Identification of human cytochrome P450s that metabolise anti-parasitic drugs and predictions of in vivo drug hepatic clearance from in vitro data

Article abstract of DOI:10.1007/s00228-003-0636-9

Objective: Knowledge about the metabolism of anti-parasitic drugs (APDs) will be helpful in ongoing efforts to optimise dosage recommendations in clinical practise. This study was performed to further identify the cytochrome P₄₅₀ (CYP) enzymes that metabolise major APDs and evaluate the possibility of predicting in vivo drug clearances from in vitro data. Methods: In vitro systems, rat and human liver microsomes (RLM, HLM) and recombinant cytochrome P₄₅₀ (rCYP), were used to determine the intrinsic clearance (CL_int) and identify responsible CYPs and their relative contribution in the metabolism of 15 commonly used APDs. Results and discussion: CL_int determined in RLM and HLM showed low (r²=0.50) but significant (P<0.01) correlation. The GL_int values were scaled to predict in vivo hepatic clearance (CL_H) using the 'venous equilibrium model'. The number of compounds with in vivo human CL data after intravenous administration was low (n=8), and the range of CL values covered by these compounds was not appropriate for a reasonable quantitative in vitro-in vivo correlation analysis. Using the CL_H predicted from the in vitro data, the compounds could be classified into three different categories: high-clearance drugs (> 70% liver blood flow; amodiaquine, praziquantel, albendazole, thiabendazole), low-clearance drugs (< 30% liver blood flow; chloroquine, dapsone, diethylcarbamazine, pentamidine, primaquine, pyrantel, pyrimethamine, tinidazole) and intermediate clearance drugs (artemisinin, artesunate, quinine). With the exception of artemisinin, which is a high clearance drug in vivo, all other compounds were classified using in vitro data in agreement with in vivo observations. We identified hepatic CYP enzymes responsible for metabolism of some compounds (praziquantel - 1A2, 2C19, 3A4; primaquine - 1A2, 3A4; chloroquine - 2C8, 2D6, 3A4; artesunate - 2A6; pyrantel - 2D6). For the other compounds, we confirmed the role of previously reported CYPs for their metabolism and identified other CYPs involved which had not been reported before. Conclusion: Our results show that it is possible to make in vitro-in vivo predictions of high, intermediate and low CL_int drug categories. The identified CYPs for some of the drugs provide a basis for how these drugs are expected to behave pharmacokinetically and help in predicting drug-drug interactions in vivo.

Full text of DOI:10.1007/s00228-003-0636-9

Eur J Clin Pharmacol (2003) 59: 429–442
DOI 10.1007/s00228-003-0636-9
PHARMACOKINETICS AND DISPOSITION
Xue-Qing Li Æ Anders Bjorkman Æ Tommy B. Andersson
¨
Lars L. Gustafsson Æ Collen M. Masimirembwa
Identification of human cytochrome P s that metabolise anti-parasitic
4
50
drugs and predictions of in vivo drug hepatic clearance from in vitro data
Received: 17 December 2002 / Accepted: 16 June 2003 / Published online: 12 August 2003
ꢀ
Springer-Verlag 2003
Abstract Objective: Knowledge about the metabolism pyrantel, pyrimethamine, tinidazole) and intermediate
of anti-parasitic drugs (APDs) will be helpful in ongoing clearance drugs (artemisinin, artesunate, quinine). With
efforts to optimise dosage recommendations in clinical the exception of artemisinin, which is a high clearance
practise. This study was performed to further identify drug in vivo, all other compounds were classified using
the cytochrome P450 (CYP) enzymes that metabolise in vitro data in agreement with in vivo observations. We
major APDs and evaluate the possibility of predicting in identified hepatic CYP enzymes responsible for meta-
vivo drug clearances from in vitro data.
bolism of some compounds (praziquantel—1A2, 2C19,
Methods: In vitro systems, rat and human liver micro- 3A4; primaquine—1A2, 3A4; chloroquine—2C8, 2D6,
somes (RLM, HLM) and recombinant cytochrome P450 3A4; artesunate—2A6; pyrantel—2D6). For the other
(
(
rCYP), were used to determine the intrinsic clearance compounds, we confirmed the role of previously re-
CL ) and identify responsible CYPs and their relative ported CYPs for their metabolism and identified other
int
contribution in the metabolism of 15 commonly used CYPs involved which had not been reported before.
APDs. Conclusion: Our results show that it is possible to make
Results and discussion: CL determined in RLM and in vitro–in vivo predictions of high, intermediate and
int
2
HLM showed low (r =0.50) but significant (P<0.01) low CLint drug categories. The identified CYPs for some
correlation. The CLint values were scaled to predict in of the drugs provide a basis for how these drugs are
vivo hepatic clearance (CL ) using the ‘venous equilib- expected to behave pharmacokinetically and help in
H
rium model’. The number of compounds with in vivo predicting drug–drug interactions in vivo.
human CL data after intravenous administration was
low (n=8), and the range of CL values covered by these Keywords Anti-parasitic drug Æ Metabolism Æ
compounds was not appropriate for a reasonable Microsomes
quantitative in vitro–in vivo correlation analysis. Using
the CLH predicted from the in vitro data, the com-
pounds could be classified into three different categories:
high-clearance drugs (>70 % liver blood flow; amodi-
aquine, praziquantel, albendazole, thiabendazole), low-
clearance drugs (<30 % liver blood flow; chloroquine,
dapsone, diethylcarbamazine, pentamidine, primaquine,
entities (NCE) in drug discovery, drug absorption, dis-
tribution, metabolism, excretion and toxicology
Introduction
In efforts to reduce attrition rates of new chemical
(
ADME-Tox) studies have become a major activity in
X.-Q. Li Æ T. B. Andersson Æ C. M. Masimirembwa (&)
Department of Drug Metabolism and Pharmacokinetics
the pharmaceutical industry [1]. To ensure that only
drugs with minimal adverse reactions (ADR) are li-
censed, drug regulatory agencies like the Food and Drug
Administration (FDA) are also increasingly demanding
data on ADME-Tox in relation to new drug applica-
tions (NDA) [2]. This not only affects the drug discovery
and development process but also the optimal use of
drugs already on the market.
and Bioanalytical Chemistry, AstraZeneca R&D Mo
4
¨
lndal,
31 83 Molndal, Sweden
¨
E-mail: collen.masimirembwa@astrazeneca.com
Tel.: +46-31-7762201
Fax: +46-31-7763786
¨
X.-Q. Li Æ A. Bjorkman Æ C. M. Masimirembwa
Unit of Infectious Diseases,
Karolinska Institute Hospital, Stockholm, Sweden
Most anti-parasitic drugs (APDs) in use had already
been introduced in the 1940s, before bioanalytical
methods were available to evaluate ADME parameters
L. L. Gustafsson
Department of Clinical Pharmacology,
Huddinge Hospital, Karolinska Institute, Sweden

4
30
and optimise the pharmacokinetics (PKs) of the drugs. reports on the success and failure of making in vitro–in
Because chemotherapy is often the principal means to vivo correlations of drug clearance which could indicate
control parasitic diseases, optimisation of their use is limitations inherent in the models used and a need for
urgently needed. In addition, use of drug combinations further research [7]. The specific aim of this study was to
to increase therapeutic efficiency and to avoid the identify the major CYPs involved in the metabolism of
emergence of drug resistance is now being advocated in APDs and to predict in vivo clearance from in vitro
tropical medicine [3]. In recent years, the gross PKs of intrinsic clearance data. This was part of our group’s
some APDs has been studied in humans [4], but studies effort to elucidate the metabolism of APDs (Fig. 1.).
are also needed on the mechanistic basis for PK prop- We have previously evaluated the inhibitory effects [8]
erties, therapeutic or toxicological outcome observed in and inducing effects [9] of APDs on CYPs. This infor-
the use of these drugs.
mation will be useful in rationalising the PKs of these
There have been many methodological developments drugs and predicting potential drug–drug interactions.
in the study of drug metabolism. The increased avail-
ability of human tissue (liver microsomes, slices and
hepatocytes) and recombinantly expressed human drug
Materials and methods
metabolising enzymes are giving an insight into how
humans metabolise drugs (reviewed in [1]). The identifica- Chemicals
tion of cytochrome P450 (CYP)-specific marker reactions,
antibodies and chemical inhibitors selective for some
CYPs have improved the capacity to qualitatively and
quantitatively estimate the role of different enzymes in the
Substrates, metabolites and inhibitors were purchased from dif-
ferent companies: phenacetin, paracetamol, coumarin, 7-hydroxy-
coumarin, diclofenac, amodiaquine (AQ), albendazole (ABZ),
dextromethorphan, debrisoquine, diethylcarbamazine, pentami-
metabolism of test compounds [5, 6]. There are many dine, praziquantel (PZQ), primaquine, pyrantel, pyrimethamine,
Fig. 1 Structures of the 15 anti-
parasitic drugs studied

