|Year : 2016 | Volume
| Issue : 4 | Page : 282-287
|A quick screening model for symptomatic bacterascites in cirrhosis
Long-Chuan Zhu1, Long Xu2, Wen-Hua He3, Wei Wu4, Xuan Zhu3
1 Department of Digestion, The First Affiliated Hospital of Nanchang University; Department of Hepatology, Infectious Disease Hospital of Nanchang University, Jiangxi, China
2 Department of Hepatology, Infectious Disease Hospital of Nanchang University, Jiangxi, China
3 Department of Digestion, The First Affiliated Hospital of Nanchang University, Jiangxi, China
4 Department of Digestion, Children's Hospital of Jiangxi Province, Jiangxi, China
Click here for correspondence address and email
|Date of Submission||12-Oct-2015|
|Date of Acceptance||29-Mar-2016|
|Date of Web Publication||3-Aug-2016|
| Abstract|| |
Background: Diagnosis of spontaneous bacterial peritonitis in cirrhosis can be made when a patient has an ascites polymorphonuclear leukocyte count ≥250/mm3. However, symptomatic bacterascites, which is a variant of spontaneous bacterial peritonitis with signs of infection but an ascites polymorphonuclear leukocyte count <250/mm3, cannot be confirmed until the time-consuming ascites culture becomes positive. Currently, early indicators for symptomatic bacterascites remain undetermined. Aims: To develop a quick screening model for early detection of symptomatic bacterascites in cirrhosis. Materials and Methods: Data on patients with cirrhotic ascites from two hospitals (from 2010 to 2014) were collected retrospectively. Patients with symptomatic bacterascites were enrolled in the case group and compared with patients without any infection in the control group. Logistic regression analysis was used to build a model for screening symptomatic bacterascites, and a receiver operating characteristics curve was used to assess the model. Results: In total, 103 patients were enrolled in the case group and 204 patients were enrolled in the control group. A screening model was constructed based on body temperature, abdominal tenderness, blood neutrophil percentage, blood total bilirubin, prothrombin time, and ascites nucleated leukocyte count. The area under the receiver operating characteristic curve was 0.939; a screening score of 0.328 was the best cutoff value. Conclusion: Patients with suspected symptomatic bacterascites can be quickly screened according to the developed model, and a screening score ≥0.328 indicates symptomatic bacterascites.
Keywords: Cirrhosis, spontaneous bacterial peritonitis, symptomatic bacterascites
|How to cite this article:|
Zhu LC, Xu L, He WH, Wu W, Zhu X. A quick screening model for symptomatic bacterascites in cirrhosis. Saudi J Gastroenterol 2016;22:282-7
Spontaneous bacterial peritonitis (SBP) is a severe complication of decompensated cirrhosis, and the inhospital mortality for SBP ranges from 21.3% to 37%.,, Bacterial translocation is the major cause of SBP; therefore, no intra-abdominal source of infection can be found., Ascites culture is the gold standard for SBP diagnosis, and a high ascites polymorphonuclear leukocyte (PMN) count is accepted as an early indicator of SBP., An ascites PMN count ≥250/mm 3 is considered to indicate empirical antibiotic therapy based on the current guidelines., However, in clinical practice, this standard may lead to misdiagnosis in certain cases.
Symptomatic bacterascites (SB) is a variant of SBP with signs of infection but an ascites PMN count <250/mm 3, and it can be only confirmed by a positive ascites culture., SB is common in clinical practice accounting for 63% of SBP episodes and leads to a high mortality rate of 48.4%. According to current guideline, SB should be treated with antibiotics. Therefore, early detection of SB is important for both patients and physicians. Apparently, the threshold of ascites PMN count mentioned above is not useful for the diagnosis of SB, and ascites culture always takes several days; therefore, it is difficult to diagnose SB as soon as it appears.
