The differential diagnosis is challenging in cases requiring distinction between acute respiratory distress syndrome (ARDS) and cardiogenic pulmonary edema (CPE), which may delay initiation of critical treatment measures (for example, lung-protective ventilation, prone positioning, neuromuscular blockade).The diagnosis of ARDS requires the exclusion of left atrial hypertension, which is usually dependent on the clinical judgment along with echocardiography and invasive hemodynamic monitoring.BNP and NT-proBNP were proved useful in excluding CPE and identifying patients with ARDS, but renal dysfunction, aging in zcritically ill patients limit the discriminative role of them.The plasma levels of presepsin may be an excellent predictor of outcome in patients with ARDS or CPE.

CD14, a cluster of differentiation (CD) marker protein expressed on the surface of various cells including monocytes, macrophages, neuutrophils, and B cells, is a multifunctional cell surface glycoprotein.CD14 is thought to be the lipopolysaccharide-binding protein (LPS-LBP) complex receptor, which exists either in a glycosylphosphatidylinositol (GPI)-anchored membrane form (mCD14), or in a circulating soluble form (sCD14)^[1-2].Circulating plasma proteases activate a cleavage of sCD14, generating a 13-kDa truncated N-terminal fragment of 64 amino acid residues named sCD14 subtype (sCD14-ST) or presepsin^[1].Recent studies showed that plasma presepsin levels were significantly increased in septic patients, and were positively correlated with the severity of sepsis^[3-7].In addition, in vitro studies indicated that sCD14 could interact with human bronchial epithelia cells (HBECs) to augment the production of IL-8 and IL-6^[8].Blockade of CD14 may attenuate the development of acute lung injury and suppress the activation of macrophage function after LPS challenge ^[9].Thus, the aim of the present study was to investigate the plasma levels of presepsin and analyze its value in patients with ARDS.

Materials and methods

Research object  The prospective, observational trial was undertaken from Jan., 2013 to Jun., 2014.A total of 108 consecutive patients with ARDS or CPE admitted into ICU of Zhongshan Hospital, Fudan University (Shanghai, China) were included.Exclusion criteria included age < 18 years, pregnant women, immunosuppression due to medication or disease and patients on haemodialysis.Healthy volunteers (n=20, free from pulmonary or cardiac disease) were defined as normal individuals.The study protocol was approved by the Ethics Committee of Zhongshan Hospital in accordance with the Declaration of Helsinki.Written informed consent was obtained from all participating people or their legal representatives.

Diagnostic criteria  Patients were enrolled within 6 h from the time of admission to the ICU.Two intensivists who were blinded to the results of presepsin reviewed all the other available clinical information integrated with progression of the disease and response to therapy, then made the final diagnosis of ARDS or CPE.Patients with ARDS met the following consensus definition:a known clinical result or new or worsening respiratory symptoms within 1 week; bilateral opacities which were not fully explained by effusions, lobar/lung collapse, or nodules; respiratory failure not fully explained by cardiac failure or fluid overload; severe hypoxemia with a ratio of arterial oxygen partial pressure and inspiratory oxygen fraction (PO₂/FiO₂) less than 300 with positive end expiratory pressure (PEEP)≥5 cm H₂O^[10].Patients with CPE was diagnosed by a combination of clinical signs (gallop rhythm, jugular venous distension, systolic hypertension); radiographic (cardiothoracic ratio>0.53 and vascular pedicle width>65 mm), electrocardiographic (new ST-segment and T-wave changes), laboratory (elevated troponin T >0.1 ng/mL), and hemodynamic findings [pulmonary arterial occlusion pressure (PAOP)≥18 mmHg, decreased ejection fraction<#60;45%, presence of severe left-sided valvular heart disease, such as aortic or mitral stenosis or regurgitation]; and the response to appropriate therapy (preload/afterload reduction, treatment of ischemia or inotropic agents)^[11].

