Ecological and Evolutionary Dynamics of Cellular Immunotherapies

Cellular therapies represent a new frontier for quantitative systems modeling. Chimeric Antigen Receptor (CAR) T cell therapy is a novel treatment for refractory lymphoma or leukemia patients that can improve response rate over five-fold. This cellular therapy uses patients’ own T cells to genetically engineer them and specifically target antigen-presenting tumor cells. During expansion ex vivo, before re-injecting the cellular drug, patients undergo the crucial step of depletion of most normal lymphocytes by chemotherapy. The loss of normal lymphocytes is not permanent but provides the engineered cells with a temporal advantage.

We integrate mathematical modeling, based on first principles of nonlinear stochastic birth-death processes, with clinical data and in vitro models of CAR T cell reaction to stress, in order to better understand CAR T expansion, CAR T interactions with the (inflammatory) tumor microenvironment, and CAR killing of tumor cells. These processes are all impacted by feedback from the tumor environment and the T cell population heterogeneity. We ask three important questions:

(1) How does the CAR T phenotype (naïve, memory, and effector) determine response? (2) Why does cellular immunotherapy fail in some patients, and how does therapy evasion evolve? (3) How do inflammatory cytokines signals determine toxicity?

To answer these questions, we combine data novel in vitro experiments and longitudinal patient data of inflammatory cytokines and circulating tumor DNA, to inform models of T cell and tumor cell kinetics. Since many of the important interactions, such as T cell expansion, decay, and tumor killing, are highly transient, this system is driven by non-equilibrium mechanics.