In vitro and in vivo CYP3A64 induction and inhibition studies in rhesus monkeys: a preclinical approach for CYP3A-mediated drug interaction studies

Evaluating the Disposition of a Mixed Aldehyde Oxidase/Cytochrome P450 Substrate in Rats with Attenuated P450 Activity

Marketed drugs cleared by aldehyde oxidase (AO) are few, with no known clinically relevant pharmacokinetic drug interactions associated with AO inhibition, whereas cytochrome P450 (P450) inhibition or induction mediates a number of clinical drug interactions. Little attention has been given to the consequences of coadministering a P450 inhibitor with a compound metabolized by both AO and P450. Upon discovering that VU0409106 (1) was metabolized by AO (to M1) and P450 enzymes (to M4-M6), we sought to evaluate the in vivo disposition of 1 and its metabolites in rats with attenuated P450 activity. Male rats were orally pretreated with the pan-P450 inactivator, 1-aminobenzotriazole (ABT), before an i.p. dose of 1. Interestingly, the plasma area under the curve (AUC) of M1 was increased 15-fold in ABT-treated rats, indicating a metabolic shunt toward AO resulted from the drug interaction condition. The AUC of 1 also increased 7.8-fold. Accordingly, plasma clearance of 1 decreased from 53.5 to 15.3 ml/min per kilogram in ABT-pretreated rats receiving an i.v. dose of 1. Consistent with these data, M1 formation in hepatic S9 increased with NADPH-exclusion to eliminate P450 activity (50% over reactions containing NADPH). These studies reflect possible consequences of a drug interaction between P450 inhibitors and compounds cleared by both AO and P450 enzymes. Notably, increased exposure to an AO metabolite may hold clinical relevance for active metabolites or those mediating toxicity at elevated concentrations. The recent rise in clinical drug candidates metabolized by AO underscores the importance of these findings and the need for clinical studies to fully understand these risks.

Use of 4β-hydroxycholesterol in animal and human plasma samples as a biomarker for CYP3A induction

Background: 4β-hydroxycholesterol (4βHC) has recently been proposed as a potential endogenous biomarker for CYP3A activity. Developing bioanalytical assays for 4βHC is challenging for several reasons, including endogenous background levels in plasma; the presence of free and ester forms; the inherent lack of MS sensitivity; and the presence of multiple positional isomers. Results: Bioanalytical assays in mouse, rat, dog and human plasma were adapted and modified from a previous published human plasma assay for 4βHC by using alkaline de-esterification, picolinic derivatization, a surrogate analyte (d7-4βHC) in authentic matrices and chromatographic conditions that showed good separation from isobaric, positional isomers. Conclusion: These assays were applied to multiple studies and demonstrated potential applications of 4βHC as a CYP3A biomarker across preclinical and clinical settings.

Muscle wasting associated with pathologic change is a risk factor for the exacerbation of joint swelling in collagen-induced arthritis in cynomolgus monkeys

BACKGROUND

Not only joint destruction but also muscle wasting due to rheumatoid cachexia has been problem in terms of quality of life of patients with rheumatoid arthritis (RA). In the present study, we performed histopathological examination and assessed relationships between characteristic parameters relating to muscle and joint swelling in a collagen-induced arthritis (CIA) model using cynomolgus monkeys (CMs).

METHODS

Female CMs were used and CIA was induced by twice immunizations using bovine type II collagen with Freund's complete adjuvant. Arthritis level was evaluated from the degree of swelling at the peripheral joints of the fore and hind limbs. Food consumption, body weight, and serum biochemical parameters were measured sequentially. Five or 6 animals per time point were sacrificed at 2, 3, 5 and 9 weeks after the first immunization to obtain quadriceps femoris specimens for histopathology. Pimonidazole hydrochloride was intravenously administered to determine tissue hypoxia in skeletal muscle.

