Decreased brain modularity after psilocybin therapy for depression.

Importance Psilocybin therapy shows antidepressant potential; our data link its antidepressant effects to decreased brain network modularity post-treatment. Objective To assess the sub-acute impact of psilocybin on brain activity in patients with depression. Design Pre vs post-treatment resting-state functional MRI (fMRI) was recorded in two trials: 1) Open-label treatment-resistant depression (TRD) trial with baseline vs 1 day post-treatment fMRI (April-2015 to April-2016); 2) Two-arm double-blind RCT in major depressive disorder (MDD), fMRI baseline vs 3 week after psilocybin-therapy or 6 weeks of daily escitalopram (January-2019 to March-2020). Setting Study visits occurred at the NIHR Imperial Clinical Research Facility. Participants Adult male and female patients with TRD or MDD. modularity one day post-treatment correlated with response the psilocybin-arm, improvements at the 6-week primary endpoint Two independent trials reveal converging evidence for action, namely, less modular brain dynamics.


Introduction
Depression is a highly prevalent mental health condition 1 , the incidence of which has increased during the Covid-19 pandemic 2 , e.g., as re ected in increased prescriptions of antidepressant medications 3 .
However, even the best performing antidepressant drugs show modest e cacy 4 , non-negligible side effects 5 , discontinuation problems 6 , and high relapse rates 7 , highlighting the need for new improved treatments 8 .
Patients with a diagnosis of depression often exhibit a negative cognitive bias, characterised by pessimism 9 poor 'cognitive exibility' 10 , rigid thought patterns 11 and negative xations regarding oneself and future prospects 9 . A number of authors have directly or indirectly taken inspiration from dynamical systems theory to describe depressive episodes as 'attractor states', i.e., stereotyped states with 'gravitational pull' 12,13 .
Neuroimaging research has consistently found examples of abnormal brain functioning in depression, resonant with its phenomenology [14][15][16] . A hierarchically supraordinate intrinsic brain network 17 , the default mode network (DMN), is associated with introspection and self-referential thinking 18 and these functions are often overactive in depression 9 . Indeed, several studies have linked excessive engagement of DMN functioning with depressive symptomatology 19,20 .
In addition to the DMN, other higher-order brain networks such as the executive (EN) and salience networks (SN) have been implicated in depression [21][22][23] . These networks are associated with the 'cognitive control' of thoughts and attention switching 24,25 between introspective thought and an external focus 26 . Such attentional switching is impaired in depression 27 . Tellingly, the serotonin 2A (5-HT2A) receptor subtype, which is the key proteomic binding-site of 'classic' serotonergic psychedelic drugs, such as 'psilocybin' 28 , is most densely expressed in a broad pattern of cortex that closely resembles a conjunction map of the DMN, EN and SN 29 , corresponding to the transmodal portion of the brain's principal hierarchical gradient 17 .
In the last 15 years, at least six separate clinical trials have reported impressive improvements in depressive symptoms with psilocybin therapy 30 . Included among these studies are: 1) an open-label trial in treatment-resistant depression 31 , and 2) a double-blind randomised controlled trial (DB-RCT) with an active comparator, the selective serotonin reuptake inhibitor (SSRI) and conventional antidepressant, escitalopram 32 . These two trials, which included pre and post-treatment functional magnetic resonance imaging (fMRI), are the focus of the present paper's analyses.
The therapeutic action of psilocybin and related psychedelics is incompletely understood. However, one model proposes that psychedelics cause a 5-HT2A receptor induced dysregulation of spontaneous population-level neuronal activity, linked to a temporary 'disintegration' of intrinsic functional brain networks 33 and a hypothesised decrease in the precision-weighting of internal predictive models instantiated by these functional modules 34 . One important corollary of modular 'disintegration' appears to be the broadening of the brain's functional repertoire of states or 'state-space' -commensurate with a broader or atter global energy landscape 35 .
Here we hypothesize that the well-replicated nding of brain network disintegration and desegregation under psychedelics 36,37 will be apparent sub-acutely 38 , in post-treatment resting-state fMRI data. We also hypothesise that this effect, consistent with a atter energy landscape, will relate to improved depression outcomes, and also that it will not be observed after a course of the selective serotonin reuptake inhibitor (SSRI), escitalopram.

