ESTIMATING THE CAUSAL EFFECTS OF CARDIOMETABOLIC FACTORS ON CORONARY ARTERY DISEASE IN BRITISH PAKISTANIS AND BANGLADESHIS: A TRANS-ANCESTRY MENDELIAN RANDOMISATION STUDY

Background and Aims : British people with South-Asian ancestry have a higher risk of coronary artery disease (CAD) than other ancestry groups. Statistical power can be the limiting factor when extending Mendelian Randomisation (MR) to non-European populations because ancestry-matched GWAS for risk factors (RFs) of interest might not be sufficiently large.Methods: We compared different strategies for trans-ancestry MR to assess the causal effect of cardiometabolic RFs (BMI, triglycerides, HDL-cholesterol, LDL-cholesterol, systolic and diastolic blood pressure) on the risk of CAD in 22,000 British Pakistani and Bangladeshi (BPB) individuals from the Genes&Health cohort. We used an ancestry-matched sample to derive instruments in a two-sample MR of CAD in G&H, with summary statistics for RF from the UK Biobank BPB group. However, insufficient numbers of genome-wide significant instruments were identified in the UK Biobank BPB population. Therefore, we used a less stringent p-value threshold (p<5x10-5) for selecting instruments, incorporated results from large European GWASs, and used a subset of loci with evidence of transferability.Results: We found that most of the associations were not significant in the ancestry-matched MR. We found a risk increasing effect for LDL-cholesterol and risk decreasing effect for HDL-cholesterol when using the variants from the large European GWAS as instruments, and also for the subset of loci that were transferable. The association of BMI with CAD was significant only for transferable loci.Conclusions: We showed that incorporating findings from large European GWAS can increase power for MR in other ancestry groups. We demonstrated the importance of considering transferability of RF loci to ensure causal inference. Background and Aims : British people with South-Asian ancestry have a higher risk of coronary artery disease (CAD) than other ancestry groups. Statistical power can be the limiting factor when extending Mendelian Randomisation (MR) to non-European populations because ancestry-matched GWAS for risk factors (RFs) of interest might not be sufficiently large. Methods: We compared different strategies for trans-ancestry MR to assess the causal effect of cardiometabolic RFs (BMI, triglycerides, HDL-cholesterol, LDL-cholesterol, systolic and diastolic blood pressure) on the risk of CAD in 22,000 British Pakistani and Bangladeshi (BPB) individuals from the Genes&Health cohort. We used an ancestry-matched sample to derive instruments in a two-sample MR of CAD in G&H, with summary statistics for RF from the UK Biobank BPB group. However, insufficient numbers of genome-wide significant instruments were identified in the UK Biobank BPB population. Therefore, we used a less stringent p-value threshold (p<5x10-5) for selecting instruments, incorporated results from large European GWASs, and used a subset of loci with evidence of transferability. Results: We found that most of the associations were not significant in the ancestry-matched MR. We found a risk increasing effect for LDL-cholesterol and risk decreasing effect for HDL-cholesterol when using the variants from the large European GWAS as instruments, and also for the subset of loci that were transferable. The association of BMI with CAD was significant only for transferable loci. Conclusions: We showed that incorporating findings from large European GWAS can increase power for MR in other ancestry groups. We demonstrated the importance of considering transferability of RF loci to ensure causal inference.