Greater global integration in the brain after psilocybin therapy for depression.

Depression is a highly prevalent mental health condition. ¹ , whose incidence increased during the COVID-19 pandemic. ² , for example, as reflected in the increase in prescriptions for antidepressant medications ³ . However, even the best-performing antidepressants have modest efficacy, significant side effects, discontinuation problems, and high relapse rates. ^{4 , 5 , 6 , 7} highlighting the need for new and improved treatments. ⁸ .

Patients diagnosed with depression often exhibit negative cognitive bias, characterized by pessimism, low cognitive flexibility, rigid thought patterns, and negative fixations regarding the self and the future. ^9 , 10 . Several authors have drawn directly or indirectly from dynamical systems theory to describe depressive episodes as 'attractor states' (stereotyped cognitive states with 'gravitational attraction'). ¹¹ ).

Neuroimaging research has consistently found examples of abnormal brain function in depression, resonant with its phenomenology. ^{12 , 13 , 14} . An intrinsically hierarchically superior brain network ¹⁵ The default mode network (DMN) is associated with introspection and self-referential thinking. ¹⁶ . These cognitive functions are often overactive in depression. ⁹ and several studies have linked excessive involvement of the DMN (Depressive Neuromuscular System) with depressive symptomatology. ¹² .

In addition to the DMN, other higher-order brain networks, such as the executive network (EN) and the salience network (SN), have been implicated in depression. ^14 , 17 . These networks are associated with 'cognitive control' and the switching between internal and external attention. ^{18 , 19 , 20} . This shift in attention is often impaired in depression. ²¹ . Notably, the serotonin 2A receptor subtype (5-HT2A), which is the main proteomic binding site of 'classic' serotonergic psychedelic drugs such as psilocybin 22, is more densely expressed. ⁱⁿ a broad cortical pattern resembling a map of the intersection of DMN, EN, and SN ²³ , corresponding to the transmodal portion of the brain's main hierarchical gradient. ¹⁵ .

Over the past 15 years, at least six separate clinical trials have reported impressive improvements in depressive symptoms with psilocybin therapy. ²⁴ . Included among these studies are (1) an open-label study in treatment-resistant depression ²⁵ and (2) a double-blind randomized controlled trial (DB-RCT) with an active comparator, the selective serotonin reuptake inhibitor (SSRI) and conventional antidepressant, escitalopram ²⁶ . These two trials, which included pre-treatment and post-treatment fMRI, are the focus of the analyses in this article.

The therapeutic action of psilocybin and related psychedelics is not fully understood; However, one model proposes that psychedelics cause a 5-HT2A receptor-induced dysregulation of spontaneous neuronal activity at the population level, linked to a temporary "disintegration" of intrinsic functional brain networks 27 and a hypothetical decrease in the precision weighting of models ^predictive encoded (at least in part) by the integrity of the functional modules. ²⁸ . An important corollary of modular "disintegration" appears to be the expansion of the functional repertoire of brain states, compatible with a broader or flatter global energy landscape. ²⁹ .

Here, we hypothesize that the well-replicated finding of disintegration and desegregation of the brain network under psychedelics ^30 , 31 It will be apparent in a subacute form, in post-treatment resting fMRI data. We also hypothesized that this effect, consistent with a flatter energy landscape, would be related to better depression outcomes and would not be observed after a course of SSRI, escitalopram.

Open test

Rapid antidepressant effect of psilocybin therapy

Patients with treatment-resistant depression (TRD) participated in a single-arm, open-label clinical trial of psilocybin therapy (Fig. 1a ). Initial clinical assessment and resting-state fMRI were followed by fixed-order dosing days (DDs) of 'low' (10 mg) and 'high' (25 mg) psilocybin therapy that were separated by 1 week. A second clinical evaluation and fMRI were performed 1 day after DD2. Remote assessments of clinical status were performed 1 week, 3 months, and 6 months after DD2. More details are available at Methods and in other places ²⁵ .

Of the 19 patients recruited, 3 were excluded due to excessive head movement on fMRI (Fig. 2a ). We first confirmed an antidepressant effect of psilocybin in this imaging sample of 16 patients (mean age, 42.75 years, SD = 10.15, 4 women) using the Beck Depression Inventory (BDI-1A). This patient-assessed measure was pre-registered for the original investigation (gtr.ukri.org). MR/J00460X/1 ). The BDI captures a wide range of symptoms and places particular emphasis on the cognitive features of depression. ³² , which may be an important target for psilocybin therapy.

