Rapid Identification of Klebsiella pneumoniae by Matrix‐Assisted Laser Desorption/Ionization‐Time of Flight Mass Spectrometry and Detection of Meropenem Resistance by Flow Cytometric Assay

Abstract

Background

The aim of this study was to develop a rapid detection method of carbapenem‐resistant Klebsiella pneumoniae (CRKP) strains both MALDI‐TOF MS and flow cytometry (FCM).

Methods

A total of 174 K. pneumoniae strains were included in this study. Molecular characterization of carbapenemase gene was performed by PCR. Bacterial identification was performed by MALDI‐TOF‐MS. Meropenem susceptibility was tested at the concentrations of breakpoints described by the Clinical and Laboratory Standards Institute (CLSI) guide by FCM.

Results

Sixty‐two CRKP were positive for at least one carbapenemase gene. A total of 174 K. pneumoniae isolates obtained from clinically relevant material were correctly identified by Bruker MALDI‐TOF MS with log (score) >2.0. These results were 100% concordant with the Phoenix^™ Automated Microbiology System (BD, MD) and conventional identification results. Based on the analysis of the receiver operating characteristic (ROC) curves, the best validity and sensitivity data were obtained with a cut‐off value of 18.88% by FCM. The concordance, sensitivity, and specificity for FCM by the selected cut‐off values were 99.4%, 98.9%, and 100%, respectively.

Conclusions

We conclude that reliable results on bacterial identification and meropenem susceptibility test can be obtained within 2 hr combined by MALDI‐TOF‐MS and FCM.

Introduction

Enterobacteriaceae species are among the most important human pathogens, causing community – and hospital – acquired infections 1. Klebsiella pneumoniae is the most common pathogenic Klebsiella species and belongs to the Enterobacteriaceae 2. In humans, K. pneumoniae most often colonizes the nasopharynx and the intestinal tract and is an important etiological agent of a wide variety of serious infectious diseases, including urinary tract, respiratory tract, and blood stream infections 2, 3.

Carbapenems such as ertapenem, imipenem, and meropenem are the first‐line antimicrobial drugs for treatment of serious healthcare‐associated infections caused by K. pneumoniae 4. During the past decade, resistance to carbapenems has been increasingly reported worldwide in K. pneumoniae. The spread of the carbapenem‐resistant K. pneumoniae (CRKP) in hospitals represents a serious threat to public health, since treatment options are very limited 5. Although the prevalence of carbapenem‐resistant Enterobacteriaceae (CRE) isolates is reported relatively low, it has dramatically increased in recent years in Turkey. Among Enterobacteriaceae, high prevalence of antibiotic resistance has been reported in K. pneumoniae isolates 6.

Rapid identification of K. pneumoniae and detection of carbapenem resistance are critical for early and targeted antimicrobial therapy and infection control 7. The delay to appropriate antibiotic treatment is a major risk factor for infection associated morbidity and mortality. Conventional methods for bacterial identification and antibiotic‐susceptibility test are time consuming, labor intensive, and still require 48–72 hr from sample collection to test result 8.

Matrix‐assisted laser desorption/ionization time of flight mass spectrometry (MALDI‐TOF MS) was first developed in the late 1980s. Since then, the system has been adapted for microorganism identification. The overall microorganism's species identification workflow is simple, time saving, and cost effective in this system 9. MALDI‐TOF MS assays to detect β‐lactamase activity have also been developed in the recent years 10. Flow cytometry (FCM) was first appeared in microbiology in the former 1970s. FCM is a rapid, powerful high‐throughput technology for the differentiation and functional analysis of microorganisms. FCM has also been used previously for antibiotic‐susceptibility testing of several clinical important microorganisms 11, 12.

The aim of this study was to develop a rapid detection method of CRKP strains using both MALDI‐TOF MS and FCM.

