Psilocybin-induced changes in neural reactivity to alcohol and emotional cues in patients with alcohol use disorder: an fMRI pilot study

CONCLUSION

The present study sought to characterize psilocybin-induced alterations in neural activity to alcohol and emotional cues which may account for therapeutic effects in patients with alcohol use disorder (AUD). Psilocybin treatment was associated with engagement of various prefrontal cortical areas (lateral and medial PFC) and the caudate, and disengagement of the insula, motor and cerebellar areas, and temporal, parietal, and occipital cortices. These post-acute effects (i.e. occurring in the days following psilocybin administration) largely implicate brain areas previously reported to be acutely affected by psilocybin. Importantly, group-by-time interactions were mostly driven by changes in the psilocybin group, suggesting that psilocybin-assisted therapy alters neural activity across the cortex and within multiple limbic structures. The high prevalence of overlapping regions across conditions suggests treatment effects were largely non-specific to stimulus type (alcohol, negative, and positive cues), and possibly reflects alterations to the saliency of visual stimuli, affective processing, or a general mood stabilizing effect. Psilocybin-treated patients displayed increased caudate, mPFC, vlPFC, and dlPFC engagement across multiple cue types, suggesting functional reorganization of structures involved in emotional regulation, response inhibition, goal-directed action, and executive functioning. However, the directionality of some of the effects are-at initial pass-inconsistent with normalization of AUD-related dysfunction as meta-analyses indicate hyperactive frontostriatal circuits in AUD. Specifically, studies have reported hyperactivity of the mPFC and dorsal striatum in response to alcohol cues, relative to healthy controls, and treatment-induced downregulation of this pathway within AUD samples. While this warrants caution when interpreting the present study findings, a few lines of evidence offer potential explanations. First, hyperactivity to alcohol cues in these regions are frequently reported in the context of hypoactive responses to other stimulus categories (i.e., negative/stress, neutral, positive stimuli). Such alcohol-specific hyperactivity supports the notions of pathological incentive salience toward alcohol cues, and concomitant devaluation of non-drug stimuli in AUD. Therefore, it is plausible that increased activity in these brain regions across alcohol and affective stimuli reflects a broadening of incentive salience and changes in general affective processing. Such a widening of the attentional scope may be critical to belief updating in predictive coding and Bayesian models of addiction, as has been posited to be a mechanism of action of psychedelics. Secondly, directionality has been mixed as studies have also reported hypoactivity within frontostriatal regions in AUD. For example, hypoactive mPFC and striatum responses to alcohol and negative/stress images, in contrast to hyperactive responses in these regions to neutral/relaxing images, have been reported in AUD compared to healthy controls. Since both hyper-and hypoactivity in the mPFC predicted drinking behavior and relapse in these studies, valence-dependent responses in the mPFC may be clinically relevant. Notably, we observed decreases in orbitofrontal cortex (OFC), a subregion of the vmPFC, and increases in the dmPFC, areas responsible for emotional and cognitive aspects of self-referential processing, respectively. In line with our findings, successful inhibition of cue-induced cocaine craving has been negatively associated with OFC activity and positively associated with vlPFC activity in the right hemisphere. Thus, we speculate psilocybin might dampen the emotional and enhance the cognitive self-relevancy of emotionally charged stimuli. It is also important to consider that mPFC and caudate were activated in concert with ventral and dorsal divisions of the lateral PFC, matching what is observed in healthy controls who show greater lateral PFC recruitment compared to AUD patients. Additionally, greater medial and lateral PFC activity during the regulation of alcohol craving and negative emotions has been observed in patients with AUD. Thus, while psilocybin-induced increases in medial PFC is inconsistent with normalization of alcohol cue sensitization in AUD, patterns match neural signatures of cognitive regulation, suggesting that psilocybin may enhance top-down executive control, rather than blunt the saliency of alcohol-related cues. Future studies should consider the complex and potentially opposing roles of ventral, dorsal, and orbital divisions of the medial PFC, and contemporaneous lateral PFC co-activation, when evaluating psilocybin modulating effects on cue-reactivity. Further support for psilocybin's putative effects on cognitive regulation can be drawn from the neurobiological underpinnings of attentional and inhibitory control in AUD. For example, IFG response is negatively associated with attentional biases to drug cues; heightened dlPFC and vmPFC is observed during alcohol interference in a Go-NoGo task; diminished dlPFC recruitment is observed when making reward-related decisions and processing negative prediction errors; and dlPFC stimulation reduces alcohol craving. In the context of psilocybin treatment, one study found increased dlPFC, vlPFC, and mPFC response in an emotional conflict Stroop task, and another found mPFC functional connectivity changes during a focused attention meditation practice. Considering this research in the context of AUD suggests that psilocybin might diminish preference for alcohol cues and engage hubs of inhibitory control. However, follow-up studies using executive functioning tasks are needed to directly test this proposition. While comparisons with other studies of psilocybin's action are difficult due to heterogeneities in clinical samples, assessment time points (acute versus post-acute), and task designs, there has been some consistency in reported brain regions, including: the mPFC, a hub of the DMN (see Gattuso et al.for a review of psychedelic effects on the DMN), the ACC and insula, nodes of the SN, and lateral PFC, a hub of the executive control network. Focusing strictly on post-acute effects, psilocybin has been shown to induce connectivity changes in the cingulum, striatum, and mPFC, with decreased mPFC-PCC connectivity predictive of positive mood 4 