The risk of chronic psychedelic and MDMA microdosing for valvular heart disease

Tagen and colleagues introduce the practice of psychedelic microdosing—regular use of sub-hallucinogenic doses of psychedelic substances—and note that its long-term safety is poorly characterised. Surveys indicate that microdoses are typically around 5–10% of a full psychedelic dose and are used for diverse goals such as improving mood, cognition, or creativity. Most clinical research has evaluated single or short courses of full doses, and observational surveys predominate for microdosing; neither approach reliably addresses rare or delayed adverse outcomes. The authors focus on valvular heart disease (VHD) as a theoretically plausible and serious long-term risk of chronic microdosing because drug-induced VHD has previously been linked to agonism at the serotonin 5-HT2B receptor. This review aims to assess available in vitro, animal, and clinical data on the 5-HT2B activity and VHD risk of five commonly microdosed compounds—LSD, psilocybin (psilocin), mescaline, DMT, and MDMA—and to apply established risk-assessment strategies to those data to determine what is known and what remains uncertain.

The paper is a narrative risk assessment that compiles and interprets in vitro, animal, and clinical evidence relevant to 5-HT2B-mediated VHD for LSD, psilocin (the active metabolite of psilocybin), mescaline, DMT, and MDMA. The extracted text does not clearly report a formal literature search strategy (databases, dates, or explicit inclusion/exclusion criteria), but the authors present data from radioligand-binding and functional assays, toxicology and repeat-dose animal studies, and available human clinical and observational studies. For in vitro assessment, the investigators reviewed ligand-binding affinities (K i or K d), functional activity assays (EC50 and E max across signalling readouts such as intracellular calcium, inositol phosphate accumulation, β-arrestin translocation, MAPK1/2 phosphorylation and NFAT activation), and results from human valvular interstitial cell (hVIC) proliferation assays. They describe the common practice of deriving a safety margin by comparing maximum unbound plasma concentration (C max) to binding affinity or potency (for example K i/C max), and discuss limitations that affect assay interpretation, including receptor reserve, functional selectivity (biased signalling), and variability between laboratories. Animal evidence was evaluated with attention to the strengths and limits of rodent toxicology for detecting valvular lesions. The authors summarise attempts at modelling drug-induced VHD in rodents and note problems such as anatomical differences, inconsistent histopathology, and the lack of a validated preclinical model. Clinical assessment emphasises echocardiography as the gold-standard method to detect early valvular changes and discusses the limitations of relying on adverse event reports or short-duration studies to detect VHD, given that valvular regurgitation may take months to years of exposure to emerge. Where clinical or animal data were sparse or absent, the authors relied on calculating in vitro-to-in vivo safety margins (K i or EC50 relative to expected plasma C max for typical microdoses) and on available pharmacokinetic data and reported microdose regimens to gauge potential exposure and risk. When metabolites were reported, their binding and functional profiles were considered where data existed.

Across the five compounds reviewed, the authors found that each can interact with the 5-HT2B receptor and, except where potency was too low or data were limited, act as partial agonists; mescaline was the exception in being insufficiently characterised. Key findings by compound follow. LSD: In transfected cell systems, LSD bound the human 5-HT2B receptor with K i/K d values between 0.91 and 30 nM, and binding to 5-HT2A ranged from equipotent to up to 10-fold stronger. Functional assays showed LSD to be a partial agonist at 5-HT2B with E max values of 13–73% and a marked bias toward β-arrestin2 signalling (EC50 for β-arrestin2 ≈ 0.68 nM versus 34 nM for calcium influx). Clinical microdosing pharmacokinetic data after a 10 µg dose reported median plasma C max values around 0.323 ng/mL (≈1.0 nM) and geometric mean 0.279 ng/mL (≈0.86 nM); using a plasma free fraction estimate of 0.1 yields a K i/C max safety margin near 10 relative to the lowest reported K i of 0.91 nM. No robust animal or clinical studies with adequate duration and cardiac assessments were identified; small or short-duration human studies and surveys reported no cardiac signals but were insufficient to detect VHD. Psilocybin/psilocin: Because psilocybin is rapidly converted to psilocin, in vitro data focus on psilocin. Psilocin bound the human 5-HT2B receptor with K i values of 4.6–8 nM, often showing higher affinity for 5-HT2B than 5-HT2A in transfected systems. Functional assays reported EC50s ranging from about 1.07 nM to 58 nM across different readouts, and E max values of 38–84%. Human PK after 15 mg psilocybin produced a psilocin geometric mean C max of 13 ng/mL (≈63.5 nM), so a typical 2 mg microdose was estimated to give a C max ≈8.4 nM. Using total concentrations and the lowest K i (4.6 nM) gives a safety margin of about 0.55, which the authors flag as concerning, particularly because some functional EC50s were lower than K i values. No clinical or animal studies with sufficient dosing duration and cardiac imaging were available to assess VHD risk. Mescaline: In vitro affinity and functional data were sparse and inconsistent; some studies suggested equipotent binding to 5-HT2A and 5-HT2B, while others showed minimal 5-HT2B binding or methodological limitations. Functional activation in calcium assays was not clearly demonstrated, possibly due to low potency and limited concentration ranges tested. New pharmacokinetic data showed geometric mean plasma C max values of 3,460 nM and 4,900 nM after 300 and 500 mg doses. Extrapolating dose-proportionally, a 40 mg microdose would be predicted to give a plasma C max ≈848 nM; compared to the lowest reported K i of 793 nM this yields a safety margin of ≈1.07 (total plasma), which the authors regard as minimal and concerning. No adequate clinical or animal cardiac assessments were identified. DMT: Binding K i values for human 5-HT2B ranged from 101 to 184 nM. Functional characterisation was limited; one intracellular calcium assay reported an EC50 of 3,400 nM with low intrinsic activity (E max 19%), but the assay has since been optimised and additional data are lacking. Repeat-dose rodent toxicology studies of oral ayahuasca up to relatively high multiples of ceremonial doses (including a 28-day study) did not reveal histological cardiac valve alterations, but the authors note that some known valvulopathogens also fail to produce VHD in rats. Human repeat-dose data adequate to assess VHD were absent. A small IV dose study that produced microdose-like effects yielded a plasma C max ≈10 ng/mL; assuming an unbound fraction of 0.677 gives a free C max of ≈36 nM and a safety margin ≈2.8 relative to the lowest K i of 101 nM. The extremely short half-life of inhaled or IV DMT produces only transient exposure, which may mitigate risk, but uncertainties about ayahuasca microdosing (MAO inhibitor dosing, dose-proportionality) prevent firm conclusions. MDMA: MDMA and its N-demethylated metabolite MDA both bind human 5-HT2B. MDA is more potent (K i lower than MDMA) and shows higher intrinsic efficacy in functional assays; reported EC50s were ≈2,000 nM for MDMA and ≈190 nM for MDA in PI hydrolysis assays, with MDA producing greater E max (90% vs 32%). Both MDMA and MDA stimulated hVIC mitogenic responses at 10 µM. Retrospective human data included an echocardiographic comparison of 29 MDMA users versus matched controls in which 28% of the MDMA group had echocardiographic signs of VHD versus 0% in controls; higher cumulative tablet counts correlated with more severe regurgitation. A case report also described drug-induced VHD with long-term MDMA use. For microdosing, reported doses vary widely; a 50 mg MDMA dose produced mean plasma C max values of 266 nM for MDMA and 28.5 nM for MDA. Using reported free fractions (0.57 for MDMA, 0.37 for MDA) yields free C max safety margins of ≈3.29 for MDMA against a K i of 500 nM and ≈8.6 for MDA against a K i of 91 nM. The authors note additional potential risk factors for MDMA, including elevated plasma serotonin and other cardiac toxicities.