Initial BDI scores indicated severe depression (mean BDI = 34.81, SD = 7.38). According to our previous report ²⁵ Rapid, substantial, and sustained reductions in the severity of depression were observed after treatment (Fig. 3a,b ). In relation to the baseline, significant reductions in BDI were observed in 1 week (mean difference, −21.0 points; t ₁₅ = 7.11, 95% confidence interval (CI) −27.30 to −14.71, P < 0.001, d Cohen = 1.78) and still evident at 6 months (mean difference, −14.19 points; t ₁₅ = 4.26, 95% CI −21.29 to −7.09, P < 0.001 d = 1.07).

Decreased brain modularity one day after psilocybin therapy.

To test our primary hypothesis, pre-processed fMRI data were used to estimate normalized network modularity from Pearson correlation functional connectivity matrices of the cortex ( Methods ). Confirming our primary hypothesis, the modularity of the brain network was significantly reduced (Fig. 4a ) 1 day after psilocybin therapy (mean difference, -0.29; t ₁₅ = 2.87, 95% CI 0.07 to 0.50, P = 0.012 d = 0.72). This result implies an overall increase in functional connectivity between the brain's main intrinsic networks.

Decreased modularity suggests better clinical outcomes.

Our hypothesis is that the decrease in brain network modularity is related to the sustained improvements in depression severity that follow psilocybin therapy. To test this, we calculated Pearson correlations between post-treatment brain modularity and BDI scores at three post-treatment time points (1 week, 3 months, 6 months). After correcting for the false discovery rate (FDR) for multiple comparisons, a strong significant Pearson correlation was observed at the primary endpoint of 6 months ( r ₁₄ = 0.64, 95% CI 0.29 to 0.84, P = 0.023; Fig. 4b ). Directionally consistent relationships were observed at 3 months ( r ₁₄ = 0.46, 95% CI 0.03 to 0.74, P = 0.114) and 1 week ( r ₁₄ = 0.29, 95% CI −0.16 to 0.64, P = 0.284), but these did not survive the correction. Pre-treatment versus post-treatment changes in modularity correlated significantly with the change in BDI score at 6 months, relative to baseline ( r ₁₄ = 0.54, 95% CI 0.14 to 0.78, P = 0.033; Fig. 4c ). These results imply that the decrease in brain modularity 1d after psilocybin therapy is related to long-term improvements in the severity of depression symptoms.

Changes in DMN functional connectivity within and between networks one day after processing.

Previous research has linked depressive symptomatology with DMN hyperconnectivity. ¹² and hypoconnectivity of the DMN with other higher-order 'cognitive' networks, including EN and SN ^14 , 17 . Therefore, we tested the evidence of these abnormalities being attenuated after treatment using functional mapping ( Methods ). Consistent with our previous hypothesis, significant reductions in DMN network recruitment (mean difference, −0.54; t ₁₅ = −2.99, 95% CI −0.92 to −0.15, P = 0.009 d = 0.75; Fig. 4d ) and increased network integration between DMN and EN (average difference, 0.53; t ₁₅ = 3.01, 95% CI 0.15 to 0.90, P = 0.01 d = 0.75) and between DMN and SN (mean difference, 0.55; t ₁₅ = 2.89, 95% CI 0.14 to 0.95, P = 0.01 d = 0.72; FDR-corrected) were observed 1 day after psilocybin therapy. An exploratory analysis of changes in network recruitment and integration between networks of other brain networks is available in Fig. 2 Supplement.

Taken together, these findings indicate a clinically relevant decrease in brain network modularity after psilocybin therapy for DRD. An analysis of the network mapping suggests that this global shift in network organization may be underpinned by a specific decrease in connectivity within the DMN and an increase in the DMN's connectivity with other higher-order networks, including the EN and SN.

Double-blind randomized controlled trial

Psilocybin therapy versus escitalopram for depression

The design of this DB-RCT (Fig. 1b This provided a new opportunity to compare not only the safety and efficacy, but also the mechanisms of action of psilocybin therapy to those of a conventional antidepressant drug, escitalopram. Patients with major depressive disorder (MDD) were randomly allocated to either a 'psilocybin arm' or an 'escitalopram arm' (Fig. 1b ). Initial clinical assessment and resting-state fMRI were followed by DD1, when patients received either 25 mg of psilocybin (psilocybin arm) or a presumed inactive dose of 1 mg of psilocybin (escitalopram arm). All patients were informed that they would receive psilocybin, but were unaware of the dosage. DD2 occurred 3 weeks after DD1 and was a double dose. Starting 1 day after DD1, patients took daily capsules for 6 weeks and 1 day in total. For both conditions, one capsule per day was taken for the first 3 weeks, and two capsules per day were taken thereafter. The capsule contents were either an inert placebo (microcrystalline cellulose in the psilocybin arm) or 10 mg of escitalopram in the escitalopram arm (10 mg daily for the first 3 weeks and 2 × 10 mg (20 mg) daily for the last 22 days). More details are available at Methods and elsewhere ²⁶ .

