Evaluation of the compatibility of Phoenix 100 and Microflex LT MALDI-TOF MS systems in the identification of routinely isolated microorganisms in the clinic microbiology laboratory

Introduction

Identification of microorganisms in clinical microbiology laboratories is primarily based on phenotypic characteristics presented by colony morphology, reproduction in different media, gram staining, and biochemical assays [1, 2]. It was a foregone conclusion that the development of modern information systems would benefit clinical laboratory studies [3]. Automation in microbiology began in the early 1970s with adoption of semi-automatic blood culture devices. This innovation was followed by the automation of bacterial identification and susceptibility testing [4]. Compared to manual methods, a, improve repeatability, enhance data management, and accelerate result generation. However, automation can also present disadvantages, such as costly equipment and consumable materials [2, 5, 6].

Matrix-assisted laser desorption/ionization time-of flight mass spectrometry (MALDI-TOF MS) is a system that does not require any special experience in the identification of microorganisms and has a high process volume and can perform identification in minutes. MALDI-TOF MS systems are becoming prevalent worldwide because of their superiority over other methods in cost-effectiveness and ease of use [7].

Adoption of new tests or devices in clinical laboratories requires a careful evaluation of whether the new systems function as expected [8]. Many centers that use the Phoenix 100 system today will probably adopt MALDI-TOF MS systems in the future. In the present study, we evaluate the compatibility of microbial identifications produced by the Phoenix 100 and a MALDI-TOF MS system.

Materials and methods

This study was conducted in the clinical microbiology laboratories of Cumhuriyet University, Faculty of Medicine, Research and Application Hospital, which is a tertiary training and research hospital with 1150 beds. Our study was completed over a 6-month period between October 2014 and March 2015. Microorganisms, which were isolated from the cultures of different samples sent from our hospital’s clinic and polyclinics, were identified simultaneously with Phoenix 100^® (Becton Dickinson, Sparks, MD, USA) and Microflex LT MALDI-TOF MS (Bruker Daltonics, Bremen, Germany). An attempt was made to confirm the isolates for which identical assignments could not be obtained on both devices, using the VITEK 2 compact system (VITEK 2 GP and VITEK 2 GN kits; bioMérieux, Marcy l’Etoile, France), BBL Crystal (BBL Crystal GP and E/NF kits; Becton Dickinson), and known conventional methods.

In compliance with their characteristics, different samples submitted to the laboratory were inoculated into one or more of the following media: 5% sheep blood agar (Salubris), eosin methylene blue (EMB) agar (Salubris), Chocolate agar (Salubris), or Salmonella-Shigella (SS) agar (Salubris). Inoculated plates were incubated in aerobic and 5% CO₂ conditions at 35.5 °C for 24–48 h and then evaluated for microorganisms.

Results

Eight hundred and twelve microbial isolates from various clinics and polyclinics requiring aerobic culture were identified simultaneously, using Phoenix 100 and Microflex LT MALDI-TOF MS devices. Recurrent isolations from the same patient were not included in the study.

Both systems simultaneously identified 359 Enterobacteriaceae isolates in the pool of clinical isolates. Both systems identified the same species for 99.1% of these samples. However, the MALDI-TOF MS could identify, at the species level, two Salmonella isolates that the Phoenix 100 did not identify at the species level. These bacteria were confirmed with antiserums (DIFCO, Becton Dickinson). The most frequently isolated bacterium in this group was Escherichia coli. One hundred and ninety-four of the 195 bacterial isolates (99.4%) tested using the Phoenix 100 were assigned the same identity when tested using the MALDI-TOF MS. A bacterial isolate identified as Escherichia coli by the Phoenix 100 was identified as Klebsiella pneumoniae by the MALDI-TOF MS device (Table 1). Motility tests on this isolate were negative in our hands, using a VITEK II Compact and BBL Crystal E/NF (Becton Dickinson). This negative result confirmed the MALDI-TOF MS identification of this isolate as Klebsiella pneumoniae. Additionally, two strains identified as Klebsiella pneumoniae and Enterobacter cloacae by the Phoenix 100 were identified differently by MALDI-TOF MS; further tests using the VITEK II Compact and BBL Crystal E/NF devices confirmed the MALDI-TOF MS identifications. Phoenix 100 and MALDI-TOF MS identifications of Enterobacteriaceae isolates are provided in Table 1.

Within the course defined in our study, 117 non-fermentative and the other Gram-negative rod-shaped clinical bacterial samples were simultaneously identified with both systems. Both systems were 100% consistent in their identifications of these 117 samples at the genus level, and 98.2% consistent at the species level. Achromobacter isolates that were identified by the Phoenix 100 at the genus level could be identified with MALDI-TOF MS at the species level. A strain identified as Acinetobacter baumannii by the Phoenix 100 was identified as Acinetobacter nosocomialis by the MALDI-TOF MS. In addition, a strain identified as Acinetobacter lwoffii by the Phoenix 100 was identified as Acinetobacter baumannii by the MALDI-TOF MS (Table 2). Because no satisfactory results were obtained from studies that were conducted with other systems, we identified these two isolates at the genus level.

