Physiologically-based pharmacokinetic modeling to predict the clinical pharmacokinetics of monoclonal antibodies.

Accurate prediction of the clinical pharmacokinetics of new therapeutic entities facilitates decision making during drug discovery, and increases the probability of success for early clinical trials. Standard strategies employed for predicting the pharmacokinetics of small-molecule drugs (e.g., allometric scaling) are often not useful for predicting the disposition monoclonal antibodies (mAbs), as mAbs frequently demonstrate species-specific non-linear pharmacokinetics that is related to mAb-target binding (i.e., target-mediated drug disposition, TMDD). The saturable kinetics of TMDD are known to be influenced by a variety of factors, including the sites of target expression (which determines the accessibility of target to mAb), the extent of target expression, the rate of target turnover, and the fate of mAb-target complexes. In most cases, quantitative information on the determinants of TMDD is not available during early phases of drug discovery, and this has complicated attempts to employ mechanistic mathematical models to predict the clinical pharmacokinetics of mAbs. In this report, we introduce a simple strategy, employing physiologically-based modeling, to predict mAb disposition in humans. The approach employs estimates of inter-antibody variability in rate processes of extravasation in tissues and fluid-phase endocytosis, estimates for target concentrations in tissues derived through use of categorical immunohistochemical scores, and in vitro measures of the turnover of target and target-mAb complexes. Monte Carlo simulations were performed for four mAbs (cetuximab, figitumumab, dalotuzumab, trastuzumab) directed against three targets (epidermal growth factor receptor, insulin-like growth factor receptor 1, human epidermal growth factor receptor 2). The proposed modeling strategy was able to predict well the pharmacokinetics of cetuximab, dalotuzumab, and trastuzumab at a range of doses, but trended towards underprediction of figitumumab concentrations, particularly at high doses. The general agreement between model predictions and experimental observations suggests that PBPK modeling may be useful for the a priori prediction of the clinical pharmacokinetics of mAb therapeutics.