A factor that complicates the use of human
A factor that complicates the use of human liver angiopoietin is one of variability, which arises in part because many of the human liver samples available for research show various stages of a variety of disease states (steatosis (fatty liver), cirrhosis, carcinogenesis in adjacent tissue), although care is taken to avoid liver infected with hepatitis, HIV or other potentially infectious agents. However, even among human livers that would be considered normal, there is tremendous inter-individual variability in the expression of CYP450 enzymes. Some of this variation is genetically based, inasmuch as some of the CYP450 enzymes, such as CYP2D6 and CYP2C19, are highly polymorphic (Parkinson, 1996a, Parkinson, 1996b). For other CYP450 enzymes, such as CYP1A2 and CYP3A4, much of the inter-subject variability likely arises from individual differences in exposure to inducers, such as cigarette smoke, diet and medications, or suppressors, such as active systemic infections (Parkinson, 1996a, Parkinson, 1996b, Maurel, 1996b). Despite the variability, species differences in CYP450 enzyme regulation suggest that it is important to utilize human hepatocytes to evaluate the ability of drugs and NCEs to induce cytochrome P450. For example, omeprazole and rifampin, which are prototypical inducers of human CYP1A2 and CYP3A4, respectively, cause little or no induction of the corresponding CYP450 enzymes in rats or mice (Diaz et al., 1990, Pichard et al., 1990). In addition, the variability observed in primary cultures of human hepatocytes better reflects the variability in drug-induced enzyme regulation observed in the general population (Flockhart, D., personal communication).
Unlike primary cultures of rat hepatocytes, primary cultures of human hepatocytes do not provide a reproducible means of assessing the enzyme-inducing capabilities of drugs and NCEs (in an absolute sense). We have prepared primary cultures of human hepatocytes that responded well to the CYP1A2-inducing effects of omeprazole, but that responded poorly to the CYP3A4-inducing effects of rifampin. It should be noted that the standard deviations for each treatment group illustrate the large degree of variation (see Fig. 9). Due to the variability in the expressed levels of CYP450 activity, it is important when testing NCEs that studies be conducted with positive controls against which unknowns can then be directly compared.
The degree to which variability in the inducibility of CYP450 enzymes in vitro reflects variability in their inducibility in vivo is not known, but there is clinical evidence to suggest that the inducibility of CYP450 enzymes in vivo does show intersubject variability. For example, treatment of healthy volunteers with omeprazole increases the CYP1A2-dependent N-demethylation of caffeine to paraxanthine in most but not all cases (Rost and Roots, 1994, Rost et al., 1994). Likewise, intersubject variation has been observed in the degree to which rifampin therapy stimulates the urinary excretion of 6β-hydroxycortisol, a marker of CYP3A activity (Ged et al., 1989). There may be cause to believe that there are individuals who are ‘poor’ and ‘extensive’ responders to CYP450 inducers, possibly due to polymorphisms in the pregnane X receptor (PXR) (Lehmann et al., 1998) (see Section 5). Further research is required to ascertain whether genetic differences exist within the population that are responsible for the variability observed in vitro. In addition, although the relationship is by no means perfect, we have observed an apparent relationship between the absolute enzyme activity in microsomes from induced hepatocytes and the corresponding control activity (McCune et al., 2000). The underlying basis and the significance of this apparent relationship also remain to be determined.
Conclusions One of the biggest challenges that continues to face pharmaceutical scientists today is the accurate prediction of in vivo disposition from in vitro data. The prospect of accurately predicting a drug’s in vivo induction potential is not unlike that which currently confronts scientists faced with predicting enzyme inhibition from data derived in vitro. Just as we have made progress in the latter area, we are becoming more capable of accurately assessing the likelihood that a particular drug will cause serious interactions in vivo via enzyme induction (Ito et al., 1998). The pharmacodynamic effects of inducers cannot be predicted accurately without consideration of the respective pharmacokinetics. Table 1 contains just a few of the parameters that are involved in determining the total disposition of a particular drug and therefore will partially determine the rate and extent of drug at a particular nuclear receptor site in the case of enzyme induction (Nies et al., 1976, Ito et al., 1998, Derendorf and Meibohm, 1999, Sun et al., 1999). Continued efforts to define and model these parameters and their impact on drug interactions in vivo are needed for proper management of patient drug therapies.