Incidence and Predictive Factors of First Episode of Spontaneous Bacterial Peritonitis in Cirrhosis with Ascites: Relevance of Ascitic Fluid Protein Concentration JOSEP LLACH,ANTONI RIMOLA,MIQUEL NAVASA, PEREGINES,JOAN M. SALMERON, ANGELSGINES, VICENTE ARROYO AND JOAN RODES Liver Unit, Hospital Clinic i Provincial, University of Barcelona, 08036 Barcelona, Spain
To investigate the long-term probability of the appearance of the first episode of spontaneous bacterial peritonitis in cirrhosis with ascites and to identify predictors of this complication, we closely followed throughout their illness 127 patients consecutively admitted to our unit for the treatment of an episode of ascites without prior spontaneous bacterial peritonitis (follow-up period 21 22 mo). Thirteen patients (10%) had the first spontaneous bacterial peritonitis episode during follow-up. The appearance probability of this complication is 11%at 1yr and 15% at 3 yr. Thirty-three variables obtained at admission (including clinical data, standard liver and kidney function test results, ascitic fluid protein concentrations and hemodynamic parameters) were analyzed in relation to their value in predicting spontaneous bacterial peritonitis development. In univariate analysis (Kaplan-Meiercurves) five variables reached statistical significance (p < 0.05) as predictive factors for the development of the first spontaneous bacterial peritonitis episode. These five variables were poor nutritional status, increased serum bilirubin levels, increased serum AST levels, decreased prothrombin activity and reduced total protein concentration in ascitic fluid. When these five variables were introduced in a multivariate analysis, only the ascitic fluid protein concentration was found to correlate independently with spontaneous bacterial peritonitis development (p = 0.002). The probability of first spontaneous bacterial peritonitis after 3 yr of follow-up was 24% and 4% in patients with ascitic fluid protein content lower than 1 gm/dl and greater than or equal to 1 gm/dl, respectively. We concluded that the most important predictor of the development of the first spontaneous bacterial peritonitis episode in cirrhosis with ascites is the total protein concentration in ascitic fluid. (HEPATOLOGY 1992;16:724-727.)
*
Received January 6, 1992;accepted May 8,1992. This work was supported by a grant from the Direccion General de Investigacion Cientfico y Tecnica (PM 91-02161,Spain. Dr.Josep Llach was a recipient of a grant from the Fundaci6 Catalana per l’Estudi de les Malalties del Fetm i Gastroenterolbgiques,Barcelona, Spain. Address reprint requests to:Autoni Rimola, M.D., Liver Unit, Hospital Clinic i Provincial, Villarroel170,08036 Barwlona, Spain. 31/1/39360
Spontaneous bacterial peritonitis (SBP) is a common and severe complication in cirrhotic patients with ascites. The incidence of SBP in cirrhosis has been reported to average 20% (1-5).A direct correlation has been found between the impairment of the local defensive mechanisms of ascites (assessed by either the opsonic activity or the total protein concentration in ascitic fluid) and the risk of SBP in cirrhotic patients hospitalized with ascites (6,7). It has also been reported that previous episodes of SBP highly predispose these patients to have new episodes of SBP, with a probability of SBP recurrence of approximately 70% 1yr after the resolution of the first SBP episode (8). On the other hand, Wang et al. from Taiwan (9) recently reported an 11%cumulative probability of SBP occurrence in ascitic patients with hepatitis B-related cirrhosis. However, at present there is no study specifically investigating the probability of the first SBP episode in patients with cirrhosis from other causes that are found more frequently than hepatitis B infection is in western countries. Furthermore, no information has been obtained about the significance of the alterations of ascitic fluid defensive mechanisms or other clinical and laboratory parameters in predicting the long-term risk of the first episode of SBP in cirrhosis with ascites. Therefore this study investigated the probability of the appearance of the first SBP episode in a large series of cirrhotic patients with ascites who were followed for a long period of time. It also investigated the predictive values of clinical parameters, standard liver and kidney function tests, ascitic fluid total protein concentration and several hemodynamic data in the development of this infective complication in these patients. MATERIALS AND METHODS The study was performed in 127 cirrhotic patients who were consecutively admitted to our unit for the treatment of an episode of ascites and who had not previously had SBP. As each patient was admitted, we obtained a detailed history, performed a complete physical examination and tapped the abdomen to measure the total protein concentration and cell count in ascitic fluid. On the first day of hospitalization, standard liver and kidney function test results were determined. The urinary sodium excretion and the glomerular
724
Vol. 16,No. 3, 1992
725
FIRST SPONTANEOUS BACTERIAL PERITONITIS IN CIRRHOSIS
TABLE 1. Variables analyzed as possible predictors of development of the first episode of SBP
