The authors have declared that no competing interests exist.
Antibiotic resistant bacterial infections are a significant problem in the healthcare setting, in many cases requiring the rapid administration of appropriate and effective antibiotic therapy. Diagnostic assays capable of quickly and accurately determining the pathogen resistance profile are therefore crucial to initiate or modify care. Matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry (MS) is a standard method for species identification in many clinical microbiology laboratories and is well positioned to be applied towards antimicrobial susceptibility testing. One recently reported approach utilizes semi-quantitative MALDI-TOF MS for growth rate analysis to provide a resistance profile independent of resistance mechanism. This method was previously successfully applied to Gram-negative pathogens and mycobacteria; here, we evaluated this method with the Gram-positive pathogen
Antimicrobials are effective and often life-saving treatment options for bacterial infections; however, rising levels of resistance and the emergence of multidrug-resistant strains are becoming major challenges in continuing to combat these pathogens. Researchers and medical professionals are pursuing several avenues for countering these problems, with increased antimicrobial stewardship among the forefront [
Matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry (MS) is an effective tool for routine bacterial and yeast identification, with increasing adoption in clinical microbiology laboratories due to its fast time to answer and low per-sample costs [
Recent studies reported an alternative MALDI-TOF MS based method for AST termed the MALDI Biotyper antibiotic susceptibility test rapid assay (MBT-ASTRA) as a possible solution to these limitations [
One drawback of the MBT-ASTRA assay is the concentration of antibiotic used and the incubation time must be optimized for each species and antibiotic combination [
Bacterial isolates were obtained from the American Type Culture Collection (ATCC), the Biological and Emerging Infections Resources Program (BEI Resources), the Children’s National Medical Center via the Food and Drug Administration (FDA), and the Centers for Disease Control (CDC). The origins of each strain are listed in
Source | Ciprofloxacin MIC (ug/mL) | Oxacillin MIC (ug/mL) | Cefepime MIC (ug/mL) | Vancomycin MIC (ug/mL) | |
---|---|---|---|---|---|
BAA-38 | ATCC | 0.19 | 16 | 64 | 2 |
BAA-39 | ATCC | 4 | 48 | 24 | 2 |
BAA-40 | ATCC | 0.125 | >256 | 192 | 2 |
BAA-41 | ATCC | 12 | >256 | >256 | 2 |
BAA-43 | ATCC | 16 | >256 | >256 | 2 |
BAA-44 | ATCC | 12 | >256 | >256 | 2 |
700698 | ATCC | >32 | >256 | >256 | 3 |
FDAARGOS_135 | CNH | 8 | 128 | 192 | 2 |
FDAARGOS_140 | CNH | 6 | 128 | 192 | 2 |
GRIM 48 | CNH | 0.19 | 0.5 | 3 | 1.5 |
FDAARGOS_159 | CNH | 0.19 | 0.5 | 2 | 2 |
FDAARGOS_169 | CNH | 1 | 32 | 64 | 1.5 |
ATCC 25923 | ATCC | 0.19 | 0.19 | 1.5 | 1.5 |
ATCC 29213 | ATCC | 0.19 | 0.25 | 2 | 1.5 |
ATCC 12600 | ATCC | 0.19 | 0.19 | 2 | 1 |
ATCC 13565 | ATCC | 0.19 | 0.25 | 1.5 | 0.75 |
ATCC 14458 | ATCC | 0.125 | 1 | 2 | 1.5 |
ATCC 8095 | ATCC | 0.19 | 0.19 | 2 | 1 |
ATCC 8096 | ATCC | 0.125 | 0.125 | 1.5 | 1.5 |
ATCC 29247 | ATCC | 0.125 | 0.38 | 2 | 1.5 |
HIP13170 | BEI | >32 | >256 | >256 | 4 |
HIP13419 | BEI | >32 | 256 | >256 | 4 |
880 (BR-VRSA) | BEI | >32 | 128 | 128 | >256 |
AIS 1000505 | BEI | >32 | >256 | >256 | 6 |
71080 | BEI | >32 | >256 | >256 | >256 |
VCU089 | BEI | 32 | 0.125 | 2 | 1.5 |
HFH-29568 | BEI | 8 | 16 | 24 | 1.5 |
MN8 | BEI | 0.25 | 0.38 | 3 | 1.5 |
HT 20020058 | BEI | 0.25 | 0.75 | 2 | 1.5 |
HT 20020470 | BEI | 0.25 | 0.38 | 2 | 1.5 |