4
31
quinine, thiabendazole, tinidazole, quinidine, a-naphthoflavone, operated in the positive ion electrospray mode (except 7-hydroxy-
reduced NADPH (Sigma Chemical Co., St. Louis, MO); bufuralol, coumarin which was monitored in negative mode). The spray
1
¢-hydroxybufuralol, S-mephenytoin, 4¢-hydroxymephenytoin, 1¢- voltage was set to 4.0 kV, heated capillary temperature to 200 ꢁC,
hydroxymidazolam, 4-hydroxydebrisoquine, sulfaphenazole source CID set to off. Nitrogen was used as the sheath and auxil-
Ultrafine, Manchester, UK); paclitaxel, 6a-hydroxypaclitaxel, iary gas and set to 70 and 20 (arbitrary units), respectively. The
(
4
¢-hydroxydiclofenac, midazolam (GenTest Co., Woburn, MA); mass spectrometer was operated in the selected ion monitoring
artemisinin, dapsone, quercetin (Aldrich Chemical Co., Milwau- (SIM) mode. Instrument control, data acquisition and data eval-
kee, WI); ketoconazole (Janssen Biotech, Flander, NJ); ticlopidine, uation were performed using Xcalibur software (version 1.2,
dextrophan (ICN Biomedicals Inc., Aurora, Ohio). Metoprolol, Finnigan). The lower limit of quantification (LLOQ) was 0.1 lM
a-hydroxymetoprolol, and demethylated metoprolol were obtained for all the APDs investigated except artemisinin (LLOQ, 1 lM)
¨
from AstraZeneca (Molndal, Sweden). Bupropion, hydroxybu- and artesunate (LLOQ, 0.5 lM). The LLOQ was 0.05 lM for all
propion, chloroquine (CLQ), desethylchloroquine and N-deseth- the metabolites of CYP marker reactions.
ylamodiaquine were gifts from Karolinska Institute (Stockholm,
¨ ¨
Sweden). Dr. Michael Ashton (Goteborg University, Goteborg,
Sweden) kindly provided artesunate. All other reagents were of
analytical or HPLC grade.
Determination of intrinsic clearance
Intrinsic clearance (CLint) studies were performed for 15 APDs
with HLM, recombinant CYP enzymes (rCYPs) and RLM using
the substrate disappearance approach [10, 11]. The typical incu-
bation mixture consisted of 0.1–2 mg/ml HLM (0.5 mg/ml RLM
or 20 pmol rCYPs), 1.0 lM substrates, 1 mM NADPH in 0.1 M
phosphate buffer pH 7.4 in a final volume of 200 ll. After adding
NADPH to initiate reaction, ice-cold acetonitrile was added at 0,
Human liver microsomes and recombinant
cytochrome P450s (rCYPs)
Twenty human liver microsomes (HLM) were obtained from an in-
house bank of liver microsomes maintained at AstraZeneca Re-
search and Development (Mo
CYP1A2, 2A6, 2B6, 2C8, 2C9, 2C19, 2D6, 2E1 and 3A4) in indi-
vidual HLM were determined using diagnostic marker substrates.
¨
lndal, Sweden). CYP activities
10, 20, 30, 45, 60 min to stop the reaction. The CLint was deter-
mined as follows:
(
The following marker reactions were used: phenacetin demethyla- CLint ¼ ðln2 ꢀ incubation volumeÞ=
À
Á
tion (CYP1A2), coumarin 7-hydroxylation (CYP2A6), bupropion
hydroxylation (CYP2B6), paclitaxel 6a-hydroxylation (CYP2C8),
T
ꢀ Protein or enzyme amount
1=2
diclofenac 4¢-hydroxylation (CYP2C9), S-mephenytoin 4¢-hydrox- T1/2 was determined from the elimination rate constant k=ln2/T
1/
ylation (CYP2C19), bufuralol 1¢-hydroxylation, debrisoquine 2. For the clearance determination of artemisinin and artesunate,
-hydroxylation, dextromethorphan O-demethylation, metoprolol the substrate concentrations used were increased to 10 lM and
a-hydroxylation and demethylation (CYP2D6), chlorzoxazone 5 lM, respectively, to increase their response for LC/MS detection.
4
6
-hydroxylation (CYP2E1) and midazolam 1¢-hydroxylation, tes- The use of low substrate concentrations, usually 1.0 lM, to esti-
tosterone 6b-hydroxylation (CYP3A4). Pooled HLM were pre- mate drug clearance stems from assumptions derived from the
pared from a set of liver pieces of patients undergoing liver Michaelis–Menten kinetics: rate of metabolism (v)=Vmax[S]/
resections. Recombinant human CYP enzymes 1A1, 1A2, 2B6, 2C8, (K +[S]) and CLint=Vmax/Km (Michaelis–Menten equation).
m
2C9, 2C19, 2D6, and 3A4 were from yeast (AstraZeneca, Sweden). Under linear conditions, when [S] is 10 % or less of Km (which
CYP enzymes CYP2A6, 1B1, 2E1, 3A5, and 4A11 were from most drugs are therefore assumed to be at in vivo), the equation
lymphoblastoid cell lines (GenTest Corp, Woburn, MA). The reduces to v=(Vmax[S])/Km which can be rearranged to:
microsomal preparations were stored at )80 ꢁC until use.
CLint ¼ Vmax=Km ¼ rate of metabolism=½Sꢁ:
Detailed derivations of these equations and discussions of the
Incubation conditions with HLM, RLM and rCYP enzymes
assumptions are as presented previously [7, 12, 13].
All reactions were performed in 96-well plates. Each reaction
mixture consisted of the appropriate enzyme, substrate, 1 mM re-
duced nicotinamide adenine dinucleotide phosphate (NADPH), in In vitro–in vivo correlations
0
2
.1 M potassium phosphate buffer pH 7.4 in a final volume of
00 ll. The reactions were started by the addition of NADPH after The CLint in the incubation is expressed as ll/min/[mg (micro-
a preincubation of 5 min at 37 ꢁC. All reactions were stopped by somes) or pmol (enzyme)]. This can be scaled to the apparent
the addition of 150 ll ice-cold acetonitrile. After centrifugation at clearance (CLint,app.) for the whole liver (70 kg person with a 1.4-kg
4
liquid chromatography/mass spectrometry (LC/MS).
500 · g for 20 min, 50 ll supernatant was analysed by means of liver, 20-g liver/kg body weight). In the scaling process, the cur-
rently used factor is 45 mg HLM per gram liver. Using these
factors, the equation is [14]:
0
:693
ml incubation
45 mg microsomes
ꢀ
ln vitro T1=2 mg microsomes g liver
Chromatography
CLint;app
¼
ꢀ
The LC/MS system consisted of an HP 1100 system (Hewlett-
Packard, Palo Alto, CA) and a Finnigan LCQ ion trap mass
spectrometer (Finnigan Mat, San Jose, CA) employing an atmo-
20 g liver
ꢀ
kg body weight
H
spheric pressure ionisation interface. Chromatography was per- To estimate hepatic clearance (CL ) due to metabolism, a number
formed on a Symmetry C18 column (3.9 · 150 mm i.d.; 5 lm, of models have been proposed and include (a) the ‘well stirred’ or
Waters, Milford, Massachusetts). The mobile phases consisted of the ‘venous equilibrium’ model, (b) the parallel tube model, and (c)
acetonitrile as eluent A and 5 mM ammonium acetate buffer the dispersion model. Most in vitro–in vivo predictions are made
(
pH 3) as eluent B, or 10 % of acetonitrile in 12 mM formic acid using the venous equilibrium model due to its simplicity and the
solution (v/v) as eluent C. Gradient runs of combinations of these fact that studies have observed little difference in the values pre-
eluents at a flow rate of 1 ml/min are shown in Table 1. The dicted by the three models [13]. The hepatic clearance, CL
effluent was split with approximately 0.3 ml/min introduced into expressed as:
the mass spectrometer. Source parameters of mass spectrometer
H
, is
Q : ꢂ fuB ꢂ CLint
H
(e.g., spray voltage, temperature, gas flow rates, etc.) were indi- CLH
¼
Q
vidually optimised for each compound. The mass spectrometer was
þ fuB ꢂ CLint
H

