PDF
Citation
Print
INTRODUCTION
Patients with liver cirrhosis tend to suffer from bacteremia due to the hepatic reticuloendothelial system and neutrophil hypoactivity, coupled with an influx of bacteria into the general circulation via portocaval shunts.123 It is thought that 30%–60% of patients with liver cirrhosis develop a bacterial infection,345 and such infection is involved in one-quarter of deaths among these patients.67
Spontaneous bacterial peritonitis (SBP) is the most common bacterial infection among liver cirrhosis patients, accounting for at least 24% of infections in that population.89 Although there is a considerable amount of information available regarding the pathogenesis, diagnosis, and treatment of SBP,810 an up-to-date, proven, and easy-to-use clinical factor is not yet available to physicians for determining the in-hospital mortality associated with SBP. The ability to predict prognosis is important with respect to identifying high-risk patients toward whom more intensive monitoring and treatment strategies can be targeted. The predictive model needs to be simple so that it can be easily applied in the real-life, clinical setting. In an attempt to identify independent prognostic factors of in-hospital mortality, a meta-analysis demonstrated that renal dysfunction is a significant independent predictor of prognosis in cirrhotic patients.11 Unfortunately, that study was limited by no estimation of the optimal cutoff point being reported for the high-risk mortality group.
Early detection of high-risk patients is essential for prognostic improvement. Therefore, finding prognostic factors is important for risk stratification, physician decision making, and individualized medical therapy for different at-risk populations. The establishment of prognostic factors that are modifiable enables early treatment strategies to be developed. Accordingly, the aim of this study was to identify the prevalence and independent predictors of in-hospital mortality in a cohort of liver cirrhosis patients with SBP receiving antibiotics and, if deemed necessary by guidelines, intravenous albumin and terlipressin therapy. A prognostic factor that is easy to implement in real-life, clinical settings was investigated.
METHODS
Patients and procedures
Four hundred and ninety-nine SBP cases occurring between August 2002 and February 2013 were identified. These SBP patients were diagnosed in four tertiary referral hospitals (Soonchunhyang University Seoul Hospital, Bucheon Hospital, Cheonan Hospital, and Gangneng Asan Medical Center), either on hospital admission or during hospitalization due to liver cirrhosis.
The inclusion criteria were as follows: 1) adult patients who were diagnosed with liver cirrhosis based on their clinical presentation, laboratory markers (liver function parameters), and compatible radiology findings (nodular contour to the liver parenchyma and evidence of portal hypertension); 2) diagnosis of SBP based on the presence of an ascitic fluid polymorphonuclear neutrophil (PMN) cell count greater than 250 cells/mm3 in the absence of any other source of intra-abdominal infection; and 3) diagnostic paracentesis within 24 hours whenever SBP was suspected. Patients were excluded if they had infectious pleural effusion, peritonitis carcinomatosa, malignant ascites, hemorrhagic ascites, septic shock, treatment before ascitic analysis, end-stage renal disease with or without hemodialysis, or previous liver transplantation. Of the 499 patients initially enrolled with suspicious SBP, 41 patients were excluded as they were in a chronic renal disease state. Another 156 patients were excluded because their ascitic PMN cell count was under 250 cells/mm3, and 13 were excluded as they were in septic shock with arterial hypotension (systolic blood pressure < 90 mmHg) or arterial hypoxemia (ratio of the partial pressure of arterial oxygen to the fraction of inspired oxygen, < 300), as evidenced by the diagnostic criteria for septic shock.12 Fourteen patients were excluded because ascitic fluid analysis was not performed before treatment with cefotaxime because of their scanty ascite, and 42 patients were excluded because serial serum creatinine measurements were not available. Ultimately, 233 patients met all of the inclusion criteria (Fig. 1).