Measures and indexes  Patients underwent an initial clinical assessment at enrollment.Invasive hemodynamic monitoring, echocardiogram, pulmonary function and CT angiography were performed according to the treating physician.Insertion of a Swan-Ganz catheter was required when the diagnosis was uncertain.Comorbidities, ventilatory data, hemodynamic and laboratory findings were recorded at enrollment.Acute Physiology and Chronic Health Evaluation (APACHE) Ⅱscore, Lung Injury Score (LIS) and the Sequential Organ Failure Assessment (SOFA) score were also calculated.The primary clinical risk factor for ARDS was classified as direct pulmonary (pneumonia, aspiration) or indirect non-pulmonary (sepsis, hemorrhagic shock, resuscitation, multiple transfusion).For some analysis, the causes of ARDS were classified into infected group (infection as the cause) and non-infected group (all other causes, including hemorrhagic shock, aspiration, resuscitation and multiple transfusion).In addition, the degree of hypoxemia of ARDS was divided into mild (200<PO₂/FiO₂≤300), moderate (100<PO₂/FiO₂≤200) and severe (PO₂/FiO₂≤100).Patients were followed for the end point of 28-day mortality over a 28-day period after enrollment.

Blood samples for determination of presepsin were collected at enrollment and 4 days later for all subjects and were centrifugated within the next 1 hour.Plasma samples were frozen at -80 ℃ for further analysis.Levels of presepsin were measured using the PATHFAST^® analysis system (Mitsubishi Chemical Medicine Corporation, Japan) based on a chemiluminescent enzyme immunoassay (CLEIA) which has a normal reference range of 60-365 pg/mL.

Statistics analysis  Continuous variables are presented as x±s or median (with interquartile range), and categorical variables as numbers and percentages.Unpaired Student's t-test or Mann-Whitney test were used between groups for continuous variables and chi-square for categorical variables.If the difference among multiple groups was indicated significant by the Kruskal-Wallis test, then the Mann-Whitney test was used to further evaluate the difference between them.Correlations among continuous variables were assessed by the Spearman rank analysis.Receiver operating characteristic (ROC) curves were utilized to evaluate the accuracy of presepsin to diagnose ARDS.Value of presepsin was used to predict 28-day mortality.The optimal cutoff value was determined when the Youden index reached the maximum value.Logistic regression was assessed by univariate and multivariate analysis to identify independent predictors of outcome.All probabilities were two tailed and P < 0.05 was regarded as significant.Data were statistically analyzed with SPSS 16.0 software.

Results

Patient characteristics  A total of 128 subjects including 81 patients with ARDS, 27 patients with CPE and 20 healthy volunteers were enrolled in the study.Causes of ARDS included pneumonia in 35 patients (43.2%), non-pulmonary sepsis in 19 (23.5%), aspiration in 10 (12.3%), resuscitation in 7(8.6%), hemorrhagic shock in 6 (7.4%) and multiple transfusion in 4 (4.9%).Causes of CPE included congestive heart failure in 14 patients (51.9%), myocardial infarction/ischemia in 7 (25.9%) and acute volume overload in 6 (22.2%).

Baseline Characteristics of the study population were presented in Tab 1 stratified according to the final diagnosis.Echocardiographic data was obtained in 79 patients (73.1%), and hemodynamic data in 40 patients (37.0%).Compared with patients with ARDS, patients with CPE were more likely to have a history of atrial fibrillation and higher PO₂/FiO₂.Echocardiographic and hemodynamic data indicated patients with CPE had lower left ventricular ejection fraction (LVEF) and higher PAOP.

Presepsin values  Patients with ARDS had significantly higher median levels of presepsin [926.89(485.41-2 662.32) pg/mL] compared with CPE patients [376.21 (247.16-568.52) pg/mL, P < 0.001, Tab 1] and with healthy subjects [116.23 (92.26-145.65) pg/mL, P < 0.001] at enrollment.There was no significant difference between CPE patients and healthy subjects in terms of presepsin value (P=0.189).The area under ROC curve for presepsin in relation to the final diagnosis of ARDS from CPE was 0.803±0.042 (Fig 1A).At a cut point >618.58 pg/mL, presepsin provided a specificity of 81.5% and a sensitivity of 70.4% for the diagnosis of ARDS.When the analysis was restricted to 87 patients (80.6%) without renal dysfunction (creatinine clearance>60 mL/min), the AUC for presepsin improved to 0.835±0.051.

