Investigation of Adverse Events Associated with Predicted GnRHR Agonists Using the FAERS Database

Area of Science:

• Pharmacology
• Computational Toxicology
• Drug Safety

Background:

• Gonadotropin-releasing hormone receptor (GnRHR) agonists are therapeutically important but have incompletely understood adverse event profiles.
• Characterizing potential adverse events for GnRHR agonists is crucial for drug development and patient safety.

Purpose of the Study:

• To develop and apply a machine learning model to predict GnRHR agonist activity in drugs listed in the FDA Adverse-Event Reporting System (FAERS).
• To identify potential safety signals associated with predicted GnRHR agonist activity using FAERS data.

Main Methods:

• A machine learning model (BalancedRandomForest) was trained using Tox21 GnRHR agonist activity data and molecular descriptors.
• The model predicted GnRHR agonist activity for 5523 FAERS-listed drugs, with 1191 drugs meeting applicability domain criteria.
• FAERS data (2004-2024) were analyzed for adverse events using reporting odds ratios (RORs) and Fisher's exact test for significant associations.

Main Results:

• 367 drugs were predicted to have GnRHR agonist activity.
• 330 unique adverse event terms (PTs) met statistical criteria, with significant associations found for respiratory, thoracic and mediastinal disorders, infections and infestations, and hepatobiliary disorders.
• Predicted GnRHR agonists showed disproportionate reporting of adverse events in these categories.

Conclusions:

• The developed machine learning framework effectively predicts GnRHR agonist activity and associated potential adverse events.
• Findings suggest a link between predicted GnRHR agonist activity and increased reporting of respiratory, infectious, and hepatobiliary adverse events.
• This approach can aid in hypothesis generation and safety signal prioritization in drug development and postmarketing surveillance.