Linking in vitro–in vivo extrapolations with physiologically based modeling to inform drug and formulation development

Quantitative in vitro–in vivo extrapolations (IVIVE) have been applied extensively to predict drug metabolism and transporter kinetics (Sodhi & Benet, 2021; Wood et al., 2017; Zamek‐Gliszczynski et al., 2013). However, initially, IVIVE were mainly based on mean in vitro data, giving no estimation of between or within subject variability (BSV/WSV) and, therefore, have a limited ability to address the extremes of risk in real patients or replicate scenarios such as bioequivalence. In other words, ADME properties were estimated considering an average subject, if such a subject exists, under a constant non‐variant condition for all the physiology and biology. Integrating IVIVE with PBPK modeling enables predictions in virtual cohorts including estimation of BSV (Rostami‐Hodjegan & Tucker, 2007), and some recent advances were also made to address the WSV that are essential elements of virtual bioequivalence studies (Bego et al., 2022). Historically, Monte Carlo simulations are used when developing population PBPK models which commonly do not consider the parameter’s inter‐dependencies. However, it is imperative to mechanistically incorporate covariates in these models and to employ correlated Monte Carlo simulations instead (Jamei, 2016; Jamei et al., 2009). In doing so, the predicted PK properties are inherently affected by the relevant covariates (Rostami‐Hodjegan & Tucker, 2007). However, translation of in vitro ADME properties within PBPK models is associated with uncertainties related not only to the knowledge gaps in system parameters, but also to the translatability (scaling) of drug parameters (Rostami‐Hodjegan, 2018). In this context, Mao and co‐workers applied PBPK modeling to predict human PK and assessed model performance retrospectively using clinical data for 18 Genentech compounds (Mao et al., 2023). Currently, the main regulatory application of PBPK modeling is to investigate enzyme‐ or transporter‐mediated drug–drug interactions (Zhang et al., 2020). In this special issue, Vijaywargi and co‐workers compared the performance of static and PBPK models to predict transporter‐mediated DDIs. The authors also discussed the usefulness of endogenous biomarkers, the use of empirical scaling factors, enzyme–transporter interplay, and acceptance criteria for model validation to meet the regulatory expectations (Vijaywargi et al., 2023). The separation between system and drug parameters is an important feature of PBPK modeling with respect to extrapolating results between species and populations. In this special issue Alsmadi and Alzughoul integrate preclinical data and a IVIVE‐PBPK model to extrapolate drug elimination in renally impaired patients (Alsmadi & Alzughoul, 2023). In turn, Liang and co‐workers successfully applied PBPK modeling to optimize azithromycin dosing to pediatrics (Liang et al., 2023) and Zhang and co‐workers developed a PBPK‐based dose decision framework to inform clinical trials with a novel anesthetic agent in special populations (Zhang et al., 2023). PBPK models can also be linked to other modeling techniques. Probably, the most common integration is between PBPK and pharmacodynamics (PD) models to assess the time course of drug effect. In this special issue, a PBPK model was linked to a tumor growth inhibition model to identify the optimal dose of selective estrogen receptor degraders in humans (Ganti et al., 2023). Furthermore, real world evidence (RWE) analysis can be used to verify PBPK‐based predictions, leveraging the predict–learn–confirm paradigm (Grillo et al., 2023). This is a powerful two‐way approach in which the PBPK model can be used to mechanistically confirm signals detected by RWE analysis as well as RWE being able to confirm PBPK predictions. Interestingly, IVIVE‐PBPK approaches are relatively new to the biopharmaceutics field but some successful applications have been reported by different groups (Cristofoletti et al., 2019; Pathak et al., 2017). In this special issue Martins and co‐ workers integrated artificial neural network and IVIVE‐PBPK modeling to inform the development of sustained‐release formulations (Martins, 2023). In the past two decades, PBPK modeling has become the fastest growing subfield in pharmacokinetics (El‐Khateeb et al., 2021). In fact, a recent invited review in this journal (El‐Khateeb et al., 2021) estimated the growth rate for PBPK modeling publications in the past 20 years to be greater than 40‐fold. This comes as no surprise, given that the mechanistic nature of PBPK models lends itself to a multitude of applications which enable them to surrogate clinical trials and to provide informative insights into novel drug disposition. Dr Rostami‐Hodjegan’s group expanded their previous analysis towards identifying patterns in the selection of different PBPK modeling tools (Aldibani et al., 2023; Rajput et al., 2023). The US FDA Modernization Act 2.0 of 2022 introduced the term “nonclinical tests” in the legal framework, emphasizing the relevance of cell‐based assays, microphysiological systems, bioprinted or computer models as alternatives to animal testing in drug discovery and development. In this context, we envision a potential role of using microphysiological systems to inform IVIVE‐PBPK models in the quantitative clinical pharmacology field. Microphysiological systems can be viewed as an innovative technology that has the potential to enhance the understanding of physiology, pathology, and