Risk Factors for Mortality of COVID-19 Patient Based on Clinical Course: A Single Center Retrospective Case-Control Study

Background: Epidemiological factors, clinical characteristics, and risk factors for the mortality of COVID-19 patients have been studied, but the role of complementary systems, possible inflammatory and immune response mechanisms, and detailed clinical courses are uncertain and require further study. Methods: In this single center, retrospective case-control study, we included all COVID-19 inpatients transferred or admitted to Wuhan Tongji Hospital from January 3 to March 30 2020 who had definite clinical outcomes (cured or deceased) with complete laboratory and radiological results. Clinical data were extracted from the electronic medical records, and compared between the cured and deceased patients. ROC curves were used to evaluate the prognostic value of the clinical parameters, and multivariable logistic regression analysis was performed to explore the risk factors for mortality. The correlation between the variables was evaluated by Spearman correlation analysis. Results: 208 patients were included in this study, 182 patients were cured and discharged, 26 patients died from COVID-2019. Most patients had comorbidities, with hypertension as the most common chronic disease (80; 38%). The most common symptoms at onset were fever (149; 72%), cough (137; 66%), and dyspnea (113; 54%). Elevated leucocytes, neutrophils, inflammatory biomarkers (CRP, ferritin, IL6, IL8, procalcitonin), PT, D-dimer, myocardial enzymes, BUN, decreased lymphocyte and subsets (T cells, CD4 T cells, CD8 T cells, NK cells, T cells + B cells + NK cells), and immunological factors (C3, C4) indicated poor outcome. PT, C3, and T cells were confirmed as independent prognostic factors for mortality by logistic regression models. IL6 and CPR were positively correlated with neutrophils, but negatively with lymphocytes and lymphocyte subsets except B cells. IL8 and ferritin were negatively related to T cells and CD4 T cells. Positive associations existed between C3 and T cells, CD4 T cells, and CD8 T cells, whereas there was no significant correlation between C4 and lymphocyte subsets. PT was found positively correlated with IL6, IL8, and CRP. Reverse correlations were explored between C3, C4, and PT, CK-MB, total bilirubin. Conclusions: T cells, C3, and PT were identified as independent prognostic factors for mortality. Decreased C3 and C4, dysregulation of lymphocyte subsets and cytokines may lead to death after SARS-CoV-2 infection.

Methods: In this single center, retrospective case-control study, we included all COVID-19 inpatients transferred or admitted to Wuhan Tongji Hospital from January 3 to March 30 2020 who had definite clinical outcomes (cured or deceased) with complete laboratory and radiological results. Clinical data were extracted from the electronic medical records, and compared between the cured and deceased patients. ROC curves were used to evaluate the prognostic value of the clinical parameters, and multivariable logistic regression analysis was performed to explore the risk factors for mortality. The correlation between the variables was evaluated by Spearman correlation analysis.
Results: 208 patients were included in this study, 182 patients were cured and discharged, 26 patients died from COVID-2019. Most patients had comorbidities, with hypertension as the most common chronic disease (80; 38%). The most common symptoms at onset were fever (149; 72%), cough (137; 66%), and dyspnea (113; 54%). Elevated leucocytes, neutrophils, inflammatory biomarkers (CRP, ferritin, IL6, IL8, procalcitonin), PT, D-dimer, myocardial enzymes, BUN, decreased lymphocyte and subsets (T cells, CD4 T cells, CD8 T cells, NK cells, T cells + B cells + NK cells), and immunological factors (C3, C4) indicated poor outcome. PT, C3, and T cells were confirmed as independent prognostic factors for mortality by logistic regression models. IL6 and CPR were positively correlated with neutrophils, but negatively with lymphocytes and lymphocyte subsets except B cells. IL8 and ferritin were negatively related to T cells and CD4 T cells. Positive associations existed between C3 and T cells, CD4 T cells, and CD8 T cells, whereas there was no significant correlation between C4 and lymphocyte subsets. PT was found positively correlated with IL6, IL8, and CRP. Reverse correlations were explored between C3, C4, and PT, CK-MB, total bilirubin.

