Clin Pharmacokinet

DOI 10.1007/s40262-017-0540-6

ORIGINAL RESEARCH ARTICLE

Pharmacokinetics of Escalating Doses of Oral Psilocybin

in Healthy Adults
Randall T. Brown1 • Christopher R. Nicholas1 • Nicholas V. Cozzi2 •
Michele C. Gassman3 • Karen M. Cooper3 • Daniel Muller4 • Chantelle D. Thomas1 •

Scott J. Hetzel5 • Kelsey M. Henriquez3 • Alexandra S. Ribaudo3 • Paul R. Hutson3

Ó Springer International Publishing Switzerland 2017

Abstract determined using both compartmental (NONMEM) and

Introduction Psilocybin is a psychedelic tryptamine that noncompartmental (WinNonlin) methods.
has shown promise in recent clinical trials for the treatment Results No psilocybin was found in plasma or urine, and
of depression and substance use disorders. This open-label renal clearance of intact psilocin accounted for less than 2%
study of the pharmacokinetics of psilocybin was performed of the total clearance. The pharmacokinetics of psilocin were
to describe the pharmacokinetics and safety profile of linear within the twofold range of doses, and the elimination
psilocybin in sequential, escalating oral doses of 0.3, 0.45, half-life of psilocin was 3 h (standard deviation 1.1). An
and 0.6 mg/kg in 12 healthy adults. extended elimination phase in some subjects suggests
Methods Eligible healthy adults received 6–8 h of hydrolysis of the psilocin glucuronide metabolite. Variation
preparatory counseling in anticipation of the first dose of in psilocin clearance was not predicted by body weight, and
psilocybin. The escalating oral psilocybin doses were no serious adverse events occurred in the subjects studied.
administered at approximately monthly intervals in a con- Conclusions The small amount of psilocin renally excreted
trolled setting and subjects were monitored for 24 h. Blood suggests that no dose reduction is needed for subjects with mild–
and urine samples were collected over 24 h and assayed by a moderate renal impairment. Simulation of fixed doses using the
validated liquid chromatography-tandem mass spectrometry pharmacokinetic parameters suggest that an oral dose of 25 mg
(LC–MS/MS) assay for psilocybin and psilocin, the active should approximate the drug exposure of a 0.3 mg/kg oral dose
metabolite. The pharmacokinetics of psilocin were of psilocybin. Although doses of 0.6 mg/kg are in excess of
likely therapeutic doses, no serious physical or psychological
events occurred during or within 30 days of any dose.
Clinical Trials Identifier NCT02163707.
& Paul R. Hutson
paul.hutson@wisc.edu
1
Department of Family Medicine, School of Medicine and Key Points
Public Health, University of Wisconsin-Madison, Madison,
WI, USA Psilocybin, as its active metabolite psilocin,
2
Department of Cell and Regenerative Biology, School of demonstrates linear pharmacokinetics over the
Medicine and Public Health, University of Wisconsin- 0.3–0.6 mg/kg oral dose range tested.
Madison, Madison, WI, USA
3
School of Pharmacy, University of Wisconsin-Madison, 777
Less than 2% of the psilocin in plasma is excreted in
Highland Avenue, Madison, WI 53705, USA urine in that form, suggesting minimal effect of renal
4
Department of Medicine, School of Medicine and Public
dysfunction in elimination of the active metabolite.
Health, University of Wisconsin-Madison, Madison, WI, A fixed oral dose of 25 mg is expected to
USA
approximate the area under the concentration–time
5
Department of Biostatistics and Medical Informatics, School curve and concentration profile of the 0.3 mg/kg oral
of Medicine and Public Health, University of Wisconsin-
Madison, Madison, WI, USA dose used in this study.
 R. T. Brown et al.

1 Introduction dosing to its active metabolite psilocin, which demon-

strated a half-life of 163 min, and the presence of
approximately 67% of the psilocin present as the
Psilocybin (4-phosphoryloxy-N,N-dimethyltryptamine) is a
glucuronide.
naturally-occurring tryptamine contained within Psilocybe
The primary objective of this pharmacokinetic analysis
mexicana Heim and other species of psychoactive mush-
was to develop a population pharmacokinetic model for
rooms used for centuries by native cultures for shamanic or
psilocybin administered to healthy adults who received
spiritual purposes. Psilocybin produces remarkable effects
sequential, escalating oral doses. Secondarily, we sought to
on consciousness, often described as ‘psychedelic’ or
identify covariates that are predictive of the pharmacoki-
‘hallucinogenic’ [1]. The effects have been characterized
netic behavior of oral psilocybin, such as renal function
as an intense dream-like state with colorful visual illusions,
and dose, and to characterize adverse events (AEs) asso-
changes in auditory, tactile, olfactory, gustatory, and
ciated with these doses. The earlier work of Hasler et al. [7]
kinesthetic perceptions, altered perceptions of time and
used a mass spectrometry assay that had a higher lower
space, changes in body image and sensations, and intense
limit of quantitation (LLOQ) than was available for this
mood changes.
study (0.8 vs. 0.5 ng/mL), and carried out their pharma-
Interest in using psilocybin as a therapeutic agent has
cokinetic sampling for only 390 min after oral adminis-
been rekindled based on clinical data indicating that, with
tration of psilocybin. In addition to extending the sampling
appropriate preparation and dosing conditions, psilocybin
duration and determining the renal excretion of psilocin,
has safety and efficacy in alleviating existential anxiety and
this study was performed following current Good Clinical
depression associated with a diagnosis of terminal disease
Practices (cGCP) and current Good Laboratory Practices
[2]. Concurrent placebo-controlled, phase II clinical trials
(cGLP) to support future attempts to obtain US FDA
at New York and Johns Hopkins Universities found clini-
approval of psilocybin as a drug.
cally significant improvement in anxiety and depression in
patients with incurable disease even 6 months after a single
oral dose of psilcocybin [3, 4]. In addition, Carhart-Harris,
et al. reported similar results in an open-label trial of 2 Materials and Methods
patients with treatment-resistant depression in which two
oral doses of psilocybin 1 week apart resulted in an 2.1 Study Design and Subject Selection
improvement in depression in all subjects, with a sustained
benefit measured 3 months after treatment in 67% of these After approval by the University of Wisconsin-Madison
previously refractory patients [5]. Health Sciences Institutional Review Board, 12 healthy
Following oral or parenteral administration, psilocybin adults were recruited for the study by word of mouth and
is rapidly dephosphorylated to psilocin [6, 7], and it is from inquiries from the ClinicalTrials.gov website
believed that it is the psilocin that is responsible for the (NCT02163707). Subjects were recruited and administered
psychoactive effects of ingested psilocybin. Binding and over a 15-month period. Subjects were first screened by
partial agonist activity at serotonin 5 HT2A receptors are telephone for general eligibility, and were then met in
required for the manifestation of psychoactive effects from person for an in-depth screening visit. Eligible subjects
psilocin and other tryptamines [8], although other serotonin were required to have had at least one substantial prior
and nonserotonergic receptors appear to be involved in the experience with a psychedelic due to concerns about the
psychoactive effects of these substances [1, 9–16]. acceptance to the repeated blood sampling from an
One requirement for developing psilocybin as a thera- indwelling catheter for this pharmacokinetic study during
peutic agent will be an assessment of the pharmacokinetic the peak effects of psilocybin, which also required shifting
profiles of psilocybin and psilocin in humans. Two studies to phlebotomy for each sample if the catheter patency was
have previously been published regarding the human lost. No subject had used a psychedelic within 1 month of
pharmacokinetics of orally administered psilocybin. In one accrual to the study and all had negative urine drug screen
study, six volunteers were administered a single oral dose results on each day of psilocybin dosing.
of psilocybin of approximately 0.2 mg/kg, and plasma Subjects underwent a clinical interview with a licensed
concentrations of psilocin were determined over a period of clinical psychologist using the Structured Clinical Inter-
6.5 h [7]. In the second study, eight subjects received a view for DSM-4 (SCID) [18] and were excluded if they
single oral dose of psilocybin of approximately 0.2 mg/kg, met criteria for substance use disorder, seizure, or a history
and plasma and urine concentrations of psilocin were of bipolar disorder, psychosis, anxiety disorder, or major
assessed over a 24-h period [17]. These studies demon- depressive disorder within the past 5 years. A first-degree
strated the immediate hydrolysis of psilocybin after oral relative with bipolar or psychotic spectrum diagnoses also
Pharmacokinetics of Oral Psilocybin

