Staphylococcus aureus

ANTIMICROBIAL AGENTS AND CHEMOTHERAPY, OCt. 1981, p. 463-469 0066-4804/81/100463-07$02.00/0 Vol. 20, No. 4 In Vitro and In Vivo Studies of Three Antibiotic Combinations Against Gram-Negative Bacteria and Staphylococcus aureus S. H. ZINNER,' J. KLASTERSKY,2* H. GAYA,:t C. BERNARD,4 J. C. RYFF,' AND THE EORTC ANTIMICROBIAL THERAPY PROJECT GROUP Department of Medicine, Roger Williams General Hospital, Brown University, Providence, Rhode Island 02908'; Department of Medicine, Institut Jules Bordet, 1000 Brussels, Belgium2; Wright Fleming Institute, St. Mary's Hospital Medical School, London, England:'; and Service des Maladies du Sang, Hospices Ciuils de Strasbourg, 67091 Strasbourg Cedex, France4 Received 17 February 1981/Accepted 7 July 1981 The activities of azlocillin, cefotaxime, and amikacin alone and in combination were evaluated in in vitro checkerboard studies, in infected neutropenic mice, and in human volunteers. The combination of cefotaxime plus amikacin was more synergistic in vitro than the others against the Enterobacteriaceae tested, and the combination of azlocillin plus amikacin was more synergistic against Pseudomonas aeruginosa and Staphylococcus aureus. Survival of neutropenic mice infected with Escherichia coli and Klebsiella pneumoniae, respectively, was greater with azlocillin plus amikacin (24 of 40 and 11 of 40) and with cefotaxime plus amikacin (21 of 40 and 17 of 40) than with azlocillin plus cefotaxime (22 of 40 and 3 of 40; P < 0.05). Median serum bactericidal activity in volunteers receiving these antibiotics alone and in combination was -1:8 with most agents and with all combinations tested against 10 strains each of E. coli, K. pneumoniae, P. aeruginosa, and S. aureus. These data suggest that clinical trials with combinations of azlocillin or cefotaxime plus amikacin deserve further study in febrile neutropenic patients. Overwhelming infections with gram-negative bacteria and Staphylococcus aureus occur frequently in granulocytopenic patients (7, 9, 24). Early empiric combination antibiotic therapy is widely used in this clinical situation (7, 10, 18, 24). Most therapeutic regimens include an aminoglycoside, such as gentamicin, tobramycin, or amikacin, and a broad-spectrum beta-lactam, such as carbenicillin or ticarcillin, or a cephalosporin. However, there is need for combinations with even broader activity and less toxicity than the above. The studies recorded in this report on the comparative activities of combinations of amikacin with azlocillin, a new broad-spectrum ureidopenicillin (29), and cefotaxime, a new broad-spectrum cephalosporin (26), were a response to that need. The studies were pursued collaboratively in three laboratories using (i) in vitro susceptibility testing, (ii) infected neutropenic mice, and (iii) serum from human volunteers who had received each drug alone and in combination. MATERIALS AND METHODS In vitro studies. A total of 197 organisms isolated at the Wright-Fleming Institute, St. Mary's Hospital Medical School, London, were tested. These included 26 strains of Escherichia coli, 39 strains of Klebsiella species, 25 strains of Proteus mirabilis, 39 strains of Pseudomonas aeruginosa, 18 strains of Serratia species, and 50 strains of Staphylococcus aureus. Strain typing was not performed, and it is possible that some redundancy occurred. In vitro checkerboard studies on the activity of azlocillin, cefotaxime, and amikacin alone and in combination were performed in Iso-sensitest broth (Oxoid Ltd.) in microtiter plates (12). Organisms were inoculated with a multipoint inoculator to give a final concentration of 104 to 105 organisms per ml. Fractional inhibitory concentration indices (FIC indices) were calculated from the formula MIC of drug A in presence of B FIC index = + MIC of drug A alone MIC of drug B in presence of A MIC of drug B alone where MIC = minimum inhibitory concentration. According to Berenbaum (5) synergism is present if this t Present address: Brompton Hospital, Fulham Rd., London SW3 6HP, England. index value of 0.5 represents a fourfold reduction in index is <1, and antagonism is present if it is >1. An 463

