Cephalosporin and Aminoglycoside Concentrations in

ANTIMCROBIAL AGENTS AND CHzMoTERAPY, Dec. 1976, p. 902-911 Copyright 0 1976 American Society for Microbiology Vol. 10, No. 6 Printed in U.S.A. Cephalosporin and Aminoglycoside Concentrations in Peritoneal Capsular Fluid in Rabbits DALE N. GERDING,* WENDELL H. HALL, ELIZABETH A. SCHIERL, AND ROBERT E. MANION Infectious Disease Section, Department of Medicine, Veterans Administration Hospital,* and University ofminnesota Medical School, Minneapolis, Minnesota 55417 Received for publication 1 July 1976 To study the penetration of antibiotics into peritoneal tissue fluid, a subcutaneous tissue capsule model was modified by implanting multiple, perforated spherical capsules in the peritoneal cavity of rabbits. Capsules became vascularized, encased in connective tissue, and filled with fluid having a mean protein concentration of 3.6 g/100 ml. Capsular fluid was obtained by percutaneous needle aspiration and assayed for antibiotic by the disk plate bioassay technique. Cephalosporins were administered intramuscularly at a dose of30 mg/kg. Mean peak concentrations of cephaloridine and cefazolin were significantly higher than cephalothin and cephapirin in capsular fluids, but the percent penetration (ratio of capsular mean peak to serum mean peak) ranged from 8.7 to 16.9% and was not significantly different among the cephalosporins. At 24 h the capsular concentration of cefazolin was significantly greater than for the other cephalosporins (P < 0.001). Lower rabbit serum protein binding observed at high in vivo concentrations may have enabled cefazolin to penetrate capsular fluid, but in vitro protein binding studies did not confirm a decrease in serum protein binding at high concentrations within the clinical range. Kanamycin and amikacin showed comparable capsular fluid peak concentrations as did gentamicin and tobramycin. The percent penetration ranged from 15.2 to 34.5% for the aminoglycosides. The only statistical difference was that amikacin penetration was significantly higher than that for tobramycin. Mean capsular concentrations of amikacin, cefazolin, and cephaloridine compared most favorably with the minimum inhibitory concentration of gram-negative bacilli at the dosages used in this study. Among the factors determining successful antibiotic therapy, the ability of the antibiotic to reach an extravascular site of infection seems particularly important. Intra-abdominal infections (abscesses and peritonitis) are commonly caused by a mixed bacterial flora and are often treated with antibiotics administered both parenterally and locally. If parenteral administration alone fails to achieve adequate levels in peritoneal tissues, intraperitoneal therapy may be required. In addition, members of certain classes of antibiotics, such as the cephalosporins and aminoglycosides, may differ in their penetration of extravascular tissue. A perforated tissue capsule model (14) was adapted for use intraperitoneally in rabbits (11) to determine the penetration of antibiotics into peritoneal capsular fluid. The purpose of this paper is: (i) to describe the peritoneal tissue capsule model used in rabbits and (ii) to evaluate the capsular penetration of single doses of four cephalosporin and four aminoglycoside drugs in this model. A limited comparison of the penetration of subcutaneous capsules and intraperitoneal capsules was also performed. (This paper was presented in part at the 14th Interscience Conference on Antimicrobial Agents and Chemotherapy, 11-13 September 1974, San Francisco, Calif., and the Symposium on Antibiotics, Section on Medicine of the New York Academy of Medicine, 20 November 1974, New York, N.Y.) MATERIALS AND METHODS Capsules. Various tissue capsules used in the study are shown in Fig. 1. Fabricated stainless-steel mesh cylinders and 5-ml plastic syringe barrels drilled with 60 to 70 holes (1.6 mm) were implanted subcutaneously. Ordinary table tennis balls were drilled with 200 to 250 holes and implanted in the peritoneal cavity. Implantation. Albino rabbits (2.5 to 5.25 kg) were anesthetized with 50 to 75 mg of intravenous sodium pentobarbital. For subcutaneous implantations the 902

VOL. 10, 1976..... l _,.*. a l B B ANTIBIOTICS IN PERITONEAL CAPSULAR FLUID g., i.ffi6 'silf jlleslul}tililgltxlilulillttlsil} il4t lililliljlltlljlll.titilttilllllilylilltielillifltlii,.. FIG. 1. Subcutaneous (stainless-steel mesh and perforated syringe barrel) and peritoneal (perforated table tennis ball) capsules. Mesh cylinders are 5.2 cm long by 1.5 cm in diameter. Syringe barrels are 6.1 cm long and 1.4 cm in diameter. r 903 flanks were shaved and the skin was swabbed with povidone-iodine followed by absolute alcohol. A 5-cm incision was made and the subcutaneous space was dissected bluntly. One to three ethylene oxide-sterilized cylinders were implanted in each flank and positioned 2 to 3 cm apart. The skin was closed with metal clips. For intraperitoneal