Prediction of metabolic clearance using fresh human hepatocytes: Comparison with cryopreserved hepatocytes and hepatic microsomes for five benzodiazepines

Article abstract of DOI:10.1080/00498250701834665

1. Predictions of in vivo intrinsic clearance from cryopreserved human hepatocytes may be systematically low. In the current study, the metabolite kinetics of a series of CYP3A4 substrates (benzodiazepines) in fresh human hepatocytes from five donors, via a major UK supplier, were investigated and compared with those previously reported (by the authors' laboratory) for cryopreserved human hepatocytes and hepatic microsomes. 2. A high incidence of autoactivation (up to tenfold) and heteroactivation (by testosterone, up to 14-fold) among the major pathways was observed. CYP capacity (Vmax) was marginally lower and 'affinity' constants (KM, S50) were marginally greater compared with cryopreserved hepatocytes. 3. Average intrinsic clearance (based on maximal clearance, CLmax) was sevenfold lower than in cryopreserved hepatocytes (reflecting sensitivity of intrinsic clearance estimation in vitro to mechanistic parameter values, particularly those involving atypical kinetics), but scaled intrinsic clearances for fresh (and cryopreserved) hepatocytes were within the range previously determined in hepatic microsomes. 4. There was no evidence from this series of studies that fresh hepatocytes provide quantitatively improved estimates of intrinsic clearance over cryopreserved hepatocytes.

Full text of DOI:10.1080/00498250701834665

Xenobiotica, April 2008; 38(4): 353–367
Prediction of metabolic clearance using fresh human
hepatocytes: Comparison with cryopreserved hepato-
cytes and hepatic microsomes for five benzodiazepines
D. HALLIFAX, A. GALETIN, & J. B. HOUSTON
Centre for Applied Pharmacokinetic Research, School of Pharmacy and Pharmaceutical Sciences,
University of Manchester, Manchester, M13 9PT, UK
(Received 29 November 2007; accepted 29 November 2007)
Abstract
1. Predictions of in vivo intrinsic clearance from cryopreserved human hepatocytes may be
systematically low. In the current study, the metabolite kinetics of a series of CYP3A4 substrates
(benzodiazepines) in fresh human hepatocytes from five donors, via a major UK supplier, were
investigated and compared with those previously reported (by the authors’ laboratory) for
cryopreserved human hepatocytes and hepatic microsomes.
2. A high incidence of autoactivation (up to tenfold) and heteroactivation (by testosterone, up to
14-fold) among the major pathways was observed. CYP capacity (Vmax) was marginally lower and
‘affinity’ constants (KM, S50) were marginally greater compared with cryopreserved hepatocytes.
3. Average intrinsic clearance (based on maximal clearance, CLmax) was sevenfold lower than in
cryopreserved hepatocytes (reflecting sensitivity of intrinsic clearance estimation in vitro to mechanistic
parameter values, particularly those involving atypical kinetics), but scaled intrinsic clearances for fresh
(and cryopreserved) hepatocytes were within the range previously determined in hepatic microsomes.
4. There was no evidence from this series of studies that fresh hepatocytes provide quantitatively
improved estimates of intrinsic clearance over cryopreserved hepatocytes.
Keywords: Clearance prediction, human hepatocytes, benzodiazepines
Introduction
The prediction of metabolic clearance of candidate drugs is of major importance to the drug-
discovery and development processes. Although the theoretical basis of clearance prediction
is well understood (Houston 1994; Venkatakrishnan et al. 2003; Ito and Houston 2004),
Correspondence: J. B. Houston, School of Pharmacy and Pharmaceutical Sciences, University of Manchester, Oxford Road,
Manchester, M13 9PT, UK. E-mail: Brian.Houston@manchester.ac.uk
ISSN 0049-8254 print/ISSN 1366-5928 online ß 2008 Informa UK Ltd.
DOI: 10.1080/00498250701834665

354
D. Hallifax et al.
there appear to be quantitative limitations in the use of the available simple experimental
models. The microsomal model may be too simple for drugs cleared significantly by phase II
metabolism or involving rate limiting cellular uptake but these potential deficiencies might
be avoided through use of hepatocytes. The value of isolated hepatocytes to provide a less
artificial representation of clearance processes in the liver has been recognized following
numerous investigations over more than 10 years (Shibata et al. 2002; Griffin and Houston
2004; Hewitt et al. 2007; Soars et al. 2007). The initially severe limitations of lack of
availability and stability of human hepatocytes appear to have been overcome and
cryopreserved hepatocytes have been available commercially for some years, offering
convenience suitable for routine use. Recent studies (Lau et al. 2002; Naritomi et al. 2003;
McGinnity et al. 2004; Hallifax et al. 2005; Brown et al. 2007; Hewitt et al. 2007) have
provided further detail on which judgement of hepatocytes as a standard system for
determination of metabolic stability can be made. The ability of hepatocytes to predict
quantitatively hepatic clearance in vivo, mechanistically (based on unbound drug in vitro and
in vivo), remains in question: under-prediction of clearance has been demonstrated for a
range of cytochrome P450 substrates from cryopreserved hepatocytes (Riley et al. 2005;
Brown et al. 2007), appearing to reflect similar concerns using hepatic microsomes (Ito and
Houston 2005; Rawden et al. 2005).
Numerous investigations on prediction of human clearance from hepatocytes, either from
cryopreserved hepatocytes (Shibata et al. 2002), fresh hepatocytes (Soars et al. 2007) or
comparison of both types (Lau et al. 2002; Soars et al. 2002; Naritomi et al. 2003;
McGinnity et al. 2004), have been based on depletion of parent drug. In our previous study
to investigate the prediction potential of cryopreserved hepatocytes (Hallifax et al. 2005), we
determined the metabolite formation kinetics for a series of benzodiazepines (alprazolam,
diazepam, flunitrazepam, midazolam and triazolam) — prototypic substrates for the major
human drug metabolizing enzyme, CYP3A4 — in order to compare the atypical kinetics
behaviour and resultant intrinsic clearance (CLint) with those previous observed in a similar
study with human liver microsomes (Rawden et al. 2005). Despite wide variability in
clearance prediction, both systems substantially under-predicted clearance in vivo (by about
ninefold). Considering that these drugs are predominantly cleared by CYP-mediated
metabolic pathways and assuming that in vivo there should is no systematic overestimation
of clearance (due to underestimation of bioavailability or in modelling of intrinsic clearance
in vivo), there was no clear indication why this extent of under-prediction occurred.
Potential underestimation of clearance due to complexities associated with atypical kinetics
was mitigated by use of maximal (autoactivated) clearance. However, potentially reduced
metabolic capacity in these in vitro systems remains a concern.
In the current study, these issues were investigated further by comparable analysis of
metabolite kinetics for the same panel of benzodiazepine CYP3A4 substrates with the
alternative in vitro system of fresh human hepatocytes, as obtained by the major independent
supplier in the UK (UK Human Tissue Bank).
Materials and methods
Chemicals
Midazolam, diazepam, 10-hydroxymidazolam, 4-hydroxymidazolam, 3-hydroxyflunitraze-
pam and N-desmethylflunitrazepam were donated by Roche Ltd (Welwyn Garden City, UK).
Triazolam, 10-hydroxytriazolam, alprazolam, 10-hydroxyalprazolam, N-desmethyldiazepam

