Collaborative evaluation of in silico predictions for high throughput toxicokinetics

High throughput toxicokinetic (HTTK) methods address chemical risk assessment data gaps but require chemical-specific values that can be obtained by in vitro measurements or in silico models. In this study, seven quantitative structure property relationship (QSPR) models were used to estimate intrinsic hepatic clearance (Cl_int), fraction of chemical unbound in plasma (f_up), and/or TK elimination half-life (t_½). Performance of the QSPR models was evaluated using literature time-course in vivo TK data, mainly from rats. Simulations of the in vivo data were made with a high throughput physiologically based TK (HT-PBTK) model using the different QSPR model predictions as inputs. We estimate that using rat in vivo data to evaluate QSPR models trained on human in vitro measured data might inflate error estimates by as much as root mean squared log₁₀ error (RMSLE) 0.8. A sensitivity analysis showed that Cl_int and f_up parameters inform predictions of area under the curve (AUC) and steady-state concentration (C_ss). We estimate that AUC can be predicted by HTTK with RMSLE 0.9 using in vitro measurements and 0.6-0.8 using QSPR model values. We anticipate that, for some novel compounds, QSPRs for HTTK input parameters will give predictions of TK similar to those based on in vitro measurements.