Personalized Boolean network dynamic profiling and machine learning approaches to identify combinatorial treatments for prostate cancer.

Introduction: Prostate cancer is the most commonly diagnosed non-cutaneous cancer in males and the second leading cause of male cancer death in the United States. Androgen deprivation therapy to block androgen receptor (AR) signaling is part of the standard of care. However, approximately 30%-40% of patients experience biochemical recurrence (BCR) after primary therapy and progress to a more lethal form of castration-resistant prostate cancer for which there currently is no curative therapy. This study focused on patients at risk of progressing to BCR with the aim of identifying novel combinatorial targets that can be used with AR inhibition to provide greater therapeutic response. Methods: Personalized Boolean network models were developed using The Cancer Genome Atlas and in-house data to integrate mutations, copy number alterations, and RNA-seq data. Asynchronous stochastic network simulations in MaBoSS (Markovian Boolean Stochastic Simulator) were used to identify potential drug targets in combination with AR inhibition, which were cross-checked using a machine learning algorithm and tested with a cytotoxicity assay across 4 prostate cancer cell lines. Results: Discussion: Dynamic profiling of personalized Boolean networks in treatment-naïve prostate cancer subjects with and without BCR identified potential drug targets in combination with AR antagonists. Future xenograft studies are needed to confirm