To date, early indicators for SB have not been determined. Although reagent strips have been tested for detecting SB, the results are unsatisfactory. The current guideline suggests that patients with ascites PMN counts <250/mm 3 and signs of infection should be treated with antibiotics while awaiting culture results, which can be considered as management for SB. However, the “signs of infection” include several clinical manifestations only and no objective laboratory indices; therefore, this strategy would be implemented subjectively, leading to the abuse of antibiotics or delays in treatment. Worse still, ascites infection can manifest as diarrhea, hepatic encephalopathy, renal inadequacy, gastrointestinal hemorrhage, or other symptoms. These different manifestations seem unrelated to ascites infection, which makes SB difficult to be detected in time. Therefore, we aim to create a quick screening model for the early detection of patients with SB using a retrospective case–control study.
| Materials and Methods|| |
We performed a retrospective, two-center and case–control study in which inpatients with cirrhotic ascites at the First Affiliated Hospital of Nanchang University and the Infectious Disease Hospital of Nanchang University were enrolled between January 2010 and December 2014. The diagnosis of cirrhosis was based on clinical, laboratory, and imaging findings. Ascites was confirmed by abdominal paracentesis.
According to previous researches,, the diagnosis of SB was based on a positive ascitic fluid culture in association with an ascites PMN count <250/mm 3 and suspicious signs of SBP. As described in the guidelines (American Association for the Study of Liver Diseases and European Association for the Study of the Liver),, suspicious signs of SBP included local symptoms and/or signs of peritonitis (eg, abdominal pain, abdominal tenderness, vomiting, diarrhea, and ileus), signs of systemic inflammation (eg, hyper- or hypothermia, chills, peripheral leukocytosis, tachycardia, and/or tachypnea), worsening of liver function, hepatic encephalopathy, shock, renal failure, gastrointestinal bleeding, and acidosis.
Case and control group definitions
Subjects of the case group met the following criteria: (1) Patients with SB; (2) no antibiotics were administered in the two weeks prior to the presentation of SB; and (3) no intra-abdominal or extra-abdominal source of infection. Subjects of the control group met the following criteria: (1) Patients without any infection; (2) no antibiotics were administered in the two weeks prior to be enrolled; (3) PMN count < 250/mm 3 in an ascites sample, which was also negative for bacteria; and (4) discharged with improved condition including remission of symptoms, significant ascites reduction and stabilization, and improved blood biochemical indices without antibiotic treatment.
The infection mentioned above was defined by the necessity of antibiotic intervention. To achieve a patient sample as representative as possible of the actual clinical situation, no exclusion criteria regarding concurrent diseases or nonantibiotic treatments were implemented.
General information and candidate indicators of SB were reviewed. The former included gender, age, concurrent diseases, liver cancer, and cirrhosis etiology, and the latter included life signs (body temperature, heart rate, respiratory rate, systolic blood pressure, and diastolic blood pressure), local signs (abdominal pain, abdominal tenderness, abdominal rebound pain, vomiting, diarrhea, and ileus), clinical events (chills, shock, hepatic encephalopathy and gastrointestinal bleeding), hematological indices (white blood cell count, neutrophil percentage, prothrombin time and levels of alanine aminotransferase, aspartate aminotransferase, total bilirubin, albumin, bicarbonate, urea nitrogen, and creatinine), and ascites examination (nucleated leukocyte count, polymorphonuclear leukocyte count, and percentage of polymorphonuclear leukocytes). Candidate indicators were chosen based on the description of suspicious signs of SBP in the guidelines , and clinical judgment. Child–Pugh scores  were calculated and classified.
Candidate indicators were gathered on the day when ascites samples were collected for the control and case groups but before antibiotic administration for the case group. If the indicators were repeatedly measured, only the first measurements were recorded. If a patient experienced more than one event that met the inclusion criteria, only data for the first event were included in the analysis.
Quantitative variables that were normally distributed are expressed as mean ± SD, and those that were abnormally distributed are expressed as (median [interquartile range]) (M [IQR]). The quantitative variables were compared using the t-test or the rank sum test. Qualitative variables were expressed as percentages and compared using the Chi-square test or Fisher's exact test. Binary logistic regression was used for the multivariate analysis. A screening model for SB was created from the candidate indicators that were significant in the multivariate analysis. To reduce the risk for overfitting the screening model, it was stated that the ratio of candidate variables to the number of observed events should be 1:5 or less, in final model. The model was assessed using a receiver operating characteristics (ROC) curve that was applied to all patients in the current study. The cutoff value that resulted in the highest Youden's index was chosen as the optimal cutoff value. Significance was established at P < 0.05. Data analysis was performed using SPSS 19.0 software. As this study was exploratory in design, sample size and power of the test were not estimated formally.
| Results|| |
The clinical features of the patients are reported in [Table 1]. A total of 307 patients (65 from the First Affiliated Hospital of Nanchang University and 242 from the Infectious Disease Hospital of Nanchang University) were enrolled and then divided into the case group (103 patients) and the control group (204 patients). There were more male than female patients (74.8% vs. 25.2% and 71.1% vs. 28.9% for the case and control groups, respectively), and the most common etiology of cirrhosis was hepatitis B in both groups (75.7% vs. 71.1%). Differences in gender, age, concurrent diseases, presence of liver cancer, and etiology of cirrhosis between the two groups were not significant (P > 0.05). Child–Pugh score and class were significantly different between the two groups (P < 0.05).
