Journal of Critical Care xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

Journal of Critical Care journal homepage: www.jccjournal.org

Interleukin-6 as an early diagnostic marker for bacterial sepsis in patients with liver cirrhosis☆ Su Lin a,1, Zhongtao Huang b,1, Mingfang Wang a, Zhiyuan Weng c, Dawu Zeng a, Yanliang Zhang d, Yueyong Zhu a,⁎, Jiaji Jiang a a

Liver Research Center, First Affiliated Hospital of Fujian Medical University, Fuzhou, Fujian, 350005, China Department of Gastroenterology, The First Hospital of Putian City, The training hospital of Fujian Medical University, Putian, Fujian, 351100, China Cardiology of the First Affiliated Hospital of Fujian Medical University, Fuzhou, Fujian, 350005, China d Department of Infectious Diseases, Nanjing Hospital Affiliated to Nanjing Medical University, Nanjing 210006, China b c

a r t i c l e

i n f o

Keywords: Interleukin-6 Cirrhosis Bacterial sepsis Procalcitonin

a b s t r a c t Objective: Liver cirrhosis is associated with frequent bacterial infections that increase the mortality rate. However, the early diagnosis and treatment of these infections are often difficult. In this retrospective-prospective observational study, the serum levels of interleukin-6 (IL-6) and procalcitonin (PCT) were measured in 233 cirrhotic patients to evaluate the early diagnostic and prognostic values of IL-6 and PCT for cirrhotic patients. Methods: Cirrhotic patients admitted to the Liver Research Center of the First Affiliated Hospital of Fujian Medical University between 1 October 2012 and 30 June 2014 were enrolled. They showed no evidence of infection on admission, and all had first onset of fever and met the systemic inflammatory response syndrome criteria 72 hours after admission. The serum IL-6 and PCT levels were determined on admission, at the onset of fever (0 hour) and 24 and 48 hours after fever onset. Results: A total of 233 cirrhotic patients, including 183 men and 50 women, with a median age of 56 (46-65) years were enrolled. A training group of 159 patients was retrospectively enrolled from 1 October 2012 to 31 December 2013, and a validation group of 74 patients was prospectively enrolled from 1 January 2014 to 30 June 2014. Among these patients, 134 were diagnosed with bacterial sepsis, 96 of whom were in the training group and 38 of whom were in the validation group; infections were ultimately ruled out in 99 patients: 63 training patients and 36 validation patients. At 0 hour, the IL-6 and PCT levels as well as the proportion of neutrophils were much higher in septic patients than in nonseptic ones. The IL-6 level and proportion of neutrophils peaked upon the onset of fever, 24 hours before the PCT levels and white blood cell count, and then sharply declined. The area under the receiver operating characteristic curve of IL-6 for diagnosing sepsis was largest at the onset of fever (area under the receiver operating characteristic curve, 0.983; 95% confidence interval, 0.967-0.999). The threshold of IL-6 for diagnosis was 135 pg/mL, with a sensitivity of 94.8% and a specificity of 93.7%. These diagnostic values were also confirmed in the validation group, with a sensitivity of 97.4% and specificity of 80.6%. Eleven (11.5%) patients died, and 85 (88.5%) patients recovered in the sepsis group of training patients after a 4-week follow-up. The IL-6 level was significantly higher in the nonsurvival group than that in the survival group (1813.00 vs 472.10 pg/mL, P = .004) at the onset of sepsis. The cutoff value for predicting prognosis was 1105 pg/mL, with a sensitivity of 81.8% and a specificity of 76.5%. Conclusions: The serum IL-6 levels increased earlier than the PCT in septic cirrhotic patients. The direct measurement of the serum IL-6 level can help to rapidly detect bacterial infection, thus allowing for early therapeutic decisions and prognostic predictions. © 2015 Elsevier Inc. All rights reserved.

1. Introduction Abbreviations: T, body temperature; R, respiratory rate; HR, heart rate; BP, blood pressure; IL-6, interleukin-6; CRP, C-reactive protein; PCT, procalcitonin; SIRS, systemic inflammatory response syndrome; WBC, white blood cell count; N%, proportion of neutrophils; HCV, hepatitis C virus; HBV, hepatitis B virus; DM, diabetes mellitus. ☆ Financial support: This study was financially supported by the Health Science and Technology Development Fund of Nanjing (QYK11173). ⁎ Corresponding author. Tel.: +86 591 87981660. E-mail address: [email protected] (Y. Zhu). 1 Contributed equally to the writing of this article.

Patients with advanced cirrhosis are prone to bacterial infections, with a prevalence of 30% on admission or during hospitalization [1]. Such infections have become the leading cause of death in these patients [2]. The in-hospital mortality rate of patients with cirrhosis who have septic shock exceeds 70%, which is much higher than the rate of the patients without cirrhosis [3,4]. Antibiotics should be intravenously administered immediately after sepsis is diagnosed. Any delay in the initiation of appropriate antibiotics in patients with severe sepsis is

http://dx.doi.org/10.1016/j.jcrc.2015.03.031 0883-9441/© 2015 Elsevier Inc. All rights reserved.