Material and Methods

Bacterial strains and antibiotic‐susceptibility testing

A total of 174 K. pneumoniae strains (87 carbapenem resistant, 87 carbapenem susceptible) isolated from various clinical specimens (blood [n = 98], urine [n = 49], wound aspirate [n = 12], catheter [n = 7], transtracheal aspirate [n = 5], and other samples [n = 3]) between May 2011 and October 2015 were included in the study. Microorganisms were firstly defined by colony morphology, microscopic morphology by Gram stain, and further phenotypic characteristics by the Phoenix System (Becton Dickinson Diagnostic Systems, Sparks, MD). All isolates were tested for resistance against carbapenems (ertapenem, imipenem, and meropenem) by the Phoenix System. Interpretation of the test results was made using the Clinical Laboratory Standards Institute (CLSI) criteria 13. The K. pneumoniae isolates were stored in 10% skimmed milk solution at −80°C. The stored bacterial isolates were sub‐cultured on 5% sheep blood agar and Eosin Methylene Blue agar plates (Salubris AS, Turkey) and incubated at 37°C for 48 hr. Additionally, susceptibility to ertapenem, imipenem, and meropenem was determined by E‐test (Oxoid, Hampshire, UK) and was interpreted in accordance with the standards of CLSI 13. K. pneumoniae ATCC 13883 and Escherichia coli ATCC 25922 were used as quality control strains for susceptibility testing.

Molecular detection of resistance genes

DNA was isolated from bacterial colonies using the boiling lysis method as previously described (boiling a fresh cultured colony in 100 μl sterile distilled water for 10 min and subsequent centrifugation at 16,000 × g for 5 min) 14. Two micro liters of supernatant were used as the template in a 25 μl PCR. The PCR assays were carried out to amplify the entire sequences of the bla _KPC, bla _NDM‐1, bla _OXA‐48, bla _VIM, and bla _IMP, genes using previously described specific oligonucleotide primers 15. Five microliters of PCR product was run in gel electrophoresis (1.5% agarose, 1× TBE, 100V) using molecular standard (K180‐250 UL; Amresco, Solon, OH). The gel was stained with ethidium bromide, visualized with UV and evaluated by Gel Doc 2000 software (Bio‐Rad laboratories, Hercules, CA). K. pneumoniae ATCC 13883 (resistance genes negative), K. pneumoniae ATCC 1705 (bla _KPC positive), K. pneumoniae ATCC BAA 2146 (bla _NDM‐1 positive), K. pneumoniae NCTC 13440 (bla _VIM positive), and K. pneumoniae (GATA‐64) (bla _OXA‐48 positive) (kindly provided by Prof. Zerrin AKTAS [Istanbul University, Turkey]) were used in the study as reference strains.

Bruker's MALDI‐TOF MS analysis

A single colony was smeared directly onto MALDI plate. One microliter α‐cyano‐4‐hydroxycinnamic acid (CHCA) matrix solution (Bruker Daltonics, Billerica, MA) was added to the sample and allowed to dry. Analysis was carried out with Microflex LT MALDI‐TOF MS device and Flex control 3.0 software (Bruker Daltonics, Germany) optimized for microorganism identification at the linear positive ion mode between 2.000 and 20.000 Da mass ranges. The spectrums were obtained by 60 Hz laser shots, which consist 40 pockets in totally 240.

Flow cytometric analysis

All bacterial strains were grown at 37°C on 5% sheep blood agar plates (Salubris AS, Istanbul, Turkey) for 18–24 hr. Bacterial cells suspensions were prepared in Mueller Hinton broth (Salubris AS) and were incubated at 37°C with shaking until reaching the log phase (about 1 hr and 15 min), correspond with 5 × 10⁶ CFU/ml. Subsequently, bacterial suspensions were exposed with 1 mg/l meropenem (Sigma Co., St. Louis, MO) which is MIC breakpoint of meropenem according to CLSI document 13. Cell suspension without antibiotic treatment was also served as antibiotic‐free control. The suspensions were incubated for 60 min at 37°C. After incubation, the cells were washed and stained with 420 nM thiazole orange (TO) (Sigma Co., Munich, Germany) and 48 μM propidium iodide (PI) (Sigma Co.) for 5 min at room temperature, protecting from light.

Flow cytometry analysis to measure the fluorescence of bacterial suspensions was performed on a BD Accuri C6 flow cytometer (Becton Dickinson, Piscataway, NJ) with 488 nM excitation light and emission collected through a 530/30 nM band pass filter for TO (FL1) and >670 nM band pass filter for PI (FL3). Usually, 10,000 cells were analyzed for each sample and light scatter and fluorescence signals were collected with log amplification.

