Advance Access published September 16, 2004 Journal of Antimicrobial Chemotherapy DOI: 10.1093/jac/dkh435 JAC Post-antibiotic effect induced by an antibiotic combination: influence of mode, sequence and interval of exposure Ronald C. Li* and Mei C. Tang Department of Pharmacy, Faculty of Medicine, The Chinese University of Hong Kong, Shatin, Hong Kong Received 20 May 2004; returned 9 July 2004; revised 18 August 2004; accepted 18 August 2004 Objectives: The effects of mode, sequence and interval of antibiotic exposure on the post-antibiotic effect (PAE) induced by rifampicin and tobramycin were studied using Escherichia coli ATCC 25922 as the test organism. Methods: In triplicate, baseline PAEs were evaluated by exposing E. coli to rifampicin and tobramycin individually and simultaneously for 1 h. PAEs were further assessed in a second study, with the organism exposed first to rifampicin for 1 h, followed by a second 1 h tobramycin exposure, commencing at the beginning, middle and end of the PAE phase induced by rifampicin. The third study was similar to the above, but with the sequence of the two antibiotics reversed, i.e. tobramycin then rifampicin. Results: The PAE produced by simultaneous exposure of the combination showed an apparent additive interaction (PAE: 5.0 6 0.3 h) when compared with the PAE of individual antibiotics (rifampicin alone: 3.0 6 0.1 h; tobramycin alone: 1.5 6 0.1 h). However, an antagonistic interaction was observed in the second study, with a more pronounced degree of antagonism at the beginning, dissipating towards the end of the previous rifampicin PAE (PAE at the beginning: 2.6 6 0.3 h; the middle: 1.5 6 0.2 h; and at the end: 1.7 6 0.3 h). By subtracting the residual contribution from the first rifampicin exposure, the net average PAEs attributed to the second tobramycin exposure actually increased, from 20.4 to 1.7 h from the beginning to the end of the rifampicin PAE. For the third study, an additive interaction was again observed when the organism was exposed to tobramycin first (PAE at the beginning: 4.7 6 0.4 h; the middle: 3.7 6 0.7 h; and at the end: 3.1 6 0.4 h). The timing of the second rifampicin exposure had no impact to the interaction; after correction, the net mean PAEs attributed to the second rifampicin exposure were maintained at 3.2, 3.2 and 3.1 h. Conclusions: The present data suggest that the expression of interaction type on PAE by an antibiotic combination was dependent on the mode, sequence and interval of exposure. The impact of these variables should not be overlooked when clinical dosing regimens are optimized. Keywords: PAE, rifampicin, tobramycin Introduction The ability of antibiotic combinations to improve clinical efficacy is based on the following premises: (1) to produce synergistic/additive antimicrobial effects, (2) to reduce the probability of emergence of bacterial resistance, or delay its formation and (3) to expand the spectrum of antibacterial activity beyond that of the individual antibiotics alone. Information in the literature pertaining to the study of antibiotic combinations mainly focuses on their bactericidal and inhibitory activities by utilizing time kill studies and MIC measurements. 1 11 Few studies report the effects of antibiotic combinations on the post-antibiotic effect 12 16 (PAE). The design of most, if not all, of these studies required the test organism to be exposed to each antibiotic alone then to the combination simultaneously. When the combination produces a PAE similar to, longer or shorter than the sum of PAEs induced by the two antibiotics alone, the respective interaction is then categorized as additive, synergistic or antagonistic. However, such experimental design does not reflect the clinical situation, in which some antibiotic combinations are not administered simultaneously. This is particularly true when their regimens (e.g. twice a day versus three times a day) are not synchronized due to the different half-lives or their pharmacokinetic behaviours, physical incompatibility (e.g. b-lactams and aminoglycosides) or profound metabolic drug drug interactions.... *Correspondence address. Clinical and Exploratory Pharmacology, International Clinical Investigations, Sanofi-Synthelabo Research, 9 Great Valley Parkway, Malvern, PA 19355, USA. Tel: +1-610-889-6398; Fax: +1-610-889-6227; E-mail: Ronald.Li@Sanofi-Synthelabo.com... Page 1 of 5 JAC q The British Society for Antimicrobial Chemotherapy 2004; all rights reserved.
