Soluble urokinase plasminogen activator receptor is compartmentally regulated in decompensated cirrhosis and indicates immune activation and short-term mortality

Authors

  • H. W. Zimmermann,

    1. Department of Medicine III, University Hospital Aachen, Aachen, Germany
    2. NIHR Biomedical Research Unit and Centre for Liver Research, School of Immunity and Infection, University of Birmingham, Birmingham, UK
    Search for more papers by this author
  • P. A. Reuken,

    1. Department of Internal Medicine IV, Jena University Hospital, Friedrich Schiller University Jena, Jena, Germany
    2. Integrated Research and Treatment Center – Center for Sepsis Control and Care (CSCC), Jena University Hospital, Friedrich Schiller University Jena, Jena, Germany
    Search for more papers by this author
  • A. Koch,

    1. Department of Medicine III, University Hospital Aachen, Aachen, Germany
    Search for more papers by this author
  • M. Bartneck,

    1. Department of Medicine III, University Hospital Aachen, Aachen, Germany
    Search for more papers by this author
  • D. H. Adams,

    1. NIHR Biomedical Research Unit and Centre for Liver Research, School of Immunity and Infection, University of Birmingham, Birmingham, UK
    Search for more papers by this author
  • C. Trautwein,

    1. Department of Medicine III, University Hospital Aachen, Aachen, Germany
    Search for more papers by this author
  • A. Stallmach,

    1. Department of Internal Medicine IV, Jena University Hospital, Friedrich Schiller University Jena, Jena, Germany
    2. Integrated Research and Treatment Center – Center for Sepsis Control and Care (CSCC), Jena University Hospital, Friedrich Schiller University Jena, Jena, Germany
    Search for more papers by this author
  • F. Tacke,

    1. Department of Medicine III, University Hospital Aachen, Aachen, Germany
    Search for more papers by this author
  • T. Bruns

    Corresponding author
    1. NIHR Biomedical Research Unit and Centre for Liver Research, School of Immunity and Infection, University of Birmingham, Birmingham, UK
    2. Department of Internal Medicine IV, Jena University Hospital, Friedrich Schiller University Jena, Jena, Germany
    3. Integrated Research and Treatment Center – Center for Sepsis Control and Care (CSCC), Jena University Hospital, Friedrich Schiller University Jena, Jena, Germany
    • Department of Medicine III, University Hospital Aachen, Aachen, Germany
    Search for more papers by this author

  • F. Tacke and T. Bruns contributed equally.

Correspondence: Tony Bruns, NIHR Biomedical Research Unit and Centre for Liver Research, Institute of Biomedical Research, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK. (fax: +44 121 415 8701; e-mail: t.bruns@bham.ac.uk).

Abstract

Objective

Patients with decompensated cirrhosis are susceptible to bacterial infections, which are associated with organ failure and a high mortality rate. Reliable biomarkers are needed to identify patients who require intensified treatment. Our objective was to study the regulation and prognostic relevance of elevated concentrations of soluble urokinase plasminogen activator receptor (suPAR) in patients with advanced cirrhosis.

Design, setting and participants

We examined the associations between serum and ascitic fluid (AF) suPAR and liver function, bacterial infection, and short-term mortality in 162 consecutive patients with decompensated cirrhosis undergoing diagnostic paracentesis in a tertiary health care centre in Germany.

Main outcome measure

Twenty-eight-day mortality.

Results

Circulating suPAR levels were increased in patients with decompensated cirrhosis and correlated with the severity of liver dysfunction and systemic inflammation but were not indicative of bacterial infection. Circulating suPAR levels >14.4 ng mL−1 predicted 28-day mortality, even after adjustment for liver function and confounders [HR = 3.05 (1.35–6.90); P = 0.0076] equal to the MELD score (AUC = 0.71; 95% CI = 0.61–0.81; P < 0.001). Cut-off levels derived from cohorts without liver disease were not applicable due to the low specificity. AF suPAR levels were elevated during spontaneous bacterial peritonitis (SBP), but not during episodes in which bacteria or bacterial DNA was translocated into the ascites. AF suPAR levels correlated poorly with systemic suPAR but were associated with a more severe course of SBP and a worse outcome. In vitro experiments revealed that monocytes, and to a lesser extent neutrophils, secrete suPAR after Toll-like-receptor ligation, which led to rapid urokinase plasminogen activator receptor cleavage followed by increased synthesis.

Conclusion

Blood and ascitic suPAR levels provide distinct, but relevant prognostic information on the severity of complications in patients with end-stage liver disease.

Abbreviations
AF

ascitic fluid

AKI

acute kidney injury

ALT

alanine aminotransferase

bactDNA

bacterial deoxyribonucleic acid

CD

cluster of differentiation

CI

confidence interval

CLD

chronic liver disease

CRP

C-reactive protein

E. coli

Escherichia coli

ELISA

enzyme-linked immunosorbent assay

G-PLD

phosphatidylinositol-glycan-specific phospholipase D

HR

hazard ratio

INR

international normalized ratio

LBP

lipopolysaccharide-binding protein

LPS

lipopolysaccharide

LTX

liver transplantation

MELD

model for end-stage liver disease

PCR

polymerase chain reaction

ROC

receiver-operating characteristic

SAAG

serum-ascites albumin gradient

SBP

spontaneous bacterial peritonitis

SIRS

systemic inflammatory response syndrome

SOFA

sequential organ failure assessment

spp.

species

suPAR

soluble urokinase plasminogen activator receptor

TLR

Toll-like receptor

TNF-α

tumour necrosis factor-alpha

uPAR

urokinase plasminogen activator receptor (CD87)

WBC

white blood cells

Introduction

Decompensated cirrhosis is associated with a 1-year cumulative mortality between 28% and 52% (median = 39%) due to liver failure and the complications of liver failure [1]. Bacterial infections, including spontaneous bacterial peritonitis (SBP), pneumonia, urinary tract infections, and bacteraemia, occur in one-third of hospitalized patients with decompensated cirrhosis and contribute to increased mortality by triggering sepsis and organ failure [2-5]. One-year mortality rates in patients with decompensated cirrhosis and bacterial infections exceed 60% and have changed little over the last three decades [2]. Reliable biomarkers that identify patients with a high risk of mortality would be invaluable in deciding when and in which patients to escalate medical care, and in selecting patients for liver transplantation.

