In silico ADME/T modelling for rational drug design

Medicilon's pharmacokinetics department offers the clients a broad spectrum of high quality of services in the areas of in vitro ADME, in vivo pharmacokinetics and bioanalysis services, ranging from small molecules to large molecules, such as protein and antibody. The animal species involved in our services are non-human primate, canine, mice, rat, rabbit and hamster. Meanwhile, non-human primate experimental platform and isotope platform for protein/antibody are certified by the Shanghai Government. Email:marketing@medicilon.com.cn Web:www.medicilon.com

In recent decades, in silico absorption, distribution, metabolism, excretion (ADME), and toxicity (T) modelling as a tool for rational drug design has received considerable attention from pharmaceutical scientists, and various ADME/T-related prediction models have been reported. The high-throughput and low-cost nature of these models permits a more streamlined drug development process in which the identification of hits or their structural optimization can be guided based on a parallel investigation of bioavailability and safety,along with activity. However, the effectiveness of these tools is highly dependent on their capacity to cope with needs at different stages, e.g.their use in candidate selection has been limited due to their lack of the required predictability. For some events or endpoints involving more complex mechanisms, the current in silico approaches still need further improvement. In this review, we will briefly introduce the development of in silico models for some physicochemical parameters, ADME properties and toxicity evaluation, with an emphasis on the modelling approaches thereof, their application in drug discovery, and the potential merits or deficiencies of these models. Finally, the outlook for future ADME/T modelling based on big data analysis and systems sciences will be discussed. Current pharmaceutical research and development (R&D) is a high-risk investment that is characterized by a high cost and increasing attrition rate at late-stage drug development. Balancing the risk-reward ratio and improving the productivity of R&D have always been major concerns of the pharmaceutical industry. To address this issue, several multidisciplinary approaches are required for the process of drug development, including structural biology, computational chemistry, and information technology, which collectively form the basis of rational drug design. Rational drug design refers to the process of finding new pharmaceutical compounds based on the knowledge of a biological target (Liljefors et al. 2002). Because this process always relies on computer modelling techniques (although not necessarily), it has been considered near-synonymous with the term ‘computer-aided drug design’. So far, a wide range of computational approaches have been applied to various aspects of the drug discovery and development process, and it has even been proposed that extensive use of the computational tools could reduce the cost of drug development by up to 50% . Rational drug design methods can be divided into two major classes: (1) methods for lead discovery and optimization, which often play an important role in the early state of R&D and help scientists to identify compounds with higher potency and selectivity to one or a few targets; and (2) methods for predicting compounds’ druggability, of which the aim is to prioritize lead molecules for further development by a comprehensive assessment of their therapeutic properties. The studies to identify leads involve target-to-hit and hit-to-lead processes. Corresponding computational methods include drug target prediction, virtual screening, molecular docking, scaffold hopping, allosteric versus active site modulation, and three-dimensional (3D) quantitative structure-activity relationship (QSAR) analyses . Several reviews of the development of these methods and their applications have been published. The efficiency of these processes and the quality of the generated leads can be significantly improved by the deliberate selection of those computational methods. By contrast, the process to optimize leads into a drug is more challenging. This situation can be easily understood if we roughly compare the number of newly reported active compounds with that of newly approved drugs during the same period. For example, ChEMBL is a database of a large number of bioactive molecules that were extracted from the literature. In 2012, the number of compounds in ChEMBL was 629 943, whereas this number has increased to 1 638 394 by November 2014 . Although millions of active compounds have been found, the number of new molecular entities that were approved by the US Food and Drug Administration (FDA) in recent years did not increase. In contrast, there was a slight decline in 2013 compared with 2012. There are many possible reasons for this decline; except for non-technical issues, the most relevant are the efficacy and safety deficiencies, which are related in part to absorption, distribution, metabolism and excretion (ADME assay) properties and various toxicities (T) or adverse side effects. However, the current evaluation methods for ADME/T properties are costly and time consuming and often require a large amount of animal testing, which is often inadequate when managing a large batch of chemicals. Accordingly, in-depth ADME/T scrutiny will not be performed until a limited number of candidate compounds have been identified, meaning that the major chemical scaffolds or preferred core structures have been established at that stage, for which it becomes difficult to make significant structural modifications based on the results of ADME/T evaluation. This disconnect between chemical optimization and ADME/T evaluation has caused many candidate compounds showing excellent in vitro efficacy to be dismissed due to poor ‘druggability’. For example, some compounds could not dissolve in aqueous solution or permeate across the membrane to reach the concentration needed at the required therapeutic level, and some others may exhibit a removal time that is too long or have an excessive number of metabolically unstable sites. In addition, either the compounds or their metabolites may raise toxicity and safety issues, which occasionally cannot be observed by in vitro assays or animal models. Consequently, these issues complicate the assessment of the in vivo efficacy and safety of the drug and hinder the development process. Improving R&D efficiency and productivity will depend heavily on the early assessment of the druggability of compounds. In this sense, the goal of rational drug design is to fully exploit all ADME/T profiling data to prioritise the candidates or, alternatively, to ‘fail early and fail cheap’. Because it is impractical to perform intricate and costly ADME/T experimental procedures for vast numbers of compounds, in silico ADME/T prediction is becoming the method of choice in early drug discovery. The establishment of high-quality in silico ADME/T models will permit the parallel optimization of compound efficacy and druggability properties, which is expected to not only improve the overall quality of drug candidates and therefore the probability of their success, but also to lower the overall expenses due to a reduced downstream attrition rate. In the last decade, a large number of ADME/T prediction models have been reported, and several reviews regarding the development of these models have been published. Between 2008 and 2012, the Office of Clinical Pharmacology at the FDA received 33 submissions containing physiologically based pharmacokinetic (PBPK) modelling applications for investigational new drugs (INDs) and new drug applications (NDAs). However, compared with the number of newly developed models, reports of the practical applications of these models in medicinal chemistry study or the drug discovery process are rarer. To solve these problems, some solutions have been reported regarding how to develop more effective models and where these models can be used (Gleeson & Montanari, 2012). More importantly, the outcome of ADME/T models can be maximized by intelligently integrating existing in silico, in vitro, and in vivo ADME/T data to guide drug discovery. In this review, we focused on the development of ADME/T prediction models and their future opportunities and challenges. The first section concerns the influences of physicochemical (PC) parameters on compound druggability and the development of some PC prediction models. The second section introduces the prediction models for some important properties, such as human intestinal absorption, metabolism, membrane transporters, and PBPK models. The third section relates to the prediction models for toxicities, including acute toxicity, genotoxicity, and human ether-a-go-go-related gene (hERG) toxicity. In the last section, we will discuss the future direction of in silico ADME/T modelling.