Mediation of Polygenic Asthma Risk Through Gene Expression.

Abstract

Existing asthma polygenic risk scores (PRSs) have minimal validation in African-ancestry populations, leaving gaps in our understanding of the wide applicability of PRSs. To widen our understanding of the applicability of asthma PRSs, we apply published PRSs in African-ancestry individuals and quantify the extent to which the PRS-asthma relationship is mediated by clinical biomarkers and gene-expression signatures of asthma. We applied 22 PRSs from the PGS Catalog in 673 individuals from the Consortium on Asthma among African-Ancestry Populations in the Americas (CAAPA) and calculated the percent of the PRS-asthma relationship that is statistically mediated by clinical and nasal epithelium transcriptomic biomarkers of asthma. Asthma case/control status was defined as ever/never having a doctor's diagnosis of disease. For gene expression mediation analysis, we limited the cases to those with current disease. The PRS (PGS001782) created by the Global Biobank Meta-analysis Initiative (N = 32,658 individuals of African ancestry) performed the best (ΔAUC = 0.104, AUC = 0.657) adjusted for age, sex, study site, and the first two genetic principal components (PC1-2). The PRS's effect on asthma was mediated by total IgE (tIgE) (38.8%, p.adj < 0.0002), multi-allergen ImmunoCAP phadiatop specific IgE (sIgE) (38.7%, p.adj < 0.0002), and eosinophils (7.3%, p.adj = 0.004). Mediation was observed for gene expression modules related to T2 inflammation (21.9%, p.adj < 0.0024), wound healing (11.9%, p.adj = 0.008), and medication response (6.8%, p.adj = 0.049). We found the best PRS to be the one derived using the largest sample size and including African-ancestry individuals. Mediation supports the well-documented biology of T2 inflammation in asthma as well as pathophysiological components of asthma like wound healing and medication response.