implantation the abdomen was cleansed in the same manner as were the flanks. A 5-cm midline incision was made just below the xiphoid and the peritoneum was opened with scissors. Two to five sterile peritoneal capsules were inserted and manually maneuvered under the abdominal wall into well-separated positions. No attempt was made to fix the capsules in place. The abdominal wall was closed with interrupted silk sutures and the skin was closed with metal clips. Antibiotics. Tobramycin and cephaloridine were furnished through the courtesy of Eli Lilly & Co., Indianapolis, Ind., and amikacin and cephapirin were furnished by Bristol Laboratories, Syracuse, N.Y. All antibiotics were supplied as sterile dry powders or solutions. Antibiotics were administered intramuscularly in the quadriceps muscle or intravenously in the marginal ear vein over 25 min with a Harvard pump (Harvard Apparatus Co., Dover, Mass.). Blood for antibiotic levels was obtained from ear or mammary veins and occasionally by cardiac puncture. Capsular fluid was obtained by percutaneous puncture with a 22-gauge needle through one of the capsular perforations. No capsules were studied until a minimum of 30 days after implantation. Blood and capsular fluid were obtained for antibiotic assay immediately prior to antibiotic administration, at 30 min after intramuscular injection, and 5 min after intravenous infusion and then hourly for 6 h and again at 24 h after administration. Occasional capsular specimens showing evidence of gross bleeding were rejected. All of the cephalosporins (cefazolin, cephaloridine, cephalothin, and cephapirin) and the aminoglycosides (gentamicin, tobramycin, kanamycin, and amikacin) were studied in three to six rabbits each. Capsular antibiotic levels were determined in 10 capsules for each of the antibiotics. Assays. Antibiotic assays were performed in triplicate by using a microbiological disk plate method with 24 h of incubation at 37 C (20, 26). Aminoglycosides were assayed using Bacillus subtilis ATCC 6633 as the indicator organism in antibiotic medium 5 at ph 8.0 (Difco Laboratories, Detroit, Mich.). B. subtilis ATCC 6633 was also used for the assay of cephalothin, using 75% synthetic agar and 25% biochemical medium 6 at ph 8.0 (20). Biochemical medium 6 is similar to antibiotic medium 5 (Difco), but contains 6 g of yeast extract per liter and 17.5 g of agar per liter. Cephapirin and cephaloridine were assayed with Sarcina lutea ATCC 9341, using either antibiotic medium one (Difco) or biochemical medium 6, respectively. Cefazolin was assayed with Staphylococcus aureus ATCC 6538P on antibiotic medium 11 (Difco) at ph 6.0. All fluid specimens were placed on 6.35-mm paper disks (Schleicher and Schuell, Inc., Keene, N.H.), using disposable 20-,ul capillary pipettes (Unopette, Becton, Dickinson &

904 GERDING ET AL. Co., Rutherford, N.J.). Initially, standard curves for capsular fluids were determined using both pooled rabbit serum and pooled capsular fluid. No differences were noted between standard curves using serum or capsular fluid for any of the eight antibiotics assayed, and therefore subsequent assays were performed using only rabbit serum as a standard. Protein binding. Binding of cefazolin to rabbit serum was measured by an ultracentrifugation method (22). Specimens were centrifuged at 295,000 x g for 3 h at a temperature of 33 to 35 C with ph maintained in the range of 7.2 to 7.6. The concentration of cefazolin was measured in the uppermost 0.5 ml of a total serum volume of 8 ml. Binding studies were performed on serum specimens obtained from rabbits 30 min after intramuscular injection with cefazolin at doses of 30 and 15 mg/kg and on rabbit serum in vitro at concentrations of 22, 88, and 300,ug/ml. Pressures. Capsular hydrostatic fluid pressures were measured using a water manometer while the rabbit was secured on its back on a holding board at 450. Pressures were measured 8 to 9 weeks after implantation. Freshly drawn iced capsular fluid specimens unexposed to air were used for ph measurements. Statistics. Serum antibiotic half-life (t1/2) was calculated by the standard formula, using the slope of the regression line determined by the method of least squares (12) from the arithmetic means of three to six experiments. Each serum antibiotic tl/2 was calculated over the period extending from 2 to 5 h after administration, except that for cephalothin the 1- to 4-h period was used because of the short tl2 in serum. Capsular fluid t412 was determined over the period from 4 to 6 h after injection, using the mean of 10 capsules for each antibiotic. The standard error of the mean was determined by the method of Mantel (21). All statistical analysis was done by the nonparametric Mann-Whitney U Test (3). A probability value of less than 0.05 was used for statistical significance. The area under the curve (AUC) was calculated in triplicate using a planimeter. RESULTS Capsule characteristics. Peritoneal capsules