BY 4.0 license Open Access Published by De Gruyter October 23, 2020

Diagnostic and prognostic role of presepsin in patients with cirrhosis and bacterial infection

Alberto Ferrarese ORCID logo, Anna Chiara Frigo, Monica Maria Mion, Mario Plebani ORCID logo, Francesco Paolo Russo, Giacomo Germani, Martina Gambato, Umberto Cillo, Annamaria Cattelan, Patrizia Burra and Marco Senzolo

Abstract

Objectives

Serum biomarkers have suboptimal accuracy for the early diagnosis of bacterial infection (BI) in cirrhosis. The aim of the study was to evaluate the diagnostic and prognostic accuracy of presepsin (PSP) in a cohort of hospitalized patients with cirrhosis.

Methods

All adult cirrhotics admitted between 03.2016 and 06.2019 were consecutively evaluated. PSP was measured using chemiluminescent enzyme immunoassay, and its accuracy was compared with that of common biomarkers.

Results

A total of 278 cirrhotic patients for a total of 448 hospitalizations were prospectively collected. Prevalence of BI at admission was 28.3%. Median (range) Log10PSP in the whole cohort was 2.83 (2.48–3.19) ng/L, significantly higher in patients with BI than in patients without (p<0.001). For a cutoff value of 2.87 ng/L, Log10PSP showed sensitivity, specificity and AUC-ROC of 0.66 (95% CI 0.57–0.74), 0.63 (95% CI 0.57–0.68) and 0.69 (95% CI 0.63–0.73), lower than that of C-reactive protein (p=0.002), but similar to procalcitonin (p=0.18) Patients with BI at hospitalization had higher probability of 28-day mortality (sub-hazard ratio [sHR] 2.65;95% CI 1.49–4.70; p=0.001). At multivariate Cox’s regression analysis, Log10PSP (sHR 2.4; 95% CI 1.22–4.82; p=0.01) together with age and severity of liver disease, was an independent predictor of short-term mortality.

Conclusions

PSP shows low diagnostic accuracy for BI in cirrhosis, but it is an independent predictor of short-term mortality. PSP may be a biomarker of systemic inflammation, commonly seen in end-stage liver disease.

Introduction

The liver actively regulates immune and inflammatory pathways, through complex mechanisms involving tolerance against toxins and immunogenic responses [1], [2]. In cirrhosis, this homeostatic role is impaired, in view of a simultaneous reduction of immune-surveillance and an increase of systemic inflammation. Therefore, cirrhosis carries a high risk of bacterial infection (BI) and overwhelming systemic responses [3]. These events often determine worsening of portal-hypertension, impairment of hepatic and extra-hepatic organs dysfunction [3], [4], and ultimately an increase in mortality [5].

Early diagnosis of BI is often difficult in patients with cirrhosis because conventional criteria – as those used for systemic inflammatory response syndrome – have several limitations, especially among the sickest ones [6]. Similarly, the commonly adopted serum biomarkers have poor accuracy [7], [8]. Leucocytes might be qualitatively and quantitatively influenced by hypersplenism and cirrhosis-associated immune dysfunction (CAID), respectively [9]; C-reactive protein (CRP) might mirror the underlying chronic inflammatory state rather than an infection [10], [11]; procalcitonin (PCT) could be falsely increased in cases of renal dysfunction or in superimposed conditions, as acute-on-chronic liver failure (ACLF) [12], [13].

Among recently proposed diagnostic biomarkers of BI, presepsin (PSP) is a 13 kDa N-terminal fragment, produced after formation and subsequent proteolysis of the complex between lipopolysaccharide and its binding protein [14]. In critically ill patients without cirrhosis, PSP has already showed a good diagnostic accuracy for sepsis and a satisfactory prognostic role [15], [16], [17], [18]. Nevertheless, its role in cirrhosis is currently poorly understood, because of small sample size and different endpoints adopted in the available studies [19], [20], [21]. Therefore, the aims of our study were to prospectively measure PSP in a large cohort of hospitalized patients with cirrhosis, according to the severity of underlying liver disease; to evaluate its diagnostic role for BI; to compare its accuracy with that of commonly used biomarkers; and to explore a prognostic role in terms of short-term survival.

Materials and methods

This was a prospective, single-center, observational study performed at Padua University Hospital. This is a Tertiary Center Hospital actively involved in the management of severely ill patients with cirrhosis, with a well-known liver transplant program. The study was approved by the local Ethical Committee. All consecutive patients admitted at Multivisceral Transplant Unit between 01.03.2016 and 30.06.2019 were prospectively evaluated. Exclusion criteria were: 18<age>80 years; absence of definite diagnosis of cirrhosis, made on conventional histological, radiological and biochemical criteria; previous solid organ transplantation (including liver transplantation); hospital admission for ≤48 h; definite or highly suspected non-bacterial (e.g., fungal, parasitic, and viral) infection at time of hospitalization. Notably, patients with acute liver failure, acute liver injury or with flares of underlying chronic liver disease (e.g., autoimmune flare, hepatitis D superinfection, hepatitis B flare) were not enrolled.

