Evaluation of Five Chromogenic Agar Media and the Rosco Rapid Carb Screen Kit for Detection and Confirmation of Carbapenemase Production in Gram-Negative Bacilli

Abstract

An efficient workflow to screen for and confirm the presence of carbapenemase-producing Gram-negative bacilli was developed by evaluating five chromogenic screening agar media and two confirmatory assays, the Rapid Carb screen test (Rosco Diagnostica A/S, Taastrup, Denmark) and the modified Hodge test. A panel of 150 isolates was used, including 49 carbapenemase-producing isolates representing a variety of β-lactamase enzyme classes. An evaluation of analytical performance, assay cost, and turnaround time indicated that the preferred workflow (screening test followed by confirmatory testing) was the chromID Carba agar medium (bioMérieux, Marcy l'Étoile, France), followed by the Rapid Carb screen test, yielding a combined sensitivity of 89.8% and a specificity of 100%. As an optional component of the workflow, a determination of carbapenemase gene class via molecular means could be performed subsequent to confirmatory testing.

INTRODUCTION

The worldwide dissemination of carbapenemase-producing Gram-negative bacilli (CPGNB) is a significant clinical and public health concern (1, 2). The rapid detection of these antibiotic-resistant pathogens by the clinical microbiology laboratory is of utmost importance in controlling nosocomial spread and for initiating appropriate antimicrobial therapy. Chromogenic media containing a carbapenem are convenient tools for the screening and rapid detection of carbapenem-resistant Gram-negative bacilli (GNB). However, growth on a chromogenic medium simply signifies carbapenem resistance (e.g., an AmpC producer with porin loss may grow) and does not confirm the production of a carbapenemase. Recently, a novel rapid chromogenic test based on the hydrolysis of imipenem, the Carba NP test, and a commercial version of the Carba NP test, the Rapid Carb screen (RCS) test, have been described (3–8). The purpose of the current study was to evaluate screening and confirmation methods that when paired would provide a streamlined workflow for the detection of CPGNB in the clinical microbiology laboratory. First, five different chromogenic media were evaluated to select a sensitive medium type for screening for CPGNB. Second, the RCS test (Rosco Diagnostica A/S, Taastrup, Denmark) was evaluated to confirm carbapenemase production through a method comparison with the current gold standard, the modified Hodge (MH) test (9). A combined workflow approach was designed to achieve the maximum sensitivity and specificity for detecting CPGNB, with a consideration for the cost and turnaround time of testing. Last, upon finding a CPGNB, it may be of value in some circumstances to determine the carbapenemase gene class, and therefore, an established multiplex PCR was also incorporated as an optional endpoint to this workflow.

MATERIALS AND METHODS

Bacterial isolates.

