In Vitro-In Vivo Extrapolation and Hepatic Clearance-Dependent Underprediction

A Combination of Machine Learning and PBPK Modeling Approach for Pharmacokinetics Prediction of Small Molecules in Humans

PURPOSE

Recently, there has been rapid development in model-informed drug development, which has the potential to reduce animal experiments and accelerate drug discovery. Physiologically based pharmacokinetic (PBPK) and machine learning (ML) models are commonly used in early drug discovery to predict drug properties. However, basic PBPK models require a large number of molecule-specific inputs from in vitro experiments, which hinders the efficiency and accuracy of these models. To address this issue, this paper introduces a new computational platform that combines ML and PBPK models. The platform predicts molecule PK profiles with high accuracy and without the need for experimental data.

METHODS

This study developed a whole-body PBPK model and ML models of plasma protein fraction unbound ( f up ), Caco-2 cell permeability, and total plasma clearance to predict the PK of small molecules after intravenous administration. Pharmacokinetic profiles were simulated using a "bottom-up" PBPK modeling approach with ML inputs. Additionally, 40 compounds were used to evaluate the platform's accuracy.

RESULTS

Results showed that the ML-PBPK model predicted the area under the concentration-time curve (AUC) with 65.0 % accuracy within a 2-fold range, which was higher than using in vitro inputs with 47.5 % accuracy.

CONCLUSION

The ML-PBPK model platform provides high accuracy in prediction and reduces the number of experiments and time required compared to traditional PBPK approaches. The platform successfully predicts human PK parameters without in vitro and in vivo experiments and can potentially guide early drug discovery and development.

Introducing the Dynamic Well-Stirred Model for Predicting Hepatic Clearance and Extraction Ratio

The well-stirred model (WSM) incorporating the fraction of unbound drug (fu) to account for the effect of plasma binding on intrinsic clearance has been widely used for predicting hepatic clearance under the assumption that drug protein binding reaches equilibrium instantaneously. Our theoretical analysis reveals that the effect of protein binding on intrinsic clearance is better accounted for with the dynamic free fraction (fD), a measure of drug protein binding affinity, which leads to a putative dynamic well-stirred model (dWSM) without the instantaneous equilibrium assumption. Using recombinant CYP3A4 as the in vitro clearance system, we demonstrate that the binding effect of albumin on the intrinsic clearance of both highly bound midazolam and highly free verapamil is fully corrected by their corresponding fD values, respectively. On the other hand, fu only corrects the binding effect of albumin on the intrinsic clearance of verapamil, and yields severe over-correction of the intrinsic clearance of midazolam. The results suggest that the traditional WSM is suitable for highly free drugs like verapamil but not necessarily for highly bound drugs such as midazolam due to the violation of the instantaneous equilibrium assumption or under-estimating the true free drug concentration. In comparison, the dWSM incorporating fD holds true as long as drug elimination follows steady-state kinetics, and hence, it is more broadly applicable to drugs with different protein binding characteristics. Here we demonstrate with 36 diverse drugs, that the dWSM significantly improves the accuracy of predicting human hepatic clearance and liver extraction ratio from in vitro microsomal clearance data, highlighting the importance of drug plasma protein binding kinetics in addressing the under-prediction of hepatic clearance by the WSM.

Predicting the in vivo developmental toxicity of fenarimol from in vitro toxicity data using PBTK modelling-facilitated reverse dosimetry approach

In vitro methods are widely used in modern toxicological testing; however, the data cannot be directly employed for risk assessment. In vivo toxicity of chemicals can be predicted from in vitro data using physiologically based toxicokinetic (PBTK) modelling-facilitated reverse dosimetry (PBTK-RD). In this study, a minimal-PBTK model was constructed to predict the in-vivo kinetic profile of fenarimol (FNL) in rats and humans. The model was verified by comparing the observed and predicted pharmacokinetics of FNL for rats (calibrator) and further applied to humans. Using the PBTK-RD approach, the reported in vitro developmental toxicity data for FNL was translated to in vivo dose-response data to predict the assay equivalent oral dose in rats and humans. The predicted assay equivalent rat oral dose (36.46 mg/kg) was comparable to the literature reported in vivo BMD10 value (22.8 mg/kg). The model was also employed to derive the chemical-specific adjustment factor (CSAF) for interspecies toxicokinetics variability of FNL. Further, Monte Carlo simulations were performed to predict the population variability in the plasma concentration of FNL and to derive CSAF for intersubject human kinetic differences. The comparison of CSAF values for interspecies and intersubject toxicokinetic variability with their respective default values revealed that the applied uncertainty factors were adequately protective.

Drug Design and Success of Prospective Mouse In Vitro-In Vivo Extrapolation (IVIVE) for Predictions of Plasma Clearance (CLp) from Hepatocyte Intrinsic Clearance (CLint)

