Pharmacokinetics and pharmacodynamics of lysergic acid diethylamide in healthy subjects

Lysergic acid diethylamide (LSD) is a prototypical classic hallucinogen with renewed interest for basic pharmacology, recreational use and potential therapeutic applications. Earlier human data on LSD pharmacokinetics are sparse: small intravenous and limited oral studies provided rough estimates of elimination half-life and time courses, but systematic concentration–time profiles after controlled oral dosing and linked pharmacokinetic–pharmacodynamic (PK–PD) analyses have been largely missing. In particular, prior functional imaging and other experimental studies often did not measure plasma LSD, leaving the relationship between exposure and effects at the time of outcome assessment uncertain. Dolder and colleagues set out to characterise the plasma pharmacokinetics of oral LSD at two commonly used doses (100 µg and 200 µg) and to quantify the exposure–response relationship for subjective and autonomic effects. The study combined data from two double-blind, placebo-controlled, cross-over trials to (1) describe concentration–time profiles and compartmental pharmacokinetic parameters, (2) derive EC50 (half-maximal effect) and effect-site equilibration parameters using PK–PD link modelling, and (3) examine whether plasma concentrations predict between-subject differences in pharmacodynamic responses at matched time points.

The analysis pooled pharmacokinetic and pharmacodynamic data from two previously conducted double-blind, placebo-controlled, cross-over studies. Study 1 randomised 24 healthy participants to oral LSD 100 µg and placebo in balanced order, and Study 2 randomised 16 participants to LSD 200 µg and placebo; washout periods were at least 7 days. The trials were registered (NCT02308969, NCT01878942). Forty healthy volunteers were recruited from a university setting; mean ages and body weights were reported separately for each study. Inclusion and exclusion criteria were identical across studies and excluded major psychiatric disorders, recent or heavy illicit drug use, significant medical illness, pregnancy, and use of interacting medications. Urine drug tests and pregnancy tests were performed at screening and before sessions. Experimental sessions lasted 25 h and were conducted in a quiet hospital room with one investigator present. LSD or placebo was administered orally at 09:00. Standardised meals were provided at 13:30 and 17:30. Blood sampling for plasma LSD concentrations occurred predose and at multiple time points up to 24 h; sampling schedules differed slightly between studies (some early time points not collected in the 100 µg study). Plasma was stored frozen and later analysed by validated liquid chromatography–tandem mass spectrometry methods; lower limits of quantification were 0.05 ng/mL for the 100 µg study and 0.01 ng/mL for the 200 µg study. Subjective effects were measured repeatedly with visual analogue scales (VASs; 0–100%) for ‘‘any drug effect’’, ‘‘good drug effect’’, and ‘‘bad drug effect’’. Vital signs (systolic and diastolic blood pressure, heart rate, tympanic temperature) were recorded at frequent intervals. Pharmacokinetic parameters were estimated using compartmental modelling in Phoenix WinNonlin 6.4. A one-compartment model with first-order input and elimination and no lag time was selected on the basis of model fit and Akaike information criterion. Predicted plasma concentrations from the PK model were used as inputs to a PK–PD link model employing a first-order equilibration rate constant (keo) to account for plasma–effect-site delay. A sigmoid Emax model (EC50, Emax, Hill coefficient) described concentration–effect relationships; upper and lower Emax bounds were set pragmatically for the VAS scales and physiological outcomes. Effects after LSD were analysed as differences from placebo at matched time points; Pearson correlations and multiple regression (to test sex and body weight effects) assessed associations across subjects.

