INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Acinetobacter baumannii is an aerobic, gram-negative, oxidase-negative, nonfermenting bacterium that has become an increasingly frequent cause of nosocomial infections, particularly in intensive care units (3, 4, 8, 11). It has a propensity to develop antibiotic resistance extremely rapidly (2). Successive surveys have shown increasing resistance in clinical isolates, and high proportions of strains have become resistant to older, commonly used antibiotics (6, 10, 15). Only newer antibiotics, such as broad-spectrum cephalosporins, imipenem, tobramycin, amikacin, and fluoroquinolones, remain useful. The recent development of more universally resistant strains of A. baumannii has made the search for effective therapies more important and urgent (9, 13, 14; M. Wolff, M. L. Joly-Gillou, R. Farionotti, and C. Carbon, Abstr. 37th Intersci. Conf. Antimicrob. Agents Chemother., abstr. B-8, 1997).

From 1996 through 1997, an outbreak of resistant A. baumannii infections involving 96 patients occurred in the intensive care units (ICUs) of Los Angeles County-University of Southern California (LAC-USC) Medical Center. The hospital clinical laboratory did routine susceptibility testing and reported that the isolates were resistant to imipenem, ceftazidime, cefotaxime, gentamicin, tobramycin, piperacillin-tazobactam, ticarcillin-clavulanate, ciprofloxacin, ofloxacin, and trimethoprim-sulfamethoxazole. The strains were moderately susceptible to ampicillin-sulbactam. Some strains were resistant to amikacin, and others were not. There was variability in susceptibility to amikacin among the strains, even among the same-patient strains, when testing was done by the clinical laboratory using an automated instrument. The amikacin susceptibility tests were repeated by using a reference method.

Since the traditional antibiotics used against gram-negative aerobic bacteria were not effective against the Acinetobacter isolates, we selected nontraditional antibiotics for potential use in eradicating these bacteria in cases where they were clinical pathogens. We determined bactericidal or bacteriostatic activities of the novel antibiotics by performing kill-curve studies on selected strains. In addition, we examined new combinations of drugs with a checkerboard configuration by looking for synergistic interactions.

(This work was presented in part at the 18th Annual Meeting of the Surgical Infection Society, 30 April to 2 May 1998, abstr. P18, p. 93.)

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

The A. baumannii strains were isolated from specimens obtained from patients in the ICUs at LAC-USC Medical Center. The specimens were processed by the clinical laboratory according to standard methods described in the Manual of Clinical Microbiology (7). Identification to genus and species levels and the original susceptibility tests were done with the Vitek automatic instrument (bioMerieux Vitek, Inc., Hazelwood Mo.). The outbreak strains were sent to a reference laboratory for typing by pulse gel electrophoresis (PGE). The medical records of patients from whom one of these outbreak strains was isolated were reviewed.

Fifteen of the A. baumannii isolates that were found to be multidrug resistant by the hospital clinical laboratory and 14 strains randomly selected were sent to the research laboratory and were tested for additional antibiotic susceptibilities by the agar dilution technique. Three of the outbreak strains plus one multiresistant, nonoutbreak strain were selected for kill-curve studies and synergy studies.

Agar dilution test. Antibiotic laboratory standard powders of azithromycin, sulbactam, ampicillin, doxycycline, trovafloxacin (Pfizer, Inc., Groton, Conn.), cefepime, amikacin (Bristol-Myers Squibb, Princeton, N.J.), meropenem (Zeneca Pharmaceuticals, Wilmington, Del.), erythromycin (Eli Lilly, Indianapolis, Ind.), minocycline, tetracycline (Wyeth Laboratories, Philadelphia, Pa.), ciprofloxacin (Bayer Pharmaceuticals, West Haven, Conn.), netilmicin, gentamicin, isepamicin (Schering-Plough, Kenilworth, N.J.), rifampin (Novartis Pharmaceuticals, East Hanover, N.J.), and polymyxin B (Sigma Chemicals, St. Louis, Mo.) were obtained from their respective manufacturers and reconstituted according to their instructions. Stock solutions were stored at 70°C. On the day of the test, serial twofold dilutions were prepared and added to molten Mueller-Hinton agar for preparation of plates. Ampicillin plus sulbactam were tested together in a 2:1 ratio and separately.

Kill curves. Four strains were selected for further studies. Three were outbreak strains isolated over a period of 2 months, each from patients in different ICUs. Strain 13009 was isolated from a blood culture from a patient in the surgical ICU; 12770 was from a blood culture in a patient in the burn ICU; and 13801 was from a sputum specimen from a medical ICU patient. The three strains were from patients in different environments who had been subjected to different antibiotic therapies and selective pressures. The fourth strain (12676) was a randomly selected blood culture isolate that exhibited a different susceptibility pattern from the pattern of the outbreak strains (susceptible to imipenem and resistant to azithromycin).

