|Home | About | Journals | Submit | Contact Us | Français|
There are currently no defined optimal therapies available for multidrug-resistant (MDR) Acinetobacter baumannii infections. We evaluated the efficacy of rifampin, imipenem, sulbactam, colistin, and their combinations against MDR A. baumannii in experimental pneumonia and meningitis models. The bactericidal in vitro activities of rifampin, imipenem, sulbactam, colistin, and their combinations were tested using time-kill curves. Murine pneumonia and rabbit meningitis models were evaluated using the A. baummnnii strain Ab1327 (with MICs for rifampin, imipenem, sulbactam, and colistin of 4, 32, 32, and 0.5 mg/liter, respectively). Mice were treated with the four antimicrobials and their combinations. For the meningitis model, the efficacies of colistin, rifampin and its combinations with imipenem, sulbactam, or colistin, and of imipenem plus sulbactam were assayed. In the pneumonia model, compared to the control group, (i) rifampin alone, (ii) rifampin along with imipenem, sulbactam, or colistin, (iii) colistin, or (iv) imipenem plus sulbactam significantly reduced lung bacterial concentrations (10.6 ± 0.27 [controls] versus 3.05 ± 1.91, 2.07 ± 1.82, 2.41 ± 1.37, 3.4 ± 3.07, 6.82 ± 3.4, and 4.22 ± 2.72 log10 CFU/g, respectively [means ± standard deviations]), increased sterile blood cultures (0% versus 78.6%, 100%, 93.3%, 93.8%, 73.3%, and 50%), and improved survival (0% versus 71.4%, 60%, 46.7%, 43.8%, 40%, and 85.7%). In the meningitis model rifampin alone or rifampin plus colistin reduced cerebrospinal fluid bacterial counts (−2.6 and −4.4 log10 CFU/ml). Rifampin in monotherapy or with imipenem, sulbactam, or colistin showed efficacy against MDR A. baumannii in experimental models of pneumonia and meningitis. Imipenem or sulbactam may be appropriate for combined treatment when using rifampin.
Acinetobacter baumannii is an important nosocomial pathogen worldwide (5, 35), with pneumonia, bacteremia, and surgical site and urinary tract infections being the most important infections caused by this organism (16). A Spanish study showed A. baumannii as the cause of nearly 9% of cases of ventilator-associated pneumonia (VAP) (2), with a crude mortality of 40% to 70% (14). A. baumannii may also cause meningitis and ventriculitis, especially in patients undergoing neurosurgical procedures or with head trauma (17), with mortality rates between 20% and 27% (5).
The well-known ability of A. baumannii to acquire resistance to almost all groups of available antibiotics leads to serious problems in the management of infections caused by multidrug-resistant (MDR) A. baumannii infections (5, 16). In these cases, carbapenems have been considered the treatment of choice. However, increasing numbers of carbapenem-resistant A. baumanii isolates have been reported worldwide (1, 28), prompting the search for other therapeutic options.
Sulbactam has been used successfully in cases of meningitis and pneumonia caused by A. baumannii (17, 21, 39). Colistin has good in vitro activity (37) but has shown contradictory results in clinical practice (12) and experimental models (23). Rifampin has demonstrated in vitro and in vivo bactericidal activities against MDR A. baumannii in an experimental pneumonia model (23), but rifampin-resistant mutants appear shortly after treatment initiation with rifampin alone (23, 27). The combination of rifampin plus imipenem has been evaluated in clinical infections caused by highly imipenem-resistant A. baumannii strains, with inconclusive results (33). Two clinical studies have shown efficacy rates of 76% to 100% for colistin plus rifampin in VAP, bacteremia, and meningitis (4, 25).
The aims of this study were to evaluate the efficacies of rifampin and its combinations with imipenem, sulbactam, and colistin in experimental pneumonia and meningitis models caused by MDR A. baumannii strains.
(The results of the manuscript were presented in part at the 13th European Congress of Clinical Microbiology and Infectious Diseases (ECCMID), Glasgow, United Kingdom, 2003, and at the 15th ECCMID, Copenhagen, Denmark, 2005.)
