Simplified Testing Method for Direct Detection of Carbapenemase-Producing Organisms from Positive Blood Cultures Using the NG-Test Carba 5 Assay

We directly tested 484 organisms from clinical (n = 310) and simulated (n = 174) positive blood cultures using the NG-Test Carba 5 assay for carbapenemase-producing Enterobacterales detection. The assay identified all but 4 of the KPC (170/171), OXA-48-like (22/22), VIM (19/21), and NDM (14/15) producers with no false positives.

ABSTRACT

We directly tested 484 organisms from clinical (n = 310) and simulated (n = 174) positive blood cultures using the NG-Test Carba 5 assay for carbapenemase-producing Enterobacterales detection. The assay identified all but 4 of the KPC (170/171), OXA-48-like (22/22), VIM (19/21), and NDM (14/15) producers with no false positives. Among the clinical Klebsiella pneumoniae organisms tested, 122 of 123 KPC, 1 of 1 OXA-48-like, and 1 of 2 VIM producers were detected by the assay. Some VIM and NDM producers yielded scant but still-readable bands with the assay. No organisms produced the IMPs that the assay was designed to detect.

TEXT

Carbapenemase-producing Enterobacterales (CPE) infections, especially Klebsiella pneumoniae, have emerged as a global public health concern over the last decade (1) because of their complex multidrug resistance phenotypes and their ability to spread rapidly within hospital and community settings (2). Because CPE infections, including bloodstream infections, are associated with increased rates of treatment failure and poor outcomes, timely and reliable detection and identification of CPE are essential to help physicians provide quick optimal therapeutic management and implement appropriate infection control measures (3, 4).

In contrast to the recent advent of molecular methods, the NG-Test Carba 5 immunochromatographic assay (NG Biotech, Guipry, France) has been developed to rapidly detect the five main carbapenemases, i.e., KPC, OXA-48-like, NDM, VIM, and IMP (www.ngbiotech.com), which belong to classes A, B, and D of the Ambler classification. Of four studies evaluating the performance of the assay to date, two used cultured bacterial isolates (5, 6), and the remaining two used positive blood cultures (BCs) that were obtained by spiking BC bottles with whole blood and bacteria (7, 8). Despite overall sensitivity and specificity that ranged from 97.3% to 100% and 95.3% to 100%, respectively (5–8), these studies did not exactly or completely reflect clinical scenarios. Furthermore, none of the studies using positive BCs tried to simplify the sample preparation method for performing the NG-Test Carba 5 assay, which, in practice, consisted of centrifugation and wash steps before obtaining the bacterial pellet for final lysis. Therefore, a testing method with no required centrifugation and wash steps would make the NG-Test Carba 5 assay more feasible for use by clinical laboratories.

Here, we tested 484 bacterial organisms directly from clinical (n = 310) and spiked (n = 174) BCs to detect the NG-Test Carba 5 assay-targeted carbapenemases, and, notably, we used the simplified method to prepare the test samples, as mentioned above. In this study, the simulation phase served to expand the gamut of testable CPE organisms, whereas the clinical phase involved BCs collected from patients hospitalized at either the Fondazione Policlinico Universitario A. Gemelli IRCCS or the Azienda Ospedaliera San Camillo-Forlanini of Rome (Italy), where KPC-producing K. pneumoniae infections are endemic. First, we spiked BacT/Alert FA Plus BC bottles (bioMérieux, Marcy l’Etoile, France), previously filled with 10-ml sterile human whole blood, with a 0.5-ml suspension (30 CFU/bottle) of each of the 174 previously characterized clinical isolates (Table 1), according to standard guidelines (https://www.asm.org/Guideline/Cumitechs). Bottles were incubated until they signaled positive by the BacT/Alert Virtuo system (bioMérieux), with bacterial concentrations ranging from 2 × 10⁷ to 6 × 10⁹ CFU/ml. Second, we included 310 consecutively positive BCs from nonduplicate patients that grew K. pneumoniae organisms (Table 2), as identified by matrix-assisted laser desorption ionization–time of flight (MALDI-TOF) mass spectrometry-based diagnostic workflow (9). By this workflow, we subjected positive-BC bottle broth aliquots to direct analysis with the MALDI Biotyper system (Bruker Daltonics, Bremen, Germany) for species-level identification. Then, only for K. pneumoniae organisms, we subjected the corresponding aliquots to direct analysis with the NG-Test Carba 5 assay (see below). Because we performed this assay immediately after organism identification, the time between BC positivity and the Carba assay was around 1 h. This related to only the BCs that signaled positive during our laboratory working hours (7 a.m. to 7 p.m. Monday through Friday and 7 a.m. to 4 p.m. Saturday). For the BCs that signaled positive during non-working hours, the Carba assay was performed as soon as the laboratory was in operation. For all 484 bacterial isolates, we determined meropenem MICs using commercial broth-microdilution antimicrobial susceptibility testing panels (MERLIN Diagnostica GmbH, Bornheim, Germany), as recommended (http://www.eucast.org/ast_of_bacteria/mic_determination/). The isolates also underwent in-house PCR sequencing of bla_TEM, bla_SHV, bla_CTX-M, bla_CMY, bla_IMI, bla_GES, bla_KPC, bla_IMP, bla_NDM, bla_OXA-48-like, bla_VIM, and bla_AmpC β-lactamase resistance genes (10). We performed the NG-Test Carba 5 assay according to the manufacturer’s instructions with some modification. Briefly, we gently mixed 40-μl aliquots from each BC bottle broth with 5 drops of extraction buffer to perform the lysis step, which required a few seconds. We then dispensed 100 μl of the suspension into the well of the immunochromatographic cassette. After 15 min incubation at room temperature, two independent blinded readers interpreted the NG-test Carba 5 assay results.

