International Journal of Antimicrobial Agents

Transcription

1 International Journal of Antimicrobial Agents 33 (2009) Contents lists available at ScienceDirect International Journal of Antimicrobial Agents journal homepage: Efficacy of monotherapy and combined antibiotic therapy for carbapenem-resistant Acinetobacter baumannii pneumonia in an immunosuppressed mouse model Joon Young Song, Hee Jin Cheong, Jacob Lee, Ah Kyeong Sung, Woo Joo Kim Division of Infectious Diseases, Department of Internal Medicine, Korea University College of Medicine, Seoul, Republic of Korea article info abstract Article history: Received 29 May 2008 Accepted 10 July 2008 Keywords: Acinetobacter baumannii Pneumonia Colistin Rifampicin In vivo Acinetobacter baumannii is an important cause of nosocomial infection with increasing carbapenem resistance. The aim of this study was to compare the efficacy of colistin + rifampicin and imipenem + rifampicin combinations with that of several other antibiotic regimens against carbapenem-resistant A. baumannii pneumonia using an immunosuppressed mouse model. Three different A. baumannii strains with diverse resistance mechanisms (OXA-51-, IMP-1- and VIM-2-type -lactamases) were used. Among the monotherapy regimens, only rifampicin significantly reduced the bacterial load in lungs 24 h after infection with the OXA-51-producing strain. Addition of rifampicin to either imipenem or colistin yielded synergistic results after 48 h. Rifampicin was bactericidal against the IMP-1-producing strain, and only the imipenem + rifampicin combination yielded synergistic effects. In contrast, rifampicin alone was not effective against the VIM-2-producing strain, but the imipenem + rifampicin combination was bacteriostatic even at 24 h post-infection. Tigecycline and amikacin were not effective against any of the three strains. Rifampicin-based combinations were effective against A. baumannii bacteraemia and improved survival regardless of the strain type. Contrary to the similar minimum inhibitory concentration results, the antibacterial effects of rifampicin were quite different according to the strains; a tailored antibiotic strategy must be considered in treatment. Addition of rifampicin to either imipenem or colistin would be effective Elsevier B.V. and the International Society of Chemotherapy. All rights reserved. 1. Introduction Acinetobacter baumannii is an important cause of nosocomial infection worldwide and the recent marked increase in multidrugresistant (MDR) strains has made A. baumannii one of the most problematic pathogens, especially in Intensive Care Units (ICUs). Owing to its extraordinary ability to acquire antibiotic resistance, a large number of A. baumannii strains have been reported to be drug-resistant, even to carbapenems, since the late 1990s. Likewise, in Korea University Guru Hospital, Seoul, South Korea, many A. baumannii isolates from the ICU were resistant to carbapenem antibiotics and showed susceptibility only to colistin over the past several years. Of note, infections caused by MDR A. baumannii strains have an associated mortality of 25 34% attributable to inappropriate antibiotic treatment [1,2]. Corresponding author. Present address: Division of Infectious Diseases, Department of Internal Medicine, Korea University Guro Hospital, 97 Guro Dong-Gil, Guro Gu, Seoul, South Korea. Tel.: ; fax: address: heejinmd@medimail.co.kr (H.J. Cheong). Although A. baumannii can cause suppurative infections in virtually every organ system, pneumonia is the most serious nosocomial infection and has limited therapeutic options. In reports characterising carbapenem-resistant A. baumannii, colistin has shown excellent in vitro antibacterial activity, but in vivo animal model studies are insufficient to draw definitive conclusions [3,4]. In addition, rifampicin and tigecycline have been considered as potential therapeutic options based on in vitro studies [5,6], but in vivo data are also lacking for these compounds. As previously reported, the therapeutic effectiveness of antibiotic regimens against carbapenem-resistant A. baumannii might differ according to the underlying mechanism of antimicrobial resistance, especially since many diverse methods exist for producing carbapenemases [6]. For example, the VIM-2 carbapenemase isolated from an A. baumannii strain from South Korea showed significantly high resistance to carbapenems, whereas the OXA-51 carbapenemase was closely related to an efflux pump [6,7]. Moreover, some carbapenemase-producing A. baumannii showed markedly enhanced biofilm-forming capacities, which might increase their antibiotic resistance; as with Staphylococcus /$ see front matter 2008 Elsevier B.V. and the International Society of Chemotherapy. All rights reserved. doi: /j.ijantimicag

