Quantitative prediction of human pharmacokinetic drug-drug interactions and drug clearance using humanized liver chimeric mice: a review.

Drug clearance and drug-drug interactions (DDIs) are important in the pharmacokinetic assessment of investigational drugs, yet predicting in vivo fraction metabolized (fm) and DDI intensity remains challenging, particularly for low-clearance compounds. This study demonstrates how human liver chimeric mice (hu-PXB mice) can predict CYP2C9-mediated drug disposition for low-clearance compounds in humans. To estimate human in vitro CYP2C9 fraction metabolized (fm,CYP2C9,in vitro), 3 CYP2C9 substrates (phenytoin, tolbutamide, and warfarin) were incubated in human hepatocytes with or without sulfaphenazole (CYP2C9 inhibitor). The fm,CYP2C9,in vitro was calculated based on hepatic intrinsic clearance. For in vivo estimation (fm,CYP2C9,in vivo), clinical DDI data obtained using CYP2C9 inhibitors were analyzed to calculate fm,CYP2C9,in vivo based on observed clearance changes. To evaluate human DDI predictability, the 3 drugs were administered intravenously to hu-PXB and SCID mice with or without CYP2C9 inhibitors (sulfaphenazole or tienilic acid). Clearance changes were calculated and compared among humans, hu-PXB mice, and SCID mice. Results showed that fm,CYP2C9,in vitro values for phenytoin and tolbutamide were overestimated compared to fm,CYP2C9,in vivo, whereas warfarin could not be evaluated under current conditions. Hu-PXB mice demonstrated a better correlation with humans in both clearance changes and absolute values compared to SCID mice. Notably, hu-PXB mice predicted CYP2C9-mediated DDI magnitude within 15% of clinical values and predicted clearance for CYP2C9 substrates within 2-fold of clinical values. These findings establish hu-PXB mice as a reliable preclinical model for predicting human CYP2C9-mediated drug disposition. SIGNIFICANCE STATEMENT: Human liver chimeric mice can accurately predict the clearance and magnitude of drug-drug interaction for CYP2C9 substrate drugs. Findings from humanized mice enable the selection of better candidates in drug discovery and facilitate the design of efficient clinical trials for investigational drugs.

Drug clearance and drug-drug interactions are essential for pharmacokinetic assessment. However, current invitro systems and animal scale-up approaches often fail to accurately predict drug disposition mediated by metabolizing enzymes, especially uridine diphosphate-glucuronosyltransferase (UGT). This study demonstrates how UGT-mediated drug disposition in humans can be predicted using hu-PXB mice (cDNA-uPA/severe combined immunodeficiency (SCID) mice transplanted with human-derived hepatocytes). To estimate human hepatic intrinsic clearance (CLh,int) invitro, UGT substrates (acetaminophen, entacapone, ketoprofen, lorazepam, oxazepam, posaconazole, and zidovudine) were incubated with cryopreserved human hepatocytes. CLh,int was calculated based on the rate of substrate disappearance. Invivo human CLh,int values were calculated based on literature. To evaluate human CLh,int predictability, the 7 substrates were administered independently and intravenously to hu-PXB and SCID mice. We calculated the CLh,int in the mice and compared it with that in humans. For predicting UGT-mediated drug-drug interactions, 2 UGT substrates were administered intravenously to hu-PXB mice with or without probenecid (a UGT inhibitor). We compared the changes in clearance with those in humans. The invitro assay using hepatocytes significantly underpredicted CLh,int in humans. Hu-PXB mice had a much better correlation with humans in CLh,int (R2= 0.95) compared with SCID mice (R2= 0.69). Hu-PXB mice predicted the CLh,int of UGT substrate drugs within 2-fold of the clinical values for every compound we evaluated. The decrease in clearance caused by probenecid in hu-PXB mice reflected that in humans. Our findings demonstrate that human drug disposition mediated by UGT can be predicted based on the invivo studies using hu-PXB mice. SIGNIFICANCE STATEMENT: Human liver chimeric mice can accurately predict the clearance of uridine diphosphate-glucuronosyltransferase (UGT) substrate drugs and are likely to predict the magnitude of UGT-mediated drug-drug interactions. Findings from invivo studies in humanized mice enable the selection of better candidates in drug discovery and allow for the more precise physiologically based pharmacokinetic modeling of UGT substrate drugs in clinical practice.