All the patients were treated with antibiotics. Additionally they were treated with albumin (1 g/kg/day) to prevent hepatorenal syndrome or to supply albumin (1 g/kg/day) if they had a serum albumin level of < 3.0 g/dL based on the guidelines of the Korean National Health Insurance Service.1314 All patients with a serum creatinine level of ≥ 2.5 mg/dL were treated with albumin and terlipressin according to the guidelines of the Korean National Health Insurance Service, which were established using the Clinical Practice Guidelines for Liver Cirrhosis, Update, and recommendations from previous studies.15
The patients with SBP were divided into non-survivors and survivors in hospital. Information was collected on sex, age, etiology of the cirrhosis, and presence of hepatic encephalopathy. Laboratory variables were obtained at the time of diagnosis, including via serum analysis, via ascitic analysis, and the glomerular filtration rate (GFR) estimated using the Modification of Diet in Renal Disease formula1617:
GFR = 170 × [Pcr]–0.999 × [Age]–0.176 × [0.762 if patient is female] × [1.180 if patient is black] × [BUN]–0.170 × [Alb]+0.318
where Pcr is plasma creatinine, BUN is blood urea nitrogen, and Alb is albumin. These data were considered baseline values. Measurement of serum creatinine levels is usually routine upon a diagnosis of SBP and is repeated frequently during the treatment in SBP patients. Therefore, all 233 patients with SBP underwent serial serum creatinine measurement (Fig. 1). The Child-Turcotte-Pugh (CTP) and model for end-stage liver disease sodium model (MELD-Na) scores, and serum ascites albumin gradient were calculated using the laboratory data at SBP diagnosis. The prognostic factors related to in-hospital mortality were compared between the two groups (i.e., non-survivors and survivors).
Statistical analysis
The patients' characteristics are described using mean and standard deviation values or proportions. Kaplan-Meier and Cox proportional hazards regression analyses were used to determine independent predictors of in-hospital mortality. The level of statistical significance was set at P < 0.05. Variables that have previously been demonstrated to be clinically relevant and/or variables with P < 0.10 on univariate analysis were entered into the multivariate model. The results of the binary logistic regression are reported as the odds ratio (OR) and 95% confidence interval (CI). A receiver operating characteristic (ROC) curve was generated for the variables that were statistically significant on multivariate analysis to assess the predictive accuracy of these models and to determine the optimal cutoff points for the prediction of in-hospital mortality. Area under the curve (AUC) analysis was conducted to determine the discriminatory ability of the data. All data were analyzed using the Statistical Package for the Social Sciences (version 17.0, SPSS, Chicago, IL, USA).
RESULTS
Characteristics of the patients
In total, 233 patients (181 men and 52 women) were included in the study. The demographic, clinical, and laboratory characteristics of the patients are listed in Table 1. The mean age of the patients was 57.1 years. The causes of liver cirrhosis were hepatitis B (44.6%), alcohol (43.8%), hepatitis C (6.0%), and cryptogenic cirrhosis (5.6%). The mean MELD-Na and CTP scores were 27.1 and 10.7, respectively. Fifty-two patients were treated with albumin and terlipressin. Thirty-eight of 52 patients were treated with albumin and terlipressin due to type 1 hepatorenal syndrome at diagnosis and 14 of 52 patients were treated with albumin and terlipressin during the therapeutic period. Eighteen patients (18/52, 34.6%) who were treated with albumin and terlipressin died in hospital with SBP. The in-hospital mortality rate was 14.6% (n = 31) and the mean hospitalization period of non-survivors was 8.3 days. Baseline MELD-Na score (P = 0.001), CTP score (P = 0.005), absolute neutrophil count (P = 0.009), international normalized ratio (P < 0.001), ammonia (P = 0.023), aspartate transaminase (P = 0.011), alanine aminotransferase (P = 0.001), GFR (P = 0.002), serum creatinine (P < 0.001), and maximal serum creatinine (P < 0.001) were different between non-survivors and survivors (Table 2).
Table 1
Baseline demographics of the patients with SBP (n = 233)
Values are presented as number of patients (%) or mean ± standard deviation.
SBP = spontaneous bacterial peritonitis, MELD = model for end-stage liver disease, MELD-Na = model for end-stage liver disease sodium model, CTP = Child-Turcotte-Pugh, ANC = absolute neutrophil count, INR = international normalized ratio, CRP = c-reactive protein, GFR = glomerular filtration rate, SAAG = serum-ascites albumin gradient, PMN = polymorphonuclear neutrophil.
aEstimated using the Modification of Diet in Renal Disease formula.
Table 2
Baseline characteristics between survivors and non-survivors
Values are presented as number of patients (%) or mean ± standard deviation.
MELD = model for end-stage liver disease, MELD-Na = model for end-stage liver disease sodium model, CTP = Child-Turcotte-Pugh, ANC = absolute neutrophil count, INR = international normalized ratio, CRP = c-reactive protein, GFR = glomerular filtration rate, SAAG = serum-ascites albumin gradient, PMN = polymorphonuclear neutrophil.
aEstimated using the Modification of Diet in Renal Disease formula.