Hepatocyte intrinsic clearance (CL_int) and methods of in vitro-in vivo extrapolation (IVIVE) are often used to predict plasma clearance (CL_p) in drug discovery. While the prediction success of this approach is dependent on the chemotype, specific molecular properties and drug design features that govern these outcomes are poorly understood. To address this challenge, we investigated the success of prospective mouse CL_p IVIVE across 2142 chemically diverse compounds. Dilution scaling, which assumes that the free fraction in hepatocyte incubations (f_u,inc) is governed by binding to the 10% of serum in the incubation medium, was used as our default CL_p IVIVE approach. Results show that predictions of CL_p are better for smaller (molecular weight (MW) < 500 Da), less polar (total polar surface area (TPSA) < 100 Ų, hydrogen bond donor (HBD) ≤1, hydrogen bond acceptor (HBA) ≤ 6), lipophilic (log D > 3), and neutral compounds, with low HBD count playing the key role. If compounds are classified according to their chemical space, predictions were good for compounds resembling central nervous system (CNS) drugs [average absolute fold error (AAFE) of 2.05, average fold error (AFE) of 0.90], moderate for classical druglike compounds (according to Lipinski, Veber, and Ghose guidelines; AAFE of 2.55; AFE of 0.68), and poor for nonclassical "beyond the rule of 5" compounds (AAFE of 3.31; AFE of 0.41). From the perspective of measured druglike properties, predictions of CL_p were better for compounds with moderate-to-high hepatocyte CL_int (>10 μL/min/10⁶ cells), high passive cellular permeability (P_app > 100 nm/s), and moderate observed CL_p (5-50 mL/min/kg). Influences of plasma protein binding (f_u,p) and P-glycoprotein (Pgp) apical efflux ratio (AP-ER) were less pronounced. If the extended clearance classification system (ECCS) is applied, predictions were good for class 2 (P_app > 50 nm/s; neutral or basic; AAFE of 2.35; AFE of 0.70) and acceptable for class 1A compounds (AAFE of 2.98; AFE of 0.70). Classes 1B, 3 A/B, and 4 showed poor outcomes (AAFE > 3.80; AFE < 0.60). Functional groups trending toward weaker CL_p IVIVE were esters, carbamates, sulfonamides, carboxylic acids, ketones, primary and secondary amines, primary alcohols, oxetanes, and compounds liable to aldehyde oxidase metabolism, likely due to multifactorial reasons. Multivariate analysis showed that multiple properties are relevant, combining together to define the overall success of CL_p IVIVE. Our results indicate that the current practice of prospective CL_p IVIVE is suitable only for CNS-like compounds and well-behaved classical druglike space (e.g., high permeability or ECCS class 2) without challenging functional groups. Unfortunately, based on existing mouse data, prospective CL_p IVIVE for complex and nonclassical chemotypes is poor and hardly better than random guessing. This is likely due to complexities such as extrahepatic metabolism and transporter-mediated disposition which are poorly captured by this methodology. With small-molecule drug discovery increasingly evolving toward nonclassical and complex chemotypes, existing CL_p IVIVE methodology will require improvement. While empirical correction factors may bridge the gap in the near future, improved and new in vitro assays, data integration models, and machine learning (ML) methods are increasingly needed to address this challenge and reduce the number of nonclinical pharmacokinetic (PK) studies.

Effect of Binding Linkers on the Efficiency and Metabolite Profile of Biomimetic Reactions Catalyzed by Immobilized Metalloporphyrin

The investigation of liver-related metabolic stability of a drug candidate is a widely used key strategy in early-stage drug discovery. Metalloporphyrin-based biomimetic catalysts are good and well-described models of the function of CyP450 in hepatocytes. In this research, the immobilization of an iron porphyrin was performed on nanoporous silica particles via ionic interactions. The effect of the metalloporphyrin binding linkers was investigated on the catalytic efficiency and the metabolic profile of chloroquine as a model drug. The length of the amino-substituted linkers affects the chloroquine conversion as well as the ratio of human major and minor metabolites. While testing the immobilized catalysts in the continuous-flow reactor, results showed that the presented biomimetic system could be a promising alternative for the early-stage investigation of drug metabolites regarding analytical or synthetic goals as well.

Predicting transporter mediated drug-drug interactions via static and dynamic physiologically based pharmacokinetic modeling: A comprehensive insight on where we are now and the way forward

The greater utilization and acceptance of physiologically-based pharmacokinetic (PBPK) modeling to evaluate the potential metabolic drug-drug interactions is evident by the plethora of literature, guidance's, and regulatory dossiers available in the literature. In contrast, it is not widely used to predict transporter-mediated DDI (tDDI). This is attributed to the unavailability of accurate transporter tissue expression levels, the absence of accurate in vitro to in vivo extrapolations (IVIVE), enzyme-transporter interplay, and a lack of specific probe substrates. Additionally, poor understanding of the inhibition/induction mechanisms coupled with the inability to determine unbound concentrations at the interaction site made tDDI assessment challenging. Despite these challenges, continuous improvements in IVIVE approaches enabled accurate tDDI predictions. Furthermore, the necessity of extrapolating tDDI's to special (pediatrics, pregnant, geriatrics) and diseased (renal, hepatic impaired) populations is gaining impetus and is encouraged by regulatory authorities. This review aims to visit the current state-of-the-art and summarizes contemporary knowledge on tDDI predictions. The current understanding and ability of static and dynamic PBPK models to predict tDDI are portrayed in detail. Peer-reviewed transporter abundance data in special and diseased populations from recent publications were compiled, enabling direct input into modeling tools for accurate tDDI predictions. A compilation of regulatory guidance's for tDDI's assessment and success stories from regulatory submissions are presented. Future perspectives and challenges of predicting tDDI in terms of in vitro system considerations, endogenous biomarkers, the use of empirical scaling factors, enzyme-transporter interplay, and acceptance criteria for model validation to meet the regulatory expectations were discussed.