INTRODUCTION
The worldwide outbreak of coronavirus disease 2019 , with the first case confirmed in Wuhan, China, in December 2019, is caused by a new strain of coronavirus, Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) (1, 2). COVID-19 has proven highly contagious and more lethal due to rapid transmission capabilities and atypical symptoms or even asymptomatic infection (3)(4)(5). As of January 8, 2021, 86.4 million confirmed cases of COVID-19 and 1.9 million deaths in total have been reported globally, which still need strong intervention (6).
The pathogenesis and pathophysiology for COVID-19 are still unclear. Immune function may play a vital role in SARS-CoV-2 infection and there is still no specific antiviral drug for the treatment of SARS-CoV-2 until now (7). Much research has shown changes of cytokines, lymphocyte subsets, and the dysregulation of host immune response for patients with different severities of COVID-19 (8)(9)(10)(11)(12)(13). However, few studies have involved the complement system, and been conducted based on the clinical courses among patients with definite outcomes (14).
In this retrospective study, we compared clinical, laboratory, and radiological characteristics of cured and deceased patients among all confirmed COVID-19 patients who were transferred or admitted to the designated hospital, Wuhan Tongji Hospital, as of March 30, 2020. We aimed to explore the risk factors for mortality of patients with definite clinical outcomes based on clinical course, especially in the complement system, and inflammatory and immune responses. Finally, we hope to find some clues for new treatment strategies. Our results suggest that decreased C3 and C4, dysregulation of lymphocyte subsets, and cytokines may lead to death after SARS-CoV-2 infection.

Study Design
From January 3 to March 30 2020, 2,977 COVID-19 patients were transferred from other designated hospitals, mobile cabin hospitals, isolation sites, or admitted to Tongji Hospital. All adult patients (≥18 years old) were confirmed to have SARS-CoV-2 real-time RT-PCR or serum IgM and IgG antibodies, and diagnosed according to the Seventh Revised Trial Version of the Novel Coronavirus Pneumonia Diagnosis and Treatment Guidance released by the National Health Commission of China (available from http://www.nhc.gov.cn/yzygj/s7653p/202003/ 46c9294a7dfe4cef80dc7f5912eb1989.shtml). One thousand, seven hundred thirty patients were enrolled with definite clinical outcomes (cured or deceased), but only 208 patients were included with complete laboratory and radiological results (Figure 1). All the cured patients had repeated negative tests for SARS-Cov-2 RNA at least 24 h apart, symptomatic relief, significant improvement in pulmonary radiology, and no need of supportive care or other treatments. Exclusion criteria: (1) missing laboratory or clinical data; (2) Patients still hospitalized; (3) Patients who died in 24 h after hospitalization. The study was approved by the Research Ethics Commission of Tongji Hospital and the requirement for informed consent was waived by the Ethics Commission (IRB ID:TJ-IRB20200343).

Cell Subsets, Cytokines, and Receptors Testing
In order to quantify T lymphocyte subsets, 1 × 10 6 PBMCs were stained with indicated antibodies (CD3-APC, CD4-PerCP, and CD8-FITC antibodies all from BD Biosciences) in the dark at room temperature for 20 min. After two washes, the cells were analyzed within 1 h. All samples were detected by BD FACS Canto II Flow Cytometry System and analyzed with the BD FACS Diva Software.
In order to explore the influence of COVID-19 on the secretion of cytokines and receptors, chemiluminescence immunoassay (CLIA) was performed. Cytokines and receptors including interleukin 2 receptor (IL-2R), IL-6, IL-8, IL-10 and TNF-α were assessed using serum samples (Immulite 1000, DiaSorin Liaison, Italy; or Cobas e602, Roche Diagnostics, Germany). The testing kit was purchased from Roche Diagnostics.

Data Collection
Demographic data, medical history, clinical manifestations, comorbidities, laboratory tests, and radiological findings were extracted from the electronic medical records. The laboratory tests included complete blood count, serum biochemical tests [containing liver and renal function, lactate dehydrogenase (LDH), and electrolytes], myocardial enzymes, NT-proBNP, coagulation profile, procalcitonin (PCT), C-reactive protein (CRP), serum ferritin, erythrocyte sedimentation rate (ESR), cytokines and receptors, lymphocyte subsets, and immune function (including C3 and C4). Chest computed tomography (CT) was done for all patients. The results of virus tests, including SARS-CoV-2 RNA real-time RT-PCR and serum IgM and IgG antibodies, were also recorded.

Statistical Analysis
We used SPSS (version 23.0) for all statistical analyses. Normally and non-normally distributed continuous variables were expressed as means with standard deviations (SD) or medians with inter-quartile range (IQR), respectively, and analyzed with a Student's t-test or Mann-Whitney U-test. Categorical variables were presented as number and proportions, and compared with a Chi-square test, or Fisher's test if cells are small in number. Variables with P < 0.05 was considered to indicate a statistically significant difference. ROC curves were used to evaluate the prognostic value of the clinical parameters. To explore the independent prognostic factors for mortality, backward stepwise, binary logistic regression analysis was performed. Kaplan-Meier analysis was conducted to further confirm the predictive value of clinical parameters on survival. Spearman correlation analysis was used to evaluate the correlation between the variables. Values of P < 0.05 were considered statistically significant.
The deceased patients tended to have worse liver function (i.e., higher total bilirubin and lactate dehydrogenase, lower albumin), renal function (increased blood urea nitrogen and creatinine), myocardial enzymes (elevated CK-MB and hs-cTnI), NT-proBNP and severer bacterial secondary lung infections (Procalcitonin) than recovered patients ( Table 2).

