Longitudinal assessment of IFN-I activity and immune profile in critically ill COVID-19 patients with acute respiratory distress syndrome

Between March and May 2020, critically ill patients admitted to three ICUs from academic hospital (Hospices Civils de Lyon, Lyon, France) who presented with pulmonary infection with SARS-CoV-2 confirmed by RT-PCR testing were prospectively included in the study. A flowchart describing the patient datasets used for the different analyses is provided in Additional file 1: Figure S1. Preliminary results from a subgroup of the cohort were published previously [9]. This project was part of an ongoing prospective observational clinical study (RICO, REA-IMMUNO-COVID). It was approved by ethics committee (Comité de Protection des Personnes Ile de France 1 - N°IRB/IORG #: IORG0009918) under agreement number 2020-A01079-30. This clinical study was registered at ClinicalTrials.gov (NCT04392401). The committee waived the need for written informed consent because the study was observational, with a low risk to patients, and no specific procedure, other than routine blood sampling, was required. Oral information and non-opposition to inclusion in the study were mandatory and were systematically obtained before any blood sample was drawn. This was recorded in patients’ clinical files. If a patient was unable to consent directly, non-opposition was obtained from the patient’s legally authorized representative and reconfirmed from the patient at the earliest opportunity. Inclusion criteria were: patients aged > 18 years, diagnosis of COVID-19 confirmed by RT-PCR testing in one respiratory sample. Inclusion criteria were (1) man or woman aged 18 or over, (2) hospitalization in ICU for SARS-CoV-2 pneumopathy, (3) first hospitalization in ICU, (4) positive diagnosis of SARS-CoV2 infection carried out by PCR or by another approved method in at least one respiratory sample, (5) sampling in the first 24 h after admission to ICU (D0) feasible and (6) patient or next of kin who has been informed of the terms of the study and has not objected to participating. Exclusion criteria were pregnancy, institutionalized patients, inability to obtain informed consent.

For each patient, demographics, comorbidities, time from onset of COVID-19 symptoms to ICU admission, initial presentation of the disease in ICU including the ratio of the arterial partial pressure of oxygen to the fractional inspired oxygen (PaO₂/FiO₂ ratio) at admission, antiviral therapy targeting SARS-CoV-2 and organ support were documented. Organ dysfunctions according to sequential organ failure assessment (SOFA) score (range 0–24, with higher scores indicating more severe organ failures) and simplified acute physiology score II (SAPS II; range 0–164, with higher scores indicating greater severity of illness) were documented. Patients were classified in the acute respiratory distress syndrome (ARDS) group if they were invasively ventilated and met the Berlin criteria for ARDS within the first 3 days after ICU admission [16]. Follow-up included ICU length of stay, in-hospital mortality, day-28 (D28) mortality, day-90 (D90) mortality, as well as occurrence secondary infection based on recommendation from Comité technique des infections nosocomiales et des infections liées aux soins [17].

Ethylene diamine tetraacetic acid (EDTA-)anticoagulated blood was drawn five times during the first month after ICU admission: within the first 48 h after admission (Day 0: D0), between 72 and 96 h after admission (D3), between D7 and D9 (D7), between D12 and D15 (D12) and between D20 and D25 (D20). Blood was stored at 4–8 °C and processed within 4 h after withdrawal. The numbers of available values for each immune parameter at each time point are presented in Additional file 1: Table S4.

Whole blood was sampled on EDTA tubes and plasma was frozen at − 20 °C within 4 h following blood collection. Cytokine measurement was taken by batches after 1 freeze/thaw cycle using standardized protocols fulfilling clinical and diagnostic laboratories accreditation requirements from the International Organization for Standardization. Plasma concentrations of IL-6, TNF-α, IFN-γ and IL-10 were measured by Simpleplex® technology using ELLA instrument (ProteinSimple®, San Jose, CA), following manufacturer’s instructions. Plasma IFNα2 concentrations were determined by single-molecule Array (SIMOA®) on a HD-1 Analyzer (Quanterix) using a commercial kit for IFN-α2 quantification (Quanterix, Lexington, Mass).

