Classical hallucinogens and neuroimaging: A systematic review of human studies: hallucinogens and neuroimaging

Serotonergic classic hallucinogens (including LSD, mescaline, psilocybin, DMT/ayahuasca) produce marked changes in perception, mood and cognition and have been reported to show anxiolytic, antidepressant and anti‑addictive effects in experimental and clinical settings. Pharmacologically, these compounds act primarily as agonists at cortical 5‑HT2A receptors and indirectly increase frontocortical glutamatergic activity, but the neural substrates mediating their subjective and therapeutic effects remain incompletely characterised. Existing human neuroimaging work has sought to map acute and longer‑term brain effects, yet before this review the authors judged that no systematic synthesis of such imaging studies had been published. Dos Santos and colleagues therefore set out to perform a systematic review of human neuroimaging studies of serotonergic hallucinogens. Rather than attempting a quantitative meta‑analysis, the investigators chose a narrative systematic approach because of heterogeneity in drugs, doses, outcome measures, secondary variables and the small number of imaging studies for some compounds; they aimed to summarise acute and non‑acute imaging findings across PET, SPECT, MRI and fMRI studies to clarify convergent neural targets and suggest directions for future research.

The review followed PRISMA guidance and searched PubMed, Lilacs and SciELO for studies published up to 12 April 2016. Search terms combined drug names (ayahuasca, DMT, psilocybin, LSD, mescaline) with imaging terms (mri, fmri, pet, spect, imaging, neuroimaging). No language restrictions were applied and reference lists of included papers were hand‑searched. Inclusion criteria required original, peer‑reviewed human studies reporting structural, functional or neurochemical neuroimaging data for acute or non‑acute effects of serotonergic hallucinogens; animal studies, reviews, qualitative pieces, case reports, conference abstracts and similar items were excluded. Two reviewers independently screened titles/abstracts and full texts, resolving disagreements by discussion and, if needed, a third reviewer. Extracted variables comprised authors, year, location, design, sample size, drug and dose, imaging modality, regions analysed, statistical thresholds and main findings. The authors classified included reports by drug (mescaline, DMT, psilocybin, LSD, ayahuasca) and by imaging technique (PET, SPECT, MRI, fMRI). From an initial yield of 279 citations (278 unique), 25 studies met inclusion after screening and full‑text review; most investigated acute administration in within‑subject, placebo‑controlled cross‑over designs, while two were retrospective observational studies. The review therefore summarises a heterogeneous set of small studies across multiple imaging modalities rather than aggregating effect sizes.