The model for end-stage liver disease (MELD) score has proven to be an accurate predictor of short- and long-term survival in hospitalized and ambulatory patients with cirrhosis [6]. Although organ dysfunction scores, such as the Sequential Organ Failure Assessment (SOFA), outperform MELD in predicting short-term mortality in critically ill patients with cirrhosis, organ dysfunction scores are rarely used outside the ICU setting [7]. Clinical and laboratory criteria, such as the presence of systemic inflammatory response syndrome (SIRS) and elevated C-reactive protein (CRP) levels, have been reported to predict mortality in patients with cirrhosis because of the association with bacterial infections and endotoxaemia [8-10]; however, the use of composite scores or clinical data is susceptible to confounders, which may limit applicability in clinical practice.

The soluble urokinase plasminogen activator receptor (suPAR) is a promising prognostic biomarker. There is evidence that elevated levels of serum suPAR are associated with an unfavourable outcome in patients with sepsis [11-13] and bacteraemia [14-16], patients admitted to the emergency department with suspected bacterial infections [17] and patients with human immunodeficiency virus-1 (HIV-1) infections [18]. Soluble urokinase plasminogen activator receptor is released into body fluids after enzymatic removal or proteolytic cleavage of cell surface CD87 (uPAR) from various immunologically active cells, acting as a chemotactic agent and as a scavenger receptor for uPA and vitronectin [19]. Despite the role of suPAR as an emerging biomarker of immune activation in infectious disease and low-grade inflammation [20], the origin and regulation of suPAR is not fully understood [21, 22].

Systemic suPAR levels correlate with liver fibrosis and inflammation in patients with chronic liver diseases in the absence of systemic inflammation and overt bacterial infection [23-25]. Furthermore, the highest circulating suPAR levels in acutely ill medical patients are detected in patients with impaired liver function and chronic liver disease [12, 17, 26]. The prognostic significance of elevated levels of suPAR in end-stage cirrhotic patients who are at risk of infectious complications, however, has not been evaluated.

Thus, we determined the levels of suPAR in matched ascitic fluid (AF) and blood as an indicator of short-term mortality in a high-risk cohort of patients with decompensated cirrhosis and suspected infection. We investigated the cellular expression and release of uPAR in neutrophils and monocytes in vitro in response to Toll-like receptor (TLR) ligands present during bacterial infections and endotoxaemia.

Methods

Study design

One hundred sixty-two consecutive patients with decompensated liver cirrhosis admitted to the Jena University Hospital between September 2010 and April 2012 were prospectively recruited for the determination of suPAR serum and AF concentrations. The inclusion criteria were as follows: cirrhosis defined by clinical, laboratory, or histological criteria; the presence of ascites accessible to diagnostic abdominal paracentesis; and ≥18 years of age and suspected bacterial infection. The exclusion criteria included acute alcoholic hepatitis, peritoneal carcinomatosis, secondary peritonitis, acute pancreatitis, and tuberculous peritonitis. The primary outcome of the study was the prediction of 28-day mortality by elevated concentrations of serum suPAR. The secondary outcomes included the association of serum suPAR with 90-day mortality and bacterial infections, the association of AF suPAR with bacterial translocation, SBP, and SBP-related complications and identification of cellular suPAR sources using in vitro models of inflammation and infection.

Written informed consent was obtained from all of the patients, and the study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki and was approved by the local Ethics Committee (26.08.2010; no. 2880-08/10).

Samples and patient data

At the time of enrolment, blood pressure, heart rate, body temperature, and samples of blood and AF were obtained. The AF total cell count was determined using an automated XE blood cell counter (Sysmex, Norderstedt, Germany). In samples with a cell count ≥ 250 mm³, the polymorphonuclear (PMN) count was manually determined using a counting chamber. Bacterial cultures were performed in BacT/ALERT blood culture bottles (Biomérieux, Durham, NC, USA) at 37 °C after inoculation with at least 10 mL of AF or blood. Protein and albumin concentrations, blood count, liver function tests, serum creatinine and sodium levels were determined by routine laboratory analysis.

The following variables were collected at study inclusion: age; gender; aetiology of cirrhosis; co-morbidities; medication; SIRS criteria within the last 24 h; evidence of bacterial infection within the last 7 days (clinical, microbiologic and imaging criteria); increased serum creatinine (>0.5 mg dL−1) or bilirubin (>2 mg dL−1) over baseline (hospitalization or within 30 days); and MELD and Child-Pugh scores. Proven bacterial infections were defined as microbiologically diagnosed infections, including, positive blood cultures (spontaneous and secondary bacteraemia), an AF neutrophil count ≥ 250 μL−1 with or without a positive AF culture (culture-negative and culture-positive SBP), an AF neutrophil count <250 μL−1 with a positive AF culture (bacterascites), a urine WBC >20 μL−1 with a positive bacterial culture >105 colony-forming units (urinary tract infection), biliary obstruction with a positive bile culture (acute cholangitis), or new infiltrates on chest x-ray with a positive quantitative culture from a bronchoalveolar lavage (pneumonia). Bacterial infections that were diagnosed using a combination of clinical and imaging criteria and did not fulfil the criteria above were categorized as clinical infections. The survival and transplantation status were assessed in July 2012 by reviewing electronic medical records and death certificate data, or by interviewing the appropriate general practitioner.