Of the 59 patients with MDD recruited, 29 were randomly allocated to the escitalopram arm. Of these, four discontinued due to adverse reactions to escitalopram, one was lost due to the COVID-19 lockdown in the UK, and three were excluded due to excessive fMRI head movement (Fig. 2b). ) . The remaining 21 patients (mean age, 40.9 years, SD = 10.1, 6 women) were included in the escitalopram imaging sample. Thirty patients were randomly assigned to the psilocybin arm. Of these, one was excluded for choosing not to take the daily capsules (placebo), two did not attend the post-treatment session due to the COVID-19 lockdown in the UK, and five were excluded due to excessive fMRI head movement. The remaining 22 patients (mean age, 44.5 years, sd = 11.0, 8 women) were included in the psilocybin imaging sample (Fig. 2b ).

BDI was a primary outcome measure for the open-label trial ( MR/J00440/1 ) and a secondary outcome measure for this DB-RCT (ClinicalTrials.gov identifier) NCT03429075 ); However, this measure proved to be a particularly sensitive index of post-psilocybin reductions in symptom severity across all trials. ²⁶ . Among the tests, the baseline BDI (Fig. 3c ) was significantly higher in the open-label TRD trial compared with the DB-RCT MDD trial (mean difference, 6.53 points; t ₅₇ = 3.01, 95% CI 2.18 to 10.88, P = 0.013 d = 0.83). This difference is probably due to the TRD being an inclusion criterion in the open-label trial, but not in this DB-RCT.

As described in our previous report ²⁶ The reductions in depressive symptom severity measured by the BDI were significantly greater after psilocybin than escitalopram, indicating superior efficacy of psilocybin therapy versus escitalopram (Fig. 3d ). Furthermore, we confirmed the statistical significance of this contrast within the smaller neuroimaging sample included in the present analyses after testing an arm × time point analysis of the variance interaction in the BDI scores ( F _3,123 , 4.47; P = 0.005). Paired comparisons corrected for FDR from baseline were significantly different at 2 weeks (mean difference, −8.73; t ₄₁ = −3.66, 95% CI −13.55 to −3.91, P = 0.002 d = 0.98), 4 weeks (mean difference, −7.79; t ₄₁ = −2.69, 95% CI −13.62 to −1.95, P = 0.013 d = 0.77) and at 6 weeks (mean difference, −8.78; t ₄₁ = −2.61, 95% CI = −15.58 to −1.97, P = 0.013 d = 0.75), all favoring the psilocybin arm.

Brain modulation was decreased for psilocybin, but not for escitalopram.

Reconfirming our primary hypothesis (Fig. 5a,b ) and replicating the analyses on the open-label study data, brain network modularity was significantly reduced at the study's primary endpoint, 3 weeks after psilocybin therapy (mean difference, -0.39; t ₂₁ = −2.20, 95% CI −0.75 to −0.02, P = 0.039 d = 0.47). Furthermore, for the psilocybin condition, post-treatment reductions in brain network modularity correlated significantly with improvements in the severity of depression symptoms at this primary endpoint ( r ₂₀ = 0.42 P = 0.025, one-tailed).

Notably, there was no significant interaction between the treatment arm and the scanning session in the network modularity ( F _1.41 = 2,719, P = 0.107); However, there was evidence that the reduction in network modularity and its relationship with depression severity may be specific to the psilocybin arm. That is, in the escitalopram group (Fig. 5d,e ), the modularity of the network has not changed since the beginning (average difference, 0.01; t ₂₀ = 0.07, 95% CI −0.35 to 0.33, P = 0.95 d = 0.02) and there was no significant correlation between changes in modularity and changes in BDI scores ( r ₁₉ = 0.08; P = 0.361, one-tailed).

The response to psilocybin correlates with network flexibility.

The specific changes in network recruitment observed 1 day after psilocybin therapy in the open-label trial were not replicated at 3 weeks in this DB-RCT. Additional Information ). However, the faster fMRI scanning protocol adopted in the DB-RCT generated twice the amount of temporal data per scan session ( Methods ). This provided a rare opportunity to examine changes in the dynamic flexibility of brain networks after psilocybin therapy.

The metric known as 'dynamic flexibility' indicates how frequently brain regions change their loyalty to a community over time, during the course of a functional magnetic resonance imaging (fMRI) scan. ^33 , 34 ( Methods ). Reduced functional capacity has previously been associated with depressive symptoms. ¹⁴ . In an exploratory analysis, post-psilocybin changes in network flexibility were correlated with changes in BDI scores (Fig. 5c ). After correction of the FDR, increased dynamic flexibility of the EN correlated strongly with greater symptom improvement at the primary 6-week endpoint for the psilocybin arm ( r ₂₀ = −0.76, 95% CI −0.90 to −0.50, P = 0.001). Strong correlations that survived FDR correction were also observed when combining EN regions with other lateral frontoparietal networks, such as the SN and the dorsal attention network (Fig. 5c ). Critically, there were no significant correlations between changes in BDI scores and changes in dynamic flexibility in the escitalopram arm (Fig. 5f ).

