Proteome-wide Mendelian randomization in global biobank meta-analysis reveals multi-ancestry drug targets for common diseases

Abstract

Proteome-wide Mendelian randomization (MR) shows value in prioritizing drug targets in Europeans but with limited evidence in other ancestries. Here, we present a multi-ancestry proteome-wide MR analysis based on cross-population data from the Global Biobank Meta-analysis Initiative (GBMI). We estimated the putative causal effects of 1,545 proteins on eight diseases in African (32,658) and European (1,219,993) ancestries and identified 45 and 7 protein-disease pairs with MR and genetic colocalization evidence in the two ancestries, respectively. A multi-ancestry MR comparison identified two protein-disease pairs with MR evidence in both ancestries and seven pairs with specific effects in the two ancestries separately. Integrating these MR signals with clinical trial evidence, we prioritized 16 pairs for investigation in future drug trials. Our results highlight the value of proteome-wide MR in informing the generalizability of drug targets for disease prevention across ancestries and illustrate the value of meta-analysis of biobanks in drug development.

Keywords: complex diseases; drug target prioritization; multi-ancestry Mendelian randomization; plasma proteome.

Graphical abstract

Figure 1

Study design of the multi-ancestry proteome-wide Mendelian randomization (MR) in Global Biobank Meta-analysis Initiative

Figure 2

Proteome MR signals showed distinguished effects in males and females The sex-combined and sex-specific MR estimates were presented for each protein-disease pair.

Figure 3

Multi-ancestry investigation identifying shared protein regions or shared pQTLs (A) Four situations to identify multi-ancestry and ancestry-specific protein regions: (1) protein regions only non-shared pQTLs across ancestries, (2) protein regions with one or more shared pQTLs across ancestries (and no non-shared pQTLs), (3) protein regions with both shared and non-shared pQTLs, and (4) protein regions with pQTL in only one of the ancestries. (B) Number of protein regions with ancestry-specific pQTLs. (C) Number of protein regions with shared pQTLs in the cis region.

Figure 4

Comparison of multi-ancestry proteome MR signals in European and African ancestries (A) Protein-disease pairs with FDR < 0.05 in multi-ancestry comparison. (B) Comparison of MR effect estimates of the seven protein-disease pairs with MR evidence (FDR < 0.05 in multi-ancestry analysis); each point refers to one protein-disease pair, the x axis refers to the MR estimate in European ancestry, and the y axis is the MR estimate in African ancestry. (C) Miami plot of the protein-disease putative causal estimates in European and African ancestries; each point refers to a protein-disease pair, the x axis is the chromosome and position of the protein, the y axis is the −log10(p) of the MR estimate in European (upper) and African ancestry (bottom); the points with colors refer to the 7, 12, and 89 protein-disease pairs with multi-ancestry, African-specific, or European-specific MR effects (FDR < 0.05 in multi-ancestry analysis); different color refers to different outcomes of the protein-disease pairs; the points with legends are the two, seven, and seven protein-disease pairs that showed MR and colocalization evidence in discovery and validation MR analyses; background colors in the legends refer to multi-ancestry (yellow), European-specific (green), and African-specific (blue) MR estimates. (D) Protein-disease pairs with MR (FDR < 0.05) and colocalization evidence in multi-ancestry comparison and validation analysis.

Figure 5

Regional genomic plots of two protein-disease pairs with MR and colocalization evidence of potential causality in European and African ancestries (A) The two theoretical models related to genetic colocalization, causality, and colocalized, as well as confounding by LD. (B) Regional plots of protein level of SERPINE2 on VTE in European and African ancestries. (C) Regional plots of protein level of ABO on VTE in European and African ancestries.

Figure 6

Drug target prioritization profiles of MR signals across European and African ancestries (A) Drug target prioritization profile of this study. (B) Evidence level for the eight prioritized drug targets (details are in Tables S17 and S18).