Treating Bipolar Depression with Esketamine: Safety and Effectiveness data from a naturalistic multicentric study on Esketamine in Bipolar versus Unipolar Treatment-Resistant Depression

Bipolar depression accounts for the majority of symptomatic time in both type I and type II bipolar disorder and is associated with high morbidity, suicidality, and limited effective treatment options. Conventional antidepressants have delayed onset, frequent poor tolerability, and risk provoking affective switches or rapid cycling in bipolar disorder. Brain stimulation techniques (ECT, rTMS, tDCS) offer alternatives for some patients, but a substantial proportion of people with bipolar depression remain partially responsive or treatment-resistant (B-TRD). Interest has therefore grown in glutamatergic agents such as ketamine and its S-enantiomer, esketamine, which have rapid antidepressant effects in unipolar TRD and a more favourable outpatient safety profile than intravenous ketamine; however, randomised data on esketamine in bipolar depression are lacking and regulatory guidance is cautious because of concerns about affective switching. This study aimed to provide preliminary real-world evidence on the antidepressant effectiveness, safety, and tolerability of esketamine nasal spray (ESK-NS) in patients with treatment-resistant bipolar depression, comparing outcomes with an age-matched sample of unipolar TRD. Secondary objectives included assessment of anxiolytic effects and the average risk of treatment-emergent affective switch in bipolar patients treated with ESK-NS. The investigation used retrospective multicentre data collected as part of an early access programme in Italy to address the current evidence gap regarding ESK-NS use in B-TRD.

The analysis used data from the REAL-ESK study, an observational, retrospective, multicentre dataset of patients with treatment-resistant depression treated with esketamine nasal spray as part of an early access programme. Thirty-five subjects with treatment-resistant bipolar depression (B-TRD) were identified and matched one-to-one by age with 35 subjects with unipolar TRD drawn from the same dataset, producing a total analysed sample of 70 patients. All B-TRD participants had a documented history of at least one hypomanic or manic episode according to DSM-5 criteria. Patients with comorbid organic conditions that the EMA lists as absolute contraindications to ESK-NS (for example, untreated hypertension or prior cerebrovascular disorders) were excluded. Retrospective anamnestic data and clinical ratings were extracted from medical records at baseline (T0) and psychometric assessments were recorded at baseline, one month (T1) and three months (T2). Depressive symptoms were measured with the Montgomery–Åsberg Depression Rating Scale (MADRS) and the 21-item Hamilton Depression Rating Scale (HAM-D-21); anxiety was assessed with the 21-item Hamilton Anxiety Rating Scale (HAM-A). Response was defined as a ≥50% reduction in MADRS or HAM-D-21 from baseline; remission was defined as MADRS <10 or HAM-D-21 <7. Adverse events and clinically relevant outcomes (hospitalisation, relapse, premature withdrawal) were documented by treating psychiatrists. Statistical analyses used SPSS and JASP. Normality was tested with Kolmogorov–Smirnov and, since distributions were normal, parametric tests were applied. Group comparisons used Student's t-tests for continuous variables and Pearson χ2 for categorical variables. Repeated-measures ANOVA (rm-ANOVA) assessed the interaction between group (B-TRD vs TRD) and time (T0, T1, T2) for MADRS and HAM-A scores, with Mauchly's test of sphericity and Greenhouse–Geisser correction as needed. Scheffé post-hoc tests controlled for multiple comparisons. Effect sizes reported included partial eta2 for rm-ANOVA and Cohen's d or Hedges' g for pairwise comparisons. Attrition reduced the analysed samples to 33 B-TRD and 32 TRD at T1, and 31 B-TRD and 28 TRD at T2. The extracted text notes that all B-TRD subjects were treated with mood stabilisers at ESK-NS initiation, while a smaller proportion of TRD subjects were on mood stabilisers.