adhered to omental or bowel surfaces and became encased in a thin layer of connective tissue that covered both their internal and external surfaces (Fig. 2). Vascularization occurred through perforations in the capsule wall and was most prominent at the point of attachment to the bowel or omentum. Sagittal microscopic sections showed connective tissue fibroblasts and mononuclear cell infiltrates, with occasional areas of fat within the capsules. Large blood vessels were seen only in areas of omental attachment (Fig. 3). Capsular fluid characteristics are given in Table 1. Specimens were obtained from 3 weeks to 8 months after implantation. There was a A ri 1% K 4. ANTIMICROB. AGENTS CHEMOTHER..t... 0 i... ok. -- FIG. 2. Gross appearance of a capsule removed 6 months after implantation. (A) External surface showing dark vascularized area at the upper pole that was dissected free from omental attachment. The thin layer of connective tissue has been reflected at the lower pole to show nodules of tissue that had penetrated the capsule perforations. (B) Sectioned capsule showing thick connective tissue lining at the inferior pole (region of omental attachment). Thin connective tissue layer at the upper pole has separated from the capsule wall during sectioning (arrow). The central dark area was filled with capsule fluid. trend toward lower leukocyte and erythrocyte counts and higher ph in older capsules, and differential counts showed fewer polymorphonuclears with an increasing duration of implantation. The total protein and albumin remained relatively high but stable. Capsular pressures ranged from -0.7 to +2.2 cm of water in a series of 20 measurements within five capsules. The mean capsular pressure was +0.9 cm of water. Immediately after the aspiration of 0.4 to 0.6 ml of fluid, the ol 0 4 t

VOL. 10, 1976 0EwS a ;r L ANTIBIOTICS IN PERITONEAL CAPSULAR FLUID '',. '';1 -.7 -, FIG. 3. Microscopic appearance of capsule tissue (x30). (A) Area of thin connective tissue extending through a chamber perforation, showing fibroblasts and mononuclear cells. The capsule wall has been destroyed by the fixative. (B) Section through area of omental attachment, showing remnant of capsule wall (black arrow) and vascular channels (white arrow) near a capsule perforation. TABLE 1. Capsule fluid characteristics (5 to 10 specimens examined) Determination Mean Range Leukocyte count/mm3 1,006 70-2,600 Polymorphonuclears (%) 36 19-60 Erythrocyte count/mm3 493 196-5,250 ph 7.54 7.49-7.61 Total protein (g/100 ml) 3.6 2.9-4.1 Albumin (g/100 ml) 2.6 2.1-3.0 pressure decreased by a mean of 8.1 cm of water (range, 6.4 to 11.1 cm). Pressures returned to base line levels within 1 h, often rising to previous levels within 15 min after aspiration. Cephalosporins. Serum and capsular pharmacokinetics are shown in Fig. 4. The cephalosporins were all administered intramuscularly at a dose of 30 mg/kg. Mean peak serum levels were highest for cefazolin and lowest for cephalothin, but there was considerable variability among the rabbits in serum levels, particularly with cephalothin. Capsular levels also showed considerable variability, even for capsules implanted in the same rabbit. For both cephaloridine and cephalothin, peak capsular levels occurred 2 h after administration, whereas capsular peaks for cephapirin and cefazolin occurred 3 and 4 h, respectively, after administration. 905 Peak serum and capsular levels (mean and range) are shown in Table 2. The peak serum concentration for cefazolin was significantly higher than for the other three cephalosporins (P range, <0.02 to <0.005). Peak serum levels were not significantly different for the other three cephalosporins when compared with each other. Mean peak capsular levels ofboth cefazolin and cephloridine were significantly higher than for cephapirin and cephlothin (P < 0.005, cefazolin versus cephapirin; P < 0.025, cefazolin versus cephalothin; P < 0.025, cephaloridine versus cephapirin; P < 0.04, cephaloridine versus cephalothin). Peak levels for cefazolin versus cephaloridine and cephapirin versus cephalothin were not significantly different. When the ratio of mean capsular peak to mean serum peak was used to determine the percent penetration, it was found to range from 8.7 to 16.9% for the cephalosporins (Table 2). There was no statistically significant difference in percent penetration among the four cephalosporins. The ratio of mean capsular peak to serum AUC was also calculated. For cefazolin, cephaloridine, and cephapirin there were no statistically significant differences. However, cephalothin, which showed the lowest serum AUC, showed significantly higher ratios of capsular

