Real-world clinical practice, effectiveness and safety of esketamine nasal spray in major depressive disorder patients: a French prospective cohort (ELLIPSE study)

Abstract

Background:

Major depressive disorder (MDD) is highly prevalent and associated with significant disability and mortality. Treatment-resistant depression (TRD), defined as inadequate response to ⩾2 antidepressant lines, remains a major therapeutic challenge. Esketamine nasal spray, approved in 2019 for TRD, demonstrated efficacy and safety in randomized trials, but real-world evidence remains limited.

Objectives:

The ELLIPSE study aimed to describe patient characteristics, treatment patterns, and 12-month outcomes of esketamine in routine clinical practice in France.

Design:

ELLIPSE was a prospective, multicenter, observational cohort study including adults with unipolar MDD initiating esketamine between Dec 2021 and Jul 2023, followed for 12 months.

Methods:

Data were collected via standardized clinical assessments. Effectiveness was evaluated using the Montgomery–Åsberg Depression Rating Scale (MADRS). Response was defined as ⩾50% MADRS reduction; remission as MADRS ⩽ 10.

Results:

A total of 200 patients (mean age 46.6 years; 56.5% female) were enrolled; 97.5% met TRD criteria. Median esketamine exposure was 4.5 months. Mean MADRS decreased from 31.9 at baseline to 12.4 at month 12 (mean change −17.9; 95% CI: −21.6 to −14.2). Response and remission rates reached a plateau at month 2, where the response rate was 55.2% and the remission rate was 33.6%, maintained up to month 12. Adverse events (AEs) occurred in 65.7% of patients, most commonly dissociative states (28.0%) and transient blood pressure increases (11.6%).

Conclusion:

This real-world study in patients with treatment-resistant depression, including those with comorbidities/prior suicide attempts, showed clinically relevant improvement from month 2, sustained through 12 months. Dissociative symptoms were the most frequent treatment-related AE.

Introduction

Major depressive disorder (MDD) is a mental disorder that is becoming increasingly prevalent in the population. It is associated with increased disability and mortality compared to the general population.^1–3 Second-generation antidepressants (e.g., selective serotonin reuptake inhibitors (SSRIs) or selective serotonin noradrenaline reuptake inhibitors (SNRIs)) are the most commonly used treatments for acute MDD. The prevailing evidence-based guidelines advocate for the utilization of these medications as a primary therapeutic modality.^4–6

In the clinical context, patients diagnosed with MDD who demonstrate an inadequate response to a minimum of two distinct therapeutic interventions are designated as treatment-resistant (TRD).^6,7 The treatment of TRD can be achieved through several therapeutic strategies. One of them, to be carried out systematically, is to increase the dose of the antidepressant (AD) to the maximum tolerated dose specified in the summary of product characteristics. Another strategy consists of switching to another AD.^6,7 An alternative involves augmenting the current AD regimen by introducing a second agent to enhance its efficacy. This augmentation strategy can be either achieved through combination strategies, which involve co-administering two ADs with distinct and complementary pharmacological profiles (e.g., (SSRI)/(SNRI) + Noradrenergic and specific serotonergic antagonist (NaSSA)), or add-on strategies, which consist of supplementing the AD with a non-AD pharmacological agent to potentiate its effect. Examples include lithium, esketamine, antipsychotics, electroconvulsive therapy (ECT), or repetitive transcranial magnetic stimulation (rTMS). It is important to note that rTMS is not yet reimbursed in France, which may limit its accessibility in certain areas.

In 2019, esketamine received approval from the U.S. Food and Drug Administration for the treatment of TRD in adults in conjunction with an oral antidepressant, ⁸ and most recently as a monotherapy for TRD. ⁹ The European Medicines Agency approved esketamine, in combination with an SSRI or SNRI, for adults with treatment-resistant MDD who have not responded to at least two different treatments with antidepressants in the current moderate to severe depressive episode. ¹⁰

Several randomized controlled trials (RCTs) have demonstrated the efficacy and safety of esketamine nasal spray (hereafter referred to as: esketamine) in patients with TRD. Both short-term and long-term (up to 12 months or longer) efficacy and safety was demonstrated against placebo in the TRANSFORM and SUSTAIN trials which reported significant improvements in depressive symptoms (mean decrease of −21.4 (12.3) of the Montgomery–Åsberg Depression Rating Scale (MADRS) from initiation to day 28 ¹¹ ), sustained symptom remission and a low relapse rate in patients who were in remission (26.7% ¹² ). Across both short- and long-term studies, esketamine exhibited a favorable safety profile, with most adverse events (AEs)—including dissociation, vertigo, sedation, and transient increases in blood pressure—being mild to moderate in intensity and resolving spontaneously shortly after administration.^11–16 The higher remission and response rates were also demonstrated against quetiapine in TRD patients ¹⁷ and corroborated by a recent meta-analysis. ¹⁸ Real-world evidence from observational studies conducted in Italy and France provided additional data on the clinical utility of esketamine by reporting consistent effectiveness and a manageable safety profile in routine clinical practice.^19,20 Beyond these clinical findings, recent translational data in humans suggest that esketamine’s rapid antidepressant effects may be driven by glutamatergic modulation and downstream neuroplastic adaptations, ²¹ providing a biological framework that bridges clinical trials, real-world outcomes, and mechanistic understanding.

The aim of the ELLIPSE study was to describe the profile of all patients treated with esketamine in real-world clinical practice, the modalities of the use of esketamine, the patient management at the site of care, and the efficacy and safety outcomes during 12 months after esketamine treatment initiation for unipolar MDD patients.

Methods

Study design and data sources

This was a prospective, multi-center, observational study conducted in France to describe the patients treated with esketamine in routine clinical practice. It was approved by the Sud-Méditerranée I ethics committee on September 27, 2021, and all participants provided verbal informed consent prior to enrollment in the study. The consent process was appropriately recorded and documented in the study records. The data source for this study was standardized physician-completed study forms and validated published clinical scales. These scales have published psychometric support/validation. Appropriate permission or licensing for use of copyrighted/proprietary instruments was obtained where required. The physician-completed study form used for descriptive variables was developed for this observational study and was not intended as a validated psychometric instrument. ³¹ psychiatric hospitals were involved in this study, distributed across the French mainland territory (Supplemental Methods S1).