Prediction of metabolic clearance
355
(nordiazepam), flunitrazepam, testosterone, ꢀ-glucuronidase (with sulfatase) and Krebs–
Henseleit buffer were purchased from Sigma-Aldrich Ltd (Poole, UK). 4-Hydroxytriazolam
and 4-hydroxyalprazolam were purchased from BIOMOL International (Exeter, UK).
3-Hydroxydiazepam was donated by Wyeth Laboratories Ltd (Maidenhead, UK).
All solvents were of HPLC grade and all reagents were of analytical grade.
Acquisition, storage and resuspension of fresh human hepatocytes
Freshly isolated human hepatocytes were purchased from the UKHTB of De Montfort
Expertise Ltd, De Montfort University (Leicester, UK). The cells had been previously
isolated and suspended in preservation medium on ice immediately before transit by road
(2–3 h) and, due to their expected limited storage stability under these conditions, were used
immediately upon receipt. The hepatocytes were washed with, and resuspended (by
centrifugation at 50g at 4ꢀC) in, ice cold Williams’ Medium E cell culture medium (WME)
or Krebs–Henseleit buffer and subsequently warmed in a shaking water bath (37ꢀC; 10 min).
The cell number and viability was determined by the trypan blue exclusion assay (Berry et al.
1991) before drug incubations; a batch of cells was deemed suitable for incubation if viability
was greater than 80%.
Incubations to determine metabolite formation kinetics
Incubations of freshly isolated human hepatocytes with either midazolam (0.1–50 mM),
triazolam (2–100 mM), diazepam, flunitrazepam and alprazolam (all 1–500 mM) were
performed under initial rate conditions — linear with respect to cell density (0.1 ꢁ 106 live
hepatocytes ml–1 for midazolam and triazolam; 0.5 ꢁ 106 live cells ml–1 for diazepam,
flunitrazepam and alprazolam) and time (5, 45 and 60 min for midazolam, triazolam and
diazepam, flunitrazepam, alprazolam respectively).
The freshly isolated human hepatocytes (from five donor livers, each on a separate
occasion) were diluted in WME (pH 7.4) to twice the required incubation density, placed
(125 ml cell suspension/well) into 24 well plates (Corning, Inc., NY, USA) and pre-
incubated for 10 min in a Heidolph TitramaxÕ 1000 plate shaker/incubator (37ꢀC). Primary
drug solutions were prepared in acetonitrile and diluted in WME to twice the required
incubation concentration and warmed (37ꢀC). Reactions were initiated by the addition
of pre-incubated drug in WME (125 ml/well) and the plates were shaken by low speed
orbital rotation under the incubation conditions described above. In order to limit the
inhibitory effects of solvent used (Hickman et al. 1998; Busby et al. 1999) — to no more
than 10% — the total proportion of acetonitrile in the incubation was 0.5%. Reactions were
terminated by immersion of the plates in liquid nitrogen. ꢀ-Glucuronidase (125 U) in
ammonium acetate (60 mM; pH 4.5) was added and co-incubated (60 min; shaking; 37ꢀC)
to cleave any glucuronide metabolites. The reaction was terminated by addition of
acetonitrile (250 ml, containing 0.5 mM clobazam as internal standard). Samples were
centrifuged (5000g; 5 min) and an aliquot analysed by LC-MS/MS. Standard curves were
generated for each metabolite using an equivalent concentration of frozen and thawed spare
hepatocytes.
Two parallel experiments were performed (using the substrates midazolam and triazolam
with cells from one liver) to test the effects of conditions that were different to previous
studies (DMF used for some substrates, Rawden et al. 2005; Krebs–Henseleit
buffer, Hallifax et al. 2005). Firstly, incubations were performed as above except for

