Comparison of five methods for detection of carbapenemases in Enterobacterales with proposal of a new algorithm

Clinical Microbiology and Infection 25 (2019) 1286.e9e1286.e15

Contents lists available at ScienceDirect

Clinical Microbiology and Infection journal homepage: www.clinicalmicrobiologyandinfection.com

Original article

Comparison of five methods for detection of carbapenemases in Enterobacterales with proposal of a new algorithm € ttig 3, A. Saleh 1, Y. Stelzer 1, L. Lucena Baeza 1, N. Pfennigwerth 2, C. Greissl 1, S. Go S.G. Gatermann 2, A. Hamprecht 1, 4, * 1)

Institute for Medical Microbiology, Immunology and Hygiene, University Hospital of Cologne, Cologne, Germany Department of Medical Microbiology, Ruhr University Bochum, Bochum, Germany Institute of Medical Microbiology and Infection Control, Hospital of Johann Wolfgang Goethe-University, Frankfurt am Main, Germany 4) DZIF (German Centre for Infection Research), Partner Site Bonn-Cologne, Germany 2) 3)

a r t i c l e i n f o

a b s t r a c t

Article history: Received 14 December 2018 Received in revised form 1 March 2019 Accepted 2 March 2019 Available online 18 March 2019

Objectives: The aim of this study was to evaluate the performance of five different carbapenemase tests and to develop an algorithm which will permit the detection of most common and rare carbapenemases in routine microbiology laboratories. Methods: The immunochromatographic tests CARBA-5 (NG), RESIST-4 O.K.N.V. (Coris), the colorimetric b-CARBA (BioRad), a newly developed carbapenem-inactivation method (CIM) supplemented with zinc (zCIM), and the Xpert Carba-R (Cepheid) were challenged with a collection of 189 molecularly characterized Enterobacterales isolates, including 146 carbapenemase producers (CPE): VIM (n ¼ 48), OXA-48like (n ¼ 40), NDM (n ¼ 29), KPC (n ¼ 13), IMI (n ¼ 9), IMP (n ¼ 9), OXA-58 (n ¼ 2), and GES (n ¼ 2). Results: The overall sensitivity/specificity values for the five carbapenemase detection tests were 84.2% (CI 77.6e89.2%)/100% (CI 91.8e100%) for RESIST-4, 88.2% (CI 82.1e92.4%)/100% (CI 91.8e100%) for CARBA-5, 88.2% (CI 82.1e92.4%)/100% (CI 91.8e100%) for Xpert Carba-R, 73.7% (CI 66.2e80.0%)/100% (CI 93.4e99.0%) for b-CARBA, and 97.4% (CI 87.9e99.6%)/97.7% (CI 87.9e99.6%) for zCIM. The four common carbapenemases (KPC, OXA-48-like, NDM, and VIM) were detected with 97.6% sensitivity by all tests except for b-CARBA (76.6% (CI 68.4e83.2%)). IMI and GES were only detected by zCIM (sensitivity 90.9% (CI 62.3e98.4%)). Based on these results a new algorithm was developed, consisting of an immunochromatographic assay as the first test followed by zCIM, which allows detection of 99.3% of all carbapenemases assessed. Conclusions: Except for b-CARBA, all methods showed excellent sensitivity/specificity for the detection of the four most frequent carbapenemases. With the new algorithm, rare variants can also be detected. It is rapid, simple, and inexpensive and can be performed in any microbiology laboratory, as no PCR equipment is required. L. Lucena Baeza, Clin Microbiol Infect 2019;25:1286.e9e1286.e15 © 2019 The Authors. Published by Elsevier Ltd on behalf of European Society of Clinical Microbiology and Infectious Diseases. This is an open access article under the CC BY-NC-ND license (http:// creativecommons.org/licenses/by-nc-nd/4.0/).

Editor: F. Allerberger Keywords: b-CARBA CARBA-5 Carbapenem-inactivation method GeneXpert Immunochromato graphic assay IMP KPC NDM OXA-48 VIM

Introduction The worldwide emergence of antibiotic resistance severely limits therapeutic options and is a significant global public health threat [1]. Resistance to carbapenems in Enterobacterales can be caused by hyperproduction of AmpC b-lactamases or extendedspectrum b-lactamases (ESBLs) combined with altered membrane

* Corresponding author: A. Hamprecht, University of Cologne, Institute for Medical Microbiology Immunology and Hygiene, Cologne, Germany. E-mail address: [email protected] (A. Hamprecht).

