ID #353 Deciphering the tumour clones, epigenetic states and transcriptomic signatures underlying fusion-driven astroblastoma

Abstract Astroblastoma (ABM) is a central nervous system tumour driven by defined gene fusions involving epigenetic and transcriptomic regulators such as the MN1::BEND2 fusion. ABMs have different disease courses including both aggressive and benign tumours. ABMs mostly occur in female patients; however, the mechanisms behind the diverging clinical features as well as the drivers of malignancy and respective therapeutic options are poorly understood. To investigate how oncofusions reshape the epigenome and transcriptome in ABM we resorted to single-cell technologies including RNA and Multiome sequencing. Covering a range of clinically more benign and aggressive tumours, we built a cohort recapitulating the diverse clinical courses. We uncovered the intra- and intertumoral heterogeneity and analysed the epigenetic and transcriptomic landscape of ABM. Distinguishing microenvironment and tumour cells, we identify distinct cell states, including neuronal- and ependymal-like states. Remarkably, whereas some states were shared among aggressive and benign tumours, others were unique to the different disease courses. Similarly, the immune cell compartment showed diverging compositions in clinically distinct patient groups, suggesting clinical features are also recapitulated on the single-cell level. Leveraging the Multiome data we found both converging and diverging cell identities in the genetic, epigenetic and transcriptomic landscape. Reconstructing tumour cell clones we could recapitulate clonal evolution and identify recurring alterations upon disease progression. Finally, by constructing gene regulatory networks, we shed lights on the molecular drivers of tumour aggressiveness and identified the PDGF-receptor pathway in aggressive tumour cell states shedding light on potential therapeutic targets. In future, we aim to uncover the developmental origin of ABM by integrating our data with neurodevelopmental datasets. Furthermore, by in-depth characterisation of gene regulatory networks we will test therapeutic vulnerabilities, especially in recurring aggressive tumours. Taken together, our multimodal atlas of ABM uncovers how both epigenome and transcriptome are reshaped by oncogenic fusions.