Trial overviews
The trial designs for and the main clinical outcomes of the open-label 31 and DB-RCT 32   Exclusion criteria were: Immediate family or personal history of psychosis, risky physical health condition (physician-assessed), history of serious suicide attempts, positive pregnancy test and MRI contraindications. The DB-RCT had the additional exclusion criteria of selective serotonin reuptake inhibitor (SSRI) contraindications or previous escitalopram use. Eligible patients undertook telephone screening interviews, provided written informed consent and their mental and physical medical histories were thoroughly evaluated.

Interventions
Patients in the open-label trial attended a 1-day pre-treatment baseline session that included eyes-closed resting-state fMRI and clinical assessment (Figure 1a). This was followed by two psilocybin therapy dosing days (DD), separated by 1 week. A low-dose of psilocybin (10mg) was orally ingested on DD1 and followed by a high-dose dose (25mg) on DD2. The follow-up fMRI and clinical assessment occurred one day after DD2. Patients attended an on-site clinical assessment at 1-week post-DD2 and completed further clinical assessment electronically at 3 and 6 months.
Of the 59 MDD patients in the DB-RCT, a random number generator allocated 30 to the 'psilocybin-arm' and 29 to the 'escitalopram-arm' (Figure 1b). Patients attended a pre-treatment baseline eyes-closed resting-state fMRI. DD1 consisted of either 25mg psilocybin (psilocybin-arm) or a presumed negligible 1mg psilocybin (escitalopram-arm) dose. All patients were informed that they would receive psilocybin but were blind to the dosage. DD2 occurred three weeks after DD1 and was a duplicate dosage. There was no dosage-crossover. Beginning one day post DD1, patients took daily capsules for 6 weeks and 1 day in total. For both conditions, one capsule per day was ingested for the rst 3 weeks and two thereafter. Capsule content was either inert placebo (microcrystalline cellulose, psilocybin-arm) or escitalopram in the escitalopram-arm, 10mg for the rst 3 weeks and 2 x 10mg = 20mg total thereafter.

Measuring depression severity
Beck Depression Inventory 1A (BDI-1A) scores were used to assess depression severity in both studies. This patient-rated measure captures a broader range of symptoms, with an additional focus on the cognitive features of depression, compared to other measures such as the QIDS-SR-16 40 . In the openlabel trial, BDI was measured at baseline and 1 week, 3 months and 6 months post DD2. For the DB-RCT, BDI was measured at baseline and 2, 4 and 6 weeks post DD1.

Measuring brain network modularity
For each scanning session, resting-state fMRI was recorded using a 3T Siemens Tim Trio MRI scanner at Invicro, London, UK (see Supplemental eMethods: MRI acquisition).
Our principal metric of interest was brain network modularity, a measure that describes the degree of segregation between the brain's functional networks (or, the communities of brain regions) 41 .
Preprocessed fMRI data were used to estimate functional connectivity (FC) matrices from 100 cortical regions as de ned by a functional atlas 42 (see eMethods: MRI preprocessing; Functional connectivity) that were subjected to a commonly used community detection algorithm 43 . This step seeks to maximise the extent to which the brain regions can be segregated into non-overlapping communities or modules 44 .
See the Supplemental eMethods: Modularity section for details on modularity estimation and normalisation.
The modularity metric used in the present work has been applied in many previous contexts, including depression studies 21,45,46 , where high modularity scores indicate a greater degree of separation between brain networks. Scores were compared between imaging sessions and correlated with depression severity scores.

