A Phase 3, Randomized, Double-Blind, Placebo-Controlled Study to Evaluate the Efficacy and Safety of Pimavanserin as an Adjunctive Treatment for the Negative Symptoms of Schizophrenia (ADVANCE-2) in Patients With Predominant Negative Symptoms

Abstract

Background and Hypotheses

Negative symptoms of schizophrenia (NSS) carry a substantial burden, and there are no treatments currently approved for NSS. The efficacy of pimavanserin, a selective 5-HT_2A inverse agonist and antagonist, in treating NSS was assessed.

Study Design

ADVANCE-2 was a phase 3, randomized, double-blind, placebo-controlled study of pimavanserin in patients with schizophrenia and predominantly negative symptoms. Patients were randomized (1:1) to receive pimavanserin (34 mg/day) or placebo alongside ongoing background antipsychotic medication. Eligible adults were aged 18–55 years and had access to a caregiver. The primary and key secondary endpoints were the change from baseline to week 26 in the Negative Symptom Assessment–16 (NSA-16) total score and Clinical Global Impression–Schizophrenia Scale-Severity (CGI-SCH-S) negative symptom score, respectively.

Study Results

Of the 454 randomized patients, 71 (39 placebo; 32 pimavanserin) discontinued and 383 (188 placebo; 195 pimavanserin) completed the study. The safety and full analysis sets comprised 453 and 446 patients, respectively. The NSA-16 change from baseline to week 26 was not significantly different between groups (least squares mean difference: −0.67; SE, 0.95; [95% CI: −2.54, 1.20]; P = .48; Cohen’s d effect size: 0.07). Treatment-emergent adverse events occurred in 30.4% with pimavanserin and 40.3% with placebo.

Conclusions

In this study, pimavanserin was well tolerated, and although it demonstrated a similar treatment effect as in the prior phase 2 study favoring pimavanserin, treatment with pimavanserin vs placebo did not result in significant differences for primary or other endpoints.

Introduction

Schizophrenia is a chronic and debilitating condition with a heterogeneous clinical presentation that includes positive symptoms, cognitive disruptions, and negative symptoms.¹ Negative symptoms of schizophrenia (NSS) include diminishing motivation and interest and reduced expression.¹ NSS can be categorized as either primary and caused by schizophrenia, or secondary and caused by some other pathology. They are also described as either enduring or transient, resolving after treatment of the causative pathology.^1,2 NSS can be further categorized as predominant if negative symptoms are pronounced and meet minimum criteria concomitant with relatively mild, well-controlled positive symptoms.³

Consistent with the heterogeneous presentation of schizophrenia, negative symptoms occur frequently and contribute substantially to the burden of the disease.^1,4,5 They are a core clinical feature of schizophrenia and are estimated to affect up to 60% of patients.^4,6 Two large, retrospective studies of electronic health records reported that 52.5% of patients with schizophrenia had at least 1,⁴ and 41.0% had 2 or more negative symptoms.⁵ The high burden associated with NSS has been attributed to reduced health-related quality of life and worse functional outcomes due to disruptions in multiple life domains, including the social domain and the livelihood domain, which includes work and academic performance.^1,7,8

Given the high prevalence and burden of NSS, there is a substantial need for treatments that effectively treat this symptom domain. Currently, available antipsychotics are primarily D₂ dopamine-targeted therapies, which are effective for positive symptoms but ineffective for reducing NSS.^9–11 Recent evidence demonstrates that relatively newer atypical antipsychotics with a higher affinity for serotonin (5-HT₂) receptors may reduce NSS; however, there are currently no approved therapies for NSS despite substantial widespread efforts.^2,10,12 Likely contributors to the lack of approved therapies are the numerous challenges of conducting clinical trials for NSS, which include many factors, such as the choice of and limitations to negative symptom assessment tools or scales, monotherapy vs adjunctive trial design, the heterogeneity of trial populations, the limited understanding of NSS pathophysiology, elevated placebo effects and changes in response related to treatment administration, the lack of agreement on severity criteria for positive symptoms and depression to limit pseudospecificity, and the lack of long-term negative symptom data, among other factors.^13,14 It is likely that several of these challenges contributed to recent trials involving multiple candidate compounds (ie, roluperidone, cariprazine, volinanserin, ritanserin, and eplivanserin) that failed to meet their endpoints and gain approval.^15–21 Another challenge for recent clinical trials investigating NSS is the COVID-19 pandemic, which generated substantial psychological stress among both the general population and vulnerable populations (including patients with schizophrenia) via pandemic-associated disruptions of life, such as disrupted daily routines, social isolation, and travel restrictions.^22,23

This trial investigates pimavanserin, which is a selective 5-HT_2A receptor inverse agonist and antagonist with lesser affinity for 5-HT_2C.²⁴ It has negligible affinity for dopaminergic, muscarinic, histaminergic, and adrenergic receptors.²⁴ Based on data from compounds with similar receptor profiles, pimavanserin is hypothesized to provide additional benefit in adults with NSS whose treatment benefit has been maximized with ongoing atypical antipsychotics.²⁵ When combined with atypical antipsychotics, pimavanserin may reduce NSS through increased dopamine efflux in the prefrontal cortex, downstream potentiating effects on n-methyl-d-aspartate (NMDA) currents, and a complementary combination of 5-HT_1A agonism (from the background antipsychotic) with 5-HT_2A inverse agonism.^26,27 Although many atypical antipsychotics have relatively high levels of 5-HT_2A receptor occupancy and an effect on negative symptoms, the relationship between 5-HT_2A receptor occupancy and therapeutic efficacy is poorly understood.²⁸ Adding pimavanserin to atypical antipsychotic medications with lower selectivity for 5-HT_2A receptors over D₂ receptors may allow titration for increased 5-HT_2A receptor occupancy relative to the D₂ receptor occupancy, thereby increasing clinical efficacy. This strategy is hypothesized to not compromise occupancy at other important targets and to not increase the risk of extrapyramidal side effects. At a speculative level, while 5-HT_2A receptor occupancy may be high in patients receiving risperidone^29–31 and olanzapine,³¹ the unique inverse agonism of both 5-HT_2A and 5-HT_2C receptors by pimavanserin at high doses (ie, 34 mg) could potentially further enhance neocortical dopamine release.²⁶