rendered the subject ineligible. Tobacco users not on a allowed a standardized breakfast in the CRU, and ingested
nicotine patch were excluded, as were individuals who the psilocybin capsule with 360 mL of water.
would typically need to take medications within the 8-h Anticoagulated (K2 EDTA) blood samples for determi-
period of drug action. Subjects were also excluded if they nation of plasma psilocybin and psilocin concentrations
were taking antidepressants or monoamine oxidase were collected from subjects just prior to each dose and at
inhibitors. 0.25, 0.5, 0.75, 1, 2, 3, 4, 6, 8, 12, 18, and 24 h postdose.
Each subject met with a trained pair of guides for 6–8 h Urine was collected for 24 h after the dose with the use of a
of preparatory counseling prior to receiving the first dose of commode insert. To stabilize psilocin, 5% ascorbic acid
psilocybin. The same guide pair (male and female dyad) was added to each plasma and urine aliquot to make a final
attended the subjects during each of their 8-h psilocybin 25 mM ascorbate concentration [17]. After addition of the
sessions. The preparatory counseling sessions involved a ascorbic acid solution, the aggregated urine collection was
review of relevant personal history, meditation and stored in crushed ice until aliquots were removed and
grounding exercises, preferred method for initiating com- frozen at the end of the 24-h collection period. Aliquoted
munication (e.g. requesting assistance, alerting participant and stabilized plasma and urine samples were stored in a -
to an impending blood sample), and strategies for opti- 70 °C freezer until thawed for assay.
mizing the potential personal benefit of the psychedelic
experience. The same guide dyad met with the subject the 2.2 Liquid Chromatography–Tandem Mass
morning after each dosing session for debriefing and inte- Spectrometry Analysis
gration of the experience. Separate, trained study staff
performed the blood sampling, collection of vital signs, and The analyses for psilocybin and psilocin were performed
reattachment and operation of a laptop-based 12-lead by Covance Laboratories (Madison, WI, USA) following
electrocardiograph (CardioCard, Nasiff Associates, Central cGLP methods. The assay included a solid-phase extraction
Square, NY, USA). step using a C18 matrix, with elution performed by a Luna
Synthetic psilocybin was prepared under appropriate Phenyl-Hexyl, 50 9 2 mm, 5 lm particle size column.
federal and state controlled substance permits. A certifi- The HPLC assay used an ammonium formate buffer
cate of analysis was prepared by the Zeeh Pharmaceutical (10 mM) and a gradient from 2 to 95% acetonitrile. Plasma
Experiment Station housed in the University of Wiscon- and urine samples were treated with 1:1 methanol:ace-
sin-Madison School of Pharmacy. The analysis included tonitrile to precipitate protein, and after centrifugation of
purity by high-performance liquid chromatography the deep-well 96-well plates an aliquot was transferred to a
(HPLC), water content, and demonstration of the absence new plate, dried under nitrogen, and reconstituted for
of residual organic solvents or metals. The purity of the injection. Psilocybin and its metabolite psilocin were
psilocybin was consistently over 99%, with approxi- quantified using mass-spectrometry [19, 20]. The Sciex
mately 1–6% residual water. Using actual body weight API spectrometer used positive ion electrospray at 1200 V
from their most recent visit or treatment, individual and 600C, with nitrogen gas used throughout. No parent
psilocybin doses of 0.3, 0.45, or 0.6 mg/kg were prepared psilocybin was found in any urine or plasma sample. The
1–4 days prior to the date of the intended dose, and were lower limit of quantitation (LLOQ) for psilocin in plasma
corrected for water content. The psilocybin was added to and urine was 0.5 and 5.0 ng/mL, respectively. Loss of
one-half of an opaque #0 methylcellulose capsule, with psilocin at room temperature for 24 h and over five freeze–
lactose USP completing the fill of that capsule half. The thaw cycles was less than 3%.
study dose was not masked.
At a minimum of 4-week intervals, subjects received 2.3 Noncompartmental Pharmacokinetics
single, oral doses of psilocybin at escalating doses of 0.3,
0.45, or 0.6 mg/kg. On the morning of each dosing day, the WinNonlin (v6.4; PharSight, Certara Corp, Princeton, NJ,
subject arrived at the University of Wisconsin Clinical USA) was used to determine maximum concentration
Research Unit (CRU) for placement of electrocardiography (Cmax), time to reach Cmax (Tmax), and area under the
(ECG) electrodes, collection of vital signs, urine drug concentration–time curve from zero to 24 h (AUC24) of
screen testing, pregnancy testing (if applicable), and psilocin in plasma. AUC was determined using the default
placement of the intravenous catheter. The subject was trapezoidal-up, log-trapezoidal-down approach.
then walked approximately 500 meters to the study room in
the School of Pharmacy. After approximately 15 min of 2.4 Population Pharmacokinetic Modeling
centering meditation, the subject emptied their bladder, a
baseline 12-lead ECG was obtained along with vital signs, Model building and evaluation was conducted using
and the baseline blood sample was collected. Subjects were NONMEM Version 7 level 3 (NONMEM; ICON
 R. T. Brown et al.

