Cytokine biomarkers of COVID-19

ABSTRACT

We used a new strategy to screen cytokines associated with SARS-CoV-2 infection. Cytokines that can classify populations in different states of SARS-CoV-2 infection were first screened in cross-sectional serum samples from 184 subjects by 2 statistical analyses. The resultant cytokines were then analyzed for their interrelationships and fluctuating features in sequential samples from 38 COVID-19 patients. Three cytokines, M-CSF, IL-8 and SCF, which were clustered into 3 different correlation groups and had relatively small fluctuations during SARS-CoV-2 infection, were selected for the construction of a multiclass classification model. This model discriminated healthy individuals and asymptomatic and nonsevere patients with accuracy of 77.4% but was not successful in classifying severe patients. Further searching led to a single cytokine, hepatocyte growth factor (HGF), which classified severe from nonsevere COVID-19 patients with a sensitivity of 84.6% and a specificity of 97.9% under a cutoff value of 1128 pg/ml. The level of this cytokine did not increase in nonsevere patients but was significantly elevated in severe patients. Considering its potent antiinflammatory function, we suggest that HGF might be a new candidate therapy for critical COVID-19. In addition, our new strategy provides not only a rational and effective way to focus on certain cytokine biomarkers for infectious diseases but also a new opportunity to probe the modulation of cytokines in the immune response.