Prediction of human clearance from animal data and molecular structural parameters using multivariate regression analysis

Ulgen K. Advances in Physiologically Based Pharmacokinetic (PBPK) Modeling of Nanomaterials

Nanoparticles (NPs) have been widely used to improve the pharmacokinetic properties and tissue distribution of small molecules such as targeting to a specific tissue of interest, enhancing their systemic circulation, and enlarging their therapeutic properties. NPs have unique and complicated in vivo disposition properties compared to small molecule drugs due to their complex multifunctionality. Physiologically based pharmacokinetic (PBPK) modeling has been a powerful tool in the simulation of the absorption, distribution, metabolism, and elimination (ADME) characteristics of the materials, and it can be used in the characterization and prediction of the systemic disposition, toxicity, efficacy, and target exposure of various types of nanoparticles. In this review, recent advances in PBPK model applications related to the nanoparticles with unique properties, and dispositional features in the biological systems, ADME characteristics, the description of transport processes of nanoparticles in the PBPK model, and the challenges in PBPK model development of nanoparticles are delineated and juxtaposed with those encountered in small molecule models. Nanoparticle related, non-nanoparticle-related, and interspecies-scaling methods applied in PBPK modeling are reviewed. In vitro to in vivo extrapolation (IVIVE) methods being a promising computational tool to provide in vivo predictions from the results of in vitro and in silico studies are discussed. Finally, as a recent advancement ML/AI-based approaches and challenges in PBPK modeling in the estimation of ADME parameters and pharmacokinetic (PK) analysis results are introduced.

Predicting human half-life for insulin analogs: An inter-drug approach

An inter-drug approach, applying pharmacokinetic information for insulin analogs in different animal species, rat, dog and pig, performed better compared to allometric scaling for human translation of intra-venous half-life and only required data from a single animal species for reliable predictions. Average fold error (AFE) between 1.2-1.7 were determined for all species and for multispecies allometric scaling AFE was 1.9. A slightly larger prediction error for human half-life was determined from in vitro human insulin receptor affinity data (AFE on 2.3-2.6). The requirements for the inter-drug approach were shown to be a span of at least 2 orders of magnitude in half-life for the included drugs and a shared clearance mechanism. The insulin analogs in this study were the five fatty acid protracted analogs: Insulin degludec, insulin icodec, insulin 320, insulin 338 and insulin 362, as well as the non-acylated analog insulin aspart.

The Comparison of Machine Learning and Mechanistic In Vitro-In Vivo Extrapolation Models for the Prediction of Human Intrinsic Clearance

Accurate prediction of human pharmacokinetics (PK) remains one of the key objectives of drug metabolism and PK (DMPK) scientists in drug discovery projects. This is typically performed by using in vitro-in vivo extrapolation (IVIVE) based on mechanistic PK models. In recent years, machine learning (ML), with its ability to harness patterns from previous outcomes to predict future events, has gained increased popularity in application to absorption, distribution, metabolism, and excretion (ADME) sciences. This study compares the performance of various ML and mechanistic models for the prediction of human IV clearance for a large (645) set of diverse compounds with literature human IV PK data, as well as measured relevant in vitro end points. ML models were built using multiple approaches for the descriptors: (1) calculated physical properties and structural descriptors based on chemical structure alone (classical QSAR/QSPR); (2) in vitro measured inputs only with no structure-based descriptors (ML IVIVE); and (3) in silico ML IVIVE using in silico model predictions for the in vitro inputs. For the mechanistic models, well-stirred and parallel-tube liver models were considered with and without the use of empirical scaling factors and with and without renal clearance. The best ML model for the prediction of in vivo human intrinsic clearance (CLint) was an in vitro ML IVIVE model using only six in vitro inputs with an average absolute fold error (AAFE) of 2.5. The best mechanistic model used the parallel-tube liver model, with empirical scaling factors resulting in an AAFE of 2.8. The corresponding mechanistic model with full in silico inputs achieved an AAFE of 3.3. These relative performances of the models were confirmed with the prediction of 16 Pfizer drug candidates that were not part of the original data set. Results show that ML IVIVE models are comparable to or superior to their best mechanistic counterparts. We also show that ML IVIVE models can be used to derive insights into factors for the improvement of mechanistic PK prediction.

Discovery of a Potent and Selective Tyrosine Kinase 2 Inhibitor: TAK-279

TYK2 is a key mediator of IL12, IL23, and type I interferon signaling, and these cytokines have been implicated in the pathogenesis of multiple inflammatory and autoimmune diseases such as psoriasis, rheumatoid arthritis, lupus, and inflammatory bowel diseases. Supported by compelling data from human genome-wide association studies and clinical results, TYK2 inhibition through small molecules is an attractive therapeutic strategy to treat these diseases. Herein, we report the discovery of a series of highly selective pseudokinase (Janus homology 2, JH2) domain inhibitors of TYK2 enzymatic activity. A computationally enabled design strategy, including the use of FEP+, was instrumental in identifying a pyrazolo-pyrimidine core. We highlight the utility of computational physics-based predictions used to optimize this series of molecules to identify the development candidate 30, a potent, exquisitely selective cellular TYK2 inhibitor that is currently in Phase 2 clinical trials for the treatment of psoriasis and psoriatic arthritis.

Development of a 2D-QSAR Model for Tissue-to-Plasma Partition Coefficient Value with High Accuracy Using Machine Learning Method, Minimum Required Experimental Values, and Physicochemical Descriptors

BACKGROUND

The demand for physiologically based pharmacokinetic (PBPK) model is increasing currently. New drug application (NDA) of many compounds is submitted with PBPK models for efficient drug development. Tissue-to-plasma partition coefficient (K_p) is a key parameter for the PBPK model to describe differential equations. However, it is difficult to obtain the K_p value experimentally because the measurement of drug concentration in the tissue is much harder than that in plasma.

