This further analysis of an fMRI study (n=19) investigated changes in brain function before versus after psilocybin (with psychological support) in patients with depression (TRD). After treatment, patients showed changes in amygdala function connectivity.
Abstract
“Background: Psilocybin has shown promise as a treatment for depression but its therapeutic mechanisms are not properly understood. In contrast to the presumed actions of antidepressants, we recently found increased amygdala responsiveness to fearful faces one day after open-label treatment with psilocybin (25 mg) in 19 patients with treatment-resistant depression, which correlated with treatment efficacy.
Aims: Aiming to further unravel the therapeutic mechanisms of psilocybin, the present study extends this basic activation analysis. We hypothesised changed amygdala functional connectivity, more precisely decreased amygdala-ventromedial prefrontal cortex functional connectivity, during face processing after treatment with psilocybin.
Methods: Psychophysiological interaction analyses were conducted on functional magnetic resonance imaging data from a classic face/emotion perception task, with the bilateral amygdala and ventromedial prefrontal cortex time-series as physiological regressors. Average parameter estimates (beta weights) of significant clusters were correlated with clinical outcomes at one week.
Results: Results showed decreased ventromedial prefrontal cortex-right amygdala functional connectivity during face processing post- (versus pre-) treatment; this decrease was associated with levels of rumination at one week. This effect was driven by connectivity changes in response to fearful and neutral (but not happy) faces. Independent whole-brain analyses also revealed a post-treatment increase in functional connectivity between the amygdala and ventromedial prefrontal cortex to occipital-parietal cortices during face processing.
Conclusion: These results are consistent with the idea that psilocybin therapy revives emotional responsiveness on a neural and psychological level, which may be a key treatment mechanism for psychedelic therapy. Future larger placebo-controlled studies are needed to examine the replicability of the current findings.”
Authors: Lea J. Mertens, Matthew B. Wall, Leor Roseman, Lysia Demetriou, David J. Nutt & Robin L. Carhart-Harris
Notes
This paper is included in our ‘Top 10 Articles on Psychedelics in the Treatment of Depression‘
Summary
Introduction
Serotonergic psychedelics have shown promising results in disorders such as obsessive-compulsive disorder, end-of-life psychological distress, addiction and depression, when administered under supportive conditions. Long-term follow-ups have revealed stable antidepressant effects for some patients.
Serotonergic psychedelics interact with multiple receptor types, but have high affinity for the serotonin (5-HT) 2A receptor, which mediates their hallucinogenic effects via full or partial agonism. This receptor’s role in increased excitability, plasticity and adaptability may be underlying their therapeutic effects.
Psilocybin may be a promising treatment for depression.
Previous work has identified some post-acute neurofunctional changes after psychedelics, including increased connectivity within the default mode network (DMN) and decreased parahippocampal-prefrontal RSFC, both predictive of treatment response at five weeks.
In the same group of patients as the present study, fMRI was used to assess amygdala responsiveness to emotional faces pre- and post-treatment with psilocybin. The results revealed increased amygdala activity to fearful faces, which was predictive of depressive symptoms at one week and treatment response at one-day, one-week, and three-weeks.
The increased amygdala responsiveness to negative faces in depressed patients and acute amygdala reactivity under psilocybin are particularly striking, as amygdala involvement and hypersensitivity to negative stimuli have long been thought to be a component of clinical depression.
In order to further investigate the therapeutic mechanisms of psilocybin, we predicted that the decreased FC between the amygdala and the ventromedial PFC during face processing after psilocybin might account for the increased amygdala responsiveness. Lastly, we predicted that this change might relate to post-treatment changes in depression severity and rumination tendencies. Alongside basic depression scores, rumination scores were used as a secondary clinical outcome of interest. Rumination has different neural underpinnings in depressed patients as compared with healthy controls.
Procedure
A total of 19 patients with TRD underwent two psilocybin-assisted therapy sessions a week apart. FC was assessed using fMRI during an emotional face paradigm pre- and post-psilocybin, and clinical outcome measures were administered at baseline and at multiple time-points after the high-dose session.
Major depression and treatment-resistance were defined as 16+ on the HAM-D and lack of improvement despite two adequate courses of pharmacologically distinct antidepressant medications over at least six weeks each, within the present depressive episode.
