![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
|
|
|
|
|
||
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Pharmacokinetics, Dynamics, and Metabolism, Pfizer, Groton, Connecticut and Sandwich, United Kingdom
(Received December 1, 2005; accepted June 8, 2006)
 |
Abstract |
|---|
 Top Abstract Materials and MethodsResultsDiscussionReferences |
|---|
).
The consequences of polymorphic enzymes on drug metabolism in relation to efficacy and side effects has been the focus of numerous studies (Vandel et al., 1999; Dandara et al., 2001). For instance, individuals lacking the expression of a polymorphic drug-metabolizing enzyme (commonly referred to as "poor metabolizers" or PMs) will have higher drug exposure if those drugs are metabolized by those polymorphic enzymes, which could lead to exaggerated pharmacology or enhanced side effects relative to the intermediate metabolizer and extensive metabolizer (EM) subjects given the same dose (Mahgoub et al., 1977). Alternatively, if a polymorphic enzyme forms a particular metabolite that contributes to the activity of a drug, then different efficacy profiles might be observed in EM and PM subjects (Poulsen et al., 1996).
Of the identified polymorphic enzymes involved in drug metabolism, CYP2D6 is considered one of the most important, with a substrate specificity typical of many new chemical entities (broadly speaking, lipophilic bases). An estimated 20 to 25% of all drugs in clinical use are metabolized at least in part by CYP2D6 (Evans and Relling, 1999). The frequency of CYP2D6 PMs in the population depends on race and is reported to be approximately 1% of Asians and 5 to 10% of Caucasians (Shimizu et al., 2003). The primarily hepatic expression of this enzyme governs first pass metabolism after oral drug administration, whereas the low levels of intestinal expression do not appear to be important (Madani et al., 1999). Numerous studies have characterized the impact of CYP2D6 polymorphism on substrate area under the curve (AUC) in EM and PM subjects, and a recent article by Dorne et al. (2002) provides a useful database of human clearance and metabolite recovery data.
Because the safety profile of a new discovery candidate is often unknown, it is beneficial to limit the contribution of polymorphic enzymes below some cutoff to avoid the requirement of phenotyping/genotyping before the initiation of drug therapy and distinct dosing regimens in EM and PM subjects. A variety of in vitro tools are available to determine the relative contribution or percent contribution to metabolism by a polymorphic enzyme. These include coincubation with specific enzyme inhibitors (chemical inhibitors/inhibitory antibodies), use of poor metabolizer in vitro reagents (either human fractions or recombinant systems), and studies in individually expressed recombinant enzymes (Bjornsson et al., 2003; Williams et al., 2003). The quantitative link between results from these assays and clinical pharmacokinetic variability has not been described.
The objective of our studies was to investigate the relationship between estimated fraction metabolized by CYP2D6 in vitro and pharmacokinetic impact observed in clinical studies. Because of the wealth of data involving in vitro and in vivo studies and the frequency of encountering CYP2D6 substrates, this polymorphism served as the basis for examining the impact of polymorphic metabolism more closely. Although less data exists on the impact of other polymorphisms, we feel that the results with CYP2D6 can be generalized to other situations in which a polymorphism contributes to interindividual variability in clearance.
|
Materials and Methods |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
Microsomal Incubations. Human liver microsomal incubations were performed at 0.5 µM P450 (1.5 mg protein/ml) and 1 µM substrate, in the presence and absence of P450 inhibitors. Each incubation (final volume 1.2 ml) comprised 50 mM potassium phosphate buffer (pH 7.4) and 5 mM MgCl2. Reducing equivalents required for P450 metabolism were provided by NADPH (1 mM), which was regenerated in situ by an isocitric acid/isocitric acid dehydrogenase system. Over the 60-min incubation period, 100-µl samples were removed and added to 100 µl of ice-cold acetonitrile containing internal standard to terminate the reaction. Samples were centrifuged at 2000g for 40 min, and 80 µl was directly injected onto a generic high-performance liquid chromatography-mass spectrometry system (Sciex API 2000 Mass Spectrometer with TurboIonSpray interface, with an OptiLynx Reliasil C18 (40-µm, 15 x 2.1 mm) column. Peak responses were judged to be within the linear range for the instrument.