Blood samples for the control group with compensated cirrhosis without ascites were obtained from patients recruited at the University Hospital of Aachen [27].

Ex vivo and in vitro analyses

In patients with culture-negative, non-neutrocytic ascites, AF was assessed for microbial DNA fragments (bactDNA) as an indicator of bacterial translocation [28] using the VYOO multiplex polymerase chain reaction (PCR) system (SIRS-Lab, Jena, Germany), as previously described [29]. Soluble uPAR, lipopolysaccharide-binding protein (LBP), interleukin (IL)-10, IL-6 and tumour necrosis factor-alpha (TNF-α) were measured by ELISA, as described previously [24, 30]. Monocytes and neutrophils were isolated from blood obtained from healthy volunteers and stimulated in vitro using recombinant human TNF-α (Peprotech, London, UK) and TLR agonists. The surface expression of CD87 was determined using flow cytometry [Cyan; Beckman Coulter, High Wycombe, UK (for details, see Data S1)].

Statistical analysis

Patient data are reported as the median/range for continuous variables or number/fraction for discrete data. The statistical differences between groups were analysed using a non-parametric Mann–Whitney U-test or Kruskal–Wallis test with Dunn's post hoc test for continuous data or Fisher's exact test for discrete data, respectively, as two-sided tests. Nonparametric measurement of statistical dependence between two variables was performed by calculating the Spearman's rank correlation coefficient [rho (rs)]. Ninety-day survival was assessed using the Kaplan–Meier method, pairwise log-rank test, and univariate and multivariate Cox proportional hazards models, as indicated. Patients were censored at the completion of short-term follow-up or on the day of liver transplantation. In vitro data are reported as the mean and standard error of the mean, and tested for significance using a paired t-test or repeated measures anova with Dunnet's post hoc test. Statistical calculations were performed using spss 18 (SPSS, Inc., Chicago, IL, USA) and Graphpad Prism 5 (La Jolla, CA, USA).

Results

Elevated levels of serum suPAR correlate with organ function and inflammation, but not with bacterial infection and SIRS

Of the 162 patients with decompensated cirrhosis in the study population, 120 were men, 24–91 years of age (median = 58 years), and the majority presented with alcoholic liver disease (Table 1). Even in the absence of overt bacterial infection and SIRS, patients with cirrhosis and ascites had significantly higher serum suPAR levels (median = 12.9 ng mL−1; range = 5.9–45.7 ng mL−1; n = 32) than an age- and gender-matched control group of patients with compensated cirrhosis without ascites (median = 8.9 ng mL−1; range = 2.5–20.0 ng mL−1; n = 39; P < 0.001; Fig. 1a). Surprisingly, in patients with decompensated cirrhosis the serum suPAR concentrations did not significantly differ with respect to the presence of SIRS or underlying bacterial infection (Table 1; Fig. 1b). In contrast, circulating suPAR correlated with the MELD (rs = 0.44; P < 0.001) and Child-Pugh scores (rs = 0.38; P < 0.001; Fig. 1c), and with the composites of those scores (serum bilirubin, albumin, creatinine, and INR), white blood cell count, CRP, and serum sodium (Table 2). There were no significant correlations between serum suPAR and heart rate, mean arterial pressure, serum IL-6, or serum TNF-α.

Table 1. Serum suPAR levels stratified for demographic and clinical baseline characteristics in patients with advanced cirrhosis
 No (%) (n = 162)Serum suPAR
Median (interquartile range)P-value
  1. P-values derived from nonparametric testing (Mann–Whitney U-test or Kruskal–Wallis test).

  2. a

    Within the last 14 days prior to inclusion.

