Performance of the Vitek 2 system software version 5 . 03 in the bacterial identifi cation and antimicrobial susceptibility test : evaluation study of clinical and reference strains of Gram-positive cocci

Introduction. The genera Enterococcus, Staphylococcus and Streptococcus are recognized as important Gram-positive human pathogens. The aim of this study was to evaluate the performance of Vitek 2 in identifying Gram-positive cocci and their antimicrobial susceptibilities. Methods. One hundred four isolates were analyzed to determine the accuracy of the automated system for identifying the bacteria and their susceptibility to oxacillin and vancomycin. Results. The system correctly identifi ed 77.9% and 97.1% of the isolates at the species and genus levels, respectively. Additionally, 81.8% of the Vitek 2 results agreed with the known antimicrobial susceptibility profi les. Conclusion. Vitek 2 correctly identifi ed the commonly isolated strains; however, the limitations of the method may lead to ambiguous fi ndings.

Gram-positive cocci are widely distributed as part of the normal fl ora in humans; however, some species are recognized as major human pathogens and cause a large variety of infections worldwide.These microorganisms are frequently isolated from bloodstream infections, skin and soft tissue infections, sepsis, urinary tract infections and lower respiratory tract infections 1 .
Automated bacterial identifi cation in the clinical laboratory provides a rapid and reliable diagnosis for most pathogens involved in infectious diseases.A previous study demonstrated the satisfactory performances of the automated methodologies, resulting in their use in routine practice with a highly acceptable level of identification accuracy; additionally, automated identification enabled the interpretation of antimicrobial susceptibility tests for the correct treatment of patients 2 .
The aim of the present study was to evaluate the performance of the Vitek 2 automated system in the identifi cation of bacteria and antimicrobial susceptibilities of Gram-positive cocci isolates recovered from clinical samples and reference strains.
The study was performed at Laboratório de Cocos Gram-Positivos (LCGP) of the Universidade Federal de Ciências da Saúde de Porto Alegre (UFCSPA) and Laboratório Qualità, Rio Grande do Sul, Brazil.The isolates included in the present study were selected from the strain collection belonging LCGP.A total of 104 isolates of Gram-positive cocci were analyzed, including 29 reference strains selected from the American Type Culture Collection (ATCC) and 75 clinical strains isolated from different patients; these strains included Staphylococcus coagulase-negative (n=36), Enterococcus spp.(n=33) and Staphylococcus aureus (n=6).All of these strains have been previously characterized by the LCGP with regard to their virulence factors using molecular methods and susceptibility profi les and were identifi ed at the species level using conventional reference methods 3,4 .For the identifi cation of staphylococci, the following characteristics were tested: catalase; colony morphology and pigmentation; Gram stain; hemolysis; susceptibility to novobiocin; polymyxin B; fosfomycin and deferoxamine; enzyme activity of arginine arylamidase; ornithine decarboxylase and urease; and acid production from trehalose, mannitol, mannose, xylose, cellobiose, arabinose, maltose, lactose, sucrose and raffi nose.For the enterococcal isolates, the following phenotypic characteristics were evaluated: catalase; colony morphology and pigmentation; esculin hydrolysis in the presence of 40% bile; growth in 6.5% NaCl; motility; and acid production from mannitol, sorbose, arginine, arabinose, sorbitol, raffi nose, sucrose, pyruvate and methyl-glucopyranoside (MGP).Susceptibility to oxacillin and vancomycin was evaluated using the broth microdilution and disk-diffusion reference methods according the Clinical and Laboratory Standards Institute (CLSI) documents M7-A6 5 and M100-S21 6 , respectively.Vitek 2 (bioMérieux, Marcy L'E`toile, France) bacterial identifi cation and antimicrobial susceptibility testing (AST) methods were evaluated according to the manufacturer's instructions.The comparative analysis of the bacterial identifi cation by Vitek 2 and the conventional methods was performed, and the accuracy (percentage of matched identifi cation) was characterized.The AST results evaluated using the Vitek 2 automated system were compared to those obtained using the reference method, and the agreement rates were analyzed.The type of error was evaluated using the following criteria: I) very major error (VME) defi ned as a resistant bacterial isolate appearing susceptible by Vitek 2; II) major error (ME) defi ned as a susceptible isolate displaying a resistant profi le by Vitek 2; or III) minor error (MiE) defi ned as microorganisms for which the AST reference indicated intermediate resistance and susceptibility or resistance by an automated AST test.
The Vitek 2 system correctly identifi ed 81 (77.9%) and 101 (97.1%) of the 104 bacterial isolates at the species and genus levels, respectively.Additionally, Staphylococcus aureus was identifi ed more quickly than the other staphylococcal isolates (p<0.05)(Table 1).
The Vitek 2 automated method was accurate for specieslevel identifi cation of the commonly isolated Gram-positive cocci.All of the clinical isolates of S. aureus and Enterococcus faecalis were correctly identified.The bacterial isolates were misidentified or showed low-level discrimination in the clinical samples of Staphylococcus epidermidis (25%), Staphylococcus haemolyticus (66.7%),Staphylococcus hominis (16.7%),Staphylococcus saprophyticus and Staphylococcus warneri (both 20%).In the Enterococcus strains, Enterococcus avium and Enterococcus durans were misidentifi ed as other enterococcal isolates, and the discordant results were found 50% of Enterococcus casseliflavus, 75% of Enterococcus faecium and 25% of Enterococcus gallinarum isolates.Among the 29 reference strains, a defi nitive species identifi cation was provided by the automated system for 22 (75.9%)strains.The incorrect identifi cation or low-level discrimination results are listed in Table 2.The minimal inhibitory concentrations (MICs) of oxacillin and vancomycin generated by the automated system were compared according to the error type, and the results are shown in the Table 3.In total, 45 of the 55 (81.8%) bacterial isolates agreed with the identifi cations made using the conventional antimicrobial susceptibility profi le.
At the species level, the automated Vitek 2 system was able to identify the commonly isolated staphylococci and enterococci strains, such as S. aureus and E. faecalis, respectively, and all of these isolates, as well as the streptococci reference strains, were correctly identifi ed.These results were in agreement with those reported by Chatzigeorgiou 7 et al., who evaluated the performance of the Vitek 2 system in comparison with other automated systems 7 .However, in the present study, the Vitek 2 version 5.03 software was not able to correctly identify all of the S. epidermidis isolates, with only 71.4% of the isolates matched at the species level.
S. haemolyticus is frequently isolated from blood cultures and has a tendency to develop resistance to multiple antimicrobial drugs 8 , similar to S. warneri, which is another CoNS that can cause catheter-related bacteremia and native and prosthetic valve endocarditis 9 .Two isolates of this bacterial species were misclassifi ed as S. warneri by the Vitek 2 system.Two S. saprophyticus were misidentifi ed as S. warneri and S. cohnii subs.urealyticus by the Vitek 2 automated system.According to the biochemical profi le of these isolates, major error in the bacterial identifi cation can be evaluated by novobiocin susceptibility, as S. saprophyticus and S. cohnii are resistant, while S. warneri is susceptible 3 10 .
In the present study, the agreement identifi cation rate of the enterococci isolates was similar to a previous study, and all of the E. faecalis isolates were correctly identifi ed 11 .However, 3 of the 5 E. faecium strains were misclassifi ed or showed low-level discrimination.The bacterial identifi cation of the Enterococcus isolates is important because this genus includes some of the most important multidrug-resistant organisms in healthcare-associated infections.These isolates usually affect patients who are debilitated by other concurrent illnesses or are undergoing prolonged hospitalization.Including the E. faecium strains, these pathogens have the ability to succeed in the hospital environment 12 .
The automated system was able to identify 11 Enterococcus, 26 Staphylococcus and 37 Streptococcus species and subspecies using the Gram-Positive Card (GP-Card -bioMeriéux).New Gram-positive cocci species have been identifi ed from clinical specimens in recent years, and these species exhibit phenotypic profi les similar to other staphylococcal strains 13 .Although these species were not included in our study, the tested strains allowed us to evaluate the accuracy of the automated system in identifying the common species isolated in the clinical setting.
Our data indicated that the Vitek 2 system provided inaccurate susceptibility test results for oxacillin and vancomycin, as the agreement rate with the reference method was very low, and the error rates, mainly VMEs and MEs, were higher compared to other studies 14 .Although our study has evaluated the antimicrobial susceptibility of commonly isolated strains, such as S. aureus and E. faecalis, most of the isolates were microorganisms that belong to coagulase-negative staphylococci (CoNS) and non-faecalis enterococci.A previous meta-analysis reported that the discordant results in bacterial identifi cation could be explained by the metabolic rate of the bacterial isolates, as slow metabolism can lead to ambiguous reactions during the short incubation times used by the automated instruments 15 .These characteristics of the samples could interfere with the AST automated method, as the staphylococci and enterococci isolates exhibited very major errors.
In conclusion, the Vitek 2 Compact system software version 5.03 correctly identifi ed the commonly isolated Gram-positive cocci; however, the limitations of the method may lead to ambiguous fi ndings and the inability to identify uncommon microorganisms.Therefore, additional phenotypic tests may be necessary to identify some strains at the species level.Additionally, critical inquiry of the AST results reported by the automated method showed discrepancies in the antimicrobial susceptibilities that might occur in uncommon isolated pathogens.

