Matrix-Assisted Laser Desorption Ionization–Time of Flight Mass Spectrometry for Antimicrobial Susceptibility Testing

ABSTRACT

The advent of matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) in clinical microbiology has dramatically improved the accuracy and speed of diagnostics. However, this progress has mainly been limited to the identification of microorganisms, whereas the practical improvement of antimicrobial susceptibility testing (AST) still lags behind. MALDI-TOF MS-based approaches include the detection of selected resistance mechanisms and the universal phenotypic AST. This minireview focuses on the discussion of those MALDI-TOF MS methods that allow universal growth-based phenotypic AST. The method of minimal profile change concentrations (MPCC) is based on detecting proteome modification in the presence of an antimicrobial. Using stable-isotope labeling, characteristic mass shifts in the presence of an antimicrobial indicate the incorporation of the isotopic labels, and thus the viability and resistance of the microorganism. For MALDI Biotyper antibiotic susceptibility test rapid assay (MBT-ASTRA), microorganisms are incubated with or without an antimicrobial, followed by cell lysis, protein extraction, and transfer of the cell lysate onto a MALDI target plate. Using the internal standard, peak intensities are correlated to the amount of microbial proteins, and the relative microbial growth is calculated. Most recent development in the field is the direct-on-target microdroplet growth assay (DOT-MGA). Here, incubation of microorganisms with antimicrobials takes place directly on spots of a MALDI target in the form of microdroplets. After incubation, nutrient medium is removed by dabbing with absorptive material. Resistant microorganisms grow despite the presence of antimicrobial, and their amplified biomass is detected by MALDI-TOF MS. Finally, an outlook is provided for further assay improvements.

INTRODUCTION

The introduction of matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) has dramatically changed the routine of clinical microbiology over the past decade (1). It has been praised as revolution or paradigm shift (2, 3), and indeed, it has allowed a major breakthrough in diagnostic laboratories (1–3). Both accuracy and speed of diagnostics have been considerably improved (1).

However, the MALDI-TOF MS-caused progress has mainly been limited to the identification of bacteria and fungi, whereas the practical improvement of another important function of a microbiological laboratory, namely, antimicrobial susceptibility testing (AST), still lags behind (4). Many academic groups and industry researchers have suggested innovative solutions to accelerate AST (4, 5) Among these approaches, efforts have been made to enable detection of microbial resistance or susceptibility by MALDI-TOF MS (6, 7). These MALDI-TOF MS-based approaches can be divided into two principal groups: (i) detection of selected resistance mechanisms and (ii) universal phenotypic AST (1, 6, 7) (Fig. 1). Defined resistance traits, e.g., carbapenemase production, can be detected very rapidly by MALDI-TOF MS (1, 6, 7). However, this approach is severely limited by the fact that a negative result doesn’t necessarily imply susceptibility toward the antimicrobial in question (4). Alternative resistance mechanisms that are not detectable by this approach can render the microorganism resistant. One of the typical examples is decreased permeability of Gram-negatives, which can lead to phenotypic resistance even in the absence of carbapenemase production (4, 8, 9). The MALDI-TOF MS-based approaches for the detection of selected resistance mechanisms have been extensively reviewed elsewhere, and we refer readers to the recent articles that comprehensively address these methods (6, 7). Here, the present review will only focus on MALDI-TOF MS-based approaches that allow universal growth-based phenotypic AST (Table 1).

FIG 1.

Main approaches to the determination of antimicrobial susceptibility and resistance using MALDI-TOF MS.

TABLE 1.

