Phenotypic Screening of Carbapenemases and Associated β-Lactamases in Carbapenem-Resistant Enterobacteriaceae

Abstract

Dissemination of carbapenem resistance among Enterobacteriaceae poses a considerable threat to public health. Carbapenemase gene detection by molecular methods is the gold standard but is available in only a few laboratories. The aim of this study was to test phenotypic methods for the detection of metallo-β-lactamase (MBL)- or Klebsiella pneumoniae carbapenemase (KPC)-producing Enterobacteriaceae and associated mechanisms of β-lactam resistance against a panel of 30 genotypically characterized carbapenem-resistant Enterobacteriaceae : 9 MBL, 7 KPC, 6 OXA-48, and 8 extended-spectrum β-lactamase (ESBL) or AmpC β-lactamases associated with decreased permeability. We used carbapenemase inhibitor-impregnated agar to test for carbapenem-resistant strains. Differences in the inhibition zone sizes of the meropenem, imipenem, ertapenem, and doripenem disks were measured between control and inhibitor (EDTA or phenylboronic acid [PBA] with or without cloxacillin)-impregnated Mueller-Hinton agar with a cutoff of 10 mm. All 9 MBL- and 7 KPC-producing Enterobacteriaceae were identified from the differences in zone size in the presence and absence of specific inhibitors, regardless of the carbapenem MICs and including isolates with low-level resistance to carbapenems. We also detected their associated β-lactam resistance mechanisms (11 ESBL-type and 5 class A β-lactamase 2b). No differences in zone size were observed for OXA-48-producing strains or other carbapenem resistance mechanisms such as ESBL and decreased permeability. We propose a new strategy to detect carbapenemases (MBL- and KPC-type) and associated mechanisms of β-lactam resistance (ESBL or class A β-lactamase 2b) by the use of inhibitor-impregnated agar. A rapid phenotypic detection of resistance mechanisms is important for epidemiological purposes and for limiting the spread of resistant strains by implementing specific infection control measures.

INTRODUCTION

Acquired resistance to carbapenems in Gram-negative pathogens such as Enterobacteriaceae is a growing problem worldwide. This resistance is mediated by carbapenemases such as Ambler class B metallo-β-lactamases (MBL), including IMP, VIM, and NDM, as well as by plasmid-mediated clavulanic acid-inhibited class A β-lactamases such as Klebsiella pneumoniae carbapenemase (KPC) and GES and the class D β-lactamase OXA-48 (5, 19). These enzymes are usually encoded by mobile DNA elements with a high capacity for dissemination (16).

Nosocomial outbreaks of carbapenemase-producing Enterobacteriaceae infection have been reported, particularly in Greece (24), and NDM-1 β-lactamase-producing bacteria have been detected in drinking water in New Delhi, India (32).

Dissemination of carbapenem resistance among Enterobacteriaceae poses a considerable threat to public health. In particular, it severely limits treatment options, as carbapenemase-producing strains are resistant not only to carbapenems but also to almost all β-lactam antibiotics except for aztreonam (ATM) for MBL and some other compounds (OXA-48). Moreover, carbapenem resistance in Enterobacteriaceae is often associated with extended-spectrum β-lactamase (ESBL) or with AmpC β-lactamase production and porin loss (15, 33). Infections due to these resistant strains are associated with higher morbidity and mortality rates (6, 27).

Rapid detection of these mechanisms of resistance is crucial for appropriate antimicrobial therapy and infection control measures.

Carbapenemase gene detection by molecular methods is the gold standard but is available in only a few reference laboratories, and phenotypic tests have therefore been developed. The modified Hodge test (MHT) is the only Clinical and Laboratory Standards Institute (CLSI)-recommended carbapenemase-screening method (10). Other tests are based on the synergy between MBL or KPC inhibitors and carbapenems (14, 23, 26). MBL inhibitors used in these methods include the metal-chelating agent EDTA, thiol compounds (mercaptopropionic acid or sodium mercaptoacetic acid), and dipicolinic acid. The different formats include the double-disk synergy test (DDST), the combined disk test, and imipenem (IPM)/imipenem-EDTA Etest strips (bioMérieux, Marcy l'Etoile, France) (1, 17). Boronic acids such as 3-aminophenylboronic acid (PBA) are used for KPC detection in the DDST or the combined disk test (30). However, none of these tests is able to detect resistance mechanisms coexisting with carbapenemase production in clinical isolates.