TABLE 1.

Details of 174 carbapenemase- and non-carbapenemase-producing bacterial organisms tested in simulated positive blood cultures by the NG-Test Carba 5 assay

Species	Carbapenemase status (no. positive/total)a and relative meropenem susceptibility
	KPC (n = 48)		OXA-48-like (n = 21)		VIM (n = 19)b		NDM (n = 15)b		Combination (n = 8)		Other carbapenemase/noncarbapenemase (n = 63)
	Type	MIC (μg/ml)	Type	MIC (μg/ml)	Type	MIC (μg/ml)	Type	MIC (μg/ml)	Type	MIC (μg/ml)	Type	MIC (μg/ml)
Klebsiella pneumoniae (total no. positive)	40		15		11		7		6		27
	KPC-2 (15/15)	4 to 16	OXA-48 (15/15)	2 to 64	VIM-1 (8/8)	4 to 32	NDM-1 (7/7)	16 to 32	KPC-2 + VIM-1 (1/1)	64	CTX-M-15 (0/9)	≤0.125
	KPC-3 (15/15)	8 to >64			VIM-1 + SHV-12 (1/1)	8			KPC-3 + VIM-1 (5/5)	32 to 64	SHV-2a (0/5)	≤0.125
	KPC-3 + CTX-M-15 (10/10)	8 to >64			VIM-4 + CMY-4 (2/2)	32					SHV-5 (0/3)	≤0.125
											SHV-12 (0/8)	≤0.125
											CMY-2 (0/2)	≤0.125
Escherichia coli (total no. positive)	8		5				7				33
	KPC-3 (8/8)	2 to 64	OXA-48 (5/5)	1 to 4			NDM-1 (5/5)	8 to 16			CTX-M-3 (0/2)	≤0.125
							NDM-1 + CTX-M-15 (1/1)	32			CTX-M-10 (0/2)	≤0.125
							NDM-5 (1/1)	>64			CTX-M-14 (0/5)	≤0.125
											CTX-M-2 + SHV-12 (0/2)	≤0.125
											CTX-M-15 + impermeability (0/5)	2
											SHV-2a (0/3)	≤0.125
											SHV-11 (0/2)	≤0.125
											SHV-12 (0/5)	≤0.125
											CMY-2 (0/6)	≤0.125
											FOX-7 + impermeability (0/1)	4
Enterobacter cloacae complex (total no. positive)			1		5				2		1
			OXA-48 (1/1)	8	VIM-1 (4/4)	8 to 32			VIM-1 + NDM-1 (2/2)	16 to 32	IMI-2 (0/1)	64
					VIM-4 + CMY-4 (1/1)	32
Citrobacter freundii (total no. positive)					1
					VIM-1 (1/1)	1
Klebsiella oxytoca (total no. positive)					2	8
					VIM-1 (1/2)	8
Proteus mirabilis (total no. positive)							1
							NDM-1 (0/1)	2
Pseudomonas aeruginosa (total no. positive)											2
											GES-2 (0/1)	16
											IMP-13 (0/1)	4

^aThe NG-Test Carba 5 immunochromatographic assay is capable of detecting the following variants from the 5 carbapenemases (KPC, OXA-48-like, VIM, IMP, and NDM) included as targets: KPC-2/-3, NDM-1/-4/-5/-6/-7/-9, IMP-1/-8/-11, VIM-1/-2/-4/-19, and OXA-48/-181/-204/-232/-244/-517/-519/-535. Because no organism produced 1 of the detectable IMPs, the assay was unable to detect the only IMP (i.e., IMP-13) among the carbapenemases produced by the organisms in the study.

^bThe NG-Test Carba 5 immunochromatographic assay failed to detect 2 (1VIM-1 and 1NDM-1) of the carbapenemases produced by the organisms in the study.

TABLE 2.

