Robust prognostic signatures derived from proteomic panel profiling of plasma from patients with severe COVID-19 complications

Abstract COVID-19 complications present a huge burden on healthcare systems and warrant a predictive risk model for disease severity of SARS-CoV-2 infection to enable early intervention, prospective decision-making and triaging of patients. We profiled plasma proteins from COVID-19 patients (severe n=50, and mild n=50) and controls (n=50) using function- and pathway-based panels developed with the highly specific proximity extension assays. Several biological pathways were specific for patients with severe complications. Based on these dysregulated profiles, we propose candidate FDA-approved drugs that target multiple upregulated proteins to treat severe complications. In addition, the set of differentially expressed plasma proteins in severe disease contained a robust 46-protein signature, the COVID-19 molecular severity score, which predicts the risk of severe complications. We cross-validated this molecular severity score in an independent cohort and found it useful within three days after hospital admission to predict COVID-19 severity and outcomes. Associated with the molecular severity score, we identified a set of clinical parameters available at admission, that act as a clinical risk score for complications. The molecular and clinical risk scores described in our study may be prognostic tools for severe COVID-19 disease and help alleviate the pressure on healthcare systems during infection peaks.