Alternative classification of glioblastoma based on BUB1B-inhibition sensitivity. Homo sapiens

We created a computational framework with which to predict BUB1B-GLEBS sensitivity based on gene expression data from BUB1B-inhibition sensitive (BUB1BS) or resistant (BUB1BR) GBM stem cells (GSCs), astrocytes, and neural progenitors. Our classifier examines the expression of 838 genes comprising the BUB1BS signature. Applying this scheme to GBM patient tumor data stratified tumors into BUB1BS and BUB1BR subtypes, revealing that BUB1BS patients have a significantly worse prognosis regardless of their tumor's development subtype (i.e., classical, mesenchymal, neural, proneural). Applying the same scheme to drug sensitivity profiles predicts that BUB1BS cells to be more sensitive to Raf inhibitors as well as other drugs such as type I and II topoisomerase inhibitors among other drugs. Functional genomic profiling of BUB1BR versus BUB1BS GBM isolates revealed differentially reliance of genes enriched in the BUB1BS classifier, including those involved in mitotic cell cycle, microtubule organization, and chromosome segregation. Taken together, our results show that BUB1BR/S classification of GBM tumors, in addition to predicting BUB1B-inhibition sensitivity, may help predict cancer aggressiveness and sensitivity to multiple anti-cancer drugs. Overall design: Examination of 9 GBM tumor tissue and the corresponding GSC.