Precision sirolimus dosing in
children: The potential for
model-informed dosing and novel
drug monitoring
Guofang Shen
1
†
, Kao Tang Ying Moua
2
†
, Kathryn Perkins
2
†
,
Deron Johnson
3
, Arthur Li
4
, Peter Curtin
1
, Wei Gao
5
and
Jeannine S. McCune
1
*
1
Department of Hematologic Malignancies Translational Sciences, City of Hope, and Department of
Hematopoietic Cell Transplantation, City of Hope Medical Center, Duarte, CA, United States,
2
Alfred E.
Mann School of Pharmacy and Pharmaceutical Sciences, University of Southern California, Los Angeles,
CA, United States,
3
Clinical Informatics, City of Hope Medical Center, Duarte, CA, United States,
4
Division
of Biostatistics, City of Hope, Duarte, CA, United States,
5
Division of Engineering and Applied Science,
Andrew and Peggy Cherng Department of Medical Engineering, California Institute of Technology,
Pasadena, CA, United States
The mTOR inhibitor sirolimus is prescribed to treat children with varying diseases,
ranging from vascular anomalies to sporadic lymphangioleiomyomatosis to
transplantation (solid organ or hematopoietic cell). Precision dosing of
sirolimus using therapeutic drug monitoring (TDM) of sirolimus concentrations
in whole blood drawn at the trough (before the next dose) time-point is the
current standard of care. For sirolimus, trough concentrations are only modestly
correlated with the area under the curve, with
R
2
values ranging from 0.52 to 0.84.
Thus, it should not be surprising, even with the use of sirolimus TDM, that patients
treated with sirolimus have variable pharmacokinetics, toxicity, and effectiveness.
Model-informed precision dosing (MIPD) will be bene
fi
cial and should be
implemented. The data do not suggest dried blood spots point-of-care
sampling of sirolimus concentrations for precision dosing of sirolimus. Future
research on precision dosing of sirolimus should focus on pharmacogenomic and
pharmacometabolomic tools to predict sirolimus pharmacokinetics and
wearables for point-of-care quantitation and MIPD.
KEYWORDS
sirolimus (rapamycin), pediatrics, therapeutic drug monitoring (TDM), sweat, saliva, dried
blood spots (DBS), pharmacogenomics, pharmacometabolomic
1 Introduction
Rapamune
®
(sirolimus) is approved to prevent organ rejection in patients aged 13 years
or older receiving renal transplants by the Food and Drug Administration (
Author
Anonymous, 2022a
). In addition, the European Medicines Agency (EMA) approved
Rapamune
®
for prophylaxis of organ rejection in adults at low to moderate
immunological risk receiving a renal transplant and for treatment of patients with
sporadic lymphangioleiomyomatosis with moderate lung disease or declining lung
function (
EMA, 2022
). Over the past 20 years since these initial approvals, sirolimus has
expanded to treat children undergoing heart (
Rossano et al., 2017
), hematopoietic cell
(
Monagel et al., 2021
), intestine (
Andres et al., 2021
), liver (
Hendrickson et al., 2019
), or lung
OPEN ACCESS
EDITED BY
Raffaele Simeoli,
Bambino Gesù Children
’
s Hospital
(IRCCS), Italy
REVIEWED BY
Paula Schaiquevich,
Garrahan Hospital, Argentina
Tamorah Rae Lewis,
University of Toronto, Canada
*CORRESPONDENCE
Jeannine S. McCune,
jmccune@uw.edu
†
These authors share
fi
rst authorship
SPECIALTY SECTION
This article was submitted to Obstetric
and Pediatric Pharmacology,
a section of the journal
Frontiers in Pharmacology
RECEIVED
18 December 2022
ACCEPTED
14 February 2023
PUBLISHED
20 March 2023
CITATION
Shen G, Moua KTY, Perkins K, Johnson D,
Li A, Curtin P, Gao W and McCune JS
(2023), Precision sirolimus dosing in
children: The potential for model-
informed dosing and novel
drug monitoring.
Front.Pharmacol.
14:1126981.
doi: 10.3389/fphar.2023.1126981
COPYRIGHT
© 2023 Shen, Moua, Perkins, Johnson, Li,
Curtin, Gao and McCune. This is an open-
access article distributed under the terms
of the
Creative Commons Attribution
License (CC BY)
. The use, distribution or
reproduction in other forums is
permitted, provided the original author(s)
and the copyright owner(s) are credited
and that the original publication in this
journal is cited, in accordance with
accepted academic practice. No use,
distribution or reproduction is permitted
which does not comply with these terms.
Frontiers in
Pharmacology
frontiersin.org
01
TYPE
Review
PUBLISHED
20 March 2023
DOI
10.3389/fphar.2023.1126981
transplant (
Hayes et al., 2014
). Furthermore, children with vascular
anomalies are treated with sirolimus (
Mizuno et al., 2017c
).
Sirolimus is a lipophilic macrocytic lactone that binds
distinctly to FK binding protein 12 (FKBP12), forming a
complex with the mammalian target of rapamycin (mTOR)
(
Cutler and Antin, 2004
). This sirolimus
–
FKBP12
–
mTOR
complex inhibits multiple cytokine
–
stimulated cell cycling
pathways by reducing DNA transcription, DNA translation,
protein synthesis, and cell signaling (
Sehgal, 2003
). It also
inhibits interleukin
–
2
–
mediated proliferation signaling, leading
to T
–
cell apoptosis (
Sehgal, 2003
).
The most common toxicities (
>
20%) in patients taking
sirolimus include hypertriglyceridemia (45%
–
57%),
stomatitis,
diarrhea,
abdominal
pain,
nausea,
nasopharyngitis, acne, chest
pain, peripheral edema, upper
respiratory tract infection, headache, dizziness, and myalgia
(
Author Anonymous, 2022a
). Sirolimus is available as tablets
(0.5,1,2mg)and,insomecountries,asaliquidsolution(1mg/
mL) formulation. The long half
–
life of sirolimus allows for
convenient once
–
daily dosing, but a loading dose is required
to rapidly achieve target drug co
ncentrations in whole blood.
Therefore, it is usually administered once daily at a
fi
xed dose in
adults (one 6
–
12 mg loading dose, followed by 2
–
4mg daily)
and as a body surface area
–
based dose in children (2.5 mg/m
2
/
day) (
Supplementary Table S1
).
Sirolimus has a narrow therapeutic window. Its product
labeling recommends monitoring sirolimus trough
concentrations for all patients, especially those likely to have
altered drug metabolism, in patients
≥
13 years who weigh less
than 40 kg, in patients with hep
atic impairment, when a change
in the sirolimus dosage form is made, and during concurrent
administration of strong cytochrome P450 3A (CYP3A) inducers
and inhibitors (
Author Anonymous, 2022a
). Therefore, its doses
are personalized dosing using therapeutic drug monitoring
(TDM) of whole blood obtained immediately before the
subsequent dose (i
.e., trough or predose samples). This
provides a useful strateg
y to optimize transplant
pharmacotherapy (
Kahan et al., 2000
). For the past 20 years,
the majority of patients have u
ndergone the following TDM
process for precision dosing of sirolimus: 1. choose the target
trough concentration in whole blood, typically between 3 and
14 ng/mL (
Claxton et al., 2005
;
Alyea et al., 2008
;
Ho et al., 2009
;
Nakamura et al., 2012
;
Khaled et al., 2013
); 2. administer a
sirolimus loading dose based on weight or body surface area;
3. Obtain a trough pharmacokinetic sample; 4. quantitate the
sirolimus concentrations in who
le blood, typically using liquid
chromatography
–
mass spectrometry (LC
–
MS); 5. use that
trough concentration to personalize the dose to achieve the
target trough concentration. No
tably, sirolimus whole blood
concentrations may be measured by either chromatographic
or immunoassay methods (
Mahalati and Kahan, 2001
;
Schmid
et al., 2009
). Due to cross
–
reactivity with sirolimus metabolites,
immunoassay methods have a positive bias ranging from 14% to
39% compared to LC
–
MS methods (
Schmid et al., 2009
). Because
sirolimus whole blood concentrations vary by the type of assay
used, concentrations are not inte
rchangeable between methods.
Therefore, sirolimus TDM should be conducted using one
consistent bioanalytical method within an institution.
2 Why should we expand the precision
dosing of sirolimus beyond TDM
Dosing sirolimus based on trough concentrations has been the
current standard of care (
Stenton et al., 2005
) for over 20 years.
However, trough concentrations only modestly correlate with
AUC
0
–
24hr
, with
R
2
values between trough concentrations and
area under the plasma concentration-time curve for sirolimus
ranging from 0.52 to 0.84 (
Schachter et al., 2004
;
Schubert et al.,
2004
;
Goyal et al., 2013
). We have not found publications regarding
the existence or results from a sirolimus pro
fi
ciency program that
evaluates the accuracy of quantitation, pharmacokinetic modeling,
and dose recommendations. Hopefully, such a pro
fi
ciency program
will be developed for sirolimus TDM because such programs have
discovered challenges with TDM of other drugs (
Neef et al., 2006
;
McCune et al., 2021b
).
