IP-10 and MCP-1 as biomarkers predicting disease severity of COVID-19

Background: COVID-19 is a viral respiratory disease caused by the severe acute respiratory syndrome-Coronavirus type 2 (SARS-CoV-2). Patients with this disease may be more prone to venous or arterial thrombosis because of the activation of many factors involved in it, including inammation, platelet activation and endothelial dysfunction. Therefore, this study focused on coagulation and thrombosis-related indicators (IP-10, MCP1 and MIP1a) in COVID-19, with the hope to nd biomarkers that can predict patients’ outcome. Methods: This is a retrospective single-center study involving 74 severe and critically ill COVID-19 patients recruited from the ICU department of the Tongji Hospital in Wuhan, China. The patients were divided into two groups: severe patients and critically ill patients. The serum IP-10, MCP-1 and MIP1a level in both groups was detected using the enzyme-linked immunosorbent assay (ELISA) kit. The clinical symptoms, laboratory test results and the outcome of COVID-19 patients were retrospectively analyzed.


Introduction
In December 2019, a new coronavirus pneumonia (COVID-19) originated in Wuhan, China, spreading across the country(1). On February 11, 2020, the International Virus Classi cation Committee announced the o cial name of this disease caused by a new coronavirus, such as "severe acute respiratory syndrome-Coronavirus type 2" (SARS-CoV-2) (2). The main source of infection is represented by pneumonia patients with new coronavirus infection. As of May 31, the new SARS-CoV-2 has spread to over 200 countries and regions around the world, with more than 6 million con rmed cases reported abroad and more than 300,000 deaths worldwide, with a mortality rate of approximately 5.44% (3).
Some studies showed that 40% of COVID-19 patients are at risk of venous thromboembolism (4), and among 30 COVID-19 deaths, 46% were affected by a disseminated intravascular coagulation, indicating that the coagulation dysfunction is one of the main cause of death in severe patients with COVID-19 (5).
SARS-CoV-2 is mainly affecting the alveolar type II epithelial cells, lung macrophages, hilar lymph nodes, spleen and testicular tissue (6). SARS-CoV-2 invades human cells by binding the angiotensin converting enzyme 2 protein distributed on the surface of cells (7) in organs such as heart, lung, kidney, testis and digestive tract (8). As a consequence of SARS-CoV-2 infection, a massive amount of in ammatory factors is released, leading to a systemic in ammatory response syndrome (9). Therefore, the microvascular system is damaged, resulting in an abnormal activation of the coagulation system, causing systemic small vasculitis and extensive microthrombosis (6). Therefore, this study aimed to discover coagulation-related factors that can predict the prognosis of patients.
Several studies showed that IL-1β, IL-6, FGF-2, MCP-1, CCL3 (MIP1a), and CXCL10 (IP-10) are cytokines related to thrombosis [9][10][11] . Mercler et al. reported that the culture medium of pulmonary endothelial cell from patients with chronic thromboembolic pulmonary hypertension contains a higher level of FGF-2, IL-1β, IL-6 and MCP-1 (10). Mir et al. reported that MIP1a may be used as a potential biomarker to predict the risk of deep vein thrombosis in patients with glioma (11). Several studies reported that MCP-1 may be involved in the recruitment of monocytes into the arterial wall during the formation of atherosclerotic plaques (9). Elevated levels of MCP-1 were detected in patients with venous thrombosis (12). Lupieri et al. reported that improved endothelial healing is a major challenge to prevent arterial thrombosis, and IP-10 can inhibit endothelial healing (13). Since IL-1β and IL-6 are routinely tested as indicators of in ammation in COVID patients, this study focused on IP-10, MCP-1, and MIP1a level in the blood serum.
IP-10(Interferon gamma inducible protein) is a small 10.8kD protein secreted by many cells in response to interferon-gamma (IFNγ). These cell types include monocytes, endothelial cells and broblasts (14). During secretion, IP-10 is cleaved into a 8.7kD bioactive protein, which acts as a chemotactic agent for T cells, NK cells, monocytes / macrophages and dendritic cells. In addition, IP-10 has antitumor activity by inhibiting bone marrow colony formation and angiogenesis. IP-10 works by binding to cell surface chemokine receptor 3 (CXCR3) (14,15). MCP-1 (monocyte chemoattractant protein-1) is a chemokine that attracts monocytes and basophils, but not neutrophils or eosinophils. MCP-1 plays a role in the pathogenesis of diseases characterized by monocyte in ltration, such as psoriasis, rheumatoid arthritis, or atherosclerosis (16). May be involved in the recruitment of monocytes to the arterial wall in the process of atherosclerosis (17). Macrophage in ammatory protein 1-alpha (MIP1a, also known as CCL3) is a monocyte cytokine with in ammatory and chemotactic properties. MIP1a can be combined with CCR1, CCR4, and CCR5 (18). In addition, it is one of the major HIV inhibitory factors produced by CD8 + t cells (19). These cytokines were measured at different time points in each patient, with the aim to verify whether these coagulation-related factors changed over time or were related to the patient's risk of death.

