The new infectious disease COVID-19 currently has no established curative treatment, and has a high mortality rate when it becomes severe. Therefore, the identification of biomarkers that can predict not only the severity but also the prognosis of this disease, is important for guiding the treatment direction and saving patients’ lives.
Recent studies on COVID-19 have shown that levels of certain coagulation/fibrinolysis markers, inflammatory markers, or cytokines are good indicators for predicting a poor prognosis or severity of this disease [26–30]. This study showed that a decrease in the platelet count, which is a coagulation marker, and an increase in concentrations of P-SEP, which is an inflammation marker, were good indicators for predicting a poor prognosis and the time course of a critical condition. We also found that the prognosis of patients with severe COVID-19 was significantly worse if inflammation and coagulation were enhanced simultaneously. Additionally, measuring these two markers simultaneously and applying those levels to our proposed panel provided a high probability of diagnosing in-hospital death. To date, there have been no reports limited to critically ill patients with COVID-19. Our findings may be useful for ICU management of patients with COVID-19, especially those in a critical condition.
Thrombocytopenia is frequently the initial feature in sepsis, and it may be followed by the prolongation of global coagulation [31]. In patients with sepsis, thrombocytopenia is associated with an increase in the rate of mortality [32, 33]. Additionally, mortality of these patients is associated with prolonged thrombocytopenia and the absence of a relative increase in the platelet count [32].
In a previous study, thrombocytopenia was detected in 5–41.7% of patients with COVID-19 [34]. Thrombocytopenia also occurs more often in patients with COVID-19 with severe disease and is potentially correlated with an increased mortality rate [13, 26]. Measurement of the platelet count is a simple, economic, rapid, and commonly available laboratory parameter [13]. Therefore, we believe that daily measurement of the platelet count can discriminate between patients with COVID-19 with and without severe disease.
In this study, patients with severe COVID-19 who had a low platelet count on ICU admission (cut-off value: 153 × 103/µL) that remained low without increasing throughout the first week of stay in the ICU had a poor prognosis. This finding suggests that daily measurement of the platelet count is a useful and essential tool for predicting in-hospital mortality. Additionally, this cut-off value is similar to the predicted value for bleeding complications reported by Al-Samkari et al. [35].
P-SEP is known as soluble CD14 subtype. P-SEP is a 13-kDa protein and is a truncated N-terminal fragment of CD14, which is the receptor for lipopolysaccharide/lipopolysaccharide binding protein complexes [36, 37]. This biomarker is mainly synthesized and released by cells of the monocyte/macrophage lineage in response to a vast array of infections [38]. P-SEP concentrations increase in the blood of septic patients. The measurement of P-SEP is reported to be useful for the initial diagnosis of sepsis, evaluating the severity of sepsis, risk stratification, and monitoring clinical responses to therapeutic interventions [39–42].
Recently, studies have reported that elevated concentrations of P-SEP may be a biomarker in the prognostic assessment of patients with COVID-19 [27, 28]. Schirinzi et al. [29] reported that P-SEP and IL-6 concentrations reflect the clinical course of COVID-19. Additionally, in a population of patients with COVID-19 and acute respiratory failure in the Emergency Department, P-SEP and troponin I are accurate predictors of 30-day mortality [43]. Additionally, P-SEP is highly specific and may enable the early identification of patients who could benefit from more intensive care as soon as they enter the Emergency Department [43]. A pooled analysis also showed a positive difference in P-SEP values between patients with or without severe/critical COVID-19 illness. A previous study showed that P-SEP values were increased by 2.74-fold in patients with COVID-19 with severe/critical illness compared with those without COVID-19 [44]. To date, there have been no reports of elevated P-SEP concentrations during viral infection. There have been a few reports that showed no increase in P-SEP concentrations under an influenza virus infection [45]. However, the cases were not severe. In our study of patients with severe COVID-19, survivors had significantly higher P-SEP concentrations than non-survivors throughout the first week of their stay in the ICU. We speculate that the reason for this finding is as follows. In patients with COVID-19, macrophages actively phagocytose SARS-CoV-2 to eliminate this virus that has invaded the body. Additionally, the active and robust invasion of SARS-CoV-2 into macrophages may contribute to an increase in P-SEP concentrations in the bloodstream. The ACE2 receptor is expressed on the surface of macrophages, and the SARS-CoV-2 spike protein binds to this receptor, eventually allowing the virus to invade macrophages. Tavazzi et al. showed that viral particles were observed within macrophages from a patient with COVID-19 and severe pulmonary inflammation and cardiogenic shock requiring ECMO [46]. Therefore, we suspect that SARS-CoV-2 actively invades macrophages, causing immunodeficiency and worsening the prognosis of patients. At the same time, blood P-SEP concentrations may be elevated.