A total of 150 isolates were tested, including an assortment of 49 isolates of GNB harboring a diverse set of carbapenemase enzyme classes and 101 carbapenemase-negative isolates. The 49 carbapenemase producers, a combination of clinical and ATCC strains, were previously molecularly characterized for β-lactamase genes and included 16 K. pneumoniae carbapenemase (KPC) producers (bla_KPC-2 [n = 9], bla_KPC-3 [n = 5], bla_KPC-11 [n = 1], bla_KPC-12 [n = 1])—Citrobacter freundii, n = 2; Citrobacter koseri, n = 1; Enterobacter cloacae, n = 1; Escherichia coli, n = 4; Klebsiella oxytoca, n = 1; Klebsiella pneumoniae, n = 5; Pseudomonas aeruginosa, n = 1; and Serratia marcescens, n = 1; 12 VIM producers (bla_VIM-1 [n = 6], bla_VIM-2 [n = 3], bla_VIM-4 [n = 2], bla_VIM-5 [n = 1])—C. freundii, n = 2; Enterobacter aerogenes, n = 1; E. cloacae, n = 1; K. oxytoca, n = 1; Proteus mirabilis, n = 1; Providencia stuartii, n = 1; and P. aeruginosa, n = 5; 4 IMP producers (bla_IMP-1 [n = 3] and bla_IMP-26) [n = 1]—E. coli, n = 1; K. pneumoniae, n = 2; and P. aeruginosa, n = 1; 9 NDM producers (bla_NDM-1 [n = 6], bla_NDM-4 [n = 1], bla_NDM-5 [n = 1], bla_NDM-7 [n = 1])—Acinetobacter baumannii, n = 1; E. coli, n = 4; K. oxytoca, n = 1; K. pneumoniae, n = 1; P. mirabilis, n = 1; and Providencia rettgeri, n = 1; 8 OXA producers (bla_OXA-24 [n = 1], bla_OXA-48 [n = 7])—A. baumannii, n = 1; C. koseri, n = 1; E. cloacae, n = 1; K. pneumoniae, n = 3; Morganella morganii, n = 1 and S. marcescens, n = 1; and one SME producer—S. marcescens, n = 1). Of note, one E. cloacae isolate possessed both VIM-4 and OXA-48. The 101 carbapenemase-negative isolates included extended-spectrum β-lactamase (ESBL) and AmpC producers (previously molecularly characterized) with and without porin mutations, clinical isolates with elevated carbapenem MICs (and negative for carbapenemases by PCR), and carbapenem-susceptible isolates. The isolates were blinded, and 75 isolates were distributed to each of the two clinical microbiology laboratory testing sites (Health Sciences Centre and St. Boniface Hospital, Winnipeg, Canada).

Chromogenic media to screen for CPGNB.

The first part of this study involved evaluating five different chromogenic medium types used to screen for CPGNB. The five chromogenic media evaluated were (i) Oxoid Brilliance ESBL (Besington, Hants, United Kingdom), (ii) Oxoid Brilliance CRE, (iii) bioMérieux chromID Carba (Marcy l'Étoile, France), (iv) CHROMagar Colorex C3Gr (Paris, France), and (v) CHROMagar Colorex KPC. The Brilliance ESBL and Colorex C3Gr media are designed to detect ESBLs, and in this study, they were evaluated both with and without an ertapenem disk (10 μg) to select for carbapenem-resistant isolates. The inocula for each of the chromogenic medium types and a nonselective blood agar (BA) plate were 10 μl (∼10⁶ CFU/ml) of a 0.5 McFarland standard prepared in sterile saline (0.85%). Brilliance ESBL, Brilliance CRE, and chromID Carba were streaked using the Isoplater (Vista Technology, Inc., Edmonton, Alberta, Canada) to mimic standard laboratory practice. The Colorex C3Gr and Colorex KPC media were evaluated as a biplate, so the media were streaked manually. Ertapenem disks were placed between the 2nd and 3rd quadrants of the Brilliance ESBL and Colorex C3Gr media following the streaking of the plate. The plates were incubated for 24 h at 37°C, and the presence/absence and color of growth were interpreted according to the package inserts. Any isolate on the ESBL chromogenic media with an ertapenem zone of inhibition of ≤27 mm was considered a putative carbapenemase producer and was considered for further evaluation using a confirmatory method. The use of the ≤27-mm cutoff was established in a previous study using an ertapenem disk (10 μg) to screen for carbapenem-resistant GNB on MacConkey medium (10).

Phenotypic confirmatory assays.

In the second part of the study, a method comparison between the MH test and the RCS test was performed on all study isolates using growth from the BA plate inoculated concurrently with the chromogenic media. The BA plate was added to perform the confirmatory tests the following day, since the RCS test cannot be performed from the chromogenic media, as the color of the colonies interferes with the interpretation of the test. The MH test using meropenem disks (10 μg) was performed and interpreted according to CLSI methodology (9). The RCS is a commercially available rapid chromogenic test for the detection of carbapenemases based on the hydrolysis of imipenem in the presence of an indicator (phenol red [5]). All the components of the assay are present within tablets called diatabs. Two sets of diatabs are contained in the kit, one containing imipenem and one without. The diatabs are dissolved in saline, and the lysate of the organism is added to a tube with a dissolved diatab with imipenem and a tube with a dissolved diatab without imipenem (negative-control tube; it should remain red in color). Initially, the RCS test was performed as previously described (4). K. pneumoniae strain ATCC BAA1706 and K. pneumoniae strain ATCC BAA1705 were the positive and negative controls, respectively. Any color change from red to yellow in the 1.5-ml test tube containing imipenem was considered a positive reaction for the RCS. Uninterpretable results were isolates that gave a slight positive reaction (peach color) in both the tubes with and without imipenem.