Four MDR A. baumannii strains (Ab506, Ab940, Ab1327, and Ab1417) isolated from patients with bacteremia and representing the most frequent clones isolated in our hospital (22) were studied. They were resistant to cefotaxime, imipenem, sulbactam, amoxicillin-clavulanate, and tetracycline and susceptible to colistin and rifampin. A. baumannii Ab1327 was used for the in vivo studies.
For in vitro assays, antimicrobials were used as standard laboratory powders. Imipenem was supplied from Merck Sharp and Dohme (Madrid, Spain), sulbactam was from PharmaSierra (Madrid, Spain), and rifampin and colistin sulfate were from Sigma-Aldrich (Madrid, Spain). For in vivo experiments, commercial vials were used: imipenem-cilastatin (Merck Sharp and Dohme), sulbactam (PharmaSierra), rifampin (Aventis, Madrid, Spain), and colistin methanosulfonate (1 mg, equivalent to 12,500 IU; Aventis, Bellon, France). The anesthetic used in the mouse pneumonia model was 5% (wt/vol) sodium thiopental administered intraperitoneally (i.p.) (B. Braun Medical S.A. Rubi, Barcelona, Spain), and ketamine (Ketolar, Parke-Davis, Madrid, Spain) and xylazine (Rompun, Bayer Healthcare, Kiel, Germany) administered intramuscularly (i.m.) were used in the rabbit meningitis model.
MICs of imipenem, sulbactam, rifampin, and colistin were determined according to standard methods (8, 9). Escherichia coli ATCC 25922 was used as a control strain. The breakpoints for resistance were those defined by the Clinical and Laboratory Standards Institute (8), except for rifampin, for which breakpoints from the French Society for Microbiology were used (9).
Bactericidal activities of antimicrobials were evaluated by the time-kill method (7). Time-kill curves were performed using concentrations of 1× the MIC, and bacterial growth was quantified after 0, 2, 4, 8, and 24 h of incubation at 37°C by plating 10-fold dilutions on sheep blood agar (SBA). The limit of detection was 10 CFU/ml, corresponding to 1 log10 CFU/ml. The in vitro synergy studies (imipenem plus sulbactam or rifampin or colistin, sulbactam plus rifampin or colistin, and rifampin plus colistin) with the four strains were performed. Concentrations of 1× the MIC of each antimicrobial in the combinations were used. Bacterial growth was measured after 0, 2, 4, 8, and 24 h of incubation at 37°C. An antimicrobial was considered bactericidal when a 3-log10 decrease in the CFU/ml was reached compared with the initial inoculum. Synergy was defined as a 2-log10 decreases in the colony count of the combination compared with that of the most active antibiotic (11).
C57BL/6 female mice, weighing 16 to 20 g (Universidad de Sevilla's Facility, Seville, Spain) were used for the pneumonia model experiment.
New Zealand female rabbits weighing 2.5 to 3 kg (Charles River Laboratories, Barcelona, Spain) were employed for the meningitis model experiment. Animals were housed in regulation cages and given free access to food and water. The studies were approved by the Ethics Committee of the University Hospitals Virgen del Rocío (Seville, Spain).
Serum antibiotic concentrations were determined in groups of 21 healthy mice after a single administration of imipenem (30 mg/kg of body weight, i.m.), sulbactam (60 mg/kg, i.m.), colistin (20 mg/kg, i.m.), or rifampin (25 mg/kg, i.p.). In sets of three animals and at 10, 15, 30, 60, 90, 120, and 150 min after administration, blood samples were obtained from anesthetized mice from the periorbital plexus.
Serum antibiotic concentrations were determined in groups of four healthy rabbits after a single administration of imipenem (120 mg/kg, i.m.), sulbactam (30 mg/kg, i.m.), colistin (12 mg/kg, i.m.), or rifampin (25 mg/kg, intravenously [i.v.]). Blood samples were drawn from the marginal vein of the ear at 5, 10, 15, 30, 60, 90, 120, 240, and 480 min. Also, in infected rabbits, serum and cerebrospinal fluid (CSF) rifampin and colistin concentrations were measured at 2, 4, and 6 h after administration of colistin (12 mg/kg, i.m.) and rifampin (25 mg/kg, i.v.).