Brain network characteristics
The modularity metric assesses a particular property of global brain function. To gain a ner-grained perspective on changes to individual networks, we employed methods from functional cartography 47 .
For study 1, we measured changes in network recruitment as the probability that brain regions of a network form communities with regions from the same network and network integration as the probability that regions form communities with regions from other networks (see eMethods: Functional cartography). For study 2, a dynamic community detection and exibility analysis 48 was applied using a sliding-window approach where FC was estimated for multiple windows in time, instead of for the entire scan. Network exibility was then de ned by the average number of times that brain regions within a given network changed their community a liation across time 47  Decreased brain modularity following psilocybin therapy Con rming our primary hypothesis prediction, brain network modularity was signi cantly reduced ( Figure  3a) one day after psilocybin therapy (mean difference, -0.29; 95% CI 0.07 to 0.50, P=.012), indicative of an increased integration of brain networks.
Decreased modularity predicts long-term clinical outcomes We hypothesised that decreased brain network modularity would relate to sustained improvements in depression severity following psilocybin therapy. To test this, we calculated Pearson correlations between the post-treatment brain modularity and the BDI scores from the 3 post-treatment timepoints (1 week, 3 months, 6 months). After false discovery rate (FDR) correction for multiple-comparisons, a strong signi cant correlation was observed at 6 months (Figure 3b -Pearson, r=0.64; P=.023). Although consistent with this, relationships at 3 months (r=0.46; P=.114) or 1 week (r=0.29; P=.284) did not survive correction. Furthermore, the pre vs post-treatment change in modularity signi cantly correlated with the change in BDI score at 6 months, relative to baseline (Figure 3c -Pearson, r=0.54; P=.033). These results indicate that decreased brain modularity relates to long-term improvements in depression symptom severity. Psilocybin therapy has greater e cacy than escitalopram for treating depression Decreased depressive symptom severity was signi cantly greater under psilocybin than escitalopram, indicating superior e cacy of psilocybin therapy vs. escitalopram (Figure 4). This was con rmed within this neuroimaging sample by a signi cant arm x timepoint ANOVA interaction for the BDI scores (F, 4.47; P=0.005). FDR-corrected pairwise comparisons relative to baseline were signi cantly different at 2 weeks (mean difference, -8.73; 95% CI = -13.55 to -3.91, P=0.002), 4 weeks (mean difference, -7.79, 95% CI = -13.62 to -1.95, P=0.013) and at 6 weeks (mean difference, -8.78, 95% CI = -15.58 to -1.97, P=0.013), all favouring the psilocybin-arm (see 32 for full sample). Increased brain network integration is speci c to psilocybin therapy Con rming our primary hypothesis (Figure 5a-b) and replicating the ndings of the open-label trial, brain network modularity signi cantly reduced following psilocybin therapy (mean difference, -0.39; 95% CI = -0.75 to -0.02, P=0.039). Individuals' decreases in brain network modularity signi cantly correlated with greater depression recovery at the 6-week primary endpoint (Pearson, r=0.42, P=.025, one-tailed).
Importantly, this replication was speci c to the psilocybin-arm; in the escitalopram group (Figure 5d-e), modularity did not change from baseline to week 6 (mean difference, 0.01; -5% CI -0.35 to 0.33, P=0.945) and there was no signi cant relationship with changes in BDI scores (Pearson, r=0.08; P=0.361, onetailed).
Depression recovery correlates with increased cognitive network exibility.
Next, we examined the dynamic exibility of the brain's canonical networks. This ner-grained metric summarises how often brain regions change their community allegiance during the course of an fMRI scan. Post-treatment change in network exibility were correlated with the changes in BDI score. Speci cally, increased EN dynamic exibility related to greater depression recovery at the 6-week primary endpoint for the psilocybin-arm (Pearson, r=-0.76, P=0.001). Signi cant relationships predominantly involved the EN, SN and dorsal attention networks (Figure 5c). No signi cant correlations between BDI and dynamic exibility were observed in the escitalopram-arm (Figure 5f).

