Cytokine Elevation in Severe COVID-19 From Longitudinal Proteomics Analysis: Comparison With Sepsis

ABSTRACT

Introduction Coronavirus disease 2019 (COVID-19) is a new viral disease. Uncontrolled inflammation called “cytokine storm” is reported to contribute to disease pathogenesis as well as sepsis. We aimed to identify cytokines related to the pathogenesis of COVID-19 through a proteomics analysis of 1463 plasma proteins, validate these cytokines, and compare them with sepsis. Materials and Methods In a derivation cohort of 306 patients with COVID-19, 1463 unique plasma proteins were measured on days 1, 4, and 8. Cytokines associated with disease severity and prognosis were derived. In a validation cohort of 62 COVID-19 patients and 38 sepsis patients treated in the intensive care unit [ICU], these derived cytokines were measured on days 1 (day of ICU admission), 2-3, and 6-8 (maximum 3 time points/patient). Derived cytokines were compared with healthy controls and between COVID-19 and sepsis patients, and the associations with prognosis were evaluated. The time to wean off mechanical ventilation (MV) was evaluated only for COVID-19. Results IL-6, amphiregulin, and growth differentiation factor (GDF)-15 were associated with disease severity and prognosis in the derivation cohort. In the validation cohort, IL-6 and GDF-15 were elevated in COVID-19 and sepsis on day 1, and the levels of these cytokines were higher in sepsis than in COVID-19. IL-6 and GDF-15 were associated with prognosis in sepsis. Cox proportional hazards model with time as a dependent covariate showed a significant relationship between plasma GDF-15 level and time to wean off MV (hazard ratio, 0.549 [95% confidence level, 0.382–0.789]). The GDF-15 level at ICU admission predicted late recovery. Conclusion GDF-15 and IL-6 derived from proteomics analysis were related with disease severity of COVID-19. Their values were higher in sepsis than in COVID-19 and were associated with prognosis in sepsis. In COVID-19 patients treated in the ICU, GDF-15 was associated with the time to wean off MV and better predicted late recovery.