Please cite this article as: Lin S, et al, Interleukin-6 as an early diagnostic marker for bacterial sepsis in patients with liver cirrhosis, J Crit Care (2015), http://dx.doi.org/10.1016/j.jcrc.2015.03.031

2

S. Lin et al. / Journal of Critical Care xxx (2015) xxx–xxx

associated with an increase in the mortality rate [5–7]. However, the early diagnosis of sepsis is extremely difficult, especially in patients with cirrhosis. In the noncirrhotic population, 2 acute-phase serum proteins, serum C-reactive protein (CRP) and procalcitonin (PCT), have been widely used as diagnostic markers of infection [8]. Several studies have demonstrated that the CRP levels negatively correlated with the extent of liver failure [9,10]. C-reactive protein has been shown to be less diagnostic in the cirrhotic population [11]. Procalcitonin has been shown to react faster with an inflammatory stimulus than CRP [12]. It is mainly produced by the liver; thus, liver dysfunction interferes with the synthesis of PCT [13].The diagnostic accuracy of PCT in cirrhotic patients remains controversial [11]. Interleukin-6 (IL-6) is produced during bacterial infections and is involved in the initiation of the acute-phase response in humans [14]. The level of IL-6 also increases earlier than that of PCT [12]. The level of PCT increases for 6 to 8 hours after exposure to bacterial products and reaches a plateau 12 hours thereafter [15,16]. Bacterial infection can be detected earlier by directly measuring the serum level of IL-6 compared to that of other acute-phase proteins, which are secreted in response to proinflammatory cytokines. Several studies have affirmed the diagnostic value of IL-6 in septic patients in the general population [17–19]. Byl et al [20] found that IL-6 level was elevated with deterioration of cirrhosis due to infection. Wang et al [21] also reported that the IL-6 levels significantly increased from the baseline levels in cirrhotic patients with hospital-acquired infections. Therefore, IL-6 may be a promising early diagnostic marker of bacterial sepsis in patients with cirrhosis. However, the best threshold for IL-6 and its role in the routine diagnosis of bacterial infection in cirrhotic patients remain undetermined because of the limited number of patients. In this cohort study, we focused on cirrhotic patients who suffered from bacterial sepsis during hospitalization by analyzing the levels of IL-6 and PCT as well as the white blood cell count (WBC) and proportion of neutrophils (N%) to evaluate the rapid diagnostic and predictive values of serum IL-6 for cirrhotic patients with bacterial sepsis. 2. Methods

(4) WBC greater than 12 000/mm3 or less than 4000/mm3 or immature forms greater than 10%. Infection was diagnosed based on the following criteria [23]: (1) clinical symptoms of infection, such as fever, hypotension, abdominal pain, diarrhea, cough, expectoration, and frequent, urgent, or painful urination; (2) pathogenic bacterial growth in blood cultures or in normally sterile sites (eg, ascites, urine); (3) radiography or computed tomography test of chest, urine sediment, or ascitic fluid cell count was positive; (4) for patients with negative cultures, infection was clinically and/or radiologically confirmed as well as based on the response to antibiotic treatment. Infection was excluded based on the following evidence: (1) no clinical symptoms of infections; (2) radiography or computed tomography test of chest, urine sediment and culture, ascitic fluid cell count and cultures, and blood culture were all negative; (3) fever stopped spontaneously without antibiotic therapy, or patients did not respond to antibiotic therapy. Sepsis was diagnosis based on the presence of SIRS and infections. 2.3. Data collection The patient’s vital signs, including the highest body temperature (T), lowest blood pressure (BP), and highest heart (HR) and respiratory rates (R), were recorded on admission and upon the onset of fever (0 hour) as well as 24 and 48 hours after the onset of fever. Routine blood tests and microbiological culture were performed at the same time point. The outcome (survival or death) was assessed after a 4week follow-up. Serum samples from all cirrhotic patients were collected on admission and stored at −80°C to measure the baseline PCT and IL-6 levels. 2.4. Bacterial culture and measurements of plasma PCT and IL-6 levels Blood cultures were obtained from separate venipuncture sites in at least 2 collection bottles. All cultures were incubated at 35.8°C in an appropriate atmosphere and observed for 7 days. The IL-6 and PCT levels were determined with a commercially available electrochemiluminescence analyzer (ECL, Roche E601, Switzerland).

2.1. Patients This work was a retrospective-prospective cohort study conducted from 1 October 2012 to 30 June 2014. Cirrhotic patients hospitalized at the Liver Research Center of the First Affiliated Hospital of Fujian Medical University were analyzed in this study. Patients were hospitalized because of gastrointestinal bleeding, hepatic encephalopathy, hepatorenal syndrome, massive ascites, a recent deterioration of liver function, or hepatocellular carcinoma. Because determining the accurate time of bacterial sepsis was difficult when the infections developed in the community, which would preclude the assessment of initial data, only patients who showed no signs of infection on admission but developed a fever and the met systemic inflammatory response syndrome (SIRS) criteria during hospitalization were included in this research. Patients showing any signs of infection within 72 hours after admission and those with alcoholic hepatitis were excluded from this study because the baseline IL-6 levels were elevated in these patients [22]. The training cohort was enrolled from 1 October 2012 to 31 December 2013, and the validation cohort was prospectively enrolled from 1 January to 30 June 2014. The same parameters were collected in both groups.

2.5. Cost-effectiveness analysis The cost-effectiveness of different biomarkers for the diagnosis of sepsis was evaluated in the training group. The effectiveness was determined based on the sensitivity for diagnosis, and the cost depended on the expense of the tests. According to the prices paid in China, the cost was $8.6 (¥54) for IL-6 detection, $38.5 (¥240) for PCT detection, and $4.8 (¥30) for a routine blood test. 2.6. Statistical analysis The data are expressed as median values with interquartile ranges. A Mann-Whitney test or Student t test was used to compare groups. The significance of the difference in proportions was tested with the χ2 statistic. The diagnostic accuracy is expressed as the area under the receiver operating characteristic curve (AUC), which was derived from a logistic regression analysis. Statistical significance was achieved if P b .05 with a 2-tailed tests. All data were analyzed with SPSS (Chicago, IL) 13.0.