Development Solutions, Ellicott City, MD, USA), and 2.4.2 Covariate and Error Model Development
followed standard model building approaches to define the
structure, intersubject variability, and covariate depen- Covariates that significantly contributed to the model when
dence of the model [21, 22]. Modeling was performed on a added singly were sought in a manner similar to that used to
Dell Inspiron 6420 i7 laptop running the current GFortran develop the structural model. These covariates were then all
compiler. R version 3.2.3 was used for statistical modeling added to the model and individually removed to determine
and graphics in combination with XPOSE4 [23, 24]. Wings their impact on the model. In addition to reductions in the
for NONMEM was used as an interface for the model fit- objective function, the residual plots of NPDE and ETA
ting and bootstrap simulations [25]. versus covariate were inspected for evidence of a covariate
effect. Furthermore, available demographic and laboratory
2.4.1 Base Model Development covariates were added to the structural model to determine if
the objective function and residual plots were improved.
A value of 0.71863 was given to F1 in the NONMEM Continuous covariates such as weight were normalized
model to assume that all psilocybin had been converted to to the median of the study population and included as a
psilocin, and to relate the measured amount of psilocin to power function, where P1 and P2 are fixed effect parame-
the equivalent molar amount of psilocybin. The resulting ters, and Rref is a reference value of the covariate (Eq. 1)
clearance and volume terms from the model still include an affecting the typical value (TV) of a given parameter:
unknown extent of conversion of the psilocin to its glu-  P2
curonide or other metabolites. Psilocin concentration data R
PTV ¼ P1 ð1Þ
were log-transformed. Rref
The subroutines ADVAN2, ADVAN4, and ADVAN6 Categorical covariates of sex, race, and dose number were
were used to model one-, two-, and three-compartment tested using a proportional model, where sex is 0 or 1 (male,
models, respectively, and the estimation method used first- female) and race or dose number (e.g. 1, 2, or 3) activated
order conditional estimation (FOCE) with interaction. potential parameters associated with that covariate, where P1
Covariates such as hepatic and renal function and weight and P2 are fixed-effect parameters (Eq. 2).
were tested against both clearance and distribution volume.
In addition to the use of the initially reported data, the M3 PTV ¼ P1 ð1 þ RP2 Þ ð2Þ
method was applied to plasma psilocin concentrations that Alternatively, this relationship was tested using the
were below the LLOQ (0.5 ng/mL) [26]. Between-subject following relationship:
(intraindividual) variability (IIV, ETA) was assessed on
each pharmacokinetic parameter using an exponential PTV ¼ P1 RP2 ð3Þ
model. Residual error was modeled as an additive error on Creatinine clearance was measured with a monitored 24-
the log scale, resulting in a proportional error model, h urine collection that was begun just prior to each dose of
although a combined proportional and additive error model psilocybin. Five percent (v/v) of 1 M ascorbic acid in water
was tested with less satisfactory results. was added immediately to each urine collection fraction to
Selection of the structural model was made by inspec- stabilize the psilocin. After subtracting the volume of
tion of residual error plots, normalized prediction distri- ascorbic acid added to the urine, the creatinine clearance
bution error (NPDE) distributions, and visual predictive was determined from Eq. 4:
check (VPC), as well as by minimization of the objective
function. A decrease in the objective function value of CLCr (mL/min)
C3.84 was considered a significant (p \ 0.05) improve- Urine creatinine ðmg=dLÞ  Urine volume ðmLÞ
¼
ment by the addition of a covariate term. Model selection Serum creatinine (mg/dL)  Collection duration (1440 min)
also included consideration of the condition number, cal- ð4Þ
culated as the square root of the ratio of the highest and
Body surface area (BSA) was estimated from the Mosteller
lowest values of the Eigen matrix from the covariance step.
equation [27], ideal body weight (IBW) was calculated from
Although a condition number less than 20 was sought,
the method of Devine [28], and lean body weight (LBW) was
condition numbers less than 100 were considered to have
calculated from the method of Janmahasatian [29].
acceptable levels of colinearity.
Given a prolonged peak and suggestion of a biexpo-
nential decay curve in some, but not all, dosing sessions, 2.4.3 Model Evaluation
conversion to a psilocin glucuronide pool was also tested.
This allowed recirculation of the psilocin glucuronide from One thousand bootstrap runs were performed using the
the plasma to the liver with hydrolysis and reabsorption. NMBS routine in Wings for NONMEM using the
Pharmacokinetics of Oral Psilocybin

parameters from the final model to provide 95% confidence Table 1 Categorical demographics
intervals (CIs) for the parameters. Covariate factors that All subjects
had 95% CIs that included null were removed and the
bootstrap rerun [30]. Additionally, a VPC [31, 32] was N %
conducted to compare the distribution of simulated obser- Total 12 100
vations from the final model with those obtained from the Sex
original data. In this visual check, the concentration–time Male 10 83
profiles were simulated using the final model parameters. Race
The VPC was stratified by the three administered dose White 11 92
levels. In addition to the plots of IIV (ETA), the NPDE was Native American 1 8
examined and plotted against covariates [33]. Hispanic 0 0

2.4.4 Renal Clearance

available. One subject was removed from the study and
Renal clearance was determined from the ratio of the replaced because no blood samples could be obtained from
amount of excreted psilocin in the 24-h urine collection the indwelling catheter or venipuncture at any timepoint
divided by the plasma psilocin AUC over the 24-h col- after the first dose despite being normotensive. One subject
lection period. Measured creatinine clearance was also received only one of the three planned doses due to ‘white-
tested as a covariate for clearance in the NONMEM model. coat’ hypertension. A third subject received only two doses
of psilocybin because of difficulty getting the time off from
2.4.5 Flat Dose Simulation work to participate in the study (Tables 1, 2).