Concentrations of imipenem, sulbactam, colistin, and rifampin were determined by using an agar diffusion bioassay with Micrococcus luteus ATCC 9341, A. baumannii ATCC 19606, Bordetella bronchiseptica ATCC 4617, and Bacillus subtilis ATCC 6633 as control strains, respectively (19). Antibiotic assays were performed in triplicate. The maximum serum concentration (Cmax, in mg/liter), area under the concentration-time curve (AUC, in mg·h/liter), elimination half-life (t1/2, in h), AUC/MIC, and Cmax/MIC were calculated by using the linear trapezoid method and the PKCALC program (34). Time during which the serum drug concentration remained above the MIC (TMIC, in h) was extrapolated from the regression line of the serum drug concentrations (13).
The doses of imipenem and sulbactam used in the pneumonia model were those that, when administered in combination with rifampin, prevented the appearance of rifampin-resistant mutants (27) and were in the ranges of plasma human drug concentrations after standard dosing for severe infections (sulbactam at 1 g, i.v., and imipenem at 500 mg, i.v.); in the meningitis model, doses of imipenem and sulbactam were chosen to reach Cmax levels similar to those in the pneumonia model. For rifampin and colistin, because there are no defined pharmacodynamic parameters that predict efficacy, doses of antimicrobials were chosen to obtain an AUC similar to that found in humans (3, 18, 32).
A previously characterized pneumonia model (31) was used as follows: anesthetized C57BL/6 mice (thiopental at 5% [wt/vol], i.p.) were infected by intratracheal instillation, using 50 μl of a final inoculum of 108 CFU/ml mixed 1:1 with 10% porcine mucin (Sigma-Aldrich). Therapy was initiated 4 h after the inoculation. Fifteen mice were randomly included as controls (no treatment), and the following treatment groups were used: imipenem, sulbactam, rifampin, colistin, imipenem plus sulbactam, rifampin plus imipenem, rifampin plus sulbactam, and rifampin plus colistin. The total daily doses of antimicrobials were the following: imipenem at 120 mg/kg, i.m.; sulbactam at 240 mg/kg, i.m.; rifampin at 100 mg/kg, i.p.; colistin at 60 mg/kg, i.m. Total daily dosages of the drugs therapies were divided in four doses for imipenem, sulbactam, and rifampin, and colistin was administered thrice daily. Mice were observed for 72 h for mortality, and the surviving animals were sacrificed 6 h after the last dose by i.p. administration of a lethal dose of sodium thiopental. Immediately after death, thoracotomy was carried out. Through a cardiac puncture, blood samples were taken and 100 μl was plated on SBA for qualitative cultures. Lungs were homogenized in 2 ml of sterile saline solution (Stomacher 80; Tekmar Co., Cincinnati, OH), and 10-fold dilutions were plated on SBA for quantitative cultures.
In five randomly selected mice, lung samples were processed for histopathological studies. The lungs were fixed with 10% formaldehyde, embedded in paraffin, and cut into 4-μm-thick sections. The slices included all pulmonary lobes to be studied by optical microscopy. They were processed according to standard methods for hematoxylin-eosin, periodic acid-Schiff, Gram, Masson's trichromic, and silver reticulin stains.
The experimental meningitis model described by Dacey and Sande (10) was used. The day before the inoculation, an acrylic helmet was affixed to the rabbit skull to provide stable fixing to the stereotactic frame. The day after, animals were anesthetized with 35 mg/kg (i.m.) of ketamine (Parke-Davis) and 5 mg/kg (i.m.) of xylazine (Bayer Healthcare). Prior to bacterial challenge, a sample of CSF was taken by an intracisternal extraction of 200 μl, to prove its sterility and ensure the proper site of inoculation was used. Meningitis was induced by inoculating the same volume of a saline suspension containing 107 CFU/ml of strain Ab1327. A CSF sample was drawn 12 h after bacterial inoculation (time zero), when CSF features of meningitis developed. Treatment groups were those that reached efficacy in the previous experimental pneumonia model. Eight rabbits were randomly included in one of five groups: control group (no treatment), rifampin (25 mg/kg, i.v.), colistin (12 mg/kg, i.m.), imipenem (120 mg/kg, i.m.) plus sulbactam (30 mg/kg, i.m.), rifampin plus imipenem (25 mg/kg, i.v., and 120 mg/kg, i.m., respectively), rifampin plus sulbactam (25 mg/kg, i.v., and 30 mg/kg, i.m., respectively), and rifampin plus colistin (25 mg/kg, i.v., and 12 mg/kg, i.m., respectively). A single dose of the antimicrobial alone or the combinations were administered. Sequential CSF samples were drawn through cisternal puncture at 2, 4, and 6 h after antimicrobial administration.
