MULTI-OMICS MENDELIAN RANDOMISATION USING EXPRESSION, PROTEIN AND SPLICING QUANTITATIVE TRAIT LOCI: IDENTIfiCATION OF NOVEL DRUG TARGETS ASSOCIATED WITH RISK OF GLIOMAGENESIS

Abstract

Abstract AIMS Genetic variants associated with molecular traits are also associated with liability to glioma. We sought to pro- vide causal evidence for the prioritization of these traits using triangulation of several causal inference approaches. METHOD We performed two-sample Mendelian randomization and genetic colocalization of molecular traits on glioma. Molecular data were taken from studies of expression quantitative trait loci (QTL) [11,803 genes]; protein QTL [7,376 proteins] and splicing QTL [13,285 genes]) derived from 15 brain tissues. Glioma data were taken from a genome-wide association meta-analysis (12,496 cases and 18,190 controls). RESULTS The MR analysis showed evidence for a causal effect of 25 molecular traits on glioma. Of these, 19 (all gene expression) were robust according to colocalization and Steiger filtering. 18/19 (95%) of these genes were previously implicated in genome-wide and transcriptome-wide association studies with glioma risk. We identified one novel locus with causal evidence in cortex tissue: HBEGF (5q31.3). HBEGF expression has been found to be significantly increased in many human cancer types, including glioma. We found genetically predicted in- creased expression of HBEGF associated with an increased risk of all glioma [OR 1.53 (95%CI 1.28 to 1.82); P =2.66 x 10-6] and the glioblastoma subtype [OR 1.68 (95%CI 1.35 to 2.08); P = 2.97 x 10-6]. CONCLUSIONS We provide robust causal evidence as validation for genes previously implicated with glioma risk in genome- wide association studies. Additionally, we show for the first-time evidence for a causal relationship between HBEGF expression and glioma risk.