2.2. Diagnosis of sepsis 2.7. Ethics Systemic inflammatory response syndrome was diagnosed when 2 or more of the following criteria were met [23]: (1). body temperature greater than 38°C or less than 36°C; (2) tachycardia greater than 90 beats per minute; (3) hyperventilation: respiratory rate greater than 20 beats per minute or arterial hypocapnia less than 32 mm Hg;

The study protocol was approved by the Institutional Ethics Committee of the First Affiliated Hospital of Fujian Medical University and complied with the Declaration of Helsinki. Written informed consent for the use of stored serum samples was obtained from all patients.

Please cite this article as: Lin S, et al, Interleukin-6 as an early diagnostic marker for bacterial sepsis in patients with liver cirrhosis, J Crit Care (2015), http://dx.doi.org/10.1016/j.jcrc.2015.03.031

S. Lin et al. / Journal of Critical Care xxx (2015) xxx–xxx

3. Results 3.1. Baseline characteristics Within 21 months, a total of 233 cirrhotic patients, 183 men and 50 women, with a median age of 56 (46-65) years were enrolled in this study. Of these patients, 159 patients were retrospectively enrolled from 1 October 2012 to 31 December 2013 as the training group; and 74 patients were prospectively enrolled from 1 January 2014 to 30 June 2014 as the validation group. The following underlying diseases were identified in the training group: 136 (85.5%) cases of hepatitis B virus (HBV)–associated cirrhosis, 12 (7.5%) cases of hepatitis C virus (HCV)–associated cirrhosis, 3 (1.9%) cases of primary biliary cirrhosis (PBC) or autoimmune hepatitis (AIH), 4 (2.5%) cases of Wilson disease, and 4 (2.5%) cases of other unknown cirrhosis. The majority of cases showed decompensated cirrhosis, with 76 (47.8%) cases of Child-Pugh stage B and 55 (34.6%) cases of stage C, whereas only 28 (17.6%) patients were compensated (Child-Pugh stage A). Thirty (18.9%) patients had diabetes mellitus (DM) (Table 1). All patients developed a fever during hospitalization. Ninety-six (60.4%) patients were ultimately diagnosed with bacterial sepsis (sepsis group), 8 (8.3%) of whom experienced septic shock. Thirty-seven (38.5%) cases of spontaneous bacterial peritonitis, 25 (26.0%) cases of pneumonia, 18 (18.8%) cases of biliary infection, and 16 (16.7%) cases of urinary infection were identified. The blood cultures were positive in 35 (36.4%) patients, and the cultures of other body fluids (ascites or urine) were positive in 28 (29.2%) patients; the other 33 (34.4%) patients yielded negative bacterial cultures. The following pathogens were identified: 15 cases of Klebsiella pneumoniae, 13 of Escherichia coli, 6 of Acinetobacter baumannii, 6 of Staphylococcus, 5 of Pseudomonas aeruginosa, 3 of Enterococcus, 3 of Proteus, 2 of Streptococcus, and 10 of other origin. All patients received empiric intravenous antibiotics within 24 hours after the onset of fever. The remaining 63 (39.6%) patients did not suffer from sepsis (nonsepsis group). They developed fevers because of the following reasons: 26 (41.3%) cases of drug- or blood transfusion–induced fever, 22

3

(34.9%) cases of viral upper respiratory tract infection, 8 (12.7%) cases of postoperative fever after radiofrequency ablation or transcatheter arterial chemoembolization of hepatocellular carcinoma, and 7 (11.1%) cases of unexplained fever origin. The vital signs (T, BP, HR, R) and laboratory parameters (WBC, N%, PCT, and IL-6 levels) did not significantly differ between the nonsepsis and sepsis patients at baseline in the training group (P N .05) (Tables 1 and 2).

3.2. Patients’ characteristics upon the onset of fever in the training group In the nonsepsis group, the WBC and IL-6 levels were higher at 0 hour than at baseline (P b .05); but the level of PCT and N% did not significantly differ between baseline and 0 hour (Table 2). In patients with sepsis, T, HR, and R were significantly higher at the onset of fever than at baseline (Table 2). The level of IL-6 sharply increased at the onset of fever by more than 20-fold compared with the baseline level (24.73 vs 538.55). The level of PCT was only four- to fivefold higher than the baseline level (0.21 vs 0.96). The level of IL-6 at 0 hour did not correlate with the temperature (r = 0.107, P = .308), septic shock (r = 0.177, P = .085), age (r = −0.002, P = .981), Child-Pugh score (r = −0.055, P = .597), or DM (r = −0.040, P = .701).

3.3. Diagnosis of bacterial infection in training group At 0 hour, T, HR, RR, BP, and WBC did not significantly differ between septic and nonseptic patients in the training group. However, the levels of IL-6, PCT, and N% were much higher in septic patients than in nonseptic ones (Table 2). The AUC of IL-6 was largest for diagnosing sepsis, followed by those of N%, PCT, and WBC (Fig. 2). The threshold of IL-6 that diagnosed bacterial sepsis at the beginning of infection was 135 pg/mL, with a sensitivity of 94.8% and a specificity of 93.7%. The diagnostic thresholds of PCT, N%, and WBC were 0.39 ng/mL, 4.65%, and 3.96 × 103/mm3, respectively (Table 3). (See Fig. 1.)

