Multi-omics Mendelian randomisation using expression, protein and splicing quantitative trait loci to identify novel drug targets associated with risk of gliomagenesis

BackgroundGenetic variants associated with molecular traits that are also associated with liability to glioma can provide causal evidence for the prioritisation of these traits as candidate drug targets. MethodsWe performed two-sample Mendelian randomisation and genetic colocalisation of a large panel of molecular traits on glioma. Molecular data were taken from studies of expression quantitative trait loci (QTL) [11,985 genes]; splicing QTL [13,285 genes] and protein QTL [7,376 proteins] derived from 15 brain tissues. Glioma summary-level data was extracted from a genome-wide association meta-analysis of 12,496 cases and 18,190 controls. ResultsOur MR analysis showed evidence for a causal effect of 85 molecular traits on glioma - 37 were robust according to colocalisation and Steiger filtering. We found causal evidence for 10 genes previously associated with glioma risk. We identified one novel genetic locus with strong causal evidence in the gene expression analysis: HBEGF (5q31.3) in all glioma [OR 1.36 (95%CI 1.19 to 1.55); P = 4.41 x 10-6]. We also identified three novel genetic loci with strong causal evidence in the splicing variation analysis: CEP192 (18p11.21) in glioblastoma [OR 4.40 (95%CI 2.28 to 8.48); P = 9.78 x 10-4]; FAIM (3q22.3) in all glioma [OR 2.72 to 3.43; P = 1.03 x 10-5 to 1.09 x 10-5] and SLC8A1 (2p22.1) in all glioma [OR 0.37 (95%CI 0.24 to 0.56; P = 5.72 x 10-6]. ConclusionsWe provide robust causal evidence for genes previously associated with glioma risk in genome-wide association studies, as well as four novel genes.