PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of jcmPermissionsJournals.ASM.orgJournalJCM ArticleJournal InfoAuthorsReviewers
 
J Clin Microbiol. 2016 January; 54(1): 35–42.
Published online 2015 December 30. Prepublished online 2015 September 2. doi:  10.1128/JCM.01643-15
PMCID: PMC4702732

Clinical Performance of a Matrix-Assisted Laser Desorption Ionization–Time of Flight Mass Spectrometry Method for Detection of Certain blaKPC-Containing Plasmids

Y.-W. Tang, Editor

Abstract

Rapid detection of blaKPC-containing organisms can significantly impact infection control and clinical practices, as well as therapeutic choices. Current molecular and phenotypic methods to detect these organisms, however, require additional testing beyond routine organism identification. In this study, we evaluated the clinical performance of matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) to detect pKpQIL_p019 (p019)—an ~11,109-Da protein associated with certain blaKPC-containing plasmids that was previously shown to successfully track a clonal outbreak of blaKPC-pKpQIL-Klebsiella pneumoniae in a proof-of-principle study (A. F. Lau, H. Wang, R. A. Weingarten, S. K. Drake, A. F. Suffredini, M. K. Garfield, Y. Chen, M. Gucek, J. H. Youn, F. Stock, H. Tso, J. DeLeo, J. J. Cimino, K. M. Frank, and J. P. Dekker, J Clin Microbiol 52:2804–2812, 2014, http://dx.doi.org/10.1128/JCM.00694-14). PCR for the p019 gene was used as the reference method. Here, blind analysis of 140 characterized Enterobacteriaceae isolates using two protein extraction methods (plate extraction and tube extraction) and two peak detection methods (manual and automated) showed sensitivities and specificities ranging from 96% to 100% and from 95% to 100%, respectively (2,520 spectra analyzed). Feasible laboratory implementation methods (plate extraction and automated analysis) demonstrated 96% sensitivity and 99% specificity. All p019-positive isolates (n = 26) contained blaKPC and were carbapenem resistant. Retrospective analysis of an additional 720 clinical Enterobacteriaceae spectra found an ~11,109-Da signal in nine spectra (1.3%), including seven from p019-containing, carbapenem-resistant isolates (positive predictive value [PPV], 78%). Instrument tuning had a significant effect on assay sensitivity, highlighting important factors that must be considered as MALDI-TOF MS moves into applications beyond microbial identification. Using a large blind clinical data set, we have shown that spectra acquired for routine organism identification can also be analyzed automatically in real time at high throughput, at no additional expense to the laboratory, to enable rapid detection of potentially blaKPC-containing carbapenem-resistant isolates, providing early and clinically actionable results.

INTRODUCTION

The spread of carbapenemase-producing organisms represents an urgent global public health threat in an era of increased international travel and limited effective antimicrobial options. The genes encoding carbapenemases are commonly carried on plasmids, and it has been hypothesized that an important factor contributing to their spread is the diversity of the plasmid contexts in which they are found. In particular, the blaKPC gene encoding the Klebsiella pneumoniae carbapenemase (KPC) has been identified in a variety of plasmid backbones in association with a mobile Tn4401 transposable element. This association may have contributed to the global dissemination of the blaKPC gene in a variety of genera and species of bacteria (1). Rapid methods to detect blaKPC-carrying plasmids in the clinical microbiology laboratory would be of great clinical and epidemiologic value. Various techniques such as the Carba NP test (2), modified Hodge test (3), and molecular assays (4) have proven useful for blaKPC detection but require additional steps beyond those performed for routine organism identification. Recently, matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) has been used to detect the presence of hydrolyzing enzymes (such as carbapenemases) by measuring the products of antibiotic hydrolysis directly (5,7); however, there have been limited data on the clinical utility of these assays to date.

Recently, the pKpQIL_p019 (referred to here as “p019”) protein, encoded by a gene present in certain blaKPC-containing plasmids, was found to be detected as an ~11,109-Da peak in MALDI-TOF MS spectra used for routine organism identification (8). The p019 gene is located adjacent to the blaKPC-containing Tn4401 transposon in the pKpQIL plasmid in which it was discovered (8), and a GenBank search demonstrated its presence in a number of other blaKPC-containing plasmids (8,11). Additional experiments involving transformation of other organisms with the pKpQIL plasmid demonstrated that the p019 protein could be expressed and detected in Escherichia coli and in other strains of K. pneumoniae (8). As a proof-of-principle study, it was demonstrated that this MALDI-TOF MS peak could be used retrospectively to track a blaKPC-pKpQIL plasmid responsible for an outbreak of carbapenem-resistant K. pneumoniae infections that occurred at our hospital in 2011 (8).