A970675 | BEI | 0.19 | 0.75 | 3 | 1.5 |
MNHOCH | BEI | 0.25 | 0.75 | 3 | 1.5 |
VCU006 | BEI | 0.19 | 0.19 | 2 | 2 |
HFH-30123 | BEI | >32 | >256 | >256 | 2 |
A890259 | BEI | 0.19 | 0.19 | 3 | 1.5 |
AR# 0215 | CDC | 4 | |||
AR# 0216 | CDC | 4 | |||
AR# 0217 | CDC | 8 | |||
AR# 0218 | CDC | 4 | |||
AR# 0219 | CDC | 8 | |||
AR# 0220 | CDC | 8 | |||
AR# 0221 | CDC | 4 | |||
AR# 0222 | CDC | 4 | |||
AR# 0223 | CDC | 4 | |||
AR# 0224 | CDC | 4 | |||
AR# 0225 | CDC | 4 | |||
AR# 0226 | CDC | 4 | |||
AR# 0227 | CDC | 4 | |||
AR# 0228 | CDC | 4 |
ATCC, American Type Culture Collection; CNH, Children’s National Hospital; BEI, Biodefense and Emerging Infections Research Resources Repository; CDC, Centers for Disease Control
Minimum inhibitory concentrations (MICs) were determined by Etest. Freshly plated bacteria were grown for approximately 24 h at 37°C before colonies were used to create 0.5 McFarland suspensions in saline solution. The solutions were used to uniformly inoculate MH agar plates via polyester swabs. Etest strips (bioMérieux, France) were placed on the plates and incubated for 24 h at 37°C. MIC values were determined from the test strips as per the manufacturer’s instructions.
Cultures of
One microliter of formic acid protein extract was spotted on a polished steel MALDI target and allowed to completely dry. One microliter of a 5 mg/mL solution of α-cyano-4-hydroxy-cinnamic acid (CHCA) in 50% acetonitrile—47.5% water—2.5% trifluoroacetic acid was then spotted over the dried spots within 30 min. After drying, the samples were analyzed with either a Microflex LRF or an Autoflex III smartbeam MALDI-TOF MS instrument (Bruker Daltonics, Bremen, Germany) in linear positive mode. Both instruments produced equally usable data. Instrument parameters for the Microflex were: mass range, 2,000–20,800 Da; ion source (IS) 1, 20.0 kV; IS2, 18.35 kV; lens, 9.0 kV; detector gain, 2,916 V; deflection up to 1,950 Da. Instrument parameters for the Autoflex were: mass range, 2,000–20,800 Da; ion source (IS) 1, 20.0 kV; IS2, 18.35 kV; lens, 7.0 kV; detector gain, 1868 V; gating, 2,000 Da; gating strength, high. Spectra were acquired at maximum frequency with 2,000 shots per spot. Instrument calibration was performed with the bacterial test standard (BTS; Bruker Daltonics). All samples were analyzed in duplicate.
Automated data analysis was performed with the MBT-ASTRA software prototype made available by Bruker [
Relative growth values for 35
Relative growth values from the MBT-ASTRA assay performed with oxacillin are shown. A relative growth cutoff of 0.5 was utilized to classify resistance, indicated by the horizontal line. Strains are arranged from left to right in order of increasing MIC, with exact values given in
Relative growth values from the MBT-ASTRA assay performed with cefepime are shown. A relative growth cutoff of 0.5 was utilized to classify resistance, indicated by the horizontal line. Intermediate resistance is abbreviated as Int. Strains are arranged from left to right in order of increasing MIC, with exact values given in
Relative growth values from the MBT-ASTRA assay performed with vancomycin for the 35
Assay reproducibility from 40 replicates with strains susceptible and resistant to each antibiotic. Relative growth values from each replicate are shown, with data points representing major and very major errors colored red. A relative growth cutoff of 0.5 was utilized to classify resistance, indicated by the horizontal line. Intermediate resistance is abbreviated as Int. For each strain, 40 replicates were obtained over at least 4 days with up to 12 replicates from a single starter culture.
