Published on 23 September 2026
Esketamine without an oral antidepressant: what does a placebo-controlled trial show?
JAMA Psychiatry · 2025; 82(9): 877-887 · Janik et al.
DOI 10.1001/jamapsychiatry.2025.1317
PMID 40601310
Scientific 67
Editorial 75
The essentials
A phase 4, randomized, double-blind, placebo-controlled trial conducted from November 2020 to January 2024 at 51 outpatient centers in the United States. Intranasal esketamine was given alone, after every oral antidepressant (OAD, in the trial’s own terminology) had been tapered off for at least two weeks. The authors describe their own work as “this first placebo-controlled, monotherapy randomized clinical trial,” and state in their introduction that, “to our knowledge, no studies have assessed esketamine’s effects as monotherapy without concomitant use of an OAD.” The efficacy analysis included 378 participants, the safety analysis 476. At four weeks, the adjusted mean difference on the MADRS scale was 5.1 points in favor of the 56 mg dose and 6.8 points in favor of the 84 mg dose, with effect sizes of 0.48 and 0.63. A measurable effect was already present at day 2. Two reservations dominate the reading. The blinded phase lasted only four weeks, and the twelve-week extension that followed it was open label and uncontrolled, so nothing here supports a conclusion about durability. And the conflicts of interest are structural: seven authors are employees of the company that markets the product, all of them are shareholders, five hold a patent on the molecule, in a trial that the same company funded and whose employees took part in designing, conducting, analyzing, and deciding to submit.
Context
Intranasal esketamine entered the management of treatment-resistant depression under one constant constraint: it had always been evaluated, approved, and administered together with an oral antidepressant. That combination was not an argued pharmacological choice, it was simply the framework of the registration trials. Nobody knew whether the oral antidepressant contributed to the result, or merely accompanied the active product.
The question is not theoretical. Some patients with treatment-resistant depression can no longer tolerate oral antidepressants, refuse to continue them after repeated failures, or have interactions that complicate the combination. Knowing whether esketamine works on its own changes the conversation with these patients.
One nuance is needed on the claim of priority. The authors themselves restrict it to the placebo-controlled design, and they cite in their own discussion an earlier, uncontrolled precedent: a post hoc analysis of a subgroup of 50 patients from the open-label extension study SUSTAIN-3 who received esketamine monotherapy for three months or longer. The claim therefore holds for a randomized trial controlled against placebo, not for the first observation of esketamine given without an oral antidepressant.
The mechanism
What the trial measured, and what it did not measure
The underlying hypothesis is that esketamine’s rapid antidepressant effect does not depend on potentiating an oral antidepressant, and that it follows a pharmacological pathway distinct from classic monoaminergic modulation. This is the hypothesis the design tests, but it tests it through a clinical outcome, not through its mechanics.
| Element | Status in this trial |
|---|---|
| Clinical effect without an oral antidepressant | Measured |
| FindingThis is the primary outcome, and the only thing the design establishes directly. | |
| Speed of onset | Measured as early as day 2 |
| FindingThe 24-hour measurement is the key secondary end point, prespecified in the protocol and in the testing hierarchy. It documents clinical kinetics, not the pharmacological pathway involved. | |
| Signaling pathway involved | Not measured |
| FindingThe protocol included pharmacokinetic and biomarker objectives, but the publication reports no biological marker, no imaging, and no mechanistic intermediate outcome. The article says nothing about the mechanism itself. | |
| Direct comparison with the combination | Absent |
| FindingThe comparator is an intranasal placebo, not esketamine combined with an oral antidepressant. The trial therefore cannot say whether monotherapy performs as well as the combination. | |
This last point is the one that gets lost fastest in commentary. Showing that esketamine alone beats placebo is not the same as showing that it equals esketamine combined with an oral antidepressant. Noninferiority is not tested here, and an absence of comparison cannot be read as equivalence.