906 GERDING ET AL. ANTIMICROB. AGENTS CHEMOTHER.. 5 CEFAZOLIN 50 CEPHALORIDINE 40-- ~~~~~~40- z 30 30-0z' 4Q 20- is 20- tr'r c_ 10 1 o0.- 0' 0 1/2 1 2 3 4 56 24 '4123456 24 Z Hours Hours 0 0 Z 70-70. 0 60--- 60- oo50 CEPHALOTHIN CEPHAPIRIN 0 50N 50 M Z 40-.40-0- 0 30-30- 20-20- 10 10 0 A 0. i R /2 1 2 3 4 5 6 24 '/4 1 2 3 4 5 6 24 Hours Hours FIG. 4. Concentrations offour cephalosporins in serum (solid line) and peritoneal capsular fluid (broken line). Administration was intramuscular at a dose of3o mglkg. Each point on the serum curve is a mean of3 to 6 determinations, and each point on the capsular curve is a mean of10 determinations. Vertical bars indicate standard errors of the mean. TABLE 2. Serum and capsular mean peak concentrations and ranges (micrograms per milliliter), percent penetration (ratio of mean capsule peak to mean serum peak x 100), and mean capsular concentration and range at 24 h (micrograms per milliliter) for four cephalosporins and four aminoglycosidesa Serum peak Capsule peak % Penetra- Capsule concn at 24 h Antibiotic Mean Range Mean Range tion Mean Range Cefazolin (6) 111.3 76-165 10.5 4.6-17 9.5 3.1 1.1-3.9 Cephaloridine (3) 73.7 66-84 12.4 3.1-27 16.9 1.3 0.8-2.3 Cephapirin (4) 60.0 50-76 5.2 4.0-9.2 8.7 0.2 0-0.8 Cephalothin (3) 40.2 16.5-78 6.1 1.0-12.5 15.1 0 0 Kanamycin (3) 21.4 18.6-27 4.7 1.0-12.5 22.0 0.6 0-1.7 Amikacin (3) 19.7 17.4-28.8 6.8 1.4-13 34.5 1.2 0-2.7 Tobramycin (5) 6.5 3.5-8.2 1.0 0.4-1.8 15.2 0.3 0-0.7 Gentamicin (5) 4.7 5.1-7.8 1.4 0.3-2.3 29.8 0.4 0.1-0.9 a Ten capsules were studied with each antibiotic. The number of animals studied is in parentheses. peak to AUC than was the case for cefazolin or cephapirin (P < 0.04). Cephalothin was not significantly different from cephaloridine in this regard. A comparison of capsular AUC to serum AUC was not performed because total capsular AUC could not accurately be determined from the available data (Fig. 4). Capsular levels at 24 h (Table 2) were signifi-

VOL. 10, 1976 ANTIBIOTICS IN PERITONEAL CAPSULAR FLUID 907 cantly higher for cefazolin than for the other three cephalosporins (P < 0.001). At 24 h the cephaloridine capsular level was significantly higher than for cephapirin (P > 0.01) and cephalothin (P > 0.001). Cephapirin and cephalothin levels were not significantly different from each other at 24 h. Because cefazolin has been reported to be highly bound to human serum protein (18), we determined the binding of cefazolin to rabbit serum protein at two different in vivo serum cefazolin concentrations. Binding was 86% at a cefazolin concentration of 30 gg/ml, but dropped to 61% at the higher concentration of 82 Ag/ml. However, in vitro rabbit serum binding determinations at 22, 88, and 300,ug/ml showed bindings of 95, 95, and 74%, respectively, much higher than the binding observed on in vivo specimens. Serum and capsular t1l2 is shown in Table 3. For the cephalosporins tl2 was 3.5 to 6.5 times longer in capsules than in serum. Cefazolin had the longest serum t1/2 and cephaloridine had the longest capsular tl/2. The t1/2 in serum and capsules was similar for cephalothin and cephapirin Ȧminoglycosides. Kanamycin and amikacin were administered intramuscularly at a dose of 7.5 mg/kg. Gentamicin and tobramycin were administered either intramuscularly or by a 25- min intravenous infusion at a dose of 2 mg/kg. Data using intramuscular and intravenous administration routes were pooled for both tobramycin and gentamicin because similar serum peaks and serum tl,2 were observed. Serum and capsular kinetics for the four aminoglycosides are shown in Fig. 5. As expected, peak levels of amikacin and kanamycin were similar, but the peak serum level of gentamicin (+ standard error) was lower (4.7 + 1.3,ug/ml) than the peak for tobramycin (6.5 + 0.9,ug/ml). This difference was largely artifactual, resulting from a distribution of serum peaks between the 30-min and 1-h specimens for gentamicin, whereas tobramycin peaks occurred consistently at 30 min. When the mean of the serum TABLE 3. Serum and capsule tl/2 Antibiotic Serum t1l2 Capsule t1,2 (h) Cefazolin 1.4 4.9 Cephaloridine 0.9 5.9 Cephapirin 0.6 3.9 Cephalothin G. 8 3.9 Kanamycin 1.1 5.1 Amikacin 2.0 12.6 Tobramycin 1.3 3.9 Gentamicin 1.3 5.8 peak concentration for gentamicin was calculated, disregarding the time of occurrence, it was 6.1 + 0.7,ug/ml, a value in close agreement with the tobramycin peak. Peak capsular concentrations occurred at 3 h for kanamycin, amikacin, and tobramycin and at 4 h for gentamicin, possibly a reflection of the later serum peaks with gentamicin. Peak capsular concentrations of amikacin versus kanamycin were not significantly different nor were peak capsular concentrations of tobramycin versus gentamicin (Table 2). The ratio of mean peak capsule concentration to mean peak serum concentration (percent penetration) ranged from 15.2% for tobramycin to 34.5% for amikacin (Table 2). The only significant difference in percent penetration was for amikacin versus tobramycin (P < 0.01). Although the value for gentamicin was nearly twice that of tobramycin, this was largely an artifact due to slurring of the gentamicin serum peak. When the actual mean serum gentamicin peak of 6.1,ug/ml was used for calculation, the percent penetration