OBJECTIVE

Instead of experiments, many researchers have sought in silico methods. Today, most of the models for K_p prediction are using in vitro and in vivo parameters as explanatory variables. We thought of physicochemical descriptors that could improve the predictability. Therefore, we aimed to develop the two-dimensional quantitative structure-activity relationship (2D-QSAR) model for K_p using physicochemical descriptors instead of in vivo experimental data as explanatory variables.

METHODS

We compared our model with the conventional models using 20-fold cross-validation according to the published method (Yun et al. J Pharmacokinet Pharmacodyn 41:1-14, 2014). We used random forest algorithm, which is known to be one of the best predictors for the 2D-QSAR model. Finally, we combined minimum in vitro experimental values and physiochemical descriptors. Thus, the prediction method for K_p value using a few in vitro parameters and physicochemical descriptors was developed; this is a multimodal model.

RESULTS

Its accuracy was found to be superior to that of the conventional models. Results of this research suggest that multimodality is useful for the 2D-QSAR model [RMSE and % of two-fold error: 0.66 and 42.2% (Berezohkovsky), 0.52 and 52.2% (Rodgers), 0.65 and 34.6% (Schmitt), 0.44 and 61.1% (published model), 0.41 and 62.1% (traditional model), 0.39 and 64.5% (multimodal model)].

CONCLUSION

We could develop a 2D-QSAR model for K_p value with the highest accuracy using a few in vitro experimental data and physicochemical descriptors.

Computational Predictions of Nonclinical Pharmacokinetics at the Drug Design Stage

Although computational predictions of pharmacokinetics (PK) are desirable at the drug design stage, existing approaches are often limited by prediction accuracy and human interpretability. Using a discovery data set of mouse and rat PK studies at Roche (9,685 unique compounds), we performed a proof-of-concept study to predict key PK properties from chemical structure alone, including plasma clearance (CLp), volume of distribution at steady-state (Vss), and oral bioavailability (F). Ten machine learning (ML) models were evaluated, including Single-Task, Multitask, and transfer learning approaches (i.e., pretraining with in vitro data). In addition to prediction accuracy, we emphasized human interpretability of outcomes, especially the quantification of uncertainty, applicability domains, and explanations of predictions in terms of molecular features. Results show that intravenous (IV) PK properties (CLp and Vss) can be predicted with good precision (average absolute fold error, AAFE of 1.96-2.84 depending on data split) and low bias (average fold error, AFE of 0.98-1.36), with AutoGluon, Gaussian Process Regressor (GP), and ChemProp displaying the best performance. Driven by higher complexity of oral PK studies, predictions of F were more challenging, with the best AAFE values of 2.35-2.60 and higher overprediction bias (AFE of 1.45-1.62). Multi-Task approaches and pretraining of ChemProp neural networks with in vitro data showed similar precision to Single-Task models but helped reduce the bias and increase correlations between observations and predictions. A combination of GP-computed prediction variance, molecular clustering, and dimensionality-reduction provided valuable quantitative insights into prediction uncertainty and applicability domains. SHAPley Additive exPlanations (SHAPs) highlighted molecular features contributing to prediction outcomes of Vss, providing explanations that could aid drug design. Combined results show that computational predictions of PK are feasible at the drug design stage, with several ML technologies converging to successfully leverage historical PK data sets. Further studies are needed to unlock the full potential of this approach, especially with respect to data set sizes and quality, transfer learning between in vitro and in vivo data sets, model-independent quantification of uncertainty, and explainability of predictions.

Prediction of Total Drug Clearance in Humans Using Animal Data: Proposal of a Multimodal Learning Method Based on Deep Learning

Research into pharmacokinetics plays an important role in the development process of new drugs. Accurately predicting human pharmacokinetic parameters from preclinical data can increase the success rate of clinical trials. Since clearance (CL) which indicates the capacity of the entire body to process a drug is one of the most important parameters, many methods have been developed. However, there are still rooms to be improved for practical use in drug discovery research; "improving CL prediction accuracy" and "understanding the chemical structure of compounds in terms of pharmacokinetics". To improve those, this research proposes a multimodal learning method based on deep learning that takes not only the chemical structure of a drug but also rat CL as inputs. Good results were obtained compared with the conventional animal scale-up method; the geometric mean fold error was 2.68 and the proportion of compounds with prediction errors of 2-fold or less was 48.5%. Furthermore, it was found to be possible to infer the partial structure useful for CL prediction by a structure contributing factor inference method. The validity of these results of structural interpretation of metabolic stability was confirmed by chemists.

Interpretation of Non-Clinical Data for Prediction of Human Pharmacokinetic Parameters: In Vitro-In Vivo Extrapolation and Allometric Scaling

Extrapolation of pharmacokinetic (PK) parameters from in vitro or in vivo animal to human is one of the main tasks in the drug development process. Translational approaches provide evidence for go or no-go decision-making during drug discovery and the development process, and the prediction of human PKs prior to the first-in-human clinical trials. In vitro-in vivo extrapolation and allometric scaling are the choice of method for projection to human situations. Although these methods are useful tools for the estimation of PK parameters, it is a challenge to apply these methods since underlying biochemical, mathematical, physiological, and background knowledge of PKs are required. In addition, it is difficult to select an appropriate methodology depending on the data available. Therefore, this review covers the principles of PK parameters pertaining to the clearance, volume of distribution, elimination half-life, absorption rate constant, and prediction method from the original idea to recently developed models in order to introduce optimal models for the prediction of PK parameters.

