Modeling stochastic phenotypic adaptation to predict enzalutamide resistance in prostate cancer

Therapeutic resistance to second-generation androgen receptor antagonists, such as enzalutamide (Enza), is a common complication in patients with advanced prostate cancer (PCa). Durable resistance mutations have been well-studied in response to long-term treatment. However, far less is known about phenotypic changes in sensitive cells that lead to persistence over smaller timescales. Recent experimental efforts can accurately track the growth dynamics of both sensitive and resistant cell types in response to a variety of temporally variable drug dosing schedules. We develop and apply a stochastic decision-making model to study the dynamics of Enza-resistant and Enza-sensitive cells. In this model, we assume that sensitive and resistant cells are capable of dynamically adjusting their phenotype based on a memory of prior environmental encounters with a time-varying drug treatment policy. We find that large memory sizes (two times the drug cycling period) generate growth dynamics that are most consistent with experimentally observed rates. By applying our model, we can identify the characteristic timescale of adaptation, which provides insights for devising more effective drug dosing schedules. Such modeling can be utilized to forecast how adaptive systems react to environmental alternations and can also be used to propose optimized therapeutic interventions.