Factors Associated with Antidepressant Effects of Ketamine: A Reanalysis of Double-Blind Randomized Placebo-Controlled Trial of Intravenous Ketamine for Treatment-Resistant Depression

Ketamine, an NMDA receptor antagonist, has shown rapid antidepressant effects in clinical trials for major depressive disorder (MDD) and bipolar depression, but real-world response rates vary widely. Previous reviews and cohort studies have suggested a range of clinical predictors of ketamine response — including family history of alcohol use disorder, body mass index, younger chronological age, number of prior treatment failures, biomarkers such as BDNF, and neuroanatomical measures — yet findings across studies have been inconsistent and study designs heterogeneous. Notably, the roles of age at illness onset and acute dissociative symptoms after infusion remain underexplored in homogeneous samples of treatment-resistant depression (TRD). Yonezawa and colleagues conducted a post-hoc analysis of a double-blind, placebo-controlled randomized trial (NCT01920555) to identify clinical factors associated with acute antidepressant response to therapeutic doses of intravenous ketamine (0.5 and 1.0 mg/kg) in adults with TRD. The investigators hypothesised that age at onset and dissociative symptoms after infusion would relate to treatment response, and they examined demographic and clinical predictors using regression models with change in the HAM-D-6 score and a binary response definition as outcomes. This study aims to clarify predictors in a diagnostically uniform TRD sample drawn from a high-quality trial dataset.

This study is a post-hoc secondary analysis of data from a previously conducted placebo-controlled, double-blind, randomised trial (NCT01920555) of single 40-minute intravenous infusions in adults with current major depressive episodes meeting DSM-IV criteria for MDD and exhibiting treatment-resistant depression (TRD). TRD was defined as less than 50% improvement after at least two adequate antidepressant courses in the current episode, and enrolled participants had a MADRS total score ≥20 at screening and baseline. In the original trial 99 participants were randomised across five arms: ketamine 0.1 mg/kg, 0.2 mg/kg, 0.5 mg/kg, 1.0 mg/kg, and midazolam 0.045 mg/kg as an active placebo. For this reanalysis the investigators restricted the sample to participants who received a single therapeutic ketamine dose (0.5 or 1.0 mg/kg) and who had baseline HAM-D-17 and HAM-D-6 assessments as well as HAM-D-6 at day 3. The primary clinical outcome here was treatment response on day 3, defined as a ≥50% reduction in HAM-D-6 total score versus baseline; percentage change in HAM-D-6 from baseline to day 3 was analysed as a continuous secondary outcome. Dissociative symptoms were measured with the Clinician-Administered Dissociative States Scale (CADSS) at 40 minutes after infusion and used as an explanatory variable. Logistic regression was used to examine associations with binary response status, and multiple linear regression was applied to the percentage change outcome. Explanatory variables included age, age at illness onset, sex, baseline HAM-D-6 total score, and CADSS total score at 40 minutes. Statistical significance was set at two-tailed p < 0.05, analyses were performed in R version 4.1.0, and variance inflation factors were inspected to assess multicollinearity. Ethical approvals and informed consent were obtained for the original trial; this secondary analysis used anonymised data and did not seek additional approval.

Thirty-one participants met the inclusion criteria for this post-hoc analysis (ketamine 0.5 mg/kg: n = 17; 1.0 mg/kg: n = 14). The sample comprised 13 women with a mean age of 48.4 years (standard deviation reported elsewhere as 10.9). Treatment response on day 3 — defined as ≥50% reduction in HAM-D-6 — was the primary endpoint for the predictive analyses. Logistic regression identified a positive association between age at illness onset and treatment response at day 3 (β = 0.08, p = 0.037), indicating that a later age of onset predicted a greater likelihood of achieving the predefined response. No statistically significant relationships were observed between response and chronological age, sex, baseline HAM-D-6 total score, or CADSS total score measured 40 minutes after infusion. In the multiple linear regression examining percentage change in HAM-D-6 from baseline to day 3, none of the evaluated variables showed a significant association. Variance inflation factors for the explanatory variables were all below five, suggesting multicollinearity was not a concern in these models.

Yonezawa and colleagues interpret their principal finding — that later age at illness onset was associated with better acute antidepressant response to therapeutic-dose ketamine at three days — as suggesting biological heterogeneity between early-onset and late-onset TRD. They propose that earlier-onset MDD may index chronicity or a longer duration of illness, both of which have been associated with poorer response to conventional antidepressant treatments and may likewise reduce responsiveness to ketamine. The authors consider neurobiological mechanisms that might underlie this effect, emphasising evidence for impaired glutamatergic signalling and diminished AMPA receptor expression and synaptic spine density in chronic stress and depression models. Because ketamine's antidepressant action is thought to involve AMPA receptor activation and synaptogenic pathways, a longer disease course or early-onset disease could plausibly attenuate ketamine’s effectiveness via reduced neuroplasticity. The discussion notes the recent development of an AMPA-specific PET tracer ([11C]K-2) as a potential tool for future studies investigating receptor expression and treatment response. Regarding dissociative symptoms, the study found no correlation between CADSS total score at 40 minutes post-infusion and antidepressant response at day 3. The authors acknowledge that prior studies report mixed results on this relationship; they suggest that the CADSS may not sensitively capture ketamine-induced dissociation and that observed discrepancies across studies may reflect differences in sample size, study design, timing, and measurement instruments. Key limitations acknowledged by the investigators include the post-hoc design, small sample size, restriction to a single early assessment time point (day 3), and lack of data on duration of the index depressive episode. They also note that other potentially relevant predictors reported in prior work — such as family history of alcohol use disorder and body mass index — were not available in the dataset. Strengths highlighted include the use of data from a rigorously conducted, double-blind randomised trial and a diagnostically uniform TRD sample. The authors recommend larger, prospectively designed studies that include measures of illness duration, more sensitive dissociation instruments, and biomarker assessments to clarify predictors and mechanisms of ketamine response.

A later age at onset of depression was associated with a better short-term antidepressant response three days after a single therapeutic infusion of ketamine (0.5 mg/kg or 1.0 mg/kg) in this sample of patients with treatment-resistant depression. The authors suggest that early-onset TRD may represent a biologically distinct subgroup with greater resistance to ketamine, potentially due to impaired glutamatergic transmission and reduced neuroplasticity after a long disease course. They call for further research to identify reliable predictors of ketamine response and to elucidate the underlying mechanisms of its antidepressant effect.