Improving the Accuracy of Predicted Human Pharmacokinetics: Lessons Learned from the AstraZeneca Drug Pipeline Over Two Decades

It is now universally recognized that an acceptable human PK profile increases the probability of a candidate drug becoming a successful therapy.A variety of tools are available to predict human PK behavior in advance of clinical data, including scaling clearance from in vitro metabolic stability data, and physiologically based scaling of volume of distribution.To improve PK prediction methods, it is crucial to continually assess the performance of predictions with first-time-in-human PK data for new candidate drugs.To date, AstraZeneca have compared observed PK to predictions for 116 candidate drugs, and since the launch of our five-dimensional framework in 2011, we have driven sustained improvements in the quality of PK predictions. During drug discovery and prior to the first human dose of a novel candidate drug, the pharmacokinetic (PK) behavior of the drug in humans is predicted from preclinical data. This helps to inform the likelihood of achieving therapeutic exposures in early clinical development. Once clinical data are available, the observed human PK are compared with predictions, providing an opportunity to assess and refine prediction methods. Application of best practice in experimental data generation and predictive methodologies, and a focus on robust mechanistic understanding of the candidate drug disposition properties before nomination to clinical development, have led to maximizing the probability of successful PK predictions so that 83% of AstraZeneca drug development projects progress in the clinic with no PK issues; and 71% of key PK parameter predictions [64% of area under the curve (AUC) predictions; 78% of maximum concentration (Cmax) predictions; and 70% of half-life predictions] are accurate to within twofold. Here, we discuss methods to predict human PK used by AstraZeneca, how these predictions are assessed and what can be learned from evaluating the predictions for 116 candidate drugs. During drug discovery and prior to the first human dose of a novel candidate drug, the pharmacokinetic (PK) behavior of the drug in humans is predicted from preclinical data. This helps to inform the likelihood of achieving therapeutic exposures in early clinical development. Once clinical data are available, the observed human PK are compared with predictions, providing an opportunity to assess and refine prediction methods. Application of best practice in experimental data generation and predictive methodologies, and a focus on robust mechanistic understanding of the candidate drug disposition properties before nomination to clinical development, have led to maximizing the probability of successful PK predictions so that 83% of AstraZeneca drug development projects progress in the clinic with no PK issues; and 71% of key PK parameter predictions [64% of area under the curve (AUC) predictions; 78% of maximum concentration (Cmax) predictions; and 70% of half-life predictions] are accurate to within twofold. Here, we discuss methods to predict human PK used by AstraZeneca, how these predictions are assessed and what can be learned from evaluating the predictions for 116 candidate drugs. In response to the recognition that a significant cause of attrition in clinical development of small molecules was due to a lack of appropriate pharmacokinetic (PK) (see Glossary) properties [1.Kola I. The state of innovation in drug development.Clin. Pharmacol. Ther. 2008; 83: 227-230sCrossref PubMed Scopus (217) Google Scholar], considerable investments were made across the pharmaceutical industry to build the scientific understanding and experimental assays to characterize and predict human PK. This produced a wealth of publications demonstrating scaling methodology, from the use of human liver microsomes and hepatocytes for predicting human hepatic clearance (CL) [2.Grime K. Riley R.J. The impact of in vitro binding on in vitro-in vivo extrapolations, projections of metabolic clearance and clinical drug-drug interactions.Curr. Drug Metab. 2006; 7: 251-264Crossref PubMed Scopus (114) Google Scholar, 3.Riley R.J. et al.A unified model for predicting human hepatic, metabolic clearance from in vitro intrinsic clearance data in hepatocytes and microsomes.Drug Metab. Dispos. 