The Effects of Psilocybin in Adults with Major Depressive Disorder and The General Population

The investigators performed a systematic review following PRISMA guidelines. Three independent reviewers searched PubMed, Google Scholar and PsycINFO for English-language articles from database inception to 17 November 2021 using MeSH terms and combinations around psilocybin, depression (including treatment-resistant depression) and neuroimaging modalities (fMRI, PET, diffusion tensor imaging). A parallel search sought neuroimaging studies of psilocybin in healthy volunteers; reference lists were also checked. Eligibility required clinician-confirmed MDD or validated diagnostic scales for the clinical group and use of a functional imaging technique (diffusion tensor imaging, fMRI or PET). The authors excluded unpublished datasets, case reports, conference abstracts, observational studies, duplicate reports from the same dataset, studies lacking clinical depression assessment, studies without functional imaging in the clinical sample and animal studies. Data extraction items included sample size, gender distribution, mean age, imaging technique, study design, psilocybin dose, intervention details (including psychological support), outcome measures and reported findings. Study quality (risk of bias) was assessed using the revised Cochrane RoB 2 tool, covering randomisation, deviations from intended interventions, missing outcome data, outcome measurement, selective reporting and conflicts of interest. The final selection comprised three studies that met inclusion criteria for MDD neuroimaging and ten studies in healthy participants. For the clinical studies the typical intervention protocol involved an initial low oral dose (10 mg) followed one week later by a high dose (25 mg), with preparatory psychological support, in-session support and an integration session the day after the high dose. Baseline fMRI scans were taken prior to intervention and post-treatment scans were typically acquired one day after the high-dose session. Healthy volunteer studies used a variety of designs (counterbalanced placebo-controlled, single-blind and double-blind paradigms), doses (fixed mg doses and several mg/kg regimens) and imaging schedules, including both resting-state and task-based fMRI (emotion-processing tasks, autobiographical recall, meditation paradigms) with scans timed variably relative to dosing.

Search yield and study characteristics: after deduplication and screening the authors identified three neuroimaging studies in participants with MDD and ten in healthy volunteers. The three MDD studies included a total neuroimaging sample of 55 participants (reported in the review as the available sample for neuroimaging analyses). Clinical studies used a 10 mg then 25 mg dosing schedule spaced one week apart with psychological support; imaging was obtained at baseline and primarily one day after the high dose. Amygdala activation: findings in the MDD studies were inconsistent. One study using cerebral blood flow measures reported reduced left amygdala activation one day post-treatment. A second MDD study that used a voxel-wise BOLD analysis during an emotional face task (Karolinska Directed Emotional Faces) reported increased right amygdala activation to fearful, happy and neutral faces; after correcting for multiple tests the fearful > neutral contrast remained significant (p = .001). By contrast, several studies in healthy volunteers generally showed attenuated amygdala responses to negative stimuli: one study reported reduced right amygdala activity during negative (p = 0.001) and neutral (p < 0.001) picture processing, and another reported robust reductions in both left (p < 0.00005) and right (p = 0.00005) amygdala activation one week after psilocybin that returned to baseline by one month. Functional connectivity: two MDD studies examined post-treatment connectivity. One found no significant change in amygdala resting-state functional connectivity post-treatment. The other reported increased task-based connectivity between the amygdala and visual regions (right lateral occipital cortex, intracalcarine/supracalcarine cortices, cuneus, precuneus) during processing of happy and neutral faces versus resting state (both p < 0.001). Regarding vmPFC-related effects, one MDD study (Carhart‑Harris et al. referenced in the review) observed increased vmPFC resting-state connectivity with bilateral inferior-lateral parietal cortex and decreased parahippocampus–prefrontal connectivity; another MDD study found decreased vmPFC–right amygdala connectivity during processing of fearful (p = .032) and neutral (p = .041) faces and task-related increases in vmPFC/amygdala connectivity with occipital-parietal cortices. Reports were mixed for subgenual ACC–PCC connectivity (one study showed increased coupling, another showed no change). Healthy volunteer studies showed diverse connectivity changes. Analyses reported reduced top-down amygdala→primary visual cortex connectivity to threat (p = 0.01), decreased striatum–amygdala or amygdala–frontal connectivity in some emotion conditions, and broad, time-dependent shifts in brain-wide connectivity: increased global connectivity with reduced fronto-parietal coupling in one report, and relative hypoconnectivity in associative regions with hyperconnectivity in sensory regions in another (associations with baseline connectivity and spatial 5‑HT2A/5‑HT1A gene expression were reported). Barret et al. found numerous resting-state connection strength changes at one week (38 increases, 10 decreases) and at one month (29 increases, 18 decreases). Additional findings included reduced medial temporal–high-level cortical coupling, decreased interhemispheric communication and salience network disintegration associated with ego-dissolution, and decoupling of mPFC–PCC (p = 0.003) and mPFC–angular gyrus (p = 0.030) in a meditation-related paradigm. Depressive symptom correlations: two MDD studies tested associations between neural changes and depressive symptoms. One found that reductions in QIDS‑SR16 scores one day after treatment correlated with reduced amygdala CBF (p = 0.01). The other reported that increased right amygdala activation during the KDEF task correlated with lower in-scanner depression ratings and with BDI and QIDS scores. Functional connectivity measures were also linked to clinical outcomes: increased vmPFC–ilPFC resting connectivity (p = 0.03) and decreased parahippocampus–prefrontal connectivity (p = 0.04) predicted treatment response at five weeks; vmPFC–occipital‑parietal task connectivity correlated with BDI one week post-treatment (p = 0.048); and reduced vmPFC–amygdala connectivity during neutral and fearful face processing associated with lower rumination one week post-treatment (p = 0.018).