permeability or by the production of carbapenemases (carbapenemase-producing Enterobacterales (CPE)). To contain the further dissemination of CPE, the rapid and accurate identification of carbapenemases is of paramount importance for both epidemiological and infection control purposes [2]. Additionally, rapid identification of some carbapenemases can help to guide therapy, as most isolates producing OXA-48-like or Klebsiella pneumoniae carbapenemases (KPC) are susceptible to ceftazidimeeavibactam. Although a number of phenotypic and genotypic diagnostic methods for CPE detection have been developed, the performance of these tests varies considerably, depending on the enzyme and species. Colorimetric tests, such as the RAPIDEC CARBA NP

https://doi.org/10.1016/j.cmi.2019.03.003 1198-743X/© 2019 The Authors. Published by Elsevier Ltd on behalf of European Society of Clinical Microbiology and Infectious Diseases. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

L.L. Baeza et al. / Clinical Microbiology and Infection 25 (2019) 1286.e9e1286.e15

rieux, Nürtingen, Germany) or the b-CARBA (BioRad, (bioMe Marnes-la-Coquette, France), show variable sensitivity for b-lactamases with low carbapenemase activity [3]. The carbapenem inactivation method (CIM) [4] or its modified version (mCIM) [5,6] require an extra overnight culture for a definitive result and hence delay the time-to-result. These phenotypic assays detect carbapenem hydrolysis but cannot identify the respective carbapenemase. Immunochromatographic tests (ICT), such as the recently developed RESIST-4 O.K.N.V. (Coris BioConcept, Gembloux, Belgium) or the CARBA-5 (CARBA-5, NG biotech, Guipry, France), represent the fastest methods available, demonstrating high sensitivity and specificity in several studies [7e9]. PCR is considered as the reference standard for carbapenemase detection, but it requires additional equipment, skilled staff and is not available in many laboratories, especially beyond core working hours. Additionally, only targeted genes can be detected, with new enzyme variants possibly being missed. The goal of this study was to evaluate the performance of five carbapenemase detection methods in Enterobacterales which are suitable for routine microbiology laboratories. Two ICTs, a colorimetric test, a modified CIM test (zCIM), and the automated qPCRbased GeneXpert Carba-R assay (Cepheid, Frankfurt, Germany) were compared. Based on our results, an algorithm for improved detection of carbapenemases in the routine microbiology laboratory was developed. Materials and methods Bacterial isolates A total of 189 clinical Enterobacterales isolates, all molecularly characterized by PCR and sequencing of carbapenemase genes, were employed to challenge the assays. All clinical isolates were obtained from hospitalized patients in Germany, either from previous studies [10e15] or from the German National Reference Center for multidrug-resistant Gram-negative bacteria at the University of Bochum. The collection was composed of 43 carbapenemase-negative isolates and 146 CPE producing a total of 152 carbapenemases, including isolates belonging to Ambler classes A (n ¼ 22), B (n ¼ 80), D (n ¼ 38), and six isolates producing two carbapenemases (Table 1). Out of all carbapenemase-negative strains (n ¼ 43), 21 (48.8%) were non-susceptible to at least ertapenem (MIC 1 mg/L), mostly by a combination of ESBL (n ¼ 27) and/or AmpC (n ¼ 12) and porin loss. For further characterization of isolates please see supplementary material. For the comparison of the assays, a suspension equivalent to a 0.5 McFarland turbidity standard was inoculated on MuellereHinton agar plates (MHA; Oxoid, Wesel, Germany) and incubated overnight at 37 C. All five methods were performed on the same day using the same inoculum. Except for the Xpert CarbaR (automated interpretation), all other tests were read by a second person who was blinded to the molecular characterization. All tests were performed according to the manufacturer's recommendations unless stated otherwise.

1286.e10

previously inoculated with the susceptible Escherichia coli indicator strain ATCC 25922 (0.5 McFarland). After an 18-hr incubation at 37 C, the inhibition zone was measured and the presence of microcolonies was recorded. If available, cut-offs published previously were used [4e6]; additionally new cut-offs were calculated based on the frequencies of CIM inhibition zone measurements (see Figs S3eS6). Among all conditions tested, TSB þ ZnSO4 (¼ zCIM) showed the overall best performance among the preliminary bacterial collection (sensitivity 97.8%, specificity 100%) using the cutoff 20 mm/>20 mm for positive/negative (Fig. S6 and Table S7). It was therefore chosen for the following evaluation of 189 isolates. Cepheid Xpert Carba-R A full 10-mL inoculation loop with a bacterial suspension equivalent to 0.5 McFarland standard was added into the sample reagent. After homogenization by vortexing for 10 s, 1.7 mL was subsequently transferred into the Xpert Carba-R cartridge (version 2.0) and run on the GeneXpert platform (Cepheid).