Study population and study period

The inclusion criteria for patients were as follows: aged 18 and over and clinically diagnosed with unipolar MDD, for whom the treating physician had decided to initiate treatment with esketamine prior to the patient's inclusion in the study, with informed consent to data collection. Patients who had participated in an interventional clinical trial within 30 days prior to study inclusion, or who were participating at the time of inclusion, or who planned to participate in an interventional clinical trial during the study, were deemed ineligible for participation. Patients were included between December 2021, and July 2023 and followed for 12 months (last patient visit on September 2024).

Assessments

In this real-world study, the commercial nasal spray formulation of esketamine (28 mg) was assessed (Supplemental Methods S2). In France, esketamine is administered according to the SmPC, which recommends 56–84 mg twice weekly during the 4-week induction phase for adults aged 18–64 years, followed by weekly dosing during weeks 5–8 and weekly to every-other-week dosing from week 9 onward. For adults aged ⩾65 years, the recommended starting dose is 28 mg.

Data were prospectively collected at study initiation, day 8, month 1 (M1), M2, M3, M6, M9, M12, and treatment discontinuation. Baseline data collected included sociodemographic characteristics, history of depression, and characteristics of the current major depressive episode (MDE), including duration, typology based on clinical feature (suicidal ideation, risk behaviors, deterioration of general physical condition, and functional disability), depression treatment history (including previous neurostimulation started and ended before esketamine initiation and well-conducted lines started more than 28 days before esketamine initiation, defined as at least one of the following: an antidepressant taken as monotherapy during at least 28 days (with or without augmentation), a combination of antidepressants (two or more) taken during at least 28 days (with or without augmentation), at least six consecutive sessions of neurostimulation (ECT, rTMS, and transcranial direct current stimulation (tDCS) ⁷ ). Psychiatric (DSM-5 criteria) and somatic comorbidities were collected.

The dosage and frequency of esketamine administration, in addition to post-administration surveillance duration, the administration setting, and concomitant antidepressant treatments (AD), were collected.

The MADRS was assessed at baseline and throughout the follow-up to assess esketamine clinical effectiveness. ²² Thresholds for the total score were: 0–6 (absence of symptoms), 7–19 (mild depression), 20–34 (moderate depression), 35–60 (severe depression). Treatment response was defined either by a ⩾50% relative reduction in MADRS total score from baseline or by an MADRS total score ⩽10. Remission was defined as a total score ⩽10. Relapse was defined as the occurrence of a MADRS score ⩾22 at any follow‑up assessment in a patient who had previously achieved remission.^11,17

The Clinical Global Impression- Suicidality Severity–Revised (CGI-SS-R, from Normal: 0 to Most extremely suicidal: 6) scale ²³ and items 4 and 5 from the Columbia-suicide severity rating scale ²⁴ (C-SSRS) in patients with CGI-SS-R > 1 were assessed at baseline and throughout the follow-up to assess suicidal ideation evolution. The Quality of Life in Depression Scale (QLDS) was assessed at baseline and throughout follow-up to evaluate depression-related quality of life; this validated patient-reported instrument ranges from 0 to 34, with higher scores indicating poorer quality of life. ²⁵

All treatment-emergent adverse events (TEAE) and special situations were collected (but not their duration) in all patients who received at least one dose of esketamine and were followed in the prospective cohort; this population was slightly broader than the Prospective Analysis Population used for descriptive and effectiveness analyses. TEAEs were considered as related to the treatment in case of a reasonable causal relationship (i.e., evidence to support a causal relationship established by the investigator) between administration of the medicinal product and the AE. Serious adverse events (SAE) criteria are defined in Supplemental Methods S3.

Statistical analyses

All available data were described at each study time point using standard descriptive statistics. Esketamine utilization was described overall and by phase (induction—week 1–4; maintenance 1—week 5–8; maintenance 2—week 9-month 12). Esketamine exposure, and the time to first response/first remission/first relapse were analyzed using the Kaplan-Meier method. Median survival was given with its 95% confidence interval (95% CI). To adjust for potential missing data, a sensitivity analysis using multiple imputation methods was conducted by regression on the MADRS score analysis. The regression models incorporated several cofactors, including the variables “visit,” “score at baseline,” “age,” “sex,” and “number of well-conducted AD treatment lines in current episode prior to esketamine initiation (<2 lines; ⩾2 lines). The imputed scores were then utilized in conjunction with the observed scores (i.e., imputed scores were employed in cases where scores were missing, and observed scores were used in cases where scores were available) to derive patient response and remission rates. Another sensitivity analysis was conducted by defining remission with MADRS ⩽ 12 instead of ⩽10. ¹¹

A descriptive sample size of 200 patients was selected to ensure acceptable precision for estimating key proportions, with a worst-case 95% CI half-width of <7% and maintained precision (<10%) for subgroup analyses.

Statistical analyses were conducted using SAS^® software, version 9.4, SAS Institute, NC, Cary, USA. The reporting of this study conforms to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement for cohort studies ²⁶ (Supplemental Materials).

Results

A total of 200 patients were enrolled across 31 psychiatric hospitals (Table S1) and followed for a median duration of 11.8 months (interquartile range (IQR): 6–12.2). Among them, 122 patients (61.0%) completed the full 12-month follow-up period, while 27 patients (13.5%) were still undergoing esketamine treatment at the end of the observation period. Thus, completion of study follow-up and treatment continuation were not equivalent, as a substantial proportion of patients completed the 12-month follow-up despite no longer receiving esketamine at study end. Study discontinuation prior to month 12 occurred in 78 patients (39.0%; Table S2). The reasons for discontinuation included one death (cause of death: drowning), 17 cases of loss to follow-up, 39 patient-initiated study discontinuation, 6 for physician decision, and 15 discontinuations for other reasons.

Patients’ baseline characteristics and depression history

Baseline characteristics of the study population are presented in Table 1. The majority of patients were under 65 years of age (88.5%, mean age: 46.6 years (standard deviation (SD) = 15.5)) and predominantly female (56.5%). The median duration of lifetime depression was 10 years (IQR: 3.2–22.8), with patients experiencing an average of 3.0 (SD = 2.9) previous MDEs. Among patients with at least one previous MDE (N = 160, 80.0%), 39.0% reported a lifetime history of suicide attempt prior to the current episode.

Table 1.

Patients’ baseline characteristics and depression history.