Written informed consent was obtained from all participants during the screening visit, and ethical approval was obtained from the National Research Ethics Service, the UK Medicines and Healthcare Products Regulatory Agency, and Imperial College London.
Patient characteristics
17 patients met the criteria for severe or very severe depression at baseline, and 2 patients were moderately depressed. All patients were asked to discontinue their antidepressant medication during the period of the trial, and all but 2 patients adhered to this request.
fMRI design
Imaging was performed on a 3T Siemens Tim Trio using a 12-channel head coil at Invicro, London, UK. Functional T2*-weighted echoplanar images (EPIs) were acquired for the functional task scans.
Patients were shown blocks of emotion faces on two occasions, baseline and the morning after the high-dose session. They had to passively view the faces and press a button to check that they were still attentive.
fMRI analysis
Blood-oxygen-level-dependent (BOLD) pre-processing was conducted with four different software packages: FMRIB Software Library (FSL), AFNI, Freesurfer and Advanced Normalization Tools (ANTS). The following pre-processing steps were applied: motion correction, brain extraction, rigid body registration, non-linear registration to 2 mm MNI brain, scrubbing, spatial smoothing and nuisance regressing.
General linear models were used to analyse the BOLD-weighted fMRI data, and time-series were derived from the functional data using FMRIB’s Automated Segmentation Tool (FAST).
Two different PPI analyses were conducted using seed regions of interest (ROIs) in the amygdala and vmPFC. These analyses included separate regressors for all facial expression conditions (happy, fearful, neutral) as compared to the rest-period, and their respective interactions with the ROI time-series.
Face regressors were convolved with a gamma function to simulate a canonical haemodynamic response function. The simple and complex PPI models provided similar results, but with different levels of detail.
In this secondary analysis, we looked at the contrasts between the ROIs for fearful and happy faces, happy versus rest, and fearful>neutral faces, as well as the contrast between the ROIs for fearful vs neutral faces.
Higher-level analyses were performed on fMRI data comparing subjects before and after a high dose of psilocybin. Significant clusters were interpreted using the Harvard-Oxford cortical and subcortical structural atlases.
The relationship between the fearful PPI contrast and the basic amygdala BOLD response was tested by correlating the response amplitudes from the fearful PPI contrast and the results from the task analysis contrast.
Relationship with clinical outcomes
To analyse the relationship between face processing information and clinical outcomes, we used the average parameter estimates from the two simple PPI models and the statistical software program IBM SPSS Statistics for Windows.
We limited the primary analyses to one clinical outcome of interest: the self-rated Beck Depression Inventory (BDI). The QIDS-SR16 was administered at more time-points for comparison with results by Roseman et al. (2018).
The Beck Depression Inventory (BDI) is a reliable and valid 21-item self-rating questionnaire for the assessment of depressive symptoms. It can be used to differentiate between absent, mild, moderate and severe depression.
The QIDS-SR16 is a 16-item self-report inventory that assesses the severity of depressive symptoms in reference to the past seven days. The total score determines the severity of depression.
A 22-item self-report scale was used to measure rumination at baseline, one week after and at three-months follow-up. A higher score reflecting more ruminative thinking and behaviour was examined at one week.
Anxiety levels were measured with the State-Trait Anxiety Inventory (STAI) at baseline and one-week post-treatment. The STAI is a widely-used 40-item self-report inventory based on a four-point Likert scale (‘almost never’ to ‘almost always’).
Statistical analysis
Changes in depression and anxiety levels were analysed posttreatment. The relationship between rumination and the primary clinical outcome was investigated by comparing the changes in BDI scores at one week with the changes in STAI scores at one week and the anxiety in-scanner rating.
We performed multiple correlation analyses and t-tests on each cluster and construct separately, in order to reduce the risk of type two errors due to an overly conservative correction.
PPI: amygdala
Results from the primary (simple) PPI model revealed increased FC between the amygdala and visual areas during face processing after psilocybin treatment.
The secondary analysis (complex) PPI model showed stronger FC between the amygdala and similar visual areas during happy and neutral faces, but not during fearful faces after treatment with psilocybin.
PPI: vmPFC
A whole-brain analysis of the primary (simple) PPI model showed that the vmPFC was more connected to areas in the left occipital and parietal lobes during face processing post-treatment versus pre-treatment.