Misconceptions and issues regarding allometric scaling during the drug development process

INTRODUCTION

Allometry is the study of size and its consequences. The simple hypothesis of allometric scaling is that all physiological parameters are proportional to body size or body mass. This review examines the development of theory-based allometry or fixed exponents (0.75 and 1.0 for basal metabolic rate and volume, respectively) and the evidence for or against the theory. The main focus of this report is to show the readers that there is enough evidence from experimental data that negate the concept of theory-based allometry in biology, physiology, and pharmacokinetics. Areas covered: In this review, the history of the development of theoretical allometry and the strong evidence against theory-based allometry demonstrated by experimental data is provided. During drug development, allometry is applied to both inter-species (from animals to humans) and intra-species (adults to children) scaling. These two forms of allometric scaling in the context of theory-based allometry are discussed and provide insight on scientific progress that refute theory-based allometry. Expert opinion: Theory-based allometry is a mere theory and experimental data and real-life observations strongly negate the existence of such a theory. Pharmacostatistical and physiological models based on theory-based allometry can be misleading and incorrect because the theory-based allometric exponent 0.75 is not universal. The exponents of allometry are data dependent and are not fixed in the universe.

Mechanistic investigations into the species differences in pinometostat clearance: impact of binding to alpha-1-acid glycoprotein and permeability-limited hepatic uptake

1. The plasma clearance of the first-in-class DOT1L inhibitor, EPZ-5676 (pinometostat), was shown to be markedly lower in human compared to the preclinical species, mouse, rat and dog. 2. This led to vertical allometry where various interspecies scaling methods were applied to the data, with fold-errors between 4 and 13. We had previously reported the elimination and metabolic pathways of EPZ-5676 were similar across species. Therefore, the aim of this work was to explore the mechanistic basis for the species difference in clearance for EPZ-5676, focusing on other aspects of disposition. 3. The protein binding of EPZ-5676 in human plasma demonstrated a non-linear relationship suggesting saturable binding at physiologically relevant concentrations. Saturation of protein binding was not observed in plasma from preclinical species. Kinetic determinations using purified serum albumin and alpha-1-acid glycoprotein (AAG) confirmed that EPZ-5676 is a high affinity ligand for AAG with a dissociation constant (K_d) of 0.24 μM. 4. Permeability limited uptake was also considered since hepatocyte CL_int was much lower in human relative to preclinical species. Passive unbound CL_int for EPZ-5676 was estimated using a correlation analysis of logD and data previously reported on seven drugs in sandwich cultured human hepatocytes. 5. Incorporation of AAG binding and permeability limited hepatic uptake into the well-stirred liver model gave rise to a predicted clearance for EPZ-5676 within 2-fold of the observed value of 1.4 mL min^-1kg^-1. This analysis suggests that the marked species difference in EPZ-5676 clearance is driven by high affinity binding to human AAG as well as species-specific hepatic uptake invoking the role of transporters.

Systematic Evaluation of Wajima Superposition (Steady-State Concentration to Mean Residence Time) in the Estimation of Human Intravenous Pharmacokinetic Profile

We present a systematic evaluation of the Wajima superpositioning method to estimate the human intravenous (i.v.) pharmacokinetic (PK) profile based on a set of 54 marketed drugs with diverse structure and range of physicochemical properties. We illustrate the use of average of "best methods" for the prediction of clearance (CL) and volume of distribution at steady state (VDss) as described in our earlier work (Lombardo F, Waters NJ, Argikar UA, et al. J Clin Pharmacol. 2013;53(2):178-191; Lombardo F, Waters NJ, Argikar UA, et al. J Clin Pharmacol. 2013;53(2):167-177). These methods provided much more accurate prediction of human PK parameters, yielding 88% and 70% of the prediction within 2-fold error for VDss and CL, respectively. The prediction of human i.v. profile using Wajima superpositioning of rat, dog, and monkey time-concentration profiles was tested against the observed human i.v. PK using fold error statistics. The results showed that 63% of the compounds yielded a geometric mean of fold error below 2-fold, and an additional 19% yielded a geometric mean of fold error between 2- and 3-fold, leaving only 18% of the compounds with a relatively poor prediction. Our results showed that good superposition was observed in any case, demonstrating the predictive value of the Wajima approach, and that the cause of poor prediction of human i.v. profile was mainly due to the poorly predicted CL value, while VDss prediction had a minor impact on the accuracy of human i.v. profile prediction.

Prediction of human clearance based on animal data and molecular properties

Human clearance is often predicted prior to clinical study from in vivo preclinical data by virtue of interspecies allometric scaling methods. The aims of this study were to determine the important molecular descriptors for the extrapolation of animal data to human clearance and further to build a model to predict human clearance by combination of animal data and the selected molecular descriptors. These important molecular descriptors selected by genetic algorithm (GA) were from five classes: quantum mechanical, shadow indices, E-state keys, molecular properties, and molecular property counts. Although the data set contained many outliers determined by the conventional Mahmood method, the variation of most outliers was reduced significantly by our final support vector machine (SVM) model. The values of cross-validated correlation coefficient and root-mean-squared error (RMSE) for leave-one-out cross-validation (LOOCV) of the final SVM model were 0.783 and 0.305, respectively. Meanwhile, the reliability and consistency of the final model were also validated by an external test set. In conclusion, the SVM model based on the molecular descriptors selected by GA and animal data achieved better prediction performance than the Mahmood method. This approach can be applied as an improved interspecies allometric scaling method in drug research and development.

