Nitrous Oxide for Treatment-Resistant Major Depression: a Proof-of-Concept Trial

Treatment-resistant depression (TRD) is described as a severe form of major depressive disorder in which patients fail multiple standard antidepressant treatments and face poor long-term prognosis; the authors note that about one in three patients with major depression meet criteria for TRD. The NMDA (N-methyl-D-aspartate) receptor has been implicated in depression neurobiology, and earlier studies have shown that NMDA receptor antagonists such as ketamine can produce rapid antidepressant effects at subanesthetic doses. Given a similar primary mechanism of action, nitrous oxide was proposed as a candidate rapid-acting treatment for TRD because it is an inhalational general anaesthetic that acts as a noncompetitive NMDA receptor inhibitor. Nagele and colleagues therefore conducted a proof-of-concept clinical trial to test whether a single session of nitrous oxide inhalation produces immediate (2-hour) and sustained (24-hour) antidepressant effects in well-characterised patients with TRD. The trial aimed to detect an early efficacy signal that could justify larger studies and to collect preliminary safety and tolerability data in this population.

The investigators designed a randomised, placebo-controlled crossover pilot trial in which 20 evaluable patients with TRD each received two 1-hour inhalation sessions one week apart: active treatment (up to 50% nitrous oxide/50% oxygen) and placebo (50% nitrogen/50% oxygen). Session order was randomised by a computer generator. To preserve blinding the teams administering gas and performing psychiatric assessments were physically separated, treatment records were kept separately until study completion, the equipment and procedures were identical between sessions, and patients were told they would receive either nitrous oxide or an air mixture with a high nitrogen component. Participants were adults aged 18–65 meeting DSM-IV-TR criteria for major depressive disorder without psychosis, with pretreatment HDRS-21 > 18, and with TRD defined as at least two failed adequate antidepressant trials in the current episode and at least three lifetime antidepressant failures. Major psychiatric exclusions included bipolar disorder, psychotic disorders, recent substance abuse, active suicidal intention, prior NMDA antagonist exposure, and ongoing electroconvulsive therapy. Medical exclusions included conditions contraindicating nitrous oxide and pregnancy/breastfeeding. Patients were required to maintain stable antidepressant or psychotherapy regimens for 4 weeks prior to and during the study. During active treatment the nitrous oxide concentration was titrated over the first 10 minutes toward a 50% inspiratory concentration (the dose commonly used in clinical sedation settings); the total gas flow was 4–8 L/min and administration used a standard anaesthesia facemask with monitoring of inhaled/exhaled concentrations. Continuous physiologic monitoring (ECG, pulse oximetry, noninvasive blood pressure, end-tidal CO2) was provided under an attending anaesthesiologist, and patients were observed in recovery for 2 hours post-treatment. Outcome assessments occurred at baseline, 2 hours and 24 hours after each session (six time points in total). The primary endpoint was change in the 21-item Hamilton Depression Rating Scale (HDRS-21) at 24 hours. Secondary endpoints included the Quick Inventory of Depressive Symptomatology Self-Report (QIDS-SR). Psychiatric safety assessments targeted suicidality and emergent psychosis; other safety outcomes included cardiovascular, respiratory, and CNS adverse events and measurement of plasma total homocysteine to estimate nitrous oxide-induced vitamin B12 inactivation. For statistical analysis the primary HDRS-21 outcome was evaluated using a repeated-measures mixed effects linear model with restricted maximum likelihood estimation. Because a carryover effect was observed, the model included a randomisation group term and a three-way interaction (treatment × time × randomisation group), and the investigators also conducted an analysis restricted to the first treatment session (parallel-group comparison). Response and remission comparisons used an exact binomial test for paired data, with odds ratios presented; the pilot sample size (20 evaluable patients) was chosen pragmatically based on prior ketamine studies in similar populations. All reported p values were two-sided and p < .05 was considered statistically significant.