Specificity of Chemical Inhibitors. Incubations (n = 2) were performed as described above with the probe substrates phenacetin (CYP1A2), diclofenac (CYP2C9), S-mephenytoin (CYP2C19), dextromethorphan (CYP2D6), and midazolam (CYP3A4). For each substrate, incubations were performed in the presence and absence of P450 inhibitors, 10 µM furafylline (CYP1A2), 10 µM sulfaphenazole (CYP2C9), 3 µM benzlnirvanol (CYP2C19), 1 µM quinidine (CYP2D6), and 1 µM ketoconazole (CYP3A4), and first order disappearance rate constants were determined.
Incubations with CYP2D6 Substrates. In a preliminary investigation, incubations (n = 2) were performed as described above, for the following CYP2D6 substrates: amitriptyline, atomoxetine, desipramine, dextromethorphan, duloxetine, flecainide, fluoxetine, fluvoxamine, imipramine, metoprolol, mexiletine, nortriptyline, propafenone, propranolol, sertraline, sparteine, tolterodine, and venlafaxine. Substrates exhibiting sufficient metabolic vulnerability in human liver microsomes were further investigated. Incubations were performed in the presence (n = 4) and absence (n = 4) of 1 µM quinidine. Each substrate was assayed over four to six separate days to assess variability of the assay.
Data Analysis. The first order disappearance rate constant was determined for each incubation by plotting ln substrate conc. (peak area ratio, drug/internal standard) against incubation time and determining the gradient of the regression line, using data for which an accurate first order decay curve could be obtained. Data were only used when there were at least three time points collected per incubation, substrate had been depleted by >20% by the final time point, and regression of the log-linear substrate declination plot gave a correlation coefficient of >0.9.
The mean disappearance rate constant data generated on each day in the presence and absence of inhibitor was used to calculate the percentage P450 contribution using the equation below:
 |
Theoretical Considerations. For this analysis, two key parameters were included in the prediction of CL or AUC in EM and PM subjects. The first parameter characterized the degree of functional impairment that the polymorphism causes and was termed EF (enzyme function). For example, homozygous CYP2D6*6 individuals (PM) do not express functional hepatic CYP2D6 enzyme caused by a gene deletion (EF = 0). Contrary to the situation with CYP2D6, a polymorphism in CYP2C9 (CYP2C9*3) has been shown to reduce the capacity of the enzyme (Vmax/Km) to 
15 to 20% (EF = 0.15) of the wild-type enzyme (Haining et al., 1996; Sullivan-Klose et al., 1996; Miners and Birkett, 1998; Kirchheiner et al., 2002). The fraction of metabolism by a polymorphic enzyme is the second key parameter that must be considered (fm(n)), which has also been referred to as relative contribution. One way to determine this parameter involves microsomal incubations and selective P450 inhibitors with the potential limitations described below.
Equations were adapted from the literature based on the predictions of dose adjustments required in patients with renal impairment (Rowland and Matin, 1973; Shaw and Houston, 1987; Rowland and Tozer, 1989). More recently, the same principles have been applied to improve the prediction of metabolismbased drug-drug interactions (Venkatakrishnan et al., 2003; Ito et al., 2005).
For the purposes of this analysis, hepatic clearance in poor metabolizers (CLPM) was defined as follows:
 | (1) |
For many drugs, clearance by P450 represents one of several elimination pathways. The clearance by a specific P450 enzyme in an EM subject was calculated by the equation below:
 | (2) |
The relative importance of other elimination pathways is calculated by the equation below:
 | (3) |
 | (4) |
 | (5) |
 | (6) |
) that RAUC can be simplified to:
 | (7) |
|
Results |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
), and these have traditionally been used to assess percentage contributions to metabolism by P450 enzymes. As a preliminary investigation, the specificity of these inhibitors was reassessed in the specific batch of human liver microsomes used for this study. A P450 concentration of 0.5 µM was chosen, which is higher than the P450 content typically used in substrate depletion assays (0.25 µM). Using 0.5 µM P450 will increase metabolic vulnerability by approximately 1.5-fold (S.D. = 0.45, n = 45) (data not shown), thereby increasing the ability to investigate more slowly metabolized compounds. Incubations (n = 2) were performed at 0.5 µM P450 and 1 µM substrate on two separate days. The effect of inhibitors on both the target P450 and cross-reactivity against the other P450s was assessed. Of the probe substrates investigated, substrate loss could be monitored for all, with the exception of S-mephenytoin, where over the course of the 60-min incubation, substrate depletion was insufficient to determine a metabolic rate (i.e., <20% substrate depletion after 60 min). Thus, effects on CYP2C19 activity were assessed based on 4-OH mephenytoin metabolite formation. In this instance, only data points describing the initial linear reaction velocity were used. Results are summarized in Table 1 and clearly demonstrate that although some cross-reactivity is observed, the inhibitors demonstrate sufficient selectivity (>80% for target substrate, <20% for other substrates) at the inhibitor concentrations selected.
