JAC Cross-resistance patterns among clinical isolates of Klebsiella pneumoniae with decreased susceptibility to cefuroxime

Transcription

1 Journal of Antimicrobial Chemotherapy (2000) 46, JAC Cross-resistance patterns among clinical isolates of Klebsiella pneumoniae with decreased susceptibility to cefuroxime Helga Schumacher a *, Jens Scheibel b and Jens Kjølseth Møller a a Department of Clinical Microbiology, Aarhus University Hospital and b Department of Clinical Microbiology, The University Hospital in Copenhagen County, Denmark The frequency of decreased susceptibility to cefuroxime and quinolones and the correlation between these drug resistance traits was investigated in clinical isolates of Klebsiella pneumoniae from two Danish counties. Eighty-three randomly selected clinical isolates of K. pneumoniae with decreased susceptibility to cefuroxime were examined for cross-resistance patterns and the production of β-lactamases. The frequency of resistance to cefuroxime and ciprofloxacin has increased from <5% in 1990 to 15% and 7% in 1998, respectively. Two of the 83 isolates were multiply resistant and seemed to produce extended-spectrum β-lactamases. However, cross-resistance to ciprofloxacin and other classes of drug in 68% of the remaining isolates indicates that other resistance mechanisms, such as penetration barriers, had probably been selected in these Danish isolates. The susceptibility to ciprofloxacin decreased successively with decreasing susceptibility to cefuroxime for K. pneumoniae. This did not occur in cefuroxime-resistant Escherichia coli. Introduction Klebsiella pneumoniae isolates are generally susceptible to cefuroxime, since the SHV-1 β-lactamase produced by the majority of strains does not hydrolyse this cephalosporin. It is a matter of debate whether acquired resistance to one cephalosporin implies resistance to all cephalosporins among K. pneumoniae. 1 The fluoroquinolones seem to be an alternative for the treatment of serious infections caused by cefuroxime-resistant K. pneumoniae. A previous investigation of 20 cefuroxime-resistant K. pneumoniae showed that five isolates produced extended-spectrum β-lactamases (ESBLs), but all isolates had a 10-fold increase in ciprofloxacin MICs. 2 It is of clinical importance to note the possibility of cross-resistance to β-lactam antibiotics and quinolones. 3 Serious infections with K. pneumoniae are associated with a higher mortality rate than infections with other Enterobacteriaceae, so efficient initial antibiotic coverage is essential. 4 The aim of this study was to investigate the frequency of decreased susceptibility to cefuroxime and quinolones and the correlation between these drug resistance traits among clinical isolates of K. pneumoniae in two Danish counties. The results were compared with similar data for Escherichia coli. Eighty-three clinical isolates of K. pneumoniae with decreased susceptibility to cefuroxime were further examined for other cross-resistance patterns and production of ESBLs. Materials and methods Data from the Department of Clinical Microbiology, Aarhus University Hospital in Aarhus county (county A) and the Department of Clinical Microbiology, Herlev University Hospital in Copenhagen county (county B) were compared. Bacterial strains The frequency of resistance to cefuroxime and ciprofloxacin was investigated for isolates from two Danish counties. In county A, 83 isolates of K. pneumoniae with decreased susceptibility to cefuroxime were randomly selected in the periods 1 January to 1 August 1997 and 1 May to 23 July Repeat isolates (identical isolates from the same specimen within 3 weeks) were excluded. E. coli ATCC and other isolates with known MICs of cefuroxime and ciprofloxacin were included as controls for susceptibility testing and MIC determinations. E. coli *Corresponding author. Tel: ; Fax: ; helga.schumacher@inet.uni2.dk 2000 The British Society for Antimicrobial Chemotherapy 215