Requirements for predicting human pharmacokinetics in drug discovery are increasing. Developing different methods of human pharmacokinetic prediction will facilitate lead optimization, candidate nomination, and dosing regimens before clinical trials at various early drug discovery stages. To develop and validate generic methods of human pharmacokinetic prediction to meet the requirements in early drug discovery. The physiologically based pharmacokinetic (PBPK) model implemented in Gastroplus™ was used for human pharmacokinetic predictions. The absorption, distribution, metabolism, and excretion properties of drugs in humans predicted from molecular structure and extrapolated from tested preclinical data were used as inputs in the PBPK model. The approaches were validated by comparison of the predicted pharmacokinetic parameters with actual pharmacokinetic parameters of 15 marketed small-molecule compounds approved by the US Food and Drug Administration. Based on the validation and reported approaches, we proposed a strategy for human pharmacokinetic prediction at different drug discovery stages. Obvious underestimation of exposure (< 1/3 of actual exposure) was not observed using in silico prediction as inputs, which may reduce the probability of missing the potential compounds with predicted false low exposure. The simulated human pharmacokinetic results using tested data as inputs were superior to those obtained via in silico prediction. Both methods similarly predicted the multiphasic shape of pharmacokinetic profiles. These generic PBPK approaches of full in silico prediction or perdition using a combination of tested in vivo and in vitro data were validated and proved useful for human pharmacokinetic predictions.

Characterization of drug disposition in humans using novel in vitro methodologies based on the Extended Clearance Model

Physiologically based pharmacokinetic (PBPK) modeling was used to predict the human pharmacokinetics and drug-drug interaction (DDI) of GDC-2394. PBPK models were developed using in vitro and in vivo data to reflect the oral and intravenous PK profiles of mouse, rat, dog, and monkey. The learnings from preclinical PBPK models were applied to a human PBPK model for prospective human PK predictions. The prospective human PK predictions were within 3-fold of the clinical data from the first-in-human study, which was used to optimize and validate the PBPK model and subsequently used for DDI prediction. Based on the majority of PBPK modeling scenarios using the in vitro CYP3A induction data (mRNA and activity), GDC-2394 was predicted to have no-to-weak induction potential at 900 mg twice daily (BID). Calibration of the induction mRNA and activity data allowed for the convergence of DDI predictions to a narrower range. The plasma concentrations of the 4β-hydroxycholesterol (4β-HC) were measured in the multiple ascending dose study to assess the hepatic CYP3A induction risk. There was no change in plasma 4β-HC concentrations after 7 days of GDC-2394 at 900 mg BID. A dedicated DDI study found that GDC-2394 has no induction effect on midazolam in humans, which was reflected by the totality of predicted DDI scenarios. This work demonstrates the prospective utilization of PBPK for human PK and DDI prediction in early drug development of GDC-2394. PBPK modeling accompanied with CYP3A biomarkers can serve as a strategy to support clinical pharmacology development plans. SIGNIFICANCE STATEMENT: This work presents the application of physiologically based pharmacokinetic modeling for prospective human pharmacokinetic (PK) and drug-drug interaction (DDI) prediction in early drug development. The strategy taken in this report represents a framework to incorporate various approaches including calibration of in vitro induction data and consideration of CYP3A biomarkers to inform on the overall CYP3A-related DDI risk of GDC-2394.

