This post hoc analysis included pooled data from three similarly designed, randomized, double-blind, placebo-controlled, Phase 3 trials of adjunctive brexpiprazole in adults with MDD and inadequate response to ADTs: Pyxis (Trial 228; ClinicalTrials.gov identifier: NCT01360645) (25), Polaris (Trial 227; NCT01360632) (26), and Sirius (Trial 214; NCT02196506) (27). All three trials were conducted in compliance with the International Council for Harmonisation Good Clinical Practice guideline and local regulatory requirements, and with the principles laid out in the Declaration of Helsinki. The protocols were approved by independent ethics committees, and all patients provided written informed consent to participate.

Detailed trial designs have been previously published (25–27). In brief, the studies enrolled outpatients aged 18–65 years with MDD as per the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision (DSM-IV-TR) criteria (28); a current depressive episode of ≥8 weeks; an inadequate response (defined as <50% improved) to 1–3 prior ADTs during the current episode; and a Hamilton Depression Rating Scale (HAM-D₁₇) (29, 30) Total score of ≥18. Exclusion criteria included suicidal ideation or behavior, substance abuse or dependence, and specified DSM-IV-TR comorbidities. Comorbid DSM-IV-TR sleep disorders were not exclusionary.

In the trials, eligible patients received single-blind placebo together with an investigator-determined, open-label ADT (sertraline, escitalopram, duloxetine, fluoxetine, paroxetine controlled-release, or venlafaxine extended-release) for 8 weeks. The purpose of the 8-week prospective treatment phase was to identify patients with inadequate response to an additional ADT. Inadequate response to ADT was defined as <50% reduction in HAM-D₁₇ Total score from the start to the end of the prospective treatment phase, HAM-D₁₇ Total score of ≥14 at the end of the prospective treatment phase, <50% reduction in Montgomery–Åsberg Depression Rating Scale (MADRS) (31) Total score from the start of the prospective treatment phase to each scheduled visit, and a Clinical Global Impression – Improvement (CGI-I) (32) score of ≥3 at each scheduled visit during the prospective treatment phase.

Patients with an inadequate response to ADT were randomized to 6 weeks of double-blind treatment with fixed-dose brexpiprazole (1, 2, or 3 mg/day, depending on the study) or placebo, adjunctive to their continued ADT. Brexpiprazole was initiated at a dose of 0.5 mg/day, titrated to 1 mg/day after 1 week, and titrated to the allocated dose after 2 weeks. Concomitant benzodiazepines and non-benzodiazepine sleep aids were prohibited except for the short-term management of treatment-emergent agitation/anxiety and insomnia, respectively; these drugs could not be taken in the 12 hours before a scheduled efficacy or safety assessment.

In this post hoc analysis, sleep disturbance was measured by the Sleep Disturbance Factor (SDF), which is the sum of three HAM-D₁₇ item scores: item 4 “insomnia – early”, item 5 “insomnia – middle”, and item 6 “insomnia – late” (33, 34). Each item is rated on a 3-point scale: 0 (absent), 1 (occasional), and 2 (frequent). Scores for the three insomnia items are summed to form the SDF score, which ranges from 0 (no sleep disturbance) to 6 (maximum sleep disturbance) (29, 30, 34). Patients were categorized by baseline (the randomization visit, prior to dosing) level of sleep disturbance, defined as in previous literature as high (SDF score ≥4), or low (SDF score <4) (34). Patients with an SDF score of 0 (indicating no sleep disturbance) were also included in the low SDF subgroup.

The primary efficacy endpoint in each of the three trials was change in MADRS (31) Total score from baseline to Week 6 of the randomized treatment phase. The MADRS, a measure of depression severity, was administered at baseline and at weekly intervals throughout the randomized phase, and was the main depression outcome of this post hoc analysis.

Change in depression severity was also assessed using the clinician-reported HAM-D₁₇ (29, 30) Total score and Clinical Global Impression – Severity (CGI-S) (32) score. MADRS response rates (defined as a ≥50% reduction in MADRS Total score from baseline to Week 6) were also assessed. Change in functioning was assessed using the patient-reported Sheehan Disability Scale (SDS) (35, 36) score. In this post hoc analysis, change in sleep disturbance was assessed using the SDF score, and the MADRS “reduced sleep” item score. Rating scales were administered by trained and experienced clinicians, who were certified for the trials to administer the MADRS and HAM-D₁₇. The number of raters within each trial center was kept to a minimum.

In this post hoc analysis, safety was assessed by the incidence of treatment-emergent adverse events (TEAEs), with a focus on sedating TEAEs (somnolence, fatigue, sedation, lethargy) and activating TEAEs (akathisia, restlessness, insomnia, initial insomnia, middle insomnia, terminal insomnia).

In this post hoc analysis, data were pooled for adjunctive brexpiprazole doses of 2 or 3 mg/day, reflecting the recommended-to-maximum brexpiprazole doses for the adjunctive treatment of MDD in the United States (12). Separately, data were pooled for adjunctive placebo.

