Ketamine-Associated Brain Changes: A Review of the Neuroimaging Literature

The authors performed a Medline search through December 2016 using combinations of terms including "depression and ketamine and neuroimaging," "ketamine and imaging," and related phrases. Only studies in English and adult human research were considered. The initial search yielded 966 records; after removing duplicates and non-human studies, 47 papers met the inclusion criteria and were reviewed. Ionescu and colleagues organised the extracted literature into three overarching result sections: (1) ketamine and neuroimaging in depression (clinical patient studies), (2) ketamine in non-depressed subjects using non-task resting-state scans, and (3) ketamine in non-depressed subjects using task-based scans. The review considered multiple imaging modalities; where studies used more than one technique, findings were synthesised across modalities rather than treated separately. The Methods text in the extracted file does not report formal quality assessment or meta-analytic pooling procedures, indicating a narrative review rather than a quantitative synthesis.

Overall, 47 neuroimaging studies were reviewed: 13 papers addressed unipolar depression and two addressed bipolar depression; the remainder studied healthy volunteers. Among patient studies using fMRI and connectivity metrics, several findings emerged. One study found that ketamine increased neural responses to positive emotions in the right caudate, with greater post‑ketamine caudate connectivity linked to clinical improvement. Abdallah and colleagues reported reduced global brain connectivity in the prefrontal cortex (PFC) at baseline in MDD versus controls, and ketamine increased global connectivity in the right lateral PFC while decreasing it in the left cerebellum; responders showed increased connectivity in lateral PFC, caudate and insula compared to non‑responders. Downey and colleagues observed increased BOLD signal in the subgenual anterior cingulate cortex (sgACC) after ketamine that predicted symptom change at 24 hours and 1 week, although that study had no significant antidepressant effect and showed strong placebo and baseline imbalance. Structural and diffusion imaging provided additional predictive markers: smaller left hippocampal volume at baseline correlated with greater antidepressant response at 24 hours in one study. Diffusion MRI measures (fractional anisotropy, mean diffusivity and radial diffusivity) in cingulum, forceps minor and frontostriatal tracts at baseline predicted symptom improvement 24 hours after infusion. MEG studies implicated the anterior cingulate cortex (ACC): higher baseline ACC cortical activity to fearful faces (particularly pregenual ACC, pgACC) and lower baseline amygdala activation predicted larger early antidepressant responses; lower pgACC engagement during increasing working-memory load and lower pgACC–left amygdala coherence likewise predicted response. Other MEG work showed decreased amygdala–insulo‑temporal connectivity post‑ketamine, and increased somatosensory cortical excitability (an index of synaptic plasticity) in responders. PET studies of whole‑brain metabolism reported heterogeneous correlations with clinical outcomes. NIMH investigators linked reduced anhedonia post‑ketamine with increased metabolism in the hippocampus and dorsal ACC (dACC) and decreased metabolism in orbitofrontal cortex (OFC). Other PET findings included decreased metabolism post‑ketamine in the right habenula, right insula, right ventrolateral PFC and dorsolateral PFC; clinical improvements correlated with increased metabolism in superior and middle temporal gyri and cerebellum and decreased metabolism in parahippocampal and inferior parietal regions. In bipolar depression samples (patients maintained on lithium or valproate), decreased anhedonia correlated with increased dACC and putamen metabolism; larger antidepressant gains associated with increased ventral striatum metabolism, and sgACC metabolism correlated positively with symptom improvement. In healthy volunteer resting‑state and non‑task studies (21 papers, mainly MRI and MRS), ketamine altered cerebral blood flow (CBF) and connectivity. Several studies reported reduced hippocampal CBF and increased CBF in ACC and prefrontal regions; other reports showed reduced CBF in OFC and sgACC, with one study finding a very strong correlation (r=0.90) between sgACC CBF reduction and dissociation measured by the Clinician‑Administered Dissociative States Scale (CADSS). Resting‑state fMRI showed decreased connectivity within auditory and somatosensory networks relative to pain‑processing regions, reduced pACC–dPCC connectivity correlated with psychotomimetic effects during infusion, and disruption of functional network connectivity among pgACC, medial PFC and bilateral dorsomedial PFC persisting 24 hours after infusion. Ketamine produced hippocampal hyperconnectivity in networks vulnerable to mood and cognitive disorders, and acute increases in hippocampal Glx (glutamate+glutamine) measured by MRS accompanied decreased fronto‑temporal and temporo‑parietal functional connectivity. Pharmacological imaging contrasts (ketamine minus placebo) showed increased BOLD in bilateral midcingulate, ACC, insula and right thalamus. MEG during infusion revealed increased gamma power, decreased beta activity (noted in thalamus, hippocampus and fronto‑cortical regions) and increased thalamocortical information transfer (measured by transfer entropy). Task‑based imaging in non‑depressed subjects (15 papers, 14 using fMRI) produced mixed but thematically consistent results. Emotion‑processing tasks often showed attenuated amygdalo‑hippocampal BOLD responses, particularly to negative stimuli; the magnitude of acute psychedelic effects predicted the reduction in responsiveness to negative images. In some cases pgACC activation increased 24 hours post‑infusion during negative picture viewing, especially in participants with lower distraction ability. Ketamine reduced insula activation (right insula across valences; left insula and right dorsolateral PFC for negative stimuli only) and impaired self‑monitoring associated with reduced left superior temporal cortex activation. Working‑memory and episodic memory tasks produced heterogeneous outcomes: some studies found increased fronto‑parietal and PFC activation associated with correct encoding under ketamine, while others documented impaired working‑memory performance with reduced PFC activation and connectivity related to poorer task performance. Verbal fluency was generally impaired under ketamine, with altered activation in temporal, frontal and parietal regions depending on task type.

