Antibiograms of resistant Gram-negative bacteria from Scottish CF patients $

Transcription

1 Journal of Cystic Fibrosis 3 (2004) Antibiograms of resistant Gram-negative bacteria from Scottish CF patients $ F.M. MacKenzie a, S.V. Smith a, K.E. Milne a, K. Griffiths b, J. Legge b, I.M. Gould a, * a Department of Medical Microbiology, Aberdeen Royal Infirmary, Foresterhill, Aberdeen AB25 2ZN, UK b Department of Respiratory Medicine, Aberdeen Royal Infirmary, Foresterhill, Aberdeen AB25 2ZN, UK Received 29 December 2003; accepted 24 March 2004 Available online Abstract Background: Over a 19-month pilot phase, 93 multiply resistant Gram-negative isolates from Scottish cystic fibrosis patients were sent to a referral laboratory for further investigation. Methods: In common with the referring diagnostic laboratories, disc diffusion testing was carried out. Antibiotic susceptibility testing was also established by MIC methodology. NCCLS methods were used throughout. Twenty antibiotics were tested. Results: Comparing disc diffusion results against MIC results, there were 167 (14%) major errors. By MIC, Pseudomonas aeruginosa (n = 59), Stenotrophomonas maltophilia (n = 16), Burkholderia cepacia (n = 10) and Alcaligenes xylosoxidans (n = 7) were susceptible to 18%, 11%, 4% and 35% of the antibiotics tested, respectively. Colistin and tobramycin were the most active agents against P. aeruginosa with 60% and 49%, respectively, testing susceptible. Minocycline and gentamicin were most active against S. maltophilia with 58% and 18%, respectively, testing susceptible. B. cepacia were most susceptible to co-trimoxazole (10%) and ciprofloxacin (10%). Five and six of the seven A. xylosoxidans isolates were susceptible to piperacillin and imipenem, respectively. Conclusions: Improved methods for susceptibility testing of such clinical isolates need to be employed in routine diagnostic laboratories. Levels of resistance in referred isolates were very high and similar to those described in the USA. D 2004 European Cystic Fibrosis Society. Published by Elsevier B.V. All rights reserved. Keywords: Gram-negative bacilli; Multiply resistant; Antibiotic susceptibility 1. Introduction While antibiotic therapy has almost certainly prolonged the life of patients with cystic fibrosis (CF), an inevitable consequence of therapy is increased colonisation and infection in this group of patients with increasingly multiply resistant bacteria [1 3]. The clinical implications of antibiotic resistance are important in this group of patients and include issues such as dilemmas on best treatment, infection control and transplant eligibility. A reference laboratory, funded by the United States (US) CF Foundation, was set up at Colombia University, New York, in 1992 to perform extended susceptibility testing of antibiotic combinations in order to define optimal therapy in individual patients [4]. $ The Common Services Agency of the National Service Division of the Scottish Executive funded this work. * Corresponding author. Tel.: address: i.m.gould@abdn.ac.uk (I.M. Gould). Saiman et al. concluded that antibiotic disc diffusion testing is the preferred susceptibility method to employ in routine diagnostic laboratories when testing single agents. This is despite the fact that there are no National Committee for Clinical Laboratory Standards (NCCLS) disc diffusion breakpoints for many of the Gram-negative non-fermenting bacterial species which colonise/infect CF patients. Interpretive breakpoints only exist for Pseudomonas aeruginosa. They also concluded that the reference method most suitable to test such CF isolates is the broth microdilution minimum inhibitory concentration (MIC) method. A similar CF referral laboratory, funded by the Common Services Agency (CSA) of the Scottish Executive, was established at Aberdeen Royal Infirmary (ARI) in April The services of this national referral centre were piloted for 1 year across the whole of Scotland. In the following 7 months, the service was further piloted to carry out an in-depth evaluation of the clinical utility of the service. This second pilot phase was restricted to isolates from the Grampian region of Scotland. We now present /$ - see front matter D 2004 European Cystic Fibrosis Society. Published by Elsevier B.V. All rights reserved. doi: /j.jcf

