Matrix-Assisted Laser Desorption Ionization–Time of Flight Mass Spectrometry Meropenem Hydrolysis Assay with NH₄HCO₃, a Reliable Tool for Direct Detection of Carbapenemase Activity

Abstract

A comparison of a matrix-assisted laser desorption ionization–time of flight mass spectrometric (MALDI-TOF MS) meropenem hydrolysis assay with the Carba NP test showed that both methods exhibited low sensitivity (approximately 76%), mainly due to the false-negative results obtained with OXA-48-type producers. The addition of NH₄HCO₃ to the reaction buffer for the MALDI-TOF MS assay dramatically improved its sensitivity (98%). Automatic interpretation of the MALDI-TOF MS assay, using the MBT STAR-BL software, generally agreed with the results obtained after manual analysis. For the Carba NP test, spectrophotometric analysis found six additional carbapenemase producers.

TEXT

Nosocomial infections caused by carbapenem-resistant Enterobacteriaceae and Pseudomonas spp. are now emerging worldwide and are difficult to treat (1). Previous data have shown that strict epidemiological intervention based on the rapid detection of carbapenemase production can prevent the spread of those bacteria (2). Therefore, the introduction of rapid and sensitive methodologies for the detection of carbapenemase-producing bacteria is of utmost importance. Recently, two new highly sensitive and rapid methods for the direct detection of carbapenemase activity were developed. In 2011, two research groups demonstrated that matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) can detect carbapenemase activity, based on mass spectral profiles obtained from carbapenem molecules (3, 4). The second method, the Carba NP test, is a biochemical method used in the detection of carbapenemase activity in Enterobacteriaceae, Pseudomonas spp., and Acinetobacter species (5–7). This test is based on a decrease in pH resulting from the hydrolysis of the β-lactam ring of carbapenem molecules, which is detected using phenol red as a pH indicator.

The aim of this study was to validate the efficiency of a MALDI-TOF MS meropenem hydrolysis assay and the Carba NP test for the detection of carbapenemase producers. The performances of both methods were compared with that of a recently published modification of the MALDI-TOF MS assay (MALDI-TOF BIC) in an effort to increase the reliability of detecting OXA-48 producers (8). Furthermore, the automatic interpretation of all methods was evaluated.

The methods were tested against a group of 124 Enterobacteriaceae and 37 Pseudomonas aeruginosa isolates from collections of the Faculty of Medicine and University Hospital in Plzen (Czech Republic), the National Medicines Institute in Warsaw (Poland), the Robert Koch Institute in Wernigerode (Germany), and the University Hospital in Larissa (Greece). The isolates were previously characterized, as described below. All isolates were tested for extended-spectrum β-lactamase (ESBL) and AmpC expression by the ESBL double-disk synergy test (DDST) with cefotaxime, ceftazidime, aztreonam, and amoxicillin-clavulanate disks in the absence and presence of cloxacillin (250 μg/ml) (9). Susceptibility to carbapenems was determined by using imipenem, meropenem, and ertapenem disks and interpreting them according to the EUCAST criteria (http://www.eucast.org/). All isolates that were nonsusceptible to at least one carbapenem were subjected to metallo-β-lactamase (MβL), Klebsiella pneumoniae carbapenemase (KPC), and OXA-48 phenotypic detection using DDST with EDTA (10), the combined disk test with phenylboronic acid (11), and a temocillin disk (12), respectively. PCR detection of various bla genes was performed by the use of specific primers for bla_SHV-, bla_CTX-M-, bla_CMY-, bla_KPC-, bla_VIM-, bla_IMP-, bla_NDM-, and bla_OXA-48-like genes, as reported previously (13–20). The PCR products were purified and sequenced on both strands, as described previously (21). The MICs of ceftazidime and meropenem were determined by the broth dilution method (22). The group included 113 isolates producing KPC- (n = 21), VIM- (n = 44), NDM- (n = 24), IMP- (n = 4), and OXA-48-type (n = 19) carbapenemases and one isolate coproducing VIM and IMP metallo-β-lactamases (23–29). The remaining 48 isolates were non-carbapenemase producers (Table 1) (21, 30).

TABLE 1.