0.4
4
Age Sex
Cause of cirrhosis Previous history of ascites Previous history of hepatic encephalopathy Previous history of gastrointestinal hemorrhage Time since the first episode of ascites Physical examination Hepatic stigmata Hepatomegaly Splenomegaly Nutritional status (good, regular or poor) Standard liver and kidney function tests Serum bilirubin
Serum albumin Prothrombin activity Serum AST Serum ALT Serum W n e phosphatase Serum y-glutamyl transpeptidase Serum cholesterol Serum y-globulin Platelet count WBC count Blood urea nitrogen Serum creatinine Serum sodium Urinary sodium excretion Glomerular filtration rate Other Child-Pugh classification (A, B and C) Total protein concentration in ascitic fluid Esophageal varices (presence or absence; small or large) Arterial pressure Portal pressure (hepatic venous pressure g-rahent) Treatment of ascites (diuretics,paracentesis or LeVeen shunt)
filtration rate (determined by inulin clearance) were measured after 4 days with a 40 mEq sodium diet and without diuretics. During the patients' hospitalization, we determined portal pressure (estimated as the hepatic venous pressure gradient) in 36 patients and performed fiberopticendoscopy in the upper gastrointestinal tract of 69 patients to establish the existence and size of esophageal varices. Twenty-six patients were treated with paracentesis associated with intravenous albumin infusion, and 101 patients were treated with diuretics (spironolactone alone or combined with furosemide. In 10 patients not responding to diuretic therapy, a peritoneovenous shunt was inserted. Sixteen of the 127 patients died during the hospitalization in which they had been included in the study. The other 111 were discharged from the hospital. Five of these 111patients were lost to follow-up, 4 immediately after being discharged and 1 after 23 mo. In 9 other patients we learned of patient survival and possible development of SBP by telephone or mail contact with their physicians. The remaining 97 patients were followed closely throughout their illness by staff members of our unit (follow-up period: 21 +- 22 mo; mean ? S.D.), with special emphasis placed on detecting the development of SBP. Because intestinal decontamination has been proved to be very effective in preventing infection in cirrhotic patients with gastrointestinal bleeding (lo), patients who had gastrointes-
: m
0.2
4
Y
0
12
24
36
48
MONTHS
FIG.1. Probability of development of the first SBP in 127 cirrhotic patients with ascites.
tinal hemorrhage during follow-up received oral nonabsorbable antibiotics (neomycin, colistin and nystatin) throughout the hemorrhagic episode. Patients with variceal hemorrhage during the follow-up period were treated with vasopressin or somatostatin administration as hemostatic measures, followed by sclerotherapy or @-blockeradministration to prevent hemorrhage recurrence. Esophageal tamponade was reserved for patients in whom pharmacological therapy failed, and surgery was done in only Child class A or B patients with uncontrollable bleeding. During follow-up, prophylactic oral nonabsorbable or partly absorbable antibiotics were not administered in circumstances other than hemorrhagic episodes. SBP was diagnosed when a patient fulfilled all of the following criteria: (a) having either abdominal pain, fever or leukocytosis with neutrophilia in the WBC count; (b) having ascitic fluid polymorphonuclear count greater than 250/mm3, and (c) having an absence of clinical, laboratory, radiologicalor ultrasonographical data suggesting secondary peritonitis. Because ascitic fluid cultures are negative in a proportion of SBP (1,111,the isolation of causative organisms was not considered essential for the diagnosis of SBP. In cases with culturenegative SBP, the existence of other abdominal disorders resembling SBP, such as pancreatitis, hemorrhage into ascites and peritoneal tuberculosis or carcinomatosis, were excluded by clinical and appropriate laboratory data. A total of 33 variables (Table 1)were analyzed as possible predictors of SBP appearance during follow-up.For qualitative variables patients were classified according to the presence or absence of each variable. For quantitative variables patients were grouped according to the median value of each variable, except in the case of the total protein concentration in ascitic fluid. For this variable the cutoff selected was 1gm/d because the risk of SBP during hospitalization of cirrhotic patients with ascites and the probability of SBP recurrence in patients recovering from the first SBP episode have been reported to be much greater in patients with an ascitic fluid protein content less than 1 gm/d than in those with higher ascites protein concentration (7, 8).After grouping the patients according to each variable, we performed univariate analysis by computing SBP appearance curves (Kaplan-Meiermethod) in each group and by comparing the curves with the Mantel-Cox test. Variables reaching statistical significance (p < 0.05) in the univariate analysis were subsequently introduced in a multivariate analysis by use of the Cox multiple regression method
726
HEPATOLOGY
LLACH ET AL.