For each patient, the following variables were evaluated at the time of hospital admission: age, gender, etiology of liver disease, diagnosis of hepatocellular carcinoma, previous hospital admissions; mean arterial pressure, PaO2, presence and grade of hepatic encephalopathy, ascites, acute kidney injury; ACLF was defined according to EASL-CLIF criteria [22]. Charlson comorbidiy index was obtained for each patient at time of admission.

Each patient underwent baseline measurement of blood biomarkers (leucocyte count, CRP, PSP); for patients with high suspicion of infection, PCT was also obtained. The Model for End Stage Liver Disease (MELD) and Child-Pugh scores were collected at admission for each patient, whereas disease-specific scores (e.g., CLIF-AD, CLIF-C ACLF) were calculated when appropriate [22].

BI was diagnosed using the commonly adopted criteria, described elsewhere [23]. Each BI was classified as hospital acquired, healthcare-associated and community-acquired according to epidemiological characteristics [24]. When available, bacterial strains were classified according to Gram stain.

PSP was measured in each patient at hospital admission. In detail, a blood sample was obtained at fasting and sent to the local Laboratory Medicine, where it was analyzed. The PSP concentration was determined using the PATHFASTTM analyzer (Mitsubishi Chemical Europe GmbH), with an analytical method based on the chemiluminescent enzyme immunoassay technique. Data from manufacturer: the measuring range was between 20 and 20.000 ng/L; the imprecision has been obtained by measuring for 20 non-consecutive days in duplicate 4 plasma samples that showed a mean value between 445 and 19.292 ng/L, with a coefficient of variation between 3.8 and 5.0%; the reference range was from 57 to 337 ng/L. Plasma PSP value was expressed in this study as its logarithmic transformation (log10PSP). The outcome of each patient was prospectively evaluated in terms of 28-day survival since hospital admission, considering liver transplantation as a competing event. This study was conducted in accordance with the World Medical Association Declaration of Helsinki regarding ethical conduct of research involving human subjects.

Statistical analysis

Variables were reported as counts and percentages or median (range), as appropriate. The diagnostic accuracy of PSP, CRP, PCT, and leucocytes in predicting BI was assessed by the area under the curve (AUC) identified by the receiver operating characteristic (ROC) and estimated by a generalized linear mixed model with logit link function and a compound symmetry covariance matrix, to take into account the correlation of the data given by the repeated hospitalizations. AUCs were estimated with the 95% confidence interval (CI) obtained with the Wald method and compared by chi-square test. Threshold values were identified applying the Youden index. The sensitivity and specificity corresponding to the previously identified threshold value of PSP, CRP, PCT, and leucocytes were estimated with the 95% CI calculated with the binomial method. Probabilities of 28-day outcome were evaluated using the Fine and Gray approach considering liver transplantation as a competing event and estimated as cumulative incidence. Follow-up time was defined as the time from date of hospitalization to date of death, liver transplantation or end of follow-up, whichever came first. Risks were expressed as sub-hazard ratios (sHR) and 95% CIs. Potential predictors of mortality were identified with univariate Cox regression for competing risks considering the hospitalizations clustered within patients to take into account the correlation. The prognostic role of log10PSP in terms of 28-day survival was investigated using a multivariate Cox’s regression model, adjusted for confounders (age, presence of BI, AKI, severity of liver disease according to Child-Pugh, MELD, presence of ACLF, CRP).

The effect of Child-Pugh classes, BI, and their interaction on log10PSP was investigated with a generalized linear model considering the Normal distribution, the clustering of the hospitalizations within patients and a compound symmetry covariance matrix. Difference between presence or absence of BI on log10PSP was estimated with least squares and 95% CI for each Child-Pugh class. The same model was applied for evaluating the effect of AKI, BI and their interaction on log10PSP. The significance level was set at the 5% bilateral value. Statistical analysis was performed with SAS 9.4 (SAS Institute Inc., Cary, NC, USA) for Windows.

Results

A total of 1,280 hospitalizations occurred in the study period, and 448 (35%) were considered (Supplementary Figure 1). Characteristics of 278 patients enrolled are presented in Table 1. In detail, 70.9% were males and nearly half had alcohol related disease. All patients were admitted due to liver-related conditions, and no active, liver-unrelated inflammatory conditions occurred at time of hospitalization. BI was diagnosed at admission in about one third of patients (127/448, 28.3%), being the lower respiratory tract, urinary tract and blood the most common sources. Prevalence of hospital-acquired and culture-positive infections was 53.5 and 62.2%, respectively. Gram +ve strains were prevalent (54/79, 68.3%).

Table 1:

Characteristics of enrolled patients and clinical features at hospitalization.

