Human neural organoid modeling of diffuse midline glioma captures the complexity of patient tumors

Diffuse Midline Glioma H3K27-altered (DMG) is an extremely aggressive and lethal childhood brain cancer that grows within the midline structure of the brain. Current treatment options are only palliative, making DMG in desperate need for therapeutic breakthroughs. One of the major challenges limiting the study of DMG is the lack of reliable preclinical models. In-vivo mouse models are expensive and technically challenging and in-vitro cell culture models lack the essential components of tumor microenvironment (TME) needed to recapitulate the complex biology of these tumors. Scalable human planar neural organoids (PNOs) with multi-cellular make-up can serve as a cost effective and reliable model system to capture DMG biology and allow effective species matched drug testing in-vitro. Using 3 separate DMG patient derived xenograft (PDX) cell lines, we spatially profiled a novel scalable human iPSC-derived PNO system containing neurons, functional astrocytes and microglia using the NanoString GeoMx spatial transcriptomics system. We found that all three cell lines interact with and integrate into the human PNOs, demonstrating favorable growth conditions in a complex co-culture. Across spatially resolved regions of interest (ROI’s) tumor cells individually interact with microglia and astrocytes and transcriptomic profiling of these mixed cell ROI’s shows differences in the genetic signatures of both the normal cells (microglia/astrocytes) and the tumor cells. When compared to biopsies obtained directly from DMG patients, DMG cells within PNOs correlate strongly at both transcriptomic and proteomic levels. The multi-cellular PNOs also enabled drug target bystander toxicity screening not possible in a traditional tumor cell only monoculture. This study provides a proof-of-concept for scalable PNO modeling for DMG and underscores the translational relevance of this model system.