Effectiveness and factors associated with esketamine response during the 4-week induction period for treatment-resistant depression: post-hoc analysis of the real-world ESKALE study

Treatment-resistant depression (TRD) is defined here as major depressive disorder that has not responded to at least two adequate antidepressant trials during the current episode. TRD affects a substantial minority of people with major depression (estimates cited in the paper range from 30% to 55%) and is associated with greater comorbidity, poorer quality of life, higher suicide risk and increased health-care utilisation. Intranasal esketamine (ESK), administered with an oral antidepressant, received regulatory approval in 2019 for adults with TRD based on several clinical trials (TRANSFORM, SUSTAIN and ESCAPE-TRD) and subsequent real-world studies that showed effectiveness and an established safety profile for this indication. Early improvement on treatment has previously been reported as a sensitive predictor of later stable response or remission, but relatively few studies have analysed which baseline or early-treatment factors predict response to ESK. Samalin and colleagues report a post-hoc analysis of the French, multicentre, non-interventional ESKALE study (a 12-month, secondary-data study) to characterise the evolution of depressive symptoms during the 28-day induction period with ESK and to identify factors associated with early response at 1 and 4 weeks after initiation. The authors note prior pooled and machine-learning analyses that have suggested candidate predictors (for example age, employment, number of prior antidepressant failures, and early clinical improvement), and set out to test potential associations in this real-world cohort of patients with highly treatment-resistant illness.

This is a post-hoc analysis of the ESKALE study, a French multicentre, non-interventional 12-month cohort using secondary data from medical records of patients who initiated esketamine between October 2019 and July 2021. The study complied with relevant French and European data-protection law and was conducted under the MR-004 methodology for secondary-data studies; it was registered in the Health Data Hub. Patients were informed and no one objected to data use. For the present analysis, eligible patients received at least one dose of ESK and had a baseline Montgomery–Åsberg Depression Rating Scale (MADRS) score and at least one MADRS assessment between Week (W)1 and W4. The primary longitudinal outcome was absolute change in total MADRS score from baseline to W1 and W4. Changes were analysed with a mixed model for repeated measures (MMRM) with medical visit (W1–W4) as a fixed effect and patient age, gender and baseline MADRS score as adjustment covariates; an unstructured covariance matrix modelled within-patient measurements. The pre-specified binary response outcome was ≥50% reduction in MADRS total score from baseline. Dissociative adverse events (AEs) were captured from physician reports and coded using MedDRA terminology; the paper defines dissociation broadly (depersonalisation, derealisation, perceptual detachment, altered sense of self or surroundings, emotional numbing or dissociative amnesia). Potential factors associated with response at W1 and W4 were investigated using logistic regression. Candidate covariates included baseline sociodemographic and clinical characteristics, prior treatments for the current episode, combined therapy at initiation, ESK dose up to W4, and occurrence of dissociation within the first week and within the 28-day induction period. Multiple imputation (fully conditional specification) was used for covariates with missing data up to a 20% threshold. Variables with p ≤ 0.20 in univariate analyses and with <20% missing data were considered for multivariate models; age, gender and baseline MADRS were retained in multivariate models regardless of univariate significance. The multivariate significance threshold was α = 0.05 (two-sided). Analyses were performed using SAS version 9.4.

