Clinical Evaluation of VITEK MS PRIME with PICKME Pen for Bacteria and Yeasts, and RUO database for Filamentous Fungi

1. Introduction

Identifying microorganisms typically demands significant effort, time, and resources using various methods like physiological, serological, biochemical, and chemotaxonomic approaches. While genomic methods offer high reliability, they are slow and require extensive expertise [1]. In contrast, Matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS) offers a swift, cost-efficient alternative capable of processing large volumes of samples simultaneously [2]. This technique has revolutionized conventional identification methods in laboratories by enabling rapid pathogen identification [3]. Although it is widely accepted for identifying bacteria and yeasts, there are still challenges when it comes to filamentous fungi [4]. The main challenge is the lack of a sufficient database and of an effective and rapid protein extraction method [5].

The VITEK MS PRIME (bioMérieux, Marcy l’Etoile, France) is a new version of the VITEK MALDI-TOF MS instrument developed by bioMérieux. This upgraded instrument features a continuous load-and-go sample loading system, prioritization for critical patient samples on urgent slides, and new internal components, all of which are designed to enhance sample processing speed and reduce handling time [6]. The VITEK MS PRIME system utilizes the VITEK MS IVD Database 3.2., but it also allows for additional analysis using the SARAMIS® Knowledge Base v4.16 (for research use only).

The VITEK-PICKME pen (hereafter referred to as PICKME, bioMérieux) is a device with a disposable nip that facilitates the user's ability to pick and smear colonies on the VITEK MS target slide, improving smear quality. Compared to conventional wooden toothpicks, PICKME provides better uniformity and thinner deposits [7]. A study highlighted by the American Society for Microbiology (ASM) Microbe 2019 showed its ability to decrease variability among users, reduce sample preparation time by 26-42%, and improve uniformity across different user experience levels [8].

In this study, bacterial and fungal isolates detected in clinical specimens were identified using the VITEK MS PRIME to confirm their identification capabilities. Both bacteria and yeast were tested using the PICKME and a conventional wooden toothpick, respectively. The results were also compared with the MicroIDSys Elite (ASTA, ASTA Corp., Suwon, Korea) MALDI-TOF MS currently used in our laboratory. For filamentous fungi, clinical isolates previously identified by molecular sequencing were analyzed using the VITEK MS IVD Database 3.2 and the SARAMIS® Knowledge Base V4.16 (RUO database), respectively to find the clinical utility of the RUO database.

2. Materials and Methods

2.1. Clinical Isolates

A total of 320 bacteria and yeast isolates were obtained from the International St.Mary`s hospital, consisting of 268 bacteria (124 gram-negative bacilli, 91 gram-positive cocci, 22 gram-positive bacilli, 16 anaerobes, 15 gram-negative cocci) and 52 yeasts (37 Candida species, 15 others). All isolates were confirmed by ASTA MicroIDSys Elite (ASTA), VITEK2 (bioMérieux), and/or molecular sequencing by Clinical and Laboratory Standards Institute (CLSI) guideline (MM18) [9].

For the evaluation of filamentous fungi, 289 fungal isolates deposited in the Korean nationwide fungal collection network by the National Culture Collection for Pathogens, were used. Fungal isolates compromising 31 genera and 79 species, were obtained from various clinical specimens from 10 hospitals in the Republic of Korea (Seoul, Incheon, Gyeonggi-do, Busan, Daejeon, Ganwon-do, and Jeju Island) between March 2017 and December 2020. They were recovered from various clinical specimens such as respiratory specimens, blood, urine, CSF, body fluids, tissue, eye, nose, ear, nail, skin, and hair. Preservation was done using liquid nitrogen in a deep freezer to minimize loss. The identification of the isolates involved macroscopic phenotypic identification/microscopic examination and molecular sequencing, as described in previous studies [2].

Approval for this study was obtained from the International St. Mary's Hospital, Catholic Kwandong University College of Medicine in Korea (IS22ESSE0006).

2.2. MALDI-TOF MS Analysis

For bacterial and yeast isolates, VITEK MS PRIME [6] and ASTA MicroIDSys Elite [2] systems were used according to the manufacturer`s instructions. Sample preparation was performed by direct smearing method, as per the manufacturer's instructions. VITEK MS PRIME was conducted using both the PICKME and a wooden toothpick. The spectra generated by VITEK MS PRIME were analyzed by the VITEK MS Software (version 1.1.0 – 203571250) and the VITEK MS IVD Database 3.2. ASTA MicroIDSys Elite experiments were conducted using the applying wooden toothpick. The MS spectra obtained by ASTA MicroIDsys Elite were analyzed with the reference library (CoreDB version 1.27.04). Duplicate spreading on a single-use target slide was performed for all samples.

