Detection of Carbapenem Resistance in Enterobacterales Directly From Positive Blood Cultures Using Matrix-Assisted Laser Desorption Ionization Time-of-Flight Mass Spectrometry

One hundred clinical Enterobacterales (Table), recovered between 2019 and 2021 from patients at a tertiary hospital in Porto Alegre city (South Brazil), were selected for this study. They were identified by MALDI-TOF MS (Bruker Daltonics). Susceptibility to meropenem (Sigma-Aldrich) was defined by broth microdilution (0.5–256 µg/mL), according to ISO (International Organization for Standardization) 20776-1,¹⁸  and isolates with minimal inhibitory concentration (MIC) above 8 µg/mL were considered resistant to meropenem.¹⁹  The presence of bla_KPC, bla_GES, bla_NDM-1, bla_IMP, bla_VIM, and bla_OXA-48-like was established by real-time multiplex polymerase chain reaction with high-resolution melting (HRM-qPCR), as previously described.²⁰  The distribution of species within our sample followed hospital epidemiology.

Bacteria were cultured on Mueller-Hinton agar and incubated overnight (35°C–37°C). To simulate blood culture–positive specimens, 1 mL of bacterial suspension with approximately 10⁷ CFUs (colony forming units) was inoculated into an aerobic blood culture bottle (BacT/ALERT FA Plus REF 410851, bioMérieux), previously inoculated with 4 mL of human blood. Thereby, a final bacterial concentration of approximately 2 × 10⁶ UFCs/mL per bottle was achieved. The vial was incubated in the automated system (BacT/ALERT), according to manufacturer’s instructions, until positivity was achieved.

For bacterial extraction, a tube with separator gel and sodium dodecyl sulfate (SDS) as lysis buffer was used.^21,22  First, 3 mL from a positive blood culture bottle was injected into a serum separator tube (BD Vacutainer Tubo SST REF 360060, Becton Dickinson) and centrifuged at 850g for 5 minutes to remove blood cells and debris at the bottom of tubes. The supernatant was discarded, and the bacterial pellet at the surface of the separation gel was carefully transferred into a 2-mL Eppendorf tube and washed twice by using 1 mL of water with 0.1% SDS (centrifuged at 18 620g for 2 minutes). Then, the pellet was resuspended with 10 µL of meropenem (Sigma-Aldrich) solution at 1 mg/mL in water with 0.01% SDS.^7,17  This suspension was incubated for 2 to 4 hours (35°C–37°C).

After incubation, the suspension was centrifuged for 2 minutes at 14 000 rpm. One microliter of the supernatant was spotted on the MALDI-TOF target plate (Bruker Daltonics).¹⁷  After drying, 1 µL of the matrix solution (α-cyano-4-hydroxy-cinnamic acid 10 mg/mL in 50% acetonitrile and 0.1% trifluoroacetic acid) was added and then left to dry at room temperature.⁷ 

A mass spectrum was acquired by the flexControl 3.4 software of the Microflex LT mass spectrometer (Bruker Daltonics) in the low mass-to-charge range (100–1000 Da) with 60-Hz laser frequency, laser power 25% to 35%, detector gain 3.2×, and 40 laser shots. Each spot was read once, and data were automatically acquired by using the MALDI Biotyper Selective Testing of Beta-lactamase Activity (MBT STAR-BL) method with autoXecute mode. The spectrometer was externally calibrated by using MBT STAR-ACS (Antibiotic Calibration Standard No. 1818702, Bruker Daltonics) for the appropriate mass range. Each isolate was evaluated in duplicate after 2, 3, and 4 hours of incubation (Figure 1).

Analysis of mass spectrum was manually performed by flexAnalysis 3.4 (Bruker Daltonics), searching for peaks of nonhydrolyzed and hydrolyzed forms of meropenem. The intensities of the respective peaks were used to calculate logRQ (measure of hydrolytic efficiency), which is the logarithm of the ratio of the sum of the intensity of the decarboxylated and hydrolyzed forms at 358 m/z and the hydrolyzed form at 402 m/z and the summed intensity of the intact forms (peak [M+H]⁺ at 384 m/z and sodium [M+Na]⁺ at 406 m/z).^7,17 

Since both peaks of meropenem and its hydrolyzed, decarboxylated forms may be observed in the mass spectra of meropenem solutions incubated with carbapenemase or non-carbapenemase producers, a cutoff value of logRQ, previously established,^7,23  was used to distinguish carbapenemase producers from nonproducer bacteria. Thereby, logRQ values greater than 0.4 or less than 0.2 indicate positive or negative results, respectively, for the hydrolysis of meropenem. Values between 0.2 and 0.4 were considered indeterminate.