Although TDM is accepted for sirolimus, trough concentrations
are limited because they fail to provide a rich, mechanistic
description
of
the
pharmacokinetic/pharmacodynamic
relationship (
Dupuis et al., 2013
) that could advance our
understanding of why certain patients experience adverse
outcomes. Model-informed precision dosing (MIPD) may
improve clinical outcomes by optimizing the personalized dose
for an individual patient (
Darwich et al., 2017
). The development
of such mechanistic models can help improve individual patients
’
clinical outcomes, which can be attributed to the pharmacokinetics
and pharmacodynamics of sirolimus. Regarding the
pharmacokinetics, low immunosuppressant concentrations or
exposure is multifactorial; they can result from insuf
fi
cient dosing
or dose personalization, aberrant pharmacokinetics due to patient
covariates, and/or non-adherence (
McCune and Bemer, 2016
;
McCune et al., 2016
;
Vaisbourd et al., 2022
). Non-adherence is
associated with the development and severity of adverse outcomes
such as graft loss in solid organ transplant (
Foster et al., 2018
) and
graft-versus-host disease (GVHD) hematopoietic cell transplant
patients (
Gresch et al., 2017
;
Ice et al., 2020
). Thus,
improvements in sirolimus dosing to minimize the between-
patient variability in sirolimus concentrations through novel
-omics techniques or point-of-care monitoring at home may be
bene
fi
cial. These will be the focus of this review.
However, the promising innovations in -omics techniques and
mathematical modeling related to the immune system necessitate
that we brie
fl
y describe factors possibly affecting the
pharmacodynamics of sirolimus. Pharmacodynamic monitoring
of the cellular targets of immunosuppressant drugs may re
fl
ect
clinical outcomes better than TDM (
Monchaud and Marquet,
2009a
;
Monchaud and Marquet, 2009b
). For example, recipient
pretransplant inosine monophosphate dehydrogenase activity is
associated with clinical outcomes after renal transplant or
allogeneic hematopoietic cell transplant recipients treated with
mycophenolate mofetil (MMF) (
Glander et al., 2004
;
Bemer
et al., 2014
). However, a drug-speci
fi
c pharmacodynamic
biomarker for sirolimus
’
s effectiveness or toxicity has yet to be
identi
fi
ed. Turning to the use of sirolimus to alter the immune
system, the ontogeny of innate and adaptive immune responses
involve more than 1,600 genes (
Simon et al., 2015
). These genes are
essential to sustain life in a hostile environment. Yet the immune
system is relatively immature at birth. It has to evolve during a
Frontiers in
Pharmacology
frontiersin.org
02
Shen et al.
10.3389/fphar.2023.1126981
lifetime of exposure to many foreign challenges through childhood,
young and mature adulthood (including pregnancy). It subsequently
declines in old age (
Simon et al., 2015
). Beyond the effects of these
genes, patients may have several transcriptomic (
Furlan et al., 2020
),
proteomic (
Cohen Freue et al., 2013
;
Levitsky et al., 2013
),
metabolomic [reviewed in
Supplementary Table S1
of
McCune
et al. (2021a)
], and lipidomic (
Liggett et al., 2022
) characteristics
that could in
fl
uence the effectiveness and toxicity of transplantation
and/or sirolimus. For example, metabolomics can offer discoveries
yielding new insights into how metabolites (here, endogenous, not
sirolimus metabolites) in
fl
uence gut physiology, organ function, and
immune function (
Wishart, 2019
). Given their signaling properties
in addition to a multitude of other functions (
Cockcroft, 2021
),
lipidomics may enable a deeper mechanistic understanding of the
T-cell migration (
Cucchi et al., 2020
), drug-target interactions, and
systems physiology from the molecular (genomic, proteomic,
metabolomic) to cellular to whole-body levels. Collectively, these
works could lead to a system-wide perspective of pathophysiology
wherein genes, proteins, metabolites, and lipids are understood to
interact synergistically to modify the functions within a patient
receiving sirolimus. These insights may provide the foundation for
enhanced pharmacokinetic/dynamic modeling to comprehensive
quantitative systems pharmacology (QSP) models (
Ayyar and
Jusko, 2020
). In homogenous and suf
fi
ciently powered
populations of patients treated with sirolimus, multi-omic tools
should be explored for precision dosing and for building QSP
models to improve clinical outcomes.
Mathematical modeling and simulation can characterize the
complexity and multiscale nature of the mammalian immune
response and provide a mechanistic understanding of the data
generated from the novel
–
omics technologies (
Palsson et al.,
2013
). As an example of such modeling and simulation, the
Fully-integrated Immune Response Model (FIRM) represents a
multi-organ structure comprised of the target organ, where the
immune response occurs, and circulating blood, lymphoid T, and
lymphoid B tissue (
Palsson et al., 2013
). FIRM can be expanded to
include novel biological
fi
ndings relevant to sirolimus, such as
incorporating novel medications that target antigen-presenting
cells (B-cells), T-cell subsets, T-cell signal transduction,
costimulatory molecules, or cytokines into QSP models. Early
steps are being taken to apply QSP models to precision dosing,
speci
fi
cally in the context of the well-characterized coagulation
cascade (
Hartmann et al., 2016
). However, the inherent
complexity of the immune system and the dif
fi
culty of measuring
many aspects of a patient
’
s immune state
in vivo
makes it
challenging to develop such QSP models of immune response
(
Laubenbacher et al., 2022
). Because the immune system has an
important role in such a wide range of diseases and health
conditions, digital twins of the immune system are of keen
interest. Advanced medical digital twins will make precision
medicine a reality (
Laubenbacher et al., 2022
).
Therefore, it is crucial to explore newer methods for precision
dosing of sirolimus (
Table 1
). Here, we summarize our
fi
ndings from
a series of literature reviews (
Supplementary Methods
and
Supplementary Figures S1
–
S5
) about various
–
omic or
point
–
of
–
care tools focusing on the sirolimus pharmacokinetics
that may improve the precision dosing of sirolimus in children.
3 Sirolimus pharmacokinetics
After oral administration, sirolimus achieves its peak whole
blood concentrations within 1
–
3.5 h (
MacDonald et al., 2000
;
Stenton et al., 2005
). The apparent oral bioavailability of
TABLE 1 Steps of and research about precision sirolimus dosing: TDM of sirolimus dosing the initial (
fi
rst) dose to achieve the target trough. The tools in the bold
font should be implemented, and those in the italicized font are not recommended for clinical use.
Steps
Current steps
Research opportunities
1. Choose and then administer the initial
sirolimus dose
Body weight
Before sirolimus administration:
Model-informed precision dosing using population
pharmacokinetic (popPK)
Djebli et al. (2006)
,
Mizuno et al.
(2017a)
,and
Darwich et al. (2017)
Pharmacogenomics:
Table 2
Pharmacometabolomics
2. Pharmacokinetic blood sampling
Trough
PopPK
–
guided limited sampling schedules for blood sampling
Djebli et al. (2006)
Dried blood spots
:
Table 3
Saliva sampling:
Table 4
Sweat sampling:
Table 5
3. Quantitation of sirolimus
concentrations
Immunoassay
Metabolite
–
speci
fi
c antibody
–
like molecularly imprinted polymers
and redox
–
active reporter nanoparticles
Wang et al. (2022)
LC-MS
4. Pharmacokinetic modeling of
concentration
–
time data
Not possible with trough concentration only
PopPK
–
guided dosing with a posterior Bayesian prediction
Mizuno et al. (2017a)
5. Determine one patient
’
s precision dose
to achieve their target trough
Personalized Dose
Target trough X
(
Initial Dose
Trough with initial dose
)
Frontiers in
Pharmacology
frontiersin.org
03
Shen et al.
10.3389/fphar.2023.1126981
sirolimus is poor (
Stenton et al., 2005
). Extensive intestinal and
hepatic
fi
rst-pass metabolism contributes to the low oral
bioavailability.
Sirolimus is highly lipophilic (log P of 4.3,
Supplementary Table
S1
). Sirolimus has a large apparent volume of distribution of
(5.6
–
22.8 L/kg) (
Brattstrom et al., 2000
;
Stenton et al., 2005
).
This large distribution volume implies extensive distribution to
organs and tissues and contributes to the long half-life. The free
fraction in plasma is 8% (
MacDonald et al., 2000
). Sirolimus is
partitioned extensively into blood cells with a blood-to-plasma ratio
of 35.6 (
Tejani et al., 2004
). In kidney transplant (KT) patients, this
ratio is independent of sirolimus concentration and exhibits a large
variability (
Ferron et al., 1997
). In humans, sirolimus is distributed
among red blood cells (94.5%), whole blood (3.1%), lymphocytes
(1.01%), and granulocytes (1.0%) (
Stenton et al., 2005
). The
sequestration of sirolimus in red blood cells is believed to be
partially due to their rich content of immunophilins (
Stenton
et al., 2005
). In the whole blood compartment, sirolimus exhibits
concentration
–
dependent binding to lipoproteins (40%) with a
minor fraction (
<
4%) bound to plasma proteins. Therefore,
whole blood is considered the most favorable matrix for TDM
(
Stenton et al., 2005
).