464 ZINNER ET AL. the MIC of each antibiotic in the presence of the other agent and is consistent with a significant synergistic effect (21). In this paper, an FIC index of <1 represents synergism, an index of I to 2 represents an indifferent or additive effect, and an index of >2 represents antagonism. These definitions more closely approximate the concave inward bowing of a synergistic isobologram and the concave outward bowing of an antagonistic isobologram (21, 22). Animal studies. Animal studies were performed by using a modification of a neutropenic mouse model previously reported from this laboratory (6). BALB/c mice (22 to 24 g) were made neutropenic with two intraperitoneal injections of 200 mg of cyclophosphamide per kg, separated by an interval of 54 to 56 h. This dosage schedule produced maximal granulocytopenia on day 4 after the first injection more consistently than did the single dose used in the earlier study. The leukocyte and granulocyte response to cyclophosphamide is shown in Fig. 1. On day 4 after the first cyclophosphamide injection, the mice were injected intraperitoneally with a bacterial inoculum in 0.2 ml of Mueller-Hinton broth. The inoculum concentrations ranged from 3 to 20 times the 50%; lethal dose, which was determined for each infecting organism in groups of 30 mice. Protective dose curves were determined for azlocillin, cefotaxime, and amikacin against each test organism, and calculations were made by the method described by Litchfield and Wilcoxon (17). A dose approximately equal to the concentration of drug expected to protect only 2.5%; of the mice (PD2.5) was extrapolated from the protective dose curves. This dose was selected for use in combination studies designed to show possible additive effects of two of these drugs used together. In practice, this selected dose of single agents allowed survival of 5 to 15% of treated mice, possibly due to inoculum variation. All infecting organisms were extremely susceptible to all three antibiotics, which necessitated the C-) UU lo T r LEUKOCYTES o NEUTROPHILS o ' A CYTOXAN 200mg/kg IP _ 0 A A DAYS FIG. 1. Reduction in total leukocyte count and neutr-ophil count caused bv 200 mg of cyclophosphamide (cytoxanj per kg gilen intraperi/oneall- on two occasions in 20 mice. Brackets represent standard errors of the means. CMM. Cubic millimeter. ANTIMICROB. AG.ENTS CHEMOTHER. use of very low doses (see Table 3) in an attempt to maximize the likelihood of showing an additive effect with the combinations. Each antibiotic was injected subcutaneously in a volume of 0.2 ml 1 h after infection. A second treatment (same dose and volume) was injected 3 to 4 h later. Azlocillin, cefotaxime, and amikacin were given alone to groups of 20 to 40 infected mice and in combinations of two agents to other groups of 20 to 40) infected animals. The MIC of each antibiotic for each organism used, the 50% lethal dose, the infecting inoculum injected, and the doses of each drug are shown in Table 3. Antibiotic levels were not measured in the mice. Results are reported as the number of survivors per treatment group, and the chi-square test for comparison of proportions was used in statistical analysis. Studies in human volunteers. Three groups of six informed and consenting adult volunteers with normal renal and hepatic function received one infusion per day of a single dose of one antibiotic or a combination. Thus, one group of six received on 3 separate days azlocillin (75 mg/kg), amikacin, (7.5 mg/ kg), and the combination at the same doses. A second group of six received on 3 separate days one infusion of cefotaxime (15 mg/kg), amikacin (7.5 mg/kg), and the combination, and a third group of six received cefotaxime, azlocillin, and the combination of the two beta-lactam agents in the same doses as given above. Within each group, the order of administration was randomized by using a Latin square design so that, for example, two volunteers received azlocillin on dav 1, two received amikacin on day 1, and two received the combination as the first infusion. Antibiotics were administered as intravenous infusions in 50 ml of 5% dextrose over 30 min. Sera were obtained at 1 and 6 h after infusion and were frozen at -20 C until used. Serum bactericidal activity of each of the 108 sera against each of 10 strains of E. coli, Klebsiella species, P. aeruginosa, and S. aureus was determined in microtiter plates with tryptic soy broth as the diluent (15). These organisms were isolated from patients at the Institut Jules Bordet in Brussels. Gram-negative