Deep reinforcement learning identifies personalized intermittent androgen deprivation therapy for prostate cancer

Abstract

The evolution of drug resistance leads to treatment failure and tumor progression. Intermittent androgen deprivation therapy (IADT) helps responsive cancer cells compete with resistant cancer cells in intratumoral competition. However, conventional IADT is population-based and ignores the heterogeneous phenotypes of individual patients. To address this challenge, we developed a time-varied, mixed-effect, and generative Lotka-Volterra (tM-GLV) model to account for the heterogeneity of the evolution mechanism and the pharmacokinetics of individual patients. Then, we proposed a reinforcement learningenabled individualized IADT framework, namely, I2ADT, to learn the patient-specific tumor dynamics and derive the optimal drug administration policy. Experiments with clinical trial data demonstrated that the proposed I2ADT can significantly prolong the time to progression of prostate cancer patients with reduced cumulative drug dosage. This research elucidates the application of reinforcement learning techniques to identify personalized adaptive cancer therapy.

Mathematical Model-Driven Deep Learning Enables Personalized Adaptive Therapy

“…In addition to parameter uncertainty, tumor behavior may not align with the virtual patient model (1), as patient tumors may exhibit heterogeneity in the rules governing their dynamics. Hence, we assess the DRL framework's robustness to variations in the underlying tumor dynamics by evaluating our framework on a modified Lotka–Volterra model introduced by Lu and colleagues ( 44 ) and reproduced in Supplementary Section S6, with model parameters given in Supplementary Table S2. This model uses a diminishing competition term, such that intratumoral competition decreases over time, with a modified progression criterion based on the growth of the resistant subpopulation alone.…”

“…The approach we have presented here is not the first DRL framework to tackle AT. The work of Lu and colleagues ( 44 ) used a different underlying prostate cancer model, but trained the network on both the PSA and the sensitive and resistant populations; although PSA is easily measured in a blood draw, there is no direct way to measure the numbers of sensitive and resistant cells in a real patient. Although they were also able to produce significantly improved outcomes for a single patient, the generalizability of our approach, and the clear path to clinical practice, make our implementation somewhat more robust.…”