In order to differentiate between susceptible and resistance of the strains to the meropenem, percentage of viable bacterial counts obtained from cells with meropenem were compared, through receiver operating characteristic (ROC) curve analysis, with those obtained from the antibiotic‐free control cells. In this manner, the optimal threshold for discriminating between susceptibility and resistance was obtained. The susceptibility results were compared to those of the E‐test results accepted as the gold standard.

Statistical analysis

Statistical analysis was performed with SPSS software version 20.0 (SPSS Inc. Chicago, IL) to generate the ROC curves and calculate the associated areas and specificities and sensitivities.

Results

Carbapenemase genes analysis

The 87 CRKP were tested for carbapenemase genes by PCR, of which 62 were positive for at least one gene. PCR results showed that 40 K. pneumoniae strains harbored only bla _OXA‐48 gene, seven strains had only bla _NDM‐1 and one strain had only bla _VIM gene. Both bla _OXA‐48 and bla _NDM‐1 were present in six strains, whereas three strains had both bla _NDM‐1 and bla _IMP genes. Two K. pneumoniae were positive for bla _OXA‐48, bla _NDM‐1, and bla _VIM genes, while bla _OXA‐48, bla _NDM‐1, and bla _IMP genes were determined in three K. pneumoniae strains.

MALDI‐TOF MS analysis

A total of 174 K. pneumoniae isolates obtained from clinically relevant material were correctly identified by Bruker MALDI‐TOF MS at the species level with log scores >2.0. These results were 100% concordant with the Phoenix identification results.

FCM analysis

All experiments were performed and compared to appropriate paired antibiotic‐free controls. Initially to evaluate auto fluorescence, tube containing only broth stained with TO and PI and tube containing antibiotic‐free cells unstained with TO and PI was differentiated by forward scatter (FSC) and side scatter (SSC) threshold from noise, medium components and cellular debris. No significant basal fluorescence was recorded stained broth and unstained antibiotic‐free cells (Fig. 1).

Figure 1.

Forward scatter (FSC) vs. side scatter (SSC) signals in tube (a) containing Mueller Hinton broth stained with TO and PI to evaluate broth fluorescence, (b) containing unstained antibiotic‐free cells to evaluate cells auto fluorescence.

Gates were set up based on live and dead cells. Forward scatter, side scatter and fluorescence intensities of the individual particles (labeled with TO and labeled with PI) were detected. The percentage of live bacteria were calculated based upon the gated population by forward and side scatter and the area under the obtained curve, which is automatically normalized by the FCM software (Becton Dickinson). Fluorescence from the FL1 (TO) and FL3 (PI), FSC, and SSC signals were monitored for each treatment at each point by FCM (Fig. 2).

Figure 2.

Forward scatter (FSC), side scatter (SSC), FL1 (thiazole orange [TO], and FL3 (propid‐ium iodide [PI]) fluorescence intensities of the antibiotic‐free control (a–d), meropenem‐resistant K. pneumoniae ATCC BAA 2146 (e–h), and meropenem‐susceptible K. pneumoniae ATCC 13883 (i–l) strains. Percentages of dead cells show in upper right (UR) quadrant in panels c, g, and k. Histograms of dead and live cells counts depicts in panels d, h, and l. The M1 area shows viable cells, and dead bacteria are in the M2 area. The proportion of dead cells recorded in panels d, h, and l were 0.9%, 1.7%, and 73.1%, respectively.