Table 1 Characteristics of patients on baseline Total (N = 233)

Age (y) Male (%) Underlying disease HBV HCV AIH or PBC Wilson disease Cryptogenic Child-Pugh score 5-6 (stage A) 7-9 (stage B) 10-12 (stage C) Reasons for hospitalization Gastrointestinal bleeding Hepatic encephalopathy Hepatorenal syndrome Massive ascites Deterioration of liver function Concomitant diseases DM Renal failure

Sepsis (n = 134)

Nonsepsis (n = 99)

56 (46-65) 104 (77.6%)

55 (47-63) 79 (80.0%)

110 (82.1%) 13 (9.7%) 4 (3.0%) 4 (3.0%) 3 (2.2%)

Training group (n = 159) P⁎

Validation group (n = 74) Sepsis (n = 38)

Nonsepsis (n = 36)

P‡

.726 .976

56 (47-69) 28 (73.7%)

56 (46-62) 29 (80.6%)

.491 .828

.726 .498

.465 .887

56 (88.9%) 3 (4.8%) 1 (1.6%) 0 3 (4.8%)

.366 .366 .822 .153 .302

30 (78.9%) 4 (10.5%) 2 (5.3%) 0 2 (5.3%)

32 (88.9%) 0 2 (5.6%) 0 2 (5.6%)

.557 .652 .891 – .891

.619 .839 .318 .577 .194

1.000 .552 .299 – .836

15 (15.6%) 42 (43.8%) 39 (40.6%)

13 (20.6%) 34 (55.6%) 16 (25.4%)

.524 .256 .061

10 (26.3%) 13 (34.2%) 15 (39.4%)

13 (36.1%) 13 (36.1%) 10 (27.8%)

.861 .107 .140

.587 .337 .903

.103 .098 .816

.047 .646 .389 .791 .064

10 (10.4%) 2 (2.1%) 1 (1.0%) 48 (50.0%) 35 (36.5%)

14 (22.2%) 1 (1.6%) 0 30 (47.6%) 18 (28.6%)

.068 .822 .416 .871 .390

6 (15.8%) 2 (5.3%) 0 19 (50.0%) 11 (28.9%)

8 (22.2%) 1 (2.8%) 0 21 (58.3%) 6 (16.7%)

.554 .637 – .646 .297

.388 .318 .528 1.000 .807

1.000 .685 – .970 .228

.734 .389

17 (17.7%) 1 (1.0%)

13 (20.6%) 0

.682 .416

9 (23.7%) 0

4 (11.1%) 0

.765 –

.430 .528

.277 –

Sepsis (n = 96)

Nonsepsis (n = 63)

.929 .799

56 (47-66) 76 (79.2%)

56 (45-64) 50 (79.4%)

88 (88.9%) 3 (3.0%) 3 (3.0%) 0 5 (5.1%)

.151 .065 .984 .083 .244

80 (83.3%) 9 (9.4%) 2 (2.1%) 4 (4.2%) 1 (1.0%)

25 (18.7%) 55 (41.0%) 54 (40.3%)

26 (22.3%) 47 (47.5%) 26 (26.3%)

.165 .506 .027

16 (11.9%) 4 (3.0%) 1 (0.7%) 67 (50.0%) 46 (34.3%)

22 (22.2%) 2 (2.0%) 0 51 (51.5%) 24 (24.2%)

26 (19.4%) 1 (0.7%)

17 (17.2%) 0

P†

P§

P

⁎ P sepsis vs nonsepsis in all enrolled patients. † P sepsis vs nonsepsis in training group. ‡ P sepsis vs nonsepsis in validation group. § P training vs validation group of sepsis patients. P training vs validation group of nonsepsis patients.

Please cite this article as: Lin S, et al, Interleukin-6 as an early diagnostic marker for bacterial sepsis in patients with liver cirrhosis, J Crit Care (2015), http://dx.doi.org/10.1016/j.jcrc.2015.03.031

4

S. Lin et al. / Journal of Critical Care xxx (2015) xxx–xxx

Table 2 Vital signs and laboratory parameters of nonsepsis and sepsis groups in training patients Nonsepsis group (n = 63)

BP (mm Hg) T (°C) HR (beat/min) R (/min) WBC (×103/mm3) N (%) PCT (ng/mL) IL-6 (pg/mL)

Sepsis group (n = 96)

Baseline

Onset of fever

P⁎

Baseline

Onset of fever

P†

P‡

P§

118/72 (110-132/67-82) 36.5 (36.3-36.8) 78 (71-80) 19 (18-21) 3.85 (2.78-6.21) 61.65 (55.30–72.80) 0.18 (0.08-0.30) 18.83 (11.92-29.41)

112/68 (104-120/62-74) 37.9 (37.4-38.5) 92 (84-100) 21.0 (20-22) 5.92 (3.46-9.30) 66.60 (57.20-74.30) 0.18 (0.10-0.38) 36.92 (15.39-76.09)

.014 .000 .000 .000 .000 .161 .070 .000

120/70 (110-134/63-78) 36.5 (36.3-36.9) 78 (72-84) 19 (18-20) 5.80 (3.89-8.02) 67.87 (57.40-75.00) 0.21 (0.08-0.68) 24.73 (11.74-46.31)

118/70 (108-123/63-78) 38.8 (38.2-39.7) 98 (84-111) 20 (19-22) 6.95 (4.90-10.95) 82.81 (75.40-88.40) 0.96 (0.31-3.31) 538.55 (271.52-1378.00)

.035 .000 .000 .000 .000 .000 .000 .000

.325 .716 .422 .093 .003 .070 .313 .259

.158 .000 .093 .132 .021 .000 .000 .000

⁎ P baseline vs onset of fever in nonsepsis group. † P baseline vs onset of fever in sepsis group. ‡ P nonsepsis group vs sepsis group on baseline. § P nonsepsis group vs sepsis group on the onset of fever.

3.4. Cost-effectiveness analysis

3.5. Changes in WBC, N%, and levels of PCT and IL-6 over time in sepsis group

Table 3 demonstrates the cost-effectiveness analysis of the 4 biomarkers. The cost of measuring PCT was the highest among these tests. The WBC was the most cost-effective biomarker, followed by N%; however, these 2 markers were least specific. Although the costeffectiveness ratio of IL-6 was approximately 50% higher than those of WBC and N%, the AUC and specificity of IL-6 were the highest among these markers.

