QSAR models using a large diverse set of estrogens.

Area of Science:

• Environmental Chemistry
• Toxicology
• Computational Chemistry

Background:

• Endocrine disruptors (EDs) pose risks to human and animal health.
• Thousands of chemicals require safety testing, necessitating efficient prioritization methods.
• Estrogen receptor (ER) binding is a key mechanism for estrogenic EDs.

Purpose of the Study:

• To evaluate quantitative structure-activity relationship (QSAR) models for predicting chemical binding to the ER.
• To develop a rapid, integrated system for prioritizing chemicals for ED testing.
• To assess the performance of CoMFA and HQSAR models for predicting relative binding affinities (RBAs).

Main Methods:

• Developed a four-phase integrated system for ED screening.
• Constructed and compared CoMFA and HQSAR models using a dataset of 130 chemicals.
• Utilized leave-N-out cross-validation and external test sets to validate model performance.

Main Results:

• The CoMFA model achieved a cross-validated r² of 0.91 and q²LOO of 0.66.
• Incorporating a phenol indicator improved the CoMFA model's q²LOO to 0.71.
• External validation demonstrated the CoMFA model's utility with q²pred values of 0.71 and 0.62.

Conclusions:

• The CoMFA model provides accurate quantitative predictions of ER binding affinities.
• This model, as part of a tiered system, enables efficient prioritization of chemicals for ED testing.
• The developed QSAR approach significantly aids in managing large chemical datasets for safety assessment.