IL-6 combined with CD8+ T cell count early predict in-hospital mortality for patients with COVID-19.

Luo, Miao; Liu, Jing; Jiang, Weiling; Yue, Shuang; Liu, Huiguo; Wei, Shuang

BACKGROUND The numbers of fatal cases of Coronavirus Disease 2019 (COVID-19) continue to increase rapidly around the world. We aim to retrospectively investigate potential roles of factors, mainly immunologic parameters, in early predicting outcomes of patients with COVID-19. METHODS A total of 1,018 patients confirmed COVID-19 were enrolled in our retrospective study from two centers. The data of clinical features, laboratory tests, immunological tests, radiological findings, and outcomes were collected. Univariate and multivariable logistic regression analysis were performed to evaluate factors associated with in-hospital mortality. Receiver operator characteristic (ROC) curves and survival curves were plotted to evaluate the clinical usefulness. RESULTS Compared to the survival patients, the counts of all T lymphocytes subsets were markedly lower in non-survivors(P < 0.001), especially in CD8+ T cells (96.89 vs 203.98 cells/μl, P < 0.001) . Among all tested cytokines, IL-6 elevated most significantly with an upward trend of more than ten times (56.16 vs 5.36 pg/mL, P < 0.001). By a multivariable logistic regression analysis, two immunological indicators were found to be associated with in-hospital mortality, including IL-6 > 20 pg/mL (OR = 9.781; 95%CI, 6.304-15.174; P < 0.001) and CD8+ T cell count < 165 cells/μl (OR = 5.930; 95%CI, 3.677-9.562; P < 0.001), after adjusting confounding factors (age, gender, and underlying diseases). All the patients were divided into four groups according to levels of IL-6 and CD8+ T cells. The group with IL-6 > 20 pg/mL and CD8+ T cell count < 165 cells/μl had more old and male patients, as well as more proportion of patients with comorbidities, ventilation, ICU admission, shock, and death than those of any other group (P < 0.001). Furthermore, the ROC curve of the model combining IL-6 (>20 pg/mL) and CD8+ T cell count(<165 cells/μl) displayed more favorable discrimination than that of CURB-65 score (area under curve (AUC) = 0.907 vs 0.843, P < 0.001). Hosmer-Lemeshow test showed a good fitting of the model with no statistical significance (P = 0.581). CONCLUSIONS We firstly identify two reliable prognostic indicators, IL-6 (>20 pg/mL) and CD8+ T cell count (<165 cells/μl), which can accurately stratify patients into risk categories and predict mortality of patients with COVID-19. Those two indicators combined may guide clinicians to evaluate patient prognosis and make appropriate decisions.