In light of the growing evidence of the antidepressant efficacy of psilocybin therapy ²⁶ These findings advance our understanding of their possible underlying brain mechanisms. In two trials, a decrease in brain modularity was observed and correlated with improvements in depressive symptoms. Furthermore, this antidepressant effect may be specific to psilocybin therapy, as no changes in modulation were observed with the conventional SSRI antidepressant, escitalopram.

Research on the acute brain action of psychedelics has revealed well-replicated changes in global brain function that are somewhat consistent with those observed here (an increased repertoire of inter-regional and inter-network functional connectivity (FC) 29 ^{, 30 , 31 )} . Our previous analysis suggested some contrasting changes in the architecture of spontaneous brain function 1 day after psilocybin treatment for depression compared to what was observed during the acute psychedelic state itself: spatially expanded DMN FC (1 day after treatment for DRUGS) versus acute "disintegration" of the DMN. ²⁵ . However, others have reported evidence of increased inter-network CF 1 week and 1 month after treatment with psilocybin. ³⁵ , as well as 1 day after ayahuasca, including increases in DMN-SN heart rate in healthy volunteers. ³⁶ . These findings are consistent with the present study, but here we show robust and reliable evidence that increased global brain network integration accompanies the antidepressant efficacy of psilocybin therapy.

Current modularity metrics may be more sensitive indicators of psilocybin's antidepressant action than previously applied intra-network and inter-network mean time FC analyses. ²⁵ . Indeed, they may be relevant to other CF metrics applied to acute-state psychedelic fMRI data. ^29 , 30 where an enhanced overall picture of global FC and an expanded dynamic state space emerged ²⁸ . In this context, the results could be understood as a 'transition' effect similar to the brain dynamics associated with the acute action of psychedelics, although at an attenuated level and in a specific population (depressed patients). To demonstrate the robustness of the analytical method, we also performed more traditional univariate mass analyses, and these produced consistent results (Complementary Figs). 3 and 4 ). However, one advantage of the network's modularity is its ability to elegantly summarize global changes in the organization of the brain's functional network. ³⁷ .

Previous research on resting-state activity in depression found high network modularity correlated with symptom severity. ^17 , 38 . Further studies suggest that increased intra-DMN cardiac function and elevated heart rate between limbic regions, such as the amygdala, and high-level cortical regions correlate with ruminative symptoms in depression. ^12 , 39 . Taken together, this gives rise to a model of abnormally modulated spontaneous brain function in depression that is effectively remedied by psilocybin therapy. According to several findings, the FC energy landscape or state space in depression can be described as abnormally restricted, paralleling the narrow, internally focused, ruminative quality of mood and cognition in the disorder. ¹¹ . In contrast, psilocybin appears to increase the brain's ability to visit a broader state space, both acutely and after psilocybin therapy in depressed patients, as shown here. Furthermore, this 'liberating' action of psilocybin is accompanied by subjective reports of 'emotional release'. ^40 , 41 as well as subacute increases in behavioral optimism ⁹ cognitive flexibility ⁴² and psychological flexibility after taking a psychedelic drug ⁴³ . In fact, increased emotional responsiveness may be specific to psilocybin therapy versus SSRIs. ²⁶ .

It is plausible that this supposed liberating effect of psilocybin on cortical activity occurs through its direct agonist action on cortical 5-HT2A receptors, deregulating activity in regions rich in its expression. We hypothesize that chronic escitalopram does not have the same effect on brain modulation due to its more generalized action on the serotonergic system and predominant action on inhibitory postsynaptic 5-HT1A receptors, which are richly expressed in the limbic circuit. ^27 , 44 .

In addition to the overall decrease in network modularity after psilocybin, we observed functional changes in DMN, EN, and SN dynamics that are consistent with neurobiological models of depression. ⁴⁵ . These high-order transmodal networks harbor the highest density of 5-HT2A receptors, the primary site of action for serotonergic psychedelics. ^22 , 23 . Higher-order networks are implicated in the acute action of psychedelics, where they exhibit reduced modularity and increased communication with regions normally outside the boundaries of their communities. ^{29 , 30 , 31} .

EN and SN have been associated with tasks that require cognitive flexibility, such as learning and task switching. ^{18 , 19 , 46 , 47} ; The impaired functioning of these networks was reported during the depression. ^14 , 17 and in other disorders that exhibit cognitive inflexibility, such as autism spectrum disorder ⁴⁸ and obsessive-compulsive disorder ⁴⁹ . Our results suggest that decreased modularity or increased flexibility of these networks after psilocybin therapy is a key component of its therapeutic mechanism of action. We did not formally assess cognitive flexibility in the clinical trials reported here, but we observed improvements in overall cognitive functioning after psilocybin treatment in the DB-RCT, as well as treatment-specific improvements in 'emotional avoidance' (an inversion of the related construct 'psychological flexibility'). ²⁶ ).