2 152 F.M. MacKenzie et al. / Journal of Cystic Fibrosis 3 (2004) Table 1 Disc diffusion and MIC breakpoints (BPs) Antibiotic Abbreviation Disc diffusion MIC Disc content (Ag) Method BPs (mm) Method BPs (Ag/ml) S z I R V S V I R z Amikacin AMI 30 NCCLS a NCCLS a Ampicillin/sulbactam AMP/SUL 10/10 NCCLS a NCCLS a 8/4 16/8 32/16 Aztreonam AZT 30 NCCLS a NCCLS a Cefepime FEP 30 NCCLS a NCCLS a Cefpirome CPO 20 BSAC b BSAC b 1 2 Cefatzidime CAZ 30 NCCLS a NCCLS a Ciprofloxacin CIP 5 NCCLS a NCCLS a Chloramphenicol CHLOR 30 NCCLS a NCCLS a Colistin COLIS 25 BSAC b BSAC b 4 8 Co-trimoxazole TMP/SMX 1.25/23.75 NCCLS a NCCLS a 2/38 4/76 Gentamicin GENT 10 NCCLS a NCCLS a Imipenem IMI 10 NCCLS a NCCLS a Meropenem MER 10 NCCLS a NCCLS a Minocycline MIN 10 NCCLS a NCCLS a Netilmicin NET 30 NCCLS a NCCLS a Piperacillin PIP 100 NCCLS c NCCLS c 64 (128) (16) (32 64) 128 Piperacillin/tazobactam PIP/TAZ 100/10 NCCLS c NCCLS c (16) (32 64) 128 Rifampicin RIF 5 NCCLS d NCCLS d Tetracycline TET 30 NCCLS a NCCLS a Tobramycin TOB 10 NCCLS a NCCLS a S = sensitive, I = intermediate, R = resistant. a NCCLS disc diffusion BPs quoted in table are for P. aeruginosa; NCCLS MIC BPs quoted in table are for P. aeruginosa and Acinetobacter spp.. b BSAC disc diffusion and MIC BPs quoted are for Pseudomonas species. c Disc diffusion BPs quoted in table are for P. aeruginosa; MIC BPs quoted in table are for P. aeruginosa (values in parenthesis are for non-p. aeruginosa and non-enterobacteriaceae). d NCCLS Staphylococcus species BPs used in the absence of specific BPs for species tested. susceptibility data on the multiresistant isolates and a comparison of NCCLS-based disc diffusion and NCCLS broth microdilution antibiotic susceptibility results of isolates sent to the referral laboratory during the 19-month pilot phase of the service. 2. Materials and methods 2.1. Test isolates Multiply resistant Pseudomonas species and other nonfermenting Gram-negative bacteria from CF patients were referred from Scottish laboratories to ARI. Isolates were referred if there were local problems identifying treatment Table 2 Definition of susceptibility testing errors Disc diffusion NCCLS MIC reference method comparative method S I R S Agree Major Major I Minor Agree Major R Minor Minor Agree S = sensitive, I = intermediate, R = resistant. regimens and if it was considered clinically beneficial to perform extended susceptibility tests, in particular, MICs or antibiotic combination testing (the latter is the subject of another report in preparation). Organisms were defined as multiply resistant by the referring laboratories if they were resistant to all agents tested in two or more classes of antibiotic. Isolates were identified at a local level, typically by API 20NE (Biomerieux, Marcy l Etoile, France), and posted to the referral laboratory where they were processed within 48 h of arrival. Susceptibility testing and PCR identification of the isolates were then performed as detailed below. Of the 126 isolates referred over the 19- month pilot period, 93 have been included for the purposes of this study. The other 33 isolates were excluded for various reasons. These included duplicate isolates sent from the same patient, isolates from non-cf patients and isolates that failed to grow Susceptibility testing Disc susceptibility and MIC testing were carried out according to NCCLS methodology [5,6]. The inocula for the MIC tests were confirmed using a microdilution method [7]. The antibiotics used in the MIC testing were amikacin/ami (Sigma, Dorset, UK), ampicillin/amp

3 F.M. MacKenzie et al. / Journal of Cystic Fibrosis 3 (2004) (Sigma), sulbactam/sul (Pfizer, Sandwich, UK), aztreonam/azt (Pfizer), cefepime/fep (Bristol Myers Squibb, Hounslow, UK), cefpirome/cpo (Hoeschst Marion Roussell, West Malling, UK), cefatzidime/caz (Glaxo Wellcome, Uxbridge, UK), ciprofloxacin/cip (Bayer, Basingstoke, UK), chloramphenicol/chlor (Sigma), colistin/colis (Pharmax, Bexley, UK), co-trimoxazole/ TMP/SMX (Glaxo Wellcome), gentamicin/gent (Sigma), imipenem/imi (MSD, Hoddesdon, UK), meropenem/mer (Astra Zeneca, Macclesfield, UK), minocycline/min (Sigma), netilmicin/net (Schering Plough, Welwyn Garden City, UK), piperacillin/pip (Sigma), tazobactam/taz (Cyanamid, Gosport, UK), rifampicin/rif (Sigma), tetracycline/tet (Sigma) and tobramycin/tob (Eli Lilly and Co, Basingstoke, UK). Standard NCCLS breakpoints were used to interpret the MIC results. P. aeruginosa NCCLS breakpoints were used to interpret all disc diffusion results except results for COLIS, CPO and RIF for all species tested. In the absence of approved NCCLS breakpoints, P. aeruginosa British Society for Antimicrobial Chemotherapy (BSAC) breakpoints were used for COLIS and CPO and the NCCLS Gram-positive breakpoints were used for RIF. Table 1 details the antibiotic disc strengths used as well as the breakpoint values used to interpret disc Fig. 1. (a) Percent isolates found susceptible to the h-lactams by NCCLS MIC testing. (b) Percent isolates found susceptible to the non-h-lactam antibiotics by NCCLS MIC testing.

