Increased levels of plasma cytokines and correlations to organ failure and 30-day mortality in critically ill Covid-19 patients

Highlights • Fifteen cytokines were increased at ICU admission in Covid-19 patients.• A relatively moderate increase in cytokine concentrations was observed.• IL-1ra, IL-6 and IP-10 correlated with respiratory failure and acute kidney injury.• IL-8 may be a future biomarker due to its correlation with 30-day mortality.


Introduction
Corona virus disease -2019 (Covid- 19) is an acute viral infection caused by severe acute respiratory syndrome virus 2 (SARS CoV-2), a novel β-coronavirus. Clinical presentation ranges from mild to severe illness associated with acute respiratory failure requiring intensive care and considerable mortality. No specific treatment exists to date thus intensive care measures are mainly supportive. ICU mortality has been reported between 16 and 78% [1]. Virally induced acute respiratory distress syndrome (ARDS) and systemic hyperinflammation appear common features of critically ill Covid-19 patients [2]. Elevated cytokine levels and infection-related biomarkers are associated with increasing severity of illness [3,4]. The pronounced inflammatory response, often described as a cytokine storm syndrome, has become a point of interest for potential anti-inflammatory treatments such as interleukin-6 antagonists and glucocorticoids [5]. Inflammation is a complex and dynamic response that is a result of an intricate network of numerous interacting mediators that alters over time [6]. No immunomodulatory treatment has thus far become mainstay treatment in ARDS and glucocorticoids are only implemented in sepsis during refractory shock [7,8]. Hence, the introduction, target and timing of immunomodulatory therapies remains elusive in Covid- 19. Further understanding of the inflammatory response in Covid-19 is required to delineate targets and timing of possible immunomodulatory treatments.
The present study aims to characterize the inflammatory profile at time of admission in a cohort of patients with PCR confirmed Covid-19 requiring intensive care treatment and to correlate plasma cytokine expression with routine infection-related biomarkers, physiological scoring, organ failure and 30-day mortality.

Materials and method
The study was approved by the National Ethical Review Agency (EPM) (No. 2020-01623). The protocol of the study was registered prior to initiation (Clinical Trials ID: NCT04316884). Informed consent was obtained from the patient or next of kin if the patient was unable to give consent. The Declaration of Helsinki and its subsequent revisions were followed. STROBE guidelines were followed for reporting. This prospective observational study was performed at the intensive care unit (ICU) of Uppsala University Hospital, Sweden, a tertiary hospital.

Data collection and patient groups
A total of 35 patients with plasma samples in the PRONMED Biobank (BbA-827-2018-009) were included in the study. The patients were 24 adults with PCR-confirmed Covid-19 admitted to the ICU. This was the initial Covid-19 cohort treated in the intensive care unit at Uppsala University Hospital between March 14th and April 8th, 2020. Controls were eleven adult patients admitted for elective abdominal surgery for peritoneal carcinosis with preoperative plasma samples in the PRONMED Biobank.
Clinical data were recorded prospectively daily. Simplified Acute Physiology Score 3 (SAPS3), Sequential Organ Failure Assessment (SOFA) score, renal function, circulatory support and respiratory support data were collected as reported in the results [9,10]. Patients with acute kidney injury (AKI) were identified according to Kidney Disease: Improving Global Outcome (KDIGO) criteria [11]. PaO 2 /FiO 2 ratio was estimated in the control cohort from registered oxygen saturation (SpO 2 ) with a mean of 55.55 (±9.57). Blood samples were collected on the day of admission to the ICU. The most recent bloodsample preoperatively was used in the control group. Routine chemistry including hemoglobin, white blood cell count (WBC) and platelets, inflammatory markers including C-reactive protein (CRP), procalcitonin (PCT) and ferritin, and kidney function tests were performed in the hospital central laboratory.

Statistical analysis
Continuous variables are presented as mean (SD) or median (IQR) and categorical variables as number of observations (percent of total number of observations). Mann-Whitney U test was used for group comparisons, Bonferroni adjusted. Correlation between variables was analyzed with Spearman Rank. A two-sided p < 0.05 was considered significant. The mean of routine chemistry values from the day of admission was calculated in the case of multiple analysis performed in the Covid-19 cohort. Cytokines with observed concentrations below limit of detection were treated as missing in the statistical analysis. All statistical analyses were performed using SPSS Statistics, Version 26.0 (IBM SPSS Statistics, IBM Corp). Figures were constructed using R version 3.5.0.

