GASTROENTEROLOGY
1991;100:499-493
A Rodent Model of Cirrhosis, Ascites, and Bacterial Peritonitis BRUCE A. RUNYON, SHIGEO SUGANO, and MARTHA A. MELLENCAMP
GARY KANEL,
Liver Unit, University of Southern California School of Medicine, Los Angeles, California; and Division of Immunology, James N. Gamble Institute of Medical Research, Cincinnati, Ohio
We sought to develop a rodent model of spontaneous bacterial peritonitis and report here the preliminary results of carbon tetrachloride-induced cirrhosis in which ascites and bacterial peritonitis predictably develop. Of 41 rats that survived the initial carbon tetrachloride toxicity, 38 (92.7%) developed cirrhosis with ascites. Of these 38,21(55.3%) developed 24 episodes of ascitic fluid infection without iatrogenic colonization. No surgically treatable source of infection was identified at autopsy in any rat; therefore, the infections were presumed to be “spontaneous.” Eight (50%) of the 18 rats with culture-positive ascitic fluid at postmortem examination also had spontaneous pleural fluid infection with the same organism. Escherichia coli and Proteus sp. were the organisms most commonly isolated. This rodent model of cirrhosis with ascites appears to be the first high-yield animal model of spontaneous bacterial peritonitis. Ascitic fluid infection in these rats resembles ascitic fluid infection in humans. This model will allow further investigation of the mechanisms of pathogenesis of ascitic fluid infection and provide insight into the prevention and treatment of spontaneous bacterial peritonitis and pleural fluid infection in patients with cirrhosis.
S mon and frequently fatal complication sis. Ten percent to 27% of patients with pontaneous
bacterial
peritonitis
(SBP) is a comof cirrhoascites are found to have infected ascites at the time of hospital admission (1,2). The hospitalization mortality rate was 78% of the 246 patients reported as of 1985 (3). The reasons for this extremely high fatality rate are unclear. Progress in reducing mortality is dependent on understanding the fundamental processes that are involved in the pathogenesis of SBP. Insight into its pathogenesis might improve the results of treatment or perhaps even prevent this infection, resulting in improved survival of these patients. To date research
in the area of ascitic fluid infection has involved clinical studies of patients (4) and in vitro studies of the opsonic activity of ascitic fluid (5-7). There are no published animal models of SBP. Some studies of the pathogenesis of ascitic fluid infection would be difficult to perform in patients or would be subject to ethical concerns. An animal model of SBP might be of value in providing insight into the pathogenesis of ascitic fluid infection. In an animal model of cirrhosis designed by Mellencamp and Preheim to study pneumococcal pneumonia, a small percentage of rats with cirrhotic ascites were noted to die of gram-negative peritonitis before pneumonia could be induced (8). The model reported here is a modification of the model of Mellencamp and Preheim. Their model was based on the classic model of Proctor and Chatamra (9). Materials and Methods Male Sprague-Dawley rats weighing 100-120 g were treated with 1.5 mmol/L phenobarbital in tap water (the only accessible water) for 10-14 days until their weight increased to >ZOO g. They were fed rodent chow (Leach Grain and Milling Co., Downey, CA) and kept in a room with an 18/6-hour light/dark cycle. After the rats had achieved this weight, weekly intragastric installations of carbon tetrachloride (Baxter Healthcare Corp. Analytical Reagent, McGaw Park, IL) were begun using light methoxyflurane anesthesia (Pitman-Moore, Inc., Washington Crossing, NJ) and a Hamilton syringe (Hamilton Co., Reno, NV) with an attached stainless steel animal feeding tube (Popper and Sons, New Hyde Park, NY). The initial carbon tetrachloride dose was 20 @rat. Subsequent doses were adjusted based on the change in body weight 48 hours after the last dose (Table 1). After development of ascites, a small stable
Abbreviation used in this paper: SBP, spontaneous bacterial peritonitis. o 1991by the American GastroenterologicalAssociation 0016-5085/91/$3.00
GASTROENTEROLOGY Vol. 100, No. 2
490 RUNYONET AL.
Table I. Dosing Schedule of Carbon Tetrachloride
Weightchange, 46 h after last dose Stable or increasing 2%-5.9% loss 6%-10% loss lO.l%-15% > 15% loss
loss
250/mm3 grew bacteria; i.e., there were no episodes of culture-negative neutrocytic ascites (11). Seventeen (70.8%) of the 24 episodes of ascitic fluid infection were episodes of SBP; 7 (29.2%) were bacterascites episodes. Five of the bacterascites episodes were diagnosed before death, and 3 episodes of SBP were detected before death. Of the 16 episodes of ascitic fluid infection that were detected postmortem, 14 were episodes of SBP and 2 were bacterascites episodes. Both bacterascites specimens grew EscheriTable 2. Ascitic Fluid Culture Results Organism(s) E. coli E.coli and Proteus sp.