Patients n=278
Male gender, n (%) 197 (70.9)
Age, years 57.4 (18–79.5)
Etiology of liver disease
 Alcohol 132 (47.5)
 HCV 42 (15.1)
 HBV 23 (8.2)
 Mixed (Alcohol + virus) 19 (6.8)
 Metabolic 22 (7.9)
 Cholestatic/autoimmune 33 (11.9)
 Other 7 (2.5)
Hepatocellular carcinoma, n (%) 69 (24.8)
Charlson comorbidity index 5 (2–13)
Hospitalizations n=448
Length of hospitalization, days 6 (2–64)
Lab tests at admission
 Hemoglobin, g/L 98.0 (8.0–173.0)
 INR 1.4 (1.0–8.1)
 Creatinine, μmol/L 85.5 (26.0–783.0)
 Bilirubin, μmol/L 50.9 (2.0–971.0)
 Na+, mEq/L 136.0 (26.5–150.0)
 Albumin, g/L 30.0 (12.0–50.0)
 Child-Pugh score 9.0 (5.0–15.0)
          A/B/C 75/151/222
 MELD score 17 (6.0–51.0)
 MELD-Na score 21 (6.0–56.0)
 CLIF-AD scorea 50.9 (24.8–73.1)
 CLIF-C ACLF score 46.8 (30.6–70.7)

  1. aOnly in patients with AD; only in patients with ACLF. AD, acute decompensation; ACLF, acute-on-chronic liver failure.

Patients with BI at admission had similar demographic characteristics than patients without, but they were sicker in terms of liver dysfunction (ACLF: 43.3 vs. 14.3%; p<0.001; median (range) Child-Pugh score: 11.0 (5.0–15.0) vs. 9.0 (5.0–15.0); median (range) MELD score: 22.2 (7.6–45.9) vs. 15.2 (6.0–50.5); each p<0.001; Table 2). Considering serum biomarkers, median (range) CRP (33.0 (2.0–276.0) vs. 9.7 (2.0–150.0) mg/L; p<0.001) and leucocytes (6.4 (0.4–31.5) vs. 5.2 (0.6–23.3)⋅109/L; p=<0.001) were significantly higher among patients with BI, whereas PCT was not (0.9 (0.0–50.0) vs. 0.4 (0.0–89.0) μg/L; p=0.61).

Table 2:

Clinical features in the whole population according to BI.

Bacterial infection n=127 No bacterial infection n=321 OR (95% CI) p-Value
Male gender, n (%) 97 (76.3) 237 (73.8) 1.12 (0.66–1.86) 0.69
Age, years 58.0 (18.1–78.5) 57.3 (18.7–79.5) 1 (0.98–1.21) 0.98
Alcoholic liver disease 60 (47.2) 142 (44.2) 1.15 (0.74–1.79) 0.54
Child-Pugh score 11.0 (5.0–15.0) 9.0 (5.0–15.0) 1.4 (1.26–1.53) <0.001
MELD score 22.2 (7.6–45.9) 15.2 (6.0–50.5) 1.10 (1.06–1.13) <0.001
MELD Na + score 25.4 (6.9–46.9) 18.7 (6.0–56.0) 1.09 (1.04–1.13) <0.001
ACLF, n (%) 55 (43.3) 46 (14.3) 4.36 (2.7–7.00) <0.001
WBC, 10 9 /L 6.4 (0.4–31.5) 5.2 (0.6–23.3) 1.09 (1.04–1.14) <0.001
CRP, mg/L 33.0 (2.0–276.0) 9.7 (2.0–150.0) 1.04 (1.03–1.05) <0.001
PCT, μg/La 0.9 (0.0–50.0) 0.4 (0.0–89.0) 1.04 (0.89–1.23) 0.61
Log 10 PSP, ng/L 3.05 (2.07–4.45) 2.74 (1.74–4.30) 3.64 (2.25–5.87) <0.001

  1. aAvailable in 254 cases. AD, acute decompensation; ACLF, acute-on-chronic liver failure; BI, bacterial infection; CRP, C-reactive protein; PCT, procalcitonin; PSP, presepsin; WBC, white blood cell count. Bold values are those with statistically significant p-value.

Presepsin

The median (range) value of plasma log10PSP at hospitalization was 2.83 (2.48–3.19) ng/L in the whole cohort. Notably, males had higher median (range) values than females (2.85 (1.74–4.45) vs 2.78 (1.83–3.80) ng/L; p=0.025) and PSP was within the pre-defined values for the general population only in a minority of cases (121/448, 27%).

Patients with BI showed significantly higher median (range) log10PSP values than patients without (3.05 (2.07–4.45) vs. 2.74 (1.74–4.30) ng/L; p<0.001). However, only 33% patients in the latter group had PSP serum levels within the pre-defined range.

Plasma log10PSP was not significantly different across different types of infection (p=0.29) and after stratification for Gram stain (Gram −ve vs. Gram +ve: 3.09 (2.28–3.98) vs. 3.00 (2.07–4.45) ng/L; p=0.64).

In the whole cohort, log10PSP rose according to severity of underlying liver disease (i.e., Child-Pugh classes; p<0.001), and presence of BI (p=0.003), even if there was no significant interaction between Child-Pugh classes and BI (p=0.25). When comparing patients with similar underlying liver dysfunction, log10PSP was significantly different according to infectious status in those belonging to Child-Pugh A class (p<0.001), but not among Child-Pugh B and C patients (each p=0.06; Figure 1).

Figure 1: 
PSP values according to severity of underlying liver disease and bacterial infection.
BI, Bacterial infection; PSP, presepsin.

Figure 1:

PSP values according to severity of underlying liver disease and bacterial infection.

BI, Bacterial infection; PSP, presepsin.