In the current work, we present a blind clinical validation of a MALDI-TOF MS assay to detect the presence of the p019 protein in a diverse collection of carbapenem-susceptible and -resistant Enterobacteriaceae isolates. We compared the performances of two different protein extraction methods (plate extraction and tube extraction) for peak detection, assessed method reproducibility, and monitored for operator-dependent factors by using two independent MALDI-TOF MS readers for blind analysis. An analysis script was also developed for automated peak detection. PCR amplification of the p019 gene was used as the reference standard.

(This study was presented in part at the 54th Interscience Conference on Antimicrobial Agents and Chemotherapy, Washington, DC, 2014.)

MATERIALS AND METHODS

Figure 1 provides a summary of the methods used for the prospective and retrospective clinical validation of the p019 MALDI-TOF MS peak detection assay.

FIG 1
Summary of the methods used for clinical validation of the MALDI-TOF MS method for p019 detection.

Isolates.

A total of 140 Enterobacteriaceae isolates (54 carbapenem-susceptible isolates and 86 carbapenem-resistant isolates; 60 contained blaKPC) representing eight genera and at least 14 species were recovered from frozen stock and passaged twice before testing (Table 1). Organisms were obtained from clinical cultures at the National Institutes of Health (NIH) Clinical Center between 2011 and 2014, College of American Pathologists Breakpoint Implementation Toolkit 2012, and the New York Presbyterian Hospital/Weill Cornell Medical Center (courtesy of S. Jenkins). Data from 27 (all blaKPC) of the carbapenem-resistant isolates included here were published previously as part of supplemental data (8). Testing of susceptibility to ertapenem, imipenem, meropenem, and doripenem was performed by the disk diffusion method (Thermo Fisher Scientific, Lenexa, KS). For the purpose of discussion, isolates testing as intermediate or resistant to two or more carbapenems following CLSI M100-S24 breakpoints (3) were defined as carbapenem resistant. Two additional strains (of K. pneumoniae and Escherichia coli) that had been transformed with the pKpQIL plasmid were used as positive controls for MALDI-TOF MS p019 detection (the methods used to transform these strains are described in reference 8).

TABLE 1
Summary of PCR and MALDI-TOF MS p019 peak analysis results for carbapenem-resistant and -susceptible Enterobacteriaceae isolates

MALDI-TOF MS.

Two different protein extraction methods (plate extraction and tube extraction) were performed for each isolate. For plate extractions, 1 to 4 colonies from three separate but same-day subcultures (biological replicates; <24 h of growth) were spotted in triplicate on the target plate (technical replicates), for a total of nine spots per isolate. A 1-μl volume of 70% formic acid and 2 μl of alpha-cyano-4-hydroxycinnamic acid (α-CHCA) were overlaid on each spot. For tube extractions, three independent extractions (biological replicates) were performed from different colonies picked from a single plate (<24 h of growth), and each extraction mixture was then spotted in triplicate on the target plate (technical replicates) for a total of nine spots per isolate. Tube extraction was performed as described previously (8). Protein extracts were stored at −70°C for future MALDI-TOF analysis. MALDI-TOF MS analysis was performed on a Bruker MicroFlex LT mass spectrometer (Bruker Daltonics, Billerica, MA). Spectra were acquired from 1,000 laser shots at a minimum intensity threshold of 30 arbitrary units (A.U.) as described previously (13). Plate calibration was performed on each run before spectral acquisition using a mixture of eight proteins contained in Protein Calibration I and Peptide Calibration II (Bruker Daltonics). Plate calibration met acceptable specifications when at least seven of the eight proteins were detected within 150 ppm of the “in last fit” parameter. Flex Analysis software (version 3.4; Bruker Daltonics) was used for analysis of acquired spectra after baseline subtraction.

Blind analysis of MALDI-TOF MS data by two independent operators.

Two independent operators were instructed on use of the Flex Analysis software for peak detection. A peak was defined as a deflection from baseline appearing greater than noise in the mass window at 11,109 ± 15 Da (11,094 to 11,124 Da). Operators were blind to p019 PCR and susceptibility results. For calculation of overall sensitivity and specificity, an isolate was considered positive for p019 when an 11,094-to-11,124-Da peak was present in any one representative spectrum (biological and technical triplicates) per isolate and negative when no peak meeting the criteria was detected in the nine analyzed spectra. Calculation of 95% confidence intervals was performed using MedCalc (https://www.medcalc.org/calc/diagnostic_test.php).

Method development for automated peak detection using Flex Analysis.

A script for automated peak detection was developed by modifying the Collect_Peaklists FAMS Method (version 1.2) that is available in all Flex Analysis software programs (Bruker Daltonics). To detect the p019 peak, the dRange(0) and dRange(1) variables were changed from 1,000.0 and 3,000.0 to 11,093.999 and 11,124.001, respectively. Within the method, the signal-to-noise ratio, minimum threshold, peak width, and peak height were set at 3, 0, 5, and 80, respectively. All other program parameters were set to default.

