Circulating KL-6/MUC1 as an independent predictor for disseminated intravascular coagulation in acute respiratory distress syndrome

Authors


Akihito Yokoyama MD, Department of Hematology and Respiratory Medicine, Kochi University, Kohasu Oko-cho, Nankoku-city, Kochi 783-8505, Japan.
(fax: 81 88 880 2348; e-mail: ayokoyama@kochi-u.ac.jp).

Abstract

Objectives.  Acute respiratory distress syndrome (ARDS) patients show high levels of circulating mucin including KL-6/MUC1 (soluble MUC1 mucin). Because cancer mucin can bind vascular endothelial cells and platelets via selectins, mucin-selectin interactions are reported to trigger platelet aggregation and intravascular coagulation. Therefore, we hypothesized that KL-6/MUC1 is involved in the pathogenesis of disseminated intravascular coagulation (DIC) in ARDS. The aim of the current study is to evaluate the association between circulating KL-6/MUC1 and DIC in ARDS patients.

Design.  Observational study with structured follow-up.

Setting.  Intensive care unit in Hiroshima University Hospital.

Subjects.  Fifty-six newly diagnosed patients with ARDS.

Interventions.  Circulating levels of KL-6/MUC1 were measured during diagnosis and serially measured during the clinical course along with indices of respiratory failure, inflammation, coagulation and fibrinolysis and multiple organ dysfunction.

Results.  Acute respiratory distress syndrome patients complicated with DIC showed significantly higher levels of serum KL-6/MUC1 than patients without DIC during the clinical course. Amongst the parameters analysed at diagnosis of ARDS, KL-6/MUC1 was an independent predictor for DIC complication. The baseline level of circulating KL-6/MUC1 at diagnosis of ARDS was significantly correlated with an increased DIC score following ARDS diagnosis. Using an optimum cutoff level of KL-6/MUC1 obtained by a receiver operating characteristic curve, the sensitivity and specificity for predicting future DIC development in ARDS patients were 88.9% and 55.3%, respectively.

Conclusions.  These results suggest that KL-6/MUC1 is associated with DIC development in ARDS patients. Elevated levels of KL-6/MUC1 at diagnosis could be a predictor of DIC complication in ARDS.

Introduction

Acute respiratory distress syndrome (ARDS) is the most serious manifestation of acute lung injury. Since its description 40 years ago, accumulating knowledge and in particular, recent progress in supportive care have improved the prognosis for this intractable disease; yet prognosis remains poor with 35–50% mortality rate [1]. There is no doubt that more information about the pathophysiology of ARDS is needed to reduce mortality. ARDS patients have increasing levels of circulating soluble mucin, and the mucin levels show a positive correlation with clinical severity [2]. We found previously that circulating KL-6, a circulating high-molecular-weight glycoprotein epitopically expressed on the soluble MUC1 mucin molecule is a good marker for various interstitial lung diseases [3–5]. More recently, KL-6/MUC1 is reported as a useful prognostic marker for patients with not only interstitial lung diseases, but also ARDS [6–8]. Higher levels of circulating and epithelial lining fluid KL-6/MUC1 reflect both more severe disease and poorer prognosis in patients with ARDS [7, 8]. In patients with idiopathic pulmonary fibrosis, not only the absolute amount but also the serial change of KL-6/MUC1 is reported to be useful for monitoring therapeutic effects [9]. Though the level of serial change in KL-6/MUC1 may reflect pathophysiology in ARDS, the significance of this measure has not been well clarified.

Another aspect of ARDS is frequent complication with coagulation abnormalities. Intra-alveolar and intravascular fibrin deposition is frequently found in ARDS patients, and coagulation cascades are activated [10–12]. Disseminated intravascular coagulation (DIC), the most severe form of coagulation abnormality has been recognized as a frequently observed complication (25–70% of ARDS patients) for a long time [13]. Development of DIC contributes to the poor prognosis of ARDS [14, 15]. Many drugs have failed to improve outcomes for ARDS, however, activated protein C therapy succeeded in improving prognosis for severe sepsis and it is considered as an expected therapy [16]. Thus, the coagulation cascade is a hopeful therapeutic target [1]. A recent study demonstrated that circulating carcinoma mucins are directly involved in Trousseau syndrome, in which the underlying pathophysiology is thought to be subclinical DIC [17]. The study clearly showed that soluble cancer mucins with selectin ligands interact with leucocyte l-selectin and platelet P-selectin and can trigger intravascular thrombosis in mice [17].

