Prediction Model Based on the Combination of Cytokines and Lymphocyte Subsets for Prognosis of SARS-CoV-2 Infection

Abstract

Background: There are currently rare satisfactory markers for predicting the death of patients with coronavirus disease 2019 (COVID-19). The aim of this study is to establish a model based on the combination of serum cytokines and lymphocyte subsets for predicting the prognosis of the disease.

Methods: A total of 739 participants with COVID-19 were enrolled at Tongji Hospital from February to April 2020 and classified into fatal (n = 51) and survived (n = 688) groups according to the patient's outcome. Cytokine profile and lymphocyte subset analysis was performed simultaneously.

Results: The fatal patients exhibited a significant lower number of lymphocytes including B cells, CD4⁺ T cells, CD8⁺ T cells, and NK cells and remarkably higher concentrations of cytokines including interleukin-2 receptor, interleukin-6, interleukin-8, and tumor necrosis factor-α on admission compared with the survived subjects. A model based on the combination of interleukin-8 and the numbers of CD4⁺ T cells and NK cells showed a good performance in predicting the death of patients with COVID-19. When the threshold of 0.075 was used, the sensitivity and specificity of the prediction model were 90.20% and 90.26%, respectively. Meanwhile, interleukin-8 was found to have a potential value in predicting the length of hospital stay until death.

Conclusions: Significant increase of cytokines and decrease of lymphocyte subsets are found positively correlated with in-hospital death. A model based on the combination of three markers provides an attractive approach to predict the prognosis of COVID-19.

Keywords: Coronavirus disease 2019; cytokines; lymphocyte subsets; prognosis; severe acute respiratory syndrome coronavirus 2.

Fig. 1

Using cytokines and lymphocyte subsets on admission for discriminating the fatal cases from the survived patients. a Scatter plots showing the levels of IL-2R, IL-6, IL-8, and TNF-α in the fatal cases (n = 51) and the survived patients (n = 688). Horizontal lines indicate the median. ***P < 0.001 (Mann-Whitney U test). b Scatter plots showing the numbers of B cells, CD4⁺ T cells, CD8⁺ T cells, and NK cells in the fatal cases (n = 51) and the survived patients (n = 688). Horizontal lines indicate the median. ***P < 0.001 (Mann-Whitney U test). c ROC analysis showing the performance of IL-2R, IL-6, IL-8, and TNF-α in distinguishing the fatal cases from the survived patients. d ROC analysis showing the performance of B cells, CD4⁺ T cells, CD8⁺ T cells, and NK cells in distinguishing the fatal cases from the survived patients. IL-2, interleukin-2 receptor; IL-6, interleukin-6; IL-8, interleukin-8; TNF-α, tumor necrosis factor-α; ROC, receiver operating characteristic curve; AUC, area under the curve

Fig. 2

Change of cytokines and lymphocyte subsets in the same patients. a Line graphs showing the levels of IL-2R, IL-6, IL-8, and TNF-α for each fatal patient on admission and death (n = 42). One line represents one patient. ***P < 0.001 (Wilcoxon’s test). b Line graphs showing the numbers of B cells, CD4⁺ T cells, CD8⁺ T cells, and NK cells for each fatal patient on admission and death (n = 16). One line represents one patient. *P < 0.05, **P < 0.01, ***P < 0.001 (Wilcoxon’s test). c Line graphs showing the levels of IL-2R, IL-6, IL-8, and TNF-α for each survived patient on admission and discharge (n = 86). One line represents one patient. *P < 0.05, **P < 0.01, ***P < 0.001 (Wilcoxon’s test). d Line graphs showing the numbers of B cells, CD4⁺ T cells, CD8⁺ T cells, and NK cells for each survived patient on admission and discharge (n = 62). One line represents one patient. **P < 0.01, ***P < 0.001 (Wilcoxon’s test). IL-2, interleukin-2 receptor; IL-6, interleukin-6; IL-8, interleukin-8; TNF-α, tumor necrosis factor-α

Fig. 3

Correlation between cytokines and lymphocyte subsets on admission and days from hospital admission to death in the fatal patients. a Correlation between the levels of IL-2R, IL-6, IL-8, and TNF-α and days from admission onset to death (n = 51). b Correlation between the numbers of B cells, CD4⁺ T cells, CD8⁺ T cells, and NK cells and days from admission onset to death (n = 51). Each symbol represents an individual fatal patient. *Days from admission onset to death. IL-2, interleukin-2 receptor; IL-6, interleukin-6; IL-8, interleukin-8; TNF-α, tumor necrosis factor-α

Fig. 4

Establishment of prediction model for prognosis of SARS-CoV-2 infection based on combination of cytokines and lymphocyte subsets. a Scatter plots showing the score of prediction model in the fatal cases (n = 51) and the survived patients (n = 688). Horizontal lines indicate the median. ***P < 0.001 (the Mann-Whitney U test). Blue dotted lines indicate the cutoff value in distinguishing these two groups. b ROC analysis showing the performance of prediction model in distinguishing the fatal cases from the survived patients. ROC, receiver operating characteristic curve; AUC, area under the curve