As there were many uninterpretable results initially using the previously described RCS protocol, troubleshooting was performed in the RCS protocol; the appropriate inoculum was determined by testing one to three calibrated loops of organism and finding the optimal tube for performing the assay by testing 1.5-ml test tubes, cryovials, and round-bottom 5-ml tubes. Thus, the modified RCS test was performed upon repeat on all isolates to include only one 10-μl calibrated loopful of organism (reduced from two) and a round-bottom 5-ml tube (instead of a conical 1.5-ml tube). As there was variability in the degree of color change for the positives with the modified RCS protocol, positives were considered those displaying any color change from that of the negative-control tubes (tubes without imipenem). The isolates that were orange by the modified RCS protocol in this study were repeated using both the RCS test and the Carba NP test, as previously described, to assess the ease of interpretation of the intermediate results (3, 5, 7).

Conventional multiplex PCR.

Last, a conventional multiplex PCR for bla_KPC, bla_NDM, bla_VIM, bla_IMP, and bla_OXA-48 and bla_OXA-181 was performed on all isolates, as previously described (11). In addition, a bla_SME PCR was performed on one isolate to confirm SME production (12).

Determination of an optimal workflow for the screening and confirmation of CPGNB.

The results of the screening and confirmatory assays (performed on all test isolates) were analyzed to develop the optimal workflow (i.e., test choice) for the detection and confirmation of CPGNB. By combining the individual data from the chromogenic media and the confirmatory assays, we were able to determine the sensitivities and specificities of the combined assays. In addition, the cost and time to perform the different combination of assays were evaluated and considered for feasibility and full optimization of the workflow within a clinical microbiology laboratory setting.

RESULTS

Chromogenic media for CPGNB.

The first part of this study involved evaluating five different chromogenic medium types used to screen for CPGNB, including two media targeting ESBLs supplemented with and without an ertapenem disk (Table 1). Of the five different chromogenic media evaluated to detect CPGNB (Table 1), Colorex C3Gr without the ertapenem disk had the highest sensitivity, at 95.9%. However, the two ESBL medium types, Colorex C3Gr and Brilliance ESBL, without the use of the ertapenem disk both had a specificity of 38.6% for CPGNB. The addition of the ertapenem disk to Colorex C3Gr and Brilliance ESBL resulted in increased specificities of 62.4% and 94.1%, respectively; however, it resulted in a loss of sensitivity, at 89.8 and 42.9%, respectively.

TABLE 1.

Performances of various chromogenic media, multiple confirmation assays, and algorithms for the detection and confirmation of carbapenemase-producing GNB^a