4
32
Table 1 Liquid chromatography (LC)/mass spectrometry (MS) chromatographic conditions for 15 anti-parasitic drugs and 14 specific
cytochrome P450 (CYP) enzyme marker reactions analysis
a
Compound
HPLC conditions
Retention MS detection
+
time (min) [M+H] (m/z)
Mobile phase Linear gradient program of mobile phase
Antiparasitic drugs
Albendazole
Amodiaquine
Artemisinin
A and B
A and B
A and B
A and B
0–6 min, A from 35% till 80%
0–6 min, A from 5% till 35%
0–5.5 min, A from 50% till 98%
0–5.5 min, A from 50% till 98%
4.9
4.6
5.0
4.2
266
356
283
402
Artesunate
+
(
4
[M+NH ] )
Chloroquine
Dapsone
Diethylcarbamazine
A and B
A and B
A and B
0–6 min, A from 5% till 35%
0–6 min, A from 10% till 95%
0–1 min A keep at 5%, 1–5 min
A from 5% till 28%
4.5
4.9
4.2
320
249
200
Pentamidine
Praziquantel
Primaquine
Pyrantel
Pyrimethamine
Quinine
Thiabendazole
Tinidazole
Specific CYP enzyme marker reactions
O-Deethylphenacetin (CYP1A2)
A and B
A and C
A and B
A and B
A and B
A and B
A and B
A and B
0–6 min, A from 5% till 45%
0–6 min, A from 30% till 75%
0–6 min, A from 5% till 65%
0–6 min, A from 5% till 40%
0–5.5 min, A from 10% till 70%
0–6 min, A from 5% till 57%
0–6 min, A from 5% till 57%
0–6 min, A from 5% till 70%
5.0
5.3
5.5
4.7
4.7
5.1
5.2
5.4
341
313
260
207
249
325
202
248
A and C
0–1 min A keep at 5%, 1–4 min
A from 5% till 45%
Constant: 0–5 min, A 20% and C 80%
0–4 min, A from 20% till 47%
0–1 min A keep at 5%, 1–8 min
A from 5% till 95%
2.5
152
7
-Hydroxylcoumarin (CYP2A6)
Hydroxylbupropion (CYP2B6)
a-Hydroxylpaclitaxel (CYP2C8)
A and C
A and B
A and C
3.3
3.3
7.4
161 ([M-H])
)
256
870
6
N-Desethylamodiaquine (CYP2C8)
A and B
A and C
0–6 min, A from 5% till 35%
0–2.5 min, A 20% till 95%, 2.5–5 min,
A 95%
4.4
4.0
328
312
4¢-Hydroxyldiclofenac (CYP2C9)
4
1
4
¢-Hydroxyl-S-mephenytoin (CYP2C19)
A and C
A and C
A and B
0–6 min, A from 5 till 70%
0–4 min, A from 5% till 30%
0–3.5 min, A from 10% till 45%
0–5 min, A from 10% till 60%
0–5 min, A from 10% till 60%
0–5 min, A from 10% till 60%
0–4 min, A from 20 till 60%
0–4 min, A from 30% till 62%
4.7
2.6
2.6
4.3
3.2
3.4
2.8
3.5
235
278
192
258
284
254
342
305
¢-Hydroxylbufuralol (CYP2D6)
-Hydroxyldebrisoquine (CYP2D6)
O-Demethyldextromethorphan (CYP2D6) A and B
a-Hydroxylmetoprolol (CYP2D6)
O-Demethylmetoprolol (CYP2D6)
A and B
A and B
A and C
A and B
1
¢-Hydroxylmidazolam (CYP3A4)
6b-Hydroxyltestosterone (CYP3A4)
a
Chromatography was performed on a symmetry C18 column acetonitrile in 12 mM formic acid solution (v/v) (C), run gradiently
(
3.9 · 150 mm i.d.; 5 lm). The mobile phase consisted of aceto- at a flow rate of 1 ml/min. The effluent was split with approxi-
nitrile (A), 5 mM ammonium acetate buffer (pH 3) (B), or 10 % of mately 0.3 ml/min introduced into the mass spectrometer
where CLint is the intrinsic clearance reflecting the actual metabolic The metabolites formed were determined by LC/MS and quantified
capacity of the enzyme system with free access to substrate, fuB is by external standardisation using authentic references. The detailed
the free fraction in whole blood and Q
(
present study, the CL
H
the value of hepatic blood flow used here is 1400 ml/min). In the
is the total liver blood flow chromatographic conditions are shown in Table 1.
H
was calculated using the total incubation
concentrations for all the drugs since the binding to HLM of the Contribution of CYPs to HLM APD metabolism
compounds was not evaluated. The CL of a drug cannot exceed
the hepatic blood flow, 1400 ml/min. In general, drugs that have
CL above 980 ml/min/70 kg (70 % Q ) are classified high-clear-
ance drugs and those below 420 ml/min/70 kg (30 % Q ) are
H
The percent contributions of CYPs to the metabolism of 15 APDs
were estimated by applying the relative activity factor (RAF) values
as proposed by Crespi [14] using the values of the activities. In
brief, the RAF factor method aims to estimate the relative quantity
of a specific CYP in HLM based on its metabolic activity on an
enzyme-specific reaction in both the pure form of the enzyme and
within the HLM. The activity can be measured as velocity (v) or
H
H
H
classified low-clearance drugs. In order to compare values across
different studies, the conventional unit—ml/min/kg—is used. Using
this unit, the limit of CL
clearance drugs having values >14 ml/min/kg, and low-clearance
drugs <6 ml/min/kg. In vitro–in vivo correlations of drug clear-
ance was done for compounds for which drug clearance was
measured after i.v. administration.
H
becomes 20 ml/min/kg with high-
m
CLint [either as Vmax/K or ln2/(T1/2 x protein concentration)]. On
dividing the activity of HLM for the marker reaction (pmol
product/min/mg HLM) by the activity of the recombinant CYP
enzyme on the marker reaction (pmol product/min/pmol rCYP),
the RAF is derived as pmol rCYP/mg HLM. The RAFs of eight
major CYPs (CYP1A2, 2A6, 2B6, 2C8, 2C9, 2C19, 2D6 and 3A4)
were determined as the ratio of the activity of each enzyme marker
Enzyme kinetics
Incubation conditions used for 14 marker reactions in different reaction. The CYP enzyme-specific marker substrates for the 8
recombinant human CYP enzymes and pooled HLMs were opti- CYPs were: phenacetin (CYP1A2), coumarin (CYP2A6), bupro-
mised for linearity with respect to time and protein concentrations. pion (CYP2B6), paclitaxel, amodiaquine (CYP2C8), diclofenac

4
33
Table 2 Comparison of relative activity factors by different marker relative activity factor calculated from the Vmax of specific marker
substrates and calculating methods. RAFCL relative activity factor reaction in HLM and rCYP enzyme, RAF
relative activity factor
calculated from the clearance (Vmax/K ) of specific marker reaction calculated from the velocity of specific marker reaction at substrate
in human liver microsomes (HLM) and rCYP enzyme, RAFVmax concentration at K
in HLM and rCYP enzyme
v
m
m
CYPs
Marker
Microsomes
V
max (pmol/
K
m
RAFCL
RAFVmax RAF
v
% of Relative
min/mg HLM (lM) (Vmax/K
or pmol rCYP)
m
)
(Vmax (velocity content
)
a
at K
m
)
in HLM
(
pmol CYP/mg HLM)
CYP1A2 Phenacetin O-demethylation
CYP2A6 Coumarin 7-hydroxylation
CYP2B6 Bupropion hydroxylation
CYP2C8 Paclitaxel 6a-hydroxylation
Amodiaquine N-desethylation
HLM
rCYP1A2
HLM
rCYP2A6
HLM
rCYP2B6
HLM
rCYP2C8
HLM
718.1
6.5
431.0
3.5
137.0
14.9
107.9
1.1
1695.9
4.7
30.0
37.8 140.5
198.9
127.6
5.8
241.3
184.0
7.2
14.7
24.7
1.0
2.4
4.6
235.5
76.0
160.6 9.7
10.9
4.9
3.4
0.8
4.6
6.1
23.1
42.6
38.6
345.3
130.8
27.3
3.4
50.1
177.1
166.2
48.0
3.1
4.5
rCYP2C8
HLM
rCYP2C9
81.8
(8.2)
18.5
6.0
CYP2C9 Diclofenac 4¢-hydroxylation
1256.4
9.4
69.4
175.9
CYP2C19 S-Mephenytoin 4¢-hydroxylation HLM
rCYP2C19 2.6
50.2 57.3
7.4
CYP2D6 Bufuralol 1¢-hydroxylation
HLM
rCYP2D6
HLM
rCYP2D6
HLM
62.9
17.0
47.7
3.1
6.0
74.1
3.0
(0.3)
4.5
Debrisoquine 4-hydroxylation
205.3 43.2
3.6
18.9
32.4
Dextromethorphan
O-demethylation
159.1
rCYP2D6
HLM
rCYP2D6
HLM
rCYP2D6
HLM
rCYP3A4
HLM
rCYP3A4
15.6
69.5
4.6
822.0
12.2
1021.7
3.3
12.1 34.5
106.0
99.2 14.0
199.0
91.6 31.1
2.7
11.1
17.8
(3.7)
(1.5)
(3.3)
26.0
Metoprolol a-hydroxylation
Metoprolol O-demethylation
CYP3A4 Midazolam 1¢-hydroxylation
Testosterone 6b-hydroxylation
Total
14.7
13.2
61.8
37.2
2.2
42.7
248.1
276.8
344.4
277.8
551.5
5327.2
15.9
111.1 871.6
953
(56.7)
HLM
CYP concentration
480pmol/mg
a
The percentage relative content of each of the eight major CYPs in reaction, the one with a substrate commonly used and normally
HLM were calculated from the total amount obtained using the given to humans was used. Shown in bold. The relative CYP content
RAFCL data in bold. When there was more than one marker calculated by other marker substrates are shown in parentheses
(
CYP2C9), S-mephenytoin (CYP2C19), bufuralol, debrisoquine, and 2 lM ketoconazole were considered suitable for a one-
dextromethorphan, metoprolol (CYP2D6), midazolam and testo- concentration screening study. The inhibitors were dissolved in
sterone (CYP3A) (Table 2). methanol. Two microliters of the stock solution were added to
Using RAFCL, the relative contribution on the clearance of 200-ll incubation mixtures. The inhibitory effect of each inhibitor
each substrate by CYPs to that in HLM was calculated using was assessed by comparing the catalytic activity of the metabolism
equations described previously [15]. Briefly, the relative contribu- of three substrates (albendazole, amodiaquine or praziquantel)
tion is calculated by multiplying the metabolic rate (velocity, pmol in HLM with and without inhibitors.
product/min/pmol rCYP or CLint, ll/min/pmol rCYP) of the test
compound by rCYP by the RAF (pmol rCYP/mg HLM), which
gives the predicted activity in the HLM due to that specific CYP
Data analysis
(pmol product/min/mg HLM or ll/min/mg HLM). To calculate
percentage contribution, this value is then divided by the activity of
HLM for the test compound and multiplied by 100. Since the
metabolites are not known for most of the drugs or authentic
standards not available, the drug clearances were determined by the
substrate depletion approach described above.
All data points represent the mean of duplicate determinations.
Estimation of in vivo clearance was made from in vitro T1/2 data
according to the venous equilibrium model (well-stirred model)
using equations described previously [10, 16]. The liver was taken
as the main site of drug metabolic clearance. Predictions from in
vitro data were compared to the reported in vivo clearances. K
m
and Vmax values of each marker substrate in pooled HLM and eight
rCYPs were determined using non-linear least-squares regression
analysis with GraFit software (version 3.0, Erithacus Software
Inhibition studies
On review of reported studies on inhibitor concentrations that Limited, Middlesex, UK) and SigmaPlot Enzyme Kinetics Module
would show inhibitor-CYP selectivity, an average of 10 lM for Windows 7.0 (SPSS, Inc., Chicago, IL). Correlations between
a-naphthoflavone, quercetin, sulfaphenazole, ticlopidine, quinidine the activities of respective CYP enzyme in individual HLMs and