Results of the MALDI-TOF MS assays and Carba NP test^a

Resistance mechanism	Bacterial species	No. of isolates	Reference/source	MIC range (μg/ml) for:		No. of carbapenemase producers detected usingb:
				Ceftazidime	Meropenem	MALDI-TOF assay	Carba NP test	MALDI-TOF BIC assay
KPC	K. pneumoniae	20	23, this study	32 to >32	8 to >16	20 (20)	20 (20)	19 (19)
	E. cloacae	1	This study	>32	>16	1 (1)	1 (1)	1 (1)
VIM	K. pneumoniae	6	23, this study	32 to >32	1 to >16	6 (6)	6 (6)	6 (6)
	E. cloacae	7	This study	24 to >32	1 to >16	7 (7)	7 (7)	7 (7)
	Serratia marcescens	2	This study	4 to 8	>16	2 (2)	2 (2)	2 (2)
	Klebsiella oxytoca	1	This study	>32	0.25	1 (1)	1 (1)	1 (1)
	Citrobacter freundii	1	This study	16	2	1 (1)	1 (1)	1 (1)
	L. adecarboxylata	1	24	>32	4	1 (1)	1 (1)	1 (1)
	P. aeruginosa	26	25, 26	16 to >32	4 to >16	16	24 (26)	25
NDM	K. pneumoniae	4	27	>32	4 to 16	4 (4)	4 (4)	4 (4)
	E. cloacae	10	27, 28	>32	0.5 to 16	10 (10)	6 (8)	10 (10)
	E. coli	10	27	>32	1 to >16	10 (9)	8 (9)	10 (9)
IMP	S. marcescens	1	This study	>32	>16	1 (0)	1 (1)	1 (0)
	P. aeruginosa	3	25	>32	>16	3	2 (3)	3
VIM + IMP	P. aeruginosa	1	25	32	>16	1	1 (1)	1
OXA-48-like	K. pneumoniae	11	29, this study	0.5 to >32	0.5 to >16	1 (1)	1 (1)	11 (11)
	E. cloacae	1	29	>32	8	0 (0)	0 (0)	1 (1)
	E. coli	6	29, this study	0.5 to 32	0.5 to >16	1 (1)	0 (0)	6 (6)
	Raoultella ornithinolytica	1	29	>32	4	1 (1)	0 (0)	1 (1)
Total carbapenemase producers	Enterobacteriaceae	83				67 (65)	59 (62)	82 (80)
	P. aeruginosa	30				20	27 (30)	29
CTX-M	K. pneumoniae	3	This study	>32	4 to 8	0 (0)	0 (0)	0 (0)
	E. coli	15	21	1 to >32	≤0.125	0 (0)	0 (0)	0 (0)
SHV	K. pneumoniae	7	This study	16 to >32	0.125 to 16	0 (0)	0 (0)	0 (0)
CTX-M + SHV	K. pneumoniae	10	This study	16 to >32	0.125 to >16	0 (0)	0 (0)	0 (0)
CTX-M + CMY	E. coli	1	21	16	≤0.125	0 (0)	0 (0)	0 (0)
DHA	K. pneumoniae	5	30	8 to >32	0.5 to >16	0 (0)	0 (0)	0 (0)
Nonec	P. aeruginosa	7	25, 26	4 to >32	8 to >16	0	0 (0)	0
Total non-carbapenemase producers	Enterobacteriaceae	41				0 (0)	0 (0)	0 (0)
	P. aeruginosa	7				0	0 (0)	0

^aRows representing the total number of isolates per category are in bold.

^bNumbers refer to the number of carbapenemase producers that were correctly detected by each method, while the numbers in parentheses mention the carbapenemase producers that were correctly identified by automatic interpretation of the MALDI-TOF MS assay, Carba NP test, and MALDI-TOF BIC assay. Automatic interpretation was not applied to the spectra of P. aeruginosa isolates.

^cNegative by PCR analysis for bla_SHV-, bla_CTX-M-, bla_CMY-, bla_KPC-, bla_VIM-, bla_IMP-, bla_NDM-, and bla_OXA-48-like genes.