The aim of this study was to test two different phenotypic methods for the detection of MBL- or KPC-producing Enterobacteriaceae and the associated mechanisms of β-lactam resistance. These tests were evaluated against a panel of 30 genotypically characterized carbapenem-resistant Enterobacteriaceae exhibiting one or more of the following β-lactam resistance mechanisms: carbapenemases, ESBL, plasmid-mediated AmpC β-lactamases, AmpC hyperproduction, and decreased permeability.

MATERIALS AND METHODS

Bacterial strains.

The following isolates were tested: Klebsiella pneumoniae (n = 18), Escherichia coli (n = 6), Enterobacter cloacae (n = 3), Providencia stuartii (n = 1), and Proteus mirabilis (n = 2). The identity of the isolates was confirmed with the API20E method (bioMérieux, Marcy l'Etoile, France).

The following mechanisms of β-lactam resistance were studied: the carbapenemases KPC (n = 7), VIM (n = 6), NDM-1 (n = 3), and OXA-48 (n = 6); decreased permeability associated with ESBL production (n = 3); AmpC hyperproduction (n = 2); and other AmpC β-lactamases (CMY-4, ACC-1, and DHA-1), with or without associated ESBL production (n = 3) (Table 1).

Table 1.

Results of genotypic and phenotypic tests

No. of clinical isolates	Clinical isolate type	Resistance mechanism(s) determined by PCR	MIC (μg/ml) for:				MHT result for b:				Effect of PBA	Effect of EDTA	Effect of cloxacillin	Difference in zone size observed with indicated inhibitor				Detection of associated resistance mechanisma	Synergy detected only with the addition of EDTA or PBA on the 3 disks
			MEM	ETP	IMP	DORI	MEM	ETP	IMP	DORI				ETP	IMP	MEM	DORI
1	E. cloacae	KPC-3, OXA-9, TEM-1	16	32	32	8	+	+	+	+	+	−	−	≥10	≥10	≥10	≥10	BLA	−
4	K. pneumoniae	KPC-2, TEM-1	>32	>32	>32	>32	+	+	+	+	+	−	−	≥10	≥10	≥10	≥10	BLA	−
5	K. pneumoniae	KPC-2, TEM-1	>32	>32	>32	>32	+	+	+	+	+	−	−	≥10	≥10	≥10	≥10	BLA	−
6	K. pneumoniae	KPC-2, TEM-1	>32	>32	6	>32	+	+	+	+	+	−	−	≥10	≥10	≥10	≥10	BLA	−
25	K. pneumoniae	KPC-2, CTX-M-15, TEM-1	>32	>32	>32	>32	+	+	+	+	+	−	−	≥10	≥10	≥10	≥10	ESBL	+
29	K. pneumoniae	KPC-2, SHV-5	2	6	1.5	1	+	+	+	+	+	−	−	≥10	≥10	≥10	≥10	ESBL	+
7	K. pneumoniae	KPC-2, SHV-12	4	16	2	2	+	+	+	+	+	−	−	≥10	≥10	≥10	≥10	ESBL	+
3	E. coli	VIM-1, SHV-5	0.125	0.125	0.75	0.125	+	+	+	+	−	+	−	≥10	≥10	≥10	≥10	ESBL	+
8	K. pneumoniae	VIM-1, CTX-M-14, TEM-1	>32	>32	8	>32	+	+	+	+	−	+	−	≥10	≥10	≥10	≥10	ESBL	+
17	P. mirabilis	VIM-1, SHV-5	1	1	4	1	±	+	+	+	−	+	−	≥10	≥10	≥10	≥10	ESBL	+
