Attenuation of antidepressant and antisuicidal effects of ketamine by opioid receptor antagonism

Suicide rates have risen in the United States over recent decades and suicide is now a leading cause of death in younger age groups, creating an urgent need for rapid-acting interventions to reduce suicidal thinking and behaviour. Ketamine, administered intravenously (racemic) or intranasally as esketamine, produces rapid reductions in suicidal ideation within hours that can last days to a week, and its clinical effects have commonly been attributed to antagonism of the N-methyl-D-aspartate receptor (NMDAR). However, emerging evidence implicates the endogenous opioid system in mood regulation and suicidality, and recent work by the study team found that ketamine’s antidepressant effect in humans depends on opioid receptor activation. Against this background, Williams and colleagues tested whether opioid receptor blockade with oral naltrexone attenuates ketamine’s antisuicidality effects. The investigation reported here is a post hoc secondary analysis of a previously completed randomised, double-blind, placebo-controlled crossover trial in which participants received oral naltrexone (50 mg) or placebo prior to a standard 0.5 mg/kg intravenous ketamine infusion. Suicidality was assessed with HDRS-17 item 3, MADRS item 10, and the Columbia Suicide Severity Rating Scale (CSSRS), and the primary contrast for this analysis focused on ketamine responders (defined as >50% reduction in HDRS-17 on postinfusion day 1).

Participants were adults meeting DSM-5 criteria for a nonpsychotic, non-atypical major depressive episode within either major depressive disorder or bipolar II disorder. Inclusion required HDRS-17 total score ≥20 at enrolment, insufficient benefit from at least four prior antidepressant or somatic treatments as defined by the Antidepressant History Treatment Form, and at least six weeks of prior psychotherapy during the current MDE. Sixteen participants provided consent; 14 received at least one ketamine infusion and 12 completed both crossover conditions. Of the 12 completers, all were diagnosed with recurrent MDD; nine had a score ≥1 on HDRS item 3 and 11 had a score ≥1 on MADRS item 10 at screening, indicative of current suicidal ideation in most participants. A randomised, double-blind crossover design was used with two conditions: ketamine plus oral placebo (K + P) and ketamine plus oral naltrexone 50 mg (K + N). Placebo or naltrexone was given 45 minutes before a 0.5 mg/kg intravenous ketamine infusion to achieve peak naltrexone levels. The order of conditions was randomised and both participants and investigators were blinded. Suicidality and depressive outcomes were assessed at baseline (day 0) and on postinfusion days 1, 3, 5, 7, and 14 using HDRS-17, MADRS, and CSSRS. The trial was planned with an a priori power calculation that estimated 30 participants would be required to detect differences in the primary antidepressant outcome (>50% HDRS-17 reduction on day 1) with a moderate effect size and alpha 0.05, but the study was terminated early after 14 participants completed at least one treatment because naltrexone markedly reduced ketamine’s antidepressant effect. For this secondary analysis the primary suicidality measure was HDRS-17 item 3, with MADRS item 10 and total CSSRS score as additional measures. The principal analysis included all participants who completed both treatments (n = 12). A linear mixed model with compound symmetry covariance structure compared mean suicidality changes over time, treatment condition, and ketamine responder status (responder defined as >50% HDRS-17 reduction on day 1). Time, treatment, responder status, and the time × treatment × responder interaction were fixed factors, with time and treatment treated as repeated within-subject measures. Bonferroni correction was applied for multiple comparisons. Associations between changes in suicidality and changes in total depression scores were tested with Pearson correlations. Missing data for the primary outcome were absent; missing secondary outcome data were not imputed.