Pimavanserin was approved by the United States Food and Drug Administration (FDA) in 2016 to treat hallucinations and delusions associated with Parkinson’s disease psychosis.³² The phase 2 ADVANCE study in adults with schizophrenia and predominant NSS demonstrated that pimavanserin was associated with significantly greater reductions in NSS than placebo, as measured by the primary endpoint of change in 16-item Negative Symptom Assessment (NSA-16) score (least squares mean [LSM] difference − 1.9 [standard error (SE) 0.95]; P = .043; Cohen’s d effect size 0.211).³³ Furthermore, subsequent exposure-response modeling of data from the ADVANCE study has shown that increasing pimavanserin plasma concentrations were associated with improvements in NSS.³⁴ Together, these results suggest that pimavanserin may effectively reduce NSS in patients with schizophrenia and predominant NSS. The objective of this study was to evaluate the efficacy of pimavanserin vs placebo on NSS in adults with schizophrenia and predominant negative symptoms.

Methods

Study Design and Participants

In this phase 3, randomized, double-blind, placebo-controlled, multicenter study, outpatient adults with predominant NSS were randomly assigned to receive pimavanserin (34 mg) or placebo once daily added to an ongoing antipsychotic (Supplementary Figure S1). Patients were screened at 91 sites and recruited from 82 outpatient clinics (hospitals and community clinics) across Europe and other regions across the world. Patients were assessed for eligibility during the screening period (≤6 weeks). After screening, eligible patients entered the 26-week treatment period, which was followed by an optional 52-week, open-label extension or a safety follow-up for patients with early termination or who did not enroll in the open-label extension (Figure 2).

Figure 2.

Least squares mean (+/− SE) change in the NSA-16 Total Score from Baseline-by-Study Visit (Full Analysis Set). LSM, least squares mean; MMRM, mixed-effect model for repeated measures; NSA-16, 16-item Negative Symptom Assessment, negative symptom assessment; SE, standard error.

LSM from MMRM with fixed effects of region (Europe or the rest of the world), planned treatment (adjunctive pimavanserin, adjunctive placebo), study visit (weeks 2, 4, 8, 14, 20, 26), treatment-by-visit interaction, baseline score (continuous covariate), and baseline-by-visit interaction.

Eligible adults (18–55 years) were outpatients diagnosed with schizophrenia (≥1 year prior to screening) and presenting with predominant NSS. Schizophrenia diagnosis was confirmed by the Structured Clinical Interview for DSM-5 Clinical Trials Version (SCID-5-CT). Patients were required to be medically stable (ie, no recent hospitalization for exacerbation of psychotic disorder) based on the investigator’s judgment for ≥12 weeks before screening (Supplementary Table S2) and to have a designated caregiver to support study completion and adherence with study procedures. The collection of demographic information included age, sex, geographic region, race, body mass index, employment status, and other measures related to schizophrenia diagnosis. The list of protocol-permitted, background atypical antipsychotic medications that did not carry a warning for prolongation of the corrected QT interval in their label at the time of the ADVANCE-1 study included aripiprazole, aripiprazole long-acting injectable (LAI), asenapine, brexpiprazole, cariprazine, lurasidone, olanzapine, risperidone, or risperidone LAI. This ADVANCE-2 study included the addition of paliperidone, including its extended-release formulation (≤9 mg/day), and paliperidone palmitate based on evidence from clinical studies indicating that they were not associated with an increased cardiovascular risk.³⁵

To control for pseudospecificity and exclude patients with depression or extrapyramidal side effects that could confound the evaluation of negative symptoms, patients with any active comorbid psychiatric disorders, other than schizophrenia, or other disorders that would interfere with study assessments were excluded; patients were also assessed using movement and depression scales (Supplementary Tables S1 and S2). To reduce the potential for secondary improvement in negative symptoms relating to improvements in other symptom domains, this study selected patients with predominant NSS—pronounced negative symptoms and minimal positive symptoms. Pronounced negative symptoms were defined at screening and baseline as a score ≥20 on the sum of the 7 Positive and Negative Syndrome Scale (PANSS) Marder negative factor items (and scores ≥4 on at least 3 of the 7 PANSS items; or ≥5 on at least 2 of the 7 PANSS items) and a Clinical Global Impression of Schizophrenia Scale–Severity (CGI-SCH-S) for the NSS score ≥4 (moderately ill or worse).³⁶ Minimal or well-controlled positive symptoms were defined at screening and baseline as a score ≤22 on the sum of the 8 PANSS Marder positive factor items (with ≤2 of the selected items [P1 delusions, P3 hallucinatory behavior, and P6 suspiciousness or persecution] rated 4, and none rated ≥5, at screening and baseline).

Patients were required to have had ≥8 weeks of treatment with an adequate dose of a selected atypical antipsychotic prior to screening (Supplementary Table S2); treatments were selected in agreement with the FDA. To confirm treatment adherence, antipsychotic blood concentrations had to be above the lowest measurable quantity level by laboratory cutoff. Eligibility was confirmed using a well-established method of independent adjudication, which consisted of a structured telemedicine interview.³⁷ Full inclusion and exclusion criteria are included in Supplementary Table S2.

