The Prognostic Utility of Cytokines in Hospitalized COVID-19 Patients

We performed a single-center prospective cohort study including 181 adult patients admitted to the 1st Infectious Diseases County Hospital of Târgu Mureș, Romania between December 2020 and September 2021. All patients were diagnosed with COVID-19, confirmed by positive real-time polymerase chain reaction (RTPCR). Patients under the age of 18 years, pregnant women, patients with HIV infection or AIDS, different types of cancer, or connective tissue disorders, transplant recipients and patients with incomplete or missing data were excluded from the study.

For each patient, clinical and demographical parameters (sex and age) were obtained from the medical records and introduced into a database. We collected data on comorbidities, serum cytokine levels, clinical severity of COVID-19, need for oxygen therapy, length of stay, ICU transfer and outcome. Cytokine (IL-6, IL-8, IL-10, IL-12, TNF-α and IFN-γ) serum levels were measured in the first 1–3 days of admission. The blood samples were collected in the first 24 h after hospitalization and, where appropriate, in the first 24 h after reclassification into another severity group, at any time during hospitalization. In the statistical analysis we used the blood samples collected in the first 24 hours after admission.

Patients were split into two age groups: ≤60 years old and older 60.

We compared cytokine serum levels between cases with COVID-19 of different severities to evaluate their predictive value of in relation to the disease severity, need for oxygen therapy, ICU transfer and outcome. We also compared serum cytokine levels according to sex, age, comorbidities and length of stay.

The severity of COVID-19 was defined based on the World Health Organization’s guidance on the clinical management of COVID-19 published on November 23, 2021. Accordingly, ‘mild stage’ was defined as a disease with few symptoms (low fever, cough, fatigue, anorexia, shortness of breath, and myalgias), without evidence of viral pneumonia or hypoxia. ‘Moderate stage’ was defined as a disease with fever and respiratory symptoms, associated with pulmonary imaging findings but no signs of severe pneumonia and an oxygen saturation (SpO₂) of ≥94% on room air. ‘Severe stage’ was defined as the presence of severe pneumonia, plus one of the following: respiratory rate of >30 breaths/min, severe respiratory distress, or SpO₂ <94% on room air. Based on this classification, study cases were divided into two categories: 1) non-severe (mild and moderate stage); 2) severe (severe stage).

The blood samples were collected by venous puncture into vacuum collectors with clot accelerator, were centrifuged and the serum was cryopreserved until analysis at −20 °C. The samples were transported to the Laboratory of Humoral Immunology of the Center for Advanced Medical and Pharmaceutical Research, where serum cytokine levels were measured using xMAP technology with a customized Human Magnetic beads panel for cytokine detection (EMD Millipore Corporation) on a Flexmap 3D analyzer. The laboratory procedures were performed following the manufacturers’ recommendations. For all cytokines, the measuring intervals ranged between 0.64 and 10,000 pg/ml, and the intra-assay coefficient of variation was below 3.5%.

We performed a descriptive analysis, expressing categorical variables with numbers and percentages; numerical variables with mean, median, range and inter-quartile range (IQR). The distribution of the data was assessed using the Kolmogorov-Smirnov test and analyzed using the non-parametric Mann-Whitney U test. Each parameter was estimated using a 95% confidence interval (95% CI). Multivariate logistic regression analysis was performed for disease severity, ICU transfer and outcome. The predictive ability of the studied parameters regarding ICU transfer and outcome was estimated using the receiver operating characteristic curve (ROC) and the area under the curve (AUC), where an AUC 0.5 indicated no predictive ability, a value of 0.8 was considered good, and a value of 1.0 was considered perfect. All statistical analyses were performed using IBM SPSS Statistics for Windows version 26 (Armonk, NY: IBM Corp). A p value <0.05 was considered statistically significant.

The study was approved by the ethics committees of the “George Emil Palade” University of Medicine, Pharmacy, Science and Technology of Târgu Mureș (1237/08.01.2021) and of the Mureș County Clinical Hospital (19038/21.12.2020). Written informed consent was obtained from each patient before inclusion, and full anonymity was preserved for all participants.

A total of 181 hospitalized patients were enrolled in the study. At admission, 87 (48.06%) patients had severe COVID-19, and 48 (26.51%) were reclassified from moderate to severe COVID-19 during hospital stay. The clinical and demographic characteristics of the patients are presented in Table 1.

The mean age of the patients was 64 years (range 21–90). There were 57 severe and 57 non-severe cases among patients above 60 years; 30 severe and 37 non-severe cases among patients below 60 years (p = 0.3). We found no correlation between age and disease severity (r = 0.042; p = 0.229). In total, 114 out of 181 (62.98%) patients were above 60 years, and we found significantly higher serum levels of IL-6 (p = 0.015), IL-8 (p = 0.06), IL-10 (p = 0.006), and TNF-α (p <0.0001) in these patients.

We found a significant difference in the distribution of sexes between patients above and below 60 years, with a male predominance among younger patients (p = 0.003). We found no correlation between sex and serum cytokine levels.

Population distribution by ethnicity: 177 (97,7%) white patients and 4 (2,2%) Indo-Aryan (gipsy). We found no correlation between ethnicity and disease severity, outcome or ICU transfer requirement.