bacilli were studied at a final inoculum of 104 organisms per ml, and the staphylococci were studied at an inoculum of 10' organisms per ml. As described elsewhere (15), the highest dilution of a given serum which resulted in a 99.9%( reduction of this inoculum represented the serum bactericidal activity. RESULTS In vitro studies. In vitro studies were done at the Wright-Fleming Institute, St. Mary's Hospital Medical School, London. Data regarding the susceptibility of the 197 strains tested to the three antibiotics are presented in Table 1 (4). MICs were lowest for cefotaxime against E. coli, Klebsiella species, P. mirabilis, and S. aureus. Amikacin was most active against Serratia species, and amikacin and azlocillin were most active against P. aeruginosa. Assays for beta-lactamase produced by these organisms were not performed.

VOL. VEVALUATION 20, 1981 OF ANTIBIOTIC COMBINATIONS 465 TABLE 1. In vitro susceptibility of 197 bacterial isolates to azlocillin, cefotaxime, and amikacin" MIC ([g/ml) Organism No. of isolates Antibiotic For % of isolates: Range 25 50 90 E. coli 26 Azlocillin 4-256 4 8 128 Cefotaxime 0.06-0.25 <0.06 0.06 0.12 Amikacin 0.5-4 0.5 2 4 Klebsiella species 39 Azlocillin 4-512 16 128 256 Cefotaxime 0.03-2 0.03 0.06 0.25 Amikacin 0.25-4 0.75 1.0 2 P. mirabilis 25 Azlocillin 1.0-512 1.0 2 256 Cefotaxime 0.015-16 0.015 0.03 0.06 Amikacin 0.5-8 1.0 2 8 Serratia species 18 Azlocillin 8-512 8 16 512 Cefotaxime 0.06-16 0.06 0.12 8 Amikacin 0.5-4 0.5 1.0 12 P. aeruginosa 39 Azlocillin 2-256 2 4 8 Cefotaxime 2-64 4 8 16 Amikacin 0.5-32 0.5 2 8 S. aureus 50 Azlocillin 1.0-256 2 4 16 Cefotaxime 0.25-32 0.5 1.0 16 Amikacin 1.0-256 1.0 2 16 (Series from St. Mary's Hospital, London. Table 2 presents the results of in vitro checkerboard studies of these drugs in combination against the 197 strains of gram-negative rods and S. aureus. Synergism was found most frequently with cefotaxime plus amikacin for E. coli and Klebsiella species and with azlocillin plus amikacin for P. aeruginosa and S. aureus. Antagonism was not found with azlocillin plus amikacin and was found against only one strain of P. aeruginosa with cefotaxime plus amikacin. Antagonism was found with azlocillin plus cefotaxime against 12 of the 108 strains (11%) of Enterobacteriaceae tested. These data also are reflected in the FIC indices in Table 2. Of 108 Enterobacteriaceae tested, cefotaxime and amikacin were synergistic for 84 (78%) compared with 64 (59%) with azlocillin plus amikacin (P < 0.01) and 59 (50%) with azlocillin plus cefotaxime (P < 0.001). The combination of azlocillin plus amikacin was more frequently synergistic against P. aeruginosa and S. aureus than were the other combinations. Animal experiments. Animal experiments were done at Roger Williams General Hospital, Providence, R.I. The inocula for the infecting doses and the doses used in the treatment of the infected neutropenic mice are presented in Table 3. The results of therapy with the study antibiotics alone and in combination are presented in Table 4. Subeffective doses approximating the PD2> of each antibiotic alone and in combination were used in these experiments. In practice, due to small variations in actual inoculum received and in recovery from neutropenia, all animals treated with low doses of single antibiotics or with saline did not always die. In animals infected with E. coli, each combination of subeffective doses was significantly more effective than was each drug alone. In animals infected with K. pneumoniae, azlocillin plus amikacin was more effective than azlocillin alone, and cefotaxime plus amikacin was more active than either drug alone. In both of these groups of mice, azlocillin plus amikacin and cefotaxime plus amikacin were more active than the double beta-lactam combination (Table 4). Similar trends were seen in the smaller number of animals infected with Serratia marcescens. Studies in human volunteers. Studies in human volunteers were done at the Institut Jules Bordet, Brussels, Belgium. When each antibiotic was given alone, mean peak serum levels ± standard deviation in micrograms per milliliter were 192 ± 45 for azlocillin, 14.5 ± 4 for cefotaxime, and 21 + 4 for amikacin. At 6 h after the infusion, serum drug concentrations were 30 ± 23 fg/ml for azlocillin, 0.8 + 0.5 yg/ml for cefotaxime, and 4 + 2 tig/ml for amikacin.