4 154 F.M. MacKenzie et al. / Journal of Cystic Fibrosis 3 (2004) Table 3 Breakdown of susceptibility results by NCCLS test methods Species Number Number (%) tests susceptible diffusion and MIC results. Definitions of the susceptibility testing errors can be found in Table 2. Details of which isolates were tested against specific antibiotics are found in Fig. 1a and b Species identification by PCR Standard chloroform-based DNA extraction was carried out on all the test isolates and four control strains (P. aeruginosa ATCC 27853, A. xylosoxidans ATCC 27061, B. cepacia ATCC and S. maltophilia ATCC 13637). PCR was then carried out to identify each isolate to species level. The primers used have been described previously and were as follows: P. aeruginosa: P16SH/ P16SIR specific to the 16S rrna gene [8], A. xylosoxidans: AXF1/AXB1 also specific to the 16S rrna gene [9], B. cepacia complex: BCR1/BCR2 which target the reca gene [10] and S. maltophilia: SM1/SM4 which are directed towards the 23S rrna gene [11]. After initial denaturation for 4 min at 65jC, 30 amplification cycles were completed, each consisting of 1 min at 94jC, 1 min at the annealing temperature and 1 min at 72jC. A final extension of 8 min at 72jC was applied. Annealing temperatures were 65, 57, 64 and 58jC for the P. aeruginosa, A. xylosoxidans, B. cepacia and S. maltophilia primers, respectively. Each clinical isolate and control underwent PCR with each of the stated primer pairs. All PCR runs included a positive and negative control. The PCR products were visualised on a 1.4% agarose gel stained with ethidium bromide. 3. Results of tests performed Disc method MIC method A. xylosoxidans (n = 7) (33%) 26 (35%) B. cepacia (n = 10) (14%) 5 (4%) P. aeruginosa (n = 59) (27%) 139 (18%) S. maltophilia (n = 16) (21%) 20 (11%) No I.D. (n = 1) 14 2 (14%) 2 (14%) Totals (n = 93) (22%) 192 (16%) The 93 isolates tested were submitted from 39 patients with between 1 and 7 isolates referred per patient. The largest group of isolates were from the Grampian region of Scotland (40%), with 24% from Lothian, 14% from Tayside, 11% from Dumfries, 6% from Highland, 3% from Strathclyde and 2% from the Western Isles. The isolates were made up of 59 P. aeruginosa, 16 S. maltophilia, 10 B. cepacia, 7 A. xylosoxidans and 1 organism that remained unidentified by PCR. Disc diffusion and NCCLS MIC testing were carried out on all 93 isolates. Table 3 details the number of individual disc diffusion and MIC tests performed per species in the referral laboratory along with the percentage that tested susceptible. Although A. xylosoxidans appeared to be the most susceptible species tested, it should be borne in mind that only seven isolates were tested. The most resistant group tested was the B. cepacia isolates, irrespective of the testing method. MIC and disc diffusion testing to between 17 and 20 antibiotics were carried out on each isolate depending on species. By MIC, the isolates of P. aeruginosa were susceptible to 18% of the antibiotics tested and the isolates of S. maltophilia, B. cepacia and A. xylosoxidans were susceptible to 11%, 4% and 35% of the antibiotics, respectively. It is generally accepted that MIC testing is more discriminatory than disc diffusion testing and discrepancies between the two were interpreted in this light. Table 4 summarises the major errors encountered. There were a total of 167 major