Journal of Clinical Microbiology, January 1999, p. 211-214, Vol. 37, No. 1
0095-1137/99/$04.00+0
Copyright © 1999, American Society for Microbiology. All rights reserved.
Identification of Yersinia Species by
the Vitek GNI Card
Hans-Jörg
Linde,1,*
Heinrich
Neubauer,2
Hermann
Meyer,2
Stojanca
Aleksic,3 and
Norbert
Lehn1
Institute for Medical Microbiology and
Hygiene, D-93042 Regensburg,1
Institute
of Microbiology, Federal Armed Forces Medical Academy, D-80937
Munich,2 and
Institute for Hygiene,
National Reference Center for Yersiniosis, D-29539
Hamburg,3 Germany
Received 12 June 1998/Returned for modification 13 August
1998/Accepted 9 October 1998
 |
ABSTRACT |
The Vitek GNI card was used to identify 212 isolates of 10 Yersinia species. Identification was correct for
96.3% of the isolates (156 of 162) to the genus level and for 57.4%
of the isolates (93 of 162) to the species level for
Yersinia spp. listed in the Vitek database. We recommend
additional identification methods for isolates assigned to the
genus Yersinia by the Vitek system.
 |
TEXT |
The genus Yersinia
belongs to the family Enterobacteriaceae and contains 11 species, three of which are pathogenic in humans. Yersinia
pestis is the bacterial agent of plague and was not included in
this study. Yersinia enterocolitica and Yersinia
pseudotuberculosis can cause gastroenteritis and mesenteric
lymphadenitis mimicking appendicitis, but the bacteria may also
cause infections at other sites, such as wounds, joints, and the
urinary tract, or invoke postinfectious sequelae such as reactive
arthritis, urethritis, pleurisy, vasculitis, cholecystitis, and
erythema nodosum. Y. enterocolitica and Y. pseudotuberculosis are the most important causes of
gastroenteritis after Salmonella spp. and
Campylobacter spp. (9). Other species
include Yersinia aldovae, Yersinia bercovieri,
Yersinia frederiksenii, Yersinia intermedia,
Yersinia kristensenii, Yersinia mollaretii, and
Yersinia rohdei, all of which were formerly considered to be
biovars of Y. enterocolitica and can act as opportunistic
pathogens (2, 5, 6, 8, 15, 16). Yersinia ruckeri
causes redmouth disease in salmonids (10).
Yersinia species are relatively slow growers among the
Enterobacteriaceae and display their biochemical
characteristics most reliably at temperatures between 25 and 32°C
(3). Traditional identification is based on a number of
biochemical reactions often not incorporated in commercially
available tests. In Y. enterocolitica, enteropathogenicity is restricted to members of some
serovar-biovar combinations harboring the 64-kDa Yersinia
virulence plasmid. Its presence is demonstrated by a positive
autoagglutination test (1).
The Vitek system (bioMérieux Vitek, Inc., Hazelwood, Mo.) is an
automated miniaturized biochemical test system operated at a fixed
incubation temperature of 37°C. The GNI card is designed to identify
members of the Enterobacteriaceae family and a select group
of nonfermenting gram-negative bacteria. Six species of the genus
Yersinia are listed in the Vitek database: Y. enterocolitica, Y. frederiksenii, Y. intermedia, Y. kristensenii, Y. pestis,
and Y. pseudotuberculosis. The Vitek system is routinely
used for identification of Enterobacteriaceae in our
laboratory. Rapid commercial identification systems generally favor the
faster-growing and biochemically more active
Enterobacteriaceae, which might result in a higher
probability of misidentification for Yersinia spp.
The present study evaluated the ability of the Vitek Gram-Negative
Identification (GNI) card
an automated rapid miniaturized biochemical
identification system
to identify members of the genus Yersinia. Evaluation of the Vitek system has already been
reported for various genera of the Enterobacteriaceae
family, including Yersinia. The present study contains the
largest collection of strains belonging to the genus
Yersinia tested by the Vitek system.
