Psilocybin pulse regimen reduces cluster headache attack frequency in the blinded extension phase of a randomized controlled trial

Cluster headache (CH) is an intensely painful primary headache disorder that affects roughly 0.1% of the population and exists in episodic and chronic forms; chronic cluster headache (CCH) denotes attacks that persist with less than three months without attacks per year and is associated with anxiety, depression and suicidal ideation. Standard prophylactic drugs such as verapamil or lithium reduce attack frequency in many patients but can produce unacceptable side-effects or inadequate benefit, such that novel prophylactic options are needed. Observational reports and case series have suggested that serotonergic psychedelics, in particular psilocybin, may reduce CH frequency, but these effects had not been prospectively evaluated in a clinical trial setting. Madsen and colleagues set out to assess feasibility, safety and preliminary prophylactic efficacy of a short course "pulse" regimen of moderate-dose oral psilocybin in people with CCH, and to explore whether treatment response relates to changes in hypothalamic functional connectivity (FC) measured with fMRI. The study therefore combines clinical diary-based outcomes (attack frequency and self-rated pain intensity) with plasma psilocin pharmacokinetics and resting-state neuroimaging to probe potential neural correlates of any clinical effect.

This was a small open-label interventional study in ten patients with verified chronic cluster headache (N=10; five females; mean age 49.4 (SD 12.9)). The within-subject design comprised a four-week baseline headache diary period, three oral psilocybin treatments administered at a dose of 0.14 mg/kg bodyweight (COMP360, pharmaceutical-grade synthetic psilocybin) spaced seven days apart, and a four-week follow-up during which patients continued daily headache diaries. Patients met with two trained session facilitators the day before the first treatment and received psychosocial support on treatment days; they also had a debrief/integration meeting on the day after each session. Prophylactic medications were not allowed during the trial. Primary clinical outcomes were feasibility/safety and change in weekly CH attack frequency from baseline to follow-up; the secondary clinical outcome was change in average daily pain intensity (0–10 numeric scale). Patients logged number of attacks, average pain intensity and acute treatment use daily for ten weeks. Blood pressure was measured immediately prior to dosing and at 40, 80, 120, 180 and 240 minutes after administration. Plasma psilocin concentration (PPL) was sampled pre-dose and at multiple time points up to 240 minutes and analysed by ultra-high performance liquid chromatography–tandem mass spectrometry; Cmax and AUC were derived. Neuroimaging comprised structural and resting-state BOLD fMRI acquired on a 3T scanner the day before the first psilocybin session and one week after the last session. Functional connectivity analysis used an 8 mm radius seed in the posterior hypothalamus (MNI 3, −9, −10), with denoising (band-pass, motion/spike regression, aCompCor) and cluster-level inference (voxel p<0.001, cluster-level family-wise error corrected pFWER < 0.05). FC estimates for clusters showing significant connectivity with the seed were extracted for statistical tests. Exploratory analyses included network-level FC (seven canonical networks using Schaefer 400 parcellation) and structural regional measures (cortical thickness and grey matter volumes) derived with FreeSurfer. Statistical analyses for clinical outcomes used the Wilcoxon signed-rank test for baseline vs follow-up; the family-wise error rate for the two main clinical comparisons was controlled using the Bonferroni–Holm method. Change in hypothalamic FC was tested with Wilcoxon signed-rank, and simple linear regression assessed correlations between percent change in attack frequency and change in FC, with FWER control for the neuroimaging tests. Exploratory correlation analyses between PPL metrics (Cmax, AUC) and clinical change were performed by linear regression.

Ten patients entered and completed the study procedures for clinical analyses; nine completed all three psilocybin sessions, one missed the third session for logistical reasons but completed the remainder of the protocol. The dosing regimen was generally well tolerated and no serious adverse reactions were observed. Acute subjective psychoactive effects varied across patients; one participant reported no subjective effects and had the lowest measured PPL Cmax (5.9 μg/L). Blood pressure measurements showed no treatment-related changes. Clinically, weekly CH attack frequency decreased from baseline to the four-week follow-up period. The mean change was −3.57 attacks/week (SD 3.87), corresponding to a mean percent change of −30% (SD 31%), and this reduction was statistically significant after FWER correction (pFWER = 0.008). One patient experienced a complete remission lasting 21 weeks, beginning one day after the first psilocybin session. Average self-rated pain intensity across attacks also decreased: mean change −0.89 (SD 0.94) on the 0–10 scale, a mean percent reduction of −13% (SD 15%), significant after FWER correction (pFWER = 0.03). Use of acute therapies (oxygen or subcutaneous sumatriptan) and the self-reported average duration of daily attacks were unchanged between baseline and follow-up. Due to a COVID-19–related interruption, fMRI data were available for nine patients. Seed-based FC analysis identified four clusters that showed significant positive connectivity with the posterior hypothalamic seed across baseline and follow-up scans: a large diencephalic cluster reported to encompass the hypothalamus, thalamus and caudate (the extracted text also includes an ambiguous fragment mentioning "brain"), and three other clusters located in the left lingual gyrus, right cerebellum and occipito-temporal white matter. Change in hypothalamic FC from baseline to follow-up was not significant at the group level (P-values > 0.6). However, a linear regression revealed a significant negative correlation between percent reduction in attack frequency and change in hypothalamic FC with the diencephalic cluster (pFWER = 0.03, Pearson R = −0.81), indicating greater clinical improvement associated with greater reductions in that FC estimate. No significant correlations were found for the other three clusters (P-values > 0.7). Plasma psilocin levels showed substantial inter-individual variability. Exploratory regression analyses did not find significant associations between PPL Cmax or AUC and percent change in attack frequency or pain intensity (all P-values > 0.9). One patient reported durable positive psychological effects extending at least six months, and another reported improved quality of life despite unchanged long-term headache burden.

Madsen and colleagues interpret these data as the first formal clinical evidence that a short regimen of moderate-dose oral psilocybin can be delivered safely in a controlled setting to people with chronic cluster headache and may reduce attack frequency. The investigators report a substantial but variable reduction in weekly attack frequency (mean −30%) and a smaller reduction in self-rated pain intensity (mean −13%), with one participant experiencing prolonged remission. They further note a negative correlation between clinical improvement and change in posterior hypothalamus–diencephalic functional connectivity, suggesting that modulation of this neural pathway may be related to the prophylactic effect. The authors situate these findings against prior imaging and mechanistic work implicating hypothalamic dysfunction in CH and against known acute psilocybin effects on functional brain organisation and possible neuroplasticity-promoting properties. They speculate that a ‘‘recalibration’’ of aberrant hypothalamic–diencephalic circuits, potentially supported by increased neuroplasticity after psilocybin, could contribute to reduced attack propensity. Key limitations acknowledged by the investigators include the small sample size (ten patients analysed versus an original target of 20), the open-label design without placebo control, recruitment from a tertiary headache centre which may limit generalisability, and exclusion of patients on regular prophylactic medications so results may not extend to those populations. The authors also caution that some observed effects—particularly on subjective pain intensity—could reflect placebo or expectancy effects. They reiterate safety and regulatory cautions: psilocybin has potent psychedelic properties, is illegal in most jurisdictions, and can lead to adverse outcomes in uncontrolled settings; therefore, self-medication outside clinical contexts is discouraged. Finally, the investigators call for larger, placebo-controlled trials in carefully controlled settings to confirm safety, establish effect sizes, determine durability of benefit, and further probe neurobiological mechanisms underlying any prophylactic action of psilocybin in cluster headache.