Ability of the MicroScan Rapid Gram-Negative ID Type 3 Panel To Identify Nonenteric Glucose-Fermenting and Nonfermenting Gram-Negative Bacilli

Abstract

The MicroScan Rapid Neg ID3 panel is designed for the identification of Enterobacteriaceae and nonenteric glucose-fermenting and nonfermenting gram-negative bacilli. We evaluated this panel for its ability to identify gram-negative non-Enterobacteriaceae bacteria. A total of 134 strains, representing 26 genera and 42 species, were taken from storage at −70^oC, passaged three times before testing, and inoculated into the panels according to the manufacturer's directions before being inserted into a Walk/Away 96 instrument loaded with version 22.28 software. At the end of the initial 2.5-h incubation period, 89 isolates (66.4%) were correctly identified at a probability level of ≥85%. After additional testing recommended by the manufacturer was completed, another 11 isolates (8.2%) were correctly identified at probability levels of ≥85%. Twenty-five (18.7%) isolates were correctly identified after additional testing, but the probability levels were less than 85%. Two isolates were unidentified, and seven (5.2%) were incorrectly identified. The seven misidentified strains were not concentrated in any one genus. With an accuracy approaching 75%, this product may be used for the identification of the commonly isolated non-Enterobacteriaceae bacteria but may present problems in identification of other non-glucose-fermenting gram-negative bacilli.

Commercial systems for bacterial identification have been available for approximately 30 years. The first manual systems were read visually but were gradually replaced by those more advanced in their substrates. Eventually automated systems that often incorporate substrates that cannot be read visually came on the market. Of the original systems, only the API 20E (bioMérieux, Inc., Durham, N.C.) remains available, while some MicroScan panels (Dade Behring, Inc., MicroScan Inc., West Sacramento, Calif.) and Vitek cards (bioMérieux, Inc.) have made the leap in technology, incorporating many enzymatic substrates that necessitate electronic reading of the tests. Other systems offer a combination of the pH-based and enzymatic tests.

Evaluations of these systems for their ability to identify members of the family Enterobacteriaceae correctly are common, while reports of their accuracy in the identification of nonenteric glucose-fermenting and nonfermenting gram-negative bacilli are not as prevalent. Shelly et al. (6) and Visser et al. (9) recently evaluated rapid panels, while Shelly et al. (6) looked specifically at the ability of conventional panels to identify members of the genus Burkholderia. van Pelt et al. (8) evaluated both conventional and rapid panels for their accuracy in identifying Burkholderia species.

MATERIALS AND METHODS

Isolates.

A total of 134 strains of biochemically typical and atypical glucose-fermenting and nonfermenting nonenteric gram-negative bacilli from the stock culture collection of the Centers for Disease Control and Prevention were used in the study (Table 1). The collection included both oxidase-negative and oxidase-positive strains.

TABLE 1.

Strains tested in Rapid Neg ID3 panel

Reference identificationa	No. of strains
	Total	Correctly identified			Not identified	Incorrectly identified
		At 2.5 h with ≥85% probabilityb	With additional testing
			≥85% probability	<85% probability
Acinetobacter baumannii	7	6				1
Acinetobacter lwoffii	5	2		3
Actinobacillus ureae	1	1
Aeromonas hydrophilia/caviae	2	2
Agrobacterium radiobacter	2	2
Alcaligenes denitrificans	2	2
Alcaligenes faecalis	4		2	2
Alcaligenes xylosoxidans	3		2	1
Bergeyella zoohelcum	3			2	1c
Bordetella bronchiseptica	3	3
Brevundimonas diminuta	2	2
Brevundimonas vesicularis	2	2
Burkholderia cepacia	7	6		1
Burkholderia pickettii	5	1		3		1
Chryseobacterium indologenes	4	4
Chryseobacterium meningosepticum	3	3
Chryseomonas luteola	2	2
Comamonas acidovorans	2	2
Comamonas testosteroni	2			1	1
Empedobacter brevis	2	1				1
Flavimonas oryzihabitans	2	1		1
Flavobacterium odoratum	2	2
Moraxella non-liquefaciens	2	1		1
Moraxella osloensis	2	2
Moraxella phenylpyruvica	2			1		1
Oligella ureolytica	2	2
Pasteurella haemolytica	2	1		1
Pasteurella multocida	3	3
Plesiomonas shigelloides	3	2				1
Pseudomonas aeruginosa	10	8	2
Pseudomonas alcaligenes	3	2		1
Pseudomonas fluorescens	4			3		1
Pseudomonas pseudoalcaligenes	3	2		1
Pseudomonas putida	6		5	1
Pseudomonas stutzeri	5	3		2
Shewanella putrefaciens	3	3
Sphingobacterium multivorum	2	2
Sphingomonas paucimobilis	2	2
Stenotrophomonas maltophilia	10	9				1
Weeksella virosa	3	3
Total	134	89 (66.4%)	11 (8.2%)	25 (18.7%)	2 (1.5%)	7 (5.2%)