4
34
the clearance of substrates, artemisinin, amodiaquine and pra- only (HLM were used in all other investigations). Ta-
ziquantel, or the formation of their major metabolites were deter-
mined by least-squares linear regression analysis using SigmaPlot
2001 (version 7.0, SPSS Science Software UK Ltd., UK). P<0.05
was considered statistically significant.
ble 3 shows the in vitro T1/2 values, from which CLint for
the 15 APDs are calculated (Table 4) for RLM and
HLM. The predicted in vivo CLH values in humans
from the in vitro CLint data are shown in Table 4. Linear
regression analysis of predicted hepatic clearance, CLH,
between RLM and HLM yielded a low correlation
Results
2
coefficient (r ) of 0.50, which was statistically significant
(P<0.01; Fig. 2). The clearances in RLM, however,
Drug clearance by RLM and HLM
showed much higher values than those in HLM (from
In this study, the in vitro clearance of 15 APDs has been 1.6-fold for pyrimethamine to 19-fold for diethyl-
done in both HLM and RLM for species comparison carbamazine).
Table 3 Identification of cytochromes P450 (CYPs) involved in the metabolism of 15 anti-parasitic drugs. Note that all rat liver micro-
somes (RLM) are used with 0.5 mg/ml and recombinant CYP enzymes (rCYPs) are used at 20 pmol per incubation. HLM human liver
microsome
Compound
T1/2,RLM HLM concentration T1/2,HLM 1A1 1A2 2A6 1B1 2B6 2C8 2C9 2C19 2D6 2E1 3A4 3A5 4A11
(min)
(mg/ml)
(min)
(% of substrate consumed in rCYPs)
Albendazole
Amodiaquine
Artemisinin
Artesunate
Chloroquine
Dapsone
31.6
22.6
28.0
12.2
106.8
91.5
0.09
0.1
0.4
1
1
1
39.2
15.8
87.4
20.9
454.0
100.8
1036
100 93
100 19 20 49
17 25
69 91 42
91
52 28 73
94
43 43 54 42
13
33
39
49
12 13
18
59
27
18
19 18 20
5
19
16
12
Diethylcarbamazine 160.7
1
17
Pentamidine
Praziquantel
Primaquine
Pyrantel
Pyrimethamine
Quinine
201.3
24.6
18.8
121.9
>1500
36.8
2
0.2
1
1
1
0.4
0.1
2
>2000 63
41.8
92.6
16 23
11 33
23
21 13
91
20
45 23
39
30
17
161.3
623.3
190.4
39.6
28
97
19 18
13
15
11
22
8
Thiabendazole
Tinidazole
69.0
161.0
100 100
100
298.0
a
Table 4 Calculated intrinsic clearance (CLint ) of 15 anti-parasitic (CYPs) to the metabolism of APDs in HLM using the relative
drugs (APDs) in rat liver microsomes (RLM), human liver micro- activity factor (RAF) method. N.D. not detectable due to immea-
somes (HLM), and the relative contributions of cytochromes P450 surably long elimination half-life
b
Compound
CLint,RLM
CLint,HLM
Predicted CL
ml/min/kg)
H
rCYP (% contribution in HLM)
(
(ll/min/mg protein)
1A2 2A6 2B6 2C8 2C9 2C19 2D6 3A4 Sum
Albendazole
Amodiaquine
Artemisinin
Artesunate
Chloroquine
Dapsone
43.8
61.3
49.4
113.6
3.2
204.4
440.1
19.8
33.2
1.5
6.9
0.7
<0.1
82.9
7.5
4.3
1.1
9.1
174.9
1.2
18.2
19.1
9.8
12.3
1.4
5.0
0.6
<0.1
16.0
5.3
3.4
1.0
6.1
17.9
1.1
53
0.3
67
3.5
48
2.2
0.6
0.2
6.5
5.4
25
65
68
42
121
120
105
N.D.
N.D.
83
83
90
N.D.
74
395
89
10
1.3
120
54
10
53
4.3
13
31
15.1
12
14
Diethylcarbamazine 8.6
Pentamidine
Praziquantel
Primaquine
Pyrantel
Pyrimethamine
Quinine
6.9
56.3
73.9
11.4
<0.9
37.7
20.1
8.6
39
60
0.2
23
90
30
3.8
70
77
Thiabendazole
Tinidazole
395
12
a
The CLint was determined by the elimination half-life of substrate with a 1.4-kg liver, 20 g liver/kg body weight. In the scaling process
[
CLint=(ln2 · incubation volume)/(T1/2 · protein or enzyme the currently used factor is 45 mg HLM per gram liver. The value
amount)]
Predicted drug in vivo hepatic clearances (CL
of hepatic blood flow used here is 1400 ml/min/70 kg
b
H
) were scaled from
in vitro CLint, HLM using the normal human body weight of 70 kg

4
35
(
Table 5). This seemingly high correlation was, however,
not robust in that omission of one or two compounds
resulted in a significant loss in correlation to coefficients
around 0.6. The data are, therefore, not of sufficient
quantity and range to perform a useful correlation
analysis. Whereas this study focuses on the role of he-
patic CYPs, for some of the drugs studied, like artesu-
nate, dapsone and albdendazole, other non-CYP and/or
extrahepatic enzymes are involved in their metabolism
and disposition (Table 5). This has important implica-
tions in efforts to estimate total in vivo clearance, CLtot,
from CYP metabolism-based hepatic clearance CLH.
Relative activity factors
H
Fig. 2 A correlation analysis of hepatic clearances (CL ) of 15
2
anti-parasitic drugs in rats and humans (n=15, r =0.50, P<0.01).
Data were calculated from drug intrinsic clearance in vitro and
employed the normal human and rat body weights of 70 kg and
Table 2 shows the RAFs determined in our laboratory
using recombinant enzymes expressed in yeast and a pool
of HLM made from 20 livers. The CYP concentration of
the pooled liver microsomes was 480 pmol/mg. The
RAFs were determined using a variety of kinetic mea-
0
.25 kg, respectively
Comparison of clearance from in vitro
and in vivo studies
sures of activity (v=velocity at K , Vmax=velocity at
m
saturating substrate concentrations, CL=Vmax/K ) and
m
Table 5 summarises the in vivo CL data obtained for with more than one marker substrate for some CYP
some of the drugs taken from PK literature reports of enzymes. Using the RAFs of some marker substrates the
intravenous administration [17, 18, 19, 20, 21, 22, 23, 24, total amount of CYP in the human liver according to
2
5, 26, 27, 28, 29, 30, 31, 32]. Tinidazole showed the RAFs was twice (953 pmol/mg) the actual estimated
lowest mean clearance and amodiaquine the highest. amount of CYP. This indicates that the RAFs do not
The interindividual range of values for each substrate give the actual amount of CYP but the relative amounts
varied from twofold for tinidazole to tenfold for amo- that would be associated with reported activities on
diaquine. Using the well-stirred model, the clearance of specific marker reactions. When analysed this way, the
each substrate was calculated from the in vitro T1/2 and RAFs obtained using CL=Vmax/Km as an activity
was scaled to in vivo units of ml/min/kg body weight, to marker (Table 2) give relative abundances of the differ-
enable direct comparison with literature data. Table 4 ent CYPs in proportions that are consistent with litera-
and Table 5 show the predictions of in vivo clearances ture data obtained from immunoquantification studies
from in vitro data. In this study, we obtained a signifi- (reviewed [5]). Use of the RAF calculated from CL was
2
cant in vitro–in vivo correlation, r =0.82 (P<0.01) for also shown to be better than those derived from velocity
the eight compounds for which there was in vivo data at any concentration or at Vmax by Nakajima et al. [15].
Table 5 Comparison of in vivo clearances of eight anti-parasitic drugs between predicted values from in vitro studies with those reported
after i.v. drug administration in humans and their metabolic profiles
Predict
CLin vivo
Clinical
in vivo CL
CYP enzymes mediating
biotransformation in humans
Main metabolite in humans Renal
(reported data)
Protein
clearance binding
(% dose) (%)
a
(
ml/min/kg)
(ml/min/kg)
Mean Range
Amodiaquine 19.1
115
41.5
12.3
27–288
13–70
6.4–18
CYP2C8
N-mono- and di-desethyl
0.15
>90
Artesunate
Chloroquine
12.3
1.4
(Esterases and hepatic enzymes) Dihydroartemisinin
CYP2C8, 2D6, 3A Mono- and di-desethyl
-chloro-4-aminoquinoline
20–60
50–64
7
N-mono-and di-oxide
Hydroxylamine (toxic)
mono-acetyl
No data available
Carboxy and others
3-Hydroxy
Dapsone
5.0
0.65
0.3–0.90 CYP3A4, 2E1, 2C,
and N-acetyltranferase
17.87 8.8–23.8 CYP2D6 and 1A1
20
50–80
Pentamidine
Primaquine
Quinine
<0.1
5.3
6.1
2–11
0.7
20
5.8
3.4
0.5
4–7.5
1.8–4.6
0.4–0.7
45–65
80–90
CYP3A4 and 2C19
Tinidazole
1.1
No data available
20
a
Data were obtained from healthy volunteers except for artesunate which was from patients with moderate malaria and pentamidine
which was from patients with late stage Trypanosoma gambiense sleeping sickness