In this study, we found that patients with severe COVID-19 who had high P-SEP values on ICU admission (cut-off value: 714 pg/mL) that remained high without decreasing throughout the first week of their stay in the ICU had a poor prognosis. Previous reports showed a predicted cut-off value of 600–800 pg/mL for the diagnosis of sepsis, which was in close agreement with the cut-off value for in-hospital death in patients with severe COVID-19 [39–42].
In our study, PCT did not reflect the clinical course of COVID-19. Therefore, PCT cannot be used as an indicator for predicting a poor prognosis and the time course of a critical condition. The reasons why PCT concentrations were not greatly elevated in both groups and why there was no difference in PCT concentrations between both groups might be as follows. PCT is usually produced in parenchymal cells, such as the lung, liver, kidney, and muscle, owing to an increase in inflammatory cytokines (e.g., IL-1 and/or tumor necrosis factor (TNF)-α) when patients develop sepsis or a serious inflammatory reaction due to a bacterial infection [47, 48]. However, the majority of cytokines that are associated with the cytokine storm in viral infections, such as IL-6 and interferon-γ, are only significantly elevated in the late stage of severe COVID-19 illness [49]. Additionally, PCT is difficult to increase in viral and fungal infections [50]. Therefore, SARS-CoV-2 infections, in contrast to bacterial infection, usually induce only a modest and delayed increase in circulating PCT concentrations [47]. These findings suggest that, in patients with COVID-19 and high PCT concentrations, high PCT concentrations may be associated with bacterial co-infection. Because of the association of PCT concentrations with bacterial co-infection and severe disease, serially testing of calcitonin levels has been recommended, particularly in ICU patients [51, 52]. Serial PCT measurement may play a role in predicting evolution toward a more severe form of disease [53, 54].
Patients with COVID-19 show markedly increased inflammatory and anti-inflammatory cytokines, such as IL-6, IL-2R, IL-10, and TNF-α concentrations [30]. Among them, IL-10 concentrations reflect the severity of patients, and IL-10 concentrations are significantly higher in the severe group of patients than in the moderate group [49, 55, 56]. In our study, IL-10 concentrations were significantly higher in the non-survivor group than in the survivor group on all 8 days. A cytokine storm in patients with COVID-19 is similar to that previously observed in patients with severe acute respiratory syndrome infected by SARS-CoV. However, a unique feature of the COVID-19 cytokine storm is the dramatic elevation of IL-10 concentrations in severe/critically ill patients [55, 57–60]. IL-10 concentrations are elevated earlier than IL-6 concentrations in patients with COVID-19 [61].
How SARS-CoV-2 infection differs from SARS-CoV in its capacity to stimulate IL-10 expression is currently unknown. However, while IL-10 is usually considered an anti-inflammatory cytokine that plays a role in suppressing an excess immune system, IL-10 may play a pro-inflammatory and immune-activating role in the pathogenesis of COVID-19 [55, 62].
In this study, concentrations of NT-proBNP, which is a marker of myocardial damage, were significantly higher in the non-survivor group than in the survivor group on 6 of the 8 days, excluding Day 1. However, KL-6 and SP-A, which are markers of alveolar injury, were not suitable for predicting the outcome. We initially expected that KL-6 and SP-A would be better predictors of the outcome than NT-proBNP, but this was not found. A recent review article described that cardiac injury caused by COVID-19 infection might be an important cause of severe clinical phenotypes [54]. This review also suggested that myocardial damage is closely related to the severity of the disease and even the prognosis in patients with COVID-19. A review article by Ali et al. [63] described the following. ACE2 was first cloned from the human heart, but not bronchi, and was found to be highly expressed throughout the endothelia of coronary and renal blood vessels. This finding suggests important biological functions of ACE2 in the cardiovascular system. On the basis of these findings, we suspected that NT-proBNP may be a better predictor of the patients’ outcomes than KL-6 and SPA. However, the levels of biomarkers of myocardial injury, such as NT-proBNP and TnT, are affected by many factors, such as infection, hypoxia, and renal function. Therefore, the potential for false positives for myocardial injury in patients with COVID-19 must be considered [54]. Consequently, the patients’ outcomes should not be predicted with just one measurement of NT-proBNP.
Generally, concentrations of D-dimer, which is a breakdown product of cross-linked fibrin, are correlated with the disease severity and predict the risk of thrombosis, the requirement for ventilator support, and mortality [64]. However, in our study, D-dimer was not a useful predictive marker of in-hospital mortality. We suspect that the reason for this finding is that when patients with COVID-19 develop a severe condition and need to be admitted to the ICU, D-dimer concentrations remain high, regardless of whether they die in the hospital.
This study has limitations. First, this was a retrospective study and the measurement of each biomarker for each patient was at different times from the beginning of COVID-19 infection. Second, approximately 55% of patients were administered anticoagulant, mainly unfractionated heparin, after ICU admission for performing ECMO. Third, because this study focused on measurable markers in our hospital, few data on other markers were examined. Fourth, this study was a relatively short-term study in Fukuoka, Japan and was a single-center study. Therefore, further research is required to confirm our findings.