Paim
TGS et al -Vitek 2 evaluation of Gram-positive cocci isolates

TABLE 1 -Descriptive measures from the automated system Vitek 2 compact.
a mean (SD -standard deviation); analysis of variance (ANOVA) followed by Student-Newman-Keuls post hoc test.The data were tested for normal distributions using the Kolmogorov-Smirnov test.b p=0.71(among the genus group).c p<0.05 (among the staphylococcal species).d p=0.06 (among the enterococcal species).e p=0.41 (between the Gram-positive cocci families).CoNS: coagulase-negative Staphylococcus non-epidermidis.

TABLE 2 -Bacterial species reported by the Vitek 2 System that were misidentifi ed or showed low discrimination.
Subs: subspecie.

TABLE 3 -Agreement of the antimicrobial susceptibility tests and the type of error among the staphylococci and enterococci isolates. a
Streptococci reference strains were not tested.bSusceptibility to oxacillin.Susceptibility to vancomycin.VME: very major error; ME: major error; MiE: minor error.phenotypicprofi les and exhibit low discriminatory power in bacterial identifi cation; both species are urease positive and novobiocin resistant.The clinical laboratory exams should be able to differentiate these strains, as S. cohnii subs.urealyticus is recognized as a pathogen in infectious diseases, such as endocarditis, septicemia and urinary tract infections; however, previous study have reported diffi culty in routinely identifying this microorganism a