Consideration is being given to the use of a flat dose of 3.1 Structural Model
psilocybin for future studies instead of the current, indi-
vidualized dosing of the drug (e.g. 0.3 mg/kg). Using the The final model for psilocin in serum following oral
parameters of the final model for psilocybin, a VPC was administration of psilocybin capsules was a one-compart-
performed with simulation of the concentrations expected ment model with linear clearance and linear absorption.
from a single 20 and 25 mg dose, and was compared with The model was parameterized in terms of absorption rate
the actual concentrations observed after the 0.3 mg/kg constant (Ka), clearance (CL/F), and volume of distribu-
dose. The psilocin AUC and calculated Cmax and Tmax tion of the central compartment (V1/F). Although the
arising from the VPC simulation of both the 0.3 mg/kg and prospective model-building criteria slightly favored the
25 mg doses were compared using an unpaired t test. selection of a two-compartment model, the plots of con-
centration versus time for some subjects appeared, by
2.5 Adverse Events inspection, to favor a one-compartment model. It was
considered that the sensitivity of the assay might have
All expected and unexpected AEs occurring from the time limited the ability of the model to distinguish a two-com-
of enrollment into the study through the 30-day visit fol- partment model; however, inclusion of the below limit of
lowing the last dose were recorded and will be reported in
detail in a subsequent manuscript. Severity of the AEs was Table 2 Summary of baseline continuous covariates for all subjects
graded using the Common Terminology Criteria for (N = 12)
Adverse Events (CTCAE) version 4 (http://evs.nci.nih.gov/ Units Mean Median Range
ftp1/CTCAE/). Laboratory values that were out of normal
Age years 43 43 24–61
limits were recorded as AEs, but determination of clinical
Height cm 179.1 177.5 169.3–187.7
significance and attribution was made by the study
Weight kg 78.1 71 60.9–119.8
physician.
BMI kg/m2 24.2 23.5 19.4–34.0
BSA m2 1.96 1.86 1.73–2.50
3 Results IBW kg 73.4 72.8 60.8–82.0
LBW kg 60.0 56.5 51.1–79.7
The final dataset contained 353 evaluable measurable CLCr mL/min 126 123 71–177
plasma psilocin concentration observations from 12 sub- CLCr/1.73 m2 mL/min/1.73 m2 73 71 41–102
jects. When values below the LLOQ were included for the BMI body mass index, BSA body surface area, IBW ideal body weight,
M3 method, an additional 48 plasma samples were LBW lean body weight, CLCr creatinine clearance
 R. T. Brown et al.

quantitation (BLOQ) plasma psilocin concentrations in the

range of 0.25–0.5 ng/mL (LLOQ) using the M3 method did
not improve the fit of the model and was not used in
covariate screening. For those subjects for whom a two-
compartment model provided a better fit than a one-com-
partment model, the volume of the peripheral compartment
was over 20,000 L.
As an alternative to a second tissue compartment, the
inclusion of a compartment for psilocin glucuronide
improved the model. The inclusion of a bile-gut transit
process did not improve the model over the addition of rate
constants to a ‘compartment’ of psilocin glucuronide.
Stepwise forward and backward removal of covariates
suggested no effect of weight on CL/F, but normalized
dose, total bilirubin, and albumin appeared to improve the
fit of CL/F, and normalized weight affected apparent vol-
ume of distribution (V/F). When the bootstrap of this
combined model was performed, the 95% CIs for these
covariate factors included null, suggesting that their
inclusion in the model was not needed. Additionally, a
bootstrap of the model with and without normalized weight
as a covariate of V/F did not significantly alter the esti-
mates for V/F or CL/F. The final model therefore included
no covariates, but did incorporate a compartment in which Fig. 1 Psilocybin pharmacokinetic model. Ka first-order rate
constant
psilocin glucuronide could be formed and hydrolyzed to
reform psilocin. This psilocin glucuronide compartment
provided the allowance for apparent biexponential decay of 3.3 Noncompartmental Analysis
psilocin concentrations noted in some subjects. Inclusion of
renal function as a covariate of the loss of psilocin glu- The pharmacokinetics of psilocin over a twofold dosing
curonide did not improve the fit of the model. range were linear. This was indicated both by the structural
Between-subject variability was applied to both the model and its diagnostic plots, and by the noncompart-
clearance and distribution volume terms (CL/F and V/F), mental evaluation of dose-normalized AUC and Cmax
and allowing for the incorporation of the off-diagonal of (Fig. 2). The slope of AUC/dose or Cmax/dose versus dose
the covariance matrix (BLOCK 2) improved the fit. The was not significantly different than 0, with 95% CIs of -
limited number of subjects and sampling strategy impaired 308 to 222 (lgh/L)/(mg/kg) and -43.6 to 61.3 (lg/L)/(mg/
the ability to estimate Ka, and without assays of psilocin kg), respectively (Table 4).
glucuronide the kinetics of this compartment were only
indirectly estimated. Confidence in the final model esti- 3.3.1 Renal Clearance
mates for other parameters improved when the interindi-
vidual variation (ETA) for Ka and K32 (glucuronide to Renal clearance of psilocin was calculated from the ratio of
psilocin) were fixed at null. the amount of this active metabolite excreted in urine over
The renal clearance of psilocin was 1.7% of the total 24 h divided by the AUC for the same 24-h interval, as
clearance of this active metabolite, and renal function was reported by WinNonlin. Only 1.7% (95% CI 1.4–1.9%) of
not a useful covariate in the pharmacokinetic model of the administered dose of psilocybin was found as psilocin
clearance. in the urine. When calculated as the amount of renally
excreted psilocin with respect to the plasma psilocin
3.2 Final Compartmental Model AUC24, the renal clearance of psilocin was 1 mL/min/kg,
or 58% that of the corresponding measured creatinine
The final pharmacokinetic model included a linear one- clearance. The addition of measured creatinine clearance as
compartment and bidirectional compartment of psilocin a covariate in the NONMEM mixed-effects model did not
glucuronide (Fig. 1). The pharmacokinetic parameters for improve the fit of the data, and did not lead to a successful
the model are provided in Table 3. convergence of the model.
Pharmacokinetics of Oral Psilocybin