The primary route of elimination occurs
via
fecal/biliary
pathways, with an estimated terminal elimination half
–
life of
approximately 62 ± 16 h (
Stenton et al., 2005
). After a single oral
dose in healthy adults, sirolimus has a half-life of 81.5 h and a total
body clearance is 278 mL/h/kg (
Brattstrom et al., 2000
).
Sirolimus is metabolized by CYP3A4 and CYP3A5 in both human
liver and small intestinal microsomes to various demethylated and
hydroxylated species (
Paine et al., 2002
). Degradation products,
including an ester hydrolysis product and a ring
–
opened isomer,
have also been described (
Paine et al., 2002
). In the liver, sirolimus is
primarily metabolized by CYP3A4, with CYP3A5 and
CYP2C8 having lesser roles (
Jacobsen et al., 2001
;
Emoto et al.,
2013
). Sirolimus is transported by the multidrug resistance gene
product pump p
–
glycoprotein (PgP) (
Stenton et al., 2005
), an
apically directed ATP
–
dependent transmembrane secretory (ef
fl
ux)
pump expressed at high levels in enterocytes (
Paine et al., 2002
).
Examples of potential drug
–
drug interaction (DDI) with
sirolimus result from concomitant calcium channel blockers,
imatinib, antibiotics, or antifungals (
Leather, 2004
;
Bernard et al.,
2014
;
Bleyzac et al., 2014
). Because of routine TDM of trough
concentrations of the sirolimus, these results can be used to
identify a DDI and appropriately personalize the sirolimus dose.
For example, the DDI between azole antifungals and sirolimus has
long been recognized (
Yee and McGuire, 1990b
;
Yee and McGuire,
1990a
) and can be managed through TDM (
Leather, 2004
). The
azoles have variable CYP3A4 inhibition, potentially affecting
CYP2C9, CYP2C19, and PgP (
Leather, 2004
). Although these
azole
–
immunosuppression DDI are well known, their
management can be variable and could bene
fi
t from improved
pharmacokinetic/pharmacodynamic modeling. The EMA changed
the drug label for sirolimus to include therapeutic monitoring
during dose adjustments when inducers or inhibitors of CYP3A
are concurrently administered and/or discontinued (
Ehmann et al.,
2014
).
Furthermore, sirolimus is also susceptible to being the victim
drug to natural products altering CYP3A or PgP activity, such as
grapefruit juice and St. John
’
s wort (
Edwards et al., 1999
;
Mai et al.,
2003
;
Brantley et al., 2013
). Sirolimus pharmacokinetics may have
circadian variability, as the maximum plasma concentration and
AUC of other CYP3A/PgP substrates (i.e., cyclosporine and
tacrolimus) are higher in the morning than in the afternoon
(
Baraldo and Furlanut, 2006
). Seasonal variation is also of
concern, as it has recently been reported that duodenal
CYP3A4 mRNA is signi
fi
cantly higher between April and
September than between October and March (
Thirumaran et al.,
2012
).
The maintenance dose of sirolimus should be adjusted in
patients with hepatic impairment or at risk of interactions with
sirolimus, either due to concomitant drugs (
Author Anonymous,
2022a
) or natural products (
Paine et al., 2018
) affecting CYP3A or
PgP activity.
4 Model-informed precision dosing of
sirolimus in children
Pharmacometrics enables developing models that describe
factors affecting the pharmacokinetics and/or pharmacodynamics
in children (
Mehrotra et al., 2016
). Children are not small adults
because of differences in biochemical, body composition, and
physiology processes (
Rodman et al., 1993
;
Murry et al., 1995
;
de
Wildt et al., 1999
;
Blanco et al., 2000
;
Kearns et al., 2003
;
Ince et al.,
2009
;
Mahmood, 2014
). Speci
fi
c to sirolimus, the data regarding the
maturation of CYP3A4 is con
fl
icting, with different age variations in
enzyme activity. For example,
Salem et al. (2014)
suggested that
hepatic CYP3A4 increases from an early age and reaches the adult
level by 2.5 years. In contrast,
Upreti and Wahlstrom (2016)
suggested that CYP3A4 maturation exceeds the adult level
between 0.1 and 11 years. Using such enzyme maturation data is
critical in creating physiologically based pharmacokinetic models
(PBPK) to predict pediatric pharmacokinetics and dosing (
Mehrotra
et al., 2016
).
A PBPK model for children aged 1 month to 2 years
demonstrated that the relationship between allometrically scaled
in vivo
sirolimus clearance and age was described by the Emax
model (
Emoto et al., 2015a
). Consistent with this increased clearance
in patients,
in vitro
intrinsic clearance of sirolimus using pediatric
liver microsomes shows a similar age-dependent increase. In
children older than 2 years, allometrically scaled apparent oral
clearance of sirolimus did not show further maturation.
Simulated clearance estimates with a sirolimus PBPK model that
included CYP3A4/5/7 and CYP2C8 maturation pro
fi
les were in
close agreement with observed
in vivo
clearance values (
Emoto et al.,
2015a
). However, further exploration of the impact of different
assumptions regarding CYP3A4 age
–
related changes on the PBPK
model predictive performance may be bene
fi
cial (
Lang et al., 2021
).
In addition, PBPK model-simulated sirolimus pharmacokinetic
pro
fi
les predicted the actual observations well (
Emoto et al.,
2015a
). The mean sirolimus clearance was 11 ± 3 L/h, 17 ± 4 L/
h, 21 ± 3 L/h, and 18 ± 6 L/h for the age groups of 1
–
8 months (
<
1),
1 year (1 to
<
2), 2 years (2 to
<
3), and 3
–
18 years (
≥
3), respectively
(
Emoto et al., 2015a
). Sex, ethnicity, or race did not show a
statistically signi
fi
cant association with sirolimus clearance in a
cohort of 44 children (
Emoto et al., 2015a
). These results
Frontiers in
Pharmacology
frontiersin.org
04
Shen et al.
10.3389/fphar.2023.1126981
demonstrate the utility of a PBPK modeling approach for predicting
the developmental trajectory of sirolimus metabolic activity and its
effects on total body clearance in neonates and infants (
Emoto et al.,
2015a
).
Complementing PBPK modeling, population pharmacokinetic
(PopPK) models (
Beal and Sheiner, 1982
) can address relevant
hurdles by accounting for variability and mitigating the resource
intensity obtaining more pharmacokinetic samples beyond trough
samples. PopPK models mathematically describe typical drug
kinetics while accounting for between subject variability and
residual unknown variability (
Holford et al., 2000
) and the role
of demographic covariates responsible for or related to variability,
such as age or gender. PopPK models also facilitate the development
of limited sampling schedules, which is essential since most
immunosuppression is administered in the outpatient clinic (
Li
et al., 2012
;
Li et al., 2013
). PopPK models could overcome the
major challenge of the
–
omics tools, speci
fi
cally the interference
from confounding factors (
Ioannidis et al., 2009
;
Gu et al., 2012
).
Pharmacokinetics can be used to address these confounding factors
by identifying factors associated with aberrant metabolism.
Theoretical allometrically scaled body weight accounted for
differences in body size (
Mizuno et al., 2017b
). Using a popPK
model and more comprehensive blood sampling of sirolimus, MIPD
was conducted in children with vascular anomalies as part of a
prospective phase II trial (
Mizuno et al., 2017a
). In 52 children, the
target trough was attained in 94%, speci
fi
cally 49 of 52 children,
across the age range of 3 weeks
–
18 years after 2
–
3 months of therapy
(
Mizuno et al., 2017a
). The mean sirolimus dose to achieve the target
trough of ~10 ng/mL for patients older than 2 years was 1.8 mg/m
2
twice daily (range 0.8
–
2.9), while it was 0.7
–
1.6 mg/m
2
twice daily
for patients 3 weeks of age to 2 years. The
fi
nal popPK model
included a maturation function for sirolimus clearance and
allometrically scaled body weight to account for size differences.
The mean allometrically scaled sirolimus clearance estimates
increased from 3.9 to 17.0 L/h/70 kg with age from shortly after
birth to 2 years of age, while the mean estimate for patients older
than 2 years was 18.5 L/h/70 kg. This MIPD can be extended to
other pediatric populations and perhaps adults (
Mizuno et al.,
2017a
).
5 Multi-omics technologies
5.1 Pharmacogenomics
The pharmacokinetics and pharmacodynamics of sirolimus can
be in
fl
uenced by genetic polymorphisms in
fl
uencing the expression
of CYP3A4, CYP3A5, or PgP (
Utecht et al., 2006
). Pre
–
emptive
genotyping, or using genotyping results to guide initial dosing before
sirolimus administration, takes a step towards the
“
right
–
dose
–
fi
rst
–
time
”
paradigm (
Minto and Schnider, 1998
).
However, preemptive genotype
–
directed dosing will not account
for non-genetic factors associated with sirolimus pharmacokinetics
(
Section 3
and
Section 4
). Although the
CYP3A4
and
MDR1
genes
may contribute to sirolimus pharmacokinetics, we focused on
CYP3A5
because it is the only gene involved in sirolimus
pharmacokinetics
with
Clinical
Pharmacogenetics
Implementation Consortium (CPIC
®
) guidelines. We reviewed
the literature regarding the association of the
CYP3A5
genotype
with the pharmacokinetic phenotype (
Supplementary Figure S1
).