Studies of indices of meningeal inflammation (white blood cell [WBC] counts and lactic acid concentration) were determined at 0 h and 6 h. WBC counts were performed using a Neubauer chamber. CSF lactic acid concentrations were determined by an enzymatic and colorimetric method (LOX/PAP; Roche Diagnostics, Mannheim, Germany). Quantitative bacterial cultures were performed at 0, 2, 4, and 6 h, through 10-fold dilutions of CSF and plating of 100 μl on SBA. Immediately after the 6-h extraction of CSF, rabbits were sacrificed with the administration of sodium thiopental. After the sacrifice a craniotomy was performed and the brain was extracted, weighed,and then desiccated in a heater for 7 days at 100°C to measure brain water. Brain edema was defined as a more than 400% increase in the weight brain water content (36).
The variables analyzed were percent survival, bacterial lung concentration (mean ± standard deviation of the log10 CFU/g of lung tissue), and percent blood sterility. The two-tailed Fisher's test, analysis of variance (ANOVA), and the Dunnet and Tukey post hoc tests were used.
For the meningitis model experiment, the analyzed variables were bacterial CSF concentration (log10 CFU/ml of CSF), WBC concentration in CSF (cells/μl), lactate concentration in CSF (mmol/liter), and brain edema (grams of water/100 g [dry weight] of brain). Results are expressed as median values (P25/P75). The nonparametric Mann-Whitney and Wilcoxon tests were used. Differences were considered significant at a P value of <0.05.
All four isolates were resistant to imipenem and sulbactam but susceptible to rifampin and colistin. The MIC and minimal bactericidal (MBC) results of the antimicrobials are shown in Table Table1.1. Time-kill experiments with the antimicrobials alone or in combination are presented in Fig. Fig.1.1. Rifampin exhibited bactericidal activities for the four strains, showing a transient bactericidal effect with regrowth in the cases of the strains Ab1327 (between 8 and 24 h) and Ab1417 (between 4, 8, and 24 h), while strains Ab506 and Ab940 appeared sterilized with rifampin exposure in vitro. Imipenem showed bactericidal activity at some time points against the strains Ab940, Ab1327, and Ab1417. Sulbactam did not show bactericidal activity against any strain. Colistin only showed bactericidal activity against strain Ab940.
In the synergy studies, the combinations with rifampin were those that exhibited the best activity, especially rifampin plus colistin, which presented synergistic activity against the four strains at 2 h and sterilized all four strains within 2 or 4 h. The synergistic effects of rifampin plus imipenem or sulbactam could not be demonstrated in some cases because of the high bactericidal activity of rifampin alone. Except for rifampin plus sulbactam with strain Ab1327, all of the antimicrobial combinations with rifampin sterilized the bacterial cultures. The combination with the least synergistic effect was imipenem plus sulbactam (only against Ab506).
The serum pharmacokinetic and pharmacodynamic parameters of each antimicrobial in mice and rabbits are shown in Table Table2.2. The Cmax values of rifampin, imipenem, and sulbactam were similar in mice and rabbits; in the case of colistin the Cmax was less than one-third that in rabbits compared to mice but reached a higher AUC/MIC ratio. The serum and CSF levels of rifampin and colistin in infected rabbits are detailed in Fig. Fig.2.2. The ratios of CSF to serum levels at 2 h were 12.9% and 27.5% for colistin and rifampin, respectively.
The efficacies of the antimicrobials, expressed as survival, bacterial lung concentration, and sterility of blood cultures, are shown in Table Table33.
