Reconsidering “dissociation” as a predictor of antidepressant efficacy for esketamine

Mathai and colleagues place this study in the context of longstanding uncertainty about whether subjective drug effects—particularly drug-induced “dissociation”—are mechanistically related to the antidepressant effects of ketamine and its enantiomer, esketamine. They review historical and conceptual debates about dissociation, noting its roots in 19th- and early 20th-century accounts and its later operationalisation in instruments such as the Clinician-Administered Dissociative States Scale (CADSS). Previous clinical work with ketamine has produced inconsistent findings: some studies report that greater acute dissociation predicts better antidepressant response, while others do not. The authors also note that classic psychedelics show clearer links between particular qualities of the acute experience (for example, mystical-type experiences) and clinical benefit, and that the subjective profile of dissociatives may differ meaningfully from that of classic psychedelics. The present paper aims to clarify whether acute dissociation predicts antidepressant efficacy for esketamine by conducting a pooled, post hoc analysis of two Phase III trials (TRANSFORM-1 and TRANSFORM-2). Specifically, the investigators test whether CADSS-measured dissociation interacts with time to predict change in depression severity during the 4-week induction phase, and whether dissociation after the first dose predicts rapid antidepressant benefit within 24–48 hours. The authors frame this as the largest formal assessment to date of dissociation’s therapeutic relevance for esketamine and argue that a comprehensive, across-trials approach is warranted given prior mixed results.

This secondary, pooled analysis combined de-identified data from two randomised, double-blind Phase III trials of intranasal esketamine plus an oral antidepressant versus intranasal placebo plus oral antidepressant: TRANSFORM-1 (fixed-dose arms: 56 mg, 84 mg, and placebo) and TRANSFORM-2 (flexible-dose arm: 56 mg or 84 mg and placebo). Participants were adults aged 18–64 with treatment-resistant depression, defined as lack of response to at least two adequate antidepressant trials in the current episode. Study drug was administered twice weekly over an initial 4-week induction phase. The Yale Open Data Access (YODA) Project provided the datasets; the Johns Hopkins IRB determined the secondary analysis did not constitute human subjects research. Depression severity was measured with the Montgomery-Åsberg Depression Rating Scale (MADRS), a 10-item clinician-rated scale (range 0–60) assessed by independent remote raters blind to protocol details. Dissociation was assessed on-site using a 23-item clinician-administered version of the CADSS, administered before dosing, at 40 minutes post-dose (approximate peak), and at 1.5 hours post-dose. The authors selected the 40-minute CADSS score for primary analyses and included CADSS data from all eight induction doses. Statistical analyses were conducted in R (version 3.6.3). The primary model was a linear mixed-effects model with total MADRS score as the outcome and baseline MADRS, treatment condition-by-time interaction, and CADSS-by-time interaction as fixed effects. Random intercepts for participant accounted for repeated measures, and random intercepts for study accounted for clustering within trials. Significance of the CADSS-by-time interaction was assessed with type II Wald chi-square tests (ANOVA in the car package). The team also generated model-predicted MADRS means at days 1 and 28 for a range of CADSS values (0, 2, 4, 8, 16) to visualise relationships. Separately, a planned simpler regression tested whether day 1 CADSS predicted day 2 MADRS, controlling for baseline MADRS and treatment condition.

The pooled analysis included 576 participants: placebo (TRANSFORM-1 + TRANSFORM-2) N = 227; fixed-dose esketamine 56 mg (TRANSFORM-1) N = 115; fixed-dose esketamine 84 mg (TRANSFORM-1) N = 116; flexible-dose esketamine 56/84 mg (TRANSFORM-2) N = 118. Mean baseline MADRS for the placebo group was reported as 37.6 (SD = 5.14). The extracted text does not clearly report mean baseline MADRS values for the other groups. Plots of raw MADRS and CADSS scores over time and scatterplots of individual CADSS values versus change in MADRS were described but not reproducible from the extracted text. In the primary linear mixed model, the CADSS-by-time interaction was not significant (p = 0.7), indicating no evidence that time-varying dissociation scores, as measured by CADSS at 40 minutes, predicted change in MADRS across the 4-week induction phase. Model-predicted MADRS values were generated across CADSS values from 0 to 16 for days 1 and 28, but the main inferential result was null for the interaction term. In the separate regression examining rapid benefit, each additional point on day 1 CADSS was associated with a -0.04 change in day 2 MADRS score (95% CI -0.08 to -0.002; p = 0.04). The authors present this as a statistically significant but very small effect size. They also note that dissociation levels observed in these esketamine trials were lower than those reported in many ketamine studies, although numerical comparisons are not provided in the extracted text.

Mathai and colleagues interpret their findings as providing no evidence that acute dissociation, as measured by the CADSS in the specific context of the TRANSFORM trials, either facilitates or interferes with esketamine’s antidepressant effects. They emphasise that this pooled analysis is the largest formal assessment to date and that its null findings are broadly consistent with a prior industry analysis that reached similar conclusions despite methodological differences. The authors argue that pooling both TRANSFORM-1 and TRANSFORM-2 and examining dissociation across all induction sessions provides a more comprehensive test than analyses limited to one trial or to fewer time points. Several explanations and caveats are discussed. The investigators note that dissociation levels in the esketamine trials were relatively low compared with many racemic ketamine studies, which could reflect pharmacological, dosing, or delivery differences; however, variable findings across ketamine studies with similar administration methods suggest other factors may also matter. The potential influence of researcher and site attitudes on patient expectancies and experiences is raised as a possible modifier of subjective effects and outcomes, and the authors highlight the broader importance of contextual and extra-pharmacological variables. They further observe that the CADSS was developed to index pathological, trauma-related dissociation and may insufficiently capture other acute subjective phenomena—such as trance-like states, absorption, or mystical-type experiences—that could plausibly relate to therapeutic change. Subcomponents of the CADSS (for example, depersonalisation items) have in some studies been more predictive of outcome, and qualitative data hint that experiences of detachment may promote perspective shifts relevant to therapy. The authors acknowledge several limitations of the current analysis. Flexible dosing in TRANSFORM-2 precluded distinguishing between 56 mg and 84 mg doses in that trial, reducing the ability to probe dose-related effects across the pooled sample. Differences between ketamine and esketamine in pharmacology, delivery, and patient preference are noted as undercharacterised in the present data. The authors recommend future research that employs instruments better tailored to capture the full range of acute subjective effects, includes subgroup and dose-response analyses, and considers the role of psychological support and therapeutic context. They also note the clinical imperative to understand and mitigate any dissociative phenomena that may pose safety risks while, conversely, attending to subjective experiences that might promote benefit. Overall, the authors conclude that more nuanced conceptualisations and measurement of subjective drug experience are needed to clarify whether—and how—acute states relate to antidepressant response with ketamine derivatives.