466 ZINNER ET AL TABLE 2. E. coli Organism Klebsiella species P. mirabilis Serratia species P. aeruginosa S. aureus ANTIMICROB. AGENTS CHEMOTHER. Combined in vitro activity of azlocillin plus amikacin, cefotaxime plus amikacin, and azlocillin plus cefotaxime against 197 strains of gram-negativle bacilli and S. aureus" No. of isolates 26 Antibiotics AZ + AM 39 AZ + AM 25 AZ + AM 18 AZ + AM 39 AZ + AM 50 AZ + AM Synergism No. of strains showing: FIC index Antagonism Median Range 0 1.0 0.5-2 0 0.52 0.19-1.0 4 1.03 0.15-2.25 20 19 0 0.75 0.05-2 30 9 0 0.75 0.31-1.5 23 15 1 0.75 0.06-4.13 18 7 0 0.63 0.19-1.5 18 7 0 0.63 0.13-1.5 12 9 4 1.0 0.13-8.1 14 4 0 0.63 0.15-1.5 14 4 0 0.56 0.14-1.25 15 0 3 0.5 0.09-8 33 6 24 14 25 14 34 16 22 28 23 27 'AZ, Azlocillin; AM, amikacin; CF, cefotaxime. Series from St. Mary's Hospital, London. 0 0.56 0.09-1.5 1 0.75 0.5-2.06 0 0.75 0.15-1 0 0.56 0.14-1.5 0 1.0 0.15-1.5 0 1.0 0.19-1.5 T'ABLE 3. Characteristics of infecting organisms and dclosage used in neutropenic mouse experiments MIC (yg/ml) LD:,, Inoculumil Dose (mng/kg) Organism Azlocillin Cefotaximne Amikacin WCFU/mil) (CFtU/nil) Azlocillin Cefotaximiie Amikacin E. coli 3 0.025 0.75 2 x 1(4 4.2 x 10' 300 3 0.75 K. pneumoniae 12 0.05 0.75 4 X 10I 2 X 10 30 0.75 0.75 S. marcescens 6 0.1 0.75 125 1.7 x 10 600 12 3 'LDr,5, 50% lethal dose; CFU, colony-forming units. 12 22 9 Indifference 14 4 13 Median peak serum bactericidal activity (SBA) was.1:16 against E. coli and Klebsiella species for all three combinations (Table 5). Most of the peak serum activity against these organisms in the cefotaxime plus amikacin group was due to cefotaxime, which produced similar activity when given alone. Median peak SBAs against these two organisms with azlocillin plus amikacin were similar to those found with each agent alone. With azlocillin plus cefotaxime, median SBA was considerably higher than with either drug alone. At 6 h after infusion (not shown in Table 5), 97 to 100% of sera from volunteers infused with cefotaxime plus amikacin and azlocillin plus cefotaxime had bactericidal activity -1:8 compared with only 55 and 39%7 for azlocillin plus amikacin against E. coli and Klebsiella species, respectively (P < 0.001). Against P. aeruginosa, azlocillin plus amikacin was slightly more active than the other two combinations. At 6 h after infusion, 80% of the sera from volunteers receiving azlocillin plus amikacin had SBA -1:8 compared with 47% for cefotaxime plus amikacin and 58c for azlocillin plus cefotaxime. For S. aureus, median peak SBA with cefotaxime plus amikacin was higher than with the other two combinations, but much of this activity was attributable to amikacin. At 6 h after infusion, none of the regimens produced much antistaphylococcal activity. DISCUSSION Although different organisms were tested in each of the three experimental studies included in this report, taken together, the data indicate that the activities of combinations of azlocillin plus amikacin, cefotaxime plus amikacin, and, to a lesser extent, azlocillin plus cefotaxime, are enhanced against the organisms most frequently

