Controlling for false positives: interpreting MBL Etest and MBL combined disc test for the detection of metallo- -lactamases

Sir, Metallo-β-lactamases (MBLs) are enzymes with versatile hydrolytic capabilities, namely the ability to hydrolyse all β-lactam antibiotics, with the exception of monobactams. Consequently, bacteria producing MBLs often cause infections that are difficult to treat due to the ineffectiveness of β-lactams. Often, the genes encoding MBLs are located on integrons that frequently harbour other resistance determinants and that may be associated with mobile genetic elements, thus increasing the likelihood of their dissemination. MBLs are particularly associated with Gram-negative pathogens, and their occurrence requires continuous monitoring due to their potential impact on patient management and the associated implication for infection control. Several methodologies have been proposed for the routine detection of MBL-producing Pseudomonas aeruginosa, particularly the MBL Etest and the MBL combined disc test, although there are concerns regarding their reliability.1–3 Recently, in order to increase the reliability of the test, some authors suggested the inclusion of ceftazidime, meropenem or imipenem resistance in the MBL-screening criteria.2 Nevertheless, in very specific circumstances, such as the case of mucoid P. aeruginosa isolates, these methods are stated as inefficient.1 Relying on EDTA for MBL inactivation, these methods are prone to false-positive results since the chelating agent itself can increase membrane permeability, thus increasing the chance of a bactericidal effect.4 Given the significance of MBL screening for routine laboratories, it seems important to present data that highlight problems with established methodologies. In the present work we assessed the reliability of the MBL Etest and combined disc test methods for the detection of MBL-producing P. aeruginosa isolates. From 2004 to 2008, 51 multidrug-resistant, non-mucoid P. aeruginosa isolates were recovered from different non-cystic fibrosis patients, hospitalized in nine distinct wards in South Hungary (Szeged). Within this time frame, three different outbreaks were detected at the intensive care unit, involving 5, 7 and 29 isolates each, and presenting identical inhibition zone profiles. Strain HMV-2, a blaVIM-2 producer (kindly provided by Professor Luis Martinez-Martinez), was used as a positive control for MBL detection, and P. aeruginosa ATCC 27853 was used both as a recommended control for susceptibility tests and as a negative control for MBL assays. Identification was performed using the VITEK 2 system and susceptibility was determined by the disc diffusion method. MBL detection included the MBL Etest (AB Biodisk, Solna, Sweden) and the MBL combined disc test, performed on Mueller–Hinton agar medium using 10 µg imipenem discs (Oxoid Ltd, Basingstoke, UK) and imipenem discs supplemented with EDTA (750 µg) in parallel with blank paper discs supplemented with EDTA alone. A bioassay and a carbapenem hydrolysis assay with cell-free extracts were performed, in order to detect carbapenemase activity as previously described.5 MBL presence was confirmed by PCR aimed at detecting blaSPM-1, blaGIM-1 and the most common blaVIM and blaIMP genes.6 Class 1 integrons were characterized by PCR and sequencing. Strain clonal relatedness was assessed by PFGE.6 All isolates were resistant to both imipenem and meropenem (MIC ≥ 16 mg/L), and, with one exception, additionally exhibited ceftazidime resistance, which was used as a marker for potential MBL production.2 PFGE typing revealed no clonal relatedness among the isolates. MBL detection results, both by MBL Etest and MBL combined disc test, were negative for two isolates. The remaining isolates fulfilled the necessary criteria for identification as an MBL-producing strain (a reduction of imipenem MIC by three or more 2-fold dilutions in the presence of EDTA for the Etest, and a zone diameter difference between imipenem + EDTA and imipenem discs of >7 mm for the MBL combined disc test). In contrast, the bioassay with cell-free extracts as well as the carbapenem hydrolysis assay with spectrophotometric analysis produced negative results. This pattern was later confirmed by PCR, since no MBL genes were detected. Nevertheless, it was found that four strains harboured different sized class 1 integrons carrying only aminoglycoside resistance genes. Summarized results are presented in Table 1. Comparison of MBL detection methods in different carbapenem-resistant P. aeruginosa strains IPM, imipenem; NT, not tested. aA, B and C are representative strains from three different outbreaks. Comparison of MBL detection methods in different carbapenem-resistant P. aeruginosa strains IPM, imipenem; NT, not tested. aA, B and C are representative strains from three different outbreaks. A high rate of false-positive results was obtained with both the MBL Etest and MBL combined disc test, even when the most recent MBL-screening criteria were adopted.1,2 Our results stress the importance of conducting a control test using a disc containing solely EDTA. In fact, the obtained EDTA inhibition zone diameters proved to be significantly bigger than expected [t-test against constant (0): N = 10, t = 16.58, P < 0.001]. Additionally, the sole usage of EDTA embedded discs produced growth inhibition effects similar to the difference between the inhibition zones obtained from imipenem alone and imipenem + EDTA [t-test for dependent samples: EDTA average = 11.80, (imipenem + EDTA)−imipenem average = 9.70, N = 10, t = 1.50, P = 0.17]. The combination of the above results clearly shows that the increased inhibition zones obtained for imipenem + EDTA seem to be due to a bactericidal effect of EDTA, probably through a membrane permeabilizing effect, rather then a chelating action upon any hypothetical MBL. In fact, the resistance phenotype to carbapenems observed in the studied strains could be linked to the overexpression of efflux pumps involved in the extrusion of carbapenems and/or OprD deficiency (data not shown) rather than the production of MBLs as could be suspected by the interpretation of the MBL Etest and MBL combined disc test. These results highlight the need for considerable caution, at least in some specific geographical regions, in the routine application of the Etest or the combined disc test as a carbapenemase-screening method. Alternatively, an easy-to-perform assay using cell-free extracts, where EDTA interference is avoided, seems preferable (see Table 1). This study was supported by a fellowship from the Federation of European Microbiological Societies (FEMS) FRF 2007-2. None to declare. This work was presented at the Forty-eighth Annual Interscience Conference on Antimicrobial Agents and Chemotherapy and the Infectious Diseases Society of America Forty-sixth Annual Meeting, Washington, DC, 2008 (Abstract C2-213). We wish to thank Professor Luis Martinez-Martinez for providing a control strain (VIM-2-producing P. aeruginosa) used in this study.