Phenotypic Detection of Carbapenemase-Producing Enterobacteriaceae by Use of Matrix-Assisted Laser Desorption Ionization–Time of Flight Mass Spectrometry and the Carba NP Test

Abstract

We compared the diagnostic accuracy of the Carba NP test with that of a straightforward matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) method for detecting carbapenemase-producing Enterobacteriaceae (CPE). Using PCR as the reference method, both tests demonstrated a sensitivity of 87% and a specificity of 100%. MALDI-TOF MS offers a potential alternative for the rapid detection of CPE in the clinical laboratory setting.

TEXT

The global spread of carbapenemase-producing Enterobacteriaceae (CPE) is a major threat to public health (1). The efficient laboratory detection of these agents requires a rapid phenotypic test that provides accurate results while awaiting confirmation by PCR. In this study, a method was developed for detecting CPE using an imipenem hydrolysis assay and matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS), using the Vitek MS research use only (RUO) system (bioMérieux, Marcy l'Étoile, France). Using PCR as a reference method, the diagnostic accuracy of MALDI-TOF MS was compared with that of the Carba NP test (CNP).

The investigation was conducted at Melbourne Pathology and Swinburne University of Technology, Melbourne, Australia, between October 2013 and April 2014. A panel of 105 genetically characterized Enterobacteriaceae was assembled from nine Australian laboratories. All strains were identified by MALDI-TOF MS (Vitek MS [bioMérieux]) using the Vitek MS Database version 2.0 (bioMérieux) and had MICs for meropenem determined by Etest (bioMérieux). Molecular characterization had been performed by various methods (2–11); at a minimum, each strain had been subjected to PCR, including targets for bla_KPC, bla_NDM, bla_IMP, bla_VIM, and bla_OXA-48 (2–11). The entire collection was assembled prior to the study, and all of the strains obtained were included.

The CNP was performed according to a previously published method (12) on strains cultured on Columbia blood agar (CBA) for 18 to 24 h. The results were read by a microbiologist who was blinded to all other results in the study.

The MALDI-TOF MS method was adapted from a previous investigation (13) and developed through small pilot studies. The strains were cultured for 18 to 24 h on CBA at 35 to 37°C. A 1-μl loopful of bacteria was suspended in 100 μl of 0.45% saline containing 0.25 mg/ml imipenem-cilastatin and incubated for 4 h at 35 to 37°C. The suspensions were then centrifuged at 12,300 × g for 3 min. One microliter of the supernatant was spotted in quadruplicate onto a FlexiMass target plate (bioMérieux), overlaid with 1 μl of α-cyano-4-hydroxycinnamic acid matrix (bioMérieux), and dried at room temperature. Control antibiotic solutions were processed identically and also spotted in quadruplicate. Immediately upon drying, the target plates were inserted into the Vitek MS RUO instrument, and their spectra were acquired by firing once per well (10 laser shots per profile). The Vitek MS RUO system consisted of the Axima Performance mass spectrometer (Shimadzu Corporation, Kyoto, Japan) and Launchpad version 2.9 (Shimadzu Corporation). Launchpad was used to optimize the Vitek MS RUO settings for the low-molecular-mass range, perform the calibration, and review the spectra.

The definition of a positive result for the imipenem hydrolysis assay was selected based on the results of the pilot studies and specified prior to the final experiment. This was defined as the complete disappearance of the imipenem peak at 300 Da, as indicated by a median area under the curve (AUC) for the test replicates of <5% of the median AUC of the control replicates. Any other result was considered negative (Fig. 1). No peak corresponding to the predicted imipenem hydrolysis product ([M_hydr + H]⁺, 318 Da) was identified.

FIG 1.

Spectra in MALDI-TOF MS method for CPE detection. Top, imipenem; middle, imipenem with K. pneumoniae harboring bla_IMP; bottom, imipenem with PCR-negative K. pneumoniae. a, area under the curve.

Sensitivity and specificity were estimated for the CNP and MALDI-TOF MS, using PCR as the reference method. The strains that were found to be discrepant by phenotypic and genotypic testing were subjected to multiplex PCR and a reverse line blot assay (11) at the Centre for Infectious Diseases and Microbiology, Westmead Hospital, Sydney, Australia. This method provided the final PCR result for analyzing these strains. A receiver operator characteristic (ROC) curve analysis was used to establish the optimal cutoff for the MALDI-TOF MS method using MedCalc. Extensive post hoc spectral analysis was also conducted using SARAMIS Premium version 4.10 (bioMérieux).

After resolution of the 12 discrepant strains, two strains that were received as PCR positive were found to be negative by the reference PCR method (11). These were analyzed as non-CPE, in accordance with the testing protocol. In total, there were 46 carbapenemase-producing strains (20 KPC, 9 NDM, 9 IMP, 2 VIM, and 6 OXA producers) and 59 strains that were negative for the gene targets tested. The collection included 11 different species. The results for all the strains are presented in Table 1. Overall, the sensitivity and specificity of both tests were found to be 87% (95% confidence interval [CI], 74.3 to 93.9%) and 100% (95% CI, 93.9 to 100%), respectively.