All treatments, alone or in combination, increased survival compared with the control group; survival with rifampin (71.4%) was higher than that observed with imipenem, sulbactam, or colistin (28.6%, 40%, and 40%, respectively; not significant [NS]). With respect to the combinations, imipenem plus sulbactam (85.7% versus 0%; P < 0.05), rifampin plus sulbactam (46.7% versus 0%; P < 0.05), and rifampin plus colistin (43.8% versus 0%; P < 0.05) improved survival in comparison with the control.
Monotherapy with rifampin cleared bacteria from lung tissue compared with the controls (3.05 ± 1.9 versus 10.6 ± 0.27; P < 0.05) and also compared with imipenem alone (3.05 ± 1.9 versus 7.87 ± 3.43; P < 0.05). Colistin also decreased the bacterial lung concentration compared to controls (6.82 ± 3.4 versus 10.6 ± 0.27; P < 0.05). Rifampin plus imipenem, sulbactam, or colistin reduced the bacterial concentration compared with the controls (2.07 ± 1.82, 2.41 ± 1.37, and 3.4 ± 3.07 versus 10.6 ± 0.27, respectively; P < 0.05). It must be noted that the combination of rifampin plus imipenem or sulbactam improved the results in terms of bacterial lung concentration compared with colistin in monotherapy. Finally, the combination of imipenem plus sulbactam also reduced the bacterial lung concentration compared to the control group (4.22 ± 2.72 versus 10.6 ± 0.27; P < 0.05).
Rifampin and colistin showed the best results among the monotherapies in sterilizing blood cultures (78.6% and 73.3%, respectively) compared with controls (0%; P < 0.05) and the other monotherapies (P < 0.05). Of the combinations, rifampin plus imipenem, sulbactam, or colistin cleared bacteria from blood compared with imipenem plus sulbactam (100%, 93.3%, and 93.8%, respectively, versus 50%; P < 0.05).
Lungs showed acute inflammation, characterized by diffuse and/or focal affectation of all lobes, with mild to severe infiltration of polymorphonuclear cells, sometimes forming segmentary abscesses, and mild to moderate infiltration of alveolar macrophages. Gram-negative bacterial colonies and alveolar hemorrhagic areas were also observed.
Table Table44 summarizes the results for bacterial CSF concentration, WBC CSF levels, and brain edema for the meningitis model.
All treatments reduced bacterial concentrations at 6 h with respect to those immediately before their administration (0-h time point; P < 0.05), with the exception of colistin, which reached only a reduction of 0.8 log10 CFU/ml. Rifampin in monotherapy or combined with colistin showed the maximal reduction in bacterial concentration (−2.6 and −4.4 log10 CFU/ml, respectively).
Lactate CSF levels were elevated at 12 h after inoculation (P < 0.05) (data not shown). WBCs were absent before inoculation, and there were increases in their levels during the experiments in all groups. All treatment groups showed brain edema.
The present study shows that rifampin is efficacious in the treatment of both experimental pneumonia and meningitis caused by imipenem-resistant A. baumannii. It is worth noting the efficacy observed with rifampin in the pneumonia model, in terms of improving survival and decreasing bacterial burden from lungs and blood. These results are in accordance with the in vitro bactericidal activities found against the four strains tested. However, it must be noted that for Ab1327 (used in the in vivo experiments) and Ab1417, there was regrowth after 8 and 4 h, respectively, which may have been due in part to the higher MBCs of rifampin against these strains compared with the other two strains. Another possible explanation is the development of rifampin resistance during the experiments, but with strain Ab1327 we have reported that the induction of rifampin resistance appears between 24 and 48 h of incubation (27).
Imipenem and sulbactam were not efficacious, as expected, based on the MIC of 32 mg/liter of both antimicrobials for the A. baumannii strain used, which prompted a TMIC of 0 h in the case of imipenem and below the value necessary to obtain efficacy in this experimental model (30). Colistin, in spite of its in vitro activity against the A. baumannii strain used in the in vivo experiments, showed a similar activity compared with imipenem or sulbactam in terms of survival or bacterial clearance in the pneumonia model.
