Sex-stratified polygenic scores for 47 inflammation and vascular stress biomarkers provide a resource for profiling chronic disease risk
摘要
Systemic inflammation underlies many chronic diseases, yet its sex-specific genetic architecture remains under-explored. We developed sex-specific polygenic scores (PGSs) for 47 inflammation and vascular stress biomarkers using 10,000 healthy individuals from the Danish Blood Donor Study and evaluated their associations across 12 chronic diseases in the Copenhagen Hospital Biobank (N ≈ 300,000). Genome-wide association studies identified significant associations for 83% of biomarkers, yielding 112 independent loci and 30 significant sex-by-genotype interactions. Individual sex-specific PGSs constructed from these discovery analyses explained an average of 2.6% of phenotypic variance upon validation. We then partitioned genetic risk into four functional domains (innate proinflammation, growth factors/vascular stress, chemokines, and T-cell-associated inflammation). Association mapping across the 12 diseases revealed distinct shared and sex-specific patterns, with genome-wide PGSs capturing broad systemic risk profiles whilst local, cis-restricted PGSs isolated unconfounded aetiological mechanisms. Subsequent clinical classification using these PGSs yielded modest absolute incremental gains in the area under the curve (ΔAUC). However, non-linear machine learning (XGBoost) optimised the added predictive value in over half of the disease-sex groups. These PGSs establish a validated, open-access resource to map baseline inflammation-related genetic liabilities and clarify sex-divergent aetiological mechanisms at a biobank scale.