TABLE 2. Probability of development of the first SBP episode in 127 cirrhotic patients with ascites ~
+ 0.3k
i
Variables
To investigate the long-term probability of the appearance of the first episode of spontaneous bacterial peritonitis in cirrhosis with ascites and to identify predictors of this complication, we closely followed throughout their illness 127 patients consecutively admitted to our unit for the treatment of an episode of ascites without prior spontaneous bacterial peritonitis (follow-up period 21 22 mo). Thirteen patients (10%) had the first spontaneous bacterial peritonitis episode during follow-up. The appearance probability of this complication is 11%at 1yr and 15% at 3 yr. Thirty-three variables obtained at admission (including clinical data, standard liver and kidney function test results, ascitic fluid protein concentrations and hemodynamic parameters) were analyzed in relation to their value in predicting spontaneous bacterial peritonitis development. In univariate analysis (Kaplan-Meiercurves) five variables reached statistical significance (p < 0.05) as predictive factors for the development of the first spontaneous bacterial peritonitis episode. These five variables were poor nutritional status, increased serum bilirubin levels, increased serum AST levels, decreased prothrombin activity and reduced total protein concentration in ascitic fluid. When these five variables were introduced in a multivariate analysis, only the ascitic fluid protein concentration was found to correlate independently with spontaneous bacterial peritonitis development (p = 0.002). The probability of first spontaneous bacterial peritonitis after 3 yr of follow-up was 24% and 4% in patients with ascitic fluid protein content lower than 1 gm/dl and greater than or equal to 1 gm/dl, respectively. We concluded that the most important predictor of the development of the first spontaneous bacterial peritonitis episode in cirrhosis with ascites is the total protein concentration in ascitic fluid. (HEPATOLOGY 1992;16:724-727.)
*
Received January 6, 1992;accepted May 8,1992. This work was supported by a grant from the Direccion General de Investigacion Cientfico y Tecnica (PM 91-02161,Spain. Dr.Josep Llach was a recipient of a grant from the Fundaci6 Catalana per l’Estudi de les Malalties del Fetm i Gastroenterolbgiques,Barcelona, Spain. Address reprint requests to:Autoni Rimola, M.D., Liver Unit, Hospital Clinic i Provincial, Villarroel170,08036 Barwlona, Spain. 31/1/39360
Spontaneous bacterial peritonitis (SBP) is a common and severe complication in cirrhotic patients with ascites. The incidence of SBP in cirrhosis has been reported to average 20% (1-5).A direct correlation has been found between the impairment of the local defensive mechanisms of ascites (assessed by either the opsonic activity or the total protein concentration in ascitic fluid) and the risk of SBP in cirrhotic patients hospitalized with ascites (6,7). It has also been reported that previous episodes of SBP highly predispose these patients to have new episodes of SBP, with a probability of SBP recurrence of approximately 70% 1yr after the resolution of the first SBP episode (8). On the other hand, Wang et al. from Taiwan (9) recently reported an 11%cumulative probability of SBP occurrence in ascitic patients with hepatitis B-related cirrhosis. However, at present there is no study specifically investigating the probability of the first SBP episode in patients with cirrhosis from other causes that are found more frequently than hepatitis B infection is in western countries. Furthermore, no information has been obtained about the significance of the alterations of ascitic fluid defensive mechanisms or other clinical and laboratory parameters in predicting the long-term risk of the first episode of SBP in cirrhosis with ascites. Therefore this study investigated the probability of the appearance of the first SBP episode in a large series of cirrhotic patients with ascites who were followed for a long period of time. It also investigated the predictive values of clinical parameters, standard liver and kidney function tests, ascitic fluid total protein concentration and several hemodynamic data in the development of this infective complication in these patients. MATERIALS AND METHODS The study was performed in 127 cirrhotic patients who were consecutively admitted to our unit for the treatment of an episode of ascites and who had not previously had SBP. As each patient was admitted, we obtained a detailed history, performed a complete physical examination and tapped the abdomen to measure the total protein concentration and cell count in ascitic fluid. On the first day of hospitalization, standard liver and kidney function test results were determined. The urinary sodium excretion and the glomerular
724
Vol. 16,No. 3, 1992
725
FIRST SPONTANEOUS BACTERIAL PERITONITIS IN CIRRHOSIS
TABLE 1. Variables analyzed as possible predictors of development of the first episode of SBP
0.4
4
Age Sex