It should be noted that psychological processes that do not reliably correlate with changes in brain modularity may have played a role in the main clinical outcomes of this study, and the inability to rule out such factors precludes making reliable inferences that the drug alone was the primary causal determinant of the imaging results or, indeed, that decreased modularity is sufficient for response to psilocybin therapy. However, the changes observed in the neuroimaging data were consistent with previous brain imaging research on the acute action of psychedelics and are plausible in light of evidence of high modularity and abnormal functioning of higher-order networks in depression. ^{12 , 17 , 38 , 39 , 50} .

Successful Phase III DB-RCTs will be necessary to obtain licensing for psilocybin therapy, but pragmatic trials may better address issues related to feasibility, specificity, and treatment optimization. ⁵¹ . Given the emerging research on psychedelic therapy, it is important that large-scale trials establish the generalizability, reliability, and specificity of the antidepressant response of psilocybin. For example, effectiveness is likely to depend on symptom severity, depression subtypes and comorbidities, as well as other important pharmacological and extrapharmacological factors. ⁵² . For brain imaging studies, we recommend network modularity analyses, such as those employed here. fMRI datasets are complex, expensive, and susceptible to noise, contributing to the challenge of detecting reliable biomarkers. Simplifying composite measures, such as network modularity, combined with a research domain, symptom-based approach to psychological phenomena, could be a particularly productive path to follow. ^37 , 51 .

It should be noted that the findings of the present study do not support baseline modularity as a predictor of response to psilocybin therapy. Patients with a range of baseline modularity values ​​showed reductions in modularity after treatment with psilocybin; However, the present results suggest that the change in modularity from the initial phase is predictive of the long-term treatment response to psilocybin therapy.

It is important to consider the potentially confounding effects of head movement when interpreting fMRI data. Here, a robust fMRI preprocessing pipeline was employed along with strict head motion criteria for patient inclusion. To examine robustness to movement, an analysis of head movement was performed and is available in the Additional Information . These analyses reinforce the current findings, as there is no evidence that head movement differs between treatment sessions or arms, or that it is correlated with network modularity. The fMRI data were collected using a closed-eye protocol, which has some advantages; However, it would be interesting if these findings were replicated in data acquired using an open-eyes protocol. Sleepiness in the scanner may be more likely during fMRI with eyes closed, which is a particular disadvantage ⁵³ . Sleep mode on the scanner cannot be ruled out here. However, analyses of head movement make it unlikely that sleep is a confounding factor. In addition, a self-reported visual analog scale of time spent sleeping and 'sleepiness' was acquired immediately after all scans in study 2. An analysis of these data is available in  Additional Information . Critically, sleep or drowsiness scores on the scanner were low and did not differ between treatment arms or the scanning session.

The primary hypothesis of this study was confirmed and replicated despite substantial differences between the design of the two trials. The severity of baseline depression was significantly greater in the open-label trial. Furthermore, the open-label fMRI scan was performed just 1 day after DD2 and was recorded with a 12-channel head coil and a repetition time (TR) of 2 seconds. In contrast, post-treatment DB-RCT scanning was performed 3 weeks after DD2 and was recorded with a 32-channel head coil and a TR of 1.25 s.

Recognizing these differences between study designs serves to strengthen the validity of the main findings, as they were replicated robustly; However, the lack of replication in more refined cartographic analyses limits specific network inferences. A supplementary analysis ( Additional Information ) confirmed that the severity of baseline depression correlated with connectivity within the DMN and between the DMN-EN and DMN-SN networks, as previously reported. ^{12 , 17 , 38 , 39 , 50} ; However, in DB-RCT, we did not replicate all the changes in network mapping that were observed in the open-label trial. Given that the observed network effects directly follow predictions from the literature on depression, it is possible that they are more pronounced in more severe cases, such as those included in the open-label TRD study. Alternatively, it may be that these effects are detectable only in a short-term subacute phase after psilocybin therapy and not, for example, 3 weeks later. Analyzing the relative contribution of baseline severity and time since treatment will be an important feature of future neuroimaging clinical trials, ideally with larger samples and repeated scanning sessions.

Dynamic analyses can be difficult to conduct. To have sufficient energy, the time series need to be long enough to be divided into several sufficiently long time windows to compute reliable heart rate measurements, and previous research guided our parameter selection. ^54 , 55 . Collecting sufficient fMRI data from patient cohorts can be challenging, but due to the appeal of dynamic analyses, efforts are being made to facilitate and improve them. ⁵⁴ . It should be noted that a sufficiently wide time window was used to estimate the Pearson FC correlation; However, ongoing work is needed to better understand how to capture the most functionally meaningful dynamic data.