R. C. Li and M. C. Tang When antibiotics in combination are not administered simultaneously, it is typical that one antibiotic is administered prior to the other. Therefore, optimal antimicrobial responses should also consider the sequence and interval of dosing. At the time of this study, the authors were not aware of any published reports examining the impact of sequential dosing of antibiotic combinations on PAE. To gain further insights in this area of research, tobramycin and rifampicin were utilized in this study as a test combination against a standard strain of Escherichia coli. The selection of this antibiotic combination was based on its known mechanisms of action rather than its clinical relevance; tobramycin acts primarily by disrupting protein synthesis leading to altered cell membrane permeability and cell death whereas rifampicin inhibits DNA-dependent RNA polymerase activity in susceptible cells. As a result of the distinct mechanisms of action, the likelihood of a pharmacological interaction between the two antibiotics is high. In this study, particular attention was given to the three variables, e.g. mode, sequence and interval of exposure of the two antibiotics, and the relevance of their interactions to PAE. Materials and methods Test organism Lyophilized E. coli ATCC 25922, purchased from Difco Laboratories (Detroit, MI, USA), was used. After initial isolation, an organism from a single colony was maintained on agar slants at 48C. Culture media Mueller Hinton broth supplemented with 25 mg of Ca 2+ and 12.5 mg of Mg 2+ per litre (MHB-S) was used throughout. Nutrient agar (lot 73271JE) was employed for the colony count assay via a pour plate technique. Both culture media were purchased from Difco Laboratories, Detroit, MI, USA and were sterilized per manufacturer s instructions. Antimicrobial agents Rifampicin and tobramycin were purchased from Sigma Chemical Co., St. Louis, MO, USA. Aqueous stock antibiotic solutions were prepared and stored frozen at 208C before use. MIC MICs of the two antibiotics and test organism were measured by the macrodilution technique 17 after 18 h of incubation at 378C. Bacterial culture The test organism maintained on an agar slant was transferred to 10 ml of MHB-S and incubated overnight at 378C. This overnight culture was diluted with MHB-S and further incubated for 2 3 h at 378C to enter logarithmic growth. The actively growing culture was subsequently diluted with MHB-S to achieve turbidity matching that of a 0.5 McFarland standard. PAE assessments To start the experiment, 0.1 ml of the adjusted culture at 0.5 McFarland was introduced to 9.9 ml of antibiotic-containing MHB- S to yield a total volume of 10 ml. Three sets of experiments, each in triplicate, were carried out; (1) E. coli was exposed to rifampicin and tobramycin individually and simultaneously for 1 h each, (2) the organism was first exposed to rifampicin for 1 h, followed by exposure for 1 h to tobramycin, at the beginning (T0; immediately after antibiotic removal of the first rifampicin exposure), middle (T1 = 2 h) and end (T2 = 4 h) of the first PAE induced by rifampicin and (3) the studies described in (2) were repeated, but with the sequence of the two antibiotics reversed, i.e. tobramycin, then rifampicin, with T0 also started immediately after removal of the first tobramycin exposure, and at the middle (T1 = 1 h) and end (T2 = 2 h) of the first PAE induced by tobramycin. With the anticipation of variability in the PAE measurements, the above time schedules would conform to the intended study conditions, i.e. exposure of the second antibiotic at the beginning, middle and end of the PAE of the first exposure. At the end of every 1 h exposure, the antibiotic(s) was removed by washing three times using sterile 0.9% saline and centrifugation at 1200 g for 10 min. For all studies, the concentrations of rifampicin and tobramycin were 25 mg/l (1.6 3.2 MIC) and 2 mg/l (1 MIC), respectively. These concentrations were established in our preliminary studies to limit