2005; 33: 1304-1311Crossref PubMed Scopus (301) Google Scholar, 4.Ring B.J. et al.PhRMA CPCDC initiative on predictive models of human pharmacokinetics, part 3: comparative assessement of prediction methods of human clearance.J. Pharm. Sci. 2011; 100: 4090-4110Abstract Full Text Full Text PDF PubMed Scopus (153) Google Scholar], to the application of animal in vivo data and allometric scaling to predict volume of distribution, CL, and bioavailability [5.Poulin P. et al.PhRMA CPCDC initiative on predictive models of human pharmacokinetics, part 1: goals, properties of the PhRMA dataset, and comparison with literature datasets.J. Pharm. Sci. 2011; 100: 4050-4073Abstract Full Text Full Text PDF PubMed Scopus (44) Google Scholar, 6.Grime K.H. et al.Application of in silico, in vitro and preclinical pharmacokinetic data for the effective and efficient prediction of human pharmacokinetics.Mol. Pharm. 2013; 10: 1191-1206Crossref PubMed Scopus (61) Google Scholar, 7.Huang Q. Riviere J.E. The application of allometric scaling principles to predict pharmacokinetic parameters across species.Expert Opin. Drug Metab. Toxicol. 2014; 10: 1241-1253Crossref PubMed Scopus (71) Google Scholar]. Physiologically based pharmacokinetic (PBPK) modelling also emerged as a mathematical framework describing the major organs and tissues of the mammalian body, connected by the arterial and venous pathways, developed with the intention of: (i) being used as the basis for integrating and simulating the time course of a compound into, through, and out of the body [i.e., absorption, distribution, metabolism, and excretion (ADME)]; and (ii) offering a more mechanistically realistic way of scaling PK properties to humans by accounting for species differences in physiology and ADME properties [8.Poulin P. et al.PHRMA CPCDC initiative on predictive models of human pharmacokinetics, part 5: prediction of plasma concentration-time profiles in human by using the physiologically-based pharmacokinetic modeling approach.J. Pharm. Sci. 2011; 100: 4127-4157Abstract Full Text Full Text PDF PubMed Scopus (147) Google Scholar,9.Jamei M. et al.The Simcyp population-based ADME simulator.Expert Opin. Drug Metab. Toxicol. 2009; 5: 211-223Crossref PubMed Scopus (403) Google Scholar]. Optimization strategies focused on achieving target potency, selectivity, and an acceptable predicted human PK profile maximize the probability of successful drug discovery. As a compound's key PK properties, such as oral absorption, distribution, susceptibility to being a substrate or inhibitor for drug-metabolizing enzymes (DMEs), and passive and active elimination processes are commonly associated with inherent physicochemical properties, such as molecular size, hydrophobicity, aqueous solubility, and ionization state at physiological pH, understanding these relationships is central to a successful outcome [6.Grime K.H. et al.Application of in silico, in vitro and preclinical pharmacokinetic data for the effective and efficient prediction of human pharmacokinetics.Mol. Pharm. 2013; 10: 1191-1206Crossref PubMed Scopus (61) Google Scholar]. By the mid-2000s, there was an extensive number of methods available [4.Ring B.J. et al.PhRMA CPCDC initiative on predictive models of human pharmacokinetics, part 3: comparative assessement of prediction methods of human clearance.J. Pharm. Sci. 2011; 100: 4090-4110Abstract Full Text Full Text PDF PubMed Scopus (153) Google Scholar,5.Poulin P. et al.PhRMA CPCDC initiative on predictive models of human pharmacokinetics, part 1: goals, properties of the PhRMA dataset, and comparison with literature datasets.J. Pharm. Sci. 2011; 100: 4050-4073Abstract Full Text Full Text PDF PubMed Scopus (44) Google Scholar,8.Poulin P. et al.PHRMA CPCDC initiative on predictive models of human pharmacokinetics, part 5: prediction of plasma concentration-time profiles in human by using the physiologically-based pharmacokinetic modeling approach.J. Pharm. Sci. 2011; 100: 4127-4157Abstract Full Text Full Text PDF PubMed Scopus (147) Google Scholar,10.Jones R.D. et al.PhRMA CPCDC initiative on predictive models of human pharmacokinetics, part 2: comparative assessment of prediction methods of human volume of