Pharmacokinetics and pharmacodynamics of sublingual microdosed lysergic acid diethylamide in healthy adult volunteers

Microdosing with LSD refers to repeated administration of sub‑perceptual doses (commonly about 10 µg) that users report may improve mood, cognition and other symptoms. Earlier experimental work in healthy volunteers has shown dose-dependent effects of low doses (5–26 µg) on physiological and behavioural measures, but pharmacokinetic (PK) characterisation at microdose levels, especially for sublingual administration, is limited. There are no published population PK models for sublingual microdoses of LSD, and uncertainty remains about concentration–effect relationships, appropriate pharmacodynamic (PD) endpoints sensitive to microdoses, and the influence of metabolic genotype (notably CYP2D6) on exposure and subjective effects. Morse and colleagues set out to describe the pharmacokinetics and pharmacodynamics of a 10 µg sublingual LSD microdose in healthy adult male volunteers, develop and validate a high‑sensitivity LC‑MS/MS assay for low plasma concentrations (including the major metabolite O‑H‑LSD and iso‑LSD), and to explore whether CYP genotype and peripheral brain‑derived neurotrophic factor (BDNF) concentrations change acutely or after a 6‑week microdosing regimen. The study aimed to provide a population PKPD model that could inform dose selection, bioequivalence studies and future clinical development of LSD microdosing.

This was a Phase I, double‑blind, placebo‑controlled, parallel‑groups trial conducted at the University of Auckland Clinical Research Centre. Eighty healthy male volunteers aged 25–60 years were randomised 1:1 to receive either 10 µg sublingual LSD or inactive treatment (placebo). Key exclusions included elevated resting blood pressure, significant medical or psychiatric disorders, recent psychedelic use (past year) or any prior microdosing. Participants provided informed consent and the protocol was registered and ethically approved. On the first treatment visit (morning sessions), an intravenous cannula was placed and participants self‑administered 1 mL syringes containing either 10 µg LSD in distilled water (0.2% ethanol) or distilled water placebo held sublingually for 30 seconds before swallowing. A titration regimen (reducing by 5 µg then increasing by 1 µg steps as tolerated) was available for participants who reported overstimulation; six entered titration (five active, one placebo) and four active participants were subsequently withdrawn due to anxiety/overstimulation. Blood samples for PK analysis were collected at 30, 60, 120, 240 and 360 minutes post‑dose; plasma and serum for BDNF and DNA for genotyping were also collected at baseline and at a final visit after 6 weeks of home microdosing (only the first‑dose PK/PD data are the focus of this report). LSD and metabolites were quantified using a validated LC‑MS/MS method developed for low plasma concentrations; assay reproducibility was demonstrated with re‑analysis of samples showing ≤15% difference. PKPD modelling used nonlinear mixed‑effects methods (NONMEM v7.5.1) with one‑ and two‑compartment structures considered, first‑order absorption/elimination, allometric scaling of body weight (exponent 3/4 for clearances, 1 for volumes), and an effect‑compartment to allow for delay between plasma concentration and effect. PD endpoints were heart rate and a visual analogue scale (VAS) item capturing a subjective 'feel effect'. Residual error models included additive and proportional components, and model selection used objective function value changes and visual predictive checks; bootstrap (100 replications) assessed parameter uncertainty. Pharmacodynamic statistical analyses used linear mixed‑effects models (lmerTest in R) with Group (LSD vs placebo) and Time as fixed effects and participant as a random effect, with interest in the group × time interaction. CYP genotyping was performed on extracted DNA using Agena MassARRAY with the Veridose panels to assess CYP2D6, CYP2C19, CYP1A2, CYP2B6, CYP2C9 and CYP3A4 and to call activity scores per contemporary CPIC guidance; copy number variants (CNVs) were also assessed. Plasma and serum BDNF were measured by ELISA (R&D Systems Quantikine), samples assayed in duplicate, with outliers and values below limit of detection retested; intra‑ and inter‑assay variability were reported ≤6.1%.

