Large meta-analysis identifies 30 new genetic loci for ankylosing spondylitis risk

Ankylosing spondylitis (AS) is a chronic inflammatory arthropathy with heritability estimated at approximately 90%, yet the effector genes and regulatory mechanisms beyond the well-established HLA-B*27 association remain incompletely defined. Translating GWAS-identified loci into biological insight requires integration of calibrated association statistics, gene-level prioritization, fine-mapping, molecular prediction, and experimental follow-up. We performed a GWAS meta-analysis of 7,551 AS cases and 1,258,581 controls from the Million Veteran Program, FinnGen, and UK Biobank whole-genome sequencing cohorts. The post-GWAS analysis combined causal transcriptome-wide association study (cTWAS) across four immune-relevant tissues, colocalization, four-method fine-mapping, Geneformer V2 in-silico perturbation, AlphaGenome variant-effect prediction, cross-trait LD-score regression (LDSC), pathway enrichment, structured druggability assessment, and siRNA knockdown in Jurkat T cells. We added LDSC calibration, per-cohort Q-Q plots, heterogeneity summaries, and leave-one-cohort-out sensitivity analyzes for revised quality control. The meta-analysis identified 30 genome-wide significant loci harboring 26,178 significant variants. After LDSC-compatible quality control, the meta-analysis showed lambda_GC = 1.09, LDSC intercept = 1.045 (SE = 0.009), and attenuation ratio = 0.365 (SE = 0.075); per-cohort lambda_GC values were 1.027, 1.080, and 1.018 for MVP, FinnGen, and UKB-WGS, respectively. cTWAS prioritized 64 causal-candidate genes (posterior inclusion probability [PIP] > 0.5), with gene expression explaining 19.5% of AS heritability. Seven genes showed convergent cTWAS and colocalization evidence (TBKBP1, TIMD4, HABP4, XCL1, USP22, ABO, ACTA2). Four-method fine-mapping identified 64 consensus variants, while heterogeneity and leave-one-cohort-out analyzes highlighted cohort-sensitive signals, particularly in the MHC and other high-I2 regions. Cross-trait LDSC showed positive genetic correlations with inflammatory bowel disease and psoriasis, but not rheumatoid arthritis or the available uveitis proxy. siRNA knockdown provided functional support for TBKBP1 and XCL1 in Jurkat T cells, while TIMD4 behaved as a context-specific myeloid/T-cell contrast. This convergent evidence framework prioritizes AS candidate genes and regulatory variants while separating statistical prioritization from experimental support. The revised analyzes strengthen calibration, heterogeneity reporting, cross-disease context, and translational interpretation, nominating immune and regulatory hypotheses for follow-up rather than definitive therapeutic targets.