Table 3 Parameter estimates for final pharmacokinetic model: psilocin in serum

Parameter PK parameter mean (SE %) IIV (%CV; % shrinkage) Bootstrap median (95% CI) Bootstrap IIV [%CV] (95% CI)
(units)

CL/F (L/h) 164 (23.2) 31.4% 164.5 (29.0–224) 31.9 (15.8–145)

(61.9; 1.5)
V/F (L) 298 (20.2) 50.4% (51.2; 2.6%) 305 (212–465) 47.3 (24.8–72.2)
CL:V covariance: -0.269 Covariance: -0.387 (-100 to 61.9)
Ka (1/h) 0.367 (9.3) NE 0.372 (0.322–0.482) NE
K23 (1/h) 0.212 (39.6) NE 0.198 (0.1–0.576) NE
K32 (1/h) 0.0175 (55.9) NE 0.017 (0.006–0.068) NE
Residual error 0.43 (6.3) NE 0.425 (0.375–0.48) NE
PK pharmacokinetic, SE standard error of the mean, CV coefficient of variation, NE not estimated, CI confidence interval, IIV interindividual
variability, CL/F apparent clearance, V/F apparent volume of distribution, Ka first-order rate constant, K23 rate of formation of psilocin
glucuronide from psilocin, K32 rate of hydrolysis of psilocin glucuronide back to psilocin

3.4 Safety Figure 5 compares the AUC and Cmax of psilocin fol-
lowing a fixed psilocybin dose of 25 mg versus a weight-
Detailed presentation of the adverse effects noted during based dose of 0.3 mg/kg. Although the regression lines
and after the dosing sessions with psilocybin will be pub- suggest that the weight-based dosing of psilocybin may
lished elsewhere. No serious AEs were reported. The side result in higher than average exposures in heavier patients,
effects of the doses reflected previously reported findings, the slopes of these lines are not significantly different than
and included mild, transient hypertension and tachycardia null. Instead, the extensive overlap of the respective pre-
[34, 35]. As previously reported, mild headaches were diction CIs supports the use of a fixed dose of psilocybin
common in the second 12 h of the 24-h study periods, and over this dose and weight range.
were successfully treated with acetaminophen [36]. No
reports of hallucinogen persisting perception disorder
(HPPD) were reported during study or at follow-up. 4 Discussion

3.5 Simulation of Fixed Psilocybin Doses Our findings corroborated previous reports of the pharma-
cokinetics of psilocybin and its active metabolite, psilocin
Simulations (N = 500) were performed using the final [7, 18]. No parent psilocybin was detectable in plasma or urine,
pharmacokinetic model to evaluate the effect of adminis- arguing for the rapid luminal and first-pass dephosphorylation
tering a fixed dose of psilocybin (20 or 25 mg) instead of of psilocybin to psilocin. Although psilocybin metabolites
administering a 0.3 mg/kg dose to subjects included in the psilocin-O-glucuronide and 4-hydroxyindole alcohol have
present study. Simulations were plotted using a VPC with been reported to be present in concentrations several-fold
an overlay of the actual concentrations observed after the greater than psilocin in plasma [17], we did not determine their
0.3 mg/kg oral dose. The VPC plots demonstrate that the concentrations in this study due to funding constraints. The
expected concentrations from the 25 mg fixed dose overlap activity of the metabolites of psilocin has not been described.
with the observed concentrations (Fig. 3). The pharmacokinetics following the administration of
Psilocin AUC and Cmax expected from a fixed 20 and escalating doses of oral psilocybin were best fit with a two-
25 mg dose are compared with the actual data points for compartment model of elimination, but the large distribu-
AUC and Cmax for study subjects who received the 0.3 mg/ tion volume of the tissue compartment suggested from a
kg dose (Fig. 4). The plotted AUC values are derived from two-compartment fit (21,500 L) is difficult to justify on a
the dose and estimated clearance from the compartmental physiologic basis. Rather than presume a tissue compart-
model. The outliers from the box and whisker plot reflect ment of psilocin, the model was adapted to allow the
the influence of the subject weighing 120 kg, and the large reversible glucuronidation of psilocin. No separate com-
number of these outliers (‘?’ symbol) arises from the 500 partment was assumed for the glucuronide, but it was
iterations performed for the VPC test. The interquartile instead assumed to share the volume of the central com-
range (box region) of AUC and the Cmax estimated for a partment. Although multiple organs have b-glucuronidase
fixed psilocybin 25 mg dose approximates the range of activity, the cleavage of the sugar to reform psilocin was
concentrations following the actual 0.3 mg/kg dose (blue assumed to occur in the the vascular compartment [37].
circles). This ‘two-compartment’ model does not necessarily
 R. T. Brown et al.