Use of artificial neural networks in the QSAR prediction of physicochemical properties and toxicities for REACH legislation

With the introduction of the REACH legislation in the European Union, there is a requirement for property and toxicity data on chemicals produced in or imported into the EU at levels of 1 tonne/year or more. This has meant an increase in the in silico prediction of such data. One of the chief predictive approaches is QSAR (quantitative structure-activity relationships), which is widely used in many fields. A QSAR approach that is increasingly being used is that of artificial neural networks (ANNs), and this chapter discusses its application to the range of physicochemical properties and toxicities required by REACH. ANNs generally outperform the main QSAR approach of multiple linear regression (MLR), although other approaches such as support vector machines sometimes outperform ANNs. Most ANN QSARs reported to date comply with only two of the five OECD Guidelines for the Validation of (Q)SARs.

The application of allometric scaling principles to predict pharmacokinetic parameters across species

INTRODUCTION

Interspecies allometric scaling provides a simple and fast option to interpolate or extrapolate drug dose or pharmacokinetic parameters to a species of interest. Over the years, new scaling methods have been developed in order to improve the performance of these predictions. It is critical to choose appropriate allometric scaling approach(es) to analyze the available pharmacokinetic data.

AREAS COVERED

This review provides updated information on the latest allometric scaling methods developed for the most frequently interpolated or extrapolated pharmacokinetic parameters. The different degrees of success and advantages/disadvantages of different methods are compared and contrasted. The pitfalls that affect the accuracy of prediction and the solutions to avoid the risk of prediction errors are discussed. The application of allometric scaling in veterinary medicine is presented.

EXPERT OPINION

Although interspecies allometric scaling needs further refinements and has limitations, it is still a potential tool and rational option for the estimate of pharmacokinetic parameters in species for which there are no data available or to better interpret preclinical efficacy and safety trials. Allometric scaling can offer insight into possible mechanisms of species-dependent drug disposition.

Nilotinib preclinical pharmacokinetics and practical application toward clinical projections of oral absorption and systemic availability

Nilotinib is a highly potent and selective bcr-abl tyrosine kinase inhibitor used for the treatment of patients who are in the chronic and accelerated phases of Philadelphia chromosome-positive (Ph+) chronic myeloid leukemia (CML). Nilotinib preclinical data and its use for practical predictions of systemic exposure profiles and oral absorption are described. The systemic clearance (CL) of nilotinib was relatively low in rodents with a value of less than 25% of hepatic blood flow (Q(H)), while it was moderate in monkeys and dogs (CL/Q(H)  = 32-35%). The steady state volume of distribution (V(ss) ) ranged from 0.55 to 3.9 l/kg across the species tested. The maximum concentration (C(max)) of nilotinib occurred at 0.5-4 h and the bioavailability was moderate (17-44%). The plasma protein binding was high (> 97.5%) in preclinical species and humans. The human CL (~ 0.1 l/h/kg) and V(ss) (~2.0 l/kg) were best predicted by the rat-dog-human proportionality method and allometric scaling method, respectively. The human intravenous pharmacokinetic profile was projected by the Wajima 'C(ss)-MRT' method. The predicted micro-constants from human intravenous profiles were incorporated into the advanced compartmental absorption and transit model within the GastroPlus program to simulate the oral concentration-time curves in humans. Overall, the simulated oral human pharmacokinetic profiles showed good agreement with observed clinical data, and the model predicted that the C(max) , AUC, t(½) , V(z) /F and CL/F values were within 1.3-fold of the observed values. The absolute oral bioavailability of nilotinib in healthy humans was predicted to be low (< 25%).

Applications of human pharmacokinetic prediction in first-in-human dose estimation

Quantitative estimations of first-in-human (FIH) doses are critical for phase I clinical trials in drug development. Human pharmacokinetic (PK) prediction methods have been developed to project the human clearance (CL) and bioavailability with reasonable accuracy, which facilitates estimation of a safe yet efficacious FIH dose. However, the FIH dose estimation is still very challenging and complex. The aim of this article is to review the common approaches for FIH dose estimation with an emphasis on PK-guided estimation. We discuss 5 methods for FIH dose estimation, 17 approaches for the prediction of human CL, 6 methods for the prediction of bioavailability, and 3 tools for the prediction of PK profiles. This review may serve as a practical protocol for PK- or pharmacokinetic/pharmacodynamic-guided estimation of the FIH dose.

Prediction of pharmacokinetic parameters

In silico tools specifically developed for prediction of pharmacokinetic parameters are of particular interest to pharmaceutical industry because of the high potential of discarding inappropriate molecules during an early stage of drug development itself with consequent saving of vital resources and valuable time. The ultimate goal of the in silico models of absorption, distribution, metabolism, and excretion (ADME) properties is the accurate prediction of the in vivo pharmacokinetics of a potential drug molecule in man, whilst it exists only as a virtual structure. Various types of in silico models developed for successful prediction of the ADME parameters like oral absorption, bioavailability, plasma protein binding, tissue distribution, clearance, half-life, etc. have been briefly described in this chapter.

Approaches for predicting human pharmacokinetics using interspecies pharmacokinetic scaling

Reliably predicting pharmacokinetic behavior in humans from preclinical data is an important aspect of drug development. The most widely used technique in this regard is allometric scaling. In this review, various approaches developed for predicting pharmacokinetic parameters in humans using interspecies scaling are introduced and discussed. Methods to predict plasma concentration-time profiles in humans after intravenous and oral administration are also reviewed. The reliable prediction of human pharmacokinetics with regard to investigational drugs is aimed, ultimately, at selecting the first in-human dose with which to begin clinical studies. Approaches for the selection of the first in-human dose are also reviewed. Although there have been many trials to compare and optimize interspecies scaling methods, no firm conclusions have been reached. Because interspecies scaling methods are still highly empirical, further effort is needed to improve the reliability of predicting human pharmacokinetics by interspecies scaling.