The MALDI-TOF MS meropenem hydrolysis assay was performed as previously described (31). For the measurement, 1 μl of the sample was applied on a stainless steel MALDI target plate (MSP 96 target; Bruker Daltonics) and allowed to dry. Each sample was overlaid with 1 μl of matrix solution (10 mg/ml 2,5-dihydroxybenzoic acid and 0.1 μM reserpine diluted in 50% ethanol). After air drying, spectra were measured within the m/z range 350 to 700 using a microflex LT mass spectrometer (Bruker Daltonics). The analysis of the spectra was performed using the flexAnalysis 3.0 software, based on the interpretation criteria shown in Table 2. The Carba NP test was carried out on strains cultured on Columbia blood agar plates (bioMérieux), as described by Dortet et al. (32). The results of all methods tested were interpreted by three independent technicians who were unaware of the identities of the isolates.

TABLE 2.

Criteria used for manual interpretation of both MALDI-TOF MS assays

Strain type	Peak at m/z ofa:
	358.5	380.5	384.5	406.5
Carbapenemase producerb	Present	Present	Absent	Absent
Non-carbapenemase producerc	Absent	Absent	Present	Present

^a358.5-m/z peak, decarboxylated product of meropenem; 380.5-m/z peak, sodium salt of the decarboxylated product of meropenem; 384.5-m/z peak, meropenem; 406.5-m/z peak, sodium salt of meropenem. A peak was interpreted as noise if its intensity was not five times higher than the intensity of the background peaks.

^bAn isolate was interpreted as a carbapenemase producer if the presence of at least one of the decarboxylated products of meropenem was detected.

^cAn isolate was interpreted as a non-carbapenemase producer if the absence of both decarboxylated products of meropenem and the presence of meropenem and/or its sodium salt were observed.

The results are summarized in Tables 1 and 3. False-positive results were not observed among non-carbapenemase producers with either the MALDI-TOF MS assay or the Carba NP test. Also, both methods performed well against the 21 KPC producers and the isolate coproducing VIM and IMP carbapenemases. The MALDI-TOF MS assay correctly detected all NDM and IMP producers but failed to detect a carbapenemase in 10 VIM-producing P. aeruginosa isolates. The false-negative results found with P. aeruginosa can be explained by the presence of numerous background peaks of significant amplitude (>20% amplitude of the target peaks; Table 2), precluding a definitive analysis of the spectra. On the other hand, the Carba NP test failed to detect three P. aeruginosa isolates expressing either VIM (n = 2) or IMP (n = 1) and six NDM-producing isolates of Enterobacter cloacae (n = 4) and Escherichia coli (n = 2). Of note, both methods experienced significant problems with the subset of the 19 isolates producing OXA-48-type β-lactamases, which are enzymes with weak carbapenemase activity. The MALDI-TOF MS assay correctly detected three out of the 19 (19%) OXA-48-type producers, while the Carba NP test detected the production of a carbapenemase in only one of these isolates (6%). However, an attempt to increase the enzyme quantity in the revealing solution (7) by using the double bacterial inoculum had no impact on the performance of Carba NP test.

TABLE 3.

Diagnostic values of the MALDI-TOF MS assays and Carba NP test for the detection of carbapenemase-producing isolates^a

Method	No. of results				Sensitivity	Specificity
	TP	TN	FP	FN
MALDI-TOF assay	87 (65)	48 (41)	0 (0)	26 (18)	77% (78%)	100% (100%)
Carba NP test	86 (92)	48 (48)	0 (0)	27 (21)	76% (81%)	100% (100%)
MALDI-TOF BIC assay	111 (80)	48 (41)	0 (0)	2 (3)	98% (96%)	100% (100%)

^aTP, true positive; TN, true negative; FP, false positive; FN, false negative. The numbers in parentheses refer to the diagnostic values for the automatic interpretation for each method. Automatic interpretation was not applied to the spectra of P. aeruginosa isolates.