errors. Tables 5 and 6 summarise the MIC results per species for the most commonly tested antibiotics. It should be noted that MIC 50 and MIC 90 values were not calculated where less than 10 isolates were tested. Fig. 1a and b summarises the percentage of each species that were susceptible by MIC to the individual antibiotics detailed. It should be noted that a blank box indicates that the specific antibiotic/organism was either not tested or less than 10 were tested. The most susceptible combination was that of A. xylosoxidans and the h-lactams with an overall susceptibility of 55%. A. xylosoxidans is, however, not shown in the figure as only seven isolates were tested. The S. maltophilia and B. cepacia isolates were almost totally resistant to the h-lactams with 96% and 97% resistance respectively. The P. aeruginosa isolates were moderately resistant to the h-lactams with an overall resistance prevalence of 80% rising to 23% susceptible to MER and 25% susceptible to PIP/TAZ which were the most active of the h-lactam agents. In contrast to the h-lactams, the seven A. xylosoxidans isolates with an overall resistance prevalence of 81% were moderately resistant to the non-h-lactams tested. Notably Table 4 Major errors established when comparing NCCLS MIC and disc diffusion testing Species Total number tests compared Numbers of tests resulting in Major errors; susceptibility results by MIC and disk diffusion I (S) R (S) R (I) A. xylosoxidans (n =7) B. cepacia (n = 10) P. aeruginosa (n = 59) S. maltophilia (n = 16) No I.D. (n =1) Totals (n = 93) S = susceptible, I = intermediate, R = resistant.

5 F.M. MacKenzie et al. / Journal of Cystic Fibrosis 3 (2004) Table 5 Summary of MIC data for commonly tested h-lactam antibiotics Antibiotics tested PIP PIP/TAZ IMI MER CPO FEP AMP/SULB CAZ AZT A. xylosoxidans (Number tested) (7) (4) (7) (3) (2) (2) NT (7) (6) Range 0.25 > > > >128 MIC 50 MIC 90 B. cepacia (Number tested) (10) (10) (10) (10) 7 (7) (8) (10) NT Range >128 > >128 >128 > >128 MIC 50 >128 > >128 MIC 90 >128 > >128 P. aeruginosa (Number tested) (57) (48) (58) (57) (47) (48) (8) (58) (58) Range 4 >128 2 > >128 4 > >128 2 > >128 MIC 50 >128 > MIC 90 >128 > >128 >128 >128 >128 S. maltophilia (Number tested) (15) (14) (14) (9) (5) (5) (1) (10) (3) Range 64 > >128 2 > > > >128 > >128 >128 MIC 50 >128 >128 > MIC 90 >128 >128 >128 >128 NT: not tested. : not determined due to too few numbers. high numbers of S. maltophilia were susceptible to MIN (58%) and high numbers of P. aeruginosa were susceptible to COLIS (60%) and TOB (49%). Overall, 85% of the B. cepacia isolates were resistant to the non-h-lactams. However, three of nine B. cepacia isolates were susceptible to MIN. Table 6 Summary of MIC data for commonly tested non-h-lactam antibiotics Antibiotics tested AMI GENT NET TOB CIP COLIS CHLOR SMX/TMP RIF MIN TET A. xylosoxidans (Number tested) (6) (6) (6) (6) (7) (2) (1) (1) (1) NT (1) Range 32 >128 4 >128 8 > MIC 50 MIC 90 B. cepacia (Number tested) (2) (2) (2) (9) (10) (2) (10) (10) (10) (9) NT Range > >128 1 >128 > > > MIC MIC 90 >128 > P. aeruginosa (Number tested) (58) (58) (54) (57) (59) (48) NT NT NT NT NT Range 4 >128 2 >128 4 > >128 V 0.06 >128 MIC MIC 90 > > S. maltophilia (Number tested) (6) (11) (5) (13) (15) (9) (8) (14) (12) (12) (3) Range 16 > > >128 8 > >128 4 > MIC MIC 90 >128 > NT: not tested. : not determined due to too few numbers.