2 H. Schumacher et al. producing the β-lactamases TEM-1 and SHV-1, SHV-2, (supplied by A. Harris, Glaxo Wellcome, Greenford, UK), SHV-3 (supplied by D. Sirot, Université de Clermont- Ferrand, France), and SHV-4 and SHV-5 (supplied by T. Medeiros, Long Island, USA) were used as reference isolates for β-lactamase investigation. All bacteria were stored at 80 C. Antibiotics and discs Antibiotics were supplied by their respective manufacturers as standard powders. Discs prepared in-house contained gentamicin 20 g, cefuroxime 40 g, cefotaxime 25 g, ceftazidime 25 g, ciprofloxacin 2.5 g, chloramphenicol 50 g or tetracycline 20 g. Tablets from Rosco Neo Sensitabs (Taastrup, Denmark) contained cefuroxime 60 g, cefotaxime 30 g, ceftazidime 30 g or amoxycillin clavulanate 30 g/15 g. Etest strips containing cefuroxime, cefotaxime, ceftazidime, ciprofloxacin or ceftazidime clavulanate were from AB-Biodisk (Solna, Sweden). Media Nutrient beef broth, a basal culture medium (Statens Seruminstitut, Copenhagen, Denmark), contained 1000 ml water, 400 g beef infusion, 0.3 g glucose, 3.0 g sodium chloride, 2.0 g disodium hydrogen phosphate 12H 2 O and 10.0 g peptone in a volume of 7 ml; the ph was adjusted to ph 7.4. MacConkey agar (Statens Seruminstitut) contained 1000 ml water, 20.0 g peptone, 10.0 g lactose, 5.0 g sodium chloride, 106 ml ox bile, g neutral red, 9.0 g agar, ph 7.2; 23 g of this medium was used in 9 cm Petri dishes. Danish blood agar plates 5 were from Statens Seruminstitut. PDM agar plates, containing 31.5 g/l PDM antibiotic sensitivity medium (obtained as a powder from AB-Biodisk), were produced at Herlev University Hospital, Herlev, Denmark. Frequency of resistance ( ) Data on the frequency of resistance to cefuroxime and ciprofloxacin in county A, retrieved from MADS, a computerized microbiology reporting system, 6 were compared with similar data from county B, obtained from the ADBact microbiological reporting system (Autonic AB; RAM- STA, Skøldinge, Sweden). Cross-resistance Cross-resistance to cefuroxime and ciprofloxacin was investigated for all K. pneumoniae and E. coli isolates from 1992 to 1998 in county B by preparing frequency histograms of ciprofloxacin zone diameters for cefuroximesusceptible, -intermediate and -resistant isolates. Similar histograms of cefuroxime zones were made for ciprofloxacin-susceptible and -resistant isolates. Repeat isolates (identical isolates obtained from the same patient within 3 weeks) were excluded. Identification of bacteria The 83 isolates were identified with the API 20 E Bio typing system (biomérieux, Mary l Etoile, France). Susceptibility tests The methods used for routine susceptibility testing differed between counties A and B. County A. Here the susceptibility to cefuroxime, ceftazidime, cefotaxime, gentamicin, chloramphenicol, tetracycline, ciprofloxacin and nalidixic acid was determined with a prediffusion test, 7,8 the routine susceptibility test method in that county. In this method, there is a period of prediffusion, in which antibiotic-containing discs are placed on the surface of the agar and kept at 5 C for 18 h before inoculation of bacteria. An antibiotic gradient is formed in the agar before growth of the bacteria begins; the influence of growth rate and inoculum on zone formation is hence diminished. The effect of introducing a prediffusion time on the dynamics of zone formation can be expressed mathematically; the amount of antibiotic (m ) that is just capable of inhibiting microbial growth under certain test conditions is called the critical concentration, and can be expressed by the general formula ln m ln m 0 (x 2 /(4D t 0 )) where m 0 is the concentration of antibiotic applied to the agar surface, x is the distance between the reservoir and the edge of the zone of inhibition, D is the diffusion coefficient for the antibiotic under the test conditions and t 0 is the time at which the position of the zone of inhibition is determined. t 0 is determined by the size of inoculum and the growth rate of the organism. When a prediffusion period is introduced, t 0 is greater and the antibiotic concentration gradient determining the zone of inhibition is formed before bacteria are added, so the influence of growth rate and inoculum on formation of the zone is reduced similarly. 8 County B. In county B, the routine susceptibility test method is the Biodisk system (AB-Biodisk) on PDM agar plates. This method was used to determine susceptibility to cefuroxime and ciprofloxacin. Detection of β-lactamases β-lactamases were detected using a tablet approximation test. 7 Neo Sensitab tablets (Rosco) containing cefuroxime, ceftriaxone or ceftazidime were placed 1.5 and 2.0 cm from the edge of a tablet containing amoxycillin clavulanate. 216

