Systems Analysis of Deeply Phenotyped GBM Cohorts of Long-and Short-Term Survivors

Most patients with glioblastoma multiforme (GBM), an aggressive brain cancer, pass away within the first 15 months of their diagnosis. Obstacles to treating GBM include localization of the tumour, rapid invasion of cancerous cells into nearby brain tissue, scarcity of available treatments (radiotherapy, chemotherapy), and difficulty of surgically removing the tumour. Since its publication in 2005, the Stupp protocol has become the gold standard of care for the treatment of GBM, and it has significantly increased survival rates (a median increase in survival of 2.5 months or a relative reduction in the risk of death of 37 percent). It involves radiotherapy and concurrent chemotherapy using the alkylating agent temozolomide. However, there have been limited advances in finding new treatments for GBM. Hence it isurgently required to identify and develop new treatment options and effective medicine therapies to treat GBM.

In this study, we performed a multi-omic profiling of primary GBM samples. Profiling included RNA-sequencing-based transcriptomics, reverse-phase protein array analysis (RPPA) of proteins and phosphoproteins, and shallow DNA sequencing for analysis of copynumber alterations. The analysis was performed to examine the molecular differences between patients with extremely short-term survival (≤9 months, short-term survivors [STS]) and patients with long-term survival (≥36 months, long-term survivors [LTS]). We selected patients from an internal GLIOTRAIN cohort, and clinically annotated and profiled n=18 STS and n=33 LTS samples, as well as an additional, n=82 intermediate-term survivors (ITS) from four European neurooncology centres (RCSI, ICM, EMC, and LIH).Results were validated in the publicly available The Cancer Genome Atlas (TCGA) cohort.

In Chapter 4, we performed unsupervised clustering of RPPA data, and identified four unique clusters across all GLIOTRAIN samples, although the overall survival of patients was not significantly associated to these clusters. In univariate analysis, however, six signalling proteins —P27, GAB1(Y627), SRC(Y527), BCLXL, BCL2(S70), and RAF(S338) —were significantly associated with overall survival, and the median levels of these proteins was significantly different between STS and LTS samples. We also identified two proteins –STAT3(Y705) AND BCL2(S70) –that were significantly associated with first progression survival. The TCGA cohort included five of these six proteins, but findings could not be validated in this dataset. 

In Chapter 5, we analysed RNA-seq transcriptomics data and identified 6410 genes differentially expressed between STS and LTS samples in the GLIOTRAIN cohort. We also found 31 genes that were associated with cilium related GO terms that showed a better prognosis with higher expression, with a significant p-value of 0.0001. This ledus to believe that the cilium genes are fundamentally responsible for the improved prognosis shown in the KM survival curve in the GLIOTRAIN cohort. It should be noted that by using immunohistochemistry, we confirmed our findings and discovered increased ciliae presence in LTS samples, suggesting a role for restricting GBM growth or invasion. 

Following analysis of shallow-seq data in chapter 6, we identified significant amplification at chromosome 7 in both STS and LTS samples from both the GLIOTRAIN and TCGA cohorts. A significant deletion was observed at chromosome 9 in both STS and LTS samples from both the GLIOTRAIN and TCGA cohorts. We than used machine learning algorithm to classify the differences between the STS and LTS samples using the GLIOTRAIN cohort as prediction dataset and TCGA as the validation dataset. There were no discernible differences in Copy Number Alteration profiles between STS and LTS samples, this was likely due to small sample sizes and an imbalance of class between STS and LTS.

Collectively, this research demonstrated that cilium gene signatures displayed a prognostic value for overall survival of samples from the GLIOTRAIN cohort, and thatmay have significant potential as novel therapeutic targets for the treatment of GBM. In conclusion, work presented here provides novel insights into the molecular landscape of GBM that can help to improve personalized medicine treatment