All participants provided written informed consent prior to study participation. This study complied with the Declaration of Helsinki, the International Conference on Harmonization Good Clinical Practice Guidelines, and local regulatory requirements. The protocol was approved by local independent ethics committees or institutional review boards at each site.

Randomization and Masking

Eligible patients were randomized (1:1) to receive pimavanserin (34 mg/day) or placebo added to their current, ongoing atypical antipsychotic treatment. Randomization was performed with a computer-generated, permuted-block sequence, stratified by region (Europe or the rest of the world). Assignments were double-blinded such that patients, caregivers, investigators, outcomes assessors (ie, study statisticians), study sponsors, contract research organizations, and other vendors were masked to treatment group assignments. Blinding was assured by restricting access to patients’ treatment assignments and using identical tablets and packaging for pimavanserin and placebo treatments.

Procedures

The pimavanserin dosage of 34 mg/day was selected based on the prior, flexibly dosed ADVANCE-1 study.³³ Treatment adherence to pimavanserin was demonstrated to be ≥95.0% in a post hoc analysis of the prior ENHANCE and ADVANCE studies.³⁸ As in prior trials, in this study, treatment adherence was confirmed using the pill quantity returned at each visit as well as pharmacokinetic and pharmacodynamic data (ie, the plasma concentration of pimavanserin, its metabolite, and the main background antipsychotic). Patients continued to receive their ongoing antipsychotic at a stable dosage for the duration of the study. Dose changes were not allowed within 4 weeks of screening for participants treated with ongoing oral antipsychotics or within 16 weeks for participants treated with ongoing LAI antipsychotics. Protocol-specified dosages of hypnotics, anticholinergics, or anxiolytics were allowed for insomnia, extrapyramidal side effects, and anxiety, respectively.

The screening evaluation consisted of the SCID-CT neuropsychiatric interview, physical examination, vital signs, electrocardiogram (ECG), and laboratory tests. All study scales were administered by trained and certified raters, and caregivers were required to attend rating visits. Scales that assessed psychopathology included the 16-item Negative Symptom Assessment (NSA-16; primary endpoint), the CGI-SCH-S of negative symptoms score (key secondary endpoint), CGI-SCH-Improvement scale (CGI-SCH-I) of negative symptom scores, PANSS, all PANSS subscales, and the Informant Questionnaire PANSS (IQ-PANSS). Scales that assessed psychopathology and functioning are described in Supplementary Table S1. Additional study procedures, including protocol amendments, schedule of efficacy assessments, and reasons for study removal are available in Supplementary Methods.

Outcomes

The primary endpoint was the change in NSA-16 total score from baseline to week 26 (Supplementary Methods).^39,40 The key secondary endpoint was the change from baseline to week 26 in the CGI-SCH-S negative symptom score. Other secondary endpoints included the CGI-SCH-I negative symptom score at week 26, CGI-SCH-I negative symptom responders (negative symptom score of 1 or 2) at week 26, the change from baseline to week 26 in the Personal and Social Performance scale (PSP) score, the proportion of NSA-16 responders (≥20% and ≥30% reduction in NSA-16 total score) at week 26, the change from baseline to week 26 in the PANSS total score, PANSS negative subscore, and PANSS Marder factor (negative symptoms) score. The responder analysis of NSA-16 used a spectrum of cutoffs to define responders, including patients with ≥20% improvement and ≥30% improvement at week 26. Other exploratory endpoints are listed in Supplementary Methods.

Safety measures, assessed at baseline and routinely during the study, included physical examination, 12-lead ECG, clinical laboratory tests, adverse events, vital signs, and Columbia-Suicide Severity Rating Scale (C-SSRS) scores. Movement scales, which were assessed at baseline and weeks 2, 14, and 26 included the Abnormal Involuntary Movement Scale (AIMS), Barnes Akathisia Rating Scale (BARS), and the Simpson–Angus Extrapyramidal Side Effects Scale (SAS); the movement disorder scales are described in Supplementary Methods.

Statistical Analysis

The sample size determination was based on an estimated common standard deviation of 9.0 for NSA-16 total scores. An estimated 191 evaluable patients per treatment group would provide at least 90% power to detect a difference of 3.0 points in mean change in NSA-16 total from baseline to week 26 between pimavanserin and placebo at a significance level of 0.05 (two-sided t-test). On the basis of a potential non-evaluable rate of up to 10%, approximately 426 patients were randomly assigned (213 patients per treatment group).

The safety analysis set included all randomized patients who received at least 1 dose of the study drug. The full analysis set included all randomized patients who received at least 1 dose of the study drug and had both a baseline and at least 1 postbaseline NSA-16 total score.

The NSA-16 total score, CGI-SCH-S, PANSS, CGI-SCH-I, and PSP were analyzed using mixed-effect model repeated measures (MMRM), with change from baseline as the dependent variable; independent variables included treatment (pimavanserin or placebo), visit (categorical variable), treatment-by-visit interaction, region (Europe or the rest of the world), baseline value of the endpoint being analyzed with the exception of CGI-SCH-I, and baseline-by-visit interaction. For CGI-SCH-I, the baseline CGI-SCH-S score and geographic region were included as covariates in the MMRM analysis. An unstructured covariance matrix was used, and the Kenward–Roger approximation was used to adjust the denominator degrees of freedom. The treatment comparison was based on the difference in the LSM at week 26 and was tested at an α level of 0.05 (two-sided) using the full analysis set. A hierarchical testing procedure was used to control for type I errors. For NSA-16, a sensitivity analysis was conducted using a pattern-mixture model (Supplementary Methods). For responder analyses, the treatment groups were compared using a Cochran-Mantel-Haenszel test stratified by geographic region (Europe or the rest of the world). The adjusted difference in the percentage of responders between the treatment groups (pimavanserin minus placebo group) was calculated using the weighting scheme of Cochran-Mantel-Haenszel and Newcombe’s 95% CI. All analyses were done using SAS version 9.4 or higher. The study is registered at ClinicalTrials.gov (NCT04531982).