Fig. 3 shows the changes in the serum markers over time. The IL-6 level and N% peaked at the onset of fever, 24 hours before the changes in the PCT levels and WBC. However, the levels of PCT reached a maximum 24 hours after the onset of fever. Although the level of IL-6 remained higher at 24 hours, it reduced by more than 50% in 39 (40.6%) patients. The PCT (9.85 [1.94-27.95] ng/mL) levels peaked at 24 hours (Table 4).

1653 patients were diagnosed as cirrhosis

623 patients met the following exclusion criteria: 249 patients had signs of infection on admission 139 had signs of infection within 72 h after admission 235 had alcoholic cirrhosis

1030 patients were eligible

771 patients did not develop fever during hospitalization 100 patients had infection but not sepsis

159 patients were enrolled

63 patients had fever and met SIRS criteria but sepsis was finally excluded (Control group)

96 patients were enrolled finally (Sepsis group)

85 patients survived

11 patients died

Fig. 1. Flowchart of study design showing patient allocation in training group.

Please cite this article as: Lin S, et al, Interleukin-6 as an early diagnostic marker for bacterial sepsis in patients with liver cirrhosis, J Crit Care (2015), http://dx.doi.org/10.1016/j.jcrc.2015.03.031

S. Lin et al. / Journal of Critical Care xxx (2015) xxx–xxx

5

A: ROC curve for early diagnosis of bacteria sepsis.t

B: ROC curve for predicting the prognosis.

Fig. 2. IL-6, PCT levels, WBC, and N% upon the onset of fever for early diagnosis and prognosis of bacterial sepsis in cirrhotic patients. A, ROC curve for early diagnosis of bacterial sepsis. B, ROC curve for predicting the prognosis. AUC of IL-6 for diagnosing sepsis at the onset of fever was largest (AUC, 0.983; 95% CI, 0.967-0.999), followed by those of N% (AUC, 0.844; 95% CI, 0.783-0.905), PCT (AUC, 0.803; 95% CI, 0.735-0.872), and WBC (AUC, 0.603; 95% CI, 0.512-0.694).

Table 3 The sensitivity, specificity, and cost-effectiveness analysis of WBC, N%, PCT, and IL-6 Parameters 3

3

WBC (≥3.96 × 10 /mm ) N% (≥74.65%) IL-6 (≥135 pg/mL) PCT (≥0.39 ng/mL)

AUC

95%CI

Sensitivity/effectiveness (E) (%)

Specificity (%)

Cost (C) ($)

C/E

0.603 0.844 0.983 0.803

0.512-0.694 0.783-0.905 0.967-0.999 0.735-0.872

83.5% 78.0% 94.8% 72.5%

42.9% 77.8% 93.7% 77.8%

4.80 4.80 8.60 38.50

5.74 6.15 9.07 53.10

Fig. 3. Changes in WBC, N%, and PCT and IL-6 levels over time in sepsis group. IL-6 level and N% peaked at the onset of fever, 24 hours before the PCT levels and WBC, and then declined sharply. However, the peak PCT levels and WBC occurred 24 hours after the first fever.

3.6. Comparison between survival group and nonsurvival group in septic patients of training group After 4 weeks of follow-up, 85 (88.5%, 70 men, 55 years old) patients survived and 11 patients (11.5%, 7 men, 65 years old) died. Gender did not significantly differ between the survival group and nonsurvival group at

baseline (P N .05). The Child-Pugh score and age of the nonsurvival group were higher than those of the survival group (Table 5). At 0 hour, T, BP, WBC, N%, and PCT levels did not significantly differ between the 2 groups (P N .05). Nevertheless, the initial level of IL-6 on the onset of sepsis was significantly higher in the nonsurvival group than in the survival group (1813.00 [1119.00-2943.00] pg/mL

Table 4 Changes of vital signs and laboratory parameters over time in sepsis patients of training group

BP (mm Hg) T (°C) HR (beat/min) R (/min) WBC (×103/mm3) N (%) PCT (ng/mL) IL-6 (pg/mL)

Baseline

0h

24 h

48 h

120/70 (110-134/63-78) 36.5 (36.3-36.9) 78 (72-84) 19 (18-20) 5.80 (3.89-8.02) 67.87 (57.40-75.00) 0.21 (0.08-0.68) 24.73 (11.74-46.31)

118/70 (108-123/63-78) 38.8(38.2-39.7) 98(84-111) 20 (19-22) 6.95 (4.90-10.95) 82.81(75.40-88.40) 0.96 (0.31-3.31) 538.55 (271.52-1378.00)

114/70 (101-121/63-76) 37.6 (36.9-38.5) 84 (80-96) 20 (19-20) 10.04 (6.29-11.85) 79.80 (73.60-89.40) 9.85(1.94-27.95) 218.85 (97.47-872.25)

112/70 (104-121/65-76) 37.0 (36.5-37.8) 83 (76-88) 19 (18-20) 5.87 (4.31-9.24) 74.13 (66.20-82.70) 1.37 (0.52-5.51) 68.86 (29.03-159.05)

Please cite this article as: Lin S, et al, Interleukin-6 as an early diagnostic marker for bacterial sepsis in patients with liver cirrhosis, J Crit Care (2015), http://dx.doi.org/10.1016/j.jcrc.2015.03.031

6

S. Lin et al. / Journal of Critical Care xxx (2015) xxx–xxx

Table 5 Clinical symptoms and laboratory parameters of survival group and nonsurvival group Survival group (n = 85) Age Male (%) Child-Pugh score SIRs at baseline n (%) HR at 0 h (beat/min) R at 0 h (/min) BP at 0 h (mm Hg) Temperature at 0 h (°C) WBC at 0 h (×103/mm3) N (%) at 0 h PCT (ng/mL) Baseline 0h 24 h 48 h IL-6 (pg/mL) Baseline 0h 24 h 48 h

Nonsurvival group (n = 11)