TABLE 1.

Results of the Carba NP test and MALDI-TOF MS (n = 105)

Species (no. of different strains)	PCR detection result	Meropenem Etest MIC (mg/liter)	Carba NP test	MALDI-TOF MS
K. pneumoniae (10)	KPC-2	>32	+	+
K. pneumoniae (5)	KPC	>32	+	+
K. pneumoniae	KPC-2	>32	−	+
K. pneumoniae	KPC-2	32	+	+
K. pneumoniae	KPC	32	+	+
K. pneumoniae	KPC-2	16	+	+
E. coli	KPC	2	+	+
E. coli	NDM	>32	+	+
K. pneumoniae	NDM-2	>32	+	+
K. pneumoniae (2)	NDM-1	>32	+	+
K. pneumoniae	NDM	>32	+	+
E. coli	NDM-1	24	+	+
E. coli	NDM-5	16	+	+
Proteus penneri	NDM	4	+	−
Morganella morganii	NDM-1	2	+	−
Serratia marcescens	IMP	>32	+	+
Citrobacter freundii	IMP-4	32	+	+
E. coli	IMP	24	+	+
K. pneumoniae	IMP-4	2	+	+
K. pneumoniae	IMP-4	1.5	+	+
K. pneumoniae	IMP	1.5	+	+
C. freundii (2)	IMP	1	+	+
E. coli	IMP-4	0.38	+	+
Proteus mirabilis	VIM-1	0.25	+	−
P. mirabilis	VIM-1	0.25	+	−
K. pneumoniae	OXA-48-like	>32	−	+
K. pneumoniae	OXA-48-like	>32	−	+
E. coli	OXA-48	2	+	+
E. coli	OXA-48	0.38	−	+
E. coli	OXA-51-like	2	−	−
E. coli	OXA-23-like	0.38	−	−
E. coli	Negative	32	−	−
E. coli (19)	Negative	0.016–1	−	−
Enterobacter cloacae	Negative	8	−	−
E. cloacae (3)	Negative	1.5	−	−
E. cloacae (8)	Negative	0.064–0.5	−	−
K. pneumoniae	Negative	1.5	−	−
K. pneumoniae (10)	Negative	0.032–1	−	−
Enterobacter aerogenes	Negative	6	−	−
E. aerogenes (3)	Negative	0.094–0.5	−	−
Klebsiella oxytoca (3)	Negative	0.047–0.064	−	−
P. mirabilis (3)	Negative	0.064–0.094	−	−
C. freundii (2)	Negative	0.094	−	−
S. marcescens (2)	Negative	0.047–0.094	−	−
Citrobacter koseri	Negative	0.047	−	−
M. morganii	Negative	0.125	−	−

For MALDI-TOF MS, more complex spectral comparisons using SARAMIS Premium yielded identical results to those of the much simpler single-peak method obtained with Launchpad for all strains tested and by all methods used (Cohen's kappa, 1 for all comparisons; data not shown). No peaks were seen in the range of 200 Da to 700 Da that were consistently present in any group of CPE types and consistently absent among the non-CPE. The ROC curve analysis revealed no alternative cutoff value with superior results to the value of 5% that was chosen prospectively.

The CNP is a rapid easy-to-use phenotypic test of carbapenemase production. It has consistently been found to be 100% specific for detecting carbapenemases in Enterobacteriaceae (12, 14); sensitivity estimates, however, range from 72.5 to 100% (12, 14). Previously, the main limitation found has been a reduced ability to detect OXA-48-producing strains (14, 15). Unfortunately, the present investigation provides further evidence of this shortcoming.

MALDI-TOF MS performed well overall and detected 97% of the meropenem-resistant strains by CLSI criteria (16). Additionally, it correctly identified all four OXA-48 producers. Given the previously published success of MALDI-TOF MS for detecting OXA production (13), this aspect of its performance warrants further investigation. While the reason MALDI-TOF MS failed to identify carbapenemase production by the three strains of Proteus spp. is not entirely clear, it may be related to improved enzyme detection after cell lysis, which forms a part of the CNP but not the MALDI-TOF MS method.

The main limitation of this study was the relatively small sample size. In particular, it was possible to incorporate only two VIM-producing strains. Additionally, care must be taken when generalizing these results to settings with differing prevalences of the various carbapenemase types described.

For clinical laboratories using Vitek MS that have access to the appropriate software, the method described can be relatively easily implemented and offers an alternative to the CNP. The main advantages of MALDI-TOF MS are the objective endpoint and the possibility of better detection of OXA-48 producers. The main disadvantage of MALDI-TOF MS is the extra time required for incubation (4 h versus 2 h), processing of target plates, and interpretation of the results (total hands-on time of several minutes). This could be minimized with efficient incorporation into the laboratory workflow and greater automation of the testing procedure by the manufacturer.

In conclusion, both the CNP and MALDI-TOF MS performed well. MALDI-TOF MS offers a potential alternative for the rapid detection of CPE in the clinical laboratory setting.