Combination treatment optimization using a pan-cancer pathway model

Area of Science:

• Computational biology
• Systems biology
• Pharmacology

Background:

• Developing effective combination therapies for complex diseases like cancer is challenging due to tumor heterogeneity and numerous drug options.
• Mechanistic ordinary differential equation-based pathway models can predict molecular-level treatment responses but are underutilized for therapy discovery.

Purpose of the Study:

• To leverage a large-scale pan-cancer signaling pathway model for identifying novel combination therapies.
• To optimize drug combinations for individual and heterogeneous cancer cell lines, minimizing proliferation and side effects.
• To develop optimized sequential treatment plans for enhanced therapeutic benefits.

Main Methods:

• Utilized a state-of-the-art pan-cancer signaling pathway model.
• Employed the Covariance Matrix Adaptation Evolution Strategy (CMA-ES) algorithm.
• Integrated a scalable sampling scheme for truncated Gaussian distributions using a Hamiltonian Monte-Carlo method.

Main Results:

• Identified candidate novel combination therapies for individual and heterogeneous cancer cell lines.
• Demonstrated the optimization of sequential drug treatment plans.
• Showcased the potential for minimizing cancer cell proliferation and drug dosage.

Conclusions:

• The developed computational method effectively identifies promising combination therapies by integrating pathway models and advanced optimization algorithms.
• This approach offers a scalable and adaptable framework for discovering novel treatments for cancer and potentially other complex diseases.
• Optimized sequential treatments provide additional therapeutic advantages over static combination therapies.