Predicting the emergence of enzalutamide resistance in prostate cancer as a consequence of memory-driven phenotypic adaptation

Cancer therapeutic resistance to targeted therapies, such as enzalutamide (Enza), is a frequent cause of disease progression and contributor to mortality. While the emergence of durable resistance mutations has been well-characterized, phenotypic changes over smaller timescales also contribute to cancer persistence. Here, we discuss our development of a continuous-time dynamic model of memory-driven cellular adaptation to study cancer adaptation. This framework models cancer growth and death, in addition to phenotypic switching that depends on a short-term history of previous drug encounters. Our model is applied in experimental collaboration to Enza-sensitive and Enza-resistant co-cultured cells under variable treatment regimens. We demonstrate that memory-driven adaptation accurately captures the mean phenotypic dynamics relative to a fixed-adaptation strategy. Using our trained model, we predict the phenotypic response of cancer cells when treated with dual Enza and P38 inhibitor, and we discuss in silico prediction of optimal dosing strategies.