Background
A team from Osaka University performed a longitudinal blood proteomics study combined with clinical information, to define a new clinically useful molecular phenotype associated with prognosis in severely ill COVID-19 patients. This extensive study included two diverse cohorts in the discovery phase, combining data from a previous COVID-19 study from Massachusetts General Hospital with new data from a Japanese cohort, both using Olink® Explore 1536 to measure plasma proteomics. Key findings were then finally verified in a third independent cohort, using ELISAs to confirm the most important proteins identified.
Outcome
This identified four key proteins (WFDC2, GDF15, CHI3L1, and KRT19) that were involved in critical pathogenesis in all three independent cohorts. Machine learning was then applied to show distinct associations with 3 different clinically relevant phenotypes that they had defined. Interestingly, the 4 main protein markers identified are not associated with immune responses, but cell adhesion functionality, offering new insights into the pathogenesis of COVID-19 and potentially, new treatment possibilities.