation and not so much the psychotic episodes. But again, that's my observations from the population that they care for. I thought Study 019 met its primary outcome and that the exploratory analysis support the findings after controlling for the protocol deviation. I found the responder analysis to be convincing in presenting -- by presenting good separation of the curves. My only question was about durability of effects, but there is that fluctuating aspect of psychosis, unquestionably so. In terms of Study 045, I found the data on SAPS, HD and CGI convincing. It is unfortunate that there were not more Alzheimer's disease subjects enrolled for better power and the study was stopped early. But the effect size in AD meets my expectations. I don't anticipate it to ever match Parkinson's disease dementia or dementia with Lewy body for that matter. And the time to relapse curves were persuasive. The Parkinson's study in its own doesn't support an indication for Alzheimer's, but the fact that the medication has been on the market and has been administered to tens of thousands of people safely does matter as our other options had a black box warning and have cognitive side effects and reduce the mobility of our patients by virtue of Parkinsonism. So there is the unmet need. We can't ignore that. And also to consider that these advanced dementia trials are very hard to conduct. And I cannot help but see more positive than negative in the data presented today. That's it.

Is there anybody else in the committee with their hands up? Everybody has lowered their hands, so I will summarize the discussion I heard so far. So from the committee, we heard mostly on -- that we heard from the committee members of the Study 019 is not adequate. People raised the protocol deviations issue, although they felt that was somewhat addressed by the agency's review. There were some concerns about the separation only at week 6. People felt this effect, it was very small, and also there was some concerns that it was not maintained. There was also questions raised about the validity and construct of the outcome measures used, lack of signal in the secondary measures and lack of functional outcome improvement was a concern. People also raised concerns about the [ ethnic ] composition. There was also I heard that people thought that it was designed as a Phase II trial, so it was not really -- didn't provide sufficient evidence as a Phase III larger trial would have done. But I also heard some positive comments on 019 that some people felt it was positive and the data was supportive but not perfect. They also felt it was persuasive despite the audit and the deviations and people -- I also heard that it was technically a win. With respect to Study 045, I heard that the randomized trial design -- by design was not the best to look at the efficacy because of the selection bias of when we're including responders and withdrawing the drug, people also felt that the premature termination of the study because it was underpowered to really gauge the efficacy in Alzheimer's disease psychosis. And that was unfortunate that it was terminated early. I also heard that the dopaminergic drug issue is not very convincing. And I also heard that the post-hoc analysis in the separate groups is not necessarily a randomized comparison to provide this clear-cut efficacy data. With respect to the Parkinson's disease psychosis 020 study is relevant to here. I felt many people today was not relevant or the relevance was unclear. People thought it was a different illness. But people also thought -- agree that there was some overlap between the 2 conditions and maybe there's some biological possibility that psychosis could be effectively treated with pimavanserin. I also heard that because this has been administered safely for a large population, PDP population, it could be reasonable to go forward. So that's kind of my summary. We would like to move to question #2. Is there anything else anybody wants to add to the summary before I move to the voting question? I do not see any other hands raised, so there's no further discussion on the discussion question. We will now move to the next question, which is a voting question. Dr. Joyce Frempong will provide the instructions for the voting.

Question 2 is a voting question. Voting members will use Adobe Connect platform to submit their votes for this meeting. After the Chairperson has read the voting question into the record and all questions and discussions regarding the wording of the vote question are complete, the chairperson will announce that voting will begin. If you are a voting member, you'll be moved to a breakout room. A new display will appear where you can submit your vote. There will be no discussion in the breakout room. You should select the radio button, that is the round circular button, and the window that corresponds to your vote, yes, no or abstain. You should not leave the no vote choice selected. Please note you do not need to submit or send your vote. Again, you only need to select the radio button that corresponds to your vote. You'll have the opportunity to change your vote until the vote is announced as closed. Once all voting members have selected their vote, I will announce that the vote is closed. Next, the vote results will be displayed on the screen. I'll read the vote results from the screen into the record. Thereafter, the Chairperson will go down the roster, and each voting member will state their name and their vote into the record. You can also state the reason why you voted as you did if you want to. However, you should also address any subparts of the voting question, if any. Are there any questions about the voting process before we begin...

