Designing QSARs for Parameters of High-Throughput Toxicokinetic Models Using Open-Source Descriptors

Area of Science:

• Environmental chemistry
• Computational toxicology
• Pharmacokinetics

Background:

• Intrinsic metabolic clearance (Clint) and unbound fraction (fup) are crucial for toxicokinetic (TK) models.
• Experimental data for Clint and fup are scarce for many chemicals.
• High-throughput screening requires reliable in silico prediction methods.

Purpose of the Study:

• Develop open-source quantitative structure-activity relationship (QSAR) models for Clint and fup.
• Provide reliable in silico predictions for diverse chemical classes.
• Evaluate the utility of these QSAR models in risk-based prioritization.

Main Methods:

• Developed open-source QSAR models for Clint and fup.
• Utilized model predictions as inputs for TK component of risk-based prioritization (Bioactivity/Exposure Ratios - BERs).
• Applied models to a subset of the Tox21 screening library (6484 chemicals).

Main Results:

• In silico predictions yielded similar proportions of chemicals with BER < 1 as in vitro data (17.5%).
• High concordance (90.4%) was observed between in silico and in vitro parameters for classifying chemicals with BER < 1 or > 1.
• QSAR models demonstrated suitability for prioritizing chemicals lacking in vitro TK data.

Conclusions:

• Open-source QSAR models for Clint and fup offer reliable in silico predictions.
• These models can effectively support risk-based chemical prioritization strategies.
• The developed QSARs are valuable tools for assessing chemicals with limited experimental toxicokinetic data.