A total of 90 (20%) cases fulfilled criteria of AKI at the time of hospitalization. Presence of AKI and BI significantly influenced log10PSP values (each p<0.001), but without a significant interaction between these two factors (p=0.42; Supplementary Figure 2).

Diagnostic accuracy of PSP for BI in cirrhosis

A log10PSP cutoff value of 2.87 ng/L was able to retrieve the best diagnostic accuracy for BI in the whole cohort, displaying an AUC-ROC equal to 0.69 (95% CI 0.63–0.73), with a sensitivity and specificity equal to 0.66 (95% CI 0.57–0.74) and 0.63 (95% CI 0.57–0.68), respectively. The accuracy, sensitivity and specificity of CRP and leucocytes (WBC) were obtained using the same method (Supplementary Table 1). For a cutoff value of 27 mg/L, CRP displayed an AUC-ROC of 0.77 (95% CI 0.72–0.82), whereas leucocytes displayed an AUC-ROC equal to 0.59 (95% CI 0.53–0.65) using a cutoff value of 7.19⋅109/L. After comparison of different biomarkers, the diagnostic accuracy of PSP was lower than that of CRP (p=0.002), but higher than that of WBC (p=0.006). A model including both PSP, CRP and WBC showed an AUC-ROC equal to 0.77, not significantly different than that of CRP alone (p=0.83; Figure 2).

Figure 2: 
ROC curves for diagnostic accuracy of bacterial infection in cirrhosis.
Model curve refers to pooled diagnostic accuracy of CRP, PSP, WBC. CRP, C-reactive protein; PSP, presepsin; WBC, white blood cell count.

Figure 2:

ROC curves for diagnostic accuracy of bacterial infection in cirrhosis.

Model curve refers to pooled diagnostic accuracy of CRP, PSP, WBC. CRP, C-reactive protein; PSP, presepsin; WBC, white blood cell count.

Comparison of presepsin and procalcitonin

We have further analyzed a cohort of patients with the highest risk of BI at hospitalization (n=251), in whom PCT was also retrieved. After calculating sensitivity and specificity in this subset of patients (Supplementary Table 2), PSP, CRP and PCT displayed an AUC-ROC of 0.61 (95% CI 0.53–0.68), 0.68 (95% CI 0.61–0.75) and 0.65 (95% CI 0.58–0.75), respectively. The diagnostic accuracy of PSP was similar to that of CRP (p=0.052) and of PCT (p=0.18), but better than that of WBC (p<0.001). A model including all biomarkers showed an AUC-ROC equal to 0.69 (95% CI 0.62–0.75), with a better accuracy than that of PSP alone (p=00.03), but similar to that of PCT and CRP alone (p=00.32 and p=0.49, respectively, Figure 3).

Figure 3: 
ROC curves for diagnostic accuracy of bacterial infection in cirrhosis, including only cases with available PCT serum levels. Model curve refers to pooled diagnostic accuracy of CRP, PSP, PCT, WBC. CRP, C-reactive protein; PCT, procalcitonin; PSP, presepsin; WBC, white blood cell count.

Figure 3:

ROC curves for diagnostic accuracy of bacterial infection in cirrhosis, including only cases with available PCT serum levels. Model curve refers to pooled diagnostic accuracy of CRP, PSP, PCT, WBC. CRP, C-reactive protein; PCT, procalcitonin; PSP, presepsin; WBC, white blood cell count.

Prognostic role of Log10PSP in patients with cirrhosis

All cases included in the study were followed-up until recording the final event (i.e., hospital discharge, liver transplantation, death) and/or for at least 28 days. Twenty-eight (10%) patients underwent LT during this period, whereas 58 (20.8%) died. All but one causes of death were liver-related, being end-stage liver disease and sepsis the commonest events (37 and 36%, respectively). At competing risks analysis, patients with BI at hospitalization had a significantly higher probability of 28-day mortality than patients without (sHR 2.65, 95% CI 1.49–4.70; p<0.001;Supplementary Figure 3).

At univariate Cox’s regression analysis, age, log10PSP, Child-Pugh score, MELD score, presence of ACLF at admission, BI, AKI and CRP were all predictors of 28-day mortality (Supplementary Table 3). At multivariate Cox’s regression analysis, log10PSP (sHR 2.44 95% CI 1.22–4.82; p=0.01), together with age (sHR 1.08; 95% CI 1.04–1.12; p<0.001), Child-Pugh score C (sHR 7.84; 95% CI 1.54–143.3; p=0.004) and presence of ACLF at any stage (sHR 2.75; 95% CI 1.28–5.8; p=0.014) were independent predictors of short-term mortality (Table 3). Notably, neither CRP nor AKI and BI at hospitalization was significantly associated with short-term mortality. In a multivariate Cox’s regression model including MELD score as a significant predictor of short-term outcome (sHR 1.08; 95% 1.03–1.15; p=0.008), we confirmed that log10PSP (SHR 1.96; 95% CI 1.01–3.81; p=0.046) was an independent predictor of 28-day mortality together with age (sHR 1.09; 95% CI 1.06–1.14; p<0.001, Supplementary Table 4).

