Genetically proxied circulating PD-1/PD-L1 levels and broadly defined myocarditis: A bidirectional Mendelian...

Myocarditis is an inflammatory myocardial disease with potentially severe outcomes. Programmed cell death protein 1 (PD-1) and programmed death-ligand 1 (PD-L1) regulate immune tolerance, but the association between lifelong genetically proxied circulating PD-1/PD-L1 levels and broadly defined myocarditis remains uncertain. We investigated these associations and explored related plasma lipid species. We conducted bidirectional 2-sample Mendelian randomization using proteomic genome-wide association data from the UK Biobank Pharma Proteomics Project and INTERVAL. FinnGen Release 10 was the primary broadly defined myocarditis outcome, and an independent myocarditis genome-wide association study (GCST90018882) provided outcome-level validation. Complementary estimators, heterogeneity and pleiotropy diagnostics, influence analyses, MR-RAPS, and supportive meta-analyses were performed. Associations with 179 plasma lipid species were examined in exploratory analyses. Higher genetically proxied circulating PD-L1 was inversely associated with broadly defined myocarditis in UKB-PPP (odds ratio [OR] 0.834, 95% confidence interval [CI] 0.698-0.995; P = .0441), and the independent INTERVAL analysis yielded a concordant inverse estimate (OR 0.619, 95% CI: 0.434-0.883; P = .0083); no clear association was observed for PD-1. The MR-RAPS estimate retained the inverse direction; estimates against the independent broadly defined myocarditis dataset were also inverse, and supportive meta-analyses across protein and outcome sources yielded inverse pooled estimates. Reverse MR did not support effects of broadly defined myocarditis liability on circulating PD-1 or PD-L1. Exploratory lipid analyses identified nominal associations requiring confirmation. Higher genetically proxied circulating PD-L1 may be associated with a lower risk of broadly defined myocarditis, supporting further investigation of PD-L1-related immune regulation. These findings do not directly estimate the effects of pharmacologic PD-1/PD-L1 blockade. The lipid findings are hypothesis-generating.