All blood cultures were performed with BD BACTEC standard/10 aerobic/F bottles and were incubated in a BD BACTEC FX40 instrument. Blood cultures were inoculated with 10 mL of fresh human whole blood (BioreclamationIVT, Maryland, USA) spiked with 400 μL of
Previous work demonstrated that the antibiotic concentration and incubation time used in the MBT-ASTRA assay must be optimized for each antibiotic / species combination, with the discriminating antibiotic concentration potentially being significantly different than the resistance breakpoint concentration [
Antibiotic | Isolates resistant by Etest | Isolates susceptible by Etest | Total replicates | Very major error rate | Major error rate | Sensitivity (95% CI) |
Specificity (95% CI) |
Overall accuracy | ||
---|---|---|---|---|---|---|---|---|---|---|
Susceptible by MBT-ASTRA | Resistant by MBT-ASTRA | Susceptible by MBT-ASTRA | Resistant by MBT-ASTRA | |||||||
Cirpofloxacin | 2 | 61 | 77 | 3 | 143 | 1.4% | 2.1% | 97% (89–100) | 96% (89–99) | 97% |
Oxacillin | 6 | 62 | 72 | 2 | 142 | 4.2% | 1.4% | 91% (82–97) | 97% (91–100) | 94% |
Cefepime | 3 | 58 | 70 | 3 | 134 | 2.2% | 2.2% | 95% (86–99) | 96% (89–99) | 96% |
Vancomycin | 0 | 8 | 111 | 5 | 124 | 0.0% | 4.0% | 100% (63–100) | 96% (90–99) | 96% |
Overall | 11 | 189 | 330 | 13 | 543 | 2.0% | 2.4% | 95% (90–97) | 96% (94–98) | 96% |
aCI, confidence interval
For ciprofloxacin, a concentration of 4 μg/mL was the lowest concentration providing consistent differentiation between resistant and susceptible strains, while an incubation time of 2 h was required to allow sufficient growth of slower growing
We chose oxacillin and cefepime as representatives of the penicillin and cephalosporin families of β-lactam antibiotics, respectively. As with ciprofloxacin, 2 h incubation yielded the best assay performance for the reasons presented above. For oxacillin, an antibiotic concentration of 2 μg/mL was optimal with 94% overall accuracy (134/142 total replicates). These conditions correctly classified strains in most instances, with 2 major errors and 6 very major errors (
In the case of cefepime, a 2 h incubation with 4 μg/mL antibiotic resulted in a 96% overall accuracy (128/134 total replicates), with 3 major errors and 3 very major errors (
A 2 h incubation with 2 μg/mL vancomycin showed optimal classification of resistance with this antibiotic; however, availability of resistant isolates limited testing to two strains, both with MICs > 256 μg/mL. These optimized conditions yielded an overall accuracy of 96% (119/124 replicates), with 5 major errors and no very major errors (
To better understand assay reproducibility within single strains, we performed an additional 40 replicates with at least two susceptible and two resistant strains for each antibiotic, as well as with intermediate resistance strains when available (
Results from the larger number of replicates for strains with intermediate resistance were in line with the previous results and highlight the difficulties of applying the MBT-ASTRA assay in its current form to intermediate resistance. Strains BAA-39 with cefepime and HIP13170 with vancomycin both performed similarly to the fully susceptible strains for those respective antibiotics, despite the assay being optimized to employ the lowest antibiotic concentration possible while consistently classifying susceptible strains correctly. In contrast, performing the assay with vancomycin on strain AR# 0217 resulted in a widely scattered distribution of relative growth values centered on approximately 0.75 but ranging from nearly zero to greater than one (
Finally, we assessed the performance of the MBT-ASTRA assay optimized for
MBT-ASTRA assay performed from spiked blood culture. Relative growth values from each replicate are shown, with data points representing major errors colored red. Strain FDAARGOS_159 represented by (●), strain ATCC 25923 represented by (■), strain BAA-44 represented by (▲), strain BR-VRSA represented by (♦). Filled symbols indicate the strain is susceptible to the antibiotic by reference standard while open symbols indicate resistance. A relative growth cutoff of 0.5 was utilized to classify resistance, indicated by the horizontal line. For each strain, 3 independent replicates were obtained on different days, with the mean and standard deviation shown for each set.
