Antibiotic Concentrations in Ascitic Fluid of Patients with Ascites and Bacterial Peritonitis DALE N. GERDING, M.D.; WENDELL H. HALL, M.D., Ph.D., F.A.C.P.; and ELIZABETH A. SCHIERL, B.S.; Minneapolis, Minnesota
Thirty-six paired specimens of serum and ascitic fluid from 21 patients with peritonitis and ascites, most with spontaneous bacterial peritonitis and alcoholic cirrhosis, were assayed for antibiotic content. Antibiotics assayed and number of determinations were gentamicin, 14; tobramycin, 7; ampicillin, 5; clindamycin, 3; penicillin G, 2; cephalothin, 2; chloramphenicol, 2; and cefazolin, 1. In 31 paired specimens the ascitic fluid antibiotic concentration was about one half or more of the simultaneous serum level and in 17 assays exceeded 9 0 % of the serum level. All antibiotics studied penetrated ascitic fluid equally well. Clinical response to antibiotic therapy was good in 12 of 16 patients with culture-proven bacterial peritonitis. Antibiotic levels in ascitic fluid exceeded the minimal inhibitory concentration of the infecting organisms in all but one patient who responded. Direct intraperitoneal instillation of antibiotics does not appear to be necessary routinely; however, there may be an initial lag of several hours before antibiotic concentrations in ascites achieve therapeutic levels.
tains little or no protein. Antibiotic concentrations may therefore be considerably different in it than in the large volumes of ascitic fluid observed in conditions such as cirrhosis, peritoneal tumors, and congestive heart failure. Furthermore, in patients undergoing peritoneal dialysis, serum kinetics of most antimicrobials is prolonged by renal failure, a condition that is usually absent or less severe in patients with ascites. Moreover, lowered concentrations of serum albumin in cirrhotic patients with ascites may alter the distribution of those antibiotics that are highly bound to serum proteins. We have measured antibiotic concentrations in 35 paired specimens of serum and ascitic fluid from 21 patients with ascites and peritonitis who received systemic therapy (intravenous, intramuscular, or oral) with gentamicin, tobramycin, ampicillin, penicillin, clindamycin, cephalothin, chloramphenicol, or cefazolin, and in one patient after intraperitoneal gentamicin treatment. In addition, antibiotic concentrations in ascitic fluid were compared with [1] the in-vitro minimum inhibitory concentration of the majority of infecting organisms and [2] the patient's clinical response.
W E ARE CONCERNED about the clinical adequacy of antibiotic therapy administered systemically in patients with ascites who develop peritonitis, particularly spontaneous bacterial peritonitis in cirrhosis (1-3). A previous study of the kinetics of cephalothin in serum and uninfected ascitic fluid of cirrhotic patients has suggested the use of both intravenous and intraperitoneal therapy ( 4 ) . Several studies of peritoneal transport of antibiotics in patients undergoing peritoneal dialysis have shown little or no clearance of tetracycline, colistimethate, and chloramphenicol from serum into dialysate fluids (5) and have recommended intraperitoneal instillation of gentamicin or tobramycin for the treatment of peritonitis complicating peritoneal dialysis (6, 7 ) . In another report, concentrations of methicillin and tetracycline diffusing from serum into peritoneal dialysis fluid were therapeutically adequate, but it was suggested that patients with peritonitis who are undergoing dialysis should receive intraperitoneal antibiotics ( 8 ) . Peritoneal dialysis fluid is rapidly exchanged and con-
Patients
• From the Infectious Disease Section, Department of Medicine, Veterans Administration Hospital, and University of Minnesota Medical School; Minneapolis, Minnesota.
708
Twenty-one patients were studied in a 3-year period ending September 1976 at Minneapolis Veterans Administration Hospital (18 cases) or St. Paul-Ramsey Hospital (three cases) (Table 1). In 20 cases the underlying cause of ascites was hepatic cirrhosis, 19 of which were due to alcoholism. The remaining patient (Case 4) had ascites secondary to renal failure and developed peritonitis after peritoneal dialysis. Of the 20 cirrhotic patients, three (Cases 5, 12, and 15) also had a pancreatic pseudocyst, one of which was infected (Case 15) and one of which ruptured (Case 12) and were the apparent cause of peritonitis. In one cirrhotic patient (Case 13) a perforated duodenal ulcer was the source of peritonitis, and another subject (Case 20) developed peritonitis after portacaval shunt surgery. Spontaneous bacterial peritonitis was the presumptive diagnosis in the remaining 16 patients. One of these patients (Case 11) was thought to have postnecrotic cirrhosis and chronic active hepatitis. He was receiving prednisone at the time of peritonitis. Patients ranged in age from 33 to 79 years (mean, 52 years), and 19 were men. Case 2 was previously reported elsewhere (9). The diagnosis of peritonitis was based on positive bacterial cultures of ascitic fluid in 16 patients (Table 2). In the five patients with negative cultures (Cases 5, 6, 7, 19, 21), the diagnosis was based on the presence of fever ( > 38 °C), abdominal pain, and the following clinical and ascitic fluid findings: one subject (Case 21) had Klebsiella pneumoniae bacAnnals of Internal Medicine 86:708-713, 1977
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19522/ by a University of California San Diego User on 04/10/2017
Table 1 . Antibiotic Concentrations in Serum and Ascitic Fluid
Antibiotic
Case
Antibiotic
Frequency
Creatinine
Dose
Day of
Time Since
Antibiotic
Therapy
Last Dose
Concentration
3 4 2 5 5 3 3 2 1 1 1 1 1§ 1§ 2 7 1 3 7 4 3 12 5 1 2 8 3 4 3 5 2 8 2 5 1 1§
h 30 4.0 7.5 0.75 5.0 15.5 2.5 8.5 3.0 0.5 3.75 4.5 3.0 0.25 5.0 1.0 1.25 1.0 4.5 1.0 4.0 3.5 2.0 4.0 2.0 4.0 4.0 2.0 5.5 1.0 2.5 3.0 6.5 12.0 3.0 1.0-2.0
Serum Gentamicin
Tobramycin
1 2 3 4 5 6 7 8 9
Penicillin G Clindamycin Cephalothin Cefazolin Chloramphenicol
mg/dl 4.4 1.6 1.6 4.1 0.9 2.7 0.9 1.9 1.2
10 11
80mgiv* 80mgimt 80mgiv 60mgim 60mgim 120mgim 75mgiv 120mgiv 120mgiv lOOmgiv 120mgiv lOOmgiv
(1.0)t (1.4) (1.0) (1.0) (1.2) (1.8) (0.9) (1.5) (1.5) (1.2) (1.2) (1.2)
12
80mgip|| 80 mg im
(1.0) (1.4)
8
1.1
(1.6) (1.1) (1.3) (0.9) (1.2) (27) (16) (8) (24)
12 12 12 8 8 4 4 6 6
2.7
14 15 16 17 18 7
120mgim 80mgim lOOmgim 60mgim 60mgiv 2 g iv 1 g iv 0.5 gpo** 2giv
1.2 0.9 2.9 2.3 1.0 0.9
19 10 1 5 8 20 6 21 11
1.6 giv 0.75 giv 600 mg iv 400mgiv 600mgiv 2 g iv 2 giv 1 g iv 1 g iv
(16) (6) (7) (7) (8) (29) (28) (12) (13)
4 4 6 6 6 6 6 6 6
0.9 1.0 4.4 0.9 1.9 1.8 2.7 6.4 0.7
13
Ampicillin
h 24-36 8 12 24-48 8 24 8 12 8 8 8 6
1.0 0.7 1.0
Ascites tig/ml
0.6 10.0 5.7 8.5 2.9 3.4 0.6 3.1 3.1 6.0 3.3 0.8 2.3 2.3 2.6 9.0 11.5 6.2 13.6 2.5 3.6 313 212 2.6 27 15 9.5 7.6 5.9 6.1 7.6 22.0 9.8 81 6.0 22.1
0.7 8.2 5.3 4.6 4.1 3.5 0.7 4.1 1.0 3.6 2.8 0.9 1.1 8.0 1.7 1.3 5.6 5.6 10.8 2.5 3.8 313 216 0.6 27 28 5.7 16.5 3.2 2.9 2.3 14.4 8.5 76 2.5 2.5
* iv = intravenous. t Dose in mg/kg body weight. t im = intramuscular. § Second episode. || ip = intraperitoneal. ** po = oral.
