EPCO-18. UNRAVELLING IDH-MUTANT GLIOMAS HETEROGENEITIES WITH SPATIAL TRANSCRIPTOMICS

Abstract Oligodendrogliomas (OL) and Astrocytomas (AS), typically affect young adults and are characterized by the mutation of the isocitrate dehydrogenase genes (IDH1/2). While IDH-mutant gliomas are of good initial prognosis, they relapse almost systematically in higher-grade gliomas. These tumors display large intra- and inter-patient heterogeneity hallmarked by dynamic adaptation of tumor cell phenotypes and microenvironment across tumors and grades. However, the spatial organization and interactions between cell-types from the brain tumor microenvironment remain elusive and is required to better understand tumor growth and relapse. Here, we profiled in-depth IDH-mutant gliomas with Visium spatial transcriptomics across 32 areas from 20 patients that underwent surgical resection. Our cohort spans tumors of grade II to IV in WHO 2021 classification. We implement state-of-the art bio-informatics methods to decipher spatial heterogeneities such as chromosomal alterations, spatial domains, cell-type distribution and cell-cell interactions. With copy number variation inference, we retrieve stereotypical chromosomal alteration describe in the literature and classify spot according to their malignant content to identify tumor core areas compare to infiltrated cortex. By performing cell-type deconvolution with cell2location, we provide a map of IDH-mutant tumors microenvironment showing tumor cell states and brain tumor microenvironment cell spatial distribution and relationships. We then use machine learning and spatial statistics to identify recurrent patterns across samples and compare their spatial distribution across IDH-mutant subtypes and grades. With matrix factorization technique, we identify cellular niches defined by recurrent cell-type co-localizations found across IDH-mutant gliomas. We also identify clusters enriched in hypoxia programs that are linked to the specific cell-type compositions uncovered with deconvolution. Finally, we provide a spatially resolved landscape of the cell-cell interaction of IDH-mutant gliomas where we identify communication patterns between cellular niches. To summarize, our study provides a novel comprehensive landscape of IDH-mutant glioma spatial organisation across subtypes and grades.