In vitro–in vivo extrapolation of hepatic clearance: using virtual experiments to identify a plausibly influential source of inaccuracies

It is understood that entanglement of different methodological and biological factors is responsible for IVIVE uncertainties, but untangling them using wet‐lab and correlation methods has proven difficult. It seems likely that the relative importance of entangled factors will often be use‐case specific, making the problem even more daunting. We are assessing the scientific utility of using virtual experiments to explore for sources of influential extrapolation inaccuracies arising from simplifying assumptions often made about the hepatic disposition of chemical entities, ceteris paribus . We anticipate that, when a significant source is identified, mitigating methods can be developed. Consequently, the prospect of utilizing IVIVEs to improve applications to model‐based personalized dosing methods may increase. To be scientifically useful, objects of virtual experiments, such as virtual mice, livers (vLivers) and hepatocytes (vHPCs) must meet demanding representational requirements. For example, mechanism components are concrete. The organization of components produce a strong behavioral analogy to the referent biology. Within reasonable constraints, events occurring during execution are biomimetic. Measurements of phenomena are quantitatively similar to referent wet‐lab measurements. Smith et al. (PMID27984590) demonstrate the scientific usefulness of virtual experiments by falsifying the prevailing explanation for the characteristic early pericentral pattern of acetaminophen‐induced hepatic necrosis. A liver model is an essential feature of IVIVE methods. The well‐stirred compartment and parallel tubes models are prevalent. Both models abstract away lobular architecture, although parallel‐tube models often describe sequential metabolism. The working hypothesis during extrapolations is that, by adequately accounting for factors responsible for the drug's intrinsic clearance, lobular architectural influences can be abstracted away. We challenge that hypothesis by experimenting on three types of virtual mice, each having a different vLiver substructure but identical vHPCs (no zonation). The lobular architecture of control vLivers (from Smith et al.) and hepatic lobules are strongly analogous. Two test‐vLivers map to the parallel tube model. They are structural variants of Control vLivers that differ only in opportunities that vCompounds have to enter (or not) a vHPC. During in vitro experiments, the exposure rate of each hepatocyte to the compound under study is the same. That is not the case in vivo or for the three vLivers. We measure each vHPC's exposure rates using two vCompounds. They represent extremes of the virtual intrinsic clearance spectrum. Measured periportal and pericentral differences in exposure rates among the three vLiver types are > 10x. That evidence is sufficient to challenge the above working hypothesis. Using liver models that do not ignore lobular structural influences may provide a strategy to reduce IVIVE inaccuracies while providing a context for future inclusion of individualizable intralobular influences. Support or Funding Information UCSF Biosystems Group This abstract is from the Experimental Biology 2019 Meeting. There is no full text article associated with this abstract published in The FASEB Journal .