teremia and an ascitic fluid leukocyte count of 20 000/mm 3 with 8 8 % neutrophils, but ascitic fluid was cultured only after several days of antibiotic therapy; one (Case 5) had pancreatitis and pneumococcal bacteremia with 2900 leukocytes/mm* and 9 5 % neutrophils in ascitic fluid; one (Case 18) had a positive Gram's stain (Gram-positive cocci) of ascitic fluid, and ascitic leukocyte count was 10 200/mm 3 with 9 5 % neutrophils; one (Case 6) had 7500 ascitic leukocytes/mm 3 with 8 5 % neutrophils; and one (Case 7) had 1300 leukocytes/mm 3 with 87% neutrophils in the ascitic fluid. In addition, one patient (Case 11) had a second episode of culture-negative peritonitis (abdominal pain, hypotension, 8900 ascitic fluid leukocytes/mm 3 , with 9 4 % neutrophils) that occurred 16 days after documented Escherichia coli peritonitis. Criteria for a favorable response to therapy were resolution of fever and abdominal signs and symptoms, negative ascitic fluid cultures on repeat paracentesis (culture-positive patients), or reduction in ascitic fluid leukocytes and percentage of neutrophils (culture-negative patients) and survival for at least 7 days from the onset of therapy. All patients except one (Case 11) received only one course of therapy. One subject (Case 11) was retreated for a culture-negative clinical relapse
16 days after the initial episode of E. coli peritonitis. Specimens for antibiotic assay were obtained by needle aspiration of ascitic fluid or by aspiration of surgical abdominal drains (Case 12). All abdominal paracenteses were done with informed consent of the patient, and a blood specimen for assay of serum antibiotic concentration was drawn at the time of paracentesis. Paracentesis was usually done to evaluate response to therapy, but eight specimens were obtained within several hours of the first antibiotic dose solely to ascertain early antibiotic penetration. In these instances the patient was specifically informed of the experimental purpose of the procedure and gave his consent. N o attempt was made to standardize the time a specimen was obtained relative to the last dose of antibiotic. Methods Antibiotics were assayed in triplicate using a microbiologic disc-plate method with either 4- or 24-h incubation at 37 °C (10, 11). Most gentamicin assays were done with a highly antibiotic-resistant strain of K. pneumoniae (University of Minnesota, K-1296), obviating the need to inactivate other antibiotics present in the specimens ( 1 2 ) . Sensitivity of the Gerding et al. • Antibiotics in Ascitic Fluid
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19522/ by a University of California San Diego User on 04/10/2017
709
Table 2. Organisms Cultured from Ascitic Fluid, Minimum Inhibitory Concentration (MIC) of the Appropriate Antibiotic, Highest Concentration of Antibiotic in Ascitic Fluid, Response to Treatment, and Hospital Survival of 16 Patients with Culture-Proven Bacterial Peritonitis
Case
Organism
Antibiotic
1 2 3 4 8 9 11 12 13 14 15 16 17 18 10 20
Escherichia coli Pasteurella multocida E. coli Proteus morganii E. coli E. coli E. coli Multiple* Multiple! Klebsiella pneumoniae E. coli Enterococcus E. coli Enterococcus Streptococcus pneumoniae Staphylococcus aureus
Gentamicin Gentamicin Gentamicin Gentamicin Gentamicin Gentamicin Gentamicin Tobramycin Tobramycin Tobramycin Tobramycin Ampicillin Ampicillin Ampicillin Penicillin G Cephalothin
MIC ng/ml
... 1.0 2.0 4.0 ... 0.5 0.25 0.5-128 0.5-> 128 1.0 2.0 2.0 2.0 ... ... ...
Ascitic Fluid Concentration
Favorable Response
Survived Hospitalization
Yes Yes No Yes Yes Yes Yes Yes No Yes No Yes Yes Yes Yes No
No No No Yes Yes No No Yes No No No No No Yes Yes No
fjig/ml
0.7 8.2 5.3 4.6 4.1 3.6 1.1 1.7 10.8 2.5 3.8 313 216 0.6 16.5 14.4
* Pseudomonas aeruginosa, E. coli, Proteus species, Enterococcus. t P. aeruginosa, E. coli, P. morganii, Enterococcus, K. pneumoniae.
assay was increased by reduction of the inoculum and incubation for 24 h. This organism is also resistant to neomycin, which may be present in the serum of patients with hepatic and renal failure who are receiving it orally. Tobramycin and several gentamicin assays were done with Bacillus subtilis ATCC 6633 or a multiresistant strain of Citrobacter as the test organism. Any penicillins or cephalosporins present in these specimens were inactivated by the addition of either penicillinase (200 000 units/0.8 ml) or cephalosporinase (13). Penicillin G, ampicillin, cephalothin, and cefazolin were assayed with either B. subtilis ATCC 6633 or Staphylococcus aureus ATCC 6538P. The latter organism was used in the presence of aminoglycosides with the pH adjusted to 5.5 to inactivate the aminoglycoside according to the method described by Sabath and associates (14). Clindamycin was assayed with Sarcina lutea ATCC 9341. In the presence of gentamicin, clindamycin was separated from the specimen using a double chloroform extraction (two parts chloroform to one part specimen). The chloroform fraction was evaporated and resuspended in 0.1 M phosphate buffer and the assay done in the usual manner using phosphate buffer to prepare the standard curve*. For each antibiotic, standard curves were initially prepared in both pooled human serum and pooled human ascitic fluid. No differences were noted in the two standards for aminoglycosides; therefore, all subsequent ascitic fluid and serum specimens were assayed against a standard curve prepared in human serum only. For ampicillin, penicillin G, cephalothin, and cefazolin, the serum and ascitic fluid standard curves differed significantly, and assays were done either by using a separate ascitic fluid standard curve or by diluting the ascitic fluid specimen at least threefold in pooled human serum and using a serum standard curve. The test bacteria used in microbiologic assays were not inhibited by pooled serum or ascitic fluid in the absence of antibiotics. Chloramphenicol was assayed using an enzymologic microassay technique (15). Minimum inhibitory concentration for each antibiotic and infective bacterial strain was measured in Mueller-Hinton broth (16). Results
Concentrations of the antibiotics studied in paired serum and ascitic fluid are shown in Table 1. In seven patients assays of two different drugs were done. Simultaneous * LUMMIS W, The Upjohn Company, Kalamazoo, Michigan, personal communication. 710
administration of more than one antibiotic did not have any detectable effect on ascitic fluid concentrations (Table 1). In general, antibiotic concentrations in ascitic fluid exceeded half of the simultaneous serum level for all antibiotics measured, and about half of the ascitic fluid concentrations were over 90% of the simultaneous serum concentration. Antibiotic concentrations in ascitic fluid less than half of the serum concentration occurred in only five patients (Cases 8, 9, 11, 12, 17). In Case 9, gentamicin measured 3 h after the first dose was only 1.0 /xg/ml in ascitic fluid (30% of serum concentration), but Vi h after the third dose (given every 8 h) the concentration had risen to 3.6 fig/ml (60% of simultaneous serum concentration). Chloramphenicol level also measured after the first dose (second episode) in Case 11 was 2.5 fig/ml in ascitic fluid between 1 and 2 h after intravenous therapy, while simultaneous serum level was 22.1 /xg/ml. The same dose of chloramphenicol in the same patient resulted in levels of 6.0 fig/ml and 2.5 fig/ml in serum and ascites, respectively, 3 h after the initial dose in the first episode of peritonitis. In Case 12, tobramycin concentration on Day 7 of therapy was 1.3 fig/ml 1 h after a dose, which was only 14% of the serum concentration. However, there was only one patient (Case 12) in whom specimens were obtained from abdominal drains, and the specimens may not have been in communication with free ascitic fluid. Ampicillin was measured in Case 18 4 h after the first dose, and the concentration in ascitic fluid was only 2 3 % of the serum level. This patient was essentially asymptomatic, but ascitic fluid culture yielded Enterococcus organisms and the ascitic fluid inflammatory response (400 leukocytes/mm 3 , 35% neutrophils) was the lowest of any patient studied. One patient (Case 8) had a clindamycin concentration of 2.3 fig/ml in ascitic fluid 2.5 h after his seventh dose (Day 2 of treatment), the only unexplained assay below half of the serum concentration. As expected, serum concentrations of aminoglycosides usually exceeded ascitic fluid concentrations in specimens
June 1977 • Annals of Internal Medicine • Volume 86 • Number 6
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19522/ by a University of California San Diego User on 04/10/2017