3 Cross-resistance in Klebsiella pneumoniae ESBL production was inferred when the inhibition zone for the test antibiotics was extended by 3 mm towards the tablet containing amoxycillin clavulanate. For resistant isolates, with small zone diameters, the extension of the zones was measured using tablets placed 1.5 cm apart; larger zone diameter extensions were clearer when discs were 2 cm apart. For ESBL-negative strains, no extension was seen at either of these distances. 7 Isolates with decreased susceptibility to ceftazidime or cefotaxime were also tested for production of ESBLs with the ESBL Etest, as described by the manufacturer. Determination of MIC Cefuroxime, cefotaxime, ceftazidime and ciprofloxacin MICs were determined by the Etest as described by the manufacturer. An inoculum resulting in semiconfluent growth was used on all plates; plates were incubated for 18 h at 35 C before zone diameters and MICs were determined. Breakpoints for susceptibility, decreased susceptibility and resistance are shown in Table I. The breakpoints for the Biodisk system used in county A were recommended by the manufacturer of PDM sensitivity test medium (AB-Biodisk). In county B, breakpoints were calculated according to regression analysis and frequency histograms of MICs (agar dilution method) and zone diameters (prediffusion method on Danish blood agar) for a variety of clinical isolates. Characterization of β-lactamases The crude enzyme preparations used for characterizing β-lactamases were prepared as follows. Danish blood agar plates were inoculated with bacteria and incubated overnight. The bacteria were harvested and suspended in 5 ml phosphate buffer, ph 6.0. Polymyxin was added to a concentration of 0.1 mg/l to cause partial disintegration of the Table I. MIC breakpoints for the prediffusion method and the AB-Biodisk system MICs (mg/l) Antibiotic susceptible resistant Gentamicin Cefuroxime 2 a / 4 b 16 Ceftazidime 2 16 Cefotaxime 2 16 Nalidixic acid 4 16 Ciprofloxacin a /<1 b 4 a / 8 b Chloramphenicol 2 16 Tetracycline 2 16 a Prediffusion method. b AB-Biodisk methods. outer membrane 9 before application of 10 freeze thaw cycles. After centrifugation at 13000g for 30 min, the crude enzymes were stored at 20 C. Isoelectric focusing was performed using the Ready Gel System (Bio-Rad) as described by the manufacturer. This method is particularly suited for screening, but not for determining precise pi values. Isoelectric focusing was repeated for enzymes with bands different from that of SHV-1 (pi 7.6). Results Frequency of resistance in general in In county A, the frequency of decreased susceptibility to cefuroxime and ciprofloxacin among K. pneumoniae isolates was 15% and 7%, respectively (n 1267), while that in county B was 14% and 8.7%, respectively (n 219). About 70% of the ciprofloxacin-resistant isolates in county A were also resistant to cefuroxime, while all ciprofloxacin-resistant isolates in county B were also resistant to cefuroxime. Figure 1 shows the distribution of ciprofloxacin zone diameters among cefuroxime-susceptible, -intermediate and -resistant isolates of K. pneumoniae. The median ciprofloxacin zone diameter was 31 and 26 mm among cefuroxime-susceptible and -resistant isolates, respectively. Figure 1 also shows the frequency histogram of cefuroxime zone diameters for ciprofloxacin-resistant isolates, where all ciprofloxacin-resistant isolates were cross-resistant to cefuroxime. The distribution of ciprofloxacin zone diameters among cefuroxime-susceptible and -resistant E. coli is shown in Figure 2. Here the median ciprofloxacin zone diameter for cefuroxime-susceptible isolates was similar to that for cefuroxime-resistant isolates. Other resistance patterns in county A Of 1267 isolates from 853 patients (86% of the isolates were from urinary tract specimens), 83 K. pneumoniae isolates with decreased susceptibility to cefuroxime were randomly selected for further investigation. Of these 83 isolates, only 75 had decreased susceptibility to cefuroxime when retested. The frequency of decreased susceptibility to cefotaxime, ceftazidime, gentamicin, tetracycline, nalidixic