19	P. stuartii	VIM-1, SHV-5	>32	32	>32	>32	+	+	+	+	−	+	−	≥10	≥10	≥10	≥10	ESBL	+
10	K. pneumoniae	VIM-4, SHV-5	>32	>32	>32	>32	+	+	+	+	−	+	−	≥10	≥10	≥10	≥10	ESBL	+
11	K. pneumoniae	VIM-4, CMY-4, CTX-M-15, TEM-1	2	4	4	2	±	+	+	+	−	+	−	≥10	≥10	≥10	≥10	ESBL	+
2	E. coli	NDM-1, CTX-M-15, TEM-1	8	>32	32	8	+	+	+	+	−	+	−	≥10	≥10	≥10	≥10	ESBL	+
9	E. coli	NDM-1, OXA-1	8	>32	16	4	−	−	±	−	−	+	−	≥10	≥10	≥10	≥10	BLA	−
18	P. mirabilis	NDM-1, VEB-6	>32	>32	>32	>32	+	+	+	+	−	+	−	≥10	≥10	≥10	≥10	ESBL	+
15	K. pneumoniae	OXA-48	0.25	0.38	0.25	0.125	+	+	+	+	−	−	−	<5	<5	<5	<5	−	−
28	E. coli	OXA-48	0.38	4	0.38	0.125	+	+	+	+	−	−	−	<5	<5	<5	<5	−	−
27	E. coli	OXA-48, CTX-M-9	6	>32	>32	4	+	+	+	+	−	−	−	<5	<5	<5	<5	−	−
16	K. pneumoniae	OXA-48, TEM-1, OXA-1	>32	>32	>32	>32	+	+	+	+	−	−	−	<5	<5	<5	<5	−	−
30	E. coli	OXA-48, CTX-M-14, TEM-1	0.25	2	0.5	0.38	+	+	+	+	−	−	−	<5	<5	<5	<5	−	−
14	K. pneumoniae	OXA-48, CTX-M-15	0.75	3	0.75	0.5	+	+	+	+	−	−	−	<5	<5	<5	<5	−	−
22	E. cloacae	Hyperproduced AmpC, decreased permeability	0.38	4	1.5	0.125	±	−	±	−	−	−	+	<5	<5	<5	<5	−	−
23	E. cloacae	Hyperproduced AmpC, decreased permeability	0.5	4	1.5	0.5	−	±	±	−	+	−	+	≥10	≥10	≥10	≥10	−	−
12	K. pneumoniae	CMY-4, SHV-5, TEM-1, decreased permeability	16	>32	16	4	−	−	−	−	+	−	+	≥10	≥10	≥10	≥10	ESBL	−
13	K. pneumoniae	ACC-1, TEM-1, decreased permeability	2	16	8	2	−	−	−	−	+	−	+	<5	<5	<5	7	BLA	−
24	K. pneumoniae	CTX-M-15, DHA-1, decreased permeability	4	>32	16	2	−	−	−	−	+	−	+	≥10	≥10	≥10	≥10	ESBL	−
20	K. pneumoniae	CTX-M-15, OXA-1, TEM-1	8	>32	3	3	−	−	−	−	−	−	−	<5	<5	<5	<5	−	−
21	K. pneumoniae	CTX-M-15, OXA-1, TEM-1	12	>32	3	4	−	−	−	−	−	−	−	<5	<5	<5	<5	−	−
26	K. pneumoniae	CTX-M-15, OXA-1	3	8	1	1.5	−	−	−	−	−	−	−	<5	<5	<5	<5	−	−

^aBLA, narrow-spectrum β-lactamases (e.g., TEM-1, SHV-1, OXA-1); ESBL, extended-spectrum β-lactamases (e.g., CTX-M-15, SHV-12, etc.).

^b±, doutful.

All the isolates have previously been characterized or are as recently described (13, 23).

Antimicrobial susceptibility.

Antimicrobial susceptibility was tested according to CLSI recommendations. The diffusion method on Mueller-Hinton agar (MHA; Bio-Rad, Marnes-La-Coquette, France) was used to test susceptibility to amoxicillin (AMX), ticarcillin (TIC), cefepime (FEP), ticarcillin-clavulanic acid (TCC), cephalotin (CF), ceftazidime (CAZ), amoxicillin-clavulanic acid (AMC), cefotaxime (CTX), ertapenem (ETP), doripenem (DORI), aztreonam (ATM), cefoxitin (FOX), meropenem (MEM), moxalactam (MOX), and imipenem (IPM).

