Gene Expression Risk Scores for COVID-19 Illness Severity

ABSTRACT

The correlates of COVID-19 illness severity following infection with SARS-Coronavirus 2 (SCoV2) are poorly understood. While several demographic and underlying clinical variables increase risk for severe outcomes, at the onset of symptoms, it is challenging to identify those who will progress to requiring intensive care support. We conducted a pilot study to understand peripheral blood gene expression correlates of COVID19 illness across the spectrum of disease severity. We assessed gene expression in 53 confirmed SCoV2-infected adult participants during acute illness (within 28 days of onset). We found global gene expression patterns in participants with mild and moderate illness were highly similar, but significantly different from participants with severe illness. When comparing gene expression in those with severe as compared to non-severe illness, we identified >4000 genes significantly differentially expressed (FDR<0.05). Biological pathways represented by genes significantly increased in severe COVID19 were associated with platelet activation and coagulation, while those significantly decreased in severe COVID19 were associated with T cell signaling and differentiation (Figure 1). We used statistical modeling with crossvalidation to identify an 18-gene signature which classified severe illness (ROC AUC=0.98) in our training cohort, and strongly predicted hospitalization in an independent test cohort (ROC AUC=0.85). A weighted gene expression risk score (WGERS) provided 100% sensitivity and 85% specificity for classifying severe illness in our training cohort, and only mis-classified (5/19) participants with moderate illness. Importantly, the WGERS demonstrated 84% sensitivity and 74% specificity for predicting hospitalization in the test cohort. These data indicate that gene expression classifiers may provide clinical utility for identifying participants likely to require intensive care following SCoV2 infection.