obtained within 2 h of a dose, reflecting slow distribution of the drug from the vascular compartment to ascitic fluid. Beyond 2 h after a dose, ascitic fluid and serum concentrations tended to be comparable, with ascitic fluid concentrations occasionally higher than serum. Ascitic fluid concentrations of ampicillin and penicillin G obtained 2 h after a dose (Cases 7, 10, 16) were equal to or in excess of simultaneous serum specimens, possibly a reflection of their shorter serum half-life as compared to that of the aminoglycosides. With the exception of Case 8 and the second episode of Case 11, ascitic concentrations of clindamycin, chloramphenicol, and cefazolin appeared to be comparable to serum concentrations in the few patients studied. Although serum creatinine concentration was elevated ( > 1.4 mg/dl) in 12 patients (Table 1), antibiotic penetrance into ascitic fluid did not appear to be altered. The dosage interval for aminoglycosides was adjusted to compensate for decreased renal function, but for ampicillin (Cases 15 and 16) and cefazolin (Case 20) no dosage adjustment was made and concentrations in both serum and ascitic fluid were quite high. A favorable response to therapy occurred in 12 of 16 patients with culture-proven bacterial peritonitis (Table 2) and in all five patients with negative culture. Of the four patients who failed to respond, one (Case 3) was thought to be in septic shock, one (Case 13) had a perforated peptic ulcer that was not surgically closed, one (Case 15) had an infected pancreatic pseudocyst, and one (Case 20) had staphylococcal peritonitis after portacaval shunt surgery. In 11 of the 16 patients with culture-proven bacterial peritonitis, the minimum inhibitory concentration of the infecting organisms was measured and compared to the highest concentration of antibiotic found in the patient's ascitic fluid (Table 2 ) . In all but one patient (Case 12, mixed infection) with a favorable response to therapy, the antibiotic concentration in the ascitic fluid exceeded the minimum inhibitory concentration of the infecting organism. However, in three of the four patients who did not respond to therapy, the antibiotic concentration in the ascitic fluid was also in excess of the minimum inhibitory concentration. Obviously the presence of appropriate antibiotics in the infected area does not necessarily insure a favorable therapeutic response. Despite a favorable response to therapy and resolution of the acute infection, the ultimate prognosis remains poor. Eleven of the 16 patients with culture-proven bacterial peritonitis died (Table 2 ) , and two of the five patients whose ascitic fluid cultures were negative also died. Most patients who recovered from their acute infection died from progressive hepatic and renal decompensation or complications of alcoholism and cirrhosis (gastrointestinal bleeding and hepatoma). Death in these patients occurred from 8 to 86 days after admission. The cause of death in those patients failing to respond to antibiotic therapy was directly related to the infection. Discussion
Direct intraperitoneal instillation of antibiotics has been
recommended for the treatment of bacterial contamination of the peritoneal cavity by a number of surgeons (17-23). Both intraperitoneal antibiotic lavage and instillation of kanamycin (17), kanamycin-bacitracin (18), ampicillin (19), cephalothin (20), tetracycline (21), cephaloridine (22), and gentamicin (23) have received favorable reports for reducing morbidity or mortality. However, these studies are compromised by the concomitant use of other systemic antibiotics (17, 18, 20, 23), use of retrospective controls (19, 20, 23), or investigation limited to laboratory animals (21). Fowler (22) has shown a significant reduction in the incidence of intraperitoneal abscesses in children with peritonitis by the use of intraperitoneal cephaloridine for 48 h. In contrast, Rambo (24) was unable to find any benefit from peritoneal lavage with cephalothin compared to saline solution in patients with peritonitis. A later study of experimental E. coli peritonitis in dogs showed no difference in therapeutic benefit between intravenous and intraperitoneal cephalothin (25). Studies of the concentration of antibiotics in peritoneal fluid after systemic therapy are limited. In an early study of experimental peritonitis in dogs, the peritoneal concentration of sulfadiazine was found to be the same as in blood, while peritoneal fluid concentrations of streptomycin and penicillin were 50% of blood levels (26). Concentrations of cephaloridine (22) in man and kanamycin (27) in dogs after systemic therapy appear to be about 20% of peak serum concentrations after a single dose. Studies in dogs with experimental ascites have shown ascitic fluid peak antibiotic concentrations ranging from 5.8% to 6 5 % of peak serum concentrations after a parenteral dose (28). In general, we found that concentrations of antibiotics in ascitic fluid are adequate in patients with bacterial peritonitis who receive only systemic therapy. Although five of the patients studied had negative bacterial cultures of ascitic fluid, we believe the clinical presentation, cellular inflammatory response in ascitic fluid, and response to antibiotic therapy justify their inclusion in this series. Ascitic fluid leukocyte and differential counts in alcoholic cirrhosis do not always clearly separate infected from uninfected patients (29). However, the leukocyte responses in our group of patients with negative cultures were well above the mean for uninfected patients, and most were above the highest recorded normal values (29, 30). Ascitic fluid leukocyte counts in the culture-negative group were no different than in our culture-positive patients, and there was no difference in the ascitic fluid antibiotic levels in the two groups. Antibiotic concentrations in patients with ascites and bacterial peritonitis are therapeutically adequate in the ascitic fluid, somewhat analogous to the situation in septic joint effusions where direct instillation of antibiotic has not been necessary (31). Although not all of our patients had a favorable response to antibiotic therapy (Table 2 ) , those who failed to respond were severely ill and usually had additional complications (shock, infected pseudocyst, perforated viscus, abdominal surgery). We believe that systemic therapy is adequate in this clinical setting but ascitic fluid concentrations of the aminoglycosides may be very near the minimum inhibitory concentration of infectGerding et a\. • Antibiotics in Ascitic Fluid
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19522/ by a University of California San Diego User on 04/10/2017
711
ing Gram-negative bacteria (Table 2 ) . Except for the sharp serum peaks, ascitic fluid levels parallel serum concentrations. Therefore it would seem advisable to assay aminoglycosides in serum (and also ascitic fluid if possible) to be certain that the prescribed dosage has resulted in therapeutic levels. The best time to draw a serum specimen appears to be at about 2 to 4 h after a dose, as serum levels at that time are most likely to reflect ascitic fluid levels (Table 1). Ascitic fluid antibiotic concentrations in an uninfected dog model have been shown to increase slowly after single intramuscular injections. Peak ascitic fluid level did not occur until 4 h after administration of most drugs, and mean peak ascites concentration ranged from 1 1 % to 27% of peak serum level for aminoglycosides, cephalosporins, and clindamycin ( 2 8 ) . Peak ascites and serum levels of antibiotics were not measured in our present study; however, seven paired specimens were obtained within 4.5 h of the first antibiotic dose, and they tend to confirm in humans with peritonitis the observations of low ascites levels in uninfected dogs. Gentamicin concentration in ascitic fluid was low in three of four specimens obtained after the first dose in three patients (Case 11 was treated twice). Ascitic fluid chloramphenicol was also measured twice after the initial dose in Case 11 and was 2.5 /xg/ml on both occasions. Ampicillin ascitic level after the first dose in Case 18 was low (0.6 fig/ml). In the uninfected ascitic dog model, the concentration of antibiotic in ascites has been shown to depend on the serum antibiotic level and the ascitic fluid volume (28). In the patient with peritonitis, peritoneal inflammation may also alter ascitic fluid penetrance as has been shown in meningitis. Three measures of ascitic inflammatory response—leukocyte count, percentage of neutrophils, and total protein— were compared with ascitic fluid gentamicin level to ascertain the possible influence of inflammation. The comparison was restricted to those cases in which gentamicin was measured after the first dose (Cases 9, 10, and 11) because multiple doses appear to raise ascitic fluid levels considerably (Table 1). There did not appear to be any relation of gentamicin ascitic fluid level to leukocyte count, neutrophil percentage, or ascitic fluid protein in this small group of patients; in particular, one patient (Case 10) (5100 leukocytes/mm 3 , 87% neutrophils, 0.4 g/dl protein) did not show evidence of greater inflammation than the others. Based on dog studies (28), we suspect ascitic fluid volume differences are the major determinant of ascitic fluid levels, but ascites volume was not measured in these patients. Wilson, Chalmers, and Madoff (4) were unable to correlate ascitic fluid cephalothin levels with ascites volume but commented that the lowest ascitic fluid levels were seen in the patient with the greatest amount of ascites. Delay in ascitic fluid penetration of the initial dose of systemic antibiotic may be important clinically if the patient is severely ill. Although routine intraperitoneal instillation of antibiotics is not necessary, it may be justified as an initial route of administration in a critically ill patient. We have administered gentamicin once by direct intraperitoneal instillation in a patient (Case 11), who presented clinical signs of septic shock with apparent relapse of E. Coli peri712
