A multispecies framework for modeling adaptive immunity and immunotherapy in cancer

Predator-prey theory is commonly used to approximately describe tumor growth in the presence of selective pressure from the adaptive immune system. These interactions are mediated by the tumor immunopeptidome (what the tumor "shows" the body) and the T-cell receptor (TCR) repertoire (how well the body "sees" cancer cells). Importantly, both tumor and T cell populations are dynamic and in competition with each other and their fraternal lineages. In particular, the immunopeptidome comprises neoantigens which can be gained and lost throughout tumorigenesis and treatment. Heterogeneity in the immunopeptidome is predictive of poorer survival and response to T-cell dependent immunotherapy in some tumor types, suggesting the TCR repertoire is unable to support a fully polyclonal response against every neoantigen. Whether between-lineage competition among T cells plays a role, and in what contexts, is unknown; moreover, longitudinal TCR profiling studies are expensive and logistically complex to conduct. In silico models may offer an inexpensive way to interrogate these phenomena ex ante and deepen our understanding of the tumor-immune axis. Here, we construct and calibrate a predator-prey-like model to preclinical and clinical data to describe tumor growth and immunopeptidome diversification. Simultaneously, we model the expansion of neoantigen-specific T cell lineages and their consumption of both lineage-specific and shared, lineage-agnostic resources. This predator-prey-like framework accurately described clinically observed immunopeptidomes as well as correlates and consequences of response to immunotherapy, including immunoediting. The model was also suitable for exploring treatment of tumors with varying growth and mutation rates.