Therapeutic plasma exchange accelerates immune cell recovery in severe COVID-19

Abstract

Background: Immunological disturbances (anti-type I IFN auto-antibody production, cytokine storm, lymphopenia, T-cell hyperactivation and exhaustion) are responsible for disease exacerbation during severe COVID-19 infections.

Methods: In this study, we set up a prospective, randomised clinical trial (ClinicalTrials.gov ID: NCT04751643) and performed therapeutic plasma exchange (TPE) in severe COVID-19 patients in order to decrease excess cytokines and auto-antibodies and to assess whether adding TPE to the standard treatment (ST, including corticosteroids plus high-flow rate oxygen) could help restore immune parameters and limit the progression of acute respiratory distress syndrome (ARDS).

Results: As expected, performing TPE decreased the amount of anti-type I IFN auto-antibodies and improved the elimination or limited the production of certain inflammatory mediators (IL-18, IL-7, CCL2, CCL3, etc.) circulating in the blood of COVID-19 patients, compared to ST controls. Interestingly, while TPE did not influence changes in ARDS parameters throughout the protocol, it proved more effective than ST in reversing lymphopenia, preventing T-cell hyperactivation and reducing T-cell exhaustion, notably in a fraction of TPE patients who had an early favourable respiratory outcome. TPE also restored appropriate numbers of CD4+ and CD8+ T-cell memory populations and increased the number of circulating virus-specific T cells in these patients.

Conclusion: Our results therefore indicate that the addition of TPE sessions to the standard treatment accelerates immune cell recovery and contributes to the development of appropriate antiviral T-cell responses in some patients with severe COVID-19 disease.

Keywords: COVID-19; adaptive immunity; anti-type I IFN autoantibodies; cytokine storm; immune response; therapeutic plasma exchange.

Figure 1

Oxygenation parameters, neutrophils/lymphocytes ratio and numbers of deaths during the study. (A) Daily fraction of oxygen (FiO2), (B) cumulative FiO2 between day 4 and day 10 and (C) duration of oxygen therapy after 2 months in patients treated with TPE (red symbols) or ST (blue symbols). (D) Evolution of neutrophils/lymphocytes ratio measured using an automated haematology analyser between baseline and day 7. (E) Numbers of deceased, intubated patients and patients with increased or decreased O2 supply and O2 weaning at day 10. (F) Numbers of deaths and survival at day 60. In (A–D), patients were further stratified according to unfavourable (empty symbols) and favourable early outcome (full symbols), as defined in Material and Methods and Supplementary Figure S2 . The number next to each symbol corresponds to the patient's assignment. In (E, F), the days notified in pie charts correspond to the date for each complication. Statistics were calculated with Wilcoxon and an adjusted risk (α’)=0.01.

Figure 2

Type I IFN neutralizing auto-Abs circulating in the plasma of TPE- and ST-treated patients between baseline and day 4. The presence of type I IFN neutralizing auto-Abs present in the plasma of TPE (red symbols)- and ST (blue symbols)-treated patients was quantified using a reporter cell-based neutralization assay and HEK293T cells transfected with a luciferase plasmid containing interferon-stimulated response elements. (A, B) Luciferase activity in HEK293T cells stimulated with low (A, 10 ng/mL) or high (B, 100 ng/mL) concentrations of IFNα2, IFNβ and IFNω and the plasma from TPE (red symbols)- or ST (blue symbols)-treated patients collected at baseline. (C–F) Changes in luciferase activity at day 4 are also shown for patients with significant type I IFN neutralizing auto-Abs (IFNω, C, E, F; IFNα2, D) detected at baseline. A luciferase activity below or above 15% was used to reflect respectively the presence or the absence (grey area) of anti-type I IFN auto-Abs. (G) The evolution of type I IFN neutralizing activity in the two groups of patients between day 4 and baseline was appreciated by dividing the luciferase activity values (IFNα2 100 pg/ml, IFNω 10 ng/ml and 100 pg/ml) on day 4 by those on day 0. In (A–F), patients were further stratified according to unfavourable (empty symbols) and favourable early outcome (full symbols). The number next to each symbol corresponds to the patient's assignment. Statistics were calculated with Wilcoxon and an adjusted risk (α’)=0.017.