VOL. 20, 1981 encountered in neutropenic patients with fever. The first multihospital EORTC study (7) and other studies (9) showed that E. coli, Klebsiella species, P. aeruginosa, and S. aureus are likely pathogens in the febrile neutropenic patient. Successful treatment of gram-negative bacillemia in neutropenic patients has been achieved with combinations of broad-spectrum penicillins plus an aminoglycoside, cephalosporins plus an aminoglycoside, cephalosporins plus penicillins, and other combinations (11, 13, 18, 27). Klastersky and colleagues (14, 16) have demonstrated higher serum bactericidal activity and associated improved outcome in bacteremic patients treated with antibiotics showing in vitro synergism than in those receiving nonsynergistic combinations. Anderson et al. (2) also showed an improved outcome in neutropenic patients with TABLE 4. Survival of infected neutropenic mice in terms of treatment with subeffective doses of azlocillin, cefotaxime, and amikacin alone and in combination' Survivors/total tested (%) Treatment F. colib K. pneumo- S. marcesniae" cens d A. Azlocillin 2/40 (5) 3/40 (8) 5/20 (25) B. Cefotaxime 5/40 (13) 3/40 (8) 4/20 (20) C. Amikacin 2/40 (5) 6/40 (15) 4/20 (20) D. Azlocillin + 24/40 (60) 11/40 (28) 11/20 (55) amikacin E. Cefotaxime + 21/40 (53) 17/40 (43) 12/20 (60) amikacin F. Azlocillin + 11/40 (28) 3/40 (8) 8/20 (40) cefotaxime Saline 0/40 3/40 (8) 3/20 (15) "Summary of 204 experiments with 10 mice in each treatment group. MICs for infecting organisms are presented in Table 3. be. coli: for A versus D, C versus D, and C versus E, P < 0.001; for B versus E and D versus F, P < 0.01; for A versus F, B versus F, and E versus F, P < 0.05. ' K. pneumoniae: for B versus E and E versus F, P < 0.001; for A versus D, C versus E, and D versus F, P < 0.05. d S. marcescens: for A versus D and B versus E, P < 0.05. EVALUATION OF ANTIBIOTIC COMBINATIONS 467 gram-negative bacillemia who received synergistic antibiotics. Overall, there is some evidence that combination therapy is superior to singledrug therapy in cancer patients (2, 11, 13-16, 18, 27) Ċombined antibiotic activity can be investigated by various methods, and Moellering (19) has commented recently that the presence of synergism may differ with these different techniques. The demonstration of antibiotic synergism in experimental infections is less well standardized than in in vitro studies, and the results are more variable. Moreover, there are no universally accepted criteria for in vivo synergism. Nonetheless, antibiotic combinations have been shown to be more effective than single agents in experimental models of rabbit endocarditis due to S. aureus (23) and rabbit staphylococcal osteomyelitis (20). Animal experiments have demonstrated synergism against P. aeruginosa with carbenicillin plus gentamicin in neutropenic rats (25) and in a rabbit endocarditis model, although very high doses of aminoglycoside alone can produce bactericidal activity in sera greater than that found with the combination (3). Winston et al. (28) have demonstrated in neutropenic rats that cefazolin plus amikacin compared with amikacin alone produced greater rates of bacterial clearance and higher levels of SBA against strains of Klebsiella pneumoniae which were affected synergistically in vitro. In the present study (Table 4), deliberately suboptimal doses of azlocillin, cefotaxime, and amikacin were used