Seventy‑nine participants contributed blood samples for PK analyses (one participant could not be cannulated). Using the LC‑MS/MS assay, LSD was quantified in 100% of collected samples from active participants and the intended sample collection rate for the active group was 96.5% (two missing samples at 120 and 360 minutes). A one‑compartment model provided the most parsimonious description of the sublingual 10 µg LSD PK data; a two‑compartment model reduced the objective function value (ΔOFV = 11.4, p = 0.003) but was judged over‑parameterised with poorly estimated intercompartmental parameters and no improvement in visual predictive checks. Final population parameter estimates (with reported variability) yielded a central volume of distribution around 32.9 L/70 kg and an elimination clearance approximately 7.78 L/h/70 kg. Derived metrics were: mean Cmax ≈ 0.20 µg/L (204 pg/mL; reported ±0.13), Tmax ≈ 1.51 h (±0.66), and elimination half‑life ≈ 3.08–3.09 h (±0.37). Allometric scaling by body weight was applied in the model and visual predictive checks and bootstrap resampling (100 replicates) were used to evaluate model performance and parameter uncertainty. Pharmacodynamic effects at this microdose were minimal. Heart rate and VAS 'feel effect' showed small maximal increases (<15% relative to baseline), which limited the ability to robustly characterise concentration–effect relationships; effect‑compartment modelling was used but estimated EMAX parameters remained small. Mixed‑effects analyses examined group × time interactions for PD measures but the low magnitude of changes and a relatively strong placebo response reduced discriminability between groups. The authors note a higher placebo response compared with some prior microdose studies and discuss potential reasons (parallel design, demographic differences, prior psychedelic experience). CYP genotyping in the active group (n = 40) showed 21/40 participants with CYP2D6 extensive metabolism, 13/40 intermediate metabolisers and no clearly classifiable poor or ultrarapid metabolisers based on core diplotypes; five participants had CNVs or hybrid alleles complicating phenotype assignment. Two participants were classified as intermediate‑weak (activity score 0.5) and both were among the four active participants withdrawn for overstimulation/anxiety during the home microdosing phase. Exploratory plots comparing concentration–time curves by genotype suggested qualitative effects for several enzymes: no clear CYP2C19 influence, possible lower concentrations in some CYP1A2 (rs762551) A/A carriers consistent with induction, and trends in expected directions for CYP2B6 and CYP2C9 variants though numbers were small. The dataset lacked sufficient numbers of poor or ultra‑rapid CYP2D6 phenotypes to permit formal covariate modelling of activity scores. BDNF analyses: for plasma, 214/240 (89%) observations were analysable after excluding missing samples, below‑limit values and outliers. Linear mixed models showed no group × time interaction at 6 h (t = 0.14, p = 0.89) or at 43 days (t = 0.40, p = 0.68). There was an effect of time at 6 h across groups (t = 2.15, p = 0.03) but no group effect. For serum BDNF, 224/240 (93%) samples were analysable; models showed no interaction at 6 h (t = 0.234, p = 0.81) but a statistically significant interaction at 43 days (t = 2.45, p = 0.016). The authors attribute that interaction to a low baseline mean in the placebo group rather than an increase in the LSD group; reported means were: placebo pre‑mean 26,496 (SD 6,676) pg/mL, placebo post‑mean 28,528 (6,661) pg/mL, LSD pre‑mean 28,412 (6,284) pg/mL, LSD post‑mean 28,482 (5,767) pg/mL. Overall, the study found no convincing evidence that a single 10 µg microdose or a 6‑week microdosing regimen altered peripheral BDNF in healthy male volunteers.

Morse and colleagues interpret their findings as providing a validated population PKPD framework and a high‑sensitivity LC‑MS/MS assay suitable for microdosing studies. The one‑compartment PK description, estimated clearance and volume of distribution, and derived Cmax, Tmax and half‑life are consistent with prior microdose work and earlier reports of LSD PK. The presented assay detected LSD and O‑H‑LSD in plasma at microdose concentrations and was reproducible across re‑analysed samples. Pharmacodynamic effects at the 10 µg sublingual dose were small, with heart rate and self‑reported 'feel effect' VAS changes under 15%, limiting the capacity of the PKPD model to characterise robust concentration–effect relationships. The investigators highlight a relatively strong placebo response and note that the parallel‑groups design may have improved blinding compared with crossover designs, potentially contributing to smaller between‑group PD differences. Demographic differences and prior psychedelic experience compared with other studies are also discussed as factors that may explain differing placebo and active responses. Genotyping results suggest that CYP2D6 and other CYP enzymes may influence LSD concentrations qualitatively, but the sample contained no clearly classifiable poor or ultra‑rapid CYP2D6 metabolisers and only two intermediate‑weak participants. The two intermediate‑weak participants were among those withdrawn for overstimulation in the home‑dosing phase, a finding the authors consider potentially relevant if replicated in larger samples: CYP2D6 activity scores might be useful biomarkers to guide dosing or to stratify risk of adverse effects. The authors caution that multiple CYPs and CNVs contribute to a complex genotype profile and that larger studies are needed to quantify the average contribution of specific SNPs and CNVs to plasma exposure. Regarding BDNF, the study found no evidence that a single 10 µg microdose or a 6‑week regimen increased peripheral BDNF in healthy men. The authors contrast this null result with prior small studies reporting increases but note those earlier studies had substantial missing data and small effective sample sizes. High baseline variability in peripheral BDNF and differences between plasma and serum matrices are discussed; most peripheral BDNF is platelet‑stored and serum processing releases platelet BDNF, so plasma (with fewer platelets) might better reflect acute effects, but the evidence remains inconsistent. The investigators therefore recommend larger, within‑subject designs and careful sample handling to probe peripheral BDNF as a biomarker. Key limitations acknowledged include the exclusive inclusion of males (restricting generalisability and not accounting for menstrual cycle effects), limited sampling during the elimination phase and lack of diversity in CYP2D6 phenotypes (few or no poor/ultrarapid metabolisers and several CNV/hybrid genotypes complicating classification). The authors call for further studies to identify PD measures sensitive to microdoses, to include females with attention to menstrual phase and life stage, and to extend PKPD modelling across formulations, doses and populations so that dosing regimens can be rationally developed for potential therapeutic use.

This study delivers a validated LC‑MS/MS assay for low plasma concentrations of LSD, a population one‑compartment PK model for a 10 µg sublingual microdose, and preliminary PKPD and pharmacogenetic observations. No convincing acute or 6‑week effects on peripheral BDNF were detected in healthy male volunteers. The authors conclude that CYP2D6 activity score merits further investigation in larger samples as a potential biomarker of response or adverse effects, and that future work should prioritise sensitive PD metrics, inclusion of females with menstrual considerations, and larger pharmacogenomic studies to inform dose titration and personalised microdosing regimens.