Microsome composition-based model as a mechanistic tool to predict nonspecific binding of drugs in liver microsomes

The purpose of this study was to investigate the ability of the microsome composition-based model to predict the unbound fraction determined in vitro in microsomal incubation system (fuinc ). Another objective was to make a comparative assessment between the proposed mechanistic method and three empirical methods published in the literature, namely the models of Austin et al. (2002, Drug Metab Dispos 30:1497-1503), Turner et al. [2007, Drug Metab Rev 38(S1):162], and Halifax and Houston (2006, Drug Metab Rev 34:724-726), which are based solely on physicochemical properties. The assessment was confined by the availability of measured fuinc data in rat and human at diverse microsomal protein concentrations for 132 compounds. The proposed microsome composition-based model can be viewed as a combination of two distinct processes, namely the nonspecific binding to neutral lipids and the ionic binding to acidic phospholipids. Across methods, the maximum success rate in predicting fuinc of all compounds was 98%, 91%, and 84% with predictions falling within threefold, twofold, and 1.5-fold error of the observed fuinc , respectively. The statistical analyses suggest that the prediction models are more effective at computing fuinc (i) for rat as compared with human, and (ii) for acids and neutral drugs as compared with strong basic drugs. In addition, on the basis of the comparisons made using all datasets, the method that made use of microsome composition data compares well with those methods that relied solely on physicochemistry. The sensitivity analysis demonstrated the importance of the compound properties and physiological parameters reflective of specific mechanistic determinants relevant to prediction of fuinc values of drugs. Overall, the results obtained with our proposed model demonstrate a significant step toward the development of a generic and mechanistic model of fuinc for liver microsomes, which should provide rationale extrapolation procedures of hepatic clearance using a physiologically-based pharmacokinetics (PBPK) modeling approach.