key: cord-1010678-6gkd7f6h authors: Kleymenov, Denis A.; Bykonia, Evgeniia N.; Popova, Liubov I.; Mazunina, Elena P.; Gushchin, Vladimir A.; Kolobukhina, Liudmila V.; Burgasova, Olga A.; Kruzhkova, Irina S.; Kuznetsova, Nadezhda A.; Shidlovskaya, Elena V.; Divisenko, Elizaveta V.; Pochtovyi, Andrei A.; Bacalin, Valeria V.; Smetanina, Svetlana V.; Tkachuk, Artem P.; Logunov, Denis Y.; Gintsburg, Alexander L. title: A Deep Look Into COVID-19 Severity Through Dynamic Changes in Blood Cytokine Levels date: 2021-11-09 journal: Front Immunol DOI: 10.3389/fimmu.2021.771609 sha: 98f1be8ef674ddd4304b6943ae502faf8ce5f363 doc_id: 1010678 cord_uid: 6gkd7f6h An excessive inflammatory response to SARS-CoV-2 is thought to be a major cause of disease severity and mortality in patients with COVID-19. Longitudinal analysis of cytokine release can expand our understanding of the initial stages of disease development and help to identify early markers serving as predictors of disease severity. In this study, we performed a comprehensive analysis of 46 cytokines (including chemokines and growth factors) in the peripheral blood of a large cohort of COVID-19 patients (n=444). The patients were classified into five severity groups. Longitudinal analysis of all patients revealed two groups of cytokines, characterizing the “early” and “late” stages of the disease course and the switch between type 1 and type 2 immunity. We found significantly increased levels of cytokines associated with different severities of COVID-19, and levels of some cytokines were significantly higher during the first three days from symptom onset (DfSO) in patients who eventually required intensive care unit (ICU) therapy. Additionally, we identified nine cytokines, TNF-α, IL-10, MIG, IL-6, IP-10, M-CSF, G-CSF, GM-CSF, and IFN-α2, that can be used as good predictors of ICU requirement at 4-6 DfSO. Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is a novel betacoronavirus that emerged in December 2019 in Wuhan (China) and resulted in the current pandemic of coronavirus disease 2019 (COVID-19) (1). By September 2021, more than 218 million people have been diagnosed with COVID-19, and approximately 4,5 million people have died during the pandemic (2) . In most cases, the disease course is mild (with or without pneumonia); however, dyspnea, hypoxia, and greater than 50% lung involvement can develop in severe cases, possibly leading to acute respiratory distress syndrome (ARDS), multiple organ failure and death (3) . Mortality in COVID-19 patients admitted to the intensive care unit (ICU) has exceeded 35.5% (4) . The host immune response to SARS-CoV-2 appears to play a critical role in the pathogenesis and progression of COVID-19 (5) ; the response is initiated when SARS-CoV-2 enters alveolar epithelial cells through ACE2 (6) (80% of ACE2-expressing cells) or through AXL (7) or CD147 (8) receptors. After internalization, the virus triggers the canonical response of the innate immune system via interaction with pattern-recognition receptors (PRRs) expressed by epithelial cells, macrophages and dendritic cells, with subsequent massive proinflammatory cytokine release and an enhanced cellular response aimed at preventing viral replication (9) . Serum concentrations of proinflammatory cytokines strongly correlate with disease outcome and were increased in patients with severe disease (10) . In severe cases, induced expression of inflammatory cytokines (especially IL-6, TNF-a) can shift from local to systemic inflammation (5) through dysregulation of immune pathways (9) and immune cell distribution (lymphopenia, T-cell exhaustion, increasing counts of macrophages and neutrophils) (11) (12) (13) . It is supposed one of the main causes of such hyperinflammation and the development of serious complications during COVID-19 is the delayed or impaired type I IFN response as the first line of antiviral defense (14) . Among possible explanations, genetic factors (15) , autoantibodies against type I IFNs (16) and viral immunosuppressive mechanisms (5) have been discussed (17) . Nevertheless, there are contradictory data (17) regarding the kinetics of early type I IFN responses. In addition to IFNs, there has been extensive research on prospective inflammation markers in COVID-19 patients through measurement of increased serum levels of cytokines, chemokines and growth factors (18) (19) (20) (21) . Moreover, several immunological cytokine profiles based on disease severity (IL-6, TNF-a, IL-8, IL-10, G-CSF) (19, 21) have been defined, as have several patient demographic characteristics, including age (IL-6, IL-8, TNF-a) (19, 22) , sex (IL-6, IL-18, IL-7) (23, 24) and the presence of noninfectious comorbidities (IL-6, IL-8, TNF-a) (19, 25) . Some of these factors have been proposed for use as predictors of severity and