Characteristic measuredb	No. (%) of enzymes detected in:											Performance for CPGNB detection (%)d		No. of confirmatory tests	Cost (Can$)e	TAT (h)f
	Carbapenemase-producing GNB (total no.)c							Carbapenemase-negative GNB (total no.)
	KPC (16)	VIM (12)	IMP (4)	NDM (9)	OXA (8)	SME (1)	Total (49)	ESBL (12)	AmpC (45)	Carb Sj (43)	Total (101)	Sensitivity	Specificity
Chromogenic media used for detection
Brilliance ESBL	14 (87.5)	11 (91.6)	4 (100)	9 (100)	4 (50.0)	0	41 (83.7)	12 (100)	32 (71.1)	18 (41.9)	62 (61.4)	83.7	38.6	NAg	5.42	18–24
Brilliance ESBL with ERT	7 (46.7)	4 (33.3)	1 (25.0)	6 (66.7)	4 (50.0)	0	21 (42.9)	2 (16.7)	2 (4.4)	2 (4.7)	6 (5.9)	42.9	94.1	NA	5.62	18–24
Brilliance CRE	15 (93.8)	8 (66.7)	3 (75.0)	9 (100)	4 (50.0)	0	38 (77.6)	3 (25.0)	9 (20.0)	1 (2.3)	13 (12.9)	77.6	87.1	NA	7.56	18–24
chromID Carba	16 (100)	10 (83.3)	4 (100)	9 (100)	5 (62.5)	1 (100)	44 (89.8)	2 (16.7)	3 (6.7)	0	5 (5.0)	89.8	95.0	NA	3.25	18–24
Colorex C3Gr	16 (100)	12 (100)	4 (100)	9 (100)	7 (87.5)	0	47 (95.9)	12 (100)	41 (91.1)	9 (20.9)	62 (61.4)	95.9	38.6	NA	2.79	18–24
Colorex C3Gr with ERT	16 (100)	10 (83.3)	4 (100)	8 (88.9)	6 (75.0)	1 (100)	44 (89.8)	2 (16.7)	34 (75.6)	2 (4.7)	38 (37.6)	89.8	62.4	NA	2.99	18–24
Colorex KPC	15 (93.8)	10 (83.3)	4 (100)	8 (88.9)	5 (62.5)	0	41 (83.7)	2 (16.7)	5 (11.1)	1 (2.3)	8 (7.9)	83.7	92.1	NA	2.79	18–24
Confirmation assays
Modified Hodge	14 (87.5)	7 (58.3)	3 (75.0)	6 (66.7)	8 (100)	0	37 (75.5)	0	7 (15.6)	0	7 (6.9)	75.5	93.1	NA	0.27	24
Rapid Carb screen	16 (100)	10 (83.3)	4 (100)	9 (100)	7 (87.5)	1 (100)	48 (98.0)	0	0	0	0	98.0	100	NA	4.57	2
Multiplex conventional PCR	16 (100)	12 (100)	4 (100)	9 (100)	7 (87.5)h	0i	47 (95.9)	0	0	0	0	95.9	100	NA	10.95	5
Algorithms for the detection and confirmation of carbapenemase producers
Brilliance ESBL and RCS	14 (87.5)	11 (91.6)	4 (100)	9 (100)	4 (50.0)	0	41 (83.7)	0	0	0	0	83.7	100	103	9.99	50
Brilliance CRE and RCS	15 (93.8)	8 (66.7)	3 (75.0)	9 (100)	4 (50.0)	0	38 (77.6)	0	0	0	0	77.6	100	51	12.13	50
chromID and RCS	16 (100)	10 (83.3)	4 (100)	9 (100)	4 (50.0)	1 (100)	44 (89.8)	0	0	0	0	89.8	100	49	7.82	50
Colorex C3Gr and RCS	16 (100)	12 (100)	4 (100)	9 (100)	7 (87.5)	0	47 (95.9)	0	0	0	0	95.9	100	109	7.36	50
Colorex KPC and RCS	15 (93.8)	10 (83.3)	4 (100)	8 (88.9)	5 (62.5)	0	41 (83.7)	0	0	0	0	83.7	100	49	7.36	50
Colorex C3Gr and multiplex PCR	16 (100)	12 (100)	4 (100)	9 (100)	6 (75.0)	0	45 (91.8)	0	0	0	0	91.7	100	103	13.74	53
chromID Carba and multiplex PCR	16 (100)	10 (83.3)	4 (100)	9 (100)	4 (50.0)	1 (100)	44 (89.8)	0	0	0	0	89.8	100	49	14.20	53

^aGNB, Gram-negative bacilli. There were a total of 150 isolates.

^bERT, ertapenem; RCS, Rapid Carb screen.

^cThe bla genes produced were the following: for KPC, bla_KPC-2, bla_KPC-3, bla_KPC-11, bla_KPC-12; for VIM, bla_VIM-1, bla_VIM-2, bla_VIM-4, bla_VIM-5; for IMP, bla_IMP-1, bla_IMP-26; for NDM, bla_NDM-1, bla_NDM-2, bla_NDM-4, bla_NDM-5; for OXA, bla_OXA-24, bla_OXA-48; and for SME, bla_SME. One E. cloacae isolate possessed both VIM-4 and OXA-48.

