Comparative Evaluation of Bruker Biotyper and ASTA MicroIDSys for Species Identification in a Clinical Microbiology Laboratory

Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI–TOF MS) has been widely used for microbial identification, because of its speed and accuracy, since its introduction to clinical microbiology laboratories. In this study, we evaluated the performance of ASTA MicroIDSys, a newly developed MALDI–TOF, and compared it with the widely used Bruker Biotyper. Microbial identification with the Bruker Biotyper system was performed by using a direct smear method and the Bruker Biotyper database (reference library version 6.0.0.0). The isolates were also tested in parallel, using the ASTA MicroIDSys system with a direct smear method and the MicroIDSys database, CoreDB v1.26. A total of 914 clinical isolates were recovered from the clinical specimens. Identical results with confidence scores (≥2.0, for the Bruker Biotyper) and acceptable scores (≥140 for the ASTA MicroIDSys) were obtained for 840 (91.9%) isolates. The minor errors were defined as misidentification at the species level, and the rate was 1.1% (9/792) for Bruker Biotyper and 1.6% (13/792) for ASTA MicroIDSys. Major errors were defined as misidentification at the genus level, and the rate was 0.3% (2/792) for both Bruker Biotyper and ASTA MicroIDSys. ASTA MicroIDSys showed reliable performance for microbial identification, which was comparable to that of the Bruker Biotyper. Therefore, ASTA MicroIDSys can be applied for the identification of microorganisms in clinical microbiology laboratories.


Introduction
The introduction of matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) has greatly improved the turnaround time for the routine identification of microorganisms, along with rapid and accurate identification [1][2][3]. MALDI-TOF MS provided consistent and accurate results when compared with those of biochemical identification methods, but no standard method for dealing with unexpected species identification has been presented. The discrepancies in the outcomes for rare bacterial species for different MALDI-TOF MS manufacturers should be examined.
In the present study, we compared the routine performance of the ASTA MicroIDSys with that of the Microflex Biotyper for identification of all microbial isolates, including bacteria and yeasts, in a clinical microbiology laboratory, during the study period.

Materials and Methods
All clinical strains, except for filamentous fungi and mycobacteria, isolated in a clinical microbiology laboratory at a 750-bed general hospital in Korea, from October to December 2018, were included in this study. The study was approved by the institutional review board of Kangdong Sacred Heart Hospital (IRB file No. Kangdong NON2018-001, 18 September 2018). The clinical specimens were inoculated in appropriate media, such as 5% sheep blood agar, MacConkey agar, or chocolate agar for bacteria; Buccella blood agar for anaerobic bacteria; and Sabouraud dextrose agar for yeast. The specimens were then incubated for 24-48 h, at 35 • C, in appropriate conditions. A total of 914 clinical isolates were recovered from clinical specimens of blood, body fluids, wounds, and pus.
Microbial identification with Bruker Biotyper system was performed by using a direct smear method, following the manufacturer's instructions. Briefly, a singly colony of the isolate was smeared and dried on a plate. Subsequently, 70% formic acid and cyano-4hydroxycinnamic acid matrix solution were added, and the target plate was analyzed by using Bruker Biotyper database (reference library version 6.0.0.0, Bruker Daltonics, Bremen, Germany). The confidence scores values over 2.0 were considered acceptable according to the manufacturer's recommendations. The isolates were also tested in parallel, using ASTA MicroIDSys system with a direct smear method. Moreover, the target plate was analyzed by using MicroIDSys database (CoreDB v1.26, ASTA Inc., Suwon, Korea). Identification scores over 140 were considered acceptable according to the manufacturer's recommendations. If the test provided a score under the target cutoff (<140 for MicroIDSys and <2.0 for Bruker Biotyper) or invalid results, we immediately repeated the test with other colonies from the same agar plate.
When the results by two systems showed discrepancies at the species level or one of the results was under the cutoff score or invalid, 16S rRNA gene sequencing was performed for bacterial identification by Macrogen (Seoul, Korea). The PCR primers for 16S rRNA were 5 -GGATTAGATACCCTGGTA-3 and 5 -CCGTCAATTCMTTTRAGTTT-3 , and the sequencing primers were 5 -AGAGTTTGATCMTGGCTCAG-3 and 5 -TACGGYTACCTT GTTACGACTT-3 . The 16S rRNA sequences obtained were compared with GenBank data, using the BLAST alignment software (blast.ncbi.nlm.nih.gov, accessed on 12 July 2021) and a threshold of ≥99% homology was used for identification to the species level.
Chi-square test or Fisher's exact test was used for analyzing differences in the identification rate. SPSS Statistics 24 (IBM SPSS Inc., Chicago, IL, USA) was used for statistical analyses, and a 2-tailed p-value of 0.05 was considered statistically significant.
Minor errors were defined as misidentification at the species level with a score above the target cutoff (≥2.0 for Bruker Biotyper and ≥ 140 for MicroIDSys). The minor error rate was 1.1% (9/792) for Bruker Biotyper and 1.6% (13/792) for ASTA MicroIDSys (p = 0.388). Major errors were defined as misidentification at the genus level (≥140 for MicroIDSys and ≥ 1.7 for Bruker Biotyper). Exceptionally, the misidentification of S. aureus at the species level was considered as a major error because of the clinical importance of this species. The major error rate was 0.3% (2/792) for both Bruker Biotyper and ASTA MicroIDSys, and there was no case of misidentification of S. aureus as other coagulase-negative staphylococci. The isolate of Weisella cibaria was misidentified as E. coli, and the isolate of Brevibacterium frigoritolerans as Lactobacillus jensenii by Bruker Biotyper. The isolate of Staphylococcus warneri was misidentified as Azotobacter nigricans and the isolate of Kluyvera ascorbata was misidentified as Raoultella ornithinolytica by ASTA MicroIDSys.
The two systems showed discrepant results for 31 isolates (3.4%). However, at the genus level, they were in agreement, except for five isolates. The identification results for the 16S rRNA gene sequencing for these isolates are shown in Table 2. The 16S rRNA sequencing could not be performed for 14 bacterial isolates due to the lack of samples and one fungal isolate. Among the 16 results for the 16S rRNA sequencing, nine and three were in agreement with those of Bruker Biotyper and ASTA MicroIDSys, respectively (p = 0.066). The Bruker Biotyper showed invalid results for 10 isolates (1.1%), and ASTA Mi-croIDSys showed invalid results for four isolates (0.4%) (p = 0.178). The identification results for the 16S rRNA gene sequencing for these isolates are shown in Table 3.