Predicting human pharmacokinetics (PK) such as clearance (CL) and volume of distribution (Vd) is a critical component of drug discovery. These predictions are mainly performed by in vitro-in vivo extrapolation (IVIVE) using human biological samples, such as hepatic microsomes and hepatocytes. However, some issues with this process have arisen, such as inconsistencies between in vitro and in vivo findings; the integration of predicted CYP, non-CYP and transporter-mediated human PK; and the difficulty of evaluating very metabolically stable compounds. Various approaches to solving these issues have been reported. Allometric scaling using experimental animals has also often been used. However, this method has also shown many problems due to interspecies differences, albeit that various correction methods have been proposed. Another approach involves the production of chimeric mice with humanized liver via the transplantation of human hepatocytes into mice. The livers of these mice are repopulated mostly with human hepatocytes and express human drug-metabolizing enzymes and drug transporters, suggesting that these mice are useful for solving the issues of IVIVE and allometric scaling, and more reliably predicting human PK. In this review, we summarize human PK prediction methods using IVIVE, allometric scaling and chimeric mice with humanized liver, and discuss the utility of predicting human PK in drug discovery by comparing these chimeric mice with IVIVE and allometric scaling.

Hepatic organic cation transporter 1 (OCT1) can be a determinant of drug clearance and distribution, which can impact drug exposure and response. OCT1 was shown recently to be the rate-determining step in the clearance of several drugs in humans, and thereby a mechanism of pharmacogenetic variability and drug-drug interactions (DDIs). OCT1 mediates metformin distribution to the liver (key biophase). As OCT1 modulation impacts metformin response, but not pharmacokinetics (PK), metformin DDI studies require pharmacodynamic endpoint(s) to inform rational metformin dose adjustment.

Expressive molecular representation plays critical roles in researching drug design, while effective methods are beneficial to learning molecular representations and solving related problems in drug discovery, especially for drug-drug interactions (DDIs) prediction. Recently, a lot of work has been put forward using graph neural networks (GNNs) to forecast DDIs and learn molecular representations. However, under the current GNNs structure, the majority of approaches learn drug molecular representation from one-dimensional string or two-dimensional molecular graph structure, while the interaction information between chemical substructure remains rarely explored, and it is neglected to identify key substructures that contribute significantly to the DDIs prediction. Therefore, we proposed a dual graph neural network named DGNN-DDI to learn drug molecular features by using molecular structure and interactions. Specifically, we first designed a directed message passing neural network with substructure attention mechanism (SA-DMPNN) to adaptively extract substructures. Second, in order to improve the final features, we separated the drug-drug interactions into pairwise interactions between each drug’s unique substructures. Then, the features are adopted to predict interaction probability of a DDI tuple. We evaluated DGNN–DDI on real-world dataset. Compared to state-of-the-art methods, the model improved DDIs prediction performance. We also conducted case study on existing drugs aiming to predict drug combinations that may be effective for the novel coronavirus disease 2019 (COVID-19). Moreover, the visual interpretation results proved that the DGNN-DDI was sensitive to the structure information of drugs and able to detect the key substructures for DDIs. These advantages demonstrated that the proposed method enhanced the performance and interpretation capability of DDI prediction modeling.

The role of pharmacokinetics (PK) in drug discovery is to support the optimisation of the absorption, distribution, metabolism and excretion (ADME) properties of lead compounds with the ultimate goal to attain a clinical candidate which achieves a concentration-time profile in the body that is adequate for the desired efficacy and safety profile. A thorough characterisation of the lead compounds aiming at the identification of the inherent PK liabilities also includes an early generation of PK/PD relationships linking in vitro potency and target exposure/engagement with expression of pharmacological activity (mode-of-action) and efficacy in animal studies. The chapter describes an exposure-centred approach to lead generation, lead optimisation and candidate selection and profiling that focuses on a stepwise generation of an understanding between PK/exposure and PD/efficacy relationships by capturing target exposure or surrogates thereof and cellular mode-of-action readouts in vivo. Once robust PK/PD relationship in animal PD models has been constructed, it is translated to anticipate the pharmacologically active plasma concentrations in patients and the human therapeutic dose and dosing schedule which is also based on the prediction of the PK behaviour in human as described herein. The chapter outlines how the level of confidence in the predictions increases with the level of understanding of both the PK and the PK/PD of the new chemical entities (NCE) in relation to the disease hypothesis and the ability to propose safe and efficacious doses and dosing schedules in responsive patient populations. A sound identification of potential drug metabolism and pharmacokinetics (DMPK)-related development risks allows proposing of an effective de-risking strategy for the progression of the project that is able to reduce uncertainties and to increase the probability of success during preclinical and clinical development.