Prediction of human oral plasma concentration-time profiles using preclinical data: comparative evaluation of prediction approaches in early pharmaceutical discovery

BACKGROUND AND OBJECTIVES

Empirically based methods remain one of our tools in human pharmacokinetic predictions. The Dedrick approach and the steady-state plasma drug concentration (C(ss))-mean residence time (MRT) approach are based on the assumption that concentration-time profiles are similar among species, including man, and that curves derived from a variety of animal species can be superimposed after mathematical transformation. In the Dedrick approach the transformation is based on the slope and intercept of the allometric relationship. The C(ss)-MRT approach is based on the implementation of measured animal and predicted human MRT and dose/volume of distribution at steady state (V(ss)). The aims of the present study were to compare the predictive performance of concentration-time profiles obtained by these approaches, to evaluate the prediction of individual pharmacokinetic parameters by these approaches and to further refine these approaches incorporating the experience from our previous work.

METHODS

A retrospective analysis using 35 proprietary compounds developed at Johnson & Johnson Pharmaceutical Research and Development was conducted to compare the accuracies of the Dedrick and C(ss)-MRT approaches for predicting oral concentration-time profiles and pharmacokinetic parameters in man. In the first step, input for the transformation was based on simple allometry. Then we assessed whether both methods could be fine-tuned by systematically incorporating correction factors (maximum life span potential, brain weight and plasma protein binding), depending on the interspecies relationship. In addition, for the C(ss)-MRT approach, we used formulas based on multivariate regression analysis as input for the transformation.

RESULTS

Inclusion of correction factors significantly improved the profile predictability for the Dedrick and C(ss)-MRT approaches. This was mainly linked to an improved prediction of terminal elimination half-life (t(½)), MRT and the ratio between the maximum plasma concentration and the concentration at the last observed time point (C(max)/C(last)). No significant differences were observed between the Dedrick approach with correction factors, the C(ss)-MRT approach with correction factors and the C(ss)-MRT approach, based on the regression equations.

CONCLUSIONS

Based on the dataset evaluated in this study, we demonstrated that human plasma concentration-time profiles and pharmacokinetic parameters could be predicted with the Dedrick and C(ss)-MRT approaches and that if correction factors were implemented, the predictions improved significantly. With the requirement of only a limited preclinical in vivo pharmacokinetic dataset, these empirical methods could offer potential in the early stages of drug discovery.

PhRMA CPCDC initiative on predictive models of human pharmacokinetics, part 3: comparative assessement of prediction methods of human clearance

The objective of this study was to evaluate the performance of various allometric and in vitro-in vivo extrapolation (IVIVE) methodologies with and without plasma protein binding corrections for the prediction of human intravenous (i.v.) clearance (CL). The objective was also to evaluate the IVIVE prediction methods with animal data. Methodologies were selected from the literature. Pharmaceutical Research and Manufacturers of America member companies contributed blinded datasets from preclinical and clinical studies for 108 compounds, among which 19 drugs had i.v. clinical pharmacokinetics data and were used in the analysis. In vivo and in vitro preclinical data were used to predict CL by 29 different methods. For many compounds, in vivo data from only two species (generally rat and dog) were available and/or the required in vitro data were missing, which meant some methods could not be properly evaluated. In addition, 66 methods of predicting oral (p.o.) area under the curve (AUCp.o. ) were evaluated for 107 compounds using rational combinations of i.v. CL and bioavailability (F), and direct scaling of observed p.o. CL from preclinical species. Various statistical and outlier techniques were employed to assess the predictability of each method. Across methods, the maximum success rate in predicting human CL for the 19 drugs was 100%, 94%, and 78% of the compounds with predictions falling within 10-fold, threefold, and twofold error, respectively, of the observed CL. In general, in vivo methods performed slightly better than IVIVE methods (at least in terms of measures of correlation and global concordance), with the fu intercept method and two-species-based allometry (rat-dog) being the best performing methods. IVIVE methods using microsomes (incorporating both plasma and microsomal binding) and hepatocytes (not incorporating binding) resulted in 75% and 78%, respectively, of the predictions falling within twofold error. IVIVE methods using other combinations of binding assumptions were much less accurate. The results for prediction of AUCp.o. were consistent with i.v. CL. However, the greatest challenge to successful prediction of human p.o. CL is the estimate of F in human. Overall, the results of this initiative confirmed predictive performance of common methodologies used to predict human CL.

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

The objective of this study was to evaluate the performance of the Wajima allometry (Css -MRT) approach published in the literature, which is used to predict the human plasma concentration-time profiles from a scaling of preclinical species data. A diverse and blinded dataset of 108 compounds from PhRMA member companies was used in this evaluation. The human intravenous (i.v.) and oral (p.o.) pharmacokinetics (PK) data were available for 18 and 107 drugs, respectively. Three different scenarios were adopted for prediction of human PK profiles. In the first scenario, human clearance (CL) and steady-state volume of distribution (Vss ) were predicted by unbound fraction corrected intercept method (FCIM) and Øie-Tozer (OT) approaches, respectively. Quantitative structure activity relationship (QSAR)-based approaches (TSrat-dog ) based on compound descriptors together with rat and dog data were utilized in the second scenario. Finally, in the third scenario, CL and Vss were predicted using the FCIM and Jansson approaches, respectively. For the prediction of oral pharmacokinetics, the human bioavailability and absorption rate constant were assumed as the average of preclinical species. Various statistical techniques were used for assessing the accuracy of the simulation scenarios. The human CL and Vss were predicted within a threefold error range for about 75% of the i.v. drugs. However, the accuracy in predicting key p.o. PK parameters appeared to be lower with only 58% of simulations falling within threefold of observed parameters. The overall ability of the Css -MRT approach to predict the curve shape of the profile was in general poor and ranged between low to medium level of confidence for most of the predictions based on the selected criteria.