Proteus sp. Shigella sp. Gamma
Total
streptococcus
No. of episodes 10 8 4 1 1
24
% of total 41.7 33.3 16.7 4.2 4.2
100
P
Culture-positive
693 rfr 550
< 0.005
2928 t 3358
63 f 55 6f4
< 0.005 < 0.0001
1033 + 1363 32 2 22
chia coli. The neutrophil count was 200/mm3 in one sample and 34/mm3 in the other. It was of interest to us to compare the natural history of untreated bacterascites with that of untreated SBP in this model. However, the numbers were too small to permit meaningful statistical analysis. All 3 rats found to have SBP before death died rapidly (within 5 hours, 2 days, and 6 days) and were found to have the same organisms (2 had E. coli; 1, E. coli and Proteus) in their ascitic fluid at postmortem examination. This rapid progression of infection may have been the reason that most examples of SBP were diagnosed postmortem. In contrast, only 1 rat with bacterascites (E. coli and Proteus) died rapidly (24 hours). The other 4 rats with premortem bacterascites (E. coli and Proteus, E. coli, E. coli and Proteus, and Proteus) lived 7, 10, 48, and 51 days, respectively, before dying. One of the rats that grew Proteus in ascitic fluid initially developed documented E. coliand Proteus-induced SBP 49 days after the initial episode of bacterascites; this rat then died with E. coli- and Proteus-induced SBP 2 days later. Another of these rats (with E. coli- and Proteus-induced bacterascites) was found to resolve the ascitic fluid colonization in 6 consecutive specimens obtained over 46 days before dying with Proteus -induced SBP. Unfortunately 3 rats with premortem bacterascites had no ascitic fluid at the time of autopsy, presumably because of dehydration immediately preceding death. Therefore, data regarding postmortem ascitic fluid cell counts and cultures were unavailable in these 3 rats. Infected pleural effusions were also common in this model. Of the 16 rats with infected ascites at autopsy, 8 (50%) had pleural fluid infection with the same organism, including the 2 rats with E. coli-induced bacterascites. There were 3 episodes each of E. coli and Proteus pleural fluid infection and 2 episodes in which both organisms were cultured. All infected pleural fluid samples (except one E. coli episode in which the neutrophil count was 134/mm3) were neutrocytic. In an attempt to reduce the very high initial fatality rate associated with carbon tetrachloride, the dose of Ccl, given to the final group of 25 rats (included in the total group of 87 rats) was based on the change in weight 72 hours after the last dose rather than the
492
RUNYON ET AL.
48-hour weight. The 72-hour weight was in general the nadir weight; therefore, this change in dosing resulted in a slightly lower dose of toxin. Unfortunately, this did not reduce the mortality (52%), but significantly slowed the development of ascites (11.1+ 2.1 weeks for 72-hour weight vs. 8.5 2 2.5 weeks for 48-hour weight; P < 0.005). Also, ascitic fluid infection appeared to be less common in rats that were dosed based on 72-hour weight change compared with those whose dose was based on 48-hour weight change; 33.3% (4 of 12 infected rats) vs. 65.4% (17 of 26 infected rats; P > 0.1). Perhaps the slightly higher dose given the rats dosed based on 48-hour weights was crucial in producing an adequate insult to the liver, resulting in rapid ascites formation and ascitic fluid infection. Discussion This rodent model of carbon tetrachlorideinduced cirrhosis with ascites seems to be the first high-yield animal model of spontaneous bacterial peritonitis. Thirty-eight (92.7%) of the 41 rats that survived the first few doses of carbon tetrachloride developed cirrhotic ascites. Of these 38 rats, 21 (55.3%)developed 24 episodes of spontaneous ascitic fluid infection, including 8 premortem and 16 postmortem episodes. Eight (50%) of the 16 rats with culture-positive ascitic fluid at postmortem examination also had spontaneous pleural fluid infection with the same organism. Rats were not colonized iatrogenitally with virulent organisms: the infections developed spontaneously. Antibiotic treatment was not given in this study. This model of ascitic fluid infection resembles the human disease in several respects: (a) gut flora, e.g. E. coli, were the most common isolates; (b) bacterascites was common (2); (c) SBP was usually fatal; and (d) bacterascites frequently resolved (12). Also, the neutrophil cut-off of 250 cells/mm3 as a criterion for SBP seemed as applicable to rats as it is humans (4). All specimens of rat ascitic fluid with a neutrophil count above this value grew bacteria. The method of ascitic fluid culture that has been validated in humans was used in this study and worked very well (10). However, there are some differences in ascitic fluid infection between cirrhotic rats and humans. Humans with SBP usually have a predominance of neutrophils in their ascitic fluid white blood cell counts. In this model neutrophils constituted a minority of the white cells in ascitic fluid. This difference may be explained by the predominance of lymphocytes and monocytes in the peripheral blood of the rat, in contrast to the neutrophil predominance in humans. Human SBP is essentially always a monomicrobial infection (10). Eight (33.3%) of 24 episodes of ascitic fluid infection