b-CARBA test The b-CARBA test (BioRad) was performed as follows: for every sample, 40 mL of reagent 1 was mixed with 40 mL of reagent 2. A full 1-mL inoculation loop was harvested from MHA, mixed with the reagents, vortexed, and incubated at 37 C for 30 min before results were read. The protocol used here deviated from the manufacturer's recommendations by the use of MHA instead of other media (e.g. Columbia blood agar (CBA)). CARBA-5 Five drops of extraction buffer were mixed with a full 1-mL inoculation loop of bacteria harvested from MHA and 100 mL of this suspension was transferred into the CARBA-5 cassette (NG biotech); results were read after 15 min of incubation at room temperature. RESIST-4 O.K.N.V The test was performed as previously published [16]. Briefly, 12 drops of LY-A buffer were mixed with a full 1-mL inoculation loop of bacteria, which has previously been demonstrated to be the optimal amount for analysis [10]. Three drops of diluted sample was added into each well of the two cassettes of RESIST-4 O.K.N.V. (Coris). Results were read after 15 min of incubation at room temperature. Statistics The sensitivity and specificity was calculated and comparison with molecular characterization which served as the reference standard. Additionally, the 95% confidence intervals (CIs) and the Youden index were calculated.

Carbapenem inactivation method

Results

For assessment of CIM and its modified versions, a series of different conditions were tested on a preliminary collection of 46 carbapenemase-positive and 11 carbapenemase-negative isolates (see Table S1). A full 10-mL inoculation loop of bacteria was suspended in 400 mL of Tryptic Soy Broth (TSB; BD, Heidelberg, Germany) or distilled water with or without ZnSO4 (0.3 mM final concentration), and a 10-mg meropenem disc (Oxoid). After a 2-hr incubation at 37 C, the disc was placed on a MHA plate

Comparison of the performances of carbapenemase-detection methods in Enterobacterales The overall sensitivity of the five tests was 73.7% (CI 66.2e80.0%) for b-CARBA, 84.2% (CI 77.6e89.2%) for RESIST-4, 88.2% (CI 82.1e92.4%) for CARBA-5, 88.2% (CI 82.1e92.4%) for Xpert Carba-R, and 97.4% (CI 93.4e99.0%) for zCIM (Table 2). When considering only the four most common carbapenemases (NDM, KPC, OXA-48,

1286.e11

L.L. Baeza et al. / Clinical Microbiology and Infection 25 (2019) 1286.e9e1286.e15

Table 1 Overview of the final bacterial isolate collection used for the evaluation of the five carbapenemase detection methods

Carbapenemase-positive Ambler class A GES GES-2 GES-25 IMI IMI-1 IMI-2 IMI-3 IMI-4 IMI-9 IMI-10 IMI-12 IMI-14 IMI-16 KPC KPC-2 KPC-3 Ambler class B IMP IMP-1 IMP-4 IMP-8 IMP-13 IMP-14 IMP-22 IMP-28 IMP-50 NDM NDM-1 NDM-3 NDM-4 NDM-5 NDM-7 NDM-8 NDM-9 VIM VIM-1 VIM-2 VIM-4 VIM-5 VIM-26 VIM-27 VIM-31 VIM-39 VIM-46 VIM-51 VIM-52 VIM-54 VIM-56 VIM-58 VIM-59 Ambler class D OXA OXA-48-like OXA-48 OXA-162 OXA-181 OXA-204 OXA-232 OXA-244 OXA-245 OXA-370 OXA-58 Isolates with 2 carbapenemases OXA-232þNDM-1 KPC-2þVIM-1 OXA-181þNDM-5 OXA-48þNDM-1 OXA-48þVIM-1-like Carbapenemase-negative Total number

K. pneumoniae

E. coli

E. cloacae

C. freundii

E. aerogenes

S. marcescens

P. mirabilis

Others*

Subtotal

47 9

32

32 9

19 2 1

2

5

3

6 2 1 1

146 22 2 1 1 9 1 1 1 1 1 1 1 1 1 11 9 2 80 9 2 1 1 1 1 1 1 1 26 17 1 1 2 2 1 2 45 21 4 5 1 2 1 1 3 1 1 1 1 1 1 1 38

1 9 1 1 1 1 1 1 1 1 1 9 8 1 18 3 1 1

1

13 1

23

1 1

1 16 2 1

5 3

1

4

1 1

1 1

1 1 1 1

10 7

1 2 5 2 1

8 4 1 1 1 1

4 3

4 4

19 9 1 2

1 1 1 1

1

1 1

13 5 1 2 1

1

3 1 1 1

2 1 1 3 1 1 1 1 1 1 17

16

17 9 2

16 9 1 3

1 1 1 2 1

1 1 1

2

2

2 2

1

2 1

2 3 1 2

10 57

3

1 1 1 20 52

6 38

1 20

4 6

5

1 4

1 7

36 20 4 3 1 3 2 2 1 2 6 1 2 1 1 1 43 189

* VIM-1 Escherichia hermannii (n ¼ 1), GES-2 Klebsiella aerogenes (n ¼ 1), Klebsiella oxytoca (VIM-2, n ¼ 1 and VIM-4, n ¼ 1), KPC-2 Citrobacter braakii (n ¼ 1), NDM-1 Raoultella ornithinolytica (n ¼ 1) and carbapenemase-negative Klebsiella oxytoca (n ¼ 1). The main carbapenemase groups are marked in bold.