Overall synthesis: Despite methodological heterogeneity and generally small samples, the reviewed studies converge on modulation of networks implicated in visual processing (occipital cortex) and in cognitive/emotional processing (frontolateral/frontomedial cortex, medial temporal lobe, amygdala). Acute neurochemical imaging (PET/SPECT) tended to show regional increases in perfusion or glucose metabolism in frontal and limbic regions, whereas many resting‑state fMRI studies reported decreases in activity or altered functional connectivity in overlapping regions. Mescaline: An open‑label SPECT study in Germany (11 healthy male volunteers, mean age 35.5) who ingested 0.5 mg mescaline reported significant increases in regional cerebral blood flow in frontal cortical regions, especially on the right, at peak effect (240–270 min), correlated with subjective intensity, and decreased activity in posterior cortical areas (P < 0.05). No significant subcortical effects or task performance differences were reported. DMT: A randomized, double‑blind, placebo‑controlled, cross‑over BOLD fMRI study in 14 healthy volunteers tested bolus plus infusion regimens of DMT (and S‑ketamine as comparator). DMT significantly slowed reaction times and blunted inhibition‑of‑return effects relative to placebo (P < 0.05, corrected). In task paradigms probing visual and auditory alertness, DMT produced reduced BOLD activity in extrastriate visual regions and temporal regions (visual effects dominant; reported at P < 0.001, uncorrected), whereas S‑ketamine increased activation in several frontal and temporal regions. Psilocybin: PET with 18F‑FDG in an open‑label sample (10 healthy volunteers) found global increases in cerebral metabolic rate of glucose, notably in frontomedial/frontolateral cortex, anterior cingulate and temporomedial cortex, with CMRglu changes correlating with dose and subjective effects (P < 0.01, uncorrected). Additional PET work reported regional increases and decreases in rMRGlu across cortical and subcortical sites, with different compounds producing partly distinct metabolic patterns. Intravenous psilocybin studies using arterial spin labelling (ASL) and resting BOLD fMRI (single‑blind, within‑subjects, n≈15) observed significant decreases in rCBF and BOLD signal in bilateral thalamus, posterior cingulate/precuneus, medial prefrontal cortex and higher‑order visual areas (P < 0.05, corrected); the magnitude of rCBF decreases correlated with subjective intensity. Re‑analyses of these fMRI data reported: increased memory‑related activations in limbic and medial temporal regions during personal memory cues under psilocybin; increased coupling between the DMN and salience/right frontoparietal/auditory networks (reduced orthogonality between DMN and task‑positive networks); reduced stability and increased repertoire (entropy) of transient connectivity states; increased BOLD signal variance and spectral power in anterior cingulate and hippocampi; and widespread decreases in low‑frequency power in frontal/parietal association regions. In an emotion‑processing study (double‑blind, randomized cross‑over, n=25), oral psilocybin (0.16 mg/kg) attenuated right amygdala responses to negative and neutral pictures (P = 0.001 and P < 0.001, corrected) and reductions in amygdala BOLD correlated with increases in positive mood. Other connectivity re‑analyses linked psilocybin to decreased top‑down amygdala→primary visual cortex modulation during threat and to connectivity changes associated with ego‑dissolution. LSD: Resting and task fMRI studies of LSD (extracted text reports a dose of "75 g. i.v."; the unit in the extraction appears unclear) in healthy volunteers found increased cerebral blood flow and functional connectivity in visual cortex and enhanced coupling between primary visual cortex and widespread cortical and subcortical regions, with correlations between visual cortex coupling and subjective imagery/hallucination ratings (P < 0.05, corrected). LSD also reduced DMN functional connectivity (so‑called DMN “disintegration”), correlated with subjective ego‑dissolution, and decreased oscillatory power in lower‑frequency bands (1–30 Hz) across the brain, with relationships between power decreases and phenomenological effects. Analyses of global functional connectivity showed increased connectivity in thalamus and high‑level association cortices overlapping spatially with known 5‑HT2A receptor density, and links between connectivity increases in temporo‑parietal and insular regions and ego‑dissolution ratings. An interaction between LSD and music increased parahippocampal→visual cortex effective connectivity and memory recollection. Ayahuasca/DMT in oral form: A randomized, double‑blind, placebo‑controlled SPECT study (n=15 healthy males, 1.0 mg DMT/kg, orally) reported bilateral activation of anterior insula/inferior frontal gyrus and right anterior cingulate/medial frontal regions (P < 0.002, uncorrected), and activation of amygdala/parahippocampal gyrus in the left hemisphere. BOLD fMRI imagery tasks in frequent ayahuasca users (n=9) found increased activation in primary and higher visual areas, parahippocampal and posterior cingulate cortices during imagery, with primary visual activation comparable to natural vision; ayahuasca also altered fronto‑occipital functional connectivity such that primary visual cortex more strongly led other cortical areas during imagery. In an open‑label SPECT trial in 17 patients with Major Depressive Disorder, a single oral ayahuasca dose (2.2 mL/kg, 0.8 mg DMT/mL) was associated with increased perfusion in left nucleus accumbens, right insula and left subgenual area (P < 0.01, corrected) and with significant reductions in depressive and anxiety scores at 1, 7 and 21 days post‑dose (P < 0.05 reported for symptom changes). Polydrug/long‑term use and structural findings: A retrospective PET/SERT and 5‑HT2A receptor study of 24 MDMA/hallucinogen users versus 21 controls found marked SERT reductions in MDMA users (pallidostriatum 19% reduction; amygdala 32%; neocortex 56% with regional variation; P ≤ 0.001) while hallucinogen users per se had largely normal SERT. Combined groups showed a small neocortical 5‑HT2A receptor reduction (~9%) that lost significance when two high‑binding controls were excluded, suggesting limited robustness. A structural MRI study of 22 long‑term Santo Daime ayahuasca users versus 22 matched controls reported cortical thinning in mesotemporal and several association regions (posterior cingulate, precuneus, superior occipital gyrus, etc.) and increased thickness in precentral gyrus and anterior cingulate (P < 0.002, uncorrected); these morphological changes correlated inversely with age of onset, intensity of prior use and measures of self‑transcendence, and were not accompanied by increased psychopathology according to the extracted text.

Dos Santos and colleagues interpret the convergent imaging findings as indicating that serotonergic hallucinogens modulate a distributed network encompassing frontolateral/frontomedial cortices, medial temporal lobe structures and occipital visual regions—areas implicated in self‑awareness, cognition, memory and emotion. Neurochemical imaging (PET/SPECT) more consistently reported regional increases in perfusion or glucose metabolism in frontal and limbic regions, while many resting‑state fMRI studies showed decreased activity or altered connectivity in overlapping hubs such as the default mode network (DMN; a set of regions more active during rest and mind‑wandering). The authors note that apparent discrepancies between PET/SPECT and fMRI may relate to the different temporal windows sampled by radiotracers (which reflect longer time‑averaged metabolic changes) versus the faster haemodynamic and BOLD measures captured by fMRI. Functional connectivity and electrophysiological findings were interpreted as reflecting both ‘‘disintegration’’ of canonical networks (reduced DMN integrity and decreased low‑frequency oscillatory power in key hubs) and increased cross‑network integration or repertoire (more transient connectivity states, higher entropy and enhanced coupling between sensory and association cortices). These network effects were linked conceptually to reported subjective phenomena such as expanded awareness, ego‑dissolution and vivid imagery, and were proposed as potential neural mechanisms supporting the psychological and therapeutic effects of the drugs. Mechanistically, the review emphasises agonism at deep‑layer cortical 5‑HT2A receptors on pyramidal neurons as a primary action leading to altered cortical synchronisation, modulated glutamatergic transmission and downstream changes in neurotrophic signalling (for example BDNF/GDNF), potentially promoting neuroplasticity. The authors argue that decreased DMN activity and reduced amygdala reactivity may underlie reductions in rumination and anxiety, providing a plausible pathway for antidepressant and anxiolytic effects observed in some studies. They also highlight associations between drug‑occasioned mystical or ‘‘self‑transcendent’’ experiences and long‑term psychological outcomes reported elsewhere. Key limitations acknowledged by the investigators include small sample sizes, high methodological heterogeneity (different compounds, doses, designs, imaging modalities and statistical thresholds), frequent lack of placebo or control groups in some reports, and challenges accurately quantifying dose/composition in retrospective studies. The authors caution that some receptor binding effects were small and not robust to sensitivity analyses. For future research they recommend improved standardisation of drug delivery and dosing, larger and better characterised samples, multimodal imaging integration, and methodological developments to clarify temporal dynamics and causal links between neural changes and clinical outcomes.