GASTROENTEROLOGY
Vol. 100,No. 2
in rodents were polymicrobial. Also, the prevalence of Proteus sp. as ascitic fluid isolates in rats (37.5% of total isolates) was much higher than in human SBP (- 0%) (10).These differences (polymicrobial infection and prevalence of Proteus sp.) in the flora of ascitic fluid between rats and humans may reflect interspecies differences in the porosity of the gut and flora of the gut, respectively (13). Perhaps the rat gut and/or mesenteric lymphatics are more permeable than those of humans and more readily permit translocation of gut lumen flora, therefore resulting in polymicrobial rather than monomicrobial infection (13). Further studies will be needed to pursue these issues. The high frequency of infected pleural effusions was unexpected in this model but intriguing. Case reports and two recent studies confirm that pleural fluid infection in cirrhotic humans is not uncommon (14-16). Most episodes of ascitic fluid infection were detected at postmortem examination in this study; only one third were detected before death. This raises the question whether most episodes represent postmortem colonization of the ascitic fluid. However, ascitic fluid infection was the only apparent cause of death in the rats found to have infected ascites at postmortem examination. In addition, the presence of an elevated neutrophil count in the fluid in all but 2 of the 16 postmortem episodes is evidence for premortem entry of bacteria into the fluid with activation of complement and entry of these phagocytes into the peritoneal cavity. The 2 postmortem bacterascites episodes were monomicrobial E. coli infections; 1 involved a neutrophi1 count that was almost 250/mm3. The predominance of monomicrobial postmortem ascitic fluid infection (75%) also provides evidence against postmortem colonization. Polymicrobial postmortem ascitic fluid cultures would be expected if the fluid were colonized after death. Finally, the presence of monomicrobial infection of neutrocytic pleural fluid also provides evidence that these infections probably began before death. It would be unlikely that a single species of gut flora would be able to traverse the diaphragm and incite a neutrophil response after death. It is possible that the carbon tetrachloride contributed to ascites formation in this model by its known nephrotoxicity (17) in addition to its fibrogenic effect on the liver. This may explain why continued administration of the toxin (after ascites developed) was needed to ensure persistence of ascites. Although carbon tetrachloride-induced gut damage has not been reported (8, 9,17,18), it is also possible that oral administration of the toxin may predispose to ascitic fluid infection by damaging the gut mucosa. The absence of reports of SBP in animal models that involve inhalation or intraperitoneal administration
February 1991
of the agent lends some credence to this possibility. However, evidence of mucosal damage was not noted in this model. Further studies of the ultrastructural and physiologic effects of carbon tetrachloride on the gut are needed. It is possible that SBP develops in humans, in part, because of an increase in gut mucosal permeability related to portal hypertension and/or gut edema. Perhaps carbon tetrachloride increases the risk of peritonitis in cirrhotic rats with fluid overload also because of an enhancement of mucosal permeability. Spontaneous bacterial peritonitis and bacterascites are common complications in patients with cirrhotic ascites (l-4). The current theory of the pathogenesis of SBP is that gut or respiratory flora invade the bloodstream of susceptible patients and then enter ascites via hepatic sinusoids (7). This colonization of ascites, i.e., bacterascites, may then resolve after eradication of bacteria by host defenses or may progress to SBP and death. It is of interest that premortem bacterascites was documented in this rat model of ascitic fluid infection and that this colonization resolved in some rats and progressed in others. Further detailed studies of this model will allow investigation of the pathogenesis, prevention, and treatment of SBP and pleural fluid infection. The source of the organisms can be proven using labeled bacteria. Gut mucosal permeability can be measured serially during induction of cirrhosis and ascites formation to determine if stepwise changes occur before the development of ascitic fluid infection. Factors that predispose some but not all rats to infection can be pursued. Prophylactic and therapeutic antibiotic treatment can be investigated. Trials of bacterial vaccines, passive transfer of immunoglobulin, and immunomodulation can be performed.