Enterobacterales included in the study were as follows: Klebsiella pneumoniae (n = 53), Enterobacter cloacae complex (n = 22), Serratia marcescens (n = 12), Klebsiella oxytoca (n = 3), Klebsiella aerogenes (n = 3), Escherichia coli (n = 3), Citrobacter freundii (n = 2), Morganella morganii (n = 1), and Providencia stuartii (n = 1). According to broth microdilution, 68.0% (68 of 100) were resistant to meropenem (MIC > 8 µg/mL), 12.0% (12 of 100) were susceptible (MIC ≤ 2 µg/mL), and 20.0% (20 of 100) were categorized as susceptible, increased exposure (4–8 µg/mL).

Based on HRM-qPCR (Table), 81 of 100 isolates carried a carbapenemase gene: 49 bla_KPC, 18 bla_NDM-1, 10 bla_OXA-48-like, 3 bla_KPC + bla_NDM-1, and 1 bla_OXA-48-like + bla_NDM-1. Of these, 81.5% (66 of 81) had positive results for hydrolysis, with 74.2% (49 of 66) presenting results within 2 hours of incubation. Spectra from 12.3% (10 of 81) of carbapenemase producers did not demonstrate hydrolysis within 4 hours (false negatives): 1 bla_KPC (E cloacae complex; MIC = 2 µg/mL; susceptible), 3 bla_OXA-48-like (E cloacae complex; MIC = 8 µg/mL; susceptible, increased exposure), and 6 bla_NDM-1. The bla_NDM-1 isolates were identified as E cloacae complex (n = 2; MIC = 8 µg/mL; susceptible, increased exposure), K pneumoniae (n = 2; MIC = 0.5 and 2 µg/mL; susceptible), M morganii (n = 1; MIC = 1 µg/mL; susceptible), and P stuartii (n = 1; MIC = 4 µg/mL; susceptible, increased exposure).

The remaining 5 isolates (6.2%; 5 of 81) carrying a carbapenemase gene (2 bla_KPC, 2 bla_NDM-1, and 1 bla_OXA-48-like) had indeterminate results with the MALDI-TOF–based hydrolysis assay. Of note, although the bla_NDM-1-positive and bla_OXA-48-like–positive isolates with indeterminate logRQ values had MIC values of 8 µg/mL or lower, both Klebsiella pneumoniae carbapenemase (KPC)–positive isolates had high MIC values, that is, 32 µg/mL and 128 µg/mL. All discrepant results were repeated and confirmed; therefore, the reasons for the false-negative and indeterminate hydrolysis results remain to be elucidated. These results are detailed in the Table.

Among the 19 isolates with negative results in HRM-qPCR (Table), 89.5% (n = 17) did not show hydrolysis at any evaluated time. Two isolates presented false-positive results (logRQ > 0.4); both were K oxytoca isolates with MIC of 4 µg/mL (susceptible, increased exposure).

Overall, sensitivity and specificity were 86.8% (66 of 76, excluding 5 indeterminate results, as sensitivity is calculated from dichotomic variables) and 89.5% (17 of 19), respectively. Stratifying by enzyme type, and, again, excluding the 5 indeterminate results, data were consistently distinct. For KPC-producing isolates (n = 47), 46 demonstrated hydrolysis (sensitivity of 97.9%), with most results (84.8%; 39 of 46) observed after 2 hours of incubation. Sensitivity decreases significantly when bla_NDM-1 (62.5%; 10 of 16) and bla_OXA-48-like (66.7%; 6 of 9) are taken into consideration. The incubation time required for each enzyme also varied significantly, with only 50.0% and 33.3% of bla_NDM-1 and bla_OXA-48-like positive isolates observed after 2 hours of incubation. No false-negative results were observed for coproducer isolates, and meropenem hydrolysis occurred quickly (2 hours) for the 3 bla_KPC + bla_NDM-1 isolates. On the other hand, although the bla_NDM-1 + bla_OXA-48-like coproducer was also correctly detected by the hydrolysis assay, it was a time-consuming reaction, requiring 4 hours of incubation.

In addition to rapid hydrolysis, bla_KPC-positive isolates also showed a higher logRQ mean (Figure 2) than bla_NDM-1 and bla_OXA-48-like, corroborating the sensitivity results. LogRQ ranged from 0.37 to 1.97 in KPC-producing isolates, except for 1 false-negative isolate with logRQ of −0.59 and MIC of 2 µg/mL. For isolates carrying bla_NDM-1, logRQ ranged from −1.37 to 0.83, while for bla_OXA-48-like, the range was −1.08 to 1.79.

MALDI-TOF MS has become an important tool in the clinical microbiology laboratory, and its application for detection of resistance mechanisms, through hydrolysis assay, has been extensively explored.^{12–14,24–27} 

Jung et al¹⁷  proposed a logarithm of hydrolyzed/nonhydrolyzed peaks (logRQ) to quantify the hydrolysis of third-generation cephalosporins for the detection of extended-spectrum β-lactamases, discriminating enzyme producers and nonproducers in 2.5 hours (after establishing the positivity of blood cultures) with 100.0% sensitivity and 91.5% specificity.