On FSC vs. SSC panels, to distinguish bacteria cells from auto fluorescence related to “background,” we marked red color the fluorescence from TO (Fig. 2a, e, i). All particles fluoresced red were accepted as bacteria cells, because TO stains just live/dead cells; remaining fluorescents were accepted as auto fluorescence. In following, the gates were established for all K. pneumoniae on the basis of their TO fluorescence for antibiotic‐free control (Fig. 2b), resistant (Fig. 2f) and susceptible (Fig. 2j) strains in FL1. The mean peak values on the basis of their PI fluorescence in FL3 were used to calculate of percentage of live and dead cells in antibiotic‐free control (Fig. 2c), resistant (Fig. 2g) and susceptible (Fig. 2k) strains. There was no significant PI fluorescence respecting dead cells observed in antibiotic‐free control cells (Fig. 2d). The meropenem‐resistant strains were not stained by PI and showed a fluorescence profile identical to the antibiotic‐free control cell (Fig. 2g and h). The typical fluorescence distribution of susceptible bacterial population is depicted in Figure 2k and l.

Figure 3 shows the percentage of bacterial count curves of meropenem susceptible, resistant K. pneumoniae incubated with meropenem and antibiotic‐free control strain. In the antibiotic‐free control, percentage of bacterial count remained nearly the same after incubation for 15, 30, 45, and 60 min with meropenem. In the resistant K. pneumonia, percentage of bacterial count obtained in the presence of antibiotic was nearly similar to those of the antibiotic‐free control. However, meropenem‐sensitive strain observed that percentage of bacterial count was significantly reduced compared with antibiotic‐free control after 1 hr incubation of meropenem (considered as 100%).

Figure 3.

Percentage of bacterial counts of susceptible isolate (K. pneumoniae ATCC 13883) with and without meropenem and resistant isolate (K. pneumoniae ATCC BAA 2146) with meropenem at concentration of breakpoint of sensitivity (1 mg/l).

In order to obtain the optimal cut‐off percentage of bacterial count for discriminating between susceptible and resistance, data obtained from isolates were analyzed by ROC curve (Fig. 4). Table 1 shows a complete sensitivity/specificity report and indicates the cut‐off value in bold.

Figure 4.

ROC curves for discriminating resistance from susceptible by flow cytometry according to established cut‐off (18.88%). Area under curve: 0.990; P < 0.001.

Table 1.

ROC Curve Analysis for Distinguishing Between Susceptible and Resistance Calculated by SPSS v.20.0 Software

Reduced percentage of bacterial counts in the cells compared the paired antibiotic‐free control	Sensitivity	Specificity
5.0577	0.931	0.000
5.2686	0.943	0.000
5.6011	0.954	0.000
6.3434	0.966	0.000
7.0593	0.977	0.000
18.8872	0.989	0.000
31.1173	0.989	0.011
32.9062	0.989	0.023
35.0892	0.989	0.034
36.3778	0.989	0.046
37.1577	0.989	0.057

This value allowed us to consider K. pneumoniae susceptible to meropenem by the FCM if the percentage of bacterial counts in the meropenem containing medium was reduced 18.8% or more compared to the number of K. pneumoniae in the paired antibiotic‐free control.

The susceptibility test performed by FCM for 174 K. pneumoniae isolates showed 99.4% concordance with the susceptibility results obtained by E‐test accepted as the gold standard. Only one discrepancy was found. The isolate was found susceptible by FCM method, but resistant by the E‐test. The carbapenems have slightly different characteristics. However, we carried out the susceptibility test for just meropenem. Since we found a correlation among meropenem, ertapenem, and imipenem susceptibilities results by E‐test, which give rise to thought flow cytometric results for the one of carbapenems give preliminary opinion for the others.

Discussion

K. pneumoniae is an important etiological agent for serious hospital—or community—acquired bacterial infections all around the world. Carbapenem antibiotics are the first line antibiotic class for the treatment of serious hospital‐acquired infection caused by K. pneumoniae 4, 16. Since its first detection in 1996 in North Carolina, CRKP has become an increasingly serious problem worldwide. Clinicians have very limited antibiotic options to use for treatment of infections, since CRKP is resistant to almost all of the antibiotics 17. Infections caused by CRKP increase in mortality, total hospital cost, and length of stay. Therefore, rapid identification of K. pneumoniae and detection of carbapenem resistance are critical steps for the infection control team to implement strict infection control practice 7, 17. However, detection of CRKP strains in the clinical laboratory remained a difficult task 18.

