Genetic Architecture of Circulating Metabolic Biomarkers Across Ancestral Populations

Circulating metabolite levels hold promise for disease risk prediction, yet the genetic architecture and cross-ancestry transferability remain incompletely characterized. Here, we conducted the largest genome-wide association study in East Asians to date, quantifying 248 blood metabolic biomarkers using nuclear magnetic resonance spectroscopy in 37,969 Japanese individuals. We identified eight novel associations and pinpointed rs75326924, causing CD36 deficiency, as a putative causal variant influencing average diameter for low-density lipoprotein particles. A cross-ancestry meta-analysis of 657,341 individuals uncovered 28 additional novel associations and 11,167 credible sets, indicating 52.6% of ancestry-specific causal variants despite highly correlated marginal effect sizes (Pearson r = 0.78 - 0.98). Integration with liver single-nucleus transcriptomics highlighted perivascular hepatocytes as a key cell type of the genetic regulation of circulating biomarkers. Finally, weighted sums of metabolic biomarkers derived from the UK Biobank predicted myocardial infarction and cirrhosis risks in Japanese individuals. Our findings refine the genetic architecture of metabolic biomarkers and underscore their clinical relevance across ancestries.