Clinical Courses of Two Typical Hospitalized COVID-19 Patients From Illness Onset
The duration of clinical symptoms, viral shedding, abnormal laboratory tests, and treatments of two typical hospitalized COVID-19 patients from illness onset were presented in Figure 3. Fever persisted longer in the deceased patients (11 days) than the recovered (7 days). The course of viral shedding in cured cases was much shorter (9 vs. 16 days). Increased IL6, lymphopenia, elevated neutrophils lasted longer in the fatal casea, who also had decreased C3. The initiation time and duration of immunomodulator, antiviral and antibacterial treatments were similar between the two patients. However, the deceased patients received systematic corticosteroid and longer oxygen therapy (25 vs. 11 days), and underwent about 18 days invasive mechanical ventilation and 19 days ICU admission after ARDS.

Temporal Changes in Laboratory Markers in Hospitalized COVID-19 Patients From Illness Onset
The trend of key laboratory markers was tracked after illness onset (Figure 4). In the non-survivors, lymphocytes, T cells, and NK cells counts were significantly lower than survivors, with the worst observed around day 20 until death. In survivors, the severe lymphopenia was occurred on day 10, T cells and NK cells count were lowest on day 15, and could gradually return to normal during hospitalization (Figures 4A-C). Levels of IL6, CRP, Ddimer, BUN, and hs-cTnI were significantly elevated in deaths compared with recovered patients throughout the clinical course, whereas C3 decreased with disease exacerbation (Figures 4D-I).
In the deceased patients, IL6 increased rapidly on day 25 from disease onset (Figure 4D), while D-dimer, BUN, and hs-cTnI increased rapidly in the early stage of illness (Figures 4G-I).
Overall, the levels of inflammatory factors between the two groups showed apparent difference even in the first 20 days from illness onset.

