Malignancy History Affected the Prognosis of COVID-19 Patients via Release of Interleukin-6

Background: Coronavirus disease 2019 (COVID-19), a newly erupted respiratory infectious disease caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), has swept across the most of countries. The laboratory characteristics of COVID-patients accompanied with cancer and the risk factors for disease progression and survival of this particular population were few reported. Methods: We enrolled 585 conrmed COVID-19 patients admitted to our hospitals with measured interleukin-6 level on admission. Laboratory tests and outcome were extracted from electronic medical records. Data was divided to cancer group and non-cancer group to explorer the risk factors of progression and survival. Findings: A total of 44 patients with different cancer type (cancer group) and 541 patients without cancer (non-cancer group) were included. Cancer group had signicant higher levels of NEUT, NLR, IL-6, and CRP than non-cancer group, but lymphocyte count and ALB were lower. Cancer group showed signicantly higher progression rate (42·1% vs 22·5%) and mortality (27·27% vs 11·91%) than non-cancer group. Elevated IL-6 and CRP were the risk factors associated with progression among moderate patients and death in-hospital (all p<0·05) in non- cancer group. This correlation was not observed in caner group. Interpretation: IL-6, CRP, NEUT, and NLR were elevated in COVID-19 patients with cancer, with lower level of LYMP and ALB. IL-6 and CRP were positively correlated with progression and poor outcome in patients without cancer. As one of combined diseases, despite malignancy history did not directly affect the prognosis of COVID-19, but it could play a role in the poorer outcome through release of IL-6 and CRP. P-values: result from Chi-square test (for gender and group), Independent t-test (for ALB) and Mann-Whitney U-test (for age, NEUT, LYMP, NLR, IL-6 and CRP). The signicant factors of univariate analysis were included into multivariable logistic regression model. and Mann-Whitney U-test (for age, NEUT, LYMP, NLR, IL-6 and CRP). The signicant factors of univariate analysis were included into multivariable logistic regression model.


Introduction
Coronavirus disease 2019 (COVID-19), a newly erupted respiratory infectious disease caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), has swept across the most of countries on planet. The majority of patients infected COVID-19 exhibit mild to moderate symptoms, while approximately 15% could develop to severe stage and about 5% eventually develop acute respiratory distress syndrome (ARDS), even multiple organ dysfunction 1,2 . Accumulating evidence suggests that patients with severe COVID-19 might have a cytokine storm syndrome and immunosuppression, with hyperin ammation, mainly characterized by increased cytokines (interleukin-2, interleukin-7…), among which, interleukin-6 (IL-6) was regarded as a predictors of fatality 3 . Elevated IL-6 was a hallmark in ammatory signature seen in serum of patients with severe COVID-19 acute respiratory distress 4 .
Elevated serum IL-6 correlates with respiratory failure, ARDS, and adverse clinical outcomes 5 . Malignancy, despite less frequently reported than COPD, hypertension, and diabetes, was notwithstanding controversial for the predictive value on COVID-19 severity. As Guan at al. reported, malignancy was one of the risk factors of disease severity 6 . While another Meta-analysis suggested that there was no correlation between malignant tumor and COVID-19 patients' aggravation 7 . It is worthy to mention that IL-6 has been proposed to play key role in pathogenesis and development of cancer. It has been reported that tumor cells could release certain cytokine, of which one important role was to induce the production of proin ammatory cytokines, such as IL-6, from the stromal cells as well as tumor cells 8 . Hence, we aimed to compare the laboratory characteristics of patients with cancer and without cancer to determine explorer the risk factors for disease progression and survival, certainly the role of IL-6 included.

Study design and participants
In this retrospective cohort study, we enrolled all laboratory con rmed COVID-19 patient admitted to our hospitals who measured IL-6 levels on admission from January 1 to March 25, 2020. Patients were diagnosed as COVID-19 according to Diagnosis and Treatment Protocol for Novel Coronavirus Pneumonia recommended by the National Health Commission (NHC) of China (version 7·0) 9 . The study was approved by The Central Hospital of Wuhan Hospital Ethics Committee and written informed consent was waived by the Ethics Commission of the designated hospital for emerging infectious diseases.

