Predictions of Hepatic Disposition Properties Using a Mechanistically Realistic, Physiologically Based Model □ S Li Yan, Shahab Sheihk-Bahaei, Sunwoo Park, Glen E. P. Ropella, and C. Anthony Hunt The UCSF/UCB Joint Graduate Group in Bioengineering, University of California, Berkeley and San Francisco, California (L.Y., S.S.-B., C.A.H.); Department of Bioengineering and Therapeutic Sciences, the BioSystems Group, the University of California, San Francisco, California (S.P., G.E.P.R., C.A.H.); and Tempus Dictum, Inc., Eagle Creek, Oregon (G.E.P.R.) Received October 28, 2007; accepted January 24, 2008 ABSTRACT: Quantitative mappings were established between drug physico- chemical properties (PCPs) and parameter values of a physiolog- ically based, mechanistically realistic, in silico liver (ISL). The ISL plugs together autonomous software objects that represent he- patic components at different scales and levels of detail. Microar- chitectural features are represented separately from the mecha- nisms that influence drug metabolism. The same ISL has been validated against liver perfusion data for sucrose and four cationic drugs: antipyrine, atenolol, labetalol, and diltiazem. Parameters sensitive to drug-specific PCPs were tuned so that ISL outflow profiles from a single ISL matched in situ perfused rat liver outflow profiles of all five compounds. Quantitative relationships were then established between the four sets of drug PCPs and the corre- sponding four sets of PCP-sensitive, ISL parameter values; those relationships were used to predict PCP-sensitive, ISL parameter values for prazosin and propranolol given only their PCPs. Rela- tionships were established using three different methods: 1) a simple linear correlation method, 2) the fuzzy c-means algorithm, and 3) a simple artificial neural network. Each relationship was used separately to predict ISL parameter values for prazosin and propranolol, given their PCPs. Those values were applied in the ISL used earlier to predict the hepatic disposition details for each drug. Although we had only sparse data available, all predicted disposi- tion profiles were judged reasonable (within a factor of 2 of refer- ent profile data). The order of precision, based on a similarity measure, was 3 > 2 > 1. The in silico liver (ISL) (Hunt et al., 2006; Yan et al., 2007) (Fig. 1) is a first-generation example of synthetic, discrete, physiologically based analogs that are intended for refining, exploring, and testing hypotheses about the details of hepatic drug disposition. Autonomous components represent spatial aspects of hepatic organization and function. Different drugs can be represented and studied simulta- neously or separately. Each ISL component can interact uniquely with any drug-representing object that enters its local environment. The consequences of simulated systemic and local interactions can be measured and studied simultaneously, analogous to how wet-lab ex- periments are conducted. In Yan et al. (2008), the simulated hepatic disposition of atenolol, antipyrine, labetalol, and diltiazem, along with coadministered su- crose, used only one parameterized ISL structure for all compounds. A subset of components interacted differently with the particular compounds. Monte Carlo ISL variants simulated compound-specific outflow profiles that matched referent profiles. The results supported two hypotheses. 1) The mappings in Fig. 2 between ISL components and corresponding liver components were sufficiently realistic for the stated model use. 2) The simulated drug-ISL component interaction events mapped to corresponding hepatic disposition events. A goal for this project has been to discover and verify ISL counterparts to the relationships between drug physicochemical properties (PCPs) and pharmacokinetic (PK) mechanisms and use them to make predictions. A vision motivating research on this class of models is identical to one that has motivated development of traditional physiologically based PK models: by “accounting for the causal basis of the observed data, . . . the possibility exists for efficient use of limited drug-specific data [italics added] to make reasonably accurate predictions as to the pharmacokinetics of specific compounds, both within and between species, as well as under a variety of conditions” (Rowland et al., 2004). PCP-sensitive, physiologically based PK model parameters necessarily conflate features and properties of the biology (aspects of histology and others) with drug PCPs. In doing so, there is a risk that “the causal basis” becomes obscured because of the conflated biolog- ical features that were especially influential in causing some property of the data. Interconnections between sinusoids might be such a feature. In the ISL, because we have built a collection of mechanisms from finer-grained components, we have precise control over confla- tion, yet the causal basis is still there in the drug-component interac- tion logic (axioms and rules). Our expectation has been that at some This work was supported in part by grants (to C.A.H.) and fellowships (to S.S.-B. and S.P.) provided by the CDH Research Foundation and abstracted in part from the Ph.D. dissertation presented by L.Y. to the Graduate Division, University of California, Berkeley, CA. Article, publication date, and citation information can be found at http://dmd.aspetjournals.org. doi:10.1124/dmd.107.019067. □ S The online version of this article (available at http://dmd.aspetjournals.org) contains supplemental material. ABBREVIATIONS: ISL, in silico liver; PCP, physicochemical property; PK, pharmacokinetic; CV, central vein; SS, sinusoidal segment; PV, parameter value; CC, correlation coefficient; FCMA, fuzzy c-means algorithm; ANN, artificial neural network; SM, similarity measure. 0090-9556/08/3604-759–768$20.00 DRUG METABOLISM AND DISPOSITION Vol. 36, No. 4 Copyright © 2008 by The American Society for Pharmacology and Experimental Therapeutics 19067/3323256 DMD 36:759–768, 2008 Printed in U.S.A. 759 http://dmd.aspetjournals.org/content/suppl/2008/01/28/dmd.107.019067.DC1.html Supplemental material to this article can be found at: at ASPET Journals on June 2, 2016 dmd.aspetjournals.org Downloaded from