2 34 J.Y. Song et al. / International Journal of Antimicrobial Agents 33 (2009) aureus,an A. baumannii biofilm-associated protein has already been identified [8]. In this study, we aimed to compare the efficacies of colistin + rifampicin and imipenem + rifampicin combinations with several other antibiotic regimens against carbapenem-resistant A. baumannii using the neutropenic mouse pneumonia model. The effects of each regimen on lung bacterial loads, bacteraemia and survival were assessed. 2. Materials and methods 2.1. Bacterial strains Carbapenem-resistant A. baumannii strains were selected that are known to be resistant to almost all known antibiotics, including imipenem; colistin was the only exception. Three different strains with diverse resistance mechanisms (OXA-51-, IMP-1- and VIM-2-type -lactamases) were employed; all had been isolated from clinical specimens. Polymerase chain reaction (PCR) detected Ambler class B carbapenemase bla IMP-1, bla IMP-2, bla VIM-1 and bla VIM-2 and Ambler class D OXA-type carbapenemases using previously described methods [6] Antibiotic susceptibility tests Antibiotic susceptibility tests were performed using the agar dilution method according to the Clinical and Laboratory Standard Institute guidelines and were used to determine minimum inhibitory concentrations (MICs) against imipenem, colistin sulfate, tigecycline, sulbactam, amikacin, rifampicin, cefepime and ceftazidime [9,10]. Strains were considered resistant to carbapenems if the MICs against imipenem were 16 mg/l. The following concentrations were considered as susceptibility breakpoints for the other tested antimicrobials: colistin, 4 mg/l; tigecycline, 2 mg/l; sulbactam, 8 mg/l; and rifampicin, 2 mg/l. Escherichia coli ATCC and Pseudomonas aeruginosa ATCC were used as controls Mouse pneumonia model Immunocompetent specific-pathogen-free CD-1 (ICR) young female mice (average weight 25 g; 6 7 weeks old) were supplied by ORIENT BIO Inc. (Seongnam, South Korea). Mice were rendered neutropenic by injecting cyclophosphamide intraperitoneally (300 mg/kg body weight) in a volume of 0.2 ml 4 days beforea. baumannii inoculation in the lung. Mice were anesthetised by nasal inhalation of enflurane and infected by transtracheal injection with a fine-calibre needle, as follows: after a midline vertical incision (1 1.5 cm) at the shoulder level, the trachea was exposed and 100 L of bacterial suspension containing 10 8 colony-forming units (CFU)/mL (spectrophotometrically controlled) was injected with a 0.5 ml insulin syringe (BD Ultra-Fine TM II, mm; BD, Rockville, MD). Mice remained suspended in a vertical position for 3 5 min until respiration was stabilised and pneumonia development was monitored by chest radiography after 24 h. Inoculum sizes were confirmed by quantitative culture. Infected mice were randomised into groups (six mice per group) corresponding to the control or each treatment regimen. Mice within each group were also randomly assigned to a 24-h (three mice) or 48-h (three mice) group. There were 13 experimental groups (6 monotherapy regimens and 7 combined regimens), as follows: imipenem; colistin; rifampicin; low-dose tigecycline; high-dose tigecycline; amikacin; imipenem + rifampicin; colistin + rifampicin; high-dose tigecycline + rifampicin; imipenem + colistin; imipenem + sulbactam; imipenem + amikacin; and high-dose tigecycline + amikacin. All regimens were tested for efficacy against the OXA-51-producing strain, whilst only tigecycline (high-dose) and the regimens that were bactericidal against OXA-51 were evaluated against IMP-1- and VIM-2-producing strains. To evaluate therapeutic efficacy on pneumonia and bacteraemia, three allocated mice from each treatment group were sacrificed at 24 h and 48 h post-infection. Blood and lung tissue