Blind analysis of MALDI-TOF MS data using automated peak detection software.

The modified script was used to analyze the 2,556 spectra generated from the 140 Enterobacteriaceae isolates and 2 control isolates after recalibration of the MALDI-TOF instrument. In addition, the script was run retrospectively against 720 spectra that were identified as representing Enterobacteriaceae at BioTyper scores of ≥2.0 that were collected as part of our routine MALDI-TOF MS identification method for clinical isolates from May through August 2014. BioTyper identification was supplemented by additional mass spectral profiles provided by several NIH-developed databases (13,16).

blaKPC, blaNDM-1, and p019 PCR.

Multiplex blaKPC and blaNDM-1 colony PCR was performed as previously published (http://www.cdc.gov/hai/pdfs/labsettings/kpc-ndm-protocol-2011.pdf). Primers and a TaqMan probe for pKpQIL_p019 PCR were designed using PrimerQuest (Integrated DNA Technologies, Inc., Coralville, IA). Colony PCR for the p019 gene was performed on a 7500 Fast real-time PCR instrument (Applied Biosystems, Foster City, CA) in a 20-μl reaction mixture consisting of 10 μl 2× TaqMan Fast Universal PCR master mix (Applied Biosystems, Grand Island, NY), 1 μM forward primer (5′-TCCTGTTGCTGTTTTTGCGG-3′), 1 μM reverse primer (5′-TGATGGTGGGCAATCATCCC-3′), 0.5 μM probe (5′-6-carboxyfluorescein [FAM]/ATGGGCGAT[ZEN]CACCATTCCATGTCT[3IABkFQ]-3′), and 5 μl of DNA. Thermal cycling conditions were 95°C for 3 min followed by 40 cycles of 95°C for 3 s and 60°C for 30 s.

RESULTS

Isolate characteristics.

Of the 140 Enterobacteriaceae examined, 86 (61%) were resistant to the carbapenems. Of these, 60 (70%) were positive for blaKPC by PCR (Table 1). p019 was detected in 26 isolates by PCR; all were positive for blaKPC, and all were carbapenem resistant.

Performance of MALDI-TOF MS to detect p019. (i) Initial analysis.

Two independent operators analyzed a total of 2,520 spectra from the 140 isolates tested (9 spectra per isolate per extraction method). The ~11,109-Da peak was detected by two independent operators in 25 of the 26 isolates that were positive for p019 by PCR (96% overall sensitivity; 95% confidence interval [CI], 80% to 100%) (Table 2). All nine replicates were detected in 21 to 24 (84% to 96%) of these isolates, depending on the extraction method, detection method, and operator (Table 2). The minimum number of replicates detected for a single isolate was four. The overall accuracy of the identifications by the two operators was high, with operator 1 and operator 2 identifying the peak in 446 (95%) and 441 (94%) spectra, respectively (98.9% agreement; Table 2). The ~11,109-Da peak was not observed by either operator for isolates where p019 was not detected by PCR (114 Enterobacteriaceae isolates; 2,052 analyzed traces; 100% specificity) (Table 2). All controls performed as expected.

TABLE 2
Summary of sensitivity and specificity for p019 MALDI-TOF MS peak detection for the 140 studied Enterobacteriaceae isolates and the different variables studied

(ii) Effect of instrument parameters on assay performance.

After our initial analysis was completed, our MicroFlex instrument underwent laser and turbo pump replacement after 4.5 years of usage, as well as annual instrument calibration and preventative maintenance. To assess the potential effects of instrument maintenance/parameters on the performance of the p019 MALDI-TOF MS assay, we obtained new spectra (in technical triplicates) using the original protein extracts from the tube extractions that had been placed in frozen storage. Upon reanalysis, the p019 peak was observed by both operators in all spectra from 26 isolates that were positive for the p019 gene (234 spectra; 9 spectra per isolate; 100% reproducibility; 100% sensitivity [95% CI, 87% to 100%]) (Table 2). All nine replicates were detected for each of the 26 p019-containing isolates. Similarly to the first analysis, no ~11,109-Da peak was observed in spectra from isolates that were negative for p019 by PCR (1,026 spectra; 100% specificity [95% CI, 99.6% to 100%]). The average signal-to-noise intensity of the collected spectra was found to have increased 3- to 30-fold after laser replacement, turbo pump replacement, cleaning, and instrument calibration.

