Abstract | There is a concerted effort to provide a predictive morphological distinction between relatively indolent and lethal forms of prostate cancer. Data from patient biopsies suggest that both the tumour cell phenotypes and the composition of the stromal compartment can be predictive of the clinical significance of prostate tumors even if the biological mechanisms relating phenotype to function are still unclear. We have implemented a mathematical model, using hybrid cellular automata (HCA), that recapitulates key interactions in nascent tumor foci between prostate tumor cells and stroma. The computer simulations demonstrate how stochastic interactions between tumor cells and tumor stroma may lead to a structural suppression of tumor growth, modest proliferation, or unopposed tumor growth. The model incorporates key aspects of prostate tumor progression including net pleiotropic growth factor activity, net matrix degrading enzyme (MDE) activity, and stromal activation. As a result of this computatinal work, cancer biologists at Vanderbilt University and Pathologists at Baylor College of Medicine have carried out experiments in vivo and tissue culture that corroborate some of the main findings from the simulations. The implications of the model underscore the need for quantitative experimental measurements and integration with computational modelling that could eventually lead to more accurate diagnoses and treatments of prostate cancer. |