Plasma proteomic associates of infection mortality in UK Biobank

BackgroundInfectious diseases are a major cause of mortality in spite of existing public health, anti-microbial and vaccine interventions. We aimed to define plasma proteomic associates of infection mortality and then apply Mendelian randomisation (MR) to yield biomarkers that may be causally associated. MethodsWe used UK Biobank plasma proteomic data to associate 2,923 plasma proteins with infection mortality before 31st December 2019 (240 events in 52,520 participants). Since many plasma proteins also predict non-infection mortality, we focussed on those associated with >1.5-fold risk of infection mortality in an analysis excluding survivors. Protein quantitative trait scores (pQTS) were then used to identify whether genetically predicted protein levels also associated with infection mortality. To conduct Two Sample MR, we performed a genome-wide association study (GWAS) of infection mortality using UK Biobank participants without plasma proteomic data (n=363,953 including 984 infection deaths). FindingsAfter adjusting for clinical risk factors, 1,142 plasma proteins were associated with risk of infection mortality (false discovery rate <0.05). 259 proteins were associated with >1.5-fold increased risk of infection versus non-infection mortality. Of these, we identified genetically predicted increasing MERTK concentration was associated with increased risk of infection mortality. GWAS for infection mortality revealed no SNPs achieving genome-wide statistical significance (p<5x10-8). However, MR supported a causal association between increasing plasma MERTK protein and infection mortality (odds ratio 1.46 per unit; 95% CI 1.15-1.85; p=0.002). InterpretationPlasma proteomics demonstrates many proteins are associated with infection mortality. MERTK warrants exploration as a potential therapeutic target.