Abstract
In a stress test, the recently introduced RapID CB Plus system (Remel Inc. [formerly Innovative Diagnostic Systems], Norcross, Ga.) was challenged with a diverse set of gram-positive rods comprising 345 strains of coryneform bacteria and 33 strains of Listeria spp. representing a total of 49 different taxa. Overall, within 4 h, the system correctly identified 80.9% of the strains on the species level and 12.2% of the strains on the genus level. Only 3.7% strains were misidentified, and for 3.2% of the strains no identification was provided. Difficulties with the system were mainly due to occasional uncertainties in reading reactions for acid production from carbohydrates and, to a lesser extent, aminopeptidase reactions. It is concluded that the system may also perform well under the conditions of a routine clinical laboratory.
Within the last few years, clinically relevant coryneform bacteria (i.e., aerobically growing, asporogenous, non-partially acid-fast, irregular, gram-positive rods) have been encountered with increasing frequency in human specimens and a plethora of new taxa belonging to the coryneform bacteria has been described (5). As a result, clinical microbiologists are faced with an increasing need to identify a larger number and a more diverse group of coryneform bacteria. Traditional biochemical identification systems for coryneform bacteria are based on macrodilution tube methods which require a large inoculum, are dependent upon the growth of the test organism, and often give delayed results. Thus, there is a clear need for reliable commercial rapid identification systems for coryneform bacteria. The RapID CB Plus system is a new identification system for coryneform bacteria which was developed by Remel Inc. (formerly Innovative Diagnostic Systems [IDS] Norcross, Ga.) and which is based on the well-established technology of the IDS RapID product line. A description of the prototype of this system was presented in 1996 (2); subsequently, the system became commercially available and the database has been expanded. We therefore considered it appropriate to evaluate the RapID CB Plus panel by a comprehensive stress test rather than by a weighted laboratory profile (7) in order to challenge the depth of the system’s database. This evaluation is the first one from Europe on this new identification system for coryneform bacteria and includes a more comprehensive spectrum of challenge organisms than a previous study (6).
MATERIALS AND METHODS
Strains, media, and growth conditions.
A total of 378 strains of mainly coryneform bacteria were included in the present study. It also included 33 strains of Listeria spp. and 4 strains of Rhodococcus equi strains as noncoryneform bacteria. Approximately 10% of all strains were fresh clinical isolates identified in the Department of Medical Microbiology, University of Zürich (DMMZ), and the remaining 90% came from the culture collection of the same institution. All strains used in the study had been extensively characterized by applying phenotypic, chemotaxonomic, and molecular genetic methods (5, 8), and some of the species tested had recently been defined by workers from DMMZ. The type strain of each species tested was also included.
All strains were subcultured twice before testing to allow optimal expression of bacterial enzymes. Strains were cultured on Columbia agar (Difco, Detroit, Mich.) supplemented with 5% sheep blood (SBA), and cells were harvested after 24 h of incubation at 37°C in a 5% CO2-enriched atmosphere.
RapID CB Plus system.
The RapID CB Plus system is composed of 4 rapid carbohydrate utilization tests and 14 single-substrate enzymatic tests (6). A catalase test and the observation of the production of yellow pigment are supplementary tests used in the system. It was inoculated with suspensions of test organisms corresponding to a turbidity equal to that of a no. 4 McFarland standard as described previously by Hudspeth et al. (6). The suspensions were used within 10 min of preparation, and special attention was given to an even distribution of the inoculation fluid in each reaction well. The inoculation fluid was cultured in parallel on SBA for 24 h at 37°C in order to check for purity (this is recommended only during an evaluation). The RapID CB Plus panels were incubated for 4 h at 37°C in ambient air. After incubation, the needed reagents were added (6) and the panels were independently read by two different persons. In case of ambiguous results (which occurred for fewer than 5% of the strains examined), a third person was asked to read the panels. A seven-digit numerical code was derived after interpretation of the reactions. This code was entered into the IDS Electronic Code Compendium, which contains RapID CB Plus identification software (March 1998 version) which was run on a personal computer.
As recommended by the manufacturer, three quality control strains were tested for each RapID CB Plus lot used. In addition, we tested two quality control strains, Corynebacterium pseudodiphtheriticum ATCC 10701 and Arcanobacterium pyogenes ATCC 19411, after the testing of every 50 panels.
Reporting of results.