Gender
Female42 (26)12.0 (9.3–17.2)0.98
Male120 (74)12.4 (9.1–17.3)
Aetiology
Alcoholic123 (76)12.4 (9.5–18.1)0.09
Nonalcoholic39 (24)10.5 (7.7–15.9)
Alcohol abstinence
Yes86 (53)11.0 (9.1–15.7)0.06
No76 (47)13.8 (9.3–19.6)
Child-Pugh class
B46 (28)9.9 (6.8–13.9)0.001
C116 (72)13.6 (9.9–19.2)
Portal vein thrombosis
No140 (86)11.8 (8.9–17.1)0.34
Yes13 (8)13.7 (10.5–16.3)
Unknown9 (6)14.5 (10.2–38.5)
Oesophageal or gastric fundal varices
No49 (30)13.0 (9.0–17.9)0.41
Yes94 (58)11.2 (9.3–16.5)
Unknown19 (12)13.5 (9.0–30.8)
Gastrointestinal haemorrhagea
No147 (91)12.0 (9.1–17–4)0.91
Yes15 (9)13.7 (9.9–17.1)
Malignant tumour
No123 (76)12.4 (9.4–17.5)0.27
Hepatocellular carcinoma24 (15)12.0 (9.3–16.7)
Other15 (9)9.4 (6.8–16.8)
Diabetes mellitus
No106 (65)12.1 (9.1–17.7)0.97
Yes56 (35)12.3 (9.2–16.7)
Systemic inflammatory response syndrome
No72 (44)10.8 (9.1–16.2)0.20
Yes90 (56)12.8 (9.3–19.6)
Bacterial infection
No67 (41)12.4 (9.1–16.8)0.95
Clinical39 (24)12.4 (8.2–17.5)
Proven56 (35)11.4 (9.2–18.1)
Bacterial infection
No67 (41)12.4 (9.1–16.8)0.27
SBP, including bacterascites37 (23)14.7 (9.8–19.4)
Pneumonia21 (13)16.1 (8.6–25.3)
Urinary tract infection7 (4)15.3 (10.6–21.6)
Other30 (19)10.0 (8.6–14.4)
Blood culture result
Negative140 (86)11.9 (8.9–17.0)0.32
Positive22 (14)14.3 (9.6–22.7)
Ascitic fluid culture result
Negative141 (87)12.2 (9.1–17.0)0.27
Positive21 (13)14.7 (10.0–30.7) 
Dialysisa
No153 (94)12.2 (9.1–17.5)0.77
Yes9 (6)11.3 (9.2–16.7)
Antibiotic therapya
No72 (44)11.6 (9.0–16.7)0.44
Yes90 (56)12.7 (9.4–18.2)
Vasoactive therapya
No144 (89)12.1 (9.1–16.9)0.65
Terlipressin10 (6)11.1 (9.9–25.9)
Other vasopressors8 (5)16.4 (8.0–23.7)
Albumin replacementa
No102 (63)11.8 (8.7–16.9)0.17
Yes60 (37)12.7 (9.8–20.6)
Table 2. Nonparametric correlation of serum and ascitic suPAR levels with markers of liver function and inflammation
 Serum suPARAscitic suPAR
All patients (n = 162)No SBP (n = 132)SBP (n = 30)
  1. Spearman's rank correlation coefficient and significance level in nonparametric correlation are indicated. MELD, model for end-stage liver disease; MR-proADM: mid-regional pro-adrenomedullin; WBC: white blood cell count; INR: international normalized ratio; LBP: lipopolysaccharide-binding protein; ALT: alanine aminotransferase; SAAG: serum-ascites albumin gradient.

MELD score0.44 (P < 0.001)−0.01 (P = 0.87)0.30 (P = 0.11)
Serum MR-proADM0.42 (P < 0.001)0.19 (P = 0.04)0.40 (P = 0.04)
Bilirubin0.38 (P < 0.001)−0.04 (P = 0.66)0.11 (P = 0.55)
Child-Pugh score0.38 (P < 0.001)−0.15 (P = 0.08)0.15 (P = 0.42)
INR0.33 (P < 0.001)−0.15 (P = 0.08)0.29 (P = 0.13)
WBC0.32 (P < 0.001)0.13 (P = 0.13)0.44 (P = 0.02)
C-reactive protein0.26 (P = 0.001)0.08 (P = 0.379)0.34 (P = 0.06)
Creatinine0.25 (P = 0.002)0.22 (P = 0.01)0.31 (P = 0.10)
Serum interleukin-100.22 (P = 0.006)0.10 (P = 0.28)0.03 (P = 0.90)
Serum LBP0.14 (P = 0.07)0.14 (P = 0.12)0.40 (P = 0.03)
Heart rate0.13 (P = 0.10)0.07 (P = 0.46)0.23 (P = 0.23)
ALT0.09 (P = 0.26)−0.18 (P = 0.04)−0.26 (P = 0.17)
Serum TNF-α0.03 (P = 0.67)0.13 (P = 0.15)−0.21 (P = 0.28)
Temperature0.01 (P = 0.91)0.07 (P = 0.40)−0.06 (P = 0.77)
Mean arterial pressure0.00 (P = 0.97)−0.03 (P = 0.76)0.06 (P = 0.75)
Serum interleukin-6−0.01 (P = 0.93)−0.01 (P = 0.88)0.11 (P = 0.55)
Platelets−0.03 (P = 0.70)0.34 (P < 0.001)0.24 (P = 0.20)
SAAG−0.05 (P = 0.58)−0.19 (P = 0.03)−0.20 (P = 0.30)
Ascitic WBC count−0.09 (P = 0.28)0.09 (P = 0.31)0.02 (P = 0.91)
Age−0.17 (P = 0.03)−0.02 (P = 0.83)0.15 (P = 0.45)
Ascitic protein−0.20 (P = 0.01)0.43 (P < 0.001)0.06 (P = 0.75)
Sodium−0.20 (P = 0.01)−0.03 (P = 0.73)−0.19 (P = 0.33)
Serum albumin−0.30 (P < 0.001)0.17 (P = 0.05)−0.21 (P = 0.26)
Figure 1.

Circulating suPAR levels in advanced cirrhosis. (a) Even in the absence of bacterial infection and systemic inflammatory response syndrome (SIRS), suPAR serum concentrations were higher in patients with cirrhosis and ascites (n = 32) than in patients with compensated cirrhosis (n = 39). Tukey's box plots are shown and P values from Mann–Whitney U-test are indicated. (b) SuPAR serum concentration in 162 patients with decompensated cirrhosis showed no significant differences with respect to the presence of SIRS and clinically diagnosed or microbiologically proven infection. Distribution and median are shown and P value from Kruskal–Wallis test is indicated (for frequencies of bacterial infections, see Table 1). (c) Circulating suPAR correlated (nonparametric correlation) with the MELD and Child-Pugh scores in patients with cirrhosis and ascites. Spearman's rank correlation coefficients are indicated. (d) Serum suPAR levels in patients who died within 28 days after inclusion were higher than patients who survived or underwent liver transplantation (LTX). Median and P values from the Mann–Whitney U-test are indicated. (e) Receiver operating characteristic curve showing the optimum cut-off (14.4 ng mL−1) to predict 28-day mortality in this cohort of patients with advanced cirrhosis and suspected infection, as well as proposed cut-offs for mortality in other cohorts between 6.4 and 9.0 ng mL−1 [12, 17, 24]. (f) Kaplan–Meier analysis of survival demonstrating higher 90-day mortality in patients with serum suPAR levels ≥14.4 ng mL−1 (n = 61) than patients with lower AF serum levels (n = 101). P value from log-rank test and censored cases due to liver transplantation are indicated. (g) Hazard ratios and 95% confidence intervals for elevated serum suPAR levels ≥14.4 ng mL−1 to predict death within 28 days are indicated for the overall cohort and for subgroups with respect to aetiology of cirrhosis or the presence of bacterial infections.