Results

Baseline Characteristics

Between July 6, 2020, and July 31, 2023, 671 patients (640 unique patients) with NSS were screened, of which 217 (32.3%) patients were determined to be ineligible (Figure 1). The remaining 454 patients were randomized (1:1) to either placebo (n = 227) or pimavanserin (n = 227). Following randomization, 71 patients (39 placebo and 32 pimavanserin) discontinued the study and 383 patients completed the study, including 188 (82.8%) patients in the placebo group and 195 (85.9%) patients in the pimavanserin group. In total 453 patients (placebo n = 226; pimavanserin n = 227) were included in the safety analysis set, and 446 patients (placebo n = 222; pimavanserin n = 224) were included in the full analysis set. Baseline characteristics were comparable between the treatment groups (Table 1). Other than antipsychotic medications, the only concomitant medications taken by >5% of patients in both treatment groups were biperiden and lorazepam. Most patients were diagnosed with paranoid schizophrenia, while others were diagnosed with a mix of undifferentiated, disorganized, residual, and simple schizophrenia.

Figure 1.

Patient disposition (safety analysis set).

Table 1.

Baseline Characteristics (Safety Analysis Set)

Characteristic	Placebo (n = 226)	Pimavanserin (n = 227)
Age at screening (years)	37.5 (0.7)	36.3 (0.6)
Sexa
Male	136 (60.2)	132 (58.1)
Regiona
Europe	187 (82.7)	188 (82.8)
Rest of world	39 (17.3)	39 (17.2)
Racea
White	224 (99.1)	225 (99.1)
Black/African-American	0 (0.0)	0 (0.0)
Asian	0 (0.0)	0 (0.0)
American Indian or Alaska Native	1 (0.4)	0 (0.0)
Other	1 (0.4)	2 (0.9)
Body mass index (kg/m2)	26.59	26.41
Employment statusa
Full-time	19 (8.4)	14 (6.2)
Part-time	22 (9.7)	21 (9.3)
Unemployed	138 (61.1)	143 (63.0)
Retired	47 (20.8)	49 (21.6)
Duration of schizophrenia (years)	12.5 (0.5)	11.6 (0.6)
Diagnosed with schizophrenia > 5 yearsa	173 (76.5)	151 (66.5)
Duration of negative symptomsa
<1 year	24 (10.6)	13 (5.7)
1 to 5 years	106 (46.9)	119 (52.4)
>5 years	96 (42.4)	95 (41.9)
Time since first antipsychotic (years)	11.45 (0.5)	10.39 (0.5)
Background antipsychotic typea
Aripiprazoleb	56 (24.8)	58 (25.6)
Brexpiprazole	1 (0.4)	0 (0.0)
Cariprazine	15 (6.6)	21 (9.3)
Lurasidone	0 (0.0)	7 (3.1)
Olanzapine	70 (31.0)	61 (26.9)
Risperidoneb	62 (27.3)	62 (27.3)
Paliperidoneb	22 (9.7)	18 (7.9)
Duration of background antipsychotic medication (months)	30.35 (2.6)	25.08 (2.5)
Other medications
Anticholinergic agentsc	24 (10.6)	26 (11.5)
Anxiolyticsd	55 (24.3)	58 (25.6)
Hypnotics and sedativese	6 (2.7)	10 (4.4)
NSA-16 total score	60.8 (0.52)	61.3 (0.53)
PSP score	44.9 (0.70)	44.8 (0.71)
Schizophrenia negative symptomsa
NSA-16 total score ≤ 55	60 (26.5)	59 (26.0)
NSA-16 total score > 55	166 (73.5)	168 (74.0)
CGI-SCH-S, negative symptom score	4.8 (0.04)	4.8 (0.04)
CGI-SCH-S, negative symptom score ≥ 5a	170 (75.2)	165 (72.7)
PANSS total score	72.6 (0.6)	72.6 (0.6)
PANSS positive symptoms subscore	11.7 (0.2)	11.7 (0.2)
PANSS negative symptoms subscore	27.6 (0.2)	27.7 (0.3)
WoRQ total score	15.7 (0.2)	15.9 (0.2)
CDSS total score	0.8 (0.1)	0.8 (0.1)

CGI-SCG-S, Clinical Global Impression of Schizophrenia Scale-Severity; NSA-16, 16-item Negative Symptom Assessment; PANSS, Positive and Negative Syndrome Scale; PSP, Personal and Social Performance scale; WoRQ, Work Rehabilitation Questionnaire.

Data are mean (SD) unless indicated as otherwise.

^aData are presented as n (%).

^bIncludes oral and long-acting injectable formulation.

^cIncludes biperiden, biperiden hydrochloride, and trihexyphenidyl.

^dIncludes alprazolam, bomazepam, chlorazepate dipotassium, clonazepam, diazepam, duloxetine, escitalopram, fluoxetine, lorazepam, paroxetine hydrochloride, phenazepam, and sertraline.

^eIncludes eszopiclone, lormetazepam, other compounds and homeopathic treatments (ie, ethyl 2-bromoisovalerate; humulus lupulus; mentha × piperita oil; menthol; menthyl valerate, and medazolam hydrochloride), phenazepam hemisuccinate, zolpidem, zolpidem tartarate, and zopiclone.