Very few studies evaluated if the
CYP3A5
genotype was associated
with the effectiveness or toxicity (i.e., pharmacodynamics) of
sirolimus
–
based immunosuppressive regimens (
Mourad et al.,
2005
;
Renders et al., 2007
;
Lukas et al., 2010
;
Zochowska et al.,
2012
;
Wang et al., 2014
;
Khaled et al., 2016
;
Rodriguez-Jimenez et al.,
2017
).
A variant in intron 3 of
CYP3A5
(rs776746) creates a cryptic
splice site which results in aberrant splicing and creates a
premature stop codon that results in transcript degradation
(
Hustert et al., 2001
;
Kuehl et al., 2001
). This allele, now known
as
CYP3A5*3
,explainstheliver
’
s highly variable expression of
CYP3A5 protein. Based on homozygosity for the
CYP3A5*3
allele, individuals are divided into CYP3A5 non-expressors
(
CYP3A5*3/*3
) and CYP3A5 expressors (
CYP3A5*1/*3
and
CYP3A5*1/*1
)(
Rodriguez-Antona et al., 2022
).
CYP3A5*3
is
the most common allele in Euro
pean and Asian populations,
but it is the minor allele in people of African ancestry. Thus,
CYP3A5 protein is only expressed in 10%
–
30% of Europeans
and Asians but in ~70% of people of African ancestry
(
Rodriguez-Antona et al., 2022
).
The
CYP3A5
rs776746 variant has been the most extensively
evaluated for its association with sirolimus pharmacokinetics in
transplant patients (
Table 2
). CYP3A5 non-expressors should have
lower sirolimus clearance and, without TDM, should have higher
dose
–
adjusted trough concentrations (
Anglicheau et al., 2005
).
Conversely, the CYP3A5 expressors should have CYP3A5 protein
and thus faster sirolimus clearance, and, without TDM, should have
lower dose
–
adjusted trough concentrations. Anglicheau
(
Anglicheau et al., 2005
) was one of the earliest and largest (
n
=
129) studies to demonstrate signi
fi
cant differences in dose
–
adjusted
trough concentrations between the
CYP3A5
genotypes. They
evaluated three different KT patient treatment groups: 1)
sirolimus rescue therapy after discontinuing calcineurin inhibitor
(CNI) therapy for concerns of nephrotoxicity (
n
= 69), 2) sirolimus-
based therapy
de novo
post
–
KT (
n
= 51), 3) sirolimus + CNI
–
based
therapy (
n
= 29). In the rescue therapy group, expressors had an
average weight adjusted trough concentration of 89 ± 65 (ng/mL)/
(mg/kg) vs. non-expressors 145 ± 93 (ng/mL)/(mg/kg) (
n
= 69,
p
<
0.02) at 3 months post sirolimus initiation. Notably, only those
patients who were transitioned onto sirolimus rescue therapy had a
statistically signi
fi
cant association between
CYP3A5
rs776746 genotype and dose
–
adjusted trough concentration
(
Anglicheau et al., 2005
).
Le Meur had a similar
fi
nding that non-expressors of
CYP3A5 consistently had higher dose
–
adjusted trough
concentrations and AUC
0
–
9hr
when measured at 1 week, 2 weeks,
1 month, and 3 months post
–
KT (
Le Meur et al., 2006
).
Miao et al.
(2008)
,
Lee et al. (2014)
, and
Li et al. (2015)
reported that the
CYP3A5
non-expressors have a higher trough concentration/(dose/
weight) than expressors. These three studies had a more even
distribution of expressors and non-expressor in their treatment
groups (i.e., Miao
n
= 21,26; Lee
n
= 36, 41; Li = 20, 23)
compared to other studies with more patients with the expressor
genotype (
Miao et al., 2008
;
Lee et al., 2014
;
Li et al., 2015
). In a
long
–
term retrospective cohort study, Rodriguez
–
Jimenez reported
an association of CYP3A5 expressor status with the sirolimus
Frontiers in
Pharmacology
frontiersin.org
05
Shen et al.
10.3389/fphar.2023.1126981
TABLE 2 Pharmacogenetic association of
CYP3A5
rs776746 genotype with sirolimus pharmacokinetics. Publications are organized in order of publication date, starting with the oldest. Gray-shade
d cells show results where the
CYP3A5
genotype was associated with a statistically signi
fi
cant difference (
p
<
0.05) in sirolimus pharmacokinetics.
Author
Study design/
Study population
Sirolimus dosing and pharmacokinetic sampling
Immunosuppressant regimen
a
CYP3A5
Allele
Sirolimus PK
endpoint
Anglicheau et al. (2005)
Retrospective
N= 149 KT
Self
–
reported race: 140 Caucasian, 4 Black, 5 Caribbean
Age: 44.9 ± 11.4 years
Child: No
Included: SIR for
≥
3 months
Excluded: (
n
= 8) DDI (i.e., nicardipine, diltiazem,
fl
uconazole)
A priori power analysis: NA
Initial dose: NA
Target trough:
SIR Based (including Rescue Therapy): 10
–
20 ng/mL
SIR + CNI
–
Based
:10
–
15 ng/mL
PK sampling: Whole blood collected 24 hours post-dose;
used troughs after at least months of SIR administration
Was PK sampling at steady state?: Yes
Quantitation: HPLC
Detection: NA
LOD/LOQ: NA
Trough/(Dose/Weight)
(ng/mL) per (mg/kg)
Mean ± SD
SIR-Based Therapy:
N=51
b
SIR, purine inhibitor (azathioprine or MMF), + Prednisolone
Steroid dosing
a
:NA
*1/*1
*1/*3
n
=13
b
147 ± 55
*3/*3
n
=18
b
171 ± 113
SIR Rescue Therapy
N=69
SIR is used in pts with suspected CNI nephrotoxicity;
Steroid dosing: NA
*1/*1
*1/*3
n
=11
89 ± 65
*3/*3
n
=58
145 ± 93
SIR + CNI
–
Based Therapy
:
N=9: SIR + TAC + Prednisolone
N = 20: SIR + Cyclosporine + Prednisolone
Steroid dosing: NA
*1/*1
*1/*3
n
=7
264 ± 221
*3/*3
n
=22
268 ± 141
Mourad et al. (2005)
Cross
–
Sectional Study
N = 85 KT
Self
–
reported race: 82 Caucasian, 2 African, 1 South Asian
Age: 52.3 ± 13 years
Child: No
Included: stable post
–
KT, 6.2
–
285.3 months post
–
KT
Excluded: History of graft rejection or altered renal function
leading to modifying drug doses during 2 months before
the study; pts taking drugs that precipitate DDI
A priori power analysis: NA
Initial dose: NA
Target trough:
5
–
15 ng/mL
PK sampling: whole blood collected 12-hour post
–
dose
Was PK sampling at steady state?: NA
Quantitation: LC
–
MS/MS
LOD/LOQ: NA
ISx regimen: SIR with Prednisolone (
n
= 81),
and MMF (
n
= 27) or Azathioprine
(
n
= 12); SIR and tacrolimus (
n
= 24)
c
Steroid dosing:
Weight
–
adjusted prednisolone dose per day was not
signi
fi
cantly different between the
CYP3A5
expressors vs. non
–
expressors
Allele
Trough/(Dose/Weight)
(ng/mL) per (mg/kg)
Median (range)
*1/*1
*1/*3
n
=7
c
176 (102
–
260)
*3/*3
n
=78
c
169 (46.2
–
1093)
Le Meur et al. (2006)
Clinical Trial
N= 21 KT
Self
–
reported race: 21 Caucasian
Age:
*3/*3
: 51.0 ± 13 years
*1/*1
&
*1/*3:
40.5 ± 17.3 years
Child: No
Inclusion: NA
Excluded: pts taking drugs that precipitate DDI
A priori power analysis: NA
Initial dose: 15 mg/day loading dose days 1 and
2, 10 mg/day ×7 days, then titrated to target trough
Target trough: 10
–
15 ng/mL
PK sampling: Whole blood was collected immediately
before dose administration, then at 0.33, 0.66, 1, 1.5,
2, 3, 4, 6, 9 hours after dose administration, on weeks 1
(W1), week 2 (W2), 1 month 1 (M1), 3 months (M3).
In W1 and W2, two additional samples were obtained
at 12 and 24 hours.
Was PK sampling at steady state?: NA
Quantitation: LC
–
MS
LOD: 0.5 ng/mL
LOQ: 1 ng/mL
ISx regimen: SIR, MMF, Thymoglobulin × 5 days, steroids
Steroid dosing: Methylprednisolone 250 mg IV pre
–
and
post
–
surgery, then oral prednisolone 1 mg/kg/day
days 1
–
7, 0.5 mg/kg/day days 8
–
14, taper by 5 mg/day
each week down to 20 mg/day, decrease by 2.5 mg/day
each week down to 10 mg/day, dose maintained for 1 month then decrease by 2.5 mg/day each week until
complete stop if possible.