Gill and colleagues report heterogeneous neuroimaging results across a very small clinical literature, with variable patterns of amygdala activation and functional connectivity following psilocybin-based psychotherapy in MDD. The investigators highlight that some MDD studies found reductions in amygdala activity and vmPFC–amygdala coupling—effects that mirror observations from healthy volunteer trials and that are reminiscent of the down-regulating effects of selective serotonin reuptake inhibitors on amygdala responsivity. Such changes are plausible neural correlates of antidepressant action, because amygdala hyperactivity has been linked to depressive episodes and because modulation of prefrontal–limbic circuits can relate to reductions in negative affect and rumination. At the same time, the authors emphasise important inconsistencies: one MDD study observed increased right amygdala activation after treatment, in contrast to the general pattern of amygdala attenuation described in several healthy volunteer studies. The review stresses that small sample sizes, methodological heterogeneity (dosing regimens, timing of scans, task versus resting-state paradigms), and differences between clinical and healthy cohorts limit the ability to draw firm conclusions. The investigators acknowledge practical challenges—high cost of fMRI, difficulty of adequate placebo control given the subjective effects of psychedelics, and variable imaging protocols—and suggest larger, longitudinal, placebo-controlled studies with harmonised imaging and behavioural assessments are needed. They additionally propose exploring more cost-effective imaging modalities such as functional near‑infrared spectroscopy to increase sample sizes and replication potential. Overall, the authors conclude that while preliminary findings point to prefrontal–limbic circuitry (notably vmPFC and amygdala) as a candidate substrate for psilocybin's antidepressant effects, the current evidence base is limited and heterogeneous; replication in larger, well-controlled and longitudinal studies is required to clarify neural mechanisms and their relationship to clinical outcomes.

This qualitative review identified limited and heterogeneous neuroimaging evidence for psilocybin’s effects in MDD. Studies in clinical samples reported changes in amygdala activation (one reporting reduced left amygdala activation, two reporting increased right amygdala activation), whereas healthy volunteer studies more commonly found reduced amygdala reactivity to negative stimuli. Functional connectivity changes centred on prefrontal–limbic and occipital‑parietal networks, and some neural measures correlated with clinical improvement or reductions in rumination. Given the small number of MDD neuroimaging studies (k = 3), modest sample sizes (n = 55 for MDD neuroimaging), and methodological heterogeneity, the investigators call for larger, longitudinal, and better-controlled studies to determine whether these neural changes reliably mediate psilocybin’s antidepressant effects.