Abstract BACKGROUND Diffuse Midline Glioma (DMG) is a rare and aggressive pediatric brain cancer with no chance of survival, highlighting a critical need for therapy development. Disease representative models can aid the search for effective treatments, but human preclinical models that reflect the unique developmental features and anatomical environment of DMG, are so far lacking. Latest research on DMG suggest a fetal neurodevelopmental origin with a stem cell-like cellular profile. Here, we developed a novel pontine hindbrain model to create de novo H3K27M DMG and applied it to study early tumorigenesis and immunotherapy response. MATERIAL AND METHODS Timely sequenced morphogens were applied to developing cerebral organoids, to create a new pontine patterned brain organoid model. Bulk sequencing and 3D imaging over a period of 16 weeks were used to determine ideal patterning conditions, confirm the correct brain regionality and reconstruct cellular developmental dynamics. Introducing H3K27M with common pontine mutations P53 and PDGFRA-D842V via electroporation resulted in de novo H3K27M DMG, which subsequently was characterized based on WHO-histopathological criteria and single cell sequencing. Exploiting this new human DMG model, we applied barcode-based genetic lineage tracing paired with single cell sequencing, to delineate tumorigenesis, and anti-GD2 chimeric antigen receptor (CAR) T cell therapy to investigate CAR T cell responses. RESULTS Our novel pontine hindbrain organoid model inherently gives rise to all relevant macroglia and pons-specific neurons, resembling the same developmental dynamics as seen in humans. Moreover, de novo DMG introduced in these pontine organoids strongly resembles patient cancer, including its cellular human-specific heterogeneity and invasive nature, outperforming existing gold standard PDX and cell line models. As a first model for investigating DMG, the applied barcoded lineage-tracing delineates cancerous transforming precursor states and how they contribute to the diverse lineage among known DMG cancer cell populations. Finally, administration of CAR T cells recreates treatment outcomes observed in patients and demonstrates a high level of CAR T cell heterogeneity. CONCLUSION We generated a bona fide DMG pontine hindbrain organoid model, the first human in vitro model for this specific region of the brain. The matching resemblance to patient tumor paired with the relevant healthy brain environment gives this model the potential for new insights into DMG early tumorigenesis and microenvironmental impact, as well as next generation therapy development.