Following the observation of the substantial effects of instrument maintenance and recalibration described above, we reran the entire analysis (including extraction) a second time to understand better the consequences of the instrument changes. All 140 Enterobacteriaceae isolates were repassaged twice from frozen stock, recoded for blind analysis, and reanalyzed in biological and technical triplicates using both plate extraction and tube extraction. Each operator independently detected the p019 peak in all 26 isolates that were positive for the p019 gene from both extraction methods (468 spectra; 100% sensitivity) (Table 2 and Fig. 2). Manual peak analysis showed 100% interoperator agreement for detection of the peak in biological and technical triplicates for both extraction methods (all nine replicates were detected for each isolate; data not shown). Specificity was 99.5%, as a peak was detected in five spectra from a single p019-negative, carbapenem-resistant isolate (isolate 137; plate extraction method only) (Table 2 and Fig. 2C).

FIG 2
Representative spectra for p019 detection. Peak intensity is displayed in arbitrary units (104 A.U.). (A and B) Presence of the ~11,109-Da peak in K. pneumoniae (A) and E. coli (B) p019-containing isolates. (C) Presence of an ~11,109-Da ...

(iii) Development and evaluation of automated peak detection software.

To standardize data analysis, a script was written to allow automatic detection of the 11,109-Da peak within a ±15-Da window using Flex Analysis software. Using this program, the ~11,109-Da peak was detected in at least one tube extraction spectral trace from each of the 26 isolates that were positive for p019 by PCR (100% overall sensitivity [95% CI, 87% to 100%]). All nine replicates were detected for 25 (96%) of these isolates (Table 2). Using the plate extraction method, the p019 peak was detected in 25 isolates (96% overall sensitivity [95% CI, 80% to 100%]); all nine replicates were detected in all 25 isolates (100%) (Table 2). The overall specificities of the automatic detection software were 99% for plate extraction (95% CI, 95% to 100%) and 95% for tube extraction (95% CI, 89% to 98%), accounting for one and six isolates, respectively, in which an ~11,109-Da peak was detected in at least one spectral trace (Table 2).

To study the utility of the automated peak detection script in unselected clinical spectra, we used the script to perform retrospective analysis of 720 routine clinical spectra that were collected between May and August 2014 and that were identified as representing Enterobacteriaceae at BioTyper scores of ≥2.0. Carbapenem susceptibility results were available for 716 (99%) isolates—41 spectra were obtained from carbapenem-resistant isolates and 675 spectra from carbapenem-susceptible isolates. Overall, the software detected a peak at 11,109 ± 15 Da in nine (1.3%) spectra—seven from carbapenem-resistant strains. PCR performed on the corresponding isolates confirmed the presence of the p019 gene in seven spectra (78%); all were from carbapenem-resistant isolates, and all were positive for blaKPC. The identifications of the 2 remaining p019-negative isolates where a peak was detected were E. coli (peak detected at 11,102 Da; total of 248 E. coli isolates tested) and Pantoea sp. (peak detected at 11,101 Da; total of 4 Pantoea isolates tested). PCR testing demonstrated that the remaining 34 carbapenem-resistant Enterobacteriaceae isolates were negative for p019, blaKPC, and blaNDM. PCR testing was not performed on carbapenem-susceptible Enterobacteriaceae corresponding to the remaining 673 spectra, and potential false negatives could thus have been missed by this analysis.

(iv) Mass of the p019 peak over different extraction methods and replicates.

The ranges of the mass attributed to p019 obtained with plate extraction and tube extraction across all biological and technical replicates after instrument tuning were 11,097 to 11,114 Da (average, 11,108 Da) and 11,097 to 11,118 Da (average, 11,104 Da), respectively. The standard deviation for technical triplicates ranged from 0 to 3.61 for plate extraction and from 0 to 8.66 for tube extraction. The standard deviation for biological triplicates ranged from 0 to 3.79 for plate extraction and from 0.58 to 8.08 for tube extraction. The ranges in standard deviations for technical and biological triplicates were similar for tube extracts pre- and post-instrument maintenance. However, a wider range in standard deviations was observed for plate extracts for both technical (0 to 11.93) and biological (0 to 6.93) triplicates prior to instrument maintenance.

DISCUSSION

We have demonstrated in this study that detection of p019 from MALDI-TOF MS profiles acquired for routine organism identification can be easily implemented in the clinical laboratory workflow, enabling real-time detection of certain blaKPC-containing plasmids that also contain the p019 gene and prediction of carbapenem resistance, with significant benefits to infection control efforts. While specific identification of all carbapenemase genes or detection of a protein signature that represents a broader range of carbapenemase-containing plasmids would be ideal, the MALDI-TOF MS p019 assay offers laboratories the opportunity to detect many blaKPC-containing organisms in real time, without additional cost, equipment, labor, or reagents. Through rigorous testing of 140 Enterobacteriaceae isolates in biological and technical triplicates (nine spectra per isolate per extraction method), we have demonstrated that p019 was detected in all replicates 96% to 100% of the time, suggesting high probability for finding the p019 peak in a single spectrum used for routine identification (Table 2). The combination of tube extraction with manual peak analysis provided the best performance (96% to 100% sensitivity and 100% specificity) (Table 2); however, these methods may not be the ideal fit for routine clinical workflow. We therefore propose an algorithm that combines plate extraction with automated peak analysis, which was shown to reliably detect the p019 peak with 99% specificity in 96% of cases (Table 2). Manual peak analysis can be available as a backup option to investigate suspect isolates such as during a known outbreak of infections with an organism carrying a p019-containing plasmid.

