Efficacy, Safety, and Durability of Repeated Ketamine Infusions for Comorbid Posttraumatic Stress Disorder and Treatment-Resistant Depression

Major depressive disorder is a leading cause of global disability and about one-third of patients fail to remit after multiple standard treatments, a condition termed treatment-resistant depression (TRD). Ketamine has emerged as a rapidly acting, novel treatment for TRD; prior trials have used intravenous racemic ketamine or intranasal S-ketamine (esketamine). Substantive phase III data exist for intranasal esketamine, but randomised controlled trials of repeated racemic ketamine have been few, small and heterogeneous in design. Choice of comparator (saline placebo versus an active psychoactive control such as midazolam) and route of administration (intravenous versus intramuscular, subcutaneous, oral or intranasal) both affect masking and measured effect sizes. Data on cumulative or longer-term safety with repeated dosing are also limited. Loo and colleagues set out to evaluate the efficacy, safety and durability of a 4-week course of repeated subcutaneous racemic ketamine in adults with TRD, using midazolam as an active comparator to improve blinding. The trial aimed to test whether repeated, adequately dosed subcutaneous ketamine would produce higher remission and response rates than midazolam while systematically assessing acute, between-session and short-term cumulative adverse effects.

This was a multicentre, randomised, double-blind, active-controlled Phase III trial (KADS) conducted at seven specialist mood-disorder centres in Australia and New Zealand. Ethical approvals were obtained from relevant committees and all participants provided written informed consent. Eligible participants were adults (≥18 years) with major depressive disorder of at least 3 months' duration confirmed by the Structured Clinical Interview for DSM-5, insufficient response to at least two adequate antidepressant trials, a baseline MADRS score ≥20, and stable antidepressant medication for ≥4 weeks if present. Participants were randomised 1:1 to receive subcutaneous racemic ketamine hydrochloride or subcutaneous midazolam, with drug vials prepared to appear identical. Blinding was maintained for participants, raters and most study personnel; only the trial pharmacist, the statistician who generated the randomisation sequence and the Data Safety Monitoring Board were unblinded. Participants and raters were asked to guess treatment allocation during the trial to assess masking. Treatments were administered into the abdominal wall twice weekly for 4 weeks (eight sessions). The study enrolled two sequential cohorts. Cohort 1 used a fixed dose (ketamine 0.5 mg/kg vs midazolam 0.025 mg/kg, identical injection volumes). After a DSMB review of the first 51 completers and low remission rates, the protocol was amended for cohort 2 to a flexible, response-guided dosing regimen: dose escalation steps for ketamine to 0.6, 0.75 and 0.9 mg/kg and matched midazolam increments (0.03, 0.0375 and 0.045 mg/kg) if a participant had not achieved ≥50% MADRS improvement at predefined sessions. The primary outcome was remission defined as MADRS ≤10 assessed 3–4 days after the final treatment (end of the 4-week randomised period). Key secondary outcomes included remission defined as MADRS ≤12, response (≥50% MADRS reduction), change in MADRS score from baseline to RCT end, and CGI-S/CGI-I scores. Safety was assessed at each session and between sessions using a prototype Ketamine Side Effect Tool (KSET), as well as CADSS and YMRS for dissociation and emergence of manic symptoms. Analyses used a modified intention-to-treat population (randomised and received ≥1 treatment). The primary outcome was modelled via penalised logistic regression on multiply imputed datasets with treatment, baseline MADRS and site as covariates; mixed-effects models were used for longitudinal MADRS change, and ordinal regression for CGI outcomes. Pre-specified heterogeneity, sensitivity and subgroup analyses were conducted according to the published analysis plan.