cultures from each time point were then compared with those in the control group. All lung samples were cultivated within 24 h after the death of each mouse. If the mouse assigned to the 48 h group died within 24 h, it was exchanged with another from the 24 h group; the order of exchange was randomly designated before infection Treatment protocol The indicated regimen was initiated 3 h after inoculation. Antimicrobial agents were given by intraperitoneal injection, with dosages as follows: colistin methanesulfonate, 1.25 mg/kg every 6 h (q6h) (daily dose 5 mg/kg); imipenem cilastatin, 50 mg/kg q6h (daily dose 200 mg/kg); sulbactam, 30 mg/kg q6h (daily dose 120 mg/kg); rifampicin, 25 mg/kg per day; low- and high-dose tigecycline, 5 mg/kg per day (daily dose 5 mg/kg) and 10 mg/kg every 12 h (q12h) (daily dose 20 mg/kg), respectively; and amikacin, 7.5 mg/kg q12h (daily dose 15 mg/kg). These doses were chosen according to previous pharmacokinetic and pharmacodynamic data from experimental models [11 14]. Since few published data exist regarding the appropriate dose of colistin in mice, the recommended dose in humans was applied Effects on lung bacterial loads Bacterial counts in the lungs were determined after 24 h and 48 h from the start of antimicrobial therapy. Six mice per regimen (three mice at each time point) were used. To eliminate the antibiotic carry-over effect, the mice in treatment groups were sacrificed no sooner than 3 h after the last dose of antibiotics. For quantitative bacteriological studies, whole lungs were removed, weighed and homogenised in 1 ml of saline. Ten-fold dilutions were performed and 100 L aliquots were plated on tryptone soy agar with 5% sheep blood for 24 h at 37 C. Once they appeared, colonies were counted for each dilution and each animal. Culture results were expressed as mean ± standard deviation (S.D.) of the differences in log 10 of CFU per gram of lung, as follows: results were determined in the control and each treated group at each of the two time points (24 h and 48 h) and the difference between two groups was calculated ( log = mean treated group mean control group ). Bactericidal activity was defined as 3 log increase in killing at each time point. For combination regimens, synergy was defined as a 2 log increase in killing compared with the most active component drug alone. Antagonism was defined as a 2 log decrease in killing compared with the most active component drug alone. No difference was defined as <1 log (increase or decrease) in killing. The lower limit of detection was 1.5 log 10 CFU/g [15] Effects on bacteraemia and survival Bacteraemia was investigated in blood samples collected by cardiac puncture. The rates of bacteraemia were estimated daily and compared between control and each antibiotic treatment group. The survival rates for all mice were recorded at 24 h and 48 h and survival was comparatively analysed among all antibiotic treatment groups and the control group.

3 J.Y. Song et al. / International Journal of Antimicrobial Agents 33 (2009) Table 1 Antibiotic susceptibilities of OXA-51-, IMP-1- and VIM-2-producing strains of Acinetobacter baumannii -Lactamase MIC (mg/l) a IMP COL CEF CAZ AMK RIF TIG OXA IMP VIM MIC, minimum inhibitory concentration; IMP, imipenem; COL, colistin; CEF, cefepime; CAZ, ceftazidime; AMK, amikacin; RIF, rifampicin; TIG, tigecycline. a Susceptibility breakpoints (mg/l) were as follows: IMP, susceptible 4, resistant 16 (according to Clinical and Laboratory Standards Institute (CLSI) criteria); COL, susceptible 4, resistant 8 (according to British Society for Antimicrobial Chemotherapy (BSAC) criteria); CEF, susceptible 8, intermediate 16, resistant 32 (CLSI); CAZ, susceptible 8, intermediate 16, resistant 32 (CLSI); AMK, susceptible 16, intermediate 32, resistant 64 (CLSI); RIF, susceptible 2, resistant 4 (according to Working Party Report of BSAC); TIG, susceptible 2, resistant 8 (Wyeth research) Statistical analysis All bacterial counts are presented as mean ± S.D. Kruskal Wallis test and Student s t-test were used to analyse intergroup differences in bacterial counts. To compare bacteraemia or mortality rates between groups, Fisher s exact test was used. In all tests, differences were considered statistically significant when the P-value was < Results 3.1. Antibiotic susceptibility of the challenge strains and characterisation of carbapenemase production MICs for imipenem, colistin, ceftazidime, cefepime, tigecycline, amikacin and rifampicin are shown in Table 1. Each of the three strains (OXA-51, IMP-1 and VIM-2) was resistant to all available antibiotics, including imipenem, but was susceptible to colistin and tigecycline. All of the three carbapenem-resistant A. baumannii strains contained bla OXA-51 in common, and two of them carried either bla IMP-1 or bla VIM-2 additionally Effects on lung bacterial loads Lung bacterial loads in antibiotic-treated and control mice are summarised in Table 2. In mice infected with the OXA-51-producing strain, rifampicin was the only monotherapy regimen that significantly reduced the bacterial load in lungs compared with the control at 24 h post-infection (P < 0.05). After 48 h, rifampicin and colistin were bactericidal ( 3 log decrease in bacterial load) and imipenem was bacteriostatic (Fig. 1). Combining rifampicin treatment with either tigecycline or amikacin did not yield a synergistic effect at either the 24 h or 48 h time points, but imipenem + rifampicin and colistin + rifampicin combinations were bactericidal (at 24 h and 48 h) and synergistic (at 48 h). The imipenem + sulbactam combination was bacteriostatic, reducing bacterial loads by 2.79 log at 48 h. Neither tigecycline nor amikacin were bacteriostatic or synergistic when combined with any other antibiotic agent. In mice infected with the IMP-1-producing strain, rifampicin was bactericidal at both the 24 h and 48 h time points. Tigecycline was totally ineffective against IMP-1-producing A. baumannii pneumonia. Addition of colistin to rifampicin did not significantly decrease lung bacterial loads more effectively than rifampicin alone, but the imipenem + rifampicin combination was synergistic (Fig. 2A). In mice infected with the VIM-2-producing strain, neither rifampicin nor tigecycline was an effective antibacterial agent in this pneumonia study (Fig. 2B), but the imipenem + rifampicin combination was bacteriostatic even at 24 h (1.56 log); this effect increased as time went on (2.52 log at 48 h). In comparison, colistin + rifampicin showed a moderate bacteriostatic effect (1.56 log) that was apparent only after 48 h from the initial infection Effects on the eradication rate of bacteraemia In bacteraemia induced by the OXA-51 strain, a statistically significant additive effect was noted when using the imipenem + rifampicin combination for 24 h (P < 0.01). At the 48 h time point, all rifampicin-based regimens provided significant blood clearance compared with the control group (P < 0.01) (Table 3). Table 2 Therapeutic effects on lung bacterial loads (expressed as mean ± standard deviation of log 10 of colony-forming units per gram of lung) at 24 h and 48 h after transtracheal injection: comparison among three different carbapenemase-producing Acinetobacter baumannii strains Therapeutic regimen OXA-51 strain IMP-1 strain VIM-2 strain 24 h 48 h 24 h 48 h 24 h 48 h Control a 9.06 ± ± ± ± ± ± 0.23 RIF 5.14 ± 0.37 b 3.93 ± 0.53 b 6.47 ± 0.36 b 5.77 ± 1.05 b 9.14 ± ± 0.61 COL 9.31 ± ± 0.98 b N.D. N.D. N.D. N.D. IMP 8.46 ± ± 1.40 b N.D. N.D. N.D. N.D. AMK 9.51 ± ± 0.55 N.D. N.D. N.D. N.D. TIG (5 mg/kg/day) 8.74 ± ± 0.37 N.D. N.D. N.D. N.D. TIG (20 mg/kg/day) 9.41 ± ± ± ± ± ± 0.41 COL + RIF 6.25 ± 0.21 b 3.42 ± 0.12 b 6.49 ± 0.69 b 5.69 ± 0.37 b 8.43 ± ± 0.12 b IMP + RIF 4.98 ± 0.25 b 2.57 ± 0.67 b,c 6.63 ± 0.53 b 2.58 ± 0.52 b,c 7.30 ± 1.09 b 6.30 ± 1.07 b TIG (20 mg/kg/day) + RIF 6.38 ± 1.37 b 3.39 ± 0.53 b N.D. N.D. N.D. N.D. AMK + RIF 6.55 ± 0.35 b 5.09 ± 1.21 b N.D. N.D. N.D. N.D. IMP + COL 9.89 ± ± 3.56 b N.D. N.D. N.D. N.D. IMP + SUL 8.47 ± ± 2.15 b N.D. N.D. N.D. N.D. IMP + AMK 8.97 ± ± 0.23 N.D. N.D. N.D. N.D. RIF, rifampicin; COL, colistin; IMP, imipenem; AMK, amikacin; TIG, tigecycline; SUL, sulbactam; N.D., not done. a At each time point, 39, 12 and 12 mice were used as controls for the OXA-51-, IMP-1- and VIM-2-producing strains, respectively. b Statistically significant differences noted in bacterial loads between the regimen and controls (P < 0.05). c Statistically significant differences noted in bacterial loads compared with the more active antibiotic alone (P < 0.05).

