Tuning HERG out: antitarget QSAR models for drug development

Area of Science:

• Computational chemistry
• Drug discovery
• Pharmacology

Background:

• Inhibition of hERG K+ channels by non-cardiovascular drugs can cause fatal heart arrhythmias.
• hERG safety testing is a mandatory FDA requirement, necessitating predictive tools for early drug development stages.

Purpose of the Study:

• To develop and validate robust quantitative structure-activity relationship (QSAR) models for predicting hERG channel blockage.
• To identify potential hERG blockers and non-blockers within marketed drugs using developed models.

Main Methods:

• Utilized the largest publicly available dataset (11,958 compounds) from the ChEMBL database.
• Developed and validated QSAR models using four descriptor types and four machine-learning techniques, adhering to OECD guidelines.
• Applied validated models to screen the World Drug Index (WDI) database.

Main Results:

• Achieved high classification accuracies (0.83-0.93) for distinguishing hERG blockers from non-blockers on an external dataset.
• Derived structure-activity relationship (SAR) rules to guide optimization of hERG blockers into non-blockers.
• Identified putative hERG blockers and non-blockers among existing marketed drugs.

Conclusions:

• Developed predictive QSAR models reliably identify hERG channel blockers and non-blockers.
• The models and a freely accessible web server can aid the scientific community in drug discovery and safety assessment.
• SAR insights facilitate structural modifications to mitigate hERG channel interactions.