Early Th2 inflammation in the upper respiratory mucosa as a predictor of severe COVID-19 and modulation by early treatment with inhaled corticosteroids: a mechanistic analysis

Abstract

Background: Community-based clinical trials of the inhaled corticosteroid budesonide in early COVID-19 have shown improved patient outcomes. We aimed to understand the inflammatory mechanism of budesonide in the treatment of early COVID-19.

Methods: The STOIC trial was a randomised, open label, parallel group, phase 2 clinical intervention trial where patients were randomly assigned (1:1) to receive usual care (as needed antipyretics were only available treatment) or inhaled budesonide at a dose of 800 μg twice a day plus usual care. For this experimental analysis, we investigated the nasal mucosal inflammatory response in patients recruited to the STOIC trial and in a cohort of SARS-CoV-2-negative healthy controls, recruited from a long-term observational data collection study at the University of Oxford. In patients with SARS-CoV-2 who entered the STOIC study, nasal epithelial lining fluid was sampled at day of randomisation (day 0) and at day 14 following randomisation, blood samples were also collected at day 28 after randomisation. Nasal epithelial lining fluid and blood samples were collected from the SARS-CoV-2 negative control cohort. Inflammatory mediators in the nasal epithelial lining fluid and blood were assessed for a range of viral response proteins, and innate and adaptive response markers using Meso Scale Discovery enzyme linked immunoassay panels. These samples were used to investigate the evolution of inflammation in the early COVID-19 disease course and assess the effect of budesonide on inflammation.

Findings: 146 participants were recruited in the STOIC trial (n=73 in the usual care group; n=73 in the budesonide group). 140 nasal mucosal samples were available at day 0 (randomisation) and 122 samples at day 14. At day 28, whole blood was collected from 123 participants (62 in the budesonide group and 61 in the usual care group). 20 blood or nasal samples were collected from healthy controls. In early COVID-19 disease, there was an enhanced inflammatory airway response with the induction of an anti-viral and T-helper 1 and 2 (Th1/2) inflammatory response compared with healthy individuals. Individuals with COVID-19 who clinically deteriorated (ie, who met the primary outcome) showed an early blunted respiratory interferon response and pronounced and persistent Th2 inflammation, mediated by CC chemokine ligand (CCL)-24, compared with those with COVID-19 who did not clinically deteriorate. Over time, the natural course of COVID-19 showed persistently high respiratory interferon concentrations and elevated concentrations of the eosinophil chemokine, CCL-11, despite clinical symptom improvement. There was persistent systemic inflammation after 28 days following COVID-19, including elevated concentrations of interleukin (IL)-6, tumour necrosis factor-α, and CCL-11. Budesonide treatment modulated inflammation in the nose and blood and was shown to decrease IL-33 and increase CCL17. The STOIC trial was registered with ClinicalTrials.gov, NCT04416399.

Interpretation: An initial blunted interferon response and heightened T-helper 2 inflammatory response in the respiratory tract following SARS-CoV-2 infection could be a biomarker for predicting the development of severe COVID-19 disease. The clinical benefit of inhaled budesonide in early COVID-19 is likely to be as a consequence of its inflammatory modulatory effect, suggesting efficacy by reducing epithelial damage and an improved T-cell response.

Funding: Oxford National Institute of Health Research Biomedical Research Centre and AstraZeneca.

Conflict of interest statement

Declaration of interests SR reports grants and non-financial support from Oxford Respiratory National Institute for Health Research (NIHR) Biomedical Research Centre (BRC), during the conduct of the study; and non-financial support from AstraZeneca and personal fees from Australian Government Research Training Program, outside of the submitted work. LED reports grants from AstraZeneca and Boehringer-Ingelheim, outside of the submitted work. PJB reports grants and personal fees from AstraZeneca and Boehringer Ingelheim, and personal fees from Teva and Covis, during the conduct of the study. REKR reports grants from AstraZeneca, and personal fees from Boehringer Ingelheim, Chiesi UK, and GlaxoSmithKline, during the conduct of the study. MB reports grants from AstraZeneca; personal fees from AstraZeneca, Chiesi, and GlaxoSmithKline; and is a scientific advisor for Albus Health and ProAxsis, outside of the submitted work. JRB, SPC, MM, DVN, and JSL declare no competing interests.

Copyright © 2022 The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY 4.0 license. Published by Elsevier Ltd.. All rights reserved.

Figures

Figure 1

Immunological features of the nasal mucosa in patients with early COVID-19 over time in the STOIC study A) Heatmap and B) volcano plot of 26 nasal mediators from 20 healthy individuals and 140 patients with community-based early COVID-19. C) Heatmap and D) volcano plot of 30 nasal mediators from patients with community based early COVID-19 at day 0 after enrolment in the trial and day 14 in the usual care group (n=60). E) Heatmap and F) volcano plot of 30 nasal mediators from patients with community based COVID-19 at days 0 and 14 in the budesonide group (n=62). Horizontal dotted line on volcano plots depicts the cutoff for statistical significance; the vertical dotted line represents the cutoff point determining whether mediator concentrations were higher significantly (right, red) or lower (left, blue) for day 14 samples compared with either healthy controls (unpaired) or paired day 14 samples. Black dots represent changes in mediator concentration. Data were analysed using Mann-Whitney t-test or Wilcoxon matched pairs signed rank test. IL=interleukin. TNF=tumour necrosis factor. IFN=interferon. TSLP=thymic stromal lymphopoietin. CXCL=CXC chemokine ligand. CCL=CC chemokine ligand. GM-CSF=granulocyte-macrophage colony-stimulating factor. VEGF=vascular endothelial growth factor. PDGF=platelet-derived growth factor. TGF=transforming growth factor.