The pneumonia model was chosen because pneumonia is the most common nosocomial infection among those caused by A. baumannii (29), and it is a well-characterized and reproducible model (31). Meningitis is not a frequent nosocomial infection, but it also has a high mortality rate, and the information about the treatment in cases produced by imipenem-resistant A. baumannii includes only a small series evaluating sulbactam (17) or cases treated with colistin (18, 26). The model described by Dacey and Sande (10) was followed in order to characterize experimental meningitis by A. baumannii for the first time. In the meningitis model, we used only the treatments that showed efficacy in the pneumonia model, to spare the use of animals. Also, the meningitis model has the limitation of using only a single dose of antimicrobials; thus, it may be that a more extended therapy would produce better results. However, the bacterial clearances observed with monotherapies in the meningitis model are in accordance with those in the pneumonia model.
The efficacy of rifampin in A. baumannii infections had been shown in experimental pneumonia models in mice (23, 38). Montero et al. (23), using three A. baumannii strains with a MIC of rifampin of 8 mg/liter, obtained 100% survival with rifampin and a significant decrease in the counts of bacteria from lungs; as in the results of the present study, imipenem and sulbactam were not efficacious in increasing survival or reducing the bacterial lung concentration when using a strain with MICs of these antimicrobials of 512 and 128 mg/liter, respectively (23). In another experimental pneumonia study in neutropenic mice (38), using two strains of A. baumannii with rifampin MICs of 8 and 4 mg/liter, rifampin significantly reduced the bacterial lung concentration, to the same degree as imipenem with both strains (38); in terms of survival, rifampin was better than imipenem in the experiments with an A. baumannii strain with intermediate susceptibility to imipenem (8 mg/liter). A recent study (35) in neutropenic mice also showed efficacy of rifampin in the pneumonia model with A. baumannii, using strains with MICs of 4 and 8 mg/liter, in reducing bacterial lung concentration and bacteremia. Although these latter results are difficult to compare, because of the small size of the groups (three mice each) and the short duration of the treatment (24 or 48 h), monotherapy with colistin was less efficacious than rifampin, and the addition of colistin to rifampin did not improve the results with rifampin alone.
Colistin was used in the present study as the methanosulfonate because this is the parenteral form used in humans (12). However, in spite of showing a bacterial lung reduction of approximately 3 log10 CFU/g with respect to the control group, colistin only improved the activities of imipenem or sulbactam in monotherapy approximately 1 log10 CFU/g, correlating with the absence of bactericidal activity in the time-kill experiments. These results are not in accordance with those found by Montero et al. (23) using three strains of A. baumannii, also with a MIC of colistin of 0.5 mg/liter; in that study, the efficacy of colistin was scarce or absent in reducing the bacterial lung concentration and it did not improve the survival in any case. This discrepancy may be explained by the lower colistin AUC (11.96 mg·h/liter) obtained with the dose employed in contrast with the AUC of 26.42 mg·h/liter obtained in the present study, which is similar to that in humans (23.43 mg·h/liter) using 1,000,000 IU (18). The moderate activity of colistin in the pneumonia model could be explained by its lack of bactericidal activity in vitro and poor distribution in tissues (20). The results in experimental studies are in accordance with clinical results. The mortality rate with colistin in cases of nosocomial pneumonia caused by colistin-susceptible A. baumannii is high, reaching 38% (15); this mortality rate is not superior to that found with imipenem (35.7%), suggesting that neither drug is an optimal treatment for severe infections caused by A. baumannii, despite their extensive clinical use.
