Eur J Clin Microbiol Infect Dis DOI 10.1007/s10096-014-2059-1

ARTICLE

Evaluation of phenotypic detection methods for metallo-β-lactamases (MBLs) in clinical isolates of Pseudomonas aeruginosa S. Peter & A. Lacher & M. Marschal & F. Hölzl & M. Buhl & I. Autenrieth & M. Kaase & M. Willmann

Received: 5 November 2013 / Accepted: 8 January 2014 # Springer-Verlag Berlin Heidelberg 2014

Abstract Metallo-beta-lactamase (MBL) production in Pseudomonas aeruginosa is a growing issue across the globe. Fast and reliable diagnostic tools are needed for appropriate implementation of infection control measures. In this study we evaluated the performance of three commercial combined disk tests, two EDTA based in-house combined disk tests and the Carba NP test in comparison to molecular detection of MBL genes on 133 meropenem non-susceptible non-duplicate Electronic supplementary material The online version of this article (doi:10.1007/s10096-014-2059-1) contains supplementary material, which is available to authorized users. S. Peter (*) : A. Lacher : M. Marschal : F. Hölzl : M. Buhl : I. Autenrieth : M. Willmann Institute of Medical Microbiology and Hygiene, University of Tübingen, Elfriede-Aulhorn-Str.6, 72076 Tübingen, Germany e-mail: [email protected] A. Lacher e-mail: [email protected] M. Marschal e-mail: [email protected] F. Hölzl e-mail: [email protected]

P. aeruginosa clinical isolates. The meropenem/DPA based commercial KPC+MBL-confirm ID kit (Rosco Diagnostica, Denmark) and the MASTDISCS™ ID carbapenemase (Enterobacteriaceae) detection disc set (MAST Diagnostics, UK) showed sensitivities of 31.1 % and 28.8 % and specificities of 69.3 % and 79.6 %, respectively. The total MBL confirm kit (Rosco Diagnostica, Denmark) contains imipenem/DPA and imipenem/EDTA combination disks. Evaluation of the single disk combinations revealed 84.4 % sensitivity and 81.8 % specificity for the imipenem/DPA assay and 86.7 % sensitivity and 51.1 % specificity for the imipenem/EDTA test. Applying both tests simultaneously resulted in a slightly higher sensitivity of 88.9 % but a lower specificity of 48.9 % when compared to the single tests alone. The Carba NP test showed 93.3 % sensitivity and 96.6 % specificity. All phenotypic combined disk tests lacked either sensitivity or specificity for the detection of MBL in P. aeruginosa. The Carba NP test showed excellent test properties, but suffers from drawbacks in handling and high costs. The optimal diagnostic approach needs to be chosen depending on the epidemiological situation, laboratory resources and availability of molecular confirmation tests.

M. Buhl e-mail: [email protected] I. Autenrieth e-mail: [email protected] M. Willmann e-mail: [email protected] S. Peter : M. Marschal : F. Hölzl : M. Buhl : I. Autenrieth German Center for Infection Research (DZIF), Partner Site Tübingen, Tübingen, Germany M. Kaase National Reference Laboratory for Multidrug Resistant Gram-negative Bacteria, Department of Medical Microbiology, Ruhr University Bochum, Bochum, Germany e-mail: [email protected]

Introduction Metallo-beta-lactamases (MBLs) belong to the class B carbapenemases and are characterized by their ability to inactivate most beta-lactam antibiotics, including carbapenems. MBLs have been described in numerous Enterobacteriaceae, but also in non-fermenters like Pseudomonas aeruginosa. Acquired MBL genes of P. aeruginosa are typically located on integron structures with mobile elements, and are often associated with additional antibiotic resistances, limiting therapeutic options for infections with these organisms [1, 2].