Based on these findings, we hypothesized that circulating KL-6/MUC1 mucin, which are increased in patients with ARDS may contribute to DIC, a frequent complication of ARDS. To test this possibility, we serially measured circulating KL-6/MUC1 mucin, platelet counts and d-dimer, which reflect intravascular coagulation, clot formation and lysis in patients with ARDS.

Materials and methods

Patients

This study was approved by the Institutional Review Board and informed consent was obtained from each study participant or from an immediate family member. Between April, 2002 and April, 2005, venous blood samples were collected from 56 consecutive newly diagnosed ARDS patients (36 males and 20 females, 65.8 ± 14.5 year) at Hiroshima University Hospital. Clinical characteristics including age, sex and the causes of ARDS were recorded at diagnosis of ARDS. Ventilator settings including peak airway pressure, mean air way pressure and positive end-expiratory pressure levels were also recorded at diagnosis of ARDS. The diagnosis of ARDS was based on recommended criteria by the American-European Consensus Conference Committee in 1994 [18]. Samples were obtained weekly and within 24 h after patients met the ARDS criteria. Patients who spent more than 24 h after the diagnosis ARDS were excluded from our study. Patients with obvious malignancy were also excluded, as malignancy affects circulating KL-6/MUC1 levels [19]. All surviving patients were provided a 60-day follow-up period.

Measurement of circulating markers

Each serum sample was analysed for KL-6/MUC1, lactate dehydrogenase (LDH), and C-reactive protein (CRP). KL-6/MUC1 levels were determined by ELISA using a commercially available kit (Eitest KL-6, Eisai, Tokyo, Japan) as described elsewhere [5, 6]. Each plasma sample was analysed for d-dimer with commercially available latex agglutination kits (LPIA ace D-D dimer, Mitsubishi Kagaku Iatron Inc., Tokyo, Japan). Arterial blood gas was also analysed in all patients.

Severity of illness scores

Lung Injury Score, Acute Physiology and Chronic Health Evaluation (APACHE) II score, and Sequential Organ Failure Assessment (SOFA) score were calculated based on previously reported definitions at diagnosis of ARDS [20–22]. Sepsis and the number of systemic inflammatory response syndrome (SIRS) criteria met by the patients were determined according to the previous definition [23]. The subjects were also given a DIC score according to DIC diagnostic criteria published by the Japanese Ministry of Health and Welfare [24]. Briefly, clinical condition, platelet counts, prothrombin time, fibrinogen and fibrin degradation products were scored from 0 to 13. Patients were monitored for DIC complication during follow-up period. The patients whose DIC score was higher than 7 in their clinical course were diagnosed as complicated with DIC. Although complication with DIC occurred after diagnosis of ARDS, the patients who showed overt DIC during follow-up period were described as ‘ARDS patients with (or complicated) DIC’. The patients who did not suffer from DIC during the follow-up period were described as ‘ARDS patients without (or uncomplicated) DIC’. To assess the relationship between changes of DIC score and other measured variables, the differences in the amount of DIC score after diagnosis of ARDS were measured and reported as delta (Δ) DIC score.

Statistical analysis

Data are shown as the mean ± SD. Differences between groups were analysed using the Mann–Whitney U-test for median values and the Fisher’s exact test for categorical data. Correlation coefficients for the markers were calculated using the Spearman’s rank correlation coefficient analysis. Multivariate logistic regression analysis was used to identify independent predictors for DIC complication. Independent predictors for DIC complication were identified from amongst the parameters at diagnosis of ARDS: gender (male = 1, female = 0), age, KL-6/MUC1 and parameters which showed P < 0.10, when compared between ARDS patients with and without DIC. Odds ratio and 95% confidence intervals were computed for these variables. The upper left corner coordinate point of receiver operating characteristic (ROC) curve was used to determine the optimum cutoff level for predicting DIC complication. Kaplan–Meier analysis was used to evaluate DIC-free survival rate, and the differences between the two groups were analysed by the log-rank test. Statistical significance was defined as < 0.05. All analyses were performed with a statistical software package (spss for Windows, version 12.0; SPSS Inc, Chicago, IL, USA).

