Replication of Ketamine’s Antidepressant Efficacy in Bipolar Depression: A Randomized Controlled Add-On Trial

Bipolar disorder is a severe and often lethal psychiatric illness for which existing pharmacological treatments do not typically produce rapid reductions in depressive symptoms or suicidal ideation within hours or a few days. Converging preclinical, postmortem, and clinical evidence has implicated glutamatergic neurotransmission and the NMDA receptor complex in mood disorders, and prior clinical trials in major depressive disorder (MDD) have shown that a single intravenous infusion of the NMDA antagonist ketamine can produce rapid antidepressant effects. Earlier work by the same group reported rapid antidepressant effects of ketamine in a small sample of treatment-resistant bipolar depression. Zarate and colleagues designed the present study to replicate that earlier finding in an independent sample of inpatients with bipolar I or II depression who were maintained on therapeutic lithium or valproate. The primary hypothesis was that a single intravenous infusion of ketamine (.5 mg/kg) would produce a rapid and clinically meaningful reduction in depressive symptoms within 230 minutes postinfusion, with secondary interest in effects on suicidal ideation and safety/tolerability.

This was a single-centre, double-blind, randomized, placebo-controlled, crossover trial conducted at the NIMH Mood Disorders Research Unit. Participants were aged 18–65, met DSM-IV criteria for bipolar I or II disorder without psychotic features, and were experiencing a major depressive episode of at least 4 weeks duration. Inclusion required a Montgomery–Åsberg Depression Rating Scale (MADRS) score of at least 20 at screening and at infusion start, prior failure of at least one adequate antidepressant trial, and failure of a prospective open mood stabilizer trial in hospital (lithium or valproate for at least 4 weeks at therapeutic levels). Key exclusions were recent substance abuse, prior ketamine treatment, serious unstable medical conditions, pregnancy or nursing, and concurrent psychotropic medications other than lithium or valproate (with a 5-week washout for fluoxetine). After a period of inpatient characterisation and mood stabiliser trial, subjects underwent two infusions two weeks apart in randomized order: ketamine hydrochloride .5 mg/kg or saline placebo, each administered intravenously over 40 minutes. Infusions were prepared in identical syringes and delivered via infusion pump by an anaesthesiologist; all staff were blind to assignment. Participants remained on lithium or valproate at therapeutic levels throughout; weekly levels were monitored. Primary outcome assessments used the MADRS; secondary measures included the 17-item Hamilton Depression Rating Scale (HDRS), Beck Depression Inventory (BDI), Visual Analogue Scale (VAS) for depression and anxiety, HAM-A, Brief Psychiatric Rating Scale, Clinician Administered Dissociative Scale, and Young Mania Rating Scale. Ratings occurred 60 minutes preinfusion, at 40, 80, 110, and 230 minutes postinfusion, and on days 1, 2, 3, 7, 10, and 14. The primary analysis followed an intent-to-treat approach using linear mixed models with time, drug, and infusion as within-subject factors; fixed effects and interactions for time and drug were included. Restricted maximum likelihood estimation with a first-order autoregressive covariance structure was used; Bonferroni-corrected simple effects tests identified timing of drug differences. Cohen's d effect sizes with confidence intervals were reported. Kaplan–Meier survival analysis was used to examine time to response and remission. Interrater reliability for key scales was high (MADRS ICC = .82; HDRS ICC = .91; YMRS ICC = .90). The investigators had initially planned a sample of about 20 to achieve 80% power based on earlier effect sizes, with allowance for an interim analysis after 60% of patients completed the study.