With these caveats in mind, it should be emphasized that the inferences from both mapping analyses converged on the higher-order networks of the brain. In particular, DB-RCT analyses showed robust correlations ( r (~0.8) between increased flexibility of the higher-order network and response to psilocybin treatment, and this converges with the open-label trial as well as previous research. ¹⁴ .

In summary, depression is a major public health problem associated with enormous burdens and costs. Here, we identified a robust, reliable, and potentially specific biomarker of response to psilocybin therapy for depression. Our results may help explain why psilocybin therapy is promising as a new treatment option in psychiatry.

Test overview

The drawings of the essays (Fig. 1 ) and the main clinical results from open-label trials ³¹ (gtr.ukri.org: MR/J00460X/1) and DB-RCT ³² (clinicaltrials.gov: NCT03429075 ) were published previously. Both trials were conducted at the National Institute for Health Research Imperial Clinical Research Facility and received sponsorship from Imperial College London, NHS research, and ethical approval from the Imperial College Joint Research and Compliance Office, the Health Research Authority, and the Medicines and Healthcare Products Regulatory Agency. This work was carried out under a Schedule 1 medicinal product license from the UK Home Office. All participants provided written informed consent. The participants were not financially compensated.

participants

For both trials, eligibility required a diagnosis confirmed by a general practitioner of unipolar MDD (16+ on the 21-item Hamilton Depression Rating Scale). The open-label study had the additional criteria for TRD, as defined by no improvement despite multiple courses of antidepressant medication (mean = 4.6 ± 2.6 prior medications; range, 2-11). ³⁹ . Patients were asked if they had any prior experience using psychedelics. In the open trial, 25% had prior experience. Similarly, in the DB-RCT, 31% of patients in the psilocybin arm and 24% in the escitalopram arm had prior experience.

The exclusion criteria were immediate family or personal history of psychosis, high-risk physical health condition (assessed by a physician), history of serious suicide attempts, positive pregnancy test, and contraindications for magnetic resonance imaging. The DB-RCT had additional exclusion criteria including contraindications to SSRIs or prior use of escitalopram. Notably, treatment resistance was not an inclusion or exclusion criterion in the DB-RCT. All eligible patients underwent telephone screening interviews, provided written informed consent, and had their physical and mental medical histories thoroughly evaluated.

Interventions

Nineteen patients with TRD were recruited for the open-label trial and participated in an initial 1-day pretreatment session that included resting-state fMRI with eyes closed and clinical assessment (Fig. 1a). ) . This was followed by two DDs of psilocybin therapy, separated by 1 week. A low dose of psilocybin (10 mg) was ingested orally on DD1 and followed by a high dose (25 mg) on ​​DD2. Follow-up fMRI and clinical evaluation were performed 1 day after day 2 (DD2). Patients attended an on-site clinical assessment 1 week after DD2 and completed the clinical assessment electronically at 3 and 6 months. Of the 19 patients with TRD, 16 were retained (mean age, 42.75 years; SD = 10.15, 4 women) for the present analysis after 3 were excluded due to excessive fMRI head movement (Fig. 2a). ) .

Of the 59 MDD patients recruited for the DB-RCT, a random number generator allocated 30 to the psilocybin arm and 29 to the escitalopram arm (Fig. 1b ). The final image samples for this investigation were n = 22 for the psilocybin arm (mean age, 44.5 years, SD = 11.0, 8 women) and n = 21 for the escitalopram arm (mean age, 40.9 years, SD = 10.1, 6 women) (Fig. 2b ). Patients underwent pre-treatment resting-state fMRI with their eyes closed. DD1 consisted of 25 mg of psilocybin (psilocybin arm) or a presumed negligible dose of 1 mg of psilocybin (escitalopram arm). All patients were informed that they would receive psilocybin, but were unaware of the dosage. DD2 occurred 3 weeks after DD1 and was a double dose. There was no overlap in dosage. Starting 1 day after DD1, patients took daily capsules for 6 weeks and 1 day in total. For both conditions, one capsule per day was taken for the first 3 weeks and two thereafter. The capsule contents were either an inert placebo (microcrystalline cellulose in the psilocybin arm) or escitalopram in the escitalopram arm, 10 mg for the first 3 weeks and 2 × 10 mg (20 mg total) thereafter.