^dCPGNB, carbapenemase-producing Gram-negative bacilli.

^eThe cost for the chromogenic media is the list price per plate. For the confirmation assays, it is the cost per assay, and for the algorithms, it is the cost per isolate.

^fTAT, turnaround time; the combined algorithms factor in the likelihood of the need for pure culture to initiate any confirmatory testing.

^gNA, not applicable.

^hOne OXA-24 producer was not identified by the multiplex conventional PCR, as it is specific for bla_OXA-48 and bla_OXA-181 (11).

ⁱbla_SME is not a target in the multiplex PCR. However, a simplex PCR for bla_SME was performed and was positive for the isolate (12).

^jCarb S, carbapenem susceptible.

Among the three chromogenic media designed to detect CPGNB, chromID Carba demonstrated the highest sensitivity and specificity, followed by Colorex KPC and Brilliance CRE. The chromID Carba medium failed to identify 5 CPGNB (3 isolates with OXA-48, 1 isolate with VIM-1, and 1 isolate with VIM-5). The Colorex KPC and Brilliance CRE media failed to support the growth of these same five isolates; the Colorex KPC medium additionally did not detect one of each of the NDM-1, KPC-2, and SME producers, and Brilliance CRE additionally did not detect one of each of the VIM-2, IMP-26, SME, and OXA-48 and VIM-4 producers. All three media designed for the detection of CPGNB performed poorly for the detection of OXA-48 producers, detecting at the most 5 of the 8 (63.5%) OXA-48 producers.

Phenotypic confirmatory assays.

In the second part of the study, a method comparison between the MH test and the RCS test was performed on all isolates. On the first attempt with the previously described RCS protocol, 29 (28.7%) of the carbapenemase-negative isolates gave uninterpretable results (i.e., a peach color in both tubes with and without imipenem [e.g., Fig. 1C, tubes 2a and 2b]). In addition, 4 of the carbapenemase producers were interpreted as negative (3 OXA-48 producers and 1 SME producer). Based on the initial results, troubleshooting was performed for the RCS protocol, and it was modified upon repeat to only include one 10-μl calibrated loopful of organism (reduced from two) and a round-bottom 5-ml tube (instead of a conical 1.5-ml test tube) to allow the diatab to dissolve completely (Fig. 1).

FIG 1.

Troubleshooting the Rapid Carb screen test with the use of different tubes and increasing inocula. Shown is the Rapid Carb screen comparing the test in round-bottom 5-ml tubes (A), cryovials (B), and test tubes (C) using an AmpC-producing E. cloacae isolate (which should yield a negative result by RCS). Tube 1a has imipenem, and tube 1b is without imipenem; these used one calibrated 10-μl loop of the organism. Tube 2a has imipenem, and tube 2b is without imipenem; these used two calibrated 10-μl loops of the organism. Tube 3a has imipenem, and tube 3b is without imipenem; these used three calibrated 10-μl loops of the organism. In addition, the color change in the test tubes (C) is not apparent, as the diatabs were not able to properly dissolve and resulted in uninterpretable results, especially with increased inoculum. The round-bottom 5-ml tubes (A) demonstrated the easiest-to-interpret results.

