Antidepressant Efficacy of Ketamine in Treatment-Resistant Major Depression: A Two-Site Randomized Controlled Trial

Major depressive disorder is a leading cause of disability worldwide and a substantial subgroup of patients fail to achieve clinically meaningful improvement despite multiple antidepressant trials and augmentation strategies. Treatment-resistant major depression is typically defined as inadequate response to at least two adequate antidepressant treatments and is associated with poor outcomes and a slow therapeutic onset for conventional agents, which generally act on monoamine systems and require 4–12 weeks to produce benefit. Converging evidence implicates glutamatergic dysfunction in depression, and small case series and crossover studies had suggested that the NMDA receptor antagonist ketamine can produce rapid antidepressant effects within hours of a single subanesthetic intravenous infusion; however, prior work was limited by small samples, inert placebos, and potential unblinding due to ketamine's psychoactive effects. Murrough and colleagues designed a two-site, parallel-arm, double-blind randomized controlled trial to test whether a single low-dose ketamine infusion produces a rapid antidepressant effect in patients with treatment-resistant major depression. To strengthen blinding and control for nonspecific anaesthetic effects they used midazolam as an active placebo comparator, selected 24 hours postinfusion as the primary endpoint to capture rapid effects while minimising contamination by acute psychoactive symptoms, and powered the study to detect clinically meaningful differences in clinician-rated depression severity.

This was a two-site (Baylor College of Medicine and Icahn School of Medicine at Mount Sinai) parallel-arm randomized controlled trial conducted between November 2010 and August 2012. Adults aged 21–80 with a primary DSM-IV diagnosis of major depressive disorder and treatment resistance (inadequate response to at least three antidepressant trials per the Antidepressant Treatment History Form) were eligible. Additional inclusion criteria required either recurrent major depressive disorder or a chronic current episode of at least 2 years plus a screening Inventory of Depressive Symptomatology-Clinician Rated score of 32 or greater. Key exclusions were lifetime psychotic illness or bipolar disorder, substance or alcohol abuse in the past 2 years, unstable medical illness, imminent suicidal or homicidal risk, a Mini-Mental State Examination score under 27, or use of contraindicated medications. Standard safety screening including physical examination, laboratory testing, urine toxicology and ECG was performed, and institutional review boards approved the protocol. Participants were medication-free for the study duration (with limited allowance for stable non-benzodiazepine hypnotics) and underwent washout prior to infusion (at least 4 weeks for fluoxetine, 1 week for other agents). Randomisation at a 2:1 ratio assigned patients to a single intravenous infusion of ketamine hydrochloride 0.5 mg/kg or midazolam 0.045 mg/kg, each administered over 40 minutes. Midazolam was chosen as an active placebo to mimic ketamine's onset and short half-life and to enhance blinding. Study personnel, clinicians, anaesthesiologists, raters and patients were masked to allocation; a research pharmacist prepared sealed envelopes with drug identity. Physiological monitoring was performed during and after infusion and trained raters who were blind to infusion-day side effects conducted symptom assessments. The primary outcome was change in depression severity on the Montgomery-Åsberg Depression Rating Scale (MADRS) at 24 hours postinfusion. Secondary outcomes included MADRS response rate (≥50% reduction), the Quick Inventory of Depressive Symptomatology-Self-Report, Clinical Global Impression (CGI) severity and improvement measures, and durability of benefit up to 7 days. Safety assessments captured general adverse events, dissociative states (Clinician-Administered Dissociative States Scale), and psychotomimetic symptoms (Brief Psychiatric Rating Scale positive symptom subscale). The protocol defined nonresponders as <50% improvement on MADRS; nonresponders were followed for 7 days while responders were followed biweekly until relapse or for an additional 4 weeks. The statistical plan used conservative effect-size estimates to power the trial: a planned sample of 72 randomized in a 2:1 ratio provided 80% power for continuous MADRS differences and 96% power for dichotomous response differences at 24 hours, assuming Cohen's d=0.71 and response rates of 60% versus 15%. Modified intention-to-treat analyses included all randomized patients with baseline and at least one postbaseline assessment. General linear modelling (mixed models) evaluated MADRS scores at 24 hours controlling for baseline MADRS and site; exact logistic regression analysed response rates adjusted for site. Secondary endpoints used analogous general linear or ordinal logistic models. Time-course analyses among responders employed mixed modelling of MADRS over follow-up. A significance threshold of p≤0.05 was applied and safety analyses were descriptive.

