Fig 1: Plasma biomarkers of long COVID compared with non-long COVID SARS-CoV-2 convalescences(A) Targeted quantitative measurements of plasma biomarkers of 34 long COVID (LC) patients and 15 post-SARS-CoV-2 infection healthy controls (HC): (A) PCA of all markers showing partial separation of LC (green) and controls (white); symbols denote sex. PC2 variable loadings indicating contributors to case-control separation (negative, higher in controls; positive, higher in LC).(B–D) Boxplots (scale in log10 pg/mL) for IFN-γ, IFN-β, and GFAP comparing LC vs. controls.(E–G) Boxplots for LC subgroup comparisons: GFAP, NFL, and IFN-β across HC/LC-1/LC-2/LC-3.Boxes show median and interquartile range; points are individuals. Numbers printed on the plots are p values with the Benjamini-Hochberg (BH)-adjusted p value in parentheses from linear models adjusted for age, sex, and days since infection. ns, not significant.
Fig 2: Plasma proteomics clustering analysis identified biomarkers distinguishing three long COVID subgroups(A) Partial least-squares discriminant analysis of plasma proteome shows a separation of LC-1 from LC-2 and LC-3 on PC1, while LC-2 and LC-3 segregated on PC2.(B and C) Gene set enrichment analysis of PC1 and PC2 loadings. Bar length shows −log10(q value); color tiles show normalized enrichment scores (NES). “pos” and “neg” indicate enrichment toward the positive or negative direction of the PC (PC1+, LC-1; PC2+, LC-2).(D) Heatmap of the top 200 loadings from PC1/PC2 clustered into six protein modules (C1–C6) with distinct patterns; example annotations include GFAP/NFL (LC-1 linked), IL-12/HIF-1α (LC-2 linked), and TNFSF9/IL4R/TGFBR1 (LC-3 linked).(E) Tissue enrichment of differentially regulated proteins (one vs. rest) for each subgroup.Black-bordered squares indicate q < 0.05. GO, Gene Ontology; PID, Pathway Interaction Database.
Supplier Page from Abcam for Recombinant Mouse GFAP protein (His tag)