55 (46-64) 70 (82.35%) 9 (7-10) 18 (21.20%) 94(84-107) 21 (20-22) 116/70 (107-123/63-78) 38.7 (38.2-39.4) 7.07(4.91-10.80) 81.80 (74.80-87.70)

65 (59-67) 7 (63.64%) 11 (9-13) 5 (45.50%) 92 (83-120) 20 (20-23) 124/72 (109-141/63-86) 38.2 (37.7-39.3) 6.35 (2.97-23.68) 87.70 (79.40-92.60)

0.23 (0.09-0.69) 0.79 (0.30-2.61) 5.93 (1.08-22.21) 1.37 (0.50-5.35)

0.09 (0.08-0.59) 3.09 (1.49-5.01) 25.00 (10.00-55.00) 59.29 (29.58-89.00)

25.97 (12.30-45.72) 472.10(240.00-1091.50) 146.60 (63.65-286.40) 67.18 (28.49-138.80)

vs 472.10 [240.00-1091.50] pg/mL, P = .004); and this difference persisted until the 24th hour. The same trend was evident for the PCT levels (Table 5). At 48 hours, the serum levels of IL-6 and PCT remained higher in the nonsurvival group than in the survival group; and this difference was significant (Table 5).

3.7. Predictive value of IL-6 The serum levels of IL-6 at 0 hour significantly differed between the survival group and nonsurvival group (Table 5 and Fig. 2). The AUCs of PCT and N% were 0.692 and 0.691, respectively, for predicting death. The AUC of IL-6 (0.763) was larger than those of the other markers. The cutoff value for predicting prognosis was 1105 pg/mL, with a sensitivity of 81.8% and a specificity of 76.5%. The cutoff value of PCT was 1.45 ng/mL, with a sensitivity and specificity of 81.8% and 63.7%, respectively.

3.8. Validation of IL-6 for diagnosis of sepsis and predictive Within 7 months (from 1 January 2014 to 30 June 2014), 74 patients were available to validate the diagnostic value of IL-6 for sepsis in cirrhosis. Similarly, 38 (51.4%) patients suffered sepsis 72 hours after admission; and 36 (48.6%) patients developed a fever without sepsis. The patients’ characteristics at baseline are shown in Table 1. The baseline diseases (HBV- or HCV-associated cirrhosis, PBC or AIH, Wilson disease, cryptogenic cirrhosis), vital signs (T, BP, HR, R), and laboratory parameters (WBC, N%, PCT, IL-6 levels) did not significantly differ between the training group and validation group. As outlined above, an IL-6 value of 135 pg/mL could accurately diagnose sepsis, with a sensitivity of 97.5% and specificity of 80.6%. Furthermore, the positive predictive value (PPV) and negative predictive value (NPV) were 77.0% and 90.6%, respectively. The values of PCT (0.39 ng/mL) were lower for the diagnosis of sepsis, with a sensitivity of 47.5% and specificity of 90%; and the PPV and NPV were 82.6% and 63.2%, respectively. Among the 38 sepsis patients, 5 (13.2%) patients died during the study. The sensitivity and specificity were 60% and 72.7%, respectively, using a cutoff value of 1105 pg/mL for predicting prognosis; and the PPV was 27.3%, whereas the NPV was 93.1%. Using the PCT value (1.45 ng/mL) for prognosis, the sensitivity was 60.0% and the specificity was 34.3%; the PPV and NPV were 20.0% and 92.0%, respectively.

14.10 (12.06-31.42) 1813.00 (1119.00-2943.00) 1340.00(500.00-1980.00) 1210.05(630.10-1790.00)

P value .045 .143 .009 .086 .638 .264 .348 .462 .983 .039 .390 .034 .013 .006 .741 .004 .000 .013

4. Discussion In this cohort study, we focused on the early diagnostic and predictive values of the serum IL-6 level in patients with bacterial sepsis. The serum IL-6 increased and decreased earlier than the levels of other infectious parameters, which was consistent with previous reports [12,24]. The level of serum IL-6 peaked upon the onset of fever, whereas those of PCT peaked 24 hours later. The IL-6 level subsequently significantly decreased after 24 hours, but the PCT levels had not yet reached a maximum at this point. Procalcitonin, WBC, and N% are the most widely used diagnostic biomarkers for infections. In the present study of cirrhotic patients with sepsis, the median level of PCT was only 0.96 ng/mL at the onset of sepsis; and the diagnostic threshold of PCT in this cohort was only 0.39 ng/mL. Previous studies have demonstrated that the best cutoff value of PCT for predicting sepsis was 2.415 ng/mL in critically ill patients in an intensive care unit [25]. According to this documented diagnostic threshold, only 28 (29.2%) patients in our cohort could be correctly diagnosed at the onset of bacterial sepsis with a PCT level of 2.415 ng/mL. Furthermore, a cost-effectiveness analysis revealed that the cost of measuring PCT was the highest among the tests for the 4 biomarkers examined in this study. Procalcitonin is unlikely to be useful as a biomarker for the early diagnosis of sepsis in cirrhotic patients. The serum IL-6 levels are associated with inflammation in the human body. The IL-6 levels have been confirmed to rapidly increase after exposure to bacterial components, and this increase occurred much earlier than that of PCT [12,26]. However, the threshold of IL-6 for the diagnosis of bacterial infection in cirrhotic patients ranged from 80 to 200 pg/mL [20,21,27]. This difference may be due to the small sample sizes of studies. In this study, an IL-6 level greater than or equal to 135 pg/mL corresponded to a sensitivity of 94.5% and specificity of 93.7% for the rapid diagnosis of bacterial sepsis, much higher than the sensitivity and specificity of PCT. This result was also confirmed in a validation cohort. Although the cost-effectiveness ratio of IL-6 was not the lowest (approximately 50% higher than those of WBC and N%), the AUC and specificity of IL-6 were the highest. Furthermore, the diagnostic thresholds of WBC and N% (3.96 × 103/mm3 for WBC and 74.65% for N%) were both within reference ranges, which may complicate decision making. However, the diagnostic threshold of IL-6 (135 pg/mL) was almost 20 times higher than the upper limit of the reference range (0-7.0 pg/mL). This difference may facilitate the diagnosis of sepsis. Therefore, IL-6 could be a useful marker for the early diagnosis of