Univariate analysis of candidate indicators for symptomatic bacterascites
The candidate indicators of SB are outlined in [Table 2] and [Table 3], which show that the incidences of abdominal pain, abdominal tenderness, and abdominal rebound pain were significantly higher in the SB patients than in the controls (13.6% vs. 3.4%, 46.6% vs. 5.9%, and 34% vs. 2.5%, respectively, P < 0.001). Although the ascites nucleated leukocyte count and PMN count of the SB patients were low, their median values were still significantly higher than those of the control group (200 vs. 100 and 70.3 vs. 24, respectively, P < 0.001). Other significant candidate indicators were body temperature, chills, hepatic encephalopathy, white blood cell count, blood neutrophil percentage, blood total bilirubin, prothrombin time, blood urea nitrogen, blood creatinine, and ascites polymorphonuclear leukocyte percentage (P ≤ 0.002). No patients experienced shock or ileus during this study.
|Table 2: Univariate analysis of candidate indicators for symptomatic bacterascites|
Click here to view
|Table 3: Univariate analysis of candidate indicators for symptomatic bacterascites|
Click here to view
Multivariate analysis of candidate indicators for symptomatic bacterascites
Indicators that were significant in the univariate analysis were further examined by multivariate analysis, and the results are shown in [Table 4]. Body temperature, abdominal tenderness (yes = 1, no = 0), blood neutrophil percentage, blood total bilirubin, prothrombin time, and ascites nucleated leukocyte count were significantly related to SB (P < 0.05).
|Table 4: Multivariate analysis of candidate indicators for symptomatic bacterascites|
Click here to view
Construction and assessment of screening model for symptomatic bacterascites
The screening score (SS) of SB for individual patients can be calculated by combining the values of six indicators with the regression coefficients reported in [Table 4] as follows: SS = exp (logit (SS))/ (1 + exp (logit (SS))), where logit (SS) = −50.325 + 1.09 × body temperature + 2.103 × abdominal tenderness (yes = 1, no = 0) +0.048 × blood neutrophil percentage + 0.008 × blood total bilirubin + 0.069 × prothrombin time + 0.013 × ascites nucleated leukocyte count. For example, for a hypothetical patient with a body temperature of 37.9°C, abdominal tenderness, blood neutrophil percentage of 60.6%, blood total bilirubin of 28.1μmol/L, prothrombin time of 16.4 seconds and ascites nucleated leukocyte count of 260/mm 3, the SS would be calculated as follows: Logit (SS) = −50.325 + 1.09 × 37.9 + 2.103 × 1 + 0.048 × 60.6 + 0.008 × 28.1 + 0.069 × 16.4 + 0.013 × 260 = 0.734; SS = exp (0.734)/ (1 + exp (0.734)) = 0.676.
The screening scores of all patients enrolled in this study were calculated. A ROC curve for this screening model was constructed, and the area under the curve (AUC) was 0.939 ([95% CI, 0.908–0.970], P < 0.001, [Figure 1]). According to the maximum Youden's index, the optimal cutoff value of SS was 0.328, and it had a sensitivity of 86.4% and a specificity of 92.2%.
|Figure 1: Receiver operating characteristic curve of the screening model for symptomatic bacterascites. Note: X-axis, 1-specificity; Y-axis, sensitivity; dashed line, reference line; solid curve, receiver operating characteristic curve of the screening model for symptomatic bacterascites, with an area under the curve was 0.939 ([95% confidence interval, 0.908–0.970], P < 0.001)|
Click here to view
| Discussion|| |
Ascites PMN count, ascitic fluid lactoferrin, serum procalcitonin, and other indicators have been reported to be useful for the early diagnosis of SBP.,, However, these methods were all tested in patients with ascites PMN counts ≥250/mm 3; thus, whether they are suitable for the screening of SB is still unknown. Ascites bacteria can be quickly detected by identification of bacterial 16S rRNA and by Raman spectroscopy,, which seem to be good techniques for diagnosing SB. Regrettably, these techniques are not recommended by the current guidelines., In the absence of effective indicators or means, the model generated in this study, with integrated factors including clinical manifestations and laboratory examinations for the screening of SB, would be useful.