MALDI-TOF MS growth-based phenotypic AST methods

Method and main characteristics	Accuracy as determined by studies
Minimal profile change concentration (MPCC)
Principle: Change of the microbial mass spectrum in the presence of an antimicrobial. The MPCC, a lowest drug concentration resulting in mass spectrum profile change, correlates with the MIC. Advantages: Correlation between MPCC and MIC. Limitations: Long incubation time required; need for the  time-consuming protein extraction from microbial cell pellets.	C. albicans (n = 17) vs fluconazole: Essential agreement of MPCC and MIC within ±1 dilution: 94%; essential agreement of MPCC and MIC within ±2 dilutions: 100%; categorical agreement in all isolates except one (14).
	Candida spp. (n = 34) and Aspergillus spp. (n = 10) vs caspofungin: Essential agreement of MPCC and MIC within ±2 dilutions: 100%. For Candida spp. (n = 34): categorical agreement 94.1% (15).
	C. albicans (n = 62): Correct classification as susceptible or resistant towards caspofungin 98.4%, using the presence or absence of FKS1 hot spot mutations as gold standard (16).
	C. glabrata (n = 80) vs anidulafungin: 100% of isolates in the susceptible category were classified as susceptible; 50.0% of isolates in the nonsusceptible (intermediate or resistant) category were classified as resistant. C. glabrata (n = 80) vs fluconazole: 97.6% of isolates in the susceptible dose-dependent category were classified as susceptible; 94.9% of isolates in the resistant category were classified as resistant (17).
	Essential agreement for: C. albicans (n = 35) vs fluconazole, voriconazole, and posaconazole: 82.9%, 74.3%, and 54.3%, respectively; C. glabrata (n = 35) vs fluconazole, voriconazole, and posaconazole: 77.1%, 60.0%, and 97.1%, respectively; C. tropicalis (n = 37) vs fluconazole, voriconazole, and posaconazole: 54.1%, 64.9%, and 78.4%, respectively (18).
	Aspergillus spp. (n = 20) vs voriconazole: At 24 h incubation, appropriate classification of A. fumigatus WT strains as WT. For the Cyp51A mutants, 2 of 4 mutant strains displaying non-WT BMD MICs were assigned as WT by the MALDI-TOF MS assay (two very major errors); accurate discrimination between WT and non-WT at incubation times of 30 h and 48 h (19).
Stable-isotope labeling
Principle: Mass spectrum of a microorganism cultured in the stable isotope-labeled medium in the presence of an antimicrobial is compared with a mass spectrum of the microorganism cultured in the non-labeled medium without the antimicrobial. Characteristic mass shifts in the presence of an antimicrobial indicate incorporation of the isotopic labels, and thus the viability and the resistance of the microorganism. Advantages: Short incubation time (3 h); applicable to a broad range of organism/antimicrobial combinations Limitations: Incorporation of the labeled lysine in susceptible isolates despite the presence of some antimicrobials; lack of correlation with MIC; need for specific culture medium without the “marker” amino acid; need for the time-consuming protein extraction from microbial cell pellets.	Exemplary testing of E. coli vs streptomycin; no. of isolates not described (20).
	Differentiation between MSSA and MRSA in S. aureus (n = 48): correct classification of all isolates but one (23).
	P. aeruginosa (n = 30) vs ciprofloxacin, tobramycin, and meropenem: correct interpretation as susceptible or resistant in all isolates (24).
MALDI Biotyper antibiotic susceptibility test rapid assay (MBT-ASTRA)
Principle: Microorganisms are incubated with or without an antimicrobial, followed by the cell lysis, protein extraction, and transfer of the cell lysate onto a MALDI target plate. Using the internal standard, peak intensities are correlated to the amount of microbial proteins. The relative microbial growth with and without an antimicrobial is calculated as the ratio of the areas under the curve. Advantages: Short incubation time (1-4 h for common bacteria). Limitations: Time-consuming workflow with laborious steps of centrifugation, washing, lysis, and protein extraction; every single combination of species and antimicrobial requires optimization regarding the incubation time and the concentration of an antimicrobial.	Detection of meropenem resistance in Klebsiella spp. from agar cultures (n = 108): sensitivity of 97.3% and specificity of 93.5%. Detection of meropenem resistance in Klebsiella spp. from spiked blood cultures (n = 18): correct classification of all isolates but one after 1 h of incubation, correct classification of all isolates after 2 h of incubation (25).
	Detection of nonsusceptibility against gentamicin and ciprofloxacin in Enterobacteriaceae from spiked blood cultures (n = 30) and detection of nonsusceptibility against cefotaxime, piperacillin-tazobactam, and ciprofloxacin in Gram-negative rods (n = 99) from patient-derived blood cultures: correct classification of all isolates for nonsusceptibility against gentamicin and cefotaxime. One misclassification for ciprofloxacin and five misclassifications for piperacillin-tazobactam (26).
	E. coli, K. pneumoniae, P. aeruginosa, Acinetobacter baumannii vs penicillins, cephalosporins, carbapenems, fluoroquinolones, and aminoglycosides: accuracy varied depending on species/antimicrobial combination and assay conditions (27).
	E. coli (n = 103) from positive blood cultures vs amoxicillin (categorical agreement 97%) and cefotaxime (categorical agreement 83%); The study failed to find optimal testing conditions for piperacillin-tazobactam (28).
	E. coli and K. pneumoniae (898 blood cultures spiked with 14 reference strains) vs cefotaxime, meropenem and ciprofloxacin: overall categorical agreement of 97% (strains of intermediate category were excluded from the data analysis) (29).
	S. aureus from agar cultures (n = 35) vs ciprofloxacin, oxacillin, cefepime, and vancomycin: overall accuracy rate of 95% (applying the optimized assay conditions). S. aureus from spiked blood cultures (n = 4) vs same four antimicrobials: accuracy of 96% (30).
	M. tuberculosis (n = 39) vs rifampin, isoniazid, linezolid, and ethambutol: 100% agreement. Nontuberculous mycobacteria (n = 33) vs clarithromycin and rifabutin: 98.5% agreement (31).
	B. fragilis (n = 2) vs clindamycin, meropenem, and metronidazole: correct classification as susceptible or resistant (32).
	Detection of tetracycline resistance in P. multocida (n = 100): sensitivity 95.7% and specificity of 100% (33).
	Detection of caspofungin resistance in C. albicans (n = 58): validity (successful growth control detection) 88%, sensitivity 100%, specificity 100%. Detection of caspofungin resistance in C. glabrata (n = 57): validity 95%, sensitivity 94%, specificity 80% (34).
	Detection of anidulafungin resistance in C. glabrata (n = 100) directly from positive blood cultures: validity 98%, sensitivity 80%, specificity 95% (35).
	Resistance detection of C. auris from agar cultures (n = 50) to caspofungin (sensitivity 100%, specificity 73%), anidulafungin (sensitivity 100%, specificity 98%) and micafungin (sensitivity 100%, specificity 95.5%) Resistance detection of C. auris from positive blood cultures (n = 20) to caspofungin (categorical agreement 90%), anidulafungin (categorical agreement 100%), and micafungin (categorical agreement 100%) (36).
Direct-on-target microdroplet growth assay (DOT-MGA)
Principle: The method is based on broth microdilution in that incubation of microorganisms with or without antimicrobials takes place directly on spots of a MALDI target in form of microdroplets. After incubation, medium is removed by dabbing with an absorptive material prior to biomass measurement by MALDI-TOF MS. Advantages: Rapid method; relatively close to CLSI and ISO standards; easy-to-perform; applicable to a broad range of organism/antimicrobial combinations; can be combined with identification; capability for automation (high-throughput feasible); expandable to further applications, e.g., susceptibility determination directly from clinical samples, and simultaneous testing of multiple antimicrobials. Limitations: Longer incubation times may be required for slow-growing species; thorough standardization of assay conditions is crucial (humidity, medium removal, MALDI-TOF MS settings, internal control, spectra analysis).	Detection of meropenem non-susceptibility in K. pneumoniae (n = 24) from agar cultures: validity 100%, sensitivity 100%, specificity 100% after 4 h using 6-μl droplets. Detection of meropenem non-susceptibility in P. aeruginosa (n = 24) from agar cultures: validity 83.3%, sensitivity 100%, specificity 100% after 5 h using 6-μl droplets (37).
	Detection of meropenem non-susceptibility in Enterobacterales (n = 28) from blood cultures: validity 96.3%, sensitivity 91.7%, specificity 100% applying the lysis/centrifugation method after a 4-h incubation (41).
	Detection of ESBL and AmpC β-lactamases in Enterobacterales (n = 50): positive and negative percent agreement values for ESBL 94.4%/100%, for AmpC 94.4%/93.8%, for ESBL+AmpC 100%/100%, compared to PCR (38).
	Differential detection of carbapenemases in Enterobacterales (7 reference strains and 20 clinical isolates): correct identification of KPC, MBL, and OXA in 100% agreement with PCR after 4-h incubation (39).
	Detection of methicillin resistance in S. aureus (n = 28): From agar cultures after 5-h incubation: validity 100%, sensitivity 100%, specificity 100%. Directly by lysis/centrifugation method from positive blood cultures after 4-h incubation: validity 96.4%, sensitivity 100%, specificity 100% (42).
	E. coli (n = 40) vs ceftriaxone: validity 82.5%, sensitivity 90%, specificity 100%. K. pneumoniae (n = 40) vs meropenem: validity 100%, sensitivity 95%, specificity 95%. S. aureus (n = 40) vs oxacillin: validity 95.8%, sensitivity 100%, specificity 100% (40).

