Limited Time Offer

Claim a 25% discount on all eLearning courses (including credentials) with code ELEARN25.

Offer is valid from March 10-31. Public courses excluded from promo. 

(620g) Uncovering Hidden and Observable Features That Contribute to Tumor Aggression and Progression

Hazboun, A. - Presenter, Northwestern University
Prybutok, A., Northwestern University
Yu, J., Northwestern University
Bagheri, N., University of Washington
Tumor aggression is a consequence of subcellular dysregulation, cell decision processes, and the microenvironment. These features collectively drive many hallmarks of cancer, including cell lifespan, enhanced proliferative capacity, increased growth rates, and ability to invade healthy tissue. Additionally, genetic heterogeneity within this dynamic microenvironment enables tumors to build resistance to therapeutic interventions. Treatment mechanisms exploit various microscopic features, such as nutrient uptake rates, cell division rates, and the extent of cell cycle synchronization throughout the tumor. Some of these features can be measured through imaging techniques (e.g., PET scans to understand metabolic demands within a tumor), while others cannot (e.g., heterogeneity between tumor cells and number of subpopulations). We bridge this gap of observables using a computational model--specifically an agent-based model--to understand how cell-level and population-level features contribute to tumor aggression and progression.

Agent-based models (ABMs) are a “bottom-up” computational framework that predict emergent cell population dynamics from individual cell agent rules. We use this framework to simulate tumor microenvironments by integrating heterogeneous tumor cell agents with healthy cell agents in a dynamic spatiotemporal environment that comprises dynamic vasculature. The ABM provides accurate, high-resolution data describing the behavior of individual cells at every timestep that cannot be captured in vitro or in vivo [Yu et al., Frontiers 2020]. We use the established model to generate simulations that vary cell agent characteristics and their decision rule parameters, providing a landscape of data describing emergent tumor cell population dynamics. Through use of various machine learning algorithms, we uncover key intervention strategies--both at the individual cell and population levels--that contribute to tumor progression and aggression. Our work will inform clinically relevant design features that could prove as effective targets for treatments that minimize tumor aggression and progression.