Two dose investigation of the 5-HT-agonist psilocybin on relative and global cerebral blood flow

The serotonin (5-HT) system modulates perception, cognition, mood and consciousness, and classic psychedelics such as psilocybin produce profound alterations in these domains. Neuroimaging studies of psilocybin and related 5-HT agonists have reported apparently conflicting results: early PET studies, after correction for global fluctuations, found regional increases in frontal metabolism (a "hyperfrontal" pattern), whereas a more recent arterial spin labelling (ASL) study reported widespread decreases in cerebral blood flow (CBF). Because serotonergic drugs can also alter vascular tone, disentangling regional neuronal effects from global haemodynamic changes is important for interpreting pharmacological neuroimaging findings. Lewis and colleagues set out to resolve this contradiction by measuring cerebral perfusion after two oral doses of psilocybin using pseudo-continuous ASL (pCASL) in a randomised, double-blind, placebo-controlled design. They compared two analytic strategies: one that normalised for global perfusion to identify relative regional CBF (rCBF) changes, and one non-adjusted analysis to capture absolute or global CBF (gCBF) effects. The authors hypothesised that normalised analyses would reproduce the hyperfrontal pattern seen in PET, that non-adjusted analyses would show global decreases similar to the recent ASL report, and that the higher dose would produce larger effects than the lower dose.

Study design was randomised, double-blind and placebo-controlled. Two separate participant groups received either a lower medium dose (0.16 mg/kg) or a higher medium dose (0.215 mg/kg) of oral psilocybin in single-dose sessions; placebo (maltose) was administered in the alternate session. Sessions were at least 10 days apart. Because regulatory approval was limited to single-dose administration per participant, the low and high dose conditions were implemented in separate groups rather than within-subject. An anatomical MRI was acquired about 60 minutes after drug intake, followed by a resting-state pCASL perfusion scan. Subjective effects were assessed using the 5D-ASC questionnaire at 360 minutes post-dose. Participants abstained from nicotine and caffeine on test days. Participants were recruited via university advertisements and underwent medical screening (medical history, physical examination, blood tests and ECG), urine toxicology and pregnancy testing; prior heavy psychedelic use was an exclusion criterion. The extracted text does not clearly report the sample size in the Participants section, though inferential statistics reported elsewhere in the paper are consistent with a total sample of 58 participants (two groups). All participants provided written informed consent and the study was approved by Swiss regulatory and ethics bodies. Neuroimaging data were acquired on a Philips Achieva 3.0T scanner. The resting pCASL sequence parameters included TR 4400 ms, TE 20 ms, field of view 240 x 240 mm, matrix 80, 23 slices at 7 mm thickness, post-labeling delay 1525 ms, label duration 1650 ms and 50 dynamics (each dynamic a control and a labelled image). High-resolution T1-weighted images were also collected. Preprocessing (performed with the ASLtoolbox and SPM12) comprised realignment, smoothing (6 x 6 x 14 mm kernel), perfusion-weighted image construction, calibration using M0 images and conversion to absolute mean CBF images; data were visually inspected for motion and all scans were reported to have less than one voxel of movement. Statistical analysis of psychometrics used repeated-measures ANOVA on 11 validated 5D-ASC scales with within-subject factors of drug and scale, a between-subject factor of dose (Low vs High) and covariates age and sex; significant effects were followed by Bonferroni-corrected pairwise comparisons. Group-level ASL analyses used a flexible-factorial general linear model in SPM12. Age and sex were included as covariates throughout and family-wise error (FWE) correction was applied at peak-level p < 0.05. To probe dose effects, the High and Low groups were compared and a t-test was applied to subtraction images (psilocybin minus placebo). The investigators used two approaches to assess relative CBF: the normalization option within the flexible-factorial design (global covariate as nuisance) and an analysis that included each participant’s mean whole-brain CBF (from an AAL mask) as an additional covariate; both yielded similar results and the normalized ANOVA is reported. Non-adjusted analyses were used to characterise gCBF changes. Treatment order effects were also tested on extracted quantitative CBF values.