According to a previous study, abdominal tenderness is common in patients with SBP. Wallerstedt et al. concluded that paying particular attention to abdominal tenderness may be the best way to become aware of the possible development of SBP. In this study, as a binary variable, abdominal tenderness had a large odds ratio, suggesting that positive abdominal signs are reliable indicators of SB when other effective indicators are not available. Presumably, abdominal rebound tenderness is also an important indicator of SB in clinical practice; however, it was not included in this model.
It has been reported that most patients with SBP have elevated body temperature. Similarly, in this study, among the continuous variables included in the model, body temperature had the largest odds ratio, which indicates that increased temperature is a powerful indicator of ascites infection. As a special pathophysiological characteristic, hypersplenism is common in decompensated liver cirrhosis and occurs with a low baseline of white blood cell count, which may not exceed the upper limit of normal even under stimulation by bacterial infection., This finding can mislead physicians attempting to determine whether infection is present; therefore, the blood neutrophil percentage, which is minimally impacted by hypersplenism, was recommended as a good substitute for white blood cell count in this study for the detection of infection.
Because an ascites PMN count cutoff value of ≥250/mm 3 is not capable of identifying SB, the ascites PMN count was not used in our screening model and was replaced by the ascites nucleated leukocyte count. In addition, Link et al. reported that an ascites nucleated leukocyte count lower than 1000/mm 3 was unlikely to indicate SBP, with a negative predictive value of 95.5%. Therefore, we hypothesize that it is feasible to determine an ascites nucleated leukocyte count cutoff value for the early detection of SB.
Generally, abnormal blood total bilirubin and prothrombin time values have not been regarded as direct indicators of SBP but rather as results or risk factors for SBP. However, these two indices were included in our screening model, as they were found to be significant indicators of SB in this study. A possible explanation may be that whether antibiotic treatment is warranted partially depends on the extent of liver function damage, and patients of suspected SB with worsening liver function cannot afford a fatal strike resulting from a missed diagnosis or delayed treatment. In other words, antibiotics should be used more readily in particular situations, such as when liver function deteriorates rapidly without an explicit cause.
Gastrointestinal bleeding is not exclusive to SB and was rejected by the univariate analysis in this study. Nevertheless, the current guideline suggests that patients with cirrhosis that are experiencing gastrointestinal bleeding should be treated with short-term (maximum 7 days) antibiotic prophylaxis (Class I, Level A) regardless of whether they have a bacterial infection. So cirrhotic patients with suspected SB and gastrointestinal bleeding should receive antibiotic therapy directly.
| Conclusion|| |
We identified several convenient indicators and developed a screening model for SB. Patients with suspected SB can be quickly screened according to the developed model, and a screening score ≥0.328 may be considered to indicate SB. Further studies are needed to validate this screening model and its rationality.
The authors thank Peng Huang, Wei Zhou and Biao Chen for their great help during this study.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| References|| |
Nobre SR, Cabral JE, Gomes JJ, Leitão MC. In-hospital mortality in spontaneous bacterial peritonitis: A new predictive model. Eur J Gastroenterol Hepatol 2008;20:1176-81.
Hung TH, Tsai CC, Hsieh YH, Tsai CC, Tseng CW, Tsai JJ. Effect of renal impairment on mortality of patients with cirrhosis and spontaneous bacterial peritonitis. Clin Gastroenterol Hepatol 2012;10:677-81.
Hassan EA, Abdel Rehim AS. Creatinine modified Child-Turcotte-Pugh and integrated model of end-stage liver disease scores as predictors of spontaneous bacterial peritonitis-related in-hospital mortality: Applicable or not. J Gastroenterol Hepatol 2015;30:1205-10.
Wiest R, Krag A, Gerbes A. Spontaneous bacterial peritonitis: Recent guidelines and beyond. Gut 2012;61:297-310.
Lutz P, Nischalke HD, Strassburg CP, Spengler U. Spontaneous bacterial peritonitis: The clinical challenge of a leaky gut and a cirrhotic liver. World J Hepatol 2015;7:304-14.
