Treatment Response With Esketamine Nasal Spray Plus an Oral Antidepressant in Patients With Treatment-Resistant Depression Without Evidence of Early Response: A Pooled Post Hoc Analysis of the TRANSFORM Studies

Major depressive disorder (MDD) causes substantial morbidity and disability, and a sizeable proportion of patients do not achieve an adequate response after multiple oral antidepressant trials. The term treatment-resistant depression (TRD) is used for patients who fail to respond to ≥ 2 antidepressants in the current major depressive episode, and for many patients the cumulative time spent on ineffective treatments is substantial. Rapid-acting therapies such as ketamine analogues have raised expectations for early improvement, yet individual trajectories vary and lack of an early response does not necessarily predict ultimate nonresponse. Turkoz and colleagues set out to examine whether adults with TRD who do not show an early response to treatment with esketamine nasal spray plus a newly started oral antidepressant (ESK + AD) might nevertheless achieve a clinically meaningful response by the end of a standard 4-week induction phase. Using pooled data from two Phase III, double-blind, active-controlled TRANSFORM studies, the analysis specifically estimated day 28 response rates among patients who failed to meet predefined early-response criteria at day 2 or at days 2 and 8.

This is a post hoc pooled analysis of TRANSFORM-1 and TRANSFORM-2, two multicentre, Phase III, double-blind, active-controlled trials comparing intranasal esketamine plus a newly initiated oral antidepressant with placebo nasal spray plus a newly initiated oral antidepressant. Each trial comprised a screening/prospective observation phase (up to 4 weeks), a 4-week double-blind induction phase and follow-up. Randomisation used a computer-generated schedule and study sprays (esketamine or placebo) were packaged identically; placebo contained a bittering agent to mimic taste. Independent, remote raters blinded to treatment assignment conducted the primary efficacy assessments. Patients were aged 18–64 with recurrent MDD or a single episode ≥ 2 years, absence of psychotic features, and moderate-to-severe depression (Inventory of Depressive Symptomatology ≥ 34 and MADRS ≥ 28). Entry required nonresponse to ≥ 1 but ≤ 5 oral antidepressants at screening, with confirmation of nonresponse to the current antidepressant during the prospective observation phase; at randomisation all patients met the study definition of TRD (nonresponse to ≥ 2 antidepressants in the current MDE). Key exclusions included recent suicidal behaviour/intent, psychotic disorder, bipolar disorder, recent moderate-to-severe substance use disorder, and positive drug screens. Intranasal study drug was self-administered at the study site under supervision twice weekly for 4 weeks. Esketamine dosing was 56 mg or 84 mg; patients assigned to esketamine started at 56 mg with option to titrate to 84 mg after day 1. All patients also initiated an open-label oral antidepressant (escitalopram, sertraline, duloxetine, or venlafaxine XR) and titrated to maximally tolerated labelled doses. The primary efficacy endpoint for the original trials was change in MADRS total score from baseline to day 28. For this post hoc analysis early response was defined two ways: ≥ 50% decrease in MADRS total score by day 2, and ≥ 50% decrease by day 2 that was maintained at day 8. The analytic population included randomized patients who received ≥ 1 dose of intranasal medication and ≥ 1 dose of oral antidepressant; 565 patients constituted the full analysis set, and 518 patients had MADRS data at day 28 and were included in the present day-28 analyses. Comparisons of response rates used the Cochran-Mantel-Haenszel test controlling for study ID, region and class of oral antidepressant. To estimate conditional probabilities of day-28 response among early nonresponders, the investigators used multiple logistic regression models adjusted for demographic and baseline characteristics, multivariate adaptive regression splines (MARS) to identify influential covariates, and repeated-measures generalized linear mixed models to leverage all visit data. Bootstrapping was used to generate confidence intervals around predicted probabilities. Safety and tolerability were summarised; adverse events and discontinuations were reported in supplementary material.