of gentamicin was 23%, which was not significantly different from the value for tobramycin. The ratio of mean capsular peak concentration to serum AUC was not significantly different for any of the aminoglycosides. Although the capsular concentration of amikacin was twice as high as that of kanamycin at 24 h (Table 2), this was not a significant difference. Similarly, gentamicin and tobramycin concentrations at 24 h were not significantly different. Serum and capsular t1,2 for the aminoglycosides is given in Table 3. The t1/2 in capsules was 3 to 6.3 times as long as in serum. For amikacin, capsular t1/2 was the longest of all antibiotics tested (12.6 h) and correlated fairly well with the concentration of the drug measured at 24 h (1.2 gg/ml). Additional capsular antibiotic studies. An extensive comparison of peritoneal to subcutaneous capsular antibiotic concentrations was abandoned when subcutaneous cylinders were found unsatisfactory for chronic use because fluids were frequently contaminated with blood or easily infected or the cylinders quickly filled with connective tissue. However, cefazolin was studied successfully with five subcutaneous cylinders in two rabbits. The mean peak serum concentration (± standard error) was 98.5 + 1.5 p.g/ml. The mean peak subcutaneous cylinder level was 11.0 + 2.0 Ag/ml and occurred at 4 h. The percent penetration was 11.2%. In similar rabbits with peritoneal capsules the serum mean peak was 111.3 + 14.8 ug/ml and the peritoneal capsular mean peak was 10.5 + 1.2

908 GERDING ET AL. ANTIMICROB. AGENTS CHEMOTHER. AMIKACIN 7.5mg/Kg IM KANAMYCIN 7.5mg/Kg IM C0' z I- 0 =- z w U z 0 w a (n 0 0 z 0 5-4- 3-2- 1 0 GENTAMICIN 2mg/Kg IM or IV 7-6- 5-4- 3. 2-1- 2 3 4 5 Hours TOBRAMYCIN 2mg/Kg IM or IV /2 1 2 3 4 5 6 ' 24 1/2 1 2 3 4 5 6' 24 Hours Hours FIG. 5. Concentrations offour aminoglycosides in serum (solid line) and peritoneal capsular fluid (broken line). Each point on the serum curve is a mean of3 to 5 determinations, and each point on the capsular curve is a mean of10 determinations. Vertical bars indicate standard error of the mean.,ug/ml, with a percent penetration of 9.5%. Therefore, serum and capsular peak levels and percent penetration for cefazolin were not significantly different in the two types of capsules. Anecdotally, during preparation of one of the rabbits for surgery, ascites was noted and on exploration the liver surface was observed to be nodular. Capsules were implanted and the animal was studied to determine the concentration of amikacin in ascitic fluid for comparison with peritoneal capsular fluid. The peak ascitic level of amikacin was 7.8 ug/ml, as compared with a mean peak capsular fluid level of 6.8 + 1.2,ug/ml. Unfortunately this animal expired before concentrations of other antibiotics in ascitic fluid could be measured. Because of a concern that the aspiration of intraperitoneal capsular fluid might artifactually induce the penetration of capsules by serum containing high concentrations of antibiotics (particularly in the first 2 h after administration), a series of experiments was performed to investigate this possibility. Some capsular aspirations were delayed until 3 to 4 h after the administration of antibiotics (when the serum antibiotic concentration was considerably lower) for a comparison with the levels achieved in capsules aspirated beginning at 30 min after administration. Tobramycin, cephalothin, and cefazolin were studied sequentially in the same animal. No increase in antibiotic levels was observed in capsules repeatedly aspirated beginning at 30 min as compared with those aspirated only at 3 to 4 h. However, in the course of these experiments it was noted that certain capsules consistently yielded higher antibiotic concentrations than others, regardless of the antibiotic used. At autopsy those capsules showing the higher antibiotic concentrations appeared to have larger areas of vascular attachment than those exhibiting somewhat lower levels. Thus, consistent variation in achievable capsular antibiotic concentrations could be demonstrated between capsules, but

VOL. 10, 1976 artifactual induction of consistently increased concentrations by repeated aspiration was not seen. DISCUSSION The use of perforated capsules as a means of measuring interstitial tissue fluid pressure in dogs was first described by Guyton (14). He implanted capsules both subcutaneously and intramuscularly, and pressures within were found to range from -2.6 mm ofhg in muscle to -7.1 mm of Hg in subcutaneous tissue. Mean capsular fluid protein was only 1.96 giloo ml. Additional studies were done in dogs by Calnan et al. to further characterize subcutaneous capsular fluid and its dynamics (7). The total protein in capsular fluid averaged 2.54 giloo ml (30 to 45% of serum proteins). Sodium was found to be freely diffusible from blood to capsular fluid and back. Albumin was also readily diffusible from blood to tissue fluid, but over a longer period of