Eur J Clin Microbiol Infect Dis

P. aeruginosa is an important nosocomial pathogen, responsible for a wide range of hospital-acquired infections such as urinary tract infections, pneumonia and bloodstream infections [3]. The increasing incidence of MBL producing P. aeruginosa strains is a major challenge for infection control strategies within the hospital setting [4, 5]. Therefore, fast, reliable and cost-efficient diagnostic tools are needed for the detection of MBL producing P. aeruginosa. A number of different confirmation methods for MBL enzymes have been described; most important among these are phenotypic confirmation assays, molecular confirmation assays and a recently described carbapenem hydrolysis assay [6]. Combined disk tests are based on the inhibition of MBL enzymes by ethylenediaminetetraacetic acid (EDTA) or dipicolinic acid (DPA), leading to differences in zone diameters of carbapenem disks with or without the inhibitor. Numerous in-house combined disk tests have previously been established and evaluated for the detection of carbapenemase production in Enterobacteriaceae [7–14], Pseudomonas sp. and Acinetobacter sp. in various clinical settings [11, 15–18]. In addition, different combined disk tests are currently commercially available, e.g. the KPC+MBL confirm ID kit and the total MBL confirm kit (both Rosco Diagnostica, Denmark) or the MASTDISCS™ ID carbapenemase (Enterobacteriaceae) detection disc set (MAST Diagnostics, UK). These assays contain meropenem or imipenem disks alone and in combination with EDTA or DPA. Commercial detection assays provide several advantages for the adherence to diagnostic quality standards in the routine diagnostic laboratory, particularly with respect to storage conditions, standardized production, and documentation of test evaluation results. These commercial assays have been tested for numerous Enterobacteriaceae species expressing a large variety of carbapenemases [7–9, 19]. So far only few well conducted studies have addressed the performance of these commercial assays for the detection of MBL production in P. aeruginosa [20, 21]. Additionally, a carbapenem hydrolysis assay for the detection of carbapenemases in Enterobacteriaceae and Pseudomonas species has recently been described by Dortet et al. [6, 22, 23]. This assay is an elegant system based on a pH change which occurs when imipenem monohydrate is hydrolyzed by carbapenemase enzymes in the supernatant of the bacterial lysate. It has been evaluated in two additional studies, including two MBL and six MBL producing P. aeruginosa strains, respectively [24, 25]. VIM and IMP enzymes are the predominant MBLs in P. aeruginosa worldwide [26], as well as in Germany [27]. While PCR-based detection techniques for these genes provide high sensitivity and specificity [28], they are rather expensive, labor-intensive and require a well-equipped molecular facility. In the present study we aimed to evaluate the performance of the KPC+MBL confirm ID kit and Total MBL confirm kit (both Rosco Diagnostica), the MASTDISCS™ ID carbapenemase

(Enterobacteriaceae) detection disc set (MAST Diagnostics), two in-house combined disk tests described by Pitout et al. [28] and the carba NP test II described by Dordet el al. [22] in meropenem non-susceptible clinical P. aeruginosa isolates.

Materials and methods Clinical isolates One-hundred thirty-three P. aeruginosa clinical strains with reduced susceptibility to meropenem (disk diffusion zone diameter ≤ 23 mm) were collected at the Institute of Medical Microbiology and Hygiene, Tübingen, Germany over a period of 3 years. Strain identification was achieved by the means of a linear MALDI-TOF mass spectrometer (bioMérieux, Marcy l’Etoile, France). Duplicate isolates from the same patient were excluded from the study. Strains were stored and cultured on Columbia agar with sheep blood (Oxoid, Germany) over night prior to testing. Antibiotic susceptibility testing Disk diffusion susceptibility testing was performed according to EUCAST (the European Committee on Antimicrobial Susceptibility Testing) recommendations with the exception of colistin whose susceptibility was evaluated according to CLSI (Clinical and Laboratory Standards Institute) recommendations [29, 30]. Apart from ceftazidime disks (Oxoid, UK), all test disks were obtained from Becton Dickinson, Germany. Using overnight cultures, disk diffusion susceptibility testing was performed on Mueller Hinton agar (Mueller Hinton dehydrated culture medium, Becton Dickinson, prepared in-house) at a 0.5 McFarland standard followed by incubation at 35 °C for 16–18 h. The results were interpreted in accordance with either EUCAST clinical breakpoints [31] or, in the case of colistin, according to the CLSI breakpoints [32]. DNA extraction and detection of blaVIM, blaIMP and blaNDM genes DNA extraction was performed using the InstaGene matrix (Bio-Rad, Switzerland) following the instructions of the manufacturer. The detection of the blaIMP or blaVIM genes was performed by a multiplex PCR approach using the primer pairs IMP-A/IMP-B and VIM2004A/VIM2004B as described previously [28]. BlaNDM was detected as described previously using the primer pair NDM-1_a_fw and NDM-1_a_rev [33]. Isolates positive for either blaIMP, blaVIM or blaNDM genes were considered to be true MBL producers. Phenotypic metallo-beta-lactamase detection assays MASTDISCS™ ID carbapenemase (Enterobacteriaceae) detection disc set (MAST Diagnostics, UK) The MASTDISCS ID™ is an inhibitor-based combined disk diffusion assay. It was performed on Mueller Hinton agar (Mueller Hinton