It was noted that different enzyme classes performed differently with the RCS test. KPC producers turned the bacterial suspension a characteristic bright yellow color (positive test) immediately or within minutes of inoculating the test, all VIM and IMP producers were positive at 30 min, and NDM and OXA producers and other CPGNB with relatively low carbapenem MICs turned color more slowly and yielded an intermediate orange hue toward the end of incubation. Three of the carbapenemase-positive isolates displayed intermediate positive results (orange hue), which included 3 of the 4 that were called negative by the unmodified RCS (2 OXA-48 producers and 1 SME producer). Those isolates that yielded an orange color toward the end of incubation in this study were repeated using both the RCS and the Carba NP tests to determine which assay was easier to interpret for the intermediate results (Fig. 2) (3, 5, 7). Although both the RCS and the Carba NP tests yielded the same intermediate orange color, it was found that the Carba NP test was easier for interpreting the intermediate positives. As there was variability in the degree of color change for the positives with the modified RCS test, the positives were considered those displaying any color change from that of the negative-control tubes (tubes without imipenem). Using this as a marker of positivity, as opposed to a strict yellow color change, the sensitivity reached 98.0% and the specificity was 100%. The RCS failed to identify one OXA-48-producing K. pneumoniae isolate.

FIG 2.

Comparison of the Rapid Carb screen with the Carba NP test for an SME-producing Serratia marcescens isolate at 2 h of incubation (intermediate positive; orange hue). (A) Rapid Carb screen results using round-bottom 5-ml tubes and one full 10-μl calibrated loop of the organism. (B) Carba NP assay using 1.5-ml conical test tubes and one full 10-μl calibrated loop of the organism. Tubes labeled with a 1 are positive controls of K. pneumoniae ATCC BAA1705, tubes labeled with a 2 are K. pneumoniae ATCC BAA1706, and tubes labeled with a 3 are SME-producing S. marcescens. Tubes with numbers followed by “a” lack imipenem, and tubes with numbers followed by “b” contain imipenem. (A) Note the dissolved diatabs settling at the bottom of the tubes.

Conventional multiplex PCR.

The multiplex PCR for the determination of carbapenemase gene class was verified, with a sensitivity and specificity of 95.9% and 100%, respectively. The PCR failed to identify one OXA-24 producer and one SME producer. It was expected that bla_OXA-24 would not be detected by the multiplex PCR, as this assay specifically targets bla_OXA-48 and bla_OXA-181. Similarly, bla_SME is not one of the targets included in the multiplex PCR. However, a PCR specific to SME was performed on the isolate and was found to be positive. Considering only the genes targeted by the multiplex PCR, the sensitivity and specificity achieved were both 100%.

Determination of an optimal workflow for the screening and confirmation of CPGNB.

The results of the screening and confirmatory assays (performed on all test isolates) were analyzed to develop the optimal workflow for the detection and confirmation of CPGNB. The best combination of tests in terms of analytical performance, assay cost, and turnaround time was the use of chromID Carba medium for screening, followed by the RCS test or multiplex PCR for confirmation of detected CPGNB (89.8%; Table 1). The Colorex C3Gr medium paired with the RCS test had the highest combined sensitivity (95.9%, which was partly due to the improved sensitivity for detecting OXA-producing strains, but it did not reach 100% sensitivity in this enzyme class); however, the low specificity of the ESBL chromogenic screening media for CPGNB would greatly impact the overall cost of testing if routinely implemented. The large proportion of false-positive CPGNB on Colorex C3Gr medium that would be encountered under routine conditions (in which ESBLs and organisms with reduced carbapenem susceptibility dominate in incidence over true CPGNB) would make it cost- and time-prohibitive.

DISCUSSION

The development of a screening and confirmatory method for the detection and confirmation of CPGNB is complicated by the scope of the genera and enzyme classes encompassed in this broad group of antimicrobial-resistant organisms. This study aimed to provide an optimal workflow for the detection of CPGNB to be implemented within the clinical microbiology laboratory. As such, a comprehensive evaluation of multiple chromogenic media and confirmatory methods was conducted to evaluate their analytical performance, assay cost, and turnaround time. Overall, we found that chromID Carba medium paired with the modified RCS protocol was the optimal approach (Fig. 3), yielding a combined sensitivity of 89.8% and a specificity of 100%. The multiplex PCR for carbapenemase genes provided an equal analytical performance when paired with the chromID Carba medium, but, of note, our laboratory system selected the RCS test as the confirmatory component for cost and time-to-reporting considerations. The RCS cost Can$6.38 less per test and was anticipated to be completed at a minimum of 3 h prior to the PCR result. However, for laboratories in which high-volume CPGNB screening may occur, the scalability of PCR (i.e., use of 96-well plates and associated infrastructure) may be more cost- and time-effective than the manual manipulation required to complete the RCS test.