|
|
Incubations in the presence and absence of 1 µM quinidine were performed for the 11 remaining CYP2D6 substrates, and the disappearance half-life was determined. Four replicate incubations were performed (with and without quinidine) over four to six separate days and the data used to calculate a percentage CYP2D6 contribution. Representative examples of the human microsomal substrate depletion data (±1 µM quinidine) are provided in Fig. 2 for atomoxetine and amitriptyline.
|
|
|
Statistical analysis of the half-life data indicates that the greatest variability is observed interday (see Table 2: CV of up to 63.2%) rather than intraday. However, since for each determination the final reported result (% CYP2D6 contribution) is the relationship between the plus and minus inhibitor incubation on a single day, the intraday variability in absolute metabolic stability is automatically corrected, as demonstrated by the low variability observed in Table 3. Further statistical analysis of the data indicates that by performing the assay in the configuration used for this analysis, and using a positive control, to ensure added confidence, the assay need only be run over 2 days to give a 95% confidence interval of less than ±12% for the predicted percentage CYP2D6 contribution.
The percent contribution data were analyzed by assuming normal errors, common variance for each substrate, and the censored values were also bound to below 100. A Bayesian analysis assuming negligible prior information was implemented using the WinBUGS program, with the mean and its standard deviation of each substrate's percent CYP2D6 response being reported. The estimated mean percent CYP2D6 contribution in vitro for each substrate is summarized in Table 3.
Theoretical Considerations. Simulations were performed to predict the ratio of AUCPM/AUCEM as a function of the fraction metabolized by a polymorphic enzyme. Figure 3 shows the theoretical relationship between the ratio of AUCPM/AUCEM (equivalent to CLEM/CLPM) and the degree of enzyme functional impairment on the x-axis. As the degree of enzyme impairment becomes more severe (and closer to zero) the distinction between PM and EM exposure was predicted to become larger. The extent of P450 metabolism relative to other elimination routes can also be considered in cases where the fraction of a dose metabolized by a particular isoform (fmn) is less than 1. For instance, if fmn is 0.5 and the degree of enzyme function is reduced to 0, the analysis suggested that a 2-fold increase in AUC would be observed in PM subjects relative to EM subjects.
|
). These simulations are intended to represent the most straightforward situation, in which linear kinetics are observed and "average" differences in exposure between EM and PM subjects are presented. It was assumed that increased drug exposure due to the lack of polymorphic enzyme function resulted in concentrations that were still in the linear capacity range of alternative elimination pathways.
Literature data for known CYP2D6 substrates are summarized in Table 4, including the major products of CYP2D6 metabolism, methods of evaluation, and the estimated importance of CYP2D6 to product formation. A variety of methods have been used to characterize the metabolism of CYP2D6 substrates to specific products. In Table 5, the estimated fmCYP2D6 from in vitro and in vivo data are calculated as the product of the fractional conversion to the recovered metabolite(s) in vitro and in vivo recovery of that specific metabolite(s). A comparison of the estimated fmCYP2D6 in vivo and in vitro using microsomes and the specific inhibitor quinidine was provided. For drugs with CYP2D6 as the major determinant of clearance (>60%), large differences in AUC or oral clearance were observed between EM and PM subjects (Fig. 4). The most extreme pharmacokinetic differences were observed with dextromethorphan and tolterodine, which showed 53- and 22-fold higher oral clearances in EM subjects compared with PM subjects. Modest differences in oral clearance (3.5- to 10-fold) were observed with atomoxetine, propafenone, desipramine, and venlafaxine between EM and PM subjects. For the five drugs that had a low estimated CYP2D6 contribution in vivo, differences between EM and PM oral clearance values were minor. For amitriptyline, and propranolol, a <3-fold difference in CL/F was observed for EM and PM subjects. The remaining drugs had <2-fold decrease in CL/F for PM subjects relative to EM subjects. In general, the observed versus predicted EM/PM differences were in agreement.
