The relationship between subjective effects induced by a single dose of ketamine and treatment response in patients with major depressive disorder: a systematic review

Mathai and colleagues frame ketamine as an N‑methyl‑D‑aspartate receptor (NMDAR) antagonist with rapid antidepressant effects that are hypothesised to arise from modulation of glutamate transmission, AMPA receptor potentiation and downstream neuroplastic changes. The introduction notes ketamine's broader pharmacology, neuroimaging findings, and historical links to psychedelic research; it highlights that ketamine can occasion pronounced alterations of consciousness that resemble, in some respects, experiences induced by serotonergic psychedelics. The authors point out that while dissociative and psychotomimetic effects are routinely measured in ketamine trials (often as adverse events), it remains unclear whether and how these subjective states relate to antidepressant benefit. This systematic review therefore set out to examine empirical evidence for a relationship between single‑dose ketamine‑induced subjective experiences and subsequent change in depression scores in adults with Major Depressive Disorder (MDD). Specifically, the investigators sought studies that measured subjective effects with validated psychometric scales (for example the Clinician‑Administered Dissociative States Scale, CADSS; the Brief Psychiatric Rating Scale, BPRS; and the 5‑Dimensional Altered States of Consciousness Rating Scale, 5D‑ASC) and statistically correlated those measures with depression outcomes following a single ketamine exposure. The goal was to clarify whether hallucinogenic or dissociative states contribute meaningfully to ketamine’s antidepressant mechanism or clinical effect.

The review followed PRISMA recommendations. Two reviewers performed searches of PsycINFO, Embase and PubMed for peer‑reviewed articles dated from 1 January 2000 to 31 May 2019. Search terms combined ketamine/esketamine with a broad set of terms capturing mystical, psychedelic, psychotomimetic and altered‑states constructs, plus depression‑related terms. Inclusion criteria required studies of adults diagnosed with MDD who received ketamine, with validated measures of depressive symptoms collected at regular intervals after administration, validated measures of subjective state collected on the day of dosing, and a reported statistical correlation between the subjective‑state measure and post‑injection depression scores. Studies limited to bipolar depression or without a single‑dose phase (i.e. only repeated dosing with no single‑dose data) were excluded. Only human, English‑language studies were eligible. Database searches initially yielded 554 records plus two cross‑references; after deduplication and screening 8 studies met the inclusion criteria. Two reviewers independently extracted data using a pre‑piloted form, resolving discrepancies by discussion. The investigators summarised study characteristics including sample age ranges, medication washout practices, ketamine route/doses and infusion durations. All included studies measured depression primarily with the Montgomery‑Åsberg Depression Rating Scale (MADRS) or versions of the Hamilton Depression Rating Scale (HAM‑D); secondary measures in some studies included the Beck Depression Inventory (BDI) or QIDS‑SR. Dissociative or psychotomimetic effects were assessed predominantly with the CADSS (a clinician‑rated scale for dissociative states) or the BPRS (a more general psychiatric symptom scale, often using a positive symptoms subscale). Two studies used additional measures, one using a Visual Analog Scale (VAS) and another the 5D‑ASC (a multi‑dimensional scale for altered states of consciousness). Timing of assessments typically captured dissociation within 20–240 minutes post‑infusion and depressive symptoms at multiple points up to 1 month post‑dose. The review also documented trial design features relevant to bias: six of eight studies were placebo‑controlled (five double‑blind, one single‑blind), two used midazolam as an active comparator, and study sample sizes and blinding integrity were recorded as potential quality concerns.

