Nitrous oxide as an adjunctive therapy in major depressive disorder: a randomized controlled double-blind pilot trial

Guimara and colleagues frame their study within limitations of monoaminergic antidepressants and growing evidence implicating glutamatergic dysfunction in major depressive disorder (MDD). Previous preclinical and clinical work has shown that antagonism of the N-methyl-D-aspartate receptor (NMDAR), notably by ketamine, can produce rapid antidepressant effects. Nitrous oxide (N2O) is another NMDAR antagonist with additional actions on opioid receptors, GABAergic sites, and serotonergic systems; a 2015 proof-of-concept crossover trial reported short-term antidepressant benefit after a single 50% N2O inhalation in treatment-resistant depression. Safety concerns with N2O include effects on vitamin B12 and homocysteine metabolism, but at therapeutic concentrations it is commonly considered safe for outpatient use. This pilot trial was designed to test whether repeated adjunctive N2O inhalation would reduce depressive symptoms in patients with MDD who remained symptomatic despite at least 4 weeks of antidepressant treatment. The investigators hypothesised that 50% N2O administered twice weekly for 4 weeks, added to ongoing antidepressant therapy, would produce greater reductions in clinician-rated depressive symptoms than placebo (100% oxygen). The study therefore examined both acute (within-session) and cumulative antidepressant effects over a 1-month course of semiweekly N2O sessions.

This was a double-blind, placebo-controlled, randomized parallel pilot trial (1:1) conducted at a university hospital between 2016 and 2018. Eligibility included adults aged 18–65 with DSM-5 major depressive disorder who had been on stable antidepressant treatment for at least 4 weeks and had a baseline 17-item Hamilton Depression Rating Scale (HAM-D17) score of ≥17. The protocol was adjusted from the initial plan to permit any class of antidepressant (mono- or combination therapy) and to shorten the requirement for no antidepressant dose adjustment from 6 to 4 weeks. Exclusion criteria included bipolar disorder, psychotic or substance use disorders (except caffeine and nicotine), prior antidepressant-induced mania/hypomania, significant suicidal ideation, relevant medical comorbidities (including chronic lung disease), pregnancy or breastfeeding, contraindications to N2O (e.g. chronic B12 deficiency, pneumothorax, elevated intracranial pressure, intestinal obstruction), and inability to breathe through the nose. Participants were recruited from outpatient services, referrals and public advertisement. Subjects were randomised to inhalation of 50% N2O (diluted in 50% oxygen) or placebo consisting of 100% oxygen, each delivered by a nasal mask for 60 minutes, twice weekly for 4 weeks (eight sessions total). The N2O concentration (50%) was chosen for outpatient safety and because it had been used in prior work. Continuous monitoring of heart rate, peripheral oxygen saturation and blood pressure was performed during sessions; women of reproductive age had a urinary pregnancy test before starting. Blinding measures included having two unblinded collaborators set the device outside the assessor's presence, covering the device display, and using flavoured nasal masks to mask N2O's slight sweet odour. The primary outcome was change in HAM-D17 score from baseline to the end of week 4. Secondary outcomes included changes in Beck Depression Inventory-II (BDI-II), rates of therapeutic response (≥50% reduction from baseline) and remission (HAM-D17 ≤7; BDI-II ≤13), suicidal ideation measured by the Columbia-Suicide Severity Rating Scale (C-SSRS), emergence of manic symptoms assessed with the Young Mania Rating Scale (YMRS), and adverse events. Assessments were performed by a psychiatrist before and after each inhalation session (16 assessments total). Randomisation used permuted blocks generated online. Statistical analysis employed repeated-measures analysis of variance (ANOVA) for HAM-D17 (factors: time, drug, time×drug), with independent t-tests when interactions were significant; similar approaches were used for BDI-II and YMRS. Non-parametric tests addressed C-SSRS, and Fisher's exact test compared response/remission rates. Effect sizes were reported using partial eta-squared (Z2p) with predefined thresholds.