Serum suPAR identifies patients at risk for short-term mortality independent of MELD score, bacterial infection, and level of inflammation

Twenty-eight days after inclusion, 34 patients (21%) had died, six (4%) underwent liver transplantation, and 122 (75%) were alive without transplantation. The documented causes of death included infection/sepsis (n = 16), renal failure and hepato-renal syndrome (n = 7), acute-on-chronic liver failure and multi-organ failure (n = 7), gastrointestinal haemorrhage (n = 2), and other causes (n = 2). Nonsurvivors presented with significantly higher levels of serum suPAR at inclusion than survivors or patients who subsequently underwent transplantation (Fig. 1d). The receiver operating characteristics (ROC) curve suggested a serum suPAR concentration of 14.4 ng mL−1 as the best cut-off for predicting 28-day mortality with a sensitivity of 71% and a specificity of 71%, whereas other proposed cut-offs (6.4–9.0 ng mL−1) [12, 17, 24] had poor discrimination of survival in this cohort (Fig. 1e). The areas under the ROC curve (AUROC) for predicting death within 28 days were 0.71 (95% CI = 0.61–0.81; P < 0.001) for serum suPAR, 0.71 (95% CI = 0.59–0.82; P < 0.001) for the MELD score, and 0.66 (95% CI = 0.56–0.79; P = 0.004) for CRP.

Using the derived cut-off, elevated levels of suPAR were associated with a HR of 4.83 (95% CI = 2.31–10.12) for 28-day mortality (P < 0.001; Table 3A, Fig. 1f) in patients with and without bacterial infections (Fig. 1g). Other parameters associated with increased 28-day mortality in univariate analysis were a higher MELD score, bacterial infection, and elevated CRP (Table 3A). Serum suPAR levels >14.4 ng mL−1 remained a significant predictor of 28-day mortality, with a HR of 2.93 (95% CI = 1.31–6.55; P = 0.009) in multivariate Cox regression after adjustment for covariates (MELD score and bacterial infection).

Table 3. Short-term survival
CharacteristicsFollow-up at 28 daysUnivariate modelMultivariate modelb
No (%) who died (n = 34)No (%) alivea (n = 128)Hazard ratio (95% CI)P-valueAdjusted hazard ratio (95% CI)P-value
(A) Risk factors for 28-days mortality
Gender
Female9 (27)33 (26)1.00 (ref)0.84
Male25 (74)95 (74)0.92 (0.43–1.98)
Age
<55 years11 (32)40 (31)1.00 (ref)0.44 (0.18c)
55–64 years10 (29)52 (41)0.74 (0.32–1.75)
≥65 years13 (38)36 (28)1.27 (0.57–2.84)
Aetiology
Alcoholic23 (67)100 (78)1.00 (ref)0.18
Non-alcoholic11 (32)28 (22)1.63 (0.80–3.35)
Child-Pugh score
<103 (9)43 (34)1.00 (ref)0.01 (<0.001c)Removed from modeln.s.
≥1031 (91)85 (66)4.63 (1.41–15.14)
MELD score
<158 (24)54 (42)1.00 (ref)0.001 (<0.001c)1.06 (1.01–1.11) per 1-point increase0.01c
15–194 (12)24 (19)1.08 (0.33–3.58)
20–246 (18)30 (23)1.30 (0.45–3.74)
≥2516 (47)20 (16)4.54 (1.94–10.62)
Malignant tumour
No24 (71)99 (77)1.00 (ref)0.51
HCC5 (15)19 (15)1.07 (0.41–2.81)
Other5 (15)10 (8)1.77 (0.69–4.65)
Diabetes mellitus
No22 (65)84 (66)1.00 (ref)0.99
Yes12 (35)44 (34)1.00 (0.50–2.03)
SIRS
No10 (29)62 (48)1.00 (ref)0.05
Yes24 (71)66 (52)2.09 (1.00–4.37)
Bacterial infection
No6 (18)61 (48)1.00 (ref)0.0041.00 (ref)0.04
Clinical8 (24)31 (24)2.31 (0.80–6.66)1.97 (0.68–5.70)
Proven20 (59)36 (28)4.56 (1.83–11.35)3.23 (1.28–8.14)
SBP in history
No31 (91)123 (96)1.00 (ref)0.24
Yes3 (9)5 (4)2.03 (0.62–6.66)
GI haemorrhage
No31 (91)116 (91)1.00 (ref)1
Yes3 (9)12 (9)1.00 (0.31–3.27)
Beta-blocker
No22 (65)62 (48)1.00 (ref)0.12
Yes12 (35)66 (52)0.57 (0.28–1.16)
C-reactive proteind
<29 mg L−17 (21)65 (51)1.00 (ref)0.003 (0.003c)Removed from modeln.s.
≥29 mg L−127 (79)63 (49)3.49 (1.52–8.02)
Serum suPAR
<14.4 ng mL−110 (29)91 (71)1.00 (ref)<0.001 (0.002c)1.00 (ref)0.009
≥14.4 ng mL−124 (71)37 (29)4.83 (2.31–10.12) 2.93 (1.31–6.55) 
 Risk factors for 28-day mortality adjusted for age and cancerRisk factors for 90-day mortality adjusted for age and cancer
Adjusted hazard ratio (95% CI)P-valueAdjusted hazard ratio (95% CI)P value
  1. MELD, model for end-stage liver disease; SIRS, systemic inflammatory response syndrome, SBP, spontaneous bacterial peritonitis; GI, gastrointestinal; suPAR, soluble urokinase plasminogen activator receptor.