Table 3:

Predictors of 28-day mortality in hospitalized patients with cirrhosis at Cox’s regression multivariate analysis.

Parameter p-Value SHR 95% CI
A ge, years <0.001 1.08 1.04 1.12
Log 10 PSP, ng/L 0.012 2.44 1.22 4.82
Child-Pugh B 0.004 1.91 0.31 36.55
Child-Pugh C 7.84 1.54 143.33
ACLF 0.014 2.75 1.28 5.87
BI 0.71 1.14 0.60 2.12
AKI 0.87 0.94 0.45 1.99
CRP, mg/L 0.84 1.00 0.99 1.01

  1. ACLF, acute-on-chronic liver failure; AKI, acute kidney injury; CRP, C-reactive protein; PSP, presepsin. Bold values are those with statistically significant p-value.

Discussion

BI represents a common event in the natural history of cirrhosis, often associated with an impairment of hepatic and extra-hepatic organ(s) function. Our study confirmed the high prevalence of BI at hospitalization (28.3% of the whole cohort, 38.9% considering only patients admitted for AD or ACLF). Sources of infection were consistent with previous reports [23], [25], whereas we confirmed the rising epidemiological trend of Gram +ve strains among culture-positive infections [26].

An early diagnosis of BI is of utmost importance in patients with cirrhosis, in order to rapidly institute adequate antibiotic and supportive therapy. Therefore, the availability of new biomarkers or composite scores is urgently needed for such patients. The lipopolysaccharide and its receptor (CD14) have already been investigated as potential biomarkers of bacterial infection, with promising results [27], [28]. In this study, we have further explored this topic using a relatively new serum glyco-peptide, PSP, which is derived from the cleavage of lipopolysaccharide-CD14 complex.

In our cohort, the median (range) log10PSP serum level was equal to 2.83 (2.48–3.19) ng/L, a value higher than that previously reported [19]. Notably, PSP was within the range established for the general population (57–337 ng/L) in a minority of cases (27%), and displayed a wide inter-individual variability. This was confirmed also in patients without infection, who showed a log10PSP within the range previously established in 33% cases, and a log10PSP>4 ng/Lin 5/321 (1.5%) hospitalizations, being significantly influenced by renal and liver dysfunction in this cohort. This might reflect the correlation between increased bacterial translocation, high serum values of lipopolysaccharide, CAID and liver dysfunction [9], [29]. Therefore, it was not surprising that PSP was significantly different according to infection only in compensated patients (e.g., Child-Pugh A), but not in decompensated ones (Figure 1). We also confirmed that PSP might be influenced by renal dysfunction (Supplementary Figure 2), as previously shown in cohorts of patients without cirrhosis [30].

For a logarithmic cutoff value of 2.87 ng/L, diagnostic sensitivity, specificity, and accuracy of PSP for the early diagnosis of BI was equal to 0.66 (95% CI 0.57–0.74), 0.63 (95% CI 0.57–0.68), and 0.69 (95% CI 0.63–0.73), respectively. Papp et al. [19] for a PSP logarithmic cut-off value (2.94 ng/L, reported a diagnostic accuracy of 0.77 (95% CI 73–84), whereas Fischer et al. [20] using a higher cut-off (log10PSP 3.16 ng/L), showed a diagnostic accuracy of 0.75 (95% CI 0.67–0.84). Patients’ baseline characteristics (i.e., liver and renal dysfunction) and underlying clinical condition (i.e., prevalence of ACLF) between our cohort and previously published studies could explain these differences. Taken together, these data confirmed the suboptimal diagnostic accuracy of PSP in patients with cirrhosis, conversely from data obtained in non-cirrhotic patients. Indeed, a previous metanalysis on nine studies showed a PSP diagnostic sensitivity and specificity ranging from 71–100% and 62–98%, respectively [31]. The fact that CD14 acts as a co-receptor not only for LPS but also for other molecular triggers [32], and that lipopolysaccharide serum levels are over-expressed in patients with cirrhosis due to increased intestinal permeability [27], may explain this difference on diagnostic performance of PSP in cirrhosis.

Using a cut-off value of 27 mg/L, CRP was the serum biomarker having the best diagnostic accuracy in our study (p=0.002 vs. PSP). Moreover, a composite model, pooling together PSP, WBC and CRP did not perform better than CRP alone (Figure 2). For a cutoff value of 0.86 μg/L, PCT displayed a diagnostic accuracy of 0.65 (95% CI 0.58–0.75), not significantly better than that of PSP (p=0.18). It is noteworthy to mention that both biomarkers are commonly influenced by renal dysfunction and by persistent systemic inflammation, highly prevalent factors among decompensated cirrhosis and ACLF [12].

Summarizing our findings, this study confirmed that CRP remains the most reliable biomarker for the early diagnosis of BI in cirrhosis, even if without an optimal accuracy (AUC-ROC 0.77), whereas PSP has a disappointing performance. Moreover, considering only culture-positive infections, PSP values were not significantly different between Gram positive and negative strains. Several studies have clearly shown a benefit from the use of PCT-guided algorithms for the monitoring (i.e., antibiotic discontinuation) rather than for the diagnosis of infection, especially in severely ill patients [33]. Therefore, since PCT and PSP share similar characteristics in our cohort of patients, further studies investigating this topic would be of interest in the next future.