the possibility of the viable bacterial densities in samples falling below the 200 cfu/ml limit of quantification over the entire study period for all conditions. Throughout the experiments, cultures were kept at 378C using a calibrated water bath. Samples (0.1 ml) were withdrawn from the test culture hourly (up to 12 h) and submitted to the pour plate assay until steady growth was observed after removal of the last antibiotic exposure. PAE was determined as the difference between the time required for the viable count in the test culture to increase by one log unit immediately after antibiotic removal and the time required after the same procedure for the control culture. To avoid overgrowth of the control culture, dilution (10 100-fold) using pre-warmed MHB-S was performed at the same time points as scheduled antibiotic exposures during the study. For clarity, data pertaining to the control cultures (all in simple logarithmic growth) are not shown in the Figures described in the Results section. Data interpretation Interpretation of the interaction expressed by the combination was based on the comparison of PAE ascribed to the antibiotic used in the second exposure and the PAE of that antibiotic alone. The combination was considered to be additive, antagonistic and synergistic when the former PAE was similar to, shorter and longer than the latter, respectively. Results MICs The MICs of rifampicin and tobramycin were 8 16 and 2 mg/l, respectively. PAE Following 1 h of antibiotic exposure, the mean (±S.D.) PAE estimates were 1.5 ± 0.1 h for tobramycin alone and 3.0 ± 0.1 h for rifampicin alone. Simultaneous exposure to both antibiotics for 1 h resulted in a PAE of 5.0 ± 0.3 h. Figure 1 shows the changes in bacterial density over time under the three exposure conditions. The combination would, therefore, be considered additive. Both the PAE and extent of bacterial killing were greatest when the organism was exposed to the combination in comparison to the antibiotics alone. In addition, the extent of bacterial density reduction produced by the combination Page 2 of 5
Antibiotic combination and PAE Figure 1. The changes in mean (±S.D.) bacterial density over time showing the extent of viable count reduction and PAE (or growth suppression) after E. coli was submitted to 1 h of exposure to tobramycin alone (T), rifampicin alone (R) and tobramycin and rifampicin simultaneously (T + R). Note: timings of the antibiotic(s) exposure and removal are indicated by vertical and horizontal arrows, respectively. immediately before antibiotic removal was approximately the sum of those by the two antibiotics alone. In the second set of experiments, when the organism was sequentially exposed to rifampicin then tobramycin, mean (S.D.) PAEs for tobramycin were 2.6 ± 0.3, 1.5 ± 0.2 and 1.7 ± 0.3 h when it was introduced at the beginning, middle and end of the PAE of the preceding rifampicin exposure. Apparently, the PAE of tobramycin was shorter when its exposure was further from the PAE of the first rifampicin exposure. However, these PAE values were not only reflective of those from the second tobramycin exposure, but also the residual PAE from the first rifampicin exposure. To generate the net apparent PAE produced by the second tobramycin exposure, the residual PAE from the first rifampicin exposure needed to be subtracted. To accomplish this, the time difference between the removal of the first rifampicin exposure and the start of the second tobramycin exposure was obtained, e.g. T0 T0, T1 T0. The residual PAE of the first rifampicin exposure was then estimated by subtracting the above time difference from the PAE of rifampicin alone observed in the first study. Taking the tobramycin exposure in the middle of the rifampicin PAE as an example, the residual PAE was [PAE rifampicin alone (T1 T0)] or [3.0 h (2.0 h 0.0 h)] = 1 h, whereas the net PAE was equal to actual PAE measured residual PAE, i.e. 1.5 h 1 h = 0.5 h. The residual PAE was zero hour for the tobramycin exposure at the end of the rifampicin PAE. Therefore, subsequent to the removal of the residual rifampicin PAE from the actual PAE measured for the second tobramycin exposure, the respective net PAEs induced by the tobramycin were thus estimated to be, on