100 (34.2e100) 100 (34.2e100) 0 (0e65.8) 0 (0.0e65.8) 0 (0.0e65.8)

Sens, sensitivity; Spec, specificity; CI, confidence interval.

100 (91.2e100) 100 (91.2e100) 100 (91.2e100) 100 (91.2e100)

100 (91.2e100)

100 (88.3e100) 93.8 (83.2e97.9) 100 (70.1e100) 100 (91.6e100) 41.4 (25.5e59.3) 75.0 (61.2e85.1) 100 (70.1e100) 100 (91.6e100) 96.6 (82.8e99.4) 100 (92.6e100) 55.6 (26.7e81.1) 95.2 (84.2e98.7) 93.1 (78.0e98.1) 100 (92.6e100) 0 (0e29.9) 95.2 (84.2e98.7)

100 (88.3e100) 100 (92.6e100) 44.4 (18.9e73.3) 95.2 (84.2e98.7)

100 (77.2e100) 50 (9.5e90.6) 100 (70.1e100) 96.5 (90.2e98.8) 100 (77.2e100) 0 (0e65.8) 0 (0e29.9) 66.3 (55.8e75.4) 100 (77.2e100) 0 (0e65.8) 0 (0e29.9) 94.2 (87.1e97.5) 100 (77.2e100) 0 (0e65.8) 0 (0e29.9) 87.2 (78.5e92.7)

100 (77.2e100) 0 (0e65.8) 0 (0.0e29.9) 94.2 (87.1e97.5)

73.7 (66.2e80.0) 100 0.73 (91.8e100) 54.2 (35.1e72.1) 88.2 (82.1e92.4) 100 0.88 (91.8e100) 54.2 (35.1e72.1) 84.2 (77.6e89.2) 100 0.84 (91.8e100) 54.2 (35.1e72.1)

88.2 (82.1e92.4) 100 0.88 (91.8e100) 54.2 (35.1e72.1)

1286.e12

All carbapenemases (n ¼ 152) Ambler class A (n ¼ 24) KPC (n ¼ 13) GES (n ¼ 2) IMI (n ¼ 9) Ambler class B (n ¼ 86) NDM (n ¼ 29) VIM (n ¼ 48) IMP (n ¼ 9) Ambler class D (n ¼ 42) OXA-48-like (n ¼ 40) OXA-58 (n ¼ 2)

zCIM

b-CARBA

Spec % (CI) Youden Sens % (CI) index Spec % (CI) Youden Sens % (CI) index

Xpert Carba R CARBA-5 RESIST-4 O.K.N.V.

In this study, we assessed the performance of five phenotypic and genotypic methods for carbapenemase detection using a large collection of clinical Enterobacterales isolates and evaluated their applicability in the clinical setting. Several studies assessing the immunochromatographic RESIST-3 O.K.N. have been published, showing high sensitivity and specificity results [10,15,17]. Only two studies analysed the performance of RESIST-4 O.K.N.V., demonstrating 100% sensitivity for KPC, OXA, and VIM variants [9,16]. For NDM a sensitivity of 83.3% was recorded in one study as a result of two false-negatives (one Proteus mirabilis and one Providencia stuartii) [9]. In the second study, the sensitivity for NDM was 96.6% when tested from MHA, but increased to 100% when tested from CBA, zinc-supplemented MHA or from the inhibition zone of an ertapenem disc [16]. The false-negative isolate from MHA was also an isolate of P. mirabilis. In the present study, the only NDM-producing P. mirabilis tested positive, but two isolates co-producing NDM were false negatives. CARBA-5 detected all KPC, OXA, VIM, and NDM (except for one K. pneumoniae co-producing OXA-232/NDM-1), which is in

Table 2 Performance of the five assays for detection of carbapenemase production

Discussion

Spec % (CI) Youden Sens % (CI) index

Since none of the assays detected all carbapenemases an algorithm was developed combining two different methods (Table S10). The overall highest sensitivity of 99.3% (CI 96.2e99.9%)/specificity 97.7% (CI 87.9e99.6%) can be achieved with the combination of zCIM with either CARBA-5, Carba-R or RESIST-4. If CARBA-5 or RESIST-4 are used as a first assay, the four most common carbapenemases are detected within 15 min (Fig. 1). If positive, no further testing is necessary. In case of a negative result, zCIM is performed as a second test for phenotypic detection of rare carbapenemases. If zCIM is also negative, a carbapenemase is highly unlikely. If zCIM results positive, further work-up is necessary, e.g. by referring to a reference centre, by additional PCRs for rare carbapenemases or by whole-genome sequencing if available.

