P2.C055: Characterization of the tumor microenvironment in diffuse midline glioma (DMG) for D3 in vitro model building

Background: Diffuse midline glioma (DMG) is an aggressive pediatric brain cancer. Few comprehensive tumor characterization studies have been conducted due to the relative scarcity of tissue samples. However, it is well known that the surrounding microenvironment, and particularly the cells of the normal brain, influence tumor outcomes. Here, we investigate the cell type distribution among patient autopsy samples, comparing the primary tumor, metastatic tumors, and normal brain tissue.

Methods: Patient samples were obtained at Children’s National through a postmortem autopsy program. Initially, 10 primary tumor samples were collected from 10 individual patients. An additional 30 patient samples were collected from 10 distinct patients, consisting of a primary tumor, a metastatic site, and normal tissue from each. All patients were diagnosed with DMG between the ages of two and thirteen. Brain sections were embedded in paraffin, sectioned into 5 μm thick slices, and stained using fluorescent immunohistochemistry. Targets included astrocytes (ALDH1L1), microglia (IBA1), neurons (NeuN), blood vessels (CD31), and tumor cells (H3K27M). Images were captured using the VS200 whole slide scanner. Automatic cell detection was conducted using QuPath, and statistical analysis was performed in R Studio.

Results: From initial primary tumor IHC analysis, microglia appear to be present in slightly higher ratios than astrocytes, though the relationship is not statistically significant. (p=0.9107). Furthermore, neurons appear in lower numbers compared to microglia as well as astrocytes (p = 0.0002 and p = 0.0023). This can be visualized in Figure 1. In addition, we analyzed intra-patient heterogeneity between three distinct tissue sites: primary tumor, metastatic tumor, and normal tissue. We looked at 10 different patients, and each patient had samples from the three distinct tissue sites. The thirty samples were then stained for the same cellular markers. When looking at microglia, there is a statistically significant difference between the primary tumor and the metastatic tumor, as well as the metastatic tumor and the normal tissue (p = 0.0019, p = 0.0001). Furthermore, when looking at neurons, there is a statistically significant difference between primary and metastatic tumors (p = 0.0220). This can be visualized in Figure 2.

Conclusion: Using this tumor composition data, we plan to create a 3D gel in vitro cell culture model of DMG. In collaboration with individuals at Children’s National and Columbia, we will use this model in conjunction with focused ultrasound and drug testing.