Transcriptome-wide Mendelian randomization study prioritising novel tissue-dependent genes for glioma susceptibility

• Objective: This study used a two-sample Mendelian randomization (MR) and colocalization approach to identify genes whose tissue-specific expression is causally associated with the risk of glioma and its subtypes (glioblastoma and non-GBM gliomas).
• Key Findings: MR analysis provided evidence for a causal link between the expression of 12 genes and glioma risk, including three novel genes: RETREG2/FAM134A, FAM178B, and MVB12B.
• Tissue Specificity: The results confirmed that effects are often tissue-dependent, such as the association for MDM4 being specific to brain tissue, emphasizing the necessity of using relevant eQTL data.

PubMed: 33504897 DOI: 10.1038/s41598-021-82169-5 Overview generated by: Gemini 2.5 Flash, 28/11/2025

Key Findings: Identifying Glioma Susceptibility Genes

This study integrated genetic data and gene expression data using Mendelian randomization (MR) to identify genes whose expression levels in specific tissues are causally linked to glioma susceptibility and its subtypes (glioblastoma (GBM) and non-GBM gliomas).

• Prioritized Causal Genes: The combined MR and colocalization approach provided evidence that genetically predicted increased gene expression of 12 genes is associated with an increased risk of glioma, GBM, and/or non-GBM risk.
• Novel Susceptibility Genes: Three of these 12 genes are novel glioma susceptibility genes: RETREG2/FAM134A, FAM178B, and MVB12B.
• Tissue Dependence: The findings demonstrated strong evidence for tissue-dependent effects. For example, the causal effect of MDM4 expression on glioma risk was found to be specific to brain tissue, whereas effects for other genes (like TP53) were more generalized (present in both brain and whole blood).
• Glioma Subtype Specificity: The study observed distinct genetic associations for the different glioma subtypes, suggesting that some genetic mechanisms may be specific to GBM or non-GBM tumors.

Study Design and Methods

Study Design

This was a two-sample Mendelian randomization (MR) study that utilized a combined MR and colocalization approach. The primary goal was to infer causal relationships between genetically predicted gene expression levels (the exposure) and glioma risk (the outcome).

Data Sources

• Outcome GWAS Data (Glioma Risk): Summary statistics were used from a large-scale glioma GWAS (7400 cases, 8257 controls), with subtype-specific analyses for glioblastoma (GBM, 3112 cases) and non-GBM gliomas (2411 cases).
• Exposure eQTL Data (Gene Expression): Expression quantitative trait loci (eQTLs) were sourced from two main sources:
  1. GTEx Data (Brain and Whole Blood): eQTLs from whole blood (effective \(n=31,684\)) and brain tissue (estimated effective \(n=1194\)).
  2. Brain-Specific GTEx Data: A more granular analysis was conducted using eQTLs from 13 distinct brain tissues (e.g., cortex, cerebellum, amygdala) with smaller sample sizes (\(n=114\) to 209).

Statistical Analysis

1. Mendelian Randomization (MR): The Inverse-Variance Weighted (IVW) method was used as the primary MR approach. This was complemented by robust sensitivity analyses, including:
   • MR-Egger regression: To test for and correct horizontal pleiotropy.
   • Weighted Median and Weighted Mode: To provide consistent causal estimates even if a subset of instruments are invalid.
   • MR-PRESSO (Pleiotropy RESidual Sum and Outlier): To detect and remove pleiotropic outliers.
2. Colocalization Analysis (moloc): The moloc method was used to confirm that the observed association between gene expression and glioma risk was due to a shared causal genetic variant, rather than two separate causal variants in linkage disequilibrium. This analysis provided strong statistical evidence for a shared signal (posterior probability, \(PP_4 > 0.8\)).

Conclusions and Recommendations

The study concludes that integrating multi-tissue eQTL data with robust MR and colocalization methods is effective for identifying biologically plausible, tissue-dependent genes involved in complex disease risk.

• The findings highlight the crucial role of tissue specificity in the genetic architecture of glioma, demonstrating that expression in the target tissue (brain) is often the most relevant exposure.
• The prioritization of genes like RETREG2/FAM134A, FAM178B, and MVB12B provides strong candidates for functional studies aimed at understanding the precise molecular mechanisms of glioma initiation and progression.
• The authors recommend that future genetic studies on brain disorders should prioritize the use of brain-specific eQTL data over more easily accessible (but less relevant) tissues like whole blood.