RAT MODEL OF BACTERIAL PERITONITIS
493
4. Runyon BA. Spontaneous bacterial peritonitis: an explosion of information. Hepatology 1988;8:171-175. 5. Runyon BA, Morrissey R, Hoefs JC, Wyle F. Opsonic activity of human ascitic fluid: a potentially important protective mechanism against spontaneous bacterial peritonitis. Hepatology 1985;5:634-637. 6. Runyon BA, Van Epps DE. Diuresis of cirrhotic ascites increases its opsonic activity and may help prevent spontaneous bacterial peritonitis. Hepatology 1986;6:396-399. 7. Runyon BA. Patients with deficient ascitic fluid opsonic activity are predisposed to spontaneous bacterial peritonitis. Hepatology 1988;8:632-635. 8. Mellencamp MA, Preheim LC. Pneumococcal pneumonia in a rat model of cirrhosis: effects of cirrhosis on pulmonary defense mechanisms against Streptococcus pneumoniae. J Infect Dis (in press). 9. Proctor E, Chatamra K. High yield micronodular cirrhosis in the rat. Gastroenterology 1982;83:1183-1190. 10. Runyon BA, Canawati I-IN, Akriviadis EA. Optimization of ascitic fluid culture technique. Gastroenterology 1988;95:13511355. 11. Runyon BA, Hoefs JC. Culture-negative neutrocytic ascites: a variant of spontaneous bacterial peritonitis. Hepatology 1984;4: 1209-1211. bacterascites: a 12. Runyon BA. Monomicrobial nonneutrocytic variant of spontaneous bacterial peritonitis. Hepatology 1990; 12:710-715. 13. Steffen EK, Berg RD. Deitch EA. Comparison of translocation rates of various indigenous bacteria from the gastrointestinal tract to the mesenteric lymph node. J Infect Dis 1988;157:10321038. 14. Flaum MA. Spontaneous bacterial empyema in cirrhosis. Gastroenterology 1976;70:416-417. 15. Xiol X, Castellote J, Baliellas C, et al. Spontaneous bacterial empyema in cirrhotic patients: analysis of eleven cases. Hepatology 1990;11:365-370. 16. Ben-Yehuda 0, Samniah N, Guetta V, Khoury M, Sikuler E, Schlaeffer F. Spontaneous bacterial pleuritis in patients with cirrhosis (abstr). Gastroenterology 1990;98:A568. 17. Zimmerman SW, Norbach DH. Nephrotoxic effects of longterm carbon tetrachloride administration in rats. Arch Path01 Lab Med 1980;104:94-99. 18. Charbonneau M, Tuchweber B, Plaa GL. Acetone potentiation of chronic liver injury induced by administration of carbon tetrachloride. Hepatology 1986;6:684-700.
References 1. Runyon BA. Low-protein-concentration
ascitic fluid is predisposed to spontaneous bacterial peritonitis. Gastroenterology 1986;91:1343-1346. 2. Pinzello G, Simonetti RG, Craxi A, Di Piazza S, Spano C, Pagliaro L. Spontaneous bacterial peritonitis: a prospective investigation in predominantly nonalcoholic cirrhotic patients. Hepatology 1983;3:545-549. 3. Hoefs JC, Runyon BA. Spontaneous bacterial peritonitis. Dis Mon 1985:31:1-48.
Received February 13,199O. Accepted August 1,199O. Address requests for reprints to: Bruce A. Runyon, M.D., Associate Professor of Medicine, Department of Internal Medicine, GastroenterologyiI-Iepatology Division, University of Iowa Hospitals and Clinics, Iowa City, Iowa 52242. This work was supported in part by the Hastings Foundation of Los Angeles and by the United Liver Association of Los Angeles. The authors thank Francine Charboneau for technical assistance.