6 156 F.M. MacKenzie et al. / Journal of Cystic Fibrosis 3 (2004) Discussion Antibiotic resistance among non-fermenting Gram-negative bacteria is a well-known phenomenon. In general, most are innately resistant to many agents due to a combination of enzymatic degradation, reduced permeability and efflux pumps [12 15]. It is well established that non-fermenting Gram-negatives, from CF patients in particular, exhibit multiple resistance phenotypes. This is almost unquestionably due to selective pressure from intensive antibiotic treatment [4]. In a recent UK-wide survey, Henwood et al. [16] showed that isolates of P. aeruginosa from CF patients were substantially more resistant than those from non-cf patients. Even then, they reported that >85% of the CF isolates were susceptible to ceftazidime, meropenem, piperacillin and piperacillin/tazobactam and 57%, 64%, 69% and 76% were susceptible to gentamicin, amikacin, imipenem and ciprofloxacin, respectively. Their isolates were clearly much more susceptible than the P. aeruginosa isolates described in the current study. P. aeruginosa resistance levels determined by MIC in the current study are very similar to those described by the US CF referral laboratory with one or two exceptions [4]. Notably, the US laboratory provides no data on colistin, as it is not licensed for use in CF patients in that country. The US data describe over 90% P. aeruginosa susceptible to tobramycin when using the higher breakpoint of 128 mg/l, based on the use of nebulised tobramycin. Using the lower breakpoint of 16 mg/l, as in the present study, very similar resistance levels were described. The BSAC breakpoints for colistin and cefpirome were used in this study, as there are currently no NCCLS interpretative criteria for these antimicrobials. The BSAC breakpoint of 8 mg/l for colistin could probably also be increased substantially to take account of its use in the nebulised form. Another notable point is the improved activity of piperacillin against P. aeruginosa when combined with tazobactam in comparison with the US data. In common with previous studies, B. cepacia was the most resistant species amongst those studied [17 19], with an average of only 1.2 agents active (fully or at an intermediate level) against this species. The problems of performing and interpreting susceptibility tests in this group of organisms have previously been well described. Biofilm production in CF patients means that it is impossible to totally eradicate infecting organisms, even if they are fully susceptible to the agent used, unless the infection is treated within the first few months. Nevertheless, the current widespread use of disc diffusion susceptibility testing, with its major errors, is likely to lead to the inappropriate use of inactive agents and consequent poor clinical response. Routine diagnostic laboratories generally perform agar disc diffusion testing on their isolates [20]. Neither the BSAC [21] nor the NCCLS [5,6] recommend agar disc diffusion be performed on species tested in this study, other than P. aeruginosa, and consequently, no breakpoints exist. As disc diffusion testing is the method of choice in most diagnostic laboratories, disc diffusion was also carried out in this study (Table 3). In the absence of suitable breakpoints, P. aeruginosa breakpoints were used as it is believed that these are generally used for interpretation of diagnostic laboratory results. When comparing disc diffusion and MIC results, MIC values were set as the gold standard against which disc diffusion results were compared. The comparisons show a very worrying number of major errors (Table 4). That is, 14% of disc diffusion results reported false susceptibility. With the increasing understanding of the importance of the MIC in predicting clinical outcome [22], it would seem that MIC testing should be the norm with isolates from this difficult to treat group of patients. Standard MIC methods are time consuming; however, the commercial E-test may provide a solution to this problem making routine determination of MICs easier. So far, results are contradictory [18,19]. Finally, with such a group of multiply resistant organisms, combination therapy is common place [23]. The referral laboratory in Scotland routinely carries out combination testing on isolates referred in a bid to facilitate the optimum choice of combination therapy, and this will be the subject of a further report. Acknowledgements Thanks are due to all the referring laboratories and clinical staff. This service has recently (July 2001) been restarted on a national (Scottish) basis with funding from the CSA. This study was presented, in part, at the 11th European Congress of Clinical Microbiology and Infectious Diseases April 1 4, 2001, Istanbul, Turkey, abstract P1383 and the 13th European Congress of Clinical Microbiology and Infectious Diseases May 10 13, 2003, Glasgow, abstract P1089. References [1] Denton M, Wilcox MH. Antimicrobial treatment of pulmonary colonisation and infection by Pseudomonas aeruginosa in cystic fibrosis patients. J Antimicrob Chemother 1997;40: [2] Hoiby N. Treatment of Pseudomonas aeruginosa infection in cystic fibrosis. Clin Microbiol Infect 1999;40:S46 8. [3] Saiman L. Epidemiology, mechanisms and implications of multiply antibiotic-resistant Pseudomonas aeruginosa in patients with cystic fibrosis. Clin Microbiol Infect 1999;40:S55 8. [4] Saiman L, Burns JL, Whitter S, Krzewinski J, Marshall SA, Jones RN. Evaluation of reference dilution test methods for antimicrobial susceptibility testing of Pseudomonas aeruginosa strains isolated from patients with cystic fibrosis. J Clin Microbiol 1999;37: [5] NCCLS. Methods for Dilution Antimicrobial Susceptibility Tests for Bacteria That Grow Aerobically; Approved Standard-Sixth Edition. NCCLS document M7-A6. NCCLS, 940 West Valley Road, Suite 1400, Wayne Pennsylvania USA, [6] NCCLS. Performance Standards for Antimicrobial Susceptibility Testing; Fourteenth Informational Supplement. NCCLS document M100-S14 [ISBN]. NCCLS, 940 West Valley Road, Suite 1400, Wayne, Pennsylvania USA, 2004.

