Large-scale brain connectivity changes following the administration of lysergic acid diethylamide, d-amphetamine, and 3,4-methylenedioxyamphetamine

Psychedelic compounds such as LSD, psilocybin and DMT have shown promise for ameliorating symptoms across a range of psychiatric disorders, but the neurobiological mechanisms that underlie their effects remain incompletely characterised. Comparing a classic serotonergic psychedelic with other psychoactive substances that share some pharmacological actions may help disentangle which connectivity changes are specific to psychedelics and which reflect more general drug effects. Resting-state functional MRI (rs-fMRI) offers a way to probe large-scale functional connectivity, including measures of within-network connectivity (integrity), between-network connectivity (segregation), seed-based whole-brain coupling of canonical resting-state networks (RSNs), and global connectivity as indexed by degree centrality (DC). Avram and colleagues set out to compare acute brain connectivity changes induced by lysergic acid diethylamide (LSD), d-amphetamine and MDMA in healthy volunteers. The primary aims were to (1) test drug effects on RSN integrity and segregation using a canonical RSN parcellation, (2) map voxel-wise whole-brain coupling of each RSN via seed-based analyses, and (3) assess changes in global connectivity with DC. The investigators also examined whether voxel-wise iFC changes related spatially to 5-HT2A receptor density from a reference atlas and whether connectivity alterations correlated with subjective effects measured by the 11D-ASC questionnaire.

The study analysed imaging and behavioural data from the registered clinical trial NCT03019822 carried out in Basel, Switzerland. A double-blind, placebo-controlled, within-subject crossover design was used. Twenty-eight healthy participants were enrolled and 25 were included in the present analyses (12 female, mean age 28.2 ± 4.35 years). Each subject underwent four sessions in randomised, counterbalanced order receiving 0.1 mg LSD, 40 mg d-amphetamine, 125 mg MDMA and placebo. Subjective effects were captured with the 11D-ASC approximately 11 hours after administration, and several physiological parameters (blood pressure, heart rate, tympanic temperature) were measured immediately before and after the fMRI scan; differences versus placebo (ΔPP) were used as covariates in control analyses. Structural and resting-state functional MRI were collected on a 3 T Siemens Prisma using standard sequences (imaging parameters in the supplement). Preprocessing followed the C-PAC pipeline (v1.7.1) and included slice-timing correction, motion correction, scrubbing (FD Jenkinson < 0.2), intensity normalisation, nuisance regression with aCompCor (five components each from white matter and CSF), Friston 24 head-motion regressors, bandpass filtering (0.01–0.1 Hz), anatomical registration and smoothing (5 mm FWHM). Given the debate about global signal regression (GSR) in psychedelic neuroimaging, analyses were performed both with and without GSR. Network integrity (within-network iFC) and segregation (between-network iFC) were assessed using Yeo's seven-network parcellation, excluding the limbic network for signal reasons. Dual regression produced subject-specific spatial maps and time series; integrity was the mean parameter estimate across voxels within each RSN, while segregation was computed as pairwise Pearson correlations between RSN time series followed by r-to-z transformation, yielding 6 × 6 matrices per condition. Voxel-wise seed-based correlation analysis (C-PAC) used the same RSN templates as seeds to map whole-brain iFC for each network. Degree centrality (DC) quantified global connectivity per voxel by correlating voxel time courses, keeping significant correlations (P < 0.001) within a gray-matter mask, and computing weighted DC as the average correlation coefficient across retained edges; DC maps were standardised as z-scores. To test spatial associations between seed-based iFC changes and 5-HT2A receptor density, the fslcc tool computed cross-correlation coefficients between each subject/condition RSN iFC map and a reference 5-HT2AR density atlas, producing one coefficient per subject, RSN and condition. Associations between connectivity measures and subjective effects (11D-ASC) used partial correlations controlling for ΔPP and changes in head motion (ΔFD); the focus was on iFC increases versus placebo, RSN integrity and DC. Statistical inference for network integrity and segregation used repeated-measures ANOVA with post-hoc pairwise t-tests and FDR correction for multiple comparisons; voxel-wise iFC and DC employed one-way repeated-measures ANOVAs in SPM12 with cluster-level familywise error correction (cluster-defining voxel P < 0.001, cluster-level FWE P < 0.05).