Characteristics, n (%) otherwise specified	N = 200
Demographics and physical examination
Age, years
Mean (SD)	46.6 (15.5)
Median (IQR)	48 (35–58)
<65 years	177 (88.5)
Female	113 (56.5)
Normal blood pressure at baseline	178/195 (91.3)
Lifetime history of depression
Time between first diagnosis of lifetime depression and inclusion (years)	N = 142
Median (IQR)	10 (3.2–22.8)
Lifetime number of previous MDE (excluding the current episode)	N = 198
Mean (SD)	3.0 (2.9)
Lifetime number of treatment-resistant depressive episodes (failure of at least 2 lines of treatment, excluding current depressive episode)	N = 158
Mean (SD)	1.8 (2.1)
Patients with at least one previous MDE	N = 160
Lifetime history of suicide attempt	78 (39.0)
Current MDE	200 (100.0)
Duration of MDE, years	N = 184
Mean (SD)	2.9 (3.7)
Median (IQR)	1.8 (0.9–3.2)
Typology	N = 200
Suicidal ideation	102 (51.0)
Risk behaviors	21 (10.5)
Deterioration of general physical condition	36 (18.0)
Functional disability	84 (42.0)
Unknown	27 (13.5)
Clinical subtype
Anxiety feature	116 (58.0)
Melancholic features	42 (21.0)
Psychotic features	4 (2.0)
Atypical features	12 (6.0)
Catatonic features	4 (2.0)
Seasonal features	5 (2.5)
At least one full-time hospitalization since the onset of the current MDE	115 (57.5)
Treatment-resistant (clinician’s judgment)	195 (97.5)
MADRS at baseline	N = 198
Total mean (SD)	31.9 (7.0)
Mild depression (7–19 points)	8 (4.0)
Moderate depression (20–34 points)	118 (59.6)
Severe depression (>34 points)	72 (36.4)
CGI-SS-R score at baseline	N = 197
Mean (SD)	1.9 (1.4)
0 Normal, no at all suicidal	46 (23.4)
1 Questionably suicidal	32 (16.2)
2 Mildly suicidal	43 (21.8)
3 Moderately suicidal	45 (22.8)
4 Markedly suicidal	28 (14.2)
5 Severely suicidal	2 (1.0)
6 Among the most extremely suicidal participants	1 (0.5)
C-SSRS at baseline in patients with CGI-SS-R > 1 at baseline	N = 146
Question 4: Active suicidal ideation with some intent to act without specific plan	79 (54.1)
Question 5: Active suicidal ideation with specific plan and intent	33 (22.6)
QLDS Score at baseline	N = 190
Mean (SD)	20.8 (3.8)
Severity of the current MDE (clinician’s judgment)
Mild	1 (0.5)
Moderate	63 (31.5)
Severe	136 (68.0)
Comorbidities at baseline
Other psychiatric comorbidity	144 (72.0)
Anxiety disorders	79/144 (54.9)
Posttraumatic stress disorder	48/144 (33.3)
Current substance use-related disorders and addictive disorders	40/144 (27.8)
Obsessive-compulsive and related disorders	11/144 (7.6)
Conduct and Impulse Control Disorders	7/144 (4.9)
Neurodevelopmental disorders	5/144 (3.5)
Endocrine disorder	19/199 (9.5)
Cardiac disease	31/199 (15.6)
Vascular disease	8/199 (4.0)
Respiratory, thoracic, and mediastinal disorders	14/199 (7.0)
Nervous system disorders	26/199 (13.1)
Hepatobiliary disorders	9/199 (4.5)
Previous well-conducted treatment lines of AD	N = 182
Median number (IQR)	3 (2–5)
⩾2	145/182 (79.7)
⩾5	52/182 (28.6)
Details of previous well-conducted lines of AD	182 (91.0)
A single AD without augmentation	129 (64.5)
A single AD with augmentation	61 (30.5)
Combination of AD without augmentation	59 (29.5)
Combination of AD with augmentation	74 (37.0)
A single augmentation therapy drug	63 (31.5)
Combination of augmentation therapy drugs	18 (9.0)
Neurostimulation (ECT, rTMS, tDCS)	45 (22.5)
No treatment	6 (3.0)

AD, antidepressant; CGI-SS-R, Clinical Global Impression- Suicidality Severity–Revised; ECT, electroconvulsive therapy; IQR, interquartile range; MADRS, Montgomery–Åsberg Depression Rating Scale; MDE, major depressive episode; QLDS, Quality of Life in Depression Scale; rTMS, repetitive transcranial magnetic stimulation; tDCS, transcranial direct current stimulation.

The current MDE had a median duration of 1.8 years (IQR: 0.9–3.2). More than half of the patients (57.5%) had experienced at least one psychiatric hospitalization during the current episode, and the current episode was considered to be treatment-resistant in nearly all (97.5%) patients. The most frequently reported typology based on clinical features of the current MDE included suicidal ideation (51.0%), functional disability (42.0%), deterioration of general physical condition (18.0%), and risk behaviors (10.5%). The mean baseline MADRS score was 31.9 (SD = 7.0). Based on clinician judgment, 68.0% of episodes were classified as severe, and 31.5% as moderate, while 36.4% met the MADRS threshold for severe depression and 59.6% met the MADRS threshold for moderate depression. Baseline suicidal ideation assessments reported a mean CGI-SS-R score of 1.9 (SD = 1.4), with nearly half of patients rated as at least mildly to moderately suicidal (mildly: 43 (21.8%), moderately: 45 (22.8%), markedly: 28 (14.2%), severe: 2 (1.0%), most extremely: 1 (0.5%)). Among those with CGI-SS-R > 1, C-SSRS findings indicated suicidal intent (with or without a specific plan) was frequent, affecting 54.1% (N = 79) of these patients. A mean baseline QLDS score of 20 (3.8) was reported, on a 0–34 scale, where higher scores indicate poorer depression-related quality of life.

Psychiatric comorbidities were present in 144 patients (72.0%) at baseline, with 54.9% reporting anxiety disorders, followed by trauma- and stress-related disorders (33.3%), and substance use or addictive disorders (27.8%).

Prior to Esketamine initiation, a substantial proportion of patients (79.7%) had undergone two or more well-conducted treatment lines, and 28.6% had received five or more well-conducted treatments. The median number of prior well-conducted lines was 3 (IQR: 2–5).