Primary Efficacy Outcome

The change in NSA-16 total score from baseline to week 26 was not significantly different between the pimavanserin and placebo groups (LSM difference: −0.67; SE, 0.95; [95% CI: −2.54, 1.20]; P = .48; Cohen’s d, effect size 0.07; Table 2, Figure 2). Similar results were seen with the pattern-mixture model sensitivity analysis of the NSA-16 total score differences between the pimavanserin and placebo groups (LSM difference: −0.68; SE, 0.94; [95% CI: −2.52, 1.16]; P = .47). For the NSA-16 domain scores, compared with the placebo group, the pimavanserin group had greater, but not statistically significant, improvements on 4 of the 5 domains (ie, communication, emotion or affect, motivation, and retardation), but not the social involvement domain. In prespecified subgroup analyses by background antipsychotic medication, there were no significant differences in NSA-16 total scores between treatment groups (Supplementary Table S3).

Table 2.

Baseline and Mean Change at Week 26 for the Primary and Secondary Outcomes (Full Analysis Set)

Secondary endpoints	Placebo (n = 222)		Pimavanserin (n = 224)
	Mean (SE) baseline	LSM (SE) change	Mean (SE) baseline	LSM (SE) change	LSM (SE) difference [95% CI]e	P value	Cohen’s d effect size
NSA-16 total scorea,b,c	60.88	−11.13 (0.68)	61.34	−11.80 (0.67)	−0.67 (0.95) [2.54, 1.20]	0.48	0.07
NSA-16 domain scoresa,b,c
Communication	11.92 (0.20)	−2.41 (0.19)	11.97 (0.20)	−2.63 (0.19)	−0.22 (0.27) [−0.74, 0.31]	0.41	0.08
Emotion or affect	12.73 (0.12)	−2.33 (0.17)	12.93 (0.12)	−2.62 (0.17)	−0.28 (0.24) [−0.76, 0.19]	0.24	0.12
Social involvement	12.63 (0.15)	−2.05 (0.17)	12.85 (0.16)	−2.00 (0.17)	0.05 (0.24) [−0.43, 0.52]	0.85	–0.02
Motivation	16.68 (0.16)	−2.64 (0.21)	16.81 (0.18)	−2.77 (0.20)	−0.12 (0.29) [−0.69, 0.44]	0.67	0.04
Retardation	6.91 (0.11)	−1.76 (0.11)	6.79 (0.10)	−1.83 (0.11)	−0.07 (0.16) [−0.38, 0.27]	0.65	0.05
CGI-SCH-S negative symptom scorea,b,c	4.85 (0.04)	−0.86 (0.06)	4.83 (0.04)	−0.87 (0.06)	−0.01 (0.09) [−0.19, 0.17]	0.89	0.01
CGI-SCH-I negative symptom scoreb,c,d	---	2.94 (0.06)	---	2.81 (0.06)	−0.13 (0.09) [−0.31, 0.04]	0.14	0.15
PSPa,b,c	44.85 (0.70)	11.18 (0.78)	44.77 (0.72)	12.03 (0.78)	0.85 (0.78) [−1.32, 3.02]	0.44	0.08
PANSS total scorea,b,c	72.56 (0.61)	−9.41 (0.68)	72.71 (0.64)	−10.66 (0.67)	−1.24 (0.95) [−3.11, 0.62]	0.19	0.13
PANSS negative symptoms subscoresa,b,c	27.59 (0.24)	−5.02 (0.32)	27.71 (0.25)	−5.29 (0.32)	−0.27 (0.45) [−1.16, 0.63]	0.56	0.06
PANSS Marder factor (negative symptoms) scorea,b,c	26.85 (0.21)	−5.72 (0.32)	26.79 (0.22)	−6.06 (0.31)	−0.34 (0.45) [−1.22, 0.54]	0.45	0.08

CGI-SCH-I, Clinical Global Impression of Schizophrenia Scale-Improvement; CGI-SCH-S, Clinical Global Impression of Schizophrenia Scale–Severity; LSM, least squares mean; MMRM, mixed-effect model for repeated measures; PANSS, Positive and Negative Syndrome Scale; PSP, Personal and Social Performance; SE, standard error.

Data are mean (SE) unless otherwise stated.

^aLSM from MMRM with fixed effects of region (Europe or the rest of the world), planned treatment (pimavanserin or placebo added to ongoing antipsychotic treatment), study visit (weeks 2, 4, 8, 14, 20, 26), treatment-by-visit interaction, baseline score, and baseline-by-visit interaction. An unstructured covariance matrix is used to model the within-subject errors. The denominator degrees of freedom are estimated using the Kenward–Roger approximation.

^bDifference between LSM changes for pimavanserin and placebo (pimavanserin—placebo) at the specified visit from MMRM analysis.

^cTwo-sided P-value for treatment difference at specified visit from MMRM analysis.

^dLSM from MMRM with fixed effects of region (Europe or the rest of the world), planned treatment (pimavanserin or placebo added to ongoing antipsychotic treatment), study visit (weeks 2, 4, 8, 14, 20, 26), treatment-by-visit interaction, baseline CGI-SCH-S score, and baseline-by-visit interaction.

^eAt week 26 the number of participants in the placebo and pimavanserin groups were 181 and 183, respectively.

Secondary Efficacy Outcomes

Pimavanserin did not achieve a statistically significant difference from placebo in any of the secondary outcomes, including the key secondary endpoint of change from baseline to week 26 in the CGI-SCH-S negative symptoms score (MMRM LSM difference: −0.01; SE, 0.09; P = .89; Cohen’s d effect size 0.01; Table 2). None of the changes from baseline to week 26 were significantly different between the treatment groups for any of the other secondary efficacy endpoints, which included the CGI-SCH-I negative symptom score (P = .14), PSP score (P = .44), PANSS total score (P = .19), PANSS negative subscores (P = .56), and PANSS Marder factor (negative symptoms) score (P = .45).