When the different facial expressions were taken into account, the vmPFC showed increased FC with the right angular gyrus and lateral occipital cortex during fearful faces, with the occipito-parietal visual areas centred around the cuneal cortex during happy faces, and with the left cerebellum during neutral faces after psilocybin treatment.
Relation with amygdala response to fearful faces (task-analysis)
The PPI fearful contrast did not correlate with the task analysis after treatment with psilocybin, but the vmPFC-right amygdala FC did correlate with a lower amygdala response to fearful faces before treatment.
Relationship with clinical outcomes
Results show that post-treatment FC changes were related to BDI scores and rumination and anxiety scores.
Clinical outcomes were significantly improved at one week post-treatment, with 63.2% of patients showing a treatment response. Rumination levels decreased from baseline to one week and then rose at three months, despite remaining significantly lower than at baseline.
vmPFC-right amygdala connectivity post-treatment was not significantly related to depression levels at one week, nor to change in BDI or rumination from baseline. However, vmPFC-occipital-parietal connectivity post-treatment was significantly correlated with BDI scores at one week, and so were the respective change scores.
Discussion
The current analyses constitute an extension to previous work that revealed increased amygdala responsiveness post treatment with psilocybin for TRD. We hypothesized that decreased amygdala-vmPFC FC in particular, would relate to improvements in depression, anxiety, and rumination scores, plus in-scanner anxiety, one-week post-treatment.
vmPFC-right amygdala FC
Our findings provide some support for the hypothesis that psilocybin treatment affects the amygdala-PFC connection. This effect was primarily driven by the viewing of fearful and neutral faces.
Increased amygdala responsiveness to fearful faces was related to improved clinical outcomes, but decreased amygdala-prefrontal connectivity was not significantly related to depression severity at one week. However, lower connectivity was associated with less rumination.
Researchers found no association between decreased prefrontal-amygdala FC and levels of anxiety, fear conditioning or PTSD, which challenges the general view of amygdala responsivity/hyperactivity and decreased amygdala-prefrontal connectivity as being indicative of psychopathology per se.
Psilocybin therapy allows patients to more fully engage with emotions, whereas conventional antidepressants down-regulate emotional responsiveness. This might be an important underlying mechanism involved in the emotional reconnection reported with successful psychedelic therapy.
While decreased vmPFC-right amygdala connectivity was observed during the processing of fearful and neutral faces, this change did not significantly correlate with increased amygdala responsiveness.
Further changes in FC
The amygdala and vmPFC showed increased connectivity with occipital and parietal cortices during face processing posttreatment with psilocybin, and this was linked to change in depression and anxiety scores at one week as well as criteria for response and remission of depression.
Although the secondary (complex) analysis did not reveal a significant difference when contrasting amygdala FC during fearful versus neutral faces, the results do support the predicted results.
Limitations
The given study was an open-label feasibility study with a small sample size, and it suffered from lack of statistical power due to multicollinearity between regressors. Furthermore, two separate PPI analyses were conducted with two ROIs, adding to the problem of multiple comparisons.
The study sample comprised a very specific patient group: TRD. To rule out that these results are sample and/or TRD-specific, future studies should explore psilocybin as a treatment for non-treatment-resistant major depression in a larger sample, using a placebo-controlled design. Based on results from the present study, it is not possible to disentangle the effects of psilocybin from those of the psychological support provided during the treatment. However, adding a psilocybin without psychological support condition would resolve this issue.
While patients were requested to down-titrate their antidepressant medication over a minimum wash-out period of two weeks, patient 2 continued taking venlafaxine throughout both dosing and scanning sessions.
FC analyses provide information on temporal correlation between activity in spatially separate regions of the brain, but do not provide information on directionality and causal interactions between brain regions.
Psilocybin treatment improved emotional processing in TRD patients, and rTMS can temporarily disrupt vmPFC function. Eye-tracking could be used to ensure that measured effects are not solely due to attentional changes.
Conclusion
Psilocybin is a novel and promising treatment strategy for TRD, with increased FC between the vmPFC and the bilateral amygdala, and occipital-parietal cortices during face processing post-treatment with psilocybin. This may be related to an increase in emotional sensitivity and acceptance post-treatment with psilocybin. The proposed therapeutic mechanisms of psilocybin treatment are quite different to those hypothesised to underlie the action of classical antidepressants. However, there are also some similarities.