Patients with ARDS were classified into infected group (52 patients, 64.2%) and non-infected group (29 patients, 35.8%).The median presepsin levels at enrollment were 934.74 (456.44-3 322.51) pg/mL and 798.12 (485.41-2 561.40) pg/mL respectively, without statistical significance (P=0.079).In the infected group, there was no marked difference in presepsin levels between Gram-positive and Gram-negative bacterial infection.Furthermore, the presepsin levels of non-infected patients with ARDS were significantly higher than that of patients with CPE [798.12 (485.41-2 561.40) pg/mL vs.376.21(247.16-568.52) pg/mL, P= 0.004].There was no significant difference between pulmonary and non-pulmonary group in terms of presepsin value.

ARDS patients were divided into mild (n=21), moderate (n=48) and severe (n=12) group according to the degree of hypoxemia.There was no significant difference in presepsin value between mild and moderate group [769.62 (428.86-1 291.51) pg/mL vs.944.90 (419.45-2 183.10) pg/mL, P=0.633].However, the presepsin value of moderate group was significantly lower than that of severe group [944.90 (419.45-2 183.10) pg/mL vs.3 681.32 (948.25-4 629.63) pg/mL, P=0.003] (Fig 2).

In patients with ARDS, presepsin levels were correlated with serum creatinine (r=0.71, P < 0.001), APACHEⅡ score (r=0.63, P=0.031), SOFA score (r=0.61, P=0.023) and PO₂/FiO₂ (r=-0.52, P=0.029) at enrollment.

The mortality was 27.2%in patients with ARDS.Compared with survivors, non-survivors had significantly higher APACHE Ⅱscore, SOFA score, arterial lactate and lower PO₂/FiO₂ (Tab 2).The presepsin levels of non-survivors were significantly higher than that of survivors [3 158.30 (963.91-4 489.33) pg/mL vs.729.09 (398.05-1 467.24) pg/mL, P < 0.001] at enrollment.ROC curve was drawn to evaluate the value of presepsin to predict 28-day mortality and the AUC was calculated as 0.776±0.062 (Fig 1B).The optimal cutoff value for predicting death was >2 024.9 pg/mL, which gave specificity of 83.1% and sensitivity of 68.2%.Kaplan-Meier curve was drawn according to the value of 2 024.9 g/mL for as a cut-point to describe death over 28 days follow-up (Fig 3).There was a significant difference in the occurrence of death (P<0.001).

Univariate logistic regression analysis showed that APACHE Ⅱ score, SOFA score, PO₂/FiO₂ and plasma presepsin levels at enrollment were the common predictors of 28-day mortality in patients with ARDS.Multivariate logistic regression analysis showed SOFA score (OR=1.81, P=0.033) and presepsin(OR=1.51, P=0.027) remained the independent predictor for mortality after adjustment for risk factors (APACHEⅡscore, PO₂/FiO₂)(Tab 3).

In ARDS patients whose plasma presepsin level increased over 4 days had a trend toward an increased risk of death compared with those whose plasma level decreased over time (mortality:36.1%vs.18.1%, P=0.024).The association between increasing presepsin and death was statistically significantin patients with ARDS (OR=1.42, P=0.035).

Discussion

There were no identified biomarkers or tools with high-quality evidence for differentiating ARDS from CPE, because there is no objective gold standard for diagnosing ARDS or CPE.In our study, we found that patients with ARDS had significant higher median levels of presepsin compared with patients with CPE.Presepsin may be useful in identifying patients with ARDS and CPE.In this way, ARDS and CPE can be identified early and treated correctly and effectively.

Univariate Logistic regression analysis showed that APACHE Ⅱ score, SOFA score, PO₂/FiO₂ and plasma presepsin levels at enrollment were the common predictors of 28-day mortality in patients with ARDS.As we all known, APACHE Ⅱ, SOFA score, PO₂/FiO₂ were associated with poor outcome.In some studies, the initial oxygenation abnormality defined by the PO₂/FiO₂ ratio did not predict mortality unless it was grossly abnormal^[23].Studys supported the use of SOFA score as an aid to identify patients with increased risk of in-hospital mortality among patients hospitalized with infection.In our study, presepsin levels of non-survivors was significantly higher than survivors.Multivariate logistic regression analysis showed presepsin at enrollment was the independent predictor for 28-day mortality.