DISCUSSION
Although the changes in the hematological tests, infectionrelated markers, and lymphocyte subsets of COVID-19 patients were described by lots of literature (9)(10)(11)(12)(13), few studies have investigated the trend through the clinical process with definite outcomes (14). In this retrospective case-control study, we reported inflammatory and immunological factors and lymphocyte subsets in the cohort of 208 COVID-19 patients with definite outcomes. Collectively, increased inflammatory cytokines, T lymphopenia, especially CD4 T cells (CD3+CD4+, abbreviated as CD4+) and CD8 T cells (CD3+CD8+, abbreviated as CD8+), decreased C3 and C4, were more pronounced in the deceased patients. Decreased C3 and C4, dysregulation of lymphocyte subsets and cytokines, may lead to death after SARS-CoV-2 infection. We noticed that complement C3 and C4 were reduced in the non-survivors, indicating that the dysregulated complement system may be activated in COVID-19. To our knowledge, the complement system is critical in innate immune response to viral infections (15)(16)(17). Several studies have investigated the role of the complement system in SARS-CoV infection and demonstrated that the activation of the complement system can induce illness deterioration. Gralinski revealed that SARS-CoV infected C3-deficient mice had lower levels of neutrophils, monocytes, immune cells, and cytokines than SARS-CoV infected controls, and concluded that in the absence of complement, SARS-CoV was unable to lead to inflammatory  response as in wild-type mice (17). Risitano indicated that decreased C3 could contribute to thrombus, inflammation, and organ damage in COVID-19 (18). Recent clinical findings showed the highly selective C3-target inhibitor, AMY-101, with good safety and tolerability in a Phase I study, and the Phase II clinical trials are currently underway (19,20). Therefore, C3 inhibition may be a prospective therapeutic approach to inhibit complement activation, and complement-mediated inflammatory and immune response in COVID-19 patients. The significant elevation of IL6, a pro-inflammatory cytokine, can lead to a "cytokine storm" and result in acute lung injury, acute respiratory distress syndrome (ARDS), MODS or even multiple organ failure (MOF) (10,12,21). Coronavirus infection of macrophages, dendritic cells, and monocytes induce their own activation and promote the secretion of IL6 and other cytokines. IL6 can function through both classic cis signaling and trans signaling pathways (10,22). In addition, IL6 could bind to membrane-bound IL6 receptor (mIL-6R)-gp130 complex in cis signaling, and soluble form of IL6R (sIL-6R) with a gp130 dimer in trans signaling, and then activate downstream signal transduction through JAKs (Janus kinases) and STAT3 (signal transducer and activator of transcription three). Activation of cis signaling can lead to multiple effects on the innate and acquired immune system and result in cytokine release syndrome (CRS) as mIL-6R is mainly expressed on the surface of immune cells (10,22). Trans signaling is always triggered in cells that do not express mIL-6R, like endothelial cells, leading to secretion of additional IL6, IL8, vascular endothelial growth factor (VEGF), and monocyte chemoattractant protein-1 (MCP-1), reduction of E-cadherin, and finally giving rise to a "cytokine storm." Moreover, IL6 can induce the expression of C-reactive protein (CRP), which is an important inflammatory factor. All of the above further confirm the vital role of IL6 in inflammation and immune response. IL-6 may play a key role in driving the COVID-19 hyper-inflammation, so blockade of IL-6 signaling is promising to improve survival efficiency.
Reduced CD4 + and CD8 + T cells were found in SARS infected patients compared with healthy individuals (23,24). Consistent with previous research (25), we also found that CD4 + and CD8 + T cells were significantly decreased in non-surviving COVID-19 patients in comparison to survivors. The critical role of CD8+ T cells has been proven in respiratory viral clearance (26,27), and CD8+ T cell response was related to the severity of Middle East respiratory syndrome coronavirus (MERS-CoV) infection (28). Effective T cell responses can help effectively eliminate MERS-CoV and control high levels of inflammatory cytokines that can cause extensive inflammatory response and even pulmonary injury. In the early stage of infection, CD4+ regulatory T cells are essential for activating CD8+ T cells to respond to acute respiratory virus infection (29). Elevated inflammatory cytokines, and loss of CD4+ and CD8+ T cells in the deceased COVID-19 patients, may cause strong inflammation, cytokine storms, and eventually more severe tissue damage. Therefore, dysregulated immune responses with consumption of lymphocytes may play a pivotal role for COVID-19 pathogenesis.
Previous studies showed that higher D-dimer and prolonged PT always existed in more severe COVID-19 patients (30,31). In our study, we also found that D-dimer and PT were higher in deceased patients as compared to the cured. Increased D-dimer, prolonged PT, and thrombocytopenia were recently reported to be associated with 28-day mortality in patients with severe COVID-19 infection (32), and the combination of these could be explained by the coagulation system activation and always indicated the possibility of disseminated intravascular coagulation (DIC) (33). Viral infections can lead to inflammatory response and imbalance between procoagulant and anticoagulant homeostasis (34). The inflammatory response induced cytokine storm elicits vascular endothelial dysfunction, disrupts the balance between the coagulation and fibrinolytic systems, and as a result leads to DIC and microcirculation disorder and finally multiple organ dysfunction syndrome (MODS) (35,36). Besides endothelial damage, additional pathogenetic mechanisms, including increased von Willebrand factor, and activated Toll-like receptor and tissue-factor pathways, are also involved (37). Hence, early detection and routine monitoring of hemostasis tests, and correction of coagulation dysfunction in time, could help to effectively establish treatment strategy and reduce mortality (38,39). D-dimer is the degradation product of fibrin, with the elevated level indicating hypercoagulating state and secondary fibrinolysis, while PT is related to exogenous coagulating system. Both can be used for early diagnosis of coagulopathy. Our study reinforced that coagulation dysfunction was more likely to occur in decreased COVID-19 patients, and PT could be used as an independent indicator to predict the mortality for COVID-19.
There were several limitations in this study. First, this was a retrospective, single center study of patients admitted to Wuhan Tongji Hospital. Multicenter research may be needed to evaluate the temporal changes of laboratory markers with COVID-19. Second, some patients might lack laboratory tests in the defined time intervals due to the retrospective study design, which would affect the estimation of their role in predicting mortality. Third, the COVID-19 patients with bacterial co-infection or sepsis might affect the outcome of immune response. Last but not least, our findings may be limited by sample size. The results should be validated in a prospective study with a larger sample.

CONCLUSIONS
In summary, our study demonstrated that T cells, C3, and PT were independent prognostic factors for the mortality of patients with COVID-19. The dysregulated inflammation and immune response in COVID-19 patients may be mainly due to lymphopenia, especially CD4 + and CD8 + T lymphopenia, activation of complement system, extensive inflammation, production of cytokine storm, and eventually severe multiorgan injury. Taken together, surveillance of lymphocyte subsets, inflammatory, and immunological factors would effectively help clinicians to predict the prognosis, and make proper strategies for patients with COVID-19.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by Research Ethics Commission of Tongji Hospital. Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.

AUTHOR CONTRIBUTIONS
CY made substantial contributions to the study design and took responsibility for obtaining ethical approval. JZ was in charge of the manuscript draft. CY and XW took responsibility for data acquisition. JZ and ZW made main contributions to data analysis and interpretation. CY, JZ, and XW participated in the diagnosis and treatment of health professionals. ZW, ZH, and PL made substantial revisions to the manuscript. All authors contributed to the article and approved the submitted version.

FUNDING
This work was supported by the National Natural Science Foundation of China (No. 81702989).