Brain multi-omic Mendelian randomisation to identify novel drug targets for gliomagenesis

Abstract Background Genetic variants associated with molecular traits that are also associated with liability to glioma can provide causal evidence for the identification and prioritisation of drug targets. Methods We performed comprehensive two-sample Mendelian randomisation (Wald ratio and/or IVW) and colocalisation analyses of molecular traits on glioma. Instrumentable traits (QTLs P < 5 × 10−8) were identified amongst 11 985 gene expression measures, 13 285 splicing isoforms and 10 198 protein abundance measures, derived from 15 brain regions. Glioma summary-level data was extracted from a genome-wide association meta-analysis of 12 496 cases and 18 190 controls. Results We found evidence for causal effect of 22 molecular traits (across 18 genes/proteins) on glioma risk. Thirteen molecular traits have been previously linked with glioma risk and five were novel; HBEGF (5q31.3) expression and all glioma [OR 1.36 (95%CI 1.19–1.55); P = 4.41 × 10−6]; a CEP192 (18p11.21) splice isoform and glioblastoma [OR 4.40 (95%CI 2.28–8.48); P = 9.78 × 10−4]; a FAIM (3q22.3) splice isoform and all glioma [OR 2.72–3.43; P = 1.03 × 10−5 to 1.09 × 10−5]; a SLC8A1 (2p22.1) splice isoform and all glioma [OR 0.37 (95%CI 0.24–0.56; P = 5.72 × 10−6]; D2HGDH (2q37.3) protein and all glioma [OR 0.86 (95%CI 0.80–0.92); P = 5.94 × 10−6)]. Conclusions We provide robust causal evidence for prioritising genes and their protein products in glioma research. Our results highlight the importance of alternative splicing as a mechanism in gliomagenesis and as an avenue for exploration of drug targets.