tonitis. The patient received 80 mg of gentamicin intraperitoneally 3 h after the first intravenous dose of 100 mg. Ascitic fluid gentamicin level was 1.1 /xg/ml before intraperitoneal instillation and 8.0 jug/ml 15 minutes later (Table 1). Serum gentamicin level remained constant at 2.3 fig/ml. No toxicity was noted, but clinical benefit could not be ascertained. Although it appears that systemic gentamicin doses (1 to 2 mg/kg) can be safely administered intraperitoneally, comprehensive studies of this procedure to evaluate possible toxicity and clinical efficacy in peritonitis associated with ascites have not been done. Optimal choice of antibiotics in the empiric treatment of peritonitis associated with ascites is not clear and cannot be ascertained from this study. The most frequent infecting bacteria in spontaneous bacterial peritonitis and cirrhosis have been Gram-negative aerobic enteric bacilli and Grampositive aerobic cocci, including pneumococci and enterococci (1-3). Because of the excellent Gram-negative spectrum of the newer aminoglycosides (gentamicin, tobramycin, amikacin), we favor use of one of these drugs plus penicillin G or ampicillin for Gram-positive bacteria in empiric therapy. Although clinical efficacy of the aminoglycosides appears to be good (Table 2 ) , this is an uncontrolled observation. We are concerned about aminoglycoside use in this setting for two reasons: the minimum inhibitory concentration of Gram-negative bacilli may be very near the achievable levels of aminoglycosides in ascitic fluid (Table 2 ) ; and patients frequently have renal failure at the initiation of treatment (Table 1) and often show deterioration of renal function while receiving aminoglycosides. We have not established a causative role of aminoglycosides in this deterioration. Whether this data on antibiotic concentrations in ascitic fluid in this group of patients can be extrapolated to patients with peritonitis and no detectable ascites is not known. Our studies in the dog suggest that antibiotic concentrations are likely to be higher if ascitic volume is small, presumably because of a lesser dilutional factor (28). Bacterial contamination may be much greater in surgical peritonitis, particularly if a viscus or abscess has ruptured, and direct antibiotic installations and irrigations may therefore be superior to parenteral administration, but this is unproved. Certainly the bacterial spectrum is different from spontaneous bacterial peritonitis in cirrhosis and probably requires a different spectrum of antibiotics, perticularly therapy for Bacteroides fragilis. ACKNOWLEDGMENTS: The authors thank the interns, residents, and fellows of the participating hospitals for their case contributions and assistance in obtaining specimens for antibiotic assay; and Richard H. Lyon, Ph.D., for chloramphenicol determinations. This paper was presented in part at the Fifteenth Interscience Conference on Antimicrobial Agents and Chemotherapy, Washington, D.C., 24-26 September 1975. Grant support: in part by a grant from Eli Lilly and Company, Indianapolis, Indiana. Received 13 December 1976; revision accepted 4 March 1977. • Requests for reprints should be addressed to Dale N. Gerding, M.D.; Infectious Disease Section, Veterans Administration Hospital, 54th St. and 48th Av. South; Minneapolis, MN 55417. References 1. CONN HO, FESSEL JM:
Spontaneous bacterial peritonitis in
cirrhosis: variations on a theme. Medicine {Baltimore) 50:161197, 1971
June 1977 • Annals of Internal Medicine • Volume 86 • Number 6
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19522/ by a University of California San Diego User on 04/10/2017
2. C U R R Y N , M C C A L L U M
RW, G U T H P H :
Spontaneous
peritonitis
in cirrhotic ascites. Am J Dig Dis 19:685-692, 1974 3. CORREIA J P , C O N N H O : Spontaneous bacterial peritonitis in cirrhosis: endemic or epidemic? Med Clin North Am 59:963981, 1975 4. W I L S O N
DE,
CHALMERS
TC,
MADOFF
MA:
The
passage
of
Manual of Clinical Microbiology, 2 n d ed., edited by L E N N E T T E E N , SPAULDING E H , TRUANT J P . Washington, D.C., American Society for Microbiology, 1974, pp. 410-417 17. D I V I N C E N T I F C , C O H N I J R : Prolonged administration of intraperitoneal kanamycin in the treatment of peritonitis. Am Surg 37:177-180, 1971
cephalothin into a n d o u t of ascitic fluid. Am J Med Sci 253:449452, 1967 5. GREENBERG P A , SANFORD J P : Removal and absorption of antibiotics in patients with renal failure undergoing peritoneal dialysis. Ann Intern Med 66:465-479, 1967
tion of peritoneal irrigation with antibiotic solution. Surgery 62:73-78, 1967 19. A U N E S, N O R M A N N E : Diffuse peritonitis treated with continuous peritoneal lavage. Acta Chir Scand 136:401-404, 1970
6. SMITHIVAS T, H Y A M S PJ, M A T A L O N R, et a l : T h e use of
20.
gen-
tamicin in peritoneal dialysis. I. Pharmacologic results. J Dis 124:S77-S83, 1971 7. W E I N S T E I N
AJ,
KARCHMER
AW,
MOELLERING
RC
Tobramycin concentrations during peritoneal dialysis. Agents Chemother 4:432-434, 1973 8. SHEAR L, SHINABERGER J H , BARRY K G :
Infect
J R , et
al:
Antimicrob
Peritoneal t r a n s p o r t of
antibiotics in man. N Engl J Med 272:666-669, 1965 9. GERDING D N , K H A N M Y , E W I N G J W , et al: Pasteurella
SABATH L D , CASEY J I , R U C H
P A , et al: R a p i d microassay
circulating nephrotoxic antibiotics. mother-1970:$3-S9, 1971 12.
LUND
ME,
BLAZEVIC D J , M A T S E N
Antimicrob JM:
Rapid
Agents
for
Che-
gentamicin
bio-
21.
26.
29.
Rapid
microassay
of
15.
LIETMAN
PS,
WHITE
JJ,
SHAW
WV:
Chloramphenicol:
an
enzymological microassay. Antimicrob Agents Chemother 10: 347-353, 1976 16. WASHINGTON J A II, BARRY A L : Dilution test procedures, in
CARIDIS D T , GADDIE J, M A T H E S O N
NA:
lavage
Continuous
perito-
using
peritoneal
HUNT
J A,
RIVLIN
ME,
CLAREBOUT
HJ:
Antibiotic
peritoneal
ROTHENBERG
S, SILVANI
H,
CHESTER
S, et al:
Comparison
of
the efficacy of therapeutic agents in the treatment of experimentally induced diffuse peritonitis of intestinal origin. Ann Surg 128:1148-1163, 1948 27. C O H N I J R : Intestinal antisepsis. Springfield, Illinois, Charles C Thomas, 1968, p . 144
14.
P A , et al:
T r e a t m e n t of
lavage in severe peritonitis. South Am Med J 49:233-238, 1975 24. R A M B O W M : Irrigation of the peritoneal cavity with cephalothin. Am J Surg 123:192-195, 1972 25. SHARBAUGH R J , R A M B O W M : Cephalothin and peritoneal lavage in the treatment of experimental peritonitis. Surg Gynecol Obstet 139:211-214, 1974
28.
SABATH L D , CASEY J I, R U C H
PELOSO OA, FLOYD VT, WILKINSON L H :
lavage in peritonitis. Eur Surg Res 1:142-146, 1969 22. F O W L E R R : A controlled trial of intraperitoneal cephaloridine administration in peritonitis. / Pediatr Surg 10:43-50, 1975
assay using a multiple-antibiotic-resistant strain of Klebsiella pneumoniae. Antimicrob Agents Chemother 4:569-573, 1973 13. STROY SA, PRESTON D A : Specific assay of aminoglycosidic- or polymyxin-type antibiotics present in h u m a n sera in combination with cephalosporins. Appl Microbiol 21:1002-1006, 1971 gentamicin, kanamycin, neomycin, streptomycin and vancomycin in serum or plasma. J Lab Clin Med 78:457-463, 1971
N O O N G P , BEALL A C J R , JORDAN G L J R , et a l : Clinical evalua-
nitis with continuous postoperative peritoneal cephalothin. Am J Surg 126:742-747, 1973
23. multo-
cida peritonitis in hepatic cirrhosis with ascites. Gastroenterology 70:413-415, 1976 10. LILLY RESEARCH LABORATORIES: Paper disc or cup plate antibiotic assay procedures. Indianapolis, Indiana, Eli Lilly and Co., 1970 11.
18.
