2003-LB: Dissecting Ancestry-Aware Molecular Causal Paths of Type 2 Diabetes

Multiple molecular mechanisms are involved in the pathogenesis of type 2 diabetes (T2D), with potentially different effects across ancestries. Recent large-scale efforts by the Type 2 Diabetes Global Genomics Initiative (T2DGGI) have broadly described the genetic architecture of T2D, data that can be subsequently used to pinpoint causal molecular mechanisms leading to T2D in an ancestry-aware manner.

In this work, we sought to explore the causal effects of gene expression and proteins on T2D across ancestries by leveraging data from the latest T2DGGI multi-ancestry genome-wide association study (GWAS). We conducted two-sample Mendelian randomization (MR) using cis-expression and protein quantitative trait loci (eQTL/pQTL) data from various datasets of four different major ancestries. To corroborate our findings, we performed statistical colocalization using PWCoCo. We then performed meta-analyses across ancestries for molecular traits showing evidence of causality in at least one ancestry (5% FDR-adjusted MR p-value and PWCoCo posterior probability ≥ 0.5).

We found causal evidence for changes in the genetically regulated expression of 267 genes and 15 proteins with T2D risk in at least one ancestry. When meta-analyzing the results across ancestries, we found, for eQTLs, evidence of causality between 148 genes and T2D, including PABPC4, a high-confidence effector gene for T2D with evidence of ancestry-correlated heterogeneity (I2 = 85%). For pQTLs, NELL1, ANXA7, PCSK1, and CTRB2 were found to be causal to T2D risk in the cross-ancestry meta-analysis, with CTRB2 showing evidence of ancestral heterogeneity (I2 = 85%).

Our findings highlight the power of large-scale GWAS and multi-omics MR to identify causal pathways involved in T2D risk. Our results also show the existence of ancestry-correlated heterogeneity, which emphasizes the need for expanding investigations into non-European ancestry populations to better understand T2D etiology.