Multi-ancestry genome-wide association study of serum creatine kinase implicates myopathy genes and muscle pathways.

Fig. 1. Overview of multi-ancestry GWAS and functional analyses of serum CK associated loci. We performed a fixed-effects meta-analysis of GWAS summary statistics for CK levels from six cohorts across diverse ancestries. Associated loci were annotated and tested for enrichment in Mendelian muscle disorder genes. Candidate genes were prioritised using eQTL and pQTL data. Shared genetic architecture with related traits was explored through LD score regression, Mendelian randomisation, and colocalisation. Loci were further clustered according to pleiotropic profiles.

Fig. 2. Multi-ancestry genome-wide association and tissue/cell-type enrichment analyses. (A) Manhattan plot of multi-ancestry GWAS results for CK across all autosomes; the red dashed line denotes the genome-wide significance threshold (P = 5 × 10−8). Previously unreported lead SNPs are shown in red; for clarity only previously unreported loci with exonic lead SNPs are annotated with mapped genes. The y axis is capped at −log10(P) = 50; peaks exceeding this value are truncated and marked with arrows, with lead variants and mapped genes rs7305678 (CD163/APOBEC1), rs11559024 (CKM), rs12975366 (LILRB5) and rs7481951 (ANO5). (B & C) MAGMA tissue and cell-type enrichment analysis based on (B) GTEx expression data, (C) Human Protein Atlas single-cell expression profiles; the dashed line indicates the Bonferroni-corrected significance threshold, and significant tissues and cell types are shown in red.

Fig. 3. Mendelian Randomisation and colocalisation analyses implicate candidate genes for serum CK. (A) Heatmap of SMR associations (−log10P) between eQTLs and pQTLs across heart, skeletal muscle, adipose, liver, whole blood, and circulating proteins versus CK GWAS signals. Asterisks indicate significant colocalisation. (B) Regional plots for ALPK3 in muscle (rs11073663), KLF5 in muscle (rs9543398), and LILRB5 in plasma proteins (rs12975366), showing QTL signals above and CK GWAS signals below. Variants are coloured by LD r2 with the lead SNP.

Fig. 4. Genetic correlations and pleiotropic effects of CK-associated loci. (A) Genome-wide genetic correlations between CK and 42 systemic traits estimated by LDSC. Traits are grouped as biomarkers (black), diseases (blue), and muscle-related traits (red). Horizontal bars denote 95% CI; red dashed line, zero correlation. (B) Heatmap summarising pleiotropic patterns across clusters. Each cell shows signed −log10(P) for the association between cluster membership and the trait (sign from the regression coefficient aligned to the CK-increasing allele). Values are truncated at ±20 for visualisation; positive indicates trait-increasing, negative indicates trait-decreasing effects. (C) Regional plot for rs12975366 showing colocalisation of IGF-1, AST, and CK signals. Dotted line, genome-wide significance (P = 5 × 10−8); points coloured by LD r2 with lead SNP. (D) Regional plot for rs1799938 demonstrating shared effects on creatinine, AST, and CK.