Sens % (CI)

Development of an algorithm for the detection of carbapenemases in the clinical laboratory.

Spec % (CI) Youden Sens % (CI) index

Spec % (CI) Youden index

VIM), the sensitivity was 77.7% (CI 69.8e84.0%) for b-CARBA, 98.5% (CI 94.6e99.6%) for RESIST-4, 99.2% (CI 95.8e99.9%) for CARBA-5, 100% (CI 97.1e100%) for Xpert Carba-R, and 97.7% (CI 93.4e99.2%) for zCIM (Table S8). KPC and OXA-48-like were detected 100% (CI 93.2e100%) by all tests, while NDM and VIM were more frequently missed, with the lowest sensitivity being recorded for b-CARBA (41.4%, CI 25.5e59.3%, and 75.0%, CI 61.2e85.1%, respectively). Of note, in isolates co-producing NDM and OXA-48-like the carbapenemase NDM was not detected by CARBA-5 in one out of three and RESIST-4 in two out of three isolates. b-CARBA and zCIM were the only assays which were positive in all IMP-producing isolates (sensitivity 100%, CI 70.1e100%), followed by CARBA-5 (sensitivity 55.6%, CI 26.7e81.1%), detecting IMP-1/-4/-8/-22 and Xpert Carba R (sensitivity 44.4%, CI 18.9e73.3%), detecting IMP-1/-4/-28), see Table S9. All IMI variants and one GES-25 were only detected by zCIM, as they are not targeted by the ICTs or Xpert Carba-R. OXA-58 was only detected using zCIM and b-CARBA. All assays showed 100% (CI 91.8e100%) specificity except for the zCIM (97.7%, CI 87.9e99.6%). Only one AmpC producing Enterobacter cloacae isolate with decreased permeability gave rise to a false-positive result for zCIM, while other tests did not produce any false-positive results. The Youden index ranged from 0.73 for b-CARBA to 0.95 for zCIM (Table 2). A comparison describing the main features of each test is shown in Table 3.

97.4 (93.4e99.0) 97.7 0.95 (87.9e99.6) 95.8 (79.8-99.3)

L.L. Baeza et al. / Clinical Microbiology and Infection 25 (2019) 1286.e9e1286.e15

1286.e13

Table 3 Comparison of main features of each carbapenemase detection method CARBA-5

b-CARBA

zCIM

XpertCarba-R

All supplied by manufacturer

All supplied by manufacturer

All supplied by manufacturer

Conventional laboratory equipment

GeneXpert instrument





Room temperature

Room temperature

4 C

4 C

Room temperature

5 min 15 min

5 min 15 min

5 min 30 min

10 min 2 h plus overnight culture

10 min 1 hour

1 min

1 min

1 min

5 min

1 min

21 min ~15V Immunochromatographic test

21 min ~15V Immunochromatographic test

36 min ~5V Colorimetric test

1 h11 min ~30V Multiplex-PCR

Result

OXA-48 ,KPC, NDM, VIM

Sample type

OXA-48, KPC, NDM, VIM, some IMP variants Bacterial colonies

Bacterial colonies, some clinical samples (e.g.urine or blood cultures [15]) Straightforward for most isolates; Straightforward for most isolates; faint bands can occur for some faint bands can occur for some NDM/VIM isolates NDM/VIM isolates Fast, simple moderate costs Fast, simple moderate costs differentiation of the 5 most differentiation of the 4 most common carbapenemases common carbapenemases

Bacterial colonies, some clinical samples (blood cultures [28])

Bacterial colonies

OXA-48, KPC, NDM, VIM, some IMP variants Bacterial colonies, some clinical samples [29]

Subjective (colour change)

Subjective if microcolonies present

Automatic interpretation

Simple, inexpensive differentiation CPE/non-CPE detection of rare/new carbapenemases

High reproducibility relatively fast differentiation of the 5most common carbapenemases

Interpretation

Strengths

Limitations

False-negatives with weak MBLs when harvested from MHA in absence of antibiotic pressure [10,16]

Reference

[9]

*

20 h 15 min ~1V Carbapenem hydrolysis assay (activity test) Carbapenemase positive/negative Carbapenemase positive/negative

Fast, simple differentiation CPE/non-CPE only detection of some rare carbapenemase variants (e.g. OXA-58) possible False-negatives with weak or low-level carbapenemase False-negatives with producers, especially withMBLs weak or low-level carbapenemase producers, especially with MBLs and IMI/GES [7] [3]

Subject to variation depending on country/distributor.