The study at a glance
| Population | |
| Adults aged 18 or older with major depressive disorder without psychotic features (DSM-5), with 25% or less improvement under at least two oral antidepressants during the current episode, and a score of at least 34 on the IDS-C30 at screening. Any antidepressant and any adjunctive antipsychotic were tapered and then stopped at least two weeks before randomization. Recruitment ran from November 2020 to January 2024 at 51 outpatient centers in the United States. The efficacy analysis set includes 378 participants, the safety analysis set 476. | |
| Intervention | |
| Fixed-dose intranasal esketamine, 56 mg or 84 mg, self-administered on site under supervision, twice weekly for four weeks, with at least two hours of observation after each dose. Efficacy analysis set: 86 participants on 56 mg and 95 on 84 mg. Safety set: 105 and 121. | |
| Comparator | |
| Matching intranasal placebo, identical in appearance, same dosing frequency. Efficacy analysis set: 197 participants. Safety set: 250. Randomization 1 to 1 to 2 among 56 mg, 84 mg, and placebo, by permuted blocks of 4, stratified by center and by antidepressant treatment status at screening. | |
| Outcomes | |
| Primary: change in total MADRS score from baseline to day 28. Key secondary: change in MADRS score at day 2, 24 hours after the first dose. Other outcomes, with no control for alpha-risk inflation: response, defined as a reduction of at least 50% in MADRS score; remission, defined as a MADRS score of 10 or less; overall severity measured by the CGI-S; PHQ-9 score; and tolerability. | |
| Design | |
| Phase 4, parallel-group, randomized, double-blind, placebo-controlled trial. Four phases: a screening period of up to seven weeks, a four-week double-blind treatment phase, an optional twelve-week open-label phase, then follow-up about one week after the last dose. Primary analysis by mixed-effects model for repeated measures on observed data, in the efficacy analysis set, which includes randomized participants who received at least one dose and met the MADRS severity criteria defined before randomization. SAS version 9.4. |
Quality control
| Point checked | Verdict |
|---|---|
| Randomization and double blinding | Present |
| FindingComputerized randomization by permuted blocks of 4, stratified by center and by antidepressant treatment status, with a matching intranasal placebo identical in appearance. Efficacy raters were kept separate from safety raters, specifically to limit functional unblinding. | |
| Integrity of blinding | Tested and imperfect |
| FindingParticipants were asked directly on day 28. Among esketamine recipients, 59 of 83 in the 56 mg arm (71.1%) and 72 of 92 in the 84 mg arm (78.3%) were strongly convinced they had received the active product. Among placebo recipients, 90 of 192 (46.9%) were strongly convinced they had received placebo. The authors explicitly acknowledge this risk of functional unblinding and list it among the trial’s limitations. Their argument against it rests on post hoc analyses and outside studies, not on an internal correction built into the design. | |
| Primary analysis | Appropriate |
| FindingA mixed-effects model for repeated measures, suited to longitudinal data with missing values. One point of vocabulary: this is not an intention-to-treat analysis. The analysis plan defines an efficacy analysis set restricted to randomized participants who received at least one dose and met the MADRS severity criteria before randomization, and the primary analysis is based on observed data. A post hoc analysis on all randomized participants, regardless of the severity criterion, found differences of 4.6 and 5.7 points, somewhat smaller but in the same direction. | |
| Multiplicity | Controlled and prespecified |
| FindingThe statistical analysis plan and supplementary material describe a sequential gatekeeping method with the Hochberg procedure, an overall alpha of 0.05, and a truncation parameter of 1. The day 2 outcome could be tested only after the null hypothesis for the primary outcome was rejected for both doses. This point is solid. No alpha-risk control, however, was planned for the other outcomes. | |
| Sample size and power | Prespecified |
| FindingThe sample size calculation assumed a standardized effect of 0.45, corresponding to a difference of 5.4 points for a standard deviation of 12, with a 20% dropout rate. It required 356 participants meeting the severity criteria to reach 85% power per dose and 93% power to reject at least one null hypothesis. The trial enrolled 378 in this set, so it reached its planned power. Worth noting: the observed effect at 56 mg, 5.1 points, runs slightly below what had been assumed. | |
| Length of the blinded phase | Four weeks |
| FindingThe twelve-week extension is open label, single arm, with no comparator. The authors themselves describe its results as exploratory. No controlled data exist beyond one month. Durability of monotherapy cannot be assessed from this trial. | |
| External validity | Restricted |
| FindingA limitation the authors acknowledge. Excluded were current psychotic disorders, bipolar and related disorders, current obsessive-compulsive disorder, intellectual disability, autism spectrum disorder, borderline and antisocial personality disorder, recent (past six months) moderate or severe substance use disorder, recent suicidal ideation or behavior, and prior exposure to ketamine or esketamine. Racial and ethnic diversity in the sample is low: 86.1% white participants in the safety analysis set. | |
| Funding | Fully industry |
| FindingThe trial was funded by Janssen Research & Development. The authors state that employees of the sponsor took part in the design and conduct of the study, in the collection, management, analysis, and interpretation of the data, in preparing and approving the manuscript, and in the decision to submit it. Writing assistance was also funded by the sponsor. | |
| Authors’ conflicts of interest | Structural |