Measuring the severity of depression

BDI-1A scores were used to assess the severity of depression in both studies. This patient-assessed measure captures a broader range of symptoms, with an additional focus on the cognitive characteristics of depression, compared to other measures such as the QIDS-SR-16 (ref. 40). ⁾ . The BDI was pre-registered as a primary outcome measure in the open-label trial (gtr.ukri.org). MR/J00460X/1 ) and was measured at baseline and 1 week, 3 months, and 6 months after DD2. For the DB-RCT, BDI was measured at baseline and 2, 4, and 6 weeks after DD1. The BDI was a secondary outcome measure for this DB-RCT (ClinicalTrials.gov). NCT03429075 ) and was used here to test the replication of the effects observed in the open-label study.

Magnetic resonance imaging acquisition

The brain imaging was performed on a 3T Siemens Tim Trio at Invicro. The anatomical images were acquired using the Alzheimer's Disease Neuroimaging Initiative, Grand Opportunity (ADNI-GO). ⁵⁶ Recommended MPRAGE parameters (1 mm isotropic voxels; TR, 2300 ms; TE, 2.98 ms; 160 sagittal slices; 256 × 256 in the flat field of view; rotation angle, 9 degrees; bandwidth, 240 Hz per pixel; GRAPPA acceleration, 2).

In both studies, resting-state fMRI data with eyes closed were collected using T2*-weighted echoplanar images with 3 mm isotropic voxels. In study 1, a 12-channel head coil was used to acquire 240 volumes in approximately 8 min: TR, 2,000 ms; TE, 31 ms; 36 axial cuts; angle of inclination, 80 degrees; bandwidth, 2,298 Hz per pixel; and GRAPPA acceleration, 2). In study 2, a 32-channel head coil was used to acquire 480 volumes in ~10 min: TR, 1,250 ms; TE, 30 ms; 44 axial cuts; angle of inclination, 70 degrees; bandwidth, 2,232 Hz per pixel; and GRAPPA acceleration, 2).

fMRI data preprocessing

The image data was pre-processed using a custom internal pipeline comprised of tools from the FMRIB Software Library packages. ⁵⁷ , Analysis of Functional NeuroImages (AFNI) ⁵⁸ Freesurfer ⁵⁹ and Advanced Normalization Tools ⁶⁰ . Patients were excluded if any fMRI scan contained > 20% of volumes with a framewise offset > 0.5 mm.

Specifically, the following preprocessing steps were performed: (1) removal of the first three volumes; (2) peak removal (3dDespike, AFNI); (3) correction of the cutting time (3dTshift, AFNI); (4) motion correction (3dvolreg, AFNI) recording each volume in the volume most similar, in the least squares sense, to all the others; (5) brain extraction (BET, FSL); (6) Rigid body registration for anatomical scans (BBR, FSL); (7) Nonlinear recording for the brain from the Montreal Neurological Institute (MNI) of 2 mm (Symmetric Normalization, Advanced Normalization Tools); (8) debugging, using a framewise offset limit of 0.5 mm, the debugged volumes were replaced by the average of the neighboring volumes; (9) total width of 6 mm at half maximum Gaussian spatial smoothing (3dBlurInMask, AFNI); (10) 0.01 to 0.08 Hz bandpass filtering (3dFourier, AFNI); (11) linear and quadratic de-trend (3dDetrend, AFNI) and (12) voxelwise annoyance regression with the six realignment motion regressors and three tissue signal regressors (Ventricles, Freesurfer, eroded in 2 mm space), drainage veins (FSL's CSF minus Freesurfer's Ventricles, eroded in 1 mm space) and local white matter (WM) (FSL's WM minus Freesurfer's subcortical gray matter structures, eroded in 2 mm space). Regarding the local regression of WM, AFNI's 3dLocalstat was used to calculate the average local WM time series for each voxel, using a 25 mm radius sphere centered on each voxel. eroded within a 1 mm space) and local white matter (WM) (WM of the FSL minus subcortical gray matter structures of the Freesurfer, eroded within a 2 mm space). Regarding the local regression of WM, AFNI's 3dLocalstat was used to calculate the average local WM time series for each voxel, using a 25 mm radius sphere centered on each voxel. eroded within a 1 mm space) and local white matter (WM) (WM of the FSL minus subcortical gray matter structures of the Freesurfer, eroded within a 2 mm space). Regarding the local regression of WM, AFNI's 3dLocalstat was used to calculate the average local WM time series for each voxel, using a 25 mm radius sphere centered on each voxel.

functional connectivity

After preprocessing, a functional atlas was used to separate the cerebral cortex into 100 regions of interest (ROIs). ⁶¹ . The correlation coefficient (FC) between each pair of ROIs was calculated using a Pearson correlation coefficient between each pair of mean signal 'time courses' (representing fluctuations in neural activity over time). This resulted in a matrix N × N FC, with each element representing the strength of the connectivity between a pair of ROIs. Positive values ​​were retained and transformed using Fisher's method into scores. z . This procedure was repeated independently for each patient and scan (baseline and post-treatment).

