Sjögren’s syndrome (SS) pathogenesis involves complex interactions between gut microbiota and immune dysregulation, yet causal relationships remain not fully elucidated. This Mendelian randomization (MR) study investigates whether gut microbiota influences SS risk through immune cell mediation. We performed univariable and mediation MR analyses using genome-wide association study data from FinnGen (3309 SS cases, Finnish), MiBioGen consortium (5959 individuals, Finnish), and GWAS Catalog (3757 individuals, Sardinian). Five MR methods were employed, with inverse-variance weighted (IVW) as the primary analysis. Robustness was assessed through Robust IVW, Penalized IVW, and Penalized Robust IVW methods. Supplementary methods further strengthened robustness. Heterogeneity and pleiotropy were assessed via Cochran Q test and MR-Egger intercept, respectively, while MR-PRESSO identified and corrected outliers. P-values were adjusted using the Benjamini-Hochberg FDR (False Discovery Rate) method. Suggestive causal relationships between gut microbiota, immune cells, and SS were identified based on nominal P 0.05). Specifically, Bifidobacteriaceae, Bacteroides A, and UBA7703 exhibited suggestive causal effects on SS. The suggestive causal effect of Bifidobacteriaceae on SS might be mediated by CD3 expression on CD28 + CD45RA- CD8 + T cells (9.28%, OR = 0.983, 95%CI 0.96–1.01). Furthermore, CD4 on secreting CD4 regulatory T cells (4.59%, OR = 1.01, 95%CI 0.997–1.03) was found to potentially mediate the causal relationship between Bacteroides A and SS. Additionally, CD8 on CD28 + CD45RA- CD8 + T cells (12.14%, OR = 1.08, 95%CI 0.978–1.20) presented a suggestive mediating effect on the causal relationship between UBA7703 and SS. This exploratory study provides hypothesis-generating genetic evidence for suggestive causal links between specific gut microbiota taxa and SS, which may be partially mediated by immune cell dysregulation. Findings highlight potential candidate therapeutic targets for SS, though mechanistic and clinical validation is needed.
Yang et al. (Fri,) studied this question.