1054 Multiomic Characterization of Meningiomas Informs Celltype of Origin and Identifies Immune Dysregulation at the Blood-CSF Barrier

INTRODUCTION: Meningiomas are the most common central nervous system neoplasm believed to derive from arachnoid cap cells. Traditionally considered an inert structural organ, meninges are increasingly recognized to have immune regulatory functions. METHODS: scRNA-seq data were analyzed from 12 meningiomas, 2 meningioma-brain interface specimens, and 2 dura 1. Clustering, differential gene expression, cell type assignment, and pathway analysis were performed (Seurat v4 and CellChat) with anchor-based mapping of meningeal and immune cells to mouse E14 meningeal and CITE-seq reference of peripheral blood mononuclear cells, respectively 2-7. 840 meningiomas and 12 dura underwent DNA methylation-classification and 857 underwent whole transcriptome sequencing 8-13. RESULTS: Meninges-derived clusters were transcriptionally homogenous within a sample. Each tumor exhibited a predominant transcriptional signature of dura, arachnoid, or pial. Meningiomas exhibited a depletion of lymphocyte populations with expansion of yolk-sac derived PU.1/SPI1+ myeloid populations without evidence of infiltration of bone-marrow derived MYB+ myeloid cells. DNA-methylation based classification assigned 809/840 (96.4%) meningiomas to one of 7 described methylation classes. High-grade meningiomas were significantly less likely to be assigned to a meningioma DNA methylation class (WHO 1/I: 497/505 (98.4%) vs. WHO 2/II & 3/III: 312/335 (93.1%), c2=37.3, df=2, p=7.91x10-9). Overall, 23 of 31 meningiomas (74%) that were assigned to non-meningioma methylation classes were WHO 2/II & 3/III. CONCLUSIONS: Meningiomas may originate from multiple layers of the meninges. Expanded tumor myeloid populations express SPI1/PU.1 indicating that these populations represent expansion of tissue resident border macrophages in the blood-CSF barrier rather than recruitment of bone-marrow derived myeloid populations. DNA-methylation based classification was able to successfully classify >95% of meningiomas, however, a disproportionate number of tumors classified as non-meningioma tumor subtypes implying an unrecognized heterogeneity exists.