Prediction of human pharmacokinetics—evaluation of methods for prediction of hepatic metabolic clearance

Methods for prediction of hepatic clearance (CLH) in man have been evaluated. A physiologically-based in-vitro to in-vivo (PB-IVIV) method with human unbound fraction in blood (fu,bl) and hepatocyte intrinsic clearance (CLint)-data has a good rationale and appears to give the best predictions (maximum ∼2-fold errors; &lt; 25% errors for half of CL-predictions; appropriate ranking). Inclusion of an empirical scaling factor is, however, needed, and reasons include the use of cryopreserved hepatocytes with low activity, and inappropriate CLint- and fu,bl-estimation methods. Thus, an improvement of this methodology is possible and required. Neglect of fu,bl or incorporation of incubation binding does not seem appropriate. When microsome CLint-data are used with this approach, the CLH is underpredicted by 5- to 9-fold on average, and a 106-fold underprediction (attrition potential) has been observed. The poor performance could probably be related to permeation, binding and low metabolic activity. Inclusion of scaling factors and neglect of fu,bl for basic and neutral compounds improve microsome predictions. The performance is, however, still not satisfactory. Allometry incorrectly assumes that the determinants for CLH relate to body weight and overpredicts human liver blood flow rate. Consequently, allometric methods have poor predictability. Simple allometry has an average overprediction potential, &gt; 2-fold errors for ∼1/3 of predictions, and 140-fold underprediction to 5800-fold overprediction (potential safety risk) range. In-silico methodologies are available, but these need further development. Acceptable prediction errors for compounds with low and high CLH should be ∼50 and ∼10%, respectively. In conclusion, it is recommended that PB-IVIV with human hepatocyte CLint and fu,bl is applied and improved, limits for acceptable errors are decreased, and that animal CLH-studies and allometry are avoided.

Physiologically based pharmacokinetics (PBPK)

Allometric scaling is widely used to predict human pharmacokinetic parameters from preclinical species, and many different approaches have been proposed over the years to improve its predictive performance. Nevertheless, prediction errors are commonly observed in the practical application of simple allometry, for example, in cases where the hepatic metabolic clearance is mainly determined by enzyme activities, which do not scale allometrically across species. Therefore, if good correlation was noted for some drugs, poor correlation was observed for others, highlighting the need for other conceptual approaches. Physiologically based pharmacokinetic (PBPK) models are now a well-established approach to conduct extrapolations across species and to generate simulations of pharmacokinetic profiles under various physiological conditions. While conventional pharmacokinetic models are defined by drug-related data themselves, PBPK models have richer information content and integrate information from various sources, including drug-dependent, physiological, and biological parameters as they vary in between species, subjects, or with age and disease state. Therefore, the biological and mechanistic bases of PBPK models allow the extrapolation of the kinetic behavior of drugs with regard to dose, route, and species. In addition, by providing a link between tissue concentrations and toxicological or pharmacological effects, PBPK modeling represents a framework for mechanistic pharmacokinetic-pharmacodynamic models.

To scale or not to scale: the principles of dose extrapolation

The principles of inter-species dose extrapolation are poorly understood and applied. We provide an overview of the principles underlying dose scaling for size and dose adjustment for size-independent differences. Scaling of a dose is required in three main situations: the anticipation of first-in-human doses for clinical trials, dose extrapolation in veterinary practice and dose extrapolation for experimental purposes. Each of these situations is discussed. Allometric scaling of drug doses is commonly used for practical reasons, but can be more accurate when one takes into account species differences in pharmacokinetic parameters (clearance, volume of distribution). Simple scaling of drug doses can be misleading for some drugs; correction for protein binding, physicochemical properties of the drug or species differences in physiological time can improve scaling. However, differences in drug transport and metabolism, and in the dose-response relationship, can override the effect of size alone. For this reason, a range of modelling approaches have been developed, which combine in silico simulations with data obtained in vitro and/or in vivo. Drugs that are unlikely to be amenable to simple allometric scaling of their clearance or dose include drugs that are highly protein-bound, drugs that undergo extensive metabolism and active transport, drugs that undergo significant biliary excretion (MW > 500, ampiphilic, conjugated), drugs whose targets are subject to inter-species differences in expression, affinity and distribution and drugs that undergo extensive renal secretion. In addition to inter-species dose extrapolation, we provide an overview of dose extrapolation within species, discussing drug dosing in paediatrics and in the elderly.

Evaluation of a basic physiologically based pharmacokinetic model for simulating the first-time-in-animal study

The objective of this study was to evaluate a physiologically based pharmacokinetic (PBPK) approach for predicting the plasma concentration-time curves expected after intravenous administration of candidate drugs to rodents. The predictions were based on a small number of properties that were either calculated based on the structure of the candidate drug (octanol:water partition coefficient, ionization constant(s)) or obtained from the typical high-throughput screens implemented in the early drug discovery phases (fraction unbound in plasma and hepatic intrinsic clearance). The model was tested comparing the predicted and the observed pharmacokinetics of 45 molecules. This dataset included six known drugs and 39 drug candidates from different discovery programs, so that the performance of the model could be evaluated in a real discovery case scenario. The plasma concentration-time curves were predicted with good accuracy, the pharmacokinetic parameters being on average two- to three-fold of actual values. Multivariate analysis was used for identifying the candidate properties which were likely associated to biased predictions. The application of this approach was found useful for the prioritization of the in vivo pharmacokinetics screens and the design of the first-time-in-animal studies.

