Evaluation of antimicrobial susceptibility of bacteria containing the qnr gene and FOX-5 β-lactamase by four automated systems

Authors


Corresponding author and reprint requests: M. C. Conejo, Microbiology, Faculty of Medicine, University of Seville, Seville, Spain
E-mail: cconejo@us.es

Abstract

The accuracy and performance of four automated instruments (BD Phoenix, MicroScan WalkAway, VITEK-2 and Wider) were evaluated for susceptibility testing of fluoroquinolones and β-lactams with four clinical isolates of Klebsiella pneumoniae and the corresponding Escherichia coli transconjugants containing a plasmid carrying the qnr gene and coding for FOX-5 production. No major or very major errors were detected with the MicroScan system. Many of the minor errors for both quinolones and β-lactams clustered around the intermediate breakpoints.

Resistance to fluoroquinolones is increasing in many bacteria of clinical importance [1]. Mechanisms of fluoroquinolone resistance in Gram-negative bacteria include mutations in genetic loci such as gyrA, parC, gyrB and parE[2]. Changes in the expression of efflux pumps and porins may also contribute to fluoroquinolone resistance [3]. Recently, a new mechanism, plasmid-mediated quinolone resistance, has been described [4]. The responsible gene is qnr, which itself confers only a low level of resistance, but which may contribute to clinical resistance when associated with other mechanism(s). Problems in the use of automated systems to detect fluoroquinolone resistance have been reported, although corrective action by equipment manufacturers appears to have resolved these problems [5]. However, the accuracy of these systems has not yet been assessed with organisms expressing the qnr gene.

Plasmids carrying the qnr gene usually also carry genes coding for β-lactamases, most frequently the AmpC-type enzyme FOX-5 [6,7]. The purpose of the present study was to evaluate the performance of four automated instruments, namely the BD Phoenix (BD Diagnostic Systems, Sparks, MD, USA), MicroScan WalkAway (Dade Behring, West Sacramento, CA, USA), VITEK-2 (bioMérieux-Vitek, Hazelwood, MO, USA) and Wider (Francisco Soria Melguizo S.A., Madrid, Spain) systems, for susceptibility testing of fluoroquinolones and β-lactams with four clinical isolates of Klebsiella pneumoniae and their corresponding Escherichia coli transconjugants. The latter contained a plasmid coding for both Qnr and FOX-5 production [4,8], but only FOX-5-expressing transconjugants were obtained from one K. pneumoniae isolate (1132) because, for unknown reasons, none of the E. coli transconjugants obtained with this donor strain were able to express FOX-5 and Qnr simultaneously.

The NCCLS broth microdilution assay [9] was used as the reference test method for the following antibiotics: ampicillin (Sigma, Madrid, Spain); cefotaxime (Sigma); cefoxitin (Sigma); ceftazidime (Glaxo, Barcelona, Spain); ciprofloxacin (Bayer, Leverkusen, Germany); nalidixic acid (Sigma); norfloxacin (Sigma); and ofloxacin (Roussel Uclaf, Romainville, France). For susceptibility testing using the automated systems, inocula were prepared according to the manufacturers' suggested procedures. The card and panels used in this study were as follows (mg/L): NMIC/ID-12 Combo panel for the BD Phoenix (ampicillin, 4–16; cefotaxime, 1–16; cefoxitin, 4–16; ceftazidime, 1–16; ciprofloxacin, 0.125–2; nalidixic acid, 8–32; and norfloxacin, 2–8); Combo Negative type 1S panel for MicroScan WalkAway 96 (ampicillin, 4–16; cefotaxime, 0.5–32; cefoxitin, 8–16; ceftazidime, 1–16; ciprofloxacin, 0.12–2; and ofloxacin, 0.5–4); Gram-negative MIC panel for Wider (cefoxitin, 4–16; ceftazidime, 0.5–16; ciprofloxacin, 0.12–4; and nalidixic acid, 4–16); and AST-N020 for VITEK-2 (ampicillin, 4–32; cefotaxime, 1–32; cefoxitin, 8–32; ceftazidime, 1–32; ciprofloxacin, 0.5–4; ofloxacin, 0.5–4; and norfloxacin, 1–32). The MICs obtained were interpreted as susceptible, intermediate or resistant according to NCCLS criteria [9]. ‘Very major errors’ were recorded when an organism was considered resistant by the reference method, but susceptible by the test method. ‘Major errors’ were recorded when an organism considered susceptible by the reference method was resistant by the test method. ‘Minor errors’ were recorded when an organism was considered susceptible or resistant either by the reference or the test method, but intermediate by the other method. All tests showing very major errors or major errors were repeated in duplicate by both test and reference methods. MICs of quinolones and β-lactams obtained by the reference method and the four automated systems are shown in Table 1. The number of errors for quinolones and β-lactams by the test method when compared with the reference method are shown in Table 2.