Regarding the short-term outcome, 20% patients died within 28 days. This was consistent with baseline characteristics of our cohort, in which 22.5% patients fulfilled criteria of ACLF. At competing risks analysis, patients with BI at admission had a higher probability of death than patients without (sHR 2.65, 95% CI 1.49–4.70; p<0.001; Supplementary Figure 3). Nevertheless, these patients were sicker at baseline, as shown by MELD score (22.2 (7.6–45.9) vs. 15.2 (6.0–50.5); p<0.001) and by prevalence of ACLF (43 vs. 14.3%; p<0.001). At multivariate Cox’s regression model, age, log10PSP (sHR 2.44; 95% CI 1.22–4.82; p=0.012), and severity of liver disease (according to Child-Pugh score, MELD score and presence of ACLF), were considered as independent predictors of death. On a clinical point of view, PSP may represent bacterial translocation and systemic inflammation, factors that correlate with the severity of underlying disease, in a similar way of lipopolysaccharide binding protein and interleukin-6 [9], [28]. Indeed, our data are consistent with that reported in cohorts of patients with and without cirrhosis, where prognostic value of PSP has been clearly demonstrated [20], [21], [34].

The strengths of our study are the prospective design, the evaluation of a new biomarker of systemic infection in a real-life cohort of hospitalized patients, using the same analytic method; moreover, this was the largest study evaluating both diagnostic and prognostic role of PSP in cirrhosis to date. Limitations could be the presence of a proportion of planned hospitalizations and the absence of PCT measurement in the whole cohort. Given that, sub-analyses have been conducted on these topics. Furthermore, although we have hypothesized that PSP was a biomarker of systemic inflammation, we did not collect other molecules, as interleukin-1 and interleukin-6, which have been already investigated in patients with cirrhosis [35], [36], for a direct comparison.

In summary, this study demonstrated that PSP could be considered a sub-optimal biomarker for the early diagnosis of BI in hospitalized patients with cirrhosis, with a lower accuracy than CRP; however, it was an independent predictor of short-term mortality, mirroring the pro-inflammatory state which correlates with poor outcome in end-stage liver disease.


Corresponding author: Dr. Alberto Ferrarese, Multivisceral Transplant Unit, Gastroenterology, Department of Surgery Oncology and Gastroenterology, Padua University Hospital, via Giustiniani 2, 35128, Padua, Italy, Phone: +39 (0)0498218727, E-mail:

  1. Research funding: None declared.

  2. Author contributions: A.F, AC.F participated in research design, performance of the research, data analysis, and writing of the manuscript; MM.M, M.P. participated in research design and preparation of the manuscript; FP.R., M.G., G.G. participated in performance of the research, data analysis, and writing of the manuscript; U.C., AM.C, P.B. participated in research design and preparation of the manuscript; M.S. participated in research design, data analysis, and preparation of the manuscript. All authors have accepted responsibility for the entire content of this manuscript and approved its submission.

  3. Competing interests: Authors state no conflict of interest.

  4. Informed consent: Informed consent was obtained from all individuals included in this study.

  5. Ethical approval: The study was approved by the local Ethical Committee.

References

1. Strnad, P, Tacke, F, Koch, A, Trautwein, C. Liver – guardian, modifier and target of sepsis. Nat Rev Gastroenterol Hepatol 2017;14:55–66. https://doi.org/10.1038/nrgastro.2016.168. Search in Google Scholar

2. Broad, A, Jones, DE, Kirby, JA. Toll-like receptor (TLR) response tolerance: a key physiological “damage limitation” effect and an important potential opportunity for therapy. Curr Med Chem 2006;13:2487–502. https://doi.org/10.2174/092986706778201675. Search in Google Scholar

3. Moreau, R, Soubrane, O, Sogni, P, Hadengue, A, Gaudin, C, Lin, HC, et al. Hemodynamic, neurohumoral, and metabolic responses to amino acid infusion in patients with cirrhosis. Gastroenterology 1992;103:601–8. https://doi.org/10.1016/0016-5085(92)90853-q. Search in Google Scholar

4. Thalheimer, U, Triantos, CK, Samonakis, DN, Patch, D, Burroughs, AK. Infection, coagulation, and variceal bleeding in cirrhosis. Gut 2005;54:556–63. https://doi.org/10.1136/gut.2004.048181. Search in Google Scholar

5. Arvaniti, V, D’Amico, G, Fede, G, Manousou, P, Tsochatzis, E, Pleguezuelo, M, et al. Infections in patients with cirrhosis increase mortality four-fold and should be used in determining prognosis. Gastroenterology 2010;139:1246–56, 56.e1-5. https://doi.org/10.1053/j.gastro.2010.06.019. Search in Google Scholar

6. Piano, S, Bartoletti, M, Tonon, M, Baldassarre, M, Chies, G, Romano, A, et al. Assessment of Sepsis-3 criteria and quick SOFA in patients with cirrhosis and bacterial infections. Gut 2018;67:1892–9. https://doi.org/10.1136/gutjnl-2017-314324. Search in Google Scholar