Between August 2016 and April 2020 (follow-up completed May 2020), 1,033 people were screened and 184 were randomised; three subsequently withdrew consent for data use. Of the remaining participants, cohort 1 comprised 73 randomised (68 received ≥1 dose) and cohort 2 comprised 108 randomised (106 received ≥1 dose). Most participants received all eight planned doses. In cohort 2 the majority were escalated to the highest dose level (30/53 ketamine, 38/53 midazolam). Primary outcome: In cohort 1 (fixed-dose) there was no significant difference in remission (MADRS ≤10) between ketamine and midazolam (remission 6.3% v. 8.8%; OR = 1.34, 95% CI 0.22–8.21, P = 0.76). In cohort 2 (flexible-dose), ketamine produced significantly higher remission than midazolam (19.6% v. 2.0%; OR = 12.11, 95% CI 2.12–69.17, P = 0.005), corresponding to a number needed to treat (NNT) of 6.01 (95% CI 3.34–30.58). The treatment effect odds in cohort 2 were greater than in cohort 1 (interaction OR = 12.96, 95% CI 1.10–152.54, P = 0.04). Missing data at the primary endpoint were minimal. Secondary outcomes: Cohort 1 showed no significant between-group differences on mean MADRS change, response (≥50% reduction) or remission defined as MADRS ≤12. In cohort 2 ketamine outperformed midazolam: the between-group difference in mean MADRS change at treatment end was 5.5 points (95% CI 2.1–8.7), response rates were 29% v. 4% (P = 0.001), and remission (MADRS ≤12) was 22% v. 4% (P = 0.007). CGI-S and CGI-I measures mirrored these findings, with significant improvements favouring ketamine in cohort 2 but not cohort 1. Pre-specified subgroup signals suggested larger treatment effects among participants with higher baseline anxiety, a greater number of prior failed antidepressant trials, and those taking concomitant antipsychotics. Follow-up: At 4 weeks after the final treatment the between-group difference in remission in cohort 2 was attenuated and no longer statistically significant (8.0% ketamine v. 2.1% midazolam; OR = 2.02, 95% CI 0.40–10.28, P = 0.4). No significant differences were observed at follow-up for MADRS-based remission/response or mean change in either cohort. Longer-term follow-up data were limited because many participants entered an open-label phase after the 4-week follow-up. Safety and masking: Serious adverse events were uncommon and mostly judged unrelated to study drug. In cohort 2 two serious events in the ketamine arm were considered related (one major dissociative episode and one auditory hallucination). Four ketamine-treated participants discontinued for non-serious adverse events. Typical acute ketamine effects (psychotomimetic symptoms, transient blood-pressure elevation) occurred—more frequently in cohort 2—and resolved within the 2-hour observation window without need for medical intervention. No cognitive impairment was detected during the trial (details to be reported elsewhere). Assessment of blinding using the Bang Blinding Index indicated incomplete masking: in cohort 1 raters and participants tended to guess midazolam correctly; in cohort 2 masking failed for midazolam groups and was partial for ketamine. Most rater guesses were driven by perceived efficacy rather than intra-treatment experience.

Loo and colleagues report the largest randomised controlled trial to date of racemic ketamine for TRD and show that adequately dosed subcutaneous ketamine administered over 4 weeks can be efficacious and reasonably safe when compared with an active psychoactive control. The antidepressant superiority of ketamine was evident only in the flexible, response-guided dosing cohort where escalation up to 0.9 mg/kg was permitted; no efficacy advantage was seen in the fixed-dose 0.5 mg/kg cohort. The patient sample was highly treatment-resistant (24% had previously failed electroconvulsive therapy), and the authors view the cohort-2 benefits as both clinically and statistically meaningful in this context. The investigators acknowledge several limitations. A planned mid-study dosing amendment, recommended by the DSMB after observing no remissions among early completers, resulted in two cohorts and meant neither reached the originally planned sample size. At the time of trial design, the bioavailability of subcutaneous ketamine (~0.66) was not known; subsequent data suggest higher subcutaneous doses (≈0.75 mg/kg) better approximate effective intravenous dosing, which helps explain the cohort differences. Recruitment was disrupted by the COVID-19 pandemic, and the trial was designed to assess effects over a 4-week treatment period with only a 4-week post-treatment follow-up, so longer-term efficacy and maintenance questions remain unanswered. Blinding was imperfect despite use of an active control, and the authors discuss the likelihood that some degree of unmasking is inevitable in trials of psychoactive treatments; they also present arguments that unmasking alone is unlikely to explain the cohort-2 treatment effect (for example, raters and participants based many guesses on observed efficacy rather than acute subjective drug effects, and cohort 1 showed unmasking without a treatment effect). The study contributes important evidence on dosing and route of administration: subcutaneous delivery of generic racemic ketamine is a simpler, lower-cost parenteral option with practical clinical translation potential. The authors highlight the value of individualised, response-guided dose escalation to optimise efficacy and safety, noting that the full antidepressant effect in cohort 2 emerged relatively late owing to the titration schedule. Comprehensive safety monitoring using the KSET showed dose-related increases in acute psychotomimetic and cardiovascular effects but an overall acceptable acute and short-term safety profile when treatment is delivered within a structured monitoring framework. Finally, the authors underline that benefits commonly attenuate after treatment cessation, supporting consideration of ongoing treatment strategies and further research comparing formulations, routes and maintenance approaches.