FIG 3.

The chosen workflow for the detection and confirmation of carbapenemase-producing Gram-negative bacilli using chromID Carba medium for detection and the Rapid Carb screen kit for confirmation. GNB, Gram-negative bacilli; IP&C, Infection Prevention and Control; IDS, Infectious Disease Service.

The first part of this study involved evaluating five different chromogenic medium types used to screen for CPGNB. Of the five different chromogenic media evaluated to detect CPGNB, Colorex C3Gr without the ertapenem disk had the highest sensitivity, at 95.9%. However, not surprisingly, the two ESBL medium types (Colorex C3Gr and Brilliance ESBL) without the use of the ertapenem disk lacked specificity, allowing the growth of most ESBL and AmpC producers. The addition of the ertapenem disk to the two medium types designed to detect ESBLs helped to substantially increase the specificity of the media for detecting CPGNB; however, it resulted in a loss of sensitivity. This is the first study, to our knowledge, that incorporated a carbapenem disk to select for CPGNB from chromogenic media designed to detect ESBLs.

Among the three chromogenic media designed to detect CPGNB, chromID Carba demonstrated the highest sensitivity and specificity, followed by Colorex KPC and Brilliance CRE. All three media that were designed to detect CPGNB performed poorly for the detection of OXA-48 producers. A reduced ability to detect OXA-48 producers was also observed by Wilkinson et al. (13) in their comparison of four chromogenic medium types used for the detection of carbapenemase-producing Enterobacteriaceae. Some OXA-48 isolates in our study coproduced other β-lactamases, but these did not appear to consistently play a role in OXA-48 detection on chromogenic media, as was previously described (13). The reduced ability of the chromogenic media evaluated in this study to detect OXA-48 producers is a known limitation of the media. Recently developed media, such as Supercarba (14) and chromID OXA-48 (bioMérieux), have been described to have better sensitivity for detecting OXA-48 producers (14, 15). Neither of these media were commercially available in Canada at the time of this study.

In the second part of the study, a method comparison between the MH test and the RCS test was performed on all isolates. Initially, many uninterpretable results occurred with the RCS test. At that point, troubleshooting was performed for the assay, and improvements were made for the modified RCS protocol by using one 10-μl calibrated loopful of organism (reduced from two) and a round-bottom 5-ml tube (instead of a conical 1.5-ml tube) to allow the diatab to dissolve completely. The diatabs did not dissolve completely in the conical-bottom test tubes, as there was not enough surface area of the diatab in contact with the small volume (150 μl) present in the tube. Also, it was noted after unblinding that if a high inoculum of a carbapenemase-negative organism was present in the tubes containing imipenem, these tests would change to a peach color toward the end of incubation. The same uninterpretable results were reported by Huang et al. (4). The modifications of a lowered inoculum and the use of tubes with increased surface area are unique to this study.

Upon repeat with the modified RCS protocol, a sensitivity of 98% and a specificity of 100% were achieved. The RCS failed to identify one OXA-48-producing K. pneumoniae isolate. Previous studies have demonstrated that the RCS test, like the Carba NP test, does not identify all OXA-48 producers (4, 8). Some intermediate positive results were obtained with the modified RCS test, and these isolates were tested using the Carba NP test for comparison. The Carba NP test was easier to interpret than the RCS test for the intermediate positives; however, the advantage of the RCS test is that it is a quality-controlled commercially available laboratory kit, whereas the Carba NP test requires the preparation of reagents in-house (3, 5, 7). Overall, the vast majority (91.8%) of our tested CPGNB were yellow at 30 min, although the current clinical incidences of those enzyme classes that demonstrated an orange positive reaction are unknown.

This is the first study to date that has compared the RCS test to the MH test. The RCS test proved to be a more sensitive and specific test than the MH test, corroborating previous findings that compared the Carba NP test with the MH test (3, 5, 7).

