Cognitive Function Mediates the Anti-suicide Effect of Repeated Intravenous Ketamine in Adult Patients With Suicidal Ideation

Suicide is a major global public-health problem and is frequently associated with mood disorders such as major depressive disorder (MDD) and bipolar depression. Conventional treatments (antidepressants, cognitive behavioural therapy, electroconvulsive therapy) can reduce suicidal ideation but typically require weeks to take effect, leaving many patients at persistent risk. Over the past two decades, repeated evidence from randomised trials has shown that subanesthetic doses of ketamine, an NMDA receptor antagonist, produce rapid antidepressant and anti-suicide effects, sometimes within hours. Concurrently, cognitive dysfunction (notably in processing speed, working memory, executive function and decision-making) is recognised as a core feature of depression and has been linked to suicidal vulnerability. Preliminary reports have also suggested that ketamine may produce pro-cognitive effects in some cognitive domains. Z. and colleagues set out to test the hypothesis that ketamine's anti-suicide effects are at least partly mediated by pro-cognitive effects. Specifically, the authors aimed to determine whether improvements in depressive symptoms and/or in selected cognitive domains (processing speed, working memory, visual learning and verbal learning) mediated changes in suicidal ideation after six intravenous ketamine infusions given over 2 weeks to adults with unipolar or bipolar depression and baseline suicidal ideation. They also examined baseline predictors of anti-suicide response to repeated ketamine infusions.

This analysis used data from a larger single-centre, open-label study of intravenous ketamine in adults with MDD or bipolar depressive disorder. From the larger cohort, 86 participants who received six ketamine infusions and completed clinician-rated measures at baseline and day 13 were included in the present mediation analysis. Eligible participants were adults (≥18 years) meeting DSM-5 criteria for MDD or bipolar depression, with moderate-to-severe depressive symptoms (HAMD-17 ≥ 17) and evidence of suicidal tendency (Beck Scale for Suicide Ideation (SSI) part I score ≥ 2 at screening). Exclusion criteria included alcohol or substance dependence, unstable medical conditions, psychotic symptoms and recent mania/hypomania (within 6 months for bipolar participants). Participants continued their prescribed psychotropic medications at stable doses during the 2-week infusion period. All participants received six adjunctive intravenous infusions of ketamine hydrochloride at 0.5 mg/kg administered over 40 minutes on a Monday–Wednesday–Friday schedule across two weeks. Vital signs were monitored during and after each infusion. Clinical assessments occurred at baseline, 1 day after the sixth infusion (day 13) and 14 days after the sixth infusion (day 26). Primary measures were clinician-rated depressive symptoms (Montgomery–Åsberg Depression Rating Scale, MADRS) and current suicidal ideation (SSI part I). Cognitive function was assessed with four domains drawn from the MATRICS Consensus Cognitive Battery (MCCB): processing speed (Category Fluency, Trail Making A, BACS), working memory (WAIS-III letter–number sequencing; WMS-III Spatial Span), visual learning (BVMT-R) and verbal learning (HVLT-R). Domain scores were standardised to T scores (mean 50, SD 10). For statistical analysis the authors used linear mixed models to test change over time in MADRS, SSI and cognitive domain scores, with measurement point (baseline, day 13, day 26) as a factor and covariates including age, sex, education, illness duration, BMI, primary diagnosis and concomitant medications (mood stabiliser/benzodiazepine/antipsychotic co-prescription; one versus two antidepressants). Mediation was tested using PROCESS v2.15 in SPSS: measurement point served as the independent variable (X), SSI score as the dependent variable (Y), MADRS total score as the first mediator (M1) and each cognitive domain T score as the second mediator (M2) in serial mediation models. Covariates listed above were included. The authors report unstandardised coefficients and p-values. Finally, binomial logistic regression (forward LR) was used to identify baseline predictors of anti-suicide response and remission on the SSI at day 13.

