Ketamine Modulates the Neural Correlates of Reward Processing in Unmedicated Patients in Remission from Depression

Major depressive disorder is associated with altered reward processing and persistent anhedonia, deficits that can remain during remission and predict onset. Earlier work shows that ketamine produces rapid antidepressant and anti-anhedonic effects detectable as early as 2 hours after infusion, but it is unclear whether these behavioural effects reflect direct engagement of reward-processing circuitry or are secondary to symptom improvement. Neuroimaging studies have implicated striatal, insular and prefrontal nodes of the mesocorticolimbic pathway in reward anticipation and feedback, and prior PET and fMRI reports have linked early post-ketamine metabolic and activation changes in such regions to symptom change, yet the temporal and mechanistic relationships remain unsettled. Kotoula and colleagues set out to test whether a single subanaesthetic ketamine infusion (0.5 mg/kg) modulates neural responses to reward anticipation and feedback independently of concurrent symptom change. They applied a well-validated monetary incentive delay (MID) task and a pre-defined set of regions of interest (ventral and dorsal striatum, ventral tegmental area (VTA), amygdala, insula, plus an exploratory sgACC analysis), scanning participants 2 hours after infusion in a double-blind crossover design. The investigators also measured plasma concentrations of ketamine and main metabolites, including (2R,6R)-hydroxynorketamine ((2R,6R)-HNK), to explore relationships between metabolite exposure and task-related brain activity.

The study used a double-blind, within-subject crossover design in which participants remitted from depression, free of antidepressant treatment and without current symptoms, received a single intravenous infusion of ketamine (0.5 mg/kg) in one session and saline placebo in the other. Infusions were delivered over a 40-minute steady-state period, sessions were separated by at least 7 days, and functional MRI scanning occurred 2 hours after the end of each infusion to coincide with the early antidepressant time window. Blood samples were collected pre-infusion, immediately post-infusion and 2 hours post-infusion to quantify ketamine, norketamine and two hydroxynorketamine isoforms ((2R,6R)-HNK; (2S,6S)-HNK). Behavioral measurement included the Psychotomimetic States Inventory (PSI) at the end of each infusion and the Snaith-Hamilton Pleasure Scale (SHAPS) pre-scan and 2 hours post-infusion to index psychotomimetic effects and anhedonia respectively. The cognitive probe was a MID task with 96 trials comprising high-win, low-win and neutral cues. Regressors modelled anticipation (high, low, neutral) and feedback (high win, low win, high no-win, low no-win) phases separately for each session. Imaging was performed on a 3-T scanner. Preprocessing used SPM12: realignment, coregistration to a T1 MPRAGE, DARTEL normalization and 8 mm smoothing. Motion parameters and framewise displacement were included as regressors; one participant was excluded for excessive motion. Region-of-interest (ROI) masks covered bilateral ventral striatum (nucleus accumbens), dorsal striatum (caudate, putamen), VTA, amygdala, insula and an sgACC mask used in an exploratory analysis. Mean beta estimates per ROI and contrast were extracted with MarsBar. Statistical analysis employed mixed-effects models to test overall treatment effects on each MID contrast, followed by paired t-tests for ROI comparisons and one-sample t-tests within sessions. Bonferroni correction set the ROI significance threshold at p=0.008. Robust regression assessed relationships between 2-hour metabolite levels and ketamine-induced ROI activation, with placebo beta values entered as covariates to account for individual baseline activation; false discovery rate (FDR) correction was applied to metabolite analyses.

Participant-level and behavioural findings: Statistical degrees of freedom reported (F tests with df 1,36) indicate a sample of 37 participants entered analyses, with one excluded for motion. Ketamine produced the expected acute psychotomimetic effects: PSI total scores were substantially higher after ketamine (mean=48.4, SD=22.9) than after placebo (mean=15.1, SD=10.6). SHAPS scores were low at baseline and did not change significantly 2 hours post-ketamine, indicating no measurable change in self-reported anhedonia in this remitted cohort. Task performance did not differ between drug conditions for total money won (ketamine mean=45.1, SD=5.5; placebo mean=43.3, SD=9.1) and reaction times showed the expected main effect of reward magnitude (faster for high-win trials, F(2,36)=23.2, p<.0001) with no drug-by-reward interaction. Whole-brain analyses did not reveal significant ketamine-versus-placebo differences. ROI analyses, however, identified several treatment effects. During anticipation, a main effect of ketamine increasing activation across ROIs was observed for all win trials versus neutral (F(1,36)=9.261, p=.003). When individual ROIs were examined for high win versus neutral anticipation, increased activation was noted in the nucleus accumbens and caudate, but these did not survive correction for multiple comparisons. Effects during the feedback phase were more robust. For low win feedback versus neutral, ketamine increased activity across ROIs (F(1,36)=4.563, p<.001), and paired tests showed significant ketamine-induced increases in the nucleus accumbens and putamen that survived Bonferroni correction. No-win feedback contrasted with neutral also showed a main effect across ROIs (F(1,36)=5.467, p<.001) and, for high no-win versus neutral, F(1,36)=5.859, p=0.016, but individual ROI effects did not survive multiple-comparison correction. Comparing all win versus all no-win trials produced a main effect across ROIs (F(1,36)=5.036, p<.001) without any single ROI meeting corrected significance. Metabolite associations: In robust regression analyses (placebo beta as covariate), plasma levels of (2R,6R)-HNK at 2 hours post-infusion positively correlated with VTA activation for the contrast low win feedback versus neutral (n=22, p_FDR=0.03). A similar positive relationship between (2R,6R)-HNK and caudate activation for high no-win versus neutral was observed but did not survive multiple-comparison correction. No significant relationships emerged between ROI activation and plasma ketamine, norketamine or (2S,6S)-HNK levels. An exploratory sgACC analysis showed no significant ketamine effects for any MID contrast.

Kotoula and colleagues interpret their findings as evidence that ketamine can directly modulate reward-related neural activity within the mesolimbic pathway approximately 2 hours after administration, independent of changes in depressive symptoms or anhedonia in this remitted sample. The most consistent and robust effects were observed during the feedback phase of the MID task, with increases in nucleus accumbens and putamen activation for low-win feedback surviving correction for multiple comparisons. The authors suggest these changes could reflect enhanced sensitivity to feedback—particularly to negative or non-reward outcomes—and propose that increasing the salience of no-win trials might boost motivation and thereby be relevant to alleviating anhedonia. The reported positive correlation between (2R,6R)-HNK plasma levels and VTA activation is presented as preliminary support for a role of ketamine metabolites in modulating reward circuitry; the authors link this observation to hypotheses that (2R,6R)-HNK can directly engage AMPA receptors and trigger plasticity-related pathways. They note that such a mechanism aligns with animal-model data and PET findings of early post-ketamine metabolic changes that correlate with symptom improvement, although direct evidence for increased synaptic plasticity in humans remains indirect. Key limitations acknowledged by the investigators include the absence of a healthy control group, which prevents establishing whether ketamine normalises remitted individuals' reward responses towards healthy levels. The MID task's design emphasises anticipation and contains relatively fewer feedback trials, which constrains power for outcome-related contrasts; the task also lacked an explicit loss condition, limiting interpretation regarding punishment-specific effects. Finally, the cohort consisted of remitted, unmedicated participants, so extrapolation to acutely depressed patients requires caution. The authors recommend future, better-powered studies that include active patient samples, tasks optimised for feedback and loss conditions, and concurrent measurement of metabolism, functional responses, symptoms and metabolite levels to build a more comprehensive mechanistic model of ketamine's antidepressant and anti-anhedonic actions.