Carbapenem MICs (IPM, MEM, ETP, and DORI) were determined with the Etest (bioMérieux, Marcy l'Etoile, France) and interpreted according to CLSI guidelines (11).

MHT.

The modified Hodge test (MHT) was performed as recommended by CLSI. Four disks were used, with IPM (10 μg), MEM (10 μg), ETP (10 μg), and DORI (10 μg) placed on the same agar plate seeded with E. coli ATCC 25922.

Phenotypic test 1.

For the first phenotypic test, with carbapenemase inhibitor-impregnated agar, the β-lactams were tested on Mueller-Hinton agar (MHA; Bio-Rad) impregnated with β-lactamase inhibitors (EDTA and phenylboronic acid [PBA]) and on a commercial MHA containing cloxacillin (250 μg/ml) (AES Chemunex, Combourg, France). EDTA-impregnated agar was prepared by spreading 2 ml of 5 mM EDTA solution (Sigma-Aldrich, St. Louis, MO), and PBA-impregnated agar was prepared by spreading 750 μl of PBA (Sigma-Aldrich) at 10 mg/ml (60 mg diluted in 3 ml of dimethyl sulfoxide [DMSO] and then in 3 ml of sterile distilled water) (12).

The possible antimicrobial activity of DMSO was tested on DMSO-impregnated MHA against 20 control wild-type strains of Enterobacteriaceae and all the test strains.

Differences in inhibition zone sizes between control and EDTA-, PBA-, or cloxacillin-impregnated agar were recorded.

Phenotypic test 2.

The second phenotypic test was a modified version of the CLSI confirmatory test for ESBL production. It was evaluated for its ability to detect the association of ESBL with MBL or KPC production (31). Briefly, ATM (10 μg), AMC (10 μg), and CAZ (10 μg) disks were placed in two identical lines 20 mm apart on an MHA plate seeded with the test strain, and 5 μl of 0.5 M EDTA was added to the second line. Marked enhancement of the inhibition zone between the ATM and AMC disks and between the AMC and CAZ disks in the presence of EDTA was considered to indicate ESBL production. This experiment was then repeated with 3 lines of the same three disks, with 10 μl and 20 μl, respectively, of 20-mg/ml phenylboronic acid solution being added to the second and third lines, with the first line serving as a control.

RESULTS

Antimicrobial susceptibility.

Twenty of 22 carbapenemase-producing strains were resistant to ertapenem, 17/22 to imipenem (2 strains were intermediate using CLSI breakpoints), 16/22 to meropenem (2 strains were intermediate), and 15/22 to doripenem (2 strains were intermediate) (Table 1).

The carbapenem MICs ranged from 0.125 μg/ml to >32 μg/ml, depending on the drug. Overall, ertapenem had higher MICs than MEM, IPM, and DORI (Table 1).

Carbapenem MICs ranged from 1 μg/ml to >32 μg/ml for the 7 KPC-producing strains. Five of these strains had MICs of ≥8 μg/ml.

Carbapenem MICs for VIM-producing Enterobacteriaceae ranged from 0.125 μg/ml to >32 μg/ml. The lowest MICs were observed with the E. coli (VIM-1 and SHV-5) and P. mirabilis (VIM-1 and SHV-5) strains.

High MICs (≥4 μg/ml) were obtained with the 3 NDM-1 producers, particularly in the case of ETP. MICs for the OXA-48-producing strains were low, particularly for the 2 isolates with no associated β-lactamase production. Only ertapenem allowed detection of a decrease in susceptibility to carbapenem for 4/6 strains producing OXA-48.

According to their MICs, strains producing AmpC β-lactamases or ESBL associated with decreased permeability were all resistant to ertapenem (n = 8). Two isolates were susceptible to meropenem and doripenem, and one isolate was susceptible to imipenem. The other 5 isolates were intermediate or resistant to these carbapenems. Thus, the efficacy of ertapenem was mainly impaired by decreased permeability.

