Understanding glioblastoma-macrophage interactions through radiomics, transcriptome sequencing, and mathematical modeling

Glioblastoma (GBM) is the most common primary brain tumor and has a poor median overall survival of just under 15 months. To combat this heterogeneous disease, the immune system initiates an inflammatory response, where both brain-resident microglia and blood-derived macrophages work to fight the tumor. However, some immune cells are co-opted by the tumor to express immune-suppressive signals, allowing for continued tumor growth and are thereby termed ‘glioma-associated macrophages’. To better understand the spatiotemporal dynamics of the interactions between tumor cells and these two macrophage phenotypes, we proposed the Proliferation-Invasion-Macrophage (PIM) model, which is a partial differential equation model that incorporates the proliferative and invasive behavior of GBM cells, as well as populations for both ‘healthy’ and ‘glioma-associated’ macrophages. Through exploring the parameter space, we classified the various dynamics of tumor progression. To apply the model to patient data, spatially-distributed image-localized biopsies were collected from a cohort of patients and RNA sequencing was performed. Correlations between normalized RNA counts of key genetic markers (i.e. CD68, CD163, SOX2, KI67) were analyzed. Patient imaging and RNA sequencing data were then utilized to train and validate a predictive machine learning model that outputs transcriptome expression maps for the aforementioned key genetic markers. This was then used to parameterize the PIM model for each patient. In doing so, this provided us with a detailed characterization of the interactions between the GBM and macrophage populations on a patient-by-patient basis. Through gaining an understanding of the interactions between glioma cells and the macrophage phenotypes, we can work towards developing personalized immunotherapies and other immune-targeted therapeutic strategies that combat this phenomenon.