Esketamine for treatment-resistant depression: Comparing two maintenance schedules in an observational real-world study

Technical Summary: Esketamine for Treatment-Resistant Depression: Comparing Two Maintenance Schedules in an Observational Real-World Study

Problem Statement
Treatment-resistant depression (TRD) imposes a substantial clinical and socioeconomic burden, characterized by high hospitalization rates, elevated suicide risk, and functional impairment. While intranasal esketamine has demonstrated efficacy in clinical trials and real-world settings for inducing rapid symptom reduction, evidence regarding the optimal maintenance scheduling remains limited. Current label recommendations suggest individualizing maintenance dosing to the lowest frequency that maintains response, yet real-world practices vary significantly between "front-loaded" strategies (concentrating sessions early) and "spaced" strategies (progressively extending intervals). There is a lack of systematic comparison between these common maintenance strategies to determine if dosing frequency impacts long-term depressive symptom trajectories. Furthermore, prior observational studies have often relied on either clinician-rated or self-report instruments exclusively, rather than integrating both perspectives within the same cohort.

Methodology
This study was a single-centre, retrospective, observational analysis conducted at the University Hospital of Psychiatry Zurich, Switzerland, involving 65 adults with TRD treated between January 2021 and April 2025.

• Participants: Patients met ICD-10 criteria for a depressive episode and had failed at least two adequate antidepressant trials plus augmentation therapy. Exclusion criteria included uncontrolled hypertension, history of stroke/aneurysm, intracranial hemorrhage, and pregnancy.
• Treatment Protocol: All patients underwent an induction phase of eight intranasal esketamine sessions over four weeks (56 mg for the first session, 84 mg thereafter). Patients showing clinical benefit (CGI-I reduction ≥2 points) proceeded to maintenance.
• Maintenance Schedules: Allocation to maintenance was based on clinical availability rather than randomization.
  • Schedule 1 (MS1): Four additional sessions administered once weekly (sessions 9–12).
  • Schedule 2 (MS2): Four additional sessions administered at progressively increasing intervals (1, 2, 3, and 6 weeks), spanning a 12-week period.
• Measures: Depressive symptoms were assessed using the clinician-rated Montgomery-Åsberg Depression Rating Scale (MADRS) and the self-rated Beck Depression Inventory-II (BDI) immediately prior to each administration. Response was defined as ≥50% reduction from baseline; remission was defined as MADRS ≤12 or BDI ≤13.
• Statistical Analysis: Analyses were exploratory. Induction phase changes were assessed via paired t-tests. The maintenance phase utilized linear mixed-effects models to compare schedules across time points (T8–T11), adjusting for patients as random intercepts. Sensitivity analyses included complete-case models and ANCOVA comparing end-of-treatment scores adjusted for baseline.

Key Contributions

• Dual-Perspective Assessment: The study integrates both clinician-rated (MADRS) and patient-reported (BDI) outcomes within a single real-world cohort, addressing a gap in prior literature.
• Real-World Maintenance Comparison: It provides one of the few direct comparisons between two distinct maintenance scheduling strategies (front-loaded vs. spaced) in a routine clinical setting, moving beyond the controlled conditions of randomized trials.
• Longitudinal Trajectory Analysis: The study tracks symptom trajectories through a 16-week observation period, capturing the dynamic nature of response, remission, and relapse during the maintenance phase.

Results

• Induction Phase: Depressive symptoms decreased significantly during the induction phase on both scales. The mean MADRS score dropped from 25.46 to 17.72 (Cohen's d = 1.03), and the mean BDI score dropped from 30.97 to 23.60 (Cohen's d = 0.72). Response rates at the end of induction were 21.1% for both scales, with remission rates of 22.8% (MADRS) and 21.1% (BDI).
• Maintenance Phase: No statistically significant differences were found between MS1 and MS2 regarding depressive symptom trajectories on either the MADRS or BDI.
  • MADRS: Main effects of time point (p = .711), schedule (p = .593), and interaction (p = .098) were non-significant.
  • BDI: Main effects of time point (p = .443), schedule (p = .954), and interaction (p = .184) were non-significant.
  • Outcomes: Among treatment completers (n=51), response rates increased to 35.3% (MADRS) and 26.0% (BDI), while remission rates reached 31.4% (MADRS) and 38.0% (BDI).
• Attrition: Attrition during the 12-week maintenance phase was low (7/58 patients) with no statistically significant difference between schedules (7.7% for MS1 vs. 21.1% for MS2; p = .20), though the relative risk for dropout in MS2 was numerically higher (2.74).
• Trajectory Patterns: Symptom improvement was most pronounced early in treatment. BDI scores plateaued around session five, whereas MADRS scores continued to decline, suggesting clinician ratings may be more sensitive to gradual changes.

Significance and Claims
The authors conclude that esketamine is associated with significant reductions in depressive symptoms during the induction phase in a real-world TRD population. Crucially, the study found no evidence that the timing of maintenance dosing intervals (weekly vs. spaced) significantly alters depressive symptom trajectories.

The paper posits that these findings suggest maintenance dosing intervals may be scheduled flexibly to align with patient preferences, clinical needs, and resource constraints without a loss of efficacy. This supports current routine practices where dosing intervals are progressively lengthened as patients stabilize. The authors emphasize that esketamine should not be discontinued prematurely in partial responders, as response and remission rates continued to increase throughout the maintenance period.

However, the authors maintain a modest tone regarding their claims. They explicitly state that due to the retrospective, observational design, the absence of randomization, the small sample size (particularly in the MS2 group), and the lack of a priori power analysis, the results should be interpreted with considerable caution. They clarify that the absence of significant differences should not be interpreted as definitive evidence of equivalence. The study is described as "hypothesis-generating," and the authors call for prospective, randomized trials with larger sample sizes and longer follow-up to confirm these findings and establish optimal maintenance strategies.