TBIO-02. Mapping the origins of pediatric brain tumors to cell type lineages in the developing cerebellum.

Abstract

Understanding the cellular origins of cerebellar tumors at single cell resolution may help identify novel therapeutic vulnerabilities to overcome current impediments in pediatric neuro-oncology. One approach to answer this important question involves the mapping of tumor profiles onto normal embryonal central nervous system development. Previous attempts in this regard focused on integrating human tumor data with developing mouse cerebellum atlases, however, these approaches were inherently limited due to species-specific differences. In order to address this critical issue, we use a high resolution developing human cerebellum single cell atlas in order to compare to extended bulk and single cell transcriptomes profiles from pediatric solid tumors. In result, we provide novel and confirmatory findings for the cellular association of medulloblastoma SHH, Group3 and Group4 as well as pilocytic astrocytoma, using comprehensive approaches to decipher the cellular composition and map the origins of childhood brain tumors. We also shed light on the cellular origins of posterior fossa ependymoma and radiation-induced glioma (secondary tumor, occurring after medulloblastoma). As a common feature among the tumor similarity to normal cell types, we identify gradients of differentiation, starting from early progenitor cells to more differentiated cell states, observed in all cerebellar tumor entities investigated at single cell level; thus reflecting possible lineages of origin. Pertaining to clinical application, we identified specific developmental genes shared between cerebellar lineages and associated tumors, and tumor-specific genes absent in all cerebellar lineages. These two categories of genes comprise candidate lineage markers for faithful modeling and, if absent from other organ systems after birth, could become an important source of potential therapeutic targets. Importantly, all analysis results are publicly available via an interactive online user interface (brain-match.org) that serves as an open valuable resource for the scientific community.