4 36 J.Y. Song et al. / International Journal of Antimicrobial Agents 33 (2009) Fig. 1. In vivo therapeutic efficacy of (A) monotherapy or (B) combined antibiotic regimens against the OXA-51-producing Acinetobacter baumannii strain: differences between the means of the treated and control groups ( log = mean treated group mean control group ), each of which included three mice. RFP, rifampicin; COL, colistin; IMI/IMP, imipenem; AMK, amikacin; TG, tigecycline; SUL, sulbactam. In IMP-1 A. baumannii-induced bacteraemia, combinations of imipenem + rifampicin (P < 0.01 versus control) and colistin + rifampicin (P = 0.03 and P < 0.01 versus control at 24 h and 48 h, respectively) effectively eradicated the bloodstream infection. Rifampicin alone did not significantly decrease bacteraemia. Similar to the results in the IMP-1-producing strain, the imipenem + rifampicin combination was the most effective against the VIM-2-producing strain at reducing bacteraemia (P < 0.01 versus control). Rifampicin (P = 0.03) and the colistin + rifampicin combination (P < 0.01) showed a significant effect on blood clearance compared with the control group after 48 h. Table 3 Bacteraemia eradication rate (n/n (%)) of each antimicrobial regimen at 24 h and 48 h after transtracheal injection: comparison among three different carbapenemaseproducing Acinetobacter baumannii strains Therapeutic regimen OXA-51 strain IMP-1 strain VIM-2 strain 24 h 48 h 24 h 48 h 24 h 48 h Control 0/39 0/39 0/12 0/12 0/12 0/12 RIF 0/3 2/3 (66.7) a 0/3 0/3 0/3 2/3 (66.7) b COL 0/3 0/3 N.D. N.D. N.D. N.D. IMP 0/3 1/3 (33.3) N.D. N.D. N.D. N.D. AMK 0/3 0/3 N.D. N.D. N.D. N.D. TIG (5 mg/kg/day) 0/3 0/3 N.D. N.D. N.D. N.D. TIG (20 mg/kg/day) 0/3 0/3 0/3 0/3 0/3 1/3 (33.3) COL + RIF 0/3 2/3 (66.7) a 2/3 (66.7) b 3/3 (100) a 0/3 3/3 (100) a IMP + RIF 2/3 (66.7) a 2/3 (66.7) a 3/3 (100) a 3/3 (100) a 3/3 (100) a 3/3 (100) a TIG (20 mg/kg/day) + RIF 0/3 2/3 (66.7) a N.D. N.D. N.D. N.D. AMK + RIF 1/3 (33.3) 2/3 (66.7) a N.D. N.D. N.D. N.D. IMP + COL 0/3 0/3 N.D. N.D. N.D. N.D. IMP + SUL 0/3 0/3 N.D. N.D. N.D. N.D. IMP + AMK 0/3 0/3 N.D. N.D. N.D. N.D. RIF, rifampicin; COL, colistin; IMP, imipenem; AMK, amikacin; TIG, tigecycline; SUL, sulbactam; N.D., not done. a Differences were statistically significant compared with the control group (Fisher s exact test, P < 0.01). b Differences were statistically significant compared with the control group (Fisher s exact test, P = 0.03).

5 J.Y. Song et al. / International Journal of Antimicrobial Agents 33 (2009) Fig. 2. In vivo therapeutic efficacy of various antibiotic regimens (monotherapy or combined) against the (A) IMP-1- and (B) VIM-2-producing Acinetobacter baumannii strains: differences between the means of the treated and control groups ( log = mean treated group mean control group ), each of which included three mice. RFP, rifampicin; TG, tigecycline; COL, colistin; IMP, imipenem Effects on the survival of infected mice Mortality rates are summarised in Table 4. In OXA-51 A. baumannii-infected mice, 11.5% (9/78) and 74.4% (29/39) did not survive longer than 24 h or 48 h without treatment, respectively. After 48 h, rifampicin-based regimens and the imipenem + sulbactam combination significantly prolonged survival compared with the control group (P = 0.03). In animals infected with the IMP-1-producing strain, mortality rates of the controls were 4.2% (1/24) and 100% (12/12) at 24 h and Table 4 Effect on mortality rates (n/n (%)) at 24 h and 48 h after transtracheal injection for each antimicrobial regimen: comparison among three different carbapenemase-producing Acinetobacter baumannii strains Therapeutic regimen OXA-51 strain IMP-1 strain VIM-2 strain 24 h 48 h 24 h 48 h 24 h 48 h Control 9/78 (11.5) 29/39 (74.3) 1/24 (4.2) 12/12 (100) 0/24 4/12 (33.3) RIF 0/6 0/3 a 0/6 0/3 b 0/6 0/3 COL 0/6 1/3 (33.3) N.D. N.D. N.D. N.D. IMP 0/6 1/3 (33.3) N.D. N.D. N.D. N.D. AMK 0/6 3/3 (100) N.D. N.D. N.D. N.D. TIG (5 mg/kg/day) 0/6 1/3 (33.3) N.D. N.D. N.D. N.D. TIG (20 mg/kg/day) 0/6 1/3 (33.3) 0/6 2/3 (66.7) 0/6 0/3 COL + RIF 0/6 0/3 a 0/6 0/3 b 0/6 0/3 IMP + RIF 0/6 0/3 a 0/6 0/3 b 0/6 0/3 TIG (20 mg/kg/day) + RIF 0/6 0/3 a N.D. N.D. N.D. N.D. AMK + RIF 0/6 0/3 a N.D. N.D. N.D. N.D. IMP + COL 0/6 1/3 (33.3) N.D. N.D. N.D. N.D. IMP + SUL 0/6 0/3 a N.D. N.D. N.D. N.D. IMP + AMK 0/6 1/3 (33.3) N.D. N.D. N.D. N.D. RIF, rifampicin; COL, colistin; IMP, imipenem; AMK, amikacin; TIG, tigecycline; SUL, sulbactam; N.D., not done. a Differences were statistically significant compared with the control group (Fisher s exact test, P = 0.03). b Differences were statistically significant compared with the control group (Fisher s exact test, P < 0.01).

