Longitudinal analyses reveal immunological misfiring in severe COVID-19
Cardiff University review clinical immunology/immunity
First Author: Carolina Lucas
Journal/preprint name: Nature
Paper DOI: doi.org/10.1038/s41586-020-2588-y
Tags: Type 1 (antiviral), Type 2 (anti-helminth), Type 3 (antifungal) immunity, cytokines, immune signature, clinical correlates, dysfunction, growth factors.
Lucas et al., present novel insight into features of the dysregulated immune response to SARS-CoV-2 infection. It is newly shown that features of immune responses to pathogens other than viruses occur in COVID-19 disease.
CD4+ and CD8+ T cells were reduced in association with disease severity, but remaining T cells were activated and T cells in moderate and severe disease produced IFNg. Increased monocytes, eosinophils and low-density neutrophils correlated with disease severity, although neutrophils did not increase over time. Circulating monocytes had reduced HLA-DR expression.
COVID-19 core immune signalling signature: moderate and severe disease had high circulating IL-1a, IL-1b, IL-12p70, IL-17a and IFNa. Further markers specifically associated with severe disease included thrombopoietin (TPO), IL-33, IL-16, IL-21, IL-23, IFNl, eotaxin, eotaxin 3 and cytokines associated with cytokine release syndrome.
Longitudinal correlates: persistence of the ‘core signature,’ particularly IFNa, IFNl, and IL-1Ra, 10 or more days from onset of symptoms, occurred in severe patients, whereas they decreased in moderate patients.
Although viral titre did not correlate with disease severity, it did correlate with levels of certain signalling molecules: IFNa IFNg and TRAIL.
Patients clustered into 3 groups which had distinct acute immune features, disease severity and outcomes. The clusters were characterised by 4 immune signatures. Signature A included growth factors and wound healing factors. Signature B consisted of a mixture of Types 2 and 3 immunity signalling mediators. Signature C involved a mixed response including type 1, type 2 type 3 immunity. Signature D mainly featured chemokines. Cluster 1 patients experienced moderate disease and limited mortality, which were associated with Signature A. Clusters 2 and 3 patients both experienced more severe disease and increased mortality. Cluster 2 had higher levels of Signature C and D markers (i.e. mixed immune responses and chemokines, respectively). Cluster 3 patients had high levels of type 2 and 3 responses (Signature B) in addition to features of Signatures C and D.
Impact for COVID-19 research:
This paper presents a novel angle on the variable immunopathology in COVID-19 disease. It suggests that poor control of SARS-CoV-2 infection may occur not only due to dysfunction of the ‘appropriate’ antiviral immune response but also due to the activation of ‘inappropriate’ immune responses to other forms of pathogen, or a mix of these and the antiviral response, reducing the efficacy of the latter.
The longitudinal clustering analysis provides important insight into immune signatures of COVID-19 disease severity, clinical course and outcome. A biomarker window was described which could inform clinical practice: increased IFNa IFNl, IL-16, IL-2 and certain chemokines within 12 days of symptom onset predicted longer hospital stays and higher mortality.
Study Type: Longitudinal, Case- control study: COVID-19 patients (n=113; 253 data points over 7 longitudinal collection points) and healthcare workers (n=108 serving as controls.
Key Techniques: SARS-oV-2 infection was detected using the US CDC RT–qPCR primer/probe sets for 2019-nCoV_N1, 2019-nCoV_N2, and the human RNase P (RP) as an extraction control. Cytokines and chemokines were measured using the Human Cytokine Array/Chemokine Array 71-403 Plex Panel (HD71). Flow cytometry cell surface and intracellular analysis antibody information is provided: manufacturer, clone and dilution used. Groups comparisons were made by ANOVA (parametric) and Kruskal-Wallis (non-parametric) statistical tests with Dunn’s and Tukey’s post hoc tests, respectively. Clustering was performed using the k-means algorithm, correlations analysis by Spearman’s coefficient and risk ratio by Poisson regression.
Recruitment of healthcare workers as controls which relied on viral testing may not have exclude those who were negative at the time of testing but who previously contracted SARS-CoV-2, which could also be more likely due to their occupation.