distribution.J. Pharm. Sci. 2011; 100: 4074-4089Abstract Full Text Full Text PDF PubMed Scopus (96) Google Scholar,11.Vuppugalla R. et al.PhRMA CPCDC initiative on predictive models of human pharmacokinetics, part 4: prediction of plasma concentration-time profiles in human from in vivo preclinical data by using the Wajima approach.J. Pharm. Sci. 2011; 100: 4111-4126Abstract Full Text Full Text PDF PubMed Scopus (44) Google Scholar], but little consensus on how these approaches should define best practice. This led to a cross-industry, precompetitive initiative led by the Pharmaceutical Research and Manufacturers of America (PhRMA)i with the aims of (i) establishing a blinded compound dataset of preclinical and clinical data provided by industry members; and (ii) compiling a list of available scaling methods and testing their accuracy against the dataset to assess the merits of certain data types and scaling methods for predicting human PK. This work was published across five manuscripts [4.Ring B.J. et al.PhRMA CPCDC initiative on predictive models of human pharmacokinetics, part 3: comparative assessement of prediction methods of human clearance.J. Pharm. Sci. 2011; 100: 4090-4110Abstract Full Text Full Text PDF PubMed Scopus (153) Google Scholar,5.Poulin P. et al.PhRMA CPCDC initiative on predictive models of human pharmacokinetics, part 1: goals, properties of the PhRMA dataset, and comparison with literature datasets.J. Pharm. Sci. 2011; 100: 4050-4073Abstract Full Text Full Text PDF PubMed Scopus (44) Google Scholar,8.Poulin P. et al.PHRMA CPCDC initiative on predictive models of human pharmacokinetics, part 5: prediction of plasma concentration-time profiles in human by using the physiologically-based pharmacokinetic modeling approach.J. Pharm. Sci. 2011; 100: 4127-4157Abstract Full Text Full Text PDF PubMed Scopus (147) Google Scholar,10.Jones R.D. et al.PhRMA CPCDC initiative on predictive models of human pharmacokinetics, part 2: comparative assessment of prediction methods of human volume of distribution.J. Pharm. Sci. 2011; 100: 4074-4089Abstract Full Text Full Text PDF PubMed Scopus (96) Google Scholar,11.Vuppugalla R. et al.PhRMA CPCDC initiative on predictive models of human pharmacokinetics, part 4: prediction of plasma concentration-time profiles in human from in vivo preclinical data by using the Wajima approach.J. Pharm. Sci. 2011; 100: 4111-4126Abstract Full Text Full Text PDF PubMed Scopus (44) Google Scholar] and provides the most comprehensive assessment to date. The preferred methods that have been used for characterizing, optimizing, and predicting the fundamental human ADME parameters and have laid the foundation of AstraZeneca's PK strategy for small molecules are briefly outlined in the following sections (Figure 1A , Key Figure). There is broad agreement between pharmaceutical companies that the assays described below represent best practice experiments to predict human PK. However, where more than one assay provides information on one of the ADME processes of a candidate drug, they can suggest conflicting pictures of what the properties could be in humans; for example, high or low bioavailability or CL. In this case, it is important not to rely on a single preferred assay or average of results; the AstraZeneca approach is to consider the weight of all the evidence together, along with an understanding of the limitations of each assay, to understand the most likely scenario for the human ADME behavior. AstraZeneca has also developed a method for evaluating the accuracy of human PK predictions against clinical data for candidate drugs, discussed in detail later, which provides learning to refine our interpretation of ADME results in predicting human PK (Figure 1E). The PK prediction assessments also inform the Right Tissue/Exposure component of the AstraZeneca five-dimensional framework, which describes the profiles of risks and opportunities of drug development projects across aspects of Right Target, Right Patients, Right Safety, and Right Commercial Potential alongside the Right Tissue/Exposure [12.Cook D. et al.Lessons learned from the fate of AstraZeneca's drug pipeline: a five-dimensional framework.Nat. Rev. Drug Discov. 