Identification of filamentous fungi was performed using the VITEK MS Mold Kit, following the manufacturer's instructions. After experiments, filamentous fungi were analyzed in the IVD database first in VITEK MS PRIME. Strains that were not identified at the species level in the IVD database were analyzed once more with the RUO database version (SARAMIS® Knowledge Base V4.16).

2.3. Interpretation and Analysis of the Results

For bacterial and yeast isolates, concordance was determined when both VITEK MS PRIME and ASTA MicroIDSys Elite provided the same species or genus identification at the species or genus level. In cases of no identification or discrepant results, 16S rRNA gene sequencing was used as a reference method. For filamentous fungi, consistency was determined if one or more of the duplicate results coincided with the sequencing result. Identification failure was recorded when two results were inconsistent or identification was not possible. Statistical analysis was performed using MedCalc Statistical Software version 19.2.1 (MedCalc Software Ltd, Ostend, Belgium), using Chi-square and Fisher's exact tests with a two-tailed p-value.

3. Results

3.1. Bacterial and Yeast Identification Using PICKME and Wooden Toothpicks

Using the wooden toothpicks, VITEK MS PRIME identified bacteria and yeast at the species level in 91.9% (294/320) of cases, while PICKME achieved species-level identification in 90.3% (289/320) (Table 1). The concordance rate of wooden toothpicks and PICKME in VITEK MS PRIME was 89.4% (286/320).

Among the 34 bacterial isolates with discrepancies between the wooden toothpick and PICK-ME (Table 2), eight isolates (2.5%) were correctly identified only by the wooden toothpick, three (0.9%) were correctly identified only by PICKME, and 23 isolates (5.9%) showed no identification (n=19) or incomplete identification (n=4).

After retesting, eight isolates showed species-level agreement between the wooden toothpick and PICKME. Ultimately, 294 isolates (91.9%) showed species-level agreement between the wooden toothpick and PICKME. Additionally, two isolates showed genus-level agreement between PICKME and the wooden toothpick.

There are a total of 15 strains of Corynebacterium included (5 Corynebacterium striatum, 3 Corynebacterium diphtheriae, 2 Corynebacterium amycolatum, 2 Corynebacterium jeikeium, 2 Corynebacterium afermentans, and 1 Corynebacterium simulans). Among these strains, 11 were successfully identified at the species level using both the PICKME and wooden toothpick methods. However, for the 2 strains of Corynebacterium jeikeium, only the wooden toothpick method achieved species-level identification, while the PICKME method failed. Additionally, for the 2 strains of Corynebacterium afermentans, both the PICKME and wooden toothpick methods resulted in failed identification.

3.2. Bacterial and Yeast Identification Using VITEK MS PRIME and ASTA MicroIDSys Elite

ASTA MicroIDSys Elite achieved a species-level identification rate of 88.1% (282/320). The concordant rates for species-level identification 85.3% (273/320) between the wooden toothpick and ASTA MicroIDSys Elite, and 83.4% (267/320) between PICKME and ASTA MicroIDSys Elite. Two isolates (Bacteroides thetaiotaomicron, Stenotrophomonas maltophilia) initially showed no identification in ASTA MicroIDSys Elite but were correctly identified upon retesting. As both isolates showed identification failure in VITEK MS PRIME, the agreement rate between ASTA MicroIDSys Elite and VITEK MS PRIME did not change after retesting. Overall, gram-positive bacilli including Corynebacterium exhibited higher accuracy in ASTA MicroIDSys Elite (95.5%) compared to VITEK MS PRIME (72.7-81.8%). The strains of Corynebacterium jeikeium and Corynebacterium afermentans that failed to be identified in VITEK MS PRIME were both successfully identified at the species level in ASTA MicroIDSys Elite.

3.3. Identification of Filamentous Fungi Using the IVD and RUO Databases

Among the 289 filamentous fungi, species-level identification using the IVD database was 71.3% (206/289), while genus or complex level identification was 75.4% (218/289) (Figure 1).

Additionally, when analyzed using the RUO database, species-level identification improved significantly to 85.8% (248/289), and genus or complex level identification increased to 88.2% (255/289). There was no significant difference in identification performance between the IVD and RUO databases for Aspergillus strains. However, for basidiomycetes, there was a significant improvement in identification performance when using the RUO database, with an increase from 35.3% to 100% (P<0.0001). Detailed results for each strain are summarized in Table 3.