To investigate further the clinical utility of the automated peak detection script for finding p019, we retrospectively analyzed an additional 720 spectra that were identified as Enterobacteriaceae in our routine clinical runs. Only a small percentage (1.3%, n = 9) of these spectra had a peak detected within the window of 11,109 ± 15 Da. Seven of these isolates contained the p019 gene, corresponding to a 78% overall positive predictive value (PPV) for the method. While the absolute sensitivity and specificity of the automated script could not be determined due to the technical infeasibility of performing p019 PCR on the 673 carbapenem-susceptible isolates lacking a 11,109 ± 15-Da peak, our results suggest that this algorithm can offer high-throughput, real-time, standardized, automatic screening of routinely collected clinical MALDI-TOF MS spectra without adding to laboratory costs or to labor, reagent, or equipment requirements, making this an ideal assay in resource-limited settings. Identification of isolates containing the p019 peak in their spectra could then trigger earlier initiation of infection control measures and potential changes in therapeutic management while waiting for results from appropriate reflex testing such as CarbaNP, colony PCR for blaKPC, and susceptibility profiles. If patient isolation decisions were based on the presence of the p019 peak alone before performing confirmatory blaKPC PCR and antimicrobial susceptibility testing, the low number of false-positive results in our study (2/9 [22%] isolates from the retrospective clinical cohort and 7/114 [6%] p019-negative isolates in the prospective cohort) suggests that only a small percentage of patients would be placed on isolation precautions unnecessarily. These precautions, however, may outweigh the consequences associated with potential outbreaks.

Although many studies have demonstrated the utility of MALDI-TOF MS in the clinical microbiology setting, the inherent variability in instrument function and its impact on assay sensitivity have not been well described in the literature. It is possible that use-dependent wear of integral instrument parts (e.g., lasers and turbo pumps) over time may not have significant consequences for routine bacterial and yeast identification, where many redundant spectral features are calculated into the overall identification algorithm. Reproducibility of MALDI-TOF MS for bacterial and yeast identification has been well demonstrated (17,19), leading to recent FDA approval of bacterial and yeast databases for the Bruker BioTyper and Vitek MS (bioMérieux, Durham, NC). In our case, declining instrument performance was not obvious from our routine bacterial runs; rather, instrument compromise was suspected when identification scores from our routine clinical mold and mycobacterial runs decreased over time. Presently, not many laboratories are performing routine identification of molds and mycobacteria with MALDI-TOF MS and our findings suggest that use of MALDI-TOF MS for identification of these organisms may result in a more sensitive marker for assessing extraction efficiency and optimal instrument adjustment. In this study, retesting of the original protein extracts after parts were replaced in the instrument improved sensitivity from 96% to 100% (Table 2); however, since several parts were changed in addition to the routine maintenance, it is unclear which particular variable led to improved results. Similar findings were described previously by Lau et al. (8) in a study in which a clearly defined peak was detected in only 78% of the original spectra obtained in 2011 from the outbreak isolates, and yet a peak was detected in all spectra when the organisms were recultured and rerun on the instrument 2 years later, following interval instrument cleanings and recalibrations (8). Since nonoptimized performance of a part (or parts) within an instrument can significantly impact the sensitivity of assays, especially those that solely detect a single spectral peak, it is conceivable that negative effects may also be seen for other tests such as antibiotic hydrolysis assays that detect small shifts in antibiotic mass (5,7). The effect on these assays may be especially important because the drug products are small and below the normal acquisition range of the Bruker MicroFlex LT MALDI-TOF mass spectrometer (5,7). Increasing use of MALDI-TOF MS assays that are sensitive to shifts in drug or protein mass or of those that rely solely on detection of a single peak in the clinical laboratory will provide insight into how common this issue is and help us better understand its potential impact on clinical diagnostics in the broader setting.

Although MALDI-TOF MS peak intensities are measured in arbitrary units, their values may provide insight into the relative levels of protein expression between isolates. In our study, one isolate that contained the p019 gene demonstrated a detectable ~11,109-Da peak only after instrument part replacement (Table 2). This isolate exhibited p019 peak intensity levels that were still 5- to 42-fold lower than those demonstrated by the remaining 25 peak-positive isolates and accounted for the differences in sensitivity of automated detection (96%) and manual detection (100%). Testing of a larger number of isolates is required to assess the frequency of organisms that express p019 at low levels and that may present as false negatives when different instruments (with their inherent function differences) are utilized.