Responder Analysis

On the basis of the responder analysis at week 26, the proportion of NSA-16 responders achieving ≥20% reduction in NSA-16 total score was 107 (47.8%) with pimavanserin vs 104 (46.8%) with placebo (P = .84; Supplementary Table S4). The number of NSA-16 responders achieving ≥30% reductions were 82 (36.6%) with pimavanserin vs 68 (30.6%) with placebo (P = .18). There was also no significant difference between the treatment groups for the CGI-SCH-I negative symptom responders (negative symptom score of 1 or 2), with 70 (31.3%) for pimavanserin and 62 (27.9%) for placebo (P = .44). It should be noted that missing values were imputed as non-responders.

Treatment-Emergent Adverse Events

The incidence of any treatment-emergent adverse event (TEAE) was 30.4% (69/227 patients) with pimavanserin and 40.3% (91/226) with placebo (Table 3). The most common TEAEs (occurring in ≥2% of either treatment group) were headache, somnolence, worsening of schizophrenia, COVID-19, diarrhea, and dizziness. TEAEs determined to be drug-related occurred in 9.7% (22/227) and 11.5% (26/226) of the pimavanserin and placebo groups, respectively. The most common treatment-related TEAE (occurring in ≥2% of either treatment group) was somnolence for the placebo group (6/226; 2.7%), whereas none of the treatment-related TEAEs occurred in ≥2% of patients in the pimavanserin group. The incidence of serious TEAEs was 0.9% (2/227) and 3.1% (7/226) in the pimavanserin and placebo groups, respectively. TEAEs leading to discontinuation occurred in 2.6% (6/227) of patients in the pimavanserin group and 6.2% (14/226) in the placebo group. There were no fatal TEAEs in either group.

Table 3.

Incidence of Treatment-Emergent Adverse Events (Safety Analysis Set)

	Placebo (n = 226)	Pimavanserin (n = 227)
Any TEAE	91 (40.3)	69 (30.4)
Drug-related TEAE	26 (11.5)	22 (9.7)
Serious TEAE	7 (3.1)	2 (0.9)
TEAE leading to death	—	—
Discontinuation due to TEAE	14 (6.2)	6 (2.6)
TEAEs occurring in ≥2% of either treatment group
Headache	16 (7.1)	9 (4.0)
Somnolence	7 (3.1)	5 (2.2)
Worsening of schizophreniaa	7 (3.1)	1 (0.4)
COVID-19	6 (2.7)	2 (0.9)
Diarrhea	1 (0.4)	5 (2.2)
Dizziness	1 (0.4)	5 (2.2)

CGI-SCH-S, Clinical Global Impression–Schizophrenia Scale–Severity; COVID-19, coronavirus disease of 2019; TEAE, treatment-emergent adverse event.

Data are presented as n (%).

^aExacerbation of psychotic symptoms requiring an increase in the level of psychiatric care required by the patient, substantial deterioration on the CGI-SCH-S negative symptom scale (score 6 “much worse” or 7 “very much worse”), or at the investigator’s discretion.

Safety Analysis

Safety measurements were assessed throughout the study to evaluate the safety and tolerability of pimavanserin. No clinically relevant effects were observed for pimavanserin vs placebo for vital signs, weight, and laboratory measures. An ECG analysis also revealed no significant changes with either pimavanserin or placebo. The mean change (SE) from baseline to week 26 was similar between the treatment groups for AIMS (pimavanserin, −0.1 [0.04]; placebo, −0.1 [0.03]), BARS (pimavanserin, 0.0 [0.02]; placebo, 0.0 [0.01]), and SAS (pimavanserin, −0.1 [0.04]; placebo, −0.1 [0.05]).

There was no overall incidence of dyskinesia (based on AIMS). On the basis of overall postbaseline incidence, akathisia (based on BARS-CGA of Akathisia Scores) was seen in 0.9% (2/222) of the pimavanserin and 0.5% (1/222) of the placebo groups, and there were no instances of parkinsonism (based on SAS) in the pimavanserin group and 4.1% (9/222) of patients in the placebo group. The rate of suicidal ideation on C-SSRS was 0.9% (2/226) with placebo and 2.6% (6/227) with pimavanserin (Supplementary Table S7).

Discussion

In this phase 3, randomized, placebo-controlled study for outpatient adults with predominant NSS, neither the primary nor the key secondary endpoint of the change in NSA-16 and CGI-SCH-S, respectively, were met at week 26. Furthermore, pimavanserin did not show significant improvements over placebo for any of the secondary outcomes. The subgroup analysis of the change in NSA-16, which was performed based on baseline antipsychotic medications, yielded similar results, showing no significant differences between the pimavanserin and placebo groups. Results from the responder analysis indicate patients who experienced greater improvements were more likely to be in the pimavanserin group, though this result is not statistically significant given the sample size. Safety outcomes also did not show any significant differences between the treatment groups, but TEAEs occurred more frequently in the placebo group (40.3%) compared with the pimavanserin group (30.4%), with the most common TEAE being headache in both groups.

The recent phase 2 study (ADVANCE-1) comparing pimavanserin with placebo in patients with schizophrenia and predominantly negative symptoms demonstrated clear improvements with pimavanserin over the placebo group (P = .043) with the greatest improvement being in the social involvement domain.³³ For the current phase 3 study (ADVANCE-2), the magnitude of the improvement in the pimavanserin group (−11.80) was larger than that seen in the prior phase 2 ADVANCE study (−10.4); however, the current study demonstrated a larger treatment effect in the placebo group (−11.13) than the prior phase 2 study (−8.5).³³ Considering its magnitude, several external or study design–related factors may have contributed to the large placebo effect. The large number of study sites (82) in this trial may be linked to a higher placebo response risk. A previous meta-analysis of NSS studies demonstrated that more study sites were associated with a greater placebo response due to greater variability in study site procedures and the inclusion of a more heterogeneous study population.⁴¹ Future studies should aim to recruit more patients from fewer—but high-quality—study sites, which may lead to increased homogeneity in approaches to care for patients with NSS. Additionally, a successful phase 2 study may have increased patients’ expectations in this trial of the first potential NSS-indicated drug.