As shown in Table 1, previous well-conducted lines of AD consisted of a single AD without augmentation (64.5%), a combination of AD with augmentation (37.0%), or a single augmentation therapy drug (31.5%). Details per line are provided in Figure S1.

In the current MDE, 63 noninvasive brain neurostimulation treatments were started and ended before esketamine initiation in 48 (24.0%) patients. rTMS was the most common modality (55.6%), followed by ECT (36.5%) and tDCS (7.9%). Neurostimulation treatment outcomes varied: only one patient achieved a complete response, while 46.0% showed no response or resistance, 42.9% had a partial response, and 9.5% were not evaluable (interruption, other).

Esketamine utilization

Patients were exposed to esketamine for a median duration of 4.5 months (IQR: 1.4–8.3; Figure 1 and Figure S2). Among patients aged 65 years and older, the most commonly used starting dose was 28 mg, administered in 15 out of 23 individuals (65.2%). A total of 148 patients (74%) had 84 mg as the most frequently used dose during the induction phase (weeks 1–4). The dose increased from the beginning (171 (85.5%) patients with 56 mg) to the end of the induction (153 (76.5%) patients with 84 mg). Nearly all patients (n = 196, 98.0%) followed a twice-weekly administration schedule as the most used frequency in the induction phase. Among the remaining 157 patients who had entered the first maintenance phase (weeks 5–8), 84 mg was still the most observed dose for 132 (84.1%) patients. Regarding dosing frequency, the majority (n = 122, 77.7%) transitioned to a once-weekly schedule as the most used frequency in this maintenance phase. For the 137 patients who remained in maintenance 2 (week 9–month 12), 84 mg was still the most observed dose for 121 (88.3%) patients. In terms of the most used administration frequency, 43 (31.4%) presented a once-weekly schedule. A further shift toward less frequent administration was observed with 84/137 (61.3%) patients presenting with a longer administration than once weekly. Esketamine was permanently discontinued in 173 (86.5%) patients over the 12 months follow-up: for clinical improvement considered as satisfactory (37.0%), an inadequate response or lack of response either during initiation or during the maintenance (33.5%), after the patient’s decision (12.7%), or for the occurrence of an AE suspected to have a causal relationship with esketamine (8.7%).

Figure 1.

Esketamine most observed posology and administration schedule.

Esketamine treatment was initiated in a partial-hospitalization and/or outpatient setting for the majority of patients (145/200, 72.5%), while 55 patients (27.5%) began treatment during an inpatient hospitalization. Of those initially hospitalized, 42 patients (76.4%) transitioned to an outpatient setting after a median of 18 days (IQR: 12–41). Although actual monitoring times varied widely (overall range approximately 10–245 min), the median post-administration observation period was 120 min (IQR: 120–120) across all treatment settings, and at least 120 min of post-administration monitoring was documented for 94.6% of administrations overall.

At the time of esketamine initiation, nearly all patients (98.0%) were receiving concurrent pharmacological treatment for depression. Antidepressants were prescribed in 95.5% of cases, with (SSRIs) used in 50.0% (100/200) and (SNRIs) in 47.0% (94/200; Table 2). Notably, 26.0% (52/200) of patients were receiving neither an SSRI nor an SNRI. Augmentation strategies were employed in 55.0% of patients, most commonly involving second-generation antipsychotics (34.5%). Other pharmacologic treatments were used in 6.0% of the population.

Table 2.

Concomitant antidepressant treatments started before or/at esketamine initiation and ended after.

Pharmacologic treatments at esketamine initiation	N = 200
Antidepressants	191 (95.5)
SSRI (Selective Serotonin Reuptake Inhibitor)
Fluoxetine	47 (23.5)
Sertraline	34 (17.0)
Paroxetine	26 (13.0)
Escitalopram	23 (11.5)
Citalopram	2 (1.0)
Fluvoxamine	2 (1.0)
SNRI (Serotonin and Noradrenaline Reuptake Inhibitors)
Venlafaxine	74 (37.0)
Duloxetine	24 (12.0)
Milnacipran	9 (4.5)
NaSSA (noradrenergic and specific serotonergic antagonist)
Mirtazapine	54 (27.0)
Mianserin	29 (14.5)
TCA (tricyclic antidepressant)
Clomipramine	38 (19.0)
Amitriptyline	11 (5.5)
Dosulepine	2 (1.0)
Maprotiline	2 (1.0)
Imipramine hydrochloride	1 (0.5)
IMAO (monoamine oxidase inhibitor)
Selegiline	4 (2.0)
Phenelzine	3 (1.5)
Moclobemide	1 (0.5)
Other
Vortioxetine	48 (24.0)
Agomelatine	7 (3.5)
Tianeptine	3 (1.5)
Bupropion	1 (0.5)
Augmentation strategy	110 (55.0)
Second-generation antipsychotic	69 (34.5)
Quetiapine	36 (18.0)
Aripiprazole	22 (11.0)
Olanzapine	17 (8.5)
Risperidone	10 (5.0)
Clozapine	3 (1.5)
Antiepileptic	46 (23.0)
Lamotrigine	39 (19.5)
Valproic acid derivatives	6 (3.0)
Carbamazepine	2 (1.0)
Oxcarbazepine	1 (0.5)
Pregabaline	1 (0.5)
Lithium	40 (20.0)
Thyroid hormone	3 (1.5)
Other	21 (10.5)
Pramipexole	17 (8.5)
Amisulpride	4 (2.0)
Other pharmacologic treatments	12 (6.0)

Percentages for antidepressant classes and agents are not mutually exclusive, as some patients received more than one antidepressant concomitantly.