2014; 13: 419-431Crossref PubMed Scopus (857) Google Scholar,13.Morgan P. et al.Impact of a five-dimensional framework on R&D productivity at AstraZeneca.Nat. Rev. Drug Discov. 2018; 17: 167-181Crossref PubMed Scopus (211) Google Scholar] (Figure 1F). These five dimensions are underpinned by the Right Culture of scientific rigor, and our philosophy of predicting human PK based on mechanistic understanding demonstrates how this culture is embedded in the Right Tissue/Exposure space. Bioavailability (F) (Figure 1B) is the product of the fraction of intestinally absorbed drug (Fa), the fraction escaping intestinal metabolism (Fg), and the fraction escaping liver extraction (Fh). Aiming for a fraction absorbed (Fa) in human of >50% is appropriate, in most instances during the drug screening/selection process, to obtain sufficient bioavailability and ensure that robust systemic exposure to a compound is achieved following oral dosing. Suitable oral drug-like physicochemical properties that deliver both sufficient passive transcellular permeability and aqueous solubility of the crystalline form can lead to confident human absorption estimates of >50%, which can be further strengthened by good in vivo absorption, using appropriate formulations, in preclinical species [6.Grime K.H. et al.Application of in silico, in vitro and preclinical pharmacokinetic data for the effective and efficient prediction of human pharmacokinetics.Mol. Pharm. 2013; 10: 1191-1206Crossref PubMed Scopus (61) Google Scholar,14.McGinnity D.F. et al.Evaluation of human pharmacokinetics, therapeutic dose and exposure predictions using marketed oral drugs.Curr. Drug Metab. 2007; 8: 463-479Crossref PubMed Scopus (95) Google Scholar]. It is necessary to not just determine F in preclinical species, but to deconvolute the contributions of intestinal permeability and efflux and first-pass intestinal and hepatic metabolism. Moreover, F, or CL, values in animals are not, per se, quality measures to predict F or CL for humans; however, low bioavailability in preclinical species may create obstacles in obtaining appropriate exposures in pharmacodynamic (PD), efficacy, and safety experiments for a candidate drug. To inform on the potential for oral absorption, the apparent transcellular permeability (Papp) measure is obtained from an assessment of drug flux across a monolayer of cells with characteristics of the intestinal barrier, typically Madin–Darby canine kidney (MDCK) or colorectal adenocarcinoma-2 (Caco-2) cells [15.Yang J. et al.Prediction of intestinal first-pass drug metabolism.Curr. Drug Metab. 2007; 8: 676-684Crossref PubMed Scopus (298) Google Scholar,16.Hidalgo I.J. et al.Characterization of the human colon carcinoma cell line (Caco-2) as a model system for intestinal epithelial permeability.Gastroenterology. 1989; 96: 736-749Abstract Full Text PDF PubMed Scopus (1952) Google Scholar]. The absorption of drugs with good permeability and solubility is unlikely to be limited by efflux via P-glycoprotein (an efflux transporter present in the gut and liver) or other intestinal drug transporters unless the dose is low [5.Poulin P. et al.PhRMA CPCDC initiative on predictive models of human pharmacokinetics, part 1: goals, properties of the PhRMA dataset, and comparison with literature datasets.J. Pharm. Sci. 2011; 100: 4050-4073Abstract Full Text Full Text PDF PubMed Scopus (44) Google Scholar,17.Murakami T. Takano M. Intestinal efflux transporters and drug absorption.Expert Opin. Drug Metab. Toxicol. 2008; 4: 923-939Crossref PubMed Scopus (167) Google Scholar]. This should not necessarily be assumed if the maximum flux (Jm) for the transporter is high and/or the solubility or permeability is marginal. For compounds that have suboptimal properties, dynamic simulations using either proprietary (GI Sim) or commercial [Simcyp (Certara)ii, Gastroplus (Simulations Plus)iii] software are useful for hypothesis generation and decision making [9.Jamei M. et al.The Simcyp population-based ADME simulator.Expert Opin. Drug Metab. Toxicol. 2009; 5: 211-223Crossref PubMed Scopus (403) Google Scholar,18.Bolger M.B. et al.Simulations of the nonlinear dose dependence for substrates of influx and efflux transporters in the human intestine.AAPS J. 2009; 11: 353-363Crossref PubMed Scopus (76) Google Scholar, 19.Sinha V.K. et al.From preclinical to human--prediction of oral absorption and drug-drug interaction potential using physiologically based pharmacokinetic (PBPK) modeling approach in an industrial setting: a workflow by using case example.Biopharm. Drug Dispos. 