Neuroimaging in moderate MDMA use: A systematic review

Mueller and colleagues frame the review around ongoing uncertainty over whether MDMA ("ecstasy") produces neurotoxic effects in humans. Much of the prior neuroimaging literature sampled heavy or long‑term users, with cumulative lifetime exposure in the hundreds or thousands of tablets and frequent polydrug use, whereas most MDMA consumers use the drug transiently and at much lower cumulative doses. The authors note additional relevance given renewed interest in MDMA as an adjunct in psychotherapy, where administration is limited to a few controlled sessions, and argue that evidence specific to moderate patterns of use is required to inform both public health debates and clinical considerations. This systematic review therefore set out to identify and synthesise human neuroimaging studies that examined non‑acute brain structure, function or neurochemistry in groups with moderate MDMA exposure. Moderate use was pre‑defined as an average of <50 lifetime episodes of ecstasy use or a lifetime consumption of <100 ecstasy tablets. The review aims to assess whether neuroimaging findings in such cohorts provide convincing evidence of MDMA‑related brain alterations and to characterise methodological limitations that affect interpretation.

The investigators performed a systematic review following PRISMA guidance. They searched PubMed through July 2015 using the search string (mdma OR ecstasy OR 3,4methylenedioxymethamphetamine) AND (mri OR fmri OR pet OR spect OR imaging OR neuroimaging), with no language restriction, and screened reference lists of included articles. Inclusion criteria required original, peer‑reviewed human studies (observational or interventional) that used structural, functional or neurochemical neuroimaging to evaluate non‑acute MDMA effects and that included at least one group with mean lifetime use below the predefined moderate thresholds (<50 episodes or <100 tablets). Titles and abstracts were screened and full texts reviewed independently by two reviewers (FM, MS), with a third reviewer (CL) resolving disagreements when necessary. For each included study the reviewers extracted centre, year, design, imaging modality, sample sizes (including overlaps between studies where reported), demographics, detailed exposure metrics (tablets, episodes, estimated mg), usual and maximum dose per occasion where available, age at onset, time since last use, duration of use, matching of control groups (including matching for other drug use), abstinence verification, domains and regions tested, statistical thresholds and principal findings. Where necessary they converted heterogeneous exposure metrics into common units: a supplementary search of tablet content studies yielded a weighted mean MDMA content of 76 mg per tablet, and a weighted mean of 1.3 tablets per occasion was used to convert session‑based reports into tablet estimates. The reviewers included studies published up to July 2015. If subject overlap between publications was suspected but not reported, the authors contacted original investigators to clarify and incorporated that information when available. When key data could be computed from a report, the reviewers derived and included those values.

The search identified 250 PubMed records plus one additional article from reference screening; after screening and application of the moderate‑use inclusion criteria, 19 articles were included (publications dated 2001–2014). Nineteen studies comprised a mixture of modalities: ten used task‑based fMRI, four were neurochemical imaging studies (primarily PET and SPECT), and the remainder used structural MRI, voxel‑based morphometry (VBM), diffusion metrics (DTI) or other MR techniques; some centres contributed multiple papers with partially overlapping samples. Most studies (15/19) were retrospective observational designs; four were prospective. Control groups were present in all but two studies, but few studies matched controls for other drug use. Abstinence was usually checked, commonly by urine screening. Across modalities the overall pattern reported by the authors was a lack of consistent, replicated MDMA‑related alterations in cohorts with moderate lifetime exposure. In the fMRI literature (10 studies) task domains included working memory, associative memory, attention, decision‑making, motor function, visual stimulation and semantic memory. Behavioural task performance differences between users and controls were generally absent (all but two studies reporting no significant performance differences). Specific neuroimaging findings were heterogeneous. Daumann and colleagues reported, using both liberal and conservative thresholds, mixed regional BOLD differences in parietal, temporal and premotor areas when comparing user groups (including polydrug and ‘‘pure’’ MDMA users) with drug‑naïve controls. Jacobsen et al. observed less deactivation in the left hippocampus during an attention‑demanding condition in adolescent MDMA users and reported a small negative correlation between left hippocampal activation and abstinence (reported r = -0.05), but no correlation with cumulative dose. Jager et al., in a prospective cohort with minimal exposure at baseline, found no significant group differences across working memory, associative memory and selective attention paradigms. Becker et al. reported decreased encoding‑related activity in the left parahippocampal gyrus that correlated negatively with interim MDMA use in a sample with limited amphetamine‑type stimulant exposure, but noted baseline differences and possible recovery effects during abstinence. Watkins et al. reported greater activation in left precuneus and right superior parietal lobule during semantic encoding, with a positive correlation between right superior parietal activation and lifetime ecstasy use (Spearman's rho rS = 0.43, p = 0.042). Karageorgiou et al. found increased activation in the right supplementary motor area during a motor tapping task and within‑group correlations between MDMA amount and BOLD in basal ganglia and precentral regions; Salomon et al. reanalysed that dataset and reported reduced intraregional thalamic coherence and altered functional connectivity related to lifetime MDMA use. Bauernfeind and colleagues reported within‑group positive correlations between cumulative MDMA use and stimulus‑evoked activation in the lateral geniculate nucleus (rS = 0.59), BA17 (rS = 0.50) and BA18 (rS = 0.48), with some effects reduced after adjustment for scanner and stimulus method. Structural MRI and other MR techniques produced mixed results. Several VBM and cortical thickness studies that included modest lifetime MDMA exposure reported either no global differences or small, spatially limited volume reductions (for example in left orbitofrontal and right occipital regions). Mackey et al., studying a sample with higher use of prescription stimulants and cocaine than MDMA, reported regional increases and decreases in subcortical and cortical volumes but noted that including MDMA in analyses did not materially change results. Two consecutive prospective studies by De Win et al. assessed diffusion metrics (fractional anisotropy, apparent diffusion coefficient) and regional perfusion: with liberal thresholds they reported various increases/decreases in FA, ADC and regional blood flow, but after correction for multiple comparisons only a decrease in dorsolateral frontal grey matter perfusion remained significant in the earlier cohort; later findings were inconsistent and potentially confounded by interim increases in use of cocaine, alcohol and cannabis. Neurochemical imaging studies were also inconsistent. Three investigations of serotonin transporter binding reported no clear alterations across a range of lifetime doses. Two studies examined 5‑HT2A receptor binding with divergent outcomes: Di Iorio et al. reported increased estimated 5‑HT2A receptor densities in several cortical regions with positive correlations to cumulative MDMA dose (duration of abstinence did not affect binding), whereas Erritzoe et al. found no increased receptor densities in a sample with higher cumulative dose and shorter abstinence. A PET study of glucose metabolism in a polydrug inpatient sample showed negative correlations between duration of MDMA use and metabolism in left postcentral/inferior parietal gyrus, right inferior frontal/dorsolateral prefrontal cortex and right superior temporal pole, but lacked a non‑using control group and was confounded by other substance use. Overall, where studies reported significant associations these were often unreplicated, based on small samples, used liberal statistical thresholds or were potentially driven by co‑consumption of other substances. The reviewers emphasised pervasive methodological limitations across the literature: polydrug confounding, variable and often incomplete exposure metrics, reliance on self‑report for dose and drug identity, self‑selection of participants, small samples, retrospective designs and inconsistent control for legal drugs such as alcohol and nicotine. The reviewers also reported their data‑standardisation steps (weighted mean tablet MDMA content 76 mg; mean 1.3 tablets per occasion) used to harmonise exposure metrics across studies.