Allele
Trough/Dose
(ng/mL) per (mg)
Mean (range)
*1/*1
*1/*3
n
=3
W1: 0.45 (0.35
–
1.23)
W2: 0.34 (0.13
–
0.37)
M1: 0.85 (0.50
–
0.87)
M3: 0.94 (0.33
–
1.30)
*3/*3
n
=18
W1: 1.53 (0.78
–
5.44)
W2: 1.61 (0.50
–
9.10)
M1: 2.16 (1.06
–
5.07)
M3: 2.56 (0.92
–
6.66)
(Continued on following page)
Frontiers in
Pharmacology
frontiersin.org
06
Shen et al.
10.3389/fphar.2023.1126981
TABLE 2 (
Continued
) Pharmacogenetic association of
CYP3A5
rs776746 genotype with sirolimus pharmacokinetics. Publications are organized in order of publication date, starting with the oldest. Gray-shade
d cells show
results where the
CYP3A5
genotype was associated with a statistically signi
fi
cant difference (
p
<
0.05) in sirolimus pharmacokinetics.
Author
Study design/
Study population
Sirolimus dosing and pharmacokinetic sampling
Immunosuppressant regimen
a
CYP3A5
Allele
Sirolimus PK
endpoint
Renders et al. (2007)
Prospective clinical study
N= 20 KT
Self
–
reported race: 20 Caucasian
Age: 54.1 ± 11.1 years
Child: No
Included: clinically stable pts, who had reached steady-state on SIR
Excluded: NA
A priori power analysis: NA
Initial dose: NA
Target trough: NA
PK sampling: trough (n=20); AUC (n=10 of the original 20 pts):
before and 0.5, 1, 2, 3, 4, 6, 8, 9, 10, 11, 12, and 24 hours after administration of a single dose
of sirolimus
Was PK sampling at steady state?: Yes
Quantitation: LC/MS
LOD/LOQ: NA
ISx regimen: SIR ± MMF ± Prednisolone
Steroid dosing: Prednisolone: 5
–
10 mg/day
Allele
Dose/Trough
(× 10
3
L)
Mean ± SD
*1/*1
n
=0
d
0.6 ± 0.1
*1/*3
n
=4
0.4 ± 0.2
*3/*3
n
=16
0.6 ± 0.4
AUC
0
–
24hr
/Dose
(ng × hr /mL per mg)
*1/*1
*1/*3
n
=3
56.3 ± 5.3
*3/*3
n
=7
118 ± 81.8
Miao et al. (2008)
Clinical Trial
N= 47 KT
Self
–
reported race: 47 Chinese (Han nationality)
Age: 42 ± 15 years
Child: No
Included: KT pts, stable graft function
Excluded: pts taking drugs that precipitate DDI with SIR,
except 3 pts receiving CNI + SIR
A priori power analysis: NA
Initial dose: NA
Target trough: NA
PK sampling: NA
Was PK sampling at steady state?: NA
Quantitation: HPLC
LOD/LOQ: NA
ISx regimen: SIR, MMF, steroids
Steroid dosing: NA
Allele
Trough/(Dose/Weight)
(ng/mL) per (mg/kg)
Mean ± SD
*1/*1
*1/*3
n
=21
318 ± 113
*3/*3
n
=26
397 ± 129
Zochowska et al. (2012)
Retrospective
N = 100 KT
Self
–
reported race: NA (Poland)
Age: 48.4 ± 11.5 years
Child: No
Included: KT
Excluded: NA
A priori power analysis: NA
Initial dose: NA
Target trough: NA
PK sampling: Whole blood
Was PK sampling at steady state?: NA
Quantitation: HPLC/UV
LOD/LOQ: NA
ISx regimen: SIR, MMF OR Azathioprine, GS (
n
= 64) OR
SIR, Cyclosporine or TAC, GS
e
(
n
= 36)
Steroid dosing: NA
Allele
Trough/(Dose/Weight)
(ng/mL) per (mg/kg)
Mean ± SD
*1/*3
n
=5
294 ± 181
*3/*3
n
=50
349 ± 209
(Continued on following page)
Frontiers in
Pharmacology
frontiersin.org
07
Shen et al.
10.3389/fphar.2023.1126981
TABLE 2 (
Continued
) Pharmacogenetic association of
CYP3A5
rs776746 genotype with sirolimus pharmacokinetics. Publications are organized in order of publication date, starting with the oldest. Gray-shade
d cells show
results where the
CYP3A5
genotype was associated with a statistically signi
fi
cant difference (
p
<
0.05) in sirolimus pharmacokinetics.
Author
Study design/
Study population
Sirolimus dosing and pharmacokinetic sampling
Immunosuppressant regimen
a
CYP3A5
Allele
Sirolimus PK
endpoint
Lee et al. (2014)
Clinical Trial
N = 85 KT
Self
–
reported race: Chinese
–
Han nationality
Age: 42.9 ± 10.4 years
Child: No
Included: Stable KT treated with SIR for
>
3 months
Excluded: pts taking drugs that precipitate DDI
A priori power analysis: yes
Initial dose: NA
Target trough:
5
–
10 ng/mL
PK sampling: whole blood samples drawn 24 hours after the
previous dose (before the next dose)
Was PK sampling at steady state?: NA
Quantitation: HPLC
Detection: NA
LOD/LOQ: NA
ISx regimen: SIR, MMF, Prednisone
Steroid dosing: NA
Allele
Trough/(Dose/Weight)
(ng/mL) per (mg/kg)
Mean ± SD
*1/*1
n
=8
*1/*3
n
=36
200 ± 75.2
*3/*3
n
=41
290 ± 92.1
Wang et al. (2014)
Open
–
label non
–
randomized clinical trial
N = 24 KT
Self
–
reported race: Chinese
Age: 39.7 ± 11.1 years
Child: No
Included: at least 2 months after primary or secondary KT,
stable sirolimus dose for
>
2 weeks
Excluded: pregnant/nursing; prior or concurrent non
–
renal
transplants; rejection in preceding 4 weeks, pts taking drugs
that precipitate DDI with SIR or affect drug absorption
A priori power analysis: NA
Initial dose: NA
Target trough: NA
PK sampling:
trough: immediately before the sirolimus dose on days 1, 2, 3.
AUC: 0.5, 1, 1.5, 2, 2.5, 3, 4, 5, 6, 8, 12, and 24 hours post-dose on day 3.
Was PK sampling at steady state?: Yes
Quantitation: LC
–
MS/MS
LOQ: 0.25 ng/mL
LOD: NA
ISx regimen: SIR + Prednisone (
n
= 23); MMF (
n
= 21); Cyclosporine (
n
=9)
Steroid dosing: NA
Allele
Apparent Oral Clearance
(L/F/hour)
Median (range)
*1/*1
n
=3
15.8 (12
–
22)
*1/*3
n
=8
10.9 (6
–
14)
*3/*3
n
=13
7.3 (3
–
16)
Khaled et al. (2016)
Retrospective case series
N = 173 HCT
Self
–
reported race: 91 Caucasian, non
–
Hispanic, 52 Hispanic, 23 Asian/Paci
fi
c
Islander, 7 other
Age: 46 (10
–
70) years
Child: No
Included: Allogeneic HCT
Excluded: 4 of the original 177 genotyped were excluded due to low-quality
genotype sample
A priori power analysis: NA
Initial dose:
12 mg (loading dose) on day
–
3, followed by 4 mg/day, with
subsequent doses personalized to target levels
Target trough:
3
–
12 ng/mL
PK sampling: whole blood samples; time of sample collection relative to dose is NA;
collected twice weekly for 100 days, reported results from
fi
rst 14 days post
–
HCT
Was PK sampling at steady state?: NA
Quantitation: microparticle enzyme immunoassay
LOD/LOQ: NA
ISx regimen: SIR + TAC ± Methotrexate
Steroid dosing: not used
Allele
Trough/Dose
(ng/mL) per (mg)
Median (Range)
*
1/*1
n
=8
f
2.6 (1.8
–
8.9)
*1/*3
n
=40
f
2.0 (0.6
–
5.6)
*3/*3
n
= 121
2.1 (0.6
–
12.3)
(Continued on following page)
Frontiers in
Pharmacology
frontiersin.org
08
Shen et al.
10.3389/fphar.2023.1126981
TABLE 2 (
Continued
) Pharmacogenetic association of
CYP3A5
rs776746 genotype with sirolimus pharmacokinetics. Publications are organized in order of publication date, starting with the oldest. Gray-shade
d cells show
results where the
CYP3A5
genotype was associated with a statistically signi
fi
cant difference (
p
<
0.05) in sirolimus pharmacokinetics.
Author
Study design/
Study population
Sirolimus dosing and pharmacokinetic sampling
Immunosuppressant regimen
a
CYP3A5
Allele
Sirolimus PK
endpoint
Li et al., (2015)
Clinical study
N = 43 KT
Self
–
reported race: Chinese
Age: 35 (34
–
46) years
Child: No
Included:
fi
rst KT, SIR for
>
1 month, stable post
–
KT without rejection
Excluded: NA
A priori power analysis: NA
Initial dose:
0.04
–
0.06 mg/kg/day
Target trough: 5
–
10 ng/mL
PK sampling: immediately before the next dose
Was PK sampling at steady state?: NA
Quantitation: automated enzyme immunoassay analyzer
LOD/LOQ: NA
ISx regimen: SIR, MMF, Prednisolone
Steroid dosing:
Methylprednisolone 1000 mg IV at the time of KT, 500 mg IV next 2 days,
followed by 80 mg/day oral Prednisone tapered to 10 mg/day to 20 mg until
3 months post
–
KT, reduced to 5 mg/day or discontinued
Allele
Trough/(Dose/Weight)
(ng/mL) per (mg/kg/day)
Median (Range)
*1/*1
*1/*3
n
=20
249 (248
–
410)
*3/*3
n
=23
389 (294
–
538)
Rodriguez-Jimenez et al.