356
D. Hallifax et al.
the use of acetonitrile/dimethylformamide (50/50, v/v) instead of acetonitrile.
Secondly, Krebs–Henseleit buffer (pH 7.4), fortified with 5 mM phosphate, was used
instead of WME.
Incubations to determine metabolite formation in the presence of testosterone
Incubations of freshly isolated human hepatocytes with each substrate (1 mM) in the
presence of testosterone at different concentrations (10, 25 and 100 mM) were performed as
described above (in WME), to test for heteroactivation. The final concentration of organic
solvent (acetonitrile) in these incubations was 1.0% v/v. As above, frozen initial incubates
were further incubated with ꢀ-glucuronidase and analysed.
Non-specific binding of substrates
The recovery of unbound drug obtained from microfiltration (VectaSpin Micro 0.02 mm
tubes, Whatman, Brentford, UK) after incubation (5 min) of each substrate (at 1 and
10 mM) with a resuspension of freeze–thawed human hepatocytes (at the same cell
concentrations used in the main incubations) from the same source, was in the range
0.9–1.0. As in a previous study (Hallifax et al. 2005), it was deemed unnecessary to adjust
the assumed drug incubation concentrations for calculation of the results.
Determination of metabolite concentration
The two major metabolites of each benzodiazepine substrate (10-hydroxyalprazolam,
4-hydroxyalprazolam, 10-hydroxymidazolam, 4-hydroxymidazolam, 10-hydroxytriazolam,
4-hydroxytriazolam, 3-hydroxydiazepam, N-desmethyldiazepam, 3-hydroxyflunitrazepam
and N-desmethylflunitrazepam) were quantified by LC-MS/MS, as detailed in our previous
study (Hallifax et al. 2005).
In vitro kinetics
Metabolite formation rates were analysed by non-linear regression analysis using GraFit
(v. 3.0, Erithicus Software, Horley, UK) using models for either Michaelis–Menten kinetics
or sigmoidal kinetics (Hill equation, equation 1). The goodness-of-fit criteria used to select
the model with the most appropriate fit comprised visual inspection, consideration of the
randomness of residuals and the standard error of the parameters:
Vmax ꢂ Sn
v ¼
ð1Þ
S5n0 þ Sn
where v is the velocity of the reaction; Vmax is the maximum velocity; S50 is the substrate
concentration at half maximum velocity; S is substrate concentration; and n is the Hill
coefficient denoting sigmoidicity.

Prediction of metabolic clearance
357
Intrinsic clearance
CLint was determined from the ratio Vmax/KM (for pathways following Michaelis–Menten
kinetics); its equivalent (for non Michaelis–Menten kinetics), CLmax, was determined using
equation (2) (Houston and Kenworthy 1999) or directly from the maximum v/S:
Vmax
S50
n ꢃ 1
CLmax
¼
ꢂ
ð2Þ
1=n
nðn ꢃ 1Þ
CLint and CLmax values were scaled to in vivo units using scaling factors of 120 ꢁ 106
hepatocytes g–1 liver (Houston and Carlile 1997) and 21 g liver kg–1 bodyweight.
Comparisons with microsomal, cryopreserved hepatocyte and in vivo clearances
This study followed the design of our previous studies using either human liver microsomes
(Rawden et al. 2005) or cryopreserved hepatocytes (Hallifax et al. 2005); the kinetic
parameter values and intrinsic clearances from freshly isolated hepatocytes were therefore
specifically compared with the equivalent scaled (g–1 liver) data from these other studies.
Comparison included paired t-test paired by drug metabolic reaction.
Scaled intrinsic clearances from freshly isolated hepatocytes were evaluated relative to a
database of in vivo human studies reported previously (Ito and Houston 2005).
Results
Cell viability
The average proportion of viable cells (between batches) as indicated by the trypan blue
exclusion assay, immediately prior to incubation, was 85.8%; all batches of hepatocytes were
considered suitable for experimental use.
Incubation conditions
The rate of metabolism of all substrates (diazepam, flunitrazepam, midazolam, triazolam
and alprazolam) was linear at cell densities below 1 ꢁ 106 live hepatocytes ml–1, and with
time up to 30 min incubation (data not shown). In order to allow valid comparison with our
previous investigations with microsomes (Rawden et al. 2005) and cryopreserved
hepatocytes (Hallifax et al. 2005), variation in incubation conditions were explored.
Clearance of midazolam by either metabolic pathway was increased by 40% (data not
shown) using either a different solvent (DMF with acetonitrile rather than acetonitrile) or
using Krebs–Henseleit buffer (instead of WME). For 10-hydroxylation of triazolam, CLint
was increased (40%, data not shown) using Krebs–Henseleit buffer (instead of WME), but
there were no effects of DMF; for the 4-hydroxylation pathway, there were no significant
differences resulting from either of the different conditions. For comparison between
this study and our previous studies with microsomes (Rawden et al. 2005, where DMF
was used for some substrates) and cryopreserved hepatocytes (Hallifax et al. 2005, where
Krebs–Henseleit buffer was used), these (non-systematic) uncertainties would not have had
a major impact overall.

