Effectiveness and safety of repeat dose subcutaneous ketamine for treatment-resistant depression, and the impact of prior ketamine treatment: open label extension of the KADS study

Treatment-resistant depression remains difficult to manage, and although ketamine has shown rapid antidepressant effects, earlier studies have often been short term, small, or limited to specific routes of administration. The paper notes that evidence for repeated subcutaneous racemic ketamine, especially over longer follow-up, is sparse, and that it is also unclear whether having received ketamine before changes how patients respond to another course. Subcutaneous administration may be more practical than intravenous infusion, but its medium-term effectiveness and safety had not been well characterised in a larger sample. Glozier and colleagues set out to examine the effectiveness and safety of a 4-week repeat course of subcutaneous racemic ketamine in adults with treatment-resistant depression, with follow-up to 6 months after treatment. A second aim was exploratory: to test whether prior ketamine exposure in the earlier randomised trial phase, or response to that earlier treatment, influenced outcomes during the open label extension. The study used data from the open label extension of the KADS trial, making it one of the larger and longer follow-up datasets for repeated subcutaneous ketamine in this population.

The researchers conducted an open label extension of a 4-week, randomised, double-blind, active-control, parallel-group, multicentre Phase III trial at seven specialist mood disorder centres in Australasia. Adults with treatment-resistant major depressive disorder were eligible if, at the 4-week post-trial assessment, they still had a Montgomery-Åsberg Depression Rating Scale (MADRS) score of at least 20. Participants were required to have depression of at least 3 months' duration, confirmation by structured clinical interview, inadequate response to at least two antidepressant trials in the current episode, and stable concurrent antidepressant or augmenting medication for at least 4 weeks before and during the study. All OLE participants received subcutaneous racemic ketamine twice weekly for 4 weeks, for a planned total of eight treatments. The initial regimen was fixed at 0.5 mg/kg. After 51 participants had completed the RCT phase, a Data Safety Monitoring Board recommended a flexible, response-guided regimen for both the RCT and the OLE, allowing dose increases to 0.6, 0.75 and 0.9 mg/kg at later sessions if the MADRS had not improved by at least 50% from OLE baseline. Participants were observed after dosing, and no additional study treatment beyond the ketamine course was provided, although usual medications and psychological therapy could continue. The primary outcome was depression severity on the MADRS, measured before each treatment, 3-4 days after the eighth dose at 4 weeks, and again at 4 weeks and 6 months after treatment. Secondary outcomes included Clinical Global Impression-Severity and Improvement ratings. Safety monitoring covered suicidality using the Columbia Suicide Severity Rating Scale, urinary symptoms, dissociative and ketamine-specific effects, and blood pressure and heart rate around each injection. For analysis, the authors described response, remission and non-response rates, and examined change in MADRS over time. For the exploratory questions, they used a linear mixed-effects model with repeated measures, adjusting for session, baseline MADRS, site, prior responder status, and interactions, to compare outcomes by prior ketamine exposure and by response in the RCT phase.

