Increased global integration in the brain after psilocybin therapy for depression

Depression is a common and costly disorder marked by cognitive inflexibility, negative bias and rigid self-referential thinking. Neuroimaging work has implicated abnormal functioning of higher-order intrinsic networks—including the default mode network (DMN), executive network (EN) and salience network (SN)—in these features, and the serotonin 2A (5-HT2A) receptor, the principal binding site for classic psychedelics such as psilocybin, is densely expressed across these transmodal cortical areas. Previous clinical trials have reported substantial antidepressant effects of psilocybin therapy, but the neural mechanisms that might underlie these effects remain incompletely understood. Daws and colleagues set out to test the hypothesis that psilocybin therapy produces a subacute, global decrease in functional brain modularity—interpreted as an increase in network integration or a broadening of the brain's functional state space—and that this change would relate to antidepressant outcomes. The paper analyses resting-state fMRI and clinical data from two trials: an open-label trial in treatment-resistant depression (TRD) with scans at baseline and 1 day after the higher psilocybin dose, and a double-blind Phase II randomized controlled trial (DB-RCT) comparing two high-dose psilocybin sessions plus placebo capsules versus very low-dose (1 mg) psilocybin plus escitalopram, with scans at baseline and 3 weeks after the second psilocybin dose. The investigators also examine network-specific cartography and dynamic flexibility metrics to identify which network changes accompany clinical improvement.

Two clinical studies provided data: an open-label trial in TRD and a double-blind randomized controlled trial in major depressive disorder (MDD). Eligibility in both trials required a clinician-confirmed diagnosis of unipolar MDD and a minimum Hamilton Depression Rating Scale score; the open-label study additionally required treatment resistance. The imaging samples comprised 16 patients in the open-label trial after motion exclusions, and 22 (psilocybin arm) and 21 (escitalopram arm) patients in the DB-RCT imaging analyses after attrition and motion-related exclusions. Interventions differed by trial. In the open-label study patients received two oral doses of psilocybin in fixed order: 10 mg then 25 mg one week apart, with clinical and resting-state fMRI assessment at baseline and 1 day after the 25 mg dose. In the DB-RCT, patients were randomised to receive either 25 mg psilocybin (psilocybin arm) or a presumed negligible 1 mg dose (escitalopram arm) on two dosing days separated by 3 weeks. Beginning 1 day after the first dosing day, patients also took daily capsules for 6 weeks: placebo capsules in the psilocybin arm or escitalopram (10 mg daily for 3 weeks then 20 mg daily) in the escitalopram arm. BDI-1A scores were used to measure depression severity at prespecified time points in both trials. Resting-state fMRI was acquired eyes-closed on a 3T scanner. The open-label study used a 12-channel head coil and acquired 240 volumes (TR = 2,000 ms); the DB-RCT used a 32-channel coil and acquired 480 volumes (TR = 1,250 ms). A comprehensive preprocessing pipeline was applied including despiking, slice timing and motion correction, brain extraction, registration to anatomical and MNI templates, scrubbing (framewise displacement threshold 0.5 mm), spatial smoothing (6 mm FWHM), 0.01–0.08 Hz band-pass filtering, detrending and voxelwise nuisance regression (motion parameters and tissue signals including ventricles, draining veins and local white matter). Functional connectivity (FC) matrices were computed from 100 cortical regions of interest by Pearson correlation of regional mean time series; positive correlations were Fisher-transformed to z scores. Global network modularity was estimated using a Louvain-like community detection algorithm repeated 100 times per scan, with the highest-scoring partition normalised against mean modularity from 100 randomly rewired FC matrices to allow valid between-scan comparisons. Functional cartography used community assignments to compute within-network recruitment and between-network integration measures. The DB-RCT's higher temporal resolution enabled exploratory dynamic analyses of network 'flexibility'—a measure of how often regions change community allegiance over time—using established time-windowing procedures guided by prior work.