The combination of rifampin with imipenem or sulbactam was evaluated because previous data showed that either imipenem or sulbactam prevented the appearance of rifampin-resistant mutants in vitro and in vivo when used concomitantly with rifampin (27). However, imipenem or sulbactam (MICs of 32 mg/liter in both cases) used in combination with rifampin did not improve the therapeutic efficacy of rifampin alone in the clearance of the infection or in survival in either experimental model. In the study of Montero et al. (23) the addition of imipenem to rifampin, using a strain with a MIC of 8 mg/liter of both antimicrobials, did not improve the results with rifampin alone. In the neutropenic pneumonia model, neither imipenem nor sulbactam (MICs of 0.5 mg/liter in both cases) improved the results of rifampin alone (38). The usefulness of the combination of imipenem or sulbactam plus rifampin in the treatment of A. baumannii infections seems to be reduced due to the prevention of rifampin resistance during the treatment, which has been shown both in vitro and in vivo when the MICs of imipenem or sulbactam are not higher than 32 mg/liter (27). In a clinical study evaluating the efficacy of rifampin plus imipenem in nosocomial A. baumannii infections, a 70% cure rate was achieved but with the development of high-level resistance to rifampin in 7 out of 10 patients, probably due to the fact that the MIC of imipenem was ≥64 mg/liter in all of the isolates (33).
In the pneumonia model, the combination of rifampin plus colistin is similar to that of rifampin in monotherapy. In the same way, in the experimental pneumonia model of Montero et al. rifampin maintained the same efficacy when colistin was added to the therapy (24). It is worth noting that in the meningitis model this combination was the best in reducing bacterial CSF concentration, resulting in a reduction of more than 4 log10 CFU/ml after 6 h of treatment. This in vivo activity agrees with the synergy observed in vitro between both antimicrobials.
Two clinical studies have evaluated the efficacy of rifampin plus colistin in severe infections caused by A. baumannii (4, 25). Motaouakkil et al. (25) found a 100% clinical cure in nine cases of bacteremia, but those authors did not provide the MIC of rifampin against the clinical isolates. Bassetti et al. (4, 25) obtained clinical and microbiologic cure in 8 out of 10 (80%) cases of bacteremia and in 14 out of 19 (73%) cases of nosocomial pneumonia. These results are similar to those found in imipenem-susceptible A. baumannii bacteremia treated with imipenem (87.5%) (6) and slightly better than that found in nosocomial pneumonia caused by susceptible A. baumannii strains when patients were treated with imipenem (survival of 64.3%) or colistin (survival of 62%) (15). Taking into account the efficacy of rifampin alone in the pneumonia model in the present study, it is possible that the good clinical results obtained with the combination of rifampin plus colistin depends on the bactericidal activity of rifampin and that colistin prevents the appearance of A. baumannii resistance to rifampin, as suggested by the results of Bassetti et al. (4).
The combination of imipenem plus sulbactam showed a marginal therapeutic effect in these experimental models. It was efficacious regarding survival in the pneumonia model, and in the meningitis model this combination slightly reduced the bacterial CSF concentration. This antimicrobial combination was tested by Song et al. (35) in the A. baumannii pneumonia model in neutropenic mice, and they also found worse results with the combination than with rifampin alone in the reduction of bacterial lung concentration and in the clearance of bacteremia, but with the limitations of this model detailed above.
In summary, the results in both models show that rifampin is efficacious in the treatment of severe infections, such as pneumonia or meningitis, caused by imipenem-resistant A. baumannii strains. However, rifampin must not be used alone because rifampin resistance appears after 24 h of monotherapy in the experimental murine pneumonia caused by A. baumannii (27). In this context, it is necessary to add another antimicrobial to rifampin to prevent the development of resistance. The results of the present study suggest that imipenem or sulbactam may be an appropriate option, if their MIC is not higher than 32 mg/liter. Finally, the addition of colistin to rifampin needs to be further evaluated, in order to know if colistin also prevents the appearance of rifampin-resistant mutants, as occurs with imipenem and sulbactam.
We are indebted to Carmen Cabellos, Sandra Ribes, and Alejandro Doménech from the Laboratory of Experimental Infection, Infectious Disease Service, Hospital Universitario de Belvitge, L'Hospitalet de Llobregat, Barcelona, Spain, for their technical assistance in the experimental meningitis model.
This work was supported in part by research grants (172/00) and (32/02) from the Consejería de Salud of the Junta de Andalucia and by the Spanish Network for the Research in Infectious Diseases (REIPI RD06/0008/0000), Instituto de Salud Carlos III, Ministerio de Ciencia e Innovación, Spain.
There are no conflicts of interest regarding this work.
Published ahead of print on 4 January 2010.