pharmacologically relevant targets in anti-cytokine therapy (IL-6, IL-10, TNF-a, IFN-g) (9, 20) . Clinical trials are underway, but there are no satisfactory data on their effect thus far (14) . To achieve appropriate implementation of new therapeutic agents for COVID-19 treatment, it is necessary to determine possible immunopathological mechanisms of action to predict complications and to determine the proper time frame in which interventions can be safely performed. Thus, longitudinal analysis performed within short time intervals can expand our understanding of the initial stages of disease and identify early markers to act as predictors of disease severity. This study represents a comprehensive analysis of immune markers (46 cytokines and Ig A, M, G antibodies) in peripheral blood in a Moscow (Russia) cohort of 444 COVID-19 patients. The aim of our research was to investigate the dynamics of cytokines and antibodies in a general sample. We found early changes in cytokine levels (during the first three days from symptom onset) between patient groups with different disease severity. Moreover, we identified some immune signatures associated with sex, age and comorbidities in COVID-19 patients. All these findings will be useful for the prognosis of COVID-19 severity and the development of different therapeutic strategies. In this study, serum samples were obtained from adult COVID-19 patients seen at Clinic of Infectious Diseases №1 of Moscow Healthcare Department during the first wave of COVID-19 incidence from May to July 2020. A cohort of 444 COVID-19 patients was classified into 5 severity groups based on clinical characteristics and guidelines in the management of COVID-19 ( Figure 1 ) (26) . The main criteria were chest imaging (computed tomography (CT) score: degree of involvement ≤50% -score 1-2, >50% score 3-4), saturation of oxygen (SpO 2 ), respiratory rate and fever. The group of ICU patients was separated due to the requirement of intensive care unit therapy (n=39): severe COVID-19 patients (n=129) -CT score 3-4, SpO 2 ≤ 93%, respiratory rate ≥22 breaths/min; moderate COVID-19 patients (n=137) -CT score 0-1-2, SpO 2 >93%, respiratory rate ≥22 breaths/min; mild-moderate COVID-19 patients (n=98) -CT score 0-1-2, SpO 2 >93%, respiratory rate <22 breaths/min, body temperature (t) ≥38°C; and mild COVID-19 patients (n=41) -CT score 0-1-2, SpO 2 >93%, respiratory rate <22 breaths/min and t <38°C. Some patients required oxygen therapy, which included nonmechanical and mechanical ventilation with oxygen. The clinical characteristics of all patients are summarized in Table S1 . Twenty-seven ICU patients developed critical illness during hospitalization and died (69%), and one patient with severe disease died without being in the ICU. Ethical approval for all patients was granted by the local Ethics Committee of Clinic of Infectious Diseases №1 of Moscow Healthcare Department, Moscow, Russia: Protocol No. 2/a from 11 May 2020. Informed consent was obtained, and a questionnaire (Table S3 ) was completed for all enrolled patients. Blood and nasopharyngeal swab samples from each COVID-19 patient were drawn three times during hospitalization: on the admission day, after 4 days (median with 95% CI 3-8) and on the discharge day (median -12 days with 95% CI 7-23). Sera were collected and stored at -30°C until use. Serum samples of healthy donors (HD; n=66, Table S1) were obtained from N.F. Gamaleya National Research Center Biobank, which was collected in Russia during the first half of 2019 before the COVID-19 pandemic, frozen and stored at -80°C without other freeze-thaw cycles. A cohort of 62 COVID-19 patients named "SCG" (seroconversion group) was selected from among all 444 patients according to antibody assay results. These patients were IgM+IgA positive and IgG negative at the first blood sampling point (at admission day) and became IgG positive at the second sampling point. SCG included mild (n=7 or 11%), mild-moderate (n=14 or 23%), moderate (n=22 or 35%), severe (n=15 or 24%) and ICU (n=4 or 6%) cases. Nasopharyngeal swabs were tested using commercial reagent kits for determining the presence of SARS-CoV-2 RNA by real-time PCR: "SARS-CoV-2 FRT" manufactured by N.F. Gamaleya National Research Center, Russia and "SARS-CoV-2/SARS-CoV" manufactured by DNA Technology, Russia. Testing of samples was carried out in accordance with the manufacturer's instructions. Levels of IgG antibodies against SARS-CoV-2 antigens (Nprotein, RBD and S1) were estimated by xMAP SARS-CoV-2 Multi-Antigen IgG Assay and xMAP SARS-CoV-2 IgG Control Kit (Luminex Corp.) using the serum samples of 223 COVID-19 patients (mild (n=20), mild-moderate (n=41), moderate (n=71), severe (n=60), ICU (n=31) according to the manufacturer's instructions. Acquisitions were performed using a MAGPIX