A prominent factor that likely contributed to the large placebo effect is the COVID-19 pandemic. Most of this trial took place during the COVID-19 pandemic, which started in late 2019 and ended in May 2023; the first patient was randomized in August 2020, and the study was completed in January 2024. The pandemic was associated with extensive measures to reduce the spread of the disease, such as social distancing, limited in-person interactions, and reduced access to pleasurable recreational activities. To explore the effect of the COVID-19 pandemic in the context of the negative study outcome, quarterly randomization data starting in August 2020 were analyzed. During a period coinciding with strict lockdowns in the actively recruiting European countries (November 1, 2020–January 31, 2021), 18 patients were randomized to pimavanserin and 20 to placebo, and the treatment effect favored placebo (6.74) with a negative effect size (−0.691). A survey of patients with psychiatric illnesses (approximately half of whom had schizophrenia) during the COVID-19 pandemic reported that 51% of patients had worsening psychiatric symptoms, such as tiredness, hopelessness, depression, “no cheer,” or worthlessness.²² COVID-19 preventative measures by their very nature induce behavioral factors associated with negative symptoms and likely led to patients with schizophrenia experiencing significantly higher negative symptoms during the pandemic than before the pandemic.²³ As such, many of the patients in the current phase 3 study had higher baseline severity of negative symptoms and lower functioning compared with patients in the prior phase 2 ADVANCE-1 study.³³ Studies have also reported that due to heightened symptom burdens associated with the COVID-19 pandemic, patient interactions with clinical staff were intensified, as was the case in ADVANCE-2, leading to elevation in mood and reduced severity of NSS, which may further explain the substantial placebo effect observed.⁴²

Although the current study did not demonstrate statistical separation between pimavanserin and placebo, slightly greater improvements with pimavanserin were observed in 4 of the 5 NSA-16 domain scores (communication, emotion or affect, motivation, and retardation) relative to the placebo group.³³ Taken together, it is likely that a larger placebo effect and a lack of an effect on the NSA-16 social involvement domain both contributed to the lack of a significant effect in this study. The improvements observed in 4 of the 5 NSA-16 domain scores (communication, emotion or affect, motivation, and retardation) with pimavanserin vs placebo, albeit not statistically significant, may also be related to the COVID-19 pandemic. This observation is noteworthy because in the prior phase 2 ADVANCE-1 study, the pimavanserin group demonstrated the largest significant difference in the social involvement domain of the NSA-16 (LSM difference: −0.6; SE, 0.24; [95% CI: −1.1, −0.1]; P = .01; Cohen’s d effect size  0.27).³³ Taken together, it is plausible that the factors described above attenuated the beneficial effects of pimavanserin on negative symptoms in this trial because those benefits have previously been demonstrated to be driven largely by improvements in social interactions.

Finally, it is possible that this effect may not have been as apparent in the pimavanserin group vs the placebo group due to a ceiling effect on improvement in negative symptoms with the background antipsychotic treatment. Several antipsychotics have shown some benefit in treating NSS.⁴³ In the current ADVANCE-2 study, on average, patients had been stable on their background antipsychotic for 28 months before enrollment. Therefore, it is plausible that the ceiling effect on negative symptoms from ongoing atypical antipsychotics with varying degrees of 5-HT_2A antagonism had already been reached after 28 months. Indeed, patients receiving risperidone and olanzapine (58.4% of participants) may have already reached near-complete 5-HT_2A occupancy and any significant additional effect may not have been feasible.^29–31 Additional reductions in negative symptoms could be attributed to pimavanserin’s 5-HT_2A and 5-HT_2C antagonism/inverse agonism and its effects on neocortical dopamine release.⁴⁴ However, given the already-reached ceiling, further benefits from pimavanserin may not be large enough to show a significant difference between treatment groups.

A post hoc analysis of the ADVANCE-1 and ENHANCE studies concluded there were high levels of adherence to both background antipsychotics and pimavanserin in both studies.³⁸ Adherence in both ADVANCE-1 and ADVANCE-2 was ensured through mandatory caregiver participation and rigorous screening prior to enrollment that excluded patients not adherent to their background antipsychotic medication. Therefore, with the high overall compliance to pimavanserin and the main antipsychotic in this study (ADVANCE-2), patient characteristics or adherence changes are unlikely to have contributed to this study’s failure to reach its primary endpoint.

Recent studies with alternative therapies for NSS have not yielded positive results. A phase 3 trial of roluperidone, a 5-HT_2A, sigma-2, and α_1A-adrenergic receptor antagonist, failed to meet the primary outcome.^15,16 Another phase 3b, 26-week study of the atypical antipsychotic cariprazine, which has partial activity at the 5-HT_1A receptor, reported greater efficacy with cariprazine vs risperidone in adults with stable schizophrenia and predominantly negative symptoms; however, the study was not placebo controlled.¹⁷ Furthermore, NMDA-targeting treatments, including memantine, the d-amino acid oxidase inhibitor luvadaxistat, and the glycine reuptake inhibitor (BI 425809) all failed to demonstrate efficacy for NSS.¹⁰ While there are currently ongoing trials that may still prove effective in treating NSS,^45–47 the repeated lack of success with pharmacologic treatments may be due to the nature of NSS being a uniquely challenging clinical problem to solve.