Equation chapter 1 section 1A new method for predicting human hepatic clearance fromin Vitro experimental data using molecular descriptors

The present study demonstrated that the information of molecular descriptors of drugs increases the accuracy of predicting human in vivo hepatic clearance from in vitro experimental data in humans and rats. A new method uses not only the experimental data but also the information of molecular descriptors. Predictions for the datasets from hepatocyte experiments and microsome experiments were made by the present method, and the prediction accuracy was compared with those of the previous methods, such as methods using in vitro-in vivo scaling factor and multiple linear regression analysis, that use only the experimental data. Results showed that the present method was the most accurate prediction model with the lowest prediction errors and the strongest correlations. These results suggest that the information of molecular descriptors is significant for predicting the human in vivo pharmacokinetic parameters from in vitro experimental data. This study also demonstrated that in vitro experimental data in humans and rats were important information for predicting human in vivo hepatic clearance, and the additional rat in vivo data were not significant for prediction with the information of molecular descriptors. These results imply that the present method can be useful for high-throughput drug candidate screening by reducing the time and cost in the early stage of the drug discovery process.

Quantitative structure-pharmacokinetic/pharmacodynamic relationships

Quantitative structure-activity relationships have long been considered a vital component of drug discovery and development, providing insight into the role of molecular properties in the biological activity of similar and unrelated compounds. Recognition that in vitro bioassay and/or pre-clinical activity are insufficient for anticipating which compounds are suitable leads for further development has shifted the focus toward integrated pharmacokinetic (PK) and pharmacodynamic (PD) processes. Over the last decade, considerable progress has been made in constructing empirical and mechanistic quantitative structure-PK relationships (QSPKR), as well as diverse mechanism-based pharmacodynamic models of drug effects. In this review, traditional and contemporary approaches to developing QSPKR models are discussed, along with selected examples of attempts to couple QSPKR and pharmacodynamic models to anticipate the intensity and time-course of the pharmacological effects of new or related compounds, or quantitative structure-pharmacodynamic relationships modeling. Such models are in accordance with the goals of systems biology and the ideal of designing drugs and delivery systems from first principles.

Prediction of human drug clearance from animal data: Application of the rule of exponents and ‘fu corrected intercept method’ (FCIM)

The objective of this study is to evaluate the predictive performance of the rule of exponents (ROE) and 'fu Corrected Intercept Method' (FCIM) for the human drug clearance. Different classes of drugs such as extensively metabolized, renally excreted, renally secreted, and biliary excreted drugs were used in this analysis. The results of the study indicated that both these methods under given conditions are extremely useful for the prediction of human drug clearance. There are certain situations under which one of these methods is more suited than the other method. Overall, it appears that a rational use of FCIM and the ROE can help a great deal in obtaining a better estimate of the human drug clearance for a wide variety of drugs. The advantages and disadvantages of these two methods are also discussed.

Quantitative structure-pharmacokinetic relationships for drug clearance by using statistical learning methods

Quantitative structure-pharmacokinetic relationships (QSPkR) have increasingly been used for the prediction of the pharmacokinetic properties of drug leads. Several QSPkR models have been developed to predict the total clearance (CL(tot)) of a compound. These models give good prediction accuracy but they are primarily based on a limited number of related compounds which are significantly lesser in number and diversity than the 503 compounds with known CL(tot) described in the literature. It is desirable to examine whether these and other statistical learning methods can be used for predicting the CL(tot) of a more diverse set of compounds. In this work, three statistical learning methods, general regression neural network (GRNN), support vector regression (SVR) and k-nearest neighbour (KNN) were explored for modeling the CL(tot) of all of the 503 known compounds. Six different sets of molecular descriptors, DS-MIXED, DS-3DMoRSE, DS-ATS, DS-GETAWAY, DS-RDF and DS-WHIM, were evaluated for their usefulness in the prediction of CL(tot). GRNN-, SVR- and KNN-developed models have average-fold errors in the range of 1.63 to 1.96, 1.66-1.95 and 1.90-2.23, respectively. For the best GRNN-, SVR- and KNN-developed models, the percentage of compounds with predicted CL(tot) within two-fold error of actual values are in the range of 61.9-74.3% and are comparable or slightly better than those of earlier studies. QSPkR models developed by using DS-MIXED, which is a collection of constitutional, geometrical, topological and electrotopological descriptors, generally give better prediction accuracies than those developed by using other descriptor sets. These results suggest that GRNN, SVR, and their consensus model are potentially useful for predicting QSPkR properties of drug leads.

truPK – human pharmacokinetic models for quantitative ADME prediction

The use of in silico prediction of absorption, distribution, metabolism and excretion (ADME) properties is gaining acceptance as a useful assessment tool for early identification of likely drug candidate failures. However, it has been difficult to locate reliable models for the prediction of human pharmacokinetics (PK) in silico Currently available methods for estimating ADME and toxicity properties, such as in vitro and animal models, are not very predictive of what is observed in the clinic. Existing in silico ADME prediction tools concentrate on physicochemical properties, such as solubility, log P, rule-of-five compliance, Caco-2 permeability, blood-brain barrier and so on, or only classify drug-like candidates as 'poor', 'medium' or 'good' for a PK parameter, without ascribing values. Although physiology-based pharmacokinetic -models can predict ADME properties, they rely on using various measured properties as input for better accuracy. Strand Genomics has developed a tool, truPK, that predicts the properties of a molecule (bioavailability, protein binding, volume of distribution, elimination half-life and absorption rate) that affect its dose and dose frequency in humans. truPK's five models built using sophisticated machine methods have predicted with > 75% accuracies in external validation sets.