The RapID CB Plus reports consisted of a probability value and a biofrequency score (a measure of typicality to the ideal) for each code processed in the study. Up to four identification choices were provided; however, in our evaluation, the biofrequency score as well as any third or fourth identification choice given were not considered. There were five different categories of results: (i) “identification on the species level with ≥95% probability” meant that the correct identification was given as the first choice with a probability of greater or equal to 95%; (ii) “identification on the species level with <95% probability” meant the same as for category (i) but that the identification probability was less than 95%; (iii) “identified on the genus level” meant that all identifications given belonged to the same genus and the probability of the first choice was less than 95%; (iv) “no choices” meant that there was no sufficient probability to allow identification; and (v) “misidentified” meant that there was a disagreement between the reference identification and the identification provided by the system.
RESULTS
Overall, the RapID CB Plus system correctly identified 73.5% of all strains on the species level with ≥95% probability, 7.4% of all strains on the species level with <95% probability, and 12.2% of all strains on the genus level (Table 1). Only 3.7% of all strains tested were misidentified, and for 3.2% of the strains no identification was provided. The RapID CB Plus system correctly identified 88.5% (first choice) of the most frequently encountered coryneform bacteria in the routine clinical laboratory (i.e., Corynebacterium amycolatum, CDC coryneform group G bacteria, Corynebacterium jeikeium, Corynebacterium minutissimum, Corynebacterium pseudodiphtheriticum, Corynebacterium striatum, and Corynebacterium urealyticum).
TABLE 1.
Results of identification with the RapID CB Plus system
| Taxon | No. (%) of strains tested
|
|||||
|---|---|---|---|---|---|---|
| Total | Identified on the species level with ≥95% probability | Identified on the species level with <95% probability | Identified on the genus level | Not identified | Misidentified | |
| Actinomyces neuii | 10 | 10 | ||||
| Genus Arcanobacterium | ||||||
| A. bernardiae | 8 | 8 | ||||
| A. haemolyticum | 11 | 10 | 1 | |||
| A. pyogenes | 5 | 5 | ||||
| Arthrobacter cumminsii | 3 | 3 | ||||
| Genus Brevibacterium | ||||||
| B. casei | 10 | 10 | ||||
| Brevibacterium spp. | 10 | 2 | 1 | 7 | ||
| Cellulomonas spp. | 8 | 8 | ||||
| Genus Corynebacterium | ||||||
| C. accolens | 3 | 3 | ||||
| C. afermentans subsp. afermentans | 9 | 4 | 1 | 4 | ||
| C. afermentans subsp. lipophilum | 7 | 3 | 4 | |||
| C. amycolatum | 14 | 8 | 2 | 4 | ||
| C. argentoratense | 10 | 7 | 1 | 2 | ||
| C. auris | 10 | 10 | ||||
| C. bovis | 5 | 3 | 2 | |||
| C. confusum | 3 | 3 | ||||
| C. coyleae | 3 | 3 | ||||
| C. durum | 4 | 4 | ||||
| CDC group F-1 bacteria | 8 | 7 | 1 | |||
| CDC group G bacteria | 18 | 9 | 6 | 1 | 2 | |
| CDC group I-1 bacteria | 5 | 5 | ||||
| C. diphtheriae | 13 | 11 | 2 | |||
| C. glucuronolyticum | 12 | 9 | 3 | |||
| C. jeikeium | 12 | 12 | ||||
| C. minutissimum | 11 | 7 | 3 | 1 | ||
| C. mucifaciens | 2 | 2 | ||||
| C. propinquum | 8 | 5 | 3 | |||
| C. pseudodiphtheriticum | 11 | 11 | ||||
| C. pseudotuberculosis | 2 | 2 | ||||
| C. renale | 3 | 3 | ||||
| C. striatum | 12 | 12 | ||||
| C. ulcerans | 3 | 3 | ||||
| C. urealyticum | 18 | 18 | ||||
| C. xerosis | 5 | 5 | ||||
| “Corynebacterium aquaticum” | 4 | 1 | 3 | |||
| Dermabacter hominis | 12 | 12 | ||||
| Erysipelothrix rhusiopathiae | 1 | 1 | ||||
| Exiguobacterium acetylicum | 5 | 4 | 1 | |||
| Genus Listeria | ||||||
| L. grayi-L. murrayi | 4 | 3 | 1 | |||
| L. innocua | 6 | 3 | 3 | |||
| L. ivanovii | 5 | 5 | ||||
| L. monocytogenes | 11 | 1 | 8 | 2 | ||
| L. seeligerii | 3 | 3 | ||||
| L. welshimeri | 4 | 3 | 1 | |||
| Microbacterium spp. | 10 | 8 | 2 | |||
| Oerskovia spp. | 9 | 5 | 3 | 1 | ||
| Rhodococcus equi | 4 | 2 | 1 | 1 | ||
| Rothia spp. | 12 | 10 | 2 | |||
| Turicella otitidis | 12 | 12 | ||||
| Total | 378 | 278 (73.5) | 28 (7.4) | 46 (12.2) | 12 (3.2) | 14 (3.7) |
The difficulties encountered with the RapID CB Plus system for the strains which were not identified or misidentified are given in Table 2. These strains were randomly distributed over different taxa. Most of the problems were due to ambiguous carbohydrate reactions, and for some taxa particular reactivities in enzymatic reactions were not included in the database or an enzymatic reaction was difficult to read.