acid and ciprofloxacin for the 75 isolates is shown in Table II together with similar frequencies for the same antibiotics among 356 cefuroxime-susceptible blood culture isolates from county A. All isolates with decreased susceptibility to ciprofloxacin were resistant to nalidixic acid. Six isolates were susceptible to ciprofloxacin but resistant to nalidixic acid. Concomitant decreased susceptibility to cefuroxime and ciprofloxacin was seen in 51 of the 75 cefuroximeresistant isolates, and further resistance to nalidixic acid, chloramphenicol and tetracycline was seen in 44 of the 75 isolates (Table II). Ten isolates had slightly decreased 217

4 H. Schumacher et al. inhibition zones for ceftazidime but MICs were 2 mg/l. Two isolates were highly resistant to cephalosporins; one of these was resistant to all antibiotics except the quinolones and the second was susceptible only to gentamicin and quinolones. Other resistance patterns were less prevalent. MIC determinations According to MICs, 69 and 36 of the 83 isolates showed decreased susceptibility to both cefuroxime and ciprofloxacin, respectively. Two isolates were resistant to cefotaxime and ceftazidime. The remaining isolates all had cefotaxime and ceftazidime MICs of 2 mg/l. β-lactamase production One of the two multiply resistant isolates was positive both in the disc approximation test and in the ESBL Etest. For all other cefuroxime-resistant K. pneumoniae there was no indication of production of ESBLs according to MICs, disc approximation tests and ESBL Etests. Isoelectric focusing of crude enzyme preparations from the 83 isolates showed Figure 1. Cross-resistance among Klebsiella pneumoniae visualized using frequency histograms of ciprofloxacin zone diameters among isolates that were (a) susceptible, (b) intermediate and (c) resistant to cefuroxime; (d) cefuroxime zone diameters among ciprofloxacinresistant isolates. The median is marked with a line on the x-axis. 218

5 Cross-resistance in Klebsiella pneumoniae Figure 2. Frequency histograms of ciprofloxacin zone diameters among (a) cefuroxime-susceptible and (b) cefuroxime-resistant Escherichia coli. The median is marked with a line on the x-axis. Table II. Cross-resistance between different antibiotics among isolates of Klebsiella pneumoniae obtained from Aarhus county which were either fully susceptible or showed decreased susceptibility to cefuroxime Isolates with decreased susceptibility Isolates susceptible to to cefuroxime (1997 8), from cefuroxime (1992 8), Antibiotic various sources (n 75) from blood (n 356) Ceftazidime 13 (17%) 1 ( 1%) Cefotaxime 3 (4%) 0 (0%) Ciprofloxacin 51 (68%) 20 (5%) Nalidixic acid 49 (65%) ND Chloramphenicol 68 (91%) 28 (7%) Tetracycline 67 (89%) 36 (9%) Gentamicin 2 (3%) 1 ( 1%) ND, not determined. that 81 isolates had bands similar to those from the reference isolate producing SHV-1, while the two multiply resistant isolates produced enzymes with different bands indicating production of other enzymes. The 10 isolates with slightly decreased ceftazidime susceptibility were not ESBL producers according to MICs, disc approximation test, ESBL Etests and isoelectric focusing of the β-lactamases. Discussion In 1990, 5% of K. pneumoniae from Danish intensive care units showed decreased susceptibility to cefuroxime and ciprofloxacin. 10 We have shown that the frequency of decreased susceptibility to cefuroxime has increased to about 15% in both counties while that to ciprofloxacin has increased to 7% and 9% in counties A and B, respectively, with a considerable degree of cross-resistance to both antibiotics. In county B the susceptibility to ciprofloxacin is not routinely determined for isolates of K. pneumoniae; it is usually only determined if a strain is also resistant to other antibiotics. This may result in the slightly higher frequency of ciprofloxacin resistance. The cross-resistance between cefuroxime and ciprofloxacin seen in county A (Table II) was illustrated differently in county B where the median ciprofloxacin zone diameter increased successively with increasing cefur- 219