Given the inherent variability in measurements of protein mass by relatively imprecise clinical MALDI-TOF mass spectrometers, biological and technical triplicate experiments using both plate extraction and tube extraction were performed to assess the reproducibility of the MALDI-TOF MS method for detecting the p019 peak. No significant difference was observed between technical and biological replicates; however, the apparently lower standard deviations among spectra obtained by plate extraction (0 to 3.61 and 0 to 3.79, respectively) compared with tube extraction (0 to 8.66 and 0.58 to 8.08, respectively) may be associated with the natural variability in plate calibration between runs since the extractions were assayed on independent plates. Nonetheless, 26 (100%) and 25 (96%) p019-containing isolates were detected within a window of 11,109 ± 15 Da using manual and automated methods, respectively, suggesting that this range is acceptable for the detection of p019. Although peptide and protein standards were used to calibrate each run in this study due to the standard operating procedures in our laboratory, it is reasonable to assume that detection of the p019 peak could also be performed on plates calibrated with the Bruker IVD Bacterial Test Standard, though this was not tested directly in this study.

MALDI-TOF MS is now an integral part of many clinical microbiology laboratories, and there is increasing interest in utilizing this technology for rapid prediction of antimicrobial susceptibility profiles to help guide therapeutic choices. Here, we have shown that identification of a p019 peak from MALDI-TOF MS spectra acquired for organism identification may be a useful predictor for some strains of carbapenem-resistant Enterobacteriaceae (78% PPV in our data set). However, we evaluated a relatively low number of isolates containing the p019 gene in this data set (30% of carbapenem-resistant isolates and 43% of blaKPC-containing isolates; Table 1), and a larger-scale analysis is warranted. Nonetheless, our data suggest that positive detection of the p019 peak could potentially provide an opportunity for clinicians and hospital epidemiologists to institute rapid enhanced infection control precautions and to modify clinical practices before traditional susceptibility test results become available. Detection of the p019 peak, however, should not be considered a replacement for antimicrobial susceptibility testing or assays that screen for carbapenemase activity. As we have noted previously (8, 20, 21), caution must be applied in extrapolating the presence of functional blaKPC carbapenemase from the presence of the p019 protein. As the two genes are carried within different genetic structures on the plasmids in which they have been observed, it is certainly possible that p019 may be found to occur independently of blaKPC as more isolates are sequenced. Work is now under way to determine whether additional protein signatures exist for the detection of other plasmids, including those that are highly associated with carbapenemases other than blaKPC.

In conclusion, the results of this study show that MALDI-TOF MS is a sensitive and reproducible method for detecting p019. This simple methodology is applicable to other clinical microbiology laboratories because it can provide real-time detection of certain carbapenem-resistant bacteria that contain a blaKPC-containing plasmid encoding an associated p019 protein on that same plasmid with significant benefits for infection control efforts, using mass spectra acquired as part of routine organism identification without placing additional constraints on laboratory resources. This methodology has proven to work well in our institution because of the known presence of pKpQIL-containing organisms in our hospital environment (8, 12, 22). This methodology has now been implemented in the routine clinical workflow of our laboratory, with all spectra scanned by the automated script for peaks within a window of 11,109 ± 15 Da using the Flex Analysis program (Bruker Daltonics). Peaks detected from Enterobacteriaceae spectra are communicated to infection control and clinical staff with appropriate reflex testing for blaKPC PCR and carbaNP and susceptibility testing. The remaining carbapenemase-containing clinical isolates that do not carry p019-containing plasmids are then detected later following susceptibility testing and confirmatory carbapenemase assays. The p019 method proposed in this study can be easily implemented in any laboratory with a Bruker MALDI-TOF MS system to detect certain blaKPC-containing plasmids or to track potential nosocomial transmissions. The applicability of the p019 assay to the Vitek MS system is yet to be determined.

ACKNOWLEDGMENTS

We thank Stephen Jenkins of the New York Presbyterian Hospital/Weill Cornell Medical Center for providing isolates for testing and Mark Fisher of ARUP Laboratories for insightful discussions.

This work was supported by the Intramural Research Program of the National Institutes of Health.

The content of this article is solely our responsibility and does not represent the official views of the National Institutes of Health.

S.K.D., K.M.F., J.P.D., and A.F.L. have been involved in a collaborative agreement with Bruker Daltonics Inc. to develop organism databases for MALDI-TOF MS. Bruker Daltonics Inc. had no involvement in the planning, execution, or interpretation of experiments or in the writing of the manuscript. The remaining authors declare that we have no competing interest.

Requests for isolates used in this study will require a Material Transfer Agreement.