ABSTRACTIntroduction: The intestinal absorption process is a combination of several events that are governed by various factors. Several transport mechanisms are involved in drug absorption through enterocytes via active and/or passive processes. The transported molecules then undergo intestinal metabolism, which together with intestinal transport may affect the systemic availability of drugs. Many studies have provided clear evidence on the significant role of intestinal first-pass metabolism on drug bioavailability and degree of drug–drug interactions (DDIs).Areas covered: This review provides an update on the role of intestinal first-pass metabolism in the oral bioavailability of drugs and prediction of DDIs. It also provides a comprehensive overview and summary of the latest update in the role of physiologically based pharmacokinetic models modeling in prediction of intestinal metabolism and DDIs in humans.Expert opinion: The contribution of intestinal first-pass metabolism in the oral bioavailability of drugs and prediction of DDIs has become more evident over the last few years. Several in vitro, in situ, and in vivo models have been developed to evaluate the role of first-pass metabolism and to predict DDIs. Currently, physiologically based pharmacokinetic modeling is considered the most valuable tool for the prediction of intestinal first-pass metabolism and DDIs.

Humanized mice that express the human UDP-glucuronosyltransferase (UGT) 1 locus have been developed in a Ugt1-null background as a model to improve predictions of human UGT1A-dependent drug clearance. Enzyme kinetic parameters (K(m) and V(max)) and pharmacokinetic properties of three probe drugs were compared using wild-type and humanized UGT1 mice that express the Gilbert's UGT1A1*28 allele [Tg(UGT1(A1*28)) Ugt1(-/-) mice]. The well characterized substrate for UGT1A1, 7-ethyl-10-hydroxy-camptothecin (SN-38), showed the greatest difference in parent drug exposure ( approximately 3-fold increase) and clearance ( approximately 3-fold decrease) in Tg(UGT1(A1*28)) Ugt1(-/-) mice after intravenous administration compared with wild-type and phenobarbital-treated animals. In contrast, the clearance of the UGT2B7 substrate (-)-17-allyl-4, 5alpha-epoxy-3, 14-dihydroxymorphinan-6-one (naloxone) was not altered in Tg(UGT1(A1*28)) Ugt1(-/-) mice. In addition, pharmacokinetic parameters with 1-(4-fluorophenyl)3(R)-[3-(4-fluorophenyl)-3(S)-hydroxypropyl]-4(S)-(4-hydroxyphenyl)-2-azetidinone (ezetimibe, Zetia; Merck & Co., Whitehouse Station, NJ), considered to be a major substrate for UGT1A1, showed small to no dependence on UGT1A1-directed glucuronidation. Enzyme kinetic parameters assessed for SN-38, ezetimibe, and naloxone using liver microsomes prepared from wild-type and Tg(UGT1(A1*28)) Ugt1(-/-) mice showed patterns consistent with the in vivo pharmacokinetic data. For SN-38 glucuronidation, V(max) decreased 5-fold in Tg(UGT1(A1*28)) Ugt1(-/-) mouse liver microsomes compared with microsomes prepared from wild-type mice, and decreased 10-fold compared with phenobarbital-treated Tg(UGT1(A1*28)) Ugt1(-/-) mice. These differences are consistent with SN-38 glucuronidation activities using HLMs isolated from individuals genotyped as UGT1A1*1 or UGT1A1*28. For ezetimibe and naloxone the differences in V(max) were minimal. Thus, Tg(UGT1(A1*28)) Ugt1(-/-) mice can serve as a pharmacokinetic model to further investigate the effects of UGT1A1 expression on drug metabolism.