Unravelling the pharmacokinetics of aflatoxin B1: In vitro determination of Michaelis–Menten constants, intrinsic clearance and the metabolic contribution of CYP1A2 and CYP3A4 in pooled human liver microsomes

Mycotoxins, fungal secondary metabolites, are ubiquitously present in food commodities. Acute exposure to high levels or chronic exposure to low levels has an impact on the human body. The phase I metabolism in the human liver, performed by cytochrome P450 (CYP450) enzymes, is accountable for more than 80% of the overall metabolism of exogenous and endogenous compounds. Mycotoxins are (partially) metabolized by CYP450 enzymes. In this study, in vitro research was performed on CYP450 probes and aflatoxin B1 (AFB1), a carcinogenic mycotoxin, to obtain pharmacokinetic data on AFB1, required for further experimental work. The CYP450 probes of choice were a CYP3A4 substrate, midazolam (MDZ) and a CYP1A2 substrate, phenacetin (PH) since these are the main metabolizing phase I enzymes of AFB1. Linearity experiments were performed on the three substrates indicating that linear conditions were achieved at a microsomal protein concentration and incubation time of 0.25 mg/ml and 5 min, 0.50 mg/ml and 20 min and 0.25 mg/ml and 5 min for MDZ, PH and AFB1, respectively. The K_m was determined in human liver microsomes and was estimated at 2.15 μM for MDZ, 40.0 μM for PH and 40.9 μM for AFB1. The associated V_max values were 956 pmol/(mg.min) (MDZ), 856 pmol/(mg.min) (PH) and 11,536 pmol/(mg.min) (AFB1). Recombinant CYP systems were used to determine CYP450-specific Michaelis–Menten values for AFB1, leading to a CYP3A4 K_m of 49.6 μM and an intersystem extrapolation factor (ISEF) corrected V_max of 43.6 pmol/min/pmol P450 and a CYP1A2 K_m of 58.2 μM and an ISEF corrected V_max of 283 pmol/min/pmol P450. An activity adjustment factor (AAF) was calculated to account for differences between microsome batches and was used as a correction factor in the determination of the human in vivo hepatic clearance for MDZ, PH and AFB1. The hepatic blood clearance corrected for the AAF CL_H,B,MDZ,AAF, CL_H,B,PH,AAF CL_{H,B,AFB1,AAF(CYP3A4)} and CL_{H,B,AFB1,AAF(CYP1A2)} were determined in HLM at 44.1 L/h, 21.7 L/h, 40.0 L/h and 38.5 L/h. Finally, inhibition assays in HLM showed that 45% of the AFB1 metabolism was performed by CYP3A4/3A5 enzymes and 49% by CYP1A2 enzymes.

Drug clearance by aldehyde oxidase: can we avoid clinical failure?

Despite increased awareness of aldehyde oxidase (AO) as a major drug-metabolising enzyme, predicting the pharmacokinetics of its substrates remains challenging. Several drug candidates have been terminated due to high clearance, which were subsequently discovered to be AO substrates. Even retrospective extrapolation of human clearance, from models more sensitive to AO activity, often resulted in underprediction.The questions of the current work thus were: Is there an acceptable degree of in vitro AO metabolism that does not result in high in vivo human clearance? And, if so, how can this be predicted?We built an in vitro/in vivo correlation using known AO substrates, combining multiple in vitro parameters to calculate the blood metabolic clearance mediated by AO (CLb_AO). This value was compared with observed blood clearance (CLb_-obs), establishing cut-off CLb_AO values, to discriminate between low and high CLb_-obs. The model was validated using additional literature compounds, and CLb_-obs was predicted in the correct category.This simple, categorical, semi-quantitative yet multi-factorial model is readily applicable in drug discovery. Further, it is valuable for high-clearance compounds, as it predicts the CLb group, rather than an exact CLb value, for the substrates of this poorly-characterised enzyme.

Evidence of the Need for Modified Well-stirred Model in In Vitro to In Vivo Extrapolation

In vitro to in vivo extrapolation (IVIVE), an approach for hepatic clearance (CL_H) prediction used worldwide, remains controversial due to systematic underprediction. Among the various probable factors, the original assumption of the hepatic mathematical model (i.e., the well-stirred model, WSM) may become problematic, leading to the underestimation of drug CL_H. Having a similar prerequisite that the well-stirred conditions are homogenous with perfectly mixed reactants, but using a different driving concentration, the modified well-stirred model (MWSM) stands apart from the WSM. However, we believe that both models should coexist so that the entire well-stirred scenario can be completely illustrated. Consequently, we collected published data from the literature and employed a logistic regression method to differentiate the optimal timing of use between WSM and MWSM in drug CL_H prediction. Generally, variances adopted in the regression, including partition coefficient (logP), fraction unbound (fu), volumes of distribution at steady-state (Vss), and mean residence time (MRT), corresponded to our assumption when protein-facilitated uptake was considered. Furthermore, a new empirical approach was introduced to allow practical use of the MWSM. The results showed that this model could provide a more precise prediction compared to previous empirical approaches. Therefore, these preliminary results not only delineated a more detailed structure and mechanism of MWSM but also highlighted its necessity and potential.

