Efficacy of intravenous ketamine treatment in anxious versus nonanxious unipolar treatment-resistant depression

Anxious depression—major depressive disorder (MDD) accompanied by high levels of anxiety—is a common clinical subtype present in roughly half of people with MDD and appears particularly prevalent and difficult to treat in treatment-resistant depression (TRD). Previous antidepressant trials, including large multisite studies, have reported lower remission and response rates in anxious versus nonanxious depression and greater suicidal ideation and side-effect burden in the anxious subgroup. Ketamine, an N-methyl-D-aspartate receptor antagonist with glutamatergic effects, has demonstrated rapid antidepressant activity in TRD in multiple controlled trials, but evidence specifically addressing its efficacy in anxious depression has been limited to a few post hoc analyses with mixed signals. Salloum and colleagues therefore used data from a multisite, randomized, double-blind, active placebo-controlled trial of single-dose intravenous ketamine (four dose levels) versus midazolam to examine whether high baseline anxiety moderated ketamine's acute antidepressant effects. The authors hypothesised, based on prior post hoc reports, that subjects with anxious depression would show greater response to ketamine than those without high anxiety at baseline. The present paper reports secondary analyses testing that hypothesis, focusing on early outcome time points up to 72 hours postinfusion.

The source trial was a multisite, randomised, double-blind, active placebo-controlled study adding a single infusion of ketamine or midazolam to ongoing, stable antidepressant therapy in adults with unipolar TRD. Eligible participants were aged 18–70 years, met DSM-IV-TR criteria for MDD with a current major depressive episode of at least 8 weeks, and had TRD defined as <50% response to at least two (and no more than seven) adequate antidepressant treatment courses of at least 8 weeks. Participants were required to be on stable antidepressant doses for at least 4 weeks prior to screening; remote raters confirmed eligibility. Randomisation was stratified by body mass index (≤30 and >30) using block methods across six clinical sites. Ninety-nine subjects were randomised 1:1:1:1:1 to receive a single intravenous infusion over 40 minutes of ketamine 0.1 mg/kg (n = 18), 0.2 mg/kg (n = 20), 0.5 mg/kg (n = 22), 1.0 mg/kg (n = 20), or midazolam 0.045 mg/kg (n = 19). Vital signs were monitored during and after infusion, with blood pressure and heart rate recorded immediately prior to infusion and at 15–20-minute intervals for 120 minutes. Anxious depression was defined a priori as a score ≥7 on the Hamilton Depression Rating Scale Anxiety–Somatisation factor (HAMD‑AS), which comprises six items from the HAMD‑17 (psychic anxiety, somatic anxiety, gastrointestinal somatic symptoms, general somatic symptoms, hypochondriasis, and insight). The primary outcome for these secondary analyses was change in the six-item Hamilton Depression Rating Scale (HAMD6) at Days 1 and 3 postinfusion; secondary outcomes included the Montgomery–Åsberg Depression Rating Scale (MADRS; measured on Day 3 but not Day 1) and the Clinical Global Impression–Severity (CGI‑S) on Days 1 and 3. Dissociative symptoms during infusion were assessed with the Clinician-Administered Dissociative States Scale (CADSS), with analyses focusing on the 40-minute time point when dissociation peaked in the main trial. Statistical analyses employed linear mixed-effects models for primary comparisons. For HAMD6 on Day 1, the model used Day 1 HAMD6 as the dependent variable, baseline HAMD6 as a covariate, and predictor variables of ANX (anxious vs nonanxious), GROUP (ketamine pooled vs midazolam) and their interaction, with a random effect for SITE (six sites). The authors also repeated models using a five-level GROUP factor (ketamine 0.1, 0.2, 0.5, 1.0 mg/kg, and midazolam) to assess dose-related moderation. Similar models were fitted for HAMD6 Day 3, MADRS Day 3, and CGI‑S Days 1 and 3. Two-sample t-tests compared CADSS scores at 40 minutes between anxious and nonanxious subjects for (a) all ketamine-treated subjects and (b) the higher-dose ketamine subgroup (0.5 and 1.0 mg/kg). The analyses were conducted using two-tailed tests at alpha = 0.05 with SAS 9.4. The paper notes these anxious-status moderator analyses were a priori planned but the trial was powered only for the main outcome, not for moderator effects.

