Development of QSAR-based two-stage prediction model for estimating mixture toxicity.

Area of Science:

• Environmental Chemistry
• Toxicology
• Computational Chemistry

Background:

• Conventional mixture toxicity models like concentration addition (CA) and independent action (IA) have limitations.
• The two-stage prediction (TSP) model integrates CA and IA but requires known modes of toxic action (MoA).
• Estimating mixture toxicity is challenging when MoAs of components are unknown.

Purpose of the Study:

• To develop and evaluate a quantitative structure-activity relationship-based TSP (QSAR-TSP) model.
• To predict mixture toxicity without prior knowledge of individual chemical MoAs.
• To integrate chemical structure similarity for improved mixture toxicity assessment.

Main Methods:

• Developed a QSAR-TSP model incorporating clustering methods based on chemical structural similarity.
• Applied computerized analysis and clustering to group mixture constituents.
• Utilized Random Forest analysis to determine the importance of molecular descriptors.

Main Results:

• The QSAR-TSP model demonstrated significant prediction power for mixture toxicity.
• Clustering methods based on structural similarity effectively categorized mixture components.
• The model successfully estimated mixture toxicity in the absence of MoA information.

Conclusions:

• The QSAR-TSP model offers a robust alternative to conventional methods when MoAs are unknown.
• Chemical structural information and clustering are valuable for predicting mixture toxicity.
• This approach enhances the ability to assess risks posed by chemical mixtures.