MHT.

The modified Hodge tests using MEM, ETP, IPM, and DORI each successfully identified 21/22 (95%) of the carbapenemase producers. The unidentified strain, an NDM-1 producer, nonetheless exhibited a disturbed edge of the inhibition zone. The inhibition pattern could also be interpreted as weakly but falsely positive for 2 E. cloacae strains that hyperproduced AmpC and exhibited decreased permeability. Despite their high carbapenem MICs, we obtained no positive results with ESBL-producing strains, contrary to other authors (7, 22).

Phenotypic test 1. (i) Carbapenemase detection.

All visible differences in the inhibition zone sizes of the MEM, ETP, IPM, and DORI disks between control and inhibitor-impregnated MHA measured at least 10 mm, and we therefore chose a cutoff of 10 mm. DMSO had no effect on the inhibition zone size of any of the antibiotics tested.

All 9 MBL-producing Enterobacteriaceae were identified from the differences in zone size in the presence and absence of EDTA, regardless of the carbapenem MICs (Fig. 1A and B); this included the VIM-1- and SHV-5-producing E. coli isolate with low-level resistance to all carbapenems.

Fig 1.

Representative phenotypes of resistance. (Aa to Ad) The VIM-4- and SHV-5-producing K. pneumoniae strain (strain 10) is shown in a, b, c, and d. (a) Antimicrobial susceptibility testing was performed on Mueller-Hinton agar (MHA) as a reference. (b) Antimicrobial susceptibility testing was performed on Mueller-Hinton agar impregnated with 2 ml of EDTA 5 × 10⁻³ M. The difference in zone size in the presence and absence of EDTA was >10 mm for ETP, MEM, DORI, and IPM, suggesting MBL production. Synergy between amoxicillin-clavulanic acid or ticarcillin-clavulanic acid and broad-spectrum cephalosporins (FEP and CTX) suggests ESBL production. (c) Antimicrobial susceptibility testing was performed on Mueller-Hinton agar impregnated with 0.75 ml of 10 mg/ml PBA. No difference in zone size was observed in the presence and absence of PBA. (d) Antimicrobial susceptibility testing was performed on MHA-cloxacillin (250 μg/ml) (AES Chemunex). No differences in zone size were observed. (Be to Bh) The NDM-1- and OXA-1-producing E. coli strain (strain 9) is shown in e, f, g, and h. (e) Antimicrobial susceptibility testing was performed on Mueller-Hinton agar as a reference. (f) Antimicrobial susceptibility testing was performed on Mueller-Hinton agar impregnated with 2 ml of EDTA 5 × 10⁻³ M. The difference in zone size was >10 mm for ETP, MEM, DORI, and IPM, suggesting MBL production. (g) Antimicrobial susceptibility testing was performed on Mueller-Hinton agar impregnated with 0.75 ml of 10 mg/ml PBA. No difference in zone size was observed. (h) Antimicrobial susceptibility testing was performed on MHA-cloxacillin (250 μg/ml) (AES Chemunex). No differences in zone size were observed. (Ci to Cl) The KPC-2-, CTX-M-15-, and TEM-1-producing K. pneumoniae strain (strain 25) is shown in i, j, k, and l. (i) Antimicrobial susceptibility testing was performed on Mueller-Hinton agar as a reference. (j) Antimicrobial susceptibility testing was performed on Mueller-Hinton agar impregnated with 2 ml of EDTA 5 × 10⁻³ M. No significant difference in zone size was observed with ETP, MEM, DORI, or IPM in the presence and absence of EDTA, indicating no MBL production. (k) Antimicrobial susceptibility testing was performed on Mueller-Hinton agar impregnated with 0.75 ml of 10 mg/ml PBA. The difference in zone size was >10 mm for ETP, MEM, DORI, and IPM, suggesting class A carbapenemase production or a combination of plasmid-mediated AmpC production and porin loss. Synergy between amoxicillin-clavulanic acid or ticarcillin-clavulanic acid and broad-spectrum cephalosporins (FEP, CTX, and CAZ) suggested ESBL production. (l) Antimicrobial susceptibility testing was performed on MHA-cloxacillin (250 μg/ml) (AES Chemunex). No differences in zone size were observed. One can conclude that the E. coli strain produces both a class A carbapenemase and an ESBL. (Dm to Dp) The E. cloacae strain exhibiting AmpC hyperproduction and decreased permeability (strain 23) is shown in m, n, o, and p. (m) Antimicrobial susceptibility testing was performed on Mueller-Hinton agar as a reference. (n) Antimicrobial susceptibility testing was performed on Mueller-Hinton agar impregnated with 2 ml of EDTA 5 × 10⁻³ M. No significant difference in zone size was observed with ETP, MEM, DORI, or IPM, indicating the absence of MBL. (o) Antimicrobial susceptibility testing was performed on Mueller-Hinton agar impregnated with 0.75 ml of 10 mg/ml PBA. The difference in zone size was >10 mm for ETP, MEM, DORI, and IPM, suggesting the production of a class A carbapenemase or a combination of AmpC hyperproduction and decreased permeability. (p) Antimicrobial susceptibility testing was performed on MHA-cloxacillin (250 μg/ml). Differences in the zone diameter were observed between native MHA and MHA-cloxacillin. Given the similar inhibition zones obtained with cloxacillin and PBA, carbapenemase production can be ruled out. (Eq to Et) The OXA-48- and CTX-M-9-producing E. coli strain (strain 27) is shown in q, r, s, and t. No significant difference in zone size was observed with ETP, MEM, DORI, or IPM in the presence and absence of EDTA, PBA, or cloxacillin. (Fu to Fx) The K. pneumoniae strain exhibiting CTX-M-15, OXA-1, TEM-1, and decreased permeability (strain 20) is shown in u, v, w, and x. No significant difference in zone size was observed with ETP, MEM, DORI, or IPM in the presence and absence of EDTA, PBA, or cloxacillin.

