CYP2C9-mediated celecoxib hydroxylation was reduced by the amino acid
substitutions of Ile^359 to Leu^359 and Arg^144 to Cys^144 in the recombinant
CYP2C9 proteins, while an in vivoprediction of the impact of CYP2C9
genotype was attempted using thein vitroresults (Tang et al., 2001).
Correlation studies have been also undertaken for metabolism primarily
mediated by Phase II metabolic enzymes, UGTs in particular (Andersson et al.,
2001; Soars et al., 2002). Soars and coworkers claimed to find significant
correlations between human hepatic clearance, predicted using compound
depletion in HLM (t1/2andke), and clearance determinedin vivofor 11 UGT
substrates, when protein binding and blood partitioning were factored into the
correlation scheme. However, thein vivoclearance was consistently under-
predicted by an order of magnitude. Similar prediction was obtained using the
results from either freshly isolated (N= 4) or cryopreserved human hepato-
cytes (N= 3). Interestingly, relative to the marked under-prediction when they
used the HLM data, the predictions based on the hepatocyte results were more
accurate (Soars et al., 2002).
In contrast, studies reported by Andersson et al. (2001, 2004) raise some
concerns as to the reliability and utility ofin vivoprediction usingin vitro
results for both Phase I and Phase II enzyme-mediated metabolism. Using
precision-cut liver slices, primary hepatocytes, and liver microsomal prepara-
tions, they demonstrated that glucuronidation is the major pathway of
biotransformation of almokalant in humans. The metabolic clearancein vivo,
however, was constantly underpredicted when the results from thesein vitro
systems were used (Andersson et al., 2001). In another study, using four well-
characterized CYP2C9 substrates, they showed generally poor predictability of
in vivo CLh using CL
0
int determined in HLMs based on the well-stirred
predictive model; these predictions were not improved by inclusion of protein
binding. Therefore, the authors questionedin vitrometabolic screening as a
compound selection tool, in the absence of a provenin vitro–in vivocorrelation
(Andersson et al., 2004).
The studies described in this section demonstrate both success and
challenges, and reflect the current understanding of prediction of hepatic
clearance usingin vitrohuman metabolic data. In contrast to the well-defined
concepts, readily applicable biochemical methods, and sensitive analytical
techniques available to studies of metabolic stability and enzyme kinetics,
efforts to accurately predict in vivo, using in vitro data, require further
refinement.
ABBREVIATIONS
AcCN acetonitrile
ADME absorption, distribution, metabolism, and elimination.
APPS adenosine 3
0
-phosphate 5
0
-phosphosulfate
ABBREVIATIONS 439