Discussion
In this study, we evaluated the performance of ASTA MicroIDSys, a newly developed MALDI-TOF system which can be routinely used in clinical microbiology laboratories, in comparison with Bruker Biotyper which is widely used. Most of the bacteria and yeasts which are commonly isolated in clinical laboratories were correctly identified by MALDI-TOF MS, while a few uncommon bacteria, including Brevibacterium spp., Pseudoglutamicibacter spp., and Janibacter spp., were given invalid results.
The performance of MicroIDSys was comparable to that of Bruker Biotyper with the overall concordance rate of 91.9%. This result is in the same line with a previous study which reported good agreement of results between Bruker Biotyper and ASTA MicroIDSys. In that study, identical results with confidence scores (≥2.0 for Bruker Biotyper) and acceptable scores (≥140 for ASTA MicroIDSys) were obtained for 86.1% from the 4919 isolates recovered from sputum, urine, and pus samples [8]. In another study which evaluated the performance of ASTA MicroIDSys compared to that of VITEK MS, the ASTA MicroIDSys correctly identified 96.7% of isolates to species level which was comparable to VITEK MS (97.3%) [9].
As microorganisms are identified by MALDI-TOF MS systems, using prerecorded protein spectra, which are mostly based on ribosomal proteins, MALDI-TOF MS systems are intrinsically limited to differentiate closely related species of Klebsiella, Enterobacter, Citrobacter, and Raoultella [10,11]. The discrepant results between Bruker Biotyper and ASTA MicroIDSys in this study also revealed the known limitation of MALDI-TOF systems. Successful identification of microorganisms by using MALDI-TOF MS relies heavily on the database containing the spectra of known organisms. It is critical that it includes a sufficient number of isolates for each species, grown under a variety of conditions such that the spectral library for the organism is sufficiently robust to account for the inherent variability expected for any organism. Most of the discrepant or invalid results in this study were from microorganisms that are not frequently isolated in clinical laboratories. Of note, the ASTA MicroIDSys misidentified the rare microorganisms with scores over 140. There should be caution when interpreting the results of ASTA MicroIDSys with microorganisms such as Paenibacillus spp. or Weissella spp. Molecular approaches could be useful for correct identification. There has been a study which applied whole genome-based bacterial identification system for clinical isolates that were not identified with MALDI-TOF MS systems [12]. It evaluated thirtysix isolates including Corynebacterium spp., Brevibacterium spp., and Brevundimonas spp. which were also not correctly identified by MALDI-TOF MS in our study. Genome-based identification may be an additional tool in the future. However, whole genome sequencing is yet burdensome in cost and methodology for clinical microbiology laboratories. Targeted sequencing of 16S rRNA, gyrB, or rpoB for bacteria, and internal transcribed spacer (ITS) or 28S region for yeasts can be the practical approach when MALDI-TOF MS cannot give the correct identification.
Except for the above mentioned studies, there are other studies that evaluated the performance of ASTA MicroIDSys on yeast [13], anaerobic bacteria [14], mycobacteria [15], and filamentous fungi [16]. As the aim of this study was to evaluate the utilization of ASTA MicroIDSys for routine identification in clinical microbiology laboratories, there was no selection on the types of microorganisms or on the types of samples. This study evaluated the performance of ASTA MicroIDSys by using microorganisms isolated from all types of samples including blood specimen at a 750-bed general hospital.
There are several limitations in this study. First, when the results obtained with the Bruker Biotyper and the ASTA MicroIDSys were identical at the species with scores above the target cutoff, we considered the results as correct identification without per-forming16S rRNA sequencing. Second, there is inborn limitation of 16S rRNA sequencing as it may show poor discrimination power for some genera in Gram-positive cocci [17], Enterobacteriaceae [18], or for Campylobacter spp. [19]. However, the BLAST alignment of 16S rRNA sequencing showed only one type of species with ≥99% homology for all the clinical isolates in this study. Third, 16S rRNA sequencing could not be performed for 14 bacterial isolates which showed discrepant results by the Bruker Biotyper and ASTA MicroIDSys due to lack of samples. Lastly, molecular testing for fungal isolates could not be performed due to our laboratory setting, which might limit the exact evaluation of performance for identification of fungi. However, as 100 (95.2%) among the 105 isolates of Candida spp. showed concordant results at the species level by the Bruker Biotyper and the ASTA MicroIDSys with scores above the target cutoff, it was demonstrated that ASTA MicroIDSys could reliably identify clinically important fungal isolates.

Conclusions
Since the introduction of MALDI-TOF MS in clinical microbiology laboratories, it has been widely used for microbial identification because of its speed and accuracy. In this study, ASTA MicroIDSys showed reliable performance on microbial identification that was comparable to that of the Bruker Biotyper. Therefore, ASTA MicroIDSys could be applied for identification of microorganism in clinical microbiology laboratories.