instrument operated with xPONENT software version 4.2 (Luminex Corp.). Assay's sensitivity and specificity characteristics: for ≤7; 8-14; >14 days from symptom onset positive percent agreement was 71.1% (55-83% 95% Cl); 80.0% (58-92% 95% Cl); 98.1% (90-100% 95% Cl) respectively, and negative percent agreement was 100% (99%-100% 95% Cl). Serum IgM and IgA in samples from all 444 COVID-19 patients were measured using a COVID-19 ELISA IgM+IgA kit (Vircell) following the manufacturer's instructions. Optical density measurements were performed using a Multiscan FC microplate photometer operated with Skanit Software version 4.1 (Thermo Scientific). Assay's sensitivity in patients 5 days after a positive PCR result was 88%, and specificity in samples from healthy pre-pandemic donors was 99%. Data were analyzed using GraphPad Prism software version 8.0.2. All datasets were tested for a normal distribution using the Shapiro-Wilk normality test. As all normality tests were negative, all data sets were compared using either nonparametric two-tailed Mann-Whitney tests, Kruskal-Wallis test with Dunn's multiple comparison test, or the Wilcoxon test, as appropriate. The prognostic validity of the cytokine model (value) was evaluated by analysis of the ROC curve and was measured using the AUC. Differences were considered significant at p<0.05 (*p<0.05; **p<0.01; ***p<0.001; ****p<0.0001). Spearman's rank correlation tests were used to reveal the association between cytokine levels and were carried out with To determine patterns and predictors of COVID-19 severity during the immune response to SARS-CoV-2, we focused our research on the dynamics of serum biomarker levels (antibodies, cytokines, chemokines and growth factors) in COVID-19 patients. We established a cohort (characterized in Table S1 ) Figure 1 ). The disease severity was determined as the most severe degree of disease during observation period in hospital. Briefly, our cohort was characterized by a median age of 60 years, with a slight quantitative preponderance of females compared with males (56% and 44%, respectively), a median hospitalization period of 12.5 days, an in-hospital mortality rate of 6% and a median disease course from symptom onset to discharge of 21 days. By analyzing levels of antibodies against SARS-CoV-2, we found that the humoral immune response in our sample generally developed according to a fairly standard scenario for COVID-19 (27) . Antibodies of all three classes, appearing in some patients already in the first week of the disease, were significantly increased in general by the end of the second to the beginning of the third week of infection, and total IgM+IgA appeared slightly earlier than IgG ( Figure S1 ). To assess COVID-19 severity risk factors (males, 60+ years, comorbiditiesobesity, diabetes) (3), we performed cytokine profiling for our cohort, as distributed by sex, by age (<60 years, 60+) and the presence of noninfectious comorbidities. These results are described in the Supplementary materials (Figures S2, S3 and Table S2 ). Longitudinal cytokine analysis was performed for all patients to determine general kinetic patterns in the COVID-19 immune response. Time points of blood sample collection were stratified into four intervals of 7 days starting from symptom onset. Patients of all severity groups were included in each time interval of dynamics equally ( Figure 2B ). Our results allowed us to identify statistically significant changes in 27 cytokines (Figure 2A and Figure S4 ). Concentrations of fifteen cytokines (including proinflammatory and anti-inflammatory cytokines, chemokines and growth factors) were the highest on 0-7 days from symptom onset (DfSO) interval, and then declined steadily after 7 DfSO (IFN-a2, IL-10, IL-27, GRO-a, MCP-1, G-CSF, M-CSF) or after 14 DfSO (IFN-g, TNF-a, IL-6, IP-10, IL-15, IL-18, MIG, GM-CSF). These markers we considered "early" cytokines. The other group of cytokines included those that showed positive dynamics and increased from 0-7 DfSO to 15-21 or 22+ DfSO (IL-4, IL-5, IL-7, IL-8, MIP-1b, VEGF-A, sCD40L, FLT-3L, TNF-b, MDC, IL-13, PDGF-AA). This group we named as "late" cytokines. Furthermore, we distinguished two phases of the disease depending on the result of the PCR test on the day of the patient's admission to the hospital. As a result, all patients diagnosed with COVID-19 were divided into 2 cohorts: one in which it was still possible to detect virus from the nasopharynx by PCR (PCR "+", n=298); the other included patients in whom the virus was no longer detected but who still exhibited symptoms of the disease (PCR "-", n=146). We compared serum cytokine levels in these two cohorts. A total of 21 cytokines were revealed, which concentrations differed between the two cohorts ( Figure 2D ). The results for the majority of cytokines