European Association for the Study of the Liver. EASL clinical practice guidelines on the management of ascites, spontaneous bacterial peritonitis, and hepatorenal syndrome in cirrhosis. J Hepatol 2010;53:397-417.
Runyon BA; AASLD Practice Guidelines Committee Management of adult patients with ascites due to cirrhosis: An update. Hepatology 2009;49:2087-107.
Campillo B, Richardet JP, Kheo T, Dupeyron C. Nosocomial spontaneous bacterial peritonitis and bacteremia in cirrhotic patients: Impact of isolate type on prognosis and characteristics of infection. Clin Infect Dis 2002;35:1-10.
Campillo B, Richardet JP, Dupeyron C. Diagnostic value of two reagent strips (Multistix 8 SG and Combur 2 LN) in cirrhotic patients with spontaneous bacterial peritonitis and symptomatic bacterascites. Gastroenterol Clin Biol 2006;30:446-52.
Enomoto H, Inoue S, Matsuhisa A, Nishiguchi S. Diagnosis of spontaneous bacterial peritonitis and an in situ
hybridization approach to detect an “unidentified” pathogen. Int J Hepatol 2014;2014:634617.
D'Amico G, De Franchis R; Cooperative Study Group. Upper digestive bleeding in cirrhosis. Post-therapeutic outcome and prognostic indicators. Hepatology 2003;38:599-612.
Garcia-Tsao G, Sanyal AJ, Grace ND, Carey W; Practice Guidelines Committee of the American Association for the Study of Liver Diseases; Practice Parameters Committee of the American College of Gastroenterology. Prevention and management of gastroesophageal varices and variceal hemorrhage in cirrhosis. Hepatology 2007;46:922-38.
Harrell FE Jr, Lee KL, Matchar DB, Reichert TA. Regression models for prognostic prediction: Advantages, problems, and suggested solutions. Cancer Treat Rep 1985;69:1071-7.
Parsi MA, Saadeh SN, Zein NN, Davis GL, Lopez R, Boone J, et al.
Ascitic fluid lactoferrin for diagnosis of spontaneous bacterial peritonitis. Gastroenterology 2008;135:803-7.
Yuan LY, Ke ZQ, Wang M, Li Y. Procalcitonin and C-reactive protein in the diagnosis and prediction of spontaneous bacterial peritonitis associated with chronic severe hepatitis B. Ann Lab Med 2013;33:449-54.
Pan HY, Chen CR, Shang SQ, Sun HY, Chen QW, Xu J, et al.
Determination of ascitic bacterial 16S rRNA by quantitative PCR-microarray in the diagnosis of spontaneous bacterial peritonitis. Zhonghua Gan Zang Bing Za Zhi 2011;19:297-300.
Kloß S, Rösch P, Pfister W, Kiehntopf M, Popp J. Toward culture-free Raman spectroscopic identification of pathogens in ascitic fluid. Anal Chem 2015;87:937-43.
Bibi S, Ahmed W, Arif A, Khan F, Alam SE. Clinical, laboratory and bacterial profile of spontaneous bacterial peritonitis in chronic liver disease patients. J Coll Physicians Surg Pak 2015;25:95-9.
Wallerstedt S, Olsson R, Simrén M, Broomé U, Wahlin S, Lööf L, et al.
Abdominal tenderness in ascites patients indicates spontaneous bacterial peritonitis. Eur J Intern Med 2007;18:44-7.
Boyer TD, Habib S. Big spleens and hypersplenism: Fix it or forget it? Liver Int 2015;35:1492-8.
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, 1256.e1-5.
Link BC, Ziske CG, Schepke M, Schmidt-Wolf IG, Sauerbruch T. Total ascitic fluid leukocyte count for reliable exclusion of spontaneous bacterial peritonitis in patients with ascites. Eur J Gastroenterol Hepatol 2006;18:181-6.
Danulescu RM, Stanciu C, Trifan A. Assessing the risk of decompensation by ascites and spontaneous bacterial peritonitis in cirrhosis. Rev Med Chir Soc Med Nat Iasi 2014;118:320-6.
Prof. Xuan Zhu
Department of Digestion, The First Affiliated Hospital of Nanchang University, Nanchang, Jiangxi 330006
Source of Support: None, Conflict of Interest: None
[Table 1], [Table 2], [Table 3], [Table 4]
| Article Access Statistics|
| Viewed||1816 |
| Printed||25 |
| Emailed||1 |
| PDF Downloaded||214 |
| Comments ||[Add] |