Predicting and Optimizing Synergistic Drug Combinations for Breast Cancer Treatment Using Machine Learning.

Area of Science:

• Computational biology
• Pharmacology
• Machine learning in oncology

Background:

• Breast cancer treatment relies on effective drug combinations.
• Identifying synergistic drug combinations is crucial for improving therapeutic efficacy.
• Traditional methods for drug synergy screening are time-consuming and resource-intensive.

Purpose of the Study:

• To leverage machine learning for accurate prediction of drug synergy scores in breast cancer.
• To identify and rank highly synergistic drug combinations for potential therapeutic use.
• To accelerate the drug discovery process for breast cancer combination therapies.

Main Methods:

• Utilized machine learning models (XGBoost, Random Forest, CatBoost) to analyze breast cancer drug combination data.
• Quantified drug interactions using four synergy metrics: ZIP, Bliss, Loewe, and HSA.
• Evaluated model performance using normalized root mean squared error (NRMSE) and Pearson correlation coefficient.

Main Results:

• XGBoost demonstrated superior performance, achieving an NRMSE of 0.074 and Pearson correlation of 0.90 for the Bliss synergy model.
• Identified top-ranking synergistic combinations, including Ixabepilone+Cladribine, SN 38 Lactone+Pazopanib, and Decitabine+Tretinoin.
• Validated top combinations based on high synergy scores and supporting biological mechanisms.

Conclusions:

• Machine learning effectively predicts synergistic drug combinations for breast cancer.
• This computational approach accelerates screening and reduces experimental burden.
• The findings provide a valuable tool for guiding future in vitro and in vivo validation of novel combination therapies.