Integrative Multi-omics Analysis Reveals Molecular Subtypes of HLA-B27-Positive Ankylosing Spondylitis: A Cross-Database Study

Abstract

Ankylosing spondylitis (AS) is a chronic inflammatory arthritis with significant heterogeneity despite over 90% of patients carrying HLA-B27. This study integrated 407,734 samples from GEO, FinnGen, IGAS, ArrayExpress, and ImmPort databases, including 13,945 HLA-B27 positive AS patients and 391,889 controls, to identify molecular subtypes and establish a precision classification system. Through unsupervised clustering of 1,285 patients with complete transcriptomic data, three distinct molecular subtypes were identified: inflammation-dominant (37.8%), fibrosis-progressive (32.1%), and immune-dysregulated (30.1%). An XGBoost classification model based on 73 core genes achieved 86.5% accuracy in independent validation (AUC=0.94). The subtypes exhibited distinct clinical characteristics and treatment responses: inflammation-dominant patients showed highest BASDAI scores (5.8±1.9) with 72.3% anti-TNF response rate; fibrosis-progressive type had highest mSASSS scores (38.5±18.2); immune-dysregulated type demonstrated best JAK inhibitor response (68.5%). A 35-gene minimal classification set maintained 85.2% accuracy while reducing detection costs. The risk scoring model showed good prognostic capability (C-index=0.78), with 5-year progression-free survival rates of 42.3%, 61.5%, and 68.9% for the three subtypes respectively (p=0.003). This molecular typing system reveals the heterogeneity and biological mechanisms of HLA-B27 positive AS, providing practical tools for individualized treatment strategies to improve clinical management and patient prognosis.