Cause of cirrhosis Previous history of ascites Previous history of hepatic encephalopathy Previous history of gastrointestinal hemorrhage Time since the first episode of ascites Physical examination Hepatic stigmata Hepatomegaly Splenomegaly Nutritional status (good, regular or poor) Standard liver and kidney function tests Serum bilirubin
Serum albumin Prothrombin activity Serum AST Serum ALT Serum W n e phosphatase Serum y-glutamyl transpeptidase Serum cholesterol Serum y-globulin Platelet count WBC count Blood urea nitrogen Serum creatinine Serum sodium Urinary sodium excretion Glomerular filtration rate Other Child-Pugh classification (A, B and C) Total protein concentration in ascitic fluid Esophageal varices (presence or absence; small or large) Arterial pressure Portal pressure (hepatic venous pressure g-rahent) Treatment of ascites (diuretics,paracentesis or LeVeen shunt)
filtration rate (determined by inulin clearance) were measured after 4 days with a 40 mEq sodium diet and without diuretics. During the patients' hospitalization, we determined portal pressure (estimated as the hepatic venous pressure gradient) in 36 patients and performed fiberopticendoscopy in the upper gastrointestinal tract of 69 patients to establish the existence and size of esophageal varices. Twenty-six patients were treated with paracentesis associated with intravenous albumin infusion, and 101 patients were treated with diuretics (spironolactone alone or combined with furosemide. In 10 patients not responding to diuretic therapy, a peritoneovenous shunt was inserted. Sixteen of the 127 patients died during the hospitalization in which they had been included in the study. The other 111 were discharged from the hospital. Five of these 111patients were lost to follow-up, 4 immediately after being discharged and 1 after 23 mo. In 9 other patients we learned of patient survival and possible development of SBP by telephone or mail contact with their physicians. The remaining 97 patients were followed closely throughout their illness by staff members of our unit (follow-up period: 21 +- 22 mo; mean ? S.D.), with special emphasis placed on detecting the development of SBP. Because intestinal decontamination has been proved to be very effective in preventing infection in cirrhotic patients with gastrointestinal bleeding (lo), patients who had gastrointes-
: m
0.2
4
Y
0
12
24
36
48
MONTHS
FIG.1. Probability of development of the first SBP in 127 cirrhotic patients with ascites.
tinal hemorrhage during follow-up received oral nonabsorbable antibiotics (neomycin, colistin and nystatin) throughout the hemorrhagic episode. Patients with variceal hemorrhage during the follow-up period were treated with vasopressin or somatostatin administration as hemostatic measures, followed by sclerotherapy or @-blockeradministration to prevent hemorrhage recurrence. Esophageal tamponade was reserved for patients in whom pharmacological therapy failed, and surgery was done in only Child class A or B patients with uncontrollable bleeding. During follow-up, prophylactic oral nonabsorbable or partly absorbable antibiotics were not administered in circumstances other than hemorrhagic episodes. SBP was diagnosed when a patient fulfilled all of the following criteria: (a) having either abdominal pain, fever or leukocytosis with neutrophilia in the WBC count; (b) having ascitic fluid polymorphonuclear count greater than 250/mm3, and (c) having an absence of clinical, laboratory, radiologicalor ultrasonographical data suggesting secondary peritonitis. Because ascitic fluid cultures are negative in a proportion of SBP (1,111,the isolation of causative organisms was not considered essential for the diagnosis of SBP. In cases with culturenegative SBP, the existence of other abdominal disorders resembling SBP, such as pancreatitis, hemorrhage into ascites and peritoneal tuberculosis or carcinomatosis, were excluded by clinical and appropriate laboratory data. A total of 33 variables (Table 1)were analyzed as possible predictors of SBP appearance during follow-up.For qualitative variables patients were classified according to the presence or absence of each variable. For quantitative variables patients were grouped according to the median value of each variable, except in the case of the total protein concentration in ascitic fluid. For this variable the cutoff selected was 1gm/d because the risk of SBP during hospitalization of cirrhotic patients with ascites and the probability of SBP recurrence in patients recovering from the first SBP episode have been reported to be much greater in patients with an ascitic fluid protein content less than 1 gm/d than in those with higher ascites protein concentration (7, 8).After grouping the patients according to each variable, we performed univariate analysis by computing SBP appearance curves (Kaplan-Meiermethod) in each group and by comparing the curves with the Mantel-Cox test. Variables reaching statistical significance (p < 0.05) in the univariate analysis were subsequently introduced in a multivariate analysis by use of the Cox multiple regression method
726
HEPATOLOGY
LLACH ET AL.
TABLE 2. Probability of development of the first SBP episode in 127 cirrhotic patients with ascites ~
+ 0.3k
i
Variables