(2017)
Retrospective cohort study
N = 48 KT
Self
–
reported race: NA
Age: 58 ± 9 years
Child: No
Included: Age
>
18 years, received KT between 2002
–
2006,
Excluded: pts taking drugs that precipitate DDI with SIR
A priori power analysis: NA
Initial dose: NA
Target trough: NA
PK sampling: Whole blood samples; time of sample collection relative
to dose is NA; collected at 1 week (W1), 2 weeks (W2), 1 month (M1),
3 months (M3), 6 months (M6)
g
Was PK sampling at steady state?: Yes
Quantitation: microparticle enzyme technique
LOD/LOQ: NA
ISx regimen: SIR, MMF, Steroid
Steroid dosing: NA
Allele
Trough/(Dose/Weight)
(ng/mL) per (mg/kg)
Mean ± SD (n)
*1/*1
n
=0
*1/*3
n
=8
W1
106 ± 29.3
(
n
=3)
W2
79.4 ± 45.9
(
n
=3)
W1+2
92.7 ± 37.5
(
n
=6)
M3
220 ± 85.9
(
n
=2)
M6
266
(
n
=1)
*3/*3
n
=39
W1
193 ± 133
(
n
= 33)
W2
140 ± 65.5
(
n
= 25)
W1+2
179 ± 116
(
n
= 39)
M3
277 ± 236
(
n
=15)
M6
233 ± 77.9
(
n
=13)
(Continued on following page)
Frontiers in
Pharmacology
frontiersin.org
09
Shen et al.
10.3389/fphar.2023.1126981
TABLE 2 (
Continued
) Pharmacogenetic association of
CYP3A5
rs776746 genotype with sirolimus pharmacokinetics. Publications are organized in order of publication date, starting with the oldest. Gray-shade
d cells show
results where the
CYP3A5
genotype was associated with a statistically signi
fi
cant difference (
p
<
0.05) in sirolimus pharmacokinetics.
Author
Study design/
Study population
Sirolimus dosing and pharmacokinetic sampling
Immunosuppressant regimen
a
CYP3A5
Allele
Sirolimus PK
endpoint
Zhang et al. (2017)
Clinical Trial
N = 31 Healthy male volunteers
Self
–
reported race: Chinese
Age: 19
–
27 years
Child: No
Included: Body mass index 18
–
24 kg/m
2
, had stopped any
other drug therapy for 2 weeks before study participation
Excluded: History of drug allergies
A priori power analysis: NA
Initial dose: 5 mg once
Target trough: Not applicable
PK sampling: before and 0.5, 1, 1.5, 2, 2.5, 3, 4, 6, 8, 12, 24, 48, 72,
96, 120, and 144 hour after dose
Was PK sampling at steady state?: NA
Quantitation: LC
–
MS/MS
LOD: NA
LOQ: 0.5 ng/mL
ISx regimen: Not applicable
Steroid dosing: Not used
Allele
AUC
0
–
144hr
(hr × ng/mL)
Mean ± SD
*1/*1
n
=2
314 ± 129
h
*1/*3
n
=14
440 ± 146
*3/*3
n
=15
550 ± 138
Abbreviations: CNI, Calcineurin inhibitor; CYP, Cytochrome P450 Enzyme; DDI, Drug
–
Drug interaction with sirolimus; expressors,
*1/*1
and
*1/*3
genotypes which encode for CYP3A5 protein expression; F, fraction of sirolimus dose absorbed; HCT, Hematopoietic
cell transplant; HPLC, high
–
performance liquid chromatography; hr, Hour; ISx, Immunosuppression; KT, Kidney transplant; LC
–
MS, Liquid Chromatography Mass Spectrometry; LLOQ, Lower limit of Quantitation; LOD, Limit of Detection; LOQ, Limit of
Quantitation; MMF, Mycophenolate mofetil; CYP3A5 protein non-Expressors,
*3/*3
genotype; NA, Not Available; PK, Pharmacokinetic; Pts, Patients; SIR, Sirolimus (rapamycin); SNP, Single nucleotide polymorphism; TAC, Tacrolim
us; UV, Ultra
–
Violet
spectroscopy.
a
Steroids affect CYP3A activity
McCune et al. (2000)
and sirolimus pharmacokinetics
Cattaneo et al. (2004)
and
Mourad et al. (2005)
.
b
The number of participants in the SIR
–
based therapy group differed between 51 stated participants, but only 31 were purportedly included based on the description of CYP3A5 expressors (
n
= 18) and CYP3A5 non-expressors (
n
= 13).
c
The additional ISx administered to the sirolimus and tacrolimus was not stated; no statistically signi
fi
cant difference was observed in this sirolimus pharmacokinetic endpoint in the 24 participants receiving sirolimus and tacrolimus.
d
This publication stated there were no
CYP3A5*1/*1
patients, but sirolimus pharmacokinetic data were reported for this genotype.
e
Assuming
“
GS
”
is an abbreviation for glucocorticoids, but this abbreviation was not de
fi
ned.
f
This study is the only one that is suf
fi
ciently powered in this table based in a power analysis, published in 2015 (
Emoto et al., 2015b
), using pre-dose concentrations simulated with the PBPK model indicated that at least 80 participants in an enrichment design, 40
CYP3A5 expressers and 40 non-expressers, would be required to detect a signi
fi
cant difference in the predicted trough concentrations at 1 month of therapy (
p
<
0.05, 80% power).
g
Time
–
points after 6 months were not reported because participants were lost to follow
–
up, leading to an insuf
fi
cient sample size for comparison.
h
Apparent oral clearance (L/hour) also differed based on
CYP3A5
genotype (
p
<
0.05).
Frontiers in
Pharmacology
frontiersin.org
10
Shen et al.
10.3389/fphar.2023.1126981
concentration/dose ratio at weeks 1 and 2 post
–
KT values
(
Rodriguez-Jimenez et al., 2017
).
In addition to
CYP3A5
rs776746, other
CYP3A5
SNPs
(i.e., rs4646453 and rs15524) have been evaluated for their
association with sirolimus pharmacokinetics (
Tamashiro
et al., 2017
;
Liu et al., 2021
). Liu reported that rs776746 is in
strong linkage disequilibrium with rs4646453 and rs15524 in
69 Chinese KT recipients. In rs4646453, homozygous AA alleles
were associated with lower dose
–
adjusted trough
concentrations when compared to homozygous CC (
p
<
0.001), with CC being the more dominant genotype (
n
= 41)
in the study population (
Liu et al., 2021
). In rs4646453,
homozygous GG (
n
=6)hadlowerdose
–
adjusted trough
concentrations compared to homozygous AA (
n
= 36) (
p
<
0.001). Tamashiro also evaluated rs15524 and only found a
signi
fi
cant association between genotype and dose
–
adjusted
trough concentration at 9 months post
–
KT (
p
<
0.05)
(
Tamashiro et al., 2017
).
Over the past 20 years, there have been 14 studies of the
CYP3A5
rs776746 genotype to sirolimus pharmacokinetic
phenotype studies in transplant patients.
Table 2
summarizes
the studies evaluating the
CYP3A5
rs776746 genotypes
associated with sirolimus pha
rmacokinetics. Nine of the
fourteen studies included i
n this review found a signi
fi
cant
association between the sirolimus adjusted trough value or
AUC and
CYP3A5
genotype (
Anglicheau et al., 2005
;
Le
Meur et al., 2006
;
Miao et al., 2008
;
Lee et al., 2014
;
Li et al.,
2015
;
Rodriguez-Jimenez et al., 2017
;
Tamashiro et al., 2017
;
Zhang et al., 2017
;
Liu et al., 2021
). The majority (80%) of these
studies were conducted with less than 100 participants. An
a
priori
power calculation could not be found in these studies,
which leaves the question if statistically insigni
fi
cant results
were due to an underpowered sample size. Notably, a power
analysis published in 2015, using pre-dose concentrations
simulated with the PBPK mode
l,indicatedthatatleast
80
participants
in
an
enrichment
design,
40 CYP3A5 expressers, and 40 non-expressers, would be
required to detect a signi
fi
cant difference in the predicted
trough concentrations at 1 month of therapy (
p
<
0.05, 80%
power) (
Emoto et al., 2015b
). Only one of the studies in
Table 2
has 40 CYP3A5 expressors (
Khaled et al., 2016
). Furthermore,
the studies were in heterogenous patient populations, with
inconsistent eligibility criteria regarding drugs that
precipitate a DDI with sirolimu
s. In addition, many studies
had varying use of corticosteroids which affect CYP3A activity
(
McCune et al., 2000
) and sirolimus pharmacokinetics
(
Cattaneo et al., 2004
;
Mourad et al., 2005
). Thus, the studies
were too heterogeneous and lacked adequately powered and
suf
fi
ciently controlled studies for i
t to be feasible to establish a
CYP3A5
genotype to sirolimus pha
rmacokinetic phenotype
association. Novel
CYP3A5
haplotypes are being identi
fi
ed
and may yield insightful results (
Rodriguez-Antona et al.,
2022
). Thus, we stress collaborative efforts to improve the
accessibility of pharmacogenetic information to the entire
pharmacogenetics communi
ty through the PharmGKB
(
Rodriguez-Antona et al., 2022
). However, more research is
needed regarding using preemptive
CYP3A5
-guided sirolimus
for children.
