Establishing in vitro–in vivo correlation for antibody drug conjugate efficacy: a PK/PD modeling approach

Abstract

The objective of this manuscript was to establish in vitro–in vivo correlation (IVIVC) between the in vitro efficacy and in vivo efficacy of antibody drug conjugates (ADCs), using a PK/PD modeling approach. Nineteen different ADCs were used to develop IVIVC. In vitro efficacy of ADCs was evaluated using a kinetic cell cytotoxicity assay. The cytotoxicity data obtained from in vitro studies was characterized using a novel mathematical model, parameter estimates from which were used to derive an in vitro efficacy matrix for each ADC, termed as ‘in vitro tumor static concentration’ (TSC_{in vitro}). TSC_{in vitro} is a theoretical concentration at continuous exposure of which the number of cells will neither increase nor decrease, compared to the initial cell number in the experiment. The in vivo efficacy of ADCs was evaluated using tumor growth inhibition (TGI) studies performed on human tumor xenograft bearing mice. The TGI data obtained from in vivo studies was characterized using a PK/PD model, parameter estimates from which were used to derive an in vivo efficacy matrix for each ADC, termed as ‘in vivo tumor static concentration’ (TSC_{in vivo}). TSC_{in vivo} is a theoretical concentration if one were to maintain in the plasma of a tumor bearing mouse, the tumor volume will neither increase nor decrease compared to the initial tumor volume. Comparison of the TSC_{in vitro} and TSC_{in vivo} values from 19 ADCs provided a linear and positive IVIVC. The Spearman’s rank correlation coefficient for TSC_{in vitro} and TSC_{in vivo} was found to be 0.82. On average TSC_{in vivo} was found to be ~ 27 times higher than TSC_{in vitro}. The reasonable IVIVC for ADCs suggests that in vitro efficacy data was correctly able to differentiate ADCs for their in vivo efficacy. Thus, IVIVC can be used as a tool to triage ADC molecules in the discovery stage, thereby preventing unnecessary scaling-up of ADCs and waste of time and resources. An ability to predict the concentration of ADC that is efficacious in vivo using the in vitro data can also help in optimizing the experimental design of preclinical efficacy studies. As such, the novel PK/PD modeling method presented here to establish IVIVC for ADCs holds promise, and should be evaluated further using diverse set of cell lines and anticancer agents.