7 F.M. MacKenzie et al. / Journal of Cystic Fibrosis 3 (2004) [7] Gould IM, Wilson D, Milne K, Paterson A, Golder D, Russell D. Interaction of imipenem with erythromycin and tetracycline assessed by microdilution checkerboard techniques. Antimicrob Agents Chemother 1991;35: [8] Wilson VL, Tatford BC, Yin X, Rajki SC, Walsh MM, LaRock P. Species-specific detection of hydrocarbon utilising bacteria. J Microbiol Methods 1999;39: [9] Liu L, Coenye T, Burns JL, Whitby PW, Stull TL, LiPuma JL. Ribosomal DNA-directed PCR for identification of Achromobacter (Alcaaligenes) xylosoxidans recovered from sputum samples from cystic fibrosis patients. J Clin Microbiol 2002;40: [10] Mahenthiralingam E, Bischof J, Byrne SK, Radomski C, Davies JE, Av-Gay Y, et al. DNA-based diagnostic approaches for identification of Burkholderia cepacia complex, Burkholderia vietnamiensis, Burkholderia multivorans, Burkholderia stabilis and Burkholderia cepacia genomovars I and III. J Clin Microbiol 2000;38: [11] Whitby PW, Carter KB, Burns JL, Royall JA, LiPuma JJ, Stull TL. Identification and detection of Stenotrophomonas maltophilia by rrna-directed PCR. J Clin Microbiol 2000;38: [12] Li Z, Zian-Zhi L, Poole K. Multiple antibiotic resistance in Stenotrophomonas maltophilia: involvement of a multidrug efflux system. Antimicrob Agents Chemother 2000;44: [13] Nakae T, Nkajima A, Ono T, Saito K, Yoneyama H. Resistance to h- lactam antibiotics in Pseudomonas aeruginosa due to interplay between the MexAB-OprM efflux pump and h-lactamase. Antimicrob Agents Chemother 1999;43: [14] Hancock R. Resistance mechanisms in Pseudomonas aeruginosa and other nonfermentative gram-negative bacteria. Clin Infect Dis 1998; 27:S93 9. [15] Burns J, Wadsworth C, Barry J. Nucleotide sequence analysis of a gene from Burkholderia (Pseudomonas) cepacia encoding an outer membrane lipoprotein involved in multiple antibiotic resistance. Antimicrob Agents Chemother 1989;33: [16] Henwood CJ, Livermore DM, James D, Warner M and the Pseudomonas Study Group. Antimicrobial susceptibility of Pseudomonas aeruginosa: results of a UK survey and evaluation of the British Society for Antimicrobial Chemotherapy disc susceptibility test. J Antimicrob Chemother 2001;47: [17] Monteil H, Harf-Monteil C. Aerobic Gram-negative bacilli: newer nosocomial pathogens. Int J Antimicrob Agents 1997;8: [18] Walsh L, Thibault M, Tullus E, Devlin HR. Susceptibility testing of B. cepacia: a comparison of agar dilution with the E. Test. 40th ICAAC Abstracts, No. 328; [19] Manno G, Belli ML, Ugolotti E, Piana S, Bassetti M, Romano L, et al. In vitro activity of antibiotic combinations against Burkholderia cepacia isolates from patients with cystic fibrosis. 40th ICAAC Abstracts, No. 528; [20] Stokes EJ, Ridgway G. Clinical Bacteriology, 5th Ed. London: Edward Arnold; [21] BSAC Working Party Report Antimicrobial Susceptibility Testing: BSAC Working Party Report. J. Antimicrob. Chemother. 2001; 48 (Suppl. S1). [22] Gould IM. Pharmacodynamics and relationship between in vitro and in vitro activity of antimicrobial agents. J Chemother 1997;9: [23] Aaron SD, Ferris W, Henry DA, Speert DP, MacDonald NE. Multiple combination bactericidal antibiotic testing for patients with cystic fibrosis infected with Burkholderia cepacia. Am J Respir Crit Care Med 1999;161: