Predicting carcinogenicity of organic compounds based on CPDB

Area of Science:

• Toxicology
• Computational Chemistry
• Drug Discovery

Background:

• Chemical carcinogenicity prediction is crucial for human health and safety.
• Quantitative Structure-Activity Relationship (QSAR) methods offer a computational approach to predict chemical toxicity.
• The Carcinogenic Potency Database (CPDB) provides valuable data for developing predictive models.

Purpose of the Study:

• To develop robust QSAR models for predicting the carcinogenicity of organic compounds in rats and mice.
• To investigate the influence of molecular structure (presence of rings) and animal sex on carcinogenicity prediction.
• To assess model performance based on OECD principles, ensuring reliability and accuracy.

Main Methods:

• Utilized QSAR methodologies and DRAGON descriptors to build predictive models.
• Developed eight localized models, classifying data by target organ (liver), species (rat/mouse), sex, and molecular structure (presence of rings).
• Assessed model fitting ability, robustness, and predictive power using external validation and OECD principles.

Main Results:

• All developed models demonstrated good predictivity, with external predictive coefficients ranging from 0.711-0.906 and overall coefficients greater than 0.8.
• Analysis of standardized regression coefficients provided insights into the mechanisms of carcinogenesis.
• A trend indicating higher tolerance to carcinogens in female rats and mice compared to males was observed.

Conclusions:

• The developed QSAR models are effective tools for predicting chemical carcinogenicity in rodents.
• Animal sex is a significant factor influencing carcinogenicity, with females exhibiting greater tolerance.
• These models can aid in prioritizing chemicals for further testing and contribute to risk assessment strategies.