Modularity of the brain network

The structure of the community or segregation between functional brain networks was measured by summarizing each FC matrix with a common community detection algorithm similar to Louvain 62, ^where The goal is to maximize the extent to which brain areas can be separated into non-overlapping communities or modules. The modularity quality function score, Q ⁶³ , tends to be high when the brain exhibits high segregation between its functional networks (such as strong clusters of functional groups within brain networks/communities with weak functional groups for the rest of the brain).

This approach has been commonly applied to fMRI data to characterize how brain function adapts in various contexts. ⁵⁵ . Here, modularity, Q ⁶³ , was defined in the standard way by:

where THE _ij represents the weight of FC (correlation) between ROI i and j , γ is the free structural resolution parameter (defined as 1) and is the expected zero FC defined with How does total FC (Functional Cash Flow) work across all connections with ROI? i w _i It is the community to which the ROI i is attributed. it is the function Kronecker δe equals 1 if ROI i and j belong to the same community and 0 otherwise ⁶⁴ and k eu k j 2m _ $\frac{{}_{}}{}$ � � � � 2 � k eu = ∑ j I already​ ${}_{}$ � � = ∑ � ⁡ � � � \delta \left( {c_i,c_j} \right) d ( c eu , c j ) $$ � ( � � , � � ) m = 1 2 ∑ eu j I already​ $$ � = 1 2 ∑ � � ⁡ � � �  It is the total FC of the network.

To allow valid comparisons between patients and scans, modularity scores were generated 100 times, and the partition with the highest modularity score was normalized by the average modularity generated from 100 randomly reconnected (shuffled) FC matrices. ⁶⁵ . This common procedure was applied to account for the non-deterministic and quasi-degenerate partitions (solutions with different but similar optimization) generated by the Louvain algorithms and to account for the modularity scores related to the total sum of FC within the network. ⁶⁶ . This process was repeated independently for each patient and scan.

functional cartography

The community detection procedure generates a community assignment for each ROI. We use these labels to determine the extent to which ROIs have been recruited into the functional network to which they typically 'belong', as defined by healthy adults (such as DMN regions should reliably form communities with each other).

First, we created a loyalty matrix, P ³³ , which represents the probability of two regions iej being assigned to the same community in the 100 iterations of the modularity algorithm, defined here as:

where THE = 100 as the number of partitions. For each partition, It equals 1 if the regions iej belong to the same community. a k , o eu , j ${}_{}^{}$ � � , � � , �

Using the loyalty matrix, we summarized how frequently ROIs formed communities with ROIs from the same functional network (network recruitment) or communities formed with ROIs between different networks (internet integration) in the partitions. ³³ . These functional mapping measures were then normalized relative to the average values ​​of 1,000 randomly shuffled network ROI assignments to account for differences in the number of regions in each network. ⁵⁵ . Finally, network recruitment and inter-network integration scores were calculated at the network level using seven predefined cortical networks. ⁶¹ .

Dynamic flexibility

The short TR used in the fMRI protocol of study 2 generated approximately twice the number of time points, and this provided an additional analysis of dynamic flexibility. Multilayer modularity estimate ³⁴ It was performed using a matrix. N × N × T 30 sliding volume windows (37.5 s of real time) with 50% overlap. This window size is typical for estimating dynamic heart rate with fMRI. ⁵⁴ . For each patient and scan, multilayer modularity, Q _ML , was estimated 100 times from each matrix N × N × T FC by:

where total of the multilayer network, and This is the total FC of layer l. THE _ijl It's the FC between ROI, i.e., iej, in layer l. This is the expected zero FC in layer l. The two free parameters γ and ω Structural and temporal resolution are used to scale the number of communities and the strength of interlayered boundaries, respectively. As is typical for fMRI modularity analyses, both were defined as 1 (refs. ^33 , 55 ). m = 1 2 ∑ eu j eu I love you I $$ � = 1 2 ∑ � � � ⁡ � � � � m eu = 1 2 ∑ eu j eu I love I​ ${}_{}$ � � = 1 2 ∑ � � � ⁡ � � � � \frac {{k_{il}k_{jl}} } {{2m_l}} k o o k j eu 2 m eu $\frac{{}_{}}{}$ � � � � � � 2 � �

The multilayer modularity estimate generates a matrix. N × T where each element represents the community's allocation of each ROI in each layer (time window). From this, the flexibility metric, f This can be simply calculated as the number of times an ROI changes their loyalty to the community, given the number of observations. ³³ :

Flexibility scores close to 0 represent rigid ROIs whose loyalty to the community is stable over time, while scores close to 1 represent flexible ROIs whose loyalty to the community changes regularly (highly flexible). Flexibility scores at the network level were defined by the average flexibility of the ROIs assigned to a given network.

Summary of the report

More information about the research project is available at Nature Research Reporting Summary linked to this article.