Aims: This study investigated the prevalence of plasmid-mediated AmpC production in selected clinical isolates of Escherichia coli, Klebsiella species and Proteus mirabilis, and compared the results of boronic acid disc screening with conventional susceptibility testing for the detection of AmpC-positive isolates.
Methods: E coli, Klebsiella species and P mirabilis with reduced susceptibility to amoxycillin-clavulanate, cefuroxime and cephalexin, but without phenotypic evidence of extended-spectrum β-lactamases were screened for AmpC activity using enzyme-extraction methods. The presence of plasmid-mediated ampC was determined by multiplex PCR. Antibiotic susceptibilities were determined using both disc and dilution-based methods. A disc-based screening method for detection of AmpC-producing strains was evaluated using boronic acid as an inhibitor of AmpC, and cefoxitin as the antibiotic substrate.
Results: Plasmid-mediated ampC was present in 26% of study isolates, with CMY-like enzymes detected predominantly in E coli and DHA-like enzymes predominantly in Klebsiella pneumoniae. Current susceptibility methods failed to detect a significant proportion of plasmid-mediated AmpC-producing isolates, with 33% of such strains interpreted as susceptible to third-generation cephalosporins using current Clinical Laboratory Standards Institute breakpoints. The boronic acid disc method showed sensitivity and specificity of 90% and 98% respectively in detecting AmpC-positive isolates.
Conclusion: The prevalence of plasmid-mediated ampC was high in the study population, and may be missed by conventional susceptibility testing methods. Inhibitor-based screening methods would improve detection of this emerging resistance phenotype.
Statistics from Altmetric.com
Resistance to expanded-spectrum cephalosporins in Enterobacter spp, Citrobacter spp and Serratia spp may develop through the expression of chromosomally encoded class C β-lactamases, also known as AmpC β-lactamases. The gene encoding for the AmpC β-lactamase is also present in Escherichia coli, but is not expressed because of the lack of a promotor region. The transfer of chromosomal genes to plasmids allowed the expression of AmpC β-lactamases in Klebsiella spp, E coli, Proteus mirabilis and Salmonella spp.
The accurate detection of plasmid-mediated ampC is important to improve clinical management of infections and provide sound epidemiological data. At present, there are no standard methods available to detect the presence of AmpC enzymes. Reduced susceptibility to cefoxitin in the Enterobacteriaceae may be an indicator of AmpC activity. However, cefoxitin resistance may also be mediated by alterations to outer membrane permeability.1 Disc-based tests for the detection of AmpC activity have been proposed, using various compounds as inhibitors of AmpC. The phenotypic “gold standard” for the detection of AmpC requires labour-intensive enzyme extraction methods that are not suitable for routine use in clinical microbiology laboratories.
This prospective study was carried out over a 1 year period in the absence of a documented outbreak in order to (a) delineate the prevalence of plasmid-mediated ampC within selected clinical isolates, (b) evaluate the results of conventional susceptibility testing methods in detecting isolates with plasmid-mediated ampC, and (c) determine the accuracy of a disc screening test utilising boronic acid as an inhibitor of AmpC.
Selection of clinical isolates
The study population was defined as unique clinical isolates of E coli, Klebsiella spp and P mirabilis that were resistant to amoxycillin-clavulanate, cefuroxime and cephalexin, but without phenotypic evidence of extended-spectrum β-lactamase (ESBL) activity. Susceptibilities to the three antibiotics were determined by disc diffusion,2 and the presence of ESBLs was excluded by the disc approximation method.3
Epidemiological information was collected from the hospital patient information system. Isolates from patient samples that were submitted >48 h after initial admission, or from patients with a previous record of hospitalisation within this institution in the last 90 days, were designated as nosocomial in origin. All other isolates were designated as possible community acquired.
Detection of AmpC activity
The presence of AmpC enzyme activity was determined for all study isolates using previously described enzyme-extraction techniques. In brief, crude enzyme extracts were prepared by freezing and thawing centrifuged cell pellets from broth cultures. The enzyme extracts were inoculated into wells on Mueller-Hinton agar plates containing cefoxitin at a concentration of 4 μg/ml.4 Enzyme extracts were also inoculated using the modified three-dimensional test, as described by Coudron et al.5 A positive result by either test method was taken to be phenotypic confirmation of AmpC activity.
Detection of plasmid-mediated ampC
Study isolates that were positive for AmpC activity were tested with a multiplex PCR assay6 that identifies six family-specific ampC genes carried on plasmids. Sequencing of the amplified ampC product was performed on a representative selection of PCR-positive isolates, and the ampC gene was determined through a BLAST search.7
Antibiotic susceptibility testing
Antibiotic susceptibilities to ceftriaxone and ceftazidime were determined by disc diffusion, as previously described. Minimum inhibitory concentrations were determined by microbroth dilution (MicroScan MICPlus 4 panels; Dade-Behring, Newark, Delaware, USA) for all isolates carrying plasmid-mediated ampC genes.
Screening tests for AmpC
Two screening methods were evaluated. Susceptibility to cefoxitin was determined by standard disc methods.2 Study isolates were also tested using a modified disk-based test,8 utilising boronic acid as an inhibitor of AmpC. Disc susceptibility testing was performed on Mueller-Hinton medium (Becton Dickinson, Franklin Lakes, New Jersey, USA) using a standard 30 μg cefoxitin disc (Becton Dickinson) and a similar cefoxitin disc supplemented with 400 μg of phenylboronic acid (Sigma-Aldrich, Singapore). An increase of ⩾5 mm in the zone of inhibition for the phenylboronic-cefoxitin disc as compared with the cefoxitin disc was interpreted as a positive test for the AmpC enzyme.