Of the 99 randomised participants, 45 (45.5%) met the predefined anxious depression criterion (HAMD‑AS ≥7) at baseline. Treatment assignment by arm was: ketamine 0.1 mg/kg (n = 18), 0.2 mg/kg (n = 20), 0.5 mg/kg (n = 22), 1.0 mg/kg (n = 20), and midazolam 0.045 mg/kg (n = 19). Baseline demographic and clinical characteristics were reported in tabular form (details referenced to the main paper). When ketamine arms were pooled and compared to midazolam, the interaction between treatment group and anxious-status was not significant for HAMD6 change on Day 1 (GROUP*ANX: F(1,84) = 0.02, P = 0.88) or Day 3 (F(1,82) = 0.12, P = 0.73). Reported mean change scores from baseline to Day 1 on HAMD6 were: anxious subjects on ketamine −5.36 ± 4.44 and on midazolam −2.89 ± 2.89; nonanxious subjects on ketamine −5.33 ± 4.04 and on midazolam −2.00 ± 2.87. MADRS was not collected at Day 1 per the trial design. Using a five-level GROUP variable to model ketamine doses separately versus midazolam, the GROUP*ANX interaction was not significant for HAMD6 on Day 1 (F(4,78) = 1.62, P = 0.18) nor for CGI‑S on Day 1 (F(4,78) = 0.20, P = 0.08) and Day 3 (F(4,76) = 2.16, P = 0.08). However, the interaction reached the conventional threshold for statistical significance for HAMD6 on Day 3 (F(4,76) = 2.53, P = 0.05). Follow-up analyses reported in the Discussion indicate that, at Day 3, the nonanxious subgroup showed a significantly greater response to the lowest ketamine dose (0.1 mg/kg) compared with the anxious subgroup, and similar dose-specific interactions were observed for MADRS on Day 3; these dose-specific findings were based on small sample sizes and were not corrected for multiple comparisons. Regarding dissociative symptoms, subjects with anxious depression exhibited lower CADSS scores at 40 minutes postinfusion compared with nonanxious subjects. The authors note a higher prevalence of benzodiazepine use in the anxious subgroup, which could have attenuated dissociative symptoms. The extracted results do not report detailed adverse event rates or additional numerical outcomes beyond those summarised above.

Salloum and colleagues interpret their findings to mean that intravenous ketamine appears to be similarly efficacious for acute treatment of TRD in patients with and without anxious depression when ketamine doses are analysed together. The pooled analyses showed no significant treatment-by-anxiety interactions on HAMD6 at Days 1 and 3, indicating comparable early antidepressant effects across anxiety subgroups. When ketamine doses were modelled separately, a dose-by-anxiety interaction emerged at Day 3 for HAMD6 and MADRS such that the nonanxious group responded better to the lowest ketamine dose (0.1 mg/kg) than the anxious group; the authors emphasise these dose-specific findings were exploratory, derived from small cell sizes, and not adjusted for multiple comparisons, so they should be interpreted cautiously. The discussion situates the results alongside two prior human reports that explored baseline anxiety and ketamine efficacy: an open‑label single‑infusion trial in 26 unmedicated TRD patients that suggested some advantages in anxious patients, and a pooled analysis from two crossover trials in bipolar depression that found no anxiety-by-drug interaction. Taken together with those reports, the present results are seen as supporting the potential role of ketamine for anxious depression. The authors speculate on possible mechanisms, noting preclinical evidence of anxiolytic-like effects of ketamine in rodent models and hypothesising that ketamine-induced changes in glutamate-mediated synaptic plasticity and downstream neurotrophic signalling (for example, increased AMPA receptor activation and brain‑derived neurotrophic factor release) may underlie both antidepressant and anxiolytic effects. Key limitations acknowledged by the investigators include the trial not being powered to test anxious-status as a moderator, small sample sizes when analysing individual ketamine dose groups, and the absence of postbaseline HAMD‑17 measurements to evaluate trajectories of anxiety symptoms specifically. The authors also note the possibility that concomitant benzodiazepine use in the anxious group may have influenced dissociative symptom measurements. They conclude that the present findings are exploratory and call for larger, adequately powered studies designed specifically to test whether baseline anxiety moderates ketamine response in TRD.