This is Dr. Stander. Just a quick question on the -- it says for the no vote, if you vote that way, we're not as [ yeses ] you're supposed to provide the rationale. Is there going to be like to take that in on the site?

No. If you explain your rationale, it would be for -- if -- when you vote. Dr. Narendran will ask you your reason why voted as you did, and you can speak. Okay. All right, Dr. Narendran?

Sorry, voting question. Question number two, does the available evidence support the conclusion that pimavanserin is effective for the treatment of hallucinations and delusions in the Alzheimer's disease psychosis population? If yes, provide the rationale, if no provide the rationale and a recommendation for what -- further evidence should be generated. Are there any questions about the question concerning the wording before we decide to vote? If you do have questions about the wording, please raise your hand. Seems pretty clear. Joyce, I'll hand it to you.

We will now move voting members to the voting breakout room to vote only. There will be no discussion in the voting breakout room. Voting has closed and is now complete. Once the vote results display, I'll read the vote result into the record. The vote results are displayed. I will read the vote totals into the record. The Chairperson will go down the list, and each voting member will state their name and their vote into the record. You can also state the reason why you voted as you did if you want to. However, you should also address any subparts of the voting questions, if any. For our results, we have 3 yeses, 9 no's and no abstain.

Thank you. We will now go down the list and have everyone who voted state their name and vote into the record. You may also want to address the subpart questions. If you provide the rationale if you voted yes. And if no, provides the rationale and the recommendation for what further evidence should be generated. We will start with Dr. Johnston.

That would be giving me a promotion. I'm actually the patient advocate.

Ms. Johnston.

I did vote yes. So my name is Colette Johnston. And what I thought was going to be a fairly easy day turned out to be very difficult for me. I have over 25 years of experience in reviewing clinical trials on various IRBs. And from that perspective, I have so many concerns, and there are all the concerns that have been addressed from patient population to inform consent, to this not being applicable to our patient population. That said, my role here today is as a patient advocate, and so I have to take myself back to the night I got a phone call that my father in a care center had just been in a physical altercation with another patient and who is the most docile kind men you would ever meet. And I'm 250 miles away, and the only thing they can do is send him to a psych ward in an ambulance. If I would have had the opportunity to use this drug, that whole scenario would have changed, and the next 2 months of his life before he passed would not have been -- could not have been -- possibly would not have been spent in a drug [ abuse ] sedation. So from a patient advocate's point, I know that desperation should not drive us. And I do feel like this is being pushed towards the market, and I have to say I was fairly safe with my yes vote because it was pretty obvious that we were going to have more no. I couldn't look at somebody in that position and justify that they couldn't have access to that drug, especially since this is being used off-label. So that's my rationale.

Sorry, I lost connection. Dr. Follmann?

Yes. This is Dean Follmann of NIAID. I voted no, and I think the reasons I articulated in the last question with a lot of what had been said. In terms of further evidence, I'd like to see a randomized trial in [ ADP ]. And just one small comment on that. I've not really seen the randomized withdrawal design before, but it seems like if you have such a design and you show a striking benefit, then you would want to give those randomized to placebo the effective drug. And you could do this in a blinded way and look at sort of what is the change in like a symptom score at the time they get the drug or in a blinded way the drug people continue to get the drugs or look at the change. If something like that had been done in 045, we would have additional evidence that -- yes, we'd have evidence whether the drug worked or not. So anyway, just a consideration for a future twist on a randomized withdrawal design. That's all.

Dr. Fiedorowicz?

Yes, so Jess Fiedorowicz. My vote was a no. A lot of the reasons weren't clear from the prior discussion. As far as further evidence it would be required, it suggests a Phase III RCT for -- in Alzheimer's psychosis, as was proposed by the original Lancet paper for Study 019. I do want to also just add that while I understood from the applicant and the agency that the original protocol specified 6 weeks, the registration is what is available to the global public and the scientific community and do want to underscore that. And for any such follow-up study, I think everyone has already touched on this, but an adequate representation, particularly racial representation of this study would be valuable.