On subjective measures, the repeated-measures ANOVA (drug x scale x dose, controlling for age and sex) produced significant main effects of drug, scale and dose, and significant interactions including scale*dose, drug*dose, scale*drug and the three-way interaction. Pairwise comparisons showed the High dose produced significantly larger subjective effects than the Low dose on four of the 11 5D-ASC scales: Disembodiment (p = 0.017), Complex Imagery (p = 0.026), Elementary Imagery (p = 0.016) and Audiovisual Synesthesia (p = 0.005). Simple main effects indicated all scales increased after psilocybin relative to placebo (all p < 0.001). Analyses of cerebral perfusion found no dose-dependent effects on either rCBF or gCBF even at uncorrected thresholds (p < 0.001), and both dose groups exhibited similar spatial patterns; consequently, subsequent imaging analyses were collapsed across dose groups. Sex had no detectable effect on rCBF or gCBF. Age showed a negative relationship with gCBF; a pCASL SPM map reported two significant age-related clusters (right frontal superior/inferior orbital area at 24 28 -18, k = 73, t = 6.57; middle frontal gyrus at 36 8 34, k = 93, t = 6.26). When the global signal was regressed out (rCBF analysis), psilocybin produced significant local increases and decreases (FWE p < 0.05). Increases were observed in bilateral inferior frontal gyrus and insula, with right-sided increases in hippocampus, operculum and superior temporal gyrus. Decreases were found predominantly in left-hemispheric structures including the amygdala, lentiform nucleus, globus pallidus, putamen, thalamus, insula and portions of frontal, parietal, occipital and temporal cortices. In the non-adjusted gCBF analysis, psilocybin elicited widespread decreases in absolute perfusion across frontal, parietal, temporal, limbic and occipital cortices as well as cingulate, insula, caudate, putamen, pallidum, amygdala, hippocampus and thalamus. No regions showed increased gCBF under psilocybin compared with placebo. Finally, testing for treatment order effects on extracted quantitative CBF values found no significant differences for placebo or psilocybin (placebo t(56) = 0.84, p = 0.84; psilocybin t(56) = -1.17, p = 0.24).

Using both global-normalized and non-adjusted ASL analyses, Lewis and colleagues reconcile prior discrepant findings by showing that analytic choice determines the apparent direction of psilocybin effects on perfusion. Normalized analyses (rCBF) identified regionally specific increases and decreases, including right-lateralised frontal and temporal increases and left-sided decreases in parietal regions and subcortical structures, a pattern consistent with early PET reports of hyperfrontality after global correction. The non-adjusted analyses demonstrated broad reductions in absolute perfusion across cortical and subcortical regions, aligning with a previous ASL study that reported widespread CBF decreases following intravenous psilocybin. The investigators emphasise that ASL remains an indirect measure of neuronal activity and that psilocybin’s vasoactive properties complicate interpretation: observed gCBF decreases could partly reflect global vascular tone changes rather than purely neuronal suppression. The absence of a dose effect on CBF, despite larger subjective effects at the higher dose, suggests that the mechanisms mediating phenomenology may be subtler than what ASL captures in this dose range or that neurovascular factors decouple subjective ratings from perfusion measures. The authors discuss potential receptor-level explanations, noting that 5-HT2A activation on excitatory pyramidal cells may produce regional increases in activity and blood flow, whereas 5-HT1A actions and indirect inhibition via interneurons could produce regionally opposing decreases; they caution that receptor-specific contributions cannot be disentangled in the present design. Methodological implications are highlighted: inclusion or omission of a global covariate meaningfully changes phMRI inferences, and reporting both normalized and non-normalized results can clarify the distinct regional and global consequences of serotonergic challenges. Limitations acknowledged by the authors include the between-subjects dose comparison, the relatively narrow dose range tested, and differences in route of administration across studies they compare. They conclude that understanding both neuronal and neurovascular effects of psilocybin is important for interpreting neuroimaging results and for informing clinical applications, and they recommend careful selection and transparent reporting of analysis approaches in pharmacological ASL research.