Table 1.  Comparison of MICs (mg/L) obtained by the reference method and automated testing methods
StrainAntimicrobial
agent
Reference
method
MicroScanBD PhoenixWiderVITEK-2
  1. NT, not tested; TC, Escherichia coli transconjugant.

UAB1Ciprofloxacin22> 221
Nalidixic acid32NT> 3216NT
Norfloxacin16NT> 8NT2
Ofloxacin164NTNT2
Cefoxitin> 128> 16> 16> 16> 64
Ceftazidime64> 16> 16> 1616
Cefotaxime881684
Ampicillin> 128> 16> 16NT> 32
KpN5Ciprofloxacin0.251120.25
Nalidixic acid32NT816NT
Norfloxacin2NT2NT0.5
Ofloxacin22NTNT2
Cefoxitin128> 16> 16> 16> 64
Ceftazidime32> 16> 161616
Cefotaxime832> 1684
Ampicillin> 128> 16> 16NT> 32
1960Ciprofloxacin11> 242
Nalidixic acid> 256NT> 32> 16NT
Norfloxacin4NT8NT8
Ofloxacin4> 4NTNT> 8
Cefoxitin256> 16> 16> 16> 64
Ceftazidime1616> 16> 1616
Cefotaxime44844
Ampicillin> 128> 16> 16NT> 32
1132Ciprofloxacin4> 2> 24> 4
Nalidixic acid> 256NT> 32> 16NT
Norfloxacin16NT> 8NT16
Ofloxacin16> 4NTNT> 8
Cefoxitin> 256> 16> 16> 16> 64
Ceftazidime64> 16> 16> 16> 64
Cefotaxime881688
Ampicillin> 128> 16> 16NT> 32
UAB1 TCCiprofloxacin0.510.521
Nalidixic acid32NT1616NT
Norfloxacin2NT22NT
Ofloxacin12NTNT2
Cefoxitin> 128> 16> 16> 16> 64
Ceftazidime32> 168816
Cefotaxime88884
Ampicillin> 128> 16> 16NT> 32
KpN5 TCCiprofloxacin0.12510.252< 0.25
Nalidixic acid32NT1616NT
Norfloxacin1NT22NT
Ofloxacin0.52NTNT2
Cefoxitin32> 16> 16> 16> 64
Ceftazidime816161616
Cefotaxime24844
Ampicillin> 128> 16> 16NT> 32
1960 TCCiprofloxacin0.12510.252< 0.25
Nalidixic acid32NT1616NT
Norfloxacin1NT22NT
Ofloxacin0.52NTNT1
Cefoxitin32> 16> 16> 16> 64
Ceftazidime888816
Cefotaxime24884
Ampicillin> 128> 16> 16NT> 32
1132 TCCiprofloxacin0.008< 0.12< 0.13< 0.12< 0.25
Nalidixic acid4NT< 8< 4NT
Norfloxacin0.06NT< 2< 0.5NT
Ofloxacin0.06< 0.5NTNT< 0.25
Cefoxitin128> 16> 16> 16> 64
Ceftazidime3216161616
Cefotaxime48884
Ampicillin> 128> 16> 16NT> 32
Table 2.  Quinolone and β-lactam susceptibility testing errors observed with the automated systems, compared to results obtained by the reference microdilution method
DrugErrorsMicroScanBD PhoenixWiderVITEK-2
CiprofloxacinMinor0142
Major0110
Very major0000
Nalidixic acidMinor00
Major00
Very major45
NorfloxacinMinor11
Major00
Very major01
OfloxacinMinor21
Major00
Very major01
CefoxitinMinor0000
Major0000
Very major0000
CeftazidimeMinor2346
Major0000
Very major0110
CefotaximeMinor1200
Major0100
Very major0000
AmpicillinMinor000
Major000
Very major000