1991;100:499-493
A Rodent Model of Cirrhosis, Ascites, and Bacterial Peritonitis BRUCE A. RUNYON, SHIGEO SUGANO, and MARTHA A. MELLENCAMP
GARY KANEL,
Liver Unit, University of Southern California School of Medicine, Los Angeles, California; and Division of Immunology, James N. Gamble Institute of Medical Research, Cincinnati, Ohio
We sought to develop a rodent model of spontaneous bacterial peritonitis and report here the preliminary results of carbon tetrachloride-induced cirrhosis in which ascites and bacterial peritonitis predictably develop. Of 41 rats that survived the initial carbon tetrachloride toxicity, 38 (92.7%) developed cirrhosis with ascites. Of these 38,21(55.3%) developed 24 episodes of ascitic fluid infection without iatrogenic colonization. No surgically treatable source of infection was identified at autopsy in any rat; therefore, the infections were presumed to be “spontaneous.” Eight (50%) of the 18 rats with culture-positive ascitic fluid at postmortem examination also had spontaneous pleural fluid infection with the same organism. Escherichia coli and Proteus sp. were the organisms most commonly isolated. This rodent model of cirrhosis with ascites appears to be the first high-yield animal model of spontaneous bacterial peritonitis. Ascitic fluid infection in these rats resembles ascitic fluid infection in humans. This model will allow further investigation of the mechanisms of pathogenesis of ascitic fluid infection and provide insight into the prevention and treatment of spontaneous bacterial peritonitis and pleural fluid infection in patients with cirrhosis.
S mon and frequently fatal complication sis. Ten percent to 27% of patients with pontaneous
bacterial
peritonitis
(SBP) is a comof cirrhoascites are found to have infected ascites at the time of hospital admission (1,2). The hospitalization mortality rate was 78% of the 246 patients reported as of 1985 (3). The reasons for this extremely high fatality rate are unclear. Progress in reducing mortality is dependent on understanding the fundamental processes that are involved in the pathogenesis of SBP. Insight into its pathogenesis might improve the results of treatment or perhaps even prevent this infection, resulting in improved survival of these patients. To date research
in the area of ascitic fluid infection has involved clinical studies of patients (4) and in vitro studies of the opsonic activity of ascitic fluid (5-7). There are no published animal models of SBP. Some studies of the pathogenesis of ascitic fluid infection would be difficult to perform in patients or would be subject to ethical concerns. An animal model of SBP might be of value in providing insight into the pathogenesis of ascitic fluid infection. In an animal model of cirrhosis designed by Mellencamp and Preheim to study pneumococcal pneumonia, a small percentage of rats with cirrhotic ascites were noted to die of gram-negative peritonitis before pneumonia could be induced (8). The model reported here is a modification of the model of Mellencamp and Preheim. Their model was based on the classic model of Proctor and Chatamra (9). Materials and Methods Male Sprague-Dawley rats weighing 100-120 g were treated with 1.5 mmol/L phenobarbital in tap water (the only accessible water) for 10-14 days until their weight increased to >ZOO g. They were fed rodent chow (Leach Grain and Milling Co., Downey, CA) and kept in a room with an 18/6-hour light/dark cycle. After the rats had achieved this weight, weekly intragastric installations of carbon tetrachloride (Baxter Healthcare Corp. Analytical Reagent, McGaw Park, IL) were begun using light methoxyflurane anesthesia (Pitman-Moore, Inc., Washington Crossing, NJ) and a Hamilton syringe (Hamilton Co., Reno, NV) with an attached stainless steel animal feeding tube (Popper and Sons, New Hyde Park, NY). The initial carbon tetrachloride dose was 20 @rat. Subsequent doses were adjusted based on the change in body weight 48 hours after the last dose (Table 1). After development of ascites, a small stable
Abbreviation used in this paper: SBP, spontaneous bacterial peritonitis. o 1991by the American GastroenterologicalAssociation 0016-5085/91/$3.00
GASTROENTEROLOGY Vol. 100, No. 2
490 RUNYONET AL.
Table I. Dosing Schedule of Carbon Tetrachloride
Weightchange, 46 h after last dose Stable or increasing 2%-5.9% loss 6%-10% loss lO.l%-15% > 15% loss
loss
250/mm3 grew bacteria; i.e., there were no episodes of culture-negative neutrocytic ascites (11). Seventeen (70.8%) of the 24 episodes of ascitic fluid infection were episodes of SBP; 7 (29.2%) were bacterascites episodes. Five of the bacterascites episodes were diagnosed before death, and 3 episodes of SBP were detected before death. Of the 16 episodes of ascitic fluid infection that were detected postmortem, 14 were episodes of SBP and 2 were bacterascites episodes. Both bacterascites specimens grew EscheriTable 2. Ascitic Fluid Culture Results Organism(s) E. coli E.coli and Proteus sp.