Fifteen patients were randomized. The extracted text does not clearly report the full demographic breakdown; all subjects were hospitalised and had stable MADRS scores during the preinfusion hospitalisation period. Four patients dropped out during the first phase (three after receiving ketamine, one after placebo), leaving 11 (73%) who completed both phases. Fourteen patients received ketamine and 12 received placebo at least once. Mean lithium levels were reported as .8 mEq/L (SD = .1); mean valproate levels were reported as 78.0, but the extracted text does not clearly report the units for valproate. Primary outcome: Using the intent-to-treat sample, a linear mixed model on MADRS scores found a significant drug-by-time interaction [F(10,187) = 5.94, p < .001]. Post hoc tests showed significantly lower depressive symptoms following ketamine compared with placebo from 40 minutes through day 3 postinfusion. When analyses were restricted to completers the drug-by-time interaction remained significant [F(10,145) = 4.95, p < .001], with differences from 40 minutes to day 3. In a first-phase–only analysis (to probe carryover), the drug-by-time interaction was also significant [F(10,66) = 5.46, p < .001]; the day-1 effect size in that analysis was large (d = 1.34, 95% CI = .60–2.07). Eight of 10 individual MADRS symptoms improved significantly with ketamine versus placebo; reduced appetite and decreased sleep did not. Response and remission: The median time to response was 40 minutes. Median time to relapse after response was 2 days; mean time to relapse was 4.5 days (SE = 1.3). Using a 50% MADRS reduction criterion, response rates for ketamine were 64% at 40 minutes, 50% at 230 minutes, and 43% at day 1. Remission (MADRS < 10) occurred in 7% at 40 minutes, 36% at 230 minutes, and 29% at day 1. Overall, 79% of subjects responded to ketamine at some point during the study, whereas 0% responded to placebo. Relative change from baseline averaged +5% for placebo at 40 minutes (– values indicate improvement in the paper; here the reported averages were 5% improvement for placebo at 40 minutes, 9% at 230 minutes, and 1% at day 1), compared with 50% improvement for ketamine at 40 minutes, 45% at 230 minutes, and 41% at day 1. Secondary measures and suicidal ideation: Drug-by-time interactions were significant for HDRS, BDI, and VAS-Depression, confirming the primary MADRS findings. Anxiety measures (HAM-A and VAS-Anxiety) also showed significant reductions with ketamine as early as 40 minutes. No significant drug effects were seen on YMRS or Brief Psychiatric Rating Scale. Suicidal ideation scores on the MADRS, HDRS, and BDI were lower following ketamine than placebo after adjusting for baseline. On the HDRS, ketamine-associated reductions in suicidal ideation were observed from 40 to 80 minutes and at day 2; on the BDI reductions were seen from 40 minutes through day 2 and at day 10. Safety and tolerability: No serious adverse events occurred. Adverse events reported in at least 10% of subjects during infusion included feeling woozy or loopy, lethargy/drowsiness, cognitive impairment, fear or anxiety, nausea, dizziness, odd sensations, blurred vision, and headache. Dissociative symptoms were short-lived and occurred primarily at the 40-minute time point. No adverse event differed significantly from placebo at 80 minutes or thereafter. There were no significant changes in electrocardiogram, respiratory function, or laboratory values.

Zarate and colleagues interpreted the results as a successful replication: a single intravenous infusion of ketamine produced a rapid and robust antidepressant response in patients with bipolar depression, with onset within approximately 40 minutes and effects lasting through about 3 days on average. The study additionally demonstrated rapid reductions in suicidal ideation on multiple rating scales, an effect that the authors report as novel for a controlled trial in bipolar depression. Improvements were observed across clinician-rated and self-rated depression measures and on anxiety scales, despite the sample having long-standing, treatment-heavy illness histories. The investigators related their clinical findings to proposed mechanisms from preclinical work: initial NMDA receptor antagonism followed by increased AMPA throughput and activation of the mammalian target of rapamycin (mTOR) pathway, leading to synaptogenesis in prefrontal cortex, a process putatively underlying rapid antidepressant effects. They noted that the replication of similar rapid effects across MDD and bipolar depression supports a shared glutamatergic mechanism for ketamine's rapid action. Strengths highlighted by the authors included prolonged inpatient characterisation and prospective mood stabiliser trial before infusion, the randomized placebo-controlled design, and close replication of prior findings. Limitations acknowledged were the small sample size, limited generalisability given the chronic and treatment-refractory nature of the sample, ambiguity around whether all participants met standard definitions of treatment-resistant bipolar depression, and potential compromise of the blind because of transient dissociative effects. The study design conserved elements of the original trial (for replication), and raters and patients were not required to guess treatment assignment, so the possibility of unblinding bias could not be formally assessed. The authors concluded that while ketamine produces rapid antidepressant and antisuicidal effects in bipolar depression, these effects are typically short-lived for most patients, and further work is needed to replicate findings independently, use active controls, and develop strategies to sustain ketamine's benefit.