Ketamine normalizes brain activity during emotionally valenced attentional processing in depression

Reed and colleagues frame the study within the unmet need for faster-acting pharmacological treatments for major depressive disorder (MDD). They note that ketamine, a glutamatergic modulator, produces rapid antidepressant effects but that its precise neural mechanisms—particularly effects on cognitive and affective processing domains such as emotion processing—remain incompletely understood. Earlier behavioural work suggests that people with MDD often show a cognitive bias toward negative emotional information, and prior neuroimaging studies using dot probe tasks and other emotion paradigms have revealed group differences in brain regions including the cingulate and insula even when behavioural bias is absent. The authors point out that most previous ketamine fMRI work lacked simultaneous, longitudinal assessments of both MDD and healthy control (HC) participants and often did not include placebo controls, limiting interpretation of ketamine-specific effects on emotional processing networks. This study therefore set out to examine how a subanesthetic ketamine infusion affects brain activity during an emotionally valenced attentional dot probe task in unmedicated, treatment-resistant MDD participants and HCs. Using a randomized, double-blind, placebo-controlled crossover design with multiple fMRI time points (baseline, ~2 days post-infusion, and ~11 days post-infusion for each drug condition), the researchers hypothesised that ketamine would alter activation in regions implicated in emotional processing and depression, and that these effects might differ between MDD and HC groups. The approach was intended to test whether ketamine produces a normalising pattern of brain function during attentional-emotional processing in MDD relative to controls.

The study used a randomized, double-blind, placebo-controlled crossover design. Participants first completed a baseline fMRI session and were then randomised to receive either ketamine (0.5 mg/kg administered intravenously over 40 minutes) or saline placebo; two weeks later they crossed over to the alternate condition. Scans of interest occurred one to three days post-infusion (95% at two days) and nine to 13 days post-infusion (85% at 11 days), yielding five nominal scanning sessions: baseline, post-ketamine, interim-ketamine, post-placebo, and interim-placebo. Depressive symptoms were measured with the Montgomery–Åsberg Depression Rating Scale (MADRS) before infusion and on the day of the post-infusion scan. Participants comprised 33 treatment-resistant MDD individuals (12 male, 21 female; mean age 36.1 ± 9.7 years) and 26 healthy controls (10 male, 16 female; mean age 33.9 ± 10.4 years), aged 18–65. MDD inclusion criteria included onset < 40 years, current episode ≥ 4 weeks, MADRS ≥ 20 at baseline, and failure to respond to at least one adequate antidepressant trial (mean of six failed trials reported). Subjects were tapered off psychiatric medications and had a drug-free period of at least two weeks prior to study procedures. Exclusion criteria included recent substance dependence/abuse, psychosis, medical conditions affecting brain physiology, MRI contraindications, and anatomical brain abnormalities; healthy controls could not have personal or first‑degree family Axis I diagnoses. Participants were studied as inpatients and provided informed consent under NIH approval. During each fMRI session participants performed a dot probe task using emotional face stimuli (angry, happy, neutral). Each trial presented two faces side-by-side for 500 ms (one emotional and one neutral, or two neutrals), followed by a lateralised dot probe for 200 ms; congruent trials had the probe replace the emotional face, incongruent trials replaced the neutral face. The design was mixed block/event-related with angry and happy blocks; two runs per session were used and stimuli sets were counterbalanced across drug arms. Behavioural inclusion criteria required ≥ 75% task accuracy per session; reaction times between 200 and 1500 ms were analysed. Imaging was acquired on a 3T GE scanner with BOLD echo-planar imaging (TR = 2500 ms, TE = 23 ms, voxel ≈ 3.75 × 3.75 × 3.5 mm). Preprocessing in AFNI included despiking, slice timing correction, realignment, anatomical-coregistration, nonlinear normalisation to Talairach space, 6 mm smoothing, motion regression and censoring; scans with > 15% censored time points were excluded. First-level regressors modelled emotion blocks and four event types (angry congruent, angry incongruent, happy congruent, happy incongruent), plus instruction screens; fixation was the baseline. Group-level analyses used linear mixed-effect models with factors emotion, congruency, scan session, and diagnostic group. Voxel-wise thresholding used p < .001 with cluster-level family-wise error (FWE) correction at p < .05 computed via 3dClustSim (ACF). Primary contrasts emphasised post-ketamine versus post-placebo (the drug-session contrast). A follow-up mixed-effect analysis tested associations between post-infusion activation and percent change in MADRS in MDD participants, using the same voxel and cluster thresholds. Behavioural reaction times and attentional bias scores (incongruent minus congruent RT per emotion) were analysed with mixed models; only correct trials were included.

