Severe COVID-19 is characterized by a systemic deluge of cytokines, in some patients associated with an acute respiratory distress syndrome (ARDS). Circulating cytokines that can stratify risks and parallel disease course are useful for more effective triage and management of the patients. In this study we first ran a machine learning algorithm on a dataset of 36 cytokines measured in plasma collected from a cohort of severe COVID-19 patients, with a goal to identify cytokine/s most useful for describing the dynamic clinical state in a multiple regression analysis. We derived from a Bayesian Information Criterion analysis that a combination of interleukin-8 (IL-8), Eotaxin and IFN-gamma is significantly linked to blood oxygenation outcome in patients over seven days following plasma sampling. On individually testing the cytokines in receiver operator characteristics curve analysis we found that plasma level of IL-8 by itself, but not of Eotaxin or IFN-gamma, is a strong stratifier for final clinical outcomes. Moreover the circulating IL-8 dynamics closely paralleled the disease course, showing that an upward kinetics was linked to worse final outcomes and vice versa. RNAseq analysis on circulating blood cells also revealed key transitions in immune transcriptome in patients having low versus high circulating IL-8 at three different timepoints. Thus we identified plasma IL-8 as a key pathogenic circulating molecule linking systemic hyper-inflammation to the clinical outcomes in severe COVID-19 patients with ARDS.
Less...