Figure 3

Concentrations of plasma cytokines/chemokines in TPE- and ST-treated patients between baseline and day 4. IL-7 (A), IL-18 (B), C-reactive protein (C), Fibrinogen (D), CCL3 (E) and CCL2 (F) concentrations measured by Simpleplex technology and single-molecule array the plasma of TPE (red symbols)- and ST (blue symbols)-treated patients at baseline and day 4. For each mediator and treatment group, a percentage of removal or decrease between baseline and day 4 was calculated as follow = ([concentration mediator X] day 0 - [concentration mediator X] day 4) / [concentration mediator X] day 0. In (A–F), patients were further stratified according to unfavourable (empty symbols) and favourable early outcome (full symbols). The number next to each symbol corresponds to the patient's assignment. The grey areas correspond to the concentrations of mediators usually detected in the plasma of healthy volunteers. Statistics were calculated with Wilcoxon and an adjusted risk (α’)=0.005.

Figure 4

T cell recovery and changes in CD4+ and CD8+ T cell distribution between baseline and day 7. (A) Changes in total lymphocyte counts measured using an automated haematology analyser in TPE (red symbols)- and ST (blue symbols)-treated patients between baseline and day 7. Lymphocyte counts data are depicted for all patients (A1) and further detailed for patients with favourable (A2, empty symbols) and unfavourable (A3, full symbols) outcome. (B–D) Variations in CD4+ (B) and CD8+ (C) T-cell frequencies among CD3+ T cell population, as detected by spectral cytometry, and variations in respective CD4+/CD8+ T cell ratio (D) are also shown. In (A–D), the number next to each symbol corresponds to the patient's assignment. The grey areas correspond to standard values usually detected in healthy donors (A) or mean +/- SD values that we detected in 10 healthy volunteers (B–D). Statistics were calculated with Wilcoxon and an adjusted risk (α’)=0.005.

Figure 5

Variations in differentiation, activation, exhausted/senescent and regulatory phenotype in the CD4+ T cell population from baseline to day 7. (A) Changes in the frequencies of naïve (A1, CD45RA+CCR7+), central memory (A2, Tcm, CD45RA-CCR7+), effector memory (A3, Tem, CD45RA-CCR7-) and terminally effector memory (A4, Temra, CD45RA+CCR7-) subsets among CD4+ T cell population measured by spectral cytometry in TPE (red symbols)- and ST (blue symbols)-treated patients between baseline and day 7. (B–E) Changes in the frequencies of activated (B, HLADR+CD38+), proliferating (C, Ki67+), regulatory (D, FoxP3+CD25+) and exhausted/senescent (E, PD1+CD57+) subsets among CD4+ T cell population are also shown. In (A–E), patients were further stratified according to unfavourable (empty symbols) and favourable early outcome (full symbols). The number next to each symbol corresponds to the patient's assignment. The grey areas correspond to mean +/- SD values detected in 10 healthy volunteers. Statistics were calculated with Wilcoxon and an adjusted risk (α’)=0.012.