6 38 J.Y. Song et al. / International Journal of Antimicrobial Agents 33 (2009) h, respectively. At the 48 h time point, significantly improved survival was achieved with all regimens containing rifampicin (P < 0.01 versus control). After 48 h, the mortality rate in animals infected with the VIM- 2-producing strain was 33.3% (4/12) in the untreated control group. Because of the low mortality rate associated with the VIM-2 strain, no statistically significant difference in survival among control and antibiotic-treated groups was observed. 4. Discussion This study was designed to compare various therapeutic options (based on previous in vitro reports) against nosocomial pneumonia induced by carbapenem-resistant A. baumannii. In this study, we found that rifampicin was the most active single agent and that imipenem + rifampicin combination therapy most effectively treated A. baumannii mouse pneumonia in our mouse model, regardless of the mechanisms by which the strains produced carbapenemase. Carbapenem resistance among nosocomial A. baumannii isolates is a serious problem, particularly in the ICU. Increasing rates of carbapenem resistance have led to widespread use of colistin to treat the diverse infectious diseases caused by A. baumannii. However, colistin treatment failures involving A. baumannii have been reported already and colistin has limitations as a therapeutic agent, i.e. a minimal post-antibiotic effect and heterogeneous colistin resistance among MDR isolates [16 18]. In addition to colistin, some potential therapeutic regimens have emerged, although none has been tested thoroughly. Carbapenem/sulbactam combination therapy has demonstrated a synergistic effect in MDR A. baumannii strains [4,6]. Although the study population was quite small and heterogeneous and did not contain a control group, a clinical trial of rifampicin in combination with either colistin or imipenem yielded encouraging results [3,19]. Tigecycline showed good in vitro bacteriostatic activity against carbapenem-resistant A. baumannii, but bacterial expression of a multidrug efflux pump might reduce tigecycline susceptibility [5,6,20]. Despite the large number of in vitro studies on carbapenemresistant A. baumannii, in vivo data are still quite limited. This in vivo study utilised a model that was designed to develop pneumonia in a more accurate manner, and this model was used to evaluate the effectiveness of various antibiotic regimens against MDR A. baumannii strains expressing three different kinds of carbapenemases. Previously, Song et al. [6] reported that antibacterial effects can differ according to resistance mechanisms, namely serine -lactamase, metallo- -lactamase (IMP- and VIM-type), efflux pumps and outer membrane protein changes. In this study, we found that rifampicin was effective against carbapenem-resistant A. baumannii, with the exception of the VIM-2 strain. Previously, all three strains used in this study showed resistant MIC levels to rifampicin (based on the British Society for Antimicrobial Chemotherapy report) and susceptible MIC levels to tigecycline (based on Wyeth research), but the results of this in vivo study showed the exact opposite for both compounds. These findings suggest that the standards of susceptible MICs of A. baumannii for rifampicin and tigecycline may need to be reconsidered. Rifampicin has been considered inactive against Gram-negative bacteria due to its hydrophobicity, negative charge and large molecular size, all of which could decrease its outer membrane permeation. However, as Li et al. reported [21], substantial changes might occur in the outer membrane of carbapenem-resistant A. baumannii isolates. Considering the relationship between lung bacterial loads and survival, we expected lower survival rates in some experimental groups within this study. Contrary to our expectation, no antibiotic-treated mice died within the first 24 h. For example, despite lung bacterial loads in amikacin- and tigecycline-treated groups that were similar to those in the controls, all of the treated mice survived for more than 24 h, but nine mice (11.5%) of the control group died within the first 24 h. However, the difference in 24 h survival between controls and amikacin/tigecycline-treated groups was not statistically significant; one mouse in each of the tigecycline groups (low/high dose) and two mice in the amikacin group were close to death at the 24-h time point. Unlike the study by Montero et