358
D. Hallifax et al.
Table I. Kinetic parameter valuesa from fresh human hepatocytes for a series of benzodiazepines.
Vmax (pmol min–1
/
CLint
(ml min–1 kg–1
d
b
c
Drug
Pathway
ðv=SmaxÞ=ðv=Smin
Þ
106 cells)
KM or S50 (mM)
n
)
Midazolam
10-OH
4-OH
1.1
2.2
52 ꢄ 26
46 ꢄ 19
11 ꢄ 7.0
72 ꢄ 26
1.2
1.1
10
Triazolam
10-OH
4-OH
1.2
2.1
73 ꢄ 53
74 ꢄ 38
280 ꢄ 200
210 ꢄ 23
1.0
1.3
1.2
Diazepam
3-OH
2.6
1.3
110 ꢄ 35
42 ꢄ 26
210 ꢄ 100
130 ꢄ 42
1.6
1.0
N-desmethyl
1.6
Flunitrazepam
Alprazolam
3-OH
4.4
1.3
64 ꢄ 36
11 ꢄ 5.0
130 ꢄ 86
76 ꢄ 19
1.7
1.4
N-desmethyl
0.92
0.24
10-OH
4-OH
2.1
3.9
10 ꢄ 6.9
69 ꢄ 48
410 ꢄ 190
560 ꢄ 310
1.3
1.4
aMean values from five donors ꢄ standard deviation (SD).
bMeasure of autoactivation. Data are the mean ꢄ SD derived from five determinations.
cS50 values shown are when autoactivation occurs and n > 1 (Equation 1). Data are the mean ꢄ SD derived from five
determinations.
dScaled CLint; see the Materials and methods section.
Metabolite formation kinetics
The kinetics of the formation of the two metabolites for each of the five benzodiazepines,
in terms of substrate concentration dependence, were characterized and the average
values for the kinetic parameters for all of the pathways are summarized in Table I.
10-Hydroxymidazolam, 3-hydroxyflunitrazepam and 4-hydroxyalprazolam were major
metabolites (>75% of total metabolic clearance for their respective drugs), whereas for
triazolam and diazepam there was a more equal contribution to total clearance from the two
pathways. The relative pathway contributions are largely consistent with previous studies
with human hepatic microsomes (Rawden et al. 2005) and cryopreserved hepatocytes
(Hallifax et al. 2005).
To compare the activity of the five individual livers and show the concentration
dependence of the clearance, the data are presented as clearance (v/S) as a function of
substrate concentration (Figure 1). Clearance by all pathways (10- and 4-hydroxylation of
midazolam (Figure 1A), triazolam (Figure 1B) and alprazolam (Figure 1C) and
3-hydroxylation and N-demethylation of flunitrazepam (Figure 1D) and diazepam
(Figure 1E)), was generally concentration dependent. For the hydroxylation pathways in
particular, the concentration dependence was either positive or both positive (at low
concentrations) and negative (at higher concentrations), signifying autoactivation (deviation
from Michaelis Menten kinetics is also indicated by values of n greater than 1; Table I).
There were differences between livers in the extent of autoactivation and the concentrations
at which maximum activation was reached. For the N-demethylation pathways, there was
only very minor evidence of autoactivation; linear clearance at low substrate concentrations
and/or simple saturation, as associated with Michaelis–Menten kinetics, was observed for all
but one liver (Figures 1D and E). For all pathways, there was a large range in clearance
between livers.
Autoactivated pathways do not yield a constant intrinsic clearance value which can be
directly extrapolated to predict clearance in vivo but, instead, yield increasing clearances at

Prediction of metabolic clearance
359
1.0
8.0
6.0
4.0
2.0
0.0
1-OH
4-OH
A
0.8
0.6
0.4
0.2
0.0
0.1
1
10
100
0.1
1
10
100
[midazolam] (mM)
[midazolam] (mM)
0.4
0.3
0.2
0.1
0.0
0.6
0.4
0.2
0.0
4-OH
1-OH
B
1
10
100
1000
1
10
100
1000
[triazolam] (mM)
[triazolam] (mM)
0.06
0.04
0.02
0.00
0.15
0.10
0.05
0.00
1-OH
4-OH
C
1
10
100
1000
1
10
100
1000
[alprazolam] (mM)
[alprazolam] (mM)
0.6
0.4
0.2
0.0
0.20
0.15
0.10
0.05
0.00
DM
3-OH
D
1
10
100
1000
1
10
100
1000
[flunitrazepam] (mM)
[flunitrazepam] (mM)
0.8
0.6
0.4
0.2
0.0
0.8
0.6
0.4
0.2
0.0
DM
3-OH
E
1
10
100
1000
1
10
100
1000
[diazepam] (mM)
[diazepam] (mM)
Figure 1. Clearance (v/S) as a function of substrate concentration in fresh hepatocytes from five
different human livers for 10-hydroxylation (left) and 4-hydroxylation (right) of: A, midazolam;
B, triazolam; C, alprazolam; and 3-hydroxylation and demethylation of D, flunitrazepam; and
E, diazepam. Each symbol represents a different liver.

360
D. Hallifax et al.
low drug concentrations which peak at a particular concentration. To illustrate the extent of
this activation, the data are presented in terms of the mean fold increase of clearance (v/S)
from the minimum observed, at the lowest substrate concentrations, to the maximum
clearance ((v/Smax)/(v/Smin); Table I). For individual livers, the range of the autoactivation
effect for the hydroxylation pathways was between about two- and tenfold, depending on
substrate.
Effect of co-incubation with testosterone
Heteroactivation was investigated by comparing the rates of formation of each metabolite
for each substrate, at low concentration, with or without the addition of testosterone
(5, 25 and 100 mM). Increased activity in the presence of testosterone was observed for both
pathways for each substrate and, in most cases, this was maximal at the lowest testosterone
concentration (5 mM) (Figure 2). The magnitude of this effect was greater, on average, for
the 4-hydroxylation pathway (up to a mean fold increase of between seven about 14)
compared with the 10-hydroxylation pathway (up to a mean fold increase of between two and
six) (Figures 2A–C). Similarly, the 3-hydroxylation pathway was more activated (up to a
mean fold increase of between four and seven), on average, than the N-demethylation
pathway (up to a mean fold increase of about two) (Figures 2D and E).
Intrinsic clearance
Kinetic parameters (Vmax, KM or S50 and n) listed in Table I were compared with those in
the previous study using cryopreserved hepatocytes (Hallifax et al. 2005). Vmax values from
fresh hepatocytes were consistently lower (by 43%, on average) than those from
cryopreserved hepatocytes (Figure 3A); although there was considerable variation between
livers, this was significant by paired t-test (p < 0.01). There was a linear relationship between
KM (or S50) in the two systems but values from fresh hepatocytes were significantly greater
(p < 0.01 by a paired t-test) by 2.2-fold on average (Figure 3B). The range and rank order of
intrinsic clearance (determined by Vmax/KM or CLmax, as appropriate) from fresh
hepatocytes was similar to that of cryopreserved hepatocytes, but the absolute values were
considerably lower (by sevenfold on average; p < 0.01 by a paired t-test) than those from
cryopreserved hepatocytes (Figure 3C); following the same study design, clearances from
cryopreserved hepatocytes were previously observed to be of a similar magnitude, on
average, to the clearances in microsomes (Hallifax et al. 2005).
Discussion
The option to use a cellular system for routine assessment of hepatic drug clearance has
arisen in recent years. Advances in cryopreservation technology have enabled human
hepatocytes to be successfully frozen, stored, and thawed with retention of both Phase I
and II enzyme activities (Hewitt et al. 2007; Soars et al. 2007) and these are widely available
commercially. As an alternative to cryopreserved hepatocytes, fresh hepatocytes have
become regularly available; in the UK this has largely been facilitated by the UKHTB.
Although, as for microsomes, a major complication in the use of human hepatocytes may be
inter-individual variability in cytochrome P450 activity, there might be additional concern