Data Collection
Demographic, laboratory ndings and outcome data were extracted from electronic medical records through a standardized data collection form. All data were collected and checked by 3 experienced clinicians independently.
Procedures SARS-CoV-2 infection was con rmed by next-generation sequencing or real-time RT-PCR performed with pharyngeal swab or bronchoalveolar lavage uid 10 . Routine laboratory examinations included blood examinations, coagulation and biochemical tests on admission. The clinical outcomes were evaluated by two experienced clinicians.

De nitions
The disease severity of COVID-19 was assessed according to the guideline of Chinese NHC 9 . Brie y, moderate cases were de ned as patients with fever, dry cough, fatigue and other symptoms, pulmonary CT ndings were exudation or consolidation, but oxygen saturation exceeded 93% without oxygen; severe grade signi ed respiratory frequency ≥ 30 times/minute, blood oxygen saturation ≤ 93%, oxygenation index ≤ 300 mmHg, and/or lung in ltration progression > 50% within 24 to 48 hours; and critical grade was de ned as appearance of respiratory failure, septic shock, and/or multiple organ dysfunction or failure. Poor progression included moderate grade progressed to severe or critical grades and even death. Statistical Analysis SPSS software (version 22·0) was used to analyze the data. Shapiro wilktest method was used to determine the distribution of continuous variables. Student-t test was used to test the score difference of each group in the normal distribution, and rank sum test was used to compare the difference in the normal distribution. The normal distribution measurement data is expressed by mean ± standard deviation (SD), and median (interquartile Range IQR) for the non-normal distribution data. Frequency (percentage) was used to express the counting data. Chi-square test was used to compare the distribution differences among groups. When the number of predicted cases was less than 5, Fisher accurate probability method was applicated. The signi cant factors of univariable analysis were included into multivariable logistic regression model. A two-sided α of less than 0·05 was considered statistically signi cant.
The median age of cancer group was 67 years, older than non-cancer patients. Gender distribution of two groups did not show signi cant difference. As for laboratory ndings, cancer group had signi cant higher levels of neutrophil count (NEUT), neutrophil-to-lymphocyte ratio (NLR), IL-6 and C-reactive protein (CRP) than non-cancer group, but lymphocyte count (LY) and serum albumin (ALB) were lower (all p < 0·05, Table 1). Among 44 patients with cancer, 38(86·4%) were moderate and 6(13·6%) were severe or critical on admission, the ratios were equal in non-cancer patients ( Table 1, p = 0·866). But cancer group showed higher mortality and higher progression rate among moderate patients, as well as longer in-hospital time.
Risk factors associated with poor progression among moderate patients To con rm which factors contributed to the different ratios of progression between cancer and noncancer group, we enrolled all 510 moderate patients, whose data was studied in two groups: stabilization group (388 stabilized patients) and progression group (122 progressed patients). As shown in Table 2, the proportion of males and malignancy in poor progression group was higher than stabilization group.
The progression group also showed older age and higher levels of IL-6, CRP, NEUT, and NLR, as well as lower levels of ALB and LYMP compared to stabilized patients (all p < 0·05). Based on the results of univariable logistic regression, factors with p < 0·05 were included for the multivariable logistic regression. Interestingly, malignancy was not a risk factor associated with progression (p = 0·757, Table 2), but age, levels of IL-6, CRP and ALB could play a role (Table 2).
To further con rm the roles of these risk factors, we then performed analyses in two groups separately.
Unfortunately, in cancer group (n = 38), none of indicators included was signi cantly related to disease progression (all p > 0·05, Table 3). Nevertheless, In non-cancer group, progressed patients showed older age, more males, higher levels of IL-6, CRP, NEUT and NLR, also lower levels of ALB and LYMP than stabilized patients (all p < 0·05, Table 3), then multivariable logistic regression con rmed that IL-6, CRP, ARB and NLR were the risk factors associated with progression among moderate patients without cancer (Table 3). Table 1 showed the mortality between cancer group and non-cancer group were signi cantly different. We divided all 585 patients into survivor group (524 patients) and non-survivor group (61 patients) ( Table 4).