plasma, the concentration of psilocin approximately dou-

bled at most of the sampling time points. In contrast, the
stoichiometric amount of psilocin estimated to be in the
glucuronide compartment in the present model is 29.4%
(SD 8.7) of the psilocin concentration. In addition,
although the present model does fit the observed psilocin
concentrations well, it does not include the direct elimi-
nation of the glucuronide, which, although presumed to
occur, cannot be fitted with the data available.
The peak concentration of psilocin was more gradually
attained in some subjects than in others, but the broad peak
psilocin concentrations did not appear to be a function of
dose. Multiple alternative models were tested to attempt to
characterize these findings, including an intermediate
absorption compartment of enterochromaffin cells that
might temporarily sequester and release the serotonin
analog psilocin. Additional models included transit com-
partments of psilocin to bile and gut, and mixed order (first
and mixed order) elimination of psilocin. Such models did
not yield improved objective functions, and were uniformly
poorly conditioned and parameters were indeterminately
resolved. Although the unavailability of plasma psilocin-O-
glucuronide concentrations made it impossible to confirm
the glucuronide as the source of the apparent two-com-
partment pharmacokinetics, the premise of an exchange-
able ‘reservoir’ of psilocin was considered to be plausible
and explained the plasma psilocin concentrations well.
The 24-h collection of urine after the administration of
psilocybin allowed the renal clearance of psilocin to be
determined from the amount of psilocin in the aggregate
24-h urine collection and the plasma psilocin AUC24. We
chose to add stabilizing ascorbic acid to the aggregate urine
collection kept in crushed ice, rather than to add ascorbate
and immediately freeze individual urine collections. The
use of DL-dithiothreitol as a reducing agent was not con-
sidered necessary due to the excellent peak resolution in
the presence of ascorbate [38]. The renal clearance of
psilocin in the present study was less than 2% of total
clearance, which is similar to the 3.4% renal excretion
reported by Hasler et al. [17]. The renal clearance of psi-
Fig. 2 Dose-normalized plasma psilocin AUC and Cmax. The dotted
locin was 58% that of measured creatinine clearance.
black line represents the least squares regression. AUC area under the Based on these findings and the lack of influence of mea-
concentration–time curve, Cmax maximum concentration sured creatinine clearance on the pharmacokinetic model,
no adjustment of the dose of psilocybin appears warranted
contradict the monoexponential model described by Hasler in subjects with impaired renal function. Furthermore,
et al. [7], which utilized a slightly less sensitive assay and moderately impaired renal function does not appear to be a
collected samples for only 6.5 h instead of over 24 h. justifiable criterion for exclusion of subjects from future
A confounder of the incorporation of a reversible psi- studies, but this presupposes that psilocin-O-glucuronide
locin-glucuronide ‘compartment’ is that we did not have and the 4-hydroxyindole metabolites are neither active nor
the resources to synthesize standard psilocin glucuronide or renally excreted. No data were found to support or refute
other metabolites that would have provided support for this activity of these metabolites.
component of the model. Additionally, the report by Hasler Only two women and two non-White subjects were
et al. [17] suggested that after enzymatic hydrolysis of recruited to this study. Advertising for the study was by
Pharmacokinetics of Oral Psilocybin

Table 4 Plasma psilocin AUC and Cmax by dose level

Dose level (mg/ Plasma psilocin Dose-adjusted plasma psilocin Plasma psilocin Dose-adjusted plasma psilocin Tmax (h)
kg) [doses] AUC (lgh/L) AUC (lgh/L)/(mg/kg) Cmax (lg/L) Cmax (lg/L)/(mg/kg)

0.3 [N = 12] 140 (102–175) 6.06 (4.61–7.34) 16 (14.5–17.2) 0.7 (0.584–0.779) 2.03 (1.15–2.07)
0.45 [N = 11] 213 (150–261) 6.84 (4.61–8.13) 26 (22.7–35.1) 0.838 (0.781–0.875) 2.03 (1.3–3)
0.6 [N = 10] 267 (201–356) 6.84 (4.61–8.13) 37.6 (27.7–43.2) 0.799 (0.645–1.096) 2.05 (1.55–2.08)
Data are expressed as median (25th and 75th percentiles) [number of doses administered]
AUC area under the concentration–time curve, Cmax maximum concentration, Tmax time to reach Cmax

found no evidence of a sex or race effect on the phar-

macokinetics of psilocybin, our ability to make such
distinctions was limited. Similarly, the psilocybin dose
range of 0.3–0.6 mg/kg was only twofold, and limited our
ability to identify nonlinear pharmacokinetics. Higher
doses were considered an excessive risk, but evaluation of
lower doses such as 0.2 mg/kg or even 0.1 mg/kg psilo-
cybin would have improved the confidence of a finding of
linearity.
Future progress to phase III trials of psilocybin will
require the use of an oral formulation meeting the FDA
and European Medicines Agency (EMA) expectations of
current Good Manufacturing Practice (cGMP). All
recent clinical studies of oral doses of psilocybin have
utilized individually prepared doses based on body
Fig. 3 VPC plot comparing observed results from psilocybin 0.3 mg/ weight (e.g. 0.3 mg/kg or 21 mg/70 kg). Oral psilocybin
kg oral dose versus simulated 25 mg oral dose. The open, blue circles doses greater than approximately 0.3 mg/kg, such as
are the observed concentrations after a psilocybin 0.3 mg/kg oral
dose; the solid and dotted-dashed blue lines are the median and 95% those evaluated in this phase I trial, are not expected in
CI of these observed concentrations; and the dashed and dotted black future clinical trials. Weight-based dose individualiza-
lines are the predicted median and 95% CI of the concentrations from tion would dramatically complicate the standardization
a flat 25 mg dose administered to the same subjects. VPC visual and validation of a cGMP product and its distribution,
predictive check, CI confidence interval, Obs observed, Pred
predicted particularly for a scheduled drug. An alternative is to
utilize a fixed dose of psilocybin. Simulations were
word of mouth and this, plus the imbalanced ethnic performed for single 20 and 25 mg oral doses of psilo-
demographics of Madison, Wisconsin, led to a low repre- cybin, and, in the present study, were compared with the
sentation of non-White races and ethnicities. Although we actual concentrations observed after a single 0.3 mg/kg

Fig. 4 Comparison of psilocin

AUC and Cmax following fixed
psilocybin 20 and 25 mg doses
versus a 0.3 mg/kg dose. Blue
circles represent the observed
AUC and Cmax in subjects; the
box represents the first and third
quartiles (25th and 75th
percentile) of 500 simulations of
AUC and Cmax from psilocybin
20 and 25 mg oral doses; and
the black crosses represent
outliers beyond the whisker
denoting 1.5 times the upper
quartile range. AUC area under
the concentration–time curve,
Cmax maximum concentration
 R. T. Brown et al.