Expensive False-positives in some isolates with Additional equipment AmpC and porin loss False-negatives with weak carbapenemases required Long time-to-result This study

[18]

L.L. Baeza et al. / Clinical Microbiology and Infection 25 (2019) 1286.e9e1286.e15

Special equipment necessary Storing conditions Time for Preparation Incubation/ reaction Reading/ Interpretation Definitive result Price per test* Test principle

RESIST-4 O.K.N.V.

L.L. Baeza et al. / Clinical Microbiology and Infection 25 (2019) 1286.e9e1286.e15

agreement with previous results [7,8]. This study demonstrated the additional detection of NDM-8, VIM-5/-26/-27/-31/-39/-46/-51/52/-54/-56/-58/-59, OXA-370, and IMP-22 (Table S9). IMP-13/-14/28/-50 were not detected, with IMP-13/-14 proving also negative as published elsewhere [8]. Our findings with Xpert Carba-R are in agreement with the manufacturer's specification and with results from several studies, which demonstrated that only IMP variants belonging to the IMP-1 family can be detected [18e20]. Xpert Carba-R was the only assay which detected all carbapenemases in double-carbapenemase producers. The b-CARBA test proved to be the least sensitive of all methods evaluated here. Several studies showed a wide range of sensitivity, ranging from 64.9% to 97.3% [3,21e24]. The performance depends on the carbapenemase but is also considerably influenced by the inoculum, incubation temperature, and time [3,21]. The overall sensitivity found in the present study was 73.7%, which is mainly the result of the poor detection of MBLs. This could be related to culture on MHA, which is a medium not recommended for this test. When retested from CBA or MHA supplemented with 50 mg/L zinc sulphate, sensitivity improved to 95.9% (CI 91.3e98.1%) and 96.6% (CI 92.2e98.5%), respectively. A better performance from CBA or zinc supplemented agars, which increase the detection rate of MBL producers due to higher enzymatic activity in the presence of zinc, has also been shown for other tests, e.g. ICTs [10,15,16]. The widest spectrum of carbapenemases was detected by zCIM, including rare variants. A positive zCIM test indicates hydrolytic activity and is therefore able to detect also carbapenemases not targeted by the ICTs or Xpert Carba-R. While CIM and mCIM have been shown to be highly sensitive and specific [4,25,26], among our isolates MBLs were only partly detected with the original CIM test.

1286.e14

For that reason, we developed zCIM, which showed improved sensitivity (97.8%) compared with the original CIM (69.6%; Table S7) and has a shorter incubation time (2 hr) than mCIM (4 hr). Inhibitor-based combination disc tests (e.g. with EDTA and boronic acid) are recommended by EUCAST [27] for carbapenemase detection and could be an easy-to-perform alternative to zCIM (please see supplementary material). Among the 22 isolates producing IMI, GES, IMP, and OXA-58 which are currently not or only partly detected by CARBA-5, RESIST-4, or Xpert Carba-R, the sensitivity of inhibitor tests was 13/22 (59.1%, CI 38.7e76.7%). When testing all 189 isolates with inhibitors, sensitivity was higher (84.9%, CI 78.2e89.8%), but specificity was moderate (76.7%, CI 62.3e86.9%). For optimized carbapenemase detection, the new algorithm incorporating an ICT þ zCIM (Fig. 1) has the advantage of being rapid, affordable, and simple. With such a combination, the four most common carbapenemases can be identified within 15e20 min, with the less frequent ones being detected within a day. Compared with a single test (sensitivity 73.7e97.4%), the combination of either CARBA-5 or RESIST-4 with zCIM improves sensitivity to 99.3%. This study has several limitations. We used MHA since additional tests are mostly done from this agar, e.g. after carbapenem non-susceptibility has been detected by a disc diffusion test. Therefore, the use of MHA is likely to produce the result most close to real life in the routine laboratory. Additionally, this was an analysis based on a collection containing many rare carbapenemases in order to challenge the tests. The performance of the assays is likely to be better in the routine laboratory, where the four main carbapenemases mostly occur. When taking into account only OXA48-like, KPC, NDM, and VIM, the overall sensitivity was higher in all

Fig. 1. Algorithm for the detection of carbapenemases in the routine laboratory.*OXA-48-like, KPC, VIM, NDM, (IMP) **Multiplex PCR targeting rare carbapenemases (e.g.IMI, GES etc.), whole genome sequencing, other tests.

