Plasma proteomic profiles correlate with organ dysfunction in COVID-19 ARDS

Abstract

Severe COVID-19 is often complicated by hypoxemic respiratory failure and acute respiratory distress syndrome (ARDS). Mechanisms governing lung injury and repair in ARDS remain poorly understood. We hypothesized that plasma proteomics may uncover protein biomarkers correlated with COVID-19 ARDS severity. We analyzed the plasma proteome from 32 patients with ARDS and COVID-19 using an aptamer-based platform of 7289 proteins, and correlated protein measurements with sequential organ failure assessment (SOFA) scores at days 1 and 7 of ICU admission. We identified 184 differentially abundant proteins correlated with SOFA at day 1 and 46 proteins at day 7. In a longitudinal analysis, we correlated dynamic changes in protein abundance and SOFA between days 1 and 7 and identified 40 significant proteins. Pathway analysis of significant proteins identified increased ephrin signaling and acute phase response signaling correlated with increased SOFA scores between days 1 and 7, while pathways related to pulmonary fibrosis signaling and wound healing had a negative correlation. These findings suggest that persistent inflammation may drive disease severity, while repair processes correlate with improvements in organ dysfunction. This approach is generalizable to future ARDS cohorts for identification of biomarkers and disease mechanisms as we strive towards targeted therapies in ARDS.

Keywords: COVID‐19; SARS‐CoV‐2; acute respiratory distress syndrome; proteomic analysis.

FIGURE 1

Schematic diagram of the cohort design. Patients with ARDS and COVID‐19 were identified from the BuckICU biorepository. Patients without available biologic samples and those with certain co‐morbid illnesses were excluded. The primary analysis was designed to compare differentially abundant proteins between survivors and non‐survivors, and 32 patients were included in the final analysis. Secondary analyses included all 32 patients and correlated differential protein abundance to SOFA scores.

FIGURE 2

Differentially Abundant Proteins (DAPs) between days 1 and 7 for all analyzed samples. (a) Volcano plot of DAPS on Day 7 compared to Day 1. Red dots indicate increased protein abundance (positive fold change) on Day 7 compared to Day 1. Green dots indicate decreased protein abundance. Significance was defined as an adjusted p‐value (FDR) <0.1 and log2FC absolute value >0.5. The upregulated DAPS are represented by red dots and the downregulated proteins are in green. (b) Ingenuity pathway analysis (IPA) of the canonical pathways identified by the longitudinal analysis. The size of bubbles correlates the number of genes overlapping the pathway and the color corresponds to the directionality of the pathway, where the orange represents a positive z‐score (increased on Day 7 compared to Day 1) and the blue represents downregulation of a pathway. Here, −log (p‐value) >2.0 (equivalent to p‐value <0.01) and z‐sore >0.5 were used to filter significant pathways.

FIGURE 3

Day 1 differentially abundant proteins by SOFA scores. (a) Volcano plot of significant DAPS by SOFA scores on Day 1. Significance was defined as an adjusted p‐value (FDR) <0.1. DAPS positively correlated with SOFA score are red; negative correlations are green. (b) Scatter plots of the top 9 most significant DAPs, ordered alphabetically, plotting Day 1 array signaling intensity relative to protein abundance (x‐axis) against Day 1 SOFA scores (y‐axis). (c) Heat map illustrating the significant DAPS (FDR <0.1) with columns ordered by SOFA score. Blue to red color scale represents log2fold change. (d) Ingenuity pathway analysis (IPA) of canonical pathways for Day 1. The size of bubbles correlates the number of genes overlapping the pathway and the color corresponds to the directionality of the pathway, where the orange represents a positive z‐score (upregulated) and the blue represents downregulation of a pathway. Here, −log (p‐value) >2.0 (equivalent to p‐value <0.01) and a z‐sore >0.5 were used to filter significant pathways.

FIGURE 4

Day 7 differentially abundant proteins by SOFA scores. (a) Volcano plot of significant DAPS by SOFA scores on Day 7. Significance was defined as an adjusted p‐value (FDR) <0.1. DAPS positively correlated with SOFA score are red; negative correlations are green. (b) Heat map illustrating the significant DAPS (FDR <0.1) with columns ordered by SOFA score. Blue to red color scale represents log2fold change. (c) Scatter plots of the top 9 most significant DAPs, ordered alphabetically, plotting Day 7 protein abundance (x‐axis) against Day 7 SOFA scores (y‐axis).

FIGURE 5

Changes in protein abundance correlate to changes in SOFA scores between Days 1 and 7. (a) Heat map of the significant DAPS (FDR <0.1) when modeling change in protein abundance as an explanatory variable for change in SOFA scores between Days 1 and 7 with columns ordered by difference in SOFA scores over time (Days 7 SOFA score minus Day 1 SOFA score). Blue to red color scale represents log2fold change. (b) Scatter plots of the top 9 most significant DAPs, ordered alphabetically, plotting log2 fold change in protein abundance (x‐axis) against change in SOFA score (y‐axis). (c) Ingenuity pathway analysis (IPA) of canonical pathways for Day 1. The size of bubbles correlates the number of genes overlapping the pathway and the color corresponds to the directionality of the pathway, where the orange represents a positive z‐score (upregulated) and the blue represents downregulation of a pathway. Here, −log (p‐value) >2.0 (equivalent to p‐value <0.01) and a z‐sore >0.5 were used to filter significant pathways.