The potential pro-cognitive effects with intravenous subanesthetic ketamine in adults with treatment-resistant major depressive or bipolar disorders and suicidality

Cognitive dysfunction is a central feature of mood disorders such as major depressive disorder (MDD) and bipolar disorder (BD), affecting attention, memory, learning, executive function and processing speed, and contributing to poor occupational and functional outcomes. Conventional antidepressants generally have limited or mood-mediated effects on cognition, and vortioxetine is the only agent with replicated evidence of direct pro-cognitive effects; by contrast, ketamine is established for rapid antidepressant and anti‑suicidal effects but its cognitive effects in clinical populations remain uncertain. Prior small and heterogeneous studies have suggested no severe persistent cognitive deficits following therapeutic ketamine and have raised the possibility of domain-specific improvements, but questions remain about whether any cognitive changes are independent of mood improvement or secondary to practice effects or sample characteristics. Zhou and colleagues set out to determine whether repeated, subanaesthetic intravenous ketamine produces direct pro‑cognitive effects in a large, real‑world sample of adults with treatment‑resistant or suicidal MDD or BD. The study aimed to assess changes across four cognitive domains (processing speed, working memory, visual learning and verbal learning) measured before treatment, 1 day after a six‑infusion course, and 14 days later, and to use mediation analysis to test whether cognitive changes were independent of reductions in depressive symptoms.

This was a single‑arm, open‑label clinical study conducted at the Affiliated Brain Hospital of Guangzhou Medical University. The project originally screened 185 patients between September 2016 and January 2018 and enrolled 100 patients to receive intravenous ketamine; the present analysis includes 136 patients who received ketamine and, more specifically, 111 patients who completed six infusions and had baseline and day‑13 cognitive measurements (13 patients, 11.7%, lacked day‑26 data). Ethical approval and informed consent were obtained. Eligible participants were adults aged 18–65 with a primary DSM‑V diagnosis of MDD or BD confirmed by SCID, currently experiencing moderate to severe depression (HAMD‑17 ≥ 17) and either treatment resistance (insufficient response to ≥2 adequate antidepressant trials) or suicidal ideation (score ≥2 on the first five items of the Beck Scale for Suicide Ideation, SSI part I). Exclusion criteria included alcohol or substance dependence, other primary serious mental disorders such as schizophrenia, or unstable/serious medical conditions; comorbid anxiety disorders were permitted if not the primary cause of illness in the prior year. Psychotropic medications had to be stable for ≥4 weeks and were maintained during the six‑infusion period. Intervention consisted of six intravenous infusions of racemic ketamine at 0.5 mg/kg diluted in 40 ml saline and administered over 40 minutes on a Monday–Wednesday–Friday schedule across 12 days. Depression severity was assessed via the Montgomery‑Åsberg Depression Rating Scale (MADRS) at baseline, day 13 (one day after the final infusion) and day 26 (14 days after the final infusion). Cognitive function was measured with the MATRICS Consensus Cognitive Battery (MCCB) at the same time points; four domains were selected and converted to T‑scores (mean 50, SD 10): processing speed, working memory, visual learning and verbal learning. Statistical analyses used IBM SPSS v22 with α = 0.05. Linear mixed models tested main effects of treatment (time) and diagnosis (MDD vs BD) and their interaction, with covariates age, sex, education, illness duration, BMI and concomitant use of mood stabilisers/benzodiazepines/antipsychotics. Treatment-by‑MADRS interactions were examined to assess whether cognitive changes were moderated by depressive symptom change. Bonferroni‑corrected post hoc comparisons and Cohen's d effect sizes were reported for baseline → day 13 and baseline → day 26 contrasts. An intention‑to‑treat approach was applied for analyses including patients missing day‑26 data. Mediation (path) analyses using PROCESS v2.15 in SPSS tested whether changes in MADRS mediated the relationship between treatment time point and cognitive scores.