Ionescu and colleagues interpret the literature as indicating recurring involvement of a set of regions—particularly the subgenual ACC (sgACC), posterior cingulate cortex (PCC), prefrontal cortex (PFC) and hippocampus—in ketamine’s neural effects, which overlap with networks implicated in the pathophysiology of depression. Across modalities, ketamine appears to reduce activity in regions tied to self‑monitoring, produce an emotional blunting or attenuation of limbic responsiveness to negative stimuli, and increase activation of regions associated with reward processing. Together these effects are hypothesised to shift processing away from internally focused, aversive states and rumination toward altered perceptual and reward‑related processing. The authors note inconsistencies in specific findings, using the sgACC as an example: some studies report immediate reductions in sgACC coupling and blood flow that predict dissociation, while PET studies at about 120 minutes post‑infusion have reported increased sgACC metabolism correlated with clinical improvement or no change in sgACC metabolism. Ionescu and colleagues suggest that timing of scans relative to infusion (acute dissociative window versus later post‑infusion) may account for some discrepancies and that dissociation itself could mediate aspects of antidepressant response by transiently "disconnecting" excessive aversive visceromotor states from cognition. They also highlight evidence that ketamine disrupts default mode network (DMN) hyperconnectivity and pACC–dPCC coupling—networks associated with rumination—and that responders show increased frontal connectivity (for example lateral PFC, caudate, insula) consistent with restored frontal regulation over limbic regions. Metabolic correlates of clinical improvement varied across studies, including increased metabolism in hippocampus and dACC and decreased OFC metabolism in some datasets, while other work linked improvements to temporal and cerebellar metabolic changes. Several limitations are emphasised. Most neuroimaging studies to date involve healthy volunteers rather than patients with depression, many task studies enrolled only male participants, and sample sizes are generally small (the largest depression imaging study in the review included 24 participants). Heterogeneity in dose regimens, use of racemic versus S‑ketamine, timing of imaging relative to infusion, and the rapid time course of ketamine’s antidepressant effect complicate chronological interpretation of which brain changes reflect direct pharmacology versus secondary mood improvement. These factors limit generalisability and interpretability. Finally, the authors recommend further controlled, larger‑scale imaging studies in depressed patients, focused on sgACC, PCC, PFC and hippocampus, to clarify ketamine’s mechanism and to inform new models of depression and potential drug discovery. They frame ketamine as a valuable probe for advancing neurobiological understanding because of its rapid and robust clinical effects.