Rapid and Longer-Term Antidepressant Effects of Repeated Ketamine Infusions in Treatment-Resistant Major Depression

Major depressive disorder (MDD) causes substantial morbidity and societal cost in part because many patients do not respond adequately to existing antidepressant treatments. Earlier clinical research has shown that the N-methyl-D-aspartate (NMDA) receptor antagonist ketamine can produce rapid antidepressant effects, including in individuals with treatment-resistant depression (TRD). Most trials have focused on a single low-dose (.5 mg/kg) intravenous infusion and report rapid but often short-lived benefit, leaving a clinical need to identify strategies that extend or maintain ketamine's antidepressant effects. Murrough and colleagues designed the present study to characterise the time course and durability of antidepressant effects when ketamine is given repeatedly. Using an open-label regimen of up to six IV ketamine infusions administered three times weekly over 12 days, the investigators aimed to: quantify the overall response rate; determine when the largest symptom change occurs; compare symptom trajectories between responders and nonresponders; estimate time to relapse after stopping infusions; and examine ketamine's effect on individual depressive symptoms in a larger sample than their prior report.

Participants were adults with chronic or recurrent MDD who met criteria for TRD, defined as failure to respond to at least two FDA-approved antidepressants in the current episode. Recruitment came from physician referral, media advertisement, and an academic outpatient clinic. Diagnostic confirmation used the Structured Clinical Interview for DSM-IV and assessment by a study psychiatrist. Key inclusion criteria included an Inventory of Depressive Symptomatology-Clinician Rated (IDS-C) score of at least 32 at screening and baseline and a negative urine toxicology screen. Exclusion criteria included uncontrolled hypertension or unstable medical illness, recent substance abuse or dependence, lifetime history of psychosis, bipolar disorder, developmental disorder, or recreational ketamine/phencyclidine use. Women of childbearing potential required a negative pregnancy test. Participants who were taking antidepressants underwent a washout (≥2 weeks, or 4 weeks for fluoxetine) and remained medication-free during the infusion period. The protocol comprised two phases. In phase I, participants received up to six IV infusions of racemic ketamine hydrochloride at .5 mg/kg administered over 40 minutes on a Monday–Wednesday–Friday schedule across 12 days. Participants were admitted for 24 hours for the first infusion and subsequently treated as outpatients; standard telemetry monitoring and an anaesthesiologist administered each infusion. Response in phase I was predefined as ≥50% improvement on the Montgomery-Åsberg Depression Rating Scale (MADRS). Phase II followed phase I responders for relapse monitoring up to 83 days after the last infusion; relapse was defined as <50% improvement on MADRS for two consecutive visits. The protocol was amended during the study: an initial cohort exited if they failed to respond after the first infusion, whereas later participants (n = 14) remained regardless of early response to permit assessment of repeated infusions in initial nonresponders. During follow-up, most responders remained off psychotropic medication; three entered a separate randomized placebo-controlled venlafaxine ER relapse-prevention study. Safety and acute effects were assessed repeatedly. Depression ratings were taken before the first infusion (−60 min) and at +120 min, +240 min and 24 hours after the initial infusion, and at −60 and +240 min for subsequent infusions. Acute psychotomimetic effects were measured with the four-item positive symptom subscale of the Brief Psychiatric Rating Scale (BPRS), dissociation with the Clinician-Administered Dissociative States Scale (CADSS), manic-like elevation with the Young Mania Rating Scale mood item, and subjective "high" with a visual analogue scale. General side effects were recorded with a systematic self-report inventory at multiple time points. Predefined vital-sign criteria guided temporary discontinuation of an infusion (for example systolic/diastolic BP >180/100 or >20% increase above pre-infusion reading, or tachycardia >110 beats/min). Statistical analysis compared baseline characteristics between responders and nonresponders using Mann-Whitney U tests for continuous variables and chi-square tests for categorical variables. Paired t tests examined changes between two time points, and Spearman correlations quantified associations between continuous variables. Random effects models summarised MADRS score trajectories and component items over time and compared temporal differences by responder status; splines were used to identify the time when no additional improvement occurred. Predictive performance of early response (2, 4 and 24 hours) for end-of-study response was summarised with sensitivity, specificity and predictive values. Time-to-relapse among responders was estimated with the Kaplan-Meier method. Analyses used IBM SPSS Statistics and SAS. The Methods section does not state a priori sample-size calculation; the Results report the enrolled sample.