7. Singer, M, Deutschman, CS, Seymour, CW, Shankar-Hari, M, Annane, D, Bauer, M, et al. The third international consensus definitions for sepsis and septic shock (Sepsis-3). J Am Med Assoc 2016;315:801–10. https://doi.org/10.1001/jama.2016.0287. Search in Google Scholar

8. Jalan, R, Fernandez, J, Wiest, R, Schnabl, B, Moreau, R, Angeli, P, et al. Bacterial infections in cirrhosis: a position statement based on the EASL Special Conference 2013. J Hepatol 2014;60:1310–24. https://doi.org/10.1016/j.jhep.2014.01.024. Search in Google Scholar

9. Albillos, A, Lario, M, Álvarez-Mon, M. Cirrhosis-associated immune dysfunction: distinctive features and clinical relevance. J Hepatol 2014;61:1385–96. https://doi.org/10.1016/j.jhep.2014.08.010. Search in Google Scholar

10. Tsiakalos, A, Karatzaferis, A, Ziakas, P, Hatzis, G. Acute-phase proteins as indicators of bacterial infection in patients with cirrhosis. Liver Int 2009;29:1538–42. https://doi.org/10.1111/j.1478-3231.2009.02088.x. Search in Google Scholar

11. Mackenzie, I, Woodhouse, J. C-reactive protein concentrations during bacteraemia: a comparison between patients with and without liver dysfunction. Intensive Care Med 2006;32:1344–51. https://doi.org/10.1007/s00134-006-0251-1. Search in Google Scholar

12. Elefsiniotis, IS, Skounakis, M, Vezali, E, Pantazis, KD, Petrocheilou, A, Pirounaki, M, et al. Clinical significance of serum procalcitonin levels in patients with acute or chronic liver disease. Eur J Gastroenterol Hepatol 2006;18:525–30. https://doi.org/10.1097/00042737-200605000-00012. Search in Google Scholar

13. Li, CH, Yang, RB, Pang, JH, Chang, SS, Lin, CC, Chen, CH, et al. Procalcitonin as a biomarker for bacterial infections in patients with liver cirrhosis in the emergency department. Acad Emerg Med 2011;18:121–6. https://doi.org/10.1111/j.1553-2712.2010.00991.x. Search in Google Scholar

14. Chenevier-Gobeaux, C, Borderie, D, Weiss, N, Mallet-Coste, T, Claessens, YE. Presepsin (sCD14-ST), an innate immune response marker in sepsis. Clin Chim Acta 2015;450:97–103. https://doi.org/10.1016/j.cca.2015.06.026. Search in Google Scholar

15. Behnes, M, Bertsch, T, Lepiorz, D, Lang, S, Trinkmann, F, Brueckmann, M, et al. Diagnostic and prognostic utility of soluble CD 14 subtype (presepsin) for severe sepsis and septic shock during the first week of intensive care treatment. Crit Care 2014;18:507. https://doi.org/10.1186/s13054-014-0507-z. Search in Google Scholar

16. Matera, G, Quirino, A, Peronace, C, Settembre, P, Marano, V, Loria, MT, et al. Soluble CD14 subtype-A new biomarker in predicting the outcome of critically ill septic patients. Am J Med Sci 2017;353:543–51. https://doi.org/10.1016/j.amjms.2017.03.036. Search in Google Scholar

17. Galliera, E, Massaccesi, L, de Vecchi, E, Banfi, G, Romanelli, MMC. Clinical application of presepsin as diagnostic biomarker of infection: overview and updates. Clin Chem Lab Med 2019;58:11–7. https://doi.org/10.1515/cclm-2019-0643. Search in Google Scholar

18. Brodska, H, Valenta, J, Pelinkova, K, Stach, Z, Sachl, R, Balik, M, et al. Diagnostic and prognostic value of presepsin vs. established biomarkers in critically ill patients with sepsis or systemic inflammatory response syndrome. Clin Chem Lab Med 2018;56:658–68. https://doi.org/10.1515/cclm-2017-0839. Search in Google Scholar

19. Papp, M, Tornai, T, Vitalis, Z, Tornai, I, Tornai, D, Dinya, T, et al. Presepsin teardown – pitfalls of biomarkers in the diagnosis and prognosis of bacterial infection in cirrhosis. World J Gastroenterol 2016;22:9172–85. https://doi.org/10.3748/wjg.v22.i41.9172. Search in Google Scholar

20. Fischer, P, Grigoras, C, Bugariu, A, Nicoara-Farcau, O, Stefanescu, H, Benea, A, et al. Are presepsin and resistin better markers for bacterial infection in patients with decompensated liver cirrhosis? Dig Liver Dis 2019;51:1685–91. https://doi.org/10.1016/j.dld.2019.05.025. Search in Google Scholar

21. Elefsiniotis, I, Tsakiris, SA, Barla, G, Tasovasili, A, Vrachatis, D, Mavrogiannis, C. Presepsin levels in cirrhotic patients with bacterial infections and/or portal hypertension-related bleeding, presenting with or without acute kidney injury. Ann Gastroenterol 2018;31:604–12. https://doi.org/10.20524/aog.2018.0292. Search in Google Scholar

