Collaborative evaluation of in silico predictions for high throughput toxicokinetics

Area of Science:

• Pharmacokinetics and Toxicokinetics
• Computational Chemistry and Cheminformatics
• Chemical Risk Assessment

Background:

• High-throughput toxicokinetic (HTTK) methods are crucial for addressing data gaps in chemical risk assessment.
• These methods rely on chemical-specific values derived from in vitro measurements or in silico models.
• Quantitative structure-property relationship (QSPR) models offer a computational approach to estimate these parameters.

Purpose of the Study:

• To evaluate the performance of seven QSPR models in estimating key toxicokinetic parameters: intrinsic hepatic clearance (Clint), fraction of unbound chemical in plasma (fup), and elimination half-life (t½).
• To assess the utility of QSPR-derived parameters as inputs for a high-throughput physiologically based toxicokinetic (HT-PBTK) model.
• To compare the predictive accuracy of QSPR models against in vitro measurements for HTTK applications.

Main Methods:

• Utilized seven QSPR models to predict Clint, fup, and/or t½.
• Evaluated QSPR model performance against literature time-course in vivo toxicokinetic data, primarily from rat studies.
• Performed simulations using a high-throughput physiologically based toxicokinetic (HT-PBTK) model, incorporating QSPR predictions as input parameters.
• Conducted a sensitivity analysis to identify key parameters influencing predictions of area under the curve (AUC) and steady-state concentration (Css).

Main Results:

• Estimated that using rat in vivo data to evaluate QSPR models trained on human in vitro data could inflate error estimates (RMSLE up to 0.8).
• Sensitivity analysis confirmed that Clint and fup significantly inform predictions of AUC and Css.
• AUC predictions using HTTK were estimated with RMSLE of 0.9 for in vitro measurements and 0.6-0.8 for QSPR model values.

Conclusions:

• QSPR models provide valuable predictions for HTTK input parameters, potentially yielding toxicokinetic predictions similar to in vitro measurements for novel compounds.
• The choice of evaluation data (e.g., rat vs. human) can impact error estimates for QSPR models.
• Accurate estimation of Clint and fup is critical for reliable predictions of AUC and Css in HTTK assessments.