Multi-omics Mendelian randomization and cartilage transcriptomics prioritize genetic loci and candidate genes for osteoarthritis
Аннотация
Osteoarthritis (OA) is a complex joint disorder in which inherited susceptibility intersects with immune and metabolic processes. We integrated gut-microbiota, 731 immune-cell-trait and 1,400 plasma-metabolite genome-wide association study summary statistics with an OA genome-wide association study ( n = 403,124) using two-sample Mendelian randomization. All screening analyses are reported with false-discovery-rate (FDR) control, and nominal findings are framed as exploratory. Associations passing prespecified sensitivity filters underwent colocalization, summary-data-based Mendelian randomization with the HEIDI test using blood cis-eQTLs, and expression assessment in a merged 43-sample cartilage resource. Screening identified 8 microbial taxa, 39 immune traits and 97 metabolites or ratios at nominal significance; no exposure–OA association survived FDR correction. Two associations, the glycine-to-serine ratio and CD45RA on TD CD8br, replicated at Bonferroni significance in FinnGen (61,356 cases, 315,115 controls) with concordant effect sizes, whereas neither colocalization signal reproduced. In the discovery analysis, colocalization provided moderate support for a shared signal between a long-chain sphingomyelin trait (GCST90200134) and OA at SPTLC3 (PP.H4 = 0.82; stable for p12 ≥ 1 × 10⁻^5 but lower under the most conservative prior), whereas the signal between a CD8 T-cell phenotype and OA at DHX16 was weak (PP.H4 = 0.49) and depended on the analysis window. Integration of SNP-annotated and SMR-prioritized genes yielded 92 candidates; 60 were assessable in cartilage; six showed nominal expression differences and one (TBC1D17) survived FDR correction. Together, the results provide a hypothesis-generating prioritization of one moderate colocalization signal and cartilage-expressed candidate genes, requiring validation in tissue-specific experimental systems.