4
36
Identification of CYPs responsible for the metabolism
of 15 APDs
The CYP enzymes involved in the metabolism of each
APD were studied at 1 lM (except for artemisinin 10 lM
and artesunate 5 lM) of the substrate concentration with
a microsomal concentration of 0.1–2.0 mg/ml and
20 pmol/200 ll incubation of rCYP. Table 3 shows the
catalytic activities of the 13 rCYPs towards the elimina-
tion of each APD. The table also shows the in vitro T1/2
data in HLMs and RLMs. The mainly hepatic CYP1A2,
2
D6 and 3A4 were involved in the metabolism of most of
these compounds. The mainly extrahepatic CYP1A1,
B1 also showed high catalytic activity with some sub-
1
strates. The formation of metabolites of amodiaquine,
chloroquine and praziquantel were also determined using
LC/MS by comparing with the authentic references. The
main metabolite of AQ in HLM was desethylAQ, which
was mediated mainly by CYP2C8 as we previously
showed [18]. Two hydroxylated metabolites were formed
from PZQ after incubation with HLMs. Formation of
4-hydroxylPZQ was catalysed by CYP1A2 and 2C19,
whereas that of a previously unidentified mono-hydrox-
ylated metabolite (X-OH-PZQ) was mediated by
CYP3A4 and 3A5 (Fig. 3a, b). According to its LC/MS/
MS chromatographic and mass spectrometric charac-
teristics, the hydroxy group of X-OH-PZQ was not
located at the cyclohexyl ring. The desethylated
chloroquine was the major metabolite of CLQ in HLM
which was catalysed by CYP2C8, 2D6 and 3A4 (Fig. 3c).
Fig. 3 Plots of enzyme metabolic activity versus substrate concen-
tration of praziquantel (PZQ) (a and b) and chloroquine (CLQ) (c)
Contribution of CYP enzymes to the metabolism
activity of substrates in HLM
in human liver microsomes (HLM). Inset Vmax and K
the formation of PZQ 4- and unidentified hydroxylated (X-OH)
Following the results from CYP identification studies of metabolites and CLQ desethylated metabolite in HLM, respec-
m
values for
tively. The Vmax (pmol/min/pmol) and K
reaction in recombinant cytochrome P450 (CYP) enzymes were
as follows: PZQ 4-hydroxylation: Vmax, CYP1A2=33, Km, CYP1A2
m
(lM) values of each
the 15 APDs (Table 3), further investigations were done
to determine their in vitro T1/2 values in the rCYP en-
=
zymes. The CLint for these compounds in HLM ranged 54; Vmax, CYP2C19=4, Km, CYP2C19=7; PZQ X-hydroxylation:
from less than 0.1 ll/min/mg for pentamidine to 440 ll/
min/mg for amodiaquine.
V
max, CYP3A4=7, Km, CYP3A4=65;CLQdesethylation:Vmax, CYP2C8
m,CYP2C8=458; max, CYP2D6=1.7, m, CYP2D6=62;
max, CYP3A4=2.3, Km, CYP3A4=582. The formation of X-OH-
=
6
.9,
K
V
K
V
PZQ was calculated semi-quantitatively using 4-OH PZQ as a
reference because of the lack of the reference compound
The RAF approach
With the exception of artemisinin and thiabendazole,
The RAFs of CYP1A2, 2A6, 2B6, 2C8, 2C9, 2C19, 2D6, the total relative contribution of the eight major CYP
and 3A4 were calculated as the ratio of the marker enzymes to the metabolism of each compound in HLM
clearance (CL=Vmax/K ) in HLM to that in rCYPs were between 65 % and 120 %, which means the
m
(
Table 2). The estimated relative contribution of each metabolism of these compounds in HLM was mainly
CYP enzyme in the clearance of these APDs by HLM is mediated by the CYPs. Table 4 also shows the relative
shown in Table 4. When calculating the relative contri- significance of each CYP to the metabolism of com-
bution of CYPs in the metabolism of a compound pounds by the relative contribution data, such as
metabolised by more than one CYP, RAFs obtained CYP1A2 to the metabolism of albendazole and thia-
from rCYPs from the same source were used, yeast or bendazole, CYP2C8 to amodiaquine, CYP2D6 to pyr-
lymphoblastoid cells. It is not recommended to mix antel, and CYP3A4 to quinine. For thiabendazole, an
RAFs derived from different sources when calculating unrealistic total contribution by CYP1A2 of 394 % was
relative contributions on the metabolism of a compound observed. At 1.0 lM substrate concentration, we found
because the RAF values also depend on the source of a large amount of hydroxylated metabolite of thiaben-
recombinant CYPs [6].
dazole formed with recombinant CYP1A2, while it was

4
37
Table 6 Relative contributions of cytochromes P450 (CYPs) to the metabolism of substrates in human liver microsomes (HLM) by
inhibition studies
a
b
Blank Control a-Naphthoflavone Quercetin Sulfaphenazole Ticlopidine Quinidine Ketoconazole Sum
(
1
10 lM)
A2
(10 lM) (10 lM)
(10 lM)
2C19
(10 lM) (2 lM)
2C8
2C9
2D6
3A4
c
Albendazole (lM)
Contribution
1.08
0.50
0.15
0.89
0.41
0.26
0.75
42.8 %
0.66
64.5 %
0.40
55.4 %
0.75
34.8 %
0.00
0.96
77.9 %
0.93
98.0 %
0.14
84.4 %
0.70
28.9 %
0.01
0.55
7.4 %
0.21
8.0 %
0.83
7.1 %
0.53
12.4 %
0.22
0.85
60.3 %
0.24
11.6 %
0.78
11.6 %
0.68
27.6 %
0.02
0.59
12.1 %
0.26
14.5 %
0.79
10.5 %
0.51
10.4 %
0.23
0.62
19.5 %
0.53
47.4 %
0.56
37 %
0.70
28.8 %
0.21
220 %
244 %
206 %
143 %
d
Amodiaquine (lM) 0.94
Contribution
Desehtyl-AQ (lM) 0.00
Contribution
e
Praziquantel
Contribution
0.98
0.00
4
-hydroxyl-PZQ
(lM)
Contribution
X-Hydroxyl-PZQ
100.0 %
0.31
94.9 %
0.30
13.5 %
0.26
90.7 %
0.37
11.2 %
0.25
16 %
0.00
327 %
90.6 %
0.00
0.29
(lM)
Contribution
)6.9 %
)3.1 %
10.8 %
)26.0 %
15.8 %
100.0 %
a
d
Blank sample: incubation without NADPH and inhibitors
Control sample: incubation without inhibitors
Incubation condition for amodiaquine (AQ): 1 lM of AQ,
0.1 mg/ml HLM, 45 min
b
c
e
Incubation condition for albendazole (ABZ): 1 lM of ABZ,
.1 mg/ml HLM, 45 min
Incubation condition for praziquantel (PZQ): 1 lM of PZQ,
0.2 mg/ml HLM, 30 min
0
not detectable in HLM. In this case, the erroneous en- overestimated contribution of CYPs to the metabolism
zyme contributions (exceeding 100 %) and the forma- of substrates using inhibition studies could therefore be
tion of enzyme-source specific metabolite (hydroxylated a reflection of the lack of selectivity of these inhibitors
thiabendazole in rCYP1A2) may have been caused by for the respective CYPs.
differences between HLM and recombinant CYP1A2 in
cofactor and supporting enzyme (cytochrome b5 and
CYP NADPH reductase) requirements in the metabo-
lism of index and test compounds. For diethylcarbam-
The activity correlation in a panel of HLM approach
Correlations between the clearances of amodiaquine,
azine, pentamidine and pyrimethamine, it was difficult
artemisinin, and praziquantel and 9 CYPs activities in a
to estimate the relative contributions of CYPs to their
panel of 20 HLM were made (Table 7). The marker
elimination in HLM because of the immeasurably long
T1/2 in rCYPs even at high enzyme concentrations of
substrates for the 9 CYPs were: phenacetin (CYP1A2),
coumarin (CYP2A6), bupropion (CYP2B6), paclitaxel
20 pmol per incubation.
(
(
(
CYP2C8), diclofenac (CYP2C9), S-mephenytoin
CYP2C19), bufuralol (CYP2D6), chlorzoxazone
CYP2E1) and midazolam (CYP3A). The correlation
The selective inhibition approach
results of CYP activities derived from the 20 HLMs are
summarized in Table 8. From our experience with CYP
2
a-Naphthoflavone (CYP1A2), quercetin (CYP2C8), activity correlation studies, r values of over 0.7 indicate
sulfaphenazole (CYP2C9), ticlopidine (CYP2C19 and metabolically significant association between enzyme
CYP2D6), quinidine (CYP2D6) and ketoconazole activity and compound metabolism. Values lower than
(
CYP2C8, 3A4, 3A5) have been shown to be potent this, even if they are statistically significant, usually are
2
diagnostic inhibitors. Three high-clearance drugs— of no biological importance. Inter-CYP correlations (r )
albendazole, amodiaquine and praziquantel—were in the HLM samples used should, therefore, be less than
chosen to evaluate the utility of using selective inhibitors 0.7 in order to differentiate metabolically relevant cor-
in CYP identification studies. The inhibitory effect by relations in the metabolism of a compound by two or
these six inhibitors on either the clearance of substrates more CYPs. Table 8 shows that the HLM samples used
or the formation of metabolites of AQ and PZQ are in this study did not exhibit any metabolically significant
shown in Table 6. When the extent of inhibition of each inter-CYP activity correlation. The highest correlation
2
enzyme was added up, the total contribution calculated was between CYP2B6 and CYP2C8: r =0.52.
for most of the drugs was over 200 %. The relative Artemisinin metabolism showed a significant corre-
contribution by the inhibition studies was also different lation (r =0.75, P<0.01) with CYP3A4 activity. A
2
2
from those obtained from the RAFs approach. Only the good correlation (r =0.96, P<0.0001) was observed
predicted CYP contribution for the formation of between amodiaquine metabolism and CYP2C8 acti-
hydroxylated metabolite, X-OH-PZQ, of praziquantel vities. In this case, CYP 2C9 activities also showed a
was close to 100 %, which was only mediated by good correlation (r =0.69, P<0.01, respectively) with
2
CYP3A and only inhibited by ketoconazole. The AQ metabolism. This apparent correlation is likely