Proteus sp. Shigella sp. Gamma
Total
streptococcus
No. of episodes 10 8 4 1 1
24
% of total 41.7 33.3 16.7 4.2 4.2
100
P
Culture-positive
693 rfr 550
< 0.005
2928 t 3358
63 f 55 6f4
< 0.005 < 0.0001
1033 + 1363 32 2 22
chia coli. The neutrophil count was 200/mm3 in one sample and 34/mm3 in the other. It was of interest to us to compare the natural history of untreated bacterascites with that of untreated SBP in this model. However, the numbers were too small to permit meaningful statistical analysis. All 3 rats found to have SBP before death died rapidly (within 5 hours, 2 days, and 6 days) and were found to have the same organisms (2 had E. coli; 1, E. coli and Proteus) in their ascitic fluid at postmortem examination. This rapid progression of infection may have been the reason that most examples of SBP were diagnosed postmortem. In contrast, only 1 rat with bacterascites (E. coli and Proteus) died rapidly (24 hours). The other 4 rats with premortem bacterascites (E. coli and Proteus, E. coli, E. coli and Proteus, and Proteus) lived 7, 10, 48, and 51 days, respectively, before dying. One of the rats that grew Proteus in ascitic fluid initially developed documented E. coliand Proteus-induced SBP 49 days after the initial episode of bacterascites; this rat then died with E. coli- and Proteus-induced SBP 2 days later. Another of these rats (with E. coli- and Proteus-induced bacterascites) was found to resolve the ascitic fluid colonization in 6 consecutive specimens obtained over 46 days before dying with Proteus -induced SBP. Unfortunately 3 rats with premortem bacterascites had no ascitic fluid at the time of autopsy, presumably because of dehydration immediately preceding death. Therefore, data regarding postmortem ascitic fluid cell counts and cultures were unavailable in these 3 rats. Infected pleural effusions were also common in this model. Of the 16 rats with infected ascites at autopsy, 8 (50%) had pleural fluid infection with the same organism, including the 2 rats with E. coli-induced bacterascites. There were 3 episodes each of E. coli and Proteus pleural fluid infection and 2 episodes in which both organisms were cultured. All infected pleural fluid samples (except one E. coli episode in which the neutrophil count was 134/mm3) were neutrocytic. In an attempt to reduce the very high initial fatality rate associated with carbon tetrachloride, the dose of Ccl, given to the final group of 25 rats (included in the total group of 87 rats) was based on the change in weight 72 hours after the last dose rather than the
492
RUNYON ET AL.
48-hour weight. The 72-hour weight was in general the nadir weight; therefore, this change in dosing resulted in a slightly lower dose of toxin. Unfortunately, this did not reduce the mortality (52%), but significantly slowed the development of ascites (11.1+ 2.1 weeks for 72-hour weight vs. 8.5 2 2.5 weeks for 48-hour weight; P < 0.005). Also, ascitic fluid infection appeared to be less common in rats that were dosed based on 72-hour weight change compared with those whose dose was based on 48-hour weight change; 33.3% (4 of 12 infected rats) vs. 65.4% (17 of 26 infected rats; P > 0.1). Perhaps the slightly higher dose given the rats dosed based on 48-hour weights was crucial in producing an adequate insult to the liver, resulting in rapid ascites formation and ascitic fluid infection. Discussion This rodent model of carbon tetrachlorideinduced cirrhosis with ascites seems to be the first high-yield animal model of spontaneous bacterial peritonitis. Thirty-eight (92.7%) of the 41 rats that survived the first few doses of carbon tetrachloride developed cirrhotic ascites. Of these 38 rats, 21 (55.3%)developed 24 episodes of spontaneous ascitic fluid infection, including 8 premortem and 16 postmortem episodes. Eight (50%) of the 16 rats with culture-positive ascitic fluid at postmortem examination also had spontaneous pleural fluid infection with the same organism. Rats were not colonized iatrogenitally with virulent organisms: the infections developed spontaneously. Antibiotic treatment was not given in this study. This model of ascitic fluid infection resembles the human disease in several respects: (a) gut flora, e.g. E. coli, were the most common isolates; (b) bacterascites was common (2); (c) SBP was usually fatal; and (d) bacterascites frequently resolved (12). Also, the neutrophil cut-off of 250 cells/mm3 as a criterion for SBP seemed as applicable to rats as it is humans (4). All specimens of rat ascitic fluid with a neutrophil count above this value grew bacteria. The method of ascitic fluid culture that has been validated in humans was used in this study and worked very well (10). However, there are some differences in ascitic fluid infection between cirrhotic rats and humans. Humans with SBP usually have a predominance of neutrophils in their ascitic fluid white blood cell counts. In this model neutrophils constituted a minority of the white cells in ascitic fluid. This difference may be explained by the predominance of lymphocytes and monocytes in the peripheral blood of the rat, in contrast to the neutrophil predominance in humans. Human SBP is essentially always a monomicrobial infection (10). Eight (33.3%) of 24 episodes of ascitic fluid infection