3.4. Identification Failure of Filamentous Fungi

A total of 34 isolates failed to be identified in VITEK MS PRIME using both the IVD and RUO databases (Table 4). Among them, six isolates were strains included in the IVD database, including Aspergillus versicolor (3), Exophiala xenobiotica (2), and Nannizzia gypsea (1). The remaining 28 isolates were strains included only in the RUO database, with Penicillium and Talaromyces strains accounting for 11 cases.

4. Discussion

Based on the species included in the VITEK MS IVD database, the performance of VITEK MS PRIME for the identification of bacteria and yeasts using the PICKME was remarkable. It provided quick and reliable identifications for 89.4% of the isolates, using the same laboratory workflow as the conventional method in various bacterial and yeast species. A previous study highlighted the significant role of the PICKME in selecting rough Nocardia colonies and creating thin deposits, essential for accurate MALDI-TOF identification [7]. The study involved adding formic acid directly to the deposit along with the cyano-4-hydroxycinnamic acid (CHCA) matrix solution to facilitate bacterial wall destruction and protein extraction. Despite being an extra step, this direct deposit analysis method proved much faster than the sample preparation recommended by the manufacturer and was compatible with standard bacterial identification workflows.

The quality of the deposit on the target can influence the spectra's quality. Proper training is needed to ensure the appropriate amount of colony transfer onto the plate [10]. Previous research reported significant differences in score distribution among testing institutions, even when using the same extraction method [11]. In this study, all unidentified isolates with wooden toothpicks and 37.5% of unidentified isolates with PICKME were correctly identified after retesting. Therefore, using the PICKME may lead to more consistent testing, unaffected by the examiner's capabilities.

Comparing VITEK MS PRIME to ASTA MicroIDSys Elite, the platform showed similar performance. Notably, the VITEK MS PRIME exhibited relatively higher rates of inaccurate identification in the Coryneform and aerobic/anaerobic gram-positive bacilli group, which aligns with a previous study [12]. This difference may be attributed to the thick peptidoglycan layer, potentially interfering with laser ionization. Interestingly, among the 19 strains not detected by VITEK MS PRIME, ASTA MicroIDSys Elite detected 7 strains at the species level and 3 strains at the genus level. Among the unidentified isolates in VITEK MS PRIME, Aeromonas dhakensis, Comamonas acidovorans, Spingomonas paucimobilis, and Yarrowia galli have been reported as unidentified isolates in previous systems and databases [13,14]. These species do not exist in the VITEK MS ver. 3.2 database, but all of them were reported as human pathogens [15,16,17,18]. Therefore, updating the database is needed to improve the distinguishing ability from other species.

For filamentous fungi, the VITEK MS PRIME using the updated IVD database successfully species-level identification in 71.3% (206/289). The use of the RUO database significantly improved the identification ability by up to 85.8% and especially, basidiomycetes showed a remarkable improvement.

The manufacturers' libraries for microbial identification are regularly updated to include more species and variations within a single species. However, they still face the challenge of incompleteness due to the vast diversity encountered in human pathology [10]. The VITEK MS v3.2 database only covers 221 fungal species. While common molds are generally well covered, identifying rarer or cryptic species becomes more challenging. To improve the identification capability of rare filamentous fungi, developing a homemade library or utilizing open public databases for analysis is possible, but it may not be straightforward to implement in actual laboratories [10]. The VITEK MS PRIME provides a practical solution, allowing a seamless transition to the RUO database for strains that failed in the existing IVD database without additional experiments. This is expected to significantly aid in the identification of rare filamentous fungi.

The VITEK MS PRIME successfully identified T. rubrum and T. Tonsurans to the species level in IVD mode. However, it failed to achieve species-level identification for T. interdigitale and T.mentagrophytes even in RUO mode. This lack of accurate identification was also observed in a previous study using VITEK MS v3.0 [19]. Indeed, some strains (Aspergillus vesicolor, Exophiala xenobiotica, and Nannizzia gypsea) included in the IVD database, could not be identified using VITEK MS v3.2. These findings were also reported in the previous study, where Zvezdánova et al. reported correct identification rates of 86.4% for Aspergillus species and 65.7% for Mucorales [20]. It will be necessary to enhance the database for Aspergillus species, Mucorales, and Trichophyton strains to improve their identification in the VITEK MS PRIME.

In this study, VITEK MS PRIME with PICKME provides reliable results with high efficiency and allows standardized sample preparation. For filamentous fungi, VITEK MS PRIME provides reliable results for diverse filamentous fungi that are commonly isolated in clinical laboratories. For further identification, combining the use of the RUO database can be beneficial, especially for basidiomycetes.