Twenty-five participants contributed high-quality rs-fMRI data with no significant differences in head motion across sessions (mean FD, F(3,72) = 1.80, P = 0.15). Analyses revealed both shared and substance-specific effects on connectivity. Network integrity: Repeated-measures ANOVAs (FDR-corrected) showed significant condition effects for the visual network (VIS; F3,96 = 31.3, P_FDR < 0.001), frontoparietal network (FPN; F3,96 = 7.7, P_FDR < 0.001), default mode network (DMN; F3,96 = 3.3, P_FDR = 0.036) and auditory-sensorimotor network (ASM; F3,96 = 28.5, P_FDR < 0.001). Post-hoc tests indicated that VIS integrity was reduced by all three active substances versus placebo, and FPN integrity was likewise reduced in all active conditions. DMN integrity decreased only under LSD compared to placebo, whereas ASM integrity decreased for d-amphetamine and MDMA but not LSD. Direct contrasts showed d-amphetamine reduced integrity in VIS and ASM more than the other drugs in specific comparisons, and MDMA reduced ASM integrity more than LSD. Inclusion of GSR magnified integrity reductions across substances, especially for LSD. Network segregation: Between-network iFC differed across conditions for most RSN pairs except ASM–DMN (F3,96 = 2.5, P_FDR = 0.06). All active substances tended to increase iFC between transmodal networks (DMN, SAL, FPN, DAN) and unimodal networks (ASM, VIS). LSD produced widespread increases in between-network connectivity across nearly all RSN pairs. By contrast, d-amphetamine showed reductions in connectivity between DMN and both DAN and FPN, and MDMA reduced connectivity between VIS and ASM. Comparisons between drugs indicated stronger transmodal interconnectivity for LSD versus the amphetamines (with one exception: DAN–FPN was reduced). d-Amphetamine showed increased DAN–FPN coupling but decreased DMN–FPN, DMN–DAN and DAN–VIS versus MDMA. GSR appreciably altered segregation findings, generally producing more extensive increases in segregation, particularly for LSD. Seed-based whole-brain iFC: Voxel-wise analyses showed that, relative to placebo, most RSN seeds displayed increased connectivity with regions from other RSNs, predominantly linking unimodal and transmodal networks, with LSD yielding the most extensive increases. Decreases in iFC were observed within most seed RSNs, except for some seeds such as DAN and ASM where voxel-wise differences were less apparent. Unlike LSD, both amphetamines decreased iFC between the unimodal VIS and ASM networks. Results were sensitive to GSR, which reduced between-network connectivity and modified effect extents. Degree centrality (DC): LSD increased DC in basal ganglia and thalamus and in regions overlapping ASM and SAL, while decreasing DC in VIS, relative to placebo. d-Amphetamine increased DC in basal ganglia and regions overlapping FPN and SAL, and decreased DC in VIS and portions of ASM (paracentral and right precentral gyri). MDMA raised DC in the left temporal pole and parts of FPN and decreased DC in VIS and ASM similarly to d-amphetamine. In drug–drug contrasts, LSD showed higher DC in some ASM regions compared to the amphetamines but lower DC in right angular gyrus (DMN overlap) and in several FPN and DMN areas compared to d-amphetamine. d-Amphetamine produced lower DC in VIS than MDMA. DC results were relatively robust to GSR compared with other analyses. Control analyses: Correlations between substance-induced connectivity changes and head motion (ΔFD) or physiological parameters (ΔPP) were generally weak and did not survive multiple comparison correction except for an association between temperature change and reduced DC under LSD. Some LSD-induced integrity and seed-based iFC changes showed nominal correlations with ΔPP; amphetamine and MDMA changes did not correlate with ΔPP but had associations with ΔFD in some tests. Associations with 5-HT2A receptor density: The spatial similarity between RSN seed-based iFC maps and a 5-HT2AR density atlas was stronger for LSD than for placebo and both amphetamines across most networks, with the exception of ASM and VIS. Both amphetamines also showed stronger positive associations than placebo for SAL and VIS seed-based iFC respectively. Associations with subjective effects: Partial correlations controlling for ΔPP and ΔFD indicated that for LSD, reductions in network integrity and decreases in DC related to some 11D-ASC subscales, with DMN integrity linked to the largest number of subjective effects. For d-amphetamine, only network integrity measures—predominantly involving SAL—correlated with subjective effects. MDMA showed associations between DMN and ASM integrity and some subjective measures. Similar patterns were reported for the 5D-ASC in supplementary analyses.