2012; 33: 111-121Crossref PubMed Scopus (59) Google Scholar, 20.Sjogren E. et al.In silico modeling of gastrointestinal drug absorption: predictive performance of three physiologically based absorption models.Mol. Pharm. 2016; 13: 1763-1778Crossref PubMed Scopus (55) Google Scholar]. Approaches to predict Fg are being developed [21.Jones C.R. et al.Gut wall metabolism. Application of pre-clinical models for the prediction of human drug absorption and first-pass elimination.AAPS J. 2016; 18: 589-604Crossref PubMed Scopus (42) Google Scholar]. The major human intestinal DMEs are cytochrome P450 (CYP)3A, CYP2C9 and uridine 5′-diphospho-glucuronosyltransferase (UGT) [22.Komura H. Iwaki M. In vitro and in vivo small intestinal metabolism of CYP3A and UGT substrates in preclinical animals species and humans: species differences.Drug Metab. Rev. 2011; 43: 476-498Crossref PubMed Scopus (60) Google Scholar, 23.Galetin A. et al.Contribution of intestinal cytochrome p450-mediated metabolism to drug-drug inhibition and induction interactions.Drug Metab. Pharmacokinet. 2010; 25: 28-47Crossref PubMed Scopus (97) Google Scholar, 24.Thelen K. Dressman J.B. Cytochrome P450-mediated metabolism in the human gut wall.J. Pharm. Pharmacol. 2009; 61: 541-558Crossref PubMed Google Scholar, 25.Paine M.F. et al.The human intestinal cytochrome P450 "pie".Drug Metab. Dispos. 2006; 34: 880-886Crossref PubMed Scopus (698) Google Scholar, 26.Zhang Q.Y. et al.Characterization of human small intestinal cytochromes P-450.Drug Metab. Dispos. 1999; 27: 804-809PubMed Google Scholar, 27.Prueksaritanont T. et al.Comparative studies of drug-metabolizing enzymes in dog, monkey, and human small intestines, and in Caco-2 cells.Drug Metab. Dispos. 1996; 24: 634-642PubMed Google Scholar]. Intestinal extraction is typically associated with metabolically unstable drugs, yet the enterocyte location of CYP3A4 and P-glycoprotein, which share substrate specificity, facilitates cycling of drugs, prolonging exposure and increasing potential for intestinal metabolism [15.Yang J. et al.Prediction of intestinal first-pass drug metabolism.Curr. Drug Metab. 2007; 8: 676-684Crossref PubMed Scopus (298) Google Scholar,28.Doherty M.M. Charman W.N. The mucosa of the small intestine: how clinically relevant as an organ of drug metabolism?.Clin. Pharmacokinet. 2002; 41: 235-253Crossref PubMed Scopus (197) Google Scholar,29.Zhang Y. Benet L.Z. The gut as a barrier to drug absorption: combined role of cytochrome P450 3A and P-glycoprotein.Clin. Pharmacokinet. 2001; 40: 159-168Crossref PubMed Scopus (492) Google Scholar]. Therefore, optimization of hepatic CL, efflux, and permeability properties will reduce factors contributing to intestinal metabolism and mitigate against intestinal metabolism being a significant issue [29.Zhang Y. Benet L.Z. The gut as a barrier to drug absorption: combined role of cytochrome P450 3A and P-glycoprotein.Clin. Pharmacokinet. 2001; 40: 159-168Crossref PubMed Scopus (492) Google Scholar]. Volume of distribution (Vd) (Figure 1B) is a measure of the relative affinity of a drug for tissues and plasma. Generally, physicochemical properties rather than specific pharmacophores regulate Vd values: acidic drug-like molecules have low (<1 l/kg); neutrals and zwitterions moderate (1–3 l/kg); and basic drugs high (>3 l/kg) Vd values [6.Grime K.H. et al.Application of in silico, in vitro and preclinical pharmacokinetic data for the effective and efficient prediction of human pharmacokinetics.Mol. Pharm. 2013; 10: 1191-1206Crossref PubMed Scopus (61) Google Scholar,14.McGinnity D.F. et al.Evaluation of human pharmacokinetics, therapeutic dose and exposure predictions using marketed oral drugs.Curr. Drug Metab. 2007; 8: 463-479Crossref PubMed Scopus (95) Google Scholar,30.Lombardo F. et al.Trend analysis of a database of intravenous pharmacokinetic parameters in humans for 1352 drug compounds.Drug Metab. Dispos. 