Of 151 RCT participants who attended the 4-week post-RCT assessment, 130 (86%) entered the OLE. Overall, 116 completed the 4-week ketamine treatment course and were assessed at end of treatment, 110 (85%) were assessed at 4 weeks after treatment, and 73 (56%) at 6 months. There was no clear evidence of attrition bias based on baseline characteristics. At OLE entry, 44 participants (34%) reported active suicidal thoughts. Depression outcomes showed a short-term benefit in a minority of participants. At the end of the 4-week OLE course, 30% (35/116) met response criteria, defined as at least a 50% reduction in MADRS, and about one quarter were in remission. Four weeks later, response had fallen to 17% (19/110), and by 6 months only 11% (8/73) remained responders. More than half of participants were non-responders at each assessment point. Mean MADRS fell from 28.8 at baseline to 20.3 at end of treatment, then rose to 23.0 at 4 weeks and 24.8 at 6 months, indicating that the largest improvement occurred after the first dose and was partly lost after treatment stopped. There were no consistent significant differences between the fixed and flexible dosing regimens on depression outcomes. Both regimens showed improvement from baseline during treatment, and the gains remained statistically better than baseline at 4 weeks and 6 months, although symptoms worsened after cessation. Regarding prior ketamine exposure, participants who had received ketamine in the RCT showed less improvement after their first OLE treatment than those who had received midazolam, with a mean MADRS difference of 3.2 points at that time point, but no other time points showed a difference. This suggests a possible attenuation only for the first re-treatment dose. By contrast, prior RCT responders tended to improve more during the OLE, and the difference was significant by the end of treatment and remained so at 6 months. In a sensitivity analysis, 13 of 58 RCT midazolam non-responders (22%) later responded in the OLE, while 11 of 44 RCT ketamine non-responders (25%) responded when retreated with ketamine. Most prior ketamine responders responded again in the OLE. Clinician-rated global outcomes also improved: the proportion rated as having mild illness or better rose from 8% at baseline to 46% at end of treatment, then declined to 31% at 4 weeks. By end of treatment, 45% were rated as much or very much improved, falling to 32% at 4 weeks and 17% at 6 months. Safety findings were reassuring. There were no deaths or self-harm events requiring admission. Three severe adverse events occurred, all in the flexible-dose group, and were judged unlikely to be related to ketamine. Clinically relevant blood pressure rises occurred in 12.9% of the fixed-dose group and 11.3% of the flexible-dose group, with no rescue medication required. Fourteen participants (11%) reported at least one new urinary symptom, but none had a clinically significant bladder symptom score after treatment. Dissociative symptoms increased transiently during sessions as expected, and the more common acute effects were consistent with known ketamine effects. Four participants reported mild craving once during treatment, and one participant reported brief recreational use after treatment ended.

The authors interpret the OLE as providing the largest medium-term dataset to date for repeated subcutaneous racemic ketamine in highly treatment-resistant depression, in a setting that is closer to routine practice because it was open label and non-randomised. They argue that the average antidepressant effect was clinically meaningful, but that this average concealed substantial variability in response, which is important for patient counselling. They emphasise that response and remission rates were similar to those seen in the earlier trial phase, but that benefits often diminished quickly after treatment stopped, with fewer than one in five still responding 4 weeks later. They also conclude that the treatment was generally well tolerated, with expected transient cardiovascular and dissociative effects but no major safety signal, including no deaths or self-harm requiring admission during treatment or within 2 months after stopping. The authors note that this is reassuring for repeated dosing up to 0.9 mg/kg, although they do not claim that longer-term safety has been definitively established. In relation to earlier research, they suggest that the response rates in this very treatment-resistant sample may appear lower than in some other ketamine studies, but they attribute this partly to the stricter and more refractory nature of the cohort, and to the fact that the OLE excluded participants who had sustained benefit after the RCT. They also note that the flexible-dose regimen was not superior to the fixed-dose regimen, and raise the possibility that higher doses, longer treatment, or different protocols may be needed, although they do not present this as established. A notable interpretation is that prior ketamine exposure did not appear to materially shape the response trajectory over 6 months, apart from a smaller effect after the first re-treatment dose. The authors suggest this means prior exposure should not currently alter patient selection or counselling. By contrast, prior response may be more informative, because most previous responders responded again, whereas some previous non-responders also improved when retreated. They explicitly state that the study was not powered for these subgroup analyses and that the findings need confirmation. The main limitations they acknowledge are selection bias from requiring consent to randomisation, the unblinded open label design, and attrition by 6 months, especially among responders. They also note that it remains unclear what the best dosing protocol is, whether prior response is a reproducible predictor, whether findings generalise across ketamine formulations and delivery systems, and whether longer treatment could sustain benefit once treatment stops. The authors say further work is needed to clarify these issues as ketamine becomes more widely used in practice.