A three-dimensional agent-based model of glioblastoma with applications towards in vitro and in vivo therapy simulation

Glioblastoma is an aggressive and highly infiltrative malignant brain tumor with a poor survival rate. Treatment options for patients are limited and have remained stagnant for almost two decades despite dozens of clinical trials. Advancements in genomic characterization methods allowed for the classification of three different subtypes of glioblastoma: proneural, classical, and mesenchymal. The distinct genetic alterations in these subtypes lead to downstream differences in cell migration speed and immune system activation status. Despite these behavioral differences, glioblastomas are treated using a uniform standard of care regardless of tumor subtype. To identify subtype-dependent features influencing tumor progression and response to treatments, we developed a three-dimensional agent-based model of glioblastoma. The model incorporates rules governing cell proliferation, migration, and intercellular interactions between cancer cells and cytotoxic T cells. The model was parameterized using data from a genetically engineered mouse model of proneural and mesenchymal glioblastomas. First, we modeled tumor development in both subtypes and found that simulated mesenchymal tumors were more diffuse and had greater tumor volumes. Simulations of anti-migratory and T cell-based immunotherapy treatments demonstrated differential efficacy between the two tumor subtypes, highlighting the need to account for subtype in glioblastoma therapy development. Next, we modified the simulator to replicate experimental results from a three-dimensional in vitro cytotoxicity assay that measures CAR T cell efficacy against glioblastoma. We adjusted T cell parameters to capture patterns of CAR T cell trafficking, proliferation, and persistence observed in this experimental system. The proliferation rate and death rate of CAR T cells were configured to dynamically change based on local IL-2 concentrations, leading to the formation of proliferating clusters of CAR T cells as observed in vitro. We simulated CAR T cell therapies against glioblastoma spheroids and were able to replicate the patterns of killing observed at different effector to target ratios using this system. By parameterizing rules governing the behavior of individual cells in our model, we were able to capture large-scale dynamics of treatment response in simulated in vitro and in vivo systems. This research enhances our understanding of the mechanisms driving therapy failure in glioblastoma and provides a strategy for predicting effective future treatments.

Author dc:creator

• Hall, Riley

Language dc:language.iso

en

Handle dc:identifier.uri

https://hdl.handle.net/11299/279115

OAI identifier oai:identifier

oai:conservancy.umn.edu:11299/279115