Data availability varied across sessions: usable scans included 52 baseline (32 MDD; 20 HC), 44 post-ketamine (27 MDD; 17 HC), 40 interim-ketamine (28 MDD; 12 HC), 42 post-placebo (26 MDD; 16 HC), and 35 interim-placebo (24 MDD; 11 HC). MADRS scores differed between groups at all time points (p < .001). In MDD participants MADRS decreased significantly from pre-infusion to the day-two post-infusion time point following ketamine (p < .001) but not placebo; post-ketamine MADRS was lower than post-placebo in MDD (p = .009). Healthy controls showed a small, clinically negligible reduction post-placebo (p = .029) and no significant change post-ketamine. Imaging main effects: There was no overall main effect of diagnostic group across sessions and conditions. A robust main effect of emotion was observed across groups and sessions, with greater activation to angry than to happy faces in a widespread set of regions including bilateral dorsal anterior cingulate cortex (dACC), precentral and medial frontal gyri, fusiform and middle temporal gyri, putamen, thalamus, and posterior cingulate (clusters p FWE < 0.001). A session main effect was also detected (p FWE < 0.05), with prominent differences between baseline and subsequent sessions, plausibly related to task/scanner novelty. Comparing drug sessions across groups, post-ketamine showed greater activation than post-placebo in the left middle occipital gyrus (p FWE < 0.001) and reduced activation post-ketamine versus post-placebo in left temporal and inferior frontal cortices (p FWE < 0.002). Two-way interactions: A diagnosis-by-emotion interaction reached significance in small peripheral clusters (p FWE < 0.05), with emotion effects more prominent in the MDD group. Session-by-emotion interactions were found notably in bilateral orbital cortex and ACC (p FWE < 0.05). A group-by-session interaction indicated that baseline differences (MDD < HC) in clusters including bilateral thalamus, caudate and right middle/inferior temporal gyri (p FWE < 0.001) were not present post-placebo or at interim scans. Crucially, when comparing ketamine versus placebo between groups, HCs generally showed increased activation post-ketamine relative to post-placebo, whereas MDD participants showed decreased activation post-ketamine relative to post-placebo, particularly in right frontal cortex, dACC, and left inferior occipital gyrus (p FWE < 0.05). No significant interactions involved congruency. Three-way interaction and drug-by-emotion-by-group effects: A significant session-by-emotion-by-group interaction appeared as a large medial prefrontal and anterior cingulate cluster (p FWE < 0.001). Extracted beta values indicated opposite response patterns across groups and drug sessions: in HCs under placebo this ACC region activated to angry and deactivated to happy trials, and this pattern reversed after ketamine. MDD participants showed the opposite starting pattern (deactivation to angry, activation to happy under placebo) and also reversed after ketamine. Overall, MDD participants post-ketamine resembled HCs post-placebo in this region. Association with symptom change: In MDD participants percent change in MADRS was associated with magnitude of activation; associations tended to be positive post-ketamine and negative post-placebo. Specifically for ketamine scans, an emotion-by-MADRS interaction occurred in left parahippocampal gyrus and amygdala (p FWE < 0.002) and in bilateral cingulate, precuneus, and left medial/middle frontal gyri (p FWE < 0.002): greater symptom improvement correlated with decreased activation to angry trials and increased activation to happy trials. Placebo scans showed a single emotion-by-MADRS interaction in left frontal gyrus (p FWE < 0.001). Behavioural results: Task accuracy was high (mean 97%) and similar across groups/sessions. Mean reaction time was 516.56 ms (SD = 82.06). There was a significant main effect of emotion on reaction time (F = 10.935; p = .001), with faster responses to angry than happy trials, and a significant session main effect (F = 11.335; p < .001), though the post-ketamine versus post-placebo contrast was not significant. A trend for congruency (F = 3.275; p = .071) suggested slightly faster responses on congruent trials. No group effects or interactions were observed for reaction time. Attentional bias scores (angry mean 3.39 ± 19.75; happy mean 6.18 ± 26.25) showed no main effects or interactions.

Reed and colleagues interpret their findings as evidence that ketamine alters brain responses during emotionally valenced attentional processing and may normalise aberrant neural activity in treatment-resistant MDD. The principal observation was that ketamine produced opposite effects on activation in certain regions in MDD participants versus healthy controls; specifically, several frontal, parietal and occipital regions and the dACC showed increased activation post-ketamine in HCs but decreased activation in MDD participants, and activity patterns in MDD participants after ketamine tended to resemble those of HCs after placebo. The authors emphasise a large fronto-cingulate (ACC/medial prefrontal) cluster that exhibited an emotion-by-drug-session-by-group interaction: healthy controls under placebo activated more to angry than happy faces but this reversed after ketamine, whereas MDD participants showed the opposite baseline pattern that also reversed post-ketamine. In linking these results to prior work, the researchers note consistency with earlier reports of anterior cingulate involvement in emotional and attentional abnormalities in depression and with other ketamine studies reporting decreased ACC/DLPFC activation to negative stimuli in healthy volunteers. They also reference an accompanying magnetoencephalography finding in the same sample showing normalisation of gamma power post-ketamine in MDD, suggesting a broader pattern of homeostatic change across modalities. The association between symptom improvement (MADRS reduction) and a shift toward reduced activation to angry stimuli and increased activation to happy stimuli in regions such as the parahippocampal gyrus and amygdala is interpreted as indicating that ketamine's antidepressant effects may attenuate neural responses to negative stimuli and enhance responses to positive stimuli. The authors acknowledge several limitations. Behaviourally, MDD participants did not show the expected negative attentional bias for angry faces, which may reflect stimulus specificity (for example, sad faces might produce different effects) or the greater sensitivity of fMRI relative to behavioural measures. The effective sample size for key contrasts was reduced because not all participants had usable data for every scan session due to motion, data quality, or task accuracy exclusions. Finally, the complex crossover design and multiple factors led to numerous contrasts, adding analytic complexity. Strengths highlighted include the inclusion of both MDD and HC groups studied longitudinally, the placebo-controlled crossover format, and inpatient monitoring that minimised confounds such as concurrent medications. The authors conclude that their results advance understanding of ketamine's mechanism of action by showing drug- and emotion-specific modulation of brain networks involved in attentional and emotional processing.

The authors conclude that ketamine produces distinct effects on brain activity during emotionally valenced attentional processing, with responses in MDD participants often opposite to those in healthy controls. A key finding is that activation patterns in MDD participants after ketamine resembled those of healthy controls after placebo, supporting the notion that ketamine may normalise neural function at the system or network level in regions implicated in attention and emotion, particularly a fronto-cingulate network. These results are presented as contributing to a better mechanistic understanding of ketamine's antidepressant action.