Utilizing virtual experiments to increase understanding of discrepancies involving in vitro-to-in vivo predictions of hepatic clearance

Predictions of xenobiotic hepatic clearance in humans using in vitro-to-in vivo extrapolation methods are frequently inaccurate and problematic. Multiple strategies are being pursued to disentangle responsible mechanisms. The objective of this work is to evaluate the feasibility of using insights gained from independent virtual experiments on two model systems to begin unraveling responsible mechanisms. The virtual culture is a software analog of hepatocytes in vitro, and the virtual human maps to hepatocytes within a liver within an idealized model human. Mobile objects (virtual compounds) map to amounts of xenobiotics. Earlier versions of the two systems achieved quantitative validation targets for intrinsic clearance (virtual culture) and hepatic clearance (virtual human). The major difference between the two systems is the spatial organization of the virtual hepatocytes. For each pair of experiments (virtual culture, virtual human), hepatocytes are configured the same. Probabilistic rules govern virtual compound movements and interactions with other objects. We focus on highly permeable virtual compounds and fix their extracellular unbound fraction at one of seven values (0.05–1.0). Hepatocytes contain objects that can bind and remove compounds, analogous to metabolism. We require that, for a subset of compound properties, per-hepatocyte compound exposure and removal rates during culture experiments directly predict corresponding measures made during virtual human experiments. That requirement serves as a cross-system validation target; we identify compound properties that enable achieving it. We then change compound properties, ceteris paribus, and provide model mechanism-based explanations for when and why measures made during culture experiments under- (or over-) predict corresponding measures made during virtual human experiments. The results show that, from the perspective of compound removal, the organization of hepatocytes within virtual livers is more efficient than within cultures, and the greater the efficiency difference, the larger the underprediction. That relationship is noteworthy because most in vitro-to-in vivo extrapolation methods abstract away the structural organization of hepatocytes within a liver. More work is needed on multiple fronts, including the study of an expanded variety of virtual compound properties. Nevertheless, the results support the feasibility of the approach and plan.

Application of empirical scalars to enable early prediction of human hepatic clearance using IVIVE in drug discovery: an evaluation of 173 drugs

The utilization of in vitro data to predict drug pharmacokinetics (PK) in vivo has been a consistent practice in early drug discovery for decades. However, its success is hampered by mispredictions attributed to uncharacterized biological phenomena/experimental artifacts. Predicted drug clearance (CL) from experimental data (i.e. hepatocyte intrinsic clearance: CLint, fraction unbound in plasma: fu,p) is often systematically underpredicted using the well-stirred model (WSM). The objective of this study was to evaluate using empirical scalars in the WSM to correct for CL mispredictions. Drugs (N=28) were used to generate numerical scalars on CLint (α), and fu,p (β) to minimize the error (AAFE) for CL predictions. These scalars were validated using an additional dataset (N=28 drugs) and applied to a non-redundant AstraZeneca (AZ) dataset available in the literature (N=117 drugs) for a total of 173 compounds. CL predictions using the WSM were improved for most compounds using an α value of 3.66 (~64%<2-fold) compared to no scaling (~46%<2-fold). Similarly, using a β value of 0.55 or combination of α and β scalars (values of 1.74 and 0.66, respectively) resulted in a similar improvement in predictions (~64%<2-fold and ~65%<2-fold, respectively). For highly bound compounds (fu,p{less than or equal to}0.01), AAFE was substantially reduced across all scaling methods. Using the β scalar alone or a combination of α and β appeared optimal; and produce larger magnitude corrections for highly-bound compounds. Some drugs are still disproportionally mispredicted, however the improvements in prediction error and simplicity of applying these scalars suggests its utility for early-stage CL predictions. Significance Statement In early drug discovery, prediction of human clearance using in vitro experimental data plays an essential role in triaging compounds prior to in vivo studies. These predictions have been systematically underestimated. Here we introduce empirical scalars calibrated on the extent of plasma protein binding that appear to improve clearance prediction across multiple datasets. This approach can be used in early phases of drug discovery prior to the availability of pre-clinical data for early quantitative predictions of human clearance.

IVIVE: Facilitating the Use of In Vitro Toxicity Data in Risk Assessment and Decision Making

During the past few decades, the science of toxicology has been undergoing a transformation from observational to predictive science. New approach methodologies (NAMs), including in vitro assays, in silico models, read-across, and in vitro to in vivo extrapolation (IVIVE), are being developed to reduce, refine, or replace whole animal testing, encouraging the judicious use of time and resources. Some of these methods have advanced past the exploratory research stage and are beginning to gain acceptance for the risk assessment of chemicals. A review of the recent literature reveals a burst of IVIVE publications over the past decade. In this review, we propose operational definitions for IVIVE, present literature examples for several common toxicity endpoints, and highlight their implications in decision-making processes across various federal agencies, as well as international organizations, including those in the European Union (EU). The current challenges and future needs are also summarized for IVIVE. In addition to refining and reducing the number of animals in traditional toxicity testing protocols and being used for prioritizing chemical testing, the goal to use IVIVE to facilitate the replacement of animal models can be achieved through their continued evolution and development, including a strategic plan to qualify IVIVE methods for regulatory acceptance.