Results

Characteristics of patients

The characteristics of ARDS patients who participated in this study are shown in Table 1. Amongst 56 patients, there were 39 patients with a direct cause (34 with pneumonia and 5 with aspiration pneumonia) and 17 patients with an indirect cause (14 with extrapulmonary sepsis, two with trauma and one with pancreatitis) of ARDS. When comparisons were made between patients complicated with and without DIC, KL-6/MUC1, d-dimer, platelet counts and DIC score at diagnosis were significantly different (Table 1). There were no significant differences in degree of respiratory failure (Pao2/Fio2 and the Lung Injury Score), inflammation (CRP and SIRS criteria), multiple organ dysfunction (APACHI II and SOFA score), ventilator settings (peak airway pressure, mean airway pressure and positive end-expiratory pressure levels) or cause of ARDS between patients with and without DIC (Table 1). Out of 18 patients complicated with DIC, 15 (83%) were deceased, whereas 12 of those without DIC were deceased out of 38 patients (32%). Therefore, complication with DIC was frequently observed in nonsurvivors (15/27, 56%), but was relatively rare in survivors (3/29, 10%; < 0.001 by Fisher’s exact test). Serum levels of KL-6/MUC1 at diagnosis were significantly different between ARDS patients with and without DIC (601 ± 563 U mL−1 vs 376 ± 328 U mL−1; = 0.018 by Mann–Whitney U-test, Table 1), and were significantly higher in nonsurvivors than in survivors (621 ± 522 U mL−1 vs 287 ± 219 U mL−1; < 0.001 by Mann–Whitney U-test).

Table 1.   Characteristics of ARDS patients at diagnosis
 ARDS (n = 56)P-value
without DICwith DIC
  1. ARDS, acute respiratory distress syndrome; DIC, disseminated intravascular coagulation; LDH, lactate dehydrogenase; CRP, C-reactive protein; Fio2, fraction of inspired oxygen; PEEP, positive end-expiratory pressure levels; APACHE, Acute Physiology and Chronic Health Evaluation; SOFA, Sequential Organ Failure Assessment; SIRS, systemic inflammatory response syndrome.

Characteristic(n = 38)(n = 18) 
Gender (male/female)24/1412/61.000
Age (years)67.0 ± 11.863.3 ± 19.20.765
KL-6/MUC1 (U mL−1)376 ± 328601 ± 5630.018
d-dimer (μg mL−1)11.3 ± 13.019.7 ± 15.20.010
LDH (IU mL−1)362 ± 167473 ± 2730.138
CRP (mg dL−1)15.1 ± 9.314.1 ± 8.90.636
Platelet counts (×109 L−1)187 ± 95106 ± 650.001
Pao2/Fio2 ratio143.7 ± 30.6146.4 ± 32.30.806
Arterial pH7.39 ± 0.127.40 ± 0.070.352
Peak airway pressure (cm H2O)21.4 ± 5.222.2 ± 5.00.755
Mean airway pressure (cm H2O)14.0 ± 3.414.7 ± 2.90.602
PEEP (cm H2O)6.5 ± 2.47.1 ± 2.20.479
Lung Injury score2.30 ± 0.442.55 ± 0.580.116
APACHE II score17.9 ± 6.018.9 ± 5.30.660
SOFA score6.2 ± 2.67.7 ± 2.30.053
SIRS criteria2.7 ± 0.82.4 ± 1.00.214
DIC score3.2 ± 1.54.9 ± 1.4≤0.00
Cause of ARDS (direct/indirect)27/1112/60.763

Serum KL-6/MUC1 is an independent predictor for DIC complication

Measurements of KL-6/MUC1 were serially evaluated in ARDS patients complicated and uncomplicated with DIC (Fig. 1). Serum levels of KL-6/MUC1 were significantly higher in ARDS patients complicated with DIC than uncomplicated with DIC during 2 weeks following diagnosis of ARDS. A multivariate logistic regression analysis revealed that KL-6/MUC1 and platelet counts at diagnosis of ARDS were significant predictors for DIC complication, whereas gender, age, d-dimer and SOFA score were not (Table 2).