4
38
Table 7 Correlation coefficients of substrate clearance and/or for- 10 lM artemisinin (0.4 mg/ml HLM, 30 min), 1 lM amodiaquine
mation of major metabolites with specific cytochrome P450 enzyme (0.1 mg/ml HLM, 20 min) and 10 lM praziquantel (0.2 mg/ml,
activities in a panel of 20 human liver microsomes (HLM). The 30 min)
incubations were performed with substrate concentrations of
2
Correlation coefficient (r )
Compound
1A2
2A6
2B6
2
C8
2C9
2C19
2D6
2E1
3A4
Artemisinin
Amodiaquine
Desethyl-AQ
Praziquantel
4-Hydroxyl-PZQ
X-Hydroxyl-PZQ
0.02
0.02
0.01
0.00
0.61***
0.05
0.08
0.13
0.10
0.01
0.01
0.03
0.40**
0.43
0.42
0.26*
0.00
0.36**
0.12
0.40**
0.69**
0.64**
0.48***
0.12
0.06
0.01
0.03
0.33**
0.45**
0.07
0.02
0.04
0.04
0.01
0.02
0.08
0.02
0.00
0.01
0.03
0.02
0.04
0.75***
0.01
0.00
0.81***
0.02
0.82***
0.96***
0.99***
0.09
0.03
0.09
0.37**
*P<0.05
**P<0.01
***P<0.001
2
Table 8 Correlation coefficients (r ) of cytochrome P450 (CYP) CYP2C9 (diclofenac 4¢-hydroxylation), CYP2C19 (S-mephenytoin
activities in a panel of 20 human liver microsomal samples. CYP 4¢-hydroxylation), CYP2D6 (bufuralol 1¢-hydroxylation), CYP2E1
activity marker reactions used were: CYP1A2 (phenacetin deme- (chlorzoxazone
thylation), CYP2A6 (coumarin 7-hydroxylation), CYP2B6 (bu- 1¢-hydroxylation)
propion hydroxylation), CYP2C8 (paclitaxel 6a-hydroxylation),
6-hydroxylation),
CYP3A4
(midazolam
2
Correlation coefficients (r )
CYPs
Activity range
pmol/min/mg)
(
1A2
1
2A6
2B6
2C8
2C9
2C19
2D6
2E1
3A4
1
2
2
2
2
2
2
2
3
A2
A6
B6
C8
C9
C19
D6
E1
51–1334
94–1054
1.3–474
14–114
289–3292
1.0–240
45–198
0.00
1
0.02
0.19
1
0.00
0.04
0.52***
1
0.00
0.01
0.24*
0.46***
1
0.05
0.03
0.01
0.01
0.06
1
0.01
0.00
0.07
0.00
0.01
0.06
1
0.08
0.00
0.03
0.00
0.03
0.00
0.12
1
0.00
0.02
0.15
0.03
0.29*
0.16
0.05
0.04
1
361–1570
27–2100
A4
*P<0.05
**P<0.01
***P<0.001
2
derived from the correlations (r =0.46, P<0.001) cation (Table 3 and Table 4) and inhibition studies
between the activities of CYP2C8 and 2C9, in the liver (Table 6), the PZQ concentration used was 1.0 lM. The
bank. The relative enzyme contribution studies, Table 3 relative contributions of CYP1A2, 2C19 and 3A4 to PZQ
and Table 4, do not support the role of CYP2C9 since metabolism activity in HLM measured at substrate
this rCYP did not metabolise amodiaquine.
concentrations of 1 lM and 5 lM showed that the con-
For the formation of praziquantel hydroxylated tribution of CYP3A4 increased from 36 % to 59 %,
metabolites, the activities of CYP1A2 and 2C19 showed while contributions of CYP 1A2 and 2C19 decreased
the highest correlation (r =0.61 and 0.45, P<0.05) with from 47 % to 35 % and 17 % to 5 %, respectively. These
2
PZQ 4-hydroxylase, and CYP3A4 showed correlation data seem to indicate that, at low concentrations of PZQ,
(
consistent with the CYP identification results (Table 3). are important and at high plasma concentration, the low
2
r =0.82, P<0.0001) with X-hydroxylase. This was the high affinity and low turnover CYPs 1A2 and 2C19
2
Interestingly, a strong correlation (r =0.81, P<0.0001) affinity and high turnover CYP3A4 takes over (Fig. 3).
was observed with PZQ elimination and CYP3A4
activity in HLMs, while almost no correlation (r =0.05)
2
was found with CYP1A2 activity. This was not in line Discussion
with the results of CYP identification and relative con-
tribution of CYPs to PZQ metabolism by RAFs and Our in vitro studies on drug clearance and the identifi-
inhibition methods, which showed that CYP 1A2 was cation of important hepatic CYPs responsible for the
more important to PZQ metabolism than CYP3A4 metabolism of APDs provide insights into the pharma-
(
Table 4 and Table 6). However, it has to be noted that, cokinetic properties of these drugs. Rigorous evaluation
in the correlation studies, 10 lM PZQ (which is equiva- of in vitro–in vivo correlations was not possible due to
lent to human plasma concentration after therapeutic the scantiness of i.v. pharmacokinetic data on most of the
doses) was used (Table 7); whereas, in the CYP identifi- APDs, and that, for the few compounds it was available,