Details of 310 carbapenemase- and non-carbapenemase-producing bacterial organisms tested in clinical positive blood cultures by the NG-Test Carba 5 assay

Species	Carbapenemase status (no. positive/total)a and relative meropenem susceptibility
	KPCb		OXA-48-like		VIMb		Noncarbapenemase
	Type	MIC (μg/ml)	Type	MIC (μg/ml)	Type	MIC (μg/ml)	Type	MIC (μg/ml)
Klebsiella pneumoniae (total no. positive)	123		1		2		184
	KPC-2 (4/4)	8 to >64	OXA-48 (1/1)	16	VIM-1 (0/1)	8	CTX-M-15 (0/72)	≤0.125
	KPC-3 (98/99)	8 to >64			VIM-1 + CTX-M-15 (1/1)	4	SHV-12 (0/12)	≤0.125
	KPC-3 + CTX-M-15 (20/20)	4 to >64					None (0/100)	≤0.125

^aThe NG-Test Carba 5 immunochromatographic assay is capable of detecting the following variants from the 5 carbapenemases (KPC, OXA-48-like, VIM, IMP, and NDM) included as targets: KPC-2/-3, NDM-1/-4/-5/-6/-7/-9, IMP-1/-8/-11, VIM-1/-2/-4/-19, and OXA-48/-181/-204/-232/-244/-517/-519/-535. No organisms produced IMP or NDM carbapenemases detectable by the assay.

^bThe NG-Test Carba 5 immunochromatographic assay failed to detect 2 (1 KPC-3 and 1 VIM-1) of the carbapenemases produced in the study.

When performed on spiked BCs (Table 1), the NG-Test Carba 5 assay yielded positive results for 109 (98.2%) of 111 Enterobacterales organisms that produced KPC (48/48), OXA-48-like (21/21), VIM (18/19), NDM (14/15), and KPC/VIM or VIM/NDM combinations (8/8). The two CPE organisms that the assay failed to detect were VIM-1 and NDM-1 producers. All of the 63 bacterial organisms (including Pseudomonas aeruginosa), that produced noncarbapenemase (n = 60) or carbapenemase other than those detectable by the assay (n = 3, including an IMP-13) tested negative. When performed on clinical BCs (Table 2), the NG-Test Carba 5 assay yielded positive results for 124 (98.4%) of 126 K. pneumoniae isolates that produced KPC (122/123), OXA-48-like (1/1), and VIM (1/2). The two K. pneumoniae organisms that the assay failed to detect were KPC-3 and VIM-1 producers, and the former had a hypermucoviscous phenotype in culture. All of the 184 non-carbapenemase-producing K. pneumoniae organisms tested negative. Consistent with these findings, we noticed that the immunochromatographic assay bands were scant (but still readable) with some VIM and NDM producers, whereas they were sharp with all KPC and OXA-48-like producers. Considering spiked and clinical BCs together, the overall sensitivity and specificity of the NG-Test Carba 5 assay were 98.3% (95% confidence interval [CI], 95.7% to 99.3%) and 100% (95% CI, 98.5% to100%), respectively. Considering only the clinical BCs, the NG-Test Carba 5 assay showed 98.1% (95% CI, 90.3% to 99.9%) sensitivity and 100% (95% CI, 95.6% to 100%) specificity among the BCs tested within 1 h of signaling positive (n = 137) and 98.4% (95% CI, 94.4% to 99.7%) sensitivity and 100% (95% CI, 97.9% to 100%) specificity among the BCs tested overall (n = 310).

Our microbiology laboratory experience with the NG-Test Carba 5 assay confirms and extends previous data from simulated BC testing. Compared with the proof-of-principle studies by Takissian et al. (7) and Bodendoerfer et al. (8), we used a more simplified lysis step on the BC broths (which requires only 2 min of manual work) before allowing the migration of their carbapenemase content. However, this procedural simplification apparently did not affect the NG-Test Carba 5 assay’s performance; the immunochromatographic bands were always readable with our method. As a result, there were only 2 undetected carbapenemase producers (1 VIM and 1 NDM) among the 111 carbapenem-resistant (meropenem MICs, >0.125 μg/ml) CPEs tested by us. Note that extensive pretreatment of BC broths did not restore a positive signal for the 3 carbapenemase producers (2 VIM and 1 NDM) that were negative on initial testing in the Takissian et al. study (7). In line with the two studies (7, 8), the specificity of the NG-Test Carba 5 assay was very high, as none of the 63 bacteria producing other carbapenemases or not producing carbapenemases gave positive results (Table 1). Similarly, we obtained promising data from clinical BC testing. We included only K. pneumoniae isolates as CPE organisms (123 KPC, 2 VIM, and 1 OXA-48-like), reflecting the fact that, in Italy, 34% of the K. pneumoniae isolates causing bloodstream infections in 2015 were carbapenem resistant (http://www.epicentro.iss.it/focus/resistenza_antibiotici/EpidItalia.asp). We showed that 98.4% (124/126) and 100% (184/184) of the K. pneumoniae isolates tested were correctly detected as producers (meropenem MIC, >0.125 μg/ml) or nonproducers (meropenem MIC, ≤0.125 μg/ml), respectively (Table 2). Unfortunately, we did not evaluate the NG-Test Carba 5 assay for IMP and (only in the clinical study) NDM types, which presents a need for further clinical validation studies.

In conclusion, we show that the NG-Test Carba 5 assay can be a reliable and convenient means of accelerating the laboratory diagnosis of CPE bloodstream infections, especially in hospital settings dominated by a carbapenem-resistant organism population with a restricted number of type/variant carbapenemase genes.