hide
Free keywords:
-
Abstract:
In over two centuries of cancer research, we have gone from knowing its cellular origin to knowing that it can be viewed as a rogue organ that can recruit immune cells and other healthy cells. While laboratory and clinical observations accumulate rapidly, the theory that could connect all the observations requires catching up and validation. Consequently, we have seen one-off successes in hematological malignancies, but theoretical and quantitative support is required to treat solid tumors. In this thesis, I present my contribution to tools in accelerating research from conceptual to laboratory and clinical phases. Each chapter in this thesis focuses on a characteristic of cancer cells while working with tumors of different organs. I distill the critical processes within and surrounding the tumor cells to gain mechanistic underpinnings and quantify population-level dynamics. I use ordinary dierential equations and agent-based models to capture tumor cell heterogeneity and Bayesian inference to perform model selection from experimental data. The thesis aims for mechanistic insights into carcinogenesis and cancer treatment with quantitative models highlighting their value for integrative cancer research. Chapter 2 originates from observing isogenic cells change their phenotype at ranges faster than the doubling time in the laboratory. The ability to have a plastic phenotype has consequences on the dissemination and treatment of cancer. I show that contemporary treatments can be improved by altering the driving factors of phenotypic plasticity. For this, I capture the eect of microenvironmental factors in one parameter of a multi-compartment ordinary differential equation capturing heterogeneous tumor phenotypes. In Chapter 3, I explicitly model the cell progression through the cell cycle to reveal that drugs with stage-specic eects lead to delayed effects. Moreover, I show that the drug's eect is proportional to the duration of the cell cycle stage where this eect occurs. I show that agent-based models bring more realism to quantitative models while being simple. Chapter 4 uses the Bayesian approach to parameter inference, parameter identifiability, and model selection. I let different assumptions on the same phenomena compete to find a suitable model of cell-cell interactions between tumors and engineered T cells. The approach reveals non-linear dependence of treatment outcome on T cell characteristics and dose dependence of the treatment dynamics. All in all, this work explores several cancer systems unified by questions in the ecology and treatment dynamics of heterogeneous tumors. The approaches used inthis work show new ways to accelerate blackboard-to-bench translation in cancer research.