Figure 6

High dimensional cell analysis of CD4+ T cell subsets during the study. FlowSOM analysis with automatic consensus clustering was performed on concatenated CD4+ T cell data (1000 cells/sample) from TPE- and ST-treated patient samples collected at baseline and day 7. Data obtained from 10 healthy volunteers were also included as controls. (A) Results were presented as a self-organizing map gathered in 10 background coloured clusters (1-10). Each cluster includes phenotypically similar cells. (B) Heat map of the integrated MFI of 25 markers across the 10 FlowSOM clusters identified in (A) The colour in the heatmap represents the median of the arcsinh for each cluster (centroid) transformed with a coefficient of 5 for marker expression. Clusters (lines) were hierarchically metaclustered using Ward’s method, and differential marker expression was used to assign each cluster and metacluster with a specific identity. (C, D) Cluster frequencies were determined for each sample from each patient and each healthy volunteer and presented as a heatmap, in which the colours represent cluster abundance among the CD4+ T cell population (C). Results are also depicted as individual scatter plots, in which the grey areas correspond to mean +/- SD values detected in the 10 healthy volunteers. TPE (red symbols)- and ST (blue symbols)-treated patients were also stratified according to unfavourable (empty symbols) and favourable early outcome (full symbols). The number next to each symbol corresponds to the patient's assignment. Statistics were calculated with Wilcoxon and an adjusted risk (α’)=0.005.

Figure 7

High dimensional cell analysis of CD8+ T cell subsets during the study. FlowSOM analysis with automatic consensus clustering was performed on concatenated CD8+ T cell data (1000 cells/sample) from TPE- and ST-treated patient samples collected at baseline and day 7. Data obtained from 10 healthy volunteers were also included as controls. (A) Results were presented as a self-organizing map gathered in 10 background coloured clusters (1-10). Each node includes phenotypically similar cells. (B) Heat map of the integrated MFI of 25 markers across the 10 FlowSOM clusters identified in (A) The colour in the heatmap represents the median of the arcsinh for each cluster (centroid) transformed with a coefficient of 5 for marker expression. Clusters (lines) were hierarchically metaclustered using Ward’s method to group subpopulations with similar phenotype, and differential marker expression was used to assign each cluster and metacluster with a specific identity. (C, D) Cluster frequencies were determined for each sample from each patient and each healthy volunteer and presented as heatmap, in which the colours represent cluster among the CD8+ T cell population (C). Results were also depicted as individual scatter plots, in which the grey areas correspond to mean +/- SD values detected in the 10 healthy volunteers. TPE (red symbols)- and ST (blue symbols)-treated patients were also stratified according to unfavourable (empty symbols) and favourable early outcome (full symbols). The number next to each symbol corresponds to the patient's assignment. Statistics were calculated with Wilcoxon and an adjusted risk (α’)=0.005.

Figure 8

Virus-specific CD4+ and CD8+ T cell-responses detected at baseline and day 7. 1x106 PBMC collected from TPE (red symbols)- and ST (blue symbols)-treated patients at baseline and day 7 after the start of treatments were restimulated in vitro for 6 hours with N & M (A1-D1) or S (A2-D2) peptides from SARS-COV2 virus, plus anti-CD28+ and anti-CD49d+ mAbs. The percentages of effector cells expressing TNF-α (A1-A2, C1-C2) or IL-2 (B1-B2, D1-D2) cytokines in both CD4+CD154+ (A, B) or total CD8+ (C, D) T cells were determined at the end of the stimulation period by flow cytometry. In (A–D), patients were further stratified according to unfavourable (empty symbols) and favourable early outcome (full symbols). The number next to each symbol corresponds to the patient's assignment. Statistics were calculated with Wilcoxon and an adjusted risk (α’)=0.006.

Figure 9

T lymphocyte improvements correlated to cytokine removal (A) Comparison of variation in immune parameters on all patients (treated by TPE or ST). Principal component analysis performed on the variation cytokine (values at day 0 – values at day 4) to variation of T cell parameters (values at day 7 – values at day 0). Results depicted groups of cytokines that varied and groups of T cell parameters that varied. The size of the symbols corresponded to the significance of the correlation while the colour corresponded to the sign of correlation (blue for positive correlation and red for negative correlation). Each dimension defined a group of marker. Together the 3 components shown defined approximately 70% of the variance (B) Correlations using Spearmen tests between all dimensions (cytokine and lymphocyte dimensions) were reported.