Pharmacokinetics: Observed and model-predicted plasma concentration–time profiles were consistent with a one-compartment, first-order elimination process up to about 12 h. Geometric mean Cmax values were approximately 1.3 ng/mL for 100 µg (Tmax ~1.4 h) and 3.1 ng/mL for 200 µg (Tmax ~1.5 h). The modelled plasma elimination half-life was about 2.6 h for both doses. Dose-normalised Cmax and area under the concentration–time curve (AUC) did not differ statistically between dose groups, indicating dose-proportional kinetics over the tested range. Plasma concentrations became unquantifiable in some participants at later time points; interindividual variability in concentrations was greater at the 100 µg dose. Subjective and autonomic effects: LSD robustly increased ‘‘any drug effect’’ and ‘‘good drug effect’’ VAS ratings in almost all subjects; ‘‘bad drug effect’’ increased only moderately and transiently in a subset of participants. For the 100 µg dose, mean onset of ‘‘any drug effect’’ was 0.8 ± 0.4 h, mean time to peak effect 2.8 ± 0.8 h, offset 9.0 ± 2.0 h, and mean duration 8.2 ± 2.1 h. For the 200 µg dose, onset was 0.4 ± 0.3 h, peak 2.5 ± 1.2 h, offset 11.6 ± 4.2 h, and mean duration 11.2 ± 4.2 h. Heart rate, blood pressure and body temperature increased modestly and to similar extents at both doses. PK–PD modelling and exposure–response: When analysed within subjects over time, a close relationship existed between plasma LSD concentrations and subjective and autonomic effects, with moderate counterclockwise hysteresis indicative of a short plasma–effect-site delay. Effect-site equilibration half-lives (t1/2 keo) for subjective and cardiovascular measures were typically on the order of tens of minutes (reported ranges around 13–48 min for several outcomes), whereas the thermogenic response showed a longer lag (reportedly ~136 min for body temperature at 200 µg). Sigmoidal EC50 estimates for subjective effects fell in the approximate range 0.67–2.5 ng/mL, with lower EC50 values for ‘‘good drug effect’’ and ‘‘any drug effect’’ than for ‘‘bad drug effect’’. Some participants reported negligible ‘‘bad drug effect’’ (ratings <5%), preventing EC50 estimation for those individuals. Across-subject analyses showed generally poor correlations between plasma concentrations and pharmacodynamic effects at matched time points within dose groups; for example, predicted Cmax did not correlate with maximal ‘‘any drug effect’’ within the 100 µg or 200 µg groups, although a modest correlation appeared in the pooled sample (Rp = 0.38, p < 0.05, n = 40). AUC of LSD did not correlate with AUC of ‘‘any drug effect’’. Multiple regression including concentration, body weight and sex did not identify predictors of effect variability across subjects. No evidence of acute tolerance (clockwise hysteresis) was observed.

Dolder and colleagues interpret the findings as demonstrating dose-proportional oral pharmacokinetics for LSD at 100 µg and 200 µg, with first-order elimination up to 12 h and a plasma half-life of about 2.6 h. The one-compartment model provided an adequate fit, and differences from previously published 200 µg results were attributed to more sensitive analytical methods and different modelling approaches in the present reanalysis. The authors highlight that the effect time course lags behind peak plasma concentrations, represented in the model by the effect-site equilibration parameter; this lag explains why subjective peak effects occurred roughly 0.6–1.1 h after Cmax on average and why MRI scanning times for peak effects should be chosen accordingly. In terms of pharmacodynamics, subjective ‘‘any drug’’ and ‘‘good drug’’ effects were strongly related to within-subject concentration changes, with EC50 estimates in the sub- to low-ng/mL range. The lack of clockwise hysteresis indicates no acute tolerance during single administrations at these doses. The limited or absent correlations between plasma concentrations and effects across subjects at peak response are interpreted as a ceiling effect: when doses produce concentrations above EC50 in most participants, between-subject variability in effect is low and cross-sectional concentration–effect correlations are weak. The authors note implications for interpreting between-subject associations in fMRI and other studies that do not measure plasma LSD. The study limitations acknowledged include that the two doses were evaluated in separate participant samples rather than within the same individuals, and that plasma samples were analysed in different laboratories. Nonetheless, the investigators consider the pharmacokinetic and PK–PD results to be broadly consistent across studies and informative for experimental and clinical work, including dosing and timing considerations in imaging and therapeutic studies.

The investigators conclude that oral LSD at 100 µg and 200 µg shows dose-proportional pharmacokinetics and first-order elimination up to 12 h, with a plasma half-life of approximately 2.6 h. Within individuals, subjective and autonomic effects closely track plasma concentrations with a short effect-site lag (moderate counterclockwise hysteresis), and EC50 values for subjective effects lie in the sub- to low-ng/mL range. By contrast, plasma concentrations generally do not predict between-subject differences in robust peak effects within a dose group, a consideration that affects interpretation of cross-sectional correlations in imaging and other experimental studies. The authors emphasise that these pharmacokinetic data are essential for interpreting clinical, imaging and intoxication findings involving oral LSD.