2018; 46: 1466-1477Crossref PubMed Scopus (64) Google Scholar,31.Rodgers T. Rowland M. Mechanistic approaches to volume of distribution predictions: understanding the processes.Pharm. Res. 2007; 24: 918-933Crossref PubMed Scopus (303) Google Scholar]. The dominant influence on the distribution of acidic drugs is the extensive binding to plasma albumin, and they are thus associated with similar distribution volumes to plasma (0.1–0.3 l/kg) unless active transport processes contribute to increase tissue concentrations; for example, via organic anion transporter polypeptide (OATP) [32.Rodgers T. Rowland M. Physiologically based pharmacokinetic modelling 2: predicting the tissue distribution of acids, very weak bases, neutrals and zwitterions.J. Pharm. Sci. 2006; 95: 1238-1257Abstract Full Text Full Text PDF PubMed Scopus (643) Google Scholar]. Bases, being positively charged at physiological pH, interact with anionic phospholipid head groups and subcellular acidic organelles; for example, lysosomes, leading to higher tissue affinity and distribution [6.Grime K.H. et al.Application of in silico, in vitro and preclinical pharmacokinetic data for the effective and efficient prediction of human pharmacokinetics.Mol. Pharm. 2013; 10: 1191-1206Crossref PubMed Scopus (61) Google Scholar,33.Rodgers T. et al.Physiologically based pharmacokinetic modeling 1: predicting the tissue distribution of moderate-to-strong bases.J. Pharm. Sci. 2005; 94: 1259-1276Abstract Full Text Full Text PDF PubMed Scopus (561) Google Scholar]. Given the dependency on physicochemical properties, Vd is not a parameter that is readily adaptable in a given chemical series; rather, boundary estimates of Vd, and specifying a desired human elimination half-life (t1/2) can help identify a target CL value to optimize against. Vd is reasonably predictable from physicochemical properties, in silico [34.Berellini G. et al.In silico prediction of volume of distribution in human using linear and nonlinear models on a 669 compound data set.J. Med. Chem. 2009; 52: 4488-4495Crossref PubMed Scopus (64) Google Scholar,35.Gleeson M.P. et al.In silico human and rat Vss quantitative structure-activity relationship models.J. Med. Chem. 2006; 49: 1953-1963Crossref PubMed Scopus (71) Google Scholar] and in vitro methods [36.Berezhkovskiy L.M. On the accuracy of determination of unbound drug fraction in tissue using diluted tissue homogenate.J. Pharm. Sci. 2012; 101: 1909-1916Abstract Full Text Full Text PDF PubMed Scopus (7) Google Scholar,37.Berry L.M. et al.Species differences in distribution and prediction of human V(ss) from preclinical data.Drug Metab. Dispos. 2011; 39: 2103-2116Crossref PubMed Scopus (49) Google Scholar]. Using preclinical in vivo PK data and correcting for species plasma protein binding (PPB) differences, Vd is the most predictable of PK parameters; if unbound volume of distribution (Vu) is consistent across species (it is typically conserved within twofold), human Vd can be predicted from animal data by estimating Vu [6.Grime K.H. et al.Application of in silico, in vitro and preclinical pharmacokinetic data for the effective and efficient prediction of human pharmacokinetics.Mol. Pharm. 2013; 10: 1191-1206Crossref PubMed Scopus (61) Google Scholar,14.McGinnity D.F. et al.Evaluation of human pharmacokinetics, therapeutic dose and exposure predictions using marketed oral drugs.Curr. Drug Metab. 2007; 8: 463-479Crossref PubMed Scopus (95) Google Scholar,31.Rodgers T. Rowland M. Mechanistic approaches to volume of distribution predictions: understanding the processes.Pharm. Res. 2007; 24: 918-933Crossref PubMed Scopus (303) Google Scholar,38.Waters N.J. Lombardo F. Use of the Oie-Tozer model in understanding mechanisms and determinants of drug distribution.Drug Metab. Dispos. 2010; 38: 1159-1165Crossref PubMed Scopus (25) Google Scholar]. As PK from multiple species is determined in drug discovery, this is the preferred predictive methodology in AstraZeneca for candidate drugs. The major drug CL (Figure 1B) routes in humans and preclinical species are hepatic metabolism, and renal and biliary elimination, and methods are required to predict these parallel elimination processes in humans, using a combination of animal data and in vitro human tools. Identification of the elimination route and rate in preclinical species and optimization of human CL (most commonly hepatic) are among the foremost challenges of many drug discovery projects. The liver is the major drug-metabolizing organ, and in vitro–in vivo extrapolation (IVIVE) can be used to estimate hepatic metabolic CL. In vitro turnover rates determined using hepatocytes, or subcellular liver fractions, such as microsomes, are scaled to predict in vivo liver CL using a mathematical model [2.Grime K. Riley R.J. The impact of in vitro binding on in vitro-in vivo extrapolations, projections of metabolic clearance and clinical drug-drug interactions.Curr. Drug Metab. 