Quantitative structure–pharmacokinetic relationships

Development of a predictive algorithm for the efficacy of half-life extension strategies

A challenge in development of peptide and protein therapeutics is rapid elimination from the body, necessitating frequent dosing that may lead to toxicities and/or poor patient compliance. To solve this issue, there has been great investment into half-life extension of rapidly eliminated drugs using approaches such as albumin binding, fusion to albumin or Fc, or conjugation to polyethylene glycol. Despite clinical successes of half-life extension products, no clear relationship has been drawn between properties of drugs and the pharmacokinetic parameters of their half-life extended analogues. In this study, non-compartmentally derived pharmacokinetic parameters (half-life, clearance, volume of distribution) were collected for 186 half-life extended drugs and their unmodified parent molecules. Statistical testing and regression analysis was performed to evaluate relationships between pharmacokinetic parameters and a matrix of variables. Multivariate linear regression models were developed for each of the three pharmacokinetic parameters and model predictions were in good agreement with observed data with r2 values for each parameter being: half-life: 0.879, clearance: 0.820, volume of distribution: 0.937. Significant predictors for each parameter included the corresponding pharmacokinetic parameter of the parent drug and descriptors related to molecular weight. This model represents a useful tool for prediction of the potential benefits of half-life extension.

Another string to your bow: machine learning prediction of the pharmacokinetic properties of small molecules

INTRODUCTION

Prediction of pharmacokinetic (PK) properties is crucial for drug discovery and development. Machine-learning (ML) models, which use statistical pattern recognition to learn correlations between input features (such as chemical structures) and target variables (such as PK parameters), are being increasingly used for this purpose. To embed ML models for PK prediction into workflows and to guide future development, a solid understanding of their applicability, advantages, limitations, and synergies with other approaches is necessary.

AREAS COVERED

This narrative review discusses the design and application of ML models to predict PK parameters of small molecules, especially in light of established approaches including in vitro-in vivo extrapolation (IVIVE) and physiologically based pharmacokinetic (PBPK) models. The authors illustrate scenarios in which the three approaches are used and emphasize how they enhance and complement each other. In particular, they highlight achievements, the state of the art and potentials of applying machine learning for PK prediction through a comphrehensive literature review.

EXPERT OPINION

ML models, when carefully crafted, regularly updated, and appropriately used, empower users to prioritize molecules with favorable PK properties. Informed practitioners can leverage these models to improve the efficiency of drug discovery and development process.

How effective are ionization state-based QSPKR models at predicting pharmacokinetic parameters in humans?

Optimizing the pharmacokinetics (PK) of a drug candidate to support oral dosing is a key challenge in drug development. PK parameters are usually estimated from the concentration-time profile following intravenous administration; however, traditional methods are time-consuming and expensive. In recent years, quantitative structure-pharmacokinetic relationship (QSPKR), an in silico tool that aims to develop a mathematical relationship between the structure of a molecule and its PK properties, has emerged as a useful alternative to experimental testing. Due to the complex nature of the various processes involved in dictating the fate of a drug, the development of adequate QSPKR models that can be used in real-world pre-screening situations has proved challenging. Given the crucial role played by a molecule's ionization state in determining its PK properties, this work aims to build predictive QSPKR models for PK parameters in humans using an ionization state-based strategy. We divide a high-quality dataset into clusters based on ionization state at physiological pH and build global and ion subset-based 'local' models for three major PK parameters: plasma clearance (CL), steady-state volume of distribution (VDss), and half-life (t1/2). We use a robust methodology developed in our lab entitled 'EigenValue ANalySis' that accounts for the stereospecificity in drug disposition and use the support vector machine algorithm for model building. Our findings suggest that categorizing compounds in accordance with ionization state does not result in improved QSPKR models. The narrow ranges in the endpoints along with redundancies in the data adversely affect the ion subset-based QSPKR models. We suggest alternative approaches such as elimination route-based models that account for drug-transporter interactions for CL and chemotype-specific QSPKR for VDss.

Characterizing the pharmacokinetics and biodistribution of therapeutic proteins: an industry white paper

Characterization of the pharmacokinetics (PK) and biodistribution of therapeutic proteins (TPs) is a hot topic within the pharmaceutical industry, particularly with an ever-increasing catalog of novel modality TPs. Here, we review the current practices, and provide a summary of extensive cross-company discussions as well as a survey completed by International Consortium for Innovation and Quality (IQ consortium) members on this theme. A wide variety of in vitro, in vivo and in silico techniques are currently used to assess PK and biodistribution of TPs, and we discuss the relevance of these from an industry perspective, focusing on PK/PD understanding at the preclinical stage of development, and translation to human. We consider that the 'traditional in vivo biodistribution study' is becoming insufficient as a standalone tool, and thorough characterization of the interaction of the TP with its target(s), target biology, and off-target interactions at a microscopic scale are key to understand the overall biodistribution at a full-body scale. Our summary of the current challenges and our recommendations to address these issues could provide insight into the implementation of best practices in this area of drug development, and continued cross-company collaboration will be of tremendous value. Significance Statement The Innovation & Quality Consortium (IQ) Translational and ADME Sciences Leadership Group (TALG) working group for the ADME of therapeutic proteins evaluates the current practices, recent advances, and challenges in characterizing the PK and biodistribution of therapeutic proteins during drug development, and proposes recommendations to address these issues. Incorporating the in vitro, in vivo and in silico approaches discussed herein may provide a pragmatic framework to increase early understanding of PK/PD relationships, and aid translational modelling for first-in-human dose predictions.