GERDING
DN,
KROMHOUT
JP,
SULLIVAN
JJ,
penetrance of ascitic fluid in dogs. Antimicrob 10:850-855, 1976 KLINE MM, MCCALLUM
et
al:
Agents
Antibiotic
Chemother
R W , G U T H P H : T h e clinical value of
ascitic fluid culture a n d leukocyte count studies in alcoholic cirrhosis. Gastroenterology 70:408-412, 1976 30. C O N N H O : Spontaneous bacterial peritonitis. Gastroenterology 70:455-457, 1976 31. N E L S O N J D : Antibiotic concentrations in septic joint effusion. N Engl J Med 284:349-353, 1971
Gerding et al. • Antibiotics in Ascitic Fluid
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19522/ by a University of California San Diego User on 04/10/2017
713
Thirty-six paired specimens of serum and ascitic fluid from 21 patients with peritonitis and ascites, most with spontaneous bacterial peritonitis and alcoholic cirrhosis, were assayed for antibiotic content. Antibiotics assayed and number of determinations were gentamicin, 14; tobramycin, 7; ampicillin, 5; clindamycin, 3; penicillin G, 2; cephalothin, 2; chloramphenicol, 2; and cefazolin, 1. In 31 paired specimens the ascitic fluid antibiotic concentration was about one half or more of the simultaneous serum level and in 17 assays exceeded 9 0 % of the serum level. All antibiotics studied penetrated ascitic fluid equally well. Clinical response to antibiotic therapy was good in 12 of 16 patients with culture-proven bacterial peritonitis. Antibiotic levels in ascitic fluid exceeded the minimal inhibitory concentration of the infecting organisms in all but one patient who responded. Direct intraperitoneal instillation of antibiotics does not appear to be necessary routinely; however, there may be an initial lag of several hours before antibiotic concentrations in ascites achieve therapeutic levels.
tains little or no protein. Antibiotic concentrations may therefore be considerably different in it than in the large volumes of ascitic fluid observed in conditions such as cirrhosis, peritoneal tumors, and congestive heart failure. Furthermore, in patients undergoing peritoneal dialysis, serum kinetics of most antimicrobials is prolonged by renal failure, a condition that is usually absent or less severe in patients with ascites. Moreover, lowered concentrations of serum albumin in cirrhotic patients with ascites may alter the distribution of those antibiotics that are highly bound to serum proteins. We have measured antibiotic concentrations in 35 paired specimens of serum and ascitic fluid from 21 patients with ascites and peritonitis who received systemic therapy (intravenous, intramuscular, or oral) with gentamicin, tobramycin, ampicillin, penicillin, clindamycin, cephalothin, chloramphenicol, or cefazolin, and in one patient after intraperitoneal gentamicin treatment. In addition, antibiotic concentrations in ascitic fluid were compared with [1] the in-vitro minimum inhibitory concentration of the majority of infecting organisms and [2] the patient's clinical response.
W E ARE CONCERNED about the clinical adequacy of antibiotic therapy administered systemically in patients with ascites who develop peritonitis, particularly spontaneous bacterial peritonitis in cirrhosis (1-3). A previous study of the kinetics of cephalothin in serum and uninfected ascitic fluid of cirrhotic patients has suggested the use of both intravenous and intraperitoneal therapy ( 4 ) . Several studies of peritoneal transport of antibiotics in patients undergoing peritoneal dialysis have shown little or no clearance of tetracycline, colistimethate, and chloramphenicol from serum into dialysate fluids (5) and have recommended intraperitoneal instillation of gentamicin or tobramycin for the treatment of peritonitis complicating peritoneal dialysis (6, 7 ) . In another report, concentrations of methicillin and tetracycline diffusing from serum into peritoneal dialysis fluid were therapeutically adequate, but it was suggested that patients with peritonitis who are undergoing dialysis should receive intraperitoneal antibiotics ( 8 ) . Peritoneal dialysis fluid is rapidly exchanged and con-
Patients
• From the Infectious Disease Section, Department of Medicine, Veterans Administration Hospital, and University of Minnesota Medical School; Minneapolis, Minnesota.
708
Twenty-one patients were studied in a 3-year period ending September 1976 at Minneapolis Veterans Administration Hospital (18 cases) or St. Paul-Ramsey Hospital (three cases) (Table 1). In 20 cases the underlying cause of ascites was hepatic cirrhosis, 19 of which were due to alcoholism. The remaining patient (Case 4) had ascites secondary to renal failure and developed peritonitis after peritoneal dialysis. Of the 20 cirrhotic patients, three (Cases 5, 12, and 15) also had a pancreatic pseudocyst, one of which was infected (Case 15) and one of which ruptured (Case 12) and were the apparent cause of peritonitis. In one cirrhotic patient (Case 13) a perforated duodenal ulcer was the source of peritonitis, and another subject (Case 20) developed peritonitis after portacaval shunt surgery. Spontaneous bacterial peritonitis was the presumptive diagnosis in the remaining 16 patients. One of these patients (Case 11) was thought to have postnecrotic cirrhosis and chronic active hepatitis. He was receiving prednisone at the time of peritonitis. Patients ranged in age from 33 to 79 years (mean, 52 years), and 19 were men. Case 2 was previously reported elsewhere (9). The diagnosis of peritonitis was based on positive bacterial cultures of ascitic fluid in 16 patients (Table 2). In the five patients with negative cultures (Cases 5, 6, 7, 19, 21), the diagnosis was based on the presence of fever ( > 38 °C), abdominal pain, and the following clinical and ascitic fluid findings: one subject (Case 21) had Klebsiella pneumoniae bacAnnals of Internal Medicine 86:708-713, 1977
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19522/ by a University of California San Diego User on 04/10/2017
Table 1 . Antibiotic Concentrations in Serum and Ascitic Fluid
Antibiotic
Case
Antibiotic
Frequency
Creatinine
Dose
Day of
Time Since
Antibiotic
Therapy
Last Dose
Concentration
3 4 2 5 5 3 3 2 1 1 1 1 1§ 1§ 2 7 1 3 7 4 3 12 5 1 2 8 3 4 3 5 2 8 2 5 1 1§
h 30 4.0 7.5 0.75 5.0 15.5 2.5 8.5 3.0 0.5 3.75 4.5 3.0 0.25 5.0 1.0 1.25 1.0 4.5 1.0 4.0 3.5 2.0 4.0 2.0 4.0 4.0 2.0 5.5 1.0 2.5 3.0 6.5 12.0 3.0 1.0-2.0
Serum Gentamicin
Tobramycin
1 2 3 4 5 6 7 8 9
Penicillin G Clindamycin Cephalothin Cefazolin Chloramphenicol
mg/dl 4.4 1.6 1.6 4.1 0.9 2.7 0.9 1.9 1.2
10 11
80mgiv* 80mgimt 80mgiv 60mgim 60mgim 120mgim 75mgiv 120mgiv 120mgiv lOOmgiv 120mgiv lOOmgiv
(1.0)t (1.4) (1.0) (1.0) (1.2) (1.8) (0.9) (1.5) (1.5) (1.2) (1.2) (1.2)
12
80mgip|| 80 mg im
(1.0) (1.4)
8
1.1
(1.6) (1.1) (1.3) (0.9) (1.2) (27) (16) (8) (24)
12 12 12 8 8 4 4 6 6
2.7
14 15 16 17 18 7
120mgim 80mgim lOOmgim 60mgim 60mgiv 2 g iv 1 g iv 0.5 gpo** 2giv
1.2 0.9 2.9 2.3 1.0 0.9
19 10 1 5 8 20 6 21 11
1.6 giv 0.75 giv 600 mg iv 400mgiv 600mgiv 2 g iv 2 giv 1 g iv 1 g iv
(16) (6) (7) (7) (8) (29) (28) (12) (13)
4 4 6 6 6 6 6 6 6
0.9 1.0 4.4 0.9 1.9 1.8 2.7 6.4 0.7
13
Ampicillin
h 24-36 8 12 24-48 8 24 8 12 8 8 8 6
1.0 0.7 1.0
Ascites tig/ml
0.6 10.0 5.7 8.5 2.9 3.4 0.6 3.1 3.1 6.0 3.3 0.8 2.3 2.3 2.6 9.0 11.5 6.2 13.6 2.5 3.6 313 212 2.6 27 15 9.5 7.6 5.9 6.1 7.6 22.0 9.8 81 6.0 22.1
0.7 8.2 5.3 4.6 4.1 3.5 0.7 4.1 1.0 3.6 2.8 0.9 1.1 8.0 1.7 1.3 5.6 5.6 10.8 2.5 3.8 313 216 0.6 27 28 5.7 16.5 3.2 2.9 2.3 14.4 8.5 76 2.5 2.5
* iv = intravenous. t Dose in mg/kg body weight. t im = intramuscular. § Second episode. || ip = intraperitoneal. ** po = oral.
