Lack of replication of interaction between EBNA1 IgG and smoking in risk for multiple sclerosis

Background: Epstein-Barr virus infection, smoking, HLA-A*02, and DRB1*15 have all been proposed as risk factors for multiple sclerosis (MS). In 2010, Simon et al. described an interaction on the multiplicative scale between EBNA1 immunoglobulin G (IgG) and smoking regarding risk of MS, a finding that we attempted to replicate. Methods: This Swedish case-control study consisted of patients with newly diagnosed MS and matched controls. Using logistic regression, we analyzed association to MS risk and interactions between EBNA1 IgG and smoking, HLA-DRB1*15, and A*02, respectively, on the multiplicative scale. In addition, we analyzed interactions on the additive scale using attributable proportion due to interaction (AP). Results: We did not observe any interaction on the multiplicative scale between EBNA1 IgG and any of the 3 risk factors, smoking, DRB1*15, or absence of A*02, although in a conditional analysis the interaction with absence of A*02 becomes significant. However, we observed interactions on the additive scale between EBNA1 IgG and DRB1*15 (AP = 0.34, 95% confidence interval 0.11-0.57, p = 5 x 10(-3)) and between EBNA1 IgG and absence of A*02 (AP = 0.36, 0.13-0.59, p = 2 x 10(-3)) but not between smoking and DRB1*15 and EBNA1 IgG. The interaction between EBNA1 IgG and DRB1*15 was not significant in the conditional analysis. Conclusion: We did not observe any interaction between EBNA1 IgG and smoking, regardless of scale used, and thus did not replicate the observations from Simon et al. Neurology (R) 2012;79:1363-1368