The analysis focused on 518 patients with day-28 MADRS observations (ESK + AD, n = 310; AD + PBO, n = 208). Across the full sample, esketamine plus oral antidepressant produced higher rates of categorical response (≥ 50% improvement in MADRS from predose) than oral antidepressant plus placebo at each visit. At day 2, response rates were 17.3% with ESK + AD versus 9.4% with AD + PBO (P = .01); at day 8, 15.8% versus 8.3% (P < .01). By day 28, 58.7% of ESK + AD patients met response criteria compared with 45.2% of AD + PBO patients (P < .001). Among patients who did not meet early-response criteria, outcomes at day 28 still favoured esketamine. For the 411 day-2 nonresponders, 54.9% of those receiving ESK + AD versus 44.3% of those receiving AD + PBO achieved response at day 28 (P < .01). In the subgroup of 382 patients who were nonresponders at both day 2 and day 8, 52.1% with ESK + AD versus 42.4% with AD + PBO achieved response by day 28 (P = .01). Multivariable logistic regression adjusting for baseline factors produced similar results: among day-2 nonresponders the odds ratio (OR) for response at day 28 with ESK + AD versus AD + PBO was 1.61 (95% CI, 1.09–2.40; estimated probabilities 0.56 vs 0.44). For day-2-and-8 nonresponders the OR was 1.56 (95% CI, 1.04–2.35; probabilities 0.52 vs 0.41). MARS identified predictors associated with day-28 response including lifetime C-SSRS ideation/behaviour (no event), baseline MADRS total score and baseline Sheehan Disability Scale score (greater disability associated with lower likelihood of response); after adjustment MARS-derived ORs favoured esketamine (day-2 nonresponders OR 1.66, 95% CI 1.08–2.56; day-2-and-8 nonresponders OR 1.72, 95% CI 1.10–2.71). Repeated-measures generalized linear mixed models, which used all visit data, likewise indicated higher likelihood of response with ESK + AD: among day-2 nonresponders 53.3% versus 43.3% responded at day 28 (OR = 1.50; 95% CI 1.10–2.00), and among day-2-and-8 nonresponders 49.6% versus 40.1% (OR = 1.43; 95% CI 1.04–1.91). Demographically, no major baseline differences were apparent between treatment groups for day-2 nonresponders who later responded; however, among day-2-and-8 nonresponders who responded at day 28 a higher proportion were female in the ESK + AD group. Safety and tolerability findings were consistent with the primary trials and no new or unexpected safety concerns were noted; treatment-emergent adverse events, serious adverse events and discontinuation rates are summarised in supplementary tables.

Turkoz and colleagues interpret their pooled post hoc findings to indicate that although esketamine plus an oral antidepressant produces rapid effects in a subset of patients, many individuals who do not show an early full response still have a meaningful chance of responding by the end of a 4-week induction. The investigators emphasise that completion of the full 4-week induction course may be appropriate to optimise treatment effect for patients with TRD who tolerate the treatment, even if no full response is evident after the first two doses. The discussion situates these results alongside prior observations with ketamine, noting small studies in which repeated infusions produced gradual improvements across multiple administrations. The authors also contrast the esketamine trials with typical adjunctive antidepressant trials: in the TRANSFORM studies a new oral antidepressant was initiated at randomisation (as recommended by regulators) rather than adding esketamine to an ongoing partially effective antidepressant, and visit frequency (twice-weekly supervised sessions) greatly exceeded that in standard oral antidepressant trials. The investigators acknowledge that the new oral antidepressant and increased clinical contact may have contributed to improvements seen in both groups, but highlight that differences between ESK + AD and AD + PBO widened over weeks 3 and 4. Key limitations noted by the authors include the post hoc nature of the analysis, the strict inclusion and exclusion criteria of the source trials which may limit generalisability to routine clinical populations, and the use of a rigorous predefined categorical response threshold that may not capture more nuanced clinical improvement recorded by investigators. They conclude that, within these trial settings, patients with TRD who do not achieve an early response still have significantly greater odds of response at day 28 with esketamine plus an oral antidepressant compared with oral antidepressant plus placebo, and that continuation of the induction course should be considered in light of individual clinical and practical factors.