Discussion
In light of growing evidence for the antidepressant e cacy of psilocybin therapy 32 , these ndings advance our understanding of possible underlying brain mechanisms. Across two trials decreased brain modularity was observed and correlated with improvements in depressive symptomatology. Moreover, this antidepressant action may be speci c to psilocybin therapy, as no changes in modularity were observed with the conventional SSRI antidepressant, escitalopram.
Research into the acute brain action of psychedelics has revealed well-replicated changes in global brain function that are somewhat consistent with those observed here, i.e., an increased repertoire of interregional and between-network FC 35,36,49 . Our previous analysis had suggested some contrasting changes in the architecture of spontaneous brain function one day following psilocybin treatment for depression relative to what has been observed during the acute psychedelic state itself; i.e., spatially expanded DMN FC (post-treatment for TRD) versus acute DMN disintegration 31 . Other teams have, however, reported some evidence suggestive of increased inter-network FC 1 week and 1-month postpsilocybin 50 , as well as 1-day post-ayahuasca 51 , including consistent increases in DMN-SN FC 51 , albeit in healthy volunteers. The present ndings greatly extend on previous work however, by showing robust, reliable and treatment-speci c decreases in brain modularity post psilocybin therapy for depression that relate to antidepressant e cacy.
The present modularity metrics may be more sensitive indices of the antidepressant action of psilocybin than previously applied time-averaged within and between-network FC analyses 31 . Indeed, they may bear relevance to other FC metrics applied to acute-state psychedelic fMRI data 35,36,49 where a general picture of increased global FC and a broadened state-space has emerged 34 . In this context, the results could be understood as a 'carryover' effect resembling brain dynamics associated with the acute psychedelic state, albeit at an attenuated level and in a speci c population (i.e., depressed patients).
Previous research on resting-state activity in depression has found abnormal community structure 52 and heightened network modularity 53 , correlating with symptom severity 21,54 . Additional work implies elevated FC between limbic regions such as the amygdala and high-level cortical regions in depression, correlating with ruminative symptoms 55 , as well as elevated within-DMN FC also correlating with rumination 19 . Taken together, a model emerges of abnormally modular spontaneous brain function in depression that is effectively remediated by psilocybin therapy. According to various ndings, the FC energy landscape or state-space in depression can be described as abnormally constricted, paralleling the narrow, internally focused, ruminative quality of mood and cognition in the disorder 13 . In contrast, psilocybin therapy appears to expand the brain's state-space, both acutely 56 and, (as shown here), postacutely in depressed patients, in a fashion that correlates with antidepressant outcomes. Moreover, this 'liberating' action of psilocybin is paralleled by subjective reports of emotional release via psychedelic therapy 57,58 as well as sub-acute increases in behavioural optimism 9 , cognitive exibility 59 , and psychological exibility post psychedelic-use 60-62 .
We believe that this 'liberating' effect of psilocybin on cortical activity occurs via its direct agonist action on cortical 5-HT2A receptors, dysregulating activity in regions rich in their expression. We believe chronic escitalopram does not have the same effect on brain modularity due to its more generalised action on the serotonin system and likely predominant effect on inhibitory postsynaptic 5-HT1A receptors, which are richly expressed in limbic circuitry 33,63 .
Beyond the global decrease in network modularity post psilocybin, we observed functional changes in default mode, executive and salience network dynamics that are consistent with neurobiological models of depression 64 . These higher-order frontoparietal networks house the highest density of 5-HT2A receptors, the principal action-site for serotonergic psychedelics 28,29 . High-level frontoparietal networks are implicated in the acute action of psychedelics, where they show reduced modularity and increased communication with regions ordinarily outside of their community limits [35][36][37]49 .
The EN and SN have been associated with tasks requiring cognitive exibility such as, planning 65 , learning 66 and task-switching 24,25 ; impaired functioning of these networks have been reported in depression 21,53 , and other disorders exhibiting cognitive in exibility such as traumatic brain injury 67 , autism spectrum disorder 68 and obsessive-compulsive disorder 69 . Our results suggest that decreased modularity, or increased exibility, of EN regions, following psilocybin therapy, is a key component of its therapeutic mechanism of action. We did not formally assess cognitive exibility in the clinical trials reported here but we did observe improvements in general cognitive functioning post psilocybin treatment in the DB-RCT (see 32 ). Phase 3 clinical trials will be required to achieve licensing for psilocybin therapy and pragmatic trials will inform questions regarding treatment practicability and optimization 70 . For brain imaging studies, we would recommend network modularity analyses like those employed here. fMRI datasets are burdensome and susceptible to noise, contributing to the challenge of detecting reliable biomarkers. Composite measures, such as network modularity, combined with a research domain, symptoms-based approach to psychological data, may be a productive way forward 70 .
The dynamic exibility analysis employed in the DB-RCT provided a useful perspective. However, it is limited by its requirement for fMRI scans with many timepoints. Timeseries need to be of su cient length to be split into multiple time-windows that are themselves long enough to compute reliable FC measures. It can be challenging to reliably collect high-quality data of su cient length in patient cohorts. Advances to fMRI temporal resolution, however, may improve this issue in the near future.