22. easloffice@easloffice.eu EAftSotLEa, Liver EAftSot. EASL Clinical Practice Guidelines for the management of patients with decompensated cirrhosis. J Hepatol 2018;69:406–60. https://doi.org/10.1016/j.jhep.2018.03.024. Search in Google Scholar

23. Ferrarese, A, Vitale, A, Sgarabotto, D, Russo, FP, Germani, G, Gambato, M, et al. Outcome of a first episode of bacterial infection in candidates for liver transplantation. Liver Transpl 2019;25:1187–97. https://doi.org/10.1002/lt.25479. Search in Google Scholar

24. Friedman, ND, Kaye, KS, Stout, JE, McGarry, SA, Trivette, SL, Briggs, JP, et al. Health care--associated bloodstream infections in adults: a reason to change the accepted definition of community-acquired infections. Ann Intern Med 2002;137:791–7. https://doi.org/10.7326/0003-4819-137-10-200211190-00007. Search in Google Scholar

25. Dionigi, E, Garcovich, M, Borzio, M, Leandro, G, Majumdar, A, Tsami, A, et al. Bacterial infections change natural history of cirrhosis irrespective of liver disease severity. Am J Gastroenterol 2017;112:588–96. https://doi.org/10.1038/ajg.2017.19. Search in Google Scholar

26. Ferrarese, A, Zanetto, A, Becchetti, C, Sciarrone, SS, Shalaby, S, Germani, G, et al. Management of bacterial infection in the liver transplant candidate World J Hepatol 2018;10:222–30. https://doi.org/10.4254/wjh.v10.i2.222. Search in Google Scholar

27. Albillos, A, de la Hera, A, González, M, Moya, JL, Calleja, JL, Monserrat, J, et al. Increased lipopolysaccharide binding protein in cirrhotic patients with marked immune and hemodynamic derangement. Hepatology 2003;37:208–17. https://doi.org/10.1053/jhep.2003.50038. Search in Google Scholar

28. Albillos, A, de-la-Hera, A, Alvarez-Mon, M. Serum lipopolysaccharide-binding protein prediction of severe bacterial infection in cirrhotic patients with ascites. Lancet 2004;363:1608–10. https://doi.org/10.1016/s0140-6736(04)16206-5. Search in Google Scholar

29. Lin, CY, Tsai, IF, Ho, YP, Huang, CT, Lin, YC, Lin, CJ, et al. Endotoxemia contributes to the immune paralysis in patients with cirrhosis. J Hepatol 2007;46:816–26. https://doi.org/10.1016/j.jhep.2006.12.018. Search in Google Scholar

30. Nagata, T, Yasuda, Y, Ando, M, Abe, T, Katsuno, T, Kato, S, et al. Clinical impact of kidney function on presepsin levels. PloS One 2015;10:e0129159. https://doi.org/10.1371/journal.pone.0129159. Search in Google Scholar

31. Wu, J, Hu, L, Zhang, G, Wu, F, He, T. Accuracy of presepsin in sepsis diagnosis: a systematic review and meta-analysis. PloS One 2015;10:e0133057. https://doi.org/10.1371/journal.pone.0133057. Search in Google Scholar

32. Landmann, R, Müller, B, Zimmerli, W. CD14, new aspects of ligand and signal diversity. Microb Infect 2000;2:295–304. https://doi.org/10.1016/s1286-4579(00)00298-7. Search in Google Scholar

33. Schuetz, P, Wirz, Y, Sager, R, Christ-Crain, M, Stolz, D, Tamm, M, et al. Procalcitonin to initiate or discontinue antibiotics in acute respiratory tract infections. Cochrane Database Syst Rev 2017;10:CD007498. https://doi.org/10.1002/14651858.CD007498.pub3. Search in Google Scholar

34. Yang, HS, Hur, M, Yi, A, Kim, H, Lee, S, Kim, SN. Prognostic value of presepsin in adult patients with sepsis: systematic review and meta-analysis. PloS One 2018;13:e0191486. https://doi.org/10.1371/journal.pone.0191486. Search in Google Scholar

35. Remmler, J, Schneider, C, Treuner-Kaueroff, T, Bartels, M, Seehofer, D, Scholz, M, et al. Increased level of interleukin 6 associates with increased 90-day and 1-year mortality in patients with end-stage liver disease. Clin Gastroenterol Hepatol 2018;16:730–7. https://doi.org/10.1016/j.cgh.2017.09.017. Search in Google Scholar

36. Lin, S, Huang, Z, Wang, M, Weng, Z, Zeng, D, Zhang, Y, et al. Interleukin-6 as an early diagnostic marker for bacterial sepsis in patients with liver cirrhosis. J Crit Care 2015;30:732–8. https://doi.org/10.1016/j.jcrc.2015.03.031. Search in Google Scholar

Supplementary Material

The online version of this article offers supplementary material (https://doi.org/10.1515/cclm-2020-1212).

Received: 2020-08-09
Accepted: 2020-10-14
Published Online: 2020-10-23
Published in Print: 2021-03-26

© 2021 Alberto Ferrarese et al., published by De Gruyter, Berlin/Boston

This work is licensed under the Creative Commons Attribution 4.0 International License.