teremia and an ascitic fluid leukocyte count of 20 000/mm 3 with 8 8 % neutrophils, but ascitic fluid was cultured only after several days of antibiotic therapy; one (Case 5) had pancreatitis and pneumococcal bacteremia with 2900 leukocytes/mm* and 9 5 % neutrophils in ascitic fluid; one (Case 18) had a positive Gram's stain (Gram-positive cocci) of ascitic fluid, and ascitic leukocyte count was 10 200/mm 3 with 9 5 % neutrophils; one (Case 6) had 7500 ascitic leukocytes/mm 3 with 8 5 % neutrophils; and one (Case 7) had 1300 leukocytes/mm 3 with 87% neutrophils in the ascitic fluid. In addition, one patient (Case 11) had a second episode of culture-negative peritonitis (abdominal pain, hypotension, 8900 ascitic fluid leukocytes/mm 3 , with 9 4 % neutrophils) that occurred 16 days after documented Escherichia coli peritonitis. Criteria for a favorable response to therapy were resolution of fever and abdominal signs and symptoms, negative ascitic fluid cultures on repeat paracentesis (culture-positive patients), or reduction in ascitic fluid leukocytes and percentage of neutrophils (culture-negative patients) and survival for at least 7 days from the onset of therapy. All patients except one (Case 11) received only one course of therapy. One subject (Case 11) was retreated for a culture-negative clinical relapse
16 days after the initial episode of E. coli peritonitis. Specimens for antibiotic assay were obtained by needle aspiration of ascitic fluid or by aspiration of surgical abdominal drains (Case 12). All abdominal paracenteses were done with informed consent of the patient, and a blood specimen for assay of serum antibiotic concentration was drawn at the time of paracentesis. Paracentesis was usually done to evaluate response to therapy, but eight specimens were obtained within several hours of the first antibiotic dose solely to ascertain early antibiotic penetration. In these instances the patient was specifically informed of the experimental purpose of the procedure and gave his consent. N o attempt was made to standardize the time a specimen was obtained relative to the last dose of antibiotic. Methods Antibiotics were assayed in triplicate using a microbiologic disc-plate method with either 4- or 24-h incubation at 37 °C (10, 11). Most gentamicin assays were done with a highly antibiotic-resistant strain of K. pneumoniae (University of Minnesota, K-1296), obviating the need to inactivate other antibiotics present in the specimens ( 1 2 ) . Sensitivity of the Gerding et al. • Antibiotics in Ascitic Fluid
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19522/ by a University of California San Diego User on 04/10/2017
709
Table 2. Organisms Cultured from Ascitic Fluid, Minimum Inhibitory Concentration (MIC) of the Appropriate Antibiotic, Highest Concentration of Antibiotic in Ascitic Fluid, Response to Treatment, and Hospital Survival of 16 Patients with Culture-Proven Bacterial Peritonitis
Case
Organism
Antibiotic
1 2 3 4 8 9 11 12 13 14 15 16 17 18 10 20
Escherichia coli Pasteurella multocida E. coli Proteus morganii E. coli E. coli E. coli Multiple* Multiple! Klebsiella pneumoniae E. coli Enterococcus E. coli Enterococcus Streptococcus pneumoniae Staphylococcus aureus
Gentamicin Gentamicin Gentamicin Gentamicin Gentamicin Gentamicin Gentamicin Tobramycin Tobramycin Tobramycin Tobramycin Ampicillin Ampicillin Ampicillin Penicillin G Cephalothin
MIC ng/ml
... 1.0 2.0 4.0 ... 0.5 0.25 0.5-128 0.5-> 128 1.0 2.0 2.0 2.0 ... ... ...
Ascitic Fluid Concentration
Favorable Response
Survived Hospitalization
Yes Yes No Yes Yes Yes Yes Yes No Yes No Yes Yes Yes Yes No
No No No Yes Yes No No Yes No No No No No Yes Yes No
fjig/ml
0.7 8.2 5.3 4.6 4.1 3.6 1.1 1.7 10.8 2.5 3.8 313 216 0.6 16.5 14.4
* Pseudomonas aeruginosa, E. coli, Proteus species, Enterococcus. t P. aeruginosa, E. coli, P. morganii, Enterococcus, K. pneumoniae.
assay was increased by reduction of the inoculum and incubation for 24 h. This organism is also resistant to neomycin, which may be present in the serum of patients with hepatic and renal failure who are receiving it orally. Tobramycin and several gentamicin assays were done with Bacillus subtilis ATCC 6633 or a multiresistant strain of Citrobacter as the test organism. Any penicillins or cephalosporins present in these specimens were inactivated by the addition of either penicillinase (200 000 units/0.8 ml) or cephalosporinase (13). Penicillin G, ampicillin, cephalothin, and cefazolin were assayed with either B. subtilis ATCC 6633 or Staphylococcus aureus ATCC 6538P. The latter organism was used in the presence of aminoglycosides with the pH adjusted to 5.5 to inactivate the aminoglycoside according to the method described by Sabath and associates (14). Clindamycin was assayed with Sarcina lutea ATCC 9341. In the presence of gentamicin, clindamycin was separated from the specimen using a double chloroform extraction (two parts chloroform to one part specimen). The chloroform fraction was evaporated and resuspended in 0.1 M phosphate buffer and the assay done in the usual manner using phosphate buffer to prepare the standard curve*. For each antibiotic, standard curves were initially prepared in both pooled human serum and pooled human ascitic fluid. No differences were noted in the two standards for aminoglycosides; therefore, all subsequent ascitic fluid and serum specimens were assayed against a standard curve prepared in human serum only. For ampicillin, penicillin G, cephalothin, and cefazolin, the serum and ascitic fluid standard curves differed significantly, and assays were done either by using a separate ascitic fluid standard curve or by diluting the ascitic fluid specimen at least threefold in pooled human serum and using a serum standard curve. The test bacteria used in microbiologic assays were not inhibited by pooled serum or ascitic fluid in the absence of antibiotics. Chloramphenicol was assayed using an enzymologic microassay technique (15). Minimum inhibitory concentration for each antibiotic and infective bacterial strain was measured in Mueller-Hinton broth (16). Results
Concentrations of the antibiotics studied in paired serum and ascitic fluid are shown in Table 1. In seven patients assays of two different drugs were done. Simultaneous * LUMMIS W, The Upjohn Company, Kalamazoo, Michigan, personal communication. 710
administration of more than one antibiotic did not have any detectable effect on ascitic fluid concentrations (Table 1). In general, antibiotic concentrations in ascitic fluid exceeded half of the simultaneous serum level for all antibiotics measured, and about half of the ascitic fluid concentrations were over 90% of the simultaneous serum concentration. Antibiotic concentrations in ascitic fluid less than half of the serum concentration occurred in only five patients (Cases 8, 9, 11, 12, 17). In Case 9, gentamicin measured 3 h after the first dose was only 1.0 /xg/ml in ascitic fluid (30% of serum concentration), but Vi h after the third dose (given every 8 h) the concentration had risen to 3.6 fig/ml (60% of simultaneous serum concentration). Chloramphenicol level also measured after the first dose (second episode) in Case 11 was 2.5 fig/ml in ascitic fluid between 1 and 2 h after intravenous therapy, while simultaneous serum level was 22.1 /xg/ml. The same dose of chloramphenicol in the same patient resulted in levels of 6.0 fig/ml and 2.5 fig/ml in serum and ascites, respectively, 3 h after the initial dose in the first episode of peritonitis. In Case 12, tobramycin concentration on Day 7 of therapy was 1.3 fig/ml 1 h after a dose, which was only 14% of the serum concentration. However, there was only one patient (Case 12) in whom specimens were obtained from abdominal drains, and the specimens may not have been in communication with free ascitic fluid. Ampicillin was measured in Case 18 4 h after the first dose, and the concentration in ascitic fluid was only 2 3 % of the serum level. This patient was essentially asymptomatic, but ascitic fluid culture yielded Enterococcus organisms and the ascitic fluid inflammatory response (400 leukocytes/mm 3 , 35% neutrophils) was the lowest of any patient studied. One patient (Case 8) had a clindamycin concentration of 2.3 fig/ml in ascitic fluid 2.5 h after his seventh dose (Day 2 of treatment), the only unexplained assay below half of the serum concentration. As expected, serum concentrations of aminoglycosides usually exceeded ascitic fluid concentrations in specimens
June 1977 • Annals of Internal Medicine • Volume 86 • Number 6
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19522/ by a University of California San Diego User on 04/10/2017