Whole blood was collected on PAXgene blood RNA tubes (BD, Grenoble, France) for IFN signature and frozen at − 80 °C until RNA extraction. IFN score was obtained using nCounter® analysis technology (NanoString Technologies, Seattle, WA) by calculating the median of the normalized count of 6 ISGs using standardized protocols fulfilling clinical and diagnostic laboratories accreditation requirements from the International Organization for Standardization. As previously described, six interferon responsive genes were monitored: SIGLEC1 (sialic acid binding Ig like lectin 1), IFI27 (interferon alpha inducible protein27), IFI44L (interferon induced protein 44 like), IFIT1 (interferon induced protein with tetratricopeptide repeats 1), ISG15 (interferon-stimulated gene 15) and RSAD2 (radical S-adenosyl methionine domain-containing 2). Three references genes were also measured: ACTB (Actin beta), HPRT1 (hypoxanthine phosphoribosyltransferase 1) and POLR2A (RNA Polymerase II Subunit A) [18].

T lymphocyte subpopulation immunophenotyping was performed on an automated volumetric flow cytometer from Beckman Coulter (Aquios CL) as previously described [19]. Monocyte HLA-DR expression and B and NK immunophenotyping were performed using antibodies from Beckman-Coulter and BD Biosciences. The expression of monocyte HLA-DR was determined using the Anti-HLA-DR/Anti-Monocyte Quantibrite assay (BD Biosciences, San Jose, USA). A total number of antibodies bound per cell (AB/C) were quantified using calibration with a standard curve determined with BD Quantibrite phycoerythrin (PE) beads (BD Biosciences) as described elsewhere [20]. B and NK lymphocyte immunophenotyping was performed using lyophilized antibody panel from Beckman Coulter (Duraclone kit). Data were acquired on a Navios flow cytometer (Beckman Coulter, Hialeah, FL), and flow data were analyzed using Navios software (Beckman Coulter). Enumeration of lymphocyte subpopulations as well as mHLA-DR measurement were performed using standardized protocols fulfilling clinical and diagnostic laboratories accreditation requirements from the International Organization for Standardization.

Semiquantitative values of SARS-CoV-2 viral load in upper respiratory samples at ICU admission were retrieved from clinical files for 40 patients. These results were obtained from accredited reference laboratory using RT-PCR technique with validated commercial kits (COBAS® SARS-CoV-2, Roche Diagnostics or reference technique from Pasteur Institute) based on recommendations from the Societé Française de Microbiology after evaluation of analytical performances of the techniques [21]. Patients were classified into three groups according to SARS-CoV-2 PCR Ct values reflecting respiratory viral load at admission (1) A significant viral excretion (ct value ≤ 33) subsequently divided into two groups: high viral load (Reference Ct value < 27, n = 12) and medium viral load (Reference Ct value = [27–33], n = 21); (2) a nonsignificant viral excretion corresponding to a low viral load (Reference Ct value = [33–37], n = 7).

Data are presented as numbers and percentage (qualitative variables) and medians and 25th/75th percentiles (quantitative variables). Chi square or Fisher’s exact test were used for qualitative variables assessment. Quantitative variables were compared with Mann–Whitney U test. For all pairs of immune parameters, Spearman’s rho correlation coefficients were estimated and summarized in a correlation matrix. Kaplan–Meier survival curves were calculated in groups of patients with or without ARDS. The p-value of the log-rank test is given. The level of significance was set at 5%. Data were analyzed using Graphpad Prism version 5.03 (Graphpad Software, La Jolla, USA).

Sixty-four patients with confirmed pulmonary SARS-CoV-2 infection admitted to three ICUs of Lyon University Hospitals (Hospices Civils de Lyon, Lyon, France) were included between March 16 and May 15, 2020. This period corresponded to the first surge in COVID-19 cases in France. Clinical characteristics are presented in Table 1. Median duration of symptoms before ICU admission was 7 [4–11] days. Forty patients (63%) presented with ARDS requiring mechanical ventilation within the first 3 days after ICU admission; 23/24 (96%) patients without ARDS needed noninvasive ventilation (including high flow nasal oxygenotherapy) for respiratory dysfunction/failure. Flowchart and the number of samples by time point are provided in Additional file 1: Figure S1A. As assessed after 28 days, mortality in this cohort was 22% with contrasting results between patients without ARDS (no mortality) and with ARDS (35%) (Additional file 1: Figure S1B).