Figure 1.

 Circulating levels of KL-6/MUC1 in acute respiratory distress syndrome (ARDS) patients complicated with or without disseminated intravascular coagulation (DIC). Serum levels of KL-6/MUC1 during 2 weeks following diagnosis in patients with ARDS who developed DIC were significantly higher than in patients without DIC. The Box-whisker plots show 25th and 75th percentiles (Box) and 10th to 90th percentiles (whiskers). Thick lines in the Box denote median, the circles denote outliers. *< 0.05, **< 0.01.

Table 2.   Predictors for DIC complication in ARDS patients
VariablesOdds ratio95% Confidence intervalP-value
LowerUpper
  1. ARDS, acute respiratory distress syndrome; DIC, disseminated intravascular coagulation; SOFA, Sequential Organ Failure Assessment.

Gender (male/female)2.1760.42611.1070.350
Age (per 1 year)0.9750.9261.0270.347
KL-6/MUC1 (per 10 UmL−1)1.0271.0041.0500.024
d-dimer (per 1 μg mL−1)0.9960.9441.0510.877
Platelet counts (per 10 × 109 L−1)0.8500.7350.9840.029
SOFA score (per 1 L−1)1.0220.6771.5430.918

Relationship between KL-6/MUC1 and DIC score

To clarify the association between KL-6/MUC1 and DIC complication, the changes in DIC score following diagnosis (ΔDIC score) were also evaluated. As shown in Fig. 2, relationships between KL-6/MUC1 at diagnosis of ARDS and ΔDIC score during 3 and 5 days following diagnosis of ARDS were investigated. The KL-6/MUC1 levels were positively correlated with ΔDIC score during 3 and 5 days following diagnosis of ARDS. Moreover, the LDH, CRP, Pao2/Fio2 ratio or Lung Injury Score were not significantly correlated with ΔDIC score (data not shown).

Figure 2.

 Correlations between KL-6/MUC1 and changes in disseminated intravascular coagulation (DIC) score. There were significant positive correlations between KL-6/MUC1 and changes in DIC score during first 3 days (a) and 5 days (b) in patients with acute respiratory distress syndrome.

Predictive value of KL-6/MUC1 for DIC development

To determine the optimum cutoff levels of KL-6/MUC1 for association with DIC, ROC curve for DIC complication were depicted (Fig. 3a). The optimum cutoff levels of KL-6/MUC1 for predicting DIC complication was 300 U/ml. According to the cutoff level determined by ROC curve, incidence of DIC between KL-6/MUC1 high (≥300 U mL−1) and low (≤300 U mL−1) groups was further evaluated by Kaplan–Meier analysis (Fig. 3b). Development of DIC in the KL-6/MUC1 high group was significantly more frequent than in that of KL-6/MUC1 low group. During follow-up, DIC developed in 16 of 33 patients (48%) in the KL-6/MUC1 high group, whereas only 2 of 23 patients (9%) in the low group developed DIC. Using the optimum cutoff level of KL-6/MUC1, the sensitivity and specificity for predicting future development of DIC in ARDS patients were 88.9% and 55.3%, respectively.

Figure 3.

 KL-6/MUC1 as a predictor of disseminated intravascular coagulation (DIC) complication in acute respiratory distress syndrome. (a) Receiver operating characteristic curve of KL-6/MUC1 for prediction of DIC. (b) Kaplan–Meier analysis for the proportion of patients who remained free of DIC in the KL-6/MUC1 high (≥300) and low (≤300) groups. During follow-up, DIC developed in 16 of 33 patients (48%) in the KL-6/MUC1 high group whereas only 2 of 23 patients (9%) in the low group developed DIC. AUC = the area under the curve.

Discussion

This study suggests that circulating KL-6/MUC1 is associated with development of DIC in patients with ARDS. A multiple logistic regression analysis demonstrated that circulating KL-6/MUC1 at diagnosis of ARDS was independently associated with development of DIC. KL-6/MUC1 at diagnosis significantly correlated with an increase of DIC score, which directly defines DIC. The development of DIC was more likely in patients who showed higher amount of KL-6/MUC1 than in patients who showed lower amount of KL-6/MUC1. Of particular note is that the severity indices of respiratory failure (Pao2/Fio2 and the Lung Injury Score), inflammation (CRP and SIRS criteria), multiple organ dysfunction (APACHI II and SOFA score) and ventilator settings (peak airway pressure, mean air way pressure and positive end-expiratory pressure levels) were not associated with development of DIC.