Of 116 patients who consented, 73 met eligibility and were randomized in a 2:1 ratio. The extracted text reports that all but one randomized patient (assigned to ketamine) received study medication and completed 24-hour assessments; 67 patients completed assessments through postinfusion day 7. The ketamine and midazolam groups were reported as well matched on demographic and clinical baseline characteristics and overall represented a chronically ill, moderate-to-severe treatment-resistant cohort. Primary outcome: At 24 hours postinfusion the ketamine group showed significantly greater reduction in MADRS scores than the midazolam group. After adjustment for baseline MADRS and site, the mean MADRS score was 7.95 points lower in the ketamine group (95% CI, 3.20 to 12.71), corresponding to Cohen's d=0.81. MADRS differences at 24 hours did not differ by site. Secondary outcomes at 24 hours: Response rates (≥50% reduction in MADRS) were higher with ketamine: 64% versus 28% with midazolam, yielding an odds ratio of 2.18 (95% CI, 1.21 to 4.14; p≤0.006) and a number needed to treat of 2.8. Self-reported depressive symptoms (Quick Inventory of Depressive Symptomatology-Self-Report) were lower in the ketamine group by 3.40 points (95% CI, 0.78 to 6.01; p≤0.02), Cohen's d=0.63. Ketamine also increased the odds of CGI-rated improvement at 24 hours. Durability: Treatment differences diminished over time. When modelling response probability across time, there were main effects of time (increasing nonresponse probability) and treatment (greater probability of response with ketamine), but no significant treatment-by-time interaction, indicating parallel trajectories. After adjustment for site and baseline scores, MADRS and self-report scores at day 7 no longer showed statistically significant differences between groups. At day 7, four midazolam patients and 21 ketamine patients still met response criteria and continued in the study; time-to-relapse over the subsequent 4 weeks is reported for this subgroup in the supplement. Safety and tolerability: The most common adverse events within 4 hours of infusion in the ketamine group were dizziness, blurred vision, headache, nausea/vomiting, dry mouth, poor coordination, poor concentration and restlessness; the midazolam group reported general malaise, dizziness, headache, restlessness, nausea/vomiting, dry mouth, decreased energy and poor coordination. Significant dissociative symptoms were observed in 8 of the 47 patients reported as receiving ketamine (17%); these resolved by 2 hours postinfusion. No severe psychotic symptoms occurred during follow-up. Mild transient increases in blood pressure were observed on average; two ketamine infusions were discontinued for hemodynamic changes. In one case infusion was stopped after 30 minutes for an elevated blood pressure (peak 187/91 mm Hg) that normalised within 10 minutes of cessation; in another case pronounced hypotension and bradycardia resolved without sequelae after overnight observation. The extracted text indicates serious adverse events are described in the online supplement but does not detail them in the main text.

Murrough and colleagues interpret the findings as evidence that a single low-dose ketamine infusion produces a rapid-onset antidepressant effect in patients with treatment-resistant major depression when compared with an active psychoactive placebo. The investigators emphasise robust clinician- and patient-rated improvements at 24 hours and note that responders commonly maintained benefit for several days, although statistically significant group differences were not present at day 7. They argue the two-site replication in an ethnically and racially diverse sample bolsters the reliability of the effect. The authors place their results in a biological context consistent with preclinical work showing rapid synaptic and functional changes after NMDA receptor antagonism, including activation of mTOR signalling and modulation of brain-derived neurotrophic factor pathways; they note emerging evidence that genetic variation in BDNF may moderate response to ketamine. They also stress that, while short-term administration at 0.5 mg/kg appeared safe in this trial for nonpsychotic patients with close monitoring, the longer-term safety, durability of effect and abuse liability of ketamine remain unknown and warrant caution regarding clinical implementation outside research settings. Cardiorespiratory monitoring is recommended given observed hemodynamic changes in a subgroup. Limitations acknowledged by the authors include stringent enrollment criteria (excluding psychotic disorders and recent substance use) and a medication washout requirement that limited participation to those able to tolerate discontinuation; these criteria were defended as clinically relevant risk-mitigation strategies but do constrain generalisability. The study tested only a single infusion with brief follow-up, leaving open the question of how to maintain and prolong the antidepressant response. The authors note that trials of riluzole after ketamine have not prevented relapse and that while repeated ketamine infusions have been explored an absence of controlled data limits conclusions. They suggest future work should evaluate repeated dosing schedules, compare ketamine to active standard treatments such as electroconvulsive therapy, and characterise longer-term efficacy and safety before routine clinical use.