On admission, critically ill COVID-19 patients presented with elevated plasma levels of both pro- and anti-inflammatory cytokines (Additional file 1: Table S1). Marked lymphopenia (median = 653 cells/µL) affecting all lymphocyte subsets (e.g., median CD4 + T lymphocytes = 298 cells/µL, with CD4/CD8 ratio in normal range) [10, 11] and moderately decreased mHLA-DR (median = 11,125 AB/C, control values > 13,500 AB/C) [22] were also hallmarks of initial critically ill COVID-19 patients’ immune profile. Collectively, these results are in agreement with previously published values in ICU cohorts [9, 23, 24]. Interestingly, we observed that IFN-I response was induced in the majority of these patients as we measured elevated levels of plasma IFNα2 (median = 385 fg/mL, control values < 20 fg/mL) and ISG score (median = 40, control values < 2.3). These concentrations were consistent with the previously published results in cohorts of critically ill patients [12]. To note, for these inaugural immune parameters, no differences in their concentrations were observed based on delay between first symptoms and sampling time (data not shown).

When correlating immune parameters at ICU admission (Fig. 1a), significant correlations were noted between increased pro-inflammatory IL-6 concentration and increased TNF-α and IL-10 levels (positive correlations), and decreased T lymphocyte number (negative correlation). No correlation was observed between IL-6 and IFNα2 levels or ISG score. In addition, initial increased IL-6 concentration was correlated with initial severity as measured by sepsis-related organ failure assessment (SOFA) score and with intensity of pulmonary dysfunction as measured by the ratio between partial pressure of oxygen in arterial blood (PaO₂) and fraction of inspired oxygen (FiO₂) (PaO₂/FiO₂ ratio, Fig. 1b). Finally, when classifying patients into three groups according to semiquantitative levels of viremia measured in upper respiratory samples, the group of patients with the highest viral load presented not only with higher plasma IL-6 concentrations at admission but also with a higher SOFA score and increased mortality (33%, Additional file 1: Figure S2). This suggests that the immune response is positively associated with viral burden and that failure to resolve both aspects may underlie severity [8].

Considering the significant difference between ARDS and non-ARDS groups regarding D28 mortality (Additional file 1: Figure S1B), we then compared immune profiles between these 2 groups of patients (Fig. 2, Additional file 1: Table S2). As most non-ARDS patients presented with a short ICU stay of less than 1 week, we focused on the first three time points after ICU admission. Non-ARDS patients presented with a different immune profile than patients with ARDS, including higher lymphocytes count and mHLA-DR expression and lower cytokine levels and severity scores. Except for IL-6 and TNF-α, these differences persisted over the first week of follow-up (Fig. 2). We observed no difference between ARDS and non-ARDS patients regarding plasma IFNα2 measurements or ISG score (Fig. 2). The correlation between these two parameters was statistically significant (r = 0.7, p < 0.0001). In addition, we observed a rapid decrease of IFN-I markers after ICU admission in both groups. This shows that IFN-I response was induced in critically ill COVID-19 patients and efficient in activating downstream genes. However, the intensity of this response was not impacted by the development of pulmonary dysfunction or disease severity upon ICU admission.

We next focused on ARDS patients who presented with a prolonged ICU length of stay to evaluate immune profiles in regard with D28 mortality during a 3-week follow-up (Fig. 3, Additional file 1: Table S3). Clinical characteristics of survivors and non-survivors are presented in Table 2. Surprisingly, we did not observe any difference between survivors and non-survivors regarding the majority of immune parameters. In this cohort, survivors and non-survivors presented with apparent common immune responses mixing gradual return to normal ranges of cellular markers (mHLA-DR, lymphocyte count) and progressive decrease of cytokines levels including IFNα2 (and related ISG score). Only plasma TNF-α presented with a specific kinetic as it was the only cytokine to slowly rise during the follow-up. Moreover, its levels were significantly higher in non-surviving patients compared with survivors at all time points. This may reflect the high rates of secondary infections occurring in ARDS COVID-19 patients [25‐27]. For example, in this cohort, 81% of survivors and 64% of non-survivors ARDS patients developed a secondary infection during their ICU stay. The constant elevation of neutrophil count during the whole monitoring in ARDS patients probably also reflect such re-stimulation of the immune system in response to secondary infectious challenges (Additional file 1: Table S3). Last, we observed that non-survivors exhibited a significant rise in CD4/CD8 ratio compared with survivors (Fig. 3). This was mainly due to the slower recovery of CD8+T cell numbers in this group (Additional file 1: Table S3).