1286.e15

L.L. Baeza et al. / Clinical Microbiology and Infection 25 (2019) 1286.e9e1286.e15

assays (98.5% for RESIST-4, 99.2% for CARBA-5, 100% for Xpert Carba-R, 77.7% for b-CARBA, and 97.7% for zCIM) (Table S8). To the best of our knowledge this is the first study to compare the two ICTs RESIST-4 and CARBA-5 and also the first to compare ICTs to the Xpert Carba-R or the b-CARBA. A large number of molecularly characterized isolates were analysed, comprising 53 different carbapenemases in 12 species. All assays were evaluated in a blinded fashion to exclude reading bias. Furthermore, an easy algorithm for improved detection was developed which will permit the detection of a wide range of carbapenemases in the routine microbiology laboratories without the need for molecular methods or additional equipment. In conclusion, most evaluated assays showed good sensitivity and specificity, at least for the common carbapenemase families OXA-48-like, KPC, VIM, and NDM and are suitable for the routine microbiology laboratory, thereby decreasing the time to detection and the need to refer these isolates to reference laboratories. However, the detection of rare carbapenemases (e.g. IMI, GES, IMP, or OXA-58) is still problematic with most of the commercially available tests. The combination of two tests, e.g. CARBA-5 þ zCIM, will enable most laboratories to detect these rare variants at low costs. This could help to optimize patient treatment and to limit the further spread of CPE. Transparency declaration A.H. has received lecture honoraria from OXOID, Becton Dickinson and Bruker Daltonics, outside the submitted work. S.G.G. rieux, received lecture honoraria from Beckman Coulter and bioMe outside the submitted work. All other authors declare no conflict of interest. This study was funded by grants from the Faculty of Medicine, University of Cologne and supported by the Robert Koch Institute with funds provided by the German Ministry of Health (grant no. 1369-402). AH was supported by the DZIF (German Centre for Infection Research). Acknowledgements RESIST-4 tests were provided by Coris BioConcept and CARBA-5 tests by Virotech Diagnostics Germany/NG Biotech free of charge. Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi.org/10.1016/j.cmi.2019.03.003. References [1] Global antimicrobial resistance surveillance system (GLASS) report: early implementation 2016-2017. Geneva: World Health Organization; 2017. [2] Nordmann P. Carbapenemase-producing Enterobacteriaceae: overview of a major public health challenge. Med Mal Infect 2014;44:51e6. [3] Mancini S, Kieffer N, Poirel L, Nordmann P. Evaluation of the RAPIDEC(R)® CARBA NP and beta-CARBA(R)® tests for rapid detection of Carbapenemaseproducing Enterobacteriaceae. Diagn Microbiol Infect Dis 2017;88:293e7. [4] van der Zwaluw K, de Haan A, Pluister GN, Bootsma HJ, de Neeling AJ, Schouls LM. The carbapenem inactivation method (CIM), a simple and lowcost alternative for the Carba NP test to assess phenotypic carbapenemase activity in gram-negative rods. PLoS One 2015;10:e0123690. [5] Miller SA, Hindler JA, Chengcuenca A, Humphries RM. Use of ancillary carbapenemase tests to improve specificity of phenotypic definitions for carbapenemase-producing enterobacteriaceae. J Clin Microbiol 2017;55: 1827e36. [6] Pierce VM, Simner PJ, Lonsway DR, Roe-Carpenter DE, Johnson JK, Brasso WB, et al. Modified carbapenem inactivation method for phenotypic detection of carbapenemase production among enterobacteriaceae. J Clin Microbiol 2017;55:2321e33.