Fig. 5 Comparison of psilocin AUC and Cmax by weight for the solid black lines and green shading show the respective AUC and
psilocybin 25 mg versus 0.3 mg/kg dose. Blue circles represent the Cmax following the psilocybin 25 mg dose, based on the 500
observed AUC and Cmax in subjects; the dashed, black lines and grey simulations of the final pharmacokinetic model. AUC area under
shading show the least-squares fit and 95% prediction interval for the the concentration–time curve, Cmax maximum concentration
subject results following their psilocybin 0.3 mg/kg oral dose; and the

dose. Although there is good overlap for both fixed doses 5 Conclusions
with the observed concentrations, the correspondence is
better with the 25 mg fixed dose at the time of peak This study demonstrated the linearity of psilocin exposure
plasma concentration and in the first 6 h, corresponding over the oral psilocybin dose range of 0.3–0.6 mg/kg in
with the peak effect of the drug. healthy adults, and suggests that the formation and
No serious AEs were noted in the 12 subjects treated hydrolysis of psilocin glucuronide may explain a biexpo-
with oral psilocybin. One subject was removed from the nential decay curve of psilocin. All doses were well tol-
study before receiving the second dose because his erated. Less than 5% of the oral psilocybin dose was
predose blood pressure (BP) exceeded initial eligibility excreted in the urine as psilocin, suggesting no need for
criteria. After several months of comparing recorded dose adjustment in patients with mild to moderate renal
home BP measurements versus those obtained at the impairment. Lastly, the use of a fixed oral psilocybin dose
CRU, it was determined that the subject demonstrated of 25 mg is expected to result in psilocin AUC and Cmax
‘white-coat’ hypertension. In consultation with CRU exposures similar to those demonstrated after the individ-
staff, it was learned that multiple BP readings were taken ualized 0.3 mg/kg oral dose.
on the initial eligibility screening visit in the successful
hope that the initially elevated BP might fall to accept- Acknowledgements The project described was supported by gifts
from the University of Wisconsin-Madison Foundation and the Usona
able concentrations. Given that pharmacokinetic sam-
Research Institute, and by the Clinical and Translational Science
pling was obtained after the first dose, this subject was Award (CTSA) program through the National Institutes of Health
considered eligible. (NIH) National Center for Advancing Translational Sciences
Another subject withdrew after completing the compo- (NCATS), Grant UL1TR000427. The content is solely the responsi-
bility of the authors and does not necessarily represent the official
nents of the second dose of psilocybin due to difficulty in
views of the NIH. The authors would like to thank Dr. Diane Mould for
getting off work midweek for the study. Although not advice on the population pharmacokinetic modeling, and Dr. Edmund
demonstrating adverse effects, another subject was Elder and the staff of the University of Wisconsin Zeeh Pharmaceutical
declared unevaluable after all attempts at drawing blood Experiment Station for their assistance in the quality affirmation of the
psilocybin API, and the preparation of the capsules for dosing.
after the first dose of psilocybin or at phlebotomy were
unsuccessful. Given that collection of postdose blood Compliance with Ethical Standards
samples was not feasible, the IRB permitted this subject to
be replaced. A subsequent report will describe relationships Funding This study was funded by gifts from the Psilocybin
between plasma psilocin concentrations and the psycho- Research Fund at the University of Wisconsin-Madison Foundation,
and by the Usona Research Institute.
logical effects of psilocybin, and associations of psilocin
concentration with mild but reportable AEs such as tran- Conflicts of interest Randall T. Brown, Christopher R. Nicholas,
sient hypertension and tachycardia. Nicholas V. Cozzi, Michele C. Gassman, Karen M. Cooper, Daniel
Pharmacokinetics of Oral Psilocybin

