Abstract
Genome-wide association studies (GWAS) of kidney function have focused almost exclusively on the estimated glomerular filtration rate derived from serum creatinine (eGFRcrea), which may also reflect muscle mass metabolism. To identify kidney-specific loci, researchers typically compare the GWAS results with those from alternative kidney function markers such as cystatin C-based eGFR (eGFRcys), uric acid (UA), and blood urea nitrogen (BUN). Assuming the existence of an underlying kidney function trait affecting the levels of multiple markers, we applied genomic structural equation modeling (GenomicSEM) combining publicly available univariate GWAS results of eGFRcrea (n=1,201,909), eGFRcys (n=460,826), UA (n=457,690), and BUN (n=852,678) from the CKDGen Consortium. Z-scores of the results were compared against 424 lead variants of each loci in the previous GWAS for eGFRcrea. The merged summary statistics across four previous GWAS included >6,000,000 SNPs. The highest genetic correlation was observed between eGFRcrea and eGFRcys (rg=.593, se=.041), followed by eGFRCys-UA (rg=–.393, se=.037), eGFRcys-BUN (rg=–.387, se=.042), eGFRcrea-BUN (rg=–.364, se=.044), eGFRcrea-UA (rg=–.217, se=.035), and UA-BUN (rg=.211, se=.039). Confirmatory factor analysis (CFA) showed good fitting (comparative fit index=.936, root mean square residual=.028) and identified factor loadings (λ) for each biomarker: λeGFRcrea=.654; λeGFRcys=.896; λBUN=–.405; and λUA=–.479. Of the 373 available SNPs which were significant in the previous eGFRcrea GWAS, the absolute Z-score of the common latent kidney trait were bigger for 119 variants (31.9%). Of these, 112 variants (94.1%) showed significant effects on eGFRcrea, eGFRCys and BUN, underlying their kidney function relevance. Among 254 variants with smaller absolute Z-score in the latent kidney trait analysis, only 86 (33.9%) were significantly associated with eGFRcrea, eGFRcys and BUN. These findings highlight the strong enrichment of loci directly relevant to the kidney obtained with GenomicSEM. In the absence of a kidney function-specific marker, GenomicSEM analysis shows the potential to prioritize kidney function-relevant loci by combining multiple kidney-related biomarkers.