Ms. Witczak?

Kim Witczak, Woodymatters consumer rep. I voted no, and I articulated a lot of the reasons in prior conversations. But again, I always say that I cannot -- without the benefit, we also must look at the harms in totality, although I know that wasn't our assignment. And then in terms of what I would like to see, and we heard it from Colette, there's a lot in some people that -- in the public speakers. But there is a desire and a need for this, but I would encourage the sponsor to do a Phase III in the proper population with the proper racial ethnic background as well as test for Alzheimer's disease. And I would love to see those results and see what comes back with it. So that would be my encouragement.

Thank you. Dr. Apostolova?

Liana Apostolova, Indiana University. I voted yes. And the rationale behind that is that despite the fact that those studies were small in terms of AD population, there was modest efficacy in both, which survived after controlling for protocol deviations and even -- it also was evident in the prematurely stopped trial to some extent. I also [ ensued ] by the fact that there is true life use data on pimavanserin. And we know it's safe. It doesn't cause the side effects that the atypical antipsychotics, which is the only other class of drugs we have available. So my vote was yes.

Dr. Thambisetty?

Thank you, Dr. Rajendran. Madhav Thambisetty, NIA. I voted no, and I think all of the reasons from my vote were described in the discussion question number one. As far as what further evidence should be generated, I would echo back what the FDA advised the applicant in the June 2021 type A review meeting as well as in the December 2021 type B guidance meeting when they advised that an additional adequate and well-controlled study in AD psychosis would likely provide the strongest data and support of a resubmission. So I would echo that advice. I think they got it spot on. If I may just add on an unrelated note, I found the patient testimony today, extremely moving and powerful. I myself, I'm a neurologist who has cared for patients for more than 20 years. I recognize the unmet need in the field. I just think that the unmet need should not be just [indiscernible] for us to cut from us. It should, on the other hand, inspire us to do the best time and apply the most rigorous standards towards analyzing the results from those studies. And in this context, I would also like to acknowledge the significant contributions that the applicants has made. I think these are incredibly difficult studies to run in very, very difficult patients. And I think the applicant also must be congratulated for doing their best to bring tangible benefits to our patients.

Dr. Cudkowicz?

Yes, Merit Cudkowicz. I voted yes. And the reason was that Study 019 was positive. It was on the primary outcome that was agreed on in advance with the FDA, and it's not a typical for looking at psychiatric type systems. I think 6 weeks and getting there faster is highly relevant for people suffering from psychosis in Alzheimer's and to their family members. About 045 was mildly supportive, no new safety issue. And I did -- I was persuaded by the disease experts' points about the similarities in the biology of hallucinations and delusions in Parkinson's and Alzheimer’s. I do think that the opening question still, but that many of those could be addressed in post-market type studies.

Dr. Stander?

Yes. I voted no. I expressed many of the concerns, which are similar to those raised by others, just relatively limited [indiscernible] population. I do think that as Dr. Thambisetty said, I think it's very commendable. The applicant is trying to conduct very difficult studies. In fact, I said earlier, I can empathize and identify with those in the community and elsewhere who acknowledge and express their deep concern and the need for effective therapies in the domain. But I do have also concerns that once medications or treatments are made available for problems like this, it's a little like the Genie being let out of the bottle and they tend to get used for a wide range of symptoms, patients that may not really be applicable, extreme cost and to negative effect. So I would recommend, as others have said, a more fine study focused entirely on Alzheimer's population and preferably from my expected longer duration of efficacy that is showing the benefit 6 weeks or the [indiscernible] in this company.

Raj Narendran. I voted no for the reasons mentioned before. I mean I think additional controlled -- randomized controlled trial, which is maybe incorporate some Alzheimer's blood markers, which is bigger sample size with better outcome measures, I think, would be reasonable and to generate strong data to support an indication going forward. That's all I have to say. Dr. Iyengar, you're next.

