Reverse Mendelian randomization separates causes from early proteomic biomarkers of glioma

Background/ObjectivesGlioma represents the largest entity of primary brain tumours in adults, with an overall survival of less than 20% over 5 years. Glioblastoma is the most frequent and aggressive glioma subtype. At present, there are few well-established pre-clinical predictors for glioma incidence. Due to the availability and size of prognostic studies in glioma, we utilised a Mendelian randomization framework to identify non-causal protein biomarkers which are associated with early-onset of glioma in the European population. MethodsWe generated polygenic risk scores (PRS) for glioma (n=12,496), glioblastoma (n=6,191), and non-glioblastoma (n=5,819) cases. We used reverse Mendelian randomization (MR) to examine the relationship between the genetic liability of glioma and 1,463 and 90 proteins were measured using an Olink panel (UKBB, n=35,571 and SCALLOP, n=21,758), additionally 4,907 and 2,994 aptamers were assayed using SOMAscan assays (deCODE n=35,559 and INTERVAL, n=3,301). We further performed a forward cis-MR and colocalization analysis leveraging the circulating protein markers in risk of glioma, glioblastoma and non-glioblastoma. ResultsReverse MR identified 161 unique proteins associated with the PRS of glioma, 79 proteins associated with the PRS of glioblastoma, and 11 proteins associated with the PRS of non-glioblastoma. Enrichment analyses identified a proportion of plasma proteins to be associated with the PRS of glioma to be correlated with response to external stimulus. A group of plasma proteins linked to the PRS of glioma and glioblastoma were related to the immune system process. Forward MR of the putative relationships were found to have little or no evidence of association on the causal pathway. Candidate markers ETFA, RIR1 and BT3A1 are evidenced in glioma risk. ConclusionOur findings identify a high genetic liability to glioma is associated with the immune system processes. Non-causal plasma biomarkers identified through PRS associations could indicate novel non-causal biomarkers of early glioma development.