A universal growth-based phenotypic AST allows that potentially every antimicrobial substance can be tested against the vast majority of microorganisms because in this case the determination of susceptibility or resistance toward an antimicrobial is independent from the underlying resistance mechanisms. Such phenotypic AST is currently the standard method and the subject of international guidelines on broth microdilution, disk diffusion, and agar dilution methods issued by the International Organization for Standardization (ISO), the Clinical and Laboratory Standards Institute (CLSI), and the European Committee on Antimicrobial Susceptibility Testing (EUCAST) (10–12).

Since these reference methods for phenotypic AST (10–12) as well as its multiple modifications in commercial assays (5) have already existed for decades, their standard conditions (inoculum size, nutrient medium, incubation temperature) could be beneficially used for MALDI-TOF MS-based universal AST approaches. However, there is a need in acceleration of AST (4). Even today, the time to result is still too long so that usually the finding report is only available on the next day (5). This applies especially to the laboratories that do not operate 24/7. Notably, 24/7 service is not provided by the majority of microbiological laboratories in Europe, according to the recent survey (13). Therefore, the “same-day” or “same-shift” AST report has not yet become reality in routine laboratories, at least in Europe, and it remains to be achieved as first priority. Furthermore, a combination of both tasks—identification and AST—on the same platform is highly desirable. An example of additional advantages is the possibility to better control for contamination in an AST set-up, because grown microorganisms can automatically be identified by MALDI-TOF MS after the assay’s incubation time, when the number of bacteria exceeds the lower level of MALDI-TOF MS detection. This particular feature of microorganism identification during the MALDI-TOF MS-based AST can also reduce the possibility of organism mix-ups in the diagnostic process.