average, 0.4, 0.5 and 1.7 h under the three exposure conditions. These estimates, thus, suggest that tobramycin failed to induce a PAE of its own when it was introduced at the beginning and during the PAE phase of the previous rifampicin exposure. By definition, antagonism was expressed by the two antibiotics when the PAE of the second tobramycin exposure was less than its own PAE alone, i.e. 1.5 h. Furthermore, the net PAE for the second tobramycin exposure increased from 0.4 to 1.7 h when it was moved from the beginning to the end of the first rifampicin PAE, thereby indicating a higher degree of antagonism at an early PAE phase of rifampicin. The different patterns of viable bacterial density versus time curves are shown in Figure 2. As the PAE of the first rifampicin exposure dissipated, tobramycin exposure towards the end showed more extensive bacterial killing, suggesting that the organism had regained its susceptibility to tobramycin. When the sequence of the two antibiotics was reversed, i.e. tobramycin/rifampicin, the respective mean (S.D.) estimates of PAE for rifampicin were 4.7 ± 0.4, 3.7 ± 0.7 and 3.1 ± 0.4 h, with the rifampicin exposure scheduled at the beginning, middle and end of the PAE induced previously by tobramycin. The net PAEs produced by the second rifampicin exposure were estimated to be 3.2, 3.2 and 3.1 h under the respective conditions. Since these PAE estimates were in line with that of rifampicin alone observed in the first set of experiments, the interaction expressed by the tobramycin/rifampicin exposure could be considered additive. Figure 3 shows the bacterial density versus time profiles under the above study conditions. From the Figure, two observations are noteworthy, (1) regardless of the timing of the second rifampicin exposure, the profiles converged at >9 h and (2) the drop in bacterial count immediately after the second rifampicin exposure was similar in terms of both extent and rate and was independent of the timing of the second rifampicin exposure. This suggests the preservation of its antimicrobial effect during the PAE of tobramycin. Figure 2. The changes in mean (±S.D.) bacterial density over time following the exposure of E. coli to rifampicin (R) first for 1 h, then another 1 h of tobramycin (T) at the beginning [R!T(B)], middle [R! T(M)] and end [R! T(E)] of rifampicin PAE. Note: timing of the antibiotic exposure and antibiotic removal are indicated by vertical and horizontal arrows vertical and horizontal arrows with matching shading patterns, respectively. Except for the first rifampicin exposure (arrows not shaded), the addition and removal of the second tobramycin exposure for each of the three conditions is labelled using a separate pattern of shading. Page 3 of 5
R. C. Li and M. C. Tang Figure 3. The changes in mean (±S.D.) bacterial density over time following the exposure of E. coli to tobramycin (T) first for 1 h, then another 1 h of rifampicin (R) at the beginning [T!R(B)], middle [T! R(M)] and end [T! R! (E)] of tobramycin PAE. For all three conditions, a parallel decline in bacterial counts following the addition of rifampicin was observed. Note: timing of the antibiotic exposure and antibiotic removal is indicated by vertical and horizontal arrows, respectively. Except for the first tobramycin exposure (arrows not shaded), addition and removal of the second rifampicin exposure for each of the three conditions is labelled using a separate pattern of shading. Discussion PAE has been extensively studied for single antibiotics; however, the study of antibiotic combinations has received much less attention. One of the reasons might relate to the constraint on the study design; the limited studies focusing on combinations mostly adopted a simultaneous exposure design in which the test organism is exposed to the two antibiotics at the same time. Such design sharply contrasts with the actual clinical situation in which individual antibiotics in a combination can be administered in a sequential manner. The effects of sequential dosing on the pharmacodynamic behaviour of the two study antibiotics have been demonstrated in this study. As the present PAE data reveal, the conclusion on the interaction type can vary depending on the experimental