6 H. Schumacher et al. oxime zone diameters (Figure 1). Such a gradual shift of decreased susceptibility to ciprofloxacin could not be demonstrated for cefuroxime-resistant isolates of E. coli (Figure 2), perhaps because there are other resistance mechanisms in E. coli, such as a higher level of chromosomal mediated β-lactamase production as shown previously. 11 Mutations in the gyrase and topoisomerase IV genes are believed to be the primary cause of quinolone resistance. 12 Less frequent mechanisms are multiple antibiotic resistance genes (mar genes), 13 active efflux 14 and changes in outer membrane proteins. 15,16 They generally result in a low level of quinolone resistance and cross-resistance to chloramphenicol, tetracyclines and β-lactams, 3,13 16 as was found in this study (Table I). The low-level resistance may be shifted to higher levels and result in failure of therapy. 17 However, it has been shown previously that threetimes-daily dosing of ciprofloxacin (with 400 mg) improved treatment of K. pneumoniae with an MIC of 0.5 mg/l. 18 Decreased susceptibility to ciprofloxacin seems difficult to detect by routine susceptibility tests used in Denmark (unpublished data). Similar problems with Enterobacter cloacae and Salmonella spp. have been reported Recently, it was shown that salmonellae with reduced ciprofloxacin susceptibility were uniformly resistant to nalidixic acid. 22 This indicates that resistance to a quinolone with lower activity is likely to be associated with decreased susceptibility or resistance to the newer and more active fluoroquinolones. Our study indicates that K. pneumoniae with a low level of resistance to ciprofloxacin may also be resistant to nalidixic acid. It will be interesting to see if this applies to other Enterobacteriaceae as well. In Denmark, only two reports of ESBL-producing K. pneumoniae have been published. 2,23 The SHV-1 enzymes usually produced by K. pneumoniae are less able to hydrolyse cephalosporins than other SHV enzymes (ESBLs). In this study, only two isolates seemed to produce enzymes other than SHV-1, presumably ESBLs. They were easily detected using an ordinary susceptibility test. The third-generation cephalosporins may still be useful for treating infections caused by K. pneumoniae with decreased susceptibility to ciprofloxacin, since they seem less affected by a penetration barrier than cefuroxime 24,25 and the use of higher doses and longer duration of therapy may be of benefit. 26 In conclusion, considerable cross-resistance was seen between cefuroxime and ciprofloxacin among the K. pneumoniae isolates tested, while multiple resistance and production of ESBLs was only detected in a few isolates. Susceptibility to ciprofloxacin decreased successively with decreasing susceptibility to cefuroxime, as shown by the frequency histograms, which are useful for detecting and comparing combinations of resistance traits. Guidelines for detection of decreased susceptibility and sufficient antibiotic therapy may need to be adjusted when a stepwise shift in resistance is seen. Acknowledgements We thank Steen Hoffmann for helpful discussions, and Sami Majeed, Marianne Bøgild and Kirsten Jensen for valuable technical assistance. References 1. Rice, L. B., Yao, J. D., Klimm, K., Eliopoulos, G. M. & Moellering, R. C. (1991). Efficacy of different β-lactams against an extendedspectrum β-lactamase-producing Klebsiella pneumoniae strain in the rat intra-abdominal abscess model. Antimicrobial Agents and Chemotherapy 35, Schumacher, H., Skibsted U., Hansen, D. S. & Scheibel, J. (1997). Cefuroxime resistance in Klebsiella pneumoniae: susceptibility to cefotaxime and ceftazidime despite production of ESBLs. Acta Pathologica Microbiologica et Immunologica Scandinavica 105, Sanders, C. C., Sanders, W. E., Goering, R. V. & Werner, V. (1984). Selection of multiple antibiotic resistance by quinolones, β-lactams, and aminoglycosides with special reference to crossresistance between unrelated drug classes. Antimicrobial Agents and Chemotherapy 26, Pedersen, G., Schønheyder, H. C. & Sørensen, H. T. (1997). Antibiotic therapy and outcome of monomicrobial gram-negative bacteraemia: A 3-year population-based study. Scandinavian Journal of Infectious Diseases 29, Jensen, K. T., Schønheyder, H., Pers, C. & Thomsen, V. F. (1992). In vitro activity of teicoplanin and vancomycin against grampositive bacteria from human clinical and veterinary sources. Acta Pathologica Microbiologica et Immunologica