Therefore, the oral clearance or the plasma-concentration-time curve for an
oral administration (AUCpo) can be directly estimated from CLintfor those
compounds that are exclusively eliminated in the liver.
Nevertheless, the reliability of prediction ofin vivohepatic clearance usingin
vitrometabolic data, due largely to its multifactorial nature, is uncertain. The
variables, which may potentially affect the accuracy of such an estimate,
include, but are not limited to, extrahepatic metabolism, biliary excretion,
active transporters, compound solubility, and permeability, gastrointestinal
milieu conditions, applicability of the predictive PK models, and interindivi-
dual variability.
13.5.3 Examples
Development ofin vivometabolic prediction usingin vitrodata, one of the
major goals for metabolic stability studies, is accorded high priority, yet
remains challenging. Current approaches are still being refined (Andersson
et al., 2004). However, successful cases using both human liver microsomal and
hepatocyte metabolism data have been reported (Ito and Houston, 2004;
Shibata et al., 2000). Recent studies suggest that while human liver microsomal
results often yield qualitative predictions (Li et al., 2003), quantitative
correlations are likely to be obtained if hepatocytes are used (Ito and
Houston, 2004; Iwatsubo et al., 1997; Soars et al., 2002). As indicators of
current understanding, several recent studies are presented to illustrate
anticipated and unexpected outcomes.
In vitro–in vivocorrelation studies have mainly been undertaken for CYP-
mediated metabolism. One example is the study reported by Li et al. (2003).
Using Eqs. 13.4,13.8 and 13.14 and physiological SFs, Li et al. predicted the
CLh values of 15 antiparasitic drugs from studies using HLM. With the
exception of a single drug with a high CLh, they were able to classify each drug
into a category of low, intermediate, or high hepatic clearances based on the
HLM CLintvalues (Li et al., 2003). A similar correlation study that, employed
the HLMs from individual donors (N= 12) and recombinant enzymes
(CYP3A4, CYP2C9 and their allelic variants), was reported as an effort for
the assessment of the genotypic impacts on CYP2C9-mediated celecoxib
metabolism in man (Tang et al., 2001). Celecoxib is predominantly metabolized
by CYP2C9 and CYP3A4 in humans; the potential changes in kinetic
properties (KmandVmax) due to CYP2C9 polymorphism were thereby negated
by the involvement of CYP3A4 (Eq. 13.12). Thus, the plasma AUC of
celecoxib was increased by up to five fold in those individuals carrying
the allelic variants, namely CYP2C9^ 2 (Cys^144 Ile^359 ) and CYP2C9^3
(Arg^144 Leu^359 ), consistent with predictions based on the in vitroresults. In
contrast, the average 2.2-fold increase for the homozygous and heterozygous
variant CYP2C9^ 3 (Arg^144 Leu^359 ) and no apparent changes for the hetero-
zygous variant CYP2C9^ 2(Cys^144 Ile^359 ) were determined in a subsequent
clinical study. The authors demonstrated that the CL
0
int (Vmax/Km) for438 DETERMINATION OF METABOLIC RATES AND ENZYME KINETICS