|
|
|
|
Discussion |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
Based on the relationship between fraction metabolized by CYP2D6 and the observed pharmacokinetic differences in CL/F between EM and PM subjects, a recommended cutoff value for CYP2D6 metabolism can be established. Figure 4 illustrates the observed ratio of AUCPM/AUCEM and the estimated fraction metabolized by CYP2D6 in vivo. From this relationship, it can be seen that for a fraction metabolized by CYP2D6 of 60%, it would be anticipated that, on average, PM subjects would have a 2.5-fold increase in exposure relative to EM subjects. Due to the nonlinear shape of the relationship between CYP2D6 contribution and EM/PM CL/F differences, small changes in percentage contribution can resulted in large EM/PM pharmacokinetic differences when the fraction metabolized then exceeds 60%.
With more information about the safety and/or efficacy of a compound, the cutoff value can be refined. However, frequently, the challenge in drug discovery is to select compounds without a full appreciation of the safety profile. Given the wide interindividual variability observed for a typical CYP3A substrate, we believe that a 2.5-fold EM/PM CL/F difference should be manageable for the majority of compounds coming from drug discovery. Alternatively, with safety information or different assumptions based on previous experience within a therapeutic class, a more precise limit could be selected to enable a single dose level appropriate for both EM and PM subjects. Until the use of pharmacogenetic tools becomes common-place in clinical practice, it will be necessary to exclude compounds from drug discovery that do not meet preset criteria for interindividual variability.
Because of the wealth of data available for CYP2D6, this polymorphism has served as the basis for examining the in vitro-in vivo relationship more closely. Although less data exists on the impact of other polymorphisms, we feel that the results with CYP2D6 can be generalized to other situations in which a polymorphism contributes to interindividual variability in clearance, in a similar manner. As discussed previously, it is important to consider the effect of the polymorphism on enzyme function in addition to the fraction metabolized by the enzyme. The CYP2C19 polymorphism, like CYP2D6, is a null-polymorphism and results in a complete absence of enzyme activity in homozygous PM. We believe that these similarities between the CYP2D6 and CYP2C19 polymorphisms will allow for generalization of the CYP2D6 results to CYP2C19.
For CYP2C9, a number of allelic variants exist (Goldstein and de Morais, 1994; Bhasker et al., 1997); however, there appear to be only three naturally occurring in Caucasians: the wild type (CYP2C9*1), CYP2C9*2, and CYP2C9*3 (Sullivan-Klose et al., 1996). The CYP2C9*2 and CYP2C9*3 polymorphisms are associated with a reduction in enzyme function rather than complete ablation of enzyme activity. Homozygous variants for both CYP2C9*2 and CYP2C9*3 have been shown to have a poor metabolizer phenotype (Sullivan-Klose et al., 1996). The clearance of tolbutamide, phenytoin, and glipizide in individuals with the CYP2C9*3 variant has been reported at 22%, 21%, and 18%, respectively, of that in normal individuals (Miners et al., 1985; Kidd et al., 1999). Using simulations, it is possible to predict the relationship between fraction cleared through an enzyme with reduced function (EF = 0.15) and differences in EM and PM exposures (refer to Fig. 3). In this instance, the percentage contribution increases to around 70% before a 2.5-fold difference in clearance between EMs and PMs is observed. A similar analysis performed using celecoxib as a probe for CYP2C9 suggested a similar EM/PM CL/F difference using simulations (Tang et al., 2001).