MALDI-TOF represents an attractive platform for the development of AST assays due to its rapid sample analysis speed and low operating costs. MALDI-TOF instruments are also already widely available in clinical microbiology laboratories, which would facilitate adoption of new diagnostic assays utilizing the platform. Semi-quantitative MALDI-TOF, in the form of the MBT-ASTRA assay, is a promising route forward; however, previous studies only applied the assay to Gram-negative and mycobacteria species. The results presented here demonstrate for the first time that the resistance profile of a Gram-positive species,
Gram-positive bacteria present a challenge for analysis by MALDI-TOF due to the thick peptidoglycan layer which has been noted to interfere in cell lysis in previous studies [
Previous studies on the MBT-ASTRA assay employed a minimum AUC cutoff for samples without antibiotic, which would potentially obviate the problem of extraction failure for susceptible strains. In optimizing the assay for Gram-positives, we chose to omit an AUC cutoff as
Extraction failures that led to errors in resistance classification appeared to increase in frequency with increased cell mass, although the correlation was not perfect with other factors likely also contributory. This was noted during the optimization stage for each antibiotic when incubation times beyond 2 h were evaluated and was more notable for faster growing strains. In addition to directly causing assay failure, this issue placed a limit on incubation time, thereby restricting further assay optimization to improve overall performance for slow growing isolates. In this context, future implementation of methods to improve extraction efficiency for the MBT-ASTRA assay would mitigate this issue. Our future optimization of this technique will focus on the addition of enzymatic pretreatment, thermal lysis, or physical disruption steps, or a combination thereof. Exploratory experiments with physical disruption via glass beads in our laboratory have so far shown some success in this regard.
Beyond extraction failure, another source of error appeared to stem from variability of independent cultures of the same strain displaying slightly more or less sensitivity to an antibiotic based on growth rate quantification by MALDI-TOF. Analysis of multiple replicates with and without antibiotics generated from a single starter culture highlighted this phenomenon. In these cases, the relative growth values derived from these replicates would trend similarly in a certain direction. The replicates of FDAARGOS_159 treated with ciprofloxacin performed to test assay reproducibility provide an example of this grouping: the mean of twelve replicates from a single starter culture was significantly higher than the mean of the rest of the replicates generated from other starter cultures (p = 0.023, see
In this study, we did not include the results from strains displaying intermediate resistance for a given antibiotic in the accuracy or error analysis, as the MBT-ASTRA assay is not currently designed to classify these strains. Modifications to the assay that would allow classification of intermediate resistance have been discussed elsewhere and include utilizing multiple antibiotic concentrations or finding a correlation between relative growth and MIC [
Improvements to the assay for
In conclusion, we demonstrate the first use of the MBT-ASTRA assay for AST in a Gram-positive bacterium. Despite the difficulties discussed above, the assay performed adequately in
A relative growth cutoff of 0.5 was utilized to classify resistance, indicated by the horizontal line. Vertical lines are drawn at the susceptibility and resistance breakpoints (susceptible ≤ 1 μg/mL; resistant ≥ 4 μg/mL).
(TIF)
A relative growth cutoff of 0.5 was utilized to classify resistance, indicated by the horizontal line. A vertical line is drawn at the susceptibility breakpoint (susceptible ≤ 2 μg/mL).
(TIF)
A relative growth cutoff of 0.5 was utilized to classify resistance, indicated by the horizontal line. Vertical lines are drawn at the susceptibility and resistance breakpoints (susceptible ≤ 8 μg/mL; resistant ≥ 32 μg/mL). Strains falling between these values are classified as intermediate resistance.
(TIF)
A relative growth cutoff of 0.5 was utilized to classify resistance, indicated by the horizontal line. Vertical lines are drawn at the susceptibility and resistance breakpoints (susceptible ≤ 2 μg/mL; resistant ≥ 16 μg/mL). Strains falling between these values are classified as intermediate resistance.
(TIF)
Group 1 represents a set of 12 replicates generated from a single starter culture while group 2 represents the remaining replicates. Of the 12 replicates in group 1, 10 showed a relative growth above 0.4 (5 of which were incorrectly classified). Meanwhile, only 6 of the remaining 28 replicates resulted in relative growth values above 0.4. The relative growth mean of group 1 (0.428) was significantly different than that of group 2 (0.291; p = 0.023, from unpaired t test with Welch’s correction).
(TIF)
The opinions, interpretations, conclusions, and recommendations contained herein are those of the authors and are not necessarily endorsed by the U.S. Army. The authors would like to thank Gongyi Shi and the Bruker Corporation for assistance with assay advice and for the prototype software for data analysis, and Lisa Cazares for helpful discussions regarding MALDI and access to instrumentation. The following reagents were obtained through the BEI Resources, NIAID, NIH: