| Summary |
Meningiomas are the most common primary intracranial tumors in humans. While most of these tumors are benign, some are malignant, rapidly recur after multimodal treatment with surgery and radiotherapy, and can ultimately be fatal. The current WHO grade system does not always identify high risk meningiomas, therefore better characterizations of the biology of aggressive tumors are needed. In order to address these challenges, we combined 13 bulk RNA-Seq datasets, corrected for batch effects, and applied Uniform Manifold Approximation and Projection (UMAP) to create a reference landscape of ~1300 meningioma tumors. Our analyses revealed multiple distinct meningioma subtypes with specific biological signatures. Clinical metadata, mutations, copy number alterations, and gene-fusion data effectively correlated with regions of the UMAP. Notably, regional distribution of time to recurrence identified major clusters as well as intra-cluster differences of meningiomas with varying patient outcomes. The most aggressive subtype, characterized by an enrichment of higher WHO grades, frequent tumor recurrences, and shorter time to recurrence, exhibited elevated proliferation rates and RNA expression resembling muscle development. To facilitate clinical applications, we developed a cross-validated nearest-neighbors-based algorithm that accurately maps new patients onto this UMAP landscape. Our study highlights the utility of transcriptomic analysis in discerning meningioma heterogeneity as well as successful combination of multiple datasets from various sources. We provide a valuable tool for understanding the disease, predicting tumor biology and patient prognosis. This resource is accessible via the open source, interactive online tool Oncoscape, where the scientific community can explore the landscape and mine clinical and genomic metadata.
|
Less...
More...