Tissue Drug Concentration

Blood flow enables the delivery of oxygen and nutrients to the different tissues of the human body. Drugs follow the same route as oxygen and nutrients; thus, drug concentrations in tissues are highly dependent on the blood flow fraction delivered to each of these tissues. Although the free drug concentration in blood is considered to correlate with pharmacodynamics, the pharmacodynamics of a drug is actually primarily commanded by the concentrations of drug in the aqueous spaces of bodily tissues. However, the concentrations of drug are not homogeneous throughout the tissues, and they rarely reflect the free drug concentration in the blood. This heterogeneity is due to differences in the blood flow fraction delivered to the tissues and also due to membrane transporters, efflux pumps, and metabolic enzymes. The rate of drug elimination from the body (systemic elimination) depends more on the driving force of drug elimination than on the free concentration of drug at the site from which the drug is being eliminated. In fact, the actual free drug concentration in the tissues results from the balance between the input and output rates. In the present paper, we develop a theoretical concept regarding solute partition between intravascular and extravascular spaces; discuss experimental research on aqueous/non-aqueous solute partitioning and clinical research on microdialysis; and present hypotheses to predict in-vivo elimination using parameters of in-vitro metabolism.

Integrating toxicokinetics into toxicology studies and the human health risk assessment process for chemicals: Reduced uncertainty, better health protection

The database of practical examples where toxicokinetic (TK) data has benefitted all stages of the human health risk assessment process are increasingly being published and accepted. This review aimed to highlight and summarise notable examples and to describe the "state of the art" in this field. The overall recommendation is that for any in vivo animal study conducted, measurements of TK should be very carefully considered for inclusion as the numerous benefits this brings continues to grow, particularly during the current march towards animal free toxicology testing and ambitions to eventually conduct human health risk assessments entirely based upon non-animal methods.

The Artemiside-Artemisox-Artemisone-M1 Tetrad: Efficacies against Blood Stage P. falciparum Parasites, DMPK Properties, and the Case for Artemiside

Because of the need to replace the current clinical artemisinins in artemisinin combination therapies, we are evaluating fitness of amino-artemisinins for this purpose. These include the thiomorpholine derivative artemiside obtained in one scalable synthetic step from dihydroartemisinin (DHA) and the derived sulfone artemisone. We have recently shown that artemiside undergoes facile metabolism via the sulfoxide artemisox into artemisone and thence into the unsaturated metabolite M1; DHA is not a metabolite. Artemisox and M1 are now found to be approximately equipotent with artemiside and artemisone in vitro against asexual P. falciparum (Pf) blood stage parasites (IC₅₀ 1.5–2.6 nM). Against Pf NF54 blood stage gametocytes, artemisox is potently active (IC₅₀ 18.9 nM early-stage, 2.7 nM late-stage), although against the late-stage gametocytes, activity is expressed, like other amino-artemisinins, at a prolonged incubation time of 72 h. Comparative drug metabolism and pharmacokinetic (DMPK) properties were assessed via po and iv administration of artemiside, artemisox, and artemisone in a murine model. Following oral administration, the composite C_max value of artemiside plus its metabolites artemisox and artemisone formed in vivo is some 2.6-fold higher than that attained following administration of artemisone alone. Given that efficacy of short half-life rapidly-acting antimalarial drugs such as the artemisinins is associated with C_max, it is apparent that artemiside will be more active than artemisone in vivo, due to additive effects of the metabolites. As is evident from earlier data, artemiside indeed possesses appreciably greater efficacy in vivo against murine malaria. Overall, the higher exposure levels of active drug following administration of artemiside coupled with its synthetic accessibility indicate it is much the preferred drug for incorporation into rational new artemisinin combination therapies.

A Quantitative Approach to the Prediction of Drug-Drug Interactions Mediated by Cytochrome P450 2C8 Inhibition

BACKGROUND

Ohno and Colleagues proposed an approach for predicting drug-drug interactions (DDIs) mediated by cytochrome P450 (CYP) 3A4 based on the use of the ratio of the inhibited to non-inhibited area under the plasma concentration time curve (AUC) of substrates to estimate the fraction of the dose metabolized via CYP3A4 (contribution ratio, CR) and the in vivo inhibitory potency of a perpetrator (inhibition ratio, IR). This study evaluated the performance of this approach on DDIs mediated by CYP2C8 inhibitors.

RESEARCH DESIGN AND METHODS

Initial estimates of CR and IR of CYP2C8 substrates and inhibitors were calculated for 33 DDI in vivo studies. The approach was externally validated with 17 additional studies. Bayesian orthogonal regression was used to refine the estimates of the parameters. Assessment of prediction success was conducted by plotting observed versus predicted AUC ratios.

RESULTS

Final estimates of CRs and IRs were obtained for 19 CYP2C8 substrates and 23 inhibitors, respectively. The method demonstrated good predictive capacity, with only two values outside of the prespecified limits.

CONCLUSIONS

The approach may help to adapt dose regimens for CYP2C8 substrates when given in combination with CYP2C8 inhibitors and to map the potential DDIs of new molecular entities.

Prediction of hepatic drug clearance with a human microfluidic four-cell liver acinus microphysiology system

Predicting human hepatic clearance remains a fundamental challenge in both pharmaceutical drug development and toxicological assessments of environmental chemicals, with concerns about both accuracy and precision of in vitro-derived estimates. Suggested sources of these issues have included differences in experimental protocols, differences in cell sourcing, and use of a single cell type, liver parenchymal cells (hepatocytes). Here we investigate the ability of human microfluidic four-cell liver acinus microphysiology system (LAMPS) to make predictions as to hepatic clearance for seven representative compounds: Caffeine, Pioglitazone, Rosiglitazone, Terfenadine, Tolcapone, Troglitazone, and Trovafloxacin. The model, whose reproducibility was recently confirmed in an inter-lab comparison, was constructed using primary human hepatocytes or human induced pluripotent stem cell (iPSC)-derived hepatocytes and 3 human cell lines for the endothelial, Kupffer and stellate cells. We calculated hepatic clearance estimates derived from experiments using LAMPS or traditional 2D cultures and compared the outcomes with both in vivo human clinical study-derived and in vitro human hepatocyte suspension culture-derived values reported in the literature. We found that, compared to in vivo clinically-derived values, the LAMPS model with iPSC-derived hepatocytes had higher precision as compared to primary cells in suspension or 2D culture, but, consistent with previous studies in other microphysiological systems, tended to underestimate in vivo clearance. Overall, these results suggest that use of LAMPS and iPSC-derived hepatocytes together with an empirical scaling factor warrants additional study with a larger set of compounds, as it has the potential to provide more accurate and precise estimates of hepatic clearance.