time, requiring 48 h to equilibrate. Other studies have shown a diffusion rate for albumin one-third that expected for pure water (15). We are unaware of reports of the characteristics of fluid in intraperitoneally implanted capsules, other than our own preliminary results (11). Our mean capsular protein concentration was 3.6 g/100 ml (55% of the serum protein level) in rabbits, nearly twice as high as values reported for subcutaneous capsules in dogs and slightly higher than for subcutaneous capsules in rabbits (29). The majority of peritoneal capsular protein was albumin (mean, 2.6 g/100 ml). Hydrostatic fluid pressures in peritoneal capsules were slightly positive, disagreeing with the subcutaneous capsule observations of Guyton. However, intra-abdominal pressure measurements are influenced by several factors, including the position of the animal, abdominal muscle tension, and gastrointestinal contents; the latter two factors were uncontrolled in the measurement of pressures in this study (17). Several authors have employed subcutaneously implanted capsules in the study of antibiotic penetration into tissue fluid (1, 4, 8, 10, 29, 31). Using capsules implanted for 1 h to 14 days, Alexander et al. showed high concentrations of cefazolin, cephaloridine, penicillin, and ampicillin in tissue fluid in dogs (1). Concentrations of oxacillin, nafcillirt, clindamycin, cephalothin, and gentamicin were considerably lower. In capsules implanted for a minimum of 4 weeks, Chisholm et al. (8) demonstrated peak concentrations of ampicillin that were about 25% of the serum peak concentrations. The ANTIBIOTICS IN PERITONEAL CAPSULAR FLUID 909 peak concentration of gentamicin in tissue fluid exceeded 50% of the serum peak concentration, whereas the tobramycin tissue fluid concentration did not appear to reach 10% of the peak serum concentration. The ddsage of gentamicin was 80 mg intramuscular, but for tobramycin it was only 50 mg intramuscular. Ehrlich and coworkers measured cephaloridine concentrations in fluid collected in 10-day-old subcutaneous mesh cylinders in rabbits and found them to be similar to the intraperitoneal capsular levels obtained in this study (10). Their maximum tissue fluid cephaloridine levels were reached 11/2 h after intravenous bolus infusion, in close agreement with the peak concentration time of 2 h observed after an intramuscular injection in the present study. Waterman and Kastan (31) found that the cephalothin concentration reached a peak of 12 gg/ml in interstitial fluid from 4-week-old subcutaneous capsules in dogs after a 40-mg/kg intravenous dose. This appeared to be approximately 10% of the peak serum concentration, again in agreement with the observations made here. The extravascular concentration of an antibiotic is thought to depend upon a number of factors, including: (i) the concentration gradient from serum to extravascular fluid, (ii) the degree of binding to protein and tissues, (iii) the lipid solubility, and (iv) the ionization state or pka. The binding of antibiotics to serum proteins is felt to be an important factor limiting penetration into interstitial fluid (19). Somewhat surprisingly, the known degree of binding of the various cephalosporins to human serum proteins did not appear to correlate with the peritoneal capsular fluid concentrations in the rabbit model. Cefazolin, which is 86% bound to human serum (18), achieved particularly high capsular concentrations. Our protein binding determinations in rabbit serum obtained in vivo confirmed the high binding of cefazolin at low concentrations (86% at 30,ug/ ml), but binding dropped to only 61% at a concentration of 82,ug/ml. Diminished protein binding at drug concentrations in excess of 100 gg/ml has previously been described by Rolinson and Sutherland (25). A decrease in serum protein binding at high clinical serum concentrations would provide an attractive explanation for the extravascular penetration of cefazolin, but we were unable to confirm the decrease in binding on in vitro studies and, in fact, found binding to be 95% in rabbit serum at drug concentrations in the clinical range. Waterman et al. (32) have also observed high capsular fluid concentrations of cefazolin in dogs, despite