The authors interpret their synthesis to mean that VHD is a plausible long-term risk from chronic microdosing for compounds that activate the 5-HT2B receptor, but the magnitude of that risk is currently uncertain. They emphasise that all five compounds examined show some ability to bind 5-HT2B and that, with the exception of mescaline (where data were limited), they act as partial agonists in in vitro systems. Variation in assay methods, receptor expression levels, and pathway-specific signalling bias make it difficult to identify which in vitro measures best predict clinical VHD. Tagen and colleagues position their findings relative to prior work by noting that previous commentaries focused mainly on LSD and psilocybin, whereas this review includes mescaline, DMT and MDMA and attempts to triangulate binding/potency data with pharmacokinetic estimates to derive safety margins. They conclude that LSD appears to have a reasonable but not robust safety margin at typical microdose exposures, psilocin (psilocybin’s active form) may present a higher relative risk given its higher 5-HT2B affinity and calculated C max estimates, mescaline’s low potency and limited data render its risk assessment inconclusive but potentially concerning on extrapolated plasma levels, and DMT’s short half-life affords some mitigation although data gaps—particularly for ayahuasca microdosing—remain. MDMA is singled out as having the strongest clinical signal for VHD risk, supported by retrospective human echocardiography data and pharmacology showing MDA is a potent 5-HT2B agonist that stimulates hVIC proliferation. They acknowledge several key limitations and uncertainties: the absence of standardised in vitro protocols and a consensus safety margin for 5-HT2B; potential over- or underestimation of potency due to receptor reserve in heterologous expression systems; incomplete characterisation of metabolites (notably 4-HIAA for psilocin and TMPAA/NAM for mescaline); the paucity of adequately powered, long-duration clinical trials with routine echocardiography; and the limited predictive validity of rodent models. The recent and heterogeneous nature of the microdosing phenomenon means that surveys may not capture long-term harms, and symptom-based surveillance is unlikely to detect early, often asymptomatic valvular changes. To reduce uncertainty, the authors recommend targeted steps: standardise and extend in vitro functional testing including assays more closely associated with mitogenic signalling (MAPK1/2 and NFAT), expand hVIC proliferation testing, and conduct placebo-controlled clinical trials lasting at least 6 months with serial echocardiography. They also suggest that animal models could be used to explore questions about intermittent versus daily dosing, while recognising the limitations of preclinical prediction for human VHD.

The authors conclude that valvular heart disease is a potential risk of chronic psychedelic and MDMA microdosing because these compounds can activate the 5-HT2B receptor. Among the substances reviewed, MDMA presents the clearest evidence of risk based on clinical case series and pharmacology, while LSD and psilocin show biochemical profiles that warrant caution. Mescaline and DMT have more uncertain risk profiles due to limited or inconsistent data; DMT’s short systemic exposure may reduce risk but is not definitive. Overall, data gaps prevent calculation of an absolute risk and justify further standardised in vitro work and adequately powered, long-duration human studies with echocardiography to better define the safety of microdosing.