Using PBA as the inhibitor, we also identified all 7 carbapenemase (KPC)-producing strains, again regardless of the carbapenem MICs. Positive results were also obtained with AmpC hyperproducers, and DHA-1 or CMY-4 also exhibited decreased permeability (3/6 strains had a difference in zone size of ≥10 mm in the presence and absence of PBA).

Using cloxacillin as an inhibitor allowed the differentiation of KPC-producing strains and AmpC β-lactamase-producing strains. Indeed, if the increases in the inhibition zone size were similar in both compounds, the presence of a KPC could be ruled out (Fig. 1C and D).

With the six OXA-48-producing strains, the differences in inhibition zone size were always <10 mm with inhibitors. Neither EDTA nor PBA nor cloxacillin inhibited OXA-48 (Fig. 1E), and no significant differences were observed in their presence or absence with the 4 ESBL-producing strains that also exhibited decreased permeability (CTX-M type) (Fig. 1F).

(ii) Detection of associated resistance mechanisms.

By inhibiting carbapenemases with specific inhibitors, we were able to detect the associated resistance mechanisms in 7 of the 7 KPC-producing strains, as follows: 3 ESBL-type (molecular characterization revealed that these ESBLs were SHV-5, SHV-12, and CTX-M-15) and 4 class A β-lactamase 2b. We also detected the associated resistance mechanisms in the 9 MBL-producing strains, as follows: 8 ESBL-type (molecular characterization revealed that these ESBLs were 4 SHV-5, 2 CTX-M-15, 1 CTX-M-14, and 1 VEB-6) and one class A β-lactamase 2b (Fig. 1A to C).

Phenotypic test 2.

In the second phenotypic test, the 8 strains producing both MBL and ESBL yielded synergy patterns (increased inhibition zones) between ATM and AMC and between AMC and CAZ in the presence of EDTA.

Among the 7 KPC-producing strains, synergy was observed with the 3 strains producing both KPC and ESBL. The synergy was more marked with 20 μl than with 10 μl of 20-mg/ml PBA solution (Fig. 2).

Fig 2.

