Abstract Glioblastoma Multiforme (GBM) is the most common of all gliomas with a median survival 14-18 months, despite aggressive treatment regimens. However, glioma is a disease that encompasses more than just cancer cells. In fact, many studies have shown that glioma can alter the brain microenvironment in ways that promote tumor survival and propagation. Within this brain tumor microenvironment is a diversity of cell types. One of particular group of interest is glioma associated microglia and macrophages (GAMMs), an important component of the immune cells in the brain. As a result, immunotherapy is emerging as a promising method to treat cancer; however, we are not able to identify early response or predict who will respond. While biopsies are the most reliable way to assess the immunological landscape within the tumor, we are limited both spatially and temporally in the number of biopsies we can obtain, particularly for brain tumor patients. The heterogeneity of the tumor-immune landscape across patients suggests that a patient-specific approach will be required to accurately assess each patient’s individual tumor-immune environment and the evolution thereof. As part of the Parent Grant, we will use non-invasive imaging, image-guided biopsies, computational modeling, and artificial intelligence to bridge spatial and temporal scales and predict the abundance of glioma associated microglia/macrophages (GAMMs) comprising each magnetic resonance image (MRI) at the voxel level. Linking the MRI to the biological heterogeneity using radiomics approaches provides an opportunity to individualize our understanding of the tumor-immune environment. Also in recent years, research has looked into how drug treatment is able to activate GAMMs to take on immunoreactive phenotypes. For this proposed supplement we will characterize myeloid – glioma cell interactions in response to immunogenic cell death induced by the chemotherapy drug topotecan. This will be done using MRI localized biopsies as well as in vitro co-culture systems, providing an additional therapeutically relevant context in which to study cellular response and signaling. The study will make use of single cell RNA sequencing to identify activational states of immune cells, and will provide us with another aspect of microglia and macrophage biology that can be incorporated into the model generated in the Parent Grant.

NARRATIVE While our inability to identify early response to immunotherapy in brain tumor patients poses a serious challenge to advancing cancer care and research, genomic information from image-guided biopsies, linked to non-invasive imaging through artificial intelligence and mechanistic modeling, is a useful way to assess tumor- associated immunological activity. Current methods to assess response to immunotherapy could benefit from this image-based predictive understanding of the tumor-immune landscape. We propose to evaluate immuno- therapeutic response based on changes depicted by a tumor-immune predictive map.