2006; 7: 251-264Crossref PubMed Scopus (114) Google Scholar,3.Riley R.J. et al.A unified model for predicting human hepatic, metabolic clearance from in vitro intrinsic clearance data in hepatocytes and microsomes.Drug Metab. Dispos. 2005; 33: 1304-1311Crossref PubMed Scopus (301) Google Scholar,39.Sohlenius-Sternbeck A.K. et al.Practical use of the regression offset approach for the prediction of in vivo intrinsic clearance from hepatocytes.Xenobiotica. 2012; 42: 841-853Crossref PubMed Scopus (73) Google Scholar, 40.Lave T. et al.Human clearance prediction: shifting the paradigm.Expert Opin. Drug Metab. Toxicol. 2009; 5: 1039-1048Crossref PubMed Scopus (40) Google Scholar, 41.Ito K. Houston J.B. Prediction of human drug clearance from in vitro and preclinical data using physiologically based and empirical approaches.Pharm. Res. 2005; 22: 103-112Crossref PubMed Scopus (230) Google Scholar, 42.Obach R.S. Predicting clearance in humans from in vitro data.Curr. Top. Med. Chem. 2011; 11: 334-339Crossref PubMed Scopus (75) Google Scholar, 43.Houston J.B. Utility of in vitro drug metabolism data in predicting in vivo metabolic clearance.Biochem. Pharmacol. 1994; 47: 1469-1479Crossref PubMed Scopus (766) Google Scholar]. The model describes hepatic metabolic CL (CLh), via liver blood flow (Qh), hepatic intrinsic clearance (CLint) and fraction unbound in blood (fub). Both hepatic microsomes and hepatocytes are used in AstraZeneca; hepatocytes contain the full complement of DMEs, while microsomes contain membrane bound DMEs such as CYPs. The method involves determination of hepatocyte (or microsomal) intrinsic clearance rate (CLint) and fraction unbound in the incubation (fuinc), to calculate unbound in vitro CLint, which can be scaled up to a whole liver unbound CLint. Evaluation of the predictive accuracy can be made against the relationship of unbound in vitro versus in vivo CLint, in animals for the chemical series of interest and in human for drugs with clinical PK. This transformation of unbound in vitro to in vivo CLint yields a systematic underprediction of in vivo CLint, which can be corrected empirically via use of a multiplicative factor or regression line, and, importantly, appears across different laboratories, species and hepatocytes/microsomes. In our experience, the empirical correction can be used for substrates of CYPs, flavin-containing monooxygenase (FMO), and some conjugative routes of metabolism, and is periodically reassessed via a test set of compounds [2.Grime K. Riley R.J. The impact of in vitro binding on in vitro-in vivo extrapolations, projections of metabolic clearance and clinical drug-drug interactions.Curr. Drug Metab. 2006; 7: 251-264Crossref PubMed Scopus (114) Google Scholar,39.Sohlenius-Sternbeck A.K. et al.Practical use of the regression offset approach for the prediction of in vivo intrinsic clearance from hepatocytes.Xenobiotica. 2012; 42: 841-853Crossref PubMed Scopus (73) Google Scholar,44.Jones B.C. et al.An investigation into the prediction of in vivo clearance for a range of flavin-containing monooxygenase substrates.Drug Metab. Dispos. 2017; 45: 1060-1067Crossref PubMed Scopus (18) Google Scholar,45.Hallifax D. et al.Prediction of human metabolic clearance from in vitro systems: retrospective analysis and prospective view.Pharm. Res. 2010; 27: 2150-2161Crossref PubMed Scopus (144) Google Scholar]. Renal elimination may involve both passive transport and active transport components. The renal filtration rate, as a CL, is a product of a drug's fraction unbound in blood (fub) and glomerular filtration rate (GFR, ~1.7 ml/min/kg in humans). As the majority of drugs are lipophilic (log D>0), and therefore highly plasma bound and passively reabsorbed from the urine, passive renal CL is usuall