Prediction of metabolic clearance
361
1-OH
4-OH
A
B
1500
3000
2500
2000
1500
1000
500
1000
500
0
0
0
20
40
60
80
100
0
50
100
[testosterone] (mM)
[testosterone] (mM)
C
2000
1500
1000
500
0
0
50
100
[testosterone] (mM)
DM
3-OH
E
D
1500
1000
1000
500
0
500
0
0
50
100
0
50
100
[testosterone] (mM)
[testosterone] (mM)
Figure 2. Heteroactivation of fresh human hepatocyte clearance (percentage control) by testosterone
for: A, triazolam; B, alprazolam; C, midazolam (ꢀ, 10-hydroxylation; ꢂ, 4-hydroxylation); D,
flunitrazepem; and E, diazepam (ꢀ, N-demethylation; ꢂ, 3-hydroxylation). Data are the
mean ꢄ standard deviation derived from five determinations.

362
D. Hallifax et al.
A
B
100
10
1
1000
100
10
1
0.1
1
10
100
1000
0.1
1
10
100
Cryo-cellular Vmax (nmol/min/gLiver)
Cryo-cellular S50/Km (µM)
C
1000
Cryo cell
Fresh cell
Cell=microsome
Cellx7=microsome
100
10
1
0.1
0.1
1
10
100
1000
Microsomal CL (ml/min/kg)
Figure 3. Comparison of Vmax (A), KM or S50 (B) from fresh and cryopreserved human hepatocytes
and CLint (C) from hepatocytes and microsomes, for two pathways of metabolism for five
benzodiazepines (cryopreserved hepatocyte data are from Hallifax et al. 2005; microsomal data are
from Rawden et al. 2005).
about the potentially detrimental effects of the isolation and storage procedures and about
the health status of the donor livers. In this respect, the availability of freshly isolated
hepatocytes provides the opportunity to use hepatocytes which have not undergone the
storage procedures inherent with cryopreservation. Whilst in recent years the performance of
cryopreserved human hepatocytes for quantitative prediction of clearance has been quite
widely investigated, there have been less reports regarding fresh human hepatocytes (e.g.
McGinnity et al. 2004). Good agreement in activity between fresh and cryopreserved
hepatocytes has been reported for rat liver (Naritomi et al. 2003; Griffin and Houston 2004).
Under-prediction of in vivo clearance using human liver microsomes is typical for many
drugs (Ito and Houston 2005), based on unbound concentrations (in vitro and in vivo).
Although prediction of clearance has not always been based on unbound drug, this
fundamental mechanistic principle is increasingly recognized (Grime and Riley 2006).
Microsomes might be expected to under-predict hepatic clearance due to the lack of full