MINIMAL PROFILE CHANGE CONCENTRATION

More than a decade ago, Marinach et al. investigated a susceptibility assay based on detecting proteome modification in cells grown with varying concentrations of an antimicrobial (14). As an example, they focused on determination of fluconazole susceptibility in Candida albicans. The yeast cells were cultured for 15 h in a microtiter plate, followed by the protein extraction from cell pellets and deposition of supernatants onto a MALDI plate. The authors observed that MALDI-TOF mass spectra of a sample with low fluconazole concentration were indistinguishable from those of the control culture. In contrast, the spectra obtained from a sample exposed to higher fluconazole concentration, considerably deviated from the control. This led the authors to establish the term minimal profile change concentration (MPCC), which is a lowest drug concentration resulting in mass spectrum profile change. While the authors assume that characteristic profiles might be associated with particular classes of antimicrobials, they also point out that the diagnostic profile shift didn’t vary with the level or mechanism of fluconazole resistance. Noteworthy, the concept of MPCC is similar to the definition of the MIC (10, 11), and indeed, there was a clear correlation between MPCC measured with MALDI-TOF MS and classical MIC determined by broth microdilution in this study (14). We therefore assume that the profile differences observed by the authors might not be caused by a specific proteome modification in response to a particular antifungal substance, but rather reflect the growth or growth inhibition of the microorganism. We therefore assume that the assay suggested by the authors may be classified as the universal growth-based phenotypic AST. Since only one antifungal was used in this study and the nature of molecules modified by fluconazole exposure was not explored in the study, this assumption remains unresolved. The study showed accurate results in good agreement with the reference method (14).

De Carolis et al. modified the protocol of Marinach et al. (14) for testing caspofungin against Candida and Aspergillus species (15). In this work, the composite correlation index (CCI), a statistical approach for analyzing the relationships between spectra, was used to compare spectra for the determination of MPCC (15). In the following study, the authors adapted the method in that shorter incubation time was applied to test breakpoint caspofungin concentrations against Candida albicans (16). Such a short exposure for only 3 h to anidulafungin was, however, not sufficient to detect echinocandin-resistant Candida glabrata isolates with FKS2 mutations (17). Another group applied similar methodology for detecting resistance to fluconazole, voriconazole, and posaconazole among Candida spp. (18) Although some promising results were observed in that study, the agreement with reference method and the reproducibility were suboptimal, and there was no significant time saving over the reference test (18). The study of Gitman et al. showed accurate detection of Aspergillus fumigatus strains with reduced voriconazole susceptibility by use of a similar CCI-based MALDI-TOF MS approach (19). However, the authors concluded that no advantages were seen with this particular method over conventional testing methods (19).

STABLE-ISOTOPE LABELING

Demirev at al. described an approach (20) where mass spectrum of a bacterium cultured in the stable isotope-labeled medium with an antimicrobial is compared with a mass spectrum of the organism cultured in the non-labeled medium without the antimicrobial. Thereby, characteristic mass shifts in the presence of an antimicrobial indicate incorporation of the isotopic labels and, hence, the viability of the microorganism and its resistance to the antimicrobial. This method represents an adaptation of the stable isotope labeling by amino acids in cell culture (SILAC) for the determination of antimicrobial resistance. SILAC is a technique commonly used in quantitative proteomics (21, 22).

Sparbier et al. suggested a similar method for rapid differentiation of resistant and susceptible bacteria by MALDI-TOF MS profiling of organisms grown in media containing isotopically labeled amino acids (23). In contrast to the study of Demirev at al., which applied completely ¹³C-labeled medium resulting in multiple peak shifts with various mass differences (20), Sparbier et al. employed the incorporation of lysine as a single isotopically labeled amino acid. Resistant bacteria grow in the presence of antimicrobials and incorporate “heavy” amino acids, resulting in an increase of protein masses and corresponding shifts in mass spectra. The applicability of this approach was demonstrated for the differentiation of methicillin-resistant Staphylococcus aureus (MRSA) from methicillin-susceptible S. aureus (MSSA). Briefly, the bacterial suspensions were incubated in media containing “normal” lysine or “heavy” lysine, as well as “heavy” lysine with oxacillin or cefoxitin, for 3 h at 37°C under agitation. After incubation, centrifugation and wash steps followed, and the bacterial cells were lysed by formic acid and acetonitrile. The cell lysates were spotted onto a MALDI target plate, overlaid with HCCA matrix, and measured by a Microflex LT/SH MALDI-TOF MS instrument (Bruker Daltonik, Bremen, Germany). The subsequent sophisticated spectrum analysis showed that an incubation time of 3 h was sufficient to discriminate between susceptible and resistant isolates (23).

In the work of Jung et al., the feasibility investigations of this technique were expanded to resistance toward meropenem, tobramycin, and ciprofloxacin in Pseudomonas aeruginosa (24). The authors reported the most rapid results obtained with tobramycin within 90 min. It was however observed that more time was required to distinguish between resistance and susceptibility to meropenem, because a significant incorporation of the labeled lysine occurred even in meropenem-susceptible isolates. To minimize this effect, the authors chose to add the labeled lysine only after a pre-incubation time of 30 min, which allowed the antimicrobial to act (24).