This is Satish Iyengar from Pittsburgh. I also voted no for the reasons I stated before. I also think that what's really needed is a well-powered study on an appropriate -- demographically appropriate and large ADP sample.

Dr. Krishna?

This is Sonia Krishna. I voted no for the discussion we've had. And I would like to add that, yes, I also would agree with a new study just in this patient population. And I'm also very interested to find out what has been going on for the past 6 years when this drug has been out and other people have used it, even anecdotal information, maybe from the community providers who have been treating these patients.

Dr. Dunn?

Walter Dunn, UCLA. I voted no based on the interpretation that the term conclusion in the question requires compelling evidence. I do think that Study 019 provides some evidence that pimavanserin can be effective in Alzheimer's dementia psychosis, but that further study is warranted to reach the level of a conclusion. But I would give Study 019 partial credit, again, technically a positive study but attenuated by: one, the positive outcome in week 6 looks like it's being driven by worsening in the placebo arm, which appears unrelated disease process; two, limitations in the primary outcome scale, capturing only a narrow view of symptoms and impairment; and then three, lack of concurrence, of course, in the secondary outcomes. As far as Study 045, the only 2 conclusions I can make is that there is a differential response between Parkinson's and Alzheimer's disease psychosis and that pimavanserin can be very effective as a maintenance treatment for Parkinson's disease psychosis. Accordingly, therefore, Study 020 in Parkinson's patients would not support efficacy in the Alzheimer's population. In terms of what additional evidence should be generated, I think there is no way around the need to run another study specifically in the Alzheimer's population. However, if the division is agreeable, I believe a positive randomized withdrawal study would provide compelling evidence. And I think that's despite what's been said about such designs enriching for responsive patients. All that being said, I'd like to return back to my original thought about narrow versus broad considerations. The questions before the committee have been narrow and precise. So -- but I -- so I trust that the agency will take a broad approach in their final decision about approval. There are many factors which we have not formally discussed today, such as safety and unmet clinical need. There's clearly a need in this highly vulnerable population. As a clinician, I am a proponent for having as many tools in the toolbox as possible, so I trust the agency will take into consideration all these factors in their final decision. I would also like to convey the final message to the sponsor and payers in advocating for our patients about improving access to this drug. As this is an approved medication already on the market, the real issue at hand with this approval is about lowering the financial barriers to access this treatment. Therefore, improving access is something well within the capability of the company and payers without having to involve agency.

So it seems like from what I heard, just to kind of summarize, many of the members wanted to see an additional controlled randomized trial. Some people thought a withdrawal -- positive with randomized withdrawal trial would suffice. People wanted to see adequate representation in terms of ethnicity and race. There's also people wanted to see a trial that -- where there's a longer duration of efficacy is being assessed. The people who have voted yes felt there was a strong unmet need. They felt there was modest efficacy and reasonable signal, although small, within the 2 trials that were done, especially 019, which many members who even voted no thought that -- had agreed that it met the positive endpoint. Other people thought who voted yes felt that the drug is already available and doesn't have any safety concerns as available with a typical antipsychotics which are used to treat patients off-label. So that's kind of my summary. Is there any other comments from the agency before we adjourn the meeting? Anybody from the agency want to comment or make any last -- Dr. Farchione, if you're there?

This is Paul. Sorry, go ahead, Bernie.

This is Bernie, Deputy for Psychiatry. Tiffany just had her call drop in opportune moment. But I just wanted to thank the members of the AC for their careful consideration. I thank the public hearing comments, and we will take all of this under consideration when making our decision.

Thank you, Dr. Fisher. We will now adjourn the meeting. Thank you, everyone, for attending. I do want to thank the sponsor. I want to thank all the people who participated in the open public hearing and gave powerful testimony. And I also want to thank the agency staff for all the hard work they do. Thank you. We will now adjourn the meeting.