Identifying drug sensitivity of multifocal primary prostate cancer

Abstract

The pronounced intra-patient variability in multifocal primary prostate cancer (PCa) has curtailed the efficacy of current treatment options. Patient-derived organoids (PDOs) have emerged as a pivotal model for functional testing due to their capability to retain the histopathological and molecular characteristics of parental tissues, allowing timely acquisition of drug response outcomes. In our study, employing twin biopsies from multiple lesions with matched PDO models in vitro, we investigated the molecular heterogeneity of PCa, and how it is linked to in vitro PDO pharmacological heterogeneity. Our functional testing approach leverages PDOs to screen standard-of-care treatment and FDA-approved compounds for other malignancies, aiming to repurpose their use in PCa and explore alternatives to androgen deprivation therapy. By integrating gene expression data from parental tissue with drug response results from PDOs, we have established a transcriptomics-based drug prediction models. The machine learning-based prediction model can predict the experimental PDO response to a specific drug, for the majority of screened drugs. This study offers a preclinical approach to potentially procure drug prediction outcomes and validate them in PDO models, as a prior step to clinical trial investigations or for selection of targeted therapeutic options.