Comprehensive profiling of COVID-19 severity using NULISAseq: Predictive insights and immune dynamics from the IMPACC study 2209

Abstract Description Objective This study evaluates the ability of NULISAseq to predict COVID-19 severity and elucidate immune response dynamics, comparing its performance with the Olink platform using a subset of the IMPACC cohort. Methods We analyzed serum samples from 503 COVID-19 patients using NULISAseq and Olink immunoassay platforms. Both baseline (Elastic Net analysis) and longitudinal data were compared to assess predictive capability. Association rule mining (ARM) was employed to identify associations of protein markers and immune cell subsets. Results NULISAseq data indicated superior detectability and/or dynamic range across various targets. Key proteins such as Amphiregulin (AREG) predicted COVID-19 severity effectively, achieving a test AUC of 0.84 by Elastic Net analysis. NULISAseq also significantly distinguished longitudinal trends in multiple proteins, including IL5 and interferons, between low and high severity groups. ARM identified robust associations, particularly in severe cases, revealing early cytokine signatures predictive of later immune cell changes. Olink also performed well, achieving a higher test AUC of 0.87 and identifying several proteins with severity differential longitudinal profiles. Conclusions NULISAseq shows robust capability in predicting COVID-19 severity and providing detailed immune response profiles. These findings underscore its potential for comprehensive immune profiling and early identification of therapeutic targets in COVID-19 management. Funding Sources Supported by NIH 2U19AI057229 Topic Categories Computational and Systems Immunology (COMP)