Twenty-four participants received at least one ketamine infusion; 22 received at least two, and 21 completed all six scheduled infusions. Three participants did not complete the full schedule: one exited after one infusion under the original protocol for nonresponse, one experienced hemodynamic elevation during the first infusion and was withdrawn, and one withdrew consent after three infusions citing perceived lack of benefit. At baseline the sample had a mean age of 48.1 ± 13.0 years, an age of MDD onset of 22.8 ± 13.6 years, and had failed a mean of 6.1 ± 3.3 antidepressant trials and 2.3 ± 2.3 augmentation trials in the current episode. Phase I (antidepressant effects during the infusion series): The overall response rate at study end was 70.8% (17 of 24 participants). Across the full sample, MADRS score decreased substantially within 2 hours of the first infusion by a mean of 18.9 ± 6.6 points (from 31.8 to 12.9, p < .001). This large early improvement was generally maintained throughout the infusion period as estimated by random effects modelling. Early response was highly predictive of later outcome: 94% of eventual responders had responded by 4 hours (sensitivity 94%), while 29% of nonresponders had responded at 4 hours (specificity 71%); positive and negative predictive values were .89 and .83, respectively. The relative risk of overall study nonresponse for those who had not responded at 2 hours was 4.0 (95% CI: 1.23–12.99). Individual symptom effects: Within 2 hours of the first infusion, each MADRS item examined (except appetite and sleep, which were not measured at 2 hours) showed a significant reduction across the total sample. Among the nonresponder subgroup, significant decreases were observed in sadness, inner tension, pessimistic thoughts and suicidal thoughts, but not in other items. The largest between-group effect size at 2 hours was for lassitude (Cohen's d = 1.34); substantial between-group differences were also seen for apparent sadness and concentration difficulty (d = .88 and .96, respectively). Phase II (relapse after response): The 17 responders were followed for up to 83 days. Median time to relapse was 18 days, with the 25th and 75th percentiles at 11 and 27 days. Four responders did not relapse during follow-up; the estimated probability of remaining relapse-free up to 83 days was 0.25 ± 0.11. Fourteen responders received no psychotropic medication during follow-up; three participated in the venlafaxine ER relapse-prevention trial (two randomised to active medication and one to placebo), with relapse experiences similar to the broader responder group. Acute dissociative and psychotomimetic effects: Ketamine produced a small but statistically significant increase in BPRS positive-symptom scores (mean 4.0 ± 0.1 pre-infusion to 4.5 ± 0.9 at infusion peak, p = .013), returning to baseline by +240 min. CADSS scores rose from a mean of 0.3 ± 0.5 pre-infusion to 7.8 ± 12.0 at peak (p = .001) and likewise returned to baseline by +240 min. Elevated mood and subjective "high" followed a similar transient pattern. There was no evidence of cumulative increases in these acute effects over the trial, no difference between responders and nonresponders on these measures, and no correlation between changes in MADRS and the acute effect measures. General side effects and haemodynamics: The most commonly reported side effects in the 4-hour window after infusions were feeling strange or unreal (58.3%), abnormal sensations (54.2%), blurred vision (50.0%) and drowsiness (45.8%); these effects were largely absent on pre-infusion morning assessments for infusions 2–6, indicating transience. Four participants (16.7%) reported any side effect that impaired functioning. Sixteen participants (67%) experienced no clinically significant vital-sign changes during infusions; eight (33%) had at least one episode of elevated blood pressure and/or heart rate per protocol criteria. One participant required discontinuation of the first infusion for elevated blood pressure (maximum 180/115) that stabilised after stopping the infusion. No serious adverse events were reported.

Murrough and colleagues present the largest prospective study to date of repeated ketamine infusions in TRD and highlight three principal findings: ketamine's antidepressant effect appears very rapidly, the effect spans a broad range of depressive symptoms, and an early response predicts subsequent sustained response during a course of infusions. The investigators report a large mean MADRS reduction 2 hours after the first infusion that was generally maintained across up to five additional infusions. Suicidal ideation decreased quickly across the sample, including among nonresponders, suggesting a potential anti–suicidal ideation effect distinct from full symptomatic remission. The authors emphasise the prognostic value of early response: response at 4 hours was 94% sensitive and 71% specific for response at the end of the infusion series, implying that patients who will benefit usually show rapid improvement, while a lack of immediate response predicts poor benefit from further infusions. They contrast this time course with standard antidepressants, in which many responders emerge only after several weeks, and note that definitive assessment of early-response validity requires future controlled trials. With respect to durability, the median time to relapse after the final infusion was 18 days, with substantial inter-individual variability. The authors interpret this as preliminary evidence that repeated infusions may afford greater durability than a single infusion, but they caution that optimal relapse-prevention strategies remain undefined. Prior work on riluzole did not produce consistent relapse prevention after ketamine, and the authors suggest that other mechanistically motivated approaches—for example lithium because of glycogen synthase kinase-3 inhibition implicated in ketamine's antidepressant mechanism—could be explored. Safety observations are presented as reassuring within the study context: acute dissociative and psychotomimetic effects were transient and mild, there was no accumulation of these effects across repeated infusions, and no participant experienced clinically significant psychosis. Nevertheless, the authors acknowledge continuing concerns about the safety and feasibility of prolonged ketamine exposure. They note preclinical reports of neurotoxicity at very high or repeated doses in animals and cross-sectional imaging studies linking heavy ketamine abuse with brain changes, while also acknowledging confounding factors in those human studies. Key limitations stated by the authors include the open-label design, which precludes causal attribution of efficacy because placebo effects cannot be excluded, and the modest sample size of 24, which limits generalisability. They position the study as an investigation of the pattern and durability of response to repeated ketamine administrations rather than a placebo-controlled efficacy trial. The authors conclude that although the data are preliminary, repeated ketamine infusions merit further investigation, with emphasis on controlled trials and studies to identify safe and effective relapse-prevention strategies before clinical adoption.