Eight studies were included. Participant ages ranged 18–70 years. Most trials used a single ketamine infusion; five studies administered intravenous 0.5 mg/kg, one used 0.54 mg/kg, one used a 0.1–1.0 mg/kg range, and one used intranasal 0.5 mg/kg. Infusion durations varied from 1 to 40 minutes. Three studies implemented a two‑week washout of other psychoactive medications, while five studies allowed maintenance of existing regimens. Subjective effects: In the five studies reporting CADSS data, ketamine produced marked dissociative effects with peak CADSS increases typically at 20–40 minutes post‑infusion. Dissociation showed a dose‑dependent pattern, with clear separation from midazolam beginning at 0.5 mg/kg and little effect at lower doses (0.1–0.2 mg/kg). For intranasal ketamine, the CADSS increase at 40 minutes was reported as 2.2 ± 3.7 versus 1.2 ± 3.3 for saline. BPRS findings were heterogeneous across studies: some reported transient increases in positive symptoms (unusual thought content, conceptual disorganisation, hallucinatory behaviour) in a minority of participants, while others observed no meaningful change or even decreases on total BPRS; intranasal BPRS increases were small (4.3 ± 0.7 versus 4.0 ± 0). One study using the 5D‑ASC documented large interindividual variability across subscales (for example vigilance reduction, dread of ego dissolution, oceanic boundlessness), with some patients meeting high thresholds on particular dimensions. Correlation between subjective effects and antidepressant response: Across the eight studies, three (37.5%) reported statistically significant associations between subjective measures and antidepressant outcomes. CADSS: of five CADSS studies, two found significant correlations. One study reported correlations between increased CADSS at 40 minutes and percent improvement in HAM‑D‑17 at 230 minutes (r = -0.35, p = 0.007) and Day 7 (r = -0.41, p = 0.01). Another study found that change in CADSS during infusion correlated with MADRS response at 24 hours (Pearson r = -0.46, p = 0.03). These coefficients indicate moderately negative linear relationships (greater dissociation associated with greater antidepressant improvement), with the dissociation measures explaining up to about 21% of variance in response. Three CADSS studies reported no significant correlations; one large analysis (n = 94 for some comparisons) found non‑significant correlations (for example r = -0.19, n = 94 at Day 1 and r = -0.13, n = 92 at Day 3) and an intranasal study reported no association. BPRS: six studies included BPRS data and five reported correlations. Only one study demonstrated a significant correlation: increases in total BPRS score correlated with MADRS at Day 7 (r = -0.40, p = 0.04), with trends at Day 1 and Day 4. That finding implies BPRS change accounted for about 16% of variance in antidepressant response in that sample. Other BPRS analyses showed no significant correlations (one reported R2 < 0.05, p > 0.65) or only non‑significant trends. 5D‑ASC: the single study that reported 5D‑ASC data did not find a significant correlation with MADRS percentage change on the day of treatment; full statistical details were not presented. Study quality and bias: sample sizes were generally small (median sample size for correlative analyses = 20; mean = 37.6 ± 41.3). Six of eight studies were placebo‑controlled and five were double‑blind; two studies used midazolam as an active placebo (reported doses 0.03 and 0.045 mg/kg). The reviewers noted potential unblinding risk given ketamine’s perceivable psychoactive effects and heterogeneity across measurement timing and scale use, which limited opportunities for quantitative meta‑analysis.

The investigators interpret the evidence as providing limited support for an association between ketamine‑induced subjective states and antidepressant response: three of eight studies (37.5%) reported weak to moderate correlations, and these three comprised a substantial portion of the pooled sample. When present, correlations were moderate and dissociation or psychotomimetic change accounted for a minority of variance in antidepressant outcome (reported estimates in correlated studies ranged roughly 12–21%). The authors emphasise that most included trials did not comprehensively examine correlations across all timepoints, potentially missing critical windows for detecting associations. Mathai and colleagues discuss measurement limitations as a central issue. They note that commonly used scales such as the CADSS and BPRS were developed for other clinical contexts (CADSS for PTSD‑related dissociation, BPRS for broad psychiatric symptom change) and may incompletely capture the phenomenology of ketamine‑induced altered states. One cited analysis suggested the depersonalisation subscale of CADSS related most closely to antidepressant response, but qualitative data indicate broader psychological changes are not well measured. The authors suggest that scales specifically designed for acute hallucinogenic effects (for example measures capturing oceanic boundlessness, ego dissolution or mystical‑type experiences) might better characterise subjective phenomena relevant to therapeutic response. The discussion places these findings alongside other clinical observations: attempts to develop NMDA‑modulating antidepressants with reduced dissociative effects (for example memantine, lanicemine) have not replicated ketamine’s efficacy, implying dissociative or psychoactive effects may be relevant; conversely, some pharmacological data indicate dissociation and antidepressant actions can be dissociated (for example naltrexone blocking antidepressant but not dissociative effects in one report) and repeated‑dose esketamine trials show antidepressant effects persist despite attenuation of dissociation over time. The authors therefore state that dissociation may be one of several contributors to ketamine’s effect but does not appear to account for the majority of treatment variance. Key limitations acknowledged include the small number of eligible studies, heterogeneous measures and timing, generally small sample sizes that precluded a robust quantitative meta‑analysis, potential unblinding across trials, and possible publication or reporting bias if negative correlations were not reported in some studies. The authors recommend individual participant data meta‑analysis to increase power and harmonise analyses, improved measurement tools tailored to ketamine’s phenomenology, and further research into whether therapeutic frameworks that facilitate meaningful subjective experiences could enhance and prolong antidepressant effects. They caution clinicians to consider expectancy effects and to recognise that dissociation is common in clinical practice (noting esketamine label‑reported dissociation rates of 61–75% as presented in the paper) and that the clinical relevance of subjective experiences requires clearer empirical characterisation.