Efficacy analyses were conducted for all patients randomized per final protocols who received at least one dose of double-blind medication and had both a baseline and at least one post-baseline MADRS Total evaluation in the randomized treatment phase. Safety analyses were conducted for all patients who received at least one dose of double-blind medication in the randomized treatment phase. Patients without a baseline HAM-D₁₇ measurement could not be categorized into SDF subgroups, and were therefore excluded. Change in MADRS Total score, change in HAM-D₁₇ Total score, change in CGI-S score, MADRS response rates, and change in SDS score were assessed in the high SDF and low SDF subgroups. Change in sleep endpoints – SDF score, and MADRS “reduced sleep” item score – were assessed in the high SDF subgroup, only.

Patient baseline demographic and clinical characteristics were summarized using descriptive statistics. For MADRS Total score, CGI-S score, SDS score, and MADRS “reduced sleep” item score, least squares (LS) mean changes from baseline were calculated using a mixed model for repeated measures (MMRM) method with model terms of study (to account for potential heterogeneity across studies), treatment, visit, treatment-by-visit and baseline-by-visit interaction. An unstructured covariance was used by default; normality and other covariance structures were examined by fitting the MMRM with alternative assumptions (such as t-distributed residuals/random effect, homogeneity or heterogeneity of variance, or autocorrelation over visits). For HAM-D₁₇ Total score and SDF score, LS mean changes from baseline were calculated using an analysis of covariance (ANCOVA) model on the last observation carried forward (LOCF) dataset, with treatment and study center as the main effects and baseline value as the covariate. For MADRS response rates, the Cochran–Mantel–Haenszel association test, controlling for study site, was conducted using LOCF. All p-values were tested at a nominal 0.05 level (two-sided) with no adjustment for multiplicity. Cohen’s d effect sizes (37) were also calculated. The incidence of TEAEs were summarized using descriptive statistics.

Analyses were performed using SAS version 9.4 (SAS Institute Inc; Cary, NC).

Data were analyzed for a total of 1,160 patients (efficacy and safety samples), of whom 689 (59.4%) were in the high SDF subgroup (ADT + brexpiprazole 2 or 3 mg/day, n=348; ADT + placebo, n=341) and 471 (40.6%) were in the low SDF subgroup (ADT + brexpiprazole 2 or 3 mg/day, n=229; ADT + placebo, n=242).

Baseline demographics and clinical characteristics were similar across the three trials (25–27). Within each pooled subgroup, baseline demographic and clinical characteristics were generally similar between ADT + brexpiprazole and ADT + placebo treatment groups ( Table 1 ). Baseline depression severity was higher in the high SDF subgroup than in the low SDF subgroup, but was similar between treatment arms within each subgroup.

On depression outcomes, the LS mean change from baseline to Week 6 in MADRS Total score was greater with ADT + brexpiprazole than ADT + placebo (p<0.0001) in the high SDF subgroup. Greater improvement between treatment groups was observed from Week 2 (p<0.01) onwards ( Table 2 , Figure 1A ). In the low SDF subgroup, greater improvement with ADT + brexpiprazole versus ADT + placebo was observed at Week 6 (p=0.0058) and at some earlier visits ( Table 2 , Figure 1B ). LS mean changes from baseline to Week 6 in HAM-D₁₇ Total score and CGI-S score were also greater with ADT + brexpiprazole versus ADT + placebo (p<0.05) in the high SDF subgroup and in the low SDF subgroup ( Table 2 ). In the high SDF subgroup, MADRS response rates at Week 6 were 27.0% (94/348) with ADT + brexpiprazole and 20.8% (71/341) with ADT + placebo (p=0.032). In the low SDF subgroup, MADRS response rates were 29.3% (67/229) with ADT + brexpiprazole and 21.5% (52/242) with ADT + placebo (p=0.052).

Regarding functioning, the LS mean change from baseline to Week 6 in SDS score was greater with ADT + brexpiprazole versus ADT + placebo (p<0.05) in the high SDF subgroup and in the low SDF subgroup ( Table 2 ).

In the high SDF subgroup, on sleep outcomes, the LS mean change from baseline to Week 6 in SDF score and in MADRS “reduced sleep” item score was greater with ADT + brexpiprazole versus ADT + placebo (p<0.05 for both measures) ( Table 2 ).

The overall incidence of TEAEs was higher with ADT + brexpiprazole than ADT + placebo in the high SDF subgroup (59.8% compared with 51.6%) and in the low SDF subgroup (62.4% compared with 40.9%) ( Table 3 ).

In the high SDF subgroup, the TEAEs with an incidence ≥5% in the ADT + brexpiprazole group and greater than ADT + placebo were akathisia, weight increase, and restlessness ( Table 3 ). In the low SDF subgroup, the TEAEs with an incidence ≥5% in the ADT + brexpiprazole group and greater than ADT + placebo were akathisia, restlessness, and somnolence ( Table 3 ).