Mueller and colleagues interpret the assembled evidence as providing little, if any, convincing support that moderate lifetime exposure to MDMA produces consistent structural, functional or neurochemical changes detectable by the neuroimaging techniques used to date. Findings across studies were heterogeneous and largely non‑replicated even within similar domains. Structural MRI and VBM studies in comparable moderate‑dose samples mostly found null or divergent regional effects that could not be confidently attributed to MDMA because analyses often did not disentangle amphetamine‑type stimulant exposure, and lifetime doses in those studies were small. The authors highlight several sources of uncertainty that limit causal inference. Polydrug use is common in MDMA users and many studies either did not match controls for other substance use or provided insufficient data to adjust adequately for these confounders; alcohol, nicotine and cannabis in particular were often neglected despite potential structural and functional effects. Observational and typically retrospective recruitment strategies create risks of selection bias and confounding by pre‑existing traits such as sensation seeking. Reliance on self‑reported drug identity and dose is another limitation because tablets sold as ecstasy vary in MDMA content and may contain other psychoactive compounds. Heterogeneity in exposure reporting (tablets, episodes, mg) and sparse reporting of per‑occasion maximal doses further complicate interpretation; the reviewers note that peak dose per occasion may be important for toxicity but is infrequently reported. The discussion acknowledges specific inconsistencies within modalities: DTI measures of white matter integrity produced conflicting results across studies; PET/SPECT measures of serotonin transporters showed no clear alterations across studies, while two 5‑HT2A receptor studies produced opposing results (Di Iorio et al. reporting increased density, Erritzoe et al. not replicating this). The authors consider possible moderators such as dose and sex differences — Di Iorio's sample was exclusively female while Erritzoe's was mostly male — but also note that differences in cumulative dose and abstinence time complicate such explanations. In terms of implications, the authors state that the current neuroimaging literature does not provide clear evidence that moderate MDMA exposure produces brain changes detectable with the modalities used, and therefore does not, on the basis of these data alone, indicate that MDMA used as an adjunct in psychotherapy should be regarded as inherently dangerous. Nevertheless, they caution that absence of evidence is not evidence of absence: some alterations may be below detection limits, some studies were not designed to assess MDMA neurotoxicity specifically, and the overall quality of evidence is limited. The authors recommend future work that better accounts for co‑consumption (including alcohol and nicotine), uses appropriately matched controls, reports more complete exposure metrics (including per‑occasion dose), and applies prospective designs where feasible to reduce bias and improve causal inference.

The reviewers conclude that studies of moderate MDMA exposure are heterogeneous in design and quality and yield inconsistent neuroimaging results. Within the moderate dose range examined, there is no clear, replicated evidence from structural, functional or neurochemical imaging that MDMA induces brain alterations. However, this review does not establish that moderate MDMA use is free of neurotoxic effects, because limitations in study design, exposure characterisation, statistical power and modality sensitivity could mask true effects. The authors urge improved future studies with better matching for other substance use, clearer dosing information and prospective designs to resolve remaining uncertainties.