Prediction of metabolic clearance
363
cellular metabolic function (at least for substrates dependent on active transport and/or non-
cytochrome P450 pathways). Under-prediction of hepatic (unbound) clearance of various
drugs by human cryopreserved hepatocytes has been reported recently (Riley et al. 2005;
Brown et al. 2007), but reasons for under-prediction by hepatocytes are not obvious. In view
of concerns about the potential detrimental effects on cell function from the various
procedures of isolation and preservation, a principal aim of the study reported here was to
evaluate the performance of fresh human hepatocytes from the major UK supplier, by direct
comparison with our previous study using cryopreserved hepatocytes (from a major
commercial supplier) and benzodiazepine substrates.
In our previous studies, both with microsomes (Rawden et al. 2005) and cryopreserved
hepatocytes (Hallifax et al. 2005), the benzodiazepines were chosen as relatively simple
pathway substrates of the major drug metabolizing enzyme, CYP3A4. With the aim of
assessing clearance prediction, these substrates presented an additional complexity
associated with atypical kinetics (Houston and Kenworthy 1999; Houston and Galetin
2005). The lack of a constant clearance (at sub-saturating substrate concentrations) due to
autoactivation is potentially confounding when extrapolating to in vivo; if clearance is
estimated at a single low concentration, as might be the case using the drug depletion
method, a relatively low value would lead to under-prediction. In our previous studies,
metabolite formation from the benzodiazepines was characterized by a high incidence of
autoactivation, in both microsomes and cryopreserved hepatocytes. The kinetic detail
provided allowed use of the maximal clearance for prediction of clearance in vivo. However,
despite this, in vivo clearance was under-predicted by about ninefold, on average, and this is
in approximate agreement with studies (Riley et al. 2005; Brown et al. 2007) which involved
a wide range (n ¼ 57 and 45, respectively) of drugs and in which the average under-
prediction of clearance from cryopreserved hepatocytes was about five- or sixfold,
respectively. Although it therefore appeared that other factors contributed to the under-
prediction of clearance, effective comparison with fresh hepatocytes would necessitate the
same level of kinetic detail.
Qualitatively, the kinetic characteristics were similar between the two types of hepatocyte:
fresh hepatocytes displayed autoactivation of metabolite formation for the benzodiazepine
substrates in all livers although the magnitude of effect (two- to tenfold increase in clearance)
was dependent on liver and pathway; the effective concentration was also dependent on liver
and pathway. However, Vmax in fresh hepatocytes was consistently less than in cryopreserved
hepatocytes, while KM (or S50) values were consistently greater in fresh hepatocytes
(Figures 3A and B). Both of these parameter differences were marginal (about twofold), but
the combined effect was a substantially (sevenfold) lower calculated intrinsic clearance, on
average, for fresh hepatocytes (Figure 3C). This large discrepancy highlights the sensitivity
of intrinsic clearance estimation to mechanistically modelled parameters. For these
substrates, this may partly reflect the additional complexities of this type of kinetics:
continuous concentration dependence of v/S, and relative obscurity of simple (Michaelis–
Menten based) parameters. The apparently lower capacity (Vmax) in the fresh hepatocytes
might reflect a relative loss of CYP activity in these cells but such an effect would appear to
have been limited to about twofold on average, based on these findings. In our previous
studies, KM values were lower in cryopreserved hepatocytes (Hallifax et al. 2005) than in
microsomes (Rawden et al. 2005) indicating possible active accumulation of unbound drug
in cryopreserved hepatocytes but which is not evident in fresh cells. Alternatively, this
discrepancy in ‘affinity’ constant values between the systems could reflect subtle differences
in the kinetics, due to differences in the environment of the active sites. However, it should
be emphasized that this degree of discrepancy is minor in the context of experimental

364
D. Hallifax et al.
assessment of kinetic constants. Although absolute under-prediction multiples might be
treated with caution, overall, the results from five substrates and five livers support the more
widely observed under-prediction phenomenon.
Although this handling of in vitro atypical kinetics minimizes its detrimental impact on
clearance prediction, the potential situation in vivo should be considered. In vivo, enzymes
may be naturally activated by endogenous (or exogenous) substrates or activators (Witherow
and Houston 1999). Heteroactivation by testosterone of the metabolism of benzodiazepines
has been demonstrated in microsomes (Rawden et al. 2001), cryopreserved hepatocytes
(Hallifax et al. 2005) and, in this study, in fresh hepatocytes. Although the circulating plasma
concentration of testosterone in vivo is <1 mM (Griffin and Ojeda 2000), other steroids may
be effective activators and their combined concentration in the liver might be significant.
Heteroactivation has also been demonstrated for other CYP3A4 substrates in human
cryopreserved hepatocytes (Maenpaa et al. 1998; Ngui et al. 2000, 2001) and hepatic
microsomes (Kenworthy et al. 2001; Schrag and Weinkers 2001; Nakamura et al. 2002;
Patki et al. 2003) and there have been studies demonstrating heteroactivation of CYP3A4
in vivo (Tang et al. 1999; Egnell et al. 2003; Chen et al. 2004). For the benzodiazepines, the
high degree of activation observed in fresh hepatocytes was qualitatively and quantitatively
similar to that found in microsomes and cryopreserved hepatocytes and as such is further
evidence that the concentration of heteroactivators of CYP3A4 in hepatocytes following
isolation is low. However, it remains possible that heteroactivators present in vivo are lost
during the processes of liver and cell isolation. The apparent similarity in the potential of
fresh hepatocytes for heteroactivation compared with cryopreserved hepatocytes suggests
that lack of activation in vitro is not a reason for the relatively low clearances observed in
fresh hepatocytes.
As for causes of under-prediction of clearance regardless of kinetics complexities, some
studies have indicated procedural loss CYP activity in hepatocytes (Binda et al. 2003);
further, human hepatocyte acquisition necessarily involves preparative delays due to
transport and surgical constraints possibly resulting in liver degradation such as ischaemia
(Suzuki et al. 2004), but there is a lack of systematic study of the possible factors involved.
1000
Microsomes
Cryopreserved hepatocytes
Fresh hepatocytes
100
10
1
0.1
0.01
Midazolam
Triazolam
Flunitrazepam Diazepam
Alprazolam
Figure 4. Comparison of individual liver total CLint from fresh human hepatocytes, cryopreserved
human hepatocytes, and human liver microsomes for five benzodiazepines (cryopreserved hepatocyte
data are from Hallifax et al. 2005; microsomal data are from Rawden et al. 2005).