A major concern in drug development is the characterization of new molecular entities (NMEs) with respect to their safety and efficacy. Both factors are determined by the drug’s exposure within the body which itself is affected by drug clearance processes. The major clearance organs are the liver and the kidney, where an interplay of metabolic enzymes and drug transporters mediates the elimination of drugs by metabolism and/or secretion. By that, inhibition of active clearance pathways, as observed from drug-drug interactions (DDIs), can result in alterations in a drug’s exposure. Therefore, the early characterization of the pharmacokinetic profile (PK) of NMEs is a major goal in preclinical drug development. However, due to lacking human in vivo PK data in this early phase of drug development, in vitro-based methods are commonly used to make a first assessment of the PK profile of NMEs. Consequently, the development, validation, and characterization of these methods is of major importance. Therefore, it was the aim of this work to investigate the prediction of human renal and hepatic drug clearances by in vitro-in vivo extrapolation (IVIVE) models and assess their feasibility to predict the DDI potential of drugs in human. To date, only few IVIVE approaches have been described to predict the human renal organ clearance based on filtration, secretion, and reabsorption. In a first study, we measured in LLC-PK1 cells the transport of 20 compounds with various physiochemical and PK properties. These data were incorporated into a novel kidney model to predict all renal clearance processes in human. Compared to reported renal clearances from clinical studies, the prediction accuracy in terms of percentage within three-fold error was 95%. Moreover, our model allowed the assessment of the contribution of filtration, secretion, and reabsorption to the net renal organ clearance in human. In a second study, we investigated the contribution of the organic anion transporting polypeptides (OATP) 1 and OATP1B3 to the net hepatic uptake clearance of statins. For this purpose, the absolute transporter protein abundances were determined by liquid chromatography-tandem mass spectrometry in cryopreserved human hepatocytes and single-transporter expressing HEK293 cells. Subsequently, uptake kinetics of eight statins and OATP1B1 and OATP1B3-specific reference substrates were determined in all expression systems. Transporter activity data generated in recombinant cell lines were extrapolated to hepatocyte values using relative transporter expression factors (REF) or relative activity factors (RAF). We showed that REF and RAF-based predictions were highly similar indicating a direct transporter expression-activity relationship. Moreover, we demonstrated that the REF-scaling method provided a powerful tool to quantitatively assess the transporter-specific contributions to the net uptake clearance of statins in hepatocytes. In a third study, we applied a recently developed IVIVE method to predict the human hepatic clearance and the DDI potential of eight statins. Application of the recently established Extended Clearance Concept Classification System (ECCCS), demonstrated a good predictability of the human hepatic clearance with six out of eight statins projected within a two-fold deviation to reported values. Furthermore, the DDI potential of the statins was assessed with respect to the impact of possible perpetrator drugs on hepatic uptake, metabolism, and biliary secretion and subsequently compared with reported clinical DDI effects. The predicted DDIs for statins showed excellent quantitative correlations with clinical observations. The ECCCS thus represents a powerful tool to anticipate the DDI potential of victim drugs based on in vitro drug metabolism and transport data. In a last study, we assessed the inhibitory potential of telaprevir, a new, direct-acting antiviral drug, on major human renal and hepatic drug transporters. By that, co-incubations of drug-transporter reference substrates and telaprevir in stable, single-transporter transfected HEK293 cells was investigated. Our data showed that telaprevir exhibited significant potential to inhibit major renal and hepatic drug transporters in human. Therefore, clinical co-administration of telaprevir together with drugs that are substrates of renal and hepatic transporters should be carefully monitored. Taken together, with the help of this work the safety profiles of NMEs can now be assessed in preclinical drug development based on in vitro methods. It is therefore expected, that the establishment, validation, and application of novel in vitro based methods, described in this work, will add significant value in the early assessment of the PK profile of NMEs.