TABLE 2.
Difficulties encountered with the RapID CB Plus identification system
| Taxon (no. of strains) | Problem |
|---|---|
| Arcanobacterium pyogenes (5) | Pyrrolidonyl aminopeptidase-positive strains not recognized |
| Corynebacterium afermentans subsp. afermentans (5) | False-positive reactions for acid production from carbohydrates, false-positive aminopeptidase activities |
| Corynebacterium argentoratense (2) | False-negative reaction for acid production from glucose |
| Corynebacterium bovis (2) | False-negative β-galactosidase reaction |
| CDC group G bacteria (2) | Proline aminopeptidase-negative strains not recognized |
| Corynebacterium glucuronolyticum (3) | Various enzymes (alkaline phosphatase-positive strains, β-glucosidase-positive strains) |
| Corynebacterium minutissimum (1) | Pyrrolidonyl aminopeptidase-positive strain not recognized |
| “Corynebacterium aquaticum” (3) | Acid production from carbohydrates not recognized |
| Exiguobacterium acetylicum (1) | Maltose-negative and esterase-positive strain not recognized |
| Listeria grayi-Listeria murrayi (1) | False-negative N-acetyl-β-glucosaminidase reaction |
| Microbacterium spp. (2) | Identified as the closely related Cellulomonas spp. and Oerskovia spp. |
| Oerskovia spp. (1) | Proline aminopeptidase-negative strain not recognized |
| Rhodococcus equi (1) | Glucose-positive and α-glucosidase-negative strain not recognized |
| Rothia spp. (2) | Glycosidase-positive strains not recognized |
The reproducibility of results was examined by retesting 20 randomly chosen strains, all of which gave identical results when they were retested. The results for the two quality control strains were always within the expected range.
The species included in the RapID CB Plus database but not covered in the present evaluation comprise Corynebacterium cystitidis, Corynebacterium kutscheri, Corynebacterium matruchotii, Actinomyces israelii, Actinomyces naeslundii, Actinomyces odontolyticus, Actinomyces viscosus, and Bifidobacterium spp.
DISCUSSION
The present evaluation of the RapID CB Plus panel was a stress test rather than a weighted laboratory profile (7) in order to explore the full depth of the system’s database. It is most likely that the RapID CB Plus system will also perform well under the conditions of a routine clinical laboratory since it performed well under the more stringent conditions of our stress test, with only 3.7% of the strains being misidentified and 3.2% of the strains being not identified.
The RapID CB Plus system is, in general, not demanding manually. The most critical point with the RapID CB Plus system is the difficulty in reading some of the reactions. This may have resulted in some wrong identifications, which are listed in Table 2. In our experience, it is important to read the strips within 1 min after application of the reagents. For the carbohydrate utilization reactions it is important to realize that orange colors must be considered negative results. However, some of the nonfermenting bacteria (Corynebacterium afermentans subsp. afermentans, “Corynebacterium aquaticum”) utilized sugars with weak acid production, which may have resulted in false-positive reactions (Table 2). For the enzymatic reactions it was important to realize that for the esterase reaction only a significant yellow color was positive, whereas for the phosphatase and the glycosidase reactions any development of a noticeable yellow color (i.e., even pale shades) signified positivity. These uncertainties in reading the panel may have led to false-positive and false-negative reactions (Table 2). We also observed that it seems to be critical not to overinoculate the system, which may result in false-positive reactions in some cases (data not shown).