| FindingSeven authors are employees of Johnson & Johnson, and all are shareholders of Johnson & Johnson. Five of them hold a patent on esketamine for the treatment of depression, whose rights are assigned to Johnson & Johnson. The academic authors also declare research funding, consulting activity, and paid engagements, several of them with the sponsor. These are not one-off honoraria, they describe an ongoing position of interest. This is the main limitation of this article. | |
| Independent replication | Not documented here |
| FindingNeither the publication nor its supplementary material reports a controlled monotherapy trial conducted by a team with no ties to the sponsor. This needs to be said plainly: this analysis did not carry out a systematic literature search on this point, and the absence of such a mention in the source article is not proof that no such trial exists. | |
Results
| Outcome | Result |
|---|---|
| MADRS at day 28, 56 mg dose | Difference 5.1 points, standard error 1.42, 95% CI 2.33 to 7.91, P < .001, Cohen’s d 0.48 |
| ReadingA moderate effect size. Statistical significance is clear; the clinical relevance of a five-point difference on a scale that runs to 60 is a matter for case-by-case judgment. The lower bound of the confidence interval, 2.33 points, sits close to the two-point threshold the authors treat as clinically significant. | |
| MADRS at day 28, 84 mg dose | Difference 6.8 points, standard error 1.38, 95% CI 4.07 to 9.48, P < .001, Cohen’s d 0.63 |
| ReadingA larger effect size than the lower dose. The comparison between the two doses was not planned in the analysis plan and cannot be read as demonstrating the superiority of one over the other. The confidence intervals of the two doses overlap substantially. | |
| Effect at day 2 | 56 mg: 3.8 points, 95% CI 1.22 to 6.29, P = .004. 84 mg: 3.4 points, 95% CI 1.00 to 5.89, P = .006 |
| ReadingA measurable effect as early as 24 hours, consistent with what is known of the molecule, and tested only after the primary outcome cleared, following the Hochberg procedure. Worth noting: at this time point, the 56 mg dose numerically outperforms the 84 mg dose, which does not support a dose-response reading. | |
| Number needed to treat, response | 6.5 for 56 mg, 95% CI 1.8 to 11.3, and 7.1 for 84 mg, 95% CI 1.7 to 12.5 |
| ReadingResponse defined as a reduction of at least 50% in MADRS score. These values and their intervals are calculated and published by the authors. The intervals are wide, reflecting the modest size of each arm. | |
| Number needed to treat, remission | 12.3 for 56 mg, 95% CI -0.4 to 25.0, and 6.7 for 84 mg, 95% CI 2.6 to 10.9 |
| ReadingRemission defined as a MADRS score of 10 or less. The decisive point is the interval for the 56 mg dose: it crosses the null value, meaning the benefit on remission is not established at this dose. The apparent gap between the two doses should not be read as an established ranking between them. | |
| Discontinuations for adverse events, blinded phase | 1.0% under 56 mg, 4.1% under 84 mg, 1.2% under placebo |
| ReadingThat is 1 patient out of 105, 5 out of 121, and 3 out of 250. Low over four weeks, with a higher figure at the high dose. No serious treatment-related adverse event was reported under esketamine. These data say nothing about long-term tolerability, or about the monitoring burden that administration requires. | |
| Most frequent adverse events, combined doses | Nausea 24.8%, dissociation 24.3%, dizziness 21.7%, headache 19.0% |
| ReadingAmong 226 patients treated with esketamine, against 8.4%, 2.8%, 7.2%, and 8.8% respectively among 250 patients on placebo. The contrast on dissociation is the one that weighs most heavily on the credibility of blinding. | |
| Other efficacy outcomes | All pointing the same direction, but with no control for multiplicity |
| ReadingResponse and remission rates were higher under esketamine at every measurement point, self-reported PHQ-9 improvement at day 28 of 3.7 points under 56 mg and 4.1 points under 84 mg, and the proportion of patients rated markedly ill or worse on the CGI-S fell from 79.0% at baseline to 26.8% and 22.5% under esketamine against 51.6% under placebo. The convergence is real, but only the primary and key secondary outcomes carry alpha-risk control. The rest are descriptive and cannot be presented as demonstrations. | |
Critical appraisal
| Domain | Judgment |
|---|---|
| Quality of the design | Solid |
| FindingRandomized, double-blind, placebo-controlled, across 51 centers, with a prespecified analysis plan, an explicit testing hierarchy, and power reached. On design alone, this is a well-built trial that fills a real evidence gap. | |
| Scope of the claim | Calibrated by the authors |
| FindingThe title announces a monotherapy, and the trial does evaluate a monotherapy. The authors explicitly restrict their priority claim to the placebo-controlled randomized design, and they themselves cite an uncontrolled precedent, the 50-patient monotherapy subgroup of the open-label SUSTAIN-3 extension. Stated in these terms, the claim is accurate. | |
| Time horizon | Short |
| FindingFour weeks under controlled conditions, for a condition managed over months. The twelve weeks that follow are open label, with no control group, and do not answer the question. What happens afterward remains unresolved. | |
| Independence | Compromised |
| FindingFunding, design, analysis, and writing all sit on the sponsor’s side, with authors personally tied to the patents. This configuration does not invalidate the result, but it lowers the degree of confidence until an independent team has reproduced it. | |
Level of evidence
Confidence is high that an effect exists in the short term without a concurrent oral antidepressant: the design fits the question, the sample reached its prespecified target, the testing hierarchy was respected, the difference is significant for both doses, and the direction of the effect is consistent. Confidence is lower on the exact size of that effect, whose confidence intervals remain wide, on its persistence beyond four weeks, and on any ranking between the two doses.