^aTaxonomy as reflected in software version 22.28.

^bAs defined by the system.

^cDatabase will not identify an arginine-negative isolate.

Reference method.

Each isolate identification was compared with that obtained by using conventional biochemical tests as performed at the Centers for Disease Control and Prevention (1-3, 10). Commercial media were used whenever possible.

Additional tests.

Additional tests necessary for the completion of an identification by the instrument were performed according to the manufacturer's instructions. These may include any of 42 commonly used biochemical tests plus staining for flagella. The tests that were performed in this study were urea; motility; arginine dihydrolase; fermentation of oxidation-fermentation (OF) glucose, OF maltose, OF mannitol, or OF xylose; oxidase; nitrate reduction; and lecithinase production on egg yolk agar (at 25°C), as well as flagellum staining.

System.

A MicroScan Walk/Away 96 instrument (Dade Behring, Inc., MicroScan Inc.) was used in the evaluation with the Rapid Neg ID3 panel. The software version was 22.28. The Rapid Neg ID3 panel is designed for the identification to species level of rapidly growing aerobic and facultatively anaerobic non-glucose-fermenting and non-Enterobacteriaceae glucose-fermenting gram-negative bacilli. The time to completion of the identification was 2.5 h.

Taxonomy.

Organism identifications are given as listed in the MicroScan database even though some taxonomic changes may have occurred since the database was compiled.

Definitions.

“Correct at 2.5 h” means that the identification is correct to genus and species level at the end of the initial 2.5-h incubation period at a probability level of at least 85%, which is the lower limit of “acceptability” as defined by the software. “Correct with additional tests” means that the identification is correct to genus and species level after additional tests requested by the software were performed. “No ID” means that the software was unable to identify the organism at the end of the incubation time and a report of “No ID” was generated. “Error” means that the identification was incorrect at either the genus or the species level when that identification was included in the database for this panel.

Strains used in the study were removed from −70°C storage and plated on tryptic soy agar with 5% sheep blood (TSAII; BD Biosciences, Sparks, Md.). Incubation was at 30°C over the weekend (2.5 days). The strains were then transferred once more on sheep blood agar before being transferred the third time onto MacConkey's agar (MacConkey II; BD Biosciences). If an organism did not grow on MacConkey's agar, the third transfer was made to sheep blood agar. These last two incubations were at 35°C for 24 h each.

Rapid Neg ID3 panels were removed from the refrigerator and allowed to warm to room temperature before being inoculated according to the manufacturer's directions and loaded into the MicroScan Walk/Away 96 instrument.

If an identification was in error, the strain was transferred again onto the appropriate agar and incubated, and panels were repeated in duplicate. If two of the three answers agreed with the reference identification, the answer was considered correct.

RESULTS AND DISCUSSION

We tested 134 strains that represented 26 genera and 42 different species. Results of this testing are shown in Table 1. At the end of the initial 2.5-h incubation period, 89 (66.4%) of the identifications were correct at a level of ≥85%. Of the 36 strains that were correct after additional testing was completed, the probability levels ranged from 3.4 to 99.9%. Eleven (8.2%) strains gave identifications with probability levels of 85% or greater, while 25 (18.7%) strains gave identifications with probability levels of less than 85% and should be classified as uninterpretable. When the probability is in the lower range, the laboratorian has to make a decision based on colonial morphology, the source of the isolate, and antimicrobial susceptibility data as to whether the identification could be correct. If the likelihood is very low, an alternative system of identification should be used.