GASTROENTEROLOGY
Vol. 100,No. 2
in rodents were polymicrobial. Also, the prevalence of Proteus sp. as ascitic fluid isolates in rats (37.5% of total isolates) was much higher than in human SBP (- 0%) (10).These differences (polymicrobial infection and prevalence of Proteus sp.) in the flora of ascitic fluid between rats and humans may reflect interspecies differences in the porosity of the gut and flora of the gut, respectively (13). Perhaps the rat gut and/or mesenteric lymphatics are more permeable than those of humans and more readily permit translocation of gut lumen flora, therefore resulting in polymicrobial rather than monomicrobial infection (13). Further studies will be needed to pursue these issues. The high frequency of infected pleural effusions was unexpected in this model but intriguing. Case reports and two recent studies confirm that pleural fluid infection in cirrhotic humans is not uncommon (14-16). Most episodes of ascitic fluid infection were detected at postmortem examination in this study; only one third were detected before death. This raises the question whether most episodes represent postmortem colonization of the ascitic fluid. However, ascitic fluid infection was the only apparent cause of death in the rats found to have infected ascites at postmortem examination. In addition, the presence of an elevated neutrophil count in the fluid in all but 2 of the 16 postmortem episodes is evidence for premortem entry of bacteria into the fluid with activation of complement and entry of these phagocytes into the peritoneal cavity. The 2 postmortem bacterascites episodes were monomicrobial E. coli infections; 1 involved a neutrophi1 count that was almost 250/mm3. The predominance of monomicrobial postmortem ascitic fluid infection (75%) also provides evidence against postmortem colonization. Polymicrobial postmortem ascitic fluid cultures would be expected if the fluid were colonized after death. Finally, the presence of monomicrobial infection of neutrocytic pleural fluid also provides evidence that these infections probably began before death. It would be unlikely that a single species of gut flora would be able to traverse the diaphragm and incite a neutrophil response after death. It is possible that the carbon tetrachloride contributed to ascites formation in this model by its known nephrotoxicity (17) in addition to its fibrogenic effect on the liver. This may explain why continued administration of the toxin (after ascites developed) was needed to ensure persistence of ascites. Although carbon tetrachloride-induced gut damage has not been reported (8, 9,17,18), it is also possible that oral administration of the toxin may predispose to ascitic fluid infection by damaging the gut mucosa. The absence of reports of SBP in animal models that involve inhalation or intraperitoneal administration
February 1991
of the agent lends some credence to this possibility. However, evidence of mucosal damage was not noted in this model. Further studies of the ultrastructural and physiologic effects of carbon tetrachloride on the gut are needed. It is possible that SBP develops in humans, in part, because of an increase in gut mucosal permeability related to portal hypertension and/or gut edema. Perhaps carbon tetrachloride increases the risk of peritonitis in cirrhotic rats with fluid overload also because of an enhancement of mucosal permeability. Spontaneous bacterial peritonitis and bacterascites are common complications in patients with cirrhotic ascites (l-4). The current theory of the pathogenesis of SBP is that gut or respiratory flora invade the bloodstream of susceptible patients and then enter ascites via hepatic sinusoids (7). This colonization of ascites, i.e., bacterascites, may then resolve after eradication of bacteria by host defenses or may progress to SBP and death. It is of interest that premortem bacterascites was documented in this rat model of ascitic fluid infection and that this colonization resolved in some rats and progressed in others. Further detailed studies of this model will allow investigation of the pathogenesis, prevention, and treatment of SBP and pleural fluid infection. The source of the organisms can be proven using labeled bacteria. Gut mucosal permeability can be measured serially during induction of cirrhosis and ascites formation to determine if stepwise changes occur before the development of ascitic fluid infection. Factors that predispose some but not all rats to infection can be pursued. Prophylactic and therapeutic antibiotic treatment can be investigated. Trials of bacterial vaccines, passive transfer of immunoglobulin, and immunomodulation can be performed.