Figure 2

Temporal changes in mediator concentrations in the usual care and budesonide groups A) Heatmap of 26 nasal mediators from 20 healthy individuals compared with samples at day 14 after recruitment (day 14) in the usual care group (n=60) and the budesonide group (n=62). B) List of significantly altered mediators compared with healthy controls at day 14 in both usual care and budesonide groups. C–E) Longitudinal analysis of mediator profiles in the usual care and budesonide groups displayed as representative best fit curves by smoothed spline analysis. Additional data are presented in appendix (p 9). IL=interleukin. TNF=tumour necrosis factor. IFN=interferon. TSLP=thymic stromal lymphopoietin. CXCL=CXC chemokine ligand. CCL=CC chemokine ligand. GM-CSF=granulocyte-macrophage colony-stimulating factor. VEGF=vascular endothelial growth factor. PDGF=platelet-derived growth factor. TGF=transforming growth factor.

Figure 3

Alterations in nasal mucosal inflammation in patients with early COVID-19 who clinically deteriorate A) Volcano plot of 26 nasal mediators from 20 healthy individuals and 11 patients who met the primary outcome of the study. Red dots represent significantly increased mediator concentrations, blue dots sifnificantly decreases, and black dots no changes. B–G) Significantly altered mediators at day 0 from patients with early COVID-19 (n=129) and those who met the primary outcome (n=11). B) GM-CSF. C) IL-10. D) IL-12. E) IL-2. F) IFN-α2a. G) IL-33. H) Comparison table of significantly altered mediators from nasal samples comparing healthy volunteers (n=20) and patients with early COVID-19 (n=129) without deterioration and with COVID-19 with deterioration (n=11). I) Change in concentrations of 30 mediators from day 0 to day 14 of 1 patient who required critical care respiratory support (see appendix p 12 for heatmap). Data were analysed using Mann-Whitney t-test. IL=interleukin. TNF=tumour necrosis factor. IFN=interferon. TSLP=thymic stromal lymphopoietin. CXCL=CXC chemokine ligand. CCL=CC chemokine ligand. GM-CSF=granulocyte-macrophage colony-stimulating factor. VEGF=vascular endothelial growth factor. PDGF=platelet-derived growth factor. TGF=transforming growth factor.

Figure 4

Persistence of systemic inflammation following 28–35 days of COVID-19 in the community Violin plots comparing some mediator concentrations in the serum of healthy individuals (n=20), those in the usual care group (n=61), and those in the budesonide group (n=62) of the study at 28–35 days following COVID-19 (see appendix p 13 for further results). (A) CCL11. (B) CCL13. (C) VEGF. (D) TSLP. (E) TNF-α. (F) IL-6. Data were analysed by Kruskal-Wallis with post-hoc Dunn's test. IL=interleukin. TNF=tumour necrosis factor. TSLP=thymic stromal lymphopoietin. CCL=CC chemokine ligand. VEGF=vascular endothelial growth factor.

Figure 5

Network analysis of mediator correlation data (A) Modularity maximisation and community detection for eigenvalue-based noise-cleaned nasal mucosal mediator data. (A1) Networks at initial SARS-CoV-2 infection. (A2) Networks after 14 days of initial SARS-CoV-2 infection. (A3) Networks after treatment with inhaled budesonide. Node size indicates eigenvalue centrality and arcs indicate non-zero noise-cleaned correlation. There are four modules of roughly equal size in each network. A1 and A2 are similar with 72% overlap indicating persistent relationship of inflammation over time. A1 and A3 have 27% overlap (which would be expected by a chance reassignment of nodes to the four modules) indicating that treatment with inhaled budesonide changes the inflammatory networks entirely. (B) Modularity maximisation and community detection for eigenvalue-based noise-cleaned serum data in participants from 28 to 35 days following SARS-CoV-2 infection. (B1) Network analysis of serum from participants in the usual care group. (B2) Network analysis of serum from participants in the budesonide group. The two networks are entirely different in terms of membership with 30% overlap (equivalent to a random node reassignment to modules). The network analysis shown here suggests that budesonide treatment dramatically rearranged the mediator pathways in serum. CCL=CC chemokine ligand. CRP=C-reactive protein. CXCL=CXC chemokine ligand. GM-CSF=granulocyte-macrophage colony-stimulating factor. IFN=interferon. IL=interleukin. PDGF=platelet-derived growth factor. TGF=transforming growth factor. TNF=tumour necrosis factor. TSLP=thymic stromal lymphopoietin. VEGF=vascular endothelial growth factor. vWF=von Willebrand factor.