Fourteen participants received at least one infusion and 12 completed both crossover conditions (K + P and K + N). Among completers, seven met the prespecified ketamine responder criterion (≥50% HDRS-17 reduction on postinfusion day 1) following K + P. The primary outcome comparison (HDRS item 3) in ketamine responders showed that naltrexone pretreatment profoundly attenuated ketamine’s antisuicidality effect: HDRS item 3 reductions after K + P were significantly greater than after K + N at day 1 (Bonferroni, p < 0.001) and remained significantly different at days 3 (p = 0.014), 5 (p = 0.019), 7 (p < 0.001), and 14 (p = 0.003). Concordant results were observed for MADRS item 10, with significant separation between K + P and K + N across all postinfusion days (day 1 p < 0.001; day 3 p = 0.002; day 5 p = 0.002; day 7 p < 0.001; day 14 p < 0.001). The CSSRS total score showed significant differences favouring K + P at days 3 (p = 0.002), 7 (p = 0.006), and 14 (p = 0.004). When analysing the full set of completers (n = 12) with the linear mixed model, HDRS-17 item 3 showed significant effects of treatment condition (F1,107 = 7.399, p = 0.008), day (F5,107 = 9.014, p < 0.001), and the day × treatment × responder interaction (F16,107 = 3.165, p < 0.001), while responder status alone was not significant. For MADRS item 10 there were significant effects of treatment condition (F1,106 = 8.825, p = 0.004), day (F5,106 = 7.968, p < 0.001), responder status (F1,10 = 6.140, p = 0.032), and the three-way interaction (F16,106 = 4.412, p < 0.001). CSSRS analyses showed a significant main effect of treatment condition (F1,104 = 12.627, p = 0.001) but no significant effects of day, responder status, or their interaction. Paired tests found no significant carryover effects when comparing baseline HDRS item 3 prior to each treatment order. After Bonferroni correction, changes in suicidality and changes in total depression scores were significantly correlated on day 1 but not on days 3, 5, 7, or 14. The extracted text reports no missing primary outcome data; secondary outcome missingness was not imputed.

Williams and colleagues interpret the data as evidence that opioid receptor activation is necessary for ketamine’s acute antisuicidality effects in patients with severe depression who respond to ketamine. Pretreatment with naltrexone markedly reduced ketamine’s capacity to lower suicidal ideation on clinician-rated measures (HDRS item 3, MADRS item 10) and on the CSSRS, supporting the conclusion that opioid signalling contributes substantially to the observed antisuicidality response. The authors place these findings in the context of unresolved questions about ketamine’s mechanism of action: although NMDAR antagonism has long been considered central, the limited success of other NMDAR antagonists in treatment-resistant depression suggests additional pharmacology may be required. Opioid receptors are a plausible contributor given prior clinical and preclinical evidence, the antidepressant and antisuicidality effects of buprenorphine in other studies, and the investigators’ earlier report that ketamine’s antidepressant effect is opioid dependent. The discussion acknowledges that ketamine’s antisuicidality effect may be mediated partly through alleviation of depressive symptoms and partly via independent processes. In this dataset, changes in suicidality correlated with changes in total depression scores only at day 1, not at later time points, consistent with prior reports suggesting partial independence between mood and suicidality effects. Nevertheless, the study was underpowered to definitively dissociate those mechanisms. Key limitations noted by the authors include the small sample size driven by early termination after observing a large effect of naltrexone on the antidepressant response, and the fact that participants were not specifically recruited for active suicidality. Mechanistically, the design cannot distinguish whether naltrexone blocked a direct interaction of ketamine with opioid receptors or blocked ketamine-induced release of endogenous opioids, nor can it parse the relative roles of mu (MOR), delta (DOR), or kappa (KOR) opioid receptors. The absence of a naltrexone-plus-placebo arm to assess whether naltrexone on its own worsens mood is discussed; the authors argue that prior controlled trials do not support a consistent dysphoric effect of naltrexone and note data that chronic naltrexone may even protect against suicide. Finally, Williams and colleagues call for independent replication in prospective blinded trials and further mechanistic work to define which opioid receptor systems contribute to ketamine’s effects, emphasising the clinical importance given the overlapping public health crises of suicide and opioid overdose and the growing clinical use of ketamine for suicidal ideation.