Of the remaining nine strains, seven were misidentified and two were unidentified. Table 2 lists the misidentified strains, along with the probability level of the answers that were given. Of the two unidentified strains, one was an isolate of Bergeyella zoohelcum that was negative for arginine. We suspect that the lack of identification was due to the fact that the database matrix has the arginine reaction for this organism set at 90% positive. The printout had this test listed as atypical. The other isolate that the instrument could not identify was Comamonas testosteroni.

TABLE 2.

Strains misidentified by the MicroScan Rapid Neg ID3 panel

Reference identification	Rapid Neg ID3 identification (probability level [%])
Acinetobacter baumannii	Oligella urethralis (88)
Burkholderia pickettii	Alcaligenes sp./xyl/CDC IVC2 (96)
Empedobacter brevis	Bergeyella zoohelcum (94)
Moraxella phenylpyruvica	Bordetella bronchiseptica (98)
Plesiomonas shigelloides	Three different answers (11-99)

As stated earlier, if an initial identification was in error, the strain was transferred again onto the appropriate agar and the panels were repeated in duplicate. Table 3 lists the 20 isolates that were repeated and the number of identifications that were correct on repeat along with the two that gave “No ID” on repeat. Of the 13 identifications that were correct, 8 were of ≥85% probability. We have unpublished data from previous studies that show that many of the initial incorrect identifications are related to the number of passages of an isolate upon removal from −70°C storage and that, upon continued passage, an identification will correct itself.

TABLE 3.

Isolates that gave initial erroneous identifications

No. repeated	Reference identification	No. correct on repeat
1	Aeromonas hydrophila	1
1	Bergeyella zoohelcum	No IDa
2	Burkholderia cepacia	2
4	Burkholderia pickettii	3
1	Chryseobacterium indologenes	1
1	Comamonas testosteroni	No ID
1	Empedobacter brevis	0
1	Moraxella phenylpyruvica	0
1	Plesiomonas shigelloides	0
1	Pseudomonas alcaligenes	1
1	Pseudomonas fluorescens	0
2	Pseudomonas putida	2
2	Pseudomonas stutzeri	2
1	Sphingobacterium multivorum	1
20	Total	13 (65%)

^aID, identification.

Table 4 lists other studies that evaluated either conventional or rapid MicroScan panels for the identification of glucose nonfermenters.

TABLE 4.

Other studies that evaluated MicroScan panels

Authors	No. of strains	% Correct	% Correct-low	% with no identification	% Error	Reference
Shelly et al.	20	75.0		10.0	15.0 (rapid)	6
Visser et al.	41	90.0	7.0		2.4 (rapid)	9
Rhoads et al.	124	92.7		4.0	3.2 (conventional)	5
Shelly et al.	169	76.9		8.2	14.8 (conventional)	6
Sung et al.	301	71.4	24.6	0.7	3.3 (conventional)	7
van Pelt et al.	111	55.0			45.0 (conventional)	8

The study by Shelly et al. (6) looked specifically at the identification of Burkholderia cepacia and showed a 15% (three misidentified isolates) error rate with the Rapid Neg ID panel. Two of the three misidentified isolates were reported as Alcaligenes xylosoxidans, and one was reported as Burkholderia gladioli.

Visser et al. (9) studied 41 strains, including Pseudomonas aeruginosa, Acinetobacter calcoaceticus (baumannii), and Xanthomonas (Stenotrophomonas) maltophilia. Only one strain, of A. calcoaceticus, was misidentified, for an accuracy rate of 97.0%.

In an earlier study of ours (4), 36 isolates of P. aeruginosa, Stenotrophomonas maltophilia, Acinetobacter baumannii, and Acinetobacter lwoffi were correctly identified 81% of the time.

The overall accuracy of the Rapid Neg ID3 panel for the identification of a challenge set of non-Enterobacteriaceae bacteria after additional testing was 74.6% with an additional 20.2% of the isolates having results that either were not able to be interpreted or were unidentified.