关键词: 强直性脊柱炎, Epstein-Barr 病毒, 抗体, 因果关系, 广义汇总数据孟德尔随机化, 异质性依赖工具检验, 工程化组织构建

Abstract: BACKGROUND: Epstein-Barr virus (EBV), a human herpesvirus, exhibits widespread prevalence in the global population. Previous observational studies have suggested an association between EBV infection and ankylosing spondylitis; however, conventional methodologies are limited by confounding factors and reverse causation bias, precluding definitive causal inference. Elucidating the causal relationship between EBV and ankylosing spondylitis is critical for uncovering the immune pathogenesis of ankylosing spondylitis and establishing a theoretical foundation for EBV-targeted preventive strategies.
OBJECTIVE: To investigate the bidirectional causal relationship between EBV infection and ankylosing spondylitis using bidirectional Mendelian randomization (MR) analysis.
METHODS: Bidirectional two-sample MR analysis was conducted using European population-wide genome-wide association study (GWAS) summary statistics. The generalized summary data MR approach was integrated with conventional methods, including inverse-variance weighted, MR-Egger, and weighted median estimators, to assess bidirectional causality between EBV infection and ankylosing spondylitis. EBV antibody data (EA-D, EBNA-1, VCA p18, and ZEBRA) were sourced from the UK Biobank, while ankylosing spondylitis data were obtained from FinnGen. Instrumental variables were selected based on genome-wide significance (P < 5×10⁻⁶), with exclusion of linkage disequilibrium and confounding single-nucleotide polymorphisms associated with smoking, rheumatoid arthritis, or psoriasis. Statistical significance was determined using Bonferroni correction (threshold: P=6.3×10-3), supplemented by sensitivity analyses for heterogeneity (Cochran’s Q), pleiotropy (MR-Egger intercept, MR-PRESSO), and robustness (leave-one-out analysis). The generalized summary data MR approach further applied HEIDI-outlier testing (P < 0.01) to eliminate pleiotropic single-nucleotide polymorphisms, ensuring causal inference validity.
RESULTS AND CONCLUSION: (1) Bidirectional MR revealed that EBV infection significantly increased the risk of ankylosing spondylitis: Elevated EBNA-1 antibody levels showed a positive association with ankylosing spondylitis [odds ratio (OR) = 1.41, 95% confidence interval (CI): 1.14–1.76, P = 0.002], while ZEBRA antibodies demonstrated a stronger effect (OR = 1.56, 95% CI: 1.31–1.85, P = 5.4×10⁻⁷), suggesting both latent (EBNA-1) and lytic (ZEBRA) EBV phases may drive ankylosing spondylitis pathogenesis via cross-reactive immune responses. (2) Reverse MR analysis indicated an inverse correlation between ankylosing spondylitis and EBV lytic-phase EA-D antibodies (OR=0.96, 95% CI: 0.94–0.98, P=3.25×10⁻⁴), indicating that ankylosing spondylitis-associated immune profiles may suppress EBV reactivation. (3) All results remained robust in heterogeneity, pleiotropy, and sensitivity analyses, with no detectable bias. (4) This study is the first to confirm EBV infection as an independent causal risk factor for ankylosing spondylitis based on international databases and European population data. Although genetic heterogeneity exists across populations, these findings in European populations provide critical insights into shared immune mechanisms underlying ankylosing spondylitis. Future validation in Chinese cohorts is warranted to characterize EBV-ankylosing spondylitis molecular interactions specific to this population. Furthermore, the generalized summary data MR framework exemplifies the utility of public GWAS data for causal inference, offering a paradigm for efficient risk factor discovery and helping the development of precision medicine. 

Key words: ankylosing spondylitis, Epstein-Barr virus, antibodies, causal relationship, Mendelian randomization of generalized summary data, heterogeneity-dependent instrument test, engineered tissue construction