Microbiological characteristics and antibiotics
A profile of the organisms isolated is listed on Table 3. Of 233 patients, 29 patients had a positive acsitic fluid culture result: Escherichia coli was isolated in 19 patients and Klebsiella pneumoniae was isolated in 6 patients. Three cases of E. coli were resistant to third-generation cephalosporin. Two hundred and thirty-three patients received empirical antimicrobial therapy: 199 (85.4%) received third-generation cephalosporin, 24 (10.3%) received fluoroquinolone, 8 (3.4%) received carbapenem, and 2 (0.8%) received fourth-generation cephalosporin.
Prognostic factors related to in-hospital mortality during treatment
During treatment in hospital, maximum serum creatinine level (OR, 1.776; 95% CI, 1.400–2.254; P < 0.001), the absolute neutrophil count (OR, 1.000; 95% CI, 1.000–1.000; P = 0.015) and the MELD-Na score (OR, 1.128; 95% CI, 1.062–1.198; P < 0.001) were significant prognostic factors of in-hospital mortality on univariate analysis (Table 2). The CTP score, total bilirubin, serum albumin, international normalized ratio, ascitic PMN cell count, and hepatic encephalopathy were not statistically significant predictors of in-hospital mortality. Multivariate analysis revealed that the maximum serum creatinine during treatment (OR, 1.680; 95% CI, 1.313–2.151; P = 0.005) was an independent predictor of in-hospital mortality (Table 4). The follow-up serum creatinine level was significantly higher in non-survivors than survivors (3.6 vs. 1.7 mg/dL).
Table 4
Predictors of in-hospital mortality during treatment
OR = odds ratio, CI = confidence interval, MELD = model for end-stage liver disease, MELD-Na = model for end-stage liver disease sodium model, CTP = Child-Turcotte-Pugh, ANC = absolute neutrophil count, INR = international normalized ratio, PMN = polymorphonuclear neutrophil.
aOR was generated by binary logistic regression.
Accuracy of prognostic factors related to in-hospital mortality
Multivariate analysis of the prognostic factors related to the in-hospital mortality of SBP patients during treatment revealed that the maximum serum creatinine during treatment was significantly associated with in-hospital mortality. ROC curves were constructed to determine the optimal cutoff points for in-hospital mortality for the remaining variable. A maximum serum creatinine level during treatment of ≥ 2.0 mg/dL had a sensitivity of 77.4%, a specificity of 74.3%, and an AUC of the ROC curve of 0.78 (Fig. 2). Thus, patients without the risk variable had an in-hospital mortality rate of 4.5% (7/157), while those with the risk variable had an in-hospital mortality rate of 31.6% (24/76; Table 5). Patients could therefore be divided into high- and low-risk groups according to the maximum serum creatinine cutoff of ≥ 2 mg/dL during treatment, with the survival rate being significantly higher for the low-risk group than for the high-risk group (P < 0.001; Fig. 3). The mean duration of time during which the level of creatinine rose maximally in patients with maximum serum creatinine level (≥ 2.0 mg/dL) was 4.03 (± 5.938) day
Fig. 2
ROC curve for determining cutoff point for maximal serum creatinine level. A maximal serum creatinine level during treatment of ≥ 2 mg/dL was associated with a sensitivity of 77.4%, a specifity of 74.3%, and an AUROC of 0.78 for determining in-hospital mortality. The solid line represents the ROC based on chance alone and has a c-statistic of 0.5.
ROC = receiver operating characteristic, AUROC = area under ROC curve.
Table 5
In-hospital mortality determined by the presence of high-risk variable
| Variables | In-hospital mortality | Proportion of the group in this category |
|---|---|---|
| Maximal serum creatinine ≥ 2 mg/dL | 31.6% (24/76) | 32.6% |
| Maximal serum creatinine < 2 mg/dL | 4.5% (7/157) | 67.4% |
Fig. 3
Mortality rate of cirrhotic patients with SBP. In-hospital mortality rate according to maximal serum creatinine level during treatment. The patients with a maximal serum creatinine level during treatment of ≥ 2 mg/dL showed higher in-hospital mortality rate than a maximal serum creatinine level during treatment of < 2 mg/dL.
SBP = spontaneous bacterial peritonitis.
DISCUSSION
The main findings of the current retrospective study are as follows: 1) renal dysfunction is a robust prognostic factor of patients with cirrhosis and SBP; 2) high serum creatinine levels during treatment are associated with in-hospital mortality; and 3) the optimal cutoff point of maximum serum creatinine during treatment is ≥ 2 mg/dL.
These results are very important because these are based on real-life data. A maximum serum creatinine level during treatment of ≥ 2 mg/dL was the most important predictor of in-hospital mortality in the present cohort. When considering all patients, the absence of this high-risk factor was associated with an in-hospital mortality rate of 4.5%, while its presence was associated with an in-hospital mortality rate of over 30% (Table 5).