Prediction of metabolic clearance
365
Although cryopreserved hepatocytes from various sources have clearly been shown to under-
predict hepatic clearance, the apparently lower total activity of the fresh hepatocytes in this
study suggests an additional factor (for fresh hepatocytes), possibly reflecting the health
status of these donor livers: fresh hepatocytes from the source used for this study are usually
obtained from liver resection in patients requiring transplant or with inoperable
hepatocarcinoma; the impact of this pathological state, and possible concomitant
pharmacological treatment, on the function of P450 enzymes is not clear, but liver quality
has been cited as a an important factor in quantitative studies (Fisher et al. 2001).
In considering the accuracy of human clearance prediction, the variability both in vivo and
in vitro is an important limitation. In this series of studies, variability was particularly high
among source livers from commercially available hepatic microsomes (Rawden et al. 2005).
For the hepatocytes, although the variability was apparently less (Figure 4), the sample was
smaller (n ¼ 5). Variability in vitro may not only reflect donor individuality (by age, health
status and genotype) but also procedural/experimental uncertainties, and so there is clearly
potential for wide disagreement with in vivo, further emphasizing the need for caution
with respect to absolute multiples of under-prediction. Nevertheless, this comparison
between in vitro systems for the benzodiazepines has further highlighted quantitative
discrepancy with in vivo.
In summary, the kinetics of metabolite formation for five benzodiazepines in freshly
isolated human hepatocytes were qualitatively and quantitatively similar to that previously
observed in cryopreserved human hepatocytes as well as human liver microsomes, in terms
of auto- and heteroactivation (by testosterone). Despite considerable inter-individual
variability, prediction of clearance in vivo from the fresh hepatocytes was significantly lower
than that from the other systems. However, the capacity of the fresh hepatocytes was only
marginally smaller, and the affinity constants only marginally greater, leading to this effect.
There was therefore no evidence that fresh hepatocytes, at least from this source, provide
a more accurate in vitro model of intrinsic clearance (at least for these substrates) than either
cryopreserved hepatocytes or microsomes. Overall, the findings provide further evidence of
negative bias in clearance prediction from human in vitro systems.
Acknowledgements
Financial support for this project was provided by the following consortium members of the
Centre for Applied Pharmacokinetic Research: Bristol Myers-Squibb, GlaxoSmithKline,
Hoffmann-La Roche, Novartis, Pfizer and Servier. The authors acknowledge the
collaboration of UKHTB that facilitated supply of fresh hepatocytes.
References
Berry MN, Edwards AM, Barritt GJ. 1991. Isolated hepatocytes: preparation, properties and applications.
In: Burden RH, Knippenberg PH, editors. Laboratory techniques in biochemistry and molecular biology.
New York, NY: Elsevier. p 46.
Binda D, Lasserre-Bigot D, Bonet A, Thomassin M, Come MP, Guinchard C, Bars R, Jacqueson A, Richert L.
2003. Time course of cytochromes P450 decline during rat hepatocyte isolation and culture: Effect of L-NAME.
Toxicology in Vitro 17:59–67.
Brown HS, Griffin M, Houston JB. 2007. Evaluation of cryopreserved human hepatocytes as an alternative in vitro
system to microsomes for the prediction of metabolic clearance. Drug Metabolism and Disposition 35:293–301.

366
D. Hallifax et al.
Busby WF, Ackermann JR, Crespi CL. 1999. Effect of methanol, ethanol, dimethyl sulfoxide and acetonitrile on
in vitro activities of CDNA-expressed human cytochromes P-450 Drug Metabolism. Disposition 27:246–249.
Chen Q, Tan E, Strauss JR, Zhang Z, Fenyk-Melody JE, Booth-Genthe C, Rushmore TH, Stearns RA, Evans DC,
Baillie TA, et al. 2004. Effect of quinidine on the 10-hydroxylation of R-warfarin: species differences and
clearance projection. Journal of Pharmacology and Experimental Therapeutics 311:307–314.
Egnell AC, Houston JB, Boyer S. 2003. In vivo CYP3A4 heteroactivation is a possible mechanism for the drug
interaction between felbamate and carbamazepine Journal of Pharmacology and Experimental Therapeutics
305:1251–1262.
Fisher RL, Gandolfi AJ, Brendel K. 2001. Human liver quality is a dominant factor in the outcome of in vitro
studies. Cell Biology and Toxicology 17:179–189.
Griffin JE, Ojeda SR. 2000. Textbook of endocrine physiology. 4th ed. New York, NY: Oxford University Press.
p 257.
Griffin SJ, Houston JB. 2004. Comparison of fresh and cryopreserved rat hepatocyte suspensions for the prediction
of in vitro intrinsic clearance. Drug Metabolism and Disposition 32:552–558.
Grime K, Riley RJ. 2006. The impact of in vitro binding on in vitro–in vivo extrapolations, projections of metabolic
clearance and clinical drug–drug interactions. Current Drug Metabolism 7:251–264.
Hallifax D, Rawden HC, Hazook N, Houston JB. 2005. Prediction of metabolic clearance using cryopreserved
human hepatocytes: Kinetic characteristics for five benzodiazepines. Drug Metabolism and Disposition
33:1852–1858.
Hewitt NJ, Gomez Lechon MJ, Houston JB, Hallifax D, Brown HS, Maurel P, Kenna JG, Gustavsson L,
Lohmann C, Skonberg C, et al. 2007. Primary hepatocytes: Current understanding of the regulation of
metabolic enzymes and transporter proteins, and pharmaceutical practice for the use of hepatocytes in
metabolism, enzyme induction, transporter, clearance, and hepatotoxicity studies. Drug Metabolism Reviews
39:159–234.
Hickman D, Wang J-P, Wang Y, Unadkat JD. 1998. Evaluation of the selectivity of in vitro probes and suitability of
organic solvents for the measurement of human cytochrome P450 monooxygenase activity. Drug Metabolism
and Disposition 26:207–215.
Houston JB. 1994. Utility of in vitro drug metabolism data in predicting in vivo metabolic clearance. Biochemical
Pharmacology 47:1469–1479.
Houston JB, Carlile DJ. 1997. Prediction of hepatic clearance from microsomes, hepatocytes and liver slices. Drug
Metabolism Reviews 29:891–922.
Houston JB, Galetin A. 2005. Modelling atypical CYP3A4 kinetics: Principles and pragmatism. Archives of
Biochemistry and Biophysics 433:351–360.
Houston JB, Kenworthy KE. 1999. In vitro–in vivo scaling of cytochrome P450 kinetic data not consistent with the
classical Michaelis–Menten model. Drug Metabolism and Disposition 28:246–254.
Ito K, Houston JB. 2004. Comparison of the use of liver models for prediction of drug clearance using
in vitro kinetic data from hepatic microsomes and isolated hepatocytes. Pharmaceutical Research 21:
785–792.
Ito K, Houston JB. 2005. Prediction of human drug clearance from in vitro and preclinical data using
physiologically based and empirical approaches. Pharmaceutical Research 22:103–112.
Kenworthy KE, Clarke SE, Andrews J, Houston JB. 2001. Multisite kinetic models for CYP3A4: Simultaneous
activation and inhibition of diazepam and testosterone metabolism. Drug Metabolism and Disposition
29:1644–1651.
Lau YY, Sapidou E, Cui X, White RE, Cheng K-C. 2002. Development of a novel in vitro model to predict hepatic
clearance using fresh, cryopreserved and sandwich-cultured hepatocytes. Drug Metabolism and Disposition
30:1446–1454.
Maenpaa J, Hall SD, Ring BJ, Strom SC. 1998. Human cytochrome P450 3A (CYP3A) mediated midazolam
metabolism: The effect of assay conditions and regioselective stimulation by ꢁ-naphthoflavone, terfenadine and
testosterone. Pharmacogenetics 8:137–155.
McGinnity DF, Soars MG, Urbanowicz RA, Riley RJ. 2004. Evaluation of fresh and cryopreserved hepatocytes as
in vitro drug metabolism tools for the prediction of metabolic clearance. Drug Metabolism and Disposition
32:1247–1253.
Nakamura H, Nakasa H, Ishii I, Ariyoshi N, Igarashi T, Ohmori S, Kitada M. 2002. Effects of endogenous steroids
on CYP3A4-mediated drug metabolism by human liver microsomes. Drug Metabolism and Disposition
30:534–540.
Naritomi Y, Terashita S, Kagayama A, Sugiyama Y. 2003. Utility of hepatocytes in predicting drug metabolism:
Comparison of hepatic intrinsic clearance in rats and humans in vivo and in vitro. Drug Metabolism and
Disposition 31:580–588.