REFERENCES

1. Cuzon G, Naas T, Truong H, Villegas MV, Wisell KT, Carmeli Y, Gales AC, Venezia SN, Quinn JP, Nordmann P 2010. Worldwide diversity of Klebsiella pneumoniae that produce beta-lactamase blaKPC-2 gene. Emerg Infect Dis 16:1349–1356. doi:.10.3201/eid1609.091389 [PMC free article] [PubMed] [Cross Ref]
2. Nordmann P, Poirel L, Dortet L 2012. Rapid detection of carbapenemase-producing Enterobacteriaceae. Emerg Infect Dis 18:1503–1507. doi:.10.3201/eid1809.120355 [PMC free article] [PubMed] [Cross Ref]
3. Clinical and Laboratory Standards Institute. 2014. Performance standards for antimicrobial susceptibility testing; twenty-fourth informational supplement, CLSI document M100-S24. Clinical and Laboratory Standards Institute, Wayne, PA.
4. Cunningham SA, Noorie T, Meunier D, Woodford N, Patel R 2013. Rapid and simultaneous detection of genes encoding Klebsiella pneumoniae carbapenemase (blaKPC) and New Delhi metallo-beta-lactamase (blaNDM) in Gram-negative bacilli. J Clin Microbiol 51:1269–1271. doi:.10.1128/JCM.03062-12 [PMC free article] [PubMed] [Cross Ref]
5. Hrabák J, Walková R, Studentová V, Chudácková E, Bergerová T 2011. Carbapenemase activity detection by matrix-assisted laser desorption ionization–time of flight mass spectrometry. J Clin Microbiol 49:3222–3227. doi:.10.1128/JCM.00984-11 [PMC free article] [PubMed] [Cross Ref]
6. Carvalhaes CG, Cayo R, Visconde MF, Barone T, Frigatto EA, Okamoto D, Assis DM, Juliano L, Machado AM, Gales AC 2014. Detection of carbapenemase activity directly from blood culture vials using MALDI-TOF MS: a quick answer for the right decision. J Antimicrob Chemother 69:2132–2136. doi:.10.1093/jac/dku094 [PubMed] [Cross Ref]
7. Vogne C, Prod'Hom G, Jaton K, Decosterd L, Greub G 12 December 2014, posting date. A simple, robust and rapid approach to detect carbapenemases in Gram negative isolates by MALDI-TOF mass spectrometry: validation with triple quadripole tandem mass spectrometry, microarray and PCR. Clin Microbiol Infect doi:.10.1111/1469-0691.12715 [PubMed] [Cross Ref]
8. Lau AF, Wang H, Weingarten RA, Drake SK, Suffredini AF, Garfield MK, Chen Y, Gucek M, Youn JH, Stock F, Tso H, DeLeo J, Cimino JJ, Frank KM, Dekker JP 2014. A rapid matrix-assisted laser desorption ionization–time of flight mass spectrometry-based method for single-plasmid tracking in an outbreak of carbapenem-resistant Enterobacteriaceae. J Clin Microbiol 52:2804–2812. doi:.10.1128/JCM.00694-14 [PMC free article] [PubMed] [Cross Ref]
9. Frasson I, Lavezzo E, Franchin E, Toppo S, Barzon L, Cavallaro A, Richter SN, Palu G 2012. Antimicrobial treatment and containment measures for an extremely drug-resistant Klebsiella pneumoniae ST101 isolate carrying pKPN101-IT, a novel fully sequenced blaKPC-2 plasmid. J Clin Microbiol 50:3768–3772. doi:.10.1128/JCM.01892-12 [PMC free article] [PubMed] [Cross Ref]
10. Chen L, Chavda KD, Al Laham N, Melano RG, Jacobs MR, Bonomo RA, Kreiswirth BN 2013. Complete nucleotide sequence of a blaKPC-harboring IncI2 plasmid and its dissemination in New Jersey and New York hospitals. Antimicrob Agents Chemother 57:5019–5025. doi:.10.1128/AAC.01397-13 [PMC free article] [PubMed] [Cross Ref]
11. Ho PL, Cheung YY, Lo WU, Li Z, Chow KH, Lin CH, Chan JF, Cheung VC 2013. Molecular characterization of an atypical IncX3 plasmid pKPC-NY79 carrying blaKPC-2 in a Klebsiella pneumoniae. Curr Microbiol 67:493–498. doi:.10.1007/s00284-013-0398-2 [PubMed] [Cross Ref]
12. Conlan S, Thomas PJ, Deming C, Park M, Lau AF, Dekker JP, Snitkin ES, Clark TA, Luong K, Song Y, Tsai YC, Boitano M, Dayal J, Brooks SY, Schmidt B, Young AC, Thomas JW, Bouffard GG, Blakesley RW, Mullikin JC, Korlach J, Henderson DK, Frank KM, Palmore TN, Segre JA 2014. Single-molecule sequencing to track plasmid diversity of hospital-associated carbapenemase-producing Enterobacteriaceae. Sci Transl Med 6:254ra126. doi:.10.1126/scitranslmed.3009845 [PMC free article] [PubMed] [Cross Ref]