(A) The K. pneumoniae strain producing VIM-4 and SHV-5 (strain 10) is shown. Two lines of disks containing ATM (10 μg), AMC (10 μg), and CAZ (10 μg) were placed 20 mm apart on an MHA plate seeded with the test strain. Then, 5 μl of 0.5 M EDTA was added along the second line. Enhancement of the zone of inhibition in the area between the ATM and AMC disks and between the AMC and CAZ disks in the presence of EDTA is considered to suggest ESBL production in addition to MBL. (B) The K. pneumoniae strain producing KPC-2, CTX-M-15, and TEM-1 (strain 25) is shown. Three lines of disks containing ATM (10 μg), AMC (10 μg), and CAZ (10 μg) were placed 20 mm apart on an MHA plate seeded with the test strain. Then, 10 μl and 20 μl of phenylboronic acid (20 mg/ml) were added to lines 2 and 3, respectively. Enhancement of the zone of inhibition between the ATM and AMC disks and between the AMC and CAZ disks in the presence of PBA suggests ESBL production in addition to KPC or AmpC expression and decreased permeability. Cloxacillin-impregnated agar is necessary to draw firm conclusions, as it is otherwise impossible to tell whether the enhancement of the inhibition zone is due to the effect of boronic acid on an AmpC β-lactamase or on a KPC carbapenemase (PBA can inhibit both enzymes).

We detected no further synergy in the presence or absence of EDTA or PBA in the case of OXA-48-producing strains or ESBL or AmpC producers that also exhibited decreased permeability.

DISCUSSION

Phenotypic detection of carbapenemase-producing organisms was originally based on tests showing reduced carbapenem susceptibility. Carbapenem MICs for carbapenemase producers are highly variable. Ertapenem has been described as the most appropriate carbapenem for detecting KPC producers with low-level resistance to carbapenems (8, 29). According to our results, it also seems to be the most appropriate carbapenem for detecting other carbapenemases (21).

Low levels of carbapenem resistance have often been observed in Enterobacteriaceae producing carbapenemases of various classes (18, 25, 29), so carbapenemase detection can be problematic, owing to factors such as low carbapenem MICs (20) and an inoculum effect (4). Indeed, the new CLSI breakpoints updated in June 2010 (9) do not always allow carbapenemase producers with low carbapenem MICs to be detected. This variability means that no single carbapenem-screening criterion can be used to identify all isolates. Therefore, confirmatory testing is essential for the detection of carbapenemase-producing Enterobacteriaceae. As these genes are usually carried by mobile genetic elements (16) with a high capacity for horizontal dissemination, these resistant strains could be responsible for epidemics. Indeed, all enterobacterial isolates with low-level resistance to carbapenems should be screened for carbapenemase production, especially as the level of carbapenemase production at sites of infection is not known (21).

Our results with the modified Hodge test are largely in keeping with previous reports. Indeed, the MHT is reported to have high sensitivity (95 to 100%) (20) and to be suitable for confirming carbapenemase production. But its interpretation can be difficult, and false-positive results can be observed with strains producing ESBLs or AmpC with decreased porins (7, 22). We found that the MHT efficiently identified carbapenemase producers and that the 4 substrates (MEM, DORI, ETP, IPM) gave similar results. However, weakly positive and false-negative results have been reported for Enterobacteriaceae producing MBLs, and especially NDM-1, as observed in this study (3, 26).

We used carbapenemase inhibitor-impregnated agar to test for carbapenem-resistant strains. Differences in the inhibition zone sizes of the MEM, ETP, IPM, and DORI disks were measured between control and inhibitor (EDTA or PBA)-impregnated MHA (with a cutoff of 10 mm). The MBL-producing Enterobacteriaceae were well identified with this method. Results were easily interpreted regardless of the carbapenem MICs and the results of the modified Hodge test.

This phenotypic test seems to be sensitive for strains producing KPC-type enzymes, although we obtained false-positive results with AmpC hyperproducers, and DHA-1 or CMY-4 also exhibited decreased permeability. As PBA can also inhibit class C β-lactamases (2), we recommend testing these strains in parallel on both PBA agar and cloxacillin agar. If the increases in the inhibition zones are similar with the two agents, the presence of a KPC can be ruled out (Fig. 1D). In contrast, with KPC-producing strains, we observed no difference in the inhibition zones in the presence and absence of cloxacillin (Fig. 1C) and therefore inferred that the β-lactamase was a class A carbapenemase. When a difference in inhibition zone diameter is observed in the presence and absence of PBA but not in the presence and absence of cloxacillin, one can conclude that the strain expresses a carbapenemase of the KPC type. However, it would be interesting to test a strain producing KPC and AmpC (plasmid-located or derepressed ampC) with the method described in this study.