Esketamine effectiveness (MADRS)

Among patients who remained under treatment, depressive symptoms improved markedly over the course of follow-up. The mean MADRS score decreased from 31.9 (SD = 7.0) at baseline to 18.9 (SD = 9.1 at the M1 visit (end of the initiation phase), corresponding to a mean absolute change of −13.0 points (95% CI: −14.6 to −11.4; Figure 2). Continued improvement was observed at M2 (end of the first maintenance phase), with a mean score of 15.6 (SD = 9.1) and a mean absolute change of −15.6 (95% CI: −17.4 to −13.8). Scores remained stable at M3 (16.0, SD = 8.9; mean change: −15.2 (95% CI: −17.2 to −13.3)) and M6 (14.9, SD = 8.6; mean change: −15.6 (95% CI: −17.8 to −13.3)). The lowest mean MADRS score was observed at M12, reaching 12.4 (SD = 9.4), with a mean absolute change of −17.9 (95% CI: −21.6 to −14.2). MADRS items all showed improvement over the 12-month follow-up. At baseline, the most severe MADRS items were Reported Sadness (4.2, SD = 1.1) and Apparent Sadness (4.0, SD = 1.2), followed by lassitude (3.7, SD = 1.1), inability to feel (3.7, SD = 1.1), and inner tension (3.3, SD = 1.3), while symptoms such as suicidal thoughts (2.5, SD = 1.5), reduced sleep (2.4, SD = 1.8), and reduced appetite (1.5, SD = 1.6) were least pronounced. Over the 12-month follow-up, the most important absolute improvements were observed for Apparent Sadness and Reported Sadness, each decreasing by more than 2 points, with Inner Tension and Lassitude also showing substantial reductions (between 1 and 2 points). In terms of relative change, all symptom improvements ranged between 20% and 60% at month 1 and were generally maintained throughout the treatment period up to month 12. Although suicidal thoughts, reduced sleep, and loss of appetite were the least severe items at baseline, they showed substantial relative improvement, with mean scores >1, reflecting near disappearance of these symptoms over 12 months of treatment (Table S3 and Figures S1, S3, and S4).

Figure 2.

Mean absolute changes from baseline in MADRS total score by visit.

MADRS, Montgomery–Åsberg Depression Rating Scale.

Response and remission rates increased progressively throughout the follow-up for the patients who were still under esketamine treatment (Figure 3). At M1, 36.8% of patients met criteria for response, and 21.6% achieved remission. These rates rose to 55.2% and 33.6%, respectively, at M2. By M6, response and remission rates were 57.1% and 32.9%, respectively (including 10/70 patients who had not been responders at M2). At M9, response was observed in 47.4% of patients, with 28.9% in remission. The highest rates were recorded at M12, with 70.4% of patients achieving a response and 51.9% reaching remission. In the full cohort analysis (all patients still in the study regardless of still being under treatment or not), response and remission rates followed the same positive evolution throughout the follow-up with a plateau around M2, although displaying slightly lower rates (Figure S6). Remission sensitivity analyses with a higher MADRS threshold (⩽12 instead of ⩽10) also showed consistent results, with slightly higher rates when the threshold was set to ⩽12 (Figures S7 and S8).

Figure 3.

MADRS response and remission rates in patients still under esketamine with MADRS available up to month 12.

MADRS, Montgomery–Åsberg Depression Rating Scale.

Among the 136 (68.0%) patients who responded at least once during follow-up, the median time to first response was 50 days (IQR: 26–84). For the 91 patients who achieved remission at least once, the median time to first remission was 62 days (IQR: 28–95). Relapse occurred in 19 patients after a median of 190 days (IQR: 87–548). Of these, 10 patients relapsed while still under treatment (median time: 159 days (IQR: 63–236)), and nine relapsed after treatment discontinuation (median time: 143 days (IQR: 104–234)).

Suicidal ideation (CGI-SS-R and C-SSRS)

Suicidal ideation analyses also suggested a progressive reduction in suicidal severity over follow-up. In patients still under esketamine, mean CGI‑SS‑R change from baseline reached −1.34 at month 2, −1.50 at month 6, and −1.72 at month 12; similar trends were observed in the full cohort (−1.21, −1.42, and −1.58, respectively). Among patients with baseline CGI‑SS‑R > 1, C‑SSRS findings also showed an overall reduction in active suicidal ideation with intent over time. In the full cohort of these patients, the proportion with active suicidal ideation with some intent to act without a specific plan decreased from 54.1% at baseline to 32.3% at month 2, 18.8% at month 6, and 14.3% at month 12, while active suicidal ideation with specific plan and intent decreased from 22.6% to 4.6%, 4.2%, and 2.4%, respectively (Figures S9–S11).

Quality of life (QLDS)

Over follow-up, QLDS scores improved progressively, with mean absolute changes from baseline of −2.8 at month 1, −4.2 at month 2, and −6.3 at month 12 in the full cohort; similar trends were observed in patients still under esketamine (Figure S12).

Safety

During the study period, at least one AE was reported in 136 patients (65.7%; Table 3). AEs were assessed as related to esketamine according to the physician’s judgment in 99 patients out of 136 (47.8% of all patients) and mainly included dissociative states (28.0%), increased blood pressure (11.6%), and hypertension (10.1%). A total of 31 AEs led to treatment discontinuation in 23 patients (11.1%), including hypertension in 6 patients (2.9%) and suicidal ideations in 3 patients (1.4%). Thirty-one SAEs were reported in 25 (12.1%) of the patients (including one death by drowning, for which the causality with treatment could not be established according to the physician). SAEs were assessed as related to esketamine according to the physician’s judgment in four patients (1.9% of all patients) and included bipolar disorder (n = 1), panic attack (n = 1), headache (n = 1), serotonin syndrome (n = 1), and unequal pupils (n = 1).

Table 3.

Safety events.