Please cite this article as: Lin S, et al, Interleukin-6 as an early diagnostic marker for bacterial sepsis in patients with liver cirrhosis, J Crit Care (2015), http://dx.doi.org/10.1016/j.jcrc.2015.03.031

S. Lin et al. / Journal of Critical Care xxx (2015) xxx–xxx

sepsis in cirrhotic patients. Because the timely diagnosis and treatment of sepsis are crucial, developing a rapid detection method, such as an IL6 reagent strip, for rapid bedside screening is critical in clinical practice. The IL-6 level has been demonstrated to correlate with the mortality rate [28]. The blood IL-6 level during the intensive care unit stay was significantly higher in nonsurvivors than in survivors [29]. Interleukin-6 may be responsible for the renal impairment and mortality of spontaneous bacterial peritonitis patients [30]. The serum IL-6 level has also been shown to correlate with organ failure in cirrhotic patients [31]. Moreover, an IL-6 level greater than 300 pg/mL was associated with significantly increased risks of complications (pneumonia, multiple organ failure, death) [32]. However, the predictive value and threshold of IL6 in cirrhotic patients with sepsis remain unknown. In this study of cirrhotic patients with sepsis, the IL-6 level was significantly higher in the nonsurvival group than in the survival group (1813.00 vs 472.10 pg/mL, P = .004). An IL-6 level greater than 1105 pg/mL at the very onset of infection, but not the PCT levels or N%, could predict the outcome at 4 weeks. Interleukin-6 is a proinflammatory cytokine produced during the early phase of inflammation [33], and its excessive production can induce SIRS. An increased IL-6 level indicates deteriorated SIRS [34], which is independently associated with in-hospital death in cirrhotic patients [35]. Therefore, the IL-6 level during the initial stage of bacterial sepsis can predict the deaths of patients with cirrhosis. In this research, only patients who developed high temperature during hospitalization were selected to obtain the data at the onset of sepsis. Therefore, many septic patients who did not have a high fever were excluded. The early diagnosis of infections in patients without apparent clinical symptoms remains difficult. High IL-6 levels may be an alarming sign of severe sepsis. However, a single measurement is not reliable for diagnosing or ruling out bacterial infection, sepsis, or severe sepsis; but these parameters can be helpful when used in conjunction with the patient’s clinical data. Furthermore, relative adrenal insufficiency (RAI) has been shown to affect 26% of patients with decompensated liver cirrhosis and is associated with severe sepsis and overall mortality [36]. The plasma IL-6 level is increased in patients with RAI [36]; therefore, RAI could be a potential confounding factor in this study. We did not examine adrenal function because short corticotropin stimulation tests cannot be performed in a retrospective cohort. Well-designed prospective studies with larger cohorts of patients are needed to enrich the data and verify the conclusion. 5. Conclusion In conclusion, IL-6 appears to be superior to PCT in diagnosing and predicting sepsis in cirrhotic patients at the onset of fever. Our observations encourage the use of IL-6 as an early marker of bacterial sepsis. Acknowledgments This study was financially supported by the Health Science and Technology Development Fund of Nanjing (QYK11173) and Tianqing Liver Disease Research Fund (TQGB20140087). References [1] Fernandez J, Navasa M, Gomez J, et al. Bacterial infections in cirrhosis: epidemiological changes with invasive procedures and norfloxacin prophylaxis. Hepatology 2002;35(1):140–8. [2] Arvaniti V, D'Amico G, Fede G, et al. Infections in patients with cirrhosis increase mortality four-fold and should be used in determining prognosis. Gastroenterology 2010;139(4):1246–1256.e1245. [3] Plessier A, Denninger MH, Consigny Y, et al. Coagulation disorders in patients with cirrhosis and severe sepsis. Liver Int 2003;23(6):440–8. [4] Foreman MG, Mannino DM, Moss M. Cirrhosis as a risk factor for sepsis and death: analysis of the National Hospital Discharge Survey. Chest 2003;124(3):1016–20.