Risk factors associated with death in-hospital
Not surprisingly, poorer survival was associated with higher proportion of males, more severe/critical presentation at admission, older age, high levels of IL-6, CRP, NEUT, NLR, and lower levels of ALB and LYMP, combination with malignancy was also concluded. Age, gender, IL-6 and CRP were proved meaningful in the subsequent multivariable logistic regression (all p < 0·05, Table 4). However, malignancy was still not a risk factor associated with death in-hospital of COVID-19 (p = 0·223, Table 4).
Subgroup analyses were then performed. As shown in Table 5, in cancer group, non-survivor demonstrated older age, lower ALB, and higher levels of IL-6, CRP and NLR. But all these factors s were meaningless in multivariable logistic regression (all p > 0·05, Table 5). While in non-cancer group, age, IL-6 and CRP were proved as risk factors associated with death in-hospital (all p < 0·05, Table 5).

Discussion
In this retrospective study, we analyzed the laboratory data of 585 COVID-19 infected cases in our hospital, including 44 cases accompanied with different cancers and 541 cases without cancer. The severity grading on admission was equal between these two groups.
In terms of laboratory tests, lymphopenia 11,12 , as well as increased NLR were common feature and thought to be a critical factor associated with disease severity and mortality in COVID-19 13,14 . Our results con rmed lymphopenia and increased NLR in patients with poor outcome in-hospital in both tumor group and non-tumor group.
Studies demonstrated that most patients with severe COVID-19 exhibit substantially elevated serum levels of pro-in ammatory cytokines, including IL-6 and IL-1β, IL-17, G-CSF, GM-CSF…, characterized as cytokine storm 1,2,15,16 as well as C-reactive protein 17 . IL-6 is a call-to-arms for some components of the immune system, including macrophages. Macrophages fuel in ammation and can damage normal lung cells 18 . Elevated IL-6 could trigger cytokine release syndrome through cis signaling or trans signaling. After binding to IL-6 receptor, downstream JAKs (Janus kinases)-STAT3 (signal transducer and activator of transcription 3) signaling results in activation of either immune system cells (B and T cells, neutrophils, macrophages, and natural killer cells) which can contribute to cytokine release syndrome 19 ; or endothelial cells, which then results in a systemic "cytokine storm" charactered by secretion of additional IL-6, vascular endothelial growth factor (VEGF) and reduced E-cadherin expression on endothelial cells 20 . The latter two enhance vascular permeability and leakage, which play an important role in pulmonary dysfunction in ARDS 5 . The severity of IL-6 elevation correlated with the need for mechanical ventilation and mortality 21 . Consistently with previous reports, our analysis showed increased IL-6 level in patients with progression and death in-hospital in both tumor group and non-tumor group. Univariable and multivariable logistic regression analysis con rmed that the increase of IL-6 was positively correlated with progression and poor outcome in non-cancer group. This relationship was not observed in cancer group, indicated that except for IL-6 or CRP, other factors Interestingly, IL-6 has been proposed to play key role in pathogenesis and development of cancer. Tumor cells themselves and immune cells around tumor can release IL-6, and hence increase plasma levels of IL-6 22 . Besides, another analysis revealed an increase in circulating levels of IL-17A in cancer patients, of which the primary role is to induce the production of proin ammatory cytokines, such as IL-6, from the stromal cells as well as tumor cells 8 .
In the other side, tumor patients are frequently combined with cachexia in various degrees according to tumor type and stage 23 . It has been shown elevated IL-6 levels correlated with weight loss and performance status as well as tumor burden 22,24 . A tumor-speci c pro le of cachexia-inducing factors analyze showed upregulated IL-6 level in pancreatic adenocarcinoma, esophageal carcinoma, head and neck squamous cell carcinoma, stomach adenocarcinoma, colon adenocarcinoma, acute myeloid leukemia, breast carcinoma, hepatocellular carcinoma 25 . Logically, our data indicated IL-6 level in cancer group was nearly 3 times of that in non-cancer group. Meanwhile, the mortality in cancer group was identically 3 times of that in non-cancer group. Thus we lanced univariable and multivariable logistic regression analysis to determine whether this increased level of IL-6 could contribute to the higher mortality in patients with cancer. Tables 2 and 4 showed that cancer, as comorbidity, was not a signi cant factor that affected the progression and survival of COVID-19 patients, which was consistent with the previous meta-analysis 7 ; nevertheless dramatically, IL-6, upregulated in cancer group, was con rmed as a marked indicator of poor survival in COVID-19 patients. In summary, despite that cancer itself could not directly play a role in the prognostic of COVID-19, it could conduce to a poorer outcome through release of IL-6.
Elevated serum C-reactive protein (CRP), a protein whose synthesis and release are regulated by IL-6, is also a biomarker of severe COVID-19 infection 5,26 . CRP ≥ 5 mg/dl was considered as an important evidence of severe in ammation. Elder patients with higher CRP had higher in-hospital mortality with a RR of 2 compared with lower CRP group 27 . Similarly in our results, CRP level was signi cantly higher in patients with progression and death in-hospital in both tumor group and non-tumor group. Univariable and multivariable logistic regression analysis con rmed that the IL-6 level was positively correlated with progression and poor outcome in non-cancer group. Furthermore, increased CRP concentrations have been reported in different cancer type 28 . As mentioned above, tumor cells have been shown to secrete IL-6, which in turn induced the production of CRP. Consequently, our data revealed the CRP level in cancer group was marked elevated related to non-cancer group, and this increase of CRP was positively correlated with progression and poor outcome in non-cancer group. Univariable and multivariable logistic regression analysis con rmed CRP, together with IL-6, was a signi cant factor that affected the mortality of COVID-19 patients.
It is worthy to mention that neither IL-6 nor CRP, or other indicator was proved to affect the progression and death within the cancer group, although some laboratory test showed signi cant difference between survivors and non-survivors in univariable logistic regression analysis. This signi ed the prognosis of COVID-19 patients with cancers might be affected by some other critical indicators that were not covered in our study. Guan