Baseline characteristics: Seventy patients were included (35 B-TRD, mean age 52.57 ± 12.77; 35 TRD, same mean age). The bipolar sample comprised 17 with BD type I and 18 with BD type II. Groups did not differ significantly in age, gender, employment, years of education, or overall illness duration. Lifetime suicide attempt rates and baseline suicidality (MADRS item 10: 2.75 ± 1.54 vs 1.73 ± 1.48) were higher in the B-TRD group (p = 0.007). Pharmacotherapy differed between groups: current SNRI use was higher in TRD (42.8% vs 20%, p = 0.032), while mood stabiliser treatment was universal in B-TRD versus 57.14% in TRD (p < 0.0001). Baseline total psychometric scores did not differ significantly between groups. Antidepressant effectiveness: Repeated-measures analyses found no significant interaction between group (TRD vs B-TRD) and time on MADRS scores (multivariate Wilks' Lambda = 0.938, F2,52 = 1.728, p = 0.188; univariate rm-ANOVA F1.723,91.334 = 2.445, ε = 0.862, p = 0.100). Mean MADRS reductions from T0 to T1 were reported as −13.03 in B-TRD and −12.21 in TRD, consistent with a rapid antidepressant effect. Response and remission rates increased from T1 to T2 in both groups: in B-TRD, responders were 9/35 (25.7%) at T1 and 24/35 (68.6%) at T2; remitters were 6/35 (17.1%) at T1 and 17/35 (48.6%) at T2. In TRD, responders were 9/35 (25.7%) at T1 and 20/35 (57.1%) at T2; remitters were 3/35 (8.6%) at T1 and 10/35 (28.6%) at T2. The authors note a marked increase in responders between one and three months, suggesting later responders after the induction phase. Anxiolytic effects: The rm-ANOVA interaction for HAM-A did not reach significance (Wilks' Lambda = 0.953, F2,42 = 1.027, p = 0.367). Nonetheless, the extract reports a larger within-group anxiolytic effect in B-TRD between T1 and T2 (p = 0.003, Cohen's d = 0.768) than in TRD (p = 0.330, Cohen's d = 0.489). HAM-A remission at T2 was higher in B-TRD (35%) compared with TRD (3%). Safety and tolerability: Reported side-effect rates were 57.14% in B-TRD versus 77.15% in TRD; types and frequencies are presented in the original tables (not fully reproduced in the extract). One affective switch was recorded in the B-TRD group; no statistically significant difference between groups was found for affective switch risk or development of manic symptoms. The authors state there was no overall increase in suicidality, need for acute hospitalisation, emergent hypomania, psychosis, or dissociation in B-TRD compared with TRD. The extract highlights that all B-TRD patients were on mood stabilisers when ESK-NS was started, which may have influenced the observed low rate of switches.

Martinotti and colleagues present this work as the first multicentre, real-world evaluation of esketamine nasal spray in a treatment-resistant bipolar depression sample, comparing outcomes with age-matched unipolar TRD patients. The investigators interpret the results as supporting good antidepressant effectiveness and tolerability of ESK-NS in both groups, with a rapid reduction in MADRS scores within the first month and further increases in response and remission rates by three months. Although the group-by-time interaction on MADRS was not statistically significant, the bipolar sample showed numerically higher remission rates at three months (48.6% vs 28.6%) and a larger within-group anxiolytic effect, leading the authors to suggest comparable or potentially greater benefit in some domains for B-TRD. The authors situate their findings within the broader glutamatergic hypothesis of mood disorders, noting neuroimaging evidence implicating glutamate dysfunction in regions such as the DLPFC and ACC and prior ketamine data in bipolar depression. They propose that ESK-NS and other glutamatergic interventions could be promising treatment options for bipolar depression, including mixed-features presentations, and that the observed anxiolytic signal may reflect efficacy on agitation/mixed symptoms as well as on comorbid anxiety disorders. Safety-wise, the low observed rate of affective switch and absence of increased severe adverse events are emphasised, but the authors acknowledge an important caveat: all B-TRD patients were concomitantly treated with mood stabilisers, which is appropriate in real-world practice but may reduce observable switch risk and thus bias safety conclusions. Key limitations acknowledged include the small sample size, retrospective observational design, and the absence of structured mania-rating scales (for example, YMRS) to systematically detect hypomanic/manic symptoms. The extracted text indicates the discussion of a further limitation was cut off; consequently, the full list of limitations and the authors' final recommendations are not completely reported in the provided extract. The authors call for randomised controlled trials with larger samples and placebo control to confirm these preliminary observations and better characterise both efficacy and switch risk of ESK-NS in bipolar depression.