4
39
the spread of clearance values was poor. Qualitative amodiaquine, praziquantel and thiabendazole are high-
predictions to high, intermediate and low clearance clearance drugs, artemisinin, artesunate and quinine,
drug categories were, however, possible. A combination medium-clearance drugs, and chloroquine, dapsone,
of various in vitro reaction phenotyping approaches led diethylcarbamazine, pentamidine, primaquine, pyrantel,
to a fairly accurate identification of major CYPs involved pyrimethamine and tinidazole, low-clearance drugs.
in the metabolism of test compounds.
These categorisations were generally in agreement with
in vivo observations (Table 4, Table 5). Considering the
various sources of error and uncertainties, for now it will
suffice to say that for those with in vivo clearance above
hepatic blood flow (e.g. amodiaquine), there are other
Interspecies differences
In the pharmaceutical industry, preclinical predictions of major non-hepatic metabolism means of drug clearance.
human PKs of a drug are based on in silico, in vitro and For some drugs, such as dapsone, the contribution of
animal studies (reviewed in [7, 12, 33]). In vitro systems the cytosolic enzyme N-acetyltranferase (NAT2) [36] is
(
liver microsomes, tissues slices or hepatocytes) derived not included in the estimation of in vivo clearance from
from different species can assist in selecting a suitable in vitro HLM-derived metabolism data. For some drugs,
animal pharmacokinetic and/or toxicological model. such as arteminisin whose metabolism by the screened
For the APDs, there was a low but significant correlation CYPs does not account for 100 % of the metabolism, it
between metabolic stability in rats and in humans but can be suspected that other yet unidenitified microsomal
the RLM were generally associated with higher CLint enzymes are involved (Table 4).
rates than humans. These data point to the limitations
of using animal-derived tissue for metabolism studies
Identification of major CYPs involved in the metabolism
with the intention of predicting metabolism in humans.
of a compound
CYPs are a major drug metabolising enzyme system.
In vitro–in vivo correlations for drug clearance
Studies to identify the major enzymes associated with a
drug’s metabolism therefore start by evaluating the role
The variation in human in vivo CL values and the ten-
of CYPs (CYP identification). In this study we applied
dency to under-predict are consistent with literature data
the three major approaches being used for reaction
for other drugs [34]. The variation in human CL values
phenotyping: (a) RAF, (b) activity correlation analysis
is, however, not the only major reason for failure to
and (c) diagnostic inhibitors. Each of these studies has
predict the in vivo situation. In a study of 1163 com-
its pros and cons which, for some drugs, require one to
pounds from 48 chemistry programmes [35] and 48
use them in combination. We reported, for the first time
compounds in the same chemical series (Andersson et al.,
to our knowledge, the CYPs involved in the metabolism
personal communication) in rats showed very poor in
of praziquantel, primaquine, tinidazole, artesunate and
vitro–in vivo correlations in CL. Many factors have
pyrantel (Table 4). For other drugs, artemisinin [37],
been proposed to explain these poor correlations. They
amodiaquine [18], quinine [28], chloroquine [32], thia-
range from protein binding, role of transporters,
bendazole [38], albendazole [39] and dapsone [40], we
uncertainty of actual drug concentration available for
confirm enzymes identified (Table 4, Table 5). For
metabolism in the cells, scaling factors and complexity
some, pentamidine [41], chloroquine [42], the reasons for
of models used, role of other enzymes, extrahepatic
some discrepancies in the role of some CYPs can either
metabolism and other means of drug clearance [7, 12,
be in vitro conditions, CYP identification method and/
3
3]. Using drug elimination rate at a substrate concen-
or enzyme system used. In this study we also report
estimations of the relative contribution of each enzyme.
This is important if one is going to consider the clinical
significance of enzyme regulation of the involved
CYP(s). Our studies also demonstrated that it is difficult
to apply the RAF factor on compounds that are me-
tabolised very slowly (pentamidine, diethylcarbamazine)
or metabolically very unstable (thiabendazole). This
might be due to failure to achieve optimal in vitro
conditions for evaluating their metabolism. Our data
also shows that in addition to the documented 4-OH
PZQ, X-OH-PZQ is also a major metabolite of PZQ.
tration of 1.0 lM to estimate drug clearance may not
always satisfy the assumptions of Michaelis–Menten
kinetics, namely when [S]<<Km CL =Vmax/K =rate
int
m
of metabolism/[S], especially for low-clearance drugs
whose plasma concentration may be higher than 1 lM.
The worrying aspect is that each compound seems to be
affected differently by one or a combination of some of
these factors, thus compromising the HTS dream. We,
however, believe that these HTS screens for metabolic
stability are still useful as a starting point from which to
explore in greater detail the contribution of the likely
causes of poor prediction of in vivo situations.
In screening campaigns, working classifications of
drugs into high-clearance (>70 % of hepatic blood Some limitations of enzyme identification approaches
flow) and low-clearance (<30 % of hepatic blood flow)
categories are used. What we can salvage from the pre- The use of liver microsomes and experimental
dictions from our in vitro data is that albendazole, conditions optimal for CYPs automatically excludes

4
40
non-microsomal and extrahepatic enzymes. This can be ment regiments for neurocysticercosis from 2 weeks to
important for some drugs, e.g. dapsone, which is also 1 day. This approach to optimise drug treatment has a
metabolised by the cytosolic NAT2 and amodiaquine, precedent in the co-administration of cyclosporin with
which is also metabolised by the extrahepatic CYP1A1 ketoconazole, which resulted in lower doses for the
and 1B1 [18]. The role of microsomal enzymes like expensive immuno-suppresant and also resulted in
UGTs that require different incubation conditions for more stable and predictable PKs [48]. That amodia-
activity will also be missed. The use of correlation quine (CYP2C8), pyrantel (CYP2D6), chloroquine
studies can suffer from intra CYP activity correlations in (CYP2D6 and CYP2C8), and artesunate (CYP2A6) to
the panel of HLM used (Table 8). For example, if only a major extent (>50 %) appear to be metabolised by
the correlation studies had been used, the intra-corre- polymorphic enzymes hints at the possibility of inter-
lation between CYP2C8 and CYP2C9 would have made individual as well as interethnic variations in the dis-
it difficult to decide which enzyme metabolises amodia- position of these drugs. Drugs which are mainly
quine. The use of diagnostic inhibitors suffers from the metabolised by one CYP, e.g. pyrantel, amodiaquine,
lack of selectivity for target CYPs. This is reflected by tinidazole and thiabendazole, could be prone to drug–
the more than 200 % calculated relative contribution drug interactions involving inhibition, induction and or
from assumed selective inhibitors (Table 6). It can also polymorphic regulation of the major route of their
lead to false CYP identification as was the case in the elimination as there are no other compensatory en-
study by Jewell et al. [43]; these authors concluded that zymes. Our results also imply that a slow rate of
CYP3A4 was responsible for the metabolism of amo- metabolism contributes to chloroquine’s long half-life,
diaquine from inhibition studies using ketoconazole. besides the traditional explanation based on its large
Ketoconazole also inhibits CYP2C8, the hepatic enzyme volume of distribution. The 98 % renal clearance of
we conclusively demonstrated to be responsible for diethylcarbamazine [49] is consistent with the metabolic
amodiaquine metabolism [18]. The RAF factor may stability we observed in this study. Our results have
suffer from its assumption that the factors that affect the also highlighted areas that need further work towards
metabolism of enzyme marker reaction are similar to improving the in vitro experimental design in order to
those of the test compound, which is not true for some make them more predictive of in vivo situations. Re-
enzymes and test compounds. This could result in sults of our study will assist in predicting drug com-
unrealistic enzyme percentage contributions of over binations that might be associated with drug–drug
100 % for some compounds (Table 4). The choice of interactions through inhibition of particular CYPs and
compound concentration used in both intrinsic clear- the potential role of CYP genetic polymorphisms in the
ances and in CYP identification is important. The rela- pharmacokinetic variability of some drugs.
tive contribution of CYPs in praziquantel metabolism
changed with increasing drug concentrations, the high Acknowledgement Dr. Xue-Qing Li is a recipient of the Wenner-
Gren Foundation postdoctoral fellowship (Stockholm, Sweden).
affinity CYP1A2 and CYP2C19 being important at
1.0 lM PZQ. At 5 lM and 10 lM, the contributions of
these enzymes reduced as the contribution of CYP3A4
became dominant. In principle, it means that the role of
enzymes continuously changes as a function of drug
concentration in vivo for a compound metabolised by
enzymes with different affinities and catalytic activities.
References
1
. Masimirembwa C, Thompson R, Andersson T (2001) In vitro
high throughput screening of compounds for favourable met-
abolic properties in drug discovery. Comb Chem High
Throughput Screen 4:245–263
2
. Davit B, Reynolds K, Yuan R, Ajayi F, Conner D, Fadiran E,
Gillespie B, Sahajwalla C, Huang S, Lesko L (1999) FDA
evaluations using in vitro metabolism to predict and interpret
in vivo metabolic drug-drug interactions: impact on labeling.
J Clin Pharmacol 39:899–910
Clinical significance of these studies
Despite the limitations discussed above, the results of
our study offer potentially useful explanations of the
observed PKs of APDs in humans. For example,
finding that at in vivo plasma concentrations, CYP3A4
is the major enzyme involved in metabolising PZQ
explains why the plasma levels reduced drastically and
sometimes with loss of therapeutic effect in subjects
taking carbamazepine or dexamethasone [44, 45]
known inducers of CYP3A4. Knowledge of the role of
CYP3A4 in PZQ metabolism also explains why Diek-
mann et al. [46] and Jung et al. [47] observed that
cimetidine, ketoconazole and miconazole inhibited the
disposition of PZQ. Co-administering PZQ with
inhibitors of its metabolism in these studies increased
the therapeutic efficacy of the drug and reduced treat-
3. White N (1999) Antimalarial drug resistance and combination
chemotherapy. Phil Trans R Soc London 354:739–749
4
. Krishma S, White N (1996) Pharmacokinetics of quinine,
chloroquine and amodiaquine: clinical implications. Clin
Pharmacokinet 4:263–299
5. Rodriques D (1999) Integrated cytochrome P450 reaction
phenotyping: attempting to bridge the gap between cDNA-
expressed cytochromes P450 and native human liver micro-
somes. Biochem Pharmacol 57:465–480
6
¨
. Stormer E, von Moltke L, Greenblatt D (2000) Scaling drug
biotransformation data from cDNA-expressed cytochrome
P-450 to human liver: a comparison of relative activity factors
and human liver abundance in studies of mirtazapine meta-
bolism. J Pharmacol Exp Ther 295:793–801
7
. Bertz R, Granneman G (1997) Use of in vitro and in vivo data
to estimate the likelihood of metabolic pharmacokinetic inter-
actions. Clin Pharmacokinet 32:210–256