Catalase-positive, Gram-positive cocci identifications were consistent between both systems (100% at the genus level; 92.1% at the species level). All of the 81 isolates identified as Staphylococcus aureus by the Phoenix 100 device were identified in the same way by MALDI-TOF MS. An isolate identified as Staphylococcus epidermidis by the Phoenix 100 was identified as Staphylococcus aureus by the MALDI-TOF MS. The applied coagulase test was positive, and BBL Crystal GP and VITEK II GP confirmed the MALDI-TOF MS result. An isolate that was identified as Staphylococcus saprophyticus by the Phoenix 100 device was identified as Staphylococcus epidermidis by the MALDI-TOF MS system. The result of the applied novobiocin test was positive at 23 mm; BBL Crystal GP and VITEK II GP confirmed the MALDI-TOF MS result. The Staphylococcus genus bacteria that presented different results at the species level in both systems could not be resolved at the species level; this isolate was identified as coagulase-negative Staphylococcus through the application of coagulase tests (Table 3).

Catalase-negative, Gram-positive cocci identifications were consistent between both systems (100% at the genus level; 95.1% at the species level). An isolate that was identified as Enterococcus faecalis by the Phoenix 100 system was identified as Enterococcus casseliflavus by the MALDI-TOF MS. Further tests using the VITEK II Compact and BBL Crystal GP devices confirmed the result of MALDI-TOF MS, identifying this isolate as Enterococcus casseliflavus . A strain that was identified as Streptococcus pyogenes by the Phoenix 100 device was identified as Streptococcus agalactiae by MALDI-TOF MS. A bacitracin test applied to this isolate was found 8 mm and VITEK II Compact and BBL Crystal GP devices identified the isolate as Streptococcus agalactiae . The isolates of the Streptococcus genus bacteria, identified differently by the two systems, were resolved at the species level (Table 4).

We tested 29 yeast isolates with both the Phoenix 100 and the MALDI-TOF MS. The Phoenix 100 identified 18 of these isolates as Candida albicans , consistent with the MALDI-TOF MS test results. Two strains identified differently at the species level by the two systems were identified as Candida species (Table 5).

Discussion

Automatic systems for microorganism identification and antibiotic susceptibility testing are increasingly mainstream, since their adoption began in the 1990s. These systems include the BD Phoenix (Becton Dickinson), VITEK1-2 (BioMérieux), MicroScan WalkAway SI (Dade Behring), and Sensititre ARIS 2x (TREK). These systems require 24 h to pass between culturing and identification. In situations requiring urgent identification and treatment, such as sepsis, this waiting period is crucial, with each additional hour of delay increasing the hospitalization duration, costs, and mortality rate [5, 6].

Over the last 20 years, mass spectrophotometry has emerged as an important tool for the identification and analysis of proteins. Following this, emerged the MALDI-TOF MS, which is used in the field of clinic microbiology. Here, proteins are measured as small protein fragments. Ions are decomposed after the ionization, and mass measurement is performed. Thus, proteomic fingerprints specific to microorganisms are leaned on the iron molds and typed [6, 9].

Simultaneously using a Phoenix automated system and a MALDI-TOF MS system, we identified 476 Gram-negative bacterial isolates. Identifications of the Gram-negative isolates were consistent between these two systems. Measurements from the Phoenix and the MALDI-TOF MS were consistent for frequently isolated Gram-negative genera such as, Klebsiella , Enterobacter , and Proteus ; this agreement is also reported in the literature. Bizzini et al. [1] compared conventional systems with the MALDI-TOF MS system, and found that the different systems of identification revealed highly consistent results (97% and 100% for genera such as Klebsiella , Proteus , Morganella , and Salmonella ). In another study, Wang et al. [10] compared conventional phenotypic identifications with those produced by MALDI-TOF MS, and reported a consistency of up to 100% for Enterobacteriaceae members. This study confirms our results.

Non-fermentative Gram-negative taxa such as Pseudomonas aeruginosa , Acinetobacter , and Stenotrophomonas were also identified with high consistency by both of our systems. A recent study by Almuzara et al. [11] compared MALDI-TOF MS identifications with identifications made using conventional methods for 396 clinical isolates of non-fermentative Gram negative bacilli. These researchers identified, with nearly 100% consistency, non-fermentative bacteria such as Pseudomonas aeruginosa , Stenotrophomonas maltophilia , and Acinetobacter baumannii , which are often isolated in clinical microbiology laboratories. Another study evaluated non-fermentative Gram-negative bacilli with by both techniques, correctly identifying 549 out of 559 clinical isolates [12]. The rates given in these studies seem consistent with our results.