Participant details: the analytic sample comprised 111 patients who completed six ketamine infusions and had baseline and day‑13 cognitive assessments; within this group, 84 had unipolar depression (MDD) and 27 had bipolar depression (BD). The MDD and BD subgroups were similar in age, education, BMI and prior psychiatric hospitalisation, while BD patients had a higher proportion of females, more family history of psychiatric disorders and a longer illness duration. The mean age of the overall sample was reported as relatively young (34.6 years) in the Discussion. Depressive symptoms: linear mixed modelling showed a significant main effect of treatment on MADRS scores (F = 111.170, P < 0.001) with significant decreases from baseline to day 13 (t = 13.737, P < 0.001) and to day 26 (t = 12.536, P < 0.001) in the total sample. There were no significant main effects of diagnosis nor treatment-by‑diagnosis interactions on MADRS. Processing speed: a significant main effect of treatment was observed for processing speed (F = 10.317, P < 0.001). Mean processing speed T‑scores increased from baseline to day 13 (effect size d = 0.501) and to day 26 (d = 0.654), indicating improvement that persisted at least two weeks. The treatment-by‑MADRS interaction for processing speed was significant, suggesting that changes in depressive symptoms moderated processing‑speed improvements. Path analysis indicated both a statistically significant direct effect of ketamine on processing speed (β = 2.444, P = 0.017) and a significant indirect effect via MADRS change (β = 1.220, P = 0.048), implying that processing‑speed gains were partly independent of mood improvement. Verbal learning: treatment also produced a significant main effect for verbal learning (F = 9.434, P < 0.001). Mean verbal learning increased from baseline to day 13 (d = 0.362) but was not maintained at day 26. The treatment‑by‑MADRS interaction was not significant. Mediation analysis showed a non‑significant direct effect of ketamine on verbal learning (β = -1.963, P = 0.052) but a significant indirect effect via MADRS change (β = 1.386, P = 0.024), consistent with verbal‑learning improvement being mediated by reduction in depressive symptoms. Working memory and visual learning: neither domain showed significant treatment main effects, diagnosis effects, treatment‑by‑diagnosis interactions nor treatment‑by‑MADRS interactions (all P > 0.05). Post hoc comparisons indicated a significant decrease in mean visual learning score from baseline to day 13, but because the linear mixed models showed no overall treatment effect for these domains, no mediation analyses were performed. Subgroup (MDD vs BD) analyses yielded similar patterns to the total sample. The text references tables and figures for detailed values, but those items are not reproduced in the extracted text.

Zhou and colleagues interpret their findings as evidence that six subanaesthetic infusions of intravenous ketamine did not impair cognitive function in this clinical sample and were associated with improvements in processing speed and transient improvement in verbal learning after the treatment course. They highlight that the processing‑speed improvement persisted to day 26 and that mediation analysis indicated this domain benefited partly independently of reductions in depressive symptoms, whereas verbal‑learning gains appeared to be mediated by mood improvement. The authors situate these results alongside earlier research that has not identified clear cognitive harm from therapeutic ketamine, and they note both concordant and discordant findings in prior studies: some reports have found domain‑specific cognitive improvements after repeated ketamine, while single‑dose studies have shown minimal or transient effects. They acknowledge the recurrent methodological concern of practice (learning) effects given repeated cognitive testing, and they contrast ketamine's preliminary pro‑cognitive signature with vortioxetine, the only drug to date recognised for a direct cognitive effect in depression. Mechanistically, the investigators discuss potential biological mediators consistent with the literature: ketamine's modulation of glutamatergic signalling, effects on brain‑derived neurotrophic factor (BDNF), synaptic connectivity and brain volumes in frontal and limbic regions, and alterations in the kynurenine pathway including increased kynurenic acid. They note prior subgroup data from the same project showing increased plasma BDNF at day 13 and day 26 relating to antidepressant outcome, and they suggest functional neuroimaging, including functional near‑infrared spectroscopy (fNIRS), could clarify neurophysiological changes underlying pro‑cognitive effects. The authors acknowledge several limitations. The open‑label, single‑arm design and lack of a control group prevent definitive attribution of cognitive changes to ketamine rather than nonspecific treatment or practice effects. The fixed protocol (six 0.5 mg/kg infusions over two weeks) leaves unanswered questions about optimal dosing and frequency; concerns remain about potential cognitive harm with frequent or long‑term use, as seen in recreational use. Concurrent psychotropic medications (including benzodiazepines, lithium and antipsychotics) and comorbidities were not controlled and may have influenced cognitive outcomes. The sample was relatively young, had a short illness duration and a high rate of antipsychotic use, which may limit generalisability. Finally, repeated testing with the same cognitive instruments may produce learning effects that confound interpretation. The investigators recommend randomised controlled designs, inclusion of active comparators or standard antidepressant treatment, control for comorbidities and concomitant medications, and incorporation of neuroimaging to further probe mechanisms and causal pathways. They also propose assessing whether procognitive changes mediate improvements in suicidality and psychosocial functioning.

The authors conclude that in this real‑world sample of adults with MDD or BD who were treatment‑resistant or expressed suicidal ideation, six infusions of subanaesthetic intravenous ketamine did not impair cognitive function and showed a potential mood‑independent pro‑cognitive effect on processing speed over a 26‑day observation. They suggest these findings could inform development of interventions targeting cognitive impairment in treatment‑resistant depression and recommend future studies to evaluate whether procognitive effects of ketamine mediate broader clinical outcomes such as suicidality and psychosocial function.