OXA-48-producing strains can exhibit low-level carbapenem resistance, as can ESBL-producing strains with decreased permeability. There is currently no phenotypic test capable of detecting OXA-48. Our test is not able to distinguish between strains expressing OXA-48-type carabapenemases and strains that are resistant to carbapenems due to ESBL production associated with decreased permeability.

The associated resistance mechanisms were easily detected with carbapenemase inhibitor-impregnated agar for MBL- and KPC-producing strains (Fig. 1A to C). Regarding the NDM-1-producing E. coli isolate (with no associated ESBL) for which the modified Hodge test was negative, our approach detected the carbapenemase and demonstrated the lack of ESBL (Fig. 1B) but the presence of a class A β-lactamase 2b. This is not directly relevant for therapeutic decisions, but aztreonam susceptibility is an important point, as this molecule may be an alternative treatment for infections due to both MBL-producing and non-ESBL-producing Enterobacteriaceae. Indeed, in studies of carbapenems and aztreonam in animal models of infection due to an MBL-producing (ESBL-negative) Enterobacteriaceae clinical isolate (6), treatment with aztreonam significantly reduced mortality (28).

The results of the second phenotypic test (Fig. 2) were in keeping with those obtained with carbapenemase inhibitor-impregnated agar and with genetic findings. This test was effective for the detection of ESBL associated with carbapenemase but not carbapenemase itself. Therefore, a phenotypic or genotypic detection of carbapenemases must be combined with this test. However, the production of ESBL in MBL-producing Enterobacteriaceae was easily observed, as the increased inhibition zone size among antibiotic disks was consistent. It was less straightforward for KPC- and ESBL-producing Enterobacteriaceae, but results were interpretable.

Finally, the phenotypic test 1 seems to be efficient for the detection of carbapenemases and associated β-lactam resistance mechanisms. However, as in all phenotypic tests, interpretation is subjective and requires some experience, but it has the advantage of being easy to carry out compared to other tests. In addition, breakpoints of 10 mm, instead of the usual breakpoints of 5 mm, seem to facilitate interpretation.

As there are no marketed carbapenem inhibitor-impregnated agars, the preparation of these plates is a manual technique which, although easy to perform, can be time-consuming in a routine laboratory.

In conclusion, the concomitant presence of several different types of β-lactamases in resistant Enterobacteriaceae makes it more difficult to detect individual mechanisms, as one mechanism can mask another. Phenotypic detection (in 2 days) of combined mechanisms of resistance, such as ESBL expression in KPC- or MBL-expressing isolates, is important for epidemiological purposes and for implementing rapid and specific infection control measures.

We propose a new strategy to detect carbapenemase resistance and associated mechanisms of β-lactam resistance by the use of specific inhibitors. Carbapenemase inhibitor-impregnated agar gave the best results and could be suitable for routine use. However, only a relatively small number of characterized isolates were examined in this study, hence the need to validate these results on more strains. It will be interesting to test this approach in a routine laboratory in which carbapenemases are frequently detected.

A strategy for identifying the type of carbapenemase is proposed in Fig. 3. This is a satisfactory and inexpensive method for characterizing the type of carbapenemase and for detecting associated resistance mechanisms in laboratories when PCR is not readily available. Although we tested a large number of antibiotics, our results suggest that placing 7 antibiotic disks (ticarcillin, aztreonam, cefepime, cefotaxime, ertapenem, imipenem, and amoxicillin-clavulanic acid in the middle) on a round MHA plate is sufficient for the detection of carbapenemases (MBL- and KPC-type) and associated mechanisms of β-lactam resistance in Enterobacteriaceae.

Fig 3.

Strategy for identifying the type of carbapenemase.