PATH-49. COMPREHENSIVE RNA-SEQ ANALYSIS REVEALS NOVEL SUBTYPES OF MENINGIOMAS AND PREDICTS PATIENT OUTCOME.

Abstract

Abstract OBJECTIVE Meningiomas are the most common intracranial tumor in humans. While most tumors are benign, some are malignant and ultimately lethal. Current grading systems based on WHO grade do not provide adequate risk stratification, therefore, better characterizations of the biology of aggressive meningiomas are needed. METHOD: We obtained 12 datasets from 9 institutions and 5 countries in North America, Europe and Asia and combined them with ~300 tumors sequenced from the University of Washington to create a set of ~1300 meningioma tumors. Raw sequencing data was aligned to hg38, batch corrected and then dimension reduction of normalized counts was performed to create a UMAP landscape. Differential expression analysis and gene set enrichment analysis were performed to elucidate the underlying biology of meningioma subtypes. RESULTS The analysis revealed nine distinct meningioma subtypes through unsupervised clustering. Clinical metadata, DNA sequencing, copy number alterations, and gene-fusion data effectively correlated with these identified clusters. These subtypes exhibited specific biological signatures. Notably, regional distribution of time to recurrence identified subtypes as well as intra-cluster differences of meningiomas with varying patient outcomes. The most aggressive subtype, characterized by higher WHO grades, frequent tumor recurrences, and shorter time to recurrence, exhibited elevated proliferation rates and RNA expression resembling embryonic limb development. To facilitate clinical applications, we developed a cross-validated nearest-neighbors-based algorithm that accurately mapped new patients onto this UMAP landscape, achieving a remarkable 95% accuracy. CONCLUSION Our study highlights the utility of RNA sequencing in discerning meningioma heterogeneity and provides a valuable tool in predicting tumor biology and patient prognosis. The integration of our UMAP landscape with patient data offers an effective approach for personalized treatment strategies.