System organ class/preferred term	Total: N = 207
	AEs			Related AEs
	Nb AEs	Nb pat	% pat	Nb AEs	Nb pat	% pat
Any AEs	1 263	136	65.7	1 071	99	47.8
Psychiatric disorders	644	99	47.8	566	66	31.9
Dissociative states	470	61	29.5	465	58	28.0
Suicidal ideation	35	31	15.0	4	3	1.4
Anxiety	36	19	9.2	25	11	5.3
Suicide attempt	8	8	3.9	0	0	0.0
Euphoric mood	7	5	2.4	5	4	1.9
Nervous system disorders	193	42	20.3	169	32	15.5
Dizziness	41	15	7.2	36	11	5.3
Headache	27	14	6.8	24	11	5.3
Sedation	27	12	5.8	22	9	4.3
Somnolence	21	8	3.9	20	7	3.4
Dysarthria	8	5	2.4	8	5	2.4
Gastrointestinal disorders	70	29	14.0	51	19	9.2
Nausea	52	20	9.7	43	15	7.2
Vomiting	10	9	4.3	6	6	2.9
Investigations	116	29	14.0	115	28	13.5
Blood pressure increased	107	25	12.1	106	24	11.6
Vascular disorders	51	26	12.6	43	23	11.1
Hypertension	48	23	11.1	41	21	10.1
General disorders and administration site conditions	63	21	10.1	43	15	7.2
Feeling abnormal	19	7	3.4	10	6	2.9
Fatigue	5	5	2.4	2	2	1.0
Ear and labyrinth disorders	29	13	6.3	28	12	5.8
Vertigo	8	6	2.9	8	6	2.9
Eye disorders	49	9	4.3	38	8	3.9
Vision blurred	19	5	2.4	21	4	1.9
Infections and infestations	10	9	4.3	1	1	0.5
COVID-19	5	5	2.4	0	0	0.0
Respiratory, thoracic, and mediastinal disorders	6	6	2.9	2	2	1.0
Serious AEs	31	25	12.1	5	4	1.9
Psychiatric disorders	20	16	7.7	2	2	1.0
Suicidal ideation	9	9	4.3	0	0	0.0
Suicide attempt	6	6	2.9	0	0	0.0
Anxiety	2	2	1.0	0	0	0.0
Bipolar disorder	1	1	0.5	1	1	0.5
Panic attack	1	1	0.5	1	1	0.5
Somatic symptom disorder	1	1	0.5	0	0	0.0
Nervous system disorders	2	2	1.0	2	2	1.0
Headache	1	1	0.5	1	1	0.5
Serotonin syndrome	1	1	0.5	1	1	0.5
Eye disorders	1	1	0.5	1	1	0.5
Unequal pupils	1	1	0.5	1	1	0.5
Gastrointestinal disorders	1	1	0.5	0	0	0.0
Pancreatitis	1	1	0.5	0	0	0.0
General disorders and administration site conditions	1	1	0.5	0	0	0.0
Drowning	1	1	0.5	0	0	0.0
Hepatobiliary disorders	1	1	0.5	0	0	0.0
Cholecystitis	1	1	0.5	0	0	0.0
Infections and infestations	1	1	0.5	0	0	0.0
COVID-19	1	1	0.5	0	0	0.0
Metabolism and nutrition disorders	1	1	0.5	0	0	0.0
Diabetic metabolic decompensation	1	1	0.5	0	0	0.0
Respiratory, thoracic, and mediastinal disorders	1	1	0.5	0	0	0.0
Chronic obstructive pulmonary disease	1	1	0.5	0	0	0.0
Surgical and medical procedures	1	1	0.5	0	0	0.0
Hip surgery	1	1	0.5	0	0	0.0
Vascular disorders	1	1	0.5	0	0	0.0
Jugular vein thrombosis	1	1	0.5	0	0	0.0

AE, adverse event; Nb, Number.

Dissociative states occurred in 61/207 patients (29.5%), and the majority of them were related to treatment (465/470, 98.9%). None of them were serious (regardless of being related to treatment or not), nor did they lead to death. All cases recovered, and a single event led to treatment discontinuation. In addition, 99.1% (466/470) did not lead to any change, and 0.6% (3/470) resulted in a decrease of the esketamine dose.

Discussion

ELLIPSE is the first prospective cohort of this size describing esketamine real-world utilization in patients with unipolar MDD, whereas previous results have come either from clinical trials or retrospective real-world cohorts. The inclusion of 31 centers across France ensures broad geographical and institutional representation, enhancing the generalizability of the findings and offering a nuanced understanding of esketamine’s role in routine care.

A key strength of this study lies in the characterization of its population, which reflects the clinical complexity and heterogeneity often encountered in psychiatric practice but underrepresented in RCTs primarily because of the highly selective eligibility criteria that tend to exclude such complex cases.^27,28 Patients in the ELLIPSE study had a long-standing history of depression (median duration: 10 years), prolonged current episodes (median: 1.8 years), and a high burden of psychiatric comorbidities (72%), including anxiety (54.9%), trauma- and stress-related disorders (33.3%), and substance use disorders (27.8%). Psychiatric comorbidities, including substance use or addictive disorders, were documented at baseline as part of the clinical profile. These conditions were not treated as primary indications for esketamine initiation and did not influence eligibility, which was strictly limited to patients with unipolar MDD. The main clinical features during the current MDE were suicidal ideations (51.0%) and functional disability (42.0%). Nearly all had undergone multiple prior treatment attempts, with 28.6% having received five or more well-conducted lines, and many had previously received neurostimulation. This profile contrasts with more selective RCT populations, such as those in TRANSFORM-2, where fewer patients had extensive treatment histories or comorbidities. ¹¹ The ELLIPSE cohort is more comparable to previous real-world cohorts such as ESKALE,^19,20,29,30 which similarly reported high levels of chronicity, suicidality, and treatment resistance. A notable difference with real-world cohorts is that patients in the ESKALE cohort appeared to be substantially more treatment-resistant, exhibiting more advanced therapeutic progression, as evidenced by a median number (IQR) of 6.0 (4.0; 9.0) well-conducted lines of treatment prior to esketamine initiation (88.3% of the cohort with at least three lines). ²⁰ The difference between six prior well-conducted lines in ESKALE and three in ELLIPSE suggests that esketamine nasal spray may be considered earlier in the treatment pathway in more recent clinical practice, although this likely also reflects the progressive implementation of esketamine over time, as the studies were conducted several years apart. Notably, 20% of patients in ELLIPSE had received fewer than two prior lines for the current episode, reflecting clinicians’ decisions to initiate esketamine based not only on the duration of the current episode but also on the patient’s lifetime history of depression, severity, and preexisting comorbidities. This finding, combined with the observation that 97.5% of patients were classified as treatment-resistant, suggests a possible confusion in clinical practice between severity and resistance. Importantly, this discrepancy does not necessarily reflect off-label use. The European SmPC does not specify any minimum duration for prior antidepressant trials. In ELLIPSE, the definition of a “well-conducted” treatment line relied on a strict methodological framework aligned with AFPBN recommendations for TRD, ⁷ which define adequate prior antidepressant trials as treatments given at an adequate dose and for at least 4 weeks once the target dose is reached. The ⩾28-day threshold was therefore selected for methodological consistency within the study, rather than because such a duration is required by the SmPC. Patients treated with esketamine in ELLIPSE displayed a clinically complex profile, notably characterized by suicidality, psychiatric comorbidities, functional impairment, and prolonged depressive episodes. These dimensions may have contributed to clinicians’ overall judgment of treatment resistance, even when the formal documentation of prior treatment duration did not fully meet standardized research thresholds. Consequently, the apparent discrepancy between clinician-defined resistance (97.5%) and the number of well-documented prior lines (⩾2 lines in 80%) may reflect the way TRD is operationalized in routine care rather than a limitation of the study. This evolution highlights how clinical practices have shifted over time, with esketamine progressively finding its place in therapeutic strategies through real-world use: patients are now less treatment-resistant, fewer have received ECT beforehand, and a growing proportion are treated in outpatient settings, reflecting improved accessibility and integration into routine care.

