Integrating Transcriptomic Data with Mechanistic Systems Pharmacology Models for Virtual Drug Combination Trials

Area of Science:

• Computational biology
• Cancer genomics
• Pharmacology

Background:

• Mutation-targeted kinase inhibitors show limited efficacy in glioblastoma (GBM) monotherapy due to challenges like blood-brain barrier penetration and tumor heterogeneity.
• Overcoming therapeutic obstacles requires exploring complex drug combinations, which is a daunting task given the vast number of potential agents.

Purpose of the Study:

• To develop a simulation-based approach for prioritizing drug combinations against glioblastoma.
• To integrate patient-specific data with a mechanistic computational model of cancer driver pathways to predict treatment responses.

Main Methods:

• A computational model was tailored to 14 glioblastoma patients using mRNA-seq transcriptome data.
• Simulations assessed responses of heterogeneous tumor cells to three FDA-approved kinase inhibitors (bosutinib, ibrutinib, cabozantinib) with known blood-brain barrier penetration.
• The model incorporated drug binding affinities to primary and off-target kinases.

Main Results:

• Single kinase inhibitors showed minimal efficacy or counterproductive effects in simulations.
• Common genetic alterations (PTEN loss, EGFR amplification, NF1 loss) weakly correlated with drug efficacy, emphasizing the need for post-genomic approaches.
• Simulated drug combinations were predominantly cytostatic or cytotoxic, rarely both, indicating a need to combine targeted and non-targeted therapies.

Conclusions:

• Computational modeling offers a viable strategy to navigate the complexity of glioblastoma drug combinations.
• Personalized approaches accounting for kinase inhibitor promiscuity are essential for effective glioblastoma treatment.
• The developed simulation framework is broadly applicable beyond glioblastoma for prioritizing cancer drug combinations.