  2. a

    Patients who underwent liver transplant within 28 days were censored at the date of transplantation.

  3. b

    Multivariate model includes significant variables from univariate regression in a stepwise backward analysis (P < 0.05).

  4. c

    P-value in Cox hazard regression, when treated as a continuous variable.

  5. d

    Cut-off derived from Cervoni et al. [10].

(B) Prognostic relevance of serum suPAR for 28- and 90-day mortality
MELD score1.06 (1.02–1.12) per 1-point increase0.008c1.07 (1.03–1.11) per 1-point increase<0.001c
Bacterial infection
No1.00 (ref)0.061.00 (ref)0.02
Clinical2.13 (0.73–6.23)2.09 (0.97–4.47)
Proven3.06 (1.20–7.79)2.51 (1.30–4.83)
Serum suPAR
<14.4 ng mL−11.00 (ref)0.0071.00 (ref)0.03
≥14.4 ng mL−13.05 (1.35–6.90)1.94 (1.09–3.46)

Elevated levels of circulating suPAR were predictive of survival, even after 90 days of follow-up in univariate (HR = 2.93; 95% CI = 1.73–4.96; P < 0.001) and multivariate analyses (HR = 1.94; 95% CI = 1.09–3.46; P = 0.03; Table 3B).

Ascitic fluid concentrations of suPAR indicate SBP and correlate with severity

Serum and AF concentrations of suPAR were weakly correlated (r = 0.16; P = 0.05), but showed an interesting pattern of differential regulation in which subgroups of patients predominantly showed AF or serum elevation (Fig. 2a).

Figure 2.

Ascitic fluid suPAR levels. (a) Serum and ascitic fluid (AF) concentrations of suPAR did not correlate well and revealed the highest AF concentrations in patients with spontaneous bacterial peritonitis (SBP). (b) Ascitic fluid (AF) suPAR concentrations were highest in the presence of culture-positive (n = 13) and culture-negative (n = 17) SBP, but did not differ between sterile ascites without bacterial DNA fragments (bactDNA; n = 85), sterile ascites with bactDNA (n = 39), and monomicrobial non-neutrocytic bacterascites (n = 8). The mean and SEM are indicated and the P value from the Kruskal–Wallis test is shown [*P < 0.05 and **P < 0.01 (Dunn's post hoc test)]. (c) SuPAR concentrations in AF correlated with local concentrations of lipopolysaccharide-bind protein (LBP) in patients with SBP. Spearman's rank correlation coefficient is indicated. (d) AF suPAR levels in patients who died within 28 days after SBP were higher than patients who survived or underwent liver transplantation (LTX) or died without presenting with SBP. The median and P values from the Mann–Whitney U-test are indicated. (e) Kaplan–Meier analysis of survival indicates the highest mortality in patients with SBP and AF suPAR levels ≥13.9 ng mL−1 (black dashed line, n = 14) compared to patients with SBP and lower AF suPAR levels (black solid line, n = 16) and to patients without SBP (grey solid line, n = 132). P value from the log-rank test is indicated. (f) Patients with SBP and AF suPAR levels ≥13.9 ng mL−1 presented numerically more often with hypotension (systolic blood pressure < 90 mmHg), decreased renal function (increase in serum creatinine ≥0.3 mg dL−1 compared with hospitalization or recent baseline values) and acute kidney injury according to the AKIN criteria (increase in serum creatinine ≥0.3 mg dL−1 within 48 h after inclusion).

Patients with SBP had higher concentrations of AF suPAR (median = 12.8 ng mL−1; range = 3.6–74.9 ng mL−1) than patients without SBP (median = 9.4 ng mL−1; range = 2.2–27.7 ng mL−1; P = 0.001). Furthermore, the levels of AF suPAR were higher in patients with culture-positive SBP (median = 17.8 ng mL−1; range = 8.0–60.9 ng mL−1) than patients with culture-negative SBP (median = 11.1 ng mL−1; range = 3.6–74.9; P = 0.025; Fig. 2b) and correlated with the concentration of AF lipopolysaccharide-binding protein (LBP) (rs = 0.54; P = 0.002; Fig. 2c). The AF suPAR concentration suggested SBP at an AUROC of 0.70 (95% CI = 0.59–0.81; P = 0.001) and culture-positive SBP at an AUROC of 0.83 (95% CI = 0.71–0.94; P < 0.001).

In 39 patients with sterile, non-neutrocytic ascites, fragments of bacterial DNA were detected, which indicated bacterial translocation (Table S1). The median levels of AF suPAR did not differ between bacterial deoxyribonucleic acid (bactDNA)-positive and bacterial deoxyribonucleic acid (bactDNA)-negative patients and were comparable to patients with non-neutrocytic monomicrobial bacterascites (Fig. 2b). In contrast to the circulating suPAR concentration, the AF concentrations were largely independent of the liver function scores (Table 2).

