Abstract
COVID-19, a novel coronavirus-related illness, has spread worldwide. Patients with apparently mild/moderate symptoms can suddenly develop severe pneumonia. Therefore, almost all COVID-19 patients require hospitalization, which can reduce limited medical resources in addition to overwhelming medical facilities. To identify predictive markers for the development of severe pneumonia, a comprehensive analysis of serum chemokines and cytokines was conducted using serial serum samples from COVID-19 patients. The expression profiles were analyzed along the time axis. Serum samples of common diseases were enrolled from a BioBank to confirm the usefulness of predictive markers. Five factors, IFN-λ3, IL-6, IP-10, CXCL9, and CCL17, were identified as predicting the onset of severe/critical symptoms. The factors were classified into two categories. Category A included IFN-λ3, IL-6, IP-10, and CXCL9, and their values surged and decreased rapidly before the onset of severe pneumonia. Category B included CCL17, which provided complete separation between the mild/moderate and the severe/critical groups at an early phase of SARS-CoV-2 infection. The five markers provided a high predictive value (area under the receiver operating characteristic curve (AUROC): 0.9-1.0, p < 0.001). Low expression of CCL17 was specifically observed in pre-severe COVID-19 patients compared with other common diseases, and the predictive ability of CCL17 was confirmed in validation samples of COVID-19. The factors identified could be promising prognostic markers to distinguish between mild/moderate and severe/critical patients, enabling triage at an early phase of infection, thus avoiding overwhelming medical facilities.