Patient characteristics
The controls had an even distribution of men and women and the most common comorbidities were hypertension and diabetes in addition to their cancer diagnosis (Table 1). In the Covid-19 cohort the majority were male (75%). The most common comorbidities were hypertension, pulmonary disease and diabetes mellitus (54%, 29% and 25% respectively). One fifth (21%) of the Covid-19 cohort were treated with corticosteroids before admission to the ICU. The median ICU admission was eleven days after the initial symptoms associated with Covid-19. Median SAPS3 score and SOFA score at the day of admission were 53(13) and 6 (6) respectively. During the first seven days of intensive care 18 (75%) patients required mechanical ventilation and 16 (67%) patients were treated with norepinephrine, no other vasopressor was administered. 24 (100%) patients developed AKI based on increased creatinine and/or loss of urine production. 17 (71%) developed AKI grade 2 and three (13%) patients progressed to grade 3. Seven (29%) patients died within 30 days after ICU admission.

Discussion
The main finding of the present study is that organ failure in critically ill Covid-19 patients is associated with increases of a broad range of inflammatory mediators at time of admission. IL-1β, IL-1ra, IL-2, IL-4, IL-  [12]. In line with this study, all but one of the mentioned mediators were significantly elevated in our critically ill cohort although some more moderately than others.
Dysregulated and exaggerated inflammatory processes are known to result in collateral organ damage in conditions such as sepsis and ARDS. In the present study, the actual concentrations of plasma cytokines, although raised in the Covid-19 cohort, are in some cases rather moderate in comparison to other inflammatory states with associated organ failure such as sepsis [13]. For example, IL-6 measured in plasma is often raised to the thousands in sepsis and septic shock while IL-6 ranged between 4 and 271 ρg/mL in our study [14]. A recent study noted differences in IL-6 levels at the moderate level between severe and critical Covid-19 and suggested its use as a biomarker [15]. Another study in patients admitted to the ICU due to septic shock had plasma IL-8 levels reaching several hundred ρg/mL yet we found a range of 6-62 ρg/mL [16].
The present result strengthens the argument that Covid-19 patients that progress to respiratory failure are in a state of inflammation with a general increase in inflammatory mediators. The term cytokine storm syndrome (CSS) is often mentioned as a propagator of severe Covid-19. CSS refers to a state of hypercytokinemia, hyperferritinemia and multiorgan-failure [17]. In our study, Covid-19 patients presented with increased levels of ferritin as well as at least dual organ failure involving lungs and kidneys. The cytokines raised in CSS including IL-1β, IL-2, IL-6, IL-17a, IL-8, TNF α and MCP-1 were significantly yet moderately elevated in our cohort. Other key characteristics involving the reticuloendothelial system such as bone marrow hemophagocytosis were lacking [3]. There remains a question regarding the dynamics of cytokine release in Covid-19 and if cytokine storm is the most apt description [18].
The TNF α and IL-10 ratio is commonly used to estimate proinflammatory vs anti-inflammatory activity in the inflammatory response [19]. In our study, this ratio was strongly correlated with respiratory failure (r = 0.731) and although significantly correlated with AKI and 30-day mortality these relationships were weaker (r = − 0.543 and r = − 0.388 respectively). This signals a warped balance between the gas and break of the inflammatory response related to organ failure and   cytokines participate in cell mediated immunity, promotes inflammation and tissue damage [40]. IL-10, IL-4, IL-13, IL-7 and IL-9 were categorized as Th2-cytokines (C). Th2-cytokine expression plays a role in humoral immunity and may act anti-inflammatory [41]. NLPR-associated mediators (D) IL-1β, IL-6, IL-8 and IL-1ra take part in the innate immune response including promoting immune cell infiltration of infected tissues. Notably, these cytokines clustered the groups perfectly [42,43]. MIP-1a, eotaxin, MCP-1, MIP-1b, IP-10 and RANTES were grouped as chemokines [44]. The color key of the calculated z-score represents a scale of − 4 to 4 SD where the individual value departs from the group mean as shown in the figure.
mortality. However, cytokines are small (<40 kDa) proteins produced in response to molecular patterns associated with pathogens or self-injury. Their effects are dynamic dependent on context and a simple pro-vs anti-inflammatory division does not always hold true [12,20]. In our study IFN γ was raised at the time of admission. Coronaviruses, such as those causing Severe Acute Respiratory Syndrome (SARS) and Middle Eastern Respiratory Syndrome (MERS) in addition to SARS CoV-2, have previously been noted to demonstrate interferon suppression that in turn has been suggested to initially delay the innate immune response and later contribute to its dysregulation and collateral lung damage [21,22]. The patients in this study were admitted at a median of eleven days after initial symptoms and respiratory failure was already present at admission and hence the IFN γ levels may reflect a later stage in the disease.
