Drug Repurposing against KRAS Mutant G12C: A Machine Learning, Molecular Docking, and Molecular Dynamics Study

Area of Science:

• Oncology
• Pharmacology
• Computational Chemistry

Background:

• KRAS^G12C mutations are prevalent in non-small-cell lung cancer (NSCLC).
• Existing KRAS^G12C inhibitors face challenges with efficacy in resistant tumors.

Purpose of the Study:

• To discover novel KRAS^G12C inhibitors using artificial intelligence (AI).
• To repurpose FDA-approved covalent drugs for KRAS^G12C-targeted therapy.

Main Methods:

• Machine learning models, specifically extreme gradient boosting (XGBoost), were developed.
• Models were trained and validated to predict KRAS^G12C inhibitory activity.
• 67 FDA-approved covalent drugs were screened for potential inhibitory effects.

Main Results:

• The XGBoost models achieved high predictive accuracy (validation = 0.85, Q²_Ext = 0.76).
• Afatinib, neratinib, and zanubrutinib were identified as potential covalent inhibitors.
• Afatinib demonstrated the highest predicted inhibitory concentration (pIC₅₀).
• Afatinib, neratinib, and zanubrutinib showed favorable binding site interactions and distance deviations.

Conclusions:

• Afatinib, neratinib, and zanubrutinib are promising drug candidates for KRAS^G12C-targeted NSCLC treatment.
• AI-driven drug repurposing offers a valuable strategy for identifying new cancer therapies.