Conclusion
This retrospective study revealed that IL-6, CRP, NEUT, and NLR were elevated in COVID-19 patients with cancer, with lower level of LYMP and ALB. IL-6 and CRP were positively correlated with progression and poor outcome in patients without cancer. As one of combined diseases, notwithstanding malignancy history did not directly affect the prognosis of COVID-19, but it could play a role in the poorer outcome through release of IL-6 and CRP.

Limitation
There were some limitations in this study. Firstly, due to the rapid pandemic outbreak and the heterogeneity of prescription of each medical worker, IL-6 concentration test was not performed in all the patients hospitalized. This could induce the possible selection bias and potentially lessened the representative value of our study. Secondly, due to the retrospective study design, not all laboratory tests were performed, especially some important immunological indicators as IL-10, IFN-γ and TNF-α, as well as lymphocyte subsets quanti cation (CD3 + CD4+/CD8 + etc.). Therefore, profound immunological interaction could not be further explored. Moreover, our study was single-central and small-sized, which may make it di cult to generalize the result, larger sample research of multiple centers should be more representative.

Declarations
Ethics approval and consent to participate The study was approved by The Central Hospital of Wuhan Hospital Ethics Committee and written informed consent was waived by the Ethics Commission of the designated hospital for emerging infectious diseases.

Consent for publication
Not applicable.

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.  Note: Quantitative vales coincided with normal distribution are expressed by mean ± SD, and median (interquartile range, IQR) for the non-normal distribution data. Frequency (percentage) was used to express the counting data.
P-values: result from Chi-square test (for gender, severity, progression among moderate patients and outcomes) and Mann-Whitney U-test (for age, time in-hospital and laboratory ndings).
P-values: result from Chi-square test (for gender and group), Independent t-test (for ALB) and Mann-Whitney U-test (for age, NEUT, LYMP, NLR, IL-6 and CRP). The signi cant factors of univariate analysis were included into multivariable logistic regression model.
P-values: result from Chi-square test (for gender and severity), Independent t-test (for normal distribution values) and Mann-Whitney U-test (for non-normal distribution values). The signi cant factors of univariate analysis were included into multivariable logistic regression model.