13. Lau AF, Drake SK, Calhoun LB, Henderson CM, Zelazny AM 2013. Development of a clinically comprehensive database and a simple procedure for identification of molds from solid media by matrix-assisted laser desorption ionization–time of flight mass spectrometry. J Clin Microbiol 51:828–834. doi:.10.1128/JCM.02852-12 [PMC free article] [PubMed] [Cross Ref]
14. Stevenson LG, Drake SK, Shea YR, Zelazny AM, Murray PR 2010. Evaluation of matrix-assisted laser desorption ionization–time of flight mass spectrometry for identification of clinically important yeast species. J Clin Microbiol 48:3482–3486. doi:.10.1128/JCM.00687-09 [PMC free article] [PubMed] [Cross Ref]
15. Saleeb PG, Drake SK, Murray PR, Zelazny AM 2011. Identification of mycobacteria in solid-culture media by matrix-assisted laser desorption ionization–time of flight mass spectrometry. J Clin Microbiol 49:1790–1794. doi:.10.1128/JCM.02135-10 [PMC free article] [PubMed] [Cross Ref]
16. Fedorko DP, Drake SK, Stock F, Murray PR 2012. Identification of clinical isolates of anaerobic bacteria using matrix-assisted laser desorption ionization–time of flight mass spectrometry. Eur J Clin Microbiol Infect Dis 31:2257–2262. doi:.10.1007/s10096-012-1563-4 [PubMed] [Cross Ref]
17. Westblade LF, Garner OB, MacDonald K, Bradford C, Pincus DH, Mochon AB, Jennemann R, Manji R, Bythrow M, Lewinski MA, Burnham CA, Ginocchio CC 2015. Assessment of reproducibility of matrix-assisted laser desorption ionization–time of flight mass spectrometry for bacterial and yeast identification. J Clin Microbiol 53:2349–2352. doi:.10.1128/JCM.00187-15 [PMC free article] [PubMed] [Cross Ref]
18. Mellmann A, Bimet F, Bizet C, Borovskaya AD, Drake RR, Eigner U, Fahr AM, He Y, Ilina EN, Kostrzewa M, Maier T, Mancinelli L, Moussaoui W, Prevost G, Putignani L, Seachord CL, Tang YW, Harmsen D 2009. High interlaboratory reproducibility of matrix-assisted laser desorption ionization–time of flight mass spectrometry-based species identification of nonfermenting bacteria. J Clin Microbiol 47:3732–3734. doi:.10.1128/JCM.00921-09 [PMC free article] [PubMed] [Cross Ref]
19. Branda JA, Rychert J, Burnham CA, Bythrow M, Garner OB, Ginocchio CC, Jennemann R, Lewinski MA, Manji R, Mochon AB, Procop GW, Richter SS, Sercia LF, Westblade LF, Ferraro MJ 2014. Multicenter validation of the VITEK MS v2.0 MALDI-TOF mass spectrometry system for the identification of fastidious gram-negative bacteria. Diagn Microbiol Infect Dis 78:129–131. doi:.10.1016/j.diagmicrobio.2013.08.013 [PubMed] [Cross Ref]
20. Partridge SR. 2014. Tn4401 carrying blaKPC is inserted within another insertion in pKpQIL and related plasmids. J Clin Microbiol 52:4448–4449. doi:.10.1128/JCM.02426-14 [PMC free article] [PubMed] [Cross Ref]
21. Lau AF, Wang H, Weingarten RA, Drake SK, Suffredini AF, Chen Y, Gucek M, Youn JH, Stock F, Tso H, DeLeo J, Cimino JJ, Frank KM, Dekker JP 2014. Reply to “Tn4401 carrying blaKPC is inserted within another insertion in pKpQIL and related plasmids.” J Clin Microbiol 52:4450. doi:.10.1128/JCM.02614-14 [PMC free article] [PubMed] [Cross Ref]
22. Snitkin ES, Zelazny AM, Thomas PJ, Stock F; NISC Comparative Sequencing Program Group, Henderson DK, Palmore TN, Segre JA 2012. Tracking a hospital outbreak of carbapenem-resistant Klebsiella pneumoniae with whole-genome sequencing. Sci Transl Med 4:148ra116. doi:.10.1126/scitranslmed.3004129 [PMC free article] [PubMed] [Cross Ref]

Articles from Journal of Clinical Microbiology are provided here courtesy of American Society for Microbiology (ASM)