RAT MODEL OF BACTERIAL PERITONITIS
493
4. Runyon BA. Spontaneous bacterial peritonitis: an explosion of information. Hepatology 1988;8:171-175. 5. Runyon BA, Morrissey R, Hoefs JC, Wyle F. Opsonic activity of human ascitic fluid: a potentially important protective mechanism against spontaneous bacterial peritonitis. Hepatology 1985;5:634-637. 6. Runyon BA, Van Epps DE. Diuresis of cirrhotic ascites increases its opsonic activity and may help prevent spontaneous bacterial peritonitis. Hepatology 1986;6:396-399. 7. Runyon BA. Patients with deficient ascitic fluid opsonic activity are predisposed to spontaneous bacterial peritonitis. Hepatology 1988;8:632-635. 8. Mellencamp MA, Preheim LC. Pneumococcal pneumonia in a rat model of cirrhosis: effects of cirrhosis on pulmonary defense mechanisms against Streptococcus pneumoniae. J Infect Dis (in press). 9. Proctor E, Chatamra K. High yield micronodular cirrhosis in the rat. Gastroenterology 1982;83:1183-1190. 10. Runyon BA, Canawati I-IN, Akriviadis EA. Optimization of ascitic fluid culture technique. Gastroenterology 1988;95:13511355. 11. Runyon BA, Hoefs JC. Culture-negative neutrocytic ascites: a variant of spontaneous bacterial peritonitis. Hepatology 1984;4: 1209-1211. bacterascites: a 12. Runyon BA. Monomicrobial nonneutrocytic variant of spontaneous bacterial peritonitis. Hepatology 1990; 12:710-715. 13. Steffen EK, Berg RD. Deitch EA. Comparison of translocation rates of various indigenous bacteria from the gastrointestinal tract to the mesenteric lymph node. J Infect Dis 1988;157:10321038. 14. Flaum MA. Spontaneous bacterial empyema in cirrhosis. Gastroenterology 1976;70:416-417. 15. Xiol X, Castellote J, Baliellas C, et al. Spontaneous bacterial empyema in cirrhotic patients: analysis of eleven cases. Hepatology 1990;11:365-370. 16. Ben-Yehuda 0, Samniah N, Guetta V, Khoury M, Sikuler E, Schlaeffer F. Spontaneous bacterial pleuritis in patients with cirrhosis (abstr). Gastroenterology 1990;98:A568. 17. Zimmerman SW, Norbach DH. Nephrotoxic effects of longterm carbon tetrachloride administration in rats. Arch Path01 Lab Med 1980;104:94-99. 18. Charbonneau M, Tuchweber B, Plaa GL. Acetone potentiation of chronic liver injury induced by administration of carbon tetrachloride. Hepatology 1986;6:684-700.
References 1. Runyon BA. Low-protein-concentration
ascitic fluid is predisposed to spontaneous bacterial peritonitis. Gastroenterology 1986;91:1343-1346. 2. Pinzello G, Simonetti RG, Craxi A, Di Piazza S, Spano C, Pagliaro L. Spontaneous bacterial peritonitis: a prospective investigation in predominantly nonalcoholic cirrhotic patients. Hepatology 1983;3:545-549. 3. Hoefs JC, Runyon BA. Spontaneous bacterial peritonitis. Dis Mon 1985:31:1-48.
Received February 13,199O. Accepted August 1,199O. Address requests for reprints to: Bruce A. Runyon, M.D., Associate Professor of Medicine, Department of Internal Medicine, GastroenterologyiI-Iepatology Division, University of Iowa Hospitals and Clinics, Iowa City, Iowa 52242. This work was supported in part by the Hastings Foundation of Los Angeles and by the United Liver Association of Los Angeles. The authors thank Francine Charboneau for technical assistance.