Avram and colleagues interpret the results as evidence that LSD produces both overlapping and distinctive alterations in large-scale brain connectivity compared with d-amphetamine and MDMA. The most salient LSD-specific findings were decreased DMN integrity, more extensive increases in between-network connectivity (particularly linking transmodal with unimodal networks), broader seed-based connectivity increases across RSNs, stronger spatial associations between iFC maps and 5-HT2A receptor density, and increased global connectivity in basal ganglia and thalamus. The authors relate these effects to prior reports of psychedelic-induced DMN disruption and enhanced global integration, and to the thalamic ‘‘filter’’ model in which thalamic disinhibition leads to increased information flow to cortex and characteristic psychedelic phenomenology. The discussion highlights that d-amphetamine and MDMA produced similar connectivity signatures to one another across several metrics, despite different primary pharmacologies. The investigators suggest that structural similarity and shared norepinephrinergic effects, and possible dopaminergic–serotonergic interactions, may explain this convergence. Specific differences between the amphetamines were noted: d-amphetamine had more pronounced effects on VIS and increased coupling between attentional (DAN) and executive (FPN) networks consistent with its cognitive-enhancing subjective effects, whereas MDMA showed some effects resembling LSD in selected transmodal interactions. Common effects across all substances included reduced integrity in VIS and FPN, increased between-network coupling of VIS with higher-order networks, and reduced DC in visual regions. The authors propose that shared dopaminergic modulation of visual processing may underlie these commonalities. They also acknowledge methodological factors that may influence results, such as differing integrity computation methods across studies. Key limitations are enumerated by the investigators. The modest sample size could limit power and may partly account for discrepancies with other studies; physiological changes and head motion could have residual confounding effects despite control analyses; inclusion of GSR substantially altered some findings and its use remains contentious in this literature; doses were not matched for equivalent pharmacological efficacy so dose differences could contribute to observed effects; the 5-HT2AR association used a separate PET-based atlas rather than simultaneous PET–fMRI, limiting direct inference about individual receptor–connectivity relationships; and the crossover design raises the possibility of carryover effects because psychedelic-induced connectivity changes may persist for weeks. Overall, the authors conclude that several imaging signatures commonly reported for classic psychedelics—DMN integrity decreases, increased between-network connectivity and elevated basal ganglia/thalamic global connectivity—appear more pronounced for LSD than for the amphetamines tested. They propose that these LSD-specific effects could help identify neural targets relevant to therapeutic mechanisms, while emphasising the need for future studies to address dosing, longitudinal effects, simultaneous PET–fMRI and potential carryover in crossover designs.

The study provides a comparative characterisation of acute large-scale functional connectivity changes produced by LSD, d-amphetamine and MDMA in healthy volunteers. Avram and colleagues report that while all three substances induce some common alterations—particularly in visual and frontoparietal networks—LSD elicits more extensive and distinctive effects, especially decreased DMN integrity, widespread increases in between-network coupling, stronger alignment with 5-HT2A receptor density and increased global connectivity in basal ganglia and thalamus. The authors suggest these LSD-specific signatures may be relevant for understanding psychedelic mechanisms of action and for guiding future clinical research, but they caution that limitations such as sample size, dosing differences, preprocessing choices and potential carryover effects temper definitive conclusions.