From 201 assessed for eligibility, 23 participants were randomised (12 to N2O, 11 to placebo). Two participants in the placebo group discontinued (one after session 1 for symptom worsening, another after session 5 for adherence difficulties), leaving 21 completers for end-of-trial outcome analyses. Baseline demographic and clinical characteristics, including HAM-D17 and BDI-II severity, were similar between groups. Primary outcome (HAM-D17): Both groups showed significant reductions in HAM-D17 scores over time, but the N2O group improved more. Groupwise mean HAM-D17 changed from 22.58±3.83 to 5.92±4.08 in the N2O group and from 22.44±3.54 to 12.89±5.39 in the placebo group; a significant time×drug interaction was observed (F15,285 = 2.26; p = 0.005; Z2p = 0.11). Within-session (pre- vs post-inhalation) reductions were significant in nearly all sessions in both groups, but only the N2O group reached mean values in the remission range at post-inhalation assessments in sessions 7 and 8. At the end of session 8, HAM-D17 remission occurred in 75% (9/12) of the N2O group versus 11.1% (1/9) of the placebo group (difference 0.64; p = 0.008, Fisher's exact test). Response rates were 91.7% (11/12) for N2O versus 44.4% (4/9) for placebo (difference 0.47; p = 0.046). Secondary outcomes: BDI-II scores fell significantly over time in both groups with large effect sizes, but there was no significant time×drug interaction (F15,285 = 1.12; p = 0.34; Z2p = 0.06). End-of-trial BDI-II remission rates were 66.7% (8/12) in the N2O group and 33.3% (3/9) in placebo (p = 0.20); response rates were 66.7% vs 55.6% (p = 0.67). Suicidal ideation (C-SSRS) distributions did not differ between groups at any assessment and no suicidal behaviours occurred. YMRS scores did not indicate manic or hypomanic episodes in either group; no participant exceeded a YMRS total score of 12. Adverse effects were more frequently reported with N2O. Common events included somnolence, paresthesia, nausea, vomiting and headache. In the N2O group (96 sessions total), four of 12 participants required temporary inhalation interruptions across 12 sessions; five sessions were discontinued before completion, three of them for a single participant. One participant in the N2O arm had increased blood pressure during one session. No serious adverse events were reported. In the placebo group (78 sessions), adverse effects were similar but fewer interruptions occurred.

Guimara and colleagues interpret their findings as indicating that adjunctive 50% N2O inhalation twice weekly for 4 weeks produced greater clinician-rated antidepressant benefit than placebo in symptomatic patients on stable antidepressant therapy. The HAM-D17 showed both acute within-session reductions and a cumulative effect across sessions in the N2O group, whereas the placebo group showed more limited cumulative change. Reported response and remission rates on the HAM-D17 were substantially higher in the N2O arm (response 91.7%, remission 75%) and reached statistical significance versus placebo. The authors note an inconsistency between clinician-rated (HAM-D17) and self-rated (BDI-II) measures: both instruments improved, but the superiority of N2O over placebo was evident only on the HAM-D17. They attribute this divergence in part to differences between clinician and self-rated scales and to known only-moderate correlations between them. Possible mechanisms for N2O's antidepressant effect are discussed, including NMDAR antagonism, interactions with the opioid system (as suggested by ketamine–naltrexone data), and effects on nitric oxide signalling. Acknowledged limitations include the small sample size, short 4-week duration, lack of post-treatment follow-up and absence of an intention-to-treat analysis. The investigators also recognise potential unblinding due to N2O-associated adverse effects and the use of 100% oxygen as placebo rather than an active comparator. Practical constraints led to sessions being scheduled at varying times; treatment-resistance status was not systematically recorded, though the high response rates suggest few participants were treatment-resistant. Measurement issues are highlighted: HAM-D and BDI-II include items (sleep, appetite, weight) that do not change rapidly, which complicates within-session comparisons. The chosen 50% N2O concentration prioritised outpatient safety, but optimal dosing and frequency remain unknown; the trial also lacked capnography or gas analysis to measure expired N2O or CO2. Finally, the authors call for larger, longer, and methodologically rigorous trials—preferably with active controls, intention-to-treat analyses, clearer distinction of treatment-resistant samples, and exploration of dose and administration schedules—to determine whether N2O is an effective antidepressant alone or as adjunctive therapy.