Synergy studies. The checkerboard method (5) was used to assess the activity of the antimicrobial combinations. Test tubes containing sulbactam plus azithromycin, doxycycline, trovafloxacin, or rifampin in 1 ml of Mueller-Hinton broth were prepared in a checkerboard configuration. The test strains were added at a final concentration of approximately 10⁵ CFU/ml. After overnight incubations, the MICs were interpreted. Tubes showing no growth were subcultured onto Mueller-Hinton agar to determine the minimal bactericidal concentration.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

There are 14 medical and 40 surgical ICU beds for adult patients at the LAC-USC Medical Center. The surgical beds include those in the general surgical, neurosurgical, and burn services. Ninety-six patients who had been in one of these ICUs had at least one specimen from which multiresistant A. baumannii was isolated. In Table 1, the specimen sources of the isolates are listed. The most frequent source of the isolate was the respiratory tract, with 41% of the patients having this as their only source of positive specimen. Wound and blood specimens accounted for 37 and 28% of specimen sources, respectively. PGE patterns of the outbreak strains showed the presence of one clone with eight subtypes based on a difference of one to three minor bands.

In the burn surgical ICU, there were 29 patients with A. baumannii infections. However, in this service, only one patient had the respiratory tract as the sole source of the organism. The majority, 96%, had more than one specimen site positive for the resistant bacteria. Twenty-one (72%) and 16 (55%) had positive wound and blood specimens, respectively.

In Table 2, the determinations of the MICs at which 50 and 90% of the isolates tested were inhibited (MIC₅₀s and MIC₉₀s) for 18 antimicrobials are listed for the outbreak strains and for the randomly selected strains. The outbreak strains of A. baumannii were resistant to ampicillin, cefepime, erythromycin, tetracycline, ciprofloxacin, and gentamicin. However, not all the strains were resistant to amikacin, with a MIC range between 8 and 32 µg/ml. The randomly selected nonoutbreak strains of Acinetobacter were resistant to ampicillin, erythromycin, tetracycline, ciprofloxacin, and gentamicin, just as the outbreak strains were. However, there were differences. Some of the randomly selected strains were resistant to azithromycin, netilmicin, and amikacin while the outbreak strains were not. Both groups of strains were susceptible to ampicillin-sulbactam, sulbactam alone at 16 µg/ml, meropenem, doxycycline, minocycline, and polymyxin B and were moderately susceptible to trovafloxacin. Ampicillin alone was not active. The nonoutbreak stains were also susceptible to cefepime in this study, and many of them were susceptible to other cephalosporins and imipenem, as determined in the hospital laboratory.

In the kill-curve studies, trovafloxacin showed some decrease in the number of organisms initially during the first 8 h (Fig. 1). However, after 24 h, the drug was essentially bacteriostatic. The same effect was seen with doxycycline, with an initial drop in numbers of organisms in the first 8 h of incubation followed by a subsequent static effect for the outbreak strains while having a bactericidal effect after 24 h on the nonoutbreak strain (Fig. 2).

View larger version (14K): [in this window] [in a new window]   FIG. 1.   The activity of trovafloxacin against strain 12676 typifies all four strains. While a slight decrease in CFU per milliliter was noted after 8 h, the drug was essentially bacteriostatic after 24 to 48 h.

View larger version (13K): [in this window] [in a new window]   FIG. 2.   The activity of doxycycline against strain 13801 typifies three of the strains. A slight decrease in CFU/milliliter was noted after 4 to 8 h, but regrowth started after 24 h, and the drug was essentially bacteriostatic. Strain 12770 was different in that doxycycline exerted a bactericidal effect after 24 h.

There were 28 kill-curve broths from which persistent colonies were grown when subcultured after 48 h and 7 days of exposure to the antibiotics. Two of the strains that survived after 1 week of exposure to trovafloxacin at one-half the MIC showed a twofold dilution increase in trovafloxacin MICs. The MICs of all antibiotics for the remaining 26 cultures were within 1 dilution of the MICs of the unexposed strains. MICs for none of the strains exposed to doxycycline were more than 1 dilution different from those for the preexposed strains.

Rifampin did not have consistent effects on all four strains tested. It showed rapid bactericidal activity against two of the outbreak strains (Fig. 3). Bacteriostatic effects were detected on the nonoutbreak strain and one outbreak strain (Fig. 4).

View larger version (12K): [in this window] [in a new window]   FIG. 3.   Rifampin showed rapid bactericidal activity against two of the outbreak isolates: strains 13081 and 12770.

View larger version (13K): [in this window] [in a new window]   FIG. 4.   Rifampin was bacteriostatic against the nonoutbreak strain 12676 and the outbreak strain 13009. The third outbreak strain was resistant to rifampin (MIC = 128 µg/ml).

Sulbactam had bactericidal effects only initially against all strains tested. After 48 h, there was regrowth (Fig. 5). MICs of sulbactam for 25 of the 28 regrowth strains were within 1 dilution of those for the preexposed strains. The MIC for one strain was 4 dilutions higher, one was 2 dilutions higher, and one was 2 dilutions lower.

View larger version (14K): [in this window] [in a new window]   FIG. 5.   The activity of sulbactam against strain 12270 typifies its activity against all the strains. There was initial bactericidal activity after 4 to 24 h, with regrowth occurring at 48 h.