Allometric scaling of pharmacokinetic parameters in drug discovery: can human CL, Vss and t1/2 be predicted from in-vivo rat data?

In a drug discovery environment, reasonable go/no-go human in-vivo pharmacokinetic (PK) decisions must be made in a timely manner with a minimum amount of animal in-vivo or in-vitro data. We have investigated the accuracy of the in-vivo correlation between rat and human for the prediction of the total systemic clearance (CL), the volume of distribution at steady state (Vss), and the half-life (t1/2) using simple allometric scaling techniques. We have shown, using a large diverse set of drugs, that a fixed exponent allometric scaling approach can be used to predict human in-vivo PK parameters CL, Vss and t(1/2) solely from rat in-vivo PK data with acceptable accuracy for making go/no-go decisions in drug discovery. Human in-vivo PK predictions can be obtained using the simple allometric scaling relationships CL(Human) approximately = 40 CL(Rat) (L/hr), Vss(Human) approximately = 200 Vss(Rat) (L), and t1/2(Human) approximately = 4 t1/2(Rat) (hr). The average fold error for human CL predictions for N = 176 drugs was 2.25 with 79% of the drugs having a fold error less than 3. The average fold error for human Vss predictions for N = 144 drugs was 1.85 with 84% of the drugs having a fold error less than 3. The average fold error for human t1/2 predictions for N = 145 drugs was 2.05 with 76% of the drugs having a fold error less than 3. Using these simple allometric relationships, the sorting of drug candidates into a low/medium/high/very high human classification scheme was also possible from rat data. Since these simple allometric relationships between rat and human CL, Vss, and t1/2 are reasonably accurate, easy to remember and simple to calculate, these equations should be useful for making early go/no-go in-vivo human PK decisions for drug discovery candidates.

Prediction of human pharmacokinetic profile in animal scale up based on normalizing time course profiles

The aim of the present study was to develop a method for predicting the concentration-time profile in humans based on pharmacokinetic data for animals. The method is based on the assumptions that concentration-time profiles of a drug are similar among species and "normalized curves" from a variety of animal species including humans can be superimposed. Normalized curves are obtained by normalizing the time axis with the MRT (mean residence time) and the concentration axis with dose/Vdss, where Vdss is the volume of distribution at steady state. The concentration-time profile in humans after intravenous injection can be simulated using the normalized curve for an animal and the predicted values of clearance (CL) and Vdss for humans. Although the general idea of our method is similar to the Dedrick plots, ours is superior in that it enables the use of predicted CL and Vdss values from any method. Our method was applied to some drugs using actual published data sets, and the assumption of the similarity of concentration-time profiles among species was found to be acceptable for these drugs. The results for the prediction of concentration-time profiles for humans were also acceptable. This method can be applied to any drug on the assumption that normalized curves from a variety of species can be superimposed.

Application of simple mathematical expressions to relate the half-lives of xenobiotics in rats to values in humans

INTRODUCTION

Previous publications from GlaxoSmithKline and University of Toledo laboratories convey our independent attempts to predict the half-lives of xenobiotics in humans using data obtained from rats. The present investigation was conducted to compare the performance of our published models against a common dataset obtained by merging the two sets of rat versus human half-life (hHL) data previously used by each laboratory.

METHODS

After combining data, mathematical analyses were undertaken by deploying both of our previous models, namely the use of an empirical algorithm based on a best-fit model and the use of rat-to-human liver blood flow ratios as a half-life correction factor. Both qualitative and quantitative analyses were performed, as well as evaluation of the impact of molecular properties on predictability.

RESULTS

The merged dataset was remarkably diverse with respect to physiochemical and pharmacokinetic (PK) properties. Application of both models revealed similar predictability, depending upon the measure of stipulated accuracy. Certain molecular features, particularly rotatable bond count and pK(a), appeared to influence the accuracy of prediction.

DISCUSSION

This collaborative effort has resulted in an improved understanding and appreciation of the value of rats to serve as a surrogate for the prediction of xenobiotic half-lives in humans when clinical pharmacokinetic studies are not possible or practicable.

A comprehensive analysis of the role of correction factors in the allometric predictivity of clearance from rat, dog, and monkey to humans

This study was conducted to comprehensively evaluate the performance of various allometric scaling methods for the prediction of human clearance. Allometric scaling was used to predict clearance for 103 compounds, for which clearance data in the rat, dog, monkey, and humans were available. Allometry was performed using all three preclinical species and with combinations of any two species. The methods employed included standard allometry and various correction factors, including brain weight, maximum lifespan potential, and glomerular filtration. Scaling was performed on all compounds universally and on segregated subsets based on allometric exponent, clearance, physicochemical property, or route of elimination. 776 allometric combinations with 27,313 individual outcomes were performed. A predicted-to-observed clearance ratio of 0.5 to twofold was preselected as the criterion for predictive success. The success rate of allometric scaling ranged from 18 to 53%; none of the correction factors resulted in substantially improved predictivity. Furthermore, none of the methods attempted in this study achieved a success rate greater than that observed by simply estimating human clearance based on monkey hepatic extraction. Prospective allometric scaling, with or without correction factors, represents a suboptimal technique for estimating human clearance based on in vivo preclinical data.