Optimal Cancer Evasion in a Dynamic Immune Microenvironment Sculpts Immunogenicity

The human T cell repertoire plays a central role in recognizing and eliminating immune threats, including cancer. Despite improvements in patient outcomes via T cell immunotherapies, which aim to enhance anti-tumor immune recognition, complete remission is limited by cancer immune evasion. We introduce a minimal stochastic model for studying the dynamics of adaptive immune evasion by quantifying the interaction between an adaptive immune system recognizing tumor-associated antigens on the surface of dominant cancer clones. We solve for the unique optimal cancer evasion strategy using stochastic dynamic programming and demonstrate that this policy results in increased cancer evasion rates when compared to a passive, fixed strategy. Our foundational model links the likelihood and temporal dynamics of cancer evasion to features of the immune microenvironment, where tumor immunogenicity is determined by balancing cancer adaptation and host recognition.

We predict that the distribution of aggressive cancer immunogenicity following immune escape varies widely based on the recognition efficiency of the adaptive immune system and the average arrival rate of new tumor-associated antigens. In contrast with a passive strategy, optimal evaders navigating fluctuating immune environments demonstrate substantial diversity in their predicted immunogenicity following escape and clinical detection. Moreover, these evolutionary trajectories agree with empirical observations and together explain effects of the tumor-immune microenvironment on the generation of immunogenically hot or cold tumors, a direct correlate of immunotherapeutic efficacy.