Azithromycin tested as bactericidal on the outbreak strains that were susceptible to the antibiotic (Fig. 6). On the nonoutbreak strain, which was resistant to azithromycin (MIC, 32 µg/ml) the effect was bacteriostatic. MICs for 27 of the 28 persistent isolates were within 1 dilution of those for the preexposure strains. The MIC for one strain was 2 dilutions higher after 1 week of incubation at two times the MIC.

View larger version (13K): [in this window] [in a new window]   FIG. 6.   Azithromycin was bactericidal against strain 13009 and the other outbreak strains after 8 to 24 h. Azithromycin was not as active against the nonoutbreak strain (MIC = 32 µg/ml) and was very slowly bactericidal after 48 h and 7 days.

The results of the checkerboard synergy testing are in Table 3. Using the checkerboard combinations of azithromycin, rifampin, trovafloxacin, or doxycycline with sulbactam, only the combinations of azithromycin with sulbactam and rifampin with sulbactam showed a synergistic effect with the nonoutbreak strain. The other combinations resulted in additive or indifferent effects.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Of all the species in the genus, A. baumannii is the main species associated with outbreaks of nosocomial infection. In recent years, the species has emerged as particularly important in nosocomial infections in ICUs, probably related to the increasingly invasive diagnostic procedures used and the increasingly greater quantity of broad-spectrum antimicrobials used.

In our institution, 40 of the 96 patients from whom a resistant Acinetobacter organism was isolated had the organism in a single site. It can be argued that these patients may have been colonized as opposed to infected. The other 56 patients had more than a respiratory source positive for the organism. The patients with A. baumannii isolated from a wound, blood, or urine source can be described as infected with that pathogen. The 15 strains included in this study were isolated during the first 3 months of the outbreak that occurred throughout all the ICUs. Although their antibiograms showed some variation among them, the PGE demonstrated them to belong to a single clone with eight subgroups. The control strains were of different PGE types and had different antibiograms. The striking characteristic of the outbreak strains was their resistance to imipenem as opposed to the imipenem-susceptible control strains.

In the early 1970s, nosocomial Acinetobacter infections were treated successfully with gentamicin, ampicillin, naladixic acid, or carbenicillin either as single agents or in combinations (2). Successive surveys have demonstrated increasing resistance. The survey presented in this study reports an Acinetobacter organism resistant to most antibiotics, with few therapeutic choices remaining.

The outbreak strains tested in this survey were resistant to erythromycin and sensitive to azithromycin. Nonoutbreak strains showed resistance to erythromycin and variable susceptibility to azithromycin. In the kill-curve studies, azithromycin, with a MIC of 4 µg/ml, demonstrated rapid bactericidal activity for most strains and, with a MIC of 32 µg/ml, demonstrated a bacteriostatic effect on one strain. However, the potential activity of azithromycin may not be predicted by comparing MICs to levels in serum since the drug is concentrated within the intracellular and interstitial compartments of tissues (13, 14). The extravascular MICs may be a better measure of the potential use of this drug since levels in tissue can be much higher than levels in serum according to pharmacokinetics studies. Neu has suggested that achievable tissue levels be used as breakpoints for susceptibility. Ultimately, the susceptibility breakpoints will be established by clinical studies. In the synergy results presented in this study, synergy was seen with azithromycin plus sulbactam, with a MIC of 32 µg/ml for one strain.

Trovafloxacin was moderately active against multiresistant A. baumannii. It had much better activity against both the outbreak and nonoutbreak strains than ciprofloxacin tested by agar dilution and ofloxacin tested with the Vitek instrument. The drug was bacteriostatic. Synergy studies with trovafloxacin and sulbactam demonstrated additive effects.

Rifampin has been reported to have a strong in vitro bactericidal effect and a synergistic effect with beta-lactamase inhibitors on multiresistant A. baumannii (Wolff et al., 37th ICAAC). In our study, rifampin did not have a bactericidal effect on all strains tested and had a synergistic effect only on the nonoutbreak strain.

Doxycycline, minocycline, and tetracycline were tested. Tetracycline was found to be ineffective while the other two were active. Further studies were done with doxycycline because it would be a nontoxic, inexpensive agent to use. Doxycycline had a bacteriostatic effect against three of the four outbreak strains tested. No effect was seen by combining doxycycline with sulbactam.

Of the aminoglycosides tested, gentamicin, tobramycin, amikacin, and netilmicin, only netilmicin was effective against all the outbreak strains. There were several outbreak and nonoutbreak strains that demonstrated resistance to amikacin. Although the outbreak strains had MICs that are defined as amikacin susceptible, the MICs determined by agar dilution were high at 8 to 32 µg/ml. The MICs determined by the automated instrument were variable for the same patient isolates. All three types of aminoglycoside-modifying enzymes have been identified within clinical Acinetobacter strains (15).

In the kill-curve studies, short-term exposure of the Acinetobacter isolates to the antibiotics for 7 days or less did not result in any significant increases in resistance.

The increasing antimicrobial resistance of A. baumannii presents a tremendous challenge. It is important to continue to test all potential drugs, alone or in combination, to find ways of treating the serious infections that this organism can cause.