alone and in combination to treat infections with gram-negative bacilli in neutropenic mice. Despite the many confounding variables mentioned above, combinationtherapy with azlocillin plus amikacin and cefotaxime plus amikacin were associated with significantly greater survival than was either drug alone at the same low doses used in the combination. The results were not as good with azlocillin plus cefotaxime. Obviously, this does not prove the presence of in vivo synergism rather TABLE 5. Median MICs and median SBA against gram-negative bacilli and S. aureus at 1 h after antibiotic infusion Azlocillin Cefotaxime Amikacin Azlocillin + Cefotax- Azlocillin Organism No. amikacin ime + ami- + cefotax- MICa SBA MIC SBA MIC SBA SBA kacin SBA ime SBA E. coli 10 3.0 1:32 0.007 1:512 1.5 1:8 1:32 1:512 >1:2,048 Klebsiella species 10 12 1:8 0.007 1:1,024 1.5 1:16 1:16 1:1,024 >1:2,048 P. aeruginosa 10 1.5 1:64 12 1:4 1.5 1:8 1:64 1:32 1:32 S. aureus 10 100 1:2 0.7 1:4 1.5 1:16 1:8 1:16 1:8 a MICs are given in micrograms per milliliter.

468 ZINNER ET AL. than an additive effect, but these data do show enhanced survival in neutropenic mice treated with these combinations. The results of in vitro checkerboard studies with these agents were consistent with these data. Antagonism was most frequently demonstrated with the double beta-lactam combination, and synergism was found for Enterobacteriaceae significantly more frequently with cefotaxime plus amikacin than with the other two combinations. The results of studies of human volunteers treated with these drugs alone or in combination parallel those from the in vitro and animal studies in that they demonstrate enhanced antibacterial activity of the combinations, at least against P. aeruginosa (Table 5). Because of the high level of activity achievable with each drug alone, the differences in SBA found with the combinations cannot be ascribed to a synergistic effect per se. However, these data do demonstrate that bactericidal activity which exceeds a dilution of 1:8 for E. coli and Klebsiella species can be achieved in patients treated with azlocillin plus amikacin, cefotaxime plus amikacin, and azlocillin plus cefotaxime. This level of SBA has been associated with a good clinical response (15). The former two combinations should provide sufficient inhibitory or bactericidal activity against P. aeruginosa, and the combinations containing amikacin should provide similar activity against S. aureus. The three approaches reported here utilize different techniques for investigating the interaction of antibiotics against bacteria frequently isolated from infected neutropenic cancer patients. Since different test organisms were used in each laboratory and since the three methods studied measure different aspects of antibiotic activity (19), it is of interest that the results obtained agree as well as they do. The in vitro checkerboard data showed more antagonism with the azlocillin plus cefotaxime combination than with the other combinations studied, especially against Klebsiella, Serratia, and Proteus species. Similarly, the neutropenic mouse model showed considerably less enhancement of survival in animals infected with E. coli and K. pneumoniae which were treated with this double beta-lactam