A total of 212 strains belonging to the genus Yersinia was
included in this study. Enteropathogenic and nonenteropathogenic strains of Y. enterocolitica and Y. pseudotuberculosis were isolated from humans. Other isolates
included one reference strain of each species and field strains from
human, animal, and environmental sources. Strains were provided
by the Institute of Hygiene (Hamburg, Germany), G. Wolf
(Munich, Germany), the Central Institute of FAF (Munich and
Berlin, Germany), B. Niederwöhrmeier (Munster, Germany), and by
the Institute for Medical Microbiology (Regensburg, Germany).
Type strains (except Y. pestis) were obtained from the American Type Culture Collection (Rockville, Md.). Strains were maintained at 4°C on nutrient agar and subcultured twice on Columbia blood agar before testing (subculture on
cefsulodin-Irgasan-novobiocin-agar [Oxoid, Wesel, Germany] did not
influence identification results [data not shown]). Conventional
biochemical tests were done as described by Bockemühl
(7) by using the differentiation scheme of Aleksic and
Bockemühl (1), including testing of
enteropathogenicity by autoagglutination. Identification by the GNI
card was carried out in accordance with the instructions of the
manufacturer, including additional tests for melibiose, rhamnose,
raffinose, and salicin (read after 48 h) when necessary. Manual
reading of Vitek reactions was done in accordance with the table
of color reactions provided by the manufacturer.
We investigated a total of 212 phenotypically characterized strains
belonging to the genus Yersinia. For the species listed in
the Vitek database, the GNI card correctly identified 96.3% (156 of
162) of all strains to the genus level and 57.4% (93 of 162 strains)
to the species level. Correct identification to the species level was
44.9% (44 of 98 strains) for Y. enterocolitica, 95.5% (21 of 22 strains) for Y. pseudotuberculosis, 77.8% (14 of 18 strains) for Y. frederiksenii, 41.7% (5 of 12 strains) for Y. intermedia, and 75% (9 of 12 strains) for Y. kristensenii. The time to identification was 9.7 ± 2.7 h (mean ± standard deviation) without and 32.4 ± 24.4 h with the additional biochemical tests recommended in the Vitek
manual. Ninety-two percent of pathogenic strains of Y. enterocolitica (24 of 26) were identified correctly to the species
level, compared to 27% (17 of 63) of nonpathogenic strains
(P < 0.001, chi-square test according to
Pearson). Eighty-four percent of Y. enterocolitica
strains (42 of 51) were misidentified as Y. frederiksenii, and 17.6% (9 of 51) were misidentified as Y. intermedia. For 18 isolates of Y. enterocolitica misidentified as Y. frederiksenii,
Y. intermedia, or Hafnia alvei, the absolute calculated likelihood was
85%. A detailed analysis of the parameters associated with the identification is shown in Tables
1 and
2.
A number of authors have discussed the shortcomings associated with
commercially available miniaturized tests for identifying members of
the genus Yersinia (see overview in Table
3). Because of their slow growth, growth
optimum at 25 to 32°C, and high biochemical similarity,
Yersinia spp. present a special challenge to standard biochemical test systems. However, conventional biochemical testing, which was used as the "gold standard" in this study, might fail to
correctly identify some isolates (11). From a clinical point of view, correct identification of Y. enterocolitica and Y. pseudotuberculosis (Y. pestis not investigated) and separation of
nonenteropathogenic strains from Y. enterocolitica and Y. pseudotuberculosis are most important. In the present study, correct identification of
Y. enterocolitica to the species level was achieved in
only 44.9% of strains and in 18 of 98 (18.4%) cases, the absolute
calculated likelihood was >85%. When the absolute calculated
likelihood is
85%, the result is not likely to be questioned by the
reader. Two of four isolates of Y. frederiksenii
were misidentified as Y. enterocolitica, as were 1 of
12 Y. bercovieri isolates, 1 of 11 Y. mollaretii isolates, and 1 of 10 Y. rohdei
isolates. However, pathogenic strains of Y. enterocolitica were correctly identified to the species level
significantly more often than were nonpathogenic strains,
presumably because of their higher metabolic activity. We made
several attempts to improve inconclusive results by modifying the
test procedure. Using an inoculum equivalent to a McFarland turbidity
standard of no. 2 instead of no. 1 failed to influence profile generation (data not shown).