910 GERDING'ET AL. high serum protein binding, and have attributed this to the slow rate of binding of cefazolin to serum proteins (33). In contrast to the cephalosporins, the aminoglycosides used in this study have little or no detectable binding to human serum protein (9, 13). Although the percent penetration of amikacin was the highest among the aminoglycosides, this was statistically significant only when compared with tobramycin. This difference could possibly be due to the higher dose of amikacin (7.5 versus 2 mg/kg). Kqnamycin and amikacin are pharmacologically very similar (9). With the same large dose of kanamycin and amikacin, capsular fluid concentrations of amikacin were slightly higher (Table 2), although this was not significant statistically. Gentamicin and tobramycin also have similar pharmacokinetics (16, 27). Serum and capsular peak concentrations were similar for the two drugs (Table 2). Although the percent penetration of gentamicin appeared to be almost twice as great as for tobramycin (Table 2), this difference could be misleading, due to the lower serum peak of gentamicin resulting from a distribution of peak concentrations between 30 min and 1 h. When the actual serum peak concentration of 6.1 Ag/ml was used, there were no statistical differences between tobramycin and gentamicin. The general pattern of slow accumulation and elimination of drugs from peritoneal capsular fluids has also been seen in subcutaneous tissue capsule models (1, 4, 8, 10, 31), as well as in other models of extravascular antibiotic penetration (5, 6, 28). Our calculations of capsular fluid tl2 (Table 3) are based on only three determinations in each capsule over a 2-h period and should not be construed as highly precise measures, but rather as representative figures for capsular fluid kinetics after the administration of a single parenteral dose. Nevertheless, all capsular tl2 values were in the range of 3.9 to 5.9 h, with the exception of amikacin (12.6 h). From capsule fluid concentrations calculated at 24 h (Table 2), the calculated capsular t1/2 of amikacin appears to be somewhat long, whereas the t1/2 of cefazolin seems too short. Both of these discrepancies are likely due to the short time period used in the calculation of the t1/2. Finally, slow capsular fluid kinetics suggest the probable accumulation of antibiotic in capsules with repeated doses. The exact degree of accumulation is likely to be greatest for those drugs with the longest capsular t1,2, but this will also depend upon the frequency of systemic administration. The basic question of the need for local antibiotic instillation in peritoneal infection cannot ANTIMICROB. AGENTS CHEMOTHER. be answered from the results of the single-dose studies performed here. Concentrations of cephalosporins and aminoglycosides in capsular fluid, although substantial, are not in excess of the minimum inhibitory concentration of some relevant organisms, particularly gram-negative bacilli (24, 30). At the dosage used in this.study, capsular concentrations of cefazolin and cephaloridine compare favorably with the minimum inhibitory concentration of the majority ofenterobacteriaceae, whereas cephapirin and cephalothin levels are somewhat low (30). Furthermore, the assay method employed in this study also measured the desacetyl metabolites of cephalothin and cephapirin, which are biologically active, but to a lesser degree than the parent compounds (34, 35). Among the aminoglycosides, only amikacin capsular concentrations are consistently in excess of the minimum inhibitory concentration of the Enterobacteriaceae (24). Although kinetics suggest that accumulation probably occurs with repeated systemic doses, this remains to be proven, and we emphasize that no attempt has been made in this study to achieve equilibrium conditions. Furthermore, the effect of inflammation in altering tissue penetration might be expected to raise capsular fluid concentrations (paralleling the situation in meningitis), but this is also an unproven hypothesis. Intraperitoneal implantation of tissue capsules has circumvented several of the technical difficulties we have encountered with subcutaneous implants (bleeding, infection, rapid tissue growth). If the similar penetration shown for cefazolin in the two models is confirmed with other antibiotics, this would suggest that either model can provide equivalent results representative of interstitial tissue fluid penetrance in general. However, peritoneal capsular concentrations of antibiotics are variable, even in the same animal, probably as a result of the inconstant degree of vascularity of capsules. Future studies comparing antibiotic penetrance should either be done with large numbers of capsules (as in this study) or by paired cross-over studies in the same capsules in the same animals. The potential of the tissue capsule model for in vivo study of the interaction of microorganisms with antimicrobials has already been emphasized (2, 23, 29). Experimental infections in intraperitoneal capsules and the effect of antibiotic treatment on those infections warrant further investigation. ACKNOWLEDGMENTS This work was supported by a grant from Eli Lilly & Co., Indianapolis.