RESULTS

Gradual joint swelling was observed and the maximum arthritis score was noted at Week 5. In histopathology, necrosis of muscle fiber in the quadriceps femoris was observed only at Week 2 and the most significant findings such as degeneration, atrophy, and regeneration of muscle fiber were mainly observed at Week 5. Food consumption was decreased up to Week 4 but recovered thereafter. Body weight decreased up to Week 5 and did not completely recover thereafter. A biphasic increase in serum cortisol was also observed at Weeks 2 and 5. Histopathology showed that muscle lesions were mainly composed of degeneration and atrophy of the muscle fibers, and ATPase staining revealed that the changes were more pronounced in type II muscle fiber than type I muscle fiber. In the pimonidazole experiment, mosaic pattern in skeletal muscle was demonstrated in the intact animal, but not the CIA animal. Increased arthritis score was accompanied by a decrease in serum creatinine, a marker that reflects muscle mass.

CONCLUSIONS

Muscle wasting might exacerbate joint swelling in a collagen-induced arthritis model of cynomolgus monkeys.

Cynomolgus monkey as a potential model to assess drug interactions involving hepatic organic anion transporting polypeptides: in vitro, in vivo, and in vitro-to-in vivo extrapolation

Organic anion-transporting polypeptides (OATP) 1B1, 1B3, and 2B1 can serve as the loci of drug-drug interactions (DDIs). In the present work, the cynomolgus monkey was evaluated as a potential model for studying OATP-mediated DDIs. Three cynomolgus monkey OATPs (cOATPs), with a high degree of amino acid sequence identity (91.9, 93.5, and 96.6% for OATP1B1, 1B3, and 2B1, respectively) to their human counterparts, were cloned, expressed, and characterized. The cOATPs were stably transfected in human embryonic kidney cells and were functionally similar to the corresponding human OATPs (hOATPs), as evident from the similar uptake rate of typical substrates (estradiol-17β-d-glucuronide, cholecystokinin octapeptide, and estrone-3-sulfate). Moreover, six known hOATP inhibitors exhibited similar IC(50) values against cOATPs. To further evaluate the appropriateness of the cynomolgus monkey as a model, a known hOATP substrate [rosuvastatin (RSV)]-inhibitor [rifampicin (RIF)] pair was examined in vitro; the monkey-derived parameters (RSV K(m) and RIF IC(50)) were similar (within 3.5-fold) to those obtained with hOATPs and human primary hepatocytes. In vivo, the area under the plasma concentration-time curve of RSV (3 mg/kg, oral) given 1 hour after a single RIF dose (15 mg/kg, oral) was increased 2.9-fold in cynomolgus monkeys, consistent with the value (3.0-fold) reported in humans. A number of in vitro-in vivo extrapolation approaches, considering the fraction of the pathways affected and free versus total inhibitor concentrations, were also explored. It is concluded that the cynomolgus monkey has the potential to serve as a useful model for the assessment of OATP-mediated DDIs in a nonclinical setting.

Use of in vivo animal models to assess pharmacokinetic drug-drug interactions

Animal models are used commonly in various stages of drug discovery and development to aid in the prospective assessment of drug-drug interaction (DDI) potential and the understanding of the underlying mechanism for DDI of a drug candidate. In vivo assessments in an appropriate animal model can be very valuable, when used in combination with in vitro systems, to help verify in vivo relevance of the in vitro animal-based results, and thus substantiate the extrapolation of in vitro human data to clinical outcomes. From a pharmacokinetic standpoint, a key consideration for rational selection of an animal model is based on broad similarities to humans in important physiological and biochemical parameters governing drug absorption, distribution, metabolism or excretion (ADME) processes in question for both the perpetrator and victim drugs. Equally critical are specific in vitro and/or in vivo experiments to demonstrate those similarities, usually both qualitative and quantitative, in the ADME properties/processes under investigation. In this review, theoretical basis and specific examples are presented to illustrate the utility of the animal models in assessing the potential and understanding the mechanisms of DDIs.