Prediction of metabolic clearance
367
Ngui JS, Chen Q, Shou M, Wang RW, Stearns RA, Baillie T, Tang W. 2001. In vitro stimulation of warfarin
metabolism by quinidine: increases in the formation of 40-hydroxylation and 10-hydroxywarfarin. Drug
Metabolism and Disposition 29:877–886.
Ngui JS, Tang W, Stearns RW, Shou M, Miller RR, Zhang Y, Lin JH, Baillie T. 2000. Cytochrome P450
3A4-mediated interaction of diclofenac and quinidine. Drug Metabolism and Disposition 28:1043–1050.
Patki KC, Von Moltke LL, Greenblatt DJ. 2003. In vitro metabolism of midazolam, triazolam, nifedipine and
testosterone by human liver microsomes and recombinant cytochromes P450: role of CYP3A4 and CYP3A5.
Drug Metabolism and Disposition 31:938–944.
Rawden HC, Carlile DJ, Tindall A, Hallifax D, Galetin A, Ito K, Houston JB. 2005. Microsomal prediction of
in vivo clearance and associated inter-individual variability of six benzodiazepines in humans. Xenobiotica
35:603–625.
Rawden HC, Tindall A, Hallifax D, Houston JB. 2001. Heteroactivation of CYP3A4 substrates: Impact on intrinsic
clearance and inter-individual variability. Drug Metabolism Reviews 33:211.
Riley RJ, McGinnity DF, Austin RP. 2005. A unified model for predicting human hepatic, metabolic clearance
from in vitro intrinsic clearance data in hepatocytes and microsomes. Drug Metabolism and Disposition
33:1304–1311.
Schrag ML, Weinkers LC. 2001. Triazolam substrate inhibition: evidence of competition for heme-bound reactive
oxygen within the CYP3A4 active site. Drug Metabolism and Disposition 29:70–75.
Shibata Y, Takahashi H, Chiba M, Ishii Y. 2002. Prediction of hepatic clearance and availability by cryopreserved
human hepatocytes: An application of the serum incubation method. Drug Metabolism and Disposition
30:892–896.
Soars MG, Burchell B, Riley RJ. 2002. In vitro analysis of human drug glucuronidation and prediction of in vivo
metabolic clearance. Journal of Pharmacology and Experimental Therapeutics 301:382–390.
Soars MG, McGinnity DF, Grime K, Riley RJ. 2007. The pivotal role of hepatocytes in drug discovery. Drug
Metabolism and Disposition 35:859–865.
Suzuki S, Sato T, Yoshino H, Kobayashi E. 2004. Impact of warm ischemia time on microsomal P450 isoforms in a
porcine model of therapeutic liver resection. Life Science 76:39–46.
Tang W, Stearns RA, Kwei GY, Iliff SA, Miller RR, Egan MA, Yu NX, Dean DC, Kumar S, Shou M, et al. 1999.
Interaction of diclofenac and quinidine in monkeys: Stimulation of diclofenac metabolism. Journal of
Pharmacology and Experimental Therapeutics 291:1068–1074.
Venkatakrishnan K, Von Moltke LL, Obach RS, Greenblatt DJ. 2003. Drug metabolism and drug interactions:
Application and clinical value of in vitro models. Current Drug Metabolism 4:423–459.
Witherow LE, Houston JB. 1999. Sigmoidal kinetics of CYP3A substrates: An approach for scaling
dextromethorphan metabolism in hepatic microsomes and isolated hepatocytes to predict in vivo clearance
in rat. Journal of Pharmacology and Experimental Therapeutics 290:58–65.