The patients’ comorbidities included arterial hypertension, diabetes mellitus, chronic obstructive pulmonary disease (COPD), asthma, pulmonary fibrosis, hyperlipidemia, obesity and chronic renal and liver disease. IL-6 (p = 0.035), IL-8 (p <0.0001), IL-10 (p = 0.009), and TNF-α (p <0.0001) serum levels were significantly higher in patients with arterial hypertension. We found no statistically significant differences when comparing serum cytokine levels according to the presence of obesity, COPD, asthma and chronic renal disease.

The average length of hospital stay was 13 days, with an equal distribution between sexes. Hospital stay was significantly longer in patients over 60 years (p = 0.016) (mean 13.5 days, range 5–36) than in patients below 60 years (mean 11.9 days, range 1–32 days). The length of stay was not significantly associated with outcome (p = 0.244). A longer hospital stay was also found in patients who required ICU transfer (p <0.0001), and the length of stay was significantly influenced by the presence of respiratory failure (p <0.0001). Furthermore, a positive correlation was found between the length of stay and serum levels of IL-10 (r = 0.273, p = 0.0002), IL-6 (r = 0.217, p = 0.003) and IL-8 (r = 0.187, p = 0.011).

With the exception of IL-12 (p = 0.341) and IFN-γ (p = 0.926), the serum levels of all other cytokines showed significant differences between cases with severe (n = 87) and non-severe (n = 94) COVID-19. We found significantly higher levels of IL-6 (p = 0.002), IL-8 (p = 0.002), IL-10 (p = 0.006) and TNF-α (p = 0.039) in severe COVID-19.

In total, 134 out of 181 (74.03%) patients needed oxygen therapy, 90 (67.16%) of which were over 60 years. We found a positive correlation between the need for oxygen therapy and age (r =0.163; p = 0.038). Patients who needed oxygen therapy presented significantly higher serum levels of IL-6 (p = 0.016), IL-8 (p = 0.001), IL-10 (p <0.0001) and TNF-α (p = 0.002). Furthermore, we found a negative correlation between serum levels of IL-6 (r = −0.278; p = 0.0002), IL-10 (r = −0.262; p = 0.0004), TNF-α (r = −0.175; p = 0.018) and SpO₂ measured on admission.

In total, 35 (19.33%) patients required ICU transfer, 25 (71.42%) from the >60 age group and 10 (28.58%) from the ≤60 age group. The average length of ICU stay was 9 days (range 1–24 days). IL-6, IL-8 and IL-10 serum levels were significantly higher in patients who required ICU transfer (Table 2).

Of the 114 patients over 60 years, 94 (82.46%) were discharged and 20 (17.54%) have died, while of the 67 patients below 60 years, 63 (94.03%) were discharged and 4 (5.97%) have died (p = 0.02). A total of 24 (13.25%) patients died during hospital stay, of which 13 (54.16%) were female and 11 (45.84%) were male. With the exception of IFN-γ and IL-12, all the other studied cytokines showed significantly higher serum levels in non-survivors. We observed a similar trend for IL-12, but the results did not reach the significance level (Table 3).

We performed a multivariate regression analysis, in which the dependent variables were represented by severity (non-severe vs. severe), survival (deceased vs. alive), and ICU transfer (transferred to the ICU vs. not transferred to the ICU) and the independent variables were represented by the different interleukins. The aim of the multivariate regression analysis was to assess the prognostic utility of interleukins based on the dependent variables. An estimated risk value of >1, expressed through the odds ratio (OR), signifies a risk or unfavorable prognosis.

The results of the multivariate regression analysis showed that depending on the ‘severity’ variable, IL-8 had a prognostic value for disease severity, increased levels being associated with severe forms (Table 4).

Depending on the ‘survival’ variable, IL-8, IL-6 and IL-10 had prognostic value, increased levels being associated with death. Of note, these three interleukins had higher values in deceased patients even at the bivariate analysis (Table 5).

Depending on the ‘ICU transfer’ variable, IL-8, IL-6 and IL-10 had prognostic value, increased levels being observed in those transferred to the ICU. Similarly, to the ‘survival’ variable, these three interleukins had higher levels in those transferred to the ICU even at the bivariate analysis (Table 6).

ROC curve analysis for IL-8 regarding disease severity yielded an AUC of 0.632 (95% CI 0.550–0.714; p = 0.002), with a cut-off level of 5.36 pg/ml. The analysis showed a good predictive ability for IL-6 (AUC 0.801; 95% CI 0.713–0.889; p <0.001) and IL-10 (AUC 0.763; 95% CI 0.684–0.842; p <0.001) regarding ICU transfer (Figure 1).

Fig. 1.

ROC curve analysis also showed a good predictive ability for IL-6 (AUC 0.853; 95% CI 0.759–0.946; p <0.001) and IL-10 (AUC 0.831; 95% CI 0.756–0.906; p <0.001) regarding outcome (Figure 2).

Fig. 2.

Cut-off levels and AUC values for IL-6 and IL-10 regarding patient outcome are shown in Table 7.