Between November 2012 and February 2014, 24 patients were enrolled; after three screen failures 21 were randomised, and one patient withdrew before outcome assessment, leaving 20 evaluable patients who completed both sessions (modified intention-to-treat). The cohort had chronic illness: the mean lifetime duration of major depressive disorder was 19 years, the median number of failed adequate antidepressant trials was eight, and patients were taking a median of two antidepressants at study entry. Median baseline HDRS-21 score was 23. Fifteen patients completed the full 60-minute nitrous oxide session. Two patients had brief 5-minute interruptions and three patients discontinued nitrous oxide early (at 55, 28, and 18 minutes) because of emotional discomfort, regurgitation, claustrophobia, or nausea/vomiting. The mean duration of nitrous oxide treatment was 55.6 ± 2.5 minutes at a median inspiratory nitrous oxide concentration of 44% (interquartile range 37%–45%). All placebo sessions were completed for the full 60 minutes. On the primary outcome the extracted text reports significant improvements in depressive symptoms after nitrous oxide compared with placebo. Specifically, the report states a mean HDRS-21 change at 2 hours of 24.8 points and at 24 hours of 25.5 points following nitrous oxide; these figures represent reductions in HDRS-21 score (improvement) although the extraction appears to have omitted explicit negative signs in some places. The corresponding reported comparisons for placebo show smaller, non-significant changes (for example, at 24 hours the placebo change is reported as 22.8 points, p = .07). The between-treatment comparisons for HDRS-21 change favoured nitrous oxide (comparison p < .001). The QIDS-SR also showed a significant reduction at 24 hours after nitrous oxide; the extracted text gives this as a mean change of 23.2 points (95% CI 21.3 to 25.0, p = .001) though the formatting suggests the sign may be missing in extraction. Using conventional thresholds, four patients (20%) met treatment response criteria (≥50% reduction in HDRS-21) 24 hours after nitrous oxide versus one patient (5%) after placebo (odds ratio 4.0, 95% CI 0.45–35.79). Three patients (15%) achieved full remission (HDRS-21 ≤ 7) after nitrous oxide compared with none after placebo (odds ratio 3.0, 95% CI 0.31–28.8). When HDRS-21 severity was collapsed into five levels, 7 of 20 patients (35%) showed at least a two-level improvement 24 hours after nitrous oxide versus 2 patients (10%) after placebo (p = .06). The crossover design exhibited a carryover effect: several patients had markedly lower HDRS-21 scores at the start of their second session, and the statistical test for carryover was significant (p = .02). To address this, the investigators analysed only the first treatment session in a parallel-group manner (10 patients who received nitrous oxide first versus 10 who received placebo first). In that first-session-only analysis, patients who received nitrous oxide first had significant HDRS-21 reductions at 2 hours, 24 hours, and at 1 week compared with those who received placebo first (reported mean reductions for the nitrous-oxide-first group were 27.1 points at 2 hours, 28.6 points at 24 hours, and 25.5 points at 1 week; the extracted text again appears to omit minus signs but these are presented as reductions). Safety data indicated no serious adverse events. Adverse events were described as brief and of mild to moderate severity; a subset of patients required interruption or discontinuation of nitrous oxide for transient symptoms (nausea, anxiety, vomiting). Plasma total homocysteine did not increase after nitrous oxide or placebo in the extracted data, suggesting minimal acute inactivation of vitamin B12-dependent metabolism after a single exposure.

Nagele and colleagues interpret the findings as providing preliminary, proof-of-concept evidence that a single 1-hour inhalation of nitrous oxide can produce rapid antidepressant effects in patients with TRD that are sustained for at least 24 hours and, in some patients, for up to 1 week. In this small pilot cohort nitrous oxide produced treatment response in 20% of patients and remission in 15% by conventional HDRS-21 criteria. The investigators emphasise that adverse events were typically mild to moderate, transient, and reversible, although some patients experienced emotional discomfort or nausea sufficient to interrupt or stop treatment. The discussion places these results in the context of prior work with NMDA receptor antagonists, particularly ketamine, noting a similar rapid onset of antidepressant action but an apparent absence of ketamine-like psychotomimetic side effects with nitrous oxide; the authors suggest that differences in pharmacokinetics and receptor pharmacology may account for some differences among NMDA antagonists. They also acknowledge that other receptor systems (for example, nicotinic acetylcholine receptors) might contribute to rapid antidepressant actions, and that differences between NMDA antagonists (for example, memantine versus ketamine) remain to be clarified. Key limitations acknowledged by the investigators include the carryover effect observed in the crossover design, which altered baseline scores at the second session and complicates interpretation of pooled crossover estimates. Blinding was another concern: nitrous oxide produces noticeable subjective effects (sedation, mild sweet smell/taste) that could have unmasked treatment assignment, and the authors did not formally assess whether patients guessed their allocation. The choice of assessment instruments was also noted as a limitation because HDRS-21 and QIDS-SR include items (sleep, weight) that are designed for changes over days to weeks and may be suboptimal for very rapid antidepressant effects; alternative short-term symptom instruments might be preferable in future studies. Dosing uncertainty was acknowledged — the trial used a conventional 50% inspiratory concentration based on sedation practice in dentistry and obstetrics but the optimal antidepressant dose and duration remain unknown. Safety considerations highlighted by the authors include the potential for adverse effects with repeated dosing, particularly nitrous oxide-induced inactivation of vitamin B12 and related haematologic or neurologic complications when exposures are frequent or in vulnerable patients. The abuse potential of nitrous oxide was also raised as a clinical concern that this pilot trial could not address. The investigators conclude that while the results are promising, the trial was an early Phase II pilot and that larger, controlled studies are required to determine optimal dosing, replicate efficacy, and thoroughly establish the risk–benefit profile before clinical adoption.