Patients
This study is a retrospective single-center study involving 74 ICU patients admitted to the Tongji Hospital, Wuhan City, China, with a diagnosis of severe and critical ill COVID-19 con rmed by polymerase chain reaction (PCR). As of February 7, the ICU of this hospital has been managed by the multidisciplinary medical team of the Peking Union Medical College Hospital, and most COVID-19 patients are severe and critically ill transferred from ICU of hospitals at all levels. The distinction between severe and critically ill COVID-19 patients was realized according to the "New Coronavirus Pneumonia Diagnosis and Treatment Program (Trial Version 7)" (20). Those who meet one of the following conditions are de ned as critical ill COVID-19 patients: (1) Respiratory failure occurs and mechanical ventilation is required; (2) Shock occurs; (3) Combined with failure of other organs, and ICU monitoring and treatment is required. Therefore, the patients were divided into severe patients and critically ill patients according to the above instructions.
This study was approved by the Ethics Committee of the Peking Union Medical College Hospital, and the informed consent to participate to this study was provided by all the enrolled patients or their families.

Cytokine determination
Serum was obtained by centrifugation of a 5 ml whole blood sample and was stored at -80 °C until further use. The amount of three in ammatory cytokines, such as IP-10 (ab173194), MCP-1 (ab179886), and MIP1a (ab214569) (all from Abcam Ltd., Cambridge, UK) was measured in the serum using the human enzyme-linked immunosorbent assay (ELISA) kit (Abcam). The assay was performed according to the manufacturer's instructions.

Statistical analysis
Statistical analysis was performed using SPSS 19.0 for Windows (SPSS Inc, Chicago, IL, USA). The gures were generated by GraphPad Prism 7.0 (La Jolla, CA, USA). Categorical variables were expressed as percentages, and frequency was compared using Pearson's χ² or Fisher's exact test. Continuous variables were expressed as median and interquartile range (IQR) values. The comparison of continuous variables between the two groups was performed with Student's t test and Mann-Whitney's U test, and that correlation analyses were performed using Spearman correlation analysis. A p value less than 0.05 was considered statistically signi cant.