The sensitivity and specificity of using cefoxitin susceptibility and the phenylboronic acid based disc test for detecting AmpC activity were calculated using the enzyme-extraction tests as the gold standard.
A total of 201 isolates of E coli (n = 153), Klebsiella spp (n = 43) and P mirabilis (n = 5) were screened. AmpC activity was present in 58 (29%) isolates, and plasmid-encoded ampC genes were detected by PCR in 52 (26%). Non-plasmid-derived AmpC activity was present in six (3%) strains of E coli, and was presumed to originate from hyperproduction of endogenous AmpC enzyme. Plasmid-encoded ampC genes belonging to the CIT family (originating from Citrobacter freundii) were detected in 41 (79%) isolates. CIT-like ampC genes were predominantly detected in E coli (n = 34, 83% of total CIT-positive isolates), but were also present in Klebsiella pneumoniae (n = 3, 7%) and P mirabilis (n = 4, 10%). Plasmid-encoded ampC genes from the DHA family (originating from Morganella morganii) were detected in the remaining 11 (21%) isolates, predominantly in K pneumoniae (n = 10, 91%) but also in E coli (n = 1, 9%). Sequencing of representative PCR amplicons demonstrated that the CIT and DHA amplicons were identical to CMY-2 and DHA-1 respectively.
Strains with plasmid-mediated ampC were isolated from the urinary tract (n = 46, 88%), blood cultures (n = 3, 6%) and other body sites (n = 3, 6%). Of the 52 test isolates identified as possessing plasmid-mediated ampC, 34 (65%) were defined as nosocomial in origin.
Antibiotic susceptibilities of ampC-positive isolates
Based on the results of standard disc susceptibility testing, a proportion of isolates carrying plasmid-mediated ampC were susceptible to ceftriaxone (n = 18, 34%), ceftazidime (n = 10, 19%) and cefoxitin (n = 1, 2%). When current Clinical Laboratory Standards Institute breakpoints were applied, an even larger proportion of these isolates were susceptible to cephalosporins by minimum inhibitory concentration (MIC) testing: ceftriaxone susceptible (n = 35, 67%), ceftazidime susceptible (n = 18, 35%), cefotaxime susceptible (n = 43, 83%) and cefoxitin susceptible (n = 2, 4%). Seventeen (33%) isolates with plasmid-mediated ampC were susceptible to all three third-generation cephalosporins based on MIC interpretation. The cephalosporin susceptibilities and MIC distributions for E coli and K pneumoniae isolates positive for plasmid-mediated ampC are shown in table 1.
Evaluation of detection methods
In this test population, the detection of AmpC activity using reduced susceptibility to cefoxitin by disc testing as a screening criterion showed a sensitivity of 98%, a specificity of 73%, a positive predictive value of 63%, and a negative predictive value of 99%. The phenylboronic-cefoxitin disc tests showed corresponding sensitivity of 90%, specificity of 98%, positive predictive value of 96%, and negative predictive value of 95%. It is important to note that phenotypic tests are unable to differentiate between plasmid-mediated AmpC activity and chromosomally encoded AmpC activity.
Plasmid-mediated AmpC-producing strains of Escherichia coli, Klebsiella spp and Proteus mirabilis may be poorly detected by current susceptibility testing methods.
Inferring the presence of AmpC activity from reduced cefoxitin susceptibility has good sensitivity but lacks specificity.
The use of boronic acid in a disc-based screening method improved the test specificity.
The results of this study demonstrate several important findings. By applying screening criteria of reduced susceptibility to amoxycillin-clavulanic acid, cefuroxime and cephalexin, and excluding isolates with detectable ESBL activity, the prevalence of plasmid-mediated ampC was high in the tested population: 23% in E coli and 30% in Klebsiella spp. Although most strains were nosocomial and isolated from the urinary tract, there was no evidence of a nosocomial source for one-third of isolates. Results from a single centre study should be interpreted with care, but the data suggest that further population-based prevalence studies are warranted in order to monitor the true spread of plasmid-mediated ampC within nosocomial and community settings.9 10
In the clinical laboratory, the expression of AmpC enzymes in Enterobacteriaceae is often inferred by resistance to third-generation cephalosporins and cephamycins in the absence of recognisable resistance mechanisms e.g. ESBL. However, in this study, one-third of bacterial isolates possessing plasmid-encoded ampC would have been reported as susceptible to third-generation cephalosporins by conventional MIC testing. The importance of detecting AmpC-producing isolates is highlighted by data showing that such strains of K pneumoniae are associated with high clinical failure rates when treated with cephalosporins.11
Based on our studied population, the use of cefoxitin resistance as an indicator for AmpC-producing strains has a high negative predictive value, but may overestimate the number of true positives. The use of phenylboronic acid in combination with cefoxitin as a phenotypic screening method showed better potential.
In summary, this study showed high prevalence of plasmid-mediated ampC in a selected population of E coli, K pneumoniae and P mirabilis. A significant number of these isolates would not have been detected by conventional susceptibility testing methods. Further developmental work is required for an effective screening method for AmpC enzymes and to determine the epidemiology of this resistance mechanism.
We thank Ms Pamela Leow for technical assistance provided during the course of this study.
Competing interests: None.
Funding: This study was partially funded by a grant from Singapore Polytechnic.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.