Renal function has been recognized as an independent predictor of mortality in several clinical settings, including SBP.1118192021 The mechanism underlying the association between renal impairment and SBP relative to mortality is probably hemodynamic in nature. Patients with cirrhosis and ascites have a circulatory dysfunction that is characterized by high cardiac output, arteriolar vasodilatation, hypotension, decreased effective arterial blood volume, and homeostatic activation of the renin-angiotensin and sympathetic nervous systems.2223 Therefore, the high frequency and severity of renal dysfunction after the onset of SBP are probably due to the combination of the circulation failure already present as an effect of cirrhosis and circulation failure caused by infection.2425 This combined effect probably disturbs the compensatory action of renal vasodilators and leads to decreases in renal function that are represented by increased serum creatinine. The finding of the present study that renal impairment is associated with in-hospital mortality can be explained by this hypothesis.
This study was based on patients who had undergone serial serum creatinine measurements and who were classified according to their maximum serum creatinine during treatment. These data could make it easier to predict the in-hospital mortality of cirrhotic patients with SBP in real-life clinical settings. Terg et al.26 measured serum creatinine levels 48 hours after treatment in SBP patients, but did not determine the cutoff value, while Soylu et al.27 used baseline creatinine levels and determined an optimal cutoff value of > 1.1 mg/dL. In addition, a meta-analysis revealed that renal dysfunction defined as a creatinine level of > 1.5 mg/dL is the main prognostic factor of in-hospital mortality in patients with SBP.28 However, cirrhotic patients with SBP frequently have elevated serum creatinine levels. Therefore, the results of these previous studies have a low discriminatory ability related to the serum creatinine level. The point of difference of the present study is the sense of reality, since maximum serum creatinine levels during treatment reflect the patients' current renal function. Thus, the present results regarding the prediction of in-hospital mortality could be more useful than those published previously. The establishment of a maximum serum creatinine level during treatment of ≥ 2.0 mg/dL is valuable for selecting cirrhotic patients with SBP who are at high risk of in-hospital mortality. In addition, the determined cutoff value (2.0 mg/dL) for the maximum serum creatinine is lower than the diagnostic serum creatinine criterion for type 1 hepatorenal syndrome (2.5 mg/dL).131429 These findings support the implementation of a serum creatinine level of ≥ 2.0 mg/dL as a predictor of in-hospital mortality in liver cirrhosis patients with SBP. Therefore, physicians should be aware that SBP patients with a maximum serum creatinine level of ≥ 2 mg/dL during treatment have a high risk of in-hospital mortality.
Albumin was administered at 1 g/kg/day to all patients in order to prevent hepatorenal syndrome or when serum albumin was < 3.0 g/dL. Furthermore, albumin (1 g/kg/day) plus terlipressin (1 mg/6 hours) was used only when serum creatinine was ≥ 2.5 mg/dL, in line with the recommendations of the Korean National Health Insurance Service regarding the use of expensive drugs such as albumin and terlipressin. Despite many of the high-risk patients being treated with antibiotics and albumin therapy, SBP mortality rates remained high (over 30%). It is therefore particularly important for physicians to accurately identify high-risk patients at the time of presentation. This encourages vigilance with respect to monitoring and treating these patients and allows more aggressive treatments to be directed toward the high-risk group. The renal dysfunction, and especially the maximum serum creatinine during treatment, should be evaluated for high-risk liver cirrhosis patients with SBP. The results of this study provide physicians with information that should enable them to distinguish cirrhotic patients with SBP who are at high risk of death, thereby allowing them to be managed more intensively.
This study was subject to some limitations: it had a retrospective design and was not a randomized controlled trial. However, the findings reflect circumstances that are closer to real life than would be possible in a planned, randomized controlled trial. Given the still high mortality rates associated with SBP in patients with liver cirrhosis, more research is needed to optimize the therapeutic algorithms for the treatment of this condition. Nevertheless, in the real-life, clinical setting, the use of simple rather than complex prognostic factors, such as that revealed by our results, can aid the physician in identifying high-risk patients toward whom more intensive monitoring and treatment strategies can be targeted.
In conclusion, we suggest that when cirrhotic patients with SBP have a maximum serum creatinine level during treatment of ≥ 2 mg/dL, the physician should identify them as being at high risk of in-hospital mortality, and monitor them intensively to prevent fatal situations such as type 1 hepatorenal syndrome.
XML Download