Results
Clinical analysis and laboratory examination of 74 severe and critically ill patients The clinical analysis and laboratory tests performed on 74 severe and critically ill patients are shown in . In terms of clinical manifestations, the proportion of males in critically ill patients was higher than that in the severe patients (P = 0.024), while the remaining clinical manifestations were not statistically different between the two groups. The content of both IP-10 and MCP-1 in the serum of the critically ill patients was higher than that in severe patients (P<0.001). In addition, critically ill patients had a higher level of IL-6 in the serum compared to that in severe patients (P<0.001). No statistical difference was found in the level of other cytokines. The hematologic indicators PLT and PCT were both lower in critically ill patients compared with severe patients (P<0.001). In addition, the critically ill patients also had a signi cantly higher level of PDW, MPV and P-LCR compared with the level in the severe patients. Furthermore, most of the coagulation indicators were all signi cantly increased in critically ill patients, including PT, INR, D-dimer and FDP.
Cytokines and coagulation parameters in 74 patients with COVID-19 strati edaccording to high (≥ median) versus low (< median) IP-10 As shown in Table 2, the IP-10 results of 74 COVID-19 patients were analyzed, grouped according to severe and critically ill, and the cutoff value was found. The sensitivity of IP-10 in the prediction of critical illness was 69.1%, the speci city was 89.5%, and the AUC was 0.806 when the Youden index was the largest. The decreased IP-10 group included 34 patients while the increased group included 40 patients. The proportion of critically ill patients in the increased group (38/40) was signi cantly higher than that in the decreased group (17/34) (P<0.001). However, the mortality between the increased group (7/40) and decreased group (3/34) was not signi cantly different. The increased IP-10 group had higher IL-6, IL-8, IL-10, TNFα, PT, INR, TT, and lower PTA compared with their values in the decreased group (P<0.05).
Cytokines and coagulation parameters in 73 patients with COVID-19 strati ed according to high (≥ median) versus low (< median) MCP-1 The MCP-1 results of 73 COVID-19 patients were analyzed, are shown in Table 3, and grouped according to severe and critically ill to nd the cutoff value. The data of one patient were missing. The sensitivity of MCP-1 in the prediction of critical illness was 78.2%, the speci city was 83.3%, and the AUC was 0.852 when the Youden index was the largest. The decreased MCP-1 group included 27 patients, while the increased MCP-1 group included 46 patients. The proportion of critically ill patients in the increased group (43/46) was signi cantly higher than that in the decreased group (12/27) (P<0.001). However, the mortality between the increased group (9/46) and decreased group (1/27) was not signi cantly different.
The level of IL-6 increased, PT, INR, D-dimer and FDP were higher, and PTA decreased in the increased MCP-1 group compared with the decreased MCP-1 group (P<0.05).
Coagulation and thrombosis-relatedELISA indicators in 71 patients with COVID-19 strati ed according to high (≥ median) versus low (< median) D-dimer The D-dimer results of 71 COVID-19 patients were analyzed (Three patients' D-dimer results were missing). Table 4 shows the grouping according to survival and death, and Table 5 shows the grouping according to severe and critically ill patients. The cutoff value was found after grouping by survival and death, and the sensitivity of D-dimer in the prediction of critical illness was 100%, the speci city was . The AUC for IP-10 was 0.8057, the AUC for MCP-1 was 0.8520, and the AUC for D-dimer was 0.7222. Then, we combined these three indicators to see whether the performance of the model can be improved, and found that the combined AUC of the three could reach 0.8998, proving a good application prospect of the joint detection index of the three.
Dynamic changes of coagulation and thrombosis-related ELISAindicators Figure 2 lists the dynamic changes of coagulation and thrombosis-related ELISA indicators in the two outcomes after the critical illness turned into severe and the critical ill patients eventually died. Patients whose multi-point indicators were greater than three time points were selected for dynamic analysis. The overall index of coagulation and thrombosis-related ELISA indicators in the death group was higher than that in the survival group. Figure  When the critical illness turned into death, the INR level decreased at rst and then increased in most patients.

Correlation analysis among coagulation and thrombosis-related ELISA indicators, cytokines and coagulation-related parameters
Supplementary Table 1 lists the results of the correlation analysis among ELISA detection indexes, cytokines and coagulation parameters. Figure 4 shows