The AF suPAR levels were higher in 28-day non-survivors after SBP (Fig. 2d), with 13.9 ng mL−1 as the best cut-off based on ROC analysis [area under curve (AUC) = 0.65; sensitivity = 65%; specificity = 77%]. The cumulative 28-day survival rates were 63% in patients with SBP and AF suPAR concentrations <13.9 ng mL−1 and 21% in patients with AF suPAR concentrations >13.9 ng mL−1 [P = 0.07 (log-rank test); Fig. 2e]. Although higher AF suPAR concentrations identified patients with an increase in serum creatinine of >0.3 mg dL−1 from baseline (P = 0.03), elevated AF suPAR concentrations failed to predict subsequent acute kidney injury according to the AKIN criteria (P = 0.46; Fig. 2e). Furthermore, six of 14 patients (43%) with an AF suPAR ≥ 13.9 ng mL−1 presented with hypotension (systolic blood pressure <90 mmHg), indicating a more severe course of SBP (Fig. 2e).

Monocytes and neutrophils release suPAR in response to ligation of TLRs

Increased levels of circulating suPAR are thought to correlate with cellular immune activation [21]. Flow cytometry of peripheral blood and AF demonstrated that amongst leucocyte subsets, only monocytes, neutrophils, and CD14+ peritoneal macrophages had detectable levels of membrane-bound uPAR (CD87) as a relevant potential source of suPAR (Fig. 3a). Based on cultures of circulating leucocyte populations isolated from healthy volunteers, monocytes released more soluble uPAR (719 ± 62 pg per 106 cells) than neutrophils (224 ± 108 pg per 106 cells) within 24 h (P = 0.008).

Figure 3.

uPAR expression and suPAR release from monocytes and neutrophils. (a) Representative flow cytometry analysis of peripheral blood and ascitic fluid demonstrating uPAR (CD87) expression on circulating monocytes (left panel, dark grey) and neutrophils (left panel, black), as well as resident peritoneal macrophages (right panel, light grey), in contrast to lymphocytes (left panel, light grey). The isotype-matched controls on the respective populations are indicated. (b) Release of suPAR from freshly isolated primary human monocytes or neutrophils into cell culture media within 24 h after in vitro stimulation with combinations of lipopolysaccharide (LPS) at 10 ng mL−1 and tumour necrosis factor-alpha (TNF-α) at 10 ng mL−1. The mean and SEM of at least three independent experiments are shown. P values derived from repeated measures anova [*P < 0.05 and ***P < 0.001 (Dunnet's post hoc test)]. (c) Dose-dependent suPAR release from freshly isolated primary human monocytes or neutrophils into cell culture media within 24 h after in vitro stimulation with TNF-α and various Toll-like receptor (TLR) agonists. TNF-α was used in concentrations up to 100 ng mL−1, LPS up to 100 ng mL−1, zymosan up to 10 particles per immune cell and polyinosinic/polycytidylic acid (poly I:C) at 10 μg mL−1. (d) uPAR expression on freshly isolated monocytes after different times of incubation with medium alone (light grey area), 10 ng mL−1 of LPS (black solid line), or 10 ng mL−1 of LPS with 0.4 μL Brefeldin A (black dashed line), indicating early cleavage of uPAR from the surface, beginning recovery after 4 h and overexpression after ≥ 8 h. The isotype-matched control is indicated at 0 h (grey dashed line). One representative plot of three independent experiments is shown. Bar diagrams show the corresponding suPAR release from monocytes or neutrophils into culture medium over 24 h, indicating that up-regulation of uPAR is necessary for significant suPAR release after initial cleavage.

TNF-α and LPS, which are important mediators of immune cell activation and inflammation in cirrhosis, have been shown to up-regulate uPAR expression on neutrophils and monocytes in vitro and in vivo [31-33], but little is known about the contribution of these cells to circulating suPAR levels. Stimulation of monocytes with LPS at a concentration of 10 ng mL−1 resulted in a 2–4-fold increase in soluble uPAR in the supernatant, whereas TNF-α did not have a comparable effect in doses up to 100 ng mL−1 (Fig. 3b,c). Compared with monocytes, cultured neutrophils released less suPAR in response to LPS and TNF-α (Fig. 3b).

Having determined that TLR4 activation by LPS is a potent stimulus for suPAR release from monocytes, we next determined whether or not engagement of other TLRs had similar effects. Stimulation of monocytes and neutrophils with zymosan, a TLR2 agonist, and of monocytes with poly I:C, a TLR3 agonist, also resulted in increased suPAR release, but did not exceed the effect observed with LPS (Fig. 3c).

Early shedding of membrane-bound uPAR from monocytes by LPS is followed by uPAR recovery on the surface

To elucidate the possible mechanisms underlying suPAR release, we stimulated monocytes with LPS (10 ng mL−1) and determined the expression of uPAR at different time-points. After 2 h of stimulation, the expression of uPAR was markedly reduced, suggesting initial cleavage. At the 4 and 8 h time-points, however, we observed that uPAR reappeared on the cell surface, and after 24 h uPAR exceeded basal expression (Fig. 3d). Monocytes treated with LPS and Brefeldin A, which inhibits translocation of secretory proteins from the endoplasmatic reticulum to the Golgi [34], failed to re-express uPAR on the cell surface and released less soluble uPAR into the cell culture medium, indicating de novo synthesis of uPAR after LPS treatment (Fig. 3d). Culture of monocytes in serum-free medium did not abrogate suPAR release upon stimulation with LPS, demonstrating that proteolytic serum enzymes are not essential for suPAR shedding.

