Image analysis and agent-based modeling of tumor-immune interactions in the glioblastoma microenvironment

Glioblastoma is a highly aggressive and deadly brain cancer with no current treatment options available that can achieve remission. One potential explanation for minimally effective treatments is the ability of gliomas to take advantage of processes within the body’s immune system to infiltrate the tumor microenvironment with myeloid-derived suppressor cells (MDSCs). These cells hinder the ability of T cells to effectively destroy tumor cells. In this work, we create an agent-based model to simulate the interactions of these three cell types. Our model also includes three substrates to influence cell behavior: oxygen, a T-cell chemoattractant, and an MDSC chemoattractant. To validate our model and quantify cell clustering patterns in glioblastoma, we use pair and cross-pair correlation functions (PCF and cross-PCF, respectively) on location data for each cell type. This data is extracted from cross-sectional tumor images of cellular biomarkers using image analysis techniques. We then conduct the same PCF and cross-PCF analysis on the cell location outputs from parameter testing trials of our ABM. These methods allow us to fit a linear model using the least squares method to the tumor image analysis results and evaluate how well the tests match the experimental data. From these results, we use our model to identify the impact of our tested parameters on cell clustering behavior within the glioblastoma tumor and its microenvironment.