Transcriptome-wide Mendelian randomisation identifying brain-specific causal genes influencing glioma


Session type:


Jamie Robinson1,Richard Martin1,Amy Howell1,Spiridon Tsavachidis2,Georgina Armstrong2,Caroline Relton1,Melissa Bondy2,Jie Zheng1,Kathreena Kurian1
1University of Bristol,2Baylor College of Medicine



The human transcriptome can provide promising insights into causal genes in glioma aetiology and help prioritise potential drug targets for glioma treatment.


We apply two-sample Mendelian randomisation (MR) and genetic colocalisation to explore the influence of genetically predicted gene expression across 13 brain tissues from GTEx (n=80 to 491) and whole blood from eQTLGen (n=31,684) on glioma risk (5,739 cases, 5,501 controls).


The MR analysis showed a strong association for 15 results on glioma risk, consisting of 10 unique genes. 11 of these results showed strong evidence of colocalisation. We identified a novel glioma susceptibility locus (2q35) with two genes associated with glioma: ZFAND2B (odds ratio (OR)=0.70; 95% confidence interval (CI): 0.59 to 0.82; P=2.17x10-05) and ATG9A (OR=0.64; 95% CI: 0.52 to 0.79; P=2.76x10-05). Furthermore, we systematically linked each gene with evidence from Open Targets to further strengthen the putative causal evidence of genes on glioma. With triangulation of evidence from the Open Targets platform and with the concept of the druggable genome, we highlight two genes as having potential use as drug targets for glioma: TP53, a well-known tumour-suppressor, and JAK1, involved in the JAK-STAT pathway implicated in glioma development.


This study evidences the value of expression QTL MR in identifying causal associations for glioma and also for the prioritising of potential therapeutic targets that can aid drug discovery methods.