The Covid-19 cohort presented with respiratory failure at the time of admission, with a mean PaO 2 /FiO 2 of 15.8 (±3.82) kPa during the first day of intensive care. During the first week in the ICU 75% of patients required mechanical ventilation and these selected patients met the criteria of moderate to severe ARDS according to Berlin criteria [23]. The mean PaO 2 /FiO 2 was lowered in the Covid-19 patients during the first week pointing towards a progression in respiratory failure. The inflammatory mediators that correlated with PaO 2 /FiO 2 included the majority of cytokines in our assay which may be attributed to the lung as the focal point of the infection and inflammatory response. The development of and poor prognosis in ARDS of other etiologies has been associated with a span of inflammatory markers including IL1β, IL-2, IL-4, IL-6, IL-8 and TNF α [24,25]. IL-2, IL-4, IP-10 and MCP-1 have specifically been associated with mortality in ARDS [26]. Our findings are in-line with these previous studies. The cytokines that displayed the strongest correlation with respiratory failure were IL-1ra, IL-6 and IP-10 (r > 0.700, p = 0.000, p = 0.000 and p = 0.000 respectively). IP-10 is a chemokine secreted by epithelial cells in the airways as well as several immune cells including lymphocytes and neutrophils in response to TNF α and IFN γ [27]. It promotes and maintains inflammation and has been designated a biomarker in airway infections, interestingly it has been suggested as protective in coronavirus infections [28].
All Covid-19 patients in our study developed AKI during the first week in the ICU. This incidence is much higher than previously reported in Covid-19 [29][30][31][32]. In ARDS in general the reported incidence of AKI is close to 50% [33]. The reason for our high incidence is unclear. It may be attributed to the large number of mechanically ventilated patients or other factors such as a restrictive fluid regime [34]. Although 67% of patients were treated with norepinephrine, they remained relatively hemodynamically stable as reflected in the SOFA score. Nearly all of the cytokines in our study correlated with AKI development. However, those with the strongest correlation were IL-1ra, IL-6, IL-8, IL-17a, IP-10 and MCP-1 (r > 0.700 respectively). IL1β, IL-6, IL-8 and TNF α have been associated with AKI in ARDS and these were all increased at time of admission in our cohort. In addition, IL-6 and IL-8 were among those with the strongest correlation. AKI in Covid-19 has been associated with increased mortality [32,35]. Furthermore, AKI development early in ARDS is independently associated with longer duration of mechanical ventilation and increased mortality [36]. The interaction between AKI and ARDS development is not completely understood to date although partly attributed to inflammation [37]. Inflammatory mediators that have been designated possible roles in this lung-kidney interaction are IL1β, IL-6, IL-8, TNF α and MIP-2 [38].
A small majority of the analyzed inflammatory mediators correlated with 30-day mortality. However, these correlations were generally weaker than those with respiratory failure and AKI. The strongest correlation could be seen for IL-8 (r = 0.503, p = 0.002). IL-8 also correlated strongly with respiratory failure and AKI (Fig. 2). IL-8 is a chemokine released by a variety of cell types including endothelial and epithelial cells as well as macrophages and monocytes. Its receptors are expressed on various leukocytes including neutrophils and natural killer cells (NK cells). Activation induces chemotaxis and in the case of neutrophils also degranulation and generation of reactive oxygen species [39]. The accumulation of activated neutrophils causing collateral damage to lung tissue has been suggested previously as a mechanism in ARDS.
There are several limitations to this study. First of all, the controls although not displaying any deterioration in organ functions preoperatively had a malignant diagnosis and may therefore have an affected inflammatory state at baseline. Furthermore, we do not study the inflammatory mediators over time. Many of these patients require long periods of intensive care before clinical improvement. Hence, it would be of interest to follow the inflammatory response during the entire ICU stay and elucidate any inflammatory profile differences in organ failure resolution compared to organ failure progression and mortality. Lastly, the small number of patients constitutes an additional limitation.

Conclusion
Organ failure in critically ill Covid-19 patients is associated with increases of a broad range of inflammatory mediators at time of admission to the ICU. However, some appear only moderately increased in comparison to other inflammatory conditions such as sepsis. A strong correlation was found between respiratory failure and IL-1ra, IL-4, IL-8 and IP-10 expression. Acute kidney injury development was frequent and correlated well with increased levels of IL-1ra, IL-6, IL-8, IL-17a, IP-10 and MCP-1. Weaker correlations were generally observed between cytokine expression and 30-day mortality. IL-8 demonstrated the strongest signal in terms of mortality. The dynamics of cytokine release during the course of severe Covid-19 needs further investigation, as do the question whether cytokine storm syndrome is the most suitable description of the pathogenesis.