The positioning of esketamine in real-world care warrants further reflection. In ELLIPSE, esketamine was frequently initiated in patients with long-standing, clinically complex depression, notably characterized by suicidal ideations, psychiatric comorbidities, functional impairment, and prolonged depressive episodes, suggesting that, in practice, it is still perceived as a treatment of last resort. At the same time, a meaningful proportion of patients initiated esketamine despite having fewer than two well-conducted treatment lines in the current episode, indicating that clinical complexity and chronicity may weigh as heavily as formal definitions of treatment resistance in decisions to escalate care. This raises an important question for future practice and research: should esketamine remain restricted to the most complex or high-risk presentations, or should its earlier use (prior to entrenched chronicity, multiple hospitalizations, or repeated treatment failures) be explored more systematically? Clarifying the optimal timing of esketamine initiation and determining whether earlier intervention may yield different or more favorable outcomes represents a meaningful direction for future real-world studies.

Esketamine utilization in ELLIPSE followed the modalities of use indicated in the summary of product characteristics (SmPC). ¹⁰ Regarding initial dosage, the majority of patients aged 65 and over received 28 mg as recommended, while patients aged 18–64 years primarily initiated treatment with 56 mg. The progression of the initial maintenance period (weeks 5–8) was consistent with the SmPC. The most prevalent doses observed were 56 mg and 84 mg, with a rapid shift from 56 mg at initiation (85.5% of patients) to 84 mg at the end of the induction period (76.5% of patients). A downward shift in the weekly administration schedule was observed. By the time of the second maintenance period (week 9 and beyond), most patients still under treatment were receiving 84 mg, albeit with a longer frequency of less than once per week. The sharp increase from 56 to 84 mg was also observed in ESKALE, where patients under 65 years reached 84 mg after a median time of 8.5 days. ²⁰ In contrast, 66% of the REAL-ESK patients were under 56 mg at 1 month after initiation, and 52.7% at 3 months. ¹⁹ Dosing patterns in INTEGRATE showed pragmatic up-titration from 56 mg to 84 mg twice weekly during induction, with nearly half of patients continuing this frequency beyond week 4. ³⁰ Although clinical practice patterns and reimbursement conditions differ across France (ELLIPSE and ESKALE), Italy (REAL-ESK), and Spain (INTEGRATE), these findings suggest a pragmatic dose-escalation pattern in routine care, with many patients rapidly titrated to and maintained on 84 mg over time. This may reflect growing clinician confidence in optimizing esketamine dosing according to response in real-world practice.

Esketamine was predominantly administered in partial-hospitalization or outpatient settings (72.5% at initiation), and 76.4% of those who began treatment as inpatients transitioned to outpatient care. This higher use of outpatient or partial-hospitalization initiation suggests that esketamine is becoming more embedded in routine ambulatory care, rather than being reserved primarily for inpatient or highly refractory hospital-managed cases. In INTEGRATE, Esketamine was administered primarily in outpatient or day-hospital settings, aligning with our observations of increased accessibility. ³⁰ Similarly, the more flexible setting reported in ELLIPSE may reflect improved accessibility to treatment while maintaining compliance with the post-administration monitoring protocol by a healthcare professional until the patient is considered clinically stable and ready to leave the healthcare setting. ¹⁰

Nearly all patients were receiving a concomitant antidepressant treatment at esketamine initiation, with SSRIs and SNRIs being the most frequently prescribed antidepressants, and a slight majority of patients also receiving augmentation strategies at initiation.

The effectiveness of esketamine in this real-world cohort was consistent with, and in some respects extended, findings from clinical trials, even if the characteristics of the different studies populations vary from one to another. In ELLIPSE, the MADRS decreased consistently in patients still under treatment during the induction phase and stabilized by M2 (mean MADRS 15.6 (9.0)), with sustained improvements through M12. This overall decrease was also consistent among the different sub-items of the score and overall with the imputation sensitivity analysis. Response and remission rates also increased progressively to 55.2% and 33.6% at M2, reaching 70.4% and 51.9% at M12. The results reported from ESKALE were comparable, with 56.2% of the patients achieving a clinical response by M3, and 43.8% remitting from MDE. Rates at M6 were also in the same range for both response (58% vs 57.1% in ELLIPSE) and remission (36% vs 32.9% in ELLIPSE). The REAL-ESK cohort also reported a significant reduction in the MADRS score in patients still under esketamine was observed at month 1 (mean, 22.27 ± 9.81) and month 3 (n = 91; mean, 14.69 ± 9.88) compared to baseline ((T0); mean, 35 ± 8.53). ¹⁹ Compared to T0, there was an increase in clinical response (month 1, 28.4%; month 2, 64.2%) and remission (month 1, 11.2%; month 2, 40.6%) with significant improvement in both at month 3 versus month 1. In the ESCAPE-TRD trial, the absolute rate of remission (score of 10 or less on the MADRS) was 27.1% at month 2. ¹⁷ Although response and remission rates stabilized around month 2, this should not be interpreted as a signal of treatment futility. Rather, month 2 represents a meaningful clinical checkpoint at which the overall management plan can be reassessed and optimized. In ELLIPSE, later responders were observed beyond month 2: among the 68 patients still under treatment at month 6, 40 had achieved response and/or remission, including 10 who had not been responders at month 2. This supports continued treatment and individualized reassessment beyond the initial maintenance phase. Evidence from other real-world cohorts also confirms the presence of slow responders; for example, in the REAL-ESK cohort, ¹⁹ more than one-third of patients who achieved remission at month 3 had not responded at month 1. Importantly, study completers and treatment retainers represented different subgroups in ELLIPSE: while 122 patients completed the 12-month follow-up, only 27 were still receiving esketamine at study end, indicating that follow-up completion should not be conflated with treatment continuation when interpreting long-term outcomes. Of note, treatment discontinuation should not be interpreted per se as a marker of lack of effectiveness, as discontinuation may also occur after satisfactory clinical improvement or for other reasons. Accordingly, the month-12 response and remission rates observed among patients still under treatment are best interpreted as outcomes in a selected subgroup of treatment retainers and likely overestimate long-term effectiveness at the overall cohort level. In contrast, the full cohort analysis, including patients who had discontinued treatment, provides a more conservative perspective on long-term outcomes, although it does not fully account for patients who exited the study prematurely. This interpretation is further supported by the treatment-retention pattern which illustrates the progressive decline in esketamine exposure over time.