The MALDI-TOF BIC assay differed from the original assay only in that the reaction buffer (20 mM Tris-HCl, 0.01% sodium dodecyl sulfate, 0.1 mM meropenem) was supplemented with 50 mM NH₄HCO₃ (pH was adjusted to 7.0). Validation of the MALDI-TOF BIC assay indicated that the addition of NH₄HCO₃ did not have an impact on the specificity (100%) of the assay, while it dramatically improved its sensitivity (98%). One hundred eleven carbapenemase producers, including all NDM-, IMP-, OXA-48-type-producing isolates and the isolate coproducing VIM and IMP carbapenemases, were correctly detected by the modified assay (Table 1). The MALDI-TOF BIC assay missed one KPC-producing K. pneumoniae isolate and one P. aeruginosa isolate expressing VIM metallo-β-lactamase. The failure of the MALDI-TOF BIC assay to detect these two isolates might be explained by low expression of carbapenemase enzymes in the respective isolates or by the inhibitory effects of NH₄HCO₃. However, the inhibitory effects should be further examined, since we are unaware of the side effects of NH₄HCO₃. The addition of NH₄HCO₃ to the Carba NP test was not tested, since it might be problematic due to the alkalization of the reaction buffer (31).

Automatic interpretation of results is important for the standardization of a diagnostic process. For Enterobacteriaceae, raw spectra that were acquired with the MALDI-TOF MS and MALDI-TOF BIC assays were automatically analyzed using the MBT STAR-BL prototype software (Bruker Daltonics). The intensity ratio of the hydrolyzed to intact meropenem signals indicates the level of carbapenemase activity. The MBT STAR-BL prototype software calculated the summed hydrolyzed and intact meropenem intensities, as well as the corresponding ratio (logRQ value). The results were normalized to the signal intensities obtained from positive- and negative-control strains and displayed as a plot (Fig. 1), which was evaluated by the unaided eye. The automatic interpretation was generally in accordance with the results obtained after manual analysis (Table 1). Discrepancies were observed only for the same two carbapenemase-producing isolates, which were positive by manual reading of both assays but interpreted as negative by the software. The failure of the software to detect the two carbapenemase-producing isolates might be explained by the fact that both of them exhibited significantly low intensity ratios. Intensity values, which are arbitrary units, may be affected by several variables, resulting in a significant influence on the intensity ratios. Automatic analysis was not applied on P. aeruginosa isolates due to the lower quality of their spectra.

FIG 1.

Image obtained after automatic interpretation of spectra by using MBT STAR-BL prototype software. The isolates with normalized logRQ values above the threshold line were interpreted as positive for carbapenemase production, while isolates with normalized logRQ values below the threshold line were interpreted as negative. Under our experimental conditions, the best sensitivity and specificity values were obtained by setting up the threshold line to 0.5.

Furthermore, an automatic reading of Carba NP test was attempted by using a PR 3100 TSC enzyme-linked immunosorbent assay (ELISA) reader (Bio-Rad). The absorbances of the samples were compared with those of the negative controls. Each sample had its own negative control. The samples with lower absorbances were read as positive, while samples with absorbances that were higher or equal to those of the negative controls were read as negative. The use of an ELISA reader found six additional carbapenemase producers, interpreted by the technicians as being non-carbapenemase producers (Table 1), and increased the sensitivity of the method to 81%. This finding underlined that the manual interpretation of Carba NP test was inaccurate, especially for samples in which the color change could not be detected by the unaided eye.

In conclusion, the MALDI-TOF MS assay and Carba NP test performed well, exhibiting excellent specificity (100%). Measuring the Carba NP test by an ELISA reader is important for archiving primary data, which is a legal requirement in most European countries. Additionally, the use of an ELISA reader increased the sensitivity of Carba NP test, indicating that automatic reading may help with subjectivity in the interpretation, especially for doubtful results. However, the Carba NP test displayed a lower sensitivity than was initially reported (76% in this study versus 100%) (5, 6), mainly due to false-negative results obtained with OXA-48-type producers. This finding is in agreement with previous studies reporting the limited ability of Carba NP test to detect OXA-48-type-producing isolates and some weak carbapenemases of class A, like GES-5 and SME-1 (33, 34). For the MALDI-TOF MS assay, the limitation in detecting OXA-48-type producers was overtaken by addition of NH₄HCO₃ to the reaction buffer. The MALDI-TOF BIC assay correctly detected 98% of the carbapenemase producers, demonstrating this method to be a reliable tool for detecting carbapenemase activity. Furthermore, the results from the automatic interpretation of both MALDI-TOF MS assays suggested that the use of the MBT STAR-BL prototype software can be helpful for nonexperienced users facing difficulties with spectrum analysis.