conditions for the same pair of antibiotics. When the test organism was exposed to the test combination simultaneously, the interaction appeared to be additive. However, exposure first to rifampicin antagonized the PAE expressed by the second (tobramycin) exposure, and such antagonism was time-dependent during the PAE of rifampicin. On the other hand, when the organism was exposed first to tobramycin, the second (rifampicin) did not elicit similar antagonism during the PAE of tobramycin, but rather an additive interaction with no timedependency. In terms of bacterial count, simultaneous exposure to the antibiotic combination produced the most extensive reduction as compared with each antibiotic alone. The extent of reduction, similar to PAE, appeared to be additive in nature (Figure 1). However, when rifampicin was applied first, the bactericidal effect exhibited by the tobramycin (applied second) increased along with a lower degree of antagonism expressed on PAE when it was introduced towards the end of the rifampicin PAE (Figure 2). In the third set of experiments, the similar and parallel reduction in bacterial count produced by the second rifampicin exposures was consistent with the additive effect observed on PAE when exposure to tobramycin was first (Figure 3). Interpretation of these data requires a broader understanding of the pharmacological activities of the two antibiotics. Rifampicin and tobramycin act on different bacterial targets; however, the former exhibits a relatively strong bacteriostatic effect against the test organism, whereas the latter is more bactericidal. Previous studies showed that bacteriostatic agents antagonize both bactericidal activity and the PAE of bactericidal agents. 18,19 Current data from the second set of studies point in the same direction, i.e. antagonism was observed when exposure to rifampicin preceded that to tobramycin, and it was not apparent when the sequence of exposure was reversed or exposure to the two antibiotics was simultaneous. In regard to the timing of the second antibiotic exposure, the complete pharmacodynamic profile has to be considered. Data showing the longest delay in bacterial regrowth from each of the three sets of experiments are presented in Figure 4. Although the test organism received a total of 1 h of rifampicin exposure plus 1 h of tobramycin exposure in all cases, the sequence of rifampicin/tobramycin (tobramycin exposure at the end of the rifampicin PAE) delayed bacterial regrowth for the longest time, whereas the delay was shortest when exposure was simultaneous. This suggests that the former was pharmacodynamically the most attractive. In terms of the extent of bacterial density reduction over the study period, the same conclusion can be Figure 4. The composite plot of mean (±S.D.) bacterial density over time for conditions showing the longest PAE observed in the three sets of experiments; tobramycin together with rifampicin for 1 h (T + R), 1 h of rifampicin then 1 h of tobramycin at the end of the rifampicin PAE [R!T(E)], and 1 h of tobramycin then 1 h of rifampicin at the beginning of the tobramycin PAE [T! R(B)]. The different pharmacodynamic profiles show a clear impact of the mode, sequence and interval of antibiotic exposure. Note: arrows identical to those in Figures 1 3 are used to label antibiotic addition and removal. Page 4 of 5
Antibiotic combination and PAE drawn. The positive relationship observed between the extent of bacterial killing and the length of PAE is in good agreement with the data we reported previously. 20 In summary, present data support the need to consider the mode, sequence as well as interval of exposure when an antibiotic combination is used because all these factors contribute to the two pharmacodynamic attributes studied, i.e. bactericidal activity and PAE. More importantly, these factors should not be overlooked when antibiotic combinations are utilized in the clinic. References 1. Sader, H. S., Huynh, H. K. & Jones, R. N. (2003). Contemporary in vitro synergy rates for aztreonam combined with newer fluoroquinolones and b-lactams tested against gram-negative bacilli. Diagnostic Microbiology and Infectious Disease 47, 547 50. 