From paradox to biology: shared genetic architecture underlies the inverse association between smoking and Sjögren's disease.

Abstract

Smoking has been consistently associated with a paradoxically lower risk of Sjögren's disease (SjD). We used genetic approaches to investigate whether this association reflects a causal effect of smoking or shared genetic architecture, and to identify shared loci that may help explain this paradox and inform disease biology. We applied an integrated cross-trait genetic framework using genome-wide association data for smoking initiation and SjD. Conventional Mendelian randomization (MR) was used as an initial test of association, followed by genome-wide causal modelling (CAUSE and latent causal variable analysis) to distinguish directional causality from shared genetic architecture. Shared loci were identified using cross-trait pleiotropy analysis, refined through colocalization, and further explored using proteome-informed mediation analyses. Conventional MR showed an inverse association between genetic liability to smoking initiation and SjD risk (OR 0.53, 95% CI 0.34-0.84). However, genome-wide modelling did not support a causal effect and instead suggested the association was driven by shared genetic influences. Cross-trait pleiotropy analysis identified 1536 variants across 11 loci, with colocalization supporting shared causal variants at five regions (3q13.32, 2q22.3, 8p21.2, 14q24.3 and 5q34). These loci implicated pathways related to neuroimmune signalling, epithelial organization and interferon regulation. Proteome-informed analyses identified no circulating proteins that robustly mediated the smoking-SjD relationship. The inverse smoking-SjD association likely reflects shared genetic liability rather than a protective causal effect. These findings highlight biological pathways relevant to SjD pathogenesis and emphasize the importance of cautious interpretation of MR estimates when traits share genetic architecture.