What is demonstrated: over four weeks, in adults with treatment-resistant depression selected under strict criteria, intranasal esketamine given alone outperforms an intranasal placebo on the MADRS score, at day 28 as well as at day 2. What is suggested: that this benefit persists beyond that point, a hypothesis consistent with what was observed in the open-label phase but not tested under blinding here. What amounts to expert opinion: the idea that the oral antidepressant could be dispensed with in practice, a claim that would require a direct comparison between monotherapy and combination therapy, absent from this trial.
One last point, and it belongs to editorial judgment more than to the data. A trial funded by industry is not a false trial, and dismissing it on that basis alone would be lazy reasoning. But when the sponsor funds, designs, analyzes, writes, and decides to submit, and when several authors hold the patent on the product being evaluated, the guarantee that method alone provides is no longer sufficient by itself. What is missing here is not fixed by reader vigilance, it is fixed by replication.
The colleague test
What an experienced colleague might say about this study in two minutes, between two consultations.
“So esketamine works without an oral antidepressant, first randomized trial against placebo on that question, decent effect size, six to seven points on the MADRS. Except seven authors are employees of the company, five hold the patent, three out of four treated patients guessed they were on the active drug, and we only have four weeks under blinding. I’ll factor it in. I’m not changing anything until someone else has redone it.”
Translated for practice: this trial is information worth keeping for later, not a reason to change a prescription today. The exact conditions of administration, and whether combination with an oral antidepressant remains a condition of the marketing authorization, depend on the regulatory framework where you practice: check them at the point of prescribing.
What you can do with this
- Know that the monotherapy question now has a placebo-controlled randomized trial behind it, and be able to say so to a patient who asks, or a colleague who raises it.
- Do not transpose this result into practice without checking local rules first: most current marketing authorizations for esketamine require combination with an oral antidepressant, and any off-label use is subject to conditions that vary by jurisdiction.
- Keep the two claims separate: esketamine alone outperforms placebo, which this trial shows, and esketamine alone equals the combination, which it does not show.
- Remember that only two outcomes carry statistical protection here, MADRS at day 28 and MADRS at day 2. Everything else, response, remission, CGI-S, PHQ-9, is descriptive and converges without demonstrating.
- Read a trial’s conflict-of-interest statement before its results section, not after. It sets the weight everything else should carry.
- Keep the time horizon in mind: four weeks under blinding. Any question about whether the benefit holds at three or six months remains unanswered by controlled data as of now.
Frequently asked questions
Can esketamine be prescribed alone, without an oral antidepressant?
Current European marketing authorization requires administration together with an oral antidepressant, but this trial was conducted entirely in the United States and does not itself alter any regulatory framework. Rules differ by country and change over time. Whether monotherapy is authorized, whether the product is actually marketed, and whether it is reimbursed or covered by insurance are three separate questions, and each should be checked separately, in the clinician’s own jurisdiction, at the time of prescribing. Using esketamine outside an approved indication is a form of off-label prescribing, and off-label prescribing is itself governed by conditions that vary from one country to the next.
Does this trial mean the oral antidepressant serves no purpose?
No. The trial compares esketamine alone against placebo, never esketamine alone against esketamine combined with an oral antidepressant. The contribution of the oral antidepressant is not evaluated, and an absence of measurement should not be read as an absence of effect.
Is a 6.8-point difference on the MADRS clinically meaningful?
