Meta-analysis of short- and mid-term efficacy of ketamine in unipolar and bipolar depression

Major depression is common, disabling and often slow to respond to conventional monoaminergic antidepressants, prompting interest in treatments that act more rapidly. Ketamine, an N-methyl-D-aspartate (NMDA) receptor antagonist that modulates the glutamatergic system, has been reported to produce marked antidepressant effects within hours of a single dose. However, the clinical literature is heterogeneous with small samples, mixed diagnoses (unipolar versus bipolar), differing outcome measures and variable follow-up times, so the time-course and reliability of ketamine’s antidepressant action remained unclear. Romeo and colleagues therefore carried out a meta-analysis to provide a quantitative, time-resolved estimate of ketamine’s efficacy in treatment‑resistant depression. The study aimed to compare ketamine with placebo on depression scores at prespecified time points (day 1, day 2, day 3–4, day 7 and day 14) and to explore whether demographic or clinical sample characteristics (age, sex, illness duration, episode duration, comorbidities, prior treatment) or route of administration influenced effect sizes. This work sought to address gaps in prior meta-analyses, particularly the lack of detailed time-course analyses based on continuous depression scores.

The investigators searched MEDLINE and PsycINFO through December 2013 without year limits, using ‘‘ketamine’’ plus a range of mood disorder and depression-related terms. Inclusion criteria required studies to be published in English in peer‑reviewed journals, to be randomized, double‑blind and placebo‑controlled trials of ketamine, and to include patients meeting DSM criteria for a major depressive episode (DSM‑III, IV or V). Trials with active comparators (for example midazolam) or electroconvulsive therapy (ECT) were excluded. Study selection was performed by one author and verified by another, and the teams of included trials were contacted to obtain original data. Data extraction compiled means and standard deviations of depression scores (Hamilton Depression Rating Scale, Montgomery–Åsberg Depression Rating Scale, Beck Depression Inventory) and Brief Psychiatric Rating Scale (BPRS) scores together with demographic and clinical variables: age, percent female, years of illness, duration of current episode, and rates of comorbid anxiety, substance and alcohol use disorders. RevMan 5.3 was used for analyses. Effect sizes were calculated as standardized mean differences (SMD; Cohen’s d: difference in means divided by pooled SD) between ketamine and placebo at baseline, day 1, day 2, day 3–4, day 7 and day 14. All meta-analyses used a random‑effects model to account for within‑ and between‑study variability. Heterogeneity was assessed with the Q statistic and I2 (with 25%, 50% and 75% indicating low, moderate and high heterogeneity). Leave‑one‑out sensitivity analyses examined influence of single studies. Secondary analyses compared results by primary diagnosis (bipolar versus recurrent unipolar depression) and by route of administration (intravenous versus intranasal), and funnel plots were planned when at least five studies contributed. Meta-regressions (simple linear models) tested whether clinical variables explained significant SMDs or heterogeneity, but were only performed when SMDs or heterogeneity were significant.