combination than in infected animals treated with either beta-lactam in combination with amikacin. In the human volunteers, SBA was somewhat lower with azlocillin plus cefotaxime than with cefotaxime plus amikacin against S. aureus. The potential for antagonism of beta-lactam agents in combination has been raised in several studies. Acar et al. (1) showed that the in vitro activity of carbenicillin or ampicillin against ANTIMICROB. AGENTS CHEMOTHER. some strains of E. coli, Proteus species, and P. aeruginosa could be antagonized by cephaloridine, cloxacillin, or 6-amino-penicillanic acid. Graham and Medeiros (8) reported that carbenicillin was antagonized in vitro by cefoxitin, cefamandole, and cephaloridine when tested against several strains of Enterobacter cloacae, Enterobacter aerogenes, S. marcescens, Morganella morganii, and Citrobacter freundii. In clinical studies, there are very few welldocumented examples of antagonism with double beta-lactam combinations. In the EORTC trial (7), although not statistically significant, there was a trend, in febrile neutropenic patients whose granulocyte count did not rise above 500 mm3, for the poorest response to empirical therapy to be found in those treated with carbenicillin plus cephalothin (9 of 23 patients) compared with those who received cephalothin plus gentamicin (17 of 39) or carbenicillin plus gentamicin (27 of 44). Although the data presented do not suggest that combinations of beta-lactams should not be used clinically, they do suggest a cautious approach for the use in these agents in treating neutropenic patients with gram-negative bacilli or staphylococci. Taken together, the results of the three studies recorded here suggest that clinical trials of combinations of azlocillin plus amikacin and cefotaxime plus amikacin in these patients may prove useful. ACKNOWLEDGMENT The authors are members of the International Antimicrobial Therapy Project Group, European Organization for Research on Treatment of Cancer. LITERATURE CITED 1. Acar, J. R., L. D. Sabath, and P. A. Ruch. 1975. Antagonism of some penicillins by other penicillins and cephalosporins. J. Clin. Invest. 55:446-453. 2. Anderson, E. T., L. S. Young, and W. L. Hewitt. 1978. Antimicrobial synergism in the therapy of Gram negative rod bacteremia. Chemotherapy 24:45-54. 3. Archer, G., and F. R. Fekety, Jr. 1977. Experimental endocarditis due to Pseudomonas aeruginosa. II. Therapy with carbenicillin and gentamicin. J. Infect. Dis. 736:327-335. 4. Bennett, J. V., J. L. Brodie, J. J. Benner, and W. M. Kirby. 1966. Simplified accurate method for antibiotic assay of clinical specimens. Appl. Microbiol. 14:170-177. 5. Berenbaum, M. C. 1978. A method for testing for synergy with any number of agents. J. Infect. Dis. 137:122-130. 6. Ekwo, E., and G. Peter. 1976. Effect of clindamycin on aminoglycoside activity in a murine model of invasive Escherichia coli infection. Antimicrob. Agents Chemother. 10:893-898. 7. EORTC International Antimicrobial Therapy Project Group. 1978. Three antibiotic regimens in the treatment of infection in febrile granulocytopenic patients with cancer. J. Infect. Dis. 137:14-27. 8. Graham, W. C., and A. A. Medeiros. 1980. Antagonism by cephalosporins in gram-negative bacilli, p. 489-491.

