Simulating Heterogeneity in Deterministic Models of Prostate Cancer Response to Immunotherapy with Standing Variations Modeling

Advanced, hormonally refractive prostate cancer is typically treated with docetaxel, a chemotherapeutic compound that inhibits cell division. However, this treatment eventually fails due to onset of resistance. Multiple mechanisms driving docetaxel resistance have been identified and several drugs targeting these mechanisms are in various stages of clinical trial, including immunotherapy in the form of a vaccine. However, optimizing strategies to overcome such resistance remains a critical challenge because the problem is inherently multiscale due to characteristic variability at the subcellular, cellular and individual levels. In this talk, I present a simple dynamical systems model of prostate cancer response to immunotherapy. I then introduce our novel modeling approach, Standing Variations Modeling, which exploits uncertainty and variability in data to inform the probability distributions - rather than specific values - from which model parameters most likely arise. This differs from traditional modeling approaches that only use static or mean expression levels and cellular responses, thereby ignoring the significant variance that exists across cell populations as well as individuals being treated. Sampling from these posterior distributions allows us to generate a virtual cohort of individuals on which in silico clinical trials are conducted to predict optimal dosing combinations and subpopulations that would benefit most from such an intervention.