P13.06 Transcriptome-wide Mendelian randomization study to identify brain-specific causal genes influencing glioma

Abstract BACKGROUND The drug treatment regimen for glioma has remained relatively static since the introduction of temozolomide, although new drugs and drug combinations are being trialled. The human transcriptome can provide promising insights into causal genes as potential drug interventions for glioma treatment and may guide drug discovery methods. MATERIAL AND METHODS We apply two sample Mendelian randomisation and colocalisation to explore the influence of genetically-predicted gene expression across 12 tissue types located in the brain (4,554 genes from GTEx) and whole blood (16,112 genes from eQTLGen) on glioma risk (5,739 cases, 5,501 controls from a meta-analysis of GICC and MDA glioma GWAS). We used the MR-Base R package to conduct these analyses. RESULTS We identify 9 genes whose genetically-predicted expression was strongly associated with glioma risk. Of these genes, 7/9 are shown to have tissue-specific expression while the other two genes showed an association with glioma across multiple brain tissues and whole blood. For example, JAK1, involved in the well-known JAK-STAT pathway, is found in the frontal cortex (OR=1.49 for glioma per standard deviation increase in gene expression; 95% CI: 1.28 to 1.73; P=1.79 × 10−7), the cerebellar hemisphere (OR=1.32; 95% CI: 1.19 to 1.47; P=2.64 × 10−7), the cerebellum (OR=1.27; 95% CI: 1.16 to 1.39; P=2.64 × 10−7) and the cortex (OR=1.38; 95% CI: 1.22 to 1.57; P=2.64 × 10−7). This pathway has been highlighted in previous research as a potential intervention target for glioma therapies. We found that 5/9 of the genes from the MR analysis are expressed in the cerebellum. However malignant cerebellar glioma is a rare tumour (~3% of all malignant gliomas). This suggests that tumourigenesis elsewhere in the brain may be affected by other tissue-specific genes, specifically in the cerebellum, though this will require further research to elucidate. We further triangulate the MR findings with evidence from the OpenTargets platform to strengthen the putative causal associations. OpenTargets aims to “generate evidence on the validity of therapeutic targets based on genome-scale experiments and analysis”. For example, JAK1 receives an overall OpenTargets score of 0.89 out of 1, with most of the evidence for this JAK1-glioma association coming from affected pathways data. CONCLUSION This study has combined genetic epidemiological approaches to the analysis of the human transcriptome on glioma incidence. We provide evidence that these genes may inform putative drug targets for tertiary treatment of glioma. Future research specifically towards this aim will be required to fully elucidate intervention targets.