Multi-ancestry genetic risk score for lung function from meta-analysis of genome-wide association studies in 580,869 individuals from five ancestry groups

Abstract

Genome-wide association studies (GWAS) of lung function and COPD to date have primarily focussed on European ancestry cohorts. Multi-ancestry studies GWAS should improve statistical power, and improve prospects to inform prediction, prevention, diagnosis, and treatment relevant to diverse populations. We obtained GWAS summary statistics from cohorts of European, African, American, East and South Asian ancestries (N=580,869 in total). In discovery we meta-analysed results for FEV₁, FVC, FEV₁/FVC and Peak Expiratory Flow at 66.8M variants. We incorporated genome-wide associated variants in genetic risk scores (GRS) with ancestry-specific weights. We identified 1020 independent signals for lung function (P <5×10^‑9). Compared with our previous GRS weighted by European ancestry effects, a GRS using weights derived in each ancestry separately increased the proportion of FEV₁/FVC variance explained from 5.02% to 7.30% in European ancestry, from 0.74% to 2.14% in African ancestry and from 2.02% to 3.75% in East Asian ancestry. Weights from a meta-regression across ancestries with continuous axes of ancestry as covariates further increased the variance explained to 7.77%, 2.35% and 3.84% in each ancestry respectively. In summary, we have increased the number of lung function signals from around 300 to over 1000. A new multi-ancestry weighted genetic risk score is more predictive for lung function across European, African and East Asian ancestries than either a GRS using weights derived from European ancestry or from matched ancestry. Our findings highlight the importance of inclusion of multiple ancestries in respiratory genomic studies.