Muller, Chantelle D. Thomas, Scott J. Hetzel, Kelsey M. Henriquez, 14. Ray TS. Psychedelics and the human receptorome. PLoS One.
Alexandra S. Ribaudo, and Paul R. Hutson declare that they have no 2010;5:e9019.
conflicts of interest that might be relevant to the contents of this 15. Rickli A, Moning OD, Hoener MC, Liechti ME. Receptor
article. interaction profiles of novel psychoactive tryptamines compared
with classic hallucinogens. Eur Neuropsychopharmacol.
Ethical approval All procedures performed in these studies involv- 2016;26:1327–37.
ing human subjects were in accordance with the ethical standards of 16. Griffiths RR, Johnson MW, Richards WA, Richards BD, McCann
the institutional research committee, and with the 1964 Helsinki U, Jesse R. Psilocybin occasioned mystical-type experiences:
Declaration and its later amendments or comparable ethical standards. immediate and persisting dose-related effects. Psychopharma-
cology (Berlin). 2011;218:649–65.
Informed consent Informed consent was obtained from all individ- 17. Hasler F, Bourquin D, Brenneisen R, Vollenweider FX. Renal
ual participants included in this study. Consent was re-established excretion profiles of psilocin following oral administration of
after any change in protocol while the subject was still on study. psilocybin: a controlled study in man. J Pharm Biomed Anal.
2002;30:331–9.
18. Williams JBW, Gibbon M, First MB, Spitzer RL, Davis M, Borus
J, et al. The Structured Clinical Interview for DSM-III-R (SCID)
References II. Multi-site test-retest reliability. Arch Gen Psychiatry.
1992;49:630–6.
1. Nichols DE. Psychedelics. Pharmacol Rev. 2016;68:264–355. 19. Milanowski D, Leahy M, Freeman D. Validation of a method for
2. Grob CS, Danforth AL, Chopra GS, Hagerty M, McKay CR, the determination of psilocybin and psilocin in ascorbic acid
Halberstadt AL, et al. Pilot study of psilocybin treatment for treated human plasma by HPLC with MS/MS detection method
anxiety in patients with advanced-stage cancer. Arch Gen Psy- validation report. Covance Study No. 8290845.
chiatry. 2011;68:71–8. 20. Milanowski D, Leahy M, Freeman D. Validation of a method for
3. Griffiths RR, Johnson MW, Carducci MA, Umbricht A, Richards the determination of psilocybin and psilocin human urine by
WA, Richards BD, et al. Psilocybin produces substantial and HPLC with MS/MS detection method validation report. Covance
sustained decreases in depression and anxiety in patients with Study No. 8290843.
life-threatening cancer: a randomized double-blind trial. J Psy- 21. Beal SL, Sheiner LB, Boeckmann AJ, Bauer RJ, editors. NON-
chopharmacol. 2016;30:1181–97. MEM 7.3.0 users guides (1989–2013). Hanover: ICON Devel-
4. Ross S, Bossis A, Guss J, Agin-Liebes G, Malone T, Cohen B, opment Solutions. 2013.
et al. Rapid and sustained symptom reduction following psilo- 22. Reporting the results of population pharmacokinetic analyses.
cybin treatment for anxiety and depression in patients with life- CHMP/EWP/185990/06. European Medicines Agency. http://
threatening cancer: a randomized controlled trial. J Psychophar- www.ema.europa.eu/ema/index.jsp?curl=pages/regulation/general/
macol. 2016;30:1165–80. general_content_001284.jsp&mid=WC0b01ac0580032ec5. Acces-
5. Carhart-Harris RL, Bolstridge M, Rucker J, Day CM, Erritzoe D, sed 25 Mar 2017.
Kaelen M, et al. Psilocybin with psychological support for 23. R Core Team. R: A language and environment for statistical
treatment-resistant depression: an open-label feasibility study. computing. R Foundation for Statistical Computing, Vienna,
Lancet Psychiatry. 2016;3:619–27. Austria. 2015. https://www.R-project.org/. Accessed 21 Apr
6. Horita A, Weber LJ. The enzymic dephosphorylation and oxi- 2016.
dation of psilocybin and psilocin by mammalian tissue homo- 24. Jonsson EN, Karlsson MO. Xpose: an S-PLUS based population
genates. Biochem Pharmacol. 1961;7:47–54. pharmacokinetic/pharmacodynamic model building aid for
7. Hasler F, Bourquin D, Brenneisen R, Bar T, Vollenweider FX. NONMEM. Comput Methods Progr Biomed. 1999;58:51–64.
Determination of psilocin and 4-hydroxyindole-3-acetic acid in 25. Holford N. Wings for NONMEM. http://wfn.sourceforge.net.
plasma by HPLC-ECD and pharmacokinetic profiles of oral and Accessed 28 June 2016.
intravenous psilocybin in man. Pharm Acta Helv. 26. Beal SL. Ways to fit a PK model with some data below the
1997;72:175–84. quantification limit. J Pharmacokinet Pharmacodyn.
8. Vollenweider FX, Vollenweider-Scherpenhuyzen MF, Babler A, 2001;28:481–504.
Vogel H, Hell D. Psilocybin induces schizophrenia-like psychosis 27. Mosteller RD. Simplified calculation of body-surface area.
in humans via a serotonin-2 agonist action. Neuroreport. N Engl J Med. 1987;317:1098.
1998;9:3897–902. 28. Devine BJ. Gentamicin therapy. Drug Intell Clin Pharm.
9. Fantegrossi WE, Reissig CJ, Katz EB, Yarosh HL, Rice KC, 1974;8:650–5.
Winter JC. Hallucinogen-like effects of N,N-dipropyltryptamine 29. Janmahasatian S, Duffull SB, Ash S, Ward LC, Byrne NM, Green
(DPT): possible mediation by serotonin 5-HT1A and 5-HT2A B. Quantification of lean bodyweight. Clin Pharmacokinet.
receptors in rodents. Pharmacol Biochem Behav. 2005;44(10):1051–65.
2008;88:358–65. 30. Parke J, Holford NH, Charles BG. A procedure for generating
10. McKenna DJ, Repke DB, Lo L, Peroutka SJ. Differential inter- bootstrap samples for the validation of nonlinear mixed-effects
actions of indolealkylamines with 5-hydroxytryptamine receptor population models. Comput Methods Progr Biomed.
subtypes. Neuropharmacology. 1990;29:193–8. 1999;59:19–29.
11. Nichols DE. Hallucinogens. Pharmacol Ther. 2004;101:131–81. 31. Karlsson MO, Savic RM. Diagnosing model diagnostics. Clin
12. Smith RL, Canton H, Barrett RJ, Sanders-Bush E. Agonist Pharmacol Ther. 2007;82:17–20.
properties of N, N-dimethyltryptamine at serotonin 5-HT2A and 32. Keizer R. Creating visual predictive checks in R. https://github.
5-HT2C receptors. Pharmacol Biochem Behav. 1998;61:323–30. com/ronkeizer/vpc. Accessed 25 Mar 2017.
13. Moreno JL, Holloway T, Albizu L, Sealfon SC, Gonzalez-Maes J. 33. Comets E, Brendel K, Mentré F. Computing normalised predic-
Metabotropic glutamate mGlu2 receptor is necessary for the tion distribution errors to evaluate nonlinear mixed-effect mod-
pharmacological and behavioral effects induced by hallucino- els: the npde add-on package for R. Comput Methods Progr
genic 5-HT2A receptor agonists. Neurosci Lett. 2011;493:76–9. Biomed. 2008;90:154–66.
 R. T. Brown et al.

34. Studerus E, Kometer M, Hasler F, Vollenweider FX. Acute, 37. Woollen JW, Walker PG. The fluorimetric estimation of b-glu-
subacute, and long-term subjective effects of psilocybin in heal- curonidase in blood plasma. Clin Chim Acta. 1965;12:659–70.
thy humans: a pooled analysis of experimental studies. J Psy- 38. Manevski N, Kurkela M, Hoglund C, Mauriala T, Court MH, Yli-
chopharmacol. 2010;25:1434–52. Kauhaluoma J, et al. Glucuronidation of psilocin and 4-hydrox-
35. Johnson MW, Richards WA, Griffiths RR. Human hallucinogen yindole by the human UDP-glucuronosyltransferases. Drug
research: guidelines for safety. J Psychopharmacol. Metab Disp. 2010;38:386–95.
2008;22:603–20.
36. Johnson MW, Sewell RA, Griffiths RR. Psilocybin dose-depen-
dently causes delayed, transient headaches in healthy volunteers.
Drug Alcohol Depend. 2012;123:132–40.