The current study did illustrate the positive safety and tolerability profile of pimavanserin. The most common TEAEs observed with pimavanserin were headache (4.0%), somnolence (2.2%), dizziness (2.2%), and diarrhea (2.2%). These TEAEs led to only 2.6% of patients in the pimavanserin group discontinuing the study drug. The safety measures were similar among the treatment groups, which included vital signs, body weight, and extrapyramidal side effects as measured by the AIMS, BARS, and SAS scales. Electroencephalography analyses also revealed no significant differences between the treatment groups.

There were numerous strengths and some limitations of the present study. The selection of patients with predominant NSS allowed this study to demonstrate a greater functional change in patients with more severe negative symptoms.⁴⁸ The independent interview confirmed patients had predominant NSS and excluded treatment-resistant patients. The study also achieved a balanced representation of sex resulting in the inclusion of a large number of female patients with NSS for both the pimavanserin (41.9%) and placebo (39.8%) groups. Different background antipsychotics allowed in the inclusion criteria offered insights on the effects of pimavanserin across a range of ongoing antipsychotics; furthermore, the use of regionally approved dosing allows generalizability to the broader clinical practice setting.

The current study also demonstrated a high quality of data as evidenced by the high rates of treatment adherence and low variation in the outcome measures, over 26 weeks of study duration. As prior studies have shown, it is difficult to achieve high treatment adherence in patients with schizophrenia, especially those with predominant NSS.⁴⁹ This study achieved high levels of adherence by implementing multiple strategies, including requiring patients to have a reliable caregiver to participate, pill counting, and the use of pharmacokinetic measures to objectively confirm adherence. The requirement for a caregiver, in addition to contributing to the high rates of adherence to pimavanserin and background antipsychotic treatments, likely improved rating. More importantly, despite recruiting from a large number of study sites, the current study had a remarkably low amount of variation in the NSA-16 primary outcome measure over 26 weeks. The extended trial period of 26 weeks was also a strength, as it allowed sufficient time to evaluate the efficacy of the treatment on NSS including improvements in functioning, as well as the safety of background antipsychotics in combination with pimavanserin.

Despite many strengths, the study has some limitations. Patients from different cultures may express symptoms differently. Although the aim was to include a balanced representation of race, sex, and region in the study population, results were limited in generalizability due to the under-representation of non-White races and patients from North America, as the study was not conducted in the United States. However, the inclusion of patients from Argentina increased the representation of patients of Hispanic/Latino ethnicity. Furthermore, although a variety of background antipsychotics were allowed, some were not well represented because of scarce regional availability.

The pimavanserin and placebo groups did not differ significantly on the primary efficacy endpoint of the change in the NSA-16 total score from baseline to week 26. Also, the treatment groups did not show any significant differences in the key secondary outcome of the change in CGI-SCH-S from baseline to week 26 or the other secondary outcomes. Safety findings were largely similar between the 2 groups, and there were no new safety signals with pimavanserin. This phase 3 study did replicate the magnitude of the effect from pimavanserin in the prior phase 2 ADVANCE study; unfortunately, this study’s placebo effect was greater than in the prior. The increased placebo response may be attributed to the anticipation generated by a successful phase 2 study for the first potential NSS-indicated drug as well as the concurrent impact of the COVID-19 pandemic. Although this study’s recruitment was largely completed by May 2022 (61%), the war between Russia and Ukraine may have introduced challenges that were difficult to quantify, including limitations to social interactions and other stressors, potentially influencing patient outcomes in ways that extend beyond the scope of this study. Future studies may incorporate new tools, such as wearable technology or modern biomarkers such as speech latency measurements, which can help exclude high placebo responders from trials.

Supplementary material

Supplementary material is available at https://academic.oup.com/schizophreniabulletin.

Funding

This study was funded by Acadia Pharmaceuticals Inc. The funder designed the study with collaboration from investigators and had a role in data analysis, data interpretation, and writing of the report. Dr Arango is supported by the Spanish Ministry of Science and Innovation, Instituto de Salud Carlos III (ISCIII), co-financed by the European Union, ERDF Funds from the European Commission, “A way of making Europe,” financed by the European Union—NextGenerationEU (PMP21/00051), PI19/01024. CIBERSAM, Madrid Regional Government (B2017/BMD-3740 AGES-CM-2), European Union Structural Funds, European Union Seventh Framework Program, European Union H2020 Program under the Innovative Medicines Initiative 2 Joint Undertaking: Project PRISM-2 (Grant agreement No.101034377), Project AIMS-2-TRIALS (Grant agreement No 777394), Horizon Europe, the National Institute of Mental Health of the National Institutes of Health under Award Number 1U01MH124639-01 (Project ProNET) and Award Number 5P50MH115846-03 (project FEP-CAUSAL), Fundación Familia Alonso, and Fundación Alicia Koplowitz.

Conflict of Interest

DB-K, I-YL are employees of Acadia Pharmaceuticals and have stock and stock options in Acadia Pharmaceuticals. CA has been a consultant to, provided expert testimony for, or received honoraria or grants from Abbott, Acadia, Angelini, BMS, Boehringer, Gedeon Richter, Janssen Cilag, Johnson and Johnson, Lundbeck, Medscape, Minerva, Otsuka, Sage, Schering Plough, Sumitomo Dainippon Pharma, and Sunovion and has acted in a leadership or fiduciary role for Schizophrenia International Research Society, European College of Neuropsychopharmacology, and Spanish Society of Psychiatry. SRM has served on advisory boards for Roche, Boehringer-Ingelheim, Merck, Sunovion, Otsuka, Neurocrine, Sunovion, Newron, and Biogen.