It corresponds to a moderate effect size, a Cohen’s d of 0.63, and it exceeds the two-point threshold the authors treat as clinically significant, drawing on the regulatory literature. On a scale running from 0 to 60, a difference of this size is noticeable for a severely depressed patient, less so for a patient whose score is already low. Mean baseline MADRS score in this trial was 37.3, ranging from 28 to 50: the population was severe. Clinical relevance is judged patient by patient, not on an average.
Should 84 mg be preferred over 56 mg?
The trial was not built to compare the two doses against each other, and no formal test opposes them. The 84 mg dose reaches a larger effect size at day 28, but the 56 mg dose numerically outperforms it at day 2, the confidence intervals overlap substantially, and discontinuations for adverse events are more frequent at 84 mg, 4.1% against 1.0%. The authors suggest in their conclusion starting at 84 mg. That recommendation goes beyond what their design can establish.
Did the blinding hold?
Imperfectly, and the trial had the honesty to measure it. Asked on day 28, 71.1% of patients on 56 mg and 78.3% of patients on 84 mg were strongly convinced they had received esketamine, against 46.9% of placebo patients convinced they had received placebo. The molecule’s dissociative effects make this situation hard to avoid. The authors list it among the trial’s limitations.
What should be expected before this changes practice?
Replication by a team with no financial ties to the manufacturer, and a controlled follow-up period well beyond four weeks. The two conditions are independent, and both are necessary.
Annotated bibliography
Source study. Janik A, Qiu X, Lane R, et al. Esketamine Monotherapy in Adults With Treatment-Resistant Depression: A Randomized Clinical Trial. JAMA Psychiatry. 2025;82(9):877-887. doi:10.1001/jamapsychiatry.2025.1317. Published online July 2, 2025, accepted March 22, 2025. ClinicalTrials.gov registration NCT04599855. Open access under a CC-BY-NC-ND license. The supplementary material includes clinical protocol 54135419TRD4005 amendment 2 dated July 14, 2020 and the statistical analysis plan dated February 16, 2024, along with three methods appendices, one results appendix, five supplementary tables, and nine supplementary figures. Funding: Janssen Research & Development.
Uncontrolled precedent cited by the authors. Fu DJ, Lopena O, Quach P, Zaki N, Sun L. A post hoc analysis of patients who received esketamine nasal spray monotherapy in an open-label safety extension study (SUSTAIN-3), abstract TH29 presented at the annual meeting of the American Society of Clinical Psychopharmacology, May 31 to June 3, 2022, Scottsdale, Arizona. This is the reference that bounds the priority claim: it covers 50 patients, open label, with no comparator.
Clinical significance threshold used by the authors. Melander H, Salmonson T, Abadie E, van Zwieten-Boot B. A regulatory apologia, a review of placebo-controlled studies in regulatory submissions of new-generation antidepressants. Eur Neuropsychopharmacol. 2008;18(9):623-627. doi:10.1016/j.euroneuro.2008.06.003. And Montgomery SA, Möller HJ. Is the significant superiority of escitalopram compared with other antidepressants clinically relevant? Int Clin Psychopharmacol. 2009;24(3):111-118. doi:10.1097/YIC.0b013e32832a8eb2. These are the two works the authors invoke to set the clinically significant difference threshold at two MADRS points.
On blinding in antidepressant trials. Lin YH, Sahker E, Shinohara K, et al. Assessment of blinding in randomized controlled trials of antidepressants for depressive disorders 2000-2020: a systematic review and meta-analysis. EClinicalMedicine. 2022;50:101505. doi:10.1016/j.eclinm.2022.101505. The authors refer to this work to situate their own guessing figures, noting that the proportion of patients correctly guessing their treatment ranges from 45% to 71% in trials of reuptake inhibitors.
Regulatory context, to be established locally. The product label, the European Medicines Agency’s public assessment report, and any national health authority opinion were not part of the sources reviewed for this analysis and were not verified here. Three questions should be checked separately, in the clinician’s own jurisdiction, before any of this is applied to practice: the exact wording of the authorized indication, whether the product is actually marketed there, and the conditions of reimbursement or insurance coverage, together with the conditions required locally for administering and monitoring the drug.
Editorial collections
Tags
Verified on August 29, 2026 against the full text of the publication and its supplementary material where available. This analysis underwent an independent double reading. The English version was checked for conformity on September 23, 2026, against the figures of the French version and against the source. How we verify what we publish
This analysis is intended for healthcare professionals. It does not constitute a prescribing recommendation and does not replace individual clinical judgment.
Analysis from Psychiatry Evidence Base, evidence-based psychiatry, explained with rigor.