Pregnancy does not increase CYP3A or P-glycoprotein activity in the non-human primate, Macaca nemestrina

Plasma concentrations of protease inhibitors are lower in pregnant women than in nonpregnant women or men. Using nelfinavir as a model protease inhibitor, we have shown that this phenomenon can be reproduced in a representative non-human primate model, Macaca nemestrina (J Pharmacol Exp Ther 329:1016-1022, 2009). Nelfinavir is cleared from the body predominantly by CYP3A metabolism and P-glycoprotein (P-gp) efflux. Therefore, using midazolam (MDZ) as a CYP3A probe and digoxin (DIG) as a P-gp probe, we determined the antepartum (73-118 days) and postpartum (61-130 days) in vivo intestinal and hepatic CYP3A or P-gp activity in the macaque. Although the systemic clearance of MDZ was significantly increased ( approximately 70%) during pregnancy after intra-arterial (IA) administration of the drug ((15)N-labeled MDZ; 40 microg/kg), pregnancy did not affect the oral clearance of the drug administered simultaneously (1 mg/kg p.o.) with the IA dose. In vitro studies in hepatic and intestinal S-9 fractions indicated no effect of pregnancy on CYP3A activity or protein expression in the small intestine or liver. In contrast, neither the oral (100 microg/kg) nor the IA (10 microg/kg) clearance of DIG was significantly altered by pregnancy, indicating no effect of pregnancy on P-gp activity. Assuming that MDZ and DIG are selective substrates of the macaque CYP3A enzymes and P-gp, respectively, these results suggest that factors other than increased CYP3A or P-gp activity contribute to the increased clearance of protease inhibitors during M. nemestrina pregnancy.

As in humans, pregnancy increases the clearance of the protease inhibitor nelfinavir in the nonhuman primate Macaca nemestrina

The apparent oral clearance of protease inhibitors (PIs) is increased in pregnant women. Although this phenomenon is reproduced in the mouse, because of the multiplicity of mouse cytochrome P450 isoforms, lack of information on their substrate and inhibitor selectivity, and lack of reagents (e.g., antibodies, purified protein), it is difficult to study the mechanistic basis of this phenomenon in this animal model. To investigate the mechanistic basis of this phenomenon in a more representative model, the nonhuman primate, we first determined whether this phenomenon could be reproduced in Macaca nemestrina, using nelfinavir as a model PI. Consistent with the human and mouse studies, we found that the apparent oral clearance of nelfinavir (NFV) in the macaques was significantly increased (3.14-fold) antepartum (n = 3) versus postpartum (n = 4). This increased apparent oral clearance was a result of an increased systemic clearance (1.9-fold) and a decreased bioavailability (approximately 45%) during pregnancy. In vitro, pregnancy significantly enhanced the rate of NFV depletion in hepatic, but not intestinal S-9 fractions. Human CYP3A inhibitors erythromycin (0.5 mM), ketoconazole (0.5 microM), and troleandomycin (0.01-1 mM), but not the CYP2C inhibitor, sulfaphenazole (3 microM), significantly inhibited the depletion of NFV in hepatic S-9 fractions and expressed rhesus CYP3A64 enzyme. Based on these data, we conclude that increased hepatic activity of NFV-metabolizing enzymes (perhaps CYP3A enzymes) results in increased clearance of PIs during pregnancy in the macaques. The M. nemestrina should be further investigated as a model to study the mechanisms by which the clearance of PIs is increased during pregnancy.

Current industrial practices in assessing CYP450 enzyme induction: preclinical and clinical

Induction of drug metabolizing enzymes, such as the cytochromes P450 (CYP) is known to cause drug-drug interactions due to increased elimination of co-administered drugs. This increased elimination may lead to significant reduction or complete loss of efficacy of the co-administered drug. Due to the significance of such drug interactions, many pharmaceutical companies employ screening and characterization models which predict CYP enzyme induction to avoid or attenuate the potential for drug interactions with new drug candidates. The most common mechanism of CYP induction is transcriptional gene activation. Activation is mediated by nuclear receptors, such as AhR, CAR, and PXR that function as transcription factors. Early high throughput screening models utilize these nuclear hormone receptors in ligand binding or cell-based transactivation/reporter assays. In addition, immortalized hepatocyte cell lines can be used to assess enzyme induction of specific drug metabolizing enzymes. Cultured primary human hepatocytes, the best established in vitro model for predicting enzyme induction and most accepted by regulatory agencies, is the predominant assay used to evaluate induction of a wide variety of drug metabolizing enzymes. These in vitro models are able to appropriately predict enzyme induction in patients when compared to clinical drug-drug interactions. Finally, transgenic animal models and the cynomolgus monkey have also been shown to recapitulate human enzyme induction and may be appropriate in vivo animal models for predicting human drug interactions.

