MRC Biostatistics Unit, University of Cambridge, Cambridge, UK.
Department of Non-Communicable Disease Epidemiology, London School of Hygiene and Tropical Medicine, London, UK.
Commun Biol. 2024 Nov 14;7(1):1512. doi: 10.1038/s42003-024-07236-9.
Meta-analysis of genome-wide association studies (GWAS) across diverse populations offers power gains to identify loci associated with complex traits and diseases. Often heterogeneity in effect sizes across populations will be correlated with genetic ancestry and environmental exposures (e.g. lifestyle factors). We present an environment-adjusted meta-regression model (env-MR-MEGA) to detect genetic associations by adjusting for and quantifying environmental and ancestral heterogeneity between populations. In simulations, env-MR-MEGA has similar or greater association power than MR-MEGA, with notable gains when the environmental factor has a greater correlation with the trait than ancestry. In our analysis of low-density lipoprotein cholesterol in ~19,000 individuals across twelve sex-stratified GWAS from Africa, adjusting for sex, BMI, and urban status, we identify additional heterogeneity beyond ancestral effects for seven variants. Env-MR-MEGA provides an approach to account for environmental effects using summary-level data, making it a useful tool for meta-analyses without the need to share individual-level data.
对不同人群进行全基因组关联研究（GWAS）的荟萃分析为识别与复杂特征和疾病相关的基因座提供了更强有力的证据。通常，人群间效应大小的异质性与遗传背景和环境暴露（如生活方式因素）相关。我们提出了一种环境调整的荟萃回归模型（env-MR-MEGA），通过调整和量化人群间的环境和遗传异质性来检测遗传关联。在模拟中，env-MR-MEGA 具有与 MR-MEGA 相似或更高的关联能力，当环境因素与特征的相关性大于遗传背景时，关联能力会有显著提高。在我们对来自非洲的 12 项性别分层 GWAS 中约 19000 个人的低密度脂蛋白胆固醇的分析中，在调整了性别、BMI 和城市状况后，我们发现了七个变体的遗传背景效应之外的额外异质性。env-MR-MEGA 提供了一种使用汇总水平数据来解释环境效应的方法，它是一种无需共享个体水平数据的荟萃分析的有用工具。