Principally, stable-isotope labeling assay may be applicable to a broad range of organism/antimicrobial combinations, although some adaptations in test conditions such as incubation time, culture medium, and analysis algorithms may be necessary. The technical limitations of this method include incorporation of the labeled lysine in susceptible isolates despite the presence of some antimicrobials, lack of correlation with MICs, and the need for specific culture medium without the “marker” amino acid (24). An important practical drawback is the need for complex sample processing with considerable hands-on time for centrifugation, washing, lysis, and protein extraction steps (23, 24).

MALDI BIOTYPER ANTIBIOTIC SUSCEPTIBILITY TEST RAPID ASSAY (MBT-ASTRA)

In 2014, Lange et al. suggested a MALDI-TOF MS method that employs internal standard for the quantitative measurement of microbial proteins, and therefore allows growth analysis in the presence or absence of an antimicrobial (25). This method was named MALDI Biotyper antibiotic susceptibility test rapid assay (MBT-ASTRA). Briefly, bacteria were incubated in brain heart infusion (BHI) medium with or without an antimicrobial, followed by the cell lysis and protein extraction with 70% formic acid and 100% acetonitrile. RNase B was added as an internal standard to facilitate the quantitative comparison of the acquired mass spectra. One μl of the cell lysate was spotted onto a polished steel MALDI target plate, overlaid with HCCA matrix, and measured with a Microflex LT/SH mass spectrometer (Bruker Daltonik). The spectra were normalized to the peak with the highest intensity, and the automated algorithms were used to correlate the peak intensities to the amount of bacterial proteins and to calculate the relative growth as the ratio of the areas under the curve (AUC) of bacteria incubated with and without an antimicrobial. In this study with 108 Klebsiella spp. isolates, a sensitivity of 97.3% and a specificity of 93.5% for the detection of meropenem resistance were achieved after 1 h of incubation (25).

Additionally, the applicability of the method was tested for direct AST from positive blood cultures. Blood culture bottles were spiked with blood and Klebsiella pneumoniae isolates, and incubated in an automated blood culture instrument. Upon positivity, a sample of blood culture broth was centrifuged, and the pellet resuspended in BHI. Subsequently, this suspension was incubated with and without meropenem for 1 h at 37°C under agitation. After incubation, the remaining human blood cells were removed by lysis, centrifugation, and washing. After that, protein extraction, measurement, and calculations were performed as described above. After 1 h of incubation, 17 of the 18 isolates were correctly classified, whereas 2 h of incubation were necessary for the sufficient growth and valid classification of one remaining isolate (25).

A follow-up study of Jung et al. expanded the investigation of MBT-ASTRA methodology directly from positive blood cultures with Gram-negative bacteria (26). The authors tested 30 spiked blood cultures for nonsusceptibility against gentamicin and ciprofloxacin and 99 patient-derived blood cultures for nonsusceptibility against cefotaxime, piperacillin-tazobactam, and ciprofloxacin. The study showed promising results for the prediction of nonsusceptibility within approximately 4 h. The highest number of misclassifications was observed for piperacillin-tazobactam (26).

Sparbier et al. evaluated the MBT-ASTRA approach for determination of susceptibility/resistance status for various antimicrobial classes in combination with different species (27). Clinical isolates of Escherichia coli, K. pneumoniae, P. aeruginosa, and Acinetobacter baumannii were tested against penicillins, cephalosporins, carbapenems, fluoroquinolones, and aminoglycosides. The authors concluded that reliable information on resistance or susceptibility was achieved after a few hours of incubation for most combinations. However, testing piperacillin versus E. coli and ceftazidime versus P. aeruginosa was difficult, and the need for assay optimization was pointed out. Generally, each single combination required optimization regarding the incubation time and the concentration of an antimicrobial (27).

Further studies on MBT-ASTRA have also focused on direct AST from positive blood cultures. Sauget et al. applied this method to test 103 positive blood cultures positive with Escherichia coli against amoxicillin and cefotaxime (28). In this study, positive blood cultures were additionally pre-incubated in a pre-warmed BHI broth for 1 h at 37°C under agitation before further incubation with an antimicrobial. The latter took 2.5 h for amoxicillin and 2 h for cefotaxime, followed by centrifugation and washing of a sample, as well as extraction of bacterial proteins. Thus, the total time of the assay was approximately 4 h. The categorical agreement between MBT-ASTRA method and the reference method was 97% and 83% for amoxicillin and cefotaxime, respectively. The attempts to find optimal testing conditions for piperacillin-tazobactam with MBT-ASTRA failed in this study (28). Axelsson et al. applied an automated and optimized MBT-ASTRA protocol for rapid detection of antimicrobial resistance in positive blood cultures for E. coli and K. pneumoniae (29). They tested 898 blood cultures spiked with 14 reference strains and reported an overall accuracy of 97% for cefotaxime, meropenem, and ciprofloxacin, while the intermediate reference strains were excluded from the data evaluation (29).