There is Only One Valid Definition of Clearance: Critical Examination of Clearance Concepts Reveals the Potential for Errors in Clinical Drug Dosing Decisions

Drug dosing decisions in clinical medicine and in introducing a drug to market for the past 60 years are based on the pharmacokinetic/clinical pharmacology concept of clearance. We used chemical reaction engineering models to demonstrate the limitations of presently employed clearance measurements based upon systemic blood concentration in reflecting organ clearance. The belief for the last 49 years that in vivo clearance is independent of the mechanistic model for organ clearance is incorrect. There is only one valid definition of clearance. Defining organ clearance solely on the basis of systemic blood concentrations can lead to drug dosing errors when drug effect sites reside either in an eliminating organ exhibiting incremental clearance or in a non-eliminating organ where intraorgan concentration is governed by transporter actions. Attempts to predict clearance are presently hampered by the lack of recognition that what we are trying to predict is a well-stirred model clearance.

Prediction of a clinically effective dose of THY1773, a novel V1B receptor antagonist, based on preclinical data

THY1773 is a novel arginine vasopressin 1B (V1B ) receptor antagonist that is under development as an oral drug for the treatment of major depressive disorder (MDD). Here we report our strategy to predict a clinically effective dose of THY1773 for MDD in the preclinical stage, and discuss the important insights gained by retrospective analysis of prediction accuracy. To predict human pharmacokinetic (PK) parameters, several extrapolation methods from animal or in vitro data to humans were investigated. The fu correction intercept method and two-species-based allometry were used to extrapolate clearance from rats and dogs to humans. The physiologically based pharmacokinetics (PBPK)/receptor occupancy (RO) model was developed by linking free plasma concentration with pituitary V1B RO by the Emax model. As a result, the predicted clinically effective dose of THY1773 associated with 50% V1B RO was low enough (10 mg/day, or at maximum 110 mg/day) to warrant entering phase 1 clinical trials. In the phase 1 single ascending dose study, TS-121 capsule (active ingredient: THY1773) showed favorable PKs for THY1773 as expected, and in the separately conducted phase 1 RO study using positron emission tomography, the observed pituitary V1B RO was comparable to our prediction. Retrospective analysis of the prediction accuracy suggested that the prediction methods considering plasma protein binding, and avoiding having to apply unknown scaling factors obtained in animals to humans, would lead to better prediction. Selecting mechanism-based methods with reasonable assumptions would be critical for the successful prediction of a clinically effective dose in the preclinical stage of drug development.

Successful and Unsuccessful Prediction of Human Hepatic Clearance for Lead Optimization

Development of new chemical entities is costly, time-consuming, and has a low success rate. Accurate prediction of pharmacokinetic properties is critical to progress compounds with favorable drug-like characteristics in lead optimization. Of particular importance is the prediction of hepatic clearance, which determines drug exposure and contributes to projection of dose, half-life, and bioavailability. The most commonly employed methodology to predict hepatic clearance is termed in vitro to in vivo extrapolation (IVIVE) that involves measuring drug metabolism in vitro, scaling-up this in vitro intrinsic clearance to a prediction of in vivo intrinsic clearance by reconciling the enzymatic content between the incubation and an average human liver, and applying a model of hepatic disposition to account for limitations of protein binding and blood flow to predict in vivo clearance. This manuscript reviews common in vitro techniques used to predict hepatic clearance as well as current challenges and recent theoretical advancements in IVIVE.

Usage of In Vitro Metabolism Data for Drug-Drug Interaction in Physiologically Based Pharmacokinetic Analysis Submissions to the US Food and Drug Administration

The key parameters necessary to predict drug-drug interactions (DDIs) are intrinsic clearance (CL_int ) and fractional contribution of the metabolizing enzyme toward total metabolism (f_m ). Herein, we summarize the accumulated knowledge from 53 approved new drug applications submitted to the Office of Clinical Pharmacology, US Food and Drug Administration, from 2016 to 2018 that contained physiologically based pharmacokinetic (PBPK) models to understand how in vitro data are used in PBPK models to assess drug metabolism and predict DDIs. For evaluation of CL_int and f_m , 29 and 20 new drug applications were included for evaluation, respectively. For CL_int , 86.2% of the PBPK models used modified values based on in vivo data with modifications ranging from -82.5% to 2752.5%. For f_m , 45.0% of the models used modified values with modifications ranging from -28% to 178.6%. When values for CL_int were used from in vitro testing without modification, the model resulted in up to a 14.3-fold overprediction of the area under the concentration-time curve of the substrate. When values for f_m from in vitro testing were used directly, the model resulted in up to a 2.9-fold underprediction of its DDI magnitude with an inducer, and up to a 1.7-fold overprediction of its DDI magnitude with an inhibitor. Our analyses suggested that the in vitro system usually provides a reasonable estimation of f_m when the drug metabolism by a given CYP pathway is more than 70% of the total clearance. In vitro experiments provide important information about basic PK properties of new drugs and can serve as a starting point for building a PBPK model. However, key PBPK parameters such as CL_int and f_m still need to be optimized based on in vivo data.

Prediction of Drug Clearance from Enzyme and Transporter Kinetics

Accurate estimation of in vivo clearance in human is pivotal to determine the dose and dosing regimen for drug development. In vitro-in vivo extrapolation (IVIVE) has been performed to predict drug clearance using empirical and physiological scalars. Multiple in vitro systems and mathematical modeling techniques have been employed to estimate in vivo clearance. The models for predicting clearance have significantly improved and have evolved to become more complex by integrating multiple processes such as drug metabolism and transport as well as passive diffusion. This chapter covers the use of conventional as well as recently developed methods to predict metabolic and transporter-mediated clearance along with the advantages and disadvantages of using these methods and the associated experimental considerations. The general approaches to improve IVIVE by use of appropriate scalars, incorporation of extrahepatic metabolism and transport and application of physiologically based pharmacokinetic (PBPK) models with proteomics data are also discussed. The chapter also provides an overview of the advantages of using such dynamic mechanistic models over static models for clearance predictions to improve IVIVE.

Comparative Assessment of Extrapolation Methods Based on the Conventional Free Drug Hypothesis and Plasma Protein-Mediated Hepatic Uptake Theory for the Hepatic Clearance Predictions of Two Drugs Extensively Bound to Both the Albumin And Alpha-1-Acid Glycoprotein

Bteich and coworkers recently demonstrated in a companion manuscript (J Pharm Sci 109: https://doi.org/10.1016/j.xphs.2020.07.003) that a protein-mediated hepatic uptake have occurred in an isolated perfused rat liver (IPRL) model for two drugs (Perampanel; PER and Fluoxetine; FLU) that bind extensively to the albumin (ALB) and alpha-1-acid glycoprotein (AGP). However, to our knowledge, there is no quantitative model available to predict the impact of a plasma protein-mediated hepatic uptake on the extent of hepatic clearance (CLh) for a drug binding extensively to these two proteins. Therefore, the main objective was to predict the corresponding CLh, which is an extension of the companion manuscript. The method consisted of extrapolating the intrinsic clearance from the unbound fraction measured in the perfusate or the unbound fraction extrapolated to the surface of the hepatocyte membrane by adapting an existing model of protein-mediated hepatic uptake (i.e., the fu_p-adjusted model) to include a binding ratio between the ALB and AGP. This new approach showed a relevant improvement compared to the free drug hypothesis particularly for FLU that showed the highest degree of ALB-mediated uptake. Overall, this study is a first step towards the development of predictive methods of CLh by considering the binding to ALB and AGP.

In Vitro Hepatic Uptake in Human and Monkey Hepatocytes in the Presence and Absence of Serum Protein and Its In Vitro to In Vivo Extrapolation

It is well documented that human hepatic clearance based on in vitro metabolism or transporter assays systematically resulted in underprediction; therefore, large empirical scalars are often needed in either static or physiologically based pharmacokinetic (PBPK) models to accurately predict human pharmacokinetics (PK). In our current investigation, we assessed hepatic uptake in hepatocyte suspension in Krebs-Henseleit buffer in the presence and absence of serum. The results showed that the unbound intrinsic active clearance (CL_u,int,active) values obtained by normalizing the unbound fraction in the buffer containing 10% serum were generally higher than the CL_u,int,active obtained directly from protein free buffer, suggesting "protein-facilitated" uptake. The differences of CL_u,int,active in the buffer with and without protein ranged from 1- to 925-fold and negatively correlated to the unbound serum binding of organic anion transporting polypeptide substrates. When using the uptake values obtained from buffer containing serum versus serum-free buffer, the median of scaling factors (SFs) for CL_u,int,active reduced from 24.2-4.6 to 22.7-7.1 for human and monkey, respectively, demonstrating the improvement of in vitro to in vivo extrapolation in a PBPK model. Furthermore, values of CL_u,int,active were significantly higher in monkey hepatocytes than that in human, and the species differences appeared to be compound dependent. Scaling up in vitro uptake values derived in assays containing species-specific serum can compensate for the species-specific variabilities when using cynomolgus monkey as a probe animal model. Incorporating SFs calibrated in monkey and together with scaled in vitro data can be a reliable approach for the prospective human PK prediction in early drug discovery. SIGNIFICANCE STATEMENT: We investigated the protein effect on hepatic uptake in human and monkey hepatocytes and improved the in vitro to in vivo extrapolation using parameters obtained from the incubation in the present of serum protein. In addition, significantly higher active uptake clearances were observed in monkey hepatocytes than in human, and the species differences appeared to be compound dependent. The physiologically based pharmacokinetic model that incorporates scaling factors calibrated in monkey and together with scaled in vitro human data can be a reliable approach for the prospective human pharmacokinetics prediction.