obtained within 2 h of a dose, reflecting slow distribution of the drug from the vascular compartment to ascitic fluid. Beyond 2 h after a dose, ascitic fluid and serum concentrations tended to be comparable, with ascitic fluid concentrations occasionally higher than serum. Ascitic fluid concentrations of ampicillin and penicillin G obtained 2 h after a dose (Cases 7, 10, 16) were equal to or in excess of simultaneous serum specimens, possibly a reflection of their shorter serum half-life as compared to that of the aminoglycosides. With the exception of Case 8 and the second episode of Case 11, ascitic concentrations of clindamycin, chloramphenicol, and cefazolin appeared to be comparable to serum concentrations in the few patients studied. Although serum creatinine concentration was elevated ( > 1.4 mg/dl) in 12 patients (Table 1), antibiotic penetrance into ascitic fluid did not appear to be altered. The dosage interval for aminoglycosides was adjusted to compensate for decreased renal function, but for ampicillin (Cases 15 and 16) and cefazolin (Case 20) no dosage adjustment was made and concentrations in both serum and ascitic fluid were quite high. A favorable response to therapy occurred in 12 of 16 patients with culture-proven bacterial peritonitis (Table 2) and in all five patients with negative culture. Of the four patients who failed to respond, one (Case 3) was thought to be in septic shock, one (Case 13) had a perforated peptic ulcer that was not surgically closed, one (Case 15) had an infected pancreatic pseudocyst, and one (Case 20) had staphylococcal peritonitis after portacaval shunt surgery. In 11 of the 16 patients with culture-proven bacterial peritonitis, the minimum inhibitory concentration of the infecting organisms was measured and compared to the highest concentration of antibiotic found in the patient's ascitic fluid (Table 2 ) . In all but one patient (Case 12, mixed infection) with a favorable response to therapy, the antibiotic concentration in the ascitic fluid exceeded the minimum inhibitory concentration of the infecting organism. However, in three of the four patients who did not respond to therapy, the antibiotic concentration in the ascitic fluid was also in excess of the minimum inhibitory concentration. Obviously the presence of appropriate antibiotics in the infected area does not necessarily insure a favorable therapeutic response. Despite a favorable response to therapy and resolution of the acute infection, the ultimate prognosis remains poor. Eleven of the 16 patients with culture-proven bacterial peritonitis died (Table 2 ) , and two of the five patients whose ascitic fluid cultures were negative also died. Most patients who recovered from their acute infection died from progressive hepatic and renal decompensation or complications of alcoholism and cirrhosis (gastrointestinal bleeding and hepatoma). Death in these patients occurred from 8 to 86 days after admission. The cause of death in those patients failing to respond to antibiotic therapy was directly related to the infection. Discussion
Direct intraperitoneal instillation of antibiotics has been
recommended for the treatment of bacterial contamination of the peritoneal cavity by a number of surgeons (17-23). Both intraperitoneal antibiotic lavage and instillation of kanamycin (17), kanamycin-bacitracin (18), ampicillin (19), cephalothin (20), tetracycline (21), cephaloridine (22), and gentamicin (23) have received favorable reports for reducing morbidity or mortality. However, these studies are compromised by the concomitant use of other systemic antibiotics (17, 18, 20, 23), use of retrospective controls (19, 20, 23), or investigation limited to laboratory animals (21). Fowler (22) has shown a significant reduction in the incidence of intraperitoneal abscesses in children with peritonitis by the use of intraperitoneal cephaloridine for 48 h. In contrast, Rambo (24) was unable to find any benefit from peritoneal lavage with cephalothin compared to saline solution in patients with peritonitis. A later study of experimental E. coli peritonitis in dogs showed no difference in therapeutic benefit between intravenous and intraperitoneal cephalothin (25). Studies of the concentration of antibiotics in peritoneal fluid after systemic therapy are limited. In an early study of experimental peritonitis in dogs, the peritoneal concentration of sulfadiazine was found to be the same as in blood, while peritoneal fluid concentrations of streptomycin and penicillin were 50% of blood levels (26). Concentrations of cephaloridine (22) in man and kanamycin (27) in dogs after systemic therapy appear to be about 20% of peak serum concentrations after a single dose. Studies in dogs with experimental ascites have shown ascitic fluid peak antibiotic concentrations ranging from 5.8% to 6 5 % of peak serum concentrations after a parenteral dose (28). In general, we found that concentrations of antibiotics in ascitic fluid are adequate in patients with bacterial peritonitis who receive only systemic therapy. Although five of the patients studied had negative bacterial cultures of ascitic fluid, we believe the clinical presentation, cellular inflammatory response in ascitic fluid, and response to antibiotic therapy justify their inclusion in this series. Ascitic fluid leukocyte and differential counts in alcoholic cirrhosis do not always clearly separate infected from uninfected patients (29). However, the leukocyte responses in our group of patients with negative cultures were well above the mean for uninfected patients, and most were above the highest recorded normal values (29, 30). Ascitic fluid leukocyte counts in the culture-negative group were no different than in our culture-positive patients, and there was no difference in the ascitic fluid antibiotic levels in the two groups. Antibiotic concentrations in patients with ascites and bacterial peritonitis are therapeutically adequate in the ascitic fluid, somewhat analogous to the situation in septic joint effusions where direct instillation of antibiotic has not been necessary (31). Although not all of our patients had a favorable response to antibiotic therapy (Table 2 ) , those who failed to respond were severely ill and usually had additional complications (shock, infected pseudocyst, perforated viscus, abdominal surgery). We believe that systemic therapy is adequate in this clinical setting but ascitic fluid concentrations of the aminoglycosides may be very near the minimum inhibitory concentration of infectGerding et a\. • Antibiotics in Ascitic Fluid
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19522/ by a University of California San Diego User on 04/10/2017
711
ing Gram-negative bacteria (Table 2 ) . Except for the sharp serum peaks, ascitic fluid levels parallel serum concentrations. Therefore it would seem advisable to assay aminoglycosides in serum (and also ascitic fluid if possible) to be certain that the prescribed dosage has resulted in therapeutic levels. The best time to draw a serum specimen appears to be at about 2 to 4 h after a dose, as serum levels at that time are most likely to reflect ascitic fluid levels (Table 1). Ascitic fluid antibiotic concentrations in an uninfected dog model have been shown to increase slowly after single intramuscular injections. Peak ascitic fluid level did not occur until 4 h after administration of most drugs, and mean peak ascites concentration ranged from 1 1 % to 27% of peak serum level for aminoglycosides, cephalosporins, and clindamycin ( 2 8 ) . Peak ascites and serum levels of antibiotics were not measured in our present study; however, seven paired specimens were obtained within 4.5 h of the first antibiotic dose, and they tend to confirm in humans with peritonitis the observations of low ascites levels in uninfected dogs. Gentamicin concentration in ascitic fluid was low in three of four specimens obtained after the first dose in three patients (Case 11 was treated twice). Ascitic fluid chloramphenicol was also measured twice after the initial dose in Case 11 and was 2.5 /xg/ml on both occasions. Ampicillin ascitic level after the first dose in Case 18 was low (0.6 fig/ml). In the uninfected ascitic dog model, the concentration of antibiotic in ascites has been shown to depend on the serum antibiotic level and the ascitic fluid volume (28). In the patient with peritonitis, peritoneal inflammation may also alter ascitic fluid penetrance as has been shown in meningitis. Three measures of ascitic inflammatory response—leukocyte count, percentage of neutrophils, and total protein— were compared with ascitic fluid gentamicin level to ascertain the possible influence of inflammation. The comparison was restricted to those cases in which gentamicin was measured after the first dose (Cases 9, 10, and 11) because multiple doses appear to raise ascitic fluid levels considerably (Table 1). There did not appear to be any relation of gentamicin ascitic fluid level to leukocyte count, neutrophil percentage, or ascitic fluid protein in this small group of patients; in particular, one patient (Case 10) (5100 leukocytes/mm 3 , 87% neutrophils, 0.4 g/dl protein) did not show evidence of greater inflammation than the others. Based on dog studies (28), we suspect ascitic fluid volume differences are the major determinant of ascitic fluid levels, but ascites volume was not measured in these patients. Wilson, Chalmers, and Madoff (4) were unable to correlate ascitic fluid cephalothin levels with ascites volume but commented that the lowest ascitic fluid levels were seen in the patient with the greatest amount of ascites. Delay in ascitic fluid penetration of the initial dose of systemic antibiotic may be important clinically if the patient is severely ill. Although routine intraperitoneal instillation of antibiotics is not necessary, it may be justified as an initial route of administration in a critically ill patient. We have administered gentamicin once by direct intraperitoneal instillation in a patient (Case 11), who presented clinical signs of septic shock with apparent relapse of E. Coli peri712