Several publications reported adaptations of MBT-ASTRA for testing different microorganisms. Maxson et al. reported testing of S. aureus against oxacillin, cefepime, ciprofloxacin, and vancomycin, including an application for spiked blood cultures (30). Ceyssens et al. evaluated MBT-ASTRA for Mycobacterium tuberculosis (rifampin, isoniazid, linezolid, and ethambutol) and nontuberculous mycobacteria (clarithromycin and rifabutin) (31). Justesen et al. performed a proof-of-concept for testing an anaerobic bacterium Bacteroides fragilis versus clindamycin, meropenem, and metronidazole (32). In a study of van Driessche, Pasteurella multocida was subjected to susceptibility testing toward tetracycline (33). Other studies sought to investigate MBT-ASTRA for antifungal susceptibility testing. This approach was applied to detect caspofungin resistance in Candida albicans and Candida glabrata (34), anidulafungin resistance in C. glabrata directly from positive blood cultures (35), as well as resistance of Candida auris to caspofungin, anidulafungin, and micafungin from agar cultures and positive blood cultures (36).

A general limitation and an important hurdle for the practical implementation of the MBT-ASTRA approach is its time-consuming workflow with significant workload for centrifugation, washing, lysis, and protein extraction. This also substantially limits the potential for automation.

DIRECT-ON-TARGET MICRODROPLET GROWTH ASSAY (DOT-MGA)

While the studies mentioned above have described some approaches to use MALDI-TOF MS for a universal growth-based phenotypic AST, there remained a need for a practical method with enhanced potential for routine implementation (4–7). Idelevich et al. recently suggested a novel MALDI-TOF MS-based AST method designated as direct-on-target microdroplet growth assay (DOT-MGA) (37). This method is based on broth microdilution, and therefore principally similar to the reference methods recommended by CLSI (10) and ISO (11) (Fig. 2). Furthermore, test conditions are kept as close as possible to the recommendations of these guidelines. These adopted conditions include standardization of inoculum size to approximately 5 × 10⁵ CFU/ml, implementation of cation-adjusted Mueller–Hinton broth (CA-MHB), as well as incubation at 35 ± 1°C in ambient air (37).

FIG 2.

Principle of the MALDI-TOF MS-based direct-on-target microdroplet growth assay (DOT-MGA).

The fundamental difference of the novel method is that incubation of microorganisms with antimicrobials (or as a growth control without antimicrobials) does not need an additional microtiter plate or another incubation container, because it is performed directly on spots of a MALDI target in form of microdroplets. To avoid evaporation of nutrient medium, incubation is performed in a humidity chamber. After incubation time, medium is removed by dabbing with an absorptive material (37). While broth is removed by capillary effects, such gentle “touching” of microdroplets does not remove bacteria, at least not to the extent that would disturb the assay performance (37–42). The mechanisms have not been investigated in detail, but it can be assumed that there is a strong adherence of bacteria to the surface of MALDI plate during the incubation, probably including biofilm formation.

Another fundamental feature of the method is that read-out is performed by MALDI-TOF MS, in contrast to turbidity reading with reference methods (10, 11). Microorganisms are multiplying without addition of an antimicrobial (growth control), and their amplified biomass is detected by MALDI-TOF MS after the incubation. The same holds true for resistant microorganisms that are growing despite the presence of the antimicrobial. In contrast, microorganisms that are susceptible to the tested antimicrobial are inhibited, and their originally inoculated, not-amplified biomass cannot be detected by a MALDI-TOF MS instrument (Fig. 2). The sensitive detection of biomass by MALDI-TOF MS allows incubation times to be considerably shortened in comparison to the reference method. Notably, in addition to the detection of microbial spectra as biomass itself, information on species identification is simultaneously available because those spectra are characteristic for particular microbial species (37).

In the first study introducing DOT-MGA, Idelevich et al. investigated the optimal testing conditions (37). Simple tools were used in this proof-of-principle study. Plastic transport boxes (Bruker) with 4 ml water added onto the bottom were applied as humidity chamber for incubation of inoculated MALDI targets to avoid evaporation of microdroplets. The broth removal after incubation was done by gently “touching” the microdroplets on the side at the bottom with a folded tissue wipe exploiting capillary effects. Also, simple criteria were applied for biomass detection and, thus, detection of bacterial growth. Successful microbial identification (score ≥1.7 by MALDI Biotyper software) was defined as detection of bacterial growth, thus indicating either a successful growth control without an antimicrobial, or a resistant isolate in the presence of an antimicrobial. In contrast, failed identification (score <1.7) signified lack of bacterial growth in the presence of an antimicrobial, and hence, determining the isolate as susceptible to the respective antimicrobial. In total, 24 Klebsiella pneumoniae and 24 Pseudomonas aeruginosa isolates were used to assess assay accuracy for detection of meropenem resistance. In a comparison of different microdroplet volumes (2, 4, 6, 8, and 10 μl), the best assay performance was achieved using 6-μL microdroplets. For K. pneumoniae, 4-h incubation was enough to generate valid results (defined as successful growth control detection) for all isolates, and the sensitivity and the specificity for detection of meropenem resistance reached each 100% at this incubation time. For P. aeruginosa, 5-h incubation resulted in 83.3% of valid tests with 100% sensitivity and 100% specificity for detection of meropenem resistance. The very good agreement with the broth microdilution reference method can be explained by the fact that critical test conditions recommended for the reference method were also observed for DOT-MGA. As described above, this includes compliance with the standard inoculum size, nutrient medium, and incubation temperature (37). Thus, the developed method was demonstrated to be feasible and to provide accurate results on resistance or susceptibility within a short time.

