Kocuria rhizophila Misidentified as Corynebacterium jeikeium and Other Errors Caused by the Vitek MS System Call for Maintained Microbiological Competence in the Era of Matrix-Assisted Laser Desorption Ionization–Time of Flight Mass Spectrometry

LETTER

Matrix-assisted laser desorption ionization–time of flight (MALDI-TOF) mass spectrometry (MS) is rapidly becoming a routine method for identification of microorganisms in the United States. Early studies showed strong performance for both commonly used systems, with identification rates of >90% and misidentification rates of <2% in some studies (1–4). One of the major advantages of using MALDI-TOF for identification is that limited information is needed prior to submitting the isolate for analysis. Indeed, laboratories may use this technology without any testing (e.g., Gram stain, oxidase) for identification of an isolate other than the use of MALDI-TOF itself. This is preferable in a time when the number of qualified microbiologists is dwindling. However, this limited approach poses a potential problem in that incorrect identifications may occur without the careful review and comparison of MALDI-TOF results and culture phenotype. The Vitek MS system (bioMérieux, Durham, NC) has been FDA cleared for the identification of nearly 200 species of bacteria and yeast, sometimes referred to as “claimed” organisms. The cleared (in vitro diagnostics [IVD]) system also reports the identification of approximately 500 additional organisms that were not cleared by the FDA for primary identification as “unclaimed.” These identifications should not be used without additional validation of the organism identification accuracy by the Vitek MS or by using an alternative identification method. We report a case of Kocuria rhizophila bacteremia that was identified by 16S rRNA gene sequencing after being incorrectly identified as Corynebacterium jeikeium bacteremia by Vitek MS (v2.0). We also report additional instances of misidentification of organisms by Vitek MS.

A 28-year-old female with hypothyroidism and short gut syndrome who has been dependent on total parenteral nutrition (TPN) for more than 20 years presented with electrolyte imbalance and metabolic acidosis. Blood cultures were drawn from a line as well as peripherally to rule out contributory infectious causes. Her two initial peripheral blood cultures were positive in the aerobic bottles. A subsequent line culture was also positive. The first aerobic bottle (Bactec Lytic; Becton Dickinson, Sparks, MD) gave a positive signal 29 h after being loaded onto the instrument (Bactec FX; Becton Dickinson, Sparks, MD). Gram stain analysis of the bottle revealed Gram-positive cocci in clusters, and the bottle was subcultured to sheep blood agar (SBA) and chocolate agar. The following day, pale yellow colonies were growing on both culture media. A portion of a single colony from SBA was identified by the Vitek MS system as Corynebacterium jeikeium. Because the identification did not match the colony morphology or Gram stain result, a second colony was analyzed to ensure that the correct patient sample was spotted the first time. The results of the second attempt also generated an identification of Corynebacterium jeikeium. There were no error messages produced by the system, and all manufacturer-recommended quality control tests had been passed. The isolate was identified by partial 16S rRNA gene sequencing (5) as Kocuria rhizophila (470/471 identities); Kocuria carniphila (480/492) and Kocuria varians (467/474) were more distant BLAST matches. Corynebacterium jeikeium strain A376/84 had only 86% (429/499) identity.

In addition to the case presented described above, we have encountered numerous other scenarios from multiple different specimen types where an organism gave a single distinct identification by the Vitek MS system that was later shown to be incorrect (Table 1). Most of the misidentified organisms did not have simple phenotypic characteristics associated with the proposed organism, often having a discrepant Gram stain morphology, allowing easy dismissal of the incorrect identification. Some misidentifications, however, required a more detailed knowledge of the microbiology, including the expected colony morphology, oxidase reaction (Ralstonia sp.), hemolysis (Bacillus sp.), and even growth rate (Facklamia sp.). These incorrect MS results could have been reported easily if the technologists had not carefully checked the identification with known phenotypic characteristics and performed additional tests to confirm improbable identifications. A majority of the MS misidentifications were identifications that are unclaimed according to the FDA clearance data on the Vitek MS system, though some (genus-level identifications of Streptococcus spp. and Bacillus spp.) had been internally validated for use. This demonstrates the need for a thorough validation of unclaimed organisms and a high index of suspicion when they are called even after validation. Additionally, caution must be used in reporting identifications from anaerobic cultures, as these were the majority of the misidentifications in our study. Taken together, even though the frequency of misidentification is very low, representing less than 0.5% of identifications in our laboratory, each result needs to be carefully assessed by a microbiologist who is familiar with phenotypic and basic biochemical characteristics in order to prevent errors from being reported and potentially poor patient outcomes.

TABLE 1.

Vitek MS misidentifications^a

Organism category and phenotype	MALDI ID	Reported identification
Claimed
Aerobic yellowish GPC	Corynebacterium jeikeium	Kocuria rhizophilab
Anaerobic GPR	Staphylococcus cohnii subsp. cohnii	Anaerobic GPR
Anaerobic GPC in chains	Staphylococcus epidermidis	Anaerobic GPC
Anaerobic GPR	Ochrobactrum anthropi	Anaerobic GPR
Unclaimed
Aerobic oxidase negative, lactose fermenting, GNR	Ralstonia insidiosa	Klebsiella pneumoniaec,d
Aerobic alpha-hemolytic GPC	Facklamia hominis	Enterococcus faeciumc,d
Aerobic beta-hemolytic large GPR	Bacillus anthracis	Bacillus cereus groupd
Aerobic slow-growing small GPR	Mycobacterium kansasii	Actinomyces gerencseriaeb
Anaerobic small GPR	Streptococcus pseudopneumoniae	Actinobaculum schaaliib
Anaerobic GPC	Bacillus sporothermodurans	Ruminococcus gnavusb
Anaerobic GPR	Serratia rubidaea	Anaerobic GPR

^aID, identification; GPR, Gram-positive rod; GPC, Gram-positive cocci; GNR, Gram-negative rod.

^bIdentified by 16S rRNA gene sequencing.

^cIdentified by Vitek 2.

^dPresent in Vitek MS v2.0 database.