tonitis. The patient received 80 mg of gentamicin intraperitoneally 3 h after the first intravenous dose of 100 mg. Ascitic fluid gentamicin level was 1.1 /xg/ml before intraperitoneal instillation and 8.0 jug/ml 15 minutes later (Table 1). Serum gentamicin level remained constant at 2.3 fig/ml. No toxicity was noted, but clinical benefit could not be ascertained. Although it appears that systemic gentamicin doses (1 to 2 mg/kg) can be safely administered intraperitoneally, comprehensive studies of this procedure to evaluate possible toxicity and clinical efficacy in peritonitis associated with ascites have not been done. Optimal choice of antibiotics in the empiric treatment of peritonitis associated with ascites is not clear and cannot be ascertained from this study. The most frequent infecting bacteria in spontaneous bacterial peritonitis and cirrhosis have been Gram-negative aerobic enteric bacilli and Grampositive aerobic cocci, including pneumococci and enterococci (1-3). Because of the excellent Gram-negative spectrum of the newer aminoglycosides (gentamicin, tobramycin, amikacin), we favor use of one of these drugs plus penicillin G or ampicillin for Gram-positive bacteria in empiric therapy. Although clinical efficacy of the aminoglycosides appears to be good (Table 2 ) , this is an uncontrolled observation. We are concerned about aminoglycoside use in this setting for two reasons: the minimum inhibitory concentration of Gram-negative bacilli may be very near the achievable levels of aminoglycosides in ascitic fluid (Table 2 ) ; and patients frequently have renal failure at the initiation of treatment (Table 1) and often show deterioration of renal function while receiving aminoglycosides. We have not established a causative role of aminoglycosides in this deterioration. Whether this data on antibiotic concentrations in ascitic fluid in this group of patients can be extrapolated to patients with peritonitis and no detectable ascites is not known. Our studies in the dog suggest that antibiotic concentrations are likely to be higher if ascitic volume is small, presumably because of a lesser dilutional factor (28). Bacterial contamination may be much greater in surgical peritonitis, particularly if a viscus or abscess has ruptured, and direct antibiotic installations and irrigations may therefore be superior to parenteral administration, but this is unproved. Certainly the bacterial spectrum is different from spontaneous bacterial peritonitis in cirrhosis and probably requires a different spectrum of antibiotics, perticularly therapy for Bacteroides fragilis. ACKNOWLEDGMENTS: The authors thank the interns, residents, and fellows of the participating hospitals for their case contributions and assistance in obtaining specimens for antibiotic assay; and Richard H. Lyon, Ph.D., for chloramphenicol determinations. This paper was presented in part at the Fifteenth Interscience Conference on Antimicrobial Agents and Chemotherapy, Washington, D.C., 24-26 September 1975. Grant support: in part by a grant from Eli Lilly and Company, Indianapolis, Indiana. Received 13 December 1976; revision accepted 4 March 1977. • Requests for reprints should be addressed to Dale N. Gerding, M.D.; Infectious Disease Section, Veterans Administration Hospital, 54th St. and 48th Av. South; Minneapolis, MN 55417. References 1. CONN HO, FESSEL JM:
Spontaneous bacterial peritonitis in
cirrhosis: variations on a theme. Medicine {Baltimore) 50:161197, 1971
June 1977 • Annals of Internal Medicine • Volume 86 • Number 6
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19522/ by a University of California San Diego User on 04/10/2017
2. C U R R Y N , M C C A L L U M
RW, G U T H P H :
Spontaneous
peritonitis
in cirrhotic ascites. Am J Dig Dis 19:685-692, 1974 3. CORREIA J P , C O N N H O : Spontaneous bacterial peritonitis in cirrhosis: endemic or epidemic? Med Clin North Am 59:963981, 1975 4. W I L S O N
DE,
CHALMERS
TC,
MADOFF
MA:
The
passage
of
Manual of Clinical Microbiology, 2 n d ed., edited by L E N N E T T E E N , SPAULDING E H , TRUANT J P . Washington, D.C., American Society for Microbiology, 1974, pp. 410-417 17. D I V I N C E N T I F C , C O H N I J R : Prolonged administration of intraperitoneal kanamycin in the treatment of peritonitis. Am Surg 37:177-180, 1971
cephalothin into a n d o u t of ascitic fluid. Am J Med Sci 253:449452, 1967 5. GREENBERG P A , SANFORD J P : Removal and absorption of antibiotics in patients with renal failure undergoing peritoneal dialysis. Ann Intern Med 66:465-479, 1967
tion of peritoneal irrigation with antibiotic solution. Surgery 62:73-78, 1967 19. A U N E S, N O R M A N N E : Diffuse peritonitis treated with continuous peritoneal lavage. Acta Chir Scand 136:401-404, 1970
6. SMITHIVAS T, H Y A M S PJ, M A T A L O N R, et a l : T h e use of
20.
gen-
tamicin in peritoneal dialysis. I. Pharmacologic results. J Dis 124:S77-S83, 1971 7. W E I N S T E I N
AJ,
KARCHMER
AW,
MOELLERING
RC
Tobramycin concentrations during peritoneal dialysis. Agents Chemother 4:432-434, 1973 8. SHEAR L, SHINABERGER J H , BARRY K G :
Infect
J R , et
al:
Antimicrob
Peritoneal t r a n s p o r t of
antibiotics in man. N Engl J Med 272:666-669, 1965 9. GERDING D N , K H A N M Y , E W I N G J W , et al: Pasteurella
SABATH L D , CASEY J I , R U C H
P A , et al: R a p i d microassay
circulating nephrotoxic antibiotics. mother-1970:$3-S9, 1971 12.
LUND
ME,
BLAZEVIC D J , M A T S E N
Antimicrob JM:
Rapid
Agents
for
Che-
gentamicin
bio-
21.
26.
29.
Rapid
microassay
of
15.
LIETMAN
PS,
WHITE
JJ,
SHAW
WV:
Chloramphenicol:
an
enzymological microassay. Antimicrob Agents Chemother 10: 347-353, 1976 16. WASHINGTON J A II, BARRY A L : Dilution test procedures, in
CARIDIS D T , GADDIE J, M A T H E S O N
NA:
lavage
Continuous
perito-
using
peritoneal
HUNT
J A,
RIVLIN
ME,
CLAREBOUT
HJ:
Antibiotic
peritoneal
ROTHENBERG
S, SILVANI
H,
CHESTER
S, et al:
Comparison
of
the efficacy of therapeutic agents in the treatment of experimentally induced diffuse peritonitis of intestinal origin. Ann Surg 128:1148-1163, 1948 27. C O H N I J R : Intestinal antisepsis. Springfield, Illinois, Charles C Thomas, 1968, p . 144
14.
P A , et al:
T r e a t m e n t of
lavage in severe peritonitis. South Am Med J 49:233-238, 1975 24. R A M B O W M : Irrigation of the peritoneal cavity with cephalothin. Am J Surg 123:192-195, 1972 25. SHARBAUGH R J , R A M B O W M : Cephalothin and peritoneal lavage in the treatment of experimental peritonitis. Surg Gynecol Obstet 139:211-214, 1974
28.
SABATH L D , CASEY J I, R U C H
PELOSO OA, FLOYD VT, WILKINSON L H :
lavage in peritonitis. Eur Surg Res 1:142-146, 1969 22. F O W L E R R : A controlled trial of intraperitoneal cephaloridine administration in peritonitis. / Pediatr Surg 10:43-50, 1975
assay using a multiple-antibiotic-resistant strain of Klebsiella pneumoniae. Antimicrob Agents Chemother 4:569-573, 1973 13. STROY SA, PRESTON D A : Specific assay of aminoglycosidic- or polymyxin-type antibiotics present in h u m a n sera in combination with cephalosporins. Appl Microbiol 21:1002-1006, 1971 gentamicin, kanamycin, neomycin, streptomycin and vancomycin in serum or plasma. J Lab Clin Med 78:457-463, 1971
N O O N G P , BEALL A C J R , JORDAN G L J R , et a l : Clinical evalua-
nitis with continuous postoperative peritoneal cephalothin. Am J Surg 126:742-747, 1973
23. multo-
cida peritonitis in hepatic cirrhosis with ascites. Gastroenterology 70:413-415, 1976 10. LILLY RESEARCH LABORATORIES: Paper disc or cup plate antibiotic assay procedures. Indianapolis, Indiana, Eli Lilly and Co., 1970 11.
18.
GERDING
DN,
KROMHOUT
JP,
SULLIVAN
JJ,
penetrance of ascitic fluid in dogs. Antimicrob 10:850-855, 1976 KLINE MM, MCCALLUM
et
al:
Agents
Antibiotic
Chemother
R W , G U T H P H : T h e clinical value of
ascitic fluid culture a n d leukocyte count studies in alcoholic cirrhosis. Gastroenterology 70:408-412, 1976 30. C O N N H O : Spontaneous bacterial peritonitis. Gastroenterology 70:455-457, 1976 31. N E L S O N J D : Antibiotic concentrations in septic joint effusion. N Engl J Med 284:349-353, 1971
Gerding et al. • Antibiotics in Ascitic Fluid
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19522/ by a University of California San Diego User on 04/10/2017
713