Scandinavica 100, Møller, J. K. (1984). A microcomputer-assisted bacteriology reporting and information system. Acta Pathologica Microbiologica et Immunologica Scandinavica B 92, Schumacher, H., Bengtsson, B., Bjerregaard-Andersen, H. & Jensen, T. G. (1998). Detection of extended spectrum β-lactamases. Acta Pathologica Microbiologica et Immunologica Scandinavica 106, Thomsen, V. F. (1962). Correlation of the plate-dilution method to the agar diffusion method (disc- and tablet-method) with a special view to the importance of pre-diffusion. Acta Pathologica Microbiologica et Immunologica Scandinavica 54, Vaara, M. & Vaara, T. (1981). Outer membrane permeability barrier disruption by polymyxin in polymyxin-susceptible and -resistant Salmonella typhimurium. Antimicrobial Agents and Chemotherapy 19, Schumacher, H., Bremmelgaard, A., Holten-Andersen, W. R., Rasmussen, H. & Højbjerg, T. (1994). Resistant Gram-negative bacteria at seven intensive care units in Denmark. Ugeskrift for Læger 156, Schumacher, H., Skibsted, U., Skov, R. & Scheibel, J. (1996). Cefuroxime resistance in Escherichia coli. Resistance mechanisms and prevalence. Acta Pathologica Microbiologica et Immunologica Scandinavica 104, Drlica, K. & Zhao, X. (1997). DNA gyrase, topoisomerase IV, and the 4-quinolones. Microbiology and Molecular Biology Reviews 61,

7 Cross-resistance in Klebsiella pneumoniae 13. Alekshun, M. N. & Levy, S. B. (1997). Regulation of chromosomally mediated multiple antibiotic resistance: the mar regulon. Antimicrobial Agents and Chemotherapy 41, Levy, S. B. (1992). Active efflux mechanisms for antimicrobial resistance. Antimicrobial Agents and Chemotherapy 36, Chen, H. Y. & Livermore, D. M. (1993). Activity of cefepime and other β-lactam antibiotics against permeability mutants of Escherichia coli and Klebsiella pneumoniae. Journal of Antimicrobial Chemotherapy 32, Suppl. B, Gutmann, L., Williamson, N. R., Moreau, N., Kitzis, M. D., Collatz, E., Acar, J. F. et al. (1985). Cross-resistance to nalidixic acid, trimethoprim, and chloramphenicol associated with alterations in outer membrane proteins of Klebsiella, Enterobacter, and Serratia. Journal of Infectious Diseases 151, Martínez, J. L., Alonso, A., Gómez-Gómez, J. M. & Baquero F. (1998). Quinolone resistance by mutations in chromosomal gyrase genes. Just the tip of the iceberg? Journal of Antimicrobial Chemotherapy 42, Bauernfeind, A. (1993). Questioning dosing regimens for ciprofloxacin. Journal of Antimicrobial Chemotherapy 31, Deguchi, T., Yasuda, M., Nakano, M., Ozeki, S., Kanematsu, E. & Nishino, Y. et al. (1997). Detection of mutations in gyra and parc genes in quinolone-resistant clinical isolates of Enterobacter cloacae. Journal of Antimicrobial Chemotherapy 40, Piddock, L. J., Ricci, V., McLaren, I. & Griggs, D. J. (1998). Role of mutation in the gyra and parc genes of nalidixic-acid-resistant salmonella serotypes isolated from animals in the United Kingdom. Journal of Antimicrobial Chemotherapy 41, Threlfall, E. J., Ward, L. R. & Rowe, B. (1999). Resistance to ciprofloxacin in non-typhoidal salmonellas from humans in England and Wales the current situation. Clinical Microbiology and Infection 5, Hakanen, A., Kotilainen, P., Jalava, J., Siitonen, A. & Huovinen, P. (1999). Detection of decreased fluoroquinolone susceptibility in salmonellas and validation of nalidixic acid screening test. Journal of Clinical Microbiology 37, Hansen, D. S., Sirot, D. & Kolmos, H. J. (1998). Extended spectrum beta-lactamases in Danish Klebsiella isolates. Ugeskrift for Læger 160, Nikaido, H. (1988). Bacterial resistance to antibiotics as a function of outer membrane permeability. Journal of Antimicrobial Chemotherapy 22, Suppl. A, Yoshimura, F. & Nikaido, H. (1985). Diffusion of β-lactam antibiotics through the porin channels of Escherichia coli K-12. Antimicrobial Agents and Chemotherapy 27, Thauvin-Eliopoulos, C., Tripodi, M. F., Moellering, R. C. & Eliopoulos, G. M. (1997). Efficacies of piperacillin tazobactam and cefepime in rats with experimental intra-abdominal abscesses due to an extended-spectrum β-lactamase producing strain of Klebsiella pneumoniae. Antimicrobial Agents and Chemotherapy 41, Received 25 June 1999; returned 2 December 1999; revised 11 January 2000; accepted 21 March

8