MALDI‐TOF MS method has been used as a new method for rapid and inexpensive identification of bacteria in routine clinical microbiology laboratories throughout the world. The system has been applied in the identification of microorganisms for the first time in 1975. MALDI–TOF MS has the potential to replace traditional phenotypic methods for identification of bacteria in most clinical bacteriology laboratory because it is rapid and inexpensive. Two commercially available MALDI‐TOF MS systems (Bruker MS [Bruker Daltonics] and VITEK MS [bioMérieux, Marcy l'Etoile, France]) have been developed to use in clinical microbiology laboratories 19. MALDI‐TOF MS assays have also been used for the rapid detection of carbapenemase activity, at a low cost. However, it is unlikely to be usable currently. The method should be investigated further and validated 10, 20, 21. There are only a few recent studies of the use of MALDI‐TOF MS for the rapid identification and typing of K. pneumoniae strains. They reported MALDI‐TOF MS was a rapid method to identify clinical K. pneumoniae strains 22, 23, 24. In the present study, all K. pneumoniae strains were identified with log score values >2.0 using the Bruker Biotyper software in MALTI‐TOF MS. Observed concordance between MALDI‐TOF MS and BD Phoenix^™ was 100%. Faster results were obtained in a few minutes from MALDI‐TOF MS rather than at least 4–6 hr from BD Phoenix^™.

Flow cytometry is a new simple and rapid approach in accurate assessment of the susceptibility of microorganisms to many antibiotics 12, 25, 26, 27, 28. Here, we presented a study to determine the susceptibility of K. pneumoniae strains to one of the carbapenem antibiotics, meropenem. The viable bacteria percentage obtained from the cells incubated with meropenem were compared, using ROC curves, to the viable bacterial percentage from paired antibiotic‐free control. It was evaluated that the viable bacterial percentage obtained from K. pneumoniae after 1 hr of incubation made it possible to define whether a K. pneumoniae isolate was resistant or susceptible to meropenem 11. This incubation time is similar to other susceptibility studies performed in Enterobacteriaceae 11, 26, 29. Based on the analysis of the ROC curves, the best validity and sensitivity data were obtained with a cut‐off value of 18.88%. The concordance, sensitivity, and specificity data for FCM vs. E‐test method using the selected cut‐off values were 99.4%, 98.9%, and 100%, respectively. Interestingly, we found only one very major error 30, with regard to the E‐test results accepted gold standard. FCM was repeated twice and the same result was found for this strain. This K. pneumoniae isolate would be reported as sensitive by FCM, although it is resistant to meropenem. We did not obtain any minor error. Similar findings have been reported by a few researchers. Nuding and Zabel performed a susceptibility test in plate‐grown colonies of 67 bacterial strains including K. pneumoniae by FCM. Ninety‐five percent of the FCM susceptibility results were concordant with results derived by the agar diffusion method and 94% with the results obtained with the WalkAway‐96 (Dade‐Behring, Marburg, Germany). The authors concluded that FCM offered a suitable technique for the rapid detection and susceptibility testing of bacteria with great potential 12. March et al. performed susceptibility test in Enterobacteriaceae showed 97% concordance between the results obtained by FCM and the results obtained by the commercial methods. They considered Enterobacteriaceae resistant to antibiotic tested if in order to obtain a bacterial count reduction of at least 14% compared to paired antibiotic‐free control after 1 hr of incubation. They concluded that they obtained reliable results on bacterial antibiotic‐susceptibility test and confirmed that FCM proposed method allows obtaining a rapid antibiotic‐susceptibility test 11.

In conclusion, Bruker MALDI‐TOF MS offers easy, rapid, and inexpensive system for identification of clinical relevant K. pneumoniae strains. FCM provides rapid, simple, and accurate results compared with conventional antibiotic‐susceptibility methods. Both methods together might easily be implemented to the clinical microbiology laboratory during daily routine. The implementation of MALDI‐TOF MS in combination with FCM can lead to decreased turnaround times for identification of clinically important K. pneumoniae strains and the detection of antibiotic susceptibility, which result in decreased mortality and morbidity associated with infections, as well spreading of infectious agent. According to our results, we also suggest that 18.88% on the analysis of the ROC curves may be accepted cut‐off value, and the dead ratio over this threshold should be evaluated as susceptible.