By Siladitya Ray Chaudhuri, Michael B. Bolger, and Michael Lawless
Establishing a therapeutically beneficial new chemical entity (NCE) can be broadly classified into research (discovery) and development phases. Drug development is generally divided into nonclinical (animal) and clinical (human) testing stages, with regulatory approval of an investigational new drug application separating the two. In contrast, drug discovery does not require regulatory oversight and follows the general path of: target validation —> assay development —> high-throughput screening —> hit to lead —> lead optimization (LO). Promising drug candidates are selected and transitioned from late LO into nonclinical development, although the specific processes for doing so vary between organizations.
Along with potency and safety, a third key factor in determining an NCE’s viability as a potential clinical drug candidate is its ADME (absorption, distribution, metabolism, and excretion) properties. In the not-so-recent past, poor ADME properties were responsible for more clinical drug trial failures than efficacy or safety. Typically ADME properties are initially evaluated in early lead optimization with in vitro experiments, while more specific and resource-intensive in vivo exposure data is collected during later stage drug candidate optimization and selection. Both in vitro and in vivo data are used to assess for liabilities such as poor bioavailability, high clearance, potential for drug-drug interactions, etc. A more thorough mechanistic understanding of how ADME properties can be optimized has evolved in the past decade or so, which has enabled pharmaceutical scientists to select more robust NCEs. This enhanced understanding also enables earlier stage in silico modeling and simulation.
Human exposure predictions during the nonclinical development phase have also improved. Historically, compartmental pharmacokinetic (PK) methods have used allometric scaling of preclinical animal data to predict exposures in humans. In compartmental PK, the body is arbitrarily represented by either one or several theoretical compartments without specifying anatomy and physiology. These analyses, however, require in vivo data, which is often unavailable during early compound optimization. Furthermore, allometric scaling in simple compartmental models has no mechanistic basis, which can limit the predictive ability.
Physiologically based pharmacokinetic (PBPK) methods are an alternative approach to address these challenges. Although the concept of PBPK modeling is not new, its use within the pharmaceutical industry has been limited until recently following regulatory acceptance by the Food and Drug Administration. The PBPK approach utilizes anatomical and physiological parameters for either in silico/in vivo extrapolation or in vitro/in vivo extrapolation, which predicts full PK in animal species or humans. These predictions require only a compound’s in silico or in vitro ADME properties. A significant disadvantage of PBPK has been the complexity, requiring hundreds of differential equations and biopharmaceutical parameters. This shortcoming, however, has been rectified in whole-body PBPK models incorporated within several commercially available software products such as PK-Sim (Bayer Technology Services), Simcyp (Certara), and GastroPlus (Simulations Plus, Inc.).
Traditionally PBPK modeling and simulation is used in the late nonclinical to early clinical development space to support dose selection for first-in-human studies, potential drug-drug interaction effects, and possible exposure differences resulting from a change in formulation. This powerful methodology can, however, also be employed in the early to late discovery stage as discussed through in case studies in the AAPS Newsmagazine June cover story from the Drug Discovery and Development Interface section.
How significant are in silico ADMET property estimations in decision making at the interface of drug discovery and development?