Suicidal ideation analyses suggest a clinically meaningful reduction in suicidal severity over follow-up. In our cohort, CGI-SS-R scores decreased progressively from baseline to month 12 in both the full cohort and patients still under treatment, indicating an overall improvement in suicidal ideations across follow-up. In addition, among patients already identified as having suicidal ideation at baseline (CGI‑SS‑R > 1), C-SSRS findings also showed an overall decline in active suicidal ideation with intent over 12 months. These descriptive trends are consistent with recent real-world studies in which intranasal esketamine was associated with reductions in suicidal ideation, hopelessness, and self-harming behaviors. ³¹ Recent retrospective data also suggest that suicidal trajectories may differ across subgroups, including sex-related patterns. ³² Overall, these findings support the clinical relevance of suicidal ideation as an outcome domain in TRD cohorts, although whether any anti-suicidal effect is independent from the antidepressant effect remains to be further clarified. This question is also in line with recent work suggesting that the anti-suicidal signal observed with esketamine may, at least partly, be mediated by its antidepressant effect. ³³

In addition to clinician-rated symptom improvement, patient-reported quality of life also improved over follow-up, as reflected by the progressive decrease in QLDS scores. This suggests that the benefits observed in ELLIPSE were not limited to depressive symptom severity alone.

The safety profile of esketamine was consistent with previously published results. The TRANSFORM 2 trial reported dissociation, nausea, vertigo, dysgeusia, and dizziness as being the most common AEs. ¹¹ Those events were the ones also reported in SUSTAIN 1 and SUSTAIN 2.^12,15 A total of 7% of the TRANSFORM 2 trial patients discontinued treatment because of the AE encountered, while 8.7% did so in our study. However, as reported in ESCAPE-TRD, TEAEs were typically transient and mild, which might have explained why treatment discontinuation was less likely versus quetiapine in this trial. ³⁴ The discontinuation rate reported in SUSTAIN 2 (3.2%) was the lowest reported. Also, while we reported serious events related to esketamine in 1.9%, 6.9% of the SUSTAIN 2 patients experienced such events. Patients from the ESKALE cohort reported ⩾1 AE for 66.2% of them, including dissociation (34.4%), somnolence (15.9%), vertigo (15.9%), sedation (14.6%), blood pressure increase (14.0%), and anxiety (14.0%). ²⁰ In the same study, AEs leading to permanent discontinuation of esketamine were seen in 13.4% of patients (n = 21) patients, including increased blood pressure (n = 6), somnolence (n = 5), dissociation (n = 4), vertigo (n = 4), suicidal ideation (n = 2), and suicide attempt (n = 1). In the REAL-ESK cohort, dissociative symptoms (39.7%), sedation (28.4%), and transitory hypertension (10.3%) were the most common side effects reported. ¹⁹ Three patients (2.58%) discontinued because of severe side effects at T1. Safety outcomes reported in INTEGRATE were consistent with our cohort: 97.5% of AEs resolved the same day, and only 1.8% required intervention. Dissociation (50.3%), somnolence (41.8%), and dizziness (40.2%) were most common. ³⁰

Regarding the major safety events that were of concern in the literature prior to this study (namely, dissociation and treatment dependence) our findings were consistent with what was already reported. No dissociative state was serious or fatal, and all resolved spontaneously. Only one dissociative event led to treatment discontinuation. A single AE of dependence was reported.

Limitations

Key limitations inherent to observational studies were carefully addressed. Because this study was descriptive in nature, no formal power analysis was performed. The sample size was instead based on the expected precision of estimates for proportions in a real‑world MDD population. In addition to validated published scales used for key clinical outcomes, some descriptive variables were collected using a study-specific physician-completed form developed for this observational study. This form was not designed as a validated psychometric instrument and was not formally pilot-tested, which may have introduced some measurement variability for non-scale descriptive variables. Multi-level selection bias was mitigated by assessing representativeness at the site, patient, and follow-up levels. While patient inclusion relied on clinician participation, geographical and institutional diversity were preserved. Although attrition (attrition rates at M6: 27.5% overall, and 65% in patients still under treatment, at M12: 40% overall, and 86.5% in patients still under treatment) could lead to bias, sensitivity analyses using multiple imputation techniques showed consistent results. Relapse was defined on the basis of a MADRS score ⩾22 observed at any follow-up assessment after remission, without confirmation across consecutive visits. Although this approach may have captured some transient fluctuations in depressive symptoms, it was considered appropriate in the observational context of the study, where assessments were performed according to routine clinical follow-up rather than a protocol-driven schedule. Information bias was minimized through standardized data collection protocols and harmonized clinical assessment tools. Because this was an observational study without a control group or randomization, causal inference cannot be established, and improvements may partly reflect regression to the mean. Confounding by indication is particularly relevant, as patients with more severe or treatment-resistant depression may be more likely to receive esketamine. In addition, the use of concomitant pharmacological therapies (e.g., antidepressants and antipsychotics) and non-pharmacological interventions (e.g., psychotherapy and neurostimulation) may have influenced treatment response. Nevertheless, the efficacy of esketamine nasal spray was already demonstrated in randomized studies such as TRANSFORM, SUSTAIN, and ESCAPE-TRD trials.^11–14,17

Conclusion

ELLIPSE is the first prospective cohort of this size describing esketamine real-world utilization in patients with MDD in France. Participants, predominantly affected with a long-lasting moderate to severe treatment-resistant depression, and were treated in accordance with the summary of the products characteristics, were recruited from almost every region of the country, ensuring national representativeness. The results reported in terms of characteristics, treatment utilization, effectiveness, and safety were consistent with previously available literature from RCT and observational studies. It thereby adds further real-world evidence to the clinical benefit of esketamine.