Proteomic signatures to predict drug response in cancer

Abstract Precision medicine in oncology requires accurate prediction of drug response to guide personalized treatment decisions. We developed a pharmacoproteomics platform that links kinome activity in liver cancer cell lines to kinase inhibitor drug response. Preliminary results indicate that our KI-Predictor’s ability to identify dysregulated kinase signaling is highly dependent on cellular context. To examine and optimize our predictive model, we will use samples of normal human liver tissue and other inputs and evaluate its performance using additional liver cell lines and the Genomics of Drug Sensitivity in Cancer (GDSC) datasets. In parallel, we propose to evaluate our KI-Predictor by profiling kinome activity in clinical liver cancer samples, selecting candidate kinase inhibitors for testing in organotypic slice cultures derived from each individual patient’s tumor. By comparing the predicted drug response with the observed response in the organotypic slice culture models, the accuracy and reliability of the KI-Predictor model can be evaluated. This validation step enables the assessment of the model's ability to capture the complexities of the tumor microenvironment, including interactions between cancer cells, stromal cells, and the extracellular matrix.