In Silico Prediction of Human Renal Clearance of Compounds Using Quantitative Structure-Pharmacokinetic Relationship Models

Renal clearance (CLr) plays an essential role in the elimination of drugs. In this study, 636 compounds were obtained from various sources to develop in silico models for the prediction of CLr. Stepwise multiple linear regression and random forest regression methods were employed to build global models and local models according to ionization state or net elimination pathways. The local models toward compounds undergoing different net elimination pathways showed good predictive power: the geometric mean fold error was close to 2, indicating the clearance of most compounds could be predicted within a 2-fold error range. Six classification methods were used to construct classification models. However, the performance of these classification models was less than satisfactory, and the mean accuracy of the top five models in test sets was 0.65. Moreover, qualitative analysis of physicochemical profiles between compounds undergoing different net elimination pathways revealed that compounds with higher lipophilicity tended to be reabsorbed more easily and showed lower CLr, while compounds with higher values of polar descriptors tended to secrete more easily and showed higher CLr.

Predicting volume of distribution with decision tree-based regression methods using predicted tissue:plasma partition coefficients

BACKGROUND

Volume of distribution is an important pharmacokinetic property that indicates the extent of a drug's distribution in the body tissues. This paper addresses the problem of how to estimate the apparent volume of distribution at steady state (Vss) of chemical compounds in the human body using decision tree-based regression methods from the area of data mining (or machine learning). Hence, the pros and cons of several different types of decision tree-based regression methods have been discussed. The regression methods predict Vss using, as predictive features, both the compounds' molecular descriptors and the compounds' tissue:plasma partition coefficients (Kt:p) - often used in physiologically-based pharmacokinetics. Therefore, this work has assessed whether the data mining-based prediction of Vss can be made more accurate by using as input not only the compounds' molecular descriptors but also (a subset of) their predicted Kt:p values.

RESULTS

Comparison of the models that used only molecular descriptors, in particular, the Bagging decision tree (mean fold error of 2.33), with those employing predicted Kt:p values in addition to the molecular descriptors, such as the Bagging decision tree using adipose Kt:p (mean fold error of 2.29), indicated that the use of predicted Kt:p values as descriptors may be beneficial for accurate prediction of Vss using decision trees if prior feature selection is applied.

CONCLUSIONS

Decision tree based models presented in this work have an accuracy that is reasonable and similar to the accuracy of reported Vss inter-species extrapolations in the literature. The estimation of Vss for new compounds in drug discovery will benefit from methods that are able to integrate large and varied sources of data and flexible non-linear data mining methods such as decision trees, which can produce interpretable models. Graphical AbstractDecision trees for the prediction of tissue partition coefficient and volume of distribution of drugs.

Quantitative structure-pharmacokinetic relationships for the prediction of renal clearance in humans

Renal clearance (CLR), a major route of elimination for many drugs and drug metabolites, represents the net result of glomerular filtration, active secretion and reabsorption, and passive reabsorption. The aim of this study was to develop quantitative structure-pharmacokinetic relationships (QSPKR) to predict CLR of drugs or drug-like compounds in humans. Human CLR data for 382 compounds were obtained from the literature. Step-wise multiple linear regression was used to construct QSPKR models for training sets and their predictive performance was evaluated using internal validation (leave-one-out method). All qualified models were validated externally using test sets. QSPKR models were also constructed for compounds in accordance with their 1) net elimination pathways (net secretion, extensive net secretion, net reabsorption, and extensive net reabsorption), 2) net elimination clearances (net secretion clearance, CLSEC; or net reabsorption clearance, CLREAB), 3) ion status, and 4) substrate/inhibitor specificity for renal transporters. We were able to predict 1) CLREAB (Q(2) = 0.77) of all compounds undergoing net reabsorption; 2) CLREAB (Q(2) = 0.81) of all compounds undergoing extensive net reabsorption; and 3) CLR for substrates and/or inhibitors of OAT1/3 (Q(2) = 0.81), OCT2 (Q(2) = 0.85), MRP2/4 (Q(2) = 0.78), P-gp (Q(2) = 0.71), and MATE1/2K (Q(2) = 0.81). Moreover, compounds undergoing net reabsorption/extensive net reabsorption predominantly belonged to Biopharmaceutics Drug Disposition Classification System classes 1 and 2. In conclusion, constructed parsimonious QSPKR models can be used to predict CLR of compounds that 1) undergo net reabsorption/extensive net reabsorption and 2) are substrates and/or inhibitors of human renal transporters.

The use of pseudo-equilibrium constant affords improved QSAR models of human plasma protein binding

PURPOSE

To develop accurate in silico predictors of Plasma Protein Binding (PPB).

METHODS

Experimental PPB data were compiled for over 1,200 compounds. Two endpoints have been considered: (1) fraction bound (%PPB); and (2) the logarithm of a pseudo binding constant (lnKa) derived from %PPB. The latter metric was employed because it reflects the PPB thermodynamics and the distribution of the transformed data is closer to normal. Quantitative Structure-Activity Relationship (QSAR) models were built with Dragon descriptors and three statistical methods.

RESULTS

Five-fold external validation procedure resulted in models with the prediction accuracy (R²) of 0.67 ± 0.04 and 0.66 ± 0.04, respectively, and the mean absolute error (MAE) of 15.3 ± 0.2% and 13.6 ± 0.2%, respectively. Models were validated with two external datasets: 173 compounds from DrugBank, and 236 chemicals from the US EPA ToxCast project. Models built with lnKa were significantly more accurate (MAE of 6.2-10.7 %) than those built with %PPB (MAE of 11.9-17.6 %) for highly bound compounds both for the training and the external sets.

CONCLUSIONS

The pseudo binding constant (lnKa) is more appropriate for characterizing PPB binding than conventional %PPB. Validated QSAR models developed herein can be applied as reliable tools in early drug development and in chemical risk assessment.