VOL. 20, 1981 In J. D. Nelson and C. Grassi (ed.), Current chemotherapy and infectious diseases. American Society for Microbiology, Washington, D.C. 9. Gurwith, M. J., J. L. Brunton, B. A. Lank, A. R. Ronald, and G. K. M. Harding. 1978. Granulocytopenia in hospitalized patients. I. Prognostic factors and etiology of fever. Am. J. Med. 64:121-126. 10. Gurwith, M., J. L. Brunton, B. Lank, A. R. Ronald, G. K. M. Harding, and D. W. McCullough. 1978. Granulocytopenia in hospitalized patients. II. A prospective comparison of two antibiotic regiments in the empiric therapy of febrile patients. Am. J. Med. 64: 127-132. 11. Keating, M. J., G. P. Bodey, M. Valdivieso, and V. Rodriguez. 1979. A randomized comparative trial of three aminoglycosides. Comparison of continuous infusion of gentamicin, amikacin and sisomicin combined with carbenicillin in the treatment of infections in neutropenic patients with malignancies. Medicine 58:159-170. 12. Kelly, M. T., and J. M. Matsen. 1976. In vitro activity, synergism, and testing parameters of amikacin, with comparison to other aminoglycoside antibiotics. Antimicrob. Agents Chemother. 9:440-447. 13. Klastersky, J., R. Cappel, and D. Daneau. 1973. Therapy with carbenicillin and gentamicin for patients with cancer and severe infections caused by Gram negative rods. Cancer 31:331-336. 14. Klastersky, J., R. Cappel, and D. Daneau. 1972. Clinical significance of in vitro synergism between antibiotics in gram-negative infections. Antimicrob. Agents Chemother. 2:470-475. 15. Klastersky, J., D. Daneau, G. Swings, and D. Weerts. 1974. Antibacterial activity in serum and urine as a therapeutic guide in bacterial infections. J. Infect. Dis. 129:187-193. 16. Klastersky, J., F. Meunier-Carpentier, and J. M. Prevost. 1977. Significance of antimicrobial synergism for the outcome of Gram negative sepsis. Am. J. Med. Sci. 173:157-167. 17. Litchfield, J. T., Jr., and F. Wilcoxon. 1949. Simplified method of evaluating dose-effect experiments. J. Pharmacol. Exp. Ther. 96:99-113. 18. Love, L. J., S. C. Schimpff, C. A. Schiffer, and P. H. EVALUATION OF ANTIBIOTIC COMBINATIONS 469 Wiernik. 1980. Improved prognosis for granulocytopenic patients with Gram negative bacteremias. Am. J. Med. 68:643-648. 19. Moellering, R. C., Jr. 1979. Antimicrobial synergisman elusive concept. J. Infect. Dis. 140:639-641. 20. Norden, C. E. 1978. Experimental osteomyelitis. V. Therapeutic trials with oxacillin and sisomicin alone and in combination. J. Infect. Dis. 137:155-160. 21. Parsley, T. L., R. B. Provonchee, C. Glicksman, and S. H. Zinner. 1977. Synergistic activity of trimethoprim and amikacin against gram-negative bacilli. Antimicrob. Agents Chemother. 12:345-352. 22. Sabath, L. D. 1968. Synergy of antibacterial substances by apparently known mechanisms, p. 210-177. Antimicrob. Agents Chemother. 1967. 23. Sande, M. A., and K. B. Courtney. 1976. Naficillingentamicin synergism in experimental staphylococcal endocarditis. J. Lab. Clin. Med. 88:118-121. 24. Schimpff, S. C., W. Satterlee, V. M. Young, and A. Serpick. 1971. Empiric therapy with carbenicillin and gentamicin for febrile patients with cancer and granulocytopenia. N. Engl. J. Med. 184:1060-1065. 25. Scott, R. E., and H. G. Robson. 1976. Synergistic activity of carbenicillin and gentamicin in experimental Pseudomonas bacteremia in neutropenic rats. Antimicrob. Agents Chemother. 10:646-651. 26. Sosna, J. P., P. R. Murray, and G. Medoff. 1978. Comparison of the in vitro activities of HR756 with cephalothin, cefotaxime, and cefamandole. Antimicrob. Agents Chemother. 14:876-879. 27. Tan, W. K., L. S. Young, R. E. Black, D. J. Winston, S. R. Linne, R. J. Weinstein, and W. L. Hewitt. 1977. Comparative efficacy and toxicity of amikacin/ carbenicillin versus gentamicin/carbenicillin in leukopenic patients. Am. J. Med. 12:959-966. 28. Winston, D. J., H. J. Sidell, and L. S. Young. 1979. Antimicrobial therapy of septicemia due to Klebsiella pneumoniae in neutropenic rats. J. Infect. Dis. 139: 377-388. 29. Wise, R., A. P. Gillett, J. M. Andrews, and K. A. Bedford. 1967. Activity of azlocillin and mezlocillin against gram-negative organisms: comparison with other penicillins. Antimicrob. Agents Chemother. 13: 559-565.