Severe COVID-19 and non-COVID-19 severe sepsis converge transcriptionally after a week in the intensive care unit, indicating common disease mechanisms

Abstract

Introduction: Severe COVID-19 and non-COVID-19 pulmonary sepsis share pathophysiological, immunological, and clinical features. To what extent they share mechanistically-based gene expression trajectories throughout hospitalization was unknown. Our objective was to compare gene expression trajectories between severe COVID-19 patients and contemporaneous non-COVID-19 severe sepsis patients in the intensive care unit (ICU).

Methods: In this prospective single-center observational cohort study, whole blood was drawn from 20 COVID-19 patients and 22 non-COVID-19 adult sepsis patients at two timepoints: ICU admission and approximately a week later. RNA-Seq was performed on whole blood to identify differentially expressed genes and significantly enriched pathways.

Results: At ICU admission, despite COVID-19 patients being almost clinically indistinguishable from non-COVID-19 sepsis patients, COVID-19 patients had 1,215 differentially expressed genes compared to non-COVID-19 sepsis patients. After one week in the ICU, the number of differentially expressed genes dropped to just 9 genes. This drop coincided with decreased expression of antiviral genes and relatively increased expression of heme metabolism genes over time in COVID-19 patients, eventually reaching expression levels seen in non-COVID-19 sepsis patients. Both groups also had similar underlying immune dysfunction, with upregulation of immune processes such as "Interleukin-1 signaling" and "Interleukin-6/JAK/STAT3 signaling" throughout disease compared to healthy controls.

Discussion: Early on, COVID-19 patients had elevated antiviral responses and suppressed heme metabolism processes compared to non-COVID-19 severe sepsis patients, although both had similar underlying immune dysfunction. However, after one week in the ICU, these diseases became indistinguishable on a gene expression level. These findings highlight the importance of early antiviral treatment for COVID-19, the potential for heme-related therapeutics, and consideration of immunomodulatory therapies for both diseases to treat shared immune dysfunction.

Keywords: COVID-19; gene expression; immune dysfunction; longitudinal analyses; sepsis.

Figure 1

Sampling times and hospitalization duration of ICU patients from the COLOBILI cohort. Diamonds indicate time of sampling for each patient, with the majority of D1 samples collected at Day 1 in the ICU, and D7 samples at Day 7 in the ICU. The solid bars represent the duration of hospital stay (cutoff at 28 days post-ICU admission). X indicates death in the ICU.

Figure 2

SARS-CoV-2 infection strongly influenced gene expression only at D1. Shown is a principal component analysis of analyzed ICU patients. The first two principal components were plotted and labelled based on COVID-19 status (yellow = SARS-CoV-2 positive, purple = SARS-CoV-2 negative) from D1 samples (A) and D7 samples (B). X’s indicate patients who died in hospital. Density plots on the sides show the distribution of samples by COVID-19 status across the two principal components. (C) Percent variance of gene expression attributed to different metadata variables at D1 and D7. Notably, the percent variance attributed to SARS-COV-2 positivity strongly decreased from D1 to D7. In contrast, variance at D7 was more attributed to ICU mortality.

Figure 3

COVID-19 and non-COVID-19 sepsis patients differed at D1 but converged to nearly identical transcriptional profiles at D7. (A) Volcano plots of genes differentially expressed (DE) between COVID-19 (Positive) and non-COVID-19 sepsis (Negative) patients at D1 (top) and D7 (bottom). Coloured dots represent DE genes (absolute fold change ≥1.5, adjusted P-value <0.05; cut-offs indicated by dotted lines). The top 5 up- and down- regulated annotated genes (lowest adjusted p-value and highest fold change) are labelled. (B) Subset of enriched Reactome pathways (top) and Hallmark gene sets (bottom) using DE genes between COVID-19 (Pos) and non-COVID-19 sepsis (Neg) patients at D1 and D7, with all enriched pathways shown in Figures S3 , S4 . No pathways were enriched amongst the 9 DE genes at D7. “Upregulated” pathways/gene sets (Δ) had genes that were overrepresented in upregulated DE genes when compared to their prevalence in the genome, suggesting an increase in their function or activity, and vice versa for “downregulated” pathways/gene sets (∇). The total number of DE genes in each comparison are shown under each label.

Figure 4

COVID-19 and non-COVID-19 sepsis patients shared immune and immune-related pathways at both D1 and D7. Subset of the enriched Reactome pathways (top) and Hallmark gene sets (bottom) using DE genes at D1 and D7 between COVID-19 (Pos) patients and healthy controls (HC), and between non-COVID-19 sepsis (Neg) patients and healthy controls. The full list of enriched pathways and gene sets are shown in Figures S3 , S4 . “Upregulated” pathways/gene sets (Δ) had genes that were overrepresented in upregulated DE genes when compared to their prevalence in the genome, suggesting an increase in their function or activity. The total numbers of DE genes in each comparison are shown under each label.

Figure 5

Gene expression trajectories of distinct and shared DE genes in COVID-19 and non-COVID-19 sepsis patients over time. (A) Subset of the enriched Reactome pathways (top) and Hallmark gene sets (bottom) using DE genes over time in COVID-19 (Positive) and non-COVID-19 sepsis (Negative) patients, separated into distinct (left) and shared (right) enriched pathways. The full list of enriched pathways/gene sets is shown in Figures S4 , S5 . “Upregulated” pathways/gene sets (Δ) had genes that were overrepresented in upregulated DE genes when compared to their prevalence in the genome, suggesting an increase in their function or activity, and vice versa for “downregulated” pathways/gene sets (∇). The total numbers of DE genes in each comparison are shown under each label. For one pathway, both up- and down- regulated genes were enriched (indicated by *); the direction with the lower adjusted p-value (more significantly enriched) is shown. The lower panels show mean DESeq2 normalized counts for representative genes involved in the antiviral response (B), heme metabolism (C), and interleukin-1 signaling (D). Lines are coloured as yellow = SARS-CoV-2 positive, purple = SARS-CoV-2 negative. Genes in the antiviral response and heme metabolism significantly changed over time in COVID-19 patients but not in non-COVID-19 sepsis patients, while genes in interleukin-1 signaling significantly decreased over time in both patient groups. Statistically significant differences in panels B-D are indicated as ***p<0.001, **p<0.01, and *p<0.05; significance values were derived from DESeq2 model results. ns, not significant.