Disposition and baseline characteristics: Of 157 patients enrolled in ESKALE, 128 (81.5%) had at least one MADRS assessment between W1 and W4 and were included in this post-hoc analysis. Of these 128, 125 (97.7%) were followed for the full 28-day induction period and 107 (83.6%) received esketamine throughout the induction period. Twenty-one patients (16.4%) discontinued ESK before day 28; reasons reported included unsatisfactory therapeutic effect (17 patients), patient request (6), and adverse events (3), with multiple reasons possible. Baseline clinical characteristics of the analysed cohort show a severely treatment-resistant population: median of three major depressive episodes (Q1–Q3: 2–5), 48.4% had at least one prior suicide attempt, and a median of six 'well-conducted' prior treatment lines for the current episode (Q1–Q3: 3–8), with 48.4% having more than six prior lines. Mean baseline MADRS total score was 32.2 (SD 7.9). MADRS change during induction: The adjusted mean change in MADRS from baseline to W1 was −7.46 points (95% CI: −9.23 to −5.69; p < 0.0001). MADRS scores continued to decline over the induction period, with an adjusted mean change at W4 of −13.55 points (95% CI: −15.3 to −11.8; p < 0.0001). Response rates over time: Reported response rates (≥50% reduction in MADRS) increased across the induction period. The proportions and sample sizes reported were: 19.4% at W1 (n = 103), 26.7% at W2 (n = 105), 33.3% at W3 (n = 102), and 47.4% at W4 (n = 95). (The extracted text indicates variable denominators across weeks corresponding to available assessments.) Factors associated with response: In univariate analyses, two baseline parameters met the p ≤ 0.20 threshold for association with response at W1: severe versus mild-to-moderate baseline depression (p = 0.0962) and occurrence of dissociation during the first week (p = 0.0394). In multivariate logistic regression, only occurrence of a dissociative event during the first week was significantly associated with response at W1 (p = 0.0213); severe baseline depression showed a trend (p = 0.0538). No factors were identified as significantly associated with response at W4 in the logistic regression analyses. Specifically, occurrence of dissociation in the first 7-day and 14-day periods was not significantly associated with response at W4 (p = 0.3033 and p = 0.5216, respectively). Incidence of dissociation: During the 28-day induction period, 43 of 128 patients (33.6%) had one or more dissociative events reported: 36 patients (28.1%) during the first week and a further 7 patients (5.5%) between W1 and W4.

Samalin and colleagues interpret these findings as further real-world evidence that depressive symptoms often improve rapidly after initiating esketamine during the 28-day induction phase in patients with TRD. They highlight adjusted mean MADRS reductions of approximately 7.5 points at W1 and 13.6 points at W4 in this highly treatment-resistant cohort. The authors contrast this with a larger MADRS reduction reported in the TRANSFORM 2 trial (−21.4 at W4), and suggest that differences in the degree of treatment resistance between samples (ESKALE patients had more prior treatment failures) may explain some of the discrepancy. The authors focus on the relationship between early dissociative experiences and early antidepressant response. They report that experiencing dissociation during the first week was significantly associated with response at W1 but not with response at W4. They note that the term dissociation encompasses heterogeneous subjective phenomena and that prior pooled and individual-study analyses have produced mixed results: some reports link short-term dissociative symptoms to ketamine response, whereas other analyses (including work by Chen et al., cited by the authors) did not find an association. The authors emphasise that dose selection should not aim to induce dissociation, but to optimise the risk–benefit balance. The paper acknowledges several limitations inherent to the non-interventional, secondary-data design. Weekly MADRS assessments were not systematically recorded in medical files (reported missing data of 33%–39% at each weekly visit), so the analysed sample was a subset (128 of 157) with sufficient data, and multivariate analyses were therefore conducted on a relatively small and selected group. Safety data including dissociation were captured from routine physician reporting rather than standardised scales such as the Clinician Administered Dissociative States Scale (CADSS), which may limit sensitivity and comparability. The authors also note that standard assessment tools may not fully capture esketamine’s unique subjective and experiential effects and that not all potentially relevant prognostic factors were available in the collected data. They suggest that future studies should incorporate esketamine- or ketamine-specific instruments and broader measures of quality of life and experiential outcomes. In summary, the authors conclude that response rates to esketamine increased across the 4-week induction period in this real-world TRD sample, that first-week dissociation was the only factor significantly associated with response at Week 1, and that no baseline or early-treatment factors were significantly associated with response at Week 4. They call for further research to investigate prognostic factors for esketamine response in TRD.