Open-label trial: In the imaging sample of 16 TRD patients (mean age 42.8 years), Beck Depression Inventory (BDI) scores indicated severe baseline depression (mean BDI = 34.81). Psilocybin therapy produced rapid, substantial and sustained symptom reductions: at 1 week post-treatment the mean BDI change was -21.0 points (t15 = 7.11, 95% CI -27.30 to -14.71, P < 0.001, Cohen's d = 1.78) and remained significantly reduced at 6 months (mean difference -14.19, t15 = 4.26, 95% CI -21.29 to -7.09, P < 0.001, d = 1.07). Consistent with the primary hypothesis, normalised brain network modularity decreased 1 day after the 25 mg psilocybin dose (mean difference -0.29; t15 = 2.87, 95% CI 0.07 to 0.50, P = 0.012, d = 0.72), implying increased global integration across intrinsic networks. Post-treatment modularity correlated with long-term clinical outcome: after false discovery rate correction a strong correlation was observed between post-treatment modularity and BDI at 6 months (r14 = 0.64, 95% CI 0.29 to 0.84, P = 0.023). Change in modularity from baseline to post-treatment also correlated with BDI change at 6 months (r14 = 0.54, 95% CI 0.14 to 0.78, P = 0.033). Functional cartography revealed a reduction in within-DMN recruitment (mean difference -0.54; t15 = -2.99, 95% CI -0.92 to -0.15, P = 0.009, d = 0.75) alongside increased between-network integration of the DMN with the EN (mean difference 0.53; t15 = 3.01, 95% CI 0.15 to 0.90, P = 0.01, d = 0.75) and with the SN (mean difference 0.55; t15 = 2.89, 95% CI 0.14 to 0.95, P = 0.01, d = 0.72). DB-RCT: The imaging samples comprised 22 patients in the psilocybin arm (mean age 44.5 years) and 21 in the escitalopram arm (mean age 40.9 years). Baseline BDI was higher in the open-label TRD sample than in the DB-RCT MDD sample. Clinical outcomes previously reported by the authors showed greater BDI reductions after psilocybin than escitalopram. In the DB-RCT imaging analyses, normalised modularity was significantly reduced 3 weeks after the second psilocybin dose in the psilocybin arm (mean difference -0.39; t21 = -2.20, 95% CI -0.75 to -0.02, P = 0.039, d = 0.47). For the psilocybin condition, decreases in modularity correlated with symptom improvement at the primary end point (r20 = 0.42, P = 0.025, one-tailed). By contrast, the escitalopram arm showed no change in modularity (mean difference 0.01; t20 = 0.07, 95% CI -0.35 to 0.33, P = 0.95, d = 0.02) and no correlation between modularity changes and BDI changes (r19 = 0.08; P = 0.361, one-tailed). Dynamic analyses made possible by the DB-RCT's faster fMRI acquisition revealed that increases in dynamic flexibility of higher-order networks related to clinical benefit in the psilocybin arm. After FDR correction, increased EN dynamic flexibility correlated strongly with greater symptom improvement at the 6-week primary end point (r20 = -0.76, 95% CI -0.90 to -0.50, P = 0.001). Similar strong correlations were observed when combining EN regions with lateral frontoparietal networks such as the SN and dorsal attention network. No analogous correlations were observed in the escitalopram arm. The network-specific cartography changes seen 1 day after treatment in the open-label trial were not replicated at the 3-week DB-RCT scan; the authors note this as a potential timing or severity-dependent effect. Strict head-motion criteria and supplementary analyses indicated that motion was unlikely to explain the observed modularity effects.

Across two clinical trials, the investigators report convergent evidence that psilocybin therapy is associated with a subacute decrease in brain network modularity—interpreted as increased global integration—and that this change correlates with clinical improvement in depressive symptoms. The effect was observed 1 day after treatment in the open-label TRD sample and at the 3-week primary end point in the DB-RCT psilocybin arm, whereas no modularity changes were observed in patients treated with escitalopram. The authors situate these findings within prior work on the acute effects of psychedelics, which has shown increased inter-regional and between-network functional connectivity, and suggest that the post-treatment findings may represent a temporally attenuated ‘carryover’ of acute-state dynamics. Daws and colleagues propose a mechanistic model in which psilocybin, via agonism at cortical 5-HT2A receptors, dysregulates entrenched functional modules and broadens the brain's functional repertoire—a putative ‘flattening’ of the functional energy landscape that could facilitate cognitive and emotional flexibility. The observed increases in flexibility and changes involving higher-order networks (DMN, EN, SN) are consistent with this model and with the prominence of 5-HT2A expression in these regions. By contrast, escitalopram's broader serotonergic effects and predominant 5-HT1A-mediated action in limbic circuitry may not produce the same modularity changes. The authors acknowledge several limitations and uncertainties. They did not formally measure cognitive flexibility within the trials, making it difficult to tie imaging changes directly to specific cognitive processes. Psychological and contextual factors associated with psychedelic therapy might contribute to clinical outcomes, so causal attribution to the drug alone cannot be made from these data. Differences in trial design and imaging protocols—timing of post-treatment scans, head coils and temporal resolution—complicate direct comparisons; these differences may explain why finer-grained cartography findings from the open-label trial were not fully replicated in the DB-RCT. Potential confounds such as head motion and in-scanner sleep were examined and found unlikely to account for results, but the authors emphasise the need for larger, repeated-scanning studies to disentangle effects of baseline severity, time since treatment and methodological factors. Finally, they call for Phase III and pragmatic trials to establish generalisability, reliability and specificity of psilocybin's antidepressant response, and recommend network modularity analyses as a useful biomarker approach for future imaging studies.