al. [22], the present study comparatively studied A. baumannii strains that produced three different kinds of carbapenemase (OXA-51, IMP-1 and VIM-2). There is a chance that A. baumannii strains bearing different kinds of carbapenemases are associated with diverse additional resistance mechanisms and virulence factors. Recently, Lee et al. [23] reported that a PER-1 carbapenemase-producing A. baumannii strain showed significantly higher biofilm formation compared with isolates without this extended-spectrum -lactamase gene [23]. When we compared the differences among the three A. baumannii strains, we noticed two important points. First, we noted that VIM-2 A. baumannii-induced pneumonia was accompanied by a lower mortality rate compared with that induced by the OXA-51- or IMP-1-producing strains. This discordance between the degree of lung infection and survival might be related to the low virulence of the strain. Second, contrary to the similar rifampicin MIC results (range 4 8 mg/l) among the three strains, the antibacterial effects of rifampicin were quite different according to the study strains. This observation suggests that antibiotic strategies that are specifically tailored to each hospital, based on the type of prevailing carbapenemases isolated therein, would yield the most effective treatment for this pathogen. Rifampicin alone was bactericidal against OXA-51- and IMP-1-producing strains, but it showed no effect on the lung bacterial load when mice harboured the VIM- 2-producing strain. We presumed that the poor in vivo response in VIM-2 A. baumannii pneumonia resulted from its high mucinproducing characteristics, which could lead to biofilm formation. Biofilms are frequently resistant to antibiotic treatment, even if the bacterial strain is sensitive to the antibiotics in an in vitro MIC test [24,25]. Recently, Schaber et al. [26] reported that P. aeruginosa formed biofilms within 8 h of acute infection in a thermally injured mouse model. Although rifampicin was quite effective against carbapenemresistant A. baumannii strains, several studies have voiced concerns about the development of rifampicin resistance during treatment. Montero et al. [22] reported that they did not detect rifampicin resistance in an A. baumannii pneumonia model after 48 h of therapy, but Pachón-Ibáñez et al. [27] reported rifampicin-resistant (MIC 128 mg/l) mutants at a frequency of after 48 h and 72 h if rifampicin was applied alone. Clinically, Saballs et al. [3] reported resistance to high levels ( 256 mg/l) of rifampicin when using an imipenem/rifampicin combination therapy. They hypothesised that imipenem was unable to prevent growth of high-level rifampicin-resistant A. baumannii because it could not achieve the sustained high serum levels necessary to suppress carbapenem-resistant strains. It would be warranted to study rifampicin resistance selection with various carbapenem-resistant A. baumannii strains regarding inoculum effect, therapeutic regimens, biofilm production, etc. In conclusion, we compared the effectiveness of various antibiotic regimens against carbapenem-resistant A. baumannii in a neutropenic mouse pneumonia model. Based on our results, we recommend the addition of rifampicin to either imipenem or colistin. However, these in vivo data highlight a need for clinical trials to investigate treating MDR A. baumannii pneumonia with combination regimens.

7 J.Y. Song et al. / International Journal of Antimicrobial Agents 33 (2009) Funding: No funding sources. Competing interests: None declared. Ethical approval: All animal procedures were performed in accordance with the institutional guidelines for the humane handling, care and treatment of research animals. References [1] Lortholary O, Fagon JY, Hoi AB. Nosocomial acquisition of multiresistant Acinetobacter baumannii: risk factors and prognosis. Clin Infect Dis 1995;20: [2] Cisneros JM, Reyes MJ, Pachon J. Bacteremia due to Acinetobacter baumannii: epidemiology, clinical findings, and prognostic features. Clin Infect Dis 1996;22: [3] Saballs M, Pujol M, Tubau F. Rifampicin/imipenem combination in the treatment of carbapenem-resistant Acinetobacter baumannii infections. J Antimicrob Chemother 2006;58: [4] Wolff M, Joly-Guillou ML, Farinotti R, Carbon C. In vivo efficacies of combinations of beta-lactams, beta-lactamase inhibitors, and rifampin against Acinetobacter baumannii in a mouse pneumonia model. Antimicrob Agents Chemother 1999;43: [5] Pachón-Ibáñez