A subsequent study investigated a direct application of DOT-MGA on positive blood cultures (BCs) without prior subcultivation of colonies (41). AST performed directly from specimens allows further acceleration of diagnostics, which is of particular importance in the case of sepsis (4). Blood samples were spiked with meropenem-nonsusceptible and meropenem-susceptible Enterobacterales isolates, inoculated into BC bottles and incubated in an automated BC system. Upon positivity, BC broth was processed using four different methods, namely, filtration/dilution, dilution, lysis/centrifugation, and differential centrifugation. After processing, samples were subjected to the DOT-MGA, as described above. A technical improvement, implemented in this study, was the use of firm filter papers as absorptive material to remove the liquid medium by “touching” microdroplets sidewise. The best test performance was achieved with the lysis/centrifugation method. After 4 h of incubation, assay validity (successful detection of growth control) was 96.3%, with 91.7% sensitivity and 100% specificity among valid tests. An additional analysis of MALDI-TOF MS spectra with a novel prototype software within this study demonstrated that accuracy can be further increased by improvement of algorithms for spectra analysis. Overall, the study demonstrated feasibility and accuracy of the direct DOT-MGA from positive BCs (41).

Correa-Martínez et al. developed a phenotypic screening panel for rapid detection of extended-spectrum β-lactamases (ESBL) and AmpC β-lactamases in Enterobacterales using the MALDI-TOF MS DOT-MGA (38). This “all-in-one” screening panel in a 96-spot format relies on the synergistic effect between cephalosporins and inhibitors of ESBL and/or AmpC β-lactamase. Synergy, which indicates β-lactamase production, was detected by determining the MIC of cephalosporins with and without β-lactamase inhibitors. The layout was designed according to the diagnostic algorithm proposed by the EUCAST guidelines for detection of resistance mechanisms and specific resistances of clinical and/or epidemiological importance (43). The panel comprises four zones: screening for resistance against third-generation cephalosporins with cefpodoxime, ESBL detection with cefotaxime and ceftazidime in presence and absence of ESBL inhibitor (clavulanic acid), AmpC detection with cefepime, and detection of AmpC plus masked ESBL with cefepime plus ESBL inhibitor or cefotaxime plus ESBL inhibitor and AmpC inhibitor (cloxacillin). The test was developed using four reference strains recommended by the EUCAST, and validated on 50 clinical Enterobacterales isolates resistant to third-generation cephalosporins. As compared to the PCR findings, the following positive and negative percent agreement values (PPA/NPA) were obtained after 4 h of incubation: for ESBL, 94.4%/100%; for AmpC, 94.4%/93.8%; for ESBL + AmpC, 100%/100%. The results were comparable with those of broth microdilution and more accurate than the results of the combination disk test. Thus, the developed DOT-MGA-based assay rapidly provided accurate results, representing an “easy-to-perform” diagnostic alternative to the currently available methods (38).

In a further study, a similar approach was applied to develop a screening panel for rapid differential detection of carbapenemases in Enterobacterales (39). MALDI-TOF MS DOT-MGA was adapted as a phenotypic method for detection of carbapenem non-susceptibility, MIC determination, and carbapenemase differentiation in a single step. The identification of specific carbapenemase classes was performed based on the synergy between meropenem and a respective carbapenemase inhibitor (≥8-fold decrease of MIC compared to meropenem alone). Moreover, high-level temocillin resistance was applied as a further parameter for the detection of OXA production. In addition, the detection of AmpC production was included in the panel, because of the decreased susceptibility to carbapenems that these enzymes may cause. The validation of the assay was performed on seven reference strains recommended by EUCAST for detection of carbapenemases (43), as well as on 20 meropenem non-susceptible Enterobacterales strains isolated from clinical samples. After 4-h incubation, DOT-MGA correctly identified KPC-, MBL-, and OXA-producing isolates, in 100% agreement with PCR. Detection of AmpC was in agreement with the broth microdilution (performed on 96-well microtiter plates reflecting the layout of the DOT-MGA panel) and the combination disk test. Two isolates were identified as AmpC-positive by these phenotypic methods, while PCR was negative. A possible explanation for this may be the presence of AmpC genes that were not part of the PCR microarray. Overall, the DOT-MGA delivered accurate results in a short time. As with other phenotypic approaches, it allows the detection of unknown or uncommon carbapenemases encoded by genes that are not covered by PCR assays (39).

For the first time, the study of Nix et al. investigated the feasibility of the DOT-MGA for AST of Gram-positive bacteria (42). As an example, an application was developed for detection of methicillin resistance in S. aureus, because of its high therapeutic importance and relevance for infection control measures. Tested clinical isolates comprised 14 MRSA and 14 MSSA isolates as well as a collection of 16 MRSA challenge strains with different SCCmec types, mec genes, and spa types. As antimicrobial, cefoxitin was applied in the EUCAST breakpoint concentration. The study consisted of DOT-MGA testing from grown agar cultures, as well as direct testing from positive BCs. For the latter, blood samples were spiked with MRSA and MSSA isolates and positive BC broth processed by three different methods (dilution, lysis/centrifugation, and differential centrifugation) prior to being used for DOT-MGA. In this study, spectra were acquired with instrument settings optimized for DOT-MGA, and a prototype MBT FAST matrix (Bruker) was used, which contains an internal standard as quality control for spectra acquisition. The best performance of DOT-MGA for testing from agar cultures was observed after 5 h of incubation. For this time point, 100% sensitivity, 100% specificity, and 100% test validity were achieved. The 4-h incubation also resulted in 100% sensitivity and 100% specificity for detection of methicillin resistance, but test validity (successful detection of growth control) was only 85.7%. For direct testing from positive BCs, lysis/centrifugation method using a 10⁻¹ dilution of the 0.5 McFarland suspension resulted in best test performance. After 4-h incubation, 96.4% test validity, 100% sensitivity, and 100% specificity for detection of methicillin resistance were achieved with this BC processing protocol for DOT-MGA. Another important finding of this study was that the on-target protein extraction using formic acid considerably improved detection of methicillin resistance in S. aureus. To sum up, this study expanded the applications of DOT-MGA to the susceptibility testing of Gram-positive bacteria (42).

The robustness and reproducibility of the DOT-MGA were demonstrated by the applications and modifications investigated by researchers other than the test developers (40). Interestingly, Horseman et al. used another instrument platform and showed good sensitivity and specificity values with DOT-MGA while testing several bacterial species (40). Simplicity and universality of DOT-MGA were emphasized by the authors as well as its cost-effectiveness and ability to be adapted for clinical diagnostics. Ultimately, the study of Horseman et al. demonstrated DOT-MGA to be a rapid and accurate tool for AST, even when performed on an alternative MALDI-TOF MS instrument (40).

Li at al. applied a method, which was some kind of combination of MBT-ASTRA and DOT-MGA (44). The sample processing was similar to the MBT-ASTRA approach (25), with the incubation of bacteria and antimicrobials in a 96-well plate, subsequent lysis of bacteria by formic acid and acetonitrile, and transfer of the lysate onto a MALDI target plate for analysis (44). However, the data analysis and interpretation were performed as previously suggested for DOT-MGA (37), i.e., isolates were defined as resistant when MALDI Biotyper software achieved correct identification with scores ≥1.7 (growth detection) and as susceptible when the identification failed (scores <1.7, growth inhibition) (44). The authors tested carbapenem-resistant A. baumannii as an example, and demonstrated accurate AST results with a short incubation time of 4 h (44).

The progress of the DOT-MGA development can be witnessed by the publications mentioned above—the methods ranged from the use of simple tools that were adapted for the application with this assay to the development of standardized assay tools with the thorough control of test conditions and results. Indeed, standardization and automation of this assay would help to overcome logistic challenges, e.g., reduce hands-on times for application of bacteria and antimicrobials to the MALDI targets, removing excess fluid, and the result read-out by improved algorithms for spectra analysis and standardized software. The assay can be conceived as an MIC test with multiple dilutions of an antimicrobial or as a breakpoint test. At the time of the writing of this review, a ready-to-use version of the assay is in the development, but not yet commercially available. The MALDI-TOF MS-based DOT-MGA has been comprehensively reviewed by Neonakis and Spandidos (45). The latest developments and new applications of the DOT-MGA have been discussed in an updated review (46). For detailed information on DOT-MGA, we refer readers to the original publications mentioned above as well as to these review articles (45, 46).

In conclusion, several approaches have been described in the last decade for the growth-based phenotypic AST using MALDI-TOF MS. Of these, DOT-MGA appears to be most promising development, with a potential to become a widely spread AST technology. DOT-MGA enables easy, rapid, and accurate AST that is independent from the underlying resistance mechanism. Combination of identification and AST on a single instrument platform is advantageous for laboratory workflows. Information on antimicrobial susceptibility can be generated within the same working shift by DOT-MGA, hence enabling laboratories to communicate findings to the clinicians on the same day. This would significantly support therapeutic decision making and patient management. Further assay improvements would be appreciated, such as pre-deposition of antimicrobials on the MALDI targets, e.g., in lyophilized form, advanced technical solutions for incubation procedures and microdroplet removal, as well as (semi-) automation of the processing.