Abstract
Background The coronavirus disease 2019 (COVID-19) pandemic increased the demand for reliable tests to predict disease severity and mortality. Methods In training cohort, we obtained traditional clinical data and plasma proteomics performed using the Olink proteomics platform from 52 fatal COVID-19 cases (COVID-19-F), 50 severe COVID-19 cases (COVID-19-S), 55 moderate/mild COVID-19 cases (COVID-19-M), and 54 healthy controls. Receiver operating characteristic (ROC) curves and logistic regression were applied to judge the accuracy of biomarkers to predict in-hospital mortality and build combined panel. An independent external cohort was used for validation. Results In total, 19 clinical parameters and 92 proteins were assessed. Traditional clinical indices did not show adequate predictive value of short-term mortality in severe COVID-19. In proteomics analysis, 75 proteins were differentially expressed among the four groups. Pathway analysis revealed an imbalance of inflammatory responses and excessive immunity in COVID-19-F. Subsequently, a novel plasma biomarker panel (including interleukin 8 and osteoprotegerin) was developed, with AUC values of 0.791 and 0.781 when comparing COVID-19-F to COVID-19-M or COVID-19-S, respectively. The predictive power of the panel was verified in an external cohort. Conclusions Our standardized assays yielded a prediction panel of mortality during hospitalization in patients with COVID-19.