CD14 plays a role in monocyte activation, signal transduction, leukocyte aggregation, and cyte adhesion on endothelial cells^[12].Blockade of CD14 with monoclonal antibodies prevents the monocyte synthesis of tumor necrosis factor-α (TNF-α)^[13].Presepsin which generated by circulating plasma proteases activating cleavage of soluble CD14 is associated with phagocytosis and cleavage of microorganisms.It has been found that presepsin levels in patients with systemic and localized bacterial infections were significantly higher than those in patients with non-bacterial infectious diseases^[5].Beyond that, sCD14 levels were specifically increased in serum and bronchoalveolar lavage fluid (BALF) of children with pneumonia compared with cystic fibrosis, asthma and healthy subjects^[14].ARDS is characterized by extensive neutrophil influx into the lungs, production of proinflammatory mediators, and damage of lung epithelial and endothelial surfaces^[15].Martin, et al^[16] found that sCD14 increased in BALF and serum of patients with ARDS.In the present study, the presepsin levels in patients with ARDS were significantly higher than patients with CPE.In addition, the difference in presepsin levels between infected and non-infected patients with ARDS was not significant.CD14 is an essential receptor in LPS-induced lung inflammation, meanwhile, it also exists in a soluble form (sCD14) which is able to mediate LPS-activation of cells devoid of membrane CD14 expression, such as epithelial and endothelial cells^[17-19].sCD14 can reconstitute the whole (physiological and biological) response to inhaled LPS in the lung.The augment of the endogenous concentration of sCD14 can enhance neutrophil inflammation in response to inhaled LPS^[20].Moreover, the release of IL-8 and IL-6 from human bronchial epithelial cells (HBECs) was increased in a concentration-dependent manner upon stimulation with sCD14 both in the presence and absence of LPS^[8].

sCD14 is deemed to act as a key component in pulmonary inflammation, meanwhile, may represent a promising marker and therapeutic target in respiratory diseases.Pulmonary edema and neutrophil emigration after LPS challenge were attenuated by blocking CD14 using an anti-CD14 monoclonal antibody^[8].A significant decrease in protein leakage into BAL fluid was also induced by CD14 blockade^[21].Our study showed that presepsin was the independent predictor for 28-day mortality in ARDS patients.Beyond that, patients whose plasma presepsin level decreased over 4 days had a decreased risk of death compared with those whose plasma level increased over time.The decreasing trend could be related to an early clearance or to a reduced production of presepsin as a consequence of the appropriate treatment.It is conceivable that early measurement of circulation presepsin will help in monitoring the appropriateness of the early therapy adopted.

CD14 has been shown to be a specific receptor of LPS, a compound from the outer cell wall of Gram-negative bacteria.In addition, CD14 may function as a receptor for peptidoglycan, the major cell wall component of Gram-positive bacteria^[1-2].The elevated value of sCD14 in Gram-positive septic patients could thus be due to enhanced liberation of sCD14 caused by cytokines such as TNF-α^[22].This may be the reason why there is no marked difference in presepsin levels between Gram-positive and Gram-negative bacterial infection in our study.

There are several limitations to our study.Firstly, the study sample was small in size and was a single-center study, thus restricting generalizability.Secondly, we did not take in-depth analysis of the relation between presepsin levels and the therapy implemented.Finally, we did not test the concentration of presepsin in alveolar fluid and compare with other biomarkers of ARDS.

In conclusion, our findings suggested that presepsin was a biomarker which could distinguish ARDS from CPE.Furthermore, it was a strong predictor for short-term mortality in ARDS.The clinical indications of presepsin in diagnosis and prognosis should be validated in large-scale patients with ARDS.

Acknowledgments  We thank the patients who participated in this study and their families.We are indebted to all doctors, nurses and support staff in our ICU for their indefatigable work on assisting patients.