2. Mandal, S., Mandal, M. D. & Pal, N. K. (2003). Combination effect of ciprofloxacin and gentamicin against clinical isolates of Salmonella enterica serovar Typhi with reduced susceptibility to ciprofloxacin. Japanese Journal of Infectious Diseases 56, 156 7. 3. Sweeney, M. T. & Zurenko, G. E. (2003). In vitro activities of linezolid combined with other antimicrobial agents against Staphylococci, Enterococci, Pneumococci, and selected gram-negative organisms. Antimicrobial Agents and Chemotherapy 47, 1902 6. 4. Critchley, I. A., Sahm, D. F., Kelly, L. J. et al. (2003). In vitro synergy studies using aztreonam and fluoroquinolone combinations against six species of Gram-negative bacilli. Chemotherapy 49, 44 8. 5. Jung, R., Husain, M., Choi, M. K. et al. (2004). Synergistic activities of moxifloxacin combined with piperacillin-tazobactam or cefepime against Klebsiella pneumoniae, Enterobacter cloacae, and Acinetobacter baumannii clinical isolates. Antimicrobial Agents and Chemotherapy 48, 1055 7. 6. Erdem, I., Kucukercan, M. & Ceran, N. (2003). In vitro activity of combination therapy with cefepime, piperacillin-tazobactam, or meropenem with ciprofloxacin against multidrug-resistant Pseudomonas aeruginosa strains. Chemotherapy 49, 294 7. 7. Dawis, M. A., Isenberg, H. D., France, K. A. et al. (2003). In vitro activity of gatifloxacin alone and in combination with cefepime, meropenem, piperacillin and gentamicin against multidrug-resistant organisms. Journal of Antimicrobial Chemotherapy 51, 1203 11. 8. Jacqueline, C., Caillon, J., Le Mabecque, V. et al. (2003). In vitro activity of linezolid alone and in combination with gentamicin, vancomycin or rifampicin against methicillin-resistant Staphylococcus aureus by time-kill curve methods. Journal of Antimicrobial Chemotherapy 51, 857 64. 9. Gunderson, B. W., Ibrahim, K. H., Hovde, L. B. et al. (2003). Synergistic activity of colistin and ceftazidime against multiantibioticresistant Pseudomonas aeruginosa in an in vitro pharmacodynamic model. Antimicrobial Agents and Chemotherapy 47, 905 9. 10. Fish, D. N., Choi, M. K. & Jung, R. (2002). Synergic activity of cephalosporins plus fluoroquinolones against Pseudomonas aeruginosa with resistance to one or both drugs. Journal of Antimicrobial Chemotherapy 50, 1045 9. 11. Kato, K., Iwai, S., Sato, T. et al. (2002). In-vitro activity of ciprofloxacin combined with flomoxef against Bacteroides fragilis, compared with that of ciprofloxacin combined with clindamycin. Journal of Infection and Chemotherapy 8, 190 3. 12. Hostacka, A. (1997). Comparison of postantibiotic effects of imipenem and netilmicin alone and in combination against Pseudomonas aeruginosa. Arzneimittelforschung 47, 965 7. 13. Sood, P., Mandal, A. & Mishra, B. (2000). Postantibiotic effect of a combination of antimicrobial agents on Pseudomonas aeruginosa. Chemotherapy 46, 173 6. 14. Chan, C. Y., Au-Yeang, C., Yew, W. W. et al. (2001). Postantibiotic effects of antituberculosis agents alone and in combination. Antimicrobial Agents and Chemotherapy 45, 3631 4. 15. Nyhlen, A., Ljungberg, B., Nilsson-Ehle, I. et al. (2002). Postantibiotic effect of meropenem and ciprofloxacin in the presence of 5-fluorouracil. Chemotherapy 48, 182 8. 16. Gudmundsson, S., Erlendsdottir, H., Gottfredsson, M. et al. (1990). The postantibiotic effect induced by antimicrobial combinations. Scandinavian Journal of Infectious Diseases Supplementum 74, 80 93. 17. National Committee for Clinical Laboratory Standards. (1993). Methods for Dilution Antimicrobial Susceptibility Tests for Bacteria that Grow Aerobically Third Edition: Approved Standard M7-A3. NCCLS, Villanova, PA, USA. 18. Gudmundsson, S., Vogelman, B. & Craig, W. A. (1994). Decreased bactericidal activity during the period of the postantibiotic effect. Journal of Antimicrobial Chemotherapy 34, 921 30. 19. Li, R. C., Schentag, J. J. & Nix, D. E. (1993). The fractional maximal effect method: a new way to characterize the effect of antibiotic combinations and other nonlinear pharmacodynamic interactions. Antimicrobial Agents Chemotherapy 37, 523 31. 20. Li, R. C., Lee, S. W. & Kong, C. H. (1997). Correlation between bactericidal activity and postantibiotic effect for five antibiotics with different mechanisms of action. Journal of Antimicrobial Chemotherapy 40, 39 45. Page 5 of 5