Cross-Platform Comparison of Highly Sensitive Immunoassays for Inflammatory Markers in a COVID-19 Cohort

Abstract

A variety of commercial platforms are available for the simultaneous detection of multiple cytokines and associated proteins, often employing Ab pairs to capture and detect target proteins. In this study, we comprehensively evaluated the performance of three distinct platforms: the fluorescent bead-based Luminex assay, the proximity extension-based Olink assay, and a novel proximity ligation assay platform known as Alamar NULISAseq. These assessments were conducted on human serum samples from the National Institutes of Health IMPACC study, with a focus on three essential performance metrics: detectability, correlation, and differential expression. Our results reveal several key findings. First, the Alamar platform demonstrated the highest overall detectability, followed by Olink and then Luminex. Second, the correlation of protein measurements between the Alamar and Olink platforms tended to be stronger than the correlation of either of these platforms with Luminex. Third, we observed that detectability differences across the platforms often translated to differences in differential expression findings, although high detectability did not guarantee the ability to identify meaningful biological differences. Our study provides valuable insights into the comparative performance of these assays, enhancing our understanding of their strengths and limitations when assessing complex biological samples, as exemplified by the sera from this COVID-19 cohort.

Promises and Challenges of populational Proteomics in Health and Disease

“…However, the same protein measurement may differ between proteomic assay approaches and this would have an impact in the transferability and generalizability of proteomic models across technology. A recent head-to-head study between SomaScan and Olink showed a range of poor to high Spearman’s correlation for overlapping proteins with a median from studies ranging from 0.20-0.44 ( 35 , 36 , 37 ) whilst in a study comparing three affinity-based platforms in a COVID-19 cohort showed median Kendell correlation ranging 0.36 to 0.70 between those platforms ( 38 ). For proteins with disparaging measures, it is difficult to establish one as more accurate than the other in the absence of gold standards.…”