MDMA-induced changes in within-network connectivity contradict the specificity of these alterations for the effects of serotonergic hallucinogens

The analysis pooled data from two nearly identical randomized, placebo-controlled, double-blind, crossover trials conducted in Basel, Switzerland. Each study included four experimental sessions with washout periods of at least 7 or 10 days. MDMA hydrochloride was given as a single oral dose of 125 mg; placebo capsules matched in appearance were used. The resting-state fMRI acquisition was scheduled to coincide with expected peak drug effects (roughly 1.5 hours after administration). Healthy volunteers aged 18–45 were recruited; after exclusions for data quality and motion the final sample comprised 45 participants (22 women, 23 men; mean age 26.2 ± SD 4.4). Inclusion and exclusion criteria targeted general medical and psychiatric health, limited prior illicit-drug exposure, MRI compatibility, and other standard safety considerations. Urine drug screens and pregnancy tests were used as appropriate before sessions. Imaging was performed on a 3 Tesla scanner with eyes-closed resting-state runs of 300 volumes (TR 1.8 s). Preprocessing in CONN/SPM included motion realignment, unwarping, slice-timing correction, tissue segmentation, MNI normalisation, 5 mm smoothing, scrubbing (ART), linear detrending and regression of six motion parameters. Physiological noise was addressed with CompCor (five principal components from white matter and CSF) and band-pass filtering (0.008–0.09 Hz). Quality control excluded subjects with <5 minutes of usable data after scrubbing or with maximum framewise displacement (FD) >0.75 mm, leaving the 45-person sample. Group independent component analysis (ICA) used a 20-component model to facilitate comparison with established RSN templates; ten components corresponding to canonical RSNs were selected by spatial cross-correlation with Smith et al. Within-network FC was compared between MDMA and placebo using paired t tests on ICA-derived maps with a voxel threshold p < 0.001 (uncorrected) and a cluster-size threshold p < 0.005 (Bonferroni-corrected to account for testing 10 components). Degree centrality was computed following Schaefer et al.: correlations between each voxel and all other grey-matter voxels were thresholded at r > 0.15, summed, z-transformed and compared with voxel p < 0.001 and cluster p < 0.05 (FWE). Because MDMA elevates heart rate, blood pressure and body temperature and can affect motion, the investigators tested potential confounding effects in two ways. First, they correlated changes in FC (MDMA–placebo) with changes in maximum FD and physiological parameters (Spearman’s rho), reporting both uncorrected and Bonferroni-corrected results. Second, demeaned changes in these variables were entered as covariates of no interest in second-level analyses. Finally, a qualitative literature search of PubMed (to 17 September 2020) was performed to compare findings with other monoaminergic drug studies that used ICA-based RSN analyses.

ICA identified components that largely matched canonical RSNs: visual networks 1–3, cerebellar, auditory, frontoparietal 1–2, executive control and a posterior-dominant default mode network (DMN) (the anterior DMN was not reliably identified in this dataset). Spatial correlations with the Smith et al. templates ranged from moderate to strong for several components. Within-network FC was reduced under MDMA relative to placebo in visual networks 1 and 2, the posterior DMN (medial–posterior cluster centred on the posterior cingulate), the sensorimotor network, and the cerebellar network (cluster-level p < 0.005, FWE, on a voxelwise p < 0.001 basis). The largest and most widespread decreases occurred in the two visual networks and the sensorimotor network; DMN reductions were spatially smaller. In contrast, parts of the frontoparietal networks 1 and 2 showed increased within-network FC. No significant changes were detected for visual network 3, the auditory network, or the executive control network. After adjustment for motion and physiological covariates, cerebellar-network decreases were no longer significant. Degree centrality analysis showed reduced centrality in cerebellar and occipital regions and increased centrality across several cortical areas (including pre/postcentral gyri, superior parietal lobule, middle and superior frontal gyri, temporal regions), the brain stem, a small anterior thalamic cluster and part of the left caudate. However, inclusion of demeaned motion and physiological covariates abolished significant degree centrality differences between MDMA and placebo. Physiological measures were significantly elevated after MDMA versus placebo (systolic BP +21.6 mmHg ± SD 11.9; diastolic BP +11.5 mmHg ± SD 9.2; heart rate +16.5/min ± SD 10.2; body temperature +0.3 °C ± SD 0.4). Using an uncorrected threshold of p < 0.10, no systematic correlations between maximum FD or physiological changes and FC cluster values were found, and none survived correction for multiple comparisons. The literature screen returned nine fMRI studies of other monoaminergic agents; most did not examine the same set of networks, though the posterior DMN was a common target and produced mixed findings across studies.

Müller and colleagues interpret their results as evidence that acute decreases in within-network FC — previously reported after serotonergic hallucinogens — are not specific to drugs that produce overt hallucinogenic subjective effects. MDMA, which is not classically hallucinogenic but strongly stimulates monoaminergic systems, produced very similar within-network FC decreases in visual networks, the posterior DMN and the sensorimotor network. These effects were largely independent of the measured potential confounders; the cerebellar finding was an exception, losing significance after covariate adjustment. By contrast, changes in degree centrality reported after an SSRI were not robustly reproduced: degree centrality differences in this study were eliminated when motion and physiological covariates were included, suggesting those results may be susceptible to nonspecific drug-induced artefacts. The authors note a potentially meaningful overlap with antidepressant-relevant circuitry: decreased FC in the posterior DMN (posterior cingulate) was observed here and in previous studies of LSD and ayahuasca, and prior literature links elevated DMN FC to depression and its reduction to several antidepressant treatments. They therefore suggest that acute destabilisation of pathological DMN connectivity might contribute to longer-term therapeutic effects seen with some serotonergic interventions, and that MDMA’s posterior-DMN decrease could be relevant in this context. Important limitations acknowledged by the investigators include the absence of dose–response data, variability in sample sizes and statistical thresholds across comparative studies, and the possibility of global pharmacological confounds such as altered neurovascular coupling that are difficult to exclude entirely. They also point out methodological constraints: low-dimensional ICA may obscure subnetwork-specific effects, and different degree-centrality computation methods hinder direct comparison across studies. Finally, because MDMA, SSRIs and LSD all engage multiple monoaminergic systems to varying degrees, the present data cannot resolve whether observed FC changes are mediated by serotonergic, dopaminergic or noradrenergic mechanisms. In conclusion, the study demonstrates that monoaminergic stimulation by MDMA can induce within-network FC alterations that closely resemble those reported after hallucinogens, calling into question the specificity of those alterations for hallucinogenic action. The authors recommend more direct linking of subjective effects to fMRI measures, stricter control for confounders, and comparative studies across related psychoactive drugs to clarify underlying neurotransmitter contributions.