confirmed the findings for dynamics in the general cohort described above. For instance, serum levels of IFN-a2, IL-6, IL-10, IP-10, and M-CSF, which tended to decrease (Figure 2A) , were also higher in the PCR "+" cohort than in the PCR "-" cohort ( Figure 2D) . Conversely, serum levels of IL-8, MIP-1b, VEGF-A, which tended to increase (Figure 2A) , were also higher in the PCR "-" cohort than in the PCR "+" cohort ( Figure 2D) . To determine which of the cytokines were elevated in the acute phase and which remained elevated on the discharge day (recovery phase), we selected a group of COVID-19 patients according to their seroconversion data, "SCG" patients. These patients were IgM+IgA positive and IgG negative at the first blood sampling point (at admission day), became IgG positive at the second sampling point and on the day of discharge. Comparative analysis of cytokine levels in "SCG" patients revealed that twelve of them were elevated on the admission day compared with the discharge day ( Figure 2C and Figure S5 ). All of them were confirmed by general cohort dynamics as "early" cytokines, which tended to decrease after 7 or 14 DfSO (Figure 2A and Figure S4 ). Conversely, serum levels of thirteen cytokines remained elevated on the discharge day compared to admission ( Figure 2C and Figure S5 ); ten of them displayed the dynamics of "late" cytokines. A summary of cytokine level changes revealed by the three approaches described above (dynamics in the general cohort, based on PCR and IgG seroconversion) is shown in the Table 1 . We performed correlation analysis for "SCG" patients to identify correlation relationships between cytokine levels at admission and discharge (in the acute and recovery phases). Multiple correlations were found among all cytokines ( Figure S6 ). On the last day of hospitalization, we identified both repeats of the data of the first correlogram and completely new correlation pairs. IFN-a2, the main cytokine of innate immunity, showed a strong correlation with the primary acutephase proinflammatory cytokines TNF-a (r=0.6, p<0.0001), IL-1b (r=0.6, p<0.0001) and IL-15 (r=0.7, p<0.0001). These connections may be illustrated by known data for the beginning of the antiviral response (21, 28) . Another example is a "correlation triangle" between IL-6, IL-10 and TNF-a. IL-10, which is known as a suppressor in the initiation phase of inflammation during COVID-19 and correlated strongly with In PCR "+" group n=298, in PCR "-" group n=146 (data for IL-27 not shown due to its serum level is out of range of plots). The groups were compared by a two-tailed Mann-Whitney U-test for nonparametric comparison. proinflammatory IL-6 (r=0.5, p<0.0001) and TNF-a (r=0.6, p<0.0001), probably performing an immune-inhibitory mechanism as a negative feedback inflammation loop (29) . We examined cytokine levels in patients with different degrees of COVID-19 severity. For this, maximum cytokine levels during hospitalization for each patient were compared. In addition to the cytokines already known to be associated with disease severity in COVID-19 (IL-6, IL-10, IL-27, IL-15, G-CSF, M-CSF, IP-10, MIG, TNF-a, IL-1RA) (19, 21) , which were higher in the patients in our cohort with more severe COVID-19, cytokines with significantly lower concentrations in ICU patients than in others were detected: IL-5, MDC, eotaxin, and IL-12 (p40). (Figure 3) . We divided all of these cytokines into three groups: the first group, IL-1RA, IL-6, IL-10 and MIG, showed increased expression together with disease severity ( Figure 3A) . The second groups included IL-15, IL-27, IP-10, TNF-a, M-CSF, G-CSF and IFN-g, and levels were significantly higher in ICU patients than in other severity groups (Figures 3B, C) . The third group of cytokines included eotaxin, MDC, IL-5 and IL-12 (p40), and their serum concentrations were significantly lower in the ICU group than in the other groups ( Figure 3D) . To evaluate the impact of disease severity on correlations between cytokine levels at the beginning of the disease (0-7 DfSO), we selected patients based on one parameter: the time from illness onset to hospitalization of no more than 7 days (n=234, selected from the general COVID-19 cohort). This new cohort included mild (n=30), mild-moderate (n=54), moderate (n=59), severe (n=69) and ICU (n=22) cases, and cytokine levels in blood at the first time point (admission day) were used for analysis in correlation matrices ( Figures 3E-G) . The mild group was characterized by the largest number of strong correlations (r>0.7) compared with the other groups. In general, the ICU group had a smaller number of high-positive (for most cytokines) but larger negative (for MIP-1b and MDC) correlations than the other groups, especially the mild group. IL-15 exhibited strong correlations in the mild group (0.67