Only five of the 476 Gram-negative isolates (1%) in the present study were assigned different identifications by our two systems. Generally, further identity tests using conventional techniques confirmed the MALDI-TOF MS assignment. Isolates with inconsistent identifications could be retested and examined with MALDI-TOF MS because of the low operating cost of this system. In these repeat experiments, the MALDI-TOF MS replicated its previous assignments. Comparative studies conducted with other automatized systems report that, besides being quick and cheap, MALDI-TOF MS produced more accurate results than competing systems [13, 14]. In our study, two Achromobacter and two Salmonella isolates could not be identified with the Phoenix 100 system at the species level MALDI-TOF MS identified these strains at the species level. Identifying Salmonella at the species level is particularly difficult for the Phoenix 100 system. Consistent with our results, the literature indicates that MALDI-TOF MS presented better results compared to the Phoenix in the identification of Salmonella species [15].

All of our study’s Gram-positive cocci were identified with 100% consistency at the genus level and 93.1% consistency at the species level. Both systems produced identical assignments for Staphylococcus aureus , a pathogen frequently isolated in clinic microbiology laboratories. Schulthess et al. [16] obtained 99.6%–100% concordance between traditional identification methods and MALDI-TOF MS-based methods for catalase positive Gram-positive cocci, and 94.7%–100% agreement for catalase negative Gram-positive cocci. Schulthess et al. stated that the consistency rate for all Gram-positive cocci is 96.9%. El-Bouri et al. [17] report 100% identical results at the type level in Staphylococcus aureus and coagulase negative Staphylococcus . Strong agreement was discovered within enterococci and streptococci. The Phoenix 100 and MALDI-TOF MS systems seem to produce quite consistent identifications of Gram-positive cocci. Previously reported results seem to be similar to the results of our study [1, 10, 14, 18]. However, some inconsistent identifications are observed in coagulase-negative Staphylococci at the species level [19]. We came across this situation in our study. We observed in our study that the MALDI-TOF MS system presented more accurate results in our confirmation studies.

Candida infections are a major constituent of hospital-based infections. Accordingly, treatment strategies that will be applied to these infections are quite important [20]. Since the required duration of incubation for Candida isolates are longer than those of other microbes, these cultured yeasts must be identified quickly. Although the number of yeast isolates that we tested is small in comparison to the number of bacterial isolates that we tested, we observed a similar level of methodological agreement for yeast isolates as we observed for bacterial isolates. Twenty-seven of the 29 Candida isolates (93.1%) were assigned to the same species by both the Phoenix 100 and the MALDI-TOF MS. Candida albicans are among the fungal pathogens that most frequently cause a disease in humans [21]. We observed that both systems gave exactly compatible results in the identification of the Candida albicans strains. Chao et al. [22] reported that the Phoenix 100 and the MALDI-TOF MS gave 100% identical results for Candida albicans . In this study’s tests on the Candida genus, the identification rate was reported as 92.2% for the Phoenix 100 system and 98.6% for the MALDI-TOF MS system. When Dhiman et al. [23] compared a MALDI-TOF MS system with phenotypic methods, they obtained identification rates similar to the results obtained in the present study. We did not confirm two strains that were identified differently in our study. However, all of the identifications that we repeated three times and conducted with the MALDI-TOF system exhibited the same results. Consistent with previous reports [24–26], we found that the MALDI-TOF MS system made Candida assignments more quickly, accurately, and reliably than competing systems.

Today disadvantages of the MALDI-TOF system include unavailability of standardized antibiotic susceptibility testing, absence of sufficient information for direct identification from clinic samples, insufficient studies on antibiotic resistance mechanisms, and limitations to the proteomic fingerprint database [5, 6]. It is desirable that this system, which is beginning to enjoy widespread and effective use, becomes applicable for study of microbial resistance mechanisms; indeed, intensive research is underway on this matter [27, 28].

In conclusion, we observed strong concordance between the species identifications produced by the Phoenix 100 and the Microflex LT MALDI-TOF MS devices. However, the microorganism identifications, which took 6–24 h using the Phoenix system, were completed in about 1 min using the MALDI-TOF MS. Moreover, some identifications that could not be done with the Phoenix system at the species level could be done with MALDI-TOF MS at the species level. A separate kit cost is at issue for each identification using the Phoenix 100 system, whereas the MALDI-TOF MS can identify each isolate several times, without additional cost. For these reasons, laboratories currently using the Phoenix 100 device may soon be interested in replacing it with a MALDI-TOF MS system. If their experiences reflect our own, then these laboratories will not encounter significant issues while using both systems during the transition period.

Author contributions: All the authors have accepted responsibility for the entire content of this submitted manuscript and approved submission.

Research funding: None declared.

Employment or leadership: None declared.

Honorarium: None declared.

Competing interests: The funding organization(s) played no role in the study design; in the collection, analysis, and interpretation of data; in the writing of the report; or in the decision to submit the report for publication.