Esketamine Treatment for Depression in Adults: A PRISMA Systematic Review and Meta-Analysis

Earlier research establishes that conventional antidepressants have proven efficacy but a substantial proportion of patients fail to achieve sustained remission, with treatment-resistant depression (TRD) commonly defined as nonresponse to two adequate antidepressant trials. Interest in rapid-acting agents has grown because of this unmet need; ketamine and its enantiomers are central to that work. Esketamine, the S-enantiomer of ketamine, has higher affinity for NMDA (and to some extent mu-opioid) receptors and was approved by the U.S. Food and Drug Administration in 2019 as an intranasal adjunctive therapy for TRD, with a subsequent indication for major depression with acute suicidal ideation and behaviour. Despite many publications, uncertainty remains about the durability of esketamine’s antidepressant effect, its true antisuicidal efficacy, and how its benefit compares with other augmentation strategies. Fountoulakis and colleagues set out to synthesise the entire esketamine literature to clarify these questions. Using a PICO framework focused on adults with depression receiving esketamine versus placebo, the investigators aimed to assess effects on depressive symptoms and suicidality, and to examine safety signals and longer-term outcomes. The study therefore undertook a broad systematic review and meta-analysis of randomized controlled trials (RCTs) and related literature to evaluate efficacy at multiple clinically relevant time points and to explore sensitivity analyses by TRD status and dose arms.

The authors performed an over-inclusive MEDLINE search on March 24, 2024, using the single keyword "esketamine" and followed PRISMA procedures for study selection; the review was registered in PROSPERO (CRD42024516053). Eligible materials comprised RCTs, case reports/series, retrospective reviews, open trials, post hoc analyses, systematic reviews and meta-analyses addressing depression and suicidality in adults; only English-language reports were considered. Two authors carried out screening and data extraction, and where necessary values were digitised from figures. Quantitative synthesis and analyses were carried out in R (v4.3.3) using the METAFOR package. The meta-analysis used both common-effect and random-effects models, applying the inverse-variance method with a restricted maximum-likelihood estimator for between-study variance (tau-squared) and Hedges' g as the effect measure (standardised mean difference, bias-corrected). Heterogeneity was assessed with I2 and the Q test. A trial was classified as "positive" at a given time point when a statistically significant outcome difference was reported; trials with very high placebo response and nonsignificant results were termed "failed." Pre-specified outcomes examined were change in depressive symptoms measured by MADRS at four time points (up to 72 hours, and at 1, 2 and 4 weeks) and suicidality measures at two time points (days 2–5 and week 4). Because published data were generally reported as mixed-effects models with repeated measures (MMRM), the meta-analysis used these MMRM-derived estimates rather than intention-to-treat with last-observation-carried-forward. Sensitivity analyses included separating TRD and non-TRD RCTs and excluding 28-mg arms (labelled subtherapeutic) to test robustness.