5.2 Pharmacometabolomics
Metabolomics, which is the study of small molecule metabolite
pro
fi
les in biological samples, is an additional promising new
technology in precision medicine (
Nicholson et al., 2002
;
Clayton
et al., 2006
;
Clayton et al., 2009
;
Phapale et al., 2010
;
Wishart, 2019
).
Metabolomic experiments are occasionally categorized as targeted
or untargeted (
Wishart, 2019
). The targeted metabolomic
analysis involves evaluating a selected group of metabolites,
often quantifying the metabolite concentrations relative to an
authentic reference standard. In untargeted experiments, an
unbiased approach is used, and all of the metabolites detected
above the sensitivity threshold of the technology employed are
analyzed. We demonstrated that pre
–
dose metabolomic
pro
fi
ling of plasma could predi
ct busulfan clearance (
Lin
et al., 2016
;
Navarro et al., 2016
;
McCune et al., 2022b
).
Although the blood concentrations of many metabolites are
tightly regulated (
Homuth et al., 2012
), we have found that the
plasma metabolome does change af
ter treatment with alkylating
agents such as busulfan or cyclophosphamide (
McCune et al.,
2022a
;
McCune et al., 2022b
).
The urinary metabolome is also of interest, as urine metabolite
concentrations can vary widely and may serve as a
“
readout
”
of
metabolic capacities that are not detected in blood (
Schlosser et al.,
2020
). For tacrolimus, which is eliminated
via
similar drug-
metabolizing enzymes and transporters as sirolimus, predose
urine metabolites are associated with tacrolimus
pharmacokinetics (
Phapale et al., 2010
). Further research is
needed in this area, especially accounting for reduced kidney
function resulting from concomitant cyclosporine or tacrolimus
with sirolimus in transplant recipients. In 1,627 participants of
the UK Biobank with reduced kidney function, the combination
of metabolite quantitative trait loci revealed novel candidates for
biotransformation and detoxi
fi
cation reactions (
Schlosser et al.,
2020
). Thus, these novel biotransformation and detoxi
fi
cation
reactions could in
fl
uence the predose urinary metabolome in
patients treated with sirolimus. Furthermore, the potential for
renal metabolism of sirolimus should be considered because renal
CYP3A can metabolize CYP3A-substrates (
Dai et al., 2004
;
McCune
et al., 2005
).
6 Point-of-care sample collection of
dried blood spots for precision dosing
of sirolimus
Another approach to improving the precision dosing of
sirolimus is simplifying the collection of the whole blood samples
used for sirolimus TDM. There has been extensive interest in using
dried blood spot (DBS) as a point-of-care method for obtaining
blood samples to be used in sirolimus TDM. DBS sampling is a
blood sampling method alternative to venipuncture and requires less
blood volume, which makes it an attractive option for children. The
DBS sampling process is not dif
fi
cult to perform and does not
require a trained phlebotomist for blood spot collection (
Fokkema
et al., 2009
;
Wagner et al., 2016
). The patient provides a
venipunctures capillary blood drop, places the blood onto a
fi
lter
card, and subsequently allows the blood spot to dry. The DBS sample
Frontiers in
Pharmacology
frontiersin.org
11
Shen et al.
10.3389/fphar.2023.1126981
is then sent to a laboratory to quantify the sirolimus
concentration with LC
–
MS (
Wagner et al., 2016
). DBS
sampling may impro
ve patient satisfaction by reducing
commute times and possibly improving the precision dosing
of sirolimus (
Dickerson et al., 2015
). For DBS to replace whole
blood samples, it is important to evaluate if they provide similar
sirolimus concentrations. Sirol
imus concentration in DBS may
belowerthaninvenouswholebloodsamplesbecausedrug
concentrations tend to be lower in capillary blood (
Klak et al.,
2019
).
Veenhof et al. (2019)
collected DBS and whole blood
samples from patients and found that 76.9% of the samples were
within acceptable limits.
The accurate quantitation of sirolimus concentration in DBS
sampling depends on various factors, summarized in
Supplementary
Table S2
. These factors range from 1. patient characteristics, 2.
depositing the blood drop on the
fi
lter card; 3. the effect of how long
it takes for the blood to dry on the
fi
lter card (i.e., drying time); 4.
storing and transporting the DBS sample from the patient
’
s home to
the laboratory; 5. punching the DBS section for quantitation; 6.
extracting and quantitating sirolimus concentrations.
Table 3
summarizes the
fi
ndings regarding the quantitation of
sirolimus concentrations (abbreviated [SIR] in
Table 3
) from DBS
samples. The patient
’
s characteristics, speci
fi
cally their hematocrit
and the sirolimus concentration at the time of the DBS collection,
appreciably change the accuracy of sirolimus concentrations in DBS.
Varying hematocrit in
fl
uences the blood viscosity, the drying time
needed for a DBS, and the potential interference with analyte
recovery (
De Kesel et al., 2013
;
Klak et al., 2019
). Low hematocrit
(less than 0.20 L/L) and high hematocrit (greater than 0.5 L/L) are
associated with lower sirolimus recovery from DBS (
den Burger et
al., 2012
;
Koster et al., 2013
). To obtain reliable sirolimus
concentrations, it is optimal that the patient
’
s hematocrit range is
between 0.23 and 0.50 L/L (
den Burger et al., 2012
;
Koster et al.,
2013
;
Koster et al., 2017
). Therefore, DBS samples with high or low
hematocrit concentrations must be corrected and interpreted
cautiously (
Klak et al., 2019
). Also, better accuracy is achieved
when sirolimus concentrations are at least 3.0 ng/mL; however,
this may limit the use of DBS as trough concentrations may be
lower than 3.0 ng/mL (
Koster et al., 2015a
;
Koster et al., 2017
).
The second factor in
fl
uencing sirolimus concentrations from
DBS is depositing the blood spot on the
fi
lter card. Hydrogen bridges
are known to form between sirolimus and cellulose
fi
lters (
Klak et al.,
2019
). Because the sirolimus concentration is in
fl
uenced by the type
of
fi
lter card used, it is recommended that a laboratory use only one
type of
fi
lter card for calibration purposes (
Koster et al., 2015a
). The
Whatman FTA DMPK
–
C, 31 ET CHR, and Whatman DMPK
–
C
fi
lter cards have consistent sirolimus extraction recovery (
Koster
et al., 2013
;
Koster et al., 2015a
). However, the Whatman 903 and
Ahlstrom 226
fi
lter cards are preferred because of their untreated
cellulose
fi
lter and because they comply with CLSI guidelines (
Klak
et al., 2019
;
CLSI document EP09-A3, 2013
). In addition to the
effects of the
fi
lter card, the blood spot volume and homogeneity of
its placement within one spot affect the accuracy of sirolimus
concentrations from DBS samples. Filter card oversaturation by
blood spot volumes equal to or greater than 100 μL (
den Burger et
al., 2012
;
Sadilkova et al., 2013
) and volumes less than 20 or 30 μL
can lead to inaccurate results (
den Burger et al., 2012
;
Koster et al.,
2013
;
Koster et al., 2017
). Patients are more likely to provide low
blood spot volumes when self
–
sampling than the recommended
50 μL blood spot volumes (
Klak et al., 2019
). Dickerson and others
evaluated point-of-care (i.e., at-home) collection by providing and
educating families about how to collect and mail DBS samples back
to the laboratory (
Dickerson et al., 2015
). A small negative, but not
statistically signi
fi
cant, bias between DBS and whole blood samples
was found. The sirolimus concentrations in the DBS samples were
within clinically acceptable limits (
Dickerson et al., 2015
). Although
this is encouraging for self-sampling, additional studies in children
are needed to evaluate if they can provide the recommended blood
spot volume studied to date (
den Burger et al., 2012
;
Koster et al.,
2013
;
Koster et al., 2017
;
Klak et al., 2019
;
Veenhof et al., 2019
).
The third and fourth factors in
fl
uencing sirolimus
concentrations are drying the
fi
lter card and storing and
transporting the DBS on the
fi
lter card, respectively. DBS
samples are recommended to be dried at room temperature and
away from light for at least 24 hr to allow for accurate hematocrit
effects during sirolimus recovery (
Koster et al., 2015b
;
Klak et al.,
2019
). DBS samples can be stored for 20
–
29 weeks in the lab
at
−
20
°
C before losing the stability of sirolimus concentrations
(
Koster et al., 2015b
;
Veenhof et al., 2019
). All DBS samples
require spot-checking for appropriate volume size (
Veenhof
et al., 2019
). Sirolimus DBS samples degrade rapidly within 24 hr
at 60
°
C, but sirolimus is relatively stable at 25
°
C(
Sadilkova et al.,
2013
;
Klak et al., 2019
). Therefore, patients using DBS sampling
must be educated on properly collecting, storing, and transporting
their DBS samples.