Discussion

We have reported that circulating suPAR levels in patients with advanced cirrhosis increase with clinical decompensation, correlate with worsening liver function, and indicate poor short-term survival independent of infections and sepsis, whereas AF suPAR concentrations are suggestive of SBP and are largely independent of blood suPAR levels and liver function. These findings suggest a compartmental regulation of suPAR in which blood levels reflect liver disease and AF levels indicate the presence and extent of local infection.

The AF levels of suPAR were only elevated in the presence of active infection, implying that the peritoneal infiltrate is responsible for the local level. In agreement with our results, increased local levels of suPAR in cerebrospinal fluid during bacterial meningitis [35] and in bronchoalveolar lavage specimens from patients with inhalation trauma [36] are associated with the extent of local inflammation. Mesothelial cells have been shown to release suPAR in response to LPS and inflammation [37]. It is possible that mesothelial cells contributed to suPAR levels in our patient cohort; however, AF suPAR was not elevated in patients with bacterascites or the presence of bactDNA and correlated with AF LBP levels only in the presence of infiltrating inflammatory cells, suggesting that activated leucocytes are the main source of suPAR. Moreover, our prospective data suggest that AF suPAR may be clinically useful in predicting complicated cases of SBP, which warrant escalation of treatment.

In contrast to AF suPAR, circulating levels of suPAR were elevated in patients with decompensated cirrhosis without evidence of bacterial infection or sepsis. These patients presented with median suPAR levels of 12.9 ng mL−1, which exceeded the level that predicts SIRS (2.8 ng mL−1) [38] or severe sepsis (6.6 ng mL−1) [17] in non-cirrhotic patients, showing that high circulating suPAR levels are associated with liver disease per se, and not merely a reflection of sepsis in patients with advanced cirrhosis. Circulating suPAR correlates with clinical decompensation, decreased liver function, and markers associated with liver-related mortality in cirrhosis, such as creatinine [6], sodium [39], CRP [10], IL-10 [40, 41], and mid-regional proadrenomedullin [42]. Despite a good correlation with suPAR and the MELD score, only 19% (rs2) of the variance of suPAR was explained by the MELD score alone, indicating valuable information beyond liver function. In fact, serum suPAR as a single continuous parameter predicted 28-day mortality slightly better than the composite MELD score. Patients with serum suPAR levels >14.4 ng mL−1 were three times as likely to die within 28 days and twice as likely to die within 90 days compared with patients with lower serum concentrations of suPAR, even after adjusting for the MELD score, overt bacterial infection, and co-morbidities.

Although activated neutrophils contribute to suPAR during bacterial infection and inflammation [11, 43, 44] and likely release local suPAR in SBP, our data suggest that monocytes are the major source of the circulating suPAR in patients with advanced liver disease. When activated via TLRs, monocytes secreted 5–10-fold more suPAR than neutrophils. Surface uPAR is rapidly cleaved from monocytes in the presence of LPS, followed by re-expression and continuing release [33]. Blocking cytoplasmic protein transport resulted in abrogated uPAR expression and release, indicating that continued cleavage and up-regulation of uPAR are necessary for ongoing release from monocytes.

Thus, we believe that activated monocytes and liver-resident macrophages are the major source of circulating suPAR in patients with cirrhosis, even in the absence of overt infection and SIRS. We and others [27, 45, 46] have previously shown activation of circulating and hepatic monocytes with progressive fibrosis, cirrhosis, and portal hypertension, and our current data have demonstrated a correlation between serum suPAR with serum IL-10, which is mainly derived from monocyte progeny [47]. Furthermore, endotoxaemia and bacterial translocation are characteristic of decompensated cirrhosis, thus providing a source of TLR agonists in patients with advanced cirrhosis independent of overt infection [28, 30, 48]. The presence of large numbers of activated monocytes and bacterial translocation in advanced cirrhosis probably explains why circulating suPAR levels are far higher than the circulating suPAR levels in other patients with suspected infection (6.4 ng mL−1) [17], bacteraemia (9–11 ng mL−1) [14-16], or sepsis (8–12 ng mL−1) [11, 12].

Beyond the role of suPAR as an indicator of liver-related short-term mortality and immune activation, what is the role of suPAR in liver disease? There is evidence that elevated suPAR may be harmful because depletion of suPAR ameliorates glomerular damage and restores kidney function in rodent models of focal segmental glomerulosclerosis [49], and perhaps surprisingly in murine E. coli peritonitis uPAR-deficient animals have a lower peritoneal and circulating bacterial load [50]. Thus, suPAR depletion may be a potential therapeutic approach in patients with advanced cirrhosis.

Conflict of interest statement

No conflicts of interest to declare.

Acknowledgements

The authors would like to thank Aline Roggenkamp for excellent technical assistance.

Contributions

HWZ, FT, and TB designed the study, performed the in vitro experiments, performed statistical analysis, analysed and interpreted the results, conducted the literature search and wrote the manuscript. HWZ, PAR, MB, FT and TB recruited patients, collected specimens, performed measurements and analysed clinical data. AK, DHA, CT and AS participated in the design of the study, interpreted the results and revised the manuscript. All authors read and approved the final version of the manuscript.

Funding

This work was supported, in part, by the Federal Ministry of Education and Research (BMBF) Germany (FKZ: 01 E0 1002). The German Research Foundation provided research funding to TB (BR 4182/1-1), FT (DFG Ta434/2-1 and SFB/TRR 57), and CT (SFB/TRR 57). HWZ and FT received funding from the START-Program and the Interdisciplinary Centre for Clinical Research (IZKF) within the Faculty of Medicine at the RWTH Aachen University. The suPARnostic™ ELISA kits were gifts from ViroGates (Birkeroed, Denmark).

Ancillary