There are only minor difficulties with the current RapID CB Plus database. CDC coryneform group I-1 bacteria do not need to be listed as a separate taxon since it has been shown that these strains are very similar to Corynebacterium striatum (1). It is unclear why Brevibacterium casei and Brevibacterium spp. are listed as separate taxa since nearly 95% of all brevibacteria found in clinical specimens are Brevibacterium casei strains (3). It is suggested that these two taxa should be summarized as Brevibacterium spp. “Corynebacterium aquaticum” is listed under the corynebacteria but should be moved to the irregular gram-positive bacilli since, as a bacterium with diaminobutyric acid as its cell wall diamino acid, it is certainly not a true Corynebacterium (5). The current database lists 98% of all Rothia spp. as catalase positive, but a very recent study demonstrated that up to one-third of all Rothia strains are catalase negative (9). Other minor modifications of the RapID CB Plus database which seem to be necessary (Table 2) include the introduction of pyrrolidonyl aminopeptidase-positive Arcanobacterium pyogenes and Corynebacterium minutissimum strains, proline aminopeptidase-negative CDC group G and Oerskovia strains, and glycosidase-positive Rothia strains. Difficulties with the identification of Listeria spp. on the species level only with commercial identification systems but not by other conventional tests have been reported before (4, 6).
The only other extant article on the RapID CB Plus system was recently published by Hudspeth et al. (6). Those investigators tested only 98 clinical strains, and 35 of the 57 taxa of the RapID CB Plus database were not included in their evaluation. In addition, their strain collection was not particularly representative of the specimens encountered in a routine clinical laboratory because 44 of the 98 clinical strains tested had been isolated from animal bite wounds. Important species like Corynebacterium diphtheriae and Corynebacterium pseudodiphtheriticum were not represented among the clinical strains. Some of the reference identifications in the study of Hudspeth et al. (6) were questionable. As an example, it was stated that Tween 80 was required for conventional carbohydrate tests of Brevibacterium spp.; however, brevibacteria are never lipophilic and rather strongly oxidize carbohydrates (5). Definitions of the criteria for classification of strains into different identification categories were not given. Hudspeth et al. (6) also tested 10 Bacillus strains with the panel in order to show that the kit will not identify strains of bacteria not included in the RapID CB Plus Code Compendium, although the manufacturer’s instructions for use state that spore-forming bacteria should not be tested with the system.
It was certainly not the aim of our study to compare the only two identification systems for coryneform bacteria which are nowadays significantly present on the market, i.e., the API Coryne system and the RapID CB Plus system. However, we are convinced that many clinical microbiologists are interested in this question, and we feel competent to judge the basic performance and use of the two systems since we have also recently evaluated the API Coryne system with database version 2.0 (4). It is desirable to have more than one identification system present on the market so that clinical microbiologists have a choice to use the better-performing and/or more cost-effective system or just simply the system that better fits the needs of the individual laboratory. The major difference between the two systems is that the RapID CB Plus system provides a definitive result after 4 h of incubation (because it tests performed enzymes and is not dependent on growth of the organisms), whereas the final identification is given after 24 h by the API Coryne system. The enzymes tested in the RapID CB Plus system include only a proprietary glycosidase, esterase, and proline, tryptophan, leucyl-glycine, and leucine aminopeptidases, whereas the API Coryne system exclusively includes a test for pyrazinamidase activity. The API Coryne version 2.0 database contains 54 taxa and the RapID CB Plus database contains 57 taxa. However, as discussed above, the inclusion of CDC group I-1 coryneform bacteria and Brevibacterium spp. in the RapID CB Plus system is not necessary. Although aerotolerant Actinomyces israelii and Bifidobacterium spp. exist, it is most unlikely that clinical microbiologists will encounter these taxa on aerobically incubated agar plates so that it is questionable whether they should be included in the RapID CB Plus database. Species not covered by the API Coryne system include Arthrobacter cumminsii, Actinomyces israelii, Actinomyces naeslundii, Actinomyces odontolyticus, Actinomyces viscosus, Bifidobacterium spp., Corynebacterium afermentans subsp. lipophilum, Corynebacterium confusum, Corynebacterium cystitidis, Corynebacterium durum, Corynebacterium matruchotii, Corynebacterium mucifaciens, Corynebacterium xerosis, and Exiguobacterium acetylicum, whereas the API Coryne system database contains both Actinomyces neuii subspecies, Actinomyces radingae, Actinomyces turicensis, Arthrobacter spp., Aureobacterium spp., Brevibacterium epidermidis, Corynebacterium macginleyi, Dietzia maris, Gardnerella vaginalis, Gordona spp., two Oerskovia species, Propionibacterium acnes, and Propionibacterium avidum, which are not included in the RapID CB Plus database. Reading of the RapID CB Plus enzymatic reactions is sometimes more difficult compared to reading of the API Coryne system reactions because of the occurrence of pale shades (see above). Reactions for acid production from carbohydrates are sometimes (for <5% of the strains tested) difficult to read in both systems. A disadvantage of both commercial identification systems is that for the inoculation of both systems a relatively heavy inoculum (RapID CB Plus system, McFarland no. 4 standard; API Coryne system, McFarland no. 6 standard) is required, so that it is often impossible to identify coryneform bacteria directly from primary growth plates. Another problematic feature of both commercial systems is that they suggest to the user that partially acid-fast bacteria (whose occurrence is often not initially considered by clinical microbiologists) could be correctly identified but partially acid-fast bacteria are sufficiently separated only by detection and analysis of mycolic acids (1, 5).