7

[5] Dellinger RP, Levy MM, Carlet JM, et al. Surviving Sepsis Campaign: international guidelines for management of severe sepsis and septic shock: 2008. Crit Care Med 2008;36(1):296–327. [6] Kumar A, Zarychanski R, Light B, et al. Early combination antibiotic therapy yields improved survival compared with monotherapy in septic shock: a propensitymatched analysis. Crit Care Med 2010;38(9):1773–85. [7] Kumar A, Roberts D, Wood KE, et al. Duration of hypotension before initiation of effective antimicrobial therapy is the critical determinant of survival in human septic shock. Crit Care Med 2006;34(6):1589–96. [8] Pfafflin A, Schleicher E. Inflammation markers in point-of-care testing (POCT). Anal Bioanal Chem 2009;393(5):1473–80. [9] Li CH, Yang RB, Pang JH, et al. Procalcitonin as a biomarker for bacterial infections in patients with liver cirrhosis in the emergency department. Acad Emerg Med 2011; 18(2):121–6. [10] Park WB, Lee K-D, Lee CS, et al. Production of C-reactive protein in Escherichia coliinfected patients with liver dysfunction due to liver cirrhosis. Diagn Microbiol Infect Dis 2005;51(4):227–30. [11] Fernandez J, Gustot T. Management of bacterial infections in cirrhosis. J Hepatol 2012;56(Suppl. 1):S1–S12. [12] Dahaba AA, Metzler H. Procalcitonin's role in the sepsis cascade. Is procalcitonin a sepsis marker or mediator? Minerva Anestesiol 2009;75(7–8):447–52. [13] Meisner M, Müller V, Khakpour Z, et al. Induction of procalcitonin and proinflammatory cytokines in an anhepatic baboon endotoxin shock model. Shock 2003;19(2):187–90. [14] Castell JV, Gomez-Lechon MJ, David M, et al. Acute-phase response of human hepatocytes: regulation of acute-phase protein synthesis by interleukin-6. Hepatology 1990;12(5):1179–86. [15] van Rossum AM, Wulkan R, Oudesluys-Murphy A. Procalcitonin as an early marker of infection in neonates and children. Lancet Infect Dis 2004;4(10):620–30. [16] Müller B, Becker KL, Schächinger H, et al. Calcitonin precursors are reliable markers of sepsis in a medical intensive care unit. Crit Care Med 2000;28(4):977–83. [17] Uusitalo-Seppala R, Koskinen P, Leino A, et al. Early detection of severe sepsis in the emergency room: diagnostic value of plasma C-reactive protein, procalcitonin, and interleukin-6. Scand J Infect Dis 2011;43(11–12):883–90. [18] Steinberger E, Hofer N, Resch B. Cord blood procalcitonin and Interleukin-6 are highly sensitive and specific in the prediction of early-onset sepsis in preterm infants. Scand J Clin Lab Invest 2014;74(5):432–6. [19] Haasper C, Kalmbach M, Dikos GD, et al. Prognostic value of procalcitonin (PCT) and/ or interleukin-6 (IL-6) plasma levels after multiple trauma for the development of multi organ dysfunction syndrome (MODS) or sepsis. Technol Health Care 2010; 18(2):89–100. [20] Byl B, Roucloux I, Crusiaux A, et al. Tumor necrosis factor alpha and interleukin 6 plasma levels in infected cirrhotic patients. Gastroenterology 1993;104(5):1492–7. [21] Wang SS, Lee FY, Chan CC, et al. Sequential changes in plasma cytokine and endotoxin levels in cirrhotic patients with bacterial infection. Clin Sci (Lond) 2000;98(4):419–25. [22] Fujimoto M, Uemura M, Nakatani Y, et al. Plasma endotoxin and serum cytokine levels in patients with alcoholic hepatitis: relation to severity of liver disturbance. Alcohol Clin Exp Res 2000;24(4 Suppl.):48S–54S. [23] Bone RC, Balk RA, Cerra FB, et al. Definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. The ACCP/SCCM Consensus Conference Committee. American College of Chest Physicians/Society of Critical Care Medicine. Chest 1992;101(6):1644–55. [24] Buck C, Bundschu J, Gallati H, et al. Interleukin-6: a sensitive parameter for the early diagnosis of neonatal bacterial infection. Pediatrics 1994;93(1):54–8. [25] BalcI C, Sungurtekin H, Gürses E, et al. Usefulness of procalcitonin for diagnosis of sepsis in the intensive care unit. Crit Care 2002;7(1):85–9. [26] DaSilva O, Hammerberg O. Diagnostic value of leukocyte indices in late neonatal sepsis. Pediatr Infect Dis J 1994;13(5):409–11. [27] Le Moine O, Deviere J, Devaster JM, et al. Interleukin-6: an early marker of bacterial infection in decompensated cirrhosis. J Hepatol 1994;20(6):819–24. [28] Wunder C, Eichelbronner O, Roewer N. Are IL-6, IL-10 and PCT plasma concentrations reliable for outcome prediction in severe sepsis? A comparison with APACHE III and SAPS II. Inflamm Res 2004;53(4):158–63. [29] Oda S, Hirasawa H, Shiga H, et al. Sequential measurement of IL-6 blood levels in patients with systemic inflammatory response syndrome (SIRS)/sepsis. Cytokine 2005; 29(4):169–75. [30] Navasa M, Follo A, Filella X, et al. Tumor necrosis factor and interleukin-6 in spontaneous bacterial peritonitis in cirrhosis: relationship with the development of renal impairment and mortality. Hepatology 1998;27(5):1227–32. [31] Rosenbloom AJ, Pinsky MR, Bryant JL, et al. Leukocyte activation in the peripheral blood of patients with cirrhosis of the liver and SIRS. Correlation with serum interleukin-6 levels and organ dysfunction. JAMA 1995;274(1):58–65. [32] Giannoudis PV, Harwood PJ, Loughenbury P, et al. Correlation between IL-6 levels and the systemic inflammatory response score: can an IL-6 cutoff predict a SIRS state? J Trauma Inj Infect Crit Care 2008;65(3):646–52. [33] Panacek EA, Kaul M. IL-6 as a marker of excessive TNF-α activity in sepsis. Sepsis 1999;3(1):65–73. [34] Guirao X, Lowry SF. Biologic control of injury and inflammation: much more than too little or too late. World J Surg 1996;20(4):437–46. [35] Cazzaniga M, Dionigi E, Gobbo G, et al. The systemic inflammatory response syndrome in cirrhotic patients: relationship with their in-hospital outcome. J Hepatol 2009;51(3):475–82. [36] Acevedo J, Fernández J, Prado V, et al. Relative adrenal insufficiency in decompensated cirrhosis: relationship to short‐term risk of severe sepsis, hepatorenal syndrome, and death. Hepatology 2013;58(5):1757–65.

Please cite this article as: Lin S, et al, Interleukin-6 as an early diagnostic marker for bacterial sepsis in patients with liver cirrhosis, J Crit Care (2015), http://dx.doi.org/10.1016/j.jcrc.2015.03.031