Mendelian randomization reveals no correlations between herpesvirus infection and idiopathic pulmonary fibrosis.

BACKGROUND

Previous studies have found that the persistence of herpesvirus significantly increases the risk of idiopathic pulmonary fibrosis (IPF), but it is unclear whether this effect is causal. We conducted a two-sample Mendelian randomization (MR) study to evaluate the causal relationship between three herpesvirus infections and IPF.

METHODS

We used genome-wide association studies (GWAS) data from three independent datasets, including FinnGen cohort, Milieu Intérieur cohort, and 23andMe cohort, to screen for instrumental variables (IVs) of herpesvirus infection or herpesvirus-related immunoglobulin G (IgG) levels. Outcome dataset came from the largest meta-analysis of IPF susceptibility currently available.

RESULTS

In the FinnGen cohort, genetically predicted Epstein-Barr virus (EBV) (OR = 1.105, 95%CI: 0.897-1.149, p = 0.815), cytomegalovirus (CMV) (OR = 1.073, 95%CI: 0.926-1.244, p = 0.302) and herpes simplex (HSV) infection (OR = 0.906, 95%CI: 0.753-1.097, p = 0.298) were not associated with the risk of IPF. In the Milieu Intérieur cohort, we found no correlations between herpesvirus-related IgG EBV nuclear antigen-1 (EBNA1) (OR = 0.968, 95%CI: 0.782-1.198, p = 0.764), EBV viral capsid antigen (VCA) (OR = 1.061, 95CI%: 0.811-1.387, p = 0.665), CMV (OR = 1.108, 95CI%: 0.944-1.314, p = 0.240), HSV-1 (OR = 1.154, 95%CI: 0.684-1.945, p = 0.592) and HSV-2 (OR = 0.915, 95%CI: 0.793-1.056, p = 0.225) and IPF risk. Moreover, in the 23andMe cohort, no evidence of associations between mononucleosis (OR = 1.042, 95%CI: 0.709-1.532, p = 0.832) and cold scores (OR = 0.906, 95%CI: 0.603-1.362, p = 0.635) and IPF were found. Sensitivity analysis confirmed the robustness of our results.

CONCLUSIONS

This study provides preliminary evidence that EBV, CMV, and HSV herpesviruses, and herpesviruses-related IgG levels, are not causally linked to IPF. Further MR analysis will be necessary when stronger instrument variables and GWAS with larger sample sizes become available.