The MEDLINE search yielded 1,115 records; after screening 87 articles were included for review. For quantitative pooling, 11 placebo-controlled RCTs of intranasal esketamine were identified, totalling 1,774 participants (979 allocated to esketamine, 795 to placebo). Most RCTs enrolled adults with unipolar TRD and delivered esketamine as an adjunct to ongoing or newly initiated antidepressants; one trial used intravenous administration and two trials included patients with active suicidal ideation. Acute-phase trials (n=10) showed a consistent early advantage for esketamine: statistically significant benefit was observed at 2–4 hours, at 24 hours, and at 48–72 hours, and possibly at day 4. However, efficacy was inconsistent from day 8 onward and most trials were negative at week 4; specifically, at week 2 four trials were negative, and at week 4 five of six trials were negative despite continued twice-weekly dosing. The lone trial in elderly patients (n=138) found no superiority at day 28. The authors noted examples where placebo response rates rose over time while the esketamine arm remained stable, suggesting the between-arm difference narrowed because placebo+TAU caught up. A single randomized withdrawal (maintenance) trial, SUSTAIN-1 (n=297), evaluated relapse after continued esketamine versus discontinuation. Time-to-relapse favoured continued esketamine: among remitters relapse occurred in 26.7% (esketamine+TAU) versus 45.3% (placebo+TAU) (p=0.003; NNT=6), and among responders relapse was 25.8% versus 57.6% (p<0.001; NNT=4). Hazard ratios were 0.49 (95% CI=0.29, 0.84) for remission and 0.30 (95% CI=0.16, 0.55) for response. The FDA raised concern that these results were driven by a single Polish site with an anomalously high relapse rate in the placebo arm; exclusion of that site abolished statistical significance. Head-to-head comparisons (n=4) reported that a single intravenous infusion of esketamine and racemic ketamine produced similar antidepressant and antisuicidal effects (small, underpowered trials), while an open-label phase 3b trial (ESCAPE-TRD, n=676) found higher remission at week 8 with esketamine plus SSRI/SNRI than with add-on quetiapine (27.1% vs. 17.6%; p=0.003). Numerous post hoc analyses (n=36) and pooled re-analyses produced heterogeneous findings: some identified early-onset benefits for particular symptom clusters (anhedonia, hopelessness, fatigue) or faster time to remission, others found no relationship between dissociation and antidepressant response, and some suggested benefits in specific subgroups (older adults, employed patients). NNT calculations based on pooled MADRS response/remission yielded values such as NNT=8 for response and NNT=6 for remission in some analyses; maintenance NNTs for relapse prevention were <10 in pooled reports. Systematic reviews and prior meta-analyses (n≈34) reached mixed conclusions. Several meta-analyses reported a small but statistically significant antidepressant effect at early time points and at 3–4 weeks, while others emphasised low quality of underlying data and inconsistent long-term findings. For suicidality, meta-analyses and pooled analyses generally found an early effect up to 4–6 or 72 hours in some reports but no reliable benefit at later time points. Safety and adverse events: the review highlighted six deaths occurring in RCTs, all in esketamine arms, including three suicides at 4, 12 and 20 days after the last dose; other deaths were a motor vehicle accident, acute respiratory/cardiac failure, and myocardial infarction, each occurring within days of dosing in patients with risk factors. Short-term worsening of mood or emergent suicidal ideation was reported in small proportions: in RCTs six participants (1.4%) on esketamine versus one (0.2%) on placebo became more depressed; five (0.9%) versus two (0.5%) expressed increased suicidal ideas. Nonfatal serious adverse events such as cerebral haemorrhage and motor vehicle accidents were reported only in esketamine arms. Meta-analytic results reported by the authors showed small but statistically significant pooled effects on depressive symptoms: Hedges' g ≈0.33 at 24 hours, 0.25 at week 1, 0.15 at week 2, and 0.23 at week 4 using random-effects models. For suicidality, pooled effect sizes were small and not statistically significant (≈0.10 for days 2–5 and ≈0.04 at week 4). Heterogeneity across analyses was variable. Sensitivity analyses separating TRD and non-TRD samples, and excluding 28-mg arms, produced similar effect sizes and did not change conclusions.

Fountoulakis and colleagues interpret the evidence as showing a reproducible early antidepressant signal for intranasal esketamine that largely attenuates over weeks, yielding at best a modest benefit at clinically standard endpoints. The meta-analysis quantifies this as a small effect size comparable to augmentation with atypical antipsychotics rather than a robust, durable antidepressant effect. The investigators further conclude that the totality of RCTs and pooled data do not support a reliable antisuicidal effect beyond the very early hours after dosing, calling into question the rationale for the regulatory second indication related to suicidality. The authors discuss possible explanations for the observed pattern: an authentic short-lived pharmacodynamic effect that diminishes with time, substantial placebo trajectories in adjunctive designs, functional unblinding because of conspicuous adverse effects (e.g., dissociation) inflating early estimates, and the trial design choice to initiate a new antidepressant at randomization which might have increased expectancy or influenced placebo response. They also raise the concern that the positive maintenance finding could represent a withdrawal or discontinuation effect rather than genuine prophylaxis, noting that almost half of relapses occurred within the first 4 weeks after stopping esketamine and that the maintenance signal depended on an outlier site. Safety and reporting limitations are emphasised: the cluster of deaths in esketamine arms, instances of emergent worsening in a small subgroup, heterogeneous and sometimes inconsistent reporting across regulatory and published documents, and the absence of controlled long-term data create substantial uncertainty about longer-term benefit-risk balance. The authors note the unresolved questions about potential dependence, cognitive effects with prolonged or high-dose exposure, and the need to characterise who may be harmed. Finally, they recommend further patient-level data analyses and trials designed to answer whether esketamine yields true antidepressant effects on core mood/anhedonia items, whether a withdrawal phenomenon explains relapse after cessation, and how esketamine compares with other augmentation options when long-term safety and cost are considered.

The systematic review and meta-analysis did not find robust evidence for clinically meaningful, durable efficacy of add-on intranasal esketamine in depression or for a sustained antisuicidal effect. A small early benefit for depressive symptoms is evident, but effects at week 4 and beyond are modest and similar in magnitude to other augmentation strategies with newer antipsychotics. The authors call for further analyses of patient-level trial data and well-designed studies to determine whether esketamine confers true antidepressant effects on core symptoms, whether relapse after discontinuation reflects withdrawal, and what its comparative value is given safety, cost, and longer-term outcomes.