For quinolones, very major errors were observed only for nalidixic acid, norfloxacin and ofloxacin. The very major errors for nalidixic acid were produced by the BD Phoenix and Wider systems (the MicroScan panels and the VITEK-2 cards do not contain this antimicrobial agent). The MIC of nalidixic acid for the strains involved in very major errors was 32 mg/L (close to the intermediate breakpoint), while the MIC obtained with the automated systems was 16 mg/L (except for strain KpN5, for which the MIC with the BD Phoenix system was 8 mg/L). The very major errors observed for norfloxacin and ofloxacin were produced by the VITEK-2 system with strain UAB1. Major errors were only observed for ciprofloxacin with the BD Phoenix and Wider systems for strain 1960. The MIC for this strain was consistently > 2 mg/L by BD Phoenix and 4 mg/L by Wider, but was 1 mg/L by the reference method. Minor errors were encountered in all the systems evaluated. The number of minor errors was higher for ciprofloxacin than for the other quinolones. Overall, MicroScan appeared to be the best system for testing susceptibility to quinolones.

For β-lactams, no errors were found for cefoxitin and ampicillin. Very major errors were found only for ceftazidime with the BD Phoenix and Wider systems. The MIC of ceftazidime for UAB1 TC (the strain responsible for the very major errors) was 32 mg/L (close to the intermediate breakpoint), while the MIC obtained with the automated systems was 8 mg/L. The only major error was observed for cefotaxime and the BD Phoenix system with strain KpN5. The MIC for this strain was consistently > 16 mg/L with the BD Phoenix system, but was 8 mg/L with the reference method. Minor errors were detected only for susceptibility testing of ceftazidime and cefotaxime.

It is noteworthy that the BD Phoenix drew attention to the possibility of extended-spectrum β-lactamase production by all of the strains tested, and, in the case of VITEK-2, a recommendation was made about changing the clinical category of all the cephalosporins to the resistant range. Plasmid-encoded AmpC β-lactamases are found increasingly among isolates of K. pneumoniae and E. coli. There is no NCCLS recommendation regarding the MIC interpretation or detection of K. pneumoniae isolates producing this type of β-lactamase, but it seems reasonable that they should be considered resistant to all β-lactams, with the exception of carbapenems and, probably, zwitterionic cephalosporins.

Many of the minor errors for both quinolones and β-lactams clustered around the intermediate breakpoints. Thus, a change in the MIC for an organism by ± 1 doubling dilution often resulted in a minor error. The panels or cards of these automated systems contain a limited number of wells for susceptibility determination. For this reason, in addition to the fact that the intermediate range for antimicrobial agents consists of a single dilution, an error in just one well could cause an inaccurate result. If errors within one dilution of the broth microdilution reference method were excluded from analysis, most minor errors would be eliminated, including the very major errors in the case of nalidixic acid, for which intermediate susceptibility does not exist.

Acknowledgements

This work was performed under the auspices of the Red Española de Investigación en Patología Infecciosa (REIPI–ISCIII–C03/14) and was supported by a grant of the Instituto de Salud Carlos III, Fondo de Investigación Sanitaria (FIS; PI020256).

Ancillary