Various CYP time-dependent inhibition (TDI) assays have been widely implemented in drug discovery and development which has led to great success in positively identifying compounds with mechanism-base inhibition liability. However, drug-drug interaction (DDI) predictions by various in-silico models utilizing kinetic parameters obtained from TDI assays have met with significant challenges including questionable kinetic data, over-simplified in-vitro models and unreliable mathematic algorithms. Although significant efforts have been made to standardize the TDI assay and refine mathematical models, recent evaluation studies have revealed that the kinetic parameters of TDI, the most important in-vitro data required by all DDI prediction models, are significantly impacted by a variety of experimental variables including microsomal protein concentration, metabolic stability, CYP-specific probes, and post-incubation time. This review attempts to provide medicinal chemists a brief overview on the current status of TDI assays, determination of kinetic parameters and in silico DDI predictions with emphasis on the complexity of the TDI kinetics and limitations of current in-vitro models and DDI prediction methodologies.

Induction of cytochrome P450 (CYP) 3A4 potentially reduces the blood concentrations of substrate drugs to less than one-tenth, which results in ineffective pharmacotherapy. Although the prediction of drug-drug interactions (DDIs) that are mediated by induction of CYP3A4 has been performed mainly on the basis of in vitro information, such methods have met with limited success in terms of their accuracy and applicability. Therefore, a realistic method for the prediction of CYP3A4-mediated inductive DDIs is of major clinical importance. The objective of the present study was to construct a robust and accurate method for the prediction of CYP3A4-mediated inductive DDIs. Such a method was developed on the basis of the principle applied for prediction of inhibitory DDIs in a previous report. A unique character of this principle is that the extent of alterations in the area under the plasma concentration-time curve (AUC) is predicted on the basis of in vivo information from minimal clinical studies without using in vitro data. The analysis is based on 42 DDI studies in humans reported in 37 published articles over the period 1983-2007. Kinetic analysis revealed that the reduction in the AUC of a substrate of CYP3A4 produced by consecutive administration of an inducer of CYP3A4 could be approximated by the equation 1/(1 + CRCYP3A4 * ICCYP3A4), where CRCYP3A4 is the ratio of the apparent contribution of CYP3A4 to the oral clearance of a substrate and ICCYP3A4 is the apparent increase in clearance of a substrate produced by induction of CYP3A4. Using this equation, the ICCYP3A4 was calculated for seven inducers (bosentan, carbamazepine, efavirenz, phenytoin, pioglitazone, rifampicin [rifampin], and St John's wort [hypericum]) on the basis of the reduction in the AUC of a coadministered standard substrate of CYP3A4, such as simvastatin, in ten DDI studies. The CRCYP3A4 was calculated for 22 substrates on the basis of the previously reported method from inhibitory DDI studies using a potent CYP3A4 inhibitor such as itraconazole or ketoconazole. The proposed method enabled the prediction of AUC reduction by CYP3A4 induction with any combination of these substrates and inducers (total 154 matches). To assess the accuracy of the prediction, the AUC reductions in 32 studies were analysed. We found that the magnitude of the deviation between the mean values of the observed and predicted AUCs of all substrate drugs was <20% of the AUCs of the respective substrate drugs before administration of the inducers. In addition, rifampicin was found to be the most potent inducer among the compounds analysed in the present study, with an ICCYP3A4 value of 7.7, followed by phenytoin and carbamazepine, with values of 4.7 and 3.0, respectively. The ICCYP3A4 values of the other CYP3A4 inducers analysed in the present study were approximately 1 or less, which suggests that the AUCs of coadministered drugs may not be reduced to less than approximately half, even if the drug is metabolized solely by CYP3A4. By using the method reported in the present study, the susceptibilities of a substrate drug of CYP3A4 to inductive DDIs can be predicted quantitatively. It was indicated that coadministration of rifampicin, phenytoin and carbamazepine may reduce plasma AUCs to less than half for a broad range of CYP3A4 substrate drugs, with CRCYP3A4 values greater than 0.13, 0.21 and 0.33, respectively.

Manifold regularized matrix factorization for drug-drug interaction prediction.