The current study demonstrated that higher levels of KL-6/MUC1 at diagnosis of ARDS were associated with development of DIC in near term. ARDS is associated not only with inflammation but also with enhanced activation of coagulation [11, 25]. Coagulation pathways become activated by bacteria or its endotoxin and a procoagulant state develops in the vascular and the lung parenchyma [11]. Tissue factor, a pivotal molecule for the extrinsic coagulation pathway has an important role in the procoagulant state in ARDS [26, 27]. Levels of tissue factor are increased in bronchoalveolar lavage fluid specimens obtained from ARDS patients, suggesting that a procoagulant state contributes to lung inflammation [28]. Furthermore, fibrinolytic processes are inhibited in ARDS, as shown by increased levels of plasminogen activator inhibitor-1 in bronchoalveolar lavage fluid specimens [25]. The elevated levels of plasminogen activator inhibitor-1 also found in plasma and pulmonary oedema fluid of patients with acute lung injury are associated with mortality [29]. As complications of DIC have been known to increase the risk of bleeding, multiple organ failure and high mortality in patients with ARDS [14, 15], this study also shows the importance of DIC as a near-death complication. The outcome is poorer with increasing DIC score, and early diagnosis and treatment are important [30]. Using the optimum cutoff level of KL-6/MUC1 for predicting future DIC development, the specificity was not sufficiently high. However, the high sensitivity of KL-6/MUC1 for predicting DIC complication could be informative to avoid delay in the treatment of DIC.

A limitation of the present study includes the lack of a direct causal relationship between KL-6/MUC1 and DIC. The present study revealed that KL-6/MUC1 was an independent predictor of DIC complication. There was no significant difference of well-known prognostic variables between ARDS complicated with and without DIC. Moreover, development of DIC in these patients was not associated with the severity of ARDS. In fact, DIC scores of ARDS patients are not different from that of acute lung injury patients [31] or patients at risk for but not developing ARDS [15]. However, one possible explanation is that the observed association may be indirect. Circulating KL-6/MUC1 in nonsurvivors of ARDS was significantly higher than in survivors [7, 8]. In nonsurvivors, DIC was complicated more often than in survivors of ARDS. Therefore, it is possible that KL-6/MUC1 and DIC may indirectly correlate with one another.

The mechanism for association between KL-6/MUC1 and DIC could be explained in two ways. Circulating KL-6/MUC1 originates from type II alveolar epithelial cells and is quite abundant in epithelial lining fluid [8]. Destruction of air-blood barrier and increased permeability would be needed for an increase of circulating KL-6/MUC1 [4]. Those patients with greater alveolar permeability may be predisposed for KL-6/MUC1 in addition to procoagulant substances (e.g. bacterial products and tissue factor) to enter the systemic circulation from the lung. These procoagulant substances may induce systemic DIC. Therefore, KL-6/MUC1 may be indirectly associated with DIC. A second explanation is that KL-6/MUC1 may directly cause DIC. A recent report showed that injection of mucin, presumably MUC1 caused platelet-rich thrombi by interaction with selectins [17]. It has been suggested that intravascular coagulation observed in ARDS patients includes platelet-rich microthrombi [13]. In anyway, the mechanism responsible for a link between serum KL-6/MUC1 and DIC in patients with ARDS should be clarified in future.

In conclusion, this study suggests that circulating KL-6/MUC1 has a relationship with the development of DIC in ARDS patients. Baseline KL-6/MUC1 at diagnosis of ARDS is an independent predictor for future DIC development, suggesting the clinical value of KL-6/MUC1 for earlier diagnosis and treatment of DIC in ARDS. These results warrants further clinical trials to establish a role for measuring KL-6/MUC1 in patients with ARDS.

Conflict of interest statement

N. Kohno has a royalty for KL-6/MUC1. The remaining authors have no conflicts of interest.

Acknowledgements

This work was partly supported by Grants-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan.

Ancillary