VOL. 10, 1976 We wish to thank Horace H. Zinneman and Lance R. Peterson for the determination of protein concentrations and protein binding. LITERATURE CITED 1. Alexander, J. W., N. S. Sykes, M. M. Mitchell, and M.V. Fisher. 1973. Concentration of selected intravenously administered antibiotics in experimental surgical wounds. J. Trauma 13:423-434. 2. Arko, R. J. 1972. Neisseria gonorrhea: experimental infection of'laboratory animals. Science 77:1200-1201. 3. Auble, D. 1952. Extended tables for the Mann-Whitney Statistic. Institute of Educational Research, School of Education, Indiana University, Bloomington. 4. Baker, G., and T. K. Hunt. 1968. Penicillin concentrations in experimental wounds. Am. J. Surg. 115:531-534. 5. Barza, M., T. Samuelson, and L. Weinstein. 1974. Penetration of antibiotics into fibrin loci in vivo. H. Comparison ofnine antibiotics: effect of dose and degree of protein binding. J. Infect. Dis. 129:66-72. 6. Barza, M., and L. Weinstein. 1974. Penetration of antibiotics into fibrin loci in vivo. I. Comparison of penetration of ampicillin into fibrin clots, abscesses, and "interstitial fluid." J. Infect. Dis. 129:59-65. 7. Calnan, J. S., J. J. Pflug, G. D. Chisholm, and L. M. Taylor. 1972. Pathophysiology of tissue fluid. Proc. R. Soc. Med. 65:715-719. 8. Chisholm, G. D., P. M. Waterworth, J. S. Calnan, and L. P. Garrod. 1973. Concentration of antibacterial agents in interstitial tissue fluid. Br. Med. J. 1:569-573. 9. Clarke, J. T., R. D. Libke, C. Regamey, and W. M. M. Kirby. 1974. Comparative pharmacokinetics of amikacin and kanamycin. Clin. Pharmacol. Ther. 15:610-616. 10. Ehrlich, H. P., V. Licko, and T. K. Hunt. 1973. Kinetics of cephaloridine in experimental wounds. Am. J. Med. Sci. 265:33-44. 11. Gerding, D. N., and W. H. Hall. 1975. The penetration of antibiotics into peritoneal fluid. Bull. N.Y. Acad. Med. 51:1016-1019. 12. Goldstein, A., L. Aronow, and S. Kolman. 1968. Principles of drug action, p. 196, 287, 295. Harper and Row, New York. 13. Gordon, R. C., C. Regamey, and W. M. M. Kirby. 1972. Serum protein binding of the aminoglycoside antibiotics. Antimicrob. Agents Chemother. 2:214-216. 14. Guyton, A. C. 1963. A concept of negative interstitial pressure based on pressures in implanted perforated capsules. Circ. Res. 12:399-414. 15. Guyton, A. C., H. J. Granger, and A. E. Taylor. 1971. Interstitial fluid pressure. Physiol. Rev. 51:527-563. 16. Harikoshi, N., M. Valdivieso, and G. P. Bodey. 1974. Clinical pharmacology of tobramycin. Am. J. Med. Sci. 266:453-458. 17. Hartman, W. 1973. Intra-abdominal pressure in small ruminants. Zentralbl. Veterinaermed. Reihe A ANTIBIOTICS IN PERITONEAL CAPSULAR FLUID 911 20:498-502. 18. Kirby, W. M. M., and C. Regamey. 1973. Pharmacokinetics of cefazolin compared with four other cephalosporins. J. Infect. Dis. 128(Suppl.):S341-346. 19. Kunin, C. M. 1974. Blood level measurements and antimicrobial agents. Clin. Pharmacol. Ther. 16:251-256. 20. Lilly Reasearch Laboratories. 1970. Paper disc or cup plate antibiotic assay procedures. Eli Lilly & Co., Indianapolis. 21. Mantel, N. 1951. Rapid estimation of standard errors of means for small samples. Am. Stat. 5:26-27. 22. Mattie, H., W. R. 0. Goslings, and E. L. Noach. 1973. Cloxacillin and nafcillin: serum binding and its relationship to antibacterial effect in mice. J. Infect. Dis. 128:170-177. 23. Rathleu, T. 1973. Cultivation of pathogenic Treponema pallidum in chambers surgically implanted in experimental animals. Acta Pathol. Microbiol. Scand. 81: 269-271. 24. Reynolds, A. V., J. M. T. Hamilton-Miller, and W. Brumfit. 1974. Newer aminoglycosides-amikacin and tobramycin: an in vitro comparison with kanamycin and gentamicin. Br. Med. J. 3:778-780. 25. Rolinson, G. N., and R. Sutherland. 1965. The binding of antibiotics to serum proteins. Br. J. Pharmacol. 25:638-650. 26. Sabath, L. D., J. I. Casey, P. A. Ruch, L. L. Stumpf, and M. Finland. 1971. Rapid microassay for circulating nephrotoxic antibiotics, p. 83-89. Antimicrob. Agents Chemother. 1970. 27. Stratford, B. C., and S. Dixson. 1974. Serum levels of gentamicin and tobramycin after slow intravenous bolus injection. Lancet i:378-379. 28. Tan, J. S., A. Trott, J. P. Phair, and C. Watanakunakorn. 1972. A method for measurement of antibiotics in human interstitial fluid. J. Infect. Dis. 126:492-497. 29. Tight, R. R., R. B. Prior, R. L. Perkins, and C. A. Rotilie. 1975. Fluid and penicillin G. dynamics in polyethylene chambers implanted subcutaneously in rabbits. Antimicrob. Agents Chemother. 8:495-497. 30. Washington, J. A., II. 1976. The in vitro spectrum of the cephalosporins. Mayo Clin. Proc. 51:237-250. 31. Waterman, N. G., and L. B. Kastan. 1972. Interstitial fluid and serum antibiotic concentrations. Arch. Surg. 105:192-196. 32. Waterman, N. G., M. J. Raff, L. Scharfenberger, and P. A. Barnwell. 1976. Protein binding and concentrations of cephaloridine and cefazolin in serum and interstitial fluid of dogs. J. Infect. Dis. 133:642-647. 33. Waterman, N. G., L. Scharfenberger, and M. J. Raff. 1974. Rate of binding of antibiotics to canine serum protein. Antimicrob. Agents Chemother. 5:294-295. 34. Wick, W. E. 1966. In vitro and in vivo laboratory comparison of cephalothin and desacetylcephalothin, p. 870-875. Antimicrob. Agents Chemother. 1965. 35. Yamanaka, Y., S. Kono, H. Tateishi, and H. Aratani. 1974. Pharmacologic studies of cephapirin. Second report. Distribution, fate, and excretion. Chemotherapy 22:1208-1218.