Last, the multiplex PCR for the determination of carbapenemase gene class was verified, with a sensitivity and specificity of 95.9% and 100%, respectively. Although the multiplex PCR evaluated had a high sensitivity and specificity if used for confirmation, the costs and technical difficulty of the assay are not ideal for routine use, and in our hands, it will remain an optional test choice for when it is of clinical value to determine the carbapenemase gene class (including for infection, prevention, and control investigations).

By pairing the top-performing screening and confirmatory tests, the preferred workflow was chromID Carba agar medium, followed by the Rapid Carb screen test, yielding a combined sensitivity of 89.8% and a specificity of 100% (Fig. 3). The combination of assays was chosen based on the analytical performance characteristics, cost of testing, and turnaround time under routine conditions. Even though the combination of Colorex C3Gr with the RCS test had a higher sensitivity than that of the chromID Carba and the RCS test, the decreased sensitivity of the Colorex C3Gr medium would necessitate substantially more confirmatory testing; in our panel, this would equate to twice as many RCS tests (51 versus 103, respectively). On the other hand, the limitation of the chromID Carba medium is the limited detection of OXA-48 producers, which is reflected in the sensitivity (89.8%) of the screening algorithm chosen. Our laboratory selected the RCS test for the confirmatory component for cost and time-to-reporting considerations, as the RCS test was Can$6.38 less per test and was anticipated to be completed at a minimum of 3 h prior to a PCR result.

Subsequent to the evaluation and verification of this optimized workflow for the detection and confirmation of CPGNB, the chromID Carba and the RCS test were utilized clinically. A patient was admitted locally after transfer from an international hospital and was placed immediately on contact precautions in a single-bed room. From a wound specimen, carbapenem-resistant P. aeruginosa was isolated from the patient using routine culture. The isolate was positive by the RCS test and identified as a VIM producer by the multiplex PCR. A rectal swab was subsequently collected from the patient to screen for CPGNB using the chromID Carba medium and a BA plate with ertapenem disks (16) (the BA plate with ertapenem disks was also selected because the new verified protocol had not yet had a verification report filed and was not systematically used subsequent to this single collection). An NDM-producing K. pneumoniae isolate grew on both medium types, and an OXA-181-producing E. coli isolate grew on the BA only. A ward screen was performed to investigate if these carbapenemase-producing organisms had spread from the isolated patient. As such, 29 rectal swabs were submitted to the clinical microbiology laboratory and plated on chromID Carba medium; 27 (93.1%) had no growth after overnight incubation. Two patient cultures had P. aeruginosa breakthrough; these were subcultured onto BA and tested negative by the RCS test the following day (performed from the BA). In addition to the P. aeruginosa, a Gram-positive coccus and a yeast were present on the same chromID Carba plates. Thus, it appeared in this single investigation that breakthrough from rectal swabs on the chromID Carba medium is rare, and a Gram stain can be performed to quickly rule out breakthrough growth other than that from Gram-negative bacilli.

The limitations of this study include the small number of CPGNB tested within individual carbapenemase classes and the use of pure cultures of stocked isolates for the evaluation, as opposed to clinical or mocked rectal swabs. However, the strengths of this study include (i) a diverse subset of CPGNB, (ii) the evaluation of multiple chromogenic media and the novel use of a ertapenem disk on ESBL chromogenic media for the selection of CPGNBs, (iii) being the first study to compare the RCS test with the MH test, (iv) being the first study to modify the RCS protocol to improve the interpretation of the results, and (v) a complete workflow approach was tested for confirmation of CPGNB detection.

Overall, we found that chromID Carba was the most sensitive and specific chromogenic medium evaluated for the detection of CPGNB, and with modification to the RCS protocol, we were able to more easily interpret the results. By pairing the top-performing screening and confirmatory tests, we found that the preferred workflow was chromID Carba medium, followed by the RCS test, yielding a combined sensitivity of 89.8% and a specificity of 100%. The limitation of the selected algorithm is the poor detection of OXA-48 producers.