[7] Boutal H, Vogel A, Bernabeu S, Devilliers K, Creton E, Cotellon G, et al. A multiplex lateral flow immunoassay for the rapid identification of NDM-, KPC-, IMP- and VIM-type and OXA-48-like carbapenemase-producing Enterobacteriaceae. J Antimicrob Chemother 2018;73:909e15. [8] Hopkins KL, Meunier D, Naas T, Volland H, Woodford N. Evaluation of the NGTest CARBA 5 multiplex immunochromatographic assay for the detection of KPC, OXA-48-like, NDM, VIM and IMP carbapenemases. J Antimicrob Chemother 2018. [9] Kolenda C, Benoit R, Carricajo A, Bonnet R, Dauwalder O, Laurent F. Evaluation of the new multiplex immunochromatographic O.K.N.V K-Set assay for the rapid detection of OXA-48-like, KPC, NDM and VIM carbapenemases. J Clin Microbiol 2018. [10] Saleh A, Gottig S, Hamprecht AG. Multiplex immunochromatographic detection of OXA-48, KPC, and NDM carbapenemases: impact of inoculum, antibiotics, and agar. J Clin Microbiol 2018;56. € ttig S, Kaase M, Steinmann J, Gatermann S, Sommer J, et al. [11] Koroska F, Go Comparison of phenotypic tests and an immunochromatographic assay and development of a new algorithm for detection of OXA-48-like carbapenemases. J Clin Microbiol 2017;55:877e83. [12] Jazmati N, Hein R, Hamprecht A. Use of an enrichment Broth improves detection of extended-spectrum-beta-lactamase-producing enterobacteriaceae in clinical stool samples. J Clin Microbiol 2016;54:467e70. [13] Gruber TM, Gottig S, Mark L, Christ S, Kempf VA, Wichelhaus TA, et al. Pathogenicity of pan-drug-resistant Serratia marcescens harbouring blaNDM-1. J Antimicrob Chemother 2015;70:1026e30. [14] Jazmati N, Jazmati T, Hamprecht A. Importance of pre-enrichment for detection of third-generation cephalosporin-resistant Enterobacteriaceae (3GCREB) from rectal swabs. Eur J Clin Microbiol Infect Dis 2017;36:1847e51. [15] Hamprecht A, Vehreschild JJ, Seifert H, Saleh A. Rapid detection of NDM, KPC and OXA-48 carbapenemases directly from positive blood cultures using a new multiplex immunochromatographic assay. PLoS One 2018;13:e0204157. [16] Greissl C, Saleh A, Hamprecht A. Rapid detection of OXA-48-like, KPC, NDM, and VIM carbapenemases in Enterobacterales by a new multiplex immunochromatographic test. Eur J Clin Microbiol Infect Dis 2019;38:331e5. https:// doi.org/10.1007/s10096-018-3432-2. [17] Glupczynski Y, Jousset A, Evrard S, Bonnin RA, Huang TD, Dortet L, et al. Prospective evaluation of the OKN K-SeT assay, a new multiplex immunochromatographic test for the rapid detection of OXA-48-like, KPC and NDM carbapenemases. J Antimicrob Chemother 2017;72:1955e60. [18] Smith M, Diederen B, Scharringa J, Leversteijn-van Hall M, Fluit AC, Cohen Stuart J. Rapid and accurate detection of carbapenemase genes in Enterobacteriaceae with the Cepheid Xpert Carba-R assay. J Med Microbiol 2016;65: 951e3. [19] Dortet L, Fusaro M, Naas T. Improvement of the Xpert carba-R kit for the detection of carbapenemase-producing enterobacteriaceae. Antimicrob Agents Chemother 2016;60:3832e7. [20] Traczewski MM, Carretto E, Canton R, Moore NM, Carba RST. Multicenter evaluation of the Xpert carba-R assay for detection of carbapenemase genes in gram-negative isolates. J Clin Microbiol 2018;56. [21] Compain F, Gallah S, Eckert C, Arlet G, Ramahefasolo A, Decre D, et al. Assessment of carbapenem resistance in enterobacteriaceae with the rapid and easy-to-use chromogenic beta Carba test. J Clin Microbiol 2016;54: 3065e8. [22] Bernabeu S, Dortet L, Naas T. Evaluation of the beta-CARBA test, a colorimetric test for the rapid detection of carbapenemase activity in Gram-negative bacilli. J Antimicrob Chemother 2017;72:1646e58. [23] Fischer J, Simon T, Hamprecht A, Hos N, Moricke A, Cernaianu G, et al. Surgical implications for diagnosis and treatment of intestinal aspergillosis in pediatric patients with ALL. Eur J Pediatr Surg 2018;28:477e83. [24] Noel A, Huang TD, Berhin C, Hoebeke M, Bouchahrouf W, Yunus S, et al. Comparative evaluation of four phenotypic tests for detection of carbapenemase-producing gram-negative bacteria. J Clin Microbiol 2017;55: 510e8. [25] Tamma PD, Opene BN, Gluck A, Chambers KK, Carroll KC, Simner PJ. Comparison of 11 phenotypic assays for accurate detection of carbapenemaseproducing enterobacteriaceae. J Clin Microbiol 2017;55:1046e55. [26] Crowe A, Brenton L, Kingston M, Jardine D, Waters MJ. Comparison of the carbapenem inactivation method (CIM) and modified carbapenem inactivation method (mCIM) for the detection of carbapenemase-producing organisms. Pathology 2018 Dec;50:764e6. https://doi.org/10.1016/ j.pathol.2018.05.009. [27] Giske CG, Martinez-Martinez L, Canton R, Stefani S, Skov R, Glupczynski Y, et al. EUCAST guidelines for detection of resistance mechanisms and specific resistances of clinical and/or epidemiological importance. 2017. p. 1e43. [28] Meier M, Hamprecht A. Rapid detection of carbapenemases directly from positive blood cultures by the beta-CARBA test. Eur J Clin Microbiol Infect Dis 2018. [29] Tato M, Ruiz-Garbajosa P, Traczewski M, Dodgson A, McEwan A, Humphries R, et al. Multisite evaluation of cepheid Xpert carba-R assay for detection of carbapenemase-producing organisms in rectal swabs. J Clin Microbiol 2016;54:1814e9.