The
fi
fth and sixth factors in
fl
uencing DBS
’
sirolimus
concentrations are punching out the DBS section for quantitation and
the subsequent extraction and quantitation of the sirolimus
concentrations, respectively. Punch size and location of the DBS do
not appear to in
fl
uence sirolimus concentrations (
den Burger et al., 2012
;
Sadilkova et al., 2013
;
Koster et al., 2017
;
Klak et al., 2019
). The highest
extraction and recovery rates of sirolimus were found at high hematocrit
and low sirolimus concentrations (
Table 3
)(
Koster et al., 2013
). When
utilizing
fl
ow-through desorption (FTD) with LC-MS-enhanced
temperature desorption, sirolimus recovery improved signi
fi
cantly
(
Hempen et al., 2015
). However, further trials are still needed to
con
fi
rm whether FTD
–
LC
–
MS
–
MS can be established as an accurate
DBS tool.
In summary, some factors (i.e., hematocrit, sirolimus concentration,
fi
lter card, drying time) in
fl
uence sirolimus concentrations from DBS.
However, patient education is necessary for parents to collect suf
fi
cient
blood spot volumes at the correct time (
Dickerson et al., 2015
;
Klak et al.,
2019
). In addition, potentially losing samples in the mail is an ongoing
concern (
Dickerson et al., 2015
;
Urquhart and Knauer, 2015
). Therefore,
precision dosing of sirolimus using DBS samples is not recommended for
children.
7 Point-of-care collection of saliva or
sweat for precision dosing of sirolimus
Other matrices, such as saliva or sweat, can also be used in TDM.
As a common alternative to a blood sample, the non-invasive and
easily accessible nature of saliva samples makes it optimal for TDM
in outpatient settings and potentially in children. Furthermore, with
the help of PBPK modeling, the system drug exposure may be
Frontiers in
Pharmacology
frontiersin.org
12
Shen et al.
10.3389/fphar.2023.1126981
TABLE 3 Studies to quantitate sirolimus concentrations ([SIR]) in dried blood spots prepared in the laboratory or obtained from patients taking siro
limus.
Publications are organized in order of publication of the researcher group, starting with the group
’
s oldest publication.
References
Methods: DBS samples and [SIR] quantitation
Results and interpretation for [SIR]
a
Den Burger et al., (2012)
VU
University Medical Center
DBS Sample Preparation
•
Lab staff performed all experiments using prepared DBS by combining
purchased EDTA whole blood with plasma.
•
Hct average: 0.33 (range 0.22
–
0.41)
•
DBS volume: 20, 40, 60, 80, and 100 μL
•
[SIR]: Not clearly stated. Appear to be the low QC of [SIR] 1.57 ng/mL and the
high QC of 23.5 ng/mL. Used an unknown volume of [SIR] 200,000 ng/mL
b
to
spike into the blood for subsequent experiments
•
Compared DBS to whole blood samples collected in EDTA
Hct Effect (1)
•
Low Hct levels were associated with poorer recovery outcomes, but the speci
fi
c
value for [SIR] was not reported.
Blood Spot Volume (2.3)
•
Volume of DBS (40
–
100 μL) was within 85%
–
115% at two [SIR]
•
20 μL DBS volume had a white edge of the unspotted paper which contributed
to inaccuracy
Punch Size and Location (5)
[SIR] Quantitation using LC-MS
•
Calibration curves made with sirolimus
–
free EDTA whole blood (Hct: NA)
•
Matrix effect =
−
0.63% and were within desired limits
•
STD curve: 1.24, 2.24, 6.71, 11.2, 17.9, 35.8 ng/mL
•
LOQ: 1.12 ng/mL, which had an accuracy and precision of 80%
–
120%
•
LOD: NA
•
QCs were acceptable with accuracy and precision of 85%
–
115% over 6 replicates
•
a priori
for acceptance: 85%
–
115% recovery
•
Punch location did not in
fl
uence accuracy. The peripheral punch/center punch
ranges from 100.2% (low QC) to 109.9% (high QC)
Sadilkova et al., (2013)
Seattle
Children
’
s Hospital
DBS Sample Preparation
•
Patient samples: EDTA whole blood sample obtained from children (
n
= 68)
taking SIR
•
Hct median level 30%
–
35% for all participants
•
Stability of [SIR] in patient
’
s samples (presumably whole blood) tested for
5 days at the following temperatures: 20
°
C, 25
°
C, 37
°
C, 60
°
C
•
DBS preparation by laboratory personnel
•
50 μL blood on Whatman 903 DBS card
•
DBS volume: 25, 35, 50, 75, and 100 μL
•
Samples dried for 3 hr at room temperature.
•
Stability of [SIR] DBS for QC evaluated at
−
20
°
C, 4
°
C, 25
°
C
Overall Conclusion
•
[SIR] in DBS correlates with [SIR] in whole blood
Hct (1)
•
No effect between Hct of 20%
–
45%
Blood Spot Volume (2.3)
•
No effect.
Stability of Analyte (4)
•
[SIR] degraded at 60
°
C within 24 hr.
[SIR] Quantitation using LC-MS
•
Calibration curves made with immunosuppressant
–
free EDTA whole blood,
with Hct of 30%
–
35%.
•
STD curve 1.2, 2.5, 5, 10, 20, and 40 ng/mL
•
LOQ and LOD: NA
•
Intra
–
run CV was 8.6% at 4 ng/mL and 5.9% at 20 ng/mL (
n
= 23 DBS)
•
Inter
–
run CVs were 14.8% at 4 ng/mL and 11.6% at 20 ng/mL (
n
= 25 DBS,
stored at
−
20
°
Cin
–
between quantitation, which occurred over 76 days)
•
a priori
for acceptance: NA
Punch Size and Location (5)
•
No effect
Dickerson et al. (2015)
Seattle
Children
’
s Hospital
DBS Sample Preparation
•
Patient samples: A trained phlebotomist collected paired capillary DBS and
venous blood samples.
•
25 sample pairs (i.e., DBS and venous blood were obtained within minutes of
each other from 34 children (median age: 13 years) who had received a solid organ
transplant
•
[SIR] compared in three different types of samples:
1. Venous blood sent to the clinical lab for quantitation;
2. That same venous blood was used to create a whole blood spot (WBS) and
stored until the DBS arrived;
3. A trained phlebotomist prepared the capillary DBS card. The DBS card was
provided to the family to take home. Families were instructed to send the DBS card
back to the hospital within 1 week.
Sirolimus dose range: 0.4
–
4 mg twice daily
[SIR] Quantitation using LC-MS, as described by
Sadilkova et al. (2013)
.
•
a priori
for acceptance: NA
Overall Conclusion
•
A small but statistically signi
fi
cant negative bias (0.6 ng/mL,
p
= 0.0011) was
observed between the venous blood to the capillary DBS mailed back to the
laboratory.
•
Analysis of [SIR] in DBS is possible, with the difference between venous and
capillary blood within clinically acceptable limits.
Extraction Recovery (6.2)
•
Comparing the venous whole blood to the DBS, the Bland
–
Altman analysis
showed a difference in the [SIR] in these samples, with a larger variation at high
[SIR]. In DBS, [SIR] were lower by a mean of 0.8 ng/mL (interquartile range =
1.9,
p
= 0.029)
•
Comparing the WBS to the DBS, there was no statistically signi
fi
cant difference
•
Comparing the venous blood to the WBS, [SIR] was lower by a mean of 1 ng/
mL (
p
= 0.003) in WBS.
•
There are varying effects on drug concentrations with collecting capillary blood.
Still, capillary draws often occur in clinical care, and the blood sample source
(i.e., capillary vs. venipuncture) is not distinguished clinically.
Hempen et al. (2015)
Spark
Holland
DBS Sample Preparation
•
Lab staff prepared using purchased whole blood
•
Varying amounts of plasma were added or removed to achieve different target
Hct values and [SIR].
•
Hct: 0.25 or 0.60
•
[SIR]: 1, 5, 50 ng/mL
•
Blood dried for at least 2 hr at room temperature.
Overall Conclusion
•
Temperature
–
enhanced desorption increased [SIR] recovery
Extraction Recovery (6.2)
•
With FTD
–
LC
–
MS
–
MS, Hct did not impact the recovery of [SIR] from
the DBS
[SIR] Quantitation using temperature-enhanced
fl
ow
–
through desorption
(FTD)
–
LC
–
MS
–
MS. FTD desorbs dried blood from the
fi
lter by perpendicular
fl
ushing solvent through the DBS in a chamber. The chamber
’
s inlet is connected
to a solvent pump, and its outlet is connected to a collection device. Using the FTD
removes the need to punch out the disc from a DBS.
•
STD curve: 0.2
–
100 ng/mL
•
LOQ and LOD: NA
•
Bias and QCs: NA
(Continued on following page)
Frontiers in
Pharmacology
frontiersin.org
13
Shen et al.
10.3389/fphar.2023.1126981