months later among health controls. In treatment-resistant depression, psilocybin altered mPFC, ACC, and PCC connectivity one day post-treatment, with decreases in mPFC connectivity predicting depressive symptoms 5 weeks later. In a negative affective task similar to the one employed in the present study, dlPFC and mPFC decoupling with the amygdala one day post-psilocybin has been shown to predict reductions in rumination 5 weeks posttreatment. Moreover, Barrett and colleagues found psilocybin increased positive affect and increased PFC response to emotionally conflicting stimuli. While we did not observe functional connectivity changes in the mPFC as has been reported in other samples, we found increases in ACC-caudate and vlPFC-precentral gyrus connectivity, suggesting psilocybin may modulate frontostriatal and motor circuits, respectively. Whether these changes reflect top-down or bottom-up modulation deserves attention in future studies using effective connectivity approaches. Our findings of increased PFC activity and functional connectivity with striatal and motor areas add to this growing body of literature, and together, independent research groups are beginning to converge on putative therapeutic substrates of psychedelics. Augmented striatal activity to alcohol cues has been most widely reported in the ventral striatum (nucleus accumbens) and putamen, responsible for reward/motivation and motor control/habitual behavior, respectively, whereas the caudate appears to contribute more to goal-directed action and cognitive control. Given this functional distinction (and concomitant PFC activation), heightened caudate response and caudate-ACC connectivity post-treatment might reflect top-down cognitive control and diminished emotional perturbation. Relatedly, diminished functional connectivity between the striatum and ACC has been associated with AUD severity in a response inhibition task, and abstainers display stronger striatal-ACC connectivity than non-abstainers. Intriguingly, we did not observe decreases in the nucleus accumbens or amygdala as expected. Decreases in the left putamen were evident in the interaction but nonsignificant for within-psilocybin comparisons. Acute reductions in left putamen have been reported following psilocybin administration. In light of these considerations, we speculate that the effects observed in the present study reflect a state of improved self-regulatory control in relation to long-term goal pursuit (sobriety or reduced drinking) and emotional equipoise irrespective of changing environmental stimuli. Psilocybin-treated patients also displayed broad reductions in insular, motor, temporal, occipital, and cerebellar activity relative to placebo controls. These findings are in line with an activation likelihood estimation metaanalysis in AUD that found hyperactivity and treatment-induced reductions in these brain regions, including after cue-exposure therapy. Overall, the patterns of deactivation observed after psilocybin point toward normalization. For example, greater activation in insular, temporal, parietal, and occipital cortices have generally been found during alcohol cues exposure in AUD versus health controls(with some inconsistencies). A role for the cerebellum in addiction and craving has also emerged, with activity positively correlating with AUD severity. Our findings of attenuated cerebellar response support a growing consensus of its contributions to higher-order cognitive functions such as negative emotionality, salience detection, executive control, memory, and self-reflection. Acutely, psilocybin has also been shown to decrease activity in the insula, hippocampus, motor cortex, and temporal areas, although directionality might be dependent on relative versus absolute measurement. Psychedelics modulate areas rich in 5-HT1A receptor expression, such as the insula, raising the possibility that psilocybin may exert inhibitory effects on the insula via agonism at 5-HT1A receptors. In relation to AUD, decreases in insular activity are in line with previous work showing insular hyperactivity and treatment-induced reductions in AUD. The insula has long been associated with interoceptive components of craving and negative affect. Psilocybin-specific decreases in insular activity were robust for alcohol and negative affective contrasts, but not for positive affective cues, suggesting that attenuation of interoceptive processing is specific to craving and negative affect states. Unique to positive affective cues, psilocybin reduced left and increased right hippocampus engagement. Interestingly, hemispheric asymmetries have been established for emotional processing, with left hippocampal lateralization occurring when viewing negative versus neutral pictures. Others have observed increases in relative cerebral blood flow in the right hippocampus acutely after psilocybin administration, raising the question whether these changes persist or undergo temporal reconfiguration that ultimately results in durable clinical effects. We speculate that these lateralized, affect-specific responses might reflect the facilitation of natural, nondrug rewards regaining reinforcing properties and a resetting of the hedonic set point as has been qualitatively reported in the parent studywww.nature.com/scientificreports/ Recent developments in establishing a neural signature of craving have included temporal, parietal, occipital, and cerebellar regions, expanding the neurobiology of addictions beyond the confines of the mesocorticolimbic circuitry which has dominated the field's focus. Koban and colleagues posit that co-activation of visual and posterior attentional areas may be critical to ascribe personal meaning to rudimentary percepts, as has been established for complex emotional states-such as fear and sadness-which are highly embedded in the visual system. From this perspective, it is possible that personal associations with alcohol and emotional contexts are attenuated though PFC engagement and contemporaneous posterior disengagement, giving rise to a decentered, nonjudgmental, and nonreactive perspective as has been reported in the early stages of mindfulness meditation interventions. However, in the absence of brain-behavior analyses and relevant fMRI paradigms, extreme caution is warranted when inferring the cognitive and psychological processes underlying these brain findings. Well-powered studies are needed to examine the relationships between these neural correlates and the proposed cognitive constructs.