Investigating the Theoretical Basis for In Vitro-In Vivo Extrapolation (IVIVE) in Predicting Drug Metabolic Clearance and Proposing Future Experimental Pathways

Extensive studies have been conducted to predict in vivo metabolic clearance from in vitro human liver metabolism parameters (i.e., in vitro-in vivo extrapolation (IVIVE)) with little success. Here, deriving IVIVE from first principles, we show that the product of fraction unbound in the blood and the predicted in vivo intrinsic clearance determined from hepatocyte or microsomal incubations is the lower boundary condition for in vivo hepatic clearance and the prerequisite for IVIVE predictions to be valid, regardless of extraction ratio. For 60-80% of drugs evaluated here, this product is markedly less than the in vivo measured clearance, a result that violates the lower boundary of the predictive relationship. This can only be explained by (a) suboptimal in vitro metabolic stability assay conditions, (b) significant error in the assumption that in vitro intrinsic clearance determinations will predict in vivo intrinsic clearance simply by scaling-up the amount of enzyme (in vitro incubation to in vivo liver), and/or (c) the methods of determining fraction unbound are incorrect. We further suggest that widely employed organ blood flow values underpredict the effective blood flow within the organ by approximately 2.5-fold, thus impacting IVIVE of high clearance compounds. We propose future pathways that should be investigated in terms of the relationship to experimentally measured clearance values, rather than model-dependent intrinsic clearance. IVIVE outcome can be improved by estimating the ratio of unbound drug concentration in the liver tissue to the liver plasma, examining the assumption of the free drug theory (i.e., there are no transporter effects at the blood cell membrane) and the finding that the upper limit of organ clearance may be greater than blood flow entering the organ.

Are There Any Experimental Perfusion Data that Preferentially Support the Dispersion and Parallel-Tube Models over the Well-Stirred Model of Organ Elimination?

In reviewing previously published isolated perfused rat liver studies, we find no experimental data for high-clearance metabolized drugs that reasonably or unambiguously support preference for the dispersion and parallel-tube models versus the well-stirred model of organ elimination when only entering and exiting drug concentrations are available. It is likely that the investigators cited here may have been influenced by: 1) the unphysiologic aspects of the well-stirred model, which may have led them to undervalue the studies that directly test the various hepatic disposition models for high-clearance drugs (for which model differences are the greatest); 2) experimental assumptions made in the last century, which are no longer valid today, related to the predictability of in vivo outcomes from in vitro measures of drug elimination and the influence of albumin in hepatic drug uptake; and 3) a lack of critical review of previously reported experimental studies, resulting in inappropriate interpretation of the available experimental data. The number of papers investigating the theoretical aspects of the dispersion, parallel-tube, and well-stirred models of hepatic elimination greatly outnumber the papers that actually examine the experimental evidence available to substantiate these models. When all experimental studies that measure organ elimination using entering and exiting drug concentrations at steady state are critically reviewed, the simple but unphysiologic well-stirred model is the only model that can describe all trustworthy published available data. SIGNIFICANCE STATEMENT: Although the dispersion model of hepatic elimination more adequately reflects physiologic reality, there are no convincing experimental data that unambiguously favor this model. The well-stirred model can describe all well-designed perfusion studies with high-clearance drugs and nondrug substrates, but the field has not recognized this because of hesitation to accept a nonphysiologic model and flawed attempts to utilize in vitro-in vivo extrapolation approaches.

In vitro-in vivo extrapolation of metabolic clearance using human liver microsomes: factors showing variability and their normalization

In vitro-in vivo extrapolation (IVIVE) using human liver microsomes has been widely used to predict metabolic clearance, but some of the factors used in the process of prediction show variability for the same compound: notably, microsomal intrinsic clearance values corrected by the unbound fraction (CL_{int, u}), physiological parameters used for scale-up, and the source of in vivo clearance data.The purpose of this study was to assess the correlation between in vitro and in vivo CL_int with a focus on factors showing variability using four cytochrome P450 (CYP)3A substrates.We surveyed in vivo clearance values in literature and also determined the microsomal CL_{int, u} values. A scaling factor (SF_direct) was defined as in vivo CL_int divided by the microsomal CL_{int, u}, which ranged from 1190 to 2310 (mg protein per kg body weight). The application of a mean SF_direct of 1600 (mg protein per kg body weight) and further normalization by the microsomal CL_{int, u} values of midazolam, the most commonly used substrate, resulted in improved prediction accuracy for CL_{int, u} values from various microsomal batches.The results suggest the normalization of variability might be useful for predicting the in vivo CL_int.

In Vitro-In Vivo Inaccuracy: The CYP3A4 Anomaly

When predicting hepatic clearance using in vitro to in vivo extrapolation (IVIVE), microsomes or hepatocytes are commonly used. Here, we examine intrinsic clearance values and IVIVE results in human hepatocytes and microsomes for compounds metabolized by a variety of enzymes. The great majority of CYP3A4 substrates examined had higher intrinsic clearance values in microsomes compared with hepatocytes, whereas the values were more similar between the two incubations for substrates of other enzymes. We hypothesize that this may be due to interplay between CYP3A4 and the efflux transporter P-glycoprotein, as they have been shown to exhibit coordinated regulation. When examining the prediction accuracy for substrates of other enzymes between microsomes and hepatocytes, average fold errors as well as overall error were similar, demonstrating once again that IVIVE methods are not adequately defined and understood. SIGNIFICANCE STATEMENT: For CYP3A4 substrates, microsomes give markedly higher predictive in vitro to in vivo extrapolation than for other metabolic enzymes, which is not found for hepatocytes. We hypothesize that this could be a result of CYP3A4-P-glycoprotein interplay or coordinated regulation in hepatocytes that would not be observed in microsomes.