Six randomized, double‑blind, placebo‑controlled crossover trials met inclusion criteria and provided primary data. The six studies comprised 110 patients enrolled across trials; 103 patients were included in the analyses reported by the original studies and thus in this meta-analysis. Diagnostic breakdown across the pooled sample was 12 patients with a first depressive episode, 64 with recurrent unipolar depressive disorder and 34 with bipolar depression. Most studies enrolled treatment‑resistant patients except one (n = 8 in that study). Individual study SMDs were tabulated in supplemental material. At baseline the pooled SMD was 0.06 (95% CI −0.22 to 0.33, p = 0.67; I2 = 0%). Ketamine showed a large, statistically significant advantage over placebo at early time points: day 1 SMD = −1.00 (95% CI −1.30 to −0.71, p < 0.00001; I2 = 0%); day 2 SMD = −1.03 (95% CI −1.45 to −0.60, p < 0.00001; I2 = 33%); day 3–4 SMD = −0.77 (95% CI −1.10 to −0.44, p < 0.00001; I2 = 22%). By day 7 the pooled effect was smaller but statistically significant: SMD = −0.36 (95% CI −0.65 to −0.08, p = 0.01; I2 = 0%). By day 14 (only two studies contributed) the effect did not reach significance: SMD = −0.38 (95% CI −0.87 to 0.11, p = 0.13). Weighted mean percent improvements from baseline in depression scores (ketamine versus placebo) were reported as follows: day 1 −41.17% versus −6.06%; day 2 −41.24% versus −0.94% (placebo value shows a formatting artefact in the extract but is reported as small); day 3–4 −34.24% versus −7.03%; day 7 −20.04% versus −7.18%; day 14 −15.38% versus −5.87%. Excluding intranasal data produced similar results: day 1 SMD = −1.04 (95% CI −1.37 to −0.71), day 2 SMD = −1.24 (95% CI −1.64 to −0.83), day 3–4 SMD = −0.87 (95% CI −1.19 to −0.55), and day 7 SMD = −0.42 (95% CI −0.74 to −0.11), all statistically significant. Subgroup analyses by diagnosis indicated that the antidepressant effect persisted through day 7 in unipolar depression but that, in bipolar depression, SMDs were significant at day 1, day 2 and day 3–4 yet lost statistical significance at day 7 and day 14. Leave‑one‑out sensitivity analyses generally showed the pooled results were not driven by any single study for day 1, day 2, day 3–4 and day 14. However, exclusion of one or two individual studies (identities are not clearly extractable in the provided text) caused loss of significance at day 7. Meta‑regression analyses testing age, sex, alcohol abuse, substance abuse, anxiety disorder, lifetime antidepressant exposure and duration of current episode or illness found no significant relationships (all p > 0.05), although the authors noted limited power because analyses relied on as few as six data points. Regarding safety, no serious adverse events were reported across included trials. Common transient adverse effects included dissociation, transient psychotic symptoms, confusion, mild increases in blood pressure, headache and anxiety; these generally resolved within about 80 minutes after ketamine administration. BPRS positive symptom scores showed no baseline difference (SMD = 0.13, 95% CI −0.14 to 0.40, p = 0.35), a significant increase at 30–40 minutes post‑dose with ketamine compared with placebo (SMD = 1.08, 95% CI 0.62 to 1.55, p < 0.00001; I2 = 56%), and no difference at 80 minutes (SMD = 0.06, 95% CI −0.37 to 0.49, p = 0.78).

Romeo and colleagues interpret the pooled findings as providing high‑level evidence that a single dose of ketamine produces a rapid antidepressant effect in treatment‑resistant major depressive episodes that is detectable from the first day and persists for about one week. The authors emphasise that ketamine appeared relatively safe in the included trials, with psychotomimetic or manic symptoms tending to be transient and resolving within roughly two hours. They note a potential diagnostic difference: antidepressant effects were sustained through day 7 in unipolar depression but failed to reach significance beyond day 3–4 in bipolar depression, suggesting possible differences in glutamatergic pathophysiology between the disorders. The discussion outlines putative mechanisms linking NMDA receptor blockade to rapid synaptic changes, BDNF‑related dendritic spine regrowth and glutamate–norepinephrine interactions, and suggests that elucidating mechanisms of early and midterm effects could inform development of rapidly acting and more durable antidepressants. The authors acknowledge important limitations: the meta‑analysis included a small number of trials and limited data, which reduces statistical power—particularly for meta‑regression analyses that relied on as few as six data points. They also caution interpretation of the apparent unipolar versus bipolar difference and call for further studies to identify predictors of response. Finally, Romeo and colleagues report that route of administration did not appear to materially affect antidepressant efficacy in these trials and that no serious harms were observed, but they frame these safety observations within the constraint of the limited dataset. The authors conclude that a single ketamine dose yields a rapid antidepressant response lasting up to one week and highlight the need for more research on durability, mechanisms and diagnostic moderators.