Sample and retention: Of 104 patients with MDD or bipolar depression and suicidal ideation who received ketamine, 86 patients completed six infusions and had MADRS, SSI and MCCB data at baseline and day 13; 10 participants (11.6%) did not complete the day 26 visit. Clinical and cognitive change over time: Six ketamine infusions were associated with significant reductions in SSI and MADRS scores. Reported SSI means (± SD) were 4.7 ± 3.0 at baseline, 0.9 ± 1.9 at day 13 and 1.0 ± 1.9 at day 26 (F = 79.643, P < 0.001). The authors also report improvements in certain cognitive domains (processing speed and verbal learning) after treatment, although full domain-wise statistics are not reproduced in the extracted text. Mediation analysis: The overall (total) effect of ketamine treatment (measurement point) on SSI score was significant (coef = -1.853, 95% CI [-2.2, -1.5]). When serial mediation models included MADRS as M1 and each cognitive domain as M2, the authors report that both direct and total indirect effects were statistically significant for models with processing speed, working memory, verbal learning and visual learning entered as M2 (direct effects ranged approximately from coef = -1.064 to -1.352; total indirect effects ranged approximately from coef = -0.501 to -0.788 with respective 95% CIs). Despite these findings, the authors concluded that, of the tested variables, MADRS total score and processing speed T score were significant partial mediators of the relationship between ketamine treatment and improvements in suicidal ideation. In contrast, working memory, verbal learning and visual learning did not demonstrate a statistically significant mediational role in the fully specified models. Predictors of response and remission: On the SSI at day 13 the response rate was 80.2% and the remission rate was 64.0%. In logistic regression models that included the four baseline cognitive domains, baseline processing speed was the only significant predictor of achieving an anti-suicide response (B = 0.062, P = 0.028) and also predicted remission when cognitive domains alone were entered (B = 0.075, P = 0.004). When baseline MADRS and SSI scores were included alongside the cognitive domains in the remission model, baseline SSI score (B = -0.275, P = 0.001) remained a significant predictor and baseline processing speed was excluded from the final model. Other notes: The extracted text does not report adverse event rates or detailed domain-by-domain numerical changes for all cognitive measures, nor does it reproduce the full tables or figures referenced by the authors.

Z. and colleagues interpret their findings as indicating that six subanesthetic ketamine infusions in adults with unipolar or bipolar depression and baseline suicidal ideation were associated with rapid improvements in mood, reductions in suicidal ideation and pro-cognitive effects (notably processing speed and verbal learning). Their mediation analyses led them to conclude that the anti-suicide effect of repeated ketamine infusions was partially mediated by improvements in overall depressive symptoms (MADRS) and by improvements in processing speed, while measures of working memory, verbal learning and visual learning did not significantly mediate changes in suicidal ideation in the fully specified models. The authors note that greater baseline processing speed predicted a higher likelihood of anti-suicide response to repeated ketamine infusions. The authors frame these findings within distinctions between cold cognition (emotion-independent functions such as memory, attention and executive function) and hot cognition (emotion-laden processes such as rumination and suicidal ideation). They suggest that ketamine may modulate both hot and cold cognitive processes, and that pro-cognitive effects may contribute to reductions in suicidal ideation. Mechanistic discussion highlights ketamine’s glutamatergic modulation, increased AMPA throughput, and downstream effects on BDNF and mTOR signalling that promote synaptogenesis; the authors also cite neuroimaging findings (changes in prefrontal connectivity, hippocampal, amygdala and nucleus accumbens volumes) as plausible substrates linking ketamine’s antidepressant, anti-suicide and procognitive effects. The authors acknowledge several important limitations. The study was a single-arm, open-label design without a control group, which permits expectation bias in clinician- and patient-rated outcomes. Only four cognitive domains were assessed from the MCCB and the study did not include dedicated decision-making tasks (e.g. Iowa Gambling Task), limiting the assessment of cognitive functions most consistently linked to suicidality. Participants with severe suicidal ideation or recent suicidal behaviour were excluded for safety, so results cannot be generalised to those at highest acute risk and cannot be extrapolated to suicidal behaviours (attempts or completion). Concomitant psychiatric medications were maintained and could confound cognitive and clinical changes, limiting applicability to medication-free patients. Finally, the authors emphasise that the study focused on reductions in suicidal ideation and that reductions in ideation do not necessarily predict reductions in suicidal behaviour; they recommend future work that assesses behavioural outcomes and integrates neuroimaging to clarify mechanisms.