There was strong agreement between the literature values for CYP2D6 contribution as determined in vivo through a variety of methods and data generated in vitro using the 1 µM quinidine/microsomal substrate depletion approach. The large disconnect (greater than 4-fold) between propranolol literature and in-house data underscores the challenge of incorporating information on other routes of metabolism into the assessment of polymorphic enzyme contribution. In vivo, other routes of elimination beyond NADPH-dependent microsomal metabolism are important to the clearance of propranolol (Walle et al., 1985), resulting in an overestimation of the importance of CYP2D6 from microsomes. The use of cryopreserved human hepatocytes may serve as a better tool for compounds with known non-P450 routes of metabolism since similar P450 inhibitor specificities can be achieved in cryopreserved hepatocytes as assessed by probe substrate incubations and those achieved in microsomal assays (Li et al., 1999).
In the initial assessment, a number of drugs that were reported to be partially metabolized by CYP2D6 could not be evaluated using the microsomal depletion approach. This remains a significant challenge for metabolically stable compounds. Other methods based upon metabolite formation, rather than substrate depletion and/or using recombinant P450 isoforms, and scaling factors may be useful. A combination of these approaches has been evaluated using amitriptyline as probe substrate (Venkatakrishnan et al., 2001a). One caveat to this approach is the potential for even greater overestimates of fraction metabolized by an individual enzyme since, in addition to non-P450 pathways being ignored, the estimates will be based solely upon the metabolites chosen to be investigated, rather than the overall P450 metabolism. An alternative approach may be to increase the incubation time for particularly stable substrates. Traditionally, it was not recommended that microsomal incubations proceed for greater than 60 min because of degradation of the P450 enzymes. However, it has been demonstrated that P450 lability can be reduced by introducing reactive oxygen species scavengers, such as superoxide dismutase, into the incubation (Foti and Fisher, 2004). Hence, this method has a potential application for characterization of more slowly metabolized compounds.
There are several additional caveats to the analysis that should also be highlighted. For compounds metabolized through multiple pathways, no consideration has been made for potential saturation of elimination pathways due to accumulation of substrate in the estimate of in vivo fm. If CYP2D6 is a high-affinity enzyme in EM subjects, its deletion in PM subjects could give rise to a situation in which the higher exposure saturates alternative clearance mechanisms. Like-wise, the in vitro assay has been performed under standard conditions of 1 µM substrate and 0.5 µM P450, which may not reflect free concentrations in vivo. If the stability of the compound was measured at concentrations that were above the Km for CYP2D6, and lower concentrations were present after oral administration, an underestimation of the role of CYP2D6 might result.
The use of microsomes as a predictive tool for clearance estimation has been well documented in the literature. For new compounds in drug discovery, it is possible to select compounds with consideration of variability arising from the complement of P450 isoforms contributing to the overall metabolic profile. Our results suggest that compounds that have more than 60% of their metabolism by CYP2D6 are likely to exhibit large differences in CL/F when comparing EM and PM subjects.
|
Footnotes |
|---|
ABBREVIATIONS: EM, extensive metabolizer; PM, poor metabolizer; fm, fraction metabolized; AUC, area under the (concentration-time) curve; CL, clearance; CL/F, oral clearance.
Address correspondence to: John P. Gibbs, Pfizer Inc, Eastern Point Road, Mail stop 8220-4186, Groton, CT 06340. E-mail: john.p.gibbs{at}pfizer.com
|
References |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  M. Kotsuma, T. Tokui, S. Freudenthaler, and K. Nishimura Effects of Ketoconazole and Quinidine on Pharmacokinetics of Pactimibe and Its Plasma Metabolite, R-125528, in Humans Drug Metab. Dispos., August 1, 2008; 36(8): 1505 - 1511. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. F. McGinnity, N. J. Waters, J. Tucker, and R. J. Riley Integrated in Vitro Analysis for the in Vivo Prediction of Cytochrome P450-Mediated Drug-Drug Interactions Drug Metab. Dispos., June 1, 2008; 36(6): 1126 - 1134. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Kotsuma, T. Tokui, T. Ishizuka-Ozeki, T. Honda, H. Iwabuchi, T. Murai, T. Ikeda, and H. Saji CYP2D6-Mediated Metabolism of a Novel Acyl Coenzyme A:Cholesterol Acyltransferase Inhibitor, Pactimibe, and Its Unique Plasma Metabolite, R-125528 Drug Metab. Dispos., March 1, 2008; 36(3): 529 - 534. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
 |
 |
|
 |
 |
 |