Discussion
A large amount of pathological evidence from autopsies is revealing that thrombosis is an important consequence of COVID-19 disease (21). The development of thrombosis in patients with COVID-19 is due to the fact that after the infection with the virus, the body reacts with an extreme immune response and a "cytokine storm", leading to the release of "messenger substances" that induce pneumonia. These substances are the ones causing thrombosis and blood vessel blockage (21). This work focused on the relationship between COVID-19 related pneumonia and thrombosis, by the evaluation of several parameters related to the risk of thrombosis in COVID-19 patients, and the dynamic changes of these indicators in patients with different outcomes.
Many clinical studies showed that COVID-19 is associated with coagulopathy, but it is different from the disseminated intravascular coagulation with normal platelets, PT and brinogen. A report demonstrated that the platelet count is lower in non-survivors than survivors (22), and our study con rmed this result, although we additionally demonstrated that more platelet associated parameters differed between the two groups. Some studies (23) showed that non-survivors have signi cantly higher levels of D-dimer and FDP, longer PT and live APTT than survivors at admission. In addition, 71.4% most of the non-survivors showed disseminated intravascular coagulation during hospitalization compared to survivors, with abnormal coagulation results in the late stage of the disease(23). Our results are consistent with these results previously published, con rming the abnormal coagulation function in COVID-19 patients. Therefore, we further evaluated the coagulation and thrombosis-related indicators in COVID-19 patients using ELISA.
Huang et al. (22) reported that patients infected with 2019-nCoV show a signi cant increase in serum proin ammatory cytokine levels, especially IL1β, IFNγ, IP-10 and MCP-1, which may cause the activation of the T-helper-1 (Th1) cell response. In addition, patients who require ICU admission have higher GCSF, IP-10, MCP-1, MIP1a, and TNFα concentrations than patients who do not require ICU admission, suggesting that the cytokine storm is associated with disease severity (22). Moreover, Qin et.al (24) reported that several in ammatory cytokines such as IL-2R, IL-6, IL-8, IL-10 and TNF-α were increased in severe patients compared with their level in the non-severe patients.
Our study did not nd a difference in serum IL-1β level between severe and critically ill patients, but our results revealed a difference in IL-6 level between these two groups. IL-6 is a potent inducer of the acute phase response. Indeed, it is an endogenous pyrogen mainly produced in the acute and chronic in ammatory sites, causing fever in people with autoimmune diseases or infections. IL-6 is then secreted into the serum to induce transcriptional in ammation through the interleukin 6 receptor alpha.
IP-10 and MIP1a, as well as IL-1β, IL-6, IL-8, IL-10, and TNFα, are also in ammatory cytokines, and therefore showed a strong positive correlation with each other. IP-10, MCP-1, and MIP1a are parameters related to thrombosis, thus having a signi cant correlation with the coagulation parameters.
The level of both IP-10 and MCP-1 was higher in critically ill patients than that in serious patients.
Therefore, in this study, the 74 enrolled patients were divided according to the level of IP-10 and MCP-1.
Our results showed an increased IL-6 level in the IP-10 + MCP-1 increased group compared to the IP-10 + MCP-1 decreased group. PT, INR increased, and PTA decreased in the IP-10 + MCP-1 increased group compared to the IP-10 + MCP-1 decreased group, also con rming the previous statement. Moreover, the proportion of critically ill patients in the IP-10 + MCP-1 increased group was higher than that in the IP-10 + MCP-1 decreased group, further indicating that IP-10 and MCP-1 are biomarkers for predicting the severity of COVID-19 disease. When the IP-10 and MCP1 level was compared between the survival group and the death group, no signi cant difference was found, which might be due to the fact that the selected patients were severe or critically ill, resulting in a too high mortality rate, with no difference between survival and death. In addition, several previous studies in Wuhan showed that the D-dimer level in nonsurvivors are higher than that in survivors (23,27), suggesting that the increased D-dimer level is an independent risk factors of death in COVID-19 patients (28). Therefore, patients were grouped according to the D-dimer level and the results showed that regardless of the clinical feature, the increased D-dimer group had higher IP-10 and MCP-1 level than the decreased group, while MIP1a was not statistically signi cant between the two groups. Our further speculation was that IP-10 and MCP-1 could be related to the risk of progress to death in COVID-19 patients.
A report demonstrated that CXCL10 (IP-10) inhibits endothelial recovery independently of any other in ammatory factor, and anti-CXCL10 antibody is under validation in a clinical trial to prevent cardiovascular events (13) because the more severe the COVID-19 patient is, the higher the serum IP-10 level is. Therefore, anti-IP-10 antibody treatment may represent a new approach in COVID-19 patients, especially the ones with thrombotic events.
Patients whose multi-point indicators were greater than three time points were selected for dynamic analysis. The analysis of dynamic changes revealed that the overall index of the death group was higher than that in the survival group. In addition, the indicators remarkably increase in patients with a poor outcome, while some indicators decreased in a later time, suggesting a disease change to a pathophysiological model, although further studies are needed to explain this phenomenon.
This is the rst study comparing the coagulation and thrombosis-related ELISA indicators, platelet-related parameters, routinely tested cytokines and coagulation indicators according to the guidelines when serious and critically ill patients are grouped. Furthermore, this study compared the dynamic changes of multiple indicators in the serum of patients with multi-point detection.
This study is a single-center retrospective study, thus these results might not be representative, in addition to the fact that all the included patients were severe and critically ill. Thus, these results could not be compared with the results in mild patients.
However, there are some limitations in the present study. First, the sample size is small due to the limited time and number of patients allocated to PUMCH. Second, only patients with more than three measurements were included in the dynamic changes analysis. Although the more time-points available, the better characterization of dynamics over time is allowed, this approach leaves only 14 patients for the analysis, and it may introduce some bias, as the patients who had blood samples obtained most frequently may also be the most critically ill and may thus not be representative for the entire cohort.
More multi-center studies are needed in the future to verify these results and for a comprehensive interpretation of the clinical results.
In conclusion, the level of both IP-10 and MCP-1 in the serum of critically ill patients was higher than that in severe patients, proving that IP-10 and MCP-1 are biomarkers predicting the severity of COVID-19 disease. Moreover, IP-10 and MCP-1 level increased in the D-dimer increased group compared with the decreased group, suggesting that IP-10 and MCP-1 could be related to the risk of death in COVID-19 patients. Thus, anti-IP-10 antibody treatment may represent a new approach in COVID-19 patients, especially the ones with thrombotic events. However, since the selected patients were severe or critically ill, the results did not show any difference between survival and death, suggesting the need of further research.