Eur J Clin Microbiol Infect Dis

dehydrated culture medium, Becton Dickinson, prepared inhouse) using overnight cultures at a 0.5 McFarland standard followed by incubation at 35 °C for 18–24 h. The difference in zone diameters between disk A (carbapenem) and disk B (carbapenem + MBL inhibitor) was determined in millimeters. Disks C (carbapenem + KPC inhibitor) and D (AmpC inhibitor) were not analyzed, since the study was designed to evaluate the detection of MBL production. The recommended cut-off for MBL producing Enterobacteriaceae is ≥5 mm. Cut-offs ranging from ≥4 mm to ≥6 mm for P. aeruginosa were evaluated. The test is referred to as MASTDISCS ID carbapenemase. KPC+MBL-confirm ID kit (Rosco Diagnostica, Denmark) The KPC+MBL-confirm ID kit is a combined disk diffusion test. The test was performed as described in the manufacturer’s instructions using Mueller-Hinton plates (Mueller Hinton dehydrated culture medium, Becton Dickinson, prepared inhouse). The differences between the zones of inhibition of the meropenem disks (MRP 10) compared to the meropenem/DPA (MRPDP) disks were determined in mm. The disks for KPC and AmpC detection were not interpreted. The recommended cut-off for MBL production is a ≥5 mm difference between meropenem and meropenem/DPA. Cut-offs ranging from ≥4 mm to ≥6 mm for P. aeruginosa were evaluated in this study. Total MBL confirm kit (Rosco Diagnostica, Denmark) The total MBL confirm kit is a combined disk diffusion test. The test was performed as described in the manufacturer’s instructions using MacConkey agar (in-house). The differences between the zones of inhibition of the imipenem disk (IM 10) compared to the imipenem/DPA (IM+DP) and imipenem/ EDTA (IM10E) disks were determined in millimeters. Synergy between meropenem, imipenem and DPA disks was not evaluated, as synergy based assays are difficult to use in daily routine diagnostics. The recommended cut-offs for MBL production are ≥5 mm between imipenem and imipenem/DPA or ≥10 mm between imipenem and imipenem/EDTA. Additional cut-offs have been evaluated in this study. The Total MBL confirm kit has since been modified. At the time of submission of the manuscript, DPA synergy testing has been replaced by meropenem and meropenem/DPA combination disks (http://www.rosco.dk/ default.asp?mainmenu=3&submenu=65&webmanage= Total%20MBL%20Confirm%20Kit%2098016). In house combined disk test Imipenem/EDTA and Meropenem/ EDTA The combined disk test was performed as described previously [28]. In brief, overnight cultures at a 0.5 McFarland standard were inoculated on Mueller Hinton agar (Mueller Hinton dehydrated culture medium, Becton Dickinson, prepared in-house). Meropenem or imipenem disks (Becton Dickinson, Germany) were supplemented with 5 μl of 0.5 M EDTA (930 μg EDTA; AppliChem, Germany) and the

differences in zone diameters with and without EDTA were measured. The cut-off evaluated by Pitout et al. was ≥7 mm difference in zone diameter [28]. Alternative cut-offs were evaluated in this study. Carba NP test II The Carba NP test is based on a pH shift which occurs during hydrolysis of imipenem monohydrate by carbapenemases. It was performed as described by Dortet et al. [22] with slight modifications. In order to decrease the costs of the test, the total volume of the test format was reduced from 130 μl to 97.5 μl. We evaluated the altered test format with 15 isolates in parallel testing; no difficulties in test interpretation occurred. In brief, one calibrated loop (10 μl) of the strain was resuspended in 100 μl bacterial protein extraction reagent B-PER II (Thermo Scientific, USA) and vortexed for 60 s, followed by centrifugation at 10 000×g at room temperature. A total of 22.5 μl of the supernatant corresponding to the enzymatic bacterial suspension were added to 75 μl of phenol red solution (Merck, Germany) containing 3 mg/ml imipenem monohydrate (Sigma-Aldrich, Germany) and 0.1 mM ZnSO4 or 3 mg/ml or imipenem monohydrate + 0.003 M EDTA (AppliChem, Germany) prepared as described previously [22]. A control line without antibiotic and a control line without bacterial lysate were included in each run. The assay was interpreted as follows: MBL production was assumed if the imipenem-containing well changed its colour from red to yellow or orange while the combined imipenem/ EDTA well retained its red colour. A strain was interpreted negative when both wells remained red. Statistics Sensitivity and specificity were calculated for different combined disk assays and the Carba NP test, using the multiplex PCR approach as reference standard. The overall diagnostic precision of the combined disk tests was determined by receiver operating characteristic (ROC) analyses. For each test, the area under the ROC curve (AUC) was computed. AUC ranges from 0.5 (diagnosis is based on chance) to 1 (perfect precision). The imipenem/DPA and the combination imipenem/DPA and imipenem/EDTA of the Total MBL confirm ID (Rosco Diagnostica, Denmark), and the impenem/ EDTA and meropenem/EDTA (in-house) [28] tests were compared by employing a Wald test following an algorithm suggested by DeLong et al. [34]. For all analyses, Stata version 12.0 (Stat Corp., College Station, TX) was used. A P