Subsequently, several studies were carried out to detect resistance to carbapenems, with similar results. Carvalhaes et al¹⁵  reported for the first time that MALDI-TOF MS could detect carbapenemase activity (ertapenem as substrate) directly from positive blood cultures for all KPC-2–producing and SPM-1–producing isolates. Ghebremedhin et al¹⁶  demonstrated that this method allowed the detection of carbapenemases in 98% of Pseudomonas and Enterobacteriaceae isolates harboring different genes. However, they relied solely on the disappearance of the imipenem + matrix peak to consider an isolate as a carbapenemase producer. More recently, Yu et al¹⁴  reported that CRE strains were well distinguished by MALDI-TOF MS–based ertapenem hydrolysis assay, with a sensitivity of 92.5% and specificity of 100%, using logRQ as parameter.

In our experience, compared with HRM-qPCR, the hydrolysis assay using meropenem as substrate had a lower sensitivity (86.8%) and specificity (89.5%) than observed with other substrates (ie, imipenem and ertapenem). For example, all studies had 100% sensitivity and specificity when using ertapenem to evaluate carbapenem susceptibility in Enterobacterales, except for 1 study that reported 92.5% specificity.^11,13,14  Moreover, excellent results were also found when using imipenem (98% sensibility and 100% specificity).⁹  Of note, when only KPC producers were evaluated, the sensitivity increased significantly, to 97.9%.

The KPC activity was detected within 2 hours in 100.0% of the evaluated isolates, while other enzymes, such as OXA-48–like and New Delhi metalobetalactamase (NDM), required longer incubation times (3 hours) and had a slightly low sensitivity.^{7,11,13,15,28}  We also observed an important relationship between MIC and logRQ values, proving that logRQ can be a good indicator of hydrolysis. Isolates resistant to meropenem (MIC >8 µg/mL) had a logRQ media value of 0.80, while susceptible (MIC ≤2 µg/mL) isolates had a logRQ median value of −0.22 (Figure 2).

A limitation of the study was the low number of isolates resistant to carbapenems through NDM or OXA-48 producers. False-negative results (MIC ≤0.5–8 µg/mL) may be a consequence of low enzyme expression or of reduced hydrolytic capacity of such enzymes, which is well recognized, mainly for OXA-48–like.^29,30  On the other hand, a positive hydrolysis result without an HRM-qPCR–positive finding may be justified by the presence of another carbapenemase gene, other than those included in the reaction, which is another limitation of this study. The incubation time required for each enzyme also varied, in agreement with previous studies.^7,14,15 

Considering turnaround time, the detection of carbapenem hydrolysis by MALDI-TOF MS is comparable to that of other rapid methods, such as Carba MP (bioMérieux), β-CARBA (Bio-Rad), NeoRapid CARB (Rosco Diagnostica), and carbapenem inactivation method (CIM) adapted for blood cultures. Sensitivity of these assays is between 97.9% and 100%, and the specificity ranges from 91.4% to 100%.^31–33  As with the hydrolysis test, to achieve better results, it is necessary to concentrate the bacteria and hemolyze the erythrocytes to avoid interference in the test results.

Other methodologies are available, such as PCR, immunochromatography test (ICT), or EUCAST rapid antimicrobial susceptibility testing. With the last method listed, resistance to carbapenems can be detected within 4 to 8 hours for E coli and K pneumoniae; however, breakpoints are not yet available for other Enterobacterales species.^34–36  Molecular assays or ICTs detect specific carbapenemases,^35,37  whereas the MALDI-TOF hydrolysis assay detects carbapenemase activity overall.

The main advantage of the MALDI-TOF MS carbapenemase assay is that it is an inexpensive and rapid test to detect carbapenemase activity directly from blood culture vials, contributing to faster adjustment of empirical antimicrobial therapy and implementation of infection control measures. Furthermore, we emphasize that the decision to perform any assay depends on the epidemiology and individual risk factors. In countries with a high prevalence of carbapenemases, additional information on carbapenemase gene production is important and impacts treatment. On the other hand, in a low endemicity setting, testing for specific carbapenemases may not be necessary and may be reserved only for patients with risk factors, such as patients with CRE colonization or those who have been transferred from a high-risk region.

In addition, it is important to highlight that the hydrolysis assay for carbapenemase detection using MALDI-TOF MS has been performed, so far, for research purposes only, so there are barriers to its implementation in the routine setting of clinical laboratories.

In conclusion, MALDI-TOF MS–based hydrolysis assay is a fast, inexpensive, and easy-to-perform method to detect carbapenemase activity; however, it still needs some improvements. This methodology, combined with other already well-established ones, can be of great value in screening for hydrolysis of carbapenems.