Conclusion
Depression presents considerable challenges to multiple stakeholders. Here, we identify a robust, reliable and potentially speci c biomarker of response to psilocybin therapy for depression that may help to explain why it could become a valuable new treatment option. Figure 1 Trial timeline schematics. a) Open-label trial. Eligible patients attended a baseline resting-state fMRI and clinical assessment visit. This was followed by two psilocybin therapy dosing days (DD) separated by 1
Patients attended a baseline resting-state fMRI and clinical assessment visit and were randomly assigned to the psilocybin-arm (top-branch) or escitalopram arm (bottom-branch). The psilocybin-arm involved 2x25mg psilocybin therapy DD's with 3 weeks of daily placebo capsules following each DD. The escitalopram-arm involved 2x1mg psilocybin therapy DD's with 3 weeks of 10mg daily escitalopram following DD1 and 20mg of escitalopram following DD2. Both groups attended a post-treatment fMRI and clinical assessment visit 6 weeks and one day after DD1.   Open-label trial: Treatment-resistant depression patient responses to psilocybin therapy relate to decreases in brain network modularity one day post-treatment. a) Brain modularity (Q -normalised) signi cantly reduced, indicating a global increase in brain network integration. The solid and dotted lines represent the mean and median, respectively. Patient's data are connected by solid lines and rendered red if modularity decreased. b) Absolute post-treatment scan modularity correlated with absolute Beck depression inventory (BDI) scores at 6 months post-treatment. c) Change in brain modularity after treatment signi cantly correlated with the change in BDI scores between baseline and at 6 months. d) DMN recruitment decreased and its integration with the EN (gold) and SN (purple) increased (bars represent the standard error of the mean) following psilocybin therapy. e) Paired t-statistics of the change in network recruitment (diagonal) and between network integration (off-diagonal). This exploratory analysis uses an uncorrected P<.05 threshold (Red=increase, Blue=Decrease). Comparisons do not survive multiple comparisons correction. DMN=Default Mode Network, DA=Dorsal attention, EN=Executive Network, LI=Limbic, SM=Somatomotor, SN=Salience Network, VS=Visual. DB-RCT: Beck depression inventory (BDI) scores from each study arm. a) Boxplot for each timepoint at which BDI scores were measured. Decreases in depression severity from baseline were signi cantly greater in the psilocybin-arm (red) than in the escitalopram-arm (blue) at each timepoint. Median values are represented by the central marks, the 25th and 75th percentiles by the box edges and the whiskers extend to the data range. b) Individual patient BDI scores in a raster plot. Each row represents a single patient, each column represents a timepoint (BL=Baseline, wk=week). Rows were ordered by each patient's total BDI score sum across timepoints. Black/grey=Severe depression, white/light grey=Mild to no depression. The solid and dotted lines on the distributions represent the mean and median, respectively. Individual patient data are represented, connected with solid lines between sessions which were rendered red if modularity decreased between sessions. b) Patient individual differences (relative to baseline) correlate with improved treatment response at the 6-week primary endpoint. c) Signi cant correlations between changes in dynamic network exibility and BDI (relative to baseline) at 6 weeks are coloured and those that survive FDR-correction are denoted with a * (white = P>.05). The equivalent analyses in the escitalopram-arm did not show signi cant session differences in brain modularity (e) and individual differences in BDI at 6 weeks (relative to baseline) did not correlate with modularity changes (e) or network exibility (f). DN=Default Mode Network, DA=Dorsal attention, EN=Executive network, LI=Limbic, SM=Somatomotor, SN=Salience Network VS=Visual.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. natMedsuppMaterials.docx