Diclofenac hydroxylation in monkeys: Efficiency, regioselectivity, and response to inhibitors

The catalytic efficiency, regioselectivity, and response to chemical inhibitors of diclofenac (DF) hydroxylation in three Old World monkey liver microsomes (rhesus, cynomolgus, and African green monkey) are different from those determined with human liver microsomes. In contrast to the high affinity-high capacity (low Km-high Vmax) characteristics of DF 4'-hydroxylation in humans, this reaction proceeded in all monkey species with catalytic efficiencies >20-fold lower. However, DF 5-hydroxylation, a negligible reaction in human liver microsomes, was kinetically favored in monkeys mainly due to the increased Vmax values. Chemical inhibitors (reversible or mechanism-based) selective to human CYP3A4 and CYP2C9 failed to differentiate monkey orthologs involved in DF hydroxylation. Immunoinhibition studies with monoclonal antibodies against human CYPs revealed the major contribution of CYP2C and CYP3A to 4'- and to 5-hydroxylation, respectively, in rhesus and cynomolgus liver microsomes. However, in African green monkeys, in addition to CYP2C, CYP3A also appeared to be involved in 4'-hydroxylation. Further studies with recombinant rhesus and African green monkey CYP2C and CYP3A enzymes (rhesus CYP2C75, 2C74, and 3A64; African green monkey CYP2C9agm and CYP3A4agm) confirmed the major role of CYP enzymes of these two subfamilies in DF 4'- and 5-hydroxylation. Clearly, while monkey CYP2C and 3A enzymes retain the same substrate selectivity towards DF hydroxylation as their human orthologs, their altered catalytic efficiency and response to chemical inhibitors may indicate different structural features of active sites as opposed to human orthologs.

Rifampin induces the in vitro oxidative metabolism, but not the in vivo clearance of diclofenac in rhesus monkeys

Effects of rifampin on in vitro oxidative metabolism and in vivo pharmacokinetics of diclofenac (DF), a prototypic CYP2C9 marker substrate, were investigated in rhesus monkeys. In monkey hepatocytes, rifampin markedly induced DF 4'-hydroxylase activity, with values for EC(50) of 0.2 to 0.4 microM and E(max) of 2- to 5-fold over control. However, pretreatment with rifampin did not alter the pharmacokinetics of DF obtained after either i.v. or intrahepatic portal vein (i.pv.) administration of DF to monkeys. At the dose studied, plasma concentrations of rifampin reached 10 microM, far exceeding the in vitro EC(50) values. Under similar treatment conditions, rifampin was previously shown to induce midazolam (MDZ) 1'-hydroxylation in rhesus monkey hepatocytes (EC(50) and E(max) values approximately 0.2 microM and approximately 2- to 3-fold, respectively), and markedly affected the in vivo pharmacokinetics of MDZ (>10-fold decreases in the i.pv. MDZ systemic exposure and its hepatic availability, F(h)) in this animal species. In monkey liver microsomes, DF underwent, predominantly, glucuronidation, and, modestly, oxidation; the intrinsic clearance (CL(int) = V(max)/K(m)) value for the glucuronidation pathway accounted for >95% (versus about 75% in human liver microsomes) of the total (glucuronidation + hydroxylation) intrinsic clearance value. In monkey hepatocytes, the hydroxylation also was a minor component (< or =10%) relative to the glucuronidation, supporting the liver microsomal finding. Collectively, our results suggest that the oxidative metabolism is not the major in vivo clearance mechanism of DF in either untreated or rifampin-treated monkeys and, conceivably, also in humans, raising a question about the utility of DF as an in vivo CYP2C9 probe.