In conclusion, the authors of the present report consider the RapID CB Plus system to be a promising tool for the rapid identification of coryneform bacteria encountered in clinical specimens. Nevertheless, we want to emphasize that commercial identification systems should not distract clinical microbiologists from basic observations like colony morphology and odor and Gram staining results. In the few cases in which commercial identification systems do not provide a satisfactory result, strains (e.g., yellow-pigmented coryneform bacteria) should be sent to a reference laboratory.
ACKNOWLEDGMENTS
We thank L. A. Eriquez (Remel Inc., Norcross, Ga.) for kindly providing the RapID CB Plus identification kits and for support in data analysis.
G.F. is a recipient of an ESCMID research fellowship.
REFERENCES
- 1.de Briel D, Couderc F, Riegel P, Jehl F, Minck R. High-performance liquid chromatography of corynomycolic acids as a tool in identification of Corynebacterium species and related organisms. J Clin Microbiol. 1992;30:1407–1417. doi: 10.1128/jcm.30.6.1407-1417.1992. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Eriquez L A, Marler J K. Abstracts of the 96th General Meeting of the American Society for Microbiology 1996. Washington, D.C: American Society for Microbiology; 1996. Rapid differentiation and identification of corynebacteria and other selected gram-positive bacilli using single-substrate enzymatic tests, abstr. C-393; p. 71. [Google Scholar]
- 3.Funke G, Carlotti A. Differentiation of Brevibacterium spp. encountered in clinical specimens. J Clin Microbiol. 1994;32:1729–1732. doi: 10.1128/jcm.32.7.1729-1732.1994. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Funke G, Renaud F N R, Freney J, Riegel P. Multicenter evaluation of the updated and extended API (RAPID) Coryne database 2.0. J Clin Microbiol. 1997;35:3122–3126. doi: 10.1128/jcm.35.12.3122-3126.1997. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Funke G, von Graevenitz A, Clarridge III J E, Bernard K A. Clinical microbiology of coryneform bacteria. Clin Microbiol Rev. 1997;10:125–159. doi: 10.1128/cmr.10.1.125. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Hudspeth M K, Hunt Gerardo S, Citron D M, Goldstein E J C. Evaluation of the RapID CB Plus system for identification of Corynebacterium species and other gram-positive rods. J Clin Microbiol. 1998;36:543–547. doi: 10.1128/jcm.36.2.543-547.1998. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Miller J M. Evaluating biochemical identification systems. J Clin Microbiol. 1991;29:1559–1561. doi: 10.1128/jcm.29.8.1559-1561.1991. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.von Graevenitz A, Funke G. An identification scheme for rapidly and aerobically growing gram-positive rods. Zentralbl Bakteriol Parasitenkd Infektionskr Hyg Abt 1 Orig. 1996;284:246–254. doi: 10.1016/s0934-8840(96)80100-9. [DOI] [PubMed] [Google Scholar]
- 9.von Graevenitz A, Pünter-Streit V, Riegel P, Funke G. Coryneform bacteria in throat cultures of healthy individuals. J Clin Microbiol. 1998;36:2087–2088. doi: 10.1128/jcm.36.7.2087-2088.1998. [DOI] [PMC free article] [PubMed] [Google Scholar]