ME, Fernández-Cuenca F, Docobo-Pérez F, Pachón J, Pascual A. Prevention of rifampicin resistance in Acinetobacter baumannii in an experimental pneumonia murine model, using rifampicin associated with imipenem or sulbactam. J Antimicrob Chemother 2006;58: [6] Song JY, Kee SY, Hwang IS. In vitro activities of carbapenem/sulbactam combination, colistin, colistin/rifampicin combination and tigecycline against carbapenem-resistant Acinetobacter baumannii. J Antimicrob Chemother 2007;60: [7] Hu WS, Yao SM, Fung CP, Hsieh YP, Liu CP, Lin JF. An OXA-66/OXA-51-like carbapenemase and possibly an efflux pump are associated with resistance to imipenem in Acinetobacter baumannii. Antimicrob Agents Chemother 2007;51: [8] Loehfelm TW, Campagnari AA. Identification and characterization of an Acinetobacter baumannii biofilm-associated protein. J Bacteriol 2008;190: [9] Clinical and Laboratory Standards Institute. Methods for dilution antimicrobial susceptibility tests for bacteria that grow aerobically. 7th ed. Approved standard. Document M7-A7. Wayne PA: CLSI; [10] Hogg GM, Barr JG, Webb CH. In-vitro activity of the combination of colistin and rifampicin against multidrug-resistant strains of Acinetobacter baumannii. J Antimicrob Chemother 1998;41: [11] Montero A, Ariza J, Corbella X. Efficacy of colistin versus beta-lactams, aminoglycosides, and rifampin as monotherapy in a mouse model of pneumonia caused by multiresistant Acinetobacter baumannii. Antimicrob Agents Chemother 2002;46: [12] Lefort A, Lafaurie M, Massias L. Activity and diffusion of tigecycline (GAR- 936) in experimental enterococcal endocarditis. Antimicrob Agents Chemother 2003;47: [13] Nannini EC, Pai SR, Singh KV, Murray BE. Activity of tigecycline (GAR-936), a novel glycylcycline, against enterococci in the mouse peritonitis model. Antimicrob Agents Chemother 2003;47: [14] Yin LY, Lazzarini L, Li F, Stevens CM, Calhoun JH. Comparative evaluation of tigecycline and vancomycin, with and without rifampicin, in the treatment of methicillin-resistant Staphylococcus aureus experimental osteomyelitis in a rabbit model. J Antimicrob Chemother 2005;55: [15] Lorian V. Antimicrobial combinations. In: Lorian V, editor. Antibiotics in laboratory medicine. 4th ed. Philadelphia, PA: Williams and Wilkins; pp [16] Li J, Rayner CR, Nation RL. Heteroresistance to colistin in multidrug-resistant Acinetobacter baumannii. Antimicrob Agents Chemother 2006;50: [17] Owen RJ, Li J, Nation RL, Spelman D. In vitro pharmacodynamics of colistin against Acinetobacter baumannii clinical isolates. J Antimicrob Chemother 2007;59: [18] Taccone FS, Rodriguez-Villalobos H, De Backer D. Successful treatment of septic shock due to pan-resistant Acinetobacter baumannii using combined antimicrobial therapy including tigecycline. Eur J Clin Microbiol Infect Dis 2006;25: [19] Motaouakkil S, Charra B, Hachimi A. Colistin and rifampicin in the treatment of nosocomial infections from multiresistant Acinetobacter baumannii. J Infect 2006;53: [20] Peleg AY, Adams J, Paterson DL. Tigecycline efflux as a mechanism for nonsusceptibility in Acinetobacter baumannii. Antimicrob Agents Chemother 2007;51: [21] Li J, Nation RL, Owen RJ, Wong S, Spelman D, Franklin C. Antibiograms of multidrug-resistant clinical Acinetobacter baumannii: promising therapeutic options for treatment of infection with colistin-resistant strains. Clin Infect Dis 2007;45: [22] Montero A, Ariza J, Corbella X. Antibiotic combinations for serious infections caused by carbapenem-resistant Acinetobacter baumannii in a mouse pneumonia model. J Antimicrob Chemother 2004;54: [23] Lee HW, Koh YM, Kim J. Capacity of multidrug-resistant clinical isolates of Acinetobacter baumannii to form biofilm and adhere to epithelial cell surfaces. Clin Microbiol Infect 2008;14: [24] Hoffman LR, D Argenio DA, MacCoss MJ, Zhang Z, Jones RA, Miller SI. Aminoglycoside antibiotics induce bacterial biofilm formation. Nature 2005;436: [25] Landry RM, An D, Hupp JT, Singh PK, Parsek MR. Mucin Pseudomonas aeruginosa interactions promote biofilm formation and antibiotic resistance. Mol Microbiol 2006;59: [26] Schaber JA, Triffo WJ, Suh SJ. Pseudomonas aeruginosa forms biofilms in acute infection independent of cell-to-cell signaling. Infect Immun 2007;75: [27] Pachón-Ibáñez ME, Jiménez-Mejías ME, Pichardo C, Llanos AC, Pachón J. Activity of tigecycline (GAR-936) against Acinetobacter baumannii strains, including those resistant to imipenem. Antimicrob Agents Chemother 2004;48: