Abstract Glioblastoma is the most devastating brain cancer with dismal prognosis both in adults and children and accounts for 16% of primary brain tumors. Despite aggressive therapeutic approaches, the majority of patients die within one year after diagnosis. Current in vivo modeling of glioblastoma involves xenografts either of cultured cell lines or human biopsies transplanted into mice. These models do not faithfully recapitulate the complex biology/pathogenesis of cancer. Here we used human pluripotent embryonic stem cells (or young children fibroblast-driven induced pluripotent stem cells) to create 3D structures with histology reminiscent of human fetal brain (the so-called cerebral organoids). Cerebral organoids combine the accurate multilineage differentiation and physiology of a developing brain with the easiness of manipulation of an in vitro system. We engineered these mini-brains to develop tumors by forced expression of combinations of epigenetic and genetic hits that have previously been shown to drive pediatric glioblastoma multiform (GMB). These include the expression of a mutant histone H3 variant H3F3A (mutated at K27M or G34R) or a gain of function of a mutant form of isocitrate dehydrogenase (IDH1R132H) in combination with the loss of ATRX histone remodeling factor and/or p53 tumor suppressor. Lentiviral transfection was used to express combination of these genetic and epigenetic hits into cerebral organoids. The mutated organoids exhibited tumorigenic growth that was histologically characterized. For example, the expression of H3K27M in combination with loss of both p53 and ATRX showed a remarkable dysplasia reminiscent of primary glioblastoma. Moreover, these organoids exhibited tumor rosette structures resembling those found in ependymoma. In contrast, organoids expressing K27M alone did not show such structures. Interestingly, IDH1R132H combined with p53 and ATRX loss resulted in different tumor morphology, suggestive of a possibility to model oncogenic dependence in pediatric GBM. Cerebral organoids could also be used as a model to study tumor invasion starting with glioma stem-like cells (GSCs, which are believed to be the tumor-initiating cells in GBM). Interestingly, the GSC cell line 923 (which does not invade mouse brain in vivo) invaded and formed a focal lesion in a cocultured organoid. The organoid-based cancer model presented here carries a promising potential to uncover the initiation and pathogenesis of pediatric tumors early in the developing brain. Note: This abstract was not presented at the conference. Citation Format: Amin Ismail, Demis Balamatsias. A cerebral organoid model of pediatric glioma [abstract]. In: Proceedings of the AACR Special Conference on the Advances in Pediatric Cancer Research; 2019 Sep 17-20; Montreal, QC, Canada. Philadelphia (PA): AACR; Cancer Res 2020;80(14 Suppl):Abstract nr A62.