4
41
8
9
. Bapiro T, Egnell A, Hasler J, Masimirembwa C (2001) Appli- 26. Constantino L, Paixao P, Moreira R, Portela M, Do Rosario
cation of higher throughput screening (HTS) inhibition assays
to evaluate the interaction of antiparasitic drugs with cyto-
chrome P450s. Drug Metab Dispos 29:30–35
. Bapiro T, Andersson T, Otter C, Hasler J, Masimirembwa C
(
V, Iley J (1999) Metabolism of primaquine by liver homogenate
fractions. Evidence for monoamine oxidase and cytochrome
P450 involvement in the oxidative deamination of primaquine
to carboxyprimaquine. Exp Toxicol Pathol 51:299–303
2002) Cytochrome P450 1A1/2 induction by antiparasitic 27. Karbwang J, Davis T, Looareesuwan S, Molunto P, Bunnag D,
drugs: dose-dependent increase in ethoxyresorufin O-de-
ethylase activity and mRNA caused by quinine, primaquine
and albendazole in HeG2 cells. Eur J Clin Pharmacol
White N (1993) A comparison of the pharmacokinetic and
pharmacodynamic properties of quinine and quinidine in
healthy Thai males. Br J Clin Pharmacol 35:265-271
5
8:537–542
28. Zhao X, Yokoyama H, Chiba K, Wanwimolruk S, Ishizaki T
(1996) Identification of human cytochrome P450 isoforms in-
volved in the 3-hydroxylation of quinine by human liver mi-
crosomes and nine recombinant human cytochromes P450.
J Pharmacol Exp Ther 279:1327–1334
1
1
0. Houston J (1994) Utility of in vitro drug metabolism data in
predicting in vivo metabolic clearance. Biochem Pharmacol
4
1. Obach R (1999) Prediction of human clearance of twenty-nine
7:1469–1479
drugs from hepatic microsomal intrinsic clearance data: an 29. Robson R, Bailey R, Sharman J (1984) Tinidazole pharmaco-
examination of in vitro half-life approach and nonspecific
binding to microsomes. Drug Metab Dispos 27:1350–1359
2. Ito K, Iwatsubo T, Kanamitsu S, Nakajima Y, Sugiyama Y
kinetics in severe renal failure. Clin Pharmacokinet 9:88–94
30. Giao P, Vries P (2001) Pharmacokinetic interactions of anti-
malarial agents. Clin Pharmacokinet 40:343–373
1
(1998) Quantitative prediction of in vivo drug clearance and 31. Lee I, Hufford C (1990) Metabolism of antimalarial sesqui-
drug interactions from in vitro data on metabolism, together
terpene lactones. Pharmacol Ther 48:345–355
with binding and transport. Ann Rev Pharmacol Toxicol 32. Projean D, Baune B, Farinotti R, Flinois J, Beaune P, Taburet
3
8:461–499
A, Ducharme J (2003) In vitro metabolism of chloroquine:
Identification of CYP2C8, CYP3A4, and CYP2D6 as the main
isoforms catalysing N-desethylchloroquine formation. Drug
Metab Dispos 31:748–754
1
1
1
3. Houston J, Carlile D (1997) Prediction of hepatic clearance
from microsomes, hepatocytes, and liver slices. Drug Metab
Rev 29:891–922
4. Crespi C (1995) Xenobiotic-metabolizing human cells as tools 33. Iwatsubo T, Hirota N, Ooie T, Suzuki H, Shimada N, Chiba K,
for pharmacological and toxicological research. Adv Drug Res
6:179–235
5. Nakajima M, Nakamara S, Tokudome S, Shimada N,
Ishizaki T, Green C, Tyson C, Sugiyama Y (1997) Prediction of
in vitro drug metabolism in the human liver from in vitro
metabolims data. Pharmacol Ther 73:147–171
2
Yamazaki H, Yokoi T (1999) Azelastine N-demethylation by 34. Carlile D, Hakooz N, Bayliss M, Houston J (1999) Microsomal
cytochrome P-450 (CYP)3A4, CYP2D6, and CYP1A2 in prediction of in vivo clearance of CYP2C9 substrates in
human liver microsomes: evaluation of approach to predict humans. Br J Clin Pharmacol 47:625–635
the contribution of multiple CYPs. Drug Metab Dispos 35. Clarke S, Jeffrey P (2001) Utility of metabolic stability
2
7:1381–1391
6. Obach R, Baxter J, Liston T, Silber B, Jones B, MacIntyre F,
screening: comparison of in vitro and in vivo clearance.
Xenobiotica 31:591–598
Rance D, Wastall P (1997) The prediction of human pharma- 36. May D, Porter J, Uetrecht J, Wilkinson G, Branch R (1990)
1
cokinetic parameters from preclinical and in vitro metabolism
data. J Pharmacol Exp Ther 283:46–58
7. White N, Looareesuwan S, Edwards G, Phillips R, Karbwang
The contribution of N-hydroxylation and acetylation to dap-
sone pharmacokinetics in normal subjects. Clin Pharmacol
Ther 48:619–627
1
1
J, Nicholl D, Bunch C, Warrell D (1987) Pharmacokinetics of 37. Svensson U, Ashton M (1999) Identification of the human
intravenous amodiaquine. Br J Clin Pharmacol 23:127–135
8. Li X, Bjorkman A, Andersson T, Ridderstro M,
Masimirembwa C (2002) Amodiaquine clearance and its 38. Coulet M, Dacasto M, Eeckhoutte C, Larrieu G, Sutra J,
cytochrome P450 enzymes involved in the in vitro metabolism
of artemisinin. Br J Clin Pharmacol 48:528–535
¨
¨
m
metabolism to N-desethylamodiaquine is mediated by
CYP2C8: a new high affinity and turnover enzyme specific
probe substrate. J Pharmacol Exp Ther 300:399–407
Alvinerie M, Mace K, Pfeifer A, Galtier P (1998) Identification
of human and rabbit cytochromes P450 1A2 as major isoforms
involved in thiabendazole 5-hydroxylation. Fundam Clin
Pharmacol 12:225–235
1
2
9. Davis T, Phuong H, Ilett K, Hung N, Batty K, Phuong V,
Powell S, Thien H, Binh T (2001) Pharmacokinetics and 39. Rawden H, Kokwaro G, Ward S, Edwards G (2000) Relative
pharmacodynamics of intravenous artesunate in severe falci-
parum malaria. Antimicrob Agents Chemother 45:181–186
0. Walker O, Salako L, Alvan G, Ericsson O, Sjoqvist F (1987)
contribution of cytochrome P450 and flavin-containing
monooxygenases to the metabolism of albendazole by human
liver microsomes. Br J Clin Pharmacol 49:313–322
The disposition of chloroquine in healthy Nigerians after 40. Winter H, Wang Y, Unadkat J (2000) CYP2C8/9 mediate
single intravenous and oral doses. Br J Clin Pharmacol
3:295–301
1. Ducharme J, Farinotti R (1996) Clinical pharmacokinetics and 41. Bronner U (1994) Pharmacokinetics of pentamidine: focus on
dapsone N-hydroxylation at clinical concentrations of dapsone.
Drug Metab Dispos 28:865–868
2
2
2
metabolism of chloroquine. Clin Pharmacokinet 31:257–274
2. Pieters F, Zuidema J (1987) The absolute oral bioavailability of
dapsone in dogs and humans. Int J Clin Pharmacol Ther
Toxicol 25:396–400
3. Mitra A, Thummel K, Kalhorn T, Kharasch E, Unadkat J,
Slattery J (1995) Metabolism of dapsone to its hydroxylamine
the treatment of trypanosoma gambiense sleeping sickness.
Thesis, Karolinska Institute, Huddinge University Hospital,
Stockholm
42. Ducharme J, Baune B, Taburet A, Farinotti R (1996) Chlo-
roquine metabolism in human liver microsomes. Exp Toxic
Pathol 48:345
2
2
by CYP2E1 in vitro and in vivo. Clin Pharmacol Ther 58:556– 43. Jewell H, Maggs J, Harrison A, O’Neill P, Ruscoe J, Park B
5
66
(1995) Role of hepatic metabolism in the bioactivation and
detoxication of amodiaquine. Xenobiotica 25:199–217
M, Rombo L (1995) Pharmacokinetics and adverse reactions 44. Na-Bangchang K, Vanijanonta S, Karbwang J (1995) Plasma
4. Bronner U, Gustafsson L, Doua F, Ericsson O, Miezan T, Rais
after a single dose of pentamidine in patients with Trypano-
soma gambiense sleeping sickness. Br J Clin Pharmacol 39:289–
concentrations of praziquantel during the therapy of neuro-
cysticercosis with praziquantel, in the presence of antiepileptics
and dexamethasone. Southeast Asian J Trop Med Public
Health 26:120–123
2
5. Mihaly G, Ward S, Edwards G, Nicholl D, Orme M, Breck-
95
2
enridge A (1985) Pharmacokinetics of primaquine in man. I. 45. Vazquez M, Jung H, Sotelo J (1987) Plasma levels of pra-
Studies of the absolute bioavailability and effects of dose size.
Br J Clin Pharmacol 19:745–750
ziquantel decrease when dexamethosone is given simulta-
neously. Neurology 37:1561–1562

4
42
46. Diekmann H, Schneidereit M, Overbosch D (1989) Inhibitory 48. Albengres E, Tillement J (1992) Cyclosporin and ketoconazole,
effects of cimetidine, ketoconazole and miconazole on the
metabolism of praziquantel. Acta Leidensia 57:217–228
7. Jung H, Medina R, Castro N, Corona T, Sotelo J (1997) 49. Ilondu N, Orisakwe O, Ofoefule S, Afonne O, Obi E, Chilaka
drug interaction or therapeutic association? Int
Pharmacol Therapy Toxicol 30:555–570
J Clin
4
Pharmacokinetic study of praziquantel administered alone and
in combination with cimetidine in a single-day therapeutic
regimen. Antimicrob Agents Chemother 41:1256–1259
K, Orish C (2000) Pharmacokinetics of diethylcarbamazine:
prediction by concentration in saliva. Biol Pharm Bull 23:443–
445