Cookies on this website
We use cookies to ensure that we give you the best experience on our website. If you click 'Continue' we'll assume that you are happy to receive all cookies and you won't see this message again. Click 'Find out more' for information on how to change your cookie settings.

Julika Neumann, Teresa Prezzemolo , Lore Vanderbeke , Carlos P. Roca , Margaux Gerbaux , Silke Janssens , Mathijs Willemsen , Oliver Burton , Pierre Van Mol , Yannick Van Herck , CONTAGIOUS co-authors , Joost Wauters , Els Wauters , Adrian Liston , Stephanie Humblet-Baron

Link to paper:

Journal/Pre-Print: MedRxiv

Tags: Immunology/Immunity, T-cell phenotypes, Bioinformatics

Research Highlights 

1. High similarity between healthy controls, mild/moderate and severe COVID-19 patients in the subtype frequency and phenotype of circulating CD4+ and CD8+ T cells.

2. Severe COVID-19 patients saw an expansion in their circulating IL-10 producing CD25high FoxP3+ regulatory T cell (Treg) population, despite lymphopenia.

3. Data are openly available at, facilitating comparisons of gating across studies and perhaps in the future allowing data to be re-analysed


Neumann et al. use machine learning to identify changes in immunological signatures in peripheral blood from 23 mild/moderate, 20 severe COVID-19 patients and 6 healthy controls by flow cytometry. Large panels (~20 parameters) were designed to investigate: the lineage composition of circulating immunity, conventional CD4+ T cell phenotype, Treg phenotype and CD8+ T cell phenotype. The only identifiable change was an increase in IL-10 producing Tregs in severe patients. There is a trend towards increased CD4+ TCM cells over TEM in severe patients and a modest increase in IFNγ and IL-17 producing CD8+ TEM. The principal strength of this paper is the publication of data in an open repository, allowing scrutiny of data, re-analysis of gating, and perhaps in the future pooling of data if methodologies are standardised.

Impact for SARS-CoV2/COVID19 research efforts

Understand the immune response to SARS-CoV2/COVID19

Study Type

· Cross-sectional observational clinical study (mild (n=23), Severe (n=20) COVID-19 patient samples)

Strengths and limitations of the paper

Novelty: High dimensional flow cytometry characterisation of T cell subtypes with machine learning-based analysis with a comparatively high number of patient cases to existing literature. Data is openly accessible for 3rd party analysis.

Standing in the field: Several studies describing circulating lymphocyte phenotypes have been released and although many are underpowered, changes in lymphocyte activation have been associated with improved clinical outcomes. Several have noted changes in T cell differentiation and activation states particularly between healthy controls and severe/critical patients. This study does not identify such changes but may be underpowered. The main strength of the paper is in the transparency of the data which can be accessed at

Appropriate statistics: Yes. Test to describe Figure 3D should be added

Viral model used: SARS-CoV-2 infected patients from Belgium between 27th march and 17th April 2020.

Translatability: Understanding the immune response will help guide target discovery, vaccine responses and therapeutic windows.

Main limitations:

· More than 6 healthy donors are required to reduce the variation in this group for high dimensional analysis.

· Seroconversion or PCR status of the healthy donor samples not included.

· The WHO severity score is a progressive scale, some patients may move between scores 4 (hospitalised, no oxygen) and 5 (hospitalised, oxygen by mask/nasal prongs) very easily making the clustering proposed of mild and severe artificial. The most severe patients have not been included (WHO 8-9). Final clinical outcome is not detailed.

· While the study evaluated a sizable number of samples (23 mild/moderate and 20 severe patients), caution is still warranted in interpreting the results. Flow cytometry studies often need large numbers of patients and more robust clinical clustering to identify changes. The lack of changes seen may simply be due to underpowering.

· Considering the dynamic temporal changes in immunity faced with a viral infection, the time from onset/from hospitalisation should be included for each patient. Addition of this information to analysis may clarify the phenotype changes.

· Although outside the scope of this paper, the inclusion of longitudinal follow-up, convalescent and asymptomatic patients, T cell antigen-specifity would be informative in future studies.

· No comparison with other virus-induced pathologies, such as influenza, to confirm or exclude the exclusive nature of the Treg phenotype in COVID-19 patients.